Meselson, Stahl, and the Replication of DNA: A History of "The Most Beautiful Experiment in Biology" 9780300129663

In 1957 two young scientists, Matthew Meselson and Frank Stahl, produced a landmark experiment confirming that DNA repli

237 95 2MB

English Pages 528 [516] Year 2008

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
Contents
Preface
Acknowledgments
Introduction
CHAPTER ONE. The Replication Problem
CHAPTER TWO. Meselson and Stahl
CHAPTER THREE. Twists and Turns
CHAPTER FOUR. Crossing Fields: Chemical Bonds to Biological Mutants
CHAPTER FIVE. Dense Solutions
CHAPTER SIX. The Big Machine
CHAPTER SEVEN. Working at High Speed
CHAPTER EIGHT. The Unseen Band
CHAPTER NINE. One Discovery, Three Stories
CHAPTER TEN. An Extremely Beautiful Experiment
CHAPTER ELEVEN. Centrifugal Forces
CHAPTER TWELVE. The Subunits of Semiconservative Replication
CHAPTER THIRTEEN. Images of an Experiment
CHAPTER FOURTEEN. Afterword
Abbreviations Used in Notes
Notes
Index
Recommend Papers

Meselson, Stahl, and the Replication of DNA: A History of "The Most Beautiful Experiment in Biology"
 9780300129663

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

MESELSON, STAHL, AND THE

REPLICATION OF DNA

Meselson, Stahl, and the Replication of DNA A History of “The Most Beautiful Experiment in Biology”

Frederic Lawrence Holmes

Yale University Press New Haven & London

Copyright  2001 by Yale University. All rights reserved. This book may not be reproduced, in whole or in part, including illustrations, in any form (beyond that copying permitted by Sections 107 and 108 of the U.S. Copyright Law and except by reviewers for the public press), without written permission from the publishers. Designed by James J. Johnson and set in Melior type by Achorn Graphic Services, Inc. Printed in the United States of America by Edwards Brothers, Inc. Library of Congress Cataloging-in-Publication Data Holmes, Frederic Lawrence. Meselson, Stahl, and the replication of DNA : a history of “the most beautiful experiment in biology”/ Frederic Lawrence Holmes. p. cm. Includes bibliographical references and index. ISBN 0-300-08540-0 (alk. paper) 1. DNA replication—Experiments—History. 2. Meselson, Matthew. 3. Stahl, Franklin W. 4. Molecular biology—Experiments—History. I. Title. QP624 .H654 2001 572.8′6—dc21 2001017701 A catalogue record for this book is available from the British Library. The paper in this book meets the guidelines for permanence and durability of the Committee on Production Guidelines for Book Longevity of the Council on Library Resources. 10 9 8 7 6 5 4 3 2 1

TO THE MEMORY OF

Mary Morgan Stahl August 21, 1934–January 22, 1996

Her graceful spirit touched the lives of all who knew her, even those who knew her too briefly

AND TO THE MEMORY OF

Harriet Vann Holmes December 21, 1932–April 14, 2000

To the very end she kept her warmth, her humor, and her deep interest in the lives of others

Contents

Preface ix Acknowledgments xi Introduction 1

The Replication Problem 11 Chapter Two Meselson and Stahl 49 Chapter Three Twists and Turns 75 Chapter Four Crossing Fields: Chemical Bonds to Biological Mutants 116 Chapter Five Dense Solutions 157 Chapter Six The Big Machine 183 Chapter Seven Working at High Speed 215 Chapter Eight The Unseen Band 272 Chapter Nine One Discovery, Three Stories 303 Chapter Ten An Extremely Beautiful Experiment 319 Chapter Eleven Centrifugal Forces 352 Chapter Twelve The Subunits of Semiconservative Replication 388 Chapter Thirteen Images of an Experiment 412 Chapter Fourteen Afterword 435 Chapter One

Abbreviations Used in Notes 448 Notes 449 Index 497

Preface

In 1957 two young scientists at the California Institute of Technology performed an experiment that provided convincing evidence that DNA replicates in the manner predicted by the model of the double helix proposed four years earlier by James Watson and Francis Crick. Its timely appearance, after several years of controversy about whether the two strands of DNA could come apart without breaking, not only settled the issue as it was originally posed but persuaded many, beyond the immediate circle of enthusiastic supporters, that the double helix was more than an “ingenious speculation.” Quickly known by the surnames of the two men who performed it, the Meselson-Stahl experiment became a classic model in the young field of molecular biology. It has been reproduced in schematic form in textbooks of molecular biology, biochemistry, and genetics for more than three decades. It is seen not only as a landmark but as possessing special qualities that lift it above the thousands of other experiments on which the modern biological sciences have been constructed. When Horace Judson discussed the Meselson-Stahl experiment with John Cairns, Cairns called it “the most beautiful experiment in biology.” The beauty of the Meselson-Stahl experiment is invariably connected with its simplicity. When reduced to its essential features, it is readily understood even by beginning students of the life sciences. Teachers look on it with fondness for the ease with which its message can be conveyed. Scientists throughout history have extolled the simplicity of nature and have admired theories and other discoveries that seem to reveal aspects of that simplicity. But simplicity in science is less a property of nature than a product of the human need to fit representations of nature within the limits of our cognitive capacities. When a simple relationship has been “revealed,” it has, in fact, been

x

P REFACE

extracted from a matrix of complexity. This generalization applies to the Meselson-Stahl experiment with particular force. The experiment originated in complexity, was surrounded by complexity, and directed the way toward the discovery of future complexities. It was the product of a complex investigative pathway. Its beautiful result can be presented as simple only by ignoring the complexity of the reasoning that led to its design, of the instrument on which it was performed, of the prior knowledge on which it was built, and of the human environment in which it was conducted. It is the central aim of this book to contrast the core simplicity of this beautiful experiment and with the many dimensions of complexity that made it possible.

Acknowledgments

The importance of the active participation of the two principal subjects of this book, Matt Meselson and Frank Stahl, is evident on every page. It is now more than a decade since I first showed up on their respective doorsteps to ask questions that required them to plumb memories of events already three decades old. Since then they have given generously of time and support, meeting with me singly and together, in Cambridge, Massachusetts, Eugene, Oregon, and Woods Hole. They have also read successive drafts and corrected my many small errors without seeking to sway the larger direction of my intentions. They must have wondered sometimes whether anything would come of their efforts, and I can only hope that the outcome will be a fair reward for their patience. Gunther Stent received me warmly in Berkeley and answered my questions with refreshing candor and warm civility. A highlight of my work on this project was the afternoon it allowed me to spend in lively conversation with John Cairns at his country home in Charlbury, England. Both Stent and Cairns read more than one version of this manuscript and contributed in important ways to its improvement. James D. Watson received me with hospitality at Cold Spring Harbor, answered my questions, and made available to me pertinent documents from his personal files. Howard Schachman spoke with me in Berkeley. Others who read the manuscript and made valuable suggestions were John W. Drake and Joseph S. Fruton. Charles A. Thomas, J. Herbert Taylor, and Robert L. Sinsheimer supplied helpful information by correspondence. The cogent recommendations of the anonymous reviewers helped to shape the final revisions. William Summers, my colleague at Yale University, who has conducted experiments similar to those described in this book, helped

xii

A CKNOWLEDGMENTS

teach me how to read the ultraviolet absorption films produced in the Model E analytical centrifuge and explained many other technical matters to me. At the University of California at Fullerton, Bruce Weber arranged for me to observe a run of one of the few of the big machines still in operation. As a historian with only undergraduate training in science, I have needed much help from those about whom I have written in this book. Despite their generous assistance, I am bound to have missed some of the deeper levels of the thought and analysis underlying the events described. If we are to give interpretations of the historical development of science that are truly revealing, rather than impositions of our own biases, historians of science must reach, as far as we can, to the levels at which our subjects thought and acted. But most of us will fall short of complete understanding of the complexity of modern scientific specialties, and we must hold ourselves responsible not to make judgments that are beyond our capacities. I have also been greatly helped, in the practical production of a manuscript, by the skillful and devoted work of the staff of the Section of the History of Medicine. Joanna Gorman astutely managed my late transition from the pen to the personal computer and continues to rescue me from the pitfalls into which, from time to time, I still fall. She also prepared the final version of the manuscript. Patricia Johnson arranged the logistics of travel related to the project, acquired material from archives, and, through her efficient management of the life of the Section, protected as much of my time as possible for scholarship. Judith Goodstein and her staff at the California Institute of Technology Archive greatly helped me to find and use documents of crucial importance to this story. The generous assistance of Denise Ogilvie, conservateur at the Service des Archives de l’Institut Pasteur, and Madeleine Brunerie enabled me to locate pertinent documents of Jacques Monod during a brief visit to Paris. A grant from the American Philosophical Society in 1987 enabled me to begin this project. My relatively modest research costs in later years have been covered through a research fund supplied to my faculty position by Yale University. During the last years of this project, my wife, Harriet, endured, bravely and with an undaunted spirit, a long illness. I was grateful that she was still here to share my pleasure in the completion of the manuscript, but saddened that she could not celebrate with me its publication.

Introduction

I In April 1953, the American biologist James D. Watson and the British physicist Francis H. C. Crick proposed in a brief paper in Nature a “structure for the salt of deoxyribonucleic acid (D.N.A.).” 1 Soon known as the double helix, their structural model attracted immediate interest. Not only did the model decisively swing opinion to the view that DNA was the chemical basis of the classical gene; it suggested also how the DNA molecule might function in genetic replication. Coupled with the recently established doctrine that genes control life by directing the synthesis of proteins, the advent of the double helix set off intensive research on the manner in which the sequence of the base pairs in DNA determines the sequence of amino acids in protein. Besides defining the coding problem, this relationship brought into prominence the putative role of the other nucleic acid, RNA, as the intermediary between the DNA contained in cell nuclei and the proteins synthesized in the cytoplasm. Within a decade the discovery of transfer and messenger RNA had resolved the latter problem, and the genetic code had been cracked. As Gunther Stent has put it, the “brilliant wedding of structural and genetic considerations embodied in the DNA helix thus opened the era of molecular biology.” 2 Peter Medawar commented in 1968 that “the great thing about” Watson and Crick’s discovery “was its completeness, its air of finality.” Watson and Crick had not groped toward a partial answer but produced the right solution in one grand stroke. This was a perspective only attainable more than a decade after the discovery, however, when many later developments had both solidified the evidence for the basic features of the model and demonstrated its immense heuristic value for further research. Michael Morange has pointed out that

2

I NTRODUCTION

the very favorable reception accorded the double helix could not conceal how fragile it was during the years following its publication.3 The fragility of the double helix was due not only to the fact, acknowledged by Watson and Crick from the outset, that their general scheme was speculative,4 that it rested mainly on their ability to build a physical model conforming to accepted atomic dimensions, bond lengths, and angles and was compatible with X-ray crystallographic pictures made by others. A more urgent problem arose through the difficulty of imagining how the two nucleotide strands wrapped many times around each other in the double helix could separate, as they were supposed by Watson and Crick to do in the process of duplication. This replication problem, first clearly formulated by Max Delbru¨ck in 1954, vexed the newly emerging field for the next three years. Some people—in particular, physicists who had moved into biology—tried to solve the problem theoretically with various topological schemes. Members of the phage group attempted to solve it experimentally by incorporating into the DNA of bacteriophage or bacteria radioactive isotopes whose distribution they hoped to trace into progeny DNA molecules. All of these efforts were ineffective. In 1957 Herbert Taylor showed by incorporating a radioactive tracer into germinating seedlings that in cell divisions the chromosomes divide semiconservatively, in conformity with the predictions of the Watson-Crick model. Taylor’s evidence was impressive, but it did not reach directly to the replicative process at the molecular level. In 1956, Matthew Meselson and Franklin Stahl began to carry out an idea Meselson had earlier had to investigate the problem by incorporating a heavy isotope into the DNA molecules of a microorganism and tracing the distribution of these atoms into progeny DNA by separating molecules of different density in a centrifuge. In October 1957 they produced the experiment, published eight months later, that quickly appeared to settle the question whether DNA replicated in the manner predicted by the Watson-Crick model. This result played a central role in the transformation of the “fragile” double helix into the robust model seen afterward as the axis around which the new molecular biology revolved. The Meselson-Stahl experiment has already taken its place as one of the mainstream events in the early history of molecular biology. In his broad survey of that history, Horace Judson included a lively account of the origins of this experiment, oriented around several stories Meselson related to him about dramatic moments that had punctuated

I NTRODUCTION

3

the investigation.5 Michael Morange’s shorter history of molecular biology also concludes the chapter on the discovery of the double helix with a summary of Meselson and Stahl’s “demonstration of the semiconservative replication of DNA.” 6 The Meselson-Stahl experiment has thus become a canonical part of the story of the Watson-Crick model of DNA, the event that conferred on the model that air of finality that Medawar attributed retrospectively to the initial announcement of the structure four years earlier.

II The central aim of the present volume is to follow, in as full detail as the surviving documents and the memories of the participants permit, the investigative program that led Meselson and Stahl to perform the classic experiment referred to ever since as the Meselson-Stahl experiment. I have previously reconstructed in a similar manner extended portions of the investigative pathways of three other scientists: Antoine Lavoisier, Claude Bernard, and Hans Krebs. The conviction underlying all these studies has been that if we are to understand deeply how major scientific discoveries originate, we must probe the “fine structure” of the research that produces them down to the level of the daily interplay between thought and action. Synoptic accounts of discovery tend either to leave the impression that scientific investigations proceed methodically, by linear sequences of logical steps to definitive solutions, or that mysterious mental leaps carry creative scientists over the conceptual barriers that do not yield to logic. In order to include the steps later deemed essential to a discovery or a novel scientific achievement, a compressed history usually excludes, for lack of space, the moves that the scientist might have omitted had she known in advance the shortest route to the goal. It is only by following research trials in the richness of their fine structure that we can recognize both that each step of the way may be guided by fathomable reasoning and that the overall pathway is cluttered with unanticipated shifts in direction, goals, and tactics. In his studies of the role of experimental systems in biological research, Hans-Jo¨rg Rheinberger has sought to capture the subtle relation between the control that an experimentalist must maintain over the direction of an investigation and the openness that the system must retain for unanticipated developments. When pursuing an investigation, the investigator never knows in advance where it will come out.

4

I NTRODUCTION

As soon as an outcome is reached, however, the events preceding it begin to reorganize themselves in the minds of the participants and other observers as logical steps leading to an inevitable conclusion.7 The case of the Meselson-Stahl experiment is a prime illustration of the ubiquity of such mental reorganization. In the textbooks that have regularly recapitulated the major outlines of the experiment, it is often depicted as a straightforward exercise in the hypotheticodeductive logic by which science is presumed to advance. A proof was needed that DNA replicates semiconservatively, and through the elegant techniques devised by Meselson and Stahl that proof was duly provided. The experiment appeared so decisive that its result seemed, in retrospect, foreordained by the logic of the situation. In reconstructing the investigative pathway prior to the performance, I have tried to recover the uncertainty about whether Meselson and Stahl would reach their goal. Opportunities arose repeatedly that might have subverted their plan by diverting their attention to other problems. Their eventual success depended on a number of circumstances that they could not know in advance would arise. The successful experiment differed in fundamental ways from the one whose outlines they had in mind when they began. That it was successful depended on a series of fortuitous conditions, some of which did not become evident until after the experiment was received by the relevant scientific community as the confirmation of semiconservative replication. There has been much interest recently, among historians of science, in what is termed “scientific practice.” This history of the Meselson-Stahl experiment can be taken as an episode in the practice of modern experimental biology. The experiment that is the subject of this story is not, however, an actor in it but the passive denouement of many actions taken—most directly by the two young scientists who performed it, indirectly by a number of other scientists who framed the problem the experiment was designed to solve, and at a greater distance by many others who contributed to the repertoire of knowledge and techniques on which the central figures drew to attain their solution. The boundaries of such a story are not sharp and clear. The shape of the experiment was the outcome of multiple interactions, some intellectual, some methodological, some personal, and some institutional. Each of the intersections connects this story with other stories, and the question of how much of the connected stories to include is not easy to resolve. I have chosen to structure the story in the form of a drama in several

I NTRODUCTION

5

acts, with two central characters and a larger cast of other individuals who enter it along the way. Several of the scientists who appear here in supporting roles were leading figures in the development that led to the formation of molecular biology. I have not attempted to summarize their own careers and achievements. Each of them has been treated extensively by other historians. Neither, however, have I limited the narrative to the narrow investigative pathway that the two leading actors followed to the performance of the Meselson-Stahl experiment. Just as scientists reorganize prior investigative moves so that they become logical precursors to the result, so historians are constrained, when we try to account for the origins of an experiment, a discovery, a field or a discipline, to select from the profusion of earlier events those which appear in some degree relevant to the culminating developments in our narratives. Some teleological shaping is inevitable. But we can come closer to the indeterminate conditions out of which other outcomes might have materialized, by allowing some flexibility in our identification of relevant prior events. One of the circumstances relevant to understanding the course of Meselson and Stahl’s investigative enterprise is that they pursued it at Caltech in close association with the phage group led there by Max Delbru¨ck. The role of the phage group in the formation of molecular biology has been discussed at length, in the reminiscences of those who participated in it8 and by historians. The prominence generally attributed to the group has recently been contested. Some historians have pointed out that Delbru¨ck’s scientific achievements have been magnified by the force of his personality. The style of his leadership created an ethos that made his laboratory at Caltech a mecca through which many of those involved in the emergence of the new molecular biology passed during the 1940s and 1950s. His influence was further enhanced by the popularity of the phage course that he taught at Cold Spring Harbor each summer during these years. I have portrayed Delbru¨ck and his group at Caltech as Meselson and Stahl experienced them, but I have not attempted a reassessment of the place of the phage group in the larger events of its time.9 At Caltech Meselson became the last graduate student of the legendary Linus Pauling. There he learned the techniques of X-ray crystallography that Pauling had used to establish the structures of biologically significant molecules. A few years earlier Pauling had established the alpha-helix model of protein structure that inspired Watson and Crick in their efforts to solve the structure of DNA. Al-

6

I NTRODUCTION

though Pauling played no direct part in the investigation that led his last student to the Meselson-Stahl experiment, he did much to inspire the scientific style that Meselson carried into the project. Several biographies and other accounts of Pauling’s life and work have recently appeared.10 Meselson has himself drawn on his experience with Pauling to provide a vivid portrait of Pauling’s personality and attributes as a mentor.11 The appearances of James Watson in the present story outline events that might serve as a potential first chapter for a sequel to the story so engagingly told by Watson in his personal memoir of the discovery of the double helix.12 Judson and others have discussed Watson’s activities during the decade in which the double helix dominated the emerging field of molecular biology, but a full treatment of his role awaits further scholarship. This book is divided into three parts. Part 1 describes the replication problem that arose in the wake of the publication of the WatsonCrick model and various efforts to grapple with it during the following years. It introduces Matthew Meselson and Franklin Stahl, describes the idea Meselson had for resolving the problem, and summarizes their separate research activities while they awaited the opportunity to carry out together a plan to implement Meselson’s idea. Part 2 follows their investigative program from the time they took it up in September 1956 until the publication of their paper “The Replication of DNA in Escherichia coli,” in June 1958. Part 3 treats the reception of the Meselson-Stahl experiment during the years following its publication, further investigations to which it gave rise, its representation in textbooks of molecular biology, biochemistry, and genetics, and some of the reasons for its reputation as a very beautiful experiment. Interwoven with the story of the Meselson-Stahl experiment are two other stories that deal with problems not directly related to the problem of DNA replication. One was the effort of Jim Watson to solve the structure of RNA by the methods that had succeeded so well for DNA. The second was a quest by Meselson and Stahl themselves for a mechanism that would explain mutagenesis at a molecular level. I have interjected these subsidiary stories partly to show that imaginative investigators often entertain multiple research possibilities, and that it is not laid out in advance which ones they will pursue with auspicious success. A second reason for their inclusion is that all three projects were stimulated by the properties of the double helix. They

I NTRODUCTION

7

illustrate the radiating research problems that are created by discoveries with such widespread consequences. I have also given, through a flashback, attention to an event that preceded the discovery of the double helix: the Hershey-Chase experiment, which convinced the group within which Watson’s scientific career was formed that DNA is the hereditary material. This event too is connected through coincidental personal contacts with the main story, but I have included it because it and the Meselson-Stahl experiment stand out as the two eponymous experiments that loom largest on the early landscape of molecular biology. Comparisons between them provide perspective on judgments about experimental beauty, as well as on the attributes that raise a very few, out of the myriad of experiments performed in an investigative field, to canonical status. The narrative of the investigation that Meselson and Stahl pursued for nearly two years is based on surviving correspondence, progress reports, the log records for the experiments performed on the analytical ultracentrifuges at Caltech, the original films that comprise the immediate results of these experiments, and extensive recorded conversations, conducted at intervals spread over more than a decade, with Matt Meselson and Frank Stahl. Full laboratory records of the experiments, if they were ever kept, have been lost. The remaining evidence nevertheless allows a relatively full reconstruction of the day-to-day experimental activity and reasoning of which the Meselson-Stahl experiment was the most dramatic (although far from the only significant) outcome. For the other events included in this book I have relied on published papers, some correspondence made available to me by James Watson from his personal files, and interviews with Watson, John Cairns, Howard Schachman, and Gunther Stent. Cairns, Stent, Jan Drake, Herbert Taylor, and Charles Thomas have supplied me with further recollections by correspondence. My reliance on the memories of participants for some of the information used in the narrative requires commentary. Historians commonly regard such memories as unreliable. They are, however, indispensable for recovering the personal aspects of such an investigative venture that leave few traces in publications or surviving documents. There are often checks on recalled events. Memories fit or do not fit with the information contained in contemporary records. From such checks we can gain a sense of how far we can trust recollections for which there is no corroborating evidence. Some of the events in which

8

I NTRODUCTION

they participated, Meselson and Stahl remember very accurately and clearly. Others they remember vaguely or uncertainly. They have given much time and effort to work with me to reconstruct from their memories, and from the documents that can confront those memories, aspects of their collaboration not otherwise recorded. Where discrepancies have arisen, we have returned repeatedly to the evidence in our efforts to resolve them. Nevertheless, some gaps inevitably remain, and some of the memories remain problematic. In constructing the narrative I have had to apply judgments of plausibility in deciding how much reliance to place on individual recollections of particular events. In most cases I have made these judgments tacitly. In one crucial example related in Chapter 9, I have, however, made explicit the difficulties encountered in reconciling a vivid memory with the surviving records. This example illustrates the general problems that occur whenever we rely on the fertile but elusive traces of past events presented to us by the memories of living participants in those events. The names of Matthew Meselson and Franklin Stahl are indelibly linked through the eponym “Meselson-Stahl experiment” by which their joint achievement is widely known. Does the order of their names reflect only the order of the alphabet, or their relative contributions to the outcome? On this question the two principals disagree. Meselson describes them as equal partners, whereas Stahl insists that the experiment belongs essentially to Meselson. If we view the events leading to the Meselson-Stahl experiment narrowly, it seems clear that Meselson provided the germinal ideas and performed the central operations from which the experiment emerged. But the collaboration in which the two young scientists engaged during the years covered in this story was multifaceted. In other aspects of their common venture Stahl took the lead. I have tried to give equal attention to the parts both men played in this enterprise, but the surviving documentary evidence gives a systematic bias toward fuller description of Meselson’s activities. The existence of the analytical ultracentrifuge log and films enables the reconstruction of nearly every experiment that he performed on those machines. Original records of the operations that Stahl performed to support the centrifuge runs and of the experiments he performed on other aspects of their collaboration have disappeared. I have been able to reconstruct only summary accounts of his activities from correspondence, progress reports, and the memories of the two partners. These

I NTRODUCTION

9

distorting factors should be kept in mind when we balance their respective roles in the events portrayed. In addition to describing their parts in their common scientific venture, I have depicted Matt Meselson and Frank Stahl as two distinct persons at a formative time in their respective careers. To give their individuality some broad contours, I have included glimpses of events in their personal lives that occurred during the time of the narrative, but I have not attempted comprehensive biographical treatments. These are snapshots of two young men at a crucial juncture in their lives, with no pretense at a deeper analysis of the motivations or the earlier developments that brought them to the point at which they entered the stage on which the actions pertinent to the scientific achievement bearing their names took place. For Meselson, as well as for James Watson and others among their contemporaries who were still single, the nonscientific events of their lives often revolved around meeting or establishing ties with women. I have not described any of their encounters with the “woman problem” (as they called it) in detail, but I do mention them repeatedly, in part as a reminder of how different from today the social circumstances of young men and young women in America often were in the 1950s, when they were much less likely to meet in the ordinary course of their daily activity, and how much of their attention was absorbed in the problem of finding one another. Detailed narratives of events on a small scale, such as this history of the Meselson-Stahl experiment, ought also to illuminate more broadly the nature of similar events. The achievement of Meselson and Stahl was singular, but their experiences along the way resonate with those of other scientists who have engaged in laboratory work of this kind. Limitations of space preclude an extended examination here of the generalizable features of this particular investigative pathway, but one of the reviewers of this text for the Yale University Press expressed cogently some of the experiences common to his own that he has found illustrated in this story. They include the observation that the most informative experiments are frequently those which met most difficulties and had, in principle, less chance to be successful, the existence of periods of time in which all the experiments are working, whereas, previously, they were delayed by numerous, different, and frequently unexplainable problems. The psychology of scientists is also depicted . . . the

10

I NTRODUCTION

useless experiments done only to reassure oneself, the difficulty for two researchers to participate fully [and equally in] a decisive breakthrough. It is perhaps the way science develops that is most acutely described; how the objectives are frequently changed, even if the previous research objectives reappear, . . . the important role of informal exchanges between scientists, the permanent, preeminent role of chance events. The coexistence in the same person of stable knowledge and interests and . . . moving goals and occupations.13 Aesthetic judgments are often more important to scientists than is sometimes recognized by those who view science as a coldly methodical activity. The special beauty of the Meselson-Stahl experiment sets it apart from many other research pathways 14 but serves also as an ideal to which scientists frequently aspire. In the last chapter I have mentioned the views of several scientists about what makes this experiment beautiful, but another reader of this text has expressed, better than I have been able to do, the value placed on such experiments by scientific communities: The experiment both confirmed a powerfully heuristic hypothesis (Watson-Crick structure/function model of DNA), and did so elegantly and with perceived simplicity and clear message. Such experiments are rare and when understood by the scientific community are celebrated as particularly noteworthy. [This book shows] us how a work of art, albeit in the form of a scientific experiment, came into being.15

C HAPTER O NE

The Replication Problem

I One of the most famous sentences in the recent literature of science is the statement near the end of the brief article in Nature in which Francis Crick and James Watson announced, in April 1953, their proposed structure for deoxyribose nucleic acid: It has not escaped our notice that the specific pairing we have postulated immediately suggests a possible copying mechanism for the genetic material.1 Crick has since written that his enigmatic assertion had been “a compromise, reflecting a difference of opinion.” He had thought that the paper should discuss the genetic implications, whereas “Jim was against it. He suffered from periodic fears that the structure might be wrong and that he had made an ass of himself.” 2 In his popular narrative The Double Helix, Watson described the same difference of opinion but in a contrasting tone: “For awhile Francis wanted to expand our note to write at length about the biological implications. But finally he saw the point to a short remark and composed the sentence [quoted above].” 3 Privately Watson commented in 1990 that his reluctance about discussing the implications in the article had probably been “a reaction to . . . Francis talking too much.” Francis “talks so much,” Watson said, “that the hope is . . . [to] get him to do an understatement.” Watson’s preference for an understatement reflected also a desire to emulate the British style that he had come to admire during his time in Cambridge.4 Retrospective explanations by these two principals must be viewed with caution because the misunderstandings that arose between Watson and Crick subsequent to the publication of their historic paper

12

T HE R EPLICATION P ROBLEM

may affect the way in which each of them describes this incident. Yet these are not necessarily conflicting accounts of what happened, for each may have experienced their difference of opinion subjectively in the way he afterward remembered it. That Watson really did harbor serious doubts about the validity of their structure for DNA, before and after he and Crick published their first paper in Nature, is clear from contemporary letters that he wrote from Cambridge to Max Delbru¨ck at Caltech. On 12 March he described “our model,” including rough diagrams of the way in which they envisioned the two complementary base pairs, thymine with adenine and cytosine with guanine, to be held together by hydrogen bonds (figure 1.1). Watson went on: The model has been derived almost entirely from stereochemical considerations with the only X-ray consideration being the spacing between the pair of bases 3.4A which was originally found by Astbury. It tends to build itself with approximately 10 residues per turn in 34A. The screw is right-handed. The X-ray pattern approximately agrees with the model, but since the photographs available to us are poor and meager (we have no photographs of our own and like Pauling must use Astbury’s photographs) this agreement in no way constitutes a proof of our model. We are certainly a long way from proving its correctness. To do this we must obtain collaboration from the group at King’s College London who possess very excellent photographs. . . . In the next day or so Crick and I shall send a note to Nature proposing our structure as a possible model, at the same time emphasizing its provisional nature and the lack of proof in its favor. Even if wrong I believe it to be interesting since it provides a concrete example of a structure composed of complementary chains.5 As Watson’s lively account of the events surrounding the elucidation of the structure in The Double Helix shows, he was not entirely candid in this letter to Delbru¨ck about the nature of the X-ray evidence on which they had relied. If they had not yet secured the “collaboration” of the King’s College group, they had already secured some critical information from X-ray photographs taken there. Maurice Wilkins had privately shown Watson a particularly revealing X-ray photograph of the “B” form of DNA made by Rosalind Franklin. Max Perutz then made available to them a report circulated privately to the Medical Research Council that included a discussion by Franklin of the crystalline forms. This information yielded for Crick the critical clue

T HE R EPLICATION P ROBLEM

13

Fig. 1.1. Sketch of base pairs sent by James Watson to Max Delbru¨ck

that the two strands in the double helix must run in opposite directions. Because Franklin was unaware that Watson and Crick had access to her data, Watson was apparently inhibited from acknowledging the way in which they had benefited from her work.6 That circumstance aside, he was here only maintaining a caution appropriate to the boldness of their proposal, its potential importance, and the fact that the structure rested heavily on exercises in model-building that were not universally regarded as sufficient grounds for drawing such conclusions. In the letter to Nature Crick was nearly as cautious publicly as Watson was privately: “The previously published X-ray data on deoxyribo-nucleic acid are insufficient for a rigorous test of our structure. So far as we can tell, it is roughly compatible with the experimental data, but it must be regarded as unproven until it has been checked against more exact results.” 7 By the time Watson sent Delbru¨ck a copy of the draft of the Nature article, on 22 March, he had already obtained additional support for one of the critical assumptions on which his and Crick’s model had been built—the “Chargaff ratios,” or equivalent quantities of the bases in DNA that were paired in the helical model. In the data of Erwin Chargaff on which they first relied, these quantities, measured on the DNA of the bacterium Escherichia coli, were approximately equal. The ratios for adenine-thymine ranged between 1.03 and 1.06, and those for guanine-cytosine varied from 0.85 to 0.93. Gerard Wyatt had published similar results for DNA obtained from insect viruses. Although Wyatt described both ratios as “constant and close to unity,” his results were also less close for guanine-cytosine than for adenine-

14

T HE R EPLICATION P ROBLEM

thymine. By comparison with the variable ratios of adenine-guanine (0.76 to 1.75 in Chargaff’s results) or thymine-cytosine (0.63 to 1.54), these were striking regularities.8 Nevertheless, Watson worried that, although the ratios were “approaching one-to-one, they were not perfect,” and Chargaff himself had not placed much stress on that aspect of his results. During the ten days between his two letters to Delbru¨ck, Watson had visited Wyatt at the Institut Pasteur in Paris, where Wyatt had told him that “the more he refines the analysis of the bases, the closer he finds the 1 to 1 equivalence. This 1 to 1 ratio also holds for [the sum of cytosine and] 5 methyl-hydroxycytosine [found by Wyatt and Seymour Cohen to replace cytosine in phage DNA], which after more careful analysis comes to be equal to guanine.” Wyatt’s new data were the first that seemed to Watson to be “super-good” for their purposes.9 Despite this helpful development, Watson felt ambivalent about his situation: “I have,” he wrote Delbru¨ck, “a rather strange feeling about our DNA structure. If it is correct, we should obviously follow it up at a rapid rate. On the other hand it will at the same time be difficult to avoid the desire to forget completely about nucleic acid and to concentrate on other aspects of life.” 10 Max Delbru¨ck had no doubt about the importance of the new DNA molecule. To Niels Bohr he wrote on April 14, “I think that Jim Watson has made a discovery that may rival that of Rutherford in 1911.” 11 On the same day, in reply to Watson’s letters, he wrote, “I understand things are going well for your DNA structure, and I am not surprised. The more I think of it, the more I become enamored of it myself.” After conversations with several of his colleagues, Delbru¨ck put down “certain considerations” that he wished to state “to see whether we are thinking along the same lines.” The first two points were as follows: (1) In your model the DNA molecule consists of two threads each of which determines the other completely. One thinks of reproduction taking place by separation of the two threads, followed by the formation of a complementary thread by each one of them. (2) The most attractive feature of this model is that for each link to be added a correct choice of only one out of four has to be made. Moreover, the structure is such as to utilize the specific end of the link (the base) directly for steric fit purposes.12 Here Delbru¨ck was not merely rephrasing what Watson and Crick had already written but succinctly drawing “genetic implications”

T HE R EPLICATION P ROBLEM

15

that they had so far refrained from discussing. In their Nature article they had written: “The sequence of bases on a single chain does not appear to be restricted in any way. However, if only specific pairs of bases can be formed, it follows that if the sequence of bases on one chain is given, then the sequence on the other chain is automatically determined.” 13 In describing the model to Delbru¨ck, Watson had only hinted that, if the idea of complementary bases “is right, then I suspect we may be making a slight dent into the manner in which DNA can reproduce itself.” 14 That Delbru¨ck so readily translated their statements about the structure into one about “reproduction taking place by the separation of the two threads, followed by the formation of a complementary thread by one of them” shows just how immediately the structure that Watson and Crick had proposed did suggest a possible copying mechanism. The next point referred to particular implications for the separation and reproduction of the DNA threads in bacteriophage. True to his reputation for raising objections to any significant scientific assertion, however, Delbru¨ck went on to bring up what he took to be a major dilemma arising from the relation between the proposed structure for DNA and its inferred biological function: If we understand your model correctly it implies that the two threads are wound around each other plectonemically (do you remember the terms plectonemic and paranemic from Huskins’ CSH paper . . . ? They are very useful terms in this connection). For a DNA molecule of MW 3,000,000 there would be about 500 turns around each other. These would have to be untwiddled to separate the threads. A feasible way to do this would be to assume the existence of an alternate equilibrium state, in which the double thread is contracted. In contracting, it forms a superhelix (like chromosomes do), and at the same time the threads arrange themselves in a paranemic manner, i.e., such that for each turn of the superhelix the threads turn around each other in a compensating turn. In such a configuration the two threads can be pulled apart sideways without interlocking.15 The two terms to which Delbru¨ck referred had been introduced by C. L. Huskins in a paper read at a conference at Cold Spring Harbor in 1941 that Delbru¨ck had attended. Huskins was describing the various coiled structures that are formed during cell divisions by “chromonemata”—that is, the strands comprising the condensed chromosomes that appear during mitosis or meiosis. “A helix consisting of two

16

T HE R EPLICATION P ROBLEM

strands twisted about each other so that they cannot be separated without uncoiling is termed a ‘plectonemic coil,’ while two helixes which are not intertwined form a ‘paranemic coil.’ ” 16 Huskins found paired chromosomes or chromomenata in both plectonemic and paranemic coils. He also observed strands composed of “major” and “minor” coils, and others containing “reversals of direction.” 17 In responding to Watson and Crick’s helical structure for DNA, Delbru¨ck evidently assumed that the properties of the morphologically visible strands of the hereditary material of cells might also be applicable at the molecular level. “In any event,” he went on in his letter to Watson, “one must postulate that the DNA opens up in some manner, both for replication and for doing its business otherwise. In the structure you describe this opening up is opposed both by the two hydrogen bonds per nucleotide, and by the interlocking of the helices, and it becomes a very important consideration to find a way out of this dilemma, or to think of a modification of the structure that does not involve interlocking. One certainly has to assume that the DNA must go through a cyclic structural change.” 18 After relaying the opinion of his colleague Robert Sinsheimer that, because wheat contains methyl-cytosine in addition to cytosine, “one has to find a partner” for the former in order to avoid a “Waterloo for the whole idea,” Delbru¨ck predicted, “I have a feeling that if your structure is true, and if its suggestions concerning the nature of replication have any validity at all, then all hell will break loose, and theoretical biology will enter a most tumultuous phase.” 19 By the time Watson received Delbru¨ck’s letter, further developments in England had strengthened the case for the DNA structure that he and Crick had worked out. Both Maurice Wilkins and Rosalind Franklin at King’s College had reacted favorably to the model. Franklin’s response “amazed” and relieved Watson. Being under the misapprehension that she was stubbornly “antihelical,” Watson had feared that she might find some reason to reject and cast doubt on his and Crick’s handiwork. The two King’s College investigators each requested permission to submit, simultaneously with Watson and Crick’s note to Nature, papers describing their evidence from X-ray diagrams for the helical structure.20 Watson and Crick were particularly impressed by Franklin’s compelling evidence that the “phosphate groups lie on the outside of the structural unit, on a helix of diameter about 20A. The structural unit probably consists of two coaxial molecules which are not equally spaced along the fibre axis.”

T HE R EPLICATION P ROBLEM

17

These characteristics, as well as the “repeat unit of 34A,” fit harmoniously with the parameters of the model.21 On 2 April, all three papers were submitted to Nature.22 In mid-April Watson had another momentary qualm. Visiting Franklin at King’s College, he found her attempting to measure the diameter of the DNA molecule. Franklin thought that the diameter differed from what Watson and Crick had assumed in their structure. Apparently, the discrepancy was quickly resolved. Watson remained nervous about the propensity of Crick and others to “talk too much” about their grand discovery, before other problems could be ironed out, but his confidence in its basic validity was becoming firm.23 When he responded to Delbru¨ck’s suggestions on 25 April (the same day that the issue of Nature containing the papers appeared), his attitude toward the questions that Delbru¨ck raised was therefore different from what it might have been when he had written Delbru¨ck several weeks earlier. He opened his letter by quoting at length the passages from Franklin’s paper that made, as she put it, “the existence of a helical structure highly probable.” Turning then to the points Delbru¨ck had made, Watson wrote: Thus I am inclined to believe that our structure has a good probability to be correct. However I’m not as yet ready to commit myself that it is right. Thus at present I’m more concerned with seeing whether it is correct than in following up its implications, though it is of course naturally impossible not to occasionally think about them. With regard to your specific points (1) we would also guess that reproduction takes place by separation of the two threads, followed by the formation of a complementary thread by each of them. (2) We are naturally worried about how the threads would untwiddle—the fact that rather frantic coiling does occur during mitosis is comforting but it is difficult to avoid considering the gigantic number of turns which must exist in a chromosome. As far as we know our helix can only be made in the right hand sense and so we cannot use this device for producing compensating coiling. At present we are basically without ideas on this subject. (3) We have to find a mechanism for breaking the two hydrogen bonds. This, I would guess occurs by a tautomeric shift in one of each pair of bases. This might result from a change in pH or possibly by chelation in the purine partner. . . . We are inclined to agree with you that the DNA must go through a cyclic structural change.24 Watson was able to dismiss Sinsheimer’s view that another partner must be found for methyl-cytosine. From Wyatt in Paris he had

18

T HE R EPLICATION P ROBLEM

learned that “the amount of 5 methyl cytosine ⫹ cytosine ⫽ the amount of guanine.” Both bases were, therefore, likely to pair with guanine. Since guanine “cannot distinguish between the two cytosines,” he guessed that the methyl group must be nonfunctional and inquired whether Sinsheimer might be interested in doing an experiment to see whether 5-methyl cytosine is incorporated randomly into the DNA of E. coli.25 Watson’s reaction to Delbru¨ck’s arguments displays a subtle blend of caution and self-assurance. The supporting evidence from King’s College for the DNA model enabled him to move from his position of early March that “we are a long way from proving its correctness” to the assertion that “our structure has a great probability to be correct”; yet he could at the same time allow sufficient remaining uncertainty to justify avoiding a full discussion of the biological implications on which Delbru¨ck had fastened his attention. Even while not committing himself to the correctness of the structure, he could invoke critical features of the structure as a defense against modifications of the model that Delbru¨ck’s compensating coiling would entail. Even while treating Delbru¨ck’s ideas with respect, he could imply that, at this point, to be “without ideas on the subject” might be better than to entertain Delbru¨ck’s idea that the two threads could be in such a configuration as to be pulled apart sideways without interlocking. Nevertheless, Watson must have taken seriously an admonition from Max Delbru¨ck that it was “very important . . . to find a way out of this dilemma.” The dominant member of the phage group in which Watson had “grown up,” Delbru¨ck had been for him the “legendary figure” discussed in Erwin Schro¨dinger’s What Is Life? After spending two summers in the presence of Delbru¨ck at Cold Spring Harbor and one at Caltech, Watson had come to admire especially Delbru¨ck’s “insistence that the results [presented in the many seminars over which he presided] fit into some form of pretty hypothesis.” 26 It was only in keeping with that style that Delbru¨ck now insisted on considering how the structure of DNA could be fitted into a hypothesis explaining how it might function. Meanwhile, Watson and Crick had decided, as Watson explained to Delbru¨ck in a letter on 5 May, that it would be “useful” to write a second letter to Nature, “in view of the abrupt nature of our first note (it was completed before we knew the contents of the notes from King’s).” They were at work on a long manuscript “of a crystallographic type in which we adequately describe the structure,” but it

T HE R EPLICATION P ROBLEM

19

would not appear until early the following year, and in the meantime “it seems logical to emphasize the biological aspects of complementary structures and not to emphasize too strongly the exact details of the structure which may in detail be proved wrong.” 27 The title of their second note, a copy of which Watson sent to Delbru¨ck, was, in fact, “Genetical Implications of the Structure of Deoxyribonucleic Acid.” 28 Within ten days, therefore, Watson appears to have reversed the priorities he had expressed to Delbru¨ck on 25 April in the statement that he was “more concerned with seeing whether [the structure] is correct than in following up the implications.” In their text, Watson and Crick wrote that it had been the “qualitative support” given to their structure “by the X-ray evidence obtained by the workers at King’s College” that made them “now feel sufficient confidence in its general correctness to discuss its genetic implications.” 29 That evidence, however, must have been available to Watson on 25 April, when he wrote Delbru¨ck still resisting temptations to take up these implications. Something else must have persuaded him now to acquiesce in the view that Crick had held from the start: that they should write at length on that subject. Perhaps it was in part that Delbru¨ck’s letters made him realize that, if they themselves did not soon do so, someone else might take the initiative from them. Years later Watson related that when he had been carrying out experiments on X-ray inactivated phage at Caltech in 1949, Delbru¨ck, who had been “only mildly interested” in those results, had “told me that I was lucky that I had not found anything as exciting as [Renato] Dulbecco had, thereby being trapped into a rat race where people wanted you to solve everything immediately.” 30 Now Watson was experiencing the converse of that comparison, and Delbru¨ck himself was among those pressing him for solutions. The second Nature paper, also written by Crick,31 first reviewed, at somewhat greater length, the description of the structure of DNA outlined in the first note. Then it drew out the inferences that this structure held for “the essential operation required of a genetic material, that of exact self-duplication”: The phosphate-sugar backbone of our model is completely regular, but any sequence of the pairs of bases can fit into the structure. It follows that in a long molecule many different permutations are possible, and it therefore seems likely that the precise sequence of the bases is the code which carries the genetical information. If the actual order of the bases on one of the pair of chains were given,

20

T HE R EPLICATION P ROBLEM

one could write down the exact order of the bases on the other one, because of the specific pairing. . . . It is this feature which suggests how the . . . molecule might duplicate itself. . . . Our model for deoxyribonucleic acid is, in effect, a pair of templates, each of which is complementary to the other. We imagine that prior to duplication the hydrogen bonds are broken, and the two chains unwind and separate. Each chain then acts as a template for the formation on to itself of a new companion chain, so that eventually we shall have two pairs of chains, where we only had one before. Moreover, the sequence of the pairs of base will have been duplicated exactly. Following a brief suggestion that this duplication could occur most simply if free nucleotides available in quantity in the cell joined up from time to time on single chains remaining in a helical configuration and were then polymerized, the paper approached the separation problem: Since the two chains in our model are intertwined, it is essential for them to untwist if they are to separate. As they make one complete turn around each other in 34A, there will be about 150 turns per million molecular weight, so that whatever the precise structure of the chromosome a considerable amount of coiling would be necessary. It is well known from microscopic observation that much coiling and uncoiling occurs during mitosis, and though this is on a much larger scale it probably reflects similar processes on a molecular level. Although it is difficult to see how these processes occur without everything getting tangled, we do not feel that this objection will be insuperable.32 While thus acknowledging implicitly the objection that Delbru¨ck had conveyed to Watson, Crick made no concession to it. The question “what makes the pair of chains unwind and separate?” was, in his view, only one of many things that remained “to be discovered before the picture of genetic duplication can be described in detail.” Although the general scheme proposed “must be regarded as speculative,” Watson and Crick felt that their hypothesis, that “the template is the pattern of bases formed by one chain of the deoxyribonucleic acid and that the gene contains a complementary pair of such templates,” might nevertheless “help to solve one of the fundamental biological problems.” 33 Since the beginning of the year, Watson had been negotiating with Delbru¨ck to come to Caltech on a fellowship after he completed his

T HE R EPLICATION P ROBLEM

21

work in Cambridge. He had also raised the question whether funds from the fellowship might be used in advance to enable him to attend the Cold Spring Harbor Symposium on Viruses, scheduled for June. By late April Delbru¨ck decided that the recently discovered structure of DNA was of such relevance to the discussions that would take place at the symposium that he arranged for Watson to be invited as a lastminute participant in the conference. He persuaded H. M. Weaver, the director for research of the National Foundation for Infantile Paralysis, which paid the expenses of all the invited participants, to cover Watson’s round-trip transportation from England and his living expenses at the symposium.34 On 1 May Delbru¨ck wrote Watson: In further explanation of the official invitation . . . let me say that the reference to “your research” (about which you are supposed to have a manuscript ready at the time of the meeting, under penalty of not getting your trip paid), is to your DNA structure, and not your work with [Bill] Hayes [on bacterial genetics]. You are invited because I swore (and Pauling seconded my oath by a long distance phone call to Weaver) that the Watson-Crick DNA structure is of basic importance in connection with at least half a dozen of the principal papers to be given at the Symposium. I also suggested, that, since we would not be able to schedule a major paper by you, that you should be commissioned to draw up a memorandum about the structure and its implications for circulation among all participants before the meeting. “Perhaps,” he suggested, “it would be sufficient to mimeograph the three letters to Nature and to send these around, and let everybody draw his own conclusions.” 35 When Delbru¨ck received from Watson a copy of the manuscript that Crick had written for the second note to Nature, he found his previous objection to the implications of the structure only reinforced. On 12 May he wrote back, Let me start out by stating what I feel about your structure. . . . I am willing to bet that the complementarity idea is correct, on the basis of the base analysis data and because of the implication regarding replication. Further, I am willing to bet that the plectonemic coiling of the chains in your structure is radically wrong, because (1) The difficulties of untangling the chains do seem, after all, insuperable to me.

22

T HE R EPLICATION P ROBLEM

(2) The X-ray data suggest only coiling but not specifically your kind of coiling. I would suggest, therefore, that your second publication deemphasize the mode of coiling. Delbru¨ck suggested further that the note be published in the Cold Spring Harbor Symposium volume rather than in Nature, “because the paper, as it stands, contains too much that is repeated from the first letter.” 36 Watson’s reply of 21 May hints that he was caught in an uncomfortable position between the divergent opinions of two powerful figures in his life—Crick, who regarded the objection to the unwinding of the chains as “not insuperable,” and Delbru¨ck, who thought just the opposite. More generally, he was now having deep pangs about the widespread publicity that the Watson-Crick model was attracting: With regard to your comments on our note: (1) biologically we are unhappy about our plectonemic coiling but (2) we believe we should consider the X-ray evidence and stereochemical consideration first and then worry about the biological complications. If it is not a plectonemic helix, then we would favor a sheet like structure in which the two chains are complementary. As yet, however, we cannot think of a neat way to pack sheets in a way as to give the X-ray pattern, and so we strongly favor a helix. However we may be blind to something obvious. The next paragraphs revealed Watson’s anxieties: Crick was very much in favor of sending in the second Nature note despite the repetition since he feels that most readers of Nature did not understand the first note. To preserve peace I have agreed to it and so it shall come out shortly since Gale (the editor of Nature) is very close to Bragg. It is all rather embarrassing to me since the Professor (Bragg) is frightfully keen about it and insists upon talking about it everywhere. Until we produced the model Bragg did not know what either DNA or genes were and his reaction to our original Nature note was “it’s all Greek to me.” After we had convinced him that DNA might be interesting, he then got out of control and I spend most of my time de-emphasizing it since I have not infrequent spells of seriously worrying about whether it is correct or whether it will turn out to be Watson’s folly. Bragg, however, remains cheerful as ever, and has even told the story to the press and so next Friday’s “News Chronicle” carries a story on how the secret of life was discovered in Cambridge. This

T HE R EPLICATION P ROBLEM

23

immediately led a reporter of “Time” to Bragg and I am dreadfully afraid that I shall see the story in gory print when I am in the States. “I am now working very hard” on a manuscript for the Cold Spring Harbor Symposium, Watson reported. “It is a difficult paper to write since it would be much prettier if we could present a crystallographic proof or disproof of plectonemic coiling. I am assuming, however, that the deadline is June 1st and so we shall emphasize (1) two chains and (2) complementary pairing.” 37 His difficulty in writing this paper was hardly diminished by the fact that Watson was preparing to present it in precisely the setting habitually dominated by Max Delbru¨ck. If he did not take sufficient account of Delbru¨ck’s “insuperable” objections to the plectonemic helix, he could expect to be subjected to the trenchant criticism that Delbru¨ck characteristically delivered on such occasions. If, on the other hand, he conceded too much to Delbru¨ck, it might become difficult for him to keep the peace with Crick. When scientists write successive papers on the same ongoing or completed investigation, the resulting texts are commonly not independent productions but variations on a theme, orchestrated for particular occasions and audiences. The paper that Watson composed on the structure of DNA during the month of May incorporated much of what had already appeared in the two Nature articles, as well as information from the accompanying papers of Franklin and Wilkins. All of this was recast to adapt it to the forum he had to address. The opening paragraph was designed clearly to connect what he wished to report with the topic of the Cold Spring Harbor meeting: It would be superfluous at a Symposium on Viruses to introduce a paper on the structure of DNA with a discussion on its importance to the problem of virus reproduction. Instead we shall not only assume that DNA is important, but in addition that it is the carrier of the genetic specificity of the virus . . . and thus must possess in some sense the capacity for exact self-duplication. In this paper we shall describe a structure for DNA which suggests a mechanism for its self-duplication and allows us to propose, for the first time, a detailed hypothesis on the atomic level for the selfreproduction of genetic material.38 After this diplomatic nod Watson made little further mention of viruses. The first four sections of the paper—“Evidence for the Fibrous Nature of DNA,” “Evidence for the Existence of Two Chemical Chains

24

T HE R EPLICATION P ROBLEM

in the Fiber,” “Description of the Proposed Structure,” and “Evidence in Favor of the Complementary Model”—repeated much of what Crick had written in the earlier papers. In places phrases extracted from the earlier pages were reorganized to shift the emphasis. As he had indicated in his letter to Delbru¨ck, Watson amplified the aspects of the argument that stressed the two chains and complementary base pairs. He also provided more details concerning the evidence on which the structure was based, particularly that drawn from the X-ray fiber diagrams of Wilkins and of Franklin. Whereas Crick had written in the first Nature note that the structure rests mainly on “published experimental data and stereochemical arguments,” acknowledging only “stimulation” from the “general nature of the unpublished results and ideas” of Wilkins and Franklin, Watson now offered the structure of DNA as one that he and Crick had proposed “to account for these findings.” 39 The incompatibility of these two statements is self-evident. Both appear to reflect Watson and Crick’s embarrassment over the way in which they had been given access to the “findings” for which their structure accounted. Now that the results and ideas of Franklin and of Wilkins were published, Watson and Crick could safely leave the impression that they based their structure for DNA on detailed evidence that had, in fact, become public knowledge only after they had constructed their model. Section V, “Genetic Implications of the Complementary Model,” expanded considerably on the corresponding discussion in the second Nature article. It was here that Watson labored to satisfy both Delbru¨ck and Crick. Rather than follow Delbru¨ck’s advice to de-emphasize the method of coiling, he and Crick had evidently decided that they must meet his objection head on and fully defend their position. In two paragraphs the first subsection described a mechanism for DNA replication very much as the Nature paper had done. In place of the single paragraph in which Crick had brushed off objections to the unwinding of the strands as not insuperable, however, Watson devoted more than a quarter of the paper to a subsection titled “Difficulties of the Replication Scheme.” 40 Watson recognized three main objections. The first, that DNA contains 5-methyl cytosine in addition to cytosine, he could readily answer with the data of Gerard Wyatt showing that the sum of the amounts of cytosine and 5-methyl cytosine is equal to the amount of guanine. The second objection, that “our scheme . . . completely ignores the role of the . . . proteins known to be combined with DNA

T HE R EPLICATION P ROBLEM

25

in most living organisms,” he deflected with the observation that “as yet nothing is known about the function of the protein.” The third difficulty involves the necessity for the two complementary chains to unwind in order to serve as a template for a new chain. This is a very fundamental difficulty when the two chains are interlaced as in our model. The two main ways in which a pair of helices can be coiled together have been called plectonemic coiling and paranemic coiling. These terms have been used by cytologists to describe the coiling of chromosomes. . . . The type of coiling found in our model . . . is called plectonemic. Paranemic coiling is found when two separate helices are brought to lie side by side and then pushed together so that their axes roughly coincide. Though one may start with two regular helices the process of pushing them together necessarily distorts them. It is impossible to have paranemic coiling with two regular simple helices going around the same axis. This point can only be clearly grasped by studying the models.41 It was perhaps because of the difficulty of visualizing these complex spatial relations from verbal descriptions, or from the twodimensional schematic diagram published in the Nature papers, that Watson decided to have built for him, in the Cambridge machine shop, a small wire model that he could carry with him to the symposium.42 Having rejected the bet Delbru¨ck had made that plectonemic coiling “is radically wrong,” Watson had now to confront the unwinding difficulty. “The difficulty is a topological one,” he wrote, “and cannot be surmounted by simple manipulation. Apart from breaking the chains there are only two sorts of ways to separate two chains coiled plectonemically.” That the two chains might be pulled apart in the axial direction he considered “highly unlikely.” They must therefore “be directly untwisted.” Addressing himself to the problem of how many turns must be made, and how is tangling avoided, Watson estimated a lower limit of one thousand turns, based on the molecular weight of isolated DNA fibers, and an upper limit of twenty thousand turns based on the total DNA in a virus. In higher organisms the number might be “1,000 fold higher.” 43 “The difficulty might be more simple to resolve,” Watson acknowledged, “if successive parts of a chromosome coiled in opposite directions. The most obvious way would be to have both right and left handed DNA helices in sequence but this seems unlikely as we have

26

T HE R EPLICATION P ROBLEM

only been able to build our model in the right handed sense.” Having thus tacitly eliminated Delbru¨ck’s suggestion that compensating coiling might alleviate the unwinding problem (as he had already explicitly done in his letter of 25 April to Delbru¨ck), Watson turned to the danger of tangling. This problem would be considerably decreased if replication began at the ends as soon as the chains started to separate. The structure would remain rigid, and “the growing end of the pair of double stranded structures might facilitate the breaking of hydrogen bonds in the original unduplicated section and allow replication to proceed in a zipper-like fashion.” He allowed also that one chain of a pair might “occasionally” break “under the strain of twisting.” The accumulated twist would then be relieved by rotation of the second chain, after which the broken ends “might rejoin.” 44 In spite of these tentative suggestions, the difficulty of untwisting is a formidable one, and it is therefore worthwhile re-examining why we postulate plectonemic coiling. . . . Our answer is that with paranemic coiling, the specific pairing of bases would not allow the successive residues of each helix to be in equivalent orientation with regard to the helical axis. This is a possibility we strongly oppose as it implies that a large number of stereochemical alternatives for the sugar-phosphate backbone are possible, an inference at variance to our finding, with stereochemical models . . . that the position of the sugar-phosphate group is rather restrictive and cannot be subject to the large variability necessary for paranemic coiling. Moreover, such a model would not lead to specific pairing of the bases, since this only follows if the glucosidic links are arranged regularly in space. We therefore believe that if a helical structure is present, the relationship between the helices will be plectonemic. Elaborating a possible alternative that he also mentioned in his letter to Delbru¨ck while working on the paper, Watson added: We should ask, however, whether there might not be another complementary structure which maintains the necessary regularity but which is not helical. One such structure can, in fact, be imagined. It would consist of a ribbon-like arrangement in which again the two chains are joined together by specific pairs of bases, located ˚ above each other, but in which the sugar-phosphate back3.4 A bone, instead of forming a helix, runs in a straight line at an angle approximately 30° off the line formed by the pair of bases. While this ribbon-like structure would give many of the features of the

T HE R EPLICATION P ROBLEM

27

X-ray diagram of structure B [the crystalline form assumed by DNA at high humidity], we are unable to define precisely how it should give a strong equatorial reflexion at 20–24 A. We are thus not enthusiastic about this model though we should emphasize that it has not yet been disproved.45 Even though his text did not mention Delbru¨ck, it is obvious in the light of the correspondence between them that Watson was composing a public answer to the private objections Delbru¨ck had raised. Essentially his response was that he and Crick had not found the way out of the dilemma but that they had compelling reasons to remain in it rather than to choose the route that Delbru¨ck offered. Delbru¨ck wanted to have complementarity without plectonemic coiling. Watson and Crick were telling him that he could not have one without the other, because the plectonemic relation between the two polynucleotide strands was essential to the structure that defined complementary base pairs. The answer to Delbru¨ck’s claim that “the X-ray data suggest only coiling but not specifically your kind of coiling” was subtler, because it could not be conveyed entirely in words. Delbru¨ck had seen only summaries of the data and schematic drawings of the double helix. He had not had the experience that Watson and Crick had had constructing models to fit the data. Few scientists besides these two had such experience. Here they were saying to him, if you try it, you will see that you cannot construct your paranemic kind of coiling in a manner that is compatible with the data. Why were Watson and Crick, who still acknowledged that their model had not been proved correct, so confident in it as to maintain that the “formidable” difficulties that their replication scheme faced were not insuperable? Here, they were, of course, not of one mind, even when they spoke publicly with one voice. Crick seems not to have felt the doubts that sporadically afflicted Watson. Watson’s determination to stick with their solution rested probably as much on feeling and aesthetics as on the strength of the evidence supporting it. Years later, in his text Molecular Biology of the Gene, he wrote of the double helix: “Before the answer was known, there had always been the mild fear that it would turn out to be dull, and reveal nothing about how genes replicate and function. Fortunately, however, the answer was immensely exciting.” 46 Both the fear and the excitement to which he alluded had been his own. Ever since reading Erwin Schro¨dinger’s What Is Life? as an undergraduate at the University of Chi-

28

T HE R EPLICATION P ROBLEM

cago, he had been “polarized toward finding out the secret of the gene.” 47 He had, therefore, a great deal of emotional investment in the outcome that he and Crick had together reached. The qualities that made the model exciting were, ironically, the same ones that made Watson sometimes feel that it might prove to be his folly. He conjured up his apprehension retrospectively, in 1990, in the elliptical comment, “You know, there was a beautiful model, but it wasn’t correct.” 48 The search for beauty is a powerful motivating force, in science as in life; but he was well aware that, in both realms, beauty is seductive. Looking forward to the chance to see the United States after an absence of three years, Watson flew across the Atlantic on 2 June and came directly to Cold Spring Harbor, where the symposium opened on 5 June.49 Two hundred and seventy-two scientists attended—the largest gathering ever in the series of symposia on quantitative biology that had been held annually since 1932. The meetings were held in the lecture hall of the laboratory.50 As he had planned, Delbru¨ck circulated the copies of the three letters to Nature by Watson and Crick, Wilkins, and Franklin before the meeting began, so that the participants would be prepared for the discussion that he expected Watson’s talk to evoke.51 In his presentation Watson showed slides of Wilkins’s and Franklin’s X-ray diffraction pictures, as well as the previously published schematic drawings of the double helix and the polynucleotide chains and scale drawings of the base pairs, and he displayed the model he had brought with him. One of the younger participants in the meeting, Franc¸ois Jacob, has described the feeling of this dramatic moment: “With an air more bewildered than ever, his shirt fluttering, wide-eyed, his nose in the air, interrupting his discourse with brief exclamations underlining the importance of his subject, Jim explained the details of the structure.” After he had finished describing the play with models, the crystallographic arguments, the physical and chemical characteristics of the molecule, and the genetic implications for replication and mutation, “for a moment the hall remained silent. There were a few questions. How, for example, can the two chains wrapped around each other separate during the replication of the double helix without breaking? But no criticism. No objections. There was in that structure such a simplicity, such a perfection and harmony, such beauty even; the biological advantages flowed from it with such rigor and such evidence, that one could not believe that it was not true.” 52

T HE R EPLICATION P ROBLEM

29

That the question about how the chains could separate did not lead to objections such as those Delbru¨ck had raised privately suggests that Watson’s thoughtful defense of his and Crick’s position against Delbru¨ck’s must have preempted Delbru¨ck from pressing the issue during the discussion. When Cy Levinthal congratulated him after his talk, Watson replied, in the English-style understatement he emulated then, that Crick’s skill at X-ray crystallography had made it all easy.53

II It was obvious to all who heard Watson’s Cold Spring Harbor talk with such intense interest that if the postulated structure were confirmed it would bring radical changes to their understanding of the replication of viruses in particular and to genetics in general. Few were persuaded on the spot, however, that the structure was firmly established. In the papers that they had so far published or presented, Watson and Crick had outlined the general features of their structure and asserted its compatibility with the characteristics of the molecule that could be inferred from X-ray diagrams, but they had as yet not revealed the detailed “stereochemical arguments” on which they claimed to have based their model. The prevailing approach among those who had read the Nature articles or had heard Watson speak was probably to await with open minds for further information—unless, like the plant physiologist Barry Commoner, who, according to Watson’s recollection, “hated the talk,” they adamantly opposed Watson and Crick’s strategy of ignoring the protein component of the gene.54 Many of the participants must also have puzzled over the conundrum of how the two strands of a coiled helix could separate. To judge from his later actions, Delbru¨ck must have been persuaded to give up his idea that paranemic coiling or compensatory winding could solve the problem, but he was still dissatisfied with Watson and Crick’s optimistic belief that the two strands could somehow untwist. Among those present who were stimulated to think about the problem was Robert Sinsheimer, a biophysicist at Iowa State College who had spent the previous year at Caltech studying the degradation of deoxyribonucleic acid to dinucleotides and mononucleotides. During the meeting Sinsheimer made the suggestion that, if DNA is a two-stranded helix as the Watson-Crick model asserted, then when it is degraded by the enzyme DNAse “there ought to be a lag in the disintegration of the

30

T HE R EPLICATION P ROBLEM

two-ply molecule as presumably more or less adjacent bands in the two chains would have to be broken before the chain length could decrease.” 55 Consulting with Paul Doty at Harvard University, Sinsheimer learned that Doty’s data on the rate of DNAse degradation of DNA measured by light-scattering methods showed “just such a lag in the decline in average molecular weight,” although his own data on the release of titratable acid showed no lag. This difference suggested that as individual bonds began splitting on each chain, some time would pass before the breaks would occur close enough together along both chains to cause the molecule to come apart. To Delbru¨ck he wrote, “Score one for Watson and Crick.” Pondering the unwinding problem, Sinsheimer thought that if “only one chain is important and the other may be discarded, then one might go further and assume that the latter chain is not really complete but is broken maybe every 30–50 nucleotides. . . . Such a state of affairs would of course greatly simplify the unraveling under some set of conditions that would release the H bonds.” Sinsheimer saw no evidence against such an idea.56 Delbru¨ck undoubtedly did not like it, because the distinction it made between the status of the two chains violated the symmetry of the DNA structure; but he may not, at the moment, have had any better thoughts on the subject. Soon after Watson returned to Cambridge, he received from Linus Pauling an invitation to attend a protein conference that Pauling had organized for September in Pasadena. Watson made arrangements to come to Caltech in time for the conference and to begin his research fellowship immediately afterward. The main task that he had to complete before leaving Cambridge was the longer paper on DNA on which he and Crick had begun to work in March, the paper that would give the coordinates and other structural details of the model. Because Crick was now busy finishing his thesis before his impending departure to the United States to take up a position at the Brooklyn Polytechnic University, the task of writing fell to Watson. He had it finished by early August. The Royal Society received it from Bragg on 24 August, for publication in the Proceedings. It would, therefore, not actually appear until the following spring.57 Although there was some overlap between the paper on the complementary structure of deoxyribonucleic acid and the three papers on the same subject that had preceded it, it was quite different in character. Not only did it concentrate on the structural details of the model,

T HE R EPLICATION P ROBLEM

31

with only brief reference to the biological implications; for the first time in print Watson and Crick made it clear that their model rested on “stereochemical arguments” and that they had arrived at it by actually constructing physical models in laboratory space. “It has seemed worthwhile,” Watson wrote, “for us to build models of idealized polynucleotide chains to see if stereochemical considerations might tell us something about their arrangement in space. In doing so we have utilized interatomic distances and bond angles obtained from the simpler constituents of DNA and have only attempted to formulate structures in which configurational parameters assume accepted dimensions. We have only considered such structures as would fit the preliminary X-ray data of Wilkins, Franklin and their co-workers. Our search has so far yielded only one suitable structure.” 58 The paper indicated how the spacings of the reflections on the Xray pictures imposed severe restrictions on the types of models that could be built and how the possibilities allowed by the X-ray data can be differentiated by building models. Although he did not mention all of the false starts and detours about which he later gave so entertaining an account in The Double Helix, Watson did describe the considerations that had led him initially to believe that the phosphate groups should be in the center. He explained how they came to realize that this approach would lead nowhere, gave up the attempt, and decided it was “most likely that the bases form the central core and that the regular sugar-phosphate backbone forms the circumference.” 59 Less conspicuously, the paper revealed also the strategic importance for Watson and Crick of a paper published in 1950 by Sven Furberg on the crystal structure of cytidine. Cytidine is the nucleoside composed of cytosine and a deoxyribose sugar molecule. Because there were no published reports of the structure of cytosine alone, Furberg’s paper served as the only source of information on the dimensions of this base. Moreover, Furberg had concluded that, contrary to the suggestion by William Astbury that the rings of the sugar and of the base are parallel, “they are oriented in such a way that they are nearly perpendicular to each other. This would seem to be a point of considerable importance for the understanding of the structure of the nucleic acids.” Furberg’s prophecy was dramatically fulfilled when Watson and Crick adopted this perpendicular orientation of the two rings in the construction of the double helix.60 The paper provided scale drawings of the base pairs, projections of the spatial arrangements of the phosphate-sugar backbone, and pho-

32

T HE R EPLICATION P ROBLEM

tographs of the simplified wire model that Watson had carried to Cold Spring Harbor.61 Short of being able to see and touch an actual physical model of the molecule, a reader of this lucidly written and carefully illustrated paper could come as close as possible to a full appreciation of the compelling stereochemical arguments, as well as the celebrated beauty of the resulting structure. Here, too, Watson revealed clearly why for him and for Crick the complementarity of the base pairs was inseparable from the helical structure of their model. “We should note the reason why the two chains cannot be linked together by two purines or by two pyrimidines. It arises from our postulate that each of the sugar-phosphate backbone chains is in the form of a regular helix.” 62 To be sure, Watson tacitly acknowledged in the discussion section that the same conclusion about base pairing might have been reached from the data of Chargaff and of Wyatt alone. “It is difficult to imagine a structural explanation for the equivalence of adenine with thymine and of guanine with cytosine which does not involve specific pairing.” 63 Their resistance to considering alternative structures that might preserve base pairing while obviating the obstacles to replication posed by the double helix was logical, in that they had experimental evidence that the helical structure existed. But their lack of enthusiasm for alternatives was also psychologically reinforced by the history of their quest for the structure. It had been by building helices that they had come to recognize the special feature of their DNA model—its restrictions on base pairing—that imparted to it its exciting genetic implications. By the time Watson reached Caltech in September, he had lost all interest in DNA. No longer doubting the correctness of the model— perhaps the experience of writing the detailed description of its structure had bolstered his confidence in the solidity of his and Crick’s arguments—he felt that that problem was now solved. He probably saw no way to attack the unsolved problem of its replication. He had earlier intended to start working on phage during his fellowship, a highly reasonable plan, because Caltech under Delbru¨ck’s leadership had long been a mecca for phage research. Already in the spring of 1952, however, he had written Delbru¨ck that he would “like to go on to the structure of RNA” when he returned to Pasadena. When he arrived in the fall of 1953 he found that Alexander Rich in the chemistry division had recently begun to take X-ray pictures of ribonucleic acid. At that point Watson became “totally fixated on RNA.” 64 In the ensuing collaboration, Rich continued to take the X-ray pic-

T HE R EPLICATION P ROBLEM

33

tures. Watson contributed a method to obtain RNA in fibers ordered in a crystalline or semicrystalline form that would yield distinct diffraction reflections. Adapting a technique that Maurice Wilkins had used with DNA, he drew RNA out into fibers more than a centimeter in length. When he and Rich saw that these fibers were, like those of DNA, birefringent, they quickly became excited, because this feature indicated to them that the nucleotide bases were probably, like the bases in DNA, perpendicular to the fiber axis.65 The X-ray pictures that Rich took with these fibers were less well resolved than those that Wilkins and Franklin had obtained with DNA; yet their pattern resembled the DNA patterns sufficiently to suggest to Watson something that looked “slightly like a double helix.” 66 In November he reported to Crick by letter, “Naturally I’ve tried model building.” 67 Not long afterward he wrote in the annual report of research in the Caltech Division of Biology that he and Rich “hoped to establish the three-dimensional shape of this compound [RNA] and, if possible, to find a relationship between its structure and function.” X-ray diffraction patterns for all RNAs examined up to that time appeared similar enough to suggest that there was only one type of RNA structure. The pattern had some resemblance to that for DNA. They were attempting to build stereochemical models with “particular attention . . . to possible helical structures.” Although the results had not been encouraging, they felt that “model building . . . may in the final analysis be the most profitable approach to a solution of the structure.” Watson was clearly betting that the assumptions and strategies that had led him and Crick to their recent triumph with DNA could be transferred to the chemically similar RNA.68 Despite picking up this scent of a possible sequel to the DNA success, Watson was thoroughly unhappy in Pasadena. The dominant cause of his somber mood was the prospect that he would be called up for military service. Almost as soon as he arrived, he was notified that he had been reclassified 1-A. Appeals made on his behalf to defer him because of his importance to the work of the virus group at Caltech were turned down by his draft board, and he felt that the army might take him “at any moment.” 69 Watson was, in addition, disappointed with Pasadena itself. Compared to the enchantment of Cambridge, the suburban area around Caltech appeared sterile, and he disliked the smog. Worst of all, there “was no social life.” Still as preoccupied with meeting pretty women as he appears from his self-portrait in The Double Helix to have been

34

T HE R EPLICATION P ROBLEM

during his three years in Europe, Watson quickly realized at Caltech that “there were no women there.” For him that made it “a pretty bleak place.” 70 Most disheartening of all to Watson was that coming back to Delbru¨ck’s virology and biophysics section turned out to be a letdown. During the summers he had spent there he had been caught up in the excitement of the phage group, and like others, he had felt the enormous charm that Delbru¨ck exerted on those whose work he liked. Having long anticipated his return to this setting, Watson arrived after three heady years at the Cavendish laboratory only to find that everything seemed different to him. Compared to people like Crick, John Kendrew, Max Perutz, and the others in Cambridge, those in the Biology Division at Caltech appeared to be good but dull workers. Enamored by the facility and wit with which the English used words, he found Pasadena “verbally boring.” Over in the Chemistry Division the great Linus Pauling, whose methods Watson and Crick had applied to such advantage to build the double helix, seemed distant and aloof. Cambridge seemed to him still the center of his world, and he in faroff exile.71 Max Delbru¨ck himself suddenly ceased to be Watson’s hero. For years Watson had written Delbru¨ck regularly about his scientific activities and plans, seeking advice and approval. Watson expected Delbru¨ck, of all people, to appreciate fully the significance of the double helix and to pursue its implications for his own long-standing interest in the replication of bacteriophage. Instead, Delbru¨ck was just at that time making a more radical shift in his research interests. Feeling that phage genetics might have become too fashionable for him, Delbru¨ck gave up his own phage work and sought a different arena in which to probe his persistent belief that biological phenomena would eventually reveal deep paradoxes in the laws of physics. He now thought he might find such phenomena in the basic mechanisms through which living organisms react to their environments. Seeking the simplest system imaginable in which such reactions occur, he began a study of the phototropism of the single-celled fungus Phycomyces. While Watson was taking up the structure of RNA in Caltech’s Kerckhoff Laboratory of Biology, Delbru¨ck was studying there the relation between changes in the intensity of illumination and transient changes in the velocity of growth of sporangiophores on a slime mold.72 Watson thought that Delbru¨ck’s work on Phycomyces was boring. Instead of stimulating Delbru¨ck to examine its implications for phage

T HE R EPLICATION P ROBLEM

35

genetics, the double helix had actually made it easier for him to leave genetics. Delbru¨ck saw that the discovery of the structure of DNA would make genetics increasingly molecular. As one who disliked biochemistry and knew little about it, he did not wish to move in that direction. Watson now began to perceive Delbru¨ck as someone who seemed to want to solve fundamental biological problems without learning the facts of biology. Unlike Francis Crick, another physicist who “switched over,” Delbru¨ck remained, in Watson’s view, “always a physicist looking at biology rather than a molecular biologist.” Consequently, Watson lost interest in what Delbru¨ck thought. Given his current location, that was an awkward situation.73 Those in Delbru¨ck’s group who still were doing phage genetics did not seem to Watson to be studying the right problems. Work such as Jean Weigle’s research on the induction of phage mutations by ultraviolet irradiation, Joe Bertani’s work on the inheritance of P2 prophage in bacterial crosses, Robert DeMars’s investigation of genetic recombination of UV irradiated phage T2, or George Streisinger’s study of the genetics of T2 and T4 host ranges continued the classical methods of studying genetic crosses, identifying genetic markers, and measuring linkages between markers. To Watson it seemed that they were working as if the double helix did not exist. No one changed what he was doing because of it.74 Feeling intellectually isolated, and missing particularly the stimulation of daily conversations with Crick, Watson fluctuated from high optimism to discouragement over his progress with RNA. Late in the fall he believed that he was about to attain the correct structure. One of the obstacles that he still had to overcome, however, was that some chemists thought that, unlike DNA, the polynucleotide chains of RNA might be branched. Branching would play such havoc with the search for a structure that Watson hoped he could somehow rule out that possibility. Early in January he wrote Delbru¨ck from Washington that he would like “to convince a chemist that RNA is an unbranched 3–5 chain, before sticking my neck out on a structure.” By late January he had come to feel that he was still far from the solution. Nevertheless, having interested the physicist Richard Feynman—in his opinion “the best person in Caltech if not in the States”—in the problem, he no longer felt isolated: “Instead, so stimulated that I cannot sleep much.” 75 In early February Watson was full of enthusiasm for a new scheme that he had devised. By reading up on the literature about base ratios

36

T HE R EPLICATION P ROBLEM

in RNA he convinced himself that in all RNA, except for that of plant viruses, the ratios are complementary. Inferring that all RNA, whether single- or double-stranded, replicates as DNA does by forming complementary strands, he decided that the reason DNA has two strands is that one of them retains the code, while the other is transformed to RNA, which then crosses into the cytoplasm to make protein. He wrote to Crick on the thirteenth that he had persuaded Feynman of his scheme “and slightly Delbru¨ck.” Although the idea “is slightly mad, as it is cute I think it is correct.” Three days after the idea came to him, Watson impulsively began to write a letter to Nature.76 Alex Rich was also caught up in the enthusiasm. He built a large helical model of RNA that contained twenty different trapezoidal holes into which, inspired by George Gamow’s coding scheme, he believed the twenty amino acids of the proteins synthesized by RNA could fit. When Gamow drove to Pasadena to inspect the model, however, he found that the combination rules that he had formulated would not work with the model. He wrote Crick that Watson and Delbru¨ck did “not believe in it very much” either.77 Delbru¨ck left Caltech for Germany in March to spend three months at Go¨ttingen. Watson recovered from his fantasy that he could solve all the mysteries of life and saw that there was much work to be done before he could dispatch another thunderbolt to Nature. In late March, Watson wrote Delbru¨ck that he and Leslie Orgel, a member of Pauling’s group, had been “giving RNA another serious going over—I believe with some success.” They were “observing a very pretty reversible change in the RNA fibre length which occurs upon raising or lowering the relative humidity. This change in fibre length can be correlated with changes in the X-ray pattern in a nice way.” The new evidence seemed to rule out the helical model that Alex Rich had constructed. New photographs that Rich had obtained now made them “suspicious that the structure may be much closer to DNA than we would have guessed. . . . The whole picture is now very queer and paradoxical and so I have great hopes that the solution will not be trivial.” 78 By May the solution to this paradoxical picture had not yet appeared. Watson and Rich decided to announce in Nature, not the grand scheme that Watson had started to write up in February, but the technical achievement of having drawn RNA into fibers enabling them to take the first X-ray diffraction photographs of RNA that showed distinctive patterns.79 More interpretative was a paper titled

T HE R EPLICATION P ROBLEM

37

“Some Relations Between DNA and RNA” that they sent, via Linus Pauling, to the Proceedings of the National Academy of Sciences. In it they addressed themselves to the same questions that Watson had treated in his letter to Crick in February. In place of the unguarded exuberance with which Watson had proclaimed his answers to these questions to his scientific partner, however, was cautious recognition of a conglomeration of solved and unsolved problems. “About the functions of RNA,” they began, “we possess little definite information. It has been implicated in protein synthesis, but only indirectly. The really interesting thing about both nucleic acids is that we know very little about how they function chemically in a cell.” 80 Pointing to the known similarities between the two polymeric compounds, they acknowledged that “Up to now we have had success in understanding only one of these two structures.” Summarizing briefly the characteristics of the two-stranded helical structure of DNA, they went on: The most attractive feature of the two-stranded complementary helix is the fact that it suggests an answer to the question of how DNA can replicate itself exactly, a function it must possess if it is a genetic material. The complementary structure fits this requirement neatly if we make the assumption that one strand can serve as a template for the formation of its complement. We visualize, then, a mechanism involving initial separation of the two strands, with each of the separated strands serving as a template for its complement—the whole process occurring in zipper-like fashion. This method of replication is likely to be very exact, as the necessity for specific pairing is absolute, and misformed pairs will not fit into the structure.81 In view of the “formidable unresolved difficulties concerning the separation of the two strands, we can see that the mechanism that Watson could visualize was far from a definitive answer to the question of how DNA can replicate itself.” It was a statement of confidence that such a mechanism eventually could be found. Watson could not describe it literally, but only through the metaphor of a zipper. Whatever the detailed mechanism might be, it must conform to the fundamental feature of the two-stranded complementary helix, because, as he recalled in 1990, “it would be very unlikely that anything better [than base-pairing] would be forthcoming.” 82 In the spring of 1954, Watson was not immediately concerned to

38

T HE R EPLICATION P ROBLEM

describe the mechanism in detail. Knowing the general principles on which it must sometime be built was a sufficient support for him as he fixed his attention on the way in which DNA may control, “either directly or indirectly, the synthesis of specific proteins.” Enumerating several objections to the direct role, Watson wrote that it was more “plausible to suppose a connection between RNA and protein synthesis. Under such a scheme, DNA could control RNA, with RNA responsible for protein synthesis.” 83 In spite of his proposing to Crick that one of the two DNA chains is transformed to RNA, Watson did not specify in the paper how DNA can “control” RNA. “We shall not be able to check a structural relationship between RNA and protein synthesis or between RNA and DNA,” he wrote, “until we know the structure of RNA.” 84 At the time, RNA appeared to be more complex in several ways than DNA, including the possibility that its chains might be branched. “The analytical composition of the bases in RNA also appears more complex.” The paper presented a table of the ratios of adenine, uracil (which could be considered the functional equivalent of the thymine of DNA), guanine, and cytosine found by several other analysts. The ratios of adenine to uracil and of guanine to cytosine were close enough to 1:1 in all except the RNAs from plant viruses to support the claim Watson had made in his letter to Crick that they are complementary. In the plant viruses, however, the distribution of the various bases appeared “fairly random.” The possible explanation that there are two types of RNA conflicted, however, with Watson and Rich’s diffraction photographs, which showed that “RNA of all sources produces the same X-ray pattern. A simple interpretation of the analytical data does not appear possible.” 85 Undoubtedly this was a part of the picture that Watson had called “queer and paradoxical” in his letter to Delbru¨ck. The paper next compared an X-ray diffraction photograph that Rich and Watson had obtained from RNA with the two well-known photographs of DNA by Wilkins and by Franklin. Drawing attention to their similar features, Watson concluded, “The X-ray pattern therefore suggests a DNA-like structure for RNA. However, since the DNA model is based upon complementary base ratios which are not found in many RNA’s, this suggestion has many difficulties. It is possible that non-complementary side chains may arise from a complementary main structure, but proof of this awaits more direct chemical evidence of branches in RNA.” 86

T HE R EPLICATION P ROBLEM

39

The suggestion that the noncomplementary overall base ratios of plant virus RNA might derive from side chains illustrates the fluidity of Watson’s position as he grappled with the various difficulties he encountered in his search for the structure of RNA. Having failed to convince himself that the chains were unbranched, he now came to see the possibility of branches as a potential advantage, a means to resolve a paradox that had in the meantime arisen. In model-building, too, Watson and Rich could only report efforts that fell short of solutions: “We have been able to construct singlechain helical models for RNA in which the free ribose hydroxyl group is satisfactorily hydrogen-bonded to a negatively charged phosphate group. However, we have not been able to form satisfactory intramolecular hydrogen bonds between the bases, which, in this model, remain free to form external hydrogen bonds.” They ended their brief summary section with the prudent admission that “further chemical and crystallographic work is necessary before we can discover the relationship between the structure of RNA and the origin of protein specificity.” 87 On 1 June, Watson wrote Delbru¨ck, “Our work on RNA is at a standstill. We need a cute idea or a much better X-ray photograph and neither possibility seems in the air.” Noting that Gamow and others were doing much work on a “protein code,” Watson commented, “I do not think the problem can be solved in this way and that we shall have to know RNA structure first. It is more prosaic this way but I’m afraid nevertheless true.” After indicating his willingness to work with a student in the laboratory on bacterial genetics, he added, “I still feel that RNA is the most important problem for us to crack but it will probably come from inspiration and not from solid concentration.” 88 For now Watson had, in any case, to interrupt his work on RNA for two months, because he had agreed to be an instructor in the general physiology summer course at Woods Hole, beginning on 15 July. At the end of the first week in June he left Pasadena in his car for a crosscountry dash with the radioactive phage stocks he had prepared to use in the course.89 It is facile to judge Jim Watson’s unsuccessful search for the structure of RNA during his unhappy year at Caltech as a futile attempt to recapture the magic of what he and Crick had done the year before with DNA. It is easy to view his hope that a new idea as clever as complementary base pairs would come to him through inspiration rather than sustained concentration as a yearning for lightning to

40

T HE R EPLICATION P ROBLEM

strike twice. But Watson was not misguided to approach the problem in the way he did. For him their auspicious solution for the structure of DNA functioned as a paradigm, in the sense that Thomas S. Kuhn has used that term. According to Kuhn, scientists typically “solve puzzles by modeling them on previous puzzle-solutions.” 90 At first the ways in which the problems are similar and the ways in which they differ are often not apparent. There was ample motive to take up the structure of RNA as a problem strongly resembling the one that had just yielded so beautiful a solution. Although he may have appeared to some observers more erratic on his own than when his enthusiasms had been controlled by his conversations with Crick, the two papers Watson published during this period show that he was circumspect enough to control in public the brash inspirations to which he gave free rein in his private correspondence. John Kendrew wrote Watson in June that Maurice Wilkins “is very kindly disposed towards yourself though he still thinks you are at times carried away by the impetuosity of extreme youth!” 91 Impetuous though Watson undoubtedly was, there are no grounds to say that, on his own, he lacked critical scientific judgment.

III Writing to Watson in November 1953, Kendrew asked, “Have you reconciled Max to the DNA structure?” 92 Undoubtedly, Watson had not been able to overcome Delbru¨ck’s resistance to unwinding. At Pauling’s protein conference in September Delbru¨ck had made a fivedollar bet with Crick that the two strands would never separate.93 He did not change his mind that winter, despite Watson’s presence in his laboratory. At Go¨ttingen in the spring Delbru¨ck gave a seminar on the new genetics and the double helix, news of which had apparently not yet spread widely in Germany.94 Perhaps it was that occasion that stimulated him to concentrate his attention on the problem of how DNA might replicate. At any rate, in mid-May he sent to the Proceedings of the National Academy of Sciences a paper titled “On the Replication of Deoxyribonucleic Acid (DNA),” which set forth a cogent theoretical analysis of the problem. After summarizing the structure proposed by Watson and Crick and their conception of its replication in a “zipper-like fashion,” Delbru¨ck wrote:

T HE R EPLICATION P ROBLEM

41

The principal difficulty of this mechanism lies in the fact that the two chains are wound around each other in a large number of turns and that, therefore, the daughter-duplexes generated by the process just outlined are wound around each other with an equally large number of turns. There are three ways of separating the daughter duplexes: (a) by slipping them past each other longitudinally; (b) by unwinding the two duplexes from each other; (c) by breaks and reunions. We reject the first two possibilities as too inelegant to be efficient and propose to analyze the third possibility.95 Before proceeding with Delbru¨ck’s analysis, we may note that he made no attempt to revive the idea that the two strands might form a paranemic coil. Watson’s argument at Cold Spring Harbor must have reconciled him at least to the requirement that the two chains be wound around each other. It is also characteristic of Delbru¨ck’s style that inelegance and inefficiency were for him sufficient grounds to reject the possibility that either of the two mechanisms that did not require breaks and reunions could operate in nature. Taking up the third possibility, Delbru¨ck reasoned that If one tries to separate the two chains of a duplex by moving the two chains laterally in opposite directions, an interlock occurs for each turn of the helix, i.e., at each tenth link. Such an interlock can be resolved in two ways: (a) by breaking one of the chains, slipping the other chain through the gap, and rejoining the broken ends; (b) by breaking both chains at each half-turn and rejoining them crisscross. We reject both these mechanisms—the first one because it introduces an asymmetry between the two chains (only one of them being broken) which is contrary to the symmetry of the structure and the second one because it rejoins chains with opposite polarity, which is chemically not permissible. We conclude that it is not feasible to separate by breaks and rejoins the two chains of a single duplex. The situation is quite different, however, when one considers a duplex during the process of replication. Let us consider a duplex in which replication has proceeded synchronously along the two chains up to link n. We will call this point the “growth point.” If we now break both the old chains between links n and n ⫹ l, we may join the lower terminals of the breaks in a crisscross fashion, not to the upper terminals of the breaks but to the open ends of the new chains of equal polarity. The upper termi-

42

T HE R EPLICATION P ROBLEM

Fig. 1.2. Replication mechanism proposed by Max Delbru¨ck

nals of the breaks now become the open ends for the continuation of the replication process.96 Delbru¨ck illustrated his conception of the situation by several diagrams, including the lateral view of the process shown in figure 1.2. His abstract approach to the replication problem is evident in Delbru¨ck’s reasoning and in his diagrams. The structure of DNA is reduced to two linear strands—a “duplex.” The polarities due to the directionality of the 3′–5′ linkages of the deoxyribose residues in the polynucleotide chains are reduced to arrows. The complementary base pairs are reduced to links, denoted n and n ⫹ l. The style was indeed that of the physicist seeking the fundamental features of a biological mechanism that did not depend on detailed descriptions of the molecules involved. Abstract though Delbru¨ck’s scheme appeared to be, it entailed a consequence that was potentially testable:

T HE R EPLICATION P ROBLEM

43

Fig. 1.3. Distribution of labeled parental DNA chains according to replication mechanism of Delbru¨ck (with breaks) and alternative (without breaks)

It is an important implication of the proposed mechanism that the chains of the daughter-duplexes consist of alternating sections of parental and assimilated nucleotides, each section with an average length of five nucleotides. If a labeled duplex replicates repeatedly at the expense of an unlabeled pool, then, according to this model, the label will be statistically equally distributed to the daughter-duplexes at each successive replication. Without the breaks and reunions the distribution of label would occur only at the first replication. At each subsequent replication one daughterduplex would receive all of the label, the other none.97 Delbru¨ck illustrated these alternatives with the scheme shown in figure 1.3. Delbru¨ck did not indicate what kind of label might be attached experimentally to a DNA duplex to test these implications, but it is hardly coincidental that he referred in the sentence following the

44

T HE R EPLICATION P ROBLEM

quoted passage to experiments by Gunther Stent on “the mortality due to the decay of incorporated P 32 of phage infective centers at various stages of the reproductive center.” From the relation between the rate at which bacteriophage incorporating 32 P are killed and the inherent rate of the decay reaction 32 P → 32 S, Stent had inferred in 1953 that the “cause of death” must be the disruption of the DNA molecule by the disintegration of a 32P atom contained in its polynucleotide chain. That the “efficiency” of killing—that is, the ratio between the number of disintegrations and deaths—was only 1/12 or less Stent thought might be explained in part by the likelihood that only a fraction of the disintegrating 32P atoms cause a break in the polynucleotide chain, and in part, as he had suggested at the Cold Spring Harbor Symposium, because “if DNA is an intertwined, interlocked double strand, as proposed by Watson and Crick, it is not inconceivable that the molecule could sustain loss of an occasional phosphate link without being broken.” 98 Stent had made another point relevant to Delbru¨ck’s concern: We do not know to what extent the atomic identity of a parental molecule is preserved when it is being duplicated: do all the original atoms remain together in one structure and do, consequently, the atoms of the duplicate consist entirely of newcomers or are the parental atoms distributed over both structures at the end of the replication act? Experiments on the P 32 mortality during duplication offer a clear operational distinction between these alternatives: If the atoms of the parent structure remain together and the duplicate contains only material synthesized de novo, then mortality due to P 32 atoms incorporated into [the] parent molecule must come to a stop as soon as the first duplicate is finished. If, on the other hand, the parental atoms are equi-distributed among the two daughter structures, then P 32 mortality would continue after duplication with one half the ultimate rate of death of the original structure before duplication. The possible non-synchronism of duplication of different parts of the parental phage DNA unfortunately obscures the conclusions which might be drawn . . . concerning this question, the answer to which would be of capital importance to an understanding of the replication mechanism.99 Stent reported to Delbru¨ck in March 1954 that his experiments were “going very slowly” but that he had “a theory now to account for the factor of 1/12 of the efficiency of P32 death”: “The idea is based

T HE R EPLICATION P ROBLEM

45

Fig. 1.4. Interrupted chains of Schachman and Dekker, drawn by Gunther Stent

on the interrupted chains of Schachman and Dekker and simply considered to be the probability that the P32 decay occurs in a nucleotide at a spot not sufficiently close on one chain to a spontaneous break on the sister strand that the ‘point heat’ generated by the decay overcomes the resistance of any hydrogen bonds which might be still in the way and causes a complete rupture of the double helix” 100 (figure 1.4). The possibility that the polynucleotide chains of DNA molecules contain breaks was clearly pertinent to the replication question. Evidently Delbru¨ck did not welcome this news, however, because he wrote to Stent somewhat later asking for an “ ‘unprejudiced’ account of the interrupted DNA chains.” Along with a long letter he sent a manuscript of his replication paper.101 Stent replied on 11 May, reporting that “Howard Schachman presented his ideas at the National Academy meeting a couple of weeks ago and, rightly or wrongly, they were greeted with general approval by Crick, Todd, and other hot shots, as just the sort of thing that ought to exist.” Stent summarized the “pretty convincing” evidence, from titration data, heat degradation, and the action of DNAase, from which Schachman had concluded that the polynucleotide chains of purified DNA in solution are interrupted.102 Stent stressed that all of these data pertained only to solutions of purified DNA. “The only in vivo ‘evidence’ is the minimum P32 killing efficiency of 1/20 which I attribute speculatively to decay in apposition to spontaneous breaks, with the corollary that the DNA can support a large number of non-lethal breaks as those induced by radioactive decay.” This was essentially the theory that Stent had outlined to Delbru¨ck in his previous letter. “I wonder why you find the interrupted chains so unappealing?” he now added; “it would make the uncoiling problem much less formidable. As you guessed correctly, I was rather more excited by your ideas on replication than by the cofactor renaissance. The manuscript seems very ingenious to me, and I will write to you about it in more detail soon.” 103 It is not hard to discern why these interrupted chains did not ap-

46

T HE R EPLICATION P ROBLEM

peal to Delbru¨ck. To envision polynucleotide strands of molecular weight 10 4 unwinding would be no less inelegant to him than to imagine the process extending over the entire molecule. His scheme assumed that breaks and reunions were integral to the process of replication and that they took place at every turn of the helix, corresponding to a molecular weight of about 650. Preexisting breaks at longer intervals along the chains were of no help to him. Stent had further news concerning his own work that was not exactly favorable to Delbru¨ck’s scheme: “Right now, I’m snowed under with work, since I’m trying real hard to get an answer quickly to this distribution-of-atoms-during-duplication question. My preliminary idea at present is that they are not distributed, but that final experiment which was supposed to have clinched everything doesn’t seem to want to come out right.” 104 Stent mailed his letter on 11 May, and Delbru¨ck’s paper on the replication of DNA was communicated to the Proceedings of the National Academy of Sciences on 18 May. It is not certain, therefore, that Delbru¨ck read Stent’s comments before sending off his manuscript. In any case he did not allow the idea of interrupted chains or Stent’s preliminary idea that the atoms are not distributed during duplication to deflect him from proposing a mechanism that had no room for either phenomenon. In his paper he simply commented, “At present it does not seem possible to discuss the bearing of this implication [that is, of the distributions of atoms that would result from his mechanism] on the experiments of Stent. . . . These mortality experiments are complicated by the phenomena of multiplicity reactivation, i.e., an interaction between different duplexes, the nature of which is uncertain.” 105 If Delbru¨ck did have in mind Stent’s latest view on distribution, then his remark is a little puzzling; these complications would involve recombination events and would be more likely therefore to contribute to a misleading appearance of distribution than to a misleading appearance of nondistribution. For Delbru¨ck the formal elegance of his mechanism seems to have been an overriding consideration. Watson was not persuaded by Delbru¨ck’s proposed mechanism. Rather, it is a prime example of what he had in mind when he remarked that Delbru¨ck lacked “biological intuition.” 106 We may note that predictions about the shape of a future solution involve multiple levels of judgment. In this case the problem itself was a product of the compelling nature of a solution that had been around for only a

T HE R EPLICATION P ROBLEM

47

Fig. 1.5. James Watson at Cold Spring Harbor, 1953. Photo courtesy of the Archives, Cold Spring Harbor Laboratory.

short while. Base pairs set the parameters within which one could now discuss replication, but the structure that held the base pairs provided the obstacle to understanding just how that replication could take place without everything getting tangled. Watson judged that the separation problem was solved in principle and that one need not worry yet about how it could be solved in specific detail. Delbru¨ck judged this to be a pressing problem and sought to overcome what he

Fig. 1.6. Max Delbru¨ck at Cold Spring Harbor, 1953. Photo courtesy of the Archives, Cold Spring Harbor Laboratory.

48

T HE R EPLICATION P ROBLEM

had formerly seen as the insuperable difficulty of unwinding a plectonemic coil by postulating a process of breaks and reunions for which he had no direct evidence. Aesthetic criteria played a strong role in the judgments of both Watson and Delbru¨ck. Just as Delbru¨ck preferred his scheme because, even with all the breaks and reunions, it seemed to him more elegant and efficient than any unwinding scheme he could imagine, so Watson was content with a metaphorical zipper-like stand-in for a specific separation mechanism in part because the beauty of the double helix gave him confidence that eventually some satisfactory means to separate its strands would be discovered. On the other hand, judgments of beauty and elegance in science are not independent of evaluations of the strength of the supporting evidence. The beauty of the double helix does not inhere only in the imposing structure of a physical model, or in schematic drawings that enable one to visualize it, but also in the sense of beauty that the scientist can discern in the harmonious fit between the many coordinates of the atoms of the molecule and the parameters set by X-ray patterns and base pair ratios. Even in the spare formality of Delbru¨ck’s replication scheme the elegance one can discern lies in large part in the fit between his diagrams and the requirement of the Watson-Crick model that the two polynucleotide chains run in opposite directions. Even when connected to critical evidence, judgments involving beauty, elegance, or efficiency are inevitably also subjective. Delbru¨ck’s paper not only presented an elegant scheme but offered, in principle, a clear-cut, objective test that could decide between it and the alternative that he personally rejected. All that was required was the ingenuity to find out how one might label the strands of a DNA duplex in such a manner that one could actually trace the distribution of their “atoms” into daughter duplexes.

C HAPTER T WO

Meselson and Stahl

I In 1953 Matthew Meselson, a first-year graduate student in the Chemistry Division at Caltech, made an appointment to introduce himself to Max Delbru¨ck. Meselson stepped into Delbru¨ck’s office with trepidation, for he had been warned in advance that Delbru¨ck might be caustic. Despite this reputation, he was not unfriendly, but he was characteristically abrupt. As soon as Meselson sat down in front of his desk, Delbru¨ck asked what he thought about the two papers Crick and Watson had recently published in Nature. When Meselson confessed that he knew nothing about them, Delbru¨ck exclaimed, “What! The most important development in biology in ten years, and you’ve never heard of it?” Tossing reprints of the papers at Meselson, he said, “Go and read these papers, and don’t come back until you have.” 1 Meselson took these remarks not as a sign of rejection but as an invitation to return when he was prepared to talk. When he began to study the papers, however, he found that he could not understand the crystallographic arguments presented in the accompanying articles by Maurice Wilkins and Rosalind Franklin. Meselson had been taking the first-year course in X-ray crystallography, but he had only learned how to treat three-dimensional crystals using Fourier analysis. The course had not treated fibers. Wilkins applied Bessel functions to interpret the structure of fibers, and Meselson, who had not studied Bessel functions, had no foundation for grasping the nature of the evidence for the helix. As he tried to work it out, he even eavesdropped on a hallway conversation between Alex Rich and Jack Dunitz about X-ray diffraction, hoping he could pick up something to help him. He realized that it would be some time before he could continue the conversation with Delbru¨ck.2

50

M ESELSON AND S TAHL

The route that led Meselson to Caltech as a graduate student of Linus Pauling had been circuitous. The only child of Hyman and Ann Meselson, he had been born on 24 May 1930 in Denver, Colorado. When he was two years old, his family moved to Los Angeles. As a small boy, Matt became interested in chemistry. His uncle Morris gave him a little home-made chemistry set, with which Matt learned to do simple experiments. He particularly enjoyed depositing a layer of copper on an iron nail by immersing the nail into a copper sulfate solution. In high school, a tough but caring chemistry teacher named Elizabeth Butcher further strengthened his interest in the subject. In his home laboratory he progressed so far as to purify rare earths from ore given to him by Herbert McCoy, a retired chemistry professor who was a friend of Butcher. Matt dreamed that he would become an electrochemist who might find some way to create life.3 By the time he finished the first semester of the eleventh grade, in the fall of 1945, Meselson had completed all of the requirements for his high school diploma, except physical education. Learning that the University of Chicago would accept students who had not finished high school, he applied and was accepted for the fall term of 1946. When he arrived, intending still to become a chemist, he was shocked to learn that he could not take specialized chemistry courses but must go into the general program, in which the natural sciences were taught from original works such as Mendel’s Experiments on Plant Hybridization and Darwin’s Origin of Species. He was required also to take courses in classical literature. These works opened up a new world for him, and for a time his main interest was diverted to the classics. He also encountered, in this dawn of the postwar era, intense discussions about what the “new world was going to be.” 4 After three years at the University of Chicago, Meselson was “thoroughly unsettled as to what I would do with myself.” In June 1949 he traveled to Europe, hoping to accompany a young woman, in whom he had become interested, who was going to France on a tour. He stayed in Europe for more than six months. The rapid shift in American sentiment toward the Russians, so recently heroic allies who had driven the Nazis out of Eastern Europe, but now suddenly viewed as adversaries, stimulated his curiosity about that part of the world. He managed to travel to Budapest and Prague to experience what was happening in the newly formed Communist countries, and he learned to distrust equally the information given out by both the American government and the Communists. During the six months he stayed in

M ESELSON AND S TAHL

51

Europe, he realized that he could not be happy in a field such as political science or psychiatry, in which one could never be sure what was correct, and decided to return to the certainties of chemistry.5 In September 1950, Meselson enrolled at Caltech as a freshman, to make up the science courses he had missed at Chicago. Living at home, he met few other students and was unhappy with the rote approach to science he encountered, though he was much impressed by the chemistry course taught by Linus Pauling. The students seemed very young and unaware of the great world issues that had gripped Meselson in Chicago and in Europe. After one year at Caltech, he returned to the University of Chicago to complete his chemistry courses. Along the way he decided that he wanted eventually to study biology, but he did not know quite how. He had not liked the standard biology courses he had taken, because they seemed to involve mostly memorization. At Chicago he took physical, inorganic, and organic chemistry. He obtained a letter from a dean stating that, although the university did not award bachelor’s degrees in chemistry, he had completed work equivalent to such a degree.6 Learning that the University of California at Berkeley planned to introduce a graduate biophysics program, Meselson thought that it might allow him somehow to combine chemistry, physics, mathematics, and biology. When he arrived in the fall of 1952, however, he found that the advertised program did not yet exist. He instead took first-year graduate courses in the Physics Department, including a class in quantum mechanics taught by Freeman Dyson. His adviser at Berkeley suggested that he go to Caltech to study with Pauling. Doubting that Pauling would accept him, however, he applied to go back, once more, to the University of Chicago to enter the program in mathematical biophysics. Back home in Pasadena for the summer of 1953, Meselson went one day to a swimming party held by Peter Pauling, whom he had come to know during his year at Caltech, in the Pauling pool. Linus Pauling approached and asked what Meselson intended to do during the next year. When Meselson announced his plans for the program in mathematical biophysics at Chicago, Pauling declared, “But Matt, that’s a lot of baloney. Why don’t you come to Caltech and be my graduate student?” Meselson was too taken aback to say anything except, “All right, I will.” And he did.7 Entering Caltech in September 1953, Meselson took during his first year Pauling’s famous course on the nature of the chemical bond. He

52

M ESELSON AND S TAHL

absorbed more quantum mechanics as well, this time from the brilliant and colorful lectures of Richard Feynman. In preparation for his work with Pauling, he took J. Holmes Sturdevant’s course on crystal structure. Soon after his arrival he was ready also to begin a research project. To find a topic, he made an appointment with Pauling. Doing research in Pauling’s laboratory often meant working out the exact bond lengths and angles of the crystal structure of a given molecule by X-ray crystallography. Pauling suggested a mineral of tellurium but warned that there was some risk of acquiring the very offensive “tellurium breath.” Unattracted to this prospect, Meselson said that, because of his interest in biology, he would prefer to work on an organic molecule, whereupon Pauling assigned him the compound N,N′-dimethylmalonamide. The purpose of the project would be to test Pauling’s views about the peptide bond contained in this molecule. Pauling had already confirmed experimentally for other compounds the prediction, based on his resonance theory, that peptide bonds are planar, and had built that assumption into his models of protein structure.8 Meselson was placed under the immediate supervision of Raphael Pasternak, who would teach him how to grow and mount the crystals he would need to begin the analysis.9 Meselson thus became part of a well-organized team whose allotted task was to “make precise determination[s] of simpler substances closely related to proteins,” the ultimate objective being to obtain information that would be used “in a further attack on the proteins themselves.” 10 On 22 October, after reacting diethyl malonate with methylamine, Meselson produced his first dimethylmalonamide crystals. After three minutes of stirring, he allowed the solution to stand for an hour and evaporated it until cooling caused it to solidify. He dissolved the solid in benzene, filtered, evaporated, and cooled it, and obtained needlelike crystals that after recrystallizing in benzene “showed good faces and extinction.” He mounted one of the crystals and took Laue X-ray photographs for orientation. One month later he took eight series of photographs, rotating the crystal around each of its three axes.11 On 2 March 1954, Meselson attended a lecture given by the French biochemist Jacques Monod. Since 1947 Monod and his associates at the Institut Pasteur had been investigating the phenomenon known as “induced enzyme synthesis,” using as their prime example the enzyme β-galactosidase in the bacterium Escherichia coli. Up until the previous year Monod had developed the hypothesis that the inducer substances caused an enzymatically inactive protein otherwise very

M ESELSON AND S TAHL

53

similar to the galactosidase to be converted to the latter, but his most recent experiments with inhibitors of the inducers appeared to refute that view. A crucial question for him then became, Do the inducers or inhibitors “really affect the synthesis of the enzyme, rather than its state of activity?” 12 Monod came to the United States in February 1954 to deliver the Jessup Lectures at Columbia University. In a series of eight lectures, he reviewed broadly the field of inducible enzymes, covering the history of the subject as well as his own investigations of the previous fifteen years. Much of the time he focused on the synthesis of β-galactosidase, as well as its induction and inhibition. After summarizing the evidence that the induced enzyme is not formed by the conversion of preexisting protein, he presented in his last lecture various hypotheses of protein synthesis, including the most recent views regarding the role of DNA as the “carrier of the information required by a cell for the synthesis of a specific protein.” Nevertheless, he pointed out, the “evidence does not imply that DNA, to say nothing of RNA, is involved directly in enzyme synthesis, induced or not.” It is clear, he concluded, that “this problem will be taken out of the realm of speculation only when induced-enzyme synthesis will be obtained in vitro.” 13 After giving the lectures at Columbia, Monod toured other American universities, speaking at Urbana, Illinois, in Madison, Wisconsin, and in Berkeley before arriving at Caltech.14 His talk at Caltech, “Some Aspects of the Biosynthesis of a Bacterial Enzyme,” was undoubtedly a compressed version of his Jessup Lectures. He now made a suggestion, however, that had not been present in his extended lectures. If induction caused new protein to be synthesized, rather than activating protein already formed, he said, the osmotic pressure in the cells ought to increase, and this change could be measured experimentally.15 Sitting in the back of the auditorium, Meselson thought to himself that the experiment Monod proposed would not be very satisfactory, because changes in metabolism or something else, such as the permeability of the cell membranes, might complicate the measurements. The idea came to him that one could perhaps establish whether new protein is synthesized by growing bacteria in deuterium heavy water, then switching them to ordinary water at the same moment that one added the inducer for the enzyme. The difference in density between the protein that incorporated deuterium from the heavy medium and

54

M ESELSON AND S TAHL

the protein made with ordinary hydrogen ought to enable one to separate them by extracting the protein from the bacteria and centrifuging it. He imagined that one could prepare a sucrose or salt solution whose density was between that of the two forms of protein, so that the protein made with deuterium would sink, while the other protein rose to the top.16 These ideas occurred to Meselson in part because he had recently been thinking about deuterium in another context. While taking Pauling’s course on the chemical bond, he had become interested in calculating the strengths of hydrogen bonds formed with deuterium in place of ordinary hydrogen. Such a pencil-and-paper problem was not immediately relevant to his experimental idea, but these associations with the isotope running through his mind made him begin to wonder what deuterium might do to living organisms if it were incorporated into their molecules. After Monod’s lecture Meselson looked up several articles by Samuel Trelese, at Columbia University, showing that organisms such as bacteria and algae can grow in heavy water. Excited about his idea, he proposed to Pauling that he try it out. Pauling found the idea interesting but told him, “You ought to do your thesis research first.” 17 Meselson worked intensely on that research. He made more dimethylmalonamide crystals, examined their suitability for his purposes, and took more X-ray photographs. By the middle of May he was making the first integrated intensity estimates for the spots on the films taken across the three faces of the crystal. His immediate objective was to determine the dimensions of the unit cell of his monoclinic crystal.18 By this time Meselson had learned enough about Bessel functions to discuss the Watson-Crick structure for DNA with Max Delbru¨ck. Because Delbru¨ck was in Germany for the spring term of 1954, their meeting must not have taken place until early June. It was, in any event, a very satisfying conversation. Delbru¨ck gave Meselson a copy of his new paper laying out the alternative possible modes of replication of DNA. Meselson realized that the idea he had had during Jacques Monod’s lecture for examining the induction of galactosidase could also be applied to this problem. The label that Delbru¨ck wished to attach to a DNA duplex in order to trace the distribution of nucleotides from parent to daughter DNA could be a heavy isotope incorporated into the DNA while the organism grew on a medium containing the isotope. The medium could then be switched to the ordinary iso-

M ESELSON AND S TAHL

55

tope while the next generation of organisms was produced. Meselson made up his mind that he would, in some such way, determine whether DNA replicated as Watson and Crick’s model predicted or according to the scheme that Delbru¨ck proposed.19

II During the academic year 1953–54, Matt Meselson and Jim Watson both lived in the Atheneum, the elegant building at the east end of the compact Caltech campus that serves as a faculty club and a shortterm residence hall. Consequently they saw each other frequently, and each came to have a high regard for the other. Watson wrote nearly seven years later, “It was obvious from my first impressions made in 1953 that Matt could think and might develop into an outstanding scientist.” 20 When Watson accepted an offer in March 1954 to be an instructor in the famous General Physiology summer school course originated by Jaques Loeb in 1898 at the Marine Biological Laboratory at Woods Hole, he expected to have the assistance of one of the graduate students at Caltech. Since it appeared to him that Meselson had no clear plans for the summer, Watson invited him to come along also and help with the course. Meselson accepted.21 Given the responsibility to organize the lab work for one of the five weeks of the course, Watson decided to use the Hershey-Chase experiment as the basis for the laboratory exercises. Besides having characteristics suitable for didactic purposes, the Hershey-Chase experiment had been a prominent personal stimulus to Watson in the events that led him to his part in the construction of the double helix. In the well-known volume Phage and the Origins of Molecular Biology, published in 1966, Watson contributed an essay titled “Growing Up in the Phage Group.” 22 The phage group was an informal network of scientists working on the general problem of virus multiplication. Its origins were associated with the collaboration between Delbru¨ck and Salvador Luria that began in 1940. The third founding member was Alfred Hershey, who began to interact with Delbru¨ck and Luria three years later.23 Bacteriophage were especially suitable types of viruses for these studies, because the relative simplicity of their bacterial host cells made them more accessible to experimentation than the plant or animal viruses were. The style of reasoning and experimentation that distinguished the phage group was set largely by Delbru¨ck and Luria. Producing myr-

56

M ESELSON AND S TAHL

iad variations on a few basic types of experiments, such as one-step growth cycles, counts of single-cell burst plaques, and multiple and mixed infections with normal and irradiated phage, they sought through sophisticated mathematical analyses to characterize the kinetics and other quantifiable properties of processes whose chemical and morphological features were largely unknown to them. Drawing on concepts that the physicists among them had brought with them from their former field, phage biologists talked about “hits” on genetic material resulting from radioactive decay. They compared their burst-size counts with the statistical predictions of a Poisson or other form of random distribution curve. They devised mathematical models for replication, recombination, and other crucial processes in the bacteriophage cycle and represented them through simple schematic diagrams of a linear genetic material. Their papers contained many equations, tables of quantitative data, and graphs drawn on logarithmic scales to highlight the exponential character of key functional relationships. The special contribution that Hershey made to the work of the phage group between 1946 and 1951 was to show that when bacterial cells were infected simultaneously with two independent phage mutants, a process of recombination took place analogous to that shown thirty years earlier by the school of Thomas Hunt Morgan for mutants of the fruit fly Drosophila melanogaster.24 Because phage were not diploid organisms, the physical explanation that Morgan had given—that recombination resulted from exchanges between portions of homologous chromosomes, a process known as “crossing-over”—was not directly applicable. Nevertheless, Hershey concluded cautiously in 1951, the recombination tests showed that the arrangement of genes in phage is linear and that bacteriophage are, therefore, “amenable to the same kind of genetic analysis that has served to elucidate nuclear organization in other organisms.” 25 By this time the basic sequence of events involved in bacteriophage replication had also been worked out. One or more phage “particles” attached themselves to the surface of a bacterium, apparently, as recent electron microscope observations had revealed, by means of “tails” that extended from the main body of the virus. A portion only of the phage was injected into the bacterium, where it eluded direct detection during an “eclipse” period of variable duration. Subsequently the bacterium “lysed,” that is, it disintegrated, releasing phage particles into the medium in multiples of the original infecting phage

M ESELSON AND S TAHL

57

that varied according to conditions over which investigators had established a large degree of control. Despite the difficulty in visualizing what occurred during the interval in which whatever component of the phage had entered the bacterium remained there, the phage group had, by 1952, attained a clear interpretation of the meaning of these invisible processes. “Vegetative phage” somehow came to control the genetic mechanism of the bacterial cell so as to divert its metabolic machinery to reproduce the components of future phage particles, which were then “assembled” into the mature form capable of lysing the bacterium. It was during this period also that the recombination events demonstrated by Hershey took place.26 What was not known then, either for phage or for higher organisms, was the chemical nature of the molecules that took part in these genetic events. Phage appeared, however, to offer an especially favorable opportunity to approach that question, because they were believed to consist almost entirely of two classes of compounds, protein and deoxyribonucleic acid. Compared to the components of the nuclei of higher organisms this appeared to be an extremely simple situation. In 1951 Hershey and his research assistant, Martha Chase, began a series of investigations of the transfer of phosphorus and sulfur from parental to progeny phage.27 Among their goals was to ascertain whether the protein or the DNA of the phage, or both, entered the bacteria to control the reproductive process. In the paper in which Hershey and Chase described their “experiment,” they actually reported an extended series of experiments that together provided interlocking pieces of evidence for their general position that “one of the first steps in the growth of [bacteriophage] T2 is the release from its protein coat of the nucleic acid of the virus particle, after which the bulk of the sulfur-containing protein has no further function.” The technique they used was to label the phage with either a radioactive isotope of sulfur,35 S, or of phosphorus,32 P, by causing the phage to go through a reproductive cycle with bacteria grown in one or the other of the two isotopes. They knew from basic chemical knowledge of protein and DNA that the sulfur was incorporated into the former, the phosphorus into the latter. They conducted several experiments mainly to confirm the view deriving from the recent work of T. F. Anderson and R. M. Herriott showing that the virus particles consist of a protein coat enclosing the DNA. The experiments that became famous, however, were the ones in which they removed the

58

M ESELSON AND S TAHL

phage coats from infected bacteria by spinning a suspension of the latter in a Waring Blendor. In one set of experiments the infecting phage contained 35 S; in another set, otherwise similar, they contained 32 P. In samples of the suspension removed at intervals, they measured the proportion of the original 35 S or 32 P that could be found in the extracellular fluid after blending. About 80 percent of the 35 S appeared in this way, compared to less than 40 percent of the 32 P. “These facts show,” they concluded, “that the bulk of the phage sulfur remains at the cell surface during infection, and takes no part in the multiplication of intracellular phage. The bulk of the phage DNA, on the other hand, enters the cell soon after adsorption of phage to bacteria,” 28 and is there able to replicate normally. Further experiments in which they induced the infected bacteria to lyse and measured the distribution of the isotopes among the fractions of the lysate separated by centrifugation allowed them to conclude that the progeny phage were nearly free of 35 S label, whereas at least 30 percent of the phosphorus of the infecting phage was transferred to the progeny.29 From their overall results Hershey and Chase inferred that “sulfurcontaining protein has no function in phage multiplication, and the DNA has some function.” This most salient outcome of the HersheyChase experiment was, nevertheless, surrounded by qualifications. Questions remained, they noted, about whether the 20 percent of sulfur-containing protein not accounted for in the experiment entered the cell, whether “any sulfur-free phage material other than DNA enters the cell,” and whether “the transfer of phosphorus to the progeny is direct . . . or indirect.” Their experiments show clearly, they concluded cautiously, that “a physical separation of the phage T2 into genetic and non-genetic parts is possible . . . the chemical identification of the genetic part must wait, however, until some of the questions asked above have been answered.” 30 As Olby and Judson have both stressed, the impact of the HersheyChase experiment emanated less from their published account of a complex investigation and qualified conclusions than from letters that Hershey sent to a number of their colleagues outlining in simpler form the main thrust of their results.31 One of these letters had reached Watson a few days before he left Cambridge for Oxford in April 1952 to attend a symposium on the nature of virus multiplication. Watson was asked to present a paper on bacteriophage multiplication written by Salvador Luria, who had been denied a visa to attend the conference.

M ESELSON AND S TAHL

59

In this paper Luria proposed that the genetic material of the phage consisted of protein. After reading the paper, Watson added that Luria’s conjecture needed modification and proceeded to read sections of Hershey’s letter.32 Watson then added that he and Ole Maaløe had, during the previous summer, labeled phage protein with 14 C and nucleic acid with 32 P and obtained results that “show a close parallel to those of Hershey and Chase.” From the combined results of Hershey and Chase and of Maaløe and himself, he said, “it is tempting to conclude that the virus protein functions largely as a protective coat for the DNA and that the perpetuation of genetic specificity is largely or entirely a function of the DNA.” 33 According to Watson’s account of the event, few of those present at the Oxford meeting showed interest in what he had to say about Hershey’s results. Other eyewitness accounts differ on the degree of interest evident.34 Some of those to whom Hershey sent letters responded very positively. Delbru¨ck praised the experiments. Maaløe wrote back that it was a “very beautiful piece of work, and it gives all of us a lot to think about.” 35 Despite the admiration that the Hershey-Chase experiment aroused in those whose thinking was compatible with its message, it was immediately noticed that the result was not entirely conclusive. Their paper itself claimed only that the bulk of the phage sulfur remains at the cell surface, and the bulk of the phage DNA enters the cell, and it ended almost anticlimactically with the statement that the “protein probably has no function in the growth of intracellular phage. The DNA has some function. Further chemical inferences should not be drawn from the experiments presented.” 36 The fact that 20–25 percent of the phage 35 S and 21–35 percent of the phage phosphorus were unaccounted for in the Waring Blendor experiments left room for others, also, to question the result of the experiment. The historian Robert Olby judged their evidence “not all that convincing.” 37 Was the Hershey-Chase experiment a “beautiful piece of work,” as Maaløe pronounced it, or a flawed effort? Was it, as Watson retrospectively judged it, a “powerful new proof” that DNA is the genetic material,38 or merely a convenient new argument for Watson and a few other strategically placed individuals to support a position they already wished to take? Was it a crucial experiment, or merely a timely one? Historians and molecular biologists have come to judge the earlier experiments of Oswald Avery and his associates, in which they identified the bacterial “transforming factor” in 1944 as deoxyribonu-

60

M ESELSON AND S TAHL

cleic acid, to have provided at least as compelling evidence that the hereditary material is DNA. The greater immediate impact of the Hershey-Chase experiment is attributed chiefly to the attention it attracted through the network of the phage group within which Watson “grew up.” 39 In any case, his choice of the Hershey-Chase experiment for the course at Woods Hole in the summer of 1954 must have reflected the fact that in Watson’s mind it had already become a historic landmark. On the eve of his departure for Woods Hole on 1 June 1954, Watson reported to Delbru¨ck further reasons for anticipating that the experiment would be a pedagogical success: “We shall try to have the class do the Hershey-Chase expt. For the last two weeks, I have been working quite hard growing hot phage, etc. and find it as before amusing and pleasant. Over the weekend we repeated the blendor expt. with good results. I think it should also work at Woods Hole as the expt. is really so simple.” 40

III Like Watson, Meselson set out for Woods Hole by car, driving across the country in five days in his 1941 Chevrolet coupe and suffering several blow-outs along the way. Ed Furshspan, a graduate student in the Biology Division, accompanied him. Meselson went to Woods Hole as an instructor assisting Watson, but he did not actually participate in the course. What Watson wanted him to do there was to carry out some titration experiments on RNA that would help him work out its structure. In 1947 the British nucleic acid chemist Masson Gulland had shown that when deoxyribonucleic acid is titrated with acid or alkali, there is little ionization of the ionizable groups in the molecule between pH 5 and pH 11, but rapid ionization in the more extreme acid and alkali ranges. When he back-titrated the solutions, the resulting curve differed from the initial curve (see figure 2.1). The interpretation that Gulland and others gave these results was that the groups titrated in the acid range were the amino groups of the purine and pyrimidine bases, in the alkaline range the ENHECOE dissociations of guanine and thymine; that in the high alkaline and acid ranges such groups were rapidly exposed by the breakage of hydrogen bonds between them; and that consequently more such groups were available to be titrated during the back-titration. An orderly

M ESELSON AND S TAHL

61

Fig. 2.1. DNA titration curves produced by Masson Gulland

structure was apparently disrupted by breaking hydrogen bonds. When the Watson-Crick structure was proposed five years later, these conclusions were seen to be “in remarkable agreement” with it, assuming that the hydrogen bonds broken were those that hold the two strands of the double helix together. If it could now be shown that RNA exhibited a similar difference in its forward and backward titration curves, that would be welcome support for Watson’s effort to find a helical structure in the RNA molecule.41 One of the difficulties in such experiments was that the ease with which DNA and RNA could be unintentionally degraded by the processes with which one extracted them from cells often led to conflicting titration results. To obviate this problem, Watson had obtained from Alex Rich RNA prepared using the gentlest methods possible. Meselson was given space to perform the experiments in the basement

62

M ESELSON AND S TAHL

of the Loeb laboratory. First, however, he had to build his apparatus. It required a potentiometer, a glass electrode, a controlled temperature, and extensive calibrations, and it took him most of the summer to get it ready.42 To perform such experiments was, of course, not the sole or even the primary reason for a young scientist to be in Woods Hole. It was also an opportunity to come in contact with other students and leaders in the field. There was plenty of time to explore the beauties of Cape Cod. Meselson particularly enjoyed the high bluff at Nobska Point, a short walk from Woods Hole, with its overlooking lighthouse, its commanding view to the south across the water to Martha’s Vineyard, and the sweeping arc of beach below it to the west. While Meselson was standing one day with several others, including Jim Watson, Sidney Brenner, and Francis Crick, near a window on the third floor of the main MBL building, Watson pointed to someone across the street on the lawn. “That’s Frank Stahl,” he said, and added a complimentary remark about Stahl’s laboratory skills. If Stahl was that good, someone else responded, they should give him the problem of “doing the Hershey-Chase experiment by himself with both isotopes on the same afternoon.” Curious to meet the subject of this banter, Meselson went downstairs and found Stahl sitting against a tree, drinking a gin and tonic while he tried to work out a mathematical problem.43 The problem was to solve an integral equation expressing the probability that a given genetic marker located on a linear structure in a bacteriophage irradiated with ultraviolet light will replicate in a crossreactivation experiment. The probability sought was to be a function of the mean number of hits on the linear structure and of the mean number of breaks inflicted on the structure. When he realized that Meselson knew nothing about bacteriophage genetics, Stahl translated the problem into one about two flocks of birds landing on a clothesline with a clothespin on it. The question became, given the mean number of blue birds and black birds, and given that the birds alighted in random order, what were the chances that the nearest bird on either side of the clothespin is blue? If he knew little about phage, Meselson had a strong enough background in statistics and integral calculus that he could offer Stahl capable help with the problem.44 Franklin W. Stahl, a graduate student in biology at the University of Rochester, had come to Woods Hole in the summer of 1954 to take courses that were not available in his home institution. Born in Octo-

M ESELSON AND S TAHL

63

ber 1929, Stahl entered Harvard College in 1947. He majored in biology with the intention to enter medical school. For a variety of reasons, including the need to commute and to work part-time to support his schooling, it became obvious to him by the end of his second year that his grades were not good enough for this goal. He contemplated teaching biology in high school, but his mother urged him to try for an advanced degree. During his college years the subject of genetics excited him in spite of the dull way that it seemed to him to be taught at Harvard.45 Graduating from Harvard in 1951, Stahl applied to three graduate schools, and went to Rochester because it was the only one of the three that accepted him and offered financial support. When he arrived, in the fall of 1951, the chairman of the biology department, Donald R. Charles, served as his adviser. A classical geneticist, Charles deepened Stahl’s commitment to genetics. Because he was very ill with Hodgkin’s disease, however, Charles sent Stahl away for the summer of 1952 to Cold Spring Harbor. There he took a course on bacteriophage taught by A. H. (Gus) Doermann. It quickly became clear to Stahl that the future of genetics must be in the study of phage or bacterial genetics, because only these organisms were simple enough to enable one to ask fundamental questions at the molecular level.46 When he returned to Rochester and said that he wanted to study phage or bacterial genetics, Charles told him that these subjects were faddish and would waste his time. Stahl learned basic cytological techniques necessary to study Drosophila chromosomes, but he considered the genetics of the venerable fruit fly to be hopelessly complicated. He applied to work with Luria at Indiana or with Joshua Lederberg in Wisconsin but was relieved of the necessity to do so when a new department chairman at Rochester, Kenneth Cooper, decided to bring into the department someone whose specialty was phage genetics. Acting partly on Stahl’s recommendation, Cooper hired Gus Doermann, who was at the time a research scientist in the Biology Division of the Oak Ridge National Laboratory in Tennessee.47 In June 1953, Stahl attended the Cold Spring Harbor Symposium at which Watson gave the first presentation in the United States of the new DNA structure. The event was exciting to him. Afterward he went on to Oak Ridge so that he could begin working under the supervision of Doermann even before the latter was due to arrive in Rochester.48 At the time Stahl joined him, Doermann had acquired a reputation

64

M ESELSON AND S TAHL

within the phage group as a brilliant experimentalist. In 1951 he had published a paper titled “The Intracellular Growth of Bacteriophages” that quickly became a classic in the field. It began, “Direct studies of bacteriophage reproduction have been handicapped by the fact that the cell wall of the infected bacterium presents a closed door to the investigator in the period between infection and lysis. As a result it was impossible to demonstrate the presence of intracellular particles during this so called latent period. . . . This barrier has now been penetrated.” 49 Doermann penetrated the barrier by devising two methods to induce the bacteria to lyse prematurely, releasing the intracellular particles so that they could be counted and analyzed. He also extended the study of recombination in bacteriophage begun by Hershey. In 1951 Hershey had represented the recombination frequencies for seven mutants of the bacteriophage type known as T4 in linear order in three linkage groups. The following year Doermann added six more r (⫽ rapid lysis) mutants and discovered a new class of mutants that produced turbid plaques (designated “tu”). The two linkage groups into which he fit these mutants Doermann thought compatible with Hershey’s linkage groups. By this time it was clear that recombinant phage mutant ratios could be used to construct genetic maps analogous to those of classical genetics.50 When Stahl joined Doermann’s group, which now included Hershey’s collaborator Martha Chase, Doermann was hoping to add “further detail to the picture of replication and recombination” with experiments using irradiated phage. During the 1940s Luria had discovered that when two or more phage particles inactivated by lethal doses of ultraviolet light infect a bacterium simultaneously, they can cooperatively produce active phage progeny. The phenomenon was called “multiplicity reactivation.” The explanation that Luria gave, but later abandoned, was that genetic recombination between undamaged portions of the irradiated phages allowed intact progeny particles to form.51 Reviving this idea, Doermann infected bacteria with irradiated wild type T4 phage, together with unirradiated mutant “carrier” phage differing from the wild type in three genetic markers. The results he obtained were compatible with the “hypothesis that ultraviolet damage is located in the genetic structure and that genetic recombination rescues loci from these damages.” 52 After Doermann and his group moved to Rochester, Stahl began a

M ESELSON AND S TAHL

65

project to extend this method of marker “rescue” to phage inactivated by different means. In 1951 Alfred Hershey had shown that 32 P incorporated into phage gradually destroys the phage as its radioactive decay causes the phage DNA to disintegrate. The phenomenon became known as “phage suicide.” As noted in Chapter 1, Gunther Stent adopted this 32 P incorporation method for an extended series of investigations of phage reproduction. In 1953 Stent found, as Doermann did for irradiated phage, that phage inactivated by 32 P can still contribute genetic markers to progeny resulting from a mixed infection. Stahl planned to study further the effects of the disintegration of 32 P on the genetic structure of phage. In a symposium at Oak Ridge in April 1954 at which Doermann presented the results of the experiments on irradiated phage that he and Chase had conducted, Stahl commented during the discussion of their paper that “the likelihood that a disintegration prevents both of two markers from appearing in the progeny formed in a single bacterium has been found to bear an inverse relation to linkage distance. It is hoped that these experiments will permit an estimate of genetic distances in molecular dimensions.” 53 It was for the purpose of analyzing data both from Doermann’s experiments on reactivation of markers from irradiated phage and from his analogous experiments with 32 P phage that Stahl needed the equation he was seeking to formulate when Matt Meselson walked out of the MBL building at Woods Hole to meet him. Stahl had, in fact, already arrived at an equation that gave a good approximate fit to the data from Doermann’s experiments when he assumed a relatively large number of crossovers per bacteriophage mating, but his solution seemed intuitive and clumsy to him. After their conversation, Meselson worked on the problem for a few days and came up with a rigorous solution, valid for all values of the parameters. The resulting integral equation was long and complicated, but when the number of crossovers or the sum of crossovers and hits was large, his equation reduced to Stahl’s simple approximation. This performance convinced Stahl that Meselson was extremely bright.54 Meselson was probably convinced, nearly as quickly, that Stahl’s experimental skills with phage and bacteria could help him to implement his idea for testing the nature of DNA replication by density centrifugation. After they had got to know each other for a little while, he raised the possibility that they might carry out experiments together on this problem.55 In light of Stahl’s experience with the reproduction of bacterio-

66

M ESELSON AND S TAHL

phage it is not surprising that they began to talk about using phage DNA as the vehicle for the experiments. In keeping with Meselson’s original idea, they proposed to employ deuterium as the heavy isotope. They contemplated growing bacteria in a deuterium medium, then infect them with phage, hoping that the replicating phage would incorporate the deuterium into the copies of the original DNA produced in the host cells, then be released as “heavy” phage particles. With these progeny they would subsequently infect bacteria grown in an ordinary medium. The next generation of progeny they would centrifuge in a salt solution chosen to have a density intermediate between the density of phage containing deuterium and phage containing only ordinary hydrogen. They expected that the heavy phage would sink to the bottom and be recoverable from the sediment that collected there, whereas the light phage would float to the top. They could then collect this top and bottom material and do plaque assays to determine the numbers of heavy and light phage. Additional centrifugation in media of appropriate densities could be used to resolve further the pattern of distribution of parental phage DNA into progeny DNA.56 Because Stahl already intended to go to Caltech for postdoctoral work, as almost everyone involved in the study of phage genetics did if he could, he and Meselson could anticipate that they would eventually be able to conduct the experiments together.57 As Meselson recalls his early association with Stahl, the two thought so much alike that, although the initial idea was clearly his, its development was a joint process and each learned from the other. In Stahl’s view, Meselson’s potent mind so dominated the conversations that he contributed little.58 These are undoubtedly perceptions rooted in each of their personalities, which partly color the way each remembers their interactions but which also undoubtedly impinged on the nature of the interactions themselves. Meselson was buoyant, quick-minded, and optimistic, good at many things and encouraged by supportive parents to believe that he could succeed at almost anything he tried. Stahl was less self-assured. He had already encountered setbacks and failures. In spite of his reputation as a laboratory whiz, he was apprehensive whenever he went into a laboratory that he would not execute his experimental operations properly.59 More deliberate in speech and manner than Meselson, he could readily perceive conversations between himself and Meselson as dominated by Mesel-

M ESELSON AND S TAHL

67

son’s views, whereas Meselson perceived the same conversations as equal exchanges of views. Their plan was obviously derived from Meselson’s general vision that one could test the alternative theories of DNA replication by incorporating a heavy isotope into the DNA of an organism during its growth. The methods of incorporation and of density centrifugation to be applied remained, however, vaguely delineated. Neither Meselson nor Stahl had direct experience with deuterium or with the use of the ultracentrifuge that would be required. The most concrete aspect of the plan as they discussed it was the adaptation to their purpose of the one-step growth experiment that had become a basic fixture of phage reproduction investigations. Here Stahl’s laboratory experience in the field was crucial to assessing the feasibility of the idea and translating Meselson’s broad goal into a structured idea for a future experimental protocol.60 Meselson and Stahl both had ample opportunity in the relaxed atmosphere around Woods Hole to discuss scientific and other problems with Jim Watson. Finding Watson very approachable, Stahl did some laboratory work for him in addition to the course exercises.61 Meselson spent some of his leisure hours speeding along the nearby winding roads with Watson in Watson’s MG roadster. Naturally Meselson asked him for his opinion of the plan to study DNA replication by density centrifugation. Watson advised him not to attempt it at Caltech, where there was only one cumbersome, home-made analytical centrifuge. He should wait until he had finished his Ph.D., then go to Sweden, where the ultracentrifuge had been invented by The Svedberg, and where the best ultracentrifuge equipment was still available.62 Although Meselson apparently did not notice it, Watson intended these remarks to be flippant. Even though he had stated in April that Delbru¨ck’s proposal that DNA untwists with multiple breaks provided an “eventually testable prediction,” 63 he obviously did not think the time was at hand when this, or the prediction of the Watson-Crick model itself, could be tested. Meselson’s scheme appeared to him to be impossible, and he really meant that Meselson ought to go to Sweden to try it because of the beauty of Swedish women. While he was at Woods Hole, Meselson approached Jan Drake, who had come there after graduating from Yale, to ask his advice about a woman problem. When he learned that Drake would be coming to

68

M ESELSON AND S TAHL

Caltech in the fall, he also talked with him about his titration experiments on RNA. Drake and Stahl did not discuss science, because Stahl’s mathematical concerns seemed too arcane for Drake, who was working in embryology, but they collaborated on an entrepreneurial venture. Together they mixed and sold gin and tonics under a tree, to pay for the liquor they themselves consumed. Meselson, Stahl, and Drake also talked about the possibility of finding living quarters together in Pasadena when Stahl arrived there.64 At the end of the six-week summer session Stahl went back to Rochester, where he received sharp criticism for choosing Watson’s laboratory course in place of the animal physiology laboratory he had been sent to attend. Undaunted, he continued to test the fit between the equation that he had worked out with Meselson and data generated by Doermann’s experiments with the “rescue” of markers from irradiated phage. He hoped that he could work out a theoretical foundation for these results. The good agreement between the observed appearance in progeny phage of genetic markers of the T4 linkage group introduced into bacteria from irradiated phage, and the theoretical probability predicted by their equation, encouraged Stahl to develop a model to account quantitatively for the cross-reactivation results.65 The model assumed that a UV photon may strike the linear structure along which genetic markers are located at any point x. A unit that regulates an observable genetic trait cannot replicate while it is attached to a hit on the structure. If a break occurs on the linear structure somewhere between the marker and the nearest hit on either side of it, and there is a crossover at this point during a recombination event with another linear structure, then the genetic marker in question may become attached to an intact structure, where it can take part in replications. Stahl wrote a short paper titled “A Formulation of the Theory by Doerman, Chase and Stahl of the Cross Reactivation of Markers of UV’d Phage,” naming himself and “Mathew Messelson” as co-authors. The paper presented the integral equations that Meselson had derived for “the likelihood that a marker will appear in a burst as a function of the mean number of radiation hits on the genetic structure.” Using this function, Stahl plotted the “total likelihood of survival” against the mean number of hits on the structure for the case in which the mean number of crossovers between the irradiated and carrier phage was forty. The resulting curve fit closely the data from one of Doermann’s cross-reactivation experiments. He sent a mimeographed copy

M ESELSON AND S TAHL

69

of the manuscript to Meselson, who was still at Woods Hole continuing his experiments on RNA.66 By mid-August, Meselson was finally ready to carry out the titration experiment on RNA for Watson. When he did so, he found no difference between the forward and backward titration curves. While relaxing in the sun at Nobska Point on 17 August, he wrote Stahl about his first result. The experiment indicates, he said, that “there is no extensive stable H-bonding between bases as in DNA.” This conclusion depended, however, “on whether or not the RNA used was degraded. I’ll try again with better RNA when it’s available.” Meselson also responded, belatedly, to the manuscript he had received from Stahl: “This long delay has been caused by the shock of receiving a paper which one has co-authored and yet has neither read nor understood (except at the aviary level).” He awaited the reprint of the paper by Doermann, Chase, and Stahl, hoping it would give him the insight into the literal application of the equation that would “enable me to comment not too stupidly on the Stahl-Meselson fantasy.” Meanwhile, he was less impressed than Stahl was with the fit between the data and the equation, because it depended on the choice of the value 40 for an adjustable parameter. He thought that if Stahl were to obtain data for another marker and then found that the same value for the mean number of crossovers still gave the best fit for such a marker, “that would be excellent support for the model.” Meselson also corrected the spelling of his name on the manuscript.67 Meselson left Woods Hole shortly afterward to attend a phage meeting at Cold Spring Harbor, where Stahl rejoined him. By then he had come to understand their “joint” paper well enough to agree to circulate it to the participants. Only a few people there seemed interested in what appeared to be a very specialized problem, and one of them thought that forty crossovers was a very large number. No one, however, commented generally unfavorably about their effort.68 Before the meeting ended, hurricane Carol struck the east coast. When Meselson returned to Woods Hole he discovered that water had flooded the basement of the Loeb laboratory, ruining the equipment he had set up for his RNA titration experiments. There was no possibility to continue them. The one experiment he had managed to perform nevertheless seemed to show that RNA could not be thought of as double-stranded. The outcome could not have supported Watson’s pursuit of his ideas about RNA as a double helix. With nothing further to do in the east, Meselson packed his Chev-

70

M ESELSON AND S TAHL

rolet coupe and he and Ed Furshspan set out for Pasadena. They gave a ride to a German scientist whose evident Nazi sympathies and antiSemitic remarks made the long journey thoroughly unpleasant.

IV During the academic year 1954–55, neither Meselson nor Stahl was in a position to pursue the ideas for DNA replication experiments that they had discussed with such enthusiasm at Woods Hole. Each of them had first to carry through the research problem he had already begun for his Ph.D. thesis. In Rochester, Stahl pursued his study of genetic markers “rescued” in cross-reactivation experiments from phage inactivated by 32 P decay. For his experimental material he chose phage T4, for which his mentor Doermann had situated the known genetic markers along three linkage groups. His intention was to establish the relation between the linear distance separating two genetic markers on one of Doermann’s linkage groups and the likelihood that the disintegration of a 32 P atom in a phage particle will prevent both of the markers from appearing in progeny resulting from recombination events between the phage and a nonradioactive phage infecting the same bacterium. To do so, he prepared radioactive stocks of wild type T4 by infecting with them E. coli bacteria grown in a medium containing inorganic 32 P. The nonradioactive phages were mutants each containing three markers detectable by their effects on the morphology of the plaques formed by their progeny in a bacterial culture. (These were the common mutant types: r ⫽ rapid lysis, m ⫽ minute plaque, and tu ⫽ turbid plaque. Each of these could arise from several independent mutations, shown by recombination experiments to be located at different sites on the genetic map of the phage.) In order to compare the situation when the nonradioactive phage contained three linked markers with that when the markers were unlinked, he prepared one mutant stock in which one of these markers was located on each of Doermann’s three linkage groups (r 48 m 42 tu 42b ), and another in which all three markers (r 51 m 42 tu 41) were located along linkage group II. Underlying the design as well as the interpretation of the experiments was the working hypothesis that genetic specificity in bacteriophage was carried in DNA and that the lethal effects of 32 P decay were, therefore, due to damage to the genetic material that controlled the replication and recombination events between the time that

M ESELSON AND S TAHL

71

phage entered the bacteria and that progeny were released from it. For Stahl these expectations rested heavily on the result of the HersheyChase experiment.69 The general procedure that Stahl followed in these experiments was to determine what fraction of the radioactive stock infecting the bacteria contributed all three genetic markers to progeny phage, compared to the fractions able to contribute only one or two of the markers. He began with the assumption that, as in the recent experiments Gunther Stent had performed on phage T2 using two markers, the three markers would be eliminated by 32 P decay independently of one another. As he subjected his accumulating raw data to statistical analysis, however, Stahl found the situation to be more complicated. Even in the case of the unlinked markers, the results deviated so far from that predicted on the basis of independent elimination that he began to think that the loss of two or more markers might somehow be correlated with a single 32 P decay. He reexamined some of Doermann’s data on irradiated T4 from that point of view, and found that it could not distinguish between independent and correlated elimination. Doermann, whose experimental skills were not matched by mathematical expertise, was surprised but readily accepted Stahl’s conclusion. Because Stahl’s original plan to estimate genetic distances between markers from the patterns of their elimination had rested on the assumption of independent elimination, he gave up this objective and followed up instead the implications that the statistical correlations he was observing might hold for theories about the process of crossreactivation.70 The theory that Stahl, with Meselson’s help, had worked out during the summer to explain Doermann’s cross-reactivation experiments had incorporated the assumption that genetic markers are independently eliminated by ultraviolet radiation. Assuming that independent elimination still held for irradiation, and that its mechanism differed from that through which 32 P decay inactivated phage, Stahl continued to work on the theory. By making an approximation concerning the term representing the fractional distance of a marker along the linear genetic structure, he was able to reduce to a simpler form the long equation Meselson had derived for the total probability that an irradiated phage will be able to replicate. He then sent Meselson a letter with a plot of the theoretical curve derived from the approximate equation. He asked for Meselson’s opinion on whether it would be feasible to introduce into the theory a factor for the multiplication of

72

M ESELSON AND S TAHL

recombinant phages within the bacteria and also to extend the theory to include linked markers.71 When Meselson received Stahl’s letter, he was too busy with his X-ray crystallography to answer immediately. Soon after returning to Caltech he had given a seminar to the phage group based on the paper that Stahl had written. It was Meselson’s first general introduction to that group, and he faced it with some apprehension. He knew the integrals well enough, but his knowledge of phage was so thin that probing questions from these experts would quickly have revealed his ignorance of their field. None were asked during the session. Shortly afterward, a man dressed in sandals, whom he did not at first recognize, came late at night into the lab where Meselson was adjusting a model of N,N′-dimethylmalonamide. The intruder turned out to be Jean Weigle, a distinguished member of the phage group. Weigle had come ostensibly to discuss the seminar but actually to learn more about Meselson. The two engaged in an intense conversation which initiated a long friendship.72 During October and November, Meselson continued to grow N,N′dimethylmalonamide crystals, mount them, and take X-ray photographs. By December he had obtained from the diffraction patterns good values for the dimensions of the unit cell and the monoclinic angle of the crystal. He was ready to begin guessing what the structure might be. The data gathered from the patterns and intensities of the spots on the X-ray films could not directly yield that information. From them one could get a cluster of vectors connecting atoms in the crystal but could not establish their origins. It then became necessary to work backward from a possible molecular structure to the vectors that it should produce, and see if the measured vectors could be made to coincide with them. As Pauling put the problem, “using his intuition and ingenuity, the chemist devises possible [atomic] arrangements that conform to the constraints imposed by cell size, symmetry, and cell content.” 73 To produce such a candidate structure, Meselson first summarized the bond distances and angles, taken from the publications of Pauling and Corey, for each of the atomic pairs contained in the molecule. In addition, he followed a few rules from The Nature of the Chemical Bond and Pauling’s assumption that amide groups, such as the two in this molecule, were planar.74 During December, Meselson worked out the parameters for trial structure I. “Instead of using packing alone to obtain the trial structure,” he recorded in his laboratory notebook, “a set of Bragg Lipson

M ESELSON AND S TAHL

73

Fig. 2.2. Diagram for test of trial structure of N,N′-dimethylmalonamide, from Matthew Meselson’s X-ray diffraction notebook

˚ were drawn for reflections 00 16, 00 18, maps to scale 10cm. ⫽ 1 A 208, 408, 802, and 208. A wire model of the group CECENECH 3 储 O was built to the same scale and was mounted on an adjustable stand in the arrangement shown” (figure 2.2). By shining the spotlight through the three-dimensional scale model onto a flat surface and rotating the model in various directions, he tested whether his trial structure gave one-dimensional projections consistent with the projections derived from the X-ray photographs of his crystal. He discovered, however, that the parameters he had worked out “do not give good agreement between calculated structure factors and observation.” On 18 December he took thirteen more Xray photographs.75 By the time Meselson finally answered Stahl’s letter, on 10 January 1955, he had to ask whether it was “too late for a reconciliation!?” To explain his slow reply, he reported that “candidacy exams are coming up soon and even the ‘vacation’ was entirely occupied with pushing my slow moving thesis research . . . except for three days in San Francisco and Berkeley at A.A.A.S meeting time.” In the meantime, he had learned about a potential experimental threat to their theoretical analysis of cross-reactivation. Robert DeMars, a research fellow in Delbru¨ck’s group who had also been doing cross-reactivation experiments with ultraviolet irradiated phage T2H, had obtained the unexpected result that with pairs of linked markers the frequency of the recombinant form was “a linear function of the UV dose.” 76 That result indicated to Meselson that a quantity that he and Stahl had assumed to be constant in their equation (the number of crossovers) was instead a function of dose. “What’s up!” he wrote to Stahl. “Am I right to be

74

M ESELSON AND S TAHL

perturbed at the alleged result . . . ?” Seeing no explanation that would obviate the problem, he asked whether they had “better prepare to go underground.” Moreover, he feared that to introduce the multiplication of recombinant phage into their equation as Stahl proposed would make it intractable.77 His qualms about the status of their theory did not dissuade Meselson, however, from including it as one of the three “propositions” he was required to submit and to defend orally on his examination for candidacy to the Ph.D. degree.78 Meselson took his examination on 7 March. His examiners were his thesis supervisor, Pasternak, and two other members of the Chemistry Division. The other two propositions he had prepared to defend dealt with a quantum mechanical treatment of polymeric sulfur and with Pauling’s theory of the formation and structure of antibodies.79 Neither Meselson nor Stahl pursued their theoretical model further. For Stahl it had been primarily an effort to “help Doermann get a good theoretical foundation for analyzing the results he was getting.” When Stahl looked back on it in 1988, he remarked, “I think to this day the concepts generated here are the best single explanation for those experiments.” For Meselson, Stahl recollected, the project was “strictly a hobby. . . . It caught his interest in algebra, and I think he was rather proud of himself that he had solved it; [but] it was never more than a peripheral interest of his.” Until I showed him Meselson’s examination prospectus in 1988, Stahl had been unaware that Meselson had put it in as part of his scientific activities. “That means,” Stahl reflected, “that I have had some influence on his career, too.” 80 For both men, these interactions also served another, perhaps unconscious purpose. They were a kind of rehearsal for the larger collaboration that they hoped to begin when they could come together in Pasadena.

C HAPTER T HREE

Twists and Turns

I While Meselson and Stahl deferred their mutual interest in the DNA replication problem to attend to more immediate tasks, others took up the challenge posed by Max Delbru¨ck’s provocative paper on the subject. Responses appeared on two widely divergent levels. On an abstract plane that appealed particularly to physicists there appeared topographical models suggesting alternatives to Delbru¨ck’s scheme for resolving the unwinding dilemma. On an experimental plane some of the members of the phage network sought to trace the patterns of distribution of parental DNA molecules into progeny DNA. Somewhere between these levels, Jim Watson himself took Delbru¨ck’s objections seriously enough to think about alternative structures for DNA that might obviate the unwinding problem altogether. In December 1954, John R. Platt, of the Physics Department at the University of Chicago, sent to PNAS a short paper titled “Possible Separation of Intertwined Nucleic Acid Chains by Transfer-Twist.” The paper contained no data but simply presented a novel idea for solving the replication problem: “To separate the strands of a twisted rope it is not necessary to unravel them from the end. It is mechanically simpler and energetically easier to make a ‘transfer-twist,’ pulling the strands apart in the middle and letting each strand twist about itself. This suggests a method of separating two twisted intertwined helical molecules without expending excessive energy. It might be an alternative to the method suggested by Delbru¨ck.” 1 The core of Platt’s alternative suggestion was a diagram illustrating how two strands could be intertwined in such a fashion that outward pulls at the midpoint can cause the strands to rotate while simultaneously unwinding. Platt gave only the most general argument for his

76

T WISTS AND T URNS

view that the “motive force to drive the twisting . . . might come from a change in the environment” favoring either hydration or dehydration of the chains.2 The irrepressible George Gamow also had a try at resolving the replication problem through abstract models. As noted in passing in Chapter 2, Gamow had already in 1954 stirred up the Watson-Crick enthusiasts by proposing a topological scheme through which the “four digital numbers” represented by the bases in the DNA structure may be “translated” into the “20-letter alphabet comprised of the amino acids contained in proteins,” thereby defining the famous “coding-problem.” 3 When Delbru¨ck’s article on DNA replication appeared, Gamow was stimulated to further thoughts about the spatial configuration of the DNA double helix. The result of his meditation was a “simpler more natural topological possibility of achieving the same result” that Delbru¨ck’s scheme did, without requiring the breaks in the chains: “In fact, a clean separation of two helices resulting from the division can take place without any breaks if, prior to the division, the long helical molecule is coiled into a spiral possessing the same repetition period as the original helix.” Gamow illustrated his idea with two sketches, one showing a “ribbon” representing the two strands simply wound around a cylinder, the second showing the twists due to the original helical structure of the ribbon compensated by the opposite winding due to the coiling. “The separation of the strands,” he proclaimed, thus “presents no difficulty.” Gamow offered the “hypothesis . . . that the chromosomes, which become visible as comparatively thick rods in preparation for cell division, actually possess such a coiled helical structure.” 4 Gamow sent Delbru¨ck a copy of his manuscript with the salutation “Another twist to the same problem. Geo.” 5 Delbru¨ck must have admonished Gamow that his hypothesis was only another version of the paranemic coil that he had originally proposed but that Watson and Crick had ruled out as incompatible with the X-ray diffraction evidence.6 Nevertheless, Gamow published his paper in PNAS in 1955, two years after Watson and Crick had rejected such arguments.7 Watson remained at Caltech as a senior research fellow during the academic year 1954–55. Although he was less unhappy than in his first year there, he still felt that he had no one to talk to and that there was no real social life there.8 Now, however, he had good prospects for leaving that situation behind, because the Biology Department at

T WISTS AND T URNS

77

Harvard University showed interest in bringing him to the American Cambridge. In January Watson traveled to Harvard to give a talk. Although “his presentation . . . was somewhat disappointing to those who sat in the rear of the room” and “could hear him only with great difficulty,” the general recognition of his “brilliance and of the importance of his work” led to an offer in March to join the staff as an assistant professor.9 Delbru¨ck hoped that Watson might stay at Caltech but realized that he was unhappy there, and was not surprised when Watson accepted the job. Not the least of its attractions for him were the women at Radcliffe.10 The appointment would begin only in the fall of 1956, however, because he had also obtained a fellowship to return to Cambridge, England, for the year 1955–56. There he was certain he would have someone with whom to talk. Watson continued during the year 1954–55 to try to solve the problem of the structure of RNA. He did not devote himself exclusively to RNA, however. The attention that Delbru¨ck’s paper on the replication problem was drawing provoked Watson sufficiently to revive his interest in DNA. He had the idea that DNA might exist in more than one form and that when it replicated it might assume a structure that made unwinding unnecessary. That insight led him to try to build models in the form of ribbons: structures that in 1953 Watson and Crick had found unattractive. Perhaps to his own surprise, he was able to construct in this form a four-stranded DNA, a structure that lent itself also to an explanation for the puzzling observations he had made in 1950 that half of phage DNA is lost in the transfer from parental to progeny phage. Moreover, the new model appealed to him for the same reason that the double helix had: “It was a beautiful structure.” When he went east for his interview at Harvard, he carried with him his current excitement over the new development.11 In April Delbru¨ck conveyed the news of Watson’s latest venture vividly in a letter to Carsten Bresch: Jim Watson has come up with a new brainstorm concerning the replication of DNA. There are two ideas involved. The first idea is that there exists a “vegetative” form of DNA, consisting of two chains tied together not by their bases but back-to-back, by pyrophosphate linkage, the bases sticking out. This structure is not coiled but a ribbon. The second idea is that mature phage consists (in the case of virulent phages) of two standard W.C. duplexes. After injection one chain of one duplex ties up back-to-back with the chain of opposite polarity in the other duplex, forming a ribbon. The other two chains get chewed away by DNA-ase (ex-

78

T WISTS AND T URNS

plaining the 50% loss of label in the transmission experiments). One of the attractive features of this model is the fact that there is no problem of untwiddling. The single chains are freely rotatable around the single bands of the backbone. Another attractive feature is the fact that the ribbon turns its specificity to the outside making it available for either replication or pairing. These ideas are still embryonic and there is feverish activity to see whether things like recombination, heterozygotes, lysogenization, transduction, and recently discovered anomalies in Neurospora tetrads look more hopeful from such a point of view.12 It is easy to see why Delbru¨ck could become as excited as Watson about a structure with the potential to eliminate the “untwiddling” problem. Despite the fever pitch to which Delbru¨ck alluded, however, Watson had reason to remain wary. The new structure postulated the existence in DNA of a phosphate triester structure for which there was no experimental evidence.13 These responses to the replication problem posed by Delbru¨ck were, like his own treatment of the problem, mental exercises, or, at most, model-building exercises. They were efforts to evade the reasoning that had led him to conclude that the DNA strands must break apart at short intervals during every replication. Implicitly, perhaps, Watson and the physicists were acting on the intuitive assumption that the feature of the Watson-Crick DNA structure most critical to its presumed genetic role was the conservation of sequences of bases along extended lengths of nucleotide backbones during their replication. These models, whether abstractly topological or constructed in the Watson-Crick style, were simple games compared to the challenge to respond to Delbru¨ck’s invitation to settle the question experimentally by marking DNA molecules and tracing their distribution in successive generations. So long as Meselson and Stahl were precluded from implementing their plan to do so, the person best positioned to attempt the task was Gunther Stent. As we saw in Chapter 1 (pp. 44– 45), Stent was already exploring the distribution of phage DNA by means of the 32P marker when Delbru¨ck formulated his analysis of the problem. Born in Berlin like Delbru¨ck, but of Jewish parentage, Stent had fled Nazi Germany on New Year’s Eve, 1938. Arriving in Chicago in 1940, he attained a doctoral degree in physical chemistry at the University of Illinois in 1948. Inspired, like a number of other physical scientists in the early postwar years, by Schro¨dinger’s What Is Life?,

T WISTS AND T URNS

79

Stent dreamed about doing research under the guidance of the figure of Delbru¨ck portrayed in that book. When he found out in 1947 that Delbru¨ck had gone to Caltech, Stent wrote to ask if Delbru¨ck had a place in his laboratory for someone like him. Delbru¨ck replied that he did not. After Stent obtained a Merck Fellowship, however, Delbru¨ck accepted him. When Stent met Delbru¨ck for the first time, in a social setting in Chicago, he was immediately captivated. Before coming to Caltech, Stent attended the Cold Spring Harbor phage course during the summer of 1948, where he was initiated into the rites and ethos of the emerging phage group. He became acquainted with Renato Dulbecco, Jim Watson, Seymour Benzer, and others, and quickly decided that phage research offered him the chance to get in on the science of the future. During the two years he spent in Pasadena, Stent became more deeply attached to Delbru¨ck. The charismatic leader of the phage group not only impressed him with his analytical rigor but embodied for him a kind of ethical integrity that Stent felt kept his own tendencies to bend rules in check. It became a central motivation of his scientific efforts to be thought well of by Delbru¨ck.14 Stent hoped to work with Andre´ Lwoff in the Institut Pasteur at the conclusion of his fellowship at Caltech but was turned down by Lwoff. On Delbru¨ck’s recommendation, he went instead with Jim Watson, in the fall of 1950, to the laboratory of Hermann Kalkar in Copenhagen. Watson also frequented there the laboratory of Ole Maaløe, where he studied the transfer of page DNA to progeny by radioactive labeling with 32P. From Maaløe, Stent learned the techniques for incorporating 32P into viral nucleic acids.15 During the two years in which he stayed in Europe, Stent was present for some of the events that later appeared as landmarks in the emergence of molecular biology. In December 1950, he and Watson attended a lecture in Copenhagen by Lawrence Bragg that was intended to be on the structure of hemoglobin. Having received a few days beforehand a letter from Linus Pauling describing the α-helix structure for protein, however, Bragg talked instead about that imposing discovery. Stent noticed Watson becoming increasingly excited. After the lecture, Watson told him, “We have to find the structure of DNA.” Stent thought the idea banal. By the fall of 1951 he had managed to be accepted at the Institut Pasteur. While working there, he attended the conference on virus replication at Oxford at which Watson announced the results of the Hershey-Chase experiment. Although impressed with this news, Stent was influenced more deci-

80

T WISTS AND T URNS

sively by an idea that Guido Pontecorvo had presented at a genetics meeting in Copenhagen the previous spring. Pontecorvo maintained that the replication of genes and genetic recombination must be so closely related that neither process can occur without the other.16 During a discussion of a paper by Lwoff titled “The Nature of Phage Replication,” Stent now offered a “mechanism of replication” inspired by Pontecorvo’s point of view. Stent’s proposal required “neither templates nor face-to-face or back-to-back attachments”: Instead, a process of “crossing over” in which homologous pieces are exchanged between contiguous structures could be the mechanism by which small building blocks are assembled into large polymers of a given specificity. Consider a mother molecule A B C D E F which is to be duplicated in an environment containing the necessary building blocks A′,B′,C′,D′,E′,F′. The mother molecule splits into two fragments ABC DEF and only that building block may be fit on to the broken end which corresponds to the one to which the end had been joined before the break, i.e. A B C D′ C′ D E F Duplication is thus under way, but the newly-joined building blocks at the ends of the fragments do not yet possess the correct “twist” to ensure that only the right block could be joined next. By “crossing-over” between the two fragments, i.e. A B C′ D C D′ E F the newly-joined blocks C′ and D′ are placed into the interior of the molecule, and the homologous blocks of the mother molecule C and D possessing the right “twist” are at the ends. Now only the correct blocks may be joined on, i.e. A B C′ D E′ B′ C D′ E F

T WISTS AND T URNS

81

Again, “crossing over” A B′ C D′ E B C′ D E′ F and finally A B′ C D′ E F′ A′ B C′ D E′ F

...

The operational distinction of this conception of replication is that the atoms of the mother molecule are necessarily distributed evenly over the two molecules resulting from duplication. Under template and attachment theories, the atoms of the mother molecule would not appear in the daughter structures. Lwoff responded tactfully to this suggestion of the eager young member of the phage group: “Dr. Stent’s theory is very attractive. Yet it does not seem to account for the face-to-face attachment of the chromosomes. Moreover it is difficult to visualize the nature of the factors responsible for the splitting of the mother structure. Nevertheless I do not think that Dr. Stent’s scheme is more devoid of any positive evidence than any other theory of replication. We are now faced with half a dozen such theories and it may be hoped that one or the other may soon serve as a working hypothesis.” 17 Stent did not identify the “mother molecule” whose replication mechanism he envisioned. In the main paper whose presentation had afforded Stent the opportunity to put forth his proposal, Lwoff had defined the “theoretical virus” as a “specific reproducible nucleoprotein.” 18 The discussion thus revolved around conceptions of the nature of the replicable material that predated the impact that the Hershey-Chase experiment announced at this very meeting would soon have on the phage community. A few months later, in the fall of 1952, Stent joined the staff of the newly opened Virus Laboratory of Wendell Stanley at Berkeley.19 Before he arrived from France, Stent had decided that his initial project there would be to apply the techniques of 32 P decay he had learned in Copenhagen to test in bacteriophage whether he could find any positive evidence for his mechanism of replication.20 His strategy depended on the fact that his mechanism predicted that the “atoms of the mother molecule” would be dispersed evenly among the “molecules resulting from the duplication.” Six months before the advent of the

82

T WISTS AND T URNS

double helix posed the replication problem at the level of a clearly defined molecular structure, therefore, Stent had already set out, on the basis of more abstract ideas, to find an experimental means to test his and other replication mechanisms. By the time he had set himself up to begin this research, the Hershey-Chase experiment had convinced Stent that he was studying the replication of DNA. Then, in the spring of 1953, he invited Neville Simon, an Australian postdoctoral fellow in Delbru¨ck’s group, to give a seminar at Berkeley. When Stent went to pick him up at the Greyhound station in San Francisco, Simon emerged from the bus shouting that the structure of DNA had been found. He had brought along a copy of the letter in which Watson described the structure of the double helix to Delbru¨ck. (See above, pp. 12–13.) By the time they crossed the Bay Bridge, Simon had “told [Stent] the whole thing.” Stent believed it in “two seconds.” Back at the Virus Lab he called people into his office to spread the news, but when it came out that the double helix was a deduction from X-ray crystallography, his colleagues expressed the view that it might soon turn out to be a mistake.21 By the time Stent presented the first report of his research on mortality due to radioactive phosphorus as an index to bacteriophage development at the Cold Spring Harbor Symposium in June 1953, his plan to use this method to study the “distribution of parental phosphorus among replicas” confronted the prediction of the Watson-Crick model of a distribution unlike that implied in his own earlier idea. By the spring of 1954 his work was transformed into the potential means to decide between the replication schemes of Delbru¨ck and of Watson and Crick. As the correspondence between him and Delbru¨ck indicates, however, Stent himself had by this time taken up a third alternative—that the parental atoms are not distributed during replication. Stent also reported to Delbru¨ck then that “the final experiment which was supposed to have clinched everything doesn’t seem to want to come out right.” Sometime in this period, perhaps in the late spring of 1954, Stent received a visit from Matt Meselson, who was visiting the Bay area and had been asked by Delbru¨ck to convey greetings to Stent. When Meselson asked about his work, Stent told him that he was trying to determine the distribution of 32P atoms of parental phage DNA in its replicas. He had got some interesting results favoring his current view that the atoms were not distributed, “but nothing was really conclusive, because it was so difficult to ascertain the distribu-

T WISTS AND T URNS

83

tion of 32P at the molecular level.” After thinking about the problem for a while, Meselson asked Stent why he didn’t use a stable isotope, like 15N, to label the parental DNA. Stent dismissed the idea, because he thought that the mass spectroscope could not resolve the 15N content of individual DNA molecules. When Meselson asked if it might be possible to measure the density difference that such a label would produce, Stent replied that the difference would be far too tiny to measure.22 When his continued efforts with 32 P labeling seemed to go nowhere, Stent began to feel that the distribution problem would be impossible to solve. At that point the Danish immunologist Niels Jerne, with whom Stent had become friends in Copenhagen and who was a visiting fellow at Caltech, came to see him in Berkeley. Stent explained the transfer experiments and his difficulties to Jerne while driving him to San Francisco. Jerne quickly saw a way in which he might succeed by letting the decay continue into the second reproductive generation, and he proposed a protocol for the investigation. Stent suggested in turn that they carry out the experiments together. Agreeing to the idea, Jerne returned to Berkeley to begin the work. Shortly afterward, however, he accepted an invitation from another colleague to drive across the country, and Stent was left to complete the experiments on his own. In July 1955, he submitted the results to PNAS in a paper, co-authored by Jerne, titled “The Distribution of Parental Phosphorus Atoms Among Bacteriophage Progeny.” 23 Unable to trace the distribution of 32 P incorporated into parental phage DNA directly by locating it in the progeny DNA molecules, Stent reached his conclusions more indirectly by statistical analysis of the survival curves of parental and progeny phage progressively inactivated by the decay of the 32 P contained in their DNA. “The basis of these experiments,” he wrote, “is that the bacteriophages lose their infectivity upon decay of radiophosphorus P 32 incorporated in their DNA, the rate of inactivation being proportional to the number of P 32 atoms per particle. . . . We have . . . endeavored to detect the presence of parental P 32 atoms in the descendant phages by observing the lethal effects of the decay of these atoms on the progeny population.” 24 To measure these rates of decay, Stent infected bacteria with T2 or T4 phage containing 32 P, caused them to lyse, then stored the lysates at 4°C for periods ranging from a few hours to forty days before using them to reinfect bacteria. The declining infectivity of the progeny

84

T WISTS AND T URNS

phage was a measure of the radioactive decay within them. It turned out that for the progeny populations as a whole the rate was no more than 0.2 percent of the rate at which parental 32P phage is inactivated. From this result he inferred that “the majority of the individuals among even the earliest progeny do not contain more than 0.2 percent of the parental phosphorus. Since the average transfer to the first intracellular progeny has been estimated to be 2 percent parental phosphorus per particle [that is, ten times as much as is transferred to the majority of individuals], it would follow that most of the transferred phosphorus resides in a minority of the progeny population.” 25 By conducting similar experiments on the infectivity of secondgeneration progeny, as Jerne had suggested, Stent tried to determine how small a minority of the progeny contained this radioactive parental DNA. Their rate of inactivation ranged between 2 and 6 percent of the rate for the parental generation. From this result, and the fact that almost all of the transferred phosphorus entered the first fifty or so progeny formed, he reasoned that “the number of phages which carry the bulk of the transferred phosphorus atoms must be . . . between 25 and 8 particles. The fact that the parental atoms are distributed over at least 8 particles indicates that the parental bacteriophage DNA experiences a certain dissociation in the course of its reduplication within the host cell.” Stent emphasized that “this dissociation proceeds in single infection, i.e., under conditions in which it is certain that the progeny are, in fact, the true offspring of the individual whose radioactive atoms they bear.” The “significance of the present findings for various proposals concerning the mechanism of DNA replication,” he concluded, “will be considered elsewhere.” 26 It is not surprising that he deferred this discussion. The conclusion that there is “a certain dissociation” was unfavorable to Stent’s own view that the atoms are not distributed in replication, yet it provided no clearcut decision between the alternative mechanisms that Delbru¨ck had proposed. Although he had good working conditions at Berkeley, Stent felt rather isolated in the Virus Lab. Still regarding Caltech as his intellectual home, he went down to Pasadena almost every other month to keep in touch with Delbru¨ck and often participated in the camping trips that Delbru¨ck organized. He also got to know Meselson well there and to reverse his initial impression, obtained from the suggestion about the use of 15 N as a label, that this young student of Pauling lacked scientific acuity.27

T WISTS AND T URNS

85

Fig. 3.1. Gunther Stent in the Virus Lab, 1955. Photo courtesy of G. S. Stent.

II At about the time that Stent and Jerne submitted their study of 32 P distribution in phage DNA to PNAS, Frank Stahl was finishing his thesis on the genetic damage done by 32 P to bacteriophage DNA.28 Much of Stahl’s investigation paralleled work that Stent had done during the previous two years, but his results diverged sufficiently to lead

86

T WISTS AND T URNS

Fig. 3.2. Fraction of contributors of one, two, or three markers. From F. W. Stahl, Ph.D. diss.

him to a critique of some of Stent’s conclusions concerning the elimination of genetic markers. Whereas Stent had found evidence for the independent elimination of each of two genetic markers on 32 P inactivated T2, Stahl’s results with three markers on 32 P inactivated T4 indicated that markers tended to be eliminated together. When he graphed the contributions of markers from the 32 P phage to the progeny obtained in his cross-reactivation experiments as three curves showing “the fraction of contributions contributing one, two, or three markers derived from the suicide parent” as functions of the amount of 32 P decay (the decay was a function of time but was expressed in the form of “mean number of lethal hits”), their form showed distinct departures from a theoretical set of curves one could construct on the assumption of independent elimination (figure 3.2).29 Stahl gained further information about the reactivation of markers from the suicide phage by characterizing the frequency distributions of the burst sizes of progeny phage for which the markers appeared. Since Delbru¨ck had analyzed the burst size distribution in the growth of bacteriophage in 1945, burst sizes and their distributions had become a staple technique for measuring the phage yields from individ-

T WISTS AND T URNS

87

ual infected bacteria. After a suspension of bacteria has been infected by phage the suspension is diluted and separated into samples small enough so that, on the average, each sample will contain less than one infected bacterium. When the bacterium is lysed in each of the samples containing one, therefore, all the progeny phage released into the sample are derived from a single infection. One can determine this number—that is, the burst size—by mixing the sample with another suspension containing many bacteria and plating them onto agar. Each progeny virus will produce a plaque surrounding the bacterium it infects, so that by counting the plaques one determines the number of progeny. If one does this for numerous samples taken from the same experiment, the individual burst sizes show a large variation, distributed around a mean or average size that typifies the conditions of that particular experiment.30 When Stahl analyzed the burst size distributions for reactivated markers in his experiments, he noted two particularly significant features: (1) “for doses higher than a few lethal decays per phage, the mean burst-size of reactivated markers is independent of dose.” From this and some other considerations, he inferred that “the proximity of an eliminated marker to a surviving marker will have no effect on the burst size of the surviving marker;” (2) “after a few lethal decays per particle have occurred the frequency distribution of burst sizes of phage carrying reactivated markers is highly non-random.” In random distributions one would expect a maximum frequency near the mean burst size, with frequencies becoming smaller the further the burst size is from the mean. Here there was, instead, “an appreciable number of bursts liberating many particles carrying a given reactivated marker and a similar number liberating very few particles.” Because this pattern differed strikingly from the random distributions of burst size typical of recombinant phage progeny when neither infecting phage is radioactive, Stahl concluded that the process in which markers from the “suicide parent” are reactivated must be separate from the process of recombination.31 In order to explain these results, Stahl developed a conceptual model of the reactivation process. The nonrandom burst size distribution suggested a clonal process. Markers are rescued infrequently, but early in the replication process, so that those that are reactivated then multiply within the bacteria, releasing numbers of progeny varying from very few to very many, depending on the number of generations that take place before the bacteria lyse. He envisioned that 32 P decay does not necessarily destroy individual genetic markers themselves,

88

T WISTS AND T URNS

as Stent and others had suggested, but fragments the DNA molecule. These fragments cannot multiply on their own but may pair with portions of the intact genome of the carrier phage, subsequently multiplying. The contrast between the clonal properties of reactivation and the nonclonal burst size distribution of ordinary recombination events suggested to Stahl that the latter occur later in the reproductive process, so that there is little or no further multiplication of the recombinant particles. The burst size distributions for 32 P suicide phages also contrasted with those for the UV cross-reactivation events that Doermann was studying, which were more like ordinary recombination events. This difference indicated, Stahl thought, that the physical nature of the damage inflicted on the phage DNA by these two means also differed.32 It may have been because he had already perceived, during the previous fall, that the multiplication of reactivated markers would be important to his interpretation of his experiments that Stahl had tried to persuade Meselson then to introduce multiplication into their quantitative model for cross-reactivation. Meselson’s opinion that it was not feasible to do so must have discouraged Stahl from applying their model to his own investigation, inducing him to settle for the qualitative model that he discussed in his thesis. In his experiments Stahl had tested the elimination patterns for unlinked T4 markers and for linked markers with recombination frequencies ranging from 12 to 23 percent. In addition, he examined the burst size distributions for two closely linked markers that he received from Seymour Benzer, from the rII region of the T4 genome, whose recombination frequencies were, respectively, 0.1 and 1.1 percent. The distributions found in the latter experiments were, he concluded, “not easily in agreement with the hypothesis by Levinthal.” 33 Cyrus Levinthal received his Ph.D. in physics at Berkeley in 1951 for research in high energy physics. Appointed to the Department of Physics at the University of Michigan, Levinthal shifted within a year to the field of bacteriophage genetics.34 Levinthal’s hypothesis was a recombination model proposed in 1954 to account for the observation by Hershey and Chase in 1951 that a small percentage of the phage progeny resulting from mixed infections are heterozygous for several of their genetic markers and homozygous for the rest. Unlike higher organisms, which generally carry a full double set of genes (known since early in the century as the diploid state) that segregate in the next sexually produced generation, phage normally behaved as

T WISTS AND T URNS

89

though they carried only a single set (the haploid state). That there now appeared to be exceptions to this generalization—that phage derived from mixed infections could give rise in the next generation to a mixture of mutant and wild type progeny for a few of their genetic markers—invited an explanation for how a small portion of their genetic material could be duplicated within a phage particle.35 On the basis of results of three-factor cross experiments on phage T2, which showed that particles heterozygous for a middle marker were generally recombinant for the end markers, Levinthal proposed a model through which “a partial duplication . . . of a haploid chromosome could occur.” Each parental type contributes a portion of its chromosome, the duplication being an overlap between homologous portions of the two chromosomes.36 From calculations of “the number of recombinants which would be contributed by this mechanism alone,” he concluded that heterozygotes are intermediates in the production of all recombinant phage.37 The mechanism of recombination in phage thus differed fundamentally, according to Levinthal, from the crossover mechanism in higher organisms, in which two chromosomes line up together, breaks occur at the same point or points on both of them, and the segments are reattached in the opposite order to one another. “If all the assumptions” in Levinthal’s theory were correct, Stahl wrote in his thesis, “a considerable degree of clonality would be expected in the burst size distribution of recombinants between closely linked markers.” That was because, once formed, the recombinants should multiply at the same rate as the rest of the phage population. The distribution for the two closely linked markers in his own experiments showed, however, “almost no clonality.” Resolution of this apparent contradiction would require a fuller understanding of the normal recombination process, including a test of Levinthal’s supposition that recombination is dependent on multiplication.38 When Stahl finished his Ph.D. during the summer of 1955, his immediate future seemed assured. He had in hand a Rockefeller postdoctoral fellowship to support him for two years at Caltech, which was for him, as for other young phage biologists, the mecca of their field. In July he met Mary Morgan, a native of Rochester and a student at Antioch College, with whom he had previously corresponded. On their first date they played tennis and felt instantly at ease with one another. Twenty years old, Mary seemed to Frank not only young but exuberant, straightforward, and athletic. Very pretty, she had blue

90

T WISTS AND T URNS

eyes, and, he noted appreciatively, “great legs.” One week later Frank and Mary were married. Shortly afterward they drove across the country together and found a small apartment in Pasadena.39 Two months later, Stahl’s new life fell apart. Mary was extremely unhappy in Pasadena, where the smog made her sick to her stomach, and the recently acquainted newlyweds who had at first seemed so comfortable with one another now seemed unable to adjust to each other. Pregnant and desperate, Mary left for Chicago, where she found a job as a typist and intended to put the expected baby up for adoption.40 Perhaps in part to rescue Stahl from his desolation, Matt Meselson and Jan Drake revived the plan that they had discussed fourteen months earlier at Woods Hole to find living quarters together. Having noticed that a large elegant house on San Pasqual Street, less than a block away from the biological laboratories, was vacant, Meselson and Drake induced Stahl, as the only postdoctoral fellow among them, to negotiate the rental. Formerly belonging to a mathematics professor, the house was now owned by Caltech. Dressed uncharacteristically in a coat and tie, Stahl persuaded the firm, prim, and somewhat dubious institutional landlady that he could be entrusted with the care of the house, as well as the payment of the rent, and the lease was signed in his name.41 The house had large living and dining rooms, a kitchen, and a guest room on the first floor, with three bedrooms upstairs. The three new tenants obtained old furniture and dishes, including quite a lot that Meselson brought over from his parents’ home in Los Angeles. After they had moved in, they divided up the chores. Meselson acted as cook, Stahl did the dishes, and Drake cleaned the house and took care of the yard. Meselson ordinarily did not come home from the laboratory until 7 or 8 P.M., and he got the dinner on the table at about 10 P.M. To keep his roommates happy until then, he maintained a wellstocked bar provided by his father. The new bachelor establishment on the northwest rim of the Caltech campus also became the scene for festive weekend parties.42 Not all of their neighbors welcomed these arrangements. The occupants of the house noticed that a woman living across the street sometimes trained a pair of binoculars on their residence and was particularly attentive if young women appeared at breakfast on their patio. Drake was stopped one day while piling lawn cuttings in the street, as was common practice in the area, by a policeman who informed him, with some embarrassment, that the practice was illegal.

T WISTS AND T URNS

91

Their reunion at Caltech did not permit Meselson and Stahl to take up at once their plans for DNA replication experiments. Although Linus Pauling had shown interest in Meselson’s initial ideas on the subject, he had told him to finish the crystallographic part of his thesis work before starting anything else.43 In September 1955, Meselson was still a long way from that point. Stahl had obtained his fellowship for the avowed purpose of learning bacterial genetics with Giuseppe (Joe) Bertani.44 Born in Italy, Bertani completed his Ph.D. dissertation at the University of Milan in 1944, while Allied warplanes sometimes flew over the buildings. Two years later he obtained a research fellowship at Cold Spring Harbor, then joined Luria’s group at the University of Indiana. A good friend of Gus Doermann, Bertani had met Stahl during a visit to Rochester. Delbru¨ck brought him to Caltech as a senior research fellow in 1954 to supervise the research of his phage group, while he himself turned to phycomyces. Before coming to Pasadena, Bertani had investigated “temperate” phage, which can infect bacterial cells and remain within them for generations without destroying them. They are incorporated into the genetic material of the bacteria, creating a “lysogen,” and transmitted to bacterial progeny in the same fashion as the genetic markers characterizing the bacteria themselves. This relationship was not truly symbiotic, as Bertani pointed out, because the bacterial cells do not carry complete infectious phage but only a “self-reproducing ‘tape,’ ” which Andre´ Lwoff had named prophage.45 In the wake of the Hershey-Chase experiment and the WatsonCrick structure, many phage biologists expected prophages to consist entirely of DNA. By infecting the bacterium Escherichia coli strain C with different mutants of the lysogenic phage P2, Bertani had found by the summer of 1955 that two or more individuals of the same species of phage can be incorporated into the bacterium. By crossing a “doubly lysogenic strain” of the bacterium with a strain containing no prophage, Bertani showed that the mutant prophages can be inherited independently of each other, suggesting that they are genetically recombined just as ordinary bacterial genetic markers are. These results led him to suggest that the two prophage strains are attached at different sites along the bacterial chromosome.46 In order to perform such experiments, Bertani had to carry out crosses between bacteria that differed in two or three genetic markers in addition to containing or not containing prophage. Bacteria normally reproduce asexually by fission, but in 1946 Joshua Lederberg

92

T WISTS AND T URNS

and Edward Tatum had found, by the analysis of mixed cultures of E. coli strain K-12, that recombinant progeny types sometimes arose. They inferred the “occurrence of a sexual process in the bacterium.” Subsequently it was found that in this process two morphologically differentiated bacterial cells collide and form a temporary attachment. In 1952 Lederberg proposed the symbol “F⫺” to denote bacterial strains that are sterile when mated and “F⫹” for strains that form fertile matings with F⫺ or other F⫹ strains. By 1955 the investigations of William Hayes in England and of Franc¸ois Jacob and E´lie Wollman at the Institut Pasteur had shown that in the mating process all or a part of the F⫹ chromosome is donated to the F⫺ cell. By analogy to the reproductive processes in higher organisms, the F⫹ donor strain was informally called the “male” and the recipient F⫺ the “female.” Bertani’s experiments required a mutant male or female strain to cross with a wild type strain of the opposite type.47 Bertani regarded Stahl as a mature experimentalist free to pursue his own interests. To provide Stahl with experience in basic bacterial genetics, however, Bertani gave him the job of finding “potent male” mutants of E. coli strain C like those recently discovered in strain K-12. In most F⫹ ⫻ F⫺ crosses the frequency of recombinant progeny was 10⫺5 or less of that of parental type progeny. Two mutant strains of K-12 had arisen, however, which gave much higher recombinant frequencies. They were designated, accordingly, “Hfr” strains. In 1955 the genetic relationship between F⫹ and Hfr strains was an unsolved problem. Bertani provided Stahl with an F⫺ and an F⫹ strain of the bacterium. Stahl’s goal was to isolate an Hfr mutant from the F⫹ population and grow from it a pure culture.48 Meanwhile, Stahl circulated his thesis through the phage group at Caltech to obtain criticisms before preparing it for publication and sent a copy also to Levinthal. Another postdoctoral fellow, George Streisinger, challenged his interpretation of his results on two grounds. The more distant genetic markers Stahl had used were so far apart that more than one recombination event per infection would take place in a large fraction of the bacterial cells. Concerning the closely linked markers in the rII region, Streisinger pointed out that Levinthal’s model predicted a delay of one generation in the replication of recombinants and that this lag would be expected to reduce the clonal character of the burst size distributions.49 Levinthal replied on 14 October, congratulating Stahl “on a very nice job” but wishing to “argue . . . about the question of the distribu-

T WISTS AND T URNS

93

tion of recombinants.” New calculations based on the assumption of “a constant pool size in which there is presumably an equilibrium between the increase in number of [phage] particles due to growth and the decrease due to maturation and extraction” had given him results that seemed to be in agreement with Stahl’s experiments. Stahl agreed that his “conclusion of no multiplication of recombinants is shaky” but countered in a letter of 18 October that he and Streisinger had found, in the data from the experiments with closely linked rII mutants that Stahl had conducted for his thesis, ratios of heterozygote/ (heterozygote ⫹ wild type) that conflicted with the prediction of Levinthal’s model.50 In order to test Levinthal’s model further, as well as to obviate Streisinger’s objection that two or more independent recombination events could occur in the same cell, Stahl and Streisinger undertook collaborative experiments on the frequency distribution of recombinants from two rII markers “10 times as close (0.01% recombination)” as the pair Stahl had previously used. These markers had also been identified by Seymour Benzer using a special trick that Stahl had used to observe recombinants occurring with such low frequencies. The rII mutants produce plaques only on E. coli strain B, whereas the wild type produces them also on strain K-12. By making the cross on strain B and plating the progeny on strain K-12, one could detect wild type recombinants in proportions as low as 10⫺5. Streisinger and Stahl also exploited this feature of rII mutants.51 Soon after Stahl arrived at Caltech, Delbru¨ck had given him the job of putting together the semiannual progress reports on work in Delbru¨ck’s laboratory that was supported by a grant from the National Foundation for Infantile Paralysis. By the time Stahl wrote the report for the period July–September 1955, he was able to include the data for the wild-type recombinants formed in the experiments carried out with Streisinger. The distribution was nonrandom. “This finding strongly indicates,” he wrote, “that recombinants may multiply in the cell in which they originate. The nature of the distribution here found is compatible with Levinthal’s theory of recombination.” He added that they planned to test Levinthal’s theory further in future experiments in which they would check whether the ratio of heterozygous to nonheterozygous recombinants fit its predictions.52 Although still inclined to doubt Levinthal’s theory, Stahl had, for the time being, only conjectural arguments to raise against it. Consequently, in revising his thesis for publication, he acquiesced to Levin-

94

T WISTS AND T URNS

thal’s request that he eliminate its criticism of that theory. He submitted the text and data with few other changes to the journal Virology, which accepted it in January 1956.53 His thesis research, its continuation in the collaborative experiments with Streisinger, and his correspondence with Levinthal were now drawing Stahl more deeply into questions about the theoretical structure of bacteriophage genetics. The immediate issue for him revolved around the distinction between clonal and random distributions of recombinants among progeny phage. In his letter of 18 October to Levinthal he mentioned a “Luria distribution with clones of size 1–16.” The reference was to an analysis that Luria had published in 1951 of the frequency distribution, not of recombinants, but of spontaneous phage mutants. Levinthal responded acerbically: “You seemed to think that on the basis of some theory one should expect a Luria type distribution. What I’m saying is that I don’t expect a Luria type distribution on any theory, assuming you reach a constant pool size. . . . I believe the confusion has been that people were thinking in terms of the extremes of the clonal distribution or a Poisson [perfectly random] distribution. The fact is that most experimental results are between the two and what you expect from the theory of a constant pool is roughly in between the two.” 54 Whether goaded by Levinthal or merely prompted by the general problem to which Levinthal here gave expression, Stahl confronted, during the following months, the implications for his ongoing investigation of the two leading current mathematical models of phage reproduction. Luria had conducted a “Herculean” set of single burst experiments on the distribution of phage mutants, for which he counted 1.8 million plaques on 2,874 plates and found 766 mutants. The clonal distribution of the mutants “fit quite well,” Luria concluded, the “hypothesis that the genes responsible for the investigated phenotypes reproduce exponentially by successive reduplications.” Luria’s model assimilated phage reproduction to the classical mode of reproduction of higher organisms, only specifying that one phage reproductive cycle involved a “series of successive acts of replication.” 55 In 1953 Delbru¨ck and N. Visconti proposed a different mechanism of phage reproduction. Treating the topic as a problem in population genetics, they assumed that during the vegetative stage, phage particles multiply and mate repeatedly, in pairs and at random with respect to partners. Particles begin soon afterward to be withdrawn from the resulting pool of mating and multiplying partners and are “set aside as mature particles.” Delbru¨ck worked out a set of mathematical equa-

T WISTS AND T URNS

95

tions from which he could predict the frequencies of recombinant phage to be expected in two- and three-factor crosses—that is, when bacteria were infected simultaneously by phage differing in two or three genetic markers. The equations included a variable for the average number of rounds of mating. For the experiments carried out by Visconti with phage T2, the values obtained were in “good agreement with an estimate of five rounds of mating.” 56 It was the difference between the distribution of the recombinants in his experiments with Streisinger on closely linked T4 recombinants and the distribution of mutants in Luria’s results that prompted Stahl and Streisinger to ponder the relationships between these respective models. There was no apparent reason why the rare recombinational events should produce clone-size distributions that differed from those produced by rare mutational events. They considered the possibility that their results might better fit a model based on “steady-state pool kinetics,” that is, if in the Visconti-Delbru¨ck theory one made the additional assumption that “the rate of maturation balances the rate of replication,” the pool of replicating phage will be constant in size.57 Because Levinthal described in his letters to Stahl in October his attempt to modify the Luria clonal distribution by making the same assumption of an “equilibrium pool,” it seems that he, too, may have helped to stimulate their thinking in this direction.58 Soon after arriving at Caltech, Stahl had met a doctoral student named Charles Steinberg, who appeared to everyone there, including Richard Feynman, to be the “cleverest person” he had ever met.59 As Stahl undertook to analyze the mathematical theories underlying the respective models of Luria and Visconti-Delbru¨ck, he enlisted Charley Steinberg’s assistance. By the spring of 1956, Steinberg was doing most of the math required to support Stahl’s inquiry and his ongoing debates with Levinthal.60 Stahl felt at home experimenting and reasoning in the distinctive style of the phage group and was attracted also to its collective personal style. Phage biologists as a group worked hard, but also played hard, and they tried at least to appear not to take themselves too seriously. Although they preferred phage research because of its relative simplicity, they knew that biology as a whole was so complicated that what they were now doing might later appear foolish.61 Having absorbed some of this style already under Gus Doermann, his mentor at Rochester, Stahl was well prepared for the ethos that prevailed at Caltech under the dominating personality of Max Delbru¨ck. In contrast to Watson, Stahl accepted wholeheartedly Del-

96

T WISTS AND T URNS

bru¨ck’s attitude toward scientific activity. Although he soon learned that Delbru¨ck’s strong opinions were “almost always wrong” on specific questions, Stahl found him “right about the most fundamental things: . . . that science is a lot more fun when it is done socially as a community rather than competitively.” Delbru¨ck appeared less concerned about the credentials that one brought to a problem than with the ideas one could contribute. Stahl felt that he was accepted immediately into this group, simply because he was interested and willing to offer whatever ideas he had to problems of common interest.62 Unlike Watson, who had judged Caltech always in comparison to Cambridge, Stahl found the other members of Delbru¨ck’s group to be engaging colleagues. Some of them, like Charley Steinberg, Stahl considered brilliant, wonderful people with whom to argue, and he thoroughly enjoyed his collaboration with George Streisinger.63 In the house on San Pasqual Street, too, life was lively. As the only student house of its kind, it became a major social center. There were many memorable parties, which attracted members of the faculty as well as graduate students. Max and Manny Delbru¨ck were frequent participants, and Richard Feynman often played his drums there. Late-night parties sometimes ended with Sunday morning breakfasts, and most of the very few women present at Caltech appeared on the scene.64 The house also became a center for more serious events. There were many avid conversations among Meselson, Stahl, Drake, and frequent visitors, on both scientific and nonscientific topics. Among the guests was Justice William O. Douglas, who accepted an invitation to a breakfast discussion. Meselson’s irrepressible enthusiasm drew the others to his own interests. Emulating Linus Pauling, Meselson became deeply involved in political activity, particularly in the efforts of the Federation of American Scientists to warn against the dangers of atomic weapons. Soon the whole house became engaged in these activities. Stahl too was drawn in by Pauling, who once asked him to share the platform for a speech about the radiation hazards posed by nuclear weapons. After Pauling had described the physics and chemistry of the bomb, Stahl spoke about the biology of radiation. Afterward Pauling sometimes asked Stahl to speak for him at occasions for which he had been invited but could not attend. When Pauling’s passport was revoked, Jan Drake organized a meeting to protest the denial of passports and visas on political grounds.65 For all the appeal that life in Pasadena held for Stahl, these months were still very hard. Lonely for a marriage that seemed to have failed

T WISTS AND T URNS

97

Fig. 3.3. Frank and Mary Stahl, spring 1956. Photo by M. Meselson.

from the outset, he sometimes drank too heavily during the long interval between the time he came home and the time Meselson had dinner on the table. Having less confidence in his abilities than others seemed to have in him, he experienced endless frustration in his experiments on bacterial genetics. After several months of attempting to produce Hfr cultures of E. coli strain C, he had obtained no useful results. His failure to execute what ought to have been a simple investigative assignment reinforced his belief that, much as he enjoyed designing and interpreting experiments, he must be singularly inept at performing them.66 While Mary Stahl awaited the birth of a baby that she did not intend to keep, Stahl’s mother urged on him the idea that it was his responsibility to care for the child himself. At Easter 1956, Stahl made the trip to Chicago to meet this daunting challenge. His roommates heard nothing from him for some time, then received a postcard that said, “I found a woman in Chicago.” They feared at first that he had further complicated his life, until they realized that he meant that his young wife had become a woman. When he arrived he had found that Mary now wanted her baby and wanted a family to go with it. She persuaded Frank that they should return together to Pasadena. The newborn boy, whom they named Richard Andrew, had some initial digestive problems. While they waited for him to be discharged from the hospital, Mary and Frank had time to work on their own relationship. A month later they came back with Andy and stayed for a few weeks in the “boys’ house” on San Pasqual Street while they looked for more permanent quarters. After one month spent in a relatively inadequate rented place, Mary found a simple but attractive old house, owned by Caltech, which rented for sixty dollars a month. There they settled down comfortably to begin anew. This time Mary stayed.67

98

T WISTS AND T URNS

III While Frank Stahl completed his thesis, encountered the pitfalls of bacterial genetics, enjoyed a collaborative venture in theoretical phage genetics, and lost and regained his marriage, Matt Meselson concentrated on mastering the craft of X-ray crystallography. Having spent much of 1954 obtaining satisfactory X-ray photographs, Meselson shifted his attention in 1955 to the complex analytical procedures required to obtain a match between trial structures and the diffraction patterns his photographs showed. Early in the year he constructed a new molecular model, “Trial Structure II,” for which he measured x and y coordinates for the five carbon atoms, two nitrogen atoms, and two oxygen atoms of the N,N′-dimethylmalonamide molecule. These he used to calculate scattering amplitudes, from which he made a Fourier-series projection of electron density down the twofold axis. “The trial structure was shifted only slightly,” he noted, “and hence looks right in its essentials.” This result was, however, only a first approximation. Reiterating the process, he altered the coordinates, made successive Fourier projections, derived parameters for the z coordinates, and gradually reduced the “discrepancy factor” between the calculated and observed “Structure factors.” 68 These tedious procedures occupied Meselson through most of the spring and summer of 1955. Finding in September that some of the y coordinates he had chosen for the carbon, oxygen, and nitrogen atoms still did not give satisfactory calculated structural factors, he decided to try to obtain better photographs. On 30 September and 1 October, he produced photographs 67–89. With these additional data he constructed a Patterson map, from which he was able to determine “a new position along [the b axis] for the molecule.” He realized also that an erroneous choice of a space group had given him mistaken values for the parameters along the x axis. A further reiteration of the Fourier projection improved the fit between calculated and observed values. As the year neared its end, Meselson was beginning to calculate the vibrational axes of the atoms of the molecule. Having attained good values for the parameters of the heavy atoms in the backbone of the molecule, he now constructed a “difference map” and built a balland-stick model of the molecule to assign positions to the hydrogen atoms.69 Writing to Jim Watson on 6 December for recommendations for the renewal of his NSF fellowship and for an application for a PHS

T WISTS AND T URNS

99

fellowship, Meselson reported that he had “been trying hard to complete the x-ray diffraction part of my thesis and at last the end is within a few weeks away.” Anticipating the completion of his student’s first research assignment, Pauling had already asked him to do some experiments in which he would introduce small amounts of amino acids into crystals of other amino acids “in order to find the energy of replacement.” Delbru¨ck told Meselson, however, that two other scientists had already done this, and Meselson thought that “the results of such experiments are not related closely enough to template synthesis to justify doing much more.” He hoped to persuade Pauling instead to “try the partial hydrolysis of Moore and Stein on several protein materials which are known to be related to each other by mutation. The various haemoglobins or perhaps even whole or tail protein from phages with different h markers might be tried.” Meselson had discussed these ideas with people at Berkeley, who had cautioned him about the complexities involved, but he still thought it might be worth a try.70 His references to “template synthesis” and to mutant hemoglobins and phage proteins suggest that Meselson was seeking ways to determine the effects of mutations on protein structure. More specifically, he sought a point of contact between protein structure and the biological implications of the double helix. The plan for studying DNA replication by density methods that he had discussed with Stahl at Woods Hole clearly did not inhibit his fertile imagination from thinking up other possible avenues toward his long-range goal of moving from physical chemistry to the study of biological problems. That he did not pursue this idea probably had less to do with the fact that “What I do during the last predoctoral year is largely up to Dr. Pauling” 71 than with the fact that he was farther from the completion of his x-ray crystallography work than he imagined. It took him nearly six more months to finish that task. Compared to Stahl, Meselson led a free and uncomplicated life at Caltech. To him it was the ideal place to be a graduate student. The small number of students, the accessibility of the faculty, and the intimacy of the compact campus encouraged a feeling of openness and informal interaction. The groups under Pauling in the Chemistry Division and Delbru¨ck in the Biology Division had little direct contact with one another, but he found no difficulty bridging the gap personally. Both Pauling and Delbru¨ck, and even George Beadle, seemed to him like loving uncles. Meselson deeply admired Linus Pauling and hoped

100

T WISTS AND T URNS

to emulate his ability to combine outstanding science with involvement in the great international questions of the day. Sociable and engaging, Meselson was also happy in Pasadena. In the “great house on San Pasqual” he and his friends ate and drank “most unusually well for Pasadena and for graduate students.” Like Watson, he missed some of the intellectual and social stimulation he had experienced elsewhere, particularly at the University of Chicago. Here he encountered none of the fierce political and intellectual debates he had enjoyed there. As it had been for Watson, the lack of women at Caltech was “worrisome,” but for Meselson less oppressive. More poised and socially self-assured, Meselson made contacts more easily. Having acquired a sleek new two-seated black Thunderbird capable of 130 miles an hour, he did not hesitate to speed off to San Francisco for a weekend with a girlfriend. In contrast to Stahl, he had no financial worries, and he had experienced few personal setbacks to dampen his optimistic temperament.

IV While Meselson and Stahl continued to direct their activities elsewhere, other scientists continued to grapple with the DNA replication problem. During the fall of 1955 two more theoretical attempts to resolve the untwisting dilemma appeared. From Copenhagen the physicist Niels Arley contributed to Nature in September the idea that the scheme that George Gamow had proposed in 1954 to show how the bases in DNA might code for the amino acids of protein could be exploited also to obviate the difficulty Delbru¨ck had identified in the separation of two DNA chains. The 1:1 correspondence between the polynucleotide chains of DNA and the polypeptide chains of proteins that Gamow had invoked to show how a protein molecule can be built up on the double helix of a DNA molecule prompted Arley to assert that by an inversion of this process the protein molecule can serve as the template around which another DNA molecule can be built. The order of the “purine-pyrimidines is now uniquely determined by the order of the amino acids,” and the new DNA molecule would be an “exact replica of the original one.” 72 A similar but more elaborate replication mechanism came from David Bloch at the Histochemical Laboratory of Columbia University. Bloch also relied on the helical structure of a DNA-protein complex as the basis of his mechanism. The protein in his scheme, however,

T WISTS AND T URNS

101

was not the protein synthesized from the DNA but the histone or protamine long known to be associated with DNA in chromosomes. “The two polynucleotide helices” are connected to each other “not only by hydrogen bonding through their complementary bases,” as postulated by the Watson-Crick model, but also “by ionic linkage through the associated” polypeptide. “Replication of the polynucleotide in the deoxyribonucleoprotein complex can be visualized by first postulating a slight rotation or vertical displacement of one of the helices relative to the other . . . in order to be able to accommodate additional strands of polynucleotide.” 73 Using diagrams, Bloch showed how the hydrogen bonds might be broken and the bases rotated 180°, placing the bases in a position to combine with unpolymerized nucleotides. Their attachment to the histone would align the nucleotides properly during the synthesis of two new polynucleotide helices. Bloch’s scheme only postponed the unwinding dilemma to a stage following, rather than preceding, the replication process. To confront that dilemma he had to resort to Gamow’s other recent suggestion: that a coiling opposite to the direction in which the helix is wound releases the coils of the complex from one another. He explained away any contradictions between his envisioned configuration and existing X-ray diffraction studies on the grounds that the latter had been carried out on material obtained from cells in which DNA was not replicating. This scheme, which Block acknowledged could “claim no direct evidence in its support,” merely joined the growing list of possibilities suggested to elude the topological conundrum posed by the double helix.74 About this time Max Delbru¨ck, whose paper eighteen months earlier had provoked much of the attention directed since then to the replication problem, decided to review the subject again. An appropriate forum, he thought, would be a symposium on the chemical basis of heredity to be held in July 1956 at the McCollum-Pratt Institute of Johns Hopkins University. Gunther Stent was, in Delbru¨ck’s view, the person most actively pursuing the problem experimentally, and Delbru¨ck accordingly invited Stent to write a paper with him for the symposium. On 13 December Delbru¨ck wrote to Stent that he had suggested to the organizers a joint paper on the mechanism of replication. Because he would be in Germany in April, he thought that they should decide sometime in February how to write this paper.75 Stent replied enthusiastically the next day: “Naturally, I am extremely grateful to you for having arranged this for me and I am very excited about the

102

T WISTS AND T URNS

prospect of writing this paper with you. (What, by the way, is the “Mechanism of Duplication,” or don’t we have to know that before writing such a paper?). I think the sooner we plan a preliminary discussion on just how we propose to go about this and what sort of things we want to say, the better.” 76 When he wrote this, Stent probably hoped that his own current experiments would soon have something to say about the mechanism of duplication. With his first graduate student, Clarence Fuerst,77 he was investigating the inactivation of bacteria by 32 P decay in a manner similar to his previous experiments on bacteriophage. Here, too, he found that the survival curves fit the interpretation that the decay of the radioactive phosphorus damages the DNA, in this case in the bacterial “nucleus.” 78 Among the questions he hoped to answer in these experiments were the following: “How are these phosphorus atoms [retained in the DNA of E. coli cells] distributed over the nuclei of daughter cells? Do some descendant nuclei contain only atoms assimilated de novo and are others endowed exclusively with phosphorus atoms of parental origin, or are the atoms of the parental nucleus dispersed among all the nuclei in its line of descendance? The fact that it is the decay of DNA-P32 atoms which is mainly responsible for the death of the bacterial cells offers a method of resolving this question.” 79 If the first possibility were the correct one, then when he inoculated a nonradioactive culture into a medium containing 32P, even though the bacteria would give rise to descendants that incorporated 32 P in their DNA and consequently died, there would still be a class of nonradioactive cells that “would never acquire any radioactivity and hence remain stable.” Conversely, a radioactive culture inoculated into a nonradioactive medium should give rise quickly to a class of cells with stable, nonradioactive nuclei. If the atoms are dispersed, on the other hand, in both of these situations all of the daughter cells will contain some radioactivity.80 By February 1956, when Fuerst and Stent submitted a paper including these experiments to the Journal of General Physiology, their results had substantiated the second of these alternatives. “It, therefore, appears,” they concluded, “that the atomic identity of the bacterial nucleus is not preserved in the course of its reduplication but that parental and newly assimilated DNA-phosphorus atoms become intermingled within daughter nuclei.” 81 This outcome did not characterize the pattern of dispersal sufficiently, however, to enable Stent to

T WISTS AND T URNS

103

answer the question he had asked in his letter to Delbru¨ck: “It is, unfortunately, not possible to infer from the present experiments the mechanism by which this dispersion occurs, i.e., whether it is due to a partition of the parental atoms in the course of the elementary replication act of the DNA itself, such as demanded by some proposals concerning this process (16, 17) or whether it is due to the randomizing effect of some postreplication event, such as ‘crossing-over’ or assortment of ‘chromosomes.’ ” 82 The literature citation “(16, 17)” in this passage referred, respectively, to Watson and Crick’s second Nature article and Delbru¨ck’s paper on the replication problem. The fact that he lumped these papers together as proposals demanding “a partition” indicates that Stent’s results could not discriminate between the mode of replication predicted by the Watson-Crick model and the thoroughgoing dispersal of DNA atoms postulated by Delbru¨ck’s replication scheme. Privately, Stent continued to favor the view that the replication process was not distributive and that the distribution he observed derived from “postreplication” recombination events.83 At the University of Michigan, Cyrus Levinthal was also turning his investigations with 32P labeled bacteriophage toward the question of DNA replication. The speculations offered to date concerning the mechanism by which the DNA molecule is duplicated, he noted, “have not been very much limited by experimental results.” To trace the distribution of DNA-32 P more directly from parental to progeny phage, Levinthal devised a method “using an electron-sensitive photographic emulsion for the measurement of the radioactivity of a single virus particle or a single DNA molecule.” Such emulsions had previously been used by particle physicists. By embedding particles containing 32 P in a layer of emulsion sandwiched between two other layers, he obtained visible “star tracks,” each star representing the disintegrations occurring within a single phage particle. The number of stars therefore indicated the number of particles present that contained 32 P, and the number of tracks radiating from the center of the star indicated the number of 32 P disintegrations that had occurred in each particle. One could use the latter as a measure of the quantity of radioactive DNA contained in that phage particle.84 Levinthal reported the results of such experiments in a paper submitted to PNAS in April 1956. In his introductory discussion he provided a lucid analysis of the replication problem. In an idealized situation one could compare the distribution of 32 P labeled DNA with the

104

T WISTS AND T URNS

Fig. 3.4. Three types of model for DNA replication, as shown by Cyrus Levinthal

predictions of three classes of theory. The three possibilities he illustrated with the same kind of diagram that Delbru¨ck had used in his paper on DNA replication (figure 3.4). Levinthal labeled these three possibilities “template-type replication,” “dispersive replication,” and “complementary type” replication, respectively. He identified the second of these with Delbru¨ck’s scheme and the third with the “results of the proposals of Watson and Crick,” but he identified no one as the author of a template-type scheme. Where Levinthal went beyond Delbru¨ck’s theoretical analysis was in his emphasis on the “extensive genetic recombination during the vegetative phase of [phage] growth.” The problem of genetic recombination in phage had not yet been solved, but “there are experimental data which can distinguish between certain classes of schemes.” Levinthal depicted two models for recombination, one based on the idea of “breakage and reunion of already formed chromosomes,” the other on a replication de novo, in which part of the new molecule is formed “under the control of one parental molecule,” the rest under the control of another parental molecule. (In the literature these two processes were usually denoted, respectively, “crossingover” and “copy-choice.”) The first model predicted a further redistribution of parental molecules, whereas the second “will not disturb

T WISTS AND T URNS

105

the label in the original structure.” In tracer experiments, he concluded, “since in one growth cycle there is extensive growth as well as several mating events, we have the problem of distinguishing among the six possibilities obtainable by combining” the three possible replication processes with the two possible recombination processes.85 In the event, Levinthal’s experimental results fell far short of distinguishing among these six possibilities. In both first- and secondgeneration progeny Levinthal identified individual phage that, calculating from the number of tracks in their stars, he determined must contain about 20 percent of the labeled DNA of a single parental phage particle. Given that the parental DNA must be multiplied by a factor of thirty to fifty in a single-step growth cycle and by a factor of at least one thousand in two growth cycles, his result seemed “to rule out any mechanism of duplication which implies repeated sharing of the atoms of the parental structure between the daughters.” 86 Levinthal introduced a new complication into the situation, however, by a further experiment in which he subjected radioactive phage particles to osmotic shock to release their DNA into solution. “The number of stars produced by the solution,” he reported, “was the same as that produced by the intact phage. However, the size of the stars [that is, the number of tracks in each star] dropped to about 40 percent of those produced by the intact phage.” When he subjected progeny phage to the same treatment, the number of stars remained unchanged, and “the star size also remained unchanged at its value corresponding to about 20 percent of the original labelled phage.” 87 “The experiments indicate,” Levinthal claimed, “that the original phage contained one large piece of DNA with about 40 percent of the P 32 of the phage and a number of small pieces too small to be detected in these experiments.” Levinthal then made the assumption that the large piece of DNA is the genetic structure of the phage. He adduced no further grounds for this assumption, which he called in his summary a suggestion,88 and it seems to rest simply on the commonsense idea that only a large piece of DNA could be a suitable site for genetic markers located along a linear structure. His apparent discovery that “the phage T2 contains in one large piece approximately 40 percent of its DNA” overshadowed the stated purpose of Levinthal’s investigation: to study directly the question of the mechanism by which the DNA that carries genetic information is duplicated. All that he could say on this issue was that “the simplest

106

T WISTS AND T URNS

interpretation of all the data is in terms of the Crick-Watson type of complementary replication.” 89 The data indicated, that is, a high concentration of parental DNA in relatively few progeny, rather than a random distribution. Further on, however, Levinthal acknowledged: The detailed predictions inferred from the Crick-Watson model . . . cannot be checked by these experiments for the following reasons: First, it is not feasible to examine the DNA molecules of the phage after one doubling, but only after a complete growth cycle which yields a burst of approximately one hundred new phages. Second, the inefficiencies of the system which produce the transfer of only 40 per cent of the total parental phosphorus [this was not the 40 percent “piece” that Levinthal had just identified but the 40 percent of parental DNA that Watson and Maaløe and others had earlier found to be transferred from parental to progeny phage] prevent one from observing whether there are really two labeled particles formed from each parental structure. For these reasons, experiments are being undertaken to study the distribution of the label of the DNA extracted from growing bacteria in which it is known that atoms once incorporated in DNA remain.90 In his introduction Levinthal gave reasons why “bacteriophage are particularly useful for investigations of this problem.” 91 In his conclusion he identified factors that limited the phage’s usefulness and moved, as Stent had already done, to bacteria to pursue the problem further. Levinthal’s experience illuminates both the ethos of the phage group and the anomalies often lurking in the way of the drive for simplicity in science. As Frank Stahl’s enthusiastic initiation into phage genetics illustrates, phage biologists adhered to the tenet that these most simple of organisms—so simple that it was often debated whether they should be treated as forms of life or as very large chemical molecules—would be the vehicles through which genetics could finally be understood at a molecular level. Bacteria were viewed as several orders of magnitude more complex and valued therefore by phage biologists less as subjects for genetic analysis than as host organisms for their phage. In exactly the respect that could be crucial to the solution of the problem of DNA replication, however, the “higher organism” could multiply by a process of generation-by-generation duplication that promised to be simpler than the multiple replications, matings, and recombinations that comprised a single growth cycle of bacteriophage. Simplicity appears in many guises, and the forms of it that will be most conducive to the solution of a given biological

T WISTS AND T URNS

107

problem often can emerge only through prolonged investigation or the competition between the diverse viewpoints of individual investigators. One month after Levinthal had submitted his experimental contribution to the DNA replication problem to PNAS, he and H. R. Crane made a significant contribution to the theoretical discussion of the problem in a brief paper titled “On the Unwinding of DNA.” The difficulty that others had felt in accepting the idea that the two strands could unwind lay, Levinthal and Crane believed, in the fact that their rotation “has been visualized as a motion involving some kind of lateral translation of the strands through the fluid, that is, a whipping or flailing motion of the parting strands.” These difficulties had led to the several alternative mechanisms that evaded the problem at the cost of departing in some essential way from the “picture of replication originally advanced by Watson and Crick.” Levinthal and Crane set out to show that “the unwinding problem . . . does not give rise to a valid objection” to the Watson-Crick mechanism.92 Levinthal and Crane visualized the Watson-Crick model as a Y, “in which the vertical part is the parent helix and . . . the two arms are the growing progeny.” If each of these parts rotates on its own axis, while the Y as a whole retains its spatial orientation, the rotations will gradually shorten the double-stranded vertical portion and lengthen the arms. Calculating the energy required to produce this rotation against viscous drag, Levinthal and Crane found that, for a molecule with six thousand turns, it is so small compared to that of the formation of the bonds that the total energy requirement increases only insignificantly. They calculated also that the tangential force created in the helix by the rotation is so small relative to the force constants in the phosphate bonds that the mechanical strength of the helix is “sufficient to withstand the . . . torque without seriously stretching the bonds.” 93 The paper on the mechanism of DNA replication that Max Delbru¨ck invited Gunther Stent to write with him during the spring of 1956 for the McCollum-Pratt Symposium was a natural sequel to Delbru¨ck’s first paper, “On the Replication of Deoxyribonucleic Acid (DNA),” but Delbru¨ck did not write much of it. After he and Stent had discussed what should be in it, he left for Germany and contributed only the section on the topology of coiling. The second paper discussed the same fundamental problem that Delbru¨ck had introduced in the first paper, but in tone and character it differed sharply from

108

T WISTS AND T URNS

the first. The first had been a succinct argument leading to Delbru¨ck’s proposed mechanism. The second was a detailed review of the mechanisms recently proposed, of the experiments intended to trace the distribution of DNA molecules, and of the degree to which these experiments supported one or another of the alternative mechanisms. The first paper was a powerful display of Delbru¨ck’s self-assured style of reasoning and his aesthetic preferences. In the second, Stent strove for balanced assessment and open-mindedness. One critical bias that Delbru¨ck and Stent did not escape was the perspective that was due to their positions within the phage groups. “Our discussion of the mechanism of DNA replication,” they began, “will center on bacteriophages, which have been the object of most of the very few experimental attacks on this problem. We believe it to be not unlikely, furthermore, that experiments on DNA replication more decisive than those which we can discuss at this time will likewise be carried out with bacteriophages.” 94 Delbru¨ck and Stent still viewed the intertwining of the two polynucleotide chains of the DNA duplex as an “obstacle to their separation which must be overcome if such a macromolecule is to act as a template for replication in the manner proposed by Watson and Crick.” They summarized the various proposals that had been made to meet this difficulty, including the transfer-twist scheme of Platt, the DNA-polynucleotide complex of Bloch, Delbru¨ck’s own scheme of separation by breaks and reunions, Watson’s unpublished idea that one of the two DNA chains might be digested enzymatically, leaving the other to serve as a template, and the “ ‘speedometer-cable’ rotation” model of Levinthal and Crane, about which they had learned prior to its publication through a personal communication from Levinthal. Instead of arguing for the Delbru¨ck scheme, however, Stent ended this section of their paper with the bland comment: “The schemes here discussed probably exhaust the simple alternatives for resolving the intertwining of the plectonemic DNA duplex. On a priori grounds we have little reason to choose between them.” 95 What had moved Delbru¨ck, from an earlier discomfort with unwinding the two threads so strong as to prompt his break and reunion scheme, to his current neutrality? According to the recollection of Stent, he had been able, in view of their divergent opinions on the matter, to talk Delbru¨ck into accepting such a neutral position for their paper. Delbru¨ck, who took a light-hearted attitude toward theories, was not so wedded to his aesthetic judgment that unwinding was “too inelegant to be efficient” to

T WISTS AND T URNS

109

insist on maintaining that view.96 Perhaps, in addition, the “encouraging answers” from Levinthal and Crane’s calculation of the mechanical aspects of unwinding had convinced Delbru¨ck that the unwinding obstacle was no longer insurmountable. If it was the news of Levinthal and Crane’s analysis of the problem that neutralized what Delbru¨ck and Stent now referred to mildly as “the chief reason for hesitating to accept the topologically simple solution of unwinding the two threads,” 97 then Delbru¨ck must have experienced this change of heart while he and Stent were planning their paper. Stent coined the terms conservative, semiconservative, and dispersive to denote, respectively, a mechanism entailing no distribution of the parental atoms to progeny molecules, the mechanism predicted by Watson and Crick, and the model proposed by Delbru¨ck.98 The paper of which Delbru¨ck was a co-author now recognized, however, as Levinthal did, that experimental tests of the actual distribution of parental DNA designed to choose between these classes were complicated by genetic recombination, “whose possible effects on this distribution may be superimposed on those inherent in replication.” Unlike Levinthal, who did not really address himself to the question whether these combined effects could be sorted out experimentally, Delbru¨ck and Stent gave a penetrating analysis of the consequences of the situation. Summarizing three currently discussed mechanisms for recombination in phage, which they designated “fragmentive crossing-over,” “non-fragmenting copy-choice,” and “fragmenting copy choice” (see figure 3.5), they noted that only in the second of these can the “integrity of the template remain undisturbed whether a replica being copied along it has switched over to another template or not.” 99 These considerations led Delbru¨ck and Stent to a strong, if sobering, conclusion. The only experimental result “capable of an unambiguous interpretation is one in which the integrity of either the entire parental DNA duplex, or of each of its polynucleotide chains separately, is found to have survived many successive rounds of replication and recombination. In that case, it may be inferred that replication proceeds by either a conservative or semi-conservative mechanism and that recombination, if it involves participation of DNA molecules, is due to non-fragmenting copying-choice. If the results are not this simple, it will be very much harder to draw conclusions.” 100 As they reviewed the experiments that had been reported on the transfer and distribution of DNA, it became clear to Delbru¨ck and Stent that the results so far attained were not simple enough to support

110

T WISTS AND T URNS

Fig. 3.5. Three possible mechanisms for DNA recombination in bacteriophage, as shown by Delbru¨ck and Stent

unambiguous conclusions. If, on one hand, both radiography (Levinthal’s star track) experiments and 32P inactivation experiments seemed to “show that DNA of the parent phage is distributed over its progeny more widely than could be reconciled with conservative or even semiconservative replication in conjunction with non-fragmentary recom-

T WISTS AND T URNS

111

bination,” 101 experiments by Walter Plaut and Daniel Mazia on the chromosomes of plants, on the other, appeared to exclude “the possibility that all the DNA molecules of the chromosome participate equally in replication and do so by either semi-conservative or dispersive mechanisms.” 102 In Delbru¨ck and Stent’s opinion the “big DNA piece” that Levinthal had found in phage only made matters worse: “The realization . . . that the DNA of the parent phage may be bipartite introduces an entirely new element into all the foregoing arguments and very much complicates the interpretation of the distribution experiments.” 103 Ambiguous as the current state of the problem appeared to Delbru¨ck and Stent, it was hardly static. Besides summarizing the experiments of Levinthal and of Stent himself discussed above, Delbru¨ck and Stent could include the latest unpublished work of Stent, as well as new “preliminary” experiments by Hershey. The field was moving, and it was hard to predict what might open up. Delbru¨ck and Stent ended their review with a flourish: “It appears to us that no definite conclusions regarding replication can be drawn at the present moment. Our review can do no more than delineate the problem and assess the evidence. We hope that it will help to clarify the issues and to speed the answers. We are confident, in fact, that the whole issue will be resolved before many a day. Who knows, perhaps even before this goes to press?” 104 Following the presentation of Delbru¨ck and Stent’s long paper at the McCollum-Pratt Symposium in Baltimore in June, Cyrus Levinthal gave a short talk that offered some additional evidence for the genetic role of the “big piece” of phage DNA that he had identified. Although his attempt “to replace the phrase ‘the big piece’ by the word ‘chromosome’ ” 105 apparently failed, Levinthal’s big DNA piece attracted, to judge from the edited proceedings, more attention than Delbru¨ck and Stent’s assessment of the replication problem.106 Privately Delbru¨ck was optimistic, not only that the DNA replication problem would soon be resolved despite all of the complexities to which he and Stent had drawn attention, but also that his literary collaborator would be central to that outcome. In August 1956, Delbru¨ck wrote to Wendell Stanley in support of a recommendation that Stent be promoted to associate professor. Amidst a generally laudatory assessment of his achievements and character, Delbru¨ck summarized Stent’s research on phage reproduction with the radioactive tracer 32P. “Dr. Stent’s contributions to the problem of DNA replication are,” Del-

112

T WISTS AND T URNS

bru¨ck asserted, “of paramount importance. Although the problem is still unsolved, it is certain that Dr. Stent’s past, current and future contributions will be decisive to its solution. He has displayed extraordinary skill, imagination, and energy in devising ever-new experimental approaches to it.” 107 It is self-evident why Delbru¨ck picked his younger associate as the person most likely to succeed with the replication problem. Stent not only shared Delbru¨ck’s assessment of the theoretical significance of the problem but deployed the kind of tracer method that Delbru¨ck had suggested in his original analysis of the problem. Moreover, his main object of study was the very bacteriophages that had long been Delbru¨ck’s favorite target for an experimental attack on such a problem.

V James Watson and Francis Crick came from Cambridge in June 1956 to participate in the McCollum-Pratt Symposium. The paper by Delbru¨ck and Stent indicated to Watson that Delbru¨ck was still overly anxious about the “untwiddling problem.” The idea of Levinthal and Crane that was circulating there—that strands of DNA could unwind by rotating like a bicycle cable—was perhaps a welcome sign to Watson that this problem would soon become less worrisome.108 Watson was more detached from that problem now, perhaps in part because of the distance between him and Caltech, but also because his attention for the past nine months had again been dominated by the structure of RNA. Late in the summer of 1955 Severo Ochoa announced that he and his associates had synthesized artificial polyribonucleotides containing sequences of one, two, or four kinds of bases. “Poly-A” was a pure polymer of adenylic ribonucleotides. “Poly-UA” was a mixed polymer of adenylic and uridylic ribonucleotides. For Watson this news “transformed” the question “of whether it would be possible to establish the RNA structure by X-ray techniques.” 109 At the National Institutes of Health, Alex Rich quickly established that the X-ray diffraction pattern of poly-A showed “similarities” to that of natural RNA, whereas that of poly-AU was “identical” to the natural RNA pattern. Moreover, the reflections in the photographs for the artificial polynucleotides were so much sharper than the diffuse bands that Watson and Rich had obtained from natural RNA back at Caltech in 1954 that they offered a much firmer basis for identifying the molecular structure.110

T WISTS AND T URNS

113

Watson left for Cambridge that fall with his enthusiasm for the problem renewed by this unexpected turn of events and his spirits rejuvenated by the prospect of his return to the Cavendish Laboratory. Except for the fact that some of the people whom he had known there were gone, and new people had arrived, everything felt just as when he had been there before. He had no sense of an anticlimactic revisit to the scene of past triumphs. He shared the same office with Francis Crick, now joined by Sydney Brenner, and again engaged in the avid conversations he had missed so sorely during his exile in Pasadena. It was hardly surprising that Watson and Crick again began to work together to see if they could finally bring the elusive RNA molecule to order.111 Watson spent much of his time taking X-ray pictures of poly-A. Rich came to the Cavendish during that year and concentrated on poly-AU. The data quickly suggested to them a helical structure. The density of the molecules and the dimensions of the unit cell were compatible only with a two-chain model. They tried to build models in which the two chains lay side by side without coiling around each ˚ reflection deother but could not devise one that gave the 15.6 A manded by the X-ray patterns. By the time Watson arrived in Baltimore, however, he could announce at the McCollum-Pratt Symposium their success in constructing “an intertwined chain model which appears more satisfactory, both from a-priori structural grounds and from considerations of the x-ray diffraction pattern.” The dimensions were different from those of the DNA double helix, but otherwise the new RNA model was quite similar to it. The two sugar-phosphate backbones were on the outside, with the adenine bases pointing inward. Watson and Crick could incorporate only one hydrogen bond into the pairing of each adenine base with its own kind on the opposite chain, but in every other respect the fit appeared to be almost as good as in the DNA structure.112 One reason Watson and Crick constructed an RNA model so similar to the DNA double helix was that Watson was no longer attempting, as he had with Alex Rich two years earlier in Pasadena, to build into the surface of the structure cavities into which the various amino acids could be fitted. That earlier effort had been made in response to Gamow’s coding scheme. In the meantime, however, Crick had radically redefined the coding problem. Breaking away from the assumption that the amino acids coded for by a sequence of nucleotide bases must somehow attach directly to the DNA molecule or an RNA copy of

114

T WISTS AND T URNS

it, Crick visualized each amino acid becoming attached to a specific small “adapter,” the other end of which formed hydrogen bonds with the RNA bases. Watson was not immediately convinced that Crick was right. Nevertheless, when he got back to the Cavendish and Crick told him they would not find any cavities in the RNA molecule, the task of finding a satisfactory structure seemed much simpler.113 Presenting these results at the symposium, Watson was both optimistic and prudent. “Though this model appears quite promising,” he said, “we must emphasize that our work is still in a preliminary stage. We have not yet made a detailed comparison between the predicted and observed x-ray intensities. These calculations are now in progress. The preliminary results seem encouraging.” 114 Watson also cautioned that “it is difficult to comment on RNA until the structure of poly-A is established.” He listed several ways in which the X-ray patterns were so similar as to lead him and Crick to believe that “if the structure of poly-A is a two-stranded hydrogen-bonded helix, it seems most likely that RNA will also be a two-stranded intertwined helix.” That did not mean that all the pieces seemed ready to fall into place: “We do not see, however, how such a structure could form regular hydrogen bonds. Not only do the purines and pyrimidines have different sizes, but in addition we must postulate a correspondence in sequence between opposing chains. It is thus possible that the RNA structure is very irregular and that the observed configuration results largely from the tendency of the backbone to assume a configuration similar to that of DNA.” 115 We can highlight, more clearly than Watson did here, the difference between this situation and the circumstances that had led him and Crick three years earlier to the structure of DNA. After he and Crick had decided to build models with the bases inside, he had tried on paper to depict configurations in which each of the four kinds of bases was paired by hydrogen bonds to a “like” base on the other chain. He had run into the obstacle, however, that the different dimensions of the pyrimidines and purines made it impossible to fit both types of base pair into a helix of uniform outside diameter.116 It was when he perceived that the two possible pyrimidine-purine pairs both had the same overall length, and could be fit precisely into a uniform two-stranded backbone, that “complementary base pairs” and the solution to the DNA structure simultaneously emerged. Poly-A had, however, only one kind of base, dictating a structure incorporating the very “like-with-like base pairs” that Watson had rejected for DNA.

T WISTS AND T URNS

115

There was no problem about an irregular backbone, since all of the pairs were alike. Even if an acceptable structure for poly-A could be found, however, it would be difficult to adapt the solution to natural RNA, which did contain two pyrimidines and two purines. In a discussion session Francis Crick further clarified Watson’s explanation of their ideas about RNA. They believed that “perhaps RNA consists of two parallel intertwined helical chains, with occasional base pairing. However, . . . the present x-ray data certainly do not prove that such a structure exists, and the quality of the pictures suggests a . . . disordered model . . . , but that may be due to the way our RNA specimens have been treated.” 117 The possibility that there was only “occasional base pairing” in the RNA structure was, from Watson and Crick’s viewpoint, not necessarily a drawback. Except in the case of certain viruses, which contained no DNA, RNA was not expected to be a self-replicating structure. Perhaps the less-ordered state of RNA would somehow be found to be functionally related to its biological role as intermediary between DNA and proteins. During a year that was for Watson and Crick a kind of sequel to the years in which they had produced the double helix of DNA, they had pursued their attack on RNA in much the same style, employing similar techniques and arguments, and had attained resonant echoes of their earlier success. But history was not about to repeat itself at the Cavendish. The beauty of the DNA double helix had been that its structure immediately illuminated its function. Hoping similarly to find a structure for RNA that would explain its function, Watson and Crick did not know how exceptional in this respect DNA is. The moral—as John Cairns has pointed out—is that “the structure of important molecules does not necessarily tell you anything about their function.” 118 While Watson and Crick were foundering in their attempts to duplicate with RNA their model-building triumph with DNA, Waldo Cohn and Alexander Todd were elucidating the mode of linkage of the nucleotides in RNA through the classical methods of biochemistry.119 Watson spent much of the summer of 1956 at Cold Spring Harbor. He was able to revive his interest in the DNA replication problem sufficiently to ask Dan Koshland to conduct some experiments to find out whether the phosphate triesters required by the ribbon models of DNA he had built at Caltech eighteen months earlier really existed.120

C HAPTER F OUR

Crossing Fields: Chemical Bonds to Biological Mutants

I “At the end of the summer,” Matt Meselson wrote to Jim Watson a little later, “I began to look for an entry point into biology by reading about the structure and chemistry of the purines and pyrimidines.” 1 Intending to study the subject systematically, he checked out a workbook, serial number 786, from the Gates and Crellin Laboratories of Chemistry. Meselson used the workbook as a repository for references to relevant books and articles and for summaries of the portions of what he read that were most pertinent to his interests.2 Sometime early in August 1956, Meselson began to read intensely in the library on the first floor of the Crellin chemistry building. The library had decorative wood paneling and a very high ceiling. During the hot summer days it remained relatively cool and free from the irritating effects of smog. Because the journals could not be taken out, they were easy to find there. Meselson enjoyed the appearance and even the smell of this library, and he was soon spending most of his working time there.3 Before he could even approach the design of the replication experiment of which he dreamed, Meselson felt that he must understand in detail the structure and chemistry of the deoxyribonucleic acid molecule. This was not because the success of the experiment would necessarily depend on such knowledge but because it was part of the culture of Pauling’s chemistry department that one would not be taken seriously until one had mastered the structure of any molecule in which one was interested. Feeling totally ignorant of the structures and properties of the purines and pyrimidines that comprised the inner structure of the double helix, Meselson set out to acquire a basic literacy in the subject.4

C ROSSING F IELDS

117

Fig. 4.1. Resonance forms of unsubstituted purine and pyrimidine molecules, in Meselson workbook

As a logical starting point from which to enter a field in which he had not previously specialized, Meselson opened the first volume of the recently published and authoritative The Nucleic Acids: Chemistry and Biology, edited by Erwin Chargaff and J. N. Davidson. Here Meselson could hope to find, as its preface claimed, “all the information presently available” about nucleic acids, collected “into a single comprehensive work.” 5 He turned to Chapter 3, by Aaron Bendich, on the chemistry of purines and pyrimidines. Meselson focused on what he could find out about the tautomeric forms so critical to the pairing of the bases in the Watson-Crick model. He took no notes on Bendich’s brief, general discussion of tautomerism,6 but his reading apparently stimulated him to write out for himself the potential resonance and tautomeric forms of the particular derivatives of purine and pyrimidine that enter into the composition of DNA and RNA (figure 4.1).7 In these drawings Meselson first laid out the two resonance forms for the unsubstituted purine and pyrimidine molecule—that is, the alternate positions in which the double bonds in the heterocyclic rings are represented. Then, for each of the relevant derivatives, he put down all of the possible “mesomeric tautomers” that could be produced by shifts of a proton between the oxygen atoms and a nitrogen

118

C ROSSING F IELDS

atom contained either in the purine or pyrimidine ring itself, or on an amide side group. In this basic exercise he was preparing himself for the large task that lay ahead: to learn which of these numerous paper possibilities might actually exist and which were pertinent to their incorporation into DNA or RNA. To find out more about these structures, Meselson turned to Chapter 13 of The Nucleic Acids, a review by D. O. Jordan of their physical properties. The first thing Meselson learned was that his path to a comprehensive understanding of the problem would not be short. “The full details of the structures of . . . the pyrimidines known to occur in nucleic acids,” Jordan commented, “have not been established by x-ray diffraction. Some conclusions may be drawn concerning their structures, however, from those of some related pyrimidine derivatives which have been studied.” Jordan summarized these studies, reproducing diagrammatic structures that showed the bond lengths and angles. Meselson copied down thirteen of the fourteen references cited for this section, setting up for himself the next phase of his reading program.8 One point in Jordan’s discussion that may have struck Meselson was a difference of opinion over whether a shift in the absorption spectrum of pyrimidines with increasing pH was due to enolization (that is, a tautomeric shift) or to ionization. Jordan’s treatment of the purines indicated that the structure of the two contained in DNA—adenine and guanine—had been studied directly by X-ray diffraction and that their bond angles and distances had been established. For guanine, however, the positions of the hydrogen atoms were not known with sufficient precision to decide which of the “four possible tautomeric forms . . . is the correct structure.” 9 Already Meselson must have realized that it would not be easy to find out exactly what the structures of the bases in DNA were. Before launching into the research papers that would treat more fully the evidence for the structures of purines and pyrimidines in general, Meselson wrote down the “Watson-Crick Pairing” scheme for the DNA double helix (figure 4.2).10 His drawing resembles, but is not a direct copy of, the published base pairs that had appeared in the original Watson-Crick papers and were also reproduced in the Jordan chapter he had just read (the bond lengths and angles here, as we shall see, Meselson added in pencil sometime afterward). Meselson was going through another orientation exercise, writing out the pairing scheme to check his knowledge of the problem. A special point of interest is that his drawing shows three hydrogen bonds between gua-

C ROSSING F IELDS

119

Fig. 4.2. Base pairing for Watson-Crick DNA double helix, in Meselson workbook

nine and cytosine, whereas previously published representations had followed Watson and Crick in postulating only two. Linus Pauling had just written with Robert Corey a paper arguing that guanine and cytosine “should form three hydrogen bonds.” Meselson had not yet seen this paper, which was still in press. There was, however, always an intense interest among Pauling’s students in what he was doing. That Meselson put three bonds in his drawing indicates that he must have learned about Pauling’s idea either during one of the weekly seminars or from conversations in the laboratory.11 The inclusion of representations of the resonance forms, a feature ordinarily omitted in the published depictions of the Watson-Crick pairing, suggests that Meselson was pondering the general relation between these forms and the hydrogen bonds central to the Watson-Crick structure and its mechanisms of replication and mutation. On the next page of his notebook Meselson continued his concentration on the resonance forms by writing them out for both the enol and the keto form of 2-hydroxypyrimidine (figure 4.3).12 Returning again to the Bendich chapter, Meselson copied from it a list of the physical properties of purines and pyrimidines. He compiled his list of the properties of twelve different bases from two of Bendich’s tables, the second consisting of dissociation constants. From the latter Meselson picked out particularly the titrimetric pKa1 and pKa2 values: that is, the first and second ionization constants for the respective bases. Afterward he added similar data from other sources. His special interest in these values, as well as in the enol and keto forms, suggests that Meselson was already moving beyond gen-

120

C ROSSING F IELDS

Fig. 4.3. Enol and keto forms of 2-hydroxypyrimidine, in Meselson workbook

eral familiarization with purines and pyrimidines and beginning to focus his attention on specific questions. One of these was, what were the correct structures of the bases in DNA? The other was the relation between these structures and a proposal that Watson and Crick had made about the molecular mechanism of genetic mutation.13 In his presentation at the Cold Spring Harbor Symposium in June 1953, Watson had outlined a hypothetical scheme, based on the structural features of the purines and pyrimidines incorporated into the complementary base pairs of the double helix, that could account for a change in the sequence of base pairs in the nucleotide chains. Under the heading “A Possible Mechanism for Natural Mutation,” Watson stated: In our duplication scheme, the specificity of replication is achieved by means of specific pairing between purine and pyrimidine bases; adenine with thymine, and guanine with one of the cytosines. This specificity results from our assumption that each of the bases possesses one tautomeric form which is very much more stable than any of the other possibilities. The fact that a compound is tautomeric, however, means that the hydrogen atoms can occasionally change their locations. It seems plausible to us that a spontaneous mutation, which as implied earlier we imagine to be a change in the sequence of bases, is due to a base occurring very occasionally in one of the less likely tautomeric forms, at the moment when the complementary chain is being formed. For example, while adenine will normally pair with thymine, if there is a tautomeric shift of one of its hydrogen atoms it can pair with cytosine [figure 4.4]. The next time pairing occurs, the adenine (having resumed its more usual tautomeric form) will pair with thymine, but the cytosine

C ROSSING F IELDS

121

Fig. 4.4. Effect of tautomeric shift on base pairing, as proposed by Watson and Crick

will pair with guanine, and so a change in the sequence of bases will have occurred.14 Meselson never questioned the basic features of Watson and Crick’s general base-pairing scheme,15 but he was aware that they had not examined deeply the chemistry of their mutagenesis scheme, and he hoped that if he were to think hard enough about the structures, in the style of the Pauling laboratory, he might be able to see further into the mechanism. At some point he noticed an original way to view the process of mispairing. As figure 4.4 indicates, Watson and Crick had considered only the case of adenine mispairing with cytosine. Here it was most natural to supply the hydrogen at the nitrogen atom (denoted N 1 in the British system of nomenclature) necessary for adenine to form a hydrogen bond with cytosine by postulating a tauto-

122

C ROSSING F IELDS

meric shift. If one were to focus instead on thymine and guanine, however, Meselson saw, “the whole trick” was to get rid of the hydrogen on N 1 of thymine or guanine. So long as both molecules held onto their hydrogen atoms, it would be impossible for them to come close enough together to fit into the double helix, a conclusion he verified by constructing rubber models of the bases. A tautomeric shift of the hydrogen from the N 1 of either guanine or thymine to the oxygen attached to C 6 would solve the problem nicely, but, Meselson supposed, so would a simple ionization of the hydrogen bond. Thanks to Pauling’s idea that three hydrogen bonds formed between guanine and cytosine, guanine and thymine could still form two hydrogen bonds, even on the assumption that the hydrogen removed from N 1 on one of these bases by ionization was lost. When he looked up the pK values in Bendich’s chapter, Meselson was delighted to find that there was “an appreciable likelihood” that thymine is ionized at physiological pH. Although he had no way to decide whether tautomerism or ionization was the more probable mechanism, he liked the ionization alternative because it was his idea.16 As he continued to study Bendich’s chapter on the chemistry of purines and pyrimidines, a section dealing with alterations of the structure of naturally occurring purines and pyrimidines that had “yielded compounds of value in biology and medicine” caught Meselson’s attention. Most of these compounds inhibited the growth of microorganisms. The cases that most interested him were reports of the apparent incorporation of the inhibitory analog into the nucleic acids of the organism affected. Bendich mentioned four such bases: 8-azaguanine, 8-azaguanylic acid, 5-bromouracil, and 2, 6-diaminopurine.17 In his summary of this information, Meselson wrote down the following: 8-Azaguanine incorporates into L. geleii, TMV and mouse nucleic acid. 5-Bromouracil incorporates into S. faecalis.18 These molecules were analogs of the two bases in DNA—guanine and thymine—whose ionization he had been thinking about as a possible cause of mispairing. It may well have occurred to him at this point that by substituting such altered bases for the normal ones in DNA he could change the base-pairing stabilities, thereby inducing mutations.19 Sometime afterward he added the first ionization con-

C ROSSING F IELDS

123

stant of 5-bromouracil to his table of the physical properties of purines and pyrimidines. Such ideas were not yet emphatic enough to redirect the next steps in his study plan, but they alerted him to a general possibility that he kept in mind as he pursued his reading.

II By now Meselson was already refining his broad initial problem— What are the correct structures of the bases in DNA?—into several more concrete questions: What are the dominant tautomeric forms of the bases in DNA? Are these forms semipermanent? What forms occur only occasionally, serving possibly as sources of mutation? To what degree might thymine and guanine be ionized in the DNA molecule? He quickly realized, however, that too little was known about the detailed structures of the specific bases paired in the double helix, particularly about the location of their hydrogen atoms, to reason from information available about these bases alone. It would be necessary to survey the evidence obtainable from current knowledge of related compounds—not only of other derivatives of purine and pyrimidine, but even of other heterocyclic molecules analogous in configuration. In this way he might be able to identify some general relationships between the existence of the alternative tautomeric forms and the positions of the tautomeric groups on ring structures, their dependence on the character and position of other substituent groups, and on the properties of the medium. He could then apply such generalizations by analogy to thymine and guanine, or adenine and cytosine. A further difficulty was that even data specific to one or the other of the four bases of the DNA molecule might not apply to their state within the molecule in solution. Meselson worried particularly, for example, about whether the pKa values published for the free bases might differ significantly from the ionization constants for these bases in DNA itself.20 The footnotes in Bendich’s review led Meselson to some suggestive clues in a series of papers from a group of chemists, led by Adrien Albert at the Australian National University, on the absorption spectra, ionization constants, and other properties of heterocyclic compounds. A 1952 study of pteridines (compounds which, like the purines, contain two condensed heterocyclic rings but in which both rings are 6-membered, whereas in purines one ring is 6- and the other 5-membered) reported that when 6-hydroxypteridine was titrated, it produced a hysteresis loop. That is, running alkali into an aqueous

124

C ROSSING F IELDS

Fig. 4.5. Keto and enol forms of 4-hydroxyacridine, as shown by Adrien Albert and L. N. Short

solution of the compound “gave pH values which are much higher than those found on back-titration with acid.” On the grounds that xanthopterin, which was thought to undergo slow tautomerism in alkaline solution, displayed a similar hysteresis loop, Albert and his group suggested that 6-hydroxypteridine underwent a tautomeric shift between an enol and a keto form. Meselson noted the hysteresis effect and jotted down the two formulas.21 In an earlier paper by Albert and L. N. Short on the absorption spectra of acridines, Meselson picked out another apparent instance of the detectable occurrence of two tautomeric forms. When placed, respectively, in absolute and dilute alcohol, 4-hydroxyacridine showed shifts in the ultraviolet and visible spectra that suggested that in the former solution it existed in a pure enol form, in the latter in a keto form (figure 4.5). This response fit with other evidence that a rise in dielectric constant “stabilizes the keto form because the latter is more polar than the enolic form.” Again copying the two forms in question into his workbook, Meselson summarized the situation: “Albert . . . finds separate existence. Keto forms more polar than enol forms.” 22 Remote as acridine and hydroxypteridine were from the structure of the bases in the double helix, Meselson found Albert’s evidence for the stable, independent existence of both enol and keto forms of these molecules reassuring. If enol and keto forms were typically interchanged so rapidly that one could not specify in which form the molecules spend most of their time under given conditions, then the question of which was the correct form in the DNA bases would lose its meaning, and the Watson-Crick mutation hypothesis would simply collapse. Albert’s finding that in these cases the two forms exist separately, that changes in conditions could make one or the other the preferred form, provided, therefore, what Meselson needed to sustain his search.23

C ROSSING F IELDS

125

Next Meselson came across what seemed to him a “very nice review of [the] chemistry of simple pyrimidines” by one of Albert’s associates, D. J. Brown. Brown emphasized the need to pay special attention to “the synthesis and chemistry of pyrimidines as exemplified in simple pyrimidines,” but Meselson fixed his attention on the four paragraphs in which Brown briefly discussed “the potential tautomerism of hydroxy, mercapto and amino groups in the 2- and 4- positions.” From comparisons of the ultraviolet spectra of pyrimidines containing these groups with the spectra of corresponding methoxy and methylmercapto derivatives and other related compounds, Brown concluded that in all cases they existed in aqueous solution in their keto forms. The method was also extended to 2- and 4- amino pyrimidines, which “were shown to exist in the ⫺NH 2 form.” 24 In his workbook Meselson further compressed Brown’s succinct summaries into synoptic sentences that included sketches of the pertinent molecular structures.25 Reading a paper by Robert Sinsheimer and Ruth Hastings titled “A Reversible Photochemical Alteration of Uracil,” Meselson summarized their experiments in one sentence: “Uracil u.v. spectrum changes on irradiation in 230–80 region but restoration is effected by acidification.” 26 The reversible decomposition reactions they discussed suggested to him nothing relevant to his own interest in tautomeric changes. As a student of Pauling who had just completed his own analysis of the crystal structure of an organic molecule, Meselson could hardly ponder the structure of purines and pyrimidines for long without turning to the crystallographic evidence for the structural details with which any base pairing or replication scheme must be compatible. Having written down the references given by Bendich to the literature in this field, Meselson was now ready to begin reading the papers cited. First he looked up a preliminary version of “Crystal Structures of 2-Amino-4-Methyl-6-Chloropyrimidine and 2-Amino-4, 6-Dichloropyrimidine,” by C. J. B. Clews and W. Cochran, published in Nature in 1947. The paper came from the Cavendish Laboratory in Cambridge, famed for its long tradition of X-ray crystallography, and it claimed to represent the first “structural work on pyrimidines” to be “reported in the literature.” 27 Before looking up the subsequent paper by these authors that did contain their data on bond lengths and angles, Meselson pulled out Pauling’s classic The Nature of the Chemical Bond and copied from it a set of Van der Waals radii for the atoms commonly

126

C ROSSING F IELDS

found in organic compounds (and for the methyl group). Returning to the paper “Fundamental Dimensions of the Polypeptide Chains,” by Pauling and Corey, which had been one of the prime references for his thesis work on N,N′-dimethylmalonamide, Meselson took down, from a long table showing the dimensions of NEH ⋅⋅⋅ O hydrogen bonds found in amino acids and peptides, the “most probable value ˚ ), and the bond angle for ∠CEN ⋅⋅⋅ O (110° for that bond” (2.79 ⫾ .12A ⫾ 15°). From another source he added the angle for the HEOEH bond and the distances for the CECl and CEN bonds.28 Meselson selected these data from sources familiar to him to serve as a handy and concise point of reference for what he intended to read next. Although such bond lengths and angles varied from compound to compound, the general types were sufficiently characteristic that dimensions gleaned from the structure of the peptides with which he had prior experience could help him evaluate the published results he expected to find concerning the class of molecules less well known to him. Van der Waals radii were useful to keep in mind in assessing the arrangements of the molecules in crystals, because they assigned optimal separations between neighboring molecules. The inclusion of the dimensions of the NEH ⋅⋅⋅ O hydrogen bond suggests that Meselson intended to be particularly attentive to whatever evidence the crystallographic literature could offer about the purine and pyrimidine hydrogen bonds relevant to base pairing. In their second paper on 2-amino-4-methyl-6 chloropyrimidine and 2-amino-4, 6-dichloro-pyrimidine, Clews and Cochran explained that they had chosen these two compounds to apply X-ray crystallography for the first time to the pyrimidines because they “seemed likely to permit of a direct determination of structure without the necessity for any assumptions as to the nature of the pyrimidine molecule” (figure 4.6, shown here by the authors without the double bonds of the aromatic rings). This “pair of isomorphous pyrimidines” 29 was well suited to their purpose because of the chlorine substituents they contained. Such “heavy atoms” enabled one to apply to best advantage the Patterson method, in which the distribution of electron densities calculated from the X-ray diffraction pattern was used to establish the vectors connecting scattering centers. The intensities of these peaks being proportional to the square of the number of electrons in the atoms, the direction and distance between two large atoms such as chlorine or chlorine and nitrogen stood out clearly. Once positions in the unit cell were assigned to these atoms, it was much easier to devise

C ROSSING F IELDS

127

Fig. 4.6. 2-amino-4-methyl-6-chloropyrimidine and 2-amino-4,6-dichloro-pyrimidine, as shown by C. J. B. Clews and A. Cochrane

an approximate structure for the whole molecule that could be further refined through successive Fourier syntheses. Meselson devoted a full page to notes on this paper, summarizing succinctly the methods employed, the dimensions determined, and the structures proposed. He commented approvingly, “good looking structure factors.” 30 The paper appeared simple and straightforward to him, partly because the methods it deployed were those with which he felt at home through practicing the same craft, partly because the techniques applied in 1947 appeared already less refined than those that had come into use by the time he took up his own crystallographic project five years later.31 Clews and Cochran, for example, gave atomic coordinates only for the atoms in the pyrimidine ring and the heavy substituents. The scattering power of the hydrogen atoms, each of which contained only a single electron, was too weak to provide discrete peaks on the electron density maps that they constructed. Moving on to a paper published by Clews and Cochran in 1948 on two other simple pyrimidine derivatives, 2-amino-4,6 dichloropyrimidine and 5-bromo-4,6-diaminopyrimidine, Meselson learned that the two British crystallographers had refined the detection of small variations in electron density far enough to locate tentatively the positions of the hydrogen atoms in the molecule, a goal thought “beyond the power of the x-ray method because of the small scattering power of its single electron.” To do so they had, following the recommendation of A. D. Booth, carried out a three-dimensional Fourier synthesis in place of the usual two-dimensional one.32 In their introduction Clews and Cochran stressed their hope that knowledge of the structures of the pyrimidines attained by X-ray methods could answer “the question as to which of the tautomeric forms of pyrimidine derivatives [in which a hydrogen atom can shift from one atom of the ring or of a substituent group to another] will be most prominent under normal conditions of chemical reaction.” In this connection they brought up the view of Louis Hunter, who be-

128

C ROSSING F IELDS

Fig. 4.7. Resonance structures of aminopyrimidine molecule, in Meselson workbook

lieved that in such cases there formed a resonance hybrid, with properties intermediate between the two extreme forms. The hybrid possessed the properties of both forms, “although neither of these tautomers has a real existence.” Hunter called this idea “the theory of mesohydric tautomerism.” Clews and Cochran provocatively summarized Hunter’s view as suggesting that “distinction between the tautomeric forms is meaningless.” They themselves thought it unlikely that a hydrogen atom could resonate between positions on an amino nitrogen and a ring nitrogen, and favored the interpretation that the molecules they studied were in the aminopyrimidine form (rather than an iminodyhydropyrimidine form). The measured bond lengths accorded most closely with a distribution of resonance structures between two polar and two nonpolar forms of aminopyrimidine.33 In his notes on this paper, Meselson again summarized the crystallographic methods and data, copying out the structures with their bond lengths and angles. He paid special attention to Clews and Cochran’s discussion of the hydrogen bonds and tautomeric forms. Reproducing the four (of five potential) resonance structures that they thought contributed significantly to the state of the three aminopyrimidine molecules, he noted: “good agreement with the observed dimensions is found for the hybrid” (figure 4.7). Meselson’s sketch of 4-amino-2, 6-dichloropyrimidine—a composite of Clews and Cochran’s drawing of the dimensions of this molecule and of their drawing of the molecule within its crystal structure— selectively featured the lengths of the hydrogen bonds connecting the molecule with its neighbors.34 His simplified sketches of the three aminopyrimidine molecules again drew attention to the intermolecular hydrogen bonds (dotted lines), seen in figure 4.8. Paraphrasing Clews and Cochran’s discussion of these bonds, Meselson wrote beneath the figures: In the case of each of the above one amino group forms one H bond to each of two ring N in adjacent atoms. . . . In (c) one NH 2 group is related to that in another molec. by the screw axis and the NEN

C ROSSING F IELDS

129

Fig. 4.8. Aminopyrimidine molecules, in Meselson workbook

˚ while the NEN distances for the other NH2 group distance is 3.4 A are 3.02 and 2.96. The H bonds in all cases lie close to the plane of the molec. to which the ENH 2 group belongs. The CENH 2 bond ˚ long for the non H bonding NH 2 and 1.32 for the H bondis 1.40 A ing one.35 This information concerning variations in the lengths of intermolecular hydrogen bonds and their influence on the bond lengths of neighboring atoms especially interested Meselson, because the presence of a hydrogen atom changed the bond lengths of adjacent atomic pairs. From the length of a covalent bond, therefore, one might be able to infer a hydrogen bond that could not be directly detected by X-ray crystallographic analysis.36 His list of crystallographic references led Meselson next to a paper titled “The Crystal Structure of 4, 6-dimethyl-2-hydroxypyrimidine” by G. J. Pitt, of Birkbeck College, London. Using two-dimensional Patterson projections and Fourier syntheses, Pitt had encountered obstacles and could present only tentative values for the bond lengths. He claimed, however, to be able to locate the molecules of the water of crystallization between those of pyrimidine and asserted that the latter were linked to one another through hydrogen “bridges” to the water molecules. These intermolecular linkages, he concluded, “seem to provide good evidence in favor of the theory of mesohydric tautomerism given by Hunter.” 37 In a summary of Pitt’s data and methods much briefer than he had made for the Clews and Cochran papers, Meselson remarked, “Only extracyclic atoms are resolved, and these only” in one projection. By the standards of Pauling’s laboratory, where “they ran crystal structures into the ground,” this paper appeared to Meselson a disgracefully incomplete analysis. He was skeptical also of the whole idea that the pyrimidine molecules in the crystal could be held together by water.38 Here, as in the previous paper, Meselson encountered Hunter’s theory of mesohydric tautomerism, and this time he responded to it in

130

C ROSSING F IELDS

Fig. 4.9. 4,6-dimethyl-2-hydroxypyrimidine, showing hydrogen bridges to water, in Meselson workbook

his notebook. First he sketched out a portion of Pitt’s scheme with molecules of water connected to the pyrimidines by hydrogen bridges with the dimensions Pitt had assigned to them (figure 4.9, left; only the portions of the pyrimidines adjacent to these bridges are shown). Then he drew a similar sketch (figure 4.9, right), except that he showed the links as hydrogen bonds, with the usual solid and dotted lines to indicate, respectively, the covalent and the electrostatic bond. Beside the second sketch he wrote, “Pitt sees evidence for Hunter’s theory, but I don’t see why conventional H bond scheme as shown shouldn’t suffice.” 39 Meselson resisted Hunter’s theory, because the existence in some compounds of tautomeric forms stable enough to isolate seemed to refute Hunter’s idea that the tautomers do not exist. Moreover, the whole idea that a bulky hydrogen atom could resonate rapidly back and forth between its relatively strong bonds with nitrogen or carbon atoms seemed distasteful to him.40 Because his reading was drawing him more and more to think about the hydrogen bonds linking pyrimidine molecules, Meselson turned to a review of the general character of hydrogen bonds in organic crystals published in 1952 by Jerry Donohue. A postdoctoral fellow of Pauling and a leading expert on this subject, Donohue had alerted Jim Watson at Cambridge, in that same year, that in his first attempts at base pairing he had assumed the wrong tautomeric forms for guanine and thymine.41 Donohue began his paper by warning that X-ray crystallographers often produced results too inaccurate or incomplete to be of value in a discussion of the hydrogen bond. There were, however, “a large number of careful determinations,” and it was on data derived from such studies that he based his “deductions and generalizations about hydrogen bonds.” 42 Donohue discussed each of the five important types of hydrogen

C ROSSING F IELDS

131

Fig. 4.10. Tautomeric forms of adenine, as represented by Jerry Donohue

bonds: NEH ⋅⋅⋅ N, NEH ⋅⋅⋅ Cl, NEH ⋅⋅⋅ O, OEH ⋅⋅⋅ N, and OEH ⋅⋅⋅ O. For each type he provided tables listing, for the molecules for which he believed reliable results had been obtained, the bond lengths (between the two atoms linked by the hydrogen bridge) and angles (between the two atoms so linked and an adjacent atom on the covalent side of the hydrogen bond). He showed how, even in cases in which the crystallographic data did not suffice to specify the positions of the hydrogen atoms, one could sometimes draw conclusions from the intermolecular hydrogen bonding scheme in the crystal. Using as one of his examples a molecule particularly relevant to base pairing, adenine hydrochloride, Donohue was able to choose between the two tautomeric forms in which the adenine molecule was usually represented (figure 4.10) by ruling out certain of the possible positions for hydrogen bonds on the grounds that the intermolecular distances and angles in the crystal were unfavorable for such a bond. Through such reasoning Donohue “assigned” structure B to adenine.43 Meselson copied out extensive portions of Donohue’s tables of the bond lengths and angles for each of the five hydrogen bond types. (In some instances he condensed by giving aggregate limits instead of the individual values.)44 He filled up the rest of the page with a table of covalent radii, Pauling electronegativities, and van der Waals radii copied from J. M. Robertson’s Organic Crystals and Molecules. This was a standard treatise on the methods used in the X-ray crystallographic study of organic molecules. The covalent radii, including both single and double bonds, were, according to Robertson, “usually found to give fairly accurate predictions of . . . observed bond lengths.” The Pauling electronegativity factors enabled one to take account of deviations from these predicted lengths due to the partial ionic character of bonds between atoms of different electronegativity. The van der Waals radii set minimum distances between atoms not linked by bonds. Meselson extracted from Robertson’s longer table only the values for nine elements commonly found in organic molecules.45 Meselson assembled these values for hydrogen and covalent

132

C ROSSING F IELDS

bonds, van der Waals radii, and correction factors to assist him in reasoning about the structures of the molecules in the crystallographic papers he was studying. He now wished to try to understand how the bond angles and distances published for various purines and pyrimidines might be applied to elucidate the probable tautomeric forms of the base pairs in DNA.46 The next paper to which Meselson turned was the first one he examined treating directly the structure of one of the actual DNA bases. He must, however, have been disappointed to find in June Broomhead’s crystallographic paper on adenine, published in 1948, no information about the hydrogen atoms or their bonds. Although hydrogen bonds are among the major features of the structure, Broomhead stated, the results were “unfortunately . . . not sufficiently accurate for reliable conclusions to be drawn about the hydrogen positions.” Taking only brief notes, Meselson moved on quickly to a more recent paper on the same molecule, the last reference on the list taken from Bendich.47 William Cochran, co-author of the earlier crystallographic papers on pyrimidines that Meselson had already read, applied more advanced methods of observation and calculation in 1950 to adenine hydrochloride to refine the structure previously studied by Broomhead in the same laboratory (the Cavendish). In order to obtain rigorous measurements of the angular deflections and intensities of the diffracted rays, he returned to the Geiger counter X-ray spectrometer method of detection instead of the more convenient photographic methods most commonly used. Rather than calculating structure factors in the usual way and deriving an electron density map from them, he constructed a “difference map”—that is, one showing the spatial distribution of differences between the observed electron density and that calculated for an assemblage of nonbonded atoms with a known electron distribution. In this way he produced an electron density map in which the only peaks remaining could be identified with the hydrogen atoms.48 Incorporating corrections for temperature factors and thermal vibrations, the map was well enough resolved so that, with the aid of some supplementary reasoning Cochran was able to locate “with confidence” the positions of the seven hydrogen atoms contained in the purine molecule and its water of crystallization (figure 4.11). ˚ ) was shorter than Because the length of the C 6 EN 10 bond (1.30 A ˚ the normal single-bond value of 1.47 A, Cochran estimated from Pau-

C ROSSING F IELDS

133

Fig. 4.11. Bond lengths and angles of adenine hydrochloride, as represented by A. Cochrane

ling’s curve, which related bond length to double-bond character, that the resonance structure (1) in figure 4.12 made a 40 percent contribution to the molecular state of adenine hydrochloride.49 Meselson read this paper with great admiration for the precision both of Cochran’s measurements and of his theoretical analysis. He took nearly two full pages of notes, indicating the special methods employed, copying out in detail Cochran’s drawings of the dimensions of the molecular and crystalline structures, and summarizing Cochran’s arguments for the attachments of the hydrogen atoms. He checked Cochran’s assertion that C 6 EN 10 had “40% double bond character” by looking up comparative bond lengths in Pauling and other sources.50 The main reason Meselson studied Cochran’s paper on adenine so thoroughly was that he wanted to understand its methodology in order to reason cogently about thymine and guanine.51 Immediately following Cochran’s paper in Acta Crystallographica was another paper by June Broomhead on the crystal structure of guanine hydrochloride and its relation to that of adenine hydrochloride. Meselson’s special interest in the structure of guanine would have

134

C ROSSING F IELDS

Fig. 4.12. Resonance structures of adenine, as represented by A. Cochrane

made this paper appear crucial to the further development of his knowledge. Broomhead aimed to “deduce the positions of the hydrogen atoms covalently bound to nitrogen atoms of the purine molecules.” Unfortunately, the paper incorporated none of the refinements that Cochran had applied to the analysis of adenine. Broomhead had relied on more conventional methods, and the resulting electron density projections were “not sufficiently accurate to give directly any information concerning hydrogen positions.” By considering what hydrogen bonds were most likely to form between the molecules within the crystals, she was able to eliminate ten of the fourteen theoretically possible tautomeric forms of guanine, leaving four remaining possibilities. By similar reasoning she eliminated all but two of the possible tautomers of adenine.52 Meselson was apparently too little impressed by this indirect way to establish the positions of the hydrogen atoms on the purines to take notes on this aspect of the paper. Guanine was nevertheless so critical to his interest in the possible roles of ionization and tautomerism in mutagenic base pairs that he summarized extensively the methods, data, descriptions, and dimensions of its molecular structure presented there.53 Before he had completed these notes, however, he was interrupted by the receipt of some superficially unrelated information that gave new direction to his whole enterprise.

III In the letter, quoted at the beginning of the present chapter, in which Meselson mentioned to Jim Watson his search for an entry point into biology, he added, “At about that time I heard of 5-halo-uracil mutagenesis and began to think about chemical differences between the halo-uracils and thymine.” 54 The news about the mutagenic action of 5-halo-uracil came to Meselson through Frank Stahl, who learned about it during a visit to Berkeley to discuss a job possibility. During

C ROSSING F IELDS

135

Fig. 4.13. 5-Bromouracil

the summer of 1954, while Stahl, Stent, Watson, and Martha Chase were riding in a large car driven by Gus Doermann from a meeting at Oak Ridge toward New York, Stent had challenged the group with a mathematical puzzle that he claimed only one person had ever solved. Stahl found the answer in about twenty minutes, after which Stent always considered him “unbelievably smart.” Hoping now to have such an intelligent colleague with whom to discuss his daily work, Stent persuaded Stanley that Stahl could greatly strengthen phage research at Berkeley and cajoled Stanley into giving his authorization to offer Stahl a staff position in the Virus Lab.55 Stahl flew to Berkeley in August to discuss the matter with Stent. While there, he met Rose Litman, who worked in the same laboratory. Litman had escaped with her family from Belgium during World War II, first to Portugal and then to the United States. As an outstanding undergraduate at the University of Indiana she was chosen to be one of the curators of the Drosophila stock collection of Herman Muller. She came to Berkeley to study for her Ph.D. with Arthur Pardee. At the time Stahl spoke with her, she and Pardee had just completed experiments showing that 5-bromouracil can be incorporated into bacteriophage DNA. This pyrimidine derivative is equivalent to a thymine molecule with a bromine atom substituted for the CH 3 group at C 5 (figure 4.13). The 5-bromouracil, they found, replaced thymine in the phage and gave rise to progeny whose mutant frequencies were significantly higher than normal. The very fact that “you could fully substitute with bromine and not kill” all the organisms struck Stahl as very remarkable. Besides showing him the experiments, Litman gave Stahl a mim-

136

C ROSSING F IELDS

eographed copy of a paper on the production of bacteriophage mutants by a disturbance of deoxyribonucleic acid metabolism that she and Pardee had submitted to Nature.56 Meselson had, as we have seen, learned already from Bendich’s review that 5-bromouracil (or 5-BU) was among the altered forms of naturally occurring purines and pyrimidines that could be incorporated into organisms, and he had probably thought of the possibility that their substitution for the normal bases in DNA might induce mutations. Consequently, when Stahl came back from Berkeley and told him that 5-BU actually was mutagenic, they both became excited. It may have been then that Meselson mentioned that he had been thinking about the possibility that the primary phenomenon leading to mutations might be ionization rather than tautomeric changes, and that 5-BU might provoke such a process. This idea came as a complete surprise to Stahl.57 In his workbook Meselson summarized Litman and Pardee’s mimeographed paper as follows: E. coli B grown at 37°C in complete medium ⫹ .25 µ moles per ml 5-Bromo uracil for 4 generations to density of 2 ⫻ 10 8 cells ml are used as hosts for phage growth after centrifugation and resuspension in various con[centration]s of 5-BU in complete medium. Complete medium is a glucose, salts, casein hydrolysate with 1mgm/ml sulfanilamide, 25 µ gm/ml xanthine, and uracil ⫽ 5% of any 5-BU added. Up to 15% of progeny give mutant plaque types. 5 Chloro and 5 iodo uracil, but not 5 amino uracil or uracil give similar mutant yields. Background mutants ⫽ 10⫺3. Selection is ruled out in controls. With “complete” replacement of thymine by 5BU, 9% phage were alive. Sulfa interferes with methyl transfer by interfering with folic acid formation.58 He also copied from the paper a graph depicting the “pattern of action of complete medium plus 5-bromouracil on phage production and mutation” 59 (figure 4.14). Meselson did not need to copy down Litman and Pardee’s assertion that “the most obvious proposal to explain the mutagenic action of 5-bromouracil is that the incorporation of this base into deoxyribonucleic acid may be directly responsible for the appearance of phage mutants.” 60 That viewpoint accorded with the direction of his own thought, but he was already contemplating an explanation at a deeper level. Knowing that the nature and placement of the substituent

C ROSSING F IELDS

137

Fig. 4.14. Graph from paper of Rose Litman and Arthur Pardee, copied in Meselson workbook. Caption in paper reads: “Pattern of action of complete medium plus 5bromouracil on phage production [ⴝ infective yield] and mutation [ⴝ % mutants].”

groups on substituted pyrimidines can alter their ionization constants and their tautomeric properties, he could now begin to focus his study on the question whether the mutagenic action of 5-BU might not be due to a propensity greater than thymine’s to ionize or to undergo a tautomeric shift. The excitement that Meselson and Stahl felt about 5-BU quickly expanded beyond this news of its mutagenic effect in phage DNA. They realized immediately that the substitution of a heavy bromine atom (at. wgt. ⫽ 79.9) for the methyl group (at. wgt. ⫽ 15) meant that DNA in which bromouracil replaced thymine would be sufficiently heavier than normal DNA to hold implications as well for the replication experiments they had first talked about at Woods Hole. That the same molecule could confer both properties on DNA made it seem to them like a dream compound.61 Meselson could appreciate and build on the chemical implications of Litman and Pardee’s paper by himself, but Stahl undoubtedly helped him to appreciate such basic practices of phage research as the single-cell burst experiment and the use of a Poisson formula to calculate “the fraction of infected bacteria which released mutant phage.” Meselson also consulted other experienced virologists in his vicinity. Renato Dulbecco told him that the “incorporation curve [not shown in the graph] follows that for the mutants.” 62 A reference in Litman and Pardee’s paper to an earlier paper in Nature by D. B. Dunn and J. D. Smith, whose method of incorporating 5-BU into the bacteriophage nucleic acid Litman and Pardee had adopted, probably led Meselson to two closely related papers immediately following it, by Stephen Zamenhof (the son of the inventor of

138

C ROSSING F IELDS

Esperanto) and Gertrude Griboff. Zamenhof and Griboff had “grown certain strains of E. coli on a medium containing 5-bromouracil” and “isolated highly polymerized deoxyribonucleic acid in which 18–28 percent of the thymine was replaced by 5-bromouracil.” Asking the question “whether such a profound change in its composition . . . would be reflected in genotypic changes such as mass-mutation,” they returned the colonies, which appeared different from normal in the 5BU medium, to a normal medium, and found no detectable differences from colonies produced by normal bacteria. “This does not preclude the induction of rare mutations by 5-bromouracil,” they concluded; “it merely means that, in spite of being built in rather large amounts into deoxyribonucleic acid, this unnatural pyrimidine does not induce demonstrable mass-mutations.” 63 Meselson summarized this result more bluntly: “Z. gets 5-halo uracil into Coli DNA but no mutations.” 64 Situated as he and Stahl were, within the ethos of the Caltech phage group, they probably did not pursue the possibility that more discriminating experiments might show 5-BU to be mutagenic also in bacteria. The mutagenic effects Litman and Pardee had obtained in bacteriophage would have reinforced in their minds the local dictum that the only acceptable way to study molecular genetics was to do it with phage.65 The new thrust that 5-BU mutagenesis imparted to Meselson’s study program did not suddenly reorient his pathway through the literature on purines and pyrimidines. Thinking about the chemical differences between the halo-uracil and thymine only sharpened the focus of his ongoing effort to understand the effects of ionization and tautomeric shifts on base pairing. The changes in the properties of the thymine molecule that might be brought about by the substitution of bromine for the methyl group must still be linked to his prior question—How might the hydrogen attached to the N 1 of thymine or guanine be removed to permit these two bases to form a mutagenic pair? After his excursion through the papers of Litman and Pardee, Dunn and Smith, and Zamenhof and Griboff on the incorporation of 5-BU into bacteria and phage, Meselson returned to the crystallographic paper of Broomhead that he had previously been reading and took further notes.66 He copied out the guanine structure, wrote “Notice the close similarity to adenine,” and took some additional notes on certain features of the two structures. Among them, he wrote: “Unlike adenine the H⫹ [in ionized guanine] cannot be at N 1 because N 1 is already pro-

C ROSSING F IELDS

139

tonated. In fact H⫹ must be at either N 7 or O from H-bonding considerations. The pK a for guanine is ⬃.9 lower than for adenine.” 67 This statement indicated Meselson’s continuing concern with the question of the removal of a hydrogen atom from guanine or thymine to permit the bases to come together. The comparison of the pKa values of guanine and adenine suggests that he was particularly interested then in the possibility that the removal of the N 1 hydrogen atom could be attributed to ionization. It points also to another reason that the ionization idea was appealing to him. The arguments that he had so far followed about the tautomeric forms of the purines and pyrimidines were mainly qualitative. The ionization constants of the bases offered, on the other hand, the possibility that he could calculate from readily available data the degree to which guanine (or thymine, or 5-bromouracil) could be expected to be ionized.68 It was probably with the expectation that he could obtain further data for this purpose that Meselson next picked up, in the latest issue of the Journal of the Chemical Society, a paper by Adrien Albert and J. N. Phillips on the ionization constants of hydroxy-derivatives of nitrogenous six-membered ring compounds. He did find there new pKa 1 and pKa 2 constants for cytosine, uracil, adenine, and guanine, which he inserted in parentheses under the older values in the table he compiled originally from Bendich. To his surprise and delight, however, he learned that Albert and Phillips also showed how one could use the ionization constants of compounds with hydroxy-substituents to calculate the tautomeric ratios between these and the keto form.69 In his workbook he gave a succinct synopsis of the aspects of their paper that most mattered for his aims: “Give acid and base ionization constants for 87 hydroxy 6 membered nitrogenous heterocyclic compounds. Tautomeric equilibria are calculated from: log R ⫽ pK OMe ⫺ pK OH, where R is the amide/enol ratio.” 70 In the equation of Albert and Phillips reproduced in Meselson’s notes, pK OMe and pK OH were the “observed values for the addition of a proton to corresponding methoxy- and hydroxy-compounds.” Although they listed in their paper ionization constants for a total of 102 compounds, they centered their discussion of tautomerism around the simplest of them—the pyridines—which contain one nitrogen atom in the heterocyclic ring. In this case, what they called the amide/enol ratio (equivalent to the keto/enol ratio because the shift of a hydrogen atom from O to N simultaneously converts the oxygen to the keto form

140

C ROSSING F IELDS

Fig. 4.15. Amide (or keto) forms (left), and enol form (right), of pyridine molecule, as represented by Adrien Albert and J. N. Phillips

and the ring nitrogen to a tertiary amide) meant the ratio between the compounds shown in figure 4.15, for example. Their method for calculating these ratios assumed, Albert and Phillips stated, that the basic ionization constant (the pK a′ value, representing protons gained) for a given hydroxy (EOH) compound would differ little from that for the corresponding methoxy (EOCH 3 ) compound. “That this is a reasonable assumption,” they remarked parenthetically, “may be gauged by reference to the values of the methoxyand hydroxy-anilines.” 71 To clarify his understanding of Albert and Phillip’s equation, Meselson restated it in the nonlogarithmic form, keto/enol ⫽ K oMe /K OH, and “established” the latter from the equations for the equilibrium constants of the respective keto/enol reactions, which he wrote out in pictorial form (figure 4.16). In shorthand he summarized several assumptions that Albert and Phillips had discussed, including the statement, quoted above, that the basic ionization constants for the hydroxy and methyl compounds were similar. Turning to the data for methyl- and hydroxy-anilines to which they had referred to make this assumption “reasonable,” he copied out the pK values they gave for the three stereoisomers of each of the compounds: hydroxyaniline o- ⫽ 4.72, m- ⫽ 4.17, p- ⫽ 5.50 methoxyaniline o- ⫽ 4.49, m- ⫽ 4.20, p- ⫽ 5.29 These figures, which gave empirical justification for Albert and Phillips’s assumption, drew Meselson’s attention to the effects of the relative positions of the substituents on the heterocyclic ring. Paraphrasing their argument, he wrote down that these data show that “inductive (-I) effect is same for OH & OCH 3 & ⫹M effects are similar, being a bit greater for OH.” 72 The “Inductive effect” (-I) was defined as the displacement of electrons in a substituted organic molecule arising from the influence of an electron-attracting substituent. The effect, transmitted both through the carbon chain of the molecule and through space, displaces elec-

C ROSSING F IELDS

141

Fig. 4.16. Keto/enol reactions of hydroxy- (above) and methoxy- (below) pyridine, and equilibrium constants for the reactions, in Meselson workbook

trons toward the substituent, leaving more distant portions of the molecule more positive. Consequently, a hydrogen ion may be more easily removed from the distant portions, resulting in increased acidity. The designation (M⫹) was given to the “Mesomeric effect” in substituted aromatic molecules. Certain substituents exert a strong ortho-para influence, that is, they cause a second substituent to become attached at the positions ortho or para to the first substituent.73 In his workbook Meselson had sketched out the isomers of hydroxy- and methoxyaniline for the respective pK values shown above. Contemplating the relations between these structures and the values, he inferred that “for O meta to a ring N, the usually preferred keto form involves charge separation and leads one to expect the enol form to be rather abundant.” He sketched out the tautomeric forms in question (figure 4.17).74 Although stated in the form of a prediction based on the steric relations, Meselson’s reasoning was guided by the knowledge that the same conclusion could be reached directly from Albert and Phillips’s formula. Albert and Phillips had calculated R only for the ortho- and para-hydroxypyridines (left and right in figure 4.17), but one could immediately see, by inspection of the figures that Meselson copied down for pK OMe and pK OH for m-hydroxypyridine, that the formula yields for it R ⬃ 1, just as Meselson entered below that molecule in the sketch. Probably Meselson recognized, almost as soon as he had followed out this reasoning, that it would not do. Below the figures, he wrote: The above argument forgets that all 3 compounds involve charge separation. A better explanation, which also predicts the sequence of R values is the simple enumeration of resonant structures in the keto form

142

C ROSSING F IELDS

Fig. 4.17. Tautomeric forms of ortho- (left), meta- (center), and para- (right) hydroxyaniline, in Meselson workbook

ortho 2 meta 1 para 3 75 His new argument was a simple application of the most general principle of resonance theory: that molecules which can resonate between two or more electronic structures are more stable than they would be if any single one of the structures represented the normal state of the molecule.76 If only one keto form could exist for metapyridine, then its keto form would be expected to be less stable than the keto forms of ortho- or para-pyridine. For his own interest in mutagenesis, the critical question for Meselson was whether he could apply Albert and Phillips’s method for calculating tautomeric ratios to the pyrimidines relevant to DNA base pairing. In their general discussion section Albert and Phillips cautioned: “It is, in general, not desirable to extend to rings containing two or more nitrogen atoms the calculations of amide:enol ratios used for the one nitrogen rings, for it cannot be assumed with any certainty that the same nitrogen is protonated in the pair of compounds being contrasted.” 77 Meselson could not altogether ignore this warning. “The calculation of R for multi nitrogen rings is risky,” he wrote, “because [the] assumption [that the reaction by which the N in the hydroxycompound has the same equilibrium constant as that reaction for the methoxy-compound] may break down because the same N may not be protonated in the two compounds.” 78 But he reasoned that Albert and Phillips’s disclaimer would not apply to the pyrimidines incorpo-

C ROSSING F IELDS

143

rated into DNA molecules, because the second nitrogen was bonded to sugar and not free to be protonated.79 In calculating these ratios, Meselson saw, one was going through a little mental cycle in which one first removed a hydrogen atom from one site on the molecule, its tendency to leave being measured by the first pKa. Then one put this hydrogen atom on the second site, against a tendency to leave that site measured by the second pKa. The difference between the two pKa’s represented the free energy of tautomerization. It seemed to him rather wonderful that, with such easily measured values as ionization constants, one could calculate tautomeric ratios. He saw that its application to the situations that interested him would be crude. The pK values given in tables represented substances in solution at standard conditions—1 molar concentration at 20°. What would happen in the case of bases incorporated into the structure of DNA in organisms might be an altogether different matter.80 Nevertheless, he thought that he might be able to use the method for his planned study of 5-bromouracil mutagenesis. If so, it would be necessary to consider in greater detail the positional and resonance effects of ortho-, meta-, and para-substituents on the ring structures, as he had already begun to do in thinking about Albert and Phillips’s paper. They discussed such effects on the pK values and the tautomeric ratios of the substances listed in their table. Meselson returned, therefore, to the crystallographic literature, beginning with two papers that Albert and Phillips had cited.81 The first of these was a paper on the crystal structure of uracil published by G. S. Parry in 1953. Parry noted that, up until then, no detailed studies comparable to those of Cochran and Broomhead on adenine and guanine had been reported for “the pyrimidine bases uracil, thymine and cytosine” derived from the hydrolysis of nucleic acids.82 A detailed knowledge of uracil, closely related as it was to both thymine and 5-BU, was strategic to Meselson’s evolving aims. Using an electronic computer to calculate the structure factors for uracil for a three-dimensional Fourier difference summation, Parry was able to resolve the density map sufficiently, in three stages of refinement, to determine the positions of all four hydrogen atoms. Two were attached to the nitrogen atoms, and two to ring-carbon atoms. This result confirmed that, when crystallized, uracil contains two keto oxygens (figure 4.18). From arguments based on Pauling’s relation between bond lengths and character, Parry concluded that only five of the fifteen possible forms contributed to the resonance structure. Mes-

144

C ROSSING F IELDS

Fig. 4.18. Uracil, as represented by G. S. Parry

elson copied the crystallographic data and drew the five resonance structures,83 but the paper offered only slight leads for his questions about mutagenic pairing. Resonance structures in which the hydrogen atoms carried positive charges might be related to the general idea that hydrogen atoms are removed by ionization, but information concerning uracil could be used, at best, with caution, to argue from position effects how 5-BU might pair more readily than thymine with guanine. The second paper was “The Electron Distribution in Crystalline α-Pyridone” by Bruce R. Penfold, also published in 1953. Pyridones differed from pyrimidines in containing only a single nitrogen atom, rather than two, in their six-membered rings. Information concerning their structures might seem, therefore, still further removed from the specific questions Meselson had formulated in his study by this time. The paper did, however, apply the same precise crystallographic methods that Cochran had devised for adenine.84 One of Penfold’s claims—that the tautomeric form predominant in solution depends on the pH—caused Meselson to doubt the author’s knowledge of physical chemistry, but Meselson welcomed one of the conclusions Penfold drew from his analysis of the crystal structure. The precise location that he established for the hydrogen atom connecting the nitrogen atom of one α-pyridone molecule to the oxygen atom of the next molecule along the helix in which the molecules were arranged in the crystal made it certain, according to Penfold, that the hydrogen was covalently bonded to the nitrogen and was not “at the midpoint of the bond.” 85 Perhaps out of courtesy to Louis Hunter, who had supplied his crystals, Penfold did not mention that this result ruled out the application of Hunter’s “mesohydric theory,” according to which the hydrogen atoms should be resonating between the two molecules. To Meselson, however, that was the tacit meaning of Penfold’s discussion. A distinct difference between the two bond lengths of a hydrogen bond

C ROSSING F IELDS

145

was one of the surest indicators of the existence of stable tautomeric forms. Beside a sketch of the “∞ helices along C-axis” that he drew to illustrate Penfold’s conclusion schematically, he stated, “Clearly resolved hydrogens definitely covalently bonded to N rather than O. Here is very clear contradiction to Hunter’s views.” 86 Meselson was doubtless cheered by this evidence that he could now banish Hunter’s objection to the existence of tautomers from his thinking about base pairs. Meselson came next upon a crystallographic study of 4, 5 diamino-2 chloropyrimidine that had just appeared. Co-authored by Miss Noel White and C. J. B. Clews, the analysis incorporated all the refinements that had by then made it routine to locate each of the hydrogen atoms by electron density difference projections. White and Clews asserted the centrality of such crystallographic studies to the elucidation of the Watson and Crick double helix by claiming that the hydrogen bonds between the pyrimidine and purine bases “must be of the type found in the simpler compounds.” The contents of their paper gave little guidance, however, on how to cross the gap between the conditions under which hydrogen bonds formed in crystals of various substituted pyrimidines or purines and the influences that might direct their formation within the double helix. Meselson copied down as usual the basic data and the dimensions of the crystal, and showed some interest in special problems raised in the paper, but there is no indication that he found in it any new generalizable relationships that he could apply to his questions about base pairing.87 From a paper published in 1952 by Joseph Singer and I. Fankuchen on the crystal structure of 2-metanilamido-5-Br-pyrimidine, Meselson may have hoped to obtain some information about the effects of bromine substitution on pyrimidine applicable to the structure of 5-bromouracil. Unfortunately, Singer and Fankuchen’s analysis was “not precise enough to contribute to the knowledge of the details of the pyrimidine ring,” and Meselson saw little of interest in it.88 Having absorbed the crystallographic details of the analysis of one pyrimidine or purine molecule after another, without finding much in them to advance his ideas about how thymine might mispair with guanine or why 5-BU may do so more easily than thymine, Meselson turned to a short paper by Jerry Donohue that offered a refreshing perspective on the whole question of base pairs in DNA. Donohue’s article “Hydrogen-Bonded Helical Configurations of Polynucleotides,” recently communicated to PNAS by Pauling, challenged the orthodoxy

146

C ROSSING F IELDS

Fig. 4.19. Table of geometrically acceptable base pairs in polynucleotides according to Gerry Donohue

already beginning to surround Watson-Crick base pairing. Because, according to Donohue, “it is by no means certain that all nucleic acids, whether DNA or ribonucleic acid, . . . will conform to” equimolar amounts of adenine and thymine, and of guanine and cytosine (that is, to the Chargaff ratios that had been built into the reasoning behind the Watson-Crick model), “it becomes of interest to enquire into the possibility of other structures.” 89 Donohue approached this inquiry by making models of each of the four bases with dimensions and hydrogen bond distances within the limits of error of the published crystallographical data. He numbered the hydrogen donor and acceptor sites for each base and explored how many ways they could fit together two by two. Within these constraints he found twenty-four “geometrically acceptable base pairs,” which he listed in a compact table (figure 4.19). With several of these sets, in addition to those used in the WatsonCrick model (sets 5 and 15), Donohue found it possible to build twochain right-handed helical polynucleotides in which the parallel ˚ apart (that is, they conformed to the planes of the bases were 3.4 A basic crystallographic spacing that underlay the Watson-Crick model). Donohue illustrated several of these possibilities. Base-pairing 2 and 9, for example involved only purines (figure 4.20). “We conclude from this study,” Donohue wrote, “that polynucleotides may assume two-chain structures other than the one proposed by Watson and Crick.” 90 Looking back after more than forty years of familiarity with the double helix, we may see Donohue’s conclusion as mildly subversive. It is easy to forget that in 1955 the evidence in support of the WatsonCrick base pairing scheme was not conclusive enough to forestall variant pairings such as Donohue, who knew the crystallographic considerations at least as thoroughly as Watson and Crick did, here proposed. The situation was still fluid. Meselson copied out the diagrams show-

C ROSSING F IELDS

147

Fig. 4.20. Example shown by Donohue of purine base pairs compatible with crystallographic data underlying the Watson-Crick model of the double helix

ing the donor and acceptor sites on the four bases and wrote out the whole list of twenty-four “acceptable H-bond pairings.” 91 He was not, however, attracted to Donohue’s view that stable DNA existed in structures other than that postulated by Watson and Crick, or without the usual equimolar ratios of adenine to thymine, and of guanine to cytosine. He was interested in Donohue’s alternative structures because occasionally they might occur and lead to mutations.92 (Commenting in 1993 on his attitude in 1956, Meselson said, “Now you could ask, why did I think the Watson-Crick structure had to be the basic one? I don’t have an answer for it. It’s just because it was too sacred, I guess. It was very high. You couldn’t throw anything high enough to hit it. It was too good to be wrong.”)93 The dimensions that Donohue assumed for his model of cytosine came from the paper published in 1950 by Sven Furberg on the crystal structure of cytidine that had played a role in the construction of the double helix (See Chapter 1, p. 31). Cytidine is the nucleoside composed of cytosine and a deoxyribose sugar molecule. Since there were no published reports of the structure of cytosine alone, Furberg’s paper served as the only source of information available on the dimensions of this base.94 For Meselson’s purpose, however, the Furberg paper was apparently only another letdown. Working just before the introduction of the methodological refinements that enabled structures to be resolved down to the level of the hydrogen atoms, Furberg could provide “no evidence from the Fourier maps with regard to the positions of the hydrogen atoms in the structure.” 95 In his workbook Meselson put down only a brief, nominal note on the paper.96 By now he must have felt that he had little to gain by putting in more time on the crystallographic literature.

148

C ROSSING F IELDS

III By late August, Meselson was ready to examine the properties of purines and pyrimidines for himself. His reading had persuaded him that the tautomeric problem was too difficult to start with and that it would be better to begin with the ionization constants of the bases. Ionization might, after all, turn out to be as important as tautomerization for the mechanism of mutagenesis. On 30 August he obtained samples of five purine and pyrimidine bases—thymine, cytosine, guanine, adenine, and 2-aminopyrimidine—from Pauling’s collaborator, Robert Corey. He tested each base by paper chromatography. During the previous decade paper chromatography had proved very successful as a method for separating the components of nucleic acids, because the intense absorption of ultraviolet light by purines and pyrimidines enabled one to detect and estimate the separated substances by spectrophotometry. One placed a small amount of the substances to be separated in this way on a spot near one end of a strip of filter paper, then hung the paper in a cylinder with the end of the strip dipped in an organic solvent. (Meselson used a standard solvent containing 1-butanol, water, and NH 4OH.) As the solvent ascended through the paper, the substances in the spot were carried along with it but lagged behind the moving front of the solvent by a proportion characteristic for each substance. The ratio Rf ⫽

movement of band movement of advancing front of liquid

was, under carefully controlled conditions, precise enough to identify the substance. The R f values for purines and pyrimidines were given in G. R. Wyatt’s article on the chromatographic separation of nucleic acid components, among other places. The outcome of Meselson’s runs on the five bases he had received was that “All seemed pure and showed expected R f values . . . except 2 Amino pyr[imidine], which couldn’t be located under u[ltra] v[iolet] after development.” 97 Testing the purity of these bases chromatographically was probably a preparation for titrating them to determine their ionization constants,98 but there is no record that Meselson took that step at this time. What he did do was redirect his reading toward the effects of substitutions in the bases on their ionizing properties. In H. B. Watson’s Modern Theories of Organic Chemistry, an intermediate-level textbook that he picked up on the Crellin library

C ROSSING F IELDS

149

Fig. 4.21. Dissociation constants for phenols, copied by Meselson from H. B. Watson, Modern Theories of Organic Chemistry, in Meselson workbook

shelf, Meselson found a section on the effects of halogen substitution in organic compounds that appeared helpful for his interest in the properties of 5-bromouracil. “The halogens,” Watson stated, “form very stable anions and exhibit, to a marked degree, the power of drawing electrons to themselves.” Halogen-substituted acids, such as chloroacetic acid, were, therefore, much stronger than their unsubstituted counterparts, due to the inductive effect (-I) of the halogen substituent. Drawing negative charge toward itself, the halogen atom left more distal portions positively charged and therefore more easily ionized. That was the elementary effect. In aromatic compounds the situation was less simple. The main influence was the inductive effect of the halogen, so that “as a class halogenated aromatic acids and phenols are stronger than the unsubstituted compounds.” Because the halogens also have mesomeric moments, and are op-directive, there are other effects, so that “the strengths of the halogenated aromatic acids, phenols, and bases are not in the order which would be predicted on the basis of inductive effects.” To illustrate these generalizations, Watson provided a table showing dissociation constants for nine classes of aromatic compounds, comparing for each class the unsubstituted compound to those containing fluorine, chlorine, bromine, and iodine, respectively, in the meta- and para- positions relative to the dissociable substituent group. Meselson picked out one of these classes, the phenols, and listed Watson’s figures for each of the substituents (figure 4.21).99 The significance to his own purpose that Meselson saw in these data lay in the similarity between the relation of the halogen substituent and the phenol group on the benzene ring and the relation of the bromine and the keto (or enol) groups on 5-bromouracil (figure 4.22).100 Reasoning by analogy from the properties of substituted benzene rings to those of heterocyclic aromatic rings, Meselson inferred from these data, showing the ionization constants for both m-Br and p-Br

150

C ROSSING F IELDS

Fig. 4.22. Meta-bromophenol (left), para-bromophenol (center), and 5-bromouracil (right)

to be substantially greater than for the unsubstituted phenol (H in his list), that the ionization of one or both of the O substituents of uracil was favored by the bromine at C 5 The analogy was only suggestive: strongly so between p-Br phenol and the oxygen at C 2 on uracil, weaker between m-Br and the oxygen at C 6. Nevertheless, there was enough here to tell Meselson that he was on the right track in suspecting ionization to be a significant factor in the mutagenic properties of 5-BU. His search for further data concerning the ionization constants led Meselson, perhaps by way of a reference in the article by Wyatt that he consulted for his chromatographic tests, to a paper by David Shugar and Jack J. Fox titled “Spectrophotometric Studies of Nucleic Acid Derivatives and Related Compounds as a Function of pH.” Shugar and Fox measured the ultraviolet absorption spectra of several substituted pyrimidines, including thymine, cytosine, 5-methylcytosine, uracil, 5 nitro-uracil, 3-methyl uracil, and 1,3 dimethyl-uracil, over the continuous range of pH values from pH 1 to pH 14. They found that, for each compound, over certain ranges these curves remained constant, then changed rather sharply to a different curve that remained steady for another range of pH values, then in some cases changed again. Comparing the transition points with dissociation constants for the compounds measured electrometrically, they showed that their absorption curves gave an alternative method to establish the pK values for the first and second dissociation constants of each compound.101 Some of these dissociation constants, Meselson saw, might be useful to him in his study of the effects of substitutions in pyrimidines on their ionization. From data scattered through the article, he picked out several values pertinent to his concern, including the following: 5–nitro has pK a1 ⫽ 5.3

pK a2 ⫽ 11.7

uracil

⬎13

thymine

9.5 9.9

102

C ROSSING F IELDS

151

Their significance for Meselson is evident. Nitro substituents, like halogens, were electron-attracting and could therefore be expected to alter the ionization properties of the molecule in the same sense. The fact that this substitution increased the strength of both the first and the second ionizations compared to those of uracil, and the first also with respect to thymine, clearly favored his view that the ionization of 5-BU might explain its mutagenic property.103 In further preparation for the titration studies he planned, Meselson reread the discussions of the acid-base properties of the nucleic acids and their components in D. O. Jordan’s chapter in the ChargaffDavidson volume. The section on titration data for DNA included the anomalous forward-backward titration curves of Gulland with which Meselson had become familiar at Woods Hole in 1954, when he tried to perform similar titrations on RNA for Jim Watson. (See pp. 60– 62.) As Meselson was already aware, these results were, according to Jordan, “in remarkable agreement with the structure proposed by Watson and Crick,” but Jordan drew a further inference: “careful examination of the titration curve shows that the anomalous behavior ceases at approximately pH 3.5 after the titration of 1.8–2.0 equivalents of amino groups, i.e., the back- and forward-titration curves are coincident at pH values more acid than 3.5, which indicates that the most acid amino group, viz., that of guanine, does not partake in the formation of hydrogen bonds as suggested by Watson and Crick.” 104 Jordan’s statement is clearer when juxtaposed with his reproduction of the Watson-Crick pairing scheme (figure 4.23). Watson and Crick represented the amino group of adenine and that of cytosine as participating in one of the two hydrogen bonds that connect the bases, whereas the corresponding amino group on guanine does not make such a connection. Unpersuaded by Jordan’s reasoning, Meselson expressed his disagreement in an extended commentary: The back and forward titration curves coincide at pH ⬍ 3.5, 7, and 12.5. At pH 3.5 about 2 equiv.[alents] of “amino group” have been titrated and further acid is supposed to titrate the guanine NH 2 group. This leads D.O.J. to conclude guanine NH 2 groups are not hydrogen bonded. Models show that NH3⫹ on guanine could hydrogen bond almost as well as NH 2 so the conclusion of D.O.J. is not necessary. Furthermore, if Broomhead is right, the H⫹ may be going on to the imidazole ring nitrogen. If that is so, then shouldn’t [the NH 2 group of] adenine protonate in the same way while in hydro-

152

C ROSSING F IELDS

Fig. 4.23. Watson-Crick base pairs, as represented by D. O. Jordan

gen bonding? The only data showing it doesn’t is the “1.8–2.2 equiv[alent]” of Jordan mentioned above.105 Meselson’s confidence that Jordan’s conclusion—and by implication, the configuration of the hydrogen bonds shown in Watson and Crick’s base pairs—could be challenged was supported by the knowledge that Linus Pauling thought there were three hydrogen bonds between guanine and cytosine. After putting these comments in his workbook, Meselson obtained a preprint of the article Pauling and Corey had recently submitted for publication, “Specific HydrogenBond Formation Between Pyrimidines and Purines in Deoxyribonucleic Acids.” The “postulate” of Watson and Crick that “the two polynucleotide chains of the double helix could separate from one another,

C ROSSING F IELDS

153

and each could then serve as the template for the synthesis of a duplicate of the other” was, Pauling and Corey wrote, “so attractive, and the problem is so important, that we have thought it worth while to analyze the available experimental information about the detailed molecular structure of the pyrimidines and purines.” 106 The experimental information that Pauling and Corey reassessed consisted of the same crystallographic papers by Clews and Cochran, Broomhead, and Parry that Meselson had just read and on the first two of which Watson and Crick themselves had relied. Through their intimate knowledge of the nature of the chemical bond, however, Pauling and Corey were able to subject these data to a more searching theoretical critique than anyone else had thus far attempted. Scrutinizing the reported bond lengths and angles from the viewpoint of those expected according to the types of bond orbitals involved, the expected single or double bond character, and characteristic tetrahedral angles, they combined “the information provided by these investigations and . . . reasonable arguments based upon considerations of electronic structure” to arrive at “reasonable structures” for each of the four bases normally contained in DNA. On this basis they reevaluated the linkages between the complementary base pairs of the WatsonCrick helical structure. Selecting hydrogen bond distances within the ˚ that the crystallographic studies had shown to exist range 2.96–3.37 A in intermolecular hydrogen bridges, Pauling and Corey depicted the adenine-thymine pair essentially as Watson and Crick had.107 On the other hand, they thought that “there is no doubt that the guaninecytosine pair involves three hydrogen bonds, and we may conclude that the difference between this pair and the adenine-thymine pair, with two hydrogen bonds, is such as to introduce greater specificity in the action of a polynucleotide as a template than was indicated by the considerations of Watson and Crick, who had assumed that two hydrogen bonds are formed in each case” 108 (figure 4.24). Meselson filled a page of his workbook with a summary of the arguments from which Pauling and Corey derived their structures, and two pages with large sketches of their diagrams of the dimensions of the molecules. He was independent enough to question some of his mentor’s assumptions. To nitrogen atoms attached to hydrogen atoms they had assigned sp 3 orbitals, “so that the preferred angle is 125°16.” Meselson wondered why “P. and C. chose the double- single tetrahedral ∠ which involves excited states, rather than the single sp 2 angle 120°,” and he pondered an alternative explanation for the angles in uracil.

154

C ROSSING F IELDS

Fig. 4.24. Cytosine-guanine base pair, as represented by Linus Pauling and Robert B. Corey

But he must have been impressed that structures arrived at by crystallographic methods required further interpretation based on a deeper understanding of the nature of chemical bonds. Across the page in which he had copied the dimensions of the first pyrimidine molecule from Clews and Cochran’s study of 2-amino-4, 6-dichloropyrimidine, he wrote with an orange crayon, “Beware of param[eters] . . . !” His implicit acceptance of Pauling and Corey’s argument for the three hydrogen bonds in the guanine-cytosine pair is indicated by the fact that to the base pair he had already drawn with three hydrogen bonds near the beginning of his workbook he now added in pencil the lengths that Pauling and Corey assigned to these bonds.109 In the summary of their paper, Pauling and Corey emphasized that their modification of the original Watson-Crick pairing scheme for guanine and cytosine “strengthens the arguments of Watson and Crick as to the role of complementariness of structure of two DNA polynucleotide chains in the duplication of the gene.” 110 The structures derived directly from the crystallographic analyses of Cochran and Broomhead assigned hydrogen atoms to the N 1 nitrogen of both adenine and cytosine, whose dimensions were considered nearly identical. The alterations that Pauling and Corey introduced into the bond lengths and angles differentiated the dimensions of these two purines just enough to justify placing a hydrogen atom at this location on guanine alone.111 The discrepancy between the presence of a hydrogen

C ROSSING F IELDS

155

atom at this location on adenine and its absence in the adeninethymine base pair can be explained away by the fact that the situation in a molecule of adenine hydrochloride may differ from that for adenine within the DNA molecule, but the more exact fit that Pauling and Corey achieved between the structure of this base and the requirements of the Watson-Crick base-pairing scheme appears to have been influenced not only by their intimate knowledge of bond lengths and angles but also by their great attraction to the double helix. Watson and Crick are reported to have regarded Pauling and Corey’s paper as trivial.112 In 1954 they had written that it was “uncertain” as to whether guanine and cytosine might form a third hydrogen bond. These uncertainties they declared to be “of only second-order importance.” In the self-assurance they had by then reached about the basic correctness of their DNA structure, they were content to leave such questions “unsettled until the configuration of both these bases are known to a greater accuracy.” 113 Nor did they express any concerns about the correctness of the hydrogen bonds they assigned to the adenine-thymine pair. Pauling believed that the importance of the structure demanded a thorough analysis of the available experimental information, but his eagerness to “strengthen their argument” 114 reflected the belief that pervaded Caltech that the double helix was too good to be wrong. The report of the Biology Division in the Annual Report of that institution for 1954–55 stated that the structure for DNA proposed two years earlier by Watson and Crick “is now widely regarded as being substantially correct. It is a double helix and accounts for gene specificity, gene reproduction (the molecular basis of all reproduction, it is believed) and gene mutation. Genes may be thought of as information coded in molecular form.” 115 In the local setting where Meselson worked, the double helix was already heralded as the advent of a new age in biology. After finishing his study of the Pauling-Corey paper, Meselson read no further in the crystallographic literature concerning purines and pyrimidines. Doubtless he felt that he had absorbed the best information available and had seen the latest word on the subject. He had overcome his initial ignorance of the structures of these molecules, at least to the extent that he could hope to be taken seriously in the Division of Chemistry when he discussed base pairing and mispairing. Moreover, he must have come to realize that, in the current state of the field, further consideration of crystallographically determined molecular dimensions would not help him think about the mechanisms

156

C ROSSING F IELDS

of mutagenesis. On the question of tautomeric shifts, for example, Pauling and Corey’s arguments merely reinforced the view he had noted in the existing literature that “the keto and amino forms respectively are preferred.” 116 During the various tacks he had taken in his rapid voyage through the literature about purines and pyrimidines, the most promising soundings he made came not from the crystallographic data but from physical chemistry. It was the possibility that ionization constants might enable him to calculate the mutagenic propensities of normal or abnormal base pairs that seemed most promising to him as he neared the end of the summer weeks spent in the comfort of the Crellin library. The absence of dates in the first twenty-seven pages of his workbook leaves some uncertainty about how many weeks had gone by since Meselson began the literature study retraced in this chapter, but it was in any case a remarkable intellectual foray. At the beginning he was seeking only to gain rudimentary knowledge of the structure of the molecule whose mode of replication he hoped eventually to test. Along the way he identified a second problem that he might take up related to another property of the Watson-Crick model, and saw in outline how he might combine its pursuit with that of his original “towering idea.” By the end he was thinking creatively about research strategies that he could carry out in the laboratory. He had found his entry point into biology.

C HAPTER F IVE

Dense Solutions

I While Matt Meselson was crossing the boundary from physical chemistry into biology during the summer of 1956, Frank Stahl was considering a move that would forestall their intended collaboration. The offer from Gunther Stent to join him at Berkeley attracted him very much. He had earlier turned down a position in the department of Geoffrey Brown at the University of Michigan because he felt “the need to get more momentum research-wise before I get hit with teaching duties.” 1 At Berkeley, however, he would do only research. Moreover, in spite of some misgivings about Stent’s recent publications, Stahl regarded him as an extremely bright person near whom it would be good to work. Mindful also of Mary’s continued unhappiness in Pasadena, where she could find no entry into a community that seemed to be made up only of scientists and wealthy retirees, Stahl wrote to Wendell Stanley on 30 August that he accepted the Berkeley offer but would defer his move until the following year.2 Stanley replied promptly: September 4, 1956 Dear Dr. Stahl: I have your letter of August 30 and am glad that you will join the staff of the Virus Laboratory on or around the first of September, 1957. All members of the staff are provided with supplies and equipment as well as technical assistance necessary for the proper prosecution of their work. It will not be necessary for you to attempt to secure outside funds. If you will let Doctor Stent or me know some months prior to your coming concerning any special requirements, every attempt will be made to secure these for you. With kindest regards, I am Sincerely yours W. M. Stanley3

158

D ENSE S OLUTIONS

Whatever reservations he may already have had concerning Stanley’s personal style, Stahl was pleased with his decision when he wrote Cy Levinthal at the University of Michigan to explain why he had chosen to go to Berkeley rather than to Michigan. In the same letter he outlined his research plans, mentioned some of Meselson’s plans, and added a note from Meselson himself. His primary aim during his last year at Caltech would be to overcome the problems that had stalled his efforts to obtain a culture of potent male E. coli bacteria. “I’m now working almost entirely,” he reported, “at understanding Hfr in Joe’s coli C. Will then study P2 transmission via Hfr’s.” 4 “At Berkeley, and perhaps to a small extent here,” Stahl planned “to look at some aspects of the genetics of brom-uracil T4. In particular to describe the genetic behavior of non-viable substituted phage in cross-reactivation experiments and also to test the idea that substituted phages which are not mutant can make mutant progeny. This will consist of examining the frequency distribution in single bursts of mutants thrown off when cells are infected with non-mutant substituted phage. (viable substituted phage, i.e.).” Stahl had already given much thought to the problems that might arise and how he might cope with them. “The biggest difficulty would seem to be in removing from consideration bursts which arise from cells infected with mutant viable phage.” To get around this problem he proposed to use as the host for phage growth a bacterial strain (Kλ) that would screen out mutant substituted phages. “The prettiest result we could hope for,” he wrote, “would be a distribution similar to that of markers cross-reactivated from P32-suicide phage.” His plan was built both on his Ph.D. thesis research and on his discussions with Meselson about the mutagenic effects of 5-bromouracil. Concerning Meselson’s activities, Stahl reported: “Matt Meselson of the Chemistry Dept. here is interested in the potentialities of brom-uracil phage. He has calculated that the pK of these phage may be sufficiently different from non-substituted that electrophoretic separations can be affected [sic]. If they can be, he’ll probably try some transfer experiments and correlate genotype from mixed infections of substituted and non-substituted T4 with electrophoretic mobility.” Meselson’s project was a logical extension of the interests he had developed during the previous month concerning the effects of 5-BU substitution on the ionization properties of the bases and of the nucleic acids into which they are incorporated. He now envisioned the change he calculated for the pK of the DNA in substituted phage as a means to follow the distribution of parental phage

D ENSE S OLUTIONS

159

DNA into progeny DNA. The difference in charge due to differences in ionization would serve as the means of separation through electrophoretic methods. His plan bore only a remote resemblance to the idea he had discussed with Stahl two years earlier in Woods Hole to investigate DNA replication by separating substituted from unsubstituted DNA by means of a difference in their densities.4 The note Meselson appended to Stahl’s letter stressed the chemical aspects of base pairing. It reveals that he had moved, during the first two weeks of September, beyond assimilation of the literature and devised a strategy for his own investigation of mutagenesis. Hello, You once asked to be informed about any N.M.R. experiments dealing with tautomeric ratios and conversion frequencies. Harden McConnell (Chem. Dept.) and I will try to look into this fairly soon and will keep you advised. From x-ray, I.R., and U.V. work it would seem that EOH and CNH forms are present in only small amounts. The lower limit of detection by NMR may be too high to pick them up. Even if this is so, the conversion frequencies may be determinable. We’ll soon have a large (150) variety of substituted purines and pyrimidines to work with. I’ll also look into pK’s of the bases because ionization as well as tautomeric shift may switch base pairing. With 5-Bromo-uracil, for example, there is likely to be more of the ionized form at pH⫽7 than of the enol form. Matt Meselson Stahl had retyped the note for his friend “because Matt went to a progressive school too.” Below it he added, “(Enclosed is a sketch by Matt of pairing between Guanine and enol-brom-uracil and between Guanine and ionized brom-uracil. Also between Adenine and the Keto-bromo-uracil).” 4 Meselson’s plan was well-defined, bold, and optimistic. All of the crystallographic, ultraviolet, and infrared absorption studies that he had read supported the opinion that purines and pyrimidines existed primarily in the amino and keto forms. That consensus was unhelpful to his effort to explore the role of tautomeric shifts in mutagenic base mispairing. Perhaps by reading a chapter in Nucleic Acids on the magnetic properties of their components he had learned that in some cases nuclear magnetic resonance measurements could provide information on structural features of organic molecules, including keto and enol forms, that was not accessible to the more common methods for studying their structures. He had then gone to see Harden McConnell, a

160

D ENSE S OLUTIONS

physical chemist at Caltech who was a world leader in the application of resonance to such structural questions. Taking advantage of the ethos at Caltech that encouraged graduate students to approach any member of the faculty who might know about a particular problem, he visited McConnell at his home to talk over these possibilities. As Meselson’s note suggests, McConnell not only encouraged him but offered to assist. To facilitate the projected study he had also ordered, probably with the help of Pauling, a large number of substituted purines and pyrimidines synthesized by the late Treat B. Johnson. A member of the Yale Chemistry Department from 1902 until his retirement in 1943, Johnson had published 182 papers on the chemistry of the pyrimidines. Sometime in September Meselson received from New Haven what appeared to be the entire Yale collection.5 Meselson could easily conduct by himself the measurements of the ionization constants of the bases, because he had already performed similar electrometric titrations on RNA at Woods Hole. His note to Levinthal implies that at this time he regarded ionization as a more promising explanation for base pairing switches than tautomerism. The explicit reason for his preference was that there was likely to be “more of the ionized . . . than of the enol form.” Another, more tacit reason was that he understood ionization better than he did tautomerism. The carbon copy of Stahl’s letter that mentions Meselson’s sketches of the base pairs does not contain a copy of these sketches. Even without seeing them, however, we can view them as the centerpieces of his current interest in the molecular mechanism of mutagenesis. Assuming that he could provide supporting evidence that 5-BU did occur more readily than thymine in the ionized or enol form, to complete his theory he would have to show how these forms would pair preferentially with guanine rather than with adenine. Pauling and Corey’s analysis of hydrogen bonding between base pairs set forth clearer criteria for making such a case than had existed when Watson and Crick made their original proposal about base switching. By the time Meselson read Pauling and Corey’s paper, he had acquired sufficient confidence in his grasp of the structural considerations involved to postulate convincingly that hydrogen bonds of acceptable dimensions and angles would form between these complementary base “mis-pairs.” To what extent had Meselson’s new preoccupation with the physical properties of the purines and pyrimidines and their role in base pairing distracted him from his original plan to study DNA replication

D ENSE S OLUTIONS

161

through density differences? It is striking that in the dense trail of literature that he had pursued during the preceding weeks, he had read none of the various publications that had appeared on the “replication problem” in the sense of the debate that Delbru¨ck’s paper of 1954 had opened. Stahl’s reference to Meselson’s intention to try “some transfer experiments” implies that Meselson retained his interest in that problem, but the remainder of the summary suggests that the approach he now had in mind did not depend on density differences between substituted and unsubstituted DNA. Why had he shifted his attention to electrophoresis as the method for separation? There were three unrelated factors that fortuitously acted together to lead Meselson in this new direction. First, his interest in the differing pK values of substituted pyrimidines and purines, aroused initially by their connection with ionization and tautomerization as possible mutagenic mechanisms, appeared afterward as a means also to separate these bases. Second, a visiting Swedish scientist from the school of Arne Tiselius happened just then to be working at Caltech in the laboratory of Jerome Vinograd on the development of a new electrophoresis apparatus that was expected to provide much greater resolution than was normally achieved. Third, in discussions of their earlier idea for density separation by centrifugation, Meselson and Stahl, who had no prior experience with these methods, began to realize how uncertain the prospects were that they could attain clear-cut results. At just the time Stahl wrote Levinthal, these worries were causing Meselson to think about other approaches, among which the electrophoresis alternative seemed for a time more promising than what he had long had in mind. The differences between the degree of ionization of 5-BU and of thymine might be sufficient that if he set up a pH gradient in an electrophoresis experiment, phage containing substituted and unsubstituted DNA would move to different isoelectric points along it.6 In preparation for such experiments, Meselson looked up two papers on the pH stability and sedimentation properties of bacteriophage, one dealing with T2, the other with T7. The multiple authors of these papers (one group referred to by Meselson as “Kerby and 6 other guys”), which emanated from the Department of Surgery at the Duke University Medical School, had placed the phage in various buffers and tested their stability at different pH levels in two ways: finding the retention of infectivity by plating on E. coli “lawns” and counting the plaques formed by lysis and determining the physical

162

D ENSE S OLUTIONS

stability of the phage by sedimenting them in an analytical ultracentrifuge and examining whether there appeared a sharp sedimentation boundary characteristic of homogenous particles. By both criteria they showed that there were broad pH ranges of stability, with a sharp decline on the acid and the alkaline sides of these ranges. Meselson took notes on the relation between pH and infectivity. He copied three sets of curves showing percent infectivity as a function of pH, noting particularly “T-2 isoelectric point at 4.2.” 7 Less than three weeks after Stahl accepted the position at Berkeley, Meselson walked into his laboratory to announce that he was finished with his X-ray crystallographic work and that they were now free to begin their joint study of DNA replication. He had no trouble persuading Stahl to drop his work on bacterial genetics to make room for a new project. Stahl had just discovered the deeply embarrassing reason for his failure to identify Hfr mutants. Early in the investigation he had unknowingly switched the labels on the F⫹ and F⫺ strains and had ever since then been looking for mutant males in the female line. Stahl was only too glad to have a reason to turn his back on this fiasco, and they agreed to team up.8 Their collaboration would be short-lived if Stahl went through with his plans to leave Caltech. Meselson urged on him the point of view that, however attractive Berkeley might be (and Meselson too was much enamored of Berkeley), it would be more interesting for him to stay where he was and help figure out how DNA replicates. Enlisting the aid of Jean Weigle, and perhaps also of Delbru¨ck, who might well have told Stahl not to be a fool, Meselson soon talked Stahl into resigning his new job. The way now seemed clear for them to “map out a fairly long-term program” for which they could pool their complementary areas of interest and expertise. At Berkeley, Stent, who had put in a lot of work to persuade Wendell Stanley that Stahl was the “the man we need,” was angry at Meselson and his colleagues for spoiling his effort.9 Although the broad aim of the new collaborative venture on which Meselson and Stahl now embarked retained the Woods Hole–era plan to study DNA replication through density methods, Meselson’s new ideas about mutagenesis and Stahl’s interest in connecting 5-BU mutagenesis to his earlier phage research altered their priorities. The common starting point for the new work was to incorporate 5-BU into the T4 phage DNA. Their choice of T4, rather than the T2r⫹ phage into which Rose Litman had already been able to incorporate 5-BU, enabled Stahl to exploit the familiarity with its genetics that he had ac-

D ENSE S OLUTIONS

163

Fig. 5.1. Protocol by Frank Stahl for clonality experiment on T4 bacteriophage, in Meselson workbook

quired during his thesis research and to take advantage of the exceptional thoroughness with which Seymour Benzer had mapped the r mutants of this phage.10 The earliest record of their efforts to incorporate 5-BU is a “Protocol from Frank” for a “Clonality Experiment” that Meselson summarized in his workbook sometime before the third week in September (figure 5.1).11 This was to be an elaborate single burst experiment. They followed, with some modifications, the procedure described by Rose Litman in the manuscript on the production of bacteriophage mutants she had given Stahl during his visit to Berkeley in August. The bacterial strain E. coli Kλ was to be grown in a medium containing, in addition to the normal nutrients, 5-bromouracil, until they reached a density of 108 cells/ml. They would then be infected with three strains of T4 phage whose concentrations were set at 106 phage/ml. The wild type r⫹ and the mutant rI were controls, the experiment itself being done with the rII mutant. After incubating the bacteria and phage for five minutes, they would inactivate the unadsorbed phage with antiserum, then dilute the infected bacteria down to a level such that when they distributed them into individual tubes there would be about 103 bacteria per tube. The ratio between bacteria and phage had been planned so that there would be “one rII per [bacterial] cell in one out of every 5–10 tubes.” 12 Each tube was to be incubated at 37° for 1.5 hours, to allow the lysis of the bacteria containing phage. Here Stahl provided an opportunity for the experimenters to rest, if necessary, by putting the tubes in the refrigerator. Next they would plate the resulting phage progeny onto cultures of another E. coli strain B, by first adding 108 /ml of the bacteria to each tube, followed by a few milliliters of “soft agar” five

164

D ENSE S OLUTIONS

minutes later, and then pouring the inoculated agar onto the surface of a solid agar medium. The number and character of the plaques formed on these plates would tell them how many bacteria had been infected, how large the yield of progeny phage had been, and how many mutants had formed.13 Stahl designed and conducted the experiment, but Meselson gave him some assistance. Stahl taught Meselson how to use a springloaded pipette so that he could perform his job—to squirt broth into the tubes into which the infected bacteria were distributed.14 Meselson did not spend much time on bromouracil mutagenesis experiments. Stahl had probably brought him in on this one mainly to teach him the rudiments of working with phage and bacteria. During the next few weeks, Stahl must have carried out similar experiments in which Meselson did not participate. From the beginning of their new collaboration, Stahl believed that “Matt’s brains” were behind it and that his own principal contribution would be to furnish the knowledge of phage genetics and the experimental experience with both phage and bacteria that Meselson lacked. “Matt,” he later recalled, “wasn’t very familiar with handling anything alive.” 15 Meselson was not quite free to spend all his time on the new venture, because he still felt obligated to do something about the request Pauling had made eight months earlier to grow amino acid crystals with trace amounts of a second amino acid present. The object was to find the probability of the crystal making a mistake. Meselson foresaw that it might be difficult to tell whether the trace amino acid was specifically replacing molecules of the “majority molecule” (as he termed the amino acid present in large amounts) or was merely included in pockets of dirt. He hoped to overcome this trouble by using a racemic mixture of the two optical isomers of the majority molecule (DL), and the L form alone of the trace amino acid. If the trace molecule took the place of an L majority molecule, and after the trace acid was removed by chromatographic methods, the crystal ought to be left with a small excess of D isomer that might be measurable with a polarimeter. Choosing to work with the pair serine-alanine, he looked up their specific optical rotations and calculated the least fraction of excess D-serine that he could detect with a standard Hilger polarimeter. To increase the specific optical rotation of the molecules, he planned to use short-wavelength blue Hg light. The measurement appeared to him to be feasible.16

D ENSE S OLUTIONS

165

For theoretical guidance in his project, Meselson copied from a paper by B. Dawson and A. Mcl. Mathieson the unit cell dimensions and space groups of ten amino acids.17 Perhaps Meselson reproduced these figures with the thought that an amino acid crystal was more likely to incorporate traces of a second amino acid that formed similar crystals than of one that formed crystals of dissimilar symmetry and dimensions. He also made a calculation, based on an analysis of errors in protein synthesis by Pauling, of the “strain energy” that might be associated with the replacement of serine molecules by alanine molecules in a crystal.18 Because both of his chemically oriented projects—that on amino acid pairs and that on substituted pyrimidines—involved similar methods of purification, identification, and crystallization, Meselson continued, during the next few days, to prepare himself in parallel stages for each line of experimentation. Probably about this time he ordered from two suppliers (California Foundation and Nutritional Biochemical) commercially prepared 5-BU, DL serine, and L-alanine.19 Meanwhile, Meselson was also searching the recent literature for more help in his effort to determine the effects of halogen substitutions on the ionization of the atoms in the pyrimidine molecule that participate in base pairing. He found much of what he sought in Determination of Organic Structures by Physical Methods. Chapter 14, by H. C. Brown, D. H. McDaniel, and O. Hafliger, discussed dissociation constants. The piece began with a succinct introductory treatment of the measurement of the strength of protonic acids and bases, including derivations of the equations that expressed ionization constants in the form of pK values. Then it reviewed systematically the effects of various polar and dipolar substituents on the pK values of organic acids and bases. The authors focused particularly on substituted aromatic acids and bases, discussing the inductive effects of electronegative and electropositive substituents at the ortho, meta, and para positions relative to the affected acidic or basic group, resonance effects, especially at the para position, and steric effects at the ortho position.20 Much of this discussion appeared to Meselson to be of immediate importance to his problem. Although he had already encountered the same general principles earlier, in works such as Watson’s Modern Theories of Organic Chemistry, he now met them in a more powerful treatment, supported by the latest data, when he was most prepared to appreciate his need for specific information on the topic. He expressed his enthusiasm with the comment, “A first class review!” Then he summarized

166

D ENSE S OLUTIONS

the general points made in the chapter that were most relevant to his concerns: Electroneg[ative] substituents (ENO2, ⬅N, OH, OCH 3, Halogens) uniformly strengthen acids and weaken bases in the meta position. If the substituent possesses unpaired electrons (EOR, halogen) there is observed a smaller electron withdrawing effect from para than for meta. In some cases (EOH, OCH 3, OΦ) an actual reversal of the simple electrical effect is observed. This effect declines as heavier elements are bound to the ring. Electropos[itive] substituents (CH3, ) in meta and para positions weaken acids and strengthen bases.21 In his workbook Meselson did not state how these rules fit his own case, but the application is self-evident. Most helpful was the rule that was most “uniform”: that an electronegative substituent at the meta position strengthens acids and weakens bases—in other words, favors the removal of a proton. This was exactly the situation with 5-bromouracil, in which the bromine is meta to the N1 of the pyrimidine ring, to which is attached the hydrogen atom supposed to participate in an N ⋅⋅⋅ HEN bond with the unprotonated N1 of adenine in the normal Watson-Crick base pair. Removal of this hydrogen due to the influence of a meta-halogen would enable the N1 atom to participate in an NEH ⋅⋅⋅ N hydrogen bond with the protonated N1 of guanine. Moreover, the opposite effect of the electropositive meta CH3 in thymine was also relevant, because that situation stabilized the H on N1 for the normal pairing. The fact that the electron-drawing effect at the para position is lessened or even reversed (due, as Meselson did not note, to resonance effects) posed no inconvenience for his case, because base pairing between guanine and 5-bromouracil assumed the keto form for the oxygen para to the halogen substituent. All in all, the rules he extracted from this review of the subject could not have been more favorable to the argument Meselson wished to make for ionization as the inducement to 5-BU mutagenesis. On the next page of his workbook Meselson copied out the pK values for some of the substituents in the three aromatic rings that he considered analogous enough to the pyrimidines that he could use the data to reason about the differences between thymine and 5-BU (figure 5.2). Phenols represented the effects of substituents on the oxygen atoms at C2 and C6 in the uracil molecule. Anilines and pyridines stood for the effects on the N1H group in the pyrimidine molecule.22

D ENSE S OLUTIONS

167

Fig. 5.2. pK a values for three aromatic ring compounds, copied by Meselson from H. C. Brown et al., Determination of Organic Structures by Physical Methods; sketches (lower right) by Meselson of structures of 5-halogen uracil molecules, in Meselson workbook

168

D ENSE S OLUTIONS

Afterward Meselson sketched beside these tables, in pencil, the structures of each of the four 5-halogen uracil molecules (5-bromouracil, 5-chlorouracil, 5-iodouracil, and 5-fluorouracil). To get an idea of the comparative magnitudes of the ionization effects of these substituents, he subtracted the pKa value of the molecule containing the appropriate halogen substituent from that of the molecule containing the corresponding CH3 substituent (thymine being 5-methyl uracil), calculating the meta effects for both aniline and pyridine. The shifts that would affect the meta NH3 and the ortho O at C6 were impressively large, whereas that for the para O at C2 was relatively small. All this was eminently compatible with Meselson’s base mispairing scheme.23

II During the second half of September the problems that preoccupied Meselson were (1) how to detect the incorporation of 5-BU into phage DNA and (2) to assess the practicality of separating phage with substituted DNA from unsubstituted phage by ultracentrifugation or by electrophoresis. The paper by Litman and Pardee on the production of bacteriophage mutants that had originally focused Meselson and Stahl’s attention on 5-BU in August (and that appeared in the September 8 issue of Nature) gave no account of how, or whether, they determined the extent to which this base replaced the naturally occurring thymine of the bacteriophage DNA. Their short list of references, however, led Meselson for the second time to a paper published in 1954 by D. B. Dunn and J. D. Smith that did treat this question.24 Dunn and Smith used paper chromatography to establish that 5bromouracil or iodouracil “replaced quantitatively part of the thymine found in normal T2r and B. coli [E. Coli B] nucleic acids.” They published a table showing the relative proportions of the normal and the substituted bases in their experiments. They were also able in both organisms to separate substituted DNA from normal DNA by electrophoresis. “The separation of 5-bromouracil deoxyribonucleotide was facilitated,” they wrote, “by the fact that 5-bromouracil bears a negative charge at pH 9 (presumably due to the dissociation of the 2 or 6 (OH) group) while the other bases are uncharged. Paper electrophoresis of both hydrolysates in 0.05 M borate buffer pH 9 gave, in addition to a band containing four natural nucleotides, a second band migrating

D ENSE S OLUTIONS

169

more rapidly toward the anode.” The ultraviolet absorption spectrum of this second band was similar to that of 5-BU.25 Meselson took detailed notes on Dunn and Smith’s paper. He summarized their experimental procedures for both “B. coli” and phage T2 r and copied their data on the relative proportions of thymine and the halogenated pyrimidines. He particularly noted that “5BU & 5IU bear a negative charge at pH 9 while the other DNA bases do not. 5CU [5-chlorouracil] replaces thymine as do 5BU & 5IU. Other bases are not affected.” 26 There are several reasons why Meselson found this paper gratifying. That 5-BU quantitatively replaced thymine was essential to its feasibility as a DNA marker that he could use for transfer experiments. That it bore a negative charge where normal DNA was uncharged supported his arguments for the mechanism of 5-BU mutagenesis. Finally, that this difference in charge was sufficient to separate the substituted DNA from ordinary DNA by electrophoresis encouraged him in his most recent plans to separate substituted and nonsubstituted bacteriophage T4 by means of their different electrophoretic mobilities. The paper by Dunn and Smith was actually the first of three short papers published in Nature under one title. The other two were by Stephen Zamenhof and Gertrude Griboff. Meselson reread them also. Zamenhof and Griboff had incorporated 5-BU only into E. coli and had relied on paper chromatography to separate the base from thymine. They gave a more detailed description of the method of separation and more positive identification by ultraviolet absorption, obtaining a “spectrum identical with that of pure 5-bromouracil.” 27 Meselson copied the most essential data for both procedures: For chromatography, “In Butanol-H 20 R f 5-BU/R f thy ⫽ 1.04, in Butanol-NH 4OH, the ratio is .49.” For ultraviolet absorption, “In H 20 Emax is at 276mµ E(276)/E(290) ⫽ 1.57; in .1 N NaOH E max is at 296mµ.” He also wrote, “Gets E. coli thymineless mutant & back mutant to incorporate ⬃ 1/3 5.B.U.” and listed the substances that Zamenhof and Griboff had found to support and not to support the growth of the bacteria with and without thymine.28 Drawing from these papers the suggestion that he too might apply chromatography to determine the extent of incorporation of 5-BU in their experiments with phage T4, Meselson looked up the basic procedures for pyrimidines and purines in a standard manual of chromatog-

170

D ENSE S OLUTIONS

Fig. 5.3. Equations copied from W. Putnam, “Ultracentrifugation of Bacterial Viruses,” in Meselson workbook

raphy and copied from the chapter by G. R. Wyatt in Nucleic Acids the R f values for the natural bases in DNA and for 5-BU. Continuing to prepare simultaneously for these experiments and for his intended crystallization experiments for Pauling, he looked up at the same time the solvents used for the separation of serine and alanine. On 24 September he received two samples of 5-BU from different chemical supply houses and chromatographed them with both of the solvents Zamenhof and Griboff had used. One lot proved to be pure, the other to contain some uracil.29 Meanwhile, Meselson also continued to prepare for both ultracentrifugation and electrophoresis experiments by reading a paper by Frank W. Putnam titled “Ultracentrifugation of Bacterial Viruses” and tracking its references back to an earlier paper by Putnam, “Molecular Kinetic and Electrophoretic Properties of Bacteriophages.” From the latter he copied out the size, shape, and molecular constants of some purified bacteriophages, including sketches of the silhouettes of those shown by Putnam: T1, T5, T2, T4, T6, T3, and T7. Meselson paid special attention to the sedimentation and diffusion constants for the phages. He summarized the methods used to determine them as well as the specific volumes of the phage. Putnam explained how the apparent diameters had been calculated indirectly from Stokes’s Law and from Einstein’s equation for the diffusion of a spherical particle, and how the “molecular weights” (that is, the particle weights of the phage) were obtained from the diffusion and sedimentation constants, “using the familiar Svedberg equation.” Meselson summarized this discussion by copying from Putnam’s text the three equations in question (figure 5.3).30 The form of these notes suggests that Meselson was only now beginning to confront the necessity to learn the basic principles of ultracentrifugation that he would need to know in order to apply the method to his problem. Meselson also made a special note of a key piece of information concerning electrophoretic properties of bacteriophage T6. Putnam

D ENSE S OLUTIONS

171

had reported that T6 migrates toward the anode over “the entire pH stability range.” Meselson summarized the quantitative data: For T6 mobility ⫽ ⫺7.3 ⫻ 10⫺5 cm2 Volt⫺1 sec⫺1 at pH 8 to ⫺3.6 ⫻ 10⫺5 cm2 Volt⫺1 sec⫺1 at pH 5.1 Isoelectric point ⱕ 4.6 perhaps 4.2 31 Meselson appears to have been searching at this point for means to evaluate whether ultracentrifugation or electrophoresis was more likely to provide the resolution necessary to separate phage containing 5-BU DNA from normal phage—or at least to decide which method to try first. To help him in his assessment, he consulted the resident ultracentrifuge expert, Jerome Vinograd. A research associate in the Chemistry Division, Vinograd had joined Linus Pauling’s group in 1951, after spending ten years as an industrial research chemist at the Shell Development Company. At Caltech Vinograd carried out his own investigations, but he was also expected to provide “advice and assistance” to other experimenters. In the fall of 1956, a new chemistry building, the Norman Church Laboratory, was nearing completion. Vinograd was allocated several rooms in the subbasement of the new laboratory to organize “a set of facilities for the investigation of biologically significant large molecules, especially by methods involving heavier equipment.” The apparatus to be set up included moving boundary electrophoresis, but the most prominent heavy equipment to be available would be an analytical ultracentrifuge.32 The application of the ultracentrifuge to study the particle weight of bacteriophage was an extension to its limit of the customary use of the instrument to determine the molecular weights of large molecules, particularly of proteins. As Putnam wrote, “The bacterial viruses (bacteriophages) are the largest biologically active particles to be obtained in a state of homogeneity rivalling that of molecular proteins.” 33 Two basic methods were applied in such studies. One required the measurement of the rate at which the solute sedimented through the solution, together with the partial specific volume of the solute and the density of the solution. The second method depended on the fact that “if the solution is centrifuged for a sufficiently long time, a state of equilibrium is finally reached in which sedimentation and diffusion balance each other.” In order to calculate molecular weights from equilibrium conditions, one needed to determine the ratio of concentrations of the solute for a known distance along a centrifugal field of

172

D ENSE S OLUTIONS

known intensity, the partial specific volume of the solute, and the density of the solution.34 Because they were not initially concerned with the particle weights of their bacteriophage, but only with a means to separate substituted from unsubstituted particles, Meselson and Stahl did not need to determine sedimentation rates or concentration differences along the gravitational field. The sedimentation rates were, however, relevant to the question of whether their idea could be developed into a workable method. Meselson probably discussed with Vinograd the general capabilities of the centrifuge to separate phage particles of differing densities and to detect them through the optical systems of the machine. Meselson summarized the pertinent results of his conversation with Vinograd succinctly: “.002% detectable in Dr. Vinograd’s machine. 1.5 is good ratio of S20 for resolution. 1.1 may work.” The first number represented the minimum concentration of DNA in the centrifuge cell that Vinograd thought could be detected in the analytical ultracentrifuge by ultraviolet absorption.35 That number posed no difficulty, as it would be easy to obtain such concentrations. The other number, S20, was the notation for the sedimentation constant “given in Svedberg units (10⫺13 sec) . . . corrected to the water basis at 20°C.” The ratios 1.5 and 1.1 represented the velocity ratios of two molecules necessary to detect their separation by sedimentation. This information led Meselson to scan Putnam’s 1954 paper to find out more about the sedimentation constants. He focused on the fact that T2 displayed two constants, one above, the other below pH 5.8. Summarizing the possible explanations for this behavior, he also noted the “restricted” range of concentrations that were optimal for ultracentrifuge studies of phage.36 Important as these considerations might be for the practicality of the procedures he might adopt, Meselson saw that they did not answer the most critical question: Would the incorporation of 5-BU into phage DNA provide a sufficient difference in the density of the phage particles to enable him to separate them in the centrifuge?37 To find out, he calculated the theoretical difference in density of phage containing normal DNA and 5-BU substituted DNA (as well as of 5-iodouracil, which would be larger, because the atomic weight of iodine is greater than that of bromine). He first had to look up the proportions of the four natural bases in phage T2 (figure 5.4).38 The central conclusion— that the ratio of densities of phage containing normal DNA and 5-BU DNA was 1.03—incorporated the optimistic assumption that thymine

D ENSE S OLUTIONS

173

Fig. 5.4. Calculation of theoretical density differences between phage with normal and with 5-bromouracil or 5-iodouracil substituted DNA, in Meselson workbook

is completely replaced by 5-BU. Dunn and Smith had only been able to replace part of the thymine in phage and bacteria. In their best case, T2r, however, they had achieved about 80 percent replacement,39 so it was not unrealistic of Meselson to base his calculation on the possibility that he and Stahl might attain a complete substitution. The outcome of Meselson’s calculation was ambiguous. For molecules of similar size and shape, the ratio of sedimentation velocities would be the same as the density ratio. The sedimentation difference in water would, therefore, be at best only a third of that which Vinograd thought minimally necessary to separate the particles. Meselson and Stahl did not plan, however, to centrifuge the phage in water but in a medium of density intermediate between the substituted and unsubstituted phages. In such a medium the particles would move much more slowly than in water, but the ratio of their rates of sedimentation would be increased. The critical question was, therefore, whether the calculated density ratio would permit them to separate the substituted from the unsubstituted phage within a length of time that it was practical to run the centrifuge. On 24 September, the same day he chromatographed 5-BU, Meselson worked out a theoretical answer to this basic question. Under the heading “High density liquids for phage centrifugation,” he first estimated the density of phage by taking the reciprocal of the specific volume reported in Putnam’s table. It was 1.51 g/ml. By now Meselson had studied the chapter on the elementary theory of sedimentation in Svedberg and Pederson’s The Ultracentrifuge.40 Meselson wrote down the fundamental Svedberg equation relating the molecular weight of a solute (M) to its sedimentation velocity (S), its specific volume (v¯ ), the molar friction constant ( f ), and the density of the solution (ρ): S⫽

M(1 ⫺ v¯ ρ) f

174

D ENSE S OLUTIONS

Fig. 5.5. Calculation of centrifugation time required to separate 5-bromouracil from normal phage by sedimentation, in Meselson workbook

He then decided to set the desired density of the solution so that the sedimentation rate of 5-BU phage would be twice that of unsubstituted phage, providing a comfortable margin above the ratio of 1.5 that Vinograd believed sufficient. To do so, he defined S′ as the sedimentation rate for 5-BU phage and S as the sedimentation rate for normal phage and set S′/S at 2. He then used the Svedberg equation shown above to solve for the density of the solution, ρ, that would satisfy this condition. The density of 5-BU phage he obtained from Putnam’s value for the specific volume of normal T2 phage, multiplied by the ratio 1.03 that he had previously calculated from the molecular weights of thymine and 5-BU. He made the simplifying assumption that the frictional coefficients (f) in the equation were equal for substituted and unsubstituted phage and arrived at the required density, ρ ⫽ 1.47 g/ml. The next step was to determine the sedimentation velocity for phage centrifuged in a medium of this density. To do so he corrected the S 20 values given by Putnam for a “water basis” by inserting in the Svedberg equation the ratio between the difference of the density between phage and the medium and the difference of the density between phage and water. The result, he wrote, was “⬃ .5 which would require centrifuge times ⬃ 5 hours.” In his workbook Meselson summarized these calculations in shorthand form (figure 5.5).41 Meselson and Stahl never seriously considered carrying out an experiment in which they would attempt to separate substituted from unsubstituted phage by the difference in their sedimentation rates. Going through this exercise helped Meselson, however, to understand the system with which he intended to work. The general outcome of his deliberations probably also encouraged him to believe that their original plan—to separate the phage in a medium in which the substituted phage sank while the unsubstituted phage floated—was feasible. The difference between the densities of phage and medium would be of similar magnitude to the difference he had used in his calculation.

D ENSE S OLUTIONS

175

The separation ought, therefore, to be attainable in a centrifuge run of reasonable duration. Now Meselson shifted from theoretical analysis of the situation to a practical search for a solution that would produce the needed high-density liquid. The search began playfully. As Meselson related the story to Horace Judson: the first experiment we did was at dinner. We had sugar on the table . . . and I just put sugar, a lot of sugar, in a glass, and filled it with water, and then cut off a piece of fingernail and dropped it in—just to see if you could float, in a solution like that, materials of the density of DNA. I mean, we didn’t know exactly the density of DNA . . . but fingernail seemed a reasonable analogy. And as I remember, the fingernail sinks, even in the strongest sugar solution. [Then we added salt (giving salt and sugar), but the fingernail still sank.] So we needed something denser. We went to [a large oilcloth periodic table that hung] in the guest room and said, “Well, we want something like table salt, sodium chloride, but very dense,” so we read straight down the chart from sodium to the heavy elements that are chemically similar—sodium, potassium, rubidium, and then there’s cesium, which is the last naturally occurring element in the group. This story, vividly remembered, logically compelling, and attractively dramatic, seems so fully to explain Meselson’s choice that it is not surprising that Judson treated it as the definitive account of an event whose significance he compares to the famous “Eureka” story of Archimedes.42 It was, however, only the first scene in a longer story, the rest of which was less theatrical but more conclusive. The serious phase of Meselson’s search began when he looked up the section on the specific gravity of aqueous solutions in the most convenient reference work available, the Handbook of Chemistry and Physics. The Handbook gave detailed tables of the relation between specific gravities and concentrations for a select list of about seventy solutes, mainly inorganic acids, bases and salts, and a few organic substances.43 The substances were arranged in alphabetical order. Working through the tables, Meselson picked out and wrote down, in the order he found them, the substances—ten neutral salts, glycerol, and sucrose—whose densities at the highest concentrations given were close to the calculated density (ρ ⫽ 1.51) of bacteriophage (figure 5.6). A further requirement for a medium suitable to their purposes was that bacteriophage be able to survive in it. They anticipated that some

176

D ENSE S OLUTIONS

Fig. 5.6. Solution densities copied from Handbook of Chemistry and Physics, in Meselson workbook

of their experiments, particularly those conducted by Stahl, could be done with very small numbers of phage, if they could later be detected through bio-assays such as the formation of plaques. It was, therefore, “important that the virus not die.” 44 On 25 September they screened each of the solutions Meselson had listed the day before for this prerequisite. With phage T4B stock given them by George Streisinger, they added 1.0 ml of each of the designated solutions to 0.1 ml portions of diluted suspensions of the phage in a standard dilution medium (M-9). They allowed the mixtures to stand at room temperature for about two hours, then diluted them slowly in nutrient broth and plated them on a bacterial lawn.45 The results of the tests of these potential centrifuge solutions were not favorable. Two of them, KBr and MgSO 4, produced about onethird as many plaques as the controls containing the dilution medium alone. The rest of them yielded few or no plaques.46 These environments were clearly not conducive to the survival of phage. (Retrospectively, Meselson attributes the death of the organisms to osmotic shock, which might have been avoided if he had diluted the solutions more slowly.)47 Ever since their conversations at Woods Hole, Meselson and Stahl had had in mind a multistep separation strategy. First they would run the phage in a medium of density intermediate between the densities of substituted and unsubstituted phage. The denser substituted phage would sink, while the lighter unsubstituted phage would float to the top. The “light-heavy” phage that they expected to result from the rep-

D ENSE S OLUTIONS

177

lication of substituted phage DNA reproducing in a normal medium might either sink or float. They would then adjust the concentration of the medium until they were able to separate, in further centrifuge runs, the light-heavy phage from the heavy or the light phage that it would have accompanied in the first run.48 Preliminary to these experiments, they planned to try to float normal virus in the centrifuge by means of a fluid medium substantially denser than either normal or substituted phage. If that proved successful, then they could subsequently lower the concentration of the solution until its density was intermediate between that of normal and substituted phage. With these criteria in mind, Meselson looked for a solution of density well above 1.55 g/ml. Knowing that the handy Handbook for Chemistry and Physics was inadequate to this purpose, Meselson turned to the comprehensive International Critical Tables of Numerical Data. He found what he needed in a four-page section containing tables of the density of saturated solutions. The tables listed the concentration and densities of each substance included over various ranges of temperature. Meselson scanned the tables for substances whose densities at temperatures near 20°C were between 1.6 and 2.0 g/ml. On the first page he came to several, but he passed them over, because they were salts of divalent metals, which he thought not suitable, because divalent cations could form cross-linkages between the DNA molecules that might interfere with their separation.49 In the first column of the second page he encountered Ca(ClO3)2, with a density of 1.729 at 18°. He wrote it down even though it was divalent, along with Sr(ClO3)2, density 1.839 at 18°, and Ba(ClO4)2, with density of 1.912 at 20°. The long list of lithium, sodium, and potassium salts included only two or three possibilities within the required range, and he did not stop even at those. Arriving at rubidium salts, he saw that RbCl at 1.4893 was a little too light, but RbBr came in at 1.6252 at 23°, and he added it to his list. He passed over RbI, even though its density at 24.3° was 1.8479, because the liberation of free I2 would kill the phage. The last group of tables was composed of the cesium salts. Three of them, CsCl (1.9095 at 20°), CsBr (1.6933 at 21.4°), and Cs2SO4 (2.0063 at 20°), fell within the prescribed limits, but he put down only the first and last of these, probably judging their greater densities to be advantageous.50 Next to the column in which he listed the densities of these salts, Meselson headed a column “Fw” (for formula weight) but filled in only the value for CsCl, an indication that perhaps, of the six possibilities, he already

178

D ENSE S OLUTIONS

favored this salt. To avoid divalent salts he eliminated the first three substances on his list, leaving only RbBr, CsCl, and Cs2SO4. The density of the saturated solution of the rubidium salt was too close to that of the substituted phage to allow much of a margin for adjusting the concentrations used, but it remained a viable possibility. The choice between the two cesium salts was probably arbitrary, and either one would have worked.

III Settling on a cesium chloride solution as the most promising ultracentrifuge medium, Meselson set out systematically to obtain further information about its physical properties. Along the top of a page of his work book headed “Data for CsCl,” he listed the types of numerical data he hoped to gather. They included the formula weight (Fw), mole fraction (N), Molality (m), Molarity (c), density at 20° (ρ 20), density at ˚ ((δη/δc)10 3), relative 25° (ρ 25), molar refractive increase at 5460.74 A viscosity (η/η 0 ), the diffusion constants, and the activity coefficients (γ). The most immediately necessary data were the densities and their corresponding molar concentrations, from which he could calculate the weight fractions in order to learn how to weigh out the relative quantities of cesium chloride and water necessary to construct a solution of the desired density. The other data, which were related to the optical properties of the solutions and to equations used in calculations frequently carried out in conjunction with the analytical ultracentrifuge, suggest that Meselson was looking forward to experiments beyond the immediate ones he expected to begin with the preparative centrifuge. To fill out his table, he looked up, first, an article by Philip Lyons and John Riley on the diffusion coefficients for aqueous solutions of calcium chloride and cesium chloride that included also some data on relative viscosities. Meselson filled in the portions of his table to which the data applied, then turned to R. A. Robinson and R. H. Stokes’s textbook Electrolyte Solutions for the activity coefficients. From the densities and concentrations Meselson calculated weight fractions covering the range of densities pertinent to the planned experiments, but he left much of the rest of the space he had provided for data blank.51 Continuing his search for relevant data, Meselson set up a similar format on the next page of his workbook and went back to Interna-

D ENSE S OLUTIONS

179

tional Critical Tables to try to fill it. He found a set of refractive indices for cesium chloride for the Sodium D line and of molal solubilities at ten-degree temperature intervals and entered both on his tables, again leaving empty much of the space for data he had intended to find. From a paper by Herbert Harned and Orion Schupp he found a list of the activity coefficients for cesium chloride in aqueous solution and copied them below his mostly empty headings. He also copied the equations used in the paper to calculate the coefficients. An indication that he still entertained the possibility of using the other cesium salt is that he wrote down in a lower corner of this page a list of densities and corresponding weight fractions for Cs 2SO 4. In an old paper in the Journal of the American Chemical Society he found some additional density data for CsCl and entered it on the next page of his work book.52 His partial success in locating the data he thought he would need illustrates that published data are generically essential but seldom fully adequate to the specific purposes of the individual investigator. Often they serve merely as a starting point, guiding the investigator mainly to learn what further data he must produce for himself. Ready by now to begin preliminary experiments, Meselson made arrangements to use a Spinco Model L preparatory ultracentrifuge belonging to Renato Dulbecco that was housed in the basement of the Church Laboratory.53 Meselson went to the stockroom to obtain some cesium, but none was available. He was able to get a salt of rubidium, the element closest to cesium on the periodic table, and decided to try it while he ordered cesium chloride from the American Potash and Chemical Company. In his workbook he wrote down the density (ρ ⫽ 1.86) and refractivity of a rubidium formate solution. The density appeared ample for their first objective, to float T4 phage, but it is more likely that the salt they were able to procure from the supply room was rubidium chloride, whose maximum density was marginal for this purpose. To stabilize the phage they added magnesium acetate. After mixing the phage with the solution, they placed it in the plastic tubes of the Model L centrifuge and spun them. When the run was over, they drew off a little bit of fluid from the meniscus at the top of the tubes and looked for phage, but it had not collected there. Rather than continue with rubidium chloride, they decided to postpone further experiments until the cesium chloride arrived.54 Meanwhile, Stahl concentrated on the problem of growing 5-BU substituted stocks of phage that would produce mutants. In his early efforts the frequency of mutation was not much above normal. Still

180

D ENSE S OLUTIONS

following the leads provided by Rose Litman’s work, he noted that she had obtained the best results when she diluted the bacteria infected with phage into “spent medium,” that is, into a medium, containing 5-BU, in which the bacteria had already grown for three hours. Following a similar procedure, Stahl was able, by mid-October, to attain “high frequencies of plaque morphology mutants . . . in T4r stocks.” Meselson did not play an active role in this effort, but he probably hung around enough to learn from Stahl some of the techniques, as well as “the lore . . . and the flavor” of phage work.55 About this time Meselson planned a trip to San Francisco, with two purposes in mind. One was to consult Howard Schachman, at Berkeley, about ultracentrifugation. The other, perhaps stronger motivation was a weekend date with a girlfriend, Anic¸a Vesel. He asked Stahl to come along, with the idea that Stahl could check in with Rose Litman about her further progress on 5-BU incorporation into phage. Never having seen San Francisco, even during his quick trip to Berkeley in August, Stahl was happy to go. With the top down on Meselson’s sleek black Thunderbird, they wound their way along the deeply indented coastline of California Route 1. For Stahl, the spectacular views of the sea and the steeply rising headlands were new and pleasurable experiences. They stopped for a while at Big Sur, then drove to Monterey, where they picked Ani up at her parents’ home. Stahl thought that Ani was stunningly beautiful, and he did not mind that the three of them fit rather tightly in the two-seated Thunderbird.56 At Berkeley, Stahl particularly wanted to find out more about how Litman and Pardee determined the extent to which 5-BU substituted for thymine in the phage DNA in their experiments, because their paper had not discussed that question. He did not receive a definitive answer, because they were not yet certain of the factors that affected the degree of substitution. He did learn that they were making progress in that direction, and he was relieved that it appeared that he and Meselson might “be spared this aspect of the work.” 57 Knowledge of the degree, and particularly the homogeneity, of the substitution would be critical to their success, because the separation of substituted from unsubstituted phage required that the density of each kind be uniform. Meselson had good reason to consult Howard Schachman about ultracentrifugation, because Schachman had extensive experience, not only with the general techniques but specifically with the Spinco Model E. One of the first biochemists to use the machine when it be-

D ENSE S OLUTIONS

181

came commercially available, he had himself contributed advances in the methods used for its application to the study of biologically important macromolecules. Together with E. G. Pickels, a staff scientist at the Special Instruments Corporation, which manufactured the Model E, and W. F. Harrington, he had devised in 1952 a special centrifuge cell into which one could introduce layered solutions of different density. “Density gradient” methods improved the ability of the centrifuge to analyze mixtures of macromolecules of differing sedimentation rates.58 Later he became particularly interested in the use of ultracentrifugation methods to determine the molecular weights of the degradation products of proteins and of DNA. In the case of DNA, he sought evidence to test the “interrupted two-strand model” he had proposed in 1954 (see Chapter 1). These objectives pressed him to devise procedures to deal with substances of lower molecular weight than those normally measured in the ultracentrifuge and to work with unusually dilute solutions.59 By 1956 Schachman had concluded that “the limitations of ultracentrifugation reside in our inability to measure precisely the partial specific volume of the materials under investigation.” In an application for a National Science Foundation grant in October, he described a new method to measure the density of solutions and solvent that involved a density gradient “produced by the pressure gradient present in any ultracentrifuge cell.” Filling a special divided cell with a silicone fluid slightly less dense than the solution and solvent whose density he wished to measure, he placed a drop of the solution in one side and a drop of the solvent in the other. The drops sank to the bottom of the fluid, but as the centrifuge rotor began to spin, the compression of the silicone fluid, increasing with the square of the speed, caused the drops to rise until they reached an equilibrium position where their density equaled that of the fluid. The density of the drops of solute and solvent could then be calculated to an accuracy of one part in a million.60 When Meselson visited him in his laboratory, Schachman was apparently carrying out similar measurements using hollow glass beads in a cesium chloride medium. He may have been preparing a way to adopt his method to denser solutions and solvents such as those containing DNA. In his grant application he wrote that he intended to place greater emphasis on the study of various types of DNA, “in particular, on phage DNA.” 61 Schachman, who was wearing his habitual white lab coat when

182

D ENSE S OLUTIONS

Meselson came in, received his visitor warmly. Meselson had once audited his undergraduate physical biochemistry course at Berkeley and had impressed him then as a brilliant student. When Meselson discussed his experimental plans with Schachman, and mentioned that he was using rubidium chloride for the solution, Schachman suggested that cesium chloride would be far better. Meselson perhaps tactfully did not say that he intended to switch as soon as he received some cesium.62 Meselson must have received his first supply of cesium chloride soon after their return from Berkeley. Sometime before the third week of October, he and Stahl put some unsubstituted T4 phage into a concentrated cesium chloride solution and centrifuged it in Dulbecco’s preparatory machine. After the run, they found the phage concentrated in the fluid that they drew from the top of the centrifuge tube. Undoubtedly somewhat elated at the success of this first step in their plan, they were now “almost certain that separation of substituted from non-substituted phage may be affected [sic] by employing a concentration which gives a density intermediate to the two types of phage.” 63

C HAPTER S IX

The Big Machine

I On 21 October 1956, Frank Stahl typed out a three-page “5-BromUracil Prospectus,” intended probably to let Max Delbrck know what he and Matt Meselson were planning.1 The title expressed the common grounding in this molecule of a series of investigative projects that were similar to those described by Stahl five weeks earlier in his letter to Levinthal. (See pp. 158–159.) Stahl’s decision to remain at Caltech enabled him to integrate what he had then described as the separate plans of Meselson and himself. Diverse as the individual research problems appeared, they could now be viewed as “lots of threads that interweave.” 2 His success in his initial efforts to produce mutant stocks of 5-BU substituted T4 phage encouraged Stahl to elaborate his plans for their genetic characterization. He intended to map the mutations to see whether they gave “classical segregation patterns,” and he had already asked Seymour Benzer to map those in the rII region for comparison with the fine structure maps Benzer had produced for normal T4 phage. With “non-viable members of a substituted stock” he planned cross-reactivation experiments to determine whether the reactivation was “primarily by crossing over or by the replacement of lost gene functions.” After that, he envisioned multiplicity reactivation experiments, whose nature would depend on the cross-reactivation results.3 Briefly Stahl summarized their expectations for “the separation by ultra-centrifugation of substituted from non-substituted phage,” based on Meselson’s calculation that the former should be about 3 percent heavier than the latter. Reflecting the encouragement that the intervening weeks had brought to their hopes for that method, Stahl now treated the electrophoretic method that Meselson had recently favored

184

T HE B IG M ACHINE

as no more than a choice to fall back on later if necessary: “Electrophoretic separation of substituted from non-substituted phage may be possible (in the event that centrifugation, which looks more promising, fails.)” 4 The auspicious start that they had made on both of the major thrusts of their new joint venture emboldened Meselson and Stahl to add some heady embellishments to their ambitions. If the separation of substituted from unsubstituted phage by centrifugation “is a roaring success,” Stahl wrote, “transfer experiments may be conducted. The ‘20% piece’ idea of Levinthal may be checked. If possible, then the experiment will be done with genetically mixed infection. A possible further refinement of the transfer experiments involves the use of a suspension medium whose density is graded along the length of the cell. A suspension of phage with varying densities should be highly fractionated under this condition. Information on the distribution of ‘small DNA pieces’ may therefore be obtained.” 5 The “20% piece” idea and the “small DNA pieces” referred to the claims Levinthal had made the previous spring, on the basis of his 32P experiments, that T2 phage contains one large piece and several smaller pieces of DNA. (See above, p. 105.) His ongoing correspondence with Levinthal kept Stahl informed about the status of this idea. In July 1956, Levinthal reported to him that “Charlie Thomas has found that T4, when shocked, shows a big piece of DNA about the same size as T2” and that “Geoffrey Brown has been able to show, using emulsions, that the big pieces and the little pieces separate on his histone columns.” 6 In his letter to Levinthal in September, Stahl asked whether he had “any new facts on genotype versus big piece.” 7 Stahl was both interested and somewhat skeptical. The prospect that ultracentrifugation could fractionate components of phage into particles of different densities provided for him an inviting means to subject the idea to a critical test. Meselson probably agreed to the inclusion of this test in their prospectus, but he was less interested in the question.8 The suspension medium of graded density that Stahl had in mind would be produced by carefully layering into a centrifuge cell solutions of decreasing density of a given solute. Although the density layers gradually diffuse into one another, the gradient lasts long enough to carry out experiments lasting several hours.9 Of the several other prospects Meselson and Stahl outlined for their young collaboration, their design for a possible sequence analysis of the bases along the DNA molecule was the most daring: “A variety of rII mutants will be screened for mutability by 5BU. If some rII mu-

T HE B IG M ACHINE

185

tants are r by virtue of the fact that a certain nucleotide pair is cytosineguanine and some by virtue of their being adenine-thymine, these two classes may show a different back mutation response to 5BU. Although both classes should give back mutants with 5BU (since guanine might be expected to direct the incorporation of 5BU about as often as 5BU directs the incorporation of guanine), only the latter class should produce phage unstable when passed through a subsequent growth cycle in the absence of 5BU.” (For a clarification of Stahl’s elliptic explanation, see below, pp. 440–441.) When Meselson copied this paragraph into a similar prospectus two days later, he added a final sentence: “By working with pairs of mutants, one might even characterize a base pair with respect to purine pyrimidine order.” 10 The great significance of the sequence of base pairs in DNA had been obvious since Watson and Crick made it the foundation for the genetic code. Speculations about the general nature of the code were rife in 1956, but no one thought that the means would soon be at hand to determine base sequences experimentally. That Meselson and Stahl imagined they might be able to establish even the order of two adjacent base pairs at a few specific sites was a type of stargazing in which bright young scientists were perhaps encouraged to indulge themselves by the ethos at Caltech. Shortly after Stahl wrote this memo—possibly on the same day— he and Meselson complemented their previous flotation experiment in the preparative centrifuge by sedimenting unsubstituted virus in a cesium chloride medium of lower density and recovering them quantitatively from the bottom of the cell.11 The next step, according to their plan, was to separate substituted and unsubstituted phage in a medium of intermediate density. There is no record of this experiment, and Meselson and Stahl differ in their recollections of whether and how it may have been carried out. Meselson remembers having planned to ascertain the locations of the phage after the centrifuge run by piercing the bottom of the plastic centrifuge tube, collecting the contents drop by drop in separate tubes as they ran through the hole, and plating each drop individually to determine whether it contained phage. When he tried it, the drops fell so rapidly that he could not catch them in separate tubes. Although it would have been easy to devise a method to slow the rate, he decided that the method was very tedious and had the further disadvantage that one could detect only viable phage. He could obviate these problems by running the experiment in the new analytical ultracentrifuge in Jerry Vinograd’s labora-

186

T HE B IG M ACHINE

tory, where he could see what was happening through its optical system. Stahl does not remember collecting drops then, and he attributes the decision solely to the desire to see what was happening.12

II The Spinco Model E analytical ultracentrifuge was a formidable machine. Standing about six feet high and more than seven feet long, it dominated the space in which it stood. Though massive in overall appearance, the analytical ultracentrifuge was also an elegant instrument of delicately refined design. Its graceful ellipsoid rotor, machined from an aluminum alloy forging and measuring 7 1/4 inches in length, spun at speeds of up to 60,000 RPM, while suspended by a slender piano wire from the electric motor that drove it (figure 6.1).13 Much of the space behind the symmetrical fac¸ade of the Model E was filled by auxiliary systems providing for safe and flexible operation and for the optics required to view what happened within the compact centrifuge cells as they underwent gravitational forces ranging up to 289,000 g. The rotor and motor themselves occupied only the middle and upper modules in the leftmost of the three sections into which the machine was divided vertically. Although the selfbalancing rotor possessed a durability resulting from long years of refinement in design, the risk that a metal object subjected to such stress might disintegrate was still sufficient that the Model E was provided with a ring of cast steel, 1 3/4 inch thick, that surrounded the rotor during its operation. This heavy cylinder occupied the lower third of the left side of the machine while the rotor compartment was open for loading and unloading. An electric motor elevated the cylinder into place before each run began.14 In order to minimize frictional heating and avoid thermal gradients, the machine contained a vacuum system that evacuated the rotor chamber to a pressure of 1 micron. A two-stage mechanical pump was first switched on, producing a partial vacuum. Then an oil diffusion pump further reduced the pressure to the final level. These pumps were located behind the bottom panel in the central section of the machine. The lower right side contained a refrigeration system that enabled the rotor to operate at any desired temperature between 0°C and that of the room. The speed control unit was located in the upper panel of the center section. Essentially a precision mechanical gear

(a)

(b)

(c) Fig. 6.1. (a) General view of Spinco Model E analytical ultracentrifuge; (b) standard rotor; (c) insertion of rotor, from Beckman/Spinco brochure, Ultracentrifuge

188

T HE B IG M ACHINE

box, this unit enabled the operator to spin the rotor at any of thirty speeds. The speeds were set by the numerical ratios of the gears and resulted in rather odd-looking sequences, such as 12,590; 13,410; 14,290; to 52,640; 60,000 RPM. Critical to the effectiveness of the Model E was that the machine was capable of maintaining these speeds to a mean accuracy of 0.1 percent.15 The materials to be centrifuged were inserted into specially designed ultracentrifuge cells. Cylindrical in shape and inserted in the rotor into a cylindrical hole whose axis was parallel to that of the axis of rotation, these cells were simple in overall appearance but complex in detail. Viewed from above, the centerpiece of the basic “single sector cell” was a small cylinder containing two quartz windows through which the light of the optical systems could pass. The cavity in the centerpiece was shaped like a sector, its sides placed at an angle to one another equal to 4 degrees of arc. The length of the sector, 15 mm, defined the “depth” of the cell in the direction of the centrifugal field. The “thickness” of the cell, that is, the length of the cylinder perpendicular to the field, varied from 1 1/2 to 30 mm, the choice of lengths being made according to the concentration and optical densities of the substances placed in the cells. When disassembled, these cells consisted of sixteen different parts, including the cell housing, the plastic centerpiece and windows, screw rings to hold the parts together, and gaskets to keep them from leaking (figures 6.2, 6.3).16 The machine housed two separate optical systems. The Schlieren system, with a cylindrical lens, was a method for representing density

Fig. 6.2. Single-sector analytical ultracentrifuge cell from Beckman/Spinco brochure

T HE B IG M ACHINE

189

Fig. 6.3. Exploded view of parts composing analytical ultracentrifuge cell

gradients along the cell by means of the changes in the refractive index associated with differences of concentration. Light shining through a cell with uniform refractive index passes undeviated. Where there is a boundary in the cell, however, the light will be deviated to a degree proportional to the refractive index gradient at that point. The various Schlieren methods, named after the German word for “streak,” function “by interfering with the deviated light in some manner or other.” 17 In the system incorporated into the Model E, the light leaving the sample cell passed first through a condensing lens, then through a “Schlieren analyzer,” consisting of a diagonal slit. Then followed a camera lens and a special cylindrical lens, the rays finally passing to a photo-

190

T HE B IG M ACHINE

Fig. 6.4. Schlieren line at successive stages in the course of sedimentation of a single component solute, from Beckman/Spinco brochure

graphic plate. By means of effects on the light rays that are too complex to be summarized here, the system produced on the plate a curve whose horizontal coordinate represented the distance along the centrifugal field of the centrifuge tube and whose vertical displacement represented the refractive index gradient, and hence the density gradient, at that point along the cell.18 A typical set of Schlieren lines from the Model E system, tracing the stages in the sedimentation of a single component, is shown in figure 6.4.19 For a solute denser than the solvent, the movement of the peak of the curve from left to right corresponds to the movement of the boundary between a high concentration of the sedimenting solute and the lower concentration of pure solvent in the portion of the cell to which the solute has not yet sedimented. In the Model E, the source of light for the Schlieren system was located below the rotor, its vertical beam aligned with the axis of the cylinder containing the centrifuge cell. After passing through the condensing lens, the light beam was deflected 90° by a mirror. The rest of the optical components were placed horizontally along the top of the machine, the photographic plate being inserted and removed at the upper right end. A swing-out mirror located in the path of the light just before the light reached the plate enabled the operator to divert the beam forward, into a viewing aperture. In this way, it was possible to watch the changing appearance of the “Schlieren line” during the course of an experiment.20 The second optical system detected the absorption of ultraviolet light by substances contained in the centrifuge cell. The ultraviolet light source was also placed below the rotor, its beam passing vertically through the centrifuge cell at a different point in its rotational pathway. The ultraviolet beam also made a 90° turn at the top of the machine, ending its trajectory at a photographic plate located at the top right of the machine. To obtain monochromatic light, a filter that permitted light of only a single wavelength to pass was inserted be-

T HE B IG M ACHINE

191

tween the mercury discharge light source and the sample. Although, in principle, the operator could insert different filters, depending on the maximum absorption characteristic of the material under study, most often the 2560A filter with which the machine came equipped was adequate.21 When the films inserted in the photographic plate holders were developed, regions of high absorption appeared light, those of low absorption dark. Because the cell was passing very rapidly through the optical beam, the photographs could not record images of objects in a cell but could only differentiate regions in the direction of the centrifugal field. Boundaries were seen sharply along that dimension.22 From his perusal of The Svedberg and Kai Pederson’s massive monograph The Ultracentrifuge, Meselson was aware of the long history of the instrument. The first low-speed ultracentrifuge had been designed and built for Svedberg’s laboratory at the Institute of Physical Chemistry in Uppsala, Sweden, in 1924. By 1927, Svedberg and his associates had advanced to a high-speed ultracentrifuge with oilturbine drive, which could operate at speeds up to 40,000 RPM. In the 1940 monograph, Svedberg described the succession of improved designs, machine by machine and rotor by rotor, recounting in particular the circumstances under which many of the early rotors had exploded.23 During the pre-war period, each ultracentrifuge machine was a custom installation, a heavy and cumbersome complex requiring a large, well-reinforced laboratory room. A new era began in 1947, when E. G. Pickels designed, at Spinco Instruments of Belmont, California, the first commercial instrument which contained all the equipment and controls necessary for precise ultracentrifuge measurements.24 The Caltech Chemistry Department had already possessed, since about 1947, an analytical ultracentrifuge designed and built in its own shop—a huge, noisy, air-driven machine that occupied a large room. The rotor and centrifuge cells were supplied by Spinco, but the heavy assembly was devised locally. Unlike the commercial Spinco instrument, the Caltech machine incorporated a bottom pin to prevent excess wobble at low speed. The machine was employed for a number of research projects during the ensuing years, but as Vinograd wrote to Pickels, its use had “been severely limited by a tendency to develop, unpredictably, precessions and vibrations. This makes it difficult to plan longer term work, limits the number of people that may qualify as operators, and besets the operation with a certain amount of nervous tension.” Vinograd hoped that Spinco might be able to modify the

192

T HE B IG M ACHINE

machine so as to stabilize it.25 Whether or not Spinco made such an effort, the homemade Caltech ultracentrifuge stood idle by the time Meselson and Stahl arrived on the scene.26 During the early months of 1956, Vinograd was in frequent correspondence with Spinco concerning the construction of the Model E that was intended for his new basement laboratory in the Church Building. That there was still some custom preparation required for each unit is suggested by a letter from Spinco in March 1956 concerning the installation of the UV system, one in April about replacement of an earlier type phase plate, and one in June from Pickels himself, reporting that since Vinograd’s last visit to Spinco, he had “checked into the question of the temperature control in our Model E machine.” In August the machine may have been in place at Caltech, but something must have gone wrong, for the institute had to pay for the replacement of its drive. When it was finally ready for operation in October, it bore the serial number 186 (figure 6.5).27 On 22 October, Meselson participated in the first experiment carried out on phage T4 on the Model E analytical centrifuge. First, he probably prepared the suspension of phage in cesium chloride in his

Fig. 6.5. Jerome Vinograd in front of his Model E Analytical Ultracentrifuge. Photo courtesy of the Archives, California Institute of Technology

T HE B IG M ACHINE

193

own laboratory. The density of the cesium chloride solution, calculated from the measurement of its refractive index, was 1.6022 at 21.3°C. Having previously estimated the density of the phage at 1.51, he chose this concentration of CsCl to produce a density difference just large enough to float the phage. In the suspension itself, Meselson calculated, there were 8.3 ⫻ 1011 particles/ml, comprised of 0.02 percent DNA and approximately 0.04 percent solid matter in all.28 Vinograd signed the centrifuge log and led Meselson through the procedures for preparing the centrifuge cell and operating the machine. The cell had to be very clean. Meselson assembled the centerpiece, the quartz windows, and the rings and gaskets in the aluminum cell housing, then tightened the screw ring with a special wrench to an exactly measured amount of torque. With a hypodermic needle he sucked some of his phage suspension out of the test tube and injected it through a small filling hole into the cell. He punched a little circle out of a piece of red plastic with a cork borer and inserted it into the hole, then tightened down a screw binding to seal it. The next step was to prepare a counterweight by placing little brass screws in it until it exactly balanced the cell. Then he was ready to insert the cell into the rotor. Again he had to ensure that both the cell and the cylinder within which it fit in the rotor were very clean; otherwise the cell might jam inside the rotor. To prevent them from binding, he coated the inside of the rotor opening with a molybdenum lubricant. After inserting the cell, he used a special tool to rotate it until it was perfectly aligned, according to matching lines on cell and rotor. After he had inserted the counterweight on the other side of the rotor, he carried the rotor to the centrifuge, opened the door in front of the rotor chamber, and used another special tool to fasten it to the shaft from which it would be suspended.29 When he closed the chamber door, Meselson heard the clunking sound which indicated that the protective ring had been lifted into place. He started the mechanical vacuum pump. Vinograd cautioned him never to switch the oil diffusion pump on until the vacuum gauge showed a pressure of one or two millimeters—otherwise, he might ruin the pump. At 2:37 P.M. Meselson turned the rotor on. He let it idle for a few minutes, while the chamber reached its operating pressure, then clicked the speed control knob through the settings that appeared in a window on the center console, until he reached 52,640 RPM, near the 60,000 RPM limit.30 The electric tachometer on the control panel showed that the machine was accelerating rapidly. Within

194

T HE B IG M ACHINE

seven minutes it reached the desired RPM, and the yellow acceleration light on the panel turned off.31 Sometime before starting the run, Meselson had gone behind a curtain into the darkroom adjoining the room in which the centrifuge stood and loaded film into several cassettes. He felt the notches on one edge of the film to make sure that he was not putting the film in backwards. Returning to the machine, he mounted the first cassette in its holder behind a door in the upper right corner of the console and set the shutter speed of the camera for the exposure he wanted. As the run began, he switched on both the Schlieren and the ultraviolet absorption optical systems.32 After the centrifuge reached its operating speed, Meselson checked the Schlieren line by viewing through an eyepiece the projection of the line onto a ground glass plate. If the cell were leaking, the Schlieren system would immediately reveal the descent of the meniscus. Apparently that was not happening. The Schlieren line began to rise a little, as would be expected, because the pressure on the fluid would compress it, making its density increase slightly with depth. As Meselson continued to observe the Schlieren line, however, he was “astounded” to see that it kept moving slowly upward for the entire 2 1/4 hours that the run lasted. (There must also have formed a narrow peak in the line representing the moving boundary of the phage.)33 The sharp boundary detected by the Schlieren line and verified when Meselson developed the first ultraviolet absorption films, which showed a boundary moving upward in the cell, indicated that the phage was “homogeneous with respect to sedimentation velocity.” This result, confirming the earlier results of Frank Putnam for several other types of phage,34 was overshadowed in Meselson’s mind by the steady rise of the Schlieren base line. That meant to him that “something else was happening. . . . The cesium [chloride] was sedimenting.” That cesium chloride could be redistributed under the high gravitational force of the ultracentrifuge, so that it formed a concentration gradient, was not itself novel information. Meselson had looked at an old paper by Svedberg’s associate, Kai O. Pederson, on the sedimentation equilibria of inorganic salts in the ultracentrifuge, which recorded results for seven salts, one of them cesium chloride. At equilibrium, Pederson had reported in 1934, the concentrations of each of these salts increased along the axis of rotation of the centrifuge cell. Pederson’s paper stated, “Die Zentrifugierungen haben gewo¨hnlich 7 bis 10 Stunden in Anspruch genommen.” Meselson, whose intuition

T HE B IG M ACHINE

195

told him that it “would take days for a little molecule like cesium chloride to come to equilibrium in a centrifuge,” was not certain what Stunden meant. After he had watched the surprising behavior of the Schlieren line, he looked up the word in a German dictionary and confirmed that it meant “hours.” Still unconvinced, he went to Delbru¨ck and asked him if there was any other definition of the word. When Delbru¨ck told him that Stunden “meant hours and that there was no way of viewing it as anything but hours,” Meselson later recalled, that “jolted me to believe that I was really seeing this density gradient, which I found difficult to accept.” 35 When he copied out the concentrations along the x axis that Pederson had reported, Meselson stated in plain English in his workbook that they had been obtained with 0.1 molar CsCl, centrifuged “at 55,000 rpm at 32.8°C for 6 1/2 hours.” 36 By the next day, Meselson must have at least partially accepted this conclusion. The first idea that occurred to him was that this phenomenon might help him confirm that the phage particles were really homogenous enough to enable 5-BU substituted phage to be separated from unsubstituted phage in the ultracentrifuge. On 23 October, he wrote out a prospectus similar to the one Stahl had written two days earlier, perhaps to keep Pauling informed about his newest research initiatives.37 After summarizing briefly the flotation and sedimentation of virus in the preparative centrifuge, Meselson added, “Part of this [carefully purified virus] was examined in CsCl in the analytical centrifuge and found to be quite homogeneous with respect to sedimentation velocity. It was felt that any natural heterogeneity in density was considerably less than the 3% difference expected between substituted and non-substituted phage. Within the next few days a virus preparation will be examined in a small density gradient in the analytical centrifuge. By proceeding to equilibrium, very small heterogeneities should be revealed if present.” 38

III The startling developments of the preceding day did not induce Meselson to recast the priorities of his joint venture with Stahl when he recapitulated them for Pauling. The “small density gradient” referred to in the above passage appeared helpful for a preliminary examination of the degree of homogeneity of the phage. Their goals, as Meselson now expressed them, still gave equal weight to replication and mutagenesis,

196

T HE B IG M ACHINE

built on the common theme of the effects of the replacement of thymine by 5-bromouracil in bacteria or bacteriophage. There were, however, subtle differences in emphasis from Stahl’s corresponding memo, reflecting the respective outlooks of the two partners in the venture. Meselson divided the “several sorts of experiment” that the biological effects of 5-BU substitution suggested into three categories. The second he disposed of in a single sentence: “II. The 5BU-produced mutants could be mapped to determine if they follow classical segregation patterns.” Under the third category, “The factors determining frequency of mutation and inactivation,” he placed first the subcategory “Mutability.” Here he outlined, very briefly, the theory he had developed during August that either the greater tendency of 5-BU to ionize, or “a tautomeric equilibrium toward the rare hydroxyl form,” could be responsible for its mutagenic action. The rest of his description of the experiments they might perform on 5-BU mutants, as well as of the mutant stocks that Stahl had already grown, Meselson extracted directly from Stahl’s memo.39 Only in his summary of experiments planned to separate substituted from unsubstituted phage or DNA did he incorporate an idea that went beyond what he or Stahl had previously mentioned on paper: DNA and/or virus might be separated into variously substituted fractions by ultracentrifugal and electrophoretic methods. Total replacement by 5BU results in a 6% density increase for DNA and a 3% increase for virus. [Afterward he made this statement more cautious by inserting “should” between “5BU” and “results.”] . . . If separations are possible, detailed studies could be made of the molecular integrity of DNA through successive cycles of replication. By working with virus and recovering separated fractions with unimpaired infectivity, one might examine the genetic function of any piece of DNA found to maintain its integrity. If a large piece is found, it could be examined for the distribution of heavy atoms in it. Perhaps it could be decided whether the heavy atoms are confined to one chain of the DNA double helix.40 The statement that “if separations are possible, detailed studies could be made of the molecular integrity of DNA through successive cycles of replication” was the first explicit declaration Meselson had put in writing of the goal about which he had begun to dream more than two years earlier. There is no indication here, or in the incipient stage of the experimental program he had so far carried out, that he

T HE B IG M ACHINE

197

yet had devised a specific strategy for these detailed studies. His first experience with the analytical centrifuge did not even prompt him to follow Stahl’s lead in describing centrifugation as more promising than electrophoresis. Moreover, the plan was interwoven with so many other threads, from his theory of the mutagenic properties of 5-BU to his and Stahl’s latest scheme to determine base sequences in DNA, that a reader of his memo would scarcely guess that Meselson’s towering motivation was to solve, by means of density differences, the much-debated DNA replication problem. His reticence may reflect the fact that the first reaction of Meselson and Stahl to the surprising formation of a CsCl density gradient in the ultracentrifuge was to worry that this phenomenon might upset their plan to separate substituted from unsubstituted DNA or phage by centrifugation. If the density spread were greater than the difference between the densities of the substituted and unsubstituted material, that would prevent the former from sinking to the bottom and the latter from rising to the top, thus defeating their method. They realized immediately that the materials probably would then collect separately in layers somewhere between the light and the heavy end of the cell. There was, however, no guarantee that the separation that could be achieved between these bands would be greater than the width of the bands. If, through diffusion or through the inhomogeneity of the phage, the bands became broader than their centers were apart, then the “game would be over,” as Meselson later put it. They proceeded hopefully but cautiously, keeping open the possibility that they might, in the end, have to turn to the electrophoresis option.41 Meselson began the second experiment in the analytical centrifuge three days or less after finishing the first one. Because the first page of the log has been lost, there is no record of the starting date and time, the material inserted, or other initial conditions. The fact that the run lasted more than twenty-four hours suggests, however, that it must have been the first attempt to proceed to equilibrium with a phage preparation in a cesium chloride density gradient. As in the previous run, Meselson ran the centrifuge at a speed of more than 50,000 RPM. He switched the Schlieren and ultraviolet optical systems on and off several times during the long course of the run. The absorption photographs showed a sharp division between a dark and a light side, with a few faint bands near one end of the latter. The bands may have been breakdown products of the phage, but the main conclusion was simply that the cell had leaked—perhaps because the high speed had forced

198

T HE B IG M ACHINE

some of the very dense solution out, or perhaps because Meselson had not yet learned how to assemble the centrifuge cells properly.42 In order to prepare a CsCl solution in which the phage would come to equilibrium within the density gradient, it would be helpful to know the exact density of the phage. To ascertain this figure, Meselson performed another sedimentation experiment. On 26 October, he began the run with phage T4 in a CsCl solution of density (ρ ⫽ 1.444), sufficiently below the calculated density of the phage (1.5) to sink the latter. The experiment was apparently aborted after thirty minutes. When he repeated the experiment he reduced the concentration of the CsCl still further, to 1.345. For the first fifty-four minutes he ran the centrifuge at the low speed of 8,225 RPM, before raising it to 52,640, but the cell had already begun to leak.43 Despite the failure of these efforts, Meselson was ready to try another equilibrium experiment. He prepared a suspension of phage T4 with a titer of 2.2 ⫻ 1011 /ml and made up a CsCl solution to a density of 1.516 at 22°C, that is, to approximately the same density that he had earlier calculated for the phage. He hoped that this was close enough to the actual density so that the solution would form a gradient whose density at the top of the cell would be somewhat less than that of the phage and at the bottom somewhat more, and that the phage would become concentrated within the cell at the level at which the densities were equal.44 To reduce the risk of leakage, which Meselson now suspected resulted from high speeds rather than improper assembly of the cells, he planned to run the centrifuge at 25,980 RPM, even though he knew that would greatly increase the time necessary to arrive at equilibrium. Shortly after the run began, he sketched on a blank space on the log page, perhaps while explaining to someone what he was doing, a curve representing the density distribution of the CsCl solution along the axis of the centrifuge cell, and he marked the point at which the density would match that of the phage (figure 6.6).

Fig. 6.6. Density distribution curve of CsCl solution along axis of centrifuge cell, sketched on centrifuge log, run no. 653, p. 1

T HE B IG M ACHINE

199

The run lasted nearly forty hours. Meselson did not stay at his post throughout that time. Once he assured himself that everything was functioning well, he left, returning from time to time to check on progress. At one point he noted “Speed and Temp control good.” For most of the run he operated only the ultraviolet optical system. He must have developed the films as soon they were taken, to see whether the contents of the cell were approaching a steady state. During the last six hours he operated the Schlieren system as well and took some Schlieren photographs.45 If he developed immediately each of the films taken successively during the course of this run, Meselson may have been hopeful after the first four hours that a band was beginning to form at the “light” end of the cell. Afterward, however, it seemed to dissipate rather than to consolidate, and by the end the cell was divided about equally between a dark and a light half. Meselson and Stahl could have concluded little more than that the phage seemed to be falling apart during the run, releasing fragments of varied density. They determined, however, to pursue experiments of a similar type, hoping to find conditions under which they could prevent or reduce the disintegration of the organism sufficiently to produce a discrete band.46 Before proceeding with this plan, Meselson tried to estimate the homogeneity of the phage by the sharpness of its sedimentation boundary in an ordinary sedimentation run in water. To prevent the phage from breaking down, he added a trace of MgSO4. It is not clear whether the analysis yielded any useful information.47 On 4 November, Meselson began another equilibrium run. He altered somewhat the conditions of the previous one. The phage suspension was more dilute—0.6 ⫻ 1011 /ml instead of 2.2 ⫻ 1011 —and the solution density was 1.4900 instead of 1.56. He ran the centrifuge at the same low RPM (25,980), but something nevertheless went awry. He terminated the run after only five hours, after noting “equilibrium reached.” The films taken during this period showed no sign that anything was forming in the cell.48 Meselson immediately began another run that went on for more than two full days. In the second experiment he used a phage suspension intermediate in concentration between the previous two (1.2 ⫻ 1011 /ml) and a CsCl solution slightly less dense than before (ρ ⫽ 1.4775). He operated both optical systems and again spun the rotor at 25,980 RPM. He took ultraviolet photos throughout the run, Schlieren photos twice during the last half. When he developed the films for this experiment, Meselson saw

200

T HE B IG M ACHINE

Fig. 6.7. Film from ultracentrifuge run 657, exposure 8. As in all subsequent films, density increases from right to left.

that they had finally attained a significant result. Much of the lighter side of the cell was filled by a broad but easily discernible band. A narrower band was located near the heavy end of the cell (figure 6.7). The identification of these bands was probably uncertain to Meselson and Stahl, because they did not yet know how to determine the extent of the density gradient across the cell. They may have thought, however, that the diffused band at the light end of the cell corresponded to the phage and that the sharp band at the heavy end might represent DNA released by decomposing phage. In an effort to identify more clearly the band they suspected to be DNA, Stahl used a Spinco Analytrol photometer that Vinograd had in his laboratory to make a tracing of the absorption pattern. When moved across a photographic film, the photometer automatically transformed differences in the intensity of light transmitted through successive segments of the film into vertical distances on a graph whose horizontal axis recorded the position along the film. When applied to films taken in the analytical ultracentrifuge, the result was a “densitometer tracing” that gave the concentration of the absorbing material as a function of its position in the centrifuge cell.49 Stahl produced densitometer tracings from one photo taken in the middle of the run and one from the last set taken. They were nearly alike (figure 6.8). The two experiments begun on the analytical ultracentrifuge on 4 November were the first ones for which Meselson signed the log book himself. He had passed his apprenticeship, and Vinograd now trusted him to operate the machine on his own. Meselson quickly came to enjoy the routines associated with the process. Performing the systematic sequence of procedures was like following a well-ordered recipe. The machine itself had a pleasant tactile feeling. Its steady hums and

T HE B IG M ACHINE

201

Fig. 6.8. Densitometer tracing of film from centrifuge run 657–8 (reduced from original size). As in subsequent tracings, distance from the center of rotation is increasing from right to left; optical density is increasing downward. Peaks on either end represent opaque portions of the centrifuge cell bounding the quartz window.

clicks were enjoyable. As he gained experience, Meselson cultivated a few operational idiosyncrasies. He always started the rotor on low speed, for example, before the chamber was evacuated, because he worried that the contents of the cell might boil at low pressure. The spinning rotor, he believed, sealed the contents more tightly and precluded such a mishap. He never tested his assumption but stuck to his precautionary habit. He always used the optional electric brake to decelerate the rotor more rapidly at the end of a run, because he saw no point in waiting the full seven minutes required for the machine to come to a stop without the brake.50 The most immediate practical problem raised by this early stage in the implementation of the research plan was that the long equilibrium runs consumed far more centrifuge time than Meselson had envisioned when he first asked for time on Vinograd’s machine. The only operational analytical centrifuge at Caltech was in considerable demand. There was, in the laboratory room, a sign-up list on which those

202

T HE B IG M ACHINE

who wished to use the centrifuge had to reserve time in advance, sometimes leading to a negotiation to swap time with someone else.51 Vinograd expected his own research program and those of his immediate associates to center around work with the analytical ultracentrifuge.52 As the Meselson-Stahl research program began to require multiple experiments running the big Model E continuously for as long as twenty-four or even forty-eight hours, that demand came increasingly into competition with the needs of other users.53 After these first marathon experiments, Meselson had to wait three weeks for his next turn on the machine. The Norman W. Church Laboratory for Chemical Biology, financed by a matching grant from the Rockefeller Foundation and a bequest from the will of the late Mr. Church, was intended to provide adequate facilities for the virologists and biophysicists of the Biology Division. The new building was located just south of the similar building housing the older Kerckoff Laboratory, with the expectation that a third unit would eventually be built to join them in an H-shaped set of three interconnected laboratories. By the time the dedication ceremonies for the Church Laboratory took place in November, those to whom the space was allotted had already moved in. Among those who attended the dedication was John D. Smith, from the Berkeley Virus Laboratory, one of the authors of the paper on the incorporation of halogens into bacterial and phage DNA that Meselson had studied in September. Meselson and Stahl took the opportunity to consult with Smith about their own progress on the incorporation of 5BU into phage and showed him the tracings from their latest centrifuge experiment on T4 phage.54 Perhaps cognizant of difficulties Howard Schachman and his associates had encountered at Berkeley with the absorption of ultraviolet light by the cesium chloride solutions in their ultracentrifuge experiments, Meselson and Stahl decided early in their experiments to purify the commercial salt before using it. Meselson devised a procedure in which he precipitated the salt with perchloric acid, forming cesium perchlorate. He then washed the powdery precipitate, removing any soluble impurities. Placing the precipitate in a platinum dish, he heated it in a furnace until the perchlorate decomposed, leaving very pure cesium chloride. The operation was rather harrowing. To melt the perchlorate he had to heat the furnace until it was cherry red, all the while thinking that organic impurities might cause the perchlorate to explode.55 No disasters occurred. The best way to find out whether the band at the heavy end of the

T HE B IG M ACHINE

203

density spectrum in their experiments with T4 phage was really DNA, Meselson and Stahl decided, was to centrifuge some actual DNA and see if it formed a band in the same place.56 They were able to obtain from Oleg Jardetsky, a postdoctoral fellow, a preparation of calf thymus DNA. Jardetsky supplied the DNA in solid form, from which Meselson and Stahl made up a 1.4 mg/ml solution in water. For the centrifuge run they added .01 ml of this solution to 0.8 ml of a stock CsCl solution that, they noted, “had not been purified.” Phosphate buffer was added to set the pH at 7.6. The density of the CsCl solution, 1.666, was significantly higher than the solutions they had used for T4 phage, because they wanted the average density of the solution to approximate the known density of DNA.57 Meselson began the long equilibrium run at 3:40 P.M. on Saturday, 24 November. He had probably bargained for extended time by agreeing to do the experiment on the weekend, when demand for the centrifuge was lower. Cautiously increasing the speed above the 25,980 RPM he had recently maintained to avoid leakage, he accelerated the rotor to 31,410 RPM. He switched on the ultraviolet optics, checked that the cell was not leaking, and took photographs at thirty-minute intervals. He left just before midnight. When he checked in at about 8 A.M. on Sunday and developed the last film taken, banding was discernible. The machine ran all through Sunday. The camera was set to take photos automatically at sixty-four-minute intervals, until the timer was turned off at 7:50 P.M.. The next photograph was taken manually at about 9 P.M.. The photos indicated that the bands were not yet approaching equilibrium, so the machine spun on through the night and into Monday morning. At 9:30 A.M. Meselson activated the Schlieren system. Finally, at 3 P.M., he ended another run that had lasted almost 48 hours.58 The calf thymus DNA “banded at almost the same place” that the band Meselson and Stahl had taken to be DNA released from T4 phage had banded in the earlier experiment, “though not as sharply.” The conclusion that the two bands were in almost the same place was only a rough estimate, because Meselson and Stahl had no fixed reference point with which to compare the density gradients formed by separately prepared CsCl solutions in different centrifuge runs. Small differences in the average density of a solution would cause significant displacements of the positions within the cell at which material of a given buoyant density would concentrate. Because they had centrifuged the T4 phage in a CsCl solution whose density was 1.4775, and

204

T HE B IG M ACHINE

the calf thymus DNA in a solution of 1.666, the bands they were comparing could not literally have been in the same place. They were confident enough in their estimates of the effect of these differences, however, to believe that the new experiment confirmed their identification of a DNA band in the phage experiment.59 That significant chemical differences existed between DNA molecules of different origins was well known from the work of Erwin Chargaff and others.60 Even without thinking through whether such differences might affect their own venture, Meselson and Stahl could immediately appreciate that their identification of the DNA band in the T4 phage preparation would be strengthened if they compared it with DNA from the same source. Accordingly they subjected some of their T4 stock to “osmotic shock”—that is, they rapidly diluted the strong salt solution in which they had suspended the phage, disrupting its protein shell and releasing its DNA. They placed what phage biologists called the “shockate” resulting from this procedure directly into the CsCl solution. At some point they decided that it was important, in preparing DNA for centrifugation, to avoid traditional extraction methods, in which some material is thrown out, because the discarded material might include some fraction of the DNA. The protein remaining in the solution would be separated from the DNA during the run, because of its different density. Trying various reagents known to denature proteins, Meselson found that guanidine hydrochloride caused the phage to “open up” and deposit its contents in the salt solution.61 Meselson probably obtained centrifuge time to make the run during the first week in December. Using a CsCl solution adjusted to approximately the same density (ρ⫽1.66) as in the previous experiment with calf thymus DNA, Meselson began the run at 6:45 [P.M.?], reached 29,500 RPM seven minutes later, and was able to bring the run to a close in just under thirty hours.62 The band was similar enough to the one obtained from calf thymus DNA to support the identification they had made of the DNA band derived from whole phage decomposing in the cell. This was not a remarkable conclusion. More significant than the finding itself was the potential that was beginning to emerge for the method they were pursuing. The density gradient that formed spontaneously within a few hours when Meselson spun cesium chloride in the ultracentrifuge, at speeds moderate enough to avoid cell leakage, caused DNA to form a fairly well-defined band in nearly the same place in several successive experiments. That outcome meant

Fig. 6.9. Matthew Meselson operating the Model E Analytical Ultracentrifuge. Photo courtesy of the Archives, California Institute of Technology.

206

T HE B IG M ACHINE

that he and Stahl no longer had to plan to separate substituted from unsubstituted DNA by causing one to float to the top and the other to sediment to the bottom. The density gradient looked like it might be just the right degree to cause the substituted and unsubstituted molecules to form bands at different positions along the axis of the centrifuge cell. That was auspicious progress for their first six weeks of work on the problem. In a letter to James Watson on 14 December, Meselson wrote hopefully about the prospects for the method he and Stahl were developing: “Separation experiments. We have looked to see if the substituted phage can physically be separated from the non-substituted ones by virtue of the 3 per cent maximum density difference. So far I have had success in banding whole phage (and DNA) in density gradients of cesium chloride at equilibrium in the ultracentrifuge. If the resolution can be improved to anywhere near the theoretical limit imposed by diffusion, we will be able to separate variously substituted phage. Various transfer experiments would be possible.” 63 His reference to the theoretical limit imposed by diffusion implies that his initial observations of banding had already induced Meselson to begin to examine more deeply the factors that underlay their formation and their character. The centrifugal forces that drove the material to concentrate where its effective density equaled that of the solution would be counteracted by the tendency of solute molecules to diffuse from areas of high concentration to those of lower concentration. It remained to be seen whether the bands could be made sharp enough and narrow enough to achieve the desired separations. With his usual blend of venturesomeness and caution, Meselson was not yet ready to place all his bets on density gradient centrifugation. In the final sentence of his summary, he held open the same alternative that had attracted him before the centrifuge runs had begun: “Electrophoresis might be tried as a separating method if the centrifuge fails.” 64

IV The progress that Meselson had made in these separation experiments did not, as 1956 drew to a close, overshadow the other threads woven into the research program for which he and Stahl had teamed up. The collaborators seemed, in fact, more excited about the prospects opening up through the progress that Stahl was making on the mutagenesis front of their joint campaign. By mid-December, Stahl had made two

T HE B IG M ACHINE

207

significant forward thrusts: one concerning the genetic character of the mutants he had produced in September and early October, the other concerning the effects of X-ray and ultraviolet irradiation on 5-BU substituted phage. Having produced mutant 5-BU stocks from two strains of T4 bacteriophage (T4B and T4Br), Stahl isolated two of the mutants, crossed them, and determined their recombination frequencies with respect to one another, as well as with a “well-known marker” in the rII region whose fine structure had been mapped in normal T4 by Benzer. These “induced mutants mapped OK, with no funny business,” Meselson and Stahl reported a little later to Smith and Schachman. Stahl “found one [of the mutants] to be linked to the rII locus and the other not. They are not linked to each other.” Encouraged by this evidence that 5-BU mutants did follow classical segregation patterns, Stahl moved on to the next phase of his research plan: performing cross-reactivation experiments, that is, infecting bacteria with a mixture of inactivated 5-BU phage and active phage differing in their genetic markers. No details of the experimental design have survived, but if he carried it out as planned in October, he would have cross-reacted an inactivated substituted rII⫹ stock (wild type) by a nonsubstituted rII (mutant) stock. In such experiments genetic material from the “dead” virus can make a genetic contribution to progeny phage because it is “rescued” by recombination events that attach undamaged parts of its genetic material to the genetic material of the active phage. In the results obtained by mid-December, Stahl found that the “reactivable fraction” of the “dead members of a 5BU stock” was “about 40 percent.” 65 The two research partners were sufficiently emboldened by these results to embark in December on the further mutagenesis experiments they hoped would lead them eventually to the sequence analysis that Stahl and Meselson had each outlined in their prospectuses in October. Only the class of mutants that were so “by virtue of the fact that a certain nucleotide pair is cytosine-guanine,” they had reasoned then (see above, p. 185), “should produce phage unstable when passed through a subsequent growth cycle in the absence of 5BU.” They were now “looking to see if substituted phage throw off additional mutants when grown in cells free of 5BU.” The experiment with which they tested this approach was “to count on K lambda, reversions to wild type in a substituted 5BU stock and also to count reversions after preadsorption to K (sensitive) followed by plating effective centers on K lambda. Unsubstituted phage are the control and several rII markers

208

T HE B IG M ACHINE

are being looked at.” 66 Their strategy was, like Stahl’s experiments on closely linked rII markers the previous fall, based on the same very sensitive method that Seymour Benzer used to detect any wild-type phage that might have arisen by reverse mutations by plating on a strain of E. coli on which the rII mutants are unable to form plaques.67 This part of the project was particularly attractive to Meselson, because the reasoning on which it was based was inferred from the theory of base mispairing that he had worked out during the summer. He did not leave all the work to Stahl but became directly involved in the phage genetics experimentation. In his letter to Watson on 14 December he wrote, “I’m just in the middle of this and should know the answer soon.” 68 There were several reasons to study the effects of radiation on the 5-BU substituted phage. Most general, irradiation was long established as a standard means to investigate the properties of phage. More immediately pertinent to the current situation was that Rose Litman informed Stahl that the substituted phage showed greater sensitivity to ultraviolet radiation than did normal phage. Whether because of this knowledge or independently of it, Meselson predicted on theoretical grounds that the substituted phage should also be more sensitive to X-rays, because of the bromine atom contained in 5-BU. As he explained his view to Watson shortly afterwards, “Considering the great photoelectric cross section of bromine in its K-edge and the very high yield of extremely short range photoelectrons as contrasted with the long range photoelectrons ejected from light elements, the increased number of ionizations in a substituted phage might” cause more phage deaths when subjected to the same dose of radiation.69 Measurements of the effects of ionizing radiations, such as X-rays, on bacteriophage were performed and evaluated in accordance with the “target theory.” This theory, devised before World War II, assumed that “the effect studied is due to the production of ionization by the radiation in, or in the immediate vicinity of, some particular molecule or structure.” Experimentally, one irradiated the phage with X-rays of a given wavelength and intensity for a period of time during which one removed samples at regular intervals (for example, every fifty seconds) and determined what percentage of the phage had survived up to that point by plating them on bacterial lawns. Experience had shown that the lethal effect of the radiation was proportional to the cumulative “dose,” that is, to the product of the intensity and the exposure time. The results were plotted in the form of a “survival curve.”

T HE B IG M ACHINE

209

For the irradiation of phage, these survival curves often turned out to be exponential, an outcome that was interpreted, in terms of the target theory, as demonstrating that a single ionization, or “hit,” was sufficient to inactivate the organism. When the logarithm of the surviving fraction of the organism was plotted against the dose, a straight line (displaying the logarithmic form of the exponential function) indicated that the lethal action of the radiation was caused by a single hit, and the slope of the line was a measure of the sensitivity of the organism. The “mean lethal dose,” defined as corresponding to an average of one hit per target (also called the 37 percent dose, allowing 36.8 percent survival, because e⫺1 ⫽ 0.368), could be used, in accord with other theoretical assumptions, to define the target size. In general, a larger target size was correlated with increased sensitivity, that is, with lower rates of survival for given doses.70 Stahl carried out the required experiments and found, as Meselson had predicted, that “phage with incorporated 5-bromouracil was . . . more sensitive to X-rays than control phage.” Moreover, “the survival curves are still exponential,” and “the increased sensitivity is maintained when the irradiation is performed on the infected cells, after penetration of the phage DNA, rather than on the free phage.” 71 This was the bare statement of the results as expressed in Stahl’s grant report. Their implicit meaning Meselson conveyed in his letter to Watson: “We found that 5BU phage show increased target size. . . . The x-ray death looks nicely one hit.” That the increased sensitivity was retained after the phage were injected into bacteria indicated that the sensitivity resided in the DNA. This confirmation may have tempted them to become further involved in experiments to characterize the action, perhaps to test Meselson’s theory about the effect of bromine. They resisted the opportunity to weave still another thread into their programmatic tapestry, however, deciding instead to exploit the result for more immediate practical purposes. Having found that the X-ray sensitivity was higher in “stocks prepared by methods which Rose [Litman] says give good chemically measured substitution than in stocks from ‘bad’ methods and in controls,” they hoped that the sensitivity would turn out to be proportional to the degree of substitution of bromouracil for thymine in the DNA. If so, they might use differences of sensitivity as a measure of the extent of 5bromouracil incorporation and avoid the more cumbersome method of chemical analysis. The ability to control this factor would be critical to the success of their plans for separating substituted from unsubstituted

210

T HE B IG M ACHINE

DNA, because, whatever method they finally used, they could obtain clean results only if the substituted material was homogeneous.72 The letter that Meselson wrote to Watson on 14 December was a general summary of his research activities, accompanying a request for a letter in support of his fellowship application. If we were to judge the intensity of Meselson’s interest in the various strands of his investigative agenda by the attention he gave to each of them here, then the mutagenesis questions seemed at this point to outweigh by far the separation experiments aimed at the replication problem. About the latter he wrote only the single relatively short paragraph (see above, p. 206) placed at the very end of his four-page description of his research objectives and progress. By comparison, he wrote a paragraph about twice as long, at the beginning of his letter, on the experiments he had begun for Linus Pauling on the crystallization of mixed amino acids—an effort that he was, in fact, on the verge of abandoning. The rest of his text, more than three pages, he devoted to the mutagenesis work. The reason for this apparent inversion of his earlier priorities can only be inferred from the circumstances of the moment. Although the first round of ultracentrifuge experiments had been promising, there were still enough uncertainties that Meselson remained guarded about the ultimate prospects. Meanwhile, the progress that Stahl had made on the genetics of 5-BU substituted phage was impressive enough to give a strong lift to their hopes for the longer-term goals of this side of their venture. For Meselson’s personal investment in the collaboration, the most exciting consequence of this development was that it seemed to be opening up investigative vistas that led toward tests of his theoretical examination of the mutagenic properties of 5-BU substitution. It was these theoretical possibilities, he told Watson, that “made me very interested in doing biological experiments to see first if 5-HU [halouracil] mutagenesis is due to local action of the unnatural base and then if base pair switching occurs.” Meselson included in the letter a lengthy summary of his analysis of the effects of halogen substitution on the propensity of the base for tautomerization and ionization. It was the only detailed discussion of the subject that he put into writing, and is reproduced here nearly in full. By referring back to Chapter 4, the reader can easily identify the sources of Meselson’s interpretation in the reading that he had pursued in the Crellin library during the previous August and September. An effect of the halogen substituent [of the halouracils] is to remove electrons from the pyrimidine ring. The effect increases in

T HE B IG M ACHINE

211

the order para, meta, ortho. In the case of thymine as compared with a 5-halo uracil, this makes more labile the proton on the number one nitrogen and will also loosen protons on either oxygen when the molecule is in one of the rarer tautomeric forms. How the tautomeric ratios themselves change can be predicted from the relative effects on proton dissociation constants at the nitrogen and at the two oxygens. Because tautomerization is thermodynamically a process equivalent to ionization from one site and association onto another, tautomeric ratios can be calculated if pKa’s are known for the separate groups. Such separate pKa’s are not directly determinable by experiment and in general are not predictable from the pKa’s of monofunctional molecules. However the change of pKa of a particular group when a halogen substituent replaces methyl can be approximated by the analogous change occurring when a monofunctional molecule is similarly substituted. If R is a tautomeric ratio for thymine and R′ is the corresponding ratio for a 5-halo uracil, then logR′/R ⫽ [pKa 1(5-HU) ⫺ pKa 1(thy)] ⫺ [pKa 2(5-HU) ⫺ pKa 2(thy)] where the proton shift is from site two to site one. The quantities in brackets are the pK changes of a given functional group brought about by substitution of halogen for methyl and may be estimated from data for benzene and pyridine compounds. In this way it is predicted that all the halogens (F, Cl, Br, I) should increase tautomerization onto both oxygen-6 and oxygen-2 with the larger increase for the latter. Quantitative statements come much harder but perhaps there is a ten-fold increase for O-6 and a hundred-fold increase for O-2. . . . As you have pointed out, proton shift to O-6 could lead to base pair switching. This would hold for O-2 as well if a hydrogen there could be directed properly away from the guanine amino group without interfering with the sugar. Meselson next mentioned that he had been looking for solvents suitable for preparing solutions of thymine and 5-halouracils for proton magnetic resonance study, from which he surmised that it might be possible to “look at the tautomeric ratios and conversion rates.” Then he turned to the other possible mutagenic mechanism, ionization, which had been his own idea: Another effect of the electron withdrawal due to halo-substitution is simply to ionize away the proton altogether. This effect is directly measurable by simple titration. The result is that 5-BU is ionized about 100 times more often than thymine. This result was predicted by the substituent approximations already mentioned

212

T HE B IG M ACHINE

and encourages me about the calculations for the tautomers. Ionized thymine or 5-HU can pair with guanine so this may be another route for base pair switching. A difference between the routes lies in the possible unstable base pairs formed in the process of reaching a stable switched pair. Ionization can give only the illegal pair G-T while both G-T and A-C could result from tautomerization.73 This passage indicates that, sometime during the preceding two months, Meselson had carried out the titrations that he had told Levinthal in September he planned to do. No further details are recorded, but it is evident that the results supported his theoretical prediction. The boldness of Meselson’s research plan may not be immediately evident from the simplicity of this straightforward summary of his reasoning. He had arrived at a testable theory of mutagenesis within a few weeks of the commencement of a reading program in which he had started out merely to teach himself the rudiments of the structure of DNA that he would be expected to know in the Chemistry Department when he began work on the replication problem. He had extracted the relatively simple but strategic information incorporated into his analysis from an intensive self-guided survey of the current state of knowledge about the properties of DNA and its constituent molecules. To judge what is relevant and cogent to the definition of a problem deeply embedded in a field one has just entered is a formidable task. It is a measure of his self-assurance that the young scientist, still months away from the completion of his Ph.D., was ready to submit his idea to the scrutiny of one of the authors of the structure of DNA, which served as the foundation for the very problem of mutagenesis to which he hoped to make a contribution. What was Watson’s reaction to a proposal that had the potential to clarify a mutagenesis mechanism that he and Crick had put forth prominently but with little supporting evidence? There is no record that Watson responded directly to Meselson’s account of this or the other aspects of the research program on which he and Stahl had embarked. That Watson was favorably impressed, however, is shown by a statement he made three years later when Meselson was under consideration for a faculty position at Harvard: “Meselson has been also concerned with the genetic problem of mutagenesis and was the first person to seriously question why [a] base analogue would be mutagenic. He realized that the presence of Br in 5 bromo-uracil lowers the pK at its ionizing group and thus a simple explanation of why its introduction into DNA should be mutagenic.” 74

T HE B IG M ACHINE

213

V For Frank Stahl everything seemed to be going better at the end of 1956. In three months he had escaped from a research impasse and begun a fruitful collaboration in which both he and his new associate had important, complementary parts to play. His young family was thriving. Their first son, Andrew, was healthy and lively, and a second child was expected. Meselson reported to Watson that “Andy is a delightfully warm-hearted and intelligent baby and soon there will be another child born to them. All goes well there.” 75 Meselson also took satisfaction in his own circumstances as the year neared its end. Stahl’s decision not to leave Caltech encouraged him to expect that Frank would be there for a number of years and that they would be able to carry out their long-term program together. Meselson did have one mishap. After a dinner with Ani in San Francisco in November, he missed a turn while approaching the Bay Bridge and nearly demolished his Thunderbird. Unhurt, he had to leave his car to be repaired, and to pay a small fortune for it, because he did not have insurance coverage. Alan Garen went to pick it up for him at the body shop where it was supposed to have been fixed, but when Alan learned from the mechanic that one of its wheels wobbled, he refused to drive it off. The Thunderbird was eventually restored to serviceable condition, but it was never the same after the accident.76 At the house on San Pasqual Street another graduate student had replaced Stahl. When Jan Drake had begun his graduate studies in experimental embryology, he met Howard Temin. Deciding to switch into a discipline that offered more modern analytical methods, Drake and Temin entered the laboratory of Renato Dulbecco, where both came under the guidance of a postdoctoral fellow, Harry Rubin. When Temin joined Drake and Meselson at the “boys’ house” he took on some of the cooking, and he too found there “an exciting and stimulating scientific atmosphere.” 77 In his letter to Watson, Meselson gave a vivid vignette of the social life that he and the other bachelors in the house shared: “From the windows of our great house on San Pasqual there often blaze forth festive lights. We eat and drink most unusually well for Pasadena and for graduate students and have a guest suite awaiting even the most unexpected visit from you. Only the worrisome woman problem remains but that at much attenuated levels.” 78

C HAPTER S EVEN

Working at High Speed

I At the dedication of the Church Laboratory in November 1956, Robert Sinsheimer, who had recently accepted a position in the Biology Division at Caltech, delivered a special lecture titled “First Steps Toward a Genetic Chemistry.” In it he warned that the evidence linking genes with the chemistry of DNA was “at present necessarily circumstantial.” Most of it served merely to “present DNA as a likely candidate for a genetic role.” There was no experimental evidence to support the “postulate that the information is carried as a nucleotide sequence—we simply lack any credible alternative hypothesis.” 1 He appeared equally skeptical about current ideas regarding the replication of DNA: “As a further bit of fancy, it has been suggested that when DNA is to be duplicated, the chains in some manner separate, perhaps only a segment at a time, and each chain serves as a template upon which to build up its complement, resulting in two chains like the original. Various modes of accomplishing this feat have been proposed which would result in various distributions of the original parental material among the daughters.” 2 Sinsheimer spoke as a member of the network surrounding Delbru¨ck but one who stood somewhat apart from the enthusiasms of the phage group. His intent was to remind his colleagues that their “hardwon recognition of the role of DNA” had brought them only to the beginning of a “new era in genetics and biochemistry.” 3 Meselson and Stahl were present for the Church Laboratory celebrations and may have heard Sinsheimer’s lecture, but his comments about the replication of DNA apparently did not lead to any discussions with them about their plans to test the various proposals for “accomplishing this feat.” They did obtain from him some of his large stock of wild-type

216

W ORKING

AT

H IGH S PEED

T4 phage for a series of ultracentrifuge runs that they began on 19 December.4

II The main purpose of the experiments Meselson and Stahl began just as the 1956 Christmas holidays began was to band T4 phage DNA repeatedly, seeking conditions that would improve the resolution and testing at the same time the general reliability of the procedure by seeing if the band came out in the same place each time.5 Perhaps they were also taking advantage of an opportunity the holidays offered for nearly continuous time on the ultracentrifuge. In preparation for these runs, on 19 December Meselson and Stahl further purified the wild T4 Sinsheimer stock that they had earlier partly purified. Taking the highly concentrated (2 ⫻ 1014 /ml) stock from the refrigerator, they added RNAse to degrade any RNA from the lysed bacteria. After refrigerating the stock again overnight, they allowed it to stand for an hour at room temperature, diluted it with suspension medium, and spun it in a preparative centrifuge for ten minutes at 5,000 RPM, then sixty minutes at 10,000 RPM, followed by two more high-speed runs, one low-speed run, and a final highspeed run. This cyclic procedure was intended to separate out any fragments remaining from the lysed bacteria. At low speeds the “junk” that sedimented easily would separate out, forming a pellet that they discarded, while leaving the phage in solution. The high-speed runs brought down the phage, leaving free protein, RNA, and degraded phage fragments in solution. They then threw that solution away and resuspended the phage for the next low-speed run. The supernatant fluid from the last low speed run they labeled “Purified Sinsheimer Wild-19 December” and put it into the refrigerator for future use. The final pellet from the last high-speed run contained pure phage with a DNA content of about 20mg, which they suspended first in .5N purified CsCl solution, then in saturated, purified CsCl solution for two hours at room temperature. To release the DNA by osmotic shock, they pipetted 1ml of the suspension into 19ml of water “with stirring.” The shockate was “fairly turbid and quite viscous and elastic.” For the first centrifuge run they placed 0.05ml of this shockate into 3 ml of a stock solution of CsCl purified by the procedure Meselson had recently adopted. The density of the solution, 1.69 at 23°C, was chosen to

W ORKING

AT

H IGH S PEED

217

match the density of DNA calculated from its known partial specific volume.6 By the time they had completed these preparations and Meselson had filled and secured the centrifuge cell, it was late in the evening. He switched on the big machine at one minute before midnight, worked the speed in stages up to 31,400 RPM, and stayed until 2 A.M.. Expecting that he would not be present for much of the next day, he left extended instructions for someone, perhaps Vinograd, who had agreed to check the machine for him. Judging from the length of time it had taken for bands to appear in the last similar run, he estimated that they should become discernible at about 4 P.M.. As he had predicted, the films developed in the early afternoon showed a light area that narrowed gradually in successive shots. By the time the run ended, after thirty-one hours, early on the morning of 21 December,7 the light region had coalesced into a well-defined, if still rather broad, single band. It was near the center of the cell, as would be expected for DNA spun in a solution of density near 1.7. Stahl made a densitometer tracing (figure 7.1).8 How would they have interpreted this result? The band appears

Fig. 7.1. Densitometer tracing of film from centrifuge run 719

218

W ORKING

AT

H IGH S PEED

rather broad and somewhat skewed, implying, perhaps, that the material was not entirely homogeneous. Perhaps the shoulder on the curve would have suggested that there were two species of DNA present. At this early stage in their trials, however, they were probably not yet certain how sharply their method could be expected to band pure, homogeneous DNA, so that such conclusions would have been tentative ones. On the day after Christmas, Meselson began another centrifuge run with Sinsheimer wild type T4, but this time using intact phage, with a CsCl solution reduced, accordingly, to a density of 1.48. In an attempt to stabilize the phage, which had tended to disintegrate in previous runs, he added 10 micrograms/ml of gelatin to the suspension. The run was “terminated after 23 hours,” and the result was similar to that of the first equilibrium run on whole phage done 6 1/2 weeks earlier: there was a broad band representing whole phage, and a sharp one at higher density, where they thought the DNA released from the phage should be found. The former was better defined and only about half as wide as that of the earlier experiment, suggesting that they had partially stabilized the phage but that it was still tending to come apart.9 During the course of this run Meselson described, to someone who happened by, the general idea for the transfer experiments he and Stahl hoped eventually to carry out. While doing so, he sketched, on the back of the ultracentrifuge log, diagrammatic representations of the main features of the plan (figure 7.2).10 From this sketch we can reconstruct what Meselson explained to the passerby. At the upper left he drew the outline of the centrifuge cell, oriented so that the centrifugal force (F) is directed downward. Deoxyribonucleic acid in which 5-BU was substituted for thymine ought to form a band below that of normal DNA, because of the 6 percent difference he had calculated for their respective densities. He drew a third band between these two, to indicate the position in which one would expect progeny DNA formed by one strand of substituted and one of unsubstituted DNA to band. To make the mechanism clearer to his listener, Meselson then represented (to the right) the four bases that form pairs in DNA by two sets of simple complementary shapes specifying the allowable base pairs. He indicated the substituted base by attaching “Br” to its diagrammatic outline. Just to the left of that diagram he illustrated the

W ORKING

AT

H IGH S PEED

219

Fig.7.2. Diagrammatic representation of DNA replication sketched on centrifuge log (slightly darkened by drawing over faint original lines)

random order in which such base pairs might be aligned along the two strands of a DNA helix. Below this, he diagrammed what would be the expected result if an organism grown in a medium such that it contained fully substituted DNA were then placed in a normal medium. The circles on the far left represented the reproduction of the organism, beside it the corresponding replication of its DNA (the latter displaced downward relative to the former, because he drew the DNA strand undergoing its first replication below his sketch of the fully substituted molecule). After the first division, all the DNA should be “H L,” that is, composed of one heavy and one light strand. After a

220

W ORKING

AT

H IGH S PEED

second replication, the bottom row of his sketch showed, there should be equal numbers of HL and of LL, the latter consisting entirely of normal, that is, “light” DNA. This sketch is particularly revealing, because it shows that Meselson had thought through the structure of the intended experiments more thoroughly than any of the verbal descriptions he had written up until then indicated. Moreover, it suggests that, well in advance of the event, he expected that the outcome would verify the prediction of the Watson-Crick model. There is, however, a puzzling feature of the visual representation. The succession of two simple duplications, paralleling the reproduction by binary fission of an organism depicted as a simple oval, strongly suggests the association of the process depicted with the replication of bacterial DNA. The reproduction of bacteriophage DNA was a more complex process, involving multiple replications and recombinations. The simplest characteristic diagram of a T series phage was not an oval, but a figure such as:

Nevertheless, there is no contemporary evidence that, at this stage in their venture, Meselson and Stahl had ever contemplated carrying out the transfer experiments on any organism other than T4 bacteriophage. The run with intact T4 ended just after midnight on 27 December. Two hours later, Meselson was already beginning another run (number 732) with the shockate, this time reduced to one-quarter of the concentration used in the previous one, and giving a calculated optical density also of about one-quarter that of the earlier one. Evidently they were searching for conditions that might improve the resolution of the DNA band expected to form. Meselson ran the centrifuge at the usual speed (31,400 RPM). By the next midnight the first signs that a single band was forming became visible on the developed film, and by early next morning it was evident that it would be clearly defined. The run ended after thirty-four hours.11 When Stahl made a tracing of the film with the densitometer, it appeared that they had achieved what they sought: a nearly symmetrical band with a single, clearly defined peak (figure 7.3).12 This result was encouraging, but the tracing also showed that the optical density below the band was greater than that above it, suggesting that the run may not have reached equilibrium. Meselson and Stahl decided, therefore, to rerun the same cell at a higher RPM. Starting the run (735) on 29 December at about 3 P.M., Meselson was able

W ORKING

AT

H IGH S PEED

221

Fig. 7.3. Central portion of densitometer tracing of film from centrifuge run 732

to accelerate the centrifuge to 44,770 RPM without incurring leaks, and he noted that the speed was controlled to “⫹/-0,1%!” Other troubles did intrude, however. At 2 A.M. he realized that all of the photographs taken until then had been “lost because of aperture stop.” Undaunted, he readjusted the aperture and reset the timer to take exposures at thirty-two-minute intervals. At 4:30 in the morning he found the refrigeration off and the control “locked on heat!” He turned the refrigeration on and jiggled the relay to free it. Despite these disturbances, Meselson knew, as soon as he developed the first series of unspoiled films, that the run was going well. A clearly defined band was forming after about twelve hours of operation, an indication that 44,770 RPM might turn out to be the best compromise between reducing the number of hours required to reach equilibrium and running the risk of leakage in the cell.13 By the time the run ended in the early afternoon, after only twentyone hours, the single band was sharper and narrower than anything they had previously attained. When Stahl made a densitometer tracing, he set the film carriage to make a slow scan, because he had found that when tracing a sharp band with the instrument set at the normal speed, the mechanical inertia in the linkage caused the pen to lag and

222

W ORKING

AT

H IGH S PEED

Fig. 7.4. Central portion of densitometer tracing of film from centrifuge run 735. The curve from run 732 is superimposed by hand.

distort the curve. He now obtained a highly symmetrical curve. To compare it with the band obtained at lower RPM, he superimposed the densitometer curve made from run 732 and drew it by hand over the new curve (figure 7.4).14 Increasingly confident now of the resolving power of their method, Meselson began another check on its reliability by running a mixture of the contents of the cell just used and the calf thymus DNA obtained from Oleg Jardetsky. He turned on the centrifuge at 1:06 A.M. on 30 December. By the time he developed the first film series at noon, he could already tell that a band similar to that from the previous run was forming. He allowed the run to continue through the rest of the day and night. The final pictures showed a band almost exactly superimposable on the previous run with T4 DNA alone, together with a less dense extension beyond each edge which suggested that the calf thymus DNA did not band quite as tightly as the T4 DNA did.15 Before the conclusion of these runs, Meselson and Stahl were already getting ready to try for the next major goal of their density gradient method: the separation of 5-BU substituted from normal phage DNA. On 28 December, they began preparing liter-scale 5-BU stocks.

W ORKING

AT

H IGH S PEED

223

After growing bacteria (the strain is not specified in the protocol) overnight in a standard medium to a concentration of 5 ⫻ 108 /ml and icing it for about twelve hours, they concentrated the bacteria by centrifugation, then placed a portion in each of two containers with 0.9 l of a special medium containing 25 mg of xanthine, 100mg of 5-BU, and 5 mg of uracil per liter of normal growth medium. The first portion they allowed to grow to 3 ⫻ 108 /ml and labeled it “A,” the second they grew to 3 ⫻ 107 /ml and labeled “B.” By this time it was midnight. Their objective was to infect the bacteria in the separate batches with phage at different multiplicities (numbers of phage particles absorbed per bacterium), to ascertain which proportion would result in greater incorporation of 5-BU into the phage progeny. The next day Meselson added to the two cultures quantities of T4 r240 calculated to give a multiplicity of phage per bacterium of 7.5 for A and 0.02 for B. After aerating the mixtures and waiting for them to clear, he assayed the lysates by the usual method of plating. From the number of plaques he calculated the numbers of phage formed in A and B, and noted “no sizable number of mutants were observed.” 16 When he reexamined this record in 1992, Stahl commented that it “probably wasn’t a successful protocol, because the first indication that you’re getting good incorporation [of the 5-BU into the phage DNA] is lots of mutants.” 17 At the time, however, they apparently did not draw that logical conclusion, because they went on to produce a purified “5BU Lysate ‘B’ (low mult[iplicity]),” which they regarded as more promising than the high-multiplicity “A” lysate. Meselson carried out the purification procedures, which were similar to those followed with Sinsheimer’s wild T4 stock. To produce the maximum degree of osmotic shock, he devised a “fancy” procedure to ensure that the final phage pellet would be suspended in totally saturated CsCl. After placing the pellet in a saturated solution, together with a little Mg, he placed them in a desiccator along with a dish of saturated CsCl and lowered the pressure. The equilibrium of water vapor pressures then ensured that the concentration of water in the two solutions would also be equal. This meticulous attention to details beyond what was essential to success was characteristic of Meselson’s emerging experimental style. “Some drying unfortunately took place,” he noted, but that did not seriously disturb the outcome. To produce the osmotic shock, he “dumped in” 2 ml of distilled water. The shockate was “turbid and contained clumps of unshocked phage.” Meselson “guess[ed] that 80% of the phage shocked, judging from the turbidity.” 18

224

W ORKING

AT

H IGH S PEED

In a similar way, Meselson prepared a shockate of purified T4 r240 “parent stock” that Stahl had already partly purified. The shockate was much more turbid and contained more clumps than the shockate from 5-BU stock. Now Meselson carried out a comparative assay of the substituted and unsubstituted stocks, scoring the respective mutants, as determined from their plaques. The mutant frequency for 5-BU stock was 3 plaques out of 200, compared to 0 out of 200 for the parent stock. Apparently he counted plaques only for a single plate for each type. The result was, therefore, only a weak indication that 5-BU had been incorporated into the phage DNA. When Meselson performed a comparative viability test by subjecting the two stocks to the same dose of ultraviolet radiation, the outcome pointed the other way. The ratio of viable to UV-inactivated phages was almost the same for the 5-BU and the parent stock. Because 5-BU substitution should have made the phage more sensitive to the radiation, this result suggested, as Stahl commented in 1992, “that there was no incorporation of 5BU, and . . . [the preparation] was a dud.” 19 Acting at the time, however, on the ray of hope provided by the three 5-BU mutant plaques, Meselson and Stahl proceeded to test whether they could separate the two DNA shockates in the ultracentrifuge. Wasting no time, Meselson prepared immediately for the first run with “.027 ml 5BU and 0.0025ml parent shockates . . . added to 1ml. CsCl of 23 deg density ⫽ 1.71,” concentrations devised to produce similar optical densities of the two types. Beginning the run (number 739) on the afternoon of 1 January, he accelerated again to 44,770 RPM and continued through the morning of 2 January. When he developed the films, he found a “mystery band.” It was nearly as sharp and narrow as the band obtained two days before from the parent shockate alone, but was displaced about one-third of the way toward the “heavy” end of the cell.20 That location might have appeared to Meselson to fit well with an expected density of 1.8 for fully substituted DNA (if he was assuming then, as he did three weeks later in a letter to Watson, that the density gradient in the cell was 0.4gm/cm4); but in that case, where was the band representing the unsubstituted stock? An alternative interpretation was that the effective density of the CsCl was somewhat less than intended (or, more likely, that the density of the CsCl in the previous run had become a little more dense through evaporation),21 that very little 5-BU had been incorporated into the phage DNA, and that “5-BU” and “parent” phage DNA had banded essentially in the same place. For now, the band remained a mystery.

W ORKING

AT

H IGH S PEED

225

Thinking that he might be able more easily to detect the incorporation of 5-BU into whole phage than into the DNA released in the shockates, Meselson prepared a suspension of 0.012 ml of 5-BU sample and .0032 ml of parent sample added to 1 ml of CsCl solution as in run 739. (Perhaps there was, here, an omission in the protocol, because whole phage should have been run in a less dense medium than he had used with the shockate.) He established by a bio-assay that “1.8 ⫻ 1010 phage are present.” He started run 740 nearly as soon as the previous one was over and continued it until 2:30 the next morning. By then it was evident that no bands were going to form, so he closed down the operation.22 Suspecting, perhaps, that there might have been some sort of interaction between the substituted and the normal phage, Meselson decided to centrifuge them in separate runs. He began with the parent stock alone. He placed half as much of it as in the previous run into the same CsCl solution used in that run, reducing further the calculated optical density of the cell. Just prior to the run, which he began on 3 January, he coated the quartz windows of the cell with a silicone compound that was thought to reduce leakage. Twenty-three hours of running time sufficed to produce a sharp band.23 The cumulative results of these experiments were by now driving home to Meselson how tricky it was to tell where the bands were. A slight change in the average density of the solution, such as could easily be caused by evaporation, would move a band representing the same DNA to a different place in the cell. He could not really compare one run with another, he thought, unless he could put in some sort of standard marker.24 In an effort to provide such a marker, Meselson reran the contents of the previous run the next day with carbon tetrachloride (CCl4) added. Knowing that the density of CCl4 was 1.59, he surmised that if it formed a band in the cell, he might be able to “get a measure of where in the gradient a certain density is.” He quickly gave up this first attempt. “No photos,” he wrote in large handwriting on the centrifuge log. “CCl4 seems to band tho’ run was terminated too soon to be sure.” 25 On the afternoon of the same day (4 January), Meselson began another run (number 743) with the same substituted whole phage, T4 r240 5-BU “B” used in run 740, placed in a CsCl solution of density 1.48. He terminated the run, after about twenty hours, late in the morning of 5 January. The films showed no discrete bands.26

226

W ORKING

AT

H IGH S PEED

Following these indecisive results, Meselson turned from the “B” substituted phage stock that he had prepared from a low multiplicity infection to the “A” stock prepared with a ratio of phage to bacteria representing high multiplicity. (See above, p. 223.) He placed both thymine and 5-BU “A” shockates in the cell. The density of the CsCl was not recorded, but was probably the same (ρ ⫽ 1.71) solution used in previous experiments with shockates. He spun the machine at the speed (44,770 RPM) that he seemed by now to have settled on as his standard, for about eighteen hours, and obtained again a single, sharp band, despite having centrifuged two stocks that should have differed in density. The band was nearly indistinguishable, in both position and width, from the one he had obtained a week earlier (run 735) from T4 wild DNA. The most likely inference was that little or no 5-BU had been incorporated into the phage DNA produced through the high multiplicity procedure and that he was, therefore, seeing only two consolidated samples of unsubstituted DNA.27 Meselson had, by now, run the ultracentrifuge daily for eleven straight days and had totally monopolized the machine for the last five days. He had to yield it to other users for four runs, but nevertheless was back two days later, on 7 January, for another turn. He used it to try again to elucidate the nature of the “mystery band” of run 739 by recentrifuging the contents of the cell used then, with carbon tetrachloride added as a marker. He terminated the run after six hours. The next day he repeated the experiment, after adding a little more CCl4 and enough CsCl of density 1.477 to reduce the overall density of the solution from 1.71 to 1.602 to match the density of the CCl4, which he wrote down as 1.60. The marker strategy failed again, as Meselson noted “CCl4 floats in cell at rest,” and he obtained only a single band located in about the same place as the original mystery band.28 On 8 January Meselson returned to Jardetsky’s calf thymus DNA, which he put into the centrifuge cell in the usual (ρ ⫽ 1.71) CsCl solution. He ended the run after thirteen hours, when the DNA was forming a band in the expected position for unsubstituted DNA but had not yet come to equilibrium. Perhaps all he needed, at that point, was to reassure himself of the regularity of that observation.29 Having apparently negotiated the sole use of the ultracentrifuge for another week, Meselson raced on. He and Stahl had produced a new 5-BU phage stock, which they labeled “L.” (There is no surviving protocol for this preparation, so it cannot be determined whether they

W ORKING

AT

H IGH S PEED

227

modified the procedures used in the preparation of the “A” and “B” stocks.) On 9 January Meselson centrifuged a mixture of “5BU ‘L’ & Thy shockates T4r240,” probably in the usual CsCl solution of density 1.71. Once more, he obtained only a single band, in approximately the position expected for unsubstituted DNA.30 Was he again finding only that the new procedure had also failed to incorporate 5-BU into the T4 DNA? To find out, Meselson next centrifuged the T4 r240 “L” 5-BU shockate alone. He began the run just after midnight on 11 January. When he ended it thirteen hours later, no bands at all were forming. The answer was not, apparently, that there was no 5-BU in the 5-BU DNA but that something else was preventing it from banding. One possibility was that it was slightly too dense to fall within the boundaries of the density gradient and had gone to the bottom of the cell. To find out, Meselson next placed a more concentrated suspension of his 5-BU “L” shockate in a somewhat more concentrated CsCl solution (ρ ⫽ 1.779 in place of the usual 1.71). In another attempt to provide a density marker, he added fourteen small drops of N-13, which was one of a series of fluorocarbons of graded density marketed by General Electric.31 The marker again failed, but the change in density paid off. Late in the evening of 13 January, after nearly twenty hours of operation, Meselson developed the first set of films that showed a definite band forming. He saw from the meniscus at the light end of the film that the cell had leaked a little, but not enough to spoil the result. When he shut down the machine an hour later, after taking two more film sequences, the band remained faint and diffuse, but its location, two-thirds of the way toward the heavy end of the cell, was consistent with the expected density (ρ ⫽ 1.8) of substituted DNA (figure 7.5). Strongly encouraged that they had, at last, been able to prepare and band 5-BU DNA, Meselson and Stahl moved in the most direct way possible back to their immediate goal, the separation of substituted from unsubstituted DNA. To do so, they simply added 0.003 ml of T4 r240 thymine shockate to the centrifuge cell without emptying the contents of the previous run. Meselson began the new run at 1:10 on the afternoon of 13 January. He encountered some technical problems with the machine. The “weak” brake seemed erratic, and the control voltage varied. By early the next morning, however, when he developed the film from the third series of photos, he knew that they had succeeded in their immediate quest. The broad 5-BU band appeared in the same place that it had when run alone, and a much sharper

228

W ORKING

AT

H IGH S PEED

Fig. 7.5. Film from centrifuge run 755, exposure 6

band showed up about where the normal DNA would be expected, near the light end of this somewhat denser CsCl solution (figure 7.6).32 The result showed that the 5-BU DNA was far less homogeneous than the thymine DNA, which meant that they had not yet achieved a uniform incorporation of the substituted base. Nevertheless, they had now demonstrated beyond a doubt that their density gradient method was capable of the basic task for which they had devised it. There was not only a clean separation of the substituted from the normal DNA, but, as they put it a week later in a report of their progress to Schachman and Smith at Berkeley, “There is plenty of room between the two peaks to detect molecules of intermediate density which may be produced by growing substituted phage in clean bacteria.” 33 This sudden success, after almost four weeks of relentless preparation and round-the-clock centrifuging of phage DNA, must have been equally satisfying to both Meselson and Stahl. Reporting the same result to James Watson three days later, Meselson wrote, We have banded 5BU DNA from a T4 stock which was about 60 per cent substituted as determined chemically. To the same centrifuge cell contents we then added [“with thorough mixing,” he added in the margin] some normal T4 DNA and found a very healthy separation of the two sorts of DNA. Plenty of room for finding intermediate pieces from transfer exps. Also this mixed run looks just like the sum of separate runs each with only one sort of DNA. This relieves us of interaction worries.34 This “healthy separation” between the two peaks was, in fact, not only more than they had expected but more than it should have been. During the course of a theoretical analysis of the density gradient method

W ORKING

AT

H IGH S PEED

229

Fig. 7.6. Film from centrifuge run 756, exposure 6

they were using (see the following section), Meselson calculated the density gradient in the centrifuge cell. Proportional to the square of the velocity, the gradient for runs at the 44,770 RPM at which he had spun the centrifuge during these experiments he found to be “about .4 g/cm4.” The density difference between the two peaks, according to Meselson’s estimate that fully substituted DNA would be 6 percent heavier than normal DNA,35 would have been 0.06 gm/cm3 for 60 percent substituted DNA. In a density gradient of 0.4gm/cm4, the two bands ought to have been separated by about 0.15 times the length of the centrifuge cell. Instead, they were at least half the length of the cell apart. Commenting on this situation to Watson, Meselson wrote, The observed density of the 5BU DNA is interesting. One can calculate a density supposing that bromine replaces methyl with no other modification in the DNA. The observed density difference is over twice the expected one. My guess is that this is due to ionization of the 5BU as might be expected in such strong salt at pH 8.5 where we run. The ionized 5BU would have cesium fairly strongly associated with it instead of hydrogen and would therefore exhibit an increased density. According to this explanation, lowering the pH should move the 5BU band up. If this can be observed, electrophoresis might do a good job of separating the two kinds of DNA.36 Despite the auspicious progress they had made with the centrifugation method, Meselson thus continued to entertain the possibility that he would turn eventually to electrophoresis for the separation.

230

W ORKING

AT

H IGH S PEED

III While they were achieving these significant results with the ultracentrifuge, Meselson and Stahl came to believe that they needed to understand at a deeper level what was happening in their experiments. They felt as if they were entering some kind of risky territory, and, therefore, that they really should know what they were doing.37 Without knowing at first what the impact on their investigation might be, Meselson undertook, sometime in January, a theoretical analysis of the processes underlying the observed phenomena. His point of departure was the treatment of the theoretical considerations underlying the standard methods for measuring molecular weights in the ultracentrifuge in the monumental monograph The Ultracentrifuge, by The Svedberg and Kai Pederson. There were two basic methods. One was to measure the rate of sedimentation of the solute molecules through a liquid of uniform density under the force of the centrifugal field. The other method, and the one most relevant to Meselson, was to allow the centrifugation to go on long enough for a state of equilibrium to be reached, “in which sedimentation and diffusion balance each other.” The equation for sedimentation Svedberg derived from the basic expression for centrifugal force: F ⫽ M (1 ⫺Vρ)ω 2x, where M ⫽ molecular weight of the solute, ω ⫽ angular velocity of rotation, V ⫽ partial specific volume of the solute (that is, the increase in volume caused by the addition of a unit weight of the solute to a very large volume of the liquid), and x ⫽ the distance from the center of rotation. The rate of sedimentation ⫽ cω 2xM(1 ⫺ Vρ)(i/f)dt, where c ⫽ the concentration at x, and f ⫽ the frictional coefficient per mol. The rate of diffusion ⫽ ⫺RTdc/dx(1/f )dt, where R ⫽ the ideal gas constant and T ⫽ the absolute temperature. By setting these two rates as equal and eliminating the terms for the frictional coefficient, Svedberg obtained the following equation: dc/c ⫽ ⫺

M(1 ⫺ Vρ)ω 2 xdx RT

(1)

Integrating between x 1 and x 2 resulted in the equation M⫽

2RTlog(c 2 /c 1 ) (1 ⫺ Vρ)ω 2 (x 22 ⫺ x 21.

(2)

W ORKING

AT

H IGH S PEED

231

Svedberg derived the same equation from the fundamental thermodynamic criterion that, under equilibrium conditions in a closed system, the change in free energy per mol of solute along the axis of rotation must be zero (otherwise the solute particles would redistribute themselves until that condition was met).38 In the conventional Svedberg method, the solution through which the sedimenting and diffusing particles moved was assumed to be a liquid of constant density (ρ). Only the movements of the large particles, whose molecular weights were to be measured, were considered. Meselson’s first theoretical problem was to adapt the Svedberg equation to calculate the density gradient formed at equilibrium in a CsCl solution in the ultracentrifuge cell by the balance between the diffusion and sedimentation of the small molecules of the salt. Neglecting “electrical effects” due to ionization of the salt, Meselson applied the differential form of the Svedberg equation, in which now M ⫽ molecular weight of CsCl, v ⫽ partial specific volume of CsCl, and ρ 0 ⫽ density “at point of interest.” In place of the concentration gradient dc/dx, he put the density gradient ∂ρ/∂r: ρ M(1 ⫺ v¯0 ρ0 )ω 2 r 0 ∂ρ ⫽ 0 ∂r ρ0 2RT (The 2 in the denominator is a mistake, and it is not clear why Meselson put it in this equation.) The value for M was known, and ρ 0 could be calculated either from the refractive index of the solution at rest or from the known quantities of solute and solvent. Meselson could use this equation, therefore, to calculate the density gradient in the cell at a given ultracentrifuge velocity. For the runs that he had recently completed at 44,770 RPM, he estimated that the gradient was about 0.4 g/cm4 along the cell.39 Because the depth of the centrifuge cell was about 1cm, that meant also a density range of about 0.4 g/cm 3. For a CsCl solution of average density 1.5, the density at the heavy end would be 1.7, at the light end 1.3. It may have been after making this calculation that Meselson returned to the first photometer tracing from the ultracentrifuge run on T4 phage, made on 4 November (see above, p. 201), and wrote those two numbers below the corresponding ends of the graph. This range supported the assumption they had then made that the two bands represented, respectively, whole phage and DNA. Sometime after writing to Watson on 24 January, Meselson adopted

232

W ORKING

AT

H IGH S PEED

a modified equation for the density gradient which took into account the activity coefficient for CsCl: dρ ∂ρ aM(1 ⫺ v¯ρ)ω 2r ⫽ , dr ∂a RT where a ⫽ activity. Recalculating the density gradient for 44,770 RPM, he found an error in his previous calculation and came out now with 0.12 g/cm4, less than one-third of his earlier result.40 That meant that the resolution of the gradient for separating bands was more than three times as great as he had previously thought. It also meant that much of what they had been searching for during the previous weeks had been founded on misconceptions, and some of their interpretations of their results were also incorrect. Their repeated efforts to band phage in the same cell with free DNA could not work, because the density difference exceeded the range of the density gradient in the cell, and their belief that they had succeeded was a mistake. Their failure to find a 5-BU substituted band in their first several tries, they could now see, was due not to a failure to incorporate the 5-BU into the phage but to the fact that Meselson’s miscalculation had led him to choose the wrong density for the CsCl solution. Probably most pertinent at this stage in the investigation, the separation between T4 DNA and 5-BUsubstituted DNA finally observed in the crucial centrifuge run (number 755) of 13 January now fit closely with theoretical expectations. There was no further need for Meselson’s hypothesis that ionization increased the density of 5-BU substituted DNA.41 To determine the theoretical density distribution of DNA in cesium chloride solution, Meselson again began with a variant of the Svedberg equation, but he now had to take into consideration a system containing three components—water and two solutes, each of the latter at an equilibrium formed by the balance between sedimentation and diffusion. Meselson’s original derivation of an equation to express the distribution of the concentration of DNA in cesium chloride has not survived (he might have worked out only afterward the formal derivation included in his thesis four months later [see below, pp. 263 ff.]), but it turned out to conform to the standard forms of the equation for a general normal curve, such as the following: y⫽



1 ⫺(x2/2σ2) e or, y ⫽ 2π



冤 冥

1 x2 exp ⫺ 2π 2σ 2

W ORKING

AT

H IGH S PEED

233

From this result Meselson concluded that the density changes of the DNA across the width of the band followed a Gaussian distribution curve. He communicated an early version of his equation to James Watson on 24 January, along with the following explanation: In such a density gradient, the centrifugal field will tend to concentrate any substance which might be introduced into the place where its density is the same as that of the CsCl solution. Diffusion will oppose this tendency and the result will be a gaussian concentration distribution of the introduced substance if it is a singular molecular species. This distribution is given by the following formula: (2) C(r) ⫽ A i



miβ exp ⫺[m i β (r ⫺ r i ) 2] π

where A i ⫽ the total amount of the species with molecular weight m i and having the same density as the CsCl located at r i ; β ⫽ a constant involving the same temperature, angular velocity, partial specific volume as the ith species, the density gradient and the distance to the axle. “If Beer’s law holds good for all species present and if the extinction coefficient is the same for all species,” Meselson wrote, then the concentration of DNA at any point (r) along the gradient “may be obtained directly” from the ultraviolet absorption.42 (Beer’s law states that the absorption of monochromatic light by a solution is described by the following relation):43 optical density ⫽ log





intensity of emergent light intensity of incident light

In a letter written three days earlier to Schachman and John Smith, Meselson wrote out an adaptation of his equation to this assumption that replaced the concentration by the optical density at a position x from the center of a DNA band: O.D.⫽ 1/⑀ ∑ n i m i



m i β ⫺miβx2 e π

where n i ⫽ number of molecules of molecular weight mi present in the cell. ∂ρro ∂x 2RT

vω 2 β⫽

where r o is the distance from x ⫽ 0 to the axis of rotation.

234

W ORKING

AT

H IGH S PEED

In addition to assuming that a given band contains material of unique v [specific partial volume], the above assumes that r o ⱭⱭ x, ∂ρ/∂x ⫽ constant [that is, constant density gradient], O.D. ⫽ 1/⑀ conc. DNA [following Beer’s law], and that DNA behaves as an ideal solute. By means of formulas integrating the appropriate values over the width of the band, Meselson added, one could also determine the number of molecules of DNA in a band, the number average molecular weight of the DNA, and the weight average molecular weight. “If v is not unique within a band,” he added, “these molecular weights are lower limits.” 44 In the equation for a normal curve indicated above, σ is defined as the standard deviation. The equivalent expression applied to the distribution of DNA in the cesium chloride gradient would define the natural band width. For Meselson it became an urgent matter to determine σ, because he still feared that the band width might turn out to be greater than the distance between substituted and unsubstituted bands.45 At some point in his analysis in January, he derived this expression for the calculated quantity: σ2 ⫽

2RT Mv¯ dρ/dr ⋅ r 0ω 2

Shortly afterward, Stahl applied this equation to the data derived from centrifuge run 752 on T4 DNA to calculate the molecular weight of the DNA. He determined σ by measurement of the parameters of the densitometer curve of the band, squared the result, and inserted it in the equation. For the computation, which he carried out on the upper left-hand corner of the graph paper on which the densitometer tracing had been made, he used the value 0.06 g/cm4 for the density gradient and obtained for the molecular weight M ⫽ 24 ⫻ 106. Later Meselson looked over Stahl’s calculation, changed the density gradient to 0.12 g/cm4, and corrected the result to M ⫽ 12 ⫻ 106.46 This outcome was exciting to both Meselson and Stahl, because it coincided exactly with the most recent estimate for the molecular weight of the “small pieces” of phage DNA that accompanied Cy Levinthal’s “big piece” 47 It may also have been one of the events that made Meselson and his associates think that they could have at hand a “pretty powerful method for examining big molecules.” 48

W ORKING

AT

H IGH S PEED

235

III Sometime probably in late January 1957, Stahl wrote up the annual report to the National Foundation for Infantile Paralysis for the period January–December 1956. He was able to report optimistically on the project titled “Incorporation of bromouracil into phage (Stahl, in collaboration with M. S. Meselson),” The physical properties of phage and of DNA from phage obtained under conditions of 5-bromouracil incorporation are being investigated (by Meselson) using a newly developed method. Both phage T4 and DNA obtained from it have been centrifuged to near equilibrium in strong aqueous solutions of cesium chloride. Under these conditions, sedimentation of the cesium chloride results in a density difference of a few percent across an ultracentrifuge cell. The phage (or DNA) collect in a band at a point where their effective density is the same as that of the cesium chloride solution. Bands observable by UV absorption or Schlieren optics are formed by quantities of DNA ca. 1µg. Attempts are being made to interpret band shapes and positions in terms of molecular weight and heterogeneity of composition of DNA’s from different sources. Preliminary results indicate that it is possible by this method to separate normal DNA from the more dense DNA obtained from phage T4 prepared under conditions of 5-bromouracil incorporation.49 His reference to a “density difference of a few percent” suggests that Stahl wrote this paragraph after Meselson had corrected the density gradient to 0.12 g/cm.4 The heady combination of these experimental and Meselson’s theoretical advances during the first half of January once again upset the order of priorities for their joint investigation, nearly reversing the way both had viewed the situation just before Christmas. Not only did density gradient centrifugation seem far more promising for its intended purpose than it had three weeks earlier, but it was beginning to appear as an important general method applicable to other studies as well. Vinograd also became interested in the possibilities the new method offered, and he encouraged Meselson to push it forward, particularly because he thought that Schachman might be pursuing similar methods at Berkeley. Vinograd’s competitive instinct was undoubtedly convenient for Meselson and Stahl, because it made it easier for Meselson to claim the large amounts of time on the ultracentrifuge that their project now demanded.50

236

W ORKING

AT

H IGH S PEED

The latest upheaval in their sense of priorities was reflected in the way Meselson and Stahl reported on their activities to colleagues near the end of January. On 21 January, Meselson, Vinograd, and Stahl jointly sent a letter, probably drafted by Meselson, to John Smith and Schachman, to summarize “our progress to date and . . . outline where we hope we’re headed. Our main aim in the 5BU business is to elucidate the mechanism of mutagenesis by 5BU. As we guess we’ve told you, we will concentrate our efforts on the occurrence of mutations among the offspring of substituted phage grown in clean bacteria. As to be expected, we’re still nowhere on this problem. Our main difficulty is that we’ve been (temporarily) sidetracked by the promise of being able to learn something about the inheritance of phage DNA as revealed by behavior in the ultracentrifuge.” 51 (When shown this letter in 1987, Meselson was so surprised that the idea of studying the inheritance of DNA by ultracentrifugation “which predates my meeting with Frank Stahl” appeared here as an offshoot from a “main aim” directed at mutagenesis, that he wondered at first whether Stahl might have written the letter. He acknowledged his own authorship when he noticed, on the second page, the equations representing his theoretical analysis of the density gradient method.52 In January 1957, Meselson was reporting a sharp shift in priorities that was taking place on a time scale of weeks. In 1987, no longer thinking about events at so fine a scale, he remembered his idea for studying DNA replication as the steady compass guiding the direction of his endeavor.) Near the end of this joint letter, the authors reported, “We plan to submit a description of the density-gradient method to Nature in the very near future. The primary purpose of the note will be to make this method (which looks pretty powerful) available to other workers.” 53 The letter conveys the strong impression that Jerry Vinograd was now a full partner in the project that Meselson and Stahl had begun five months earlier. The nominal authors of the letter were unaware of the degree to which they differed in their perceptions of the extent to which that had happened. In Stahl’s view, Meselson had been responsible for all of the key developments, both theoretical and experimental, concerning the density gradient method. In Meselson’s view, Stahl shared in these developments, but Vinograd contributed little beyond the generous donation of time on his ultracentrifuge. Vinograd quickly came to view himself, however, as one of the co-inventors of the new method.54 The sense that the events of early January had diverted Meselson

W ORKING

AT

H IGH S PEED

237

and Stahl from their previous intentions was reiterated, with somewhat different emphasis, in Meselson’s letter to Watson on the twentyfourth: “The experiment you once recommended be done only in Sweden has temporarily sidetracked Frank and me from our main project of elucidating the mechanism of 5-BU mutagenesis. Instead, we are well on the way to doing transfer experiments based on the separability of 5-BU DNA from normal DNA in the ultracentrifuge.” 55 As Meselson continued to describe details of the density gradient method for Watson, however, the transfer experiments themselves seemed to be overshadowed, at least in the short run, by an anticipation that a powerful use for the new method might prove to be the determination of molecular weights for large molecules such as DNA. After reproducing his equation for the distribution of DNA in CsCl (see above, p. 233), Meselson discussed this prospect: The assumption [made in his equation] of there being only one [molecular] species is . . . definitely not valid but this just requires that a summation over i be performed in calculating the concentration. The one hooker is that both the band shape influencing factors of molecular weight and density may vary from molecule to molecule. If it weren’t for the possibility of small density differences (order of .005) existing between molecules, we could calculate very accurate molecular weights or, for weight-heterogeneous mixtures, molecular weight means. However, if density heterogeneity is neglected, one still gets a minimum estimate of molecular weights. Of course, if the bands are skewed or poly-modal, there is density heterogeneity.56 Meselson enclosed a densitometer tracing of T4 shockate DNA showing little of either of these deviations from a normal distribution curve and a tracing of calf thymus DNA that “shows skewness and does not band sharply compared to shockate.” The prospects for the method as a means to determine molecular weights thus appeared enticing but uncertain. They rested on the degree to which they could neglect the possibility that the molecules differed in density. Among the incentives that induced them, nevertheless, at this juncture to shift their emphasis to the exploration of such a method was that the more conventional ultracentrifuge techniques so useful for the measurement of protein molecular weights had proven less adequate when applied to DNA. Both the sedimentation and the diffusion coefficients required for those methods varied more rapidly with the concentration for nu-

238

W ORKING

AT

H IGH S PEED

cleic acids than for proteins. The extrapolation to infinite dilution customary to obtain the most accurate values for these coefficients for proteins was sometimes not practicable for the results obtained with nucleic acids. These difficulties had prompted Paul Doty at Harvard to develop, in the early 1950s, light-scattering methods that yielded more consistent values, but these values were approximately four times as great as the values obtained by sedimentation methods.57 The unsettled state of the molecular weight of DNA at a time when this particular species of macromolecule had suddenly acquired great biological prominence invited the entry of new methods into the field. To test the potential of their method, Meselson wrote Watson, “We would like very much to make measurements on a DNA which Paul [Doty] has studied by light scattering.” He requested that Watson act as the intermediary to procure a “suitably fresh preparation” from Doty.58 Although the determination of accurate molecular weights was the most prominent of the new uses that Meselson and Stahl now saw for the density gradient method, their fertile imaginations quickly added other possibilities for it. “From the mean molecular weight of viral nucleic acid,” they inferred, “one may calculate the number of pieces of DNA present in the virus, providing a representative DNA preparation is available. The only other necessary information is the amount of nucleic acid per virus particle.” Having found evidence for Levinthal’s little pieces of phage DNA, they thought they might be able to isolate the big piece. They even believed that they might be able to do something about “sweet and sour DNA.” 59 This phrase was used informally among members of the phage group to refer to a distinction between two forms of phage DNA. Jerry Wyatt and Seymour Cohen had found in 1954 that phage DNA contained 5-hydroxymethylcytosine in place of the usual base, cytosine. Later that same year, Sinsheimer discovered that in some of the DNA in phage T2r⫹ glucose is substituted for the hydroxyl group on 5-hydroxymethylcytosine. Jean Weigle, George Streisinger, and others in the Caltech phage group became very interested in what Delbru¨ck called, in a 1956 letter to Sinsheimer, the “sweet-and-sour T2, T4 story.” 60 Having read a paper describing some complexes that certain sugars can form with borates, Meselson thought that it might be possible, by attaching borate ions to the glucose in “sweet” phage DNA, to exaggerate the difference in density between it and “sour” DNA sufficiently to separate them in the density gradient. In his workbook he took notes on the article, under the heading, “Binders for Sweet and Sour DNA.” 61

W ORKING

AT

H IGH S PEED

239

With youthful verve, Meselson and Stahl leaned hard into the latest curve in their research trail. On 29 January Stahl wrote to his mentor Gus Doermann, “We’re presently going full time on the fundamental characterization of Matt’s density-gradient technique. It’s a powerful way to detect density in-homogeneities and a lovely way to determine molecular weights—considerably superior to other techniques available for sloppy molecules like DNA. In our ‘spare’ time we’re pushing the DNA transfer experiment. We should know soon if Levinthal’s ‘20%’ piece is half as dense or half as big as the ‘40’ piece.” 62 The newest directions they were taking would not, Meselson and Stahl seemed certain, cause them to lose sight of their earlier goals. These latest topics, Meselson assured Watson, were “less dear to our hearts than the 5BU business.” After mentioning the “healthy separation” of 5-BU from normal DNA that they had just achieved in their centrifuge experiments, he added that “our main effort now is to learn how to get homogeneously substituted stocks so that intermediate DNA’s from transfer experiments may be resolved.” 63 This effort Meselson left mainly to Stahl. By chromatographic analysis, they had found that in the 5-BU DNA used in the centrifuge experiments in January the average substitution of 5-BU for thymine was only about 60 percent. Moreover, the shape of the 5-BU bands suggested that “the substitution was not very homogeneous.” 64 Besides attempting to improve on both counts by varying the protocol for the production of substituted phage, Stahl worked further on the development of a method to use X-ray survival curves as a measure of percentage of substitution. For a long time the curves he obtained were erratic. By the end of January he thought that the trouble seemed to have been photoreactivation (that is, an effect in which otherwise inactivated phage are able to produce viable progeny in bacteria illuminated by visible light) and reported, “My X-ray survival curves seem at last to be reproducible.” 65 He had not been able fully to determine the relationship between percentage substitution and X-ray sensitivity, but he supposed them to be more or less proportional. The nature of the X-ray survival curve for the stocks they had used for the successful experiments in January strengthened their belief that the substitution was heterogenous. The rate of inactivation of bacteriophage by X-ray irradiation was known to be exponential.66 That is, if the phage were homogeneous in their sensitivity to the irradiation, the relation between the fraction of phages surviving after exposure to X-rays at a given dose, and the time of exposure, followed

240

W ORKING

AT

H IGH S PEED

Fig. 7.7. Reconstruction of curve obtained by Stahl when plotting phage survival against time of exposure to X-rays. Drawn by Stahl for the author in 1996.

a curve of the form n/n 0 ⫽ e⫺kt. When plotted in log form, the results should give a straight line with a negative slope equal to ⫺k. If, however, the population of phage was not uniform in its sensitivity, then those of differing sensitivity would have survival curves with different constants. A log plot would begin to descend with a slope representing the more sensitive phage and then level off. The curve that Stahl obtained suggested that “some phage (about 20% of them) were considerably more substituted than the majority of the population” 67 (see figure 7.7). Meselson had speculated in December that 5-BU substituted phage would be more sensitive to X-rays than unsubstituted phage because of the high X-ray cross-section of bromine. If so, then they “should be able to increase the effect with an X-ray tube having a high output near the K-shell edge of bromine.” The substituted phage would absorb more X-rays, which would also have a very short range. A tube with a molybdenum anode ought to be able to supply such rays. “If we can effect an appreciable increase in this way,” he wrote in the letter to Berkeley in late January, “X-rays should be a very nice tool for detecting heterogeneous incorporation.” 68 Having no such tube readily available, Stahl tested Meselson’s idea by irradiating 5-BU phage through a layer of a solid containing bromine to absorb the specific X-ray wave-length that would be the cause of the increased sensitivity. He and Meselson nearly choked on the bromine. The outcome of the experiment—that 5-BU substituted phage were not more sensitive than normal phage to X-rays lacking the wavelength absorbed by bromine—supported Meselson’s view that it was the bromine-specific wavelengths in the unfiltered X-rays that had caused the difference

W ORKING

AT

H IGH S PEED

241

in sensitivity. The next step should have been to procure an X-ray source, such as the molybdenum tube, that would be rich in the bromine-specific wavelength, or to filter out the other wavelengths from the copper anode they were presently using and then to irradiate the phage for a very long time to compensate for the low density of the bromine-absorbing wavelength and find out if the power of X-rays to discriminate between 5-BU and normal phage would be increased. This step they never took.69

IV During the two weeks following his marathon runs of early January, Meselson used the ultracentrifuge only four times. During this period the machine was run more than thirty times by other users. In his continuing sporadic efforts to create a density reference for the cesium chloride gradients, Meselson placed a glass ball along with DNA and CsCl in a centrifuge cell on 17 January (“that was a horrible thing,” he commented when looking over the log in 1992), but he aborted the run early. The next day he added three drops of fluorocarbon N-43, another in the family of graded density markers produced by General Electric, to 0.01 ml of T4r 240 shockate, in a CsCl solution of density 1.71. The known density of the fluorocarbon was the same as the expected density of the DNA. The appearance of a single band on the film reassured him that “we had the densities measured about right.” 70 In an attempt to double the efficiency of the ultracentrifuge runs, Meselson placed two cells with different preparations in the rotor on 26 January. One of the cells was fitted with an optical wedge that displaced the light so that the image of a wedge cell appeared separate from the image of the other “non-wedge” cell spun at the same time. In one of the cells he placed 5-BU phage shockate “A” from the preparations made in December, together with a very small amount of fluorocarbon. In the other he placed shockate from unsubstituted T4r 240 phage. Early in the run the non-wedge cell leaked. Meselson took photographs alternately of the wedge and the non-wedge cell, but no useful result came from this elaborate effort.71 About this time Meselson adopted a simpler way to purify CsCl. Because the only purpose for the purification was to rid the solution of substances that would absorb ultraviolet light, he tried out the decolorizing charcoal “Norite,” marketed by a Dutch firm, that had long

242

W ORKING

AT

H IGH S PEED

been used for whitening sugar and for many other similar purposes. This treatment, he found, removed an impurity whose absorption maximum otherwise interfered with the photometry of the DNA bands. It also eliminated the hazardous process in which Meselson had been heating cesium perchlorate in a red-hot furnace.72 The task of determining molecular weights that Meselson and his associates now envisioned for the density gradient method set more rigorous demands on the method than their previous objectives had required. The first of these was that the equations relating the band width to the molecular weight assumed a true equilibrium between sedimentation and diffusion. Hitherto, Meselson had continued his runs until the gradually contracting bands on his films appeared to change no further. Because the centrifugal force exerted on the molecules continually diminished as they neared the place at which their buoyant density equaled that of the solution, however, he would expect the final approach to equilibrium to be so slow that he could not be certain of having attained it even during the longest runs so far conducted. In his letter to Watson on 24 January, Meselson wrote, “We are in process of doing careful runs (ensuring equilibrium, etc.) in order to get molecular weights for shockate DNA.” 73 Strictly speaking, on 24 January Meselson and Stahl were only planning such a run. It was to be, even by their own standards, so lengthy that they may have had to wait in line for some time until they could reserve the machine for more than a week. The fundamental test for any equilibrium state is that it be the same from whatever direction it is reached. In the case of the density gradient, the equilibrium bandwidth was a function of the speed of rotation of the centrifuge. The equilibrium for a given speed could be approached from both a higher or a lower speed. At the higher speed the band would be thinner, at the lower speed it would be wider. If the final band width were the same when approached from either direction, that would constitute an “absolutely rigorous demonstration that it is at equilibrium.” 74 The Spinco Model E became available for the long run on the evening of 31 January. Meselson placed the shockate from T4r 240 phage into a CsCl solution whose density, after recrystallization and treatment with Norite, was 1.697. Starting the run at eight minutes before midnight, he accelerated the rotor initially to 39,460 RPM. The series of photos made between 8:46 A.M. and 1 P.M. the next day showed already a clear single band forming. He maintained the run at this speed

W ORKING

AT

H IGH S PEED

243

Fig. 7.8. Film from centrifuge run 793, exposure 6 (the band is in the left third of the film, the thin line to the right is the meniscus)

until the next morning, for a total of thirty-one hours. The band was sharp and “tight” (figure 7.8). At 7 A.M. Meselson lowered the RPM to 27,690 RPM. The first series of pictures taken over the next four hours showed the band growing slightly wider and more diffuse, after which it stayed constant for the rest of the thirty-three hours that he kept the centrifuge running at this speed (figure 7.9). At 3:47 on the afternoon of 3 February, Meselson

Fig. 7.9. Film from centrifuge run 793, exposure 10

244

W ORKING

AT

H IGH S PEED

Fig. 7.10. Film from centrifuge run 793, exposure 14

dropped the speed to 19,160 RPM. Within a few hours the band became noticeably wider and looser. Near the end of the thirty-three hours during which the centrifuge ran at this low speed, the bands appeared as in figure 7.10. For the last stage of the run Meselson returned at 2 A.M. on 5 February to the 27,690 RPM setting of the second stage. Indefatigably he decided to keep the durable machine spinning for at least two more full days. Some signs of trouble showed up by midnight the next evening. Worried about the temperature regulation, he decided to take a quick series of graded exposures to check the conditions in the cell. The next morning, Vinograd developed for him the films taken through the night with the automatic timer and saw that they were extensively underexposed. The cause, he found, was a thick layer of material deposited on a quartz guard disc through which the ultraviolet light passes into the vacuum chamber. After he cleaned it the problem was solved, but he noticed a further deposit the next morning at 2 A.M. and removed the disc. At 7 A.M. (6 February) the temperature had “dropped about 2.5°C,” and Vinograd had to “revive” a relay to restore control. The films taken that day were ruined by “white light in the dark room.” The centrifuge ran through the early morning of 7 February, but no pictures were taken. At 7 A.M. the temperature had again fallen, this time by two degrees, and again there were regulator worries. The situation was stabilized by 8 A.M., and Meselson himself noted a little later, “speed control good.” He took two more series of pictures before finally ending the marathon run at 8:30 A.M. (figure 7.11).

W ORKING

AT

H IGH S PEED

245

Fig. 7.11. Film from centrifuge run 793, exposure 23

Despite all these technical distractions during the last two days of the run, the outcome was decisive. At 27,690 RPM the band had again tightened from the broader form it had displayed at the lower RPM, until it appeared visually identical in width to what it had been at the same RPM after the expansion of the tight band initially formed at a higher RPM. That result proved definitively, for Meselson and Stahl, that the DNA molecules in their CsCl density gradient were actually at equilibrium.75 This knowledge was essential to their quest for the molecular weights of the molecules. It could only help them, however, if they could also verify that the density distribution at equilibrium fitted exactly a Gaussian curve. If it did, then the width of the band, as defined by the standard deviation, σ, could be used to calculate a molecular weight for the DNA. This value can be represented directly on such a curve as the horizontal distance from the mean to the curve at approximately two-thirds (0.6) of its height (see figure 7.12).76

Fig. 7.12. A normal distribution curve (see text at n. 76). The value of σ is the horizontal distance from 0 to 1 on the x axis, or the distance from the vertical axis to the point at which the arrow shown vertically above 1 touches the curve.

246

W ORKING

AT

H IGH S PEED

About this time Stahl recognized that there was a rigorous way to test whether the symmetrical curves they had obtained for the distribution of DNA molecules were true Gaussian curves. From the equation for the curve,



y ⫽ exp ⫺



x2 , 2σ 2

it followed that a plot of ln y against x 2 yields a straight line with the slope ⫺1 2σ 2 By graphing the data from densitometer measurements in this form, a straight line would not only confirm that the curve was Gaussian, but its slope would yield a value for the standard deviation derived from all of the data, rather than from only two points on the curve. An accurate value for σ was essential to an accurate calculation of the molecular weight from the equation whose variables included σ 2. If the line turned out not to be straight but showed downward concavity, Stahl inferred, that would indicate that the density of the molecules was heterogeneous, whereas upward concavity would be presumptive evidence for heterogeneous molecular weights.77 To make measurements of the density distribution that would be accurate enough for these purposes, however, Meselson and Stahl first had to find a way to produce more accurate tracings of the optical density variations across the band than was possible with the limited resolution of the Spinco Analytrol instrument available in Vinograd’s laboratory. From the subbasement where it had been stored, Meselson and Stahl were able to borrow a Sinclair Smith microphotometer that had formerly been used to measure the light intensities of stars on photographic plates. They set the instrument up in a closet. When a film was placed in the photometer, the light source in the instrument passed through a slit adjusted to 0.25 mm in width, producing an image of a small portion of the film enlarged fourteen times. The light emerging from the film entered a photoelectric cell connected to a Wheatstone bridge. When one balanced the resulting current, one obtained a galvanometer reading proportional to the intensity of the light passing through that portion of the film. By moving the slit across the

W ORKING

AT

H IGH S PEED

247

film and taking readings at small intervals, one could obtain precise measures of the variations in the optical density of the film in the direction of movement.78 Stahl took on the task of operating the photometer, which proved rather cumbersome to use. His first attempts gave erratic results. When Meselson and Stahl, who were sitting together at the instrument, realized that the trouble arose from smoke getting into the light path, Stahl promptly gave up smoking. Thereafter the instrument functioned reliably. Before applying it to their intended objective, he tested various aspects of the system. By measuring the ratio of transmission of two chromatically neutral filters over a range of incident illuminations, he established the linearity of the photometer readings. By taking photographs through known DNA concentrations he determined that the densities of the films were directly proportional to the optical densities in the centrifuge cell. He verified that DNA in the CsCl solution they used obeyed Beer’s law at each of the three wavelengths of ultraviolet light produced by the mercury light and the chlorine-bromine filter with which the analytical ultracentrifuge was equipped.79 Even though they were now devoting “ca 85% of [their] current effort” to the molecular weight method, Meselson and Stahl were still putting “about 10%” of their time into the search for optimal conditions for the preparation of homogeneously highly substituted 5-BU phage stock for transfer experiments.80 They had, since drawing up their original research plans in the fall, planned to cause 5-BU phage to reproduce in bacteria grown in a normal medium to study the back mutations. The same strategy could serve their long-held goal to use density separation methods to study DNA replication. As Stahl put it in a letter to Robert Edgar in late February, “The hope [was], of course, that 1/2 dense pieces arise when 5-BU phage grow in clean bacteria.” That is, in accordance with the predictions of the Watson-Crick model, they hoped to find progeny DNA molecules whose density was midway between that of the substituted and unsubstituted DNA.81 Early in February Meselson and Stahl made a “one cycle” T4r 240 5-BU stock “with a high input and low burst size”—that is, they infected bacteria grown in a medium containing 5-BU with the T4 phage and isolated the progeny resulting from the first reproductive cycle. On 10 February Meselson centrifuged a sample of DNA obtained from this stock, in a CsCl solution of density 1.76, at the usual speed of 44,770 RPM, for about twenty hours. The resulting film showed a rather faint, narrow band at the light end of the gradient, a somewhat

248

W ORKING

AT

H IGH S PEED

Fig. 7.13. Film from centrifuge run 812, exposure 3. The two thin lines on the far right represent the meniscus. The narrow band is just to the left of them.

denser but diffuse band on the heavy side, and a still broader, denser band about midway between them (figure 7.13).82 It is not clear what Meselson and Stahl’s intentions in centrifuging this 5-BU stock were, or whether they expected the result they obtained. Afterward the meaning of the outcome was, however, quite clear to them. In his letter to Edgar two weeks later, Stahl wrote, “Out of this effort [to produce homogeneous, highly substituted 5-BU stock] has come a sort of cheap transfer experiment. . . . The DNA from this stock is rich in pieces having just 1/2 the density of the heaviest DNA in the stock. Suggestive but not yet tight.” 83 In a progress report written about six weeks after the event, Meselson gave a fuller description of the experiment and their interpretation of its significance: “A preliminary experiment has been performed in which thymine phage was allowed to duplicate a few times in bacteria growing in 5-bromouracil medium. The density distribution of the progeny DNA showed peaks corresponding to zero (very small peak), 50 (large peak), and 100% (large peak) substitution

W ORKING

AT

H IGH S PEED

249

of thymine by 5-bromouracil. This result is consistent with the replication scheme proposed by Watson and Crick in which the two complementary strands of a DNA double helix separate and each acts as template for the formation of progeny molecules.” 84 As auspicious as this result seemed for the eventual realization of Meselson’s long-held primary goal, it did not prompt him to reorder once again his and Stahl’s immediate priorities. The research partners continued to put their main effort “into determining the MW of phage DNA by the gradient centrifuging technique.” While Stahl worked to get all of the bugs out of the procedures for making accurate microphotometer tracings of the UV pictures of phage shockate banded to equilibrium, Meselson decided to use his spare time to pursue his ideas for separating sweet and sour phage DNA. Sinsheimer had found that all T4 phage DNA contains glucose. To procure “sour DNA,” therefore, Meselson had to turn to T2, 23 percent of which, according to Sinsheimer, was glucose-free. On 12 February he centrifuged “T2 Sour Shockate” in CsCl of density 1.704 at the usual speed and obtained within six hours a well-defined single band. Five days later he added some “T2 Sweet Shockate” to the same cell and spun them together for a day and a half, but found only the same band that had appeared in the first run. Evidently the differences in density were not sufficient to separate them without some form of enhancement.85 These were the only two runs that Meselson made between 10 February and the end of the month. Meanwhile, Vinograd himself tried out the new density gradient method on other macromolecules and cellular particles. He began on 8 February with a preparation of ribosomes. On the thirteenth he spun m-perfluoroheptane in a solution of density 1.704 and did another run with the same material on the next day. Between 18 and 22 February he made two runs with tobacco mosaic virus and obtained bands with “very sharp” edges. The next day he turned to RNA derived from TMV and made two runs in CsCl solution (ρ ⫽ 1.70). Even though RNA could not have been successfully banded in this solution, Vinograd was satisfied after this burst of activity that the method was suitable for the study of a variety of biological materials and chemical polymers. He did no further experiments with CsCl gradients for six weeks.86 Stahl was making good progress on the microphotometer tracings. By 22 February he could report to Edgar that “all is going well now” and that “these tracings will give MW of here-to-for un-equaled preci-

250

W ORKING

AT

H IGH S PEED

sion.” He was concentrating his effort on films obtained from the long equilibrium run of 31 January–8 February on T4 shockate DNA. The photometer allowed him to take density readings at 0.045 mm intervals across the film, corresponding to 0.025 mm along the axis of the cell. This fineness of resolution gave him eighty points across the width of the DNA band. Plotting these points in log form, he obtained a straight line, from which he determined σ. “Looks,” he wrote Edgar, “as though our value for MW is going to be about 12 million.” 87 While Stahl worked on the analysis of the first long equilibrium run (number 793) with T4 shockate, Meselson began, on 2 March, another one that, he noted on the log, was “like Run 793 but with half as much DNA.” In fact, the purposes of the two runs seem to have been rather different. Whereas formerly he had spun the cell for long stretches at high, medium, low, and again medium speeds to demonstrate the equilibrium state, here he maintained the run for the first fifty-three hours at 27,640 RPM, raised it for only three hours to 44,770 RPM, then went back to 27,690 for forty-five more hours. Evidently the main objective this time was to produce films optimally suited to Stahl’s photometric analysis of the density distribution in the band.88 Stahl measured the density distribution of the new bands with great care. He made duplicate readings in two directions and found that they were identical. He extended the readings beyond the band edge to establish a base line. The logs of the observed points plotted against the squares of the distances from the center of the bands fit exactly along a straight line. Initially Meselson was disappointed that Stahl did not carry the points further down than he did, but Stahl was able to persuade him that the closer one got to the base line, which had to be subtracted out, the less certain the values became. On 6 March, the same day that the second long run was completed, Stahl again reported to Edgar, with unveiled excitement, the “latest on Molecular weight experiment. We have analysis of banded T4 DNA. The band is a perfect Gaussian.” According to his latest calculation, “MW of the DNA is ca. 10 ⫻ 10 6.” 89 To demonstrate in a visually more dramatic way the perfect Gaussian distribution of the bands, Stahl or Meselson later plotted the points in the standard form of a normal curve and compared them to a calculated Gaussian curve drawn with the same parameters. All of the observed points fell on the theoretical line (figure 7.14). Even though his own interest and expertise in his collaboration with Meselson was centered on the 5-BU mutagenesis aspect of their

W ORKING

AT

H IGH S PEED

251

Fig. 7.14. Density plot for observed distribution of phage T4 DNA, superimposed on the calculated Gaussian distribution, as reproduced in Meselson’s Ph.D. thesis. Meselson, “Equilibrium Sedimentation,” p. 4.

original plan, Stahl was being drawn, with Meselson, relentlessly into a fixation on the molecular weight problem. Already at the end of January he had sent some of their mutant stocks to his mentor, Gus Doermann, in Rochester, hoping they would be able to identify some additional genetic markers that Stahl would not have the time to find.90 By late February he was expressing some regret that the new directions were distancing him from the projected mutation studies: “We’re itching to do the SB [single burst] on mutations arising in clean growth of substituted phage. And to pick these mutants etc. etc. in our effort to write a base-pair sequence for the rII region. 5% of our efforts are directed at bemoaning the fact that we’re too busy to do it immediately.” 91 By early March there appeared at least to be more equality in the fates of the projects “dearest to the hearts” of the two collaborators, as the transfer experiments central to Meselson’s endeavor suffered postponement along with the mutation experiments cherished by Stahl: “Transfer exp[eriment]s and genetics with 5BU,” Stahl wrote, “have been neglected in our big push to get out a paper on the application of density gradient CFG for detecting density heterogeneities and

252

W ORKING

AT

H IGH S PEED

determining molecular weights.” 92 A few days later Meselson ended a letter to Cyrus Levinthal with the comment that “last year I thought I could turn into a biologist by working with Frank. Instead, to our mutual horror, he’s become another physical chemist.” 93

V In his derivation of the Gaussian distribution equation for the concentration of a substance in the CsCl density gradient, Meselson had assumed that there was only one “species” of macromolecule present. He had recognized in January that this assumption was definitely not valid. If there were density or weight heterogeneities neglected in the calculation, “one still gets a minimum estimate of molecular weights.” Skewed or bimodal curves, such as those they had already found with calf thymus DNA, indicated density heterogeneity, but T4 DNA “showed little if any skewness, and no polymodality.” 94 By March the disclosure that the distribution curve for T4 phage was a perfect Gaussian persuaded him and Stahl that, in this case at least, they were dealing with molecules of uniform density and molecular weight. Aside from a sharp boost in their confidence in the new method, the most immediate consequence Meselson and Stahl drew from this conclusion was that it conflicted with Cy Levinthal’s claims about big and little pieces of bacteriophage DNA. In his letter to Edgar of 6 March, Stahl hinted at what they were thinking: “Therefore, in our experiment all pieces of DNA are of the same size. Hmmm. We will try to get Cy to see if his big piece falls apart in Cesium chloride.” 95 A week later Meselson wrote directly to Levinthal to break the news to him: Dear Cy, DNA from T4r shockate banded in CsCl in the analytical Spinco seems to contain only one molecular weight species rather than big and little pieces. We expect each mw species will distribute itself at equilibrium in a band of Gaussian shape. If several species are present, the observed band will be a sum of the individual Gaussians. The one band we find in the cell is exactly Gaussian with a half-width corresponding to about ten million mw. We have yet to calculate out the exact mw. We have tried to make sure that conditions theoretically necessary for a Gaussian distribution of each species are in fact fulfilled but there are still a couple of con-

W ORKING

AT

H IGH S PEED

253

trols we must do. However, I’m fairly sure they will not turn up the big piece. This developing result makes us wonder where the big piece is. Perhaps this phage has none. Do you know? Or perhaps the big piece is out of our view at the top or bottom of the cell. We are checking this. We use whole shockate so at least we’ve not thrown them down the sink. Meselson suggested that the big pieces might be aggregates that exist inside the phage but come apart in the cesium chloride experiments or might be artifacts that aggregate during Levinthal’s nuclear emulsion experiments. He thought that some of these questions could be laid to rest by experiments on the effect of CsCl on the “stars” produced in Levinthal’s experiments (see above, p. 103). In case someone in Levinthal’s laboratory were inclined to do so, Meselson summarized the procedures he and Stahl used to prepare DNA shockates and purified CsCl solutions.96 During this time Meselson was also on the lookout for opportunities to exploit the capacity of the method to separate similar molecules whose densities might differ by small amounts. On 8 March he tried an experiment with DNA from a leukemia patient, to see if persons with cancer might have DNA of density different from that of normal people. After twenty hours he got a rather fat band, but the result could not be evaluated until he ran a comparative test with normal human DNA.97 Another such endeavor, which Meselson resumed on the evening of the same day, was the effort to separate sweet and sour T2 phage DNA. Beginning with sour shockate, he boldly accelerated the centrifuge after about two hours to 57,000 RPM. When the films he took the next morning showed that half the contents of the cell had leaked out, he throttled back to his customary 44,770 RPM and continued the run until 7 P.M. on 10 March. A faint but well-defined band formed in the solution still left in the cell. An hour and a half later, Meselson began another run with T2 shockate, but this time in the presence of 19 mg H 3 BO 3. Evidently he was beginning to implement his plan to form borate complexes to separate sweet from sour DNA, but he began with the control—the complexes were expected to form with sweet DNA. No bands appeared, but when he repeated the run with the same cell, beginning on the evening of 11 March, he obtained a sharp band.98 Meselson performed no more centrifuge runs for the next ten days.

254

W ORKING

AT

H IGH S PEED

The machine was operated nineteen times during that period. Whether he had again exhausted his quota until other users had the chance to catch up on their needs, or whether he was away for some other reason, the promise of the new density gradient method and the pressure it placed on the use of the analytical ultracentrifuge at Caltech had by this time reached such proportions that the chairman of the Division of Biology, George Beadle, decided the situation warranted the immediate purchase of a second machine. In a letter to Ralph Meader, at the National Institutes of Health, Beadle asked on 12 March for approval to use funds expected from a pending grant application for that purpose. “Subsequent to the time the application was made,” he explained, another rather urgent need has arisen. Dr. Vinograd and associates have developed a new technic of analysis in which an equilibrium density gradient of cesium chloride is employed for separation of molecules of different densities. Each species seeks its property [sic] density level in the centrifuge cell. In addition the width and sharpness of the band in which the molecules of a given density are concentrated is a function of molecular weight since greater diffusion of smaller molecules makes the band wider. Advantages of this technic now appear to be very great. All possible effort is being made to push its development and application as far as possible. The density gradient method depends on the establishment of equilibrium conditions and this means long runs. Currently our one analytical ultracentrifuge is being used almost full time—24 hours a day—on the new work. As a result another machine is right now needed for conventional analysis. The Division of Chemistry is, I understand, proposing to make application for such a machine. Even this second ultracentrifuge would not suffice for long, Beadle added, because when Robert Sinsheimer joined the Biology Division on 1 July, “still another machine would be needed for his group.” It was to procure that third machine that Beadle proposed to expend $20,000 from the grant that he hoped NIH would soon approve.99 In his letter to Levinthal on the thirteenth, Meselson connected his latest experiments on sweet and sour DNA with his questions about the existence and whereabouts of Levinthal’s DNA pieces: “We are trying to separate sweet from sour DNA in the centrifuge in hopes of getting a look at the separate (big and little?) pieces.” 100 When his next turn on the centrifuge came around, however, his priorities had again

W ORKING

AT

H IGH S PEED

255

been altered, this time by the arrival of the salmon DNA he had requested from Paul Doty. The opportunity to test material whose molecular weight had been established by another method was particularly welcome to Meselson and Stahl just at that time. Their new method was stirring such interest at Caltech that they felt pressure to publish sooner than they wished. As Meselson, put it later, they wanted “to be sure the bridge won’t collapse when we put our weight on it.” 101 Meselson would have liked to subject the salmon sperm DNA to the same rigorous test for equilibrium that he had applied to T4 DNA, but he was unable to reserve the seven or eight days on the machine that that experiment had required. The most he could bargain for was about four days. That amount of time enabled him to centrifuge the salmon sperm DNA for substantially longer than had been found adequate to bring phage DNA to equilibrium at the same speed and temperature. Even then the heavy demand for the machine caused him to have to begin the run at 2:45 A.M. on Saturday, 23 March. The density of the CsCl solution was 1.70, and he set the speed at the same speed—27,690 RPM—that he had used for the previous long equilibrium runs. The solution filled only two-thirds of the cell, but did not appear to leak further during the run. During most of the next day Meselson’s attempts to take pictures were stymied by jamming of the film transport mechanism. When he finally got a successful series, they showed a well-defined band that was conspicuously broader than the ones obtained with T4 DNA. That difference would immediately have signified that the molecular weight was lower than that of the phage DNA, a favorable outcome. Doty’s light-scattering measurement of the molecular weight for the salmon sperm DNA was about 8 million, whereas their own current estimate for the T4 DNA was between 10 and 13 million. Meselson continued the run until just after noon on Wednesday, 27 March, a total of 106 hours.102 Two days later Meselson ran some of the same DNA from a leukemia patient that he had first tested three weeks earlier, this time together with “normal” human DNA. The result was a band with a “halo” on the heavy side, which might have indicated some density heterogeneity. There was, however, no clear-cut separation such as one might expect between distinct molecular species. He pursued the question no further.103 After this unsatisfying run, Meselson had to give up experimentation for several weeks to write his thesis. Stahl was also diverted, al-

256

W ORKING

AT

H IGH S PEED

though less completely. He had been co-opted to edit a chapter for a textbook on bacteriophages. The book was to be based on a manuscript that Mark Adams, who had for more than a decade taught the annual phage course at Cold Spring Harbor, left unfinished at his untimely death in October 1956. The phage group had formed a committee, chaired by Al Hershey, who asked for volunteers to finish the task. They were motivated to leave a monument to their departed colleague and to fill the need for a broad introductory survey of their emerging scientific discipline.104 Those to whom Meselson and Stahl relayed by letter the news of their density gradient determinations of the molecular weight of phage DNA did not immediately accept, or even understand, their capsule reports of the method and its first fruits. In response to Stahl’s letter of 6 March, Bob Edgar queried why the Gaussian curve showed that the DNA molecules possessed a uniform molecular weight. On 18 March (in a letter dealing mainly with Edgar’s forthcoming fellowship at Caltech), Stahl answered, “When you arrive, I can explain why normal distribution of DNA in CsCl gradient means all pieces same size (or it could mean a normal distribution of densities). But, for all same density, the DNA should be distributed as the sum of gaussians all with the same origin [if there were several species with different molecular weights]. Our band is a perfect Gaussian, and not the sum of two or more Gaussians with same origin.” 105 There is an undercurrent here of concern that the assumption on which the method was built— that all the DNA in the band was of uniform density—might not hold up. In view of their “perfect” early result, however, it was sensible to believe that the assumption was reasonable.106 Cyrus Levinthal strongly resisted the conclusion that Meselson sent him on 18 March that DNA from T4 shockate banded in the analytical Spinco seemed to contain only one molecular species rather than big and little pieces. He responded almost immediately that the DNA might not have reached equilibrium in their experiments, and that something must be “deeply wrong” with them.107 Meselson believed that the experiment in which he had obtained the same band when he started from an even distribution of material as when he went “from an extremely tight band produced by running for a while at very high speed” constituted “excellent experimental verification of equilibrium.” 108 Nevertheless, Levinthal’s objections caused Meselson to strain to find a mathematical proof that equilibrium could be attained within the time limits of his centrifuge runs.

W ORKING

AT

H IGH S PEED

257

He required an equation that would describe the time necessary for the distribution of the macromolecules to approach closely its equilibrium value. For help in thinking the problem through, he turned to Girair (Jerry) Nazarian, a doctoral student in the Department of Chemistry who was completing his dissertation on the theory of the transient state in the ultracentrifuge. Nazarian led him to a paper published in 1924 by Max Mason and Warren Weaver that had set up differential equations to describe “the settling of small particles in a fluid” under the normal force of gravity. The Mason-Weaver solution had been adopted also for describing the motions of particles in the ultracentrifuge. With “the very great help” of Nazarian, Meselson followed the general form of the Mason-Weaver treatment “except that the density ρ is assumed to be a linear function of the distance x.” Beginning with equations of “flow” of DNA: 1)

dJ dC dc ⫽⫺ , and 2) J ⫽ ⫺D ⫹ CV dx dt dx

where C ⫽ Conc. of DNA, J ⫽ flow of DNA, and D ⫽ diffusion constant, he derived, through steps too long to summarize here, a cumbersome equation for the time course of the concentration distribution containing Hermite polynomials, but dependent, for a given column length, only on D and on a term representing the reciprocal of the final band width. Defining “equilibrium” as the time when the concentration at the center of the band (x ⫽ 0) is within 1 percent of its final value, he calculated that for T4 DNA “t ⫽ 54 hrs!” That result, well within the duration of their longest equilibrium runs, clearly seemed to Meselson to vindicate his confidence in the method. On 1 April he wrote Levinthal, “Believe me, the bands are at equilibrium!” 109 Along with his derivation of the DNA distribution in the cell as a function of time, Meselson sent Levinthal “some qualitative remarks designed to wear down your intuitive feeling that equilibrium should come much more slowly than we find”: The production of a band from an initially even distribution in a density gradient in a centrifugal field is a process of sedimentation. It is certainly opposed by the frictional coefficient hence it is natural to feel that a low diffusion coefficient will mean a slow band formation. But a high molecular weight goes with the low diffusion coefficient of DNA and this acts to speed the sedimentation. The reason I was surprised at the small time requirement was not that

258

W ORKING

AT

H IGH S PEED

the diffusion coeff. is low, but that as the molecules approach the center of the band where they have zero effective mass, the driving force of the centrifugal field falls off to zero. In this region even sedimentation might be hopelessly slow. Forming a band from an initially tighter one is another proposition. There, diffusion is the transport process and I’m still surprised (intuitively, no longer theoretically!) how fast that goes.110 By now, Meselson was less inclined than he had been two weeks earlier to admit that Levinthal’s big piece might be out of sight at the top or the bottom of the cell. All of the forms of DNA they had so far tested, including now Doty’s salmon sperm DNA, had given bands in exactly the same position, “so I’d be surprised to find some at the top or bottom of the cell. But it’s possible.” Explaining to Levinthal once more that the fact that the T4 band was Gaussian to within their best measurements indicated that it consists of fairly homogeneously sized particles, very different from what a combination “of your big and little pieces would look” like, Meselson offered several possible ways to reconcile their results with Levinthal’s star size distribution. He dismissed as very doubtful to only possible the idea that CsCl makes all the pieces into pieces of uniform weight, or that it makes either the big or little pieces into uniform ones while the others remain out of view. The only suggestion that he presented more positively was that Levinthal’s big pieces were the same size as the little ones but hotter— that is, that some fractions of DNA had taken up the radioactive label more than others did. “We don’t know of any evidence showing that the various molecules in a phage are evenly labelled when there is both hot inorganic and cold organic phosphorus in the medium.” This idea was undoubtedly not attractive to Levinthal, because it would explain his results while depriving them of the significance he attributed to them. A controversy had arisen whose resolution could not leave both Levinthal’s and Meselson and Stahl’s positions intact. At the time Meselson wrote Levinthal, he and Stahl were still refining the estimates of the molecular weights based on their completed centrifuge runs. The expression for the molecular weight required a figure for the partial specific volume, a quantity known in NaCl solutions but not in CsCl solutions. “We are just getting around to measuring the exact psv of DNA in CsCl solutions,” he reported, “and until we get it our M.W. value isn’t guaranteed.” He had reasons to believe that the result would not be much different from the value for NaCl, but wanted to “wait and see. I’m willing to guarantee our MW value

W ORKING

AT

H IGH S PEED

259

right now to ⫾30 percent and that’s going to drop to 10 percent and maybe even lower depending on the p.s.v. measurement accuracy. THE VALUE IS NOW 19 MILLION.” 111 One week later Stahl reported to Edgar that “re-calculations of the phage DNA data gives us the estimate 20 million, with all molecules the same size,” an outcome that was not entirely satisfying to him— it was, in fact, “an awkward conclusion, indeed.” 112 The reason for his discomfort was that comparisons of this result with estimates of the total quantity of DNA contained in a single bacteriophage led to the calculation that there were about twelve molecules per virus particle. From Stahl’s perspective this was a paradoxical situation, because the genetic recombination studies of Doermann, which had also underlain his own thesis research, indicated that there were only three linkage groups. The natural inference was that the genetic material should be divided into no more than three pieces of DNA. The molecular weight data were, therefore, in conflict with the genetic data.113 Between 23 March and his letter of 7 April, Stahl probably spent much of his time on the analysis of the data from the 106-hour centrifuge run with Doty’s salmon sperm DNA. For this task he abandoned the tedious method of direct reading of the galvanometer on the Sinclair Smith photometer and connected it to a continuously recording Brown potentiometer. Comparing the two methods, he found them to give identical results. He also checked the resolution of the system by taking a photograph of a fine wire in a resting centrifuge cell. The image projected onto the slit stage of the potentiometer was sharply defined. When he used the new system to plot out the density curve for the salmon sperm, however, he encountered another discomforting surprise. The curve fit a Gaussian distribution only on the light side of the mean. On the heavy side it was skewed (figure 7.15).114 That this result indicated that there must be some density heterogeneity did not deter Meselson and Stahl from using it to make a calculation of the molecular weight. They measured σ as the standard deviation of the Gaussian curve that could be fitted to the low density side of the observed distribution curve. On 7 April Stahl announced to Edgar, “We have determined the MW of salmon sperm DNA supplied by Paul Doty. We get 8 ⫻ 106. This is elegant agreement with his estimate by light scattering.” Meselson telephoned Doty to give him the good news.115 Their satisfaction was ephemeral. Several days later Meselson found an error of a factor of two in the calculations and had to revise

260

W ORKING

AT

H IGH S PEED

Fig. 7.15. Density plot for observed distribution of salmon sperm DNA, superimposed on the calculated Gaussian distribution, as reproduced in Meselson, “Equilibrium Distribution,” p. 36

the molecular weight estimate to four million. He did not hurry to convey this setback to Doty. The revised result did not put their new molecular weight method into direct conflict with the results of an existing method, because they could easily fall back on their assumption that when density heterogeneities are present, only a minimal estimate of the molecular weight is possible. Moreover, because lighter molecules diffuse more rapidly and sediment less rapidly, it was possible that the 106 hours that would have been more than adequate for T4 DNA to reach equilibrium may not have been sufficient for the smaller salmon sperm molecules.116 “Bigger DNA molecules,” he had just written to Levinthal, “band up faster than small ones.” 117 Meselson and Stahl continued to believe that their density gradient method could be developed into a viable means to determine molecular weights for DNA and other macromolecules, but the luster that had so quickly surrounded it at Caltech was already beginning to fade.

VI Writing up his thesis probably occupied most of Meselson’s time through April and the first half of May. It was composed of two entirely unrelated halves. Part 2, “The Crystal Structure of N,N′Dimethyl Malonamide,” represented research long since completed. In a summary of previous and proposed research,‘‘ written probably while he was working on his dissertation, Meselson summarized the outcome succinctly: ‘’The crystal structure of N,N′ dimethyl malonamide has been determined by x-ray diffraction. The bond lengths and angles conform to expectations based on studies of similar molecules.

W ORKING

AT

H IGH S PEED

261

Fig. 7.16. Schematic drawing of N,N′-dimethylmalonamide, showing bond lengths and angles, from Meselson, “Crystal Structure,” Ph.D. diss., pt. 2

A method of parameter refinement was devised and successfully used to improve resolution in two-dimensional projections.‘‘118 Schematically he represented the molecule in a diagram typical of the Pauling school (figure 7.16).119 Part 1, “Equilibrium Sedimentation of Macromolecules in Density Gradients with Application to the Study of Deoxyribonucleic Acid,” was, in contrast, an effort to present systematically the early results of a project in which he was still deeply engaged. Astutely presuming that the chemists to whom he would have to present his thesis would be uninterested in his biological objectives, and having no definitive biological results to display anyway, Meselson emphasized what he thought would seem most appropriate to them in a chemistry thesis.120 Density gradient centrifugation appeared as a method generally applicable to macromolecules. The results that had been obtained for bacteriophage DNA, calf thymus DNA, and salmon sperm DNA were presented as illustrations of the method, with no discussion of the purposes for which Meselson had chosen to experiment on just these substances. Knowing the importance of equations in the chemistry department, he formalized his derivations of those he had devised for

262

W ORKING

AT

H IGH S PEED

the equilibrium distribution of the macromolecules, and the time course of their approach to equilibrium, and devoted half of his space to their presentation. As mentioned earlier, The Svedberg had derived the same equation for the conditions of equilibrium in the ultracentrifuge cell from the relation between sedimentation and diffusion and from thermodynamics.121 According to Schachman, the thermodynamic approach yields, “with rigor and scarcely any assumptions,” the identical result that the kinetic approach produces “with approximations and some equivocation in terms.” 122 Working from the thermodynamic foundation that the total potential of any component at equilibrium in a closed system at constant temperature must be uniform throughout the system, Meselson applied, for the case of a centrifugal field, the equation: M i (1 ⫺ v¯ i ρ)ω 2 r dr ⫺

∂µ i

冱 ∂C K

dC k ⫽ 0

k

This equation, whose two terms represent the familiar sedimentation and diffusion terms of the Svedberg equation, Meselson arrived at by summation of terms representing a sequence of infinitesimal components along the centrifugal field. The value µ represents the partial molal Gibbs free energy, and the subscripts represent the kth and ith components of the sequence. The density of the solution at distance r from the center of rotation is represented by ρ.123 Meselson’s equation was very similar to one published by Richard Goldberg in 1953 that was, according to Schachman, the first “rigorous description of a system at sedimentation equilibrium in terms of a continuous sequence of phases of fixed volume and infinitesimal depth in the direction of the centrifugal field.” The resemblance between the Goldberg equation and the above led Schachman to infer in 1959 that Meselson “used [the Goldberg equation] with the modification that ρ itself is a function of x due to the concentration gradient in the cesium chloride solution.” 124 Knowing nothing about the Goldberg equation, however, Meselson had actually derived his equation straight from “the unadorned Svedberg equation” by making the density a function of r.125 The incident only illustrates that scientists, as well as historians, sometimes construct logical connections between scientific contributions that do not coincide with the personal experiences underlying them. Meselson first used the Svedberg equation to calculate the distribution of CsCl at equilibrium, ignoring the high-molecular weight com-

W ORKING

AT

H IGH S PEED

263

ponent. Then, simplifying the problem by assuming a linear gradient of the salt concentration over the distance of a polymer band, he used his modified Svedberg equation to find the distribution of the latter in the former. The expression at which he arrived for the concentration of the polymer along the gradient (C PXn (r)) was:





(r ⫺ r 0 ) 2 2σ 2

C PXn (r) ⫽ C PXn (r 0 ) exp ⫺

This equation was in the form of a Gaussian normal distribution curve (see above, p. 233), where σ is defined as the standard deviation. The expression Meselson derived for σ that permitted him to put the equation in this form was: RT

σ2 ⫽ M PXn v¯ PXn

冢 冣 dρ dr

ω2r0

r0

For the equations describing the time course of the approach to equilibrium, Meselson simplified the derivation he had worked out with Nazarian. Beginning with an equation of continuity relating the diffusion constant and the sedimentation velocity to the change in concentration with time, expressing the sedimentation velocity by means of Svedberg’s equation, and introducing an expression for the concentration gradient, ρ⫽

1 dρ ⫹ x, v¯ dx

Meselson combined them to yield a time-dependent differential equation:





1 ∂C ∂ ∂C xC ⫹ 2 ⫽ ∂x ∂x σ D ∂t Solving the equation with the use of Hermite polynomials, setting certain simplifying conditions, setting t* as the time required to reach, at a distance from the center ⫽ 2σ, a concentration within 1 percentage point of equilibrium, and retaining only the most slowly decaying term in the equation, he solved for t*: t* ⫽





σ2 L ln ⫹ 1.26 D σ

264

W ORKING

AT

H IGH S PEED

where D ⫽ the diffusion constant and L ⫽ the length of the fluid column and is much longer than σ.126 By mid-May Meselson had probably finished his thesis, and he had a few days left over before his Ph.D. examination, scheduled for 23 May. He used the time to try another long equilibrium run on Doty’s salmon sperm DNA. Beginning at 6 P.M. on 17 May, he used the same quantity of DNA in CsCl of slightly higher density, but ran the centrifuge at a higher speed than in the first test (37,020 instead of 27,690 RPM). The run lasted only a little longer (110 hours instead of 106), but at the higher speed he could expect to achieve equilibrium more quickly. The run ended on the afternoon of 22 May. There probably was not time for Stahl to begin the photometer analysis of the result prior to Meselson’s examination, but visual inspection would have shown immediately that the band was very similar to that obtained in the first run.127 Following a well-established custom in the Chemistry Division, Meselson was required to prepare eleven propositions for problem solutions or potential experimental investigations, around which his examination would center. The candidate was not expected to state in detail how to carry out the investigations or solve the problems. The examining committee would draw out, through their questions to him, what the candidate had in mind. Three of the propositions Meselson presented related to the crystallographic portion of the thesis and one to the project on which he had worked for a time for Pauling on substitution defects in crystals. One proposition was customarily humorous. The other six were connected to various aspects of the joint investigative program Meselson was pursuing with Stahl. Of these, the first referred to the theory of mutagenesis that he had worked out during the fall: 1. A specific molecular mechanism is proposed for the powerful mutagenic action of 5-bromouracil on bacteriophage. Some critical experiments are suggested. Another derived from his recent confrontation with Levinthal’s big and little pieces: 7. Levinthal, on the basis of auto-radiography of DNA containing radioactive phosphorus, has concluded that DNA prepared from bacteriophage by osmotic shock consists of at least two very different molecular weight species. Another explanation is proposed for

W ORKING

AT

H IGH S PEED

265

Levinthal’s findings which does not require molecular heterogeneity of the DNA. A simple experiment is proposed which should distinguish between the two explanations. The experiment Meselson proposed might have been based on a question he had asked Levinthal himself: “Do you ever look at shockates from phage grown in the absence of organic phosphate?” The purpose of doing so would be to test whether in the experiments in which Levinthal grew the phage in a medium containing radioactive inorganic and nonradioactive organic phosphorus “the various molecules in a phage are evenly labelled.” The eighth proposition was an inference drawn from the exact Gaussian curve obtained from the banding of T4 DNA: 8. The apparent monodispersity of DNA from bacteriophage T4 suggests that the molecules of this material are in some sense also discrete in vivo. This speculation is made more plausible by the molecular weight value itself which corresponds to about 12 molecules of DNA per phage particle, a number consistent with the apparent symmetry of the virus. A method is proposed for directly determining the size of the heritable sub-units. The point of special interest in this idea is that it suggests that Meselson and Stahl reacted differently to the outcome of their measurement of the molecular weight of T4 DNA. Whereas Stahl worried that the evidence for twelve molecules of DNA was inconsistent with the genetic evidence for no more than three, Meselson perceived in the twelve molecules an opportunity to investigate the molecular architecture of the virus. These attitudes were consistent with their training, respectively, in phage genetics and in structural chemistry. The tenth proposition was a very succinct statement of Meselson’s longcherished goal: 10. It is proposed that the technique of density gradient centrifugation of DNA lends itself to experiments able to test critically some present models of the mechanism of DNA replication.128 Meselson entered the Crellin Conference Room for his examination at 1:30 P.M. on 23 May. Linus Pauling chaired his committee, which also included Jerry Vinograd, the physicist Richard Feynman, Harden McConnell, the physical chemist whom Meselson had consulted about magnetic resonance methods, and Dan Campbell, an immunologist who had provided some questionable evidence in support of Pau-

266

W ORKING

AT

H IGH S PEED

ling’s antibody theory. As the questioning began, Feynman, who had not previously read Meselson’s thesis, sat in a corner of the room and swiftly scanned it. Then he told the candidate that there was a much simpler way to calculate the time course of the distribution of macromolecules in the density gradient. The first and second moments of the distribution, which could be calculated directly from the differential equation of flow for the system, were simple in form. Stepping to the blackboard, Feynman quickly showed Meselson how to make the calculations. This impressive performance was not due to Feynman’s brilliance alone but also to his familiarity with the mathematically similar form of the well-known quantum mechanical equations for a particle in a harmonic potential.129 The examination itself went smoothly. Afterward, when the rest of the committee had left the room, Pauling stayed behind and told Meselson that he was “fortunate to be entering a field on the verge of so many advances.” 130 Whatever regrets Pauling may have felt that Meselson would not continue to work on molecular structures was more than compensated by his admiration for the “powerful method of studying macromolecules” that his student had invented and for the “thorough way, both theoretically and experimentally,” in which he had studied this new technique.131 The next day Meselson probably had a party to celebrate his twenty-seventh birthday. Soon afterward he extracted the first twenty pages from part 1 of his thesis, made a few stylistic alterations, and turned it into the paper that he and Stahl had been urged to prepare on the density gradient method. The title was also identical to the first portion of the title for that part of his thesis: “Equilibrium Sedimentation of Macromolecules in Density Gradients,” but there were now three co-authors listed—Meselson, Stahl, and Vinograd. On 27 May Pauling communicated the paper to PNAS.132

VII While Meselson worked on his thesis in April, Jerry Vinograd conducted several density gradient runs with microsomes and with RNA from baker’s yeast. Meselson assisted Vinograd in two runs with hemoglobin, noting on the centrifuge log that it “looks like big pieces Hb band first, then small ones at more nearly the expected rate.” 133 At the end of part 1 of his thesis, he listed briefly under “additional applications” some of the materials to which Vinograd had applied

W ORKING

AT

H IGH S PEED

267

the method. In his acknowledgments, Meselson stated that “Dr. Franklin Stahl has been a close companion as well as a research partner. Part of this work has been done in collaboration with him and with Dr. Jerome Vinograd, to whom I am grateful for his encouragement and friendship.” 134 Sometime during the spring of 1957, Vinograd wrote a three-page memorandum titled “Applications of Stable Density Gradients in Virus and High Polymer Chemistry.” In the introductory paragraph he discussed gradients that arise in the ultracentrifuge and referred to a recently published paper by Nazarian, Raphael Pasternak, and himself that had shown mathematically that the gradients “are reached in much shorter periods of time than had previously been supposed” (a claim which suggests that Vinograd was unfamiliar with the paper by Kai Pederson from which Meselson had learned about the rapidity with which the gradients are established). He referred to the same paper as source for the example that “a strong CsCl solution is essentially at equilibrium in eight hours” (even though there is no mention of CsCl in the paper cited).135 The purpose of this note, Vinograd wrote, was to report briefly on three applications of such gradient systems. The first was “the determination of number average and weight average molecular weights and the density of dissolved macromolecules homogeneous in density.” After giving a synoptic derivation of the equation for calculating such molecular weights, Vinograd mentioned that “this method is being used to determine the molecular weight of DNA from T-4 bacteriophage” and presented a density curve whose symmetry showed that the DNA was homogeneous in density. To illustrate the second application, “the detection of heterogeneity in density of dissolved macromolecules,” he presented a curve for calf thymus DNA, whose skewed band showed that the material was not homogeneous. Finally, to show how the method could separate two or more dissolved materials of different densities, he showed a photometer tracing of normal T-4 DNA and 5-bromouracil DNA bands in a single cell. Nowhere in his text or in his list of three references did Vinograd mention Meselson or Stahl.136 Although this note is in the form of a manuscript to be submitted for publication, there is no evidence that Vinograd did so, or even that he ever showed it to anyone else. It has been preserved in the Vinograd collection of the Caltech Archive, where I saw it in 1991. When I showed it to Meselson in May 1992, he thought at first that the reference to the Nazarian, Pasternak, and Vinograd paper was a mistake

268

W ORKING

AT

H IGH S PEED

and that the intended reference was the Meselson, Stahl, and Vinograd paper. When he realized that it was not, he remarked that he couldn’t imagine that “it would be about density gradient centrifugation without me and Frank having something to do with it.” As he examined the manuscript further, he saw there was no reference anywhere to him or Stahl, but the experiments described were all his. We superimposed one of the curves, that for the separation of 5-BU and ordinary phage DNA, on the densitometer tracing in the centrifuge log and found them to coincide exactly. The other curves were evidently drawn independently, but all were derived from the data of the experiments carried out by Meselson in January and February 1957.137 For Meselson the realization of the existence of this document, thirty-five years after the fact, was a painful moment of recognition that the relation between Vinograd, Stahl, and himself had been quite different from what he had imagined it to be. He took it as evidence that an association that had seemed to give him and Frank much pleasure had simultaneously “given great pain to someone else.” It is painful also to discuss the situation here, but unavoidable, because the document that reveals it is part of an archival record available to all interested scholars. Why did Vinograd write it? Meselson speculated that Vinograd, who had spent much time assisting a graduate student who was not his own student, was struggling to find his intellectual place at Caltech, and that he was presenting to himself the development of the new method that had been discovered on his machine as he might have wished it to have happened.138 Stahl, to whom I had shown the manuscript three weeks earlier, commented more bluntly that it seemed like an appropriation of what really belonged to Meselson.139 Vinograd is not alive to tell his side of the story, but it is not difficult to imagine how he could come to imagine that he really was, as he later represented himself, the “co-discoverer, with Meselson and Stahl, of the method of separating and characterizing viruses and macromolecules.” 140 He had lent Meselson not only his machine but his expertise in the theory and practice of the analytical ultracentrifuge. He was often present for parts of Meselson’s centrifuge runs and sometimes took pictures or otherwise participated in the process. The discovery of the rapid formation of a cesium chloride density gradient probably took place while he was around, because it was Vinograd who repeatedly told the story of Meselson’s astonishment when the Schlieren line continued to move upward.141 Senior to Meselson and

W ORKING

AT

H IGH S PEED

269

Stahl, as well as custodian of the laboratory in which Meselson performed the experiments, Vinograd was easily seen at Caltech—as the letter of Beadle to the NIH indicates—as the leader of a group developing the new density gradient method. His interest in the project extended to exploring its wider application himself. His inclusion as a co-author of the paper describing the method in print validated the claim that he had made a contribution. Is it possible that Meselson and Stahl did not notice, or did not remember, the extent of his role? In his famous Traite´ e´le´mentaire de chimie, Antoine Lavoisier, one of the first great collaborators in the history of science, wrote that the custom he and his associates had formed “to live together, to communicate our ideas, our observations, and our points of view, has established among us a sort of community of opinions in which it is difficult, even for us, to distinguish what belongs to each of us individually.” 142 Scientific investigation has, ever since, thrived on such communal activity. The difficulty in identifying the originators of the ideas that arise in group research comes easily into conflict, however, with the scientific norm that individual scientists must be recognized appropriately for their original work.143 There is no way to recover the day- by-day conversation in Vinograd’s laboratory in which he may have made some suggestions that helped Meselson to perform the experiments and to formulate the theories that underlay the invention of CsCl density gradient centrifugation. The written record, however, provides unequivocal evidence that Meselson performed the fundamental experiments and produced, with help from Jerry Nazarian, the theoretical analysis on which the new method was based.

VIII Through the heady spring of 1957, Meselson and Stahl held fast to the conviction with which they had entered it: that the development of the density gradient method itself was only a detour along the path to their primary goals. Meselson made this point clearly in a summary he prepared, probably about the last week in March, of his previous and proposed research. After a brief summary of his completed crystallography work (quoted above), he spent three pages recapitulating the theory and experiments underlying the new method. Then he discussed the way in which 5-BU substituted DNA could be used to follow the fate of parental DNA through successive replications of virus

270

W ORKING

AT

H IGH S PEED

to check the predictions of the several replication schemes that had been proposed. Describing the preliminary experiment of this type already performed (see above, pp. 248–249), he continued, This leads to the experiments for which the density gradient centrifugation method was originally designed and which I hope to carry out during the coming year. My program includes the refinement and extension of the above experiments bearing on the replication mechanism of DNA. In addition to studying the DNA of bacteriophage, I plan to use similar methods to investigate bacterial DNA which can also be grown with 5-bromouracil. Bacterial systems have the advantage that divisions may be synchronized and stopped after any number of cycles. This makes possible the isolation of DNA from cells which have undergone a known number of divisions. The Watson Crick expectation, for example, is that the first generation DNA from initially heavy parents grown in the absence of 5-bromouracil will be 50 per cent heavy and that half of the second generation molecules will be 50 per cent heavy and the other half will be completely light.144 Here Meselson placed on paper for the first time the schematic plan that he had probably outlined verbally during a centrifuge run in January. The advantage he foresaw for carrying out the experiments on bacteria appears here so compelling that one must ascribe to the continuing power of the phage group ethos the priority he still gave to further experiments on bacteriophage. With the same irrepressible enthusiasm that had marked the earlier planning of his collaboration with Stahl, Meselson outlined a series of further steps he would take if these experiments were to turn out as he expected. If the 50 percent progeny continued to appear, he would try to learn how the 5-bromouracil was distributed in the molecule. “The Watson Crick scheme demands that the heavy analogue be confined to only one of the two chains but other distributions might be found.” In order to study these distributions, they might treat the progeny molecules with agents that cut them transversely and examine the density distribution of the fragments.145 Beyond all this, Meselson envisioned experiments separating whole virus particles by density differences. If the same bacteria were infected by virus particles differing both in genetic constitution and in density, it might be possible to determine, through examination of the progeny, “whether the portions of DNA which remain intact during replication carry genetic information.” 146

W ORKING

AT

H IGH S PEED

271

The expansiveness of the program Meselson set out here is a mark of the enthusiasm that the successes of the density gradient method so far had kindled in his expectations for it as a tool to investigate biological questions. What had happened to his earlier enthusiasm for the theoretical mechanism of 5-BU mutagenesis that he had hoped, with Stahl, to test through genetic experiments? Although that plan went unmentioned here, there is evidence that it remained in his mind, beyond the transfer experiment, the ultimate goal of their collaboration. In his letter to Levinthal on 13 March, at the height of his preoccupation with the general density gradient method, he had written, “We are hoping to get back to the transfer experiments very soon and then to our biggest interest in the 5BU business, the mutations.” 147 Stahl yearned to get on with the mutagenesis side of their venture as soon as possible but understood that it was likely to be deferred so long as the density gradient method and the transfer experiments dominated Meselson’s attention. Wistfully he suggested to Edgar in April that it might be possible during the summer to persuade one or two students to “initiate the mutation analysis,” while he and Meselson concentrated on the transfer experiments.148

C HAPTER E IGHT

The Unseen Band

I As the new opportunities created by the density gradient method deferred Meselson and Stahl’s attack on the DNA replication problem at the beginning of 1957, the problem was not left waiting for them to get around to it. Contrary to the prediction by Delbru¨ck and Stent that further elucidation would emerge from work on bacteriophage, the next significant advance came from experiments on bean plant seedlings. Several investigators had attempted, during the early 1950s, to introduce radioactive isotopes into dividing cells to study the manner in which chromosomes are organized and synthesized. Alma Howard and S. R. Pelc, for instance, incorporated 32 P into dividing cells in the meristem of the English broad bean plant Vicia faba in 1950. They showed by autoradiographs that the phosphorus was concentrated in the nucleus during the resting stage and inherited by the daughter cells in successive divisions.1 Their resolution was not sufficient to trace the distribution of the label in the chromosomes during their replication. At Oak Ridge National Laboratory in Tennessee, J. Herbert Taylor had already begun to study chromosome replication when he moved in 1950 from the University of Tennessee to Columbia University. He continued his investigation at the Brookhaven National Laboratory as a research collaborator, using 32 P to label nucleic acids. In 1955 he read some papers on the use of tritiated water to label cells and had the idea that tritium-labeled thymidine (the nucleoside composed of thymine attached to the deoxyribose sugar) would be more suitable for his purposes. Because tritium decays by emitting beta particles with relatively short paths, he thought that, after applying radioactive thymidine for less than one cell generation, he could obtain autoradio-

T HE U NSEEN B AND

273

graphs with sufficient resolution to locate the radioactive thymidine within the chromosomes of subsequent generations. Influenced by the Watson-Crick model for the replication of DNA, Taylor expected to find that one of the daughter chromosomes would be labeled “to the exclusion of the other.” When Max Delbru¨ck’s “On the Replication of Deoxyribonucleic Acid (DNA)” appeared, Taylor found Delbru¨ck’s ideas so attractive that for several months he “postponed further thinking” about his contemplated experiments. A more serious cause of delay was that he had no facilities for handling large amounts of tritium. In June 1956, he made arrangements with Walter L. Hughes to share some of the tritiated thymidine that Hughes planned to synthesize in the Medical Department at Brookhaven for use against cancer cells. Taylor grew seedlings of Vicia faba—the same plant that Howard and Pelc had used—in a mineral medium containing the radioactive thymidine. He carried out the work as a summer project in the laboratory of Phillip S. Woods.2 After growing the seedlings in the radioactive medium for a length of time corresponding to about one-third of a cell division cycle, Taylor planned to wash them in a nonradioactive medium containing colchicine. This alkaloid was known to suppress the anaphase of mitosis, that is, the stage in which the daughter chromosomes move apart toward their eventual places in the nuclei of the daughter cells. Consequently, the chromosomes lined up along the equatorial plate at metaphase, duplicated, and separated, but remained in the same nucleus. After a time they duplicated a second time. The number of duplications that had taken place could therefore be determined by counting the number of daughter chromosomes, or “sister chromatids,” as they were called, contained in each cell.3 The experiments carried out with the first sample of thymidine failed because it was not radioactive enough to label the chromosomes. With a second sample prepared by Hughes, Taylor obtained convincing results. Autographs of chromosomes that had duplicated no more than once showed an even distribution of labeling. In cells whose chromosomes had duplicated twice, “all chromosomes were labeled; however, only one of the two sister chromatids of each was radioactive.” In cells in which the chromosomes had duplicated three times, “approximately one-half of the chromosomes of a complement contained one labeled and one nonlabeled chromatid, while the remainder showed no label in either chromatid.” 4 These results Taylor interpreted as indicating “(1) that the thymi-

274

T HE U NSEEN B AND

dine built into the DNA of a chromosome is part of a physical entity that remains intact during succeeding replications. . . . (2) that a chromosome is composed of two such entities probably complementary to each other; and (3) that after replication of each to form a chromosome with four entities, the chromosome divides so that each chromatid (daughter chromosome) regularly receives an ‘original’ and a ‘new’ unit.” 5 Taylor presented his results at a meeting of the American Institute of Biological Sciences at the University of Connecticut in early September. News of the experiments spread rapidly, and his colleagues at Columbia pressed him to publish immediately. The paper “The Organization and Duplication of Chromosomes as Revealed by Autoradiographic Studies Using Tritium-Labeled Thymidine” by Taylor, Woods, and Hughes was communicated to PNAS in October.6 To make his conclusions about the replication of chromosomes clear in his paper, Taylor used a diagram to show how the “strands” in a chromosome separated (figure 8.1). “It is immediately apparent,” Taylor wrote, that “this pattern of replication is analogous to the replicating scheme proposed for DNA by Watson and Crick. We cannot be sure, of course, that separation of the two polynucleotide chains in the double helix is involved, for the chromosome is several orders of magnitude larger than the proposed double helix of DNA.” 7 Despite this reservation, Taylor, like others before him, offered a “provisional” model, in this case to explain the duplication of chromosomes at the “microscopic level.” He visualized the chromosome as composed of two complementary multistranded ribbons, with flexible materials in their outer edges giving them a tendency to coil. The stability of the larger units would, he asserted, facilitate their rapid separation as intact units. He left open the question of whether the “multiple, identical [cross-bonded] strands” he postulated as composing the

Fig. 8.1. Diagram by J. Herbert Taylor to illustrate replication of chromosomes

T HE U NSEEN B AND

275

Fig. 8.2. Diagram by Taylor illustrating his conception of separation of DNA double helices

larger units corresponded to Watson-Crick double helices. If the separation of these strands did involve “the separation of intertwined double helices of DNA, unwinding presents a problem, but perhaps not an impossible one.” He clarified his conceptions with another schematic diagram (figure 8.2).8 In October and November, Francis Crick, Max Delbru¨ck, and Alex Rich came to Taylor’s laboratory to view his results directly. In January 1957 he gave a seminar at Caltech at which George Beadle managed the slide projector and Linus Pauling sat near the front row. Taylor’s work captivated especially Delbru¨ck, who called a meeting the next morning to discuss an implication that he thought Taylor’s observations held for the polarity of the chains of the Watson-Crick double helix.9 Following up the discussion after Taylor’s departure, Delbru¨ck

276

T HE U NSEEN B AND

composed, on 1 February, “A Memo on the Sister-Strand Exchanges Observed by Taylor et al.” Summarizing the distribution of label in the “sister chromosomes” that they had found at the three successive metaphases, Delbru¨ck concluded that “the experiments show that the new DNA synthesized during I1 [the first interphase] consists of two elements which are segregated at M1 [the next metaphase], but which retain their integrity during subsequent mitoses. Since the new DNA consists of two elements, we conclude that also the old DNA, present at the beginning of I1, consists of two elements, similarly segregated at M1, and similarly retaining their integrity during subsequent mitoses.” 10 Delbru¨ck depicted this conclusion in a schematic diagram in which he represented DNA present at the first interphase by solid lines, labeled DNA synthesized during that interphase by dotted lines, and DNA synthesized during the second interphase by dashed lines (figure 8.3). In this statement and diagram, Delbru¨ck implicitly abandoned the dispersive replication model that he had put forth in 1954. What most interested him now was not that retreat forced on him by Taylor’s general result but an observation that Taylor had described in passing in his PNAS paper as a phenomenon secondary to the predominant pattern of replication. In cells in which the chromosomes had duplicated twice or more, Taylor noted, a “few chromatids were labeled along only a part of their length, but in these cases the sister chromatids were labeled in complementary portions.” This observation demonstrated that his method enabled him to observe crossing over between sister chromatids. No such exchanges were observed, however, in cells in which the chromosomes had duplicated only once. There were, in other words, no decisive cases of “half-chromatid exchange” following the first metaphase after the incorporation of the isotope.11 For Delbru¨ck these “sister exchanges” raised the question whether the two elements present at the beginning of the first mitotic interphase “are of such a nature that they can undergo a reciprocal exchange, or whether they have in some sense an opposite polarity, as for instance, the two polynucleotide chains in the Watson-Crick model.” Abstracting the term “polarity” from that model, he defined polarity simply by the criterion that elements of opposite polarity for some reason or other cannot undergo a reciprocal exchange (because they cannot be joined together end to end). “If the two elements revealed in Taylor’s experiment have opposite polarity,” he asserted, then “every

T HE U NSEEN B AND

277

Fig. 8.3. Diagram of DNA replication and synthesis by Max Delbru¨ck in a letter to J. H. Taylor, 1 February 1957. Reproduction courtesy of the Archives, California Institute of Technology.

sister exchange observed at the second metaphase in one pair of sisters must be accompanied by an identical sister exchange in the corresponding pair of sisters in the sister nucleus [sic].” If the sister exchanges are observed to be always twinned, “it can be concluded that the two elements referred to do not have opposite polarity.” 12 Delbru¨ck “proved” this proposition with the aid of schematic diagrams showing that a single exchange cannot occur in the two-element stage, since they have opposite polarities. In the four-element stage, such an exchange can take place, but only if two such exchanges occur at the identical level along the strands. Otherwise there could not be the equal distribution of label shown by the experimental data. He sent his memo to Taylor, who responded enthusiastically, on 4 February, that “everything I have seen to date is consistent with” the hypothesis that the two strands have opposite polarity. He sent Delbru¨ck sketches of the chromosomes of ten diploid cells from roots of bulbs of Bellevalia romana that showed exchanges of radioactive segments paired as Delbru¨ck had predicted.13 Delbru¨ck responded by return mail, on 11 February, “nobody could be more amazed and excited than I by your wonderful finding. It is so rare in biology that a simple logical argument points the way to a discovery and that an experimental procedure is so powerful that it can right away give an unambiguous answer.” The result proved, he thought, both the polarity of the elements and that “the sister exchanges occur obligatorily twinned.” Taylor’s earlier results “clearly disproved my model for W. C. replication,” but now they demonstrate

278

T HE U NSEEN B AND

for “the mysterious large elements” the same kind of twinned exchange that Delbru¨ck had invented in his attempt to disentangle the interlocks of the Watson-Crick duplex during replication. “Strange world,” he thought.14 During the next week Delbru¨ck decided that he had been too precipitous in abandoning the DNA replication model that he had proposed in 1954 to obviate the unwinding problem. He thought of a way in which he could reconcile the idea of breaks and reunions with the experimental evidence that labeled DNA is not evenly dispersed into the successive generations of DNA molecules. He wrote Taylor another letter on 18 February, withdrawing his surrender of the previous week and replacing it with a scheme modified according to his latest thinking: “On second thought I would like to take back a statement in my last letter. I said that your findings disprove my model of DNA replication, because my model leads to 1 :1 distribution of label at each replication. However, my model predicts this result only if one assumes that the breaks and reunions occur statistically every half turn of the helix. If they occur exactly at every half turn (every fifth nucleotide) and if the break points are the same at every replication, then my model gives the same distribution as the Levinthal scheme, as illustrated on the attached sheet.” His newest scheme illustrated “the distribution of label in two successive replications for a WC helix 1.5 turns long” (figure 8.4). Delbru¨ck acknowledged at once some difficulties with his revised scheme: It seems of course pretty extravagant to assume that the breaks occur at exactly reproducible points at each replication, but, who knows, there may be landmarks along the way to keep things orderly. I find your discovery of the postulated type of double exchange sufficiently encouraging to make me want to hesitate before discarding my mechanism of DNA replication. A decision could probably only be reached by enzymatic breakdown of DNA after one cycle of replication in 100% labelled medium. For the dinucleotides the ordinary schemes predict that one should find hot-hot and cold-cold ones, whereas my model predicts that one should find 20% hot-cold dinucleotides.15 The revival of his replication model depended on the selectivity with which Delbru¨ck transferred properties upward or downward along the several orders of magnitude that, as Taylor had remarked, separate reasoning about DNA from reasoning about chromosomes. At

T HE U NSEEN B AND

279

Fig. 8.4. Model of DNA replication proposed by Max Delbru¨ck in a letter to J. H. Taylor, 18 February 1957. Reproduction courtesy of the Archives, California Institute of Technology.

first Delbru¨ck transferred Taylor’s evidence for the separation of intact chromosome strands to the molecular level, a move that led him temporarily to give up his dispersive mechanism. But Taylor’s observation of a particular type of double exchange at the chromosome level, which encouraged him to think that such a type might take place also at the level of the nucleotide strands of the double helix, then prompted him to disregard the first analogy and cling to the second. Seeking alternative experimental evidence for his view, he took refuge in the “Levinthal scheme,” that is, in the ambiguous result of the experiments that Cyrus Levinthal had carried out on bacteriophage replication with 32 P, which showed a high concentration of parental DNA in relatively few progeny, not a random distribution among the progeny (see above, pp. 103–107). By devising a scheme in which some of the duplexes resulting from replications were completely labeled and others were only partly labeled (see figure 8.4), Delbru¨ck could conform to Levinthal’s finding and still save the essential feature of his scheme. Both of Delbru¨ck’s new schemes were saved from being pure speculation by the fact that he drew from them testable consequences. In the case of sister exchanges, preliminary results seemed brilliantly to confirm his reasoning. In the case of his modified DNA replication scheme, the critical experiments remained to be done. Nevertheless, one can see in these ideas the tendencies that made Delbru¨ck the unique figure he was within the phage group. They suggest why he

280

T HE U NSEEN B AND

was seen as a trenchant, playful critic who was frequently wrong but who often stimulated others to constructive responses. Delbru¨ck was not only interested in his own model, however. He also liked the model Taylor had proposed to connect the replication of chromosomes with that of DNA. He ended his letter with the comment: “Your model of chromosome structure, with a double backbone and DNA chains attached to both sides, continues to be discussed widely. It has very attractive possibilities beyond those we discussed when you were here. It helps one for thinking about lysogenization, transduction, and transformation. I would urge you to formally propose it soon. (We call it the centipede model).” 16 The model to which Delbru¨ck referred was a refinement of the one described in Taylor’s recently published paper, and it may have emerged during the discussions that the two held at Caltech. The centipede model adapted to the simpler organization of bacteriophage DNA a scheme with which Taylor had originally described the organization of DNA in eukaryotic chromosomes. In a letter written in February in response to the memo on Taylor’s sister exchanges, which Delbru¨ck had also sent him, Gunther Stent wondered if Taylor’s investigation in general “makes the case for W-C replication even stronger.” Ironically, he added, “I received a bunch of reprints of our, possibly already outdated, replication epic. . . . What do you propose to do about mailing them out?” 17 Stent was, however, no more prepared than Delbru¨ck to capitulate. As we shall see, he was pursuing the “conservative” alternative of the three schemes that he and Delbru¨ck had outlined in their discussion of the replication problem, the reprints of which they had just received. Taylor and Delbru¨ck continued to exchange views on the chromatid exchanges in Taylor’s experiments on Vicia faba. Pointing out that single chromatid exchanges should be possible beginning with the second metaphase, Taylor wondered why he had not yet observed any and predicted different ratios between single and twin exchanges depending on whether there are polarity restrictions on the chromatids that can exchange material or whether the recombinations are random. By early April he had acquired further experimental data that included both single and double exchanges, but not in the expected ratios. From the diagrams of the exchanges Taylor sent him, Delbru¨ck revised the expected ratios. “The main conclusion,” he wrote to Taylor on 10 April, is that “in the unrestricted case the ratio of Twins to Singles is necessarily smaller than half,” a proportion incompatible with the observations.18

T HE U NSEEN B AND

281

Taylor accepted Delbru¨ck’s analysis of the expected ratios. During a trip to Washington in April to speak at the National Academy, Delbru¨ck stopped off in New York to thrash out the statistical questions with Taylor, and in May Taylor sent his completed manuscript on the question to Delbru¨ck for criticism.19 Delbru¨ck went over the manuscript with his usual critical eye and made numerous suggestions for changes, to which Taylor responded with gratitude. He submitted the revised manuscript, “Sister Chromatid Exchanges in Tritium-Labeled Chromosomes” to Genetics in August. The frequency ratio of twin exchanges to single exchanges, Taylor concluded there, “can be explained only on the hypothesis that the two strands of the chromosome are unlike, i.e., are not free to unite at random. Therefore, the chromosome has two features in common with the Watson-Crick model of DNA. It has two strands and the strands are different in some structural feature that restricts reunion to like strands when chromatid exchanges occur.” 20 Although the main object of his paper was to describe the “sister exchanges” between chromatids that had so engaged Delbru¨ck in February, Taylor included a summary of the model of the chromosome as “an array of DNA double helices” that he had proposed in his previous paper: The core is visualized as a double ribbon with DNA double helices attached along the edges so that each helix has one polynucleotide chain attached to one ribbon and the other chain of the same helix attached to the other ribbon, perhaps by a terminal phosphate group. Each ribbon with its attached DNA chains, possibly on both edges to make it symmetrical, would comprise one unit of the duplex. The duplex chromosome would separate into its components during duplication. Another ribbon would be built along each separating ribbon and the DNA double helices would begin to pull apart, with the separation from each other starting at their points of attachment to the ribbons. With the unattached free end free to rotate and with rotation of strands possible at single bonds along the polynucleotide chains in the separating regions, replication could proceed according to the Watson and Crick scheme. All new polynucleotide chains would be attached to the new ribbons and would segregate as units in future duplications of the chromosomes, while the sister chromatid exchanges would represent breaks and exchanges along the axis.21 Taylor did not draw attention to the fact that his new description provided a far more specific scheme for the location and replication

282

T HE U NSEEN B AND

of the polynucleotide chains than had his original proposal. Neither did he provide a revised diagram to illustrate his account of the structures. This was, however, evidently what Delbru¨ck referred to as the “centipede model.” What Taylor described as “one unit of the duplex” was called “the Taylor unit” at Caltech.22 Delbru¨ck did not bring Meselson and Stahl into any detailed discussions of the relation between his collaboration with Taylor and their experimental plans, nor did he share with them the recent shifts in his own ideas about DNA replication. Although they felt close to Delbru¨ck and were confident of the warmth of his support for their work, conversations with him were infrequent. They went to see him in his office only on special occasions, when they needed to consult him about a particular problem.23 Nevertheless, they were very aware of Taylor’s work with Delbru¨ck and of the implications of his earlier results for their own plans. Sometime during the summer of 1957, Meselson wrote in his workbook: “Taylor’s exp. has no rigorous implication for a molecular transfer exp. because it is not known what fraction of DNA in a chromosome is actually going to be ‘parental’ by next doubling.” 24 That is, because Taylor did not know how much of the DNA in the chromosomes had become labeled, he could not tell whether all or only some unknown fraction of the DNA was distributed in the manner revealed by his autoradiographs. Moreover, depending on the molecular model one chose to represent the results obtained at a level several orders of magnitude larger, one could interpret the double helices as replicating either conservatively or semiconservatively. The critique was probing. Taylor and Delbru¨ck both tended to slide too easily by analogy from the morphological level of the chromosomes to the molecular level of DNA. It was for Meselson, also an effort to reassure himself that the experiments he had for so long had in mind were still worth doing,25 that he and Stahl had not been forestalled while they had attended to other problems and opportunities. Meselson was also much interested in the sister exchanges and in Taylor’s centipede model. He thought it really wonderful that “you could tell something by the ratio of the twins and singles about the polarity of the subunits.” The model, on the other hand, as we shall see further on, struck him and Stahl as important but flawed.26

II At the beginning of 1957, Meselson had planned to go immediately to Europe after receiving his Ph.D. “and stay the summer before re-

T HE U NSEEN B AND

283

turning to Pasadena to work on the mutations.” 27 By the end of March, however, he had decided he would spend most of the summer working with Stahl on 5-bromouracil phage, but he still felt the need for some kind of break. In June he went off to the music festival in Aspen, Colorado. The first two nights he put down his sleeping bag in one of the many Victorian houses then standing empty. Afterward he moved into the old Hotel Jerome and stayed for another week. Once while sitting at the back of the tent in which the concerts were held, he was struck by the appearance of one of the flute players. Afterward he met her. Her name was Katherine, and he thought she was marvelous.28 When Meselson returned to Pasadena during the last week in June, Robert Sinsheimer had just arrived to take up his new position at Caltech. Sinsheimer’s new Model E analytical centrifuge was installed in time for his arrival.29 Almost as soon as Sinsheimer and his wife could move their furniture into the house they had rented in Pasadena, Meselson arranged for time on the ultracentrifuge. Despite their common interests, however, Meselson did not develop much interaction with Sinsheimer, who seemed to him a rather shy and formal man, somewhat isolated from the “rough and tumble” interactions of the phage group.30 Meselson had also been given the responsibility to introduce an undergraduate student, Eugene Robkin, to summer research.31 He began to teach Robkin to use the ultracentrifuge. On 24 June they focused the ultraviolet optical system of the Sinsheimer machine. The next day Meselson led Robkin through a centrifuge run, using a DNA preparation (of unspecified type—Robkin, who filled in the “materials” blank on the log, put down only “DNA ⫹ CsCl ρ ⫽ 1.66.”). They used a “partition” cell, that is, a centrifuge cell whose length was divided into two segments by a fixed partition with small holes in it. This arrangement allowed one to separate material contained in the two halves of the density gradient after the run. The run began in the late afternoon. During the evening, Meselson explained the Schlieren optics and the expected pattern of bands to Robkin, using the wide margin of the log page to sketch curves representing the density gradient and the band that would be expected to form in the “heavy” section of the partitioned cell. The developed films showed the anticipated band. The next day, Meselson extracted the contents (.4 ml) of the top of the cell, added 3 ml of denser CsCl (ρ ⫽ 1.716) solution, placed them in another partition cell, and made another run. This time a band formed in the top section of the cell. Evidently the aim of these experiments was to capture the banded material, controlling the half of the cell in which it appeared by altering the density of the CsCl

284

T HE U NSEEN B AND

solution; but their reason for doing so is not evident. On the twentyseventh he and Robkin centrifuged “hot T4 r240 shockate” in ρ ⫽ 1.716 CsCl, in an ordinary cell. The material may have been a slightly radioactive DNA preparation they had obtained from Sinsheimer.32 On the same day, Meselson finally wrote to Paul Doty to inform him that the two molecular weight determinations he and Stahl had made on Doty’s salmon sperm DNA had not yielded a figure in agreement with the 8 million that Doty had reached with light-scattering methods, as he had thought when he telephoned Doty in April (see above, p. 259), but only half as much. “The molecular weight estimate,” he said, “must be revised to 4 million or greater.” When there was density heterogeneity, he explained, their method gave only a minimum molecular weight. Because the “light side” of the distribution curve fit a Gaussian curve, Meselson thought that the value of 4 million “may well refer to the correct weight of some of the sample,” but he admitted that “this is a rather weak argument.” Reporting that a second run (carried out on May 17–22; see above, p. 264) had also yielded the same result, he asked Doty whether there “was anything done with the material since you scattered light from it which might have degraded it.” He also asked if Doty would be interested in determining, by the light-scattering method, the molecular weight of T4 bacteriophage DNA, which would subject the density gradient method to a critical comparison, and offered to prepare samples for him.33 Without waiting for a reply, Meselson took up another question connected both with Doty and with the molecular weight of DNA. Several biochemists had found that, when subjected to heat or to acids, DNA underwent physical changes analogous to those that took place in the denaturation of proteins. Their viscosity was altered, the absorption of ultraviolet light increased, and their light-scattering properties changed. In Brussels, Rene´ Thomas had been studying the denaturation of DNA for several years. Thomas concluded in 1954 that the increases in ultraviolet absorption caused by denaturing DNA— he had studied the effects of lowering the saline concentration, altering the pH, and heating the molecules—could best be explained by the supposition that “denaturation consists in the collapse of an ordered secondary structure, transforming, for example, the pair of polynucleotide helices into random threads.” Thomas suggested that the hydrogen bonds that held together the two polynucleotide threads according to the Watson and Crick structure were disrupted.34 In the same year K. A. Stacey and P. Alexander reported that in 4Molar urea

T HE U NSEEN B AND

285

the molecular weight of DNA was halved, and they interpreted the phenomenon as the separation of the two strands of the Watson-Crick structure. Doty and his collaborator, Stuart Rice, reported in 1955, however, that denatured DNA, “produced either by exposure to pH 2.6 (with or without 8M urea) . . . or by heating the neutral saline solution to 100°C for 15 minutes,” has the same molecular weight as “native DNA.” 35 In 1957 they claimed that when DNA is denatured, some of the hydrogen bonds that maintain the structure of the double helix are broken, but enough of them are left in place to hold the two strands together in a less ordered state.36 Having read earlier some of Thomas’s work on the denaturation of DNA, Meselson was deeply impressed with the beauty of the papers in which Thomas presented his results. Influenced more by Thomas’s view than by Doty’s, Meselson and Stahl now hoped that they could separate the two chains by denaturation and document that change with their new molecular weight method.37 On Saturday, 29 June, Meselson prepared to make runs on denatured and native DNA on the ultracentrifuge. By comparing the position and the width of bands formed by the substance in these two states, he might obtain information both about their relative molecular weights and their relative densities. The first run Meselson carried out on T4r240 shockate “heated 5 min[utes] in boiling H2O bath.” This was a conservative treatment. Doty and Rice had stated that fifteen minutes at the temperature of boiling water was required to produce a “completely denatured” product. Perhaps Meselson boiled his preparation only one-third as long to eliminate all possibility of further degradation, preferring to deal with a product that might be only partially denatured. Alternatively, he might have thought either that Doty was wrong in believing denaturation to be gradual, or else that the rate of denaturation in cesium chloride solution was not comparable to that in an ordinary buffer.38 He put the DNA into CsCl of density ρ ⫽ 1.716 and ran the centrifuge at his standard 44,700 RPM, beginning at 5 P.M. The run lasted for twenty-four hours and produced a well-defined band. Immediately he began a second run with the unheated T4r240 shockate added to the contents of the previous run. He now obtained two bands, amply separated. The band on the “heavier” side was also a little broader and more diffuse. That might indicate a lower molecular weight (figure 8.5). Which band represented the heated DNA? According to Doty, denatured DNA contracted in volume, with no change in molecular

286

T HE U NSEEN B AND

Fig. 8.5. Film from centrifuge run B-11, exposure 6

weight. It might be expected, therefore, to be more dense than native DNA. That it might also have a lower molecular weight was contrary to Doty’s results for salmon sperm DNA and favorable to the view that the chains separated. Some sketches of band widths and locations that Meselson drew on the margin of the log suggest that he may have been trying to interpret the result while the run continued, but it is not clear whether he reached any conclusions (figure 8.6). To clarify the situation, Meselson reran the preparation after diluting it fourfold. Starting on Monday afternoon (2 July), at four o’clock, he continued until early Wednesday afternoon. The two bands reappeared, similarly spaced but fainter, and now about equal in intensity and width—an outcome that favored Doty’s view that denaturation increased the density without changing the molecular weight.39 On 5 July Meselson ran another preparation of “hot T4r240 shockate” in a partition cell and again obtained a band in the “heavy” section of the cell. The next day he centrifuged in Vinograd’s machine a mixture of heated and unheated T2 “sour” DNA. He used this time only half as much unheated as heated DNA. That he did so even though denatured DNA had been found to absorb ultraviolet light more strongly than native DNA suggests that, in spite of Doty’s view, he suspected that, as in the first denaturation run, the band of heated DNA would turn out to be broader than that of unheated and would therefore require more material to reach the same concentration at

T HE U NSEEN B AND

287

Fig. 8.6. Sketches from centrifuge log sheet for run B-11

the absorption peak. Running into difficulties with the light source, he reran the preparation on the Sinsheimer machine. The heavier band did appear again wider than the lighter band. By now Meselson probably concluded tentatively that the molecular weight of denatured T4 DNA is lower than that of the unheated material, in contrast to what Doty had found for salmon sperm DNA. Leaving this question, Meselson started one more run on hot T4 DNA in a partition cell, aborted it, and moved on, finally, to take up again the long-deferred continuation of the experiments with 5-bromouracil DNA begun in January.40 The occasion for the renewal of their effort was that Stahl was now able to prepare T4 stocks in which the thymine was almost completely replaced by 5-BU. The critical change from the procedures he had used in January was that he added aminopterin, a folic acid antagonist, to the growth medium of the bacteria, along with 5-BU. Aminopterin blocked the thymidine metabolism of the organisms, leaving them with no capacity to manufacture thymine. This addition appeared to Stahl and Meselson to be the “magic step” that produced the fully 5-BU substituted DNA they needed for a successful transfer experiment.41 When the bacterial cells were nearing the end of their exponential growth period, he added enough phage “to infect several cells out of each 100.” After about eight hours, the bacteria spontaneously

288

T HE U NSEEN B AND

Fig. 8.7. Film from centrifuge run B-16, exposure 3

lysed, releasing phage whose DNA he found to have 5-BU substituted for more than 90 percent of their thymine. On 6 July, Stahl prepared in this manner “T4 wild 5-BU shockate in two batches, labeled “I” and “II.” He may have tested the sensitivity of the stocks to X-rays and found them about twice as sensitive as unsubstituted stocks.42 Assisted by Robkin, Meselson centrifuged some of batch II of the new 5-BU T4 DNA in ρ ⫽ 1.779 CsCl solution in Vinograd’s machine, beginning at 6:45 P.M. on 8 July. At eight o’clock the next morning the run was ended when the pictures were lost. The two films saved showed a band beginning to form. The next afternoon they filled two cells with 5-BU T4 DNA preparations from batch I but ran only one of them, on the Sinsheimer machine. When Robkin developed the photographs taken the next morning, he found a single band, with a hazy edge on the lighter side (figure 8.7). Compared to the similar band resulting from the run of 13 January (see above, p. 228), this one was stronger and more distinct. It was also closer to the heavy end of the density gradient (both runs used a CsCl solution of ρ ⫽ 1.779)—heavy enough to have a “density characteristic of DNA fully substituted with 5-bromouracil (1.80).” 43 These results must have encouraged Meselson and Stahl.44 As he had done in January, Meselson added to a portion of the

T HE U NSEEN B AND

289

Fig. 8.8. Film from centrifuge run B-17, exposure 4

solution from the cell used for this run a solution containing normal T4 DNA (T4r240, rather than the wild type used in the preparation of the 5-BU DNA) and began another run at 9 P.M. on the same day. Thirtysix hours later, on the morning of 12 July, Robkin developed films showing two distinct, widely separated bands (figure 8.8).45 Although the one on the left, representing 5-BU DNA, was still a little soft on its light side, Meselson must have been convinced by now that the results were almost good enough to begin transfer experiments. The opportunities to make long centrifuge runs now suddenly expanded with the installation, in Vinograd’s laboratory, of the third analytical centrifuge that Beadle had requested in March (see above, p. 254). At 12:30 A.M. on 15 July, Meselson began, again with Robkin, the first recorded experiment on the new centrifuge, which was designated the “C” machine. (Sinsheimer’s machine was the “B” machine, whereas the runs on the first Vinograd machine, now well over one thousand, were still designated by number alone.) They ran a new 5-BU DNA shockate prepared on 10 July, together with normal T4 DNA. As the run continued the next morning, Meselson explained to Robkin, with curves drawn on the log, the relation between the position and density distribution of a band and the density and molecular weight of the DNA (see the uppermost drawing in figure 8.9). He also illustrated the X-ray sensitivity test that Stahl used to measure the degree and homogeneity of the incorporation of 5-BU (see the lower figure, where a straight line in a plot of the log of the percent surviving against the time indicates homogeneity, and the steeper slope of “5-BU” T4 indicates its greater sensitivity). Robkin took many exposures, but no films have survived. There were probably some technical problems with the new centrifuge. While these were attended to, Meselson and Robkin continued the

290

T HE U NSEEN B AND

Fig. 8.9. Sketches drawn on centrifuge log sheet for run C-1

experiments on Sinsheimer’s machine. On 16 July, they ran the new 5-BU shockate together with a sample of normal T4r240 DNA that Meselson had kept since January. The result was very similar to that of 12 July.46 On 18 July Meselson and Robkin reran the contents of the previous experiment after diluting it with more CsCl solution. Preparing Robkin to carry out centrifuge runs alone, Meselson wrote out on the back of the log page a list of reminders:

T HE U NSEEN B AND

1. 2. 3. 4. 5. 6.

291

Keep log records Close Spinco doors Never leave refrigeration on with temperature control off. Think! Plate type ⫽ comm. Log no. not oper. run no.

Robkin was able to produce a 5-BU band that was slightly more clearly defined than the previous ones.47 Still not entirely satisfied with the resolution of the 5-BU DNA bands, Meselson centrifuged another combination of 5-BU and normal DNA on 19 July, using this time the batch II 5-BU shockate prepared on July 6, together with the T4r240 shockate left from 1 January. The result was disappointing. The 5-BU band was wider and “fuzzier” than previous ones.48 By 22 July the C machine was again ready for operation. Meselson apparently had high priority in its use and no longer needed to borrow time regularly on Sinsheimer’s machine. During the first two days Robkin performed solo runs on the new ultracentrifuge, practicing techniques with cells containing CsCl solutions, and in one case only water.49 While he did so, Meselson and Stahl were getting ready for the experiment toward which they had long aspired. In preparation for their first formal transfer experiment, they infected bacteria grown in a normal medium with phage (presumably T4) into which 5-BU had been incorporated by previously infecting bacteria grown in a 5-BU medium. The multiplicity ratio was 13 phage per cell. After twenty-one minutes they lysed the bacteria, releasing the progeny phage. The ratio of phage to lysed cells was about 15: 1. They purified this phage by centrifuging it and adding enzymes to degrade bacterial material, then subjected the phage to osmotic shock to release its DNA. At 9:52 P.M. on Wednesday, 24 July, Meselson started the C machine on the centrifuge run for “Transfer Exp.#1.” That this was a special event for him is suggested by the unusually neat hand in which he filled out the information at the top of the centrifuge log (figure 8.10). If he had developed the films taken just after midnight right away, Meselson would have known before going to bed that two widely separated bands were beginning to form. In the series shot the next morning the bands were well defined. He kept going through the rest of the day, then turned the run off at 11:56 P.M. Thursday after twenty-six hours. The results were promising. The two bands formed were simi-

292

T HE U NSEEN B AND

Fig. 8.10. Top of the first centrifuge log sheet for run C-5

lar to those formed when he had centrifuged mixtures of 5-BU and normal DNA (figure 8.11).50 The main disappointment was that no intermediate band was visible. That result was not unduly discouraging, however, because, even if the replication was semiconservative, the amount of hybrid DNA that would be expected to remain after the multiple replications that constitute a complete bacteriophage reproductive cycle would be only 1 percent or less of the quantity of “normal” DNA. On 27 July Stahl answered a letter from Cy Levinthal informing him that Levinthal had accepted a position at MIT. After congratulating Levinthal and reporting on some experiments that Bob Edgar, who had recently arrived at Caltech, was currently running, Stahl wrote, that “One (1) transfer experiment has been run.” After summarizing the preparative steps, he described the result: “The DNA formed two bands in the centrifuge—1 band unsubstituted and the other fully substituted. There was no evidence for DNA of intermediate substitution. However, there wasn’t much total DNA in the centrifuge cell, and all the biological controls were not done. But, the system is working, and we should have a believable result soon.” 51 Meselson was less optimistic than Stahl that the system was working. Without evidence for an intermediate band, they had discovered little about the replication process. The weaker fully substituted band was expected, but fortuitous, because there should have been little or no fully substituted DNA left in the progeny phage after the multiple replications of a single reproductive cycle of the phage in unsubsti-

T HE U NSEEN B AND

293

Fig. 8.11. Film from centrifuge run C-5, exposure 6

tuted bacteria, whether the process was semiconservative, conservative, or dispersive. That band represented phage that had failed to inject the bacteria or to replicate, and its presence served only as a convenient marker.52 Merging two investigative strands, Meselson now used shockate left from the transfer experiment to pursue the denaturation question. He heated 0.014 ml of the “5BU pilot trans. #1 shockate” for five minutes at 100°C, then added 0.010 ml more of the same shockate unheated and centrifuged them together on 28 July. Robkin, who completed the run, obtained two bands, both toward the “light” end of the cell. The upper of the two, representing native DNA, was intense, whereas the lower one, representing the heated material, was faint. Having begun with approximately equal quantities of both, Meselson must have concluded that—unless he had made a mistake in preparing the samples—either the heating had caused a large loss of material or the denaturation process was so gradual that after five minutes most of the heated material was still in native form. Because there was no way to account for a loss, except for the doubtful possibility that the heated material stuck to the sides of the tube, he probably chose the second explanation, and decided to prolong the heating.53 The next day Meselson and Robkin repeated the experiment, heating the trans-

294

T HE U NSEEN B AND

Fig. 8.12. Film from centrifuge run C-8, exposure 4

fer shockate for fifteen minutes at 100°. This was the same treatment recommended by Doty for the complete denaturation of DNA in ordinary buffer solution. Robkin kept the run going for three days. The films he took are, however, missing.54 While this experiment continued, Meselson used the other Vinograd ultracentrifuge to begin, on 31 July, another transfer experiment. (He called it “Transfer Exp. #3,” but there is no record of a transfer experiment 2.) Aside from the fact that he used a little more shockate (0.72 ml. instead of 0.64 ml), and made the pH 6.5 instead of 7.5, the experiment was the same as the first one. Robkin finished out the run, twenty-four hours later, on 1 August, while his long denaturation run was still under way. No films survive from this experiment, but it is likely that they showed this time a single band on the light side of the cell, probably representing fully unsubstituted DNA. (For the transfer experiments Meselson used CsCl of ρ ⫽ 1.756, a density that would cause 5-BU substituted DNA to appear near the heavy end of the cell and unsubstituted near the light end.) To check that interpretation, Robkin reran the contents of the cell with a small amount of 5-BU shockate added to produce a marker. The result is shown in figure 8.12. This outcome was satisfactory in the sense that a strong unsubsti-

T HE U NSEEN B AND

295

tuted band appeared together with a faint 5-BU marker band. There was, however, still no sign of the intermediate band they would have liked to see.55 Either because he now considered the transfer experiments less promising or because he had received a new supply of Doty salmon sperm DNA (made by Michael Litt, a predoctoral student at Harvard), Meselson turned again, on 3 August, to denaturation. He heated 2 ml of the salmon sperm DNA for six minutes at 97–98°C and added an equal amount of unheated DNA, placing them in a CsCl solution of density 1.716. Running the centrifuge at 35,600 RPM for about thirtysix hours, he obtained a strong single band with a fainter extension on its “heavy” side. Probably he interpreted the strong band as undenatured DNA and the faint band as denatured DNA whose density was greater than that of the undenatured. Its faintness might have been due either to the DNA having been only partially denatured in the length of time he had heated it, or to the CsCl not having been dense enough to float the denatured DNA. To check the first possibility, he heated the contents of the run to 99° for fifteen minutes and centrifuged it again on 5 August. (He also must have added some unheated DNA that he did not record.) This time he found two clearly separated bands. They were of similar width, but the “heavy” band was now stronger.56 The result fit Doty’s view that salmon sperm DNA increases in density, without changing its molecular weight, when it is denatured. On 11 August, Meselson centrifuged some chicken fibroblast DNA made by Edward Simon, a younger graduate student working in the laboratory of Renato Dulbecco. Two days later he came back to the transfer experiment, looking once more in the remaining transfer pilot shockate for the elusive intermediate band. Using a somewhat less dense CsCl solution (ρ ⫽ 1.7), he obtained a single band in the middle of the cell.57 (It was, Meselson commented in 1992, “another nice, clean, negative” result. “I don’t understand,” he asked in retrospect, “why I didn’t put in a lot more” of the shockate.)58 On 16 August, Meselson tried again, not using more shockate, but adding a buffer to bring the pH to 9.4, to see if the molecules would aggregate. The outcome was unchanged. Two days later, the same shockate at pH 7.0 produced the same band. (Reviewing these runs in 1992, Meselson depicted himself as “desperately trying to get” the experiment to work.)59 This was the last centrifuge experiment that Meselson himself per-

296

T HE U NSEEN B AND

formed during the summer. He left instructions for Robkin to carry out more runs with the transfer experiment shockate and salmon sperm DNA,60 while he and Stahl left for Cold Spring Harbor to, as Delbru¨ck put it, “present the West Coast view” at the annual phage meeting. In a letter to Carsten Bresch, Delbru¨ck gave his view of the state of Meselson and Stahl’s experiments on the replication of DNA: Meselson has made two transfer experiments (as yet not quite perfect) with 5-bromouracil phages reproducing in normal medium. These experiments did not show any evidence for “hybrid” DNA molecules (one strand 5BU, the other thymine). These hybrids, if they are formed during replication, should be clearly demonstrable in the ultracentrifuge by his density-centrifugation method. . . . He may yet find such hybrid molecules by improved techniques (using N15 instead of Br, to alter the density of the parental DNA, since the 5BU DNA molecules may be physiologically too different from normal ones), but the fact is that as yet we do not know how phage DNA segregates on replication. Earlier, Delbru¨ck might have taken Meselson’s negative result as reinforcement for his skepticism that DNA can replicate by the separation of the two strands of a Watson-Crick duplex. Now, he inferred that “we know less about” the replication of phage DNA “than we know (since Taylor) about the replication of chromosomes.” Still strongly supportive of Taylor’s work, Delbru¨ck was now somewhat more cautious than he had been in the spring about drawing inferences from the replication of chromosomes to the replication of DNA: “Whether the two elements of Taylor’s chromosome are single polynucleotide chains is of course dubious. They may be threads of a nonDNA backbone to which the DNA molecules are attached sidewise (the centipede model) or the chromosome structure may be still more complex. What we have to require, though, is this: that all the DNA in one chromosome is organized before replication into two units, not more and not less, such that each of these units stays in one piece through an indefinite number of generations (except for the sister exchanges).” Here Delbru¨ck recognized, as Meselson had noted in his workbook, that one of the limitations of Taylor’s experiment was that it did not necessarily account for the distribution of all of the DNA in the chromosome. At present it would be best, Delbru¨ck thought, “to discuss phage genetics quite separate from that of higher organisms, let each run its course and hope for the best that some day in the future

T HE U NSEEN B AND

297

a way will be found to relate them to each other.” 61 This was a stark retreat from the attitude that he and Stent had expressed in their review of the replication question the year before, when they had predicted that the decisive experiments on DNA replication would be carried out on bacteriophages.62 Delbru¨ck’s reference to the possibility that Meselson might perform a future transfer experiment with 15N in place of 5-BU was the result of a brainstorm that Meselson had one day in August, while leafing through a chemical supply catalogue, where he came across an advertisement from the Isomet Corporation of Palisades Park, New Jersey, offering commercially the stable heavy isotope of nitrogen.63 Like all sudden creative insights, this one had been prepared by more gradual developments, not all of which may have been on Meselson’s mind when the idea occurred to him. The basic idea of using a heavy isotope went back all the way to the day in 1954, during Jacques Monod’s lecture, when Meselson had thought of growing bacteria in deuterium water. Afterward he had actually entertained the idea that a stable isotope like 15N might be used to label parental phage DNA, but he had been deterred by the concern that the density differences would be far too small to detect (a concern possibly reinforced by Gunther Stent’s reaction to the proposal).64 Among the reasons he and Stahl had chosen 5-BU for the transfer experiments was that the high atomic weight of bromine provided a much greater difference in density between the substituted and unsubstituted DNA, and, therefore, a greater chance that they could be clearly separated in a centrifuge. When Meselson centrifuged normal and 5-BU DNA, however, the two density peaks were widely separated, and as he worked through the density gradient method in greater depth during the spring of 1957, he realized that it was capable of finer resolution than he had at first anticipated. By the time he happened on the 15N advertisement in August, therefore, his initial skepticism about using it had been dissipated by the demonstrated resolving power of the method. Then it would have been evident that the isotope of nitrogen had the huge advantage that it would have no biological effects on the organisms that could introduce artifacts into the experiments.65 In his letter to Bresch at the end of August, Delbru¨ck reported that among the chief concerns in the lab during the summer were “the (unsuccessful) efforts of Meselson and Charles Thomas to identify in the analytical centrifuge the big piece of Levinthal. Its nature remains a complete mystery. We suspect that it has something to do with an

298

T HE U NSEEN B AND

ability of the ‘small pieces’ (which have a remarkably homogeneous molecular weight of 14 million) to aggregate and to disaggregate under as yet un-understood conditions.” 66 Despite the impression Delbru¨ck’s report leaves, this was not a collaborative venture. When Cy Levinthal left the University of Michigan for MIT, his associate Charles Thomas was left behind. Sinsheimer wrote Delbru¨ck in March that he thought Thomas “would welcome an opportunity to work with me at Cal Tech for a few years as a Research Fellow. I think he is quite capable and I would be glad to have him.” Delbru¨ck encouraged the idea and thought there would be no trouble getting a fellowship for him.67 Matters were quickly arranged, and Thomas arrived at Caltech during the summer. While there, he tried to account for the contradiction between Levinthal’s experimental evidence for one large piece of DNA and the evidence from Meselson and Stahl’s molecular weight determinations that phage DNA molecules were all of the same molecular weight. Thomas left Caltech at the end of the summer without resolving the problem.

III At Cold Spring Harbor, according to Gunther Stent, the star of the phage meeting was Seymour Benzer, who “gave a really terrific talk on the topological construction of the T4rII map.” The young prince, however, was “Meselson, whom everyone recognizes as a coming light.” 68 Meselson made remarks there “about the calculation of the enol-keto ratio of 5 bromouracil,” and he and Stahl explained to various participants the theory of the mechanism of 5-BU mutagenesis that they hoped to test in their future experiments.69 On the way home from the meeting, Meselson went to Cambridge to discuss the molecular weights and denaturation of proteins with Paul Doty and his associates. Among those he met was Michael Litt, who was studying the morphology of DNA with the electron microscope. While there, Meselson showed the group how to use the density gradient method for determining molecular weights and applied it with them to tobacco mosaic virus particles.70 While Meselson and Stahl were away, Robkin helped Edward Simon begin a project to label with 5-bromouracil the DNA of tissue culture cells of a human carcinoma of the cervix (strain HeLa). Knowing of Meselson and Stahl’s work with 5-BU phage DNA, Simon started out on his own to apply a similar approach to the transfer of

T HE U NSEEN B AND

299

DNA from parent to progeny in HeLa cells. Admiring the boldness with which the younger graduate student undertook a venture in which they themselves were still bogged down, Meselson and Stahl gave Simon encouragement. On 28, 29, and 30 August, Robkin did three centrifuge runs for Simon: one with 5-BU HeLa DNA and two with controls. Meanwhile, the further experiments that Robkin performed on pilot shockate from Meselson’s 5-BU phage transfer experiments yielded nothing new.71 When Meselson returned to Pasadena, he met a newcomer at the house on San Pasqual Avenue. John Cairns, a thirty-five-year-old influenza virus researcher, had arrived from Australia in August, after an exhausting transpacific flight, to begin a four-month Rockefeller Foundation fellowship in the laboratory of Renato Dulbecco. Ill himself with influenza when he set foot on the Caltech campus, Cairns spent two days in bed at the Athenaeum before discovering that the cost of staying there had already consumed half of the money he had brought along. Getting himself up, he walked over to the laboratories to find somewhere else to stay and encountered Jan Drake. Drake invited Cairns to stay in the San Pasqual house but warned him that Matt Meselson, the real organizer of the establishment, was away, and would have to approve of him when he returned. Apprehensive that he might not be approved, Cairns found another place to lodge, but he continued to come to the house to eat.72 After Meselson returned to the laboratory, he assigned Robkin to experiments with denatured DNA that were probably inspired by his conversations with Doty’s group. Perhaps encouraged to try for more complete denaturation, they heated T4 DNA “a la Doty” for twenty minutes at 99°C. On 4 September, Robkin centrifuged this preparation in CsCl (ρ ⫽ 1.92, overall calculated density 1.70) and obtained a sharp band. Attempting to apply the same techniques to a protein, Meselson had Robkin centrifuge some bovine serum albumin the next day in a CsCl solution diluted to a density of 1.282. Then Robkin reheated the contents of the first run to 58° for thirty minutes and spun it also at a solution density of 1.283. On 9 September he ran serum albumin heated for 30 minutes at 58° at approximately the same density of solution. Finding out nothing of interest about proteins from these experiments (there are no films for any of these three runs),73 they returned to DNA. Two days later Robkin was finally ready to compare the T4 DNA heated for twenty minutes for the first of the runs with unheated T4 DNA. In (ρ ⫽ 1.707) CsCl, two well-defined and separated

300

T HE U NSEEN B AND

bands formed, the one on the denser side being broader than the other. This result must have suggested that the prolonged heat had both decreased the molecular weight and increased the density of the molecules, confirming Meselson’s earlier result and strengthening his belief that phage DNA must behave differently from what Doty had reported about salmon sperm DNA under heat denaturation.74 During the summer centrifuge runs that included heated DNA, Meselson noticed that the heating reduced the time necessary for the bands to form. In each of the runs that included heated and unheated T4 DNA, the earliest films taken showed the band of heated material already well-defined while the band from the unheated DNA was still diffusely concentrated in the region where the band was forming. This effect could be explained by an increase in the diffusion coefficient owing to a collapse of the native structure. Only in cases where there was independent evidence for a decrease in molecular weight, however, could they use this observation to support the view that the two strands had come apart, as opposed to the Doty claim that the structures became less orderly but left the two strands attached to one another. In the experiments they had so far performed, this evidence remained ambiguous.75 On 12 and 13 September, Robkin banded tobacco mosaic virus particles for a molecular weight determination to confirm the result Meselson and Doty had obtained in Cambridge. During the first run, at 20,410 RPM, Robkin noticed that the machine was “chattering badly.” A few days later a different counterbalance was installed to correct the problem, and Robkin checked the operation of the machine at several speeds.76 On 20 September Meselson prepared a CsCl solution with phosphate buffer at pH 7. He treated it for one hour in a shaker at 37° with Norit A, filtered it, added 0.001 percent gelatin, and adjusted the density of the solution to 1.486. Then he added whole T4 wild phage. The intended centrifuge run apparently never got started. Robkin wrote on the log “apparent denaturation of phage. . . . Hearsay & Heresy from Matt.” The following Tuesday Meselson did carry out a run containing phage T2 in a CsCl solution containing gelatin. He obtained two bands, the “heavier” one being fainter and less than half as wide as the “lighter” one.77 What was the purpose of these experiments with whole phage? In his letter to Levinthal in March, Meselson had written that they had at last “learned to band whole phage. Now we can supplement our

T HE U NSEEN B AND

301

transfer experiments with genetic information.” 78 What he had in mind is clearer in a paragraph that he included at the end of the statement of previous and proposed research that he wrote later that month (for a summary of the rest of the statement, see above, pp. 269– 270): Finally, virus particles themselves can be separated according to their density and they may be recovered from CsCl solutions with unimpaired infectivity. Two genetically different virus particles will be allowed to infect the same bacterium. One virus will contain completely heavy DNA and the other light. The progeny virus will be examined to see whether heavy virus always carry some of the genetic markers originally present in the heavy parent. In this way it can be shown whether the portions of DNA which remain intact during replication carry genetic information.79 If the experiments on whole phage performed in September were preliminary explorations for this ambitious plan, their continuation was forestalled by other changes in direction that soon intervened. Before he could “supplement” the transfer experiments, Meselson had still to make the transfer experiment itself work. To that recalcitrant problem he returned on 26 September. Because Robkin filled in the entry for material on transfer experiment 4 only with the uninformative notation “DNA ⫹ CsCl,” the experiment cannot be described in detail. From the appearance of the films of the following experiment, Meselson judged in 1992 that this time they must have put “a whale of a lot” of the shockate derived from the progeny phage into the CsCl solution. The experiment itself was confusing. They took some of the photographs on commercial film and some on X-ray film, and many of them did not come out. The best series, taken after about thirty hours, showed a single, very wide but still slowly contracting band.80 Two days later, Meselson took the pellicle remaining at the top of the centrifuge cell after the preceding run, washed it in CsCl solution (ρ ⫽ 1.71), added buffer, and centrifuged it again. He began the run just after midnight on Saturday. On Saturday afternoon the very wide, dark band first visible was gradually narrowing. By early Sunday morning, however, the band was not growing sharper but beginning to dissolve into multiple bands of varying intensity and width.81 What was Meselson attempting when he centrifuged the pellicle material? In 1992 he commented, “Maybe the hybrid DNA is stuck

302

T HE U NSEEN B AND

in the pellicle. Try buffering, try heat, try anything. Blast it out. You see, there will always be at the top, the meniscus—proteins, soapy stuff, whatever, and maybe it is trapped up there for some ineffable reason.” 82 While this run was still in progress, Meselson began, at 2:26 A.M. on Sunday, on the other Vinograd machine, another run with shockate from transfer experiment number 4. He finished on Monday morning.83 Then the whole long, frustrating attempt to follow DNA replication by labeling bacteriophage DNA with 5-bromouracil came to a sudden halt.

C HAPTER N INE

One Discovery, Three Stories

I At the end of September Meselson received the 15N he had ordered from the Isomet Corporation. The isotope arrived in the form of ammonium nitrate, contained in small vials.1 He and Stahl decided not only to switch from 5-BU to 15 N to produce heavy DNA but at the same time to carry out the transfer experiments with bacteria in place of phage.2 Although their shift was abrupt, it was not unpremeditated. In early February, Meselson had already contemplated using bacteria, when he still envisioned doing the experiments only with 5-BU. In the research statement he submitted then, he had outlined the strategy he would use. The plan was to grow the organisms in the heavy medium, transfer them to the light medium, withdraw samples at time intervals representing successive cell generations, lyse the bacteria to release their DNA, and centrifuge the latter in the density gradient.3 The first step in the new experimental design was to test whether Escherichia coli would grow normally in a medium containing heavy nitrogen. Preparing a medium in which the sole source of nitrogen was 15 NH4 Cl, they learned that the bacteria had no trouble in such an environment. Their next preliminary step was to lyse and centrifuge DNA from E. coli grown in an ordinary medium. As with phage (see above, p. 204), they wanted to prepare the DNA by a method that did not lose any of the constituents of the cells, so that they would not risk throwing out some fraction of the DNA. Guanidine hydrochloride, which had served well for the phage, was not effective with the bacteria. Meselson tried various other denaturing agents, including urea, before coming upon the detergent sodium dodecyl sulfate (called “Duponal” by its producer, the Du Pont Corporation), which worked perfectly. On Wednesday, 2 October, Meselson spun “N14 Coli DNA” in a

304

O NE D ISCOVERY, T HREE S TORIES

cesium chloride solution of density 1.742 and obtained a sharp single band.4 Meselson interrupted this promising start to perform some experiments on whole phage λ obtained from Jean Weigle. The special interest of this phage for Weigle and other virologists was that it was temperate, that is, under specified conditions it did not destroy the bacteria that it infected but became incorporated as a “prophage” into the bacterial genome, where it was replicated along with the reproduction of the bacteria. By various experimental means, however, the bacteria could be induced to lyse, releasing progeny phage. In 1955, M. L. Morse, a graduate student in the laboratory of Joshua Lederberg at the University of Wisconsin, discovered that phage λ could infect a strain of bacteria able to ferment galactose, be induced to lyse, and then transduce into another strain of bacteria that was unable to utilize galactose the ability to do so. Examining the phenomenon in further detail, Lederberg and Morse found that the Gal marker was apparently the only gene that phage λ could transduce. Weigle, who had been working with phage λ at Caltech since 1951, began also to investigate the nature of this transduction. Early in 1957 he published a paper suggesting a model of the mechanism by which the Gal⫹ marker becomes attached to the phage genome and transferred to the genome of the acceptor bacteria.5 Although he may have run phage λ at this particular time simply because Weigle had material conveniently available, Meselson was interested in this phage because he thought it might be more suitable than T4 for his plan to band whole phage. The tendency of T4 to break up in the centrifuge cell had hindered their early efforts, and he hoped that the smaller phage λ would be more stable. That Weigle had stocks of Gal transducing and nontransducing phage λ would enable Meselson eventually to investigate genetically different virus particles in the manner he had envisioned in his statement of proposed research of the previous spring (see above, p. 301). On 7 October, he centrifuged “λ from Jean Weigle” in a CsCl solution of ρ ⫽ 1.852, to which he added water to give a solution of ρ ⫽ 1.4928, and obtained a band that was sharper than any he had been able to attain with T4. At 1:08 A.M. on Friday, 11 October, he began a run on “λ Gal at 2 ⫻ 1010 ” (a preparation that also included a “helper phage”—wild type phage λ—because phage λ Gal is defective and cannot replicate in bacteria that it infects in isolation) in a solution of the same density as the previous run. The run yielded a faint but well-defined band whose position, compared

O NE D ISCOVERY, T HREE S TORIES

305

to the band produced by the nontransducing strain, was displaced to the heavier side of the cell.6 The sharpness of these bands and the separation of the λ band from the presumed λ Gal band quickly persuaded Meselson that phage λ offered the opportunity he had been looking for to supplement transfer experiments with studies of the genetics of whole phage. His first priority was, however, still to make the new design for a bacterial transfer experiment work. Two and a half hours after ending this experiment he returned, early on Friday afternoon, to a critical test of that plan: centrifuging a mixture of DNA from bacteria grown on 15 N and bacteria grown in normal medium. Although Meselson had realized, when he worked through the theory of the density gradient method, that it was capable of finer resolution than he had been able to count on during the early stages of the experiments with 5-BU substitution, the fact that the difference between the densities of ordinary DNA and DNA incorporating the isotope of nitrogen was much less than that between ordinary and 5-BU DNA left it uncertain whether “N14 and N15 Coli DNA” could be clearly separated. On 11 October, Meselson ran the experiment and obtained two sharp, fully separated bands.7 (See figure 9.1.) This result was highly encouraging, because the distance between the centers of the bands suggested that it would be just possible also to distinguish an intermediate band composed of hybrid DNA from the heavy and the light bands. Three days later Delbru¨ck wrote enthusiastically to Gus Doermann, “Frank and Matt are very busy with 5BU mutagenesis and with replication of N15 substituted bacterial DNA. The N15 DNA can be separated beautifully from normal DNA in the CsCl2 [sic] density gradient.” 8 Torn between the excitement of this beautiful start on the transfer experiment and the excitement of the beautiful bands formed by phage λ, Meselson turned once more to the latter before proceeding with the former. On 14 October, he again centrifuged λ Gal. Perhaps to bring out the second band, representing wild-type phage λ that he expected to be present, he doubled the quantity of the preparation that he had used in the previous run. He got more than he bargained for. In place of the single band, there were now four of them. Evidently surprised by the result, he wrote on the envelope in which he stored the film, “λ Gal: 4 bands!” 9 Despite the auspicious progress of their collaborative project, Frank Stahl was not happy in October 1957. Sensing the potential success of his long-held dream, Meselson was pressing forward with ever greater intensity. By now he was nearly monopolizing Vinograd’s sec-

306

O NE D ISCOVERY, T HREE S TORIES

Fig. 9.1. Film from centrifuge run C-44, exposure 3

ond centrifuge machine and, as it seemed to Stahl, working “manically.” 10 The latest turn in their plan for the transfer experiments was, however, drawing their combined venture further than ever from that part of their original project—the mutagenic effects of 5-bromouracil—that had been closest to Stahl’s own research interests. As Delbru¨ck’s comment suggests, Stahl had recently been able to make some progress on that part of the old plan. He obtained better incorporation of 5-BU into phage DNA and enough better growth cycles to prepare concentrated stocks of several mutant strains. Nevertheless, he had come to feel that he could not keep up with Meselson’s pace, that he was losing his own momentum and identity as a scientist and becoming little more than a cog in Meselson’s scientific machine. Whatever suggestions he could provide for their joint enterprise seemed to him so quickly rejected or improved upon by Meselson that his contributions were barely visible. If the transfer experiments should succeed, it would, he believed, be Meselson’s achievement, notwithstanding that his name would appear on the paper. The more his admiration for Meselson’s fearless attitude toward science grew, the less comfortable he became at the prospect of continuing indefinitely their partnership. The more avidly Meselson pursued what he believed was the

O NE D ISCOVERY, T HREE S TORIES

307

dream that he and Stahl shared, the more Stahl felt his own self-esteem diminish. Moreover, Stahl had many more complexities to contend with in his life than Meselson appeared to have. Whereas Meselson, who had neither financial concerns nor dependents, seemed to orient his whole being around his scientific quest, Stahl had to struggle to provide enough money for a growing family.11 Stahl’s unhappiness began to show in his relations with others around him. In early October his old mentor, Gus Doermann, had complained to Delbru¨ck that Stahl’s “attitude in scientific matters” was making Bob Edgar, another of his former students who had recently come to Caltech, “quite miserable in the laboratory,” even though Edgar “appreciated all of the practical things Frank did for him.” Doermann wondered, “What is troubling Frank?” Delbru¨ck had not noticed anything wrong and was “very much puzzled” by the charge.12 (The problem, as Meselson commented after reading the above, may have been as much due to the ease with which Edgar could be irritated, and his inability to appreciate teasing, as with Stahl’s purportedly “vicious” remarks.) 13 But Stahl really was troubled, so much so that he had decided he must get out soon. He was already looking for another job. Stahl did not explain to Meselson why he wanted to leave. He felt, probably correctly, that Matt would not have understood.14 In Meselson’s view, he and Stahl thought very much alike. He admired not only Stahl’s experimental skills but his analytical clarity. In their endless daily discussions about what to do next in the laboratory, he relied on Stahl’s critical judgment. In fact, probably more often than Stahl noticed, when their opinions differed Stahl’s may have been the one that quietly prevailed. Meselson felt very much “at one with Frank.” 15 In his thesis acknowledgments he had described Stahl not only as his “valued research partner” but as “a close companion.” 16 In his singleminded concentration on what he took to be the central thrust of their collaboration, Meselson was little aware of the doubts and frustrations overtaking his talented but troubled research companion. Meanwhile, Meselson and Stahl were ready to plan their first transfer experiment with E. coli and 15 N DNA. They devised together a protocol in which they would grow the bacteria in the heavy medium until the culture reached a titer of between 1 and 2 ⫻ 108 (a concentration high enough to give adequate amounts of DNA, but not so high as to retard the growth of the bacteria) and maintain this concentration afterward by successive additions of fresh medium. By taking colony

308

O NE D ISCOVERY, T HREE S TORIES

counts during such a growth period, they had ascertained that the population doubled in about 0.8 hours, a period that they would use in the experiment as the approximate generation time. They planned to switch the bacteria abruptly to a normal medium by adding a large excess of a solution containing ordinary ammonium chloride and to remove one sample immediately. That would represent the “0” generation. They would chill it to prevent further change. They would remove further samples representing several stages during the first and second generation times. They would then centrifuge and resuspend the samples in the cold, lyse the bacteria with Duponol to release their DNA, and store the samples until they were ready to place each one in turn in a cesium chloride solution for a run in the ultracentrifuge.17 Meselson and Stahl must have worked out this experimental strategy immediately after the trial run with mixed heavy and light E. coli, if not earlier. Before they could carry it out, Stahl had to leave for Missouri for a job interview. Too eager to wait for Stahl’s return, Meselson decided to perform the experiment by himself. Before Stahl’s departure Meselson suggested that it would be good to run the experiment in two directions—one beginning in heavy medium and switched to normal, the other beginning in normal and switched to heavy. Stahl cautioned him not to try to do both at once, because there was too much risk that he would mix up the samples.18 After Stahl had gone, Meselson decided that he could color-code the tubes in which he placed the samples and carry out a double experiment without confusion. To be absolutely thorough, he may even have added one or two control experiments in which he switched cultures from one heavy to another heavy medium, and from one light to another light medium.19 Late on the evening of 15 October he began a centrifuge run on the first of the samples he had collected, finishing it the following afternoon. This first run was probably one of the controls. It showed a single sharp band. That same evening he began the second run.20 Was there a definite moment when Meselson realized that the result was going to be clear-cut and decisive? I asked him that question during the second extended interview I held with him, in 1987. His recollection of such a moment was vivid: “That moment, whenever it was, when I developed that film, and saw those three sharp bands. I knew there had to be only two bands, I mixed up the samples, but seeing those sharp bands, particularly the one in the middle, I knew

O NE D ISCOVERY, T HREE S TORIES

309

that settled it. And I remember I went yelling across the street to the party that was going on in the house.” 21

II Did it really happen that way? Several years after this conversation, Meselson retrieved the centrifuge log and film negatives for the experiments he had conducted during this period. It was possible from this record to identify with confidence the sequence of centrifuge runs for the first transfer experiment with E. coli and 14N. Unfortunately, none of the films contains three bands. When the scientist’s memory conflicts with contemporary documents, historians often discard the former in favor of the latter. Were we to reconstruct the experiment from the surviving record without regard for Meselson’s recollection of the event, the story might look like this: Ignoring Stahl’s admonition, Meselson not only set up the experiment to go in both directions at once, intending to transfer one culture from a 15 N to a 14 N medium and a second culture from a 14 N to a 15 N medium, but in a compulsive desire to control for the possibility that the suspension and centrifuging might alter things,22 he added one or more control experiments in which he transferred a culture from its initial growth medium into another with the same nitrogen isotope. In order to distinguish the samples, he devised a notation that indicated the overall experiment as G, followed by a roman numeral, sometimes a capital letter, and finally the designation “14” or “15.” There is no surviving record of the meaning of these notations, and it has not been possible to infer their meaning unambiguously from those designated on the centrifuge logs. The roman numeral and letter may have referred, respectively, to the four initial cultures (including two controls) and the sequence of samples taken after transfer into the second medium, and the “15” or “14” referred to the fact that the medium from which the culture was transferred was, respectively, heavy or light. The first sample he centrifuged, on 15 October, was designated G III B-15. It yielded a sharp band. It was not possible to determine from its position alone whether this was a heavy or a light band until it could be compared with later runs, but because it probably represented a sample grown in 15 N just after transfer to 14 N, Meselson may have assumed the former. The next evening he reran the cell at a lower

310

O NE D ISCOVERY, T HREE S TORIES

Fig. 9.2. Sketch drawn on centrifuge log sheet for run C-50

speed and obtained, as he would have expected, a broader, less sharp band.23 In the early afternoon of Saturday, 19 October, Meselson began a centrifuge run with a cell in which he had mixed samples of G III B-15 and G I-15. Because it has so far not been possible to crack fully this notation code, it is not evident exactly what he expected to find. Late that night he drew on the back side of the log sheet, perhaps for the benefit of someone who had wandered in, a sketch of three bands (figure 9.2), labeled HH (for Heavy-Heavy), HL (Heavy-Light), and LL (Light-Light). This was not what he anticipated the films would show but the trace of an explanation he was giving of the general idea of the experiment.24 When he developed the films in the early hours of Sunday morning, there were two strong bands whose centers were about as far apart as those in the preliminary experiment with mixed 15 N and 14 N Coli DNA (see above, p. 306). The samples represented, therefore, either controls, or DNA from cultures grown, respectively, in heavy and in light medium just after transfer to the other. So far he had obtained

O NE D ISCOVERY, T HREE S TORIES

311

nothing that would tell him how the experiment as a whole was going to turn out.25 Late in the afternoon of that same Sunday, Meselson inserted a sample from the tube labeled G III A-14 into a centrifuge cell and placed the cell in the rotor. Before he started the rotor he noticed that the cell had leaked. He refilled it, using three small gaskets to seal it more securely, and began the run. The sample was probably taken from the culture transferred from heavy to light medium a short time after that switch, and would therefore have been of special interest to him. At 10:40 Monday morning, seventeen hours into the run, when he could expect the cell to be approaching equilibrium, he checked to make sure the optical system was at the best possible focus and took photographs. He began the last series of photographs at 9:07 P.M. and turned off the machine at 9:30.26 When he developed the films, Meselson noted, close beside a strong single band, and in the direction of lower density, the faint shadow of a second band. It looked very much like the beginning of a transfer. Despite his incipient excitement, he recognized that he might merely be seeing the halo that sometimes bordered a band and that it was too soon to decide what this film meant.27 So anxious was Meselson now to obtain a more decisive result that he began the next run (C-52), with a sample labeled G II-14, just twenty minutes after shutting down the centrifuge at the end of the previous one. He developed the films showing the cell at equilibrium on Tuesday afternoon (figure 9.3). Now he saw two well-defined bands of nearly equal intensity, separated so narrowly that one of them must be “heavy-light,” the hybrid product of the first replication after switching media that he had long sought. There is nothing on the centrifuge log expressing the emotion he must have experienced. At 5 P.M. he turned off the centrifuge.28 It was time to go home to prepare dinner for his housemates. Intent now on finishing the long series of centrifuge runs required by his complicated multiple experiment as quickly as possible, Meselson commandeered both of Vinograd’s machines on Tuesday to do two on the same day. Long before dawn he placed a mixture of sample G I-14 and G I-15 in the A machine. He started the centrifuge at 1:52 A.M. but soon noticed the cell leaking. He continued the run anyway, until nearly two o’clock in the afternoon, but took only two photographic exposures and did not save any films from the ruined effort. Three hours later he started the B machine with a sample from tube G

312

O NE D ISCOVERY, T HREE S TORIES

Fig. 9.3. Film from centrifuge run C-52, exposure 3. The single narrow band to the right is probably carbohydrate.

II-15. By Wednesday evening he had another suggestive but imperfect result. The film showed what looked like one unusually fat band or two closely spaced, incompletely resolved bands. Meselson thought that he was probably seeing hybrid and fully labeled DNA, but was disappointed that he had not obtained better bands. Figure 9.4 presents a flow chart of the first transfer experiment.29 By now Meselson was convinced that the eventual outcome of his quest was already settled. One of the runs had yielded clear evidence of the formation of a hybrid DNA, intermediate in density between DNA formed entirely with 15 N and ordinary DNA, and two other runs had given indications that could be interpreted similarly. He had also learned, however, to appreciate the wisdom of Stahl’s caution not to try to do everything at once. In at least one of the runs the result— either that a heavy band appeared when he had expected a light one, or the converse—showed that he had mixed up some of the tubes, just as Stahl predicted he would.30 Even while carrying out these last runs, Meselson was making preparations for a simpler experiment, one in which a single culture would be grown in heavy medium and transferred to normal medium. Without bothering to spin the remaining samples, Meselson started on the next day to begin the runs for “Transfer Experiment No. 2.”

III Suppose that the historian does not easily discard the scientist’s recollection, because he believes that it is implausible that the scientist would have so vivid a memory of something that never happened. In that case he may seek to modify details of the story that might well

|

Thurs. 18

15

N– 14 N E. coli transfer experiment

Fri. 19

Sun. 21

|

| G II ⫺ 14

Tues. 23

|

Wed.

G II ⫺ 15

| | G I-14 ⫹ G I-15

|

|

Mon. 22

G III ⫺ A14

| | G III B-15 ⫹ G I-15

Sat. 20

Duration of Run and Tube Notations

| | Cell of C47, lower RPM

|

Wed. 17

G III B-15

Tues. 16

Fig. 9.4. Summary of the first

C53

1306

C52

C51

C50

C48

C47

October 15

Centrifuge Run

|

1 wide or 2 incompletely sep. bands

leaked

2 narrowly separated bands

1 strong band and halo

2 widely separated bands

1 more diffuse band

1 sharp band

Result

314

O NE D ISCOVERY, T HREE S TORIES

have been inaccurate, while fitting what he takes to be the essence of the remembered event as best he can into the narrative constraints set by the documentary evidence. Where in the second story told above is there room for an event resembling the one recounted by Meselson in 1987? There is only one plausible location. The centrifuge run for the sample marked G II-14, a label indicating a control experiment in which the bacteria had been transferred only from one 14 N medium to another, was completed at 5 P.M. on Tuesday, 21 October. Busy with something else, Meselson was unable to develop the films until late in the evening. He returned to the laboratory, took the film cartridge into the dark room, removed the film, and began processing it. This time he saw two well-defined bands of nearly equal intensity, separated so narrowly that one of them had to be “heavylight.” Something was wrong, because the control experiment should have produced only a single “light” band. He immediately attributed the discrepancy to his having mislabeled the tubes, just as Stahl had said he would. Nevertheless, he could see, almost at a glance, that the matter was settled. Waving the film in his hand, he ran out of the laboratory and across the street, yelling that it was exactly as Watson and Crick had said. Even though it was a weeknight, one of the frequent parties in the house was in progress. Meselson hurried in and told everyone in sight that the experiment was mixed up but still a great success.31

IV None of these three stories, not even the one that eliminates Meselson’s memory of the central event, has been reconstructed independently by the historian. They are the product of an extended, active collaboration between scientist and historian and rest heavily on the scientist’s deeper understanding of the general experimental situation. Neither, however, has the historian been subordinated to the scientist, for the stories rest also on the historian’s experience in reconstructing narratives of scientific investigation from evidence based partly on memory and partly on contemporary records. I first raised the problem of the absence of three bands in any of the films produced from this experiment in the summer of 1992, during a conversation in which both Meselson and Stahl took part. Stahl noted that it was also hard to imagine, “according to our present understanding,” how Meselson could have produced three bands merely by con-

O NE D ISCOVERY, T HREE S TORIES

315

fusing the tubes. He would have had to pour two samples into the same centrifuge tube.32 Two years later I proposed an interpretation that I thought might, with an inversion of certain crucial details in Meselson’s memory, make it compatible with the record. Meselson was able to show easily why my explanation could not be right33 and suggested another one, which also turned out to be problematic. In March 1995, Meselson and I sat down together, with all of the relevant documents before us, to see if we could infer what must have taken place. We went over each of the centrifuge logs and examined each of the films. Meselson made several hypotheses concerning the coding of the tubes and tested each one by attempting to interpret the sequence of films accordingly. Each time, he encountered a contradiction. We searched through the box in which the films are stored, looking for a film containing three bands that might have been placed out of order. We entertained the possibility that some additional experiments had been performed that were not included in the loose-leaf notebook containing the centrifuge log, but Meselson concluded from the sequence and dating of the runs that nothing is missing. After nearly three hours of concentrated effort, he felt a little tired. “It’s one thing to have two hypotheses and not to be able to decide between them,” he remarked. “We don’t have any. It’s very frustrating. What are we going to do?” When I suggested that if we couldn’t solve it I could write it all up “as a good example of the indeterminacy of historical interpretation,” he was not satisfied.34 The failure of our efforts to solve the puzzle left Meselson more inclined than I was to doubt the validity of his memory. At one point he said, “I must admit after all this that . . . once you begin to question memory a lot, at least in my case . . . you begin to wonder if it really happened.” A little later he stated resignedly, “We cannot support . . . the memory of the party and we cannot support the memory of the three bands, so what is left?” Near the end of our conversation he thought that, even if he had seen three bands, that would not have led to the immediate conclusion that the tubes were mixed up, because the most natural initial inference would be that the bacteria were replicating at different rates. The three bands story, he said, “I think is a myth.” 35 Carrying out afterward my idea that the situation be presented as one in which the surviving evidence is insufficient for a definitive interpretation, I wrote a version of the above three stories that Meselson read and annotated. My third story contained an inference that

316

O NE D ISCOVERY, T HREE S TORIES

he showed was untenable and that led him to modify my account of the C-52 centrifuge run. Explaining his own reconstruction, Meselson wrote, on the margins of the earlier draft, “But two sharp close bands (i) for sure demonstrate hybrid DNA and (ii) could have been unexpected in sample G II-14 if it were mislabeled. These are the two main parts of the memory, i.e., proof of hybrid DNA and realization of a mistake in labeling.” Although still relying on an assumption not further documented (that the run was mislabeled as a control experiment), Meselson’s resolution seemed to me more plausible than any other alternative. Therefore, I accepted it in place of my version. In a commentary about an earlier stage of my joint efforts with Meselson to solve these problems, John Heilbron has pointedly asserted that documents written by the historical subject “long after the events it concerns” have “doubtful authenticity.” To “save Meselson’s . . . experience,” according to Heilbron, I have become “complicit” in his story.36 It must be acknowledged that efforts by the subject to reconstruct an account reconciling his memory with contemporary evidence have a status at least as problematic as the spontaneous memories themselves. Is the original memory important enough to go to such lengths to save? If the scientist was ready at one point to abandon his own memory, why should the historian cling to it? An easy answer in this case might be that I do not want to lose this story, because it forms the dramatic peak of the narrative. There are, however, more compelling reasons to think that the memory as a whole is not a myth. Much evidence exists that memories can be incorrect in many details, yet accurate in their core meaning. Moreover, recent studies have shown that, contrary to popular intuitions, the degree of confidence that a subject has in his own memories is largely unrelated to their reliability.37 The memories that are most indelible, it is generally agreed, are the ones in which an event is associated with a strong affect. Meselson’s memory has this quality. When I asked him, in our first conversation on the subject, for further details about the party, he could not supply any. “I frankly don’t remember,” he replied. “It’s a blur. It’s like getting drunk. It’s funny, you remember certain things very clearly. I remember lysing the bacteria with sodium dodecyl sulfate. I was, as far as I know, the person who devised that method, using detergent, to lyse bacteria. And I remember doing that. I remember loading the centrifuge cells, and I remember developing the film. It’s sort of like an automobile accident. But I don’t remember anything

O NE D ISCOVERY, T HREE S TORIES

317

afterward except seeing the bands and getting enormously elated, and running across the street.” 38 The procedural steps Meselson recalled were ones that he had repeated many times. It is possible that his memory of them is a reconstruction from generic experiences, rather than a direct recollection of the steps he took at that particular time. But the elation and the running are singular events, exactly the type that we are most likely to remember long afterward. Sometimes such events might become retrospectively attached to the wrong contexts, but in this case it is hard to imagine any alternative event during the time Meselson spent at Caltech to which he would have been more likely to respond in this way. Why does it matter to a history of the Meselson-Stahl experiment whether this memory is right or wrong? If we were interested only in the cognitive and operative pathway through which the two collaborators arrived at their historical achievement, it would seem sufficient to reconstruct this first flawed but decisive experiment from the archival record. That record verifies that within the period of several days that it took him to perform it, Meselson would have recognized both that it was imperfectly carried out and that it nevertheless had already settled the replication question. Is there anything significant to be gained by saving his recollection that this recognition came in a climactic moment of discovery? Popular images of science often exaggerate the significance of such “Eureka” experiences. When examined closely, these peak events sometimes dissolve into a more complex sequence of developments, in which the apparent suddenness of discovery is replaced by incremental stages in a gradually evolving or clarifying picture. Scientists do, nevertheless, sometimes experience authentic moments of high drama, when the expected or the unexpected is revealed within an interval of time so compressed that a flash of recognition remains ever afterward as a highlight of a lifetime. Thirty years later, Meselson described it as “a mystical kind of thing, just to see that there. . . . When I saw that, the feeling was—it was exactly like Watson and Crick said. . . . Furthermore, it was such a clear-cut result. I had the strong feeling that, in a sense, nature wanted us to know this, and she was now going to open all the secrets.” 39 For such an experience to seem as momentous as it was brief requires not only the special outward circumstances of the experimental

318

O NE D ISCOVERY, T HREE S TORIES

situation but a personality open to its emotional dimensions. Had Meselson been of a different temperament, had he not dreamed of such an event for so long, had he not concentrated his life so intently on it during the weeks leading up to it, the sight of that film might not have been received with the impact that sent him dashing across the street. The nature of Meselson’s experience was as much an expression of his subjective inner life as it was the product of an objective experimental event that he and his absent research partner had generated. It is because scientists do have such experiences, along with the reasoning and the actions that lead them to their insights and discoveries, that creative scientific investigation is so deeply a human activity.

C HAPTER T EN

An Extremely Beautiful Experiment

I Frank Stahl flew back from his job interview in Missouri hopeful about his prospects there. A former classmate from Rochester who was already at the university depicted it to him as a pleasant place in which to work. Its rural atmosphere appealed to Stahl. The university had been strong in classical genetics but was off the beaten track for the newer molecular genetics—a situation that seemed to him advantageous, because he would be able to test his abilities independent of the ideas of other people. He thought also that he would have an opportunity to take students who were not necessarily outstanding in their own estimations and turn them into real scientists.1 Upon Stahl’s return Meselson showed him the photographs resulting from the transfer experiment begun in his absence. The outcome, Meselson said, was ludicrous—an artifact due to his having mixed up the DNA samples just as Stahl had teasingly said he might. But the appearance of a “half-heavy band” must be real; there was no way he could have messed things up so as to make that happen. Immediately they made preparations for a second experiment, and Meselson agreed this time to simplify matters by running it only in one direction. They chose to grow the bacteria first on 15N and then switch them to 14N, rather than the other way around, because that strategy minimized the quantity of the expensive isotope required. It was used only in the early stages of bacterial growth, when the culture was smaller.2 On 21 or 22 October, Stahl and Meselson began growing Escherichia coli B in the standard M9 glucose salts medium, with 15 NH4 Cl as the only source of the nutrient nitrogen. They followed the growth of the culture by standard assay methods and microscopical cell counts. After about twelve hours, when the titer had reached 2 ⫻ 10 8,

320

A N E XTREMELY B EAUTIFUL E XPERIMENT

representing fourteen bacterial generations, they switched the bacteria abruptly to 14 N by adding a tenfold excess of medium containing 14 NH4 Cl. This medium included additional sources of nitrogen in the form of ribosides of the four bases contained in DNA. Meselson and Stahl had hoped to use deoxyribosides, which would have provided more immediate precursors for the DNA. They were worried that a leftover pool of 15 N might compete with the 14 N precursors added and diffuse the effects of the shift in the medium. The presence of immediate 14 N precursors, they hoped, would bypass this problem. No deoxyribosides were available at Caltech, however, and in the event the problem did not arise. At the time of the shift they withdrew the first sample from which they would extract the DNA. They chilled it immediately, carried out the procedures for lysing the bacteria, placed the lysate in a tube labeled 2-I-A (for second experiment, initial medium, and first sample), and stored it in the cold.3 As the bacteria continued to grow, Stahl and Meselson added fresh medium as necessary to keep the titer between 1 and 2 ⫻ 10 8. They removed five more samples at approximately fifteen-minute intervals, lysed them, and stored them in tubes labeled 2-II-A to 2-II-E. While they were making these preparations, Meselson was still completing, on 22 October, the last centrifuge run with a sample from the first experiment.4 On Wednesday, 23 October, at 6:27 P.M., Meselson started the ultracentrifuge up again with the sample (2-I-A) taken just before the medium had been switched in the second experiment. Twenty-two hours later he switched off the machine and took the exposed photographic films into the darkroom. When developed they showed a single sharp band, representing “pure heavy” DNA. Working now around the clock, he began at 6:32 on the same evening a run with the first sample drawn during the bacterial generation following the switch (2-II-A). That run, completed just before noon on Friday, yielded two bands— an intense heavy band and a fainter band just to its lighter side. By 3:55 that afternoon he was ready to begin centrifuging the second posttransfer sample, 2-II-B.5 The centrifugation of sample 2-II-B was finished on Saturday morning. It showed two bands also. Now, however, the band in the position of heavy DNA was faint, whereas that in the “heavy-light” position was stronger. Within forty-five minutes of the completion of this run Meselson switched the centrifuge back on with 2-II-C, the sample taken from the bacteria at approximately the end of the first generation of their growth in the medium containing ordinary nitrogen. By Sun-

A N E XTREMELY B EAUTIFUL E XPERIMENT

321

day afternoon he knew that this lysate produced the anticipated single band of heavy-light DNA. Moving on that same afternoon to sample 2-II-D, he had found by late Monday afternoon that midway through the second generation of growth a band representing “pure light” DNA was appearing in its proper place above the heavy-light band, and there was no fully heavy band. Continuing without letup, he began on Monday evening the run with 2-II-E, drawn at the end of the second generation. Late Tuesday he obtained the remarkable final result: there were two bands that appeared visually to be of equal intensity, in the positions of heavy-light and pure light DNA. By then Meselson and Stahl knew that they had produced an experimental demonstration of semiconservative replication so nearly perfect that it was almost embarrassing.6 This result immediately raised the question, what were the units that were replicated semiconservatively? If, for example, each of the two subunits separated in the process was a complete double helix, and the units were pairs of double helices, then the latter might be held together by proteins. To test this possibility, Meselson ran two experiments, on Tuesday and Wednesday, in which he placed samples of 2-II-C (which had during the previous run shown a single intermediate band) in the centrifuge cell with guanidine hydrochloride added to the usual cesium chloride solution. Guanidine hydrochloride would be expected to dissociate molecules that might be conjoined by protein. The band remained intact, supporting the view that the units were single DNA molecules, but Meselson resisted the conclusion that the experiment had confirmed that the subunits were, therefore, single polynucleotide strands.7 The whole house on San Pasqual Street had by this time been absorbed into the “throes of the N 15 -N 14 transfer experiment.” Meselson explained the procedures and the problem the experiment was intended to solve in great detail over dinner, with numerous diagrams drawn on paper napkins. To John Cairns, who came with little knowledge of genetics and little recollection of what he had learned of chemistry, Meselson’s “seminars” were “seamless and impregnable.” Meselson thought that Cairns grasped the situation clearly and immediately, but he saw that Cairns also wanted to understand the problem very deeply. He suggested, therefore, that Cairns might best teach himself by building a model of DNA. Using the coordinates published by Maurice Wilkins, Cairns built two models, one space-filling and the other open. Sinsheimer found him working on them and expressed skepticism that the effort was worthwhile, but Cairns had the last

322

A N E XTREMELY B EAUTIFUL E XPERIMENT

laugh. Wanting to see how the glucose molecules fit into “sweet” T4 DNA, he added them to the cytosine bases on the model and found that they extended into the “big groove” of the double helix. When Sinsheimer came around again, he acknowledged that he had not realized that they did.8 When Cairns, who was also improving his background by taking the phage course at Caltech, had learned enough to grasp the full significance of the transfer experiment, he was deeply impressed. Meselson spent much more time explaining its details and its implications to him.9 Either during or immediately after the successful transfer experiment, Meselson called up his old University of Chicago roommate, Ernest Callenbach, now in Berkeley, to tell him about the results. A leading environmentalist and founding editor of the journal Film Quarterly, Callenbach kept in close touch with Meselson’s work. Callenbach then called Gunther Stent to let him know that Meselson had shown that DNA replication is semiconservative. Stent was crushed by the news.10 It was not because Meselson and Stahl had solved an experimental problem that had eluded him that Stent was tremendously disappointed, but because he had, during the eighteen months since writing with Delbru¨ck the paper on the mechanism of DNA replication, become fixated on the idea that the process must be conservative.11 Finding his experiments to trace the distribution of DNA atoms in phage progeny by means of 32P decay always frustrated by the fact that replication, recombination, and mating processes seemed to occur together, he set out to devise a molecular scheme that could account for genetic recombination as well as replication.12 The idea that Stent came up with was that a third polynucleotide chain could grow within the “deep groove” of the DNA double helix (figures 10.1 and 10.2). This third strand, composed of RNA, would then unwind in the manner of the Levinthal and Crane “speedometer cable” rotation (see above, p. 107). Two such molecules could then “mate,” and this RNA duplex could form the template on which progeny DNA duplexes were formed. Recombination events could be accounted for by matings between RNA strands that had formed on different DNA duplexes. With the help of the crystallographer Jerry Donohue, who had in 1953 supplied Jim Watson with the critical information that guanine and thymine occur primarily in the keto form,13 Stent worked out the idea that the Watson-Crick base pairs could form hydrogen bonds with the bases of a third polynucleotide chain. Stent de-

A N E XTREMELY B EAUTIFUL E XPERIMENT

323

Fig. 10.1. Purine pyrimidine base triplets proposed by Gunther Stent for a conservative DNA replication scheme

scribed his scheme in detail and adduced some experimental evidence in favor of it in a review article titled “Mating in the Reproduction of Bacterial Viruses,” intended for publication in Advances in Virus Research. He had just finished writing the manuscript when Callenbach’s call shattered the assumption about conservative replication of DNA duplexes that Stent had built into this elaborate mechanism.14 Despite this blow, the warm-hearted Stent admired Meselson and Stahl too much not to be at the same time excited about the brilliant experiment they had just pulled off. He was, therefore, pleased when Meselson called him to say that he was driving to Berkeley with John Cairns and hoped to talk with him about the results. Meselson had, as usual, combined professional and personal reasons for making the trip. He wanted to see Stent, “because he was at the heart of all these interesting subjects,” but his main motivation was to see his girlfriend Ani.15 Before leaving, Meselson conducted one more centrifuge experiment, in which he mixed together DNA from the initial samples drawn in the first and second experiments. Perhaps his purpose was to use the results of the clear-cut second experiment to help straighten out the mislabeled samples from the first. The run lasted from Friday, 1 November, until almost noon on Sunday. The combination of 2-I-[A] and G-I-14 formed two well-separated bands, indicating that the latter

324

A N E XTREMELY B EAUTIFUL E XPERIMENT

Fig. 10.2. Illustration of growth of a third polynucleotide chain within the groove of a DNA duplex in the conservative DNA replication scheme proposed by Gunther Stent

had been correctly identified as the DNA withdrawn just before the bacteria grown initially on a 14 N medium were switched to the heavy medium.16 Meselson and Cairns left for San Francisco in the black Thunderbird on Sunday afternoon. When they arrived, Cairns was “sent to stay with the Stents to get me out of the way,” while Meselson went to see Ani. A candlelight dinner prepared by her, in a softly lit room, with drinks and a perfectly set table, melted him. The next day Meselson showed up at the Virus Lab in Berkeley, bringing with him the photographic films from his latest experiment. When Stent saw these films, it was quickly obvious to him that they proved that bacterial DNA did replicate semiconservatively. During a long talk with Meselson and Cairns, Stent argued at first that the experiment did not exclude conservative replication. When he was eventually persuaded that it did, he complained cheerfully about this “devilish, hellish experiment” that was too watertight to challenge. He seemed pleased about Meselson and Stahl’s success, however, and his joking complaint about the experiment was so civilized that Meselson did not notice how devastating the result had been to Stent’s own ideas about DNA replication.

A N E XTREMELY B EAUTIFUL E XPERIMENT

325

Inviting Meselson to discuss the experiment at a quickly arranged seminar in the Virology Laboratory, Stent introduced him as “The Mozart of Molecular Biology,” a form of cultured high praise that somewhat embarrassed the speaker. After a dinner with Ani at the top of the Mark Hopkins Hotel, Meselson and Cairns left Berkeley early on Thursday. They drove down the coast, stopping at a vineyard where they had been instructed to purchase some wine for Delbru¨ck’s wife Manny, and stayed overnight at a picturesque but chilly hotel at Big Sur run by Norwegians who busied about in their stocking feet. At Carmel, Cairns received an American crew cut. On the way back to Pasadena they made an unscheduled stop when Meselson received a ticket for speeding. Writing Delbru¨ck about their visit on the same day that Meselson and Cairns departed, Stent gave expression to his strongly ambivalent feelings: We’ve had a few terrific days with Matt and John Cairns; they just headed back for Pasadena this morning—Naturally, I am not unexcited over the N 15 results and am mindful of your statement: “We are confident that the whole issue will be resolved before many a day. Who knows perhaps even before this goes to press?” Things do look good for the W.C. mechanism, I must admit (very generous of me, isn’t it?) but I still can’t see how the intimate connection between replication and genetic exchange can be accounted for from this point of view—there is one very long shot, i.e. a very remote possibility, that Matt’s bands are not DNA but the bacterial RNA—protein particles, which would have about the same density as DNA and could have been denatured from spherical into linear bodies by the Duponol treatment. But this will be easy to eliminate, of course, by DNase and RNase treatments. In the meantime, we are still going to look a little into “information” transfer during phage reproduction. I have thought of a “definitive” experiment which, no doubt, if it works at all will give ambiguous results. We may have an answer in a few weeks.17 With some bravado, Stent persuaded himself that there might be two kinds of DNA replication. The Meselson-Stahl experiment had detected the more trivial type, such as in ordinary bacterial multiplication, where no genetic change occurs, whereas his mechanism could explain the more important types in which replication is closely connected with genetic recombination. He allowed his manuscript to go to press without retracting his replication mechanism, but he felt that he received little support for his views, even from his friends in the field.18

326

A N E XTREMELY B EAUTIFUL E XPERIMENT

The Meselson-Stahl experiment was no kinder to the dispersive DNA replication mechanism favored by Stent’s mentor than to his own conservative mechanism, but that outcome did not dampen Max Delbru¨ck’s excitement about the event. When Meselson showed him the films, Delbru¨ck was immediately convinced of the formal result that DNA replicated semiconservatively. Like Meselson, Delbru¨ck wondered whether the subunits were the chains of a Watson-Crick model or something else. On 5 November he wrote to Helga Harm, “Meselson is making earth-shaking discoveries on the replication of DNA, and every afternoon he and Sinsheimer and I have been having endless discussions (with tea) about this.” Meselson felt for the first time that he was “really treated as an equal” by Delbru¨ck. Two weeks later Delbru¨ck gave his assessment to Carsten Bresch that “Meselson and Stahl have made an important discovery.” 19 The theoretical adjustment was probably less difficult for Delbru¨ck to make than for Stent. As Stent recalled in 1992, Delbru¨ck took “a light-hearted view of theory, so I probably was much more committed to theories than he was.” 20 On the first day after his return trip from Berkeley, Meselson wrote to Jim Watson, 8 November, 1957 Dear Jim, A transfer experiment with bacterial DNA has been completed—E. coli was grown from 10 4 to 10 8 cells per ml. in N 15 M9. The generation time was 45 minutes in this medium and was the same in a parallel N 14 culture. Upon reaching 10 8, a 20-fold excess of N 14 was added to the N 15 culture along with adenosine and uridine. Samples of bacteria were withdrawn just before the change of medium and afterwards for two generation times. The generation times, as measured both by colony formation assays and direct particle counts, remained constant at 45 minutes. The bacterial samples were chilled and centrifuged immediately upon their withdrawal. The sedimented cells were resuspended in versene, treated with lysozyme and then duponol and placed in the refrigerator. This treatment yields a clear lysate which is added to CsCl and centrifuged. Nothing is thrown out. DNA bands of three discrete densities were found in the various samples as shown in the following table [figure 10.3]. Times are given in units of one division time. Clean as a whistle! Who would have imagined that, with all the other great good luck we’ve had, the DNA molecules would replicate all at the same rate?21

A N E XTREMELY B EAUTIFUL E XPERIMENT

327

Fig. 10.3. Table from Meselson’s letter to Watson, 8 November 1957

Not knowing how many DNA molecules, or chromosomes, each bacterial cell contained, Meselson and Stahl had no reason to anticipate that the DNA replicated that way. Had the DNA not all replicated at the same rate, they would still have found just three types of bands, but the light band might have appeared before the heavy band had disappeared. They often talked afterward about what kind of clock there could be that was so accurate that no molecule replicated twice before all the molecules had replicated once.22 Meselson explained to Watson that they had not yet measured the intensities of the bands with the photometer, because the one they had been using was too much trouble to operate, and they were awaiting delivery of a new one. Therefore, “the films have been judged by eye” as the basis for the above estimates of the fractions of total DNA in each band. “The bands all seem to have the same width,” he added, “and the separations are clean looking. The intermediate band seems to be just in the middle.” 23 His ability to judge quantitative fractions of the DNA by the relative darkness of the bands on the films was a skill Meselson had acquired through his training as an X-ray crystallographer. It was customary in the analysis of diffraction patterns to estimate the relative intensities of the spots on the films by comparing them visually with a standard composed of a series of spots of graded densities.24 He also made a theoretical calculation of what the fractions ought to have been (figure 10.4).25 The resulting values were remarkably close to his visual estimates. Before they carried out the decisive experiments, Meselson and Stahl had sometimes sat around thinking about the possible outcomes and making up little verses to express them. The lines Meselson liked best turned out to be inapplicable to the actual result, but he could not resist putting them anyway into this letter to Watson:26 I was all set to send you a collection of verses the overall mood of which is set by the lines “Now N 15 by heavy trickery Ends the sway of Watson-Crickery . . .’’

328

A N E XTREMELY B EAUTIFUL E XPERIMENT

Fig. 10.4. Calculated fractions of heavy, heavy-light, and light DNA for the transfer experiment

But now we have WC with a mighty vengeance . . . or else a diabolical camouflage. Returning to a serious vein, Meselson gave a cautious interpretation of the result: “Formally, what the centrifuge says is that the nitrogen of the units which form a band is divided equally between two sub-units and that upon replication the sub-units separate from each other and become associated with new sub-units built from nitrogen quite recently present in the medium.” 27 In his next three paragraphs, Meselson discussed what these units might be. Here he could conveniently draw on the experience that he and his associates had gained using the density gradient method to determine molecular weights. Even without the densitometer readings and a formal calculation, he could tell visually from the width of the bands alone roughly what the molecular weight of the molecules contained in them should be: The units which form a band have a molecular weight of about 8 million and a frictional constant related to their MW as would be expected from centrifuge experiments with phage, calf, and

A N E XTREMELY B EAUTIFUL E XPERIMENT

329

other DNA samples. Also, these units have the same behavior upon denaturation as phage and salmon DNA. Nevertheless this does not prove that they are single DNA duplexes. Until the photometer arrives I can’t say whether these units are monodisperse. Aside from the question of the nature of the units themselves, we are also ignorant of the nature of the sub-units. Even if the units are WC duplexes, the sub-units need not be single strands. In all of this the temptation is great to believe in pure, simple WC explanations but we’ll try to determine more definitely the identity of the units and sub-units. I’ll write to you later of how we will attempt this. Please give it some thought and send along any ideas you come up with.28 Both Meselson and Stahl were personally convinced that the experiment had proven the position Watson and Crick had taken in 1953 to be right: that what they had observed was the separation and conservation of the two polynucleotide strands of the DNA double helix.29 Mention of a molecular weight, a frictional coefficient, and denaturation behavior consistent with those of DNA molecules was intended to suggest cautiously the likelihood that the “units” of the experiment were such molecules. Meselson’s resistance to the temptation to state his belief, even in an informal letter, that they actually were “WC duplexes” was due to a concern not to draw conclusions that went beyond what his data said.30 Whether Delbru¨ck reinforced Meselson’s own caution or whether, as Stahl comments, “It is more likely Matt’s approach to science in general, that when he makes a conclusion he makes sure that he has ruled out all other conceivable possibilities,” 31 is difficult to determine so long after the conversations that may have taken place between them on this subject. His circumspection about what the transfer experiment had demonstrated did not inhibit Meselson from wondering out loud to Watson about the broader implications that could be explored through further experiments. The line of thought he pursued most extensively was the connection between the subunits defined by their experiment and the subunits of Taylor’s structural model of the chromosome. “If we assume,” Meselson wrote, that the centrifuge experiment would have the same outcome if it had been done with the DNA from Vicia Faba [the bean seedlings], then it is necessary that one sub-unit of the centrifuge unit be associated with one sub-unit of the Taylor unit. This model of the chromosome includes Taylor’s proposal

330

A N E XTREMELY B EAUTIFUL E XPERIMENT

Fig. 10.5. Models for Taylor units sketched by Meselson in a letter to Watson, 8 November 1957

In Taylor’s model there are two sorts of attachment of the DNA to the Taylor unit: a 3′ and a 5′ attachment. In the right-hand lower model both attachments are the same. Also, in this model, new material might be laid down on each strand always in the same chemical direction, say the 3′ to 5′ direction. The two-fold rotational symmetry of the molecule would then be conserved even during replication. A “disadvantage” might be that, with this kind of symmetrical replication, half of the molecule would be single strand when replication is half completed. Maybe the strands could nevertheless be kept from collapsing into an undesirable configuration by some accessory structure. The unwinding of the DNA in the symmetrical chromosome might be made difficult by the requirement that, if the molecules remain attached to the Taylor units, uncoiling requires simultaneous breakage of all the interchain hydrogen bonds. Taylor’s own model still allows residueby-residue breakage. I wonder if the synthesis of new DNA might not take place without unwinding—some four stranded creature being an intermediate. Then the Taylor units might comprise some sort of machine for unwinding these. We shall try to look at DNA just after DNA doubling in synchronized cultures in hopes of finding interesting intermediates. We will survey the entire density spectrum so as not to miss interesting DNA-RNA-protein complexes. N 15 of course makes possible transfer experiments for nearly any cellular component which will band. Already we have found some interesting “mystery bands” which seem to be very large but homogeneous nucleoprotein structures. Do you have any ideas for preventing nucleo-protein disassociation in the CsCl [. . .] We will try adding Mg⫹⫹, keeping the pH low and using low temperatures and fresh samples.32 This speculation offers an illuminating insight into Meselson’s viewpoint, and mood, in wake of the experimental tour de force he and Stahl had achieved. Most prominent is his optimistic belief that

A N E XTREMELY B EAUTIFUL E XPERIMENT

331

they had developed an experimental method so powerful that they could attack with it all the central questions embedded in the “replication problem.” Displayed here also is a disposition to join in the theoretical play with structural models that had become popular during the three years since the advent of the double helix. Meselson would have had good reason to feel that the combination of this new experimental method and his training in structural chemistry under Pauling placed him in an advantageous position to enter that arena. There is here a distinct echo, too, of the difficulties that the “unwinding” conundrum had posed for all those who had thought about it since 1953, and a willingness to try his hand at circumventing the difficulty. The sketch with which Meselson depicted Taylor’s proposal must have been intended to represent the model that Taylor had described in the paper sent to Genetics in August (see above, p. 281). Meselson’s interpretation included details, such as the spatial orientation and asymmetry of the attachments of the chains to the ribbons, that Taylor had not mentioned and may not have thought through but that seem to be implications that Meselson and Stahl must have worked out as they contemplated his model. They had found a “deep inconsistency” between Taylor’s evidence for the opposite polarity of the two strands of the helix and the way in which it seemed to them that each strand must be attached “by a terminal phosphate group” to one of the core ribbons. The variant models that Meselson proposed in his letter, especially the one sketched on the lower right of the group, represented their attempts to devise a modification that would allow both terminal attachments to be the same. Meselson and Stahl were not satisfied, however, with the model Meselson proposed here. It avoided the asymmetry inherent in Taylor’s model only by incurring another difficulty: the necessity for a lot of single-stranded DNA when the replication is halfway done. That feature “seemed ugly” to them.33 (One might well have applied to this model Watson and Crick’s 1953 acknowledgment about their original proposal for the replication of DNA: that “it is difficult to see . . . how these processes occur without everything getting tangled.”) While they looked out for another model that might avoid both defects, Meselson was driven toward a suggestion that there might be a four-stranded DNA intermediate involved that bore some resemblance to ideas with which Watson himself had toyed during his last year at Caltech (see above, pp. 77–78). Meselson may have gotten the idea of a four-stranded DNA directly from Watson or indi-

332

A N E XTREMELY B EAUTIFUL E XPERIMENT

rectly through conversations with Delbru¨ck, who had been attracted to Watson’s four-stranded model “because there is no problem of untwiddling.” Finally, running through much of Meselson’s letter to Watson is a sense that the experimental result they had attained—clean though it was—remained incomplete. Unless they could identify the units and subunits more definitely, they would not fully have solved the problem they had set out to solve. As the paragraph following the discussion of Taylor units reveals, however, Meselson’s experimental aims also extended well beyond that problem: The questions regarding organization of DNA molecules into larger structures perhaps of importance in ordering the genetic determinants and making recombination possible are the ones that interest us most just now. I think a good program in this area, in addition to the above experiments with cell lysates, would be to characterize the DNA of lytic lambda hoping to find more than one piece. Lambda can be banded in CsCl very beautifully. If there is more than one piece of DNA in the virus, then N 15 transfer experiments with whole phage will tell if these molecules remain associated during their biological career. If they do, we can perhaps find gentle enough preparative methods (injection on bacterial membranes) to obtain these interesting aggregates. We might then find how they are built. If the molecules come apart during their careers, we have the puzzle of keeping order on the genetic map, unless each piece or one piece is a complete (known) genome. A result very interesting in itself and which will provide a very good handle to the problems of DNA organization is that whereas lambda gives one sharp band, lambda plus lambda Gal gives two!! 34 It is evident that Meselson was building these ideas on the experiments with phage that he had carried out just as he was getting under way with the 15 N– 14 N transfer experiments (see above, pp. 304–305). (None of the three recorded runs had produced just two bands, however. Either Meselson must have carried out another one whose result has not survived, or he was inferring that “lambda plus lambda Gal” should give two bands because they were biologically distinct.) Those experiments now appeared to him to provide the starting point for a new program to examine the organization of DNA molecules into a simple genome. After finishing his letter to Watson, Meselson wrote a nearly identical letter to Cy Levinthal. In doing so, he polished the sentences some-

A N E XTREMELY B EAUTIFUL E XPERIMENT

333

what and clarified a few points. The verse he had offered to Watson he reduced in this letter to the declaration “Perfect Watson-Crickery, or else a diabolical camouflage.” He specified now that the “we” in his letter to Watson meant “Frank and I.” The only other significant information he added was a hint of the methods he and Stahl were thinking about using to determine the identity of the units and subunits. “Perhaps we’ll try various hydrogen-bond breaking treatments followed by centrifuge examination . . . perhaps electron microscopy. We would very much welcome any suggestions you have.” 35 There is no indication here that Meselson and Stahl had yet devised a clear plan for tackling this critical question. After drafting his letter to Watson, Meselson noticed that the equal division of the nitrogen of the DNA unit into two subunits was informative not only about their mode of replication but also about their internal organization. In a marginal afterthought, he added, “so the thymines are not highly concentrated in one subunit.” In his letter to Levinthal he incorporated this idea into his main text and stated it more fully: “(incidentally, the existence of only one intermediate band shows that the nitrogen-poor thymines are not highly concentrated in one sub-unit).” 36 He did not need to elaborate his reasoning fully in writing, because the explanation would be evident to Watson and Levinthal. Thymine is “deficient” in nitrogen relative to the other three bases, because it contains only the two ring nitrogens, whereas cytosine has three nitrogen atoms, and adenine and guanine each have four. If thymine were concentrated heavily in one of the subunits, therefore, less of the heavy isotope would enter that subunit than would enter the subunit containing greater amounts of the nitrogenrich bases. Two types of hybrid molecules of differing overall density could then form, which would, if the difference were great enough to be resolved in the centrifuge, have produced two hybrid bands (or what would look like a broadened hybrid band).37 The judgment that Meselson expressed in these letters that the intermediate band “seems to be just in the middle” was significant, because that was where it ought to be if this band contained DNA that was exactly half labeled. Because the three bands did not appear together on any single film, however, the accuracy of this conclusion was not directly observable. To show more clearly that the intermediate band was exactly in the middle, Meselson ran the next day (9 November) a mixture of samples 2-I[-A] and 2-II-E from the second transfer experiment (that is, containing DNA from just before the switch in

334

A N E XTREMELY B EAUTIFUL E XPERIMENT

Fig. 10.6. Film from centrifuge run C-63, exposure 5

medium and DNA at the end of the second generation). The resulting film verified, as far as visual inspection could tell, that the band really was halfway between the other two bands (figure 10.6).38 On 10 November, Meselson described the transfer experiment for the third time in a letter, this one to Paul and Helga Doty. The language was similar to that of the first two, except that he simplified the description of the experiment and added that “the density difference between the two [varieties of DNA] is just what can be calculated under the assumption that 14.3 per cent of the mass of the molecule is due to nitrogen.” This agreement indicated to him that there must be little or no solvation of DNA molecules. Once again Meselson sought advice about identifying the units and subunits: “We can’t at all be sure that the molecules are WC duplexes. Even assuming that they are, we are ignorant of the nature of the sub-units: they need not be single polynucleotide strands. The identification of the molecules and the sub-units is a chemical problem on which we’d very much appreciate your advice. In my letter to Jim, there are some phantasies about chromosome structures; I hope you’ll ask him to show it to you.” 39 The last sentence suggests that, however much thought Meselson

A N E XTREMELY B EAUTIFUL E XPERIMENT

335

may have given to possible chromosome structures, he did not want his scientific friends to believe that he took these ideas too seriously, even as he invited further comment about them. In the rest of his letter Meselson reported on the results carried out in September with the tobacco mosaic virus that he had brought home after his stay with Doty in Cambridge. He ended by declining an offer Doty had made to him of an instructorship at Harvard: “From those very pleasant days this summer, I know what a stimulating place it would be for me. Yet there is a great deal of work, most of it involving equipment finally smooth-running at Cal Tech, which I want to do before starting to teach. I simply think that I couldn’t keep up enough momentum to teach, set up new facilities, and do research at the same time. Presently, therefore, I plan to remain in Pasadena this year and next.” 40 It is a measure of the freedom he felt at Caltech, and the optimism of the young scientists of that era, that Meselson did not hesitate to turn down an offer to become an instructor at Harvard. Those whom he consulted suggested to him that if he went he would become a slave to teaching. No one at Caltech told him what to do from day to day, and the thought of giving up such freedom frightened him. He had everything he needed for the next year or two and little worry about what he would do in the longer term.41 A few days after completing the centrifuge run with mixed samples from the transfer run that showed all three bands, Meselson made a composite photograph to show at a glance the changes in the bands as they appeared at the successive times in which the lysates had been taken. After making positive prints from the negatives, he cut out one exposure from each film, lined them up, and slid the individual strips back and forth until all the corresponding bands in successive pictures matched, then cut off the ends of the strips to make a uniform edge. Despite the absence of an independent determination of the density of the points in the solutions at which the bands formed, there was no sleight of hand here, because the structure of the experiment left no doubt of the identity, in each successive picture, of one of the bands contained in both. The cleanness of the bands, their clear-cut separation, the position of the light-heavy band midway between the heavy and light bands, and the immediacy with which the succession portrayed visually the temporal course of the experiment make it immediately obvious why those to whom it was shown found the photograph compelling (figure 10.7).

Fig. 10.7. Composite photograph of positives printed from films for centrifuge runs of the second transfer experiment with E. coli and 15N DNA. The films were reversed, so that the density increases here from left to right.

A N E XTREMELY B EAUTIFUL E XPERIMENT

337

Among the first to see this composite photograph was Linus Pauling. On the back of a copy given him by Meselson, Pauling wrote: “19 Nov. 1957. Duplication of nucleic acid- E. coli Matt Meselson N 15, then N 14 in CsCl solution.” Pauling undoubtedly understood immediately that Meselson and Stahl had carried out an “extraordinary experiment, which showed that the first-generation molecules of nucleic acid are newly synthesized to the extent of 50 percent, the other 50 percent being half-molecules of the parent nucleic acid.” 42 Meselson encountered Richard Feynman walking across the campus one day and showed him the results of the experiment. Sensing that this was a fundamental discovery, Feynman found the news “tremendously exciting and very important.” He urged Meselson to state his claim that the experiment confirmed the mode of replication predicted by the double helix, but even the enthusiasm of this brilliant physicist could not induce Meselson to abandon his reserve about the nature of the subunits.43

II Jim Watson probably did not write a detailed reply to Meselson’s letter, but he was excited enough about the result to call up various people who would be interested in it. Cy Levinthal, now at MIT, and therefore Watson’s neighbor, got the phone call from Jim before he received his own letter. Levinthal wrote Meselson on 18 November: “Again my congratulations. The experiment is extremely beautiful.” 44 The scientists associated with phage genetics and early molecular biology particularly admired beauty in experimentation. We may recall that Ole Maaløe called the Hershey-Chase experiment “a very beautiful piece of work.” In a discussion at the McCollum-Pratt symposium of Seymour Benzer’s experiments to establish fine structure genetic maps of T4, George Beadle agreed with others that Benzer’s was “a very beautiful work.” In 1957 Delbru¨ck described Levinthal’s experiment appearing to show that heterozygotes are a necessary intermediate in the formation of recombinant phage “a beautiful experiment.” 45 Levinthal’s response to the Meselson-Stahl transfer experiment was in keeping with the ethos of the field. His use of the superlative “extremely” suggests that he felt, however, that even within the class of work awarded the accolade “beautiful,” this experiment stood out. Levinthal’s high praise must have gratified Meselson and Stahl,

338

A N E XTREMELY B EAUTIFUL E XPERIMENT

because Levinthal was himself prominent among those who had been grappling with the DNA replication problem. He was no less impressed with the initial results from phage λ. “The multiple bands with Galducing lambda is also extremely exciting,” he wrote, “and should certainly be informative.” He proffered advice on how Meselson and Stahl could extend the study of replication to further generations: “One suggestion which could be useful to show the conservation of the subunits in prolonged growth would be to combine N 15 and P32 (at very low level) labeling to show that after prolonged growth when the intermediate density band is so diluted with the light that it cannot be seen optically the P32 is still all contained in it. How difficult is it to extract a band after the centrifuge has stopped running?” 46 Levinthal was evidently thinking about some composite of the Meselson-Stahl density gradient method and his own 32P emulsion methods. His idea did not much interest Meselson and Stahl. It would be easy to extract a band by running the experiment in a preparative centrifuge with swinging buckets, in which gravity would maintain the gradient after the run had stopped, but they were not convinced that carrying the replication out to “remote generations” would show anything different from what they had already found.47 Characteristically, Levinthal offered critiques along with his accolades, this one concerning Meselson’s estimate of the size of the DNA units and their correspondence with Taylor units: “In thinking about models I still think one should be careful about taking the weights of the stuff you obtain too seriously. Here (as possibly in the case of phage) you may be breaking a pre-existing structure either at a weak point or at a point determined by the configuration in the cell at the time of breakage. I don’t see any evidence yet that all the DNA in a Vicia Faba chromosome may not be held together by phosphate ester linkages, that is, be a single DNA molecule.” 48 Nor was Levinthal yet ready to concede that their earlier molecular weight determinations on T4 phage DNA had undermined his claims about the “big piece”: “As to the status of the big piece, I don’t know of anything which changes the conclusions in [my earlier] paper. Your M.W. determination adds a new fact, namely, that it can easily be broken into subunits which is certain to be important biologically but the big piece is a unique thing and the transfer experiments apply to it.” 49 Meselson and Stahl may have felt fortunate that their shift from transfer experiments on phage to those on bacterial DNA obviated for now the need to contest further with Levinthal the existence of his

A N E XTREMELY B EAUTIFUL E XPERIMENT

339

big and little pieces of phage DNA. Levinthal’s prescient caution that the molecular weights they had measured could turn out to be those of broken pieces of DNA molecules turned out to have more serious consequences for his own position than for their research program (see below, pp. 394–395).50 Around the time Meselson and Stahl were conducting the transfer experiments that Max Delbru¨ck described as “earthshaking discoveries,” Delbru¨ck accepted an invitation to give a lecture at MIT in association with a series of six lectures that the distinguished physicist Niels Bohr had been invited to deliver during the month of November on the subject of complementarity in quantum physics. On the day following Bohr’s fifth lecture, on the application of the complementarity principle to biology and psychology, Delbru¨ck was to present a seminar lecture on a related topic.51 Delbru¨ck accepted this invitation, even though it was offered on very short notice, because both Bohr and the theme of Bohr’s projected lectures were dear to his heart. Ever since spending a summer at his institute in Copenhagen in 1931, Delbru¨ck had been deeply taken with the principle of complementarity that Bohr had invented to deal with contradictory points of view that had merged recently in physics. The disturbing fact that electrons can be treated sometimes as particles and sometimes as waves, but that the two conceptions cannot be applied simultaneously, was elevated by Bohr’s principle to a general insight into the nature of science. Delbru¨ck’s pre-war shift from physics to biology was partly motivated by his desire to seek some specific form of the complementarity principle in biology.52 Delbru¨ck had clung steadfastly to this quest without notable success during his years as a phage biologist, and it underlay also his decision in 1953 to shift from phage to the light-sensitive fungus phycomyces. As he explained in a letter to Bohr in December 1954, his “ulterior motive” in this research, as “in biology from the beginning,” was his suspicion, “you might also say hope, that when this analysis is carried sufficiently far, it will run into a paradoxical situation analogous to that into which classical physics ran in its attempts to analyze atomic phenomena.” 53 Having submitted the title “Atomic Physics in 1910 and Molecular Biology in 1957” for his MIT lecture, Delbru¨ck was, as he wrote to Helga Harm on 5 November, “in desperate straights [sic] to put something under the title.” 54 So far he had not encountered in his phycomyces work the paradoxical situation for which he was looking. He

340

A N E XTREMELY B EAUTIFUL E XPERIMENT

decided, therefore, to illustrate his point of view with contemporary biological investigations from which he could draw implications about possible future paradoxes. One of the three cases he chose involved the transfer experiment just completed by Meselson and Stahl. The biology of 1957 could be compared with physics in 1910, Delbru¨ck asserted in his lecture, because the program of physics seemed “perfectly clear” then; yet within a year afterward it became evident, at first to a few persons, and then to many, “that there was something radically wrong . . . in classical physics.” Currently in biology “We are at the height of enthusiasm for building molecular models to interpret biological functions.” Wonderful techniques had been devised for the characterization of macromolecules. “The identity of the principal carrier of genetic information has been established,” its structure ascertained, “and this structure is such as to suggest simple molecular mechanisms for its replication, and for the transfer of this information to proteins. . . . The possibility for further advance seems unlimited, and he who has his doubts finds them to be ill-founded on every turn.” Nevertheless, the example of physics offered a warning: “Complementarity may be just around the corner!” 55 The first example Delbru¨ck invoked to illustrate his belief was the contractile mechanism of muscle as revealed by the recent electron microscope investigations of Hugh Huxley and Jean Hanson. Delbru¨ck had recently learned about this work while teaching his course in biophysics. The mechanism seemed to consist of a system of interdigitating rods that slide past each other—a “gross mechanical motion.” On the other hand, the energy needed to produce this motion came from a chemical reaction. The molecular dimensions of the mechanical and the chemical aspects of contraction were far apart. Perhaps a refined ultrastructure analysis could eventually bridge this gap and provide “a complete molecular picture of the process”; but it might turn out, Delbru¨ck thought, that the mechanical approach to the problem and the chemical are “mutually exclusive in principle.” In that case it “may be senseless to ask for an ultra-ultra-structure of active muscle.” 56 The second example Delbru¨ck discussed was the replication of DNA. In order to outline the problem, he described succinctly the “well-known structure” of the molecule, the sequence of the pairs of side groups, “which is random, from the chemical point of view,” and the current belief that this chemical randomness is “the vehicle in which genetic information is coded.” If the structure is to serve as a

A N E XTREMELY B EAUTIFUL E XPERIMENT

341

repository of genetic information, he continued, “this sequence must be copied accurately during each replication.” How, Delbru¨ck asked, does such a structure replicate? “The first question we may ask in this respect is how the atoms of the parental molecule are distributed. Do they entirely, or in part, reappear in the daughter molecules, and all in one, or half in one, half in the other? First attempts to answer these questions by means of various labelling experiments led to inconclusive results, but recently two highly significant partial answers have been obtained. The first, due to Taylor, Woods and Hughes, concerns the distribution of DNA between daughter chromosomes.” 57 Briefly describing Taylor’s experiments, Delbru¨ck stressed that, although they demonstrated that the DNA in chromosomes exists in two units which in replication “are segregated, one to one daughter, one to the other,” the result “does not say anything about the nature of this unit.” The other recent contribution Delbru¨ck discussed was “due to Meselson and Stahl.” After summarizing their experiment with his usual lucidity, Delbru¨ck gave his view of its meaning: These experiments prove that the DNA molecule consists of two units, equal in size, segregated to the two daughters on replication. They do not prove that these two units are the two chains of the Watson-Crick structure. In fact, one may have doubts. First of all, the “molecule” studied in these experiments may consist of two Watson-Crick duplexes. Second, even if each “molecule” is a single duplex one may have doubts that the units are the two single chains. These doubts are connected with the fact that the two chains in the Watson-Crick model are wound around each other, with a very large number of turns. Various mechanisms have been proposed to explain the separation of the molecule into two half molecules. One of these schemes involves numerous breaks and reunions between the chains and this is equally compatible with the results. Presumably this question will be resolved presently by a closer study of the distribution of N15 [sic] within the hybrid molecules.58 The “earth-shaking” nature of the Meselson-Stahl experiment had obviously so far failed to free Delbru¨ck from the old doubts about the unwinding of the many turns in the Watson-Crick structure that had induced him to propose his original dispersive scheme of DNA replication. Yet his position had been shaken enough so that he now entertained at once both of the alternatives to semiconservative replication:

342

A N E XTREMELY B EAUTIFUL E XPERIMENT

on one hand, that there were breaks and reunions during the process, and on the other, that the duplexes were conserved intact, replication being a separation of two duplexes contained in a four-stranded molecule. His comments can be interpreted as highly compressed allusions to the evolution of his own views since the publication of his first paper on the replication problem. The sentence referring to the various mechanisms that had been proposed cited his joint paper of 1956 with Gunther Stent. The scheme involving breaks and reunions “compatible with the results” appears, however, to be an implicit reference to the revised (and unpublished) scheme that he had proposed in one of his letters to Taylor in the spring of 1957 (see above, pp. 278– 279). His sentence concerning closer studies of the distribution of 15 N within hybrid molecules bears some resemblance to plans for further investigation of the nature of the “units” of DNA mentioned in the Meselson letters announcing the transfer experiment. There are enough parallels between Delbru¨ck’s brief discussion of these issues and the ideas Meselson expressed in his letter to Watson to suggest that both of them were influenced by the conversations between them that took place during the course of the experiment. Stahl’s 1992 recollection that “we actually considered the far-out possibility that the dispersive mechanism was true, but the points at which the chains break and join in each generation were precisely the same, from one generation to the next,” 59 resonates also with interchanges that must have taken place between them and Delbru¨ck, for this was the same possibility that Delbru¨ck had raised in his letter to Taylor. Delbru¨ck’s doubts about current efforts to describe the mechanism of DNA replication ran deeper, however, than the question of what the units that separated were. He went on: Together, the experiments of Taylor et al., and of Meselson and Stahl, prove that the Taylor unit in the chromosome contains half of each of the DNA molecules in the chromosome. This is then what we know now about the replication of DNA. It is far from telling us all about the molecular mechanism of replication. The principal thing that is missing is the origin of the new DNA material. What are the immediate precursors, and how are they incorporated? As to the precursors, we are still in the dark. On the one hand, there are strong hints of a direct tie-up with the synthesis of the other type of nucleic acid, RNA. On the other hand, there is strong evidence that one of the nucleotides, that having thymine as a side chain, generally does not pass through a ribose state. As

A N E XTREMELY B EAUTIFUL E XPERIMENT

343

to the manner of incorporation, we know still less. We can draw up paper schemes in which a synthesis occurs step-wise in zipperlike fashion, one nucleotide at a time, and we can persuade ourselves that such schemes might have sufficient specificity to ensure exact replication, or rather the synthesis of exactly complementary copies. There is no telling, though, whether these paper schemes are anywhere near the truth. As a biochemical problem, the situation is certainly quite unique, and must be characterized further before we can even tell what the nature of the problem is.60 Delbru¨ck explored these deeper doubts further in his third example, the relation of the DNA molecule to the genetic map. Although he regarded it as firmly established that DNA is the carrier of genetic information, it was “merely a suggestion derived from the structure of DNA” that this information is carried in the sequence of bases. He mentioned the evidence that mutations result from replacement of single amino acids and reflect “a similarly localized change in the gene” and summarized the investigations of Seymour Benzer indicating that a point on a genetic map “cannot be larger than a few unit links of the molecule.” Further evidence for the identification of mutations with the chemical structure of the DNA molecule came from studies showing that the mutability of local regions of the map can be increased by “a very slight chemical substitution” in which “a methyl group on one of the side chains is replaced by a bromine atom.” Delbru¨ck did not refer here to Meselson and Stahl’s still-incipient investigation of mutagenesis in phage with 5-bromouracil substituted DNA but to unpublished work on the same subject by Benzer and Ernst Freese. “This looks,” Delbru¨ck acknowledged, “very encouraging for the idea that points on the genetic map can be equated to points on the molecule. . . . Does it follow that the genetic map is a direct image of the sequence in the DNA molecule? I believe that this does not necessarily follow.” Even though classical genetic recombination experiments had long been consistent with the assumption that all genetic markers fit into a linear map, there were, when one reached the fine structure, a number of “oddities.” These oddities, which Delbru¨ck summarized, suggested to him that genetic mapping and molecular mapping may not be simple images of each other. Either the topology of the material carrier is not really a simple line: Branches, ladders, etc., may be involved, and the mechanism

344

A N E XTREMELY B EAUTIFUL E XPERIMENT

of recombination may be a little more involved than simple breaks and reunions, but the ultimate picture will still be a clearly visualizable tinker toy; or, the non-simplicity results from the fact that we are confronted with a true complementarity. . . . A rational account which includes both sides of the picture, chemical structure and genetic map, may turn out to be an abstract one, not a visualizable one.61 Delbru¨ck must have shown his MIT lecture to Matt Meselson before, or shortly after, delivering it. When Meselson read it he became concerned about Delbru¨ck. Much as he admired Delbru¨ck’s willingness to question currently fashionable views, this seemed to him to be “going beyond where Max normally questioned.” To doubt that the genetic map and the molecular map were collinear seemed to him an extreme form of pessimism. It meant that Delbru¨ck felt “nature was deliberately trying to deceive us.” He seemed to be expressing a kind of “Wagnerian gloom” that made Meselson feel that Delbru¨ck must be depressed. Other signs he noticed around this time reinforced Meselson’s worry that the MIT lecture was an expression of Delbru¨ck’s current state of mind rather than his long-held convictions.62 Meselson’s inability to understand why Delbru¨ck thought as he did reflects both a generation gap and the stern personality of Delbru¨ck, which prevented the younger man from seeing into his inner world. Far from dismayed at the prospect that molecular mechanisms would “in the limit” fall short of full biological explanations, Delbru¨ck ended his lecture with the comment that “it is pleasant to think that further progress may demand from us a deeper analysis of the scientific method itself.” 63 Even if Meselson had been more aware that such a viewpoint had been the leitmotiv of Delbru¨ck’s twenty-five-year sojourn in biology, he probably could have had little sympathy for it. Entering molecular biology under the influence of Linus Pauling, one of the most irrepressible enthusiasts for building molecular models to interpret biological functions, Meselson fully shared his mentor’s enthusiasm.64 That there could soon arise a crisis similar to that which had afflicted classical physics a half century earlier was remote from his experience and from his mood. Delbru¨ck’s lecture evoked at MIT an avid discussion lasting three hours, but he did not subsequently publish it. According to his biography it “was not greeted with enthusiasm by his friends.” Niels Bohr considered his expression of the complementarity principle in biology inadequate.65 Few biologists probably shared or even understood his

A N E XTREMELY B EAUTIFUL E XPERIMENT

345

viewpoint. To cast doubt on the direction of a rapidly moving scientific frontier places one in a lonely position, even when the person who does so is personally well-established in his field. In 1957 the way of the future seemed to molecular biologists to lie less with the vision of Max Delbru¨ck than with that of Francis Crick. In the September 1957 issue of Scientific American Crick reviewed for a wider audience the argument that nucleic acids “appear to carry the pattern of living matter from one generation to the next.” Crick acknowledged that much remained unknown, for example, about the structure of RNA, and that it was still an unsolved problem to learn how DNA reproduces itself. For the latter he proposed the broad hypothesis that “the two chains of the DNA, which fit together as a hand fits into a glove, are separated in some way and the hand then acts as a mold for the formation of a new glove while the glove acts as a mold for a new hand.” He, too, admitted that the experimental evidence supporting this idea—he cited particularly the work of Levinthal and of Taylor—did not tell us “whether the duplicating process operates at the level of the helical chains. Further, we still have the difficult problem of trying to imagine what sort of mechanism the chromosomes can employ to unwind the two DNA chains to free them for replication.” Far from inclining him toward Delbru¨ck’s type of skepticism that ultimate molecular mechanisms could fully explain biological functions, these problems were for Crick merely temporary obstacles. In a triumphant concluding paragraph he proclaimed: “From every point of view biology is getting nearer and nearer to the molecular level. Here in the realm of heredity we now find ourselves dealing with polymers, and reducing the decisive controls of life to a matter of the precise order in which monomers are arranged in a giant molecule.” 66 Our knowledge of later developments in molecular biology should not tempt us to treat the skeptics of 1957 as less perceptive than the enthusiasts. If few shared Delbru¨ck’s philosophical conviction that the field was heading toward a deep paradox, more were cautious about the length of the road still to be traveled. The cautions that Robert Sinsheimer had expressed in the lecture delivered at Caltech in November 1956 (see above, p. 215), were published as an article in Science in June 1957.67 Those who stood further outside the small group who constituted a nascent coalescing community of molecular biologists remained uncommitted. The second edition of the widely used textbook General Biochemistry, by Joseph S. Fruton and Sofia Simmonds, represented a view then common among biochemists when it

346

A N E XTREMELY B EAUTIFUL E XPERIMENT

stated that the “ ‘pairing’ of the nitrogenous bases in DNA preparations, together with X-ray diffraction data, have provided the basis for an ingenious speculation by Crick and Watson, who have proposed a helical structure for DNA.” 68

III During November, Ann Roller, the second woman to enter the graduate school at Caltech (the first being Elizabeth Bertani, the wife of Joe Bertani), became Meselson’s first graduate student. Highly recommended by Delbru¨ck, Roller came to Meselson for a research project and received from him the difficult assignment to study the problem on which he and Stahl had just given up: the replication of DNA in T4 phage. She accepted the challenge with a brave enthusiasm that she showed for every aspect of an environment in which she was nearly the only woman.69 In order to learn to run the ultracentrifuge, she assisted Meselson on some of the runs he carried out in November. With Roller, Meselson tested, on 15 November, a new, denser CsCl solution composed of 0.80 ml of the solution that had been his standard, together with 0.16 ml of CsCl of density 1.85 g/cm3. As a reference marker he ran the “pure heavy” (15 N) DNA from the second transfer experiment. The change in the solution moved the band from the lower third of the centrifuge cell to its midpoint. Apparently finding the new solution convenient, he used it in his subsequent runs.70 Following up what he had written Watson a week earlier—that what most interested him and Stahl just then was “to characterize the DNA of lytic lambda”—Meselson resumed experiments on phage λ on 20 November. Again with the help of Ann Roller, he centrifuged a mixture of “λ⫹ λ Galt .” 71 As in the run of 14 October with λ Gal, he observed four bands. This time, however, the second band from the heavy end was noticeably stronger than the others. On Saturday, 23 November, he centrifuged wild-type phage λ obtained by induced lysis and saw a sharp single band. On the twenty-seventh he centrifuged another strain of λ Gal⫹ and again found four bands, this time the densest one being stronger than the others.72 These results were interesting but puzzling to Weigle and Meselson. The discrete band they were able to attain for phage λ would enable them to calculate the density of the whole phage, and from that and the density of protein to estimate the proportion of protein and nucleic acid it contained. But the resolution of the transducing λ phage

A N E XTREMELY B EAUTIFUL E XPERIMENT

347

into four bands only opened questions for further study. Because the phage were intact, such a result did not mean that they were finding separate pieces of DNA but that there were, for some reason, several “density species” of the phage. How these differences in density might be connected with differences in the DNA content of the phage was not immediately obvious.73 Meselson broke off this nascent venture the next day to begin the centrifuge runs for a new 15 N transfer experiment. Perhaps by then Delbru¨ck was beginning to press him and Stahl to publish a paper on a result that had quickly impressed everyone who had heard about it informally. Even though the second of the two experiments already performed appeared entirely unproblematic, it would be important to show, before going into print, that it was repeatable. For transfer experiment number 3 Meselson used the new, denser CsCl solution. He modified the coding of the sample tubes slightly, using a number in place of a letter to designate the lysates taken from the bacterial culture at successive intervals. There is no record of the total number taken. On 28 November, the tube labeled 3-II-2 (equivalent to 2-II-B of the previous experiment) yielded two sharp bands representing the initial heavy and the hybrid DNA, the latter being more intense. The next day, the contents of tube 3-II-1 produced the same two bands, but the heavy one was darker. These were the expected results. Meselson had run them in reverse order from the temporal events, so that they recorded the forming hybrid band and the fading pure heavy band.74 During the next week Meselson did no further centrifuge runs, but Ann Roller spun a sample of tube 3-I-2, the lysate from the latest transfer experiment, taken just before the switch from 15 N to 14 N medium, mixed with DNA lysate from E. coli grown in a normal medium and shockate from normal phage T4. She obtained two strong, wellseparated bands, with a weaker band between them but closer to the “heaviest” band (figure 10.8).75 The heavier of the two dark bands represented 15 N coli DNA. The faint band represented 14 N coli DNA, the lighter dark band the 14 N T4 DNA. This first comparison by the density gradient method of the densities of phage T4 and bacterial DNA thus showed that the phage DNA was less dense. They might have predicted this result, because of the glucose molecules attached to this “sweet” DNA and because the A ⫹ T/G ⫹ C ratio of T4 DNA is greater than that of E. coli DNA.76 No further centrifuge runs were performed on transfer experiment

Fig. 10.8. Film from centrifuge run C-71, exposure 2

Fig. 10.9. Social life at the house on San Pasqual Street, fall 1957. Max Delbru¨ck and John Cairns at the piano, Manny Delbru¨ck in center background. Photo courtesy of John Cairns.

A N E XTREMELY B EAUTIFUL E XPERIMENT

349

Fig. 10.10. Party at the house to celebrate completion of the phage course. Left to right: Harry Rubin, Max Delbru¨ck, unidentified, Matt Meselson, Frank Stahl. Photo courtesy of John Cairns.

number 3. It is not clear whether they had been intended or whether Meselson and Stahl had deliberately produced only three lysates to check the equivalent runs from the previous experiment. In any case, the almost exact match between these partial results and the corresponding results from the more complete experiment must have enhanced the confidence they felt in the reliability of what they had found. On 8 December Meselson did one more experiment on a mixture of wild-type phage and transducing λ Gal⫹ and again saw four bands.77 He then tried out a new density gradient solution. In his letters to Watson and Levinthal, Meselson had written that they would “try to look at DNA just after DNA doubling in synchronized cultures in hopes of finding interesting intermediates.” They intended to “survey the entire density spectrum from protein to RNA so as not to miss any interesting structures.” 78 He gave only brief thought to the idea of synchronizing the replication of bacterial DNA, but, in keeping with the plan to survey the density spectrum, he looked in the critical tables for a salt solution whose density would be sufficiently greater than

350

A N E XTREMELY B EAUTIFUL E XPERIMENT

Fig. 10.11. A house “seminar.” Charles Steinberg explaining the experiment of Meselson and Stahl to Howard Temin. Photo courtesy of John Cairns.

that of a saturated CsCl solution (ρ ⫽ 1.923) to do experiments with RNA. A saturated cesium formate solution, he found, has a density of ρ ⫽ 2.3365 (the difference is due to the greater solubility of the formate).79 On 10 December he placed an E. coli lysate in a cesium formate solution of initial density (ρ ⫽ 2.32), diluted with enough water to reduce the solution to about ρ ⫽ 1.92. The films showed only a diffuse shadow at the extreme light end of the cell, suggesting that the concentration of the solution was a little too high to band the RNA. Repeating the experiment on 12 December with the solution adjusted to ρ ⫽ 1.80, he observed a single band, broader and more diffuse than those ordinarily obtained for DNA in cesium chloride.80 Assuming that this difference reflected a lower molecular weight for RNA, he or Stahl

A N E XTREMELY B EAUTIFUL E XPERIMENT

351

measured the density gradient across the band. Applying Stahl’s method of graphing the log of the intensities against the square of the distance from the center of the band, and obtaining a linear plot, Meselson inferred that the curve was Gaussian. From its slope he calculated that the molecular weight of the sodium salt of coli RNA was 460,000.81 After four relentless months of centrifuging, Meselson left Pasadena for a long cross-country holiday. The main purpose of his trip was to spend time in New York with Katherine, the flute player he had met in Aspen.

C HAPTER E LEVEN

Centrifugal Forces

I By the time Matt Meselson interrupted his work in mid-December, word of the spectacular experiment he and Frank Stahl had performed was already spreading through the community of scientists concerned with the biological role of DNA. In December, Paul Zamecnik invited Meselson to be a featured speaker at the Gordon Conference on Nucleic Acids and Proteins, to be held in New Hampshire the following June, and to talk on the DNA density transfer experiment. He was to share with Herbert Taylor the opening session on DNA synthesis, with Gunther Stent as commentator and Cy Levinthal as moderator.1 By this time also, scientists in other laboratories were beginning to inquire how they might apply the density gradient method to their own problems. On 27 December, for example, Salvador Luria wrote to Meselson for suggestions about how density gradient analysis could help him to separate particles of P1 transducing phage that appeared to lack part of the normal phage genome and to separate mutants of another phage that he suspected of containing “different amounts of bacterial DNA.” 2 To Frank Stahl, the ideas for further work on DNA replication that Meselson was announcing as what “Frank and I” planned to do were really all Meselson’s plans. When Meselson laid them out for him he said they sounded great, but privately he felt more than ever peripheral to the project. Nevertheless, he continued to help with the biological preparations and with the analysis of the data emanating from Meselson’s centrifuge runs. He also rejoiced in the achievement itself and took pride in the part he felt he had played.3 Stahl, who had the general task of preparing the annual report on work supported by Delbru¨ck’s polio foundation grant, wrote the account of their progress for January–December 1957. For the benefit of physicians not familiar

C ENTRIFUGAL F ORCES

353

with the field, he managed to include, in his succinct statement, a brief primer on the Watson-Crick DNA structure and its predicted mechanism of duplication, followed by an extremely condensed description of their experiment and its results: Mechanism of DNA replication in E coli (Meselson and Stahl) A DNA molecule consists of a pair of polynucleotide chains wound helically about a common axis. Each nucleotide on one chain is hydrogen bonded to the nucleotide at the same level on the other chain. Of the ten possible pairwise combinations of the four nucleotides, only the pairs adenine-thymine and guanine-cytosine occur, the others being prohibited by structural requirements. This self-complementarity suggested to Watson and Crick a mechanism by which the molecule might duplicate. According to this idea, the polynucleotide strands separate from one another so as to expose their specific hydrogen-bonding sites. Then, in accordance with the pairing restrictions, a new polynucleotide chain is formed along each of the parental chains. Accordingly, each daughter molecule contains one of the original parental chains paired with one new chain. Upon further replication, the two original parent chains remain intact, so that there will always be found two molecules, each with one parental chain. Employing the technique of density gradient centrifugation, we have observed the distribution of N 15 among molecules of bacterial DNA following the transfer of a fully N 15 labeled logarithmicallygrowing E. coli population to N 14 medium. When the cellular population has doubled, all DNA molecules are found to be half-labeled with the heavy isotope. Through subsequent generations, the number of such half-labeled molecules remains constant. These observations demonstrate that the nitrogen of a DNA molecule is divided equally between two sub-units and that upon replication each daughter molecule receives one sub-unit. The sub-unit remains intact through successive replications. It remains to be demonstrated whether the sub-units which are transferred intact to the daughter molecules are the polynucleotide chains of the Watson-Crick structure.4 It was not just for simplicity that Stahl described the transfer experiment only as a direct test of Watson and Crick’s original idea. Alternative mechanisms, including Delbru¨ck’s contribution to the discussion, had seemed to him irrelevant diversions.5 He took the same straightforward view several weeks later in a letter to Doermann: “The density separation method for DNA has been successfully applied to the study

354

C ENTRIFUGAL F ORCES

of DNA duplication in coli (we used N 15 rather than 5BU). . . . The result is exactly that predicted by Watson-Crick.” 6 Although careful to mention in the report that one could not yet identify the subunits conserved in the experiment with the Watson-Crick polynucleotides, Stahl was, as his letter to Doermann indicates, privately less resistant than Meselson was to “the temptation . . . to believe in pure, simple WC explanations.” Neither of them, in fact, really doubted that the units and subunits of their experiment were “single DNA duplexes” and polynucleotide strands,7 but whereas Stahl was inclined to accept the experiments already performed as complete, Meselson was more concerned to find other ways to strengthen their case. Reflecting on this situation while they were together in 1992, Meselson commented that “for Frank this [experiment] did a lot to settle the question, and I was still. . . Maybe he was less interested in mopping up.” Stahl then added, “Matt, like a bull, will have an answer that’s absolutely adequate in the minds of many people—one of the reasons he is a terrific scientist is because he realizes that he hasn’t yet clinched [the case] and he will hit it from another direction. I would have been satisfied long before Matt.” Meselson then interjected a more personal motivation: “Partly, maybe, it was the sadness of breaking off what was in a sense a dialogue, you really don’t want it to end.” 8 Colored though these recollections were by the passage of more than thirty years, by the esteem each continued to feel for the other, and by a certain nostalgia for the time about which they were reminiscing, these comments nevertheless seem to me to offer deep insight on the nature of their scientific partnership. Even as his role in the most successful of the several directions of the collaborative investigation he had initiated with Meselson more than a year earlier seemed to him to fade, Stahl took on more of the pursuit of the aspect of their work that had at the outset appeared most prominent: the study of mutagenesis in 5-bromouracil substituted bacteriophage. Meselson continued to discuss this work daily with Stahl and to contribute ideas, but as the pursuit of the transfer and other related centrifuge experiments claimed more and more of his attention, he left the implementation of the phage mutagenesis project almost entirely to his research partner.9 Having during the previous summer established techniques that assured full incorporation of 5-BU into the phage DNA (see above,

C ENTRIFUGAL F ORCES

355

p. 287), Stahl began in the fall to prepare the mutant stocks he would need for the mutagenesis project itself. When wild-type T4 infected E coli B under conditions favoring the incorporation of 5-BU into the progeny phage, 5 percent of the bacterial cells liberated at least one mutant phage with a mutant marker in the rII region of its genome (a mutation frequency twelve times as large as the frequency for unsubstituted controls). By this means he isolated “a large number of independently occurring mutations.” 10 With these stocks prepared, he was ready to embark on his main object: the study of the biological properties of these 5-BU substituted phages. While Meselson was away in December and January, Stahl concentrated on these experiments. On 28 January he wrote to Doermann: “We have collected a large number of induced rII’s in T4B. We have attempted to determine whether 5BU phage are genetically unstable (i.e. have an abnormal mutation rate) in normal medium. This was a good experiment and permits us to set a (rather low) upper limit to the mutation rate. However, the experiment was of a design which does not permit us to conclude that 5BU is totally ineffective as a mutagen when it is within the DNA; we are planning other experiments to probe further this issue.” 11 To test the stability of 5-BU substituted phage mutants in a normal medium had been a central object of the mutagenesis investigation from the time Meselson and Stahl first envisioned it in the fall of 1956. Meselson’s structural prediction that 5-bromouracil induced mutations by ionization and by shifting tautomeric forms more readily than thymine did had been the theoretical foundation for their plan to test the theory of Watson and Crick that mutations were due to tautomeric shifts that led to incorrect base pairing. If so, the base pairs in which 5-BU participated should be unstable, giving rise in the next replication by means of a “back mutation” to a different pair of “legal bases” (see above, pp. 184, 207). So faithful did Stahl remain to the research program projected by Meselson’s original analysis of these possible mechanisms that Charley Steinberg teasingly called it Stahl’s search for the Holy Grail.12 His pursuit of this goal did not consume all of Stahl’s time. Together with Steinberg, he also continued the theoretical analysis of the distribution of the clone sizes of mutant phages, based on the view that there is a steady-state pool of vegetative phage (see above, p. 95). In some respects Stahl felt that this was a more equal collaboration than the one he enjoyed with Meselson. Steinberg supplied more of the mathematical expertise, but he could not have done the analysis without Stahl’s understanding of the questions to be asked.13

356

C ENTRIFUGAL F ORCES

When Meselson told Jerry Vinograd that he had agreed to talk about the DNA transfer experiment at the Gordon Conference, Vinograd wrote Paul Zamecnik, the program chairman for the conference: “I think it would be appropriate at this meeting to have presented as well some of our other work on density gradient sedimentation. I have in mind a presentation of the general methodology with a summary of its applicability in studies of proteins, nucleic acids and viruses and perhaps as well a report on our density gradient work on RNA from microsomes. If you consider this material appropriate, I would appreciate your arranging for my attendance.” 14 Vinograd recommended also that Howard Dintzis, who was working with him on the last mentioned topic, be invited. The requested invitations were not issued. This letter not only fits Vinograd’s later view of himself as a cofounder of the density-gradient method but betrays the fact that he felt inadequately recognized for the role he had played, and unhappy that the attention focused on the transfer experiment overshadowed a broader methodological contribution in which he had come to view himself as a full partner (see above, p. 236). The jealousy he may have felt over the special recognition Meselson was receiving did not, however, prevent Vinograd from admiring his scientific talents. When Meselson asked him to fill out a recommendation for his NSF fellowship application, Vinograd supplied an appraisal that would surely have pleased its subject if he could have seen it then: It is a pleasure to recommend Mr. Meselson, whom I have known for three years. For the last six months we have worked together closely and have maintained almost daily contact. The candidate is of high quality in all attributes mentioned [on the NSF form]. . . . I regard him to be among the top 5% of the men I have known at this stage of professional development. Quite extraordinary are the following characteristics which I have come to know quite well in the last six months: a) He thinks out his problems energetically, thoroughly, and independently. He does not willingly accept ideas on authority. b) He performs experiments with attention to significant detail, and will put himself to much trouble to do things right rather than almost right. c) He is audacious in science. This manifests itself in an interest in important, though difficult, problems. d) He is energetic, mentally and experimentally. I regard him as a man likely to exercise leadership in his field.15

C ENTRIFUGAL F ORCES

357

Fig. 11.1. Matt Meselson at the reunion in Chicago, New Year’s Day 1958. Photo courtesy of Ernest Callenbach.

II While Meselson spent the Christmas holiday with Katherine in New York, rumors floated through Delbru¨ck’s group in Pasadena that he might return engaged. His immediate intention, however, was only to get to know Katherine better, and he left New York, as he later reported to Watson, with “no change in my customary marital status.” 16 From the East Coast, Meselson flew to Chicago to join a New Year’s reunion with six of his undergraduate classmates from the University of Chicago. The others included Ernest Callenbach, Jack Burgess, Horace Judson, Bruce Draper, Richard Sawyer, and James Powers (figure 11.1).

358

C ENTRIFUGAL F ORCES

While consuming together large quantities of steak and Jack Daniels, each told the story of his life since graduation. Believing that there would be a nuclear war, they worked out a contingency plan to make their way to a common meeting spot in the event that the United States was devastated by radiation. At breakfast on New Year’s Day, Meselson showed his friends the composite photograph made from the films of the DNA transfer experiment. Although none of the others was a scientist, they understood immediately the meaning of the experiment.17 Heading west, Meselson stopped off for a job interview with Arthur Kornberg at Washington University in Saint Louis. There was, at the time, much interest in Kornberg’s enzymatic synthesis of “DNA” from nucleotides of the four bases together with a DNA “primer” molecule. In a preliminary report of this synthesis at the McCollum-Pratt Symposium on the chemical basis of heredity in 1956, he had written, “We know relatively little about this enzyme reaction and the nature of the DNA product.” Since then he and his associates had made impressive further progress, but there was still controversy about whether the product was true DNA and whether it was synthesized de novo or as a “salvage process” from components of partially broken down molecules.18 Meselson was curious to find out whether Kornberg was “making [Watson-Crick] double helixes or something else.” 19 When he arrived, he was “much impressed with the present performance of Kornberg’s system. Howard Schachman [who had worked with Kornberg at Saint Louis during a sabbatical leave] has been there and has found that the synthesized DNA resembles the primer in regard to sedimentation constant and (more significantly) intrinsic viscosity.” Meselson discussed with Kornberg the possibility of using heavy nitrogen for the primer and “looking for half-heavy molecules in CsCl gradient.” He offered to make heavy T4 DNA for them.20 It must have been a gratifying experience for the younger scientist to find potential new applications of his method to the work of eminent colleagues. His stay in Saint Louis ended, however, rather lamely. Invited to dinner at the home of Arthur and Sylvy Kornberg, he was served several martinis in rapid succession and became so sick he had to lie down on their bed. They did not offer him a job.21 When he arrived back at Caltech during the third week of January, Meselson faced an acute dilemma. The great interest that the transfer experiment had already received made it more urgent than ever to complete and to publish the work, but first he was anxious to find out

C ENTRIFUGAL F ORCES

359

something more about the identity of the DNA units and subunits, and it seemed also essential to confirm his and Stahl’s splendid single result by repeating the full experiment at least once. He learned on his return that it appeared impossible in the near future to do either. Having tied up Vinograd’s ultracentrifuges for most of the past year, he found that Vinograd and his associate Howard Dintzis now wanted to get on with their work. In view of Vinograd’s letter to Zamecnik, about which Meselson knew nothing, we may surmise that Vinograd’s sense of having been left out now made him feel less generous with his machines than he had been a year earlier. Sinsheimer’s machine was also solidly booked. Worse still, having run out of cesium chloride, they were unable to replenish their supply. A supplier in Philadelphia had not replied to letters ordering more, and when Meselson tried a long distance call he found out that the supplier’s telephone had been disconnected. They found that the American Potash and Chemical Company could make the salt for them, but he could not expect it to be delivered in less than a month. A week later these problems were resolved. Perhaps Delbru¨ck, who was now pressing them hard to finish, was able to help expedite their short-term problems. Somehow the cesium chloride was quickly procured, and Meselson was able to schedule four consecutive days of centrifuge time, divided between the Sinsheimer and Vinograd machines, beginning on 25 January. For the long term Delbru¨ck was even more helpful. Deciding that the lengthy equilibrium runs required Meselson to have a machine of his own, Delbru¨ck arranged for a new one to be bought. The machine would be kept in the phage group, and with luck it could be set up and ready to run within two months.22 Delbru¨ck not only supported Meselson and Stahl strongly but viewed what they were doing as the most exciting current development in his group. In December, in a letter to the biochemist Otto Warburg, he had written: “At our institute, most of the actual research work centers on the implications of the discoveries of the structure of DNA and of its role in genetics. I enclose a reprint of a lecture which discusses some of the recent findings, and a preprint of a paper by Meselson, Stahl, and Vinograd concerning their fascinating new centrifuge technique.” 23 In January, he suggested to Ernst Freese that his wife Elizabeth might study “DNA behavior during meiosis, by the Meselson-Stahl method.” At Caltech itself, graduate students were increasingly interested in projects that would use the method. In addition to Edward

360

C ENTRIFUGAL F ORCES

Simon and Ann Roller, who were already at work, a Japanese graduate student, Noboru Sueoka, wanted to apply the method during the next year to study DNA replication in the unicellular green alga Chlamydomonas.24 In the midst of all this enthusiasm, Delbru¨ck had every incentive to ensure that Meselson and Stahl were not blocked in the continuation of their work. The new experiment, for which Meselson and Stahl must have begun growing bacteria in 15 N medium on 24 January, was designated “Transfer Exp. #5.” There is no record of a transfer experiment number 4. A fourth experiment may have been aborted before any centrifuge runs were needed. Meselson and Stahl repeated the transfer experiment not primarily to confirm a result whose validity no one doubted but to extend it. They made two small procedural changes. Adding four ribosides instead of two to the 15 N medium made no difference in the bacterial growth curve. Using the slightly denser “CsCl #2” solution they had adopted since the first successful experiment only moved the bands closer to the center of the cell. The most important advance they hoped to make was to carry out to four generations the bacterial duplications that they had previously followed only to the second generation past the shift from a 15 N to a 14 N medium. They were, in fact, sufficiently content with the results of the first experiment over the interval that it covered not to duplicate the five samples they had collected during the course of the first two generations. They merely took one sample (5.0) just before the shift in the medium and one (5.1) at exactly the end of the first generation. They collected three samples (5.2 to 5.4) between the second and fourth generations.25 What they would expect to find was that the “heavy light” band, which at the end of the first experiment was equal in intensity to the light band, should become progressively weaker in comparison to it. Because there would in each generation be only two hybrid molecules descended from each original molecule of heavy DNA, whereas the number of light molecules would continue to double at each generation, the ratio of hybrid to light DNA should be 1 :4 at the end of the third generation and 1 :8 at the end of the fourth. That Meselson and Stahl had fixed their attention primarily on the generations beyond those included in the previous experiment is further indicated by the fact that Meselson did not, as before, centrifuge the samples in the order in which they had collected them. He began, at 12:30 A.M. on Saturday, 25 January, with sample 5–2, the first of the

C ENTRIFUGAL F ORCES

361

Fig. 11.2. (a) Film from centrifuge run B-157, exposure 3 (transfer experiment 5–2); (b) film from centrifuge run B-158, exposure 3 (transfer experiment 5–3)

samples representing the continuation of bacterial replication past the point where the first experiment had stopped. The result nicely fit expectations. There were still two bands, but now the band representing heavy-light DNA was noticeably weaker than the band representing the light DNA.26 Continuing his night-shift schedule, and wasting as little as possible of his precious centrifuge time, Meselson began the next run, with sample 5–3, taken at the end of the third generation, at 3 A.M.. He finished shortly after noon on Saturday. Again the result fell perfectly into line—the heavy-light band had grown still weaker relative to the light band, which looked slightly stronger than at the previous stage (figure 11.2). Taking a few hours off on Sunday afternoon and evening, he did not begin his third run until 1:34 A.M. on Monday. By early Tuesday morning he could see that at the end of the fourth generation the heavy-light band had become very faint, and the light band was still stronger. All was in excellent order.27 So confident of the outcome of the experiment that he did not feel pressed to verify that the first two samples (5.0 and 5.1) confirmed the results of the earlier experiment, Meselson used his next available centrifuge time to pursue the vexing question of the identity of the

362

C ENTRIFUGAL F ORCES

units and subunits of the DNA forming the bands. Having received, in response to his inquiries to his colleagues, no novel suggestions about how to answer this question, he could only follow up on his own idea that a “hydrogen-bond breaking treatment” might provide some insight . Accordingly, he planned to subject some of the DNA samples collected in the experiment to heat denaturation, to see whether he could dissociate units into subunits. He would not have been assured of success. His heat denaturation experiments on salmon sperm DNA during the previous summer had produced a change of density with no apparent change in molecular weight. They had seemed to support the conclusion of Doty and his associates that heat denaturation caused some sort of collapse of the DNA molecule without separating the two strands composing the double helix. The similar experiments that Meselson and Stahl had performed on T4 DNA had, on the other hand, given evidence that its molecular weight was reduced by half when it was denatured (see above, pp. 286 ff.). As he now prepared to carry out the same type of experiment on E. coli DNA, Meselson’s curiosity about which pattern would apply to this crucial case must have been high. To obtain centrifuge time, Meselson had again to do the work on weekends, when the demand for the machines was low. For the weekend beginning Friday, 31 January, he was able to line up all three of the analytical ultracentrifuges on the Caltech campus, and he planned to crowd five runs into that time. In the first of them he leaped, in a single step, to test whether the hybrid DNA would dissociate. Sample 5.1, collected at the end of the first generation, ought to contain pure hybrid DNA. Even before running the sample by itself to see that it did, he prepared on Friday evening to centrifuge a mixture of sample 5.1 and of 5.1 heated for thirty minutes at 100°C. By 11:06 he had started the old Vinograd ultracentrifuge, which he set for the unusually high speed of 50,740 RPM. He risked the leakage that sometimes occurred at that speed, in order to increase the range of the density gradient in the cell. After he had taken a first set of photos at 11:12, he wrote down on the log, “I will be in 10 A.M. to take down run—unless I am telephoned . . . . to do so earlier. mm.” In the event, he returned an hour earlier than that to take more photographs, but the run went on.28 While this run continued, Meselson started a run with sample 5.1 alone just after noon on Saturday on the second Vinograd machine. The first run (containing heated and unheated 5.1) ended at midafternoon. The cell had leaked, but not enough to spoil the result. The

C ENTRIFUGAL F ORCES

363

Fig. 11.3. Film from centrifuge run 1476, exposure 3 (transfer experiment 5–1)

films showed three distinct bands, the “lightest” one being sharp and narrow, the other two being a little broader (figure 11.3). The result was highly suggestive. The sharp band appeared to be the unheated hybrid DNA, the other two the heavy and light portions into which the DNA had dissociated by heating. That these two were wider than the other was consistent with the lower molecular weight one would expect for the two dissociated strands. As in the runs made the previous summer with T4 DNA, the bands representing the heated bacterial DNA had formed sooner than the band from the unheated DNA, indicating that there had been a collapse in the structure of the molecule.29 On Sunday afternoon Meselson used the Sinsheimer machine to examine the effects of heat denaturation of the hybrid DNA, this time placing a sample of 5.1 heated for thirty minutes at 100°C alone into the cell. Two hours later he started, on the old Vinograd machine, a run with heated and unheated samples of the “pure heavy DNA” (5.0, taken at the time the medium was switched). While these runs were still in progress, he probably developed the film for the run with unheated sample 5.1 and observed a single, sharp, hybrid band.30 Meselson began the last run of his marathon weekend at 1:00 on Monday morning. In it he mixed together samples 5.0 and 5.4, representing pure heavy and pure light DNA. His objective was to establish density reference points for the other runs by getting both bands on the same film. By about 2:30 Monday morning the run begun on Sunday afternoon with the heat-treated hybrid DNA was over. Still having enough

364

C ENTRIFUGAL F ORCES

stamina to develop the films immediately afterward, he observed a very encouraging result. In place of the single band given by the unheated sample of this first generation DNA, the heated sample yielded two distinct, well-spaced bands, indicating that the hybrid unit had separated cleanly into a heavy and a light subunit. Moreover, the bands were again sufficiently broader than the single band from the unheated sample to indicate that the molecular weights of the heavy and light units were smaller than that of the hybrid unit. In these last two experiments he had, therefore, by carrying out separate runs on heated and unheated hybrid DNA, verified the interpretation of the first run, carried out with both of them together. At 3:05 A.M. Meselson took the last photographs on the centrifuge running the reference mixture of heavy and light DNA, closed down the machine, and walked home to sleep for whatever was left of the night. When Stahl later measured with the densitometer the intensity gradient across the width of the bands, he confirmed that the molecular weights of the denatured DNA molecules were just half the weight of the unheated hybrid molecules.31 The relentless pace at which Meselson ran the three ultracentrifuges over the weekend of 1–3 February was not due entirely to his difficulty getting time on them. Another reason to squeeze so much work into so little time was that he was about to leave for the East Coast to attend a Biophysical Society symposium at MIT, for which he had been invited to present a paper on the replication of DNA. James Watson had also asked him to give a seminar on the same topic at Harvard. He wanted to gather as much data as he could before then, to support his presentation and especially to have the heatdenaturation results in hand, so that he could discuss with Doty, Watson, Levinthal, and others their implications concerning the identity of the units and subunits. He did not have time to complete a critical control experiment, centrifuging a mixture of heated pure heavy and pure light DNA. If the two strands of the molecule did separate in denaturation, then a mixture of denatured heavy and light DNA should give exactly the same result as denatured hybrid DNA. He had not quite clinched his case, but he did have something to show his colleagues.32

IV Meselson flew from Los Angeles almost immediately after completing the last experiment on Monday morning to reach Cambridge in time

C ENTRIFUGAL F ORCES

365

for the seminar at Harvard on Tuesday. Paul Doty introduced him with what seemed such exaggerated compliments that the recipient felt somewhat embarrassed, but Meselson’s presentation was received with strong interest. The next day the symposium on microsomal particles and protein synthesis began at the Massachusetts Institute of Technology. It was the first meeting held by the newly formed Biophysical Society. The morning sessions on 5, 6, and 8 February were filled with the usual short papers, and the afternoon sessions were devoted to review papers that presented more general problems and progress in selected areas of the field. An afternoon highlight was the invited paper by George Palade titled “Microsomes and Ribonucleoprotein Particles.” Among the short morning papers was one by Meselson’s local colleagues, Howard Dintzis, Henry Borsook, and Jerome Vinograd. Density gradient centrifugation played an inconspicuous part in their presentation of the methods they had used to study microsomal structure and hemoglobin synthesis in the rabbit reticulocyte.33 It was a measure of the importance already attached to the experiment Meselson and Stahl had performed that the chairperson of the session at which he was to speak said that Meselson would be given thirty minutes instead of the usual fifteen. He thought to himself that he was not sure he could fill up that much time.34 During the meetings Meselson discussed the experiment with Herbert Taylor, who had already followed his and Stahl’s progress with interest. When Delbru¨ck had written Taylor about it at the beginning of November, just before the first success, he had replied, “The work with separation of DNA by the ultracentrifuge is most exciting. I shall naturally be interested to know how the experiment which you described with bacteria finally comes out.” 35 Now Taylor was “delighted to hear the details of the elegant experiments with N 15-labeled DNA.” He did not see how the units with which they were dealing could be “anything else” but “single double helices,” but Meselson maintained that they could not yet say.36 (Taylor’s somewhat confusing term was meant to contrast with the “double” double helices, or the fourstranded units that some, including Watson, Delbru¨ck, and Meselson, had entertained to get around the unwinding problem. His point would have been clearer if he had said he did not see how the subunits could be anything else but single polynucleotide strands.) Taylor had, in the meantime, continued his experiments on sister exchanges during the replication of Bellevalia chromosomes. By changing the time at which he put in colchicine, he was able to obtain frequencies of single and twin exchanges approaching the expected

366

C ENTRIFUGAL F ORCES

Fig. 11.4. Model for Taylor units drawn in a letter from J. H. Taylor to Max Delbru¨ck, 29 April 1958

1:1 ratio. In a long “bull session” at the Biophysical Society meeting, Taylor, Meselson, and Ernst Freese discussed Taylor’s structural model for the organization of DNA into chromosomes. Undoubtedly they considered ways to avoid both the asymmetry difficulty in Taylor’s centipede model and the single strand difficulty in Meselson’s variation. Taylor particularly liked a model, which he thought was “largely Ernst’s idea,” that he reproduced afterward in a letter to Delbru¨ck (figure 11.4).37 Meselson stayed at the house of Paul and Helga Doty through the week of these meetings, remaining until Saturday to take in a party in Cambridge. He was back in Pasadena in time to begin another centrifuging weekend on Valentine’s Day. He operated only one machine and worked at a comparatively leisurely pace, completing only three runs. On Friday night he returned to the Gal transducing phage and once again found four bands. This time there were two strong bands and two weak bands. The distribution of the “density species” of the phage thus appeared to vary markedly from experiment to experiment.

C ENTRIFUGAL F ORCES

367

After finishing this run on Saturday morning, he turned in the afternoon to the effect of heat denaturation on herring sperm DNA that Mike Litt had given him while he was in Cambridge. A mixture of unheated DNA and DNA heated for twenty-five minutes at 100°C yielded two bands, the “heavy” one being a little broader. On Sunday evening he repeated the run using heated herring sperm DNA alone and obtained a band corresponding to the “heavy” band of the preceding experiment. These results were generally similar to those Meselson had observed the previous summer with Doty’s salmon sperm DNA, but the somewhat wider “denatured” band might have left it in doubt whether the molecular weight remained unchanged as it had then. After closing down this last run on Monday afternoon, he had no further centrifuge time until the next weekend.38 On Saturday, 22 February, Meselson tested the effect of brief heating—for five minutes at 100°—on the hybrid E. coli. DNA (sample 5.1) from the recent transfer experiment. A single sharp band showed him that this treatment was not enough to alter the molecules. On Sunday he performed the control experiment for which he had not had time before traveling to Cambridge. He heated 15 N DNA (sample 5.0) and 14 N DNA (sample 5.4) each for thirty minutes and centrifuged them together. The two resulting bands were nearly identical to the bands yielded earlier by heated hybrid DNA (figure 11.5).39 The proof was now nearly as strong as he could make it that heating hybrid molecules of the bacterial DNA dissociated them into two molecular subunits of densities corresponding respectively to those of the subunits of 15 N and 14 N DNA. After this decisive run Meselson received no further centrifuge time for more than two weeks, and he performed only two more runs during the whole month of March. The most likely reason, in addition to the machines being in use by their owners and the prospect that he would soon have his own to work with, was that he and Stahl were now under sufficient pressure to get on with their paper on the replication experiments that it had become more urgent to analyze and interpret the data so far obtained than to gather further data. By the time Meselson returned from Cambridge, Delbru¨ck had grown impatient with his repeated assurances that they “were going to write it.” 40 In the meantime, Meselson and Stahl received the welcome news that, even in the absence of a publication, their experiment had made a strong impact in the place where the double helix itself had originated.

368

C ENTRIFUGAL F ORCES

Fig. 11.5. (a) Film from centrifuge run B-166, exposure 4 (heated hybrid E. coli DNA; (b) film from centrifuge run B-203, exposure 2 (heated 15N and 14N E. coli DNA)

From the Cavendish Laboratory in Cambridge, England, Sydney Brenner wrote on 16 February: Dear Matt, We were all very excited to hear about your wonderful experiment with the light and heavy DNA. As you say, perfect WatsonCrickery, and we eagerly await further news. Since hearing about your experiment, Francis [Crick] and I have been investigating a new coding scheme and have also been thinking about the nonrandomness of DNA sequences. As you know, the evidence suggests bunching of purines and we would be interested to know whether you could test the distribution of purines between the two chains by your method. Brenner then gave suggestions for carrying out such an experiment by growing bacteria in heavy purines and switching them to light ones. At the end of one generation there should be a single band if the purines were equally partitioned between the chains, otherwise “you should get two bands.” Similar experiments could be conducted with pyrimidines. In a handwritten sentence added afterward, he commented that the experiment “would help here” if the result turned out to be that there was “unequal partitioning.” They were also interested

C ENTRIFUGAL F ORCES

369

in “whether you intend to see whether the N 15 duplex is the same as the DNA duplex, or whether it is structure at a higher level,” and asked “Have you tried digesting the material with DNAase . . . ?” 41 Meselson was more cheered by Brenner and Crick’s enthusiasm than moved by their proposals. In view of the statement he had earlier made to Watson and to Levinthal—that the single intermediate band in the transfer experiment ruled out the possibility that the pyrimidine thymine could be concentrated in one of the chains (for the argument, see above, p. 333)—he could anticipate that the outcome of the experiments they suggested would not be helpful to them. They were also apparently unaware of the degree to which the Caltech group had already worried about the question of whether the “N 15 duplex,” by which Brenner apparently meant what Meselson and Stahl called the DNA “unit,” was a “DNA duplex” (that is, a double helix), and how hard it appeared to be to arrive at a decisive answer. The still-unpublished Meselson-Stahl experiment was not only exciting the foremost investigators in the fields directly concerned with the structure and function of DNA but was already beginning to be knitted into the fabric of that field. The earliest evidence of the rapidity with which the experiment was acquiring such status was that sometime in January or February, Christian B. Anfinsen, a biochemist at the National Institutes of Health, wrote to them requesting photographs of their experimental results. Anfinsen wished to incorporate a description of their work into a book he was writing titled The Molecular Basis of Evolution. A protein biochemist himself, Anfinsen was learning about the latest developments in other fields of chemistry and of genetics so that he could depict the “excitement and promise” of these fields for a fuller understanding of the processes underlying biological evolution. Even though he was not a member of the close network of molecular biologists and phage geneticists among whom the first news of the Meselson-Stahl experiment was circulated, Anfinsen must have learned quickly through informal communication that the experiment provided what he considered “the most impressive support” so far for the double-stranded, complementary structure of DNA. Meselson and Stahl did not immediately answer Anfinsen’s request, because they hoped they would soon have a completed manuscript about the experiment to send him.42 During February, Meselson once again had to summarize his research experience and his research plans in connection with the renewal of his fellowship. Under the first heading he compressed into

370

C ENTRIFUGAL F ORCES

three brief paragraphs a statement about his work on the crystal structure of N,N′-dimethylmalonamide, the collaborative development of the density gradient method, and its use to demonstrate that “a DNA molecule of E. Coli is composed of two sub-units and that upon duplication each daughter molecule receives one sub-unit.” About his plans he was also succinct but characteristically audacious. The further work on DNA replication and structure that he had outlined in the letters he had written during the fall now became only the first step in a more ambitious program: “During the course of the proposed fellowship I would continue to develop and apply the method of density gradient centrifugation. This work would include further investigations of the duplication of DNA, especially in its relationship to genetic recombination and chromosome structure. In addition, bacterial RNA would be investigated in order to determine its molecular weight and degree of heterogeneity. Experiments would be performed with this RNA to discover the mechanism of polymerization and the eventual fate of the macromolecule. The overall objective of the studies of RNA would be the isolation and characterization of the primary gene product.” 43 His apparent initial success in banding RNA in December had clearly made a large impact on Meselson’s new research objectives. If we compare this plan with the earlier plans for collaborative investigations that he had worked out with Stahl in the fall of 1956, however, we can also see a striking inversion of ends and means. Meselson and Stahl had begun with a set of theoretical problems, including the mechanism of mutagenesis and the replication of DNA, and contemplated bringing a variety of methods to bear on these problems. Meselson had, of course, from the beginning associated the replication problem with the innovative idea that he would follow the process by some kind of density separation method, but he did not know in detail how he would carry out the separations. Early in the course of their investigations he had devised a powerful method that enabled them to solve what was for him the most compelling of the problems they had originally taken up. Now the great promise of the method was beginning to direct Meselson’s choice of future problems. His auspicious success with the method, and the prospect that he would soon have available his own analytical centrifuge to use as often as he liked, made it natural that at this point in his career the new methodology became the driving force in his research. These expansive research plans remained for now in Meselson’s future, while he and Stahl struggled to finish up what they had already begun. When Stahl had finished plotting the quantitative density

C ENTRIFUGAL F ORCES

371

curves for the heat denaturation experiments (using the accurate but tedious old astronomical densitometer, because the automated one they had expected in November arrived with its wedges damaged),44 showing that the heated bacterial DNA was close to half the molecular weight of the unheated material, that meant, for the case of the hybrid band, which had separated into two bands, that “it is demonstrated that the two molecular sub-units are each physically continuous.” 45 Implied in this view was that the result eliminated Delbru¨ck’s replication scheme. Privately Meselson was convinced that these denaturation experiments left little doubt that the subunits were the two strands of the DNA double helix, but he still resisted going beyond what his data demanded. Delbru¨ck also remained unconverted. On 28 February he wrote to Watson, “in this respect the situation is still in a state of maximum confusion which may be characterized by the fact that Matt and I have a bet according to which I have to pay $10 if these subunits are single strands, and I will receive $10 if they are not.” 46 During March, Meselson and Stahl must finally have begun working seriously on their paper. By this time endless discussions about the exact meaning of the various parts of the experiment among the residents and visitors to the house on San Pasqual Street had already helped them to polish their understanding of the result they had reached.47 Stahl, who enjoyed the interpretation of data more than the actual performance of an experiment, carried through the laborious densitometer readings point by point from the bands and plotted them on graph paper. Drawing on his recognition from the previous spring that if the DNA in the bands was monodisperse—that is, homogeneous in molecular weight—then a plot of the square of the distance across the width of a band against the logarithm of the relative concentration of the DNA should be a straight line (see above, p. 246), he verified that the banded DNA of the E. coli transfer experiments was homogeneous. The molecular weight of the Cs 14 N DNA salt was 9.4 ⫻ 106. Meselson, meanwhile, began writing a draft of the paper. His progress was slow. He wrote and rewrote pieces of it, and sometimes he procrastinated.48 When several weeks had gone by since they had received the request from Anfinsen for photographs from their transfer experiment, Meselson and Stahl recognized that they could not keep him waiting until they were able to get their manuscript in final form. Meselson therefore sent him two photographs to use in his book. One was a series reproduced directly from the original negative, taken at twohour intervals during one of the runs made in early October on a mixture of 15 N and 14 N E. coli DNA. The sequence displayed the way in

372

C ENTRIFUGAL F ORCES

which a dark area consolidated and slowly narrowed, until two discrete bands appeared and grew sharper as the centrifuge cell approached equilibrium. The second was the composite photograph of the runs made during the first successful transfer experiment (see above, p. 336). Meselson included an explanation of the results the pictures represented. He mentioned that the experiment depicted had since been repeated, to four generations, and interpreted the overall results with the same caution evident in his earlier letters: From them, we can say that the nitrogen of a DNA molecule is equally divided between two sub-units. Upon duplication, each daughter molecule receives one of these. Sub-units are conserved throughout successive generations. Although this is in exact accord with Watson-Crickery, we cannot say from our work that the molecules we study are double helices or that the sub-units are single polynucleotide strands. We are trying hard now to settle these questions and have done one experiment (repeatedly) which says a little about the sub-units. Half labeled DNA obtained from coli cells one generation after transfer was heated to 100°C for 30 minutes in the CsCl centrifuging medium. This treatment results in the loss of the original half-labeled material and in the appearance in equal amounts of two density species upon heating. Heating, therefore, has resulted in the dissociation of the heavy subunits from the light ones. The molecular weight of the heated DNA (with or without N 15) is half of that of the unheated stuff. Thus it must be concluded that the sub-units which are conserved during each replication are physically continuous . . . not too unusual! 49 Perhaps Meselson was finding it hard to complete the manuscript because he still hoped to settle the questions about the subunits before he and Stahl committed themselves to publication. Even Delbru¨ck seemed to think that when “Matt is further along in characterizing the conditions under which the subunits stay together and under which conditions, if any, they come apart, undenatured,” 50 it might become clearer what the subunits were. Time, however, was now running against Meselson and Stahl. Still determined to leave Caltech as soon as he could find a suitable position elsewhere, Stahl had written to Doermann at the end of January, asking him to “suggest me to anyone who is hunting!” 51 He traveled to Yale and Columbia for job interviews. At both places he gave seminars on the DNA replication experiment, but feeling that the work belonged intellectually to Meselson, he gave in each case a short, fac-

C ENTRIFUGAL F ORCES

373

tual presentation with little interpretation. At Columbia he discussed the experiment in greater detail with Herbert Taylor, who seemed to him such a pleasant, easy-going southern gentleman that Stahl wondered how Taylor had been able to pursue the ambitious experimental program on chromosome replication that now appeared to complement so nicely their own experiment. Even though Stahl was convinced that Delbru¨ck’s ideas about DNA replication were wrong, he also resisted Taylor’s belief that the subunits must be single DNA strands, upholding the Caltech position that they did not yet know. His East Coast venture yielded no jobs, but during March his earlier trip to the University of Missouri culminated in the offer of an associate professorship.52 Delbru¨ck, who had the highest regard for Stahl as a scientist, tried to dissuade him from going. He hinted that if Stahl were willing to wait while he worked things out, he might be able to arrange a permanent position for him at Caltech. Stahl was adamant. Mary wanted to get away as soon as possible. After two years in Pasadena she still felt a stranger there. Stahl himself felt as strongly as ever that he must distance himself from Meselson’s dominant influence if he were to survive as an independent investigator. Not deterred by comments from others that it was crazy for a phage biologist to go from the mecca of phage research to an out-of-the way place in Missouri,53 he planned to depart by the beginning of May. On 28 March, Delbru¨ck wrote to Bob Edgar, “Frank is leaving, much to my chagrin, but there was no holding him.” 54 By the end of March, Meselson’s Spinco Model E Ultracentrifuge was installed across the hall from Delbru¨ck’s office, with its own darkroom nearby, and Delbru¨ck predicted that “Matt will be going ‘full’ blast, centrifuging away with the new machine.” 55 Meselson named his new machine “Daisy.” The experiments he carried out with it during its first months are not recorded, but one of the effects was apparently to divert him once again from finishing his manuscript on DNA replication. When nearly another month had gone by with no sign of its completion, Delbru¨ck decided to take more drastic action. In a letter to Ernst Freese on 28 April he remarked, “I am pushing [Matt and Frank] furiously to finish the N 15 MS, hoping to get it done in a very few days.” 56 Repeating a stratagem that he had used successfully with other young scientists in the past, including particularly Seymour Benzer, Delbru¨ck made them pack up their notebooks, their films, their drafts, and their typewriter, drove them to the marine biological

374

C ENTRIFUGAL F ORCES

station belonging to Caltech at Corona del Mar, put them in a room in a tower overlooking the beach, and told them to stay there until they were finished. Meselson and Stahl worked closely together for about three days at Corona del Mar. There was no one else around to distract them. Although Meselson’s vision shaped the text, Stahl’s critical judgment was equally essential. They “revised and revised.” Stahl concentrated on the analysis of the data and its representation in graphs and diagrams.57 The experience of being taken to Corona del Mar to overcome procrastination, and being required to write there in a few days a complete scientific paper, became a part of the phage group lore about Max Delbru¨ck. In his contribution to the Delbru¨ck Festschrift, Phage and the Origins of Molecular Biology, Seymour Benzer wrote, “The urge to do experiments was always so strong that we could not get ourselves to sit down and write up the results. Delbru¨ck had a solution for this. He assembled all who had papers to write and whisked us off to Cal Tech’s Marine Biology station at Corona del Mar. There we were locked up . . . and ordered to write . . . and in three days everyone had a completed paper.” 58 As with all shared community rituals, these stories have acquired some mythic overtones. Meselson and Stahl did work intently on their manuscript and data during the days they spent at the seaside station, and they completed a first draft; but they returned to Pasadena with much work on the paper still to do. By the second week in May they were struggling against more stringent time limits. Delbru¨ck was, as Meselson reported shortly afterward, “in a near fury driving us to finish writing in time for the May PNAS deadline.” To add further pressure, Stahl was by now packed and ready to leave Pasadena with his family. His departure schedule allowed him only a few more days to help Meselson meet Delbru¨ck’s ultimatum. In their rush to finish, Meselson felt afterward, “The part on heat denaturation and especially the associated figures have suffered.” They were finally through around 10 or 11 May. As Meselson wrote to Watson a few days later, “Frank and I finished this paper literally twenty minutes before the Stahl’s began their long drive to Columbus, MO.” 59 Recalling their work on the paper, Meselson said in 1987 that the thing he remembered most clearly is that I wanted to write the paper, not the way we approached the experiment, but the other way back. What I mean by that is that we did the experiment in order to distinguish between different models for DNA replication. But I wanted to put all that aside and say, what

C ENTRIFUGAL F ORCES

375

if all you knew was what was in the experiment[?] . . . So I spent a long time trying to write down sentences which would say in the [most] rigorous and briefest form, what it was that the experiment said; and finally got down to this notion that it says there is something which has two sub-units, which contain heavy nitrogen, and which come apart . . . and become associated with newly synthesized sub-units. . . . And of course in our minds we were convinced that it meant that [the] Watson and Crick [model] was right. But if you read the paper you can get the impression that we were distancing ourselves from Watson and Crick. [That was] because of this desire to rewrite the paper from the experimental results outward instead of from the possibilities toward the results.60 The two short introductory paragraphs of Meselson and Stahl’s paper “The Replication of DNA in Escherichia coli” reflect the thought and care that they had to put into it: Introduction. Studies of bacterial transformation and bacteriaphage [sic] infection strongly indicate that deoxyribonucleic acid (DNA) can carry and transmit hereditary information and can direct its own replication. Hypotheses for the mechanism of DNA replication differ in the predictions they make concerning the distribution among progeny molecules of atoms derived from parental molecules. Radioisotopic labels have been employed in experiments bearing on the distribution of parental atoms among progeny molecules in several organisms. We anticipated that a label which imparts to the DNA molecule an increased density might permit an analysis of this distribution by sedimentation techniques. To this end, a method was developed for the detection of small density differences among macromolecules. By use of this method, we have observed the distribution of the heavy nitrogen isotope N 15 among molecules of DNA following the transfer of a uniformly N 15-labeled, exponentially growing bacterial population to a growth medium containing the ordinary nitrogen isotope N 14.61 The brevity with which Meselson and Stahl summarized the problem to which they had addressed themselves and the means they had chosen to examine it was forced on them by the strict page limits for articles in the PNAS.62 But the clarity of their summary matched that of the experiment it was written to introduce. The distance from Watson and Crick that Meselson wanted to attain is evident. In contrast to Stahl’s progress report, which had placed the Watson-Crick structure

376

C ENTRIFUGAL F ORCES

prominently at the beginning, it is here not even mentioned. Meselson’s desire to write from the experimental results outward is reflected in the reduction of the theoretical background to two short sentences, followed immediately by a succinct description of the experimental strategy that he and Stahl had followed. To begin with the experimental results, and to restrict himself to what it was that the experiment said, was, of course, an ideal impossible for Meselson literally to realize. The original purpose of the investigation, to distinguish between different models for DNA replication, had been embedded in the very design of the experiment. By minimizing the discussion of this background, Meselson merely left implicit what in Stahl’s progress report had been made explicit. Whether acknowledged or not, the Watson-Crick model was the point of departure for all that was to be described in the paper. To try to write the paper “the other way back” from the way they had actually approached the experiment was a continuation of Meselson’s need to guard himself from the temptations of “pure Watson-Crickery.” The second introductory paragraph reduces the three years of reasoning and experience that Meselson and Stahl had traversed in reaching the experimental method presented in their paper to its barest essentials. That is what a well-constructed scientific research paper is expected to do. What it also does is to detach the outcome of the investigation from the rich local context of its origins. The complex historical pathway along which we have followed Meselson and Stahl has vanished from view. For scientific purposes the travail, the meandering, the tentative forays, the accidental circumstances, the shift from one organism to another and from one means of providing density differences to another, as well as the dramatic moments of insight and discovery, have all become irrelevant. The body of the paper was organized according to the usual conventions of scientific writing. The sections titled “Density-Gradient Centrifugation” (corresponding to the ordinary “Methods” section), “Experiment,” and “Results” were lucid and straightforward, with an economy of description reflecting the care with which they were written. Given the nature of the experiment, the direct visual presentation of the results was as critical to understanding and persuasion as was the text. Meselson and Stahl included three sets of photographs: a series showing the long approach to equilibrium of a single band of E. coli DNA during a forty-three-hour run; a single photograph of the “resolution of N 14 DNA from N 15 DNA,” the bands being shown beside

C ENTRIFUGAL F ORCES

377

the microdensitometer tracing made from them; and a composite of twelve photographs showing the successive formation of heavy, heavy-light, and light bands during the course of four generation times from the switch to 14 N medium. Here too they showed the bands beside the microdensitometer tracings (figure 11.6).63 As the numbers to the left of the photographs indicate, Meselson and Stahl combined the separate results of the experiment carried out in October and the one carried out during January and February to represent a composite result that might in principle have been derived from a single experiment. Because of the way in which the second experiment had been designed to complement the first, the results could be fitted nicely together so as to include all but one run of both experiments without duplication. The integration of the figure with the text was somewhat less complete, as the latter made reference only to the elapse of “two generation times.” 64 It appears as if Meselson wrote the “Results” section before he and Stahl decided to incorporate the second experiment into the paper and did not get around to revising that section. The small lapse is a hint of the time pressure under which the two collaborators were working. The “Discussion” section stated in the form of three propositions the conclusions that could “be drawn regarding DNA replication under the conditions of the present experiment.” 1. The nitrogen of a DNA molecule is divided equally between two subunits which remain intact through many generations. The observation that parental nitrogen is found only in half-labeled molecules at all times after the passage of one generation time demonstrates the existence in each DNA molecule of two subunits containing equal amounts of nitrogen. The finding that at the second generation half-labeled and unlabeled molecules are found in equal amounts shows that the number of surviving parental subunits is twice the number of parent molecules initially present. That is, the subunits are conserved. 2. Following replication, each daughter molecule has received one parental subunit. The finding that all DNA molecules are halflabeled one generation time after the addition of N 14 shows that each daughter molecule receives one parental subunit. If the parental subunits had segregated in any other way among the daughter molecules, there would have been found at the first generation some fully labeled and some unlabeled DNA molecules, representing those daughters which received two or no parental subunits, respectively.

Fig. 11.6. Composite photograph containing positives made from films of the transfer experiment conducted 23–30 October 1957 (here labeled “1”) and the transfer experiment conducted 24 January to 3 February 1958 (here labeled “2”), published in Meselson and Stahl, “Replication of DNA,” p. 675. On this copy of a reprint, Meselson and Stahl added, in 1996, the dates and centrifuge runs of the films from which the composite was constructed.

C ENTRIFUGAL F ORCES

379

3. The replicative act results in a molecular doubling. This statement is a corollary of conclusions 1 and 2 above, according to which each parent molecule passes on two subunits to progeny molecules and each progeny molecule receives just one parental subunit. It follows that each single molecular reproductive act results in a doubling of the number of molecules entering into that act.65 It is these three statements that Meselson had principally in mind when he recalled in 1987 that he had “spent a long time trying to write down sentences which would say in the most rigorous and briefest form, what it was that the experiment said.” That process had begun long before he and Stahl sat down at Corona del Mar to draft their paper for PNAS. In his letter to Watson immediately following the first 15 N transfer experiment, Meselson had already written that what it “formally says is that the nitrogen of the units which form a band is divided equally between two sub-units and that upon replication the sub-units separate from each other and become associated with new sub-units.” He had hoped then to determine more definitely the identity of the units and subunits (see above, pp. 328–329). The denaturation experiments were intended to meet this objective, and they did persuade Meselson and Stahl that the subunits of the DNA molecule defined by the transfer experiment were “single, continuous structures.” 66 Was that not enough to infer that they were the single polynucleotide strands composing a double helix? Some of those who learned about the density transfer experiment had little trouble reaching that conclusion. Meselson chose greater restraint. When he reminded himself to “forget all those preconceptions, ask what the data tell you, don’t listen to anything but the data,” 67 he was affirming that the situation had not changed substantially since November. They had not been able to collect the kind of data that they had then hoped might say for them what the subunits were. Having decided to stick with the noncommittal language of units and subunits, Meselson put the conclusion already reached into a form that was as rigorous and as compact as he could make it. In July 1992, Meselson, Stahl, and I discussed together an earlier draft of the preceding account in which I had attributed more of Meselson’s caution to the influence of Delbru¨ck than now appears to me justified. The question prompted a richly revealing conversation between them. Because of its intimate blend of memory, of reflection long afterward, and of the rapport that they have maintained over the

380

C ENTRIFUGAL F ORCES

years since then, the exchange is best quoted as it occurred, without attempting to integrate it directly into my narrative: FLH: Do you have any comments on this sentence about “Meselson’s resistance to the temptation, reinforced by Delbru¨ck’s insistence . . .”? FWS: I would say it’s more likely Matt’s approach to science in general, that, when he makes a conclusion he makes sure that he has ruled out all other conceivable possibilities. MM: Or else Frank would bring them up. I don’t think Max insisted on it at all. FLH: Well, I’m guessing here—obviously I didn’t talk to him. ... MM: Here’s my memory, it could be wrong. I think it was when we decided to start to try to write something—the time when we actually finished writing was long after we started—but when we started, I think I remember Frank and I having a conversation like, “Let’s try to put down on paper just what comes out of the experiment itself.” FWS: Absolutely. MM: Without anything else in the world, you just don’t know anything except what’s in the experiment, and write that down. You may not end up writing a paper that way, but only by writing that down will you understand what you’ve really done. And then it came right out of that—and we kept writing the language, and it had become more and more general, and so finally you got to sub-units. FWS: You couldn’t say single chains, etc. That’s correct. MM: Because our experiment didn’t—we didn’t see chains. That’s what I remember, and part of the feeling . . . was the same feeling that with thermodynamics . . . you look at it, and there’s nothing in it except what you put in it. . . . And it happened when we sat down and started to write. FWS: But do you think this was an approach to analysis that you adopted at that moment, or had it characterized your life previously? MM: I think it crystallized at that moment. I hadn’t really been a scientist until that moment. The X-ray crystallography was something that I was not able to grasp fully enough to, you know, do things with.68 It is difficult to locate the moment that Meselson was recalling within the detailed chronology established from the surviving records of their activity. Could it have taken place during the few days that

C ENTRIFUGAL F ORCES

381

elapsed between the completion of the first fully successful density transfer experiment on 30 October and the letter from Meselson to Watson on 8 November that already used the terms “units” and “subunits”? Or during one of the many conversations about it at the house? Was what seemed to Meselson thirty-five years later a “moment” in reality a more gradual development lasting several weeks or months? These difficulties in the integration of a memory into the fine structure of historical events do not deprive the memory of the deep insight it reveals concerning the meaning of those events. The conclusions that Meselson drew in the text were also represented schematically in a diagram (the published version of which was probably drawn for them by an artist at the Caltech Graphic Arts Unit). To “make sure that nobody [would] read anything into the picture that was not supported by the experiments,” they represented the units and subunits as simple rectangles that implied as little as possible about their identity.69 This visual representation of what the experiment said is both abstract and concrete, and for many readers it must have seemed simpler and more compelling than even the most carefully chosen words could be (figure 11.7).70 Neither the verbal nor the visual representation of their conclusion extended past the second generation of daughter cells. Even though they had taken the trouble to extend the experiment to four generations, the additional data were, in the end, superfluous to their argument, because there was logically no way that these results could change the conclusion drawn from the first two generations.71 According to the standard format for a scientific research paper, this one had now reached its conclusion and ought to have ended with a brief summary restatement of its major points. Instead there were interpolated two additional sections, somewhat like a long coda following the main developmental structure of a musical sonata. The first of these, a revision of the paragraphs Stahl had written for their polio foundation progress report a few months earlier, (see above, p. 353), was titled “The Watson-Crick Model.” It began, “A molecular structure for DNA has been proposed by Watson and Crick.” After a brief description of the structure, the section went on to say that the complementariness between the two chains suggested to Watson and Crick a definite and structurally plausible hypothesis for the duplication of the DNA molecule. According to this idea, the two chains separate, exposing the hydrogen-bonding

382

C ENTRIFUGAL F ORCES

Fig. 11.7. Schematic representation of conclusions concerning DNA replication, in Meselson and Stahl, “Replication of DNA,” p. 677

sites of the bases. Then, in accord with the base-pairing restrictions, each chain serves as a template for the synthesis of its complement. Accordingly, each daughter molecule contains one of the parental chains paired with a newly synthesized chain. The results of the present experiment are in exact accord with the expectations of the Watson-Crick model for DNA duplication. However, it must be emphasized that it has not been shown that the molecular subunits found in the present experiment are single polynucleotide chains or even that the DNA molecules studied here correspond to single DNA molecules possessing the structure proposed by Watson and Crick.72 Meselson and Stahl illustrated the Watson-Crick mechanism with another schematic diagram, probably drawn for them also at the Graphic Arts Unit (figure 11.8). The striking visual correspondence

C ENTRIFUGAL F ORCES

383

Fig. 11.8. “Illustration of the mechanism of DNA duplication proposed by Watson and Crick,” in Meselson and Stahl, “Replication of DNA,” p. 678

between this and the preceding diagram must for some readers have overshadowed Meselson and Stahl’s verbal caution against identifying their molecular subunits with single nucleotide chains. Following this disclaimer, Meselson and Stahl presented some information that they had been able to gather about the molecules and subunits. First, they had prepared highly purified DNA molecules from E. coli and determined its apparent molecular weight to be 7 ⫻ 106. To show that the banded DNA was monodisperse, they presented Stahl’s linear plot of the relation between the square of the distances across the band and the log of the relative concentrations of DNA.73 Their information about the subunits was contained in a section titled “Heat Denaturation.” Summarizing the evidence from Paul Doty’s laboratory that heating salmon sperm DNA at 100°C for thirty minutes

384

C ENTRIFUGAL F ORCES

causes it to collapse with no apparent change in its molecular weight, they reported that their own density gradient experiments “corroborated” these findings. In contrast, their experiments on the denaturation of bacterial DNA showed that “the apparent molecular weight of the heated bacterial DNA is reduced to approximately half that of the unheated material.” They included microdensitometer curves showing the resolution of hybrid DNA into two bands represented by a double peak in the curve, and compared these with two curves obtained from the heat-denaturated 14 N and 15 N DNA centrifuged together. The inference from the data to the conclusion that hybrid DNA separated into two subunits close to half the molecular weight of the intact molecule was less transparent than the relation between data and conclusions in the previous sections of the paper, because readers needed to understand the detailed description of the density gradient method for determining molecular weights contained in the authors’ earlier paper with Vinograd in order to see how the broader density curves of the denatured DNA translated into smaller molecular weights. From the “behavior” of the hybrid DNA on heating they drew the further conclusion that “the subunits of the DNA molecule are single, continuous structures. The scheme for DNA duplication proposed by Delbru¨ck is thereby ruled out.” They further complicated this section by ending it with two possible interpretations of the difference between the effects of heating salmon sperm DNA and heating E. coli DNA.74 In 1989 Meselson explained the reasoning behind the organization of the paper: “I think it was that we wanted to put ourselves first and Watson and Crick second—in the sense that we had done something that stood by itself. It was pure . . . I mean, you could have done this, without even knowing about Watson and Crick. It was self-contained, and we felt strongly about it. And we felt that the best way to explain it pedagogically was just not to let the reader even think about Watson and Crick until we’re ready to let him, . . . and then trot [Watson and Crick] out and say, look here, . . . its just the same.” 75 Meselson’s recollection of how he and Stahl had felt thirty-two years earlier rings true, both to the psychological situation in which they were placed and to the actual arrangement of the paper they wrote. The reason for their strong feeling is also easy to fathom. They had brought forth something that was, in its own way, as beautiful as the Watson-Crick structure. To present their work as a straightforward confirmation of the Watson-Crick model would be to subordinate a brilliant experimental achievement to a conceptual tour de force that was already exerting a hegemonic hold on their field.

C ENTRIFUGAL F ORCES

385

The strategy was effective. The elegance of the experiment is manifest in its presentation, its self-contained quality evident. To the reader alerted to the context within which it was written, however, the unresolved tension inherent in the “pedagogical” organization its authors chose is also clearly discernible. Perhaps someone might have done the experiment without knowledge of Watson and Crick, but Meselson and Stahl had undertaken it in the full knowledge of the Watson-Crick model and its prediction about DNA replication. Moreover, the cogency of their result to the community of scientists to whom it was addressed depended on its capacity to distinguish between the various replication schemes that had been evoked by the double helix. Meselson and Stahl themselves would have been happy to connect their result more tightly to that structure, if only they had been able to find a way to identify their subunits. They were caught in ambivalent circumstances that are reflected in the text of the paper. Despite the care with which they wrote and rewrote, the various sections comprising the paper do not maintain a uniform point of view. It is almost as though the two ways in which they might have written the entire paper—one as a self-contained experiment independent of Watson and Crick, the other as a response to the replication problem raised by Watson and Crick—compete within the overall structure of the paper. This tension is clearest in the final two sections. The first is titled “Conclusion,” the second “Summary,” but they can be seen instead as alternative endings, each of them capturing one of the two viewpoints that jostled for Meselson and Stahl’s allegiance: Conclusion—The structure for DNA proposed by Watson and Crick brought forth a number of proposals as to how such a molecule might replicate. These proposals make specific predictions concerning the distribution of parental atoms among progeny molecules. The results presented here give a detailed answer to the question of this distribution and simultaneously direct our attention to other problems whose solution must be the next step in progress toward a complete understanding of the molecular basis of DNA duplication. What are the molecular structures of the subunits of E. coli DNA which are passed on intact to each daughter molecule? What is the relationship of these subunits to each other in a DNA molecule? What is the mechanism of the synthesis and dissociation of the subunits in vivo? Summary—By means of density-gradient centrifugation, we have observed the distribution of N 15 among molecules of bacterial DNA following the transfer of a uniformly N 15-substituted exponentially

386

C ENTRIFUGAL F ORCES

growing E. coli population to N 14 medium. We find that the nitrogen of a DNA molecule is divided equally between two physically continuous subunits; that, following duplication, each daughter molecule receives one of these; and that the subunits are conserved through many duplications.76 Meselson handed the final manuscript to Delbru¨ck, who read it through, offered no changes, and communicated it on 14 May to the National Academy. There is no indication of whether Delbru¨ck accepted the conclusion that the heat-denaturation experiments had “ruled out” his replication scheme, but he did not object to the statement that they did. The guarded treatment of the subunits in the paper meant that Delbru¨ck had, at least, not yet lost his bet with Meselson over the question of what they were. When both the Stahls and the Meselson-Stahl paper had left Pasadena, Meselson wrote to Watson, “Frank’s departure was disappointing to all of us but especially to me. I’m sure the reasons for it were not altogether simple. Yet, it may be best for him to be essentially alone and under conditions in which his work will even more completely be a self-expression.” 77

VI The annual report on the activities of the Division of Biology at Caltech for the academic year 1957–1958 selected the density gradient method as one of four research projects of its faculty that were of sufficient general interest to describe in “a less technical way” in the introduction to the volume. “New experimental methods in science,” the description began, “often lead to rapid advances.” It was now clear that the density gradient method, like paper chromatography, “will be widely useful not only in biology but in other fields as well.” After a brief summary of the method itself, the report described at considerable length, and as “a specific example of how the method is used,” the “investigation made on the replication of deoxyribonucleic acid (DNA).” There is some irony in the way in which the replication investigation was represented here as an application of a previously invented general method, when, as we have seen, the method had in fact been the by-product of a previously initiated investigation of the replication of DNA. In any case, it is evident that the method and its most successful application were together considered to rank among the most important recent advances made within the Biology Division.78

C ENTRIFUGAL F ORCES

387

Fig. 11.9. Frank Stahl teaching at Cold Spring Harbor, summer 1958. Max Delbru¨ck is standing at right. Photo courtesy of the Archives, Cold Spring Harbor Laboratory.

Among the scientists to whom Meselson sent a mimeographed copy of the paper soon after its submission was Maurice Wilkins. Besides having played a subordinate role to Watson and Crick in the development of the original double helix model, Wilkins had been mainly responsible for the further refinement of the model to conform to more accurate X-ray coordinates. Wilkins replied, on 3 June, Thank you very much for your letter and the mss describing your elegant and definitive experiments. Doubtless you have had many opinions and comments on your work but may I give mine too? Its importance seems to lie in the fact that it studies replication on a molecular level and is therefore distinct from previous work. The only difficulty I can see is that the characterization of single molecules of DNA in solution may not be so clear as some of the workers in the field have maintained and I do not know therefore how certain you can be that you are studying single double-helical molecules. But this does not worry me much, and as a result of your experiments I personally begin to feel real confidence in the Watson and Crick duplication hypothesis.79 Stahl stopped for only one day in Columbus, Missouri, to leave off some of the family belongings, and then drove on to Cold Spring Harbor for the summer. While teaching a course there, he asked Jim Watson to give a historical talk on the double helix. Watson described the origins of the model, which he called “an interesting idea.” But “the important part,” he added, “is that its central feature has been tested” experimentally. “You can hear about that from Frank Stahl.” 80

C HAPTER T WELVE

The Subunits of Semiconservative Replication

I The immediate impact of the publication of the Meselson-Stahl experiment in PNAS in the summer of 1958 extended well beyond the circle of biologists already engaged with the replication of DNA. When Joseph Fruton read their paper, for example, he concluded that what he and his wife had treated in their biochemistry textbook as an ingenious speculation was a real thing. He remained skeptical that all DNA is in helical form, but he now regarded the complementary base pairing mechanism as fundamentally important.1 Carolyn Walch was one of eight biology majors who had graduated from Swarthmore College in the spring of 1958. While on an expedition in Panama that summer to participate in a study of tropical ecology, she received a letter from one of her classmates who had just read Meselson and Stahl’s paper. Informing her that, in spite of her remote location, she should know about the latest scientific news, her friend summarized the experiment in her letter. After drawing out for herself the diagrams illustrating the result of the experiment, Walch became too excited to sleep. She knew then that she would go into genetics instead of ecology.2 Within the group who had participated in or followed the debate over conservative, semiconservative, and dispersive replication, it was rapidly accepted that Meselson and Stahl’s experiment settled the question in favor of semiconservative replication but not that it immediately confirmed the Watson-Crick prediction that the molecule replicated by separation of the two nucleotide strands constituting a double helix. Meselson and Stahl themselves contributed to this ambiguity in a paper on the replication of DNA that they gave at a session titled “Replication and Recombination of Genetic Material” at the

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

389

Cold Spring Harbor Symposium held in June 1958. The overall theme of the symposium that year was “Exchange of Genetic Material: Mechanisms and Consequences.” A few of the inside members of the phage group, such as James Watson and Salvador Luria, were present, but the participants also included classical geneticists and the eminent immunologist Macfarlane Burnet.3 The occasion was, therefore, auspicious for Meselson and Stahl to present their interpretation of their experiment to a broader biological community. Now Meselson and Stahl dropped the approach they had worked so long to establish in their PNAS paper, where they had presented the results of their experiment as far as possible independent of presuppositions. Beginning their second paper on the experiment with the paragraph (based in turn on Stahl’s progress report from the previous December) with which they had begun the section on the WatsonCrick model placed near the end of their PNAS paper (see above, p. 000), they transformed the role of their experiment into a “direct experimental test” of the Watson-Crick hypothesis. The statements that they had presented in the PNAS paper as the conclusions to be drawn from the experiment (see above, pp. 381–383) they now treated, in compressed form, as “striking predictions” that the Watson-Crick hypothesis makes concerning its outcome.4 After a brief summary of the experiment itself, Meselson and Stahl focused on the implications for the Watson-Crick hypothesis. Again presenting (this time on the same page, where the comparison was more striking) the two diagrams from the PNAS paper, they stated that the “results of the N15 transfer experiment are in exact agreement with the predictions of Watson and Crick.” Again, however, they deflected the conclusion that this agreement constituted verification. “We wish to point out,” they added, “that our results support any hypothesis which embodies the following features: (1) The nitrogen of a DNA molecule is equally divided between two physically continuous subunits. (2) DNA molecules replicate by duplication. (3) Following duplication, each daughter contains one parental sub-unit. (4) Sub-units are conserved through many duplications.” 5 Although an “acceptable DNA duplication scheme must involve the established features of the Watson-Crick structure,” Meselson and Stahl continued, there remained “unanswered questions regarding the structure of DNA whose answers may be of considerable relevance to the problem of replication.” Little was known about the ends of the structure, and other irregularities had not yet been ruled out. “End-

390

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

to-end or even side-by-side super structures may exist.” For the sake of definiteness, they added, “it could be supposed that DNA in coli (and perhaps in other creatures) exists as a pair of Watson-Crick molecules associated laterally, or perhaps end-to-end, and that it is these intact Watson-Crick molecules which are the subunits in the N15 transfer experiments. . . . So far as we can see, however, such models can be accepted only at the expense of abandoning the reasonable and detailed molecular mechanism for the specific synthesis of new DNA offered by the Watson-Crick hypothesis.” 6 For an audience composed of molecular biologists and geneticists, Meselson and Stahl took the occasion also to draw attention to the connection between their result and the experiments of Herbert Taylor on Vicia root tip chromosomes: The DNA of E. coli and the chromosomes of Vicia are each composed of two sub-units which are assorted into progeny units according to the same set of rules. If we assume that the DNA of Vicia would behave in a DNA transfer experiment exactly as does that of coli, we can deduce the following relationship between DNA molecules and chromosomes: The two subunits of each DNA molecule segregate with different sub-units of the chromosome. The establishment of this rule tempts one to construct models of chromosomes with specified structural relationships between chromosomal and molecular sub-units. As we have seen, Meselson had himself been tempted into proposing such a model to several of his colleagues in the heady mood following the first successful transfer experiment. Now, however, tempered, perhaps, by the conversations in Boston with Taylor and Freese, in which several alternative models had come up, he and Stahl drew back from that attempt: “Very little effort in such a venture,” they averred, “brings one to the conclusion that there exists a number of reasonable possibilities.” Instead of describing any such models, they suggested that “chromosome transfer experiments employing protein-specific label would considerably diminish the number of possibilities.” 7 During the discussion of their paper, Charles Thomas, who had continued at Johns Hopkins the efforts begun at Caltech the previous summer to reconcile the DNA molecular weight determinations made by the density gradient centrifugation method with the large piece observed through radiographic methods by Levinthal, reported that his new work “only confirms the contradiction.” 8

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

391

Their discussion of structural models of the kind Taylor had originally put forth, and, more crucially, the interpretation of their own experiment, can be seen as manifestations of the scientific style that Stahl later attributed to Meselson: never to grasp at an experimental conclusion without giving full consideration to all the possible alternatives. Searching in 1996 for another way to express their shared attitude, Meselson called it “almost a kind of arrogance that Frank and I, I think, share, which is that, if we think something out, and we see that there is a logical possibility that it exists, we want to [keep it open]. . . . It wasn’t reticence, it was a certain amount of pedantic rigor—intellectual discipline with a touch of insolence, . . . keeping as close as possible to what nature allowed to be revealed of herself.” To their regret, a few of their colleagues took the rigor with which Meselson and Stahl restrained their conclusions about the units and subunits of DNA replicated as “obscurantist” 9 or, worse yet, as an opening to champion the alternative that they supposed “for the sake of definiteness” (and had once even entertained), that coli DNA exists as a pair of Watson-Crick molecules. At the Sloan-Kettering Institute for Cancer Research, Liebe Cavalieri took up the challenge to find out the nature of the molecular subunits of which Meselson and Stahl’s experiments had shown the DNA molecules of E. coli to be composed. It is not clear whether he took from Meselson and Stahl’s Cold Spring Harbor paper or arrived independently at the idea that the subunits might consist of “two strands, rather than the single polynucleotide chain necessitated by the Watson-Crick replication hypothesis.” 10 Repeating Meselson and Stahl’s procedures for the growth of E. coli bacteria and their centrifugation in a Model E analytical ultracentrifuge, Cavalieri and his two associates, Barbara Rosenberg and Joan Deutsch, observed a band identical with that obtained by Meselson and Stahl. (They obtained only a single band because they were producing only normal DNA in an ordinary medium. The band would be equivalent to the “light” band in the Meselson-Stahl experiment.) By Doty’s light-scattering method they found the molecular weight of this DNA to be 11 ⫻ 106. Heating the DNA in CsCl caused its molecular weight to drop to 5.6 ⫻ 106. Heating it in the presence of the proteolytic enzyme chymotrypsin or shaking it with a chloroform-octanol mixture dropped the weight to 2.6 ⫻ 106. On the basis of these measurements, associated measurements of the physical properties, and enzyme kinetics of the molecules, Cavalieri reported, “We deduce that the unag-

392

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

gregated, unit DNA molecule of E. coli is actually a dimer composed of two double helices, laterally bonded together; [and] that each double helix is conserved intact during cell division.” 11 In October, 1959—several weeks after Cavalieri’s conclusions were published in the first volume of Biochemical and Biophysical Research Communications—James Watson led a discussion titled “Current Concepts of the Fine Structure of Chromosomes and the Nature of the Coding Mechanism” at a Macy Conference on genetics at Princeton, New Jersey. Watson devoted nearly one-third of his presentation to a description of the Meselson-Stahl experiment. “The technique devised by Meselson and Stahl” he called “very elegant.” One explanation for their results, Watson showed, using the diagram taken from Meselson and Stahl’s paper, was the mechanism of DNA replication based on the separation of the strands of the double helix. One other possible interpretation of these experiments was that the hybrid DNA was a dimer, composed of two DNA molecules aggregated end to end. If these molecules split crosswise, “the results of this experiment would be the same as under the previous hypothesis.” Watson reported that Meselson had since done a control experiment to settle this point. He had sonically vibrated hybrid DNA molecules. Such treatment should transect the molecules. If the dimer hypothesis were correct, some of the resulting fragments should contain 15 N, and some 14 N. The result, that all the molecules remained hybrid, ruled out the dimer hypothesis. (The experiments were actually carried out by one of Meselson’s early graduate students at Caltech, Ronald Rolfe. The further development of this investigation became the basis for Rolfe’s dissertation.) 12 Two of the participants at the conference, Sol Spiegelman and Joshua Lederberg, brought up Cavalieri’s recent report (because Cavalieri postulated that the dimers were connected side by side, rather than end to end, Rolfe’s sonication experiment did not necessarily apply to it), but Watson dismissed it. “I can think of no reason,” he commented, “why double-stranded DNA would pair.” 13 He summarized the current situation as he saw it: the structure of DNA is known at the micro level, in the sense that we know that a small unit will look like the complementary double helix. More difficulties arise at the macro level, particularly in connection with the structure of chromosomes and how DNA molecules are incorporated into them. I regard the general problem of DNA replication as solved. Some

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

393

might be more hesitant, but I think the evidence of the Meselson and Stahl experiments is conclusive. Ever since we have known what the structure is, there has always been doubt as to whether the strands could come apart easily. The intuition of some of us was that it might be difficult, and of others that it was just hopeless. Against this view that it was hopeless, there has always been Dr. Levinthal. [Watson was referring here to the “speedometer cable” model of DNA unwinding that Levinthal had proposed in 1956 to show the plausibility of the separation (see above, p. 107).] He seems to be vindicated, in that the Meselson and Stahl experiment proves that the strands do come apart.14 According to one of the co-discoverers of the double helix, therefore, the replication problem raised by the publication of their DNA structure in 1953 was now settled. The Meselson-Stahl experiment vindicated not only Levinthal but the confidence that Watson and Crick had maintained from the start that intuitive difficulties imagining how the two strands of the DNA molecule could be “untwiddled” were not sufficient grounds to abandon the genetic implications of the base-pairing mechanism built into their model. Later in the discussion, Levinthal himself returned to that question: “If the two chains of a very long piece of DNA are wound around each other, is there any way in which they can be unwound without getting into enormous difficulties?” Attempts to understand the motions involved in unwinding the DNA strands were, he pointed out, complicated by the “enormous mathematical difficulties involved in the hydrodynamic problem as well as the fact that our intuition does not apply to the type of motion involved.” When one tried to depict the problem by comparing it with macroscopical objects such as rubber tubes, one imagined that “inertia and gravity play the dominant role in determining the motion.” At the molecular level, the motion that will take place is, however, determined almost entirely by viscous forces. Explaining his unwinding model, Levinthal assured the discussants that the amount of torque required to rotate the strands of the helix with respect to one another was reasonable.15 Levinthal’s model could do no more than make a mechanism for separating the strands of the double helix reasonable. The most important point he made was that, in judging the issue, intuition failed. It had, in fact, been such an intuition that had led Max Delbru¨ck to pose

394

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

the replication problem in the first place. The primary psychological effect of the Meselson-Stahl experiment was to defeat objections based on intuition. It cleared the way for those who accepted its verdict to take up the more difficult problem: What is the real mechanism through which the semiconservative replication of DNA can take place? While the Meselson-Stahl experiment was winning the adherence of the scientific community, recent estimates of the molecular weight of DNA, including those made by the Meselson-Stahl-Vinograd density gradient method, were losing the confidence of that community. Levinthal’s concern that DNA became fragmented in the centrifuge cell, and the conflict between the finding of Meselson and Stahl that the DNA in T4 phage was of a uniform molecular weight with his own finding that each phage particle contained one big piece of DNA and several small pieces, had prompted him to have one of his associates at MIT examine the effects of shear forces on the degradation of phage T4 DNA. Measuring changes in the sedimentation constant as the indicator of changes in the molecular weights of the DNA, Peter Davison varied the degree of shear force applied to the molecules by varying the velocity with which he injected the DNA into the centrifuge cells with a hypodermic needle. His results demonstrated the relative fragility of the DNA. The most reasonable explanation, Davison asserted, was to “assume that the hydrodynamic shear is transecting the DNA double helix.” Because molecules should break roughly in half when a critical shear gradient is exerted, “a very long molecule should be halved repeatedly until a fairly uniform population of chains stable to the shear gradient is obtained.” 16 These conclusions implied that all the estimates made until then about the molecular weights of DNA molecules might have been measured on fragments of molecules. At the Macy conference discussion, Watson acknowledged that “we have been shaken by the observation of Davison that DNA is rather fragile.” The relative stability of DNA with respect to such factors as temperature had caused them to neglect the fact that it should not be shaken and left many questions about whether the “molecular weight distribution which we observe is artificial.” 17 Although all of the current methods were put in question, the density gradient method used by Meselson and Stahl was particularly susceptible to doubt. Because they had used hypodermic needles to inject the DNA into the centrifuge cells, the DNA they were analyzing in their system was not likely to be composed of the original intact

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

395

molecules (there were additional questions raised about the validity of their molecular weight analyses even of the pieces).18 This development also abolished Meselson and Stahl’s reasons for doubting the existence of Levinthal’s “big piece” but at the same time undermined Levinthal’s own evidence for the molecular weight of his big piece. Davison’s result led Levinthal instead to propose that “the DNA of the chromosome is all in one piece.” Even though “all of our intuitive feelings about molecules indicate that this preposterously long single molecule could not occur,” the new reasons to believe that all prior molecular weight determinations had been made on fragmented DNA molecules “made all the evidence to the contrary . . . very questionable.” 19 Within a year, Alfred Hershey and his associates learned how to extract “this preposterously long single molecule” from phage T2 and T4 without breaking it. They extracted the DNA in water-saturated phenol. At relatively high concentrations of DNA they could shake the extracted phage for thirty minutes with no appreciable breakage, whereas at lower concentrations “breakage is very rapid.” When they wished to study radioactive DNA present in small concentrations, therefore, they added “carrier DNA” to protect it. They also removed impurities by dialysis, and when they had to deal with lower concentrations they transferred them by pouring rather than by pipetting, to avoid the shear forces that Davison had shown to fracture the molecules.20 Then, together with I. Rubenstein and Charles Thomas, Hershey showed, using autoradiographic methods similar to those of Levinthal, that “the bacteriophage T2 contains a single molecule of DNA which accounts for virtually all of the phosphorus of the virus particle.” 21 Adopting Hershey’s methods for extracting DNA without fragmenting it, Levinthal, Davison, and their associates at MIT also reapplied their earlier autoradiographic methods and reported, simultaneously with Hershey’s group, that “all the DNA in each particle [of T2 bacteriophage] was present in one structural entity which might be identified as a DNA molecule.” Levinthal now attributed his coveted large pieces of T2 DNA, as well as the smaller pieces observed by Meselson, Stahl, and Vinograd, to degradation of the molecule through shearing.22 The discovery of the fragility of DNA molecules did not leave the Meselson-Stahl experiment itself untouched. The “heavy,” “hybrid,” and “light” DNA bands were now seen as comprised, not of whole molecules, but of fragments of uniform size produced in the handling

396

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

of the DNA. That realization did not diminish the significance of the experiment. Rather, it provided a needed explanation for its most prominent feature. The sharp hybrid band with density exactly halfway between that of the heavy parental DNA and the light progeny was the main source, not only of the immediate persuasive power, but also of the beauty perceived in the experiment. To explain its sharpness, however, required one to argue that the replication process occurred so rapidly that, at any given time, there was a very small probability of finding unfinished chains on a given piece of DNA or on any piece whose chains were part 15 N and part 14 N. But in 1960 Fred Abbo and Arthur Pardee confirmed that E. coli bacteria, growing in a culture in which their reproductive cycles are synchronized, synthesize DNA continuously.23 Now the sharp hybrid band became paradoxical. The following year R. G. Wake and R. L. Baldwin studied the in vitro replication of primer DNA by means of DNA polymerase (sometimes called the Kornberg enzyme). Using density gradient methods to separate the products, they attained results that strongly supported a semiconservative mechanism for replication. In contrast to the sharp hybrid band of the Meselson-Stahl experiment, however, they found “broad hybrid peaks.” To account for this “striking difference,” they applied an explanation based on the assumption that Meselson and Stahl’s procedures had fragmented the DNA molecules: The amount of DNA in an E. coli chromosome (3 ⫻ 109 in units of molecular weight) is roughly 400 times the size of the DNA molecules studied by Meselson & Stahl (7 ⫻ 106). According to Jacob and Wollman (1958) the E. coli chromosome contains a single linkage group and it is possible that it is a single DNA molecule. Consider the situation when such a DNA molecule is broken into 400 more or less equal pieces. Suppose that one strand is 14 N and that the other is part 14 N, part 15 N, covalently linked at a single point. The intact molecule would band at a density between that of [14 N] DNA and a hybrid which is half 14 N, half 15 N. However, after breakage all but one of the 400 pieces will be either completely 14 N or exactly half 14 N, half 15 N.24 If the beautiful hybrid band in the film sequence that so strikingly represented the result of the Meselson Stahl experiment was, therefore, in one sense an artifact produced by breakage, this revelation did not weaken the experiment in the eyes of contemporaries. Far from it. In the same paper in which they offered this explanation, Wake

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

397

and Baldwin referred to “the classical study by Meselson & Stahl of the replication of E. coli DNA in vivo.” 25 In a review of the mechanism of DNA replication in 1963, Saul Kit wrote, “The primary stumbling block to mechanisms positing Conservative Replication is the classic experiment of Meselson & Stahl which clearly shows that the parental DNA molecule is divided equally between two subunits which remain intact through many generations, and that following replication, each daughter molecule receives one parental subunit.” 26 Many landmark experiments in the history of science are eventually described as classic. To have attained that status less than five years after its publication makes the Meselson-Stahl experiment exceptional.

II Liebe Cavalieri did not give in easily. In 1961 he published in the Biophysical Journal a set of three papers that spelled out in detail the evidence on which he based his conclusion that the unit of DNA conserved during replication is a double helix. With heretical selfassurance, he summed up his position: “There is no evidence to support the unzippering of the double helix proposed by Watson and Crick, whereas we have presented here direct evidence . . . to show that the double helix is preserved intact.” To make his concepts concrete, Cavalieri presented a schema for the replication of “biunial” DNA molecules held together by protein links. To explain how the double helixes can reproduce conservatively, he drew on the scheme proposed by Bloch in 1955 (see above, p. 100), whereby the polynucleotide strands, stabilized by protein strands, could be rotated outward to form templates.27 In a review of the nucleic acids for Annual Review of Biochemistry in 1962, where he continued to maintain that he had shown that the conserved unit was the double helix, Cavalieri also revealed the motivation behind his view. Discussing the possibility that there is a single-stranded DNA template for RNA, he wrote, “then there must be mechanisms in vivo for untwisting the strands of native DNA. . . . When one also considers that there is protein associated with the DNA, the problem in mechanics appears to be intractable.” 28 In the same mode that Delbru¨ck had initiated a decade before him, Cavalieri was still proposing alternative DNA replication schemes because his mechanical intuitions could not countenance the unwinding of a very long double helix.

398

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

Cavalieri was not able to convert the scientific community to his position, but he did manage to keep the question open at the beginning of the sixties. Even though Wake and Baldwin found strong support for a semiconservative mechanism for replication, they acknowledged in 1962 that “conservative mechanisms for replication should be considered seriously in view of the work of Cavalieri and his coworkers.” 29 Kit commented in 1963 that “although the results of Meselson & Stahl have generally been identified with each of the single strands of the DNA double helices, Cavalieri and Rosenberg have proposed that they are, in fact double stranded.” Kit reported recent evidence in “excellent agreement with the Watson-Crick model,” and findings “in apparent contradiction to the Cavalieri model of replication,” but he did not declare the case closed.30 Meanwhile, another of Meselson’s graduate students, John Menninger, sought to answer the question whether the subunits conserved in E. coli replication were single or double polynucleotide strands by means of low-angle X-ray diffraction. The sonication experiments by Rolfe having ruled out the possibility of end-to-end attachment, Menninger planned to estimate the mass per unit length of the DNA to determine whether the molecule was two-stranded or four-stranded. When Meselson moved to Harvard in 1961, Menninger went with him. The values of mass per unit length that Menninger determined were “in agreement with the value predicted for a two polynucleotide strand helix.” 31 In an attempt to resolve his differences with Cavalieri, Meselson several times visited him at the Sloan-Kettering Institute but could come to no agreement with him. It seemed to Meselson coincidental that Cavalieri kept getting “just the right numbers to support four strands.” 32 Looking back on this controversy, John Cairns commented sympathetically in 1996 on the “resistance in people’s minds to the idea that the strands of long DNA molecules could ever unwind. Can you imagine a molecule 1mm long spinning inside a bacterium at 5000 rpm? Indeed we now know that there are special enzymes for unwinding and winding up DNA. It is important to emphasize that it was the unwinding problem that drove people like Cavalieri to question the nature of semiconservative replication.” 33 In 1963 Cairns himself arrived at a persuasive proof that the replication of DNA involves separation of the strands. His presence at Caltech just at the time Meselson and Stahl performed their decisive experiment in October 1957 was, Cairns recalled in 1965, “to determine

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

399

much of what I did later.” Meselson sent Cairns a copy of the manuscript of their PNAS paper as soon as it was finished in May 1958. Cairns quickly read the manuscript three times with great admiration. “You must feel continuously pleased about it,” he wrote to Meselson. The discussion of the difference between the effects of denaturation on salmon sperm and E. coli DNA led him, however, to make a rather unusual suggestion: that salmon sperm is haploid, whereas “bacteria may well be diploid.” Cairns wondered whether “(1) haploid cells . . . have their DNA in the form of a double helix (2) diploid cells, in the form of a pair of double-helices.” Drawing a little sketch to illustrate his suggestions, Cairns added a verse in the same spirit, but with the opposite thrust, as the one Meselson had put in his letters announcing the transfer experiment.34 Cairns did not pursue these ideas (when he saw a copy of his letter in 1993 he had forgotten about it), but he followed with interest the similar questions raised by Meselson and Stahl themselves, and by Cavalieri and others, about the identity of the subunits conserved in the Meselson-Stahl experiment. Before coming to Pasadena, Cairns had studied the kinetics of infection of embryonic tissue cells by the influenza virus. His main purpose at Caltech was to learn the techniques that Renato Dulbecco used to study animal viruses in tissue cultures. Dulbecco was away during most of the time he was there, but he learned how to culture cells from Dulbecco’s able collaborator, Marguerite Vogt. Cairns returned to Canberra determined to apply to some RNA-containing virus some of the techniques for analyzing bacteriophage that he had seen in use at Caltech. After a fruitless year spent trying to find a virus that would infect an amoeba, he returned to studies of the multiplication of a DNA-containing virus in animal tissue that he had interrupted when he left Australia. In place of influenza virus, he now chose the vaccinia virus. Cairns also used the technique of autoradiography to locate, within the cells that the virus infected, the sites at which they began to synthesize DNA. Growing mammalian cells on a surface on which autoradiographic film could be laid, he infected them with the virus, then placed them for a fixed interval in medium containing thymine or thymidine made radioactive by incorporation of the heavy hydrogen isotope tritium (3H). After various further intervals he fixed and stained the cells and exposed the film. Radiation from the tritiated thymine incorporated into DNA identified the places in the cells at which the virus had begun to synthesize DNA. Within a few hours of the infection, he found, several isolated areas within the cytoplasm

400

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

showed evidence of synthesis. Cairns concluded that each infecting virus particle begins a separate center of synthesis within the cytoplasm of a cell.35 The head of the Department of Microbiology in which Cairns worked at the Australian National University decided at about this time to groom Cairns for a tenured position in molecular genetics that he had obtained for his department and urged Cairns to spend his sabbatical leave in 1960 learning genetics in the laboratory of Joshua Lederberg. Not wanting to go abroad again, Cairns persuaded his chief to allow him to apply instead to work with Alfred Hershey at Cold Spring Harbor. Having heard that Hershey was a solitary person, Cairns rather hoped to be rejected, but Hershey returned his letter with a note at the bottom saying only “Sure, any time.” 36 When Cairns arrived at Cold Spring Harbor during the summer of 1960, Hershey and his postdoctoral students were developing the techniques described above to extract intact DNA molecules from T2 and T4 phage and E. coli bacteria. Because of Cairns’s seniority, however, he and Hershey agreed that he ought to work independently. Cairns devised a complicated transfer experiment using 5-BU and 32 P, which he thought would enable him to “observe the dispersal of parental DNA to progeny phage and the genotype of the progeny at the same time.” After doing a few preliminary experiments, he explained his plan to Bob Edgar, who had just finished teaching the summer phage course. Edgar improvised a little square dance rendition of the experiment that quickly drove home to Cairns that “the experiment was ludicrously complicated and, worse, that I was really rejoicing in its complexity rather than considering its ultimate object (which was obscure).” 37 Trying to think of something else to do, Cairns suddenly realized that by combining the autoradiographic methods he had used in Australia with the methods that Hershey was developing to isolate intact DNA molecules, he might be able to determine the length of the molecules. The New England Nuclear Corporation had just begun to produce tritiated thymine and thymidine in which the label was confined to a single hydrogen atom at the 5 position on the pyrimidine ring. Their specific activity was so high that it seemed possible that he could use them to visualize individual DNA molecules.38 In order to incorporate the radioactive thymine into T2 phage, Cairns had to “engineer the situation so that phage would only be made if thymine was present and would then be made in an amount

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

401

which was proportional to the amount of thymine present.” He chose a thymine-dependent strain of E. coli bacteria, but even then he had to block an enzyme that enables these bacteria to synthesize thymine after they are infected with the phage. Using Hershey’s new method, he extracted highly labeled intact DNA without problems. It took several months, however, to develop methods to “fix this DNA in an extended state so that its contours could be followed by autoradiography.” Even the most satisfactory method of drawing out the fibers, which he discovered by accident, produced only “localized regions where the DNA was suitably extended.” Then it took two months of exposure of the autoradiographic film to the fixed molecules to produce films in which images of the molecules appeared as series of grains.39 The maximum length of the extended DNA molecules turned out to be slightly more than 50 µ. Calculating the molecular weight of a molecule of such length on the basis of the spacing of bases in the double helix model and the molecular weight of one base, Cairns arrived at a figure close enough to the accepted values for the molecular weight and phosphorus content of the molecule to postulate “an uncomplicated double helix as the form of the T2 DNA molecule.” 40 “In passing,” Cairns measured the suicide rate of the tritium labeled phage and calculated the number of “lethal” thymine H atoms per phage. To his surprise, the value turned out to be nearly the same as that for P atoms per phage in 32 P suicide. That result seemed to him extraordinary, because of the great difference in the energies of the electrons released in these two kinds of decay.41 His time in Hershey’s laboratory deeply impressed Cairns, confirming his view of Hershey as a man of few words and no pretense but of great intellectual honesty. Moreover, he witnessed at close range that Hershey was an exceedingly careful experimentalist. In a sense Hershey became his scientific model. “For those who were lucky enough to know him,” Cairns wrote recently, “the memory is of a giant figure watching over us all to make sure that we would get it right.” 42 Shortly after Meselson arrived in Cambridge to take up a position at Harvard, Cairns visited him there. Besides discussing their work, Cairns let it be known that he would like to find a position in the United States.43 On the way back to Australia, Cairns and his family stopped off at Caltech, where they stayed with the Delbru¨cks. One morning Delbru¨ck announced, with no previous warning, that they must get going, because Cairns was scheduled to give a seminar in ten

402

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

minutes. With no slides or preparation, Cairns described his recent results—the length of the T2 DNA molecule and the coincidence of the suicide rates. Strongly disliking the idea that electrons of such unlike energies could act so similarly, Delbru¨ck asked how many points Cairns had on his survival lines. Having learned how do deal with Delbru¨ck during his earlier stay at Caltech, Cairns simply asked the audience if there were any other questions.44 Soon after he had begun the work with T2 DNA, Cairns had “resolved that if all went well I should turn as quickly as possible to bacterial DNA.” If he were able to estimate the length of E. coli DNA, Cairns would be able to determine whether it, like T2 phage DNA, was two-stranded. Then he might settle the debate about whether the subunits conserved in the Meselson-Stahl experiment were single DNA strands or double helices. Knowledge that T2 DNA was twostranded did not solve the problem, because “the transfer of parental T2 DNA to the progeny is accompanied by so much fragmentation that only small sections of the recipient molecules are known to be hybrid.” 45 Before he could carry out this intention in Canberra, a paper published by Meselson and Jean Weigle on phage λ offered Cairns another opportunity to put together the two conditions necessary to resolve this dilemma. Following up the experiments on phage λ in the winter of 1957–58 that had shown the transducing version of the phage to exist in several “density species” (see above, pp. 346–347), Meselson and Weigle had shown by the fall of 1959 that each strain of transducing phage λ has a different density, which remains stable in successive generations. During this investigation Meselson did not carry out the density gradient experiments in the analytical ultracentrifuge but returned to a preparative ultracentrifuge equipped with “swinging bucket rotors,” like the one he and Stahl had used for their first experiments with T4 bacteriophage in October 1957. The phage in its CsCl solution he placed in a “lusteroid” tube. Piercing the bottom of the tube after the run and collecting the contents drop by drop, just as he had attempted then, but now with better control, he could recover the phages from each band separately. Assaying the phages contained in the successive drops and graphing the phage content against the drop number gave a curve of the density distribution of the phage corresponding to the curves made by a densitometer from the ultraviolet absorption films taken in a run with the analytical ultracentrifuge. Be-

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

403

cause a single phage particle can be assayed as an infective agent, the method could locate as little as one phage particle in a drop.46 Meselson and Weigle continued their study of the genetics of phage λ in 1960. Their main objective now was to decide between the two mechanisms of genetic recombination in phage that had earlier been proposed: copy choice or breakage and reunion. Again Meselson used density gradient methods. Phage recombination was customarily investigated by infecting bacteria with two strains of phage differing in their genetic markers. Meselson prepared one of the strains in ordinary medium, the other in a medium in which they would produce heavy DNA. During experiments carried out during the summer of 1960, while Weigle was away in Geneva, Meselson incorporated into the DNA, in addition to 15 N, a heavy isotope of carbon, 13 C, not then readily available in the West. His supply had come from the Soviet Union, through ties that Linus Pauling had established after being made a member of the Soviet Academy of Sciences. Again Meselson used the preparative centrifuge and the successive drop method to determine the density distribution of progeny phage.47 In some of their crosses, Meselson and Weigle found the phage resulting from a single growth cycle distributed in three discrete density modes. They inferred that these densities represented phage containing, respectively, parental, half-parental, and nonparental DNA. The first two they referred to as “conserved” and “semi-conserved” DNA. Conserved DNA was found only with high multiplicity infections, and they regarded it as preserved in phage that had infected the bacteria but had not replicated. “The finding of semiconserved phage indicates,” they concluded, “that the DNA complement of λ is equally divided between two subunits which may separate from one another and appear in progeny along with newly synthesized subunits.” 48 The existence of the conserved phage led them to infer that “the DNA complement of λ is a single structure capable of remaining intact throughout the processes of infection and maturation. . . . We are led to conclude that the entire DNA complement of λ is a single semiconservatively replicating structure. In this respect, the DNA of λ behaves as a single molecule possessing the Watson-Crick structure according to the scheme proposed by them.” 49 These passages confirmed that the semiconservative mode of replication demonstrated in E. coli in the original Meselson-Stahl experiment extended also to at least one type of bacteriophage. They reveal

404

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

also that Meselson had now abandoned his reluctance to identify the two subunits separated in the process firmly with the Watson-Crick structure. He did choose, however, to make this shift in his attitude remarkably inconspicuous. Not only did he avoid drawing attention to the change, but he buried it as a subordinate result of the study of genetic recombination. “Our main finding relevant to the mechanism of genetic recombination,” he and Weigle wrote, “is that discrete amounts of original parental DNA appear in recombinant phages. This suggests that recombination occurs by breakage of parental chromosomes followed by the reconstruction of genetically complete chromosomes from the fragments.” 50 Meselson’s greater confidence that the phage λ DNA molecule traced in these experiments behaved like a Watson-Crick structure did not constitute proof that they were not pairs of double helixes in the manner that Cavalieri contended. When Meselson and Weigle published their paper in April 1961, it was John Cairns who saw in this work the opportunity to clinch that case. “If this molecule can be shown to have the ratio of total length to total mass of two stranded DNA, then strand separation must occur.” The possibility of doing so had just been made easier by the fact that, using the new methods that did not break DNA molecules, Hershey and others had been able to extract the entire DNA of λ phage in a single, intact molecule.51 Extracting the DNA intact, collecting it on a microscope slide, and producing an autoradiographic image (figure 12.1) proved “uneventful”: “Though most labelled molecules were seen to be folded or tangled, some could be measured. These ranged up to 23 µ in length. This estimate of length, when combined with the reported value [by Hershey] of about 46 ⫻ 106 for the molecular weight gives a ratio of about 2 ⫻ 106 mol. wt./µ. This is the expected ratio for a DNA double helix in the B configuration.” 52 Cairns sent off to Nature a note that appeared in the issue of 30 June 1962. After summarizing his procedure and result, he concluded: “In short, λ-DNA is known to form hybrids on replication. It is shown here to have the ratio of total length to total mass of two-stranded DNA. Therefore, the replication of DNA must involve separation of the polynucleotide strands and the Watson-Crick model for DNA replication seems to be correct.” 53 Despite his own brief flirtation with ideas about “double-double helices,” Cairns was convinced that DNA replicated according to the Watson-Crick model long before delivering this “proof” of its correct-

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

405

Fig. 12.1. Autoradiograph of two molecules of phage DNA, from Cairns, “Replication of DNA,” p. 1274. Scale represents 10 ␮.

ness. In the introduction to his paper he described the WatsonCrick mechanism as a hypothesis well supported by the result of the Meselson-Stahl experiment, but “not beyond the realm of cavil. The DNA which divides between the two daughters might, so the argument goes, consist of a pair of double helices,” eliminating the need for unwinding and suggesting “a mechanism of replication that would be radically different from that proposed by Watson and Crick.” 54 According to his own view of the situation, therefore, Cairns had merely put an end to a cavil. In 1965 he wrote, “Any doubts about the nature of semiconservative replication were now settled.” 55 In 1993 he gave an even more modest appraisal of his achievement: “I was merely showing what people believed. There was no intellectual content to what I was doing. It was a tidying up operation.” 56 Showing that what people believe is correct is, however, what much of creative scientific investigation is about. Contemporaries judged Cairns to have contributed an elegant denouement to the original replication problem and another landmark in the formative decade of molecular biology. At the same time he was completing his experimental proof that the DNA in phage λ is double-stranded, Cairns performed his first experiments on what he regarded as the “ultimate subject for autoradiography,” the bacterial DNA molecule. In applying his autoradiographic method to estimate the length of the DNA of Escherichia coli he sought to answer the question whether the DNA of the bacteria is contained in a single, very long molecule, of molecular weight around 4 ⫻ 109, or in the much smaller pieces that recent molecular weight measurements had seemed to indicate. The uniformity of the lower molecular weights measured could be seen, since Davison’s discovery of the fra-

406

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

gility of DNA molecules, as artifacts produced by breakage into uniform fragments. To Cairns, the best arguments for a single linear structure were that the genome acts as a single linkage group and that the Meselson-Stahl experiment had shown that all of the DNA “replicates semiconservatively in an orderly fashion.” In the many dinner-table conversations at the house on San Pasqual Street just after the experiment had been performed, in October 1957, Cairns had been particularly struck by the implications of the fact that no “light” DNA band formed until after the “heavy” band had disappeared. This meant, Meselson had told him then, that “everything is replicated once before anything is replicated twice.” The knowledge that the estimates of the molecular weight of DNA before 1958 had been based on fragmented molecules now suggested that the explanation for that orderliness could be that in the intact bacteria all the DNA was contained in a single molecule along which the replication proceeded also in an orderly fashion.57 In order to extract the DNA unbroken from the bacteria, Cairns modified the procedures of Meselson and Stahl to minimize the shear forces and turbulence that might fracture the molecules. He lysed E. coli labeled with tritium by dialysis against Duponol in a medium containing 2 M sucrose to avoid exploding the bacteria, with phenolextracted T2 carrier DNA added to protect the bacterial DNA from fragmentation. After dialyzing the DNA, he collected it on a filter, dried it, placed the filter on a microscope slide, and covered it with autoradiographic film.58 The exposed film showed that the bacteria had lysed, but most of the DNA molecules had “not unravelled enough to display any features of its form.” There were, however, some that had untangled enough to display “a long unbroken thread.” The longest of these measured about 400µ. Because he could not identify the two ends with certainty, he reported this figure as the minimum length of the DNA molecule. It corresponded to a molecular weight of about 109, or about one fourth of the DNA content of a bacterial cell.59 To determine whether E. coli DNA was two-stranded by this method, Cairns needed to be able to estimate its actual, rather than its minimum, length. But the appearance of the autoradiographic pictures remained “extremely confusing.” Everything was “frightfully tangled.” That it should be so was not surprising, because the threads were so much longer than a bacterial cell that they must be very convoluted to fit inside the cell, and once extracted there was nothing but

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

407

Brownian movement, convection currents, and the slow movement of the meniscus that could serve to disentangle them. For a time he thought that he might have to count the number of autoradiographic grains per unit length along short stretches of the molecule. By getting in this way some idea of the quantity of DNA contained in a certain length he might be able to decide whether he was dealing with a Watson-Crick duplex or “something more complicated.” 60 While he struggled with this problem, Cairns and his family spent a summer holiday by the sea. Walking along the beach, he saw nylon fishing lines washed up on the shore and wondered how many he would have to look at before he found one that was not in a tangle. Then it dawned on him that there was a way to study the replication of the DNA molecules autoradiographically that was easier than searching until he found a molecule that happened to have become untangled. Because E. coli had recently been shown to synthesize DNA virtually continuously, exposing the organisms for limited time periods to media containing tritiated precursors of DNA would label whatever segments of the DNA molecule were engaged during that time in the synthesis of new DNA. After this “pulse-labelling,” he would not need to find completely untangled molecules but only short untangled lengths where he could examine whether single or double strands were being made.61 After determining that, when he transferred the bacteria from an ordinary medium to one containing tritiated thymidine, the label gets into the DNA so quickly that the error in timing a pulse would be very small, Cairns made autoradiographs of E. coli DNA prepared by lysing the bacteria after a three-minute pulse, after a three-minute pulse followed by fifteen minutes in a “cold” medium, and after a six-minute pulse followed by fifteen minutes in a cold medium. Only a few of the DNA molecules “chose” to untangle sufficiently for him to observe lengths of the molecule that were replicating. The DNA lysed immediately after the pulse labeling showed autoradiographic images of “two pieces lying in close association.” Those lysed fifteen minutes afterward showed pieces that “have moved apart and [that] can be photographed separately. It seems, therefore,” Cairns concluded, “that two labelled molecules are being formed in the region of replication.” 62 Because he had still not determined the total length of the E. coli DNA molecule, Cairns had not been able to determine directly that it was two-stranded. This he now inferred indirectly, by comparing the density of the autoradiographic grains along the observed lengths to

408

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

the density of grains in labeled T2 and phage. The density of the first being about half of the latter two, he assumed that the latter were labeled in twice as many strands and inferred that “the two molecules being created at the point of duplication [in the E. coli bacteria] must be each labelled in one strand.” 63 That outcome extended to bacteria Cairns’s “proof” that DNA replicates by separation of the strands of a Watson-Crick double helix, but this proof was no longer the central aim of his investigation. The decision to switch to pulse labeling allowed him to move beyond the question of whether the two strands of the double helix separate in semiconservative replication. Together with his ability to isolate intact the “single DNA molecule” constituting the bacterial genome, pulse labeling enabled him now to catch the chromosome in the act of replicating. This capacity enabled him to pose questions about how, and where on the DNA molecule, the two strands come apart. His study now became that of “the shape and form of replicating DNA.” 64 For this purpose Cairns undertook longer pulse labeling experiments. To observe the patterns they revealed, he had again to search out the relatively few molecules that had untangled sufficiently to permit interpretation. The details of his analysis would take us beyond the scope of the replication problem as we have followed it in this volume. It was, in fact, the beginning of an investigation of the detailed mechanism of DNA replication that continues to the present day. The conclusion that Cairns reached—that E. coli DNA comprises a single molecule forming a closed circle and that the replication takes place at a fork that advances around the circle—became germinal to these later investigations.65 Biologists who hesitated to accept James Watson’s view in 1959 that the Meselson-Stahl experiment had already settled the replication problem posed by the advent of the double helix were persuaded that Cairns’s demonstration ended the controversy in 1962. No proof is final, and there was still much opportunity to extend the conclusion to eucaryotic organisms, but scientists in this field were ready by 1963 to accept that DNA had been shown to replicate as the Watson-Crick model predicted it would. Discussing in 1987 the paper in which he and Stahl presented the results of their transfer experiment in 1957, Meselson mused, I’ve often wondered, “What if we wrote the paper differently?” and said, “This is the structure of DNA. There are three possibilities

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

409

for how it replicates. Here is an experiment, here is the outcome, possibility number one is the right one. Period.” Then I think it would have been accepted just as well, and people would not have emphasized that it could have meant something else. It would have the same outcome, but . . . In a sense we raised doubts that weren’t very great in our mind, but maybe greater in other people’s minds.66 It was not pure hindsight that led Meselson to ponder alternate historical possibilities. Richard Feynman had advised him at the time to take this course. Meselson had rejected that recommendation but had expected others to view the reference to subunits instead of strands of the double helix as scientific rigor rather than true doubts about their identity. Most of them had, and many contemporaries quickly made the identification from which the paper had refrained. But the fact that one active participant in the field chose to interpret the subunits as unidentified, and to offer an identification different from the one intended, surprised and, to some degree, disappointed Meselson. Would it have made any difference if Meselson had chosen the other option? When he stated that the outcome would have been the same, he probably meant that one way or another the case would have been made for the semiconservative replication of DNA double helices. The difference for which he might have hoped would be to avoid the debate raised by Cavalieri’s attempt to prove that the subunits conserved were double helices rather than single polynucleotide strands. The question then becomes, Would Cavalieri have taken up this question if the paper of Meselson and Stahl had not implied that the identification of the units and subunits remained an open question? His first paper on the topic does suggest that he took their paper as an invitation to determine the chemical identities of entities so far defined only instrumentally. That Cavalieri described the mechanical problems involved in the unwinding of the double helix in 1961 as “intractable” suggests, however, that, even if Meselson and Stahl had been unequivocal, he might have resisted their conclusion. At a different level, Meselson’s question pertains to the historical place of the Meselson-Stahl experiment. Had they been less intent on stating only “what the experiment itself really said,” would their result have been viewed as resolving a problem that required instead four more years until John Cairns completed the case? Such feelings are inevitable and normal. Few individuals are so detached as not to

410

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

ask, when they have lived to see outcomes that they had not expected, what might have happened if they had, when they could have, made a move that would have altered that denouement. Nine years later, in 1996, Meselson reviewed these events again, but in a different mood. In a previous draft of the preceding paragraph I had referred to the aftermath of their decision not to identify their units and subunits as Watson-Crick helices and polynucleotide strands as a “source of regret.” He penciled in “glad of the rigor,” and maintained in our conversation about it that “I don’t regret it at all.” 67 In the end, in fact, Meselson and Stahl lost very little by their choice, and the field of molecular biology gained the elegant experiments through which John Cairns converted his admiration for the MeselsonStahl experiment into a worthy complement to what it had achieved. After reading the present chapter, Meselson himself summarized in such a way the relation between Cairns’s work and their own: “We had shown by density methods that DNA replicates semi-conservatively. John Cairns, more decisively [than Meselson’s own student John Menninger,] showed that the linear density was that of the DNA duplex not a DNA quadruplex. So putting those two things together, John closed the logical circle and that ended the debate.” 68 The debate had ended, even though the semiconservative replication of the DNA double helix had been demonstrated primarily on a single bacterial organism (with supporting evidence from two types of bacteriophage that would have been unconvincing without the initial result on E. coli). That few doubted that the conclusion would be found to apply to organisms in general illustrates a deeply held tradition in biological research. For many problems there is an organism of choice, best suited to reveal a phenomenon or process more difficult to discern in most other organisms. The Barbary ape used by Galen in antiquity as a stand-in for human anatomy, the frog in nineteenth-century physiology, and the fruit fly in classical-twentieth century genetics are only among the most prominent examples of the power of this principle throughout the history of the life sciences.69 Among the phage group it was assumed that bacteriophage would be the organism of choice for revealing the molecular mechanisms of genetics. For the demonstration of semiconservative replication, however, that prediction proved mistaken. The particular choice of T4, based on Stahl’s prior experience with that organism and Benzer’s analysis of its rII gene, proved especially ill-suited to the task. The exceptionally close link-

S UBUNITS OF S EMICONSERVATIVE R EPLICATION

411

age between replication and recombination greatly complicated the study of replication in T4 during this period, as has been noted repeatedly in this story, and it was this characteristic that provided the most intractable obstacle to Meselson and Stahl’s efforts to conduct the transfer experiments with T4. One knowledgeable reader of this manuscript has commented that the choice was particularly unlucky. If they had begun with phage λ, whose DNA replication is not as dependent on recombination as that of T4 is, Meselson and Stahl would have encountered less difficulty and could have succeeded (as Meselson and Weigle later did) in their original plan to demonstrate semiconservative replication with bacteriophage. Frank Stahl strongly disagrees with this assessment. “I don’t think,” he replied, “the choice of T4 was bad luck. If we had used lambda and obtained Meselson-Weigle type results, we would have published and never done the E. coli experiments, but the data would not have been as beautiful as were those from E. coli.” 70 This response not only illustrates the essential unpredictability of scientific research pathways but confirms that beauty in science is seldom produced purely by human design.

C HAPTER T HIRTEEN

Images of an Experiment

I In 1965 James Watson published a molecular biology book for introductory students. It would have been unwise to attempt such a book five years earlier, he wrote in the preface, but now that biology has a “sound basis,” it is “time to reorient our teaching and to produce new texts” that will give new rigor, perspective, and enthusiasm to biologists of the future.1 Based on lectures he had given at Harvard, Watson’s book Molecular Biology of the Gene transformed the events from which molecular biology had emerged into a pedagogically structured compendium of the field. After eight chapters on classical Mendelian genetics, cellular biology, and intermediary metabolism, Watson entered the heart of his subject in chapter 9, “The Replication and Genetic Organization of DNA.” In eleven pages he covered the evidence that the gene is “almost always” DNA and that “DNA is usually a double helix.” In a section titled “The Complementary Shape Immediately Suggests SelfReplication,” he summarized in simplified form the argument that he and Crick had given ever since 1953, illustrated with a schematic diagram (figure 13.1).2 The problems that Delbru¨ck and others had raised a decade earlier about this process Watson now waved aside. “No difficulty arises” from the need to break the hydrogen bonds. “Nor does the need to untwist the DNA molecule to separate the two intertwined strands ˚ ), and present a real problem. The DNA molecule is very thin (20 A rotation about its axis involves almost no energy.” 3 After explaining on theoretical grounds why base pairing should permit very accurate replication and summarizing experiments on the enzymatic synthesis of DNA (the experiments of Kornberg, although

I MAGES OF AN E XPERIMENT

413

Fig. 13.1. Schematic representation of replication of DNA, from Watson, Molecular Biology of the Gene, p. 267

Watson did not identify them as such) which proved that DNA replication does not involve synthesis of specific proteins, Watson came to the section on “solid evidence in favor of DNA strand separation.” 4 “To prove this point,” Watson stated, “methods had to be found to separate physically parental from daughter DNA molecules. This separation was first accomplished through the use of heavy isotopes such as N15.” In one succinct paragraph, he summarized the procedures followed in the experiment, again without identifying the authors. Then he compressed its result into four sentences: “If DNA replication involves strand separation, definite predictions can be made about the density of the DNA molecules found after various growth intervals in light medium. After one generation of growth, all the DNA

414

I MAGES OF AN E XPERIMENT

molecules should contain one heavy and one light strand and thus be of intermediate hybrid density. This result is exactly what is observed. Likewise, after two generations of growth, half the DNA molecules are light and half hybrid, just as strand separation predicts.” 5 To make the experiment and its outcome more concrete for the introductory students expected to read his text, Watson supplemented his brief verbal description with a schematic diagram (figure 13.2).6 With masterful clarity, attained in part by eliminating all the qualifications surrounding the historical events, Watson reduced the long story we have followed in this volume to a paradigmatic example of theoretical prediction confirmed by appropriate experimentation. The Meselson-Stahl experiment appeared here as the method that “had to be found” to prove the point that DNA strands separate in the replication of the double helix. Just as the text was kept short enough barely to mention the essential points, so the schematic diagram reduced the experiment to its barest features. A simple shape surrounding a double helix represented by two intertwined strands signified a bacterium containing its DNA molecule (in the original the heavy strand is shown in light brown, the light strand in black). In order to make the centrifuge easier for beginning students to visualize, the swinging buckets of an ordinary table-top centrifuge with which they were likely to be familiar was substituted for the sophisticated centrifuge cells of the analytical ultracentrifuge. “Heavy,” “light-heavy,” and “light” DNA were represented as though they could be seen directly within the centrifuge tube, rather than as bands on a film exposed to the ultraviolet light passing through a complex optical system. The statement that “the location of DNA molecules within the centrifuge cell can be determined by ultraviolet optics” only faintly qualified the visual impression that the locations can be seen in the tube itself. Watson presented this and other subjects in such simplified form not only because the book was aimed at undergraduates with little prior relevant knowledge but because he had struggled to present a comprehensive picture of the new biology within a book of moderate length. “Often,” he acknowledged, “I present a fact, and because of lack of space, I cannot outline the experiments that demonstrate its validity. Given the choice between deleting an important principle or giving an experimental detail, I am inclined to state the principle.” 7 In view of these constraints and priorities, Watson was singling out for prominent attention the experiment that demonstrated strand sep-

Fig. 13.2. Schematic representation of the Meselson-Stahl experiment from Watson, Molecular Biology of the Gene, p. 272

416

I MAGES OF AN E XPERIMENT

aration in the duplication of DNA. Nevertheless, we may question whether students who learned about the experiment through this influential text were exposed to the real Meselson-Stahl experiment or to a pallid stand-in. Just as Watson’s textbook as a whole served as a model for several generations of textbooks in the burgeoning field of molecular biology, so his treatment of the replication of DNA seems to have set the framework for discussions of the phenomenon in textbooks of biology, molecular biology, biochemistry, and genetics during the next three decades. Following a discussion of the double helix, the textbooks mention that the model predicts its own mode of replication, then bring in the Meselson-Stahl experiment as the critical experimental test of the theoretical prediction. There have been, however, nearly as many variations on this common theme as there have been textbooks, and special ingenuity has been shown in redesigning the ubiquitous schematic diagrams thought to make the experiment more readily understandable. In contrast to Watson’s original discussion, most of the subsequent discussions identify Meselson and Stahl as the authors of the experiment. An early example of these discussions is J. J. W. Baker and Garland Allen’s The Study of Biology, the first edition of which appeared in 1967. Under the section headed “Testing the Model,” Baker and Allen wrote, “One of the elegancies of the Watson-Crick hypothesized model of DNA replication is the way it immediately suggests a method of selfreplication.” After briefly describing the “unzippering” mechanism as a “mental picture,” they went on to say that “an elegant experimental test of the means of DNA replication was performed by two California Institute of Technology scientists, M. Meselson and F.W. Stahl.” Baker and Allen’s description of the experimental procedure was somewhat longer than Watson’s, mainly because they sought to include a brief explanation for the formation of the density gradient. In a curious twist apparently intended to make the situation easier for a student to grasp, they separated the Meselson-Stahl experiment into two stages, the first extending only over one bacterial generation after the switch to light medium. This procedure produced only the hybrid band that “according to the hypothesis of Watson and Crick should be” the result of such replication. The experimental results thus “verified” the prediction. Then Baker and Allen described as “another test of the Watson-Crick hypothesis” an experiment in which the bacteria underwent “two generations in a medium containing N14.” The sche-

I MAGES OF AN E XPERIMENT

417

Fig. 13.3. Schematic representation of the Meselson-Stahl experiment from J. J. W. Baker and G. Allen, The Study of Biology

matic drawing accompanying their text looked very different from that of Watson (figure 13.3).8 The representation of DNA double helices in centrifuge tubes was obviously copied from the Watson scheme, but the Baker-Allen scheme shifted the attention from the mechanical aspects of centrifugation to a representation of the replication process as it was assumed to be taking place in the bacterial DNA. In accord with the text, the scheme implies that the first- and second-generation replications were followed in separate experiments. This version also omits all mention of the fact that the DNA was not directly visible in the centrifuge cells, as the diagram suggests. In 1970 Albert Lehninger, chairman of the Department of Physiology at the Johns Hopkins University School of Medicine, published the first edition of a text titled Biochemistry that dominated the teaching of this subject for the next decade. He wrote his book, he said, for “students who are taking their first and perhaps their only course in biochemistry, whether as undergraduates or graduate or medical students.” 9 Only after covering many of the more “classical” biochemical topics did Lehninger come, in chapter 29, to the “Replication and Transcription of DNA.” His treatment of the first of these two phenomena bore a strong family resemblance to the account in Watson’s Molecular Biology of the Gene, which Lehninger described as “an excellent elementary account of the principles of molecular genetics.” 10 Unlike Watson, however, Lehninger alluded to the debate that had taken place between 1953 and 1957 over the mode of replication. Instead of introducing the Meselson-Stahl experiment as a test of the

418

I MAGES OF AN E XPERIMENT

prediction of the double helix alone, Lehninger presented it as the arbiter between the three possible mechanisms defined during that debate. “The most striking feature of the Watson-Crick hypothesis from the genetic point of view,” he wrote, is its postulation that the two strands of double-helical DNA are complementary and that replication of each to form complementary daughter strands results in two daughter duplex DNA molecules, identical to the parent DNA, each of which contains one strand from the parental DNA. This process is called semiconservative replication. However, there are two other mechanisms by which two complementary strands of the parental duplex DNA could yield daughter DNA duplexes chemically identical to those of the parent, conservative and dispersive replication. . . . Ingeniously contrived experiments carried out by Meselson and Stahl in 1957–1958 conclusively proved that in intact living E. coli cells DNA is replicated in the semiconservative manner postulated by Watson and Crick. After giving a straightforward narrative description of the experiment, Lehninger concluded that “these results . . . are exactly those expected from the hypothesis of semiconservative replication proposed by Watson and Crick; whereas they are not consistent with the alternative hypotheses of conservative or dispersive replication. Thus, the Meselson-Stahl experiment gave profound support to the WatsonCrick hypothesis and was particularly convincing because it was carried out in intact dividing cells without intervention of inhibitors or other injurious agents.” 11 Lehninger illustrated the three possible mechanisms of replication of DNA with the diagram shown in figure 13.4. (In the original “newly replicated strands” were shown in red, parental strands in black.)12 It should be evident that Lehninger devised his diagrammatic representation by adapting for all three possible replication modes the diagram that Meselson and Stahl had used in their original publication to depict the prediction of the Watson-Crick model. This tripartite scheme was copied repeatedly in later textbooks. Lehninger also showed the results of the Meselson-Stahl experiment “schematically” (figure 13.5).13 This rendition is particularly interesting when compared with the results as published by Meselson and Stahl and as represented in Watson’s textbook. The figures on the left were clearly intended to represent centrifuge tubes, but they are

I MAGES OF AN E XPERIMENT

419

Fig. 13.4. Illustrations of three possible mechanisms of DNA replication, from A. Lehninger, Biochemistry

far more abstract than in the Watson depiction. The locations of the DNA are shown not as little symbols of a double helix but as “bands.” These bands, and the horizontal position of the “tubes,” causes the diagrams to resemble the films on which the bands were shown in the original experiment more than they do the real tubes of a conventional centrifuge that Watson’s diagram featured. Neither the caption nor the text, however, explained that the bands were “seen” on films rather than directly in the centrifuge cell. To a reader introduced to the Meselson-Stahl experiment through Lehninger’s textbook, this abstract diagram would appear only to reduce the results to their simplest essential form. To someone familiar with the original publication of Meselson and Stahl, the ambivalence of the diagram can be startling. Visual images with double meanings have been regarded as characteristic of art rather than of science. At the end of his description of the Meselson-Stahl experiment Lehninger added: “That semiconservative replication of chromosomal DNA also occurs during cell division in eucaryotic organisms was shown in conceptually similar experiments by Taylor and his colleagues on roots of been seedlings grown in tissue culture.” 14 Taylor’s experiments thus appeared in Lehninger’s text as the extension to

420

I MAGES OF AN E XPERIMENT

Fig. 13.5. Schematic representation of the Meselson-Stahl experiment from A. Lehninger, Biochemistry

higher organisms of what Meselson and Stahl had shown for procaryotic organisms. This inversion of the historical order in which the experiments were performed and published is typical of the way in which textbooks restructure the flow of scientific investigation to replace contingent events by logical coherence. By the 1970s the formulations of the historical role of the Meselson-Stahl experiment illustrated by the preceding examples were becoming canonical. The penetration of these themes into genetics textbooks can be illustrated by the treatment in Ursula Goodenough and Robert Paul Levine’s Genetics, a book published in 1974 and “intended to accompany a one semester course in genetics for college or

I MAGES OF AN E XPERIMENT

421

medical students who have a knowledge of introductory biology and chemistry.” 15 In a section titled “The mode of DNA replication,” Goodenough and Levine, too, began with the prediction suggested by the WatsonCrick model of DNA structure. “Experimental support for this mode of DNA replication came in 1958 from M. Meselson and F. Stahl.” Like Baker and Allen, they divided their very compact description of the experiment into two stages. Replication for one generation produced a “single species with the intermediate density of a 15 N-14 N hybrid, exactly as one would expect from the mode of replication proposed by Watson and Crick.” If the daughter cells were allowed to divide again, the results, which they summarized briefly, are again “compatible with the mode of replication proposed by Watson and Crick.” 16 Only after describing the experiment as a direct test of the WatsonCrick prediction did Goodenough and Levine mention alternative possibilities. “The Meselson-Stahl experiment is particularly valuable,” they wrote, “in that it not only established the mode of replication of DNA but it also eliminated the possibility of other modes.” They then described the conservative and dispersive modes so ruled out.17 In their limited space, Goodenough and Levine chose barely to mention that the method used to establish the densities of the DNA was density gradient centrifugation. Correspondingly, their schematic representation showed no centrifuge tubes. It did, however, move closer to the original representation of the results by depicting the density distributions not as visible bands but as ultraviolet absorption peaks resembling the densitometer curves Meselson and Stahl had included in their paper (figure 13.6).18 In their compromise between brevity and clarity, therefore, Goodenough and Levine chose to remain closer to one of the original published forms of the results but omitted the connections between the positions of DNA molecules in a centrifuge cell and tracings made from bands formed through an optical system on a film. There is no need to follow in detail further variations on this standard theme that appeared in the textbooks of the 1970s and 1980s.19 A few of the more advanced textbooks provide more detailed descriptions than those summarized above, including clarification of the fact that the positions of the bands are registered by ultraviolet light optics on photographic plates. One elaboration that several authors added to

422

I MAGES OF AN E XPERIMENT

Fig. 13.6. Schematic representation of the Meselson-Stahl experiment from U. Goodenough and R. P. Levine, Genetics

the earlier textbooks’ division of the Meselson-Stahl experiment into two phases was to suggest that the first-generation result ruled out conservative replication, whereas the second-generation results ruled out dispersive replication. This interpretation was depicted in a boldly pedagogic manner in the schematic representation of the experiment in Neil Campbell’s Biology textbook (figure 13.7).20 A notable exception to the usual textual and visual schematizations used to simplify and dramatize the presentation of the MeselsonStahl experiment is Lubert Stryer’s Biochemistry, first published in 1975. Within a little more than one page of text Stryer managed to summarize all the elements of the experiment, retaining in com-

I MAGES OF AN E XPERIMENT

423

Fig. 13.7. Schematic representation of the Meselson-Stahl experiment from N. Campbell, Biology

pressed form much of the language of Meselson and Stahl’s original publication and quoting a key passage from their carefully stated conclusions: “that the nitrogen of a DNA molecule is divided equally between two physically continuous subunits; that the subunits are conserved through many duplications.” Preserving the spirit of Meselson and Stahl’s rigorous restraint, Stryer ended with the statement that “their results agreed perfectly with the Watson-Crick model for DNA replication.” Stryer directly reproduced the figure in which Meselson and Stahl had grouped the series of films showing the bands over four generations, with the densitometer curves alongside. He included also a redrawn but otherwise faithful reproduction of their schematic diagram of the semiconservative replication of a double helix.21 Although Stryer placed the Meselson-Stahl experiment in the conventional sequence, following a description (in their own words) of Watson and Crick’s hypothesis for the replication of DNA, and introduced the experiment as a “critical test of this hypothesis,” he refrained from recasting his summary of what he called this “incisive” experiment to invoke pedagogical lessons.22 Stryer’s example suggests that it is sometimes possible to compress and simplify without losing the qualities that gave the original publication of a scientific achievement its luster. There is some indication from the most recent textbooks that the Meselson-Stahl experiment may have run its course as a standard fix-

424

I MAGES OF AN E XPERIMENT

ture in the pedagogical structure of molecular biology. The second edition of the Molecular Biology of the Cell (a multiauthored descendant of Watson’s original textbook), published in 1989, omits mention of the experiment. Under the section “The structure of DNA provides an explanation for heredity,” the text describes the separation of the strands and the formation of new strands by base pairing. The section concludes with the statement that “the mechanism of DNA replication is said to be semiconservative,” without adducing experimental evidence for that conclusion. The book reproduces the familiar schematic representation of semiconservative replication, but without tracing its lineage to the paper in which Meselson and Stahl first published it.23 In the monumental textbook Genes and Genomes, which appeared in 1991, Maxine Singer and Paul Berg included schematic diagrams of all three possible modes of replication: semiconservative, conservative, and dispersive. In their text they wrote, “Since its proposal [by Watson and Crick], the basic nature of the template mechanism has been verified by a large body of evidence obtained from many different types of organisms, in vivo and in vitro. As the model predicted, replication of all double strand DNA is semiconservative. Alternatives such as conservative or dispersive modes of double strand DNA replication are not known to occur.” 24 Long familiarity with the double helix, and the many successful investigations built on the belief that DNA replicates as the structure predicted it would, seem to be eliminating the need to provide what the molecular biologists of the 1950s hailed as decisive support for what had been for several years a contested implication of a theoretical model. Scientists regularly bury their past, unconscious of the degree to which their assumptions have grown from that past. The survival of the diagram Meselson and Stahl once used to illustrate the concordance between their result and the Watson-Crick replication mechanism, now without attribution to its originators, is also a symbol of the process through which the outcomes of scientific investigations eventually become detached from their historical sources. Simplification is vital to the progress of science. Without it the bounded cognitive powers of the human intellect could not make sense of a boundlessly complex natural world. The Meselson-Stahl experiment revealed a simple regularity within a process of DNA replication that ongoing investigations have shown to be far more complicated than scientists of the 1950s could imagine. If the experiment disappears from future textbooks, the most compelling reason will be

I MAGES OF AN E XPERIMENT

425

that growing knowledge of this complex mechanism will have crowded it out of the limited space available in the chapter, or section of a chapter, that can be devoted to the replication of DNA. Even the original paper presenting the Meselson-Stahl experiment vastly simplified the historical process of investigation that formed, as we have seen, the context from which the experiment emerged. Readers of that paper were, however, at least at one point in direct contact with the results of the experiment, because Meselson and Stahl reproduced a set of ultraviolet absorption films that constituted a selected sample of the immediate data of the experiment. With rare exceptions, such as Stryer, or Frank Stahl himself, who wrote in 1964 a textbook titled The Mechanics of Inheritance in which he reprinted the same plate used in the original paper,25 textbook writers have substituted various simplified representations for these results. Some of these schematic diagrams attempt to combine several levels of representation. They represent the experimental procedure, the experimental results, and the replication of DNA. Some of them provide matching sequences of experimental results and their “interpretation” that invite the reader to “see” the semiconservative replication of DNA in the result without intervening inference. As we have noted, the difficulty of incorporating all three of these levels into a format that can be readily scanned sometimes results in images so abstract as to provide only bare glimpses of the robust experimental operations and the vivid film traces that underlay the actual experiment. The narrative paragraphs that these diagrams accompany supply some of the detail missing from the latter, but the rich visual character of the original experimental results is often sacrificed to the drive for simplicity. The schematic diagrams now ubiquitous in the textbooks of the life sciences provide convenient entry into worlds of bewildering complexity, but they are sometimes little more than icons of what they represent.

II Because the analytical ultracentrifuge was large and expensive, the Meselson-Stahl experiment did not lend itself to undergraduate laboratory exercises. It has long been, however, a favorite subject for lectures that demonstrate the ideals of biological experimentation along with the concept of semiconservative replication. Its appeal to teachers of the biological sciences is pungently exemplified by the reasons

426

I MAGES OF AN E XPERIMENT

Fig. 13.8. Diagrams drawn by Gunther Stent on large sheets of paper for an undergraduate lecture on the Meselson-Stahl experiment at Berkeley. Photograph by the author.

that Gunther Stent gives for having heartily enjoyed teaching the very experiment that had once spoiled his own efforts to solve the replication problem. Using a technique he had learned from Delbru¨ck, Stent illustrated his lectures for many years with diagrams that he drew by hand on large sheets of butcher paper. For the Meselson-Stahl experiment he prepared four figures (three of which are reproduced in figure 13.8). The first was the usual intertwined helices illustrating conservative and semiconservative replication. The third figure he used to explain how heavy, hybrid, and light DNA molecules would collect in that portion of the centrifuge cell at which their buoyant density equaled

I MAGES OF AN E XPERIMENT

427

that of the solution. The fourth was a schematic representation of the placement of the bands on the photographic films during the first three bacterial generations. Stent favored these hand-crafted but permanent figures, which he put up before his class year after year, because he was not good at the blackboard, but he could feature the particular points he wished to make more easily than with photographic slides.26 There were many aspects of the experiment that made it satisfying for Stent to teach. The first was that “semi-conservative replication is a problem which can be stated very clearly,” so that even the slowest students understood it. They could easily see how the two strands were supposed to separate in the Watson-Crick structure. He could readily explain densities and the way one marked the molecules with heavy nitrogen. “When you take them through it, and when they see the bands, they have learned something that is really meaningful, in a way that I think gels are not, because [in gels] there are too damn many bands.” 27 Stent not only found that “even the dumbest Sophomores” could understand the experiment and its consequences but that he could use it to teach a tremendous number of concepts, including the nature of chemical equilibrium through the equilibrium between gravitation and diffusion. “I bring in Archimedes in the bathtub, and so forth. It is rich in that it proves an important point, it has a diversity of physical [principles, and] it . . . is within the range of people of average intelligence.” 28 As in textbooks, so in classes, the Meselson-Stahl experiment, along with others of its era, is now losing the place it has held in the education of more than a generation of students. Typical of such trends is the experience of Charles Hill, at the Penn State University College of Medicine in Hershey, Pennsylvania. For Hill, the experiment is unique in molecular biology, because it is so “graphic and pictorial.” Hill taught the experiment in his molecular biology course, “in some detail,” from 1968 until 1989. Then the “crush of total information” that it was necessary to convey, which allowed him “to devote less and less time to these really classical experiments,” pushed the Meselson-Stahl experiment out of his lectures.29

III What do scientists mean when they call an experiment beautiful? By what standards did John Cairns judge the Meselson-Stahl experiment to be “the most beautiful experiment in biology”?30 If beauty is “that

428

I MAGES OF AN E XPERIMENT

quality . . . which affords keen pleasure to the senses . . . or which charms the intellectual faculties,” 31 may we infer that the beauty that scientists see in experiments bears no relation to their utility? If, on the other hand, we agree with Meselson’s view that “all really important experiments are beautiful,” 32 is beauty a functional attribute of the experiments that display it? When scientists call an experiment “beautiful” do they mean something different from what they mean when they call it “elegant”? That aesthetic sensibilities play a role in the “structure and style of the scientific process” has long been recognized.33 Although scientists are often attracted to scientific theories, models, or objects by their perceived beauty, many commentators would agree with Helge Kragh that “beauty is essentially subjective and hence cannot serve as a commonly defined tool for guiding or evaluating science.” 34 Others, including most recently James McAllister, contend that beauty is frequently a significant factor in determining which theories scientists prefer. According to McAllister, one can isolate several “indicators of beauty,” including particularly simplicity and symmetry, which constitute the aesthetic criteria scientists attach to their theories and hypotheses. Neither symmetry nor simplicity can be defined, however, independently of the context in which scientists in a given field, during a given era, recognize these qualities in their theories and in their reasoning. The aesthetic qualities of theories that have previously proved successful guide the sensibilities of a community of scientists in the perception of beauty in their current reasoning. Aesthetic preferences thus arrived at inductively can be successful in formulating precepts to guide future assessments of scientific theories.35 McAllister’s analysis appears applicable also to the aesthetic qualities scientists recognize in experiments. My intention here is not to discuss the general role of beauty in science but to add to previous accounts of the aesthetic experiences that scientists themselves describe.36 The most widely known personal accounts of beauty in science come from mathematicians and theoretical scientists. The powerful aesthetic response of biological scientists and students to the Meselson-Stahl experiment offers a strategic opportunity to expand the exploration of scientific beauty to its role in experimentation. In order to begin such an exploration I have asked a number of scientists associated in one way or another with this experiment to comment on its beauty. When I asked John Cairns in 1993 about the statement that Judson

I MAGES OF AN E XPERIMENT

429

had quoted fourteen years earlier, he at first played down its significance by recalling that “I just said that having lunch at a pub.” I pressed him to say what is beautiful about the experiment. “In a sense,” he said, everything comes together. First of all, the experiment was needed at the time. Secondly . . . the situation was such that that was the result. Suppose [that DNA replicated] in tiny little pieces as Delbru¨ck and Stent said, then the result would not have been so beautiful. . . . But I don’t think Meselson and Stahl get any credit for that sector of the beauty, nature gets the credit for that. Then there was the beauty of the way they did it, because they did it with the utmost precision. Lastly, there is the fact that they knew every part of the implications of [what] they found. They had realized, for example, that it meant that everything is replicated once before anything is replicated twice. . . . Matt and Frank wrung every bit out of [the result], as elegantly as possible. So it has elegance on every count, and the fact that it has it on every count is itself an elegance.37 Gunther Stent responded hesitantly to my question. “I don’t know. It really tells the whole story. I mean, you’ve got the heavy band, you’ve got a light band, which are the controls, then you have the band in between halfway. I don’t know whether I can explain, but it’s in its simplicity. Any fool can see that immediately, you don’t have to be very sophisticated.” 38 Stent’s appreciation for the simplicity of the experiment and its interpretation is the more striking in the light of his own early reputation for complicated experiments requiring sophisticated interpretation. Frank Stahl was characteristically blunt when I asked him in 1988 about the beauty that other people see in the experiment. “Yes,” he said, and “what’s very important to me is, I feel that way. I think it’s one of the most beautiful experiments I’ve ever seen, and I’m proud as hell to have been involved in it. Yes, it’s gorgeous. And I’ve been trying to do something half as pretty since. It’s a nice thing to have as a target.” 39 In another comment that might have served as his explanation of what made the experiment so beautiful, Stahl said about the cleanness of the result, “It’s very rare in biology that anything comes out like that. It’s all so self-contained. All so internally self-supporting. Usually, if you are lucky to get a result in biology, you then spend the next year doing all those plausible controls to rule out other explanations; but this one was just a self-contained statement.” 40

430

I MAGES OF AN E XPERIMENT

One time these qualities of the experiment even worked to Stahl’s disadvantage. During the seminar at Yale University to which he had come for a job interview soon after it had been performed, Stahl found himself unable to fill the fifty minutes allotted for his talk. “I was finished in 25 minutes, because that is all it takes to tell that experiment. It’s so totally simple and contained. . . . I didn’t get a job offer from it.” 41 Howard Schachman answered the question of what is beautiful about the Meselson-Stahl experiment in May 1994: It’s conceptually simple and elegant. You had a question, and you had the opportunity to ask, if two strands were to separate and you built another one for each one, and that was a different density from the original ones, would you get a molecule that was halfway along in the density gradient? Then when you did it in the next generation, you would just have dilution of half-size molecules, you would never have them increase. Just every step in the experiment was thought out very carefully, and the execution was perfect, and they got the result that you would expect from the model. It starts out with a presumption of a model, which is a good way to do experiments, and then designs a very pretty experiment to do it. . . . So many experiments always have complications. It doesn’t go 100 percent in that direction, only 90 percent, . . . and you have to rationalize away the 10 percent. . . . Theirs was clean, it didn’t require any kind of fudge factors.42 Instead of asking James Watson what was beautiful about the Meselson-Stahl experiment, I asked, in March 1990, what qualities of the experiment had led him to call it a classic experiment in the second edition of his Molecular Biology of the Gene. He responded that “it really showed that the DNA molecule split into two.” He then moved on to mention questions left unresolved until it was proved that DNA was a two-stranded rather than a four-stranded molecule. The sense in which the experiment was classic seemed to depend, for him, on its placement “in the whole.” The Meselson-Stahl experiment was “the formal proof of separation. . . . When we found the double helix we thought it was very plausible, but we didn’t ever expect so soon a formal proof of the fact.” 43 To the graphic, pictorial qualities that Charles Hill sees as unique to the Meselson-Stahl experiment he also applies the word “beauty.” That is not to say that he limits beauty to the visual context: “In molecular biology there is usually some sort of numerical interpretation of

I MAGES OF AN E XPERIMENT

431

data, or even in experiments which do involve some sort of photographic image, there is a much more abstract scheme. In this case . . . the only number really necessary is the very simple concept of a quantitative doubling, and its all visual, it’s graphic. It is a direct, visual demonstration of one of the most important relationships of all—the intrinsic suitedness of the DNA molecule to be the genetic substance.” 44 My sample of five responses to the Meselson-Stahl experiment suggests both shared and particular images of its beauty. The subjective qualities of the aesthetic experiences are self-evident. No two of the scientists to whom I addressed the question gave the same set of reasons for their feeling of beauty. What impressed each of them as beautiful depended as much on their individual sense for what is beautiful as on describable properties of the experiment. At the same time there are recurrent themes, including simplicity, precision, cleanness, and strategic importance. One has the impression that, in attributing beauty to this experiment, these scientists were seeing in it a realization of their ideals about how experiments should be conducted in their science. It was beautiful just because, in practice, such ideals were so seldom fully achieved. “It is easy to do a beautiful experiment to solve a minor problem,” Stahl once remarked to me. “The difficulty is to devise a beautiful experiment that will solve an important problem.” 45 Unlike some highly regarded experiments such as that of Hershey and Chase, which were actually composites of multiple individual experiments, Meselson-Stahl really was a singular historical event, an experiment performed in two parts during two identifiable segments of time, the first during October 1957, the second during January 1958. To my knowledge it was never afterward repeated, though numerous related experiments were subsequently performed using similar techniques. Only one person—Matthew Meselson—ever experienced the full range of the sensible qualities of the experiment. Its aesthetic features included, for him, the tactile sensations and sounds associated with the elegant Model E analytical ultracentrifuge, the first appearances in the darkroom of the sharp absorption bands—one set after another on successive days—just where he expected them to be, and other untransferable sights and sounds. To what extent did the colleagues who immediately saw the experiment as beautiful perceive the same experiment that Meselson did? The first recorded judgment that the experiment was “extremely beau-

432

I MAGES OF AN E XPERIMENT

tiful” came from Cyrus Levinthal, in response to Meselson’s letter about it. Meselson’s spare account did not provide a full description of the experiment. Nor had Levinthal seen the films that constituted its immediate, visual result. On what grounds, therefore, did the experiment appear beautiful to him? Like any professionals who share training and experience, scientists communicate with one another in shorthand. Much of what they leave out of their condensed descriptions their colleagues fill in mentally from their own similar knowledge. Levinthal could undoubtedly visualize the preparatory stages of the experiment, because he routinely performed experiments that relied on similar preparations. He could less readily have visualized the observations associated with the centrifuge run, because he had not himself used the new density gradient method, but his earlier correspondence with Meselson about that method would have given him some background knowledge to supplement Meselson’s outline description of the experiment. Levinthal’s sense of the beauty of the method probably depended, however, less on the ability to visualize it in concrete detail than on his recognition that in a single stroke it provided a decisive answer to a crucial question to which he had previously applied his own distinctive methods with far less success. Abstracted from both the tangible procedures of the centrifuge and the visual pattern of the results, the beauty was concentrated in the sharpness and significance of the outcome. Those who first encountered the experiment in the PNAS paper that Meselson and Stahl published in 1958 may have experienced more directly the aesthetic qualities, not of the experiment itself, but of their carefully crafted verbal and graphic representation of its result. The lucidity of their economical description of the problem and their method, and their rigorous statement of the conclusions they had drawn from the experiment, possess considerable literary beauty. Their diagrammatic representation of the replication of DNA units and subunits displays, in a visual mode, the two most canonical attributes of beauty: simplicity and symmetry. The best testimony for the aesthetic quality of the corresponding diagram with which they represented the replication of the Watson-Crick double helix is the frequency with which that diagram has been repeated in a succession of textbooks and its survival even in the recent textbooks from which the experiment itself has vanished. The simplicity so often associated with the beauty of the Meselson-

I MAGES OF AN E XPERIMENT

433

Stahl experiment is a surface phenomenon. The result can be expressed in various simple forms, and the central experimental operations can be compactly outlined. Underlying each experimental operation, however, are numerous complexities of theory and technique. Fully to explain the experiment entails understanding the operation of a powerful and sophisticated machine, whose design and function involve much physics, optics, theory of solutions, and chemical and hydrodynamic equilibrium. The optics in turn are connected to knowledge of absorption spectra and the physical and chemical properties of DNA. Extensive knowledge of the physiology and genetics of bacteria is incorporated into the experiment. The choice of 15 N relies on knowledge of the physical and chemical properties of isotopes. Beneath the replication problem itself are the esoteric theoretical structure and arcane experimental practices of X-ray crystallography and the nature of the chemical bond that Watson and Crick had built into the double helix. That this intrinsically complex experiment can, nevertheless, appear extremely simple to experienced molecular biologists is because all of these attendant factors can be relegated to background knowledge. They were shared with other contemporary experiments with which these scientists were already familiar. When scientists in the same or related fields form a mental image of the Meselson-Stahl experiment, they need include in it only the way in which standard components are combined in it to produce the novel features of this particular experiment. The cleanness and simplicity of the experiment appear immediate to them, because they need not recapitulate to themselves all the steps in reasoning that connect, for example, a sequence of dark bands on a series of photographic films with the categories “heavy,” “hybrid,” and “light” DNA. The pedagogical simplicity of many textbook presentations of the experiment is of a different order. Students introduced to molecular biology or to modern genetics for the first time are generally not prepared to connect the many background components to the salient features of the experiment, so these complications are eliminated. The resulting description is elementary, a simplicity that is superficial. The simplicity of the original Meselson-Stahl experiment, built on complexity, is profound. Reviewers of this chapter have wished that I might connect the foregoing discussion of the beauty of the Meselson-Stahl experiment

434

I MAGES OF AN E XPERIMENT

more explicitly to the general topic of beauty in science and to the lengthy literature on that subject. I recognize the desirability of such an extension of the treatment provided here, but neither space limitations nor my present knowledge of that literature are conducive to including such a discussion in this volume. Just prior to publication, my attention was drawn to a recent book by Ernst Peter Fischer on the aesthetic impulse in science. Fischer describes the Meselson-Stahl experiment to illustrate his claim that “form plays a role,” not only in the presentation of science but in its outcomes themselves, and not only in “the case of beautiful theories, but also, and in particular, of elegant experiments.” Among the elements of the beauty of the experiment Fischer includes the concept of isotopes themselves and the refinement of the analytical ultracentrifuge. “The picture that the use of these procedures yielded, and the picture that one can produce of the primary process of interest itself,” he concludes, “are so closely related to one another that their aesthetic appeal and the tangibly perceptible power of persuasion of the ‘Meselson-Stahl experiment’ speak for themselves, and make any further words superfluous.” 46 In somewhat more elegant language, Fischer’s summation mirrors that of the comments of the scientists I have included above and suggests that their immediate reactions to the experiment do capture much of what can be said about the beauty of such an experiment.

C HAPTER F OURTEEN

Afterword

I What part has the Meselson-Stahl experiment played in the development of molecular biology in the postwar period? In his history of molecular biology, the French biochemist Michel Morange has invoked two other pivotal experiments in the identification of the genetic role of DNA as representative of two fundamental roles of experiment: “There exist in science two types of experiment, of different nature and function. The experiment of Avery [identifying the bacterial transforming factor as DNA in 1944] . . . is an example of the first category: without a priori ideas, the investigator discovers a surprising, novel, unexpected phenomenon. The experiment of Alfred Hershey and Martha Chase in 1952 is an example of the second category: at the limit, the result is already known. The objective of the experiment is simply to demonstrate it in a clear manner.” 1 These two categories clearly do not exhaust the possible types of experiment (Morange himself acknowledges as much in the English translation of his book by adding the phrase “at least” two types, and when Meselson read this passage he quickly added a third type—“Expecting a certain result, a totally different result is found”),2 but Morange’s dichotomy nevertheless provides a useful framework for discussion. As his phrase “at the limit” hints, even these two types of experiment are less sharply distinguished than the rest of his statement asserts. The degree to which the Meselson-Stahl experiment fits into the second category depends on the degree to which one already “knew,” in 1958, that DNA replicates semiconservatively. To those who, like Watson and Crick themselves, appeared confident from the time they proposed the double helix that their model predicted the manner in which DNA replicates, the Meselson-Stahl experiment was

436

A FTERWORD

a supremely clear demonstration of a result already known. To biochemists and others more skeptical of the status of an inference drawn from a structure reached mainly by model-building, the MeselsonStahl experiment seemed more like the discovery of the semiconservative replication of DNA. It did not, however, produce a phenomenon surprising or unexpected to those concerned with the replication problem. As its textbook representations often suggest, the experiment fits the classic role that philosophers of science have commonly attributed to experiment: that of testing previously formulated theories. Some have portrayed it as confirming the Watson-Crick model, others as providing an experimental decision between three alternative hypotheses: conservative, semiconservative, and dispersive replication. In any case, the experiment marks, as prominently as any other single event, the transition between an era in which the reality of the replication mechanism suggested by the double helix was contested and one in which the assumption of its fundamental correctness sustained investigations of the details of the process. That search, continuing to the present, has revealed a mechanism far more complex than could be imagined by those who attempted, during the first years of the double helix, to picture how its two strands might come apart and serve as templates for the construction of further complementary strands.

After reading the above in 1998, Meselson wrote in the margin of a copy of the manuscript the cogent comment: “Yes, in detail [the mechanism was] not at all known. But one often deliberately ignores the complexity one suspects, since it is likely to be at a separable level of explanation. Nature, at least as conceptualized by scientists, has a layered structure.” Concerning their own expectations, Meselson commented, “In retrospect it all seems very simple. But at the time it was not sure that unimagined complexities would not come into view as research continued.” In conversation he added, “We were aware that there were all kinds of things behind—complexities out there. But you can solve a problem without caring about the huge complexities at the next level.” 3 Meselson’s reflections substantiate the subjective experience corresponding to the general themes about simplicity and complexity in science outlined in the preface to this book.

A FTERWORD

437

In a draft chapter since removed from the present book, I wrote that I had hoped “not only to account for the historical origins of the Meselson-Stahl experiment, but to recreate a picture of the everyday world of two young scientists entering their careers at a time not far removed from our own.” Jan Drake objected that the phrase “ ‘not far removed’ is inappropriate because of the profound changes in the climate of biology over the past 40 years. I cannot imagine a graduate student or postdoc now daring to take the risks Meselson and Stahl took, because the requisite number of papers to get a job could not be assured.” 4 Both views are, I believe, appropriate from their respective vantage points. To a historian who has written about the investigative pathways of scientists ranging back to the eighteenth century, the ventures of two young scientists that took place during the years just after I had studied undergraduate biology appear as very recent events. To a scientist of approximately the same age who has remained in biology, the changes that have taken place in that field since the mid-1950s make them remote events. Meselson and Stahl themselves look back on these events as part of a golden age of their youth, when it was possible to concentrate on an absorbing scientific problem with no pressure to begin a professional career.5 Were one inclined to draw normative inferences about the pursuit of science from this case history, one might well ponder how to retrieve something important that has been lost from the ethos of molecular biology. The atmosphere that was particularly strong at Caltech, but characteristic more generally of early postwar biology, encouraged two fledgling scientists to pursue a grand idea of their own invention that some of those who gave them advice, including Watson himself, doubted could work. We might well ask whether the powerful technical methods and conceptual orthodoxies of the molecular biology of today do not tend to suppress, rather than encourage, such creative adventures as Meselson and Stahl were able to conceive and carry out.

II One of the aims of the story told in the preceding chapters has been to show that the investigative pathway that Meselson and Stahl followed to the performance of their most famous experimental achievement was not linear. Not only did the form and nature of the experiment they first imagined in 1954 change several times before the design of the definitive experiment emerged, but its pursuit was en-

438

A FTERWORD

meshed in several other investigative goals that they took up along the way. Nor was the experiment the only success to which these intermingled goals eventually led. The auspicious apparent success that temporarily diverted them from their plan for testing replication mechanisms—that is, the application of their new density gradient method to determine molecular weights—proved ephemeral. That method itself, however, acquired an importance far beyond the replication question, which prompted its inadvertent discovery. Meselson, his students, and his collaborators found so many uses for it that it seemed, he wrote to Stahl in September 1959, that “CsCl has an inexhaustible number of golden eggs to lay.” A few months later he helped Stahl order a Model E ultracentrifuge for his laboratory.6 During the following years their density gradient method spread widely in other laboratories of molecular biology, where it found many applications. The mechanism of 5-bromouracil mutagenesis, which had for a time overshadowed Meselson and Stahl’s original project, faded in turn from their collaborative venture with the great success of transfer experiments using 15 N instead of 5-BU. That story, however, was not over, and they were not the only investigators interested in mutations induced by incorporating such substituted bases into DNA. At Harvard, Ernst Freese used 2-aminopurine and 5-bromouracil to study the differences between the mechanism of mutations induced by baseanalogues and that of spontaneous mutations in phage T4. At Berkeley, David Pratt and Gunther Stent incorporated 5-BU into T2 and T4 to study the formation of mutant heterozygotes. Their results also provided further support for the semiconservative replication of phage DNA.7 Meselson and Stahl tried to renew their collaboration by correspondence as soon as Stahl settled in at the University of Missouri in the fall of 1958. Both took up again the 5-BU mutagenesis program that they had outlined in the memoranda each had written at the beginning of their joint venture two years earlier (see above, pp. 183– 185). In October Meselson wrote that he had assembled nine compounds, including 5-iodo-uracil, “which may be mutagenic.” He suggested they exchange mutagens and mutant stocks and discussed some ideas for the “clean growth tests” they had earlier planned. A little later he carried out a “BU → Thy (as well as a Thy → BU) transfer exp[eriment],” and “in both cases at gen[eration] II there are the ex-

A FTERWORD

439

pected two bands in CsCl.” Stahl made the revival of the 5-BU mutagenesis facet of their project the main thrust of his research in Missouri. In March 1959 he wrote to Meselson, “I am impressed (more every day) by the potential for analogue-induction of mutation.” 8 Both Meselson and Stahl were imaginative as ever in proposing new ideas for the further development of the investigation, but neither made much progress on the problem during the academic year 1958– 59. Having been appointed an assistant professor in the Division of Chemistry at Caltech, Meselson had to teach a course in physical chemistry. Perhaps because of the meticulous care he gave to every task, he found the preparation of lectures and grading of papers to be “terribly hard work.” Very unhappy about the restriction it placed on research, he nevertheless decided to “put the course first.” In the time he did have left for experimentation, he concentrated on the collaboration with Jean Weigle on phage λ whose results have been described in Chapter 12.9 Stahl quickly found the University of Missouri unconducive to his hopes both for teaching and for research. A conservative genetics faculty reacted with hostility to his efforts to introduce molecular genetics into his course. Intellectually his surroundings seemed to him barren and corrupt, and he lacked much of the basic equipment necessary to pursue his experimental activity. Before he was halfway through his first year there, Stahl was looking for other job offers.10 Supported by James Watson and others aware of his talent, Stahl received and accepted a position at a newly formed Institute for Molecular Biology at the University of Oregon in Eugene. In November 1959, he summarized in a letter to Meselson the reasons why he and his whole family were “thoroughly pleased with our good fortune.” The people were “sociable and imaginatively so. They are enthusiastic about their work and their recreation.” The faculty “have faith in the future of the University.” Moreover, “the countryside is glorious and accessible.” 11 Mary was equally happy with her new surroundings. Stahl had finally landed in an environment in which he could pursue good science in a relaxed mood, and his whole family could prosper. At Missouri and during his first year at Oregon, Stahl investigated further the discovery he had made late in 1956—in confirmation of Meselson’s prediction—that 5-BU substituted phage is more sensitive to X-ray radiation than is normal phage (see above, pp. 209–210). What had been then mainly a tool to determine the degree of substitution now became the subject of the research. Before he left Missouri,

440

A FTERWORD

he was probably able to establish that substituted phage T4 are more sensitive also to visible light emitted from an ordinary fluorescent light fixture.12 In Oregon the work accelerated, as Stahl quickly attracted graduate students and postdoctoral fellows. They confirmed the original observation by Litman and Pardee that 5-BU phage are more sensitive to ultraviolet radiation. To estimate the extent to which 5-BU had replaced thymine in the phage stocks, Stahl turned to density gradient centrifugation, employing a preparatory centrifuge and collecting drops through a hole pierced in the centrifuge tubes, as Meselson had first tried to do in 1956 and he and Weigle had done in their phage λ studies in 1958. The result showed that most of the DNA released by osmotic shock from the phage had a density “very close to the most dense fraction which contains a significant number of particles.” This outcome, in accord with an earlier study he had done with Meselson and Walter McNutt at Caltech, suggested that the particles were “in large part fully substituted” and, consequently, that full substitution was compatible with the viability of the phage. Stahl and his associates submitted a paper reporting these results to Virology in October 1960.13 In this work one sees an inversion of means and ends typical of ongoing investigative pathways. Stahl had previously used an observed proportionality between X-ray sensitivity and degree of substitution to test the extent of substitution of 5-BU phage to be used for transfer experiments in the ultracentrifuge. Now the centrifuge provided in turn a test for the degree of substitution of phage used to study their sensitivity to several types of radiation. Thus the strands interwoven in Stahl’s earlier enterprise with Meselson could be unwoven and rewoven into a different pattern. The main objective of Stahl’s renewed encounter with 5-BU substituted phage was to test the prediction concerning 5-BU mutagenesis that Meselson had worked out in 1956 (see above, pp. 210–212). For that purpose his central problem was to “demonstrate the existence of ‘clean-growth’ mutagenesis in phage T4.” The reason for this desideratum, which Meselson and Stahl had already discussed several times in memoranda and correspondence at the beginning of their venture, was that mutation in a medium free of the 5-BU could be caused only by 5-BU previously incorporated into the phage DNA, thus distinguishing its action from mutagenic agents that increase by some other means the frequency of spontaneous mutations. Clean-growth

A FTERWORD

441

mutations “must be presumed to represent transitions of the type AT → GC,” because it would be a 5-BU base present in the DNA molecule in place of the normal thymine that would be likely to pair during replication with guanine as an incoming nucleotide instead of adenine, the guanine subsequently directing the addition of a cytosine on the complementary chain.14 During his first year at Oregon, Stahl did not get far with this effort, because the mutagenic properties of the substituted DNA caused many mutants to arise while the phages were still being grown in the 5BU medium. The following year, however, his former collaborator at Caltech, George Streisinger, rejoined him in Eugene and showed him how to get rid of these mutants, so that he could “see” the mutants that arose later in the clean medium. From Sydney Brenner they received a strain of E. coli (B3), which required thymine in its nutrient medium, a condition that favored the production of fully substituted phage DNA. They used the drop-collecting cesium chloride density gradient method to select the most dense fractions of the phage. With these tools, they were able to identify six 5-BU mutants that reverted at a high rate to the wild type. Four of the six required 5-BU in the medium, whereas two of them fulfilled the long-sought condition that they reverted also in the absence of 5-BU in the medium. The existence of these two classes, they wrote in a paper submitted in July 1962 to PNAS, “supports the widely held notion that 5BU induces mutations via the mechanism of base-pairing mistakes proposed by Watson and Crick.” In a footnote they added that “Loss of H⫹ from the #1 nitrogen of the 5BU by ionization might have similar consequences (M. Meselson, 1956, personal communication; P. D. Lawley and P. Brookes, J. Mol. Biol. 4 (1962)).” Thus, by the time Stahl finally was able to provide experimental support for the idea that Meselson had worked out through his meticulous study of the literature six years earlier, it was too late to claim more than informal priority for the idea. In a paragraph at the end of the text, Stahl pointed out that the predictions about 5-BU mutagenesis had been brought to his attention in 1956 by Meselson and stated, “We wish to acknowledge also the many assistants, students, and colleagues who have accompanied one of us (F.W.S.) in his six-year search for the Holy Grail.” 15 In 1964 Stahl published a textbook titled The Mechanics of Inheritance, intended to introduce the primary concepts of a genetics that had been growing “explosively” since the elucidation of the chemical structure of the genic material. Much of the chapter on the mutation

442

A FTERWORD

of DNA treated the Watson-Crick “suggestion” that rare tautomeric alternatives for each of the bases provides a mechanism for mutation during the replication of DNA. Providing structural diagrams of each of the bases in both forms, of all of the permitted and “forbidden” base pairs, Stahl described the mechanism with elegant clarity. Drawing on his own research with 5-BU and that of others using other mutagens, he showed succinctly how base pair transitions induced by a structural analogue of thymine “made possible an experimental challenge of the Watson-Crick hypothesis for the mechanism of mutation.” Revising the emphasis in the paper in which he had published these results, he wrote that it “now appears that most 5BU-induced mutagenesis is caused by ionization.” The outcome of these experiments, as well as similar experiments in which the deamination of cytosine by HNO2 converts this base to uracil, were “in fine agreement with our picture of the nature of . . . mutagenesis and of a semiconservative mode of replication.” 16 This outcome, reached at about the same time that John Cairns and others delivered convincing evidence that in the semiconservative replication of DNA the subunits are single polynucleotide strands, brought the collaborative venture on which Frank Stahl and Matt Meselson had embarked just a decade earlier to a satisfying closure. Now both of the major clusters of interwoven research threads, those oriented around the replication problem and those oriented around 5-BU mutagenesis, had come to successful conclusions. Just as their paper of 1958 had produced results in striking agreement with the mechanism of replication predicted by the double helix, so Stahl’s faithful experimental pursuit of the 5-BU mutation mechanism originally formulated by his research partner produced in 1962 strong supporting evidence for the mechanism of mutation suggested by the double helix. And just as James Watson incorporated the Meselson-Stahl experiment in 1965 into his Molecular Biology of the Gene as solid evidence that the DNA strands separate in replication, so he featured 5-bromouracil (and HNO2 ) mutagenesis in support of the case that mutations are changes in the sequence of base pairs.17 The two achievements were not of the same magnitude. Had Meselson and Stahl been able to carry out in 1957 what Stahl and his associates in Oregon completed in 1962, the work might have attracted comparable attention. By the time it appeared, however, other investigations of base-analogue mutagenesis had overtaken their lead, and

A FTERWORD

443

their contribution only fit into a broader pattern of circumstantial, but less than decisive, evidence for a mechanism that still rested largely an the inferences originally drawn by Watson and Crick from the stereochemistry of the bases. Stahl was himself well aware of the indecisive nature of this evidence. With characteristic modesty, he looks back on the 5-BU mutagenesis work as “inadequate science,” because the experimental strategy could never disprove the proposed mechanism. A negative result could always be explained away as a mere failure to show it. “We expected to find it, because we anticipated that Watson and Crick were probably right, and we looked long enough until we found it—it’s not the best science.” In accord with the “falsification” view of the philosopher Karl Popper, Stahl believes that the best science is an “experiment that can rule out a widely held belief.” 18 Meselson and Stahl had also, of course, anticipated that Watson and Crick were probably right when they designed the experiment to test semiconservative replication, but their result did rule out the only two alternatives to semiconservative replication—dispersive and conservative replication—that had been proposed. Moreover, the further visions that Meselson and Stahl had entertained for using 5-BU mutagenesis experiments to identify sequences of base pairs (see above, p. 185) were overrun when Marshall Nirenberg broke the genetic code in 1962 and opened up possibilities for determining base sequences undreamed of in 1956. But Stahl’s sixyear quest for his Holy Grail does not fully measure his scientific accomplishments during these years. Thriving in an environment that proved as happy for him and his family as his first impressions led them to expect, Stahl quickly established himself as a leading phage geneticist. He played a major part in the construction of the circular genetic map of the T4 genome19 and the characterization of the “amber” mutants, and he determined the direction of transcription of several T4 genes. During the last twenty years his research has focused particularly on the mechanisms of recombination, to which he has made deep experimental and theoretical contributions.20 Frank and Mary raised their three children, Andrew, Joshua, and Emily in Eugene and participated actively in outdoor life in Oregon. Mary became a skilled research technician and worked for many years together with Frank in his laboratory. The majority of his publications after 1970 were co-authored with her. Diagnosed with lymphoma in 1991, Mary bore her illness and the rigors of the treatments she under-

444

A FTERWORD

went with characteristic grace and optimism. In January 1996, Mary died. Matt Meselson resigned his assistant professorship at Caltech after two years and became a senior research fellow in order to have more time for the experimental opportunities that his mastery of the new density gradient method opened for him. “For a young man,” he explained to a colleague in 1960, “it is a mistake” to take an academic job unless it will allow him to “conduct really active research. . . . It is better to gain a sound reputation as a research worker while working on fellowships. Then really good academic jobs will become available.” This view, he added, was “an extremely prevalent one.” 21 He, too, began to train graduate students, who followed up the questions about semiconservative replication left open by the Meselson-Stahl experiment. During the summer of 1960, Franc¸ois Jacob and Sydney Brenner came to Caltech to use density gradient centrifugation for an experiment intended to demonstrate the existence of “messenger” RNA. As they were about to begin the experiment in late June, Meselson wrote that he felt it “will be either a total flop, or, less probably, a real smash.” After he, Jacob, and Brenner had encountered many failures, Meselson left Pasadena to propose to Katherine. During his absence, and near the end of their stay at Caltech, Jacob and Brenner finally performed the first successful experiment. Even though he was not present for this result, the method that Meselson had invented had given him a part in another of the landmark discoveries of early molecular biology.22 Despite the auspicious development of his research there, Meselson was, by 1960, nearly as ready to leave Caltech as Stahl had been two years earlier. The crowding and air pollution of southern California were becoming oppressive to him as well, and he anticipated that the future would only become worse. Los Angeles, he wrote to Stahl in March 1959, “will provide the nation with a sad lesson in the dangers of chaotic city-planlessness in the face of rapid urban growth. For me, the solution is to get out.” For a while he entertained the possibility that he might join Stahl in Oregon, but eventually he accepted his second offer from Harvard and arrived in Cambridge with Katherine in February 1961.23 At Harvard, Meselson continued research in molecular genetics, including important studies of recombination, restriction enzymes, ribosomes, and DNA error correction. Unlike Stahl, Meselson turned increasingly after 1970 to the molecular genetics of eukaryotic organ-

A FTERWORD

445

isms, including the classic fruit fly, Drosophila. Most recently he has been investigating the question of how a class of rotifers that have no sexual generation can have evolved in the absence of a process considered essential to natural selection. Still modeling his career after Pauling’s, Meselson began during the 1960s to devote large proportions of his time to international issues. He has made major contributions to the goal of eliminating chemical and biological weapons. During the 1980s he established, through field work in Southeast Asia, that yellow droplets alleged by the U.S. government to represent chemical warfare instigated by Communists were really only bee droppings.24 Meselson’s marriage with Katherine did not last. With his second wife, Sarah, who had worked as a technician in his laboratory before they married, he raised two children, Zoe and Amy. Their marriage ended in divorce in 1980. Six years later, Meselson married Jeanne Guillemin, a sociologist at Boston College, whom he had met also at Aspen. Matt’s relentless work habits proved compatible with Jeanne’s equally busy professional life, and they have pursued common interests and worked together. Recently they have collaborated on an epidemiological analysis of the outbreak of anthrax of 1979 in the Soviet Union that American agencies had attributed to an accident at a military facility engaged in activities related to biological warfare.25

III During my first conversations with Meselson and with Stahl, I asked each of them what effect their early success in producing the Meselson-Stahl experiment has had on their later careers. Meselson answered, in Cambridge, in 1987: It had a tremendous effect on me. In all kinds of ways. It made me feel very confident. Maybe I would have anyway, who knows? You can’t do the control. Leo Szilard used to say that scientists who were successful could be trusted in political things because they didn’t need to prove anything. I sort of believe that. It gave me the confidence to do other things that I really wanted to do. It got me job offers all over the place. It got me the friendship of . . . interesting people. It just feels good. You feel that you did something that really works. And it was a good thing, and it didn’t hurt anybody.26

446

A FTERWORD

Fig. 14.1. Matt Meselson and Frank Stahl in 1996, standing at the place where they had met in 1954 at Woods Hole. Photograph by the author.

Frank Stahl answered, in Eugene, in 1988: It set a standard for me in science which I don’t know I can ever achieve again, but it’s worth shooting for. Anytime I write a paper I remember that one, and say . . . come as close as you can. That’s been awfully important. Secondly, it allowed me to make some youthful mistakes and recover, like going to Missouri. Without that paper, going to Missouri would have been death . . . I was able to escape to here . . . and have that second chance.27 Stahl made the most of his second chance. In July 1992, Meselson and Stahl sat in the living room of Meselson’s elegant nineteenth century summer home in Woods Hole, both

A FTERWORD

447

reading a draft of one of the chapters of the present book. After we had discussed various specific points, I asked whether reading it had conjured up any images of them actually being back at Caltech doing the things related in the narrative. Matt answered, It brought it all back. I can remember being in that world—imperfect, but a time machine. Realize how lucky we were. What a great place it was, great time. It makes me sad. Frank added, It brought back something to me. It makes me very glad that I left when I did, because I’ve been able to have Matt’s friendship for a lifetime, and if I’d stayed there I couldn’t have handled it, and that would have been lost. To that thought Matt responded, “It’s a paradox I didn’t fully understand at the time, but I do now.” 28

Abbreviations Used in Notes

Recorded interviews are indicated as follows: Name or initials of person interviewed/FLH, date, number of tape (where there is more than one), pages of typed transcription. (The tapes and transcriptions are presently in the author’s personal file.) Notes of unrecorded interviews are as above, except that the date is preceded by NUI, which designates Notes of Unrecorded Interview. Numbers following citations to the Max Delbru¨ck archive refer to box numbers: MD, 3.29. All words italicized in passages quoted in the text are emphasized in the original sources. CSHS CTA FWS JDW MD MM MMmc PNAS UL

Cold Spring Harbor Symposia on Quantitative Biology Archives of the California Institute of Technology Franklin W. Stahl, personal file James D. Watson archive, Cold Spring Harbor Max Delbru¨ck archive, CTA Matthew Meselson, personal file Comment by Meselson in margin of previous draft Proceedings of the National Academy of Sciences Ultracentrifuge Log, MM

Notes

Introduction 1. J. D. Watson and F. H. C. Crick, “A Structure for Deoxyribose Nucleic Acid,” Nature 171 (1953): 737. 2. Gunther S. Stent, “The DNA Double Helix and the Rise of Molecular Biology,” in James D. Watson, The Double Helix: A Personal Account of the Discovery of the Structure of DNA, ed. Gunther S. Stent (New York: W. W. Norton, 1980), p. xviii. 3. P. Medawar, “Lucky Jim,” in Ibid., p. 218; Michel Morange, Histoire de la biologie mole´culaire (Paris: Editions la De´couverte, 1994), p. 152. 4. J. D. Watson and F. H. C. Crick, “Genetical Implications of the Structure of Deoxyribonucleic Acid,” Nature 171 (1953): 967. 5. Horace Freeland Judson, The Eighth Day of Creation: The Makers of the Revolution in Biology (New York: Simon and Schuster, 1979), pp. 186– 93. 6. Morange, Biologie mole´culaire, pp. 152–55. 7. Hans-Jo¨rg Rheinberger, Toward a History of Epistemic Things: Synthesizing Proteins in the Test Tube (Stanford: Stanford University Press, 1997). 8. These reminiscences are most concentrated in Phage and the Origins of Molecular Biology, ed. John Cairns, Gunther Stent, and James D. Watson (Cold Spring Harbor, N.Y.: Cold Spring Harbor Laboratory, 1966). 9. For descriptions and interpretations of the role of Delbru¨ck and the phage group, see Judson, Eighth Day; Morange, Biologie mole´culaire, pp. 55– 68; Robert Olby, The Path to the Double Helix (Seattle: University of Washington Press,1974), pp. 225–40; Lily E. Kay, The Molecular Vision of Life: Caltech, the Rockefeller Foundation, and the Rise of the New Biology (New York: Oxford University Press, 1992), pp. 243–56; Pnina Abir-Am, “Themes, Genres, and Orders of Legitimation in the Consolidation of New Scientific Disciplines: Deconstructing the Historiography of Molecular Biology,” History of Science 23 (1985): 73–117; Muriel Lederman and Sue E. Tolin, “OVATOOMB: Other Viruses and the Origins of Molecular Biology,” Journal

450

N OTES TO

PAGES 5 – 14

of the History of Biology 26 (1993): 235–54; and William C. Summers, “How Bacteriophage Came to Be Used by the Phage Group,” Ibid., pp. 255–268. 10. See, among others, The Pauling Symposium, ed. Ramesh S. Krishnamurthy (Corvallis: Oregon State University Libraries, 1996); Judson, Eighth Day, pp. 70–93; Olby, Path, pp. 267–95; Kay, Molecular Vision, pp. 147–90, 256–64; John Servos, Physical Chemistry from Ostwald to Pauling: The Making of a Science in America (Princeton: Princeton University Press, 1990), pp. 275–98; Mary Jo Nye, Before Big Science: The Pursuit of Modern Chemistry and Physics, 1800–1940 (London: Twayne, 1996), pp. 178–88. 11. Matthew Meselson, “Linus Pauling as an Educator,” in Pauling Symposium, pp. 91–101. 12. Watson, Double Helix. The original edition was published by Atheneum Press in 1968. See also Olby, Path, pp. 297–423; Judson, Eighth Day, pp. 97–186. 13. Michel Morange to Jean Black, October 1999. 14. Ernst Peter Fischer has invoked the Meselson-Stahl experiment as a ¨ sprime example of “the elegant experiment” in Das Scho¨ne und das Biest: A thetische Momente in der Wissenschaften (Munich: Piper, 1997), pp. 46–49. 15. [anonymous] to Jean Black, August 1999.

Chapter One. The Replication Problem 1. J. D. Watson and F. H. C. Crick, “A Structure for Deoxyribose Nucleic Acid,” Nature 171 (1953): 737. 2. Francis Crick, What Mad Pursuit: A Personal View of Scientific Discovery (New York: Basic Books, 1988), p. 66. 3. James D. Watson, The Double Helix (New York: Atheneum, 1968), p. 220. 4. JDW/FLH, 5 March 1990. 5. Watson to Delbru¨ck, 12 March 1953, MD, 23.22. 6. Watson, Double Helix, pp. 167–69, 181–82. For further discussions of these episodes, see Robert Olby, The Path to the Double Helix (Seattle: University of Washington Press, 1974), pp. 395–405; and Horace Freeland Judson, The Eighth Day of Creation (New York: Simon and Schuster, 1979), pp. 158– 67. After the publication of the double helix, questions were raised about the ethics of the actions of Wilkins and Perutz. Both defended their conduct in letters to the editor of Science in 1969, and Watson apologized for presenting the episode in a way that implied, incorrectly, that the information was confidential. See Watson, Double Helix, ed. Gunther S. Stent (New York: Norton, 1980), pp. 207–12. 7. Watson and Crick, “Deoxyribose Nucleic Acid,” p. 737. 8. Berta Gandelman, Stephen Zamenhof, and Erwin Chargaff, “The Desoxypentose Nucleic Acids of Three Strains of Escherichia Coli,” Biochimica

N OTES TO

PAGES 14 – 21

451

et Biophysica Acta 9 (1952): 399–401; G. R. Wyatt, “The Nucleic Acids of Some Insect Viruses,” Journal of General Physiology 36 (1952): 201–5. 9. Watson to Delbru¨ck, 22 March 1953, Delbru¨ck Coll., MD 23.22; JDW/ FLH, 5 March 1990, p. 12. 10. Watson to Delbru¨ck, 22 March 1953. 11. Delbru¨ck to Bohr, 14 April 1953, MD, 3.29. 12. Delbru¨ck to Watson, 14 April 1953, MD, 23.22. 13. Watson and Crick, “Deoxyribose Nucleic Acid,” p. 737. 14. Watson to Delbru¨ck, 12 March 1953. 15. Delbru¨ck to Watson, 14 April 1953. 16. C. L. Huskins, “The Coiling of Chromonemata,” in Genes and Chromosomes: Structure and Organization, CSHS, vol. 9 (Cold Spring Harbor, N.Y.: Biological Laboratory, 1941), p. 13. 17. Ibid., pp. 13–18. 18. Delbru¨ck to Watson, 14 April 1953. 19. Ibid. 20. Watson, Double Helix, pp. 208–12. 21. Rosalind E. Franklin and R. G. Gosling, “Molecular Configuration in Sodium Thymonucleate,” Nature 171 (1953): 740–41. 22. Ibid; Watson and Crick, “Deoxyribose Nucleic Acid;” M. H. F. Wilkins, A. R. Stokes, and H. R. Wilson, “Molecular Structure of Deoxypentose Nucleic Acids,” Nature 171 (1953): 738–40. 23. JDW/FLH, March 5, 1990, pp. 11–12. 24. Watson to Delbru¨ck, 25 April 1953, MD, 23.22. 25. Ibid. 26. J. D. Watson, “Growing Up in the Phage Group,” in Phage and the Origins of Molecular Biology, ed. John Cairns, Gunther S. Stent, and James D. Watson (Cold Spring Harbor: Laboratory of Quantitative Biology, 1966), pp. 239–45. 27. Watson to Delbru¨ck, 5 May 1953, MD, 23.22. 28. J. D. Watson and F. H. C. Crick, “Genetical Implications of the Structure of Deoxyribonucleic Acid,” Nature 171 (1953): 964–67. 29. Ibid., p. 965. 30. Watson, “Growing Up,” p. 243. 31. JDW/FLH, 5 March 1980, [p. 10]. 32. Watson and Crick, “Genetical Implications,” pp. 965–66. 33. Ibid., p. 966. For emphasis, I have rearranged the order in which these phrases occur in the original. 34. Delbru¨ck to Watson, 28 January 1953, 4 March 1953, 29 March 1953; Watson to Delbru¨ck, 13 April 1953; Weaver to Delbru¨ck, 21 April 1953, MD, 23.22; Max Delbru¨ck, “Introductory Remarks about the Program,” in Viruses, CSHS, vol. 18 (1953), pp. 1–2. 35. Delbru¨ck to Watson, 1 May 1953, MD.

452

N OTES TO

PAGES 22– 33

36. Delbru¨ck to Watson, 12 May 1953, MD. 37. Watson to Delbru¨ck, 21 May 1953, MD. 38. J. D. Watson and F. H. C. Crick, “The Structure of DNA,” in Viruses, CSHS, vol. 18 (1953), p. 123. 39. Ibid., pp. 123–27. 40. Ibid., pp. 127–29. 41. Ibid., p. 128. 42. JDW/FLH, 5 March 1990, [pp. 14–15]. 43. Watson and Crick, “Structure of DNA,” pp. 128–29. 44. Ibid., p. 129. 45. Ibid. 46. J. D. Watson, Molecular Biology of the Gene, 2d. ed. (New York: Benjamin, 1970), pp. 66–67. 47. Watson, “Growing Up,” p. 239. 48. JDW/FLH, 5 March 1990 [p. 12]. 49. Watson to Delbru¨ck, 21 May 1953, MD. 50. M. Demerec, “Foreword,” Viruses, CSHS, vol. 18 (1953), p. v. 51. Delbru¨ck to Watson, 1 May 1953, MD. 52. Franc¸ois Jacob, La Statue inte´rieure (Paris: Odile Jacob, 1987), p. 301. 53. Stahl to Holmes, May 14, 1997. 54. JDW/FLH, 5 March 1990 [p. 14]. 55. Sinsheimer to Delbru¨ck, 17 June 1953, MD, 20.3. 56. Ibid. 57. Pauling to Watson, 29 June 1959; JDW/FLH, 5 March 1990, [p. 15]; Watson to Delbru¨ck, 22 March 1953, May 5 1953, MD; F. H. C. Crick and J.D. Watson, “The Complementary Structure of Deoxyribonucleic Acid,” Proceedings of the Royal Society of London, ser. A, 223 (1954): 80–96. 58. Watson and Crick, “Complementary Structure,” p. 81. 59. Ibid., pp. 85, 86. 60. Ibid., p. 90; S. Furberg, “The Crystal Structure of Cytidine,” Acta Crystallographica 3 (1950): 325–33, esp. 330. 61. Watson and Crick, “Complementary Structure,” pp. 87–93. 62. Ibid., p. 86. 63. Ibid., p. 94. 64. JDW/FLH, 5 March 1990, [p. 16]; Watson to Delbru¨ck, 20 February 1953, 20 May 1952, MD. 65. Watson to Delbru¨ck, 20 May 1952, MD; Alexander Rich and J. D. Watson, “Physical Studies on Ribonucleic Acid,” Nature 173 (1954): 995; JDW/ FLH, 5 March 1990 [p. 17]. 66. Alexander Rich and J. D. Watson, “Some Relations Between DNA and RNA,” PNAS 40 (1954): 759–63, esp. 762; quotation is from JDW/FLH, 5 March 1990 [p. 17]. 67. Watson to Crick, quoted in Judson, Eighth Day, p. 262.

N OTES TO

PAGES 33 – 42

453

68. California Institute of Technology, Biology Annual Report, 1954, pp. 157–58. 69. JDW/FLH, 8 March 1990, [p. 18]; Beadle to Selective Service Board, 22 September 1953; Delbru¨ck to National Foundation for Infantile Paralysis, 8 October 1953; NFID to Selective Service, 15 October 1953, MD; Kendrew to Watson, 8 November 1953, JDW. 70. JDW/FLH, 8 March 1990 [p. 19]. 71. Ibid., [pp. 19, 23]. 72. Ibid.; Ernst Peter Fischer and Carol Lipson, Thinking About Science: Max Delbru¨ck and the Origins of Molecular Biology (New York: Norton, 1988), pp. 236–39; Cal Tech Biology Annual Report, 1954, p. 156; Delbru¨ck to Bohr, 1 Dec. 1954, MD, 3.29. 73. JDW/FLH, 8 March 1990 [pp. 19, 23–24]; Fischer and Lipson, Thinking About Science, p. 234. 74. JDW/FLH, 8 March 1990 [pp. 19, 23–24]; Fischer and Lipson, Thinking About Science, p. 234. 75. JDW/FLH, 8 March 1990 [pp. 17, 19]; Watson to Delbru¨ck, January 4, 1953 [actually 1954], MD; Watson to Kendrew, 22 January 1954, Kendrew Archive, National Catalog Unit for the Archives of Contemporary Scientists, Bath, U.K. 76. Judson, Eighth Day, pp. 262–63. 77. Ibid., pp. 263–64. 78. Watson to Delbru¨ck, 25 March 1954, MD. 79. Rich and Watson, “Physical Studies,” pp. 995–96. 80. Rich and Watson, “Relations Between DNA and RNA,” p. 759. 81. Ibid., pp. 759–60. 82. JDW/FLH, 8 March 1990 [pp. 25–26]. 83. Rich and Watson, “Relations Between DNA and RNA,” p. 760. 84. Ibid. 85. Ibid., pp. 760–61. 86. Ibid., pp. 761–763. 87. Ibid., p. 763. 88. Watson to Delbru¨ck, 1 June 1954, MD. 89. Watson to Delbru¨ck, 25 March, 1954, 1 June 1954. 90. Thomas S. Kuhn, The Structure of Scientific Revolutions, 2d ed. (Chicago: University of Chicago Press, 1970), p. 189. 91. Kendrew to Watson, 6 June, 1954, JDW. 92. Kendrew to Watson, 8 November, 1953. 93. JDW/FLH, 8 March 1990 [p. 25]. 94. Fischer and Lipson, Thinking About Science, p. 263. 95. Max Delbru¨ck, “On the Replication of Deoxyribonucleic Acid (DNA),” PNAS 40 (1954): 783–88, on 784. 96. Ibid., p. 785.

454

N OTES TO

PAGES 43– 53

97. Ibid., pp. 787–88. 98. Gunther S. Stent, “Mortality Due to Radioactive Phosphorus as an Index to Bacteriophage Development,” in Viruses, CSHS, vol. 18 (1953), 255– 59. 99. Ibid., p. 258. 100. Stent to Delbru¨ck, 27 March 1954, MD. 101. Stent to Delbru¨ck, 11 May 1954, MD. 102. Ibid. See C. A. Dekker and H. K. Schachman, “On the Macromolecular Structure of Deoxyribonucleic Acid: An Interrupted Two-Strand Model,” PNAS 40 (1954): 894–909. 103. Stent to Delbru¨ck, 11 May 1954, MD. 104. Ibid. 105. Delbru¨ck, “Replication,” p. 788. 106. JDW/FLH, 8 March 1990 [p. 24].

Chapter Two. Meselson and Stahl 1. MM/FLH, 2 December 1987, pp. 8–9; 5 February 1988, pp. 38–39. 2. MM/FLH, 2 December 1987, pp. 8–9; 5 February 1988, pp. 38–39; J. D. Watson and F. H. C. Crick, “A Structure for Deoxyribose Nucleic Acid,” Nature 171 (1953): 737–38; M. H. F. Wilkins, A. R. Stokes and H. R. Wilson, “Molecular Structure of Deoxypentose Nucleic Acids,” ibid., pp. 738–40; Rosalind E. Franklin and R. G. Gosling, “Molecular Configuration in Sodium Thymonucleate,” ibid., pp. 740–41; J. D. Watson and F. H. C. Crick, “Genetical Implications of the Structure of Deoxyribonucleic Acid,” ibid., pp. 964–67; MM/FLH, 14 July 1998, 1:5. 3. MM/FLH, 5 February 1988, pp. 12–18. 4. Ibid., pp. 19–25. 5. Ibid., pp. 26–28. 6. Ibid., pp. 30–31. 7. Ibid., pp. 32–34; December 2, 1987, pp. 21–22. 8. R. B. Corey and L. Pauling, “Fundamental Dimensions of Polypeptide Chains,” Proceedings of the Royal Society of London B141 (1953): 10–20; L. Pauling and R. B. Corey, “Stable Configurations of Polypeptide Chains,” ibid., pp. 21–33. 9. MM/FLH, 5 February 1988, pp. 36–38; April 19, 1993, pp. 32–33. 10. California Institute of Technology, Annual Report, 1953–54, p. 89. 11. [Meselson], “Gates and Crellin Laboratories of Chemistry Workbook Serial No. 583,” pp. 5–7. 12. MM/FLH, 2 December 1987, p. 5, 20 May 1992, 1:2–3; Bernardino Fantini, “Monod, Jacques Lucien,” Dictionary of Scientific Biography, Supp. II, ed. Frederic L. Holmes (New York: Scribner’s, 1990), pp. 636–49; Jacques

N OTES TO

PAGES 53 – 57

455

Monod and Melvin Cohn, “La Biosynthe`se induite des enzymes (Adaptation enzymatique),” Advances in Enzymology 8 (1952): 67–119; Melvin Cohn and Jacques Monod, “Specific Inhibition and Induction of Enzyme Biosynthesis,” in Adaptation in Micro-Organisms, ed. R. Davies and E. F. Gale (Cambridge: Cambridge University Press, 1953), pp. 132–49; quotation is from p. 145. 13. Jacques Monod, “Jessup Lectures,” Service des Archives de l’Institut Pasteur, Fonds Monod, Mon. MSS. 03, nos. 2, 1–8, quotations from pp. 179– 81. 14. Monod to Bonner, 1 December 1953, Archives de l’Institut Pasteur, Fonds Monod, MON. Cor. 02; Monod to Stanier, 22 December 1953, ibid., Cor. 16. 15. Annual Report of the Division of Biology, California Institute of Technology, 1954, p. 160; MM/FLH, 2 December 1987, p. 5. 16. MM/FLH, 2 December 1987, p. 5. 17. Ibid., pp. 5–6; MM, FWS/FLH, 12 July 1992, 3:6–7. 18. MM/FLH, 19 April 1993, 2:17–18; Workbook 583, pp. 8–13. 19. MM/FLH, 5 February 1988, pp. 39–42. 20. MM/FLH, 2 December 1987, p. 12; quotation is from Watson to Stanley, 1 March 1960, JDW. Despite the retrospective nature of these comments, and the fact that Watson’s opinion is part of a recommendation written for Meselson, there appears no reason to doubt their accuracy. 21. Watson to Delbru¨ck, 25 March 1954, MD, 23.23; JDW/FLH, March 5, 1990, p. 4. 22. J. D. Watson, “Growing Up in the Phage Group,” in Phage and the Origins of Molecular Biology, ed. John Cairns, Gunther S. Stent, and James D. Watson (Cold Spring Harbor, N.Y.: Cold Spring Harbor Laboratory of Quantitative Biology, 1966), pp. 239–45. 23. Michel Morange, Histoire de la biologie mole´culaire (Paris: Editions la De´couverte, 1994), pp. 55–68. 24. A. D. Hershey, “Mutation of Bacteriophage with Respect to Type of Plaque,” Genetics 31 (1946): 620–40; A. D. Hershey and Raquel Rotman, “Genetic Recombination Between Host-Range and Plaque-Type Mutants of Bacteriophage in Single Bacterial Cells,” Genetics 34 (1948): 44–71. 25. A. D. Hershey and Martha Chase, “Genetic Recombination and Heterozygosis,” Genes and Mutations, CSHS, vol. 16 (1951), 471–72. 26. For an integrated interpretation of these events as they were understood in 1952, see S. E. Luria, “An Analysis of Bacteriophage Multiplication,” in The Nature of Virus Multiplication: Second Symposium of the Society for General Microbiology, ed. Paul Fildes and W. E. van Heyningen (Cambridge: Cambridge University Press, 1953), pp. 99–112. 27. A. D. Hershey, “The Injection of DNA into Cells by Phage,” in Phage and the Origins, pp. 100–8.

456

N OTES TO

PAGES 58– 64

28. A. D. Hershey and Martha Chase, “Independent Functions of Viral Protein and Nucleic Acid in Growth of Bacteriophage,” Journal of General Physiology 36 (1952–1953): 39–49. 29. Ibid., pp. 49–54. 30. Ibid., p. 54. 31. Robert Olby, The Path to the Double Helix (Seattle: University of Washington Press, 1974),p. 319; Horace Freeland Judson, The Eighth Day of Creation (New York: Simon and Schuster, 1979),p. 130. 32. Stent to Holmes, 29 December 1997. Stent was present at this meeting. 33. Luria, “Bacteriophage Multiplication” (Discussion), pp. 113–16. Watson’s comments as recorded in this volume may not be exactly the same as what he said during the actual discussion, because the discussions were not directly recorded but reconstructed from forms on which the participants were asked to write down what they had said. See Fildes and van Heyningen, Virus Multiplication, p. ix. Gunther Stent, who was present, does not remember Watson having claimed that his and Maaløe’s results paralleled those of Hershey and Chase. Stent to Holmes, 29 December 1997. 34. James D. Watson, The Double Helix (New York: Atheneum), p. 119; Stent to Holmes, 29 December 1997; Cairns to Elworthy,19 May 1997. 35. Olby, Path, p. 319. 36. Hershey and Chase, “Independent Functions,” p. 56. 37. Olby, Path, p. 318. 38. Watson, Double Helix, p. 119. 39. Morange, Histoire, pp. 65–68. 40. Watson to Delbru¨ck, 1 June 1954, MD. 41. MM/FLH, 5 February 1988, pp. 46–47; D. O. Jordan, “The Physical Properties of Nucleic Acids,” in The Nucleic Acids, ed. Erwin Chargaff and J. N. Davidson (New York: Academic Press, 1955) 1: 475–80. 42. MM/FLH, 5 February 1988, p. 47. 43. MM/FLH, 21 November 1989, pp. 5–6; FWS/FLH, 21 November 1988, p. 3. 44. MM/FLH, 21 November 1989, pp. 5–6; FWS/FLH, 21 November 1988, p. 3; Matthew Meselson and Franklin W. Stahl, “Demonstration of Semiconservative Mode of DNA Replication,” in Phage and the Origins, p. 246. 45. FWS/FLH, 21 November 1988, pp. 1–2, 73–74. 46. Ibid., pp. 74–75. 47. Ibid., pp. 75–76. 48. Ibid., pp. 2,76. 49. A. H. Doermann, “The Intracellular Growth of Bacteriophages,” Journal of General Physiology 35 (1952): 645. 50. A. H. Doermann and M. B. Hill, “Genetic Structure of Bacteriophage T4 as described by Recombination Studies of Factors Influencing Plaque Morphology,” Genetics 38 (1953): 79–90.

N OTES TO

PAGES 64 – 73

457

51. A. H. Doermann, Martha Chase, and Franklin W. Stahl, “Genetic Recombination and Replication in Bacteriophage,” Journal of Cellular and Comparative Physiology 45 (1955): 51–60. 52. Ibid., pp. 60–67. 53. Ibid., p. 72. 54. FWS/FLH, 21 November 1988, p. 2, 4 May 1992, pp. 15–16. 55. FWS/FLH, 21 November 1998, p. 3. 56. Ibid., p. 4; Meselson and Stahl, “Demonstration of Semi-Conservative Mode,” p. 246. 57. FWS/FLH, 21 November 1988, pp. 3–4. 58. Ibid., p. 4; MM/FLH, 2 December 1987, p. 14. 59. FWS/FLH, 22 November 1988, pp. 85–86, 91. 60. FWS/FLH, 21 November 1988, p. 10. 61. Ibid., p. 40. 62. MM/FLH, 2 December 1987, pp. 11, 13. 63. J. D. Watson, “Biological Consequences of the Complementary Structure of DNA,” Journal of Cellular and Comparative Physiology 45 (1955): Supp. 24, p. 110. 64. FWS/FLH, 21 November 1988, pp. 3–4; Drake to Holmes, 16 September 1997. 65. FWS/FLH, 21 November 1988, pp. 5–6, 22 November 1990, pp. 81– 82. 66. Franklin W. Stahl and Mathew Messelson [sic], “A Formulation of the Theory by Doerman [sic], Chase, and Stahl of the Cross Reactivation of Markers by UV’d Phage,” FWS. 67. Meselson to Stahl, 17 August 1954, FWS. 68. Ibid.; FWS/FLH, 22 November 1988, p. 42, 4 May 1992, p. 19. In recalling this comment in 1992, Stahl pointed out that “subsequent analysis supports” the view developed in the manuscript. 69. Franklin W. Stahl, “The Effects of the Decay of Incorporated Radioactive Phosphorus on the Genome of Bacteriophage T4,” Virology 2 (1956): 206– 12; FWS/FLH, 21 November 1988, p. 5. 70. Stahl, “Effects,” pp. 212–21; FWS/FLH, NUI, 4 May 1992, p. 1. 71. Meselson to Stahl, 10 January 1955, FWS. 72. MM/FLH, 20 May 1992, 2:4–5. 73. Linus Pauling, General Chemistry (New York: Dover Publications, 1988), p. 886. 74. MM/FLH, 19 May 1993, 2:19–20; Linus Pauling, The Nature of the Chemical Bond, 3d. ed. (Ithaca: Cornell University Press, 1960), p. 281. The second edition of the book, which Meselson would have consulted, did not state, as does the third, that the peptide bond is “completely planar.” 75. [Meselson,] Workbook no. 583, pp. 21–22. 76. Meselson to Stahl, 10 January 1955; Robert E. Demars, “Genetic Re-

458

N OTES TO

PAGES 74– 80

combination of UV Irradiated Phage T2,” Biology Annual Report, California Institute of Technology, 1954, pp. 151–52. 77. Meselson to Stahl, 10 January 1955. 78. M. Meselson, “Propositions” (Examination for Ph.D. Candidacy), pp. 2–3, MM. 79. Ibid., p. 1. 80. FWS/FLH, 21 November 1988, pp. 6, 44–45.

Chapter Three. Twists and Turns 1. John R. Platt, “Possible Separation of Intertwined Nucleic Acid Chains by Transfer-Twist,” PNAS 41 (1955): 181. 2. Ibid. 3. G. Gamow, “Possible Relation Between Deoxyribonucleic Acid and Protein Structures,” Nature 193 (1954): 318; Horace Freeland Judson, The Eighth Day of Creation: Makers of the Revolution in Biology (New York: Simon and Schuster, 1979), pp. 250–53. 4. G. Gamow, “Topological Properties of Coiled Helical Systems,” manuscript intended for PNAS, MD, 8.21. 5. Ibid. 6. Max Delbru¨ck and Gunther S. Stent, “On the Mechanism of DNA Replication,” in A Symposium on the Chemical Basis of Heredity, ed. William D. McElroy and Bentley Glass (Baltimore: Johns Hopkins University Press, 1972), p. 702. I make the assumption here that Delbru¨ck would have expressed privately to Gamow when he received this manuscript the same opinion that he later expressed in public. 7. G. Gamow, “Topological Properties of Coiled Helical Systems,” PNAS 41 (1955): 7. 8. Brenner to Watson, 12 April 1955, JDW. 9. Raper to Delbru¨ck, 2 March 1955, MD, 23.23. 10. Luria to Watson, 25 April 1955, JDW; J. D. Watson/FLH, 5 March 1990, p. 19. 11. JDW/FLH, 5 March 1990, p. 20. 12. Delbru¨ck to Bresch, 21 April 1955, MD, 4.20. 13. JDW/FLH, 5 March 1990, p. 20. 14. Gunther S. Stent, Nazis, Women, and Molecular Biology: Memoirs of a Lucky Self-Hater (Kensington, Calif.: Briones Books, 1998), pp. 289–369. GS/FLH, 5 May 1992, 1:6–7. 15. GS/FLH, 6 May 1992, 1–2:39–42. 16. Ibid., 2:42–46; Stent to Holmes, 19 February 1998. For an overview of Pontecorvo’s views on genes and their replication, see G. Pontecorvo, “Genetic Formulation of Gene Structure and Gene Action,” in Advances in Enzymology 13 (1952): 121–49.

N OTES TO

PAGES 81 – 91

459

17. Andre´ Lwoff, “The Nature of Phage Reproduction,” in The Nature of Virus Multiplication (Second Symposium of the Society for General Microbiology, Oxford University, April 1952) (Cambridge: Cambridge University Press, 1953), pp. 173–74. 18. Ibid., p. 162. 19. For the origins of the Virus Lab, see Angela N. H. Creager, “Wendell Stanley’s Dream of a Free-Standing Biochemistry Department at the University of California, Berkeley,” Journal of the History of Biology 29 (1996): 331–60. 20. Stent, “Autobiography,” p. xx–8; GS/FLH, 6 May 1992, 2:46. 21. GS/FLH, 5 May 1992, 1:15. 22. Stent to FLH, 11 March 1992. 23. GS/FLH, 7 May 1992, 1:18–20; Gunther S. Stent and Niels K. Jerne, “The Distribution of Parental Phosphorus Atoms Among Bacteriophage Progeny,” PNAS 41 (1955): 704–9. 24. Ibid., pp. 704–5. 25. Ibid., pp. 705–6. 26. Ibid., pp. 707–9. 27. GS/FLH, 5 May 1992, 1:6. 28. FWS/FLH, 21 November 1988, p. 5. 29. Franklin W. Stahl, “The Effects of the Decay of Incorporated Radioactive Phosphorus on the Genome of Bacteriophage T4,” Ph.D. diss., University of Rochester, 1955, pp. 14–19; published in Virology 2 (1956): 206–34, on pp. 212–16. 30. Max Delbru¨ck, “The Burst Size Distribution in the Growth of Bacterial Viruses (Bacteriophages),” Journal of Bacteriology 50 (1945): 131–35. 31. Stahl, “Effects of the Decay,” diss. pp. 19–27; Virology, pp. 217–23. 32. Ibid., diss. pp. 34–44, Virology pp. 227–31. 33. Ibid., diss. pp. 22–23, 41–42, 49. 34. Barry Honing, “In Memoriam: Cyrus Levinthal,” Proteins 11 (1991): 239–41. 35. Cyrus Levinthal, “Recombination in Phage T2: Its Relationship to Heterozygosis and Growth,” Genetics 39 (1954): 169–70. 36. Ibid., pp. 170–79. 37. Ibid., p. 183. 38. Stahl, “Effects of the Decay,” diss. pp. 42–43. 39. FWS/FLH, 21 November 1988, pp. 5–6, 80; 12 July 1992, 3:13, 21– 22; Stahl to Holmes, 14 May 1997. 40. FWS/FLH, 21 November 1988, pp. 5–6, 80. 41. Ibid., pp. 6–7; MM, FWS/FLH, 9 May 1997, 2:22. 42. FWS/FLH, 21 November 1988, p. 49; MM/FLH, 21 November 1988, p. 37; MM, FWS/FLH, 18 May 1993, 2:9–10. 43. MM/FLH, 2 December 1987, pp. 5–6. 44. FWS/FLH, 21 November 1988, p. 7.

460

N OTES TO

PAGES 91– 95

45. G. Bertani, “The Role of Phage in Bacterial Genetics,” Brookhaven Symposia in Biology, ed. R. C. King, vol. 8 (1956), pp. 50–51. See also Joshua Lederberg, “Recombination Mechanisms in Bacteria,” Symposium on Genetic Recombination (Oak Ridge: Oak Ridge National Laboratory, 1954), pp. 81–83; Bertani/FLH, 25 February 2000, Notes. 46. Bertani, “Role of Phage,” pp. 51–55. 47. Ibid., p. 52; Joshua Lederberg and E. L. Tatum, “Gene Recombination in Escherichia coli,” Nature 158 (1946): 558; L. L. Cavalli, J. Lederberg, and Esther M. Lederberg, “An Infective Factor Controlling Sex Compatibility in Bacterium coli,” Journal of General Microbiology 8 (1953): 89–102; E.-L. Wollman, F. Jacob, and W. Hayes, “Conjugation and Genetic Recombination in Escherichia coli K-12,” in “Genetic Mechanisms: Structure and Function,” CSHS 21 (1956), pp. 141–62. 48. Wollman, Jacob, and Hayes, “Conjugation,” pp. 142,157–60; FWS/ FLH, 21 November 1988, pp. 7, 59–60; 4 May 1992, NUI, pp. 1–2; MM, FWS/ FLH, 9 May 1997, 2:25–28; Bertani/FLH, 25 February 2000, Notes. 49. FWS/FLH, 4 May 1992, NUI, p. 2; Stahl to Holmes, 12 April 1997. 50. Levinthal to Stahl, 14 October 1955; Stahl to Levinthal, 18 October 1955, FWS. 51. Stahl to Levinthal, 18 October 1955; F. Stahl and G. Streisinger, “The Frequency Distribution of Rare Recombinants in Single Cell Bursts,” in Division of Biology, California Institute of Technology, “Semi-Annual Report to the National Foundation for Infantile Paralysis, July-December 1955,” FWS, p. 12; Seymour Benzer, “Fine Structure of a Genetic Region in Bacteriophage,” PNAS 41 (1955): 344–54. 52. Stahl and Streisinger, “Frequency Distribution,” pp. 12–13. 53. Stahl to Levinthal, 18 October 1955; Levinthal to Stahl, 9 November 1955, 28 December 1955, FWS; Stahl, “Effects of Decay,” Virology 2 (1956): 206–34. 54. Levinthal to Stahl, 8 November 1955. 55. S. E. Luria, “The Frequency Distribution of Spontaneous Bacteriophage Mutants as Evidence for the Exponential Rate of Phage Reproduction,” CSHS, vol. 16 (1951), pp. 463–70; Gunther S. Stent, Molecular Biology of Bacterial Viruses (San Francisco: Freeman, 1963), pp. 193–95. 56. N. Visconti and M. Delbru¨ck, “The Mechanism of Genetic Recombination in Phage,” Genetics 38 (1953): 5–33. 57. Charley Steinberg and Frank Stahl, “The Clone-Size Distribution of Mutants Arising from a Steady-State Pool of Vegetative Phage,” Journal of Theoretical Biology 1 (1961): 488–89. 58. Levinthal to Stahl, 14 October 1955. 59. Cairns to Elworthy, 19 May 1997. 60. Stahl to Levinthal, 23 July 1956; Stahl to Holmes, 12 April 1997, 23 February 1998.

N OTES TO PAGES 95– 105

461

61. FWS/FLH, 21 November 1988, p. 78. 62. Ibid., pp. 38–39, 78–79. 63. Ibid., p. 79. 64. MM, FWS/FLH, 18 May 1993, 2:10. 65. Ibid., pp. 83–84; MM, FWS/FLH, 9 May 1997, 2:29; Drake to Holmes, 23 August 1997. 66. FWS/FLH, 21 November 1988, pp. 7, 49, 60. 67. FWS/FLH, 12 July 1992, 3:13–14; Drake to Holmes, 23 August 1997. 68. Meselson, Workbook 583, pp. 24–61. 69. Ibid., pp. 61–84. This nominal summary of Meselson’s progress during 1955 is intended only to give a glimpse of the kinds of problems with which he dealt, without entering into the very complicated methods of X-ray diffraction analysis. 70. Meselson to Watson, 6 December 1955, JDW. Nine months after Meselson wrote down this plan, Vernon Ingram reported on his landmark investigation of the chemical difference between normal hemoglobin and sickle-cell anemia hemoglobin. See V. M. Ingram, “A Specific Chemical Difference Between the Globins of Normal Human and Sickle-Cell Anaemia Haemoglobin,” Nature 178 (1956): 792–94. 71. Meselson to Watson, 6 December 1955, JDW. 72. Niels Arley, “The Duplication Mechanism of Deoxyribonucleic Acid,” Nature 176 (1955): 465–66; G. Gamow, “Possible Relation Between Deoxyribonucleic Acid and Protein Structures,” Nature 193 (1954): 318. 73. David P. Bloch, “A Possible Mechanism for the Replication of the Helical Structure of Desoxyribonucleic Acid,” PNAS 41 (1955): 1058–64. 74. Ibid. 75. Delbru¨ck to Stent, 13 December 1955, MD. 76. Stent to Delbru¨ck, 14 December 1955, MD. 77. GS/FLH, 5 May 1992, 2:47. Stent now regards Fuerst as the best graduate student he has had. A Canadian, Fuerst went to the Institut Pasteur after completing his Ph.D. and later moved into the field of animal cancer viruses. 78. Clarence R. Fuerst and Gunther S. Stent, “Inactivation of Bacteria by Decay of Incorporated Radioactive Phosphorus,” Journal of General Physiology 40 (1956): 73–90. 79. Ibid., p. 84. 80. Ibid. 81. Ibid., p. 86. 82. Ibid., p. 87. 83. GS/FLH, 6 May 1992, 2:52. 84. Cyrus Levinthal, “The Mechanism of DNA Replication and Genetic Recombination in Phage,” PNAS 42 (1956): 394–404. 85. Ibid., pp. 394–96. 86. Ibid., pp. 400–1.

462

N OTES TO

PAGES 105 –115

87. Ibid. 88. Ibid., pp. 401, 403. 89. Ibid., pp. 396, 402. 90. Ibid., p. 403. 91. Ibid., p. 395. 92. C. Levinthal and H. R. Crane, “On the Unwinding of DNA,” PNAS 42 (1956): 436. 93. Ibid., pp. 437–438. 94. Max Delbru¨ck and Gunther Stent, “On the Mechanism of DNA Replication,” in A Symposium on the Chemical Basis of Heredity, ed. William D. McElroy and Bentley Glass (Baltimore: Johns Hopkins University Press, 1957), pp. 699–736, on p. 699. 95. Ibid., pp. 700–7, 710. 96. GS/FLH, May 6, 1992, 2:50–51. 97. Delbru¨ck and Stent, “Mechanism of DNA Replication,” p. 702. 98. GS/FLH, 5 May 1992, 1:22–23, 6 May 1992, 2:50. 99. Delbru¨ck and Stent, “Mechanism of DNA Replication,” pp. 707–14. 100. Ibid., p. 714. 101. Ibid., p. 719. 102. Ibid., p. 729. 103. Ibid., p. 722. 104. Ibid., p. 735. 105. C. Levinthal and C. A. Thomas Jr., “The Molecular Basis of Genetic Recombination in Phage,” in Chemical Basis of Heredity, pp. 737–42. 106. “Discussion,” in Chemical Basis of Heredity, pp. 743–755. 107. Delbru¨ck to Stanley, 5 August 1956, MD. 108. JDW/FLH, 5 March 1990, pp. 21, 25. 109. James D. Watson, “X-Ray Studies on RNA and the Synthetic Polyribonucleotides,” in Chemical Basis of Heredity, pp. 552–53. 110. Ibid. 111. Ibid., p. 552; JDW/FLH, 5 March 1990, pp. 21, 26–29. 112. Ibid.; Watson, “X-Ray Studies,” pp. 553–55. 113. JDW/FLH, 5 March 1990, p. 29. See also Horace Freeland Judson, The Eighth Day of Creation (New York: Simon and Schuster, 1979), pp. 280–84. 114. Watson, “X-Ray Studies,” p. 555. 115. Ibid., p. 556. 116. James D. Watson, The Double Helix (New York: Atheneum, 1968), pp. 182–88. 117. “Discussion,” in Chemical Basis of Heredity, p. 747. 118. Cairns to Elworthy, 19 May 1997. 119. See Joseph S. Fruton, Molecules and Life (New York: Wiley-Interscience, 1972), pp. 210-15. 120. JDW/FLH, 5 March 1990, p. 20. The triesters in question do not exist.

N OTES TO PAGES 116 –125

463

Chapter Four. Crossing Fields 1. Meselson to Watson, 14 December 1956, FWS. 2. MM/FLH, 21 November 1989, pp. 9–10; Gates and Crellin Laboratories of Chemistry Work Book No. 786, MM. 3. MM/FLH, 19 April 1993, 2:25–26. 4. Ibid., 1:8, 2:28–29. 5. Erwin Chargaff and J. N. Davidson, eds., The Nucleic Acids: Chemistry and Biology (New York: Academic Press, 1955), 1:vii. 6. Aaron Bendich, “Chemistry of Purines and Pyrimidines,” in The Nucleic Acids, 1:114–16. 7. Work Book No. 786, pp. 4–5. 8. D. O. Jordan, “The Physical Properties of Nucleic Acids,” in The Nucleic Acids, 1:447–51; Work Book No. 786, pp. 5–6. 9. Jordan, “Physical Properties,” pp. 450, 452. 10. Work Book No. 786, p. 6. 11. Linus Pauling and Robert B. Corey, “Specific Hydrogen-Bond Formation Between Pyrimidines and Purines in Deoxyribonucleic Acids,” Archives of Biochemistry and Biophysics 65 (1956): 164–81, on 180; MM/FLH, 19 April 1993, 2:30–31. 12. Work Book No. 786, p. 7. 13. Ibid., p. 8; Bendich, “Chemistry of Purines,” pp. 108–9, 113; MM/ FLH, 19 April 1993, 3:4. 14. J. D. Watson and F. H. C. Crick, “The Structure of DNA,” CSHS, vol. 18 (1953): 129–30. 15. MM/FLH, 18 May 1993,1:25–26; 10 May 1997, 3:1; 14 July 1988, 1:1. 16. MM/FLH, 19 April 1993, 1:11; 18 May 1993, 1:4, 17–18. 17. Bendich, “Chemistry of Purines,” pp. 104–7. 18. Work Book No. 786, p. 9. 19. MM/FLH, 18 May 1993, 1:10, 22. 20. MM/FLH, 19 April 1993, 1:8–9; 2:32, 3:16. 21. Adrien Albert, D. J. Brown, and Gordon Cheeseman, “Pteridine Studies. Part II. 6- and 7- Hydroxypteridines and Their Derivatives,” Journal of the Chemical Society 155 (1952): 1620–30, on p. 1622; Work Book No. 786, p. 9. 22. Adrien Albert and L. N. Short, “Absorption Spectra of Acridines,” Journal of the Chemical Society (1945): 760–63, on p. 762. 23. MM/FLH, 19 April 1993, 3:6; 14 July 1998, 1:1. 24. D. J. Brown, “The Simple Pyrimidines,” Reviews of Pure and Applied Chemistry 3 (1953): 115–33. 25. Work Book No. 786, pp. 9–10. 26. Ibid., p. 10. 27. C. J. B. Clews and W. Cochran, “Crystal Structures of 2-Amino-4Methyl-6-Chloropyrimidine and 2-Amino-4, 6-Dichloropyrimidine,” Nature 159 (1947): 264–65.

464

N OTES TO

PAGES 126 –133

28. Linus Pauling, The Nature of the Chemical Bond, 3rd ed. (Ithaca: Cornell University Press, 1960), p. 260; R. B. Corey and L. Pauling, “Fundamental Dimensions of Polypeptide Chains,” Proceedings of the Royal Society B141 (1953): 10–20, on p. 18; Work Book No. 786, p. 10. 29. C. J. B. Clews and W. Cochran, “The Structures of Pyrimidines and Purines. I. A Determination of the Structures of 2-amino-4-methyl-6-chloropyrimidine and 2-amino-4, 6-dichloropyrimidine by X-ray methods,” Acta Crystallographica 1 (1948): 4–11; “Editorial Preface,” Ibid., pp. 1–2. 30. Work Book No. 786, p. 11. 31. MM/FLH, 19 April 1993, 3:9. 32. C. J. B. Clews and W. Cochran, “The Structures of Pyrimidines and Purines. III. An X-ray Investigation of Hydrogen Bonding in Aminopyrimidines,” Acta Crystallographica 2 (1949): 46–57; A. D. Booth, “The Accuracy of Atomic Co-ordinates Derived from Fourier Series in X-Ray Structure Analysis,” Proceedings of the Royal Society A188 (1946): 77–92. 33. Clews and Cochran, “Structures III. Hydrogen Bonding in Aminopyrimidines,” pp. 46–47, 56; Louis Hunter, “Mesohydric Tautomerism,” Journal of the Chemical Society 148 (1945): 806–9. 34. Work Book No. 786, p. 12. 35. Ibid., p. 13. 36. MM/FLH, 10 May 1997, 3:5. 37. G. J. Pitt, “The Crystal Structure of 4, 6-dimethyl-2-hydroxpyrimidine. I,” Acta Crystallographica 1 (1948): 168–74. 38. Work Book No. 786, p. 13; MM/FLH, 19 April 1993, 3:13–14. 39. Work Book No. 786, p. 13. 40. MM/FLH, 19 April 1993, 2:24–25; July 14, 1998, 1:2. 41. James D. Watson, The Double Helix (New York: Athenaeum, 1968), p. 190. 42. Jerry Donohue, “The Hydrogen Bond in Organic Crystals,” Journal of Physical Chemistry 56 (1952): 502–3. 43. Ibid., pp. 503–10. 44. Ibid., p. 509; Work Book No. 786, pp. 14–15. 45. J. Monteath Robertson, Organic Crystals and Molecules: Theory of X-Ray Structure Analysis with Applications to Organic Chemistry (Ithaca: Cornell University Press, 1953), pp. 217–18, 226–27; Work Book No. 786, p. 15. 46. MM/FLH, 19 April 1993, 3:14–15. 47. June M. Broomhead, “The Structures of Pyrimidines and Purines. II. A Determination of the Structure of Adenine Hydrochloride by X-Ray Methods,” Acta Crystallographica 1 (1948): 324–29; Work Book No. 786, p. 16. 48. W. Cochran, “The Structures of Pyrimidines and Purines. V. The Electron Distribution in Adenine Hydrochloride,” Acta Crystallographica 4 (1951): 81–89.

N OTES TO PAGES 133 –142

465

49. Ibid., pp. 85–92. 50. Work Book No. 786, pp. 16–17; MM/FLH, 19 April 1993, 3:16; 18 May 1993, 1:1–2. 51. MM/FLH, 18 May 1993, 1:1–2, 12. 52. June M. Broomhead, “The Structures of Pyrimidines and Purines. IV. The Crystal Structure of Guanine Hydrochloride and Its Relation to That of Adenine Hydrochloride,” Acta Crystallographica 4 (1951): 92–100. 53. Work Book No. 786, pp. 18–19. 54. Meselson to Watson, 14 December 1956, MM. 55. GS/FLH, 5 May 1992, 12–13, 18; FWS/FLH, 18 May 1993, 1:42–43; MM, FWS/FLH, 10 May 1997, 3:12–13. 56. Stent to Holmes, 28 September 1993; Sidney Altman to Holmes, 18 May 1999; FWS/FLH, 18 May 1993, 7, 12, 22. Molecular formula reproduced from Joseph S. Fruton and Sofia Simmonds, General Biochemistry, 2d ed. (New York: John Wiley, 1958), p. 903. 57. MM/FLH, 18 May 1993, 1:22; FWS/FLH, 18 May 1993, 1:19. 58. Work Book No. 786, p. 19. 59. Rose M. Litman and Arthur B. Pardee, “Production of Bacteriophage Mutants by a Disturbance of Deoxyribonucleic Acid Metabolism,” Nature 178 (1956): 529–31; Workbook No. 786, p. 19. 60. Litman and Pardee, “Production of Bacteriophage Mutants,” p. 531. 61. MM, FWS/FLH, 18 May 1993, 1:28–29. 62. Work Book No. 786, p. 19. 63. Both the Zamenhof and Gribhoff papers and the Dunn and Smith papers were presented under the common title “Incorporation of Halogenated Pyrimidines into the Deoxyribonucleic Acids of Bacterium Col. and Its Bacteriophages,” Nature 174 (1954): 305–8. Meselson listed in addition two papers by Zamenhof “in press.” 64. Work Book No. 786, p. 19. 65. MM/FLH, 18 May 1993, 1:29; MM, FWS/FLH, 18 May 1993, 1:16. 66. Workbook No. 786, p. 20. 67. Ibid. 68. Ibid., p. 8; MM/FLH, 19 April 1993, 2:9; 18 May 1993, 1:17–18. 69. Workbook No. 786, p. 8; MM/FLH, 18 May 1933, 1:18. 70. Workbook No. 786, p. 21. 71. Adrien Albert and J. N. Phillips, “Ionization Constants of Heterocyclic Substances. Part II. Hydroxy-Derivatives of Nitrogenous Six-Membered Ring-Compounds,” Journal of the Chemical Society 159(1956): 1297–98. 72. Work Book No. 786, p. 21. 73. H. B. Watson, Modern Theories of Organic Chemistry (Oxford: Clarendon Press, 1941), pp. 31, 87. 74. Work Book No. 786, p. 22. 75. Ibid.

466

N OTES TO

PAGES 142 –151

76. Linus Pauling, The Nature of the Chemical Bond, 3d ed. (Ithaca: Cornell University Press, 190), pp. 185–86. 77. Albert and Phillips, “Ionization Constants,” p. 1299. 78. Work Book No. 786, p. 22. 79. MMmc. 80. MM/FLH, 19 April 1993, 1:14–15; 18 May 1993, 1:18. 81. Albert and Phillips, “Ionization Constants,” p. 1299. 82. G. S. Parry, “The Crystal Structure of Uracil,” Acta Crystallographica 7 (1954): 313. 83. Ibid., pp. 313–320; Workbook No. 786, p. 23. 84. Bruce R. Penford, “The Electron Distribution in Crystalline α-Pyridone,” Acta Crystallographica 6 (1953): 591–600. 85. Penfold, “Electron Distribution,” p. 597. 86. Work Book No. 786, p. 24; MM/FLH, 19 April 1993, 3:11. 87. Noel E. White and C. J. B. Clews, “The Crystal and Molecular Structure of 4, 5-Diamino-2-Chloropyrimidine,” Acta Crystallographica 9 (1956): 586–93; Work Book No. 786, pp. 25–26. 88. Joseph Singer and I. Fankuchen, “The Crystal Structure of 2-Metanilamido-5Br-Pyrimidine,” Acta Crystallographica 5 (1952): 99–103; Work Book No. 786, p. 26. 89. Jerry Donohue, “Hydrogen-Bonded Helical Configurations of Polynucleotides,” PNAS 42 (1956): 60. 90. Ibid., pp. 60–65. 91. Work Book No. 786, p. 27. 92. MM/FLH, 18 May 1993, 1:25–26. 93. Ibid. 94. S. Furberg, “The Crystal Structure of Cytidine,” Acta Crystallographica 3 (1950): 325–33, esp. p. 330. 95. Ibid., p. 331. 96. Work Book No. 786, p. 28. 97. Ibid.; G. R. Wyatt, “Separation of Nucleic Acid Components by Chromatography on Filter Paper,” in The Nucleic Acids, ed. Chargaff and Davidson, 1: 243–55. See also Joseph S. Fruton and Sofia Simmonds, General Biochemistry, 2d ed. (New York: John Wiley, 1958), pp. 115–19. 98. MMmc. 99. H. B. Watson, Modern Theories of Organic Chemistry (Oxford: Clarendon Press, 1941), pp. 101–3; Work Book No. 786, p. 28; MM/FLH, 18 May 1993, 1:27. 100. MM/FLH, 18 May 1993, 1:27. 101. David Shugar and Jack J. Fox, “Spectrophotometric Studies of Nucleic Acid Derivatives and Related Compounds as a Function of pH,” Biochimica et Biophysica Acta 9 (1952): 199–218. 102. Work Book No. 786, p. 28.

N OTES TO PAGES 152 –163 103. 104. 105. 106.

467

Shugar and Fox, “Spectrophotometric Studies,” pt. II, pp. 369–84. Jordan, “Physical Properties,” p. 480. Work Book No. 786, p. 29. Pauling and Corey, “Specific Hydrogen-Bond Formation” pp. 164–

65. 107. Ibid., pp. 165–81. 108. Ibid., p. 179. 109. Work Book No. 786, pp. 6, 11, 30–31, 33. 110. Pauling and Corey, “Specific Hydrogen-Bond Formation,” p. 180. 111. Ibid., pp. 168–178. 112. MM/FLH, 19 April 19, 1: 28. 113. F. H. C. Crick and J. D. Watson, “The Complementary Structure of Deoxyribonucleic Acid,” Proceedings of the Royal Society of London, ser. A, 223 (1954): 90. 114. Pauling and Corey, “Specific Hydrogen-Bond Formation,” p. 165. 115. California Institute of Technology, Annual Report, 1954–55, p. 67. 116. Work Book 786, p. 30.

Chapter Five. Dense Solutions 1. Stahl to Levinthal, 15 September 1956, FWS. 2. FWS/FLH, 4 May 1992, NUI, pp. 1, 3; 12 July 1992, p. 22. 3. Wendell Stanley papers, Bancroft Library, Berkeley, 78/18c, carton 13. 4. Stahl to Levinthal, 14 September 1956. 5. MM/FLH, 19 April 1993, 1:9–10, 15–16; Hubert Bradford Vickery, “Treat Baldwin Johnson: 1875–1947,” Biographical Memoirs of Fellows of the National Academy of Sciences 27 (1952): 83–119. 6. MM/FLH, 19 April 1995, 1:1–6. 7. Ibid., 2:12; Work Book 786, p. 34. 8. FWS/FLH, 21 November 1988, pp. 7, 60; 4 May 1992, NUI, p. 7; Meselson to Watson, 14 December 1956; Stahl to Holmes, 11 March 1998. 9. FWS/FLH, 21 November 1988, p. 62; 4 May 1992, NUI, p. 3; GS/FLH, 5 May 1992, p. 12; MM, FWS/FLH, 18 May 1993, 1:35; Meselson to Watson, 14 December 1956, MM; Stahl to Holmes, 11 March 1998. 10. MM/FLH, 12 July 1992, 5:19. 11. Work Book No. 786, p. 35. 12. Ibid.; Rose Litman and Arthur B. Pardee, “Production of Bacteriophage Mutants by a Disturbance of Deoxyribonucleic Acid Metabolism,” Nature 178 (1956): 529–31; Franklin W. Stahl, Jean M. Crasemann, Larry Okun, Evelyn Fox, and Charles Laird, “Radiation-Sensitivity of Bacteriophage Containing 5-Bromodeoxyuridine,” Virology 13 (1961): 99; Franklin W. Stahl, The Mechanics of Inheritance (Englewood Cliffs, N.J.: Prentice-Hall, 1964), pp. 51– 52.

468

N OTES TO

PAGES 164 –171

13. Work Book No. 786, p. 35. 14. MM/FLH, 19 April 1993, 2:13; MM, FWS/FLH, 18 May 1993, 2:7. 15. FWS/FLH, 21 November 1988, p. 10. 16. Meselson to Watson, 6 December 1955, 14 December 1956; Work Book No. 786, pp. 32, 35; MM/FLH, 19 April 1993, 2:26; Jesse P. Greenstein, Sanford M. Birnbaum, and M. Clyde Otey, “Optical and Enzymatic Characterization of Amino Acids,” Journal of Biological Chemistry 204 (1953): 307–21. See also Joseph S. Fruton and Sofia Simmonds, General Biochemistry, 2d ed. (New York: John Wiley, 1958), pp. 74–76. 17. Work Book No. 786, p. 36; B. Dawson and A. McL. Mathieson, “The Crystal Structures of Some α-amino acids. A Preliminary X-Ray Examination,” Acta Crystallographica 4 (1951): 475–77. 18. Work Book No. 786, p. 38. 19. Ibid., p. 43. 20. H. C. Brown, D. H. McDaniel, and O. Hafliger, “Dissociation Constants,” in Determination of Organic Structures by Physical Methods, ed. E. A. Braude and F. C. Nachod, vol. 1 (New York: Academic Press, 1955), pp. 567– 662. 21. Work Book No. 786, p. 36. 22. Ibid., p. 37. 23. Ibid. 24. D. B. Dunn and J. D. Smith, “Incorporation of Halogenated Pyrimidines into the Deoxyribonucleic Acids of Bacterium coli and Its Bacteriophages,” Nature 174 (1954): 305–6. 25. Ibid., p. 305. 26. Work Book No. 786, p. 39. 27. Stephen Zamenhof and Gertrude Griboff, “E. coli Containing 5-Bromouracil in Its Deoxyribonucleic Acid,” Nature 174 (1954): 306–8. 28. Work Book No. 786, pp. 39–40. 29. Ibid., pp. 40,43; Richard J. Block, Emmet L. Durrum, and Gunter Zweig, A Manual of Paper Chromatography and Paper Electrophoresis (New York: Academic Press, 1955), pp. 206–21. 30. Frank W. Putnam, “Molecular Kinetic and Electrophoretic Properties of Bacteriophages,” Science 111 (1950): 481–88; Frank W. Putnam, “Ultracentrifugation of Bacterial Viruses,” Journal of Polymer Science 12 (1954): 391– 400; Work Book No. 786, pp. 38–42. 31. Putnam, “Molecular and Electrophoretic Properties,” pp. 486–87; Work Book No 786, p. 42. 32. William J. Hagan, “Vinograd, Jerome Ruben,” in Dictionary of Scientific Biography, Supp. 2, ed. F. L. Holmes, vol. 18 (New York: Scribners, 1990), pp. 965–66; “Curriculum Vitae—Jerome Vinograd,” Vinograd Collection, CTA, 25.3; Vinograd to Pauling, 26 July 1956; Vinograd to Koepfil, 19 March 1957, ibid., 4.18.

N OTES TO PAGES 171 –179

469

33. Putnam, “Ultracentrifugation,” p. 391. 34. The Svedberg and Kai O. Pederson, The Ultracentrifuge (Oxford: Clarendon Press, 1940), pp. 3–12. 35. Retrospectively, Meselson has pointed out that .002 percent gives an optical density of 1.0, which is easily detectable. MM, FWS/FLH, 10 May 1997, 3:20. 36. Work Book No 786, p. 42. 37. MM/FLH, 4 June 1996, 1:22–25. 38. Work Book No. 786, p. 42. 39. Dunn and Smith, “Incorporation of Halogenated Pyrimidines,” p. 305. 40. Svedberg and Peterson, Ultracentrifuge, pp. 5–15. 41. Work Book No 786, p. 43. Svedberg called the sedimentation rate (s) the sedimentation constant, because “this quantity is a characteristic constant for a given molecular species in a given solvent at a given temperature.” Svedberg and Peterson, Ultracentrifuge, p. 5. 42. Horace Freeland Judson, The Eighth Day of Creation: Makers of the Revolution in Biology (New York: Simon and Schuster, 1979) pp. 189–90. I have omitted the final sentences of this passage, ending with “we chose cesium chloride.” As told, the choice seems to have been made at once. Meselson has commented, however, that these sentences refer to “later” events (Meselson, MMmc.) Meselson added the sentence in brackets to his statement to Judson. 43. Charles Hodgman, Robert S. Weast, Robert S. Shankland, and Samuel M. Selby, Handbook of Chemistry and Physics (Cleveland: Chemical Rubber Publishing, 1961), pp. 1996–2111. I have used for convenience an edition published several years later. All the numbers written down by Meselson are identical with those found in its tables. 44. FWS/FLH, 21 November 1988, p. 51. 45. Work Book No. 786, p. 44. 46. Ibid. 47. MM/FLH, 20 May 1992, 2:17. 48. FWS/FLH, 4 May 1992, NUI, pp. 3–4. 49. FWS/FLH, 21 November 1988, p. 11. 50. Work Book No. 786, p. 45; International Critical Tables of Numerical Data (London: McGraw Hill, 1928), 3:104–7; MM, FWS/FLH, 10 May 1997, 3:23–24. 51. Work Book No. 786, pp. 56–57; Philip A. Lyons and John F. Riley, “Diffusion Coefficients for Aqueous Solutions of Calcium Chloride and Cesium Chloride,” Journal of the American Chemical Society 76 (1954): 5216– 20; R. A. Robinson and R. H. Stokes, Electrolyte Solutions (London: Butterworths, 1955), p. 480. 52. Work Book No. 786, pp. 58–60; Herbert S. Harned and Orion E. Schupp Jr., “The Activity Coefficients of Cesium Chloride and Hydroxide in

470

N OTES TO

PAGES 179 –186

Aqueous Solution,” Journal of the American Chemical Society 52 (1930): 3886–92. 53. MM/FLH, 4 June 1996, 1:1, 36. 54. Work Book No. 786, p. 60; MM/FLH, 19 April 1993, 1:25; MM, FWS/ FLH, 10 May 1997, 3:27, 29. 55. Litman and Pardee, “Bacteriophage Mutants,” p. 530; [F.W. Stahl], “5Brom-Uracil Prospectus,” 21 October 1953, FWS; MM/FLH, 2 December 1987, p. [20]. 56. FWS/FLH, 4 May 1992, NUI, p. 3. 57. [Stahl], “5-Brom-Uracil Prospectus,” p. 3. 58. E. G. Pickels, W. F. Harrington, and H. K. Schachman, “An Ultracentrifuge Cell for Producing Boundaries Synthetically by a Layering Technique,” PNAS 38 (1952): 943–48. 59. Howard K. Schachman, “Application for Renewal of National Science Foundation Grant G 616,” 1 October 1956, Bancroft Library, University of California, Berkeley, 78/18C, Carton 4. 60. Ibid., pp. 7–8. 61. Ibid., p. 7; MM/FLH, 19 April 1993, p. 21. 62. MM/FLH, 19 April 1993, pp. 21–24; Howard Schachman/FLH, 2 May 1994, p. 6. Meselson’s and Schachman’s memories of the time and place of this meeting are fragmentary. I have reconstructed it by fitting what they do recall into the most logical circumstances. 63. [Stahl], “5-Brom-Uracil Prospectus,” p. 3.

Chapter Six. The Big Machine 1. FWS/FLH, 21 November 1988, p. 47. 2. Ibid., p. 52. 3. [F.W. Stahl], “5-Bromo-Uracil Prospectus,” 21 October 1956, FWS. 4. Ibid., pp. 2, 3. 5. Ibid., p. 2. 6. Levinthal to Stahl, 19 July 1956, FWS. 7. Stahl to Levinthal, 14 September 1956, FWS. 8. MM/FLH, 21 November 1989, p. 19. 9. For a succinct description of the method, see Kensal Edward van Holde, Physical Biochemistry (Englewood Cliffs, N.J.: Prentice-Hall, 1971), p. 115. 10. [Meselson], untitled, 23 October 1956, FWS. 11. This inference is based on the fact that Stahl mentioned only the flotation experiment in his memo, whereas Meselson mentioned both experiments in his memo two days later. See [Meselson], untitled, FWS. 12. MM/FLH, 2 December 1987, p. 18, 4 June 1996, 1:1–5, 35–36 ; MM, FWS/FLH, 10 May 1997, 29; MMmc, 1995, 1997; Stahl to Holmes, 20 August 1996.

N OTES TO PAGES 186 –194

471

13. Beckman/Spinco, Analytical Ultracentrifuge: Model E (Palo Alto: Beckman Instruments, Inc., n.d.), pp. 6, 8, 12. 14. Ibid., pp. 8–9. 15. Ibid., pp. 7–9. 16. Ibid., p. 13; Beckman, Analytical Rotors, Cell, Cell Components and Accessories for Model E Ultracentrifuge (Palo Alto: Beckman Instruments, Inc., 1970), pp. 4–5. 17. Howard K. Schachman, Ultracentrifugation in Biochemistry (New York: Academic Press, 1959), p. 35. 18. Beckman/Spinco, Ultracentrifuge, p. 10. For descriptions of the Schlieren method that explain its optical system, see Schachman, Ultracentrifugation, pp. 34–38; van Holde, Physical Biochemistry, pp. 91–94. 19. Ibid., p. 4. 20. Beckman/Spinco, Ultracentrifuge, p. 10. 21. Ibid.; MM/FLH, 20 May 1992, 1:21. 22. Ibid., 1:20. 23. The Svedberg and Kai O. Pederson, The Ultracentrifuge (Oxford: Clarendon Press, 1940), pp. 80–212. 24. Beckman/Spinco, Ultracentrifuge, p. 2. 25. Vinograd to Pickels, 30 September 1954, Vinograd Papers, CTA, 2.3. 26. MM/FLH, 2 December 1987, p. 18. 27. Gray to Vinograd, 21 March 1956; Mansfield to Vinograd, 16 April 1956; Pickels to Vinograd, 13 June 1956; Vinograd to Wilson, 13 August 1956; Gray to Vinograd, 7 January 1956, Vinograd Papers, CTA, 2.3. 28. Meselson, UL 637. 29. MM/FLH, 20 May 1992, 2:7–8; MMmc. 30. MM/FLH, 20 May 1992, 2:8–9. 31. Beckman/Spinco, Ultracentrifuge, p. 6. 32. UL 637; MM/FLH, 20 May 1992, 2:10. 33. MM/FLH, 2 December 1987, pp. 18–19. There is no contemporary documentary evidence to support Meselson’s recollection of this event, but it is corroborated by Stahl’s independent recollection of the events just before and after the first experiment in the analytical centrifuge (FWS/FLH, 21 November 1988, p. 48): FWS: We started with just a high concentration and already the T4 floated. That implied, if we used more dilute concentrations, we would be able to find one where T4 floated, but substituted T4 would sink. FLH: Is this before you knew that it would establish a density gradient? FWS: Yes. Soon afterward Matt must have done this flotation experiment in a centrifuge where he could watch what was happening. He was watching for the T4 to go up, but instead the first thing he saw was the cesium was reconcentrating itself.

472

N OTES TO

PAGES 194 –203

34. Films corresponding to UL 637; [Meselson], untitled, 23 October 1956, p. 3; Frank W. Putnam, “Ultracentrifugation of Bacterial Viruses,” Journal of Polymer Science 12 (1954): 393–94. ¨ ber das Sedi35. MM/FLH, 2 December 1987, p. 19; Kai O. Pederson, “U mentationsgleichgewicht von anorganischen Salzen in der Ultrazentrifuge,” Zeitschrift fu¨r physikalische Chemie A170: (1934): 41–61, quotation from p. 53; MMmc, December 1995. 36. Workbook No. 786, p. 62. 37. FWS/FLH, 21 November 1988, p. 50. 38. [Meselson], untitled, 23 October 1956, p. 3. 39. Ibid., pp. 1–4. 40. Ibid., p. 1. 41. MM/FLH, 19 April 1993, 1:2; 4 June 1996, 1:15–16; MM, FWS/FLH, 10 May 1997, 3:31–32. 42. UL 639, pp. 2–4; MM/FLH, 12 July 1992, 5:11. Most of the ultraviolet absorption films have survived and are contained in a plywood box constructed for Meselson by the Caltech shop. The films are identified with the centrifuge runs recorded in the log, by means of corresponding number codes written on the cellophane jackets into which the films were placed and etched also onto the film negatives themselves. 43. UL 639, 648; MM/FLH, 12 July 1992, 5:10–11. 44. UL 653. 45. UL 653, pp. 1–5. 46. Film corresponding to UL 653; MM/FLH, 2 December 1987, pp. 19– 20, 12 July 1992, 5:11–12. 47. UL 655. 48. UL 656, pp. 1–2, and corresponding films. 49. UL 657; Beckman/Spinco, Ultracentrifuge, p. 11; MM/FLH, 12 July 1992, 5:12. 50. MM/FLH, 20 May 1992, 2:11. 51. Ibid., p. 6. 52. Vinograd to Committee on Laboratory Maintenance and Facilities, 19 March 1957, Vinograd papers, CTA, 4.18. 53. MM/FLH, 12 July 1992, 5:17. In 1992 Meselson recalled, “The way this machine was used previously it was typically a 30 minute or 1 hour’s run. We evolved a technique which required at least two days, sometimes . . . even longer—so it was unpopular.” 54. Biology 1956 at the California Institute of Technology: A Report for the Year 1955–1956 on the Research and Other Activities of the Division of Biology (privately circulated bound report, n.d.), p. 1. 55. MM/FLH, 12 July 1992, 5: 14, 4 June 1996, 2:5. 56. MM/FLH, 12 July 1992, 5:14. 57. Typed protocol inserted in UL before run 679.

N OTES TO PAGES 203 –213

473

58. UL 679, pp. 1–4; UL 719, p. 1. 59. Meselson, Vinograd, and Stahl to Smith and Schachman, 21 January 1957, p. 1; MM/FLH, 12 July 1992, 5:13. Meselson commented in 1992, “Telling where the bands is, is tricky, because, if you make up the solution with a slightly different density, then it will be in a different place, so you can’t really compare one run with another, unless you put in some sort of standard marker. Is it really DNA? how do we know it is in the same place? . . . So we were involved in some, I must say, not very interesting business for a while.” Meselson’s retrospective skepticism about what he and Stahl had been attempting to do thirty-five years earlier reflects the vantage point they later reached. They were undoubtedly aware, in November 1956, of these sources of uncertainty but had, at that point, little more to go on. 60. Erwin Chargaff, “Isolation and Composition of the Deoxypentose Nucleic Acids and of the Corresponding Nucleoproteins,” in The Nucleic Acids, ed. Erwin Chargaff and J. N. Davidson (New York: Academic Press, 1955), 1: 348–58. 61. MM/FLH, 4 June 1996, 1:12–14. 62. UL 693, pp. 1–3. 63. Meselson to Watson, 14 December 1956, MM. 64. Ibid. 65. Division of Biology, California Institute of Technology, Annual Report to the National Foundation for Infantile Paralysis, January–December 1956, pp. 16–17; Meselson to Watson, 14 December 1956, pp. 2–3; Meselson, Vinograd, and Stahl to Smith and Schachman, 21 January 1957, p. 4; Stahl, “5Brom-Uracil Prospectus,” p. 1. For a contemporary general description of cross-reactivation, see Mark H. Adams, Bacteriophages (New York: Interscience Publishers, 1959), pp. 362–63. 66. Meselson to Watson, 14 December 1956, p. 3. 67. See Seymour Benzer, “Fine Structure of a Genetic Region in Bacteriophage,” PNAS 41 (1955): 345. 68. Meselson to Watson, 14 December 1956, p. 3. 69. Ibid.; MM, FWS/FLH, 10 July 1992, 1:1. 70. D. E. Lea, Actions of Radiations on Living Cells (Cambridge: Cambridge University Press, 1955), pp. 69–99. 71. Division of Biology, Annual Report to the National Foundation for Infantile Paralysis, pp. 16–17. 72. Meselson to Watson, 14 December 1956, p. 3. 73. Ibid., pp. 1–2. 74. J. D. Watson, untitled manuscript, JDW. Watson made several changes in the two sentences. I have quoted them as revised. 75. Meselson to Watson, 14 December 1956, p. 5. 76. Ibid.; MM/FLH, 21 November 1989, p. 11. (Al Garen, personal comment, April 27, 1998).

474

N OTES TO

PAGES 213 –223

77. Geoffrey M. Cooper, Rayla Greenberg Temin, and Bill Sugden, eds., The DNA Provirus: Howard Temin’s Scientific Legacy, (Washington, D.C.: ASM Press, 1995), p. xvii; Drake to Holmes, 16 September 1997. 78. Meselson to Watson, 14 December 1956, p. 5.

Chapter Seven. Working at High Speed 1. Sinsheimer to Delbru¨ck, 17 December 1956, MD; Robert L. Sinsheimer, “First Steps Toward a Genetic Chemistry,” Science 125 (1957): 1123– 28. 2. Ibid., p. 1125. 3. Ibid., p. 1128. 4. UL, protocol inserted before run 719. 5. MM/FLH, 12 July 1992, 5:13. 6. Protocol inserted in UL before run 719; MM/FLH, June 4, 1996, 2: 3–4. 7. UL 719, pp. 1–4. 8. Tracing inserted in UL after run 719. 9. UL 657 and corresponding films. As Meselson has pointed out to me (MM/FLH, 4 June 1996, 2:8), DNA and intact phage bands could not actually have been found together this way, because the density difference between the light and heavy end of the cell could be no larger than 0.1gm/cm, whereas the difference between phage and DNA is about 0.2. At that time, however, Meselson was estimating the density gradient to be larger than it actually was (0.4 gm/cm⫺4 instead of 0.12). My reason for interpreting their identification this way is that in a letter of 21 January 1957, Meselson, Stahl, and Vinograd wrote, “In our attempts to band whole phage in cesium chloride density gradients, we noted, in addition to a diffuse band corresponding to the phage, a very sharp band about where DNA would be expected to band.” Meselson, Smith, and Vinograd to Schachman and Smith, 21 January, 1996, MM. 10. Sketch drawn on back of UL 731, p. 2. 11. UL 732 and corresponding films. 12. UL 732 corresponding films, and tracing “732–6 #2” inserted following. 13. UL 735, pp. 1–3, and corresponding films; MM/FLH, 14 July 1998, 1:9. 14. MM, FWS/FLH, 12 July 1992, 4:5, 4 June 1996, 2:9–12; Tracing “735 #7, slow scan,” following UL 735. 15. UL 736, pp. 1–3, and corresponding films; MM/FLH, 4 June 1996, 2:14. 16. “Liter-Scale Preparation of r240 5BU Stocks, 28 Dec. ‘56,” inserted in UL following run 736. 17. MM, FWS/FLH, 12 July 1992, 4:1. 18. “Purification of 5BU Lysate ‘B’ (low mult.) of 30 Dec.,” inserted in

N OTES TO PAGES 223 –234

475

UL after run 736; MM/FLH, 4 June 1996, 2:17–18; 14 July 1998, 1:12. Unless the protocol misstated what he had done, Meselson here did not follow the usual procedure, which was to dump the phage into distilled water. 19. “Purification of 5BU Lysate”; MM, FWS/FLH, 12 July 1992, 4:2; MM/ FLH, 4 June 1996, 2:18–21. 20. UL 739, pp. 1–3 and corresponding films; “Purification of 5BU Lysate.” 21. MM/FLH, 4 June 1996, 2:21–23. 22. UL 740, pp. 1–2 and corresponding films; “Purification of 5BU Lysate”; MM/FLH, 4 June 1996, 2:26. 23. UL 741 and corresponding films, protocol for run 741 inserted following. 24. MM/FLH, 12 July 1992, 5:13, 4 June 1996, 2:25. 25. UL 742; MM/FLH, 12 July 1992, 5:14. 26. UL 743, corresponding films, and protocol inserted preceding. 27. UL 744 and corresponding films. 28. UL 749, 750, and films corresponding to 750. 29. UL 751 and corresponding films. 30. UL 752, pp. 1–2, and corresponding films. 31. UL 753 and corresponding films; MM/FLH, 4 June 1996, 3:10–11. 32. UL 755, pp. 1–2; UL 756, pp. 1–3, corresponding films, and protocol inserted preceding run 756. 33. Meselson, Vinograd, and Stahl to Smith and Schachman, 21 January 1957, MM. 34. Meselson to Watson, 24 January 1957, MM, pp. 3–4. 35. [Meselson], untitled manuscript, 23 October 1956, MM, p. 1. 36. Meselson to Watson, 24 January 1957, MM, p. 4. 37. MM/FLH, 4 June 1996, 3:2. 38. The Svedberg and Kai O. Pederson, The Ultracentrifuge (Oxford: Clarendon Press, 1940), pp. 5–9. 39. Meselson to Watson, 24 January 1957, MM, p. 1; MM/FLH, 14 July 1998, 1:14. 40. Matthew Meselson, “Equilibrium Sedimentation of Macromolecules in Density Gradients with Application to the Study of Deoxyribonucleic Acid,” pt. 1 of Ph.D. thesis, California Institute of Technology, Pasadena, 1957, pp. 6, 26–27; MM/FLH, 14 July 1998, 1:15. 41. MM, FWS/FLH, 10 May 1997, 4:7–9. 42. Meselson to Watson, 24 January 1957, pp. 1–3, MM; MM/FLH, 14 July 1998, 1:15. 43. Joseph S. Fruton and Sofia Simmonds, General Biochemistry, 2d ed. (New York: John Wiley, 1958), p. 72. 44. Meselson, Vinograd, and Stahl to Schachman and Smith, 21 January 1957, p. 2. 45. MM/FLH, 19 April 1993, 2:7.

476

N OTES TO

PAGES 234 –238

46. Densitometer tracing inserted following run 756, UL; MM/FLH, 4 June 1996, 3:4. 47. MM/FLH, 12 July 1992, 5:15; Cyrus Levinthal, “The Mechanism of DNA Replication and Genetic Recombination in Phage,” PNAS 42 (1956): 401; Stahl to Holmes, 20 August 1996. Levinthal had reported the molecular weight of the small pieces in 1956 as seven to eight million. The estimate of twelve million was published in 1959 by Charles Thomas. See Charles A. Thomas, “The Release and Stability of the Large Subunit of DNA from T2 and T4 Bacteriophage,” Journal of General Physiology 42 (1959): 503–23. It is most likely that Stahl and Meselson learned of Thomas’s figure through correspondence well before its publication. 48. Meselson to Watson, 24 January 1957, p. 3. 49. Division of Biology, California Institute of Technology, Annual Report to the Foundation for Infantile Paralysis, January–December 1956, p. 17. 50. MM/FLH, 12 July 1992, 5:17–18, 19 April 1993, 2:7. 51. Meselson, Vinograd, and Stahl to Smith and Schachman, 21 January 1957, p. 1. 52. MM/FLH, 2 December 1987, pp. 3, 5. 53. Meselson, Vinograd, and Stahl to Smith and Schachman, 21 January 1957,p. 4. 54. FWS/FLH, 4 May 1992, NUI, p. 4; MM, FWS/FLH, July 10, 1992, 1: 6. In a curriculum vitae prepared sometime after 1965, Vinograd listed among his “principal contributions” that “Together with his collaborators he has invented two significant methods for ultracentrifugation studies.” One of these was the cesium chloride density gradient method. “Curriculum Vitae-Jerome Vinograd,” Vinograd Collection, CTA, Folder 25.3. 55. Meselson to Watson, 24 January 1957, p. 1. 56. Ibid., p. 2. 57. D. O. Jordan, “The Physical Properties of Nucleic Acids,” in The Nucleic Acids, ed. Erwin Chargaff and J. N. Davidson (New York: Academic Press, 1954), 1: 470–74. 58. Meselson to Watson, 24 January 1957, p. 3. 59. Ibid. 60. G. R. Wyatt and S. S. Cohen, “The Bases of the Nucleic Acids of Some Bacterial and Animal Viruses: The Occurrence of 5-Hydroxymethylcytosine,” The Biochemical Journal 55 (1954): 774-82; Robert L. Sinsheimer, “Nucleotides from T2r⫹ Bacteriophage,” Science 120 (1954): 551–53; G. Streisinger and J. Weigle, “Properties of Bacteriophage T2 and T4 with Unusual Inheritance,” PNAS 42 (1956): 504–10; Delbru¨ck to Sinsheimer, 13 January 1956, MD, 20.3. 61. Joseph X. Khym and Leonard P. Zill, “The Separation of Sugars by Ion Exchange,” Journal of the American Chemical Society 74 (1952): 2090– 94; Meselson, Work Book No. 786, p. 62; MM/FLH, 12 July 1992,5:19.

N OTES TO PAGES 239 –249

477

62. Stahl to Doermann, 29 January 1957, FWS. 63. Meselson to Watson, 24 January 1957, pp. 3–4. 64. Meselson, Vinograd, and Stahl to Smith and Schachman, 21 January 1957, pp. 2–3. 65. Stahl to Doermann, 29 January 1957, p. 1. Stahl was incorrect in attributing his difficulty to photoreactivation, because his later studies failed to detect such an effect. He now explains the trouble as probably due to “poorly controlled exposure to fluorescent light,” causing photoinactivation. Stahl to Holmes, 9 February 1996. 66. Mark H. Adams, Bacteriophages (New York: Interscience Publishers, 1959), p. 77. 67. Meselson, Vinograd, and Stahl to Schachman and Smith, 21 January 1957; MM/FLH, 4 June 1996, 3:8. 68. Meselson, Vinograd, and Stahl to Smith and Schachman, 21 January 1957, p. 3. 69. Stahl to Holmes, 12 December 1995, 20 August 1996; MM/FLH, 4 June 1996, 3:8–10. 70. UL 759, 760, pp. 1–2; MM/FLH, 12 July 1992, 5:15, 4 June 1996, 3: 10–11. 71. UL 777. 72. MM/FLH, 4 June 1996, 3:12; Meselson, “Equilibrium Sedimentation,” p. 26. 73. Meselson to Watson, 24 January 1957, p. 3. 74. MM/FLH, 12 July 1992, 5:16. 75. UL 793 and corresponding film. 76. William Feller, An Introduction to Probability Theory and Its Applications, 3rd ed. (New York: John Wiley, 1968), p. 179. 77. Meselson, “Equilibrium Sedimentation,” pp. 5–9; FWS, MM/FLH, 10 July 1993, 1: 6; MM/FLH, 4 June 1996, 4:1. 78. MM/FLH, 2 December 1987, p. 27, 4 June 1996, 4: 2; Meselson, “Equilibrium Sedimentation,” pp. 30–31. 79. Ibid.; MM, FWS/FLH, 10 May 1997. 80. Stahl to Edgar, 22 February 1957, FWS. 81. Ibid. 82. Stahl to Edgar, 22 February 1957; UL 812 and corresponding film. 83. Stahl to Edgar, 22 February 1957. 84. [Matt Meselson], “Previous and Proposed Research,” n.d., MM. One copy of this manuscript has the retrospective notation “about Feb. ‘57” in Meselson’s hand. I have dated it late March, because of a reference to an experiment in progress that was, according to UL, performed on 23 March (on Doty salmon sperm DNA). There is no inconsistency between Meselson’s reference to duplication “a few times” and Stahl’s to a “1 cycle” stock, because one cycle of viral infection includes several rounds of DNA replication.

478

N OTES TO

PAGES 249 –259

85. Robert L. Sinsheimer, “The Glucose Content of the Deoxyribonucleic Acids of Certain Bacteriophages,” PNAS 42 (1956): 502–4; UL 819, 835 and corresponding films. 86. UL 809, 821, 827, 835, 838, 857, 858, and 859; Jerome Vinograd, “Applications of Stable Density Gradients in Virus and High Polymer Chemistry,” Vinograd collection, CTA, p. 3; MM, FWS/FLH, 10 May 1997, 4:13. 87. Stahl to Edgar, 22 February 1957; Meselson, “Equilibrium Sedimentation,” p. 31. 88. UL 879, pp. 1–6. 89. Meselson, “Equilibrium Sedimentation,” pp. 3–4, 31; Stahl to Edgar, 6 March 1957, FWS;MM, FWS/FLH, 10 May 1997, 4:15. 90. Stahl to Doermann, 29 January 1957, FWS. 91. Stahl to Edgar, 22 February 1957. 92. Stahl to Edgar, 6 March 1957. 93. Meselson to Levinthal, 13 March 1957, MM, p. 2. 94. Meselson to Watson, 24 January 1957, p. 2. 95. Stahl to Edgar, 6 March 1957. 96. Meselson to Levinthal, 13 March 1957. 97. UL 886 and corresponding film. 98. UL 888, 890. 99. Beadle to Meader, 12 March 1957, Division of Chemistry 1.6, CTA. 100. Meselson to Levinthal, 13 March 1957. 101. MM, FWS/FLH, 18 May 1993, 1:36. 102. UL 909, pp. 1–3 and corresponding films; Meselson, “Equilibrium Sedimentation,” p. 37. 103. UL 922 and corresponding film; MM/FLH, 12 July 1992, 5:19. 104. Meselson to Levinthal, 1 April 1957, MM, p. 7; Mark H. Adams, Bacteriophages (New York: Interscience, 1959), pp. v–viii. 105. Stahl to Edgar, 18 March 1957, FWS. 106. In 1988 Stahl commented, “I don’t know why we opted for the idea that the densities would be uniform, and therefore it was a good technique for measuring molecular weight. Perhaps because it was useful to assume that.” FWS/FLH, 21 November 1988, p. 68. 107. MM/FLH, 2 December 1987, 14. 108. Meselson to Levinthal, 1 April 1957, p. 1. 109. Ibid., pp. 2–6; Girair M. Nazarian, “Theory of the Transient State in the Ultracentrifuge,” Journal of Physical Chemistry 62 (1958): 1607–8; Max Mason and Warren Weaver, “The Settling of Small Particles in a Fluid,” Physical Reviews 23 (1924): 412–26; MM/FLH, 2 December 1987, p. 14. 110. Meselson to Levinthal, 1 April 1957, p. 1. 111. Ibid., pp. 1–7. 112. Stahl to Edgar, 7 April 1957, FWS. 113. FWS/FLH, 21 November 1988, p. 70.

N OTES TO PAGES 259 –268

479

114. Meselson, “Sedimentation Equilibrium,” pp. 31–31, 36–37. 115. Stahl to Edgar, 7 April 1957; Meselson to Doty, 27 June 1957, MM. 116. Meselson to Doty, 27 June 1957; Meselson, “Sedimentation Equilibrium,” p. 37. 117. Meselson to Levinthal, 1 April 1957, p. 5. 118. [Meselson], “Previous and Proposed Research,” p. 1. 119. Matthew Meselson, “The Crystal Structure of N,N′-Dimethylmalonamide,” pt. 2 of Ph.D. thesis, California Institute of Technology, Pasadena, 1957, p. 64. 120. MM/FLH, 2 December 1987, p. 13. 121. Svedberg and Pedersen, Ultracentrifuge, p. 8. 122. Howard K. Schachman, Ultracentrifugation in Biochemistry (New York: Academic Press, 1959), p. 201. 123. Matthew Meselson, Franklin W. Stahl, and Jerome Vinograd, “Equilibrium Sedimentation of Macromolecules in Density Gradients,” PNAS 43 (1957): 584. 124. Schachman, Ultracentrifugation, p. 202; Richard J. Goldberg, “Sedimentation in the Ultracentrifuge,” Journal of Physical Chemistry 57 (1953): 196. 125. MM/FLH, 4 June 1996, 4:5–6. 126. Meselson, “Sedimentation Equilibrium,” pp. 20–26; Meselson, Stahl, and Vinograd, “Equilibrium Sedimentation,” pp. 584–587; MM-FLH, June 4, 1996, 4:7. 127. UL 996, pp. 1–2 and corresponding films. 128. “Ph.D. Examination—Matthew Meselson,” 23 May 1957, mimeographed pages bound with Meselson’s dissertation. 129. Ibid.; MM/FLH, 2 December 1987, p. 14; MM, FWS/FLH, 10 May 1997, 5:17; Matthew Meselson and Girair M. Nazarian, “The Transient State in Density-Gradient Centrifugation,” Conference on the Ultracentrifuge (New York: Academic Press, 1963), pp. 134–35. 130. Meselson to Pauling, 30 May 1980, MM. 131. Pauling to Featherstone, 19 October 1961, Ava Helen and Linus Pauling Papers, Oregon State University Library. 132. MM/FLH, 14 July 1998, 1:17; Meselson, Stahl, and Vinograd, “Equilibrium Sedimentation,” 581–588. 133. UL 937, 940, 941, 942, 958, 960. 134. Meselson, “Equilibrium Sedimentation,” p. 39. 135. R. A. Pasternak, Girair M. Nazarian, and Jerome R. Vinograd, “A Fast Method for Reaching Equilibrium in the Ultracentrifuge,” Nature 179 (1957): 92–94. 136. Vinograd, “Applications,” pp. 1–3, plus five figures. 137. MM/FLH, 20 May 1992, 1:15–20, 2:13–14. 138. Ibid., 1:18–20.

480

N OTES TO

PAGES 268 –280

139. FWS/FLH, NUI, 4 May 1992, p. 4. 140. Vinograd, Biographical Information Form, September 1969, Vinograd collection, CTA, 25.2. 141. MM/FLH, 21 November 1989, p. 15. 142. Antoine Lavoisier, Traite´ e´le´mentaire de chimie (Paris: Cuchet, 1789), 1:xxviii–xxix. 143. See Robert K. Merton, “Priorities in Scientific Discovery,” in The Sociology of Science: Theoretical and Empirical Investigations, ed. Norman W. Storer (Chicago: University of Chicago Press, 1973), pp. 286–324. 144. [Meselson], “Previous and Proposed Research,” p. 5. 145. Ibid. 146. Ibid., p. 6. 147. Meselson to Levinthal, 13 March 1957, MM. 148. Stahl to Edgar, 7 April 1957, FWS.

Chapter Eight. The Unseen Band 1. Alma Howard and S. R. Pelc, “Nuclear Incorporation of P32 as Demonstrated by Autoradiographs,” Experimental Cell Research 2 (1951): 178–87. 2. J. Herbert Taylor, Philip S. Woods, and Walter L. Hughes, “The Organization and Duplication of Chromosomes as Revealed by Autoradiographic Studies Using Tritium-Labeled Thymidine,” PNAS 43 (1957): 122; Taylor to Delbru¨ck, 17 February 1957, MD, 21.25; J. Herbert Taylor, “A Brief History of the Discovery of Sister Chromatid Exchanges,” in Sister Chromatid Exchanges, symposium held at Brookhaven National Laboratory, Basic Life Sciences 29 (1984): 1–10; Taylor to Holmes, 21 March 1997. 3. Taylor, Woods, and Hughes,“Organization,” pp. 122–23. 4. Ibid., pp. 123–24; Taylor, “Brief History,” p. 2. 5. Taylor, Woods, and Hughes, “Organization,” p. 125. 6. Taylor, “Brief History,” p. 3. 7. Taylor, Woods, and Hughes, “Organization,” p. 125. 8. Ibid., pp. 125–126. 9. Taylor, “Brief History,” pp. 3–4; Taylor to Delbru¨ck, 4 February 1957, MD, 21.25. 10. M. Delbru¨ck, “A Memo on the Sister-Strand Exchanges Observed by Taylor et al.,” MD, 21.25, p. 1. 11. Taylor, Woods, and Hughes, “Organization and Duplication of Chromosomes,” pp. 123–24. 12. Delbru¨ck, “Sister-Strand Exchanges,” p. 2. 13. Taylor to Delbru¨ck, 4 February 1957, MD, 21.25. 14. Delbru¨ck to Taylor, 11 February 1957, MD, 21.25. 15. Delbru¨ck to Taylor, 18 February 1957, MD, 21.25. 16. Ibid.

N OTES TO PAGES 280 –288

481

17. Stent to Delbru¨ck, 25 February, 1957, MD, 20.20. 18. Taylor to Delbru¨ck, 4 April 1957, 10 April 1957; Delbru¨ck to Taylor, 10 April 1957, 12 April 1957, MD, 21.25. 19. Taylor to Delbru¨ck, 17 February 1957, 5 April 1957, 15 April 1957, 10 May 1957, 22 May 1957; Delbru¨ck to Taylor, 10 April 1957, 12 April 1957, 15 May 1957, MD, 21.25. 20. J. Herbert Taylor, “Sister Chromatid Exchanges in Tritium-Labeled Chromosomes,” Genetics 43 (1958): 515–29, quotation on p. 528. 21. Ibid., p. 527. 22. Meselson to Levinthal, 8 November 1957, MM, p. 2. 23. MM/FLH, 20 May 1992, 1:3–5; MM, FWS/FLH, 11 July 1992, 2:10. 24. Workbook 786, p. 75. 25. MM/FLH, 20 May 1992, 2:20; 14 July 1998, 2:12. 26. MM/FLH, 4 June 1996, 4:11–12. 27. Meselson to Watson, 24 January 1957, MM, p. 5. 28. Delbru¨ck to Sinsheimer, 28 March 1957, MD, 20.3; MM/FLH, 12 July 1992, 6:9. 29. Sinsheimer to Delbru¨ck, 10 April 1957, MD, 20.3. 30. MM/FLH, 4 June 1996, 2:2. 31. MM/FLH, 12 July 1992, 6:1. 32. UL, [blank], 264–67, B-8, B-9; MM/FLH, 4 June 1996, 4:22; MM, FWS/ FLH, 10 May 1997, 4:24–26; MMmc, FWS mc. 33. Meselson to Doty, 27 June 1957, MM. 34. R. Thomas, “Recherches sur la de´naturation des acides desoxyribonucle´iques,” Biochimica et Biophysica Acta 14 (1954): 231–40. 35. Paul Doty and Stuart A. Rice, “The denaturation of desoxypentose nucleic acid,” Biochimica et Biophysica Acta 16 (1955): 446–48. 36. Stuart Rice and Paul Doty, “The Thermal Denaturation of Desoxypentose Nucleic Acid,” Journal of the American Chemical Society 79 (1957): 3937–47. 37. MM/FLH, 14 July 1998, 2:7–8. 38. MM/FLH, 4 June 1996, 5:1–2. 39. UL, B-10, B-11, B-12; Rice and Doty, “Thermal Denaturation,” pp. 3941, 3943. 40. UL, B-13, B-14, 1125, B-15; MM/FLH, 4 June 1996, 22–23. 41. MM/FLH, 14 July 1998, 2:8; Joseph S. Fruton and Sophia Simmonds, General Biochemistry, 2d ed. (New York: John Wiley, 1958), pp. 898, 1001. 42. Division of Biology, California Institute of Technology, Annual Report to the National Foundation for Infantile Paralysis, January–December 1957, pp. 7–8. This report does not specify the date in 1957 at which the work was done, but logically it must have preceded the sequence of density-gradient centrifuge runs begun on 8 July. 43. Ibid., p. 8.

482

N OTES TO

PAGES 288 –299

44. UL, 1135, B-16. 45. UL, B-17. 46. UL, B-18, C-1. 47. UL, B-19. 48. UL, B-20. 49. UL, C-2, C-3, C-4. 50. UL, C-5. 51. Stahl to Levinthal, 27 July 1957, FWS. 52. MM/FLH, 12 July 1992, 6:2. 53. UL, C-6; MM/FLH, 4 June 1996, 4:26–27; 14 July 1998, 2:9. The reasoning attributed to Meselson here actually occurred in 1996 while he reexamimed these films in conversation with me. That he may have reasoned in the same manner in July 1957 is made plausible by the relation between the two experiments in question. 54. UL, C-7. 55. UL, C-8; MM/FLH, 12 July, 1992, 6:1–2. 56. UL, C-9; MMmc. 57. UL, 1197, C-15. 58. MM/FLH, 12 July 1992, 6:5. 59. UL, C-15, C-17; MM/FLH, 12 July 1992, 6:5. 60. UL, 1212, C-19, C-20, C-24 to C-32; MM/FLH, 12 July 1992, 6:7. 61. Delbru¨ck to Bresch, 28 August 1957, MD, 4.20. 62. Max Delbru¨ck and Gunther S. Stent, “On the Mechanism of DNA Replication,” A Symposium on the Chemical Basis of Heredity, ed. William D. McElroy and Bentley Glass (Baltimore: Johns Hopkins University Press, 1957), p. 699. 63. MM/FLH, 2 December 1987, pp. 4–5. 64. Stent to Holmes, 11 March 1992. 65. FWS/FLH, 21 November 1988, pp. 18, 68. 66. Delbru¨ck to Bresch, 28 August 1957. 67. Sinsheimer to Delbru¨ck, 26 March 1957, 10 April 1957 ; Delbru¨ck to Sinsheimer, 28 March 1957, MD, 20.3. 68. Stent to Weigle, 5 September 1957, Weigle papers, CTA, Box 2. 69. Freese to Delbru¨ck, 5 May 1958; Delbru¨ck to Freese, 8 May 1958, MD, 8.5. 70. Meselson to Doty, 27 June 1957, MM; MM/FLH, 12 July 1992, 6:7–8; Cecil E. Hall and Michael Litt, “Morphological Features of DNA Macromolecules as Seen with the Electron Microscope,” Journal of Biophysical and Biochemical Cytology 4 (1958): 1–4. 71. Edward H. Simon, “Transfer of DNA from Parent to Progeny in a Tissue Culture Line of Human Carcinoma of the Cervix (Strain HeLa),” Journal of Molecular Biology 3 (1961): 101–9; UL, C-20 to C-23; MM/FLH, 12 July 1992, 6:6–7.

N OTES TO PAGES 299 –307

483

72. John Cairns, “The Autoradiography of DNA,” in Phage and the Origins of Molecular Biology, ed. John Cairns, Gunther S. Stent, and James D. Watson (Cold Spring Harbor: Cold Spring Harbor Laboratory of Quantitative Biology, 1966), pp. 252–53; J. Cairns/FLH, 13 July 1993, 1:2–4. 73. MM/FLH, 4 June 1996, 5:2. 74. UL, C-24 to C-28. The phrase “T4 DNA heated a la Doty” is written on the envelope containing the films for C-28. 75. MMmc; UL, B-11, B-14, C-6, C-9, C-28; Matthew Meselson and Franklin W. Stahl, “The Replication of DNA in Escherichia Coli, PNAS 44 (1958): 679. 76. UL, C-29 to C-31. 77. UL, C-33, C-36. 78. Meselson to Levinthal, 13 March 1957, MM. 79. [Meselson], “Previous and Proposed Research,” p. 6. 80. UL, C-37; MM/FLH, 12 July 1992, 6:8. 81. UL, C-38. 82. MM/FLH, 12 July 1992, 6:7. 83. UL, 1252. The films from this run have not survived.

Chapter Nine. One Discovery, Three Stories 1. MM/FLH, 18 May 1993, 2:16. 2. FWS/FLH, 21 November 1988, p. 18. 3. [Matthew Meselson], “Previous and Proposed Research,” p.[5], MM. 4. UL, C-40; MM/FLH, 4 June 1996, 1:13. 5. Armin D. Kaiser, “A Genetic Study of the Temperate Coliphage,” Virology 1 (1955): 424– 43; M. L. Morse, Esther Lederberg, and Joshua Lederberg, “Transduction in Escherichia Coli K-12,” Genetics 41 (1956): 142–56; J. Weigle, “Transduction by Coliphage of the Galactose Marker,” Virology 4 (1957): 14–25. 6. UL, C-42, C-43; MM/FLH, 4 June 1996, 5:14, 14 July 1998, 2:12; MM, FWS/FLH, 10 May 1998, 5:1. 7. UL, C-40 to C-44. 8. Delbru¨ck to Doermann, undated, judged from answer of Doermann on 21 October 1957 to be 14 October. 9. UL, C-46. 10. MM/FLH, 5 February 1988, p. 5, 15 March 1995, 1:14; FWS/FLH, 21 November 1988, pp. 21, 87. 11. FWS/FLH, 21 November 1988, pp. 36, 89–90. 12. Doermann to Delbru¨ck, 8 October 1957; Delbru¨ck to Doermann, [14 October 1957], MD, 6.17. 13. MM/FLH, 4 June 1996, 5:6. 14. FWS/FLH, 21 November 1988, pp. 36–37. 15. MM/FLH, 2 December 1987, p. 40; 10 July 1991, p. 2.

484

N OTES TO

PAGES 307 –319

16. Matthew Meselson, “Equilibrium Sedimentation of Macromolecules in Density Gradients with Application to the Study of Deoxyribonucleic Acid,” Ph.D. diss., California Institute of Technology, 1957, “Acknowledgements.” 17. Matthew Meselson and Franklin W. Stahl, “The Replication of DNA in Escherichia Coli,” PNAS 44 (1958): 673–74; MM/FLH, 14 July 1998, 2:14. 18. MM/FLH, 2 December 1987, p. 24. 19. Meselson, memorandum to FLH, 2 March 1995. 20. UL, C-47 to C-48. 21. MM/FLH, 2 December 1987, p. 29. 22. MM/FLH, 15 March 1995, 2:14. 23. UL, C-47 to C-48. 24. MM/FLH, 15 March 1995, 1:17. 25. UL, C-50. 26. UL, C-51. 27. Ibid.; MM/FLH, 15 March 1995, 3:4. 28. UL, C-52. 29. UL, C-53, 1306. 30. MM/FLH, 15 March 1995, 3:10. 31. MM/FLH, 2 December 1987, pp. 32, 35, 47; 15 March 1995, 3:10; MMmc. 32. MM, FWS/FLH, 12 July 1992, 4:8–9. 33. Frederic L. Holmes, “Historians and Contemporary Scientific Biography,” paper presented at a conference titled “Life and Work of Linus Pauling (1901–1994): A Discourse on the Art of Biography,” Corvallis Oregon, 2 March 1995; Meselson, memorandum to FLH, 2 March 1995. 34. MM/FLH, 15 March 1995, 3:6. 35. Ibid., 2:18, 3: 5, 12. 36. John L. Heilbron, “Remarks on the Writing of Biography,” in The Pauling Symposium: A Discourse on the Art of Biography, ed. Ramesh S. Krishnamurthy (Corvallis: Oregon State Libraries, 1996), pp. 231–33. 37. R. Edward Geiselman, “The Cognitive Interview,” paper presented at conference titled “Interviews in Writing the History of Science,” Stanford University, 29 April 1994. 38. MM/FLH, 2 December 1987, p. 35. Ten years later, Meselson recalled the additional detail that he met Denise Cohen, the wife of a scientist at Caltech, and made some excited remarks to her on his way out of the dark room. MM, FWS/FLH, 10 May 1997, 4:5–6.. 39. MM/FLH, 2 December 1987, pp. 32, 47.

Chapter Ten. An Extremely Beautiful Experiment 1. FWS/FLH, 4 May 1992, pp. 4–5; FWS, MM/FLH, 12 July 1992, 2:2–3. 2. FWS/FLH, 21 November 1988, pp. 21–22.

N OTES TO PAGES 320 –327

485

3. Matthew Meselson and Franklin W. Stahl, “The Replication of DNA in Escherichia Coli,” PNAS 44 (1958): 673–74; UL, C-54; FWS, MM/FLH, 11 July 1992, 2:3; 10 May 1997, 5:3– 4. 4. UL, C-53 to C-59; untitled photograph filed with copy of Meselson’s Ph.D. thesis. 5. UL, C-54 to C-56. 6. UL, C-56 to C-59; FWS/FLH, NUI, 4 May 1992, p. 4. 7. UL, C-60, C-61; MM/FLH, 4 June 1996, 5:9–10; MM, FWS/FLH, 10 May 1997, 5:5–6. 8. John Cairns, “The Autoradiography of DNA,” in Phage and the Origins of Molecular Biology, ed. John Cairns, Gunther S. Stent, and James D. Watson (Cold Spring Harbor: Cold Spring Harbor Laboratory of Quantitative Biology, 1966), pp. 252–53; J. Cairns/FLH, 13 July 1993, 1:14–15; MM/FLH, 4 June 1996, 5:11. 9. Cairns, “Autoradiography,” p. 253; J. Cairns/FLH, 13 July 1993, 1:7. 10. GS/FLH, 5 May 1992, pp. 25–26. 11. Ibid., p. 26. 12. GS/FLH, 7 May 1992, pp. 1–4. 13. James Watson, The Double Helix: A Personal Account of the Discovery of the Structure of DNA (New York: Atheneum, 1968), pp. 190–92. 14. Gunther S. Stent, “Mating in the Reproduction of Bacterial Viruses,” in Advances in Virus Research, ed. Kenneth M. Smith and Max A. Lauffer (New York: Academic Press, 1958) 5:95–149, esp. pp. 135–45; Stent to Delbru¨ck, 30 October 1957, MD, 20.3. 15. MM/FLH, 21 November 1989, pp. 22, 25. 16. UL, C-62. 17. Stent to Delbru¨ck, 7 November 1957, MD, 20.3; J. Cairns to Holmes, 18 June 1996; J Cairns/FLH, 13 July 1993, 1:1. 18. GS/FLH, 5 May 1992, p. 26, 7 May 1992, p. 4. 19. Delbru¨ck to Harm, 5 November 1957, MD, 10.5; Delbru¨ck to Bresch, 18 November 1957, MD, 4.20; MM/FLH, 21 November 1989, pp. 26–27. Neither Sinsheimer nor Meselson remembers such conversations occurring as frequently as Delbru¨ck’s letter suggests. According to Sinsheimer, “the discussions were sporadic, coming with each new experimental result.” Sinsheimer to F.L. Holmes, 9 December 1992. Meselson recalls only one conversation during which Delbru¨ck and Sinsheimer were both present. FWS, MM/FLH, 11 July 1992, 2:8–9. 20. GS/FLH, 6 May 1992, p. 51. 21. Meselson to Watson, 8 November 1957, MM, JDW. 22. FWS, MM/FLH, 10 July 1992, 1:23–24; 10 May 1997, 5:8. 23. Meselson to Watson, 8 November 1957, MM, JDW. 24. FWS, MM/FLH, 11 July 1992, 2:19–20; MM/FLH, 4 June 1996, 5:13. 25. Inserted in the binder with Meselson’s Ph.D. dissertation. 26. Meselson to Watson, 8 November 1957; MM/FLH, 2 December 1987, p. 31.

486

N OTES TO

PAGES 328 –341

27. Ibid. 28. Ibid. 29. MM/FLH, 2 December 1987, p. 38. 30. FWS/FLH, 4 May 1992, NUI, p. 7; MM/FLH, 4 June 1996, 5:13. 31. FWS, MM/FLH, 12 July 1992, 2:21. 32. Meselson to Watson, 8 November 1957. I have corrected here several trivial typographical errors in the original. 33. MM/FLH, 4 June 1996, 4:12. 34. Meselson to Watson, 8 November 1957, MM, JDW. 35. Meselson to Levinthal, 8 November 1957, MM. 36. Meselson to Watson, Meselson to Levinthal, 8 November 1957, MM. 37. MM/FLH, 4 June 1996, 5:15–16. 38. UL, C-63; FWS, MM/FLH, 11 July 1992, 2:7. 39. Meselson to Paul and Helga Doty, 10 November 1957, MM. 40. Ibid.; UL, C-29, C-30, C-39. 41. FWS, MM/FLH, 11 July 1992, 2:25. 42. Pauling to Featherstone, 19 October 1961; photograph, Pauling Papers. 43. MM, FWS/FLH, 11 July 1992, 2:1; Richard P. Feynman, ‘‘Surely You’re Joking, Mr. Feynman!” Adventures of a Curious Character (New York: Norton, 1985), pp. 234–35. The garbled nature of the account given here raises some doubt about how fully Feynman had understood what Meselson showed him. 44. Levinthal to Meselson, 18 November 1957, MM. 45. “Discussion,” in A Symposium on the Chemical Basis of Heredity, ed. William D. McElroy and Bentley Glass (Baltimore: Johns Hopkins University Press, 1957), p. 129; M. Delbru¨ck, “Atomic Physics in 1910 and Molecular Biology in 1957,” lecture, MD, 35.2, p. 18. 46. Levinthal to Meselson, 18 November 1957. 47. FWS, MM/FLH, 11 July 1992, 2:29–30. 48. Levinthal to Meselson, 18 November 1957. 49. Ibid. 50. Levinthal was later proven to be partially correct. The DNA molecules were not broken during the centrifuge run, as he thought, but as they were loaded into the centrifuge cell. See Chapter 13. 51. Delbru¨ck, “Atomic Physics in 1910,” p. 1. 52. Ernst Peter Fischer and Carol Lipson, Thinking About Science: Max Delbru¨ck and the Origins of Molecular Biology (New York: Norton, 1988), pp. 41–45, 56–57. 53. Delbru¨ck to Bohr, 1 December 1954, MD, 3.29. 54. Delbru¨ck to Harm, 5 November 1957, MD, 10.5. 55. Delbru¨ck, “Atomic Physics in 1910,” pp. 7–9. 56. Ibid., pp. 9–11. 57. Ibid., p. 12.

N OTES TO PAGES 341 –353

487

58. Ibid., pp. 13–14. 59. FWS, MM/FLH, 11 July 1992, 2:33. 60. Delbru¨ck, “Atomic Physics in 1910,” pp. 14–15. 61. Ibid., pp. 15–19. 62. MM/FLH, 21 November 1989, pp. 28–29; 4 June 1996, 5:18–19. 63. Delbru¨ck “Atomic Physics in 1910,” p. 22. 64. FWS, MM/FLH, 11 July 1992, 2:35. 65. Stent to Delbru¨ck, 9 December 1957, MD, 20.20; Fischer and Lipson, Thinking About Science, p. 222. 66. F. H. C. Crick, “Nucleic Acids,” Scientific American 197 (1957): 188– 200. 67. Robert L. Sinsheimer, “First Steps Toward a Genetic Chemistry,” Science 125 (1957): 1123–28. 68. Joseph S. Fruton and Sofia Simmonds, General Biochemistry, 2d ed. (New York: John Wiley, 1958), pp. 200–201; J. Fruton/FLH, conversation, 5 January 1996. 69. FWS, MM/FLH, 10 July 1992, 1:18–21. 70. UL, C-64. 71. UL, C-65. 72. UL, C-46, C-65 to C-67. 73. J. J. Weigle and M. S. Meselson, “Studies of the Phage Lambda by the Method of Density-Gradient Centrifugation,” Report for the Year 1957–1958 on the Research and Other Activities of the Division of Biology at the California Institute of Technology, p. 119; “Studies of the Transducing Lambda Phage by Density-Gradient Centrifugation,” ibid., pp. 119–20. 74. UL, C-68, C-69. 75. UL, C-71. 76. MMmc. 77. UL, C-72. 78. Meselson to Levinthal, 8 November 1957, MM. 79. FWS, MM/FLH, 10 July 1996, 1:24; Workbook No. 786, p. 61. 80. UL, C-73, C-74; MM/FLH, 14 July 1998, 2:19–20 81. The graph is captioned “C74–5 #7 Coli RNA,” MM.

Chapter Eleven. Centrifugal Forces 1. Program of Gordon Research Conference on Nucleic Acids and Proteins, New Hampshire, 16–20 June 1958. 2. Luria to Meselson, 27 December 1957, MM. 3. FWS/FLH, 21 November 1988, pp. 35–36; 4 May 1992, p. 7. 4. [Franklin Stahl], Report for the Year 1957–1958 on the Research and Other Activities of the Division of Biology at the California Institute of Technology, pp. 5–6.

488

N OTES TO

PAGES 353 –364

5. FWS/FLH, 21 November 1988, p. 39. 6. Stahl to Doermann, 28 January 1958, FWS. 7. Meselson to Watson, 8 November 1957, MM. 8. FWS, MM/FLH, 10 July 1992, 1:28. 9. FWS/FLH, 12 July 1992, 3:26. 10. Stahl, Report for the Year 1957–1958, p. 8. 11. Stahl to Doermann, 28 January 1958. 12. Betty E. Terzaghi, George Streisinger, and Franklin W. Stahl, “The Mechanism of 5-Bromouracil Mutagenesis in the Bacteriophage T4,” PNAS 48 (1962): 1519–24, esp. p. 1523; FWS/FLH, 4 May 1992, p. 2. 13. Charley Steinberg and Frank Stahl, “The Clone-Size Distribution of Mutants Arising from a Steady-State Pool of Vegetative Phage,” Journal of Theoretical Biology 1 (1961): 488–97; FWS/FLH, 12 July 1992, 3:20–21. 14. Vinograd to Zamecnik, 20 December 1957, Vinograd papers, CTA, 8.1. 15. [Vinograd to National Science Foundation], undated; Meselson to Vinograd, undated, Vinograd papers, CTA, 5.17. 16. Meselson to Watson, 8 November 1957, 18 May 1958, JDW; Delbru¨ck to Freese, 3 January 1958, MD, 8.5; FWS, MM/FLH, 10 July 1992, 1:26–27; MM/FLH, 4 June 1996, 4:17. 17. Horace Freeland Judson, The Eighth Day of Creation: Makers of the Revolution in Biology (New York: Simon and Schuster, 1979), pp. 188, 192; MM/FLH, 2 December 1987, pp. 30–31; FWS, MM/FLH, 10 July 1992, 1:7–9. 18. Arthur Kornberg, “Pathways of Enzymatic Synthesis of Nucleotides and Polynucleotides,” in A Symposium on the Chemical Basis of Heredity, ed. William D. McElroy and Bentley Glass (Baltimore: Johns Hopkins University Press, 1957), pp. 579–608; Arthur Kornberg, For the Love of Enzymes: The Odyssey of a Biochemist (Cambridge: Harvard University Press, 1989), pp. 147–63. 19. FWS, MM/FLH, 10 July 1992, 1:7. 20. Meselson to Luria, 18 January 1958, MM. 21. FWS, MM/FLH, 10 July 1992, 1:8. 22. Meselson to Luria, 18 January 1958, MM. 23. Delbru¨ck to Warburg, 23 December 1957, MD, 23.16. 24. Delbru¨ck to Freese, 3 January 1958, MD, 8.5. 25. UL, B-157 to B-159, B-166, 1476–1477, C-95. 26. UL, B-157. 27. UL, B-158, B-159. 28. UL, 1476. 29. UL, C-95; MMmc; Matthew Meselson and Franklin W. Stahl, “The Replication of DNA in Escherichia Coli,” PNAS 44 (1958): 679. 30. UL, B-166, 1477. 31. UL, C-96; FWS, MM/FLH, 10 July 1992, 1:12. Stahl’s later densitometer tracings have generally not survived, but the tracing of C-95–3, run on the unheated hybrid DNA, is preserved in a loose-leaf notebook containing vari-

N OTES TO PAGES 364 –375

489

ous other notes, calculations, photographs of bands, and graphs from this period kept by Meselson. MM. 32. FWS, MM/FLH, 10 July 1992, 1:14; Meselson to Watson, postcard, undated, JDW. 33. Richard B. Roberts, ed., Microsomal Particles and Protein Synthesis (New York: Pergamon Press, 1958), pp. v–vi, 36–61, 95–99. 34. MM/FLH, 4 June 1996, 5:21. 35. Taylor to Delbru¨ck, 12 November [1957], MD, 21.25. 36. Taylor to Delbru¨ck, 29 April 1958, MD, 21.25. 37. Ibid. 38. Meselson to Watson, undated; UL, C-108 to C-110. 39. UL, C-112, B-203. 40. MM/FLH, 2 December 1987, p. 36. 41. Brenner to Meselson, 18 February 1958, MM. 42. Meselson to Anfinsen, undated, MM; Christian B. Anfinsen, The Molecular Basis of Evolution (New York: John Wiley, 1959), pp. vii, 52–55. 43. Untitled manuscript, dated “Feb. ‘58,” MM. 44. MM/FLH, 2 December 1987, p. 30; FWS, MM/FLH, 10 July 1992, 1: 16–17. 45. Manuscript dated “Feb. 58”; Meselson and Stahl, “Replication of DNA,” pp. 680–81. 46. Delbru¨ck to Watson, 28 February 1958, MD, 23.24; FWS, MM/FLH, 11 July 1992, 2:1. 47. John Cairns to Sam Elworthy, 19 May 1997. 48. Meselson to Anfinsen, undated; FWS/FLH, 21 November 1988, p. 91, 4 May 1992, p. 4. 49. Meselson to Anfinsen, undated. 50. Delbru¨ck to Watson, 28 February 1958. 51. Stahl to Doermann, 28 January 1958, FWS. 52. Taylor to Delbru¨ck, 29 April 1958; FWS/FLH, 4 May 1992, p. 26, NUI, p. 6. 53. FWS/FLH, 4 May 1992, p. 25, 12 July 1992, 3:22. 54. Delbru¨ck to Edgar, 28 March 1958, MD, 7.1. 55. Ibid. 56. Delbru¨ck to Freese, 28 April 1958, MD, 8.5. 57. FWS/FLH, 21 November 1988, p. 26; FWS, MM/FLH, 12 July 1992, 3:23. 58. Seymour Benzer, “Adventures in the rII Region,” Phage and The Origins of Molecular Biology, ed. John Cairns, Gunther Short, and James D. Watson (Cold Spring Harbor: Cold Spring Harbor Laboratory), p. 158. 59. Meselson to Watson, 18 May 1958, JDW. 60. MM/FLH, 7 December 1987, pp. 38–39. 61. Meselson and Stahl, “Replication of DNA,” pp. 671–72. 62. FWS, MM/FLH, 12 July 1992, 3:2.

490

N OTES TO

PAGES 377 –392

63. Meselson and Stahl, “Replication of DNA,” pp. 672–76. 64. Ibid., p. 676. 65. Ibid., pp. 676–77. 66. Ibid., p. 681. 67. MM/FLH, 21 November 1989, p. 28. 68. FWS, MM/FLH, 11 July 1992, 2:21–22; Stahl to FLH, 31 December 1999; Meselson to FLH, telephone interview, 3 January 2000. 69. FWS/FLH, 21 November 1988, pp. 32–33. 70. Meselson and Stahl, “Replication of DNA,” p. 677. 71. MMmc. 72. Meselson and Stahl, “Replication of DNA,” pp. 677–78. 73. Ibid., pp. 678–79. 74. Ibid., pp. 679–81. In 1992 Meselson commented, “If I could change every copy in the world, I would take all that salmon sperm stuff out of the paper.” MM/FLH, 2 July 1992, 3:5. He explained further, in 1996, that Doty had been wrong and that he regretted that he had failed to challenge Doty. MM/FLH, 4 June 1996. 75. MM/FLH, 21 November 1989, p. 35. 76. Meselson and Stahl, “Replication of DNA,” pp. 681–82. 77. Meselson to Watson, 18 May 1958, MM. 78. “Introduction,” Report for the Year 1957–1958, pp. 3–4. 79. M. Wilkins to Meselson, 3 June 1958, MM. 80. FWS/FLH, 21 November 1988, p. 42.

Chapter Twelve. The Subunits of Semiconservative Replication 1. Personal conversation, 5 January 1996. 2. Personal conversation, 5 May 1996. 3. “List of Those Attending the Symposium,” CSHS, vol. 23 (1958): ix– xiv. 4. M. Meselson and F. W. Stahl, “The Replication of DNA,” CHSH, vol. 23 (1958), p. 9. 5. Ibid., p. 10. 6. Ibid., pp. 10–11. 7. Ibid., p. 11. 8. Ibid., pp. 11–12; C. Thomas to FLH, 1 January 1997. 9. MM, FWS/FLH, 10 July 1992, 1:32; 4 June 1996, 5:28. 10. Liebe F. Cavalieri, Barbara Hatch Rosenberg, and Joan F. Deutsch, “The Subunit of Deoxyribonucleic Acid,” Biochemical and Biophysical Research Communications 1 (1959): 124–28. 11. Ibid. 12. Ronald Rolfe, “The Molecular Arrangement of the Conserved Subunits

N OTES TO PAGES 392 –399

491

of the Deoxyribonucleic Acid,” Ph.D. diss., California Institute of Technology, Pasadena, 1961. 13. H. Eldon Sutton, ed. Genetics: Genetic Information and the Control of Protein Structure and Function (“Transactions of the First Conference October 19, 20, 21, and 22, Princeton, N.J.,” New York: Madison Printing Co., 1960), pp. 28–35, 52. 14. Ibid., p. 35. 15. Ibid., pp. 47–49. 16. Peter F. Davison, “The Effect of Hydrodynamic Shear on the Deoxyribonucleic Acid from T2 and T4 Bacteriophages,” PNAS 45 (1959): 1560– 68. 17. Sutton, Genetics, p. 38. 18. MMmc. 19. Sutton, Genetics, pp. 49–50. 20. Ibid., p. 29; Joseph D. Mandell and A. D. Hershey, “A Fractionating Column for Analysis of Nucleic Acids,” Analytical Biochemistry 1 (1960): 66– 67, 74–75. 21. I. Rubenstein, C. A. Thomas Jr., and A. D. Hershey, “The Molecular Weights of T2 Bacteriophage DNA and Its First and Second Breakage Products,” PNAS 47 (1961): 1113–22. 22. P. F. Davison, D. Freifelder, R. Hede, and C. Levinthal, “The Structural Unity of the DNA of T2 Bacteriophage,” PNAS 47 (1961): pp. 1123–29. 23. Fred E. Abbro and Arthur Pardee, “Synthesis of Macromolecules in Synchronously Dividing Bacteria,” Biochimica et Biophysica Acta 39 (1960): 478–85. 24. R. G. Wake and R. L. Baldwin, “Physical Studies on the Replication of DNA in vitro,” Journal of Molecular Biology 5 (1962): 201–16. 25. Ibid., p. 204. 26. Saul Kit, “Deoxyribonucleic Acids,” Annual Review of Biochemistry 32 (1963): 64. 27. Liebe Cavalieri and Barbara Hatch Rosenberg, “The Replication of DNA: III. Changes in the Number of Strands in E. coli DNA During Its Reproduction,” Biophysical Journal 1 (1961): 344–47. 28. Liebe Cavalieri and Barbara H. Rosenberg, “Nucleic Acids: Molecular Biology of DNA,” Annual Review of Biochemistry 31 (1962): 258, 260. 29. Wake and Baldwin, “Replication of DNA in vitro,” p. 312. 30. Kit, “Deoxyribonucleic Acids,” p. 64. 31. John Robert Menninger, “A Determination of the Mass per Length of DNA Using X-ray Diffraction,” Ph.D. diss., Harvard University, 1963, pp. 1– 2, 33–34. 32. MM/FLH, 4 June 1996, 6:5. 33. J. Cairns to Holmes, 18 June 1996. 34. Cairns to Meselson, 29 May 1958, MM.

492

N OTES TO

PAGES 400 –406

35. H. J. F. Cairns, “The Asynchrony of Infection by Influenza Virus,” Virology 3 (1957): 1–14; John Cairns, “The Initiation of Vaccinia Infection,” ibid., 11 (1960): 603–23; John Cairns, “The Autoradiography of DNA,” in Phage and the Origins of Molecular Biology, ed. John Cairns, Gunther S. Stent, and James D. Watson (Cold Spring Harbor: Cold Spring Harbor Laboratory of Quantitative Biology, 1966), pp. 252, 254. 36. J. Cairns/FLH, 13 July 1993, 1:26. 37. Ibid.; Cairns, “Autoradiography of DNA,” p. 255; Cairns to Holmes, 19 June 1996. 38. J. Cairns/FLH, 13 July 1993, 1:27; Cairns, “Autoradiography of DNA,” pp. 254–55. 39. John Cairns, “An Estimate of the Length of the DNA Molecule of T2 Bacteriophage by Autoradiography,” Journal of Molecular Biology 3 (1961): 756–61. 40. Ibid., p. 760. 41. Ibid., pp. 758–59. 42. John Cairns, “The Size of the Unit of Heredity,” in We Can Sleep Later: Alfred Hershey and the Origins of Molecular Biology, ed. Franklin W. Stahl (Cold Spring Harbor, N.Y.: Cold Spring Harbor Laboratory Press, 2000), pp. 49–58, on 56. 43. Meselson to Weigle, 12 March 1961, Weigle archive, CTA. 44. J. Cairns/FLH, 13 July 1993, 1:27–28. 45. John Cairns, “Proof That the Replication of DNA Involves the Separation of the Strands,” Nature 194 (1962): 1274. 46. J. Weigle and Matthew Meselson, “Density Alterations Associated with Transducing Ability in the Bacteriophage Lambda,” Journal of Molecular Biology 1 (1959): 379–86. 47. M. Meselson and J. J. Weigle, “Chromosome Breakage Accompanying Genetic Recombination in Bacteriophage,” PNAS 47 (1961): 857–68. 48. Ibid., pp. 859–65. 49. Ibid., p. 865. 50. Ibid. 51. Cairns, “Replication of DNA,” p. 1274. 52. Cairns, “Autoradiography of DNA,” p. 255. 53. Cairns, “Replication of DNA,” p. 1274. 54. Ibid. 55. Cairns, “Autoradiography of DNA,” p. 256. 56. J. Cairns/FLH, 13 July 1993, 2:10. 57. Cairns, “Autoradiography of DNA,” p. 256; John Cairns, “A Minimum Estimate for the Length of the DNA of Escherichia Coli Obtained by Autoradiography,” Journal of Molecular Biology 4 (1962): 407; J. Cairns/FLH, 13 July 1993, 1:14.

N OTES TO PAGES 406 –418

493

58. Cairns, “Minimum Estimate,” pp. 407–8. 59. Ibid., pp. 408–9. 60. Cairns, “Minimum Estimate,” p. 408; “Autoradiography of DNA,” p. 256; J. Cairns/FLH, 13 July 1993, 2:13. 61. J. Cairns/FLH, 13 July 1993, 2:13. 62. John Cairns, “The Bacterial Chromosome and Its Manner of Replication as Seen by Autoradiography,” Journal of Molecular Biology 6 (1963): 208– 13. 63. Ibid., pp. 210–11. 64. Ibid., p. 208. 65. Ibid., pp. 211–13; John Cairns, “The Chromosome of Escherichia coli,” CSHS, vol. 28 (1963): 43–46. Soon afterward it was recognized that there are two forks, one at the growing point and the other at the origin of replication. For a succinct review of the subject that integrates his own contribution with other contemporary investigations, see John Cairns, “DNA Synthesis,” The Harvey Lectures 66 (1970–71): 4–7. 66. MM/FLH, 2 December 1987, p. 35. 67. MM/FLH, 4 June 1996, 6:5. 68. Ibid., 5:29. 69. For further examples and general discussion, see “Special Section: The Right Organism for the Job,” Journal of the History of Biology 26 (1993): 233–367. 70. Stahl to Holmes, 31 December 1999.

Chapter Thirteen. Images of an Experiment 1. J. D. Watson, Molecular Biology of the Gene (New York: W. A. Benjamin, 1965), p. ix. 2. Ibid., pp. 255–66. 3. Ibid., p. 268. 4. Ibid., pp. 268–71. 5. Ibid., p. 271. 6. Ibid., p. 272. 7. Ibid., p. x. 8. J. J. W. Baker and Garland E. Allen, The Study of Biology (Reading, Mass.: Addison-Wesley, 1967), pp. 394–96. 9. Albert L. Lehninger, Biochemistry: The Molecular Basis of Cell Structure and Function (New York: Worth, 1970), p. vii. 10. Ibid., p. 657. 11. Ibid., pp. 659–61. 12. Ibid., p. 660. 13. Ibid., p. 661.

494

N OTES TO

PAGES 419 –428

14. Ibid. 15. Ursula Goodenough and Robert Paul Levine, Genetics (New York: Holt, Rinehart and Winston, 1974), p. v. 16. Ibid., pp. 95–96. 17. Ibid., pp. 96–97. 18. Ibid., p. 97. 19. For other examples, see Arthur Kornberg, DNA Replication (San Francisco: W. H. Freeman, 74), pp. 347–48; Roger L. P. Adams, John T. Knowler, and David P. Leader, The Biochemistry of the Nucleic Acids (London: Chapman and Hall, 1986), pp. 136–38; Benjamin Lewin, Genes, 3d ed. (New York: John Wiley and Sons, 1987), p. 49; Geoffrey Zubay, Genetics (Menlo Park, Calif.: Benjamin Cummings, 1987), pp. 101–3; Norman V. Rothwell, Understanding Genetics, 4th ed. (New York: Oxford University Press, 1988), pp. 294–96; Monroe W. Strickenberger, Genetics, 2d ed. (New York: Macmillan, 1976), pp. 72–75; Leon A. Snyder, David Freifelder, and Daniel L. Hartl, General Genetics (Boston: Jones and Bartlett, 1985), pp. 122–24. 20. Neil A. Campbell, Biology, 3d ed. (Redwood City, Calif.: Benjamin/ Cummings, 1993), p. 308. 21. Lubert Stryer, Biochemistry (San Francisco: W. H. Freeman, 1975), pp. 568–70. 22. Ibid., pp. 566–70. 23. Bruce Alberts, Dennis Bray, Julian Lewis, Martin Raff, Keith Roberts, and James D. Watson, Molecular Biology of the Cell (New York: Garland Publishing, 1989), pp. 96–97, 100. 24. Maxine Singer and Paul Berg, Genes and Genomes: A Changing Perspective (Mill Valley, Calif.: University Science Books, 1991), pp. 75–76. 25. Franklin W. Stahl, The Mechanics of Inheritance (Englewood Cliffs, N.J.: Prentice-Hall, 1964), p. 41. 26. GS/FLH, 5–6 May 1992, p. 36. 27. Ibid., pp. 34–35. 28. Ibid., p. 35. 29. C. Hill/FLH, 15 April 1993 (recorded but untranscribed conversation). 30. Horace Freeland Judson, The Eighth Day of Creation: Makers of the Revolution in Biology (New York: Simon and Schuster, 1979), p. 188. 31. The Shorter Oxford English Dictionary, 3d ed. (Oxford: Clarendon Press, 1973) 1:171. 32. MM/FLH, 2 December 1987, p. 44. 33. Judith Wechsler, “Introduction,” in On Aesthetics in Science, ed. Judith Wechsler (Cambridge: MIT Press, 1981), p. 6. See also Alfred I. Tauber, ed., The Elusive Synthesis: Aesthetics and Science (Dordrecht: Kluwer, 1996). 34. Helge Kragh, Dirac: A Scientific Biography (Cambridge: Cambridge University Press, 1990), p. 287. 35. James W. McAllister, “Truth and Beauty in Scientific Reason,” Syn-

N OTES TO PAGES 428 –440

495

these 78 (1989): 23–51; McAllister, Beauty and Revolution in Science (Ithaca: Cornell University Press, 1996). 36. See M. Alexenberg, Aesthetic Experience in Creative Process (Ramal Gan Israel: Bar Ilan University Press, 1981), pp. 148–202. Alexenberg seems to conflate aesthetic with emotional experience in scientific creativity. Several of his subjects, however, refer prominently to experiences of beauty in the course of their investigations. 37. J. Cairns/FLH, 13 July 1993, 2:4–5. 38. GS/FLH, 5–6 May 1992, 2:34. 39. FWS/FLH, 21 November 1988, p. 94. 40. Ibid., p. 25. 41. Ibid., p. 57. 42. H. Schachman/FLH, 2 May 1994, pp. 29–30. 43. JDW/FLH, 5 March 1990, pp. 2–3. 44. C. Hill/FLH, 15 April 1993. 45. FWS/FLH, 10 July 1992. ¨ sthetische Momente 46. Ernst Peter Fischer, Das Scho¨ne und das Biest: A in der Wissenschaft (Munich: Piper, 1997), pp. 46–49. I thank Manfred Laubichler for bringing this book to my attention.

Chapter Fourteen. Afterword 1. Michel Morange, Histoire de la biologie mole´culaire (Paris: Editions de la De´couverte, 1994), p. 65. 2. MMmc; MM/FLH, 14 July 1998, 1:17. 3. MMmc; MM/FLH, 14 July 1998, 1:19. 4. J. Drake to Holmes, 23 August 1997. 5. MM, FWS/FLH, 11 July 1992, 3:8–9. 6. Meselson to Stahl, 25 September 1959, 22 January 1960; Stahl to Meselson, 15 January 1960, FWS. 7. Ernst Freese, “The Difference Between Spontaneous and BaseAnalogue Induced Mutations of Phage T4,” PNAS 45 (1959): 622–33; David Pratt and Gunther Stent, “Mutational Heterozygotes in Bacteriophages,” Ibid., pp. 1507–15; Stent to Holmes, 29 December 1997. 8. Meselson to Stahl, 9 October 1958, undated; Stahl to Meselson, 2 March 1959, FWS. 9. Meselson to Stahl, 2 October [1958], undated, 18 March 1959, 18 May 1959, FWS. 10. Stahl to Meselson, 2 March 1959; FWS/FLH, 21 November 1988, pp. 40–41, 88; 4 May 1992, p. 9. 11. Stahl to Meselson, 12 November 1959; FWS/FLH, 4 May 1992, pp. 9– 12. 12. FWS/FLH, NUI, 4 May 1992; Franklin W. Stahl, Jean M. Crasemann,

496

N OTES TO

PAGES 440 –447

Larry Okun, Evelyn Fox, and Charles Laird, “Radiation Sensitivity of Bacteriophage Containing 5-Bromodeoxyuridine,” Virology 13 (1961): 99. 13. Stahl et al., “Radiation Sensitivity,” pp. 98–104; Stahl to FLH, 20 April 1998. 14. Betty E. Terzaghi, George Streisinger, and Franklin W. Stahl, “The Mechanism of 5-Bromouracil Mutagenesis in the Bacteriophage T4,” PNAS 48 (1962): 1519–24. 15. Ibid.; FWS/FLH, 4 May 1992, NUI, and p. 25. 16. Franklin W. Stahl, The Mechanics of Inheritance (Englewood Cliffs, N.J.: Prentice-Hall, 1964), pp. vi, 47–56. 17. James D. Watson, Molecular Biology of the Gene, 2d ed. (New York: W. A. Benjamin, 1970), pp. 300–6. 18. FWS/FLH, 4 May 1992, NUI and pp. 2–3. 19. Franklin W. Stahl, R. S. Edgar, and Jacob Steinberg, “The Linkage Map of Bacteriophage T4,” Genetics 50 (1964): 539–52. 20. For a compact overview of Stahl’s career, see John W. Drake, “The 1996 Thomas Hunt Morgan Medal: Franklin W. Stahl,” Genetics 145 (1997): 1–4. 21. Meselson to Robert Haselkorn, 18 June 1960, MM. 22. Ibid.; S. Brenner, F. Jacob, and M. Meselson, “An Unstable Intermediate Carrying Information from Genes to Ribosomes for Protein Synthesis,” Nature 190 (1961): 576–81. For accounts of the discovery of messenger RNA, see Horace Judson, The Eighth Day of Creation (New York: Simon and Schuster, 1979), pp. 427–35; Morange, Histoire, pp. 180–93; Franc¸ois Jacob, La statue inte´rieur (Paris: Seuil, 1987), pp. 344–54. 23. Meselson to Frank and Mary Stahl, 18 March 1959, FWS; Meselson to Weigle, 9 February 1961, Weigle archive, CTA. 24. Thomas D. Seeley, Joan W. Mowicke, Matthew Meselson, Jeanne Guillemin, and Pongthep Akranakul, “Yellow Rain,” Scientific American 253 (1985): 128–37. For a general summary of Meselson’s career in science and issues of warfare and disarmament, see John W. Drake, “The Thomas Hunt Morgan Medal: Matthew Meselson,” Genetics 142 (1996): 1–2. 25. Matthew Meselson, Jeanne Guillemin, Martin Hugh-Jones, Alexander Langmuir, Ilona Popova, Alexis Shelikov, and Olga Yampolskaya, “The Sverdlovsk Anthrax Outbreak of 1979,” Science 266 (1994): 1202–8. 26. MM/FLH, 2 December 1987, pp. 43–44. 27. FWS/FLH, 21 November 1988, p. 38. 28. MM, FWS/FLH, 12 July 1992, 3:10–11.

Index

Abbo, Fred: on bacterial synthesis of DNA, 396 Albert, Adrien: on properties of heterocyclic compounds, 123–124; on ionization of bases, 139–141 Alexander, P.: on denaturation of DNA, 284–285 Allen, Garland: textbook of, 416 Amino acids: substitution of in crystals, 99, 164–165, 210 Anderson, Thomas F.: on virus particles, 57 Anfinsen, Christian B.: and MeselsonStahl exp., 369 Arley, Niels: and DNA replication, 100 Astbury, William: on structure of DNA, 31 Autoradiography: and length of DNA molecule, 400, 402, 404–408 Avery, Oswald: and transforming factor, 59–60, 435 Bacteria: reproduction of, 91–92, 158 Bacteriophage: experimentation on, 55–56; genetic maps of, 56, 70; reproductive cycle of, 56–57, 94–95, 109–110, 355; recombination theory of Levinthal of, 88–89; as genetic model, 106, 108, 138; sedimentation properties of, 161–162, 170; separation of 5-bromouracil substituted, 196, 300; equilibrium centrifugation

of, 197–200, 216–217, 220–222; sensitivity of to radiation, 208–210, 239– 241, 288–289, 439–440 Baker, J. J. W., textbook of, 416 Base pairs in DNA, 118–119, 151–155; and helical structure of, 30–31, 32, 114; sequence of, 184–185, 368; alternatives to Watson-Crick pairs, 145–146 Beadle, George, 99, 254, 275, 337 Beauty: in science, ix–x, 7, 10, 48, 411, 428; of Meselson-Stahl exp., ix–x, 10, 427–434; of double helix, 28, 48; in experimentation, 337 Bendich, Aaron: on purines and pyrimidines, 117, 122 Benzer, Seymour: rII mutants of, 88, 93, 183; experiments of on genetic fine structure, 208, 298, 337, 343; on writing of papers, 373–374 Berg, Paul: textbook of, 424 Bernard, Claude, 3 Bertani, Giuseppe (Joe): experiments of on bacterial reproduction, 35, 91–92 Bloch, David: and DNA replication, 100–101, 397 Bohr, Niels, 14; on complementarity, 339, 344 Bragg, Lawrence: and double helix, 22 Brenner, Sydney, 113, 441; and Meselson-Stahl exp., 368–369; and messenger RNA, 444

498

I NDEX

5-Bromouracil: incorporation of into nucleic acids, 122, 135–138, 162–164, 172–173, 179–180, 208–210, 222– 224, 227–229, 239–240, 287–288, 354–355; 438; and mutagenesis, 135– 138, 158–160, 438–443; and density centrifugation, 137, 297 Broomhead, June: and crystal structure of guanine, 133–134, 138–139 Brown, D. J.: on pyrimidines, 125 Burnet, Macfarlane, 389 Cairns, John: on beauty of MeselsonStahl exp., ix, 428–429; on structure and function, 115; at Caltech, 299, 321–322, 324–325, 399; and Meselson-Stahl exp. 321–322, 398– 399; and units of DNA replication, 398; and Hershey, 400–401; and Delbru¨ck, 401–402; autoradiographic measurement of DNA molecule by, 404–408; demonstration of Watson– Crick model of replication by, 404– 408, 410; and mechanism of DNA replication, 408; photo, 348 Callenbach, Ernest, 322, 357 Campbell, Neil: textbook of, 422–423 Cavalieri, Liebe: and units of semiconservative replication, 393–393, 397–398 Cesium chloride, as medium for ultracentrifugation: choice of by Meselson, 175–179; use of by Schachman, 181, 202; formation of density gradient in, 194–195; purification of, 202, 241–242. See also Density gradient method Chargaff, Erwin: and base ratios in DNA, 13–14, 32; textbook of on nucleic acids, 117 Chase, Martha: and Hershey-Chase exp., 57; and Doermann, 64 Chromosomes: experiments of J. H. Taylor on replication of, 272–274, 341, 419–420 Clews, C. J. B.: on crystal structure of pyrimidines, 125–129, 145 Cochran, William: on crystal structure of pyrimidines, 125–129; of guanine, 132–133 Coding problem, 1, 36, 76, 100, 368 Cohen, Seymour: and 5-hydroxymethylcytosine, 14, 238 Cohn, Waldo: and structure of RNA, 115 Cold Spring Harbor: phage course taught there, 5; Symposium on Viruses, 21, 23, 28–29

Commoner, Barry: opposition of to double helix, 29 Corey, Robert: on chemical bonds, 125– 126; on base pairs, 152–155, 160 Crane, H. R.: “Speedometer cable” model of DNA replication of, 107, 109 Crick, Francis H. C., 30, 35: and double helix, ix, 1–3, 11–13, 18–21; and mutagenesis, 120–121, 442; and RNA, 115; on molecular biology, 345; and coding problem, 368 Cytidine: structure of, 31, 147 Davison, Peter: and shearing of DNA, 394–395, 405 Delbru¨ck, Max: and DNA replication problem, 2, 12–13, 25–27, 77–78, 107–112, 341–342; and phage group, 5, 32, 34, 35, 55–56, 95–96; and Niels Bohr, 14, 339, 344; scientific style of, 18, 95–96; and J. D. Watson, 12–21, 32, 34–35; and Phycomyces, 34, 339; and search for paradoxes in biology, 34, 339–340; mechanism of for DNA replication, 40–48, 276, 278–279, 296; and M. Meselson, 49, 54–55; theory of phage reproduction of, 94–95; and F. Stahl, 95–96, 307, 373; and J. H. Taylor, 275–282, 296, 342; and “sister exchanges” of chromatids, 276–278, 280–281; and Meselson-Stahl exp., 326, 340–341, 371, 372, 374; lecture of on molecular biology, 339–344; photo, 47, 348, 349, 387 Demars, Robert: on genetic recombination, 35; cross-reactivation exp. of, 73 Density gradient: formation of in cesium chloride, 195–195, 197 —method of separation, 196, 206; formation of bands in, 200–204; separation of 5–bromouracil from normal DNA in, 222–229, 289; theory of method, 231–234, 261–264; application to molecular weight determinations, 234, 236–238, 239, 242–247, 249–252, 255, 256, 258–260, 265, 300, 351, 364; centrifugation of RNA in, 349–351, 370; as general analytical method, 370, 386, 438 Deoxyribonucleic acid (DNA): double helix model of, ix, 1–3, 12–20, 24, 31, 32, 322; ratios of bases in, 13–14, 17–18, 24, 32; alternative structure models of, 26–27, 77–78; as genetic material, 58–60; titration of, 60–61;

I NDEX “big piece” of Levinthal of, 105, 111, 184, 234, 252–253, 256–258, 264– 265, 297–298, 338, 390, 395; physical properties of, 118; formation of bands of in density gradient, 200, 203–204; heat denaturation of, 284–287, 293– 294, 299–300, 362–364, 367, 383– 384; enzymatic synthesis of, 358, 412; bacterial synthesis of, 396; shearing of, 394–396; autoradiographic measurement of, 400, 402, 404–408 —replication of, 37, 75–84, 100–107; Delbru¨ck and, 2, 40–48, 77–78, 107– 112; semiconservative, 4, 109, 219– 220, 321, 402–405, 412–413, 418– 419; conservative, 109, 322–323, 418–419; dispersive, 109, 325–326, 384, 418–419; model of H. J. Taylor of, 274–275, 279–282, 329–331, 366. See also Base pairs, 5-Bromouracil, Density gradient Deutsch, Joan: collaborator of Cavalieri, 391 Doermann, August G. (Gus), 63, 307; exp. of on bacteriophage reproduction, 63–64, 70 Donohue, Jerry: on tautomeric forms, 130–131; on alternative base pairs, 145–146; and Stent, 322 Doty, Paul, 335, 365, 366: light-scattering method of, 30; and molecular weight of DNA, 238, 255, 259, 284; and denaturation of DNA, 284–286, 298, 299–300, 362 Double helix: research stimulated by, 6–7; genetic implications of, 11–29, 353; beauty of, 28, 48. See also Deoxyribonucleic acid, Watson-Crick model Douglas, William O., Justice of the Supreme Court, 96 Drake, John W. (Jan), 67–68, 90, 96, 213, 437 Dulbecco, Renato, 19, 137, 179, 399 Dunn, D. B.: and incorporation of 5bromouracil in phage DNA, 137, 168–169 Edgar, Robert, 292, 307, 400 Electrophoresis: as alternative method for separation of substituted and unsubstituted DNA, 161, 170–171, 184, 197, 229 Feynman, Richard: 35, 36, 52, 96, 265– 266: and Meselson-Stahl exp., 337, 409 Fischer, Ernst Peter: on beauty of Meselson-Stahl exp., 434

499

Franklin, Rosalind: x-ray photograph of, 12; and helical structure of DNA, 16, 17, 23–24, 28, 31, 49 Freese, Ernst, 343, 359, 365, 438 Fruton, Joseph S.: on Watson-Crick structure, 345–346; and MeselsonStahl exp. 388 Fuerst, Clarence: student of Stent, 102 Furberg, Sven: on structure of cytidine, 31, 147 Gamow, George: and coding problem, 36, 76, 100; and replication of DNA, 76 Gaussian distribution of densities, 245– 246, 250–252, 256, 259–260, 263 Goodenough, Ursula: textbook of, 420– 422 Griboff, Gertrude: collaborator of Zamenhof, 137–138, 169 Guillemin, Jean, 445 Gulland, Masson: and titration of DNA, 60–61 Hayes, William, 92 Heilbron, John L.: on historical complicity, 316 Hershey, Alfred: and phage group, 55– 56; and heterozygous phage markers, 88; and extraction of intact DNA molecules, 395, 400, 404. See also Hershey-Chase exp. Hershey-Chase experiment, 57–60, 337, 435; and experimental beauty, 7; J. D. Watson and, 58–60 Hill, Charles: on teaching MeselsonStahl exp., 427; on beauty of, 430– 431 Howard, Alma: and incorporation of 32P into plant cells, 272 Hughes, Walter L.: collaborator of J. H. Taylor, 273 Hunter, Louis: theory of mesohydric tautomerism of, 127–130 Huskins, C. L.: on coiling of chromosomes, 15–16 Huxley, Hugh: mechanism of muscle contraction of, 340 Ionization: —of bases, 118, 119; effects of substitution on, 149–151, 159–160, 165–168, 210–211 —as mechanism of mutagenesis, 122, 136, 139–141, 211–212 Investigative pathways in science, 9–10

500

I NDEX

Jacob, Franc¸ois, 28, 92, 444 Jerne, Niels: collaboration with G. Stent, 83–84 Johnson, Treat B.: collection of purines and pyrimidines of, 160 Jordan, Denis O.: on physical properties of DNA, 118; on base pairs, 151–152 Judson, Horace Freeland: and MeselsonStahl exp., ix, 2; on J. D. Watson, 6; on Hershey-Chase exp., 58; on choice of centrifuge solution, 175 Kit, Saul: on Meselson-Stahl exp., 397, 398 Kornberg, Arthur: 358, 412 Koshland, Daniel E., 115 Kragh, Helge: on beauty in science, 428 Krebs, Hans Adolf, 3 Kuhn, Thomas S.: on puzzle-solving, 40 Lavoisier, Antoine-Laurent, 3: on collaboration in science, 269 Lederberg, Joshua, 392: on sexual reproduction in bacteria, 91–92; and phage λ, 304 Lehninger, Albert: textbook of, 417– 419 Levine, Robert Paul: textbook of, 420– 422 Levinthal, Cyrus, 29, 352: theory of phage recombination of, 88–89, 92– 93, 337; and F. Stahl, 92–93; and replication of DNA, 102–107, 279, 393; “big piece” of DNA of, 105, 111, 184, 234, 252–253, 256–258, 264–265, 297–298, 338, 390, 395; “speedometer cable” model of, 107, 109, 322, 393; on beauty of Meselson-Stahl exp., 337–338, 431–432; and Taylor units, 338; and molecular weight of DNA, 338–339, 395 Litman, Rose: and incorporation of 5bromouracil in phage DNA, 135–138, 163, 168, 179–180 Litt, Michael, 295, 298, 367 Luria, Salvador, 389: and phage group, 55–56; on bacteriophage reproduction, 58–59; on clonal distribution of phage recombinants, 94; and Meselson-Stahl exp., 352 Lwoff, Andre´: and G. Stent, 79, 80; on prophage, 91 Lysogeny, 91 Maaløe, Ole, 59, 79, 337 McAllister, James: on beauty in science, 428

McConnell, Harden, 159–160 Medawar, Peter: on Watson and Crick, 1, 3 Memory: and narrative, 7–8; and first Meselson-Stahl exp., 308–309, 312– 318 Menninger, John: student of Meselson, 398 Meselson, Matthew, 2: idea of for DNA replication exp., 6, 8–9, 53–54, 65– 66; and Delbru¨ck, 49, 54–55, 326, 344, 373–374; early life and education of, 49–53; and x-ray crystallography, 49, 52, 72–73, 98, 260–261; and Pauling, 51–52, 54, 96, 99–100, 266– 267, 445; and J.D. Watson, 55, 212; at Woods Hole, 60–62, 65–68, 69; and titration of RNA, 60–62, 69; initiation of collaboration of with Stahl, 65–66, 162, 195–197; examinations taken by, 74, 264–266; and G. Stent, 82; and international issues, 96, 445; and mutagenesis, 121–122, 136, 138, 141–143, 158–160, 166, 211–212, 264; and amino acid substitution in crystals, 99, 164–165, 210; and ultracentrifugation, 172–174, 193–194, 200–201, 359, 373; and choice of cesium chloride, 175–179; and theory of densitygradient centrifugation, 231–234, 262–264; and Taylor model of DNA replication, 282, 329–331, 366, 390; experiments of with phage λ, 304–305, 332, 338, 346–347, 349, 366, 402– 404; on simplicity and complexity in science, 436; later life and career, 444–445; photo, 205, 349, 357, 446 Meselson-Stahl experiment: beauty of, ix–x, 10, 337, 427–434; historical place of, ix, 2–3, 435–436; textbook descriptions of, 4, 413–425; roles of Meselson and of Stahl in, 8–9; performance of, 319–321, 333–334, 360– 361; informal communication of results of, 326–329, 332–337, 353, 358, 371–372; composite photograph of results of, 336, 378; first responses to, 337–338, 352, 368–369, 387; publication of, 373–379, 381–386, 389–390; conclusions drawn from, 377–379, 382, 389; and fragility of DNA, 395– 397; early reputation of, 396–397; schematic representations of, 415, 417, 420, 422, 423, 425, 426; use of in course lectures, 425–427 Molecular weight: density gradient method for determining, 234, 235,

I NDEX 236–238, 239, 242–247, 249–252, 254, 256, 258–260, 265, 300, 351; light-scattering method of Doty, 238, 255, 259, 284 Monod, Jacques: lectures on induced enzyme synthesis, 52–53 Morange, Michel: on double helix, 1–2; on types of experiments, 435 Morgan, Thomas Hunt: and genetic recombination, 56 Morse, M. L.: student of J. Lederberg, 304 Mutagenesis: scheme of Watson and Crick of, 120–121, 442; views of Meselson on, 121–122, 136, 158–160, 264; theory of Meselson of, 198, 264, 442; plan of Meselson and Stahl for investigation of, 183, 184–185, 298; experiments on by Stahl, 206–207, 438–443. See also 5-Bromouracil Nazarian, Girair (Jerry): and theory of ultracentrifuge equilibrium, 257, 263 Ochoa, Severo: synthesis of artificial polynucleotides by, 112 Olby, Robert: on Hershey-Chase exp., 58, 59 Organisms: choice of for experiments, 410–411 Orgel, Leslie: and RNA, 36 Paranemic coiling, 16, 25–26, 27 Pardee, Arthur: and incorporation of 5bromouracil in phage DNA, 135–138, 163, 168; on bacterial synthesis of DNA, 396 Parry, G. S.: on structure of uracil, 143– 144 Pasternak, Raphael: as supervisor to Meselson, 52 Pauling, Linus, 403: as mentor to Meselson, 5–6, 51–52, 54, 90, 99, 264–266, 344; organization of protein conference by, 30; and J. D. Watson, 34; and x-ray crystallography, 72; on base pairs, 119, 152–155, 160; on chemical bonds, 125–126; and Meselson-Stahl exp., 337 Pederson, Kai: and theory of ultracentrifugation, 173; on formation of density gradients, 194–195 Pelc, S. R., and incorporation of 32P into plant cells, 272 Penfold, Bruce R.: and structure α-pyridone, 144 Perutz, Max F., 12

501

Phage group: leadership of by Delbru¨ck, 5, 32, 34, 35, 55–56, 95–96; research style of, 55–57 Phillips, J. N.: on ionization of bases, 139–141 Pickels, Edward G.: designer of Model E analytical ultracentrifuge, 181, 191 Pitt, G. J.: on structure of dimethylhydroxypyridine, 129–130 Platt, John R.: and replication of DNA, 75 Plectonemic coiling, 16, 21–22, 25–26, 27 Pontecorvo, Guido: on genetic replication and recombination, 80 Popper, Karl: on falsification, 443 Purines: resonance structures of, 117; crystal structures of, 130, 132–134 Putnam, Frank W.: on sedimentation of bacteriophage Pyrimidines: resonance structures of, 117; chemistry of, 125; crystal structures of, 125–130 Radiation: sensitivity of 5-bromouracil substituted bacteriophage to, 208– 210, 239–241, 288–289, 439–440 Rheinberger, Hans-Jo¨rg: on experimental systems, 3–4 Ribonucleic acid (RNA): intermediary between DNA and protein, 1; x-ray photography of, 32–33, 112–113; structure of, 32–40, 112–115; models of, 33, 36, 39; density gradient centrifugation of, 349–351, 370; discovery of messenger, 444 Rice, Stuart: collaborator of P. Doty, 285 Rich, Alexander: and structure of RNA, 32–33, 36, 112–113; and preparation of RNA, 61 Robkin, Eugene: ultracentrifugation by, supervised by Meselson, 283–284, 293, 298–299 Rolfe, Ronald: student of Meselson, 392 Roller, Ann: study of replication of DNA in bacteriophage, 346; ultracentrifugation experiments by, 346, 347 Rosenberg, Barbara: collaborator of Cavalieri, 391 Schachman, Howard, 358; theory of interrupted strands of DNA by, 45–46; and ultracentrifugation, 180–181, 262 Schlieren optics of ultracentrifuge, 189–190; and formation of density gradient, 194, 235

502

I NDEX

Schro¨dinger, Erwin: author of What Is Life?, 18, 27–28 “Shockate”: definition of, 204 Simon, Edward: transfer experiment by with HeLa cells, 295, 298–299 Simplicity in science, ix–x, 429, 430, 436 Singer, Joseph: on structure of metanilamido pyrimidine, 145 Singer, Maxine: textbook of, 424 Sinsheimer, Robert: and double helix, 16, 17, 29–30; on role of DNA, 215– 216; and discovery of glucose substitution in phage DNA, 238, 249; analytical ultracentrifuge of, 254, 283; arrival at Caltech, 283; and J. Cairns, 321; and Meselson-Stahl experiment, 326 “Sister chromatid exchanges” in chromosome replication, 276–278, 280– 281, 365 Smith, John D., 202; and incorporation of 5-bromouracil in phage DNA, 137, 168–169 Spiegelman, Sol, 392 Stacey, K. A.: on denaturation of DNA, 284–285 Stahl, Franklin W.: role in MeselsonStahl exp., 2, 8–9; early life and education of, 62–65; at Woods Hole, 62, 65–68; and cross-reactivation of markers , 62–63, 71–72, 73, 87–88, 207; and Meselson, 62, 162, 305–307, 308, 352–354, 447; experiments of on bacteriophage reproduction, 63–65, 70– 71, 85–88; marriage of, 89–90, 96–97; family, 213, 439; and C. Levinthal, 92–94; experiments of on bacterial mating, 92, 97, 162; and Berkeley, 157–158, 162; experiments on mutagenesis by, 158–159, 179–180, 206– 207, 236, 251, 354–355; and densitometer analysis of density gradient bands, 200–201, 217–218, 220–222, 234, 237, 246–247, 249–251, 258– 260, 364, 371; decision to leave Caltech, 307, 372–373; and University of Missouri, 319, 373, 374, 387, 438– 439, 446; and theoretical analysis of phage reproduction, 355; at University of Oregon, 439–440; on beauty of Meselson-Stahl exp., 429–430, 431; photo, 97, 387, 447 Stahl, Mary Morgan, 89–90, 96–97, 439, 443–444; photo, 97 Stanley, Wendell, 81, 157 Steinberg, Charles, 95, 96, 355; photo, 350

Stent, Gunther, 352; on molecular biology, 1; experiments of on incorporation of 32P in DNA, 44–45; on replication of DNA, 44–46, 80–85, 101, 107–112, 280, 322–323; early life and education of, 78–79; and Delbru¨ck, 79, 84, 107–112; and Meselson, 82; and Stahl, 134–135, 157, 162; and Meselson-Stahl exp., 324–325; on teaching Meselson-Stahl exp., 426– 427; on beauty of, 429; and incorporation of 5-bromouracil in phage DNA, 438; photo, 85 Streisinger, George, 92–93, 441; and genetics of phage host range, 35 Stryer, Lubert: textbook of, 422–423 Sueoko, Noboru: plan to study DNA replication in algae, 360 Svedberg, Theodor (The): and theory of ultracentrifugation, 173, 230–231, 262; and development of ultracentrifuge, 191 Szilard, Leo, 445 Target theory: and radiation of bacteriophage, 208–209 Tatum, Edward: on sexual reproduction in bacteria, 91 Tautomerism, 118, 124; mesohydric theory of Hunter of, 127–130, 144–145; tautomeric forms of adenine, 131; tautomeric ratios, 139–141, 211 Taylor, James Herbert: and replication of chromosomes in bean plant seedlings, 2, 272–275, 296, 390, 419–420; model of DNA replication of, 274– 275, 282, 329–331, 366, 390; and Meselson-Stahl exp., 365, 373 Temin, Howard, 213; photo, 350 Thomas, Charles, 395; and “big piece” of phage DNA, 184, 297–298, 390 Thomas, Rene´: on denaturation of DNA, 284 Todd, Alexander: on structure of RNA, 115 Transfer experiment: Meselson plan for, 218–220, 237, 269–270, 296–297; with 5-bromouracil substituted phage T4 DNA, 247–249, 291–296, 301– 302; with HeLa cells, 298–299; with 15 N substituted E. coli DNA, 307–314, 319–321, 329, 347–349, 360–361 Ultracentrifuge: —preparatory Model L, 179, 402–403 —Spinco Model E analytical, 180–181; description of, 186–191; acquisition

I NDEX of by Caltech, 191–192; J. Vinograd and, 191–192; operation of, 193– 194, 200–201; photo, 187, 192, 202 Ultracentrifugation: limits of resolution of, 172; theory of, 173–174, 230– 231, 262; test for equilibrium, 242– 245, 256–258; of “sweet and sour” DNA by, 249, 253; separation of 5– bromouracil substituted from unsubstituted phage DNA by, 288–291; separation of 15N substituted from unsubstituted E. coli DNA by, 305; separation of denatured from undenatured DNA by, 285–287, 362– 364, 367, 383–384. See also Transfer experiment, Density gradient Units of semiconservative replication of DNA: view of Meselson and Stahl of, 321, 328–329, 334, 353–354, 362, 365, 371–372, 375, 377–379, 381– 383, 389–390; view of L. Cavalieri of, 392–393; evidence for nucleotide strands as, 398, 404–408 Vinograd, Jerome, 161: and ultracentrifuge, 171–172, 192, 359; and density gradient method, 235, 236, 254, 266– 269, 356; and Meselson, 356 Visconti-Delbru¨ck theory of phage reproduction, 94–95 Vogt, Marguerite, 399 Wake, R. G.: on replication of DNA and Meselson-Stahl experiment, 396–397, 398 Walch, Carolyn (Slayman): and Meselson-Stahl exp., 388 Watson, James Dewey, 6: and double helix, ix, 1–3, 12, 22–23, 27–28, 30–32; on replication and genetic implica-

503

tions of DNA, 11–29, 37, 46–47, 412– 413; and Delbru¨ck, 12–21, 32, 34–35; and alternative structures for DNA, 26–27, 77–78; at Caltech, 32–40; and structure of RNA, 32–36, 38, 108– 115; and Cavendish laboratory, 34, 113; as instructor at Woods Hole, 39; 60, 61, 62, 67; and Hershey-Chase exp., 55, 58–60; and Meselson, 55, 212; appointment of to Harvard, 76– 77; scheme of mutagenesis of, 120– 121, 442; and Meselson-Stahl exp., 364, 387, 392–393, 413–416, 430; photo, 47 Watson-Crick model of DNA: schematic representation of replication of, 383, 413, 419, 420, 426; MeselsonStahl exp.and, 327, 353, 375, 378, 381–384, 385, 389; demonstration of replication according to, 404–408. See also Double helix Weigle, Jean: induction of phage mutations, 35; and Meselson, 72, 439; and phage , 304, 346, 402–404 White, Noel: on structure of diamino chloropyrimidine, 145 Wilkins, Maurice: on helical structure of DNA, 16, 23–24, 28, 31, 49; and J. D. Watson, 12, 40; and MeselsonStahl exp., 387 Wollman, E´lie: on bacterial mating, 92 Woods, Phillip S.: collaborator of J. H. Taylor, 273 Wyatt, Gerard: and ratios of bases in DNA, 13–14, 17–18, 24, 32; and 5hydroxymethylcytosine, 238 Zamecnik, Paul, 352 Zamenhof, Stephen: and incorporation of 5-bromouracil in bacterial DNA, 137–138, 169