175 21 38MB
English Pages 352 [516] Year 1991
The Bone-Biomaterial Interface
To Lyndsey
Edited by J.E. Davies
The BoneBiomaterial Interface
UNIVERSITY OF TORONTO PRESS Toronto Buffalo London
www.utppublishing.com University of Toronto Press 1991 Toronto Buffalo London Printed in the USA ISBN 0-8020-5941-4
Printed on acid-free paper
Canadian Cataloguing in Publication Data Main entry under title: The Bone-biomaterial interface Based on the proceedings of the Bone-Biomaterial Interface Workshop held in Toronto, Ont., Dec. 3-4, 1990. Includes index. ISBN 0-8020-5941-4 1. Orthopedic implants - Materials - Congresses. 2. Orthopedic implants - Biocompatibility Congresses. 3. Biomedical materials - Congresses. I. Davies, I.E. (John Edward), 1948- . II. Bone-Biomaterial Interface Workshop (1990 : Toronto, Ont.). RD755.5.B65 1991
617.4710592
C91-095118-7
Cover Illustration: Interface between woven bone (left) and bioactive glass (right) formed in vitro. New bone has been elaborated by osteoblasts which have migrated from a rat parietal bone (bottom). This scanning electron photomicrograph is of a resin-embedded tissue block. The block was plasma cleaned before gold coating (from work by the editor and Dr Tetsuaki Matsuda). The superimposed interface graphic is from an original design by Luke Davies.
Contents
Foreword Preface
D.F. Williams
ix
xi
Acknowledgments xiii
Part 1: The Material Surface 1
Surface Characterization of Implant Materials: Biological Implications 3 D.C. Smith
2
The Biomaterial-Tissue Interface and Its Analogues in Surface Science and Technology B. Kasemo and J. Lausmaa
3
Surface Reaction Kinetics and Adsorption of Biological Moieties: A Mechanistic Approach to Tissue Attachment 33 L.L. Hench
4 Titanium and Its Oxide Film: A Substrate for Formation of Apatite T. Hanawa
19
49
5
Titanium: Immersion-Induced Surface Chemistry Changes and the Relationship to Passive Dissolution and Bioactivity 62 P. Ducheyne and K. Healy
6
Kinetics of Mineralization, Demineralization, and Transformation of Calcium Phosphates at Mineral and Protein Surfaces 68 M.S-A. Johnsson, E. Paschalis, and G.H. Nancollas
7
Substrate Surface Dissolution and Interfacial Biological Mineralization R.Z. LeGeros, I. Orly, M. Gregoire, and G. Daculsi
8
High-Resolution Electron Microscopy of a Bone Implant Interface W. Bonfield and Z.B. Luklinska
76
89
Part 2: Bone Proteins and Other Macromolecules 9
Non-Collagenous Bone Proteins and Their Role in Substrate-Induced Bioactivity 97 J. Sodek, Q. Zhang, H.A. Goldberg, C. Domenicucci, S. Kasugai, J.L. Wrana, H. Shapiro, and J. Chen
vi Contents 10 Role of Adhesive Proteins and Integrins in Bone and Ligament Cell Behavior at the Material Surface 111 J.J. Sauk, C.L. Van Kampen, and M.J. Somerman 11 Non-Endocrine Regulation of Bone Cell Activity 120 H.C. Tenenbaum, C.A.G. McCulloch, H.F. Limeback, and P. Birek 12 Osteogenesis Induced by BMP-Coated Biomaterials: Biochemical Principles of Bone Reconstruction in Dentistry 127 Y. Kuboki, H. Yamaguchi, A. Yokoyama, M. Murata, H. Takita, M. Tazaki, M. Mizuno, T. Hasegawa, S. lida, K. Shigenobu, R. Fujisawa, M. Kawamura, T. Atsuta, A. Matsumoto, H. Kato, H.-Y. Zhou, I. Ono, N. Takeshita, and N. Nagai 13 Ceramic Synthesis using Biological Processes 139 B.J. Tarasevich, P.C. Rieke, and G.L. McVay Parts 1 and 2 - General Discussion
147
Part 3: Cellular Activity at the Interface 14 Inflammatory Cell Response to Bone Implant Surfaces P. Thomsen and L.E. Ericson
153
15 Modulation of Cell Activity by Titanium Peroxy Compounds 165 L.M. Bjursten and P. Tengvall 16 Behaviour of Osteoblasts on Micromachined Surfaces D.M. Brunette, J. Ratkay, and B. Chehroudi
170
17 Osteoblast Reactions to Charged Polymers 181 R.M. Shelton and J.E. Davies 18 Cell-Mediated Bone Regeneration A.I. Caplan
199
19 The Influence of Sputtered Bone Substitutes on Cell Growth and Phenotypic Expression 205 A.S. Windeler, L. Bonewald, A.G. Khare, B. Boyan, and G.R. Mundy 20 Early Extracellular Matrix Synthesis by Bone Cells 214 J.E. Davies, P. Ottensmeyer, X. Shen, M. Hashimoto, and S.A.F. Peel 21 Transmission Electron Microscopical Identification of Extracellular Matrix Components using Immunocytochemistry 229 H. Magloire, M. Bouvier, P. Exbrayat, M.B. Andujar, M.L. Couble, A. Joffre, H. Poly, M .H. Veron, D. Seux, and D.J. Hartmann 22 Molecular Biological Approaches to Investigate Cell/Biomaterial Interactions S.R. Goldring and J.-T. Wang
241
23 Biological Cascades of Fracture Healing as Models for Bone-Biomaterial Interfacial Reactions 250 S. Jingushi and M.E. Bolander
Contents vii Part 4: The Tissue-Material Interface 24 Tissue Responses to Bone-Derived and Synthetic Materials J. Glowacki and M. Spector
265
25 Hard and Soft Connective Tissue Growth and Repair in Response to Charged Surfaces M. Krukowski, B. Eppley, T. Mustoe, and P. Osdoby
275
26 Deposition of Cement-like Matrix on Implant Materials 285 J.E. Davies, N. Nagai, N. Takeshita, and D.C. Smith 27
Polymer Reactions Resulting in Bone Bonding: A Review of the Biocompatibility of Polyactive 295 C.A. van Blitterswijk, S.C. Hesseling, J. van den Brink, H. Leenders, and D. Bakker
28 Comparative Morphology of the Bone Interface with Glass Ceramics, Hydroxyapatite, and Natural Coral 308 U.M. Gross, C. Muller-Mai, and C. Voigt 29
Interfacial Reactions to Bioactive and Non-bioactive Bone Cements H. Oonishi
321
30 Modulation of Bone Ingrowth by Surface Chemistry and Roughness J.L. Ricci, J.M. Spivak, N.C. Blumenthal, and H. Alexander
334
31
Comparative Push-out Data of Bioactive and Non-bioactive Materials of Similar Rugosity M. Niki, G. Ito, T. Matsuda, and M. Ogino
350
32
Quantified Bone Tissue Reactions to Various Metallic Materials with Reference to the So-called Osseointegration Concept 357 T. Albrektsson and C. Johansson
Part 5: Mechanical Effects on Interfacial Biology 33
Effect of Mechanical Stress on Tissue Differentiation in the Bony Implant Bed D.R. Carter and NJ. Giori
367
34
Quantitative Evaluation of the Effect of Movement at a Porous Coated Implant-Bone Interface R.M. Pilliar
35
Bone Ingrowth into Porous Coatings Attached to Prostheses of Differing Stiffness D.R. Sumner, T.M. Turner, R.M. Urban, and J.O. Galante
36
Influence of Biomechanical Factors at the Bone-Biomaterial Interface J.B. Brunski
37
Bone Bonding Behavior of Biomaterials with Different Surface Characteristics under Load-Bearing Conditions 406 T. Yamamuro and H. Takagi
380
388
391
Part 5 - General Discussion: The Effect of Micromotion on Bone Healing
415
viii Contents
Part 6: Retrieval Analysis for Interpreting Interfacial Phenomena 38 Bone-Biomaterial Interfaces of Retrieved Implants 419 J.E. Lemons 39 Ultrastructural Investigation and Analysis of the Interface of Retrieved Metal Implants 425 L.E. Ericson, B.R. Johansson, A. Rosengren, L. Sennerby, and P. Thomsen 40
Synovial Cells at the Interface with Retrieved Implants 438 P.A. Revell and P.A. Lalor
41
Phenotypic Characteristics of Inflammatory Cells Derived from Hip Revision Capsules N.A. Athanasou, J.T. Triffitt, C.J.K. Bulstrode, and J. Quinn
42
Bone Bonding to Retrieved Hydroxyapatite-Coated Human Hip Prostheses 450 D.C.R. Hardy, P. Frayssinet, I. Primout, E. Yasik, M.A. Lafontaine, and P.E. Delince
444
Part 6 - General Discussion 457
Part 7: The Industrial Perception 461 Session chairman: Sydney Pugh Panellists: Richard Kenley, Gerald Niznick, Makato Ogino, Jack Taylor, Jack Parr, John Cresser Brown, Mike Hyjek, Craig van Kampen, Ann Burgess, Jim Benedict, Mark Skarsted, Eyal Ron Biographical Sketches of Invited Attendees at the Bone-Biomaterial Interface Workshop, Toronto, December 3 and 4, 1990 471 Affiliations of Contributing Authors 481 Index
487
Foreword
One of the central challenges in biomaterials science is the mechanistic explanation of the establishment of an interface between tissue and implant materials. Since there are so many different materials currently in use, and since their location in the body involves several types of tissue, it is not surprising that this problem is difficult to resolve. While, from a materials science point of view, both the theoretical and experimental approaches to elucidating interfacial structure and behaviour may seem self-evident, the problem is by no means trivial when biological milieux are involved. One major drawback is that very few biological experimental procedures have been established specifically to investigate these interfacial reactions with a resolution comparable to those of current surface analysis techniques. Moreover, the "biological" side of the interface is far more dynamic than the material side. These difficulties are found whatever type of tissue location is involved, although there will be both major and subtle variations in different tissues. The situation with bone is particularly interesting, since mechanical effects take on a significant role and since there are so many procedures that involve the establishment of a bone-biomaterial interface. Current statistics indicate that approximately half a million artificial hips and knees are implanted annually worldwide, while single dental root implants far exceed this number. The surgical success with orthopaedic devices in older patients during the past decade has encouraged the use of prostheses in younger patients, and it can reliably be anticipated that numbers of prostheses will increase even further in the future. Historically the selection of biomaterials has been based on the concept of inertness: that is, the belief that the best performance will be obtained if materials are chosen that do not interact with the tissues of the body. Hence the use of the so-called corrosion-resistant metals and bio-inert ceramics.
The inability to determine whether this has been the best approach has largely been due to the lack of understanding of the interface. It is of the utmost importance that we be able to define the nature of the interface and the role of the many mediators of its development if we want to critically assess the performance of the "inert" materials and, even more, if we want to explore, seriously, the possibilities of controlling the interfacial characteristics through active rather than inert materials. The various chapters of this book address the questions which have arisen during this period of evolution from the inert to active materials with specific reference to the bone-biomaterial interface. The nature of the interface here provides a different dimension to the problems encountered in other implantation situations, since the performance of devices in contact with bone is mainly dependent on the bond that is achieved at the interface. Failure to achieve a mechanically robust "bond," and to maintain this bond as the implant functions over months and years, is the main cause of failure in orthopaedic, dental, and maxillofacial implant surgery. A rather confusing terminology reflects the uncertainty over the mechanisms that are involved. The materials tend now to be divided into the socalled bonding and non-bonding categories, although the term "bond" should not necessarily imply a molecular interaction. The term "bioactive" has a general meaning, referring to the ability of a material to actively influence the surrounding tissue and the development of the response from that tissue, and in the context of the bone-biomaterial interface it has become associated with the ability of a material to facilitate the development of a bond to bone. Classically, bone bonding with a bioactive material results in the interdigitation of collagen fibres, produced by bone cells, with the material surface. The mechanisms are unclear and especially we have failed so far to
x Foreword understand what is happening, and the significance of what is happening, at the ultrastructural level. The role of collagen interdigitation is crucial here, since if such interdigitation is a requirement for bioactive materials and bone bonding, these terms should not logically be used to describe the situation with metals (and their oxide surfaces) or inert oxide ceramics. With materials such as titanium, the interface has been considered, until recently, to comprise a non-collagenous connective tissue ground substance. However, as demonstrated herein, in vitro experiments have now revealed a thin non-collagenous, calcified, extracellular matrix layer directly apposed to the implant. Our knowledge and understanding of the interface is not only increasing but changing; not all calcium phosphates are bioactive, while it may be possible for metal oxides and ceramics to display the characteristics of bone bonding. These difficulties of interpretation and the confusion over the mechanisms provide the essential rationale for the present volume. The organization of the chapters reveals the different aspects of the problem, since they deal sequentially with the material substrate, the bone-specific proteins, the cellsubstrate interactions, the gross tissue-substrate interactions, and the mechanical influences on the developing response. These are supplemented with
a final section on the information that can be gained through the retrieval of bone implants. No one can pretend that this volume will have solved all of the problems or indeed resolved the majority of the uncertainties. What it has done, however, through the collection and integration of contributions from leading laboratories concerned with the evaluation of the bone-biomaterial interface, is to provide an exposition of the current thinking about these problems and an indication of how mechanisms are gradually being elucidated. It is of course bone-cell-substratum interactions that have provided the focus of John Davies's own research during the past decade. His multidisciplinary background and research experience have been of immense value in allowing him to convene the Bone-Biomaterial Interface Workshop upon which this book is based. The edited texts and discussions provide an excellent, informative, and fascinating account of the position taken by the contributing authors in 1990; it will be interesting to see how these positions reflect state-of-the-art opinion in ten or even two years' time. D.F. Williams Institute of Medical and Dental Bioengineering University of Liverpool May 1991
Preface
When young I did eagerly frequent Doctor and saint and Heard great argument about it and about But evermore came out the same door as in I went. With these words of Omar Khayyam ringing clearly in one's ears and, doubtless, them being as true in the twentieth century as when they were written in the eleventh, it may seem foolhardy to want to bring together a group of specialists to discuss anything. This is especially true in a relatively new field of multidisciplinary endeavour, such as biomaterials. Nevertheless, as a result of clinical and commercial successes, this embryonic discipline has been catapulted into a social prominence that ignores our lack of understanding of fundamental reactions which take place when artificial materials are introduced into biological milieux. This problem is further compounded when one realizes that the explosion in bone implant usage is only a reflection of what is happening in the biomaterials field in general. Thus, while one could be forgiven for thinking that the field is already sufficiently burgeoned with international conferences which enable interested individuals to keep abreast of current thinking, many are becoming too large to permit in-depth discussions of our knowledge, or ignorance, of individual fields. With this rather negative view of the current state of affairs, and despite the ringing in my ears, I sought to bring together a multidisciplinary group of active research scientists to focus on the complexities of the bone-biomaterial interface. The focus on materials implanted in bone was important not only because it reflects my own research interests but also because a revolution is occurring in our understanding of bone tissue biology. Thus, bone biology has evolved from being an addendum to nineteenth-century histology into an exciting field that has benefited enormously from advances in cell and molecular biology. Cell lineages and differentiation pathways are now reasonably clearly defined while genes which regulate extracellular matrix component expression are being sequenced and Pharmaceuticals produced from recombinant
proteins. The merging of this high-tech biology with biomaterials and implantology undoubtedly has the potential to create a new revolution in these areas of health care. To understand the mechanisms of interfacial reactions between artifical materials and bone one must also strive to understand how the surfaces of materials will influence bone cell and tissue behaviour. Thus, a balance must be struck between several disciplines, each with its own champions, and this I attempted in convening the Bone-Biomaterial Interface Workshop in December 1990. Contrary to the trend of the majority of scientific conferences, the purpose was to stimulate in-depth discussion, and thus each participant was provided with the following premises upon which the workshop was based: • Little is understood of the differences in biological reactions to bonding and non-bonding materials. • Not all calcium phosphate-based ceramics generate bone bonding to their surfaces. • In spite of considerable emphasis being attached, in the literature, to the importance of surface analytical techniques in characterizing implant materials, no evidence has emerged to link surface analysis at the atomic or molecular level with bone tissue reaction. • Little is known concerning the ionic or macromolecular reactions which are generated at the solidliquid interface on implantation of a bone-substitute material. • Little is known concerning the changes in the mechanisms of normal bone healing which may be brought about by the presence of an implant material. • Little is understood concerning the reactions of bone cells to implant surfaces despite the obvious facts that bone can only be made by one cell type, the osteoblast, and that the colonization of implants by these cells and their differentiation
xii Preface underpin the phenomena of osteoconduction and osseointegration. Provided with this, one might hope, provocative start, the invited participants were asked to submit a paper in advance, which would be distributed to, and perhaps read by, the other participants before the workshop took place. Only one invitee, Tomas Albrektsson, was unable to attend, because of illness, but fortunately his paper had already been circulated, and it was therefore possible to include his contribution. The meeting itself, the format of which was inspired by the early Tooth Enamel conferences organized by my mentor Professor Ron Fearnhead, then provided only a few minutes for each author to summarize the central points of his or her paper which could be critically discussed by the assembly. The participants were seated about a table, summit-meeting style, with a gallery of about one hundred observers. Two video cameras filmed the entire meeting and a team of court stenographers produced a verbatim documentation. To enable such a format to work effectively the number of invited active participants had to be restricted. The final number was forty, and these were joined by six industry representatives to ensure that the pragmatic aspects of biomaterials research and development were not overlooked. The meeting was organized into seven sessions, dealing with materials issues, protein adsorption, cell and tissue reactions, mechanical influences on interfacial biology, retrieval analysis, and the industrial context, which are reflected in the seven parts of this book. In light of the workshop discussions the authors were given the opportunity to
revise their manuscripts before the editing process was begun. Some comments concerning my editorial procedures may be helpful to the reader at this point. The two days of almost continuous discussion generated over 600 pages of transcript. This was condensed and, where relevant, attached to the appropriate chapter. One additional chapter has been included (chapter 26) as a result of discussion of new material which was presented at the workshop. As some reordering has occurred in the book chapters, relative to the order in which the papers were discussed at the workshop, generalized discussions which evolved have been appended to the relevant parts of the book and this has required some radical restructuring of the transcripts. In all cases I have strived to maintain the sense of the discourse during this restructuring and the attendant transition from a spoken to written dialogue. Even more radical rearrangement has occurred in the final section, which is explained in the introduction to Part 7: The Industrial Perception. Throughout the book I have endeavoured to provide cross-references and also interject editorial comment where appropriate. In many cases only the authors themselves will realize that this has been done since my aim was not to disrupt the flow of a chapter but rather to provide links between chapters and the disciplines this work represents. I hope that the format, as it has emerged, will provide more interesting reading than conventional conference proceedings or multi-authored texts and that the content of both the chapters and the discussions will open new doors for those who eagerly explore the field of biomaterials.
Acknowledgments
During the Society for Biomaterials annual meeting in Charleston, in May 1990, I asked some colleagues if they would support the idea of an international conference focused specifically on the bonebiomaterial interface. They were all, unhesitatingly, enthusiastic. Encouraged by this response I undertook to convene what was subsequently called the "First International State of the Art Workshop on the Bone-Biomaterial Interface," and the first letters of invitation were mailed. The date for the workshop was set for the beginning of December and thus, at least on paper, the workshop was scheduled to go ahead. Then came the tricky part: in exactly six months the workshop had to be designed; the faculty assembled; the necessary funds raised; pre-workshop papers solicited, received, and distributed; and the locale chosen. I rapidly came to understand why international conferences are organized by committees rather than individuals. Thus, I have to bear the responsibility of having driven people insane by either imposing improbable deadlines or coercing financial and professional commitments from them before sufficient funds had been raised to transform the idea into a reality. But it is those who had faith in the idea, who changed their schedules, who wrote papers in record time, who undertook contracts before money was available to pay for them, and who started the ball of financial support rolling, who deserve the credit for this venture. It is they whom I should like to acknowledge, both for their considerable efforts and for the success attached to the Bone-Biomaterial Interface Workshop. I initiated this project in the knowledge that I could count on my research group not only in the myriad tasks that lay ahead, but also in creating an atmosphere where frenzied activity would not translate into frazzled nerves. I should like to acknowledge particularly the constant help and support of Yvonne Bovell for, without the commitment, energy, and long, late hours which she devoted to the
project, our deadlines could not have been met. My group became my impromptu "committee" but concomitantly kept our laboratories active. I should especially like to thank my technicians and research fellows Bob Chernecky, Beate Lowenberg, Ed Roberts, AugustinRodrigues, Xue-ying Shen, Amy Shiga, Melissa Troemel, and Nancy Valiquette; my students Gail Anderson, Janice Gladstone, Zhou Hong, and Sean Peel, and our visitors within the group from other universities who joined, and contributed to the team, namely Joost de Bruijn, Masaki Hashimoto, and Mikinori Ogura. As I write these acknowledgments, two organizations that were central to the success of the workshop have undergone dramatic change. It is a sad reflection on this period of economic recession (in Canada) that the Holloway Group has been disbanded. Meanwhile the Technology Institute for Medical Devices for Canada (TIMEC) has undergone complete restructuring. Lynette Holloway and Darlene Stewart from the former and David Lartigue from the latter were instrumental in ensuring the smooth management of the workshop project. I thank them for the enormous support which they gave me and hope that they will each find continuing opportunities to play similar key roles for which they are so well suited. As explained elsewhere, papers were submitted in advance and the authors were given the opportunity to modify their manuscripts in light of the workshop discussions. This also provided an opportunity for editorial change at the revision stage. Thus, it was that Kate Troemel, copy-editor and proof-reader, attended the workshop, initiated discussion with the authors, and laid the groundwork for production of the copy for this book. However, the final structure of the book was still being discussed months after the meeting in response to the copy with which we had to deal. It was thus a daunting task and also an uncommon, and difficult, burden for a copy-editor to bear. To make matters
xiv Acknowledgments worse, I suggested that we should not only endeavour to address scientific and copy-editing issues but also strive to create at least a semblance of stylistic continuity in this multi-author and multi-disciplinary book. Kate Troemel carried out these tasks with considerable energy and humour. She was a source of thoughtful and considered advice which, in an unobtrusive way, kept the project skimming over the crests of the waves in the dangerous seas that confront first-time editors such as myself. I have greatly appreciated, and relied upon, her professional support for and personal commitment to this project and I am grateful that she was part of our team. She also provided an essential link between myself and the University of Toronto Press. At the Press, Lorraine Ourom, executive editor, meticulously moulded our copy into book format. Her tireless and expert support was invaluable, especially in view of the schedule to which we wished to adhere, and the depth of her knowledge lent clarity to the most confusing issues. Of course, the whole project had to be funded. Not only was the financial support forthcoming
from both institutions and commercial organizations an essential ingredient of a successful meeting but the speed with which the targets were met was a source of continual encouragement to me during the short planning stage. By listing these organizations below I would like to reiterate my thanks, and those of the participants, for this essential support and, more important, the vision of those listed below in wanting to be part of such a new venture. My colleagues Mike Lee, Bob Pilliar, Dennis Smith, and Rana Sodhi provide a stimulating and thoroughly enjoyable working environment within the Centre for Biomaterials at the University of Toronto. Their support has been pivotal in my undertaking of this project and the atmosphere they generate has continually encouraged my meagre efforts. Finally, I am indebted to Lyndsey, my wife, to whom this book is dedicated, for her support and friendship, and to Luke, Charlotte, and Megan for the constant reminder that there are also other things in life which are as exciting as Bone-Biomaterial Interfaces.
COMPANIES AND INSTITUTIONS WHICH PROVIDED FINANCIAL SUPPORT FOR THIS PROJECT Apatite International Bio-Research Laboratories Biomet Boehringer Mannheim Calcitek CAM Implants Centre for Biomaterials, University of Toronto Ciba-Geigy Canada Core-Vent DePuy DSM Howmedica Inznova Intermedics Orthopaedics Johnson & Johnson Natural Science and Engineering Research Council of Canada (NSERC)
Nikon Ontario Centre for Materials Research (OCMR) Ontario Ministry of Industry, Trade and Technology Ontario Ministry of Intergovernmental Affairs Ontario Technology Fund of the Government of Ontario Osteonics Q-life Systems Smith & Nephew Richards Sulzermedica Synthes (Canada) Technology Institute for Medical Devices for Canada (TIMEC) 3M/Orthopaedic Products Division Zimmer
Part 1 The Material Surface
This page intentionally left blank
1
Surface Characterization of Implant Materials: Biological Implications D.C. Smith
A wide variety of metals, ceramics, and polymers are used for bone implants in many different applications including joint replacement, fracture fixation, and tissue augmentation subsequent to tumour surgery [ 1 ]. The bone-biomaterial interaction may be significantly different for all these materials. For load-bearing implants the superior fracture and fatigue resistance has made metals the materials of choice up to the present [2]. Development is proceeding on polymers, ceramics, composites, and biologically derived materials such as collagen, but suitable alternatives for general use have yet to be introduced [2J. Thus, this paper will focus on current materials in the "state of the art" and particularly on metals for use in two load-bearing types of implant-joint arthroplasty components and dental endosseous implants. More than 500.000 total hip and knee replacements are performed yearly - 150,000 in North America alone. The advent of modern technology (improved alloys, prosthesis designs, and improved fixation techniques) has led to an expansion of the indication for the procedures such that they are now being applied to younger and younger populations. Since the demographics of most Western countries indicate also an aging of the population over the next two decades with an increasing number of survivors into old age, the number of total joint replacements is expected to rise dramatically [3J. Successful functional outcomes can be predicted in a majority of these cases following arthroplasty but, in spite of improvements in materials and surgical techniques, complications do still occur related to late aseptic loosening of joint components, late infections, mechanical deficiencies leading to wear products, and adverse responses to wear debris and released metal ions. The established method of joint-implant fixation involves acrylic bone cement. In recent years "cementless" fixation involving bone ingrowth into a
porous implant surface also has become widely used. This concept of "biological" fixation has been deemed superior to the traditional fixation with bone cement since there is no longer the risk of cement fatigue failure or loosening as the result of senile osteoporosis in which the endothelial aspect of the femur is eroded. As a result of the developments of the past few years [4, 5] a variety of endosseous dental-implant systems have become available with widely different designs, surface textures, and materials of construction. A few systems have received official sanction for clinical use but, in general, there is little scientific evidence for the safety and efficacy of the majority of systems available. Claims for osseointegration, fibro-osteal integration, bone bonding, and bony ankylosis abound, but there is little precise knowledge of the actual interface between implant and tissue and of the factors which influence host response and the long-term integrity of the implant system. More fundamental research is needed on both materials and design for rational progress [6]. It is well known, as discussed by Ratner [7], that, in general, the surface chemistry, surface energy, and surface topography govern the biological response to an implanted material. Such a response may involve physical factors such as size, shape, surface texture, and relative interfacial movement, as well as chemical factors associated with the composition and surface structure [8-12], However, the details of these interactions are still imperfectly understood. In addition, evidence has continued to accumulate that implanted metallic, ceramic, and polymeric materials degrade in the body, releasing constituent ions or components that may lead to an adverse response [13]. Again, the extent and mechanisms of these effects are uncertain. Cellular response to implant materials may be affected also by adsorbed surface species that affect
4 The Material Surface the surface composition and charge [14-17]. These adsorbed species may be, initially, contaminant films arising from preparative procedures that may result in low-energy surfaces and adverse responses [17, 18]. Such films may be removed at least in part by cleaning procedures such as gas plasma (glow discharge) treatment or UV/ozone treatment [17, 18]. Surface contamination has not received great attention until quite recently in implantology research in spite of the importance to protein binding, thrombogenecity, and cell attachment [19]. The majority of studies on implants have not characterized, in a precise manner, the surface and bulk characteristics of the materials being used. Furthermore, the cellular effects of the particular cleaning and sterilization systems have received little detailed investigation [17, 18, 20]. For example, plasma cleaning of dental implants for a low-energy surface has been recommended [17, 18], but whereas rapid bone opposition using an argon plasma was reported by Hartman et al. [21] Carlsson et al. [22] found poor results using an air plasma. Brunette [23] has illustrated the pronounced effects of dental-implant surface topography on orientation and migration. Thus, it appears essential in future studies to fully characterize the surface of materials physically and chemically before in vitro and in vivo evaluation. Compositional Considerations Characterization of implant materials to evaluate tissue reaction necessarily includes analysis of surface composition and contamination [9-12]. The surface composition of most implant materials is substantially different from the bulk, and a variety of techniques have been used to characterize such surfaces [7]. Since no one technique provides complete information, several such techniques are needed to fully characterize implant surfaces. Such a combined approach is only just beginning to be used in biomaterials. The close apposition of bone to titanium, for example, has been ascribed to the characteristics of the oxide layer [15, 21] which may vary between cpTi and Ti-6AMV. Many studies [14, 15, 2428] have applied X-ray photo electron spectroscopy (XPS) and Auger analysis to the surface characterization of titanium and stainless steel (dental) implants and found, in common with our own work [29-34], that topographical and compositional changes occur in the surface of such implants as a
result of both preparative treatments and in vivo implantation. Geis-Gerstorfer and Weber [35] observed significant differences in the composition and corrosion behaviour of flame-sprayed titanium coatings from different manufacturers due to contamination. Klauber et al. [36] found considerable variation in surface contamination on assupplied dental implants with some showing high levels of silicates in the oxide layer. The in vivo significance of such variations has yet to be determined. Reported data do suggest that active surfacedissolution processes occur in vivo. Thus, McQueen et al. [24] observed an increase in the thickness of the oxide layer on titanium with time in vivo. Similarly Luthy and Strub [37] have observed in vivo (dogs) a decrease in the thickness of plasma flame-sprayed titanium coatings on titanium implants on retrieval after 36 months. We also observed [38] extensive changes in the surface composition and topography of nickel chromium alloys after implantation for at least 3 months in animal models. An additional factor in this context is the release of paniculate material or wear debris at the tissue interface with a consequent enhanced propensity to dissolution and ionic release and diffusion of material away from the implant site. Fracture and migration of sintered-bead porous surfaces has been observed [39], and titanium particles have been demonstrated in tissues adjacent to titanium implants with plasma flame-sprayed coatings [40]. Considerable concern has been expressed recently [41] over the possible relationship between wear debris and osteolysis. Few studies have provided long-term (years) data for implant ion release and its local and remote cellular effects [10, 12] in relation to the surface and/or bulk composition of the implant. In vivo experiments have demonstrated that metal-ion release can affect cell function [42], and in vitro assays have monitored the influence of increased metal-ion concentration on both the proliferation of cells [43] and the synthesis of extracellular matrix [44]. Inhibition of apatite formation by Ti and V ions has been demonstrated by Blumenthal and Cosma [45] confirming an earlier study by Gerber and Perren [46], which raises the question that metal-ion release from such implants could compromise biological mineralization at the implant interface. Recent clinical findings associated with wear debris of pain and tissue breakdown [47, 48] in the orthopaedic joint prostheses suggest that the metal ion release issue may be important to long-term biological response.
Smith: Surface Characterization of Implant Materials 5 Metal Ion Release Considerations The importance of metal ions to biocompatibility has been reviewed recently by Michel [10]. He points out that the applicable quality standards for implants (for example, ISO) do not control impurities in the less than 0.1% range, which nevertheless may represent a differential distribution concentrated in the surface. Michel states that current implant materials may contain more than 35 metals, most of which are present in the human body only in trace (10 to 10 g/g) concentrations. These trace elements may be essential, toxic, or incidental. At present, the following 14 trace elements are recognized as essential: V, Si, B, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Mo, I; and five more are possibly essential: Li, Cd, Sn, W, and Pb. For these elements there is a dose dependence outside the normal range (maintained by homoeostasis), with pathology related to both deficiency and excess (toxicity). For some implant-relevant elements - including Ti, Zr, and Al - there are no known physiological functions although toxicity is possible, for example, in the effect of aluminum on bone metabolism and its potential in Alzheimer disease. In addition to toxicity, oncogenic/mutagenic and hypersensitivity effects must also be considered. According to recent reviews, at least ten metals, including Co, Cr, Ni, Pb and Ti, can be considered as chemical carcinogens with another ten possible. Among the important allergens are Ni and Co. One of the most common hypersensitive responses is Ni allergy. Recent U.S. data suggest that as high as 20% incidence in women and a 4% to 5% incidence among men could be detected in prosthodontic patients. Nickel has also been involved in reports of tumour induction to implant alloys, the induction time being greater than 10 years. Michel cites 18 elements present in implants whose effects on biological action have to be considered. Unfortunately, the status of our knowledge, both of analytical techniques and of normal concentrations, is inadequate. Modern techniques of trace element analysis of local and remote tissues (relative to the implant) seek to provide evidence for an association between tissue levels and "adverse" effects. However, these attempts are complicated by two factors: (1) much of the analytical data in the literature was acquired with poor analytical techniques, (2) recent changes have been introduced in the definition of "normal" values in various tissues. With respect to the first point, only
a few studies have accurately controlled sample collection to avoid contamination and have validated the data by proper quality control of the analytical procedures. With respect to the second point, recent studies call into question previous normal values of trace elements in human tissue and body fluids, for example, chromium [49, 50]. The elemental analysis of biological materials has been considered in detail by lyengar [49]. He details the problems and methods involved in sample collection, tissue analysis, acquirement of reference (normal) values, and data interpretation. Because of environmental concerns trace element studies have developed rapidly in recent years, and the need for rigour in all stages of the analytical process has become apparent [51]. Routine reference to Standard Materials and collaborative multilaboratory testing have contributed to standardization of techniques and reduction of errors, as in the procedures published by the German Science Foundation [52]. As a result the sensitivity and precision of analytical techniques have improved profoundly compared with even a decade ago. Clinical indications for trace element analyses are discussed by Kruse-Jarres [53], who points to the difficulties in understanding the connections between trace elements and known, as well as still unrecognized, symptoms of disease. Factors such as age, diet, location, and disease may influence levels of specific elements. Analytical values may also be influenced by sample procurement from human subjects [54]. Versieck [55] in an extensive critical review and lyengar [49] have considered in detail the probable values and ranges for trace elements in human body fluids and tissues in the "normal" subject. lyengar discusses the meaning of "normal" in relation to essential and non-essential elements and suggests the concept of "reference values" which refer to a specific defined group of subjects living in distinct global regions. Versieck [55] reviews the "devastating effect on the result of trace element analysis of improper sample collecting and handling" and cites data which show that' 'normal'' levels of trace metal ions relevant to orthopaedic implants are very low (ultra-trace level), previous data having been in error by as much as an order of magnitude. We have critically reviewed these issues in relation to biomaterials. [50]. There is insufficient space to discuss these issues in more detail, but the data base is small or missing for the implant ions of interest here. The available data indicate that "normal" values may be summarized as follows:
6 The Material Surface 1 The level of Ti in serum in still doubtful but is probably less than 10 /xg/L; that for Al is less than 10 fJig/L in serum and whole blood likewise; for V less than 1 /ag/L; for Co, Cr, and Mo all less than 1 /xg/L in both serum and blood; for Ni about 1 jUg/L in both. 2 Values for these elements in urine are similar or in the range 1-5 /Ag/g of creatinine. 3 Levels in various organs such as the liver, kidney, or brain are variable but in the 1-10 fJLg/g dry weight range. Extremely careful measurements are therefore necessary to detect changes due to implantation rather than the effects of, for example, other environmental contamination. Although there is little information on the effects of the surface chemistry and topography of implant materials on the material-tissue reaction [32, 33] it is well established that metallic ion release occurs from implant materials [10, 76]. Trace metal ion concentration arising from corrosion processes at the implant-material interface may be exacerbated not only by the increased surface area arising from porous or irregular surfaces [56-60] but also by fine particulate debris arising from wear and surface breakdown [47, 48, 61, 62]. The results of both animal and human studies [10, 13, 61-64] have shown that local, remote, and systemic tissue distribution of released ions may occur. Systemic effects of released ions have been reviewed by Black [13, 65]. Several recent papers have illustrated the possible effects of such ions [66, 67] including interactions with lymphocyte surfaceantigen components [68]. Mechanisms of transport, distribution, and accumulation of such ions are at present little understood. However, evidence is accumulating of clinical sequelae subsequent to placement of implant systems. There is an increasing concern with the occurrence of wear debris in joint arthroplasties [47, 49, 62, 69], stained tissue on revision being increasingly reported. Furthermore, numerous cases of loose beads arising from sintered porous surfaces have been encountered [38, 39]. There is an increasing number of case reports of sarcomas developing in close proximity to hip prostheses [70-75]. While the incidence of these neoplasms is small, the increased use of "cementless' fixation in recent years raises the question whether such high-surface-area implants will be associated with increased tissue levels of titanium, aluminum, and vanadium and what will be the potential sequelae [13]. As another example, alumina (aluminum oxide) bioceramics have been widely used in recent years as hip joint compo-
nents, especially as bearing surfaces, because of their mechanical characteristics and (presumed) inertness. Recently, however, LewandowskaSzumiel and Komender [76] have found significantly increased levels of aluminum in animal femoral and mandibular bone 6-8 months after implantation of alumina bioceramic. A case of sarcoma associated with an aluminum oxide ceramic total hip has been reported [77]. There is only a small amount of metal-ion release data in the biomaterials literature related to Ti, Ti-6Al^V, and Co-Cr-Mo, as cited earlier. Animal studies starting from the classic work of Ferguson et al. [78, 79] have indicated that local accumulations of these ions may occur as well as storage in various organs. Some later animal studies have indicated increased Ti concentrations in lung and in spleen as well as local accumulations which tended to be higher for the Ti-6Al^V alloy as opposed to pure Ti. However, the possibility of contamination in these studies and the absence of reference materials result in some uncertainty. Studies with reliable human data are scant; however, some recent studies [59, 61] have suggested small elevations in blood, serum, urine, and tissue levels of implant ions such as Ti, Al, V, Co, Cr, Ni, and Mo, and large increases as a result of artificial-joint failure involving wear and accumulation of particulate debris [62, 80]. Black and coworkers [81] have shown release of metal ions from several types of prosthesis, and recently small elevations have been found in serum and urine in some patients with titanium alloy total hip replacements [49]. In a small prospective study Stulberg et al. [57] found no statistical difference between unoperated controls and patients having cemented or cementless knee prostheses for Cr, Co, Ti, V, and Al in urine up to 2 years after operation. However, standard deviations were high and some patients, notably one with loose beads from a porous surface, had high Cr or Al levels. Stulberg et al. [62], in an investigation of metal-backed patella failure, found massive local depositions and elevated serum and urine levels for these ions. Sunderman et al. [61], in a carefully controlled and standardized study of 28 patients with poroussurfaced hip and knee implants in place for up to 2.5 years, found slight increases in Co in serum and urine, substantially so for two patients with loose prostheses, one of whom had elevated Cr also. These workers also noted post-operative increased Ni in serum and urine for both Co-Cr-Mo and Ti-Al-V prostheses. They speculated that the latter could have been due to contamination of the operative site by surgical instruments. They noted
Smith: Surface Characterization of Implant Materials 7 also that increased urine Cr levels in three patients may have been a manifestation of the enhanced urinary excretion of Cr that occurs following acute trauma. An interesting finding was slightly elevated Co levels in pre-operative patients as compared with healthy controls, perhaps reflecting dietary effects. In this study, however, the concentrations of Ti, Al, and V in serum and urine were not addressed. Black et al. [48] have recently illustrated and discussed the phenomenon of apparent adverse responses to a metallic implant secondary to metallic corrosion and release of wear debris (metallosis). In this clinical case involving cemented total hip prostheses made of titanium-based alloy and ultra-high-molecular-weight (UHMW) polyethylene, corrosion and wear debris were associated with pain and a need for revision, even though the prosthesis was seemingly well-fixed.The authors concluded that wear debris alone can produce clinically important pain. In general then, the data base on released ions of orthopaedic or dental interest is small and incomplete. There are little or no data on storage in remote organs to accompany the limited information on adjacent tissue accumulation. Few implants have been studied for periods in excess of 10 years, and there are few morbidity data. There are no studies except that of Borckhaus et al. [82] on analysis of tumours for relevant ions. No normal (reference) ranges have been established for orthopaedic or dental-implant patients. Thus, much work remains to be done in this field, if the controversies over possible long-term effects arising from various implant modalities are to be resolved. Even a low release of, for example, Al could be important if it continues over many years. Since total-joint replacement is now being carried out in patients under 30 years of age. this factor assumes considerable clinical significance. At the biological level these and other reactions at the tissue-implant interface are poorly understood. Studies on the biocompatibility of various implant materials indicate, nevertheless, that various local, remote, and systemic reactions may occur following placement of implant materials. Such responses may be a function of both the physical (topographic) and chemical characteristics of the surface [13. 83]. Observations on Surface Characterization Recent work carried out by us provides practical evidence of the variables that occur in currently used implant materials and which hitherto have
received little attention or standardization in biological experiments with such materials. We have subjected four materials to commonly used preparative techniques and investigated the effects on other surface characteristics [32, 33]. We have also implanted specimens of these materials in a rabbit model for up to 2 years and measured the implant ion levels in various organs and evaluated the post-implantation interface [84]. The following materials were investigated: (1) Ti-6Al-4V ELI, conforming to ASTM F-136; (2) Co-Cr-Mo, conforming to ASTM F-75 (Canox); (3) single-crystal A12O3 (Kyocera, Kyoto) (AO); (4) dense hydroxyapatite (HA) (Durapatite; Sterling Winthrop Laboratories). Chemical analyses of these materials are given in Table 1.1. Discs 3.5 mm in diameter and 1.5 mm thick were fabricated by cutting from rod stock and finished by wet grinding to 600 grit on SiC paper. Similar porous-surfaced alloy discs were prepared by sintering spherical particles of 50-300 /um diameter onto the solid substrate as described elsewhere [32]. After fabrication, the specimens were washed in distilled water and air-dried. This was the "as received" condition. Groups of three specimens for each material were subjected to the following cleaning regimens: A. Retained in the "as received" condition B. Washing in an ultrasonic cleaner as follows: 1 Sonicated for 60 min in a 2% solution of Decon detergent (B.D.H. Chemicals Canada) 2 Then sonicated in deionized water for 2 min, three changes 3 Air-dried in a closed container C. As in B, followed by steam sterilization by autoclaving at 121° C for 30 min D. As in B, followed by radiation sterilization E. As in B, followed by treatment with 40% vol HNO3 at room temperature (ASTM F-86) for 60 min in the sonicator, followed by similar rinsing with deionized water for 3 min, three changes F. As in B, followed by exposure to an argon plasma for 30 min at 0.3 Torr (Harrick Model PDC-3XG Plasma Cleaner) The materials were coated with Au for SEM and carbon for energy dispersive X-ray analysis (EDX) in a Polaron (E5100) sputter coater. For comparison, specimens of titanium and alumina commercial dental and orthopaedic implants were also scanned. Advancing water contact angles were determined using a sessile drop method. Surface analyses were carried out on duplicate samples using ion scatter-
8 The Material Surface Table 1.1 Composition of implant materials Material
Composition
Ti-Al-V
Ti Bal. Trace: H, Y
A15.96
V4.04
Fe 0.184
C0.18;N0.1
Co-Cr-Mo
Co Bal. Fe 0.54
Cr 27.77; Ni 0.08
Mo 5.38
Mn 0.79
Si 0.72
A1203
A12O3 99.9
HA
Ca, P, O, H, Bal. Sr 0.05 Trace: B, Si, Mn, Fe, Mg, Al, Cu
ing spectroscopy (ISS), secondary ion mass spectrometry (SIMS), and XPS. ISS and SIMS measurements were made using a 3M-Kratos Model 555 BX combined ISS/SIMS spectrometer, with the assistance of Dr. G.R. Sparrow (Advanced R and D, St. Paul, Minn.). A surface area of about 6 mm2 was examined for SIMS and about 2 mm2 for ISS. The positive-ion beams used were 3He or ^Ar with a beam energy of 3 keV or, occasionally, 4 keV. At 2 keV the beam current was typically 3-500 mA/cm2 over the irradiated area. XPS spectra were obtained using a Surface Science Labs SSX-100 spectrometer using monochromatized Al Ka radiation at a pressure of 10 -10 Torr; a small flood gun compensation was applied until satisfactory charge compensation was obtained. More detailed information on materials and methods is given elsewhere [32, 33].
(a)
Results Topography The SEM observations and the contact-angle data demonstrated that the various preparative procedures resulted in differences in surface topography and surface energy [32]. A marked change in contact angle especially for the alloys was induced by ultrasonic cleaning the "as received" surfaces with an effective detergent. There were no pronounced changes in surface roughness, but the grooves in the surface appeared cleaner in the SEM micrographs — suggesting that removal of contaminant layers and/or paniculate occurred (Fig. l.la,b). Little change in surface topography was produced by radiation sterilization after detergent cleaning, but it was surprising that significant structure could be seen after autoclaving for the Tiand Co-base alloys. This may have been due to chemical effects of the autoclave cycle. Areas of
(b) Fig. 1.1 (a) Ground surface (600 grit) of as-received sintered HA plate, (b) Surface as in (a) after ultrasonic cleaning.
contamination or reaction product were seen on the Co-base alloy, as previously reported also by Baier et al. [85]. Table 1.2 indicates a decreased wettability (higher contact angle) from process B. A substantial fused neck region was evident between the Co-base alloy spheres (Fig. 1.2) in
Smith: Surface Characterization of Implant Materials 9 Table 1.2 Water contact angles of dental implant materials Contact angle (°) Preparative method
TJ-6A1-4V
Co-Cr-Mo A12O3
HA
A B C D E F
67 ± 3 44 ± 4 59 ± 4 67 ± 16 60 ± 4 32 ±4
80 ± 7 54 ± 4 69 ± 5 46 ± 5 64 ± 4 25 ± 3
48 ± 5 43 ± 6 50 ± 12 25 ± 7 17 ± 7
40 ± 5 32 ± 8 35 ± 6 62 ± 8 35 ± 5 7±3
Fig. 1.3 Detail of bead junction in sintered porous surface of Ti-6Al-4V specimen.
Fig. 1.2 Detail of sintered bead junction in porous surface of Co-Cr-Mo implant.
contrast to the deep fissures in the same region of the porous-surfaced Ti alloy (Fig. 1.3). Thermal etching was evident on the Ti-Al-V alloy (Fig. 1.4). Thus, there are substantial topographical differences between the two porous materials. The nitric acid passivation process had pronounced effects especially on the Ti-Al-V alloy (Fig. 1.5). The process produced only slightly more wettable surfaces for the two alloys, suggesting residual-reaction products. The alumina was little affected. Plasma cleaning was clearly the most effective means of reducing water contact angle, as suggested by Baier and Meyer [ 18 ]. It was evident that
(b) Fig. 1.4 Thermal etching as a result of sintering porous surface in Ti-6Al^lV alloy: (a) unsintered (x 1,000); (b) sintered ( x 1,500).
10 The Material Surface
Fig. 1.5 BSE image of Ti-6Al-4V surface passivated with nitric acid (process E) showing prior /3-grain boundaries and a-plate colonies (x 128).
some surface erosion/etching occurs during the process, especially for the metals and HA. None of the contact angles was 0, though we have observed [86] values 200° C, some NO molecules break apart (dissociate) into one oxygen and one nitrogen atom. We can now see a possible route for how CO becomes a CO2 molecule. Because of the thermal motions of all Pt atoms the CO molecule and the N and O atoms can move around on the surface by surface diffusion, as in a two-dimensional gas or liquid. Eventually a CO molecule and an oxygen atom will collide and will then, with a certain temperature-dependent probability, react to form a CO2 molecule. When this happens the carbon-platinum bond, holding the CO molecule to the surface, weakens drastically and the CO2 molecule is "kicked out" from the surface into the exhaust gas by the thermal vibrations of the Pt atoms. (This process is called desorption.) This completes the first part of reaction 1 above. The second part, N2 formation, takes place in a similar manner. The residual N atom from the NO molecule eventually collides with another N atom produced in the same way and forms an N2 molecule. Just like the CO2 molecule, the N2 molecule does not bind to the surface sufficiently strongly to remain there - it desorbs into the exhaust gas and thereby reaction 1 is completed. New CO and NO molecules will "land" on the surface and react to form CO2 and N2 and so on. This is essentially how a car exhaust catalyst removes NO from the exhaust gases. Some parallel reactions contribute as well but we would not learn anything new in principal by inspecting them. One point is important, however. If the air/fuel ratio is not regulated carefully, there may be too much oxygen and CO will be oxidized away by O2 instead,
the consequence being that NO is not reduced. There is thus a detailed balance required between CO, NO, and O2 concentrations for the whole process to function well. How unique are Pt and Rh in making these reactions possible? The answer is - very unique! There is just no alternative available at present, although many efforts are being made to find replacements. What then is so special about Pt and Rh? If we tried other, less noble, metals like Ni, Fe, Mo, Cr, Ti, the water and oxygen molecules in the exhaust gas would oxidize the surface and prevent NO from dissociating. If, however, we tried gold as a catalyst it would be too noble - it would not bind CO and it would not dissociate NO. The functioning catalyst is obtained by a very delicate balance between counteracting reactions the catalyst must not be too reactive towards O2, H2O, and combusted hydrocarbons, neither can it be too inert towards CO and NO. These requirements at present leave only Pt and Rh as alternatives - and even then the catalyst can be poisoned, for example by lead in the fuel. How has this very detailed knowledge about the quite complex car exhaust catalysis reactions been obtained? It has required massive research efforts over 10-15 years at all levels from very basic research to pure product development. Catalysts of increasing complexity have been synthesized and tested both in simple flow reactors with simplified exhaust gases, and in real engine exhaust gases. In basic research the catalysts have been simplified; pure Pt and Rh single crystals or foils have been studied as well as small particles on various supports. A variety of analytical methods have been employed and new ones developed. Examples of advanced methods are Auger electron spectroscopy (AES) and X-ray photoelectron spectroscopy (XPS) for surface analysis; secondary ion mass spectroscopy (SIMS), infrared spectroscopy, gas chromatography, mass spectrometry, and electron microscopy are among the many other new advances. Detailed understanding has emerged from a mutual interaction between basic and applied research, and analytical and preparationoriented work, and by employing methods and theoretical knowledge from many different disciplines. The Biomaterial-Tissue Interface: Problem Formulation and Research Opportunities Why are we going into such detail about the catalytic conversion of CO and NO to CO2 and N2 in a publication ostensibly dealing with the bone-
Kasemo and Lausmaa: Analogues in Surface Science and Technology 25 biomaterial interface? The answer is: It serves as an example of a very complex, technologically important process, which after systematic research efforts now has an explicitly solid scientific basis. This is the type of development which is necessary for new and better biomaterials and implants to be invented. From this example we may learn something about how the questions formulated above can be addressed. It is obvious that to answer them requires a combination of research efforts, from clinical research via tissue-cell level research and basic biomolecule-surface interaction studies to pure bio(material) synthesis and analysis work. These very different types of studies must be performed in parallel and with frequent communication between the engaged disciplines, that is, a cross-disciplinary approach is necessary. Let us be a little more specific. Figure 2.5 is an attempt to picture a possible working scheme for the most basic type of studies. The choice of biomaterial and how its surface is prepared will influence how water interacts with the surface in vitro and in vivo. (For a review of water interaction with metal and metal oxide surfaces, see [21].) Water molecules arrive at the surface on the nanosecond time scale and may bind to the surface via the oxygen atom or by hydrogen bonding. Alternatively the water molecules may dissociate to hydroxyl molecules which then form a hydroxylated surface. The water or hydroxyl layer may be ordered or disordered. What in fact happens depends on both the actual material and its preceding preparation. The first monolayer of water (or hydroxyl) and incoiporated hydrated ions will influence the bonding and structure of the second layer of water and ions. When proteins reach the surface within a few milliseconds or so, they will interact differently depending on the bonding and structure of the water layer and on the presence of ions in the water layer [1, 6]. For example, these will Influence the bond strength, orientation, and degree of denaturation, if any, of the protein. New types of biomolecules will continuously appear at the surface and the previous water bonding/protein bonding history will influence the rate of exchange of biomolecules [6]. Eventually cells will appear. They will sense the layer(s) of biomolecules on the surface and react differently depending on its composition. For some surface layers the cells may react in a "friendly" manner, and for others the reaction may be "hostile." The above scenario is an admittedly hypothetical, but logical, working hypothesis. (A complementary interface scenario, discussing the cell
biology aspects in more detail, is presented by Thomsen and Ericson in ch. 14 herein.) We can now easily imagine how the implantation of a biomaterial device is a series of coupled events [1, 2, 22] starting with the initial preparation of the implant, whose initial surface properties may influence the later interface evolution as a ' 'memory effect" by determining the nature of the water layer, which in turn determines the protein-surface and cell-surface interactions, which eventually govern the ultimate success or failure of the implant. A research program based on the preceding discussion and on Figs. 2.2 and 2.5 has much in common with the research process behind the present understanding of the CO + NO reaction described earlier. It will involve adsorption studies of water and proteins employing kinetic methods, using a variety of spectroscopic methods and atomic resolution, as well as microscopic methods to obtain the bonding, orientation, and structure of water and proteins on the surface. Simplified model systems of simple molecules and wellcharacterized surfaces as well as complex mixtures of biomolecules and heterogeneous surfaces will be employed. Cell-level interaction must also be incorporated as well as real in vivo experiments. Some studies will concentrate on static situations (snapshot pictures); others will attack the more difficult time evolution as outlined in Fig. 2.2. There will be demand for new and improved experimental methods. The scenario outlined above could equally well have started from the most highly complex, in vivo, situation and then approached the molecular level situation in descending order of complexity and length scales. It will be the mutual interaction between research efforts at all these levels of complexity and magnitude that will eventually allow us to draw a real rather than hypothetical interface scenario like the one in Fig. 2.2. Questions, methods, and research procedures addressing the cell- and tissue-level interactions are discussed, in detail, in several other chapters herein. Properties and Preparation of Implant Surfaces In this section we will briefly describe some stateof-the-art procedures by which implant surfaces can be prepared and analysed. We choose titanium and alloyed titanium as our examples because they are probably the biomaterial surfaces for which the most detailed characterization has been made. We do not intend to give a detailed account of the methods employed or of the results, since they are
26 The Material Surface
SURFACE + WATER + PROTEINS + CELLS Adsorbed water and proteins
Fig. 2.5 Illustration of basic studies of biomaterialbiosystem interaction at different levels. The original surface properties will influence the macroscopic
tissue response, via water bonding, protein adsorption, and cell behaviour,
Kasemo and Lausmaa: Analogues in Surface Science and Technology 27 both available in the literature [23-32]. We only intend to exemplify the type of results that can be obtained using current preparation and analysis methods. Properties We have recently completed an extended study of the preparation and characterization of pure Ti and TJ-6AMV [23, 24, 28-31], The main results are summarized in Table 2.1. The preparation methods employed were: clinical preparation methods (machining followed by cleaning and heat sterilization or autoclaving) [23, 24, 29]; thermal oxidation up to 450° C [28]; anodic oxidation producing oxide thickness in the range 5-200 nm [23, 29, 30, 31]: mechanical polishing; electropolishing. In addition to these methods we have recently incorporated glow discharge plasma preparation methods. The main characterization methods were scanning AES, XPS, depth profiling by AES and XPS, SIMS, and transmission electron microscopy (TEM). Additional information was obtained for selected samples using Rutherford scattering spectroscopy (RBS), scanning electron microscopy (SEM), scanning tunnelling microscopy (STM), and optical microscopy. The results summarized in Table 2.1 demonstrate that it is possible to systematically vary the properties of oxidized titanium surfaces. This could constitute a basis for future systematic biological studies of the interaction between material surfaces and proteins/cells/ tissues. "Clinical" implants have a very thin, amorphous, homogeneous, and non-porous oxide layer of essentially TiO: composition. Thicker thermal oxide films are also TiO 2 but are crystalline and show up a surface texture related to the underlying titanium microstructure. Electropolished surfaces are very smooth and have a very thin oxide layer like the "clinical surfaces." but this is doped with chlorine from the electropolishing electrolyte. Anodically oxidized surfaces show up a systematic variation in properties as the oxide thickness (i.e. the anodizing voltage) is increased. The thinnest films are amorphous, while thicker films (>100 nm) become increasingly more crystalline as the thickness increases (for pure Ti only). The anodized surfaces are quite heterogeneous with dense oxide regions alternating with porous regions. [An interesting comparison exists here with observations of differential phase structures reported by Smith in the discussion of ch. I . ) The pore size increases as the film thickness increases and is
approximately half of the film thickness. The anodized films can be deliberately doped (e.g. by sulphur or phosphorus) by choosing different electrolytes. The oxides on alloyed titanium (TJ-6A1-4V) have a thickness approximately corresponding to those of pure Ti. The major differences between pure and alloyed Ti are (1) the oxide on the alloy is enriched in the alloying elements Al and V, (2) the surface texture and microstructure are more complex and heterogeneous on the alloy, and (3) even thick anodic oxide films do not crystallize as do those on pure Ti. Even with careful cleaning and clean-room procedures and employing hermetically sealed sterile packages, implant surfaces contain at least of the order of one monomolecular layer of contaminants [see response to Brunette's questions in the discussion of this chapter]. The contamination layer is dominated by organically bound carbon (hydrocarbons, etc.), probably deriving from trace impurities in the ambient and/or from liquid cleaning agents. Other typical impurities occurring at the percentage or lower level are Ca, P, Si, Cl, S, Na. The role of these contaminants in the implant function is totally unknown. There are, however, strong theoretical reasons to keep a careful control on them [1, 22]. An interesting subject of future research is to determine if and how the biological systems are sensitive to the kind of variation in surface properties of a single material that was described above. Is a specific tissue or cell culture sensitive to whether the oxide is amorphous or crystalline, or 5 or 500 nm thick? Are they sensitive to dopants like S or P in the surface layer, or to the occurrence of porous regions? Is the presence of alloying elements influencing the tissue response? Can totally clean surfaces be implanted, and what consequences will that have? Preparation As more knowledge about how biological systems react towards different implant surfaces becomes available, control over the preparation procedures becomes increasingly more important. Surfaces may then be deliberately prepared with properties that have been shown to promote a positive biological response. A variety of preparation techniques will be available to produce such "tailor-made" surfaces. The microarchitecture of the surface can be built up in great detail using lithographic plus etching techniques initially developed for the microelec-
28 The Material Surface Table 2.1 Summary of some oxide characteristics for differently prepared Ti surfaces. Important preparation parameters which influence the final surface properties are also listed Parameters
Thickness
Composition
Structure
Impurities'*
"Clinical" oxides [23, 24]
Machining, cleaning, sterilization, handling
4-6 nmb
TiO2 + suboxides (Ti2O3, TiO) + small amounts TiN,
Amorphous, dense
C, Ca, P, Si, Cl, S, Na at % level
Thermal oxides [28]
Temperature, pressure, time, oxidizing gas (02, 03, H20)
10-40 nmc
TiO2 suboxides (+ OH in humid atmospheres)
Amorphous to crystalline (anatase or rutile)d dense
Depends on gas composition
Anodic oxides [23, 31]
Anode voltage, electrolyte, temperature
Up to several /tm
Ti02
Amorphous to crystalline,6 heterogeneous, dense to porous
Anions from electrolyte, (so432-, po4 -)
Anions (% levels)
Plasma oxides
Plasma gas composition and pressure, sample voltage, temperature, time
Up to several 100 nm
TiO2, variable
Amorphous to crystalline, dense
None (but care in handling is important)
Almost any
Electropolished Ti [31]
Electrolyte, temperature, time
3-5 nm
TiO2
Amorphous
Cl
Alloyed Ti, Ti-6Al-4V... [29, 30]
Bulk metal composition, and as above
As for pure Ti
TiO2 + % levels of A1O* and VO^
As for pure Ti
Al, V, and as above
Dopants
As for pure Ti
"For all preparation methods, the impurity levels depend on environmental handling; impurities listed are inherent for the method. b Thickness depends on method of sterilization. Thicknesses refer to oxidation in air at 200-450° C, for 1 h. d Depending on oxidation temperature; increasing temperature leads to increasing degree of crystallinity. e Depending on thickness; increasing thickness leads to increasing degree of crystallinity. Thin oxides can be crystallized by thermal treatment.
tronics industry [18]. The surface chemistry can be controlled employing methods of surface coating and preparation [33-35] developed for a variety of fields such as for catalysis, corrosion protection, tribology, semiconductor technology, cutting tools, and adhesive joints. Below we briefly outline how such systematic preparation can be performed, referring to our previous example of titanium / titanium oxide implants. To modify the oxide properties one can employ various methods of oxidation (thermal oxidation in different gases, anodic oxidation, glow discharge plasma oxidation [36], hydrogen peroxide oxidation [37, 38], acid passivation, ion implantation [35], etc.), and vary the parameters involved. Properties that can be relatively easily varied are the following (parameters used are given in parenthesis):
Oxide thickness (temperature, anodic voltage, plasma pressure and voltage, concentration of oxidizing agent, ion energy) Oxide crystallinity (temperature, anodic voltage, plasma pressure, and voltage) Oxide composition (anodic electrolyte, gas composition in plasma oxidation, bulk material composition, composition of oxidizing agent, ion species) Surface cleanliness (post-oxidation treatment, storage) Surface topography (lithographic techniques, micromachining) Employing such schemes makes it possible to produce surface oxides containing controlled amounts of, for example, nitride, carbide. One can also include "dopants" such as Ca, S, P in elemental form or as compound. Initially these possibilities
Kasemo and Lausmaa: Analogues in Surface Science and Technology 29 will be valuable mainly as research tools for evaluation of how surface properties influence biological response. Later, as new knowledge emerges, they will be useful for production of a new generation of implants. Future Outlook and Summary The most urgent need for the future is an increased understanding of material-tissue interactions. Important general questions are: Which of these interactions are material related and which, if any, occur independently of the material? How are material-related interactions influenced by various surface properties? At which stage(s) of the time evolution is the influence of the surface properties largest? Only when our understanding of the tissuematerial interface increases can we expect an accelerated development of new, better, and more advanced implants with regard to their biochemical/biological performance. This will require extensive research efforts in all disciplines involved. The alternative, purely empirical, trial-and-error approaches would be even more costly in all respects. A true cross-disciplinary approach is a prerequisite for a high success rate-when taking the scientific approach. Material and surface scientists will participate with their knowledge and methods of preparing and analysing materials. Protein and cell-surface interaction studies in vitro will form a basis for relevant in vivo experiments which will in turn suggest directions for clinical research and applications. The clinical results will suggest new in vivo experiments which in turn will suggest new in vitro experiments and so on in a closed cycle of mutual interactions between the most basic and the most applications-oriented work. This is quite an obvious thing to say. To achieve it is an enormous challenge. It involves breaking down many barriers between disciplines and subcultures to facilitate communication of knowledge between natural, technical, and medical sciences and industry. The outcome may be fascinating not only for the biomaterials area but also for many other areas in medicine, biology, chemistry, and physics. Acknowledgments We are grateful to our numerous colleagues and friends in biology and medicine for many fruitful and stimulating discussions. Funding of this work by the Swedish National Board for Technical
Development (contract nos. 90-01476P, 8901146P, and 88-00487) is gratefully acknowledged. References 1 Kasemo B, Lausmaa J (1986) Surface science aspects of inorganic biomaterials. CRC Crit Rev Biocompat 2:335-380 2 Kasemo B, Lausmaa J (1988) Biomaterials from a surface science perspective. In: Ratner BD (ed) Surface characterization of biomaterials. Progress in biomedical engineering series, Elsevier, Amsterdam, pp 1-12 3 Williams DF (ed) (1981) Fundamental aspects of biocompatibility, vols 1, 2. CRC Press, Boca Raton 4 Hench LL (1991) Biomaterials. Science 208:826831 5 Thomsen P, Bjursten LM, Ericson LE (1986) Implants in the abdominal wall of the rat. Scand J Plast Reconstr Surg 20:173-182 6 Ivarsson B, Lundstrom I (1986) Physical characterization of protein adsorption on metal and metal oxide surfaces. CRC Crit Rev Biocompat 2:1-96 7 Baier RE, Meyer AE, Natiella IR, Natiella RR, Carter JL (1984) Surface properties determine bioadhesive outcomes: methods and results. J Biomed Mater Res 18:327-355 8 Eriksson RA (1984) Heat-induced bone tissue injury. Ph.D. thesis, University of Gothenburg 9 Brunski JB (1988) The influence of force, motion and related quantities on the response of bone to implants. In: Fitzgerald R Jr (ed) Non-cemented total hip arthroplasty, chap. 2. Raven Press, New York, pp 7-21 10 Gristina AG (1987) Biomaterial-centered infection: microbial adhesion versus tissue integration. Science 237:1588-1595 11 Ducheyne P, Williams G, Martens M, Helsen J (1984) In vivo metal-ion release from porous-titanium fibre material. J Biomed Mater Res 18:293308 12 Williams DF (1981) Electrochemical aspects of corrosion in the physiological environment. In: Williams DF (ed) Fundamental aspects of biocompatibility, vol 1. CRC Press, Boca Raton, pp 11-42 13 Williams DF (ed) (1981) Fundamental aspects of biocompatibility, vols 1, 2. CRC Press, Boca Raton 14 Johansson CB, Hansson HA, Albrektsson T (1990) Qualitative interfacial study between bone and tantalum, niobium or commercially pure titanium. Biomaterials 11:276-280 15 Ericson LE, Bjursten LM, Engstrom G, Lausmaa J, Kasemo B, Thomsen P (1988) The ultrastructure of the intact interface between soft tissue and implants of titanium, zirconium, gold and PTFE.
30 The Material Surface Symposium on retrieval and analysis of surgical implants (Abstract) 16 Bjursten LM, Ericson LE, Lausmaa J, Mattsson L, Rolander U, Emanuelsson L, Thomsen P, Kasemo B (1990) Method for ultrastructural studies of the intact tissue-metal interface. Biomaterials 11:596601 17 Rostlund T, Thomsen P, Bjursten LM, Ericson LE (1990) Difference in tissue response to nitrogenion implanted titanium and c.p. titanium in the abdominal wall of the rat. J Biomed Mater Res 24:847-860 18 Cherhoudi B, Gould TRL, Brunette DM (1990) Titanium-coated micro-machined grooves of different dimensions affect epithelial and connective tissue cells differently in vivo. J Biomed Mater Res 24:1202-1219 19 King DA, Woodruff DP (eds) (1983) The chemical physics of solid surfaces and heterogeneous catalysis, vols 1^4. Elsevier, New York 20 Zangwill A (1988) Physics at surfaces. Cambridge University Press, Cambridge 21 Thiel PA, Madey TE (1987) The interaction of water with solid surfaces: fundamental aspects. Surface Sci Rep 7:211-385 22 Kasemo B, Lausmaa J (1988) Biomaterial and implant surfaces: on the role of cleanliness, contamination and preparation procedures. J Biomed Mater Res 22:145-158 23 Lausmaa J, Kasemo B, Rolander U, Mattsson L, Bjursten LM, Ericson LE, Rosander L, Thomsen P (1988) Preparation, surface spectroscopic and electron microscopic characterization of titanium implant materials. In: Ratner B (ed) Surface characterization of biomaterials. Progress in biomedical engineering series. Elsevier, Amsterdam, pp 161-174 24 Lausmaa J, Kasemo B, Mattsson H (1990) Surface spectroscopic characterization of clinical titanium implant materials. Appl Surf Sci 44:133-146 25 Smith DC, Pilliar RM, Murray G (1985) Preliminary studies on the surface characterization of dental implant materials. Trans llth Ann Meeting Soc Biomater, p 8 26 Maeusli PA, Bloch PR, Geret V, Steinemann SG (1986) Surface characterization of titanium and
27
28
29 30 31
32 33 34 35 36 37
38
titanium alloys. In: Christel P, Meunier A, Lee AJC (eds) Biological and biomechanical performance of biomaterials. Elsevier, Amsterdam, p 57-62 Gardella Jr JA, Vargo TG, Hook TJ, Grobe HI GL, Salvati Jr L, Hautaniemi J (1988) Electron and vibrational spectroscopic studies of the surfaces of biocompatible materials. In: Ratner B (ed) Surface characterization of biomaterials. Progress in biomedical engineering series. Elsevier, Amsterdam, pp 145-160 Radegran G, Mattsson L, Lausmaa J, Rolander U, Kasemo B (1991) Preparation of ultra-thin oxide windows on titanium for TEM analysis. J Elec Micros Tech (in press) Ask M, Lausmaa J, Kasemo B (1989) Preparation and surface spectroscopic characterization of oxide films on Ti6A14V. Appl Surf Sci 35:283-301 Ask M, Rolander U, Lausmaa J, Kasemo B (1990) Microstructure and morphology of surface oxide films Ti6A14V. J Mater Res 5:1662-1667 Lausmaa J, Kasemo B, Mattsson H, Odelius H (1990) Multi-technique surface spectroscopic characterization of electropolished and anodized Ti. Appl Surf Sci 45:189-200 Kasemo B, Lausmaa J (1988) Biomaterial and implant surfaces: a surface science approach. Int J Oral Maxillofacial Implants 3:247-259 Chapman BN, Anderson JC (eds) (1974) Science and technology of surface coatings. Academic Press, New York Vossen JL, Kern W (eds) (1978) Thin film processes. Academic Press, New York Picraux ST (1984) Ion implantation metallurgy. Phys Today 37:38^4 Chapman BN (1980) Glow discharge processes. Wiley, New York Tengvall P, Elwing H, Sjoqvist L, Lundstrom I, Bjursten LM (1989) Interaction between hydrogen peroxide and titanium: a possible role in the biocompatibility of titanium. Biomaterials 10:118120 Tengvall P, Elwing H, Lundstrom I (1989) Titanium gel made from metallic titanium and hydrogen peroxide. J Coll Interface Sci 130:405^13
DISCUSSION Brunette: Dr. Kasemo, I was wondering about a statement in your paper that implant surfaces contain at least a monomolecular layer of contaminants. Is the evidence for that good? Kasemo: Yes, very good. I would say that any surface I have been looking at, any surface that has been exposed to the ambient for just a few tenths of a second - this holds for biomaterials as
well as any other surfaces although there are a few exceptions - is always contaminated with hydrocarbons from the air. Always. Some of them may be so loosely bound that immediately as you dip them in water or solution they will be replaced. There are some exceptions, for example, graphite. The only surfaces or the few surfaces that work in air to give atomic resolution are those of golden
Kasemo and Lausmaa: Analogues in Surface Science and Technology 31 graphite, and this is because these surfaces are exceptions to this rule of covering themselves with contaminants. Brunette: And is it your view that it is a continuous layer of contamination, or is it spotty? Kasemo: It's very heterogeneous because, depending on one could say surface energy or depending on how unsaturated the chemical bonds are on the surface, there may be spots which are not covered by contaminants, but they will certainly be on, say, between 10% and 100% of a monolayer. Sauk: How dynamic is that absorbed protein layer on the surfaces of most bone materials that we know of today with regard to the quality of the absorbed surface? Some of the people in the cardiovascular field in particular talk about tremendous dynamics which exist in the qualities of those surfaces. Kasemo: Protein absorption is not yet my area of expertise, but one of my good friends, Ingemar Lundstrom, and other people are looking into this problem, and I think the answer is, as you indicate, that it's very dynamic - the first proteins that are absorbed on the surface are very likely not to be there later on. And these are processes occurring on a very short time scale. Say the time scale for water is nanoseconds. Then the time scale for the smallest proteins, I would say, is milliseconds, and after 10 s there may be different proteins on the surface, and after 1 day, still different proteins, and so on. So I think it's a very dynamic interface, but it doesn't mean that the memory is forgotten. I mean, it's a boundarycondition problem that there may be a connection between what was there first and what is there later on on the surface. Niznick: Dr. Kasemo, dealing with the question of contaminants on the surface and whether some contaminants are desirable or undesirable, I want to ask you a question about a paper you filed. Actually, it was a European patent in 1986, patenting titanium dioxide with certain contaminants in it, and you claimed in that that calcium, sodium, and zinc are the desirable surface contaminants. The question is: What exactly did you discover in 1986, and do you believe it still to be true? Because I am familiar with titanium dioxide prior to 1986. Kasemo: I don't think I claimed that these contaminants were good. Because we know so little, I cannot say, I don't think anybody can say what is a good and what is a bad contaminant. I am not prepared to make any value judgment. What I think this paper expressed was that it
relied entirely on clinical documentation that this was the surface that had worked. It doesn't prove that these contaminants are good or bad or just unimportant. I think if you had a surface which has been shown to work well, and you characterize it very well, this is the working surface, and you have to be careful to make changes in that surface before you know what are the species that were there in the first place, what their roles are. So that's the only claim; no value judgment. Lemons: I agree with your comment that everything is contaminated, but I also believe that the biological pathways and biological recognition mechanisms can operate at below parts-per-billion and perhaps parts-per-trillion concentrations. Your emphasis is on the first reaction in what's laid down, but we also realize that everything is dynamic and that surface is changing over time. So my question is: How much emphasis should be given to the other side of the equation, and should we in fact try to track what happens to these interfaces over time to find out if these first reactions are predictive or not? Is that true? Kasemo: No. I wish I had the answer to your question: How important are the early stages and how important are the later ones? Again, I am not prepared to make a value judgment there. I think that the evidence we have is that the surface is very dynamic and that later stages look very different from the first stages. If there is a memory effect or not I am not prepared to say. It means that we have to have the time axis as an important axis in our studies. It may be, for example, that if the surface denatures proteins with fragments which are irreversibly stuck to it, that creates a new surface, and I think that surface might have an influence on later times. I also think there will be cases with another material where the memory facts are very poor. So I think there is no general statement you can make about this dynamic interface. Some will have a memory effect from the very beginning; some will forget what properties they had from the beginning; and the surface will be very much remodelled, as we will hear in the next talk. McVay: I would like to address the question of whether the oxide layer grows with time, and do you have any clues as to whether that plays a role or does the surface texture change with time, and could that be important? Or do you have reactions of the outside layer with minerals or elements in the body which change it to become more acceptable? Do you have any clues to any of those? Kasemo: Again, not real clues. The way we work
32 The Material Surface just now is that we have made a very broad characterization of oxidized titanium surfaces, but crystalline, non-crystalline heterogeneous, homogeneous surfaces, very thick oxide films, very thin, very pure, and doped oxide films, several of these samples are sitting in vivo. We are waiting for the results. I am not prepared to say that one of these titanium surfaces is better than the other, but this is I think the direction to go. By making
these types of systematic studies we will get the answers. Now, I don't have a clue, but if you want to hear my opinion certain groups on the surfaces, certain chemical groups, sulphate groups, calcium ions, and so on, will I think have an influence. They almost have to have a profound influence because they create new bonding sites for biomolecules.
3
Surface Reaction Kinetics and Adsorption of Biological Moieties: A Mechanistic Approach to Tissue Attachment L.L. Hench
A strongly adherent, mechanically bonded interface develops between certain inorganic implant materials and bone; this is well established. [1-17, 34]. The materials on which such a bonded interface forms are called bioactive. The mechanical strength of the bonded interface has been shown, subject to testing methods, to equal or exceed that of the bone with which the bioactive implant is bonded [6, 12, 17-19, 34]. For a few glass compositions the bioactivity is sufficiently high and the rate of bonding sufficiently rapid that a strongly adherent bond will also form with soft tissues [20-23]. The tissue bonding process has been described in some detail at a physical-chemical, histological, and ultrastructural cellular level (as reviewed in [1, 4, 6, 7, 24, 25]). However, as yet there is meager understanding of the interfacial effects occurring prior to the establishment of a well-differentiated layer of adherent cells actively generating collagen within the bonding layer [26]. The purposes of this chapter are (1) to review what is or is not known about the earliest stages of the development of a bioactive bond and (2) to propose a hypothesis regarding the role of adsorbed acellular biological moieties in bioactivity and bonding. Bioactivity: A Kinetic Phenomenon Our starting point in understanding interfacial bonding is to recognize that bioactive implants exhibit a time-dependent compositional change of their surface during implantation [27, 28]. Thus, there is a dynamic interaction between the changing implant surface and the repairing tissue site. The kinetics of this surface change can be summarized by the following rate equation:
[Where /?s is the overall rate constant and the indi-
vidual rate constants are: kl for cation exchange and silica hydrolysis, k2 for the interfacial reaction, A:3 for the condensation reaction, and k4 for the HCA precipitation reaction. Note that the time exponent, v, for stage 4 is still of an unknown value.] Note that there is a new term in this equation compared with earlier publications [27, 30]. Evidence for the new 5th stage t5 term, which is the time for onset of crystallization of the amorphous calcium phosphate film formed on the surface (£4), is discussed later. Also note the negative sign for stages 1 and 2 reactions, which involve a loss of chemical species from the glass surface by dissolution. The sign of stage 3 is positive because the polycondensation reaction results in the formation of a silica-rich layer, and stage 4 is positive because of precipitation of a calcium phosphate layer on the surface. Equation 1 describes the time-dependent changes of only a single phase amorphous or glassy material. If one desires to understand a multiphase bioactive implant, such as Ceravital, apatite/Wollastonite (A/W) glass-ceramic, or polycrystalline sintered hydroxylapatite, it is necessary to establish a kinetics equation for each phase and each interface between the phases. For example, consider the following hypothetical interface for a bioactive implant with two crystalline phases, a—(3, and a residual glassy phase, y: Tissue fluid Interface l Y | a | Y l p | y | a | Y l p l Y | o c | y l P l y l Implant The relative area fractions (Af) of each of the three phases are approximately: A f (a) = 0.4, Af(/3) = 0.4, and A f (y) = 0 . 2 . Each phase will have a reaction rate equation corresponding to Eq. 1; however, some of the terms may be 0, if, for example, there is no alkali or silica in a phase. Thus, the overall rate equation for the implant will be the sum of the reaction rates of the phases
34 The Material Surface Table 3.1 Reaction stages of a bioactive implant Stage 1 Rapid exchange of Na + or K + with H+ or H3O+ from solution, Si—O—Na+ + H+ + OH" -> Si—OH + Na+(solution) + OH" Stage 2 Loss of soluble silica in form of Si(OH)4 to the solution resulting from breaking of Si—O—Si bonds and formation of Si—OH (silanols) at the glass solution interface, 2(Si—O—Si) + 2(OH~) -» Si—OH + OH—Si Stage 3 Condensation and repolymerization of a SiO2-rich layer on the surface depleted in alkalis and alkaline earth cations, 2(Si—OH) + 2(OH—Si) -> —Si—O—Si—O—Si—O—Si—O— Stage 4
Migration of Ca2+ and PO43" groups to the surface through the SiO2-rich layer forming a CaO-P2O5-rich film on top of the SiO2-rich layer, followed by growth of the amorphous CaO-P2O5-rich film by incorporation of soluble calcium and phosphates from solution.
Stage 5 Crystallization of the amorphous CaO-P2O5 film by incorporation of OH", CO32~, or F" anions from solution to form a mixed hydroxyl, carbonate, fluorapatite layer.
times their area fractions: i.e., RS = OARa + QARft + 0.2Ry
(2)
Unfortunately, chemical reactions at liquidinterphase boundaries between glass/crystal and crystal/crystal are often accelerated owing to mismatch of the respective lattices. Consequently, two additional interface reaction terms, Ra_y and Rftr must be added, multiplied by their respective area fraction terms, which are usually in the range of 0.01 to 0.05, depending on grain size and grain shape (roughness). Thus, for a three-phase bioactive glass-ceramic
Since each of these five terms can contain as many as five individual reaction rates (e.g., Eq. 1), a full kinetics description of a bioactive implant could involve as many as 25 separate but interrelated reaction rates (kl-k25) contributing to the time-dependent change of the surface. Consequently, it is not hard to understand why most existing data are only for single-phase bioactive glasses, such as are discussed below. The five stages of reaction present in Eq. 1 are summarized in Table 3.1. The literature base for stage 1 and stage 2 reactions (ion exchange (A:,) and silica network dissolution (Jt2)) is quite extensive (as reviewed in [27-30]). Measurements of stage 3, silica repolymerization, are less extensive but thoroughly documented [31-33]. Surface compositional profiles resulting from stage 1
through stage 4 reactions have been measured for numerous glass compositions (as reviewed in [28-30]). The effects of adding sparingly soluble cations, such as Mg, Ca, Al, to glasses and/or reaction solutions are also well documented [28, 30]. Even the effects of glass/crystal interfaces on overall reaction rates have been measured for simple systems [35]. Thus, the general behavior of glasses in contact with various liquids is understood. A thermodynamic basis for predicting the relative stability of complex glass compositions in terms of free energy of hydration has been developed and is very effective. This method is described by Jantzen and colleagues, who applied it to more than 300 different glass compositions (see Fig. 3.1) [36, 37]. The locations of the free energy of hydration of a bioactive glass 45 S5 and a glass just outside the bone-bonding boundary 60S5 [38] have been calculated by Jantzen and are shown in Fig. 3.1. Example of Glass Surface Reaction Kinetics (45S5 Bioglass) The "grandfather" bioactive glass 45S5 has been used as a baseline for most surface studies, largely because it is single phased and also because it has the highest in vivo bioactivity index 7B, as discussed below. Recent investigations [32, 33, 39] show quite clearly the reaction sequence depicted by Eq. 1 for 45S5 Bioglass exposed to a Trisbuffer solution. Figures 3.2 and 3.3 summarize these findings, determined using Fourier transform
Fig. 3.1 Linear regression plot of over 300 experiments relating glass composition (AG h v d ) to glass durability (expressed as Si lost from the glass to the leachate solution in a 28-day laboratory experiment) [36].
infrared (FTIR) spectroscopy of the reacted surface as a function of exposure time. All five reaction stages are clearly delineated by changes in the vibrational modes of the chemical species in the surface. Figure 3.2 shows the 45S5 glass surface before reaction, after 1 h, and after 2 h. The peaks are identified. The alkali-hydrogen ion exchange and network dissolution (stages 1 and 2) very rapidly reduce the intensity of the Si—O—Na and Si—O—Ca modes and replace them with Si—OH bonds with one non-bridging oxygen (NBO) ion. Alkali is depleted to a depth of —0.5 /mi within a few minutes. As the stage 1 and 2 processes continue the Si—OH single NBO modes are replaced with — Si — OH ;
OH
i.e., Si—2NBO, stretching vibrations which are in the range of 930/cm decreasing to 880/cm. By 20 min the Si—2NBO vibrations are largely replaced by a new mode assigned to the Si—O—Si bond vibration between two adjacent SiO4 tetrahedra. Thus, this new vibrational mode corresponds to the formation of the silica gel layer by the stage 3 polycondensation reaction between neighboring surface silanol groups. This mode decreases in frequency until it is hidden by the growing apatite layer after 1 h. As early as within 10 min a P—O bending vibration associated with formation of an amorphous calcium phosphate layer appears. This is
Fig. 3.2 FTIR spectra of 45S5 at 0 h, 1 h, and 2 h reacted in 37° C Tris-buffer solutions.
due to precipitation from solution, stage 4 in Eq. 1. In fact, Clark et al. show, using Auger electron spectroscopy (AES), that even by 2 min calcium and phosphate enrichment has occurred on the sample surface to a depth of approximately 20 nm [40]. Ogino et al. show that by 1 h the calcium phosphate layer has grown to a thickness of 200 nm [38]. Within 40 min the P—O bending vibration is strong with a continually decreasing frequency as the layer builds. At about 1.5 ± 0.2 h the P—O bending vibration associated with the amorphous calcium phosphate layer is replaced with two P—O modes assigned to crystalline apatite [41]. Concurrent with the onset of apatite
36 The Material Surface
log time (sec.) Fig. 3.3 Time dependence of wavenumber of oxide IR modes in a 45S5 Bioglass implant reacted in 37° C Tris-buffer solution.
crystallization, stage 5, is the appearance of a C—O vibrational mode associated with the incorporation of CO32~ in the apatite crystal lattice as described by Kim et al. [33] and LeGeros et al. [42]. The C—O mode decreases in wavenumber as the hydroxylcarbonate apatite (HCA) layer grows. By 10 h the HCA layer has grown to a thickness of 4 ^im [38], which is sufficient to dominate the FTIR spectra and mask most of the vibrational modes of the silica gel layer or the bulk glass substrate. By 100 h the polycrystalline HCA layer is thick enough to yield X-ray diffraction results showing the primary 26° and 33° 26
peaks with considerable line broadening [24]. By 2 weeks the crystalline HCA layer is equivalent to that of biological apatites [43]. Glass Compositional Effects When fluorine ions are present they also are incorporated in the HCA crystals [32, 33], distorting the crystal structure and forming fluorapatite [43]. Otherwise, the substitution of 40% CaF2 for CaO in the 45S5 composition has little effect on the sequence or rate of surface reactions. The onset of hydroxylcarbonate fluorapatite (HCFA) crystalliza-
Hench: Surface Reaction Kinetics 37 tion is increased very slightly to 1.8 ± 0 . 1 h (as described in [39]). Tissue response, both hard and soft, is also similar for the fluoride-containing bioactive glass (as described by Wilson and coworkers [20, 21]). Substitution of 2% B2O3 for 2% CaO and addition of 1% P2O5, thereby altering the Ca/P molar ratio in basically a 45S5 glass, yields glass S45P7 (described by Andersson et al. [44]), which also rapidly develops a strong bond to bone. (The soft-tissue bonding of this glass is as yet untested.) The reaction sequence observed for S45P7 is identical to that for 45S5; Eq. 1 is followed, for example, but at a somewhat slower rate [39], The onset of crystallization of HCA stage 5, is at 2.5 ± 0.5 h. However, the change in Ca/P ratio of the glass did not affect the rate of the silica condensation reaction, stage 3, which occurred after approximately 20 min of reaction for 45S5, 46SF, and S45P7 glasses. Kokubo [25, 45] has shown that a Ca- and Prich layer is also present at the bonding interface between the polycrystalline A/W-glass-ceramic and bone. However, the silica-rich layer was not present, even though a substantial concentration of soluble Si was lost to solution. Likewise Kokubo and co-workers have shown that the Ca- and Prich layer is present for Ceravital-type glassceramics. Hohland and co-workers report [46] a Ca- and P-rich layer at the bone interface with a glass-ceramic composed of phologopite and apatite crystals. Kokubo has also demonstrated [25] that a phosphate-free CaO-SiO2 glass will form an apatite layer on its surface on exposure for 2 days to a simulated body fluid that contains only 1.0 mM HPO42 . The CaO-SiO2 was confirmed to bond to living bone by the surface apatite layer. Previously Ogino and co-workers [38] showed that P2O5-free Na2O-SiO2 glasses form an apatite layer on their surface when exposed to an aqueous solution containing calcium and phosphate ions. Recently, Li and co-workers [47] reported that glasses containing primarily SiO2, with only 10 mole % of CaO and P2O5 and no Na2O, will form apatite layers in a Tris-buffer solution. Earlier, Walker [48] demonstrated than even nearly pure SiO2 will eventually form a bone bond if the substrate has a very high surface area, >400 m2/g. Unfortunately, the interface was not analyzed for the presence of an interfacial apatite layer which could very well have been nucleated on the surface by dissolution of soluble silica. For years it has been shown that synthetic hydroxyapatite implants will bond to bone by forming a new epitaxial apatite phase at the interface [15].
[Epitaxy was discussed but not demonstrated by Jarcho in [15].] Consequently, we are forced to conclude that bioactivity occurs only within certain compositional limits and very specific ratios of oxides in the Na2O-K2O-CaO-MgOP2O5-SiO2 systems; however, the extent of these compositional limits and the physical-chemical and biochemical reasons for the limits are poorly known at present. We do know that for a bond with tissues to occur a layer of biologically active HCA or HCFA must form. This is perhaps the only common characteristic of all the known bioactive implant materials [see also van Blitterswijk et al., ch. 27 herein]. It is the rate of HCA formation (stage 4) and the time for onset of crystallization (stage 5) that vary so greatly. When the HCA formation rate becomes excessively slow no bond forms and the material is no longer bioactive. Our next step in understanding is to relate the implant surface reactions to in vivo response. Relation of Surface Kinetics to Rate of Bone Bonding By changing the compositionally controlled reaction kinetics (Eq. 1), the rates of formation of hard tissue at a bioactive implant interface can be altered (as shown in Fig. 3.4). Thus, the relative bioreactivity of the material (also shown in Fig. 3.4) is compositionally dependent. The level of bioactivity of a specific material can be related [27] to the time for more than 50% of the interface to be bonded (t05bb), for example, 7B = (100/r05bb). It is necessary to impose a 50% bonding criterion for an 7B since the interface between an implant and bone is irregular. Gross's group have shown that the initial concentration of stem cells, osteoblasts, chondroblasts, and fibroblasts varies as a function of the fit of the implant and the condition of the bony defect [6, 7]. Consequently, all bioactive implants require an incubation period before bone proliferates and bonds (Fig. 3.4). The length of the incubation period at which this process occurs varies over a wide range depending on composition. The primary hypothesis of this paper is that the concentration and type of adsorbed proteins control this incubation period. The compositional dependence of 7B suggests that there are iso/B contours within the bioactivity boundary for any specific compositional system. Comparative data from the literature and extrapo-
38 The Material Surface
BIOREACTIVITY SPECTRUM
Fig. 3.4 Effect of time and composition on rate of bone bonding to bioactive implants. lation based upon relative reaction kinetics suggest the iso/B contours shown in Fig. 3.5, for the SiO2-Na2O-CaO-P2O5 glass system [27]. The change of 7B with the SiO2/(Na2O + CaO) ratio is very large as the bioactivity boundary is approached (as is illustrated in Fig. 3.6). The data in Fig. 3.6 were obtained by Ogino et al. [38] using AES analyses of the compositional dependence of rates of HCA formation. Compositions with tm in diameter, relative to the HA in bone, which is in the nanometre scale. Whether the HA actually extends over the polyethylene surface is something that our current studies are looking at. Lemons: I was taken by the significant differences between what appeared to be the monolithic ceramic HA and the directly adjacent biological apatite. I would have assumed that you would have gotten a moire fringe pattern from the monolithic structure because of the high degree of orientation, assuming equal thinness in the preparation. I didn't see that. Is it something that you are seeing, or can you actually go across the interface to the synthetic analogue? Bonfield: Yes. We have tracked from one to the other. We can actually do a sort of depth profile and show continuity at local areas from the synthetic to the natural material. Yes. Pilliar: It's really surprising to see these sections staying intact through the preparation processes. It must be very tedious. Could you comment a little bit on the problems associated with getting these thin films for examination in this form? Just how often do you see the region which really gives you this information rather than breakage at that interface? Bonfield: It's like all techniques. It needs a good Ph.D. project, and this actually started with a Ph.D. student who developed the technique. Now, Dr. Luklinska can routinely prepare the sections. They are very thin. They have to be somewhere between 100 and 200 A. They have to be very thin; they are very fragile. If you in fact have a fibrous encapsulation it would disintegrate. There has to be a degree of bonding between the two. Pilliar: It's just straight ultramicrotomy, then? Bonfield: It is ultramicrotomy, but with a little bit of black magic and experience. It's not the thing
you can rush out into the lab and do. Because part of our program is to look across a range of materials we are now asking the obvious question: What happens with a bioactive glass or with an AAV glass ceramic? Can one actually use this technique to provide additional information on the bonding between the synthetic and the natural surface? Kasemo: If I understand correctly, the basic idea is to use a polymer matrix and embed the HA grains as nucleation centres for epitaxial growth. Bonfield: Well, the original idea was to put HA in a polymer to make it fracture-tough and then to see whether a material of comparable modulus, but superior fracture toughness, could still get the tissue to recognize the HA. Provided you have more than 20% by volume the tissue will recognize the HA, as these experiments seem to confirm. Kasemo: In the initial preparation, how do you ascertain that the active surface of the material is not covered by the polymer? Is it polished? Bonfield: It is machined or ground to expose the HA, yes. Kasemo: I think your result is really a fantastic one if the interpretation is correct, that you see the epitaxial growth on the HA grains. Is there any way that interpretation could be wrong, or do you really think that there is epitaxial growth with perfect structure from the grain into the extracellular matrix? Bonfield: The answer to the second part of your question is, I think, yes. We need to do more work, but I think it's addressing the real crunch issue, which is if you want a long-term interface between the implant and bone which is stable and by this I mean not just 2 years, but 20, 30, 40 years - it seems to me this is one of the key factors to look for, because essentially there is no interface. And remember, the other feature is in that little microbiomechanical situation you have actually got materials that are really quite similar. I want actually to extend this into looking at the effects of time on that interface, and I hope perhaps that at the next meeting that Dr. Davies will organize I can report the results to you. [The next Bone Biomaterials Interface Workshop will be held in Leiden, The Netherlands, and will be organized by Dr. Clemens van Blitterswijk. - Ed.]
Part 2 Bone Proteins and Other Macromolecules
This page intentionally left blank
9
Non-Collagenous Bone Proteins and Their Role in Substrate-Induced Bioactivity J. Sodek, Q. Zhang, H.A. Goldberg, C. Domenicucci, S. Kasugai, J.L. Wrana, H. Shapiro, and J. Chen
Since the introduction of dissociative extraction procedures almost 10 years ago the majority of the major proteins of the mineralized bone matrix have been isolated and characterized. A number of these proteins are known to be derived from tissue fluids and appear to be concentrated in bone tissue through their affinity for hydroxyapatite (HA). However, those proteins in bone that are synthesized by osteoblasts are believed to be important, not only in the formation of the organic matrix, but also in the formation, growth, and regulated dissolution of the HA crystals. Several of the osteoblast-derived proteins are either unique to mineralized connective tissues or are produced in a form that is unique to these tissues. As such, they provide a valuable means of studying the differentiation of the cells that form mineralized tissue as well as for studying the formation of the mineralized tissue itself. Moreover, in the study of osseous tissue integration with biomaterials the expression of bone proteins can provide important insights into the response of bone cells to the environment created by the presence of an artificial material. In this chapter current information on the major bone-cell-derived proteins of bone is reviewed and the importance and potential function of these proteins evaluated from studies of bone formation in vivo and in vitro. These same proteins appear to be present in the different forms of bone tissue (woven and lamellar) and in bone produced by different pathways (endochondral versus intramembranous), although qualitative differences may be expected. In addition to the major proteins discussed here a large number of minor proteins including growth factors and proteins with boneinductive activity have been described. While many of these proteins are likely to be important, especially in modulating cellular activity during bone formation and in bone tissue homoeostasis,
and therefore merit attention, discussing them further is beyond the scope of this review. Since degradation of bone proteins is thought to occur coincidentally with the increased mineralization of mature bone [1], dissociative extraction procedures [2] are generally used to isolate the proteins from foetal and neonatal periosteal bone. To isolate non-covalently associated proteins from bone tissues in an intact form, procedures have been developed in which a strong dissociative agent, 4 M guanidine hydrochloride (GuHCl), is used first to extract proteins from the soft tissues and osteoid and is then used in combination with 0.5 Mtetrasodium ethylenediaminetetraacetic acid (EOTA) to extract the mineral-associated proteins [2]. In all the extraction solutions an inhibitor cocktail, comprising 1 mM phenylmethanesulphonyl fluoride, 5 mM benzamidine, 5 mM Nethylmaleimide and 100 mM e-aminocaproic acid, is included to prevent proteolysis during the isolation procedures. A modification of the extraction procedure employed in our laboratory [3] uses EOT A alone after the first guanidine (Gl) extraction, and 4 M GuHCl is used again (G2) after demineralization to isolate those proteins that are preferentially associated with the demineralized collagenous matrix. Following an exhaustive washing in PBS, the tissue residue is then digested with highly purified bacterial collagenase to release additional proteins which appear to be tightly bound to the collagen fibrils. Analysis of the protein extracted from foetal porcine calvarial bone has revealed that approximately 90% of the non-collagenous protein is extracted in the demineralizing step, the remaining 10% being distributed in the guanidine and collagenase extracts. The SDS-PAGE profiles of proteins obtained in these extracts are shown in Fig. 9.1, and the characteristics of the major non-collagenous proteins are presented in Table 9.1. It should be noted that
98 Bone Proteins and Other Macromolecules Mr G l
E
Gl
G2
E
G2
CD
1 2
93K 67K 45K 31K 21K
14.4K
Fig. 9.1 SDS-PAGE analysis of proteins from foetal porcine bone by dissociative extraction. Proteins were first extracted from the soft tissues with 4 M guanidine hydrochloride (Gl extract) and then the mineral-associated proteins released by extraction with 0.5 M EDTA (E). Proteins remaining in the collagenous matrix were released by a second extraction with 4 M guanidine hydrochloride (G2) and by digesting the remaining collagenous matrix with bacterial collagenase (CD). The proteins from each fraction were separated using SDS-PAGE on 15% cross-linked gels and stained with Coomassie blue (panel A) or "Stains-All" (panel B). Mr, molecular mass markers; 1 and 2, purified SPP-1 and BSP, respectively. Table 9.1 Major non-collagenous proteins of mineralized bone Bone extracts Size (kDa) Proteoglycans CS-PGI (biglycan) Core protein CS-PG II (decorin) Core protein CS-PG III Core protein Sialoproteins Secreted phosphoprotein 1 (SPP-1)/ osteopontin, BSP I Bone sialoprotein (BSP)/BSP II SPARC/osteonectin Osteocalcin
S04
P04
Gl
E
G2
CD
-200 46 -120 45, 47 -110 30-38 32 34 30 6
while the use of dissociative conditions has facilitated the isolation and purification of the bone proteins, protein unfolding during the isolation can preclude the analysis of the structure and functional properties of these proteins. Collagenous Proteins The collagenous matrix of mineralized bone tis-
sues is formed almost entirely of type I collagen which represents ~70% of the organic constituents and —90% of the protein in bone. The collagen is assembled into fibrils and fibres within and between which HA crystals form. The intertwined collagen fibres in woven bone form a contrast to the more regular arrangement of fibres in lamellar bone. As in other connective tissues, the fibres of type I collagen are responsible for the structural
Sodek et al.: Non-Collagenous Proteins in Substrate Bioactivity 99 integrity of the tissue. However, in bone and other mineralized connective tissues the collagen is capable of nucleating the formation of HA crystals, which have been observed to appear initially in the "gap" region of the collagen fibrils. In addition to the type I collagen, a small amount (—3%) of type V collagen comprising cd(V), «2(V), and al(XI) chains in a 1:1:1 ratio is present [4]. However, the precise distribution of the type V collagen and its function in bone are unknown. Of interest, al pN-propeptide of type I collagen, which is released during the proteolytic processing of procollagen, is also present in significant amounts in the bone matrix [5, 6]. The propeptide associated with mineral appears to be phosphorylated [5]. A second, possibly nonphosphorylated, form of the propeptide which is associated with the organic matrix has also been identified [6]. These propeptide forms have the potential to regulate fibril-associated HA formation and also, if released during bone resorption, they can regulate osteoblast activity by inhibiting collagen synthesis. Two forms (28 and 25 kDa) of a small collagenous protein, termed "small collagenous apatite-binding" (SCAB) proteins, have also been identified in demineralizing extracts of porcine bone [7]. Since these proteins, like the al(I)pN-propeptide, bind strongly to HA and are partially degraded by bacterial collagenase, it is conceivable that these proteins might also represent pN-propeptides, perhaps those of type V collagen. Proteoglycans The major proteoglycans (PGs) that have been isolated from bone all contain chondroitin sulphate (CS) as the attached glycosaminoglycan (Fig. 9.2). A large 1,000 kDa CS-PG is present in the non-mineralized bone matrix with at least three smaller CS-PGs associated with the mineral phase [8-14]. One of the small proteoglycans isolated from human and bovine bone, known as CS-PG I or biglycan, has an Mr of —350 kDa and a protein core of 46 kDa, and two CS chains attached to the core protein. The amount of CS-PG I appears to be generally low and variable in the different species examined but was not observed in porcine bone [13]. A more prominent small proteoglycan (Mr ~ 120K-200K), CS-PG II (decorin), has a single CS chain and a core protein that migrates as a doublet on SDS-PAGE at 45 and 47 kDa. The CS-PG I and CS-PG II proteoglycans have been shown to be homologous to the DS-PG I and DS-
PG II proteoglycans of cartilage [15]. Using monoclonal antibody probes, Goldberg et al. [13] have shown that CS-PG II core protein is homologous to the small soft connective tissue proteodermatan sulphate proteoglycan. It appears, therefore, that osteoblasts lack the epimerase activity that converts glucuronic acid to iduronic acid in the formation of dermatan sulphate glycosaminoglycan chains. A third class of small CS-PG that binds more strongly to HA ion-exchange resins is evident in demineralizing extracts of bovine and rat bone and has recently been characterized in foetal porcine bone [13]. Two species, originally called HAPG2 and HAPG3 [13], that appear to have the same protein core have subsequently been described as CS-PG III [16] in keeping with current nomenclature. This proteoglycan (Mr ~ 110K) has a protein core of 30-38 kDa which does not stain with either Coomassie blue or silver stains but produces a turquoise-blue colour with "Stains-All." The acidic nature of the protein core and staining properties are similar to those of the bone sialoproteins (see below), but the size and other properties of the protein core indicate that it may be unique. Consequently, the CS-PG III proteoglycan may be characteristic of mineralized tissue. The large CS-PG and both CS-PG I and CS-PG II are present in the soft tissues of bone, including osteoid, and we have found proteoglycan with properties similar to CS-PG II present in the G2 extracts of demineralized bone. However, most of the CS-PG I and CS-PG II, together with all of the CS-PG III, is recovered following demineralization with EOTA, demonstrating their strong association with the mineral crystals. Using a single injection of J5SO4, Prince et al. [17] have shown that there are two routes for the deposition of proteoglycan in newly formed rat bone. Approximately one-half of the proteoglycan is found at the mineral front within 2 h of synthesis, whereas the remainder is deposited with the collagen in the newly formed osteoid. Although it is apparent that a significant amount of 35SO4 label is also incorporated into bone sialoproteins, the existence of these two routes is supported by results obtained from in vitro studies of proteoglycan synthesis in foetal porcine calvaria [16]. In this study, the radiolabelled proteoglycan that associates rapidly with the mineral phase of bone was identified as CS-PG III, whereas most of the CS-PG II proteoglycans were deposited initially in the soft tissue. Although chondroitin sulphate proteoglycans have been shown to be synthesized by some rat bone cells in culture [18], in cultures
100 Bone Proteins and Other Macromolecules BONE PROTEOGLYCANS
Large CS-PG
CS-PG I biglycan
BONE SIALOPROTEIN
SPARC/OSTEONECTIN
CS-PG II decorin
CS-PG
SECRETED PHOSPHOPROTEIN 1 OSTEOPONTIN
OSTEOCALCIN
Fig. 9.2 Diagrammatic representation of the structures and binding properties of the major bone proteins. In contrast to the large CS-PG the three small CS-PGs all bind to the mineral in bone. However, whereas CS-PG III is extracted almost completely on demineralization, CS-PG I and CS-PG II are also present in the soft tissue where the CS-PG I can influence the formation of collagen fibrils and CS-PG II can bind to the collagen fibrils through its core protein. Bone sialoprotein is shown as an open structure with polyglutamic acid sites that may bind to HA and an RGD sequence that can bind cells. Sites of serine phosphorylation are shown as black dots, sites of N-linked glycosylation as small branched structures, and sulphated tyrosines at the C-terminus as open circles in a region rich in aromatic amino acids (shown as a double line here and also at the N-terminus). The structure of SPP-1 is also shown as an open structure with a polyaspartic acid HA-binding sequence and RGD cell-binding site. Note the numerous potential sites of phosphorylation, the sites of N-linked glycosylation, and the position of the aromatic amino acids (boxes). The nature and position of the sulphate in SPP-1 is not known. The SPARC structure is redrawn from Engel et al. [69] and reveals an a-helical HA binding domain followed by a disulphide bridged segment, a second a-helical region that is susceptible to proteolytic attack (arrow), and an EF-hand Ca2+-binding region flanked by short a-helices. The osteocalcin is redrawn from Hauschka [74] and shows an a-helix with the three "gla" groups interacting with the HA crystal, followed by a /3-turn and a second a-helix.
Sodek et al.: Non-Collagenous Proteins in Substrate Bioactivity 101 of human bone cells only the dermatan sulphate forms of biglycan and decorin were produced [19, 20], which may reflect synthesis by undifferentiated osteoblasts and fibroblasts. Based on the properties of the different classes of proteoglycans it appears that the CS-PG I, which can influence the formation of collagen fibres, may be important in organizing the collagenous matrix of bone. The ability of the homologous DS-PG II core protein, found in soft connective tissues, to bind in the gap region of the collagen fibrils [21] indicates that the CS-PG II proteoglycan of bone could inhibit HA crystal formation in the osteoid while binding CA 2+ ions to the GAG chains. The presence of CS-PG II in the demineralizing extracts indicates that as the HA crystals are formed this proteoglycan binds to the crystal surface, likely through the CS chain. In contrast, the CS-PG III appears to associate with preformed mineral, possibly through its acidic core protein, where it can influence the growth and physical properties of the crystals. Bone Sialoproteins The presence of highly sialylated glycoprotein in bone was first revealed with the isolation of a 25 kDa sialoprotein by Herring [22]. With the advent of dissociative extraction procedures and the use of proteinase inhibitors larger sialoproteins were isolated that migrate on SDS-PAGE at 65-80 kDa. Two sialoproteins have been isolated from bones of several species [14, 23-28]. These were originally termed bone sialoprotein I (BSP I) and bone sialoprotein II (BSP II) [24]. A third sialoprotein migrating at 75 kDa on SDS-PAGE has been isolated from rat bone [29] and termed bone acidic glycoprotein (BAG-75). However, it is not known whether this protein is a constituent of bones of other species. Although the majority of the sialoproteins are released from bone on demineralization, we have found small but significant amounts of both sialoproteins in the subsequent G2 extractions, and small amounts are also released together with a 35 kDa protein from bacterial collagenase digestions of the extracted tissue residue, indicating that some of the sialoprotein is bound to the collagenous matrix. Rat BSP I was purified as a 44 kDa phosphoprotein and shown to contain 12 phosphoserines and one phosphothreonine in addition to 5 O-linked oligosaccharides and an N-linked oligosaccharide [26]. The complete primary sequence of rat bone sialoprotein I was determined
from the cDNA sequence and shown to contain a polyaspartic acid segment, through which the protein is believed to bind to HA, and a (G)RGD(S) cell-attachment sequence [30], which has been shown to mediate the attachment of bone cells and fibroblasts [30-32]. In view of these properties the protein was suggested to bridge between bone cells and HA and was called osteopontin [30]. However, immunolocalization and in situ hybridization studies have shown that the protein is expressed in non-mineralized tissues including kidney and certain epithelia [33-35]. Moreover, characterization of a mouse gene (2ar) induced by tumour promoters [36] has revealed that this same protein is expressed in transformed cells and is known as the transformation-associated protein or pp69 [37]. Consequently, a more descriptive name, secreted phosphoprotein 1 (SPP1), has been introduced for this protein [38]. Although SPP-1 is expressed by non-mineralized tissues, recent studies have revealed that several forms of the protein, which differ through post-translational modification, exist, and that the protein synthesized by bone cells is characteristically sulphated [16, 39^1]. In addition, a different promoter appears to regulate the transcription of the gene in macrophages and T-lymphocytes ([42] and Q. Zhang, J.L. Wrana, and J Sodek, in preparation), and evidence for a differentially spliced gene transcript has been presented [43]. Thus, while the tissue-specific expression of this protein is regulated at the transcriptional level, the form of the protein which may be important functionally is regulated post-transcriptionally and by post-translational modifications that include glycosylation and phosphorylation as well as sulphation. By comparing the amino acid sequences obtained from various species it has been noted that the polyaspartic acid segment, the cell attachment motif, and the sites of phosphorylation are conserved, indicating their importance in the structure and function of the protein. Based on secondary structure predictions ([44] and C. Holt, personal communication), relatively short sections of a-helix and /3-loop structure are interspersed in what is anticipated to be an open, extended structure (Fig. 9.2). The number of potential phosphorylation sites far exceeds the 13 determined and almost all are in flexible, hydrophilic (surface) regions of the molecule together with the predicted sites of glycosylation. A site of thrombin susceptibility has been identified [45, 46] at a centrally located Arg-Ser together with sites of
102 Bone Proteins and Other Macromolecules "tryptic" cleavage which generate 23 and 20 kDa fragments that have been identified in EDTA extracts [28]. That SPP-1 is involved in bone formation is indicated through its expression by pre-osteoblasts and osteoblasts, by its presence in osteoid, and by its stimulated synthesis in the presence of hormones and cytokines, such as glucocorticosteroids and TGF-/3 [41, 47], which promote bone matrix formation. Although bone SPP-1 has many of the characteristics of a nucleator of HA, biosynthetic studies have failed to detect its presence in the collagenous matrix prior to mineralization [40, 41]. However, once mineralization has been initiated SPP-1 associates rapidly with the preformed mineral where it may regulate the growth of the HA crystals. Whether the specific and selective proteolytic fragmentation of the protein [28] is important for this function remains to be determined. It has also been postulated that SPP-1 functions in the attachment of osteoblasts to osteoid during bone formation [48]. The preference of SPP-1 for the vitronectin receptor [49], which is enriched in osteoclasts, together with the strong stimulation of SPP-1 synthesis by vitamin D3 [50] and retinoic acid (authors, in preparation) which promote bone resorption, indicates a role for SPP-1 in osteoclastic bone resorption [53]. The apparent concentration of SPP-1 beneath lining cells [52] and co-localization of SSP-1 and the vitronectin receptor in the clear zone of osteoclasts [53] provide strong support for this proposed role. Bone sialoprotein II has been purified from bones of several species including human [14], steer [23, 24], rabbit [54], and pig [28]. It has higher levels of sialic acid and glutamic acid and is more heavily sulphated than SPP-1, the sulphate being primarily linked to tyrosine as tyrosine sulphate [39, 55]. Notably, rabbit BSP has a keratan sulphate side chain [54], which appears to be peculiar to this species. Although BSP is also phosphorylated, based on 32PO4 incorporation the degree of phosphorylation is much lower than for SPP-1 [40, 41]. An Mr of 57K has been determined for bovine BSP by sedimentation analysis while the size of the nascent protein in several species is close to 34.5 kDa [56]. The difference reflects the high degree of glycosylation of the protein. The protein is characterized by its ability to bind strongly to HA [56] and to mediate cell attachment through an ROD site [30, 32] that recognizes the vitronectin receptor [49], features that it shares with SPP-1. Comparison of the primary sequences, obtained
from humans [27], the pig (H. Shapiro, J.L. Wrana, Q. Zhang and J. Sodek, unpublished data), and the rat [56] shows a high degree of sequence conservation. The protein is characterized by terminal segments that are enriched in aromatic amino acids, especially tyrosines which, based on consensus sequence analysis, are expected to be sulphated in the C-terminal region surrounding an ROD cell attachment motif (Fig. 9.2). In the central region of the molecule the polar amino acids are almost entirely negatively charged and include two long stretches of polyglutamic acid, which are believed to be involved in the binding to HA [56]. Several sites of serine phosphorylation and N-linked glycosylation are also conserved in the central region. The paucity of hydrophobic groups indicates that BSP, like SPP-1, is largely unfolded and flexible, as also revealed by rotary shadowing [24]. Unlike SPP-1, expression of BSP is essentially restricted to mineralized connective tissues [27, 56] and immunolocalization studies have shown that the protein is confined to the bone matrix and directly associated cells [52]. Studies on the expression of BSP in rat bone by in situ hybridization have shown a strong, specific expression by osteoblasts at sites of new bone formation [57]. Studies on the biosynthesis of BSP have demonstrated that the expression of BSP is directly linked to bone formation [41] and that, while most of the protein associates with pre-existing mineral, some protein is present in the collagenous matrix prior to mineralization [40, 41]. Notably, both BSP and SPP-1 are present in G2 extracts of foetal bone and both are released when the collagenous matrix is digested with collagenase. However, in radiolabelling experiments used to study bone formation BSP was the only sialoprotein detected in these two extracts, indicating that the association of SPP-1 with the collagenous matrix occurs after mineralization has been initiated. Since BSP is a phosphorylated and sulphated calcium-binding protein with segments of repeating acidic (glutamic acid) amino acids and associates with the collagenous matrix of bone prior to mineralization, it has all the requisites [58, 59] of a specific epitactic nucleator of HA formation. Collectively, therefore, these observations indicate a potential role for BSP in the promotion of mineralization and in the regulated growth of the HA crystals. Whether BSP also has a function in bone cell attachment is not clear. However, the immunolocalization of BSP to osteoblasts and osteocytes is indicative of such a role.
Sodek et al.: Non-Collagenous Proteins in Substrate Bioactivity 103 Bone SPARC Protein SPARC protein was first isolated from subperiosteal bovine bone and originally called osteonectin, based on its ability to bind collagen and promote HA formation in vitro [2, 60]. The protein represents up to 25% of the noncollagenous proteins in the bones of larger mammals and is also highly enriched in tubular dentine [61]. However, SPARC protein represents only a small percentage of the non-collagenous proteins in rodent bone and is difficult to detect in the dentine of the rodent incisor [62] despite the fact that the protein is highly expressed by osteoblasts [63, 64]. Although originally thought to be specific to bone, SPARC was shown to be synthesized by fibroblasts [61, 65] and subsequently shown to be expressed by a variety of cell types including parietal endoderm [66], endothelial cells [67], and epithelial cells [68]. Although the bone protein has been described as a phosphoprotein [2, 69], several studies of bone formation [16, 40, 41] have failed to observe any incorporation of [32PO4] into either rat or pig SPARC. The protein sequence for SPARC has been determined from human [70], bovine [71], and mouse [66] cDNA sequences. The sequence is highly conserved and comprises four distinct domains (Fig. 9.2). An amino terminal domain comprising two glutamate-rich segments can bind >8 Ca2 ions. On binding Ca 2 " the protein chain in rat SPARC has been suggested to undergo a coil-to-helix transition [69]. The amino acid sequence in this region is the least well conserved and is believed to bind to HA. Thus, the differences in the sequence in this region are thought to relate to the differences in the ability of the SPARC to bind to HA and, consequently, the differences in the amount of SPARC in the bones of animals of different species [3]. The second domain shows some homology to ovomucoid and is characteristically rich in disulphide bridges that stabilize the protein structure. This is followed by a segment which is susceptible to proteolysis and is predicted to form an a-helical structure. The fourth domain contains a single high-affinity EFhand Ca 2 * binding site with the characteristic helix-loop-helix structure [69]. This site is expected to be fully occupied at physiological concentrations of Ca 2 * and is, therefore, unlikely to be involved in Ca 2 * regulation. The expression of SPARC by a variety of cell types [72] and its variable amounts in bones of different species do not support the original
proposal [60] that this protein might be involved in the mineralization of bone. Moreover, biosynthetic studies in vivo [62] and in vitro [16, 40, 41] have failed to show any significant incorporation of SPARC into newly formed bone. Consequently, it would appear that SPARC is slowly incorporated into bone, with much of the protein synthesized by the bone cells being cleared through the blood. Nevertheless the binding of SPARC to HA could influence the later growth and dissolution of the crystals. However, this is not likely to be the principal function of the protein, which has been implicated in developmental processes [72], wound healing, and tissue remodelling [73]. Osteocalcin Also known as bone gamma-carboxyglutamic acid (gla) protein [74, 75], osteocalcin [76, 77] is the most completely characterized of the non-collagenous bone proteins although its precise function is yet to be established. Since recent reviews provide a detailed description of this protein [74, 77] it will only be described briefly here. Osteocalcin is a small 5.8 kDa protein that represents 1% to 2% of total bone protein in rat [75]. However, the amounts of osteocalcin vary considerably in bone of different species, with human bone containing approximately one-tenth of the amount found in rat [77]. The primary sequence of osteocalcin has been obtained from a number of species and high sequence conservation has been noted. It is characterized by the presence of gla residues that are formed by a vitamin-K-dependent posttranslational carboxylation of glutamic acid. The "gla" groups bind Ca 2+ ions in solution with a binding constant (kd) ~ 1 mM and also bind strongly to HA (kd ~10~7 M). A three-dimensional model of the osteocalcin structure has related the periodic spacing of the "gla" groups with the 5.45 A spacing of the Ca2+ ions in the 001 plane of the HA crystal [74]. Inhibition of gamma carboxylation by the vitamin K antagonist warfarin results in a dramatic reduction of osteocalcin in the bone [78]. Although hypermineralization has been observed under these conditions bone formation is not impaired. The hypermineralization may relate to the ability of osteocalcin to inhibit HA formation from supersaturated solutions of Ca 2+ and PO4. Using specific antisera to determine tissue distribution [79], it has been revealed that osteocalcin is specific to mineralized connective tissues.
104 Bone Proteins and Other Macromolecules
Fig. 9.3 Analysis of non-collagenous proteins incorporated into newly formed bone tissue formed in vitro. Cultures of stromal cells derived from adult rat bone marrow were cultured in the presence of 10~ 8 M dexamethasone for 14 days and then the proteins labelled with [35S]-methionine (lanes A) or [35SO4] (lanes B) for 24 h. The culture medium (M) was collected and the cells removed by treatment with 0.5 M ammonium hydroxide (Cell) before the tissue layer containing nodules of bone-like tissue was sequentially extracted with 4 M GuHCl (Gl), 0.5 M EDTA (E), and electrophoresis sample buffer (not shown). The collagenous tissue residue remaining was then digested with bacterial collagenase (CD) and samples of radiolabelled protein analysed using SDS-PAGE on 15% cross-linked gels. The prominence of the sialoproteins (large arrows) and proteoglycans (at the top of the gel), which can be discriminated by their incorporation of [35SO4], can be seen in the E extract. In addition, the BSP (upper large arrow) is also evident in the Gl extract and together with a 35 kDa sulphated protein (small arrow) is released from the collagenous tissue residue by collagenase digestion.
Notably, the expression of osteocalcin in rat bone is initiated after bone formation has occurred [80]. Although the presence of osteocalcin in serum has been used as a measure of bone remodelling its function appears to involve the resorption of bone. The synthesis of the protein is increased significantly when osteoblasts are stimulated with 1,25 dihydroxyvitamin D3 [81], which is known to stimulate bone resorption. Also, it has been shown that the presence of osteocalcin is necessary for osteoclastic resorption to occur [82]. Although the exact function of the protein in the resorption process is not known, a proteolytic fragment released from the C-terminal end of the molecule is chemotactic for the monocyte precursors of the osteoclasts [83].
Bone Formation on Artificial Surfaces Populations of cells derived from foetal and adult bone of various animal species will form mineralized tissue nodules with the characteristics of bone when grown on tissue culture plastic [84]. We have utilized rat bone cells isolated from foetal rat calvariae [85] and stromal cells from adult rat bone marrow [86] to study the relationship between the synthesis of non-collagenous protein and bone formation [40, 41]. In the marrow cell system, bone formation occurs in the presence of 10 8 M dexamethasone, which induces the concomitant expression of BSP and osteocalcin, reflecting the expression of these proteins by fully differentiated osteoblasts. In addition, the consti-
Sodek et al.: Non-Collagenous Proteins in Substrate Bioactivity 105 tutive expression of collagen, alkaline phosphatase, SPP-1, and SPARC proteins that are synthesized by pre-osteoblasts as well as osteoblasts is increased by dexamethasone [41]. Analysis of the tissue matrix formed in these cultures has revealed that the sialoproteins, SPP-1 and BSP, are the major constituents of the newly formed bone. In the absence of mineralization small amounts of BSP are present in the collagenous matrix, but SPP-1 could not be detected. The BSP in the pre-mineralized tissue appears to be bound tightly to the collagenous matrix and some is released with a sulphated 35 kDa protein by collagenase digestion of the collagenous matrix following demineralization (Fig. 9.3). Following mineralization, both sialoproteins bind rapidly to the newly formed mineral. In cultures of foetal calvarial bone cells similar results have been observed [40], but in this system the level of BSP synthesis relative to SPP-1 is significantly lower. Using pulse-chase protocols, it has also been shown that the sialoproteins associate with the bone mineral within 2 h of synthesis. However, some of the BSP appears transiently in the Gl compartment and accumulates in the G2 compartment. With continuous labelling in long-term cultures some bone SPARC could be detected with the mineral-associated proteins. These studies on the expression of the bonecell-derived, non-collagenous proteins of the bone matrix provide insights into the potential role of these proteins in the formation of bone tissue and also help define the phenotype of osteoblastic cells. Collectively, these proteins provide a powerful tool for studying the differentiation of osteogenic cells and for demonstrating the formation of bone tissue which will be important in evaluating cellular responses to bone biomaterials. References 1 Fisher LW, Termine JD (1985) Non-collagenous proteins influencing the local mechanisms of calcification. Clin Orthop Rel Res 200:362-385 2 Termine JD, Belcourt AB, Conn KM, Kleinman HK (1981) Mineral and collagen-binding proteins of fetal calf bone. J Biol Chem 256:10403-10408 3 Domenicucci C, Goldberg HA, Hofmann T, Isenman D, Wasi S, Sodek J (1988) Characterization of porcine osteonectin extracted from fetal calvariae. Biochem J 253:139-151 4 Niyibizi C, Eyre DR (1989) Bone type V collagen: chain composition and localization of a trypsin cleavage site. Conn Tissue Res 20:247-250 5 Fisher LW, Gehron Robey P, Tuross N, Otsuka
AS, Tepen DA, Esch FS, Shimasaki S, Termine JD (1987) The Mr 24,000 phosphoprotein from developing bone is the NH2-terminal propeptide of the a-1 chain of type I collagen. J Biol Chem 262:13457-13463 6 Goldberg HA, Maeno M, Domenicucci C, Zhang Q, Sodek J (1988) Identification of small collagenous proteins with properties of procollagen a-l(I)pN-propeptide in fetal porcine calvarial bone. Collagen Rel Res 8:187-197 7 Kuwata F, Maeno M, Yao K-L, Domenicucci C, Goldberg HA, Wasi S, Aubin JE, Sodek J (1987) Characterization of a monoclonal antibody recognizing small collagenous proteins in fetal bone. Collagen Rel Res 7:39-55 8 Fisher LW, Termine JD, Dejter AW, Whitson SW, Yanagishita M, Kimura JH, Hascall VC, Kleinman HK, Hassell JR, Nilsson B (1983) Proteoglycans of developing bone. J Biol Chem 258:6588-6594 9 Fisher LW (1985) The nature of the proteoglycans of bone. In: Butler WT (ed) The chemistry and biology of mineralized tissues. Ebsco Media, Birmingham, Alabama, pp 188-196 10 Franzen A, Heinegard D (1984) Characterization of proteoglycans from the calcified matrix of bovine bone. Biochem J 224:59-66 11 Franzen A, Heinegard D (1984) Extraction and purification of proteoglycans from mature bovine bone. Biochem J 224:47-58 12 Sato S, Rahemtulla F, Prince CW, Tomana M, Butler WT (1985) Acidic glycoproteins from bovine compact bone. Conn Tissue Res 14:51-64 13 Goldberg HA, Domenicucci C, Pringle GA, Sodek J (1988) Mineral-binding proteoglycans of fetal porcine calvarial bone. J Biol Chem 263: 12092-12101 14 Fisher LW, Hawkins GR, Tuross N, Termine JD (1987) Purification and partial characterization of small proteoglycans I and II, bone sialoproteins I and II, and osteonectin from the mineral compartment of developing human bone. J Biol Chem 262:9702-9708 15 Fisher LW, Termine JD, Young FM (1989) Deduced protein sequence of bone small proteoglycan I (biglycan) shows homology with proteoglycan II (decorin) and several non connective tissue proteins in a variety of species. J Biol. Chem 264:4571^576 16 Nagata T, Goldberg HA, Zhang Q, Domenicucci C, Sodek J (1991) Biosynthesis of bone proteins by fetal porcine calvariae in vitro. Rapid association of sulfated sialoproteins (secreted phosphoprotein I and bone sialoprotein) and chondroitin sulfate proteoglycan (CS-PG III) with bone mineral. Matrix (in press) 17 Prince CW, Rahemtulla F, Butler WT (1984) Incorporation of (35S-)sulphate into glycosaminoglycans by mineralized tissues in vivo. Biochem J 224:941
106 Bone Proteins and Other Macromolecules 18 Hunter GK, Heersche JNM, Aubin JE (1983) Isolation of three species of proteoglycan synthesized by cloned bone cells. Biochemistry 22:831-837 19 Beresford JN, Fedarko NS, Fisher LW, Midura RJ, Yanagishita M, Termine JD, Gehron Robey P (1987) Analysis of the proteoglycans synthesized by human bone cells in vitro. J Biol Chem 262:17164-17172 20 Fedarko NS, Termine JD, Young MF, Gehron Robey PG (1990) Temporal regulation of hyluronan and proteoglycan metabolism by human bone cells in vitro. J Biol Chem 265:12200-12209 21 Scott PG, Nakano T, Dodd CM, Pringle GA, Kuc IM (1989) Proteoglycans of the articular disc of the bovine temporomandibular joint. II. Low molecular weight dermatan sulphate proteoglycan. Matrix 9:284-292 22 Herring GM (1972) The organic matrix of bone. In: Bourne GH (ed) The biochemistry and physiology of bone, vol. 1. Academic Press, New York, pp 127-189 23 Fisher LW, Whitson SW, Avioli LV, Termine JD (1983) Matrix sialoprotein of developing bone. J Biol Chem 258:12723-12727 24 Franzen A, Heinegard D (1985) Isolation and characterization of two sialoproteins present only in bone calcified matrix. Biochem J 232:715-724 25 Franzen A, Heinegard D (1985) Proteoglycans and proteins of rat bone. Purification and biosynthesis of major noncollagenous macromolecules. In: Butler WT (ed) The chemistry and biology of mineralized tissue. Ebsco Media, Birmingham, Alabama, pp 132-141 26 Prince CW, Oosawa T, Butler WT, Tomana M, Bhown AS, Bhown M, Schrohenloher RE (1987) Isolation, characterization and biosynthesis of a phosphorylated glycoprotein from rat bone. J Biol Chem 262:2900-2906 27 Fisher LW, McBride OW, Termine JD, Young MF (1990) Human bone sialoprotein. Deduced protein sequence and chromosomal location. J Biol Chem 265:2347-2351 28 Zhang Q, Domenicucci C, Goldberg HA, Wrana J, Sodek J (1990) Characterization of fetal porcine bone sialoproteins, secreted phosphoprotein I (SPPI, osteopontin), bone sialoprotein, and a 23kDa glycoprotein. J Biol Chem 265:7583-7589 29 Gorski JP, Shimizu K (1988) Isolation of a new phosphorylated glycoprotein from the mineralized phase of bone that exhibits limited homology to the adhesive protein osteopontin. J Biol Chem 263:15938-15945 30 Oldberg A, Franzen A, Heinegard D (1986) Cloning and sequence analysis of rat bone sialoprotein (osteopontin) cDNA reveals an Arg-Gly-Asp cell binding sequence. Proc Natl Acad Sci USA 83:8819-8823 31 Somerman MJ, Prince CW, Sauk JJ, Foster RA,
Butler WT (1987) Mechanism of fibroblast attachment to bone extracellular matrix: Role of a 44 kilodalton bone phosphoprotein. J Bone Miner Res 2:259-265 32 Somerman MJ, Fisher LW, Foster RA, Sauk JJ (1988) Human bone sialoproteins I and II enhance fibroblast attachment in vitro. Calcif Tissue Int 43:50-53 33 Mark MP, Prince CW, Oosawa T, Gay S, Bronckers ALJJ, Butler WT (1987) Immunohistochemical demonstration of a 44 kD phosphoprotein in developing rat bones. J Histochem Cytochem 35:707-715 34 Mark MP, Prince CW, Gay S, Austin RL, Butler WT (1988) 44-kDal bone phosphoprotein (osteopontin) antigenicity at ectopic sites in newborn rats: kidney and nervous tissues. Cell Tissue Res 251:23-30 35 Nomura S, Wills AJ, Edwards DR, Heath JK, Hogan BLM (1988) Developmental expression of 2ar (osteopontin) and SPARC (osteonectin) RNA as revealed by in situ hybridization. J Cell Biol 106:441^50 36 Craig AM, Smith JH, Denhardt DT (1989) Osteopontin, a transformation-associated cell adhesion phosphoprotein, is induced by 12-O-tetradecanoylphorbol 13-acetate in mouse epidermis. J Biol Chem 264:9682-9689 37 Senger DR, Perruzzi CA, Gracey CF, Papadopoulos A, Tenen DG (1988) Secreted phosphoprotein associated with neoplastic transformation. Close homology with plasma proteins cleaved during blood coagulation. Cancer Res 48:5770-5774 38 Fet V, Dickinson ME, Hogan BLM (1989) Localization of the mouse gene for secreted phosphoprotein 1 (Sppl) (2ar, osteopontin, bone sialoprotein 1, 44-kDa bone phosphoprotein, tumor-secreted phosphoprotein) to chromosome 5, closely linked to Ric (Rickettsia resistance). Genomics 5:375-377 39 Nagata T, Todescan R, Goldberg HA, Zhang Q, Sodek J (1989) Sulfation of secreted phosphoprotein I (SPPI, Osteopontin) is associated with mineralized tissue formation. Biochem Biophys Res Commun 165:234-240 40 Nagata T, Bellows CG, Kasugai S, Butler WT, Sodek J (1991) Biosynthesis of bone proteins, SPP-1 (secreted phosphoprotein-1, osteopontin), BSP (bone sialoprotein) and SPARC (osteonectin) in association with mineralized tissue formation by fetal rat calvarial cells in culture. Biochem J 274:513-520 41 Kasugai S, Todescan RJr, Nagata T, Yao K-L, Butler WT, Sodek J (1991) Expression of bone matrix proteins associated with mineralized bone tissue formation by adult rat bone marrow cells in vitro. Inductive effects of dexamethasone on the osteoblastic phenotype. J Cell Physiol (in press) 42 Miyazaki Y, Setoguchi M, Yoshida S, Higuchi Y,
Sodek et al.: Non-Collagenous Proteins in Substrate Bioactivity 107 Akizuki S, Yamamoto S (1990) The mouse osteopontin gene. Expression in monocytic lineages and complete nucleotide sequence. J Biol Chem 265:14432-14438 43 Young MF, Kerr JM, Termine JD, Wewer UM, Wang MG, McBride OW, Fisher LW (1990) cDNA cloning, mRNA distribution and heterogeneity, chromosomal localization, and RPLF analysis of human osteopontin (OPN). Genomics 7:491-502 44 Prince CW (1989) Secondary structure predictions for rat osteopontin. Conn Tissue Res 21:15-20 45 Senger DR, Perruzzi CA, Papadopoulos A (1989) Elevated expression of secreted phosphoprotein 1 (osteopontin, 2ar) as a consequence of neoplastic transformation. Anticancer Res 9:1291-1299 46 Senger DR, Perruzzi CA, Papadopoulos A, Tenen DG (1989) Purification of a human milk protein closely similar to tumor-secreted phosphoproteins and osteopontin. Biochem Biophys Acta 996:43-48 47 Wrana JL, Kubota T, Zhang Q, Overall CM, Aubin JE, Butler WT. Sodek J (1991) Regulation of transformation-sensitive secreted phosphoprotein (SPP-1/Osteopontin) expression by transforming growth factor-p. Comparisons with SPARC expression. Biochem J 273:523-531 48 Butler WT (1989) The nature and significance osteopontin. Conn Tissue Res 23:123-136 49 Oldberg A, Franzen A, Heinegard D, Pierschbacher M, Ruoslahti E (1988) Identification of a bone sialoprotein receptor in osteosarcoma cells. J Biol Chem 263:19433-19436 50 Prince CW, Butler WT (1987) 1,25-Dihydroxyvitamin D3 regulates the biosynthesis of osteopontin, a bone-derived cell attachment protein, in clonal osteoblast-like osteosarcoma cells. Collagen Rel Res 7:305-313 51 Oldberg A, Jirskog-Hed B, Axelsson S, Heinegard D (1989) Regulation of bone sialoprotein mRNA by steroid hormones. J Cell Biol 109:3183-3186 52 Chen J, Zhang Q, McCulloch CAG, Sodek J (1991) Immunohistochemical localization of bone sialoprotein (BSP) in fetal porcine bone tissues: comparisons with secreted phosphoprotein 1 (SPP1/osteopontin) and SPARC (osteonectin). Histochem J (in press) 53 Reinholt FP, Hultenby K, Oldberg A, Heinegard D (1990) Osteopontin - a possible anchor of osteoclasts to bone. Proc Natl Acad Sci USA 87:4473-4475 54 Kinne RW, Fisher LW (1987) Keratan sulfate proteoglycan in rabbit compact bone is bone sialoprotein II. J Biol Chem 262:10206-10211 55 Ecarot-Charrier B, Bouchard F, Delloye C (1989) Bone sialoprotein II synthesized by cultured osteoblasts contains tyrosine sulfate. J Biol Chem 264:20049-20053 56 Oldberg A, Franzen A, Heinegard D (1988) The
primary structure of a cell-binding bone sialoprotein. J Biol Chem 263:19433-19436 57 Chen J, Shapiro HS, Wrana JL, Reimers S, Heersche JNM, Sodek J (1991) Localization of bone sialoprotein (BSP) expression to the site of mineralized tissue formation in fetal rat tissues by in situ hybridization. Matrix (in press) 58 Addadi L, Moradian H, Shay E, Maroudas NG, Weiner S (1987) A chemical model for the cooperation of sulfates and carboxylates in calcite crystal nucleation: relevance to biomineralization. Proc Natl Acad Sci USA 84:2732-2736 59 Glimcher MJ (1989) Mechanisms of calcification in bone: role of collagen fibrils and collagenphosphoprotein complexes in vitro and in vivo. Anat Rec 224:139-153 60 Termine JD, Kleinman HK, Whitson SW, Conn KM, McGarvey ML, Martin GR (1981) Osteonectin, a bone-specific protein linking mineral to collagen. Cell 26:99-105 61 Tung PS, Domenicucci C, Wasi S, Sodek J (1985) Specific immunohistochemical localization of osteonectin and collagen types I and III in fetal and adult porcine dental tissues. J Histochem Cytochem 33:531-540 62 Zung P, Domenicucci C, Wasi S, Kuwata F, Sodek J (1986) Osteonectin is a minor component of mineralized connective tissues in rat. Biochem Cell Biol 64:356-362 63 Kubota T, Zhang Q, Wrana JL, Ber R, Aubin JE, Butler WT, Sodek J (1989) Multiple forms of SPP1 (secreted phosphoprotein, osteopontin) synthesized by normal and transformed rat bone cell populations: regulation by TGF-(3. Biochem Biophys Res Commun 162:1453-1459 64 Wrana JL, Maeno M, Hawrylyshyn B, Yao K-L, Domenicucci C, Sodek J (1988) Differential effects of transforming growth factor-(3 on the synthesis of extracellular matrix proteins by normal fetal rat calvarial bone cell populations. J Cell Biol 106:915-924 65 Wasi S, Otsuka K, Yao K-L, Tung PS, Aubin JE, Sodek J, Termine JD (1984) An osteonecton-like protein in porcine periodontal ligament and its synthesis by periodontal ligament fibroblasts. Can J Biochem Cell Biol 62:470-478 66 Mason IJ, Taylor A, Williams JG, Sage H, Hogan BLM (1986) Evidence from molecular cloning that SPARC, a major product of mouse embryo parietal endoderm, is related to an endothelial cell culture shock glycoprotein of Mr 43000. EMBO J 5:1465-1472 67 Sage H, Johnson C, Bornstein P (1984) Characterization of a novel serum albumin-binding glycoprotein secreted by endothelial cells in culture. J Biol Chem 259:3993-4007 68 Mann K, Deutzmann R, Paulsson M, Timpl R (1987) Solubilization of protein BM-40 from a basement membrane tumor with chelating agents
108 Bone Proteins and Other Macromolecules
69
70
71
72
73
74
75
76
and evidence for its identity with osteonectin and SPARC. FEES Lett 218:167-172 Engel J, Taylor W, Paulsson M, Sage H, Hogan B (1987) Calcium binding domains and calciuminduced conformational transition of SPARC/BM40/osteonectin, an extracellular glycoprotein expressed in mineralized and nonmineralized tissues. Biochemistry 26:6958-6965 Lankat-Buttgereit B, Mann K, Deutzmann R, Timpl R, Krieg T (1988) Cloning and complete amino acid sequences of human and murine basement membrane protein BM-40 (SPARC, osteonectin). FEBS Lett 236:352-356 Bolander M, Young MF, Fisher LW, Yamada Y, Termine JD (1988) Osteonectin cDNA sequence reveals potential binding regions for calcium and hydroxyapatite and shows homologies with both a basement membrane protein (SPARC) and a serine proteinase inhibitor (ovomucoid). Proc Natl Acad Sci USA 85:2919-2923 Holland P, Harper S, McVey J, Hogan BLM (1987) In vivo expression of mRNA for the Ca + + binding protein SPARC (osteonectin) revealed by in situ hybridization. J Cell Biol 105:473^82 Salonen J, Domenicucci C, Goldberg HA, Sodek J (1990) Immunohistochemical localization of SPARC (osteonectin) and dentured collagen and their relationship to remodeling in rat dental tissues. Arch Oral Biol 35:337-346 Hauschka PV (1985) Osteocalcin and its functional domains. In: Butler WT (ed) The chemistry and biology of mineralized tissues. Ebsco Media, Birmingham, Alabama, pp 149-158 Hauschka PV, Lian JB, Gallop PM (1975) Direct identification of the calcium-binding amino acid ?carboxyglutamate, in mineralized tissue. Proc Natl Acad Sci USA 72:3925-3929 Price PA, Otsuka AS, Poser JW, Kristaponis J, Raman N (1976) Characterization of a gamma-
carboxyglutamic acid-containing protein from bone. Proc Natl Acad Sci USA 73:1447-1451 77 Price P (1983) Osteocalcin. In: Peck WA (ed) Bone and mineral research annual 1. Excerpta Medica, Princeton, pp 157-190 78 Price PA, Williamson MK (1981) Effects of warfarin on bone. Studies on the vitamin K-dependent protein in rat bone. J Biol Chem 256:1275412759 79 Mark MP, Butler WT, Prince CW, Finkelman RD, Ruch J-V (1988) Developmental expression of 44 kDa bone phosphoprotein (osteopontin) and bone gamma-carboxyglutamic acid (Gla)-containing protein (osteocalcin) in calcifying tissues of rat. Differentiation 37:123-136 80 Yoon K, Buenaga R, Rodan GA (1987) Tissue specificity and developmental expression of rat osteopontin. Biochem Biophys Res Commun 148:1129-1136 81 Price PA, Baukol SA (1980) 1,25 Dihydroxyvitamin D3 increases synthesis of the vitamin-K dependent bone protein by osteosarcoma cells. J Biol Chem 255:11660-11663 82 Glowacki J, Rey C, Cox K, Lian J (1989) Effects of bone matrix components on osteoclast differentiation. Conn Tissue Res 21:121-129 83 Mundy GR, Poser JW (1983) Chemotactic activity of the gamma-carboxyglutamic acid containing protein in bone. Calcif Tissue Int 35:164-168 84 Sodek J, Berkman FA (1987) Bone cell cultures. Methods Enzymol 145:303-324 85 Bellows CG, Aubin JE, Heersche JNM, Antosz MA (1986) Mineralized bone nodules formed in vitro from enzymatically released rat calvaria cell populations. Calcif Tissue Int 38:143-154 86 Maniatopoulos C, Sodek J, Melcher AH (1988) Bone formation in vitro by stromal cells obtained from bone marrow of young adult rats. Cell Tissue Res 254:317-330
DISCUSSION Gross: Dr. Sodek, we reported some years ago an experiment using antibodies provided by Dr. Termine and Larry Fisher in the NIH lab. We tried to find out if there are differences between the non-bonding and the bonding glass ceramics we used in our experiments, and we couldn't find differences between the two species. So the question is: Why do you think the bone sialoprotein or the other proteins are of significance for bonding or non-bonding of implant material to bone? Sodek: I am not sure exactly what you are saying in that you didn't see a difference, but I will answer the second part of your question first. My
impression is that a protein like the osteopontin, secreted phosphoprotein-I, which has a high affinity for HA or is secreted by bone cells and is an attachment protein, could likely mediate the attachment of bone cells to these bioactive surfaces that have HA associated with them. The protein may also bind with other mineral groups. I don't know if that's really what you mean when you say you couldn't see any difference in your experiments. Do you mean you couldn't see any protein binding to any of these surfaces with the antibodies? Gross: Yes. If there are bone cells around an implant the cells make the protein, but this does
Sodek et al.: Non-Collagenous Proteins in Substrate Bioactivity 109 not explain the bonding or the non-bonding mechanisms. Sodek: For what reason? I don't understand. Gross: In non-bonding there are no bone cells. Ricci: I am a little unclear from the staining in your histology whether the sialoprotein is showing up in the osteoid or in the mineralized tissue. The fact that it's extracted with mineral indicates that it would be attached to the mineral, but do you see it present only in the mineralized tissue or do you see it in the non-mineralized osteoid? Sodek: Other than in the high-resolution section that I showed of the osteopontin-staining osteoid it is not terribly clear. Bill Butler's lab has shown better staining for osteopontin in osteoid tissue, but when we do extractions of the protein from the tissue we don't get any appreciable amount of osteopontin coming out of bone until we demineralize, and I am wondering if what he's seeing in the osteoid is actually protein that is bound to early crystallites that are being formed in the osteoid material rather than it being part of the organic matrix. Ricci: How does this relate - you mentioned about osteoclastic resorption - how does it relate to osteoclastic resorption? Sodek: This is a current hypothesis that Dick Heinegard has put forward based on the preference of the osteopontin to bind to the vitronectin receptor, which is enriched in osteoclasts, that it may mediate the attachment of osteoclasts to bone surfaces, and those surfaces that I was showing there are likely to be surfaces where osteoclasts could resorb. Ricci: Thank you. LeGeros: We have always observed that the carbonate apatite microcrystals that form in relation to the ceramic calcium phosphate materials are always intimately associated with an organic matrix whether or not the materials were implanted in bone sites or non-bone sites and even after suspension in serum. I was wondering if you could speculate on what kind of organic matrix this would be. Sodek: Not knowing your system it would be very dangerous for me to speculate. I don't know what cell types you have there. LeGeros: No, no. I am talking about implants that we put in bony sites, not in non-bony sites, and they are always associated with an organic matrix. Sodek: There are so many cell types in those systems and so many proteins that are present that it would be pure guesswork, but there are procedures available for trying to analyse what proteins
you do have there. There are antibodies certainly to the bone proteins that one can use to determine if bone proteins are involved. Bolander: Dr. Sodek, you mentioned that you had discovered another small proteoglycan, a third proteoglycan, and I think there is a recent report by Paul Price about a protein called matrix gla protein, which has been I think recently discovered in bone. Do you think that there are other proteins yet to be discovered in bone which have not yet been identified? Sodek: I think essentially all the major proteins have pretty well been identified - of the ones produced by bone cells. Bolander: Is there a clear concept of how these fit together? You mentioned osteopontin. Is it clear how the proteins interact in the matrix? Sodek: No, but we have some information from biosynthetic studies. We have analysed for what proteins are present in collagenous matrix in the absence of mineralization, and that matrix will mineralize given the right conditions. But based on those criteria the only proteins that we have been able to find - of the non-collagenous proteins - present in the collagenous matrix are bone sialoprotein, BSP, and a 35 kDa protein that comes out of the matrix on collagenous digestion that is also sulphated just as the sialoproteins are, Hench: First of all, your summary figure on the various bone proteoglycans, for those of us that don't work intimately in that area, is enormously helpful. My question is: Is it known what the bonding sites are between the HA crystals and the bone sialoproteins, and second, do those bonding sites differ if you have a carbonate apatite or fluoroapatite or just plain apatite? Sodek: I might be able to answer part of your question. It's believed that the polyaspartic acid groups in the osteopontin are involved in the binding to HA and polyglutamic acids in the bone sialoprotein. There are two major stretches of polyglutamic acid in the bone sialoprotein. That was the reason for my asking the question earlier about the structure of the mollusk proteins because I suspect these are very similar because both of those proteins are sulphated, and the bone sialoprotein in particular has a lot of sulphate associated with it. When we label with sulphate in boneforming systems as much as 50% of the sulphate actually goes into the sialoproteins with the other 50% going into proteoglycans. So when one considers how much sulphate there is on a proteoglycan in terms of protein ratios there is much more sialoprotein there than there is proteoglycan.
110 Bone Proteins and Other Macromolecules Hench: One of the things that came to mind while reading your paper last night was an observation we made many years ago and which we had forgotten because there was no way to fit it in with any of the current thinking of 10 or 15 years ago. We did an EDX analysis across the bonebioglass bridge using a special detector for the TEM that was designed to make it possible to do low-Z material analysis. And we saw some concentration of sulphur in addition to the calcium and phosphorus in that bridge. With the data points being taken at O.l-jum intervals across, there was a particular concentration region of high sulphur in that bone bridge. I gather from the conclusions for the bone proteoglycans that sulphonated groups might be associated with apatitebonding sites and that therefore these EDX analyses might not be irrational. It may be reasonable to see a high concentration of sulphur at a critical point of a bone-bridge bond between one of the bioglass samples. Is that drawing too much conclusion? Sodek: I think it's an interesting hypothesis, and that's why I want to be clear with the question that Dr. Gross was trying to ask me or results that he was trying to give me. In terms of what kinds of proteins anyone has actually seen in these surfaces, have people looked? You mentioned sulphate. The sulphate could come from many different kinds of molecules, and it would be purely guesswork at this stage, unless you had a more specific way of looking at what that sulphate is attached to. I think it's dangerous at this time
really to speculate, but Dr. Gross was trying to say that he had used antibodies to try to show what materials were on these surfaces. Triffitt: I think there are so many materials, biomolecules present in these sites that it's impossible to identify them just on a sulphate analysis. Could I ask Dr. Sodek if he might comment on the particular proteins that he spoke about. I wonder if he might indicate whether some may be more important than others. Sodek: It's a difficult question to answer because when you don't really know what is important for the system it's very difficult to speculate what would be good to design the system with, and we don't know an awful lot about these proteins yet. We have an idea of some of the properties. We certainly know that they bind very strongly to HA, and, as I mentioned, two of them are cell-attachment proteins. John Sauk has talked about this as well. I think that it might be worthwhile looking for the involvement of those proteins in mediating attachment of bone cells to implant surfaces. Whether they prove to be good mediators of that interaction or not remains to be seen. But those are the ones that I suspect could have a role because what we do know about those proteins, and also the one proteoglycan that I spoke about, is that as soon as they are synthesized they essentially go to the mineral surface; they don't hang around in the osteoid material. If you look at it kinetically, most of those proteins are associated with the mineral within about 1 h or 2 h after synthesis.
10 Role of Adhesive Proteins and Integrins in Bone and Ligament Cell Behavior at the Material Surface /./. Sank, C.L. Van Kampen, and M.J. Somerman
Cellular interactions at material surfaces are complex molecular processes that just now are being scrutinized. These associations closely resemble cell-substratum relationships of in vitro cell culture systems in that they are mediated through several different families of receptors at the cell-polymer surface interface. In addition to directing cell adhesion to specific extracellular matrices and ligands on adjacent cells, these receptor interactions modulate various aspects of cell behavior including growth, differentiation, protein production, and stress tolerance [1-3]. The initial event in the utilization of almost all biomedical polymers involves adsorption of blood and plasma proteins onto the material surfaces [4]. These initial events are followed by temporal changes in type and concentration of proteins adsorbed. Furthermore, cellular and plasma proteins may be concurrently adsorbing onto and desorbing from surfaces with similar or different rates resulting in distinct surface concentrations of proteins at various points in time [5]. Consequently, several families of adhesion receptors may be involved in the interaction between material surfaces and adhering cells. Included among these receptors are the heterodimeric molecules which function as cell adhesion receptors, the integrins; as well as a group of non-integrin cell substratum receptors, for which a uniform terminology has yet to be established. The latter include: (1) an 85-kDa membrane protein that binds hyaluronic acid [6]; (2) the family of 67-kDa proteins which bind collagen, elastin, and laminin [7]; and (3) heparan sulfate proteoglycans. These proteoglycans participate in several cellular events, such as the formation of focal contacts and modification of the extracellular matrix, as well as serving as reservoirs for growth factors andcytokines [8-11]. Chemically, integrin molecules are heterodimers of noncovalently associated alpha (a) and beta (/3) subunits [12-14]. These molecules are defined
structurally by the presence of distinct, but homologous /3 subunits termed j8,, /32, /33 [13, 14], and recently /34 [15] and B5(x) [16]. Each /3 subunit associates with one of a number of a subunits [13, 14]. The organization of integrins into subfamilybased beta subunits has been proposed by Hynes [13] and has been modified recently by Albelda and Buck [1]. In essence, the /8j subfamily is composed of a minimum of 6 a-related complexes, each having a common /3j chain but a distinct a chain. Each of these heterodimers promotes different ligand binding specificity and affinity. This subfamily consists of those receptors that bind fibronectin, laminin, and collagens. Those receptors with unique specificity are the a 5 pj receptor for fibronectin, the a^ fibronectin receptor, and the a^ laminin receptor. The remaining members of this subfamily can bind to more than one ligand depending on the cell in which they are expressed [1, 14]. The second subfamily of receptors having a P2 subunit and several related a complexes mediate leukocyte cell interactions [17]. Receptors possessing a p3 subunit have also been termed "cytoadhesins." This subfamily of receptors includes the a^pj complex and the