Molecular Geomicrobiology 0939950715, 9780939950713

Citation preview

MOLECULAR GEOMICROBIOLOGY 59

Reviews in Mineralogy and Geochemistry

59

FROM THE SERIES EDITOR This volume was prepared in advance of a short course entitled “Molecular Geomicrobiology.” The short course, sponsored by the Mineralogical Society of America, the Geochemical Society, the US Department of Energy, and NASA Astrobiology Institute, was held at the University of California, Berkeley, December 3-4, 2005 prior to the fall AGU meeting in San Francisco, California. Errata (if any) can be found at the MSA website www.minsocam.org. Jodi J. Rosso, Series Editor West Richland, Washington October 2005

PREFACE As geomicrobiologists, we seek to understand how some of nature’s most complex systems work, yet the very complexity we seek to understand has placed many of the insights out of reach. Recent advances in cultivation methodologies, the development of ultrahigh throughput DNA sequencing capabilities, and new methods to assay gene expression and protein function open the way for rapid progress. In the eight years since the irst Geomicrobiology volume (Geomicrobiology: Interactions between microbes and minerals; volume 35 in this series) we have transformed into scientists working hand in hand with biochemists, molecular biologists, genome scientists, analytical chemists, and even physicists to reveal the most fundamental molecular-scale underpinnings of biogeochemical systems. Through synthesis achieved by integration of diverse perspectives, skills, and interests, we have begun to learn how organisms mediate chemical transformations, the ways in which the environment determines the architecture of microbial communities, and the interplay between evolution and selection that shapes the biodiversity of the planet. This volume presents chapters written by leaders in the rapidly maturing ield we refer to as molecular geomicrobiology. Most of them are relatively young researchers who share their approaches and insights and provide pointers to exciting areas ripe for new advances. This volume ties together themes common to environmental microbiology, earth science, and astrobiology. The resesarch presented here, the associated short course, and the volume production were supported by funding from many sources, notably the Mineralogical Society of America, the Geochemical Society, the US Department of Energy Chemical Sciences Program and the NASA Astrobiology Institute. We thank Jodi Rosso for her editorial contributions. October 2005 Jillian F. Ban¿eld Javiera Cervini-Silva Kenneth H. Nealson

1529-6466/05/0059-0000$05.00

DOI: 10.2138/rmg.2005.59.0

Reviews in Mineralogy & Geochemistry Vol. 59, pp. 1-7, 2005 Copyright © Mineralogical Society of America

1

The Search for a Molecular-Level Understanding of the Processes that Underpin the Earth’s Biogeochemical Cycles Jillian F. Banfield, Gene W. Tyson, Eric E. Allen, Rachel J. Whitaker Department of Earth and Planetary Sciences and Department of Environmental Science, Policy, and Management, University of California Berkeley Berkeley California, 94720-4767, U.S.A. [email protected]

CHARACTERIZING BIOGEOCHEMICAL SYSTEMS Evidence of connections between microbial activity and the Earth’s biogeochemical cycles is all around us, and motivates our interest in the mechanisms of microbial transformations, their rates, and the distribution of microbial activities across environment types and over Earth history. In general, the approach to investigating a geomicrobiological process begins with biological and geochemical characterization of the environment of interest. Geochemical characteristics constrain available metabolisms (e.g., McCollom and Shock 1997) and patterns can reveal processes not recognized initially to be microbially mediated. The membership of microbial communities can be assayed through cultivation and cultivation-independent methods. However, this task is not without its challenges. There is little consensus about the ways in which organisms should be grouped into relevant ecological units such as species (Gevers et al. 2005). Even using standard classiication techniques, the extent of microbial diversity appears vast, and recent analyses suggest that current estimates may tremendously under predict the amount of genetic diversity in the biosphere. In addition, a single organism type may contain far more genes than expected based on genomic sequencing of an isolate of that species because species populations can exhibit internal heterogeneity. Thus, it seems that comprehensive characterization of the microbial membership of an environment over space and time is a problem of almost incomprehensible magnitude. Furthermore, microbial census taking is only the irst step. Beyond documenting the assemblage of organisms present, we need to know how they are distributed, what are they doing, how they are doing it, and the ways that their activities impact the physical and chemical characteristics of their surroundings. In the near future, the only systems in which it will be plausible to tackle the level of characterization required for relatively comprehensive analyses are the simplest ones. For example, in samples of relatively low geochemical and biological complexity, it is now possible to reconstruct the genomes of the dominant organism types using environmental genomic approaches. These data can be used to classify the individual members of a community into groups based on their genome structure and gene content and to infer some aspects of their metabolism (Tyson et al. 2004). In more complex environments, analyses are likely to be restricted to studies of a subset of biogeochemical processes or organism types. In very complex environments, genomic characterization of a community may be restricted to functional proiling (Tringe et al. 2005). Other proiling using methods such as multi-locus sequence analysis (e.g., Whitaker et al. 2003) may be useful to sample speciic genes from 1529-6466/05/0059-0001$05.00

DOI: 10.2138/rmg.2005.59.1

2

Banęeld, Tyson, Allen, Whitaker

a much larger number of individuals in a population and to resolve microbial distribution patterns as a function of geochemical conditions and over time. Information about community structure and metabolic potential must be interpreted in the context of ine-scale measurements of temperature, pH, metal and oxygen concentrations, redox potential, etc. A current challenge in uniting geochemical parameters with community structure observations is to document community variation over scales comparable to those over which geochemical gradients occur. The discrepancies in sampling volumes for biological and geochemical characterization may be resolved when new methods such as single cell genomic analysis from environmental samples are combined with micron-scale methods such as microelectrode chemical analyses. Through such approaches it may be possible to decipher the extent to which geochemical gradients structure microbial populations and possibly identify important selective pressures. One of the most severe limitations in our understanding of biogeochemical systems arises from lack of information about the genetic basis for metabolic pathways involved in tasks such as iron oxidation and biomineral formation. Such gaps in our knowledge also restrict derivation of functional inferences from newly acquired genome sequences. A key strategy for deciphering how metabolic tasks are conducted and how these pathways are regulated begins with the development of genetically tractable model organisms (see Newman and Gralnick 2005). Once the sequences of genes involved in processes of interest are known, comparative analyses may provide clues to the evolutionary processes that gave rise to the function (see Raymond 2005). Puriied gene products can be assayed to determine substrate speciicity, intermediate products, and reaction kinetics, and this information can be incorporated into models for natural systems. Through reconstructions of complex pathways such as those involved in electron transport it is possible to learn how organisms derive energy and shape their surroundings (see DiChristina et al. 2005). For example, the electron transport chain is central to the catalysis of redox reactions involving geochemically abundant chemical species such as nitrate and iron, and can profoundly change the forms of nitrogen and iron in the environment(see Kappler and Straub 2005). Another example involves the microbial production and excretion of speciically tailored iron-binding molecules referred to as siderophores in order address the challenge of iron limitation (see Kraemer et al. 2005). In cases such as this, genetic, genomic, and biochemical approaches can be complemented by high-resolution data for minerals that can now be acquired using spectroscopic methods to reveal the ways in which organic ligands bind to, and interact with, minerals at the molecular level. Spectroscopic methods that provide molecular-scale insights include nano-scale secondary ion mass spectrometry (Guerquin-Kern et al. 2005), micro-scale infrared spectroscopy, nuclear magnetic resonance spectroscopy, and other X-ray-based analytical and imaging methods (see Gilbert et al. 2005). These approaches enable identiication of tiny quantities of organic compounds contained in biomineral structures (e.g., Chan et al. 2004) and can reveal surface coordination geometries and reaction mechanisms. Spectroscopic approaches may also provide insights in a wide variety of other systems, including those involving organic molecules and mineral surfaces that may have had special relevance to prebiotic synthesis early in Earth history (see Ferris 2005). Spectroscopic methods also yield isotopic information and can be used to constrain reaction dynamics. The integration of high-resolution, spatially and chemically resolved data from natural inorganic and biological materials is only in a relatively early stage. Such syntheses will certainly generate fundamental new insights into how microorganisms shape their surroundings. Molecularly resolved spectroscopic approaches are also essential tools in efforts to deine the structure and constrain mechanisms of reactions involving the mineral products of microbial metabolism. Mineral byproducts of anaerobic respiration (e.g., uranium oxide,

Molecular-Level Processes in Biogeochemical Cycles

3

zinc sulide) are frequently no more than a few nanometers in diameter (e.g., Moreau et al. 2004). As a consequence of their ultra-small size, these particles may exhibit structures and properties (possibly including redox potential) that depend on the particle diameter (see Gilbert and Banield 2005). Furthermore, the particle size and size and organization of colloidal aggregates of these particles may depend on the microbial metabolic rate, which itself depends upon solution chemistry (Jin and Bethke 2005). Recent advances in methods for study of nanoparticle structure and properties and for analysis of metabolism make it possible to analyze such connections and predict outcomes in biogeochemical systems.

MOLECULAR GEOMICROBIOLOGY: OPPORTUNITIES AND CHALLENGES One of the goals of molecular geomicrobiology is to understand how microorganisms function in, respond to, and shape their environments. Complete microbial genomes reveal an organism’s metabolic potential and provide information about genome structure and gene content (see Nelson and Methé 2005). Thus, they provide an important starting point for biogeochemical analyses. However, it should be noted that the complete genomes derive from cultivated strains, so each analysis represents only a snapshot of the genome of a single representative of a population. More importantly, only a small subset of organisms can be cultivated. Despite many successful predictions of new functional capabilities from genomic data (Ramesh et al. 2005), there are innumerable cases where the function of predicted genes cannot be inferred because the gene bears no signiicant similarity to one for which a function can be assigned. This is not a small problem (Roberts et al. 2005). Typically 30-50% of genes in sequenced genomes are annotated as either hypothetical (a predicted gene with no similarity to any other gene in the databases) or conserved hypothetical (a predicted gene that bears signiicant sequence similarity to genes in the database, but for which no function has been determined). Given that no more than a few tens of novel genes are ascribed functions per year as the result of typical biochemical and genetic experiments (M. Thelen, pers. comm.), complete functional analysis based on genomic information is a long way off, even for the best studied microbes (e.g., E. coli). This may be one of the largest roadblocks facing geomicrobiologists. Genomic data from isolates have been used extensively in comparative studies for functional and evolutionary analyses (see Nelson and Methé 2005; Raymond 2005). These studies have provided a wealth of evidence that indicates that lateral gene transfer is an important force in genome evolution. It is apparent that genes are moved between organisms that are distantly related, at least occasionally, and that the frequency of transfer increases with decreasing evolutionary distance (e.g., between different species of the same genera; Gogarten and Townsend 2005). Remaining questions relate to the mode and frequency of genetic exchange within populations of very closely related individuals. Are genes taken up from naked DNA or acquired from phage or other organisms with high frequency, only to be discarded in subsequent generations because they rarely, if ever, confer an adaptive advantage? Or are such events still quite rare but their consequences often profound? Genomic data from coexisting members of natural populations may throw light on this question. Genomic analyses can reveal how metabolic traits distribute across lineages and environment types, the rates and mechanisms by which this movement occurs, and the roles that lateral gene transfer can play in biogeochemical cycling. For example, a recent important discovery is that photosystem II genes are found in phage genomes (Lindell et al. 2005). These genes impact levels of a protein involved in photosynthesis, implying that phage can inluence

4

Banęeld, Tyson, Allen, Whitaker

or even control one of the key steps in the ocean carbon cycle. Other laterally transferred genes with biogeochemical signiicance include those that are spread on plasmids to confer the ability to degrade speciic hydrocarbon contaminants (Wilson et al. 2003; Springail and Top 2004) or resistance to antibiotics or heavy metals (Coombs and Barkay 2004; Nielsen and Townsend 2004). The mobility of genes between lineages has presented a serious challenge to the species concept because the extent to which members of the same “species” share a common evolutionary history is unclear. It is our view that both the genetic characteristics that unite groups and the processes that subdivide them to yield divergent phenotypes are best evaluated through analysis of population genetic heterogeneity. With such information in hand, we may be better positioned to meaningfully delineate organism groups (“species”) with likely ecological relevance. In addition to having the potential to tell us about evolutionary processes that occur over small times scales, environmental genomic data can reveal metabolic traits of organisms that have previously deied cultivation. This information may itself lead to growth of the organism in the laboratory, opening the way for more comprehensive physiological characterization. For example, recently Tyson et al. (2005) designed a cultivation approach based on the recognition that only a single organism type in a community was capable of nitrogen ixation. Population genomic data also reveal differences in metabolic potential that may be responsible for adaptation of otherwise closely related organisms to subtly different environment types. Whitaker and Banield (2005) describe how methods developed for comparative analysis of (clonal) isolate genomes can be used to identify differences in gene content and sequence amongst closely related individuals and the ways in which population genetic methods may be adapted from macroscopic biology to reveal patterns of selection genome wide. With genomic information for the dominant organisms in a community in hand (whether obtained from isolates or natural populations) it is possible to monitor in situ microbial activity levels in multi-species consortia. For example, gene sequence information can be used to design microarrays to detect mRNA transcripts (see Nelson and Methé 2005) or to identify peptides via mass-based measurements (e.g., mass spectrometry). Functional analyses may reveal the relative contributions of speciic organisms or strains to geochemical transformations such as carbon ixation, nitrogen ixation, and metal reduction at a given time, and to deduce how this changes over time. An important limitation for use of genomic data in functional analyses arises if these data do not accurately or fully capture the genetic potential of the community. For example, Ram et al. (2005) used genomic data from one sample to identify proteins in a closely related (but not identical) bioilm sample. Although over 2,000 proteins derived from all of the dominant organism types were detected, it is certain that an important subset were under-detected or missed because the predicted peptides differed from the observed peptides in their amino acid sequence. Thus, full analyses of function in microbial consortia will beneit from comprehensive genomic datasets that capture the full range of sequence types present in a community. Even if complete genomic data are available for one environment type, it is unclear whether they will enable functional analyses in a similar environment due to site-to-site genetic variability. With the exception of studies of organisms that cause plant and animal disease (Bhattacharyya et al. 2002; Holden et al. 2004), population-level differences in the same environment type in geographically separated locations have been little investigated (Whitaker et al. 2003; Escobar-Paramo et al. 2005). Consequently, it is dificult to predict whether the challenges associated with functional analyses of natural communities will be best addressed via better genomic databases, mass spectrometry method development, or both.

Molecular-Level Processes in Biogeochemical Cycles

5

Environmental genomic studies have the potential to constrain the rates of inter-site microbial dispersal and to reveal the relative importance of dispersal vs. in situ diversiication in adaptation as environmental conditions change. Clearly, the relative importance of dispersal vs. in situ evolution will vary with environment type, organism type, and the rate and magnitude of the perturbation. However, this balance will impact the level of global-scale biodiversity for each species. Understanding the timescales of microbial evolutionary processes is a tremendous challenge that is perhaps be addressed via studies of geologic systems. In the past, this has been undertaken through direct examination of the fossil record, including analysis of organic biomolecules preserved in ancient rocks. The opportunities and obstacles associated with attempts to tie together biological and geochemical evolution over Earth history are discussed by Brocks and Pearson (2005). Future studies may focus on modern systems for which both a recent geological and biological history can be reconstructed.

CONCLUDING COMMENTS The knowledge sought through molecular geomicrobiological studies will ind application in a variety of basic and applied ields such as astrobiology and environmental bioremediation. For example, understanding of the range of conditions in which life can persist, and the biochemical bases for the limitations, will constrain the variety of habitat types that may be targeted in the search for life signs on extraterrestrial planets. Molecular-level understanding of biogeochemical processes may be central to an assessment of the biogenicity of mineral biosignatures found there. Furthermore, information about the metabolic pathways for synthesis of organic compounds will assist in evaluation of the sources of organic biosignatures that persist in the geologic record. There has been great deal of interest in the possibility that the enormous problem of contamination of the environment by organic compounds, metals, and radionuclides may be tackled by harnessing the ability of microorganisms to change the chemical speciation of pollutants. Much research has been devoted to exploring this possibility (Madsen 2001; Lovley 2003; Nevin et al. 2003). In addition, understanding of the biogeochemical reactions that occur during stimulated bioremediation can be used to design methods to follow the progress of the treatment and identify key reaction products. For example, recently developed geophysical approaches can sense changes in subsurface transport properties, allowing progress of bioremediation to be evaluated and post treatment changes monitored (Williams et al. 2005). The merging of ields as disparate as molecular microbiology and geophysics is becoming routine, but the results are as yet mostly unknown. We can look forward with optimism to the product of the next decade of research in molecular geomicrobiology.

ACKNOWLEDGMENTS Funding for research and development of ideas described here derived from the National Science Foundation Biocomplexity Program, the Department of Energy Genomics: Genomes to Life Program and Basic Energy Sciences Chemical Sciences Program, and the NASA Astrobiology Institute. The contributions of our colleagues and collaborators to this work are gratefully acknowledged.

6

Banęeld, Tyson, Allen, Whitaker REFERENCES

Bhattacharyya A, Stilwagen S, Ivanova N, D’Souza M, Bernal A, Lykidis A, Kapatral V, Anderson I, Larsen N, Los T, et al. (2002) Whole-genome comparative analysis of three phytopathogenic Xylella fastidiosa strains. Proc Natl Acad Sci USA 99:12403-12408 Brocks JJ, Pearson A (2005) Building the biomarker tree of life. Rev Mineral Geochem 59:233-258 Chan CS, De Stasio G, Nesterova M, Welch SA, Girasole M, Frazer B, Banield JF (2004) The role of microbial polymers in templated mineral growth. Science 303:1656-2658 Coombs JM, Barkay T (2004) Molecular evidence for the evolution of metal homeostasis genes by the lateral gene transfer in bacteria from the deep terrestrial subsurface. Appl Environ Microbiol 70:1698-1707 DiChristina TJ, Fredrickson JK, Zachara JM (2005) Enzymology of electron transport: energy generation with geochemical consequences. Rev Mineral Geochem 59:27-52 Escobar-Paramo P, Ghosh S, DiRuggiero J (2005) Evidence for genetic drift in the diversiication of a geographically isolated population of the hyperthermophilic archaeon Pyrococcus. Molec Bio Evol 22: 2297-2303 Ferris JP (2005) Catalysis and prebiotic synthesis. Rev Mineral Geochem 59:187-210 Gevers D, Cohan FM, Lawrence JG, Spratt BG, Coenye T, Feil J, Stackenbrandt E, Van de Peer Y, Vandamme P, Thompson FL, Swings J (2005) Opinion: Re-evaluating prokaryotic species. Nat Rev Microbiol 3: 733-739 Gilbert B, Banield JF (2005) Molecular-scale processes involving nanoparticulate minerals in biogeochemical systems. Rev Mineral Geochem 59:109-155 Gilbert PUPA, Abrecht M, Frazer BH (2005) The organic-mineral interface in biominerals. Rev Mineral Geochem 59:157-185 Gogarten JP, Townsend JP (2005) Horizontal gene transfer, genome innovation and evolution. Nat Rev Microbiol 3:679-687 Guerquin-Kern JL, Wu TD, Quintana C, Croisy A (2005) Progress in analytical imaging of the cell by dynamic secondary ion mass spectroscopy (SIMS microscopy). Biochim Biophys Acta 1724: 228-238 Holden MT, Titball RW, Peacock SJ, Cerdeno-Tarraga AM, Atkins T, Crossman LC, Pitt T, Churcher C, Mungall K, Bentley SD, et al. (2004) Genomic plasticity of the causative agent of melioidosis, Burkholderia pseudomallei. Proc Natl Acad Sci USA 101:14240-14245 Jin Q, Bethke CM (2005) Predicting the rate of microbial respiration in geochemical environments. Geochim Cosmochim Acta 69:1133-1143 Kappler A, Straub KL (2005) Geomicrobiological cycling of iron. Rev Mineral Geochem 59:85-108 Kraemer SM, Butler A, Borer P, Cervini-Silva J (2005) Siderophores and the dissolution of iron-bearing minerals in marine systems. Rev Mineral Geochem 59:53-84 Lindell D, Jaffe JD, Johnson ZI, Church GM, Chisholm SW (2005) Photosynthesis genes in marine viruses yield proteins during host infection. Nature, doi:10.1038/nature04111 Lovley DR (2003) Cleaning up with genomics: applying molecular biology to bioremediation. Nat Rev Microbiol 1:35-44 Madsen EL (2001) Intrinsic bioremediation of organic subsurface contaminants. In: Subsurface Microbiology and Biogeochemistry. Fredrickson JK, Fletcher M (eds) Wiley-Liss Inc., New York, p. 249-278 McCollom TM, Shock EL (1997) Geochemical constrains on chemolithautotrophic metabolism by microorganisms in sealoor hydrothermal systems. Geochim Cosmochim Acta 61:4375-4391 Moreau JW, Webb RI, Banield JF (2004) Ultrastructure, aggregation-state, and crystal growth of biogenic nanocrystalline sphalerite and wurtzite. Am Mineral 89:950-960 Nelson KE, Methé B (2005) Metabolism and genomics: adventures derived from complete genome sequencing. Rev Mineral Geochem 59:279-294 Nevin KP, Finneran KT, Lovley DR (2003) Microorganisms associated with uranium bioremediation in a highsalinity subsurface sediment. Appl Environ Microbiol 69:3672-3675 Newman DK, Gralnick JA (2005) What genetics offers geobiology. Rev Mineral Geochem 59:9-26 Nielsen KM, Townsend JP (2004) Monitoring and modeling horizontal gene transfer. Nat Rev Biotechnol 22: 1110-1114 Ram RJ, Verberkmoes NC, Thelen MP, Tyson GW, Baker BJ, Blake RC II, Shah M, Hettich RL, Banield JF (2005) Community proteomics of a natural microbial bioilm. Science 308:1915-1920 Ramesh MA et al. (2005) A phylogenomic inventory of meiotic genes: evidence for sex in Giardia and an early eukaryotic origin of meiosis. Current Biology 15:185–191 Raymond J (2005) The evolution of biological carbon and nitrogen cycling—a genomic perspective. Rev Mineral Geochem 59:211-231 Roberts RJ, Karp P, Kasif S, Linn S, Buckley MR (2005) An Experimental Approach to Genome Annotation. Critical Issues Colloquia Report, Washington D.C. USA: American Academy of Microbiology Springail D, Top EM (2004) Horizontal gene transfer and microbial adaptation to xenobiotics: new types of mobile genetic elements and lessons from ecological studies. Trends Microbiol 12:53-58

Molecular-Level Processes in Biogeochemical Cycles

7

Tringe SG, von Mering C, Kobayashi A, Salamov AA, Chen K, Chang HW, Podar M, Short JM, MathurEJ, Detter JC, Bork P, Hugenholtz P, Rubin EM (2005) Comparative metagenomics of microbial communities. Science 308:554-557 Tyson GW, Chapman J, Hugenholtz P, Allen EE, Ram RJ, Richardson PM, Solovyev VV, Rokhsar DS, Banield JF (2004) Community structure and metabolism through reconstruction of microbial genomes from the environment. Nature 428: 37-43 Tyson GW, Lo I, Baker BJ, Allen EE, Hugenholtz P, Banield JF (2005) Genome-directed isolation of the key nitrogen ixer Leptospirillum ferrodiazotrophum sp. nov. from an acidophilic microbial community. Appl Environ Microbiol 71:6319-6324 Whitaker RJ, Grogan DW, Taylor JW (2003) Geographic barriers isolate endemic populations of hyperthermophilic archaea. Science 301:976-978 Whitaker RJ, Banield JF (2005) Population dynamics through the lens of extreme environments. Rev Mineral Geochem 59:259-277 Williams KH, Ntarlagiannis D, Slater LD, Dohnalkova A, Hubbard SS, Banield JF (2005) Geophysical imaging of stimulated microbial biomineralization. Environ Sci Technol 39:7592-7600 Wilson MS, Herrick JB, Jeon CO, Hinman DE, Madsen EL (2003) Horizontal transfer of phn-Ac dioxygenase genes within one of two phenotypically and genotypically distinctive naphthalene-degrading guilds from adjacent soil environments. Appl Environ Microbiol 69:2172-2181

Reviews in Mineralogy & Geochemistry Vol. 59, pp. 9-26, 2005 Copyright © Mineralogical Society of America

2

What Genetics Offers Geobiology Dianne K. Newman Division of Geological and Planetary Sciences California Institute of Technology and the Howard Hughes Medical Institute Pasadena, California, 91125, U.S.A. [email protected]

Jeffrey A. Gralnick Department of Microbiology and The BioTechnology Institute University of Minnesota St. Paul, Minnesota, 55108, U.S.A. [email protected]

INTRODUCTION For over 50 years, the Parker Brothers’ board game “Clue” has maintained its position as the classic family detective game. A murder has been committed in the mansion, but we don’t know where, by whom, or how. Was it Professor Plum in the study with a knife, or Miss Scarlett in the ballroom with a candlestick? Through rolls of the dice, fragments of information patiently accumulated piece-by-piece, and the application of logic, players construct a case to igure out “whodunit”. Because there are several potential solutions to the problem, the key challenge is to igure out what happened by understanding how it happened. As for the players of “Clue,” scientists seeking to understand the co-evolution of life and Earth are often confronted with the dilemma of having to parse multiple solutions to an ancient biogeochemical event. For example, in trying to explain the genesis of Archean Banded Iron Formations, we must ask whether it was cyanobacteria in the near shore-environment producing O2, or anoxygenic phototrophs in the oceans directly oxidizing iron (Kappler et al. 2005)? Again, in parallel to “Clue,” typically all we have to work with are isolated scraps of evidence—metamorphosed pieces of rock collected from remote locales on Earth, that contain morphological and/or chemical fossils whose origin and/or meaning is enigmatic. Nevertheless, the legacies of billions of years of evolution—genetic rolls of the dice, subject to natural selection—provide us with a means to interpret these putative biosignatures. By applying the principle of uniformitarianism, we assume that the study of modern organisms can provide us with insights into the composition and behavior of their ancient relatives, thereby allowing us to reconstruct ancient events. This, of course, is a necessary assumption that may not be true, so in the end, all we can really claim is to construct satisfying stories that it the available data. So how does one go about solving the mysteries of geobiology? Multiple approaches are covered in this book, but our focus in this chapter will be on how the logic of bacterial genetics can be applied to geobiological problems. Because genetics is not often a discipline that geologists are familiar with, we begin our discussion with some deinitions. From there, we go on to discuss how genetics can help us understand the past, both generally and through speciic examples; we do not discuss how genetics can help us understand modern biogeochemical processes, because we have recently reviewed this elsewhere (Croal et al. 2004a). Finally, 1529-6466/05/0059-0002$05.00

DOI: 10.2138/rmg.2005.59.2

10

Newman & Gralnick

we close with practical information about how to develop genetic systems in newly-isolated strains of geobiological interest, to guide those seeking to incorporate genetic analysis into their own research.

DEFINITIONS What is genetics? Classical bacterial and phage genetics was pioneered in the 1940s and 50s by Max Delbrück, Salvador Luria, Oswald Avery, Maclyn McCarty, Alfred Hershey, Martha Chase, Joshua Lederberg, Sydney Brenner, Seymour Benzer, Arthur Pardee, François Jacob, and Jacques Monod to name only a few of the key players. The extraordinary history of the development of this discipline (and molecular biology more generally) has been told well by Horace Judson in the book The Eighth Day of Creation (Judson 1996). Thanks to these scientists, genetics became a powerful tool for understanding how basic biological phenomenon worked (e.g., the nature of the gene (Avery et al. 1944), recombination (Lederberg 1946), the regulation of gene expression (Pardee et al. 1959), the nature of the genetic code (Crick et al. 1961), and mutations (Benzer and Champe 1962)). In each of these cases, genetic analysis lead to insights into how things happened, and was predicated upon the construction and analysis of mutants (Beckwith and Silhavy 1992). Accordingly, when we talk about applying genetics to geobiology, we mean performing experiments to understand geomicrobiological processes in mechanistic detail, either by mutagenizing model organisms (e.g., strains that can catalyze a particular geochemical transformation of interest, such as manganese oxidation (van Waasbergen et al. 1996), iron reduction (Coppi et al. 2001; DiChristina et al. 2002; Myers and Myers 2002), arsenate reduction (Saltikov and Newman 2003), methanogenesis (Pritchett and Metcalf 2005)) or by cloning DNA from the environment and expressing it in a foreign host (this is sometimes called “metagenomics” (Beja et al. 2000; Riesenfeld et al. 2004)). For the remainder of this chapter, we will focus our discussion on bacterial genetics to illustrate the more general theory and practice of genetics, whose logic is the same, regardless of the organism in which it is applied. In the context of geobiology, however, it is important to also recognize the recent contributions several labs have made in advancing archaeal genetics (Metcalf et al. 1997; Peck et al. 2000); because archaea catalyze a variety of geochemically signiicant reactions, that representatives from this group now can be manipulated genetically bodes well for future studies aimed at understanding their impact on the environment.

How is genetics different from molecular biology and genomics? Although modern bacterial genetics is molecular (e.g., gene composition can be readily determined by automated sequence analysis), originally it was not. The key to classical bacterial genetics was the use of deductive reasoning to understand the order and behavior of genetic elements in a genome, accomplished often through elegant assays that required little more than “toothpicks and logic” (Shuman 2003). While sequence information greatly facilitates genetic analysis today, the cornerstone of modern bacterial genetics is essentially the same as it was a half century ago: genetics deconstructs how a system works by making mutants that either eliminate/attenuate the ability of a strain to perform a certain function, or that confer a new property upon it. The challenge and satisfaction of this approach lies in being able to design simple experiments whose results will provide an explanation for a process. With a collection of different mutants, for example, complex biosynthetic processes can be broken down into components, each of which can be reconstructed and understood in detail. Genetic analysis goes hand in hand with physiological, cell biological and/or biochemical approaches that enable the phenotypes (i.e., physical characteristics or behavior) of mutants to be explored in depth.

What Genetics Oěers Geobiology

11

In contrast, molecular biology is the science of understanding the chemical composition of important biomolecules such as DNA and protein, and being able to manipulate them. Molecular biology commonly inds application in geobiology through microbial ecology surveys where the 16S gene for ribosomal RNA is cloned and sequenced to determine what types of organisms are present in a given environment (Pace 1997); another application is the use of molecular probes to identify organisms in natural samples through luorescent in situ hybridization (FISH) (Schrenk et al. 1998; Orphan et al. 2001; Michaelis et al. 2002). In effect, molecular biology permits geobiologists to apply genetics to the environment—to search for the presence and/or expression of particular genes once their function is known (KarkhoffSchweizer et al. 1995; Malasarn et al. 2004). Finally, genomics is the study of genomes with respect to their gene content and organization (also see in this volume Nelson and Methé 2005 and Whitaker and Banield 2005). It relies heavily upon computational analyses to compare different sequences (from one or more organisms) to each other and to identify motifs in the genes or their translated protein products that are predicted to have a speciic function. For example, hypotheses can be generated about what types of reactions a given protein might catalyze, or the conditions under which the gene that encodes it might be expressed; sometimes, genomic analysis can even be used to make predictions about the behavior of entire microbial communities (Tyson et al. 2004). The special advantage of environmental genomic data is that it allows gene expression in communities to be monitored in situ (Ram et al. 2005). It should be emphasized, however, that although much can be learned from genomics, ultimately, predictions about an organism’s (or a community’s) potential to perform a certain function must be tested through classical genetic and/or biochemical analyses to prove that the connection between the presence of a particular gene and a given geochemical state is actually causal as opposed to correlative.

What is a mutant? A mutant is a bacterial strain that differs genetically in some way from the parent strain of the species. While the genotype (e.g., DNA) of the mutant must, by deinition, be different from the parent, this is not necessarily the case phenotypically. A single base pair change in the genome could have no effect on the phenotype of the strain, however, genotypically, this strain is now different from the parent and is therefore a mutant.

What is mutagenesis? The capacity to alter the activities of single, or many proteins, from an organism by eliminating the gene(s) that encode them is critical for identifying proteins involved in a process of interest. Mutagenesis is the process of altering the genotype of a strain to make it different from the parent strain (i.e., a mutant). Traditional biochemical methods of identifying an activity in a cell extract can be a complementary method to genetics, but cannot unambiguously identify proteins required for an activity in vivo. If a protein is required for the activity of interest catalyzed by an organism, then removing the capacity of the strain to produce the protein will eliminate the activity. Several methods are used today to mutagenize bacteria, each with different strengths and weaknesses. These will be discussed in detail below.

TYPES OF GEOBIOLOGICAL PROBLEMS THAT GENETICS CAN SOLVE How can genetics help us learn about the geobiology of the past? To answer this question, we must irst deine what “geobiology of the past” means. Although a wide array of subjects— ranging from dinosaurs to ediacara—could it this description, we will limit our discussion to microorganisms and how their evolution affected Earth’s near surface environment (i.e.,

12

Newman & Gralnick

subsurface down to a few kilometers). This choice is justiied if one seeks to understand what life was like on this planet for the majority of its history because microorganisms have been in existence much longer than macroscopic organisms. Microorganisms (especially bacteria and archaea) are distinguished by their metabolic diversity rather than their morphological diversity, thus studying the geobiology of the past essentially is an exercise in understanding the evolution of metabolism as recorded in ancient rocks. Because our knowledge of the metabolic diversity of microbial eukaryotes is very limited, we will not consider them here, although we note that this is an area of opportunity for future students of geobiology. Modern microorganisms appear to be capable of generating metabolic energy from any redox reaction that is thermodynamically favorable so long as the constituents involved in the reaction are available in a habitable environment. Their metabolic diversity is based upon their ability to harvest energy from oxidation and reduction reactions, where the oxidant and/or the reductant may be organic or inorganic compounds. In some cases, the substrates and/or products of microbial metabolism are minerals, whereas in others, they are gases. Regardless of what form they come in, microbial substrate consumption or product formation can have a dramatic affect on the geochemistry of the environment. A classic example of this is the evolution of photosystem II, which enabled cells to produce molecular oxygen from water and thereby oxidize the Earth. Prior to this event, however, microbial life had to subsist anaerobically for millions and perhaps billions of years. How did cells cope? What electron acceptors and electron donors did microorganisms use for energy generation? And can we decipher a record of these primitive metabolisms in ancient rocks? These are hard questions, and at irst blush, it is not obvious whether genetics can provide the answers. Genetics is an experimental discipline, requiring geobiologists to work with modern microorganisms that we assume behave in much the same way as their ancient relatives. How reasonable is this assumption? One argument in its favor is that the forces of natural selection are conservative: once a particular metabolism is “invented” and is successful, only a limited set of mutations in the genes that confer this metabolism are possible in order for it to be preserved. While evolutionary history records myriad instances in which genetic changes led to the development of novel proteins and hence novel metabolisms, if we focus on a particular metabolism, and the biochemistry of its catalytic core, it is reasonable to infer that biology has only a inite number of solutions to make it work (Kauffman 1993). Moreover, as the complexity or dificulty of a metabolic process increases, we might expect the repertoire of solutions to become even more limited. This conclusion appears to be robust, albeit facilitated through horizontal gene transfer, given the conservation of metabolic genes in the genomes of phylogenetically distant organisms (Doolittle 1999; Friedrich 2002; Nixon et al. 2002; Malasarn et al. 2004; Simonson et al. 2005). Interestingly, microbiologists of the Delft school anticipated these indings nearly a century ago, noting the “manifest unity” in the biochemistry that forms the basis for the ecological relationships of microorganisms in nature (Kluyver 1924). If biochemistry is essentially conservative with respect to metabolism, then using genetics to understand how modern metabolisms work can help us develop a basis for deciphering their origins and how organisms that utilized them may have altered the chemical and physical features of our planet. So what does this mean in practice? If understanding the evolution of metabolism is the goal, there are only two ancient records to work with: one that is recorded in rocks, and one that is recorded in genomes. Let us irst consider the former. Rocks preserve two different types of fossils: morphological and chemical. Morphological fossils are the more familiar, as features that stand out from the parent rock are relatively straightforward to identify, and are becoming ever more so given recent innovations in imaging technologies (Watters and Grotzinger 2001; Corsetti and Storrie-Lombardi 2003; Kemner et al. 2004). Once identiied, however, whether these features are truly biogenic can be the subject of intense debate, be it at the scale of

What Genetics Oěers Geobiology

13

nanoparticles such as magnetite (McKay et al. 1996), micron-sized putative cellular structures (Schopf 1993; Brasier et al. 2002), or centimeter-scale stromatolites (Grotzinger and Knoll 1999). To develop criteria whereby to evaluate the biogenicity of particular structures in ancient rocks, it is helpful to understand how these structures form. This is where genetics can help. For example, if certain conditions prove to be required for the biological formation of a particular structure in modern organisms, and traces of these conditions are absent in an ancient sample, this would argue against its biogenicity. Such an argument was recently made with respect to the magnetite in the Martian meteorite ALH84001, which did not contain a magnetic signature that supported a biogenic origin (i.e., alignment of magnetite in chains) (Weiss et al. 2004). The key assumption in Weiss et al.’s paper was that bacteria organize magnetite into chains by direct molecular control—an assumption that was based on phenomenological observations (Gorby et al. 1988). Recently, genetic analysis has begun to reveal the speciic molecular components responsible for this organization (Komeili et al. 2005). The power of bacterial genetics lies is its ability to provide clear and deinitive proof that a particular protein is involved in a given function. The case of magnetite is only one example of where genetic analysis can guide our interpretation of the biogenicity of ancient samples. As stated above, in addition to morphological fossils, rocks preserve chemical fossils. These, in turn, come in two varieties: organic and inorganic. It is fair to assume that all organic fossils are of biological origin (the likelihood that prebiotic organic synthesis left preservable traces is extremely small), but it is much harder to know what they mean when we ind them, as discussed in the chapter by Brocks and Pearson (2005). Here too, genetics can help. For example, hydrocarbon molecules known as 2-methylhopanes in the sedimentary record can unambiguously be recognized as the molecular fossils of 2-methyl bacteriohopanepolyols (2MeBHPs) that are found in selected modern bacteria. Because cyanobacteria—the only bacteria that engage in oxygenic photosynthesis—are the only known, quantitatively important, source of 2-MeBHPs in the modern environment, it has been inferred that 2-methylhopanes can be used as biomarkers for oxygenic photosynthesis itself (Summons et al. 1999). Thus, Brocks et al. (1999, 2003) interpreted the presence of 2-methylhopanes in sediments of the Archaean Fortescue Group as evidence that photosynthetically-derived O2 irst appeared on Earth at least 2.7 billion years ago. However, there is presently no evidence that 2-MeBHPs and oxygenic photosynthesis are functionally related. Our conidence in this critical assumption, as well as in the use of 2-methylhopanes as biomarkers for cyanobacteria (or any other organism in which they exist), would be signiicantly improved by an understanding of the biochemical function of 2-MeBHPs. Keeping to the theme of O2 evolution, the second class of chemical fossils—inorganic biosignatures—also can be used to shed light on when this critical event occurred. Recently, through the work of Farquhar et al. (2000), mass independent sulfur isotopic signatures from sulide and sulfate in Precambrian rocks were used to date a major change in the change in the sulfur cycle between 2090 and 2450 million years ago, likely attributable to the rise of O2. Canield and colleagues have provided additional support for this conclusion, through their work on sulfur isotopic fractionation by archaea and bacteria (Canield et al. 2000; Shen et al. 2001; Habicht et al. 2002). Central to these studies is the knowledge that sulfur isotope fractionation responds to metabolism—for example, uptake and reduction of sulfate involves kinetic isotope effects that result in the lighter isotope of sulfur being enriched in the sulide product. The extent of enrichment depends on the growth rate of the organism, which can be controlled by temperature, the nature of the electron donor, and the concentration of sulfate among other factors (Jones and Starkey 1962; Kemp and Thode 1968; Shen et al. 2001)). While great strides have been made in this area without the involvement of genetics, knowledge of the biochemical pathway responsible for sulfate reduction has greatly facilitated interpretations of microbial sulfur isotopic fractionation by bacteria. It is thought that the majority of isotopic

14

Newman & Gralnick

fractionation occurs when S-O bonds are broken, such as during reduction of adenosine 5cphosphatosulfate (APS) to sulite by the enzyme APS reductase, with subsequent enzymatic reduction of sulite to sulide (Canield et al. 2005). Genetics affords a means to identify such pathways for more recently discovered geomicrobial organisms that might leave an imprint in the rock record, where the mechanism(s) of isotope effects are not yet fully understood. For example, iron-oxidizing anoxygenic phototrophs have been implicated in the direct deposition of Banded Iron Formations, but at present is it dificult to distinguish their activities from those of cyanobacteria on the basis of iron isotopic fractionation alone (Croal et al. 2004b). Knowledge of what enzymes or molecular components catalyze iron oxidation, where they are localized, and other details of how anoxygenic phototrophs trafic in iron, will position us to better interpret the mechanism of iron isotope fractionation by these bacteria, and thereby develop criteria with which to identify the products of their metabolism in ancient rocks. Even if iron isotopes prove not to provide a unique signature for a particular microbial metabolism, genetic analysis can still be very useful in pointing to potentially novel biosignatures. For example, recent genetic and biochemical results from our laboratory indicate that the enzymes that catalyze iron oxidation are soluble proteins that are localized inside the cell (Croal et al. unpub. data). If true, this implies that the cell has a mechanism for preventing the intracellular precipitation of iron oxide, possibly by chelating ferric iron with an organic molecule that helps release it to the outside where it then precipitates. In the event such a molecule were to exist, and it were shown to be preservable over geologic time scales, this would be an example of a metabolically-speciic biomarker discovered through genetics. Whether or not genetic analysis will ultimately lead to the discovery of physiologicallyspeciic biomarkers, identiication of the genes involved in geobiological processes will provide insight into their evolutionary origin. In this respect, DNA itself is a fossil, as phylogenetic relationships between sequences can provide information about their evolution. A good example of this is the interpretation of the antiquity of anoxygenic photosynthesis based on phylogenies of proteins involved in bacteriochlorophyll biosynthesis (Xiong et al. 2000). Although it is very dificult to date the divergence of groups of proteins, it is reasonable to use phylogeny in tandem with the rock record to infer the relative temporal evolution of different metabolisms (House et al. 2003; Kirschvink et al. 2000). Again, the role genetics plays in this process is to provide the proof that speciic genes encode proteins that perform speciic functions. Only after this is understood can phylogenetic comparisons be meaningful. As stated above, we caution against the danger of inferring function on the basis of phylogeny alone. Evolution is rife with instances where small sequence changes in an active site of a protein change its substrate speciicity (such as in the case of the directed evolution of a fucosidase from a galactosidase; Zhang et al. 1997), thus we cannot be certain that a putative protein actually performs the function we think it does until we do an experiment to prove it. Moreover, there are cases where different proteins have independently evolved that catalyze a similar reaction, yet on the basis of their sequence, they appear to have little in common. A good example of this are the serine proteases, subtilisin and trypsin (Kelly et al. 2005). Thus the absence of a particular gene in the genome of an organism should not be taken as evidence that it cannot perform a certain function. Even for organisms such as Escherichia coli and Salmonella, upon whose DNA the science of bacterial genetics was built, there remain a large number of genes of unknown function. Finally, genetic analysis can provide insights into the conditions that regulate a particular process. As described below, it is straightforward to use molecular reporters to assay for the expression of a gene of interest by exposing the bacteria that harbor the gene to different chemicals, temperatures, or pressures. This has the potential to be useful in making inferences about the paleoenvironment. For example, if evidence were found in the rock record that a particular biomolecule was present that was known only to be produced under conditions

What Genetics Oěers Geobiology

15

when oxidized molybdenum [Mo(VI)] was available, this would suggest that the pH of that environment must have been greater than ive and the redox potential greater than zero because these are the conditions where Mo(VI) exists in signiicant quantities (Anbar 2004).

PRACTICAL CONSIDERATIONS FOR CREATING GENETIC SYSTEMS As explained in the previous section, genetics has the potential to be a powerful tool for geobiology, offering insights into: i.) what structures to look for in the rock record, ii.) what they mean when we ind them, iii.) what enzymes catalyze their production, and iv.) what conditions regulate their expression. To be able to convert this theory into practice, it is helpful to know where to begin in the laboratory. In this section, we outline the key steps that would need to taken to make an organism genetically tractable. To briely summarize, the irst step is to isolate an organism that will be amenable to genetic analysis; this strain will serve as the standard (or “wild-type”) to which all subsequent mutants will be compared. This of course imposes a limitation on what genetics can offer geobiology, as not all strains can easily be coaxed into growing in the laboratory, much less be straightforward to mutagenize once isolated. Nevertheless, with perseverance and creativity, these dificulties can usually be overcome, leading to the second step: mutagenesis of the strain. Various methods for mutagenesis exist, offering the potential to eliminate/delete genes entirely, introduce pointmutations into speciic genes, or introduce genes into an organism. The effects of these different types of mutations can be far ranging, from altering the amino acid composition of a protein and thereby affecting its substrate speciicity, to eliminating the ability to make a set of proteins, to changing the regulation of an entire network of genes. After mutagenesis is performed, the third step is to identify the mutants either through a selection or a screen. A selection permits only those mutants that have the desired properties to grow, whereas a screen requires characterizing the behavior of thousands of mutants to identify only rare ones that have the properties/behavior of interest. Depending on the manner in which one has identiied candidate mutants, secondary screens may be required to narrow the pool of candidates down to only those that are interesting. For example, if one performs a screen to ind genes that control various steps in a biochemical reaction, if the assay for mutant identiication involves looking at the rate at which a reaction proceeds, “false” mutants could be identiied by the screen that are simply slow to grow but which do not have a speciic defect in the reaction of interest. These mutants could be sorted out by measuring the growth rate of all candidates and only continuing to study those whose growth is normal with respect to the parent strain. Once interesting mutants are identiied, the fourth step is to determine the nature of the mutation through sequencing and genetic veriication. Sequence analysis can help generate hypotheses to explain why the mutant behaves the way it does, and thus infer what affects the process of interest. To test these hypotheses, however, the inal step requires physiological, biochemical, or cell biological experiments to be performed in order to study the phenotype of the mutant in detail.

Step 1: Isolation and growth Developing a genetic system in an organism can be a tedious, albeit rewarding process. Development can be greatly enhanced when the organism of interest is a close relative to a microbe with an established genetic system. This was the case for the arsenic-respiring Gramnegative bacterium Shewanella sp. strain ANA-3, which resulted from a targeted-isolation of a strain that could grow strictly anaerobically on arsenate in minimal medium and also make single colonies overnight on Luria-Broth (LB) plates on the bench top. LB is a widely used rich medium in bacterial genetics, as it supports rapid growth and is easy to make. Because this enrichment strategy imposed a strong selection for bacteria that had respiratory versatility, it was not surprising that it resulted in the isolation of a new strain of Shewanella, a genus

16

Newman & Gralnick

renowned for this property (Nealson and Scott 2004). Since this organism is closely related to other strains of Shewanella that have established genetic systems, Saltikov et al. adapted strategies that had been successfully used in S. oneidensis strain MR-1 to their new isolate (Saltikov et al. 2003; Saltikov and Newman 2003). Several years prior to this, one of the authors had isolated a bacterial strain (Desulfotomaculum sp. strain OREX-4) that could also respire arsenate (Newman et al. 1998; Newman et al. 1997). However, because attempts to grow this strain on plates failed, a genetic system could never be established. Two main lessons regarding the development of a genetic system are illustrated by this example: 1.) Enrich for an organism in a targeted fashion so that a strain can be isolated that exhibits the desired properties and 2.) The ability to readily form colonies on plates is a highly desirable trait, for reasons that will be discussed below. One major beneit of growth on agar plates is facilitating strain isolation. Bacterial colonies on an agar plate typically form from a single cell, meaning that every cell that comprises the colony is identical at the genetic level, or of the same genotype. If a single colony is picked from a solid-surface medium, streaked across a new plate with the purpose of dispersing cells so that individual cells are isolated from their neighbors, and allowed to incubate, all subsequent colonies arising should be both morphologically and genetically identical to the original colony (Fig. 1). This process is typically called ‘colony puriication’ and can yield pure cultures of the bacterial strain of interest. Using solid surfaces to culture bacteria is, for all practical purposes, essential for the isolation of mutant strains after they are generated. Solid or liquid medium then can be used to perform basic physiological studies, such as determining optimal growth temperature, nutritional requirements and sensitivity to antibiotics or some other selectable marker. Characterizing the susceptibility of a strain to a selectable marker, such as a heavy metal (e.g., tellurium) or an antibiotic (e.g., kanamycin) is important for many genetic techniques. These techniques require the ability to isolate individuals within a population that carry a genetic difference from the overall population. Often, transposons (or “jumping genes”) that carry resistance determinants for a particular toxic compound are used to make mutants by

Figure 1. Streak plate. Example of a strain of S. oneidensis streaked for single colonies on solid medium.

What Genetics Oěers Geobiology

17

disrupting the chromosome at random, but mutants in speciic loci can also be generated by replacing the wild-type gene with a resistance determinant. Regardless of the method of mutagenesis, when these resistance determinants are inserted into the chromosome, they confer upon the resulting mutant strain resistance to the toxic compound. With the appropriate solid medium containing this compound, these mutants can be spatially separated and selected from a population that also contains the wild-type (the wild-type, lacking the resistance determinant, will die, so only the mutants will grow). When possible, it is helpful to make low endogenous resistance to several antibiotics a requirement in the isolation of an organism for genetic analysis. If the isolate is naturally resistant to many of the typical antibiotics (e.g., ampicilin, gentamycin, kanamycin, chloramphenicol and tetracycline), however, other strategies can be devised to make mutants—such as employing resistance to heavy metals (Gupta et al. 2002). Two additional properties are also beneicial to establishing genetics in an organism. First is the ability to introduce foreign DNA into the strain, which can be accomplished through transformation, transduction, or conjugation (Madigan et al. 2003). Transformation is when the cell takes up DNA directly—this can be facilitated by electroporation (using electric current to transform DNA into bacteria) or by generating chemically-competent cells (using high concentrations of salt, typically calcium chloride, and inducing the transformation via heat-shock). Transduction is the process whereby phage (i.e., viruses) infect bacterial cells and inject DNA into them which is then incorporated into the chromosome. Finally, conjugation if the process of transferring DNA from one bacterium to another by means of matings that involve the transfer of mobilizable plasmids. Introduction of foreign DNA is important for many targeted and random methods of mutagenesis, and critical for verifying the causality of the phenotype. The second beneicial property is speed of growth. When choosing a new strain for genetic work, the faster the organism grows, the faster its secrets will be unraveled (assuming the creativity of the scientist is not the limiting factor)!

Step 2: Methods of mutagenesis There are three types of mutagenesis that are common in bacterial genetics: chemical, transposon, and targeted. Chemical and transposon mutagenesis typically generate random mutations that are useful when one has no preconceptions about how a system works and seeks to cast a wide net to identify all possible genes involved in a process. In contrast, targeted gene “knockouts” are made when one has an idea of what gene(s) might be involved in a process and wants to test them speciically. We briely review these methods here, discussing their strengths and weaknesses. I. Chemical and UV mutagenesis. This method of generating mutants is rapid, inexpensive and fairly easy. Cells are treated with a mutagenic agent (e.g., ethyl or methyl methanesulfonate, nitrosoguanidine or ultraviolet light (Madigan et al. 2003)), grown briely, and then plated for isolated colonies. A balance must be struck in how much to mutagenize the cells: too little treatment results in few mutants in the population and makes it dificult to ind them among the remaining wild-type cells, while too much treatment frequently results in multiple mutations in a single cell which complications determinating causality later. Critical parameters to consider are mutagen concentration, mutagen type (some mutagens are stronger than others) and exposure time to the mutagen. In the case of UV mutagenesis, signiicant killing typically occurs, but this is a necessary side effect of achieving a suficiently high frequency of mutation in the remaining viable population. The major downside to this type of mutagenesis is the dificulty in identifying the gene, or genes, disrupted by the mutation. This means that more cells must be screened for the defect of interest because a large proportion will be both phenotypically and genotypically wild-type. On the positive side, however, this method of mutagenesis enables subtle phenotypes (such as residues on proteins that affect their substrate speciicity, or interactions with other proteins), as well as conditional

18

Newman & Gralnick

phenotypes (e.g., temperature sensitive mutations) or partial defects to be identiied. This is particularly useful in the identiication of essential genes, as they mutants can be generated under a condition that permits them to grow, and then shifted to a different condition that renders the mutation lethal. Another beneit of chemical/UV mutagenesis is that it does not depend on introducing foreign DNA into the strain, as the other techniques do (although later, this will be necessary to verify the nature of the mutation—see below). II. Transposon mutagenesis. Transposons are genetic elements that can move in either random, or non-random fashion into and out of chromosomes (Madigan et al. 2003). Facilitating this movement is an enzyme called transposase. These elements are believed to play a role in inluencing evolution and can be found in all forms of life. Researchers have modiied these elements to be used as tools to generate random mutations. Often times these modiications streamline the element to one or two genes, with one typically encoding resistance to an antibiotic. Plasmids used for transposon mutagenesis will contain both the transposon sequence, and a separate gene encoding the speciic transposase. When the plasmid is transformed into a strain, the transposase can be made, which will then facilitate integration of the transposon on the plasmid into the chromosome in a random fashion. The optimal plasmids used for such a procedure are unable to replicate without speciic genes, therefore subsequent selection of the population for strains resistant to the antibiotic will eliminate the parent strain. This leaves only mutant strains that have successfully integrated the transposon into their genome. A downside of transposon mutagensis is that genes that are essential for a process under the conditions where the transposon insertion is selected will be missed. This either requires the growth conditions for the selection of the transposon insertion to be different from those where the mutants will be identiied, or the use of chemical or UV mutagenesis that permits the study of conditional/partial phenotypes. When a transposon mutant with the desired phenotype is identiied, several methods exist to determine the disrupted gene. One way is by cloning the transposon from puriied, digested (or sheared) genomic DNA using the antibiotic resistance property. This method will yield genomic DNA lanking the transposon. Primers designed to the transposon can then be used to generate sequence into the lanking region. Alternatively, a process called arbitrary PCR can identify a small portion of sequence adjacent to the transposon directly from genomic DNA without cloning (Caetano-Anolles 1993). If working with a fully sequenced strain, as little as 20 base pairs of sequence is suficient to identify the location of the transposon. The process of identifying the transposon-mutated gene in an unsequenced strain is more cumbersome, but still straightforward. Two approaches are possible. One can make a genomic library from the mutant strain (a library comprises either plasmids or cosmids or fosmids, the latter holding signiicantly more DNA than plasmids), introduce this library into an appropriate host, and select for growth of cells that carry the transposon. Alternatively, the sequence identiied through arbitrary PCR can be used to probe a genomic library of the wild-type. Using hybridization techniques, a probe consisting of DNA lanking the transposon can identify plasmids in the genomic library containing homologous sequence. Once these have been identiied, the plasmids themselves can be sequenced, open reading frames (genes) identiied, and the genomic region surrounding the transposon reconstructed. III. Targeted gene knockout. In situations where a particular gene is suspected of being involved in a process (for example, when the genome of an organism of interest has been sequenced, and one can perform genomic analysis on it), it is often helpful to mutagenize that gene to test its involvement. This is called making a targeted gene “knockout.” Several methods exist to speciically eliminate a gene of interest. However, to take advantage of this technique, the sequence of the both the gene, and its surrounding region must be known (Fig. 2A). Simple inactivation of a gene can occur by inserting an antibiotic resistance gene into the gene targeted for knockout. This can be accomplished by cloning the gene, and some lanking sequence, if

What Genetics Oěers Geobiology

19

Figure 2. Diagram of targeted gene knockout using a suicide vector. Refer to text for description of igure.

required, into a plasmid that will only replicate in a speciic genetic background (Fig. 2B). Good examples of this class of plasmids are those that require the S protein (encoded by the pir gene, derived from the R6K plasmid (Kolter 1981) to replicate. By engineering the plasmid in an E. coli strain that contains the pir gene, one can construct such a vector. The idea is to clone the gene of interest, then modify the gene by inserting an antibiotic resistance cassette into the gene (which can be accomplished either by cloning or by fusion PCR, Fig. 2C). Ideally, this antibiotic resistance cassette will have at least 1,000 base pairs of sequence from the host strain on either side. This is important to facilitate homologous recombination into the genome of the strain of interest. Once the plasmid is constructed that contains the disrupted gene, the next step is to transform the strain with the newly constructed plasmid using a method described above. Strains that become resistant to both of the antibiotics encoded by the antibiotic resistance genes in the plasmid (antiA and antiB) have undergone a single crossover event (Fig. 2C), resulting in the integration of the plasmid into the genome (Fig. 2D). The strain cannot maintain the plasmid itself because it does not produce the S protein. By growing the strain without selecting for the endogenous plasmid resistance (antiA, Fig. 2), a second recombination

20

Newman & Gralnick

event can occur in some individuals within the population, resulting in the elimination of the plasmid DNA from the genome, along with the wild-type gene of interest (Fig. 2D, result 1). If selection for resistance to antibiotic B is maintained, the second recombination event (Fig. 2D, result 2) that reverts the strain back to wild-type, cannot occur. Variations on this technique can yield mutations such as total gene replacements, or even in-frame (non-polar) deletions of the gene of interest.

Genetic polarity in bacteria Bacteria typically contain a single, circular chromosome. Genes can be arranged in either direction in the genome and are typically clustered into operons. Genes organized in operons tend to be involved in the same process, although this is not always the case (Salgado et al. 2000). An operon is deined as multiple genes sharing the same genetic regulatory elements. A mutation that alters the capacity of the regulatory element to express downstream genes is called a “polar mutation.” This mutation can be any of the types discussed above, but is most often associated with transposon mutagenesis. Because of polarity, a gene disrupted by a transposon may not cause the identiied phenotype itself, but a gene (or genes) downstream may be responsible. To attribute a speciic process to a speciic gene, the problem of polarity must be taken into account, and complementation experiments must be done to demonstrate that a defect can be restored by provision of a particular gene.

Step 3: Identifying mutants Identiication of mutants defective in the process of interest is usually limited only by the robustness of the selection/screen, meaning that careful planning and thought should go into its design. We briely illustrate this with four examples from our laboratory. I. Screen for mutants defective in reducing anthraquinone-2,6-disulfonate (AQDS) in S. oneidensis strain MR-1 (Newman and Kolter 2000). The irst screen we performed involved the identiication of S. oneidensis mutants that were defective in reducing the soluble humic acid analog AQDS. Because reduced AQDS is orange in color, wells containing mutants defective in this process remained clear whereas other wells turned orange. Mutants were grown overnight, then inoculated into minimal medium containing AQDS in 96-well microtiter plates (96 independent mutant strains per plate) and covered with mineral oil to limit oxygen diffusion into the wells. Two classes of mutants were isolated from this screen: mutants that were completely unable to reduce AQDS, and mutants that reduced AQDS slowly. A screen that can be visually monitored over time can facilitate identiication of several classes of mutants with varying degrees of defectiveness. An example of this screen is shown in Figure 3. Note the lighter wells around the outside of the plate are likely due to re-oxidation of AQDS by oxygen diffusing into the plate. These effects could have been avoided by incubating the plates in an anaerobic chamber, which illustrates the importance of screen design for maximal eficiency in identifying potential mutants. II. Screen for iron hydr(oxide) reduction mutants in Shewanella oneidensis strain MR-1 (Gralnick and Newman, unpublished). To identify mutant strains of S. oneidensis defective in the ability to reduce insoluble iron hydr(oxide), we grew mutant strains overnight aerobically in 96-well microtiter plates in LB. These cultures were then transferred to minimal medium that contained iron hydr(oxide) as the sole electron acceptor for growth. Plates were incubated without shaking overnight, then a compound (ferrozine) was used to detect the presence of ferrous iron [Fe(II)], the product of iron hydr(oxide) reduction. Because this is a colormetric assay (when ferrozine binds to Fe(II), a purple color is formed), putative mutants were easily identiied by eye as wells that did not signiicantly change color. These mutants were retested by restreaking from the initial overnight culture, then checked again for their capacity to reduce iron hydr(oxide) to conirm the phenotype. Retesting is very important in this process, as it

What Genetics Oěers Geobiology

21

Figure 3. Mutant screen for AQDS reduction-deicient mutants in S. oneidensis. Screen was performed as indicated in the text. Dark wells represent reduced AQDS, circled well represents a mutant defective in AQDS reduction. See (Newman and Kolter 2000) for further details.

allows one to be liberal in the initial round of mutant identiication, which permits not only false positives to come through, but also mutants with subtle phenotypes. Using this method, we screened over 8,000 mutants for defects in iron hydr(oxide) reduction, yielding about 60 mutants. It was only after identifying several mutants that we realized a law in our screen design—any strain that was unable to grow in minimal medium would be identiied as defective in iron hydr(oxide) reduction. Therefore, many of the mutants we isolated were simply unable to grow in minimal medium because they lacked the capacity to produce certain essential amino acids or vitamins absent from the medium (this class of mutants are called auxotrophs). In the next version of this screen, we performed both our initial mutant selection and our screen in a medium containing amino acids and vitamins. This allowed auxotrophic strains to appear phenotypically wild-type rather than appear as mutants defective in iron hydr(oxide) reduction. This example illustrates the importance of screen construction to maximize the probability of identifying interesting mutants. III. Screen for mutants defective in photosynthetic Fe(II) oxidation in Rhodopseudomonas palustris strain TIE-1 (Jiao et al. 2005). In this screen, we used a similar approach to that described to identify mutants defective in iron hydr(oxide) reduction; however, some additional steps were required to maximize the eficiency of the screening process. Transposon-insertion mutants of R. palustris strain TIE-1 were pre-grown aerobically in 96-well microtiter plates, then transferred to photosynthetic medium and grown with hydrogen as the electron donor. Once strains had reached suficient density (3 days), they were centrifuged and resuspended in a buffer containing Fe(II). After several hours, ferrozine was added to visually observe the presence or absence of Fe(II) in each well. In this screen, wells containing mutants defective in phototrophic Fe(II) oxidation appeared purple (indicating the presence of Fe(II)), whereas clear wells revealed strains with wild-type activity. A key difference from the S. oneidensis screen is that this was a cell suspension assay, because growth of the strain was not required (i.e., the length of the assay was shorter than the doubling time of the cells). In this screen, apparent loss of Fe(II) oxidation activity could have resulted

22

Newman & Gralnick

from a blockage in a step required for Fe(II) uptake or iron oxidation. To differentiate between these two possibilities, we performed a secondary screen measuring total iron and Fe(II) for iltered and uniltered samples. In this manner, we were able to narrow the pool of candidate mutants down to only those with defects in Fe(II) oxidation. IV. Selection/screen for mutants defective in magnetite production in Magnetospirillum sp. AMB-1 (Komeili et al. 2004). Finally, perhaps our favorite mutant hunt due to its simplicity was one designed to identify genes required for magnetite formation in Magnetospirillum sp. AMB-1 (Komeili et al. 2004). Cells were irst mutagenized and grown under a condition where they did not produce magnetite. They then were pooled and transferred en mass to a condition where they could produce magnetite. Magnets were placed next to the tubes containing the entire mutagenized population to remove magnetic individuals, thus allowing for “mnm” mutants (magnetosome mutants) to be enriched. Individuals that passed through this selection, were screened individually under conditions that promoted formation of magnetite in microtiter plates. Once strains achieved the proper cell density, the entire plate was placed on top of a set of 24 magnets. The magnets were arranged so that they were positioned at the intersection of four individual wells of the microtiter plate. Magnetic strains were pulled toward the edge of their individual well, whereas non-magnetic (or poorly magnetic) strains remained at the center of the well.

Step 4: Mutant verification Regardless of the method of mutagenesis, after identifying a mutant, it is important to verify that a particular gene is responsible for the mutant phenotype (as opposed to the phenotype being due to a random mutation elsewhere in the chromosome). The process of “complementation” is used for this purpose. In complementation, the particular gene, or set of genes thought to be required for a process (usually determined by sequencing around the site of a transposon insertion), can be cloned from the wild-type into a plasmid that has the capacity to replicate in the mutant strain. If the plasmid contains suficient information to promote expression of the cloned gene, the phenotype of the mutant should be reversed, or complemented, and the mutant’s phenotype should be restored to that of the wild-type. This experiment demonstrates causality, verifying the role of the disrupted gene in the process of interest. Complementing point mutants is a more dificult task. A genomic library must be constructed from a wild-type background and then plasmids (or cosmids) containing the library transferred into the mutant strain of interest. Transformants are then screened for complementation of the mutant phenotype. Once a plasmid is identiied that will complement the mutant defect, it can be sequenced to identify the genes it contains. Individual genes can then be cloned to test for complementation, or the original plasmid can be fragmented and sub-cloned to determine the minimal amount of DNA required for complementation. Once the affected gene is identiied, the mutant version of the gene can be ampliied and sequenced to determine the nature of the original mutation.

A brief note on phage Phage (bacterial viruses) have played a critical role in not only our understanding of genetics and molecular biology, but also facilitating genetic work in a number of organisms. Modiied versions of transducing phage that package host genomic DNA at a high frequency can be used to perform many genetic tasks, from generating isogenic strains (two strains that differ genotypically in a single locus) to mapping point mutations and even generating mutant libraries. Because our goal in this chapter is merely to provide an introduction to developing a genetic system, we will not cover genetic techniques associated with phage beyond noting that developing a robust and eficient transducing phage system can add an additional level of sophistication to a genetic system.

What Genetics Oěers Geobiology

23

Step 5: Mutant analysis Once mutants have been identiied and the involvement of particular genes in the process of interest veriied, many new avenues become open for exploration. By identifying a variety of mutants with similar phenotypes, one can begin to construct a model for how a process works at the molecular level. Speciic studies can be initiated to study regulation of genes to determine the precise environmental conditions that trigger the organism to catalyze the process of interest. Putative marker genes for this activity may be identiied, potentially leading to the design of molecular tools to monitor when this process is active in a given environment (Malasarn et al. 2004). Once regulatory elements (e.g., promoters) have been identiied for a gene, a strain can be engineered to “report” when it is expressing that gene. For example, a promoter from a gene of interest can be cloned into a plasmid and used to drive expression of a protein that can be detected by luorescence or colormetric assay (e.g., green luorescent protein, GFP or betagalactosidase; reviewed by Kohler et al. 2000). Manipulating various environmental conditions in the laboratory can provide precise information regarding when the engineered strain is expressing the gene of interest by quantifying luorescence. Finally, the biochemical properties and subcellular localization of the gene product can be studied, either within the host strain, or by cloning and over-expressing the gene that encodes it in another organism. Help in understanding the speciic function of genes identiied through mutagenesis can come from analyzing the amino acid sequence encoded by the gene. A tool that can provide signiicant clues is the BLAST (Basic Local Alignment Search Tool) search engine available at NCBI (National Center for Biotechnology Information – www.ncbi.nlm.nih.gov). BLAST can be used to infer functional and evolutionary relationships between amino acid or nucleotide sequences. This program compares sequences entered by the user to a selected database, which can include all known sequences. The program assigns a statistical signiicance to matches within the database. If the gene of interest encodes a protein with a signiicant match to a protein of known function in the database, this may suggest it has a similar activity. If the protein has no signiicant match or is only similar to other proteins of unknown function, there are several additional analyses that can be performed on the sequence to gain insight into its function. Other useful types of searches are motif, post-translational modiication and topology. Several programs in each category can be found on the ExPASy (Expert Protein Analysis System) web server (www.expasy.org/tools/), which is maintained by the Swiss Institute of Bioinformatics. Programs found here will allow further analysis of the protein sequence of interest to predict characteristics such as subcellular localization, co-factor binding and transmembrane domains. For example, if the protein is predicted to bind a redoxactive cofactor such as heme, we may hypothesize that it plays a role in electron transfer. As we noted previously, however, it is imperative to remember that database predictions are only suggestive, and must be demonstrated experimentally. However, programs such as these greatly help formulate testable models for the function of a protein.

CONCLUSIONS In this chapter, we have focused our discussion on how bacterial genetics can help unravel the geobiology of the past, and have provided an introduction to basic genetic principles that we hope will encourage those less familiar with genetics to use it as a tool in their research. Not only is making a connection between genetics and geobiology a stimulating intellectual endeavor, but it is also great fun in practice. To close by returning to the analogy with which we started, when it comes to understanding the biogeochemical evolution of the Earth, we must accept that we will never know “whodunit” with absolute certainty, short of the invention of a time machine. But even if after making millions of mutants, we still don’t have a clue about the past, without question, applying genetics to geobiology affords us an excellent

24

Newman & Gralnick

opportunity to make fundamental discoveries about how modern microorganisms shape the geochemistry of their environment.

ACKNOWLEDGMENTS We wish to thank the students of the USC International Course in Geobiology (sponsored by the Agouron Institute), whose enthusiasm for genetics fed our own, and compelled us to think more critically about how genetics can help solve problems in geobiology. In addition, we would like to express our gratitude to the GPS division at Caltech, for nurturing our vision and our work, and the students and postdocs of the Newman lab (past and present) for putting it all together. Special thanks to Laura Croal, Arash Komeili and Doug Lies for constructive comments on the manuscript. We acknowledge the Luce Foundation, Packard Foundation, Agouron Institute, ONR, DARPA, NSF, Beckman Institute, and HHMI for inancial support.

REFERENCES Anbar AD (2004) Molybdenum stable isotopes: observations, interpretations and directions. Rev Mineral Geochem 55:429-454 Avery OT, MacLeod CM, McCarty M (1944) Induction of transformation by a desoxyribonucleic acid fraction isolated from Pneumococcus type III. J Exp Med 79:137-158 Beckwith J, Silhavy TJ (1992) The Power of Bacterial Genetics: a Literature-Based Course. Cold Spring Harbor Laboratory Press, New York Beja O, Aravind L, Koonin EV, Suzuki MT, Hadd A, Nguyen LP, Jovanovich SB, Gates CM, Feldman RA, Spudich JL, Spudich EN, DeLong EF (2000) Bacterial rhodopsin: evidence for a new type of phototrophy in the sea. Science 289:1902-6 Benzer S, Champe SP (1962) A change from nonsense to sense in the genetic code. Proc Nat Acad Sci USA 48:1114-1121 Brasier MD, Green OR, Jephcoat AP, Kleppe AK, Van Kranendonk MJ, Lindsay JF, Steele A, Grassineau NV (2002) Questioning the evidence for Earth’s oldest fossils. Nature 416:76-81 Brocks JJ, Buick R, Summons RE, Logan GA (2003) A reconstruction of Archean biological diversity based on molecular fossils from the 2.78 to 2.45 billion-year-old Mount Bruce Supergroup, Hamersley Basin, Western Australia. Geochim Cosmochim Acta 67:4321-4335 Brocks JJ, Logan GA, Buick R, Summons RE (1999) Archean molecular fossils and the early rise of eukaryotes. Science 285:1033-1036 Brocks JJ, Pearson A (2005) Building the biomarker tree of life. Rev Mineral Geochem 59:233-258 Caetano-Anolles G (1993) Amplifying DNA with arbitrary oligonucleotide primers. PCR Methods Appl 3: 85-94 Canield DE, Habicht KS, Thamdrup B (2000) The Archean sulfur cycle and the early history of atmospheric oxygen. Science 288:658-661 Canield DE, Thamdrup B, Kristensen E (2005) Aquatic Geomicrobiology, Advances in Marine Biology. Elsevier, San Diego Coppi MV, Leang C, Sandler SJ, Lovley DR (2001) Development of a genetic system for Geobacter sulfurreducens. Appl Environ Microbio 67:3180-3187 Corsetti FA, Storrie-Lombardi MC (2003) Lossless compression of stromatolite images: A biogenicity index? Astrobiology 3:649-655 Crick FHC, Barnett L, Brenner S, Watts-Tobin RJ (1961) General nature of the genetic code for proteins: Nature 4809:1227-1232 Croal LR, Gralnick JA, Malasarn D, Newman DK (2004a) The genetics of geochemistry. Annu Rev Genet 38: 175-202 Croal LR, Johnson CM, Beard BL, Newman DK (2004b) Iron isotope fractionation by Fe(II)-oxidizing photoautotrophic bacteria. Geochim Cosmochim Acta 68:1227-1242 Croal LR, Jiao Y, Newman DK (2005) unpublished data DiChristina TJ, Moore CM, Haller CA (2002) Dissimilatory Fe(III) and Mn(IV) reduction by Shewanella putrefaciens requires ferE, a homolog of the pulE (gspE) type II protein secretion gene. J Bacteriol 184: 142-151 Doolittle WF (1999) Phylogenetic classiication and the universal tree. Science 284:2124-2128

What Genetics Oěers Geobiology

25

Farquhar J, Bao H, Thiemens M (2000) Atmospheric inluence of Earth’s earliest sulfur cycle. Science 289: 756-758 Friedrich MW (2002) Phylogenetic analysis reveals multiple lateral transfers of adenosine-5-phosphosulfate reductase genes among sulfate-reducing microorganisms. J Bacteriol 184:278-289 Gorby YA, Beveridge TJ, Blakemore RP (1988) Characterization of the bacterial magnetosome membrane. J Bacteriol 170:834-841 Grotzinger JP, Knoll AH (1999) Stromatoites in Precambrian carbonates: Evolutionary mileposts or environmental dipsticks? Annu Rev Earth Planet Sci 27:313-358 Gupta A, Meyer JM, Goel R (2002) Development of heavy metal-resistant mutants of phosphate solubilizing Pseudomonas sp. NBRI 4014 and their characterization. Curr Microbiol 45:323-7 Habicht KS, Gade M, Thamdrup B, Berg P, Canield DE (2002) Calibration of sulfate levels in the Archaen ocean. Science 298:2372-2374 House CH, Runnegar B, Fitz-Gibbon ST (2003) Geobiological analysis using whole genome-based tree building applied to the Bacteria, Archaea, Eukarya. Geobiology 1:15-26 Jiao Y, Kappler A, Croal LR, Newman DK (2005) Isolation and characterization of a genetically-tractable photoautotrophic Fe(II)-oxidizing bacterium, Rhodopseudomonas palustris strain TIE-1. Appl Environ Microbiol 71:4487-4496 Jones GE, Starkey RL (1962) Some necessary conditions for fractionation of stable isotopes of sulfur by Desulfovibrio desulfuricans. In: Biogeochemistry of Sulfur Isotopes. NSF Symposium. Jensen ML (ed.) Yale University Press, New Haven, CT, p 61-79 Judson HF (1996) The Eighth Day of Creation. Cold Spring Harbor Laboratory Press, New York Kappler A, Pasquero C, Konhauser KO, Newman DK (2005) Deposition of banded iron formations by phototrophic Fe(II)-oxidizing bacteria. Geology 33(11):in press Karkhoff-Schweizer RR, Huber DPW, Voordouw G (1995) Conservation of the genes for dissimilatory sulite reductase from Desulfovibrio vulgaris and Archaeoglobus fulgidus allows their detection by PCR. Appl Environ Microbiol 61:290-296 Kauffman S (1993) The Origins of Order. Oxford University Press, Oxford Kelly SD, Laskowski M, Qasim MA (2005) Tje rp;e pf scaffp;domg om standard mechanism serine proteinase inhibitors. Protein Peptie Lett 12:465-471 Kemner KM, Kelly SD, Lai B, Maser J, O’Loughlin EJ, Sholto-Douglas D, Cai ZH, Schneegurt MA, Kulpa CF, Nealson KH (2004) Elemental and redox analysis of single bacterial cells by X-ray microbeam analysis. Science 306:686-687 Kemp ALW, Thode HG (1968) The mechanism of the bacterial reduction of sulfate and of sulite from isotope fractionationstudies. Geochim Cosmochim Acta 32:71-91 Kirschvink JL, Gaidos EJ, Bertani LE, Beukes NJ, Gutzmer J, Maepa LN, Steinberger RE (2000) Paleoproterozoic snowball Earth: Extreme climatic and geochemical global change and its biological consequences. Proc Natl Acad Sci USA 97:1400-1405 Kluyver AJ (1924) Unity and diversity in the metabolism of microorganisms. Chemisch Weekblad 21:226 Kohler S, Belkin S, Schmid RD (2000) Reporter gene bioassays in environmental analysis. Fresenius J Anal Chem 366:769-79 Kolter R (1981) Replication properties of plasmic R6K. Plasmid 5:2-9 Komeili A, Vali H, Beveridge TJ, Newman DK (2004) Magnetosome vesicles are present before magnetite formation, and MamA is required for their activation. Proc Natl Acad Sci USA 101:3839-3844 Lederberg J (1946) Gene recombination and linked segregations in E. coli. Genetics 32:505-525 Madigan MT, Martinko JM, Parker J (2003) Bacterial Genetics, Biology of Microorganisms. Prentice Hall, Upper Saddle River, NJ, p. 264-320 Malasarn, D, Saltikov W, Campbell KM, Santini JM, Hering JG, Newman DK (2004) arrA is a reliable marker for As(V) respiration. Science 306:455-455 McKay DS, Gibson Jr. EK, Thomas-Keprta KL, Vali H, Romanek CS, Clemett SJ, Chiller XDF, Maechling CR, Zare RN (1996) Search for past life on Mars: possible relic biogeneic activity in martian meteorite ALH84001. Science 273:924-930 Metcalf WM, Zhang JK, Apolinario E, Sowers KR, Wolfe RS (1997) A genetic system for Archea of the genus Methanosarcina: liposome-mediated transformation and construction of shuttle vectors: Proc Natl Acad Sci USA 94:2626-2631 Michaelis W, Seifert R, Nauhaus K, Treude T, Thiel V, Blumenberg M, Knittel K, Gieseke A, Peterknecht K, Pape T, Boetius A, Amann R, Jorgensen BB, Widdel F, Peckmann J, Pimenov NV, Gulin MB (2002) Microbial reefs in the Black Sea fueled by anaerobic oxidation of methane. Science 297:1013-1015 Myers JM, Myers CR (2002) Genetic complementation of an outer membrane cytochrome omcB mutant of Shewanella putrefaciens MR-1 requires omcB plus downstream DNA. Appl Environ Microbiol 68:2781-93 Nealson KH, Scott J (2004) Ecophysiology of the Genus Shewanella. In: The Prokaryotes. Dworkin M (ed) Springer-Verlag, New York. http://141.150.157.117:8080/prokPUB/chaphtm/394/COMPLETE.htm

26

Newman & Gralnick

Nelson KE, Methé B (2005) Metabolism and genomics: adventures derived from complete genome sequencing. Rev Mineral Geochem 59:279-294 Newman DK, Ahmann D, Morel FMM (1998) A brief review of microbial arsenate respiration. Geomicrobiology 15:255-68 Newman DK, Kennedy EK, Coates JD, Ahmann D, Ellis DJ, Lovley DR, Morel FMM (1997) Dissimilatory arsenate and sulfate reduction in Desulfotomaculum auripigmentum sp. nov. Arch Microbiol 168:380-388 Newman DK, Kolter R (2000) A role for excreted quinones in extracellular electron transfer. Nature 405:94-97 Nixon JEJ, Wang A, Field J, Morrison HG, McArthur AG, Sogin ML, Loftus BJ, Samuelson J (2002) Evidence for lateral transfer of genes encoding ferredoxins, nitroreductases, NADH oxidase, and alcohol dehydrogenase 3 from anaerobic prokaryotes to Giardia lamblia and Entamoeba histolytica. Eukaryotic Cell 1:181-190 Orphan J, House CH, Hinrichs KU, McKeegan KD, DeLong EF (2001) Methane-consuming archaea revealed by directly coupled isotopic and phylogenetic analysis. Science 293:484-487 Pace NR (1997) A molecular view of microbial diversity and the biosphere. Science 276:734-740 Pardee AB, Jacob F, Monod J (1959) The genetic control and cytoplasmic expression of “inducibility” in the synthesis of B-galactosidase by E. coli. J Mol Biol 1:165-178 Peck RF, Dassarma S, Krebs MP (2000) Homologous gene knockout in the archaeon Halobacterium salinarum with ura3 as a counterselectable marker. Mol Micro 35:667-676 Pritchett MA, Metcalf WM (2005) Genetic, physiological and biochemical characterization of multiple methanol methyltrasferase isozymes in Methanosarcina acetivorans C2A. Mol Micro 56:1183-1194 Ram RJ, VerBerkmoes NC, Thelen MP, Tyson GW, Baker BJ, Blake RC, Shah M, Hettich RL, Banield JF (2005) Community proteomics of a natural microbial bioilm. Science 308:1915-1920 Riesenfeld CS, Schloss PD, Handelsman J (2004) Metagenomics: genomic analysis of microbial communities. Annu Rev Genet 38:525-552 Salgado H, Moreno-Hagelsieb G, Smith TF, Collado-Vides J (2000) Operons in Escherichia coli: genomic analyses and predictions. Proc Natl Acad Sci USA 97:6652-7 Saltikov CW, Cifuentes A, Venkateswaran K, Newman DK (2003) The ars detoxiication system is advantageous but not required for As(V) respiration by the genetically tractable Shewanella species strain ANA-3. Appl Environ Microbiol 69:2800-2809 Saltikov CW, Newman DK (2003) Genetic identiication of a respiratory arsenate reductase. Proc Natl Acad Sci USA 100:10983-10988 Schopf JW (1993) Microfossils of the early archean apex chert - new evidence of the antiquity of life. Science 260:640-646 Schrenk MO, Edwards KJ, Goodman RM, Hamers RJ, Banield JF (1998) Distribution of thiobacillus ferrooxidans and leptospirillum ferrooxidans: implications for generation of acid mine drainage. Science 279:1519-1522 Shen YA, Buick R, Canield DE (2001) Isotopic evidence for microbial sulphate reduction in the early Archaean era. Nature 410:77-81 Shuman H (2003) Just toothpicks and logic: How some labs succeed at solving complex problems. J Bacteriol 185:387-390 Simonson AB, Servin JA, Skophammer RG, Herbold CW, Rivera MC, Lake JA (2005) Decoding the genomic tree of life. Proc Natl Acad Sci USA 102:6608-6613 Summons RE, Jahnke LL, Hope JM, Logan GA (1999) 2-Methylhopanoids as biomarkers for cyanobacterial oxygenic photosynthesis. Nature 400:554-557 Tyson GW, Chapman J, Hugenholz P, Allen EE, Ram RJ, Richardwon PM, Solovyev VV, Rubin EM, Rokhsar DS, Banield JF (2004) Community structure and metabolism through reconstruction of microbial genomes from the environment. Nature 428:37-43 van Waasbergen LG, Hildebrand M, Tebo BM (1996) Identiication and characterization of a gene cluster involved in manganese oxidation by spores of the marine Bacillus sp. strain SG-1. J Bacteriol 178:35173530 Watters WA, Grotzinger JP (2001) Digital reconstruction of calciied early metazoans, terminal Proterozoic Nama Group, Namibia. Paleobiology 27:159-171 Weiss BP, Kim SS, Kirschvink JL, Kopp RE, Sankaran M, Kobayashi A, Komeili A (2004) Magnetic tests for magnetosome chains in Martian meteorite ALH84001. Proc Natl Acad Sci USA 101:8281-8284 Whitaker RJ, Banield JF (2005) Population dynamics through the lens of extreme environments. Rev Mineral Geochem 59:259-278 Xiong J, Fischer WM, Inoue K, Nakahara M, Bauer CE (2000) Molecular evidence for the early evolution of photosynthesis. Science 289:1724-30 Zhang JH, Dawes G, Stemmer WPC (1997) Directed evolution of a fucosidase from a galactosidase by DNA shufling and screening. Proc Natl Acad Sci USA 94:4504-4509

Reviews in Mineralogy & Geochemistry Vol. 59, pp. 27-52, 2005 Copyright © Mineralogical Society of America

3

Enzymology of Electron Transport: Energy Generation With Geochemical Consequences Thomas J. DiChristina School of Biology Georgia Institute of Technology Environ Sci Technol Building Atlanta, Georgia, 30332, U.S.A. [email protected]

Jim K. Fredrickson and John M. Zachara Fundamental Sciences Division Paciic Northwest National Laboratory P.O. Box 999, MSIN P7-50 Richland, Washington, 99352, U.S.A. [email protected]

[email protected]

INTRODUCTION Dissimilatory metal-reducing bacteria (DMRB) are important components of the microbial community residing in redox-stratiied freshwater and marine environments. DMRB occupy a central position in the biogeochemical cycles of metals, metalloids and radionuclides, and serve as catalysts for a variety of other environmentally important processes including biomineralization, biocorrosion, bioremediation and mediators of ground water quality. DMRB are presented, however, with a unique physiological challenge: they are required to respire anaerobically on terminal electron acceptors which are either highly insoluble (e.g., Fe(III)- and Mn(IV)-oxides) and reduced to soluble end-products or highly soluble (e.g., U(VI) and Tc(VII)) and reduced to insoluble end-products. To overcome physiological problems associated with metal and radionuclide solubility, DMRB are postulated to employ a variety of novel respiratory strategies not found in other gram-negative bacteria which respire on soluble electron acceptors such as O2, NO3−, SO42−, and CO2. The novel respiratory strategies include 1) direct enzymatic reduction at the outer membrane, 2) electron shuttling pathways and 3) metal solubilization by exogenous or bacterially-produced organic ligands followed by reduction of soluble organic-metal compounds. The irst section of this chapter highlights the latest indings on the enzymatic mechanisms of metal and radionuclide reduction by two of the most extensively studied DMRB (Geobacter and Shewanella), with particular emphasis on electron transport chain enzymology. These advances have drawn signiicantly upon genomic data for isolated microorganisms from the genera Geobacter and Shewanella (see chapter by Nelson and Methé 2005). The second section emphasizes the geochemical consequences of DMRB activity, including the direct and indirect effects on metal solubility, the reductive transformation of Fe- and Mn-containing minerals, and the biogeochemical cycling of metals at redox interfaces in chemically stratiied environments.

ENZYMATIC BASIS OF IRON AND MANGANESE REDUCTION The electron transport systems of gram-negative bacteria are generally described as inner membrane (IM)-associated electron and proton carriers that 1) mediate electron transfer from 1529-6466/05/0059-0003$05.00

DOI: 10.2138/rmg.2005.59.3

DiChristina, Fredrickson, Zachara

28

primary donor to terminal electron acceptor and 2) conserve energy released during electron transfer to the generation of ATP (Madigan and Martinko 2006). Figure 1 displays the electron transport chain enzymology of Escherichia coli respiring high concentrations of dissolved O2 as electron acceptor. The E. coli electron transport system is modular in design with a membrane-soluble quinone pool (Q) linking dehydrogenase complexes at the head end with terminal reductase complexes at the terminus. Dehydrogenase complexes include electron donor-speciic oxido-reductases (e.g., NADH dehydrogenase) that couple oxidation of speciic electron donors to reduction of a series of membrane-associated electron carriers arranged in order of increasingly more positive electric potential (E0c). These electron carriers include lavoproteins (Fp) and FeS proteins that translocate protons across the IM to the periplasm and direct electrons to the Q pool, respectively. Reduced Q is subsequently protonated to QH2 at the inner aspect of the IM. QH2 carries protons across the IM to the periplasm and transfers electrons to cytochrome b556 and b562, two components of the terminal reductase complex that transfer electrons to cytochrome o (topologically located at the inner aspect of the IM) and ultimately to O2. Cytochrome o catalyzes both the translocation of a proton across the IM to the periplasm and the terminal reduction of O2 to H2O. A proton motive force (PMF) is generated by 1) proton translocation across the IM by dehydrogenase complexes, QH2 and cytochrome o, and 2) proton consumption during the terminal reduction of O2 to H2O by cytochrome o. PMF generated in this manner drives ATP synthesis as protons are translocated back into the cytoplasm through an IM-localized ATPase, catalyzing the phosporylation of ADP to ATP (Madigan and Martinko 2006). Fe(III)- and Mn(IV)-respiring DMRB, on the other hand, are presented with a unique physiological problem: they are required to respire anaerobically on terminal electron acceptors found largely in crystalline form or as amorphous (oxy)hydroxide particles presumably unable to contact IM-localized electron transport systems. A DMRB culture actively respiring solid Mn(IV) oxides as anaerobic electron acceptor is displayed in Figure 2. To overcome the problem of respiring solid electron acceptors, Fe(III)- and Mn(IV)-respiring DMRB are postulated to employ a variety of novel respiratory strategies not found in other gram-negative bacteria that respire on soluble electron acceptors such as O2, NO3−, SO42− and CO2 including 1) direct enzymatic reduction of solid Fe(III) and Mn(IV) oxides via outer membrane (OM)-localized metal reductases (Myers and Myers 1992, 2003a; Beliaev and Saffarini 1998; DiChristina et al. 2002) 2) a two-step, electron shuttling pathway in which exogenous electron shuttling compounds (e.g., humic acids, melanin, phenazines, antibiotics, AQDS) are irst enzymatically reduced and subsequently chemically oxidized by the solid Fe(III) and Mn(IV) oxides in a second (abiotic) electron transfer reaction (Lovley et al. 1996;

2H+




NADH + H+

2H+ 2e-

FeS

NAD+

2e-

Q QH2 2H+

H+

H+ 2e- b556 2eb562 1/

2O2

o + 3H+

H2O ADP H+ ATP

ELECTRON TRANSPORT SYSTEM

ATPase

Figure 1. Electron transport and proton translocation processes of the E. coli aerobic respiratory chain at high O2 concentrations with NADH2 as electron donor. Fp, lavoprotein; FeS, iron-sulfur protein; Q, quinone pool; b556 and b562, b-type cytochromes; o, cytochrome o.

Enzymology of Electron Transport

29

Figure 2. DMRB Shewanella putrefaciens strain 200 actively respiring solid Mn(IV) oxides as anaerobic electron acceptor. (A) Anaerobic cell suspensions at the beginning (tube on left side) and end (tube on right side) of a 24-hour anaerobic incubation period (note color change indicative of reductive dissolution of black Mn(IV) particles to clear, soluble reduced Mn), (B) phase contrast micrograph of cells coating the surface of a Mn(IV) oxide particle at the beginning of the anaerobic growth period, (C) epiluorescence micrograph of same ield of view as in (B) with acridine orange-stained cells, (D) epiluorescence micrograph of acridine orange-stained cells at the end of the 24-hour anaerobic incubation period (note the absence of the solid Mn(IV) particles).

Coates et al. 1998, 2002; Newman and Kolter 2000; Turick et al. 2002; Hernandez et al. 2004; DiChristina et al. 2005) 3) an analogous two-step reduction pathway involving endogenous, electron shuttling compounds (Newman and Kolter 2000; Saffarini et al. 2002) and 4) a twostep, Fe(III) solubilization-reduction pathway in which solid Fe(III) oxides are irst dissolved by exogenous or bacterially-produced organic complexing ligands, followed by uptake and reduction of the soluble organic Fe(III) forms by periplasmic Fe(III) reductases (Arnold et al. 1988; Lovley and Woodward 1996; Pitts et al. 2003). Although the number of DMRB species continues to increase rapidly and has now reached nearly 100 (Lovley et al. 2004), the enzymatic basis of electron transfer to metals has been most extensively studied in metalrespiring members of the genera Geobacter and Shewanella. The following section highlights the latest indings on the enzymatic basis of Fe(III) and Mn(IV) reduction by Geobacter and Shewanella, with particular emphasis on electron transport chain enzymology.

Direct enzymatic reduction at the outer membrane Shewanella and Geobacter catalyze the direct enzymatic reduction of solid Fe(III) and Mn(III,IV) oxides via an electron transport chain arranged in a canonical, highly branched fashion. Hydrogenase and lavin-containing dehydrogenase complexes of both Shewanella and Geobacter (Myers and Myers 1993a; Lloyd et al. 2000) oxidize a variety of electron donors (e.g., H2, NAD(P)H) and transfer electrons to a menaquinone pool (Myers and Nealson 1990; Myers and Myers 1993b; Nevin and Lovley 2002; Saffarini et al. 2002). In S. oneidensis,

30

DiChristina, Fredrickson, Zachara

menaquinol diffuses within the IM to the quinol oxidation site of CymA, a 21 kDa tetraheme cytochrome c that oxidizes menaquinol and is thought to transfer electrons to MtrA, a 32 kDa decaheme cytochrome c located in the periplasm (Myers and Myers 2000; Schwalb et al. 2003). In G. sulfurreducens, PpcB (a 36 kDa diheme cytochrome c) essentially carries out the same function as CymA (although CymA and PpcB display little or no amino acid sequence homology), oxidizing the menaquinol pool and transferring electrons to PpcA, a 10 kDa triheme cytochrome c located in the G. sulfurreducens periplasm (Lloyd 2003). As with PpcB, G. sulfurreducens PpcA does not display signiicant sequence similarity to any S. oneidensis c-type cytochromes, suggesting that they have a different evolutionary origin (Lloyd 2003). Electrons from the S. oneidensis menaquinol pool are transferred to one of four electronaccepting, c-type hemes within CymA, followed by inter-heme electron transfer (according to decreasing heme redox potential) until a inal transfer is made to subsequent electron carriers in the periplasmic space (Harada et al. 2002). cymA-deicient mutants of S. oneidensis are unable to reduce NO3−, Fe(III), Mn(IV) or fumarate as electron acceptor (Myers and Myers 1997), an indication that CymA is a central branchpoint of the S. oneidensis electron transport system. ppcB-deicient mutants of G. sulfurreducens, on the other hand, are unable to reduce Fe(III), but retain the ability to reduce fumarate (Butler 2003). The principles of, and rationale for genetic manipulation (including generation of metal respiration-deicient mutants) is discussed in the chapter by Newman and Gralnick (2005). Electron transport from MtrA in S. oneidensis and PpcA in G. sulfurreducens to solid Fe(III) oxides is postulated to proceed via an electron transport chain that spans the periplasmic space and terminates on the outside face of the OM (Myers and Myers 1993a; Leang et al. 2003). Electron transfer to solid Fe(III) and Mn(IV) in G. sulfurreducens proceeds via OmcB, an 87 kDa, 12-heme cytochrome c tentatively assigned to the inner aspect of the OM (Leang et al. 2003). Correspondingly, Fe(III)-grown G. sulfurreducens cells display higher levels of omcB transcripts (Chin et al. 2004; Methé et al. 2005). All G. sulfurreducens OM cytochromes, however, are not necessarily involved in electron transport to Fe(III) since OmcC, an OM cytochrome displaying 73% identity to OmcB, is not required for Fe(III) reduction (Leang et al. 2003; Leang and Lovley 2005) and correspondingly, Fe(III)-grown G. sulfurreducens cells do not display higher levels of omcC transcripts (Chin et al. 2004). The terminal electron transfer step to solid Fe(III) and Mn(IV) oxides in G. sulfurreducens is postulated to be catalyzed by either OmcD or OmcE, two c-type cytochromes that may be exposed on the cell surface (Methé et al. 2005). Results of DNA microarray analysis (as described in the chapter by Nelson and Methé 2005) will identify electron transport chain components with elevated transcript levels during growth on speciic electron acceptors. Similar to Geobacter, the Shewanella OM proteins involved in terminal steps of electron transfer to solid Fe(III) and Mn(IV) oxides have not been deinitively identiied, yet most likely include several c-type cytochromes (Myers and Myers 1992, 2003a). Fe(III) reduction activity is detected in wild-type Shewanella OM fractions (Myers and Myers 1993a), an activity that is severely impaired in Shewanella mutants lacking OM proteins, including several multi-heme c-type cytochromes. The S. oneidensis genome encodes 42 predicted c-type cytochromes (Heidelberg et al. 2002), including those in the mtrDEF-omcA-mtrCAB gene cluster. MtrA and MtrD are decaheme c-type cytochromes that display 99% similarity (Pitts et al. 2003), suggesting they may provide complementary function. MtrD is OM-associated, but may be oriented toward the periplasm (Pitts et al. 2003) and therefore not in position to contact solid Fe(III) directly. MtrB is a putative beta-barrel protein postulated to be involved in OM localization of the c-type cytochromes OmcA and MtrC that are involved in electron transfer to Fe(III) and Mn(IV) (Beliaev and Saffarini 1998; Myers and Myers 2002). mtrB mutants display a complete inability to reduce Mn(IV) and are severely, but not completely, impaired in Fe(III) reduction activity, yet retain the ability to reduce all other electron acceptors (Beliaev

Enzymology of Electron Transport

31

and Saffarini 1998). MtrC is an OM-localized decaheme, c-type cytochrome required for both Fe(III) and Mn(IV) reduction activity. OmcA, on the other hand, is an OM decaheme c-type cytochrome involved in electron transport to Mn(IV) and not Fe(III) (Myers and Myers 2001). omcA-deicient mutants reduce Mn(IV) at 45% wild-type rates. Interestingly, mtrC overexpression in an omcA-deicient mutant restores Mn(IV) reduction activity to greater than wild-type rates, an indication that the functional roles of MtrC and OmcA at least partially overlap in the electron transport pathway to Mn(IV) (Myers and Myers 2003b). The functions of MtrC and OmcA in Fe(III) reduction remain unclear, yet they are postulated to be major components of the Fe(III) terminal reductase. Some of the most convincing genetic evidence supporting the hypothesis that Shewanella localizes Fe(III) and Mn(IV) reductases to the OM has been derived from genetic studies with S. putrefaciens (DiChristina and DeLong 1994; DiChristina et al. 2002). Genetic mutant complementation analyses (as outlined in chapter by Newman and Gralnick 2005) indicated that a 23.3 kb S. putrefaciens wild-type DNA fragment conferred Fe(III) reduction activity to a set of 10 Fe(III) reduction-deicient mutants of S. putrefaciens. The smallest complementing DNA fragment contained one open reading frame (ORF) whose translated product displayed 87% sequence similarity to Aeromonas hydrophila ExeE, a member of the GspE family of proteins found in Type II protein secretion systems. GspE insertional mutants (constructed by targeted replacement of wild-type gspE with an insertionally inactivated gspE construct) are unable to respire anaerobically on solid Fe(III) or Mn(IV) oxides, yet retain the ability to respire all other electron acceptors including soluble complexes of Fe(III) and Mn(III) (Kostka et al. 1995; Pitts et al. 2003). Nucleotide sequence analysis of regions lanking gspE revealed one partial and two complete ORFs whose translated products displayed 55-70% sequence similarity to the GspD-G homologs of other Type II protein secretion systems. A heme-containing protein complex displaying Fe(III) reductase activity is present in the peripheral proteins loosely attached to the outside face of the wild-type OM, yet is missing from this location in the gspE mutants. Membrane fractionation studies with the wild-type strain support this inding: the heme-containing Fe(III) reductase complex is detected in the OM but not the IM or cytoplasmic fractions. These indings provide the irst genetic evidence linking anaerobic Fe(III) and Mn(IV) respiration to Type II protein secretion and provide additional biochemical evidence supporting OM localization of Shewanella Fe(III) and Mn(IV) reductases (DiChristina and DeLong 1994; DiChristina et al. 2002). Gram-negative bacteria secrete soluble exoproteins to the cell periphery or exterior via ive known protein secretion systems (Desvaux et al. 2004). Type II protein secretion is part of the main terminal branch of the general secretory (GSP) pathway (Pugsley 1993; Pugsley et al. 1997; Filloux 2004) and is generally comprised of 12-to-16 proteins encoded by a contiguous cluster of moderately-to-highly conserved pul (or gsp) genes, usually in the same order. Pullulanase secretion by the plant cell wall-degrading microorganism Klebsiella oxytoca is one of the best characterized Type II protein secretion systems and a working model for pullulanase secretion has been proposed (Pugsley et al. 1997). Nascent pullulanase is irst directed into and across the cytoplasmic membrane where it folds and is transiently anchored to the periplasmic aspect of the cytoplasmic membrane. After processing (signal peptide cleavage, disulide bond formation, fatty acylation), the mature pullulanase is guided across the periplasmic space by the Type II secretion pseudopilus (GspG, H, I, J complex) and interacts with the OMassociated, multimeric GspD channel. Pullulanase is subsequently attached to the outside face of the outer membrane via a fatty acid tail. The ferE homolog, gspE, is postulated to encode a secretion ATPase that drives the secretion process, including the rapid polymerization and depolymerization reactions associated with pseudopilus extension and retraction (Filloux 2004). Peripherally attached pullulanase cleaves alpha-1,6 linkages in branched maltodextrin polymers such as glycogen or amylopectin of plant cell wall material, thereby releasing linear

32

DiChristina, Fredrickson, Zachara

dextrins for bacterial cell uptake and metabolism. Based on the K. oxytoca Type II pullulanase secretion model and the previously reported involvement of S. putrefaciens outer membrane proteins in dissimilatory Fe(III) and Mn(IV) reduction, it has been postulated that the Fe(III) and Mn(IV) respiratory deiciencies of Type II protein secretion mutants are due to their inability to secrete Fe(III) and Mn(IV) terminal reductases to the outside face of the S. putrefaciens outer membrane (DiChristina et al. 2002). A working model of the direct enzymatic pathway for reduction of solid Fe(III) oxides in Shewanella is displayed in Figure 3.

Electron shuttling pathways A variety of Fe(III)-respiring DMRB, including Shewanella and Geobacter, can employ redox-active compounds (e.g., humic acids, melanin, phenazines, antibiotics, AQDS) as exogenous electron shuttles to reduce extracellular Fe(III) oxides (Lovley et al. 1996). The Fe(III) and Mn(IV) reduction-deiciencies of Shewanella Type II protein secretion mutants are rescued by addition of AQDS (DiChristina et al. 2005). S. oneidensis gspD insertional mutants are unable to respire anaerobically on solid Fe(III) or Mn(IV), yet retain the ability to respire all other electron acceptors, including AQDS. The ability to respire 50 mM solid Fe(III) or Mn(IV) is rescued in the S. oneidensis gspD insertional mutants by addition of 50 μM AQDS, an indication that the AQDS electron shuttling pathway is able to overcome the defect in the Type II protein secretion-linked pathway for respiration on solid Fe(III) and Mn(IV). AQDS is toxic to Shewanella cells above a critical threshold concentration and the eflux pump protein TolC protects Shewanella cells from AQDS toxicity by mediating AQDS eflux (Shyu et al. 2002). Electron transfer to AQDS also requires the OM protein MtrB, although its role in AQDS reduction remains unknown (Shyu et al. 2002). Solid Fe(III) reduction by Shewanella is also stimulated by redox-active antibiotics and phenazines (Hernandez et al. 2004). Phenazines are similar in structure to AQDS and function as electron shuttles between Shewanella cells and solid Fe(III) oxides. Redox-active antibiotics (e.g., bleomycin) also function as shuttles for extracellular electron transfer to solid electron acceptors. Bacterially-produced phenazines (e.g., synthesized by Pseudomonas chlororaphis PCL1391) stimulate Fe(III) reduction by bacteria unable to produce them (e.g., S. oneidensis MR-1) (Hernandez et al. 2004). In addition, melanin (a humic acid-like compound synthesized by S. algae BrY in the presence of high concentrations of tyrosine) will enhance rates of Fe(III) oxide reduction (Turick et al. 2002). Melanin may have a dual function by acting as both an electron shuttle and an Fe(II)-complexing agent that prevents Fe(II) from adsorbing to and blocking Fe(III) oxide surface sites. A working model of the exogenous electron shuttling pathway for AQDS-mediated reduction of solid Fe(III) oxides by Shewanella is displayed in Figure 4. Shewanella (and Geothrix fermentans) may also synthesize and release endogenous compounds that shuttle electrons to solid Fe(III) oxides (Newman and Kolter 2000; Nevin and Lovley 2002). Fe(III)-reducing G. metallireducens, on the other hand, does not appear to produce endogenous electron shuttles (Nevin and Lovley 2000). S. algae BrY produces melanin as a soluble electron shuttle for reduction of solid Fe(III) oxides (Turick et al. 2002). S. algae-produced melanin oxidizes c-type cytochromes at the cell surface and reduces solid Fe(III) oxides extracellularly (Turick et al. 2002). S. oneidensis MR-1 mutants defective in menC (encoding o-succinylbenzoic acid synthase) are deicient in menaquinone production and are unable to reduce AQDS, fumarate, thiosulfate, sulite, DMSO or solid Fe(III) and Mn(IV) (Newman and Kolter 2000). Menaquinone is detected in the spent media of the wild-type strain, but not the menC mutants. Spent medium from the wild-type strain restores Fe(III) reduction activity to the menC mutant, while spent media from menC mutant does not. S. oneidensis MR-1 mutants defective in either menD or menB (encoding components of the menaquinone biosynthetic pathway) are also unable to reduce solid Fe(III) oxides (Saffarini et al. 2002). Vitamin K2 (a menaquinone analog) restores the ability of the menD or menB

Enzymology of Electron Transport

33 Secretion of Fe(III) terminal reductase complex

Fe(II)

Solid Fe(III)

OM D S

Fe(III) terminal reductase complex

+

H

e-

ec-type cytochromes e-

H+ Menaquinone Pool

D S

G H I J

H+

IM

CymA L

e- donor

F

M C K

N O

E

Dehydrogenase

ADP H+ ATP

ELECTRON TRANSPORT SYSTEM

ATPase

TYPE II SECRETION APPARATUS

Figure 3. Working model for type II protein secretion-linked, direct enzymatic reduction of solid Fe(III)-oxides at the outer membrane.

Solid Fe(III) Fe(II)

AQDS

AH2DS OM 2e-

c-type cytochromes H+

e-

H+ Menaquinone Pool

AQDS Terminal reductase eH+

e-

IM

CymA

e- donor Dehydrogenase ELECTRON TRANSPORT SYSTEM

ADP

H+

ATP

ATPase

Figure 4. Working model for electron shuttling pathway with AQDS as electron shuttle.

34

DiChristina, Fredrickson, Zachara

mutants (and corresponding membrane fractions) to reduce either Fe(III) or Mn(IV) (Saffarini et al. 2002). It should be noted that the endogenous electron shuttle pathway may be the consequence of cell lysis and inadvertent spillage of menaquinol into the culture medium. Since shuttles can undergo redox cycling they can be effective at low concentrations and therefore even a small fraction of cell lysis could have a signiicant effect. Lipid-soluble menaquinol or vitamin K2 then diffuses into bacterial membranes and functionally complements the menB, C or D mutants. Deinitive evidence on the identity of the endogenous electron shuttle requires further research and will be challenging as exceedingly low concentrations may be involved.

Fe(III) solubilization by exogenous or bacterially-produced organic ligands followed by reduction of soluble organic-Fe(III) A strong electrochemical signal indicative of soluble organic-Fe(III) is detected in a variety of marine and freshwater environments with Au/Hg voltammetric microelectrodes (Taillefert et al. 2002). Soluble organic-Fe(III) may therefore represent a dominant, yet under appreciated electron acceptor in anaerobic aquatic systems. Microbial Fe(III) reduction rates are higher with soluble organic-Fe(III) in pure cultures of S. putrefaciens (Arnold et al. 1988) and in freshwater sediments amended with Fe(III)-chelating compounds such as nitrilotriacetic acid (Lovley and Woodward 1996). S. putrefaciens reduces soluble organic-Fe(III) complexes at rates three orders of magnitude faster than amorphous or crystalline Fe(III) forms (Arnold et al. 1988). The mechanism of formation of soluble organic-Fe(III) generally involves non-reductive dissolution of amorphous Fe(III) oxides by multidentate organic ligands (forming mononuclear complexes with the Fe(III) oxides) at circumneutral pH. The strength of binding between Fe(III) and the complexing organic ligands inluences soluble organic-Fe(III) reduction activity: organic ligands with strong Fe(III)-binding capability decrease (and in some cases totally inhibit) Fe(III) reduction activity by S. putrefaciens (Haas and DiChristina 2002). Some Fe(III)-reducing bacteria such as S. algae BrY and G. fermentans generate relatively high concentrations of soluble organic-Fe(III) in the absence of exogenous chelating compounds, an indication that such bacteria synthesize and release organic ligands to solubilize Fe(III) prior to reduction (Nevin and Lovley 2002). Soluble organic-Fe(III) is detected electrochemically in S. oneidensis and S. putrefaciens cultures incubated anaerobically with either ferrihydrite or goethite (Taillefert and DiChristina 2005). Detection of soluble organicFe(III) prior to detection of Fe(II), suggests that soluble organic-Fe(III) is an intermediate in the reduction of solid Fe(III) oxides. Since lactate is the only organic ligand added to the Shewanella batch cultures and lactate-Fe(III) complexes do not react with Au/Hg electrodes, electrochemical detection of soluble organic-Fe(III) suggests that Shewanella synthesizes and releases organic ligands that complex and dissolve Fe(III) prior to reduction. The identity of the bacterially-produced, Fe(III)-solubilizing organic ligands remains unknown. A respiration-linked, soluble organic-Fe(III) terminal reductase has yet to be deinitively identiied. As described above, Shewanella Type II protein secretion mutants are unable to reduce solid Fe(III) oxides, yet retain the ability to respire all other electron acceptors, including soluble organic-Fe(III). This inding suggests that soluble organic-Fe(III) may be reduced by terminal reductases located in subcellular compartments other than the OM. The S. oneidensis decaheme c-type cytochrome MtrA is a candidate terminal reductase for soluble organic-Fe(III): MtrA is located in the S. oneidensis periplasm and displays soluble organicFe(III) reductase activity when expressed in E. coli (Pitts et al. 2003). The requirement for MtrA in anaerobic respiration of soluble organic-Fe(III), however, has yet to be demonstrated in vivo. In S. frigidimarina, the transcriptional activator IfcR is translated in the presence of soluble organic-Fe(III) and is essential for expression of ifcO and ifcA. IfcO is a putative OM beta-barrel protein postulated to function as a soluble organic-Fe(III) transporter. IfcA is a lavin-containing c-type cytochrome with a small (10 kDa) tetraheme cytochrome domain that

Enzymology of Electron Transport

35

displays soluble organic-Fe(III) reductase activity (Pitts et al. 2003). A working model of the two-step, Fe(III) solubilization-reduction pathway in Shewanella is displayed in Figure 5.

ENZYMATIC BASIS OF URANIUM REDUCTION Members of the genera Shewanella (Lovley et al. 1991), Desulfovibrio (Lovley et al. 1993), Clostridium (Francis et al. 1994), Geobacter (Caccavo et al. 1992), Thermus (Kieft et al. 1999), Pyrobaculum (Kashei and Lovley 2000), and Desulfosporosinus (Suzuki et al. 2002) display enzymatic U(VI) reduction activity. Shewanella and Geobacter enzymatically reduce U(VI) to U(IV) via a respiratory process that supports anaerobic growth. Although several puriied c-type cytochromes display U(VI) reductase activity in vitro, a respirationlinked, U(VI) terminal reductase has yet to be deinitively identiied in vivo. Enzymatic U(VI) reduction activity is affected by U(VI) chemical speciation, electron donors, and competing electron acceptors. In the following section, the most recent indings on the enzymatic basis of U(VI) reduction by Shewanella and Geobacter are presented along with a discussion of the environmental factors affecting enzymatic U(VI) reduction activity.

Involvement of c-type cytochromes in enzymatic U(VI) reduction Cytochrome c3 of several Desulfovibrio species is involved in electron transfer to U(VI). Cytochrome c3 of U(VI)-reducing (but non-respiring) Desulfovibrio vulgaris Hildenborough displays U(VI) reductase activity in vitro with H2 as electron donor (Lovley et al. 1993). Cytochrome c3 mutants of D. desulfuricans strain G20 are unable to reduce U(VI) with H2 as

Solid Fe(III)

Soluble Organic-Fe(III)

Fe(III)

Organic Ligand

Fe(III)

Fe(II) OM

Soluble Fe(III) Organic-Fe(III)

H+

e-

H+ Menaquinone Pool

c-type cytochromes e-

Fe(II) Organic Ligand eSoluble Fe(III) reductase

H+ IM

CymA

e- donor Dehydrogenase ADP ELECTRON TRANSPORT SYSTEM

H+

ATP Organic Ligand

ATPase

Figure 5. Working model for Fe(III) solubilization-reduction pathway with endogenous organic ligand as Fe(III)-chelating compound.

36

DiChristina, Fredrickson, Zachara

electron donor and are partially impaired in U(VI) reduction activity with lactate or pyruvate as electron donor (Payne et al. 2002). After growth of wild-type D. desulfuricans strain G20 in medium containing uranyl acetate, cytochrome c3 is tightly associated with insoluble U(IV) particles (uraninite) found in the periplasm (Payne et al. 2004). Cytochrome c7 of G. sulfurreducens also displays U(VI) reductase activity in vitro, however, mutants deicient in either cytochrome c3 or c7 retain U(VI) reduction activity in vivo (Lloyd et al. 2003). These indings suggest that either cytochrome c3 and c7 are not the physiological U(VI) reductases in G. sulfurreducens or that the electron transport pathway to U(VI) is highly branched and consists of multiple U(VI) terminal reductases. The highly branched nature of the U(VI) reduction pathway in G. sulfurreducens is relected by the inding that Fe(III) reduction-deicient ppcA mutants (see above) are also deicient in U(VI) reduction activity (Lloyd et al. 2003). A genetic complementation system has recently been developed to examine the enzymatic mechanism of U(VI) reduction by S. putrefaciens (Wade and DiChristina 2000). S. putrefaciens respiratory mutants unable to reduce U(VI) have been isolated and tested for the ability to respire on a suite of alternate compounds as electron acceptor, including oxygen O2, NO3−, fumarate, trimethylamine-N-oxide (TMAO), dimethyl sulfoxide (DMSO), Mn(IV), Fe(III), chromate (Cr(VI)), arsenate (As(V)), selenite (Se(IV)), pertechnetate (Tc(VII)), thiosulfate (S(II)), and sulite (S(IV)) (Wade and DiChristina 2000). All U(VI) reduction-deicient mutant strains also lacked the ability to respire NO2−. In particular, U(VI) reduction-deicient mutant strain U14 retained the ability to respire all electron acceptors except U(VI) and NO2−. These results suggest that the electron transport chains terminating with the reduction of NO2− and U(VI) share common respiratory components.

Effect of U(VI) chemical speciation on enzymatic U(VI) reduction activity U(VI) chemical speciation is an important variable controlling enzymatic U(VI) reduction activity. In oxidizing aqueous environments at circumneutral pH (and in the absence of phosphate), U(VI) is found as soluble uranyl ion (UO22+), often in carbonate complexed form (e.g., UO2(CO3)22−, UO2(CO3)34−, CaUO2(CO3)20) or as crystalline solids such as metaschoepite (UO3·2H2O), uranyl phosphates and uranyl silicates. U(IV) precipitates in reducing environments as uraninite (UO2). The relative insolubility of U(IV) (10−8 M at pH > 5; Rai et al. 1990) compared with U(VI) is the basis of alternate bioremediation strategies (Lovley et al. 1991). The uranyl ion readily complexes with either inorganic (e.g., hydroxyl, carbonate, phosphate, sulfate and calcium) or organic (e.g., acetate, malonate, citrate and oxalate) ligands in aqueous solution (Grenthe 1992), and complexation markedly enhances its solubility. The type of complexing ligand changes the reduction potential of U(VI) thus affecting enzymatic reduction activity. In terms of reduction potential, hydroxo complexes are the most easily reduced forms of complexed U(VI), while complexation by carbonate decreases the reduction potential of U(VI). Complexation of U(VI)-carbonate by calcium (forming CaUO2-CO3 complexes) decreases the reduction potential to such an extent that enzymatic U(VI) reduction by S. putrefaciens CN32 nearly ceases (Brooks et al. 2003). Enzymatic U(VI) reduction activity by D. desulfuricans and G. sulfurreducens is also inhibited by formation of Ca-UO2-CO3 complexes. The effect of Ca2+ complexation on enzymatic U(VI) reduction activity are speciic to U(VI) reduction since the enzymatic reduction of fumarate and Tc(VII) activities are not inhibited by Ca2+. In the absence of carbonate or at pH < 6 in the presence of carbonate, organic ligands bound to U(VI) also dramatically impact enzymatic U(VI) reduction activity. Citrate, for example, binds U(VI) with varying strength as a function of pH (Pasilis and Pemberton 2003). At pH > 6 and at low citrate concentrations, the highly soluble (UO2)3Cit2 species predominates over the (UO2)2Cit2 species. S. alga BrY reduces U(VI) bound to citrate and other multidentate aliphatic complexes such as malonate and oxalate more rapidly than U(VI)

Enzymology of Electron Transport

37

bound to monodentate aliphatic complexes such as acetate, while the opposite trend is found with D. desulfuricans (Ganesh et al. 1997). U(VI) also adsorbs to carboxyl, phosphoryl and amine functional groups on the S. putrefaciens 200 cell surface, and a ligand exchange reaction may take place between the cell surface or U(VI) terminal reductases and the U(VI) complexes prior to reduction (Haas and DiChristina 2002).

Electron donors and competing electron acceptors U(VI) reduction by Shewanella is coupled to oxidation of hydrogen, lactate, formate or pyruvate (Lovley et al. 1991). U(VI) reduction rates are highest with H2 as electron donor (Liu et al. 2002b). Two explanations have been proposed to account for the increased rate of U(VI) reduction coupled to H2 oxidation (Aubert et al. 2000; Liu et al. 2002b). First, electron low through the electron transport chain may be more rapid when coupled to H2 rather than lactate oxidation. Periplasmic H2 hydrogenases may pass electrons through the electron transport chain more rapidly than those generated from cytoplasmic membrane-localized lactate dehydrogenase. Secondly, mass lux of neutrally charged H2 to the enzymatic site of oxidation may be faster than negatively charged lactate. The negative charge of the lactate ion inhibits diffusion across the cell surface to the cytoplasmic membrane, thereby requiring an active transport system. The presence of competing terminal electron acceptors also interferes with microbial U(VI) reduction. Thermodynamic calculations predict that electron acceptors should be utilized in order of highest free energy yield, a possible explanation for the inhibition of U(VI) reduction in the presence of nitrate (Finneran et al. 2002). Although the reduction of U(VI) coupled to the oxidation of organic compounds should yield greater free energy than Fe(III) (Cochran et al. 1986), the half-cell potentials of both U(VI) and Fe(III) can vary markedly with their coordination environment and whether they exist in the form of aqueous complexes or solid phases. For example, the half-cell potentials at pH 7 for many common, environmental Fe(III) forms vary from +0.35 V to −0.30 V (Stumm 1992), while those for U(VI) vary from +0.284 V to −0.042 V (Brooks et al. 2003). These variations in half-cell potential are compounded by the effects of reactant concentrations and other aqueous complexants, the similarity in half-cell potential of many Fe(III) and U(VI) forms and their uncertainty, the interfacial chemistry of solid phase electron acceptors, and the poorly understood redox chemistry of surface complexed Fe(II). All of these considerations complicate a rigorous thermodynamic analysis. The effects of pH are also strong because of the proton stoichiometry of reaction, and the redox stability of U(VI) over Fe(III), or vise-versa, may change if pH is not controlled, if reactant concentrations are varied appreciably, or if mineral biotransformation products exhibit different redox chemistry. In spite of these complexities and chemical interrelationships, there are some consistent thermodynamic observations. Ferrihydrite, with its higher redox potential (~−0.070 V at pH = 7 and Fe(II) = 10−5 mol/L) was observed to inhibit bacterial U(VI) reduction, while goethite did not (~-0.250 V at pH = 7 and Fe(II) = 10−5 mol/L) (Wielenga et al. 2000). Electron transport to Mn(IV) provides a greater free energy yield than electron transport to U(VI), and is therefore predicted to be a preferred electron acceptor (Cochran et al. 1986; Langmuir 1997). Bioavailable Mn(IV)-oxides such as birnessite and bixbyite follow this prediction, however, U(VI) is reduced concurrently with less soluble forms of Mn(IV) (Fredrickson et al. 2002). To determine if this inding is due to electron acceptor competition or abiotic oxidation of U(IV) by Mn(IV), S. putrefaciens CN32 was incubated with U(VI) and pyrolusite (E-MnO2) (Liu et al. 2002b). Extracellular, cell surface-associated, and periplasmic UO2(s) aggregates were detected by Transmission Electron Microscopy (TEM) when cells were incubated only with U(VI). Upon addition of pyrolusite, extracellular UO2(s) was depleted but periplasmic and cell surface-associated UO2(s) remained. These results suggest

38

DiChristina, Fredrickson, Zachara

that U(IV) functions as an electron shuttle and is oxidized by the extracellular pyrolusite. U(VI) is completely reduced provided the OM of intact cells physically separates (sequesters in the periplasmic space) UO2(s) from extracellular pyrolusite. Humic acids have recently gained attention for their potential role as shuttles for electron transfer between anaerobically respiring Shewanella and solid Fe(III)-oxides (see above). Addition of AQDS to S. putrefaciens CN32, however, does not enhance the reduction rate of either soluble or insoluble forms of U(VI) (Fredrickson et al. 2000). AQDS actually inhibits U(VI) reduction activity, possibly by diverting electrons away from the U(VI) reduction pathway.

Subcellular location of enzymatic U(VI) reduction activity The subcellular location of enzymatic U(VI) reduction in Shewanella has also been recently examined: Insoluble U(IV) particles are detected extracellularly, on the cell surface and within the periplasmic space of S. putrefaciens CN32 after reduction of soluble U(VI) (Liu et al. 2002a). U(IV) is not detected in the cytoplasm (Fig. 6). U(VI) reductases may therefore be localized within the OM, diffuse (or be transported) across the OM to contact U(VI) reductases located in the periplasm or IM, or both. U(VI) reduction products of Desulfosporosinus have been detected as nanometer-sized UO2(s)-particles (Suzuki et al. 2002). Nanoparticles produced in the periplasm either diffuse or are exported to the cell exterior where they organize extracellularly to form larger aggregates. Aggregation of U(IV) particles prior to export from the cell may result in the periplasmic deposits detected on TEM images of U(VI)-respiring cells (Liu et al. 2002a). U(IV) particles detected in the culture supernatant also leads to the intriguing possibility that anaerobically-respiring Shewanella are able to actively secrete U(IV) particles as a means of avoiding build-up of toxic insoluble U(IV) end-products during U(VI) reduction. S. putrefaciens CN32 is also capable of reducing solid forms of U(VI) such as metaschoepite (Fredrickson et al. 2000), although solid forms of U(VI) have been found to be resistant to microbial reduction in situ (Ortiz-Bernad et al. 2004). The mechanism by which Shewanella species reduce metaschoepite is unknown, but U(VI) terminal reductase localization to the OM to contact solid U(VI) is possible.

Figure 6. Transmission electron microscopy image of an unstained thin section from Shewanella putrefaciens strain CN32 cells incubated with H2 and U(VI) in pH 7 bicarbonate buffer, illustrating the accumulation of nano-size U(IV)O2 particles extracellularly and in association with the periplasmic space and cell surface.

Enzymology of Electron Transport

39

ENZYMATIC MECHANISM OF TECHNETIUM REDUCTION Enzymatic studies on Tc(VII) reduction have largely been focused in E. coli, however the ability to reduce Tc(VII) has been recently found in S. putrefaciens CN32, S. oneidensis MR-1 and S. putrefaciens 200 (Lyalikova and Khizhnyak 1996; Lloyd et al. 1997; Wildung et al. 2000; Payne and DiChristina 2005). Tc(VII) is also reduced under acidic conditions by Thiobacillus thiooxidans (Lyalikova and Khizhnyak 1996), under alkaline conditions by Halomonas strain Mono (Khijniak et al. 2003) and at high temperature by Pyrobaculum islandicum (Kashei and Lovley 2000). Reduction of soluble Tc(VII) results in formation of Tc(IV) which precipitates as insoluble TcO2·nH2O (hereafter termed TcO2) and may be immobilized in situ. In the absence of aqueous complexing agents, Tc(IV) may also be immobilized via formation of strong surface complexes with hydroxylated surface sites on Al and Fe oxides and clays (Rard 1983; Haines et al. 1987; Meyer et al. 1991; Eriksen et al. 1992; Wildung et al. 2000).

Involvement of hydrogenases in Tc(VII) reduction E. coli possesses four hydrogenases, designated as hydrogenases 1-4. Hydrogenases-1 and 2 share little homology to hydrogenases-3 and 4. Hydrogenases 3 and 4 share high homology to each other and are both expressed as part of the formate-hydrogen lyase complex in E. coli (Bagramyan and Trchounian 2003). The Tc(VII) reductase in E. coli has been identiied as the Ni-Fe hydrogenase-3 component of the formate-hydrogen lyase complex (Lloyd et al. 1997). Hydrogenase expression is determined by pH: hydrogenase-4 (encoded by the hyf operon) is expressed under alkaline conditions while hydrogenase-3 (encoded by the hyc operon) is expressed under acidic conditions (Bagramyan and Trchounian 2003). The formate-hydrogen lyase complex in E. coli is composed of formate dehydrogenase plus multiple components of the respective hydrogenase encoding operons (i.e., hyf and hyp operons). The bi-directional nature of hydrogenase-3 (HycE) enables both the production of H2 during formate oxidation and the direct oxidation of H2 under other conditions. S. oneidensis MR-1 does not possess a formate-hydrogen lyase complex and possesses only two hydrogenases, neither of which share signiicant homology to hydrogenases-3 or 4 of E. coli. The irst S. oneidensis MR-1 hydrogenase (Locus SO2098; HyaB) displays high homology to the IM-bound Ni-Fe hydrogenase HydB of Wolinella succinogenes, while the second S. oneidensis MR-1 hydrogenase (Locus SO3920; HydA) displays high homology to the putative D subunit of the NADP-reducing hydrogenase of Thermotoga maritima (Payne and DiChristina 2005). In terms of hydrogenase function, S. oneidensis MR-1 hydrogenases appear most similar to those of Alcaligenes eutrophus in which the cytoplasmic, soluble hydrogenase (HydA) regenerates NADH, while the membrane bound Ni-Fe hydrogenase (HyaB) generates reducing power (Lengeler et al. 1999). In organisms containing only membrane bound hydrogenases, reducing power is generated by reverse electron transport, generally carried out by membrane bound bi-directional hydrogenases (e.g., hydrogenases3 and 4 in E. coli). S. oneidensis MR-1 therefore appears to share close similarity to the hydrogen uptake and utilization systems of A. eutrophus and little sequence or physiological similarity to the H2 uptake and utilization systems of E. coli. Further work needs to be carried out to determine if the S. oneidensis MR-1 hydrogenases display Tc(VII) reductase activity.

Subcellular location of enzymatic Tc(VII) reduction activity The enzymatic reduction of Tc(VII) is electron donor-speciic. H2 serves as electron donor in all known Tc(VII)-reducing organisms (Lloyd et al. 2000; Wildung et al. 2000; De Luca et al. 2001), while the ability to couple the oxidation of other carbon sources to the reduction of Tc(VII) occurs in only a small subset of organisms. G. sulfurreducens and D.

DiChristina, Fredrickson, Zachara

40

fructosovorans have an exclusive requirement for H2 as electron donor for Tc(VII) reduction while E. coli is limited to formate and H2 as electron donor. S. oneidensis MR-1 and S. putrefaciens CN32 couple the oxidation of formate, lactate, and H2 to Tc(VII) reduction (Wildung et al. 2000; Payne and DiChristina 2005), but Tc(VII) reduction rates are markedly higher with H2 as electron donor. The reduced Tc(IV) product is generally nanometer-sized TcO2(s) in buffers or media without high carbonate. The identity of the electron donor does not seem to inluence the mineralogic nature of the reduction product. The black-colored precipitate is observed in the periplasm, and as 20-50 nm dome-like structures consisting of aggregates of many individual crystallites on the cell surface (Fig. 7). The TcO2(s) is nanocrystalline and exhibits insuficient long-range order to yield a discernable diffraction pattern. The precipitate maintains a Tc solubility (≈10−8 mol/L) that approximates measured values for TcO2·xH2O (as reported by Rard 1999 and associated citations). The physiologic relationship between subcellular and surface associated TcO2(s) is unclear. Limited evidence implies that bioreduced Tc [e.g., Tc(IV)] may exist in the form of carbonate aqueous complexes, or perhaps carbonate precipitates, in high-bicarbonate media (Wildung et al. 2000). Further research on the biogeochemistry of Tc(VII)/Tc(IV) in bicarbonate-containing media and other ligand solutions of geochemical relevance is needed.

A

B

C Figure 7. Transmission electron microscopy image of an unstained thin section from Shewanella oneidensis MR-1 cells incubated with H2 and Tc(VII) in pH 7 PIPES buffer. Tc accumulates predominantly in association with the cell envelope (A, B) as discreet aggregates (C) consisting of Tc(IV)O2 particles approximately 2.2 nm in diameter (arrow).

Enzymology of Electron Transport

41

MICROBIAL REDUCTION-INDUCED CHANGES IN METAL BIOGEOCHEMISTRY Direct enzymatic effects of dissimilatory metal-reducing bacteria (DMRB) on metal solubility The majority of electron acceptors commonly used by prokaryotes (oxygen, nitrate, sulfate, carbon dioxide) exhibit relatively high levels of solubility before and after reduction. In contrast, many of the metals used as microbial electron acceptors exhibit substantially different solubility properties in the oxidized (e.g., pH = 10) versus the reduced (pH = 4 or 2) states (Table 1). Because Fe(III) and Mn(III,IV) exist predominantly as oxyhydroxide minerals in oxic environments (e.g., ferrihydrite and goethite, or birnessite and manganite), DMRB must overcome the fundamental problem of engagement of the cell electron transport system (ETS) with the mineral surface across a solid-liquid interface. DMRB have developed several novel mechanisms for overcoming this problem, as described in preceding sections, including the “shuttling” of electrons by humic acids (Lovley et al. 1996; Lovley and Woodward 1996) or cell metabolites (Newman and Kolter 2000) from terminal points of the ETS to the mineral surfaces, possibly the direct transfer of electrons to metal in the centers of mineral surfaces by multiheme cytochromes associated with the OM (Richardson 2000; Leang et al. 2003), or the solubilization of solid phase-associated Fe(III) as subsequent engagement of Fe(III) reductase(s) as a soluble Fe(III)-organic complex. Regardless of the mechanism, the microbial reduction of Fe and Mn has a profound impact on the geochemical behavior of these metals as Table 1. Solubilities (aqueous concentrations) of select phases of Fe, Mn, U, and Tc at pH 7 in water as a function of pe (pCO2 = 10−3.46 atm, I = 0.01). Solid phase

Oxidizing pH = 10 mol/L

Reducing (pH = 4) mol/L

(pH = 2) mol/L

Fe(OH)3 (ferrihydrite)

2.06u10−8 (Fe3+)

2.06u10−8 (Fe3+) 1.29u10−7 (Fe2+)

2.06u10−8 (Fe3+) 1.29u10−5 (Fe2+)

D-FeOOH (goethite)

8.40u10−13 (Fe3+)

8.36u10−13 (Fe3+) 5.24u10−12 (Fe2+)

8.36u10−13 (Fe3+) 5.24u10−10 (Fe2+)

MnO1.8 (birnessite)

6.03u10−5 (Mn2+)

Soluble as Mn2+ a

Soluble a

J-MnOOH (manganite)

2.87u10−6 (Mn2+)

Soluble as Mn2+ a (2.83 mol/L maximum)

Soluble a

2.67u10−7 [U(VI)O22+] b

2.67u10−7 [U(VI)O22+] b

1.46u10−7 [U(VI)O22+] 2.0u10−17 [U(IV)(OH)4(aq)] c

Soluble as Tc(VII)O4−

Soluble as Tc(VII)O4−

10−8 [Tc(IV)O(OH)2(aq)]

UO2 (uraninite)

TcO2·nH2O a. b.

c.

Will precipitate as rhodochrosite [MnCO3(c)] 1u10−4 mol/L precipitated as schoepite [E-UO3·2H2O(c)], the primary aqueous complex under the given conditions is UO2(CO3)22−, solubility increases with CO2(g) partial pressure and total aqueous carbonate concentration 9.98u10−5 mol/L precipitated as uraninite [UO2(c)], pH is 1.8, U(VI) concentrations decrease with decreasing pH.

42

DiChristina, Fredrickson, Zachara

well as broader impacts on the overall geochemical and mineralogic properties of solids and sediments where these processes occur. In contrast to Fe and Mn, U, Tc, and Cr are relatively soluble in oxic environments and typically exist as anionic uranyl carbonate UO2(CO3)34−, UO2(CO3)22− complexes, pertechnetate (TcO4−), or chromate (CrO42−), respectively. The solubility of U(VI) at circumneutral pH is strongly dependent on dissolved carbonate concentration and other associated ligands such as silica or phosphate. Upon reduction to the +4 oxidation state and in the absence of strong complexants, U and Tc can precipitate as the hydrous oxides, UO2 (uraninite) and TcO2, phases that have been identiied in anaerobic suspensions of DMRB cells incubated with U(VI) (Gorby and Lovley 1992) or Tc(VII) (Wildung et al. 2000) and appropriate electron donors. The direct enzymatic reduction of U(VI) results in the formation of relatively uniformly-sized nanoparticles (Fredrickson et al. 2002; Suzuki et al. 2002), a factor that can impact their subsequent reactivity and transport. For example, if U(IV) nanoparticles are less than approximately 2-to-5 nm in diameter, they may behave as large molecular clusters and be mobile in solution, while U(IV) particles with larger diameters may not be transported as readily (as described in chapter on nanoparticles by Gilbert and Banield 2005). It is these marked changes in solubility that have prompted consideration of manipulating the activities of DMRB for the bioremediation of soils and sediments contaminated with metals and radionuclides (Lovley 1995).

Indirect effects of DMRB on metal solubility DMRB can also indirectly inluence the biogeochemical behavior of redox active and non-redox active metals. Because the mass content of Fe and Mn is typically higher than trace metals and contaminants in most soils and sediments they tend to have a dominant effect on redox reactions and function as a major sink for electrons from DMRB respiration. Due to their relatively higher mid-point potential, Mn oxides can provide a “buffer” against the net reduction of other metal ions including Fe(III) (Lovley and Phillips 1988; Myers and Nealson 1988) and U(VI) (Liu et al. 2002b). Because Mn(III,IV) oxides are relatively strong oxidants they can also oxidize reduced forms of metals such as Cr(OH)3 (Fendorf and Zasoski 1992) and UO2 (Fredrickson et al. 2002), hence impeding their net microbial reduction unless there are mechanisms that prevent their physical interaction such as the accumulation and isolation of UO2 nanoparticles in the cell periplasm (Fredrickson et al. 2002). Although Fe(III) oxides are not as effective oxidants as Mn oxides, they can potentially impede the reduction of other metals such as U(VI) via competition mechanisms (Wielinga et al. 2000). Hence, the microbial reduction of Mn and Fe oxides can result in redox conditions that are more favorable for the reduction for trace metal contaminants. In addition, Fe(II) and Mn(II) can potentially provide buffer against re-oxidation. Biogenic Fe(II) can also function as a facile reductant of trace metal and radionuclides including U(VI) (Liger et al. 1999), Tc(VII) (Lloyd et al. 2000; Wildung et al. 2004), and Cr(VI) (Wielinga et al. 2001). The rate of Tc(VII)O4− reduction by sediment-associated biogenic Fe(II) was shown to be related directly to the extent of sediment Fe(III) reduction but there was extensive variation among different sediments indicating that the effectiveness of Fe(II) as a reductant was highly dependent upon molecular speciation as opposed to Fe(II) concentration alone (Fredrickson et al. 2004; Hansel et al. 2004). In general, aqueous Fe(II) is poorly reactive with Tc(VII) (Cui and Eriksen 1996) and U(VI) (Fredrickson et al. 2000), probably due to kinetic limitations. In contrast, the rates of chromate reduction by Fe(II)aq are relatively rapid (Wielinga et al. 2001). One area that warrants further investigation is whether biosorbed Fe(II), which can form complexes and precipitates on cell surfaces (Liu et al. 2001a), can also function as a reductant in a manner similar to Fe(II) sorbed on mineral surfaces. The fact that Fe oxides can impede i.e., via competition (Wielinga et al. 2000)

Enzymology of Electron Transport

43

or promote, i.e., via surface complexation of Fe(II) (Fredrickson et al. 2004), relects the similarity of the mid-point potentials of these metals and the need to pay careful attention to factors including speciation, concentration, and solubility that can greatly impact the direction and extent of such redox reactions. Trace metals can associate with Fe or Mn oxides as adsorbed or co-precipitated species and are therefore subject to biogeochemical reactions resulting from utilization of the oxide as a terminal electron acceptor by DMRB. Ni2+ and Co2+ co-precipitated with goethite were released when oxide suspensions were subject to reduction by S. putrefaciens CN32, resulting in a net increase in aqueous concentrations of the metal ions (Zachara et al. 2001). Similarly, Ni2+ was also released from a Ni-substituted hydrous ferric oxide upon reduction by strain CN32 although under select conditions a Ni-substituted magnetite (FeIII2FeII1−xNixO4) formed (Fredrickson et al. 2001). Ni2+ was found to inhibit the overall reduction reaction by an undeined chemical mechanism that could be circumvented by addition of AQDS as an electron shuttle. Aluminum release during the bioreduction of an Al-substituted goethite associated with an Atlantic coastal plain sediment was congruent with the production of Fe(II) but the released Al was associated with a sorbed phase (Kukkadapu et al. 2001). DMRB can also promote the mobilization of arsenic as arsenate via the reductive dissolution of the ferric arsenate mineral scorodite (FeAsO4·2H2O) and from iron oxide sorption sites within sediments (Cummings et al. 1999). It is interesting to note that some DMRB are also capable of dissimilatory reduction of arsenate (AsV) to arsenite (AsIII) (Saltikov et al. 2003) but such a reduction reaction was not observed. Although there is considerable potential for mobilization of trace elements associated with metal oxides during bioreduction, the extent to which trace metals remain associated with the solid phase or are released to solution will be a function of the aqueous and solid-phase geochemical composition that ultimately controls the adsorption and precipitation reactions.

REDUCTIVE TRANSFORMATION OF Fe- AND Mn-CONTAINING MINERALS The rate and extent of microbial reduction of Fe(III) and Mn (III,IV) oxides in soils and sediments is a function of complex and highly coupled biological, chemical, and physical factors. Mineralogy plays a critical role with factors such as surface area (Roden and Zachara 1996), extent of structural disorder (Zachara et al. 1998), surface speciation (Roden and Urrutia 2002), and thermodynamics (Liu et al. 2001b) all inluencing, to some extent, the reduction process. The physiological state of the organisms, including effects resulting from growth medium composition (Glasauer et al. 2003) and electron donor-acceptor ratios (Zachara et al. 2002), are other key variables that can affect bioreduction of Fe and Mn minerals but are currently poorly understood. The role of cell physiology on metal oxide reduction is currently under-appreciated and its importance warrants further research using physiologically and compositionally deined cultures that better represent and span the range of environmental conditions.

Laboratory studies A general observation made from laboratory studies is that poorly crystalline Fe(III) oxides such as ferrihydrite (Lovley and Phillips 1986) exhibit a greater degree of bioavailability than more crystalline phases such as lepidocrocite (J-FeOOH), goethite (D-FeOOH), or hematite (Fe2O3), although as a crystalline phase, lepidocrocite is far more bioavailable than the others. The same general trend appears to hold true for Mn oxides in that more highly crystalline phases such as pyrolusite (E-MnO2) are reduced more slowly than amorphous MnO2 or birnessite (Burdige et al. 1992). This effect has been attributed to differences in solubility of these phases and is supported by experiments demonstrating that the maximum rate of Fe(III) reduction was found to correlate positively with the solubility of the oxide (Bonneville et al. 2004). Caution must be exercised when using rates of abiotic reduction of Fe oxides as an

44

DiChristina, Fredrickson, Zachara

indicator of their susceptibility to enzymatic reduction as the surface area-normalized rates of bacterial reduction of ferrihydrite, lepidocrocite, goethite, and hematite were found to be quite similar, in contrast to reduction of the same phases by ascorbic acid (Roden 2003). Naturallyoccurring (geologic) Fe oxides have been found to be equally or more reducible than their synthetic counterparts with crystalline disorder and microheterogeneities potentially being dominant factors controlling microbial reduction (Zachara et al. 1998). The presence of solid phase sorbents and organic complexants can also facilitate microbial reduction of Fe oxides, presumably via the removal of reduction product (Fe2+) from oxide and bacterial surfaces (Urrutia et al. 1999). A similar enhancement in the extent of Fe oxide reduction can be achieved by continual replacement of the aqueous phase in semi-continuous cultures (Roden and Urrutia 1999) or in continuous low columns where soluble Fe(II) is constantly removed (Roden et al. 2000). The products of oxide bioreduction can hence impede further reduction by passivating oxide and cell surfaces (Liu et al. 2001a,b) or can promote reduction by removing products from solution via secondary precipitation reactions. Transformation of Fe- and Mn-bearing minerals to secondary phases is similarly a function of a complex set of biogeochemical variables. A number of laboratory studies have probed the bioreductive transformation of poorly crystalline ferrihydrites to a wide range of phases including more highly ordered Fe(III) oxides such as 6-line ferrihydrite, goethite, lepidocrocite (Zachara et al. 2002, Fredrickson et al. 2003; Hansel et al. 2003; Kukkadapu et al. 2003, 2005), mixed valence oxides such as magnetite and green rust [FeII(6−x)FeIII(x)(OH)12]x+[(A2−)x/2·yH2O]x−] (Lovley et al. 1987; Fredrickson et al. 1998), or Fe(II) phases including siderite (FeCO3), vivianite (Fe3(PO4)2·8H2O), and ferrous hydroxy carbonate (Fe2(OH)2CO3) (Fredrickson et al. 1998; Hansel et al. 2003; Kukkadapu et al. 2003). The extent to which each of these phases may form is inluenced by a range of factors including pH and aqueous solution composition (Fredrickson et al. 1998), the relative concentrations of the oxide (electron acceptor) and electron donor (Zachara et al. 2002; Fredrickson et al. 2003), the presence of co-precipitated ions (Fredrickson et al. 2001; Kukkadapu et al. 2004), and hydrodynamic-induced distributions of reduction products (Hansel et al. 2003). The formation of secondary mineral phases is less common or extensive when more highly ordered phases such as hematite or goethite are reduced by DMRB but nonetheless Fe(II) biominerals such as siderite and vivianite have been observed to form under conditions consistent with their solubility (Zachara et al. 1998). The lower solubility of more highly ordered phases such as goethite and hematite support lower concentrations of Fe(III)aq and DMRB-generated Fe(II)aq, relative to ferrihydrite, and therefore hinder precipitation of secondary minerals such as magnetite beyond reorganization on the mineral surface into nanometer-size spinel-like domains (Hansel et al. 2004). In addition to Fe oxides and oxyhydroxides, DMRB can also reduce structural Fe in clay minerals facilitating their dissolution (Kostka et al. 1996, 1999) or changes in clay morphology, structure, and composition (Dong et al. 2003a,b). The extent of microbial reduction of structural Fe(III) in smectite can be substantial, ranging to >90% (Kostka et al. 1999). The microbial reduction of structural Fe in clay can signiicantly alter clay chemical and physical properties such as was reported for studies with smectite where reduction resulted in reduced swelling pressure, total surface area, and surface charge density (Kostka et al. 1999). Structural Fe(II) in bioreduced ferruginous clays can also promote the reductive dehydrochlorination of organic contaminants such as pentachloroethane and trichloroethane (Cervini-Silva et al. 2003).

Field studies Controlled single-phase, single organism laboratory experiments can provide important mechanistic insights but it is often dificult to predict ield behavior from such results due to environmental complexities and heterogeneities. One of the more well studied ield sites is

Enzymology of Electron Transport

45

a crude oil-impacted shallow groundwater aquifer located near Bemidji, Minnesota, USA. Research at this site has documented aspects of microbial community structure (Rooney-Varga et al. 1999) and associated geochemical changes (Baedecker et al. 1993) over the length of the plume as the predominant respiratory process shifted from Fe(III) and Mn(III,IV) reduction to methanogenesis (Anderson and Lovley 1999). A detailed analysis of sediments from within the plume and from the pristine aquifer revealed signiicant differences in the mass content and identify of Fe(III) oxides consistent with microbial-driven reduction processes (Zachara et al. 2004). Comparisons between the texturally-similar source where bioavailable Fe(III) had been exhausted and Fe(III)-reducing zone sediments where bioavailable Fe(III) remained indicated that dispersed crystalline Fe(III) oxides and a portion of the poorly crystalline Fe(III) oxide fraction had been depleted from the source zone sediment. The presence of residual ferrihydrite in the anoxic plume sediment indicated that some fraction of the Fe(III) oxides were biologically inaccessible, possibly due to their residence in microfractures in the interior of lithic fragments. Interestingly, little evidence was found for biogenic ferrous mineral phases with the exception of thin siderite or ferroan calcite surface precipitates. It is clear that additional ield-based research including characterization of samples in concert with modeling and laboratory-based experiments is needed to improve our ability to predict biogeochemical behavior of redox active metals in natural and engineered systems, particularly with regard to mineral biotransformation products and biominerals.

ROLE OF MICROBIAL METAL REDUCTION IN REDOX CYCLING As is the case for most microorganisms, DMRB rarely function alone but are members of complex communities whose collective, intertwined activities are responsible for catalyzing the cycling of elements in the biosphere. Many DMRB, particularly members of the genus Shewanella, are well-adapted to geochemically stratiied environments where there is a gradient in electron acceptors available for oxidizing organic matter and H2. These adaptations include a relatively robust and diverse electron transport system that can engage a wide range of electron acceptors and an extensive network of regulatory genes, both two-component and transcriptional regulators (Heidelberg et al. 2002), that allow the organisms to sense and respond to their environment. These organisms play a critical role in such environments by oxidizing fermentation products, such as low molecular weight organic acids, coupled to respiration of Fe and Mn.

Redox cycling in chemically stratified environments The Black Sea is a prime example of a chemically (redox) stratiied environment that exists over distances of tens of meters in the water column and where Fe and Mn undergo biogeochemical redox cycling (Nealson and Myers 1992). Fe and Mn are chemically unique ions in redox gradient environments because of their relatively low solubility in the oxidized state. As these metal ions are oxidized, either microbially or abiotically, they can precipitate and be subjected to gravitational settling into anoxic zones (Nealson and Saffarini 1994). As they enter the anoxic zones these precipitates can be reduced by DMRB coupled to organic matter oxidation at which point soluble species of Fe(II) and Mn(II) can diffuse upward into oxic zones. A number of excellent reviews have been published on this subject and the reader is directed to those for more information and examples (Burdige 1993; Nealson and Saffarini 1994; Nealson and Little 1997; Nealson et al. 2002). Previous investigations of redox stratiied environments have justiiably focused on detailed geochemical characterization and baseline investigations into associated microbial properties using a combination of cultivation and cultivation-independent methods. The availability of whole genome sequences and the ability to sequence entire microbial communities (metagenomics) provide powerful tools to probe

46

DiChristina, Fredrickson, Zachara

microbial processes in these environments in the future (for additional discussion see chapters by Whitaker and Banield 2005 and Nelson and Methé 2005).

Microscale redox cycling In addition to participating in redox cycling in chemocline environments that can span many meters, more recent research indicates that DMRB also participate in redox cycling over much shorter length scales. Because Fe(II) is rapidly oxidized by O2, neutrophilic microbial Fe oxidation is constrained to microaerobic environments where the abiotic oxidation of Fe(II) is limited by O2 availability and microbiologic rates of Fe(II) oxidation are competitive with abiotic rates. Hence, there is a potential for tight coupling between metal reduction and oxidation steps over short distance scales in environments with sharp microaerobic and anaerobic boundaries. In fact, it has been proposed that Fe(II)-oxidizing organisms localize themselves into a narrow band of cells and associated Fe(III) oxides to facilitate interfacing with DMRB (Roden et al. 2004). Such a coupling was experimentally investigated in microcosms consisting of ferrihydrite coated sand and a co-culture consisting of a lithotrophic Fe(II)-oxidizing bacterium (strain TW2) and the DMRB Shewanella alga strain BrY (Sobolev and Roden 2002). The co-culture exhibited minimal Fe oxide accumulation at the sand-water interface despite measurable dissolved O2 to a depth of 2 mm below the interface whereas a distinct layer of Fe oxide formed at this same interface in microcosms containing BrY alone. Direct microscopic observations revealed close juxtapositioning of both organisms in the upper few mm of sand. Subsequent investigations using the identical experimental system noted relatively low concentrations of Fe(II) in the co-culture relative to the microcosm containing BrY alone and suggested that Fe(III)-binding ligands impeded the formation of Fe(III) oxides and were responsible for a soluble/colloidal Fe(III) phase that facilitated the redox cycling of Fe (Roden et al. 2004). These results established the potential for a tight coupling between microbial metal reduction and oxidation processes to promote rapid microscale cycling of Fe. More research, however, is needed to better deine the nature and role of Fe(III)-complexing ligands in microscale Fe cycling and whether interactions between metal-reducing and metaloxidizing extend beyond simply Fe(II)-Fe(III) cycling.

SUMMARY Most of the electron acceptors respired by prokaryotes (O2, NO3−, SO42−, and CO2) are soluble both before and after reduction, while many of the metals respired by DMRB exhibit substantially different solubility properties in the oxidized versus the reduced states. Because Fe(III) and Mn(III,IV) exist predominantly as oxyhydroxide minerals in oxic environments, DMRB must overcome the fundamental problem of engagement of the electron transport system with poorly soluble minerals. Other metals, such as U(VI) and Tc(VII), are relatively soluble in oxic environments, typically as anionic uranyl carbonate complexes and as pertechnetate, respectively. Aqueous Tc(VII) and U(VI) and other soluble electron acceptors are therefore free to enter the cell periplasm through porins or channels in the OM. Upon reduction to the +4 oxidation state, however, U and Tc precipitate as uraninite and hydrous Tc(IV) oxides, phases that have been identiied in anaerobic suspensions of DMRB cells incubated with U(VI) or Tc(VII) and appropriate electron donors. The dilemma of reducing soluble electron acceptors to insoluble end-products is no less serious than the one dealing with reduction of solid electron acceptors. The irst section of this chapter has highlighted the latest indings on the novel respiratory strategies employed by DMRB to overcome this dilemma, including direct enzymatic reduction, electron shuttling pathways and metal solubilization by exogenous or bacterially-produced organic ligands followed by reduction of soluble organicmetal compounds. The second section has emphasized the geochemical consequences of DMRB activity, including the direct and indirect effects on metal solubility, the reductive

Enzymology of Electron Transport

47

transformation of Fe- and Mn-containing minerals, and the biogeochemical cycling of metals at redox interfaces in chemically stratiied environments.

ACKNOWLEDGMENTS The authors wish to thank David Bates, Justin Burns, Jason Dale, Amanda Payne and Tara Hoyem for help in manuscript preparation. The authors also with to thank Alice Donalkova, Dwayne Elias, David Kennedy, Matthew Marshall, and Andrew Plymale for providing the transmission electron microscopy images. Financial support for this work was provided by the National Science Foundation and the U.S. Department of Energy, Ofice of Biological and Environmental Research.

REFERENCES Anderson RT, Lovley DR (1999) Naphthalene and benzene degradation under Fe(II)-reducing conditions in petroleum-contaminated aquifers. Biorem J 3:121-135 Arnold RG, DiChristina TJ, Hoffmann MR (1988) Reductive dissolution of iron-oxides by Pseudomonas sp. 200. Biotech Bioeng 32:1081-1096 Aubert C, Brugna M, Dolla A, Bruschi M, Giudici-Orticoni MT (2000) A sequential electron transfer from hydrogenases to cytochromes in sulfate-reducing bacteria. Biochim Biophys Acta 1476:85-92 Baedecker MJ, Cozzarelli IM, Eganhouse RP, Siegel DI, Bennett PC (1993) Crude-oil in a shallow sand and gravel aquifer. 3. Biogeochemical reactions and mass-balance modeling in anoxic groundwater. Appl Geochem 8:569-586 Bagramyan K, Trchounian A (2003) Structural and functional features of formate hydrogen lyase, an enzyme of mixed-acid fermentation from Escherichia coli. Biochemistry-Moscow 68:1159-1170 Beliaev AS, Saffarini DA (1998) Shewanella putrefaciens mtrB encodes an outer membrane protein required for Fe(III) and Mn(IV) reduction. J Bacteriol 180:6292-7 Bonneville S, Van Cappellen P, Behernds T (2004) Microbial reduction of iron(III) oxyhydroxides: effects of mineral solubility and availability. Chem Geol 212:255-268 Brooks SC, Fredrickson JK, Carroll SL, Kennedy DW, Zachara JM, Plymale AE, Kelly SD, Kemner KM, Fendorf S (2003) Inhibition of bacterial U(VI) reduction by calcium. Environ Sci Technol 37:1850-1858 Burdige DJ (1993) The biogeochemistry of manganese and iron reduction in marine-sediments. Earth-Science Reviews 35:249-284 Burdige DJ, Dhakar SP, Nealson KH (1992) Effects of manganese oxide mineralogy on microbial and chemical manganese reduction. Geomicrobiol J 10:27-48 Butler A (2003) Iron acquisition: straight up and on the rocks? Nature Struct Biol 10:240-241 Caccavo F, Blakemore RP, Lovley DR (1992) A hydrogen-oxidizing, Fe(III)-reducing microorganism from the Great Bay Estuary, New-Hampshire. Appl Environ Microbiol 58:3211-3216 Cervini-Silva J, Kostka JE, Larson RA, Stucki JW, Wu J (2003) Dehydrochlorination of 1,1,1-trichloroethane and pentachloroethane by microbially reduced ferruginous smectite. Environ Toxicol Chem 22:10461050 Chin KJ, Esteve-Nunez A, Leang C, Lovley DR (2004) Direct correlation between rates of anaerobic respiration and levels of mRNA for key respiratory genes in Geobacter sulfurreducens. Appl Environ Microbiol 70: 5183-5189 Coates JD, Cole KA, Chakraborty R, O’Connor SM, Achenbach LA (2002) Diversity and ubiquity of bacteria capable of utilizing humic substances as electron donors for anaerobic respiration. Appl Environ Microbiol 68:2445-52 Coates JD, Ellis DJ, Blunt-Harris EL, Gaw CV, Roden EE, Lovley DR (1998) Recovery of humic-reducing bacteria from a diversity of environments. Appl Environ Microbiol 64:1504-9 Cochran JK, Carey AE, Sholkovitz ER, Suprenant LD (1986) The geochemistry of uranium and thorium in coastal marine sediments and sediment pore waters. Geochim Cosmochim Acta 50:663-680 Cui DQ, Eriksen TE (1996) Reduction of pertechnetate by ferrous iron in solution: Inluence of sorbed and precipitated Fe(II). Environ Sci Technol 30:2259-2262 Cummings DE, Caccavo F, Fendorf S, Rosenzweig RF (1999) Arsenic mobilization by the dissimilatory Fe(III)reducing bacterium Shewanella alga BrY. Environ Sci Technol 33:723-729 De Luca G, De Philip P, Dermoun Z, Rousset M, Vermeglio A (2001) Reduction of technetium(VII) by Desulfovibrio fructosovorans is mediated by the nickel-iron hydrogenase. Appl Environ Microbiol 67: 4583-4587

48

DiChristina, Fredrickson, Zachara

Desvaux M, Parham NJ, Scott-Tucker A, Henderson IR (2004) The general secretory pathway: a general misnomer? Trends Microbiol 12:306-309 DiChristina TJ, Adiga M, Bates D, Burns J, Haller CA (2005) Submitted for review. DiChristina TJ, DeLong EF (1994) Isolation of anaerobic respiratory mutants of Shewanella putrefaciens and genetic analysis of mutants deicient in anaerobic growth on Fe3+. J Bacteriol 176:1468-74 DiChristina TJ, Moore CM, Haller CA (2002) Dissimilatory Fe(III) and Mn(IV) reduction by Shewanella putrefaciens requires ferE, a homolog of the pulE (gspE) type II protein secretion gene. J Bacteriol 184: 142-51 Dong HL, Kostka JE, Kim J (2003a) Microscopic evidence for microbial dissolution of smectite. Clays Clay Mineral 51:502-512 Dong HL, Kukkadapu RK, Fredrickson JK, Zachara JM, Kennedy DW, Kostandarithes HM (2003b) Microbial reduction of structural Fe(III) in illite and goethite. Environ Sci Technol 37:1268-1276 Eriksen TE, Ndalamba P, Bruno J, Caceci M (1992) The solubility of TcO2•nH2O in neutral to alkaline-solutions under constant pCO2. Radiochim Acta 58-9:67-70 Fendorf SE, Zasoski RJ (1992) Chromium(III) oxidation by delta-MnO2. Characterization. Environ Sci Technol 26:79-85 Filloux A (2004) The underlying mechanisms of type II protein secretion. Biochim Biophys Acta 1694:163179 Finch R, Murakami T (1999) Systematics and paragenesis of uranium minerals. Rev Mineral 38:91-180 Finneran KT, Anderson RT, Nevin KP, Lovley DR (2002) Potential for Bioremediation of uranium-contaminated aquifers with microbial U(VI) reduction. Soil Sediment Contam 11:339-357 Francis AJ, Dodge CJ, Lu FL, Halada GP, Clayton CR (1994) XPS and XANES studies of uranium reduction by Clostridium Sp. Environ Sci Technol 28:636-639 Fredrickson JK, Kota S, Kukkadapu RK, Liu CX, Zachara JM (2003) Inluence of electron donor/acceptor concentrations on hydrous ferric oxide (HFO) bioreduction. Biodegradation 14:91-103 Fredrickson JK, Zachara JM, Kennedy DW, Dong HL, Onstott TC, Hinman NW, Li SM (1998) Biogenic iron mineralization accompanying the dissimilatory reduction of hydrous ferric oxide by a groundwater bacterium. Geochim Cosmochim Acta 62:3239-3257 Fredrickson JK, Zachara JM, Kennedy DW, Duff MC, Gorby YA, Li SMW, Krupka KM (2000) Reduction of U(VI) in goethite (alpha-FeOOH) suspensions by a dissimilatory metal-reducing bacterium. Geochim Cosmochim Acta 64:3085-3098 Fredrickson JK, Zachara JM, Kennedy DW, Kukkadapu RK, McKinley JP, Heald SM, Liu C, Plymale AE (2004) Reduction of TcO4− by sediment-associated biogenic Fe(II). Geochim Cosmochim Acta 68:3171-3187 Fredrickson JK, Zachara JM, Kennedy DW, Liu CX, Duff MC, Hunter DB, Dohnalkova A (2002) Inluence of Mn oxides on the reduction of uranium(VI) by the metal-reducing bacterium Shewanella putrefaciens. Geochim Cosmochim Acta 66:3247-3262 Fredrickson JK, Zachara JM, Kukkadapu RK, Gorby YA, Smith SC, Brown CF (2001) Biotransformation of Nisubstituted hydrous ferric oxide by an Fe(III)-reducing bacterium. Environ Sci Technol 35:703-712 Ganesh R, Robinson KG, Reed GD, Sayler GS (1997) Reduction of hexavalent uranium from organic complexes by sulfate- and iron-reducing bacteria. Appl Environ Microbiol 63:4385-4391 Glasauer S, Weidler PG, Langley S, Beveridge TJ (2003) Controls on Fe reduction and mineral formation by a subsurface bacterium. Geochim Cosmochim Acta 67:1277-1288 Gilbert B, Banield JF (2005) Molecular-scale processes involving nanoparticulate minerals in biogeochemical systems. Rev Mineral Geochem 59:109-155 Gorby YA, Lovley DR (1992) Enzymatic uranium precipitation. Environ Sci Technol 26:205-207 Grenthe I (1992) Chemical Thermodynamics of Uranium. North Holland, Amsterdam Haas JR, DiChristina TJ (2002) Effects of Fe(III) chemical speciation on dissimilatory Fe(III) reduction by Shewanella putrefaciens. Environ Sci Technol 63:373-380 Haines RI, Owen DG, Vandergraaf TF (1987) Technetium-iron oxide reactions under anaerobic conditions: a Fourier transform infrared, FTIR study. Nuclear J Canada 1:32-37 Hansel CM, Benner SG, Neiss J, Dohnalkova A, Kukkadapu RK, Fendorf S (2003) Secondary mineralization pathways induced by dissimilatory iron reduction of ferrihydrite under advective low. Geochim Cosmochim Acta 67:2977-2992 Hansel CM, Benner SG, Nico P, Fendorf S (2004) Structural constraints of ferric (hydr)oxides on dissimilatory iron reduction and the fate of Fe(II). Geochim Cosmochim Acta 68:3217-3229 Harada E, Kumagai J, Ozawa K, Imabayashi S, Tsapin AS, Nealson KH, Meyer TE, Cusanovich MA, Akutsu H (2002) A directional electron transfer regulator based on heme-chain architecture in the small tetraheme cytochrome c from Shewanella oneidensis. FEBS Lett 532:333-7 Heidelberg JF, Paulsen IT, Nelson KE, Gaidos EJ, Nelson WC, Read TD, Eisen JA, Seshadri R, Ward N, Methé B, et al. (2002) Genome sequence of the dissimilatory metal ion-reducing bacterium Shewanella oneidensis. Nat Biotechnol 20:1118-23

Enzymology of Electron Transport

49

Hernandez ME, Kappler A, Newman DK (2004) Phenazines and other redox-active antibiotics promote microbial mineral reduction. Appl Environ Microbiol 70:921-8 Kukkadapu RK, Zachara JM, Fredrickson JK, Kennedy DW, Dohnalkova AC, McCready DE (2005) Ferrous hydroxy carbonate is a stable transformation product of biogenic magnetite. Am Mineral 90:510-515 Kashei K, Lovley DR (2000) Reduction of Fe(III), Mn(IV), and toxic metals at 100 degrees C by Pyrobaculum islandicum. Appl Environ Microbiol 66:1050-1056 Khijniak TV, Medvedeva-Lyalikova NN, Simonoff M (2003) Reduction of pertechnetate by haloalkaliphilic strains of Halomonas. FEMS Microbiol Ecol 44:109-115 Kieft TL, Fredrickson JK, Onstott TC, Gorby YA, Kostandarithes HM, Bailey TJ, Kennedy DW, Li SW, Plymale AE, Spadoni CM, Gray MS (1999) Dissimilatory reduction of Fe(III) and other electron acceptors by a Thermus isolate. Appl Environ Microbiol 65:1214-1221 Kostka JE, Haefele E, Viehweger R, Stucki JW (1999) Respiration and dissolution of iron(III) containing clay minerals by bacteria. Environ Sci Technol 33:3127-3133 Kostka JE, Stucki JW, Nealson, KH, Wu J (1996) Reduction of structural Fe(III) in smectite by a pure culture of Shewanella putrefaciens MR-1. Clays Clay Mineral 44:522-529 Kostka JE, Luther GW, Nealson KH (1995) Chemical and biological reduction of Mn(III)-pyrophosphate complexes: potential importance of dissolved Mn(III) as an environmental oxidant. Geochim Cosmochim Acta 59:885-894 Kukkadapu RK, Zachara JM, Fredrickson JK, Kennedy DW (2004) Biotransformation of two-line silicaferrihydrite by a dissimilatory Fe(III)-reducing bacterium: formation of carbonate green rust in the presence of phosphate. Geochim Cosmochim Acta 68:2799-2814 Kukkadapu RK, Zachara JM, Fredrickson JK, Smith SC, Dohnalkova AC, Russell CK (2003) Transformation of 2-line ferrihydrite to 6-line ferrihydrite under oxic and anoxic conditions. Am Mineral 88:1903-1914 Kukkadapu RK, Zachara JM, Smith SC, Fredrickson JK, Liu CX (2001) Dissimilatory bacterial reduction of Al-substituted goethite in subsurface sediments. Geochim Cosmochim Acta 65:2913-2924 Langmuir D (1997) Aqueous Environmental Geochemistry. Prentice Hall, Upper Saddle River, New Jersey Leang C, Coppi MV, Lovley DR (2003) OmcB, a c-type polyheme cytochrome, involved in Fe(III) reduction in Geobacter sulfurreducens. J Bacteriol 185:2096-2103 Leang C, Lovley DR (2005) Regulation of two highly similar genes, OmcB and OmcC, in a 10 kb chromosomal duplication in Geobacter sulfurreducens. Microbiology-Sgm 151:1761-1767 Lengeler JW, Drews G, Schlegel HG (1999) Biology of the Prokaryotes. Georg Thieme Verlag, Stuttgart, Germany Liger E, Charlet L, Van Cappellen P (1999) Surface catalysis of uranium(VI) reduction by iron(II). Geochim Cosmochim Acta 63:2939-2955 Liu CG, Zachara JM, Gorby YA, Szecsody JE, Brown CF (2001a) Microbial reduction of Fe(III) and sorption/ precipitation of Fe(II) on Shewanella putrefaciens strain CN32. Environ Sci Technol 35:1385-1393 Liu CX, Gorby YA, Zachara JM, Fredrickson JK, Brown CF (2002a) Reduction kinetics of Fe(III), Co(III), U(VI) Cr(VI) and Tc(VII) in cultures of dissimilatory metal-reducing bacteria. Biotechnol Bioeng 80: 637-649 Liu CX, Kota S, Zachara JM, Fredrickson JK, Brinkman CK (2001b) Kinetic analysis of the bacterial reduction of goethite. Environ Sci Technol 35:2482-2490 Liu CX, Zachara JM, Fredrickson JK, Kennedy DW, Dohnalkova A (2002b) Modeling the inhibition of the bacterial reduction of U(VI) by beta-MnO2(S)(g). Environ Sci Technol 36:1452-1459 Lloyd JR (2003) Microbial reduction of metals and radionuclides. FEMS Microbiol Rev 27:411-425 Lloyd JR, Cole JA, Macaskie LE (1997) Reduction and removal of heptavalent technetium from solution by Escherichia coli. J Bacteriol 179:2014-2021 Lloyd JR, Leang C, Myerson ALH, Coppi MV, Cuifo S, Methé B, Sandler SJ, Lovley DR (2003) Biochemical and genetic characterization of PpcA, a periplasmic c-type cytochrome in Geobacter sulfurreducens. Biochem J 369:153-161 Lloyd JR, Sole VA, Van Praagh CVG, Lovley DR (2000) Direct and Fe(II)-mediated reduction of technetium by Fe(III)-reducing bacteria. Appl Environ Microbiol 66:3743-3749 Lovley DR (1995) Bioremediation of organic and metal contaminants with dissimilatory metal reduction. J Indust Microbiol 14:85-93 Lovley DR, Coates JD, BluntHarris EL, Phillips EJP, Woodward JC (1996) Humic substances as electron acceptors for microbial respiration. Nature 382:445-448 Lovley DR, Holmes DE, Nevin KP (2004) Dissimilatory Fe(III) and Mn(IV) reduction. Adv Microb Physiol 49:219-86 Lovley DR, Phillips EJP (1986) Availability of ferric iron for microbial reduction in bottom sediments of the fresh-water Tidal Potomac River. Appl Environ Microbiol 52:751-757 Lovley DR, Phillips EJP (1988) Novel mode of microbial energy-metabolism - organic-carbon oxidation coupled to dissimilatory reduction of iron or manganese. Appl Environ Microbiol 54:1472-1480

50

DiChristina, Fredrickson, Zachara

Lovley DR, Phillips EJP, Gorby YA, Landa ER (1991) Microbial reduction of uranium. Nature 350:413-416 Lovley DR, Stolz JF, Nord GL, Phillips EJP (1987) Anaerobic production of magnetite by a dissimilatory ironreducing microorganism. Nature 330:252-254 Lovley DR, Widman PK, Woodward JC, Phillips EJP (1993) Reduction of uranium by cytochrome-C(3) of Desulfovibrio Vulgaris. Appl Environ Microbiol 59:3572-3576 Lovley DR, Woodward JC (1996) Mechanisms for chelator stimulation of microbial Fe(III)-oxide reduction. Chem Geol 132:19-24 Lyalikova NN, Khizhnyak TV (1996) Reduction of heptavalent technetium by acidophilic bacteria of the genus Thiobacillus. Microbiology 65:468-473 Madigan MT, Martinko JM (2006) Brock Biology of Microorganisms Prentice Hall, Upper Saddle River, NJ Methé BA, Webster J, Nevin K, Butler J, Lovley DR (2005) DNA microarray analysis of nitrogen ixation and Fe(III) reduction in Geobacter sulfurreducens. Appl Environ Microbiol 71:2530-2538 Meyer RE, Arnold WD, Case FI, O’Kelley GD (1991) Solubilities of Tc(IV)-oxides. Radiochim Acta 55:11-18 Myers CR, Myers JM (1992) Localization of cytochromes to the outer membrane of anaerobically grown Shewanella putrefaciens MR-1. J Bacteriol 174:3429-38 Myers CR, Myers JM (1993a) Ferric reductase is associated with the membranes of anaerobically grown Shewanella Putrefaciens MR-1. FEMS Microbiol Lett 108:15-22 Myers CR, Myers JM (1993b) Role of menaquinone in the reduction of fumarate, nitrate, iron(III) and manganese(IV) by Shewanella Putrefaciens MR-1. FEMS Microbiol Lett 114:215-222 Myers CR, Myers JM (1997) Cloning and sequence of cymA a gene encoding a tetraheme cytochrome c required for reduction of iron(III), fumarate, and nitrate by Shewanella putrefaciens MR-1. J Bacteriol 179:1143-1152 Myers CR, Myers JM (2002) MtrB is required for proper incorporation of the cytochromes OmcA and OmcB into the outer membrane of Shewanella putrefaciens MR-1. Appl Environ Microbiol 68:5585-94 Myers CR, Myers JM (2003a) Cell surface exposure of the outer membrane cytochromes of Shewanella oneidensis MR-1. Lett Appl Microbiol 37:254-258 Myers CR, Nealson KH (1988) Microbial reduction of manganese oxides - interactions with iron and sulfur. Geochim Cosmochim Acta 52:2727-2732 Myers CR, Nealson KH (1990) Respiration-linked proton translocation coupled to anaerobic reduction of manganese(IV) and iron(III) in Shewanella putrefaciens MR-1. J Bacteriol 172:6232-8 Myers JM, Myers CR (2000) Role of the tetraheme cytochrome CymA in anaerobic electron transport in cells of Shewanella putrefaciens MR-1 with normal levels of menaquinone. J Bacteriol 182:67-75 Myers JM, Myers CR (2001) Role for outer membrane cytochromes OmcA and OmcB of Shewanella putrefaciens MR-1 in reduction of manganese dioxide. Appl Environ Microbiol 67:260-9 Myers JM, Myers CR (2003b) Overlapping role of the outer membrane cytochromes of Shewanella oneidensis MR-1 in the reduction of manganese(IV) oxide. Lett Appl Microbiol 37:21-5 Nealson KH, Belz A, McKee B (2002) Breathing metals as a way of life: geobiology in action. Antonie Van Leeuwenhoek Int J Gen Molec Microbiol 81:215-222 Nealson KH, Little B (1997) Breathing manganese and iron: solid-state respiration. Adv Appl Microbiol 45: 213-239 Nealson KH, Myers CR (1992) Microbial reduction of manganese and iron: new approaches to carbon cycling. Appl Environ Microbiol 58:439-43 Nealson KH, Saffarini D (1994) Iron and manganese in anaerobic respiration - environmental signiicance, physiology, and regulation. Annu Rev Microbiol 48:311-343 Nelson KE, Methé B (2005) Metabolism and genomics: adventures derived from complete genome sequencing. Rev Mineral Geochem 59:279-294 Nevin KP, Lovley DR (2000) Lack of production of electron-shuttling compounds or solubilization of Fe(III) during reduction of insoluble Fe(III) oxide by Geobacter metallireducens. Appl Environ Microbiol 66: 2248-2251 Nevin KP, Lovley DR (2002) Mechanisms for accessing insoluble Fe(III) oxide during dissimilatory Fe(III) reduction by Geothrix fermentans. Appl Environ Microbiol 68:2294-2299 Newman DK, Gralnick JA (2005) What genetics offers geobiology. Rev Mineral Geochem 59:9-26 Newman DK, Kolter R (2000) A role for excreted quinones in extracellular electron transfer. Nature 405:94-97 Ortiz-Bernad I, Anderson RT, Vrionis HA, Lovley DR (2004) Resistance of solid-phase U(VI) to microbial reduction during in situ bioremediation of uranium-contaminated groundwater. Appl Environ Microbiol 70:7558-7560 Pasilis SP, Pemberton JE (2003) Speciation and coordination chemistry of uranyl(VI)-citrate complexes in aqueous solution. Abstr Papers Am Chem Soc 225:U807-U807 Payne AN, DiChristina TJ (2005) Submitted for review. Payne RB, Gentry DM, Rapp-Giles BJ, Casalot L, Wall JD (2002) Uranium reduction by Desulfovibrio desulfuricans strain G20 and a cytochrome c3 mutant. Appl Environ Microbiol 68:3129–3132

Enzymology of Electron Transport

51

Payne RB, Casalot L, Rivere T, Terry JH, Larsen L, Giles BJ, Wall JD (2004) Interaction between uranium and the cytochrome c3 of Desulfovibrio desulfuricans strain G20. Arch Microbiol 181:398-406 Pitts KE, Dobbin PS, Reyes-Ramirez F, Thomson AJ, Richardson DJ, Seward HE (2003) Characterization of the Shewanella oneidensis MR-1 decaheme cytochrome MtrA: expression in Escherichia coli confers the ability to reduce soluble Fe(III) chelates. J Biol Chem 278:27758-65 Pugsley AP (1993) The complete general secretory pathway in gram-negative bacteria. Microbiol Rev 57:50108 Pugsley AP, Francetic O, Hardie K, Possot OM, Sauvonnet N, Seydel A (1997) Pullulanase: model protein substrate for the general secretory pathway of gram-negative bacteria. Folia Microbiologica 42:184-192 Rai D, Felmy AR, Ryan JL (1990) Uranium(IV) hydrolysis constants and solubility product of UO2·xH2O(Am). Inorganic Chemistry 29:260-264 Rard JA (1999) Chemical Thermodynamics of Technetium. Elsevier, Amsterdam. Rard JA (1983) Critical review of the chemistry and thermodynamics of technetium and some of its inorganic compounds and aqueous species. Livermore, California Richardson DJ (2000) Bacterial respiration: a lexible process for a changing environment. Microbiology-Sgm 146:551-571 Roden EE (2003) Fe(III) oxide reactivity toward biological versus chemical reduction. Environ Sci Technol 37: 1319-1324 Roden EE, Sobolev D, Glazer B, Luther GW (2004) Potential for microscale bacterial Fe redox cycling at the aerobic-anaerobic interface. Geomicrobiol J 21:379-391 Roden EE, Urrutia MM (1999) Ferrous iron removal promotes microbial reduction of crystalline iron(III) oxides. Environ Sci Technol 33:2492-2492 Roden EE, Urrutia MM (2002) Inluence of biogenic Fe(II) on bacterial crystalline Fe(III) oxide reduction. Geomicrobiol J 19:209-252 Roden EE, Urrutia MM, Mann CJ (2000) Bacterial reductive dissolution of crystalline Fe(III) oxide in continuous-low column reactors. Appl Environ Microbiol 66:1062-1065 Roden EE, Zachara JM (1996) Microbial reduction of crystalline iron(III) oxides: inluence of oxide surface area and potential for cell growth. Environ Sci Technol 30:1618-1628 Rooney-Varga JN, Anderson RT, Fraga JL, Ringelberg D, Lovley DR (1999) Microbial communities associated with anaerobic benzene degradation in a petroleum-contaminated aquifer. Appl Environ Microbiol 65: 3056-3063 Saffarini DA, Blumerman SL, Mansoorabadi KJ (2002) Role of menaquinones in Fe(III) reduction by membrane fractions of Shewanella putrefaciens. J Bacteriol 184:846-8 Saltikov CW, Cifuentes A, Venkateswaran K, Newman DK (2003) The ars detoxiication system is advantageous but not required for As(V) respiration by the genetically tractable Shewanella species strain ANA-3. Appl Environ Microbiol 69:2800-2809 Schwalb C, Chapman SK, Reid GA (2003) The tetraheme cytochrome CymA is required for anaerobic respiration with dimethyl sulfoxide and nitrite in Shewanella oneidensis. Biochemistry 42:9491-9497 Shyu JB, Lies DP, Newman DK (2002) Protective role of tolC in eflux of the electron shuttle anthraquinone2,6-disulfonate. J Bacteriol 184:1806-10 Sobolev D, Roden EE (2002) Evidence for rapid microscale bacterial redox cycling of iron in circumneutral environments. Antonie Van Leeuwenhoek Int J Gen Molec Microbiol 81:587-597 Stumm W (1992) Chemistry of the Solid-Water Interface. Wiley Interscience, New York. Suzuki Y, Kelly SD, Kemner KM, Banield JF (2002) Radionuclide contamination - nanometer-size products of uranium bioreduction. Nature 419:134-134 Taillefert M, DiChristina TJ (2005) Submitted for review. Taillefert M, Hover VC, Rozan TF, Theberge SM, Luther III GW (2002) The inluence of sulides on soluble organic-Fe(III) in anoxic sediment porewaters. Estuaries 25:1088-1096 Turick CE, Tisa LS, Caccavo F Jr. (2002) Melanin production and use as a soluble electron shuttle for Fe(III) oxide reduction and as a terminal electron acceptor by Shewanella algae BrY. Appl Environ Microbiol 68: 2436-44 Urrutia MM, Roden EE, Zachara JM (1999) Inluence of aqueous and solid-phase Fe(II) complexants on microbial reduction of crystalline iron(III) oxides. Environ Sci Technol 33:4022-4028 Wade R, DiChristina TJ (2000) Isolation of U(VI) reduction-deicient mutants of Shewanella putrefaciens. FEMS Microbiol Lett 184:143-148 Whitaker RJ, Banield JF (2005) Population dynamics through the lens of extreme environments. Rev Mineral Geochem 59:259-277 Wielinga B, Bostick B, Hansel CM, Rosenzweig RF, Fendorf S (2000) Inhibition of bacterially promoted uranium reduction: Ferric (hydr)oxides as competitive electron acceptors. Environ Sci Technol 34:21902195

52

DiChristina, Fredrickson, Zachara

Wielinga B, Mizuba MM, Hansel CM, Fendorf S (2001) Iron promoted reduction of chromate by dissimilatory iron-reducing bacteria. Environ Sci Technol 35:522-527 Wildung RE, Gorby YA, Krupka KM, Hess NJ, Li SW, Plymale AE, McKinley JP, Fredrickson JK (2000) Effect of electron donor and solution chemistry on products of dissimilatory reduction of technetium by Shewanella putrefaciens. Appl Environ Microbiol 66:2451-2460 Wildung RE, Li SW, Murray CJ, Krupka KM, Xie Y, Hess NJ, Roden EE (2004) Technetium reduction in sediments of a shallow aquifer exhibiting dissimilatory iron reduction potential. FEMS Microbiol Ecol 49:151-162 Zachara JM, Fredrickson JK, Li SM, Kennedy DW, Smith SC, Gassman PL (1998) Bacterial reduction of crystalline Fe3+ oxides in single phase suspensions and subsurface materials. Am Mineral 83:1426-1443 Zachara JM, Fredrickson JK, Smith SC, Gassman PL (2001) Solubilization of Fe(III) oxide-bound trace metals by a dissimilatory Fe(III) reducing bacterium. Geochim Cosmochim Acta 65:75-93 Zachara JM, Kukkadapu RK, Gassman PL, Dohnalkova A, Fredrickson JK, Anderson T (2004) Biogeochemical transformation of Fe minerals in a petroleum-contaminated aquifer. Geochim Cosmochim Acta 68:17911805 Zachara JM, Kukkadapu RK, Fredrickson JK, Gorby YA, Smith SC (2002) Biomineralization of poorly crystalline Fe(III) oxides by dissimilatory metal reducing bacteria. Geomicrobiol J 19:179-207

Reviews in Mineralogy & Geochemistry Vol. 59, pp. 53-84, 2005 Copyright © Mineralogical Society of America

4

Siderophores and the Dissolution of Iron-Bearing Minerals in Marine Systems Stephan M. Kraemer Department of Environmental Sciences ETH Zürich 8092 Zürich, Switzerland [email protected]

Alison Butler Department of Chemistry and Biochemistry University of California, Santa Barbara Santa Barbara, California, 93106-9510, U.S.A.

Paul Borer Swiss Federal Institute of Aquatic Science and Technology 8600 Dübendorf, Switzerland

Javiera Cervini-Silva Department of Earth and Planetary Science University of California, Berkeley Berkeley, California, 94720-3110, U.S.A. INTRODUCTION Scope of this review Metal ions have critical functions in biological processes and provided important biological feedbacks with the environment throughout earth history. For example, Fe, Ni, Mg, Mn, Mo, Cu, W, V, and Zn play an essential role as catalysts in key compounds involved in respiration, photosynthesis, nitrogen ixation, and many other enzymatic processes (da Silva and Williams 2001). It is likely that some of these metal bearing enzymes evolved early in the history of life. However, the availability of metal ions has changed dramatically in the last 3.8 billion years due to changes in atmospheric and marine chemistry (Canield 1998; Anbar and Knoll 2002; Saito et al. 2003). Homeostasis of metal ions (i.e., the maintenance of an approximately constant intracellular concentration) became a major problem over geologic time scales and in contemporary environments. Iron is an essential nutrient for almost all known organisms due to its important role in important enzymatic processes. While iron is the fourth most abundant element in the Earth crust, its low bioavailability limits primary production in various terrestrial and marine environments. The limitation of primary production in important ecosystems has signiicant implications for the global carbon cycle and the world climate. This review focuses on geochemical aspects of biological iron acquisition in iron limited “high nutrient low chlorophyll” (HNLC) ocean regions. The purpose of this review is to discuss the effect of biogenic iron speciic ligands, the so called siderophores, on the iron speciation, and the dissolution of iron-bearing minerals in the presence of siderophores in these marine systems. 1529-6466/05/0059-0004$05.00

DOI: 10.2138/rmg.2005.59.4

54

Kraemer, Butler, Borer, Cervini-Silva

Important iron sources for algal growth in HNLC ocean regions are the upward mixing of iron rich subsurface waters to the euphotic zone and the atmospheric deposition of dust particles on the sea surface followed by the dissolution of iron from these particles into the surface water. Iron sinks include the scavenging of iron onto inorganic and organic particles in the water and subsequent settling and sedimentation of these particles out of the water column. A number of thermodynamically or kinetically controlled processes inluence the redistribution of iron among various chemical species in the water and on the water-solid interfaces along the water column. Among those processes are dissolution and formation of iron-bearing minerals, iron adsorption on and desorption from inorganic and organic particles, (photo-) reduction of Fe(III) and oxidation of Fe(II), and iron complexation by organic and inorganic ligands. Marine micro-organisms and phytoplankton may inluence or even regulate these processes through 1) the biological exudation of iron binding ligands; 2) reduction of Fe(III) at the cell surface or Fe(III) photo reduction in the presence of biogenic chromophores which may change the redox speciation of iron; 3) transformation of colloidal iron to more bioavailable forms in the digestive system of grazers; 4) biological recycling of iron from sinking biomass. In iron limiting environments, such biogeochemical processes are critical to enhance iron uptake and/or to decrease iron bioavailability for competing species. Unfortunately, very little is known about the kinetics and luxes associated with many of these processes. An important impediment to ield research in this area is the astoundingly low iron concentration in the HNLC surface waters. Dissolved iron in these ocean areas have a nutrient-like vertical distribution with average sea surface concentrations in the subnanomolar range and increasing concentrations with depth (Johnson et al. 1997). Considering the tremendous problems involved in contamination free sampling and chemical analysis in this concentration range, a remarkable range of information has been gathered on dissolved organic and inorganic iron speciation. A key process of iron cycling in HNLC ocean regions and the focus of this review is the dissolution of iron from dust as part of biological iron acquisition strategies. We will summarize pertinent information from ield and laboratory studies and discuss important dissolution mechanisms. For in-depth discussion of related issues the reader is referred to a number of excellent review articles (Jickells 1999; Boyd 2002; Morel and Price 2003).

The iron limitation hypothesis Over glacial/interglacial time scales, CO2 levels in the earth atmosphere are strongly inluenced by phytoplanktonic photosynthesis in the oceans. In HNLC ocean areas phytoplankton productivity and consequently the eficiency of the biological CO2 pump is not limited by macronutrients such as phosphate or nitrate, since their concentrations at the HNLC sea surface waters are high. Martin and Fitzwater (1988) proposed that primary productivity in these ocean areas is limited by the low availability of the micronutrient iron. Iron fertilization experiments in the Equatorial Paciic (IRONEX I and II), the Southern Ocean (SOIREE, EisenEx, SOFeX-N and SOFeX-S) and the Sub-Arctic North Paciic (SEEDS, SERIES) have decisively supported this hypothesis (Martin et al. 1994; Coale et al. 1996, 2004; Boyd and Law 2001; Gervais et al. 2002; Tsuda et al. 2003; Boyd et al. 2004). Vast ocean areas are affected by iron limitation and it may be as important as nitrogen and phosphorous limitation of global marine phytoplankton productivity (Moore et al. 2002).

BIOLOGICAL IRON ACQUISITION STRATEGIES Iron acquisition by bacteria Considering the importance of iron as a limiting nutrient, biological iron acquisition is a key factor determining the ecology of HNLC ocean regions. An important iron acquisition

Siderophores & Dissolution of Iron Bearing Minerals

55

strategy among both freshwater and marine cyanobacteria and heterotrophic bacteria involves the production of siderophores under iron limiting conditions (Gonye and Carpenter 1974; Trick 1989; Wilhelm and Trick 1994; Wilhelm 1995; Granger and Price 1999). Siderophores are low molecular weight organic ligands (0.5–1.5 kDa) with high afinity and speciicity for iron. The siderophore mediated uptake of iron involves the recognition of the siderophore complex and the transport of the ferric siderophore complex across the cell membrane (Reid and Butler 1991; Butler 1998; Murakami et al. 2000; Guan et al. 2001; Armstrong et al. 2004). Although siderophore exudation is an important bacterial response to iron limitation, not all cyanobacteria and heterotrophic bacteria produce siderophores. While some cyanobacteria have been reported to produce multiple siderophores (Wilhelm and Trick 1994), other strains produce none (Wilhelm 1995). Similar results are reported for heterotrophic bacteria. In a recent study, from a total of 421 strains of heterotrophic marine bacteria which were isolated from marine sponges and seawater, 223 strains were observed not to produce siderophores under iron limiting conditions (Guan et al. 2001). However, the growth of 134 stains out of the total non-siderophore producing strains was stimulated by both cross-streaking of siderophore producing strains and by the addition of siderophores. Similarly, Trick (1989) showed that some marine bacteria produced siderophores which promoted growth of unrelated isolates, while other siderophores only satisied the iron requirements of the strains that produced them. Granger and Price (1999) isolated strains of heterotrophic bacteria and found that not all of them produced siderophores under the assay conditions, but all took up Fe bound to siderophores. The utilization of ferric siderophore complexes by non-siderophore producing bacterial strains may be a typical pattern in iron limited aquatic systems, since siderophore excretion is considered to be metabolically expensive (Völker and Wolf-Gladrow 1999). However, siderophore production may simply also be a “survive at all cost” response to low iron availability and the high metabolic costs are offset by the potential for survival (Wilhelm 1995).

Iron acquisition by eukaryotic phytoplankton Fe(III)-siderophore complexes may not only be an essential iron source for heterotrophic and phototrophic bacteria (prokaryotes), but also for eukaryotic phytoplankton. It has been reported that that some eukaryotic species are able to acquire iron from various iron complexes and siderophores (Allnutt and Bonner 1987; Soria-Dengg and Horstmann 1995; Kuma et al. 2000; Maldonado and Price 2001), even if they generally do not produce siderophores themselves (for exceptions see Trick et al. 1983; Benderliev and Ivanova 1994; Benderliev 1999). An important process in this context is the reduction of organically bound Fe(III) by a plasma membrane ferrireductase, which promotes the dissociation of Fe(II) from the siderophore complex (Jones et al. 1987; Weger 1999). The inorganic iron is then taken up by membrane transporters. However, the acquisition of iron from iron-siderophore complexes by eukaryotic phytoplankton (e.g., diatoms) is controversially discussed in literature. It has been shown that ferric organic complexes, including model Fe-siderophore complexes like Fe-DFOB are utilized as iron sources by some eukaryotic phytoplankton species (Soria-Dengg and Horstmann 1995; Maldonado and Price 1999, 2001). Several studies found that uptake of iron is largely inhibited by DFO-B additions or at least insuficient to satisfy cellular requirements of eukaryotic phytoplankton (Hutchins et al. 1999; Wells 1999; Timmermans et al. 2001; Eldridge et al. 2004). It is important to note in this context that adaptation of former iron-replete eukaryotic phytoplankton to artiicially induced iron limitation may require the activation of membrane bound ferric chelate reductases or high-afinity transport systems over timescales that are longer than typical shipboard incubation experiments (Maldonado and Price 2001; Wells and Trick 2004). Hutchins et al. (1999) observed iron uptake from weaker complexes (e.g., Fe-porphyrin). In contrast to most siderophores, the Fe(III)porphyrin complex exhibits a tetradentate structure, so that iron bound to porphyrin may be

56

Kraemer, Butler, Borer, Cervini-Silva

more accessible to surface reductase of eukaryotic plankton and may be reduced more easily. Consequently, the complexation of iron by siderophores may increases the bioavailability of iron for bacterial species, but also may reduce iron availability for eukaryotic phytoplankton (depending on the nutritional status), unless the iron-siderophore complexes are transformed by secondary processes (e.g., by photochemical processes). Marine algae can change iron speciation by excretion of low and high molecular weight organic substances (Lancelot 1984; Fuse et al. 1993; Myklestad 1995). In iron-limited cultures of the coccolithophore Emiliania huxylei, the release of strong iron-chelating ligands (with conditional stability constants comparable to those of siderophores) was observed under ironlimiting conditions and increased after inorganic iron was added to the cultures (Boye and van den Berg 2000). These observations suggest that the observed ligands were not siderophores (where the release is triggered by iron limitation). An increase of the concentrations of iron binding ligands was also observed as a response to iron fertilization in mesoscale experiments (Rue and Bruland 1997). Potential sources of strong Fe-chelators from phytoplankton in seawater are exudation or the rupture of intact cells. Cell rupture could lead, for example, to the release of compounds with porphyrin-type moieties with a high afinity for binding iron (Witter et al. 2000).

Role of protozoan grazers in the cycling of iron Radiotracer studies indicate that iron is rapidly cycled within the planktonic community by linked biological processes such as grazing, excretion, viral lysis, and bacterial respiration of organic matter (Hutchins et al. 1993). Regeneration of iron among the phytoplankton community does not necessarily require grazing of phytoplankton by heterotrophic protozoan grazers. Regeneration of iron is also accomplished by the grazing of bacteria through photosynthetic protozoan grazers, so called mixotrophs. Mixotrophy—deined as the ability to assimilate organic compounds as carbon sources while using inorganic compounds as electron donors for energy metabolism (Madigan et al. 2000)—is a widespread phenomenon in aquatic habitats and is observed in many ciliates and lagellates (Stoecker 1998). It has been suggested that phagotrophic ingestion of bacteria may be an adaptive strategy for photosynthetic algae to obtain iron for growth in iron limited regions of the sea. In a recent study, it has been shown that the photosynthetic lagellate Ochromonas sp. can obtain iron directly by ingesting bacteria (Maranger et al. 1998). As Ochromonas also excretes some of the Fe it ingests, phagotrophic phytolagellates may in general play an important role in the Fe cycle by regenerating Fe for themselves and for other microorganisms. Heterotrophic protozoan grazers may also generate bioavailable iron by digestion of refractory iron phases in the acidic food vacuoles. Barbeau et al. (1996) have demonstrated several grazer mediated effects on colloidal ferrihydrite, including a decrease in colloid size, an increase in colloid lability as determined by competitive ligand exchange techniques, and an increase in the bioavailability of colloids to iron limited diatoms. These results indicate that protozoan grazers may signiicantly enhance the supply of iron to marine phytoplankton from terrestrial sources. It has been estimated that protozoan grazing of colloidal particles (e.g., ferrihydrite) may equal or exceed photoreductive dissolution in increasing iron availability to phytoplankton (Barbeau et al. 1996; Barbeau and Moffett 2000). The effect of grazing on more crystalline iron oxides is not known.

SOURCES OF IRON IN HNLC OCEAN REGIONS The most important iron sources in HNLC ocean areas are upwelling and atmospheric deposition of iron derived from continental dust (Archer and Johnson 2000; Moore et al. 2002, 2004). The atmospheric lux of iron to the remote HNLC ocean areas is relatively low.

Siderophores & Dissolution of Iron Bearing Minerals

57

However, iron deiciency is not only triggered by low total iron concentrations but also by the low bioavailability of iron from atmospheric inputs. As a consequence of its low bioavailability, only a small fraction of the iron input by atmospheric aerosol is solubilized before sedimentation (Zhuang et al. 1990; Fung et al. 2000). The bioavailability of iron is strongly inluenced by the mineralogy of aerosol particles which in turn is inluenced by the characteristics of soils in the source area of the particles. Iron speciation can further be modiied by the chemical environment during atmospheric transport (Jickells 1999). Iron-bearing minerals can be dissolved by protonpromoted and photoreductive dissolution and re-precipitated during drying cycles (Spokes et al. 1994; Siefert et al. 1999; Johansen et al. 2000).

Atmospheric dust as a source of iron A strong link between dust deposition and primary production has been established by observations of increased chlorophyll and particulate organic carbon concentrations at the sea surface during episodic dust deposition events (Ditullio and Laws 1991; Lenes et al. 2001; Bishop et al. 2002). The most important source for atmospheric dust are the arid or semiarid regions of the continents (Duce et al. 1980; Duce and Tindale 1991; Tegen and Fung 1995; Mahowald et al. 1999; Perry et al. 1999; Ginoux et al. 2001). The iron content of atmospheric dust is roughly related to the average crustal abundance of iron (3.5%; Taylor 1964) and depends on the continental source area (Hand et al. 2004) and on size fractionation during transport (Claquin et al. 1999; Jickells 1999). A number of researchers have estimated aeolian iron luxes to the oceans based on measurements or models of atmospheric dust distributions over the oceans (Duce and Tindale 1991; Tegen and Fung 1995; Mahowald et al. 1999) and on global assumptions or databases of local measurements of the iron content in dust (Archer and Johnson 2000; Gao et al. 2001; Moore et al. 2002).

Iron mineralogy of atmospheric dust The distribution of iron among iron-bearing mineral phases has an important effect on its solubility and lability (Cornell and Schwertmann 2003). The fate of iron during atmospheric transport and after deposition at the sea surface will therefore strongly depend on the mineralogy of aerosol particles. Nanoparticulate hematite (11–170 nm) was observed in Mössbauer studies of atmospheric aerosols collected in a rural area of Poland (Kopcewicz and Kopcewicz 1994, 1998). Reid et al. (2003) report single particle analysis of aerosol particles of Saharan origin collected at Puerto Rico. They found most iron associated with particles with elemental abundances corresponding to illite. The elemental composition of a smaller fraction of high Fe particles suggested kaolinite aggregated with iron oxides. However, the authors noted the dificulty of assigning mineralogical structures to aggregates based on elemental abundances. Falkovich et al. (2001) used SEM-EDS and XRD to analyze individual dust particles of North African origin collected over Israel. They found most iron on particle surfaces and suggested that hematite aggregated with clay minerals were coating particle surfaces. Direct evidence of the presence of hematite and goethite was provided by diffuse relectance spectrometry of aerosol particles collected from Bermuda, Barbados, and Izaña (Arimoto et al. 2002). Crystalline iron oxides of primarily aeolian origin were also found in deep sea sediments (Bloemendal et al. 1992; Balsam et al. 1995).

Transformation of iron-bearing minerals during atmospheric transport The bioavailability of iron is strongly inluenced by it’s speciation in the aerosol particles. Iron-bearing minerals can be transformed by proton-promoted and photoreductive dissolution as well as precipitation during drying cycles (Spokes et al. 1994; Siefert et al. 1998). The observation of signiicant concentrations of dicarboxylic acids in marine aerosols may promote ligand-controlled and photoreductive dissolution mechanisms (Stephanou and Stratigakis 1993; Sempere and Kawamura 2003).

Kraemer, Butler, Borer, Cervini-Silva

58

Reduction of Fe(III) to Fe(II) dramatically increases its solubility. Reduced iron species are thermodynamically unstable under atmospheric conditions, but photoredox reactions can lead to the signiicant transient Fe(II) concentrations, particularly in acidic aerosols (Behra and Sigg 1990). Even though Fe(II) is rapidly reoxidized in seawater (Millero et al. 1987; Millero and Sotolongo 1989; King et al. 1995; King 1998), the precipitating Fe(III)-polymers are expected to be thermodynamically and kinetically less stable than iron-bearing minerals of the atmospheric dust. For these reasons, the redox speciation of iron in aerosols is a subject of ongoing research. Most studies of iron redox speciation show that the fraction of Fe(II) in acid extracts of aerosols represents only a small fraction of the total iron (Zhuang et al. 1990; Zhu et al. 1993, 1997; Siefert et al. 1999; Johansen et al. 2000; Chen and Siefert 2004). Large variations of Fe(II) fractions in aerosols collected over the Atlantic and Paciic Oceans were observed by Hand et al. (2004), with signiicantly higher Fe(II) fractions in ine aerosols ( Fe(OH)3

Microbial Fe(II)-oxidation by anoxygenic phototrophs CO2+Fe2+ + hυ --> Fe(OH)3 + CH2O

Banded Iron Formation

Figure 5. Proposed mechanisms for the deposition of Precambrian banded iron formations in the presence or absence of molecular oxygen: oxidation of Fe(II) either indirectly by cyanobacterially produced O2 or directly by anoxygenic photosynthetic Fe(II)-oxidizing microorganisms.

Anaerobic Fe(II)-oxidizing nitrate-reducing bacteria Furthermore, it was discovered that microorganisms are capable of coupling oxidation of ferrous iron to dissimilatory reduction of nitrate (Hafenbradl et al. 1996; Straub et al. 1996). At pH 7, all redox pairs of the nitrate reduction pathway can accept electrons from ferrous iron because their redox potentials are more positive than that of the redox couple Fe(III)/Fe(II) (Tables 4 and 5). The irst observations of this metabolism were made with a lithotrophic enrichment culture that was transferred successively several times in medium that contained ferrous iron as sole electron donor (Straub et al. 1996). In this culture, ferrous iron oxidation coupled to nitrate reduction deinitely supported cell growth; no oxidation of Fe(II) occurred in the presence of heat-inactivated cells or when nitrate was omitted. This type of metabolism is likely to be more abundant than ferrous iron oxidation by anoxygenic phototrophs since it is not restricted to habitats that are exposed to light. Furthermore, most-probable-number studies combined with molecular techniques indicated that the ability to oxidize ferrous iron with nitrate as electron acceptor is widespread among bacteria: members of the D-, E-, J- and G- subgroup of the Proteobacteria as well as gram-positive bacteria are probably able to oxidize ferrous iron (Straub and Buchholz-Cleven 1998; Straub et al. 2004). For these reasons, it was not surprising that enrichments of ferrous iron-oxidizing nitrate reducers were successfully established with a variety of marine, brackish or freshwater sediment samples. However, continuous cultivation

Geomicrobiological Cycling of Iron

93

with ferrous iron as sole electron donor Table 4. Redox potentials of redox pairs relevant turned out to be impossible for most of for microbial nitrate reduction at pH 7.0 and these enrichments. After a few transfers, 25 °C (Thauer et al. 1977). ferrous iron was oxidized only in the presence (of low concentrations) of an Redox pair E0c [mV] organic substrate, e.g., 0.5 mM acetate. − − NO3 / NO2 +430 Accordingly, most Fe(II)-oxidizing nitrate reducers isolated so far need +350 NO2−/ NO an organic co-substrate for growth, i.e. NO/ N2O +1180 grow only mixotrophically with ferrous +1350 N2O/ N2 iron (Straub et al. 1996; Benz et al. 1998; Straub and Buchholz-Cleven 1998; Lack et al. 2002; Straub et al. 2004). For many Table 5. Redox potentials1 of some redox pairs of these strains that need an additional relevant for microbial reduction of iron oxides at organic substrate, it is questioned whether pH 7.0 and 25 °C (Thamdrup 2000). ferrous iron oxidation is beneicial and supports cell growth or whether iron is Redox pair E0c [mV] just oxidized in a rather unspeciic side 2+ reaction. Experiments with Azospira Fe5HO8·4H2O (ferrihydrite)/Fe +2 oryzae strain PS (formerly known as −88 J-FeOOH (lepidocrocite)/Fe2+ Dechlorosoma suillum) undoubtedly −274 D-FeOOH (goethite)/Fe2+ showed that oxidation of ferrous iron 2+ initiated only after the organic (co-) −287 D-Fe2O3 (hematite)/Fe substrate was completely oxidized Fe3O4 (magnetite)/Fe2+ −314 (Chaudhuri et al. 2001). However, at 1 Slightly varying data can be found in the literature because least for some mixotrophically ferrous redox potentials strongly depend on pH, temperature, iron-oxidizing strains, i.e. Acidovorax concentrations of reactants, crystal size of the iron oxide and thermodynamic data chosen for calculations. sp. strain BrG1, Aquabacterium sp. strain BrG2 and Thermomonas sp. strain BrG3, the situation was more complex because the oxidation of ferrous iron seemed to be regulated. Only if electrons from the organic substrate exceeded those from ferrous iron by a factor of ten or if the concentration of nitrate was limited, ferrous iron oxidation ceased completely (Straub et al. 2004). Recently, some strains were isolated from the deep sea that oxidized Fe(II) with nitrate in the absence of an additional organic substrate. Unfortunately, it is not clear whether these strains can actually grow with ferrous iron as the sole electron donor for several successive generations (Edwards et al. 2003).

Mechanisms of microbial Fe(II) oxidation The mechanism of microbial Fe(II) oxidation has been studied best with the acidophilic Fe(II) oxidizer Thiobacillus ferrooxidans. According to a present model, Fe(II) is oxidized to Fe(III) at the outer membrane of the cell (Blake and Johnson 2000). The electron is then transferred to a copper-containing protein (rusticyanin) which in turn transfers it to a periplasmic c-type cytochrome. From such cytochromes, electrons are inally passed on to O2 via cytochrome oxidase to form water. The exact pathway of the electron transfer from ferrous iron to oxygen is still not completely understood, and slightly varying models are described in the literature. However, there is general agreement that the initial step, i.e. the oxidation of ferrous iron, occurs outside the cell (Blake and Johnson 2000). In addition, it was shown for neutrophilic aerobic Fe(II)-oxidizing Leptothrix spp. that the oxidation of Fe(II) is catalyzed by Fe(II)-oxidizing compounds that are actively secreted by

Kappler & Straub

94

the cell (De Vrind-de Jong et al. 1990); an Fe(II)-oxidizing protein with a molecular weight of 150 kDa was identiied from spent culture medium of strain Leptothrix discophora (Corstjens et al. 1992). For anaerobic Fe(II) oxidation, it is unknown where in the cell or at the cell surface Fe(II) is oxidized, and it is not understood how the bacteria deal with the poor solubility of the product. In particular, it is unclear how Fe(II)-oxidizing microorganisms either avoid encrustation with ferric iron minerals (such as the phototrophic Fe(II)-oxidizer ‘Rhodobacter ferrooxidans’ strain SW2) or overcome encrustation such as the nitrate-reducing Fe(II)oxidizing strain BoFeN1 (Fig. 6). A microenvironment of lowered pH values in vicinity of the cells was observed around colonies of phototrophic Fe(II) oxidizers (‘Rhodobacter ferrooxidans’ strain SW2) ixed in semi-solid agarose (Kappler and Newman 2004). Such an acidiication could explain why these microorganisms do not become encrusted with ferric iron minerals during oxidation of Fe(II) (Fig. 6). With the aerobic Fe(II)-oxidizing strain TW2, deposition of Fe(III) minerals was observed not at the cell surface but at a certain distance from the cells. It was suggested that Fe(III) was released in a ligand-bound dissolved form. The dissolved Fe(III)-ligand complex is thought to

A

B

500 nm

500 nm

Figure 6. Scanning electron micrographs showing (A) cells of the nitrate-reducing Fe(II)-oxidizing strain BoFeN1 highly encrusted with Fe(III) minerals and (B) anoxygenic photosynthetic Fe(II)-oxidizing microorganisms that are associated but not encrusted with Fe(III) minerals.

Geomicrobiological Cycling of Iron

95

diffuse away from the cells. Destabilization of the Fe(III)-ligand complex would inally lead to hydrolysis and precipitation of Fe(III) minerals distant from the metabolically active cells (Roden et al. 2004). The nature of the Fe(III)-ligand and the trigger necessary for destabilizing the dissolved Fe(III)-ligand complex are unknown so far. However, this hypothesis is supported by energetic calculations. The estimated biomass yield for growth was 0.15 mol cell-C per mol oxidized Fe(II), and hence approximately 7.5u more than experimentally observed in gradient cultures. This suggests that a substantial amount of energy is available for synthesis of other cellular components, including Fe(III)-binding ligands.

Formation of Fe(III) minerals by microbial Fe(II) oxidation Microbial oxidation of Fe(II) and precipitation of Fe(III) minerals might be better understood by comparing observations from microbial cultures to results from chemical Fe(II) oxidation experiments (e.g., Cornell et al. 1989). Mono- and dinuclear dissolved species of ferrous iron such as [FeOH]2+ and [Fe2(OH2)]4+ are formed initially during abiotic oxidation of Fe(II). Subsequently, these dissolved species transform into polymeric Fe(III) colloids before they precipitate as poorly crystalline ferrihydrite particles with a size of ~2–5 nm in diameter. Depending on the reaction conditions, the initial precipitation might be followed by further transformations of ferrihydrite. Either “solid-state conversion” to hematite (Fe2O3) by internal rearrangement of iron and oxygen atoms is induced or dissolution to low-molecular weight polynuclear iron species occurs which then transform to better crystalline iron oxides such as goethite (“dissolution-reprecipitation mechanism”) (Hansel et al. 2003; Schwertmann and Cornell 2003). Transformation of ferrihydrite to goethite via dissolution-reprecipitation could be facilitated in particular by enhanced proton activities close to cell surfaces. Lowered pH values and transformation of ferrihydrite to goethite were indeed observed in the vicinity of anoxygenic phototrophic Fe(II)-oxidizing bacteria (Kappler and Newman 2004). The formation of crystalline iron oxides during microbial Fe(II) oxidation might accelerate the speed of Fe(II) oxidation by an autocatalytic mechanism. Excess dissolved Fe(II) has a high afinity for surface-OH groups of iron oxides. These surface OH-groups are electron-donor ligands that increase the electron density of the adsorbed ferrous iron. An increased electron density stabilizes +3 charged iron better than +2 charged iron. Therefore, adsorption of Fe(II) on iron oxide surfaces increases the rate of Fe(II) oxidation (Wehrli et al. 1989; Elsner et al. 2003). An electron transfer from surface-adsorbed Fe(II) through the underlying iron oxide to the cell (where electrons could be accepted by outer membrane compounds) would abolish the need for the Fe(II)-oxidizing microbe to be in direct contact with the dissolved Fe(II). The irst evidence for such an electron transfer between adsorbed Fe(II) and Fe(III) from the underlying ferric iron oxide was recently reported by Williams and Scherer (2004). Formation of a variety of different iron minerals by different Fe(II)-oxidizing microorganisms indicates that, apart from medium composition, concentration of possible co-substrates and incubation conditions, the mechanism of Fe(II) oxidation, metabolic rates and the presence of nucleation sites inluence (and maybe even control) the mineralogy of the Fe(III) minerals produced. As an example, in a recent report polysaccharide strands were suggested to be extruded to act as a template for formation of akaganeite pseudo–single crystals (Chan et al. 2003)

MICROBIAL DISSIMILATORY REDUCTION OF Fe(III) Microbial reduction of ferric iron was known as a phenomenon for many decades before its (bio)geochemical relevance was recognized. It was presumed that microorganisms cause reduction of Fe(III) only indirectly, e.g., by lowering the redox potential or the pH. In addition,

96

Kappler & Straub

only few bacteria were known that transferred just few electrons to Fe(III) during fermentative growth (for details see Lovley 1991). This perspective changed notedly with the discovery of bacteria that respire ferric iron and thereby reduce substantial amounts of it (Balashova and Zavarzin 1979; Lovley and Phillips 1988; Myers and Nealson 1988). Today, it is generally accepted that dissimilatory ferric iron-reducing prokaryotes, i.e. organisms that gain energy by coupling the oxidation of organic or inorganic electron donors to the reduction of ferric iron, have a strong inluence on the geochemistry of many environments (e.g., Lovley 1997; Thamdrup 2000).

Acidophilic Fe(III)-reducing microorganisms The ability to reduce Fe(III) to Fe(II) under acidophilic conditions seems to be widespread among acidophilic microorganisms, but the degree of Fe(III) reduction varies signiicantly (Johnson and McGinness 1991). Chemolithotrophic and heterotrophic prokaryotes (bacteria and archaea) are able to couple the reduction of Fe(III) to the conservation of energy. Interestingly enough, acidophilic iron reduction does not require strict anoxia in some strains and proceeds most rapidly even under microoxic conditions (Johnson et al. 1993). Studies with Acidiphilium sp. strain SJH showed that this bacterium is able to reduce a variety of different Fe(III) forms, with the highest reduction rates observed for dissolved Fe(III) (Bridge and Johnson 2000). Barely soluble poorly crystalline iron oxides (e.g., ferrihydrite) were reduced faster than better crystalline iron oxides (e.g., goethite). Apparently, Acidiphilium sp. strain SJH causes dissolution of ferric iron indirectly since direct contact between bacterial cells and solid ferric iron was not necessary for ferric iron reduction to occur. The strain appears to produce an extracellular compound that accelerates Fe(III) dissolution but not reduction. The nature of this extracellular compound and further details of the dissolution process are still unknown (Bridge and Johnson 2000).

Microbial reduction of Fe(III) at neutral pH In the past decade, numerous strains of dissimilatory ferric iron-reducing bacteria and archaea have been isolated from a vast range of habitats. A comprehensive list of Fe(III)reducing microorganisms was recently published by Lovley et al. (2004). The widespread occurrence of Fe(III)-reducing prokaryotes correlates with the ubiquitous presence of ferric iron. Many sediments and soils may contain ferric iron minerals in the range of 50-200 mmol per kg dry matter. Ferric iron is therefore often the dominant electron acceptor although it is barely soluble at neutral pH. According to experimental observations, Fe(III)-reducing microorganisms developed three different strategies to cope with the dificulty of transferring electrons from the cell to the surface of a barely soluble electron acceptor (Fig. 7) (reviewed by Hernandez and Newman 2001; Lovley et al. 2004): A. Physical contact between cell surface/cell surface compounds and ferric iron allows direct delivery of electrons. B. Iron chelators increase the solubility of Fe(III) and hence alleviate Fe(III)reduction. C. Electron-shuttling compounds transfer electrons from the cell to Fe(III) without the necessity of physical contact between cells and ferric iron. Considering the complexity of natural environments and the wealth of microbial capabilities, it is not surprising that different organisms as well as single organisms developed different strategies in order to reduce diverse Fe(III) compounds under varying conditions. For example, some evidence indicates that Shewanella algae and Geothrix fermentans produce and release both Fe(III)-chelators and electron shuttles (Nevin and Lovley 2002a; Lovley et al. 2004). Furthermore, evidence in Geobacter spp. indicates that different cellular compounds are involved in reduction of dissolved Fe(III)-citrate and barely soluble ferrihydrite (Straub

Geomicrobiological Cycling of Iron

97

A Fe(III)-mineral crust on soil particle

Fe(II)

B

Chelator

Chelator

Fe(III)

Shuttlered

Shuttleox

Fe(II)

C

Figure 7. Schematic illustration of different microbial strategies to transfer electrons to ferric iron: (A) Physical contact between cell surface and ferric iron allows direct delivery of electrons; (B) Iron chelators increase solubility of ferric iron; (C) Electron-shuttling molecules transfer electrons to ferric iron. Note: single crystals of ferric iron oxides might be smaller than bacteria (see Fig. 3). However, in nature iron oxides may form crusts on soil particles as depicted here.

and Schink 2004a, Leang et al. 2005). As the methods to study microbial mechanisms of Fe(III) reduction are pivotal, they will be discussed in the next section.

Methods to study mechanisms of microbial Fe(III) reduction Physiological studies of microbial ferric iron reduction at neutral pH are rather dificult. Low solubility of ferric iron is the most prominent obstacle. It impedes the monitoring of cell growth by means of optical density and the separation of cells from iron minerals by simple centrifugation. To circumvent this dificulty, iron chelators (e.g., citrate, EDTA) were applied in many studies to keep iron in solution. However, chelators change the redox potential of ferric iron, may enter the periplasm and can react unspeciically with electron-releasing cellular compounds (reviewed by Straub et al. 2001). In addition, there is growing awareness that culturing microorganisms in rich medium (in particular with other electron acceptors than ferric iron) may cause production of cell compounds which will not be produced under iron-reducing conditions in natural habitats (Glasauer et al. 2003). Caution in the interpretation of results is also necessary when supernatants were prepared either by iltration or centrifugation. Cells of Geobacter spp. were shown to artiicially release compounds (e.g., cytochromes) by iltration with 0.2 μm ilters as well as by centrifugation (Straub and Schink 2003). In other studies, semi-permeable membranes were used to separate cells and iron oxides physically in order to determine whether prokaryotic cells produce Fe(III)-chelators or electron-shuttling molecules. However, it was recently shown that Fe(III)-chelators and electron-shuttling molecules were unable to diffuse freely through dialysis membranes with the largest pore size available (Nevin and Lovley 2000). Therefore, results from studies with semi-permeable membranes need critical assessment, in particular when positive controls with known electron-shuttling molecules are lacking. To minimize artifacts that might be induced by centrifugation or iltration, further methods were developed to study production of Fe(III)-chelators or electron-shuttling

98

Kappler & Straub

molecules in vivo. In a simple one, ferric iron is entrapped in medium solidiied with 1% agar (Straub and Schink 2003). Technically more elaborate is the use of iron containing microporous alginate (Nevin and Lovley 2000) or iron-containing glass beads (Lies et al. 2005).

Microbial mechanisms of Fe(III) reduction at neutral pH For Fe(III) reduction, species of the genus Geobacter appear to require physical contact to ferric iron oxides (Nevin and Lovley 2000; Lovley et al. 2004). The latest study with Geobacter sulfurreducens showed that pili (a special type of cell appendages) were produced during growth with poorly soluble Fe(III), but not with dissolved Fe(III)-citrate as electron acceptor. In addition, experiments with a pilus-deicient mutant implied that those pili were not just required for attachment of cells to ferric iron, and conducting-probe atomic force microscopy indicated that the pili were highly conductive. Together these results suggest that Geobacter sulfurreducens attaches and delivers electrons to the surface of ferric iron oxides via pili (Reguera et al. 2005). Initially it was thought that such a physical contact between Fe(III)-reducing prokaryotes and ferric iron minerals is mandatory for the delivery of electrons from the cells to the minerals. Today, it is generally accepted that Fe(III)-reducing microorganisms also use Fe(III)-chelators or electron-shuttling molecules to reduce barely soluble ferric iron oxides (e.g., Hernandez and Newman 2001; Rosso et al. 2003; Lovley et al. 2004). Diffusible chelators and shuttling molecules help to bridge spatial distance between cells and ferric iron oxides (Fig. 7). This is of particular importance since microorganisms and ferric iron oxides are not evenly distributed in natural environments. Plant root exudates and plant debris can release organic acids which are known to chelate Fe(III), e.g., oxalate or citrate. Accordingly, highly elevated levels of dissolved Fe(III) in the range of 20 to 50 μM were reported for soils in laboratory incubations with rice (Ratering and Schnell 2000). In comparison to the nM range of dissolved Fe(III) at neutral pH (Fig. 2), signiicantly elevated levels of dissolved, presumably chelated Fe(III) in the range of 4 to 16 μM were reported furthermore for freshwater sediment and groundwater samples (Nevin and Lovley 2002b). Plant debris is also the source for phenolic compounds and humic substances which can act as electron-shuttling molecules (e.g., Lovley et al. 1996, 1998). The oxidized form of an electron-shuttling molecule is used as the electron acceptor by the metabolically active cell. The electrons are then transferred from the reduced shuttling molecule in a chemical reaction to ferric iron. It is important that this chemical reaction regenerates the oxidized form of the shuttling molecule (Fig. 7). Prokaryotes that reduce ferric iron oxides only via electronshuttling molecules are not ferric iron-reducing bacteria in a strict sense as their electron acceptor is the shuttling molecule rather than ferric iron. In that respect it is worth mentioning that sulfur-reducing bacteria also can beneit from indirect reduction of ferric iron oxides via sulfur cycling, with sulide as reductant of ferric iron (Straub and Schink 2004b). The impact of prokaryotes that reduce ferric iron oxides only indirectly with the help of naturally occurring electron shuttles on the total Fe(III) reduction in anoxic environments has not yet been evaluated. Finally, it is useful to discuss what advantages, if any are available to iron-reducing microorganisms that speciically produce and excrete Fe(III)-chelating or electron-shuttling molecules. For a single bacterium, production and release of such specialized molecules might be too expensive, in particular if such molecules are lost or degraded before the costs of biosynthesis have been compensated. However, in bacterial communities (e.g., bioilms, cell aggregates) such expenses might be balanced: each cell contributes just few chelator or shuttle molecules and the whole community beneits from the accessibility of ferric iron as electron

Geomicrobiological Cycling of Iron

99

acceptor. To date, no Fe(III)-chelating or electron-shuttling O compound that was speciically produced and excreted in C NH2 ferric iron-reducing cultures in vivo has been identiied. N However, some evidence indicates that Shewanella algae and Geothrix fermentans produce and release both Fe(III)-chelators and electron shuttles (Nevin and Lovley 2002a; Lovley et al. 2004). Furthermore, it was recently N demonstrated that some antibiotics, e.g., phenazine-1Figure 8. Chemical structure of carboxamide (PCN), bleo-mycin and pyocyanine, function phenazine-1-carboxamide, PCN, as electron shuttles between bacteria and Fe(III) minerals a redox-active antibiotic produced (Hernandez et al. 2004). These redox-active antibiotics, by Pseudomonas chlororaphis exempliied in Figure 8 by PCN, structurally resemble that functions as electron shuttle between different microorganhumic substances with regard to aromaticity and redoxisms and ferrihydrite. active functional groups. Pseudomonas chlororaphis can transfer electrons to ferric iron oxides only due to the production and reduction of PCN. In addition to the PCN-producing strain, Shewanella oneidensis, Escherichia coli, Pseudomonas fluorescens, Pseudomonas aeruginosa and Vibrio cholerae were able to reduce PCN and thus indirectly reduce Fe(III) minerals (Hernandez et al. 2004). So far it is unknown whether the antibiotic-producing and/or antibiotic-reducing strains actually gain energy through this indirect Fe(III) reduction. It might just as well be a new microbial mechanism to acquire iron for assimilatory processes (compare to Kraemer et al. 2005). Interestingly enough, appreciable concentrations of phenazines in the range of 27 to 43 ng per g root (with soil) were found in the rhizosphere of wheat plants (Thomashow et al. 1990).

MICROBIAL IRON CYCLING Many reactions relevant to geochemistry are driven and/or accelerated by the activity of prokaryotes. Examples are manifold and include carbon mineralization, nitrogen ixation and sulfate reduction as well as iron transformations. In particular, prokaryotes that gain energy through oxidation of Fe(II) or reduction of Fe(III) have a strong inluence on the global iron cycle (for details see Kraemer et al. 2005). For example, in most acidic aerobic environments, Fe(II) would persist if not oxidized by acidophilic Fe(II) oxidizers.

Microbial iron cycling under acidic conditions Understanding microbial cycling of iron at acidic pH has implications for the leaching of ores and the development of (bio)remediation techniques for acid mine drainage. Evidence for in vivo iron cycling was obtained from mining sites and was investigated in more detail in the laboratory (Johnson et al. 1993). Mixed cultures of acidophilic Fe(II)-oxidizing and Fe(III)reducing microorganisms cycled iron between the oxidation states +II and +III when the concentration of dissolved oxygen luctuated and suficient electron donor for Fe(III) reducers was supplied. Similarly, iron cycling could also be demonstrated in pure cultures since some acidophiles, e.g., Thiobacillus ferrooxidans and Sulfobacillus acidophilus catalyze both Fe(II) oxidation and Fe(III) reduction under appropriate conditions (reviewed by Johnson et al. 1993; Blake and Johnson 2000).

Microbial iron cycling at neutral pH Ferric iron is the dominant electron acceptor for the mineralization of carbon particularly in anoxic freshwater habitats (Thamdrup 2000). Therefore, processes that regenerate Fe(III) minerals that are again available for Fe(III)-reducing prokaryotes have become of signiicant interest. Since microbial Fe(III) reduction and Fe(II) oxidation were recognized, microbial

Kappler & Straub

100

cycling of iron seemed plausible and is hypothesized for many environments. For example, it was estimated that in marine coastal sediments each iron atom cycled approximately 100 to 300 times before being buried in the sediment (Canield et al. 1993). However, the natural complexity of habitats aggravates direct measurements of microbial iron transformation reactions and thus microbial iron cycling in vivo has not yet been clearly demonstrated.

Prerequisites for microbial iron cycling at neutral pH Microbial iron cycling needs iron plus appropriate supplementary substrates, i.e., electron donors for Fe(III) reduction and electron acceptors for Fe(II) oxidation (Fig. 9). Furthermore, the nature of the iron minerals formed is crucial for an eficient cycling since not all iron minerals are equally good substrates. For instance, the redox potential of an iron redox couple determines whether it is available as electron donor or acceptor in terms of energetics (Table 5). At pH 7, molecular oxygen and all redox pairs of the nitrate reduction pathway (Fig. 4) can accept electrons from ferrous iron, independently from the Fe(III) mineral produced. The situation is more complex with ferric iron oxides as electron acceptor. The oxidation of acetate (CO2/acetate, E0c = 290 mV) is energetically favorable just with iron oxides such as lepidocrocite or ferrihydrite. On the other hand, for the reduction of goethite, hematite or magnetite, electron donors with a lower redox potential are necessary, e.g., molecular hydrogen (2H+/H2, E0c = 414 mV) or formate (CO2/formate, E0c = 432 mV). Hence, theoretically acetate can fuel microbial cycling of iron only if ferrihydrite or lepidocrocite is the product of microbial Fe(II) oxidation. Furthermore, it is essential that supplementary electron donors and acceptors can diffuse since ferric iron is barely soluble and thus rather immobile in natural environments. The solubility of Fe(III) in equilibrium with ferrihydrite is in the range of 10−9 M (Fig. 2). The solubilities of goethite and hematite are even lower and the Fe(III) concentrations in the presence of these minerals is in the range from 10−10 M to 10−13 M (Kraemer 2004). In natural environments, the concentration of dissolved Fe(II) is controlled by adsorption or precipitation and is therefore insigniicant in comparison to solid Fe(II). Dissolved Fe(II) adsorbs to soil particles, cell surfaces and also to the surface of ferric iron oxides (e.g., Liu et al. 2001; Cornell and Schwertmann 2003); in model calculations for a coastal sediment, adsorbed Fe(II) exceeded the concentration of freely dissolved Fe(II) 30fold (Van Cappellen and Wang 1996; Thamdrup 2000).

Oxygen-dependent microbial cycling of iron The product of microbial aerobic Fe(II) oxidation is often identiied as poorly crystalline ferrihydrite, a ferric iron oxide that is a favorable electron acceptor for ferric iron-reducing prokaryotes. However, traces of oxygen may repress iron respiration in facultatively anaerobic Fe(III) reducers and can even inhibit the activity of strictly anaerobic ferric iron-reducing

Electron donor,

Fe(III)

e.g. N2

e.g. benzoate

e.g. CO2, H2O

Fe(II)

Electron acceptor, e.g. nitrate

Figure 9. Schematic illustration of microbial iron cycling.

Geomicrobiological Cycling of Iron

101

microorganisms, as shown e.g., for Geobacter spp. (Straub and Schink 2004a). Hence, oxygen-dependent microbial cycling of iron (most likely) always depends on a transition between oxic and anoxic conditions. In natural environments, such transitions are supported by temporary oxygen release by roots, bioturbation by burrowing and boring animals and mixing of sediments by waves or storm events. Of particular interest for oxygen-dependent iron cycling are microaerophilic Fe(II) oxidizers since they thrive in oxic-anoxic transition zones, allowing for microscale microbial redox cycling. Such oxic-anoxic transition zones are characterized by the simultaneous presence of ferrous iron which was produced during anaerobic Fe(III) reduction and of low concentrations of oxygen which reached this zone via diffusion from overlying oxic zones (Sobolev and Roden 2002).

Oxygen-independent microbial cycling of iron The identiication of ferrihydrite as the primary product of anaerobic Fe(II) oxidation by phototrophs (Straub et al. 1999; Kappler and Newman 2004) or nitrate-reducing bacteria (Straub et al. 1996, 1998) indicated the possibility of anaerobic iron cycling. Biologically produced ferrihydrite has been shown to be an excellent electron acceptor for Fe(III)-reducing bacteria, which reduced it completely to the ferrous state (Straub et al. 1998, 2004). Similar to ferric iron, nitrate is used as electron acceptor only in anoxic zones after oxygen is depleted. In contrast to iron, nitrate is soluble at pH 7. Finally, it was feasible to show an anaerobic iron cycling in laboratory co-culture experiments (Straub et al. 2004). For these experiments, Fe(II)-oxidizing nitrate reducers were chosen that were unable to oxidize benzoate. As a counterpart, an Fe(III) reducer was selected that utilized benzoate with Fe(III) but not with nitrate as the electron acceptor. Only in experiments that were inoculated with Fe(II) oxidizers plus Fe(III) reducers was benzoate completely oxidized with nitrate in the presence of iron (Fig. 9). Although the transient iron phases in such co-cultures were not analyzed, stoichiometric considerations suggest that iron cycled 6 times between the oxidation states +II and +III in these experiments (Straub et al. 2004). Clearly, the relevance of anaerobic nitrate-dependent iron cycling for the complex low of electrons in anoxic environments still needs to be determined. Microbial anaerobic iron cycling is possible with the participation of anoxygenic Fe(II)oxidizing phototrophs. Light-dependent, anaerobic cycling of iron may occur in top layers of shallow sediments that are reached by light or in (iron rich) microbial mats.

ENVIRONMENTAL IMPLICATIONS Microbial reduction of Fe(III) and oxidation of Fe(II) may have left geological imprints during Earth‘s history, and continues to signiicantly affect modern environments. Due to the considerable amount of iron in soils and sediments, Fe(III) usually represents the most abundant electron acceptor in anoxic soils and freshwater sediments; only in marine sediments is this dominance counterbalanced by the high sulfate concentration of seawater (Thamdrup 2000). Carbon cycling, mobility of micronutrients and in particular the degradation, transformation and (im)mobilization of organic and inorganic pollutants are closely linked in many environments to the microbial iron cycle.

Degradation of organic compounds coupled to dissimilatory Fe(III) reduction In pristine environments, Fe(III)-reducing microorganisms typically couple the reduction of Fe(III) to the oxidation of H2 or other fermentation products such as simple fatty acids or ethanol. Some ferric iron-reducing strains have in addition the ability to oxidize aromatic, organic pollutants such as benzene, toluene, ethylbenzene, phenol, p-cresol and o-xylene (e.g., Lovley et al. 1989; Lovley and Lonergan 1990; Lovley and Anderson 2000; Jahn et al.

102

Kappler & Straub

2005). If at contaminated sites ferric iron oxides are available for dissimilatory iron-reducing bacteria and other essential nutrients for microbial growth (e.g., nitrogen, phosphorous, sulfur) are present, microbial Fe(III) reduction has the potential to signiicantly contribute to the degradation of aromatic pollutants in a process termed ‘natural attenuation’.

Iron minerals as adsorbents Many ferric iron mineral surfaces are positively charged at neutral pH due to their high points of net zero charge (ZPC). The pH ZPCs are ~7.9 for ferrihydrite, ~8.5 for hematite and ~9.0–9.4 for goethite (Cornell und Schwertmann 2003). Such iron oxides therefore constitute good adsorbents for negatively charged compounds like phosphate (PO43−), bicarbonate (HCO3−) and oxyanions of toxic metal ions such as arsenate (AsO43−), arsenite (AsO33−) or chromate (CrO42−). Furthermore, negatively charged natural organic matter (humic substances) also binds strongly to ferric iron mineral surfaces (Stumm and Morgan 1996; Cornell and Schwertmann 2003). Anions were shown to adsorb to ferrihydrite surfaces via replacement of surface hydroxyl groups, leading to tight bonds of almost covalent character. For weak organic acids also an outer-sphere adsorption via weak electrostatic interactions was observed. Cations usually adsorb to iron oxides via hydroxyl-bridged inner-sphere complexes at the oxide surface. A comprehensive overview on adsorption processes on iron oxides is given by Cornell and Schwertmann (2003). Transformation of iron minerals and pH changes in the environment both inluence the adsorption of cations and anions to ferric iron oxides. In some cases this has dramatic consequences, as in the well-documented example of arsenic: Arsenite and arsenate both strongly bind to ferric iron oxides (Dixit and Hering 2003). There is evidence from extended X-ray absorption ine structure (EXAFS) studies for inner sphere complexation but the nature of the surface complexes is still controversial (e.g., Waychunas et al. 1993; Shermann and Randall 2003) The microbially induced reductive dissolution of arsenic-loaded iron oxides is thought to play a key role in As-release into the groundwater, which leads to enormous drinking water contaminations observed in countries such as Bangladesh and India (Cummings et al. 1999; Smedley and Kinniburgh 2002; Islam et al. 2004; Harvey et al. 2005). In addition, arsenate can be released into the groundwater when high-surface area ferrihydrite transforms into hematite or goethite with signiicantly lower surface areas (Ford 2002).

Immobilization of toxic metal ions by microbial Fe(II) oxidation and Fe(III) reduction Reductive dissolution of metal-loaded iron oxides releases adsorbed metal ions into the environment. In contrast, Fe(II) oxidation can lead to the immobilization of toxic metal ions. Either co-precipitation during Fe(II) oxidation (Gunkel 1986; Richmond et al. 2004) or adsorption to synthetic or natural iron oxides potentially provides an applicable biotechnological method to remove toxic metal ions such as arsenic eficiently from drinking water. Natural removal of arsenic by iron oxides was observed when ferrihydrite was precipitated together with arsenic from arsenic- and iron-rich hydrothermal luids (Pichler and Veizer 1999). In addition to mobilization of adsorbed compounds by dissolution of Fe(III) minerals or immobilization of pollutants by adsorption to or co-precipitation with biogenic iron minerals, iron-metabolizing microorganisms can also have a more direct effect on the fate of pollutants. For example, Fe(III)-reducing microorganisms were shown to convert toxic metal ions from more soluble forms (e.g., Cr(VI) and U(VI)) to less soluble forms that are likely to be immobilized in the subsurface (e.g., Cr(III) and U(IV)) (e.g., Lovley 1993; Lovley and Phillips 1992).

Formation of reactive iron minerals During microbial Fe(III) reduction, different minerals are formed depending on the chemical composition of the medium, on the substrate concentrations and on the incubation

Geomicrobiological Cycling of Iron

103

conditions (e.g., Roden and Zachara 1996; Lovley 1997; Fredrickson et al. 1998; Urrutia et al. 1999; Benner et al. 2002; Zachara et al. 2002; Hansel et al. 2003; Kukkadapu et al. 2004). In particular, the presence of different counter ions leads to the precipitation of different Fe(II) minerals, e.g., iron mono- or disulides (‘FeS’ or FeS2), ferrous iron phosphate (vivianite), carbonate (siderite) or magnetite (Cornell and Schwertmann 2003). Also transformation of ferrihydrite to the more crystalline iron oxides hematite and goethite was observed during Fe(III) reduction (Hansel et al. 2003). Knowing the products of microbial Fe(III) reduction is quite important since the Fe(II)-species formed as the result of microbial Fe(III) reduction (either Fe(II) minerals or mineral-adsorbed and thus activated Fe(II)-species) can be eficient reductants in contrast to free aqueous Fe(II). They were shown to reduce organic contaminants such as nitroaromatic and chlorinated organic compounds (Hofstetter 1999) but also to reduce inorganic compounds such as U(VI) and Cr(VI) (Buerge and Hug 1999; Liger et al. 1999; Lovley and Anderson 2000; Jeon et al. 2005). Because different Fe(II)-species show different reactivities with respect to reductive pollutant transformation, understanding the mechanisms and conditions leading to different Fe(II)-species is necessary (Haderlein and Pecher 1999; Pecher et al. 2002; Elsner et al. 2003).

SOME TASKS FOR FUTURE INVESTIGATIONS Prokaryotes that gain energy from iron redox transformations have a strong inluence on the geochemistry of pristine or polluted environments. This was recognized only in the recent past and still needs to be studied in more detail. In particular the inluence of the microbial cycle of iron on the global cycles of other elements, such as carbon, nitrogen or sulfur is not completely understood. The majority of ferrous iron-oxidizing and ferric ironreducing prokaryotes were isolated during the last decade. Therefore, it is not surprising that our knowledge of these prokaryotes and their iron metabolism is still in its infancy. One major task in microbial physiology is to explain how prokaryotes transfer electrons from or to iron minerals. In this context, the combination of physiology with molecular genetics to track the activity of certain proteins is very promising as recently summarized by Croal et al. (2004) and reviewed by Newman and Gralnick (2005). Anticipated genomic data from isolates (as described by Nelson and Methé 2005) and natural communities (as discussed by Whitaker and Banield 2005) will assist in the identiication of targets. Furthermore, the identiication of Fe(III)-chelating and electron-shuttling molecules intentionally produced and released by prokaryotes is key to understand the biological and ecological importance of these postulated mechanisms. The advancement of different microscopic methods, e.g., cryo transmission electron or environmental scanning electron microscopy, will help to describe intimate interactions between microorganisms and iron minerals. Finally, consequences of microbial iron transformations for the fate of organic and inorganic pollutants have to be explored in more detail to better understand the process of natural attenuation and to foster remediation of polluted sites. An interdisciplinary approach as pursued in the emerging ield of geomicrobiology which comprises such diverse ields as microbial physiology, molecular genetics, geochemistry and mineralogy will certainly help to answer many open questions.

ACKNOWLEDGMENTS Parts of the work for this chapter were done by AK in Dianne Newman’s lab at the California Institute of Technology (Caltech) and by KLS in Bernhard Schink’s lab at the University of Konstanz (Germany). The electron micrographs were taken by AK at Caltech/JPL with the help of M. Chi and R.E. Mielke. We would like to thank B. Schink for reviewing the manuscript. AK is supported by an Emmy-Noether fellowship from the German Research Foundation (DFG).

104

Kappler & Straub REFERENCES

Anbar AD, Holland HD (1992) The photochemistry of manganese and the origin of banded iron formations. Geochim Cosmochim Acta 56:2595-2603 Baker BJ, Banield JF (2003) Microbial communities in acid mine drainage. FEMS Microbiol Ecol 44:139152 Balashova VV, Zavarzin GA (1979) Anaerobic reduction of ferric iron by hydrogen bacteria. Microbiologiia 48:773-778 Benner SG, Hansel CM, Wielinga BW, Barber TM, Fendorf S (2002) Reductive dissolution and biomineralization of iron hydroxide under dynamic low conditions. Environ Sci Technol 36:1705-1711 Benz M, Brune A, Schink B (1998) Anaerobic and aerobic oxidation of ferrous iron at neutral pH by chemoheterotrophic nitrate-reducing bacteria. Arch Microbiol 169:159-165 Blake R II, Johnson DB (2000) Phylogenetic and biochemical diversity among acidophilic bacteria that respire iron. In: Environmental microbe-mineral interactions. Lovley DR (ed) ASM Press, Washington D.C., p 53-78. Bridge TAM, Johnson DB (2000) Reductive dissolution of ferric iron minerals by Acidiphilium SJH. Geomicrobiol J 17:193-206 Buerge IJ, Hug SJ (1999) Inluence of mineral surfaces on chromium (VI) reduction by iron (II). Environ Sci Technol 33:4285-4291 Buresh RJ, Moraghan JT (1976) Chemical reduction of nitrate by ferrous iron. J Environ Qual. 5:320-325 Cairns-Smith AG (1978) Precambrian solution photochemistry, inverse segregation, and banded iron formation. Nature 276:807–808 Canield DE, Thamdrup B, Hansen JW (1993) The anaerobic degradation of organic matter in Danish coastal sediments: iron reduction, manganese reduction, and sulfate reduction. Geochim Cosmochim Acta 57: 3867-3885 Chan CS, De Stasio G, Welch SA, Girasole M, Frazer BH, Nesterova MV, Fakra S, Banield JF (2004) Microbial polysaccharides template assembly of nanocrystal ibers. Science 3003:1656-1658 Chaudhuri SK, Lack JG, Coates JD (2001) Biogenic magnetite formation through anaerobic biooxidation of Fe(II). Appl Environ Microbiol 67:2844-2848 Clayton RK, Sistrom WR (1978) The Photosynthetic Bacteria. Plenum Press, New York Cornell RM, Schneider W, Giovanoli R (1989) The transformation of ferrihydrite into lepidocrocite. Clay Minerals 24:549-553 Cornell RM, Schwertmann U (2003) The Iron Oxides: Structures, Properties, Reactions, Occurrences and Uses. Wiley-VCH, Weinheim Corstjens PLAM, de Vrind JPM, Westbroek P, de Vrind-de Jong EW (1992) Enzymatic iron-oxidation by Leptothrix discophora: identiication of an iron-oxidizing protein. Appl Environ Microbio l58:450-454 Croal LR, Gralnick JA, Malasarn D, Newman DK (2004) The genetics of geochemistry. Ann Rev Genetics 38: 175-202 Cummings DE, Caccavo F, Fendorf S, Rosenzweig, RF (1999) Arsenic mobilization by the dissimilatory Fe(III)-reducing bacterium Shewanella alga BrY. Environ Sci Technol 33:723-729 De Vrind-de Jong EW, Corstjens PLAM, Kempers ES, Westbroek P, de Vrind JPM (1990) Oxidation of manganese and iron by Leptothrix discophora: use of N,N,N’,N’-tetramethyl-p-phenylenediamine as an indicator of metal oxidation. Appl Environ Microbiol 56:3458–3462 Dixit S, Hering JG (2003) Comparison of arsenic(V) and arsenic(III) sorption onto iron oxide minerals: Implications for arsenic mobility. Environ Sci Technol 37:4182-4189 Druschel GK, Baker BJ, Gihring TH, Banield JF (2004) Acid mine drainage biogeochemistry at Iron Mountain, California. Geochem Trans 5:3-32 Edwards KJ, Rogers DR, Wirsen CO, McCollom TM (2003) Isolation and characterization of novel psychrophilic, neutrophilic, Fe-oxidizing, chemolithoautotrophic alpha- and, gamma-Proteobacteria from the deep sea. Appl Environ Microbiol 69:2906-2913 Ehrenreich A, Widdel F (1994) Anaerobic oxidation of ferrous iron by purple bacteria, a new type of phototrophic metabolism. Appl Environ Microbiol 60:4517-4526 Ehrlich HL (2002) Geomicrobiology. Marcel Dekker, New York Elsner M, Haderlein SB, Schwarzenbach RP (2003) Reactivity of Fe(II)-bearing minerals towards reductive transformation of organic contaminants. Environ Sci Technol 38:799-907 Emerson D, Revsbech NP (1994) Investigation of an iron-oxidizing microbial mat community located near Aarhus, Denmark: ield studies. Appl Environ Microbiol 60:4022-4031 Emerson D, Moyer C (1997) Isolation and characterization of novel iron-oxidizing bacteria that grow at circumneutral pH. Appl Environ Microbiol 63:4784-4792 Emerson D (2000) Microbial oxidation of Fe(II) and Mn(II) at circumneutral pH. In: Environmental Microbemineral Interactions. Lovley DR (ed) ASM Press, Washington D.C., p 53-78

Geomicrobiological Cycling of Iron

105

Ford RG (2002) Rates of hydrous ferric oxide crystallization and the inluence on coprecipitated arsenate. Environ Sci Technol 36:2459-2463 Francois LM (1986) Extensive deposition of banded iron formations was possible without photosynthesis. Nature 320:352–354 Fredrickson JK, Zachara JM, Kennedy DW, Li S, Hinman N W (1998) Biogenic iron mineralization accompanying the dissimilatory reduction of hydrous ferric oxide by a groundwater bacterium. Geochim Cosmochim Acta 62:3239-3257 Gilbert B, Banield JF (2005) Molecular-scale processes involving nanoparticulate minerals in biogeochemical systems. Rev Mineral Geochem 59:109-155 Glasauer S, Weidler PG, Langley S, Beveridge TJ (2003) Controls on Fe reduction and mineral formation by a subsurface bacterium. Geochim Cosmochim Acta 67:1277-1288 Gunkel G (1986) Studies of the fate of heavy metals in lakes. I. The role of ferrous iron oxidizing bacteria in coprecipitation of heavy metals. Arch Hydrobiol 105:489-515 Haderlein SB, Pecher K (1999) Pollutant reduction in heterogenous Fe(II)-Fe(III) systems. In: Mineral-Water Interfacial Reactions: Kinetics and Mechanisms. Sparks DL, Grundl TJ (eds) ACS Symposium Series, Washington, D.C., p 342-357 Hafenbradl D, Keller M, Dirmeier R, Rachel R, Roßnagel P, Burggraf S, Huber H, Stetter KO (1996) Ferroglobus placidus gen nov., sp. nov., a novel hyperthermophilic archaeum that oxidizes Fe2+ at neutral pH under anoxic conditions. Arch Microbiol 166:308-314 Hallberg R, Ferris FG (2004) Biomineralization by Gallionella. Geomicrobiol J 21:325-330 Hanert, H.H. (1981) The genus Gallionella. In: The Prokaryotes. Starr MP, Stolp H, Trueper HG, Balows A, Schlegel HG (eds) Springer, Berlin, p 509-515 Hansel CM, Benner SG, Neiss J, Dohnalkova A, Kukkadapu RK, Fendorf S (2003) Secondary mineralization pathways induced by dissimilatory iron reduction of ferrihydrite under advective low. Geochim Cosmochim Acta 67:2977-2992 Hartman H (1984) The evolution of photosynthesis and microbial mats: a speculation on banded iron formations. In: Microbial Mats: Stromatolites. Cohen Y, Castenholz RW, Halvorson HO (eds) Alan Liss Inc., New York, p 451-453 Harvey CF, Swartz CH, Badruzzaman ABM, Keon-Blute N, Yu W, Ali MA, Jay J, Beckie R, Niedan V, Brabander D, Oates PM (2005) Groundwater arsenic contamination on the Ganges Delta: biogeochemistry, hydrology, human perturbations, and human suffering on a large scale. Geosciences 337:285-296 Heising S, Richter L, Ludwig W, Schink B (1999) Chlorobium ferrooxidans sp. nov., a phototrophic green sulfur bacterium that oxidizes ferrous iron in coculture with a ‘Geospirillum’ sp. strain. Arch Microbiol 172:116-124 Hernandez ME, Newman DK (2001) Extracellular electron transfer. Cell Mol Life Sci 58:1562-1571 Hernandez ME, Kappler A, Newman DK (2004) Phenazines and other redox active antibiotics promote microbial mineral reduction. Appl Environ Microbiol 70:921-928 Hofstetter TB, Heijman CG, Haderlein SB, Holliger C, Schwarzenbach RP (1999) Complete reduction of TNT and other (poly)nitroaromatic compounds under iron-reducing subsurface conditions. Environ Sci Technol 33:1479–1487 Islam FS, Gault AG, Boothamn C, Polya DA, Charnock JM, Chatterjee A, Lloyd JR (2004) Role of metalreducing bacteria in arsenic release from Bengal delta sediments. Nature 430:68-71 Jahn MK, Haderlein SB, Meckenstock RU (2005) Anaerobic degradation of benzene, toluene, ethylbenzene, and o-xylene in sediment-free iron-reducing enrichment cultures. Appl Environ Microbiol 71:3355-3358 Jambor JL, Dutrizac JE (1998) Occurrence and constitution of natural and synthetic ferrihydrite, a widespread iron oxyhydroxide. Chem Rev 98:2549-2585 Janney DE, Cowley JM, Buseck PR (2000) Structure of synthetic 2-line ferrihydrite by electron nanodiffraction. Am Mineral 85:1180-1187 Janney DE, Cowley JM, Buseck PR (2001) Structure of synthetic 6-line ferrihydrite by electron nanodiffraction. Am Mineral 86:327-335 Jeon BH, Dempsey BA, Burgos WD, Barnett MO, Roden EE (2005) Chemical reduction of U(VI) by Fe(II) at the solid-water interface using natural and synthetic Fe(III) oxides. Environ Sci Technol 39:5642-5649 Jiao Y, Kappler A, Croal L, Newman DK (2005) Isolation and characterization of a genetically tractable photoautotrophic Fe(II)-oxidizing bacterium Rhodopseudomonas palustris strain TIE1. Appl Environ Microbiol 71:4487-4496 Johnson DB, McGinness S (1991) Ferric iron reduction by acidophilic heterotrophic bacteria. Appl Environ Microbiol 57:207-211 Johnson DB, Ghauri MA, McGinness S (1993) Biogeochemical cycling of iron and sulphur in leaching environments. FEMS Microbiol Rev 11:63-70 Kappler A, Newman DK (2004) Formation of Fe(III)-minerals by Fe(II)-oxidizing photoautotrophic bacteria. Geochim Cosmochim Acta 68:1217-1226

106

Kappler & Straub

Kappler A, Pasquero C, Konhauser KO, Newman DK (2005) Deposition of banded iron formations by photoautotrophic Fe(II)-oxidizing bacteria. Geology. In press. Konhauser KO (1997) Bacterial iron biomineralization in nature. FEMS Microbiol Rev 20:315-326 Konhauser KO, Hamade T, Raiswell R, Morris RC, Ferris FG, Southam G, Canield DE (2002) Could bacteria have formed the Precambrian banded iron formations? Geology 30:1079-1082 Kraemer SM (2004) Iron oxide dissolution and solubility in the presence of siderophores. Aquat Sci 66:3-18 Kraemer SM, Butler A, Borer P, Cervini-Silva J (2005) Siderophores and the dissolution of iron-bearing minerals in marine systems. Rev Mineral Geochem 59:53-84 Kukkadapu RK, Zachara JM, Fredrickson JK, Kennedy DW (2004) Biotransformation of two-line silicaferrihydrite by a dissimilatory Fe(III)-reducing bacterium: formation of carbonate green rust in the presence of phosphate. Geochim Cosmochim Acta 68:2799-2814 Lack JG, Chaudhuri SK, Chakraborty R, Achenbach LA, Coates JD (2002) Anaerobic biooxidation of Fe(II) by Dechlorosoma suillum. Microb Ecol 43:424-431 Leang C, Adams LA, Chin KJ, Nevin KP, Methe BA, Webster J, Sharma ML, Lovley DR (2005) Adaptation to disruption of the electron transfer pathway for Fe(III) reduction in Geobacter sulfurreducens. J Bacteriol 187: 5918-5926 Lennie AR, Vaughan DJ (1996) Spectroscopic studies of iron sulide formation and phase relations at low temperatures. In: Mineral Spectroscopy: A Tribute to Roger G. Burns. Dyar MD, McCammon C, Schaefer MW (eds) The Geochemical Society, St. Louis, p 117-131 Lies DP, Hernandez ME, Kappler A, Mielke RE, Gralnick JA, Newman DK (2005) Shewanella oneidensis strain MR-1 uses overlapping pathways for iron reduction at a distance and by direct contact under conditions relevant for bioilms. Appl Environ Microbiol 71:4414-4426 Liger E, Charlet L, Van Capellen P (1999) Surface catalysis of uranium(VI) reduction by iron(II). Geochim Cosmochim Acta 63:2939-2955 Liu CX, Kota S, Zachara JM, Fredrickson JK, Brinkman CK (2001) Kinetic analysis of the bacterial reduction of goethite. Environ Sci Technol 35:2482-2490 Lovley DR, Phillips EJP (1988) Novel mode of microbial energy metabolism: organic carbon oxidation coupled to dissimilatory reduction of iron or manganese. Appl Environ Microbiol 54:1472-1480 Lovely DR, Baedecker MJ, Lonergan DJ, Cozzarelli JM, Phillips EJP, Siegel DJ (1989) Oxidation of aromatic contaminants coupled to microbial iron reduction. Nature 339:297-300 Lovley DR, Lonergan DJ (1990) Anaerobic oxidation of toluene, phenol, and p-cresol by the dissimilatory ironreducing organism, GS-15. Appl Environ Microbiol 56:1856-1864 Lovley DR (1991) Dissimilatory Fe(III) and Mn(IV) reduction. Microbiol Rev 55:259–287 Lovley DR, Phillips EJP (1992) Reduction of uranium by Desulfovibrio desulfuricans Appl Environ Microbiol 58:850–856 Lovley DR (1993) Dissimilatory metal reduction. Annu Rev Microbiol 47:263-290 Lovley DR, Coates JD, Blunt-Harris EL, Phillips EJP, Woodward JC (1996) Humic substances as electron acceptors for microbial respiration. Nature 382:445-448 Lovley DR (1997) Microbial Fe(III) reduction in subsurface environments. FEMS Microbiol Rev 20:305-315 Lovley DR, Fraga JL, Blunt-Harris EL, Hayes, LA, Phillips EJP, Coates JD (1998) Humic substances as a mediator for microbially catalyzed metal reduction. Acta Hydrochim Hydrobiol 26:152-157 Lovley DR, Anderson RT (2000) The inluence of dissimilatory metal reduction on the fate of organic and metal contaminants in the subsurface. J Hydrol 238:77-88 Lovley DR, Holmes DE, Nevin KP (2004) Dissimilatory Fe(III) and Mn(IV) reduction. Adv Microb Phys 49: 219-286 Manceau A, Drits VA (1993) Local structure of ferrihydrite and feroxyhite by EXAFS spectroscopy. Clay Minerals 28:165-184 Moraghan JT, Buresh RJ (1977) Chemical reduction of nitrite and nitrous oxide by ferrous iron. Soil Sci Soc Am J 41:47-50 Myers CR, Nealson KH (1988) Microbial reduction of manganese oxides: interactions with iron and sulfur. Geochim Cosmochim Acta 52:2727-2732 Nelson KE, Methé B (2005) Metabolism and genomics: adventures derived from complete genome sequencing. Rev Mineral Geochem 59:279-294 Newman DK, Gralnick JA (2005) What genetics offers geobiology. Rev Mineral Geochem 59:9-26 Nevin KP, Lovley DR (2000) Lack of production of electron-shuttling compounds or solubilizaiton of Fe(III) during reduction of insoluble Fe(III) oxide by Geobacter metallireducens. Appl Environ Microbiol 66: 2248-2251 Nevin KP, Lovley DR (2002a) Mechanisms for accessing insoluble Fe(III) oxide during dissimilatory Fe(III) reduction by Geothrix fermentans. Appl Environ Microbiol 68:2294-2299 Nevin KP, Lovley DR (2002b) Mechanisms for Fe(III) oxide reduction in sedimentary environments. Geomicrobiol J 19:141-159

Geomicrobiological Cycling of Iron

107

Nordstrom DK, Southam G (1997) Geomicrobiology of sulide mineral oxidation. Rev Mineral 35:361-390 Pecher K, Haderlein SB, Schwarzenbach RP (2002) Reduction of polyhalogenated methanes by surface bound Fe(II) in aqueous suspensions of iron oxides. Environ Sci Technol 36:1734-1741 Pichler T, Veizer J (1999) Precipitation of Fe(III) oxyhydroxide deposits from shallow-water hydrothermal luids in Tutum Bay, Ambitle Island, Papua New Guinea. Chem Geol 162:15-31 Ratering S, Schnell S (2000) Localization of iron-reducing activity in paddy soil by proile studies. Biogeochemistry 48:341-365 Reguera G, McCarthy KD, Mehta T, Nicoll JS, Tuominen MT, Lovley DR (2005) Extracellular electron transfer via microbial nanowires. Nature 435:1098-1101 Richmond WR, Loan M, Morton J, Parkinson GM (2004) Arsenic removal from aqueous solution via ferrihydrite crystallization control. Environ Sci Technol 38:2368-2372 Roden EE, Zachara JM (1996) Microbial reduction of crystalline iron(III) oxides: inluence of oxide surface area and potential for cell growth. Environ Sci Technol 30:1618-1628 Roden EE (2003) Fe(III) oxide reactivity toward biological versus chemical reduction. Environ Sci Technol 37: 1319-1324 Roden EE, Sobolev D, Glazer B, Luther GW III (2004) Potential for microscale bacterial Fe redox cycling at the aerobic-anaerobic interface. Geomicrobiol J 21:379-391 Rosso KM, Zachara JM, Fredrickson JK, Gorby YA, Smith SC (2003) Nonlocal bacterial electron transfer to hematite surfaces. Geochim Cosmochim Acta 67:1081-1087 Shermann DM, Randall SR (2003) Surface complexation of arsenic(V) to iron(III) (hydr)oxides: structural mechanism from ab initio molecular geometries and EXAFS spectroscopy. Geochim Cosmochim Acta 67:4223-4230 Sobolev D, Roden EE (2002) Evidence for rapid microscale bacterial redox cycling of iron in circumneutral environments. Anton van Leeuw 181:587-597 Sobolev D, Roden EE (2004) Characterization of a neutrophilic, chemolithoautotrophic Fe(II)-oxidizing βProteobacterium from freshwater wetland sediments. Geomicrobiol J 21:1-10 Smedley PL, Kinniburgh DG (2002) A review of the source, behaviour and distribution of arsenic in natural waters. Appl Geochem 17:517-568 Straub KL, Benz M, Schink B, Widdel F (1996) Anaerobic, nitrate-dependent microbial oxidation of ferrous iron. Appl Environ Microbiol 62:1458-1460 Straub KL, Buchholz-Cleven BEE (1998) Enumeration and detection of anaerobic ferrous iron-oxidizing, nitrate-reducing bacteria from diverse European sediments. Appl Environ Microbiol 64:4846-4856 Straub KL, Hanzlik M, Buchholz-Cleven BEE (1998) The use of biologically produced ferrihydrite for the isolation of novel iron-reducing bacteria. System Appl Microbiol 21:442-449 Straub KL, Rainey FA, Widdel F (1999) Rhodovulum iodosum sp. nov. and Rhodovulum robiginosum sp. nov., two new marine phototrophic ferrous-iron-oxidizing purple bacteria. Int J Syst Bacteriol 49:729-735 Straub KL, Benz M, Schink B (2001). Iron metabolism in anoxic environments at near neutral pH. FEMS Microbiol Ecol 34:181-186 Straub KL, Schink B (2003) Evaluation of electron-shuttling compounds in microbial ferric iron reduction. FEMS Microbiol Lett 220:229-233 Straub KL, Schink B (2004a) Ferrihydrite reduction by Geobacter species is stimulated by secondary bacteria. Arch Microbiol 182:175-181 Straub KL, Schink B (2004b) Ferrihydrite-dependent growth of Sulfurospirillum deleyianum by electron transfer via sulfur cycling. Appl Environ Microbiol 70:5744-5749 Straub KL, Schönhuber WA, Buchholz-Cleven BEE, Schink B (2004) Diversity of ferrous iron-oxidizing, nitrate-reducing bacteria and their involvement in oxygen-independent iron cycling. Geomicrobiol J 21: 371-378 Stumm W, Morgan JJ (1996) Aquatic Chemistry: Chemical Equilibria and Rates in Natural Waters. John Wiley & Sons, New York Thamdrup B (2000) Bacterial manganese and iron reduction in aquatic sediments. In: Advances in microbial ecology. Schink B (ed) Kluwer Academic/Plenum Publishers, New York, p 41-84 Thauer RK, Jungermann K, Decker K (1977) Energy conservation in chemotrophic anaerobic bacteria. Bacteriol Rev 41:100-180 Thomashow LS, Weller DM, Bonsall RF, Pierson LS (1990) Production of the antibiotic phenazine-1carboxylic acid by luorescent pseudomonas species in the rhizosphere of wheat. Appl Environ Microbiol 56:908-912 Urrutia MM, Roden EE, Zachara JM (1999) Inluence of aqueous and solid-phase Fe(II) complexants on microbial reduction of crystalline iron(III) oxides. Environ Sci Technol 33:4022-4028 Van Cappellen P, Wang YF (1996) Cycling of iron and manganese in surface sediments – A general theory for the coupled transport and reaction of carbon, oxygen, nitrogen, sulfur, iron, and manganese. Am J Sci 296:197-243

108

Kappler & Straub

Waychunas GA, Rea BA, Fuller CC, Davis JA (1993) Surface chemistry of ferrihydrite: part 1. EXAFS studies of the geometry of coprecipitated and adsorbed arsenate. Geochim Cosmochim Acta 57:2251-2269 Wehrli B, Sulzberger B, Stumm W (1989) Redox processes catalyzed by hydrous oxide surfaces. Chem Geol 78:167-179 Whitaker RJ, Banield JF (2005) Population dynamics through the lens of extreme environments. Rev Mineral Geochem 59:259-277 Widdel F, Schnell S, Heising S, Ehrenreich A, Assmus B, Schink B (1993) Ferrous iron oxidation by anoxygenic phototrophic bacteria. Nature 362:834-835 Williams AGB, Scherer MM (2004) Spectroscopic evidence for Fe(II)-Fe(III) electron transfer at the iron oxide-water interface. Environ Sci Technol 38:4782-4790 Zachara JM, Kukkadapu RK, Fredrickson JK, Gorby YA, Smith SC (2002) Biomineralization of poorly crystalline Fe(III) oxides by dissimilatory metal reducing bacteria (DMRB). Geomicrobiol J 19:179-207

Reviews in Mineralogy & Geochemistry Vol. 59, pp. 109-155, 2005 Copyright © Mineralogical Society of America

6

Molecular-Scale Processes Involving Nanoparticulate Minerals in Biogeochemical Systems Benjamin Gilbert Earth Sciences Division Lawrence Berkeley National Laboratory 1 Cyclotron Road MS 90R1116 Berkeley, California, 94720, U.S.A. [email protected]

Jillian F. Banfield Earth and Planetary Sciences University of California Berkeley Berkeley, California, 94720-4767, U.S.A. [email protected]

INTRODUCTION Mineral particles with diameters on the scale of nanometers (nanoparticles) are important constituents of natural environments. The small size of such particles has a host of consequences for biogeochemical systems, which we will review in this chapter. We begin by briely reviewing what is known about how and when nanoparticles form and the ways in which nanoparticles impact natural processes. Nanoparticles form via a variety of inorganic and biological pathways and may be introduced into the environment as a consequence of human activity. They are widespread in the environment (Banield and Navrotsky 2001; Penn et al. 2001; Kennedy et al. 2003a,b; van der Zee et al. 2003), although few quantitative studies of their abundance are available. While all crystals begin as very small particles, an important subset retain small size at the Earth’s surface over relatively long time scales, because the combination of low temperature and low solubility inhibits growth. As a consequence, nanoparticles have the potential for a long lifetime in the environment, and widespread transport under certain circumstances. Processes that result in the removal of nanoparticles from an environment include dissolution, settling from air, transport in solution, and crystal growth. Particle aggregation may be an important component of these processes because it will promote settling, limit dispersal via solution transport, and can lead to aggregation-based crystal growth. The presence of nanoparticles can profoundly inluence biological systems. Because they are frequently formed in environments that are populated by microorganisms, nanoparticles often adhere to cell surfaces or cell-associated polymers (see Fig. 1 for examples). These coatings can have important consequences for metabolic activity, for example, by restricting communication between the cell and its surroundings. They may also provide protection from predators, inhibit desiccation, screen cells from ultraviolet radiation, and alter the cell buoyancy (e.g., Tebo et al. 1997; Phoenix et al. 2001). The controlled precipitation of nanoscale minerals can lead to formation of integrated organic-inorganic structures with diverse morphologies. In some cases, the shapes provide a record of past microbial activity (i.e., serve as biosignatures, possibly by preserving cell 1529-6466/05/0059-0006$05.00

DOI: 10.2138/rmg.2005.59.6

110

Gilbert & Banęeld

Figure 1. Microorganisms can produce copious quantities of nanoparticles. a). Field-emission scanning electron microscopy (SEM) micrograph of spherical aggregates of ZnS nanoparticles (with trace amounts of Fe, As, and Se) in a sulfate-reducing bacteria dominated bioilm growing on wood in neutralized acidmine drainage (Moreau et al. 2003). Draped over the aggregates is a dehydrated ilamentous microbial cell of a morphology commonly observed in ZnS-dense regions within the bioilm. Image courtesy of John Moreau. b). High-resolution SEM image of a fractured sulfate reducing bacterium encrusted in aggregates of compositionally mixed ZnS and FeS nanoparticles (Williams et al. 2005). Inset: Higher magniication view of biomineral coating around a fracture cell. Images courtesy of Ken Williams. c). SEM micrograph of iron oxyhydroxide nanoparticle coatings on microbial structures from bioilms from the Piquette Mine, Tennyson, WI. The twisted stalks and cylindrical sheaths are characteristic products of Gallionella ferruginea (an iron oxidizer) and Leptothrix spp. (a putative iron oxidizer), respectively. Image courtesy of Susan Welch and Clara Chan. d). Transmission electron micrograph of aggregated UO2 nanoparticles adhering to the surfaces of sulfate reducing bacteria taken from the Midnite Mine, Washington, USA (Suzuki et al. 2002). Image courtesy of Yohey Suzuki.

morphology, as seen in the iron oxyhydroxide nanoparticle coatings on extracellular sheaths of Leptothrix, Fig. 1). Biominerals constructed from nanoparticles can be generated through deliberate action of the organism in order to create cell architecture (Mann 2000), and magnetic nanoparticles can serve a role in navigation (Bazylinski and Frankel 2004). However, in many cases the function is unclear or the particles may serve no function, and their existence is purely a byproduct of microbial metabolism (e.g., ZnS and UO2 nanoparticles, Fig. 1, and Mn(IV) oxides). Binding of nanoparticles to cell surfaces can also affect the fate and distribution of nanoparticles in the environment, by restricting or facilitating their transport, aggregationbased growth, and mineral transformation pathways. The interactions between nanoparticles, individual cells, and extracellular biomolecules can also act to bind bioilms together (Mayers and Beveridge 1989).

Molecular-Scale Processes Involving Nanoparticulate Minerals

111

Sources of nanoparticles in the environment A major pathway for nanoparticle formation in aqueous environments is the precipitation of sparingly soluble dissolved ions derived from both inorganic and biological processes. Precipitation is possible when the concentrations of ions in solution exceed the solubility of a mineral (e.g., see De Yoreo and Vekilov 2003). However, formation of a crystal nucleus is inhibited by an energy penalty associated with creation of a solid-liquid interface. Consequently, ion concentrations must generally exceed the saturation state of the solution for precipitation to occur. The relative energy cost associated with this interface is reduced as the particles grow, providing a driving force for coarsening. Materials that form ultra-small particles are frequently very insoluble with low energy barriers for nucleation. Examples include UO2, with a solubility product for bulk material logKsp ≈ −60, which forms 1–3 nm diameter particles (Suzuki et al. 2002; O’Loughlin et al. 2003); and ZnS, with a solubility product of logKsp ≈ −24, which typically forms 2–3 nm diameter particles (Labrenz et al. 2001; Moreau et al. 2004). In fact, a wide range of both natural and synthetic nanoparticles initially forms in this size range. Inorganic sources of environmental nanoparticles. A variety of inorganic pathways can lead to nanoparticle formation. Chemical weathering reactions that occur when minerals in rocks are exposed to water and air at the Earth’s surface liberate ions that can reprecipitate as nanometer-scale silicate clay minerals, oxides, oxyhydroxides, and phosphates. Common examples include smectite (e.g., montmorillonite (Na,Ca)(Al,Mg)6(Si4O10)3(OH)6˜nH2O), anatase (TiO2), hematite (D-Fe2O3), and rhabdophane (CePO4˜H2O). Aqueous clusters of metal sulides and aluminum oxide are found in some lake and marine environments (Rozan et al. 2000; Casey and Swaddle 2003). Nanoparticulate minerals may nucleate at sites on another mineral surface (Stack et al. 2004). Additional inorganic sources of nanoparticles include impacts, combustion, vaporization (e.g., breakdown of meteorites as they enter the atmosphere), evaporation of sea spray, erosion, faulting, and other mechanical processes. Biological sources of environmental nanoparticles. Microbial activity is a major source of nanoparticles in the environment, and a summary of biogenic minerals is given by Frankel and Bazylinksi (2003). Central to microbial metabolism is the process of energy generation, which harnesses the free energy liberated as the result of coupled oxidation and reduction reactions. Reactions used for energy generation must be thermodynamically favorable, but are often kinetically inhibited. Organisms utilize enzymes to overcome these barriers and may speed up geochemical reactions by several orders of magnitude. In many cases, minerals can serve as either reductant or oxidant in microbial metabolism. In the presence of reduced organic carbon, oxygen or nitrate often serve as the electron acceptor for microbial carbon respiration. However, electrons can be passed to ferric iron, Mn(III)/IV, or sulfate when oxygen and nitrate are not available (Banield and Nealson 1997; Konhauser 1998; Edwards et al. 2000; Ehrlich 2002). Uranium ions can also be biologically reduced. In either metabolic or co-metabolic processes (Lovely et al. 1991; Abdelouas et al. 2000; Fredrickson et al. 2000), electrons are passed from organic carbon to aqueous uranyl (UO22+) ions, resulting in the formation of insoluble uraninite (UO2) nanoparticles. Ferric iron and manganese minerals that act as electron acceptors are typically ine grained oxides that can dissolve upon reduction (Roden and Zachara 1996; Lovely 1997; Bratina et al. 1998; Quantin et al. 2001). In contrast, for sulfate reduction, aqueous sulfate ions are converted to HS−, which, in the presence of a suitable counter-ion, will precipitate as sulide nanoparticles. Because reduction of sulfate is quantitatively linked to the oxidation of carbon, these metabolic pathways can generate copious quantities of nanoparticulate sulides over short time periods (see Fig. 1a,b).

112

Gilbert & Banęeld

In the absence of organic carbon for respiration and light for photosynthesis, both carbon ixation and energy generation can be driven by energy harvested from inorganic chemical reactions alone (Lovely 1993). In such chemoautotrophically based ecosystems, bacteria such as Thiobacillus ferrooxidans oxidize sulide and/or ferrous iron that is present in pyrite (FeS2) (Fowler et al. 1999). The resulting ferric iron and sulfate ultimately precipitate, often as nanoparticulate iron oxyhydroxide (ferrihydrite or goethite, D-FeOOH) or iron sulfate compounds (schwertmannite, Fe16O16(OH)y(SO4)z˜nH2O, and others). Anthropogenic nanoparticles. Industrial manufacturing and combustion processes can lead to nanoparticle emission into the air or into wastewater efluent. For example, numerous groups are seeking to manufacture stable water-soluble magnetic iron oxide nanoparticles as an injectable diagnostic agent for enhancing contrast in magnetic resonance imaging (Tartaj et al. 2003). Cerium dioxide nanoparticles are of interest as catalysts during combustion processes (e.g., in automotive engines), because of the capacity of the CeO2 lattice to buffer changes in oxygen partial pressures and oxidize carbon monoxide (Bekyarova et al. 1998). These nanoparticles may be released during manufacturing, use, or as the result of product disposal, and could ultimately accumulate in the environment in high enough concentrations to have important ecosystem impacts, such as accumulation in and toxicity for aquatic organisms (Oberdörster et al. 2005).

Impacts of nanoparticles on their surroundings The formation of nanoparticles can inluence the local chemical environment in which organisms live. Nanoparticles sequester ions when they precipitate, and can decrease or increase the porosity and permeability of sediments. For example, formation of sulide nanoparticles removes both metal ions (e.g., Cu, Zn, As, Cd, Fe) and toxic sulide ions from solution, improving habitability of the environment. Nanoparticle precipitation and dissolution reactions can be sources or sinks for protons, and thus can inluence environmental mineralogy through the impact of pH on mineral solubility (Langmuir 1996). For example, the precipitation of iron oxyhydroxide nanoparticles and dissolution of sulide nanoparticles both lead to environmental acidiication, promoting dissolution of surrounding minerals. The solubilization of manganese oxides may liberate adsorbed trace metals that provide nutrients in oligotrophic environments (Bratina et al. 1998). Nanoparticles, with their high surface areas, can play especially important roles in adsorption. For example, nanoparticles of iron oxyhydroxide formed during the neutralization of acidic, ferrous iron-rich solutions can sorb phosphate ions, possibly limiting biological productivity in some environments. Nanoparticle surfaces can also sequester protons or toxic ions such as arsenate (Waychunas et al. 2005).

Nanoparticles—special properties and implications Recently, a great deal of research has shown that nanoscale inorganic solids may exhibit substantially modiied properties relative to their bulk counterparts (Alivisatos 1996; Murray et al. 2000; Trindade et al. 2001), with consequences for the inorganic and biological interactions of nanoparticles in the environment. As a foundation for studies of coupled geochemical processes involving nanoparticles, we examine some principles describing how size inluences nanoparticle properties and reactivity. The present chapter introduces the physical and chemical consequences of small particle size in minerals, and discusses the effect small particle size has on redox and photochemical reactions. The concepts introduced here can be used for understanding the environmental impact and fate of both natural and engineered nanoscale materials.

Molecular-Scale Processes Involving Nanoparticulate Minerals

113

Overview of small size effects in minerals While molecular geochemistry has always been a “nano-science,” the science of nanoscale clusters and nanoparticles is distinct in an important way. The motivating tenet of contemporary nanoscience is that the chemical and physical properties of a solid inorganic material may vary as a function of particle dimensions below a critical size. The deinition of the critical size depends both upon the material under consideration and the property of interest. It may refer to the start of a size dependent trend, such the onset of electronic coninement, or to an abrupt change, such as a switch in the thermodynamically stable structure of a mineral. We stress that, in almost all cases, the description of the unmodiied (i.e., bulk) material provides an excellent guide to the properties of a nanosized material (even below a critical size). Thus, knowledge of the relevant solid state physics and chemistry for a given mineral is an indispensable foundation for understanding small-size effects. We emphasize, however, that important small size effects are not limited to modiications of nanoparticle structure and properties alone, which we denote as static effects. Static small-size effects in mineral particles include the stabilization of structural phases that are metastable for bulk materials, and the presence elevated strain and disorder, particularly at nanoparticle surfaces. Electrons within solids respond to shifts in the equilibrium positions of atoms, surface-solvent interactions, and the presence of a conining surface itself. The latter effect—quantum coninement—is striking in some materials, but (when present) is neither the only, nor always the dominant, size effect in mineral nanoparticles. A large number of surface effects (e.g., coninement, surface charge, solvent interactions, and surface reconstruction) modulate electronic structure. The kinetics of charge, energy, and material transfer also change at the nanoscale. Geochemical processes are dynamic, and signiicant size-dependent changes in reaction kinetics may affect processes in natural systems. The impact of nanoparticles on biogeochemical processes can depend on the kinetics of competing pathways. For example, the ability of photoexcited electrons in nanoparticles to reduce biomolecules is governed by the rates of several size-dependent processes that may enhance reactivity or dissipate energy without reaction. Furthermore, the products of various (photo)reduction experiments can be different for nano- versus micron-sized particles, indicating a size effect on reaction pathways (Müller et al. 1997). We start by discussing small-size effects in individual particles. However, nanoparticles are frequently observed to be intimately aggregated, with signiicant consequences for their behavior in biogeochemical systems.

PHYSICAL STRUCTURE AND COMPOSITION OF NANOSCALE MINERALS Thermodynamic constraints on the structure of nanoparticles One of the better-understood ways that size can inluence the structure of nanomaterials is through interfacial energy effects on phase stability. This topic was reviewed in detail by Banield and Zhang (2002) and Navrotsky (2002). In brief, when a compound can exist in more than one structural form (polymorph), a change in the relative stabilities of the structural variants may occur as the consequence of differences in their surface energies. This effect is only likely at small particle sizes, for which surface areas are large. For example, sizedependent phase stability may explain the precipitation of the hexagonal polymorph of ZnS (wurtzite) in place of the cubic form (sphalerite) in sediments or aqueous solutions, despite the

114

Gilbert & Banęeld

higher stability of sphalerite compared to wurtzite in bulk materials up to temperatures over 1000 °C (Scott and Barnes 1972; Qadri et al. 1999; Zhang 2003). However, it should be noted that kinetic effects may also lead to the production of metastable phases (amorphous or highly disordered materials), especially in low-temperature systems where precipitation rates are fast (Schwertmann et al. 1999; Wolthers et al. 2003). It can be dificult to assess whether an observed nanophase is thermodynamically stable or metastable. It can also be dificult to evaluate the structure and natural abundance of amorphous or highly disordered materials.

The nature of the initial precipitates and subsequent aging For many minerals formed at low temperature or by biological processes, the initial precipitates are reported to be amorphous or highly disordered, possibly hydrated nanoparticles. Examples include amorphous iron sulides, hydrous ferric oxide (HFO), and layered Mn(IV) oxides. These initial materials themselves do not grow beyond the nanoscale, and instead undergo crystallization, dehydration, or other transformations to mineral phases that may subsequently grow. It is widely thought that extracellular bacterial surfaces provide sites for heterogeneous nucleation that lower the free energy barrier for precipitation (Fortin et al. 1997; Warren and Ferris 1998). This model is widely used to describe the precipitation of ferric iron as a result of the activity of iron oxidizing bacteria (Douglas and Beveridge 1998; Ferris 2005). The model is challenged, however, by a recent study of Rancourt et al. (2005), on the precipitation of HFO in the presence of nonmetabolizing bacteria. They showed that while Fe(III) ions adsorb onto functional groups on bacterial surfaces, they remain external to the structure of HFO particles that subsequently form. Rancourt et al. argue convincingly that both inorganic and biological HFO formation always occurs via fast homogeneous precipitation. There is also uncertainty as to the role of biological factors in affecting transformations that occur in nanoparticles after precipitation. While aqueous conditions (solution chemistry, pH, temperature), rather than their inorganic or biological origins, should govern the evolution of mineral precipitates (Konhauser 1998), comparative studies of the structure and reactivity of biominerals versus synthetic analogs frequently reveal differences. The presence of organic matter is thought to be a crucial factor (Ferris 2005). Furthermore, microbial metabolism frequently generates aqueous ions (e.g., Fe(II), or small organic molecules) that can interact with and stabilize or destabilize mineral surfaces during growth, phase transformation, and dissolution (Cornell and Schwertmann 1979; Urrutia et al. 1999; Davis et al. 2000; Thomas et al. 2004). One interesting biological effect is the observation that the association of 2-line ferrihydrite nanoparticles with bacterial cell walls confers signiicant stability to the mineral against hydrothermal coarsening and transformation into hematite (Kennedy et al. 2004). The stabilization effect is observed with both biogenic and abiotic ferrihydrite in the presence of bacteria. It is inferred that the nanoparticles principally grow by an aggregation-based pathway that is hindered when the particles are immobilized on cell walls. Orientated aggregation (OA) is a signiicant pathway for nanoparticle growth in which individual particles achieve a common crystallographic orientation (Penn and Banield 1998). Subsequent elimination of the interfaces between oriented particles can generate larger single crystals, some of which may have unusual morphologies and properties (Banield et al. 2000). An example of joined UO2 nanoparticles is given in Figure 4.

Size dependence of mineral solubility From the point of view of biogeochemical systems, one of the most important sizedependent materials properties is solubility. As can be seen from the following equation

Molecular-Scale Processes Involving Nanoparticulate Minerals

115

(Stumm and Morgan 1996), solubility increases as particle size decreases because the interfacial energy,J in J m−2, is a positive contribution to the total energy. ’ log K sp = log K spBULK +

2 3γ S 2.3RT

(1)

In this equation, S is the surface area per mole, R = 8.314 J mol−1 K−1 and T is the temperature in K. For ZnS, this equation predicts that the solubility of 3 nm particles of wurtzite is an order of magnitude higher than for bulk material at room temperature, assuming an effective surface area of ~100 m2/g, log KspBULK = −22.85 (Stumm and Morgan 1996) and J | 0.5 J m−2 (Zhang et al. 2003). Sphalerite nanoparticles of the same dimensions are predicted to be two orders of magnitude more soluble, assuming log KspBULK = −24.83and J | 0.8 J m−2. The accuracy of Equation (1) for quantitative predictions is limited. First, it assumes that interfacial energy is not itself size dependent, an assumption that has been questioned previously (Zhang et al. 1999). Second, accurate measurements of the interfacial energy are frequently unavailable, especially for hydrated surfaces that may additionally be coated by organic molecules. More realistic descriptions of the solubility of natural particles as a function of size, phase, and the surface chemical environment are needed.

Characterization studies of biogenic nanoparticles We briely review recent studies of the structure of important biogenic minerals. Precise structural characterization of biogenic nanoscale materials is challenging due to the small particle size, presence of disorder, and dificulties in isolating mineral precipitates from biomass. Powder X-ray diffraction (XRD) can be a convenient approach for determining nanoparticle structure, but has limitations. First, the width of peaks in diffraction patterns depends inversely upon the number of unit cells of material that make up the diffracting atomic planes. When the number of unit cells is very small (as in nanoparticles), peaks become extremely broad, obscuring structural details and the resulting diffraction pattern resembles those from materials with only short-range order. Given these considerations, structure analysis based upon Bragg’s law can be inaccurate at the smallest particle sizes (Palosz et al. 2002). Direct simulation of diffraction patterns from small structural models can be a more accurate method (Drits and Tchoubar 1990). Furthermore, when nanoscale particles possess signiicant disorder, XRD-based methods of particle size determination (e.g., via Scherer’s equation; Patterson 1939) are extremely inaccurate (Wolthers et al. 2000). Pair distribution function (PDF) analysis of short-range order in nanoparticles is a promising extension of the powder XRD method (Gilbert et al. 2004; Billinge and Thorpe 1998). Nanoparticle imaging and structure analysis may be performed in the electron microscope by high-resolution transmission electron microscopy (HRTEM) and selected area electron diffraction (SAED). Bright, energy-tunable X-ray sources (synchrotrons) offer additional methods of structural and chemical analysis, including extended X-ray absorption ine structure (EXAFS), and X-ray absorption near-edge structure (XANES) spectroscopies. XANES spectra can be acquired with lateral resolution in scanning transmission X-ray microscopy (STXM) and X-ray photoelectron emission microscopy (X-PEEM). In addition, laboratory-based Mössbauer spectroscopy provides information on the chemical environment of the 57Fe nucleus. Iron oxides and oxyhydroxides. Biological and inorganic processes can lead to the precipitation of hydrous ferric oxides, as discussed above. Subsequent mineral transformations can produce nanocrystalline iron oxides including goethite or hematite (Benner et al. 2002; Hansel et al. 2003). Many environmental factors affect the mineral evolution. For example, low-temperature aging in the presence of water favors goethite formation; hematite is preferentially formed in warmer, dry environments (Cornell and Schwertmann 2003).

116

Gilbert & Banęeld

Magnetite (Fe3O4) is a common and mineralogically signiicant nanocrystalline iron biomineral product of the biological reduction of ferric iron. Dissimulatory iron reducing bacteria (DFeRB) utilize Fe(III) as an electron acceptor for the reduction of organic carbon (Lovely 1987; Bazylinski and Moskowitz 1997). HFO and crystalline ferric iron phases can provide a bioavailable source of ferric iron (Roden and Zachara 1996). Fe(II) released from the mineral surface can re-adsorb onto unreacted HFO, forming mixed-valence green rusts that age into nanoparticulate and poorly crystalline magnetite (Lovley et al. 1987; Sparks et al. 1990; Hansel et al. 2003). Alternative Figure 2. Thin hydrated surface layers can be ferrous iron-bearing minerals may be nanosized products of microbial metabolism. A formed, depending upon the aqueous low thin (< 1 nm) layer of hydrated magnetite (Fe3O4) rate, the relative concentration of HFO forms on the surface of colloidal goethite particles and the organic electron donor, and the as a result of dissimilatory iron reducing bacterial presence of additional solution species metabolism. [Used by permission of Elsevier from Hansel et al. (2004) Geochimica et Cosmochimica such as phosphate and carbonate, and Acta, Vol. 68, Fig. 7e, p 3217-3229.] ferrous iron complexants (Urrutia et al. 1999; Roden et al. 2000; Fredrickson et al. 2003; Hansel et al. 2003). The rate of ferric iron reduction by Shewanella putrefaciens is correlated more with the surface area of the ferric iron substrate than the solubility of the particular ferric iron phase (Roden and Zachara 1996). However, differences in the solubility of crystalline (goethite and hematite) versus disordered (HFO) materials does affect the quantity and form of magnetite produced. Magnetite formation during the bioreduction of goethite and hematite is apparently limited to approximately 1 nm thick rinds on the initial ferric iron particles (Hansel et al. 2004), as shown in Figure 2. Similar mineral transformations are produced by Fe(II) adsorption onto synthetic iron oxides (Tronc et al. 1984, 1992). Manganese oxides. Numerous microorganisms can enzymatically or indirectly facilitate the oxidation of aqueous Mn(II) by O2 at far higher rates than inorganic pathways, although the biological function(s) of this activity is still unclear (Tebo et al. 1997, 2004). Two recent studies of fresh biogenic material (from the bacterium Pseudomonas putida and from spores of the marine Bacillus sp. strain SG-1) reached similar conclusions on the structure of the irst-formed product (Villalobos et al. 2003; Bargar et al. 2005). The material is a hexagonal phyllomanganate assembled from stacked layers of Mn(IV) octahedra with considerable rotational stacking disorder (turbostratic disorder), plus a high negative structural charge due to Mn(IV) vacancies. (For a complete introduction to the complex crystal chemistry of manganese oxide materials, see Burns and Burns 1979, Villalobos et al. 2003, and Tebo et al. 2004). These freshly formed biogenic minerals contained very little Mn(III). By contrast, aged biogenic oxides (Bargar et al. 2005) and Mn(IV) oxides found in soils were observed to have vacancies or Mn(IV) substitution by Mn(III) or heterovalent cations (Fig. 3b) (Isaure et al. 2005; Manceau et al. 2005). Mn(III) incorporation follows the autocatalytic oxidation of surface-sorbed Mn(II), which is an important factor in abiotic transformation of the biogenic mineral. The precipitation and aging of biogenic Mn oxides is depicted in Figure 3a. Sulfides. Transition metal sulide minerals are common products of the metabolism of sulfate-reducing bacteria. In the case of iron sulides, the irst formed material is thought to be

Molecular-Scale Processes Involving Nanoparticulate Minerals

Figure 3. a). Scheme of the biogenic oxidation of Mn(II), the formation of disordered MnO2 and subsequent aging to layered Mn(IV) and Mn(III) mineral products in the presence of aqueous Mn2+. [Reprinted from Bargar et al. 2005.] b). Layered Mn(IV) oxides found in soils are commonly found to host Mn(III) in octahedral coordination and can incorporate Zn in either tetrahedral or octahedral coordination. The presence of lower valence Mn or other elements will have the potential to introduce electrons into the semiconducting MnO2 sheet. [Used by permission of Elsevier from Isaure et al. (2005) Geochimica et Cosmochimica Acta, Vol. 69, Fig. 7f, p. 1173-1198.]

117

amorphous, but further growth and mineral transformation leads to the production of numerous crystalline phases (Benning et al. 2000; Wolthers et al. 2003). Observed biogenic iron sulide minerals include mackinawite (tetragonal FeS), pyrite (cubic FeS2), marcasite (orthorhombic FeS2) greigite (Fe3S4) and pyrrhotite (Fe7S8). The exact mineral product is highly dependent on solution chemistry (Benning et al. 2000). It has been proposed that the sorption of ferrous iron to functional groups on cell surface membranes provides sites for heterogeneous nucleation, but as discussed above, sorption and precipitation processes may be independent. Iron sulide minerals including FeS and FeS2 are generally found to be sulfur deicient (Luck et al. 1989), but it is not known whether this is enhanced or suppressed in nanoscale particles. Disordered biogenic phases concentrated by magnetic separation were reported to have a high capacity for cation sequestration (Watson et al. 2000). Disordered nanocrystalline mackinawite exhibits an expanded lattice that incorporates water molecules, and possibly hydroxyl groups, cation impurity atoms, and sulfur vacancies (Wolthers et al. 2003).

Even in the presence of ferrous iron, alternative minerals may precipitate if thermodynamically favored. Labrentz et al. (2000) observed that highly pure ZnS nanoparticles are formed in close proximity to sources of dissolved ferrous iron. A common feature of biogenic ZnS nanoparticles is the presence of stacking faults and wurtzite nanoparticles, probably relecting the low free-energy difference between the sphalerite and wurtzite phases (Figure 4, and see Moreau et al. 2004). Interestingly, recent studies on synthetic mixed-phase ZnS nanoparticles have indicated a signiicant effect on the proile and size of the semiconducting band gap (H. Zhang, personal communication). Other minerals. The reduction of U(VI) in the aqueous uranyl ion (UO22−) to form insoluble uraninite (UO2) nanoparticles can take place as a by-product of microbial metabolism involving sulfate (Fig. 4; Suzuki et al. 2002) or iron reduction (Lovely et al. 1991; Fredrickson et al. 2000). Uraninite nanoparticles can also form abiotically following the reduction of uranyl by ferrous iron (O’Loughlin et al. 2003). Suzuki et al. (2002) used HRTEM to document the presence of biogenic UO2 nanoparticles as small as 1 nm in diameter, and EXAFS spectroscopy to infer that the average particle size is ~1.5 nm. Given the presence of 2–4 nm particles, this result suggests that ultra-small particles or molecular clusters are abundant. The U-O bond length exhibited a 0.007 nm contraction

118

Gilbert & Banęeld

Figure 4. Left. High resolution transmission electron microscope (HRTEM) image of biogenic ZnS nanoparticles formed by sulfate reducing bacteria (SRB). The upper nanoparticle has the wurtzite ZnS structure. The lower ZnS nanoparticle contains a mixed sphalerite-wurtzite stacking sequence. [Reprinted from Moreau et al. (2003).] Right. HRTEM of UO2 nanoparticles formed by SRB. Two joined UO2 nanoparticles show a characteristic necking proile indicating that oriented aggregation is an active growth pathway (top left). Image courtesy of Yohey Suzuki.

relative to bulk UO2, indicating the presence of signiicant surface strain, and an associated increase in solubility. By estimating a relationship between surface stress and interfacial energy, and assuming negligible size-dependent changes in compressibility, this observation was used to predict a billion-fold increase in the solubility of biogenic uraninite relative to the bulk mineral. The uncertainties in the assumptions required for this calculation emphasize the need for additional studies on nanoparticle structure, elastic properties, and solubility.

The effects of water and other surface-bound molecules on nanoparticle structure As the medium in which minerals form, water itself is a key player at every stage in the precipitation and evolution of nanoparticles at and near the Earth’s surface. Rapid precipitation pathways frequently lead to structural incorporation of water molecules. While the interactions between water and bulk mineral surfaces have been an area of intense study (Hochella and White 1990; Henderson 2002), there are relatively few structural or calorimetric studies on solvent interactions with nanoparticle surfaces (but see Navrotsky 2004). Nevertheless, it has been shown that nanoparticle surface interactions with water can be strong and decisive in stabilizing particular mineral structures. Diffraction studies in controlled environments have shown that surface hydration is an important factor in the surface and interior structure of ZnS and J-Fe2O3 nanoparticles (Zhang et al. 2003; Belin et al. 2004). In these examples, changes in surface hydration caused dynamic structural responses in the nanoparticles. Zhang et al. (2003) demonstrated that adsorption of water drove a structural transformation in ZnS nanoparticles at room temperature without any change in particle size. The waterdriven reaction was not reversible. However, the desorption and re-adsorption of a more volatile solvent (methanol) does cause reversible structural changes, and it is inferred that the activation barriers for structural rearrangements in these nanoparticles are small enough that they can be overcome by surface interactions at room temperature. The indings from the study of ZnS nanoparticles do not imply that natural nanoparticles will always ind their energy minimum state, given their size and surface chemical environment.

Molecular-Scale Processes Involving Nanoparticulate Minerals

119

But they do emphasize that the dynamic response of the atoms in a nanoscale particle will permit this under certain circumstances. Other groups have sought a description of the phase stability of mineral nanoparticles, based upon surface interactions in the framework of equilibrium thermodynamics. It was recently argued that, because of the high surface area of nanoparticles, the adsorption of hydroxyl groups or oxyanions onto HFO nanoparticles represents a signiicant change in the stoichiometry, one that can affect the thermodynamic stability of the nanoparticle-sorbate system (Fukushi and Sato 2005). Vayssiéres et al. (1998) interpreted the pH dependence of the phases of iron oxide nanoparticles synthesized with aqueous methods to indicate that the enthalpy of surface adsorption is a governing thermodynamic contribution. Lodziana et al. (2004) have used molecular simulations to describe the very high stability of the hydroxylated T-Al2O3 (110) surface as a thermodynamically stabilized (i.e., negative interfacial energy) material. However, such characterizations remain controversial, because in the absence of clearly reversible transitions it is dificult to assess whether the observed material is the true thermodynamically stable state. The observations that surface ligands can direct the structure of a nanomaterial may be directly relevant to low-temperature biogeochemical systems. In these environments, small mineral particles are closely associated with solute molecules including small organic compounds, and high molecular weight polymers including proteins and polysaccharides. These compounds may represent a biological inluence on the structure of nanoparticles in aqueous systems. In general, molecular simulations that have incorporated realistic surfacesolvent or surface-ligand interactions have revealed strong interactions that stabilize both the surface and interior structure of nanoparticles (Pokrant and Whaley 1999; Rabani 2001; Zhang et al. 2002; Kerisit et al. 2005).

Incorporation of impurity atoms Contaminants (e.g., Zn, As) and nutrients (e.g., phosphate) are readily adsorbed on, coprecipitated with, and incorporated into mineral nanoparticles (Watson et al. 2000; Gunnars et al. 2002; Hochella et al. 2005a,b) and (photo)redox cycles of nanoscale iron and manganese oxides and oxyhydroxides can cause the incorporation and release of these species (Isaure et al. 2005). Since it is widely observed that the availability and transport of aqueous ions is highly correlated to their interactions with colloidal particles (Kimball et al. 1995; Brown et al. 1999), important aspects of nutrient and contaminant biogeochemistry depend on their association with mineral nanoparticles. Furthermore, the reactivity of the nanoparticles themselves can be strongly modiied by the incorporation of even low concentrations of impurity atoms. For example, the incorporation of 0.2 mol% Zn into pyrite drastically affects surface photochemical behavior, with no detectable structural modiications (Büker et al. 1999). As explained below, impurity atoms can introduce additional electronic energy levels that can affect mineral reactivity. An understanding of the factors controlling impurity atom incorporation in nanoparticles remains incomplete, both in the materials and earth sciences. As with bulk materials, the ability of nanoparticles to host impurities depends on the structural characteristics of the solid, the size and charge discrepancy between the impurity and intrinsic ion, and the response of the structure to the structural perturbation associated with ion incorporation (Cornell and Schwertmann 2003; Fistul 2004). Careful combined X-ray absorption, X-ray diffraction and structure modeling studies are required to elucidate the crystal chemistry of incorporated atoms in disordered minerals (Isaure et al. 2005; Manceau et al. 2005) Defects such as ion substitutions or vacancies have an associated enthalpy from which their mean concentration can be calculated, assuming thermodynamic equilibrium. However,

120

Gilbert & Banęeld

distributing the same number of defects in nanoparticles rather than bulk material would result in the vast majority of nanoparticles being defect-free. Furthermore, there may be an impurity exclusion effect, as solid-state diffusion times to the surface are likely to be quite short (Tang et al 2003). Consequently, the presence of detectable impurity concentrations in nanoparticles is likely to indicate favorable kinetic pathways, rather than thermodynamic equilibrium. An important advance in modeling and predicting impurity incorporation is the recent proposal that it is the afinity of dissolved impurity ions for the surface of a growing nanoparticle that is the key factor determining eventual incorporation (Erwin et al. 2005; commentary by Galli 2005). The excess energy of the impurity once it is in the host lattice is not the dominant factor. Rather, it is the rate of impurity adsorption and desorption at speciic surface sites relative to nanocrystal growth kinetics. The implication is that the fast precipitation reactions that follow microbially mediated changes in the oxidation state of environmental ions such as iron, manganese and sulfur might be effective at scavenging reactive aqueous species. There is considerable empirical evidence for signiicant impurity concentrations in natural nanoparticles. For example, disordered Mn(IV) and Fe(III) oxides and oxyhydroxides are commonly observed to account for the majority of transported contaminant ions (Hochella et al. 2005a,b). Certain impurity atoms can affect the phase stability or enhance the crystallinity of biominerals (Davis et al. 2000; Webb et al. 2005). It is of interest to determine whether the miscibility of substances that form solid solutions is affected by small particle size. For example, bulk ferric iron oxides can accommodate homovalent cation substitution (e.g., Cr3+ and V3+) to 5–10 mol% (Schwertmann et al. 1989; Schwertmann and Pfab 1994), and bulk ZnxFe1−xS forms a solid solution for all values of x (Vaughan and Craig 1978). There are presently no investigations into the stability of these materials as nanoparticles. Most effort has been directed to the production of technological materials. In particular, the introduction of Fe3+ and Cr2+ into TiO2 nanoparticles creates photocatalysts with higher yields (Zhang et al. 1998; Bryan et al. 2004), and the doping of Mn2+ into ZnS nanoparticles enhances their luminescence eficiency (Yu et al. 1996).

The surfaces of nanoscale minerals The surfaces of materials have many distinct structural and electronic properties relative to the bulk or the interior. Surfaces host, mediate, or participate in all relevant biogeochemical processes involving minerals. Decades of surface science provided considerable knowledge about mineral surface structure and surface chemical interactions. However, all such work has been performed on the surfaces of bulk minerals, and it can be argued that the surface science of nanoscale minerals is just beginning. It might be reasonable to expect, a priori, that the constituents of nanoparticle surfaces will include facets and defects (such as edges or vertices) very similar to those found on bulk mineral surfaces, perhaps with a few additional types. However, the results of certain experimental and theoretical studies indicate that this model may be seriously misleading. Zinc sulide nanoparticles in the 1–5 nm diameter size range are observed to possess signiicant interior distortion, which is inferred to be provoked by extreme reconstruction at the surface (Gilbert et al. 2004). This model is supported by molecular dynamics calculations showing highly distorted surfaces that, because of high surface curvature, bear little resemblance to facets of bulk minerals (see Fig. 5). Large-scale molecular simulation of goethite nanoparticles within a dissociating model of water, and with dimensions below 10 nm, accumulate protons at highly charged edge sites to a far greater extent than is anticipated from models with simple slab geometries (Fig. 5) (Rustad and Felmy 2005). Nanoparticle surfaces may additionally be stoichiometrically distinct from the interior, as has been determined recently for the surfaces of bulk hematite (Glenn Waychunas, personal communication).

Molecular-Scale Processes Involving Nanoparticulate Minerals

121

Figure 5. Molecular modeling studies predict that the surfaces of mineral nanoparticles will exhibit local structural and charge that are distinct from the surfaces found on bulk materials. Left. A cross-section through a molecular dynamics (MD) simulation of the structure of a fully hydrated ~ 3 nm diameter ZnS nanoparticle. No recognizable facets are present on the high curvature surface. Zinc (sulfur) atoms are black (gray); water molecules are white. MD simulation (unpublished) performed by H. Zhang. Right. MD simulations of hydrated goethite nanocrystals predict high protonation along edges at the high angle intersections of (110) faces. The inhomogeneous surface charge distribution could not have been predicted from experimental or theoretical studies of bulk surfaces, and can dominate the effective charge of nanoparticles with certain morphologies. The high edge protonation is favored by ready solvation at high angle edges (high dielectric sites) and by the local high bacisity of edge Fe-O sites (sites labeled 1–3). [Used by permission of Elsevier from Rustad and Felmy (2005) Geochimica et Cosmochimica Acta, Vol. 69, Fig. 6, p. 1405-1411.]

Also, partial transformation caused by surface oxidation or reduction can lead to the formation of maghemite (J-Fe2O3) layers on magnetite surfaces, and magnetite ilms on HFO or hematite (cf. Fig. 2) (Hansel et al. 2004). Studies of the mechanism and geometry of adsorbate binding to nanoparticle surfaces are one approach to compare the surfaces of bulk and nanocrystalline minerals. Mercury, arsenic, and copper are known to attach to the surface of goethite (D-FeOOH) via inner sphere coordination and thus are suitable test sorbates for goethite nanoparticles. The coordination environment of Hg adsorbed to 5 nm diameter goethite nanoparticles is modiied relative to sorption sites on larger particles, with an expansion of the 2nd and 3rd shell He-O and Hg-Fe distances (Waychunas et al. 2005). However, no signiicant size-dependent changes in the sorption geometry of As(V) or Cu(II) were detected. At present, there is no general framework for describing nanoparticle surfaces, and both practical and theoretical approaches require considerable development. Modern X-ray methods that probe the structure of water and the chemistry at mineral surfaces must be adapted to the surfaces of nanoparticles (Eng et al. 2000; Cheng et al. 2001).

ELECTRONIC STRUCTURE OF NANOSCALE MINERALS Introduction to electronic structure of solids Electronic structure is key to the reactivity of materials. The electrochemical potential of electrons in solids drives chemical reactions that underpin many of the Earth’s biogeochemical

122

Gilbert & Banęeld

cycles. Thus, we begin this section with an introduction to the electronic structure of solids in a form that will permit size-dependent effects and molecular scale reactivity to be addressed. The quantum theory of solid-state materials gives the best quantitative description of electrons in solids and is presented in numerous texts (Ashcroft and Mermin 1976). It is briely recapped here with an emphasis on the analogies between bonding in molecules and in periodic crystals. Approximate descriptions of electrons in solids. The enormous number of atoms and electrons in solids renders exact solutions of electronic and nuclear motion utterly intractable, and requires drastic simplifying assumptions. The following are two of the most important. First, the nuclear motion is considerably slower than electronic motion, and hence the nuclei are assumed to be static during calculations of electronic structure. Second, the behavior of only a single electron is considered, with an averaged description of the inluences of all other electrons and nuclei. Within the framework of this single-electron approximation, Schrödinger’s Equation can be written:

[ Ek + V (r )] ψi (r ) = Eiψi (r )

(2)

Equation (2) relates the total energy, Ei, and wavefunction, \i(r), of an electron in a state i to its kinetic energy, Ek, and to the potential, V(r), within which it moves. The potential energy term principally contains the electrostatic potential of each atomic site, and deines the energy landscape in which the electron moves. An important additional contribution to V(r) describes (approximately) the correlated interactions with all other electrons. Collectively, the energy terms are called the Hamiltonian, H, which completely deines the possible electronic states for a given system. The wavefunction is the most complete description attainable for a quantum particle and represents the probability of inding the particle within an ininitesimal volume of space. The simplest model of size effects on electronic structure relies on the description of an electron traveling freely in space. For a given electron momentum, described using the wave vector k, the energy of a free electron is given by E=

=2 k 2 2 me

(3)

where me is the mass of the electron and = = h/2S, where h is Planck’s constant. For numerous materials, the electrons that participate in chemical reactions can be considered as free electrons, but with a modiied effective mass, me*. The effective mass captures the effect of the periodic structural environment on the propagation of wave-like electrons within the material. Assuming that the mobility of an electron in a material does not vary with particle size, me* is obtained from a irst-principles calculation of the electronic structure of a bulk material (i.e., by obtaining the solutions of Eqn. 2). Solutions of Schrödinger’s equation. There are two important concepts that permit effective solution of the single electron Schrödinger’s equation for bulk materials. First is the realization that the solution wavefunctions (or orbitals) possess the same symmetry as the spatial distribution of atoms with which they are associated. Atoms (and hence the wavefunctions of atomic electrons) possess spherical symmetry. Orbitals that are centered on atoms in a molecule possess the point symmetry of that site; and, as bulk crystals ill space with identical unit cells, the electronic wavefunctions of periodic crystals possess translational symmetry. The last statement is known as Bloch’s theorem, and provided the foundation for an explosive growth in solid-state physics during the 20th century. The second principle is that a wavefunction can be expressed as a linear combination of any set of basis functions that are mathematically complete. This is

Molecular-Scale Processes Involving Nanoparticulate Minerals

123

exactly analogous to the use of Fourier series to represent an arbitrary function. An example of this approach is molecular orbital theory, in which a linear combination of atomic orbitals is used to describe new electronic states formed when atoms bond. It permits the eficient numerical simulation of electronic structures, because realistic electronic wavefunctions can be built up as a combination of computationally convenient basis functions. As wavefunctions are (in principle) independent of the choice of basis functions, there is an equivalence between choices. This can be seen by comparing the results of real-space cluster calculations and momentum-space band-structure calculations for the same mineral. Band structure calculations give similar results for the energy positions of electronic levels (bands) in solids that can be derived from cluster calculations, but additionally show how the energy-momentum relation varies with direction of travel in a crystal. An example for sphalerite, the cubic modiication of ZnS, is given in Figure 6. The bulk electronic structure of a material is the starting point for understanding the properties and reactivity of nanoparticles of that material.

Energy levels in semiconductor minerals An important focus of this chapter is the role that nanoparticles play in redox reactions in biogeochemical systems. It is necessary, therefore, to introduce the concepts used to describe the behavior of electrons in bulk and nanosized systems. With a few exceptions, biogenic minerals are semiconductors or insulators.

Figure 6. Small cluster (ZnS46−) molecular orbital and full band structure calculations provide approximately equivalent descriptions of electronic bonding in ZnS. The band structure shows how the momentum-energy relation for a delocalized electron traveling within the crystal is modulated by the periodic lattice for various directions of travel. The wave vector, k, is related to the momentum and the symbols on the abscissa axis represent speciic directions with respect to the crystal lattice. Integrating over all directions at each electron energy gives the density of states (DOS), i.e., the number of electronic states within a small energy width. Cluster calculations are less accurate than band structure calculations for delocalized electron semiconductors, but correctly show the atomic contributions to the electronic bands. The band gap, which separates the occupied and unoccupied electronic levels is shaded gray. ZnS is a direct semiconductor: no change in momentum is required for electronic transitions from the top of the valence band (VB) to the bottom of the conduction band (CB). The Fermi level, EF lies between the VB and CB. Effective Mass Approximation (EMA) considers only the electronic states at the top of the VB and the bottom of the CB, as indicated by the dashed oval. The band structure is approximately parabolic in this region (E v k2), indicating that the electrons and holes can be modeled as free particles with an effective mass, m*. Cluster energy levels after Vaughan and Sherman (1980). Band structure and DOS calculations used by permission of the American Physical Society, from Wang and Klein (1981) Physical Review B, Vol. 24, Fig. 5, p. 3393-3416.

124

Gilbert & Banęeld

The valence and conduction bands and the band gap. The valence band (VB) and conduction band (CB) in solids are the exact analogs of the highest-occupied molecular orbitals (HOMO) and lowest-unoccupied molecular orbitals (LUMO) in molecules. The Fermi energy, EF, can be considered the electronic electrochemical potential within the material. For insulators and semiconductors, EF lies between the (completely occupied) VB and (empty) CB. When a valence electron is excited (by thermal energy or by light absorption) and acquires suficient energy to jump to the conduction band, a charge deicit, or hole, is created in the VB that acts just like a mobile positive charge. The electronic band gap is the energy required to excite an electron into the conduction band and move it out of the electrostatic ield associated with the hole that is created. The spatial scale that deines this distance is the Bohr radius, the radius of the lowest-energy bound state of an electron-hole pair that are mutually attracted because of their opposite electrostatic charges. Such a bound pair is called an exciton. The conductivity of minerals. Electronic conductivity requires mobile charge carriers in bands that are not completely occupied. Energy bands in solids are occupied with electrons up to the Fermi level. Hence, at zero temperature, metallic behavior is expected in a pure solid if the Fermi level lies within a band, while insulating behavior is expected if it lies between bands. At higher temperatures, and for suficiently small band gaps, thermal energy can excite electrons in the vicinity of the Fermi level. The promotion of a small number of electrons to the conduction band can enable conduction via both CB electrons and the associated VB holes. Pure materials that conduct by this mechanism are called intrinsic semiconductors. For some crystal structures, electronic transitions between the top of the VB and the bottom of the CB require exchange of momentum between the electron and vibrations of the lattice (phonons). Materials in which excitation across the band gap is phonon-assisted are called indirect semiconductors. Solids in which bonding arises from the interactions of s or p atomic levels follow the above description closely. However, other materials are anticipated to be metallic according to the above criteria, but are measured to be very poor conductors (Cox 1995). This is because when bonding involves partially occupied d or f levels, two additional factors can limit electron and hole mobility. First, low overlap between atomic orbitals of neighboring atoms can lead to the localization of outer shell electrons on a single atomic site. Second, energetic barriers may exist that limit charge transfer between neighbors. The origins of such barriers may be associated with electron repulsion or correlation energies, or may be structural. There is a continuum from highly localized to highly itinerant electronic behavior within which many environmentally relevant transition metal oxides and chalcogenides fall. In numerous iron-bearing materials, magnetic ordering of electrons below a threshold temperature can abruptly change the electronic properties from those of an itinerant semiconductor to localized insulator. Even for some bulk minerals such as hematite (Fujimori et al. 1986) and magnetite (Park et al. 1997; Todo et al. 2001), controversy remains with regard to the nature of the VB and CB electronic states. These distinctions, and controversies, persist in nanoscale particles. An additional important class of materials are the extrinsic semiconductors. Impurity atoms or point defects can provide electronic states within the band gap of an insulator or intrinsic semiconductor. The energy required to excite electrons or holes to or from these states can be signiicantly lower than transitions across the full band gap. Consequently, even very low concentrations of impurities can dominate the conductivity of bulk minerals. The charge carriers introduced into a solid by impurities or defects can be itinerant or localized just as the carriers introduced from atoms intrinsic to the material. Consequently, the incorporation of impurities during nanoparticle nucleation and growth can have a signiicant effect on reactivity.

Molecular-Scale Processes Involving Nanoparticulate Minerals

125

Many of the partly occupied d-shell conigurations found in iron- and manganese-bearing compounds have associated magnetic properties. A review of nanoscale magnetism is given by Rancourt (2002). Electron hopping conduction mechanisms. Localized electrons or holes experience an energy barrier against transfer to neighboring atomic sites, but at inite temperatures the thermal energy of the system may be suficient to overcome the barrier. Thus, localized charge carriers can be mobile through a “hopping” mechanism from site to site. An important example is that of the small polaron, an electron or hole that is transiently trapped at each site because of the distortion of the immediate lattice that it provokes, and which follows it. As with any process with an activation energy barrier, mobility increases with increasing temperature. A theoretical treatment of hopping conductivity (Cox 1995; Rosso et al. 2003) originated from a model electron-transfer reactions between ions in solution (Marcus 1993; Barbara et al. 1996), and is also applicable to charge and exciton transfer between nanoparticles (Adams et al. 2003; see below). Quantum mechanical tunneling is an alternative mechanism that may dominate for acceptor-donor distances less than 14 Å, and is utilized by many proteins and electron shuttles (Moser et al. 1992; Page et al. 1999). Tunneling processes do not pass through an intermediate higher energy state and are therefore distinct from the hopping mechanism. Consider the transfer of an electron from a donor (D) to an acceptor (A). The donor could be a reductant in solution or a site in a crystal that has trapped an electron. D + A l D+ + A−

(4)

The rate of electron transfer, ket, has an Arrhenius-type dependence: ⎛ ΔG * ⎞ ket = κν exp ⎜ − ⎟ ⎝ kBT ⎠

(5)

where 'G* is the Gibbs free energy of an intermediate state (the activation energy), kB is Boltzmann’s constant, and N is a prefactor for a given system. Thermal motions of the atomic nuclei bring the system into the most favorable coniguration for charge transfer with a certain frequency, given by Q. In the intermediate state, the electron overlaps with both the donor and acceptor. The major contribution to the activation energy is the reorganization energy associated with the geometry of the intermediate state. The activation energy also includes electrostatic interactions between the electron and the atoms or molecules that coordinate the donor and acceptor, such as near-neighbor atoms in a crystal or hydrating solvent molecules. Hopping-based conductivity in minerals is generally highly anisotropic, as determined by the crystal structure and electron spin ordering. For example, conduction in hematite occurs predominantly along the (001) basal planes (Rosso et al. 2003). Conduction in magnetite involves the octahedral sites on which Fe(II) and Fe(III) ions are distributed to the exclusion of purely Fe(III) tetrahedral sites (Cox 1995). As motion of the atoms is required for the system to reach the intermediate state, modiications in the vibrational properties of nanoparticles at small size (e.g., Gilbert et al. 2004) may signiicantly affect charge (or impurity) transport.

Electronic structure of nanoparticles Pure size-dependent modiications of electronic structure substantially depend on the extent to which the valence electrons are delocalized. When the dimensions of a semiconductor are similar to the Bohr radius, an electron and hole can never be so far apart that the interaction between them can be neglected. Consequently, the lowest energy excitation is not equivalent to the bulk band gap, as deined above, but contains additional terms, principally the (positive) coninement energy and the (negative) electron-hole Coulomb interaction. The coninement energy dominates and the nanocrystal exhibits an increase (or blue-shift) in its band gap. Band

126

Gilbert & Banęeld

gap opening is readily observed with ultra violet–visible (UV-vis) absorption spectroscopy, but the individual shifts of the VB and CB are not obtained with this approach. Combined Xray absorption and X-ray emission spectroscopies can resolve the size dependent trends in the energy positions of the occupied and unoccupied bands (Fig. 7). The electronic structure of bulk materials with delocalized electrons is well modeled by band structure methods (using translation symmetry). It is obvious, however, that even the most crystalline nanoparticle lacks long-range periodicity. To date, two approaches have been adopted to deal with this. Effective mass approximation (EMA). Virtually all geochemically relevant optical and redox behavior of materials involves the charge carriers in the vicinity of the top of the VB or the bottom of the CB (region of interest circled in Fig. 6). The effective mass approximation (EMA) is an approach that describes how a perturbation in a material (such as inite size) affects these electrons and holes, disregarding all others. It assumes that the mobility of

Figure 7. Quantum coninement causes shifts in the absolute energy positions of both the valence band maximum (VBM) and the conduction band minimum (CBM). Left. Soft X-ray emission (SXE) and soft X-ray absorption (SXA) spectroscopy provide information on, respectively, the density of occupied and unoccupied electronic states of a material. Spectroscopy at the sulfur L-edge provides the density of states (DOS) with d-character. The d-weighted DOS ine structure and position varies with size for CdS nanoclusters. Right. The energy shifts in the VBM and CBM are plotted versus particle size. The experimental results are compared to theoretical calculations using the ininite potential (IP) vs. inite potential (FP) effective mass approximation (EMA) and tight-binding real-space cluster calculations (TB). Ultraviolet absorption spectroscopy provides only the energy spacing between VBM and CBM. [Used by permission of Elsevier, from Lüning et al. (1999) Solid State Communications, Vol. 112, Figs. 1 & 2, p. 5-9.]

Molecular-Scale Processes Involving Nanoparticulate Minerals

127

electrons and holes (represented by the effect mass, and determined by the crystal lattice) is unchanged in nanoparticles. The size-dependent shifts in absolute energy positions of the VB and CB are given by the coninement energy of a hole and electron, respectively. The simplest EMA theory considers an ininite conining potential at the nanoparticlesolvent interface. Simple analytical expressions are obtained for spherical nanoparticles of radius d: 0 ∞ EVB ( d ) = EVB −

2 = 2 π2 ; d 2 mh*,in

0 ∞ ECB ( d ) = ECB +

2 = 2 π2 d 2 me*,in

(6)

In this expression, ECB0 and EVB0 are the CB and VB energy positions for the bulk material, and me,in* and mh,in* are the effective masses of the electron and hole inside the nanoparticle. Brus (1984) was the irst to consider quantum coninement effects on the redox potential of nanoscale solids, and his predictions for CdSe with an ininite potential are shown in Figure 8. The energy step at the interface must result from either the electron afinity of a solid (~5 eV) or the band gap of the surrounding medium (3.8 eV for aqueous systems, Memming 2001). Several authors have shown that a signiicant quantitative improvement to the EMA is obtained when a inite conining potential is considered (e.g., see Fig. 7) (Tran Thoai et al. 1990; Schoos et al. 1994; Lüning et al. 1999). Exact analytical expressions are no longer obtained, and electron and hole energies are generally obtained numerically. However, Ferreyra and Proetto

Figure 8. a). Effective mass approximation (EMA) calculations of the size-dependent redox potentials of CdS nanoparticles participating in (photo)redox half-reactions. [CdS]− = negatively charged CdSe nanoparticle. [CdS]= = doubly charged particle. [CdS]* = photoexcited nanoparticle. [Used by permission of the American Institute of Physics, from Brus (1984), Journal of Chemical Physics, Vol. 80, Fig. 2, p. 4403-4409.] b). The size dependence of the energy positions of the valence band (solid lines) and conduction band (dotted lines) of ZnS, FeS2 and hematite at pH 2. Calculations based on the EMA (Eqn 6, main text) with a 4 eV conining potential and effective masses obtained from Wang and Klein (1981), Edelbro et al. (2003), and Guo (2005). Also shown are the stability limits of water. As calculated by Sherman (2005), the redox potential for the reduction of ferric iron in hematite (Fe2+/D-Fe2O3) lies below the hematite CB at pH 2. Hence photoreductive dissolution following photoexcitation of the mineral is possible under these conditions. AVS = absolute vacuum scale; NHE = normal hydrogen electrode.

128

Gilbert & Banęeld

(1999) provide an approximate expression for the coninement energies in a inite conining potential. For a spherical particle of diameter d and a conining potential Vout outside: 0 ∞ EVB ( d ) = EVB − EVB (1 − δh );

0 ∞ ECB (d ) = ECB + ECB (1 − δe )

( 7)

where Ge = with an equivalent expression for Gh. For particles in water or air, it is generally assumed that me,out* = mh,out* = me, the mass of the free electron. Holes are perfectly conined, so that Vh,out = f and Gh = 0. [=/(dme,in*)](8me,out*/Ve,out)½,

In the so-called strong coninement regime (for nanoparticle radius similar to or less than the Bohr radius), the electron and hole can be considered as individual particles. Then the band gap is obtained from difference in the size dependent VB and CB energy positions, plus the additional term resulting from the electrostatic interaction of the electron and hole pair. The band gap, Eg, for the same particle considered above is: ⎧⎪ 2 =2 π2 ⎡ 1 ⎫⎪ e2 ⎡ ( δ + δh ) ⎤ 1 ⎤ Eg ( d ) = Eg0 + ⎨ 2 ⎢ * + * ⎥ − Ee∞ δe − Eh∞ δh ⎬ − 3.6 ⎢1 − e ⎥ εd ⎣ 4 ⎦ ⎩⎪ d ⎢⎣ me,in mh,in ⎥⎦ ⎭⎪

(8)

where e is the charge on an electron and H is the high frequency dielectric constant of the bulk semiconductor. In this expression, Eg0 is the band gap of the bulk, the second term considers the kinetic energy of the electron and the hole (i.e., the coninement energy—this is an alternative version of Eqn. 7) and the smaller third term is the Coulomb attraction between electron and hole (see discussion in Nanda et al. 2004). Effective mass values may be found directly in the literature (e.g., Landolt-Bernstein 1983) or estimated from band structure calculations (e.g., Edelbro et al. 2003). Limits and extensions of the EMA. The EMA tends to overestimate the band gap in quantum conined materials, even when a inite conining potential is used, an effect that is exacerbated at the smallest sizes. Consistent values for the effective mass can be hard to obtain from the literature; however, the EMA is only weakly dependent on the precise values for the effective mass (Gaponenko 1998; Pelligrini et al. 2005). The EMA is believed to be valid for both direct and indirect semiconductors. Absorption measurements on indirect semiconductors have shown that they retain this characteristic property down to very small sizes (~100 atoms; Delerue et al. 2001) and that coninement effects on band edge positions and band gaps are similar to those in direct semiconductors (Tolbert et al. 1994). However, the UV-vis absorption edge is generally not sharp for the indirect semiconductors, and even for well studied materials such as TiO2, this transition (which reveals the true band gap) has been confused with stronger, direct transitions to higherenergy CB states (Serpone et al. 1995; Monticone et al. 2000). The EMA assumes that there are no size-dependent changes in particle structure, but changes in structure may overwhelm coninement effects in wide band-gap, low-mobility materials. Furthermore, in narrow-band-gap semiconductors, such as PbS (Eg = 0.41 eV), the band edge states are not well approximated as free electrons, and hence quantitative agreement is poor. In such cases, the EMA underestimates the band gap (Pellegrini et al. 2005). Additional reinements of the EMA approach have been performed. Examples include the treatment of multiple bands in the VB and CB (Efros and Rosen, 2000), and inclusion of surface polarization terms (Brus, 1986). Nanda et al. (2004) discuss variations of the inite potential EMA for nanoparticles with slab (2D) and rod (1D) geometries. Many of the above methods assume that important material constants for nanoparticles, such the dielectric constant or the effective mass of the relevant charge carrier(s), are unchanged relative to bulk materials. These assumptions are presently untested. However,

Molecular-Scale Processes Involving Nanoparticulate Minerals

129

recent calculations indicate that the dielectric response of a material change very little with decreasing size, with the exception of small near-surface effects (Cartoixa and Wang 2005). Real-space cluster models. As discussed above, band structure and real-space depictions of electronic structure are highly complementary approaches. Real-space methods discard all assumptions of symmetry and are thus suited to nanoparticles. However, they carry a signiicant price, because whole-nanoparticle simulations must treat hundreds of non-equivalent atoms, while band-structure calculations consider only the number of atoms in a unit cell. The most important input to a cluster calculation is the atomistic real-space structural model itself. Nanoparticle structures are generally assembled by hand, or with classical molecular dynamics structure optimization, and the uncertainty in the structures derived in these ways are a major limitation for subsequent electronic structure calculations, no matter how sophisticated. The nature and structure of nanoparticle surfaces remain particularly obscure, because no experimental method is presently able to directly visualize it. Obviously, uncertainties surrounding the true structures of nanoscale materials cause equally great uncertainties in anticipating their electronic properties. Nevertheless, atomistic simulations provide the clearest visual depictions of nanoparticle structure (Rabani 2001; Pokrant and Whaley 1999) and have played an invaluable role in evaluating theoretical approximations such as the EMA (Lippens and Lannoo 1987; Wang and Zunger 1996; Franceschetti and Zunger 1997). Simulations that are performed with care are experiments in silicio, useful for evaluating and predicting mechanisms or trends in behavior that can be tested experimentally (Zhang et al. 2003; Rustad and Felmy 2005; Kerisit et al. 2005; Erwin 2005). Real-space cluster calculations will prove essential for understanding the electronic structure in complex environmental nanoparticles for which simple models such as the EMA are not applicable (O’Connor and Sposito 2004). Ideally, electronic and physical structure would be simultaneously optimized, as implemented by the Car-Parinello method (Car and Parrinello 1985; Galli and Parinello 1992). The “Quantum Monte Carlo” simulations by the Galli group have demonstrated the value of this approach for showing the effects of surface reconstructions on the electronic structure of nanoclusters (Puzder et al. 2003). However, nanoparticles of diameter greater than ~2 nm remain large even for eficient irst-principle calculations. Solvent effects on nanoparticle electronic structure. An interesting consequence of a inite conining potential is that tails of the electron or hole wavefunctions extend into the medium surrounding the particle, as depicted in Figure 9. This generally has a slight effect on the energies of the electron and hole states. However, solvents with a high dielectric constant can stabilize nonuniform charge distributions at the surface (Rustad and Felmy 2005), enhancing the strength of dipolar interactions between nanoparticles (Rabani 1999) and can permit solvent-mediated conductivity (Brus 1996), as discussed below. The simplest approach for estimating solvent effects on electronic energy levels in nanoparticles is to treat the solvent as a continuum characterized by a dielectric constant (Qu and Morais 1999; Rabani et al. 1999; Franschetti et al. 2000). However, it is clear from decades of research on mineral-water interfaces that a continuum description of water is severely limited (Hochella and White 1990). In fact, we require a much more complete description of the nanoparticle–electrolyte interface. Interfacial electrochemistry of nanoparticles. When a solid material is immersed in an electrolyte such as water, redistribution of charge carrying species occurs on both the solid and liquid sides of the interface. This process can produce signiicant shifts in the absolute energy positions of electronic states at the interface, and hence affect the redox properties of the bulk mineral or nanoparticle.

130

Gilbert & Banęeld

Figure 9. The effect of the height of conining potential on the energy levels and wavefunctions of an electron in a nanoparticle can be illustrated by the simple “particle in a box” model. Shown are the three lowest energies, Ei, and wavefunctions, ERed. Thus, the reduced and oxidized states of the redox system are analogous to the occupied (valence band) and unoccupied (conduction band) states in a solid, and we may deine an effective Fermi energy for the solution, EF,redox, that lies midway between EOx and ERed. The electrochemical potential, μi , of an ion, electron, or other species, i, is equal to its chemical potential, Pi, plus an additional term for each charged species if it resides within an electric ield. μi = μi + zi F φ

(10)

The additional energy term is the product of the charge on the species, zi, the local electrostatic potential, I, and Faraday’s constant, F. The relation between the chemical potential, Pi, and the activity of ion i is given in many textbooks (Lyklema 2001). The electrochemical potential for the electron in Equation (9) is deined as the difference in the values for the oxidized and reduced species, μ e,redox = μ Red − μOx. (Memming 2001). The electrolyte Fermi energy, EF,redox (in eV), is then related to μ e,redox (in J mol−1), by EF ,redox =

e μ e,redox F

(11)

For the solid phase and aqueous phase Fermi level concepts to be directly comparable, they must be expressed in the same units on the same energy scale, typically in eV on the absolute vacuum scale (AVS). The redox potential associated with a half reaction such as

Molecular-Scale Processes Involving Nanoparticulate Minerals

131

Equation (9) is usually expressed in V relative to the normal hydrogen electrode (NHE) or other electrode. As discussed by Xu and Schoonen (2000), the energy of an electron in an energy band can be converted to the associated redox potential by the relation E (NHE, V) = −E (AVS, eV) − 4.5

(12)

At equilibrium, EF = EF,redox, which requires charge redistribution. Electrons low into the solution from the material and change the concentration of redox species if EF > EF,redox, and vice versa. This changes the potential at the interface, lowering the free-energy difference due to the term ziFI in Equation (10) until equilibrium is attained. The VB and CB states at the interface are affected by the local potential, a phenomenon called band bending, as depicted in Figure 10. Energy levels further from the interface than the Debye length, LD, are shielded from the effect of the interfacial potential because of the dielectric properties of the solid. LD is typically greater than 100 Å, depending on the density of charge carriers. Thus, surface charges are not well screened in particles of dimensions smaller than this, and the electronic bands are lat (but shifted) throughout a nanoparticle, as illustrated in Figure 10. This depiction assumes that the kinetics of the redox couple are fast, but it is well known that many environmental systems are far from thermodynamic equilibrium (Schüring et al. 2000). Electric double layer. The distribution of aqueous ions near the surface of a mineral in water is affected by the presence of an electrostatic potential at this interface. Such surface

Figure 10. Semiconductor particles immersed in water reach equilibrium by donating or accepting electrons from solution until the electronic electrochemical potential (the Fermi level) is constant throughout. The Fermi level in the solution is determined by the redox couple(s) in solution. In micrometerscale particles, the change in local charge carrier density causes shifts in VB and CB energy levels (band bending) over a space charge region, dSC. The direction of band bending depends on whether holes (p-type) or electrons (n-type) are the dominant charge carriers (the results for an n-type semiconductor are shown). For nanometer-scale particles of diameter d < dSC the electronic bands inside the entire particle are shifted. The positions of the nanoparticle CB and VB are shown shifted both due to surface charging (which lifts the CB and VB energy positions) and by quantum coninement (which increases the CB-VB separation). After Memming 2001.

132

Gilbert & Banęeld

potentials can be created by equalization of the Fermi levels in the solid and the electrolyte (as described above) and by the chemisorption of charged solution species to the mineral surface. For a given mineral surface, strongly interacting charged species are called potential determining ions (PDI) (Lyklema 2001). For example, H+ and OH− are PDI for metal oxides, while HS− can be a PDI for sulide minerals (Bebie et al. 1998). The pH driven shifts in oxide band energies are quantitatively described by the Nernstian relation that predicts a shift of 0.059 V/pH at 25 °C and 1 atmosphere pressure. However, there are apparently no experimental tests of this relation for nanoscale particles. Similarly, although nanoparticles (and colloids in general) act as ready sorbents for inorganic and organic ions and molecules, the effects of different sorbates on band positions are generally not known. The adsorption of redox active solution species seldom affects the energy band positions.

REDOX BEHAVIOR OF NANOPARTICLES Having completed a brief review of the factors that affect the energies of electronic bands in solids, we are now able to apply this knowledge to the (photo)chemical reactions of nanoparticles. The fundamental reactions in which nanoparticles can participate are depicted in Figure 11.

Size effects on nanoparticle redox behavior The modiication of the absolute valence and conduction-band energy levels is a predominant effect on redox behavior when it occurs. As discussed above, doping of a semiconductor, sorption of potential determining ions, and inite particle size may all contribute to such effects. Examples of size effects on redox potential. As shown by Brus (1984), and reproduced in Figure 8a, inite size effects can have a large impact on the redox potentials of semiconductors (see also Franschetti et al. 2000). However, the extent to which valence electrons are delocalized and hence susceptible to inite size effects is unclear for several environmentally

Figure 11. Scheme of the possible redox, photochemical and charge or energy transfer reactions that can take place at the surfaces of semiconductor mineral nanoparticles. D = electron donor (or reductant); A = acceptor (or oxidant); S = surface adsorbed sensitizing ligand; VB = valence band; CB = conduction band; Eg = band gap. z− (z+) represents a negatively (positively) charged nanoparticle;z* represents a nanoparticle containing a photoexcited electron-hole pair.

Molecular-Scale Processes Involving Nanoparticulate Minerals

133

relevant materials. For example, sphalerite (ZnS) and pyrite (FeS2) are, respectively, direct and indirect delocalized electron semiconductors in which quantum coninement effects will occur. Figure 8b plots the size-dependent shifts in the VB and CB levels for these materials predicted by the EMA. It is presently dificult to anticipate quantum coninement effects in environmental iron and manganese compounds for which electron and hole effective masses have not been tabulated. Nevertheless, a recent X-ray spectroscopic study identiied an increase in the band gap of hematite nanorods ~ 4 nm in diameter by approximately 0.3 eV (Guo 2005). UV-vis spectroscopy of encapsulated iron oxide nanoparticles also indicated size-dependent bandgap opening (Iwamato et al. 2000). In common with most iron (III) and manganese (IV) (oxyhydr)oxide minerals, the conduction and upper valence bands have the character of cation d-states, while the lower valence band is principally composed of oxygen p-like orbitals, although covalency in the metal-oxygen bond causes mixing in the valence band states (Cox 1992; Sherman 1984, 1985, 2005). In Figure 8b, we predict the size dependence of the hematite band gap using an estimate for the effective masses of charge carriers (assuming me* = mh*) consistent with the observations of Guo (2005). Band-gap opening, as depicted in Figure 8, will signiicantly affect mineral reactivity (Rodriguez et al. 1998). The predictions of Figure 8 require further experimental testing, particularly the use of combined X-ray absorption and emission spectroscopic measurements of nanoparticle band gaps (Lüning et al. 1999; Sherman 2005), and more sophisticated theoretical treatments (O’Connor and Sposito 2005). A complementary approach for understanding the electronic properties of mineral nanoparticles will be studies of the kinetics of surface reactions. Kinetics studies have been used to determine the relative reactivities of different iron minerals to surface redox or complexation reactions (Hering and Stumm 1990; Elsner et al. 2004; Poulton et al. 2004; Peak 2005). Such studies provide a method for determining size-dependent changes in surface reactivity (see below; Madden and Hochella 2005). The roles of surface states. Atoms at the surface of a material are not generally able to attain the same coordination environment that is present within the interior. Thus, atomic sites at a surface may exhibit modiied local electronic structure that in some cases can introduce energy levels within the band gap of a semiconductor or insulator (Morrison 1980). Surface states can therefore have the same effect as interior impurity atoms, either facilitating the creation of mobile electron or hole charge carriers or acting as traps for them. For example, underbonded surface anions can donate electrons into the CB of the mineral (cf. Fig. 16). Surface states can also mediate the transfer of charge between an adsorbate and states within a mineral or between two aqueous or adsorbed reagents. In addition, the trapping of electrons at surface states can affect the lifetime of photoexcited electrons and holes, and can determine the rate of electron transfer between neighboring particles (see below). The above phenomena occur at the surfaces of both bulk minerals and nanoparticles. The surfaces of nanoparticles are structurally more diverse, and molecular simulations indicate that inhomogeneous charge distributions can occur at nanoparticle surface and edge sites, modifying the surface Lewis acid or base characteristics relative to large particles (Lucas et al. 2001; Noguera et al. 2002; Rustad and Felmy 2005). Scanning tunneling spectroscopy (Preisinger et al. 2005) and optical luminescence spectroscopy (Chen et al. 1997) are approaches for detecting surface states that lie in the electronic band gap. However, environmental nanoparticles and their synthetic analogs are poorly studied. Undercoordinated surface sites tend to be more reactive and hence are frequently the sites at which molecules bind to nanoparticle surfaces. In place of the initial surface states, new

134

Gilbert & Banęeld

surface-ligand molecular orbitals are formed, with discrete energy levels that may no longer reside within the band gap. While ligand binding has been extensively studied for the removal of mid-gap states in engineered nanoparticles (e.g., Green and O’Brien, 1999), ligand binding can also create (photo)redox active mid-gap energy states (e.g., Rajh et al. 2002)

Examples of nanoparticle redox behavior Nanoparticles as molecular-like redox active solution species. Nanoparticles can accept or donate electronic charge, and in this sense can be considered redox active species. Figure 12 shows that CdS nanoparticles can diffuse to, and react with an electrode in an electrochemical cell in a manner similar to an aqueous ion (Kukur et al. 2003). The potential at which the nanoparticles can be reduced (i.e., charged by a single electron) varies with particle size in agreement with coninement effects. However, even for nonaggregated nanoparticles, diffusion rates are considerably slower than dissolved ions or molecules (Scholz and Meyer 1998). The Brownian diffusion rate for nanoparticle transport is given by the Stokes-Einstein equation: D=

kBT 6 πηr

(13)

where K is the solution viscosity, and r is the particle radius. The implication of the results of Figure 12 is that nanoparticles are available to participate in molecular redox reactions with aqueous ions and biomolecules. Below, we discuss the important issue of the stability of individual nanoparticles during (photo)redox reactions. The example above is one of a number of experimental investigations of the charging of nanoparticles that was conducted at electrochemical electrode (Haram et al. 2001; Kukur et al. 2003; McKenzie and Marken 2001). In nature, redox active solution species can inject electrons into the conduction band of a mineral, provided that the redox potential of the aqueous redox reaction is more negative than the position of the CB minimum. For example, Cd(II) and Co(II) can be oxidized on the surface of ZnO and manganese (IV) oxides, respectively (Murray and Dillard 1979; Manceau et al. 1997). In the opposite direction, electrons may transfer from the mineral VB to a strongly oxidizing organic species such as ascorbic acid, which can cause the direct oxidation of manganese oxide minerals (Stone and Morgan 1984; Stumm and Morgan 1996). A negatively charged nanoparticle is free to act as a reductant with a suitable acceptor species. If further charge transfer to the nanoparticle carries an energy penalty (see below),

Figure 12. Semiconductor nanoparticles exhibit molecular-like redox behavior with size-dependent redox potentials. Electrochemical oxidation of a solution of CdSe nanoparticles in acetonitrile at a gold electrode shows a clear trend with increasing particle size (a to d). The position of the oxidation peak (Ep) indicates the valence band maximum and the trend is in quantitative agreement with an effective mass approximation calculation of electronic coninement energies. a = 3.23 nm diameter; b = 3.48 nm; c = 3.73 nm; d = 3.8 nm. [Used by permission of the American Institute of Physics, from Kucur et al. (2003) Journal of Chemical Physics, Vol. 119, Fig. 3, p. 2333-2337.]

Molecular-Scale Processes Involving Nanoparticulate Minerals

135

additional reactions with donor species are not favored the excess charge has been lost. An important question is whether oxidant and reductant must be simultaneously surface bound, or whether charged nanoparticles can be formed that are suficiently stable in solution to act as reactive intermediates. The direct reaction scheme can be written D < O > A o D+ < O > A−

(14)

D < O o D+(aq) + [O]−

(15a)

where O > D represents donor species D adsorbed onto a nanoparticle. Alternatively: [O] + A(aq) o O > A −

−

(15b)

For electron transfer reactions, the donor (or acceptor) species and the nanoparticle must (1) have a redox potential that coincides with electron energy bands in the minerals, and (2) attain suficient wavefunction overlap with these bands (Huber et al. 2000). Many singleelectron transfer reactions can occur via electron tunneling to or from species bound to the surface by outer-shell adsorption. However, ligand-particle electron transfer rates are greatly enhanced by surface complexation (Moser et al. 1991); and certain charge transfer reactions, particularly involving Fe d-electron states, require inner-sphere surface coordination to proceed at all (e.g., Ennaoui and Tributsch 1986). The possibility of multiple oxidation states of nanoparticles. The energy required to singly charge a solvated nanoparticle (Fig. 11a) is sensitive to both the nanoparticle size and the dielectric constant of the solvent (Franceschetti et al. 2000). The addition of subsequent electrons to an already charged nanoparticle is possible, but additional energy may be required to compensate for Coulomb repulsion if there is electronic overlap between CB electrons (Brus 1984). In this case, nanoparticles can behave more like atoms with variable redox states (Banin et al. 1999), than bulk minerals within which excess charges may diffuse apart. Atomic-like redox behavior would lead to a quenching of bioreductive processes involving nanoparticles, because the successive charging energies would eventually exceed the reducing power of extracellular electron shuttle molecules. However, this is not observed for the biological reduction of ferric iron minerals, since phases with higher surface areas (i.e., smaller particle size) are more completely reduced (Roden and Zachara 1996; Hansel et al. 2004). For minerals composed of atoms susceptible to valence changes, mineral dissolution is an effective pathway for shedding excess charge. Furthermore, in the small polaron model of charge transport in localized carrier materials, an additional electron at an iron site is spatially localized within a radius less than the near-neighbor bond length (Cox 1992). This implies that interactions between multiple ferrous iron sites will be weak, and that the energy required to add an electron to a ferric iron nanoparticle will not depend on the number of excess electrons already hosted, in contrast to the behavior of delocalized band semiconductors. Reductive transformation of halogenated organic contaminants. Many relevant investigations have been motivated by the desire to identify natural pathways for immobilizing or transforming organic or inorganic contaminants. For example, Fe(II) adsorbed to the surface of ferric iron and non-iron-containing minerals can be a powerful reductant, one that can transform hazardous chlorinated hydrocarbons such as carbon tetrachloride (CT) (Pecher et al. 2002; Elsner et al. 2004) and chromate (Wielinga et al. 2001). Mineral surfaces that stabilize the Fe(III) product of the oxidation of surface-bound Fe(II) can lower the Fe(II)Fe(III) redox potential (Stumm and Morgan 1996; Amonette et al. 2000; Pecher et al. 2002). Structural Fe(II) in mixed-valence iron oxide compounds such as magnetite, particularly biogenic magnetite nanoparticles, can also transform CT. As shown by Figure 13, numerous electron shuttle molecules used by anaerobic DFeRB also possess reduction potentials that can degrade CT, and co-metabolic pathways

Gilbert & Banęeld

136

Figure 13. Structural Fe(II) in biogenic magnetite nanoparticles (shown in TEM image, left) is a potent reductant of halogenated organic solvents. Although the redox potentials of common biological electron shuttle molecules are also suficient to drive the reduction of carbon tetrachloride (right), the mineral nanoparticle driven process is almost 100 times more effective. [Used by permission of the American Chemical Society, from McCormick et al. (2002), Environmental Science and Technology, Vol. 36, Figs. 1 & 4, p. 403-410.]

do contribute to the natural attenuation of this compound. However, in laboratory studies, biogenic magnetite is approximately two orders of magnitude more effective than biomolecular pathways (McCormick et al. 2002; McCormick and Adriaens 2004). Nanoscale magnetite also reduces U(VI) to U(IV) under anaerobic conditions (following surface adsorption) much more rapidly than occurs in solution under reducing conditions (Missana et al. 2003). Surface-promoted redox reactions. Bulk mineral surfaces can mediate charge transfer between solution species that otherwise interact too weakly for effective redox pathways. For example, sulide minerals—including pyrite, galena, and several doped sphalerite minerals— catalyze the oxidation of thiosulfate to tetrathionate by dissolved molecular oxygen (Xu and Schoonen 1995). A signiicant nanosize effect was recently observed for hematite-promoted Mn oxidation (Madden and Hochella 2005). While the oxidation of aqueous Mn(II) by oxygen is very slow at pH < ~8.5, it may be promoted following adsorption to mineral surfaces. Surface hydroxyl groups (denoted >OH) mediate electron transfer from molecular oxygen to adsorbed manganese, facilitating the reaction: Mn 2+ +

1 3 OH O2 + H 2O ⎯>⎯⎯ → Mn(III)OOH + 2H + 4 2

(16)

As shown in Figure 14, the kinetics of this reaction, normalized to surface area, exhibit an increase of more than one order of magnitude for 7 nm diameter hematite nanoparticles compared with 37 nm particles. Madden and Hochella (2005) discuss the possible origins of this striking enhancement of reactivity. Hematite energy bands do not play a direct role in this process; hence, electronic coninement effects are unlikely to be responsible. Possible explanations of the rate enhancement may be understood from Marcus’s original description of the rate of electron transfer (Marcus 1993). By setting 'G* = ('G° + O)2/4O in Equation (5),

(

⎛ ΔG o + λ ket = κν exp ⎜⎜ − 4λkBT ⎜ ⎝

)

2

⎞ ⎟ ⎟ ⎟ ⎠

(17)

where 'G° is the standard free energy of the reaction (positive or negative) and O is the reorganization energy (always positive). Size-dependent changes in either 'G° or O are plausible and would modify the reaction kinetics, provided that electron transfer is the rate-limiting step.

Molecular-Scale Processes Involving Nanoparticulate Minerals

137

Figure 14. The rate of heterogeneous oxidation of Mn(II) promoted by 7 nm and 37 nm diameter hematite nanoparticles. The smaller hematite nanoparticles promote oxidation at a rate that is almost two orders of magnitude larger than the larger particles. [Reprinted with permission of Elsevier, from Madden and Hochella (2005) Geochimica et Cosmochimica Acta, Vol. 69, Fig. 5, p. 389-398.]

For example, it is likely that the redox potential of Mn(II) adsorbed to the smaller particles is shifted because of a size-dependent modiication in the Lewis base character of oxygen atoms on the surface of the hematite to which Mn adsorbs (Noguera et al. 2002, Lewis et al. 2001). If this drives the 'G° of Reaction (16) more negative, Mn oxidation would not only be more favorable, but ket would increase. Alternatively, small particles may exhibit a higher density of surface sites at which the coordination geometry of adsorbed ions is distorted from the perfectly octahedral coniguration preferred by Mn(II). However, Mn(III) complexes tend to prefer a distorted octahedral coordination. Therefore, if the smaller nanoparticles possess greater surface disorder, less structural reorganization may be necessary for the reaction to proceed. This effect would reduce O, thereby increasing ket The results of Madden and Hochella (2005) demonstrate that the redistribution of both charge and atoms at the surfaces of nanoparticles may strongly inluence their reactivity. EXAFS investigations into changes in the coordination environment of Mn(II) bound to hematite nanoparticles in the absence of oxygen may help evaluate these two models.

PHOTOCHEMISTRY A photon of energy greater than the band gap can excite a valence electron to the CB, which leaves a vacant orbital (hole) in the VB. The excited electron (hole) has the ability to reduce (oxidize) chemical species at the surface of the nanoparticle. As with redox chemistry, the ability to do this depends on the absolute electron or hole energy, and hence can be affected by particle size (and pH), as described above.

Size effects on nanoparticle photochemistry Kinetics of recombination and reaction. Following photoexcitation of a nanoparticle, several processes can occur that facilitate or prevent reaction, and these processes may exhibit distinct small-size effects (Gratzel and Frank 1982; Gerischer 1993). Diffusion of an excited electron or hole to the surface and transfer to surface species competes with the recombination of the electron-hole pair and trapping at surface states (Zhang 2000). The transit time to the surface varies as the square of the particle radius, and is generally less than 1 ps for few-nm

138

Gilbert & Banęeld

diameter particles (Gratzel and Frank 1982; Huber et al. 2000). The recombination time is principally material dependent, with a weak size dependence caused by enhanced electronhole overlap. Since the recombination time is generally in the range 0.1–1 ns, exited electrons have a far greater probability of reaching the surface of nanoparticles than larger colloidal particles, such as >100 nm diameter iron oxide colloids, in which most electron-hole pairs recombine before reaching the surface (Leland and Bard 1987). Nanoparticle surface states can act as traps for excited electrons with lifetimes that may vary by many orders of magnitude for different nanoparticles. If a photoexcited electron or hole is scavenged by a solution or surface species, recombination within the nanoparticle is no longer possible, and the nanoparticle can remain excited for a considerable time (e.g., minutes) (Leland and Bard 1987). Because both a hole and electron are created following light absorption, both cathodic (i.e., reduction) and anodic (i.e., oxidation) reactions are possible at a nanoparticle surface and may proceed in very close proximity. The kinetics of these reactions are seldom equivalent, providing an opportunity for reaction intermediates to interact. Several studies have concluded that competition between the above processes, plus variation in surface area:volume ratio, leads to an optimum particle size for the maximum eficiency of a given photoreaction that is frequently in the 5–20 nm diameter range (Wang et al. 1997; Almquist and Biswas 2002). Reactions requiring multiphoton absorption. The probability of a photoexcited nanoparticle absorbing a second photon declines with particle size for statistical reasons (Wang et al. 2003), and hence reactions requiring rapid multielectron transfer are highly unlikely for nanoscale particles under environmentally relevant illumination conditions. For example, the products of the photooxidation of ethanol are different for micron-sized versus nanometer-sized ZnS colloids (Müller et al. 1997). Under constant illumination, transfer of two photoexcited holes to adsorbed ethanol can occur readily within 200 ns on the surface of the larger particles. By contrast, the mean time to create two holes in a nanoparticle reaches several seconds, permitting partially oxidized ethanol radicals to diffuse into solution and form more complex organic species. Shifts in electronic band energy positions. The effect of particle size on the photoredox activity of nanoparticles can be illustrated through analogy, with the effect of pH on the photoreduction of methylviologen ions (MV2+) at the surface of colloidal TiO2 particles (Duongdong et al. 1982). The half-reaction, MV+ l MV2+ + e−, has the redox potential E0 = −440 mV versus NHE that is independent of pH. By contrast, the electrochemical potential of the TiO2 CB varies with pH as ECB (TiO2) = ECB (TiO2, pH 0) – 0.059(pH) V (vs. NHE)

(18)

As shown in Figure 15, photoexcited electrons in the CB are suficiently reducing only above pH around 4–5. Photoreduction of MV2+ is thermodynamically forbidden below this pH threshold. As shown below, for materials that exhibit size-dependent shifts in CB position, thresholds in particle size can exist below which photoredox reactions are enabled. For completeness, it should be recalled that changes in pH can strongly affect the afinity of ionic sorbates to surface binding sites, particularly around the point of zero surface charge (pHzpc) for a material (Kormann et al. 1991). Such effects can complement or compete with shifts of the nominal redox potentials of CB electrons and VB holes.

Nanoparticle interactions with biomolecules It is clear from the study of environmental colloids that mineral surfaces can exhibit high afinities for organic molecules (Yariv and Cross 1979; Amal et al. 1992; Tiller and O’Melia 1993), and many groups have demonstrated effective surface binding of molecules to sulide

Molecular-Scale Processes Involving Nanoparticulate Minerals

139

Figure 15. The photoreduction of methylviologen (MV) at the surface of TiO2 colloids shows a strong pH dependence. a). The energy position of the TiO2 conduction band (CB) displays a linear Nernstian response to increasing pH, while the redox potential of the MV+/MV2+ couple is pH independent. b). The yield of the reduced aqueous ion MV+ vs. pH shows a strong increase around the pH at which electrons in the TiO2 CB are more reducing than MV+. Used by permission of the American Chemical Society, from Duongdong et al. (1982), Journal of the American Chemical Society, Vol. 104, Pages 2977-2985, Figure 4.

and oxide nanoparticles via assorted terminal functional groups. In particular, biomolecules such as amino acids, phospholipids, siderophores, and even DNA have been shown to stabilize sites on nanoparticle surfaces (Konovalova et al. 1999; Jones et al. 2000; Torres-Martínez et al. 2001; Dwarakanath et al. 2004). Recent studies have addressed the detailed electronic structure associated with functional groups on bacterial surfaces, including the position of the Fermi level (Vyalikh et al. 2004; Ireta et al. 1998), which will permit quantitative treatments of nanoparticle-microorganism interactions. As discussed above, strong chemical interactions can create new electronic states in the mineral, modifying (photo)redox reactivity. The coupled redox reactions of iron and manganese oxides and biomolecules play a vital role in the transformation of organic matter and the production of humic materials. Interestingly, while hydrated bioilms may considerably coat the surfaces of minerals, they appear not to seriously hinder the surface adsorption and reaction of either small molecules or metal cations (Templeton et al. 2003; Toner and Sposito 2005), although certain aqueous ions may partition between mineral surface sites and organic functional groups depending upon pH (Warren and Haack 2001). Roles of adsorbed surface species in redox and photochemistry. The complexation of surface atoms of a nanoparticle by adsorbed molecules can introduce additional electronic states within the semiconducting band gap (see Figs. 11f and 16), a phenomenon termed sensitization. Sorbates can signiicantly increase the rate of photochemical reactions, because the photon energy threshold for creating an excited electron or hole in electronic bands of the nanoparticle can be greatly reduced (Konovalova et al. 1999; Rajh et al. 2002). Charge injection from the adsorbate to the nanoparticle is generally extremely fast (< 0.1 ps), while electron transfer back to the adsorbate, which would permit electron-hole recombination, may take milli- or even microseconds (Moser et al. 1991; Huber et al. 2000).

140

Gilbert & Banęeld

A series of important experiments by the group of Rajh have shown that organic molecules possessing enediol groups (–CHOH=COH–) bind to and hybridize with several metal oxide nanoparticles (Fig. 16, Rajh et al. 2002). This process is particularly eficient because undercoordinated cations on the surfaces of Fe2O3 and TiO2 nanoparticles sit within considerably distorted sites. Ligand binding restores local octahedral symmetry and is energetically very favorable. Furthermore, Rajh et al. (2004) showed that single-stranded DNA binds to the surface of TiO2 nanoparticles and retains the ability to hybridize with complementary DNA. Following direct photoexcitation of the nanoparticle, charge transfer occurs readily onto double stranded DNA, but not single-stranded DNA.

Figure 16. Organic molecules containing the endiol group sorb readily to the surfaces of hematite nanoparticles and introduce electronic states into the mineral band gap. These surface ligands “sensitize” the nanoparticles: much less energy is required to excite electrons in these mid-gap states to the mineral CB, enhancing the probability of generating excited electrons with high reducing power following light illumination. After Rajh et al. 2002.

Examples of nanoparticle photochemistry Photofixation of CO2. There is a considerable technological effort behind using colloidal particles for harvesting solar energy, not only for electrical energy production, but also for performing benign photochemistry, such as the splitting of water (Bard and Fox 1995) or the photocatalytic degradation of organic environmental pollutants (Hoffmann 1995). Following Inoue et al. (1995), the photoreductive “ixation” of atmospheric CO2 has been studied by many groups. In most engineered systems, this pathway has always been rather ineficient, although several groups have claimed improvements using chalcogenide nanoparticles including ZnS and CdS (Fujiwara et al. 1997; Fujiwara et al. 1998). The reaction begins with the step: >CO2 ⎯hν ⎯→ >CO2• −

(19)

where >CO2• − indicates a surface-bound CO2 radical. CO2• − is extremely reactive and, following desorption into water at pH 7, produces formic acid (HCOOH), CO, and H2 (Fujita and DuBois 2003). First-principles calculations have shown that two effects (one structural, one purely electronic) can together explain the experimental observation that nanoparticles of CdSe are capable of photoixation of CO2 (Wang et al. 2002), a process not observed for bulk CdSe (Nedeljkovic et al. 1986). As shown in Figure 17, CO2 molecules may chemisorb at Se

Molecular-Scale Processes Involving Nanoparticulate Minerals

141

Figure 17. Nanoparticles of CdSe can catalyze a step in the photostimulated ixation of molecular CO2, a role that bulk CdSe cannot play. Left. CO2 adsorbed at a site of a selenium vacancy on the nanocrystal surface achieves electronic overlap with the nanoparticle band structure. Following light absorption by the nanoparticle, a conduction band (CB) electron can be transferred to the CO2. The reactive charged molecule has a low barrier for desorption. Right. Energetic diagram for the photoexcitation of CO2. Nonbonding electrons on surface Se atoms produce localized occupied states above the valence band (VB) maximum in the band gap. Light absorption can excite these electrons to the CB. Due to coninement effects, CB electrons in nanoparticles less than ~5 nm in diameter are suficiently high in energy to transfer to the lowest unoccupied molecular orbital of CO2. The CB minimum in bulk CdSe lies at too low an energy for this step to proceed. After Wang et al. 2002.

vacancies on a nanocrystal surface. Because the CB lies at higher energy in the nanocrystals than in the bulk, photon absorption creates an excited electron with a redox potential capable of reducing the sorbed molecule to produce a reactive charged radical, with a low energy barrier to desorption. Subsequent reactions can create small organic molecules such as formic acid. There is no evidence that nanoparticle production fulills a biological role of encouraging CO2 ixation. Nevertheless, the above mechanism is one way that nanoscale minerals affect organic carbon transformations. Moreover, it illustrates the capacity for nanoscale minerals to generate reactive radicals. Photogeneration of reactive oxygen species. Photogenerated radicals derived from water or oxygen are commonly found to be reactive intermediates during heterogeneous photochemical reactions (Gerischer 1993; Hoffmann et al. 1995; Riegel and Bolton 1995; Hall 2000; Torres-Martínez et al. 2001; Garrett et al. 2005). The principal reactive species − + ) or hole (hVB ) by surface-adsorbed are formed by the capture of a photoexcited electron (eCB molecular oxygen or water, respectively: ν + >O2 ⎯h⎯ → >O2• − + hVB

(20a )

ν − → >OH• + H + (aq) + eCB >H 2O ⎯h⎯

(20 b)

Hydroxyl radicals typically remain surface bound and can be considered as holes trapped at surface states (Lawless et al. 1991), while the superoxide anion, O2• −, may diffuse into solution. Further reduction of O2• − can generate hydrogen peroxide, H2O2, but H2O2 is not evolved under anaerobic conditions. All of the reactive oxygen species can cause cellular damage (Hall 2000), and interactions between nanoparticles and biomolecules can increase the type and number of reactive oxygen species produced photochemically. For example, the yield of O2• − during photoexcitation of TiO2 is greatly enhanced by the presence of low-concentration carotenoid sensitizers (Konovalova et al. 2004). Furthermore, extracellular electron shuttles such as quinones

142

Gilbert & Banęeld

(Nevin and Lovely 2000; Newman and Kolter 2000) can act as diffusive secondary reactive intermediates that propagate cell toxicity (Bolton et al. 2000). However, certain strongly bound surface ligands such as halogenated acetic acids may be oxidized directly without a role for reactive intermediates (Pehkonen et al. 1995). The photoactivity of speciic minerals cannot always be predicted from thermodynamic considerations and many environmental minerals are not well studied. The combination of sunlight and dissolved oxygen is required for signiicant photogeneration of cytotoxic radicals, limiting the impact of these processes to near-surface environments.

The stability of nanoparticles during redox chemistry and photochemistry Stability during redox reactions. An excess or deiciency of charge in a nanoparticle may be delocalized within electronic bands or localized at atomic sites, depending upon the mobility of charge carriers within the mineral. Redox active atoms within a nanoparticle may capture the electron or hole through a valence state change, which can lead to nanoparticle dissolution. Dissolution of insoluble ferric iron and manganese oxide minerals may occur following charge injection from a solution donor (Fig. 11a) (Hering and Stumm 1990), which can include biological electron-transfer molecules (Newman and Kolter 2000; Cheah et al. 2003) or molecular “nanowires” (Reguera et al. 2005), suficiently reducing inorganic (Missana et al. 2003) or organic (McCormick et al. 2002) species, or photoexcited ligands (Voelker et al. 1997). Reductive dissolution of hematite. Direct electrochemical measurements of Fe2O3 nanoparticles drive dissolution, once the electrode potential is suficiently negative to reduce the iron oxide particles (McKenzie and Marken 2001). The chemical reductive dissolution of hematite can be driven by numerous reductants (Stumm and Morgan 1996), including dissolved sulide (Dos Santos Afonso et al. 1992; Poulton et al. 2004). When the reductive dissolution of ferric iron oxides was quantiied (using a radiologically generated reductant), the amount of dissolved Fe(II) plus adsorbed Fe(II) could not account for all the reductant consumed (Mulvaney et al. 1988). The reductive dissolution of hematite can be written: Fe 2O3 + 6H + + 2e − ⎯⎯ → aFe 2+ (aq) + bFe 2+ (sorb)+ cFe 2+ (inc) + 3H 2O

(21)

where the prefactor a is the fraction of reduced iron liberated into solution, b is the fraction adsorbed at the surface, and c is the fraction of electrons incorporated in the interior of the colloidal particles (a + b + c = 2). The prefactors exhibit a strong pH dependence as expected, and vary with colloidal particle size. Around neutral pH, c ≈ 0.25 for 5 nm diameter particles, but c ≈ 0.7 for micron-sized colloids. Thus, some ferrous iron sites are indeinitely stabilized within ferric iron minerals. The partially reduced minerals are likely an intermediate step in the solid-state transformation to magnetite, perhaps with a disordered structure analogous to the that found as a thin layer upon hematite colloids (Fig. 2; Hansel et al. 2004). The partially transformed nanoparticles are reductants themselves, with an activity that depends both on the effective redox potential of structurally incorporated Fe(II) and its mobility, and partitioning between the surface and interior (Williams and Scherer 2004). As expected, higher-surfacearea particles are more effectively dissolved, but even the smallest nanoparticles retain a number of ferric sites. Stability during photoredox reactions. Many semiconductors that are stable against reductive surface processes can be subject to photocorrosion (Gerischer 1980; Memming 2001). If a nanoparticle participates in the light-stimulated reduction of an acceptor molecule, a hole remains in the VB (Fig. 11e or f). From a molecular perspective, a hole is a partially broken bond (albeit a potentially mobile one). Similarly, a photo-oxidation event (Fig. 11d) entails the production of excited electrons that may drive a local valence state change at Fe(III) and Mn(IV) sites and detachment of the atom (i.e., dissolution).

Molecular-Scale Processes Involving Nanoparticulate Minerals

143

Photoreductive dissolution of hematite. Sherman (2005) combines oxygen X-ray absorption and emission spectroscopy to determine the conduction and valence-band energy positions of bulk iron and manganese oxides and oxyhydroxides. The position of the hematite CB relative to the redox potential for hematite dissolution determines whether photoexcitation of the mineral (Fig. 11c) can lead to dissolution. Since at pH 2, E0 = 0.655 V for Fe2+/D-Fe2O3, assuming an aqueous concentration of ferrous iron to be [Fe2+] = 10 nM, and ECBhematite = 0.3 eV, direct photodissolution will occur at pH 2, but not at circumneutral pH. At pH 8.3, the bulk hematite CB is predicted to lie ~0.4 eV below the Fe2+/D-Fe2O3 couple on the AVS scale (Sherman 2005). However, as indicated in Figure 8b, hematite nanoparticles with a diameter of ~2 nm are predicted to possess suficiently reducing CB electrons for photodissolution to proceed at pH 8. Because of the increase in the band gap, higher energy photons would be required to drive photodissolution of the smaller particles. The presence of additional solution species can also favor the photoreduction of hematite, either by sensitizing the mineral surface (Pehkonen et al. 1995) or by complexing soluble Fe(II) so that the electrochemical potential for iron reduction is made more positive (Kraemer 2004) and lies beneath the CB. Ligand promoted photoreductive dissolution of manganese oxide. Manganese oxides in the presence of dissolved organic matter are sensitive to photoreduction (Sunda et al. 1983; Scott et al. 2002; Haack and Warren 2003). Following light absorption by an adsorbed ligand (e.g., humic and fulvic acids; ascorbic acid), fast injection of an electron into the mineral VB (Fig. 11f) can drive reduction of Mn(IV) to Mn(II), which is subsequently soluble. Since this is a two-electron process, the dissolution rate will depend upon the mobility of electrons in the mineral and is likely to exhibit a signiicant dependence on particle size. Manganese reduction performs an important role in the redox cycling of Mn, and can impact the bioavailability of nutrient and contaminant adsorbates by releasing them into solution (Crerar et al. 1976). Daily cycles in the quantity of oxidized Mn in a bioilm of Mn oxidizers are consistent with sunlightdriven photodissolution (Bourg and Bertin 1996; Haack and Warren 2003). Photodecomposition of sulfides. Metal sulide particles are susceptible to photodecomposition in the presence of oxygen (which acts as an electron acceptor), producing soluble metal and sulfate ions. Although stability is greater in anaerobic environments, other solution species, such as sulite ions, can act as electron donors. For example: CdS(s) + SO32 − + H 2O ⎯hv ⎯→ Cd 0 + SO24 − + SH − + H +

(22)

While the thermodynamics of photodecomposition reactions are known in many cases (Gerischer 1980; Memming 2001), it can be hard to predict particle stability for arbitrary solution chemistry (Kormann et al. 1989).

Nanoparticle interactions with microorganisms The possibility of nanoparticle uptake. When biomineralization serves no known structural, navigational, or nutrient storage function, microorganisms almost universally attempt to restrict the precipitation of solids to the extracellular regions. The cell walls of both Gram positive and Gram negative bacteria (both negatively charged at circumneutral pH) present a considerable barrier to nanoparticle transport (Beveridge 1981; Fortin et al. 1997). However, as illustrated by Figure 18, nanoparticle precipitation frequently occurs to high densities in the periplasm. Presently, most studies on nanoparticle uptake are limited to mammalian cell lines, which in contrast to prokaryotic cells are able to acquire nanoparticles by endocytosis (Parak et al. 2002; Alivisatos et al. 2005). Mechanisms of prokaryotic nanoparticle uptake are (1) nonspeciic diffusion through membranes, (2) nonspeciic membrane damage-mediated uptake and (3) speciic uptake following membrane binding. Metal chalcogenide nanoparticles

144

Gilbert & Banęeld

Figure 18. Transmission electron microscopy (TEM) micrograph of a 70 nm section of a sulfate reducing bacterium showing periplasmic accumulation of metal sulide nanoparticles (dark contrast, labeled “ZnS”). Outer and inner cell membranes are marked by arrows and a mineral-free section of the periplasm is labeled “p.” Aqueous metal cations are likely brought into the periplasm as a result of weak chelation by organic acids (e.g. citrate and lactate) during nutrient uptake, where they precipitate with sulide ions produced by sulfate reduction. Image courtesy of Ken Williams; see Williams et al. 2005.

(such as CdSe) can be internalized within both Gram positive and Gram negative bacterial species via a pyrine-dependent mechanism provided that the surface is labeled with adenine or (AMP) and that the diameter of the coated nanoparticle is less than 5 nm (Kloepfer et al. 2005). As particle uptake exhibits both a dependence on surface label and light exposure, it is likely that a combination of mechanisms 2 and 3 above is responsible for the nanoparticle uptake. Nanoparticle uptake may facilitate gene transfer. It has been shown that the association of DNA with nanoscale precipitates of calcium phosphate minerals greatly enhanced the eficiency of gene transfer into human cell cultures over what was observed for aqueous DNA in the absence of minerals (Shen et al. 2004). It was unclear whether nanoparticles acted as a vector that carried DNA into the cell, or whether the DNA was released from the nanocomposite before uptake. Nevertheless, such studies indicate that extra-cellular biomineral-associated DNA has the potential for colloid-facilitated transport within the environment, while retaining the capacity for subsequent gene transfer.

Nanoparticle aggregation and its consequences The electron micrographs of Figure 1 suggest that biogenic nanoparticles are seldom present as individual particles, but exhibit a marked tendency for aggregation and deposition onto cell bodies. These processes are governed by particle-particle and particle-cell interaction forces that include electrostatic Coulombic forces (resulting from net charge on the nanoparticle surfaces) and short-range dispersion or van der Waals forces. The classical description of colloid stability—DVLO theory—has been intensely investigated for microscopic colloids (Verwey and Overbeek 1948; Lyklema 2001), although quantitative discrepancies remain, particularly at high ionic strength (Israelachvilli 1992; Boström et al. 2001). In particular, suspensions of colloids separated by net repulsive forces are generally found to be less stable against aggregation than predicted by DVLO theory. Furthermore, the validity of this description has not been established for nanoscale particles, for which a continuum description of the electrical double layer may not be appropriate. Nanoparticle transport and settling behavior is highly affected by the colloid properties of the individual particles, but remains a dificult process to quantify (Tiller and O’Melia 1993; Elimelich et al. 1995). Nanoparticle aggregation processes are beyond the scope of this chapter, so we simply state that numerous experimental studies have shown that mineral particles can be unstable against aggregation

Molecular-Scale Processes Involving Nanoparticulate Minerals

145

at circumneutral pH, and that the resulting aggregates commonly possess a complex interior porosity with the characteristics of geometric fractals (Meakin 1988; Dickinson 1989; Amal et al. 1990; Mylon et al 2004). Charge transport in aggregated nanoparticles. Aggregates of semiconductor nanoparticles have been shown to be surprisingly eficient at transporting electrons that have been excited to the conduction band (Fig. 11g) (Wang et al. 2003). This process is initiated particularly effectively by photoexcitation of surface sensitizer molecules (Fig. 11f, where A = second nanoparticle). When the rate of back transfer is slow, the lifetime of the excited electron is suficiently long to permit its transport to a CB of a neighboring nanoparticle. Particle-particle charge transfer separates the original electron and hole, which then persist until consumed in a chemical step, such as redox reaction or radical formation. Charge transfer between two nanoparticles occurs via a mechanism analogous to electron hopping between localized sites in a crystal. The analogy is quite good, because electronic states on the nanoparticle surfaces trap the electron between particle-particle transfer steps. Typical shallow electron traps on TiO2 surfaces are 25–50 meV below the CB (Hoffmann et al. 1995), so thermally activated transport is possible at room temperature (kBT | 25 meV at 25 °C). However, while a single activation energy barrier typically limits hopping between atomic sites in a crystal, there is generally a broad range of surface states on nanoparticles and hence a distribution of activation energies (Nelson 1999). There are now numerous measurements of the conductivity of nanoparticle aggregates. Theoretical descriptions of conductivity are based on percolation (for fractal aggregates; Avnir 1989) or diffusion models (for nonfractal compact aggregates; Nelson 1999). These approaches are not too sensitive to the details of the particle-particle transfer mechanism and can provide good agreement with experiment. The factors that determine whether electron transport occurs predominantly via nanoparticle surface or interior states remain unclear. This issue has been addressed with temperature-dependent conductivity measurements across arrays of magnetite nanoparticles. In one study on 5.5 nm diameter nanoparticles, a conductivity drop was seen below metalinsulator transition temperature, a clear sign of bulk-mineral-like conductivity (Poddar et al. 2002). In a second independent study of 14 nm particles, no temperature-dependent effect was observed, suggesting that surface-mediated conduction dominated (Redl et al. 2004). Signiicantly, many investigations of the conductivity through aggregated nanoparticles have been performed on dried layers of deposited particles. However, surface hydration can increase electron transport by orders of magnitude, even when direct conductivity through the liquid is not possible. As shown by Brus (1996), in a series of calculations for porous Si, electrons can couple to the dynamics of solvent molecules, which provides an activated intermediate state for transfer between particles. Charge transfer is considerably more favorable in a polar solvent than in air or vacuum because as a consequence of the solvent, the donor and acceptor energy levels do not have to be closely spaced. Exciton transport in aggregated nanoparticles. Long-range energy transfer (also known as Förster energy transfer) is an electrostatically mediated transfer of energy between two molecules or nanoparticles that have resonant excitation energies (Fig. 11h). Signiicant energy transfer is only possible if the acceptor has a fast-energy-loss pathway to trap the excitation that otherwise can transfer back to the donor. No activation energy is required, but diffusion of excitations is the dominant process for separations less than ~5 Å. Long-range energy has been demonstrated among aggregated nanoparticles (Kagan et al. 1996) and between nanoparticles and a surface (Achermann 2004), but it is presently not known to be signiicant in biogenic nanoparticle aggregates.

146

Gilbert & Banęeld CONCLUSIONS

Nanoparticles are integral constituents of water bodies, sediments, and soils, and many microorganisms that rely on an inorganic component to their metabolism inevitably interact with nanoscale minerals. Individual nanoparticles have the capacity to act as mobile redox active species that participate in molecular-style redox reactions. The redox potentials of valence and conduction band electrons, and the kinetics of charge transport and particle diffusion can exhibit a strong dependence on particle size. However, the susceptibility of minerals to dissolution upon valence change means that, in contrast to true molecular reagents, charged or photoexcited nanoparticles generally have mechanisms for transformation and energy loss that are not observed in molecular species. Biogenic nanoparticles form extended aggregates with complex structural, surface chemical and charge transport properties. The surfaces of nanoparticles and their aggregates generally exhibit strong afinity for aqueous nutrients, contaminants and organic biomolecules, including DNA, and can promote (bio)geochemical redox reactions, in some cases at signiicantly greater rates than observed at the surfaces of bulk minerals.

ACKNOWLEDGMENTS We thank Pupa de Stasio, Glenn Waychunas, Ken Nealson and Dan Hawkes for reviewing the manuscript, and Clara Chan, John Moreau, Yohey Suzuki, and Ken Williams for generously permitting reproduction of electron micrographs.

REFERENCES Abdelouas A, Lutze W, Gong WL, Nuttall EH, Strietelmeier BA, Travis BJ (2000) Biological reduction of uranium in groundwater and subsurface soil. Sci Total Environ 250:21-35 Achermann M, Petruska MA, Kos S, Smith DL, Koleske DD, Klimov VI (2004) Energy-transfer pumping of semiconductor nanocrystals using an epitaxial quantum well. Nature 429:642-646. Adams MA et al. (2003) Charge transfer on the nanoscale: Current status. J Phys Chem B 107:6668-6697 Alivisatos AP (1996) Perspectives on the physical chemistry of semiconductor nanocrystals J Phys Chem 100: 13226-13239 Alivisatos AP, Gu W, Larabell C (2005) Quantum dots as cellular probes. Annu Rev Biomed Eng 7:8.1-8.16 Almquist CB, Biswas P (2002) Role of the synthesis method and particle size of nanostructured TiO2 on its photoactivity. J Catalysis 212:145-156 Amal R, Raper JA, Waite TD (1992) Effect of fulvic acid adsorption on the aggregation kinetics and structure of hematite particles. J Colloid Interf Sci 151:244-257 Amonette JE, Workman DJ, Kennedy DW, Fruchter JS, Gorby YA (2000) Dechlorination of carbon tetrachloride by Fe(II) associated with goethite. Environ Sci Technol 34:4606-4613. Ashcroft NW, Mermin ND (1976) Solid State Physics. Brooks Cole Avnir D (1989) The Fractal Approach to Heterogeneous Chemistry: Surfaces, Colloids, Polymers. Wiley, Chichester Banield JF, Nealson KH (eds) (1997) Geomicrobiology: Interactions Between Microbes and Minerals. Rev Mineral Geochem Vol 35, Mineralogical Society of America, Washington, DC Banield JF, Navrotsky A (eds) (2001) Nanoparticles in the Environment. Rev Mineral Geochem Vol 44, Mineralogical Society of America, Washington, DC Banield JF, Welch SA, Zhang H, Ebert TT, Penn RL (2000). Aggregation-based crystal growth and microstructure development in natural iron oxyhydroxide biomineralization products. Science 289:751–754 Barbara PF, Meyer TJ, Ratner MA (1996) Contemporary issues in electron transfer research. J Phys Chem 100: 13148-13168 Bard AJ, Fox MA (1995) Artiicial photosynthesis – solar splitting of water to hydrogen and oxygen. Acc Chem Res 28:141-145 Bargar JR, Tebo BM, Villinski JE (2000) In situ characterization of Mn(II) oxidation by spores of the marine Bacillus sp. strain SG-1. Geochim Cosmochim Acta 64:2775-2778 Bargar JR, Tebo BM, Bergmann U, Webb SM, Glatzel P, Chiu VQ, Villalobos M. (2005) Biotic and abiotic products of Mn(II) oxidation by spores of the marine Bacillus sp. strain SG-1. Am Mineral 90:143-154

Molecular-Scale Processes Involving Nanoparticulate Minerals

147

Bazylinski DA, Frankel RB (2004) Magnetosome formation in prokaryotes. Nature Rev Microbio 2:217-230 Bebie J, Schoonen MAA, Fuhrmann M, Strongin DR (1998) Surface charge development on transition metal sulides: an electrokinetic study. Geochim Cosmochim Acta 62:633-642 Bekyarova E, Fornasiero P, Kaspar J, Graziani M (1998) CO oxidation on Pd/CeO2–ZrO2 catalysts. Catal Today 45:179-183 Belin T, Millot N, Villieras F, Bertrand O, Bellat JP (2004) Structural variations as a function of surface adsorption in nanostructured particles. J Phys Chem B 108:5333-5340 Benner SG, Hansel CM, Wielinga BW, Barber TM, Fendorf S (2002) Reductive dissolution and biomineralization of iron hydroxide under dynamic low conditions. Environ Sci Technol 36:1705–1711 Benning L, Wilkin RT, Barnes HL (2000) Reaction pathways in the Fe-S system below 100 °C. Chem Geol 167:25-51 Berhault G, Lacroix M, Breysse M, Maugé F, Lavalley JC, Nie H, Qu L (1998) Characterization of acidic sites of silica-supported transition metal sulides by pyridine and 2,6 dimethylpyridine adsorption: relation to activity in CH3SH condensation. J Catal 178:555–565 Beveridge TJ (1981) Ultrastructure, chemistry and function of the bacterial cell wall. Int Rev Cytol 72:229-317 Billinge SJL, Thorpe MF (eds) (1998) Local Structure from Diffraction. Plenum Press, New York Bolton JL, Trush MA, Penning TM, Dryhurst G, Monks TJ (2000) Role of quinones in toxicology. Chem Res Toxicol 13:135-160 Bourg ACM, Bertin C (1996) Diurnal variations in the water chemistry of a river contaminated by heavy metals: Natural biological cycling and anthropic inluence. Water Air Soil Poll 86:101-116 Boström M, Williams DRM, Ninham BW (2001) Speciic ion effects: Why DLVO theory fails for biology and colloid systems. Phys Rev Lett 87:168103 Bratina BJ, Stevenson BS, Green WJ, Schmidt TM (1998) Manganese reduction by microbes from oxic regions of the Lake Vanda (Antarctica) water column. Appl Environ Microbiol 64:3791-3797 Brown GE Jr., Foster AL, Ostergren JD (1999) Mineral surfaces and bioavailability of heavy metals: a molecular scale perspective. Proc Natl Acad Sci 96:3388-3395 Brus LEA (1983) Simple model for the ionization potential, electron afinity, and aqueous redox potentials of small semiconductor crystallites. J Chem Phys 79:5566-5571 Brus LEA (1984) Electron-electron and electron-hole interactions in small semiconductor crystallites: The size dependence of the lowest excited electronic state. J Chem Phys 80:4403-4409 Brus LEA (1996) Model for carrier dynamics and photoluminescence quenching in wet and dry porous silicon thin ilms. Phys Rev B 53:4649-4656 Bryan JD, Heald SM, Chambers SA, Gamelin DR (2004) Strong room-temperature ferromagnetism in Co2+doped TiO2 made from colloidal nanocrystals. J Am Chem Soc 126:11540-11647 Büker K, Fiechter S, Eyert V, Tributsch H (1999) Photochemistry of highly Zn-doped pyrite as compared with isostructural FeS2. J Electrochem Soc 146:261-265 Burns RG, Burns RM (1979) Marine minerals. Rev Mineral 6:1-46 Car R, Parrinello M (1985) Uniied approach for molecular dynamics and density functional theory. Phys Rev Lett 55:2471-2473 Cartoixa X, Wang LW (2005) Microscopic dielectric response functions in semiconductor quantum dots. Phys Rev Lett 94:236804 Casey, WH, Swaddle, TW (2003) Why small? The use of small inorganic clusters to understand mineral surface and dissolution reactions in geochemistry. Rev Geophys 41:1-20 Cheah SF, Kraemer SM, Cervini-Silva J, Sposito G (2003) Steady state dissolution kinetics of goethite in the presence of Desferrioxamine-B and oxalate ligands: Implications for the microbial acquisition of iron. Chem Geol 198:63–75 Chen W, Wang ZG, Lin ZJ, Lin LY (1997) Absorption and luminescence of the surface states in ZnS nanoparticles. J Appl Phys 82:3111-3115 Cheng L, Fenter P, Nagy KL, Schlegel ML, Sturchio MC. (2001) Molecular-scale density oscillations in water adjacent to a mica surface. Phys Rev Lett 87:156103 Cornell RM, Schwertmann U. (1979) Inluence of organic anions on the crystallization of ferrihydrite. Clays Clay Miner 27:402-410 Cornell RM, Schwertmann U (2003) The Iron Oxides: Structure, Properties, Reactions, Occurrences and Uses. Wiley-VCH,Weinheim Cox PA (1995) Transition Metal Oxides: An Introduction to Their Electronic Structure and Properties. Clarenden Press Davis KJ, Dove PM, De Yoreo JJ (2000) The role of Mg2+ as an impurity in calcite growth. Science 290:11341137 Delerue C, Allan G, Lannoo M (2001) Electron-phonon coupling and optical transitions for indirect-gap semiconductor nanocrystals. Phys Rev B 64:193402 De Yoreo JJ, Vekilov PG (2003) Principles of nucleation and growth. Rev Mineral Geochem 54:57-93

148

Gilbert & Banęeld

Dickinson E (1989) Structure of simulated colloidal deposits. Colloids Surf 39:143-159 Dos Santos Afonso M, Stumm W (1992) Reductive dissolution of iron(III) (hydr)oxides by hydrogen sulide. Langmuir 8:1671-1675 Douglas S, Beveridge TJ (1998) Mineral formation by bacteria in natural microbial communities. FEMS Microbiol Ecol 26:79-88 Drits VA, Tchoubar C (1990) X-ray Diffraction by Disordered Lamellar Structures: Theory and Applications to Microdivided Silicates and Carbons. Springer Verlag, Berlin Duonghong D, Ramsden J, Grätzel M (1982) Dynamics of interfacial electron-transfer processes in colloidal semiconductor systems. J Am Chem Soc 104:2977-2985 Dwarakanath S, Bruno JG, Shastry A, Phillips T, John A, Kumar A, Stephenson LD (2004) Quantum dotantibody and aptamer conjugates shift luorescence upon binding bacteria. Biochem Biophys Res Commun 325:739–743 Edelbro R, Sandström A, Paul J (2003) Full potential calculations on the electronic bandstructure of sphalerite, pyrite and chalcopyrite. Appl Surf Sci 206:300-313 Edwards KJ, Bond PJ, Gihring TM, Banield JF. (2000) An archaeal iron-oxidizing extreme acidophile important in acid mine drainage. Science 287:1796-1799 Efros AL, Rosen M (2000) The electronic structure of semiconductor nanocrystals. Annu Rev Mater Sci 30: 475–521 Ehrlich HL (2002) Geomicrobiology. Marcel Dekker, New York Elimelech M, Gregory J, Jia X, Williams R (1995) Particle Deposition and Aggregation: Measurement, Modelling and Simulation. Butterworth Heinemann, Oxford Elsner M, Schwarzenbach RP, Haderlein SB (2004) Reactivity of Fe(II)-bearing minerals to reductive transformation of organic contaminants. Environ Sci Technol 38:799-807 Ennaoui A, Tributsch H (1986) Light-induced electron transfer and photoelectrocatalysis of chlorine evolution at FeS2 electrodes. J Electroanal Chem 205:185-195 Eng PJ, Trainor TP, Brown GE, Waychunas GA, Newville M, Sutton SR, Rivers ML (2000) Structure of the hydrated alpha-Al2O3 (0001) surface. Science 288:1029-1033 Oberdorster G, Oberdorster E, Oberdorster J (2005) Nanotoxicology: An emerging discipline evolving from studies of ultraine particles. Environ Health Perspect 113:823-839 Erwin SC, Zu L, Haftel MI, Efros AL, Kennedy TA, Norris DJ (2005) Doping semiconductor nanocrystals. Nature 436:91-95 Ferreyra JM, Proetto CR (1999) Quantum size effects on excitonic Coulomb and exchange energies in initebarrier semiconductor quantum dots. Phys Rev B 60:10672-10675 Ferris FG (2005) Biogeochemical properties of bacteriogenic iron oxides. Geomicrobiology J 22:79-85 Fistul VI (2004) Impurities in Semiconductors : Solubility, Migration, and Interactions. CRC Press, Boca Raton, Florida Fortin D, Ferris FF, Beveridge TJ (1997) Surface-mediated mineral development by bacteria. Rev Mineral Geochem 35:161–180 Fowler TA, Holmes PR, Crundwell FK (1999) Mechanism of pyrite dissolution in the presence of Thiobacillus ferrooxidans. Appl Environ Microbiol 65:2987-2993 Franceschetti A, Zunger A (1997) Direct pseudopotential calculations of exciton Coulomb and exchange energies in semiconductor quantum dots. Phys Rev Lett 78:915-917 Franceschetti A, Williamson A, Zunger A (2000) Addition spectra of quantum dots: The role of dielectric mismatch. J Phys Chem B 104:3398-3401 Frankel RB, Bazylinski DA Biologically induced mineralization by bacteria. (2003) Rev Mineral Geochem 54: 95-114 Fredrickson JK, Zachara JM, Kennedy DW, Duff MC, Gorby YA (2000) Reduction of U(VI) in goethite (D-FeOOH) suspensions by a dissimilatory metal-reducing bacterium. Geochim Cosmochim Acta 64: 3085-3098 Fredrickson JK, Kota S, Kukkadapu RK, Liu C, Zachara JM (2003) Inluence of electron donor/acceptor concentrations on hydrous ferric oxide (HFO) bioreduction. Biodegradation 14:91–103 Fukushi K, Sato T (2005) Using a surface complexation model to predict the nature and stability of nanoparticles. Environ Sci Technol 39:1250-1256 Fujimori A, Saeki M, Kimizuka N, Taniguchi M, Suga S (1986) Photoemission satellites and electronic structure of D-Fe2O3. Phys Rev B 34:7318-7327 Fujita E, DuBois DL (2003) Carbon dioxide ixation. In: Photoconversion of Solar Energy Photochemical and Photoelectrochemical Approaches to Solar Energy Conversion. Archer MD, Nozik AJ (eds) Imperial College Press, London Fujiwara H, Kanemoto M, Ankyu H, Murakoshi K, Wada Y, Yanagida S (1997) Visible-light induced photoixation of carbon dioxide into aromatic ketones and benzyl halides catalyzed by CdS nanocrystals. J Chem Soc Perkin Trans 2:317-321

Molecular-Scale Processes Involving Nanoparticulate Minerals

149

Fujiwara H, Hosokawa H, Murakoshi K, Wada Y, Yanagida S (1998) Surface Characteristics of ZnS nanocrystallites relating to their photocatalysis for CO2 reduction. Langmuir 14:5154-5159 Galli G (2005) Doping the undopable. Nature 436:32-33 Galli G, Parrinello M (1992) Large scale electronic structure calculations. Phys Rev Lett 69:3547-3550 Gaponenko SV (1998) Optical Properties of Semiconductor Nanocrystals. Cambridge University Press, Cambridge. Garrett BC et al. (2005) Role of water in electron-initiated processes and radical chemistry: Issues and scientiic advances. Chem Rev 105:355−389 Gerischer S (1980) Photodecomposition of semiconductors: Thermodynamics, kinetics and application to solar cells. Faraday Disc Chem Soc 70:137-151 Gerischer S (1993) Photoelectrochemical catalysts of the oxidation of organic molecules by oxygen on small semiconductor particles with TiO2 as an example. Electrochim Acta 38:3-9 Gilbert B, Huang F, Zhang HZ, Waychunas GA, Banield JF (2004) Nanoparticles: Strained and stiff. Science 305:651-654 Gilbert B, Frazer BH, Belz A, Conrad PG, Nealson KH, Haskel D, Lang JH, Srajer G, De Stasio G (2003) Multiple scattering calculations of bonding and X-ray absorption spectroscopy of manganese oxides. J Phys Chem A 107:2839-2847 Gratzel M, Frank A (1982) Interfacial electron-transfer reactions in colloidal semiconductor dispersions. Kinetic analysis. J Phys Chem 86:2964-2967 Green M, O’Brien P (1999) Recent advances in the preparation of semiconductors as isolated nanometric particles: new routes to quantum dots. Chem Commun 22:2235-2241 Guo J (2004) Synchrotron radiation, soft-X-ray spectroscopy and nanomaterials. Int J Nanotechnol 1:193-225 Gunnars A, Blomqvist S, Johansson P, Andersson C (2002) Formation of Fe(III) oxyhydroxide colloids in freshwater and brackish seawater, with incorporation of phosphate and calcium. Geochim Cosmochim Acta 66:745-758 Haack EA, Warren LA (2003) Bioilm hydrous manganese hydroxides and metal dynamics in acid rock drainage. Environ Sci Technol 37:4138-4147 Hall EJ (2000) Radiobiology for the Radiologist. Lippincott Williams and Wilkins Hansel CM, Benner SG, Neiss J, Dohnalkova A, Kukkadapu RK, Fendorf S (2003) Secondary mineralization pathways induced by dissimilatory iron reduction of ferrihydrite under advective low Geochim Cosmochim Acta 67:2977-2992 Hansel CM, Benner SG, Nico P, Fendorf S (2004) Structural constraints of ferric (hydr)oxides on dissimilatory iron reduction and the fate of Fe(II). Geochim Cosmochim Acta 68:3217-3229 Haram SK, Quinn BN, Bard AJ (2001) Electrochemistry of CdS nanoparticles: A correlation between optical and electrochemical band gaps. J Am Chem Soc 123:8860-8861 Henderson MA (2002) The interaction of water with solid surfaces: Fundamental aspects revisited. Surf Sci Rep 46:5-308 Hering JG, Stumm W (1990) Oxidative and reductive dissolution of minerals. Rev Mineral 23:427-464 Hill NA, Whaley KB (1995) Size dependence of excitons in silicon nanocrystals. Phys Rev Lett 75:1130-1133 Hochella MF Jr., White AF (eds) (1990) Mineral-water interface geochemistry. Reviews in Mineralogy Vol 23, Mineralogical Society of America, Washington DC Hochella MF Jr., Moore JN, Putnis C, Putnis A, Kasama T, Eberl DD (2005a) Direct observation of toxic metal-mineral association from a massive acid mine drainage system: Implications for metal transport and bioavailability. Geochem Cosmochim Acta 69:1651-1663 Hochella MF Jr., Kasama T, Putnis A, Putnis C, Moore JN (2005b) Environmentally important, poorly crystalline Fe/Mn hydrous oxides: Ferrihydrite and a vernadite-like mineral from a massive acid mine drainage system. Am Mineral 90:718-724 Huber R, Spörlein S, Moser JE, Grätzel M (2000) The role of surface states in the ultrafast photoinduced electron transfer from sensitizing dye molecules to semiconductor colloids. J Phys Chem B 104:8995-9003 Inoue H, Moriwaki H, Maeka K, Yoneyam H (1995) Photoreduction of carbon-dioxide using chalcogenide semiconductor nanocrystals. J Photochem Photobiol A 86:191-196 Ireta J, Galván M, Cho K, Joannopoulos JD (1998) Local reactivity of charybdotoxin, a K+ channel blocker. J Am Chem Soc 120:9771-9778 Isaure MP, Manceau A, Tamura NO, Marcus MA (2005) Zinc mobility and speciation in soil covered by contaminated dredged sediment using micrometer-scale and bulk-averaging X-ray luorescence, absorption and diffraction techniques. Geochim Cosmochim Acta 69:1173-1198 Israelachvili J (1992) Intermolecular and Surface Forces. Academic Press, San Diego. Iwamato M, Abe T, Tachibana Y (2000) Control of bandgap of iron oxide through its encapsulation into SiO2based mesoporous materials. Hoffmann MR, Martin ST, Choi W, Bahneman DW (1995) Environmental applications of semiconductor photocatalysis. Chem Rev 95:69-96

150

Gilbert & Banęeld

Jones F, Cölfen H, Antonietti M (2000) Iron oxyhydroxide colloids stabilized with polysaccharides. Colloid Polym Sci 278:491-501 Kagan CR, Murray CB, Bawendi MG (1996) Long-range resonance transfer of electronic excitations in closepacked CdSe quantum-dot solids. Phys Rev B 54:8633-8643 Kennedy CB, Scott SD, Ferris FG (2004) Hydrothermal phase stabilization of 2-line ferrihydrite by bacteria. Chemical Geology 212:269-277 Kennedy CB, Scott SD, Ferris FG (2003a) Characterization of bacteriogenic iron oxide: Deposits from Axial Volcano, Juan de Fuca Ridge, Northeast Paciic Ocean. Geomicrobiol J 20:199-214 Kennedy CB, Scott SD, Ferris FG (2003b) Ultrastructure and potential sub-sealoor evidence of bacteriogenic iron oxides from Axial Volcano, Juan de Fuca Ridge, north-east Paciic Ocean. FEMS Microbiol Ecol 43: 247-254 Kerisit S, Cooke DJ, Spagnoli D, Parker SC (2005) Molecular dynamics simulations of the interactions between water and inorganic solids. J Mater Chem 15:1454-1462 Kimball BA, Callender E, Axtmann EV (1995) Effects of colloids on metal transport in a river receiving acidmine drainage, Upper Arkansas River, Colorado, USA. Appl Geochem 10:285-306 Kloepfer JA, Mielke RE, Nadeau JL (2005) Uptake of CdSe and CdSe/ZnS quantum dots into bacteria via purine-dependent mechanisms. Appl Environ Microbiol 71:2548-2557 Konhauser KO (1998) Diversity of bacterial iron mineralization. Earth Sci Rev 43:91-121 Konovalova TA, Konovalov VV, Kispert LD (1999) Surface modiication of TiO2 nanoparticles with carotenoids. EPR Study. J Phys Chem B 103:4672-4699 Konovalova TA, Lawrence J, Kispert LD (2004) Generation of superoxide anion and most likely singlet oxygen in irradiated TiO2 nanoparticles modiied by carotenoids. J Photochem Photobiol A 162:1–8 Kormann C, Bahnemann DW, Hoffmann MR (1989) Environmental photochemistry: Is iron oxide (hematite) an active photocatalyst? A comparative study: D-Fe2O3, ZnS, TiO2. J Photochem Photobiol A 48:161-169 Kormann C, Bahnemann DW, Hoffmann MR (1991) Photolysis of chloroform and other organic molecules in aqueous TiO2 suspensions. Environ Sci Technol 25:494-500 Kraemer SM (2004) Iron oxide dissolution and solubility in the presence of siderophores. Aquati Sci 66;3-88 Kucur E, Riegler J, Urban GA, Nann T (2003) Determination of quantum coninement in CdSe nanocrystals by cyclic voltammetry. J Chem Phys 119:2333-2337 Labrenz M, Druschel GK, Thomsen-Ebert T, Gilbert B, Welch SA, Kemner KM, Logan GA, Summons RE, De Stasio G, Bond PL, Lai B, Kelly SD, Banield JF (2000) Sphalerite (ZnS) deposits forming in natural bioilms of sulfate-reducing bacteria. Science 290:1744-1747 Landolt-Bernstein (1983) Numerical data and functional relationships in science and technology, New Series Springer, Berlin Langmuir D (1996) Aqueous Environmental Geochemistry. Prentice Hall Lawless D, Sepone N, Meisel D (1991) Role of OH• radicals and trapped holes in photocatalysis. A pulse radiolysis study. J Phys Chem 95:5166-5170 Leland JK, Bard AJ (1987) Photochemistry of colloidal semiconducting iron oxide polymorphs. J Phys Chem 91:5076-5083 Lewis EA, Smith JRL, Walton PH, Archibald SJ, Foxton SP, Giblin GMP (2001) Tuning the metal-based redox potentials of manganese cis, cis-1,3,5-triaminohexane complexes. J Chem Soc Dalton Trans 2001:11591161 Lippens PE, Lannoo M (1989) Calculations of band gap for small CdS and ZnS crystallites. Phys Rev B 39: 10935-10941 Lodziana Z, Topsoe NY, Norskov JK (2004) A negative surface energy for alumina. Nature Mater 3:289-293 Lovely DR (1987) Organic matter mineralization with the reduction of ferric iron: A review. Geomicrobiol J 5: 375-399 Lovley DR, Stolz JF, Nord GL, Phillips EJP (1987) Anaerobic production of magnetite by a dissimilatory ironreducing microorganism. Nature 330:252-254 Lovely DR, Phillips EJP, Gorby YA, Landa ER (1991) Microbial reduction of uranium. Nature 350:413-416 Lovely DR (1993) Dissimilatory metal reduction. Ann Rev Microbiol 47:263-290 Lovely DR (1997) Microbial Fe(III) reduction in subsurface environments. FEMS Microbiol Rev 20:305-313 Lucas E, Decker S, Khaleel A, Seitz A, Fultz S, Ponce A, Carnes C, Klabunde KJ (2001) Nanocrystalline metal oxides as unique chemical reagents/sorbents. Chem Eur J 7:2505-2510 Lüning J, Rockenberger J, Eisbitt S, Rubensson JE, Karl A, Kornowski A, Weller H, Eberhardt W (1999) Soft X-ray spectroscopy of single size CdS nanocrystals: Size coninement and electronic structure. Solid State Commun 112:5-9 Lyklema J (2001) Fundamentals of Interface and Colloid Science Volume II: Solid-Liquid Interfaces. Academic Press, San Diego Madden AS, Hochella MF Jr. (2005) A test of geochemical reactivity as a function of mineral size: Manganese oxidation promoted by hematite nanoparticles. Geochim Cosmochim Acta 69:389-398

Molecular-Scale Processes Involving Nanoparticulate Minerals

151

Manceau A, Drits, VA Silvester E, Bartoli C, Lanson B (1997) Structural mechanism of Co2+ oxidation by the phyllomanganate buserite. Am Mineral 82:1150-1175 Manceau A, Tommaseo C, Rihs S, Geoffroy N, Chateigner D, Schlegel M, Tisserand D, Marcus MA, Tamura M, Chen ZS (2005) Natural speciation of Mn, Ni, and Zn at the micrometer scale in a clayey paddy soil using X-ray luorescence, absorption, and diffraction. Goechim Cosmochim Acta 69:4007-4034 Mann S (2000) Biomineralization. Oxford University Press, Oxford Marcus MA (1993) Electron transfer reactions in chemistry. Theory and experiment. Rev Mod Phys 65:599610 Mayers IT, Beveridge TJ (1989) The sorption of metals to Bacillus subtilis walls from dilute solutions and simulated Hamilton Harbour (Lake Ontario) water. Can J Microbiol 35:764-770 McCormick ML, Bouwer EJ, Adriaens (2002) Carbon tetrachloride transformation in a model iron-reducing culture: Relative kinetics of biotic and abiotic reactions. Environ Sci Technol 36:403-410 McCormick ML, Adriaens P (2004) Carbon tetrachloride transformation on the surface of nanoscale biogenic magnetite nanoparticles. Environ Sci Technol 38:1045-1053 McKenzie KJ, Marken F (2001) Direct electrochemistry of nanoparticulate Fe2O3 in aqueous solution and adsorbed onto tin-doped indium oxide. Pure Appl Chem 73:1885-1894 Meakin P (1988) Fractal aggregates. Adv Colloid Interface Sci 28:249-331 Memming R (2001) Semiconductor Electrochemistry. Wiley-VCH, Winheim Missana T, Mafiotte C, Garcîa-Gutiérreza M (2003) Surface reaction kinetics between nanocrystalline magnetite and uranyl. J Colloid Interf Sci 261:154-160 Monticone S, Tufeu R, Kanaev AV, Scolan E, Sanchez C (2000) Quantum size effect on TiO2 nanoparticles: Does it exist? Appl Surf Sci 162-163:565-570 Moreau JW, Webb RI, Banield JF (2004) Ultrastructure, aggregation state and crystal growth of biogenic nanocrystalline sphalerite and wurtzite. Am Mineral 89:950-960 Morrison SR (1980) Electrochemistry at Semiconductor and Oxidized Metal Interfaces. Plenum Press, New York Moser J, Punchihewa S, Infelta PP, Gratzel M (1991) Surface complexation of colloidal semiconductors strongly enhances interfacial electron-transfer rates. Langmuir 7:3012-3018 Moser CC, Keske JM, Warncke K, Farid RS, Dutton PL (1992) Nature of biological electron transfer. Nature 355:796-802 Müller BR, Majoni S, Memming R, Meissner D (1997) Particle size and surface chemistry in photoelectrochemical reactions at semiconductor particles. J Phys Chem B 101:2501-2507 Mulvaney P, Cooper R, Grieser F, Meisel D (1988) Charge trapping in the reductive dissolution of colloidal suspensions of iron(III) oxides. Langmuir 4:1206-1211 Murray JW, Dillard JG (1979) The oxidation of cobalt(II) adsorbed on manganese dioxide. Geochim Cosmochim Acta 43:781-787 Murray CB, Kagan CR, Bawendi MG (2000) Synthesis and characterization of monodisperse nanocrystals and close-packed nanocrystal assemblies. Annu Rev Mater Sci 30:545-610 Mylon SE, Chem KL, Elimelech M (2004) Inluence of natural organic matter and ionic composition on the kinetics and structure of hematite colloid aggregation: Implications to iron depletion in estuaries. Langmuir 20:9000-9006 Nanda KK, Kruis FE, Fissan H, Behera SN (2004) Effective mass approximation for two extreme semiconductors: Band gap of PbS and CuBr nanoparticles. J Appl Phys 95:5035-5041 Navrotsky A (2004) Energetic clues to pathways to biomineralization: Precursors, clusters and nanoparticles. Proc Natl Sci Acad 101:12096-12101 Nedeljkovic JM, Nenadovic MT, Micic OI, Nozik AJ (1986) Enhanced photoredox chemistry in quantized semiconductor colloids. J Phys Chem 90:12-13 Nelson J (1999) Continuous-time random walk model of electron transport in nanocrystalline TiO2 electrodes. Phys Rev B 59:15374-15380 Newman DK, Kolter R (2000) A role for excreted quinines in extracellular electron transfer. Nature 405:94-97 Nevin KP, Lovley DR (2000) Potential for nonenzymatic reduction of Fe(III) via electron shuttling in subsurface sediments. Environ Sci Technol 34:2472-2478 Noguera C, Pojani A, Casek P, Finocchi F (2002) Electron redistribution in low-dimensional oxide structures. Surface Scie 507-510:245-255 Oberdorster G, Oberdorster E, Oberdorster J (2005) Nanotoxicology: An emerging discipline evolving from studies of ultraine particles. Environ Heath Perspect 113:823-839 O’Connor MV, Sposito G (2004) Investigation of biogenic manganese oxide by density functional theory. Abstr Papers Am Chem Soc 228:U548-549 043-ENVR O’Loughlin EJ, Kelly SD, Cook RE, Csencsits R, Kemner KM (2003) Reduction of uranium(VI) by mixed iron(II/iron(III) hydroxide (green rust): Formation of UO2 nanoparticles. Environ Sci Technol 37:721-727

152

Gilbert & Banęeld

Palosz B, Grzanka E, Gierlotka S, Stel’Mahk S, Pielaszek R, Bismayer U, Neuefeind, J, Weber HP, Palosz W (2002) Diffraction studies of nanocrystals: Theory and experiment. Acta Physica Polonica 102:57-82 Page CC, Moser CC, Chen X, Dutton L (1999) Natural engineering principles of electron tunnelling in biological oxidation-reduction. Nature 402:47-52 Parak WJ, Boudreau R, Le Gros M, Gerion D, Zanchet D, Micheel CM, Williams SC, Alivisatos AP, Larabell C (2002) Cell motility and metastatic potential studies based on quantum dot imaging of phagokinetic tracks. Adv Mater 14:882-885 Patterson AL (1939) The Scherrer formula for X-ray particle size determination. Phys Rev 56:978-982 Park H, Choi W (2005) Photocatalytic conversion of benzene to phenol using modiied TiO2 and polyoxometalates. Catalysis Today 101:291-297 Parak WJ, Boudreau R, Le Gros M, Gerion D, Zanchet D, Micheel CM, Williams SC, Alivisatos AP, Larabell C (2002) Cell motility and metastatic potential studies based on quantum dot imaging of phagokinetic tracks. Adv Mater 14:882-885 Park JH, Tjeng LH, Allen JW, Metcalf P, Chen CT (1997) Single-particle gap above the Verwey transition in Fe3O4. Phys Rev B 57:12813-12817 Peak D (2005) In situ ATR-FTIR spectroscopic studies of the kinetics of oxyanion/metal oxide adsorption. Abstr Papers Am Chem Soc 228:U900 113-GEOC Pecher K, Haderlein SB, Schwartzenbach RP (2002) Reduction of polyhalogenated methanes by surfacebound Fe(II) in aqueous suspensions of iron oxides. Environ Sci Technol 36:1734-1741 Pehkonen SO, Seifert RL, Hoffman MR (1995) Photoreduction of iron oxyhydroxide and the photooxidation of halogenated acetic acids Environ Sci Technol 29:1215-1222 Pellegrini G, Mattei G, Mazzoldi P (2005) Finite depth square well model: Applicability and limitations. J Appl Phys 97:073706 Penn RL, Banield JF (1998) Oriented attachment and growth, twinning, polytypism, and formation of metastable phases: Insights from nanocrystalline TiO2. Am Mineral 83:1077-1082 Penn RL, Zhu C, Xu H, Veblen DR. (2001) Iron oxide coatings on sand grains from the Atlantic coastal plain: High-resolution transmission electron microscopy characterization. Geology 29:843-846 Phoenix VR, Konhauser KO, Adams DG, Bottrell SH (2001) Role of biomineralization as an ultraviolet shield: Implications for Archean life. Geology 29:823-826 Poddar P, Fried T, Markovich G (2002) First-order metal-insulator transition and spin-polarized tunneling in Fe3O4 nanocrystals. Phys Rev B 65:172405 Pokrant S, Whaley KB (1999) Tight-binding studies of surface effects on electronic structure of CdSe nanocrystals: The role or organic ligands, surface reconstruction and inorganic capping shells. Eur J Phys 6:255-267 Poulton SW, Krom MD, Raiswell R (2004) A revised scheme for the reactivity of iron (oxyhydr)oxide minerals towards dissolved sulide. Geochim Cosmochim Acta 68:3703-3715 Preisinger M, Krispin M, Rudolf T, Horn S, Strongin DR (2005) Electronic structure of nanoscale iron oxide particles measured by scanning tunneling and photoelectronic spectroscopies. Phys Rev B 71:165409 Puzder A, Williamson AJ, Reboredo FA, Galli G (2003) Structural stability and optical properties of nanomaterials with reconstructed surfaces. Phys Rev Lett 91:157405 Qadri SB, Skelton EF, Dinsmore AD, Yang J, Gray HF, Ratna BR (1999) Size-induced transition-temperature reduction in nanoparticles of ZnS. Phys Rev B 60:9191-9193 Quantin C, Becquer T, Rouiller JH, Berthelin J (2001) Oxide weathering and trace metal release by bacterial reduction in a New Caledonia Ferralsol. Biogeochemistry 53:323-340 Rabani E, Hetenyi B, Berne BJ (1999) Electronic properties of CdSe nanocrystals in the absence and presence of a dielectric medium. J Chem Phys 110:5355-5368 Rabani E (2001) Structure and electrostatic properties of passivated CdSe nanocrystals. J Chem Phys 115: 1493-1497 Rajh T, Chen LX, Lukas K, Liu T, Thurnauer MC, Tiede DM (2002) Surface restructuring of nanoparticles: An eficient route for ligand-metal oxide crosstalk. J Phys Chem B 106:10543- 10552 Rajh T, Saponjic Z, Liu JQ, Dimitrijevic NM, Scherer NF, Vega-Arroyo M, Zapol P, Curtiss LA, Thurnauer MC (2004) Charge transfer across the nanocrystalline-DNA interface: Probing DNA recognition. Nano Lett 4:1017-1023 Rancourt DG (2001) Magnetism of earth, planetary and environmental nanomaterials. Rev Mineral Geochem 44:217–292 Rancourt DG, Thibault PJ, Mavrocordatos D, Lamarche G (2005) Hydrous ferric oxide precipitation in the presence of nonmetabolizing bacteria: Constraints on the mechanism of a biotic effect. Geochim Cosmochim Acta 69:553-577 Redl FX, Black CT, Papaefthymiou GC, Sandstrom RL, Yin M, Zeng H, Murray CB, O’Brien SP (2004) Magnetic, electronic, and structural characterization of nonstoichiometric iron oxides at the nanoscale. J Am Chem Soc 126:14583-14599

Molecular-Scale Processes Involving Nanoparticulate Minerals

153

Reguera G, McCarthy KD, Mehta T, Nicoll JS, Tuominen MT, Lovley DR (2005) Extracellular electron transfer via microbial nanowires. Nature 435:1098-1101 Roden EE, Zachara JM (1996) Microbial reduction of crystalline iron (III) oxides: inluence of oxide surface area and potential for cell growth. Environ Sci Technol 30:1618-1628 Roden EE, Urrutia MM, Mann, CJ (2000) Bacterial reductive dissolution of crystalline Fe(III) oxide in continuous-low column reactors. Appl Environ Microbiol 66:1062-1065 Rodriguez JA, Chaturvedi S, Kuhn M, Hrbek J (1998) Reaction of H2S and S2 with metal/oxide surfaces: Bandgap size and chemical reactivity. J Phys Chem B 102:5511-5519 Rosso KM, Smith DMA, Dupuis M (2003) An ab intio model of electron transport in hematite (D-Fe2O3) basal planes. J Chem Phys 118:6455-6466 Rozan TF, Lassman ME, Ridge DP, Luther GW (2000) Evidence for iron, copper and zinc complexation as multinuclear sulphide clusters in oxic rivers. Nature 406:879-882 Rustad JR, Felmy AR (2005) The inluence of edge sites on the development of surface charge on goethite nanoparticles: A molecular dynamics investigation. Geochim Cosmochim Acta 69:1405-1411 Scholz F, Meyer B (1998) Voltammetry of solid microparticles immobilized on electrode surfaces. Electroanal Chem 20:1-86 Schooss D, Mews A, Eychmüller A, Welller H (1994) Quantum-dot quantum well CdS/HgS/CdS: Theory and experiment. Phys Rev B 49:17072-17078 Schüring J, Schulz HD, Fischer WR, Böttcher J, Duijnisveld WHM (2000) Redox: Fundamental Processes and Applications. Springer-Verlag, Berlin Schwertmann U, Gasser U, Sticher H (1989) Chromium-for-iron substitution in synthetic goethites. Geochim Cosmochim Acta 53:1293-1297 Schwertmann U, Pfab G (1994) Structural vanadium in synthetic goethite. Geochim Cosmochim Acta 58: 4349-4352 Schwertmann U, Friedl J, Stanjek, H (1999) From Fe(III) ions to ferrihydrite and then to hematite. J Colloid Interf Sci 209:215-223 Scott SD, Barnes HL (1972) Sphalerite-wurtzite equilibria and stoichiometry. Geochim Cosmochim Acta 36: 1276-1295 Scott DT, McKnight DM, Voelker BM, Hrncir DC (2002) Redox processes controlling manganese fate and transport in a mountain stream. Environ Sci Technol 36:453-459 Serpone N, Lawless D, Khairutdinovt R (1995) Size Effects on the photophysical properties of colloidal anatase TiO2 particles: Size quantization or direct transitions in this indirect semiconductor? J Phys Chem 99:16646-16654 Shen H, Tan J, Saltzmann WM (2004) Surface-meditated gene transfer from nanocomposites of controlled texture. Nature Materials 3:569-574 Sherman DM (1984) Electronic structure of manganese oxide minerals. Am Mineral 69:788-799 Sherman DM (1985) Electronic structures of Fe3+ coordination sites in iron oxides: applications to spectra, bonding and magnetism. Phys Chem Mineral 12:161–175 Sherman DM (2005) Electronic structures of iron (III) and manganese (IV) (hydr)oxide minerals: Thermodynamics of photochemical reductive dissolution in aquatic sediments. Geochim Cosmochim Acta 69:3249-3255 Sparks NHC, Mann S, Bazylinski DA, Lovely DR, Jannasch HW, Franekl RB (1990) Structure and morphology of magnetite anaerobically-produced by a marine magnetotactic bacterium and a dissimilatory ironreducing bacterium. Earth Planet Sci Lett 98:14-22 Stack AG, Erni R, Browning ND, Casey WH (2004) Pyromorphite growth on lead-sulide surfaces. Environ Sci Technol 38:5529-5534 Stone AT, Morgan JJ (1984) Reduction and dissolution of manganese(III) and manganese(IV) oxides by organics. 2. Survey of the reactivity of organics. Environ Sci Technol 18:617-624 Stumm W, Morgan JJ (1996) Aquatic chemistry: Chemical Equilibria and Rates in Natural Waters. Wiley, NY Sunda WG, Huntsman SA, Harvey GR (1983) Photoreduction of manganese oxides in seawater and its geochemical and biological implications. Nature 301:234-236 Suzuki Y, Kelly SD, Kemner KM, Banield JF (2002) Nanometre-size products of uranium bioreduction. Nature 419:134-134. Tang J, Myers M, Bosnik KA, Brus LE (2003) Magnetite Fe3O4 nanocrystals: Spectroscopic observation of aqueous oxidation kinetics. J Phys Chem B 107:7501-7506 Tartaj P, del Puerto Morales M, Veintemillas-Verdaguer S, González-Carreño T, Serna CJ (2003) The preparation of magnetic nanoparticles for applications in biomedicine. J Phys D: Appl Phys 36:R182-R197 Tebo BM, Bargar JR, Clement BG, Dick GJ, Murray KJ, Parker D, Verity R, Webb SM (2004) Biogenic manganese oxides: Properties and mechanisms of formation. Annu Rev Earth Planet Sci 32:287-328

154

Gilbert & Banęeld

Tebo BM, Ghiorse WC, van Waasbergen LG, Siering PL, Caspi R (1997) Bacterially-mediated mineral formation: insights into manganese(II) oxidation from molecular genetic and biochemical studies. Rev Mineral Geochem 35:225-266 Templeton AS, Spormann AM, Brown GE (2003) Speciation of Pb(II) sorbed by Burkholderia cepacia/goethite composites. Environ Sci Technol 37:2166-2172 Thomas TN, Land TA, Johnson M, Casey WH (2004) Molecular properties of adsorbates that affect the growth kinetics of archerite (KDP). J Colloid Interf Sci 280:18-26 Tiller CL, O’Melia CR (1993) Natural organic matter and colloidal stability: Models and measurements. Colloids Surf A 73:89-102 Todo S, Takeshita N, Kanehara T, Mori T, Mri N (2001) Metallization of magnetite (Fe3O4) under high pressure. J Appl Phys 89:7347-7349 Tolbert SH, Herhold AB, Johnson CS, Alivisatos AP (1994) Comparison of quantum coninement effects on the electronic absorption spectra of direct and indirect gap semiconductor nanocrystals. Phys Rev Lett 73:3266-3269 Torres-Martínez CL, Kho R, Mian OI, Mehra RK (2001) Eficient photocatalytic degredation of environmental pollutants with mass-produced ZnS nanocrystals. J Colloid Interface Sci 240:525-532. Tran Thoai DB, Hu YZ, Koch SW (1990) Inluence of the coninement potential on the electron-hole-pair states in semiconductor microcrystallites. Phys Rev B 42:11261-11266 Trindade T, O’Brien P, Pickett NL (2001) Nanocrystalline semiconductors: Synthesis, properties, and perspectives. Chem Mater 13:3847-3858 Tronc E, Belleville P, Jolivet JP, Livage J (1992) Transformation of ferric hydroxide into spinel by Fe(II) adsorption. Langmuir 8:313-319 Tronc E, Jolivet JP, Lefevre J, Massart R (1984) Iron adsorption and electron transfer in spinel-like iron oxide colloids. J Chem Soc 80:2619-2629 Urrutia MM, Roden, EE, Zachara, JM (1999) Inluence of aqueous and solid-phase Fe(II) complexants on microbial reduction of crystalline iron(III) oxides. Environ Sci Technol 33:4022-4028 van der Zee C, Roberts DR, Rancourt DG, Slomp CP (2003) Nanogoethite is the dominant reactive oxyhydroxide phase in lake and marine sediments. Geology 31:993-996 Vaughan DJ, Craig JR (1978) Mineral Chemistry of Metal Sulides. Cambridge University Press, Cambridge, MA Vayssiéres L, Chaneac C, Tronc E, Jolivet JP (1998) Size tailoring of magnetite particles formed by aqueous precipitation: An example of thermodynamic stability of nanometric oxide particles. J Colloid Interf Sci 205:205-212 Verwey EJW, Overbeek JThG (1948) Theory of the Stability of Lyophobic Colloids. Elsevier, Amsterdam Villalobos M, Toner B, Bargar JR, Sposito G (2003) Characterization of the manganese oxide produced by Pseudomonas putida strain MnB1. Geochim Cosmochim Acta 67:2649-2662 Voelker BM, Morel FMM, Sulzberger B (1997) Iron redox cycling in surface waters: Effects of humic substances and light. Environ Sci Technol 31:1004-1011 Vyalikh DV, Dänzenbacher S, Mertig M, Kirchner A, Pompe W, Dedkov YS, Molodtsov SL (2004) Electronic structure of regular bacterial surface layers. Phys Rev Lett 93:238103 Waite TD (1990) Photo-redox processes at the mineral-water interface. Rev Mineral 23:559-603 Wang CC, Zhang Z, Ying JY (1997) Photocatalytic decomposition of halogenated organics over nanocrystalline titania. Nanostruct Mater 9:583-586 Wang CS, Klein (1981) First-principles electronic structure of Si, Ge, GaP, GaAs, ZnS, and ZnSe. I. Selfconsistent energy bands, charge densities, and effective masses. Phys Rev B 24:3393-3416 Wang CY, Böttcher C, Bahnemannn DW, Dohrmann JK (2003) A comparative study of nanometer sized Fe(III)-doped TiO2 photocatalysts: synthesis, characterization and activity. J Mater Chem 13:2322-2329 Wang LG, Pennycock SJ, Pantelides ST (2002) The role of the nanoscale in surface reactions: CO2 on CdSe. Phys Rev Lett 89:075506 Wang LW, Zunger A (1996) Pseudopotential calculations of nanoscale CdSe quantum dots. Phys Rev B 53: 9579-9581 Warren LA, Ferris FG (1998) Continuum between sorption and precipitation of Fe(III) on microbial surfaces. Environ Sci Technol 32:2331-2337 Warren LA, Haack EA (2001) Biogeochemical controls on metal behaviour in freshwater environments. Earth Sci Rev 54:261-320 Watson JHP, Cressey BA, Roberts AP, Ellwood DC, Charnock JM, Soper AK (2000) Structural and magnetic studies on heavy-metal-adsorbing iron sulphide nanoparticles produced by sulphate-reducing bacteria. J Magnetism Magnetic Mater 214:13-30 Waychunas GA, Kim CS, Banield JF (2005) Nanoparticulate iron oxide minerals in soild and sediments: Unique properties and contaminant scavenging mechanism. J. Nanoparticle Res 7:409-433

Molecular-Scale Processes Involving Nanoparticulate Minerals

155

Webb SM, Tebo BM, Bargar JR (2005) Structural inluences of sodium and calcium ions on the biogenic manganese oxides produced by the marine Bacillus sp., strain SG-1. Geomicrobiol J 22:181-193 Wielinga B, Mizuba, MM, Hansel CM, Fendorf S (2001) Iron promoted reduction of chromate by dissimilatory iron-reducing bacteria. Environ Sci Technol 35:522-527 Williams AGB, Scherer MM (2004) Spectroscopic evidence for Fe(II)-Fe(III) electron transfer at the iron oxide – water interface. Environ Sci Technol 38:4782-4790 Williams KH, Ntarlagiannis D, Slater LD, Dohnalkova A, Hubbard SS, Banield JF (2005) Geophysical imaging of stimulated microbial biomineralization, Environ Sci Technol in press Wolthers M, van der Gaast SJ, Rickard D (2003) The structure of disordered mackinawite. Am Mineral 88: 2007–2015 Xu Y, Schoonen MAA (1995) The stability of thiosulfate in the presence of pyrite in low-temperature aqueous solutions. Geochim Cosmochim Acta 59:4605-4622 Xu Y, Schoonen MAA (2000) Absolute energy positions of conduction and valence bands of selected semiconducting minerals. Am Mineral 85:543-556 Yariv S, Cross H (1979) Geochemistry of Colloid Systems. Springer-Verlag, Berlin Yu I, Isobe T, Senna M (1996) Optical properties and characteristics of ZnS nanoparticles with homogeneous Mn distribution. J Phys Chem Solids 57:373-379 Zhang H, Penn RL, Hamers, RJ, Banield JF (1999) Enhanced adsorption of molecules on the surface of nanocrystalline particles. J Phys Chem B 103:4656-4662 Zhang H, Gilbert B, Huang F, Banield JF (2003) Water-driven structure transformation in nanoparticles at room temperature. Nature 424:1025-1029 Zhang H, Huang F, Gilbert B, Banield JF (2003) Molecular dynamics simulations, thermodynamic analysis, and experimental study of phase stability of zinc sulide nanoparticles. J Phys Chem B 107:1305113060 Zhang JZ (2000) Interfacial charge carrier dynamics of colloidal semiconductor nanoparticles. J Phys Chem B 104:7239-7253 Zhang Z, Wang CC, Zakaria R, Ying JY (1998) Role of particle size in nanocrystalline TiO2-based photocatalysts. J Phys Chem B102:10871-10878

Reviews in Mineralogy & Geochemistry Vol. 59, pp. 157-185, 2005 Copyright © Mineralogical Society of America

7

The Organic-Mineral Interface in Biominerals P. U. P. A. Gilbert*, Mike Abrecht, Bradley H. Frazer University of Wisconsin-Madison Department of Physics, and Synchrotron Radiation Center 3731 Schneider Drive Stoughton, Wisconsin, 53589, U.S.A. [email protected] * previously publishing as Gelsomina De Stasio

BIOMINERALS: TOUGH STRUCTURES OF LIFE Introduction to biominerals Numerous living organisms form minerals. These biogenic minerals, or biominerals, are composite materials containing an organic matrix and nano- or micro-scale amorphous or crystalline minerals. In this chapter we will review the molecular aspects of biomineralization and describe as completely as is currently possible the organic-mineral interface, the location in which organic-mineral interactions occur. Biomineral composite materials include bone, dentine, enamel, statoliths, otoliths, mollusk and crustacean shells, coccolith scales, eggshells, sponge silica skeletons, algal, radiolarian and diatom silica micro-shells, and a variety of transition metal minerals produced by different bacteria (Lowenstam and Weiner 1989; Addadi and Weiner 1997; Banield and Nealson 1997; Fortin et al. 1997; Fitts et al. 1999; Templeton et al. 1999; Lower et al. 2001a; Mann 2001; Glasauer et al. 2002; De Yoreo and Vekilov 2003; Weiner and Dove 2003; De Yoreo and Dove 2004). From a materials science perspective, organic molecules are soft, compliant and fracture resistant while inorganic crystals are hard and brittle. Biomineral composites combine the best of these properties and minimize the weaknesses: they are both hard and fracture resistant (tough) (Currey 1977; Jackson et al. 1988; Schäffer et al. 1997; Kamat et al. 2000; Gao et al 2003). This is due to several factors: structure, nano-size and chemical composition. Only recently materials scientists have begun to learn how to build a synthetic composite material that outperforms each component taken separately, and have done so inspired by shell nacre (Tang et al. 2003). The mechanisms of biomineral formation are not fully understood (Mount et al. 2004) and while they are of interest in their own right, they may also provide models for new materials concepts, inspire design solutions and give new insights into the genetic control of biological structure (e.g., Schäffer et al. 1997). Lowenstam (1981) introduced the distinction between the biologically induced mineralization, which is enacted extracellularly or on the cell surface by many algae and bacterial species, and the organic-matrix mediated mineralization performed by many animals (later termed biologically controlled mineralization) (Bazylinski and Frankel 2003; Frankel and Bazylinski 2003; Veis 2003). Eukaryotic biominerals often show complex hierarchical structure from the nanometer to the macroscopic scale. This structure confers mechanical strength and toughness: despite being highly mineralized, with the organic component constituting not more than a few percent of the composite material, the fracture toughness exceeds that of single crystals of the 1529-6466/05/0059-0007$05.00

DOI: 10.2138/rmg.2005.59.7

158

Gilbert, Abrecht, Frazer

pure mineral by two to three orders of magnitude (Kamat et al. 2000). The recent discovery of the silica skeleton in Euplectella sp., and its seven levels of structural organization illustrates one brilliant such hierarchy (Aizenberg at al. 2005), as reported in Figures 1 and 2. In eukaryotes biominerals provide a variety of functions including mechanical protection, movement, grinding, gravity or magnetic ield sensing. Conversely in prokaryotes biominerals are often formed as a byproduct of a biochemical pathway in which the bacteria oxidize or reduce transition metals or other species in solution, often for metabolic energy generation (Nealson and Stahl 1997; Frankel and Bazylinski 2003).

Why biominerals For prokaryotes and eukaryotes the complex bioinorganic chemistry involved in biomineralization constitutes a distinct evolutionary advantage for the organism performing it. That advantage is the reason biomineralization became as widely spread as is observed in the three kingdoms of life: Archaea, Bacteria, and Eukarya (protists, fungi, plants, and animals), although very few Archaea are known to be biomineral producers. In the cases of eukaryotic biominerals, the biomineral products are clearly of direct use and beneit. In the case of microbial biomineralization, metal oxidation or reduction can be induced - or exploited - by the bacterium, but mineralization itself may have only indirect advantages or even disadvantages. Biomineral formation often occurs extracellularly and is subsequent to oxidation or reduction. In some cases it is detrimental: entombment of the bacterium in its own biomineral products is possible, and the cell either dies or develops an evasion strategy, such as the formation of mineral sheaths (e.g., Leptothrix spp) or stalks (Gallionella ferruginea).

The organic-mineral interface Many biomineralization mechanisms are poorly understood at the molecular level. These include all cases in which highly oriented crystals are formed with the growth of a particular crystal phase, or polymorph (Falini et al. 1996; De Yoreo and Dove 2004). Mollusk shell, bone and some bacterial ilaments are examples of such biomineralization: a highly specialized organic matrix directs the formation of a speciic crystal phase, habit, size and orientation. In these composites the organic-mineral interaction is so specialized that a mechanism of epitaxial overgrowth, or templation, can be invoked. The particular matrix of organic molecules, produced by the living organism, acts as a template upon which crystals grow epitaxially, or simply, growth is nucleated, and crystal structure, phase, orientation and often habit of the mineral are determined by the organic matrix. Figure 3 shows a biomineralization paradigm. The paradigm of Figure 3 is not general: it applies to some prokaryotic and many eukaryotic biominerals. It is simply intended to guide our reasoning and give a visual model to refer to in this discussion; it is by no means intended to describe and include every biomineral formation mechanism. Many prokaryotic biominerals, in fact, do not follow such a model. Consistently, however, whether the paradigm applies or not, the organic macromolecules are formed irst and they direct or induce the growth of speciic minerals and their polymorphs. To this day, the organic macromolecular components have been identiied in only a few biominerals. This paradigm, therefore, is to be interpreted as a conceptual mechanism, not as a detailed model of interaction between known molecules. The present chapter discusses the possibility of investigating the organic mineral interface, and the chemical bonds formed at that interface, in essence, zooming in on the interface as shown in Figure 3D. In both biologically controlled and mediated biomineralization (Lowenstam 1981), the organic components are formed irst, then these bind a few ions, which serve as nucleation sites for crystal growth (Lowenstam and Weiner 1989; Falini et al. 1996; Gotliv et al. 2005). Self-assembly and epitaxial crystal growth subsequently complete the composite structure discussed in Figure 3. We propose that the exploitation of such templation mechanisms can be

Figure 1

Figure 1. Structural analysis of the mineralized silica skeleton of Euplectella sp., a deep sea sponge from the Paciic Ocean. (A) Photograph of the entire skeleton, showing the cylindrical glass cage. (B) Fragment of the cage structure showing the square-grid lattice of vertical and horizontal struts with diagonal elements arranged in a chessboard manner. Orthogonal ridges on the cylinder surface are indicated by arrows. (C) Scanning electron micrograph (SEM) showing that each strut (enclosed by a bracket) is composed of bundled multiple spicules (the arrow indicates the long axis of the skeletal lattice). (D) SEM of a fractured and partially HF-etched single beam revealing its ceramic iber-composite structure. (E) SEM of the HF-etched junction area showing that the lattice is cemented with laminated silica layers. (F) Contrast-enhanced SEM image of a cross section through one of the spicular struts, revealing that they are composed of a wide range of different-sized spicules surrounded by a laminated silica matrix. (G) SEM of a cross section through a typical spicule in a strut, showing its characteristic laminated architecture. (H) SEM of a fractured spicule, revealing an organic interlayer. (I) Bleaching of biosilica surface revealing its consolidated nanoparticulate nature. Reprinted with permission of AAAS, from Aizenberg et al. (2005), Science, Vol. 309, Fig. 1, p. 275-278.

Organic-Mineral Interface in Biominerals 159

160

Gilbert, Abrecht, Frazer

Figure 2. The Euplectella sp. Skeletal system structure (left) resembles that of the Swiss Re Tower in London (top right), the Hotel Arts in Barcelona (center right), and the Eiffel Tower in Paris (fragment shown bottom right). The characteristic structure with vertical and horizontal struts to form a checkerboard pattern, and diagonal struts at every other square is optimized for mechanical strength. Image courtesy of Joanna Aizenberg.

Figure 3. Paradigm for the epitaxial overgrowth, or templation, mechanism in biomineralization. The organic matrix (A) is composed of macromolecules, which depending on the particular biomineral may include a single organic molecule, e.g., a polysaccharide or a complex arrangement of proteins and glycoproteins. In all cases the organic components have charged functional groups that attract ions from solution (B). The steric arrangement of organic macromolecules, their sequence, and folding determines the precise position in three dimensions of the ions. Such positions are only compatible with a speciic mineral, even more: they are only compatible with a well-determined polymorph of a speciic mineral (C). The crystal structure shown (C) is aragonite, the large white ions in (B) are Ca2+, while the small-white and large-dark atoms are C and O, respectively in (C). (D) Zooming in on the organic-mineral interface: the inter-atomic bonds are indicated by dashed lines. A similar mechanism of epitaxial overgrowth takes place in many matrix-mediated eukaryotic and some biologically induced prokaryotic biomineralization mechanisms.

Organic-Mineral Interface in Biominerals

161

considered a “genome shortcut”, naturally selected for minimizing the amount of information the organism must transfer down the lineage, while maximizing the performance of the inal composite material. Speciically, self-assembly and epitaxial crystal growth are harnessed by the organism, therefore the only information stored in the genome is that involving the synthesis of the organic macromolecules of Figure 3A.

Zooming in on the organic-mineral interface Several authors suggested that the negatively charged amino acids, aspartate and glutamate, along their protein sequences attract positive ions from solution and initiate crystal nucleation and growth (Mann 2001; Weiner and Dove 2003; Gotliv et al 2005). Certainly the concentration of these amino acids in the known and sequenced organic matrix proteins is very high. They usually constitute between 30 and 40 mol% of the matrix protein, and in some cases even more. The recently discovered “Asprich” family of proteins from the bivalve mollusk Atrina rigida contain more than 50 mol% of aspartate and 10 mol% glutamate, hence their name (Gotliv et al. 2005). Therefore, the paradigm by which negatively charged amino acids collect ions from solution, provide the nucleation sites, and direct the epitaxial growth of biominerals deserves further investigation. A novel set of tools is necessary to discover exactly which molecules interact at the organic-mineral interface, and at which speciic molecular sites the irst chemical bonds are formed, that is, how biogenic mineral formation begins. X-ray spectromicroscopy, used in combination with other microscopic and biological methods, is one such novel tool to explore the chemistry of templation mechanisms at this interface.

SPECTROMICROSCOPY OF BIOMINERALS XANES spectroscopy of biominerals Only recently has the study begun of templation mechanisms in biominerals using X-ray absorption near edge structure (XANES) spectroscopy. We believe that the understanding of organic-mineral templation can be signiicantly improved with XANES spectroscopy, because this powerful chemical analysis is sensitive to elemental composition, oxidation state, coordination number, and crystal or molecular structure and orientation of minerals and organic molecules (Stöhr 1992). XANES spectroscopy is a technique irst introduced in the early 1980s—under X-ray illumination the sample emits electrons and photons, which constitute a spectrum as the X-ray energy varies. In this section a more detailed description will introduce the biomineralogist to this spectroscopy, to the microscopy version of it, and to the combination of spectroscopy and microscopy, termed spectromicroscopy. The best tool for understanding the chemical and physical properties of any material is one that reports on the x,y,z coordinates of atoms in the material. This can be done with great accuracy (1 Å resolution) using X-ray diffraction on periodic structures, such as bulk mineral crystals or crystallized organic macromolecules, when these can be crystallized. For all other systems, which are not periodic—for example, organic macromolecules that cannot be crystallized—one is left with information that can be retrieved with spectroscopies. Various kinds of spectroscopies detect different characteristics of the specimen: the resonances of nuclei, the vibrational states, the electronic structure and many more. All these approaches are limited, compared to diffraction, but they are all that is available for the majority of organic molecules. XANES is one such spectroscopy, and a very powerful one compared to others, although not quite as informative as diffraction. Its main advantage is its wide range of applicability: it can in fact detect and report on the electronic structure of ordered and disordered materials, minerals surfaces, organic macromolecules, their molecular structure, composition, and the chemical bonds that these molecules form with minerals and nano-crystalline mineral precursors.

162

Gilbert, Abrecht, Frazer

For a complete review of the XANES approach see Stöhr 1992. Briely, XANES spectroscopy probes a speciic element according to its absorption of X-ray photons, at energies that are characteristic of the element and the absorption edge. Absorption edges correspond to transitions between occupied atomic-like electronic shells (e.g., 1s, 2p, 3d, etc.) to unoccupied electronic states (molecular antibonding orbitals such as S*, V*) that are strongly affected by both the absorbing atom itself and by its neighboring atoms. Transitions from the 1st(e.g., 1s), 2nd(e.g., 2s or 2p), 3rd, etc. atomic shells to molecular orbitals are called K- or L- or M-edges. These probe bonds of the absorbing atom to intra-molecular and, to a lesser degree, extramolecular neighbors. For example the transition from 1s to S* of the C=O in the carboxyl group (COOH) has a distinct resonance (peak in the XANES spectrum) at 288.6 eV. XANES spectroscopy can also detect the presence of speciic bonds in molecules, and determine the orientation of molecules or functional groups on the surface of solids. Since the absorption edges of low-Z elements (up to atomic number Z = 30, including all non-gaseous elements from Li through Zn), which are the most relevant for biominerals, are in the 10-1000 eV range of binding energies, a source of photons with such energies must be used to observe such edges with XANES. The only tunable sources of 10–1000 eV photons, called the softX-ray range, are synchrotrons. Synchrotron radiation is also emitted by accelerated ions in the universe such as galaxies and nebulas, but those sources are too distant, their photon lux is therefore too low, screened by the atmosphere, and not easily tunable; therefore galaxies and nebulas are not useful as illuminating sources for XANES of materials on Earth. Thus, XANES spectra are acquired at synchrotron facilities while scanning the photon energy across the absorption edges of the relevant elements. When the photon energy reaches or exceeds the binding energy of electrons in a speciic atomic shell, the sample photoemits, that is, it emits electrons by the photoelectric effect. The photoelectric effect was irst observed by H. Hertz in 1887 (Hertz 1887); but only explained in 1905 by A. Einstein (Einstein 1905; the centennial celebration is undergoing this year). With his explanation Einstein introduced the concept of a photon, which is a quantum of electromagnetic energy. For this fundamental leap in the understanding of light and matter, and not for the more famous special and general relativity theories, Einstein earned the Nobel prize in 1921. In an effort to avoid a quantum-mechanical presentation, we provide in Figures 4-6 an overview of all the information afforded by XANES, and provide via examples an introduction to this extremely powerful approach, tailored to entice the biomineralogist and the geomicrobiologist. Figure 4 illustrates the sensitivity of XANES spectromicroscopy to the elemental composition of samples, and the possibility of spatially mapping the elemental distributions. Figure 5 shows the sensitivity of XANES spectra to the oxidation states of two transition metals at the L-edges. The dramatic differences introduced in the XANES spectral lineshape by changes in oxidation states make it possible to identify with this spectroscopy the chemical species present in the sample. By combining this information with the imaging capability demonstrated in Figure 4, the spatial distribution of elements and their oxidation states can be determined. As mentioned above, XANES spectroscopy is also sensitive to the crystal structure. Figure 6 shows the difference in spectra acquired from CaCO3 and ZnS mineral polymorphs, that is, minerals in which the chemical formula is identical but the crystal unit cell has a different arrangement. Slight differences in the local structural and electronic environment of elements in alternative crystal polymorphs can give clear ingerprints in XANES spectra. XANES spectroscopy is also sensitive to coordination, that is, the number of atoms to which the element under analysis is bonded. Calcium in calcite is 6-coordinated, that is, each Ca atom is bonded to six oxygen atoms all at the same distance (2.35 Å). In aragonite Ca

Organic-Mineral Interface in Biominerals

163

Figure 4. X-ray photoelectron emission spectromicroscopy (X-PEEM) image (top left) of Calothrix cyanobacteria embedded in epoxy and microtomed to a 60 nm thick section. In the top left portion of the ield of view a Calothrix ilament, containing seven bacteria, is visible, while at the center is the cross section of a single-cell, or possibly a ilament extending along the direction perpendicular to the image. The distribution maps of sodium, uranium (from uranyl acetate stain), sulfur, phosphorus and calcium are also reported. The distribution maps were obtained by digital ratio of two images, on- and off-peak at the Na L-edge, U O-edge, S L-edge, P L-edge and Ca L-edge respectively, and represent the local concentration of the relevant element: darker gray levels indicate greater elemental density. Notice the high concentration of Na, S, U and Ca in the outer sheaths, and the high density of P corresponding to bacterial DNA. Sample courtesy of Susanne Douglas.

has 9-fold coordination, with ive distinct bond lengths; these are: one oxygen at 2.32 Å, two oxygens at 2.43 Å, two oxygens at 2.51 Å, two oxygens at 2.57 Å and two oxygens at 2.66 Å. Consequently, the differences in the crystal ield peaks (arrows in Fig. 6A) between aragonite and calcite may be due to both coordination and crystal structure. In both ZnS polymorphs, sphalerite and wurtzite, sulfur atoms are 4–coordinated (as are the Zn atoms), and the nearneighbor environments are almost identical. The crystal structures are different (cubic and hexagonal), therefore the distribution of atoms in the third and higher shell of atoms around the sulfur sites are distinct, and this is the origin of the spectral differences in Figure 6B. XANES spectroscopy has been successfully used to reveal the presence and oxidation state of speciic elements in geologic minerals (Sturchio et al. 1998), the structure of synthetic

164

A

Gilbert, Abrecht, Frazer

B

Figure 5. (A) Manganese L-edge X-ray absorption near edge structure (XANES) spectra of manganese oxides. The formal Mn oxidation states are given on the right. (Data from Gilbert et al. 2003a). (B) Iron L-edge XANES spectra from ferric (III), ferrous (II) and metallic iron (0), in the minerals and metal indicated. (Data from Frazer et al. 2005)

A

B

Figure 6. (A) Ca L-edge XANES spectra from calcite and aragonite, trigonal-rhombohedral and orthorhombic polymorphs of CaCO3, respectively. The two main peaks, common to both spectra are the L3 and L2 edges, respectively. Two additional peaks at ~346 and ~350 eV (arrows), due to the crystal ield, are prominent in calcite but have lower intensity and different line shape in aragonite. (B) XANES sulfur L-edge spectra from sphalerite and wurtzite, cubic and hexagonal polymorphs of ZnS, respectively. Notice the difference in line shape between 165 and 170 eV. Used with permission of the American Chemical Society, from Gilbert et al. (2003), J. Phys. Chem. A, Vol. 107, Fig. 1, p. 2839-2847.

Organic-Mineral Interface in Biominerals

165

materials (Bozek et al. 1990), elemental speciation in soils and sediments (Myneni et al. 1997; Beauchemin et al. 2003; Zawislanski et al. 2003) and other environmentally relevant samples (Myneni 2002a,b; Myneni et al. 1999). Many other experiments on the microlocalization of trace elements in eukaryotic cells (De Stasio et al. 1993, 1996, 2001; Gilbert et al. 2000) and the identiication of prokaryotic biomineral products (Labrenz et al. 2000; Lawrence et al. 2003; Lopez-Garcia 2003; Chan et al. 2004), also attest to the power and breadth of XANES spectroscopy and spectromicroscopy. As mentioned, the lineshape of XANES spectra gives information on the molecular and/ or crystal structures surrounding the element under analysis. The interpretation of spectral lineshape and peak assignment, however, can be complicated. When the molecular or crystal structure is known, and relatively simple, ab initio calculations can be used to simulate the XANES spectrum. A comparison of experimental and calculated spectra enables peak assignment to speciic molecular structures. Speciic peaks can be considered “spectral signatures” of speciic molecular features. XANES is extremely sensitive to carbon chemistry: examples of molecular features that generate well-established spectral signatures are C{C, C=C, C–O, C=O, C–O, as well as C–C–C bond angles, conjugation of adjacent bonds, etc. A material that contains several of these molecular features exhibits a XANES spectrum resulting from the combination of their corresponding spectral signatures: the “building blocks” (Stöhr 1992). For other edges, e.g., Si or S at the L-edge, simulations of XANES spectra are not currently adequate because the electronic structure is too complex to be calculated. In these cases, the spectral signatures do exist and are measurable, but they are not univocally assigned to speciic bonds or molecular structures. Unknown minerals, however, such as sub-micron silicate inclusions, can still be identiied by empirical comparison with spectra from known, macroscopic, reference silicate minerals (De Stasio et al. 2003; Gilbert et al. 2003b).

XANES microscopy of biominerals XANES spectroscopy has been used to study the same kind of molecular interactions discussed hereafter, but without spatial resolution. Examples include organic-mineral interaction at the binding sites in metalloproteins (Benfatto et al. 2003) or between metal ion and humic macromolecules (Myneni et al. 1999; 2002a). There are practical reasons that, until recently, completely precluded the spectromicroscopy of biomineralized structures, as described below irst from a spectroscopy, then from a microscopy point of view. XANES spectroscopy can be performed in two ways: by detecting either luorescence photons or photoemission electrons (photoelectrons) from a solid sample surface. Fluorescence XANES signal is most intense for high Z elements (Z > 30). These elements have their core shell electrons at binding energies much greater than 1000 eV, therefore the corresponding absorption edges detectable by XANES spectroscopy can only be detected in the “hard-X-ray” regime. On the contrary, low Z elements, which include all the organic elements C, N and O, have their absorption edges below 1000 eV: the C K-edge is at 285 eV, the N K-edge at 400 eV, and the O K-edge at 531 eV. Since none of these edges is easily accessible to the hard-Xray luorescence range, the organic components of biominerals have never before been studied with luorescence XANES. Photoelectron XANES, also known as total electron yield or TEY-XANES, is much more intense than luorescence below 1 keV, where the Si, P, S, and Ca L-edges, and the C, N and O K-edges are located. In this spectral region, a strongly space-averaged TEY spectroscopy

166

Gilbert, Abrecht, Frazer

has always been possible on an insulating biomineral. This is, however, not particularly informative, given the highly organized microscopic structure of biominerals. Spectromicroscopy with X-ray PhotoElectron Emission Spectromicroscope (X-PEEM) adds spatial resolution to the TEY-XANES experiment, down to the 10 nm level (Frazer et al. 2004). Until recently, however, X-PEEM could only image and analyze the chemistry of conductive sample surfaces. Insulating samples such as minerals and biominerals could not be analyzed without major charging problems. Transmission X-ray microscopy experiments (e.g., scanning transmission X-ray microscopy, STXM (Kilcoyne 2003; Tyliszczak 2004), which do not suffer from charging, are limited to very thin solid samples (few atomic monolayers) or dilute liquid samples. Most biominerals, therefore, are excluded from this powerful analysis.

Overcoming charging effects We recently optimized a differential-thickness coating method (De Stasio et al. 2003) that enabled us to extensively study mineral and biominerals surface with X-PEEM and do highresolution imaging and XANES analysis on them (35 nm or better) (Gilbert et al. 2003-2). The coating approach is shown in Figure 7. We have used this coating approach on a variety of insulators, including wood, quartz, zircons, glass slides, tribological polyphosphate and nano-diamond ilms, cells in culture, mollusk shells and bone. In all these cases the coating completely removed charging and enabled micro- and nano-XANES spectroscopy of insulators. Figure 8 shows a representative example of the results enabled by differential-thickness coating. As aforementioned, the combination of

Figure 7. Schematic diagram showing the preparation steps (top to bottom) for the differential thickness coating. First the biomineral (e.g., a mollusk shell) is embedded in epoxy, then the surface is polished with grit, down to 50 nm if high-resolution imaging is desired, then a thick coating (500 Å) of platinum is deposited by magnetron sputtering on the sample, while masking and not coating the central area, typically 3 mm in diameter, which will then be analyzed by X-PEEM. Finally, a thin coating (10 Å) is deposited on the whole sample surface. The photoelectron escape depth is on the order of 30 Å at the C K-edge, therefore photoelectrons from the shell can be collected through the 10 Å coating at the center. The thicker coating layer around the central region ensures perfect conductivity and a good electrical contact with the sample holder, therefore the sample can be kept at a reliable and stable voltage, it does not charge when electrons are extracted by X-ray illumination, and XANES analysis can be performed (De Stasio et al. 2003).

Organic-Mineral Interface in Biominerals

167

Figure 8. Distribution map of Ca in the nacreous layer of pinctada margaritifera, the Tahitian black pearl oyster. Dark indicates higher Ca concentration. The Ca-poor regions between nacre tablets are thicker organic matrix strata that could be due to seasonal changes, as suggested by other researchers of abalone nacre (Lin and Meyers 2005), although the thickness and spacing are different in pinctada. This map was acquired using the spectromicroscope for photoelectron imaging of nanostructures with X-ray s (SPHINX) instrument, which is an XPEEM (Frazer et al. 2004), on a fragment of nacreous layer from pinctada embedded in epoxy and polished.

XANES spectroscopy and X-PEEM microscopy is called spectromicroscopy. From an image such as the one in Figure 8, speciic regions of interest can be selected (with the computer mouse), and spectra (e.g., C K-edge XANES spectra) can be extracted, showing different spectroscopic signatures characteristic of the crystals and the organic matrix. This powerful technique can investigate both the organic and the inorganic components of biominerals. Real-time, full-ield imaging can be done with a maximum ield of view of 180 μm in diameter. At this low magniication the area of interest in a biomineral can easily be identiied, then zooming in to higher magniication down to a ield of view of 1.7 μm allows highresolution imaging and spectroscopic analysis of biomineral nanostructures. The usual mode of data acquisition consists of acquiring stacks of images while scanning the photon energy, therefore obtaining “movies” that can then be played independent of the synchrotron source. In these movies the third coordinate is energy rather than time, and each pixel (typically 512 u 512 pixels in each image) contains the full XANES spectrum. The number of spectra simultaneously acquired is therefore 2 u 105. The resulting complexity in data analysis and interpretation initiated a considerable effort in software design, which is in constant evolution. From each one of these movies, all the elemental composition, oxidation state, coordination number, molecular or crystal structure information is available, and can be retrieved after data acquisition. Once carbon XANES spectra from the bound mineral-templating and unbound organic matrices of biominerals are obtained, the difference between those spectra reveals the organic-mineral interaction. Interpretation of the data is then done by comparison with the extensive literature on carbon XANES spectroscopy in individual amino acids and organic compounds (Stöhr 1992; Kaznacheyev et al. 2002; Carravetta et al. 1998; Myneni 2002b; Lawrence et al. 2003), or by comparison with reference molecules prepared and analyzed separately for a speciic interaction. Two main limitations remain for XANES spectromicroscopy with the X-PEEM approach: the samples must be compatible with ultra-high vacuum, and must be lat. The vacuum compatibility requirement arises from the necessity to collect photoelectrons, which would recombine with gas molecules if these were present in the experimental chamber. The latness requirement arises from the necessity to keep the sample at high voltage (typically –20 keV) to accelerate electrons away from the sample surface and towards the electron optics column. If the samples have high surface corrugation, greater than ~1 μm, severe distortions of the electric ield provoke imaging artifacts and distortions, and in extreme cases even arching and

168

Gilbert, Abrecht, Frazer

sparking which preclude analysis. Surface corrugations lower than 0.5 μm in height, however, are very easy to obtain on solid samples such as minerals and biominerals by conventional surface polishing. Another complication in the XANES X-PEEM approach is the dificulty in separating co-localized mixed phases. In the presence of multiple proteins in a biomineral (for example bone), carbon K-edge spectra may be too complicated to interpret. In that case it is necessary to acquire spectra from separate single-components and deconvolve individual contributions to XANES spectra of the mixture. Separation and/or puriication of single components may not be possible. Furthermore, the components may not be spectroscopically distinguishable. If the individual organic components contributing to XANES spectra are known and spectroscopically distinct, singular value decomposition or cluster analysis methods can be used to deconvolve and quantify their contributions (Pickering et al. 2000; Lerotic et al. 2004).

THE ORGANIC-MINERAL INTERFACE IN MICROBIAL BIOMINERALS Prokaryotic biominerals Microbes or prokaryotic cells, which include Bacteria and Archea, are single-celled organisms. They are small, ranging in size from 200 nm to 7 μm, and lack the tissue differentiation and sophisticated external structures immediately apparent in single- and multicelled eukaryotes. They also lack nuclei and membrane-bound internal organelles, with the notable exception of magnetosomes, surrounded by a phospholipid bilayer, in magnetotactic bacteria. Another exception are Gemmatata obscuriglobus, recently discovered bacteria with a double membrane surrounding their nucleoid, making them appear very similar to eukaryotes (Fuerst and Webb 1991; Lindsay et al. 2001). Prokaryotes, however, are among the most abundant organisms on Earth and can be found in virtually every known environment. In the driest location on Earth, the Atacama desert in Chile, 103 bacteria per gram of soil, can be found in the immediate underground (Maier et al. 2004). That number increases dramatically in more hospitable locations, up to 109 bacteria/g of soil in the rolling hills of Tuscany or the rain forests. Bacteria have also been found as deep under the Earth’s crust as man has drilled: over 6000 m underground in a South African mine (Takai et al. 2001; Newman and Banield 2002). Prokaryotes not only inhabit all natural waters, soils and sediments, they are also capable of surviving in extremes of temperature, pH, or salinity. Additionally, unlike eukaryotes, which depend on glycolysis and require glucose as an energy source and oxygen as an oxidant, prokaryotes adapt to extract energy from diverse and often even multiple chemical reactions (Nealson and Stahl 1997). Sources of metabolic energy include redox reactions of minerals and ions in solution, as well as other inorganic molecules. One of the most eclectic of bacteria, Shewanella putrefaciens, can extract energy from reducing iron and manganese oxides, or sulfur, or fumarate or nitrate or many other compounds, in anaerobic conditions, depending on their availability. If oxygen is available instead, Shewanella becomes an aerobic organism using molecular oxygen to oxidize its energy source (organic carbon or hydrogen) (Myers and Nealson 1990). Returning to the deinitions of the different biomineralizations, microbes mostly perform biologically induced mineralization (Lowenstam 1981; Frankel and Bazylinski 2003). Magnetotactic bacteria are an exception, and are, together with coccoliths, the most studied microbes to exhibit biologically controlled mineralization (Bazylinski and Frankel 2003). Biologically induced mineralization is especially signiicant for bacteria in anaerobic habitats, because in these conditions bacteria respire with sulfate and/or various metals as terminal electron acceptors in electron transport (Frankel and Bazylinski 2003).

Organic-Mineral Interface in Biominerals

169

Bacterial communities and bioilms thrive in environments rich in metal ions in solution and play an important role in mineral and/or rock dissolution, formation and deposition. From a materials science point of view, prokaryotes can be considered rock-catalysts: they enact or induce chemical transformations that lead to geochemical cycling and biomineral formation. Minerals formed by biologically induced mineralization are generally nucleated and grown extracellularly as a result of metabolic activity of the organism and subsequent chemical reactions involving metabolic byproducts. Microbes secrete one or more organic macromolecules that react with ions or compounds in the environment, resulting in the subsequent deposition of mineral particles. This biomineralization may be unintended (Frankel and Bazylinski 2003) or advantageous for the organism (Chan et al. 2004). The minerals formed are often nanoparticles with considerable particle-size distributions (Frankel and Bazylinski 2003). A more thorough review of the general characteristics of prokaryotic nanoparticulate biominerals is given elsewhere in this volume (Gilbert and Banield 2005). As catalysts of biomineralization, however, prokaryotes are ideally conigured. The larger the volume of an organism, the smaller the surface/volume ratio. Therefore the smallest organisms, microbes, are the most eficient in rapidly exchanging nutrients and waste byproducts with the surrounding environment. This metabolic advantage also implies that every bacterium can produce many times its body weight in biominerals. This eficiency has a price: bacteria, more likely than other larger organisms, are prone to become encrusted in their biomineral products. Furthermore, the microbial cell walls have a strong negative charge, with multiple sites available for metal binding. Metal ions in solution interact with the charged surface of the cell wall and initiate the formation of minerals. In other microbes, additional structures such as sheaths, capsules, S-layers and ilaments provide binding and nucleation sites for mineralization. In addition, bacteria can induce mineralization by secreting extracellular polysaccharides and enzymes that, when released into the surrounding environment, transform minerals already present or induce the precipitation of new minerals and metastable mineral precursors. In the irst case, the organic-mineral interface of Figure 3D is located on the surface of bacteria or on extruded but still connected structures, whereas in the second case the biomineralization occurs entirely extracellularly and away from the cell bodies. In both cases, the organic macromolecules induce nucleation and growth of the minerals, and are formed irst. In prokaryotic biomineralization, however, a combination of the cell physiology and the chemistry of the surrounding environment determine the mineralization process and the inal mineral product. No general statements, therefore, can conidently be made, and the paradigm of Figure 3A is certainly not widely applicable to prokaryotic biomineralization. The only general conclusion, perhaps, is that as a result of prokaryotic biomineralization the mineral changes redox state and the microbe gains energy, while in eukaryotic mineralization there is seldom a redox change, and the organism expends energy to form the biomineral. The structure and dynamics of the microbe-mineral interface can be studied with atomic force microscopy (Lower et al 2001a). The interactions at that interface were also reviewed by Juniper and Tebo (1995). Several groups did spectroscopic analysis of the minerals formed by microbes. Among these, several studies used extended X-ray absorption ine structure (EXAFS) spectroscopy, which explores the structure of the 2-3 nearest neighboring shells of atoms in minerals of biogenic origin. These studies include Suzuki et al. (2002), Tebo et al. (2004), and Villalobos et al. (2005). Another article analyzed the changes in elemental concentrations between adhering and suspended bacteria using hard X-ray luorescence spectromicroscopy (Kemner et al. 2004). Other studies used XANES or STXM-XANES spectromicroscopy, to analyze the oxidation states of biomineral products (Grush et al. 1996; Tonner et al. 1999; Toner et al. 2005). The latter studies, being all in the soft-X-ray region have the potential of analyzing both the mineral and the organic components of biominerals. This is, again, due to the location in energy of the absorption thresholds of organic elements, C, N, O, etc. However, to

Gilbert, Abrecht, Frazer

170

the best of our knowledge all those studies have focused on the mineral components of bacterial biomineralization. We will later describe the irst two cases in which the organic and mineral components, and the interface between them was analyzed using XANES spectromicroscopy (Chan et al 2004; Lawrence et al. 2003). Biomineralization on various prokaryotic structures is reviewed hereafter, including cell walls, capsules, S-layers, sheaths, and ilaments.

Bacterial cell walls Bacterial cell walls can be classiied into one of two groups based on their reaction to Gram’s stain, a stain used for visible light microscopy. The cell walls of both Gram-positive and Gram-negative bacteria are negatively charged and may induce biomineral formation. However, Gram-negative cells have been shown to precipitate only a fraction of the quantity of minerals produced by Gram-positive cells (Beveridge and Fyfe 1985). The cell wall for Gram-positive bacteria is made up of a layer of peptidoglycans and is separated from the interior of the cell by the plasma membrane. Peptidoglycans are composed of repeating dimers of N-acetylglucosamine and N-acetylmuramic acid. Each N-acetylmuramic acid molecule exhibits a side stem, which is a peptide with four or ive amino acids. These stems covalently bond with other stems on neighboring peptidoglycan strands to form a strong and enduring 3-dimensional macromolecular structure that surrounds the bacterium. This cell wall is 15-25 nm thick (Fortin and Beveridge 2000). Both the glycan strands and peptide stems of peptidoglycans are rich in carboxyl groups and give the cell wall a net negative charge. Secondary polymers like teichoic or teichuronic acids, which contain negatively charged phosphoryl groups, are also bound in the peptidoglycan structure and increase the negativity of the cell surface. The large number of anionic reactive sites provided by the peptidoglycan layer is the main source of surface catalysis or mineralization in Gram-positive bacteria. The cell walls of Gram-negative bacteria are more complex, both structurally and chemically. The peptidoglycan layer is much thinner than in Gram-positive cells (3 nm), contains no secondary polymers and is bound on both sides by membranes composed of lipid-protein bilayers. The outer membrane of Gram-negative bacteria is unique and asymmetric: the inner layer is composed of phospholipids but the outer layer contains an unusual lipopolysaccharide (LPS) layer, which is found uniquely in prokaryotes. LPS is a large complex molecule with three components: a lipid core, a core polysaccharide and a short polysaccharide chain that contains unique and species-speciic sugar sidechains. The core polysaccharide is rich in anionic phosphate and carboxyl groups and gives the cell wall a net negative charge. The sugar sidechains can extend up to 40 nm from core polysaccharides and may also contain negatively charged carboxyl groups (Langley and Beveridge 1999). In contrast to Gram-positive bacteria, the peptidoglycan layer is not only considerably thinner, but also shielded by the outer membrane in Gram-negative bacteria. Metal ions in the environment, therefore, cannot reach the peptidoglycans, presumably by the same mechanism excluding the Gram stain, and the biomineralization site is constituted of the numerous phosphate and carboxyl groups in the LPS layer (Fig. 9). Active cell metabolism can slow down biomineral formation on the cell wall. A clear example of this behavior is given by Bacillus subtilis cells. During metabolism, a membraneinduced proton motive force continuously pumps protons into the cell wall. Therefore metal ions must compete with protons for anionic cell wall sites, and the result is that these bacteria bind more minerals dead than alive (Urrutia et al. 1992).

Capsules Both Gram-positive and Gram-negative cells can possess additional outer layers that also

Organic-Mineral Interface in Biominerals

171

Figure 9. Transmission electron micrograph of Shewanella putrefaciens, a Gram-negative bacterium, exposed to nanocrystalline hematite. The crystals adhere to the cell wall due to its negative charge. This example illustrates that the bacterial cell wall not only binds ions from solution, but also alreadyformed mineral crystals. Reproduced with permission of the American Society of Microbiology, from Glasauer et al. (2001), Applied and Environmental Microbiology, Vol. 67, Fig. 5, p. 5544-5550.

induce biomineralization. Among these, capsules are highly hydrated amorphous matrices of exopolysaccharides or polypeptides, and strongly attached to the cell wall (see Fig. 10A). Capsules extend up to 1 μm away from the cell, and serve as protective shields for bacteria and, as cell walls, contain numerous carboxyl groups (see Fig. 10B). They contain 99 % water and allow for eficient transport of nutrients and waste products (Schultze-Lam et al. 1993). The negatively charged polysaccharides ilter and capture the positive cations from solution and induce precipitation away from the cell, thereby protecting the organism from becoming encrusted with minerals.

A B

Figure 10. (A) TEM micrograph of bacteria, surrounded by exopolysaccharide (EPS) capsules, to which clay nanoparticles adhere. Reproduced with permission from www.nwri.ca/envirozine/images/bacteria_ e.gif. (B) SEM image of another bacterium exhibiting the remains of a capsule. Bacteria in this ground water sample were not ixed, nor treated in any way, therefore the morphology of the 99% water-containing capsule is altered by dehydration. Sample courtesy of Clara S. Chan.

172

Gilbert, Abrecht, Frazer

Capsules also stabilize the metal ion concentration around the cell wall. This is particularly advantageous when the metal ion concentration in the surrounding environment, which naturally luctuates, reaches toxic levels. Experiments have shown that mutated forms of Klebsiella aerogenes, which do not produce capsules, were unable to survive in concentrations of metals in which the capsule-forming wild-type strains thrived (Bitton and Freihofer 1978).

S-layers S-layers are paracrystalline surface layers 5 to 25 nm thick, containing many ordered repeats of a single protein or glycoprotein. Both Archea and Bacteria may form S-layers. These self-assemble into a well-ordered two-dimensional shell around the bacterium (Sleytr 1997). S-layer proteins or glycoproteins assemble into regular patterns in which the unit cell has 2-, 3- 4- or 6-fold rotational symmetry. The well ordered lattice contains pores that are identical in size and morphology. Since the S-layer becomes the outermost layer for the bacterium, it can serve several functions. In addition to determining the external morphology and shape of the cell, S-layers can be extremely resistant to external chemical challenges such as salts, detergents and even enzymes, and thus provide a protective armor for the cell (Schultze-Lam and Beveridge 1994). An S-layer with well-deined pore size is a barrier for compounds with a large molecular weight and therefore acts as a molecular sieve. S-layers may promote cell adhesion to crystalline surfaces and can also provide a method of surface recognition. Before S-layer formation, the proteins and glycoproteins forming this layer have negative charges, while after formation, as the proteins self-assemble into the ordered structure, the charged amino acids are embedded within the layer, and in most cases the inal S-layer presents a net neutral charge at the cell surface. However, some S-layer proteins retain exposed anionic residues and are capable of inducing biomineralization (Schultze-Lam et al. 1993). The cyanobacteria Synechococcus spp, have a 6-fold symmetry S-layer as their outermost surface. This strain showed that strontium and calcium carbonates and other minerals can form on the S-layer (Fortin and Beveridge 2000).

Sheaths Sheaths are well-deined biomineralized structures, such as hollow cylinders, that often surround chains of ilamentous cells, and can be sites of biomineralization. Once biomineralized, the sheaths can remain long after the bacteria have died and decomposed. Leptothrix spp oxidizes ferrous iron in solution by secreting a complex matrix of heteropolysaccharides that catalyzes Fe oxidation and precipitation as iron oxyhydroxide (FeOOH) nanoparticles (Banield et al. 2000). This bacterium thrives in high concentration of Fe and Mn, and leaves behind long sheaths as shown in Figure 11.

Filaments Other bacteria induce the formation of biomineral ilaments. The microbial mineral ilaments of Figure 12 are formed by iron-oxidizing bacteria that have not yet been isolated nor phylogenetically identiied. All these ilaments show an unprecedented ~ 2 nm wide, up to 10 Pm-long, curved pseudo-single crystals of akaganeite (E-FeOOH) in their cores (Chan et al. 2004), as presented in Figure 12B. The ilaments are 20-200 nm wide, tangled, and composed of 2-line ferrihydrite (FeOOH·nH2O), surrounding the akaganeite cores (Fig. 12B). Formation of akaganeite in solution requires the presence of chloride, and is unexpected in fresh water. Chan et al. (2004) therefore suggested that akaganeite formation is catalyzed by organic polymers extruded by the bacteria. In this model, chemical bonds are formed between an organic molecule and ions in solution or amorphous nanoparticles, which are precursors of the crystal cores. As in other biominerals, the organic molecule acts as a template for a particular mineral polymorph, in this case, akaganeite. The paradigm of Figure 3A therefore

Organic-Mineral Interface in Biominerals

173

Figure 11. SEM micrograph of the FeOOH sheaths formed by Leptothrix spp in the Piquette mine, Tennyson, WI. Sample courtesy of Clara S. Chan.

B A

Figure 12. (A) SEM image of mineralized ilaments produced by Fe-oxidizing bacteria in the Piquette abandoned and looded mine in Tennyson WI. The ilaments, approximately 100 nm in diameter are mineralized by FeOOH adhesion to the polysaccharide chains immediately after being extruded by the bacterium. On the right hand-side of the image a thinner, faint, strand is visible, possibly a non-mineralized polysaccharide ibril. Image used with permission of the American Journal of Science from De Stasio et al. (2005), American Journal of Science, Vol. 305, Fig. 4. Sample courtesy of Clara S. Chan. (B) TEM micrograph of a mineralized ilament similar to the one in (A). The outer structure is formed by 1-2 nm wide 2-line ferrihydrite nanoparticles, while the central core of each ilament exhibits a ~2 nmwide crystalline core of akaganeite (E-FeOOH). This crystal core is only 2-3 unit cells wide, and can be identiied as akaganeite by its distinctive crystal spacing (0.75 nm). Image courtesy of Jillian F. Banield.

174

Gilbert, Abrecht, Frazer

applies to this biomineral formation, although akaganeite crystal cores are formed upon aging of the mineral ilaments, not as FeOOH nanoparticles are nucleated and grown. Bacterially extruded polymer ibrils were analyzed using the SPHINX spectromicroscope, and identiied as polysaccharides by comparison of their carbon K-edge XANES spectra with those from representative reference compounds (Fig. 13). Mineralized ilaments also revealed a polysaccharide spectrum at the carbon K-edge. FeOOH nanoparticles form a ~50 nm thick coating around the polysaccharide ibrils, hence the carbon signal is much lower, relative to the uncoated ilaments. Most interestingly, carbon spectroscopy from mineralized ilaments revealed a new peak, which was absent from spectra of non-mineralized polysaccharide ibrils. This spectral signature was interpreted as a V* resonance of a C-O single bond involved in FeOOH binding. It is likely that the C-O groups that interact with FeOOH originate from the carboxyl groups (O=C-O−) of acidic polysaccharides (e.g., alginate). Acidic polysaccharides have an excess of COO− groups that have high afinity for binding positive ions. Chan et al. (2004) concluded that carboxyl groups in the unidentiied bioilm polysaccharide chains must be the sites at which FeOOH amorphous nano-precipitates form chemical bonds, templating for the formation of akaganeite crystal cores upon aging (Fig. 13). This is a relatively simple biomineral, in which the biomineral composite has only three components: an unidentiied COO−-rich polysaccharide, akaganeite crystal ibers and ferrihydrite nanoparticles. Because of its simplicity, its analysis (still incomplete) suggested a possible templation mechanism, which could be inferred at the molecular level (Chan et al. 2004). Biomineralization of these bacterial ilaments has common features with many other biominerals. As a careful reader may have already noticed in all the biominerals reviewed thus far, it is always negatively charged groups along the organic macromolecules that direct

Figure 13. SPHINX image and spectra of the ilaments produced by iron-oxidizing bacteria. (A) mineralized ilaments from the bioilm contain the akaganeite crystal core described in the text. (B) Carbon K-edge XANES spectra from non-mineralized (NM) ibrils and the mineralized (M) ilament in (B), and reference organic molecules: alginate, albumin, lipid and DNA. Notice the similarity of the spectra from the NM ibrils and M ilaments with the polysaccharide spectrum, and the additional structure in the one from the M ilament: the peak at 292.4 eV was assigned to the C-O bond in carboxyl groups. Data from Chan et al. 2004.

Organic-Mineral Interface in Biominerals

175

the interaction with positively charged mineral ions, such as Fe3+ or Ca2+. In the case of the microbial acidic polysaccharides, negatively charged COO− groups are responsible. In cell walls, capsules, S-layers and sheaths, either acidic polysaccharides (rich in COO− groups) or peptide sequences rich in negatively charged amino acids (also exhibiting carboxyl groups) enact the nucleation and biomineral growth. The paradigm of Figure 3, therefore, has one more identiied component: at the interface of the inorganic and mineral components is most frequently, perhaps always, a carboxyl group. We will now discuss the biomineralization paradigm in eukaryotes, and highlight the similarities of the core mechanisms.

THE ORGANIC-MINERAL INTERFACE IN EUKARYOTIC BIOMINERALS Eukaryotic biominerals The majority of animals mineralize at least part of their bodies, usually as internal skeletons or external armors, using a variety of proteins and minerals with calcium carbonates, calcium phosphates, and silica being the most common (Currey 2005). Other eukaryotic biominerals contain a variety of elements, including barium, strontium, iron, manganese, magnesium, copper, zinc, and sulphur. These complex composites, often hierarchically organized, include bone, teeth, eggshell, mollusk shells, crustacean shells, corals, sponge skeletons, the statoliths through which trees sense gravity and grow vertically even on the steepest mountain slopes, the otoliths in the inner ear of most animals, from humans to zebra ish (Söllner et al. 2003), warm jaws (Lichtenegger et al. 2002), and many more composites, in excess of 70 biominerals known nowadays. See Weiner and Dove (2003) and Mann (2001) for the most recent complete lists of biominerals. Eukaryotic biominerals can be distinguished from their abiotic counterparts because of their uniform crystal size and habit and the regular nanostructures that result from biologically controlled mineralization (Weiner and Dove 2003). The control, again, is enacted by the organic matrix and its macromolecules: proteins, glycoproteins, and carbohydrates. Mollusk shells, and in particular that of red abalone (Haliotis rufescens), have been widely studied for their very regular repeating crystalline domains and astounding properties. The nacre layer, or mother of pearl, at the inner surface of the abalone shell has a fracture resistance 3000 times greater than that of aragonite, the pure mineral of which it is composed. The toughening effect is due to well-deined nanolayers of organics at the interfaces between micro-tiles of aragonite (Kamat et al. 2000; Currey 2005). In nacre and many other eukaryotic biomineral structures, the stiff mineral tiles absorb the bulk of the externally applied loads. The alternating organic layers, in turn, provide toughness, prevent the spread of the cracks into the interior of the structure, and even confer a remarkable capacity for recovery after deformation (Smith et al. 1999). Two other structural characteristics of eukaryotic biominerals contribute to the superior mechanical properties of skeletons made from them. First, at the lowest level, they are often made of tiny crystals that are smaller than the “Grifith length” necessary for cracks to spread (Gao et al. 2003). Second, the precision with which they can be laid down (changing their main orientation over a few micrometers, for instance) allows exquisite adaptations to the loads to which the skeletons are subjected (Currey 2005). Nacre is composed of approximately 95 mass percent aragonite and 5 mass percent organic macromolecules. We note that various groups have studied other systems of marine biominerals. For instance, in studies on marine sponges such as Tethya aurantia that form silica needles, research has focused on the role of the proteins and their possible use in organosilicon chemistry. The ultimate goal there is to manufacture silicon based polymeric materials in milder

Gilbert, Abrecht, Frazer

176

conditions than those used in today’s industry. The proteins responsible for biological silica synthesis have received a lot of attention recently, including their very own name, “silicateins” (Shimizu et al. 1998; Shimizu and Morse 2000; Weaver and Morse 2003; Pozzolini et al. 2004). In this section, we will focus on nacre and promote our opinion that the key to nacre formation lies at the organic-mineral interface. Understanding the role of that interface is thus pivotal to the development of biomimetics, that is, the ield that imports biologically inspired concepts and mechanisms into the design and fabrication of new materials.

The nano-structure of nacre Mollusk shell and pearl nacre presents a highly regular brick and mortar arrangement in which aragonite tiles, 500 nm thick along the c-axis, 10-20 μm wide along the a and b axes (Mann 2001), and polygonal in shape, form extremely lat layers (Fig. 14). Subsequent layers of aragonite tiles and organic matrix, composed of silk-like proteins and glycoproteins, keep alternating across the entire thickness of the nacreous layer (Currey 1977; Jackson et al. 1988; Schäffer et al. 1997). The regularly repeating layering of nacre, the semi-transparency of aragonite and the pitch of this periodic structure (500 nm), which falls in the middle of the visible light wavelength range (400-700 nm), all combine to generate the iridescence typical of mother of pearl. As the observation angle varies, the color perceived changes due to the variation in apparent spacing between the semi-transparent layers of crystals. Furthermore, there is considerable crystallographic alignment, with the c-axes of most tiles lying in the direction perpendicular to the tiled planes. Aragonite is an orthorhombic polymorph of CaCO3, whereas the outer prismatic layer of all mollusk shells is formed by columns of the trigonal-rhombohedral calcite polymorph. In the prismatic layer the c-axes are along the long axis of each prismatic column, perpendicular to the shell surface and parallel to the nacreous layer c-axes. Epithelial cells form a layer along the inner surface of the shell, called mantle, and secrete all the macromolecules of the organic matrix (see Figs. 15 and 16). Mechanically nacre is stiff and resistant to fracture; it therefore combines the behavior of lexible materials that can absorb energy by rearranging their molecular conformation (distortion and deformation), and that of hard and stiff materials. On the other hand, it does not suffer from the limitations of its components, as it is neither compliant (as most soft materials)

Figure 14. SEM micrograph of red abalone nacre tiles seen at a fractured edge.

Organic-Mineral Interface in Biominerals

177

Figure 15. Schematic, not to scale, of a vertical cross-section of the outer edge of the shell and mantle of red abalone (Haliotis rufescens) with an enlargement indicating the thickness of each shell structure. The size of the extrapallial space is exaggerated for clarity. Used by permission of the American Chemical Society, from Zaremba et al. (1996) Chemistry of Materials, Vol. 8, Fig. 1a, p. 680.

Figure 16. A proposed model for the organic matrix structure in nacre of the bivavlve shell Atrina serrata, observed in the hydrated state by cryo-TEM. Note that silk was found to be present in both phases, the water-soluble and water-insoluble matrices. Reproduced with permission of Elsevier, from Levi-Kalisman et al. (2001), J. Structural Biology, Vol. 135, Fig. 1, p. 8-17.

nor brittle (as most hard materials). Jackson et al. (1988) reported the Young’s modulus of nacre in the bivalve Pinctada umbricata to be approximately 70 GPa and 60 GPa for dry and wet samples, respectively, whereas the tensile strength is a corresponding 170 MPa and 140 MPa. The work of fracture varies between 350 and 1240 J/m2 (up to 3000u higher than that of CaCO3) (Jackson et al. 1988).

178

Gilbert, Abrecht, Frazer

Interestingly, the organic layers are thick along the c-axis, very thin and hard to detect, or non-existent on the lateral surfaces of tiles (see Fig. 16). The thick organic layers between tile layers are consistent with the model in which sliding of the tiles give nacre its resistance to fracture (Lin and Meyers 2005). It is conceivable that upon sliding, the chains of looped organic macromolecules stretch and break loops without breaking the main molecular chain, thus conferring nacre with its elastic behavior (Smith et al. 1999).

Start and stop signals in nacre growth Several groups studied the growth of abalone nacre and other mollusks (Addadi and Weiner 1985; Lowenstam and Weiner 1989; Belcher et al. 1996; Zaremba et al. 1996; Lin and Meyers 2005). In nacre the organic matrix is a true matrix: its continuous sheets are formed irst (Fig. 3A) and they provide the many nucleation sites, which initiate crystal growth to ill the voids in the three-dimensional organic matrix. This the “start” signal. The position and nature of the nucleation points determines the crystal species and polymorph (Falini et al. 1996), while the structure of the voids in the matrix, presumably, determines crystal habit and size. Laterally, along the a and b axes, crystal growth is stopped by crystal-crystal contacts. At the surface of nacre, which is the growth front, tiles are piled up as stacks of coins, or cones, or Christmas trees, as many authors have called them (see Fig. 15). Lateral growth along the a and b (in plane) directions occurs in these cones until adjoining terraces come in contact. This is one of the “stop” signals, and explains the polygonal appearance of nacre tiles. Vertically, however, the reproducibly perfect thickness of 500 nm must be controlled by another extremely accurate “stop” signal, transduced by the preformed matrix macromolecules. Such signals, and the matrix molecules involved in the growth cessation, are still unknown (Lowenstam and Weiner 1989). Since the aragonite tiles have a relatively small thickness in the c direction (the pure mineral aragonite crystals are much more elongated along that direction, and are much longer than 500 nm), there must be a signal stopping this growth. This signal may be linked to stereochemical adsorption of proteins in the growth of calcite crystals demonstrated by Addadi and Weiner (1985) and Addadi et al. (1987). It can be speculated that the host animal produces the proteins that stop growth in a periodic manner (Lin and Meyers 2005). Another relevant observation is that the size of the aragonite tiles does not depend on the size of the animal. The growth of nacre in space and time has been analyzed in vivo and in vitro using the flat pearl system (Fritz et al. 1994). They found that nacre growth begins with the secretion of proteins that mediate the precipitation of calcite. Other proteins then induce a phase transition from calcite to aragonite (Zaremba 1996; Belcher and Gooch 2000; Lin and Meyers 2005). Some of the matrix macromolecules involved in nacre formation have been identiied. Among the known molecules are the insoluble E-chitin central sheet in each organic matrix layer (Addadi and Weiner 1985), insoluble silk ibroin protein layers above and below this sheet, and unidentiied soluble acidic macromolecules. Even without identiication, however, these macromolecules can be extracted from nacre, exposed to non-shell E-chitin and silk ibroin in a saturated solution of CaCO3 and induce nucleation and growth of aragonite, not calcite (Falini 1996). Aragonite formation is induced by the macromolecules even when seeding calcite crystals! (Thompson et al. 2000). This is particularly clear proof of the role of these unknown acidic macromolecules in polymorph selection, since aragonite is much less stable than calcite. The Thompson et al. (2000) experiment proves that nucleation and polymorph selection are independent in nacre formation. The recent discovery and sequence of Asprich proteins contributes to the clariication of the nature of these acidic macromolecules (Gotliv et al. 2005). Again, as already noted in microbial biominerals, it is the negatively

Organic-Mineral Interface in Biominerals

179

charged amino acids aspartate and glutamate in acidic glycoproteins in mollusk shell that are believed to initiate biomineral formation (Mann 1988, 2001; Mann et al. 2000; Weiss et al. 2000; Weiner and Dove 2003; Gotliv et al. 2005). Stereochemical recognition determines the speciic interactions between aspartic acidrich proteins and certain faces of various calcium dicarboxylate crystals, which are used as model systems. The speciic faces have carboxylate groups oriented perpendicular to the face and can therefore optimally complete the coordination polyhedron around the protein bound calcium ions (Addadi and Weiner 1985). A more recent study, which explored the subtle links between atomic scale dynamics and macroscopic crystal faces, clariies further this issue and reconciles the stereochemical recognition model with the simple mechanistic model of crystal growth by step propagation across crystallographic faces, the terrace-ledge-kink model (De Yoreo and Dove 2004). A possible stereochemical recognition model for abalone nacre is reported in Figure 17.

Synergy of mechanisms for nacre growth Several mechanisms likely conjoin in the formation of nacre. These are:

x Heteroepitaxial nucleation: in this case nucleation and growth of each aragonite tile are detrmined by the organic matrix sheet beneath it (Schäffer et al. 1997).

Figure 17. Unit cell of aragonite: (a) perspective view (b) normal view showing schematic position of (Asp-Y)n and E-pleated organic matrix sheet. Notice protruding Ca ions on (001) face: black atoms are Ca, small black are C and gray are oxygen. This model is in perfect agreement with the paradigm of Figure 3. Reproduced with permission of Elsevier, from Lin and Meyers (2005), Materials Science & Engineering, Vol. 390, Fig. 5, p. 27-41.

180

Gilbert, Abrecht, Frazer x Epitaxial crystal growth of the ith crystal layer, connected to the (i − 1)th crystal layer

by mineral bridges. In this case the crystal would be uninterrupted across different tiles (Schäffer et al. 1997).

x After nucleation on the organic matrix sheet, the growth of aragonite may be mediated or catalyzed by proteins in solution (Falini et al. 1996).

All of these mechanisms, and possibly others not yet discovered, are likely to control in synergy the growth and architecture of nacre.

BIOMINERAL GLUE: THE CARBOXYL GROUP. We have already highlighted the similarity of all organic-mineral interface in prokaryotic biominerals. In the few biominerals thus far analyzed at the molecular level, including prokaryotic and eukaryotic, it is often negatively charged carboxyl groups (COO-) that attract positive ions from solution, and these nucleate for biomineral crystal growth. The carboxyl groups are located either along polysaccharide chains or in acidic amino acids, along the sequences of protein and glycoprotein rich in aspartate or glutamate (Mann 1988, 2001; Mann et al. 2000; Weiss et al. 2000; Weiner and Dove 2003; Gotliv et al. 2005). The very reason the latter are commonly called aspartate and glutamate, and not aspartic and glutamic acid, highlights the fact that they are nearly always deprotonated, and therefore their carboxyl group terminations are negatively charged at physiological pH (the typical pK for both is 4.4; Stryer 1995). Carboxyl-group-rich proteins and/or polysaccharides are the most common and most effective cation-binding macromolecules that any organism can assemble to bind mineral precursors and either control or induce biomineralization. We hypothesize that this is why this molecular functional group was selected in many biominerals as the organic-mineral interface of Figure 3D. In this hypothesis, the carboxyl group is a molecular “glue” of choice for biominerals. Interestingly, even when this hypothesis is not conirmed in speciic biominerals, by analyzing with XANES an intact and pristine biomineral from both the mineral and the macromolecule perspective, it is always possible to identify spectral signatures, even if the bond sites and functional groups involved are not those expected.

CONCLUSION Frequently in biology a gene or a protein is identiied but its function is unknown for decades. Even now, in the most intensely studied genomes, 25% or more of the genes are yet to be associated with a function. In biomineralization it is quite the contrary: most often the function, namely, the formation of a speciic biomineral structure is identiied, but the molecule or molecules responsible for it are unknown. We know, however, that composite biominerals form as a result of complex chemical interactions between organic and inorganic matrices, and that the former acts as a template for the latter, according to a paradigm presented in Figure 3. Few approaches enable the simultaneous analysis of both the organic and mineral components in biominerals and their interface. XANES spectromicroscopy studies of that interface might reveal some of the molecular details of templation mechanisms (e.g., Figs. 13 and 17). We note that at this interface, in diverse eukaryotic and prokaryotic biomineralization, there is frequently a carboxyl group. Acidic amino acids or polysaccharides with excess carboxyl groups are the most common and most effective cation-binding chemical species that any organism can assemble to bind mineral precursors and initiate templation. In this sense COO- is a biomineral preferred “glue”.

Organic-Mineral Interface in Biominerals

181

Ultimately, once the molecular-scale chemistry of the interface is elucidated in more biominerals, it may be possible to harness it and synthesize novel biomimetic composite materials that self-assemble and, as natural biominerals, outperform the sum of their components. Two conceivable avenues towards bio-inspired synthetic materials are: (i) templation by structured organic surfaces, such as self-assembled monolayers or Langmuir-Blodgett ilms and functionalized polymers; and (ii) precipitation from solution with growth modiiers, such as ions, proteins, and synthetic polymers (Han and Aizenberg 2003). The irst biomineral-inspired man-made material synthetically reproduced the nacre assembly with alternating organic and inorganic matrices, using synthetic organic molecules and clay crystals. Remarkably, the tensile strength of the prepared multilayers was similar to that of nacre, and the Young’s modulus approached that of lamellar bone (Tang et al. 2004). We envision a future with many more of these synthetic materials, assembling and structuring themselves at different scales as biominerals have done for well over 500 million years. Impact resistant cars, trains and spacecrafts, in which cracks do not propagate might one day have an attractive mother of pearl luster. In the meantime, the best we can do is to analyze and understand at the molecular level the formation mechanisms of biominerals. The paradigm introduced here includes some prokaryotic and many eukaryotic biomineralization mechanisms. In prokaryotic biominerals, however, the organic components are fewer and simpler to analyze, while the mineral diversity is enormous. This is a distinct advantage offered by prokaryotes for understanding biomineral formation. Following that paradigm as an incomplete but useful starting point, and analyzing as many prokaryotic biominerals as possible, we anticipate that the mechanisms of biomineral formation will be further elucidated. The general rules, the exceptions and the anomalies that are characteristic of the living world will eventually be clear for the biomineral world.

ACKNOWLEDGMENTS We thank Jill Banield and Clara Chan for their expert, friendly and continued collaboration, and most importantly for bringing GDS into the exciting adventure of discovering templation of akaganeite crystal ibers. That experiment sparked her interest in biomineralization mechanisms, a ield now impossible to abandon! We thank Ben Gilbert and Ronke Olabisi for critically reviewing this manuscript. GDS acknowledges the support of the UW-Graduate School, the Department of Physics, the Synchrotron Radiation Center, Air Force grant FA9550-05-1-0204 and NSF grant PHY-0523905. X-PEEM experiments were performed at the UW-Synchrotron Radiation Center, supported by NSF-DMR 0084402.

REFERENCES Addadi L, Moradian J, 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 Addadi L, Weiner S (1985) Interactions between acidic proteins and crystals: stereochemical requirements in biomineralization. Proc Natl Acad Sci USA 82(12):4110-4114 Addadi L, Weiner S (1997) Biomineralization: a pavement of pearl. Nature 389:912-915 Aizenberg J, Weaver JC, Thanawala MS, Sundar VC, Morse DE, Fratzl P (2005) Skeleton of Euplectella sp.: structural hierarchy from the nanoscale to the macroscale. Science 309:275-278 Banield JF, Nealson KH (eds) (1997) Geomicrobiology: Interactions Between Microbes and Minerals. Reviews in Mineralogy, Volume 35. Mineralogical Society of America, Washington D.C.

182

Gilbert, Abrecht, Frazer

Banield JF, Welch SA, Zhang H, Thomsen-Ebert T, Penn RL (2000) Aggregation-based crystal growth and microstructure development in natural iron oxyhydroxide biomineralization products. Science 289:751754 Bazylinski DA, Frankel RB (2003) Biologically controlled biomineralization in prokaryotes. Rev Mineral Geochem 54:217-247 Beauchemin S, Hesterberg D, Chou J, Beauchemin M, Simard RR, Sayers DE (2003) Speciation of phosphorus in phosphorus-enriched agricultural soils using X-ray absorption near-edge structure spectroscopy and chemical fractionation. J Environ Qual 32(5):1809-1819 Belcher AM, Gooch EE (2000) Protein components and inorganic structure in shell nacre. In: Biomineralization. Baeuerlein E (ed) Wiley-VCH, Weinheim, Germany, p 221–249 Belcher AM, Wu XH, Christensen RJ, Hansma PK, Stucky GD, Morse DE (1996) Control of crystal phase switching and orientation by soluble mollusk-shell proteins. Nature 381:56-58 Benfatto M, Della Longa S, Natoli CR (2003) The MXAN procedure: a new method for analysing the XANES spectra of metalloproteins to obtain structural quantitative information. J Synchro Radiat 10:51-57 Beveridge TJ, Fyfe WS (1985) Metal ixation by bacterial cell walls. Can J Earth Sci 22:1893-1898 Bitton G, Freihofer V (1978) Inluence of extracellular polysaccharide on the toxicity of copper and cadmium toward Klebsiella aerogenes. Microb Ecol 4:119-125 Bozek JD, Bancroft GM, Cutler JN, Tan KH (1990) Vibrationally resolved core-level photoelectron spectroscopy: Si 2p levels of SiH4 and SiF4 molecules. Phys Rev Lett 65(22):2757-2760 Carravetta V, Plashkevych O, Ågren H (1998) A theoretical study of the near-edge X-ray absorption spectra of some larger amino acids. J Chem Phys 109:1456-1464 Chan CS, De Stasio G, Welch SA, Girasole M, Frazer BH, Nesterova MV, Fakra S, Banield JF (2004) Microbial polysaccharides template assembly of nanocrystal ibers. Science 303:1656-1658 Currey JD (1977) Mechanical properties of mother of pearl in tension. Proc R Soc Lond B 196:443-463 Currey JD (2005) Hierarchies in biomineral structures. Science 309:253-254 De Stasio G, Casalbore P, Pallini R, Gilbert B, Sanita F, Ciotti MT, Rosi G, Festinesi A, Larocca LM, Rinelli A, Perret D, Mogk DW, Perfetti P, Mehta MP, Mercanti D (2001) Gadolinium in human glioblastoma cells for gadolinium neutron capture therapy. Cancer Res 61:4272-4277 De Stasio G, Dunham D, Tonner BP, Mercanti D, Ciotti MT, Coluzza C, Perfetti P, Margaritondo G (1993) Aluminum in rat cerebellar neural cultures. Neuroreport 4:1175-1178 De Stasio G, Frazer BH, Gilbert B, Richter KL, Valley JW (2003) Compensation of charging in X-PEEM: a successful test on mineral inclusions in 4.4 Ga old zircon. Ultramicroscopy 98:57-62 De Stasio G, Mercanti D, Ciotti MT, Droubay TC, Perfetti P, Margaritondo G, Tonner BP (1996) Synchrotron spectromicroscopy of cobalt accumulation in granule cells, glial cells and GABAergic neurons. J Physics D29:259-262 De Stasio G, Schmitt MA, Gellman SH (2005) Spectromicroscopy at the organic-inorganic interface in biominerals. Am J Sci 305: De Yoreo JJ, Dove PM (2004) Shaping crystals with biomolecules. Science 306:1301-1302 De Yoreo JJ, Vekilov PG (2003) Principles of crystal nucleation and growth. Rev Mineral Geochem 54:57-93 Einstein A (1905) Über einen die Erzeugung und Verwandlung des Lichtes betreffenden heuristischen Gesichtpunkt (On a Heuristic point of view about the creation and conversion of light). Annalen der Physik 17:132-148 Falini G, Albeck S, Weiner S, Addadi L (1996) Control of aragonite or calcite polymorphism by mollusk shell macromolecules. Science 271:67-69 Fitts JP, Persson P, Brown GE Jr, Parks GA (1999) Structure and bonding of Cu(II)-glutamate complexes at the gamma-Al2O3-water interface. J Colloid Interface Sci 220:133-147 Fortin D, Beveridge TJ (2000) Mechanistic routes to biomineral surface development. In: Biomineralization. Baeuerlein E (ed) Wiley-VCH, Weinheim, Germany, p 7-24 Fortin D, Ferris FG, Beveridge TJ (1997) Surface-mediated mineral development by bacteria. Rev Mineral 35: 161-180 Frankel RB, Bazylinski DA (2003) Biologically induced mineralization by bacteria. Rev Mineral Geochem 54:95-114 Frazer BH, Girasole M, Wiese LM, Franz T, De Stasio G (2004) Spectromicroscope for the PHotoelectron Imaging of Nanostructures with X-rays (SPHINX): performance in biology, medicine and geology. Ultramicroscopy 99:87-94 Frazer BH, Waychunas GA, Xu H, De Stasio G (2005) Quantitative mapping of the ferrous/ferric iron ratio in oxide minerals using synchrotron spectromicroscopy. Am Mineral, submitted Fritz M, Belcher AM, Radmacher M, Walters DA, Hansma PK, Stucky GD, Morse DE (1994) Flat pearls from biofabrication of organized composites on inorganic substrates. Nature 371:49-51 Fuerst JA, Webb RI (1991) Membrane-bounded nucleoid in the eubacterium Gemmatata obscuriglobus. Proc Natl Acad Sci 88:8184-8188

Organic-Mineral Interface in Biominerals

183

Gao H, Ji B, Jäger IL, Arzt E, Fratzl P (2003) Materials become insensitive to laws at nanoscale: lessons from nature. Proc Natl Acad Sci 100:5597-5600 Gilbert B, Banield JF (2005) Molecular scale processes involving nanoparticulate minerals in biogeochemical systems. Rev Mineral Geochem 59:109-156 Gilbert B, Frazer BH, Belz A, Conrad PG, Nealson KH, Haskel D, Lang JC, Srajer G, De Stasio G (2003a) Multiple scattering calculations of bonding and X-ray absorption spectroscopy of manganese oxides. J Phys Chem A 107:2839-2847 Gilbert B, Frazer BH, Naab F, Fournelle J, Valley JW, De Stasio G (2003b) X-ray absorption spectroscopy of silicates for in situ sub-micrometer mineral identiication. Am Mineral 88:763-769 Gilbert B, Frazer BH, Zhang H, Huang F, Banield JF, Haskel D, Lang JC, Srajer G, De Stasio G (2002) X-ray absorption spectroscopy of the cubic and hexagonal polytypes of zinc sulide. Phys Rev B 66(245205) Gilbert B, Perfetti L, Fauchoux O, Redondo J, Baudat PA, Andres R, Neumann M, Steen S, Gabel D, Mercanti D, Ciotti MT, Perfetti P, Margaritondo G, De Stasio G (2000) The spectromicroscopy of boron in human glioblastomas following administration of BSH. Phys Rev E 62:1110-1118 Glasauer S, Langley S, Beveridge TJ (2001) Sorption of Fe (hydr)oxides to the surface of shewanella putrifaciens: cell-bound ine-grained minerals are not always formed de novo. Appl Environ Microbiol 67:5544-5550 Gotliv BA, Kessler N, Sumerel JL, Morse DE, Tuross N, Addadi L, Weiner S (2005) Asprich: a novel aspartic acid-rich protein family from the prismatic shell matrix of the bivalve Atrina rigida. Chembiochem 6: 304-314 Grush MM, Chen J, Stemmler TL, George SJ, Ralston CY, Stibrany RT, Gelasco A, Christou G, Gorum SM, Penner-Hahn JE, Cramer SP (1996) Manganese L-edge X-ray absorption spectroscopy of manganese catalase from Lactobacillus plantarum and mixed valence manganese complexes. J Am Chem Soc 118: 65-69 Han Y-J, Aizenberg J (2003) Effect of magnesium ions on oriented growth of calcite on carboxylic acid functionalized self-assembled monolayer. J Am Chem Soc 125:4032-4033 Hertz H (1887) Über einen Einluß des ultravioletten Lichtes auf die electrische Entladung (An effect of ultraviolet light on electrical discharge). Annalen der Physik Chemie 31:983-1000 Jackson AP, Vincent JF, Turner RM (1988) The mechanical design of nacre. Proc R Soc London B 234:415440 Juniper SK, Tebo BM (1995) Microbe-metal interactions and mineral deposition at hydrothermal vents. In: The Microbiology of Deep-Sea Hydrothermal Vents. Karl DM (ed) CRC Press, New York, p. 219–253 Kamat S, Su X, Ballarini R, Heuer AH (2000) Structural basis for the fracture toughness of the shell of the conch strombus gigas. Nature 405:1036-1040 Kaznacheyev K, Osanna A, Jacobsen C, Plashkevych O, Vahtras O, Ågren H, Carravetta V, Hitchcock AP (2002) Innershell absorption spectroscopy of amino acids. J Phys Chem A 106:3153-3168 Kemner KM, Kelly SD, Lai B, Maser J, O’loughlin EJ, Sholto-Douglas D, Cai Z, Schneegurt MA, Kulpa CF Jr, Nealson KH (2004) Elemental and redox analysis of single bacterial cells by X-ray microbeam analysis. Science 306:686-7 Kilcoyne ALD, Tyliszczak T, Steele WF, Fakra S, Hitchcock P, Franck K, Anderson E, Harteneck B, Rightor EG, Mitchell GE, Hitchcock AP, Yang L, Warwick T, Ade H (2003) Interferometer-controlled scanning transmission X-ray microscopes at the Advanced Light Source. J Synchrotron Rad 10:125-136 Labrenz M, Druschel GK, Thomsen-Ebert T, Gilbert B, Welch SA, Kemner KM, Logan GA, Summons RE, De Stasio G, Bond PL, Lai B, Kelly SD, Banield JF (2000) Formation of sphalerite (ZnS) deposits in natural bioilms of sulfate-reducing bacteria. Science 290:1744-1747 Langley S, Beveridge TJ (1999) Effect of O-side-chain-lipopolysaccharide chemistry on metal binding. Appl Environ Microbiol 65:489-498 Lawrence JR, Swerhone GD, Leppard GG, Araki T, Zhang X, West MM, Hitchock AP (2003) Scanning transmission X-ray, laser scanning, and transmission electron microscopy mapping of the exopolymeric matrix of microbial bioilms. Appl Environ Microbiol 69:5543-5554 Lerotic M, Jacobsen C, Schafer T, Vogt S (2004) Cluster analysis of soft X-ray spectromicroscopy data. Ultramicroscopy 100:35-57 Levi-Kalisman Y, Falini G, Addadi L, Weiner S (2001) Structure of the nacreous organic matrix of a bivalve mollusk shell examined in the hydrated state using cryo-TEM. J Struct Biol 135(1):8-17 Lichtenegger HC, Schöberl T, Bartl MH, Waite H, Stucky GD (2002) High abrasion resistance with sparse mineralization: copper biomineral in worm jaws. Science 298:389-392 Lin A, Meyers MA (2005) Growth and structure in abalone shell. Mater Sci Eng A 390:27-41 Lindsay MR, Webb RI, Strous M, Jetten MS, Butler MK, Forde RJ, Fuerst JA (2001) Cell compartmentalisation in planctomycetes: novel types of structural organisation for the bacterial cell. Arch Microbiol 175:413429

184

Gilbert, Abrecht, Frazer

Lopez-Garcia P, Duperron S, Philippot P, Foriel J, Susini J, Moreira D (2003) Bacterial diversity in hydrothermal sediment and epsilonproteobacterial dominance in experimental microcolonizers at the Mid-Atlantic Ridge. Environon Microbiol 5:961-976 Lowenstam HA (1981) Minerals formed by organisms. Science 211(4487):1126-1131 Lowenstam HA, Weiner S (1989) On Biomineralization. Oxford University Press, Oxford Lower SK, Hochella MF Jr, Beveridge TJ (2001a) Bacterial recognition of mineral surfaces: nanoscale interactions between Shewanella and -FeOOH. Science 292:1360-1363 Lower SK, Tadanier CJ, Hochella MF Jr (2001b) Dynamics of the mineral–microbe interface: use of biological force microscopy in biogeochemistry and geomicrobiology. Geomicrobiology J 8:63-76 Maier RM, Drees KP, Neilson JW, Henderson DA, Quade J, Betancourt JL (2004) Microbial life in the Atacama desert. Science 306:1289 Mann K, Weiss IM, Andre S, Gabius HJ, Fritz M (2000) The amino-acid sequence of the abalone (Haliotis laevigata) nacre protein perlucin. Eur J Biochem FEBS 267:5257-5264 Mann S (1988) Molecular recognition in biomineralization. Nature 332:119-124 Mann S (2001) Biomineralization: Principles and Concepts in Bioinorganic Materials Chemistry. Vol 17. Oxford University Press, Oxford Mount AS, Wheeler AP, Paradkar RP, Snider D (2004) Hemocyte-mediated shell mineralization in the eastern oyster. Science 304(5668):297-300 Myers CR, Nealson KH (1990) Respiration-linked proton translocation coupled to anaerobic reduction of manganese(IV) and iron(III) in Shewanella putrefaciens MR 1. J Bacteriol 172:6232-8 Myneni SC, Tokunaga TK, Brown GE Jr (1997) Abiotic selenium redox transformations in the presence of Fe(II,III) oxides. Science 278:1106-1110 Myneni SCB (2002a) Formation of stable chlorinated hydrocarbons in weathering plant material. Science 295: 1039-1041 Myneni SCB (2002b) Soft X-ray spectroscopy and spectromicroscopy studies of organic molecules in the environment. Rev Mineral Geochem 49:485-579 Myneni SCB, Brown JT, Martinez GA, Meyer-Ilse W (1999) Imaging of humic substance macromolecular structures in water and soils. Science 286:1335-1337 Nealson KH, Stahl DH (1997) Microorganisms and biogeochemical cycles: what can we learn from stratiied communities? Rev Mineral 35:5-34 Newman DK, Banield JF (2002) Geomicrobiology: how molecular-scale interactions underpin biogeochemical systems. Science 296:1071-1077 Pickering IJ, Prince RC, Salt DE, George GN (2000) Quantitative, chemically speciic imaging of selenium transformation in plants. Proc Natl Acad Sci USA 97:10717-10722 Pozzolini M, Sturla L, Cerrano C, Bavestrello G, Camardella L, Parodi AM, Raheli F, Benatti U, Muller WEG, Giovine M (2004) Molecular cloning of silicatein gene from marine sponge Petrosia iciformis (Porifera, Demospongiae) and development of primmorphs as a model for biosiliciication studies. Marine Biotech 6:594-603 Schäffer TE, Ionescu-Zanetti C, Proksch R, Fritz M, Walters DA, Almqvist N, Zaremba CM, Belcher AM, Smith BL, Stucky GD, Morse DE, Hansma PK (1997) Does abalone nacre form by heteroepitaxial nucleation or by growth through mineral bridges? Chem Mater 9:1731-1740 Schultze-Lam S, Beveridge TJ (1994) Physicochemical characteristics of the mineral-forming S-layer from the cyanobacterium synechococcus strain GL24. Can J Microbiol 40:216-223 Schultze-Lam S, Thompsom JB, Beveridge TJ (1993) Metal ion immobilization by bacterial surfaces in fresh water environments. Water Poll Res J Canada 28:51-81 Shimizu K, Cha J, Stucky GD, Morse DE (1998) Silicatein alpha: Cathepsin L-like protein in sponge biosilica. Proc Natl Acad Sci USA 95:6234-6238 Shimizu K, Morse DE (2000) The biological and biomimetic synthesis of silica and other polysiloxanes. In: Biomineralization. Baeuerlein E (ed) Wiley-VCH, Weinheim, Germany, p 207–220 Sleytr UB (1997) Basic and applied S-layer research: an overview. FEMS Microbiol Rev 20:5-12 Smith BL, Schaffer TE, Viani M, Thompson JB, Frederick NA, Kindt J, Belcher A, Stucky GD, Morse DE, Hansma PK (1999) Molecular mechanistic origin of the toughness of natural adhesives, ibres and composites. Nature 399:761-763 Söllner C, Burghammer M, Busch-Nentwich E, Berger J, Schwarz H, Riekel C, Nicolson T (2003) Control of crystal size and lattice formation by Starmaker in Otolith biomineralization. Science 302:282-286 Stöhr J (1992) NEXAFS Spectroscopy. Springer-Verlag, Berlin Stryer L (1995) Biochemistry, 4th edition, WH Freeman and Co, New York. See page 22. Sturchio NC, Antonio MR, Soderholm L, Sutton SR, Brannon JC (1998) Tetravalent uranium in calcite. Science 281:971-973 Suzuki Y, Kelly SD, Kemner KM, Banield JF (2002) Radionuclide contamination: Nanometre-size products of uranium bioreduction. Nature 419:134

-

Organic-Mineral Interface in Biominerals

185

Takai K, Moser DP, DeFlaun M, Onstott TC, Fredrickson JK (2001) Archaeal diversity in waters from deep South African gold mines. Appl Environ Microbiol 67:5750-5760 Tang Z, Kotov NA, Magonov S, Ozturk B (2003) Nanostructured artiicial nacre. Nature Mat 2:413-418 Tebo BM, Bargar JR, Clement BG, Dick GJ, Murray KJ, Parker D, Verity R, Webb SM (2004) Biogenic manganese oxides: properties and mechanisms of formation. Ann Rev Earth Planet Sci 32:287-328 Templeton AS, Ostergren JD, Trainor TP, Foster AL, Traina SJ, Spormann A, Brown GE Jr (1999) XAFS and XSW study of the distribution of Pb(II) sorbed to bioilms on alpha-Al2O3 and alpha-FeOOH surfaces. J Synchrotron Radiat 6:642-644 Thompson JB, Paloczi GT, Kindt JH, Michenfelder M, Smith BL, Stucky GD, Morse DE and Hansma PK (2000) Direct observation of the transition from calcite to aragonite growth as induced by abalone shell proteins. Biophys J 79:3307-3312 Toner B, Fakra S, Villalobos M, Warwick T, Sposito G (2005) Spatially resolved characterization of biogenic manganese oxide production within a bacterial bioilm. Appl Environ Microbiol 71:1300-1310 Tonner BP, Droubay T, Denlinger J, Meyer-Ilse W, Warwick T, Rothe J, Kneedler E, Pecher K, Nealson KH, Grundl T (1999) Soft X-ray spectroscopy and imaging of interfacial chemistry in environmental specimens. Surf Interface Anal 27:247-258 Tyliszczak T, Warwick T, Kilcoyne ALD, Fakra S, Shuh DK (2004) Soft X-ray scanning transmission microscope working in an extended energy range at the Advanced Light Source. AIP Conference Proceedings, SRI. San Francisco Urrutia M, Kemper M, Doyle R, Beveridge TJ (1992) The membrane-induced proton motive force inluences the metal binding ability of Bacillus subtilis cell walls. Appl Environ Microbiol 58:3837-3844 Veis A (2003) Mineralization in organic matrix frameworks. Rev Mineral Geochem 54:249-289 Villalobos M, Bargar J, Sposito G (2005) Mechanisms of Pb(II) sorption on a biogenic manganese oxide. Environ Sci Technol 39(2):569-76 Weaver JC, Morse DE (2003) Molecular biology of demosponge axial ilaments and their roles in biosiliciication. Microscopy Res Tech 62(4):356-367 Weiner S, Dove PM (2003) An overview of biomineralization processes and the problem of the vital effect. Rev Mineral Geochem 54:1-29 Weiss IM, Kaufmann S, Mann K, Fritz M (2000) Puriication and characterization of perlucin and perlustrin, two new proteins from the shell of the mollusk haliotis laevigata. Biochem Biophys Res Commun 267: 17-21 Zaremba CM, Belcher AM, Fritz M, Li Y, Mann S, Hansma PK, Morse DE, Speck JS, Stucky GD (1996) Critical transitions in the biofabrication of abalone shells and lat pearls. Chem Mater 8:679-690 Zawislanski PT, Benson SM, Terberg R, Borglin SE (2003) Selenium speciation, solubility, and mobility in land-disposed dredged sediments. Environ Sci Technol 37(11):2415-2420

Reviews in Mineralogy & Geochemistry Vol. 59, pp. 187-210, 2005 Copyright © Mineralogical Society of America

8

Catalysis and Prebiotic Synthesis James P. Ferris NY Center for Studies on the Origins of Life and Department of Chemistry Rensselaer Polytechnic Institute Troy, New York, 12180-3590, U.S.A. [email protected]

INTRODUCTION Little is know about the origins of life on Earth. Most scientists believe this event occurred some time within a billion years after the Earth formed 4.6 billion years ago (Ga). It is also possible that the Earth was “seeded” by life transported here by another body like a meteorite or by extra-terrestrials. Most scientists in this ield assume that life originated on Earth or in our solar system because there would be little data on which to base a proposal for the origin of life at a location outside our solar system. As it is we have a very rudimentary knowledge of the environments on the primitive Earth in the irst billion years on our planet. Some relevant books and reviews include Brack (1998), Fry (2000), Zubay (2000), Orgel (2004), and Ferris (2005). Ten years ago it appeared that we had made good progress on understanding about when life arose and what the environmental conditions on the Earth were at that time. Carbon isotope studies on rocks present on the Earth 3.8 Ga suggested life arose in or slightly after that time period (Mojizsis et al. 1996). In addition, microfossils found in rocks dated to be 3.5 Ga suggested were consistent with the presence of life 3.5 Ga (Schopf 1993). These data have been challenged recently (Brasier et al. 2002) so it is not certain the proposed microfossils were originally living organisms. Also the carbon isotope studies have been challenged (Moorbath 2005). But new and entirely different indings suggest that that the Earth had liquid water and an environment suitable for life 4.3 Ga. (Watson and Harrison 2005). In addition it has been proposed that the early Earth had an atmosphere with a mixing ratio of hydrogen of 0.3 (Tian et al. 2005). This suggests the possibility of an atmosphere compatible with reduced organic compounds. So the good news is that this is an active area of research and there is a chance that enough data will accumulated so that a more accurate picture of the conditions on the early Earth will arise from the current confusion. One of the irst problems to consider in studies on origin of life is what is life? How would the irst, very primitive life form be recognized? This life form would have just barely transited from the non-living to the living so may be lost in the large excess of inanimate material from which it arose. The irst life was probably expending all its energy just staying alive so would be hard to recognize. Another problem is the only models of life we have are the highly evolved forms that surround us on the Earth today. Was the biochemistry of the irst life just a simpler form of our protein – DNA world or was it entirely different? We don’t know. Many scientists have proposed a deinition of life. The deinitions proposed often describe a model of the irst life that the individual is investigating. Those postulating an RNA world propose the need for RNA or structures that were precursors to RNA. Those investigating the need for cell membranes propose the need for the origin of life in a contained like a vesicle to 1529-6466/05/0059-0008$05.00

DOI: 10.2138/rmg.2005.59.8

188

Ferris

protect it from losing the essential molecules to maintain this irst life. My favored deinition is a system of biomolecules that it capable of replication and mutation. This deinition does not require that the assemblage of molecules be conined within a membrane nor does it specify the nature of the replicating system present. It has the advantage of being applicable to a variety of different replicating systems that may live in a variety of different environments.

FORMATION OF THE SOLAR SYSTEM The Big Bang, the starting point for the formation of the universe occurred 13.7 Ga. This event has been described as the simultaneous appearance of plasmas of primary particles everywhere at the same time in the Universe. The next stage in the evolution in the Universe was the spontaneous formation of the lighter elements (hydrogen, helium, lithium and beryllium) from the primary particles by the spontaneous combination of neutrons and protons. Star formation occurred next where there was a greater density of particles and elements in the Universe. This process was the result of a greater density of elements and particles in a region of the Universe that condensed to form stars. Once star formation took place there was a suficiently high concentration of particles and elements to form carbon in a three-body reaction. The next stage in the evolution of the universe was star formation where the primary particles were concentrated in a star with the initial elements where the close proximity resulted in the formation of carbon via a three-body reaction. Once carbon formed the elements up to and including iron-58 formed spontaneously. Additional energy was required to form the elements with atomic masses greater that iron. This additional energy was provided when the star’s fusion reactor shut down because the nuclear fuel, hydrogen and helium, were consumed. The star then collapsed and then exploded violently and the energy released powered the formation of the elements with masses greater than that of iron-58. The energy released during the supernova blew the elements that formed by the nuclear fusion reactions into the interstellar medium. Some stars containing high levels of carbon distributed large amounts of carbon into the interstellar medium. These violent explosions distributed clouds of dust in the interstellar medium that contained carbon, silicates and an array of other elements. The action of cosmic rays and other energy sources on the carbon initiated reactions with the abundant interstellar hydrogen to generate simple hydrocarbons. These dust clouds have a lifetime of about 108 years. They then collapse, possibly as a result of an external force like a supernova, and star formation together with the formation of the planets and other bodies in the solar system was repeated. When a dust cloud collapses to form a solar system the bulk if the material in the cloud forms the protostar and the remainder of the dust forms small bodies called planetismals that continue to accrete dust and smaller bodies to eventually form plants as they orbit the protostar. Comets form in the outer solar system where they accumulate gases, water, ice, silicates, and organic compounds. The material in comets has undergone the least change of all the orbiting material because it is far away from the radiation released by the protostar when it accumulated suficient mass to initiate fusion reactions. Comets may have delivered organics and water to the surface of the primitive Earth. After the comets formed they orbited the Sun in the vicinity of the giant planets, Jupiter, Saturn and Uranus. The strong gravitational forces of these planets accelerated many of these comets out of the Solar System to the Oort cloud. This process probably resulted in the passage of comets near the inner solar system and some of the comets may have impacted with the Earth and brought organics and water to its surface. Asteroids orbit the Sun between Mars and Jupiter in our solar system. This collection of small bodies never coalesced to planets because the strong gravitational ield emanating from

Catalysis & Prebiotic Synthesis

189

Jupiter perturbs them. There are collisions between asteroids as they orbit the Sun. These collisions resulted in the formation of dust and smaller bodies (meteorites) that are ejected from the asteroid belt. Some of this material impacts on the Earth. It is postulated from spectral studies that that some of the asteroids contain organic compounds and the dust and meteorites may released from these asteroids may have brought the organic compounds and water to the primitive Earth that initiated the origin of life. This process of accumulating dust and meteorites on the Earth continues today but at a much slower rate than it occurred on the primitive Earth. It is estimated that about 30,000 tons of dust per year is accreted on the surface of the Earth today. This general model for the formation of our solar system suggests that there will be an abundance of solar systems around other stars in the universe. This appears to be the case since over 150 extrasolar planets have been discovered. Most of the extrasolar planets are equal to or greater than the size of Uranus while few even approach the size of Earth. Planets the size of Uranus or greater may be gas giants and would not be likely to have life, as we know it, on them. The failure to detect Earth-size planets is probably not due to their absence but rather to the detection method used. It is based on the gravitational perturbation of the motion of their star, which is very small for low mass, Earth-size planets. If Earth-size planets are discovered does this mean that the conditions on them are conducive to life? Not necessarily, but it does seem likely that some of these rocky bodies will at least have microbial life. It should be noted that some scientists believe that the Earth was formed under a very restricted set of favorable conditions that may not be present on comparable Earth-size bodies. They also feel that civilizations of the type present on Earth are unlikely to be present on these Earth-size planets (Ward and Brownlee 2000).

THE EARLY EARTH It is dificult to obtain data greater than 4 Ga about the primitive Earth’s from the rock record. This is because plate tectonics has resulted in the subduction of most of the Earth’s crust where it was subjected to high temperatures and pressures. This resulted in the destruction of much of the evidence in the rock record of earliest life on Earth. One thing that appears to be known from the rock record is that the oxidation level of the Earth’s crust and mantle 3.8 Ga is the same as it is today (Delano 2001). The discovery of zircons that are 4.0–4.3 Ga old opened up a new window on the primitive Earth in the 4–4.5 Ga time period (Watson and Harrison 2005). This suggests that the impacts of larger bodies, such as comets or meteorites that would have heated the Earth to over 100 °C had decreased to a low level by 4.3 Ga. The high temperatures and pressures resulting from plate tectonics do not alter these refractory zircons. They also contain inclusions of other elements that may provide additional insight into the chemical processes on the primitive Earth prior to 4 Ga. There is evidence from lunar material of a sharp increase in the impacts with the Moon at 3.9 Ga. If it is assumed that the same heavy bombardment occurred on the Earth at 3.9 Ga it could have extinguished much of the life on Earth at that time. If this happened it might have required that the origin of life occurred again after these impact decreased in intensity. It is possible that some life survived the impacts in a niche, like the deep ocean, and this life was the ancestor of life on Earth today. Alternatively, the irst life on Earth may have arisen after 3.9 Ga.

Atmosphere If the oxidation level of the Earth’s crust and mantle was the same 3.9 Ga as today then it is unlikely that a reducing atmosphere was present on the primitive Earth (Delano 2001). If the recent claim that the Earth’s atmosphere contained 30% hydrogen is correct (Tian et al. 2005)

190

Ferris

this hydrogen may have altered the overall oxidation state of the atmosphere but may not have changed the oxidation states of the other gases present in the atmosphere. The atmosphere of the Earth is believed to have formed as a result of outgasing the Earth’s crust and mantle. It is unlikely that the gases emitted 3.8 Ga would differ from the carbon dioxide, water, sulfur dioxide and other oxidized gases emitted from volcanoes today. It is not clear whether the presence of 30% hydrogen in an atmosphere containing oxidized gases atmosphere could have been a precursor to the reduced organics that resulted in the protein and nucleic acids, central to life on the Earth today. The large deposits of limestone and other carbonates on Earth suggest that carbon dioxide was an important constituent of the primitive atmosphere. The oceans may have been slightly acidic as a consequence the dissolved carbon dioxide and the acidic oxides of sulfur emanating from volcanoes. The rate of precipitation carbonate minerals would have determined the length of time the oceans and other bodies of water were acidic and were probably not favorable for the origin of life. It is encouraging that there are an increasing number of new indings about the ancient Earth. The ield of the origins of life suffers from the lack of deinitive data about the environment on the primitive Earth. Consequently, there are a plethora of hypotheses but few facts the support or refute them.

Primary sources of simple organics Earth’s atmosphere. The Miller-Urey experiment was the irst laboratory experiment designed to investigate routes to organic compounds on the primitive Earth. In this experiment Stanley Miller passed an electric discharge through a mixture of methane, ammonia, hydrogen and water vapor at 100 °C. After allowing the experiment to proceed for one week the water was analyzed and Miller detected the presence of amino acids. Later he also found hydroxy acids, carboxylic acids and other products. This experiment was carried out with reduced carbon and nitrogen compounds because Harold Urey believed that the atmosphere of the early Earth contained these gases. Up to the present time scientists did not think the primitive Earth had a reducing atmosphere and it was generally agreed that he Miller–Urey reaction conditions are not a valid model for the source of simple organics on the primitive Earth. The proposal that the atmosphere of the primitive Earth contained 30% hydrogen reopens the debate on whether reduced carbon and nitrogen compounds were formed by the Miller–Urey reaction. Even if reduced organics are formed, the rapid photolysis of carbon, nitrogen and sulfur compounds such as methane, ammonia and hydrogen sulide by solar ultraviolet light suggests that they were present in very small amounts. Meteorites. Meteorites emanating from the asteroid belt may have been an important source of organics on the primitive Earth. It is estimated that greater than 1021 kg of the original asteroid belt reached the Earth’s surface in the form of dust and meteorites (Vokrouhlicky and Farinella 2000). This would correspond to a layer of material weighing 5 u 106 kg/m2 if spread evenly over the surface of the Earth. If the carbon content of that material is about ~1% this would have been equivalent to a layer of carbon 25 m thick over the surface of the Earth (Private communication from Michael Gaffey). Small meteorites are not destroyed when they hit the Earth’s atmosphere or surface since they break into small chunks. Large meteorites also survive their impact with the atmosphere but they generate so much energy when they hit the Earth’s surface that the organics present in them are destroyed. Dust particles from asteroids and comets loat down through the atmosphere and make a soft landing on the Earth’s surface. Meteorites are particularly important since they are a direct source of the extraterrestrial material delivered to the primitive Earth. The Murchison meteorite, which fell in the town of Murchison, Australia, was collected shortly after it fell and was therefore less likely to have been contaminated organic compounds indigenous to the Earth. It contains about 2% organic material

Catalysis & Prebiotic Synthesis

191

with the bulk of it present as a polymeric organic matrix called kerogen and the remainder of the material present as soluble small organic molecules. The latter is a complex mixture of organic compounds including amino acids, purines, pyrimidines, and carboxylic acids (Table 1) (Pizzarello et al. 2001). Other carbonaceous meteorites contain some of the same compounds so it appears likely that these compounds were present on the primitive Earth. Comets. As noted previously comets formed in the outer regions of the solar system where the solar radiation is lower so the structures of the organics more closely relect those in the dust cloud from which the solar was formed. It has been possible to determine the structures of some of the compounds by spectroscopic analysis of the coma and tails of comets (Table 2). The energetic particles and UV radiation of the Sun not only melts the ice in a comet to release the organics it also degrades high molecular weight organics and generates the lower molecular weight degradation products. Missions are in progress to directly collect cometary dust and return it to the Earth for structure analysis. For example, the Stardust mission sent a probe that arrived at comet Wild2 in January 2004. It collected dust emanating from the comet and is now returning to Earth where the captured dust will be parachuted to Earth on January 15, 2006. The European Space Agencies Rosetta Mission launched a probe to comet in March 2004. Upon arrival in 2014

Table 1. Soluble organics in the Murchison meteorite.a Class Aliphatic hydrocarbons Aromatic hydrocarbons Dicarboxylic acids Carboxylic acids Pyridine carboxylic acids Dicarboximides Sulfonic acids Amino acids Amines Amides Hydroxy acids a

Concentration (ppm)

Compounds Identiied

> 35 15-28 >30 >300 >7 >50 67 60 8 n.d. 15

140 87 17 20 7 3 4 74 10 4 7

Adapted from (Pizzarello 2001)

Table 2. Some organic compounds observed in comets. Name

Formula

Name

Formula

Methanol

CH3OH

Formic acid

HCOOH

Formamide

HCONH2

Methyl formate

HCOOCH3

Methane

CH4

Acetylene

C2H2

Ethylene

C2H4

Ethane

C2H6

Methylacetylene

CH3C2H

Hydrogen cyanide

HCN

Acetonitrile

CH3CN

192

Ferris

it will launch a lander to the comet to analyze it’s surface and subsurface. The probe will undertake extensive analysis of the comet and volatiles emanating from it. Hydrothermal systems. Hydrothermal systems provide an ecosystem where heat from the Earth’s interior, rather than sun light, provides the energy to support life at the bottom of the sea. These “black” and “white smokers” occur at crustal spreading centers where heat driven circulation of water through crustal material brings dissolved compounds in proximity to iron-containing magma. Here the iron reduces oxidized substances in the water. For example, dissolved sulfate is converted to sulides and the resultant smoker is a precipitate of metal sulides formed in the neighborhood of the vents when they exit the 300 °C vent into the 4 °C ocean water. Laboratory studies have demonstrated the reduction of molecular nitrogen at temperatures of 500-700 °C and pressures of 0.1 GPa. Nitrogen oxides have also been reduced under high temperature and pressures of hydrothermal systems using iron sulides and magnetite as reductants (Brandes et al. 1998). More complex organics that may have been destroyed by the high temperatures in a hydrothermal vent would have been stable if formed a short distance away from the vent. The heat low from the vent also drives the circulation of ocean water through vent regions that is “off axis” or away from the area close to the magma. The combination of the reducing agents, e.g. metal sulides that precipitated in the vicinity of the vent, the possible mineral catalysts in the crust and the lower temperatures could have served to generate complex structures that were stable under these reaction conditions. Some laboratory studies have been successful modeling the chemistry in hydrothermal systems. Laboratory indings have demonstrated the possibility of generating methane thiol (CH3SH) from carbon dioxide (Heinen and Lauwers 1996). Carbon dioxide is reduced to acetic acid in the presence of methane thiol (Huber and Wachtershauser 1997). Dipeptides are formed in the reaction of amino acids with carbon monoxide, an iron and nickel sulide catalyst or methane thiol at 100 °C (Huber and Wachtershauser 1998). Reaction of carbon monoxide with iron sulfate at 250 °C generates the Krebs cycle compound pyruvate (Cody et al. 2000). So far it has not been possible to demonstrate the formation of more complex biomolecules in simulations of the reactions in hydrothermal systems but studies of this type are in progress. At the present time it appears that hydrothermal systems may have served as a source of simple organics that were converted to more complex structures in other environments on the primitive Earth.

PREBIOTIC ROUTES TO BIOPOLYMER PRECURSORS RNA world Biopolymers are essential structures in life today. Protein enzymes catalyze the synthesis and transformation of chemical processes that drive metabolic and other processes in the cell. DNA stores the genetic information in its sequences. DNA transfers this information to RNA, which in turn brings it to the ribosome, the site of protein synthesis. RNA brings genetic information to the ribosome and it also catalyzes a key step in protein synthesis in the ribosome. The observation that RNA could both store genetic information in its sequences and catalyze reactions, process that are carried our by the DNA and proteins in contemporary life, led to the proposal that the irst life could have had one essential biopolymer, RNA. This was the genesis of the RNA world. A big advantage of this postulate was only one biopolymer would have to be formed by prebiotic processes rather than two. Perhaps some simple peptides may have served as catalysts in the RNA world. Eventually the RNA evolved the ability to catalyze the synthesis of proteins that in turn evolved to catalyze the synthesis of DNA.

Catalysis & Prebiotic Synthesis

193

Since the emphasis in this review is a world where one type of biopolymer that can both store information and catalyze reactions the prebiotic synthesis of proteins or protein-like structures will not be emphasized.

The structure of and prebiotic synthesis of RNA monomers RNA is a biopolymer composed of monomers linked together. The monomer is composed of three units, bases, ribose and phosphate (Fig. 1). Bases. Two purines, adenine and guanine and two pyrimidines, uracil and cytosine are the principal bases in RNA. Purines are present in small amounts in the Murchison and other carbonaceous meteorites. They are also formed from hydrogen cyanide (HCN) via two routes. The polymerization of HCN generates a black substance called the “HCN Polymer” from which adenine may be extracted in low yields after water hydrolysis. The second route is via the HCN tetramer, an intermediate in the polymerization process, that when photolyzed yields a substituted imidazole (Fig. 2). This imidazole may also be prepared by the reaction of the HCN tetramer with formamidine. Reaction of the substituted imidazole with HCN generates adenine. A variety of other purines can be prepared by the reaction of the imidazole formed by photolysis of the HCN tetramer with other simple molecules.

B (a)

RNA Monomers

HO 5'

O 3'

B

O P O 5'

-O

2'

O-

OH OH nucleoside

N Bases

B= N N H adenine

N

N

N

2'

O

N

O

N H cytosine

O

NH2

HN

NH2 RNA Oligomer

N N HO 5'-end

N N

O 3' O OH O P O 5' O OO O P O-

O N

NH

N

N

OH O

O O

NH2 O

HN N

NH2 N

O OH O O P O O O3'-end

Figure 1. RNA structural elements.

N H uracil

Pyrimidines

Purines

(c)

2'

NH2 NH

N N H guanine

O 3'

OH OH activated nucleotide

O N

P O

5'

O-

OH OH nucleotide

NH2 (b)

O 3'

B

O

N

OH OH

194

Ferris

Figure 2. The reaction pathway from HCN to some purine bases.

Pyrimidines may be synthesized by the reaction of cyanoacetylene or cyanoacetaldehyde, (the hydrolysis product of cyanoacetylene) with cyanate, guanidine or urea. Pyrimidines have also been isolated from the product mixture obtained by the hydrolysis of the HCN polymer (Ferris and Hagan 1984). Purines and pyrimidines have also been isolated by the hydrolysis of the products formed the polymerization of a mixture of HCN and ammonia that was kept at −78 °C for 27 years (Miyakawa et al. 2002). Most of the products obtained in this study were comparable to those obtained in the reactions carried out for shorter times at room temperature (Ferris et al. 1978). The various sources observed for purines and pyrimidines described above suggest that it is likely that they were present on the primitive Earth. Ribose. The ive-carbon sugar ribose is the backbone of nucleoside monomers. Initially it was proposed that the prebiotic synthesis of ribose proceeded by the Formose reaction, the self-condensation of formaldehyde catalyzed by divalent metal ions like calcium. This route dropped from favor because the yield was very low and about 35 other sugars that would have reactivity comparable to that of ribose were also formed so that it was not obvious why these didn’t compete with ribose in subsequent reactions to form nucleosides. Recent investigations suggest that ribose could have been present on the primitive Earth. It may have been possible to selectively isolate the cyanamide (NH2C{N) adduct of ribose from other reaction products (Springsteen and Joyce 2004). The adduct forms 7-fold faster than the comparable adduct with arabinose and 30 times faster than the adduct with glucose. It crystallizes from water under conditions where none of the other adducts crystallize and this adduct is much more stable than free ribose. The only problem is that cyanamide reacts with formaldehyde, glycolaldehyde and glyceraldehydes, compounds present in the Formose reaction mixture to give adducts that do not react further to give the ribose adduct. So it is necessary to postulate that cyanamide appeared in these mixtures after these intermediates were consumed. Three other routes to ribose have been reported that suggest that ribose was present on the primitive Earth. In one approach the use of a magnesium ion - lead ion mixture as a catalyst for the Formose reaction. This catalytic cocktail reduced the number of products from 35 to the 4 possible pentoses (Zubay 1998). Also of interest was the observation that the catalyst also directs the conversion of any one of the 4 pentoses to a mixture of the all four of them. This lead-magnesium ion mixture also catalyzes the conversion of hexoses to the 4 pentoses. A second prebiotic approach to ribose is the reaction of formaldehyde with calcium hydroxide in the presence of borate. The borate forms a complex with the hydroxyl groups

Catalysis & Prebiotic Synthesis

195

of ribose and ribose precursors so that the subsequent reaction with formaldehyde is to form hexoses and polymeric products are inhibited (Ricardo et al. 2004). A third route to ribose is a stepwise synthesis by irst reacting glycolaldehyde phosphate with formaldehyde to generate glyceraldehydes phosphate. This in turn reacts with glycolaldehyde to give ribose 2,4- diphosphate. Unfortunately neither phosphate is in 5c-position where it may have functioned as a point for linking nucleotides together via the 5c-phosphate group (Müller et al. 1990). Nucleosides. There is an early report of a prebiotic synthesis of purine nucleosides. It is a dry phase heating reaction of purine bases with ribose in the presence of sea salts or MgCl2 to get 1–8% yields of purine nucleosides (Fuller et al. 1972). No nucleosides were observed when the same reaction procedure was performed using pyrimidine bases in place of the purine. The low yields of purine nucleosides and the absence of formation of pyrimidine nucleosides indicates that this is not a likely prebiotic pathway to nucleosides. Limited progress has been made in other approaches to the synthesis of nucleosides but so far none of the proposed prebiotic syntheses has generated nucleosides in suficiently high yields to believe that these monomers would be present in amounts large enough to have been starting materials for RNA synthesis. This is an area that requires new ideas and experiments. Nucleotides. Nucleotides can be synthesized from nucleosides by heating them in the solid phase with acid phosphates like ammonium dihydrogen phosphate (NH4H2P04) (Osterberg and Orgel 1972; Osterberg et al. 1973). The reactions are catalyzed by amides like urea. The reaction probably proceeds by driving off the ammonia of the ammonium dihydrogen phosphate to give phosphoric acid that catalyzes the phosphorylation. When this dry heating reaction is carried out with uridine about a 70% yield of a mixtures of the phosphorylated adducts of uridine is obtained. Linear polyphosphates are formed by heating acid phosphates like sodium dihydrogen phosphate (NaH2PO4) (Osterberg and Orgel 1972). The linear polyphosphates are converted to cyclic trimetaphosphate (Fig. 3), which reacts with nucleosides in basic solution to yield triphosphates or with 5c-nucleotides under less basic conditions to give triphosphates in a series of reaction steps (Fig. 3). The yields of phosphorylated products formed by dry heating suggest that these may have been abundant on the primitive Earth if there was a plausible prebiotic synthesis of nucleosides. Another concern is that the conditions under which ribose synthesis and

Figure 3. A reaction pathway from a 5c-AMP and trimetaphosphate to ATP.

196

Ferris

phosphorylation reactions occur are very different. It is dificult to imagine how this series of steps required for nucleoside synthesis occurred on the primitive Earth. More experimental studies are required here.

Vesicles Vesicles, also called lipsomes, are small enclosed, compartments separated from their aqueous environments by a lipid bilayer. It is proposed that enclosed compartments similar to vesicles may have contained a suite of molecules that constituted the irst life. It has been observed that an organic fraction obtained from the Murchison meteorite formed vesicles and other structures when mixed with water (Deamer and Pashley 1989). It is not likely that these vesicles are enclosed within a lipid bilayer. It is known that under the proper conditions vesicles form from linear carboxylic acids containing 9 or more carbon atoms when the pH of the solution is close to the pKa of the carboxylic acids. Since carboxylic acids are present in the Murchison meteorite this may be a possible explanation for the formation of vesicles. As discussed in detail below, montmorillonite clay, which enhances the rate of formation of vesicles from carboxylic acids, also catalyzes the formation of RNA oligomers (Hanczyc et al. 2003; see discussion in “Prebiotic Polymerization of RNA Oligomers.” The prebiotic synthesis of linear carboxylic acids may have proceeded in hydrothermal systems where the high temperatures, pressures and the presence of iron may have produced them from carbon monoxide and carbon dioxide (McCollum et al. 1999; Rushdi and Simoneit 2001). The above discussion assumes that the irst life required a container to maintain the integrity of the living system. Since this irst replicater was extremely simple and probably was unable to catalyze the formation of the fatty acids needed to form vesicles it may have replicated while attached to a surface. In the simplest case the community of RNAs may have been attached to a mineral surface where it captured the necessary nutrients of life as they lowed past. There they could have also bound the metal ions that may have been required for the functioning of this extremely primitive system.

CHIRALITY Life on Earth is composed of a speciic handedness of the molecules that it utilizes. For example the proteins are composed of L-amino acids and the RNA and DNA of D-nucleotides (Fig. 4). An L-amino acid is the mirror image of the corresponding D-amino acid and each mirror image molecule is called an enantiomer. It is not known why contemporary life has amino acids with the L-coniguration and nucleotides with the D-conformation. One proposal is that it was a chance event that occurred during evolution when life based on D-amino acids and L-nucleotides was not able to compete with L-amino acids and D-nucleotides because of a favorable mutation in the latter. The latter life forms gradually took over the Earth and these conigurations have been frozen in place ever since. Another theory is based on the observation of circularly polarized infrared light in the vicinity of interstellar dust clouds (Bailey et al. 1998). Circularly polarized light can be either left- or right-handed. It is proposed that if there is circularly polarized infrared radiation then there will also be circularly polarized UV radiation as well. It is known from laboratory studies that irradiation of a mixtures of enantiomers with one handed circularly polarized UV light results in the faster rate of loss of one of the enantiomers so that an excess of the other enantiomer is left behind (Flores et al. 1977). This postulate has some support by the observation of an excess of the L-enantiomer in some of the amino acids found in the Murchison meteorite (Pizzarello and Cronin 2000). If it is assumed that there is a correlation between the handedness of the circularly polarized light and the L-conigurations of the interstellar amino acids reaching the Earth via meteorites then one could conclude that the

Catalysis & Prebiotic Synthesis

197

Figure 4. Enantiomers. (a) the amino acid alanine, (b) the nucleoside adenosine.

observed conformations resulted from the interstellar circularly polarized light. So far there has been no proof of this postulate.

PREBIOTIC POLYMERIZATION OF RNA MONOMERS The condensation of RNA monomers to oligomers in aqueous solution at pHs near neutrality is not favored energetically. The reactivity of the 5c-phosphate group must be activated for the reaction to proceed. In contemporary biochemistry the 5c-phosphate group is activated by the attachment of a diphosphate group to the 5c-nucleotide. The nucleotide triphosphate group is quite stable in aqueous solution and only generates RNA polymers in the presence of a polymerase enzyme. It was observed that some amine derivatives of the 5cphosphates (phosphoramidates) are effective activating groups for the monomeric nucleotides. Adducts of imidazole (Fig. 5a,b) and 1-methyladenine (Fig. 5c) have been observed to be effective activating agents (Weimann et al. 1968; Prabahar and Ferris 1997). Some montmorillonite clays have been shown to be effective catalysts of the oligomerization of both phosphorimidazolides and phosphoro-1-methyladenylides. It was possible to generate oligomers (short polymers) that contained 10 monomer units (10 mers) (Ferris and Ertem 1992). It was then observed that a synthetic 10 mer could be elongated to a 40-50 mer using “feeding reactions” where the activated monomer is added daily to the reaction mixture. The 40-50 mers were observed after feeding for 14 days (Ferris et al. 1996;

Figure 5. Activated nucleotide monomers where B is a purine or pyrimidine base. (a) nucleoside 5cphosphorimidazolide (ImpB), (b) nucleoside 5c-phosphoro-2-methylimidazolide (2-MeImpB), (c) nucleoside 5c-phosphoro-1-methyladenium (1-MeadpB), (d) D- nucleoside 5c-phosphorimidazolide (D-ImpB).

198

Ferris

Ferris 2002). The previous studies were performed using imidazole as the activating group. It was recently observed that 40 mers were formed in 1 day when 1-methyladenine is used as the activating group (Huang and Ferris 2003). It is possible to form longer oligomers of RNA because the catalyst selectively directs the reaction pathway to the formation of a limited number of reaction products. If the catalyst were not selective then the formation of longer oligomers would not occur because all possible isomers would be formed and these side reactions would consume all the activated monomers before 40-50 mers are formed. It has been calculated that it would be required to form 1048 isomers weighting 1028 grams, an amount equal to the mass of the Earth, to prepare a mixture that contained two identical 40 mers (Joyce and Orgel 1999). Sequence selectivity was demonstrated in the montmorillonite-catalyzed reaction of a mixture of the four activated nucleotides. Here 80 % of the dimers formed were 8 of the 16 possible dimers (Ertem and Ferris 2000; Miyakawa and Ferris 2003). In the second study of the reaction of active ImpA with ImpC the pentamer fraction contained 4 main products in yields of 4.3–13 % while random synthesis would have generated 512 products with a 0.2% yield of each (Miyakawa and Ferris 2003). Selectivity was also observed in the formation of phosphodiester bonds. The montmorillonite-catalyzed reaction of ImpA generates oligomers in which about 67% of bonds formed were 3c,5c-linked. In the absence of catalysis the percentage of 3c,5c-phosphodiester bonds are about 20%. The outcome of the reaction of D,L- ImpA has been investigated to determine if oligomers form from a D,L-mixture. It was not expected that there would a selective reaction of only one of the enantiomers during montmorillonite catalysis because montmorillonite does not have D- and L-conformations. A preponderance of homochiral oligomers over what was expected was observed. That is there was an excess of both all D and all L-enantiomers over the expected yields of D,L- and L,D- dimers. Similar results were observed with the trimers formed. Heterochiral products predominate in the reaction of D,L-ImpU.

NON-ENZYMATIC TEMPLATE-DIRECTED SYNTHESIS OF RNA RNA is essential to early life because information is stored in its sequences of purines and pyrimidines. The process of template-directed synthesis preserves this information. Information is preserved in contemporary biological systems because the sequences in the RNA can be replicated in a two-step process from the complementary chain (Fig. 6a). The key to the replication process is the selective hydrogen bonded interactions with G and C (Fig. 6b) and those with A and U. In contemporary biology the monomers of the nucleotides that are activated

Figure 6. Template-directed synthesis. (a) schematic replication of a template of the 8 mer of C and the formation of an 8 mer of G, (b) Watson-Crick hydrogen bonding between G and G.

Catalysis & Prebiotic Synthesis

199

with the triphosphate group interact with their complementary bases in the RNA and then the monomers are coupled by a polymerase enzyme to form the complementary chain. The original chain and the complementary chains dissociate and the template directed synthesis takes place on the complementary chain to regenerate the original chain. The errors in this process provide a mechanism by which evolution occurs. Leslie Orgel and coworkers discovered that template-directed synthesis does occur to a limited extent in the absence of catalysis. The best example of this is the formation of oligomers of G by the reaction of 2-MeImpG (Fig. 5b) on a poly(C) template (Fig. 6a) (Inoue and Orgel 1981). Unfortunately the other possible template-directed syntheses are not as successful. The synthesis of oligo(A)s on a poly(U) template gives lower yields of shorter of oligo(A)s. The reactions of activated pyrimidine nucleotides on polypurine templates does not yield oligopyrimidines. Replication of an RNA chain is not possible in the G–C system because the formation of the complementary oligo(C)s on a poly(G) template was not successful. These observations suggest that the RNA world will need a catalyst (RNA?) or some changes in the reaction conditions or reagents to initiate the replication of RNA. It has been estimated that a library consisting of about 1020 sequences of 40 mers is likely to contain at least one self-replicating RNA molecule (Joyce and Orgel 1999). Mineral and/or metal ion catalysis may have generated the catalytic RNAs that catalyzed RNA replication but to date none been discovered.

ALTERNATIVE GENETIC SYSTEMS It was noted in “Nucleosides” that there are deiciencies with the currently proposed prebiotic syntheses of nucleosides and nucleotides. In addition, a plausible prebiotic formation of the activated monomers has not been accomplished. These deiciencies prompted the search for simpler monomers from which to form a replicating system with the expectation that this simpler system would in turn invent the RNA world. Carboxylic acid ester groups, polypeptides with interacting charged side chains, link some of the proposed polymers. In one instance it was proposed that a replicating clay mineral initiated the irst life that catalyzed the formation of organic molecules that were more eficient than the clay as catalysts and they took over clay life to form life based on organic molecules (Cairns-Smith 1982). There are no experimental data to support the later hypothesis. A more complicated group of alternative biopolymers have incorporated some of the structural features of RNA, such as being linked by phosphoester bonds and the utilization of the same bases and base pairings as RNA. Eschenmoser and coworkers initiated a research program to undertake a systematic investigation of structures that are similar to RNA and not the prebiotic synthesis of biopolymers (Eschenmoser 1999). Their irst synthetic target was homo-DNA in which the ive-membered ring of RNA was replaced with a six-membered ring. Next they prepared pyranose-RNA (p-RNA), Figure 7. Nucleotide which also had a six-member ring in which the phosphodiester bond of threose. differed in location from that of homo-DNA. p-RNA also differed from homo-DNA in that would form a double helix with a complementary strand. A recent addition to the RNA analogs is TNA, which is based on the four-carbon sugar threose (Fig. 7). This is especially interesting since it not only forms double helices with complementary TNA strands it also forms a double helix with a complementary RNA strand even though threose only contains one less carbon atom than ribose (Schoning et al. 2000). This structure has the advantage of a simple evolution of RNA from TNA by template-directed synthesis of RNA monomers on a TNA template. The TNA would have to have catalyzed the biosynthesis of the RNA monomers.

200

Ferris

A similar approach to preRNAs is peptide RNA (PNA), which was prepared as an analog of RNA, since it has about the same distances between the monomer units as RNA (Fig. 8). The other advantage of PNA is that it has no chiral centers. It forms double helices with itself and with RNA. Once it forms a double helix with RNA it assumes the chirality of the RNA to which it is bound. It can be used as a template for the formation of the complementary RNA (Böhler et al. 1995; Schmidt et al. 1997). One problem is that it has no charges on its backbone so longer oligomers are not likely to be soluble in water. There are no complete prebiotic synthesis of any of these pre-RNAs so it has not been demonstrated how they could have been formed on the primitive Earth (Miller 1997).

Figure 8. PNA. (a) monomer, (b) oligomer.

EXAMPLES OF MINERAL AND METAL ION CATALYSIS IN PREBIOTIC CHEMISTRY Minerals and metal ions have been investigated as potential catalysts for prebiotic reactions. There are few general principles to guide the experimentalist in choosing a particular metal ion to catalyze a proposed prebiotic process. Their selection has been mainly based on (1) their ability to bind reactants (2) their utilization in contemporary biological systems to catalyze reactions similar to proposed prebiotic reactions (3) just try available metal ions. Examples of minerals and metal ions used to bring about prebiotic reactions and what is known about the source of their catalytic activity will be outlined here. It is my view that these catalysts were central to the formation of the biopolymers that were essential to the origin of life since it is unlikely that polymers could have formed without them.

Non-catalytic formation of biopolymers; polypeptides The simplest role that a mineral could play in the formation of biopolymer is to serve as a matrix on which polymers would bind and growth. These minerals are not catalysts so the key to the polymer growth is the increase in the strength of binding of the oligomer to the mineral as it increases in length. In addition, the rate of oligomer formation must be greater than its rate of hydrolysis. This procedure requires that the substrate binds to the mineral but does not require that the mineral catalyze the reaction (Ferris et al. 1996; Orgel 1998). Feeding reactions in which a condensing agent that is used to drive polypeptide formation from amino acids on mineral surfaces may generate polypeptides containing up to 50 monomer units. Hydroxyapatite, Ca5(PO4)3(OH), binds the acidic amino acid glutamic acid via the Ca2+ of the mineral. After 50 feedings of the glutamic acid together with the condensing agent, 1,1c-carbonyldiimidazole (CDI) (Fig. 9), in the presence of hydroxyapatite yields polypeptides containing up to 45 monomer units. Similar results

Figure 9. The condensing agent 1,1c-carbonyldiimidazole (CDI).

Catalysis & Prebiotic Synthesis

201

were obtained using the clay mineral illite (Hill et al. 1998). Aspartic acid oligomers were formed on hydroxyapatite using 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDAC) as the condensing agent but oligomer formation decreased when CDI was used as the condensing agent (Liu and Orgel 1998). The positively charged amino acid arginine was elongated on both illite and FeS2 using CDI as the condensing agent. Variation of the structure of the amino acids used and the surface of the mineral indicates that the extent of oligomer formation depends on a number of factors (Hill and Orgel 1999). This procedure has not been successful with amino acids with not net charge, such as alanine, since they have weak binding interactions with the mineral surfaces.

Montmorillonite catalysis of RNA synthesis Structure and properties of montmorillonite. Montmorillonite is a 2D sheet of corner linked SiO4 tetrahedra bound to layers of edge-linked AlO6 octahedra (Fig. 10). The montmorillonite platelets associate with each other via interlayer cations and van der Waals forces. Montmorillonite is a clay mineral Key: formed by the weathering of volcanic Oxygen ash. Its composition varies depending on the other elements present in the Hydroxyl exchangeable cations environment where it is formed. Mag2+ 2+ water layers Aluminum nesium ion (Mg ) ferrous ion (Fe ) and ferric ion (Fe3+) are often incorporated Silicon into the octahedral layer positions. In addition, Al3+ may be substituted for Magnesium, Iron tetravalent silicon (Si4+) in the tetrahedral silicate layer. While the theoretical formula is Al4Si8O20OH)4, the actual formula for a Wyoming montmorillonite, with Fe3+ and Mg2+ in the octahedral layer and Al in the tetrahedral layer and 0.67 monovalent exchangeable cations is (Al2.33Fe0.68Mg0.47)(Si7.71 Figure 10. The layer structure of montmorillonite. Al0.29O20(OH)4X0.67. Since the number of oxygen atoms in the montmorillonite sheet is constant the lattice has a net negative charge that is balanced by the charge of 0.67 cations. The associated cations that neutralize this negative charge usually reside in the interlayers between the montmorillonite sheets. Dry montmorillonite expands when water is added due to the solvation of the interlayer cations. When organics bind in the interlayer the sheets come further apart if the binding energy between the sheets is less than the binding energy of the organic compound. Van der Waals interaction between organic molecules and the silicate layer is often the force that attracts organic compounds to bind in the clay interlayer. Montmorillonite found in deposits on Earth usually contains a mixture of cations in its interlayer that relect those present in the environment where the montmorillonite is formed. Na+, Ca2+ usually predominate. It is possible to exchange this mixture of cations with a single cation in the laboratory to obtain a homoionic montmorillonite. This is usually done before investigating possible catalytic activity so as to avoid chemical processes due to the interlayer cations. Substances with multiple negative charges may bind at the acidic edges of the montmorillonite sheets. These include polyphosphates, dicarboxylic acids and polyanionic polymers. The edges have Al+3 with three oxygens bound to it. The fourth oxygen atom is not there because that is the bond where the sheet was broken. A water molecule will bind at Al+3 via the lone pair of electrons on the water molecule. This coordination enhances the acidity

202

Ferris

of the water molecules and generates an acidic site that can transfer a proton to a basic or a negatively charged molecule. RNA oligomer formation. In most of our studies on the montmorillonite-catalyzed formation of RNA oligomers homoionic Na+-montmorillonite was used. It was observed that other alkali and alkaline earth cations also gave catalytically active montmorillonites with the exception of Mg2+. When other divalent cations like Ni2+ and Cu2+ serve as the exchangeable cations the montmorillonite does not catalyze RNA oligomer formation. It was also observed that the acid titration exchange procedure of (Banin 1973; Banin et al. 1985) was essential for the preparation of most of these catalytically active montmorillonite. In most instances an alternative procedure where the clay is treated with an excess of NaCl to give Na+montmorillonite does not generate a catalytic montmorillonite. Not all montmorillonites catalyze oligomer formation. There may be a correlation between the increase in the iron content of the clay lattice and the increase in catalytic activity (Ferris et al. 1990). This potential correlation breaks down for the nontronites where Fe3+ replaces almost all the Al3+. Binding studies established that purine nucleotides bind more strongly to montmorillonite than do the pyrimidine nucleotides. This is consistent with the greater van der Waals interactions between the purine ring and the silicate surfaces of the montmorillonite than the smaller pyrimidine ring (Kawamura and Ferris 1999; Ertem and Ferris 2000). That the reaction of activated mononucleotides occurs in the clay interlayer was determined by irst treating the montmorillonite with tetraalkyl ammonium salts (Ertem and Ferris 1998). Dodecyltrimethylammonium cations (Fig. 11a) inhibit oligomer formation while tetramethyl ammonium ions (Fig. 11b) did not. The inhibition resulting from the substituted quarternary ammonium salts is due to the positively charged quarternary ammonium group binding to the negatively charged clay lattice and hydrophobic interactions between the long alkyl groups of the quarternary ammonium salts in the interlayer. These alkyl groups ill up the interlayer and bind so strongly that they are not replaced by the activated RNA monomers.

Figure 11. Quarternary ammonium salts. (a) dodecyltrimethylammonium, (b) tetramethylammonium.

It has also been observed that the deoxypyrophosphate derivative, dA5cppdA, (Fig. 12) inhibits oligomer formation (Wang and Ferris 2001). It is proposed that the strong inhibition by dA5cppdA is the result of the van der Waals interaction of both adenine rings to the silicate layers so that it is not possible for an RNA monomer, with only one adenine ring, to displace it from the interlayer. The chemical reactivity of the 3c-OH of deoxynucleotides is much less than that of the 2c,3c-hydroxyl groups of ribonucleotides so that reaction at the 3chydroxyls of dA5cppdA does not occur (Ferris and Kamaluddin 1989). The possibility that the catalysis occurred at the edge sites of montmorillonite was also

Figure 12. Deoxyadenosine pyrophosphate (dA5cpp5cdA).

Catalysis & Prebiotic Synthesis

203

investigated (Ertem and Ferris 1998). Trimethylsilyl groups bound to the edge silanol groups or luoride ion bound to the trisubstituted edge aluminum groups (Fig. 10) resulted in some levels of inhibition as determined from the slightly shortened chain lengths of the oligomers formed. This inding suggests the absence of edge catalysis because the formation of oligomers was not strongly inhibited. As noted in “Prebiotic Polymerization of RNA Monomers,” the preparation of oligomers containing 40–50 mers has been achieved by two experimental approaches. These are feeding reactions where activated monomers are added daily for 14 days to a decamer primer yielded 50 mers and reactions where 1-methyladenine is used instead of imidazole as an activating group generate 35–40 mers in a one day (Huang and Ferris 2003). It is not clear why it has not been possible to generate oligomers longer than 40–50 mers. One possibility is the binding of the oligomers to montmorillonite increases, as the oligomers grow longer. Since oligomers must be mobile on the clay surface for them to react with activated monomers to increase in length, very strong binding may decreases their mobility on the surface so they block the catalytic sites in the interlayer. Another possibility is that the oligomers formed ill up the interlayer so no more activated monomers can be accommodated there. Uranyl ion catalysis of RNA oligomer formation. Uranyl ion (UO22−) catalyzes of the formation of oligo(C)s, oligo(A)s and oligo(U)s with chain lengths up to 10, 16 and 10 mers, respectively, from the corresponding phosphorimidazolides of nucleotides starting compounds (Sawai et al. 1989) (Sawai et al. 1992). There are some similarities between the reactions catalyzed by UO22− and montmorillonite. The optimal pH for both catalysts is 8; cyclic dimers are formed in high yields from the activated pyrimidine nucleotides. Mainly 2c,5c- phosphodiester bonds form in reactions of activated pyrimidine nucleotides catalyzed by either UO22− or montmorillonite. There are some differences in the reaction of ImpA catalyzed by UO22− or montmorillonite. Neither cyclic dimers nor high yields of 3c,5c-linked oligomers are formed in the UO22− catalyzed reactions of ImpA. Higher yields of oligomers are observed in the UO22− catalyzed reactions because the extent of the hydrolysis reaction of the activated monomers is much lower than that observed with montmorillonite. Catalysis of RNA oligomer formation by lead and other metal ions. Lead (Pb2+), zinc (Zn2+) and lanthanide metal ions have been observed to catalyze the formation of 5–10 mers from ImpA. Pb2+ is second only to UO22− as a catalyst. The ratio of the yields of oligomers formed from the lanthanide metal ions to the yield of hydrolyzed activated monomer increases with the atomic weight of the metal ion where the highest ratio is 49 with Lutitium (Lu3+) (Sawai 1988; Sawai and Yamamto 1996). Zn2+ gave results similar to those of Pb2+ except the longest oligomers observed were only tetramers and the oligomers formed had mainly 2c,5clinks (Sawai and Orgel 1975). The Pb2+ catalysis is enhanced when the reaction is performed in the eutectic liquid phase of water at −18 °C for 20–40 days when most of the water is present as ice crystals. Ice predominates reactants and the Pb2+ catalysts are concentrated in the small liquid phase. Here oligomers as long as 17 mers are formed with overall yields of 80–90% (Kanavarioti et al. 2001; Monnard et al. 2002, 2003). The higher yields and longer oligomers are due in part to the high concentrations of reactants and the slower rate of hydrolysis of the activated monomers. It is postulated that base stacking and ordered monomer assemblies on the ice surfaces enhances the chain lengths in the eutectic phase but there is no speciic data that supports this claim.

Metal ion catalysis of template-directed synthesis Zn2+ and Pb2+ catalyze the template-directed synthesis of RNA oligomers from ImpG and ImpA (Fig. 6a) (Sleeper et al. 1979; Lohrmann et al. 1980). Pb2+ catalyzes the poly(C) template-directed synthesis of mainly 2c,5c-linked oligo(G)s that containing up to 40 mers.

204

Ferris

Pb2+ also catalyzes the formation oligo(A)s on a poly(U) template but the maximum chain length is only 7 mers. Here 75% of the phosphodiester bonds are 3c,5c-linked. Zn2+ catalysis of the template-directed synthesis of oligo(G)s in the reaction of ImpG on a poly(C) template gives oligomers as long as 30 mers that are 75% 3c,5c-linked. When the reaction is performed in the absence of Zn2+ the oligomers formed are mainly 2c,5c-linked. Zn2+ does not catalyze the template-directed formation of oligo(A)s. In the investigation of the reaction pathway it was observed that 2 Zn2+ per ImpG gave optimal oligomer yields. One of the zinc ions can be replaced with Mg2+. X-ray structure analysis of the Zn2+-nucleotides complexes show Zn2+ binding at both N7 and the phosphate of the nucleotide. It is tentatively proposed that in this complex one Zn2+ binds to N(7) and either Zn2+ or possibly Mg2+ binds to the imidazole (Birdson and Orgel 1980). No postulates on the reaction pathway of the template-directed reactions catalyzed by Pb2+ have been proposed other than the coordination of the lead ion with the 2c-hydroxyl of the nucleotide. The main basis for this suggestion is the preferential formation of 2c,5c-phosphodiester bonds (Sleeper et al. 1979). Other metal ions have been tested as catalysts for the template-directed synthesis of oligo(G)s from ImpG besides Pb2+ and Zn2+ (van Rode and Orgel 1980). Of the 17 tested, 4 (Bi3+, Sn2+, Sb3+, and Mn2+) exhibited catalytic activity. All of these generated lower yields of oligomers than were formed by Pb2+catalysis and all the oligomers were linked by 2c,5cphosphodiester bonds. The divalent transition metal ions that would be expected to coordinate with imidazole (Co2+, Ni2+, Cu2+and Cd2+) protect ImpG from hydrolysis and also inhibited catalysis. It is proposed that these metal ions bind to the imidazole of the activating group where they protect ImpG from hydrolysis and at the same time they inhibit oligomer formation. The research on the metal ion catalysis of template-directed synthesis stopped when it was observed that when the imidazole activating group was replaced with 2-methylimidazole metal ions were not required to enhance the formation 30 mers of oligo (G) s on a poly(C) template (Inoue and Orgel 1981).

POSSIBLE CATALYTIC REACTION PATHWAYS Since metal ions, RNA templates and montmorillonite clay all catalyze the reactions of nucleotides activated with imidazole at the 5c-position it may be possible to gain insight into the reaction pathway on montmorillonite by reviewing the mechanisms proposed for the catalysis by metal ions and RNA templates.

Metal ions UO22− and Pb2+ are effective while Zn2+ and lanthanides metal ions are much less effective catalysts of the reactions of ImpN where N is A, U, G, C and I. In most cases the products formed are mainly 2c,5c-linked. It is proposed that UO22− forms a complex with the activated monomers that binds them in the correct orientations for the 2c-hydroxyl group of one monomer to react with the activated phosphate of the other monomer (Shimazu et al. 1993). In studies where the nucleotide base is inverted in the nucleotide, a D-nucleotide, (Fig. 5d) the longest oligomers formed are half as long as those formed from the natural E-nucleotide and the bonds formed were linked mainly by 3c,5c-phosphodiester bonds (Sawai et al. 1997). It was not noted whether the UO22− complex, formed from UO22− monomers, binds D-ImpA in the proper orientations for reaction and whether this complex can also bind oligomers in the correct orientation for elongation. Reactions catalyzed by soluble metal ions are probably initiated by complex formation between the metal ion and the activated nucleotide. If the metal ion binds two or more activated

Catalysis & Prebiotic Synthesis

205

nucleotides then they may be held in the proper orientation for reaction in the complex. An alternative proposal is the metal ion binds the activated nucleotide in a way that activates the 2c-hydroxyl group of the activated nucleotide. This postulate is supported by the observation that the best catalyst among the lanthanide series of metal ions is Lu3+, in which the bound water has the greater acidity, and as such should be the most effective metal ion for activating the 2c-hydroxyl group of the nucleotide (Sawai and Yamamto 1996).

Metal ion catalysis of template-directed synthesis of RNA oligomers Metal ion catalyzed template-directed synthesis enhances the length of the oligomers formed over that of the catalysis by either the metal ions or the templates alone. The most successful metal ion catalysts are Pb2+ and Zn2+. Pb2+ catalyzes the poly(C) template-directed synthesis of mainly 2c,5c- linked G oligomers and the synthesis of mainly 3c,5c-linked an oligomers on a poly (U) template. Zn2+ catalyzes the template-directed synthesis of a mainly 3c,5c-linked G oligomers on a poly(C) template. The observation that metal ions were not responsible for catalysis when the mononucleotide is activated with 2-methylimidazole suggests that they may not have a direct role in the activation of the mononucleotide for reaction. Rather the metal ion may just change the geometry of the double helix of the poly(C) template with the growing G oligomers so that the activated nucleotides are lined up for reaction (Birdson and Orgel 1980). This favorable orientation does not require the activation of either the 2c- or 3c-hydroxyl groups by a metal ion.

A postulate for montmorillonite catalysis If it is assumed that similar pathways are followed in oligomer formation for metal ion, template-directed and montmorillonite-catalyzed synthesis of RNA oligomers it may be possible propose a possible mechanism for the montmorillonite-catalyzed synthesis of RNA oligomers. The observation that a 2-methylimidazole activating group is all that is needed to generate long oligomers from activated nucleotides suggests that the optimal orientation of the activated monomers for reaction is the key factor in the montmorillonite-catalyzed reaction. The failure of certain montmorillonites to be catalysts may be due to differences in the geometry of their interlayer from those of the catalytic clays. The proper orientation of the activated bound nucleotides has also been proposed to be an important factor in UO22− and Pb2+ catalysis as well. While activation of the 2c-hydroxyl group may be a factor with some of the metal ions it appears not to be a factor in the montmorillonite catalysis. The selectivity observed for reaction at either the 2c- or 3c-hydroxyl groups may be due to factors speciic for each catalyst. The difference in selectivity for montmorillonite may be due to a difference in the orientation of the activated nucleotide when bound to the clay interlayer. This difference in orientation may be responsible for the selectivity for the formation of the 3c, 5c-link with purine nucleotides and 2c, 5c- links with pyrimidine nucleotides.

POTENTIAL STEPS TO THE ORIGIN OF LIFE FROM OLIGOMERS Research on the formation of RNA oligomers is based on the assumption that the requisite activated RNA monomers formed spontaneously on the primitive Earth. Progress has been made in their formation but no plausible prebiotic synthesis of the activated RNA monomers or the monomers that would the basis of any other genetic polymer has been reported at the time of this writing (see discussion in “Alternative Genetic Systems”). In this Section we will assume that a catalyst was found that generated a mixture of informational biopolymers on the primitive Earth that were suficiently long to store genetic information and to catalyze reactions.

206

Ferris

The next question is how far away are these biopolymers from the formation of the irst life? Since these polymers were formed catalytically it is reasonable to assume that there was selectivity in the synthesis of these oligomers so a limited number of isomers were produced (see discussion in “Prebiotic Polymerization of RNA Monomers”). Because it is a catalytic process a continuous supply of these structurally related oligomers were formed continuously. If the irst life was based on a self-replicating system capable of mutations then the key step was the origin of the replication of these oligomers. In one scenario it is assumed that the irst life is an array of oligomers bound on the mineral that catalyzed their formation. This would require that a subgroup of these oligomers, or the mineral-oligomer-complex, had the ability to catalyze their own synthesis while bound to the mineral. This group of catalytic oligomers would rapidly take over the mineral surface if their rates of replication were greater than the catalytic formation of oligomers by the mineral. In this scenario the subgroup of oligomers would serve as the catalysts and the templates for the synthesis. It is assumed that the rate of catalysis by the biopolymer will increase as a consequence of the selection of the more rapidly forming oligomers. Another important catalyst was one that catalyzed the ligation of the oligomers formed by mineral catalysis. This generates longer biopolymers with greater capability of storing more genetic information. Greater information storage will be essential for the evolution of the more complex biochemical machinery needed for the formation of more sophisticated forms of life. At some point in the above process the biomolecules will have to become independent from the mineral catalyst to which they are bound. This is because the higher molecular weight oligomers will bind more strongly to the minerals surface and block the catalytic sites. This dissociation could be a stepwise process where the oligomer-catalyst complex is encapsulated within a vesicle and then the oligomers are released from the catalyst. The direct incorporation of the catalyst-oligomer complex into a vesicle is an alternative to life originating on the surface of a mineral catalyst. A possible scenario for such an event has been described (Hanczyc et al. 2003; also see discussion in “Vesicles”). Here montmorillonite catalyzes both the formation of RNA oligomers and the vesicle that encapsulated the oligomercatalyst complex. The one law with this particular scenario is that the conditions necessary for the formation of the biopolymer destroys the vesicle (Monnard et al. 2002). It seems likely that there may be an alternative approach that will be successful in the incorporation of the mineral-catalyst inside a vesicle without destroying it. One can imagine the ingestion of molecules in the vesicle that elute the oligomers from the catalyst. Of course these molecules should do this without inhibiting the catalyst. The stages leading to a self-replicating system after the initial trapping of the biopolymer inside a vesicle will be similar to the ones described above for life originating in a family of biopolymers bound to a catalytic mineral. The main difference will be the need for the ission of the vesicle into two or more vesicles once it has reached a critical size (Walde et al. 1994).

PROPOSED EXPERIMENTS Selection of oligomers that bind to other biomolecules The oligomers formed by montmorillonite catalysis are long enough to fold into threedimensional structure that can bind to other RNA oligomers or other biomolecules (Joyce and Orgel 1999). Studies should be performed to determine if this binding does indeed occur. The selectivity observed in the oligomers formed by montmorillonite catalysis suggests that it is likely, if binding is observed, that an appreciable fraction of these oligomers will bind. If

Catalysis & Prebiotic Synthesis

207

this occurs it possible that a small fraction of these oligomers will catalyze reaction(s) of the biomolecules to which they are bound.

Catalysis of template-directed synthesis One of the essential catalytic processes required for the irst life if the replication of RNA by template-directed synthesis. This is essential for the reservation of genetic material as well as the preservation of catalytic RNA. As noted in “Non-enzymatic Template – Directed Synthesis,” the non-enzymatic template-directed synthesis is only successful in the formation of G oligomers on a C-template or on a template that contains more cytidine nucleotides than guanosine nucleotides. Catalysis is needed to enhance the rates of formation of RNAs that contain A, U and C nucleotides. The potential catalysts include RNA oligomers and minerals. There are examples of the template-directed synthesis of G-oligomers on a polyC template bound to minerals (Schwartz and Orgel 1985; Holm et al. 1993). So far no minerals have been detected that enhance the rates of template-directed synthesis of activated nucleotides over that from RNA templates alone.

Catalysis of RNA ligation Some of the oligomers formed by montmorillonite catalysis are long enough to be catalysts but the bulk of the oligomers are not. A ligation catalyst that would link these smaller oligomers would enhance the pool of longer oligomers. An RNA catalyst of this type would generate longer oligomers that have similar structures because it would have a template that would simultaneously bind two oligomers close enough so that they could form a phosphodiester bond between them. While it is likely that the ligation catalyst would be another RNA molecule it is also possible that a mineral would catalyze phosphodiester bond formation between two oligomers.

ACKNOWLEDGMENTS Research support was provided by NSF grant CHE-0413739 and NASA grant NAGS12750 that supports the NY Center for Studies on the Origins of Life.

REFERENCES Bailey J, Chrysostomou A, Hough JH, Gledhill TM, McCall A, Clark S, Menard F, Tamura M (1998) Circular polarization in star-formation regions: Implications for biomolecular homochirality. Science 281:672-674 Banin A (1973) Quantitative ion exchange process for clays. U.S. Patent Nd 3,725,528 Banin A, Lawless JG, Mazzurco J, Church FM, Margulies L, Orenberg JB (1985) pH proile of the adsorption of nucleotides onto montmorillonite. Origins Life Evol Biosphere 15:89-101 Birdson PK, Orgel LE (1980) Catalysis of accurate poly(C)-directed synthesis of 3c-5c-linked oligoadenylates by Zn2+. J Mol Biol 144:567-577 Böhler C, Nielsen PE, Orgel LE (1995) Template switching between PNA and RNA oligonucleotides. Nature 376:578-581 Brack A (1998) The Molecular Origins of Life: Assembling the Pieces of the Puzzle. Cambridge, Cambridge University Press Brandes JA, Boctor NZ, Cody GD, Cooper BA (1998) Abiotic nitrogen reduction on the early Earth. Nature 395:365-367 Brasier MD, Green OR, Jephcoat AP, Kleppe AK, Van Kranendonk MJ, Lindsay JF, Steele A, Grassineau NV (2002) Questioning the evidence for the Earth’s oldest fossils. Nature 416:76-81 Cairns-Smith AG (1982) Genetic Takeover and the Mineral Origins of Life. Cambridge (Great Britain), Cambridge University Press Cody GD, Boctor NZ, Filley TR, Hazen RM, Scott JH, Sharma A, Yoder Jr. HS (2000) Primordial carbonylated iron-sulfur compounds and the synthesis of pyruvate. Science 289:1337-1340 Deamer DW, Pashley RM (1989) Amphiphilic components of the Murchison carbonaceous chondrite: Surface properties and membrane formation. Origins Life Evol Biosphere 19:21-38

208

Ferris

Delano J (2001) Redox History of the Earth’s Interior since ~3900 Ma: implications for prebiotic molecules. Origins Life Evol Biosphere 31:311-341. Ertem G, Ferris JP (2000) Sequence- and regio-selectivity in the montmorillonite-catalyzed synthesis of RNA. Origins Life Evol Biosphere 30:411-422 Ertem G, Ferris JP (1998) Formation of RNA oligomers on montmorillonite: site of catalysis. Origins Life Evol Biosphere 28:485-499 Eschenmoser A (1999) Chemical etiology of nucleic acid structure. Science 284:2118-2124 Ferris JP (2005) Origins of life, molecular basis of. In: Encyclopedia of Molecular Cell Biology and Molecular Medicine. Meyers RA (ed), Weinheim, Germany, Wiley-VCH Verlag, p 1-27 Ferris JP (2002) Montmorillonite catalysis of 30-50 mer oligonucleotides: laboratory demonstration of potential steps in the origin of the RNA world. Origins Life Evol Biosphere 32:311-332. Ferris JP, Ertem G (1992) Oligomerization reactions of ribonucleotides on montmorillonite: reaction of the 5’phosphorimidazolide of adenosine. Science 257:1387-1389 Ferris JP, Hagan Jr. WJ (1984) HCN and chemical evolution: The possible role of cyano compounds in prebiotic synthesis. Tetrahedron 40:1093-1120 Ferris JP, Hill Jr. AR, Liu R, Orgel LE (1996) Synthesis of long prebiotic oligomers on mineral surfaces. Nature 381:59-61 Ferris JP, Joshi PC, Edelson EH, Lawless JG (1978) HCN: A plausible source of purines, pyrimidines, and amino acids on the primitive Earth. J Mol Evol 11:293-311 Ferris JP, Kamaluddin (1989) Oligomerization reactions of deoxyribonucleotides on montmorillonite clay: The effect of mononucleotide structure on phosphodiester bond formation. Origins Life Evol Biosphere 19: 609-619 Ferris JP, Kamaluddin, Ertem G (1990) Oligomerization reactions of deoxyribonucleotides on montmorillonite clay: the effect of mononucleotide structure, phosphate activation and montmorillonite composition on phosphodiester bond formation. Origins Life Evol Biosphere 20:279-297 Flores JJ, Bonner WA, Massey GA (1977) Asymmetric photolysis. J Amer Chem Soc 99:3622-3625 Fry I (2000) The Emergence of Life on Earth. New Brunswick, N. J., Rutgers University Press Fuller WD, Sanchez RA, Orgel LE (1972) Studies in prebiotic synthesis IV. Synthesis of purine nucleosides. J Mol Biol 67:25-33 Hanczyc MM, Fujikawa SM, Szostak JW (2003) Experimental models of primitive cellular compartments: Encapsulation, growth and division. Science 302:618 - 622 Heinen W, Lauwers AM (1996) Organic sulfur compounds resulting from interaction of iron sulide, hydrogen sulide and carbon dioxide in an aqueous environment. Origins Life Evol Biosphere 26:1-20 Hill Jr. AR, Böhler C, Orgel LE (1998) Polymerization on the rocks: negatively-charged D-amino acids. Origins Life Evol Biosphere 28:235-243 Hill Jr. AR, Orgel LE (1999) Oligomerization of L-J-carboxyglutamic acid. Origins Life Evol Biosphere 29: 115-122 Holm NG, Ertem G, Ferris JP (1993) The binding and reactions of nucleotides and polynucleotides on Iron oxide polymorphs. Origins Life Evol Biosphere 23:195-215 Huang W, Ferris JP (2003) Synthesis of 35-40 mers of RNA oligomers from unblocked monomers. A simple approach to the RNA world. Chem Commun:1458-1459 Huber C, Wachtershauser G (1998) Peptides by activation of amino acids with CO on (Ni,Fe)S surfaces: Implications for the origin of life. Science 281:670-672 Huber C, Wachtershauser G (1997) Activated acetic acid by carbon ixation on (Fe,Ni)S under primordial conditions. Science 2776:245-247 Inoue T, Orgel LE (1981) Substituent control of the poly(C)-directed oligomerization of guanosine 5cphosphoroimidazolide. J Am Chem Soc 103:7666-7667 Joyce GF, Orgel LE (1999) Prospects for understanding the origin of the RNA world. In: The RNA World: The Nature of Modern RNA Suggests a Prebiotic RNA. Gesteland RF, Cech TR, Atkins JF (eds), Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, p. 49-77 Kanavarioti A, Monnard P-A, Deamer DW (2001) Eutectic phases in ice facilitate nonenzymatic nucleic acid synthesis. Astrobiology 1:271-281 Kawamura K, Ferris JP (1999) Clay catalysis of oligonucleotide formation: kinetics of the reaction of the 5cphosphorimidazolides of nucleotides with the non-basic heterocycles uracil and hypoxanthine. Origins Life Evol Biosphere 29:563-591 Liu R, Orgel LE (1998) Polymerization on the rocks: E-amino acids and arginine. Origins Life Evol Biosphere 28:245-257 Lohrmann R, Bridson PK, Orgel LE (1980) Eficient metal-ion catalyzed template-directed oligonucleotide synthesis. Science 208:1464-1465 McCollum TM, Ritter G, Simoneit BRT (1999) Lipid synthesis under hydrothermal conditions. Origins Life Evol Biosphere 29:153-166

Catalysis & Prebiotic Synthesis

209

Miller SL (1997) Peptide nucleic acids and prebiotic chemistry. Nature Structural Biology 4:167-169 Miyakawa S, Cleaves HJ, Miller SL (2002) The cold origins of life: B. Implications based on pyrimidines and purines produced from frozen ammonium cyanide solutions. Origins Life Evol Biosphere 32:209-218 Miyakawa S, Ferris JP (2003) Sequence- and regioselectivity in the montmorillonite-catalyzed synthesis of RNA. J Am Chem Soc 125:8202-8208 Mojizsis SJ, Arrhenius G, McKeegan KD, Harrison TM, Nutman AP, Friend CRL (1996) Evidence for life on Earth before 3,800 million years ago. Nature 384:55-59 Monnard P-A, Apel CL, Kanavarioti A, Deamer DW (2002) Inluence of ionic inorganic solutes on selfassembly and polymerization processes related to early forms of life: implications for a prebiotic aqueous medium. Astrobiology 2:139-152 Monnard P-A, Kanavarioti A, Deamer DW (2003) Eutectic phase polymerization of activated ribonucleotide mixtures yields quasi-equimolar incorporation of purine and pyrimidine nucleobases. J Am Chem Soc 125:13734-13740 Moorbath S (2005) Dating earliest life. Nature 434:155-155 Müller D, Pitsch S, Kittaka A, Wagner E, Wintner CE, Eschenmoser A (1990) Chemie von DAminonitrilen Aldomerisierung von Glycolaldehyd-phosphat zu racemischen Hexose-2,4,6-triphosphaten und (in Gegenwart von Formaldehyd) racemischen Pentose-2,4-diphosphaten: rac-Allose-2,4,6-triphosphat und rac-Ribose-2,4-diphosphat sind die Reaktionshauptprodukte. Helvetica Chimica Acta 73:1410-1468. Orgel LE (2004) Prebiotic chemistry and the origin of the RNA world. Critical Revs Biochem Mol Biol 39: 99-123 Orgel LE (1998) Polymerization on the rocks: theoretical introduction. Origins Life Evol Biosphere 28:227234 Osterberg R, Orgel LE (1972) Polyphosphate and trimetaphosphate formation under potentially prebiotic conditions. J Mol Evol 1:241-248 Osterberg R, Orgel LE, Lohrmann R (1973) Further studies of urea-catalyzed phosphorylation reactions. J Mol Evol 2:231-234 Pizzarello S, Cronin JR (2000) Non-racemic amino acids in the Murray and Murchison meteorites. Geochim Cosmochim Acta 64:329-338 Pizzarello S, Huang Y, Becker L, Poreda RJ, Nieman RE, Cooper G, Williams M (2001) The organic content of the Tagish Lake meteorite. Science 293:2236-2240 Prabahar KJ, Ferris JP (1997) Adenine derivatives as phosphate-activating groups for the regioselective formation of 3c,5c-linked oligoadenylates on montmorillonite: possible phosphate-activating groups for the prebiotic synthesis of RNA. J Am Chem Soc 119:4330-4337 Ricardo A, Carrigan MA, Olcott AN, Benner S (2004) Borate minerals stabilize ribose. Science 303:196 Rushdi AI, Simoneit BRT (2001) Lipid formation by aqueous Fischer-Tropsch-type synthesis over a temperature range of 100-400 °C. Origins Life Evol Biosphere 31:103-118 Sawai H (1988) Oligonucleotide formation catalyzed by divalent metal ions. The uniqueness of the ribosyl system. J Mol Evol 27:181-186 Sawai H, Higa K, Kuroda K (1992) Synthesis of cyclic and acyclic oligocytidylates by uranyl ion catalyst in aqueous solution. J Chem Soc Perkin I:505-508 Sawai H, Itoh T, Kokaji K, Shinozuka K (1997) An approach to prebiotic synthesis of D-oligoribonucleotides and description of their properties: selective advantage of E-RNA over D-RNA. J Mol Evol 45:209-215 Sawai H, Kuroda K, Hojo T (1989) Uranyl ion as a highly effective catalyst for internucleotide bond formation. Bull Chem Soc Jpn 62:2018-2023 Sawai H, Orgel LE (1975) Oligonucleotide synthesis catalyzed by the Zn2+ ion. J Amer Chem Soc 97:35323533 Sawai H, Yamamto K (1996) Lanthanide ion as a catalyst for internucleotide bond formation. Bull Chem Soc Jpn 69:1701-1704 Schmidt JG, Nielson PE, Orgel LE (1997) Information transfer from peptide nucleic acids to RNA by templatedirected synthesis. Nucleic Acids Res 25:4797-4802 Schoning K-U, Scholz P, Guntha S, Wu X, Krishnamurthy R, Eschenmoser A (2000) Chemical etiology of nucleic acid structure: The D-threofuranosyl-(3 -2c) oligonucleotide system. Science 290:1347-1351 Schopf WJ (1993) Microfossils of the early Archean Apex chert: new evidence of the antiquity of life. Science 260:640-646 Schwartz AW, Orgel LE (1985) Template-directed polynucleotide synthesis on mineral surfaces. Letter to the Editor. J Mol Evol 21:299-300 Shimazu M, Shinozuka K, Sawai H (1993) UO22+-catalyzed oligothioadenylate synthesis with high regio- and stereoselectivity. Angew Chem Int Ed Engl 32:870-872 Sleeper HL, Lohrmann R, Orgel LE (1979) Template-directed synthesis of oligoadenylates catalyzed by Pb2+ ions. J Mol Evol 13:203-214

210

Ferris

Springsteen G, Joyce GF (2004) Selective derivatization and sequestration of ribose from a prebiotic mix. J Am Chem Soc 126:9578-9583 Tian F, Toon OB, Pavlov AA, De Sterck H (2005) A hydrogen-rich early Earth atmosphere. Science 308: 1014-1017 van Rode JHG, Orgel LE (1980) Template-directed synthesis of oligoguanylates in the presence of metal ions. J Mol Biol 144:579-585 Vokrouhlicky D, Farinella P (2000) Eficient delivery of meteorites to the Earth from a wide range of asteroid parent bodies. Nature 407:606-608 Walde P, Wick R, Fresto M, Mangone A, Luisi PL (1994) Autopoietic self-replication of fatty acid vesicles. J Am Chem Soc 116:11649-11654 Wang K-J, Ferris JP (2001) Effect of inhibitors on the montmorillonite clay-catalyzed formation of RNA: Studies on the reaction pathway. Origins Life Evol Biosphere 31:381-402 Ward PD, Brownlee D (2000) Rare Earth. Copernicus, New York Watson EB, Harrison TM (2005) New thermometer reveals minimum melting conditions on earliest Earth. Science 308 Weimann BJ, Lohrmann R, Orgel LE, Schneider-Bernloehr H, Sulston JE (1968) Template-directed synthesis with adenosine-5c-phosphorimidazolide. Science 161:387-388 Zubay G (2000) Origins of Life on the Earth and in the Cosmos. Academic Press, San Diego, CA Zubay G (1998) Studies on the lead-catalyzed synthesis of aldopentoses. Origins Life Evol Biosphere 28: 13-26

Reviews in Mineralogy & Geochemistry Vol. 59, pp. 211-231, 2005 Copyright © Mineralogical Society of America

9

The Evolution of Biological Carbon and Nitrogen Cycling—a Genomic Perspective Jason Raymond Microbial Systems Division Biosciences Directorate Lawrence Livermore National Laboratory Livermore, California, 94550, U.S.A. [email protected]

INTRODUCTION Carbon and nitrogen are essential to all living organisms, owing to their abundance and remarkable characteristics when participating in chemical bonds. Their essentiality dates back to the very origin of life, where current theories hypothesize either a prebiotic abundance of organic compounds rich in carbon and nitrogen, or an ability to assimilate them inorganically through abiotic reactions that might have been catalyzed on ancient mineral surfaces. This chapter details the core reactions essential to the assimilation of these elements in biologically useful forms—the so-called ixation of carbon and nitrogen—focusing on recent literature and insights from comparative genomics and phylogenetics. Though considerable debate continues on the antiquity of these pathways, especially whether or not they might have been present in the last common ancestor (LCA) of modern organisms, it is clear that carbon and nitrogen ixation pathways were of crucial importance to the primitive ancestors of extant life. Furthermore, the biological assimilation of inorganic carbon (autotrophy) and atmospheric nitrogen (diazotrophy) represent pivotal juxtapositions of biological and geological cycles. It is thought that atmospheric CO2 concentrations have decreased substantially since the proposed origin of life some 3.8 billion years ago, due in large part to either primary (ixation) or secondary (e.g., weathering) inluence by biota (Hayes 1994; Rye et al. 1995; Des Marais 1997; Lowe and Tice 2004). Though the biosphere accounts for a relatively small fraction of the total carbon on Earth, the rate of carbon lux through the biosphere far exceeds that through any geological reservoirs (Des Marais 1997). Biological carbon ixation is closely balanced to carbon recycling through biological oxidation, and the future stability of this and other CO2 reservoirs (and our ability to inluence or understand them) depends critically on these biological underpinnings (e.g., Falkowski et al. 2000). Conversely, nitrogen, especially the atmospheric N2 reservoir, is remarkably stable, owing largely to the stability of the N-N triple bond and the relative inertness of the molecule. In fact, many environments are considered nitrogen limited, meaning that biologically available nitrogen is essentially locked up in biomass. Thus many ecosystems are dependent upon diazotrophs, prokaryotes that can convert atmospheric N2 into ammonia by way of the enzyme nitrogenase. As is detailed below, this enzyme is thought to be ancient and is extremely sensitive to oxygen; nitrogen-ixing bacteria and archaea are either anaerobic or have evolved elaborate mechanisms for shielding nitrogenase from molecular oxygen. It is estimated that as nitrogenase ixes of the same order per year as all anthropogenic and abiotic processes combined, including the industrial Haber-Bosch process and the lightning-catalyzed production of nitrate that (prior to Haber’s revolutionary invention) formed the basis for proitable mining industries, especially in arid areas such as Chile’s Atacama. 1529-6466/05/0059-0009$05.00

DOI: 10.2138/rmg.2005.59.9

212

Raymond USING GENOMICS TO UNDERSTAND THE PRESENT (AND INFER THE PAST)

At present, 254 prokaryotic and 20 eukaryotic genomes have been completed, with over 400 and 200 more, respectively, progressing through various pipelines around the world. The term comparative genomics—once an applied discipline unto itself—is arguably redundant, as the wealth of sequence data now available obliges genomics to be inherently comparative. More speciically, given a single gene or chromosome sequence, ab initio prediction of a protein’s structure, function, and interactions are still far beyond current capabilities, and simply predicting the position and arrangement of coding sites given a “universal” genetic code is imprecise even in the simplest prokaryotic genomes. The bootstraps upon which our vast repositories of genetic information are propped stem ultimately from the comparison of new sequences with previously characterized evolutionary “relatives”—homologous proteins that have been painstakingly expressed and characterized. Thus our understanding hinges on our knowledge of molecular evolution, how processes such as natural selection and horizontal gene transfer alter gene sequences and the function of an organism’s proteins. Widely used function prediction front-ends, such as NCBI’s BLAST sequence vs. database comparison tool, are essentially distilled algorithms for determining a protein’s phylogeny, in particular for inding homologs with known functions. Fast database search tools are typically suficient for assessing homology between sequences and annotating a newly sequenced genome, but a wealth of additional information can be obtained by looking not just at whether a group of sequences is related, but how they have changed through time. The textbook example is how natural selection has promulgated variability in the immunoglobulin antigen-recognition domain, thereby increasing the robustness of vertebrate immune systems (Tanaka and Nei 1989). Phylogenetic analysis of enzyme families, such as those presented in this chapter for proteins involved directly in carbon and nitrogen cycling, give insight into how metabolic capabilities and pathways have evolved. For example (as is discussed below), since diverging from its homologs in chlorophyll biosynthesis pathways, it is possible to elucidate how the enzyme nitrogenase has become progressively more speciic for atmospheric dinitrogen as a substrate, even though the mechanism, structure, and cofactor complement of the enzyme has been largely retained in these functionally diverse pathways. Likewise, one can envisage how the reductive tricarboxylic acid cycle, used by diverse and deeply-branching organisms for “ixing” CO2 into biomass, likely evolved from a much simpler pathway by duplication of a few ancient genes followed by improved substrate speciicity. Taken as a whole, the function of the expressed protein complement encoded within an organism’s genome comprises its metabolome and, in essence, comprises its phenotype: how and under what conditions an organism can thrive. Connecting the dots between the genome-encoded genotype and the functional phenotype represents the next major frontier in biology, with two major hurdles that so-called functional genomics must overcome. The irst is that, at a given time and under varying conditions, not all of the protein complement of a genome is translated into protein. High-throughput determination of this “business end” is presently being attacked both at the transcriptional level, using microarrays and gene chips to quantify mRNA levels, and also at the proteome level, in particular using high resolution mass spectrometers to identify fragments of proteins being expressed by an organism under varying conditions. Theoretical advances have also been made using metabolic network modeling methods such as lux balance analysis, which can correctly segregate important versus redundant metabolic pathways by optimizing lux through a metabolic network under a given set of (environmentally-imposed) boundary conditions. The second major hurdle in connecting genotype to phenotype is the remarkable paradox presented by so-called hypothetical proteins of unknown function, paradoxical because their

Evolution of Biologic C & N CyclingȰGenomic Perspective

213

presence and relative abundance has continued almost unabated since the irst genomes became available a decade ago. Speciically, this means that every new genome sequenced will have on average of 1/3 of its putative proteins with no clear homologs in any protein database or in any other genome (Bork 2000)—a veritable Red Queen principle that appears only to be solvable by bench-top biochemical characterization. However, it is not dificult to envision how high-throughput proteomics and microarray analysis will give insight into the function of hypothetical proteins, giving insight into whether and when a particular protein is expressed (Ram et al. 2005). As we learn how to decode and interpret their genomes and proteomes, microorganisms present an unrivalled diagnostic tool for understanding the physical chemistry and dynamics of natural environments, highly-tuned to their surroundings and possessing a remarkable capacity to adjust their metabolic repertoire and/or community organization in response to changing conditions. This raises the attractive possibility that an understanding of microbial evolution can provide a glimpse into the past. While this appears to hold true for many cases, for example the near-concurrence in the geological and biological records of the evolution of oxygenic photosynthesis and mitochondrial-based respiration, the plasticity with which organisms, pathways, and individual genes have evolved and, in the latter case, been horizontally transferred stand as important caveats. In particular, relying on so-called “universal” phylogenies to infer characteristics of ancient organisms is fraught by the assumption that modern lineages have somehow remained metabolically “frozen,” a dificult tenet to hold in light of the remarkable geological changes the Earth has undergone and given the proclivity and non-linearity through which evolution sometimes acts.

BIOLOGICAL NITROGEN CYCLING AND DIAZOTROPHY In modern organisms, assimilation of inorganic nitrogen is universally shunted through ammonia (as the ammonium cation), representing the central inorganic nitrogen source and further suggestive of a pivotal role in the early evolution of life. Though nitrate and nitrite are important inorganic nitrogen sources for extant organisms, these compounds were of leeting abundance on the early Earth and probably played a minor role until their concentration increased following the oxidation of Earth’s atmosphere (see discussion below). Numerous transmembrane permeases transport ammonium into cells, whereafter it is enzymatically incorporated into carbon skeletons of central metabolites. This juxtaposition of the carbon and nitrogen cycles is carried out by a highly conserved set of enzymes comprising the GS/GOGAT cycle, illustrated in Figure 1. The enzyme glutamine synthetase (GS) adds ammonium to glutamate to form glutamine, which then is used as a substrate in the amination of 2-oxoglutarate by glutamine:2-oxoglutarate aminotransferase (GOGAT or glutamate synthase). This results in the cyclic formation of two molecules of glutamate—regenerating the original glutamate substrate and freeing a second glutamate to be used in other metabolic reactions. The products of this central cycle, glutamate and glutamine, serve as the ultimate nitrogen donors for all additional nitrogen-containing metabolites through a cascade of aminotransferase reactions. Regulatory response to combined nitrogen availability is tightly queued to the intracellular ratio of glutamine to 2-oxoglutarate, further underscoring the centrality of this intersection between carbon and nitrogen assimilatory pathways. This overlap between pathways has also garnered substantial interest in the evolutionary history of the key enzymes, glutamine synthetase and glutamate synthase. Glutamine synthetase is a multisubunit enzyme (homododecameric in bacteria, homooctamer in many eukaryotes, including Homo sapiens) that catalyzes the ATP-dependent condensation of ammonium with glutamate. The wide distribution of this enzyme across all three domains of life argues that the enzyme is ancient, notably comprising two major classes

214

Raymond

Figure 1. GS/GOGAT cycle, as discussed in the text. Boxes over arrows indicate enzymes responsible for the biochemical reactions shown. Pathways are abbreviated for simpliication.

(GSI and GSII) of enzymes whose duplication origin has been argued to predate the LCA. Early attempts to date major divergences on the tree of life used this early duplication as time zero, along with an argued “clock-like” evolution among GS proteins that remains controversial (Pesole et al. 1991; Brown et al. 1994). So far, the GSI family is exclusive to bacteria, whereas eukaryotes possess only GSII. Some bacteria, most notably the rhizobia, express both forms of GS (Turner and Young 2000). Archaea also contain GSI, with euryarchaeotes sharing a highly similar GSI (GSI-D) with Gram-positive bacteria and Thermotoga, with horizontal gene transfer as a very likely explanation for this distribution (see below). Sulfolobus and other crenarchaeotes, however, possess GSI’s that are phylogenetically distinct from the GSI-D or GSI-E (bacterial) clades, with functional and phylogenetic features intermediate between the bacterial and euryarchaeal enzymes (Cabello et al. 2004). Within both classes of GS are examples of HGT, potentially confounding their evolutionary history. Some of the strongest early evidence for HGT came from analysis of the GSI family, where interdomain transfer between Gram-positive bacteria and euryarchaeotes is clear) which, intriguingly, is similar to the pattern of interdomain (bacteria l archaea) transfer evident in analysis of nitrogenase genes (discussed below) (Pesole et al. 1995; Nesbo et al. 2001). It is not yet clear whether these similar patterns of gene transfer are in fact linked—for example, as an adaptation speciic to biological nitrogen cycling—or the result of a more extensive, nonspeciic exchange of genetic information (as is suggested by recent whole genome analysis of methanogens; e.g., Deppenmeier et al. 2002). HGT also obfuscates the phylogeny of the GSII family, noted originally among the rhizobia (Turner and Young 2000). The evolution of GOGAT/glutamate synthase, shown in Figure 2, is even more perplexing. The enzyme itself comes in several different lavors that vary by their obligatory (two) electron donor, which can be ferredoxin, NADH, or NADPH. Most bacteria possess NADH- or NADPH-dependent GOGAT’s, with some speciicity within taxa such as NADH dependence in many proteobacteria versus NADPH dependence in Gram-positive Bacillus subtilis (Suzuki and Knaff 2005). The GOGAT from cyanobacteria is ferredoxin-dependent, as is that found in the plastids of photosynthetic eukaryotes. Alternatively, non-photosynthetic eukaryotes

daii

0.2

du sd

no

ge n

2 M.maripaludishii sc n na M.ja eri ark i M.b aze .m M

is1 lud i pa ni ari o m t M. ur .b M

NADPHdependent (truncated)

F.acidarmanus ilum P.aeroph

S.solfataricus

S.toko

lgi

the

P. sm fur 43 ios us es 04 T.m 19 arit 5 ima msb 8 M.ther moacet ica

D.e

ut o

M .th

er m oa

vo r an s

M. ac eti m S.

ilo el

tro ph

ti1

ic um 2

ge lla tus

nd ler i M.k a

M.th e r m o auto troph utotroph icum icum1 3

A. fu

G.sulfurreducens1 e n s1 educ tallir G.me gh orou m enb cans lu i hild r is el r u lga ulf oc D.vu es m r d . e D th C.

NADPH (NADH)dependent

olicu

Ferredoxindependent

T.elongatus C.w Sy atsonii S.e necho cy s l on N. tis ga 6 80 tus A pun 31 .v ar ctifo ia r m bi e lis

G.violaceus

A.ae

s 4 MED rinus a 2 .m P s WH810 cu oc oc Synech

e ns ie fn tilis a h b s D. .su ran B lodu is a e ns h y . e B O.ih A 2 52 i MRS n ju reus C.je ureus au S.a

cs

M. fla

m

Figure 2. GOGAT (glutamate synthase) protein evolutionary tree, based on homologs found in completed prokaryotic genomes. The different clades bear strong correlation with the electron donor, as shown. There are exceptions, for example the cyanobacterium Synechocystis PCC6803 has two GOGAT homologs, one of which is ferredoxin-dependent and the other NADH-dependent, though it clusters with NADPH-dependent enzymes. Truncated sequences are typically less than half the size of their ferredoxin- and NADPH-dependent homologs and are characteristic of archaeal genomes, though are also found among diverse bacteria. Numbers following binomen indicate duplicate gene copies within a single genome. Alignments for all phylogenies were constructed with TCoffee and trees subsequently inferred and bootstrapped (500 replicates) using MEGA v3, using the Neighbor-Joining model and a pairwise deletion of gap-aligned sites. ic u

09 A0 CG ris ium st lu zob p a hi R. sor oti2 Me.melil S us ulat aps ns M.c iodura D.rad a D.aromatic B.fungorum R.me tall id u ra ns C.ace to b u tylicu C.hu m tchin s on Pir ii ellu l a C C.a sp S .te ggre y ne pidu gatu m ch m oc y B. j

ap on

G. su lfu r r ed G.m etall uc ir e duc ens ens 2 2 D.ace to xi da ns2 T.thermophil us HB27

2 68 03

st is

sto No

M.therm oa

Evolution of Biologic C & N CyclingȰGenomic Perspective 215

s m de oi cu er acti a t h o t e sp R. agn 1 m m ns M. ubru ida r tox H7 R. e c 57 D.a O1 li co eri 2a . Eexn S.fl S.typhi

p

C.violaceum M.degradans A.vine landii P.a er u gi n osa

216

Raymond

and the non-plastid (e.g., nuclear) component of plants and algae contains NADH-dependent GOGAT’s, suggesting a linear recruitment of cyanobacterial GOGAT by algae following plastid endosymbiosis. Archaea contain NADPH-dependent GOGAT enzymes, though highly truncated in structure and evidently not possessing the two-subunit heterodimeric structure endemic to bacterial GOGAT’s (Vanoni and Curti 1999). Temple et al. (1998) have suggested that this abridged gene may represent an ancestral, minimal form of the enzyme. Subsequently, Dincturk and Knaff (2000; Dincturk 2001) and Nesbo et al. (2001) noted patterns of homology suggesting one or multiple interdomain transfers between archaea and bacteria. The centrality of these enzymes in ammonia assimilation strongly suggests that the GS/GOGAT cycle—or an analogous pathway—was operational in the LCA, which leads to the conclusion that observed HGT’s were in fact orthologous replacements, whereby the acquisition of a more eficient pathway assumes the role of a preexisting one, which is eventually lost due to purifying selection. Biologically-utilizable nitrogen compounds are often in limited supply in natural systems, and a complex array of mechanisms for assimilating and recycling nitrogenous compounds are known. For instance, while ammonia represents the nucleus of the biological nitrogen cycle, most organisms have mechanisms for transporting nitrogenous compounds (particularly nitrogen-rich amino acids such as lysine and glutamine) across the membrane for assimilation, effectively bypassing potential dependence upon ammonia. Found exclusively among prokaryotes (more speciically, bacteria and methanogenic archaea) are a diverse group of so-called diazotrophs that are able to assimilate nitrogen from atmospheric N2 using a heterotetrameric enzyme known as nitrogenase. Using a high potential electron donor such as ferredoxin, the nitrogenase complex successively reduces atmospheric dinitrogen into two molecules of ammonia, with an optimal net reaction of: N2 + 8H+ + 8e− + 16 ATP o 2NH3 + H2 + 16 ADP + 16Pi

The metabolic “cost” of dinitrogen reduction—16 ATP’s hydrolyzed—is unparalleled among biochemical reactions and may be two or more times as high under natural conditions. Active nitrogen ixation is thought to account for as much as 40% of the total ATP synthesized in diazotrophs (Daesch and Mortenson 1968). Therefore it is not surprising that nitrogenase is expressed in diazotrophs as a last-resort enzyme, when other more readily assimilated nitrogenous compounds are exhausted. It is worth noting that the net reduction of dinitrogen as shown is an exergonic reaction (dG° = −15.2 kJ/mol) and at least in theory could be used to generate energy (the high cost of running nitrogenase is thought to be involved with breaking the highly stable N–N triple bond) (Howard and Rees 1996). The evolutionary history of nitrogenase has been of considerable interest since sequences for its component proteins irst became available. Though prebiotic sources of nitrogen might have been prevalent on the early earth as a result of planetary accretion and impact events, in particular NH3 would have been subject to ultraviolet photolysis and thereby progressively depleted (Kuhn and Atreya 1979; Raven and Yin 1998). The abiotic synthesis of nitrates and nitrites, an important source of ixed nitrogen generated by lightning strikes in the modern oxidized atmosphere, is thought to have been limited under an early neutral-to-mildly-reducing atmosphere, especially as Archean CO2 levels decreased (Navarro-Gonzalez et al. 2001). Navarro-Gonzales et al. argue that these conditions culminated in a nitrogen crisis during the Neoproterozoic or Archean, and that such conditions would have favored the development of the nitrogenase enzyme complex. It is feasible that alternative sources of nitrogen might have been abiotically synthesized under conditions hypothesized for the early Earth, such as cyanide if methane were present in quantities assumed by some models (Pavlov et al. 2000), and potentially offset or delayed the onset of this ixed nitrogen crisis. While such hypotheses are still in need of validation, some interesting correlations can be drawn from examination of the biological record. The evolutionary history of nitrogenase

Evolution of Biologic C & N CyclingȰGenomic Perspective

217

suggests that it likely evolved from a complex not directly involved in reduction of dinitrogen, and which may have had a broad speciicity that may (or may not) have been associated with nitrogen assimilation (Silver and Postgate 1973; Burke et al. 1993; Fani et al. 2000; Raymond et al. 2004). For example, site-directed nitrogenase mutants as well as nitrogenase itself have been shown to have some competency for reducing cyanide, generating methane as well as ammonia, which could then serve as a nitrogen source (Kao et al. 2003; Pickett et al. 2004). This idea is intriguing given the hypothesis that Archean nitrogen reservoirs might have progressed from ammonia to cyanide and ultimately to dinitrogen, transitions quite possibly captured in the evolutionary history of nitrogenase and other nitrogen assimilating enzymes. The importance of ammonia to early organisms, arguments for an early Earth nitrogen crisis, and the solitary role of nitrogenase as a shunt into the atmospheric dinitrogen reservoir suggest that this enzyme might have been of critical importance to early life, especially as increasing numbers of organisms depleted the limited supply of bio-available nitrogen. The presence of the nitrogenase enzyme among diverse phyla of bacteria and in methanogenic archaea have led some authors to argue that nitrogenase was present in the last common ancestor (LCA) of extant organisms (Fani et al. 2000; Normand et al. 1992). While compelling based on the arguments above, the evolution of nitrogenase has been very complex and includes examples of: gene duplication resulting in numerous families of nitrogenase homologs; horizontal gene transfer within and between the bacterial and archaeal domains; and apparent loss of nitrogenase from many lineages. Raymond et al. (2004) recently proposed a scenario whereby nitrogenase was not present in the LCA, but was “invented” in methanogenic archaea and subsequently horizontally transferred into bacteria. While dificult to argue whether this late-origin proposal is more parsimonious than the LCA hypothesis, the former has an advantage in not invoking the perplexing disappearance of nitrogenase from many early branching lineages, such as eukaryotes, crenarchaeotes, and most phyla of early branching bacteria.

GEOLOGICAL CLUES TO THE EARLY NITROGEN CYCLE Phylogenetic and genomic analyses, the essential role of nitrogenous compounds in biochemistry, and the ubiquity of nitrogen-assimilating enzymes suggests that biological interactions with N2, NH3, and possibly other nitrogenous compounds such as cyanide began early in the evolution of life. Determining which of these pathways might be the most primitive or could have been dominant among early organisms remains considerably more ambiguous. The GS/GOGAT pathway, universally required for the synthesis of nitrogenous compounds by amine and amide transfers, appears to have been an early invention, possibly predating the LCA as it is found across all three domains of life. This system might have been suficient to support the synthesis of nitrogenous compounds if ammonia was present in ample concentrations on the Archean Earth. However, arguments for an early nitrogen crisis (or, at the very least, as a limited supply of nitrogenous compounds became locked up in biomass) suggest a rapidly increasing necessity for alternative shunts for assimilating nitrogen from the environment, such as nitrogenase for reducing atmospheric dinitrogen to ammonia (Kasting and Siefert 2001). Though there remains debate as to whether nitrogenase was present in the LCA, it seems likely that the enzyme evolved during the Archean, prior to the evolution of cyanobacteria and the ensuing oxidation of the Earth’s atmosphere (Navarro-Gonzalez et al. 2001). Unfortunately the arguably diagnostic isotopic signatures in carbon fractionation that result from different modes of autotrophy have thus far been substantially less forthcoming for nitrogen isotopes. Nitrogenous compounds are typically very soluble and characterized by a short residence time in sediments and crustal rocks. Beaumont and Robert (Beaumont and Robert 1998, 1999) determined isotopic compositions of cherts ranging from 3.5 billion to 500 million years in age, noting a transition from mostly negative to all positive values

218

Raymond

occurring during the transition from the Archean to the Proterozoic. This concurs with the invention of oxygenic photosynthesis and the ensuing oxidation of Earth’s atmosphere, likely representing the increasing availability (due to abiotic ixation of nitrogen as well as O2dependent biological nitriication) of soluble oxidized nitrogen such as NO3−2. Importantly, though negative excursions during the Archean are consistent with assimilation of NH3 as well as the biological ixation of dinitrogen by nitrogenase, these signatures typically not regarded as diagnostic for either process.

BIOLOGICAL CARBON CYCLING AND AUTOTROPHY Autotrophy is typically deined as the autocatalytic, reductive conversion of CO2 into the low-molecular-weight building blocks of biosynthesis (e.g., Hugler et al. 2003), and by root derivation (self-nourishing) applies to organisms that derive energy for this reaction from light (photosynthesis) or inorganic compounds (chemotrophy). The following discussion gives a broad glimpse of the four known pathways used by autotrophs to assimilate CO2, focusing in brief on the unique or key enzymes of each pathway and discussing the current evolutionary understanding of these enzymes. The four pathways are summarized in Figure 3, and it must be emphasized that key steps in the 3-hydroxypropionate pathway are still being elucidated. Finally, the distribution of these pathways in the three domains is discussed, as is the possible relevance to the origin and early evolution of life. Though not covered in detail here, it is worthwhile to briely detail and clarify some of the other carbon assimilatory pathways that have been elucidated. A remarkable diversity of organisms exist which can use various C1 compounds—organic or inorganic compounds containing, among other elements, only a single carbon atom—as sources of both energy and cell carbon. Methylotrophs are an evolutionary distinct class of organisms that can utilize C1 compounds in oxidation states ranging from formaldehyde to methanol to methane (including aminated, sulfurated, and halogenated methane). Methanotrophs are a subset of methylotrophs that speciically metabolize methane using either O2-dependent methane monooxygenases, as is found in recently sequenced Methylococcos capsulatus and other proteobacteria, or a recently elucidated anaerobic pathway found in domain Archaea that roughly appears to operate as methanogenesis in reverse (Hallam et al. 2004). C1 assimilation pathways are also known for numerous other compounds, such as formate, cyanide, and carbon monoxide, and organisms able to use CO as both sole carbon and energy source raise intriguing exceptions to the long-held “CO2-only” deinition of autotrophy. Carbon ixation in methanotrophs, on the other hand, takes place through one of two non-autotrophic pathways, the serine or ribulose monophosphate/allulose pathways, by which formaldehyde (the common product of C1-utilization) is incorporated. Intriguingly, these two pathways also incorporate CO2, but are effectively mixotrophic because of their dependence on formaldehyde. However, some methanotrophs, such as M. capsulatus, also ix carbon via the Calvin cycle and thereby are true autotrophs. Yet another class of so-called anaplerotic pathways are important in both autotrophs and heterotrophs, and anaplerotic enzymes catalyze the direct incorporation of CO2 into organic compounds for the regeneration of essential intermediates of central metabolism. The familiar Krebs/TCA cycle, which carries out the energy-generating breakdown of pyruvate into three CO2 molecules, occupies a central position in metabolism as its products serve as biosynthetic precursors for an enormous repertoire of compounds, including amino acids, nucleotides, cofactors, and lipids. Anaplerotic enzymes allow the direct replenishment of Krebs/TCA cycle intermediates, effectively “recapturing” CO2 that is lost by running the cycle. The very presence of these anaplerotic shunts gives some insight into how CO2-ixing pathways might

Evolution of Biologic C & N CyclingȰGenomic Perspective

219

Figure 3. Four known pathways for autotrophic carbon ixation. Enzymes are in boxes, denoting in particular those that are discussed in the text. The 3-hydroxypropionate cycle schema adapted from (Herter et al. 2002).

have evolved. Indeed, there are over distinct 400 reactions—almost 5% of the total—annotated in the KEGG database that directly consume or produce CO2, arguing that biological interactions with CO2 have been of foremost importance in the evolution of life.

WOOD-LJUNGDAHL (REDUCTIVE ACETYL-COA) PATHWAY In recent years, considerable progress has been made in enzymatic and mechanistic understanding of the Wood-Ljungdahl or reductive acetyl-CoA pathway. This progress has come particularly in the way of crystal structures for two key enzymes in the pathway. The Wood-Ljungdahl pathway and enzymes involved are of particular interest to evolutionary biologists, bearing many hallmarks that suggest an ancient origin and an important role among primitive autotrophs. For instance, the stepwise condensation of C1 units to form C2 and C3 compounds such as acetate and pyruvate, respectively, represents schematically the simplest form of autotrophy. Furthermore, the enzymes catalyzing each step bear arguably “primitive” characteristics—unique metal clusters and a corresponding sensitivity to molecular oxygen, as well as a distribution among many early-branching lineages on the tree of life. Note that the

220

Raymond

pathway is not cyclic/autocatalytic, though the Wood-Ljungdahl pathway has the distinction among all known autotrophic pathways in being net exergonic, according to: 2CO2 + 4H2 + n ADP + nPi o CH3COOH + 2H2O + n ATP

with an overall change in free energy of dG° = −95 kJ/mol (Muller 2003). This pathway is unique among known autotrophic assimilation pathways in allowing ixation concomitant with generation of ATP. Importantly, this dual role of energy generation and carbon ixation is not universal among microorganisms, many of which have components of the Wood-Ljungdahl pathway but are not autotrophs. Whereas methanogens, acetogens, and aerobic carboxydotrophs depend upon the pathway for chemoautotrophy and subsequent biosynthesis (and thereby actively take up CO2 from the environment), many non-autotrophic anaerobic fermenters utilize this pathway to further reduce CO2 generated from pyruvate decarboxylation to CO (Ragsdale 2004). The key enzyme for the reduction of CO2 to CO is carbon monoxide dehydrogenase (CODH), which catalyzes this reversible reaction using a variety of electron donors. The enzyme is one of a handful of enzymes that requires elemental nickel, though an oxygen-tolerant CODH that apparently utilizes molybdenum rather than nickel has been characterized from carboxydotrophs (Meyer et al. 1990). The other key enzyme in this pathway, acetyl-CoA synthase (ACS), has also been shown to contain two nickel atoms complexed to an ironsulfur cluster (despite their common nickel-containing metal clusters, ACS and CODH are not thought to be homologous) (Hegg 2004). In autotrophs, the ACS enzyme forms a bifunctional complex with CODH, where CO generated by CO2 reduction is passed through a protein “channel” from CODH to ACS, then combined with methane and coenzyme A (CoA) to form acetyl-CoA (Hegg 2004; Ragsdale 2004). Figure 4 shows the CODH phylogeny, inferred from sequences of completely sequenced genomes. As can be observed, gene duplication has been extensive, and there is a lack of overall taxonomic cohesion, as in the paraphyly of methanogens. Bootstrap support and conserved sequence signatures suggest that the phylogeny is robust overall, and that horizontal gene transfer is a good explanation for some of the topology. A number of more distant CODH homologs are excluded from this tree. These include the hydroxylamine reductase/hybrid cluster protein—to which CODH is still closely related enough to be “converted” into by sitedirected mutagenesis (Heo et al. 2002)—along with very distantly related CODH homologs from other bacteria and archaea that may carry out unrelated functions.

THE rTCA CYCLE With the exception of aerobic carboxydotrophs briely mentioned above, both the WoodLjungdahl pathway and reverse or reductive tricarboxylic acid (rTCA) cycle are typically the realm of anaerobic or microaerobic autotrophs, thanks in part to O2-sensitivity of many of their key enzymes. The rTCA cycle, as its name implies, is the conceptual “reverse” of the familiar TCA (also Krebs or citric acid) cycle which, rather than oxidizing acetate (to two molecules of CO2) concomitant with the production of reducing equivalents, uses reducing equivalents to ix CO2 ultimately to acetyl-CoA. The net production of acetyl-CoA is also found in the WoodLjungdahl pathway, though the rTCA cycle differs in being endergonic and cyclic/network autocatalytic (cycle intermediates can serve as templates for their own synthesis and—as “carriers” of assimilated carbon—are not consumed as the cycle propagates). Three key enzymes of the rTCA cycle distinguish it from the archetype Krebs cycle: fumarate reductase, ATP citrate lyase, and 2-oxoglutarate:ferredoxin oxidoreductase. 2-oxoglutarate:ferredoxin oxidoreductase catalyzes the carboxylation of succinyl-CoA, forming 2-oxoglutarate. ATP

Evolution of Biologic C & N CyclingȰGenomic Perspective 99 100 71 50

65 99

100 71 99 100 100 100 100 86 100

100 93 100 70

221

D.hafniense2 R.rubrum H.mobilis2 M.thermoacetica1 D.hafniense1 A.vinelandii G.sulfurreducens pca D.desulfuricans D.vulgaris M.barkeri2 M.acetivorans2 M.mazei2 M.thermoacetica2 C.thermocellum G.metallireducens M.jannaschii M.kandleri C.acetobutylicum1 H.mobilis1 C.acetobutylicum2 100 D.hafniense3 a. fulgidus dsm4304 M.barkeri1 M.acetivorans1 100 M.mazei1 91

0.1

Figure 4. CODH protein phylogeny, constructed as described in the legend for Figure 2. Numbers indicate duplicate CODH copies in a single genome. Many organisms in fact have two or more CODH paralogs, and the interspersed bacterial-archaeal phyla indicate a highly nonvertical evolution likely beset by horizontal gene transfer. Methods as given in Figure 2.

citrate lyase catalyzes the citrate “cleavage” that closes the cycle, forming oxaloacetate and acetyl-CoA. 2-oxoglutarate:ferredoxin oxidoreductase is a member of the diverse thimine diphosphate-dependent 2-oxoacid oxidoreductases and, along with its homolog pyruvate: ferredoxin oxidoreductase (PFOR), is able to “reverse” the oxidative decarboxylations carried out by unrelated dehydrogenases of the Krebs cycle, incorporating CO2 into biomolecules. The evolutionary history of PFOR, key in synthesizing pyruvate from acetyl-CoA in WoodLjungdahl as well as rTCA autotrophs, is shown in Figure 5. The phylogeny is congruent in many respects with the 16S rRNA-based tree of life, but the enzymes are clearly overrepresented in organisms that have an obligately or facultatively anaerobic lifestyle (with notable exceptions to both of the previous points). As with the Wood-Ljungdahl pathway, the rTCA cycle is also found among diverse prokaryotes, many of which are early-branching lineages on the tree of life. The pathway was irst described from the obligately anaerobic, non-oxygen producing phototroph Chlorobium tepidum, where the electrons generated during photosynthesis eventually enter into the rTCA cycle for carbon ixation (Evans et al. 1966). So-called Knallgas bacteria (metabolizing via 2H2 + O2 o 2H2O), such as members of phylum Aquiicales, are microaerophiles that use the rTCA cycle. The pathway is also found in autotrophic Crenarchaeota, sulfate-reducing bacteria, and in some chemolithotrophic epsilon Proteobacteria (Shiba et al. 1985; Schafer et al. 1986; Schauder et al. 1987; Hugler et al. 2005). Remarkably, several of the key enzymes from the rTCA cycle are found among a broad range of aerobic heterotrophs. Citrate lyase, for example, has homologs in humans and other vertebrates that function in fatty acid metabolism

Raymond

222

A.variabilis|1964 Nostoc sp|2284 C.watsonii|1367 100 Synechocystis 6803|2805 99 75 A.variabilis|4948 Nostoc sp|3176 100 87 S.elongatus|858 C.aggregatum|1493 36 C.tepidum tls|1594 100 100 D.acetoxidans|1527 D.acetoxidans|5881 74 G.metallireducens|1832 100 G.sulfurreducens pca|97 100 D.aromatica|1842 97 R.rubrum|2304 60 C.aurantiacus|1199 68 M.capsulatus|13637 91 S.typhi|1244 100 E.coli O157H7|2086 100 100 S.flexneri 2a|1720 L.lactis|422 100 D.desulfuricans|1201 90 D.vulgaris hildenborough|2993 90 M.thermoacetica|732 C.jejuni|1392 95 H.mobilis|1691 76 100 C.perfringens|2124 88 C.acetobutylicum|2199 D.hafniense|204 38 C.acetobutylicum|2466 97 C.thermocellum|1873 100 D.hafniense|2349 54 100 S.solfataricus|1088 46 S.tokodaii|1662 S.tokodaii|1937 74 S.solfataricus|2501 100 A.fulgidus dsm4304|2021 M.thermoacetica|326 65 C.thermocellum|2458 * M.thermoacetica|2093 49 100 T.acidophilum|618 T.volcanium|849 24 100 M.barkeri|2496 100 M.mazei|1340 100 M.acetivorans|32 24 60 M.burtonii|1125 A.fulgidus dsm4304|1678 M.thermoautotrophicum|1700 35 M.kandleri|82 45 99 M.jannaschii|271 48 M.maripaludis|1505 P.furiosus|966 61 P.abyssi|1344 100 P.horikoshii|710 99 100

73

Cyanobacteria

Green sulfur bacteria

Proteobacteria/ ChloroÀexus

Gram-positive (Clostridia/Bacilli)

Crenarchaeotes *

Euryarchaeotes

0.1

Figure 5. Pyruvate:ferredoxin oxidoreductase (PFOR) protein phylogeny. As discussed in the text, PFOR is a decarboxylating enzyme found in both heterotrophs and autotrophs and functions reversibly as a key carboxylating enzyme in the rTCA and Wood-Ljungdahl pathways. There is a clear stratiication of bacteria (top clade) and archaea (bottom clade), with distributions loosely consistent with phylum-level taxonomy (shown in boxes; asterisks indicate notable deviations). Methods as given in Figure 2.

Evolution of Biologic C & N CyclingȰGenomic Perspective

223

and biosynthesis (Wahlund and Tabita 1997). Thus detection of the rTCA cycle as a suspected autotrophic pathway in new organisms must be based on multiple factors—phenotype and the presence of all key genes—rather than the presence of any single diagnostic gene. The rTCA cycle has been argued by some authors to have been a pivotal pathway not only in the evolution of life but in the origin of metabolism. The pathway occupies a unique position at the hub of most metabolic networks, with metabolites used as precursors in biosynthesis of all major classes of biochemical compounds. Several of the enzymatic steps have plausible prebiotic analogs, most notably discussed by Wächtershäuser in his hypothesis for primitive sulfur-linked “thioanalogs” of modern rTCA cycle intermediates (Wächtershäuser 1990). This had the argued advantage of overcoming the unfavorable kinetics of several reactions, e.g., beta-carboxylations, of the rTCA cycle as well as inevitably linking this early metabolism with a sulfur/pyrite-rich interface. Thermodynamically unfavorable reactions could thereby be coupled to pyrite oxidation, much as coupling to ATP hydrolysis provides the basis for dificult modern biochemical reactions (Bebie and Cody 2000). These arguments have also been carried forth by Smith and Morowitz (2004), who argue that such a network-autocatalytic (self-replicating), redox-intermediate cycle is exactly what natural selection might be expected to produce. However, in the absence of catalysis, detractors point out that several critical steps in the pathway are simply not expected as favored outcomes (Orgel 2000). Because many of these arguments focus on pre-LCA metabolism, they are simply beyond the resolution of comparative genomics and phylogenetics. While the centrality of the rTCA cycle in metabolic charts is striking, it could just as likely represent a favorable but relatively recent reorganization of pre-existing metabolic pathways.

THE CALVIN-BENSON-BASSHAM CYCLE The Calvin-Benson-Bassham (CBB) cycle represents the best known of the autotrophic pathways, most notably because it is the only one present in eukaryotes, making it amenable to now-infamous large scale preparations from e.g., spinach for biochemical and biophysical studies. The key enzyme of the pathway, ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO), is the key enzyme of the cycle. The enzyme itself falls somewhere between evolutionary enigma and embarrassment, having a turnover rate of only a few carboxylation reactions per second and also carrying an intrinsic, energetically wasteful proclivity for using molecular oxygen, rather than carbon dioxide, as a substrate. Photoautotrophs, plants in particular, attempt to compensate for RuBisCO’s ineficiencies by overexpressing the enzyme, and as a result RuBisCO is by far the most abundant protein in perhaps the largest biomass group on the planet, making it arguably the most abundant enzyme on Earth. The CBB cycle is in essence split into three separate phases. The irst stage, carboxylation or carbon ixation, takes place by the RuBisCO-catalyzed addition of CO2 to ribulose bisphosphate, a C5 sugar that is synthesized from ribose-1,5-bisphosphate by the other key/ unique enzyme of the Calvin cycle, phosphoribulokinase (PRK). The carboxylation step is exergonic and essentially irreversible, though the overall cycle is net endergonic. The resultant C6 sugar is subsequently cleaved to two C3 3-phosphoglycerates, which are converted to glyceraldehydes 3-phosphates (G3P)—a highly utilized carbon “currency” in many cells— during the so-called reductive phase of the CBB cycle. The reductive phase requires a net input of ATP and NADPH. Because a 3C compound is output, the Cycle must turn three times per G3P generated, so that cycle intermediates are not depleted. The regeneration phase of the cycle then incorporates a number of ATP-dependent aldolase and isomerase steps to regenerate the 5C precursor to RuBisCO carboxylation. The competing oxygenase activity of RuBisCO results in a process termed photorespiration, whereby O2 rather than CO2 combines with ribulose-1,5-bisphosphate, yielding one molecule

224

Raymond

of 3-phosphoglycerate and another of 2-phosphoglycolate. This latter compound must be salvaged—reconverted into usable 3-phosphoglycerate in a series of reactions that require both ATP and NADH. As this salvage pathway does not result in ixed carbon and requires energy and reducing equivalents to amend, photorespiration may reduce the eficiency of the Calvin cycle by as much as 50%. To counter the effects of photorespiration, a suite of novel carbon concentrating mechanisms (CCM’s) have evolved to ramp up the ratio of CO2 to O2 inside cells and chloroplasts (e.g., Badger et al. 2002), responsible for, among other adaptations, the carboxysome that is among the largest known enzyme aggregates. Notably, some recent evidence suggests that photorespiration may in fact be important in higher plants as an energy sink or source of metabolites (Wingler et al. 2000). The evolutionary history and diversity of the Calvin cycle has been studied in great detail, as evidenced by the enormous repository of RuBisCO gene sequences (~25,000) available in GenBank. A disproportionately large number of these sequences are from eukaryotes, so the focus here is constrained to the more diverse RuBisCO (large subunit) sequences available from completed prokaryotic genomes. As illustrated in Figure 6, several distinct clades are evident in the phylogenetic tree, and importantly some of these represent proteins not known to function within the Calvin cycle, as evident based on active site substitutions, absence of PRK, and experimental veriication. True RuBisCO’s fall into two groups, so-called form I and form II (with subgroups within each) that highlighted on the tree. Form III and IV homologs of RuBisCO, the so-called RuBisCO-like-proteins (RLP’s), have become a recent focus of several investigations. A suite of recent experiments by Tabita and colleages has illustrated quite compellingly that the form III RuBisCO homologs are able to ix atmospheric CO2, using a functional analog of PRK to generate a substrate for the enzyme (Finn and Tabita 2004). While it is not evident that these methanogens use this alternative pathway (most are Wood-Ljungdahl autotrophs) as a primary mode of carbon ixation (Sprott et al. 1993), this may serve an important anaplerotic role and provides an exciting window into RuBisCO evolution and possible engineering. Conversely, the distantly-related form IV RuBisCO homologs, found in a diverse range of organisms including several anoxygenic phototrophs (Ashida et al. 2003), is argued to be involved with methionine salvage pathways and is not able to ix CO2. Importantly, with the exception of RuBisCO and PRK, all of the enzymes used in the Calvin cycle are found in functions known from other pathways. At least schematically, the Calvin cycle bears tantalizing similarities with the aforementioned ribulose monophosphate (RuMP) pathway found in many methanotrophs, functioning to incorporate C1 units in the form of formate rather than CO2 (analogously driving C1 + C5 o C6 ixation in the process). Thus it can be argued that most of the enzymes of the Calvin cycle were already present in the ancestors of the irst Calvin cycle autotrophs, and that “invention” of RuBisCO and PRK, combined with recruitment of enzymes from the RuMP pathway, could have resulted in a new, O2-tolerant form of autotrophy. The timing of the appearance of the Calvin cycle appears to have been closely linked to the development of bacterial aerobic respiration and photosynthesis, events that were both tied to the progressive and irreversible oxidation of Earth’s atmosphere. As the rTCA cycle and Wood-Ljungdahl pathways are both inhibited at high oxygen concentrations, it seems reasonable that RuBisCO came about in response to the challenge of an atmosphere that was increasingly oxic and with decreasing CO2 availability. The taxonomically-diverse RLP’s, recently discovered to be involved with methionine salvage, present a plausible ancestral pathway from which RuBisCO was recruited and came to function as a carboxylase, and structure-function analyses are presently providing a testbed for this hypothesis (Ashida et al. 2005). The unique dual oxygenase/carboxylase role of the enzyme would have been much less of a disadvantage as the Precambrian CO2:O2 ratio was still high, and feasibly could have served a useful additional role in ameliorating high levels of oxygen.

Evolution of Biologic C & N CyclingȰGenomic Perspective

225

C.watsonii Synechocystis 6803 T.erythraeum 46 N.punctiforme Form IA 35 100 A.variabilis (cyanobacteria) N.ostoc sp 100 83 G.violaceus 100 T.elongatus S.elongatus 100 R.metallidurans 92 100 N.europaea Form IB M.capsulatus (ȕ,Ȗ-proteobacteria, Synechococcus WH8102 99 marine cyanos) 100 P.marinus MED4 100 100 P.marinus CCMP1375 58 P.marinus MIT9313 B.fungorum Form II R.sphaeroides 100 100 (Į,ȕ-proteobacteria) S.meliloti B.japonicum 36 R.palustris CGA009 100 97 P.abyssi 100 P.horikoshii P.furiosus A.fulgidus dsm4304 Form III M.jannaschii (Archaea: M.acetivorans euryarchaeotes) M.barkeri 100 M.mazei 95 M.burtonii Form IV (b) M.magnetotacticum (Bacteria, D.aromatica 100 Archaea) R.rubrum 92 C.tepidum tls 100 C.aggregatum B.bronchiseptica Form IV (a) H.mobilis (Bacteria) Exiguobacterium B.subtilis 100 B.anthracis Ames 0581 99 100 B.cereus ATCC14579 100

65

61

49 100 89 42

100

40

53

0.1

Figure 6. RuBisCO large subunit phylogeny, showing the four different clades or Forms of RuBisCO that have been established in previous evolutionary studies. As discussed in the text, Forms I and II are key enzymes in Calvin-Benson-Bassham cycles, whereas Forms III and IV are RuBisCO-like proteins, recently shown to function in alternative carbon ixation or methionine salvage pathways, respectively. Methods as given in Figure 2.

3-HYDROXYPROPIONATE CYCLE This pathway was irst discovered only fairly recently in the ilamentous anoxygenic phototroph Chloroflexus aurantiacus, of evolutionary interest because of its early-branching position on the tree of life. In fact many of the steps in this unique pathway are only now being fully resolved, particularly by the investigations of Fuchs and colleagues (Herter et al. 2002). However, it has also recently become clear that the cycle is more widespread than previously thought, apparently functioning in many aerobic autotrophic crenarchaeota (Hugler

226

Raymond

et al. 2003). Schematically, the pathway as presently elucidated incorporates a combination of novel reactions with enzymatic steps recruited from other metabolic pathways, including the rTCA cycle and propionyl-CoA pathway which carries out odd-chain fatty acid oxidation. Carbon ixation ultimately leading to pyruvate actually comprises a bicyclic pathway; in the irst cycle, two carboxylations take place to give rise to a glyoxylate intermediate, which then condenses with an intermediate product from the irst half of the irst cycle (which accounts for the third carbon) (Herter et al. 2002). Several rearrangements, schematically similar to a reversed glyoxylate cycle but technically like a reversed citramalate pathway, eventually give rise to the cleaved pyruvate end-product. Though this remarkable pathway has now been solved in considerable detail, its evolution still remains enigmatic. Somewhat perplexing is that, despite its discovery in type strain C. aurantiacus, some close relatives of this organism (within phylum Chlorolexi) apparently do not have a 3-hydroxypropionate cycle, instead apparently having a functioning CBB cycle (Ivanovsky et al. 1999). Furthermore, many of these organisms are content to grow (photo)heterotrophically and are often found in microbial mats with abundant biomass to facilitate such a lifestyle. As shown in Figure 7, this pathway is found only in a few different groups of aerobic prokaryotes, suggesting that it, as with the Calvin cycle, might have arisen as an ad hoc autotrophic solution to the oxidation of Earth’s atmosphere, bypassing the oxygen-sensitive enzymes that catalyze critical roles in the rTCA cycle and Wood-Ljungdahl pathway. This speculation has some support from genomic data as it appears that, while homologs to several of the key enzymes that C. aurantiacus uses are present in crenarchaeotes, some enzymatic steps may be different in different crenarchaeota, suggesting evolutionary convergence or some plasticity in pathway operation (Menendez et al. 1999). The relatively recent discovery of this cycle and the apparently diverse means by which organisms carry it out suggest that there are pathways for autotrophic carbon ixation that have yet to be discovered, and resolving these as well as known pathways will no doubt shed additional light on the origin and evolution of microbial-environmental interactions.

AUTOTROPHY, HETEROTROPHY, AND THE ORIGIN OF METABOLISM For several decades, origin of life research focused on establishing a diverse synthetic library of prebiotic compounds upon which early life might have been built. A so-called heterotrophic origin assumes that a relatively complex milieu of chemical compounds would have been present on the early earth, supplied exogenously (e.g., during cometary accretion) or through abiotic reactions involving prevalent inorganic precursors such as N2, CO2, and H2 (and typically involving a mineral catalyst and/or an energy source such as lightning to overcome energetic barriers). Plausible synthetic sources for many essential ingredients of prokaryotic cells have been illustrated, and it is thought that as early cells became more complex, enzymatic routes of synthesis eventually replaced their prebiotic counterparts as their prebiotic precursors became increasingly scarce. While impressively successful in the breadth of biologically-relevant metabolites that have been obtained, heterotrophic theories have not been without criticism. Opponents cite that reaction conditions often invoke conditions that are quite different than geochemical evidence suggests might have been available on the early Earth, as well as the fact that it is dificult to imagine a single environment where all of the necessary catalysts and conditions would allow precursors to accumulate. Thus about two decades ago, autotrophic origin of life theories proposed that prebiotic reactions might have involved only inorganic precursors, catalyzed for example on mineral surfaces or within FeS “membranes” (Russell et al. 1988; Wächtershäuser 1988). Such schemes arguably allow for locally high concentrations of metabolites to accumulate and invoke a comparably simple set

Figure 7. Distribution of autotrophs on the tree of life, based on the established carbon ixation pathways discussed in the text, signature genes present in completed genomes, and experimental detection of pathways as inferred from current literature. squares=reductive TCA cycle, circles=Calvin-Benson-Bassham cycle, triangles=Wood-Ljungdahl pathway, diamonds=3-hydroxypropionate cycle. Gradient-illed shapes indicate pathways that are found in some, but not all, members of a particular clade. For example, members of the gamma proteobacteria, which includes A. vinelandii and P. aeruginosa, are known to use the Calvin cycle, though neither of these two organisms speciically is an autotroph.

Evolution of Biologic C & N CyclingȰGenomic Perspective 227

228

Raymond

of necessary catalysts and precursors, for example envisaging environments not unlike modern hydrothermal systems. Experimental corroboration of the metabolic world attainable by autotrophic prebiotic pathways is providing support, but a number of important gaps remain to be illed (see e.g., Cody 2004). Both schools of thought at least broadly agree that the individual abiotic steps within an increasingly complex metabolic network were progressively assumed by enzyme (or ribozyme) catalysts. Of relevance to this discussion, if the autotrophic theory were true, the simplest—though by no means the most likely—assumption would be that primitive carbon and nitrogen cycles were functioning in the earliest organisms so as to directly assimilate N2 and CO2. Certainly enzymes such as nitrogenase and carbon monoxide dehydrogenase have been noted for their unique metal cluster complement (Rees and Howard 2003), which poses the intriguing idea (though without much support) that these metal clusters represent the stillfunctional remnants of prebiotic mineral catalysts that carried out analogous reactions. While these putative glimpses of prebiotic chemistry through modern enzymology are tantalizing, it seems just as likely that many generations of overprinting have occurred; distinct and unrelated enzymes have evolved to carry out these ancient reactions more eficiently than their predecessors. The rTCA cycle and the Wood-Ljungdahl pathway both bear hallmarks of a primitive autotrophy, as compellingly argued by a number of authors (Wächtershäuser 1990; Morowitz 1999; Cody 2004; Smith and Morowitz 2004). As Figure 7 illustrates, both pathways operate in a number of deeply-branching anaerobes, whereas the CBB and 3hydroxypropionate cycles are found in clades closer to the “tips” of the tree of life, suggesting more recent origins. Furthermore, the rTCA and W-L pathways contain enzymes with diverse metal clusters that show sensitivity to molecular oxygen, and concomitantly are found predominantly in anaerobes and microaerobes. Each pathway also boasts speciic attractive features such as being network autocatalytic (rTCA), net exergonic (W-L), centrally connected to diverse biosynthetic pathways (rTCA), and catalyzing reactions between arguably “primitive” metabolites (W-L). Importantly however, their sporadic though widespread distribution on the tree of life prevents either of these pathways from being deinitively argued to have been present in the last common ancestor, and the possibility remains open that they originated as speciic, convergent CO2-ixing adaptations in early Earth niches—still ancient, but far removed from the de facto origin of life. Contrarily, several lines of reasoning suggest that the Calvin-Bensen-Bassham cycle represents the youngest autotrophic pathway. This is especially remarkable as the pathway is the dominant mechanism for autotrophic carbon ixation on the modern Earth, primarily as a result of the cyanobacterial endosymbiosis that gave rise to plastids and chloroplasts in modern algae and plants. Based on phylogenetic analysis of RuBisCO and its diverse homologs not involved in the Calvin-Benson-Bassham cycle, the CBB cycle appears likely to have originated during the anoxic-oxic transition, in close conjunction with the invention of oxygenic photosynthesis and O2-based respiration. This new autotrophy permitted a new legion of aerobic autotrophs the ability to ix carbon while using O2 as a high potential terminal electron acceptor and was tightly linked to both photosynthesis and respiration, allowing CO2 ixation to occur within a much more favorable energetic milieu than anaerobic habitats previously afforded. In the 2-2.5 billion years that have ensued since the appearance of the Calvin cycle, it is perhaps surprising that autotrophy has maintained any predominance, especially as heterotrophy—the ability to degrade and recycle organic biomass—has assumed hegemony in many families of microorganisms and most higher eukaryotes. However, as our understanding of global carbon cycling improves, it seems clear that the role of autotrophs has anything but diminished.

Evolution of Biologic C & N CyclingȰGenomic Perspective

229

REFERENCES Ashida H, Danchin A, Yokota A (2005) Was photosynthetic RuBisCO recruited by acquisitive evolution from RuBisCO-like proteins involved in sulfur metabolism? Res Microbiol 156(5-6):611-618 Ashida H, Saito Y, Kojima C, Kobayashi K, Ogasawara N, Yokota A (2003) A functional link between RuBisCO-like protein of Bacillus and photosynthetic RuBisCO. Science 302(5643):286-290 Badger MR, Hanson D, Price GD (2002) Evolution and diversity of CO2 concentrating mechanisms in cyanobacteria. Funct Plant Biol 29(2-3):161-173 Beaumont V, Robert F (1998) Nitrogen isotopic composition of organic matter from Precambrian cherts: new keys for nitrogen cycle evolution? B Soc Geol Fr 169(2):211-220 Beaumont V, Robert F (1999) Nitrogen isotope ratios of kerogens in Precambrian cherts: a record of the evolution of atmosphere chemistry? Precambrian Res 96(1-2):63-82 Bebie J, Cody G (2000) Pyruvic acid condensation chemistry on transition-metal sulide surfaces. Abstr Pap Am Chem S 219:U697-U697 Bork P (2000) Powers and pitfalls in sequence analysis: the 70% hurdle. Genome Res 10:398-400 Brown JR, Masuchi Y, Robb FT, Doolittle WF (1994) Evolutionary relationships of bacterial and archaeal glutamine synthetase genes. J Mol Evol 38(6):566-76 Burke DH, Hearst JE, Sidow A (1993) Early evolution of photosynthesis: clues from nitrogenase and chlorophyll iron proteins. Proc Natl Acad Sci 90(15):7134-8 Cody GD (2004) Transition metal sulides and the origins of metabolism. Annu Rev Earth Pl Sc 32:569-599 Deppenmeier U, Johann A, et al. (2002) The genome of Methanosarcina mazei: evidence for lateral gene transfer between bacteria and archaea. J Mol Microbiol Biotechnol 4(4):453-61 DesMarais DJ (1997) Long-term evolution of the biogeochemical carbon cycle. Rev Mineral 35:429-448 Dincturk HB (2001) Glutamate synthase: An archaeal horizontal gene transfer? J Biosciences 26(1):13-14 Dincturk HB, Knaff DB (2000) The evolution of glutamate synthase. Mol Biol Rep 27(3):141-148 Evans MC, Buchanan BB, Arnon DI (1966) A new ferredoxin-dependent carbon reduction cycle in a photosynthetic bacterium. Proc Natl Acad Sci USA 55(4):928-34 Falkowski P, Scholes RJ, et al. (2000) The global carbon cycle: a test of our knowledge of Earth as a system. Science 290(5490):291-6 Fani R, Gallo R, Lio P (2000) Molecular evolution of nitrogen ixation: the evolutionary history of the nifD, nifK, nifE, and nifN genes. J Mol Evol 51(1):1-11 Finn MW, Tabita FR (2004) Modiied pathway to synthesize ribulose 1,5-bisphosphate in methanogenic archaea. J Bacteriol 186(19):6360-6366 Hallam SJ, Putnam N, Preston CM, Detter JC, Rokhsar D, Richardson PM, DeLong EF (2004) Reverse methanogenesis: testing the hypothesis with environmental genomics. Science 305(5689):1457-62 Hayes JM (1994) Global methanotrophy at the Archean-Proterozoic transition. In: Early life on Earth. Bengtson S (ed), New York, Columbia University Press, p 220-236 Hegg EL (2004) Unraveling the structure and mechanism of acetyl-coenzyme A synthase. Accounts Chem Res 37(10):775-783 Heo J, Wolfe MT, Staples CR, Ludden PW (2002) Converting the NiFeS carbon monoxide dehydrogenase to a hydrogenase and a hydroxylamine reductase. J Bacteriol 184(21):5894-5897 Herter S, Fuchs G, Bacher A, Eisenreich W (2002) A bicyclic autotrophic CO2 ixation pathway in Chlorolexus aurantiacus. J Biol Chem 277(23):20277-83 Howard JB, Rees DC (1996) Structural basis of biological nitrogen ixation. Chem Rev 96(7):2965-2982 Hugler M, Huber H, Stetter KO, Fuchs G (2003) Autotrophic CO2 ixation pathways in archaea (Crenarchaeota). Arch Microbiol 179(3):160-73 Hugler M, Wirsen CO, Fuchs G, Taylor CD, Sievert SM (2005) Evidence for autotrophic CO2 ixation via the reductive tricarboxylic acid cycle by members of the epsilon subdivision of proteobacteria. J Bacteriol 187(9):3020-7 Ivanovsky RN, Fal YI, et al. (1999) Evidence for the presence of the reductive pentose phosphate cycle in a ilamentous anoxygenic photosynthetic bacterium, Oscillochloris trichoides strain DG-6. Microbiol-Sgm 145:1743-1748 Kao CM, Liu JK, Lou HR, Lin CS, Chen SC (2003) Biotransformation of cyanide to methane and ammonia by Klebsiella oxytoca. Chemosphere 50(8):1055-61 Kasting JF, Siefert JL (2001) Biogeochemistry. The nitrogen ix. Nature 412(6842):26-7 Kuhn WR, Atreya SK (1979) Solar radiation incident on the Martian surface. J Mol Evol 14(1-3):57-64 Lowe DR, Tice MM (2004) Geologic evidence for Archean atmospheric and climatic evolution: Fluctuating levels of CO2, CH4, and O2 with an overriding tectonic control. Geology 32(6):493-6 Menendez C, Bauer Z, Huber H, Gad’on N, Stetter KO, Fuchs G (1999) Presence of acetyl coenzyme A (CoA) carboxylase and propionyl-CoA carboxylase in autotrophic Crenarchaeota and indication for operation of a 3-hydroxypropionate cycle in autotrophic carbon ixation. J Bacteriol 181(4):1088-1098

230

Raymond

Meyer O, Frunzke K, Gadkari D, Jacobitz S, Hugendieck I, Kraut M (1990) Utilization of carbon-monoxide by aerobes - recent advances. FEMS Microbiol Rev 87(3-4):253-260 Morowitz H (1999) A theory of biochemical organization, metabolic pathways, and evolution. Complexity 4(6):39-53 Muller V (2003) Energy conservation in acetogenic bacteria. Appl Environ Microbiol 69(11):6345-53 Navarro-Gonzalez R, McKay CP, Mvondo DN (2001) A possible nitrogen crisis for Archaean life due to reduced nitrogen ixation by lightning. Nature 412(6842):61-4 Nesbo CL, L’Haridon S, Stetter KO, Doolittle WF (2001) Phylogenetic analyses of two “archaeal” genes in thermotoga maritima reveal multiple transfers between archaea and bacteria. Mol Biol Evol 18(3):36275 Normand P, Gouy M, Cournoyer B, Simonet P (1992) Nucleotide sequence of nifD from Frankia alni strain ArI3: phylogenetic inferences. Mol Biol Evol 9(3):495-506 Orgel LE (2000) Self-organizing biochemical cycles. Proc Natl Acad Sci U S A 97(23):12503-12507 Pavlov AA, Kasting JF, Brown LL, Rages KA, Freedman R (2000) Greenhouse warming by CH4 in the atmosphere of early Earth. J Geophys Res 105(E5):11981-90 Pesole G, Bozzetti MP, Lanave C, Preparata G, Saccone C (1991) Glutamine synthetase gene evolution: a good molecular clock. Proc Natl Acad Sci U S A 88(2):522-6 Pesole G, Gissi C, Lanave C, Saccone C (1995) Glutamine synthetase gene evolution in bacteria. Mol Biol Evol 12(2):189-97 Pickett CJ, Vincent KA, Ibrahim SK, Gormal CA, Smith BE, Fairhurst SA, Best SP (2004) Synergic binding of carbon monoxide and cyanide to the FeMo cofactor of nitrogenase: relic chemistry of an ancient enzyme? Chemistry 10(19):4770-6 Ragsdale SW (2004) Life with carbon monoxide. Crit Rev Biochem Mol Biol 39(3):165-95 Ram R, VerBerkmoes N, Thelen MP, Tyson GW, Baker BJ, Blake RC, Shah M, Hettich R, Banield JF (2005) Community proteomics of a natural microbial bioilms. Science 308(5730):1915-20 Raven JA, Yin Z-H (1998) The past, present and future of nitrogenous compounds in the atmosphere, and their interactions with plants. New Phytology 139:205-219 Raymond J, Siefert JL, Staples CR, Blankenship RE (2004) The natural history of nitrogen ixation. Mol Biol Evol 21(3):541-54 Rees DC, Howard JB (2003) The interface between the biological and inorganic worlds: iron-sulfur metalloclusters. Science 300(5621):929-31 Russell MJ, Hall AJ, Cairnssmith AG, Braterman PS (1988) Submarine hot springs and the origin of life. Nature 336(6195):117-117 Rye R, Kuo PH, Holland HD (1995) Atmospheric carbon dioxide concentrations before 2.2. billion years ago. Nature 378:603-5 Schafer S, Barkowski C, Fuchs G (1986) Carbon assimilation by the autotrophic thermophilic archaebacterium Thermoproteus-Neutrophilus. Arch Microbiol 146(3):301-308 Schauder R, Widdel F, Fuchs G (1987) Carbon assimilation pathways in sulfate-reducing bacteria .2. Enzymes of a reductive citric-acid cycle in the autotrophic Desulfobacter-Hydrogenophilus. Arch Microbiol 148(3):218-225 Shiba H, Kawasumi T, Igarashi Y, Kodama T, Minoda Y (1985) The Co2 assimilation via the reductive tricarboxylic-acid cycle in an obligately autotrophic, aerobic hydrogen-oxidizing bacterium, Hydrogenobacter-Thermophilus. Arch Microbiol 141(3):198-203 Silver WS, Postgate JR (1973) Evolution of asymbiotic nitrogen ixation. J Theor Biol 40(1):1-10 Smith E, Morowitz HJ (2004) Universality in intermediary metabolism. Proc Natl Acad Sci U S A 101(36): 13168-73 Sprott GD, Ekiel I, Patel GB (1993) Metabolic pathways in Methanococcus-Jannaschii and other methanogenic bacteria. Appl Environ Microbiol 59(4):1092-1098 Suzuki A, Knaff DB (2005) Glutamate synthase: structural, mechanistic and regulatory properties, and role in the amino acid metabolism. Photosyn Res 83(2):191-217 Tanaka T, Nei M (1989) Positive Darwinian selection observed at the variable-region genes of immunoglobulins. Mol Biol Evol 6(5):447-59 Temple SJ, Vance CP, Gantt JS (1998) Glutamate synthase and nitrogen assimilation. Trends Plant Sci 3(2): 51-56 Turner SL, Young JP (2000) The glutamine synthetases of rhizobia: phylogenetics and evolutionary implications. Mol Biol Evol 17(2):309-19 Vanoni MA, Curti B (1999) Glutamate synthase: a complex iron-sulfur lavoprotein. Cell Mol Life Sci 55(4): 617-38 Wächtershäuser G (1988) Before enzymes and templates - theory of surface metabolism. Microbiol Rev 52(4): 452-484 Wächtershäuser G (1990) Evolution of the 1st metabolic cycles. Proc Natl Acad Sci U S A 87(1):200-204

Evolution of Biologic C & N CyclingȰGenomic Perspective

231

Wahlund TM, Tabita FR (1997) The reductive tricarboxylic acid cycle of carbon dioxide assimilation: Initial studies and puriication of ATP-citrate lyase from the green sulfur bacterium Chlorobium tepidum. J Bacteriol 179(15):4859-4867 Wingler A, Lea PJ, Quick WP, Leegood RC (2000) Photorespiration: metabolic pathways and their role in stress protection. Philos T Roy Soc B 355(1402):1517-1529

Reviews in Mineralogy & Geochemistry Vol. 59, pp. 233-258, 2005 Copyright © Mineralogical Society of America

10

Building the Biomarker Tree of Life Jochen J. Brocks Research School of Earth Sciences The Australian National University Canberra, ACT 0200, Australia [email protected]

Ann Pearson Department of Earth & Planetary Sciences Harvard University Cambridge, Massachusetts, 02138, U.S.A. [email protected]

INTRODUCTION It is the great challenge of geomicrobiology to study microorganisms in the context of their environments, both in Earth’s distant past and in the present. Planet Earth and its biosphere have evolved together, and a chronicle of Earth’s ecosystems and their geochemical cycles is recorded in sedimentary rocks spanning billions of years. A relatively new and very powerful approach to read these subtle microbial and environmental signatures in ancient rocks is the study of molecular fossils, or biomarkers, within the context of the biochemistry and phylogeny of their origins. Biomarkers are organic compounds (primarily lipids) that have particular biosynthetic origins and can be preserved in sediments and sedimentary rocks. The most valuable biomarkers are taxonomically speciic, i.e., they can be assigned to a deined group of organisms, and are resistant to degradation. Reading the biomarker signatures in rocks can give information about the ancient record of anoxic conditions in the water column (e.g., Summons and Powell 1986), the intensity of UV radiation penetrating lakes (Leavitt et al. 1997), hypersalinity in evaporitic environments (Grice et al. 1998), and the function of microbial communities at methane seeps (Hinrichs et al. 1999). Biomarkers have helped to reconstruct the irst appearance of major groups of organisms (e.g., McCaffrey et al. 1994; Moldowan et al. 1994; Moldowan and Talyzina 1998; Brocks et al. 2005), elucidate events of global climate change (e.g., Brassell et al. 1986), record major perturbations and reorganization of geochemical cycles (e.g., Logan et al. 1995; Kuypers et al. 1999) and document catastrophic losses in biodiversity (e.g., Grice et al. 2005). They are even used as tools to help in the discovery of major new petroleum reservoirs (for a review see Peters et al. 2004). The ield of biomarker research is young and many new applications wait to be discovered. This review will explain how biomarkers form, how they are extracted from sedimentary rocks, and how they are used to reconstruct ancient and modern microbial ecosystems. It ends with a look to the future of biomarker research and the on-going efforts to reconcile the biomarker record with the tree of life. The review includes examples of the detection of important metabolic pathways, of the appearance of new biomarkers (and by inference new taxonomic groups) in the rock record, and of the reconstruction of geochemical processes. We will concentrate on biomarkers from bacteria, archaea and unicellular eukaryotes, excluding 1529-6466/05/0059-0010$05.00

DOI: 10.2138/rmg.2005.59.10

234

Brocks & Pearson

the molecular fossils of plants and other non-microscopic eukaryotes. We highlight biomarker research using outstanding recent examples with a pedagogic emphasis on concepts but with no claim for completeness. More encyclopedic reviews were given by Peters et al. (2004) and Brocks and Summons (2004). For readers who desire more background in organic geochemistry, excellent introductions to the nomenclature, chemistry, and biology of lipids can be found in textbooks by Killops and Killops (2005) and Madigan and Martinko (2005).

The biomarker principle The origin of biomarkers. In lakes and oceans, the organic matter from dead organisms usually is almost quantitatively (> 99.9%) recycled back into carbon dioxide and water (Hedges and Keil 1995). The biological degradation of most proteins, nucleic acids and carbohydrates proceeds rapidly as dead biomass sinks through the water column, and it continues in the surface layers of the sediments. However, a small fraction of organic matter escapes the remineralization process and accumulates. Molecules that are especially recalcitrant, such as pigments, lipids and many structural macromolecules, will become concentrated (Tegelaar et al. 1989). With the onset of reducing conditions, the remaining sedimentary organic matter is degraded further by anaerobic heterotrophic organisms such as sulfate reducers, fermenters and methanogens (Megonigal et al. 2004); the chemical structure of the remains is altered by biological and chemical processes (Hedges and Keil 1995; Hedges et al. 1997; Rullkötter 1999). These alterations are referred to collectively as diagenesis. Smaller molecular units and degradation-resistant macromolecules are cross-linked and form kerogen, an amorphous and exceedingly complex structural network of biochemical subunits (e.g., Derenne et al. 1991; de Leeuw and Largeau 1993). During the formation of kerogen, vulcanization reactions mediated by sulfur and polysulphides often play an important role in connecting smaller molecular units, such as lipids, to the macromolecular aggregate, thus protecting them against further structural alterations (Sinninghe Damsté and de Leeuw 1990). Over millions of years, and with increasing burial depth and geothermal heat, most lipids will undergo structural rearrangement via cracking and isomerization reactions. These processes create a vast range of homologues and stereo- and structural isomers. Through reduction, elimination and aromatization the biomarkers typically lose all of their functional groups. The resultant products are geologically-stable hydrocarbon skeletons. Structures 1 and 2; 3 and 4; and 5 and 6 show examples of biolipids and their diagenetic hydrocarbon products. Bitumen is deined as the fraction of organic matter that can be extracted from sediments and sedimentary rocks using organic solvents, and it includes diagenetic components that have been thermally cracked, or released, from the kerogen. With increasing burial temperature and pressure, the thermal degradation of kerogen in organic-rich sedimentary rocks will generate enough liquid bitumen and natural gas for the expulsion of hydrocarbons in the form of petroleum. Petroleum reservoirs are, in fact, gigantic accumulations of biomarkers and other cracking products of sedimentary organic matter. However, at burial temperatures in the sedimentary unit exceeding 150–250 qC, most residual gas and liquid hydrocarbons have been expelled and the kerogen dehydrogenated to a highly aromatic, black carbon phase. This is the upper survival temperature for biological molecules over geologic time (Brocks and Summons 2004). The thermal destruction of biomarkers with deep burial is the primary complication in the search for biogenic molecular remains in very ancient, billion-year-old sedimentary rocks (Brocks et al. 2003a). During the experimental analysis of biomarkers, organic-rich, black sedimentary rocks are crushed to powder, and the powder is extracted with solvents such as methanol and dichloromethane using conventional relux extraction or automated solved extractors (ASE). The bitumen extracts are usually yellow to dark brown, highly complex mixtures, containing hundreds of thousands of compounds. To simplify further analyses, the bitumen is fractionated into saturated hydrocarbons, aromatics, and polar compounds (usually those containing the

Building the Biomarker Tree of Life

1

isorenieratene

2

isorenieratane

3

235

lanosterol HO

4

14

lanostane 4

OH OH

X

5

bacteriohopanepolyols (BHP)

Y 2 3

OH

Z

X, Y = H, OH Z = OH, OR, NHR

22 30

*

6

pentakishomohopane

R1 R2

35

28 2 3

a: R1 = H; R2 = H b: R1 = H; R2 = Me c: R1 = Me; R2 = H

heteroatoms O, N, and S) using normal-phase (SiO2-gel) chromatography. The fractions are then analyzed by gas chromatography-mass spectrometry (GC-MS). The ratios of different homologues and stereo- and structural isomers contained in a sample of bitumen may contain a plethora of information about conditions during burial and diagenesis, as well as the source organisms. For example, the relative abundance of stereoisomers such as 22S vs. 22R hopanes (see 6), 20S vs. 20R steranes (see 7), or triaromatic steroids 8 with intact and cleaved side chain, often helps to estimate relative burial temperatures (e.g., Brocks et al. 2003a). A high relative abundance of biomarkers such as aromatic carotenoids (e.g., 2), 28,30-bisnorhopanes (hopanes 6 lacking the C-28 and C-30 methyl groups) (Schoell et al. 1992), or the pentacyclic triterpane gammacerane 9, produced by ciliates grazing on bacteria in the anoxic zone (Sinninghe Damsté et al. 1995), may indicate anoxic and/or sulidic conditions. Molecular fossils as markers for biosynthetic pathways. For many geobiologists, the most interesting application of biomarkers is the reconstruction of ancient microbial ecosystems and the concurrent environmental conditions, at time-scales of millions to billions

Brocks & Pearson

236

R 20

*

7

24

steranes

R

8

triaromatic steroids

R = H, Me, Et 4

9

gammacerane

of years. Although lipids usually lose their functional groups during diagenesis, and often also stereochemical speciicity, the remaining hydrocarbon skeletons retain useful biological and ecological information. For instance, the aromatic carotenoid okenane 10, a molecule detected in 1,640 million-year-old sedimentary rocks of the McArthur Basin in northern Australia, is regarded as a biomarker for purple sulfur bacteria of the family Chromatiaceae (Brocks et al. 2005). The only known biological precursor of okenane is the red-colored phototrophic pigment okenone 11, and okenone is found exclusively in Chromatiaceae. As the Chromatiaceae have a very speciic ecology, the presence of okenane has been used to predict that the waters of the McArthur Basin 1,640 million years ago were sulidic up into the photic zone.

10

okenane

11

okenone

O

Me O

The distribution of carotenoids in the biosphere is comparatively well known due to their intense color. Therefore, the interpretation of okenane as a biomarker for purple sulfur bacteria appears to be robust. However, the interpretation of many other biomarkers is more complex. Strictly speaking, biomarkers are not markers for taxonomic groups or for environmental conditions. Biomarkers are the products of biosynthetic pathways that may occur in unrelated organisms. An example of potential misinterpretation is the carotenoid isorenieratane 2. Isorenieratane 2 is commonly interpreted as a speciic biomarker for green sulfur bacteria (Chlorobiaceae). However, the precursor of isorenieratane 2, isorenieratene 1, also occurs in the gram positive bacterial group Actinomycetales (Krügel et al. 1999; Phadwal 2005), and a genome library search for genes involved in isorenieratene biosynthesis suggests that cyanobacteria may have this capacity as well (Woodward Fischer, personal communication).

Building the Biomarker Tree of Life

237

In general, the distribution of genes for carotenoid biosynthesis in Bacteria is characterized by horizontal gene transfer and gene duplication events (Phadwal 2005). Thus the capacity for biosynthesis of isorenieratene 1 might appear in other lineages as well. Additionally, the aromatic end-group of isorenieratene also may form by abiotic, diagenetic aromatization of cyclohexyl (ionene) end-groups (Koopmans et al. 1996). Thus, isorenieratane may form in sedimentary environments by alteration of a wide range of carotenoids, including E-carotene. Therefore, isorenieratane 2 is not a biomarker solely for Chlorobiaceae. Isorenieratane is a product of all biosynthetic and abiotic pathways that can produce and alter suitable precursors. A second example of the ambiguities in the interpretation of biomarkers is the phylogenetic distribution of the biosynthetic pathway leading to sterols. The C30 steroid hydrocarbon lanostane 4 has two common diagenetic precursors, lanosterol (produced by animals and fungi) and cycloartenol (biosynthesized by plants). The enzyme oxidosqualene cyclase (OSC) catalyzes the formation of lanosterol 3 and cycloartenol from the precursor compound oxidosqualene 12. These compounds are the initial steroidal products that feed the long, complex biosynthetic pathways that produce all known eukaryotic steroids. However, although most biology textbooks state that steroid biosynthesis is one of the deining characteristics of eukaryotes, lanostane 4 is not always a biomarker for Eucarya. OSC also is expressed in at least three groups of bacteria: the Methylococcales (Bird et al. 1971), Myxococcales (Kohl et al. 1983), and Planctomycetales (Pearson et al. 2003). Several of these species produce detectable amounts of lanosterol, as well as down-stream (modiied) sterols. Therefore, the compound lanostane is a biomarker for eukaryotes and bacteria. However, fossil steroids that have an additional alkyl substituent at position C-24 (see 7) in the side chain must still be regarded as diagnostic for eukaryotes, as no group of bacteria is known (yet) to possess the biosynthetic capacity to alkylate the steroid side chain (Brocks et al. 2003b; Volkman 2003).

12

oxidosqualene O

The carbon isotopic composition of biomarkers Fractionation of the stable isotopes of carbon, 12C and 13C, occurs in association with biological reactions and remains imprinted in the isotopic signatures of biomarker molecules. In addition to the intrinsic taxonomic utility of certain lipids, the isotopic ratios of these compounds can provide insight about the environmental conditions and metabolic capabilities of the source organisms. As such, compound-speciic isotopic analysis is a valuable additional tool for understanding the modern and ancient geologic record. Carbon occurs naturally as three isotopes, 12C, 13C, and 14C; with fractional abundances of 0.989, 0.011, and 10−12, respectively. The last, 14C, is radioactive, and its half-life of 5730 years yields useful radiocarbon chronologies only over the most recent few tens of thousands of years. Therefore, most isotopic analyses of individual biomarkers focus primarily on the two stable isotopes of carbon. Isotopic composition is expressed most commonly as the ratio of 13C to 12C in the substance, relative to the ratio in a standard material. The units of isotopic fractionation are parts per thousand, or “permil” (‰), and the standard reference material is a carbonate rock (VPDB; Vienna Pee Dee Belemnite), which by deinition has a G13CVPDB value of 0‰.

Brocks & Pearson

238

⎛ 13 C 12 C Sample δ13C = ⎜ 13 12 ⎜ C C Standard ⎝

⎞ ⎟ k1

Building the Biomarker Tree of Life

239

When inorganic carbon is ixed into biomass, these reactions necessarily are controlled by the activity of speciic enzymes, each of which has an associated KIE. The biological transformation of substrate (A) to product (B) is accompanied by a fractionation factor that most commonly is written using the H notation, for the subsequent convenience of relating HA-B to GA and GB when H is a small number. Because 13C preferentially remains in A, HA-B always is a positive number and products are “lighter” than reactants. HA-B { (DA-B − 1)1000 HA-B ≈ GA − GB

The enzymes utilized by organisms to ix carbon are categorized most easily by the metabolic pathways in which they are used and by the species of inorganic carbon substrate for which they are speciic (see Bott and Thauer 1989; and Hayes 2001, for compilations of substrates and isotopic fractionations, respectively). The latter is important, as it is clear that organisms utilizing pathways speciic for HCO3− necessarily begin with a substrate that is heavier isotopically than organisms that ix CO2 directly. The recent discovery of novel metabolic pathways in some species of microbes (e.g., the newly discovered 3-HP pathway; Strauss and Fuchs 1993; Raymond, this volume)—and the possibility that mixotrophic or hybrid metabolisms are expressed by species in the environment—also complicates the picture considerably. However, the currently available information can be divided into pathways that typically yield isotopically “heavy” biomass (rTCA and 3-HP); pathways yielding isotopically intermediate biomass (Calvin-Benson cycle and some methanogens); and pathways yielding isotopically light biomass (homoacetogens, some methanogens, other Acetyl-CoA pathway organisms, and all methanotrophs). The above discussion does not include strict heterotrophs, because the carbon isotopic composition of their lipids and biomass broadly relects their carbon sources. The G13C values of biomass of heterotrophic microbes and macroscopic heterotrophs appear to be united by the principle “you are what you eat, plus 1‰” (DeNiro and Epstein 1978). This rule of thumb is based on the weak carbon isotopic fractionation associated with the glycolytic respiratory metabolism of most heterotrophs. Methanotrophy and methylotrophy, on the other hand, may be considered here as special cases of autotrophy rather than heterotrophy, as they involve ixation of C1 metabolites and thus are accompanied by large H values. For further discussion of fractionations associated with aerobic methanotrophy see Summons (1994); and for separate treatment of methylotrophic metabolism see Summons (1998). Anaerobic methanotrophy has not yet been explained mechanistically, although the irst in situ isotopic measurements (Hinrichs et al. 1999) immediately indicated large values of HCH4-biomass. Application of δ13C analyses at the molecular level. Isotopic analysis of lipid biomarker molecules can provide valuable information about the geobiology and biogeochemistry of contemporary and ancient systems. Compound-speciic measurements usually employ the “continuous-low,” isotope-ratio mass spectrometric methods developed by Hayes (Matthews and Hayes 1978; Hayes et al. 1990). Initial studies focused on lipids extracted from samples of ancient geologic age (e.g., Freeman et al. 1990) and showed the utility of this approach to describe a diversity of biomarkers and their origins. Such molecular-level analyses rely on the intrinsic advantages provided by lipids: their volatility permits separation by gas chromatography, and lipids are thermally and diagenetically stable.

An important, but complicating, factor in the interpretation of G13C values of individual lipids is the extent to which the lipids themselves are fractionated isotopically relative to the total biomass of the source. Intracellular fractionations are the consequence of the diversion of metabolic intermediates such as pyruvate and acetate, which are necessary for the biosynthesis of lipids, into other pathways such as the citric acid cycle and/or for the biosynthesis of amino

240

Brocks & Pearson

acids. A detailed mathematical treatment of the isotopic consequences of such branched pathways is given in Hayes (2001), but in general, the fractionation between biomass and lipids (Hbiomass-lipid) results in a more negative value of G13C for the lipid due to fractionation during the decarboxylation of pyruvate to acetate. This is true for organisms expressing the Calvin cycle, but it is not necessarily true for species that synthesize acetate directly or for those that have alternative biosynthetic routes to acetate (e.g., van der Meer et al. 2001) or isoprenoids (Schouten et al. 1998a). Organisms that have unusual metabolisms have not yet been studied thoroughly, and the range of variability in expression of Hbiomass-lipid needs further exploration.

BIOMARKERS IN GEOMICROBIOLOGY There are far too many examples of the informative use of biomarkers in geobiology to review all of them in this chapter. Therefore, in the section that follows we present a series of case studies and outstanding examples of biomarker geochemistry. These examples are grouped by metabolic pathway, biogeochemical process, and/or by the phylogeny of the source organism. Each is an attempt to illustrate the unique contributions that the analysis of lipid biomarkers can make to the ield of geomicrobiology.

Biomarkers as indicators of metabolism Methanotrophic methanogens: the anaerobic oxidation of methane. The need for an anaerobic sink for the methane produced in marine sediments was recognized by geochemists as early as the 1970s-1980s (e.g., Reeburgh 1976; Alperin and Reeburgh 1985). Proiles of dissolved CH4 in sedimentary pore waters indicated that in most environments it did not reach the sediment-water interface and therefore was not oxidized by O2 in the deep ocean. Zehnder and Brock (1979) irst proposed—and later Hoehler et al. (1994) expanded upon—the hypothesis that the anaerobic oxidation of methane (AOM) could be achieved by methanogenic archaea utilizing a specialized metabolism designed to run “in reverse.” They suggested that the energetic expense of this reaction could be overcome by consumption of the H2 by-product or other reducing equivalents by syntrophic, sulfate-reducing bacteria (SRB). Such a process would require a close physical association in the form of an aggregate or consortium. However, it was only recently that the microorganisms were identiied that putatively are involved in this process. Currently they are classiied broadly among three new categories of Euryarchaeota, ANME-1, ANME-2, and ANME-3 (Boetius et al. 2000; Orphan et al. 2001, 2002; Knittel et al. 2005). At least one of these groups (ANME-2) indeed appears to live in consortia with SRB. Much still remains to be learned about the biology of AOM, including the possibility that there are ANME-group archaea or SRB that are capable of oxidizing CH4 independently, without a syntrophic partner. Prior to the discovery of the consortia that mediate AOM, however, there was evidence from G13C values of biomarker lipids that archaea were involved in AOM. The history behind the discovery of this process represents an outstanding case in which isotopic analysis of biomarkers led directly to conclusions about geomicrobiology and environmental metabolisms. Bian (1994) irst observed (but could not yet explain) depleted values of G13C for crocetane 13, a C20 isoprenoid isomer of phytane, in sediments of the Kattegat Strait between Denmark and Sweden. Subsequent work by Hinrichs et al. (1999) on sediments from coastal California, USA, showed similarly extreme isotopic depletions for both archaeal and bacterial biomarkers from a CH4-rich, anaerobic sediment. Hinrichs et al. (1999) showed that the lipids archaeol 14 and sn-2-hydroxyarchaeol, which are glycerol diethers typical of methanogenic euryarchaeota, had G13C values ≤ −100‰. Such light values indicated that consumption of isotopically-depleted CH4 must serve as the primary carbon source for the archaea from which the lipids were derived. The earlier G13C values for crocetane 13 and the related isoprenoid

Building the Biomarker Tree of Life

241

hydrocarbon 2,6,10,15,19-pentamethylicosane (PMI), could inally be explained as products of similar CH4-consuming archaea (Bian et al. 2001). tail-to-tail

13

crocetane

O

14

archaeol O OH

Recent work has suggested that crocetane and C20 isoprenoids derived from archaeal diethers may be biomarkers speciic for the archaea associated with ANME-2-type consortia. The ANME-1 archaea (frequently observed as individual cells and ilaments; Orphan et al. 2002) may be typiied by the C40 isoprenoids associated with tetraethers 15 (Blumenberg et al. 2004). There may be further taxonomic potential to be discovered within archaeal biomarkers, and more may be learned if ANME archaea are brought into pure or enrichment culture. head-to-head

HO

15

Crenarchaeol

O O

O O HO

After the early reports of the assimilation of CH4 by relatives of the normally methanogenic archaea, numerous other cases of AOM were documented using lipid biomarkers. Isotopicallydepleted isoprenoid ethers have been found in association with Mediterranean mud volcanoes (Pancost et al. 2000); cold-temperature CH4 seeps (Pancost et al. 2001; Zhang et al. 2003); hot-temperature CH4 seeps (Teske et al. 2002; Schouten et al. 2003b); in the Black Sea (Thiel et al. 2001; Michaelis et al. 2002; Wakeham et al. 2003); and in the spectacular carbonate chimneys of the serpentinite-hosted hydrothermal system of the Lost City (Kelley et al. 2005). There is also considerable evidence for the association of bacterial groups with the AOM process. These microbes are probably involved in the metabolism of the organic end-products of AOM and/or in the incorporation of 13C-depleted DIC. However, it remains possible that some unidentiied groups of bacteria assimilate CH4 directly. Hopanoids (Elvert et al. 2000; Pancost et al. 2000; Thiel et al. 2001, 2003) and fatty acids derived from phospholipids (e.g., Hinrichs et al. 2000) with highly depleted G13C signatures are frequently found in the same samples as the archaeal biomarkers for AOM. Anammox: the discovery of autotrophic denitrification. In a now-classic paper entitled “2 kinds of lithotrophs missing in nature”, Broda (1977) postulated the anaerobic oxidation of ammonia by nitrate or nitrite-reducing bacteria. Oxidation of NH4+ with subsequent reduction

Brocks & Pearson

242

of NO2− would represent both a means to provide energy for lithoautotrophic ixation of CO2 to biomass and a form of denitriication with associated effects on the global nitrogen cycle. NH4+ + NO2− o N2 + 2H2O

'G° = −360 kJ/mol

The elegance and simplicity of Broda (1977) make the paper virtually required reading for the student of geobiology; however, the predicted “anammox” reaction and more speciically the associated organisms remained undiscovered for a further 20 years. Anammox was inally revealed in association with nitrogen-rich wastewater reactors (Mulder et al. 1995; Strous et al. 1997; van de Graaf et al. 1997), and the organisms responsible for the reaction were identiied as members of an unusual bacterial group, the Planctomycetales (Strous et al. 1999). Recently it has been suggested that anammox may be responsible for up to 30–50% of the denitriication occurring in the global ocean (Dalsgaard et al. 2003, 2005; Kuypers et al. 2005), all of which had been previously assigned to the activity of denitrifying, anaerobic heterotrophs. Not all members of the Planctomycetales are capable of anammox metabolism; most members of this group appear to be heterotrophs or chemoorganotrophs. However, it is within the anammox genera (Brocadia, Kuenenia, Scalindua) that the unique lipids known as ladderanes 16 are found (Sinninghe Damsté et al. 2002). Ladderanes are believed to be critical components of the anammox organelle, the “anammoxosome” (van Niftrik et al. 2004). Speciically, ladderanes may provide a diffusional barrier to the toxic intermediate of ammonium oxidation, hydrazine (N2H4). For this purpose, ladderanes possess a unique ‘ladder’ of concatenated cyclobutane rings (see 16) that can be stacked into a dense membrane; and due to a lack of branched methyl groups, they are apparently biosynthesized from acetate rather than from isoprene. The C20 moieties may be connected via ether or ester linkages to a glycerol backbone (Sinninghe Damsté et al. 2002).

16

OMe

[5]-ladderane O

It is unknown at present to what extent intact ladderanes, or more likely, their degradation products, are preserved in the geologic record. Because of the highly strained nature of cyclobutane rings, these high-energy structures should be unstable and prone to rapid degradation. However, the ring-opening products of the cyclobutane groups may yield diagnostic and geologically stable biomarker products. Searching for these products, it may eventually be feasible to determine how far back into the geologic record the anammox process persists. Simple consideration of biogeochemical cycles suggests that anammox would not have developed as a signiicant process until after the advent of oxygenic photosynthesis, since the reaction is dependent on suficient quantities of NO2−. Nitrite is a product of the aerobic oxidation of NH4+ by O2 (nitrosiication, by genera such as Nitrosomonas or Nitrosobacter).

The limited G13C data that are currently available for ladderane lipids are consistent with an autotrophic metabolism for the anammox bacteria. Schouten et al. (2004) observed G13C values between −55‰ and −58‰ for samples of ladderanes from the water column of the Black Sea. The isotopic fractionation of lipid relative to substrate (CO2) was 32‰ to 49‰ (lipids depleted relative to substrate) for the samples from the Black Sea and for samples taken from laboratory enrichment cultures. These data could be consistent with a metabolic pathway dependent on autotrophic synthesis of Acetyl-CoA. Such metabolism is also broadly consistent with an ancient origin of these species and of this chemoautotrophic pathway. Ladderanes are not the only unusual lipids to be found within the Planctomycetales. Species of Pirellula and Planctomyces (Kerger et al. 1988) contained abundant C18:1Z9 fatty

Building the Biomarker Tree of Life

243

acid 17, more commonly a component of eukaryotes (in the above nomenclature for fatty acids, C18 refers to the total number of carbon atoms in the molecule, ‘:1’ to the number of C=C double bonds in the chain, and ‘Z9’ designates the position of the double bond counted from the last (Z omega) carbon atom in the chain). They also contained unique distributions of 3-hydroxy-fatty acids, which were suggested to be suficiently diagnostic biomarkers for planctomycetes in some environmental settings. Unlike most bacteria, the Planctomycetales lack peptidoglycan in their cell walls and as such contain no muramic acid (e.g., König et al. 1984; Stackebrandt et al. 1986). Although many planctomycetes, including the anammox genera, contain “nucleoids” which conine the cellular DNA, Gemmata obscuriglobus is the only species known to contain a “nucleoid” which is double membrane-bound as found in eukaryotes (Fuerst and Webb 1991; Lindsay et al. 2001). G. obscuriglobus also is among the few bacterial species that biosynthesize sterols, and it is the only species (prokaryotic or eukaryotic) in which those sterols are not subsequently demethylated at position C14 (see lanostane 4) (Pearson et al. 2003). It remains unknown what intracellular roles all of the unusual planctomycete lipids serve, or how far back in the geologic record the molecular fossils of this group may extend. O

17

C18:1ω9 fatty acid

ω2 18

ω9

2

1

OH

The Planctomycetales represent an unusual and divergent microbial lineage which traditionally has been dificult to place within the tree of life. Using approaches based on alternative phylogenetic treeing methods for 16S rRNA genes (Brochier and Philippe 2002) and on the relationships among C1 metabolic pathway genes (Chistoserdova et al. 2004), it has been suggested that the Planctomycetales may be the most deeply branching of the bacterial taxa. By these analyses, the planctomycetes are basal to the rest of the bacteria and are the closest bacterial relatives to the archaea and eukaryotes. Aromatic carotenoids, biomarkers for phototrophic oxidation of sulfide. Anoxic conditions in aquatic systems, and particularly the enigmatic Oceanic Anoxic Events such as OAE1b in the mid-Cretaceous, have become critical research areas for the geosciences. Periods of prevailing anoxia in large basins might be responsible for the widespread deposition of black shales, increased accumulation of petroleum source rocks, changes in global biogeochemical cycles, extreme shifts in climate, major mass extinctions, and concomitant biological radiations. Currently the only biomarker proxies available to study the most extreme form of anoxia, photic zone euxinia, are biomarkers of the phototrophic green and purple sulfur bacteria (Summons and Powell 1986; Requejo et al. 1992; Brocks et al. 2005; Grice et al. 2005). Green sulfur bacteria (family Chlorobiaceae) are only distantly related to other phototrophic bacterial groups (Fig. 1). Chlorobiaceae only use photosystem I (PS I) and are strictly anaerobic, exploiting reduced sulfur species, such as hydrogen sulphide, as electron sources. In microbial mats, the requirement for sulide and light restricts their habitat to the anoxic zone millimeters below the mat surface. In planktonic environments they live in a layer below the anoxic-oxic boundary, but within the photic zone. To adjust to the wavelength distribution and attenuated intensity of light at depth, Chlorobiaceae commonly possess an abundance of accessory carotenoid pigments. Green-pigmented species of planktonic Chlorobiaceae grow in a thin layer at water depths up to ~13 m, and their major carotenoid pigments are chlorobactene 18 and hydroxychlorobactene (Imhoff 1995), both of which yield the sedimentary biomarker, chlorobactane 19. Brown-pigmented Chlorobiaceae inhabit a zone deeper than the green-pigmented species; they are usually found at water depths up to 18 m,

Brocks & Pearson

244

but were also observed in the Black Sea at 80 m (Repeta et al. 1989). Their pigment system is dominated by the carotenoids isorenieratene 1 and E-isorenieratene, the precursors for the hydrocarbon biomarkers isorenieratane 2 and E-isorenieratane (Liaaen-Jensen 1965).

18

chlorobactene

19

chlorobactane

Purple sulfur bacteria of the family Chromatiaceae are a sub-group of the J-proteobacteria and utilize the PS II photosynthetic reaction center. Although unrelated to Chlorobiaceae, the Chromatiaceae have similar environmental requirements. Their preferred physiology is phototrophic oxidation of reduced sulfur under anoxic conditions. However, they are generally more tolerant to oxygen and can exploit a more versatile range of electron donors, including hydrogen. In microbial mats, as well as in planktonic environments, they grow in a thin layer directly above the zone of green sulfur bacteria but below the oxycline. Several species of planktonic Chromatiaceae have an accessory pigment system based on the red-

Eucarya 8, 4 Animals Flagellates Plants

Microsporidia

Ciliates

9 Fungi

Slime moulds

Bacteria

Diplomonads

Green non-sulfur bacteria

55

Crenarchaeota

15

11, 6b

Proteobacteria

Desulfurococcus Thermotoga

Thermoproteus

Pyrodictium

Pyrobaculum

Nitrospira

6c

Archaea

Sulfolobus

Gram positive bacteria

Cyanobacteria

Methanobacterium

Pyrococcus

Thermoplasma

2, 19

Green sulfur bacteria

Archaeoglobus Thermodesulfobacterium

Methanopyrus

Halobacterium

Methanococcus

Euryarchaeota

Aquifex

Methanosarcina

13, 14

Figure 1. SSU rRNA phylogenetic tree annotated with structure numbers of biomarkers discussed in the text (adapted after Brocks and Summons 2004).

Building the Biomarker Tree of Life

245

colored monoaromatic carotenoid okenone 10, which has a 2,3,4-trimethylaryl end-group. Planktonic species are commonly observed at water depths around 10 meters or less, and very rarely deeper than 20 meters (Van Gemerden and Mas 1995). Okenone 11 is the only known precursor for the hydrocarbon biomarker okenane 10, which represents the sole known proxy for purple sulfur bacteria in the fossil record. Biomarkers of green and purple sulfur bacteria are particularly valuable tracers for the study of marine anoxic conditions in the Precambrian (> 542 million years (Ma) ago), as body fossils of animals that give evidence about anoxic conditions in younger rocks do not exist in this period.

Biomarkers as indicators of the evolution of life and the environment Phototrophic sulfur bacteria and a sulfidic ocean in the Proterozoic. Today, Earth’s oceans are teeming with complex life, and even deep marine trenches contain enough oxygen to support macroscopic organisms. However, oceans in the distant past were fundamentally different. For the irst two billion years of its existence, the ocean-atmosphere system was almost entirely anoxic (Fig. 2) (Holland 1994), but around 2450 to 2320 Ma ago, the disappearance of mass-independent fractionation of sulfur isotopes indicates that the concentration of atmospheric oxygen rose from previously trace levels to at least 10−5u the present level (Farquhar et al. 2000; Bekker et al. 2004; Holland 2004). Soon after this initial rise of oxygen, fossil soils (paleosols) begin to show typical oxic weathering patterns that suggest atmospheric O2 quickly may have reached 15% of its present value (Rye and Holland 1998). However, the deep oceans remained mostly or entirely anoxic until at least ~1,800 Ma ago, the point in geological history when the last Paleoproterozoic banded iron formations (BIFs; iron silicates and iron carbonates) disappeared (Fig. 2). Surprisingly, the state of the ocean in the following “mid-Proterozoic” interval (~1,800 to ~800 Ma) remains particularly mysterious. One model suggests that deposition of BIFs ceased ~1,800 Ma ago because FeII emitted from mid-oceanic ridges was precipitated immediately on the oxygenated sea-loor as FeIII-hydroxides (Holland 1994). However, according to a competing model (Canield 1998), FeII was not removed as oxidized rust but precipitated as FeII-sulides in sulidic ocean waters. Evidence is accumulating from the isotopic composition and distribution of sulides (Canield 1998; Shen et al. 2003; Poulton et al. 2004), sulfates (Kah et al. 2001) and molybdenum (Arnold et al. 2004), that appears to support Canield’s model. It is also possible that a hybrid of both models existed, resembling a ‘marble cake’ ocean (A. H. Knoll, personal communication). If large areas of the world ocean were euxinic (anoxic and sulidic) in the midProterozoic, then our understanding of more than one ifth of Earth history would change

(c) (a)

(f) ‘mid-Proterozoic’ (e)

Paleo-

Archean 4.5

(d)

(b)

Meso-

Neo-

Proterozoic 2.5

1.6

(g)

1.0

P

M

C

Phanerozoic

0.54

Ga

Figure 2. Geological time chart beginning with (a) the formation of Earth ~4.6 billion years ago (Ga); (b) anoxic, non-sulidic oceans; (c) onset of oxygenation of the atmosphere (Bekker et al. 2004; Holland 2004); (d) disappearance of banded iron formations as indicator of changing ocean chemistry (Holland 2004); (e) the informally deined “mid-Proterozoic” interval with possible widespread, anoxic and sulidic marine conditions (Canield 1998); (f) major radiation of eukaryotic algae (Knoll 1992); (g) irst appearance and major radiation of multicellular organisms and animals. 0.54 Ga marks the PrecambrianCambrian boundary. P = Paleozoic, M = Mesozoic, C = Cenozoic.

246

Brocks & Pearson

radically. Geochemical cycles would have been altered, and many bioessential elements, such as nitrogen, molybdenum, and copper would have been rare. Trace-metal limitation may explain why familiar forms of life, such as modern algae and animals, arose so late in Earth history (Anbar and Knoll 2002). In the euxinic ocean, organisms requiring oxygen would have been marginalized, restricted to surface waters and shorelines. The ocean would have promoted extensive growth of green and purple sulfur bacteria wherever sulidic conditions rose into the photic zone. The earliest evidence for the existence of phototrophic sulfur bacteria comes from a biomarker study on the 1,640 Ma Barney Creek Formation in the McArthur Basin, northern Australia (Brocks et al. 2005). Well preserved, organicrich dolostones of the Barney Creek Formation were deposited in deep waters of the rift basin. The lipids extracted from these sedimentary rocks contain some of the oldest, clearly indigenous biomarkers known to date (Summons et al. 1988b). Signiicantly, the samples contain relatively high concentrations of isorenieratane 2, chlorobactane 19, and okenane 10. This indicates that the basin was stratiied, and euxinic conditions extended—at least episodically—into the photic zone of the water column. This ancient assemblage of biomarkers had two further characteristics which were radically different from any younger bitumen. Steroids alkylated at C-24 in the side chain and diagnostic for eukaryotic organisms were present at levels close to or below detection limits. In contrast, aromatic steroids without side chain alkylation but which were methylated at C-4 (see 8) were very abundant. These biomarkers, together with high relative concentrations of 3E-methyl-hopanes, suggest aerobic type-I methanotrophic bacteria were abundant members of the population. Aerobic methanotrophs are typically found in sulfate-starved environments (