Evolutionary formation of new protein folds is linked to metallic cofactor recruitment

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/26707494 Evolutionary formation of new protein folds is linked to metallic cofactor recruitment Article in BioEssays · September 2009 DOI: 10.1002/bies.200800201 · Source: PubMed CITATIONS READS 10 36 4 authors, including: Ying-Ying Jiang Hong-Yu Zhang 19 PUBLICATIONS 201 CITATIONS 164 PUBLICATIONS 3,081 CITATIONS Universität Heidelberg SEE PROFILE Huazhong Agricultural University SEE PROFILE All content following this page was uploaded by Ying-Ying Jiang on 16 April 2015. The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document and are linked to publications on ResearchGate, letting you access and read them immediately. Hypothesis DOI 10.1002/bies.200800201 Evolutionary formation of new protein folds is linked to metallic cofactor recruitment Hong-Fang Ji,1 Lei Chen,1 Ying-Ying Jiang,1,2 and Hong-Yu Zhang2* 1 Shandong Provincial Research Center for Bioinformatic Engineering and Technique, Center for Advanced Study, Shandong University of Technology, Zibo 255049, P. R. China 2 Institute of Bioinformatics and State Key Laboratory of Crop Genetic Improvement, College of Life Science and Technology, Huazhong Agricultural University, Wuhan 430070, P. R. China To explore whether the generation of new protein folds could be linked to metallic cofactor recruitment, we identified the oldest examples of folds for manganese, iron, zinc, and copper proteins by analyzing their fold-domain mapping patterns. We discovered that the generation of these folds was tightly coupled to corresponding metals. We found that the emerging order for these folds, i.e., manganese and iron protein folds appeared earlier than zinc and copper counterparts, coincides with the putative bioavailability of the corresponding metals in the ancient anoxic ocean. Therefore, we conclude that metallic cofactors, like organic cofactors, play an evolutionary role in the formation of new protein folds. This link could be explained by the emergence of protein structures with novel folds that could fulfill the new protein functions introduced by the metallic cofactors. These findings not only have important implications for understanding the evolutionary mechanisms of protein architectures, but also provide a further interpretation for the evolutionary story of superoxide dismutases. Keywords: evolution; geochemistry; metal bioavailability; metallic cofactor; protein fold Introduction In the generation of life, proteins not only provide a structural scaffold but also perform biological processes, so that it is of great interest and significance to explore how the protein space expanded. In particular, as the vast number of peptide sequences have to fit the very limited topological patterns of secondary structures (also termed folds),(1–3) protein folding has been very conserved during evolution, which implies that the generation of new folds is a very rare event in the long history of biological evolution.(4) Therefore, it is challenging to reveal the mechanisms underlying the expansion of the protein fold universe.(4,5) *Correspondence to: H.-Y. Zhang, Institute of Bioinformatics and State Key Laboratory of Crop Genetic Improvement, College of Life Science and Technology, Huazhong Agricultural University, Wuhan 430070, P. R. China. E-mail: zhy630@mail.hzau.edu.cn BioEssays 9999:1–6, ß 2009 Wiley Periodicals, Inc. Recently, through the analysis of the ligand-protein mapping patterns, we found that a correlation exists between the early history of protein-cofactor binding and protein fold formation,(6) which suggested that the generation of some primitive protein folds [such as P-loop containing nucleoside triphosphate hydrolases (c.37 fold), TIM beta/alpha-barrel (c.1 fold), and NAD(P)-binding Rossmann-fold domains (c.2 fold)] could be linked to recruitment of organic cofactors, i.e., ATP and NAD/P(H).(6) This link could be explained in terms of the functional importance of these cofactors to primordial proteins and the protein structural requirements to bind the cofactors.(5,6) It is a natural step to speculate that if cofactor recruitment played a role in the formation of new protein folds, this phenomenon should be observed not only for organic cofactors, but also for metallic counterparts. Thus, it is of great interest to explore whether some very frequently used metallic cofactors, such as manganese, iron, zinc, and copper ions, can be linked to the evolutionary generation of new protein folds. Generation of a cupredoxin-like fold could be linked to copper bioavailability A recent analysis on fold distribution patterns in domain space of copper proteins revealed that the most ancient copper protein fold is a cupredoxin-like fold (b.6, which belongs to the all beta protein class).(7) In the evolutionary order of protein architectures established by Caetano-Anollés and co-workers (http://www.manet.uiuc.edu/download/foldAncestryVal2_ 0.txt),(8–11) the b.6 fold is no. 164 in the 776 folds. As most of the proteins containing this fold are cupredoxins, this finding is consistent with the opinion proposed by Rydén 20 years ago that a primordial single-domain cupredoxin is the progenitor of multidomain copper oxidases.(12) The cupredoxin-like fold is found in six annotated families, four of which are copper proteins and the others non-copper proteins (Table 1). As the two non-copper protein families are exclusively found in mammals, it is reasonable to infer that the initial generation of b.6 fold was tightly coupled with copper ions. In fact, no protein 1 Hypothesis H.-F. Ji et al. Table 1. Family characteristics of the most ancient folds for some metalloproteins Metalloproteins Most ancient folds Family numbersa (metal þ non-metal) Copper proteins Manganese proteins Heme iron proteins Non-heme iron proteins Zinc proteins Cupredoxin-like (b.6) Nucleotide-diphospho-sugar transferases (c.68) Globin-like (a.1) Ferredoxin-like (d.58) Zincin-like (d.92) 6 (4 þ 2)b 14 (14 þ 0) 5 (4 þ 1)b 111 (7 þ 104) 16 (16 þ 0) a Families without comments in Swiss-Prot/TrEMBL are excluded. Non-metal protein families are exclusive to mammals. b architectures prior to no. 164 are dominated by copper cofactors. The question is thus raised: why did the newly evolving copper proteins need a new fold? Considering the functional selection of protein structures (or co-evolution between protein structure and function),(5,13) a likely explanation is that some new enzymatic functions introduced by copper ions could not be fulfilled by proteins containing the old protein folds; therefore, a new architecture was needed. In fact, copper enzymes have some very special properties. For instance, copper enzymes require dioxygen as an oxidant (functioning as oxidases),(14) and thus copper is not essential in strict anaerobes.(15) According to the records in the Molecular ancestry network (MANET) database,(11) the most ancient enzyme with the cupredoxin-like fold (b.6) is cytochrome C oxidase (EC 1.9.3.1), which is a component of the respiratory chain that catalyzes the reduction of oxygen to water (Table 2). Analysis of the Swiss-Prot database(16) revealed that this function is not shared by all of the enzymes that use the folds prior to b.6 (no. 164). Hence, it seems that the formation of the b.6 fold likely results from the structural requirements to fulfill the novel (oxygen-associated) enzymatic functions introduced by the copper cofactor. This explanation is supported by our recent finding that the fold usage of enzymes involved in aerobic metabolism differs greatly from that in anaerobic metabolism (unpublished results). The close relationship between copper enzymes and oxygen could be explained not only by the strong redox potentials of copper ions(14) but also from the viewpoint of geochemistry. According to the widely accepted geochemical model that describes the variations of element abundances through time,(14,17–21) the bioavailability of copper in the ancient anoxic world was extremely low (because of the extreme insolubility of copper sulfides) until the sudden rise of oxygen levels (when soluble copper sulfates came into being). Although the exact time of the rise in oxygen is contentious,(22–25) this event was correlated with the rise of cyanobacteria, and thus postdates the emergence of prokaryotes.(25) This suggests that copper proteins have a rather late origin in the evolutionary history of life (in fact, the b.6 fold is the latest architecture in the common redox proteins; see below for details). To test this inference, we examined whether the ancient enzymes participating in anoxic reactions preferred to use the folds prior to no. 164. A set of relatively early proteins was recently derived from a yeast proteome.(26) The oldest age group contains proteins that can be traced back to eubacterial genomes, and consists of 1806 members [according to Saccharomyces Genome Database (SGD), Oct 18, 2006].(27) Among these proteins, 463 enzymes were identified as specific to anoxic reactions, and these contain 682 domains. According to the annotation of SCOP 1.73,(28) most (87%) of these domains use the folds prior to no. 164. Taken together, accumulated evidence suggests a selfconsistent scenario for cupredoxin-like (b.6) fold formation. Following the sudden rise in oxygen levels several billion years ago, copper was released from its sulfides and became bioavailable. It then began to participate in the enzymatic reactions as a cofactor. As the protein architectures were not Table 2. Functions for the most ancient enzymes covered by the initial folds of some metalloproteins Metalloproteins Most ancient functionsa Copper proteins Manganese proteins Catalyzing the reduction of oxygen to water in the respiratory chain (EC 1.9.3.1 for b.6 fold)b Catalyzing the formation of dTDP-glucose, from dTTP and glucose 1-phosphate, as well as its pyrophosphorolysis (EC 2.7.7.24 for c.68 fold)b Catalyzing fumarate and succinate interconversion (EC 1.3.99.1 for a.1 fold)b For fumarate respiration using formate or sulfide as electron donor (EC 3.2.1.52 for d.92 fold)b Heme iron proteins Zinc proteins a According to the records in MANET database. Using the Swiss-Prot database, it was found that these functions are not shared by all of the enzymes that use the folds prior to that shown in the parentheses. b 2 BioEssays 9999:1–6, ß 2009 Wiley Periodicals, Inc. H.-F. Ji et al. suitable to fulfill the new functions introduced by this new cofactor, a novel fold (b.6) evolved to exploit it. The conclusion is thus that copper ions are strongly linked to the formation of b.6 fold. Generation of primitive manganese, iron and zinc protein folds could be linked to metal bioavailability To further evaluate the relationship between evolutionary generation of protein architectures and the metallic cofactor recruitment, it is necessary to examine the origin of other metalloprotein architectures, such as manganese, iron, and zinc protein folds. Using SCOP 1.73,(28) 253 types of manganese protein domains, 149 types of heme iron protein domains, 282 types of non-heme iron protein domains and 731 types of zinc protein domains were collected (only domains binding the metal were selected and the domains containing two metals were counted for each metal), which belong to 97, 38, 94, and 172 folds, respectively (Tables S1– S4). As shown in Fig. 1, the fold-domain mapping patterns for these proteins follow a power law, i.e., the number of folds ( F) Hypothesis decays with the increase of the number of domains (D) covered by the fold. The power law distribution of protein folds in domain space has been explained in terms of the preferential attachment principle,(29) which implies that the more prevalent the folds, the earlier they originated. To examine whether this explanation is still valid for manganese, iron, and zinc protein architectures, we assigned the ages to these protein folds by the following methods. If an architecture is shared by prokaryotic/archaebacterial and eukaryotic species, it is an early fold; if an architecture is only owned by eukaryotic species, it is a late fold. During this process, both the proteins recorded in PDB, and the corresponding homologous proteins in Swiss-Prot database were examined,(16) in an attempt to avoid the limitation of protein information recorded in PDB. As shown in Tables 3 and 4, on average the early folds cover many more domains than the late counterparts ( p < 0.0001) and most of the super folds (covering three or more domains) originated early, while most of the ordinary folds (covering one or two domains) appeared late. Therefore, it can be concluded that the preferential attachment principle still underlies the power law behavior of fold-domain mapping for these metalloproteins, implying that the most widely Figure 1. The number of folds (N) decays with the increase of the number of domains covered by the fold (D) and follows power law equation N ¼ aD
b ( p < 0.001). a: Manganese proteins, b: heme-iron proteins, c: non-heme iron proteins, d: zinc proteins. BioEssays 9999:1–6, ß 2009 Wiley Periodicals, Inc. 3 Hypothesis H.-F. Ji et al. Table 3. Numbers of early and late folds for some metalloproteins and average numbers of domains covered by the folds Metalloproteins Early foldsa (domains) Late foldsb (domains) Manganese proteins Heme iron proteins Non-heme iron proteins Zinc proteins 37 19 37 74 42 16 33 71 (4.8) (6.3) (5.3) (7.4) (1.3) (1.6) (1.3) (1.9) a Shared by prokaryotic/archaebacterial and eukaryotic species. Only found in eukaryotic species. b shared architectures represent the most ancient folds for corresponding metalloproteins. The most ancient fold for manganese is the nucleotide-diphospho-sugar transferases fold [c.68, which belongs to alpha and beta (a/b) protein class]; for heme-iron, the globin-like fold (a.1, which belongs to all alpha protein class); for non-heme iron, the ferredoxinlike fold [d.58, which belongs to alpha and beta (a þ b) protein class]; and for zinc proteins, the zincin-like fold [d.92, which belongs to alpha and beta (a þ b) protein class] (Table 1). To explore the correlation between these primitive metalloprotein architectures and the corresponding metals, we examined the family characteristics for these folds. As shown in Table 1, except for the families of non-heme iron proteins, the families containing these folds are dominated by corresponding metalloproteins, which strongly suggests that c.68, a.1, and d.92 folds have evolved for binding manganese, heme iron and zinc, respectively. According to the evolutionary sequence of protein architectures proposed by Caetano-Anollés et al. (http:// www.manet.uiuc.edu/download/foldAncestryVal2_0.txt),(8–11) c.68 fold is no. 14, a.1 is no. 74 and d.92 is no. 105. It is interesting to note that these folds are indeed the earliest protein architectures that are dominated by corresponding metals. More interestingly, the emerging order for these folds, i.e., manganese and iron protein folds appeared earlier than zinc and copper counterparts, is in excellent agreement with the putative bioavailability of these metals in the ancient anoxic ocean, i.e., manganese and iron ions were relatively Table 4. Numbers of super and ordinary folds for some metalloproteins and corresponding early and late folds Metalloproteins Super foldsa (early foldsb) Manganese proteins Heme iron proteins Non-heme iron proteins Zinc proteins 31 14 34 72 a (29) (12) (29) (62) Ordinary foldsc (late foldsd) 66 (40) 24 (13) 60 (32) 100 (65) Covering three or more domains. Shared by prokaryotic/archaebacterial and eukaryotic species. c Covering one or two domains. d Only found in eukaryotic species. b 4 abundant, copper ions were almost absent and zinc ions, although more available than copper ions, were more scarce than manganese or iron ions.(14,17–21) This accordance provides more evidence to support the view that the generation of c.68, a.1, d.92, and b.6 folds was linked to the recruitment of manganese, iron, zinc, and copper cofactors, respectively. A search of the MANET(11) and Swiss-Prot databases(16) showed that the generation of these initial metalloprotein folds also likely results from the structural needs of executing the novel enzymatic functions introduced by the new metallic cofactors, because the functions for the most ancient enzymes covered by these folds are not shared by all of the enzymes that use the prior folds (Table 2). As manganese, iron and copper are redox-active elements, they are widely used as redox cofactors in organisms. However, a recent analysis revealed that the primitive redox proteins depended more on organic redox coenzymes than on metallic cofactors.(30) This opinion is supported by the present finding, because most of the primordial redox proteins belong to NAD(P)-binding Rossmann-fold domains (c.2, no. 3) or TIM beta/alpha-barrel (c.1, no. 4) folds(30,31), and these folds appeared much earlier than the oldest folds for manganese, iron, and copper proteins (i.e., c.68, a.1 and b.6). Origin and evolution of superoxide dismutases The emergence of oxygen was a critical event in the evolutionary history of life, because aerobic respiration could produce more energy(32) and generate many novel metabolites.(33) As a result, the sudden rise in oxygen probably facilitated the leap from prokaryotes to eukaryotes,(32,34) and triggered the Cambrian explosion of the animal community.(35,36) However, the achievements of oxygen were gained at the cost of production of reactive oxygen species (ROS), such as superoxide, hydrogen peroxide and hydroxyl radical, that can damage organisms by oxidative reactions.(37) To guard against the toxic effects of ROS, organisms have evolved a defense system, in which superoxide dismutases (SODs), which can efficiently scavenge superoxide, are very important components.(37) Three major types of SODs have been identified in organisms, each containing a distinct metallic cofactor, i.e., iron, manganese, and copper (accompanied by zinc).(37) The distributions of these SODs in the biological kingdom and subcellular compartments are quite inhomogeneous. For instance, Fe-SOD is present in obligate anaerobes, aerobic diazotrophs, and chloroplasts of eukaryotic algae and higher plants; Mn-SOD is widely spread in bacterial, plants (localized in thylakoid or mitochondria), fungi, and animals (concentrated in mitochondria); and Cu, Zn-SOD is present in almost all eukaryotic cells (mainly in cytosol).(37,38) The phylogenetic BioEssays 9999:1–6, ß 2009 Wiley Periodicals, Inc. H.-F. Ji et al. distribution and the subcellular localization of these SODs are compatible with the endosymbiotic theory of chloroplasts and mitochondria, i.e., that these organelles of eukaryotic cells originated from endosymbiosis of prokaryotes,(38) and also imply that Fe-SOD and Mn-SOD emerged earlier than Cu, ZnSOD. This evolutionary theory for SODs is further supported by the present analysis results. As stated above, in the ancient anoxic world, iron and manganese putatively had a greater bioavailability than copper and zinc.(14,17–21) Therefore, at the beginning of oxic period, organisms had to select the former as the cofactors of SODs. In fact, the sequence and structure of Fe-SOD and MnSOD are rather similar to each other, which implies that they have a common ancestor.(37,38) With the elevation of oxygen concentration, copper, and zinc became more bioavailable, enabling the formation of Cu, Zn-SOD. The C-terminal domains of Fe-SOD and Mn-SOD belong to d.44 fold (no. 171 in the established protein architectural chronology of Caetano-Anollés et al.), which is adjacent to the most ancient copper protein fold (b.6 fold). As the d.44 fold is exclusive to SOD and the b.6 fold is shared by all cupredoxins, the present findings strongly suggest that the formations of Fe-SOD, Mn-SOD, and cupredoxins were contemporary events. An obvious question is why was the initially bioavailable copper not used to make Cu, Zn-SOD but to produce cupredoxins? We think the answer may lie in the fact that, for producing Cu, Zn-SOD, zinc (in addition to copper) is needed, a material that is also in short supply. Although Cu, Zn-SOD uses an rather early fold (b.1, no. 51 in protein architectural chronology), this fold was not invented for Cu, Zn-SOD, because Cu, Zn-SOD is a late member containing this fold according to the records of the MANET database.(11) Recently, Stevens also pointed out that, although Cu, Zn-SOD and cupredoxins are homologues, the latter emerged earlier than the former.(39) Thus, all the evidence from metallic bioavailability and fold usage suggests that Fe-SOD and MnSOD appeared earlier than Cu, Zn-SOD, although the latter recruited an early fold. Conclusions The most ancient architectures for manganese, iron, zinc, and copper proteins were identified by analyzing fold-domain mapping patterns in metalloproteins. An analysis of families harboring these folds showed that their generation was tightly coupled with corresponding trace metals. Moreover, the emerging order of these protein folds (those for manganese and iron appearing before those for zinc and copper) coincides with the putative bioavailability of these metals in the ancient anoxic ocean. Manganese and iron had a relatively high bioavailability, whereas zinc and copper did not.(14,17–21) Taken together, we conclude that the generation BioEssays 9999:1–6, ß 2009 Wiley Periodicals, Inc. Hypothesis of new protein folds could be linked to the recruitment of metallic cofactors. This link could be explained by the emergence of protein structures with novel folds that could fulfill the new protein functions introduced by the metallic cofactors. The present study not only has implications for the evolution of SODs, but is also significant for understanding the mechanistic link between evolution and environments.(40) It has been widely recognized that environmental factors, such as nutrient availability and temperature, could have impacts on genome and proteome compositions.(19,41–46) However, our study suggests that even protein architectural evolution could be influenced by environmental factors, i.e., the bioavailability of metals. Although the present hypothesis seems reasonable and self-consistent, it is challenging to test it experimentally. In the previous study, we proposed that the correlation between the recruitment of organic cofactors and and the generation of primordial protein architectures could be tested by in vitro selection,(6) a technique that can select structured proteins from a random protein sequence pool by cofactors.(47,48) We think that this methodology is also applicable to testing the present hypothesis. If manganese, heme-iron, zinc, and copper-selected proteins could use c.68, a.1, d.92, and b.6 folds, respectively, to execute the corresponding initial enzymatic functions (as listed in Table 2), our hypothesis will be largely supported. In addition, if the non-metal ancestors for the primitive metal proteins could be identified, one could start from the non-metal proteins to establish the random protein sequence pools. For instance, through searching PDB, it was found that the primitive copper protein azurin (PDB entry: 1CUO, b.6 fold) is similar to a non-copper protein (PDB entry: 2NV6) with an ancient function (enoylACP reductase) and a very early fold (c.2, no. 4) (Fig. 2). Thus, an evolutionary relationship might exist between both proteins, which implies that one could fix the common sequences of azurin and enoyl-ACP reductase and randomly change the other parts to constitute a random protein sequence library. Immobilized copper-ion affinity chromatography could then be used to select copper-binding proteins. If some of the selected proteins belong to b.6 fold and have cytochrome C oxidase-like activity, the link between copper ion recruitment and b.6 fold formation could be largely validated. Figure 2. Aligned sequences of azurin (PDB entry: 1CUO) and enoyl-ACP reductase (PDB entry: 2NV6) with identity of 33%. 5 Hypothesis Acknowledgments: This study was supported by the National Basic Research Program of China (grant 2003CB114400), the National Natural Science Foundation of China (grants 30870520 and 30700113) and Outstanding Youth Foundation of Shandong Province (grant JQ200812). References 1. Murzin, A. G., Brenner, S. E., Hubbard, T. and Chothia, C., SCOP: a structural classification of proteins database for the investigation of sequences and structures. J Mol Biol 1995. 247: 536–540. 2. Orengo, C. A., Michie, A. D., Jones, S., Jones, D. T., Swindells, M. B., et al. CATH-a hierarchic classification of protein domain structures. Structure 1997. 5: 1093–1108. 3. Chothia, C., Proteins. One thousand families for the molecular biologist. Nature 1992. 357: 543–544. 4. Caetano-Anollés, G., Wang, M., Caetano-Anollés, D. and Mittenthal, J. E., The origin, evolution and structure of the protein world. Biochem J 2009. 417: 621–637. 5. Tokuriki, N. and Tawfik, D., Protein dynamism and evolvability. Science 2009. 324: 203–207. 6. Ji, H. F., Kong, D. X., Shen, L., Chen, L. L., Ma, B. G., et al. Distribution patterns of small-molecule ligands in the protein universe and implications for origin of life and drug discovery. Genome Biol 2007. 8: R176. 7. Ji, H. F. and Zhang, H. Y., Bioinformatic identification of the most ancient copper protein architecture. J Biomol Struct Dyn 2008. 26: 197–201. 8. Caetano-Anollés, G. and Caetano-Anollés, D., An evolutionarily structured universe of protein architecture. Genome Res 2003. 13: 1563– 1571. 9. Caetano-Anollés, G. and Caetano-Anollés, D., Universal sharing patterns in proteomes and evolution of protein fold architecture and life. J Mol Evol 2005. 60: 484–498. 10. Wang, M., Boca, S. M., Kalelkar, R., Mittenthal, J. E. and CaetanoAnollés, G., Phylogenomic reconstruction of the protein world based on a genomic census of protein fold architecture. Complexity 2006. 12: 27– 40. 11. Kim, H. S., Mittenthal, J. E. and Caetano-Anollés, G., MANET: tracing evolution of protein architecture in metabolic networks. BMC Bioinformatics 2006. 7: 351. 12. Rydén, L., Evolution of blue copper proteins. Prog Clin Biol Res 1988. 274: 349–366. 13. Yomo, T., Saito, S. and Sasai, M., Gradual development of protein-like global structures through functional selection. Nat Struct Biol 1999. 6: 743–746. 14. Williams, R. J., A system’s view of the evolution of life. J R Soc Interface 2007. 4: 1049–1070. 15. Egami, F., Origin and early evolution of transition element enzymes. J Biochem 1975. 77: 1165–1169. 16. Boeckmann, B., Bairoch, A., Apweiler, R., Blatter, M. C., Estreicher, A., et al. The SWISS-PROT protein knowledgebase and its supplement TrEMBL in 2003. Nucleic Acids Res 2003. 31: 365–370. 17. Williams, R. J. and Fraústo Da Silva, J. J., Evolution was chemically constrained. J Theor Biol 2003. 220: 323–343. 18. Saito, M. A., Sigman, D. M. and Morel, F. M. M., The bioinorganic chemistry of the ancient ocean: the co-evolution of cyanobacterial metal requirements and biogeochemical cycles at the Archean-Proterozoic boundary? Inorg Chim Acta 2003. 356: 308–318. 19. Dupont, C. L., Yang, S., Palenik, B. and Bourne, P. E., Modern proteomes contain putative imprints of ancient shifts in trace metal geochemistry. Proc Natl Acad Sci USA 2006. 103: 17822– 17827. 20. Anbar, A. D., Elements and evolution. Science 2008. 322: 1481–1483. 21. Glass, J. B., Wolfe-simon, F. and Anbar, A. D., Coevolution of metal availability and nitrogen assimilation in cyanobacteria and algae. Geobiology 2009. 7: 100–123. 6 H.-F. Ji et al. 22. Canfield, D. E., The early history of atmospheric oxygen: Homage to Robert M. Garrels. Annu Rev Earth Planet Sci 2005. 33: 1–36. 23. Kato, Y., Suzuki, K., Nakamura, K., Hickman, A. H., Nedachi, M., et al. Hematite formation by oxygenated groundwater more than 2.76 billion years ago. Earth Planet Sci Lett 2009. 278: 40–49. 24. Hoashi, M., Bevacqua, D. C., Otake, T., Watanabe, Y., Hickman, A. H., et al. Primary haematite formation in an oxygenated sea 3.46 billion years ago. Nat Geosci 2009. 2: 301–306. 25. Konhauser, K., Deepening the early oxygen debate. Nat Geosci 2009. 2: 241–242. 26. Prachumwat, A. and Li, W. H., Protein function, connectivity, and duplicability in yeast. Mol Biol Evol 2006. 23: 30–39. 27. Cherry, J. M., Ball, C., Weng, S., Juvik, G., Schmidt, R., et al. Genetic and physical maps of Saccharomyces cerevisiae. Nature 1997. 387 (6632 Suppl): 67–73. 28. Andreeva, A., Howorth, D., Chandonia, J. M., Brenner, S. E., Hubbard, T. J. P., et al. Data growth and its impact on the SCOP database: new developments. Nucleic Acids Res 2008. 36: D419–D425. 29. Koonin, E. V., Wolf, Y. I. and Karev, G. P., The structure of the protein universe and genome evolution. Nature 2002. 420: 218–223. 30. Ji, H. F., Chen, L. and Zhang, H. Y., Organic cofactors participated more frequently than transition metals in redox reactions of primitive proteins. BioEssays 2007. 30: 766–771. 31. Ma, B. G., Chen, L., Ji, H. F., Chen, Z. H., Yang, F. R., et al. Characters of very ancient proteins. Biochem Biophys Res Commun 2008. 366: 607– 611. 32. Baudouin-Cornu, P. and Thomas, D., Oxygen at life’s boundaries. Nature 2007. 445: 35–36. 33. Raymond, J. and Segrè, D., The effect of oxygen on biochemical networks and the evolution of complex life. Science 2006. 311: 1764– 1767. 34. Chen, L. L., Wang, G. Z. and Zhang, H. Y., Sterol biosynthesis and prokaryotes-to-eukaryotes evolution. Biochem Biophys Res Commun 2007. 363: 885–888. 35. Knoll, A. H. and Carroll, S. B., Early animal evolution: emerging views from comparative biology and geology. Science 1999. 284: 2129–2137. 36. Falkowski, P. G. and Isozaki, Y., The story of O2. Science 2008. 322: 540–542. 37. Halliwell, B. and Gutteridge, J., Free radicals in biology and medicine, 4th edition, Oxford, Oxford University Press, 2007. 38. Grace, S. C., Phylogenetic distribution of superoxide dismutase supports an endosymbiotic origin for chloroplasts and mitochondria. Life Sci 1990. 47: 1875–1886. 39. Stevens, F. J., Homology versus analogy: Possible evolutionary relationship of immunoglobulins, cupredoxins, and Cu, Zn-superoxide dismutase. J Mol Recognit 2008. 21: 20–29. 40. Dietrich, L. E., Tice, M. M. and Newman, D. K., The co-evolution of life and Earth. Curr Biol 2006. 16: R395–R400. 41. Cambillau, C. and Claverie, J. M., Structural and genomic correlates of hyperthermostability. J Biol Chem 2000. 275: 32383–32386. 42. Suhre, K. and Claverie, J. M., Genomic correlates of hyperthermostability, an update. J Biol Chem 2003. 278: 17198–17202. 43. Foerstner, K. U., von Mering, C., Hooper, S. D. and Bork, P., Environments shape the nucleotide composition of genomes. EMBO Rep 2005. 6: 1208–1213. 44. Baudouin-Cornu, P., Surdin-Kerjan, Y., Marlière, P. and Thomas, D., Molecular evolution of protein atomic composition. Science 2001. 293: 297–300. 45. Elser, J. J., Fagan, W. F., Subramanian, S. and Kumar, S., Signatures of ecological resource availability in the animal and plant proteomes. Mol Biol Evol 2006. 23: 1946–1951. 46. Bragg, J. G. and Wagner, A., Protein carbon content evolves in response to carbon availability and may influence the fate of duplicated genes. Proc R Soc B 2007. 274: 1063–1070. 47. Roberts, R. W. and Szostak, J. W., RNA-peptide fusions for the in vitro selection of peptides and proteins. Proc Natl Acad Sci USA 1997. 94: 12297–12302. BioEssays 9999:1–6, ß 2009 Wiley Periodicals, Inc.