* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download - Wiley Online Library
Ancestral sequence reconstruction wikipedia , lookup
Transcriptional regulation wikipedia , lookup
Biosynthesis wikipedia , lookup
Genetic engineering wikipedia , lookup
Molecular ecology wikipedia , lookup
Gene desert wikipedia , lookup
Metabolic network modelling wikipedia , lookup
Transposable element wikipedia , lookup
Gene regulatory network wikipedia , lookup
Point mutation wikipedia , lookup
Ridge (biology) wikipedia , lookup
Evolution of metal ions in biological systems wikipedia , lookup
Promoter (genetics) wikipedia , lookup
Biochemistry wikipedia , lookup
Genomic imprinting wikipedia , lookup
Community fingerprinting wikipedia , lookup
Microbial metabolism wikipedia , lookup
Silencer (genetics) wikipedia , lookup
Non-coding DNA wikipedia , lookup
Gene expression profiling wikipedia , lookup
Genomic library wikipedia , lookup
Citric acid cycle wikipedia , lookup
Amino acid synthesis wikipedia , lookup
Endogenous retrovirus wikipedia , lookup
Artificial gene synthesis wikipedia , lookup
FEMS Microbiology Reviews 28 (2004) 335–352 www.fems-microbiology.org A challenge for 21st century molecular biology and biochemistry: what are the causes of obligate autotrophy and methanotrophy? Ann P. Wood a, Jukka P. Aurikko a, Donovan P. Kelly a b,* Department of Life Sciences, KingÕs College London, Franklin Wills Building, 150 Stamford Street, London SE1 9NN, UK b Department of Biological Sciences, University of Warwick, Coventry CV4 7AL, UK Received 5 August 2003; received in revised form 11 November 2003; accepted 17 December 2003 First published online 30 December 2003 Abstract We assess the use to which bioinformatics in the form of bacterial genome sequences, functional gene probes and the protein sequence databases can be applied to hypotheses about obligate autotrophy in eubacteria. Obligate methanotrophy and obligate autotrophy among the chemo- and photo-lithotrophic bacteria lack satisfactory explanation a century or more after their discovery. Various causes of these phenomena have been suggested, which we review in the light of the information currently available. Among these suggestions is the absence in vivo of a functional a-ketoglutarate dehydrogenase. The advent of complete and partial genome sequences of diverse autotrophs, methylotrophs and methanotrophs makes it possible to probe the reasons for the absence of activity of this enzyme. We review the role and evolutionary origins of the Krebs cycle in relation to autotrophic metabolism and describe the use of in silico methods to probe the partial and complete genome sequences of a variety of obligate genera for genes encoding the subunits of the a-ketoglutarate dehydrogenase complex. Nitrosomonas europaea and Methylococcus capsulatus, which lack the functional enzyme, were found to contain the coding sequences for the E1 and E2 subunits of a-ketoglutarate dehydrogenase. Comparing the predicted physicochemical properties of the polypeptides coded by the genes confirmed the putative gene products were similar to the active a-ketoglutarate dehydrogenase subunits of heterotrophs. These obligate species are thus genomically competent with respect to this enzyme but are apparently incapable of producing a functional enzyme. Probing of the full and incomplete genomes of some cyanobacterial and methanogenic genera and Aquifex confirms or suggests the absence of the genes for at least one of the three components of the a-ketoglutarate dehydrogenase complex in these obligate organisms. It is recognized that absence of a single functional enzyme may not explain obligate autotrophy in all cases and may indeed be only be one of a number of controls that impose obligate metabolism. Availability of more genome sequences from obligate genera will enable assessment of whether obligate autotrophy is due to the absence of genes for a few or many steps in organic compound metabolism. This problem needs the technologies and mindsets of the present generation of molecular microbiologists to resolve it. Ó 2003 Federation of European Microbiological Societies. Published by Elsevier B.V. All rights reserved. Keywords: Nitrosomonas; Methylococcus; Obligate autotrophy; Methanotrophy; Methanogens; Krebs cycle; Genome probing Contents 1. 2. 3. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Genome sequences and bioinformatic databases as untapped resources. . . . . . . . . . . . . . . . Background to obligate metabolic modes of life . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Bacterial obligate autotrophy and methanotrophy and their relationship to autotrophy in Archaea. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . * Corresponding author. Tel.: +44-24-7657-2907; fax: +44-24-7652-3701. E-mail address: [email protected] (D.P. Kelly). 0168-6445/$22.00 Ó 2003 Federation of European Microbiological Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.femsre.2003.12.001 336 337 337 337 336 A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 3.2. 3.3. 3.4. Survival benefits of being an obligate or a facultative autotroph . . . . . . . . . . . . . . A unitary cause of obligate autotrophy and methanotrophy? . . . . . . . . . . . . . . . . Origin and evolution of the Krebs cycle: why should it not function in an obligate autotroph? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5. Testing hypotheses about the incomplete Krebs cycle . . . . . . . . . . . . . . . . . . . . . . Testing the a-ketoglutarate hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. The a-ketoacid dehydrogenase family . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Using the partial and complete bacterial genomes and protein sequences available on the public databases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3. In silico detection of the E1 and E2 subunits of a-ketoglutarate dehydrogenase in target genomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4. Evaluation of the validity of the probing technique . . . . . . . . . . . . . . . . . . . . . . . 4.5. Refinement of the identification criteria for the a-ketoglutarate dehydrogenase E2 subunit sequence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6. BLAST probing of target genomes for E1 and E2 genes . . . . . . . . . . . . . . . . . . . . 4.7. Positive identification of the E1 and E2 genes in the genomes of N. europaea and M. capsulatus and computed physicochemical properties of the polypeptides encoded by the genes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8. A basis for obligate autotrophy and methanotrophy? . . . . . . . . . . . . . . . . . . . . . . Testing hypotheses about the nature of obligate autotrophy and methanotrophy and the absence of a-ketoglutarate dehydrogenase activity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Detecting the diagnostic coding sequences for a-ketoglutarate dehydrogenase in obligate autotrophs and methanotrophs: bioinformatic approaches . . . . . . . . . . . . 5.2. Is there a functional transcription product in obligate genera with an E1 coding sequence: is there an mRNA? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3. Is there an enzymically inactive translation product from E1 and E2 genes in obligate genera? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4. Expression of the E1 and E2 genes of obligate genera in heterotrophic hosts and vice versa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Concluding remarks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338 338 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349 4. 5. 6. 1. Introduction This review is primarily concerned with an ‘‘old problem’’, namely the attempts to explain the phenomena of ‘‘obligate autotrophy and methanotrophy’’ exhibited by diverse bacteria which otherwise share little phylogenetic commonality. In the review we concentrate primarily on the chemolithotrophs, cyanobacteria and methanotrophs, the ‘‘obligate’’ metabolism of which was studied in detail up to and during the 1960s and 1970s. Autotrophy among the Archaea and anaerobic phototrophic bacteria is considered only for comparative purposes as those organisms were not prime subjects of the ‘‘obligate autotrophy’’ debate at that time. For the purposes of the review we have used the definition of autotrophs (and autotrophy) proposed by Whittenbury and Kelly [137] to encompass all those obligately autotrophic and methanotrophic bacteria which ‘‘synthesize all their cellular constituents from one or more C1 compounds’’. The organisms central to our discussion all assimilate carbon by a C1 + C5 condensation and are 340 341 341 341 342 342 343 344 344 345 346 347 347 347 347 347 348 (a) the thiobacilli (deriving energy from inorganic sulfur compound oxidation), Nitrosomonas (deriving energy from ammonia oxidation to nitrite) and the phototrophic cyanobacteria, which all use the Calvin ribulose bisphosphate carboxylase cycle and (b) the obligate methanotrophs (such as Methanomonas methanica and Methylococcus capsulatus), which use the Quayle ribulose monophosphate (RuMP) cycle, deriving both energy and carbon from methane. We suggest approaches to the problem of obligate autotrophy using molecular biology techniques, which are based on discoveries and information largely achieved since the time at which attempts to explain obligate autotrophy virtually ceased [71,104,112,113,137]. The advent of these techniques and the consequent explosion of information about bacterial genomes and the control of gene expression means that this problem can be looked at from viewpoints that were unthinkable 20 years ago. A holistic approach, combining historical observations, molecular methods and new biochemical methods, is needed to solve this long-standing and intricate problem. A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 2. Genome sequences and bioinformatic databases as untapped resources The application of sequencing techniques has resulted in the accumulation of almost inconceivable quantities of information about the bacterial genome. The public databases now contain the sequences of billions of nucleotides coding for entire genomes, as well as for individual structural and functional genes, while the protein databases contain the amino acid sequences of the polypeptides encoded by many of these genes. The massively powerful computing systems and search engines, capable of making almost instant comparisons among these archived sequences and with newly acquired data, have revolutionized bacterial taxonomic procedures and have helped rewrite taxonomy from Kingdom to species level. Moreover, evolutionary and functional relationships among enzymes with the same catalytic function but from taxonomically diverse organisms can be analyzed in the same way, generating phylogenetic trees of polypeptides from which much can be deduced regarding the evolutionary history and conservation of function of numerous enzymes [18,50,62,77]. Two applications deriving from the availability of techniques to identify genes encoding specific functions have been (1) the ability to identify such functional genes in the partial or complete genomes of bacteria and (2) to use molecular probes to detect those genes in environmental samples, as well as in cultured isolates or uncultured molecular clones [3,5,53,60,91,102]. Current molecular biology is making significant use of gene probes to establish the presence not only of specific organism types in environmental samples, but to draw conclusions about possible metabolic processes that might be occurring in the habitats studied [56,59,88]. These molecular techniques are useful tools in indicating what might be possible, but in themselves cannot prove that any organism types so identified, or metabolic processes indicated, contribute to the dynamics of a given habitat. As Palleroni [92] has pointed out ‘‘Modern approaches based on the use of molecular approaches presumed to circumvent the need for culturing prokaryotes fail to provide sufficient and reliable information for estimation of prokaryotic diversity’’. Similarly, the presumption that the use of sequence data allows the deduction of the properties of uncultivated organisms [90] is true only if the genes detected can actually be expressed in vivo [42]. Thus, for a full understanding of a natural habitat, molecular and biochemical analysis need to be applied conjointly. The corollary of this is that molecular analysis can be applied to known organisms or systems to establish how observable biochemical processes are likely to be underpinned and encoded at the level of the genome. 337 Our aim is to show how the use of molecular methods and the public databases can help to understand obligate autotrophy and methanotrophy. 3. Background to obligate metabolic modes of life 3.1. Bacterial obligate autotrophy and methanotrophy and their relationship to autotrophy in Archaea Diverse chemolithotrophic, photolithotrophic and methanotrophic bacteria and some Archaea are characterized by their inability to grow as heterotrophs on either simple or complex organic media. These are the obligate autotrophs and type I methanotrophs, whose one-carbon (C1 )-compound-dependent metabolism requires the fixation of carbon dioxide using energy from inorganic oxidations or light, or uses methane (or in some cases methanol) as both energy and carbon substrate [69,71,112,137]. Type I methanotrophs are members of the c-proteobacteria and are characterized by having bundles of disc-shaped internal membrane vesicles, lacking a complete Krebs cycle, and using the ribulose monophosphate (Quayle) cycle to assimilate carbon from C1 -compounds [78]. Some metabolic abilities are common to both obligate autotrophs and type I methanotrophs. They use similar cyclic sugar–phosphate pathways for the assimilation of carbon dioxide or reduced C1 -compounds: the ribulose bisphosphate Calvin cycle or the Quayle cycle, respectively [78,108]. Several strains of the obligate methanotroph M. capsulatus also possess the key enzyme of the Calvin cycle, ribulose bisphosphate carboxylase, and are capable of autotrophic growth with hydrogen as the energy substrate [6,126]. Similarly, the obligate chemolithoautotroph, Thiobacillus thioparus, can oxidize and assimilate carbon from methylated sulfides [63,115]. The obligately anaerobic methanogenic Archaea and some aerobic and facultatively anaerobic sulfur-oxidizing Archaea can also be classified as chemolithoautotrophic. These derive energy from coupling the oxidation of hydrogen to the reduction of carbon dioxide to methane (or by methane production from C1 -compounds such as methanol or methylated sulfides), or from the oxidation of elemental sulfur or sulfides to sulfate, and derive cell-carbon from carbon dioxide or C1 -compounds [33,108,109]. The methanogens and sulfur-oxidizing Archaea employ a variety of pathways for autotrophic carbon dioxide fixation, none proved to be like those of the eubacterial autotrophs, some of which involve acetate or pyruvate as an intermediate, and some of which use a reductive tricarboxylic acid cycle to fix carbon dioxide [34–36,55]. Central to the C1 -metabolism of methanogens are carbon monooxide dehydrogenase and the Wood–Ljungdahl pathway, enabling the energy-efficient synthesis of acetyl 338 A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 CoA from carbon dioxide and hydrogen as chemolithotrophic growth substrates [30–32,34–36]. A reductive tricarboxylic acid cycle supports the growth of the obligately anaerobic photolithoautotroph Chlorobium [34,35], and yet another cyclic carbon dioxide-fixation pathway, with 3-hydroxypropionate as the key intermediate, is found in the phototroph Chloroflexus aurantiacus [52,121]. Clearly, obligately methanogenic growth, obligately anaerobic photolithoautotrophy and the autotrophic growth by some sulfur- or sulfate-reducing eubacteria and Archaea, and some sulfur-oxidizing Archaea, are likely to present problems distinct from those of obligate chemolithotrophy and methanotrophy in aerobes such as Nitrosomonas and Methylococcus. Indeed, the diverse metabolic types represented by these organisms were little considered from the point of view of obligate autotrophy at the time of studies in the 1950s–1970s on obligate methanotrophy and obligate autotrophy in aerobic chemolithotrophs and cyanobacteria. For this reason, we have devoted most of this review to a reconsideration of the causes of obligate methanotrophy and autotrophy in aerobic eubacteria and cyanobacteria, using examples from the Archaea where parallels can be drawn and when possible biochemical or evolutionary links can be proposed. 3.2. Survival benefits of being an obligate or a facultative autotroph The extreme specialization and restricted metabolism exemplified by obligate chemolithotrophy has been shown to give selective competitive advantage in the case of some aerobic sulfur-oxidizing bacteria [75,114]. In natural or laboratory environments where the ratio of inorganic substrate to organic nutrients is high, the obligate species tend to dominate over facultatively mixotrophic species [75,114,137]. Several factors can affect such competition but in general the obligate species exhibit much greater flexibility in responding to variable supplies of their inorganic substrate, frequently also showing higher affinities for the substrate, ensuring their survival in fluctuating environments [75]. Unlike simple competition between two heterotrophs for a single substrate, when one will eventually exclude the other completely, the evidence indicates that an obligate chemolithotroph will be maintained, albeit sometimes at a low level, in mixture with a facultative species over long periods of time, especially under fluctuating conditions of nutrient supply [45,46,75,114]. Such observations indicate how the obligate mode of metabolism could have been of selective advantage over geological time. While the survival benefit of being an obligate species shows that the ‘‘rigidity’’ of autotrophic metabolism can be selectively advantageous, it does not help answer the question of how obligate autotrophy is imposed. 3.3. A unitary cause of obligate autotrophy and methanotrophy? Obligate autotrophy among the chemo- and photolithotrophic bacteria still lacks satisfactory explanation over 100 years since its discovery [69,71,99,137,140]. The cause(s) of obligate autotrophy have never been fully identified, but the four principal hypotheses proposed from studies on thiobacilli and nitrifying bacteria up to the 1970s were: (1) the ‘‘submarine hypothesis’’, which viewed the bacteria as impermeable to organic nutrients [132,137,140]. Impermeability alone has, however, been discounted as various chemolithotrophs, phototrophs and methanotrophs do incorporate organic compounds into cell carbon, but only as a supplement to their normal unique metabolic processes [54,65,69,70,74,80,103,104, 107,112,113,124,137,146]. In some cases the amounts incorporated supply only a small fraction of the cell biomass, and in others a compound may be totally excluded (e.g., glucose by some thiobacilli), probably because of the absence of specific transport systems. In other examples, some amino acids for protein synthesis can be derived entirely from exogenous sources, and compounds such as acetate can provide a large part of the total cell-carbon, for example, to Halothiobacillus neapolitanus growing chemolithotrophically on thiosulfate [66,69,112]; (2) that organic nutrients were toxic or inhibited specific metabolic steps [132,137]. Indeed Winogradsky [140] defined such toxicity as a characteristic of the autotrophic ‘‘Anorgoxydant’’ [71]. Relatively few organic nutrient compounds have been shown to exert such toxicity, and growth inhibition by some amino acids for example is typically due to regulatory effects on branched biosynthetic pathways, preventing the synthesis of the other product(s) of a branched pathway [27–29,67,70,137]; (3) that they were unable to derive energy from NADH oxidation [54,113]. This can be discounted as NADH oxidation through the cytochrome system occurs [130] and endogenous ATP levels are in many cases sustained by the metabolism of organic storage materials. Several obligate species have been shown to store polyglucose and other carbohydrates as carbon and energy reserves, so must be able to derive energy from their dissimilation [7,8]; (4) in many of these organisms, a contributory cause of their obligate metabolism has been ascribed to the single enzymatic lesion of the absence of a-ketoglutarate dehydrogenase activity. Activity of this enzyme is absent from various thiobacilli, Nitrosomonas europaea, M. capsulatus and type I methanotrophs, some cyanobacteria and some Archaea [24,54,69,78,83,96, 111–113,137]. A corollary of the absence of activity of this key enzyme of the Krebs cycle in some obligate autotrophs is that the reactions of the Krebs cycle function as a biosynthetic ‘‘horseshoe’’, channelling carbon from carbon dioxide or methane into central biosynthetic pathways (Fig. 1). A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 CO2 fixation via the Calvin Cycle (autotrophs) 339 CH4 fixation via the RuMP cycle (Type I and X methanotrophs) Phosphoenolpyruvate Pyruvate acetyl CoA CO2 oxaloacetate malate citrate fumarate cis-aconitate succinyl CoA BLOCKED isocitrate α-ketoglutarate BIOSYNTHESIS BIOSYNTHESIS Fig. 1. Simplified schematic to illustrate the biosynthetic ‘‘horseshoe’’ reactions of the incomplete Krebs cycle in obligate autotrophs and methanotrophs [65,66,69,96,112,137]. The summary given above represents the state of knowledge some 20 years ago and has not subsequently been substantially advanced at the biochemical level. Attempts to grow various obligate autotrophs on defined media, which seek to replace the intermediates of autotrophic metabolism, have all failed, preventing full analysis of their possible ‘‘heterotrophic’’ control mechanisms. Recently, the genome of N. europaea has been shown to contain only limited numbers of genes for the transport and metabolism of organic molecules [16], underlining the possibility that its obligate autotrophy is due to a complex of enzyme ‘‘deficiencies’’ rather than any single genetic lesion. It remains possible that obligate autotrophy results from a complex of metabolic controls and deficiencies: low ability to transport external nutrients in the absence of the normal energy substrate coupled with low rates of energy generation from their dissimilation and possible repression of essential ‘‘autotrophic’’ processes either by the external nutrients or as a consequence of the induction of enzymes for the metabolism of those nutrients [75,103]. Obligate autotrophy could result from the organism being able to vary the synthesis of enzymes of its central, autotrophically based, metabolism only within narrow limits, thus preventing it from responding positively to the supply of exogenous nutrients. There can obviously be unlimited speculation along these lines, but limited evidence is yet available to answer the key questions. This topic is in need of research at the biochemical and molecular levels using the techniques not available 20 or more years ago, when these possibilities were considered. In looking for a primary and unitary cause of obligate autotrophy and methanotrophy, a biosynthetic or degradative lesion in central metabolism (such as the lack of a functional a-ketoglutarate dehydrogenase) would seem more probable in evolutionary terms than organic substrate exclusion or toxicity, and would be consistent with the occurrence of obligate autotrophy throughout the phylogenetic tree. This diversity has only recently become apparent with the use of 16S rRNA sequence analysis to reveal phylogenetic relationships. Thirty years ago, physiological properties were taken to indicate closer taxonomic relationships than can now be accepted. Extreme examples are the nitrite-oxidizing Nitrospira and iron-oxidizing Leptospirillum genera, which are now known to be members of a phylum distinct from the proteobacteria, in which physiologically similar bacteria are located [78]. We explore the viewpoint that obligate autotrophy results from lesions of central metabolism because it is only in that area that new information has become available from studies of the genomes of obligate autotrophs and methanotrophs. For this reason we have used the absence or repression of a-ketoglutarate dehydrogenase as an example of how a single enzyme of central metabolism could be the key to understanding obligate autotrophy. This may be over-simplistic but does at least provide a starting point from which new research can begin and might even reveal that as a result of ‘‘higher-level’’ regulatory phenomena in obligate organisms, lack of expression of a-ketoglutarate 340 A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 dehydrogenase is a consequence rather than a cause of obligate autotrophy. It is recognized that the lipoamide-dependent NADþ reducing a-ketoglutarate dehydrogenase typical of most mitochondria, eukaryotes and aerobic eubacteria is not the only mechanism for the decarboxylation of a-ketoglutarate and production of succinyl-CoA. In various archaeons, including Halobacterium, Haloferax, Archaeoglobus and a Sulfolobus strain, ferredoxin-dependent 2-oxoacid:ferredoxin oxidoreductases catalyze the functions associated with eubacterial a-ketoglutarate and pyruvate dehydrogenases [37,38,58,61,72,73,145]. Interestingly, a mutant of Bradyrhizobium japonicum, lacking a-ketoglutarate dehydrogenase, ‘‘closes the gap’’ in the Krebs cycle by means of a CoA- and NAD(P)þ independent a-ketoglutarate decarboxylase, producing succinic semialdehyde [47]. Neither of these types of enzyme has been sought in obligate autotrophs such as Nitrosomonas, but their presence in such organisms is very unlikely, and their absence is indicated by pulselabelling experiments as described below. The absence of a cyclic Krebs cycle from some of these obligate genera results in the incorporation of 14 C-labelled acetate into protein only as leucine and those amino acids derived from a-ketoglutarate: glutamate, proline and arginine [51,65,66,112,124]. In some cases, including N. europaea, [14 C]acetate is also incorporated into amino acids derived from aspartate, requiring a C4 -precursor for its formation [136]. This was explained by the formation of C4 -carboxylic acids from isocitrate by the activity of the glyoxylate bypass enzyme, isocitrate lyase [69,136], which is negligible in some obligate autotrophs [69,94,136]. Indeed, the facultatively heterotrophic green alga Chlamydomonas dysosmos was rendered obligately autotrophic with respect to growth on acetate by a mutation which eliminated isocitrate lyase [89]. The facultatively heterotrophic methylotroph, Methylobacterium extorquens, strain AM1 contains a complete Krebs cycle with a functional a-ketoglutarate dehydrogenase when grown heterotrophically, but a-ketoglutarate dehydrogenase is repressed during growth on C1 -compounds and the incomplete Krebs cycle then acts biosynthetically as in obligate chemolithotrophs and methanotrophs (Fig. 1) [19]. M. extorquens could be rendered obligately methylotrophic by a mutation that deprived it of a-ketoglutarate dehydrogenase, supporting the view that the basis of naturally occurring obligate methylotrophy/methanotrophy could be the absence of this enzyme [80,125]. The latter work was recently extended by Van Dien et al. [133] to show that mutants of M. extorquens could be made that produced neither a-ketoglutarate dehydrogenase nor pyruvate dehydrogenase and consequently became greatly impaired in their ability to grow on succinate and pyruvate. It has also recently been shown that N. europaea (which does not produce active a-ketoglutarate dehydrogenase) appears to contain the coding sequences for this enzyme in its genome [16]. Conversely, introducing genes coding for heterotrophic metabolism into obligate autotrophs may enhance their ability to use exogenous organic materials. Acidithiobacillus thiooxidans is virtually unable to assimilate glucose during growth with elemental sulfur as energy substrate, but transconjugants into which the pfkA gene for phosphofructokinase had been inserted (and expressed) could consume glucose, and showed an increased growth rate and biomass production when grown chemolithotrophically on sulfur and glucose compared with the transconjugant grown on sulfur alone [127]. The growth rate and yield of the transconjugant were, however, considerably lower (by 30–55%) than those of the wild type A. thiooxidans growing on sulfur alone [127]. Heterotrophy of the A. thiooxidans was thus not shown (making the paper somewhat misleading in implying that ‘‘conversion of an obligate autotroph to heterotrophy’’ was shown), and the effect of the pfkA insert was no more than to enhance the potential for ‘‘mixotrophy’’. Mixotrophy is the wellestablished ability of obligate chemolithoautotrophs and methanotrophs to obtain carbon simultaneously from carbon dioxide and organic compounds, but only at the expense of their normal energy sources [66,69,70,103,144]. 3.4. Origin and evolution of the Krebs cycle: why should it not function in an obligate autotroph? The initial steps of the Krebs cycle, from citrate synthase to isocitrate dehydrogenase, are believed to be very ancient in origin, having entered the textbooks as possibly being ‘‘a remembrance of the early RNA world’’ [122]. The cycle itself probably evolved from a reductive biosynthetic tricarboxylic acid pathway which had originated in organisms in the pre-oxic biosphere [134]. One possibility is that a primordial autocatalytic anabolic chemoautotrophic pyrite-generating reductive cycle evolved into the oxidative catabolic Krebs cycle [134]. An intermediate stage in this evolution could have been a pathway primarily for amino acid biosynthesis, by means of a non-cyclic ‘‘horseshoe’’ pathway lacking an enzymatic step to form succinate (see Fig. 1) [82]. Its evolution into an energy-generating system was thus an opportunistic secondary development, producing the most biochemically robust system possible from existing metabolic processes [39,41,82,86,87], dating at least from the advent of the oxic atmosphere. In that the earliest self-sustaining metabolism was rooted in chemolithotrophy and that the first autotrophic carbon dioxide fixation pathway was probably a reductive citric acid cycle [71,134,135], it is among extant chemoautotrophs that relicts of these initial processes may be found. A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 In its extant form, the Krebs cycle enables the efficient conservation of biochemical energy from the oxidation of pyruvate to carbon dioxide. At maximum efficiency, as in the mitochondrion, the complete oxidation of a pyruvate molecule by pyruvate dehydrogenase and the Krebs cycle drives the synthesis of 15 ATP molecules [108]. In bacteria, the efficiency of electron transport phosphorylation coupled to NADH oxidation is lower, typically with a P/O ratio of 2 rather than 3 [108]. Consequently, pyruvate oxidation in an aerobic bacterium might generate only 10 ATP during pyruvate oxidation. For an obligate autotroph such as Nitrosomonas to operate a fully functional Krebs cycle, driven by pyruvate formed from carbon dioxide, would be ‘‘metabolic suicide’’ and could not occur in nature. The fixation of three carbon dioxide molecules via the Calvin cycle to form phosphoglycerate requires 9 ATP and 6 NADH, with one ATP being regained in the conversion of phosphoglycerate to pyruvate [100]. NADþ reduction in the obligate chemolithotrophs (with inorganic sulfur, iron or nitrogen compounds as substrates) requires energy-dependent ‘‘reverse’’ electron flow from cytochromes. This requires the equivalent of a minimum of one ATP per NADþ reduced. The synthesis of one molecule of pyruvate from carbon dioxide thus ‘‘costs’’ at least 14 ATP. For an autotroph to reoxidize this pyruvate would at best create a futile cycle of no metabolic energy or carbon benefit to the organism. As outlined above (Fig. 1), the reactions of a ‘‘horseshoe’’ Krebs cycle would be expected primarily to have a biosynthetic role in obligate autotrophs and methanotrophs [28,65,66,69,107,112,120,137]. The question arises as to whether this ‘‘horseshoe’’ is a relict of the evolutionarily ancestral pathway or has resulted from later deletion of function in the evolution of obligate metabolism from heterotrophic ancestors. In the case of the anoxygenic autotrophic methanogenic Archaea, it is easy to conclude that these never possessed a complete Krebs cycle if the archaea are relicts of very ancient forms of life (>3000 million years (My) before the present; [79,108,141,142]). If, however, the controversial, minority view of Cavalier-Smith [15] is correct that the archaebacteria arose only recently (850 My before present), with methanogenesis as a secondary character, then they too could have evolved the incomplete, biosynthetic, Krebs cycle by deletion of functions present in eubacterial ancestors [15]. While the evolutionary timing of archaeal origins is beyond our scope, it obviously affects deductions about the absence of Krebs cycle genes. Regardless of the date of their origin, the modern archaea could be descendants of a line of anaerobic, fermentative, heterotrophic eubacteria which had never developed a functional oxidative Krebs cycle. While modern archaea contain citrate synthase and succinate thiokinase [20,21], they do not seem to produce a functional E1:E2:E3 pyruvate dehydrogenase 341 complex, even though some do contain coding sequences showing high identities to the required genes [61]. This provides an interesting parallel with the failure of some autotrophs and methanotrophs to produce a-ketoglutarate dehydrogenase, although absence of a functional pyruvate dehydrogenase from archaea is bypassed by the conversion of pyruvate to acetyl CoA by an oxidoreductase [61]. 3.5. Testing hypotheses about the incomplete Krebs cycle Early studies demonstrated that a-ketoglutarate dehydrogenase activity was absent from diverse obligate chemo- and photo-lithoautotrophs and some obligate methanotrophs. It could not at that time necessarily be concluded that the genes encoding this enzyme were also absent from the bacteria [54,68,113], although this was presumed to be a possibility. This presumption was based on the view that modern obligate autotrophs and methanotrophs had probably evolved from ancestors in the pre-oxic biosphere which had never possessed a functional Krebs cycle, or from an ancestor which had lost the genes for a-ketoglutarate dehydrogenase. With the advent of total sequencing of bacterial genomes it has become feasible to test these assumptions and to probe more deeply the causes of obligate autotrophy and methanotrophy. We have used in silico genome search procedures to seek the genes encoding the components of a-ketoglutarate dehydrogenase in organisms known, or likely, to possess the functional enzyme and some which do not produce detectable activities of it during growth. We show that two such organisms, N. europaea and M. capsulatus, appear to be genomically competent with respect to the genes necessary to encode a-ketoglutarate dehydrogenase, but are apparently not able to achieve heterotrophic growth by their expression. 4. Testing the a-ketoglutarate hypothesis 4.1. The a-ketoacid dehydrogenase family This comprises three cytoplasmic enzymes: a-ketoglutarate dehydrogenase, pyruvate dehydrogenase and 2-oxoisovalerate dehydrogenase, each containing subunits which exhibit evolutionary relationships and highly conserved amino acid sequences [81,93,101]. The a-ketoglutarate dehydrogenase complex has been purified from a number of prokaryotes and eukaryotes and invariably consists of three subunits [106]: these are E1, E2 and E3, corresponding to a-ketoglutarate dehydrogenase (EC 1.2.4.2), dihydrolipoamide acyltransferase (EC 2.3.1.61) and dihydrolipoamide dehydrogenase (EC 1.8.1.4), respectively. The overall reaction catalyzed by the a-ketoglutarate dehydrogenase complex is 342 A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 a-ketoglutarate þ CoA þ NADþ ! succinyl-CoA þ CO2 þ NADH þ H þ The (simplified) component reactions are: E1 : a-ketoglutarate + lipoamide ! succinyl-lipoamide + CO2 E2 : succinyl-lipoamide + coenzyme A ! succinyl-CoA + dihydrolipoamide E3 : dihydrolipoamide + NADþ ! lipoamide + NADH + Hþ The E3 subunit is very highly conserved and is common to all a-ketoacid dehydrogenases [106,118,122] and it might thus be expected to occur in the genomes even of obligate autotrophs and methanotrophs. The E1 subunit, however, displays significant interspecies differences among the a-ketoacid dehydrogenase family members [81] and might thus be redundant in obligate species. In a fully functional Krebs cycle, the a-ketoglutarate dehydrogenase complex converts a-ketoglutarate to succinyl coenzyme A, which is required for some succinylation reactions and is an essential precursor for porphyrin synthesis. Succinate formation from succinylCoA is catalyzed by succinyl-CoA synthetase (EC 6.2.1.4), which can also catalyze the reverse reaction. Succinyl-CoA synthase is also absent from some autotrophically grown bacteria (which lack a-ketoglutarate dehydrogenase), consistent with its principal role in the forward direction of the Krebs cycle [82,95,97], indicating alternative CoA-transferase systems for succinylCoA formation in autotrophs [95,96]. The genes encoding the three proteins of the a-ketoglutarate dehydrogenase complex have been identified, sequenced and characterized from various organisms, including Escherichia coli (sucA, sucB, lpdA) and Bacillus subtilis (odhA, odhB, pdhD) [22,23,84,117,119]. The availability of full gene and translated protein sequences for the E1 and E2 subunits thus provided the possibility to search for these genes in other genomes. Several features of the E1 and E2 subunits of the E. coli a-ketoglutarate dehydrogenase complex can serve as biomarkers during in silico-based identification of the corresponding encoding genes, as was used to show conservation of these markers between E. coli and Rhodobacter capsulatus [23]. For E1, the markers are two phosphorylation sites (residues 391–405 and 440–456) and the cofactor (pyrophosphate thiamine)binding site (residues 362–389). For E2, the markers are two conserved residues assumed to be located in the active site (His-375 and Asp-379) and the cofactor (a covalently linked lipoic acid moiety)-binding site (Lys-43) [14,23]. In E. coli, sucA and sucB are part of an operon sharing the same promoter, transcribed on a common mRNA, whereas lpdA is part of an unrelated operon, thus potentially allowing differential regulation of the E1 + E2 and the E3 components of the a-ketoglutarate dehydrogenase complex. The odhA and odhB genes for the E1 and E2 subunits of the B. subtilis a-ketoglutarate dehydrogenase are similarly contiguous [14]. 4.2. Using the partial and complete bacterial genomes and protein sequences available on the public databases Molecular probing techniques not available when the a-ketoglutarate dehydrogenase hypothesis was conceived now enable us to test the possibility that ‘‘absence of enzyme means absence of gene’’. The following section (Sections 4.3–4.7) are a case study of this approach. Partial and complete genomes for several well-characterized obligate autotrophs and methanotrophs, some facultative methylotrophs, phototrophs and Archaea were probed for the presence of sequences for the E1 and E2 subunits of a-ketoglutarate dehydrogenase (Table 1). The amino acid sequences used to seek the presence of homologous sequences in the target organisms were those of the E1 and E2 subunits of E. coli strain K12. Control searches for a-ketoglutarate dehydrogenase subunits were conducted against the genomes of both E. coli strain 0157:H7 (EDL933), genome Accession No. NC_002655 [9], and B. subtilis, genome Accession No. NC_000964 [76]. 4.3. In silico detection of the E1 and E2 subunits of a-ketoglutarate dehydrogenase in target genomes The general procedure used to locate the genes encoding the E1 and E2 subunits employed the basic local alignment search tool (BLAST) as TBLASTX and TBLASTN (http://www.ncbi.nlm.nih.gov), allowing the probing of the translated genomes with appropriate protein sequences. The amino acid sequences selected were those from the SwissProt database for the E. coli E1 (a-ketoglutarate dehydrogenase) subunit, Accession No. p07015 (933 amino acids), and E2 (dihydrolipoamide succinyl transferase) subunit, Accession No. p07016 (404 amino acids) [1,22,23,117]. BLAST results were analyzed both in silico and manually. Putative identifications were first based on high scores in terms of ‘‘hits’’ to the sequence being sought. Score bits were computed from the raw scores using the blosum62 substitution matrix [49]. Entry into further rounds of analysis was restricted to those hits whose score exceeded 150 bits, where the ‘‘Expect’’ probability value (E-value; [48]) was low, and which were equivalent to at least half the probe in physical length. All the putative primary amino acid sequences identified in this way were subjected to manual analysis involving the identification of the conserved residues theoretically serving biological functions. The criteria for the identification of the E2 subunit were more rigorous and relied on the identification of the E1 subunit as a prerequisite. Initially criteria analogous to those for the E1 subunit were used, but the A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 343 Table 1 Organisms whose genome sequences were used to study genes of the a-ketoglutarate dehydrogenase complex Organism Genome status Source Escherichia coli K12 strain MG1655 Escherichia coli O157:H7 Bacillus subtilis Bacillus halodurans strain C-125 Nitrosomonas europaea ATCC25978 Methylococcus capsulatus Complete Complete Complete Complete Complete Incompleteb Methylobacterium extorquens strain AM1a Ralstonia metallidurans strain CH34a Acidithiobacillus ferrooxidans ATCC23270 Prochlorococcus marinus strain MED4 Nostoc punctiforme ATCC29133 Synechocystis sp. strain PCC6803 Synechococcus sp. Aquifex aeolicus strain VF5 Methanobacterium thermoautotrophicum DeltaH Methanococcus maripaludis strain LL Methanococcus jannaschii DSM2661 Methanosarcina barkeri strain Fusaro Methanosarcina mazei strain GOE1 Incomplete Incomplete Incomplete Incompletec Incomplete Complete Incomplete Complete Complete Incomplete Complete Incomplete Incomplete Blattner et al. [9] Perna et al. [98] Kunst et al. [76] Takami et al. [123] Chain et al. [12] The Institute for Genomic Research and University of Bergen, Norway University of Washington University of Washington The Institute for Genomic Research Joint Genome Institute Joint Genome Institute Kaneko et al. [64] Joint Genome Institute Deckert et al. [25] Smith et al. [116] University of Washington Bult et al. [13] Joint Genome Institute G€ ottingen Genomics Laboratory Completed sequences, sequences in preparation, and experimental and unfinished genomic sequences may be accessed at http://www.tigr.org/tbd/ mdb/complete.html, http://www.tigr.org/tbd/mdb/inprogress.html and at http://pedant.gsf.de. a Facultative methylotrophs. b Completed subsequent to our analyses [105]. c Completed subsequent to our analyses, but not yet fully edited (http://www.ncbi.nlm.nih.gov: taxonomy ID 243233). gene was also required to map adjacent to the E1 gene within the physical genomic sequence. Subsequent to identifying the likely E1 and E2 sequences in target organisms, the genomic sequences of those organisms corresponding to the BLAST hit locations were recovered as open reading frames (ORFs) by using the NCBI ORF finder (http://www.ncbi.nlm.nih. gov/gorf.html). This yielded the putative coding sequences corresponding to the E1 and E2 subunits of the target a-ketoglutarate dehydrogenase genes within the genomes of the target organisms. These ORFs were subjected to a translation procedure using the expert protein analysis system (ExPASy) proteomics server (http://ca.expasy.org/tools/dna.html) to produce the amino acid sequences encoded by the ORFs. The Unixbased Genetic Computer Group BESTFIT pairwise alignment procedure was used to compare these to the translated sequences from E. coli. The molecular masses and isoelectric points (pI) of the translated sequences were calculated for comparison with published data for a-ketoglutarate dehydrogenase (http://ca.expasy.org/ tools/pi_tool.html). 4.4. Evaluation of the validity of the probing technique Control searches using the E. coli E1 and E2 amino acid sequences (p07015 and p07016) were undertaken against the E. coli 0157:H7 and B. subtilis genomes. For E. coli the BLAST search provided the expected positive hits, with p07015 scoring 1896 bits, and p07016 scoring 735 bits (E-values ¼ 0.0). An analogous BLAST search of the NCBI microbial genome database was conducted using the E. coli 0157:H7 sucA and sucB gene sequences. The search with sucA returned hits only to the E1 subunit of a-ketoglutarate dehydrogenase (e.g., for Pseudomonas putida, AAC23516, score ¼ 2864 bits). The search with sucB returned a variety of hits including the entire a-ketoglutarate dehydrogenase family and the 2-oxoisovalerate dehydrogenase of Bacillus halodurans (NP_243627). When the B. subtilis genome was probed with p07015 and p07016, the expected odhA (E1) and odhB (E2) genes were successfully located (with scores of 595 and 341, respectively). The conserved biomarkers were also identified within the hit sites. Secondary hits were also obtained for the odhB BLAST (score 237 bits) with B. subtilis, but these were identified as the corresponding subunit of 2-oxoisovalerate dehydrogenase in each of B. subtilis and Staphylococcus aureus (NP_372039; score 601) and the E2 subunit of the pyruvate dehydrogenase of B. halodurans (NP_241081;score 642). The identification criteria of high score, low E-value and detection of the conserved biomarkers were thus not necessarily adequate for high confidence identification of the E1 and E2 subunits of a-ketoglutarate dehydrogenase across species. BLAST searches with the B. subtilis odhA (E1) did, however, give exclusive hits to the E1 sequences in other organisms, ranging from B. halodurans (NP_243072; score 3293) to Homo sapiens (BAA01393; score 1345). 344 A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 4.5. Refinement of the identification criteria for the a-ketoglutarate dehydrogenase E2 subunit sequence In B. subtilis, the odhA and odhB genes are contiguous on the genome (genomic locations: 2,108,043–2,110,856 and 2,106,760–2108,013 bp, respectively). In contrast, the secondary hit by the p07016 (E2) BLAST probing of B. subtilis (the 2-oxoisovalerate dehydrogenase E2 sequence) was located remotely from the a-ketoglutarate dehydrogenase genes, at 2,497,143–2,496,061 bp (bfmBaa gene) on the genome. We therefore decided to add a further criterion to be met in the identification procedure, namely a gene position parameter, requiring close genomic proximity of the E1 and E2 coding sequences, as occurs in E. coli and B. subtilis. Inclusion of the positional criterion made high confidence inter-species E1 + E2 identification feasible. 4.6. BLAST probing of target genomes for E1 and E2 genes Probing the genomes of a diverse range of organisms with the E. coli sequences for E1 (p07015) and E2 (p07016) gave strong positive as well as negative results (Table 2). Strong positive results were given for the heterotrophs E. coli and B. subtilis, the facultative methylotroph Ralstonia metallidurans, the obligate methanotroph M. capsulatus and obligate chemolithoautotroph N. europaea (Table 2). Negative results or very poor scores were found for Prochlorococcus, Synechococcus, Methanobacterium thermoautotrophicum, Methanococcus jannaschii and Aquifex aeolicus, which are thus unlikely to contain the E1 gene. The complete genome sequence of Prochlorococcus marinus SS120 was also found to lack genes for E1 and E2, and for the succinyl-CoA genes [26]. Our probing also confirms the genome analysis by Smith et al. [116], who failed to find the gene for a-ketoglutarate dehydrogenase in M. thermoautotrophicum, although other Krebs cycle enzyme genes were identified. The only a-ketoacid dehydrogenase complex component identified in M. jannaschii was the common E3 subunit, dihydrolipoamide dehydrogenase (EC 1.8.1.4; [13]; located at 567,071–564,878 bp; http://pedant.gsf.de). Only a low score for E1 was seen with the facultative methylotroph Methylobacterium extorquens (Table 2), which was surprising as this organism does contain a well-defined E1 gene (NCBI Accession No. AF497852). This probably indicates a weakness of the technique which may be due to relative sequence identities of the genes from different sources (see data in Table 2). It is also possible that our inclusion of positional criteria, while giving higher confidence identification, may result in false negatives if the positional criteria are unjustified in some cases. The very poor score for the obligate chemolithoautotroph Acidithiobacillus ferrooxidans (Table 2) showed that the E1 sequence was absent from the unfinished genome sequence. Table 2 BLAST probing for the presence of the a-ketoglutarate dehydrogenase E1 and E2 genes in the genomes of a range of heterotrophic, facultatively methylotrophic, obligately autotrophic or methanotrophic bacteria, and methanogenic Archaea Organism Escherichia coli 0147:H7 Bacillus subtilis Ralstonia metallidurans Methylobacterium extorquensa Acidithiobacillus ferrooxidans Aquifex aeolicus Methanobacterium thermoautotrophicum Methanococcus jannaschii Methanococcus maripaludis Methanosarcina barkeri Methanosarcina mazei Nostoc punctiforme Prochlorococcus marinus Synechococcus sp. Synechocystis sp. Nitrosomonas europaeab Methylococcus capsulatus E1 gene E2 gene Score (bits) E-value Score (bits) E-value 1896 595 450 37 29 0 32 0 28 31 32 31 0 0 56 1011 945 0 e171 e130 0.007 7.8 0 0.099 0 9.8 7.7 0.4 1.1 0 0 5.1 0 0 735 341 430 99 108 69 26 66 nt 31 375 130 129 128 269 401 407 0 7e95 e121 e121 4e24 1.1 2.7 4.5 nt 3 e100 6e31 3e31 3e31 7.3e63 e113 e114 The TBLASTN programme (http://www.ncbi.nlm.nih.gov/blast) used the SwissProt database sequences for E1 (p07015) and E2 (p07016) from E. coli as probes against the genome listed. nt, not tested. a BLAST2 comparison of the gene for KGD from M. extorquens (AF497852) and E. coli (X00661) showed 80.5% identity for 161/200 bases, corresponding to amino acids 734–800. b BLAST2 comparison of the E1 sequence from N. europaea with the E. coli gene for KGD showed a discontinuous identity of 75% for 486/648 bases, corresponding to 214 amino acids with an identity by BLASTX of 54% to that segment of the E. coli E1 polypeptide. A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 4.7. Positive identification of the E1 and E2 genes in the genomes of N. europaea and M. capsulatus and computed physicochemical properties of the polypeptides encoded by the genes The most striking finding was the demonstration of strong hits for both the E1 and E2 genes for a-ketoglutarate dehydrogenase in both the obligate chemolithoautotroph N. europaea and the obligate methanotroph M. capsulatus, neither of which has a-ketoglutarate dehydrogenase activity in vivo. Subsequent to our finding putative E1 and E2 genes in N. europaea, using the then incomplete genome sequence (http://pedant.gsf.de), the complete genome sequence was published ([16]; NCBI nucleotide database sequence AABB02000001). In their discussion of the genome, Chain and co-authors [16] noted that genes for all enzymes of the Krebs cycle were present, including the E1 and E2 of a-ketoglutarate dehydrogenase. We were concerned that putative identifications of genes for specific functions, using BLAST and other partial alignment methods, are not absolute proof of identity. Consequently we followed through our findings by making total sequence comparisons between the putative E1 and E2 genes of M. capsulatus and N. europaea and the genes from heterotrophs, and compared the translated amino acid sequences and predicted physicochemical properties of the polypeptides encoded by the target genes. The BLAST hits for the E1 and E2 genes were used to locate and extract open reading frames (ORFs) from their ATG start codons (Table 3). Translated amino acid sequences of the ORFs were obtained by in silico 345 translation of the gene sequences and were aligned against the corresponding E. coli p07015 (E1) and p07016 (E2) amino acid sequences to assess sequence homology and biomarker conservation. The pairwise alignments using GCG BESTFIT confirmed that the E1 and E2 genes of N. europaea and M. capsulatus had indeed been detected and identified correctly (Table 4). The polypeptide sequences obtained from the translated sequences of the E1 and E2 genes in the genomes of N. europaea and M. capsulatus were used to compute their molecular masses and isoelectric points (pI; http:// ca.expasy.org/tools/pi_tool.html), and to compare these features with those of the functional E1 and E2 subunits of the E. coli a-ketoglutarate dehydrogenase (Table 5). The polypeptides encoded by the E1 and E2 genes from all three organisms had very similar properties, showing that the obligate autotroph and obligate methanotroph appeared genetically competent to produce a-ketoglutarate dehydrogenase, but do not do so in vivo. Using the completed genome sequence of N. europaea ATCC 25,978 (at http://pedant.gsf.de/cgi-bin/wwwfly.pl?Set ¼ Nitrosomonas_europaea_ATCC_25978& Page ¼ index) enabled us to confirm the findings of our indirect analyses. The coding sequences for the E1 and E2 components of a-ketoglutarate dehydrogenase were positively identified in the genome (orf431 and orf430). The E1 gene was located at contig positions 420,453– 423,317 bp, comprising 955 amino acids with a molecular mass of 108,487.0 and pI 5.9; while the E2 gene was located at contig positions 419,154–420,428 bp, comprising 425 amino acids with a molecular mass of 46,188.9 and pI 5.5. These values are comparable to the Table 3 Locations of the open reading frames for the a-ketoglutarate dehydrogenase E1 and E2 genes in the genomes of M. capsulatus and N. europaea and the derived amino acid content of the polypeptides they encode Organism Subunit Contig location (bp) Number of amino acids coded Contig accession Methylococcus capsulatus E1 E2 29841–32655 32620–33766 937 381 gn1jTIGR_414jmcapsul_bmc_79 gn1jTIGR_414jmcapsul_bmc_79 Nitrosomonas europaea E1 E2 6952–9751 9772–11050 932 425 microbe1_fasta.screen.Contig453 microbe1 fasta screen. Contig453 The open reading frames contain the TGA (stop) codon and the amino acid count excludes this non-coding codon. Table 4 BESTFIT pairwise alignment statistics for the translated a-ketoglutarate dehydrogenase E1 and E2 open reading frames identified in M. capsulatus and N. europaea against the a-ketoglutarate dehydrogenase E1 and E2 sequences of E. coli (p07015 and p07016) Organism Subunit Number of nucleotides in deduced orf Translated sequences Similarity (%) Identity (%) Methylococcus capsulatus E1 E2 2813 1143 63.0 68.8 54.5 57.5 Nitrosomonas europaea E1 E2 2796 1275 65.0 65.3 56.5 56.6 For both organisms the biomarkers (E. coli numbering) were well conserved for E1 between amino acids 362 and 405 (75% and 69% identity for M. capsulatus and N. europaea, respectively), and 440 and 456 (71% and 75% identity, respectively); and for E2 at lys-43, his-375 and asp-79. 346 A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 Table 5 Computed physicochemical properties of the a-ketoglutarate dehydrogenase E1 and E2 polypeptides encoded by the respective genes in E. coli, M. capsulatus and N. europaea located in this study Organism Subunit Molecular mass of encoded polypeptide (Dalton) pI of encoded polypeptide (pH units) Escherichia coli E1 E2 105061.7 43880.2 6.04 5.58 Methylococcus capsulatus E1 E2 105388.4 41553.8 6.06 5.98 Nitrosomonas europaea E1 E2 105923.3 44987.5 5.96 5.60 data obtained by our independent methods (Tables 3 and 5). The closest homologues to the Nitrosomonas E1 database sequence by BLASTP and TBLASTN were the odhA gene (E1) of Ralstonia solanacearum and Ralstonia eutropha, with scores of 1145 and 1138 (E-value ¼ 0.0). By TBLASTN these showed 566/955 (59%) and 519/952 (58%) amino acid sequence identity, respectively, to the N. europaea sequence. Identity to E. coli E1 (X00661) was 517/948 (54%), which was comparable to the identity of the E. coli E1 sequence to those of Xanthomonas (NP_641868.1) and Xylella (NP_778980.1) at 55% (526/ 941 and 521/937, respectively). TBLASTN amino acid sequence identity of the N. europaea E1 sequence to that of Methylobacterium extorquens was lower, at 429/990 (43%). The closest homologue of the Nitrosomonas E2 sequence was the dihydrolipoamide succinyltransferase of R. eutropha. It is also noteworthy that the genes encoding both the a- and b-chains of succinyl-CoA synthetase occur in the N. europaea genome (Scaffold_1 contig: orf128 and orf129; http://pedant.gsf.de). The presence of this enzyme, which normally functions to produce succinate from succinyl-CoA (the product of a-ketoglutarate dehydrogenase activity), has not been reported in Nitrosomonas, but clearly merits further study. 4.8. A basis for obligate autotrophy and methanotrophy? The reason for the failure of N. europaea and M. capsulatus to synthesize functional a-ketoglutarate dehydrogenase and the significance of this to their obligate physiologies remain to be established. Among the possibilities are: (i) failure to transcribe one or both of the genes into mRNA, (ii) deletion of regulatory parts of the operons, (iii) that the obligate nature of their central metabolism results from regulatory controls causing permanent repression of the genes, or (iv) some kind of post-transcriptional gene silencing, possibly by endogenous small interfering RNAs [17,131]. A parallel is seen in the facultative autotroph and methylotroph, P. versutus strain A2, when autotrophic and heterotrophic growth are compared. Both a-ketoglutarate dehydrogenase and succinyl-CoA synthetase are completely repressed during chemolithotrophic growth with thiosulfate, whereas high activities occur in heterotrophically grown bacteria [97]. During autotrophic growth of P. versutus on thiosulfate or formate, 14 C-acetate is incorporated heavily into leucine, glutamate, proline and arginine, but when grown heterotrophically, increased labelling of aspartate, alanine and other amino acids occurs in cultures ‘‘spiked’’ with [14 C]acetate [137]. Transcriptional repression of the expression of genes for a-ketoglutarate dehydrogenase in response to anaerobiosis is well known in enterobacteria [2,10,93], and a permanently ‘‘switched off’’ state may exist in obligate autotrophs. Alternatively, expression may occur but is accompanied by deficiencies at the translational or polypeptide assembly levels, resulting in failure to produce enzymatically active a-ketoglutarate dehydrogenase protein. Such a situation appears to occur in the archaeon Haloferax volcanii where an insert in the E2 gene of the pyruvate dehydrogenase complex appeared to be the cause of failure of the complex to assemble to produce a functional enzyme [61]. Another possible situation is represented by the ushB gene of Salmonella (encoding a UDP–sugar-hydrolase), which has an orthologue, ushB(c) in E. coli, which does not produce this enzyme activity. In fact the ushB(c) gene shares 100% identity with the E. coli gene for CDP-diglyceride hydrolase (cdh) but the ushB(c) gene does not produce any detectable protein in vivo despite being transcribed normally [110]. The possibility thus remains that the sequences found in obligate autotrophs and methanotrophs, seemingly encoding the E1 of a-ketoglutarate dehydrogenase may in fact not even be capable of that function if expressed. The corollary of course is that in some species in which active a-ketoglutarate dehydrogenase is found, the protein catalyzing the reactions may be structurally unrelated to the E. coli enzyme, and encoded by genes of analogous rather than homologous function, as has been extensively demonstrated by Galperin et al. [40]. Identifying which, or what combination, of these possibilities is responsible may require an extensive and multi-targeted approach. A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 5. Testing hypotheses about the nature of obligate autotrophy and methanotrophy and the absence of a-ketoglutarate dehydrogenase activity 5.1. Detecting the diagnostic coding sequences for a-ketoglutarate dehydrogenase in obligate autotrophs and methanotrophs: bioinformatic approaches To date, the number of obligate species (including methanogens) to which it has been possible to apply in silico methods to seek the E1 and E2 sequences is very limited. As more complete genome sequences become available it will be possible to screen these to see whether other genera also possess genes which do not translate in vivo into active enzyme. When such sequences are found, the same diagnostic procedures described above for Nitrosomonas and Methylococcus can be applied to detect key amino acid sequences in the translated polypeptides and to predict their molecular masses and pI values. In the short term the genomes of A. ferrooxidans, A. aeolicus, Synechocystis PCC6803, Synechococcus, Nostoc, Prochlorococcus, Methanoccous spp. and Methanosarcina spp., which all appear to lack the E1 gene (Table 2), will be of particular interest. Some of the genomes which are unfinished (Table 1) may eventually be shown to contain the E1 coding sequence. The unfinished genome of A. ferrooxidans appears to have the E2 coding sequence (Table 2) but apparently not that for E1, although these two genes would be expected to be contiguous on the genome (see Section 4.5). The E1 sequence is probably absent from A. ferrooxidans, as we have used BLASTN to probe the essentially complete genome ([128]; 3164654 bp) using the E1 nucleotide sequence (bases 2,580,584–2,583,451 of the complete genome of N. europaea), and found no significant sequence similarity. The search for E1 and E2 coding sequences, using specific gene probes, needs to be extended to genera such as Thiobacillus, Thermithiobacillus, Halothiobacillus, Methylomonas and Methylobacter as well as facultative genera and organisms such as Nitrobacter and Methylosinus, which can produce a functional a-ketoglutarate dehydrogenase enzyme. 5.2. Is there a functional transcription product in obligate genera with an E1 coding sequence: is there an mRNA? Having located the E1 and E2 genes within the genome, sequencing of the contiguous regions of DNA may give information about the regulatory elements associated with the genes. Failure to transcribe the genes could result from: (1) a missing promoter or one that is shared with and overridden by another gene or (2) the gene being in some way ‘‘faulty’’, possibly resulting in an untranslatable or nonsense polypeptide or (3) the absence or a transcriptional failure of an essential chaperone gene. If there is a regulatory failure such as deletion 347 of part of an operon, then no mRNA would be produced. This could be tested using probes from the known Escherichia and Nitrosomonas sequences and standard techniques (Northern blots, labelled probe hybridization) used to seek transcriptional mRNA products. If mRNA were to be produced, amplification of the cDNA using revere transcriptase (RT)-PCR would be a powerful means of identifying and analyzing the product. 5.3. Is there an enzymically inactive translation product from E1 and E2 genes in obligate genera? If the genes appeared to be transcribed into mRNA, standard methods could be applied to seek polypeptide products based on the predicted properties of these products, and by using antibodies raised against homologous polypeptides. Fractionation of cell-free extracts to concentrate polypeptides of the expected molecular mass followed by two-dimensional electrophoresis [138,139] using bacteria such as Methylococcus and Nitrosomonas (with E. coli or Ralstonia as controls) might reveal polypeptides of the predicted size and charge, which could be subjected to N-terminal sequencing and to fragmentation pattern analysis and internal amino acid sequencing by mass spectrometric methods. A combination of in silico prediction from the gene sequences of the likely secondary and tertiary folding of the polypeptides and direct analysis of the actual translated product and its active site(s) might reveal that the gene product could not function in vivo as part of an active a-ketoglutarate dehydrogenase. 5.4. Expression of the E1 and E2 genes of obligate genera in heterotrophic hosts and vice versa Using a recipient such as E. coli (which is capable of fermentative growth without use of the Krebs cycle) from which the E1 and E2 genes of a-ketoglutarate dehydrogenase had been deleted, the corresponding genes from an obligate genus could be inserted and their expression sought. This should restore the ability of the mutant E. coli to grow aerobically with a functional Krebs cycle. Alternative recipients could be other gram-negative bacteria capable of facultative fermentative growth. If this approach worked, it would enable subsequent overexpression of the autotroph genes and production of quantities of protein for further analysis. It would also give leads into the regulatory factors missing or controlling expression in autotrophs. In this context, the citrate synthase gene from Sulfolobus solfataricus was expressed at high levels when introduced into E. coli [20]. A parallel approach could be to introduce the E1 and E2 genes of E. coli or another heterotroph (such as Paracoccus, Methylobacterium or Ralstonia) into an obligate species such as Methylococcus or Halothiobacillus to see whether they could be transcribed and 348 A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 translated into functional a-ketoglutarate dehydrogenase, and whether the presence of active enzyme allowed heterotrophic growth on any organic substrates. Such an approach would be essential in establishing whether an overriding regulatory control existed that prevented the expression of the introduced genes. That such an approach might have success is exemplified by the succesful introduction of the gene for phosphofructokinase into A. thiooxidans [127]. The phenotype of such recombinants would be of considerable interest, such as in assessing stimulatory or deleterious effects of the inserted genes on methanotrophic or chemolithotrophic growth, both in the presence and absence of added nutrients such as Krebs cycle intermediates. 6. Concluding remarks The dual purposes of this review have been (1) to emphasize how bioinformatics can be exploited to help solve and provide new approaches to unresolved metabolic problems and (2) to stimulate the interest of ‘‘modern molecular microbiologists’’ in a neglected area of microbiology: the causes of obligate autotrophy and methanotrophy. We have stressed studies on a-ketoglutarate dehydrogenase but acknowledge that absence of this functional enzyme may not be the only, or even most important, factor in bestowing obligate metabolism. This was recognized by Whittenbury and Kelly [137] in 1977, when they noted that the ‘‘absence of 2-oxoglutarate dehydrogenase in itself does not impose obligate autotrophy, but obligate autotrophy can result if the Ômetabolic environmentÕ into which a lesion such as lack of this enzyme in introduced is such that no alternative metabolic process can be growth-supporting’’. It may transpire that a number of genes for central metabolic pathways essential for heterotrophic growth and some transport processes are commonly absent from, or are present and not expressed in, obligate genera, and that obligate autotrophy is imposed by the ‘‘concerted absence’’ of these functions. In the present state of knowledge we have concentrated on the historical candidate, a-ketoglutarate dehydrogenase, as a model for the investigation of some obligate species. a-Ketoglutarate dehydrogenase is a strong contender for a role in obligate autotrophy as its apparent absence is widespread among chemolithoautotrophs and methanotrophs throughout the phylogenetic tree. Most such a-ketoglutarate dehydrogenase-negative species do not seem to lack other enzymes of the Krebs cycle or of other central metabolic pathways. A parallel line of enquiry needs to be undertaken with the facultative autotrophs and methylotrophs such as Paracoccus and Methylobacterium in which a-ketoglutarate dehydroge- nase activity is repressible during growth on C1 -compounds, but in which little is known about the mechanism of repression. Understanding this could throw light on the obligate autotrophy phenomenon. It may be that the regulation of autotrophic metabolism in obligate genera is much more complex than the simple regulation of one or two genes, and involves cross-regulatory processes from their central carbon pathways (the Calvin or Quayle cycles). This idea requires an answer to the question ‘‘can genomics and proteomics ever tell us that an obligate species is so because a gene or gene product is non-functional, or that it is obligate because there is a higher level of genomic organization that prevents expression of genes such as a-ketoglutarate dehydrogenase?’’. That is to say, is obligate autotrophy determined or serendipitous? Once the complete genome sequences are known for a wider range of obligate chemo- and photo-lithoautotrophs and methanotrophs, it is likely that they will fall into two categories: those with genes encoding a-ketoglutarate dehydrogenase and other central metabolic enzymes, and those without. The explanation of obligate metabolism in the latter group will require survey of their overall metabolism to see if lack of specific functional enzymes is the prime cause. Absence of the genes raises the question of why they are lacking: does the organism come from an ancestral line which never possessed them or were they lost by a process of evolutionary reductionism, akin to that proposed for mycoplasmas? It may be that at least some obligate autotrophs may have evolved by the progressive loss of non-essential or potentially deleterious genes. A trend seems to be emerging to indicate that obligate chemolithoautotrophs do indeed possess small to medium-sized genomes. Examples are A. ferrooxidans, A. thiooxidans and Leptospirillum ferrooxidans with chromosomes that are 2.29–3.33 (eight strains), 3.14 and 1.90 Mb in size [4,57,128]. A. thiooxidans also contains five plasmids totalling 475 kb, indicating a maximum genome size of 3.62 Mb. Similarly the chromosome of M. capsulatus comprises about 3.31 Mb [129]. While these genome sizes are far greater than those of metabolically fastidious symbionts and pathogens such as Buchnera (0.45 Mb) and Mycoplasma (0.58 Mb), they are significantly smaller than those of heterotrophs including E. coli (4.64 Mb), B. subtilis (4.21 Mb) and Streptomyces (8.66 Mb) [43,78,85]. The genomes of strains of the cyanobacterium Prochlorococcus (the smallest oxygen-evolving phototroph known) are also very small at 1.66–2.4 Mb [26,105]. The genome of N. europaea is also compact (2.8 Mb) and Chain et al. [16] proposed that is indicative that a small genome is sufficient to support an obligate autotroph. They also suggest that N. europaea may be continuing to evolve a ‘‘more compact genome’’ by loss of functional genes [16]. Gene inactivation and subsequent erosion and deletion may be the major force that maintains relatively small A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 genome size in bacteria and selects against insertion of laterally transferred genes unless these are selectively advantageous [85]. In the case of a-ketoglutarate dehydrogenase in obligate genera, its permanent repression or deletion would be selectively advantageous in avoiding lethal or futile carbon cycling during autotrophic growth. An intriguing question is how the selective pressure has been exerted to prevent retention or expression of functional a-ketoglutarate dehydrogenase genes, possibly acquired repeatedly by lateral transfer, during evolutionary time, given that horizontal DNA transfer appears to have been (and continues to be) a major means for the acquisition of genes by bacteria [143]. Thus, in that it is estimated that 17.6% of the genes of E. coli have been acquired by horizontal transfer [11], a comparable proportion could have arisen in gram negative obligate autotrophs and methanotrophs such as Nitrosomonas and Methylococcus. This may be indicated by a BLAST search using the N. europaea genome which showed 12% and 31% of its predicted open reading frames also to be present, respectively, in Pseudomonas aeruginosa and R. solanacearum [16]. Equally significant is the likelihood of gene acquisition from archaea: as many as 24% of the genes of Thermotoga are reportedly archaeal [12]. While modern autotrophs and methanotrophs may be subject to a constant flux of gain and loss of environmental DNA, their fundamental genetic makeup may have been established early in their evolutionary history, when horizontal gene transfer might also have played a significant role [44]. Acknowledgements Some of this work used preliminary sequence data obtained from The Institute of Genomic Research website at http://tigr.org. We thank Phil Cunningham for advice on Unix-based computer systems. We thank Noel Carr, Colin Murrell and Ian McDonald (Warwick) for valuable critical comment on drafts of the review, and anonymous reviewers for helpful criticism. References [1] Altschul, S.F., Madden, T.L., Schaffer, A.A., Zhang, J., Miller, W. and Lipman, D.J. (1997) Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25, 3389–3402. [2] Amarasingham, C.R. and Davis, B.D. (1965) Regulation of a-ketoglutarate dehydrogenase formation in Escherichia coli. J. Biol. Chem. 240, 3664–3668. [3] Amann, R. (2000) Who is out there. Microbial aspects of biodiversity. Syst. Appl. Microbiol. 23, 1–8. [4] Amils, R., Irazabel, N., Moreira, D., Abad, J.P. and Marin, I. (1998) Genomic organization of acidophilic chemolithotrophic bacteria using pulsed field gel electrophoretic techniques. Biochimie 80, 911–921. 349 [5] Baxter, N.J., Scanlan, J., De Marco, P., Wood, A.P. and Murrell, J.C. (2002) Duplicate copies of the genes encoding methanesulfonate monooxygenase in Marinosulfonomonas methylotropha strain TR3 and detection of methanesulfonate utilizers in the environment. Appl. Environ. Microbiol. 68 (1), 289–296. [6] Baxter, N.J., Hirt, R.P., Bodrossy, L., Kovacs, K.L., Embley, T.M., Prosser, J.I. and Murrell, J.C. (2002) The ribulose-1,5bisphosphate carboxylase/oxygenase gene cluster of Methylococcus capsulatus (Bath). Arch. Microbiol. 177, 279–289. [7] Beudeker, R.F., Kerver, J.W.M. and Kuenen, J.G. (1981) Occurrence, structure and function of intracellular polyglucose in the obligate chemolithotroph Thiobacillus neapolitanus. Arch. Microbiol. 129, 221–226. [8] Beudeker, R.F., de Boer, W. and Kuenen, J.G. (1981) Heterolactic fermentation of intracellular polyglucose by the obligate chemolithotroph Thiobacillus neapolitanus under anaerobic conditions. FEMS Microbiol. Lett. 12, 337–342. [9] Blattner, F.R. and 16 co-authors (1997) The complete genome sequence of Escherichia coli K-12. Science 277, 1453–1474. [10] Bott, M. (1997) Anaerobic citrate metabolism and its regulation in enterobacteria. Arch. Microbiol. 167, 78–88. [11] Buchanan, F. (2002) Microbial genomes and DNA exchange. In: Lateral DNA Transfer, Mechanism and Consequences (Cold Spring Harbor Laboratory), pp. 129–169. Cold Spring Harbor Laboratory Press. [12] Buchanan, F. (2002) DNA transfer among the domains of life. In: Lateral DNA Transfer, Mechanism and Consequences (Cold Spring Harbor Laboratory), pp. 365–385. Cold Spring Harbor Laboratory Press. [13] Bult, G.J. and 23 co-authors (1996) Complete genome sequence of the methanogenic archaeon, Methanococcus jannaschii. Science 273, 1057–1073. [14] Carlsson, P. and Hedersted, T.L. (1989) Genetic characterization of Bacillus subtilis odhA and odhB, encoding 2-oxoglutarate dehydrogenase and dihydrolipoamide transsuccinylase, respectively. J. Bacteriol. 171, 3667–3672. [15] Cavalier-Smith, T. (2002) The neomuran origin of archaebacteria, the negibacterial root of the universal tree and bacterial megaclassification. Int. J. Syst. Evol. Microbiol. 52, 7–76. [16] Chain, P. and 14 co-authors (2003) Complete genome sequence of the ammonia-oxidizing bacterium and obligate chemolithoautotroph Nitrosomonas europaea. J. Bacteriol. 185, 2759–2773. [17] Chi, T.T., Chang, H.Y., Wang, N.N., Chang, D.S., Dumphy, N. and Brown, P.O. (2003) Genomewide view of gene silencing by small interfering RNAs. Proc. Nat. Acad. Sci. USA 100, 6343–6346. [18] Christen, P. and Mehta, P.K. (2001) From cofactor to enzymes. The molecular evolution of pyridoxal-50 -phosphate-dependent enzymes. Chem. Record 1, 436–447. [19] Christoserdova, L., Chen, S.-W., Lapidus, A. and Lidstrom, M.L. (2003) Methylotrophy in Methylobacterium extorquens AM1 from a genomic point of view. J. Bacteriol. 185, 2980–2987. [20] Connaris, H., West, S.M., Hough, D.W. and Danson, M.J. (1998) Cloning and overexpression in Escherichia coli of the gene encoding citrate synthase from the hyperthermophilic archaeon Sulfolobus solfataricus. Extremophiles 2, 61–66. [21] Danson, M.J., Black, S.C., Woodland, D.L. and Wood, P.A. (1985) Citric acid cycle enzymes of the Archaebacteria – citrate synthase and succinate thiokinase. FEBS Lett. 179, 120–124. [22] Darlison, M.G., Spencer, M.E. and Guest, J.R. (1984) Nucleotide sequence of the sucA gene encoding the 2-oxoglutarate dehydrogenase of Escherichia coli K-12. Eur. J. Biochem. 141, 351–359. [23] Dastoor, F.P., Forrest, M.E. and Beatty, J.T. (1997) Cloning, sequencing and oxygen regulation of the Rhodobacter capsulatus 350 [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 a-ketoglutarate dehydrogenase operon. J. Bacteriol. 179, 4559– 4566. Davey, J.F., Whittenbury, R. and Wilkinson, J.F. (1972) The distribution in the methylobacteria of some key enzymes concerned with intermediary metabolism. Arch. Mikrobiol. 87, 359–371. Deckert, G. and 14 co-authors (1998) The complete genome of the hyperthermophilic bacterium Aquifex aeolicus. Nature 392, 353–358. Dufresne, A. and 20 co-authors (2003) Genome sequence of the cyanobactertium Prochlorococcus marinus SS120, a nearly minimal oxyphototrophic genome. Nature 100, 10021–10025. Eccleston, M. and Kelly, D.P. (1972) Assimilation and toxicity of exogenous amino acids in the methane-oxidizing bacterium Methylococcus capsulatus. J. Gen. Microbiol. 71, 541–554. Eccleston, M. and Kelly, D.P. (1973) Assimilation and toxicity of some exogenous C-1 compounds, alcohols, sugars and acetate in the methane-oxidizing bacterium Methylococcus capsulatus. J. Gen. Microbiol. 75, 211–221. Eccleston, M. and Kelly, D.P. (1973) Inhibition by L -threonine of aspartokinase as a cause of threonine-toxicity to Methylococcus capsulatus. J. Gen. Microbiol. 75, 223–226. Ferry, J.G. (1995) Carbon monoxide dehydrogenase. Ann. Rev. Microbiol. 49, 305–333. Ferry, J.G. (1999) Enzymology of one-carbon metabolism in methanogenic pathways. FEMS Microbiol. Rev. 23, 13–38. Ferry, J.G. (2003) One-carbon metabolism in methanogenic anaerobes. In: Biochemistry and Physiology of Anaerobic Bacteria (Ljungdahl, L.G., Adams, M.W., Barton, L.L., Ferry, J.G. and Johnson, M.K., Eds.), pp. 143–156. Springer-Verlag, New York. Finster, K., Tanimoto, Y. and Bak, F. (1992) Fermentation of methanethiol and dimethylsulfide by a newly isolated methanogenic bacterium. Arch. Microbiol. 157, 425–430. Fuchs, G. (1989) Alternative pathways of autotrophic CO2 fixation. In: Autotrophic Bacteria (Schlegel, H.G. and Bowien, B., Eds.), pp. 365–382. Science Tech Publishers and SpringerVerlag, New York. Fuchs, G. (1994) Variations of the acetyl-CoA pathway in diversely related microorganisms that are not acetogens. In: Acetogenesis (Drake, H.L., Ed.), pp. 506–538. Chapman & Hall, New York. Fuchs, G. and Stupperich, E. (1984) CO2 reduction to cell carbon in methanogens. In: Microbial Growth on C1 Compounds (Crawford, R.L. and Hanson, R.S., Eds.), pp. 199–202. American Society for Microbiology, Washington, DC. Fukuda, E. and Wakagi, T. (2002) Substrate recognition by 2oxoacid:ferredoxin oxidoreductase from Sulfolobus sp. strain S7. Biochim. Biophys. Acta 1597, 74–80. Fukuda, E., Kino, H., Matsuzawa, H. and Wakagi, T. (2001) Role of highly conserved YTITP motif in 2-oxoacid:ferredoxin oxidoreductase. Eur. J. Biochem. 268, 5639–5646. Galperin, M.Y. and Koonin, E.V. (1999) Functional genomics and enzyme evolution – homologous and analagous enzymes encoded in microbial genomes. Genetica 106, 159– 170. Galperin, M.Y., Walker, D.R. and Koonin, E.V. (1998) Analagous enzymes: independent inventions in enzyme evolution. Genome Res. 8, 779–790. Gest, H. (1981) Evolutionary roots of the citric acid cycle in prokaryotes. Biochem. Soc. Symp. 54, 3–16. Gest, H. (2003) Anaerobes in the recycling of elements in the biosphere. In: Biochemistry and Physiology of Anaerobic Bacteria (Ljungdahl, L. and 4 co-Eds.), pp. 1–10. Springer, New York. Gil, R., Sabater-Mu~ noz, B., Latorre, A., Silva, F.J. and Moya, A. (2002) Extreme genome reduction in Buchnera spp.: towards [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] the minimal genome needed for symbiotic life. Proc. Natl. Acad. Sci. USA 99, 4454–4458. Gogarten, J.P., Hilario, E. and Olendzenski, L. (1996) Gene duplication and horizontal gene transfer during early evolution. In: Evolution of Microbial Life (Roberts, D.McL., Sharp, P., Alderson, G. and Collins, M.A., Eds.), pp. 268–292. Cambridge University Press. Gottschal, J.C., de Vries, S. and Kuenen, J.G. (1979) Competition between the facultatively chemolithotrophic Thiobacillus A2, an obligately chemolithotrophic Thiobacillus, and a heterotrophic Spirillum for inorganic and organic substrates. Arch. Microbiol. 121, 241–249. Gottschal, J.C., Nanninga, H. and Kuenen, J.G. (1981) Growth of Thiobacillus A2 under alternating conditions in the chemostat. J. Gen. Microbiol. 126, 85–96. Green, L.S., Emerich, D.W., Bergerson, F.J. and Day, D.A. (2000) Catabolism of alpha-ketoglutarate by a sucA mutant of Bradyrhizobium japonicum: evidence for an alternative tricarboxylic acid cycle. J. Bacteriol. 182, 2838–2844. Gribskov, M. and Devereux, J. (1991) Sequence Analysis Primer. Stockton Press, New York. pp. 140–141. Henikoff, S. and Henikoff, J.G. (1992) Amino acid substitution matrices from protein blocks. Proc. Natl. Acad. Sci. USA 89, 10915–10919. Hipp, W.M., Pott, A.S., ThumSchmitz, N., Faath, I., Dahl, C. and Tr€ uper, H.G. (1997) Towards the phylogeny of APS reductase and sirohaem sulfite reductases in sulfate-reducung and sulfur-oxidizing prokaryotes. Microbiology UK 143, 2891– 2902. Hoare, D.S., Hoare, S.L. and Moore, R.B. (1967) The photoassimilation of organic compounds by autotrophic blue-green algae. J. Gen. Microbiol. 49, 351–370. Holo, H. and Sirev a g, R. (1986) Autotrophic growth and CO2 fixation of Chloroflexus aurantiacus. Arch. Microbiol. 145, 173– 180. Holmes, A.J., Owens, N.J.P. and Murrell, J.C. (1995) Detection of novel marine methanotrophs using phylogenetic and functional gene probes after methane enrichment. Microbiology UK 141, 1947–1955. Hooper, A.B. (1969) Biochemical basis of obligate autotrophy in Nitrosomonas europaea. J. Bacteriol. 97, 776–779. H€ ugler, M., Huber, H., Stetter, K.O. and Fuchs, G. (2003) Autotrophic CO2 fixation pathways in archaea (Crenarchaeota). Arch. Microbiol. 179, 160–173. Hugenholz, P. and Pace, N.R. (1996) Identifying microbial diversity in the natural environment: a molecular phylogenetic approach. Trends Biotechnol. 14, 190–197. Irazabal, N., Marin, I. and Amils, R. (1997) Genomic organization of the acidophilic chemolithoautotrophic bacterium Thiobacillus ferrooxidans ATCC 21834. J. Bacteriol. 179, 1946– 1950. Iwasaki, T., Wakagi, T. and Oshima, T. (1995) Ferredoxindependent redox system of a novel reduced ferredoxinreoxidizing iron–sulfur flavoprotein. J. Biol. Chem. 270, 17878–17883. Jeanthon, C. (2000) Molecular ecology of hydrothermal vent microbial communities. Antonie van Leeuwenhoek 77, 117–133. Jensen, S., Holmes, A.J., Olsen, R.A. and Murrell, J.C. (2000) Detection of methane oxidizing bacteria in forest soil by monooxygenase PCR amplification. Microbial Ecol. 39, 282– 289. Jolley, K.A., Maddocks, D.G., Gyles, S.L., Mullan, Z., Tang, S.L., Dyall-Smith, M.L., Hough, D.W. and Danson, M.J. (2000) 2-Oxoacid dehydrogenase multienzyme complexes in the halophilic Archaea. Gene sequences and protein structural predictions. Microbiology UK 146, 1061–1069. A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 [62] Jordan, I.K., Rogozin, I.B., Wolf, Y.I. and Koomin, E.V. (2002) Microevolutionary genomics of bacteria. Theoret. Pop. Biol. 61, 435–447. [63] Kanagawa, T. and Kelly, D.P. (1986) Breakdown of dimethyl sulphide by mixed cultures and by Thiobacillus thioparus. FEMS Microbiol. Lett. 34, 13–19. [64] Kaneko, T. and 23 co-authors (1996) Sequence analysis of the entire genome of the unicellular Cyanobacterium synechocystis sp. strain PCC6803. DNA Res. 3, 109–136. [65] Kelly, D.P. (1967a) Problems of the autotrophic microorganisms. Sci. Progr. 55, 35–51. [66] Kelly, D.P. (1967) The incorporation of acetate by the chemoautotroph Thiobacillus neapolitanus strain C. Arch. Mikrobiol. 58, 99–116. [67] Kelly, D.P. (1969) Regulation of chemoautotrophic metabolism III. DAHP synthetase in Thiobacillus neapolitanus. Arch. Microbiol. 69, 360–369. [68] Kelly, D.P. (1969) Continued debate on autotrophy. Nature 221, 1198. [69] Kelly, D.P. (1971) Autotrophy: concepts of lithotrophic bacteria and their organic metabolism. Annu. Rev. Microbiol. 25, 177– 210. [70] Kelly, D.P. (1974) Growth and metabolism of the obligate photolithotroph Chlorobium thiosulfatophilum in the presence of added organic nutrients. Arch. Microbiol. 100, 163–178. [71] Kelly, D.P. and Wood, A.P. (2000) The chemolithotrophic prokaryotes. In: The Prokaryotes, An Evolving Electronic Resource for the Microbiological Community (PKIIIE) (Dworkin, M., Falkow, S., Rosenberg, E., Schleiffer, K.-H. and Stackebrandt, E., Eds.). Springer-Verlag, New York (http:// www.prokaryotes.com). [72] Kerscher, L. and Oesterhalt, D. (1981) Purification and properties of two 2-oxoacid:ferredoxin oxidoreductases from Halobacterium halobium. Eur. J. Biochem. 116, 587–594. [73] Klenk, H.-P. and 50 co-authors (1997) The complete genome sequence of the hyperthermophilic, sulphate-reducing archaeon Archaeoglobus fulgidus. Nature 390, 364–370. [74] Kuenen, J.G. and Veldkamp, H. (1973) Effect of organic compounds on growth of chemostat cultures of Thiomicrospira pelophila, Thiobacillus thioparus and Thiobacillus neapolitanus. Arch. Mikrobiol. 94, 173–190. [75] Kuenen, J.G. and Beudeker, R.F. (1982) Microbiology of thiobacilli and other sulphur-oxidizing autotrophs, mixotrophs and heterotrophs. Phil. Trans. Roy. Soc. London B298, 473–497. [76] Kunst, F. and 150 co-authors (1997) The complete genome sequence of the gram-positive bacterium Bacillus subtilis. Nature 390, 249–256. [77] Lopez, P., Casane, D. and Philippe, H. (2002) Bio-informatics (5) – molecular phylogeny and evolution. Med. Sci. 18, 1146–1154. [78] Madigan, M.T., Martinko, J.M. and Parker, J. (2003) Brock – Biology of Microorganisms, 10th ed. Prentice-Hall, Pearson Education International, London. [79] Martin, W. and Russell, M.J. (2003) On the origins of cells: a hypothesis for the evolutionary transitions from abiotic geochemistry to chemoautotrophic prokaryotes, and from prokaryotes to nucleated cells. Phil. Trans. Roy. Soc. B 358, 59–83. [80] Matin, A. (1978) Organic nutrition of chemolithotrophic bacteria. Annu. Rev. Microbiol. 32, 433–469. [81] Matuda, S., Nakano, K., Ohta, S., Saheki, T., Kawanishi, Y. and Miyata, T. (1991) The a-ketoacid dehydrogenase complexes. Sequence similarity of rat pyruvate dehydrogenase with Escherichia coli and Azotobacter vinelandii a-ketoglutarate dehydrogenase. Biochim. Biophys. Acta 1089, 1–7. [82] Melendez-Hevia, E., Waddell, T.G. and Cascante, M. (1996) The puzzle of the Krebs citric acid cycle: assembling the pieces of chemically feasible reactions, and opportunism in the design of metabolic pathways during evolution. J. Mol. Evol. 43, 293–303. 351 [83] Mendzhul, M.I., Lysenko, T.G., Shainskaia, O.A. and Busakina, I.V. (2000) Activity of tricarboxylic acid cycle enzymes in the cyanobacterium Spirulina platensis. Mikrobiol. Z. 62, 3–10. [84] Miles, J.S. and Guest, J.R. (1987) Molecular genetics aspects of the citric acid cycle of Escherichia coli. In: KrebsÕ Citric Acid Cycle – Half a Century and Still Turning. Biochem. Soc. Symp. 54, 45–65. [85] Mira, A., Ochman, H. and Moran, N.A. (2001) Deletional bias and the evolution of bacterial genomes. Trends Gen. 17, 589– 596. [86] Mittenthal, J.E., Clarke, B., Waddell, T.G. and Fawcett, G. (2001) A new method for assembling metabolic networks, with application to the Krebs citric acid cycle. J. Theor. Biol. 208, 361–382. [87] Morohashi, M., Winn, E.E., Borisuk, M.T., Balouri, H., Doyle, E.J. and Kitano, H. (2002) Robustness as a measure of plausibility in models of biochemical networks. J. Theor. Biol. 216, 19–30. [88] Murrell, J.C. and Radejewski, S. (2000) Cultivation-independent techniques for studying methanotroph ecology. Res. Microbiol. 151, 807–814. [89] Neilson, A.H., Holm-Hansen, O. and Lewin, R.A. (1972) An obligately autotrophic mutant of Chlamydomonas dysosmos; a biochemical elucidation. J. Gen. Microbiol. 71, 141–148. [90] Pace, N.R. (1996) New perspectives on the natural microbial world: molecular microbial ecology. ASM News 62, 463–470. [91] Pacheco-Oliver, M., McDonald, I.R., Groleau, D., Murrell, J.C. and Miguez, C.B. (2002) Detection of methanotrophs with highly divergent pmoA genes from arctic soils. FEMS Microbiol. Lett. 209, 313–319. [92] Palleroni, N.J. (1997) Prokaryotic diversity and the importance of culturing. Antonie van Leeuwenhoek 72, 3–19. [93] Park, S.-J., Chao, C. and Gunsalus, R.P. (1997) Aerobic regulation of the sucABCD genes of Escherichia coli, which encode a-ketoglutarate dehydrogenase and succinyl coenzyme A synthase: roles of ArcA, Fnr, and the upstream sdhCDAB promoter. J. Bacteriol. 179, 4138–4142. [94] Pearce, J. and Carr, N.G. (1967) The metabolism of acetate by the blue-green algae, Anabaena variabilis and Anacystis nidulans. J. Gen. Microbiol. 49, 301–313. [95] Pearce, J. and Carr, N.G. (1967) An incomplete tricarboxylic acid cycle in the blue-green alga Anabaena variabilis. Biochem. J. 105, 45. [96] Pearce, J., Leach, C.K. and Carr, N.G. (1969) The incomplete tricarboxylic acid cycle in the blue-green alga Anabaena variabilis. J. Gen. Microbiol. 55, 371–378. [97] Peeters, T.L., Liu, M.S. and Aleem, M.I.H. (1970) The tricarboxylic acid cycle in Thiobacillus denitrificans and ThiobacillusA2. J. Gen. Microbiol. 64, 29–35. [98] Perna, N.T. and 27 co-authors (2001) Genome sequence of enterohaemorrhagic Escherichia coli O157:H7. Nature 409, 529– 533. [99] Pfeffer, W. (1897) Pflanzenphysiologie, 2nd edn., Vol. 1. W. Engelmann, Leipzig. [100] Prescott, L.M., Harley, J.P. and Klein, D.A. (1996) Microbiology. William C. Brown, Dubuque. [101] Reed, I.J. (1974) Multienzyme complexes. Acc. Chem. Res. 7, 40–46. [102] Ritchie, D.A., Edwards, C., McDonald, I.R. and Murrell, J.C. (1997) Detection of methanogens and methanotrophs in natural environments. Global Change Biol. 3, 339–350. [103] Rittenberg, S.C. (1969) The roles of exogenous organic matter in the physiology of chemolithotrophic bacteria. Adv. Microb. Physiol. 3, 159–196. [104] Rittenberg, S.C. (1972) The obligate autotroph – the demise of a concept. Antonie van Leeuwenhoek 38, 457–478. 352 A.P. Wood et al. / FEMS Mircobiology Reviews 28 (2004) 335–352 [105] Rocap, G. and 23 co-authors (2003) Genome divergence in two Prochlorococcus ecotypes reflects oceanic niche differentiation. Nature 424, 1042–1047. [106] Roche, T.E. and Patel, M.S. (Eds.) (1989) Alpha-keto acid dehydrogenase complexes: organization, regulation, and biochemical ramifications: a tribute to Lester J. Reed. Ann. N. Y. Acad. Sci. 573. [107] Schlegel, H.G. (1975) Mechanisms of chemoautotrophy. In: Marine Ecology (Kinne, O., Ed.), Vol. 2, pp. 9–60. Wiley, London. part 1. [108] Schlegel, H.G. (1992) General Microbiology, seventh edition. The University Press, Cambridge. [109] Scholten, J.C.M., Kelly, D.P. and Murrell, J.C. (2003) Growth of sulfate-reducing bacteria and methanogenic archaea with methylated sulfur compounds: a commentary on the thermodynamic aspects. Arch. Microbiol. 179, 135–144. [110] Schroder, W., Burger, M., Edwards, C., Douglas, M., Innes, D., Beacham, I.R. and Burns, D.M. (2001) The Escherichia coli orthologue of the Salmonella ushB (ushB(c)) gene produces neither UDP–sugar hydrolase activity nor detectable protein but has an identical sequence to that of Escherichia coli cdh. FEMS Microbiol. Lett. 203, 63–68. [111] Shishkina, V.N., Yurchenko, V.V., Romanovskaia, V.A., Malashenko, Y.R. and Trotsenko, Y.A. (1976) Alternativity of methane assimilation pathways in obligate methylotrophs. Mikrobiologiya (Moscow) 45, 417–419. [112] Smith, A.J. and Hoare, D.S. (1977) Specialist phototrophs, lithotrophs, and methylotrophs: unity among a diversity of prokaryotes. Bacteriol. Rev. 41, 419–448. [113] Smith, A.J., London, J. and Stanier, R.Y. (1967) Biochemical basis of obligate autotrophy in blue-green algae and thiobacilli. J. Bacteriol. 94, 972–983. [114] Smith, A.L. and Kelly, D.P. (1979) Competition in the chemostat between an obligately and a facultatively chemolithotrophic Thiobacillus. J. Gen. Microbiol. 115, 377–384. [115] Smith, N.A. and Kelly, D.P. (1988) Mechanism of oxidation of dimethyl disulphide by Thiobacillus thioparus strain E6. J. Gen. Microbiol. 134, 3031–3039. [116] Smith, D.R. and 36 co-authors (1997) Complete genome sequence of Methanobacterium thermoautotrophicum DeltaH: functional analysis and comparative genomics. J. Bacteriol. 179, 7135– 7155. [117] Spencer, M.E., Darlison, M.G., Stephens, P.E., Duckemfield, I.K. and Guest, J.R. (1984) Nucleotide sequence of the sucB gene encoding the dihydrolipoamide succinyltransferase of Escherichia coli K-12 and homology with the corresponding acetyl transferase. Eur. J. Biochem. 141, 361– 374. [118] Steiert, P.S., Stauffer, L.T. and Stauffer, G.V. (1990) The lpd gene product functions as the L protein in the Escherichia coli glycine cleavage enzyme system. J. Bacteriol. 172, 6142–6144. [119] Stephens, P.E., Stauffer, L.T. and Stauffer, G.V. (1983) Nucleotide sequence of the lipoamide dehydrogenase gene of Escherichia coli K-12. Eur. J. Biochem. 135, 519–527. [120] Still, G.G. (1965) The role of some of the Krebs cycle reactions in the biosynthetic functions of Thiobacillus thioparus. Ph.D. Dissertation, Oregon State University. [121] Strauss, G. and Fuchs, G. (1993) Enzymes of a novel CO2 fixation pathway in the photosynthetic bacterium Chloroflexus aurantiacus, the 3-hydroxypropionate cycle. Eur. J. Biochem. 215, 633–643. [122] Stryer, L. (1988) Biochemistry, third ed. Freeman, New York. [123] Takami, H. and 11 co-authors (2000) Complete genome sequence of the alkaliphilic bacterium Bacillus halodurans and genome sequence comparison with Bacillus subtilis. Nucleic Acids Res. 28, 4317–4331. [124] Taylor, B.F. and Hoare, D.S. (1971) Thiobacillus denitrificans as an obligate chemolithotroph. Arch. Mikrobiol. 80, 262– 276. [125] Taylor, I.J. and Anthony, C. (1976) A biochemical basis for obligate methylotrophy: properties of a mutant of Pseudomonas AM1 lacking 2-oxoglutarate dehydrogenase. J. Gen. Microbiol. 93, 259–265. [126] Taylor, S.C., Dalton, H. and Dow, C.S. (1981) Ribulose-1,5bisphosphate carboxylase/oxygenase and carbon assimilation in Methylococcus capsulatus (Bath). J. Gen. Microbiol. 122, 89–94. [127] Tian, K.L, Lin, J.Q., Liu, X.M., Liu, Y., Zhang, C.K. and Yan, W.M. (2003) Conversion of an obligate autotrophic bacteria to heterotrophic growth: expression of a heterologous phosphofructokinase gene in the chemolithotroph Acidithiobacillus thiooxidans. Biotechnol. Lett. 25, 749–754. [128] TIGR (2003a) NCBI Entrez genomes: Acidithiobacillus ferrooxidans, unfinished sequence. Accession: NC_002923. [129] TIGR (2003b) NCBI Entrez genomes: Methylococcus capsulatus, unfinished sequence. Accession: NC_002977. [130] Trudinger, P.A. and Kelly, D.P. (1968) Reduced nicotinamide adenine dinucleotide oxidation by Thiobacillus neapolitanus and Thiobacillus strain C. J. Bacteriol. 95, 1962–1963. [131] Turner, M. and Schuch, W. (2000) Post-transcriptional genesilencing and RNA interference: genetic immunity, mechanisms and applications. J. Chem. Technol. Biotechnol. 75, 869–882. [132] Umbreit, W.W. (1947) Problems of autotrophy. Bacteriol. Rev. 11, 157–182. [133] Van Dien, S.J., Okubo, Y., Hough, M.T., Korotkova, N., Taitano, T. and Lidstrom, M.E. (2003) Reconstruction of C3 and C4 metabolism in Methylobacterium extorquens AM1 using transposon mutagenesis. Microbiology UK 149, 601–609. [134] W€achtersh€auser, G. (1990) Evolution of the first metabolic cycles. Proc. Natl. Acad. Sci. USA 87, 200–204. [135] W€achtersh€auser, G. (1990) The case for the chemo-autotrophic origin of life in an iron-sulfur world. Origins Life 21. [136] Wallace, W., Knowles, S.E. and Nicholas, D.J.D. (1970) Intermediary metabolism of carbon compounds by nitrifying bacteria. Arch. Mikrobiol. 70, 26–42. [137] Whittenbury, R. and Kelly, D.P. (1977) Autotrophy: a conceptual phoenix. Symp. Soc. Gen. Microbiol. 27, 121–149. [138] Wilkins, M.R., Sanchez, J.-C., Gooley, A.A., Appel, R.D., Humphery-Smith, I., Hochstrasser, D.F. and Williams, K.L. (1995) Progress with proteome projects: why all proteins expressed by a genome should be identified and how to do it. Biotechnol. Genetic Eng. Rev. 13, 19–50. [139] Wilkins, M.R. and Williams, K.L. (1997) Cross-species protein identification using amino acid composition, peptide mass fingerprinting, isoelectric point and molecular mass: a theoretical evaluation. J. Theor. Biol. 186, 7–15. € [140] Winogradsky, S. (1887) Uber Schwefelbakterien. Bot. Ztg. 45, 489–600, 604–616. [141] Woese, C.R. (1987) Bacterial evolution. Microbiol. Rev. 51, 221– 271. [142] Woese, C.R. (2000) Interpreting the universal phylogenetic tree. Proc. Natl. Acad. Sci. USA 97, 8392–8396. [143] Wolska, K.I. (2003) Horizontal DNA transfer between bacteria in the environment. Acta Microbiol. Polon. 52, 233–243. [144] Wood, A.P. and Kelly, D.P. (1984) Growth and sugar metabolism of a thermophilic iron-oxidizing mixotrophic bacterium. J. Gen. Microbiol. 130, 1337–1349. [145] Zaigler, A., Schuster, S.C. and Soppa, J. (2003) Construction and usage of a onefold-coverage shotgun DNA microarray to characterize the metabolism of the archaeon Haloferax volcanii. Mol. Microbiol. 48, 1089–1105. [146] Zhang, C.C., Jeanjean, R. and Joset, F. (1998) Obligate autotrophy in cyanobacteria: more than a lack of sugar transport. FEMS Microbiol. Lett. 161, 285–292.