Multiple genome sequences reveal adaptations of a phototrophic bacterium to sediment microenvironments
Yasuhiro Odaa, Frank W. Larimerb, Patrick S. G. Chainc,d,e, Stephanie Malfattic,d, Maria V. Shinc,d, Lisa M. Vergezc,d, Loren Hauserb, Miriam L. Landb, Stephan Braatschf, J. Thomas Beattyf, Dale A. Pelletierb, Amy L. Schaefera, and Caroline S. Harwooda,1
aDepartment of Microbiology, University of Washington, Seattle, WA 98195; bGenome Analysis and Systems Modeling, Oak Ridge National Laboratory, Oak Ridge, TN 37831; cJoint Genome Institute, Walnut Creek, CA 94598; dLawrence Livermore National Laboratory, Livermore, CA 94550; eDepartment of Microbiology and Molecular Genetics, Michigan State University, East Lansing, MI 48824; and fDepartment of Microbiology and Immunology, University of British Columbia, Vancouver, British Columbia V6T 1Z3, Canada
Edited by Robert Haselkorn, University of Chicago, Chicago, IL, and approved October 14, 2008 (received for review September 13, 2008) The bacterial genus Rhodopseudomonas is comprised of photo- exist in soils and sediments, but on a microscale that is generally synthetic bacteria found widely distributed in aquatic sediments. too small for human observation. The genus Rhodopseudomonas Members of the genus catalyze hydrogen gas production, carbon consists of photosynthetic Alphaproteobacteria of extreme met- dioxide sequestration, and biomass turnover. The genome se- abolic versatility. Members of the genus are ubiquitous in quence of Rhodopseudomonas palustris CGA009 revealed a sur- temperate aquatic sediments (7–9), and isolates classified as prising richness of metabolic versatility that would seem to explain Rhodopseudomonas spp. can grow with or without light or its ability to live in a heterogeneous environment like sediment. oxygen, fix nitrogen, and have highly developed biodegradation However, there is considerable genotypic diversity among Rhodo- abilities. The sequenced genome of Rhodopseudomonas palustris pseudomonas isolates. Here we report the complete genome strain CGA009 revealed much of the genetic basis for this sequences of four additional members of the genus isolated from versatility and, we presumed, its ability to grow in heterogeneous a restricted geographical area. The sequences confirm that the environments typical of sediments (10). However, expanding isolates belong to a coherent taxonomic unit, but they also have beyond the analysis of this single strain, a genotypic character- significant differences. Whole genome alignments show that the ization of 75 isolates of Rhodopseudomonas from sediment circular chromosomes of the isolates consist of a collinear back- samples at three different sites revealed significant strain-to- bone with a moderate number of genomic rearrangements that strain differences (11). That study showed that the genus Rho- impact local gene order and orientation. There are 3,319 genes, dopseudomonas consists of distinct populations and raised the 70% of the genes in each genome, shared by four or more strains. possibility that each population has distinctive physiological Between 10% and 18% of the genes in each genome are strain characteristics. To investigate this possibility we sequenced the specific. Some of these genes suggest specialized physiological genomes of four genotypically distinct isolates of Rhodopseudo- traits, which we verified experimentally, that include expanded monas and analyzed selected physiological traits. Our findings light harvesting, oxygen respiration, and nitrogen fixation capa- show that although the isolates share many characteristics in bilities, as well as anaerobic fermentation. Strain-specific adapta- common, each strain has a unique set of genes for physiologies tions include traits that may be useful in bioenergy applications. that define them as distinct ecotypes. The ecotypes have likely This work suggests that against a backdrop of metabolic versatility evolved to take advantage of microenvironments in sediments. that is a defining characteristic of Rhodopseudomonas, different Strain-specific adaptations that allow anaerobic fermentation,
ecotypes have evolved to take advantage of physical and chemical expanded biodegradation, or expanded light-harvesting capabil- MICROBIOLOGY conditions in sediment microenvironments that are too small for ities are also potentially useful in applications for biohydrogen human observation. production by Rhodopseudomonas.
alphaproteobacteria ͉ ecotype ͉ genomes ͉ photosynthesis ͉ Results rhodopseudomonas Selection of Rhodopseudomonas Strains for Genome Sequencing. Rhodopseudomonas strains isolated by direct plating from three atural populations of closely related bacteria are commonly freshwater sediment samples from the Netherlands belonged to Ncomprised of physiologically and genetically distinct vari- several distinct clades based on 16S rRNA analysis (Fig. 1). We ants, referred to as ecotypes (1). Ecotypes are thought to have selected three strains (BisB18, BisB5, and BisA53) isolated from evolved by adapting to environmental conditions in the natural the top 0.5 cm of claylike sediment that was present 1–2 cm below habitats from which they derive. Some have suggested that ecotypes are the fundamental biological units, rather than species, which have no generally agreed upon theoretical basis in Author contributions: Y.O., P.S.G.C., and C.S.H. designed research; Y.O., S.M., M.V.S., L.M.V., S.B., and A.L.S. performed research; Y.O., F.W.L., P.S.G.C., L.H., M.L.L., J.T.B., D.A.P., microbiology (2, 3). Perhaps the best example of the analysis of and C.S.H. analyzed data; and Y.O. and C.S.H. wrote the paper. ecotypes at the genome level comes from studies of the marine The authors declare no conflict of interest. cyanobacterial genus Prochlorococcus. Prochlorococcus isolates This article is a PNAS Direct Submission. from various depths in ocean waters vary in growth responses to Data deposition: The sequence and annotations of the complete Rhodopseudomonas light intensity and have been classified as high-light- or low- chromosomes have been deposited in GenBank/EMBL/DDBJ [accession nos. CP000250 light-adapted ecotypes. The genome sequences of a collection of (strain HaA2), CP000283 (strain BisB5), CP000301 (strain BisB18), and CP000463 (strain these isolates revealed the molecular basis for the high-light and BisA53)]. low-light ecological differentiation of natural populations (4–6). 1To whom correspondence should be addressed. E-mail: [email protected]. In contrast to open ocean environments, which tend to be This article contains supporting information online at www.pnas.org/cgi/content/full/ homogeneous on a large scale, soils and sediments are hetero- 0809160105/DCSupplemental. geneous on a large scale. Homogeneous environments likely © 2008 by The National Academy of Sciences of the USA
www.pnas.org͞cgi͞doi͞10.1073͞pnas.0809160105 PNAS ͉ November 25, 2008 ͉ vol. 105 ͉ no. 47 ͉ 18543–18548 Downloaded by guest on September 25, 2021 0.01 Clade TIE-1 100 NCIB8288 585 514 420 859 794 AP1 316 40 437 438 BIS3 80 405 488
59 tnecreP fo seneg 492 CGA009 515 462 73 KD1 355 510 335 95 529 DCP3 60 544 429 507 92 WS17
92 BIS6 40 85 90 BIS10 DX-1 BisB18 2,752 2,740 2,751 2,746 2,760 84 BisA53 20 55 HaA2 100 BisB5 0 100 96 NCIMB8252 CGA009 HaA2 BisB18 BisB5 BisA53 BIS23
(4,833) (4,683) (4,886) (4,397) (4,884) 86 BIS18 BIS17 Fig. 2. Comparative gene inventories of five strains of Rhodopseudomonas. BIS14 BIS11 Ortholog categories were determined using OrthoMCL (37). Each category is color coded as follows: genes shared by all five genomes including paralogs 97 Bradyrhizobium japonicum USDA 110 92 Bradyrhizobium sp. ORS278 (black bars); genes shared by four genomes including paralogs (red); genes Bradyrhizobium sp. BTAi1 shared by three genomes including paralogs (green); genes shared by two Nitrobacter winogradskyi Nb-255 genomes including paralogs (yellow); and strain-specific genes including in- paralogs (light blue). Numbers in the bars represent number of genes in each Fig. 1. Phylogenetic relationships of Rhodopseudomonas and Bradyrhizo- category. bium strains based on partial (1,256 bp) 16S rRNA gene sequences. Bootstrap values (100 replicates) are given at branch points. Bar represents substitutions per site. Nitrobacter winogradskyi Nb-255 was used to root the tree. Se- which encode phagelike elements, and CGA009 has a vanadium quenced strains are indicated in red. nitrogenase gene cluster. In general, the genomes are so similar that we used only CGA009 in our comparisons. When the the surface of a river along its bank. Strains BisB18 and BisB5 genomes of pairs of strains are aligned, some pairs, such as were isolated from the same 0.5 g of sediment sample, whereas BisB18 and BisA53, have similar genome architectures. Other BisA53 was from a sample taken about 5 m away. The river was pairs, such as the comparison of CGA009 with HaA2, have large not obviously polluted but it was near a small industrial area. A inversions of DNA relative to each other. In most pairwise fourth strain (HaA2) came from a site roughly 240 km from the comparisons, many rearrangements and sequence inversions are first two locations. It was obtained from a 1–2 mm-thick patch observed even though the general gene order and overall of leaf litter, roots, and sediment present Ϸ2 cm below the genome architecture is preserved (supporting information (SI) surface of a shallow pond that was formed by the accumulation Fig. S1). All of the genomes exhibit homogenized GC skews (Fig. of rainwater in a depression (11). These four isolates as well as S2), indicating that the observed high levels of genome rear- previously sequenced Rhodopseudomonas palustris strain rangements may be mostly due to old events. Only one of the CGA009 had 16S rRNA sequence identities of 97.3% or greater strains (CGA009) has an extrachromosomal element in the form (Table 1). Because this degree of divergence is slightly more than of a small plasmid. that considered by many to be the cutoff for a species (2, 3), we have elected to classify the strains simply as Rhodopseudomonas. Gene Inventories. The 23,683 predicted proteins encoded by the Cells of each strain tend to form rosettes, and each has the five genomes were clustered into categories according to the budding type of cell division that is typical of the species number of orthologs shared by groups of strains (Fig. 2). About Rhodopseudomonas palustris. 2,750 genes are shared by all five genomes, which is about 55% of the genes from each strain. We designated the 3,319 genes General Genome Features. The genomes of the five strains are each (without paralogs) that are shared by four or more strains as core about 5.5 Mb in size, with the exception of BisB5, which is slightly genes (Table S1). Among these, 1,591 genes (47.9%) had top hits smaller with a genome of 4.9 Mb (Table 1). Also included in to Bradyrhizobium japonicum USDA 110 genes, and 786 (23.7%) Table 1 are the key features of an additional recently completed and 283 (8.5%) genes had top hits to Nitrobacter hamburgensis Rhodopseudomonas palustris genome, that of strain TIE-1 X14 and Nitrobacter winogradskyi Nb-255 genes, respectively. (http://genome.ornl.gov/microbial/rpaltie1/). The genomes of Only 45 and 30 core genes had top hits to the purple nonsulfur TIE-1 and CGA009 are 97.9% identical at the nucleotide level phototrophs Rhodobacter sphaeroides 2.4.1 and Rhodospirillum over 5.28 Mb of shared DNA. The two genomes each have rubrum ATCC 11170 genes, respectively. The genomes have between 5 and 10 large indels (insertions/deletions), many of between 420 (BisB5; 9.6% of the genome) and 859 (BisB18;
Table 1. General features of sequenced strains of Rhodopseudomonas Feature CGA009 (RPA) HaA2 (RPB) BisB18 (RPC) BisB5 (RPD) BisA53 (RPE) TIE-1 (Rpal)
Size (bp) 5,459,213 5,331,656 5,513,844 4,892,717 5,505,497 5,744,041 G ϩ C percentage 65.1 66.0 65.0 64.8 64.4 64.9 Coding sequences 4,833 4,683 4,886 4,397 4,884 5,326 Pseudogenes 17 29 57 20 36 75 rRNA operons 2 1 2 2 2 2 16S rRNA gene sequence identity, %* — 97.6 97.3 97.5 97.8 100 Plasmid 1 (8,427 bp) 0 0 0 0 0
*Relative to strain CGA009.
18544 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0809160105 Oda et al. Downloaded by guest on September 25, 2021 Table 2. Ortholog comparisons between pairs of Rhodopseudomonas genomes Genome CGA009 HaA2 BisB18 BisB5 BisA53
CGA009 (4,734) 3,650 (77.1%) 3,293 (69.6%) 3,524 (74.4%) 3,376 (71.3%) HaA2 (4,585) 81.7% 3,209 (70.0%) 3,531 (77.0%) 3,306 (72.1%) BisB18 (4,695) 74.6% 75.1% 3,132 (66.7%) 3,497 (74.5%) BisB5 (4,311) 81.5% 87.0% 75.8% 3,246 (75.3%) BisA53 (4,746) 74.9% 75.3% 79.1% 76.0%
The upper-right half of the matrix shows numbers and percentage in parenthesis of orthologs (paralogs were excluded) and the lower-left half of the matrix shows the average percent amino acid identities between orthologs from pairs of genomes.
17.6% of the genome) strain-specific genes (Table S2). The organism to elucidate a central anaerobic benzoate degradation pan-genome includes about 8,000 genes (12). The percentage of pathway that involves novel steps of benzene ring reduction and shared genes between pairs of genomes varies from 66.7% to cleavage in the absence of oxygen (20, 21). The cluster of genes 77.1%, and the average percent amino acid identity between that encodes this pathway (RPA0650-RPA0673) is found in four orthologs from pairs of genomes varies from 74.6% to 87.0% of the five Rhodopseudomonas genomes. Strain HaA2 lacks this (Table 2). anaerobic capability. CGA009 has a well-developed ability to use medium- and Predicted Core Metabolic Features of Rhodopseudomonas Isolates. long-chain fatty acids and dicarboxylic acids for growth (22), and The combination of genes shared by all strains for photosynthe- this capability appears to be characteristic of the Rhodopseudo- sis, respiration, nitrogen fixation, biodegradation, and inorganic monas group. All five strains have a conserved cluster of fatty compound oxidation defines Rhodopseudomonas as a group acid -oxidation genes (RPA3712–3717) (22), and all encode distinct from any other described bacterial taxonomic unit enzymes of the glyoxylate shunt. BisA53 does however have a (Table S3). transposase gene (RPE3777) inserted between two of the genes in its -oxidation gene cluster. Phototrophy. Phototrophy is a hallmark characteristic of Rhodo- pseudomonas that is reflected in a conserved cluster of 50 genes Autotrophic Growth. Each of the strains encodes ‘‘red’’ form IC (RPA1505–1554; CGA009 numbering) for carotenoid and bac- CbbLS (RPA1559–1560) and form II CbbM (RPA4641) ribu- teriochlorophyll synthesis, the structural components of the lose-1,5-bisphosphate carboxylases (RubisCO) for carbon diox- photosynthetic reaction center, and a light-harvesting 1 (LH1) ide fixation. Other genes of the Calvin cycle are also conserved system. These genes are contiguous except for a small insertion among the five strains and adjacent to cbbM. CGA009 differs of unrelated genes in the photosynthesis cluster in BisA53 and a from other purple nonsulfur bacteria in that it grows poorly with larger insertion of unrelated genes in BisB18. This gene region CO2 but well when supplied with inorganic carbon as NaHCO3 also encodes a conserved combination of regulatory proteins (23). Two carbonic anhydrases that may be involved in conver- that is characteristic of Rhodopseudomonas. These include sion of NaHCO to CO gas are conserved in all five strains. PpsR1 and PpsR2 transcription factors, which repress photosys- 3 2 Electron donors used by the Rhodopseudomonas isolates for tem development under conditions of high aeration, and a photoautotrophic growth include thiosulfate (all strains except bacteriophytochrome (RpBphP1) that responds to far-red light BisB18 encode thiosulfate oxidase) and hydrogen (all strains to antagonize the repressive activity of PpsR2 when oxygen except HaA2 encode NiFe uptake hydrogenase). levels are sufficiently low to stimulate the transcription of
photosynthesis genes (13, 14). Each of the strains encodes a MICROBIOLOGY Respiration. Rhodopseudomonas is well adapted to microaerobic light-harvesting 4 (LH4) complex and at least three variants of light harvesting 2 (LH2). LH4 complexes predominate at growth in light, and it also grows aerobically in dark. Genes low light and may absorb light more efficiently than LH2 encoding a cytochrome cbb3 oxidase (RPA0014–0019), a cyto- complexes (15–17). The LH4 region in each Rhodopseudomonas chrome aa3 oxidase (RPA0831–0837), and a cytochrome bd strain encodes two bacteriophytochromes (RpBphP2 and 3; ubiquinol oxidase (RPA4793–4794) are conserved in the five RPA3015 and 3016) that function as light-regulated signal strains. Nos genes to permit anaerobic-dark growth with nitrous transduction histidine kinases in conjunction with neighboring oxide as a terminal electron acceptor (RPA2060–2066) are response regulator proteins that are presumed to control LH4 present in all strains except BisB5. All five strains encode two gene expression (18, 19). The presence of multiple LH polypep- paralogous NADH dehydrogenase complexes (RPA2937–2952 tides and bacteriophytochromes indicates that Rhodopseudomo- and RPA4252–4264). Genes encoding a FixLJ-K two- nas has developed a sophisticated set of photoreceptors to component low-oxygen sensor-regulator system are adjacent to respond to changes in light intensity and light quality as com- the second dehydrogenase gene cluster in all strains. This system pared with other purple nonsulfur phototrophs. has not been described in other purple nonsulfur bacteria but is widely distributed in the rhizobiaceae (24). Carbon Source Utilization. The most distinctive carbon utilization characteristic that sets the Rhodopseudomonas group apart from Nitrogen Fixation and Hydrogen Production. A cluster of 32 genes for many other bacteria is its ability to degrade aromatic compounds the assembly, synthesis, and activity of molybdenum-dependent (10). All five strains encode oxygenase-dependant ring cleavage nitrogenase (RPA4602–4633) is completely conserved among the pathways for homogentisate (RPA4670–4675) and phenylace- strains (25). All strains also have a conserved cluster of genes tate (RPA1723–1724 and RPA3765–3768), compounds that are predicted to be involved in regulating nitrogen metabolism derived from diverse sources, including amino acids and plant (RPA2591–2598). Although the biological function of nitrogenases secondary metabolites. All strains except BisB18 encode aerobic is to convert nitrogen gas to ammonia, these enzymes also generate degradation pathways for protocatechuate (derived from lignin hydrogen gas as an obligatory aspect of their catalytic mechanism. monomers; RPA4695–4703). Strain CGA009 served as a model We found that all of the Rhodopseudomonas strains grew under
Oda et al. PNAS ͉ November 25, 2008 ͉ vol. 105 ͉ no. 47 ͉ 18545 Downloaded by guest on September 25, 2021 Table 3. Major predicted distinguishing characteristics of Rhodopseudomonas strains Strain Major characteristic Description
CGA009 Nitrogen fixation and acquisition Vanadium nitrogenase gene (vnf) cluster (RPA1370–1380) HaA2 Oxidative metabolism Additional terminal oxidases ͓cytochrome bd-quinol oxidases (RPB0205–0206 and RPB0929–0930) and cytochrome o ubiquinol oxidase (RPB0239–0242)͔ BisB18 Anaerobic growth in the dark Pyruvate-formate lyase and associated proteins (RPC1163–1180) Formate-hydrogen lyase (formate dehydrogenase coupled hydrogenase) complex (RPC4554–4579) Carbon monoxide dehydrogenase and carbon monoxide-tolerant hydrogenase (RPC4493–4506) Dimethylsulfoxide reductases (RPC0656–0658, RPC1749–1751, and RPC1875–1877) BisB5 Expanded anaerobic aromatic Anaerobic phenylacetate degradation (RPD1506–1521) compound degradation BisA53 Expanded light harvesting and Additional LH polypeptide sets (RPE2713–2714, RPE2831–2832, and exopolysaccharide synthesis RPE4547–4548) Unique genes and gene clusters encoding glycosyl transferases (RPE0806–0813, RPE1088–1120, RPE 2215, RPE3492–3510, RPE3652*, and RPE4328–4330)
*Encodes a putative cyclic  -glucan synthase.
nitrogen-fixing conditions and had significant nitrogenase and the anaerobic, dark, oxidation of carbon monoxide. We con- hydrogen production activities (Table S4). firmed predictions based on its gene inventory that BisB18 grows photoheterotropically on methanol or acetone, carbon com- Features Common to Two or Three Strains. Approximately 20% of pounds that are atypical for other Rhodopseudomonas strains. the genes in any genome have an ortholog in just one or two of Although its genome sequence predicts that BisB18 should grow the other of the four (Fig. 2). Among these are an acyl homo- well by aerobic respiration, this is not its preferred mode of serine lactone (HSL) synthase gene (RPA0320) that specifies the growth in the laboratory. It grows very slowly in the presence of synthesis of a novel quorum-sensing signaling compound, p- air (Table S5). coumaroyl-HSL, from exogenously supplied p-coumarate rather The HaA2 gene inventory predicts that this strain is better than from endogenously produced fatty acids, as is the case with adapted for growth by aerobic respiration or possibly microaero- all other described HSL synthases (26). The use of p-coumarate, bic respiration in light than for anaerobic photoheterotrophic a plant metabolite, for quorum sensing allows bacterial cells to growth. In addition to the three aerobic terminal oxidases that integrate the two environmental cues of cell density and carbon are encoded by all five strains, HaA2 encodes an additional two compound availability to promote gene expression by its asso- cytochrome bd-quinol oxidase paralogs and a cytochrome o ciated p-coumaroyl-HSL-dependant transcriptional regulator ubiquinol oxidase. Genes for this latter enzyme are also present (RPA0321). Besides CGA009, in which it was characterized, this in BisB5. HaA2 grows faster with succinate under aerobic system is present and functional only in HaA2 and BisB5. The conditions than anaerobically in light (Table S5). It is also the circadian clock genes (RPA0008–0009) that were identified in one strain that lacks the genes necessary for the degradation of CGA009 and are now known to be present in other photosyn- aromatic compounds in the absence of oxygen. thetic bacteria, including Rhodospirillum rubrum, Rhodobacter A distinguishing characteristic of CGA009 is that it has highly sphaeroides, and photosynthetic Bradyrhizobium strains, are developed nitrogen fixation and nitrogen acquisition abilities. It found only in BisB5 among the newly sequenced Rhodopseudo- encodes active vanadium and iron nitrogenases in addition to the monas strains. A final example, out of many that could be expected molybdenum nitrogenase (10, 25). BisB18 and BisA53 mentioned, is the pio operon for iron (II) oxidation (27). encode molybdenum and iron nitrogenases only, and HaA2 and Characterized in Rhodopseudomonas palustris strain TIE-1 (28), BisB5 encode only molybdenum nitrogenase. HaA2 and BisB5 this operon encodes a c-type cytochrome, a putative outer encode a single ammonia transporter, whereas the other three membrane protein and a high-potential iron sulfur protein that strains encode two of these proteins (RPA0273 and RPA0275). enables photoautotrophic growth on reduced iron. This operon CGA009 and HaA2 encode four glutamine synthetases, whereas is present in CGA009 (RPA0744–0746), BisB18, and BisA53. other strains encode only one or two versions of this key ammonia-assimilating enzyme. Strain-Specific Genes Reveal Physiological Adaptations Relevant to Strain BisB5 is unique among the Rhodopseudomonas strains Life in Sediments. The set of genes that is unique to each strain in having a gene cluster for anaerobic phenylacetate degradation includes many hypothetical and conserved hypothetical genes, similar to that described in the denitrifying bacterium Azoarcus but also genes that have predicted functions that may give each sp. strain EbN1 (29). BisB5 grows well anaerobically in light with strain a competitive advantage over the others in specialized phenylacetate as a carbon source (Table S5). Phenylacetate is environments (Table 3). Of the five strains, BisB18 is predicted formed in the degradation of phenylpyruvate and flavonoids. Its to be the best adapted to anaerobic dark growth. It encodes degradation pathway involves activation by CoA and oxidative pyruvate-formate and formate-hydrogen lyases that permit decarboxylation to benzoyl-CoA. The benzoate degradation growth by pyruvate fermentation. It also has three sets of gene cluster lies adjacent to the phenylacetate gene cluster in dimethyl sulfoxide reductase genes to allow for anaerobic res- BisB5 and is followed by an operon (RPD1546–1549) encoding piration of organic compounds with dimethyl sulfoxide as a an additional ‘‘greenlike’’-type IAq RubisCO (30), which is also terminal electron acceptor (Table 3 and Table S5). Additionally, unique to BisB5. the BisB18 genome indicates the potential to derive energy from In the laboratory, BisA53 cells aggregate during growth and
18546 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0809160105 Oda et al. Downloaded by guest on September 25, 2021 0.9 lateral gene transfer. In addition to the gross changes that we see CGA009 0.8 in gene inventories, mutations and alterations of the location, HaA2 orientation, and the coding strand of existing genes have likely 0.7 BisB18 BisB5 changed the enzymatic activities and expression levels of many 0.6 ecnabrosbA BisA53 genes to provide increased adaptive fitness. 0.5 We selected for sequencing the four strains described here
0.4 because their rRNA phylogenies predicted that these would be among the most divergent strains in our collection from three 0.3 sites in the Netherlands. As noted previously, the percentage of 0.2 shared genes between pairs of genomes is about 72% Ϯ 5% for 0.1 each comparison, and the average percent amino acid identities between orthologs from pairs of genomes varies from 75% to 0 700 800 900 1000 87%. When we applied a similar analysis to nine fully sequenced Wavelength (nm) Prochlorococcus strains (5) we found that the percentage of Fig. 3. Absorption spectra of intact Rhodopseudomonas cells grown under shared genes between pairs of strains varies from 67% to 93%, anaerobic conditions in low light. and the average percent amino acid identities between orthologs from pairs of strains varies from 57% to 97% (Table S6). Thus, by these criteria, the sequenced Rhodospeudomonas strains are form thick biofilms on the walls of culture vessels. Scanning not as divergent as sequenced members of the Prochlorococcus electron micrographs showed evidence of amorphous extracel- group. The four new genome sequences discussed here are from lular material that could be exopolysaccharide. This likely re- strains that were isolated from environmental samples by direct flects the expanded set of exopolysaccharide synthesis genes that plating on nonselective media (11). Most previously described is encoded by this strain (Table 3). BisA53 also has seven sets of Rhodopseudomonas strains were isolated following enrichment light-harvesting genes, as compared with four sets in the other in selective media. Such strains tend to be more closely related strains. Absorption spectra of strain BisA53 revealed a unique to each other. For example, CGA009 and all of the strains, absorption peak at 810–820 nm that is similar to a type 3 including TIE-1, that group in the same clade with CGA009 in light-harvesting complex (LH3) described for Rhodoblastus aci- Fig. 1 were isolated by enrichment. TIE-1, the most recently dophila strain 7050 (31) (Fig. 3). This indicates that BisA53 has sequenced Rhodopseudomonas strain, was isolated following a competitive advantage over other Rhodopseudomonas strains enrichment for phototropic iron oxidizing bacteria (28), and as at 810–820 nm wavelengths of light. noted previously, is extremely similar to CGA009. Enrichment conditions typically select for the fastest growing members of a Discussion group that may be present in an inoculum of soil or sediment. It Our analysis suggests that the Rhodopseudomonas isolates are appears that as a consequence of this, the ecotypic diversity of different ecotypes that evolved by radiating into microenviron- the Rhodopseudomonas group may have been underestimated. ments with distinctive characteristics of light, oxygen, and nu- Rhodopseudomonas and other purple nonsulfur bacteria have trient availability while retaining a high degree of metabolic received attention as potential biocatalysts for nitrogenase- flexibility. CGA009 is well adapted to an environment with little mediated biohydrogen production because they can derive ATP to fixed nitrogen, and is depleted in molybdenum but possibly drive this thermodynamically unfavorable reaction from the abun- relatively rich in vanadium. BisB5 has the potential to better take dant resource of solar energy (32, 33). Currently, large quantities of advantage of decaying plant material in oxygen-depleted envi- hydrogen gas are used in petroleum refining and ammonia pro- ronments that are exposed to light than the other strains. HaA2 duction for fertilizer, and in the future there is the potential for has a highly developed ability to grow aerobically or microaero- hydrogen to be used in huge amounts as a transportation fuel.
bically and can likely outcompete other strains under these Although still under development, biological processes for hydro- MICROBIOLOGY conditions. BisB18, in contrast, can thrive in dark anaerobic gen production have the advantage of being environmentally conditions under which none of the other stains can grow. benign compared with the petroleum-based processes that are now BisA53 can use a slice of the electromagnetic spectrum that is not used for hydrogen production. The diversity of Rhodopseudomonas available to other strains. Although the microenvironments from strain-specific attributes provides new opportunities to analyze the which each of these Rhodopseudomonas ecotypes derive are on integrated functioning of the hundreds of proteins involved in a scale that may be too small to allow accurate measurement of photophosphorylation, carbon metabolism, reductant transfer, and their physical and chemical characteristics, we can perhaps infer nitrogenase synthesis that are needed for hydrogen production. For these characteristics from the genome sequences. example, studies of how BisA53 synthesizes and controls the Gene loss or acquisition is a major mechanism of bacterial activities of its record number of seven peripheral light-harvesting adaptive evolution to local conditions in heterogeneous envi- complexes in response to light intensity and quality should expand ronments that contain a range of ecologically distinct habitats. our current limited understanding of how synthetic biology and Evidence for the acquisition and retention of phenotypes via genetic engineering might be applied to maximize the efficiency of lateral gene transfer can be inferred for some of the strain- light absorption for photosynthesis. Similarly, studies of how specific gene islands listed in Table 3. For example, an integrase CGA009 differentially regulates the expression of its three nitro- gene (RPC4490) and a transposase gene (RPC4507) flank the genase isozymes should lead to an improved ability to manipulate cluster of BisB18 genes that codes for a carbon monoxide nitrogenase synthesis according to external cues of metal availabil- dehydrogenase and a carbon monoxide-tolerant hydrogenase. A ity and nitrogen starvation. The expanded abilities of strain BisB5 large cluster of putative exopolysacharide synthesis genes to use aromatic compounds of the type derived from low-cost (RPE1088–1120) is flanked by a phage integrase gene agricultural feed stocks provides opportunities to study and im- (RPE1075) and a transposase gene (RPE1134) in BisA53. In prove the ability of cells to simultaneously use diverse electron- CGA009, the vanadium nitrogenase gene cluster (RPA1370– donating substrates for hydrogen production. Finally, our discovery 1403) is part of an indel that is not present in the highly from its genome sequence that BisB18 can produce substantial homologous Rhodopseudomonas palustris TIE-1 genome se- amounts of hydrogen by pyruvate fermentation in the dark (Table quence, inferring that vanadium nitrogenase was acquired by S4), as well as via nitrogenase in light, has obvious implications for
Oda et al. PNAS ͉ November 25, 2008 ͉ vol. 105 ͉ no. 47 ͉ 18547 Downloaded by guest on September 25, 2021 developing a hydrogen production process that operates continu- Genome Sequencing and Analyses. Genomic DNA was sequenced using a ously over a 24 h day. conventional whole-genome shotgun strategy. The Gentra genomic DNA purification kit (Gentra Systems) was used to isolate DNA from cells grown in 30 ml PM containing 10 mM succinate and 0.3% yeast extract. Library con- Materials and Methods struction, sequencing, and assembly were performed as described for Rhodo- Bacterial Strains and Growth Conditions. The four strains used for genome pseudomonas palustris strain CGA009 (10). Automated annotation was car- sequencing were isolated from freshwater sediments from De Biesbosch and ried out as described (36). Haren, The Netherlands (11). The Rhodopseudomonas strains were routinely cultivated anaerobically at 30 °C in closed 15 ml screw-cap tubes containing Comparative Genome Analysis. Orthologous and paralogous groups of pro- defined mineral medium (PM) (34). Carbon sources tested as growth sub- teins were identified using OrthoMCL (37), which provides a scalable method strates were added from separately prepared stock solutions. For strain HaA2, for constructing ortholog groups across multiple genomes using a Markov 0.05% yeast extract (wt/vol final concentration) was routinely added to Cluster algorithm as described (36). Proteins shared among the Rhodopseu- domonas genomes were searched against the KEGG database containing all growth media. For photoautotrophic growth with hydrogen and NaHCO3,PM eubacterial and archaeal complete genome peptide sets with the Rhodopseu- was supplemented with 1 M NiCl2. Cells were grown under nitrogen-fixing conditions in PM without ammonium sulfate (NFM) in tubes with a nitrogen domonas strain data sets removed (1,415,335 peptides from a total of 453 genomes) to identify top hits to other bacterial genomes. For this, blastp gas phase (25). For hydrogen production measurements, 27 ml anaerobic (v2.2.15) was run using an e-value cutoff of 1e-05, with low-complexity culture tubes containing 10 ml anaerobically prepared NFM (17 ml headspace) filtering (SEG) off. The phylogenetic analysis of 16S rRNA sequences was were used. Hydrogen production values and nitrogenase were determined carried out as described previously (7). from cultures in the exponential phase of growth (OD660 0.30–0.35) as de- scribed (25). Pyruvate (40 mM) fermentation was tested by growing cells ACKNOWLEDGMENTS. The Biological and Environmental Research (BER) pro- anaerobically in dark in PM medium that also contained 0.05% yeast extract. gram of the U.S. Department of Energy’s (DOE) Office of Science funded this Hydrogen yields under fermentative conditions were from cultures in the research. The Joint Genome Institute managed the overall sequencing effort. early exponential phase of growth (OD660 0.11–0.13). For absorption spectra Lawrence Livermore National Laboratory carried out genome finishing under analyses, cells were grown in PM/YPS medium (35) (1:1) at 30 °C under anaer- the auspices of the DOE. Computational annotation was carried out at the Oak obic, low-light conditions with halogen bulbs providing irradiance measured Ridge National Laboratory, managed by UT-BATTELLE for the Department of Energy. C.S.H. received support from the DOE Office of Science (BER Grant Ϫ2 Ϫ1 at the culture level as 8 mol photons m s . Cells grown to early stationary DE-FG02–07ER64482 and Office of Basic Energy Sciences Grant DE-FG02- phase were harvested by centrifugation and resuspended in 22.5% (wt/vol) 05ER15707) and from the U.S. Army Research Office (Grant W911NF-05–1- BSA. Light scattering at 650 nm was used to normalize the spectra. 0176).
1. Coleman ML, Chisholm SW (2007) Code and context: Prochlorococcus as a model for 20. Egland PG, et al. (1997) A cluster of bacterial genes for anaerobic benzene ring cross-scale biology. Trends Microbiol 15:398–407. biodegradation. Proc Natl Acad Sci USA 94:6484–6489. 2. Achtman M, Wagner M (2008) Microbial diversity and the genetic nature of microbial 21. Harwood CS, Burchhardt G, Herrmann H, Fuchs G (1999) Anaerobic metabolism of species. Nat Rev Microbiol 6:431–440. aromatic compounds via the benzoyl-CoA pathway. FEMS Microbiol Rev 22:439–458. 3. Cohan FM, Perry EB (2007) A systematics for discovering the fundamental units of 22. Harrison FH, Harwood CS (2005) The pimFABCDE operon from Rhodopseudomonas bacterial diversity. Curr Biol 17:R373–R386. palustris mediates dicarboxylic acid degradation and participates in anaerobic benzo- 4. Dufresne A, et al. (2003) Genome sequence of the cyanobacterium Prochlorococcus ate degradation. Microbiology 151:727–736. marinus SS120, a nearly minimal oxyphototrophic genome. Proc Natl Acad Sci USA 23. Romagnoli S, Tabita FR (2006) A novel three-protein two-component system provides 100:10020–10025. a regulatory twist on an established circuit to modulate expression of the cbbI region 5. Kettler GC, et al. (2007) Patterns and implications of gene gain and loss in the evolution of Rhodopseudomonas palustris CGA010. J Bacteriol 188:2780–2791. of Prochlorococcus. PLoS Genet 3:2515–2528. 24. Cosseau C, Batut J (2004) Genomics of the ccoNOQP-encoded cbb oxidase complex in 6. Rocap G, et al. (2003) Genome divergence in two Prochlorococcus ecotypes reflects 3 bacteria. Arch Microbiol 181:89–96. oceanic niche differentiation. Nature 424:1042–1047. 25. Oda Y, et al. (2005) Functional genomic analysis of three nitrogenase isozymes in the 7. Oda Y, et al. (2004) Analysis of diversity among 3-chlorobenzoate-degrading strains of Rhodopseudomonas palustris. Microb Ecol 47:68–79. photosynthetic bacterium Rhodopseudomonas palustris. J Bacteriol 187:7784–7794. 8. Okubo Y, Futamata H, Hiraishi A (2006) Characterization of phototrophic purple 26. Schaefer AL, et al. (2008) A new class of homoserine lactone quorum-sensing signals. nonsulfur bacteria forming colored microbial mats in a swine wastewater ditch. Appl Nature 454:595–599. Environ Microbiol 72:6225–6233. 27. Jiao Y, Newman DK (2007) The pio operon is essential for phototrophic Fe(II) oxidation 9. van der Woude BJ, et al. (1994) Anaerobic degradation of halogenated benzoic acids in Rhodopseudomonas palustris TIE-1. J Bacteriol 189:1765–1773. by photoheterotrphic bacteria. FEMS Microbiol Lett 119:199–207. 28. Jiao Y, Kappler A, Croal LR, Newman DK (2005) Isolation and characterization of a 10. Larimer FW, et al. (2004) Complete genome sequence of the metabolically versatile genetically tractable photoautotrophic Fe(II)-oxidizing bacterium, Rhodopseudomo- photosynthetic bacterium Rhodopseudomonas palustris. Nat Biotechnol 22:55–61. nas palustris strain TIE-1. Appl Environ Microbiol 71:4487–4496. 11. Oda Y, et al. (2002) Genotypic and phenotypic diversity within species of purple nonsulfur 29. Rabus R, et al. (2005) The genome sequence of an anaerobic aromatic-degrading bacteria isolated from aquatic sediments. Appl Environ Microbiol 68:3467–3477. denitrifying bacterium, strain EbN1. Arch Microbiol 183:27–36. 12. Lawrence JG, Hendrickson H (2005) Genome evolution in bacteria: order beneath 30. Badger MR, Bek EJ (2008) Multiple Rubisco forms in proteobacteria: Their functional chaos. Curr Opin Microbiol 8:572–578. significance in relation to CO2 acquisition by the CBB cycle. J Exp Bot 59:1525–1541. 13. Braatsch S, et al. (2006) Rhodopseudomonas palustris CGA009 has two functional ppsR 31. McLuskey K, Prince SM, Cogdell RJ, Isaacs NW (2001) The crystallographic structure of genes, each of which encodes a repressor of photosynthesis gene expression. Biochem- the B800–820 LH3 light-harvesting complex from the purple bacteria Rhodopseudo- istry 45:14441–14451. monas acidophila strain 7050. Biochemistry 40:8783–8789. 14. Giraud E, et al. (2002) Bacteriophytochrome controls photosystem synthesis in anoxy- 32. Harwood CS (2008) in Bioenergy, eds Wall JD, Harwood CS, Demain A (ASM Press, genic bacteria. Nature 417:202–205. Washington, DC), pp 259–271. 15. Hartigan N, et al. (2002) The 7.5-Å electron density and spectroscopic properties of a 33. Rupprecht J, et al. (2006) Perspectives and advances of biological H production in novel low-light B800 LH2 from Rhodopseudomonas palustris. Biophys J 82:963–977. 2 microorganisms. Appl Microbiol Biotechnol 72:442–449. 16. Scheuring S, Gonc¸alves RP, Prima V, Sturgis JN (2006) The photosynthetic apparatus of 34. Kim M-K, Harwood CS (1991) Regulation of benzoate-CoA ligase in Rhodopseudomo- Rhodopseudomonas palustris: Structures and organization. J Mol Biol 358:83–96. 17. Tharia HA, Nightingale TD, Papiz MZ, Lawless AM (1999) Characterisation of hydro- nas palustris. FEMS Microbiol Lett 83:199–204. phobic peptides by RP-HPLC from different spectral forms of LH2 isolated from Rps. 35. Wall JD, Weaver PF, Gest H (1975) Gene transfer agents, bacteriophages, and bacte- palustris. Photosynth Res 61:157–167. riocins of Rhodopseudomonas capsulata. Arch Microbiol 105:217–224. 18. Evans K, et al. (2005) A bacteriophytochrome regulates the synthesis of LH4 complexes 36. Starkenburg SR, et al. (2008) Complete genome sequence of Nitrobacter hamburgensis in Rhodopseudomonas palustris. Photosynth Res 85:169–180. X14 and comparative genomic analysis of species within the genus Nitrobacter. Appl 19. Giraud E, et al. (2005) A new type of bacteriophytochrome acts in tandem with a Environ Microbiol 74:2852–2863. classical bacteriophytochrome to control the antennae synthesis in Rhodopseudomo- 37. Li L, Stoeckert CJ, Jr., Roos DS (2003) OrthoMCL: Identification of ortholog groups for nas palustris. J Biol Chem 280:32389–32397. eukaryotic genomes. Genome Res 13:2178–2189.
18548 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0809160105 Oda et al. Downloaded by guest on September 25, 2021