One Carbon Metabolism in SAR11 Pelagic Marine Bacteria

The SAR11 Alphaproteobacteria are the most abundant heterotrophs in the oceans and are believed to play a major role in mineralizing marine dissolved organic carbon. Their genomes are among the smallest known for free-living heterotrophic cells, raising questions about how they successfully utilize complex organic matter with a limited metabolic repertoire. Here we show that conserved genes in SAR11 subgroup Ia (Candidatus Pelagibacter ubique) genomes encode pathways for the oxidation of a variety of one-carbon compounds and methyl functional groups from methylated compounds. These pathways were predicted to produce energy by tetrahydrofolate (THF)-mediated oxidation, but not to support the net assimilation of biomass from C1 compounds. Measurements of cellular ATP content and the oxidation of 14C-labeled compounds to 14CO2 indicated that methanol, formaldehyde, methylamine, and methyl groups from glycine betaine (GBT), trimethylamine (TMA), trimethylamine N-oxide (TMAO), and dimethylsulfoniopropionate (DMSP) were oxidized by axenic cultures of the SAR11 strain Ca. P. ubique HTCC1062. Analyses of metagenomic data showed that genes for C1 metabolism occur at a high frequency in natural SAR11 populations. In short term incubations, natural communities of Sargasso Sea microbial plankton expressed a potential for the oxidation of 14C-labeled formate, formaldehyde, methanol and TMAO that was similar to cultured SAR11 cells and, like cultured SAR11 cells, incorporated a much larger percentage of pyruvate and glucose (27–35%) than of C1 compounds (2–6%) into biomass. Collectively, these genomic, cellular and environmental data show a surprising capacity for demethylation and C1 oxidation in SAR11 cultures and in natural microbial communities dominated by SAR11, and support the conclusion that C1 oxidation might be a significant conduit by which dissolved organic carbon is recycled to CO2 in the upper ocean.


Introduction
C1 metabolism takes place through a network of interrelated biochemical reactions that involves the transfer of one-carbon units from one compound to another.C1 units can be donated in the form of methyl (-CH 3 ), methylene (-CH 2 -), methenyl (-CH = ), formyl (-CHO) and formimino (-CH = NH) groups [1,2].A few specialized bacteria oxidize methyl groups and C1 compounds, such as methanol, formaldehyde, formate and methylamine, to derive energy and cellular carbon.The most well known of these organisms are methylotrophs, which assimilate C1 carbon into biomass via the ribulose monophosphate (RuMP) or serine cycle pathways [3][4][5][6][7].Less well known are organisms that have C1 oxidation pathways for energy production, but lack pathways for the net synthesis of biomass from C1 precursors [8].
Marine dissolved organic carbon (DOC) includes a diverse array of C1 and methylated compounds that are potential substrates for C1 oxidation.The most common methylated compounds in marine environments are osmolytes such as GBT, TMAO, and DMSP [9][10][11].Methanol is a major component of oxygenated volatile organic chemicals in the oceans and atmosphere [12,13].Air measurements over the Pacific Ocean indicate that sea surface methanol concentrations are about 100 nM and that central ocean regions are net sinks for methanol deposited from the atmosphere [12][13][14].Formaldehyde is ubiquitous in seawater.Likely sources of seawater formaldehyde are atmospheric deposition from industrial emissions and the photo-oxidation of atmospheric hydrocarbons [15,16], and the photo-oxidation of dissolved organic carbon in the ocean surface [17].The metabolism of methylated compounds in mammals also produces formaldehyde [18].Formaldehyde is a key reactive intermediate in bacterial metabolism of C1 growth substrates like methane or methanol [19][20][21], and it is also a central intermediate of GBT methyl group oxidation [22].Due to its nonspecific reactivity with proteins and DNA, formaldehyde is toxic to cells, and thus many studies have examined mechanisms by which organisms can remove this potentially lethal compound [23,24].
Marine bacteria of the SAR11 clade are the most abundant aerobic, free-living, heterotrophic bacteria in ocean surface waters [25,26].SAR11 was first discovered in the Sargasso Sea in 1990 [27] and is now considered to be one of the most successful organisms on the planet.Candidatus Pelagibacter ubique strain HTCC1062, the first cultured SAR11 strain, has one of the most compact genomes known for a free-living organism (1,308,759 base pairs), and has dispensed with many functions that are common in other free-living bacteria, apparently in response to selective pressure for small genome size [28].The SAR11 clade is divided into subclades that have distinct temporal and spatial distributions in the environment that are believed to represent ecotypes or species [29].Most strains now being studied in laboratories belong to the Group Ia subclade, which is common in the ocean euphotic zone [29].
In this study, we observed a suite of genes for demethylation and C1 oxidation in three SAR11 Ia genomes (Ca.P. ubique strains HTCC1062, HTCC1002, and HTCC7211) for which no role in SAR11 had yet been identified.Genomic and metagenomic data were used to explore the potential for C1 metabolism in the three SAR11 isolates and natural SAR11 populations, and experiments with pure cultures of cells were used to confirm predictions.Together, our data demonstrate that SAR11 cells are capable of oxidizing a range of C1 compounds and methyl groups from methylated compounds to produce energy.

Comparative Genome Analyses and Proposed Metabolic Pathways
Genes in the HTCC1062 genome associated with C1 oxidation are shown in Figure 1.The proposed metabolic pathways for these genes are discussed in detail below (Fig. 2), although for clarity we include only the HTCC1062 gene identifiers (e.g, ''SAR11_'') in the text.Table S1 lists C1 metabolism genes by gene identifier, COG number, organism, and function to reduce confusion caused by the common practice of assigning different gene names to homologous genes that have related functions.To evaluate the conservation and diversity of C1 oxidation pathways in SAR11 we examined the distribution of genes for C1 metabolism among three SAR11 genomes of the Group Ia subclade, and these genes were found in all three SAR11 Ia genomes (Table 1).A survey of the global ocean survey (GOS) metagenomic data for SAR11 genes involved in C1 metabolism also supported the conclusion that these genes occur frequently in SAR11 genomes.Averaged across ocean sampling locations worldwide, the ratio of SAR11 C1 genes, relative to SAR11 recA, ranged from 0.2 to 2.1 (Fig. S1 and Table S2).
Putative genes that encode C1 oxidation activities are found in many bacterial phyla, archaea and eukaryotes.We used the phylogenomic pipeline HAL [35] to study the distribution of C1 oxidation genes in the class Alphaproteobacteria and found evidence for many of these genes throughout the genomes investigated (Fig. S2).For example, homologs of fdhF and fdhD were observed in 77 and 58 of 127 genomes, respectively.However, outside of the methylotrophs, which can grow using C1 compounds as a sole carbon and energy source, C1 oxidation has received relatively little attention.FDH genes have been reported in Rhizobium japonicum and Agrobacterium tumefaciens [36][37][38].Interestingly, several non-methylotrophs, including A. tumefaciens, are capable of utilizing methylamine as a nitrogen source, but not as a sole carbon source.As with SAR11 Ia, these bacteria contain the proposed THFlinked C1 oxidation pathway and lack specialized pathways for the assimilation of C1 compounds [8,36].These observations suggest that, as we report below for SAR11 Ia strains, R. japonicum and A. tumefaciens, and probably many other bacteria, may utilize C1 and methylated compounds for energy production.
All three SAR11 Ia genomes possess an alcohol dehydrogenase (ADH) gene (SAR11_1287; Fig. 1B) that was initially annotated as methanol dehydrogenase.We propose that this enzyme catalyzes the oxidation of short chain alcohols, including methanol, to the corresponding aldehydes (Fig. 2B).In the case of methanol, the oxidation product is formaldehyde, which can be converted to CO 2 by the methyl-THF linked oxidation pathway described above.This gene encodes a member of the iron-containing alcohol dehydrogenase (Fe-ADH, PF00465) protein family found in many microorganisms.Oxidation of methanol rarely is observed in nonmethylotrophic bacteria that lack the classical pyrroloquinolinequinone (PQQ)-containing methanol dehydrogenase.Methanol dehydrogenase activity has been confirmed in ADH proteins from some Firmicutes and Archaea [39][40][41].A phylogenetic tree of the Fe-ADH proteins from three SAR11 Ia strains (Fig. 3, light green) and closely related homologs from other organisms shows that the proteins with demonstrated methanol dehydrogenase activity (arrows) are more highly diverged than their close relatives, and that the SAR11 Fe-ADH proteins branch nearby (Fig. 3).As we report below (Tables 2 and 3), physiological evidence from ATP assays supports the conclusion that the SAR11 Fe-ADH is not specific for methanol oxidation, but can oxidize other short chain primary alcohols as well.Similar conclusions were reported previously for other bacterial ADH genes in the same family [42][43][44].Only short-chained, primary alcohols (methanol, ethanol, and 1-propanol) stimulated ATP production in HTCC1062.
A predicted GBT degradation pathway operon was identified in the SAR11 Ia genomes.In this pathway, the three N-methyl groups of GBT are removed in sequence by three enzymes: betaine-homocysteine methyltransferase (BHMT), dimethylglycine dehydrogenase, and sarcosine dehydrogenase (encoded separately by SAR11_1173, SAR11_1253, and SAR11_1221 in HTCC1062).These methyl groups have different predicted fates.The first methyl group is transferred to methionine by BHMT, an enzyme that is common and well studied in mammals, but is rarely found in prokaryotes [45,46].The second and third methyl groups are predicted to be transferred to THF in reactions that are coupled to partial oxidation and then further oxidized to CO 2 by the THF-linked oxidation pathway described above (Fig. 2A, C).A previous study showed that the uptake and metabolism of GBT by bacteria-sized organisms in seawater was correlated with salinity [47].Consistent with that study, metagenomic analysis showed relatively constant ratios of SAR11 GBT gene frequencies in ocean populations, and absence of these genes in metagenomes from fresh water (Table S2, ID GS20).
Genes for methylamine dehydrogenase are missing from SAR11 Ia genomes, but they contain genes necessary to encode an alternative, glutamate-mediated pathway for methylamine oxidation [48].Latypova and colleagues showed that N-methylglutamate synthase (NMGS) and N-methylglutamate dehydrogenase (NMGD), encoded by mgsABC and mgdABCD in Methyloversatilis universalis, participate in methylamine oxidation [48].When the M. universalis NMGS protein sequences were blasted against the predicted HTCC1062 proteome, it returned a glxBCD operon (SAR11_1313-1315) as the top hit.This operon is annotated as a glutamate synthase and shares gene order with the M. universalis NMGS operon.We hypothesize that this operon may encode a holoenzyme that has NMGS activity.NMGS catalyzes transfer of a methyl group from methylamine to glutamate to produce Nmethyglutamate as a product.
We postulate that the next step in the pathway, oxidation of Nmethyglutamate to glutamate and 5, 10-methylene-THF, with the reduction of FAD to FADH 2 , is catalyzed by the products of SAR11 genes that are annotated as heterotetrameric forms of sarcosine oxidase (Fig. 2D).Two paralogous operons of four sarcosine oxidase genes, soxBDAG (SAR11_1064-1067) and soxB 2 D 2 A 2 G 2 (SAR11_1281-1284), are located in separate locations of HTCC1062 genomes with the same gene orders and conserved functional domains as those encoding NMGD in M. universalis (Fig. 1C).Metagenome surveys showed that these two sets of paralogs are highly represented in SAR11 populations globally (Fig. S1).We postulate that one of these sets of paralogs encodes the sarcosine oxidase holoenzyme, and the other encodes NMGD; however, which operon encodes which function is unclear.Heterotetrameric sarcosine oxidases (a.k.a sarcosine dehydrogenase) participate in the GBT degradation pathway by converting sarcosine to glycine.In HTCC1062, SAR11_1304 is predicted to encode the monomeric form of sarcosine oxidase, and is adjacent to two genes annotated as an aminomethyltransferase (AMT) and a GBT transporter (SAR11_1303 and SAR11_1302; Fig. 1E).The presence of two paralogous sarcosine oxidase operons and an individual gene encoding a monomeric sarcosine oxidase is noteworthy considering that paralogs are rare in highly compact SAR11 genomes [28].We postulate that SAR11 Ia strains may use these genes for the reactions described above and possibly for the demethylation of unknown substrates in oxidation reactions that are likely to produce the intermediate 5, 10-methylene-THF [49].
A different pathway for methylamine demethylation that involves the intermediate gamma-glutamylmethylamide was also postulated by Latypova and colleagues [48].We found the homolog of gamma-glutamylmethylamide synthetase (GMAS) from this pathway in HTCC1062, which was annotated as glnT, a glutamine synthetase III (SAR11_1316, Fig. 1C) and shares the conserved Gln-synt_C domain (Pfam PF00120.17) with GMAS.However, the complete pathway has not been identified and the enzyme was considered to be not essential for oxidation of methylamine in M. universalis [48].
The three SAR11 Ia genomes include a family of paralogous AMT genes (Fig. 1E).Phylogenetic analysis provided strong support for placing the four AMT genes from HTCC1062 into three functionally distinct subgroups (Fig. 4).SAR11_0666 encodes the glycine cleavage system T-protein (GcvT), part of the glycine cleavage multi-enzyme complex (GCV), which is present in most bacteria and mitochondria [50].GCV catalyzes the degradation of glycine to form 5, 10-methylene-THF, CO 2 and NH 3 [51] (Fig. 2F).SAR11_0246 encodes DmdA, a member of the AMT protein family that catalyzes removal of the first methyl group from DMSP to produce methylmercaptopropionate (MMPA) in a pathway for DMSP catabolism [52,53] (Fig. 2G).However, the enzyme that subsequently demethylates MMPA to mercaptopropionate (MPA) has not been identified.
SAR11_1265 and SAR11_1303 are annotated as probable AMTs with unknown substrates.Three genes (opuAB, opuAA, and opuAC) are located near SAR11_1303 and encode putative homologs of the GBT transport system permease protein, GBT transport ATP-binding protein and substrate-binding region of ABC-type GBT transport system (Fig. 1E).Therefore, we postulate that SAR11_1303 may encode an AMT involved in GBT metabolism.Notably, SAR11_1265 is adjacent to the putative metF (SAR11_1264), encoding a homolog of methylene-THF reductase.MetF in Methylobacterium chloromethanicum CM4 enables the oxidation of methyl-THF to methylene-THF [20].SAR11_1265 is closely related to SAR11_0246 (dmdA) (Fig. 4), and was observed at similar abundances in GOS metagenomic analysis (Fig. S1), leading us to speculate that this gene might catalyze removal of the second methyl group in the DMSP degradation pathway.All of the four AMTs tended towards a 1:1 ratio with single-copy gene recA (Fig. S1), suggesting that these genes may be part of the SAR11 Ia core genome.
Another gene, SAR11_0621, was initially annotated as an AMT, but later was identified as ygfZ, which encodes a folatebinding protein.Studies on YgfZ in Escherichia coli and Arthrobacter globiformis revealed THF-binding folds in the structure of this protein, but otherwise little similarity to the GcvT family of AMT proteins [54,55].Recent studies indicated that YgfZ in E. coli has methylase activity [56] or may be involved in the regulation of C1 metabolism [54].There is currently no direct evidence for a role in C1 metabolism for YgfZ in SAR11.Some of the three SAR11 Ia genomes had additional genes implicated in C1 metabolism (Table 1).Genes encoding glutathione-dependent formaldehyde activating enzyme (GFA), glutathione-dependent formaldehyde dehydrogenase (GD-FALDH) and S-formyl-glutathione hydrolase (FGH) are present in HTCC7211, but not in HTCC1062 or HTCC1002.In Paracoccus denitrificans, these proteins catalyze the well-studied glutathione (GSH) dependent pathway that converts formaldehyde to CO 2 [7,[57][58][59].We note that some of the predicted C1 oxidation functions of these genes are redundant with genes for the THF-linked oxidation pathway that are also present in the HTCC7211 genome; however, unlike strain HTCC1062, strain HTCC7211, an isolate from the oligotrophic Sargasso Sea, is predicted to be able to oxidize formaldehyde by using GFA to catalyze the first step in the pathway.
Methane monooxygenases, which catalyze the first reaction of methane oxidation pathways, were not present in any SAR11 genome, indicating that these cells are unlikely to be methane oxidizers.Karl and coworkers reported that methane is produced from methylphosphonate in seawater by the activity of the microbial C-P lyase pathway [60].This pathway is present in the genome of HTCC7211, but there are no known metabolic pathways by which methyl groups from methylphosphonate can be diverted into THF-linked C1 oxidation.

Direct Experimental Evidence for C1 Oxidation in SAR11 Strain HTCC1062
Genome analysis suggested that C1 metabolism evolved in SAR11 Ia for energy production, rather than as a means to accumulate biomass.Testing this hypothesis was challenging because SAR11 Ia cells require a variety of unusual organic growth factors and cannot be cultured on ''sole'' carbon sources, a common paradigm in microbiology.Some required compounds, notably glycine, can be oxidized to produce energy as well as being needed for biomass production [61].Thus, we were not surprised when the addition of various C1 and methylated compounds did not change growth rates or yields of cultures (data not shown).To test for C1 oxidation activities predicted by genome analysis, we turned to direct measurements of energy production and substrate oxidation.
Measurements of cellular ATP content supported the conclusion that HTCC1062 can produce energy from a wide variety of C1 and methylated compounds (Table 3).ATP levels were assayed in HTCC1062 cells that were first grown in the presence of C1 and methylated compounds to be certain that pathways involving C1 oxidation genes were induced.Concentrations of compounds used for ATP assays and 14 C experiments (below) were determined in advance by growth assays, as described in the materials and methods.Pyruvate, which was shown previously to be actively metabolized as an energy and carbon source by HTCC1062 cells [62], was used as a positive control.ATP content increased in cells incubated with all of the methylated compounds tested with the exception of formate, relative to negative controls.Methanol and TMAO caused the greatest increases in ATP content (3-and 2.4-fold, respectively).We speculate that formate did not enhance ATP levels because it was not transported into cells.
Consistent with predictions from the genome sequence, radioisotope studies with 14 C-[methyl]-GBT, 14 C-TMA, 14  The phylogenomics pipeline HAL and manual searches were used to detect orthologs among the genomes.Genes for C1 oxidation were present in all three genomes.HTCC7211 possesses three genes for the glutathione (GSH) dependent C1 oxidation pathway that are not present in the other two SAR11 Ia genomes.doi:10.1371/journal.pone.0023973.t001methanol and 14 C-formaldehyde demonstrated that HTCC1062 oxidized methyl groups from methylated compounds and C1 compounds to CO 2 , but incorporation of the compounds into biomass was negligible (Fig. 5).Among these compounds, the oxidation rate was fastest with 5 mM 14 C-TMA (3.82 nmol610 10 cells 21 6h 21 ), while 100 nM 14 C-formaldehyde was oxidized to 14 CO 2 at the lowest rate (0.01 nmol610 10 cells 21 6h 21 ), likely reflecting the lack of formaldehyde activating enzyme in HTCC1062.The rate of 20 mM methanol oxidized to 14 CO 2 was 0.50 nmol610 10 cells 21 6h 21 .The ability of cultured Ca.P. ubique to oxidize C1 compounds, and the conservation of genes for C1 metabolism in streamlined SAR11 genomes, suggests that C1 oxidation pathways may contribute significantly to the energy budget of these cells in nature.To examine this hypothesis, we measured the potential rates at which 14 C-labeled C1 and methylated compounds were oxidized and incorporated into biomass by cells concentrated from the Sargasso Sea upper euphotic zone during the period of summer stratification.Long-term time series measurements have shown that SAR11 cells range from 30-40% of cells at this ocean site, and that the distinctive microbial community that forms in the upper euphotic zone during the summer is dominated by the SAR11 Ia subclade [25,29].The data from bacterioplankton populations collected in the western Sargasso Sea revealed rates of carbon assimilation and oxidation of the same order observed with pure SAR11 cultures in the laboratory (Fig. 6 and [62]).Similar methanol oxidation rates (,0.8 -2.5 nmol610 10 cells 21 6h 21 ) were measured in off-shelf northeast Atlantic seawaters [14], which are also dominated by SAR11 cells [25].Notably, in the natural bacterioplankton community less than 6 % of the 14 Cmethanol, 2 % of 14 C-formate, and 3 % of 14 C-formaldehyde were assimilated into biomass, with the remainder oxidized to CO 2 by the natural community.In contrast, 35 % of glucose and 27 % of pyruvate were assimilated into bacterioplankton biomass in the field experiments (Fig. 6), similar to the fraction of organic carbon assimilation into biomass observed previously with growing SAR11 cultures and typical of bacterial organic carbon assimilation in general [62].One noteworthy difference between the results obtained with cultured strain HTCC1062 and measurements made in the field is the higher rate of formaldehyde oxidation in the field study.During comparative genome sequence analysis the protein GFA (formaldehyde activating enzyme) was noted in the HTCC7211 genome (Table 1), an isolate obtained from the Sargasso Sea.It may be that ecotypes of SAR11 found in the Sargasso Sea summer upper euphotic zone microbial community have a potential to oxidize formaldehyde that is not found in coastal isolates.Overall, these findings suggest that energy production by C1 oxidation, rather than methylotrophy, in which a large fraction (i.e., 62 %; [63]) of the C1 compound is incorporated into biomass, is the predominant mode of C1 oxidation in surface waters of temperate oligotrophic oceans.

Conclusions
It is a paradox that the SAR11 clade evolved small genomes while becoming the most successful heterotrophs known, because DOC is a complex substrate that would appear to require complex metabolic pathways for oxidation.The most obvious solution to this conundrum is substrate specialization by SAR11 and other heterotrophic species in microbial communities, dividing the oxidative side of the carbon cycle into niches.In this report we show that a considerable part of the SAR11 Ia genomes is devoted to C1 metabolism that produces energy but not biomass.C1 and methylated compounds have not generally been regarded as a large fraction of the marine carbon cycle, but there have been a number of reports of C1 compounds and C1 oxidation activity in marine systems [14,47].We speculate that C1 metabolism evolved in SAR11 because it gave these cells the ability to catabolize a variety of compounds -ranging from low molecular weight photolysis products to osmolytes carrying multiple methyl groups, such as DMSP and GBT.DMSP catabolism in SAR11, and its use as a source of reduced sulfur for growth, has been the subjects of previous studies that did not examine the fate of methyl groups [53,64].
The concept of cells producing energy but not biomass from C1 compounds has received little attention.The best analogy is the term ''carboxidovory'', which was coined to describe the metabolism of cells that produce energy by oxidizing carbon monoxide, and to distinguish them from ''carboxidotrophs'' [65,66], which can use carbon atoms from carbon monoxide for the net assimilation of biomass.Based on this precedent, we propose a new term, ''methylovores'', to distinguish cells, such as SAR11 and likely many other bacteria, that can utilize C1 compounds as a source of energy, from methylotrophs, which are able to use C1 compounds as sole sources of energy and carbon.
Phylogenomic studies have placed the mitochondria deep within the Alphaproteobacteria, close to Rickettsiales and the SAR11  clade [67].Although respiration is a central feature of mitochondria that is assumed to have driven the original endosymbiotic event, mitochondria make complex contributions to cellular metabolism, including vitamin biosynthesis [68], and the metabolism of C1 compounds [69,70], which evidence supports as being present in the protomitochondrion [71].In light of the data presented here it is not unreasonable to speculate that the C1 metabolism of mitochondria was contributed by free-living methylovorous ancestors of modern SAR11.The findings we report show that SAR11 Group Ia strains can produce cellular energy from a broad range of C1 and methylated compounds by oxidative metabolism that we here refer to as methylovory to distinguish it from methylotrophy.The data we present also show that the potential for methylovory was highly expressed in a natural oligotrophic ocean surface microbial community dominated by SAR11.Based on these findings, we speculate that C1 oxidation pathways contribute significantly to the marine carbon cycle and in part explain the broad success of SAR11.

Bioinformatics analysis and phylogeny
Gene calls and functional assignment of SAR11 Group Ia genomes were performed by an automated pipeline at Oregon State University's Center for Genome Research and Biocomput-ing (http://bioinfo.cgrb.oregonstate.edu/microbes/index.html).Conserved domains were determined with Pfam [72].KEGG [73] and MicrobesOnline [74] were used for predicting physiological and metabolic pathways.
For the phylogeny of the Fe-ADHs, Pfam was used to confirm the orthology of the Fe-ADH domain in the three SAR11 strains with that of the known methanol dehydrogenase genes (Pfam Fe-ADH PF00465).The phylogenetic tree was then constructed using these six sequences and those of over 70 of the 100 top hits identified using PSI-BLAST with the HTCC1062 gene (YP_266695) as the query sequence.Sequences were aligned using MUSCLE [75,76], and manually edited to remove columns with .85%gaps.Substitution modeling was completed using ProtTest [77].The alignment was then analyzed with RAxML [78] using the WAG substitution model and nodal support was estimated based on 100 bootstrap replications.
Phylogeny of the AMTs was performed by MicobesOnline website and MacVector software.The top 10 matched protein sequences of each AMT (YP_265671, YP_266089, YP_266673, YP_266710) were downloaded using BLAST tools from Microbe-sOnline.AMTs were aligned using the ClustalW module of MacVector version 10.5.2, and Phylogenetic trees were constructed using the neighbor-joining method.One-hundred bootstrap replicates were used to estimate the robustness of branches in the phylogenetic tree.Accession numbers are in Table S3.Homology assessment of methylovory genes was completed using the HAL pipeline [36,79] (http://aftol.org/pages/Halweb3.htm; http://sourceforge.net/projects/bio-hal/)followed by manual examination of predicted orthologous clusters.For this study, the HAL pipeline directed the following analyses: Protein sequences from 127 publically available Alphaproteobacteria genomes (for a complete list, see [67]) were imported in FASTA format from IMG (http://img.jgi.doe.gov) and subjected to all vs. all BLASTP with the output E-values provided to the program MCL [80], which grouped proteins into orthologous clusters using 13 inflation parameters from 1.1 -5.0.For homology assessment in Figure S2, we examined clusters formed at the relatively

Metagenomic analysis
The reciprocal best BLAST (RBB) approach [81] was used to identify environmental fragments in the global ocean survey (GOS) database originating from SAR11 and encoding genes related to C1 and methyl group oxidation.The amino acid sequences for the genes discussed in this paper were searched with BLASTP (e-value threshold = 1610 25 ) against peptide sequences called from GOS metagenomic nucleotide reads.The taxonomies of matching peptides were determined by querying the peptide against the NCBI non-redundant protein database (NR) using BLASTP (e-value threshold = 10).Sequences were only retained if this second search against NR found one of the three Ca.P. ubique strains to be the best match to the environmental peptide.Peptides were lastly searched with BLASTP (e-value threshold = 1610 25 ) against all Ca.P. ubique strain HTCC1062 amino acid sequences to ensure that the initial Ca.P. ubique query protein, rather than a paralog, was the best match to the environmental peptide sequence.The number of environmental peptides passing the RBB test, or hits, per gene at each GOS sampling site were normalized to the length of the single-copy essential gene recA and the number of hits to recA at the same sampling site using the formula P g = (L recA / L g )6(H g / H recA ), where P g is the percentage abundance related to recA for gene g, L recA is the length of recA in base-pairs (bp), L g is the length of gene g in bp, H g is the number of hits for gene g, and H recA is the number of hits to recA [82].GOS sampling sites that had fewer than five hits to recA were removed from downstream analysis.

ATP measurements
HTCC1062 cultures for ATP assays were grown in autoclaved, filtered seawater amended with 1 mM NH 4 Cl, 100 mM KH 2 PO 4 , 1 mM FeCl 3 , 80 mM pyruvate, 40 mM oxaloacetate, 40 mM taurine, 1 mM GBT, 50 mM glycine, 50 mM methionine, and excess vitamins [62].C1 and methylated compounds to be tested were added as follows: 1 mM GBT, 20 mM formate, 10 mM methylamine, 5 mM TMAO, 1 mM DMSP, 20 mM methanol and 100 nM formaldehyde.The concentration of GBT used in the assay was the same as routinely used in Ca.P. ubique growth media (1 mM).The concentrations of other compounds utilized in assays were chosen by testing the growth of HTCC1062 with differing concentrations of each compound and selecting concentrations that were not inhibitory (Fig. S3).Cultures were grown in the dark at 16uC.When cultures entered early stationary phase (,3610 8 cells mL 21 ), 10 mL of each culture were collected and cells were harvested via centrifugation (50 min at 43,700 g, 10uC).Following centrifugation, cells were washed twice in artificial seawater (ASW) [62] and finally resuspended in 5 mL ASW.Then cells were distributed into 1.7 mL tubes (500 mL in each tube) and starved overnight (20 hours in the dark, 16uC).Test treatments and controls were done in triplicate with identical cell suspensions.After 2 hours incubation in the dark, ATP content was measured using a luciferase-based assay (BactTiter Glo, Promega, Madison, WI) as follows: 90 mL of BactTiterGlo reagent were dispensed into white 96 well plates (Tissue culture-treated, BD Biosciences, San Jose, CA).20 mL of each sample were added per well, and luminescence was measured after 4 min using a multi-functional plate reader (TECAN, Infinite M200) with a 1 s integration and 10 ms settle time.An ATP standard curve was used to calculate the concentration of ATP in the samples.Student's t-test was used to assess statistical significance (p-value,0.01) between controls and treatments.The ATP measurements for other alcohols (20 mM ethanol, 20 mM 1-propanol, 20 mM 2-propanol, 20 mM 1-butanol, 20 mM 2-pentanol, and 20 mM iso-amyl alcohol) were performed as described above.

Radioisotope assays
To distinguish between 14 C-labeled compounds incorporated into biomass and oxidized to 14 CO 2 , a method was devised for volatile 14 C compounds (e.g., 14 C-labeled methanol).Trichloroacetic acid (TCA)-precipitation was used to assess 14 C-incorporation into cellular material, and 14 CO 2 was precipitated by addition of NaOH, Na 2 CO 3 and BaCl 2 , forming Ba 14 CO 3 and BaOH.Filtration removed unincorporated 14 C compounds and BaOH, thus minimizing potential quenching effects of the base.
For experiments with radioisotopes, HTCC1062 was grown in seawater medium as described above for ATP measurements.Cultures were amended with unlabeled test compounds to induce activities (5 mM TMA, 20 mM methanol, or 100 nM formaldehyde; 1 mM GBT was in the seawater medium).Cells were harvested in log phase by centrifugation (1 hour at 43,700 g, 10uC) and resuspended in ASW to about 4610 7 cells mL 21 .Negative controls (''Killed'') were incubated in 10% formalin for 1 hour before the addition of the isotope to the sample. 1 mM 14 C-[methyl]-GBT, 5 mM 14 C-TMA, 20 mM 14 C-methanol, or 100 nM 14 C-formaldehyde were added to both ''Live'' and ''Killed'' culture samples.Inoculated cultures (4 mL) were aliquoted into 40 mL sealed vials.Samples were incubated at room temperature.At each time point, reagents were added to cultures using syringes inserted through stoppers.For 14 Cincorporation, 2.2 mL 100% w/v cold TCA was added; for trapping 14 CO 2 , 1 mL 1N NaOH, 0.5 mL 0.1 M Na 2 CO 3 and 1 mL 1 M BaCl 2 were added.All samples were collected by filtration after incubation at 4uC for 12 hrs.Filters were transferred to vials containing 15 mL Ultima Gold TM XR scintillation fluid (Perkin-Elmer) and kept in dark overnight before counting (Beckman LS-6500 liquid scintillation counter).
For field studies of radiolabeled compound utilization by bacterioplankton in the western Sargasso Sea, seawater was collected from 10 m at Bermuda Hydrostation S using Niskin bottles, and transferred to an acid-washed, autoclaved polycarbonate carboy.For each experiment, microbial plankton were concentrated from 80 L seawater by tangential flow filtration to a final volume of 600 mL, and isotopic labeling was carried out as described above.The concentrations of the tested compounds were 3 mM glucose, 1.7 mM pyruvate, 0.3 mM formate, 1.1 mM formaldehyde, 50 mM methanol and 0.5 mM TMAO.A high concentration of methanol was used because the specific activity of the labeled compound was very low.Some unexpected evaporation of the radio-labeled compounds occurred during transport causing some variability between concentrations of the other compounds used.Nevertheless, in all cases the concentrations used were likely to be substantially higher than typical seawater concentrations, and the results represent potential rates of compound utilization.

Supporting Information
Figure S1 The abundance of SAR11 C1 metabolism genes in GOS data, relative to SAR11 recA genes.Genes were identified as SAR11 by a reciprocal best BLAST (RBB) approach.SAR11 C1 genes with frequencies less than SAR11 recA (,1x) may indicate that only subpopulations of SAR11 cells possess that gene; genes greater than 1x suggest that multiple copies of that gene are present per cell.Boxes encompass points between the 25th and 75th percentiles, with the median represented as a thick horizontal line.Whiskers span the minimal distance needed to include all points within 1.5 x the interquartile range beyond the interquartile boundary, with points outside of this range rendered individually as circles.For each gene, n = 40.Abbreviations: adj., genomically adjacent genes.(TIF) Table S1 Methylovory pathways and associated genes in Candidatus Pelagibacter ubique SAR11 HTCC1062.Table includes genes involved in methylovory in SAR11 HTCC1062 and their associated homologs in other organisms mentioned in the text of the paper.Associated proteins are grouped together.Identifiers and annotated functions are taken from the Integrated Microbial Genomes database (IMG) where available, and from the literature where marked.Clusters of Orthologous Groups assignments (COGS) were taken from the IMG database, where available, and were generated using the COGnitor tool (available from http://www.ncbi.nlm.nih.gov/COG/) for sequences referred to the literature sources.Locus names are taken from IMG where available.(XLS) Table S2 The percent abundance of SAR11 C1 metabolism genes at each sample site in GOS data, relative to recA.

(XLS)
Table S3 Accession numbers used in Figure 3 and

Figure 3 .
Figure 3. Phylogenetic tree of Fe-ADH proteins.Coloration is according to 16S rRNA gene phylogeny, as shown in the boxed legend.Bootstrap values were omitted for clarity; nodes with less than 60% support were collapsed.Arrows indicate Fe-ADH proteins for which methanol dehydrogenase activity has been demonstrated experimentally.Scale bar = 0.4 changes per position.doi:10.1371/journal.pone.0023973.g003

Figure 4 .
Figure 4. Phylogeny of SAR11 AMT proteins.Four paralogous AMTs in HTCC1062 were placed into three functional subgroups: DmdA-like, GcvT, and an AMT of unknown function.All four AMTs were also identified in HTCC1002 and HTCC7211 genomes.This phylogenetic tree was generated using the neighbor-joining method.Bootstrap values are based on 100 iterations.doi:10.1371/journal.pone.0023973.g004

Figure
Figure S2 Distribution of C1 gene homologs throughout the Alphaproteobacteria.The number of genomes containing homologs of C1 oxidation genes reported by gene and divided by Order.The total number of genomes examined for each order is in parentheses.(TIFF) Figure S3 Culture experiments to determine the concentrations of C1 and methylated compounds for ATP and radioisotope assays.HTCC1062 cells were cultured in seawater medium amended with 10 mM NH 4 Cl, 1 mM KH 2 PO 4 , 10 nM FeCl 3 , vitamins, and C1 and methylated compounds at different concentrations.(PDF)

Table 1 .
Distribution of genes involved in C1 metabolism among three SAR11 Ia genomes. C-

Table 2 .
ATP response of starved cells to addition of various alcohols a .
a.For each assay, cells were grown in media containing the test compound, then washed and starved for 20 hrs.Cellular ATP content was measured after cells were exposed for 2 hrs to the test compound (T), to no compound added (N), and to pyruvate (P; positive control to confirm metabolic activity of cells).b.Asterisk indicates statistical significance (p-value ,0.01) between ''no compound added'' and ''test compound'' treatments.doi:10.1371/journal.pone.0023973.t002

Table 3 .
ATP response of starved cells to addition of C1 and methylated compounds a .