The 380 kb pCMU01 Plasmid Encodes Chloromethane Utilization Genes and Redundant Genes for Vitamin B12- and Tetrahydrofolate-Dependent Chloromethane Metabolism in Methylobacterium extorquens CM4: A Proteomic and Bioinformatics Study

Chloromethane (CH3Cl) is the most abundant volatile halocarbon in the atmosphere and contributes to the destruction of stratospheric ozone. The only known pathway for bacterial chloromethane utilization (cmu) was characterized in Methylobacterium extorquens CM4, a methylotrophic bacterium able to utilize compounds without carbon-carbon bonds such as methanol and chloromethane as the sole carbon source for growth. Previous work demonstrated that tetrahydrofolate and vitamin B12 are essential cofactors of cmuA- and cmuB-encoded methyltransferases of chloromethane dehalogenase, and that the pathway for chloromethane utilization is distinct from that for methanol. This work reports genomic and proteomic data demonstrating that cognate cmu genes are located on the 380 kb pCMU01 plasmid, which drives the previously defined pathway for tetrahydrofolate-mediated chloromethane dehalogenation. Comparison of complete genome sequences of strain CM4 and that of four other M. extorquens strains unable to grow with chloromethane showed that plasmid pCMU01 harbors unique genes without homologs in the compared genomes (bluB2, btuB, cobA, cbiD), as well as 13 duplicated genes with homologs of chromosome-borne genes involved in vitamin B12-associated biosynthesis and transport, or in tetrahydrofolate-dependent metabolism (folC2). In addition, the presence of both chromosomal and plasmid-borne genes for corrinoid salvaging pathways may ensure corrinoid coenzyme supply in challenging environments. Proteomes of M. extorquens CM4 grown with one-carbon substrates chloromethane and methanol were compared. Of the 49 proteins with differential abundance identified, only five (CmuA, CmuB, PurU, CobH2 and a PaaE-like uncharacterized putative oxidoreductase) are encoded by the pCMU01 plasmid. The mainly chromosome-encoded response to chloromethane involves gene clusters associated with oxidative stress, production of reducing equivalents (PntAA, Nuo complex), conversion of tetrahydrofolate-bound one-carbon units, and central metabolism. The mosaic organization of plasmid pCMU01 and the clustering of genes coding for dehalogenase enzymes and for biosynthesis of associated cofactors suggests a history of gene acquisition related to chloromethane utilization.


Introduction
Chloromethane (CH 3 Cl) is a volatile organic compound emitted by oceans, plants, wood-rotting fungi and biomass burning, estimated to account for 17% of chlorine-catalyzed ozone degradation in the stratosphere [1]. Chloromethane-utilizing bacteria have been isolated from a wide variety of environments such as seawater, soil, sludge, and recently from plant leaf surfaces [2], and represent a potential biotic filter for chloromethane emissions. Many chloromethane degraders are facultative methylotrophic Proteobacteria [3] growing in aerobiosis with chloromethane and other C 1 carbons such as methanol as unique source of carbon and energy. Complete and assembled genomes of two chloromethane-utilizing strains, Methylobacterium extorquens strain CM4 and Hyphomicrobium sp. strain MC1, are available [4,5]. The only known microbial aerobic utilization pathway for chloromethane is tetrahydrofolate (H 4 F)-dependent [6]. This pathway was identified in the alpha-Proteobacterium M. extorquens CM4 using minitransposon random mutagenesis [7] and its chloromethane dehalogenase activity characterized in detail [8,9]. The first step of the cmu (chloromethane utilization) pathway is catalyzed by the two-domain methyltransferase/corrinoid-binding CmuA protein that transfers the methyl group from chloromethane to a corrinoid cofactor [9,10]. The methylcobalamin:H 4 F methyltransferase CmuB enzyme subsequently catalyzes the transfer of the methyl group from the corrinoid cofactor to H 4 F [8]. The H 4 F-bound C 1  4 MPT-dependent enzyme-mediated steps are depicted in blue and pink, respectively. Carbon assimilation operates via the serine cycle (Ser) coupled with the ethylmalonyl-CoA pathway (EMCP) [67]. Spontaneous condensation of HCHO with H 4 F or H 4 MPT, and formaldehyde oxidation to methylene-H 4 F are shown with broken line. In the cmu pathway, the methyl group enters a specific H 4 F-oxidation pathway for energy production driven by the FolD and PurU enzymes. Protein-encoded genes or genes located on plasmid pCMU01 are shown in bold. Boxes and circles highlight proteins more abundant in chloromethane-and methanol grown-cultures, respectively. CmuA, methyltransferase/corrinoid-binding two-domain protein; CmuB, methylcobalamin:H 4 F methyltransferase; Fae, formaldehyde activating enzyme; Fch, methenyl-H 4 F cyclohydrolase; FDHs, formate dehydrogenases; Fhc, formyltransferase-hydrolase complex; FolD, bifunctional methylene-H 4 F dehydrogenase/cyclohydrolase; FtfL, formate-H 4 F ligase; Gck, glycerate kinase; GcvT, H 4 F-dependent aminomethyltransferase; HprA, hydroxypyruvate reductase; MDH, methanol dehydrogenase; MetF, methylene-H 4 F reductase; MtdA, bifunctional NAD(P)-dependant methylene-H 4 F and methylene-H 4 MPT dehydrogenase; MtdB, NAD(P)-dependent methylene-H 4 MPT dehydrogenase; Mch, methenyl-H 4 MPT cyclohydrolase; MtkA, malate thiokinase large subunit; MxaF, MDH alpha subunit, PurU, 10-formyl-H 4 F hydrolase; Sga, serine-glyoxylate aminotransferase [12]. Plasmid pCMU01 encoded proteins with predicted functions include putative uncharacterized methyltransferases CmuC and CmuC2, the putative PaaE-like oxidoreductase, and the putative PQQ-linked dehydrogenase of unknown specificity XoxF2. GvcT may serve to transfer methyl groups from a wide range of substrates to H 4 F, as proposed for members that belong to the COG0354-related enzymes such as YgfZ [68]. doi:10.1371/journal.pone.0056598.g001 Known pathways for tetrahydromethanopterin (H 4 MPT)-and H 4 F-dependent C 1 substrate oxidation in Methylobacterium strains are compared in Figure 1.When growing on methanol, M. extorquens CM4 uses the H 4 MPT formaldehyde oxidation pathway first discovered in M. extorquens AM1 [12] and subsequently found to be widespread among methylotrophs.
Growth with chloromethane depends on the presence of cobalt in the medium [9] since CmuA methyltransferase activity requires a vitamin B 12 -related corrinoid cofactor that incorporates cobalt. As described for adenosylcobalamin (AdoCbl), the corrinoid cofactor may be synthesized de novo by one of Nature's most complex metabolic pathways requiring around 30 enzymemediated steps [13,14]. Of those, only cobUQD genes found adjacent to cmu genes have been described in M. extorquens CM4 [10]. Many microorganisms synthesize vitamin B 12 -related compounds from imported corrinoid intermediates [14] or from precursors such as dimethylbenzimidazole (DMB) [15] by pathways that have not been identified in chloromethanedegrading bacteria.
In this work, combined experimental and bioinformatics analysis was performed to gain a better understanding of the genes and proteins specifically associated with chloromethane utilization in M. extorquens CM4. A differential proteomic approach compared M. extorquens CM4 proteins under methylotrophic growth conditions with either chloromethane or methanol as the sole carbon and energy source. Gene clusters specific to the chloromethane response were identified, and compared to previously published clusters involved in the response of M. extorquens DM4 to dichloromethane [16], or involved in the methylotrophic growth of M. extorquens AM1 to methanol [17]. We found that growth with chloromethane elicits a specific adaptive response in M. extorquens CM4. In addition, the genome sequence of the chloromethane-degrading strain CM4 was compared to available complete sequences of other M. extorquens strains unable to grow on chloromethane (strains AM1, PA1, BJ001 and DM4; [5,11]). Genomic analysis revealed that additional gene homologs of chromosome-encoded cognate genes for coenzyme biosynthesis, as well as specific genes such as bluB2, which is predicted to be involved in both H 4 F and vitamin B 12 cofactor biosynthesis, were found nearby previously characterized genes cmuA and cmuB on a 380 kb plasmid.

Biological Materials, Media and Growth Conditions
The composition of Methylobacterium mineral medium M3 was adapted from that given in Vannelli et al. [7] with 0.2 g.L 21 (NH 4 ) 2 SO 4 final concentration and substitution of ZnCl 2 by ZnSO 4 in the trace element solution. M. extorquens CM4 was grown aerobically at 30uC either with chloromethane or with methanol as carbon substrate, on a rotary shaker (140 rpm) in 1.2liter Erlenmeyer flasks containing 200 mL M3 medium, closed with gas-tight screw caps with mininert valves (Supelco). Methanol (sterile-filtered) was added to a final concentration of 40 mM. Chloromethane gas was added to a final concentration of 15 mM in the liquid phase, assuming a Henry constant of 0.0106 m 3 .atm.mol 21 at 30uC [19], as previously described [6]. Acetone was added to a final concentration of 5 mM from a sterile-filtered solution at 250 mM. Chloromethane and acetone degradation were quantified using a CP 3800 gas chromatograph connected to a flame ionization detector (GC-FID; Varian, USA) equipped with a GC column (CP-Sil 5 CB, length 15 m; Varian).

Protein Extraction
Triplicates of M. extorquens CM4 cultures were harvested by centrifugation (10 min at 10,000 g) in mid-exponential growth phase of chloromethane-and methanol-grown cultures, using 100 mL at OD 600 of 0.2 and 33 mL at OD 600 at 0.6, respectively. Cell pellets were resuspended in 10 mM Tris, 1 mM EDTA buffer pH 7.6 (TE buffer), washed once and resuspended in 400 mL of the same buffer in the presence of benzonase (250 units; GE Healthcare) and 4 mL protease inhibitor mix 1006 (GE Healthcare). Cells were disrupted using glass beads (0.1 mm in diameter, 1 g per 0.4 mL extract) in a MM2 mixer mill (Retsch Haan, Germany) at maximal speed for 6 cycles of 30 sec, and then placed on ice for one hour. Cell debris and beads were removed by centrifugation at 14,000 g for 15 min at 4uC, and the supernatant was centrifuged again at 14,000 g for one hour at 4uC. Protein concentration in the supernatant was assayed using a commercial Bradford assay (Biorad) with bovine serum albumin as a standard, and subsequently adjusted to 1 mg/mL in TE buffer.

Two-dimensional Fluorescence Differential Gel Electrophoresis (2D-DIGE)
Samples were labeled with CyDye DIGE Fluor minimal dyes (Cy2, Cy3 or Cy5, GE Healthcare) according to the manufacturer's instructions. After acetone precipitation, the resulting protein pellet was resuspended in DIGE rehydration buffer (DRB) (7 M urea, 2 M thiourea, 4% wt/vol CHAPS, Tris 20 mM, pH 8.5). The protein concentration was quantified using a slightly modified Bradford method (as described above except for use of DRB), standardized with known concentrations of bovine serum albumin, and adjusted to 2 mg/mL using DRB. For each DIGE experiment, 8 samples were labeled, corresponding to four independent cultures for each condition. Labeling was performed by mixing 50 mg total protein from each of the four samples with either Cy3 (two samples) or Cy5 (two other samples) DIGE minimal dye (400 pmol) (GE Healthcare), to take into account biases resulting from different labeling efficiency. A pooled set of internal standards, comprising 25 mg aliquots from each of the 8 samples (200 mg total), was labeled with Cy2 DIGE minimal dye (1600 pmol total). Labeling was performed for 30 minutes on ice and in the dark, and quenched by the addition of 10 mM lysine (1 mL for Cy3-or Cy5-labeled extracts, and 4 mL for Cy2-labeled extracts, respectively). Samples were incubated for 10 min on ice in the dark. Finally, protein samples that were separated on the same gel were mixed (one Cy3-and one Cy5-labeled sample each together with one-fourth volume of the pooled set of internal standards), and supplemented with 0.6% IPG pH 3-10 NL (nonlinear) and 65 mM DTT. Buffer RB was added to each mix to final volume of 350 mL before IEF separation. Proteins were loaded on 18-cm IPG strips (non-linear gradient pH 3-10) and submitted to separation steps as described above. Gels were fixed and proteins stained as described above. Scanning of gels was performed with an Ettan DIGE Imager (GE Healthcare).

Proteome Image Analysis
Differential analysis was performed using ImageMaster 2D Platinum software (v. 6.0, GE Healthcare). Six gels were grouped in two classes of three independent gels depending on the two compared conditions (chloromethane or methanol growth conditions). Gels were matched with one reference gel (master gel) following spot detection. For each spot, the relative volume corresponded to the normalized volume of the spot compared to the normalized volume of the entire gel coloration. Statistical analysis was performed by calculating the Student t value for each spot, as well as a ratio value defined as the mean of the relative volume of the spot obtained in the different replicates for growth with chloromethane divided by the mean of the relative volumes obtained for growth with methanol. Spots with a Student t value higher than 1.9 (corresponding to a p-value of ,0.1) and ratios $2.0 or #22.0 were analyzed by mass spectrometry. DIGE images were analyzed with DeCyder software (v. 7.0, GE Healthcare). A total of twelve images obtained from 4 gels (three images each) were analyzed. Student's t test was used to determine differential abundance of proteins. In this procedure, the p-values were corrected for false discovery rate [20]. Spots with a p-value ,0.01 and ratios $2.0 or #22.0 were considered to be differentially abundant.

Mass Spectrometry Protein Identification
The procedure of Muller et al. [16] was followed for spot identification, with minor adjustments. Mass spectrometry analyses were performed in reflector positive mode on a Biflex III (Bruker-Daltonik GmbH, Bremen, Germany) matrix-assisted laser desorption/ionization time-of-flight mass spectrometer (MALDI-TOF TOF) and on an Autoflex III Smartbeam (Bruker-Daltonik GmbH, Bremen, Germany) matrix-assisted laser desorption/ ionization time-of-flight mass spectrometer (MALDI-TOF TOF). A saturated solution of a-cyano-4-hydroxycinnamic acid in 50% water/50% acetonitrile was used as matrix for MALDI mass measurement on the Biflex III. Peptide mass fingerprinting data (PMF) and peptide fragment fingerprinting data (PFF) were combined by Biotools 3 software (Bruker Daltonik) and transferred to the search engine MASCOT (Matrix Science, London, UK). Peptide mass error was limited to 100 ppm for the Biflex III and to 50 ppm for the Autoflex III Smartbeam. Proteins were identified by searching against the NCBI non-redundant protein sequence database and the predicted proteins of strain CM4 (GenBank accession no. CP001298, CP001299, CP001300).

The 380 kb Episome in M. extorquens CM4 Harbors cmu Genes and Associated Genes
The repertoire of known cmu genes and genes conserved in chloromethane-degrading strains includes genes essential for dehalogenation of chloromethane (cmuA, cmuB), genes essential for growth with chloromethane as the sole carbon and energy source (cmuC, metF, purU), and genes found in the vicinity of genes cmuA, cmuB and cmuC in methylotrophic chloromethane-degrading strains (fmdB, paaE, hutI, and cbiD) [2]. All these genes co-localize on a 138 kb region flanked by transposable elements nested within the 380 kb plasmid pCMU01 in M. extorquens CM4. This plasmid encodes proteins associated with growth on chloromethane, such as the enzymes for chloromethane degradation and for metabolism of the two essential dehalogenase cofactors AdoCbl and H 4 F, as well as transport proteins for coenzyme B 12 precursors (Table 1). Thus, plasmid pCMU01 can be designated as a chloromethane catabolic plasmid that harbors the cognate essential genes for growth on chloromethane.

Overview of Plasmid pCMU01 Gene Content
Plasmid pCMU01 is characterized by a somewhat lower GC content (66.3%) than the chromosome of strain CM4 (68.2%). To a large extent (41%), it features unique genes encoding predicted proteins without close homologs (.40% aa Id, .80% of the protein length) in the genomes of four other M. extorquens strains unable to degrade chloromethane [5] (Table 1). Of its 386 predicted CDS, 56% belong to at least one COG group [21] related to metabolism (enzymes, 69 CDS; transporters, 20 CDS), plasmid functions (23 CDS), adaptive response (regulation, 34 CDS; stress, two CDS), and genomic plasticity (mobile DNA elements, 71 CDS with a total of 18 identified IS elements representing 4% of the predicted CDS of the plasmid). Pseudogenes may account for 9% of the total predicted CDS on the plasmid with 35 pseudogenes detected.
Chloromethane is not the only organic molecule for which the plasmid allows to transform for growth. A complete acetonecatabolic gene cluster encoding the acetone carboxylase subunits (b subunit, acxA; a subunit, acxB; c subunit, acxC) and its cognate transcriptional activator (gene acxR) was found. Acetone carboxylase is the key enzyme of bacterial acetone metabolism in Xanthobacter autotrophicus strain Py2, catalyzing the ATP-dependent pCMU01 Plasmid-Driven Chloromethane Metabolism PLOS ONE | www.plosone.org carboxylation of acetone to form acetoacetate [22]. High sequence conservation was found between the acxRABC gene clusters of strains CM4 and X. autotrophicus Py2 (.82% aa Id for enzyme subunits and 53% for the regulator). The ability of M. extorquens CM4 to degrade acetone was tested in aerobic liquid cultures in M3 medium. When 5 mM acetone was supplied as the unique source of carbon and energy, strain CM4 grew up to an OD 600 of 0.4 at stationary growth phase, with total degradation of acetone as measured using GC-FID. No acetone degradation was observed in the abiotic control or in cultures of M. extorquens DM4 lacking the acx cluster under the same conditions. Thus, the plasmid pCMU01-encoded acx cluster seems to be functional in strain CM4.
Evidence of the mode of replication, maintenance, and conjugation of plasmid pCMU01 was suggested from sequence similarity searches. A combined replication and partitioning repABC unit (Mchl_5615-5617) including the incompatibility antisense RNA (ctRNA) between the repB-repC genes was found [23]. The gene products of repA, repB and repC share at least 43% aa Id with the corresponding proteins of the characterized Rhizobium etli p42d plasmid repABC cassette [24]. In a recent review, plasmid pCMU01 was classified within the RepABC family plasmids of large low-copy-number plasmids found exclusively in Alphaproteobacteria [25]. The plasmid harbors components of a core type IVB secretion/conjugation system complex often used for horizontal propagation, including the gene encoding the conserved central component of the DNA transport activity core complex (Mchl_5595), and traD (Mchl_5572) which lies within a traGDCA gene cluster (Mchl_5572-5575) conserved in other Alphaproteobacteria plasmids including Agrobacterium tumefaciens Ti plasmids (encoded proteins TraG, D, C and A sharing 43, 53, 38 and 44% aa Id, respectively). Finally, a putative restrictionmodification system encoding protein Mchl_5634 shares 48% aa Id with a bifunctional DNA methyltransferase/type II restriction endonuclease MmeI [26] (Table 1). Taken together, the described genomic features indicate that plasmid pCMU01 represents a lowcopy plasmid, vertically transmitted via a RepABC replication and partitioning unit, and most probably able to propagate by horizontal transfer within Alphaproteobacteria.
Extensive Plasmid-encoded Gene Redundancy Associated with Vitamin B 12 Metabolism M. extorquens CM4 is able to synthesize coenzyme B 12 , a cofactor essential for activity of chloromethane dehalogenase CmuAB [9]. A complete set of cob genes homologous to those described in P. denitrificans for the aerobic biosynthesis pathway of AdoCbl [14] is found on the chromosome of CM4 as well as on the chromosomes of four other M. extorquens strains (Table 2). Remarkably, strain  Probability-based mowse score calculated using MASCOT software (Matrix Science, London, UK); error refers to mass accuracy; coverage refers to the percentage of the protein sequence covered by the matched peptides. Only found in strain CM4 (among the 8 Methylobacterium strains for which the complete genome sequence is known; [5,11]) and localized on plasmid pCMU01. CM4 also contains plasmid-borne copies of 13 cob genes and genes coding for cobalt and preformed corrinoid transporters beyond to the close chromosomal homologs of these genes shared by M. extorquens strains. These include the putative cobalt transporter CzcA-related RND transporter [27] (the plasmid-borne gene product Mchl_5715 displays 43% aa Id with the chromosomeencoded Mchl_1072; Table 2), and a homolog of the preformed corrinoid specific transporter Btu [28]. Unlike the plasmid-borne btu gene cluster, the chromosome-encoded btuFCD cluster lacks the btuB gene preceded by a cobalamin riboswitch [29] (Mchl_misc_RNA_1, Table 2), suggesting that expression of the plasmid-borne btu gene cluster is controlled by cobalamin in its coenzyme form (AdoCbl).

Experimental Identification of Gene Clusters Specific of the Chloromethane Response
Differential analyses of proteins extracted from chloromethaneand methanol-grown cultures of M. extorquens CM4 were performed using 2D-E and 2D-DIGE. Overall, 88 protein spots showing differences in abundance between the two compared conditions were detected, resulting in the identification of 49 proteins (Table 3; Fig. S1). In total, 33 proteins were specific of chloromethane-grown cultures, whereas sixteen proteins were more abundant in methanol-grown cultures.
Many of the identified proteins with differential abundance have known or suspected roles in chloromethane utilization and methylotrophy ( Table 3). Many of these proteins allowed to define chloromethane-specific clusters encoding proteins more abundant during growth with chloromethane ( Fig. 2, Clusters A-F), clusters responding both to chloromethane and methanol (Clusters G-H), and or to methanol only (Clusters I-J). The two-domain methyltransferase/corrinoid binding protein CmuA, the methylcobalamin:H 4 F methyltransferase CmuB, and the formyl-H 4 F hydrolase PurU shown to be essential for chloromethane metabolism in strain CM4, were identified in the chloromethane proteome only (Fig. S1) as expected [6,9,10]. Experimental evidence for chloromethane-enhanced expression of a protein involved in cobalamin biosynthesis (precorrin-8X methylmutase CobH2), and of a putative oxidoreductase with FAD/NAD(P)binding domain encoded by a paaE-like gene often associated with cmu genes [2], was obtained here for the first time. Overall, only cluster A encoding proteins more abundant during growth with chloromethane (CmuA; CmuB; CobH2; PaaE-like; PurU; Hss2; Table 3) was localized on plasmid pCMU01.

Proteomic Identification of Stress-related Proteins
Upon dehalogenation, each mole of chloromethane yields one mole of hydrochloric acid with concomitant decrease in pH and increase in chloride concentration [7]. Chloromethane-associated Figure 2. Gene clusters associated with the chloromethane response. Sequence positions are indicated for each gene cluster. All but cluster A are located on the chromosome. Some DNA segments are omitted for clarity (double slashes), with their size indicated in kb. Gene arrows are drawn according to functional category: transport (dots); regulation, sensing or signaling (stripes); unknown (white). Protein products more abundant in cultures grown with chloromethane (C labeled circles) or with methanol (M labeled circles) are indicated, with black or white symbols used for those proteins observed exclusively or more abundant in one condition, respectively. Proteins homologous to induced genes, or proteins more abundant in a previous study of M. extorquens DM4 grown with dichloromethane compared to methanol [16], are indicated with circles labeled by a ''D''. doi:10.1371/journal.pone.0056598.g002 pCMU01 Plasmid-Driven Chloromethane Metabolism Figure 3. Gene redundancy in the biosynthesis of cofactors required for chloromethane utilization in Methylobacterium extorquens CM4. Cbi, cobinamide; Cbl, cobalamin; Ado, adenosyl; DMB, dimethylbenzimidazole; NaMN, nicotinate mononucleotide. AdoCbl and tetrahydrofolate are essential cofactors of the cmu pathway [6,9]. Transport and enzymatic reactions are shown with dotted and full arrows, respectively. Genes indicated in bold are located on the 380 kb plasmid pCMU01. Circled gene names encode proteins more abundant in chloromethane cultures. AdoCbl can be synthesized de novo by an aerobic biosynthesis pathway that incorporates cobalt (diamond), or obtained from a salvage pathway after internalization of preformed Cbi or Cbl. In prokaryotes, the cobalt needed for corrin ring synthesis may be incorporated into cells using the CorA transport system [69], the putative transmembrane proteins CbtA and CbtB [14], the Resistance-Nodulation-Division (RND)type Co 2+ /Zn 2+ /Cd 2+ efflux system CzcA [27], or the Icu transporter [70]. The TonB-dependent Btu system imports preformed corrinoid compounds [28]. We hypothesize that BluB-related proteins link AdoCbl and H 4 F de novo synthesis. doi:10.1371/journal.pone.0056598.g003 pCMU01 Plasmid-Driven Chloromethane Metabolism proteins with homologs characterized in E. coli for their role in osmoprotection were identified (Proteomic data, Table 3). Among these, protein MdoG may be associated with metabolism of osmoregulated periplasmic glucans [30], the putative dTDP-4dehydrorhamnose 3,5-epimerase RfbC may be involved in the synthesis of surface polysaccharides [31], and the putative nucleotidase SurE may be associated with survival at high NaCl concentrations, as observed in E. coli [32], where the corresponding gene lies within a survival operon conserved in Gram-negative bacteria [33].
Production of reactive oxygen species also seems associated with chloromethane utilization. One representative of each class of catalases known to catalyze disproportionation of hydrogen peroxide (H 2 O 2 ) [34] was more abundant in the chloromethane proteome. Mchl_3002 is a putative non-haem manganesecontaining catalase. Mchl_3534 is a KatA-like protein, whose gene is found next to a putative H 2 O 2 activator gene sharing 43% aa Id with E. coli OxyR. In E. coli, OxyR induces the Suf system (sulfur mobilization [Fe-S] cluster) to combat inactivation of the [Fe-S] Isc assembly system by H 2 O 2 [35]. Moreover, E. coli mutants lacking the Suf machinery are hypersensitive to cobalt at high concentrations of 200 mM [36]. In this study, SufS, a selenocysteine lyase homolog, was found more abundant in the chloromethane proteome. Similarly, the CysK cysteine synthase homolog more abundant in chloromethane cultures suggests the probable importance of reactivation systems to maintain chloromethane dehalogenase activity under aerobic conditions, as cysteine is involved in maintaining the catalytic activity and structure of many proteins with [Fe-S] clusters including ferredoxins [37].
Taken together, these data suggest that growth with chloromethane may elicit stress responses, and in particular an oxidative stress response.

Discussion
This work reports genomic and proteomic data demonstrating that cmuA and cmuB genes are plasmid-borne, and that plasmid pCMU01drives the previously defined pathway for H 4 F-mediated chloromethane dehalogenation [6]. Specifically, plasmid pCMU01 harbors cognate genes involved in chloromethaneassociated H 4 F metabolism not found in other M. extorquens genomes (folC2, folD, metF2 and purU; Table 1) [6,10].
H 4 F metabolism is likely to be strongly modulated during growth on chloromethane since proteins linked to H 4 F such as CmuA, CmuB, MetF and PurU were exclusively detected during growth with chloromethane (Table 3), whereas proteins associated to methanol oxidation with the metabolic intermediate formaldehyde and the C 1 carrier H 4 MPT were more abundant during growth with methanol (proteins Fae, Fch and MtdA; Fig. 1).
Here, the interplay of chloromethane and other methylotrophic pathways was evidenced for the first time. Two components of the glycine cleavage complex involved in the conversion of H 4 F and glycine to 5,10-methylene-H 4 F [38], the key C 1 intermediate for entry in the serine cycle, were either more abundant with chloromethane or with methanol (GcvT and Lpd, respectively; Table 3). This suggests that enzymes implied in central metabolism such as the glycine cleavage complex might be involved in integrating contradictory signals during growth with C 1 compounds, to fine-tune metabolic conditions required for growth, and to even out variations in available carbon sources.
Our proteomic study also revealed that essential serine cycle enzymes (Sga, HprA and MtkA) were more abundant in methanol-grown cultures (Table 3). These enzymes are encoded by a chromosomal region (Fig. 2, cluster J), highly conserved in Methylobacterium [11]. Acetyl-CoA, glyoxylate and NADP + have been demonstrated to decrease binding of QscR, a key regulator of C 1 metabolism [39] to the sga promoter, thereby inhibiting transcription of the major operon of the serine cycle (sga-hpr-mtdAfch, [40]). The higher level of acetyl-CoA synthetase in chloromethane-grown cultures (Table 3) may explain the observed lower abundance of five enzymes encoded by cluster J.
PaaE-like Oxidoreductase, PntAA, MetF and Acs are Proteins with Predicted Functions for Growth with Chloromethane Proteomic data provided first experimental evidence for the involvement of four previously undetected proteins, identified here as more abundant during growth with chloromethane, in chloromethane utilization.
The PaaE-like protein encoded by plasmid pCMU01 features a ferredoxin reductase-type FAD binding domain and a 2Fe-2S ferredoxin-type iron-sulfur binding domain. The PaaE-like protein is the only iron-sulfur enzyme more abundant in the chloromethane proteome. This PaaE-like oxidoreductase was suggested to be responsible for the observed methanethiol oxidase activity in the chloromethane-degrading strain, Aminobacter lissarensis CC495 [41,42]. It is also conceivable that the PaaE-like protein acts in the reactivation of the corrinoid cofactor from the inactive Co(II) to the Co(I) form (Fig. 1), as corrinoid-dependent methyltransferases are prone to inactivation by oxidation, and bacteria often require an efficient reactivation system to maintain such proteins in an active form [43,44]. The implication of PaaE in chloromethane utilization may be linked to the detection of genes associated with the oxidation stress response in chloromethanegrown M. extorquens CM4.
The transhydrogenase protein PntAA (Fig. 2 cluster D) couples the transfer of reducing equivalents between NAD(H) and NADP(H) to the translocation of protons across the membrane [45]. Previous transcriptomic and proteomic studies showed that PntAA was up-regulated in succinate-vs methanol-grown cultures of M. extorquens AM1 [46,47], indicating possible differences in energy and reducing equivalent production occurring in M. extorquens strains grown on different carbon sources. Here, the higher abundance of the PntAA complex may be the consequence of higher requirements for reducing equivalents coupled to proton extrusion during chloromethane assimilation.
The 5,10-methylene-H 4 F reductase MetF identified in the chloromethane-proteome is the chromosome-encoded protein and not the plasmid-borne protein MetF2 previously shown to be essential for chloromethane utilization [6]. The protein product of metF2 with a calculated pI of 9.5 is at the limit of the pH range studied in our experiments, which may explain why MetF2 was not detected here. In E. coli, MetF provides one-carbon precursors for methionine synthesis [48] and operates in the opposite direction of the chloromethane degradation pathway. If both MetF and MetF2 homologs share the same metabolites as substrates and products, regulatory processes in the expression of the corresponding genes arising from differences in availability of metabolites may explain the observed increased abundance of MetF with chloromethane. Further experiments are required to clarify the implications of MetF homolgs in chloromethane metabolism.
Identification of the chromosome-encoded acetyl-CoA synthetase Acs as more abundant in the chloromethane proteome was initially surprising. This protein is predicted to catalyze ATPdependent conversion of acetate to acetyl-CoA (78% aa Id with the characterized Bradyrhizobium japonicum Acs enzyme, [49]).
(Mchl_5589, 50 residues share 70% aa Id with the N-terminal part of the replication protein RepA, Mchl_5615, 408 residues) suggests that plasmid pCMU01 was assembled by acquisition of parts of different episomes.

Concluding Comments
Our proteomic analysis showed that the adaptive response of M. extorquens CM4 to chloromethane mostly involves functions which are common to M. extorquens strains, as observed previously for adaptation of M. extorquens DM4 to dichloromethane [16]. Indeed, out of five identified gene clusters responding specifically to chloromethane (Fig. 2), only the catabolic gene cluster essential for growth with chloromethane is encoded by plasmid pCMU01. When these five gene clusters responding specifically to chloromethane in M. extorquens CM4 were compared to the seven gene clusters responding specifically to dichloromethane in M. extorquens DM4, only one chromosomal gene cluster common to chloromethane-degrading strain CM4 and dichloromethane-degrading strain DM4 was identified (cluster B in Fig. 2; cluster C in [16], suggested to be involved in cell structure). Based on these findings, the adaptive response to growth with chloromethane is clearly quite different from that for growth with dichloromethane, in line with the completely different pathways for metabolism of these two halogenated C 1 compounds.
On a final note, it is striking that the organization of the cmu genes on plasmid pCMU01 of M. extorquens CM4 is different from that found in bacteria utilizing the cmu pathway known to date [2]. M. extorquens CM4 constitutes the first representative strain of the M. extorquens species for which a plasmid-encoded carbon utilization function has been clearly established, and its plasmid pCMU01features the only known catabolic gene cluster for the metabolism of halogenated methanes that is not chromosomeborne [4,11,16,66]. The state of plasmid pCMU01 as an autonomous replicating episome may have enabled the efficient acquisition of relevant resources for growth with chloromethane, and the shaping of unique genetic features not observed in other genomes of chloromethane-degrading bacteria. Figure S1 2D-DIGE master gel image of total protein extracts from chloromethane-and methanol-grown M. extorquens CM4 labeled with Cy2 (internal standard). Highlighted spots (circles) displayed differential abundance between chloromethane and methanol conditions, and were identified by mass spectrometry. 1, CmuA; 2, CmuB; 3, PurU; 4, PaaE-like oxidoreductase; 5, Fch; 6, Sga; 7, MtdA; 8, putative UspA-like protein; 9, KatA; 10, MetK; 11, Hss; 12, Acs; 13, PntAA; 14, putative endoribonuclease (Mchl_4437) (See Table 3 and Fig. 1 legend). (TIF)