Genome and catabolic subproteomes of the marine, nutritionally versatile, sulfate-reducing bacterium Desulfococcus multivorans DSM 2059

Short chain alcohols

The genome of D. multivorans harbours five genes (dmul_C14490, 16490, 16540, 36750 and adh) encoding predicted alcohol dehydrogenases. Since the product of only one of them (Adh) was found to be specifically increased in abundance during anaerobic growth with n-butanol and 1-propanol (compounds 12 and 16 in Fig. 2), it is likely to be responsible for converting these alcohols to their respective aldehydes. Four aldehyde dehydrogenases (Aor1-3, aldehyde:ferredoxin oxidoreductase; Dmul_14960, NAD-dependent aldehyde dehydrogenase) are encoded in the genome, but only the product of aor3 was detected under all tested growth conditions (Fig. 3), pointing towards its involvement in aldehyde catabolism in D. multivorans.

β-Oxidation of unbranched aliphatic carboxylates and cyclohexane carboxylate

D. multivorans utilizes fatty acids up to a chain length of C₁₄ via classic β-oxidation and cyclohexane carboxylate via the modified β-oxidation route of the central benzoyl-CoA pathway (Fig. 2). Accordingly, the genome contains a wide range of genes associated with β-oxidation-like functions: (i) Seven (3 identified) predicted paralogs of long-chain-fatty-acid-CoA-ligases (FadD) as well as 10 (7 identified) predicted AMP-dependent synthetases and ligases for CoA-activation of carboxylates; (ii) 15 (6 identified) predicted acyl-CoA dehydrogenase (Acd), (iii) 5 (2 identified) predicted enoyl-CoA hydratases (EchA) and (iv) one (identified) predicted 3-hydroxyacyl-CoA dehydrogenase (FadB). Out of these 19 proteins, seven could be assigned to specific pathways as discussed below.

The anaerobic degradation of cyclohexane carboxylate (compound 8 in Fig. 2) was recently elucidated for G. metallireducens GS-15. Initial activation to cyclohexanoyl-CoA is followed by 1,2-dehydrogenation to cyclohex-1-ene-1-carboxyl-CoA and subsequent 1,4-dehydrogenation to cyclohex-1,5-diene-1-carboxyl-CoA [41]. Genes (>80 % protein sequence identity to orthologs of G. metallireducens GS-15) encoding putative cyclohexanoyl-CoA dehydrogenase (dmul_C00970) and cyclohex-1-ene-1-carboxyl-CoA dehydrogenase (dmul_C00980) were detected in the genome of D. multivorans, directly adjacent to an AMP-dependent synthetase (dmul_C00990), a membrane-bound FeS oxidoreductase (dmul_C01000) and an electron transfer flavoprotein (etfA1B1) (Fig. 5d). All corresponding proteins were specifically detected during anaerobic growth with cyclohexane carboxylate supporting their involvement in this pathway also in case of D. multivorans. The formed cyclohex-1,5-diene-1-carboxyl-CoA enters the β-oxidation-like branch of the benzoyl-CoA pathway as supported by specifically increased abundances of respective proteins (Dch, Oah, BzdZ) (Fig. 3).

Myristinate (tetradecanoic acid; compound 10 in Fig. 2) conversion to five acetyl-CoA and one butyryl-CoA is assumed to involve CoA-activation by FadD5, AcdA9-catalyzed dehydrogenation, acetyl-CoA acyltransfer by FadA and oxidation of 3-hydroxyacyl-CoA by FadB, inferred from myristinate-specific protein formation. Respective encoding genes colocalize on a stretch of 10.3 kbp at 3.19 Mbp of the chromosome (Fig. 5f), except for fadAB (at 1.03 Mbp). Notably, this genomic locus also includes genes for a transcriptional regulator FadR2 (specifically detected in myristinate-adapted cells) as well as genes (etfA5B5) for an electron transfer flavoprotein (ETF) and a constitutively formed membrane-bound FeS oxidoreductase (fadF) (Figs. 2 and 3). The latter may be involved in myristinate-specific electron transfer (see below section “Electron transfer flavoproteins”). The TetR-type regulator FadR2 may mediate the observed myristinate-specific protein formation.

Butanoate (compound 11 in Fig. 2) is commonly degraded via crotonyl-CoA to acetoacetyl-CoA which is thiolytically cleaved into two acetyl-CoA. An increased abundance of acetyl-CoA dehydrogenases AcdA1/6 in butanoate- as well as n-butanol- and myristinate-adapted cells indicates an involvement of these proteins in conversion of butanoate and unbranched fatty acids in general (Fig. 3). However, none of the many β-oxidation-related genes could be assigned unequivocally to butanoate catabolism based on genomic co-localization and/or substrate-specific product formation, indicating a rather constitutive formation of these enzymes.

Propanoate (compound 15 in Fig. 2) and odd-numbered fatty acids are commonly degraded also via β-oxidation which however yields one propionyl-CoA in addition to the acetyl-CoA moieties. Further degradation of propionyl-CoA proceeds via the methylmalonyl-CoA pathway. Initially, propionyl-CoA is carboxylated to methylmalonyl-CoA, by ATP-dependent propionyl-CoA carboxylase (PccAB3). Methylmalonyl-CoA is transformed from D- to L-conformation by methylmalonyl-CoA epimerase (methylmalonyl-CoA racemase, Mce) prior to being intramolecularly rearranged to succinyl-CoA by methylmalonyl-CoA mutase (Sbm). Succinyl-CoA is further degraded employing enzymes of the tricarboxylic acid cycle (succinyl-CoA ligase, SucCD2; succinate dehydrogenase, SdhABC; fumarate hydratase, FumAC) followed by an oxidative decarboxylation (malic enzyme, MaeA) yielding pyruvate. Genes for all required enzymes are rather scattered across the genome of D. multivorans contrasting the clustered organization in D. autotrophicum HRM2 [14]. Respective gene products of D. multivorans were detected under all aliphatic substrate conditions (except for PccA and FumC) pointing to a constitutive presence of the methylmalonyl-CoA pathway (Figs. 2 and 3).

Branched-chain fatty acids

Isobutanoate (compound 13 in Fig. 2) is probably degraded according to the pathway suggested by Stieb and Schink [42] including initial CoA-activation followed by two oxidation and one hydrolysis step(s) yielding methylmalonic semialdehyde. The latter could be subsequently oxidized and decarboxylated forming propionyl-CoA (Fig. 2). Initial CoA-activation of branched-chain fatty acids in D. multivorans is probably mediated by AcsA2 (Fig. 3). The unambiguous assignment of genes involved in further conversion was, however, not possible and also no isobutanoate-specific formation of respective proteins was observed.

In case of 2-methylbutanoate (compound 14 in Fig. 2), the methyl-branching is compatible with classical β-oxidation (see short chain alcohols), but thiolytic cleavage of intermediary 2-methylacetoacetyl-CoA yields acetyl-CoA and propionyl-CoA (Fig. 2), with the latter feeding into the methylmalonyl-CoA pathway.

3-Methylbutanoate (isovalerate; compound 9 in Fig. 2) is probably degraded in analogy to the pathway previously proposed for a mixed anaerobic marine co-culture [43]. Initially formed isovaleryl-CoA is oxidized to methylcrotonyl-CoA, which is subsequently carboxylated to methylglutaconyl-CoA, following hydration to 3-hydroxy-3-methylglutaconyl-CoA, and final cleavage into acetoacetyl-CoA and acetyl-CoA. Genome analysis revealed a gene cluster at 3.20 Mbp potentially encoding enzymes for isovalerate degradation as well as a transcriptional regulator (Fig. 5e). Differential proteomics corroborated this prediction by demonstrating exclusive formation of acyl-CoA synthase (AcsA6), enoyl-CoA hydratase (EchA2), methylcrotonyl-CoA carboxylase (Mcc) and 3-hydroxy-3-methylglutaconyl lyase (HmgL) in 3-methylbutanoate-adapted cells of D. multivorans (Fig. 3).

Lactate and pyruvate

Lactate (compound 17 in Fig. 2) is probably taken up by L-lactate permease (LldP). The latter is encoded in an operon-like structure together with genes for pyruvate-forming lactate dehydrogenase (ldhAB), L-lactate utilization proteins (lutABC) predicted to transfer electrons from lactate dehydrogenase to membrane-embedded redox partners in Bacillus subtilis [44], and a GntR-type transcriptional regulator (dmul_C17480) (Fig. 5g). These lactate-related proteins were substrate-specifically formed, supporting their predicted function in D. multivorans.

Pyruvate, which is also generated from the C₄-dicarboxylate malate by malic enzyme (Mae) as the final step of the methylmalonyl-CoA pathway (see section above), is oxidatively decarboxylated to acetyl-CoA by the pyruvate:ferredoxin oxidoreductase (Por). Agreeing with its central metabolic role, Por was found to be constitutively formed.

QmoABC and DsrMKJOP complexes

Among the various described transmembrane complexes in SRB, only the three-subunit, quinone-interacting, membrane-bound oxidoreductase complex QmoABC and the five-subunit DsrMKJOP complex are strictly conserved in currently genome-sequenced SRB [50]. In D. multivorans the qmoABC and aprAB genes colocalize while dsrMKJOP and dsrABC genes do not. In analogy to other genome sequenced SRB, the QmoABC complex is proposed to channel electrons from the membrane-localized menaquinone pool to AprAB, while the DsrMKJOP performs the analogous function towards DsrAB via soluble DsrC [13, 51]. All QmoABC and DsrMKJOP subunits could be identified in rather high abundance under all tested substrate conditions (Fig. 4), corroborating their predicted essential function in the energy metabolism.

HmcABCDEF, TmcABC(D) and QrcABCD complexes

The six-subunit HmcABCDEF [52] complex and the potentially Hmc-derived four-subunit TmcABCD [53] complex often occur in parallel in SRB and are functionally associated with electron transfer between the cytoplasm and the periplasm. While the Tmc complex is assumed to channel electrons from periplasmic hydrogen oxidation to cytoplasmic dissimilatory sulfate reduction, the Hmc complex operates probably in reverse direction to build up reduced conditions in the periplasm [13]. Notably, the genome of D. multivorans lacks genes encoding periplasmic hydrogenases (e.g., hyn, hys, hyd), agreeing with the original finding of the strain not being able to utilize hydrogen as an electron donor [31].

The genome of D. multivorans encodes a complete Hmc complex at 0.90 Mbp, and formation of several subunits was observed in phenylacetate- (HmcABCEF) as well as phenylpyruvate-adapted (HmcABEF) cells. The comparatively selective and apparently lower-level formation of the Hmc complex indicates an only minor (non-essential) role in general electron transfer, while a potentially specific function in phenylacetate and phenylpyruvate catabolism remains to be demonstrated.

Two paralogous gene clusters for the Tmc complex (tmcA1B1C1D at 0.89 Mbp and tmcA2B2C2 at 2.38 Mbp) are present in the genome of D. multivorans sharing 35–64 % (Tmc1) and 27–35 % (Tmc2) amino acid identity with the orthologs from D. vulgaris Hildenborough [53]. Hmc complex encoding genes are located downstream of tmcA1 (Additional file 1: Figure S3) directly adjacent to genes encoding three DNA-binding response regulator proteins (Dmul_C08090/180/210), a DNA-binding universal stress protein (Dmul_C08170) and four signal transduction histidine kinases (Dmul_C08190/200/220/230), resembling the operon-like structure previously described for D. vulgaris Hildenborough [53]. In contrast to the above described Hmc complex, most constituents of the two encoded Tmc complexes were identified in the membrane protein-enriched fraction under all 17 substrate adaptation conditions, displaying, however, rather low abundances (Fig. 4). Notably, the periplasmic TmcA1 subunit and cytoplasmic TmcB2 were not detected. While the overall low Tmc complex abundance may render the small (15 kDa) and periplasmic TmcA1 not detectable, the cytoplasmic TmcB2 (50 kDa) is apparently not formed. However, the absence of genes encoding periplasmic formate dehydrogenases and hydrogenases challenges a function of the D. multivorans Tmc complexes in the respective electron transfer as described for D. vulgaris Hildenborough [53].

The Qrc complex contains three periplasmic (ABC) and one integral membrane (D) subunit(s) and has been associated with connecting periplasmic components, TpIc ₃, and non-membrane-anchored hydrogenase or formate dehydrogenase (for overview see [13, 54, 55]). The genome of D. multivorans harbours a qrcABCD gene cluster at 3.91 Mbp colocalizing with cytochrome biogenesis genes (ccsAB); they share high sequence similarities (33–75 % on amino acid level) with their orthologs from D. vulgaris Hildenborough [54], D. autotrophicum HRM2 [14] and D. alkenivorans AK-01 [16]. All four Qrc subunits of D. multivorans were abundantly detected in the membrane protein-enriched fraction of all 17 substrate adaptation conditions. The inability of D. multivorans to utilize hydrogen and formate, both of which are predicted to serve as periplasmic electron donor for the Qrc complex in D. vulgaris Hildenborough [54], suggests a different, though essential function of the Qrc complex in D. multivorans energy metabolism. The predicted physiological partner may be the periplasmic TpIc ₃ cytochrome (Dmul_C11180).

Nuo complex

Genes for the multi-subunit NADH:quinone oxidoreductase complex (Nuo, respiratory complex I; nuoA-N) were identified at 2.03 Mbp of the chromosome. Among Deltaproteobacteria, complete nuo gene sets seem to be characteristic for Desulfobulbaceae and have so far been reported only for seven representatives of Desulfovibrionaceae (for overview see [13, 50]). Thus, D. multivorans is the first member of Desulfobacteraceae possessing the complete nuo gene set in a single cluster (Additional file 1: Figure S3). However, only the cytoplasmic fusion protein NuoBCD and the integral membrane subunit NuoL could be detected under most of the tested substrate conditions at rather low abundance (Fig. 4). Hence, the Nuo complex in D. multivorans may play only a minor role in energy metabolism with its function still to be resolved.

RnfABCDEG complex and Na⁺-based bioenergetics

The Rnf complex is an electron transporting, membrane-localized ferredoxin:NAD⁺-oxidoreductase widespread among Deltaproteobacteria. The six subunit RnfABCDEG complex consists of three integral membrane proteins (RnfADE), two membrane associated (RnfBG) and one soluble subunit(s) (RnfC). It is assumed to link the cellular NADH and ferredoxin pools via bidirectional electron transfer, i.e., (i) oxidation of reduced ferredoxin by RnfB and transfer via RnfDG to RnfC for NAD⁺ reduction coupled to the generation of a Na⁺-gradient, or (ii), vice versa, from NADH to oxidized ferredoxin driven by the membrane potential [50, 56]. Notably, a different type of the Rnf complex was recently reported to be present in D. toluolica Tol2 [49]. Here, the RnfB subunit involved in ferredoxin oxidation/reduction is larger (~700 amino acids) as compared to characterized RnfB of Clostridium ljungdahlii or Acetobacterium woodii (~300 amino acids) and contains NAD⁺-binding sites in addition to the 4Fe-4S cluster and ferredoxin-type iron-sulfur binding domain present in all RnfBs. Furthermore, also the RnfC subunit of D. toluolica Tol2 differs, lacking the NADH-ubiquinone oxidoreductase and soluble ligand-binding domains necessitating an RnfC electron acceptor/donor other than NAD⁺/NADH, which has to be identified. Due to these differences, a mode of action different to the classical Rnf complex was suggested [49] that awaits experimental verification.

The genome of D. multivorans contains two colocalized rnf gene cluster (rnf1 at 3.09 Mbp and rnf2 at 3.11 Mbp) encoding an A. woodii-type Rnf complex with a small RnfB subunit (Rnf1; 35–55 % amino acid identity) as well as a D. toluolica Tol2-type Rnf complex with a large RnfB subunit (Rnf2; 41–84 % amino acid identity), a genetic repertoir similar to D. autotrophicum HRM2 [14]. Differential proteomics revealed the formation of all six subunits of Rnf1 and five subunits of Rnf2 under all tested substrate conditions in the membrane protein-enriched fraction; the only exception is the integral membrane protein RnfA2 being only detected under 6 substrate conditions (Fig. 4). The latter may be due to its size and detectability of the resulting tryptic peptides. The cytochrome encoded in the rnf2 cluster (Dmul_C27670) was only detected in lactate-adapted cells. However, Dmul_C27670 is a small (15 kDa) acidic (pI 5.3) protein, with only few detectable peptides, rendering its identification difficult and implicating its presence may not be completely excluded under the other tested growth substrates.

In contrast to D. toluolica Tol2 [15], the genome of D. multivorans does not contain genes for a Na⁺-potential-forming NADH:quinone oxidoreductase (NqrA-F).

Hdr/Mvh complex

A protein complex involved in electron bifurcation is the HdrABC/MvhADG complex, which couples the oxidation of H₂ to the reduction of ferredoxin and heterodisulfide (CoM-S-S-CoB) in methanogens [58]. HdrA contains a FAD-binding site and is therefore predicted to be involved in electron bifurcation/confurcation and to represent the site of ferredoxin reduction. In contrast, the Mvh subunits probably catalyze the H₂ oxidation (MvhA) and serve in electron transfer (MvhDG) to the FAD centre of HdrA. HdrB catalyzes the reduction of heterodisulfide and is supplied with electrons by HdrC [59]. The hdrA gene is highly conserved in methanogenic archaea and orthologs are present in the genomes of a broad range of anaerobic organisms including SRB. Indeed, some SRB genomes harbour a rather large number of hdrA orthologs, e.g., D. autotrophicum HRM2 (7 orthologs) [14] and D. toluolica Tol2 (13 orthologs) [15].

The genome of D. multivorans harbours only two hdrA homologs (hdrA1 at 0.32 Mbp and hdrA2 at 3.25 Mb). The hdrA1 gene colocalizes with mvhD1 and pyridine nucleotide-disulphide oxidoreductase (Dmul_C02800), whereas hdrA2 is located adjacent to hdrBC and mvhD2 similar to the archaean gene organization. Analysis of the membrane protein enriched-fraction of D. multivorans revealed rather substrate-specific protein formation: HdrA1 was only detected in cell extracts of cells adapted to anaerobic growth with n-butanol, 1-propanol and butanoate, whereas all HdrA2BC subunits were found in cyclohexane carboxylate-adapted cells and HdrA2 was only sporadically observed with other substrate conditions (Fig. 4). These findings agree with the non-detection of HdlA5BC subunits in D. toluolica Tol2 and preclude those protein complexes from serving a general function in cytoplasmic electron transfer. They are more likely involved in specifically linking individual redox reactions to the reducing-equivalent pool.

NfnAB transhydrogenase

Another protein complex involved in FBEB is the transhydrogenase NfnAB (NADH-dependent reduced ferredoxin:NADP⁺ oxidoreductase). This complex was originally discovered in Clostridium kluyveri and reported to catalyze the reduction of ferredoxin with NADPH in the presence of NAD⁺ which was also reduced to NADH [60, 61]. Comparative genome analysis revealed the nfnAB genes to be present in many SRB [50]. Deletion of the nfnAB genes slightly affected growth of Desulfovibrio alaskensis G20 with fumarate and malate, suggesting NfnAB to oxidize NADPH (derived from fumarate and malate conversion) and to reduce NAD⁺ and ferredoxin potentially coupled to energy conservation via Rnf [56].

In D. multivorans the nfnAB genes are located at 0.69 Mbp of the genome. Both protein subunits were identified in all tested substrate-specific subproteomes, with only NfnA revealing slightly fluctuating abundances (Fig. 4). This constitutive formation indicates a rather central role in the energy metabolism of D. multivorans and other Deltaproteobacteria (e.g., D. alaskensis G20 [56]). In D. multivorans, NfnAB may oxidize NADPH derived from acetyl-CoA oxidation by the Wood-Ljungdahl pathway yielding NADH and reduced ferredoxin (Fig. 2).