The modular respiratory complexes involved in hydrogen and sulfur metabolism by heterotrophic hyperthermophilic archaea and their evolutionary implications


Correspondence: Michael W.W. Adams, Department of Biochemistry and Molecular Biology, Life Sciences Bldg., University of Georgia, Athens, GA 30602-7229, USA. Tel.: 706 542 2060; fax: 706 542 0229; e-mail:


Hydrogen production is a vital metabolic process for many anaerobic organisms, and the enzyme responsible, hydrogenase, has been studied since the 1930s. A novel subfamily with unique properties was recently recognized, represented by the 14-subunit membrane-bound [NiFe] hydrogenase from the archaeon Pyrococcus furiosus. This so-called energy-converting hydrogenase links the thermodynamically favorable oxidation of ferredoxin with the formation of hydrogen and conserves energy in the form of an ion gradient. It is therefore a simple respiratory system within a single complex. This hydrogenase shows a modular composition represented by a Na+/H+ antiporter domain (Mrp) and a [NiFe] hydrogenase domain (Mbh). An analysis of the large number of microbial genome sequences available shows that homologs of Mbh and Mrp tend to be clustered within the genomes of a limited number of archaeal and bacterial species. In several instances, additional genes are associated with the Mbh and Mrp gene clusters that encode proteins that catalyze the oxidation of formate, CO or NAD(P)H. The Mbh complex also shows extensive homology to a number of subunits within the NADH quinone oxidoreductase or complex I family. The respiratory-type membrane-bound hydrogenase complex appears to be closely related to the common ancestor of complex I and [NiFe] hydrogenases in general.

Hydrogenase diversity

Hydrogenases are the key enzymes of microbial hydrogen metabolism. They catalyze the reversible interconversion of protons and electrons with H2 and can be found in all three domains of life. The importance of H2 metabolism for anaerobic organisms is illustrated by the fact that the majority of microbial genomes encode one or more hydrogenases. Hydrogenases can be divided into three major groups based on the metal composition of their catalytic center. Two of the classes typically contain one or more iron–sulfur clusters and are differentiated by their active site metalloclusters which contain either nickel and iron ([NiFe] hydrogenases; (Volbeda et al., 1995)) or only iron ([FeFe] hydrogenases; (Nicolet et al., 2000)). The third group is comprised of [Fe]-only hydrogenases (5,10-methenyl tetrahydromethanopterin hydrogenase, Hmd) and is devoid of iron–sulfur clusters but contain an iron–guanylylpyridinol cofactor in the active site (Shima & Thauer, 2007; Hiromoto et al., 2009). They have only been identified in hydrogenotrophic methanogens (McGlynn et al., 2010; Thauer et al., 2010). The [NiFe] and [FeFe] hydrogenases contain the dinuclear ligands CO and CN coordinated to an active site iron atom while the Hmd hydrogenases only contain CO ligands (van der Spek et al., 1996; Happe et al., 1997; Lyon et al., 2004). In spite of this distinguishing characteristic of diatomic ligands (CN and/or CO) attached to an iron site, these three enzyme classes are evolutionarily unrelated, suggesting that their properties arose through convergent evolution (Posewitz et al., 2008). All three groups of hydrogenase require a suite of accessory proteins for maturation of their complex active sites. Intriguingly, maturation of the active site clusters/cofactors of these different hydrogenases requires a suite of distinct accessory proteins for each class of enzyme and this occurs by different mechanisms, consistent with their independent evolutionary origins (Bock et al., 2006; Forzi & Sawers, 2007; McGlynn et al., 2007; Thauer et al., 2010).

[FeFe] hydrogenases have thus far only been identified in the bacterial domain and in some unicellular eukaryotes but are absent from the archaeal domain (Vignais & Billoud, 2007). [FeFe]-hydrogenase are generally involved in the production of hydrogen by anaerobic heterotrophic organisms such as Clostridium species, which use low potential ferredoxins or flavodoxins as electron carriers (Demuez et al., 2007). The diversity of [FeFe] hydrogenases can be catagorized on the basis of both primary sequence and structural considerations, including the size of the catalytic subunit and the presence of accessory subunits that typically harbor additional FeS clusters (Vignais et al., 2001). The catalytic center of [FeFe] hydrogenases consist of a [4Fe-4S] cluster connected to a binuclear iron site, forming the so-called H cluster (Nicolet et al., 2000). The conserved residues that coordinate the H cluster are arranged in three conserved cysteine-containing motifs termed L1, L2, and L3 (Vignais et al., 2001). While these signature motifs are conserved in [FeFe]-hydrogenase, significant variation exists in the composition and abundance of additional FeS clusters (so-called F clusters) as well as in the presence of accessory subunits (Meyer, 2007; Meuser et al., 2011).

In contrast to [FeFe]-hydrogenases, [NiFe] hydrogenases are widely distributed among both the bacterial and archaeal domains but have yet to be identified in the eukarya. Although the subunit composition of [NiFe]-hydrogenase vary, they consistently include a so-called large subunit that contains the active [NiFe] site and a small subunit that contains one or more FeS clusters. The large and small subunits show homology with the subunits NuoD and NuoB of respiratory complex I, respectively, and the additional subunits in the multimeric [NiFe] hydrogenases show sequence similarity to other Nuo subunits (Friedrich & Scheide, 2000). The [NiFe] active site is coordinated by four strictly conserved cysteine residues organized as two CXXC motifs located in the N- and C-terminus of the large subunit, termed the L1 and L2 motifs (Volbeda et al., 1995, 2002). Conservation in regions of the protein flanking the L1 and L2 motifs has been used to categorize the [NiFe] hydrogenases into four primary classes termed Groups 1–4. These show an apparent relationship with the inferred function of the enzyme and exhibit phylogenetic coherence (Fig. 1) (Vignais & Billoud, 2007).

Figure 1.

Bayesian phylogram of [NiFe] hydrogenases based on representative sequences of the large catalytic subunits that divide these enzymes into four major groups. Group 4 also contains related complexes that are not hydrogenases represented by Nuo (Complex I, NADH ubiquinone oxidoreductases) and Mbx (the unknown membrane-bound oxidoreductase involved in S0 reduction). Representative homologs (Table S1, Supporting information) were aligned using ClustalX (version 2.0.8) with the Gonnet 250 protein matrix and default gap extension and opening penalties (Larkin et al., 2007). ProtTest (version 2.0) (Abascal et al., 2005) was used to select WAG + I + G + F as the best-fit protein evolutionary model and the phylogeny was evaluated using MrBayes (version 3.1.2) (Huelsenbeck & Ronquist, 2001; Ronquist & Huelsenbeck, 2003), using the WAG evolutionary model with fixed (F) amino acid frequencies and gamma-distributed rate variation with a proportion of invariable sites (I+G). Tree topologies were sampled every 500 generations over 5 × 105 generations (after a burnin of 4 × 105) at likelihood stationarity and after convergence of two separate MCMC runs (average standard deviation of split frequencies <0.03) by MrBayes. The composite phylogram constructed from 1000 trees was projected with FigTree ( The depth of clades proportionally reflects the extent of the diversity within those clades.

The Group 4 hydrogenases are membrane bound and are distinct in that their catalytic subunits show a surprisingly low sequence similarity to hydrogenases that comprise the other three Groups, indicating a distinct evolutionary history for these membrane-bound enzymes (Vignais & Billoud, 2007). The Group 4 hydrogenases also include a number of distinct enzymes such as the formate hydrogen lyase or hydrogenase 3 (Hyc) of E. coli, which oxidizes formate and evolves H2 (Sawers et al., 1985; Sauter et al., 1992), and the CO-induced hydrogenases of Rhodospirillum rubrum and Carboxydothermus hydrogenoformans, which are involved in generating energy from the oxidation of CO to CO2 coupled with the production of H2 (Soboh et al., 2002; Singer et al., 2006). The Group 4 hydrogenases also include the six-subunit energy-converting hydrogenase (EchA-F), which was originally characterized in Methanosarcina barkeri where it was found to be required for growth on acetate (Meuer et al., 1999). This hydrogenase can reversibly generate H2 from reduced ferredoxin with the concomitant generation/utilization of an ion gradient; homologs of this enzyme are found among anaerobic bacteria and some methanogenic archaea. Group 4 [NiFe]-hydrogenase also include the ferredoxin-reducing, energy-converting hydrogenases of hydrogenotrophic methanogens (Eha and Ehb) that contain 17–20 subunits. These hydrogenases are thought to provide reduced ferredoxin for biosynthesis and for the first step of methanogenesis: the ferredoxin-dependent reduction in CO2 to form formylmethanofuran (Tersteegen & Hedderich, 1999; Major et al., 2010). Finally, the Group 4 enzymes include the Mbh-type energy-converting H2-producing enzymes that were first described in hyperthermophile Pyrococcus furiosus within the order Thermococcales (Sapra et al., 2000, 2003; Silva et al., 2000). These hyperthermophilic enzymes are the focus of this review.

Hydrogen and sulfur metabolism in the Thermococcales

Species of Pyrococcus and Thermococcus constitute the order of Thermococcales. These organisms are generally heterotrophic thermophiles that reduce elemental sulfur (S0) and ferment both simple and complex carbohydrates as well as peptide-based substrates. Many species can also grow on carbohydrates in the absence of S0 (Kengen et al., 1996). The pathways of carbohydrate degradation have been well studied in members of the Thermococcales, especially in Thermococcus kodakarensis and P. furiosus (Sakuraba et al., 2004; Verhees et al., 2004). During growth on sugars, such as the disaccharide maltose, the main fermentation products are H2, CO2 and acetate. The lack of other fermentation products such as lactate or ethanol indicates that all generated reducing equivalents are disposed of as H2. Glycolysis in the Thermo-coccales proceeds through a modified version of the Embden-Meyerhof pathway in which the classical glycolytic enzymes glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and phosphoglycerate kinase (PGK) are replaced by one ferredoxin-linked enzyme, glyceraldehyde-3-phosphate ferredoxin oxidoreductase (GAPOR), which does not generate ATP via substrate-level phosphorylation (Mukund & Adams, 1995; van der Oost et al., 1998). Omitting the ATP generating step of PGK therefore reduces the overall substrate-level ATP yield of glycolysis in the Thermococcales to zero (Verhees et al., 2004). Subsequently, pyruvate is oxidized to acetyl-CoA by pyruvate ferredoxin oxidoreductase (POR) which also uses the low potential electron carrier protein ferredoxin as acceptor (Blamey & Adams, 1993). This means that NADH is not formed during sugar catabolism and the only ATP generated (2 per mol glucose) is as a result of acetate production by acetyl-CoA synthetase (Mai & Adams, 1996).

It has been reported that in Tkodakarensis the conversion of phosphoenolpyruvate (PEP) to pyruvate, normally catalyzed by pyruvate kinase, can be performed by PEP synthetase. PEP synthetase uses AMP and phosphate instead of ADP for ATP generation and results in an additional yield of up to 2 ATP per glucose (Sakuraba & Ohshima, 2002; Sakuraba et al., 2004). A deletion strain of pyruvate kinase still displayed growth on carbohydrates but the deletion of PEP synthetase did not, indicating that PEP synthetase is essential for growth on carbohydrates (Imanaka et al., 2006). However, kinetic data indicate that PEP synthetase in vitro functions in the thermodynamically more favorable gluconeogenic direction (Hutchins et al., 2001). Currently, the amount of substrate-level ATP formed during sugar degradation in the Thermococcales is not clear. It is likely to be either (1) 2 ATP produced per glucose molecule using only pyruvate kinase or (2) 4 ATP produced using PEP synthetase. However, it is possible that both enzyme systems are utilized in concert depending on the physiological state of the cells, enabling metabolic flexibility and allowing for a greater energy yield under favorable conditions.

Ferredoxin is the electron acceptor for all oxidative steps in the glycolytic pathway in Thermococcales. Because of its low redox potential (−480 mV), H2 production (−420 mV) with ferredoxin as the electron donor is thermodynamically favorable, and therefore, all reducing equivalents generated in sugar metabolism can be disposed of as H2 (Park et al., 1991; Smith et al., 1995; Hagedoorn et al., 1998; Verhees et al., 2004). The enzyme responsible for H2 formation in the Thermococcales is a membrane-bound hydrogenase complex (Mbh). In P. furiosus, Mbh is a complicated enzyme encoded by a 14-gene cluster (PF1423-PF1436, MbhA-N), in which the MbhL subunit encodes the catalytic subunit (Fig. 2). Mbh is unique among [NiFe]-hydrogenases in that it generates both molecular H2 and can use the energy from this exergonic reaction to create an ion gradient (either H+ or Na+) across the membrane (Sapra et al., 2003). For this reason, Mbh has been postulated to represent the simplest form of respiration in modern day metabolism. In effect, ion pumping enables Mbh to regain some of the ATP that is not synthesized owing to the absence of the phosphorylation step in the modified glycolytic pathway with an estimated yield of 0.3 mol of ATP generated per mol H2 formed (Sapra et al., 2003). Deletion mutagenesis studies with T. kodakarensis show that Mbh is essential for growth in the absence of S0 as no other pathway for cofactor recycling is available to sustain growth (Kanai et al., 2011). As ferredoxin is used as the only electron acceptor during sugar oxidation, there must be a mechanism of generating NADPH to supply reducing equivalents for biosynthesis. In P. furiosus, two cytosolic [NiFe] hydrogenases (SHI and II) are present, and these can supply NADPH using H2 as electron donor (Ma & Adams, 2001). However, there must exist additional routes to generate NADPH because the deletion of either or both SHI and SHII did not display a growth defect (Lipscomb et al., 2011). Moreover, not all Thermococcales species contain both cytoplasmic hydrogenases and their true physiological functions are unclear.

Figure 2.

(a) Highly homologous multisubunit complexes and their catalytic subunits involved in evolving H2, reducing S0 and oxidizing electron carriers such as ferredoxin and NADH. Subunits homologous to Mrp are indicated in blue, hydrogenase subunits are displayed in red, the MbhH-type subunits with both homology to Mrp and hydrogenase modules are colored purple, the MbhI subunit, unique to Mrp-Mbh, is colored white, and active or potential active site containing subunit is boxed with red. The H subunit of Mrp-Mbh and Mrp-Mbx are most homologous to MrpD and an MrpA homolog is absent. (b). Comparison of the [NiFe] binding motifs (termed L1 and L2) in Pyrococcus furiosus Mbh with the comparable region in homologous proteins lacking hydrogenase activity. Cysteine residues are indicated in red and additional conserved residues in blue, Rc: Rhodobacter capsulatus. (c). Schematic representation of the P. furiosus Mrp-Mbh complex. The coloring scheme is the same as in Fig. 2a and subunits predicted to contain iron–sulfur clusters are indicated in blue.

The genome of P. furiosus contains a 13-gene cluster termed Mbx (PF1453-PF1441, MbxA-M) that is highly homologous to Mbh (see Table 1). Interestingly, the ion translocation modules as well as the [FeS] binding motifs present in Mbh are conserved in Mbx, suggesting that this complex could accept electrons from ferredoxin and create an ion gradient analogous to Mbh (Silva et al., 2000; Sapra et al., 2003). Previous biochemical studies have shown that Mbx has a distinct role in S0-dependent growth of P. furiosus (Schut et al., 2007). The addition of S0 elucidated a primary response that included the down-regulation of all hydrogenase-related genes (encoding Mbh and two cytosolic hydrogenases SHI and SHII) and the concomitant up-regulation of a number of genes presumably involved in the reduction of sulfur, including the genes encoding the Mbx complex, a pyridine nucleotide-disulfide oxidoreductase domain containing protein (PF1186), a glutaredoxin-like protein proposed to function as a protein disulfide oxidoreductase (Pdo, PF0094) (Ladenstein & Ren, 2006), and an uncharacterized transcriptional regulator cluster (PF2051-PF2052) (Schut et al., 2007). The gene product of PF1186 was purified from S0-grown cells and was characterized as a CoASH-dependent NADPH elemental sulfur reductase [Nsr, (Schut et al., 2007)]. This enzyme was implicated as the key enzyme in the S0 reduction pathway. It was further proposed that Mbx couples the oxidation of ferredoxin to NADPH formation, which can donate electrons to the soluble Nsr (Fig. 3). However, the exact physiological function of the Mbx complex has been hard to define, as no ferredoxin-dependent NADP reduction or ferredoxin-dependent S0 reduction activities could be measured in cell extracts or membrane preparations of P. furiosus (Schut et al., 2007). Confirmation of Mbx involvement in S0 reduction was recently provided by the disruption mutant of Mbx (deletion of MbxL, potentially the active subunit) whose growth in the presence of S0 is highly impaired (Bridger et al., 2011). Surprisingly, the Nsr deletion strain showed no obvious growth phenotype when grown on S0, and the amount of sulfide produced via S0 reduction was not significantly affected. In contrast, the MbxL deletion mutant produced no sulfide (above the abiotic background) and exhibited a clear growth defect in the presence of S0 with a reduction in cell yield of ~70% (Bridger et al., 2011). These results demonstrate that Nsr is not essential for growth with S0, but Mbx is an essential component of the S0 reduction system. It is possible that the NADPH produced by Mbx in Nsr deletion strain is used by other enzymes in a promiscuous manner to reduce S0, and such activity has been previously shown in vitro for ferredoxin NADPH oxidoreductase (FNOR) and both soluble hydrogenase I and hydrogenase II (Ma et al., 1993; Ma & Adams, 1994). Nevertheless, these results illustrate that Mbh and Mbx are highly analogous complexes that function to re-oxidize the reduced ferredoxin produced in energy metabolism in the absence and presence of S0, respectively.

Figure 3.

Schematic representation of the proposed physiological roles of MBH and MBX in the production of H2 and hydrogen sulfide in Pyrococcus furiosus and related organisms. Fdox and Fdred indicate oxidized ferredoxin and reduced ferredoxin, respectively.

Table 1. Properties and homologies of Mrp-Mbh and Mrp-Mbx genesa
Mbh locusGene# Cys (mot)Location (TMD)bMbx homologEch homologNuo homologMrp homologc
  1. a

    Homology searches were performed using NCBI blast and JCVI CMR (Altschul et al., 1997; Peterson et al., 2001).

  2. b

    The predicted cellular location is listed as Cyto, cytoplasm; Mem, membrane. The number of predicted transmembrane domains (TMD) is indicated [SOSUI, (Hirokawa et al., 1998)].

  3. c

    No MrpA homologs are included in the Mrp-Mbh type complexes (see text for details).

  4. d

    MbhG has a low homology to NuoK (Mathiesen & Hagerhall, 2003).

  5. e

    NuoL, M and N are homologous (Mathiesen & Hagerhall, 2002).

  6. f

    Homology between EchD, MbhK, NuoC and MbxK is weak.

  7. Mrp (multiple resistance and pH adaptation), Na+/H+ antiporter module; Mbh, membrane bound hydrogenase module; Mbx, membrane bound oxidoreductase module (unknown substrate); Ech, energy converting hydrogenase; Nuo, NADH ubiquinone oxidoreductase.

PF1423MbhA1Mem (3)MbxA  MrpE
PF1424MbhB0Mem (3)MbxB  MrpF
PF1425MbhC0Mem (3)MbxC  MrpG
PF1426MbhD0Mem (3)MbxD  MrpB
PF1427MbhE0Mem (2)MbxF  MrpB
PF1428MbhF0Mem (4)MbxF  MrpB
PF1429MbhG1Mem (3)MbxG NuoKdMrpC
PF1430MbhH2Mem (13)MbxH, H'EchANuoL, M, NeMrpD
PF1431MbhI0Mem (2)    
PF1432MbhJ5 [FeS]CytoMbxJEchCNuoB 
PF1434MbhL5 [NiFe]CytoMbxLEchENuoD 
PF1435MbhM0Mem (8)MbxMEchBNuoH 
PF1436MbhN8 2x[FeS]CytoMbxNEchFNuoI 

Regulation of hydrogen and sulfur metabolism

The response to the availability of S0 in P. furiosus was shown to be orchestrated by a transcriptional regulator termed SurR that contains a CXXC motif redox switch (Lipscomb et al., 2009). In its reduced state, SurR activates transcription of the genes encoding soluble hydrogenases I and II (SHI and SHII) and the membrane-bound hydrogenase (Mbh), as well as a number of other genes involved in H2 metabolism (Lipscomb et al., 2009). At the same time, reduced SurR represses the expression of genes including those that are potentially involved in the reduction in S0: Nsr, Mbx, Pdo, and the regulator cluster PF2051-PF2052. By in vitro transcription and binding studies it was shown that in the presence of S0 the CxxC motif in SurR becomes oxidized and renders the protein unable to bind to promoter DNA, resulting in the de-regulation of genes in the SurR regulon. As a result, the three hydrogenases (Mbh, SHI and SHII) are no longer expressed, while the genes involved in S0 metabolism are no longer repressed and their transcript levels increase (Lipscomb et al., 2009; Yang et al., 2010). The SurR regulator binds to a specific sequence (GTTn3AAC) in its target promoter (including its own promoter) and most of the genes observed in the primary S0 response contain this SurR binding motif. Bioinformatic analyses indicate that this regulator is only found in Thermococcales, indicating that the SurR S0/hydrogen regulation uniquely evolved within this order.

Based on biochemical and transcriptional data, we propose a core set of genes involved in the combined H2 and sulfur metabolism of P. furiosus and of members of the Thermococcales in general involving Mbh, SHI, SHII, Mbx, Nsr, SurR, Pdo and PF2051/52 (Table 2). This core set of genes is conserved within the Thermococcales, with the exception of T. gammatolerans, which does not encode any cytosolic [NiFe]-hydrogenases, and T. kodakaresis and P. horikoshii, which contain only a single cytoplasmic hydrogenase. The genes encoding the glutaredoxin-like protein (Pdo) and SurR share the same promoter region in all Thermococcales and both are subjected to SurR control. However, the precise role of Pdo in sulfur metabolism is not clear and no biochemical activity related to S0 reduction has been demonstrated (Schut et al., 2007). The Pdo gene product is characterized as a protein disulfide oxidoreductase, capable of reducing disulfide bonds in insulin using DTT as a reductant (Ren et al., 1998; Pedone et al., 2004). Homologs of Pdo are widespread in both the archaeal and bacterial domains and are also found in organisms that do not utilize sulfur to support their metabolism, indicating that Pdo is not likely to be uniquely involved in S0 reduction. In the bacterium Thermotoga maritima, a Pdo homolog was characterized and it was shown that a thioredoxin reductase could transfer electrons from NADPH to Pdo (Yang & Ma, 2010). The genomes of the Thermococcales also encode a homolog of the thioredoxin reductase and the homolog in P. furiosus (PF1422) is also part of the SurR regulon. This suggests that thiol-based electron flow might also be involved in S0 reduction by the members of the Thermococcales, although the enzymes and pathways are unclear. Although the core set of H2 and sulfur-related genes linked by SurR appear to be restricted to the order of Thermococcales, homologs of both Mbh and Mbx are found in other microbial systems in which they also are related to H2 metabolism. These will be discussed later, using the P. furiosus Mrp-Mbh as the point of reference.

Table 2. Distribution of the core set of genes involved in hydrogen and sulfur metabolism in Thermococcalesa
  1. a

    Homology searches were performed using NCBI blast and JCVI CMR (Altschul et al., 1997; Peterson et al., 2001).

  2. b

    The genome of Thermococcus AM4 is unfinished and the ORF numbers are not consecutive.

Pyrococcus furiosusPF1423-36PF0891-94PF1329-32PF1453-41PF1186PF0095PF0094PF2051-52
Pyrococcus abyssiPAB1884-94PAB1784-87PAB0638-41PAB0485-96PAB0936PAB0108PAB2245PAB229-3022
Pyrococcus horikoshiiPH1427-40PH1290-94-PH1456-46PH0572PH0180PH0178PH0062-S001
Pyrococcus NA2PNA2_0015-28PNA2_1665-68PNA2_1555-51PNA2_0044-32PNA2_1248PNA2_0699PNA2_0698PNA2_0599-98
Pyrococcus yayanosiiPYCH_11230-360PYCH_08370-400PYCH_00020-50PYCH_11530-410PYCH_07890PYCH_16310PYCH_16300PYCH_14930-940
Thermococcus kodakarensisTK2080-93TK2072-69-TK1226-14TK1299TK1086TK1085TK1261-60
Thermococcus gammatoleransTGAM_0047-34--TGAM_0703-15TGAM_1050TGAM_1303TGAM_1302TGAM_0437-38
Thermococcus onnurineusTON_1582-95TON_0537-34TON_0055-52TON_0498-86TON_0305TON_0318TON_0319TON_0525-24
Thermococcus sibiricusTSIB_1309-96 TSIB_1295-82TSIB_0282-85TSIB_1520-23TSIB_1917-05TSIB_0263TSIB_0990TSIB_0991TSIB_1263-64
Thermococcus AM4bTAM4_1289-1255TAM4_579-655TAM4_307-399TAM4_474-490TAM4_984TAM4_2127TAM4_1905TAM4_989-549
Thermococcus barophilusTERMP_01498-85 TERMP_01484-71TERMP_00539-36TERMP_00067-70TERMP_00865-53TERMP_02185TERMP_00656TERMP_00657TERMP_00904-03
Thermococcus sp. 4557GQS_07585-520 GQS_07515-445GQS_02460-75GQS_04975-990GQS_02710-770GQS_04305GQS_03680GQS_03675GQS_02525-530

Mrp-Mbh-type complexes and their taxonomic distribution

The 14 subunits of the Mbh complex show modular composition, a mosaic of defined components that include Mrp-type Na+/H+ antiporter homologs (MbhA-H) and hydrogenase components (MbhH-N). MbhH-N exhibit homology to all of the subunits of Ech-type hydrogenases as well as to a number of components of complex I (see Table 1) (Soboh et al., 2004; Hedderich & Forzi, 2005; Swartz et al., 2005). The Mrp-like system was first genetically characterized in the alkaliphile Bacillus halodurans, where the Mrp Na+/H+ antiporter was identified as indispensable for pH homeostasis under alkaline conditions (Hamamoto et al., 1994). The Mrp system in this organism consists of seven genes (MrpA-G), all with transmembrane domains. Some degree of redundancy is apparent: MrpA contains modular elements that are homologous to MrpB and MrpD (Swartz et al., 2005), suggesting that MrpA evolved from a gene fusion involving the B and D genes. Intriguingly, the Mrp module in the P. furiosus Mbh cluster lacks MrpA and only encodes homologs of MrpB-G (Table 1 and Fig. 2). It should be noted that P. furiosus also contains another Mrp-like cluster (MrpB-MrpG: PF1153-PF1149) of unknown function that is not connected to Mbh or Mbx domains. This type of Mrp cluster is conserved throughout the Thermococcales as well as in many other archaea. Although the function of this Mrp-like cluster is unknown, its presence suggests that the Mrp-type systems in the Thermococcales also evolved independently and under different selective pressure from the hydrogenase components. It was demonstrated that the A1A0-type ATPase in P. furiosus forms ATP using sodium rather than a proton gradient. This suggests that the Mrp system could function in the exchange of a proton gradient generated by the hydrogenase module for a sodium gradient, enabling the ATPase to generate ATP (Pisa et al., 2007).

For the purposes of this review, we use the P. furiosus nomenclature and its 14-subunit (MbhA-N) composition as a template, defining Mbh by the modular composition of Mrp-type genes (MrpB-G homologs) and Group 4 [NiFe] hydrogenase-type genes as annotated in Table 1. Here we define the Mbh and Mbx clusters as Mrp-Mbh and Mrp-Mbx to emphasize and distinguish modular composition. MbhL contains the [NiFe] catatlyic site of the hydrogenase and this is coordinated by two pairs of cysteine residues termed L1 and L2 (Vignais & Billoud, 2007). The Group 4 [NiFe] hydrogenases can be readily identified based on sequence conservation in the motif flanking regions, as well as by phylogenetic clustering (Fig. 1). As discussed later, based on their catalytic subunits, the various Mbh-type enzymes can also be differentiated from other Group 4 [NiFe] hydrogenase in genome sequences and also from each other. While the subunit composition of the Mbx cluster is very similar to the Mbh cluster and the genes comprising the Mbx cluster have homologs in Mbh (Table 1), these two complexes can be readily differentiated from each other. The Mbx cluster is composed of 13 genes and the difference in the number of genes when compared to the Mbh cluster (14 genes) is owing to the fact that MbxE resembles a fusion of MbhE and MbhF, MbhH is represented by two homologous subunits in Mbx (MbxH and MbxH'), and no homolog of MbhI (which is of unknown function) is present in the Mbx cluster. Mbx does not have significant hydrogenase activity based on biochemical data (Adams et al., 2001). Moreover, the L1 and L2 [NiFe] coordinating residues in Mbh are not conserved in Mbx as the second cysteine in each motif is absent (Fig. 2b) and significant diversification is present in the L1 and L2 flanking regions relative to Group 4 hydrogenases (Schut et al., 2007; Vignais & Billoud, 2007).

Mrp-Mbh-type complexes in Euryarchaeota

The Mrp-Mbh and Mrp-Mbx clusters are represented by highly similar gene clusters in the available Thermococcales genome sequences, (Table 3). This conservation suggests that these gene clusters evolved early in the evolution of the Thermococcales and that they perform a central role in the metabolism of these organisms. Interestingly, Thermococcus sp. 4557, T. sibiricus, and T. barophilus contain a second copy of the Mrp-Mbh cluster (Mrp-Mbh2) directly adjacent to the first copy (Table 3). Mrp-Mbh2 differs from Mrp-Mbh1 in that the second [NiFe] binding motif (L2) contains an additional cysteine (CMCC). The function of this second Mrp-Mbh cluster is not clear. Most methanogens also contain homologs of Group 4 energy-converting membrane-bound hydrogenases of either the Ech- or Eha-/Ehb-type. These proteins are similar to the Mrp-Mbh-type complex but have a different subunit composition (Vignais & Billoud, 2007; Thauer et al., 2010). Interestingly, three methanogens of the order Methanomicrobiales (Methanoplanus petrolearius, Methanospirillum hungatei and Methanocorpusculum labreanum) do encode a bonafide Mrp-Mbh homolog (Table 3); however, the presence of Mrp-Mbh in these organisms is puzzling considering that their genomes also encode Eha- and Ech-type hydrogenases (Anderson et al., 2009).

Table 3. Distribution of Mrp-Mbh type oxidoreductases in the domains archaea and bacteriaa
Species/Mbh-type clusterMrp-MbhMrp-Mbh2Mbh-NAD(P)HMrp-Mbh-Fdh1Mrp-Mbh-Fdh2Mrp-Mbh-CodhMrp-MbxMbx-NAD(P)HEnergy sourceElectron acceptors
  1. a

    Homology searches were performed using NCBI blast and JCVI CMR (Altschul et al., 1997; Peterson et al., 2001)

  2. b

    S. marinus Mbh-NAD(P)H cluster contains homologs of all Mrp-Mbh subunits and three homologs to the MbhH subunit

  3. c

    K. olearia Mbh-NADPH cluster contains homologs of Mrp-Mbh subunits except for MbhI.

Pyrococcus furiosusPF1423-36     PF1453-41 Pep, CBHH+, S0
Pyrococcus AbyssiPAB1884-94  PAB1389-02  PAB0485-96 PepH+, S0
Pyrococcus horikoshiiPH1427-40     PH1456-46 PepH+, S0
Pyrococcus NA2PNA2_0015-28     PNA2_0044-32 PepH+, S0
Pyrococcus yayanosiiPYCH_11240-360   PYCH_11030-200 PYCH_11530-410 Pep, CBH 
Thermococcus kodakarensisTK2080-93     TK1226-14 Pep, CBHH+, S0
Thermococcus gammatoleransTGAM_0047-34   TGAM_0065-48 TGAM_0703-15 Pep, formateH+, S0
Thermococcus onnurineusTON_1582-95  TON_0281-66TON_1563-80TON_1017-31TON_0498-86 Pep, formate, COH+, S0
Thermococcus sp. AM4TAM4_1289-1255    TAM4_1057-1048TAM4_474-690 Pep, COH+, S0
Thermococcus barophilusTERMP_01498-85TERMP_01484-1471   TERMP_01154-39TERMP_00865-53 Pep, COH+, S0
Thermococcus sibiricusTSIB_1309-96TSIB_1295-82    TSIB_1917-05 Pep, CBHH+, S0
Thermococcus sp. 4557GQS_07585-520GQS_07515-445 GQS_02455-380  GQS_02710-770   
Aciduliprofundum booneiAboo_0664-77     Aboo_0386-74 PepH+, S0 or Fe3+
Methanoplanus petroleariusMpet_1584-71       H2CO2
Methanospirillum hungateiMhun_2579-92       H2CO2
Methanocorpusculum labreanumMlab_0948-61       H2CO2
Desulfurococcus kamchatkensisDKAM_0419-07 DKAM_1015-23   DKAM_0956-45 Pep, CBHH+, S0
Desulfurococcus mucosusDesmu_0818-06 Desmu_0061-53   Desmu_1002-13 Pep, CBHH+, S0
Thermosphaera aggregansTagg_0624-36 Tagg_0059-50   Tagg_1025-36 PepH+
Ignisphaera aggregansIgag_1902-14       Pep, CBHH+
Staphylothermus marinusSmar_0018-30 Smar_1055-71b   Smar_0645-57 PepH+, S0
Staphylothermus hellenicus      Shell_0146-34 Pep, CBHS0
Thermotoga maritima       TM1205-17Pep, CBHH+, S0, math formula
Thermotoga neapolitana       CTN_1366-54Pep, CBHH+, S0, math formula
Thermotoga sp. RQ2       TRQ2_1613-01Pep, CBHH+, S0, math formula
Petrotoga mobilis       Pmob_0897-85Pep, CBHH+, S0, math formula
Thermosipho melanesiensis       Tmel_1210-22Pep, CBHH+, S0, math formula
Thermosipho africanus       THA_1696-82Pep, CBHH+, S0, math formula
Fervidobacterium nodosum       Fnod_1748-35Pep, CBHH+, S0, math formula
Kosmotoga olearia  Kole_0574-60c    Kole_2063-51Pep, CBHH+, S0, math formula
Coprothermobacter proteolyticusCOPR5265_0920-34       Pep, CBHH+, S0, math formula
Aminobacterium colombienseAmico_1275-62       Pep, amino acidsH+
Thermoanaerobacter wiegelii      Thewi_0034-46 CBHH+

Aciduliprofundum boonei (Reysenbach et al., 2006; Reysenbach & Flores, 2008), which so far is the only cultivable representative of the Archaeal DHVE2 order, also contains Mrp-Mbh and Mrp-Mbx homologs (Table 3). It also contains a homolog of cytoplasmic hydrogenase SHI but appears to lack SurR and Nsr. Aciduliprofundum boonei is a strict anaerobe and is proposed to utilize peptides as carbon and energy sources using a metabolic pathway similar to that found in the Thermococcales with sulfur or ferric iron as the electron acceptor (Reysenbach & Flores, 2008). Representatives of the order of DHVE2 are thought to be very abundant thermoacidophiles in hydrothermal vent systems and are most closely related to the Thermoplasmales (Reysenbach et al., 2006).

Analysis of the available Thermococales genome sequences has yielded additional gene clusters encoding multi-domain Mrp-Mbh complexes (Fig. 4). For example, T. onnurineus harbors four gene clusters containing Mrp-Mbh modules (Lee et al., 2008). Two of these clusters contain genes potentially coding for a formate dehydrogenase module. We will term these Mrp-Mbh-Fdh1 and Mrp-Mbh-Fdh2 (see Table 3) (Lim et al., 2010). The conversion of formate to bicarbonate and H2 was until recently not considered a means of energy generation but rather a response to avoid accumulation of formate (Bohm et al., 1990; Takacs et al., 2008). However, it was shown that T. onnurineus could derive enough energy from the conversion of formate to bicarbonate and H2 to sustain growth (Kim et al., 2010). Through transcript and mutational analysis it became clear that only the Mrp-Mbh-Fdh2 cluster (TON_1563-TON_1580) is essential for growth on formate. Mrp-Mbh-Fdh1 and Mrp-Mbh-Fdh2 are very similar in both sequence and gene synteny but Mrp-Mbh-Fdh2 harbors an additional gene potentially coding for a formate transporter. On the basis of the gene organization of Mrp-Mbh-Fdh2 it was proposed that formate is transported into the cell where the Fdh module converts formate to bicarbonate and electrons are shuttled to the Mrp-Mbh module, which produces H2 with the concomitant generation of an electrochemical gradient ultimately leading to ATP formation (Kim et al., 2010). Pyrococcus yayanosii and Thermococcus gammatolerans also contain Mrp-Mbh-Fdh2 and may also have the potential to utilize formate as an energy source in a manner similar to T. onnurineus (Zivanovic et al., 2009; Jun et al., 2011). The presence of an Mrp-Mbh-Fdh1 lacking a formate transporter in P. abyssi and T onnurineus indicates that formate potentially is formed intracellularly and that this can be utilized to generate additional energy by conversion to hydrogen. The generation of formate intracellularly could be accomplished through the activity of a pyruvate-formate lyase; however, only the genomes of T. kodakarensis and T. sibiricus encode such an enzyme (TK0289, TSIB_0631) and neither harbors an Mrp-Mbh-Fdh1-type gene cluster. The possible presence and source of any intracellular formate therefore remains unknown (Fukui et al., 2005; Mardanov et al., 2009). Interestingly, T. kodakarensis contains a cluster of soluble formate dehydrogenase-related genes just upstream of its Mrp-Mbh locus (TK2076-TK2079) but their function is also unknown.

Figure 4.

The utilization of the Mrp-Mbh-type complexes in enzymes involved in metabolism of formate, carbon monoxide, S0 and NAD(P)H by the inclusion of additional functional modules. Based on the nomenclature of Pyrococcus furiosus Mrp-Mbh (Table 1) the modules are defined as follow: Mrp, MbhA-H; Mbh, MbhH-N (MbhH is common to both Mrp and Mbh modules); Mbx, MbxH-N; Mrp-Mbh-Fdh, the formate dehydrogenase module is represented by TON_1563 (FdhA) and TON_1564 (Lee et al., 2008); Mrp-Mbh-CodH, the carbon monoxide dehydrogenase module is represented by TON_1017 (CooS) and TON_1018 (CooF) (Lee et al., 2008); Mrp-Mbh-NAD(P)H, the FAD/NAD(P)H binding module is represented by Tagg_0059 [pyridine nucleotide-disulfide oxidoreductase domain, IPR013027 (Hunter et al., 2009)] and Tagg_0058 ([FeS] containing (IPR001450, Hunter et al., 2009); Mrp-Mbx-NAD(P)H, the FAD/NAD(P)H binding module is represented by TM1217 [pyridine nucleotide-disulfide oxidoreductase domain (IPR013027, Hunter et al., 2009)]. The small light blue bar for the Mrp module in Mbh-NAD(P)H represents two genes homologous to MbhH.

Bioinformatic analyses also reveal an Mrp-Mbh-type modular system with a carbon monoxide dehydrogenase module in T. onnurineus, which we will term Mrp-Mbh-Codh, (Lee et al., 2008) (Table 3). Growth experiments indeed demonstrated that T. onnurineus can grow in the presence of CO and form H2 in an energy yielding fashion (Yun et al., 2011). Carboxydotrophic growth has also been shown for T. barophilus and Thermococcus AM4 and their genomes also harbor clusters encoding an Mrp-Mbh-Codh (Sokolova et al., 2004; Vannier et al., 2011), Table 3). The cluster is composed of three functional domains, Mrp, Mbh and a Codh domain (CooF and CooS), together with a potential regulator and a Codh maturation factor (CooC) encoded within the same cluster (Lim et al., 2010). This cluster is analogous yet very distinct from the CO-oxidizing, H2-forming enzyme complex of the bacterium C. hydrogenoformans. This contains an Ech-type hydrogenase and two CO dehydrogenase subunits [CooF and CooS; (Soboh et al., 2002)] but lacks an Mrp module.

Mrp-Mbh-type complexes in Crenarchaeota

Homologs of Mrp-Mbh and Mrp-Mbx orthologs have a limited distribution within the phylum Crenarchaeota. Only one order, the Desulfurococcales, contain complete homologs of Mrp-Mbh and Mrp-Mbx (Table 3). Its members are characterized by a S0-dependent and heterotrophic life style analogous to that of the Thermococcales, although a few strains are reported to not utilize S0 and some can grow autotrophically (Vieille & Zeikus, 2001; Boyd et al., 2007; Prokofeva et al., 2009). The genome sequences of several members of the Desulfurococcales are available as a number of type strains were sequenced as part of the Genomic Encyclopedia of Bacteria and Archaea project ( The homologs of Mrp-Mbh and Mrp-Mbx found in the Desulfurococcales are present in gene clusters that are rearranged compared with those found in the Thermococcales (Table 3). For example, both Mrp-Mbh and Mrp-Mbx contain a fused homolog of the E and F subunit, the homolog to MbhI is poorly conserved, a fused homolog of MbxJ and MbxK is present, there is only one homolog to MbxH rather than two, and in the Mbx cluster an additional gene encoding an unknown membrane-bound subunit is present.

For other members of the Crenarchaeota, the presence or absence of homologs of Mrp-Mbh and Mrp-Mbx is indicative of their metabolic capability (Table 3). For example, Ignisphaera aggregans does not contain an Mrp-Mbx homolog and does not utilize elemental sulfur, in fact, its growth is inhibited by S0 (Niederberger et al., 2006). Ignisphaera aggregans was isolated from a terrestrial geothermal spring in New Zealand and grows by fermenting carbohydrates. As a homolog of glyceraldehyde-3-phosphate ferredoxin oxidoreductase (GAPOR) is encoded in the genome of I. aggregans (Igag_1482), it is presumed to utilize a modified EM-pathway analogous to that in the Thermococcales where reducing equivalents in the form of reduced ferredoxin are disposed of as H2 (Niederberger et al., 2006; Goker et al., 2011). Interestingly, I.aggregans does not encode any other hydrogenase besides Mrp-Mbh, indicating a relatively simple H2 respiratory-type metabolism (Goker et al., 2011). In contrast, Staphylothermus hellenicus is unusual in that its genome does not seem to encode any recognizable hydrogenase, although an Mrp-Mbx cluster is present. Perhaps not surprisingly, S. hellenicus is obligatory dependent on S0 for growth and does not form H2 at all (Arab et al., 2000). Staphylothermus hellenicus was isolated from a marine environment and is reported to grow only on complex media in the presence of S0 (Arab et al., 2000). In contrast, the growth of Thermosphaera (Ta) aggregans, which has both Mrp-Mbx and Mrp-Mbh clusters, is inhibited by the presence of S0 (Huber et al., 1998). These results suggest that the Mrp-Mbh and Mrp-Mbx play important roles in H2 and sulfur metabolism in several Desulfurococcales, although their exact role in sulfur reduction has yet to be definitively demonstrated.

In addition to Mrp-Mbh and Mrp-Mbx homologs, Desulfurococcus kamchatkensis, D. mucosus and Ta. aggregans contain another member of the Group 4 hydrogenases (Table 3). This complex consists of homologs of subunits MbhH-MbhN of the Mbh hydrogenase module and a potential NAD(P)H interaction module (Fig. 4). The subunits for the Mrp antiporter module are absent with the exception of two homologs of the MbhH subunit. The NAD(P)H module consists of two ORFs, one with homology to a glutamate synthase subunit (pyridine nucleotide-disulfide oxidoreductase domain) and the second with four potential iron–sulfur cluster-binding motifs. Considering that this potential NAD(P)H-linked hydrogenase, which we termed Mbh-NAD(P)H, contains an energy-converting module, it may be the case that it can use an electrochemical gradient to couple the oxidation of NAD(P)H to H2 production (Table 3). This is utilized during the fermentation of peptides, where NAD(P)H is formed by the deamination of amino acids but the reoxidation of NAD(P)H coupled to H2 production is thermodynamically unfavorable (Thauer et al., 1977). This is precisely the reason that an external electron acceptor such as S0 is needed during peptide-dependent growth in the Thermococcales (Adams et al., 2001). In the absence of external electron acceptors, Ta. aggregans ferments peptides to H2, CO2 and various organic acids presumably with the Mbh-NAD(P)H complex encoded in its genome (Niederberger et al., 2006).

Interestingly, S. marinus contains a variant of an Mbh-NADPH system, in which a complete Mrp module is encoded along with three copies of the MbhH subunit (Table 3). This organism exhibits an obligatory S0-dependent mode of growth on complex substrates (yeast extract and peptone). When S0 is limiting, the growth rate and final cell yield decrease and cells produce H2 and hydrogen sulfide simultaneously (Hao & Ma, 2003). However, it should be noted that growth was observed using a very low sulfur concentration (0.01 g L−1) with minimal H2S production relative to H2 production [87 μM H2S vs. 1.78 mM H2 (Hao & Ma, 2003)]. This indicates that the Mbh-NAD(P)H complex might function as an energy-converting NAD(P)H-linked hydrogenase with H2 evolution from NAD(P)H using an ion gradient to drive the reaction. This organism might still need a minimal amount of S0 for other purposes. In contrast, Hyperthermus butylicus and Ignicoccus hospitalis, also within the order of Desulfurococcales, exhibit S0-dependent growth; however, they do not contain homologs of Mrp-Mbx or Mrp-Mbh (Brugger et al., 2007; Podar et al., 2008). These organisms most likely utilize a membrane-bound molybdopterin-containing sulfur reductase (Igni_0801-Igni_0803, Hbut_0373-Hbut_0371) as described for Acidianus ambivalens (Laska et al., 2003). Both organisms can generate energy by the oxidation of H2 using a Group 1 membrane-bound uptake hydrogenase (Igni_1367-Igni_1369, Hbut_1368-Hbut_1371) coupled to S0 as the electron acceptor but only I. hospitalis has been shown to grow autotrophically (Zillig et al., 1990; Paper et al., 2007).

Of the Crenarchaeota, members of the order Desulfurococcales therefore make interesting examples of the manner in which Mbh, Mrp and Mbx modules have been utilized. Members of the order Acidilobales (Acidilobus saccharavorans and A. sulfurireducens), a group of sulfur-reducing organisms that characteristically inhabit acidic, terrestrial geothermal springs, do not contain homologs of Mbx or Mbh (E.S. Boyd, unpublished data), although a close homolog [NADH dehydrogenase subunit D (NuoD)] was incorrectly annotated as MbxL (Mardanov et al., 2010). It is likely that these organisms utilize a membrane-bound molybdopterin-containing sulfur reductase (ASAC_1394-ASAC_1397) as suggested previously for H. butylicus and I. hospitalis. These results are consistent with the physiological characterization of the isolates as S0-dependent heterotrophs that are unable to grow and produce H2 in the absence of an external electron acceptor (Boyd et al., 2007; Prokofeva et al., 2009). Thermofilum pendens is so far the only organism within the order Thermoproteales that contains representatives of the Group 4 hydrogenases (Tpen_1070-Tpen_1077, Tpen_0190-Tpen_0180) (Anderson et al., 2008). However, most of the genes encoding subunits of the Mrp module are missing and the functions of these hydrogenases are not known.

Mrp-Mbh-type complexes in Bacteria

The bacterial order Thermotogales is comprised of thermophiles and genome sequences of its members indicate a relatively high percentage of genes obtained through horizontal transfer (Nelson et al., 1999; Zhaxybayeva et al., 2009). These also include Mrp-Mbh-type gene clusters (Table 3). This order is characterized by the ability of its members to metabolize a wide variety of carbohydrates (Huber et al., 1986) to produce acetate, CO2 and H2 from carbohydrate metabolism (Schroder et al., 1994). Thermotoga maritima is the type species of this order and oxidizes a wide variety of both simple and complex carbohydrates (Conners et al., 2006). However, the pathway of electron flow is different from that found in the Thermococcales, which have a modified, ferredoxin-dependent glycolytic pathway. In contrast, T. maritima uses the classical Embden-Meyerhof and Entner-Doudoroff pathways to oxidize glucose (Selig et al., 1997). This means that both ferredoxin and NAD function as physiological electron carriers and that both are recycled with the generation of H2. The generation of H2 is accomplished by a novel so-called bifurcating [FeFe] hydrogenase that simultaneously oxidizes ferredoxin and NADH with concomitant production of H2 using the exergonic reaction of ferredoxin oxidation to drive the endergonic oxidation of NADH (Schut & Adams, 2009). One of the genes clusters that was transferred from the Thermococcales to the Thermotogales is an Mrp-Mbx-type cluster in which all 13 subunits found in P. furiosus Mrp-Mbx are conserved both in sequence and synteny. However, the last subunit in the operon (Tm1217, the N-terminus is homologous to Mrp-MbxN) appears to be fused with a pyridine nucleotide-disulfide oxidoreductase domain in the Thermotogales (Table 3 and Fig. 4). Interestingly, the NAD(P)H input module from T. maritima is homologous to the input module of the Mbh-NAD(P)H-type hydrogenase found in some of the Desulfurococcales mentioned previously, suggesting that they have a common origin. However, the function of this cluster, which we have termed Mrp-Mbx-NAD(P)H, is not known.

Although several members of the Thermotogales can use alternative electron acceptors for growth, such as S0 and thiosulfate, very little is known about how these compounds are metabolized (Ravot et al., 1995). Because Mrp-Mbx is involved in S0 metabolism in the Thermococales (Schut et al., 2007; Bridger et al., 2011; Kanai et al., 2011) it is tempting to speculate that the Mrp-Mbx homolog plays a role in S0 or thiosulfate reduction in the Thermotogales. However, this Mrp-Mbx-NAD(P)H cluster is not universally conserved within the Thermotogales. It is not present in T. petrophila, T. lettingae, and T. naphtophila; yet, these organisms are still capable of reducing S0 and/or thiosulfate (Takahata et al., 2001; Balk et al., 2002). This observation indicates that the mechanism of S0 reduction might not be dependent on these Mrp-Mbx homologs and serves to emphasize our lack of understanding of S0 metabolism in the Thermotogales.

Most of the Thermotogales species produce H2 as their primary reduced product of fermentation and most of them contain a homolog of the bifurcating [FeFe] hydrogenase found in T. maritima (Schut & Adams, 2009). One exception is Kosmotoga olearia, which lacks the trimeric [FeFe] hydrogenase and contains two potential monomeric [FeFe] hydrogenases (Kole_0172 and Kole_1794), a Mrp-Mbx-NAD(P)H homolog (Kole_2063-Kole_2051), and a Mrp-Mbh homolog with a NAD(P)H input module (termed Mrp-Mbh-NAD(P)H, Kole_0574-Kole_0560, Table 3). Interestingly, the NAD(P)H input module consists of three subunits. One contains an iron–sulfur motif and is homologous to that in Mbh-NADPH system of the Desulfurococcales, while the other two share homology with two of the subunits (gamma and beta) of the cytosolic hydrogenases (such as P. furiosus SHI) present in the Thermococcales. Although very little information is available on K. olearia, it ferments carbohydrates to acetate, CO2 and H2 like other members of the Thermotogales, (Dipippo et al., 2009). The absence of a bifurcating hydrogenase indicates that the Mrp-Mbh-NAD(P)H hydrogenase complex, possibly in combination with one or both monomeric [FeFe] hydrogenases, could convert most reducing equivalents generated in sugar degradation to H2 in which the Mrp-Mbh_NAD(P)H complex functions as a NAD(P)H-linked hydrogenase using the ion gradient to drive H2 production.

Two other bacteria, in this case non-thermophilic, contain homologs of Mrp-Mbh, Aminobacterium colombiense, and Coprothermobacter proteolyticus [Amico_1275-Amico_1262, COPRO5265_0920- COPRO5265_0934, Table 3 (Chertkov et al., 2011)]. Both organisms were isolated from waste water and ferment both complex substrates and a variety of sugars mainly to acetate, H2, and CO2 (Ollivier et al., 1985; Baena et al., 1998; Etchebehere & Muxi, 2000). Aminobacterium colombiense contains an additional cytosolic [NiFe] hydrogenase cluster (Amico_1558-1553) of the Group 3d type (Vignais & Billoud, 2007). Interestingly, from sequence comparisons, the [NiFe] active site subunit of the Mrp-Mbh in C. proteolyticus (COPRO5265_0933) does not appear to be synthesized as an inactive preprotein. It lacks the C-terminal extension that is normally cleaved to activate the NiFe-site, and accordingly, an associated protease is not encoded in the adjacent cluster of accessory genes (COPRO5265_0935- COPRO5265_0941). Coprothermobacter proteolyticus contains two additional [FeFe] hydrogenases, one potentially of the bifurcating type (COPRO5265_0177- COPRO5265_0181) and a monomeric type (COPRO5256_0174). Strikingly, both bacteria encode all homologs of the Thermococcales Mrp-Mbh cluster. In A. colombiense, the synteny is conserved while in C. proteolyticus the [NiFe] maturation genes are co-localized with the Mrp-Mbh cluster. Given the limited distribution of the Mbh type within the domain bacteria and the organization in these two bacteria it is tempting to speculate that the gene clusters encoding these Mrp-Mbh homologs have been horizontally transferred from an archaeal source relatively recently in evolutionary time. Another striking case of potential horizontal transfer is represented by the presence of an Mrp-Mbx homolog in the moderate thermophilic anaerobe, Thermoanaerobacter wiegelii (Table 3). It is not known whether this Mrp-Mbx cluster is active or whether this organism is capable utilizing S0 as an electron acceptor.

Phylogeny of Mrp-Mbh-type complexes in Archaea and Bacteria

Phylogenetic analysis based on the active site containing L-subunit of the Mbh/Mbx module indicates that these complexes likely originated in the archaeal domain, as indicated by paraphyletic nesting of monophyletic bacterial sequences within these lineages. Aquisition of Mrp-Mbh/Mbx complexes in bacteria likely occurred via horizontal gene transfer (HGT) early in the evolutionary history of bacteria, with several additional HGT events occurring more recently (Fig. 5). For example, the Mrp-Mbh found in A. colombiense (Amico_1275-Amico_1262), clearly clusters together with the Mrp-Mbh complexes of the Thermococcales and the synteny of the genes in cluster is preserved, indicative of a relatively recent lateral transfer event between these lineages (Fig. 5). This analysis also shows that the monophyletic Crenarchaeal and Euryarchaeal Mrp-Mbh and Mrp-Mbx lineages nest with bacterial lineages, suggesting that acquisition of Mrp-Mbh and Mrp-Mbx in bacteria occurred after the split between the Crenarchaea and the Euryarchaea (Fig. 5). Intriguigingly, the Mrp-Mbh complexes that contain additional catalytic domains (Mrp-Mbh-Fdh, Mrp-Mbh-Codh and Mrp-Mbh-NADPH) appear to cluster together, suggesting that these complexes have a common ancestor despite the fact that Mrp-Mbh-NADPH is found in Crenarchaeota while Mrp-Mbh-Fdh and Mrp-Mbh-Codh are present in the Euryarchaeota and are so far restricted to the Thermococales (Fig. 5). The most parsimonious interpretation for this observation is that the additional Fdh, Codh or NADPH module was initially recruited to an ancestral Mrp-Mbh module, likely in a member of the Euryarchaeota, and that this resulted in the diversification of this lineage away from the Mrp-Mbh module found in Euryarchaeota today. Importantly, the conservation in Mrp-Mbh modular content observed among the two primary lineages comprising Mrp-Mbh, and the primary lineage that comprises Mrp-Mbh-Fdh/Mrp-Mbh-Codh/Mrp-Mbh-NADPH, coupled with the monophyletic nature of Mrp-Mbh in Crenarchaeota and Euryarchaeota, suggests that Mrp-Mbh emerged very early during the evolution of Archaea.

Figure 5.

Phylogenetic distribution of Mrp-Mbh/Mbx-type complexes and their relationship to Complex I. The terminal labels correspond to taxa identified in Tables 2 and 3. The phylogram includes representative sequences from a number of Mrp-Mbh/Mbx functions (Table 3) with additional examples for Ech (Thermoanaerobacter tengcongensis and Methanosarcina mazei), Nuo (Escherichia coli K12, Rhodobacter capsulatus, Aeropyrum pernix, and Acidilobus saccharavorans), Fpo (M. mazei) and Eha and Ehb (Methanobacterium thermoautotrophicum). These were aligned as described in Fig. 1. The composite radial Bayesian phylogram constructed from 1600 trees was projected with Dendroscope (Huson et al., 2007). Terminal labels are colored to distinguish Bacteria (red) and Archaea (black), and branches are colored cyan, navy, and green to indicate Mbh-, Mbx-, and Nuo-type complexes, respectively.

Mrp-Mbh-type complexes and the evolution of respiratory chains

The protein modules discussed herein can be thought of analogous to modern day circuit boards that have been assembled together through evolutionary time to form complexes with unique and selectable functions in metabolism. In this regard, the Mrp-Mbh-type complex is simply a circuit board with a H+/Na+ antiporter module (Mrp) and a Group 4 hydrogenase module (Mbh) that can be supplemented with a formate dehydrogenase module (Fdh), a CO dehydrogenase module (Codh) or a NAD(P)H input module (FAD/NADP, see Fig. 4). Through evolutionary time, these modules appear to have been rearranged and assembled into multiple enzyme complexes that yielded unique evolutionary advantages to various organisms by encoding different metabolic functions. Interestingly, this analogy can also be extended to our understanding of the evolution of NADH ubiquinone oxidoreductase (Nuo or Complex I), where the large and small subunits of [NiFe]-hydrogenases show homology with the subunits NuoD and NuoB of Complex I, respectively (Bohm et al., 1990; Friedrich & Scheide, 2000).

Among the [NiFe]-hydrogenases, those that comprise the Group 4 membrane-bound enyzmes (Ech, Mbh, etc.) are clearly related to Complex I (Fig. 1) (Mathiesen & Hagerhall, 2003; Vignais & Billoud, 2007). The energy-converting hydrogenase (Ech) from Methanosarcina sp. contains six subunits (EchA-F) that all show extensive homology to elements from Complex I. It was proposed that these Ech homologs represent the catalytic core of Complex I (Hedderich, 2004; Hedderich & Forzi, 2005). Indeed, Ech might actually represent the ‘core’ complex of Group 4 hydrogenases that contains the catalytic capability for the reversible generation of H2 from reduced ferredoxin linked with the translocation of ions. The EchA subunit connects the Mrp and the hydrogenase modules with Complex I, as it shows homology to MrpAD and NuoLMN, as well as to Mrp-MbhH subunits. This indicates that these genes share a common ancestor, and these proteins contain a conserved ion-translocating unit, as proposed for NuoLMN (Friedrich & Scheide, 2000; Mathiesen & Hagerhall, 2002; Swartz et al., 2005). While MrpA and MrpD are homologs, the larger MrpA protein is composed of an MrpD-like domain as well as an added MrpB domain, suggesting that MrpA evolved from a duplication and fusion of MrpD and MrpB (Swartz et al., 2005). In all of the Mbh-type complexes, homologs of MbhH only contain an MrpD domain, while both MrpD and MrpA domains are present in the Complex I. It has been proposed that an ancestral Mrp-type H+/Na+ antiporter (MrpA, C, B and D) together with a soluble [NiFe] hydrogenase formed an ancestral membrane-bound hydrogenase, which likely contained all three proton-translocating transmembrane subunits (NuoLMN-like) (Mathiesen & Hagerhall, 2003). This ancestral membrane-bound hydrogenase is thought to have been the progenitor to Complex I and various types of membrane-bound hydrogenases. In addition, it was proposed that NuoK shares homology with MrpC and MbhG (Table 1), further strengthening the case that Nuo and Mrp-Mbh complexes have a common evolutionary origin (Mathiesen & Hagerhall, 2003).

Although one cannot retrace how the membrane-bound hydrogenases, complex I, and Mrp-type antiporters evolved, it is clear that these complexes use homologous functional modules to catalyze similar chemical reactions. Considering that the ancestral membrane-bound hydrogenase could have arisen by fusion of an Mrp-type antiporter with an ancestral [NiFe] hydrogenase (e.g. the ancestor of Mbh and Ech), the Mrp-Mbh-type hydrogenase resembles this more closely than does Ech. In this scenario, the other Group 4 hydrogenases, such as Ech, could have evolved from an Mrp-Mbh-type hydrogenase by shedding the Mrp-like subunits to a more minimal complex as illustrated in Fig. 6. This would include loss of an MrpC homolog, which has been retained in Complex I. On the other hand, the more complex Eha and Ehb hydrogenases found in hydrogenotrophic methanogens, which consist of 17–20 subunits, appear to have evolved from an Mrp-Mbh-type complex by acquiring additional subunits or modules to function in the methanogenic electron flow network (Tersteegen & Hedderich, 1999; Thauer et al., 2010). The redox link between CO oxidation and H2 production appears to have evolved as two different but related complexes; a CO dehydrogenase module fused with an Ech-type hydrogenase, as found in C. hydrogenoformans, and the fusion of an Mrp-Mbh-type hydrogenase with a CO dehydrogenase, as found in T. onnurineus (Soboh et al., 2002; Lim et al., 2010; Yun et al., 2011). The evolution of complex I would have involved a more drastic rearrangement, such as modification of the [NiFe]-hydrogenase active site, addition of a quinone-binding site, as well as the acquisition of a NADH input module (NuoEFG) as well as additional subunits (NuoAJ) (Friedrich & Scheide, 2000). The ancestral Mrp-Mbh might have contained both MrpA and MrpD units, which in turn gave rise to NuoLMN of Complex I. In support of such a scenario, sequence comparisons indicates that NuoL is more similar to MrpA, and NuoM is more similar to MrpD, while NuoN could have evolved by gene duplication of NuoM (Mathiesen & Hagerhall, 2003). In the Mbh-type hydrogenases, it is not clear whether MrpA or MrpD is the precursor to MbhH, although MrpD is the more likely candidate.

Figure 6.

Hypothetical scheme of the evolutionary shuffling of Mrp-Mbh-type modules leading to: Mrp-Mbh-type hydrogenases, Mrp-Mbx-type oxidoreductases, Ech (energy-converting) hydrogenases, and Nuo-type (Complex 1, NADH quinone oxidoreductase) complexes. Present day complexes are shown with solid lines. An ancestral Group 4 [NiFe] hydrogenase might have evolved from the fusion of a [NiFe] hydrogenase with an Mrp-type Na+/H+ antiporter. The present day Mrp-Mbh and Mrp-Mbx complexes might be closely related to the ancestral complex and likely evolved to form the various subgroups by acquiring additional modules. This ancestral complex is likely to have contained both MrpA and MrpD as these are homologous to NuoL and NuoM in Complex I. The Ech hydrogenase could have lost several of the Mrp subunits to represent a minimal complex. The ancestral Nuo complex could have arisen from an ancestral Mrp-Mbh by shedding most Mrp subunits, retaining only MrpA, MrpD, and MrpC (indicated by a small light blue bar), loss of the [NiFe] site, and acquisition of a quinone-binding site and of additional subunits (NuoA and NuoJ). This ancestral Nuo complex could then have served to yield the archaeal 11 subunit Nuo complex, and NADH-linked complex I which include the NADH input module NuoEFG as found in present day Bacteria and Eukarya (but not in the archaea). The NADH input module shows an intriguing similarity to [FeFe] hydrogenases, which leads to the hypothesis that an ancestral NADH input module might have evolved by fushion to a [FeFe]-type hydrogenase leading to an ancestral form of a bifurcating [FeFe] hydrogenase. This then served as the precursor for the present day multimeric bifurcating [FeFe] hydrogenases as well as the complex I input module NuoEFG.

Interestingly, based on the homology of the active catalytic subunits, MbhL, MbxL, FpoD and NuoD, the S0-related Mrp-Mbx-type clusters seem to be more closely related to the Nuo (and Fpo-type) clusters than they are to Mrp-Mbh (Fig. 5). In adddition, one of the two conserved cysteine residues comprising each of the L1 and L2 motifs in Mbh is conserved in MbxL and FpoD, but neither cysteine is conserved in NuoD (Fig. 1). This suggests an evolutionary trajectory whereby Cys ligands were sequentially lost from the ancestral Group 4 hydrogenase subunit ultimately leading to the Complex I module in which all conserved Cys residues in MbhL are absent. On the other hand, the NADH input module (NuoEFG) of Complex I is unrelated to the input module found in Mrp-Mbh-NADPH or Mrp-Mbx-NADPH-type complexes. Interestingly, the N-terminal iron–sulfur cluster-binding domain (N1b, N4, and N5) of NuoG shows distinct homology with [FeFe]-hydrogenases (Albracht et al., 1997; Vignais et al., 2001; Sazanov & Hinchliffe, 2006). In addition, two accessory subunits of multimeric [FeFe] hydrogenase, such as the bifurcating [FeFe] hydrogenases of T. maritima, are homologous to NuoE and NuoF of Complex I (Verhagen et al., 1999; Schut & Adams, 2009). Importantly, in contrast to aerobic bacteria, the genomes of aerobic archaea lack recognizable homologs to the NuoEFG module, indicating that their Complex I might not accept electrons from NADH but rather from another electron carrier such as ferredoxin. In this way, the archaeal Complex I might be similar to the Mrp-Mbh complexes and most likely electron flow occurs through the interaction of ferredoxin with the NuoI subunit. This is a homolog of MbhN and is predicted to contain two iron–sulfur clusters (Silva et al., 2000; Sapra et al., 2003; Moparthi & Hagerhall, 2011). In these aerobic archaea, both ferredoxin and NADH are generated in the degradative pathways as well as in the TCA cycle (Zaparty et al., 2009). While the reduced ferredoxin that is generated can be reoxidized by the archaeal-type Complex I, a single subunit NDH type II has been described which can divert electrons from NADH into the quinone pool without any ion translocation (Bandeiras et al., 2003; Kerscher et al., 2008). This collection of fascinating observations suggests that components of Complex I share a common ancestor with elements of both the [NiFe]- and the [FeFe]-hydrogenases. Moreover, bacteria encode homologs of the NuoEFG module as well as [FeFe]-hydrogenase but archaea lack homologs of NuoEFG and [FeFe]-hydrogenase, which strongly suggests that NuoEFG and/or [FeFe]-hydrogenase evolved after the divergence of archaea and bacteria from the last universal common ancestor.

As discussed previously, the evolution of Complex I is likely to have involved the ancestral Group 4 [NiFe]-hydrogenase, as well as an ancestral form of present day [FeFe]-hydrogenases (Fig. 6). However, unlike the N-terminal domain of the catalytic subunit of [FeFe]-hydrogenase (i.e. the F cluster), the C-terminus (i.e. the H cluster) is not conserved in the G module (i.e. NuoG) of Complex I. The simplest interpretation of these observations is that an ancestral NADH input module interacted with an ancestor of the present day [FeFe] hydrogenase yielding both bifurcating [FeFe] hydrogenases and the NADH input module NuoEFG. Taken together, all of these data suggest that the NADH input module represented by NuoEFG (and domains therein) and the Mrp-Mbh-type complexes represent prime and ubiquitous examples of how electron flow pathways evolve to connect a variety of redox-driven modules. In addition, the incorporation of Mrp and [NiFe] hydrogenase modules in the form of an ancestral Mrp-Mbh complex can be seen as an evolutionary mechanism to link electron flow to energy conservation in the form of an ion gradient. The distinct electron flow pathways from a variety of electron donors through conserved iron–sulfur clusters to catalytic sites in bifurcating [FeFe] hydrogenases and Group 4 [NiFe] hydrogenases illustrates an amazing evolutionary path of electron flow mechanisms that ultimately coalesce in present day Complex I.


This research was supported by grants from the Chemical Sciences, Geosciences and Biosciences Division (FG05-95ER20175) and the Office of Biological and Environmental Research (FG02-08ER64690) of the Office of Basic Energy Sciences, Office of Science, U.S. Department of Energy, and by a grant from the Low Temperature Geobiology and Geochemsitry program of the National Science Foundation (EAR-1123689). The Astrobiology Biogeocatalysis Research Center was supported by a grant from the NASA Astrobiology Institute (NNA08C-N85A). We thank Sanjeev Chandrayan, Gina Lipscomb, Michael Thorgersen, Patrick McTernan, and Chris Hopkins for invaluable discussions.