Energy Conservation via Hydrogen Cycling in the Methanogenic Archaeon Methanosarcina barkeri

Three distinct types of hydrogenases are encoded by Methanosarcina barkeri Fusaro (see Fig. S1 in the supplemental material) (19). The F420-reducing hydrogenase (Frh) is a cytoplasmic, three-subunit (α, β, and γ) enzyme encoded by the frhADGB operon, which also includes a maturation protease, FrhD (20). This enzyme couples the oxidation/reduction of the deazaflavin cofactor F420 with production/consumption of H₂. The membrane-bound Vht hydrogenase utilizes the quinone-like electron carrier, methanophenazine, as a cofactor (21). Like Frh, Vht is a three-subunit enzyme encoded by a four-gene operon (vhtGACD) that includes a maturation protease, VhtD (19). M. barkeri also contains genes that encode homologs of both the frh and vht operons (the freAEGB and vhxGAC operons, respectively); however, multiple lines of evidence suggest that these genes are incapable of producing active hydrogenases (16, 22). Thus, the presence of these genes has no bearing on the results presented herein. The final hydrogenase encoded by M. barkeri is a membrane-bound, energy-converting hydrogenase (Ech), which couples the oxidation/reduction of ferredoxin and H₂ to the production/consumption of a proton motive force (23, 24). Thus, the enzyme can use proton motive force to drive the endergonic reduction of ferredoxin by H₂, which is required for CO₂ reduction during hydrogenotrophic methanogenesis and for biosynthesis during growth by H₂-dependent reduction of C₁ compounds (methyl-reducing methanogenesis). During both methylotrophic and aceticlastic methanogenesis, Ech is believed to couple oxidation of reduced ferredoxin to production of proton motive force and H₂. The hydrogen thus produced would need to be recaptured by Vht in a putative H₂ cycling process that contributes to proton motive force (Fig. 1) (17).

A number of studies have shown that assorted Methanosarcina strains produce H₂ during growth on methylotrophic and aceticlastic substrates (9, 25–30); however, to our knowledge, this has never been assessed in M. barkeri strain Fusaro. To test this, we quantified the accumulation of CH₄ and H₂ during growth on methanol-containing medium (Fig. 2). Consistent with the hydrogen cycling hypothesis, we observed significant H₂ production, which reached a maximum partial pressure of ca. 20 Pa near the end of exponential growth. As expected, the culture also produced substantial levels of methane. As previously observed (16), a mutant lacking Frh (strain WWM115 [Table S1]) grew at a lower rate than its isogenic parent and produced somewhat smaller amounts of methane. Very little H₂ (<4 Pa) was produced during growth of the Δfrh mutant; however, after growth ceased, the H₂ concentration slowly rose, reaching a maximum level of 7 Pa. Thus, Frh is responsible for most hydrogen production during growth of M. barkeri Fusaro on methanol, although some hydrogen is still produced in the Δfrh mutant. As will be shown below, Ech is probably responsible for the low levels of H₂ seen in the Δfrh mutant.

To investigate the role of Vht during growth of M. barkeri, we attempted to delete the vhtGACD operon via homologous gene replacement (31, 32). However, despite numerous attempts, including selection on a variety of media, with and without supplementation of potential biosynthetic intermediates, no mutant colonies were obtained. We also attempted to delete the vht operon using the markerless deletion method of genetic exchange (33). This method relies on construction of a merodiploid strain with both mutant and wild-type alleles. Upon segregation of the merodiploid, 50% of the recombinants are expected to be mutants if there is no selective pressure against the mutant allele. However, if the mutation causes a reduction in growth rate (with lethality being the most extreme case), the probability of obtaining recombinants with the mutant allele is severely reduced. We tested 101 haploid recombinants obtained from a vhtGACD⁺/ΔvhtGACD merodiploid; all carried the wild-type vht allele. Taken together, these data suggest that the vhtGACD operon is critical for normal growth of M. barkeri.

To test whether Vht is essential, we constructed a mutant in which the vht operon was placed under control of a tightly regulated, tetracycline-dependent promoter (34). We then examined the viability of the mutant and its isogenic parent by spotting serial dilutions on a variety of media, with and without tetracycline. As shown in Fig. 3, the P_tet::vht mutant is unable to grow in the absence of the inducer but grew well when tetracycline was added, whereas the isogenic parent grew with or without the addition of tetracycline. These phenotypes were observed on a variety of media, including media containing (i) methanol, (ii) methanol plus H₂, (iii) H₂/CO₂, and (iv) acetate, which were chosen because they encompass growth conditions that require each of the four known methanogenic pathways used by M. barkeri (Fig. 4). It should be stressed that the P_tet::vht mutant used in this experiment was pregrown in the presence of inducer. Thus, at the start of the experiment, all cells have active Vht. However, during cultivation in the absence of tetracycline, preexisting Vht is depleted by protein turnover and cell division, thereby allowing characterization of the Vht-deficient phenotype. The absence of growth of the diluted cultures in all media shows that Vht is essential for growth via the methylotrophic (methanol), methyl-reducing (methanol plus H₂), hydrogenotrophic (H₂/CO₂), and aceticlastic (acetate) methanogenic pathways.

To help understand why Vht is essential, we quantified production of H₂ and CH₄ in cultures of the P_tet::vht strain with and without tetracycline (Fig. 2). When the strains were grown in methanol-containing medium in the presence of tetracycline, the accumulation of H₂ and CH₄ was essentially identical to that of the isogenic parent. Cultures in which vht is not expressed (i.e., without tetracycline) grew initially but growth rapidly slowed and reached an optical density that was less than half of that obtained when vht was expressed. The optical density subsequently dropped, suggesting cell death and lysis. Similarly, methane accumulation in cultures not expressing vht was much slower than in induced cultures and only reached half of that seen under inducing conditions. In contrast, H₂ accumulation was much higher in the absence of Vht, with final levels nearly sixfold higher than those seen in cultures that express Vht. These data clearly show that Vht is required for efficient recapture of H₂ produced by Frh and Ech. Moreover, they suggest that H₂ loss is responsible for the lethal consequences of vht repression.

If the inability to recapture H₂ is responsible for the essentiality of Vht, then it should be possible to delete the vht operon in strains that do not produce hydrogen. As described above, Frh is responsible for the majority of H₂ production during growth. Thus, we attempted to introduce a Δvht allele into the Δfrh host. In contrast to our prior unsuccessful attempts to create a Δvht single mutant, the Δvht Δfrh double mutant was isolated in the first attempt. Therefore, Vht is not required when Frh is absent. Like the Δfrh single mutant, the Δvht Δfrh double mutant grows slowly on methanol and produces lower levels of methane (Fig. 2). Significantly, the double mutant does not produce the excessive level of H₂ seen in the uninduced P_tet::vht strain, instead accumulating H₂ at levels similar to those of the parental strain (ca. 20 Pa). Because Ech is the only active hydrogenase remaining in the Δvht Δfrh double mutant, it must be responsible for H₂ production in this strain. This begs the question of why H₂ accumulation stops at 20 Pa in the double mutant, while the uninduced P_tet::vht strain produces much higher levels. We suggest that the coupling of Ech activity to generation of proton motive force thermodynamically restrains excessive H₂ production, even in the absence of H₂ uptake by Vht. This would also explain the viability of the Δvht Δfrh double mutant. This situation is in stark contrast to that seen in the vht-depleted strain, where the F420-dependent Frh is responsible for most of the H₂ production (see above). Accordingly, at the low H₂ partial pressures observed in our experiments, reduction of protons with F420 is strongly exergonic, allowing excessive hydrogen accumulation. This is also consistent with the observation that the redox state of F420 is in rapid equilibrium with H₂ (35). Interestingly, the smaller amount of H₂ accumulation in the Δfrh mutant relative to that seen in the Δvht Δfrh double mutant shows that Vht also consumes H₂ produced by Ech. This supports previous studies indicating potential energy conservation via Ech/Vht H₂ cycling during acetate metabolism (17, 23).

We previously showed that M. barkeri has a branched electron transport chain, with Frh- and F420 dehydrogenase (Fpo)-dependent branches (16). The data reported here extend our understanding of the Frh-dependent branch and are fully consistent with the model depicted in Fig. 1. Thus, during growth on methylotrophic substrates such as methanol, reduced F420 is preferentially oxidized via an energy-conserving, H₂ cycling electron transport chain that requires Frh. However, in the absence of Frh, reduced F420 is channeled into the Fpo-dependent electron transport chain, which supports growth at a significantly lower rate (Fig. 1 and 2). This alternate pathway accounts for the viability of the Δfrh mutant, which is lost when both frh and fpo are deleted (16). Similar but less severe phenotypes have been observed in fpo and frh mutants of Methanosarcina mazei, thus it seems likely that H₂ cycling also occurs in this closely related species (36). However, many Methanosarcina species, especially those that inhabit marine environments, are devoid of hydrogenase activity, despite the presence of hydrogenase-encoding genes. We, and others, have interpreted this to be an adaptation to the marine environment, where H₂-utilizing sulfate reducers are likely to disrupt H₂ cycling due to the superior thermodynamics of H₂ oxidation coupled to sulfate reduction (19, 37).

A similar branched electron transport chain may also explain the contradictory evidence regarding H₂ cycling in Desulfovibrio species. Thus, the viability of Desulfovibrio hydrogenase mutants and the inability of excess H₂ to suppress substrate catabolism can both be explained by the presence of alternative electron transport mechanisms. Indeed, metabolic modeling of Desulfovibrio vulgaris strongly supports this interpretation (13). Thus, it is critical that experiments designed to test the H₂ cycling mechanism be interpreted within a framework that includes the possibility of branched electron transport chains. With this in mind, it seems likely that many anaerobic organisms might use H₂ cycling for energy conservation. Indeed, since it was originally proposed, H₂ cycling has been suggested to occur in the acetogen Acetobacterium woodii (10) and in the Fe(III) respiring Geobacter sulfurreducens (8).

Although the data presented here strongly support the H₂ cycling model, they raise additional questions regarding H₂-dependent methanogenesis that are not easily explained. In particular, it is not readily apparent why the uninduced P_tet::vht mutants are inviable during hydrogenotrophic or methyl-reducing growth. As shown in Fig. 4, it should be possible to channel electrons from H₂ oxidation into the electron transport chain via Frh and Fpo. Indeed, Thauer et al. have proposed that this alternate pathway is functional in Methanosarcina (38). Nevertheless, the P_tet::vht mutant does not grow under repressing conditions on either H₂/CO₂ or methanol plus H₂. It should be stressed that we use high concentrations of hydrogen during growth on these substrates. Thus, it is expected that reduction of F420 via Frh should be exergonic in our experiments, which would favor this pathway. (This is in contrast to the methylotrophic or aceticlastic growth conditions described above, under which the reverse reaction [i.e., hydrogen production] is favored.) Thus, a thermodynamic argument cannot easily explain the results. Further, based on available evidence (16, 39, 40), energy conservation via the Vht-dependent pathway should be identical to that of the alternate Frh/Fpo-dependent pathway. Thus, an energy conservation argument also cannot explain the phenomenon. One might argue that faster kinetics of the Vht-dependent pathway could be responsible, but in our opinion, the growth (albeit slower than wild type) of the Δfrh and Δvht Δfrh mutants during methylotrophic growth, which depends on Fpo, argues against this explanation. Therefore, as yet unknown regulatory and/or biochemical constraints on hydrogen metabolism in Methanosarcina await discovery.

The construction and genotypes of all Methanosarcina strains are presented in Table S1 in the supplemental material. Methanosarcina strains were grown as single cells (41) at 37°C in high-salt (HS) broth medium (42) or on agar-solidified medium as described previously (43). Growth substrates provided were methanol (125 mM in broth medium and 50 mM in agar-solidified medium) or sodium acetate (120 mM) under a headspace of either N₂/CO₂ (80/20%) at 50 kPa over ambient pressure or H₂/CO₂ (80/20%) at 300 kPa over ambient pressure. Cultures were supplemented as indicated with 0.1% yeast extract, 0.1% Casamino Acids, 10 mM sodium acetate, or 10 mM pyruvate. Puromycin (CalBioChem, San Diego, CA) was added at 2 µg/ml for selection of the puromycin transacetylase (pac) gene (33). 8-Aza-2,6-diaminopurine (8-ADP) (Sigma, St. Louis, MO) was added at 20 µg/ml for selection against the presence of hpt (33). Tetracycline was added at 100 µg/ml to induce the tetracycline-regulated PmcrB(tetO3) promoter (34). Standard conditions were used for growth of Escherichia coli strains (44) DH5α/λ-pir (45) and DH10B (Stratagene, La Jolla, CA), which were used as hosts for plasmid constructions.

Standard methods were used for plasmid DNA isolation and manipulation using E. coli hosts (46). Liposome-mediated transformation was used for Methanosarcina as described previously (47). Genomic DNA isolation and DNA hybridization were performed as described previously (32, 42, 43). DNA sequences were determined from double-stranded templates by the W. M. Keck Center for Comparative and Functional Genomics, University of Illinois. Plasmid constructions are described in the supporting information (Tables S2 and S3).

The markerless genetic exchange method (33) using plasmid pGK4 was employed to delete frhADGB (Δfrh) in the Δhpt background of M. barkeri Fusaro (Tables S1, S2, and S3) using methanol/H₂/CO₂ as the growth substrate. The Δvht Δfrh mutant was constructed by deleting vhtGACD in the Δfrh markerless mutant by the homologous recombination-mediated gene replacement method (32). To do this, the 5.6-kb XhoI/NotI fragment of pGK82B was used to transform the Δfrh mutant to puromycin resistance on methanol-containing medium. The mutants were confirmed by PCR and DNA hybridization (data not shown).

The tetracycline-regulated PmcrB(tetO3) promoter was employed to drive conditional expression of the vht operon in M. barkeri WWM157 (34). This strain was constructed by transforming strain WWM154 to puromycin resistance using the 7-kb NcoI/SpeI fragment of pGK61A (Tables S1, S2, and S3). The transformants were selected on methanol plus H₂/CO₂ medium in the presence of puromycin and tetracycline. The P_tet::vht strain was confirmed by DNA hybridization (data not shown). To ensure that the native vht promoter (Pvht) did not interfere with expression from PmcrB(tetO3), 382 bp upstream of vhtG were deleted in P_tet::vht. This left 1,038 bp intact for the expression of the hyp operon, which is upstream of the vht operon and expressed in the opposite direction.

Growth of strains WWM157 (P_tet::vht) and WWM154 (isogenic parent) on methanol, methanol/H₂/CO₂, H₂/CO₂, and acetate were analyzed by the spot-plate method (48). Cultures were first adapted for at least 15 generations to the substrate of interest; tetracycline was added to each medium for growth of strain WWM157. Upon reaching stationary phase, 10 ml of culture was washed three times and resuspended in 5 ml HS medium that lacked growth substrate. Subsequently, 10 µl of 10-fold serial dilutions was spotted onto the following: three layers of GB004 paper (Whatman, NJ), two layers of GB002 paper (Schleicher & Schuell BioScience, NH), one layer of 3 MM paper (Whatman, NJ), and a 0.22 mM nylon membrane (GE Water and Process Technologies, PA) soaked in 43 ml of HS medium containing the substrate of interest with and without tetracycline. The plates were sealed and incubated at 37°C for at least 2 weeks in an intrachamber anoxic incubator (49). Growth on acetate and methanol was tested under an atmosphere of N₂/CO₂/H₂S (80/19.9/0.1 ratio), while growth on methanol/H₂/CO₂ or H₂/CO₂ was tested under an atmosphere of H₂/CO₂/H₂S (80/19.9/0.1 ratio).

M. barkeri WWM85 (isogenic parent), WWM157 (P_tet::vht; grown in the presence of tetracycline), WWM115 (Δfrh), and WWM351 (Δvht Δfrh) were grown on methanol until mid-exponential phase (optical density at 600 nm [OD₆₀₀] of ca. 0.5) and then 1 ml (WWM85 and WWM157) or 5 ml (WWM115 and WWM351) was inoculated into 100 ml HS-methanol in a 500-ml serum bottle. For WWM157, the culture was washed once prior to inoculation with or without tetracycline. To measure H₂ and CH₄, ca. 1-ml or 2-ml headspace sample was withdrawn aseptically from the culture at various time points with a syringe that had been flushed with sterile, anaerobic N₂. The gas sample was then diluted into 70 ml helium. A gas-tight syringe flushed with helium was subsequently used to withdraw 3 ml of the diluted sample, which was then injected into an SRI gas chromatograph, equipped with a reduction gas detector (RGD) and a thermal conductivity detector (TCD) at 52°C. The RGD column was a three-foot-long 13× molecule sieve, whereas the TCD column was a six-foot HayeSep D porous polymer column. The RGD column was used to detect H₂ by peak height, and the TCD column was used to detect CH₄ by peak area. Helium was used as the carrier gas. OD₆₀₀ was also measured during the growth curve.