ABSTRACT
The methanogenic archaeon Methanosarcina barkeri encodes three distinct types of hydrogenase, whose functions vary depending on the growth substrate. These include the F420-dependent (Frh), methanophenazine-dependent (Vht), and ferredoxin-dependent (Ech) hydrogenases. To investigate their physiological roles, we characterized a series of mutants lacking each hydrogenase in various combinations. Mutants lacking Frh, Vht, or Ech in any combination failed to grow on H2/CO2, whereas only Vht and Ech were essential for growth on acetate. In contrast, a mutant lacking all three grew on methanol with a final growth yield similar to wild-type, produced methane and CO2 in the expected 3:1 ratio, but had a ca. 33% slower growth rate. Thus, hydrogenases play a significant, but non-essential, role during growth on this substrate. As previously observed, mutants lacking Ech fail to grow on methanol/H2 unless supplemented with biosynthetic precursors. Interestingly, this phenotype was abolished in the Δech/Δfrh and Δech/Δfrh/Δvht mutants, consistent with the idea that hydrogenases inhibit methanol oxidation in the presence of H2, which prevents production of reducing equivalents needed for biosynthesis. Quantification of methane and CO2 produced from methanol by resting cell suspensions of various mutants supports this conclusion. Based on global transcriptional profiles, none of the hydrogenases are upregulated to compensate for loss of the others. However, transcript levels of the F420 dehydrogenase operon were significantly higher in all strains lacking frh, suggesting a mechanism to sense the redox state of F420. The roles of the hydrogenases in energy conservation during growth with each methanogenic pathway are discussed.
IMPORTANCE
Methanogenic archaea are key players in the global carbon cycle due to their ability to facilitate the remineralization of organic substrates in many anaerobic environments. The consequences of biological methanogenesis are far reaching, with impacts on atmospheric methane and CO2 concentrations, agriculture, energy production, waste treatment and human health. The data presented here clarify the in vivo function of hydrogenases during methanogenesis, which in turn deepens our understanding of this unique form of metabolism. This knowledge is critical for a variety of important issues ranging from atmospheric composition to human health.
INTRODUCTION
The ability to metabolize molecular hydrogen (H2) is a key metabolic feature in methanogenic Archaea (1). This trait is conferred by a class of enzymes known as hydrogenases, which catalyze the reversible oxidation of H2 coupled to various electron donors/acceptors (2, 3). At least five distinct types of hydrogenases are found in methanogenic archaea. These enzymes differ with respect to their redox partners, their cellular localization, and whether their activity is linked to production or consumption of membrane potential (4). Biochemical characterization of these diverse hydrogenases led to proposed functions for each enzyme class that differ substantially between methanogens with and without cytochromes (5).
Methanogens without cytochromes produce at least four types of hydrogenase; including (i) the electron-bifurcating Mvh hydrogenase, which couples oxidation of hydrogen to reduction of ferredoxin and a mixed coenzyme M-coenzyme B disulfide, (ii) the coenzyme F420-dependent hydrogenase, (iii) the [Fe] hydrogenase, which couples hydrogen oxidation to reduction of methenyltetrahydromethanopterin, and (iv) the ferredoxin-dependent, energy-converting hydrogenases (4). The first three are cytoplasmic enzymes, which supply the electrons needed to reduce CO2 to methane. The last is a membrane bound multi-subunit complex that couples hydrogenase activity to the production/consumption of the ion-motive force across the cell membrane (hence the designation as “energy-converting”). In non-cytochrome-containing methanogens, these energy-converting hydrogenases are believed to provide low-potential electrons, in the form of reduced ferredoxin, needed for anaplerotic reactions (6).
Methanogens with cytochromes, typified by Methanosarcina barkeri, encode a different set of hydrogenases that includes one cytoplasmic and two membrane-bound enzymes (Fig 1) (7). Like the non-cytochrome containing methanogens, M. barkeri produces a cytoplasmic, three-subunit F420-dependent hydrogenase known as Frh (for F420-reducing hydrogenase). Frh is encoded by the four-gene frhADGB operon, which encodes the α, β and γ subunits (FrhA, FrhB and FrhG, respectively), along with the maturation protease FrhD (8). A second locus, freAEGB, encodes proteins that are 86-88% identical to FrhA, FrhB and FrhG, but lacks the gene for the maturation protease FrhD, instead encoding a small protein of unknown function (FrhE). It is not known whether the fre operon is capable of producing an active hydrogenase (8-11). A membrane-bound hydrogenase linked to the quinone-like electron carrier methanophenazine has, to date, been found only in Methanosarcina species. This enzyme, known as Vht because it was initially identified as a viologen-reducing hydrogenase, is encoded by the vhtGACD operon, which encodes the biochemically characterized enzyme comprised of VhtA and VhtG, along with a putative membrane-bound cytochrome, VhtC, that does not co-purify with the active enzyme, and a maturation protease, VhtD (7). As with the F420-reducing hydrogenase, a second locus that lacks the maturation protease is encoded in M. barkeri strains. This operon (vhxGAC) encodes proteins that display ca. 50% amino acid identity with those encoded by vhtGACD. Like freAEGB, it is not known whether the vhx operon produces an active hydrogenase (7). Finally, M. barkeri encodes a membrane-bound, ferredoxin-dependent energy-converting hydrogenase (Ech) (12). This five-subunit enzyme complex is much simpler than, and only distantly related to, the energy-converting hydrogenases of the non-cytochrome methanogens, which typically contain more than a dozen subunits (3). Homologs of the electron bifurcating- and methenyltetrahydromethanopterin-reducing hydrogenases are not known to occur in cytochrome-containing methanogens.
A key difference between the cytochrome-containing and non-cytochrome-containing methanogens is the ability of the former to use one-carbon (C-1) compounds and acetic acid, in addition to H2/CO2, as growth substrates. Catabolism of these chemically diverse substrates involves four distinct methanogenic pathways: the CO2 reduction pathway, the methyl reduction pathway, the methylotrophic pathway and the aceticlastic pathway (13, 14). While several of these pathways share common steps, they differ substantially with respect to involvement of key enzymes and the direction of metabolic flux during methane production (Fig 2). Surprisingly, it now appears that some Methanosarcina species (e.g. M. barkeri) use hydrogenases in each of the four pathways, regardless of whether external H2 is provided as a growth substrate (15).
During the CO2 reduction pathway, wherein CO2 is reduced to CH4 in a stepwise manner, hydrogenases produce the electron-donating cofactors required for four distinct reduction steps (Fig 2) (5). Initial reduction of CO2 to formyl-methanofuran requires reduced Fd (Fdred), which is produced by Ech. This reaction is dependent on ion-motive force because reduction of Fd by H2 is endergonic under physiological conditions (16). Subsequent reduction of methenyl-tetrahydrosarcinapterin (H4SPT) to methylene-H4SPT, and of methylene-H4SPT to methyl-H4SPT, requires reduced coenzyme F420 (F420red), which is supplied by Frh. Finally, reduction of a methyl group to methane using coenzyme B produces a heterodisulfide of coenzyme M and coenzyme B (CoM-S-S-CoB), which must be reduced to produce the free CoM and CoB needed for continued methanogenesis. This reaction is catalyzed by a membrane bound heterodisulfide reductase (HdrED), which uses the reduced form of a membrane-bound cofactor, methanophenazine, as the source of electrons (17). H2, in turn, serves as the reductant for generation of reduced methanophenzaine via membrane-bound Vht hydrogenase. Thus, all three types of hydrogenase are predicted to be required for growth via CO2 reduction, a conclusion that has been validated by the phenotypic analysis of single mutants (11, 15, 16).
In contrast, methanogenesis via the methyl reduction pathway is expected to require only Vht. In this pathway, methyl-CoM derived from C-1 compounds, such as methanol or methylamines, is directly reduced to methane using CoB as the electron donor. As with the CO2 reduction pathway, this produces a CoM-S-S-CoB disulfide that must be regenerated in a pathway requiring HdrDE and reduced methanophenazine, which is presumably generated by Vht (Fig 2). This idea is supported by the analysis of conditional vht mutants (15). Neither Frh nor Ech is required for methanogenesis in this model, a finding that is consistent with experimental data from Δfrh and Δech mutants (11, 16). Nevertheless, the M. barkeri Δech strain cannot grow via the methyl reduction pathway unless the media are supplemented with acetate and/or pyruvate. Thus, Ech plays an essential biosynthetic role under these conditions, which is probably the H2-dependent synthesis of reduced ferredoxin needed for synthesis of acetyl-CoA and pyruvate (16).
During aceticlastic methanogenesis, both the Ech and Vht hydrogenases play a critical role in methanogenesis. In this pathway, acetyl-CoA is split into methyl-H4SPT and enzyme-bound [CO] by the acetyl-CoA decarbonylase/synthase (ACDS) enzyme complex. CO is then further oxidized to CO2 with concomitant reduction of ferredoxin (12, 16). It is believed that the exergonic oxidation of ferredoxin by Ech produces H2 and contributes to the proton motive force by transferring protons across the membrane. The proton motive force is further enhanced by a putative H2 cycling mechanism, in which the H2 produced by Ech diffuses across the membrane, where it is oxidized by Vht to produce reduced methanophenazine. This unusual electron transport chain is completed when the reduced methanophenazine produced by Vht is used by HdrDE to regenerate free CoM and CoB from the CoM-S-S-CoB heterodisulfide (Fig 2) (18). Participation of these hydrogenases in the aceticlastic pathway is supported by mutagenic studies showing that ech and vht mutants do not grow with acetate as the sole substrate, regardless of whether biosynthetic precursors were supplied (15, 16).
Finally, all three types of hydrogenases are thought to be involved in methylotrophic methanogenesis via a H2 cycling mechanism similar to that described for aceticlastic growth (15). In this pathway, F420red and Fdred, produced by the stepwise oxidation of methyl groups to CO2, are converted to molecular H2 by Frh and Ech, respectively (Fig 2). H2 then diffuses to the outer surface of the cell membrane where it is oxidized by Vht, releasing protons on the outside of the cell and contributing to the generation of an ion-motive force (Fig 3) (15). Nevertheless, M. barkeri Δfrh and Δech strains are capable of methylotrophic growth, indicating the presence of alternative pathways for transfer of electrons from F420red and Fdred to the electron transport chain (11, 16). The membrane-bound F420 dehydrogenase complex (Fpo) has been identified as the alternate mechanism of electron transfer from F420red (11). This enzyme couples the exergonic reduction of methanophenazine by F420red, with the generation of proton motive force in a H2 independent manner. However, the M. barkeri Δfrh strain exhibits slower growth rates than wild type M. barkeri, showing that the H2-independent electron transport chain is less effective than electron transport via H2 cycling. The observation that Δfpo/Δfrh double mutants are incapable of methylotrophic growth shows that additional electron transport routes are either not present, or not sufficient for methylotrophic growth (11). Mutants lacking Vht are inviable under all growth conditions, including methylotrophic growth, unless Frh is also removed. This phenotype is probably due the inability to recapture H2 produced in the cytoplasm, which causes redox imbalance and cell lysis (15).
Although the three M. barkeri hydrogenases have been studied in vitro and in certain mutants, a complete analysis of their role during growth on various substrates has yet to be reported. In this study, all five hydrogenase operons were systematically deleted in all viable combinations, and the physiological ramifications of these mutations examined by measuring growth, methanogenesis and hydrogenase activity on various growth substrates. We also performed global transcriptional profiling to assess the possibility that alternate electron transport chain components might be upregulated to compensate for the loss of specific hydrogenases. The data suggest that hydrogenases are not required for methylotrophic methanogenesis, but are essential for CO2 reductive, methyl reductive, and acetoclastic methanogenesis. Additionally, an inhibitory effect of H2 on the methyl oxidative pathway appears to be mediated by all three hydrogenases.
RESULTS
Construction of hydrogenase deletion mutants
To assess the role of the M. barkeri hydrogenases during growth on various substrates, we constructed mutants lacking the frhADGB, freAEGB, vhtGACD, vhxGAC and echABCDEF operons, individually and in all possible combinations (Fig S1). Because mutants lacking vht are only viable in Δfrh mutants (15), we also created a series of conditional mutants that have the vht promoter replaced by the synthetic Ptet promoter, which is expressed in the presence of tetracycline and tightly repressed in its absence (19). To simplify the isolation of strains lacking the adjacent vht and vhx loci, we constructed a mutant allele (denoted Δvht-vhx) that deleted both operons along with two intervening genes that encode a putative peptidoglycan binding protein (Mbar_A1842) and an uncharacterized hypothetical protein (Mbar_A1843). The full set of strains containing deletions for hydrogenase operons in all possible combinations were successfully generated and verified by either Southern hybridization or PCR (Figures S2-S5).
Characterization of growth phenotypes in hydrogenase deletion mutants
The generation time, growth yield and duration of lag phase for each mutant was determined by monitoring optical density during growth in a variety of media, providing clues as to the function of each hydrogenase during utilization of various carbon and energy sources (Table 1). With the exception of the Δfre and Δvhx mutations, which had no discernable phenotypes alone or in combination with other gene deletions, each of the mutations caused significant growth defects in one or more media.
Strains containing the Δech mutation were unable to grow in either H2/CO2 or acetate media, regardless of whether the other hydrogenase genes were deleted. However, with the exception of the Ptetvht/Δech mutant discussed below, these mutants all grew on methanol medium, albeit with reduced growth rates. Consistent with previous reports (16), the Δech single mutant was unable to grow on unsupplemented methanol/H2/CO2 medium. Interestingly, the Δech/Δfrh double mutant regained the ability to grow in this medium, but with diminished rate and yield and the longest lag phase observed in any of our experiments. These phenotypes were substantially minimized in the Δech/Δfrh/Δvht triple mutant, suggesting that both Frh and Vht inhibit methanol oxidation, which is needed to provide reducing equivalents for biosynthesis, when H2 is present.
As previously reported, we were unable to obtain a mutant lacking only vht, suggesting that loss of this locus is lethal in otherwise wild-type strains (15). This conclusion was supported by the phenotype of the Ptetvht and Ptetvht/Δech mutants, which were incapable of growth on any medium when tetracycline is absent (i.e. under repressing conditions). However, as previously noted, when cells were grown on methanol it was possible to delete the vht operon if frh was deleted first. The Δfrh/Δvht strains, including ones that also carried an ech deletion, had methanol growth phenotypes similar to that of the wild type. Thus, hydrogenases are not required for growth on methanol, although vht-deficient strains are inviable in the presence of an active Frh hydrogenase. In contrast, strains lacking vht alone or in combination with other mutations were unable to grow on either H2/CO2 or acetate media. A more graded response was observed when various vht mutants were grown on methanol/H2/CO2. Accordingly, on this substrate combination, the vht single mutant was inviable, while the Δfrh/Δvht double mutant grew very poorly and the Δech/Δfrh/Δvht strain had phenotypes that were nearly wild-type. Again, these data are consistent with the idea that with certain mutant backgrounds Frh, Vht and Ech inhibit methanol oxidation in the presence of H2.
Finally, mutants lacking only the frh operon grew on three out of four substrates tested, failing to grow only on H2/CO2 medium. When methanol was the sole substrate, the Δfrh mutant had an extended lag phase and a generation time approximately double that of the parental strain. However, during growth on either acetate or methanol/H2/CO2 the growth phenotypes of this strain were equivalent to the parental strain, suggesting Frh enhances growth on methanol, but is not required for growth on the two latter substrate combinations.
Methane and CO2 production by hydrogenase deletion mutants
To probe the underlying mechanisms behind the growth phenotypes, we also examined production of methane and CO2 by resting cell suspensions incubated with various substrates (Tables 2 & 3). The Ptetvht and Ptetvht/Δech mutants were not examined because they do not grow in any medium under non-inducing conditions. Similarly, we did not assay production of methane from acetate, because prior growth on acetate is required to induce the enzymes needed for this activity and most of the hydrogenase mutants are unable to grow under these conditions (20, 21).
Consistent with their lack of growth phenotypes, the Δfre and Δvhx single mutations did not affect the levels of methane produced from any substrate tested. Neither did these mutations affect the ratio of methane/CO2 produced from methanol or from methanol/H2. However, the Δvhx mutation slowed the rate of methane production. The Δfre mutation also slowed the rate of methane production, but only when combined with the Δfrh mutation.
As seen in previous studies, Δech mutants produced only minor amounts of methane from H2/CO2 (<2% relative to the parental strain), but produced wild-type levels from both methanol and methanol/H2. During incubations with methanol, methane and CO2 were produced in a 3:1 ratio, consistent with disproportionation of the substrate via the methylotrophic pathway (Fig 2). Cell suspensions incubated with methanol and H2 produced only methane, showing that addition of hydrogen inhibits methanol oxidation. The rate of methane production by the Δech mutant was somewhat slower than wild-type using both methanol and methanol/H2. This rate was further reduced when the ech deletion was combined with mutations removing the frh or vht operons. Accordingly, the Δech/Δfrh double mutant produced methane nearly 5 times slower than the wild-type strain. Interestingly, this mutant produced a small amount of CO2 in addition to the wild-type level of methane, indicating a small amount of methanol was oxidized. When the ech, frh, and vht hydrogenases were deleted together, the quantity and stoichiometry of methane and CO2 production was identical to that observed on methanol alone.
The Δfrh single mutant produced similar levels and ratios of methane and CO2 relative to the parental strain with methanol or methanol/H2. When H2/CO2 was the substrate, methane production was reduced ca. 10-fold, but, significantly, not abolished. Combining the Δfrh mutation with deletions of vht and ech reduced methane production from H2/CO2 to negligible levels. In contrast, minimal affects on methane and CO2 production or stoichiometry were observed when methanol was the sole substrate. However, when combined with deletions of vht or ech, the Δfrh mutants produced significant levels of CO2 when incubated with methanol/H2, and the triple Δech/Δfrh/Δvht mutant produced similar levels to that seen in assays incubated with methanol alone. Rates of methane production were substantially slower than wild-type for all Δfrh mutants.
Enzyme activity in hydrogenase mutants
The hydrogenase activity for selected deletion mutants was measured in the forward direction (H2 oxidation) to allow estimation of the contributions of each enzyme to overall activity (Table 4).
The hydrogenase activity of mutants lacking Fre or Vhx was not statistically different from the parental strain. Moreover, hydrogenase activity of the Δech/Δfrh/Δvht mutant, which still encodes Fre and Vhx, was not statistically different than the Δech/Δfrh/Δfre/Δvht-vhx mutant that lacks all five hydrogenase operons. Thus, the freAEGB and vhxGAC operons do not, by themselves, produce detectable levels of hydrogenase. Because Fre and Vhx are essentially inactive, the hydrogenase levels in the Δfrh/Δvht and Δech/Δfrh strains can be attributed solely to Ech and Vht, respectively. Accordingly, Ech has the lowest activity of the three hydrogenases, accounting for ca. 4% of total activity, and with Vht activity being ca. 6-fold higher. Consistent with this conclusion, deletion of ech did not significantly affect hydrogenase activity, whereas the Δfrh strain had drastically diminished activity as compared to the parental strain. Additionally, activity from the Δfrh strain, which encodes both Vht and Ech, is roughly equivalent to the combined activities of strains encoding only Vht or only Ech. Because strains expressing only Frh hydrogenase are inviable, the activity of this hydrogenase cannot be directly determined from a mutant strain. However, the relative contribution of Frh can be estimated from the hydrogenase activities of other mutants. Thus, by subtracting the activities of Vht and Ech from that of the parental strain, we estimate that roughly 75% of hydrogenase activity can be attributed to Frh.
Effect of hydrogenase deletions on mRNA abundance
Our estimate of the relative activities of the individual hydrogenases assumes that the expression levels for each hydrogenase are unaffected by deletion of the others. To explicitly examine this possibility, we determined the global mRNA abundance profiles for each mutant using RNA seq (Table S4 and Dataset S1). Importantly, the RNA used in this analysis was isolated from the same cultures that were assayed for hydrogenase activity.
As expected, the mRNA levels for the deleted genes in each mutant were significantly and substantially lower than the parental strain, providing an important validation that the correct strains were used in these assays. Moreover, no significant differences in mRNA abundance from the parent were observed for the remaining hydrogenases in any strain, showing that the expression of individual hydrogenase operons is not regulated by the presence/absence of other hydrogenase genes. Thus, the hydrogenase activities found in the various mutants accurately reflect the combined activities of each enzyme in all strains.
Large numbers of M. barkeri genes showed significant changes in mRNA abundance in the hydrogenase mutants, relative to the parental strain. Accordingly, 2.7% of all genes were differently regulated in strains with one or two deleted hydrogenases, whereas the Δech/Δfrh/Δvht and Δech/Δfrh/Δfre/Δvht-vhx strains showed in 17.4% and 22.5% differently regulated genes, respectively (Dataset S1). Of these, most encode proteins with unknown functions or with annotated functions that do not appear to be related to energy conservation. One exception was the F420 dehydrogenase (fpo), whose mRNA abundance increased significantly in all strains lacking frh (Table S4). This result suggests that the cell has a mechanism to sense the redox state of F420, which is altered upon deletion of frh.
DISCUSSION
While fully consistent with the proposed functions of the M. barkeri hydrogenases, our phenotypic characterization of mutants lends new insight into the flexibility and interconnected nature of methanogenic metabolism. For example, reduction of CO2 to CH4 is expected to require three kinds of electron donors: Fdred, F420red, and reduced methanophenazine (1). Consistent with this idea, mutants lacking hydrogenases that reduce Fd (Ech), F420 (Frh) or methanophenazine (Vht) are unable to grow on H2/CO2. Thus, we were surprised to observe production of methane from H2/CO2 in cell suspensions of Δfrh mutants. Assuming this process involves the standard CO2 reduction pathway, this would require an alternative source of F420red for reduction of methenyl- and methylene-tetrahydrosarcinapterin (Fig 2). Two alternative sources can be envisioned: first, Fpo could produce F420red using reduced methanophenazine as the electron donor via reverse electron transport driven by proton motive force; second, a soluble heterodisulfide reductase could produce F420red via electron bifurcation using CoM-S-S-CoB and Fdred as substrates (as suggested in (22, 23)). In the former mechanism, reduced methanophenazine would be derived from H2 using Vht; in the latter, Fdred would be derived from H2 via Ech. Interestingly, double mutants lacking Frh and either Ech or Vht produce much less methane than the Δfrh single mutant, thus both alternate pathways may contribute to this phenotype. The inability of the Δfrh mutant to grow on H2/CO2 suggests that this alternate methane-producing pathway does not provide sufficient energy for growth, or that it fails to provide an essential biosynthetic precursor.
Similar metabolic flexibility is seen during methylotrophic methanogenesis, which can occur via H2-dependent or -independent mechanisms (Figures 2, 3) (11, 12, 15, 16). We previously showed that a hydrogen cycling mechanism involving Frh and Vht is the preferred mode of electron transport in M. barkeri. Nevertheless, M. barkeri is also capable of methylotrophic growth in the absence of Frh and Vht (11, 15). Data reported here reveal that methylotrophic growth in M. barkeri is possible when all three hydrogenases are deleted. Thus, we have created an M. barkeri strain similar to Methanosarcina acetivorans, which has no detectable hydrogenase activity, but which grows well on methylotrophic substrates (7, 14). During methylotrophic growth in M. acetivorans, an electron transport chain comprised of Fpo and HdrDE is used to capture energy from F420red produced by the oxidative branch of the methanogenic pathway (Figures 2, 3). Our genetic analyses suggest that Fpo is also used to metabolize F420red in M. barkeri Δfrh/Δvht mutants (11, 15). The oxidative branch of the methylotrophic pathway also produces Fdred, which in M. acetivorans is oxidized by membrane-bound, ion-pumping Fdred:methanophenazine oxidoreductase known as Rnf (24). However, because M. barkeri does not encode Rnf, this energy-conserving electron transport pathway in not available to the Ech mutants characterized here. Thus, an alternative Fdred:heterodisulfide oxidoreductase system must exist to allow growth of these mutants on methanol. Ithas been suggested that this alternate Fdred:heterodisulfide oxidoreductase activity is catalyzed by a cytoplasmic, electron-bifurcating heterodisulfide reductase (HdrABC), similar to the electron-bifurcating heterodisulfide reductase of non-cytochrome containing methanogens (Fig 3) (22). Biochemical data from a homologous M. acetivorans enzyme supports this possibility (23).
Interestingly, the stoichiometry of methane and CO2 produced from methanol/H2 in Δfrh/Δvht and Δfrh/Δfre/Δvht-vhx mutants lends additional support for an alternate Fdred:heterodisulfide oxidoreductase. These strains, which encode only Ech, might be expected to disproportionate methanol to CH4 and CO2 in a 3:1 ratio, as was seen in the strains lacking all three active hydrogenases. Instead, they produced CH4 and CO2 at an approximate ratio of 10:1, suggesting that a substantial portion of methanol was reduced directly to CH4 using electrons obtained by H2 oxidation. Because Ech is the sole remaining hydrogenase in these strains, electrons from Fdred must be involved in this process.
H2 also inhibits oxidation of methanol when both substrates are present, via a mechanism that is clearly mediated by hydrogenase activity. Accordingly, in the presence of H2, methanol is solely reduced to methane by cell suspensions of strains that contain all three hydrogenases, while it is disproportionated to methane and CO2 in a 3:1 ratio when all three are absent. The hydrogenase-mediated inhibition of methanol oxidation is graded, with Vht having the largest effect and Ech the least.Similarly, Ech mutants are only able to grow on methanol/H2 when supplemented with acetate and pyruvate, which has been interpreted to mean they cannot produce the reducing equivalents needed for biosynthesis by oxidizing methanol to CO2 (16). We showed here that this effect is alleviated by deletion of genes encoding Frh and Vht, with the Δvht mutation having a much larger effect. Because protein synthesis was blocked by addition of puromycin in the cell suspension experiments, these effects cannot have been mediated by changes in the concentration of enzymes in the methanogenic pathways. Moreover, because inhibition requires the hydrogenase enzymes to be present, it is likely that a product of the enzymatic reaction mediates inhibition: namely the reduced enzyme cofactors. Thus, in the presence of high H2 partial pressures and the appropriate hydrogenase, we would expect the levels of oxidized methanophenazine, F420ox and Fdox to be kept at very low levels. Interestingly, the graded inhibition in response to loss of Vht, Frh and Ech mimics the thermodynamics of the hydrogenase reactions, with methanophenazine being the most energetically favorable electron acceptor and Fd being the least. This suggests at least two possible mechanisms that might account for the inhibition: i) allosteric inhibition or covalent inactivation of a key enzymatic step in the oxidative branch of the pathway could be triggered by one or more reduced cofactors, or ii) simple changes in the availability of F420ox and Fdox, which are needed for three discrete steps in the oxidative branch of the methyloptrophic pathway (Fig 2). Note that in the second mechanism, the major inhibitory effect of Vht on methyl oxidation can only be explained if high levels of reduced methanophenazine influence the levels of F420ox, which could occur by changing the equilibrium of the Fpo reaction (Fig 2, 3).
In addition to affecting flux through methanogenic pathways, the levels of reduced or oxidized cofactors may be used as a sensory input to modulate gene regulation. Transcriptional profiling of hydrogenase mutants showed that in all strains lacking frh, the fpo operon was significantly up-regulated. Without Frh, Fpo is solely responsible for the F420red:methanophenazine oxidoreductase activity required to transfer electrons from the oxidative to reductive portions of the methylotrophic electron transport pathway. Elevated abundance of fpo mRNA in Δfrh strains indicates that the cell has a mechanism to sense and respond to F420 redox imbalance. A previous study identified MreA as a global regulator in Methanosarcina with the ability to bind and repress the fpo promoter region during aceticlastic growth (25). This regulator was shown to affect gene expression based on growth substrate, however the mechanism and sensory input are unknown. Systems for gene regulation based on detected redox imbalance of F420 and other electron carriers are a potential source for future studies.
The levels of hydrogenase activity for the three enzyme types have significant ramifications for the hydrogen cycling model on energy conservation (15). We have shown that Δvht mutations are lethal when Frh is present, but not when it is absent. Moreover, when vht expression is turned off using a regulated promoter, cell lysis is concomitant with H2 accumulation, implying that the inability to recapture H2 produced in the cytoplasm is responsible for the lethal phenotype. With this in mind, it seems clear that the cytoplasmic activities of Frh must be carefully balanced against the periplasmic activity of Vht. Interestingly, our data show that Frh activity is ca. 3-fold higher than that of Vht. Thus, it appears that ability of Frh to produce H2 is much higher than the ability of Vht to take it up. We recognize that our assays were not conducted with the native substrates (which are not commercially available), therefore we approximated the in vivo activity of each enzyme based on available literature values in which a variety of natural and artificial cofactors were used (Table S5). These data suggest that the relative activities of Frh and Vht are more similar than our assay data suggest, with Frh activity ca. 1.5-fold higher than Vht. While this extrapolation must be interpreted with caution, it still suggests that Frh capacity is higher than that of Vht. In this regard, both Vht and Ech are coupled to ion-motive force; thus, activity in whole, metabolically active cells could be substantially different.
Finally, unlike Frh and Vht, Fre and Vhx are not able to provide sufficient levels of F420red and reduced methanophenazine, respectively, for growth via CO2 reduction. Additionally, the Δech/Δfrh/Δvht strain that only encoded for the Fre and Vhx hydrogenases had no detectable hydrogenase activity. This could be due to low expression of fre and vhx operons, absence of post-translational processing, mutations in structural or catalytic residues or some combination of these (7). Analysis of RNA sequencing data from wild type M. barkeri grown methylotrophically indicated the abundance mRNA for fre approximately 50-lower than frh (Dataset S1), similar to the relative abundance observed by Vaupel and Thauer (8). Additionally, the abundance of vhx transcripts was more than 200-fold lower than those of vht. We note that our enzymatic assays would have easily detected hydrogenase activity at levels 200-fold lower than we observed for the strains encoding only vht. Thus, poor gene expression cannot explain the lack of activity in strains expressing only Fre of Vhx. Hydrogenases require several maturation steps to become active enzymes, including processing by the maturation proteases encoded by the frhD and vhtD genes. Thus, it remains possible that Fre and Vhx could encode active enzymes if FrhD and VhtD are trans-acting maturation proteases. Given that the mutants characterized here removed the entire frh and vht operons, our data do not address this possibility.
MATERIALS AND METHODS
Media and growth conditions
Methanosarcina strains were grown as single cells (26) at 37 °C in high salt (HS) broth medium (27) or on agar-solidified medium as described (28). 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 N2:CO2 (80:20 v/v) at 50 kPa over ambient pressure, H2:CO2 (80:20 v/v) at 300 kPa over ambient pressure, or a combination of methanol plus hydrogen. Cultures were supplemented as indicated with 0.1% yeast extract (YE), 0.1% casamino acids (CAA), 10 mM sodium acetate, 10 mM pyruvate or 100 mM pyruvate. Puromycin (CalBioChem, San Diego, CA) was added at 2 mg ml-1 for selection of the puromycin transacetylase (pac) gene (29). 8-Aza-2,6-diaminopurine (8-ADP) (Sigma, St Louis, MO) was added at 20 mg ml-1 for selection against the presence of hpt (29). Tetracycline (Tc) was added at 100 mg ml-1 to induce the tetracycline-regulated PmcrB(tetO1) promoter (19). Standard conditions were used for growth of Escherichia coli strains (30) DH5α/λ-pir (31) and DH10B (Stratagene, La Jolla, CA), which were used as hosts for plasmid constructions.
DNA methods and plasmid constructions
Standard methods were used for plasmid DNA isolation and manipulation in E. coli (32). Liposome mediated transformation was used for Methanosarcina as described (33). Genomic DNA isolation and DNA hybridization were as described (27, 28, 34). 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 Tables S1 and S2.
Strain construction in M. barkeri
The construction and genotype of all Methanosarcina strains is presented in Table S3. Hydrogenase encoding genes were deleted sequentially in a specific order (Figure S1) because certain hydrogenase deletion mutants are only viable when other hydrogenase genes are deleted first (15). To simplify isolation of strains that lack hydrogenase operons vhxGAC and vhtGACD, the genes between the two operons (Mbar_A1842 and Mbar_A1843) were also deleted (Figure 1). All mutants were confirmed by either PCR or DNA hybridization (Figures S2-S5).
Determination of growth characteristics
For growth rate determinations, cultures were grown on methanol or methanol plus H2/CO2 (Δfrh and Δfrh/Δfre) to mid-log phase (optical density at 600 nm [OD600] ca. 0.5). Approximately 3% inoculum of the culture (or 10%, in case of acetate) was then transferred to fresh medium in at least four replicates and incubated at 37 °C. Growth was quantified by measuring OD600. With the exception of samples grown on acetate, all OD600 were measured with a Spectronic 20 spectrophotometer (Thermo Fisher Scientific, Waltham, MA); those grown on acetate were measured with a Hewlett Packard 8453 spectrophotometer (Agilent,Santa Clara, CA). Note that an OD of 1.0 on the Hewlett Packard 8453 is equivalent to an OD of ˜ 0.2 on the Spectronic 20. Generation times were calculated during exponential growth phase and lag phase was defined as the time required to reach half-maximal OD600.
Cell suspension experiments
Cells grown on methanol or methanol plus H2/CO2 (Δfrh and Δfrh/Δfre) were collected in late exponential phase (OD600 = 0.6-0.7) by centrifugation at 5,000 x g for 15 minutes at 4 °C. The cells were washed once with anaerobic HS PIPES buffer (50 mM PIPES at pH 6.8, 400 mM NaCl, 13 mM KCl, 54 mM MgCl2, 2 mM CaCl2, 2.8 mM cysteine, 0.4 mM Na2S) and resuspended in the same buffer to a final concentration of 109 cells/ml. Cells were counted visually using the Petroff-Hausser counting chamber (Hausser Scientific, PA). All assay mixtures contained 2 ml of the suspension and were conducted under strictly anaerobic conditions in 25 ml Balch tubes sealed with butyl rubber stoppers using 250 mM methanol as the methanogenic substrate under a headspace of N2, H2, or H2/CO2 (80/20%) at 250 kPa over the ambient pressure, as indicated. Puromycin (20 μg/ml) was added to prevent protein synthesis. Cells were held on ice until initiation of assay by transfer to 37 °C. For rate determination, gas phase samples were withdrawn at various time points and assayed for CH4 by gas chromatography at 225 °C in a Hewlett Packard gas chromatograph (5890 Series II) equipped with a flame ionization detector. The column used was stainless steel filled with 80/120 CarbopackTM B/3% SPTM-1500 (Supelco, Bellefonte, PA) with helium as the carrier gas. For total CH4 and CO2 production, assays were incubated at 37 °C for 36 hours prior to withdrawal of gas phase samples for analysis by GC at 225 °C in a Hewlett Packard gas chromatograph (5890 Series II) equipped with a thermal conductivity detector. A stainless steel 60/80 Carboxen-1000 column (Supelco, Bellefonte, PA) with helium as the carrier gas was used. Total cell protein was determined using the Bradford method (35) after 1 ml of the cells was lysed by resuspension in ddH20 with 1 mg/ml RNase and DNase.
Hydrogenase assays
Strains were grown at 37 °C in HS medium supplemented with 125 mM methanol and cells were harvested from 10 ml mid-exponential phase culture at 1228 x g for 15 min in an IEC MediSpin (Needham Heights, MA) benchtop centrifuge. Preparation of cell extract was performed in an anaerobic chamber under an atmosphere of H2/N2 (4/96%). Cells were washed once in 10 ml HS-MOPS [2 mM dithiothreitol (DTT), 400 mM NaCl, 13 mM KCl, 54 mM MgCl2, 2 mM CaCl2, 50 mM MOPS, pH 7.0] and lysed in 1 ml lysis buffer (2 mM DTT, 0.5% n-dodecyl β-D-maltoside, ca. 50 Kunitz units bovine pancreas deoxyribonuclease I, 50 mM MOPS, pH 7.0) on ice for 30 min. Enzyme-containing supernatant was separated from cell debris by centrifugation at 13600 x g for 2.5 min (Fisher Scientific Micro Centrifuge Model 235C, Waltham, MA). Protein concentration was measured via the Bradford method (35).
Assays were performed anaerobically in 1.7 ml quartz cuvettes sealed with rubber stoppers. A total reaction volume of 1 ml was used, and included cell extract mixed with 50 mM MOPS buffer (pH 7.0) containing 2 mM DTT and 2 mM benzyl viologen (BV). The cuvette headspace was pressurized to 30 kPa with 100% H2 after being flushed for 2 min. Cuvettes with reaction mixture were pre-warmed to 30 °C before the reaction was initiated by the addition of BV. Hydrogenase activity was determined by quantifying the change in absorbance of BV at 578 nm (extinction coefficient, 8.65 cm-1 mM-1) with a Cary 50 UV-Vis Spectrophotometer (Agilent, Santa Clara, CA). One unit (U) of hydrogenase activity was defined as the oxidation of 1 μmol H2 per minute, based on the fact that 2 μmol BV are reduced for each H2 oxidized. A minimum of three independent measurements from biological replicates was performed for each strain.
RNA sequencing
Immediately prior to cell harvest for hydrogenase assays, 2.5 ml of the same culture was harvested for total RNA isolation. An equal volume of TRIzol reagent (Ambion, Carlsbad, CA) was added to the culture to lyse cells and samples were incubated at room temperature for 5 min. RNA was then isolated with a Direct-zol RNA MiniPrep kit from Zymo Research (Irvine, CA) according to the manufacturer’s directions. RNA samples were stored at −80 °C.
To increase coverage of mRNA during sequencing, rRNA was removed from samples via subtractive hybridization. The method of Stewart et al. (36) was utilized with the following modifications. Templates for 16S and 23S rRNA probes were generated by PCR from strain WWM85 with primers 16SFor, T716SRev, 23SFor, and T723SRev. In vitro transcription with the MEGAscript High Yield Transcription kit (Ambion) was used for the production of biotinylated antisense rRNA probes from 400 ng of the purified PCR products in separate reactions. After removal of template with DNAseI, probes were purified with the Zymo Research RNA Clean & Concentrator kit. Hybridization reactions (30 μl) for each sample contained the following: 20% formamide, 1X SSC buffer, 20 U SUPERase inhibitor, 2 μg total RNA, 4 μg 16S probe, and 4 μg 23S probe. Reactions were denatured at 70 °C for 10 min, ramped down to 25 °C (−0.1 °C sec-1), and incubated at room temperature for 10 min. rRNA hybridized to biotinylated probe was removed via streptavidin-coated magnetic beads (New England Biolabs, Ipswich, MA). Beads (500 μl per sample) were washed twice with 500 μl 1X SSC buffer prior to the addition of hybridized RNA sample diluted to 250 μl in 1X SSC buffer with 20% formamide. Samples were incubated for 1 hour at room temperature with gentle shaking before separation of beads on a magnetic rack. The supernatant was removed, beads were washed with 250 μl 1X SSC, and supernatant and wash were pooled and cleaned with the Zymo Research RNA Clean & Concentrator kit.
Preparation and sequencing of RNAseq libraries was performed at the Roy J. Carver Biotechnology Center at the University of Illinois at Urbana-Champaign.Libraries were made with the TruSeq Stranded mRNA Sample Prep kit, sequenced with a HiSeq 2000 using the TruSeq SBS v3 kit, and processed with Casava 1.8.2, all per the manufacturer’s directions (Illumina, San Diego, CA). All sequencing data was further processed and analyzed as previously described (37) with CLC Genomics Workbench 7 (Qiagen). Within this program, the Empirical analysis of Differential Gene Expression (EDGE) tool was used for statistical analysis (38). Differently regulated genes were considered significant when up- or down-regulated at least 3-fold with a p-value ≤ 0.05. Three biological replicates were sequenced and analyzed for each strain. Raw and processed data have been deposited in the Gene Expression Omnibus (GEO) under the accession number GSE98441.
ACKNOWLEDGEMENTS
The authors acknowledge the Division of Chemical Sciences, Geosciences, and Biosciences, Office of Basic Energy Sciences of the U.S. Department of Energy through Grant DE-FG02-02ER15296 for funding of this work.