Effects of H₂ and Formate on Growth Yield and Regulation of Methanogenesis in Methanococcus maripaludis

Abstract

Hydrogenotrophic methanogenic Archaea are defined by an H₂ requirement for growth. Despite this requirement, many hydrogenotrophs are also capable of growth with formate as an electron donor for methanogenesis. While certain responses of these organisms to hydrogen availability have been characterized, responses to formate starvation have not been reported. Here we report that during continuous culture of Methanococcus maripaludis under defined nutrient conditions, growth yields relative to methane production decreased markedly with either H₂ excess or formate excess. Analysis of the growth yields of several mutants suggests that this phenomenon occurs independently of the storage of intracellular carbon or a transcriptional response to methanogenesis. Using microarray analysis, we found that the expression of genes encoding coenzyme F₄₂₀-dependent steps of methanogenesis, including one of two formate dehydrogenases, increased with H₂ starvation but with formate occurred at high levels regardless of limitation or excess. One gene, encoding H₂-dependent methylene-tetrahydromethanopterin dehydrogenase, decreased in expression with either H₂ limitation or formate limitation. Expression of genes for the second formate dehydrogenase, molybdenum-dependent formylmethanofuran dehydrogenase, and molybdenum transport increased specifically with formate limitation. Of the two formate dehydrogenases, only the first could support growth on formate in batch culture where formate was in excess.

INTRODUCTION

Methanogenic Archaea (methanogens) play a key role in the anaerobic breakdown of organic matter. Five of six known orders of methanogens, termed “hydrogenotrophic,” derive their energy through the use of H₂ to reduce CO₂ to methane. Hydrogenotrophic methanogens lack a cytochrome-containing electron transport chain and are proposed to conserve energy by the mechanism of electron bifurcation (1, 2). H₂ availability has profound physiological and regulatory effects in hydrogenotrophic methanogens. For example, in Methanothermobacter thermautotrophicus, when H₂ is present in excess, growth in relation to methane production (Y_CH4, grams [dry weight] of cells per mole of methane) decreases (3, 4), suggesting that energy spilling (otherwise known as growth uncoupling or overflow metabolism [5]) occurs. In M. thermautotrophicus, Methanococcus maripaludis, and other methanogens, H₂ availability also alters the expression of key genes of methanogenesis (4, 6–10). The mechanisms for the effects on growth yield and gene expression are unknown.

In addition to H₂, formate may also be a primary electron donor for methanogenesis in the environment (11, 12), and several hydrogenotrophic species can use formate without the addition of H₂. For the catabolic process of methanogenesis, formate and H₂ are interchangeable electron donors; all four reductive steps can proceed with electrons derived directly from either electron source (13–16). H₂, however, is essential for the anabolic process of CO₂ fixation (17, 18) and for an anaplerotic input to methanogenesis (16). During growth on formate, a formate:H₂ lyase activity generates the necessary H₂ (14–16).

Since formate acts independently of H₂ as an electron donor for methanogenesis, we asked whether its availability similarly affects growth yield and gene expression. We identify four different response patterns with regard to four conditions, H₂ limitation and excess and formate limitation and excess. Interestingly, a discrete set of genes responds specifically to formate limitation.

MATERIALS AND METHODS

Strains and batch culture conditions.

Strains, plasmids, and primers are listed in Table S1 in the supplemental material. Deletion mutants were either taken from previous studies or generated in this study as described previously (13, 19). Briefly, where possible, deletion constructs were digested out of the plasmid pCRprtneo and ligated directly to pCRuptneo (14); otherwise, PCR products containing the deletion of interest were generated and ligated into pCRuptneo. Strain MM901 was then transformed with the deletion constructs and subjected to a selection/counterselection system dependent on neomycin (1 to 5 mg ml⁻¹) and 6-azauracil (250 μg ml⁻¹) to generate mutants of interest (13). Strains were grown on McCas medium as described before (19). To test for auxotrophy of the cdh and ehb mutants, sodium acetate (normally 10 mM) or Casamino Acids (normally 0.2% [wt/vol]) were omitted, respectively. When formate instead of H₂ was supplied as the electron donor, McCas medium was amended with 200 mM sodium formate (with an equivalent reduction in NaCl to maintain sodium osmolarity) and 200 mM morpholinepropanesulfonic acid (MOPS) buffer (pH 7); the headspace was 140 kPa N₂-CO₂ (80:20) (14). For growth curves, a 4% inoculum was used and optical density at 660 nm (OD₆₆₀) was monitored. For all experiments, cultures were grown at 37°C.

Growth and microarray analysis of M. maripaludis grown in a chemostat under defined nutrient limitation.

M. maripaludis strain MM901 was grown in a chemostat on H₂ as described previously (8, 20, 21). For the ehb mutant, 0.1% (wt/vol) Casamino Acids was added to chemostat medium and 100 μg ml⁻¹ of ampicillin was included to prevent bacterial contamination. For unknown reasons, when Casamino Acids (Fisher BioReagents catalog number BP1424-500) were used, limitation occurred at a lower phosphate level (0.02 mM versus 0.12 mM in minimal medium). For growth on formate, the chemostat setup is diagramed in Fig. 1. Conditions were modified as follows unless otherwise indicated. Medium contained 380 mM sodium formate in place of NaCl. Although Cl⁻ is important for cellular function, replacing NaCl with sodium formate is standard practice and has no known deleterious effects on growth (22), suggesting that other Cl⁻-containing salts in our medium (e.g., MgCl₂) are sufficient. MOPS buffer was not included during continuous culture on formate. Formate utilization results in a pH increase by the following equation: 4NaCOOH + 2H₂O → CH₄ + 3CO₂ + 4NaOH. Therefore, a pH probe was used to monitor and maintain pH at 6.95 with the automated addition of 10% (vol/vol) H₂SO₄. The agitation rate was 250 rpm. The gas mixture initially supplied to the chemostat was H₂-Ar-CO_2-1% H₂S (21, 125, 40, and 14 ml min⁻¹) (8). After the OD increased above 0.6 (24 h), the medium flow was turned on at 0.083 liter h⁻¹. Medium was either nonlimiting (described above), formate limiting (modified to contain 200 mM sodium formate and 180 mM NaCl), nitrogen limiting (2 mM NH₄Cl), or phosphate limiting (0.08 mM K₂HPO₄). Also at this point, gas flow through the vessel was shut off, a 7-kPa check valve was installed on the exhaust line to maintain positive pressure within the vessel while also allowing for exhaust of gaseous metabolic by-products, and a 1% Na₂S · 9H₂O feed was started into the vessel at 0.83 ml h⁻¹. Forty-eight hours later, samples were taken for physiological measurements or microarray analysis. If conditions were nutrient limited, nonlimiting medium was then reintroduced. A subsequent increase in OD confirmed that nutrient limitation had been imposed. If the chemostat was allowed to run longer, iron sulfides precipitated several days after the start of Na₂S addition and attached to surfaces within the culture vessel (data not shown).

Fig 1.

Schematic of chemostat setup with formate as the electron donor. A more detailed diagram of the chemostat setup for H₂-based growth can be found in reference 20.

Samples were collected for microarray analysis as follows. A total of 1.5 ml of chemostat culture (OD₆₆₀, ∼0.65) was centrifuged at 13,000 × g for 30 s, and supernatant was discarded. The pellet was immediately placed in a dry ice-ethanol bath and stored at −80°C until RNA extraction and microarrays could be performed. Microarrays were run as described previously (23).

Growth yield calculations.

To calculate growth yield of M. maripaludis during H₂ limitation or excess, culture OD₆₆₀, gas flow rate, and percent methane in the gas outflow were determined. Methane concentrations were measured on a Buck Scientific model 910 GC equipped with a flame ionization detector. Carrier gases were air (110 kPa), H₂ (180 kPa), and He (170 kPa). Growth rate based on the chemostat dilution was one doubling per 499 min. Finally, an OD₆₆₀ of 1.0 was assumed to equal 0.34 g (dry cell mass) per liter (15). The following equation was used to determine dry cell mass produced per mole of CH₄ (Y_CH4): (OD₆₆₀/CH₄ [ml/min]) × (0.34 g/OD₆₆₀) × (1/499 min) × (22,400 ml/mol).

To determine the yield of grams (dry mass) per mole of formate consumed, the offset of pH in the chemostat vessel by 10% (vol/vol) H₂SO₄ addition was monitored (Y_H2SO4). H₂SO₄ consumption was monitored over ∼18 h, and Y_H2SO4 was calculated as follows: (OD₆₆₀/[milliliters of acid/hour]) × (0.34 g/OD₆₆₀) × [1/(499/60)] × (543.5 ml of acid/mol).

Y_H2SO4 was converted to Y_CH4 assuming that each mole of H⁺ from H₂SO₄ offsets the pH change from 1 mol of NaOH produced by sodium formate utilization. Therefore, 2 mol of H₂SO₄ corresponds to 4 mol of formic acid and 1 mol of CH₄.

Microarray data accession number.

Microarray data have been deposited under GEO accession number GSE42111.

RESULTS

Chemostat system for growth of M. maripaludis with formate.

Continuous culture of M. maripaludis on formate required modification of our protocol for growth on H₂. Even though the medium contained formate, it was necessary to first establish a culture in the chemostat vessel by sparging with gas containing H₂. When H₂ was subsequently excluded, culture OD steadily decreased (Fig. 2). M. maripaludis requires H₂ for methanogenesis and anabolism; normally, formate:H₂ lyase activity generates this H₂ (14–18). Maintaining H₂-free gas flow likely flushed all H₂ from the vessel, inhibiting growth. We also tried changing the agitation rate of the vessel (Fig. 2), as this affects the rate of H₂ removal from the system; however, this had no noticeable effect. We therefore completely stopped gas flow in order to maintain physiologically relevant levels of autochthonous H₂ (Fig. 2). A 7-kPa check valve was installed to maintain positive pressure within the vessel while allowing escape of gaseous metabolic products. In addition, M. maripaludis requires sulfide as a reducing agent and a sulfur source for growth (24); therefore, a system for the addition of Na₂S, instead of gaseous H₂S, was devised. OD rebounded under these conditions, and reliable steady-state growth of M. maripaludis on formate was achieved with a constant OD of 0.8 (Fig. 2). Finally, phosphate limitation was introduced, and culture OD dropped to a steady-state value of 0.6, confirming our ability to impose nutrient limitation of M. maripaludis grown on formate (Fig. 2) similarly to nutrient limitation on H₂ (20).

Fig 2.

Growth behavior of M. maripaludis in a chemostat with formate. Plot shows OD over time in response to various perturbations. Initially the vessel was sparged with gas containing H₂ (H₂-Ar-CO₂-1% H₂S; 21, 125, 40, and 14 ml min⁻¹). At 121 h (black arrow), medium flow was started and H₂ was turned off while flow of other gases was continued (Ar-CO₂-1% H₂S; 20, 20, and 6 ml min⁻¹); these gassing rates were stepped down to 5, 20, and 2 ml min⁻¹ at hour 170. At 195 h (gray arrow), all gas flow was terminated and 1% Na₂S · 9H₂O addition was started. Finally, at 243 h (white arrow), the medium flowing into the vessel was switched to a limiting level of phosphate. Medium flow was maintained at 0.083 liter/h throughout except between hours 165 and 195, when it was 0.062 liter/h. Agitation was 250 rpm throughout except after hour 121, when it was maintained between 50 and 100 rpm.

Changes in growth yield occur with both H₂ and formate as electron donors.

Our chemostat system allowed us to determine whether H₂ excess affects growth yield in M. maripaludis. We compared three continuous cultures: H₂ limited, phosphate limited, and ammonia limited. Growth rate and cell density were held similar under all conditions. Since H₂ was in excess when phosphate or ammonia limited growth, two comparisons yielded information on the effect of H₂ limitation. Y_CH4 (grams [cell mass] per mole of CH₄ produced) under H₂ limitation was 2.86 ± 0.58 (mean ± standard deviation [SD]), compared to 0.77 ± 0.20 under phosphate or ammonia limitation (Fig. 3A). Hence, H₂ excess results in a marked decrease in growth yield in M. maripaludis.

Fig 3.

Growth yields for M. maripaludis under different nutrient limitations. (A) Cell yield (grams [dry weight]) per mole of CH₄ (Y_CH4) in steady-state culture grown with H₂ or formate excess or limitation. Error bars represent 1 SD around the mean. P and N limitation samples were pooled to represent H₂ or formate excess growth. Gray bars, formate growth; white bars, H₂ growth. Diamonds, formate-limited samples; circles, H₂-limited samples; squares, N-limited samples; triangles, P-limited samples. Open symbols indicate samples used for microarray analysis. (B) Changes to Y_CH4 for various mutants upon transition from H₂ excess (phosphate limitation) to H₂ limitation at time zero.

We then asked whether formate limitation would have the same effect. A decrease in Y_CH4 due to H₂ excess has previously been demonstrated in hydrogenotrophic methanogens, but whether it occurs with formate excess has not been determined. Again, we compared three continuous cultures: formate limited, phosphate limited, and ammonia limited. Since continuous culture on formate required the cessation of gas flow through the system, we could not accurately measure the rate of CH₄ generation directly. Instead, we used the rate of acid addition, reflecting the rate of formic acid utilization, as an indirect measure for calculating Y_CH4. Y_CH4 under formate limitation was 2.31 ± 0.29, compared to 1.50 ± 0.26 under phosphate or ammonia limitation (Fig. 3A). Hence, the lowest Y_CH4 levels were observed with H₂ excess, intermediate levels occurred with formate excess, and the highest levels occurred with formate limitation or H₂ limitation (H₂ excess versus formate excess, P = 0.002; H₂ limitation versus formate limitation, P = 0.218; formate excess versus formate limitation, P = 0.006).

Changes in mRNA abundance or energy storage capacity cannot account for differences in Y_CH4.

A variety of hydrogenotrophic methanogens display altered Y_CH4 values in response to changes in H₂ availability (4, 6, 7), suggesting a mechanism common to all organisms. In all methanogens tested, the coenzyme F₄₂₀-dependent steps of methanogenesis are also transcriptionally regulated in response to H₂ (4, 8–10). Therefore, we tested the importance of these H₂-regulated genes in altering growth yields on M. maripaludis. Genes for both of the F₄₂₀-reducing hydrogenases (fru and frc) or the F₄₂₀-dependent methylene-tetrahydromethanopterin (H₄MPT) dehydrogenase (mtd) were genetically eliminated. The gene for the F₄₂₀-dependent methylene-H₄MPT reductase (mer) is essential and, therefore, was overexpressed from the replicative vector pLW40 (25) rather than genetically eliminated. Each of these mutant strains displayed a marked change in Y_CH4 upon transition from H₂ excess to H₂ limitation similar to the wild type (Fig. 3B).

The membrane-bound, energy-converting hydrogenase Eha acts to anaplerotically sustain methanogenesis by reducing ferredoxin that can be used by formylmethanofuran dehydrogenase to reduce CO₂ to formylmethanofuran (16). This reaction consumes membrane potential and could account for differences in growth yields between H₂ excess and H₂ limitation. As eha is required for methanogenesis in wild-type M. maripaludis, we elected to place it under the control of the nitrogenase promoter (P_nif) in an effort to control and repress its expression in the presence of ammonium. When grown in batch culture with formate as the electron donor (i.e., low H₂ concentrations), this strain has a slow growth phenotype, suggesting that Eha levels are indeed repressed (see Fig. S1 in the supplemental material). The P_nif-eha construct also displayed a marked difference in growth yield in response to H₂ availability (Fig. 3B).

Finally, alterations in growth yield in M. maripaludis could be due to changes in cellular composition such that energy storage occurs under H₂ excess. As Y_CH4 decreases with H₂ excess and phosphate limitation, energy storage as polyphosphates is unlikely despite the fact that this capability has been demonstrated for methanogens from the order Methanosarcinales (26). M. maripaludis was shown to synthesize glycogen (27); therefore, a deletion mutant for glycogen synthetase (glgA) was generated (28). This mutant still displayed a marked difference in Y_CH4 between H₂ excess and H₂ limitation, suggesting that the lowered growth yield under H₂ excess is not due to an investment in glycogen synthesis. Additionally, deletion mutants for carbon monoxide dehydrogenase (cdh, essential for the first step of CO₂ fixation) and the energy-converting hydrogenase ehb (important for generating reduced electron carriers for CO₂ assimilation) were generated. As expected, the cdh and ehb mutants were auxotrophic for acetate and Casamino Acids, respectively (17, 18). With these mutants, too, trends in Y_CH4 values under the two conditions were similar to those with the wild type, suggesting that differential investment in CO₂ assimilation cannot account for the differences in growth yield (Fig. 3B). However, we cannot discount the possibility that under H₂ excess, resources are diverted toward higher energy content in some form, such as cellular lipid content or composition.

Transcriptional response of genes encoding F₄₂₀-dependent steps of methanogenesis.

We compared formate-grown cultures to H₂-grown cultures. In addition, we determined if the effect of formate limitation when formate is the electron donor is similar to the effect of H₂ limitation when H₂ is the electron donor. Microarray analysis was performed using samples from the six nutrient-limited cultures described above. In each case, nutrient supplementation after sampling resulted in an increase in OD, verifying that the nutrient in question had been limiting (Table 1). A seventh culture was grown on formate, and high levels of all nutrients were provided; in this case OD was continuously high. In addition, expression levels of key genes responded as expected, since transcript levels for genes for nitrogenase (nifH) and phosphate transporter (pstB) and mtd increased with ammonia, phosphate, and H₂ limitation, respectively (8, 9, 29).

Table 1.

Nutrient-limited conditions for M. maripaludis grown in a chemostat

Sample	Condition				OD660		Log2 expressiona
	Formate (mM)	H2 (ml/min)	Phosphate (mM)	Ammonia (mM)	Limited	Non-limited	mtd	fdhB2	nifH	pstB
Undefined limitation	380	0	0.8	10	0.89		2.07	2.14	−0.33	−0.1
Formate limitation	200	0	0.8	10	0.64	0.87	1.87	3.33	−0.56	−0.12
Formate: P limitation	380	0	0.08	10	0.69	0.86	0.82	0.23	−0.47	4.71
Formate: N limitation	380	0	0.8	2	0.66	0.89	1.78	0.28	5.33	−0.14
H2 limitation	0	21	0.8	10	0.68	>1.0	3.08	0.41	−0.34	0.05
H2: P limitation	0	110	0.12	10	0.67	>1.0	−3.84	0.29	0.06	4.06
H2: N limitation	0	110	0.8	2.8	0.59	>1.0	−2.89	0.02	5.89	−0.1

^aLog₂ expression ratios are relative to a single arbitrary standard (23). Boldface type indicates expression levels consistent with the imposed nutrient limitation.

With H₂ limitation, nearly all of the genes for the F₄₂₀-dependent steps of methanogenesis showed a marked increase in mRNA abundance, in agreement with previous results (8). Thus, fdh1 (encoding formate dehydrogenase), mtd (encoding F₄₂₀-dependent methylene-H₄MPT dehydrogenase), mer (encoding methylene-H₄MPT reductase), and fru (encoding F₄₂₀-reducing hydrogenase) had higher mRNA abundance with H₂ limitation than with either condition of H₂ excess (phosphate limitation or ammonia limitation) (Fig. 4; see also Table S2 in the supplemental material). In contrast, the formate-grown cultures showed a different pattern: nearly all of the F₄₂₀-dependent activities that showed increased mRNA abundance under H₂ limitation also had high mRNA abundance during formate growth under all three conditions (Fig. 4) as well as in the seventh culture, in which all nutrients were present at high levels (see Table S2). fru displayed a single exception, showing increased mRNA abundance with formate except when nitrogen was limiting. Hence, growth on formate as the electron donor, whether formate was limiting or not, had an effect similar to that of growth on H₂ limitation.

Fig 4.

Expression of methanogenesis genes in response to nutrient limitations. Bars and numbers show log₂ expression ratios relative to an arbitrary standard (23). Values are for genes encoding active-site enzymes (except fdh; see below). ΔμNa⁺, change in Na⁺ across membrane; H₄MPT, tetrahydromethanopterin; MFR, methanofuran; CoM-SH, coenzyme M; CoB-SH, coenzyme B; CoM-S-S-CoB, heterodisulfide of CoM and CoB; F₄₂₀/F₄₂₀H₂, coenzyme F₄₂₀; Fd_ox/Fd_red, ferredoxin; Eha/Ehb, energy-converting hydrogenase A or B; Fdh, formate dehydrogenase; Ftr, formyltransferase; Fmd/Fwd, formylmethanofuran dehydrogenase; Fru/Frc, F₄₂₀-reducing hydrogenase; Hdr, heterodisulfide reductase; Hmd, H₂-dependent methylene-H₄MPT dehydrogenase; Mch, methenyl-H₄MPT cyclohydrolase; Mcr, methyl-CoM reductase; Mer, methylene-H₄MPT reductase; Mtd, F₄₂₀-dependent methylene-H₄MPT dehydrogenase; Mtr, methyl-H₄MPT:CoM methyltransferase; Vhu/Vhc, Hdr-associated hydrogenase. *, either H₂ or formate acts as the electron donor for the reaction. (Note: fdhB transcript abundance [MMP0139 and MMP1297] was used to assess fdh expression. There is a known artifact where portions of fdhA2 [MMP0138] sometimes appear highly expressed under H₂ limitation in microarrays, most likely due to probe cross-hybridization with fdhA1 [MMP1298] transcripts which share high [∼73%] nucleotide identity [23]. However, both gene expression and protein abundance analyses have demonstrated that fdhA2 is not differentially expressed between H₂ limitation and excess [8, 9]. Due to this documented ambiguity in determining fdhA2 gene expression using microarray data, fdhA was ignored in our analysis in favor of fdhB.)

Formate limitation leads to increased mRNA levels of a formate dehydrogenase and other molybdenum-associated functions.

In contrast to most F₄₂₀-dependent steps of methanogenesis, certain genes did respond specifically to formate limitation: a second formate dehydrogenase operon fdh2 (Fig. 4), the genes for the molybdenum-containing formylmethanofuran dehydrogenase (fmd), and several molybdenum transport genes (see Table S2 in the supplemental material). In contrast, neither fdh1 (which followed the pattern of other genes for F₄₂₀-dependent steps of methanogenesis) nor the genes for tungsten-containing formylmethanofuran dehydrogenase (fwd) showed differences in mRNA abundance during formate-limited growth (Fig. 4). Since the two fdh operons were expressed differently, we checked the ability of each to support growth on formate. Deletion of fdhA1 eliminated growth on formate, while deletion of fdhA2 had no effect in batch culture (Fig. 5A). Deletion mutants for fdhA1 and fdhA2 should have no polar effects on formate transport (fdhC) or carbonic anhydrase (CA) (Fig. 5B).

Fig 5.

Growth of M. maripaludis fdh mutants on H₂ or formate. (A) Open symbols, growth on H₂; filled symbols, growth on formate; diamonds, wild type; squares, ΔfdhA1; circles, ΔfdhA2; triangles, ΔfdhA1-ΔfdhA2. Data are averages of duplicate growth experiments. Similar curves were observed in replicate experiments. (B) Structure of the operons for fdh1 and fdh2.

A distinct pattern occurred with hmd, which encodes the H₂-dependent methylene-H₄MPT dehydrogenase (Fig. 4). Although regulation appeared complex, H₂ limitation clearly resulted in low mRNA abundance, in agreement with previous results (9, 30). Formate limitation had a similar effect.

DISCUSSION

We observed four different response patterns with regard to H₂ and formate (Table 2). First, mRNA transcript levels for the F₄₂₀-dependent steps of methanogenesis were high under three conditions, H₂ limitation, formate limitation, and formate excess, and were low only with H₂ excess. One possibility is that the presence of formate could act independently of H₂ limitation. However, since the same set of genes is affected, we favor the hypothesis that the presence of formate or H₂ limitation acts through the same set of regulatory signals (see below). The second response pattern, that of growth yield, was similar: Y_CH4 was high with H₂ limitation or formate limitation, had an intermediate level with formate excess, and was lowest with H₂ excess. The third response pattern was shown with only one gene: hmd had decreased expression with H₂ limitation or formate limitation compared to formate excess or H₂ excess. Fourth, one set of genes responded specifically to formate limitation: fdh2, fmd, and genes for molybdenum transport increased.

Table 2.

Summary of regulatory and physiological responses to H₂ and formate^a

Gene(s)	H2 excess	Formate excess	H2 limitation	Formate limitation
F420-dependent genesb	—	↑	↑	↑
hmd	— or ↑	— or ↑	↓	↓
YCH4	Low	Intermediate	High	High
fdh2, fmd, molybdenum transport genes	—	—	—	↑

^a—, baseline expression; ↑, increased expression; ↓, decreased expression.

^bGenes for F₄₂₀-dependent steps of methanogenesis (except fdh2).

During growth on formate, a low level of H₂ is present due to its production and reutilization (14). It is tempting to speculate that H₂ is a primary signal and that different response thresholds determine the different regulatory patterns. In this model, the highest H₂ levels occur with H₂ excess. The next step down in H₂ occurs with growth on formate supplied in excess; at this point transcript levels for F₄₂₀-dependent steps in methanogenesis increase and Y_CH4 increases to an intermediate level. The next lower H₂ level occurs with H₂ limitation, where Y_CH4 reaches a high level and hmd mRNA decreases in abundance. Hmd is an unusual [Fe] hydrogenase that has a low affinity for H₂ (31), so when H₂ levels are low, Hmd is of limited utility.

F₄₂₀H₂ concentrations are directly affected by H₂ concentrations in the growth medium (32); therefore, it was previously suggested that when H₂ limits growth, the genes for F₄₂₀-dependent processes are upregulated to maintain a constant availability of F₄₂₀H₂ (8). When M. maripaludis is grown with formate as the electron donor, F₄₂₀H₂-dependent formate:H₂ lyase activity is essential for growth, explaining the need to maintain high levels of F₄₂₀-dependent enzymes under these conditions (14–16). A previous study that analyzed protein abundance of M. maripaludis cultures grown under different H₂ gassing regimens demonstrated that changes in mRNA levels correspond to changes in protein abundance for these enzymes (9).

The remaining set of genes, fdh2, fmd, and molybdenum transport genes, could respond to the lowest extreme of H₂ or could respond directly to a limiting level of formate. Fdh and Fmd both contain a molybdopterin cofactor (33). However, it is unlikely that these genes responded to molybdenum limitation as a primary signal, since a switch to medium containing higher formate and identical molybdate concentrations resulted in an increased OD, verifying that formate, not molybdenum, was the limiting nutrient (Table 1). We suggest that formate limitation leads to fdh2 expression and that molybdenum transport and fmd expression are secondary and tertiary responses, in which increased demand for molybdenum leads to increased molybdenum transport and increased intracellular molybdenum induces fmd transcription. Consistent with this, increased molybdate in cultures of M. thermautotrophicus resulted in increased fmd expression (34).

The different patterns of mRNA abundance for fdh1 and fdh2 suggest different roles for the two formate dehydrogenases. In a previous study by Wood et al., our lab reported that either Fdh was sufficient for growth on formate, whereas a later study by Lupa et al. reported that only Fdh1 supported growth on formate unless a Δfdh1 mutant was incubated for prolonged periods, ∼70 h (15, 35). This suggests that enrichment of a suppressor mutation may have accounted for the initial observations of Wood et al.; the data in our present study agree with this interpretation. We propose that Fdh1 functions for sustained growth on formate or for a transition from growth on H₂ to growth on formate as H₂ becomes limiting, while Fdh2 functions only at low formate concentrations. However, since a mutant containing only fdh2 (the fdh1 deletion mutant) could not be grown in the lab on formate without a probable suppressor mutation altering its expression pattern, we could not test the growth of this mutant under formate-limited conditions.

The differences in growth yield that occurred in M. maripaludis were striking, reflected in Y_CH4 values that varied as much as 4-fold. The fact that mutants for several genes in carbon metabolism (glgA, cdh, and ehb) display this phenotype suggests that energy storage does not account for differences in Y_CH4 between H₂ excess and limitation or formate excess and limitation. Additionally, mutation of each of the F₄₂₀ metabolizing enzymes of methanogenesis had no impact on the ability of M. maripaludis to alter growth yields in response to H₂. Therefore, we tentatively propose that differences in Y_CH4 values in hydrogenotrophic methanogens are due to energy spilling, or the dissipation of energy when the metabolic electron donor is present in excess. The energy yield of hydrogenotrophic methanogens relies on electron bifurcation at the heterodisulfide reductase complex (1), where exergonic and endergonic electron flows are coupled via reduced ferredoxin (Fig. 4). One possibility is that this coupling partially breaks down. To make up for the endergonic requirement, electron flow through the energy-converting hydrogenase Eha, which normally functions anaplerotically (16), could increase with a concomitant consumption of membrane potential. However, if this is the case, we saw no evidence in the form of any regulation of Eha or any other steps of methanogenesis at the transcriptional level. Indeed, a promoter switch mutant for eha still displayed differences in cell yield in response to H₂ availability. As an alternative, energy spilling in methanogenic Archaea may be more commonplace, perhaps involving a futile cycle that consumes ATP or membrane potential. Such cycling has been observed in Escherichia coli starved for potassium (36). Indeed, organisms from all three domains of life have documented examples of energy spilling metabolism (5). In methanogenic Archaea, energy spilling could be beneficial to maintain low concentrations of H₂ or formate in order to prevent nutrient use by competitors or to maximize the energetics of metabolism during syntrophic growth: rapid nutrient consumption shifts the equilibrium of syntrophic metabolism, allowing for robust growth of both partners despite energy spilling on the part of the methanogen.