Growth Kinetics, Carbon Isotope Fractionation, and Gene Expression in the Hyperthermophile Methanocaldococcus jannaschii during Hydrogen-Limited Growth and Interspecies Hydrogen Transfer.

Hyperthermophilic methanogens are often H2 limited in hot subseafloor environments, and their survival may be due in part to physiological adaptations to low H2 conditions and interspecies H2 transfer. The hyperthermophilic methanogen Methanocaldococcus jannaschii was grown in monoculture at high (80 to 83 μM) and low (15 to 27 μM) aqueous H2 concentrations and in coculture with the hyperthermophilic H2 producer Thermococcus paralvinellae The purpose was to measure changes in growth and CH4 production kinetics, CH4 fractionation, and gene expression in M. jannaschii with changes in H2 flux. Growth and cell-specific CH4 production rates of M. jannaschii decreased with decreasing H2 availability and decreased further in coculture. However, cell yield (cells produced per mole of CH4 produced) increased 6-fold when M. jannaschii was grown in coculture rather than monoculture. Relative to high H2 concentrations, isotopic fractionation of CO2 to CH4 (εCO2-CH4) was 16‰ larger for cultures grown at low H2 concentrations and 45‰ and 56‰ larger for M. jannaschii growth in coculture on maltose and formate, respectively. Gene expression analyses showed H2-dependent methylene-tetrahydromethanopterin (H4MPT) dehydrogenase expression decreased and coenzyme F420-dependent methylene-H4MPT dehydrogenase expression increased with decreasing H2 availability and in coculture growth. In coculture, gene expression decreased for membrane-bound ATP synthase and hydrogenase. The results suggest that H2 availability significantly affects the CH4 and biomass production and CH4 fractionation by hyperthermophilic methanogens in their native habitats.IMPORTANCE Hyperthermophilic methanogens and H2-producing heterotrophs are collocated in high-temperature subseafloor environments, such as petroleum reservoirs, mid-ocean ridge flanks, and hydrothermal vents. Abiotic flux of H2 can be very low in these environments, and there is a gap in our knowledge about the origin of CH4 in these habitats. In the hyperthermophile Methanocaldococcus jannaschii, growth yields increased as H2 flux, growth rates, and CH4 production rates decreased. The same trend was observed increasingly with interspecies H2 transfer between M. jannaschii and the hyperthermophilic H2 producer Thermococcus paralvinellae With decreasing H2 availability, isotopic fractionation of carbon during methanogenesis increased, resulting in isotopically more negative CH4 with a concomitant decrease in H2-dependent methylene-tetrahydromethanopterin dehydrogenase gene expression and increase in F420-dependent methylene-tetrahydromethanopterin dehydrogenase gene expression. The significance of our research is in understanding the nature of hyperthermophilic interspecies H2 transfer and identifying biogeochemical and molecular markers for assessing the physiological state of methanogens and possible source of CH4 in natural environments.

for assessing the physiological state of methanogens and possible source of CH 4 in natural environments. KEYWORDS Methanocaldococcus, RNA-Seq, Thermococcus, carbon isotope fractionation, hydrogen, hyperthermophiles, methanogenesis, syntrophs E ach year, approximately 1 Pg of CH 4 is produced globally through methanogenesis, largely by methanogens growing syntrophically with fermentative microbes that hydrolyze biopolymers (1), but little is known about the magnitude or mechanism of methanogenesis through thermophilic H 2 syntrophy or interspecies H 2 transfer. Deepsea hydrothermal vents are known habitats for thermophilic methanogens (2). It was also estimated that 35% of all marine sediments are above 60°C (3), suggesting that these environments likewise provide a large global biotope for thermophiles. Microcosms containing low-temperature hydrothermal fluid as well as an archaeal coculture derived from a high-temperature oil pipeline each produced CH 4 through interspecies H 2 transfer at 80°C when supplemented with organic compounds, both without added H 2 (4,5). Both showed that CH 4 was produced from a mixed microbial community consisting of the hyperthermophilic H 2 -producing heterotroph Thermococcus and the (hyper)thermophilic, hydrogenotrophic methanogens Methanocaldococcus, Methanothermococcus, and Methanothermobacter.
In this study, growth and CH 4 production kinetics, carbon isotope fractionation, and gene expression data were examined together for a hyperthermophilic methanogen under conditions ranging from monoculture growth at high and low H 2 concentrations to coculture growth with an H 2 -producing partner. The hyperthermophile Methanocaldococcus jannaschii was grown in monoculture in a chemostat under H 2 -replete and H 2 -limited conditions based on previous kinetic experiments (9). It was also grown with the H 2 -producing hyperthermophilic heterotroph Thermococcus paralvinellae using maltose and formate separately as the growth substrates (22). The purpose was to determine if M. jannaschii cell yield (amount of biomass produced per mole of CH 4 produced, or Y CH4 ) increases when cultures are shifted from H 2 -replete to H 2 -limited growth conditions and if Y CH4 remains high or increases further during interspecies H 2 transfer. This study also examined if interspecies H 2 transfer stimulates the growth rate or cell yield of T. paralvinellae or ameliorates its H 2 inhibition relative to its growth in monoculture. Furthermore, isotopic carbon fractionation was examined to determine if CH 4 is isotopically lighter when H 2 flux is reduced, as previously observed in moderately thermophilic methanogens (23)(24)(25). Finally, differential gene expression analysis using transcriptome sequencing (RNA-Seq) was used to determine if changes occur in M. jannaschii for the expression of genes for carbon assimilation, CH 4 production, or energy generation when H 2 decreases in availability. This study demonstrates the utility of measuring growth kinetic parameters, carbon isotope fractionation, and differential gene expression patterns for two species grown in coculture. The data elucidate how hyperthermophilic methanogens behave in a high H 2 flux environment, such as those found at some hydrothermal vents, versus a low H 2 flux environment, such as petroleum reservoirs.

RESULTS
Growth parameters for mono-and cocultures. A summary of the growth conditions is provided in  (Fig. 1A). Cell concentrations in the medium and H 2 and CH 4 concentrations in the headspace remained constant throughout growth in the chemostat (see Fig. S1 in the supplemental material). Attempts to grow M. jannaschii in coculture with T. paralvinellae in the chemostat when either maltose or formate was the energy source, with and without stirring and gas sparging of the medium with CO 2 and N 2 , were unsuccessful. This was likely due to the open reactor that permits gas to flow out of the reactor without any gas pressure increase. Coculture growth was readily established in sealed bottles that contained 1 atm of gas pressure at room temperature. At 82°C, the gas pressure in the bottle was 1.2 atm, which slowed H 2 efflux from the growth medium to the headspace. Therefore, the cocultures were grown in sealed bottles with the same volume of medium and headspace as the chemostat. The growth rates of M. jannaschii decreased further when it was grown in coculture with T. paralvinellae to 0.12 Ϯ 0.01 h Ϫ1 and 0.22 Ϯ 0.03 h Ϫ1 when T. paralvinellae was grown on maltose and formate, respectively (Fig. 1A). Relative to H 2 concentrations when T. paralvinellae was grown in the bottles in monoculture, nearly all the H 2 was removed from the coculture bottles and CH 4 was produced (Fig. S2).  (Fig. 1C). Summaries of the growth and CH 4 production kinetics data for M. jannaschii are available in the supplemental material ( Fig. S1 and S2 and Tables S1 and S2).
There was no change in the specific growth rate or maximum cell concentration of T. paralvinellae when it was grown with or without M. jannaschii or with a change in carbon source ( Fig. 2A and Fig. S2). The specific growth rates of T. paralvinellae grown on maltose in monoculture and in coculture were 0.16 Ϯ 0.01 h Ϫ1 and 0.22 Ϯ 0.02 h Ϫ1 , respectively, while growth rates on formate in monoculture and in coculture were 0.18 Ϯ 0.05 h Ϫ1 and 0.16 Ϯ 0.02 h Ϫ1 , respectively ( Fig. 2A). Furthermore, when grown on maltose, there was no change in the growth yield (Table S3) or cell-specific acetate production rate of T. paralvinellae when grown in monoculture (0.94 Ϯ 0.16 pmol cell Ϫ1 h Ϫ1 ) relative to growth in coculture (1.05 Ϯ 0.15 pmol cell Ϫ1 h Ϫ1 ) (Fig. 2B). However, when grown on maltose, T. paralvinellae produced formate (in addition to H 2 and acetate) when grown in monoculture (0.60 Ϯ 0.18 pmol cell Ϫ1 h Ϫ1 ) but not when grown in coculture (Fig. 2C). The cell-specific H 2 production rate was higher when T. paralvinellae was grown in monoculture on formate (130.9 Ϯ 11.1 fmol cell Ϫ1 h Ϫ1 ) than for monoculture growth on maltose (0.9 Ϯ 0.1 fmol cell Ϫ1 h Ϫ1 ) (Table S3). A summary of the growth and metabolite production kinetics data for T. paralvinellae is available in the supplemental material ( Fig. S2 and Table S3). There was no growth of M. jannaschii when it was incubated in monoculture in medium supplemented with only 0.01% yeast extract or 0.1% sodium formate and 0.01% yeast extract with N 2 :CO 2 in the headspace. These additions also did not stimulate the growth of M. jannaschii in monoculture when an H 2 :CO 2 headspace was provided.
Carbon isotope fractionation. The final carbon isotopic composition (␦ 13 C CO2 ) values were Ϫ24.4 to Ϫ21.6‰ in the coculture bottles and Ϫ33.3 to Ϫ28.2‰ in the   (Table 1). Similarly, ␦ 13 C CH4 values became more negative with increasing H 2 limitation during coculture cell growth. In monoculture with 1.92 atm of initial H 2 in the headspace at 82°C, the ␦ 13 C CH4 from M. jannaschii was Ϫ34.2 to Ϫ32.9‰ (Table 1). ␦ 13 C CH4 values decreased to Ϫ91.2 to Ϫ89.0‰ when M. jannaschii was grown in coculture with T. paralvinellae on maltose and to Ϫ99.4‰ when grown in coculture on formate. The corresponding CO2-CH4 values increased from 22.1 to 23.0‰ during monoculture growth in a serum bottle to 73.5 to 85.1‰ during growth in coculture with T. paralvinellae (Table 1).
Transcriptomic analyses. RNA-Seq mapped 1,866 transcripts to the M. jannaschii genome. The thirteen samples that span four growth conditions were analyzed based on principal-component analysis (PCA) (Fig. S3A) and t-distributed stochastic neighbor embedding (t-SNE) (Fig. S3B) results. Pairwise comparisons of M. jannaschii grown in monoculture on high and low H 2 showed up to 12 genes to be differentially expressed (adjusted P value of Ͻ0.01 and log 2 fold change [|log 2 FC|] of Ͼ1) with 1 gene downregulated and 11 genes upregulated during growth on low H 2 relative to growth on high H 2 (Table S4) S5) for M. jannaschii grown in monoculture on high and low H 2 in the chemostat. The genes that code for a GTP binding protein (MJ_RS01180), bacteriohemerythrin (MJ_RS03980), radical SAM protein (MJ_RS04390), a signal recognition particle (MJ_RS05550), a transcriptional regulator (MJ_RS06225), and four hypothetical proteins were upregulated on low H 2 , while a gene that codes for a histone (MJ_RS04990) was upregulated on high H 2 (Table S4).
For cocultures grown on maltose, 97% of the reads mapped unambiguously to the T. paralvinellae genome and 1.5% mapped to the M. jannaschii genome. For cocultures grown on formate, 67% of the reads mapped unambiguously to the T. paralvinellae genome and 29% mapped to the M. jannaschii genome. These proportions generally matched the proportions of T. paralvinellae and M. jannaschii cells in each coculture  (Table S5). However, we cannot rule out the possibility that some of these gene expression changes are caused by the switch from the chemostat to bottles.
F 420 -dependent methylene-H 4 MPT dehydrogenase (mtd, MJ_RS05555) gene expression was upregulated 4.3-fold in coculture relative to that of M. jannaschii grown under monoculture conditions (Fig. 3A). In contrast, gene expression of H 2 -dependent methylene-H 4 MPT dehydrogenases (hmd, MJ_RS04180; hmdX, MJ_RS03820) were both downregulated 2.1-fold in M. jannaschii grown in coculture relative to that of M. jannaschii grown in monoculture ( Fig. 3B and Fig. S4). There was no change in gene expression for the methyl-CoM reductase I and II genes (Fig. S5). Gene expression for a hypothetical protein with a predicted RNA-binding domain (MJ_RS03480) showed a 22.5-fold increase in cocultures relative to monocultures (Fig. S6). Expression of 6 of the 9 M. jannaschii genes that code for a V-type ATP synthase (MJ_RS01130 to MJ_RS01165 and MJ_RS03255) were downregulated when cultures were grown in coculture relative to expression in M. jannaschii grown in monoculture (Fig. 4). Similarly, expression of 14 genes in a putative operon for membrane-bound, ferredoxin-dependent hydrogenase was also downregulated in M. jannaschii cultures grown in cocultures relative to cultures grown in monoculture (Fig. 4). These genes include Eha subunits A and B (MJ_RS02795 to MJ_RS02800), an oxidoreductase (MJ_RS02755), a dehydrogenase (MJ_RS02765), and a catalytic subunit (MJ_RS02730).

DISCUSSION
Microorganisms in nature live in complex communities and biogeochemically impact their environment through interspecies metabolic interactions. Most of what is known about the kinetics and physiology of methanogenesis at various H 2 concentrations and in coculture comes from studies of the thermophile Methanothermobacter thermoautotrophicus and the mesophile Methanococcus maripaludis. Growth rates of both organisms decreased when they were H 2 limited relative to H 2 -replete growth. However, growth yields (Y CH4 ) increased when the cultures were H 2 limited (26-28). Prior to this study, growth yields had not been measured for any methanogen during interspecies H 2 transfer or for any hyperthermophilic methanogens under various H 2 concentrations.
To determine M. jannaschii metabolism and kinetics under H 2 -replete and H 2 -limited growth conditions, as defined in a previous study (9), continuous growth in chemostats was established. The decrease in specific growth rate and cell-specific CH 4 production rate of M. jannaschii when grown in monoculture under H 2 -limited conditions show that growth and methanogenesis rates are limited by H 2 concentration. This trend continued when M. jannaschii was grown in coculture with T. paralvinellae, suggesting that interspecies H 2 transfer led to further H 2 limitation of methanogenesis. However, the cell yield for M. jannaschii increased when the cells were grown in coculture relative to growth in monoculture. This is consistent with previous studies that show higher cell yields for M. thermoautotrophicus and M. maripaludis upon H 2 limitation, but there is no consensus on a physiological explanation (26)(27)(28). During methanogenesis, methyl-H 4 MPT is either converted to methyl-CoM for production of CH 4 and energy generation on the cytoplasmic membrane or to acetyl-CoA for biosynthetic reactions (Fig. 5). Depending on the H 2 concentration, hydrogenotrophic methanogens decide between maximum growth rate and maximum growth yield. This pattern can be explained by the rate-yield trade-off, which creates two divergent ecological strategies, namely, (i) slow growth but efficient metabolism and high yields when resources are scarce, and (ii) fast growth but inefficient metabolism and low yields upon rich resources. The rate-yield trade-off is suggested to be integral to evolution and the coexistence of species (29). It was proposed previously but not demonstrated that syntrophic growth of methanogens with a fermentative partner is optimized for cell yield rather that growth rate (27). In this study, M. jannaschii grew and produced CH 4 solely on the H 2 produced by T. paralvinellae, and the cell yield of M. jannaschii increased in coculture compared to that of growth in monoculture.
Thermococcus species use maltose for biosynthesis and energy generation that yields acetate and CO 2 as well as H 2 and a proton/sodium-motive force via a membrane-bound hydrogenase (30,31). However, they are auxotrophic for certain amino acids that must be supplied from the environment (32,33). T. paralvinellae increased gene expression of a membrane-bound formate hydrogenlyase operon and produced formate when inhibited by exogenous H 2 , suggesting that it converts H 2 to formate when H 2 is inhibited (22). T. paralvinellae also separately used formate as an energy source in the absence of maltose, produced H 2 , and generated a proton/ sodium-motive force but required 0.01% yeast extract in the growth medium (22). Consequently, the cell-specific H 2 production rate was ϳ100-fold higher when cultures were grown on formate.
Morris et al. (34) defined microbial syntrophy as obligately mutualistic metabolism and included coculture growth between the hyperthermophilic H 2 producer Pyrococcus furiosus and various hyperthermophilic methanogens, including M. jannaschii, as an example based on increased cell concentrations of both organisms in coculture relative to each in monoculture (35). Unlike T. paralvinellae, P. furiosus lacks formate hydrogenlyase as a mechanism to overcome H 2 inhibition (36) and may be more dependent upon syntrophy to ameliorate H 2 inhibition. In this study, when T. paralvinellae was grown with M. jannaschii, growth in coculture did not stimulate the growth rate, growth yield, or maximum cell concentration of T. paralvinellae. This suggests the relationship between T. paralvinellae and M. jannaschii is not obligately mutualistic and therefore more accurately represents interspecies H 2 transfer rather than syntrophy. However, there was no formate production when T. paralvinellae was grown in coculture on maltose with M. jannaschii, and M. jannaschii cannot grow on formate (37 and this study), so M. jannaschii does appear to ameliorate H 2 inhibition in T. paralvinellae when grown in coculture.
It was shown previously that the fractionation of carbon isotopes between CO 2 and CH 4 increased with decreasing concentrations of H 2 availability or, more accurately, with decreasing Gibbs energy for the methanogenesis reaction (23). The CO2-CH4 fractionation factor for the thermophile Methanothermobacter marburgensis increased from 22 to 39‰ at high H 2 concentrations to 58 to 64‰ at limiting H 2 concentrations (23,24). It was proposed that variations in the carbon isotopic fractionation factor are controlled by the extent of reversibility of the methanogenesis pathway, which was proposed to increase with decreasing Gibbs energy availability (23). In this study, the CH 4 produced was isotopically more negative and the CO2-CH4 fractionation factor increased when M. jannaschii was grown in the chemostat with low H 2 relative to high H 2 conditions. Similarly, in bottles, CH 4 was isotopically more negative and CO2-CH4 was much larger when M. jannaschii was grown in coculture with T. paralvinellae than when it was grown in monoculture with an initial estimated aqueous H 2 concentration of 1.2 mM. The most negative CH 4 in this study was produced when M. jannaschii was grown in coculture and H 2 fluxes are presumably at their lowest rates.
Previous studies showed that during CO 2 fixation and methanogenesis (Fig. 5) in M. thermoautotrophicus and M. maripaludis, gene expression for H 2 -dependent methylene-H 4 MPT dehydrogenase (hmd) decreased while expression of cofactor F 420 -dependent methylene-H 4 MPT (mtd) increased when growth was H 2 limited relative to that of H 2 -replete growth (27,28,38). It was suggested that the Mtd reaction is the more reversible of the two methylene-H 4 MPT dehydrogenase reactions, which facilitates enhanced carbon isotope fractionation by methanogenesis pathway reversal in these methanogens under H 2 -limited conditions (23). The proteome of M. jannaschii contained a lower abundance of Hmd and higher abundances of Mtd and four flagellar proteins in early logarithmic growth phase when grown in batch phase under H 2 -limited conditions than under H 2 -replete conditions, but both Hmd and Mtd were found at high relative abundances in late logarithmic growth phase when grown under H 2 -replete conditions (39). During H 2 syntrophy, the M. thermoautotrophicus proteome had more Mtd and less Hmd than were seen with monoculture growth under H 2replete conditions (40). There were no significant changes in gene expression or protein abundance for Hmd and Mtd in M. maripaludis during H 2 syntrophy relative to that of an H 2 -limited monoculture (41).
In this study, RNA-Seq was used to determine changes in gene expression profiles in M. jannaschii for carbon assimilation, CH 4 production, and energy generation pathways when there were changes in H 2 availability. When M. jannaschii was grown under H 2 -limited conditions and in coculture, mtd expression was significantly upregulated and hmd expression was significantly downregulated in coculture cells compared to that of monoculture cells. This suggests a preference for F 420 as an electron carrier in the methanogenesis pathway under H 2 -limited conditions. The increase in cell yield in coculture was not supported by a change in the expression of genes in the carbon assimilation and methanogenesis pathways. No significant changes were detected in the expression of methyl-CoM reductase I and II and methyl-H 4 MPT:CoM methyltransferase, which catalyze the last two steps of methanogenesis (Fig. 5). Previously, changes in the relative abundances of methyl-CoM reductases I and II were observed in M. thermoautotrophicus with H 2 availability and growth during syntrophy (27,40,42). Moreover, there was no change in expression in our study in the carbon monoxide dehydrogenase/acetyl-CoA synthase genes, which code for the enzyme that converts methyl-MPT to acetyl-CoA.
In coculture, there was up to a 22.5-fold increase in the expression of a putative RNA binding protein that is only found in methanogens and the Thermococcales and has been proposed to regulate cellular activity at the translation level (43). The decrease in the expression of genes in the putative membrane-bound, ferredoxin-dependent hydrogenase operon and in the membrane-bound, Na ϩ -translocating V-type ATPase operon supports the kinetic observations that M. jannaschii is energy limited when grown in coculture. Under H 2 -limited coculture conditions, the cell must direct more of its methyl-H 4 MPT toward biosynthesis. Furthermore, there was no change in the expression of the genes for flagella. This was different from what was previously observed for M. jannaschii using proteomics (39) and may be due to the use of a chemostat in this study instead of a batch reactor.
In environments such as low-H 2 hydrothermal vents along subduction zones and some mid-ocean ridges, oil reservoirs, and high saline shale beds where organic compounds are present and H 2 efflux rates are low, thermophilic methanogens like M. jannaschii likely can grow and produce CH 4 through interspecies H 2 transfer with hyperthermophilic H 2 -producing heterotrophs, like T. paralvinellae, with high cell yields and large carbon isotope fractionations, but they do so at very low rates. This likely explains the presence of thermophilic H 2 producers and thermophilic, hydrogenotrophic methanogens in petroleum reservoirs and may be a source of CH 4 in that habitat. In contrast, high-temperature methanogens in high-H 2 hydrothermal vents, such as those supported by serpentinization and following volcanic eruptions (2), may subsist entirely from abiotic H 2 with elevated cell-specific CH 4 production rates and smaller carbon isotope fractionations. Metatranscriptomic analyses coupled with carbon isotope analyses of native CH 4 will help to determine what fraction of methanogenesis in a high-temperature environment is due to interspecies H 2 transfer relative to growth on abiotic H 2 . In this manner, we will be better equipped to model cooperative, competitive, and neutral interactions between different species in an environment and predict the biogeochemical outcome of a mixed community living in a habitat.

MATERIALS AND METHODS
Growth media and culture conditions. Methanocaldococcus jannaschii DSM 2661 (37) and Thermococcus paralvinellae DSM 27261 (44) were purchased from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ). The growth medium for pure cultures of M. jannaschii was based on DSM medium 282 (9). For the cocultures of M. jannaschii and T. paralvinellae and monoculture of T. paralvinellae, the base medium was amended with 0.01% (wt vol Ϫ1 ) yeast extract (vitamin B 12 fortified; Difco), 1 M Na 2 WO 4 ·2H 2 O, 0.26 M (NH 4 ) 2 Fe(SO 4 ) 2 ·6H 2 O, and 0.25 M (NH 4 ) 2 Ni(SO 4 ) 2 ·6H 2 O. The primary carbon and energy source added for T. paralvinellae was either 0.5% (wt vol Ϫ1 ) maltose (Sigma) or 0.1% (wt vol Ϫ1 ) sodium formate (Fluka). All media were pH balanced to 6.00 Ϯ 0.05 and reduced with 0.025% (wt vol Ϫ1 ) each of cysteine-HCl and Na 2 S·9H 2 O before inoculation. To test if M. jannaschii can use formate or yeast extract for growth in the absence of H 2 , or if they stimulate growth in the presence of H 2 , M. jannaschii was incubated in monoculture on the base medium amended with 0.1% formate and 0.01% yeast extract or 0.01% yeast extract only as described above, each in serum bottles with 1 additional atm (100 kPa) of either H 2 :CO 2 (80%:20%) or N 2 :CO 2 (80%:20%) added to the headspace at room temperature prior to incubation.
M. jannaschii was grown in monoculture at 82°C and under high and low H 2 concentrations in a chemostat to measure its growth and CH 4 production kinetics and to generate biomass for gene expression analysis. A 2-liter bioreactor (all-in-one benchtop reactor; Ace Glass) with gas flow, temperature (Ϯ0.1°C), and pH (Ϯ0.1 unit; Eutech Instruments pH 200 Series) controls was used with 1.5 liters of growth medium. The medium was maintained at pH 6.0 Ϯ 0.1 by the automatic addition of 0.25 mM HCl. For high-H 2 conditions, the bioreactor was gassed with a mixture of CO 2 (20.5 ml min Ϫ1 ) and H 2 (132 ml min Ϫ1 ). For low-H 2 conditions, the bioreactor was gassed with a mixture of CO 2 (20.5 ml min Ϫ1 ), N 2 (130 ml of gas min Ϫ1 ), and H 2 (2.5 ml min Ϫ1 ). Pure gases were blended using a mass flow controller (Matheson Tri-Gas) and added to the bioreactor through a single submerged fritted bubbler (70 to 100 m; Ace Glass; ASTM certified). The reactor is an open system and remains at ambient gas pressure. It was stirred at 150 to 180 rpm using a four-blade open impeller (6-cm diameter) with a glass shaft and Teflon blades. Aqueous H 2 and CH 4 concentrations were measured before and after inoculation by drawing 25 ml of medium from the bottom of the bioreactor directly into anoxic 60-ml serum bottles and measuring the headspace gas. H 2 was measured using a gas chromatograph fitted with a thermal conductivity detector (Shimadzu GC-8A) and a 60/80 Carboxen 1000 column (15 feet by 1/8 inch; Supelco). CH 4 was measured using a gas chromatograph fitted with a flame ionization detector (Shimadzu GC-17A) and a 5A 80/100 molecular sieve column (6 feet by 1/8 inch; Alltech). The aqueous H 2 concentrations in the bioreactor prior to inoculation were 80 to 83 M for the high-H 2 condition and 15 to 27 M for the low-H 2 condition (Table 1).
The media were inoculated with 50 to 100 ml of a logarithmic-growth-phase culture of M. jannaschii. During growth, liquid samples were drawn from the bioreactor and cell concentrations were determined using phase-contrast light microscopy and a Petroff-Hausser counting chamber. The growth rate (k) was determined by plotting cell concentration against time and fitting a logarithmic curve to the growth data. M. jannaschii was grown in batch reactor mode until the culture reached mid-logarithmic growth phase, and then the bioreactor was switched to chemostat mode by pumping sterile growth medium into the bioreactor from a sealed 12-liter reservoir that was degassed with N 2 through a submerged glass tube and heated to 75°C. Simultaneously and at the same rate, spent growth medium was pumped out of the bioreactor using a dual-channel peristaltic pump. The H 2 and CH 4 concentrations in the headspace of the bioreactor were measured using gas chromatography as described above. At high and low H 2 concentrations, cells were grown in the reactor at low enough cell concentrations such that there was excess H 2 in the headspace and the cells were not H 2 limited (see Fig. S1 in the supplemental material).
Growth of M. jannaschii was stable in the chemostat after three volume replacements of the medium within the reactor (ϳ5 h for high H 2 , ϳ14 h for low H 2 ) and was monitored for an additional ϳ0.5 volume replacements to obtain kinetic data. The CH 4 production rate per cell (q) was calculated from the sum of the CH 4 concentration in the headspace times the gas flow rate and the CH 4 concentration in the medium times the medium dilution rate (i.e., CH 4 production rate), which was normalized by the total cell concentration in the reactor. The cell yield per mole of CH 4 produced (Y CH4 ) was calculated by dividing the cell production rate (dilution rate times cell concentration) by the CH 4 production rate. The complete contents of the bioreactor then were drained into ice-cooled centrifuge bottles, spun in a centrifuge at 10,000 ϫ g and 4°C for 60 min, resuspended in 1 ml of TRIzol (Invitrogen), and frozen at Ϫ80°C until processed. Chemostats were run in triplicate for both conditions.
M. jannaschii and T. paralvinellae were grown in coculture at 82°C in 2-liter gas-tight flasks (Pyrex bottles sealed with rubber lyophilization stoppers) containing 1.5 liters of medium with ambient pressure of N 2 :CO 2 (80%:20%) in the headspace at room temperature without agitation and either maltose or formate as the energy source (Table 1). Separate logarithmic-growth-phase cultures of M. jannaschii and T. paralvinellae were combined to inoculate the bottles. The coculture was established immediately and did not require prior coculture transfers. At various times during growth, total cell concentration in bottles was determined using a Petroff-Hausser counting chamber and phase-contrast light microscopy. The M. jannaschii cell concentration was determined by counting the number of autofluorescent cells using epifluorescence microscopy and UV light excitation (45). The concentration of T. paralvinellae cells was calculated by subtracting the concentration of M. jannaschii cells from the total cell concentration. The pH change was Ͻ0.1 pH units during growth. For comparison, T. paralvinellae was grown separately in the same bottles and conditions in monoculture on 0.5% maltose and separately on 0.1% sodium formate, both with ambient pressure of N 2 :CO 2 in the headspace at room temperature. Cell concentrations were measured as described above.
The growth rates (k) of M. jannaschii and T. paralvinellae were determined by plotting cell concentration against time and fitting a logarithmic curve to the growth data. The total amounts of CH 4 and H 2 in the bottles were determined by gas chromatography. The concentrations of formate, acetate, butyrate, isovalerate, and 2-methylbutyrate were measured from aliquots of syringe-filtered (0.2-m pore size) spent medium from each coculture and T. paralvinellae monoculture incubation at various time points (for maltose growth only) using ultra-high-pressure liquid chromatography (UHPLC) as previously described (46). Methanogen cell yields (Y CH4 ) were determined from the linear slope of the number of methanogen cells per bottle plotted against the amount of CH 4 per bottle (47). The rate of CH 4 production per cell is calculated from k/(0.693 ϫ Y CH4 ) as previously described (47). Similarly, T. paralvinellae cell yields based on acetate and formate produced and for H 2 produced (for monoculture only) were determined from the linear slope of T. paralvinellae cell concentration plotted against acetate, formate, or H 2 concentration. When the cocultures reached late logarithmic growth phase, the cells were harvested for transcriptome analysis as described above (T. paralvinellae cells were not harvested when grown in monoculture). Cocultures grown on maltose were grown in triplicate, while cocultures grown on formate were grown in quadruplicate.
Carbon isotope fractionation. At the start (T o ) and end (T f ) of each chemostat run, 20 ml of chemostat headspace was transferred in triplicate into evacuated vials (Labco Exetainer). M. jannaschii also was grown in monoculture in 245-ml serum bottles containing 100 ml of medium and 1 additional atm (100 kPa) of H 2 :CO 2 (80%:20%) added to the headspace at room temperature prior to incubation. M. jannaschii was also grown in coculture with T. paralvinellae in 245-ml serum bottles containing 100 ml of either 0.5% maltose medium or 0.1% sodium formate medium as described above. The isotopic signatures of CH 4 were determined using a gas chromatography-combustion-isotope ratio mass spectrometer (GC-C-IRMS; Thermo Scientific) equipped with a GS-CarbonPlot column (30 m long, 0.320-mm inner diameter, 1.50-m film thickness; Agilent). Isotopic signatures were determined using external CH 4 standards of known isotopic signatures (Ϫ57.40 Ϯ 0.06‰) that were obtained from Arndt Schimmelmann (Indiana University). The error of the analysis was determined from external standards, and the standard deviation of multiple injections was 0.3‰. At T o and T f of the chemostat runs and the serum bottles, triplicate samples of dissolved inorganic carbon (DIC) were drawn from the growth medium. Each DIC sample (either 0.8 or 1.0 ml) was syringe filtered (0.2 m pore size) and injected into prepared vials (Labco Exetainer) that had been flushed with He and contained 100 l of phosphoric acid. Samples were analyzed by GasBench-IRMS. DIC standards were prepared in concentrations from 0.5 to 7.0 mM using KHCO 3 and Li 2 CO 3 of known isotopic composition (Ϫ38.1‰ and Ϫ1.1‰, respectively). The error of analysis was determined from external standards, and the standard deviation of multiple injections was 0.3‰. The ␦ 13 C CO2 value was calculated from the ␦ 13 C DIC value using the relationship of Mook et al. (48) at the temperature of the cultures (82°C).
Carbon isotopic compositions are presented as ␦ 13 C in the per mille notation (‰) relative to the VPDB (Vienna Pee Dee Belemnite) standard: where R sample is the 13 C/ 12 C ratio of the sample and R standard is 0.0112372. The notation is used to express isotope fractionation factors in per mille (‰): The fractionation factor, ␣, is defined as the ratio between the isotopic ratio in the substrate and product: where R CO2 is the 13 C/ 12 C ratio of the initial CO 2 and R CH4 is the 13 C/ 12 C ratio of the CH 4 produced. The propagated error of the fractionation factors was 0.4‰, except in the case of the M. jannaschii monoculture.
The inorganic carbon in the M. jannaschii monoculture serum bottles was extensively drawn down, substantially altering the 13 C signature of the remaining reactant. The fractionation factor was therefore calculated by setting the initial CO 2 isotopic signature equal to that in serum bottles without cells (Ϫ26.1 Ϯ 0.8‰) and reacting it stepwise under different fractionation factors. To obtain final isotopic compositions that match the remaining CO 2 (ϩ18.9 and ϩ15.5‰) and the final accumulated product, CH 4 (Ϫ32.9‰ and Ϫ34.2‰), fractionation factors of 22.1 Ϯ 1.3‰ and 23.0 Ϯ 1.3‰ were required in the two different experiments.
RNA-Seq analysis. Total RNA was extracted from 13 cell pellets from each growth condition (Table  1) using a Direct-zol RNA extraction kit (Zymo). RNA quantity was determined using Qubit fluorometry. RNA integrity was checked using an Agilent 2100 bioanalyzer, a NanoDrop 2000 spectrophotometer, and gel electrophoresis of the RNA, followed by staining with ethidium bromide. Removal of rRNA, library construction, multiplexing, and sequencing of the mRNA using an Illumina HiSeq2500 sequencer with two 150-bp paired ends was performed commercially by GENEWIZ, LLC (South Plainfield, NJ, USA), as described by the company. Sequencing depths ranged from 30,751,946 to 41,634,527 sequence reads per sample, with a median of 34,532,231 and a mean of 35,155,474 reads per sample. The RNA-Seq reads were mapped to both M. jannaschii and T. paralvinellae genomes using BBSplit from the BBMap package (https://sourceforge.net/projects/bbmap/). BBSplit is an aligner tool that bins sequencing reads by mapping them to multiple references simultaneously and separates the reads that map to multiple references to a special "ambiguous" file for each of them. For further analyses, we removed all ambiguously mapped reads to both genomes and worked with only the reads that unambiguously map to the M. jannaschii genome. Two to 5% of the reads were lost in this step.
The mapped reads for M. jannaschii were aligned to the M. jannaschii genome and sorted using the STAR aligner, version 2.5.1b (49). Aligned sequence reads were assigned to genomic features and quantified using the featureCounts read summarization tool (50). The output of the analyses generated BAM files containing the sequence of every mapped read and its mapped location. An unsupervised t-SNE algorithm (51) and PCA were used to predict outliers among the total RNA sample replicates.
Genes that were differentially expressed were identified using DESeq2 in the Bioconductor software framework (https://www.bioconductor.org) in R (version 3.3 [http://www.r-project.org]) and on a Galaxy platform using DEBrowser (52)(53)(54)(55). Relative log expression normalization was performed by using the R package DESeq2. The DESeq2 package allows for sequencing depth normalization between samples, estimates gene-wise dispersion across all samples, fits a negative binomial generalized linear model, and applies Wald statistics to each gene. The genes were reported as differentially regulated if the |log 2 FC| value was Ͼ1 and the adjusted P value was Ͻ0.01. Heatmaps were plotted in R (version 3.3 [http:// www.r-project.org]) using the pheatmap package. The heatmap color scale represents the z-score, which is the number of standard deviations the mean score of the treatment is from the mean score of the entire population.
Data availability. The count files and raw sequences are available in the NCBI Gene Expression Omnibus (GEO) database under accession no. GSE112986.

SUPPLEMENTAL MATERIAL
Supplemental material for this article may be found at https://doi.org/10.1128/AEM .00180-19.