Hydrogen and carbon isotope systematics in hydrogenotrophic methanogenesis under H₂-limited and H₂-enriched conditions: implications for the origin of methane and its isotopic diagnosis

Abstract

Hydrogen and carbon isotope systematics of H₂O–H₂–CO₂–CH₄ in hydrogenotrophic methanogenesis and their relation to H₂ availability were investigated. Two H₂-syntrophic cocultures of fermentatively hydrogenogenic bacteria and hydrogenotrophic methanogens under conditions of <10² Pa-H₂ and two pure cultures of hydrogenotrophic methanogens under conditions of ~10⁵ Pa-H₂ were tested. Carbon isotope fractionation between CH₄ and CO₂ during hydrogenotrophic methanogenesis was correlated with pH₂, as indicated in previous studies. The hydrogen isotope ratio of CH₄ produced during rapid growth of the thermophilic methanogen Methanothermococcus okinawensis under high pH₂ conditions (~10⁵ Pa) was affected by the isotopic composition of H₂, as concluded in a previous study of Methanothermobacter thermautotrophicus. This “${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect” is a possible cause of the diversity of previously reported values for hydrogen isotope fractionation between CH₄ and H₂O examined in H₂-enriched culture experiments. Hydrogen isotope fractionation between CH₄ and H₂O, defined by (1000 + ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$)/(1000 + ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$), during hydrogenotrophic methanogenesis of the H₂-syntrophic cocultures was in the range 0.67–0.69. The hydrogen isotope fractionation of our H₂-syntrophic dataset overlaps with those obtained not only from low-pH₂ experiments reported so far but also from natural samples of “young” methane reservoirs (0.66–0.74). Conversely, such hydrogen isotope fractionation is not consistent with that of “aged” methane in geological samples (≥0.79), which has been regarded as methane produced via hydrogenotrophic methanogenesis from the carbon isotope fractionation. As a possible process inducing the inconsistency in hydrogen isotope signatures between experiments and geological samples, we hypothesize that the hydrogen isotope signature of CH₄ imprinted at the time of methanogenesis, as in the experiments and natural young methane, may be altered by diagenetic hydrogen isotope exchange between extracellular CH₄ and H₂O through reversible reactions of the microbial methanogenic pathway in methanogenic region and/or geological methane reservoirs.

Background

Methane is a greenhouse gas, a primary product of methanogenic archaea, a useful fuel for human activity, and the simplest organic molecular. The geochemical origins of methane in environmental samples have been investigated in terms of global warming (e.g., Patra et al. 2011), limits of the deep biosphere (e.g., Inagaki et al. 2015), exploration and development of energy resources (e.g., Kvenvolden 1988), and prebiotic chemical evolution on the early Earth (e.g., McCollom 2013). The geochemical origin of methane can be classified into four types in terms of the carbon source (inorganic vs. organic) and the generation process (chemical vs. microbial): abiotic, thermogenic, aceticlastic and methylotrophic, and hydrogenotrophic methane (e.g., Kawagucci et al. 2013a). Because hydrogenotrophic methanogenesis is one of the most important energy metabolisms in living ecosystems in the ancient Earth and habitable extraterrestrial environments (McCollom 1999; Takai et al. 2004; Hsu et al. 2015), distinguishing hydrogenotrophic methanogenesis from the other methanogenic processes is a key to discuss whether life is present or absent there.

The stable isotope ratios of carbon (¹³C/¹²C) and hydrogen (D/H) in methane have been used as geochemical tracers to deduce its origin (e.g., Schoell 1980; 1988; Whiticar et al. 1986; Whiticar 1999). Although the usefulness of the stable isotope tracers for methane origins have been long debated (Martini et al. 1996; Waldron et al. 1999; Tang et al. 2000), the stable isotope diagnosis becomes more accurate by enhancing the understanding of stable isotope systematics with respect to each of the methanogenic processes in addition to subsequent alteration of the imprinted isotope signature. For this purpose, carbon isotope fractionation between CH₄ and CO₂ in hydrogenotrophic methanogenesis has been well studied by laboratory incubations (e.g., Valentine et al. 2004; Penning et al. 2005; Takai et al. 2008). The magnitude of the carbon isotope fractionation is thought to depend on the thermodynamic state of the methanogenic environment: a greater ¹³C-depletion in the produced CH₄ occurs at a lower H₂ availability. This relationship has been explained by “differential reversibility” in the multistep pathway of methanogenesis from H₂ and CO₂ (Valentine et al. 2004): higher reversibility exhibits at a lower H₂ availability.

In contrast to the carbon isotopic characteristics, the hydrogen isotope characteristics of hydrogenotrophic methanogenesis remain less understood. In general, the hydrogen isotope ratio of produced CH₄ is related solely to that of H₂O (e.g., Daniels et al. 1980). Based on this finding, numerous studies have examined the CH₄–H₂O hydrogen isotope fractionation both in experiments using various methanogen cultures (Balabane et al. 1987; Sugimoto and Wada 1995; Valentine et al. 2004; Yoshioka et al. 2008; Hattori et al. 2012; Stolper et al. 2015; Wang et al. 2015) and by observations on methane-bearing environments (e.g., Nakai et al. 1974; Whiticar et al. 1986; Burke 1993; Waldron et al. 1999). Large variation of the CH₄–H₂O hydrogen isotope fractionation has been reported in previous studies (Additional file 1), and the possible relationship between the fractionation and the H₂ availability in the methanogenic environment has been discussed (Burke 1993; Yoshioka et al. 2008; Hattori et al. 2012; Stolper et al. 2015; Wang et al. 2015). On the other hand, Kawagucci et al. (2014) recently demonstrated that the hydrogen isotope ratio of substrate H₂ also affects the hydrogen isotope ratio of the produced CH₄ during the rapid growth of a thermophilic methanogen, Methanothermobactor thermautotrophicus strain ∆H, in H₂-enriched batch culture. If this “δD_H2 effect” occurs ubiquitously in other hydrogenotrophic methanogens and under other growing conditions such as low H₂ availability, it may be necessary to reconsider the geochemical implications of the traditional interpretation based on the CH₄–H₂O hydrogen isotope ratio fractionation.

In this study, we therefore investigated the hydrogen and carbon isotope systematics of H₂O–H₂–CO₂–CH₄ during the growth of different hydrogenotrophic methanogens under various growth conditions and the effect of H₂ availability. We tested two different species of methanogens under syntrophic growth conditions with fermentatively hydrogenogenic bacteria that attained very low H₂ availability (corresponding to <10² Pa-H₂ in the headspace gas) and two different species of hydrogenotrophic methanogens under H₂-enriched growth conditions (~10⁵ Pa-H₂ in the headspace gas). This is the first attempt to monitor both hydrogen and carbon isotope systematics in low-pH₂ cultures, although the hydrogen and carbon isotope systematics have been investigated separately (Penning et al. 2005; Yoshioka et al. 2008) (Additional file 1). Compilation of the hydrogen and carbon isotope fractionations from the experimental results and environmental observations reveals three provinces in the dataset and provides new insights into stable isotope diagnosis of the origin and fate of environmental methane.

Methods

Organism details and medium compositions

Table 1 provides information on the four culture sets and the experimental conditions used in this study. To assess the growth of methanogens under low H₂-availability conditions at different temperatures, two species of butyrate-oxidizing hydrogenogenic fermentative bacteria were cocultured with two species of methanogens as syntrophic consortia at 55 °C and 25 °C (thermophilic and mesophilic cocultures). The partial pressure of H₂ for the cocultures was lower than 10² Pa-H₂ during the incubation (see Results). In addition to the coculture sets, two hydrogenotrophic methanogen species were grown as pure culture under high H₂-availability conditions at 60 °C and 30 °C (thermophilic and mesophilic pure cultures). The initial partial pressure of H₂ for the pure culture sets was 2.0–2.2 × 10⁵ Pa-H₂. The metabolic reactions in the culture sets are as follows:

Table 1 Experimental and organismal information examined in this study

• 4H₂ + CO₂ → CH₄ + 2H₂O (R1: Hydrogenotrophic methanogenesis),

• CH₃CH₂CH₂COO⁻ +2H₂O → 2CH₃COO⁻ + H⁺ + 2H₂ (R2: Butyrate oxidation),

• 2CH₃CH₂CH₂COO⁻ + 2H₂O + CO₂ → 4CH₃COO⁻ + 2H⁺ + CH₄ (R3: Consortium).

For the thermophilic coculture experimental set (hereafter, TC), we purchased a pure coculture of Syntrophothermus lipocalidus strain TGB-C1 and Mtb. thermautotrophicus strain ΔH (DSM 12681) from Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ; Braunschweig, Germany). Mtb. thermautotrophicus is one of the best-studied methanogens in terms of its genetic, biochemical, and biogeochemical characteristics (e.g., Smith et al. 1997; Kawagucci et al. 2014). S. lipocalidus was originally isolated from a granular sludge in a thermophilic upflow anaerobic sludge blanket reactor (Sekiguchi et al. 2000). The　syntrophism between these microbes was previously studied by Sekiguchi et al. (2000). In addition, the CH₄–H₂O hydrogen isotope fractionation for this syntrophic consortium has been previously investigated (Yoshioka et al. 2008). The syntrophic consortium was cultivated in the medium as follows (L⁻¹): 0.54 g NH₄Cl, 0.14 g KH₂PO₄, 0.20 g MgCl₂–6H₂O, 0.15 g CaCl₂–2H₂O, 25 g NaCl, 2.0 g NaHCO₃, 0.5-mL resazurin solution (1 g L⁻¹), 1-mL vitamin solution, and 1-mL trace element solution. The composition of the vitamin solution was as follows (L⁻¹): 4.88 mg biotin, 8.82 mg folic acid, 4.12 mg pyridoxine–HCl, 6.74 mg thiamine–HCl, 7.52 mg riboflavin, 2.44 mg nicotinamide, 9.54 mg calcium pantothenate, 27.1 mg vitamin B12, 2.74 mg 4-aminobenzic acid, and 4.12 mg lipoic acid. The trace element solution contained (L⁻¹) the following composition: 1.27 g FeCl₂–4H₂O, 0.20 g MnCl₂–4H₂O, 0.13 g CoCl₂, 0.14 g ZnCl₂, 1.3 mg CuCl₂–2H₂O, 13.3 mg AlCl₃, 6.2 mg H₃BO₃, 24.2 mg Na₂MoO₄–2H₂O, 13.0 mg NiCl₂, 1.7 mg Na₂SeO₃, and 3.3 mg Na₂WO₄–2H₂O. Batches of the culture in 120-mL serum bottles, each containing 40 mL of medium, were flushed with mixture gas of N₂/CO₂ (80:20, v/v) for 5 min and sealed with butyl rubber septums and aluminum crimp seals. After autoclaving at 121 °C for 20 min, the headspace was pressurized with N₂ gas up to 150 kPa. After gas injection, 0.5 mL of the filter-sterilized reducing solution was added to the bottles. The reducing solution contained (L⁻¹) 6.25 g Na₂S–9H₂O and 6.25 g cysteine-HCl. Butyrate solution sterilized in an autoclave was added to the culture bottles at a final concentration of 20 mmol/L immediately before the inoculation. The amount of inoculum for each TC batch was 0.55 mL. Batches of TC were incubated at 55 °C in the dark without shaking.

For the mesophilic coculture experimental set (hereafter, MC), we used a butyrate-oxidizing enrichment culture from our laboratory. The culture was obtained from subseafloor sediments collected off the Shimokita Peninsula, Japan (Imachi et al. 2011; Saito et al. 2014). The culture is a highly purified enrichment culture that constitutes only three types of microorganism: a mesophilic hydrogenotrophic methanogen, Methanobacterium sp. strain MO-MB1; a mesophilic butyrate-oxidizing hydrogenogenic bacterium, Syntrophomonas sp.; and a heterotrophic anaerobic bacterium, Geosporobacter sp. The procedure for culture batch preparation for MC was almost the same as that for TC with some differences, as follows. The base medium composition for MC was as follows (L⁻¹): 0.54 g NH₄Cl, 0.10 g KH₂PO₄, 4.0 g MgCl₂–6H₂O, 1.0 g CaCl₂–2H₂O, 25 g NaCl, 2.0 g NaHCO₃, 0.5-mL resazurin solution (1 g L⁻¹), 1-mL vitamin solution, and 1-mL trace element solution. The butyrate solution was added to the bottles at a final concentration of 10 mmol/L. The amount of inoculum for each MC batch was 1.5 mL. Batches of MC were incubated at 25 °C without shaking in the dark. The preliminary experiment showed that the growth of the consortium was optimal at 25 °C with minimum growth of the contaminant bacterium Geosporobacter sp. (Saito et al. 2014). We also confirmed that butyrate oxidation occurred together with methane production stoichiometrically, which is explained by the eq. R1, 2, and 3 (see Additional file 2).

For the thermophilic pure culture experimental set (hereafter, TP), we used a thermophilic hydrogenotrophic methanogen, Methanothermococcus okinawensis strain IH1 (JCM 11175), from the culture collection of JAMSTEC. Mtc. okinawensis was isolated from a deep-sea hydrothermal vent chimney at the Iheya Ridge, Okinawa Trough (Takai et al. 2002). The base medium composition for TP was as follows (L⁻¹): 0.09 g KH₂PO₄, 0.09 g K₂HPO₄, 3.0 g MgCl₂–6H₂O, 4.0 g MgSO₄–7H₂O, 0.8 g CaCl₂, 0.33 g KCl, 30 g NaCl, 0.25 g NH₄Cl, 10.0 mg NiCl₂–6H₂O, 0.24 mg Na₂MoO₄–2H₂O, 4.9 mg NaSeO₄, 10.0 mg Fe₂(SO₄)₃, 0.5-mL resazurin solution (1 g L⁻¹), 10-mL trace element solution, and 2.5 g NaHCO₃. The trace element solution contained (L⁻¹): 1.0 g NaCl, 0.5 g MnSO₄–H₂O, 0.18 g CoSO₄–2H₂O, 0.1 g CaCl₂–2H₂O, 0.18 g ZnSO₄–7H₂O, 0.01 g CuSO₄–5H₂O, 0.02 g KAl(SO₄)₂–12H₂O, 0.01 g H₃BO₃, 0.025 g NiCl₂–6H₂O, and 0.3 mg Na₂SeO₃–5H₂O. Batches of the culture in 160-mL pressure-resistant serum bottles, each containing 20 mL of medium, were flushed with N₂ for 5 min, sealed with butyl rubber septums and aluminum crimp seals, and autoclaved at 121 °C for 20 min. The headspace for the TP comprised 100 kPa N₂, 100 kPa CO₂, and 200 kPa H₂. The medium was reduced by the addition of autoclaved 5 % (w/v) Na₂S–9H₂O solution to a final concentration of 2.1 mmol/L. The amount of inoculum for each TP batch was 0.1 mL. Batches of TP were incubated at 60 °C without shaking in the dark.

For the mesophilic pure culture experimental set (hereafter, MP), we used Methanobacterium sp. MO-MB1 isolated from subseafloor sediment in our laboratory (Imachi et al. 2011). The base liquid medium composition for MP was almost the same as that for MC, except that butyrate solution was not added. The MP batches were prepared using almost the same procedure as for the TP batches, except that the headspace gas composition was 160 kPa N₂, 40 kPa CO₂, and 200 kPa H₂. The amount of inoculum for each TP batch was 0.1 mL. Batches of MP were incubated at 30 °C without shaking in the dark.

General notation of isotope ratios

The term “hydrogen” in this paper includes both deuterium (D or ²H) and protium (H or ¹H). The relative abundance of D to H or ¹³C to ¹²C was noted R:

$$ R = \left[\mathrm{D}\right]/\left[\mathrm{H}\right]\ \mathrm{or}\kern0.5em \left[{}^{13}\mathrm{C}\right]/\left[{}^{12}\mathrm{C}\right] $$

(1)

We used delta (δ) notation to express the relative difference between the isotope ratio of a sample and a standard in the permil scale:

$$ \updelta = \left[\left({R}_{\mathrm{sample}}/{R}_{\mathrm{standard}}\right) - 1\right] \times 1000\ \left({\mbox{\fontencoding{U}\fontfamily{wasy}\selectfont\char104}} \right). $$

(2)

The δD and δ¹³C values were calculated with respect to the international standards VSMOW and VPDB, respectively. The isotope fractionation factor, α, is generally defined as the division between the isotope ratios of two molecules as follows:

$$ {\upalpha}_{\mathrm{A}\hbox{-} \mathrm{B}} = {R}_{\mathrm{A}}/{R}_{\mathrm{B}}, $$

(3)

$$ {\upalpha}_{\mathrm{A}\hbox{-} \mathrm{B}} = \left(1000 + {\updelta}_{\mathrm{A}}\right)/\left(1000 + {\updelta}_{\mathrm{B}}\right) $$

(4)

In this study, each C or H is superscribed with α to indicate carbon or hydrogen isotope fractionation, respectively, for convenience. The fractionation factors at isotopic equilibrium between H₂ − H₂O (${\upalpha^{\mathrm{eq}}}_{{\mathrm{H}}_2-{\mathrm{H}}_2\mathrm{O}}$), CH₄ − H₂O (${\upalpha^{\mathrm{eq}}}_{{\mathrm{CH}}_4-{\mathrm{H}}_2\mathrm{O}}$), and CH₄ − CO₂ (${\upalpha^{\mathrm{eq}}}_{{\mathrm{CH}}_4-{\mathrm{CO}}_2}$), and their temperature dependency (Bardo and Wolfsberg 1976; Horita and Wesolowski 1994; Horibe and Craig 1995; Horita 2001) are used.

Experimental settings of isotope ratios

To investigate the hydrogen isotope systematics during methanogen growth, the initial δDH₂O and δDH₂ values in the batch cultures were adjusted by adding D₂O and DH molecules (Table 2). For the coculture experiments, the initial ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ values were adjusted to three levels: nonD₂O-labeled (−62‰ to −58‰), moderately D₂O-labeled (+1406‰ to +1639‰), and highly D₂O-labeled (+3286‰ to +3787‰) (see the next section Sampling and analyses for δDH₂O determination). When the H₂–H₂O equilibrium is attained with these ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ values, the equilibrated ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values (${{\updelta \mathrm{D}}^{\mathrm{eq}}}_{{\mathrm{H}}_2}$) will be −755‰ to −709‰ (nonD₂O-labeled), −369‰ to −189‰ (moderately D₂O-labeled), and +120‰ to +479‰ (highly D₂O-labeled) (see Additional file 2), which are calculated from the temperature-dependent ${\upalpha^{\mathrm{eq}}}_{{\mathrm{H}}_2\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values (0.309 and 0.260 at 55 °C and 25 °C: Horibe and Craig 1995). Duplicate batches were examined for each of the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ levels in both coculture experiments (batches A1–A6 for TC and batches B1–B6 for MC). Precultures of the coculture experiments were conducted using the same medium in terms of ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ levels in order to reduce carry-over δD signatures from the precultures.

Table 2 Initial stable hydrogen isotopic conditions for the hydrogenotrophic methanogenesis experiments

For the pure culture experiments, the initial ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ values were also adjusted to three levels: nonD₂O-labeled (−59‰ to −43‰), moderately D₂O-labeled (${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ = +1097‰ to +1427‰), and highly D₂O-labeled (${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ = +2317‰ to +3142‰). In addition, the initial ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values were adjusted with respect to the theoretical ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values expected from the H₂–H₂O equilibrium, i.e., we prepared initial ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values that were higher (DH-rich), lower (D₂O-rich), or almost identical to the equilibrated ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ value at the incubation temperatures (Table 2). Moreover, we predicted that the initial δDH₂ value of the batch during methanogen growth would approach a ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ value isotopically equilibrated with the medium H₂O (Valentine et al. 2004; Kawagucci et al. 2014) because it is known that the functions of hydrogenases promote isotope exchange (Vignais 2005; Campbell et al. 2009; Yang et al. 2012; Walter et al. 2012). Six batches for TP (C1–C6) and seven batches for MP (D1–D7) were examined (Table 2). In the same manner as the coculture experiments, precultures for the pure culture experiments were conducted with media containing the same ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ and ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ levels in order to reduce carry-over δD signatures from the precultures.

Sampling and analyses

To determine the partial pressures (pH₂ and pCH₄) and isotope ratios (${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$, ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$, ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{H}}_4}$, ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{O}}_2}$), the headspace gas of each batch was collected during incubation. The sampling frequency was determined on the basis of pCH₄ monitoring in preculture to include both the exponential and stationary growth phases of the methanogens. For the TC and MC, 8 mL of the headspace gas was subsampled into an evacuated 8-mL glass vial at each sampling time to quantify pH₂ and ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ at low H₂ level. In contrast, for the TP and MP, 0.5 mL of the headspace gas was subsampled into an 8-mL glass vial filled with ultrapure grade helium (purity is >99.9999 %: Iwatani Gasnetwork Corporation, Osaka, Japan) at 100 kPa for convenience for the isotope ratio analyses. In an attempt to assess the change of ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ during the growth, 2 mL of the medium fluid was collected immediately after the inoculation and at the end of the experiment, filtered with 0.2-μm pore-size filters (ADVANTEC, Tokyo, Japan), and measured within 12 h of sampling. Because of slight changes before and after the incubation (see Additional file 2), we used the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ value at the end of incubation to evaluate the isotope systematics (Additional file 2). For only TP, cell density was determined using 0.1 mL of medium subsample by a direct cell count with DAPI (4',6-diamidino-2-phenylindole), to evaluate the growth rate and the cell-specific H₂ consumption rate (Kawagucci et al. 2014 and references therein). The cell densities of the other three experimental sets (TC, MC, and MP) could not be determined using this procedure because the cells did not diffuse in the medium but were tightly attached to the surface of the culture bottle.

The partial pressures of H₂ and CH₄ were determined by gas chromatography using a helium ionization detector (GC-HID) with an in-house standard comprising 100 ppm H₂ and CH₄ over a He matrix. Overall errors for pH₂ and pCH₄ analyses were expected to be within 10 %.

The stable hydrogen and carbon isotope ratios for CH₄ were determined by continuous-flow isotope ratio mass spectrometry (CF-IRMS) with an on-line gas preparation and introduction system connected with a mass spectrometer, MAT253 (Thermo Fisher Scientific, Bremen, Germany), based on a previous study (Umezawa et al. 2009). Details of the system and analytical procedure are as follows. A helium-purged purification line made of stainless-steel tubing and including several two-position valves (VICI Precision Sampling, Inc. Louisianna, USA) with chemical and cold traps was used as the on-line gas preparation and introduction system. Ultra-pure-grade helium (purity >99.9999 %: Iwatani Gasnetwork Corporation) was used with further purification by a Molecular Sieve 5A column at −196 °C (liquid-N₂ bath). The sample gas including CH₄ was introduced via a gas-tight syringe (PRESSURE-LOK® series, VICI Precision Sampling, Inc., Louisiana, USA) into a 30 mL/min (~ + 0.2 MPa) helium stream, named the “precon stream,” of the purification line. CH₄ in the sample gas in the precon stream was separated from CO₂ and H₂O by a stainless-steel tubing coil trap held at −110 °C (ethanol/liquid-N₂ bath) and a chemical trap filled with magnesium perchlorate (Mg(ClO₄)₂; Merck KGaA, Darmstadt, Germany) and Ascarite II (sodium-hydroxide-coated silica; Thomas Scientific, Swedesboro, New Jersey, USA). Subsequently, CH₄ was condensed on a stainless-steel tubing trap filled with HayeSep-D porous polymer (60/80 mesh, Hayes Separations Inc., Texas, USA) held at −130 °C (ethanol/liquid-N₂ bath) mounted on a two-position six-port valve. After the turn of the valve position to introduce another helium stream set at 1.0 mL/min, named the “GC stream,” into the Hayesep-D trap, the condensed CH₄ was released at >80 °C (hot water bath). The CH₄ on the GC stream was again condensed on a capillary trap made of PoraPLOT Q (20 cm long, 0.32 mm i.d.) held at −196 °C (liquid-N₂ bath) for cryofocus and finally released at room temperature. After complete separation of CH₄ from the other molecules by a HP-PLOT Molesieve capillary column (30 m long, 0.32 mm i.d.) at 40 °C, the effluent CH₄ went to the pyrolysis or combustion units (Thermo Fisher Scientific, Massachusetts, USA) to be converted into H₂ or CO₂, respectively. During the analyses, the pyrolysis and combustion units were maintained at 1440 °C and 960 °C, respectively. The pyrolysis unit was conditioned twice a month by repeated injection of 0.2 mL of pure CH₄ to form a graphite coat on the inner wall of the tubing for quantitative conversion of the sample CH₄ to H₂. The CH₄-derived H₂ and CO₂ were finally introduced via an open-split interface, GC Combustion Interface III (Thermo Fisher Scientific, Massachusetts, USA), into MAT253. Hydrogen or carbon isotope ratios were obtained through simultaneous monitoring of H₂ ⁺ isotopologues at m/z = 2 and 3 or CO₂ ⁺ isotopologues at m/z = 44, 45, and 46, respectively. Stable hydrogen isotope ratio for H₂ was also determined with MAT253 by CF-IRMS in another publication (Kawagucci et al. 2010). Stable carbon isotope ratio for the CO₂ was determined with an isotope-ratio mass spectrometer, DELTA Plus Advantage (Thermo Fisher Scientific), connected to a universal on-line gas preparation and introduction system, GASBENCH II (Thermo Fisher Scientific). Data acquisition for all the IRMS analyses was based on the ISODAT software package (Thermo Fisher Scientific). The amount of H₃ ⁺ ions generated in the ion source, the so-called H₃ ⁺ factor, in the analysis of hydrogen isotope ratios was determined on ISOTAD and was <4.5 (ppm/mV) during the study. The analytical precisions for the ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{O}}_2}$, ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{H}}_4}$, ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$, and ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ values were estimated by repeated analyses of standard gases to be within 0.5‰, 0.3‰, 10‰, and 5‰, respectively. The determined δ values were calibrated with commercial and/or in-house standard gases as follows: −40.56‰ and −3.57‰ for ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{O}}_2}$, −74.01‰ and −39.03‰ for ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{H}}_4}$, −758.22‰ and −11.09‰ for ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$, and −185.9‰ for ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$. To minimize analytical errors, amplitudes of ion beams were matched between samples and standards by regulating amounts of the analyses introduced into IRMS. The ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ of the mediums were analyzed using a liquid water isotope analyzer (Los Gatos Research, Inc.) with an analytical precision of 0.5‰. The determined ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ values were calibrated with three commercial standard waters (−154.3‰, −96.4‰, and −9.5‰). Negligible memory effects were confirmed on the analyses that alternated between D-labeled and nonD-labeled samples. For the D-enriched batches, which had δD values far from the calibration ranges (particularly ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$), the analytical accuracy was presumably worse and the calculated ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values from the batches with D-enriched H₂O should be carefully considered.

Results

All the experimental results are illustrated in Fig. 1 and listed in Additional file 2. The results of each experiment are summarized separately below for each culture type.

Fig. 1

Temporal changes in H₂, CH₄, and CO₂ in four culture experimental sets. The experimental sets for the thermophilic coculture (a–f; TC), the mesophilic coculture (g–l; MC), the thermophilic pure culture (m–r; TP), and the mesophilic pure culture (s–x; MP) consisted of experimental batches A1–A6, B1–B6, C1–C6, and D1–D7, respectively. The pH₂ values (panels a, g, m, and s), the pCH₄ values (panels b, h, n, and t), the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values (panels c, i, o, and v), the ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ values (panels d, j, p, and v), the ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{O}}_2}$ values (panels e, k, q, and w), and the ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{H}}_4}$ values (panels f, l, r, and x) were determined for a certain interval of growth from immediately after inoculation to the stationary phase of growth. The symbols are common within each experimental set (TC, MC, TP, and MP)

Results of Thermophilic Coculture (TC)

The pH₂ values of the TC batches were 4.0–6.8 Pa immediately after the inoculation, increased during the early period of growth, and decreased toward the initial level at the later period (Fig. 1a). The highest pH₂ of each TC batch was somewhat different but within the same order of values (28.6–64.9 Pa). The timings of the pH₂ peaks were slightly variable between the batches (5–8 days). Note that the observed pH₂ values were close to the threshold at which hydrogenotrophic methanogens cease H₂ consumption (10⁰–10¹ Pa; Lovley 1985; Thauer et al. 2008), suggesting that our experiment covered the lower end of H₂ availability for microbial methanogenesis. The pCH₄ generally showed an exponential increment with the time of growth and reached 5.5–6.2 kPa (Fig. 1b). The increment of pCH₄ accompanied or followed that of pH₂ in each batch (5–8 days). This implies that the growth of hydrogenotrophic methanogens is triggered by the preceding metabolic function (H₂ production) and growth of the hydrogenogenic bacterial counterpart.

The ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values of TC batches were determined only when the pH₂ values were sufficiently high for the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ analysis (pH₂ of >10 Pa) (Fig. 1c). The ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values were virtually identical within batches that had almost identical initial ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ values regardless of pH₂ and time (Fig. 1c). The ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values ranged from −714‰ to −698‰, −231‰ to −213‰, and +374‰ to +413‰ for batches with nonlabeled ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ (A1 and A2), moderately labeled ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ (A3 and A4), and highly labeled ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ (A5 and A6), respectively (Additional file 2). All of these ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values were close to the theoretical ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ value at the isotopic equilibrium between H₂ and H₂O (${{\updelta \mathrm{D}}^{\mathrm{eq}}}_{{\mathrm{H}}_2}$: see Additional file 2). Our observation is consistent with previous studies examining biologically produced H₂ that also revealed ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values close to the equilibrium (Walter et al. 2012; Yang et al. 2012). Note that the measured ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}\hbox{--} {\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ relationship showed increased deviation from the theoretically expected relationship in batches with higher ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ values (deviations of <26‰ for A1 and A2, <42‰ for A3 and A4, and <100‰ for A5 and A6). The deviations were beyond the analytical precision of <10‰. Moreover, the deviations were possibly derived from the worse accuracy for the excessively D-enriched ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ measurement resulting from the over range of calibration by using the commercial standards having relatively lower isotopic values (see Methods) than the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ values of the experimental batches (Table 2).

The ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ values ranged from −356‰ to −366‰ in A1 and A2, +720‰ to +722‰ in A3 and A4, and +1984‰ to +2092‰ in A5 and A6 (Additional file 2). The ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ values were similar within batches that had similar initial ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ values regardless of pH₂ and time (Fig. 1d); this was also observed for the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values (Fig. 1c). The ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values calculated from the measured ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ and ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ values were in the ranges 0.673–0.683, 0.655–0.656, and 0.637–0.659 for the nonlabeled, moderately labeled, and highly labeled ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ batches, respectively (Fig. 2 and Additional file 2). There appeared to be a negative correlation between the ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ and ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ values. However, when the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ value calculated from the measured ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ value and ${\upalpha^{\mathrm{eq}}}_{{\mathrm{H}}_2\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ was used instead of the measured ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ value, the resulting ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values fell in a narrower range (0.668–0.697) regardless of the D₂O labels (Fig. 2 and Additional file 2). This ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ range is similar to the ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values obtained from the nonlabeled batches using the measured ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ value (0.673–0.683: A1 and A2, Additional file 2). These facts imply that the measured ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ values in the D-enriched batches (e.g., > +1000‰) would be overestimates of the actual values, probably due to the unreliable analytical accuracy in these cases that caused the underestimation of the ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ value.

Fig. 2

Relations between ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ and ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values in the two coculture experiments (TC and MC). For TC batches, the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ (and ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$) values calculated from the measured ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values and the H₂–H₂O isotope fractionation factor at the temperature are also illustrated for comparison

The ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{O}}_2}$ value for each of the TC batches increased with the time of growth from −23.5‰ to −11.5‰ (Fig. 1e). The ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{H}}_4}$ values were nearly constant throughout growth, between −86.3‰ and −82.0‰ (Fig. 1f). The resulting ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ values generally decreased with time from 0.936 to 0.925 (Fig. 3 and Additional file 2). The ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ decreases appeared to be synchronized with the pH₂ decrease (Fig. 4a and Additional file 2), which is harmonious to the $p{\mathrm{H}}_2{{\hbox{--} \upalpha}^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ relationship previously reported (Valentine et al. 2004; Penning et al. 2005; Takai et al. 2008).

Fig. 3

Relations between ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{O}}_2}$ and ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{H}}_4}$ values during growth in each of the four experiment sets (a–d). Symbols are the same as those in Fig. 1. The black arrow in each panel shows orders of the values along with the time of growth. Diagonal dashed lines indicate the theoretical isotope fractionation factors between CO₂ and CH₄

Fig. 4

The isotope fractionations of carbon (a) and hydrogen (b) with pH₂. The horizontal axis uses a logarithmic scale. Datasets are derived from the thermophilic coculture (TC), the mesophilic coculture (MC), the thermophilic pure culture (TP), and the mesophilic pure culture (MP) in this study

Results of Mesophilic Cocultures (MC)

The pH₂ values of the MC batches showed trends similar to those of the TC batches: an initial increase during the early phase of growth and a subsequent decrease to the initial level during the later period of growth (Fig. 1g). The highest pH₂ values of each MC batch were 6.8–10.7 Pa, approximately one-fifth lower than those of the TC batches. The pH₂ peaks occurred around 9–12 days after inoculation. The pCH₄ showed an exponential increase after the pH₂ peak and rose to 3.4–3.9 kPa at the end of incubation (29 days, Fig. 1h).

The ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values of MC batches were obtained from only three samples; B1-11 h, B3-12 h, and B4-12 h, because the pH₂ values of the other samples were below the detection limits of ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ measurement (Fig. 1i). The δDH₂ values of MC batches (−746‰ for B1-11 h, −357‰ for B3-12 h, and −394‰ for B4-12 h) were similar to those of H₂ isotopically equilibrated with the medium H₂O, as for the TC batches.

The ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ values of the MC batches were virtually identical through the growth in each of the batches except B6 (Fig. 1j). The ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ values ranged from −371‰ to −354‰ in B1 and B2, +560‰ to +589‰ in B3 and B4, +1934‰ to +1959‰ in B5, and +1608 to +1802‰ in B6, whereas reason(s) of the large variation in B6 remain unclear. The measured ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ and ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ values resulted in ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values of 0.671–0.687, 0.645–0.650, and 0.628–0.645 for the nonlabeled, moderately labeled, and highly labeled ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ batches, respectively (Fig. 2 and Additional file 2). As for the TC batches, the MC batches showed a negative correlation between ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ and ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ (Fig. 2). The ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values for the nonD₂O-labeled batches were similar between TC and MC (0.673–0.683 and 0.671–0.687, respectively) in spite of the different species, growth temperatures, and pH₂ conditions (Fig. 4b and Additional file 2).

The ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{O}}_2}$ values of the MC batches increased with the time of growth from −24.7‰ to −17.2‰ (Fig. 1k). The ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{H}}_4}$ values of the MC batches increased until 12 days and subsequently decreased through time. The δ¹³C_{CH 4} values of the MC batches ranged from −97.9‰ to −84.3‰ (Fig. 1l). The resulting ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ values of the MC batches generally decreased through time from 0.935 to 0.919 (Fig. 3b and Additional file 2). The ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ range of the MC batches overlapped with but was partly lower than those of the TC batches (Fig. 4a). The last samples of each batch showed the lowest ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ values (~0.920) and pH₂ (≤1 Pa), which is harmonious with the positive $p{\mathrm{H}}_2{{\hbox{-} \upalpha}^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ relationship.

Results of Thermophilic Pure cultures (TP)

The sampling and analysis for the TP batches were conducted with hourly frequency because of the rapid growth of Mtc. okinawensis under high pH₂ conditions (Fig. 1). The pH₂ of the TP batches decreased with time from 210 kPa to 94 kPa for one batch, C2, but to ~160 kPa in the other five batches (Fig. 1m and Additional file 2). The pCH₄ increased exponentially to 30 kPa in batch C2, but to ~12 kPa in the other five batches (Fig. 1n). The H₂ decrement and CH₄ increment roughly matched the stoichiometry in the net reaction of hydrogenotrophic methanogenesis (R1). The cell densities were ~4 × 10⁵ cells/mL at the time of inoculation and exponentially increased to ~5 × 10⁷ cells/mL within 6–15 h in all the batches (Additional file 2). Specific growth rates during the exponential growth phase were estimated to be 0.19 ± 0.05 h⁻¹ (n = 6). Cell-specific H₂ consumption rates during the exponential growth phase were estimated from the specific growth rate, pCH₄, and the methanogenic stoichiometry (R1) to be 0.14–1.6 fmol-H₂ cell⁻¹ s⁻¹ (n = 6). These values are within cell-specific H₂ consumption rates evaluated in incubation of Mtb. Thermautotrophicus under 10⁵ Pa-H₂ (0.1–58 fmol-H₂ cell⁻¹ s⁻¹), in which the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect was exhibited (Kawagucci et al. 2014).

The temporal changes in the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values showed different directions in the batches that had different initial isotope compositions of H₂ and H₂O (Fig. 1o and Table 2). For the batches that initially had isotopically unequilibrated conditions between H₂ and H₂O (batches C1, C2, C4, and C5), the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values unidirectionally shifted, either upward or downward, from the initial values toward the equilibrated values (Figs. 5a–c). In the cases of batches that initially had isotopically equilibrated conditions between H₂ and H₂O (batches C3 and C6), the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values changed little during growth (Fig. 5b and c). The difference between the theoretically expected ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values (equilibrated with ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$) and the measured ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values of the last samples were within 100‰ except for the batch with the highly DH-labeled H₂ (C4, Table 2) (~600‰).

Fig. 5

Relations between ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ and ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ in the two pure culture experiment sets (TP and MP). The results were divided into three subsets within each of the experimental sets according to the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ label conditions: nonD₂O-labeled (panels a and d); moderately D₂O-labeled (panels b and e); and highly D₂O-labeled conditions (panels c and f). The black arrow in each panel marks the trend in the temporal changes of the values. The horizontal dashed lines in each panel indicate the theoretical fractionation factors between CH₄ and H₂O. The vertical gray bar in each panel indicates the equilibrated ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ value calculated according to Horibe and Craig (1995)

The ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ values of two batches with highly DH-labeled H₂ (Table 2) decreased with the time of growth from −191‰ to −416‰ in C2 and +585‰ to +360‰ in C4 (Figs. 1p, 5a–c, and Additional file 2). The ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ value of the slightly DH-rich batch (C1) varied in a smaller range (−375‰ to −414‰) than the values of batches C2 and C4. The ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ value of a D₂O-rich batch (C5) and batches C3 and C6 also showed slight changes, which were initially in H₂–H₂O isotopic equilibrium, (Additional file 2). The measured ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ and ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ values resulted in ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values of 0.558–0.850 (Figs. 4b, 5a–c and Additional file 2).

The ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{O}}_2}$ values for each of the TP batches slightly increased with time of growth from −25.7‰ to −16.8‰ (Fig. 1q and Additional file 2). The ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{H}}_4}$ values ranged from −62.0‰ to −52.1‰ (Fig. 1r and Additional file 2). The ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ values ranged from 0.960 and 0.968 through growth (Fig. 3 and Additional file 2), which were higher than the values of the TC and MC batches (Fig. 4a).

Results of Mesophilic Pure cultures (MP)

The pH₂ values of MP batches decreased with time of growth from ~200 to 1.8–37 kPa (Fig. 1s). The pCH₄ values showed an exponential increase during the early stages of growth and finally plateaued after >30 days of incubation (Fig. 1t). The pCH₄ increment of the MP batches was more than one order of magnitude slower than that of the TP batches. Although the cell-specific H₂ consumption rate cannot be evaluated due to the lack of cell density information, the pCH₄ increment rate implies an H₂ consumption rate that is more than one order of magnitude slower than that of the TP batches (i.e., <~0.1 fmol-H₂ cell⁻¹ s⁻¹ for MP batches).

Temporal changes of ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values in the MP batches were variable between batches that had different initial isotope compositions of H₂ and H₂O (Fig. 1u and Table 2), as observed in the TP batches. The ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values of batches that initially had isotopically unequilibrated conditions between H₂ and H₂O (D1, D2, D3, and D6) (Table 2) shifted toward the equilibrated values (Fig. 1u). The ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values of batches with initially in isotopic equilibrium between H₂ and H₂O (D4, D5, and D7) showed little change during growth. The differences between the theoretically expected ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values (equilibrated with ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$) and the measured ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values of the last samples were less than 100‰ (Additional file 2).

The ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ values of the MP batches were at around −358‰ in D1, ranged from +536‰ to +581‰ in D2–D5, and were between +1663‰ and +1728‰ in D6 and D7 (Fig. 5d–f and Additional file 2), respectively. The ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ values were highly variable among all the batches but showed small variations between batches that had the same initial ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ values regardless of the time of growth (Fig. 1v). This result suggests that the ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ values depend mostly on the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ values. The ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ and ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ values resulted in ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values of 0.677–0.679 in D1, 0.647–0.654 in D2–D5, and 0.642–0.657 in D6 and D7 (Fig. 4b and Additional file 2).

The ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{O}}_2}$ values of the MP batches increased with time of growth from −24.9‰ to −5.7‰ (Fig. 1w). The ${\updelta}^{13}{\mathrm{C}}_{{\mathrm{C}\mathrm{H}}_4}$ values also increased from −67.0‰ to −48.0‰ with time (Fig. 1x). The resulting ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ values ranged from 0.955 and 0.968 (Fig. 3 and Additional file 2). The ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ values in both TP and MP were similar to each other, although the experiments were conducted using different methanogen species with different growth and metabolic kinetics and under different temperature conditions (Fig. 4a). This result suggests that the stable carbon isotopic fractionation of these hydrogenotrophic methanogens is regulated little with these factors.

Discussion

Carbon isotope fractionation of pure culture and coculture experiments

Comparing the four experimental sets with each other, the ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ values were markedly different between the cocultures (0.919–0.936: TC and MC) and the pure cultures (0.949–0.968: TP and MP) (Fig. 4a) in spite of same methanogen used in mesophilic cultures (MC and MP). The ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ values in our incubations sometimes overlap but are not always consistent with those at carbon isotope equilibrium (Fig. 4a; 0.933–0.943 at 25–60 °C: Horita 2001). All the ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ values obtained in this study (0.919–0.968) fall into the ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ range previously determined by experiments between 0.913 (Penning et al. 2005) and 0.989 (Takai et al. 2008) (Fig. 6). The ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ values show a positive correlation with the pH₂ condition (Fig. 4a), as determined in previous studies (Valentine et al. 2004; Penning et al. 2005; Takai et al. 2008). These facts allows us to regard the ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ value as a proxy for H₂ availability during hydrogenotrophic methanogenesis. As previous and current incubations examining the ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ values have covered a full spectrum of pH₂ conditions from the methanogenic threshold (10⁻¹–10¹ Pa: e.g., the MC batches in this study) to unrealistic enrichment (10⁷ Pa: Takai et al. 2008), the potential range of ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ variation corresponding to pH₂ change has been fully covered.

Fig. 6

Compilation of the isotope systematics of hydrogen (${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$) and carbon (${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$) associated with in hydrogenotrophic methanogenesis. Symbol colors of reddish, greenish, and bluish respectively represent datasets of laboratory incubations under high pH₂ conditions, incubations under low pH₂ condition and observations of “young” methane reservoirs in nature, and observations of “aged” methane reservoirs. Horizontal and vertical gray bars respectively represent the fractionation factors at the equilibriums of hydrogen (${\upalpha^{\mathrm{eq}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$) and carbon (${\upalpha^{\mathrm{eq}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{CO}}_2}$) at the indicated shown temperatures (0 °C, 122 °C, and 400 °C). All the data sources are listed in the main text (see Discussion)

Hydrogen isotope systematics of syntrophic consortium incubation

The cocultures experiments (TC and MC) exhibited ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values of 0.63–0.69 through the growth phase shift, although the experiments were conducted under different temperatures and with different syntrophic couples (Fig. 6). The ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values of 0.67–0.69 from the nonD₂O-labeled batches represent the most reliable range, as discussed above (Fig. 2). A slightly higher ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ range (0.71–0.74) was obtained in previous coculture experiments with the same hydrogenotrophic methanogenic consortium, S. lipocalidus and Mtb. thermautotrophicus at 55 °C (Yoshioka et al. 2008) (Fig. 6). In addition to these syntrophic consortiums reconstructed in the laboratory, similar ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values have been reported from experiments using potential H₂-syntrophic communities in natural environments such as in soil (0.68) (Sugimoto and Wada 1995), groundwater (0.66–0.73) (Hattori et al. 2012), and landfill waste (0.66–0.69) (Waldron et al. 1998) (Fig. 6). Note that there was considerable variation in ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ (0.66–0.74) in these studies compared to the typical analytical errors of ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ and ${\updelta \mathrm{D}}_{{\mathrm{CH}}_4}$ values (≤ ~ 10‰), corresponding to the ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ deviation of ≤ ~0.01. The reason why such variation of stable hydrogen isotope fractionation occurs in various potentially H₂-syntrophic communities is currently unknown. The hydrogen isotope equilibrium at methanogenic temperatures of our culture (0.775–0.809 at 25 °C – 60 °C: Horibe and Craig 1995) does not account for the ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values observed (Fig. 4b). Nevertheless, the ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ range of 0.66–0.74 probably represents the potential hydrogen isotopic signature of methane derived from hydrogenotrophic methanogenesis that occurs in H₂-limited habitats at methanogenic rates within the experimental time scale (typically shorter than a year).

Hydrogen isotope systematics of pure culture incubation

Stable hydrogen isotope fractionation in hydrogenotrophic methanogenesis (${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$) under H₂-enriched conditions (>10⁴ Pa-H₂) has been examined using pure cultures of methanogens (Additional file 1): 0.55–0.86 for Mtb. marburgensis (Valentine et al. 2004; Stolper et al. 2015); 0.60–1.69 for Mtb. thermautotrophicus (Yoshioka et al. 2008; Kawagucci et al. 2014; Wang et al. 2015); 0.528 for Methanosarcina barkeri (Stolper et al. 2015); 0.59–0.63 for Methanobacterium formicicum (Balabane et al. 1987); 0.64–0.68 for Methanobacterium sp. MO-MB1 (MP in this study); and 0.56–0.85 for Mtc. okinawensis (TP in this study) (Fig. 6). In addition to the pure cultures, natural groundwater was also incubated with 10⁵ Pa-H₂ supply that yielded ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values of 0.63–0.66 (Hattori et al. 2012). The variation of ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values from the H₂-enriched condition (0.55–1.69) is much greater than that from the H₂-limited cultures (0.66–0.74) (Fig. 4b). The high variability of ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values under high pH₂ conditions makes it difficult to estimate the relationship between the ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values and pH₂ conditions during hydrogenotrophic methanogenesis (Burke 1993; Sugimoto and Wada 1995; Valentine et al. 2004; Yoshioka et al. 2008; Hattori et al. 2012; Stolper et al. 2015; Wang et al. 2015). Instead, the large ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ variation seems to be regulated by some other factors than pH₂.

One possible factor is the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect: the stable isotope signature of the substrate H₂ affects that of the product CH₄ during methane production by rapidly growing hydrogenotrophic methanogens (Kawagucci et al. 2014). In fact, TP batches in this study clearly showed a positive correlation between ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ and ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ values during the early growth phase (Fig. 5), as also observed in a previous study of Mtb. thermautotrophicus (Kawagucci et al. 2014). In addition, the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect has been detected during reexamination of experimental datasets published in the literatures. For example, the ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ variation was observed during the growth of Mtb. marburgensis in a flow-through culture fed with D-enriched H₂ (${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ = −187‰) and slightly D-depleted H₂O (${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ = −93‰), and the highest ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ value of 0.86 was obtained from the early growth phase (Valentine et al. 2004). The lowermost ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ value of 0.59 was obtained during the growth of Mtb. formicicum in a batch culture fed with D-depleted H₂ (${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ = −728‰) and slightly D-enriched H₂O (${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ = +39‰) (Balabane et al. 1987). These results suggest that the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect may have a significant impact on the ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values under high pH₂ conditions. The ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect potentially leads to a great variation of ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values as a result of the contribution from ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ variation.

In contrast, it was notable that little ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect was exhibited in MP batches, which cultivated a hydrogenotrophic methanogen under high pH₂ conditions (Fig. 5a–c). This finding implies that the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect is not directly associated with the pH₂ conditions of growth. The hydrogen isotope fractionation in hydrogenotrophic methanogenesis is thought to be related to the methanogenic rate (Valentine et al. 2004; Stolper et al. 2015). As the MP batches reached the stationary growth phase after several weeks, which is slower than the typical growth rates of other pure culture methanogens at >10⁴ Pa-H₂ with favorable growth conditions (timescales of hours to days: Additional file 1), the δDH₂ effect may instead be related to the metabolic and growth kinetics of methanogens.

As discussed in previous studies (Burke 1993; Sugimoto and Wada 1995; Kawagucci et al. 2014), the incorporation of intracellularly produced H⁺ from the H₂ into CH₄ may drive the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect. From this viewpoint, a previous study that cultivated Mtb. thermautotrophicus under high-pH₂ conditions attempted to yield a quantitative understanding of the magnitude of the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect (Kawagucci et al. 2014). We can know only two quantitative pieces of information relating to the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect: the cell-specific H₂ consumption rate (i.e., the intracellular H⁺ production rate) of 10⁻¹–10² fmol/cell/s and the quantity of intracellular H₂O, which is 10¹–10² fmol-H₂O. If the H₂ consumption rate is comparable with that trans-membrane H₂O and H⁺ flux, which remains unknown, the intracellular ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ value should be affected by H₂-derived H₂O (H⁺) that results in exhibition of the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect. Moreover, if the trans-membrane H₂O flux is constant regardless of the cell growth situation, the slower H₂ consumption rate would lead to a smaller ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect. It would be reasonable for a significant ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect to be apparent in the TP batches and Mtb. thermautotrophicus batches reported previously (Kawagucci et al. 2014), in which the H₂ consumption rate was rapid (10⁻¹–10² fmol/cell/s), whereas little ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect was detected in the MP batches, in which the H₂ consumption rate was slow (presumably < ~10⁻¹ fmol/cell/s). This finding seems consistent with the decrement in magnitude of the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect along with the growth phase shift from exponential growth to the stationary phase (Kawagucci et al. 2014).

Comparison between methanogenic experiments and environmental methane

To compare the experimental results with natural environments, we here review hydrogen and carbon isotope fractionations that have been obtained from environmental observations in addition to the incubations (Additional file 1). As the timescale of methanogenesis is a probable key factor controlling hydrogen isotope fractionation, as aforementioned, the compiled dataset of natural isotope fractionations is classified by considering the timescales of natural methane reservoirs (Fig. 6). Methane in the bovine rumen, one of the natural environments in which methane generation occurs as rapidly as in laboratory incubations, showed the ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ range between 0.68 and 0.74 (Burke 1993; Wang et al. 2015). The pH₂ condition of the rumen was estimated at ~10² Pa (Burke 1993; Wang et al. 2015), higher than the values exhibited in our cocultures (<65 Pa: Fig. 1) and the methanogenic threshold (<10¹ Pa: Lovley 1985; Thauer et al. 2008). Terrestrial sediments in freshwater systems of Caltech pond and Swamp Y (Stolper et al. 2015; Wang et al. 2015) possessed ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values of ~0.69. The ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values in these environmental samples from relatively “young” methane reservoirs (0.68–0.74) are consistent with those obtained from the regulated laboratory cocultures of hydrogenotrophic methanogens (≤0.74). Note that aceticlastic and/or methylotrophic methanogenesis may also occur in the wild environment.

Hydrogen and carbon isotope fractionations have also been investigated in geological methane reservoirs: natural gas fields (Nakai et al. 1974; Whiticar et al. 1986); continental groundwater at the Elk Valley coalbed methane field (Aravena et al. 2003); terrestrial groundwater associated with the accretionary prism in south-east Japan (Kimura et al. 2010); deep-sea sediments (Whiticar et al. 1986); and subseafloor methane hydrate deposits from the Northern Cascadia Margin (Wang et al. 2015), the Umitaka Spur, and the Joetsu Knoll of the Japan Sea (Hachikubo et al. 2011) (Fig. 6). We hereafter methane in geological reservoirs is named as “aged” methane, for comparison with “young” methane in laboratory cultures and natural environments, as aforementioned. Almost all of these aged samples show a ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ range of 0.91–0.95, suggesting that hydrogenotrophic methanogenesis under H₂-limited conditions is plausible as the geochemical origin of these methane. Almost all of the ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values of the aged methane samples fall within a narrow range of 0.79–0.84 (Fig. 6). The ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values are not consistent with the values obtained from the H₂-limited experiments and young environmental samples (≤0.74; Fig. 6) in spite of the commonly low ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ values (≤0.95).

Possible mechanisms exhibiting variable isotope fractionation

We define three major provinces in the compilation of isotope fractionations (Fig. 6) and infer the metabolic mechanisms functioning to yield each of the provinces as follows (Fig. 7). The first province has an ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ range higher than 0.95 with highly variable ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values (Fig. 6). This province has occurred in >10⁴ Pa-H₂ culture but never observed in any natural samples from biologically functionable temperature environments (≤122 °C). This finding is reasonable because such high-pH₂ environments are rare in nature, except for several H₂-enriched geofluid systems (Charlou et al. 2002; Proskurowski et al. 2006; Ishibashi et al. 2014). The high ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ value in >10⁴ Pa-H₂ cultures has been thought to be a result of an almost straightforward progress of carbon reduction from CO₂ to CH₄ in a multistep methanogenic pathway (Valentine et al. 2004; Penning et al. 2005; Fig. 7a and b). The large ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ variation probably results from the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect (see discussion above), and the magnitude of the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect appears to be linked to the H₂ consumption rate (Fig. 7a and b). The ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ shift, probably due to H₂–H₂O isotope exchange toward the isotope equilibrium by reversible function of hydrogenase, suggests that H₂ reproduction functioned even during the almost straightforward methanogenesis (Fig. 7b).

Fig. 7

Cartoons of intra- and trans-membrane fluxes of the molecules in hydrogenotrophic methanogenesis relating to hydrogen and carbon isotope systematics. Each five panel represents methanogenic characteristics with respect to growth rate and pH₂ condition for the growth. Red arrows represent significant molecular fluxes relating with CH₄ isotope ratios. Light green and blue arrows represent significant molecular fluxes not relating with CH₄ isotope ratios. Grey arrows represent negligible molecular fluxes. Dash-dotted circles are cell membrane, and arrows enclosed by dotted lines represent multistep methanogenic reaction (numbers of the arrows have no meaning). A term “H₂O” in cartoon represents both H₂O and H⁺. See main text for details

The second province has an ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ range lower than 0.95 and an ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ range of 0.67–0.74 (Fig. 6). This province has been found in both laboratory cultures of <10⁴ Pa-H₂ conditions and also in natural reservoirs of relatively young methane such as the bovine rumen and terrestrial sediments. The low ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ values are thought to result from the extent of reversibility of the methanogenic pathway under low-pH₂ conditions (Valentine et al. 2004; Fig. 7c). The narrow ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ range is probably caused by a combination of the higher reversibility and the slower H₂ consumption rate. The slow H₂ consumption would alter the intracellular ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ value little and leave it the same as the environmental (extracellular) ${\updelta \mathrm{D}}_{{\mathrm{H}}_2\mathrm{O}}$ value (Fig. 7c). The repeated formation and cleavage of the C–H bonds of methane precursors through the reversible methanogenic pathway lead to hydrogen isotope equilibrium between the precursor and intracellular water. Note that the ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values of this province (0.67–0.74) are out of the ${\upalpha^{\mathrm{eq}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ range between 0 °C and 400 °C (Horibe and Craig 1995). This finding suggests that some kinetic processes are also involved in the hydrogen isotope fractionation (Fig. 7c).

The third province has an ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ range lower than 0.95 and an ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ range higher than 0.80 (Fig. 6). The third province overlaps with the second province in ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ values but differs in the ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ value. This province has been observed typically in geological reservoirs of aged methane such as natural gas and seafloor methane hydrate. It was thought that methanogenesis at an extremely slow rate could produce the third province because of isotopic equilibrium between H₂O, H₂, CO₂, and methane precursors through a highly reversible methanogenic pathway (Stolper et al. 2015; Wang et al. 2015; Fig. 7d). Unfortunately, we cannot examine this possibility experimentally because microbial methanogenesis at a rate as slow as geological timescale is not reproducible by experiments. In addition, we are aware of the ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ variation in spite of the narrow ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ range within the third province (Fig. 6). The ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ variation may be inconsistent with this explanation because methanogenesis through the geological timescale is expected to occur under constantly low pH₂ conditions and slow growth rate, resulting in the constantly low ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ value of approximately 0.92.

We here hypothesize that the higher ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values of the third province may be caused by a post-methanogenesis process over geological timescales (Fig. 7e) rather than the methanogenesis through geological timescale (Fig. 7d). This process is diagenetic hydrogen isotope exchange between extracellular CH₄ and H₂O promoted by the reversible function of the multistep methanogenic pathway in methanogenic region and/or methane reservoirs after the migration (Fig. 7e). Multiple lines of circumstantial evidence based on biological and geochemical viewpoints can be cited to put forward the possibility of the diagenetic hydrogen isotope exchange as following. First, the microbial methanogenic pathway is also able to operate reversibly for methane consumption; e.g., it is known that anaerobic methanotrophic archaea can oxidize CH₄ via potential reverse methanogenesis (Knittel and Boetius 2009) and the methanogens can perform trace methane oxidation even during active methanogenesis (Moran et al. 2005). In addition, trans-membrane exchange of CH₄ would be not negligible in geological timescale. Thus, the reversible enzymatic reactions, in particular at the later steps of the methanogenesis, i.e., cleavage and reproduction of C–H bonds of CH₄, could result in promoting the hydrogen isotope exchange between extracellular CH₄ and H₂O (Fig. 7e). Second, the ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values in the third province (0.79–0.86) overlap with the ${\upalpha^{\mathrm{eq}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values in the known biologically-functional temperature range (0.758–0.843 at 0–122 °C: Horibe and Craig 1995) (Fig. 6). Third, a one-year hydrothermal experiment at 323 °C (Reeves et al. 2012) revealed negligible hydrogen isotope exchange between CH₄ and H₂O. This finding suggests that hydrogen isotope exchange is extremely sluggish in the habitable temperature range without effective catalysts such as enzymes in the reversible methanogenic pathway. Fourth, the third province is only detected in the aged methane samples obtained from both marine and terrestrial environments (Fig. 6) while relatively young methane samples such as from the rumen and laboratory experiments are never located in the third province. This age-isotope relationship is also found in a “clumped” isotope composition, i.e., the abundance of ¹³C–D bonds in CH₄, which may yield information about the temperature at which the C–H bonds were formed or last equilibrated (Stolper et al. 2014, 2015; Ono et al. 2014; Wang et al. 2015). The clumped isotope composition so far observed in aged methane reservoirs indicates the CH₄–H₂O equilibrated signature, whereas clumped signatures that cannot be explained by expected formation temperature have been found frequently in young biogenic methane from natural reservoirs and laboratory incubations (Stolper et al. 2014, 2015; Wang et al. 2015). Fifth, the diagenetic hydrogen isotope exchange is not in conflict with the broad ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ range of the third province. If a methane reservoir incorporates multiple methane sources, each of which has distinct isotope signatures, the diagenetic hydrogen isotope exchange could unify the ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values regardless of the sources but leave the ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ value as a mixed signature because of incomplete molecular equilibrium between CO₂ and CH₄ (Fig. 7e). Such diagenetic unification of the hydrogen isotope ratio while maintaining the ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ variation is evident in deep-sea, high-temperature hydrothermal fluids (Fig. 6: Proskurowski et al. 2006; McCollom 2008; Kawagucci et al. 2013a; 2013b; Wang et al. 2015). Methane in the hydrothermal fluids so far observed in various geological settings shows a broad ${\upalpha^{\mathrm{C}}}_{{\mathrm{C}\mathrm{H}}_4\hbox{--} {\mathrm{C}\mathrm{O}}_2}$ range (0.953–0.992) but a narrow ${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ range (0.88–0.90), which are close to the ${\upalpha^{\mathrm{eq}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ values at the endmember fluid temperatures (~0.87 at 300 °C–400 °C: Horibe and Craig 1995).

These pieces of circumstantial evidence put forward that the hydrogen isotope signature of CH₄ imprinted at the time of hydrogenotrophic methanogenesis (${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ = 0.66–0.74), belonging to second province, can be altered by the reversible reactions of the multistep methanogenic pathway, which leads to CH₄–H₂O isotopic equilibrium at habitable temperatures (${\upalpha^{\mathrm{H}}}_{{\mathrm{CH}}_4\hbox{--} {\mathrm{H}}_2\mathrm{O}}$ > 0.76) belonging to the third province. If this is true, the hydrogen isotope ratio of methane in geological samples would be useless as a tracer to deduce the origin of methane. However, it is also possible to criticize this hypothesis: the diagenetic isotope exchange proposed here is not required when methanogenesis through geological time actually exhibits isotope fractionation in the third province, as aforementioned. Furthermore, an in vitro biochemical study of methyl coenzyme M reductase activity, which is an enzyme that catalyzes the last step of methanogenesis, showed that little hydrogen isotope exchange between CH₄ and H₂O occurred during the enzyme reaction for several tens of minutes (Scheller et al. 2013). It is still uncertain whether diagenetic hydrogen isotope exchange can be promoted by methanogens or not and how rapid the isotope exchange is if it occurs. To solve these uncertainties, experiments using methanogenic cells concentrated in mediums with isotopically disequilibrated CH₄ and H₂O under thermodynamic states favorable to each of methane generation and methane consumption will be conducted. That work will provide experimental justification of the possible post-methanogenesis isotope diagenesis hypothesis.

Conclusions

This study confirmed five facts as follows.

• 1. Carbon isotope fractionation between CH₄ and CO₂ during hydrogenotrophic methanogenesis is correlated with pH₂, as suggested in previous studies (Valentine et al. 2004; Penning et al. 2005; Takai et al. 2008).

• 2. The hydrogen isotope ratio of CH₄ produced by a thermophilic methanogen, Mtc. okinawensis, under high pH₂ conditions (~10⁵ Pa-H₂) is affected by the isotope ratio of H₂, as pointed out in a previous incubation of Mtb. thermautotrophicus (Kawagucci et al. 2014). This pattern is named as the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect. This effect also appears to account for the diverse hydrogen isotope fractionation between CH₄ and H₂O previously observed in H₂-enriched culture incubations (Balabane et al. 1987; Valentine et al. 2004; Stolper et al. 2015; Wang et al. 2015).

• 3. A mesophilic methanogen, Methanobacterium sp. MO-MB1, showed little ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect even under high-pH₂ conditions (~10⁵ Pa). This result suggests that methanogenic and growth rates rather than the pH₂ condition are significant factors controlling the ${\updelta \mathrm{D}}_{{\mathrm{H}}_2}$ effect.

• 4. The hydrogen isotope fractionation between CH₄ and H₂O in hydrogenotrophic methanogenesis under low pH₂ conditions (<10² Pa) observed in experiments is in the range of 0.66–0.74.

• 5. Hydrogen isotope fractionation exhibited in laboratory incubations under low-pH₂ conditions is consistent with that observed in “young” methane reservoirs but inconsistent with that observed in “aged” methane reservoirs.

In conclusion, we propose that diagenetic hydrogen isotope exchange between extracellular CH₄ and H₂O catalyzed by a reversible methanogenic pathway of methanogenic populations alters the hydrogen isotope signature of CH₄ imprinted at the time of generation.