Hydrogen Does Not Appear To Be a Major Electron Donor for Symbiosis with the Deep-Sea Hydrothermal Vent Tubeworm Riftia pachyptila

Despite the presence of hydrogenase genes, transcripts, and proteins in the “Ca. Endoriftia persephone” genome, transcriptome, and proteome, it does not appear that R. pachyptila can use H₂ as a major electron donor. For many uncultivable microorganisms, omic analyses are the basis for inferences about their activities in situ. However, as is apparent from the study reported here, there are dangers in extrapolating from omics data to function, and it is essential, whenever possible, to verify functions predicted from omics data with physiological and biochemical measurements.

ABSTRACT

Use of hydrogen gas (H₂) as an electron donor is common among free-living chemolithotrophic microorganisms. Given the presence of this dissolved gas at deep-sea hydrothermal vents, it has been suggested that it may also be a major electron donor for the free-living and symbiotic chemolithoautotrophic bacteria that are the primary producers at these sites. Giant Riftia pachyptila siboglinid tubeworms and their symbiotic bacteria (“Candidatus Endoriftia persephone”) dominate many vents in the Eastern Pacific, and their use of sulfide as a major electron donor has been documented. Genes encoding hydrogenase are present in the “Ca. Endoriftia persephone” genome, and proteome data suggest that these genes are expressed. In this study, high-pressure respirometry of intact R. pachyptila and incubations of trophosome homogenate were used to determine whether this symbiotic association could also use H₂ as a major electron donor. Measured rates of H₂ uptake by intact R. pachyptila in high-pressure respirometers were similar to rates measured in the absence of tubeworms. Oxygen uptake rates in the presence of H₂ were always markedly lower than those measured in the presence of sulfide, as was the incorporation of ¹³C-labeled dissolved inorganic carbon. Carbon fixation by trophosome homogenate was not stimulated by H₂, nor was hydrogenase activity detectable in these samples. Though genes encoding [NiFe] group 1e and [NiFe] group 3b hydrogenases are present in the genome and transcribed, it does not appear that H₂ is a major electron donor for this system, and it may instead play a role in intracellular redox homeostasis.

IMPORTANCE Despite the presence of hydrogenase genes, transcripts, and proteins in the “Ca. Endoriftia persephone” genome, transcriptome, and proteome, it does not appear that R. pachyptila can use H₂ as a major electron donor. For many uncultivable microorganisms, omic analyses are the basis for inferences about their activities in situ. However, as is apparent from the study reported here, there are dangers in extrapolating from omics data to function, and it is essential, whenever possible, to verify functions predicted from omics data with physiological and biochemical measurements.

INTRODUCTION

Hydrogen gas (H₂) is an electron-donating substrate for microbial energy generation in a variety of chemosynthetic habitats, such as mangroves, shallow sediments, hydrothermal vents, hydrocarbon seeps, and whale falls (1–3). At vents, H₂ can reach micromolar concentrations up to tens to hundreds, depending on the site (4–6). In these environments, H₂ is used by methanogens, acetogens, and a diversity of other microbes for energy generation (7–9).

Hydrogenases mediate the reaction H₂⇌ 2 H⁺ + 2e⁻ and are found across all three domains of life (10). Hydrogenases are classified according to their catalytic metal centers: the [FeFe] hydrogenases, the [NiFe] hydrogenases, and the [Fe] hydrogenases, the last being found only in some methanogenic archaea. The three classes differ with respect to gene organization, location within the cell, and function. Many of these enzymes are known to be bidirectional, but in general the [FeFe] hydrogenases are involved in H₂ evolution, whereas the [NiFe] hydrogenases are involved in H₂ consumption (10, 11). Studies in the last few years have shown that some hydrogenases are capable of electron bifurcation and are even involved in extracellular electron transfer (12–14). However, the physiological function of hydrogenases in many organisms remains putative due to their limited representation among cultivated microbes, the modular nature of how hydrogenases are assembled, and the fact that the direction of electron flow through these enzymes is dependent on the electrochemical potential near their catalytic sites (10, 15–17).

Symbioses between chemoautotrophic bacteria and eukaryotic hosts are ubiquitous in many of the ecosystems in which H₂ is found (18). Among most of these symbioses, inorganic electron donors and acceptors are used to regenerate ATP and reducing equivalents for microbial growth and biosynthesis. Moreover, the symbionts assimilate inorganic carbon and “fix” it into metabolic precursors (e.g., simple organic acids) that are typically translocated to the host (19–22). The animals, in turn, use this carbon for their own bioenergetic and biosynthetic needs. The efficacy of these associations is apparent in the abundance of symbioses in many chemosynthetic habitats; indeed, carbon fixation by these associations contributes substantially to net primary productivity at vents (23).

To date, many studies have focused on the use of hydrogen sulfide (here referred to as ΣH₂S) by the symbionts of these associations (24), but more recently the palette of electron donors has grown to include H₂ (1, 25). Symbionts of the mussel Bathymodiolus puteoserpentis have been shown via genomic analyses, environmental geochemical measurements, and H₂-dependent carbon fixation to definitively use hydrogenases for harnessing energy from H₂ oxidation (1). Transcriptomic and proteomic studies have even found that the symbionts of the giant hydrothermal vent tubeworm Riftia pachyptila express putative hydrogenase genes and proteins, respectively (2, 26–28). R. pachyptila flourishes at deep-sea hydrothermal vents in the Eastern and Southeastern Pacific and is a foundational species that supports high productivity rates and provides habitat for many other vent fauna. R. pachyptila is a siboglinid worm that forms an obligate symbiosis with “Candidatus Endoriftia persephone” (27, 29–31), a member of Gammaproteobacteria. R. pachyptila lacks a mouth, gut, and anus and derives most, if not all, of its nutrition from its symbionts. Early studies of R. pachyptila tested the utilization of multiple potential electron donors, including H₂, but it was observed that only ΣH₂S stimulated carbon fixation under their test conditions. That said, earlier studies were limited to measuring substrate uptake and carbon fixation in excised symbiont-containing tissues incubated at 1 atm (32). The importance of ΣH₂S to R. pachyptila’s symbionts was further underscored by the abundance of sulfide-binding proteins in the worm’s blood (33, 34), as well as the presence of elemental sulfur among the symbionts (35, 36). These observations, along with the measured stimulation of carbon fixation in the presence of ΣH₂S, emphasized the role of ΣH₂S as the symbiont’s primary electron donor (23, 37, 38).

Nevertheless, the more recent transcriptomic and proteomic studies call into question the relevance of H₂ to symbiont function. H₂ is present in the hydrothermal fluid bathing R. pachyptila in situ, at concentrations ranging up to ∼25 μM (4, 39). In basalt hosted vent sites where R. pachyptila is found, the majority of the available catabolic energy comes from ΣH₂S; nevertheless, H₂ oxidation does occur and represents ∼5 to 20% of total available energy in the fluid surrounding R. pachyptila (4). Supplementation of sulfide-driven primary productivity with H₂ use by R. pachyptila could be an important aspect of this symbiosis and could plausibly contribute to R. pachyptila’s remarkable growth rates, which, in turn, lead to its flourishing at high densities around vents and playing a key role in structuring the vent ecosystem. The aforementioned presence of putative hydrogenases and the recognition that earlier studies lacked the ability to replicate in situ conditions led us to reevaluate the possible relevance of H₂ oxidation to the energy metabolism of the R. pachyptila symbiont.

Therefore, the purpose of this study was to determine if the R. pachyptila symbionts could use H₂ as an electron donor for chemolithoautotrophy. With that in mind, we conducted live-animal and symbiont-containing tissue incubations with or without H₂, comparing rates of O₂, ΣH₂S, and H₂ consumption, as well as carbon fixation. In addition, hydrogenase assays were performed to measure enzyme activities in symbiont-containing tissues, and the presence and transcription of hydrogenase genes from the symbiont genomes were noted.

RESULTS

Respirometry.

H₂ consumption in aquaria with R. pachyptila was not significantly different from that in control aquaria without R. pachyptila (Fig. 1). The two control aquaria (H₂ with low and high O₂ concentrations) were aggregated for this comparison, as no significant difference was found between them using a Welch two-sample test (α = 0.05). The least square (LS) means of H₂ uptake between aquaria with R. pachyptila and the aggregated aquaria without tubeworms (control aquaria) were calculated using a generalized least square model in R, and these least square means were compared using a pairwise-adjusted Holmes test, with an α value of 0.05 (40). O₂ uptake rates in aquaria supplied with H₂ were lower than in the case of treatments with ΣH₂S. This was true for both high and low O₂ concentrations (Fig. 2).

FIG 1.

H₂ uptake by R. pachyptila compared to vessels without R. pachyptila (control), calculated using the least square (LS) means of the difference in H₂ concentrations in the seawater entering and leaving the pressure vessels. For the control, vessels without R. pachyptila were supplied with H₂ gas (∼100 μM) with either low (∼30 μM) or high (∼250 μM) O₂ concentrations. Data from these control vessels were aggregated, as no significant difference of uptake was found between them. Furthermore, no significant difference was found in H₂ uptake when comparing control vessels to those containing R. pachyptila (α = 0.05).

FIG 2.

Least squares means of O₂ uptake rates by R. pachyptila incubated in vessels in the presence of H₂ or ΣH₂S. Letters indicate data that differed statistically significantly (α = 0.05).

Carbon fixation by R. pachyptila symbioses maintained in high-pressure aquaria with either H₂ or ΣH₂S as an electron donor.

Dissolved inorganic carbon (DIC) incorporation into biomass (i.e., carbon fixation) by R. pachyptila incubated in the presence of H₂ gas was 5- to 23-fold lower than in treatments in which R. pachyptila was given adequate ΣH₂S and O₂ (Fig. 3 and 4). The rates of carbon fixation for both H₂ treatments were indistinguishable from those measured when R. pachyptila was maintained under low-O₂ conditions (Fig. 4).

FIG 3.

LS means of inorganic carbon incorporation rates (C_inc) by R. pachyptila incubated in vessels in the presence of H₂ or ΣH₂S. Means were computed from data from individuals in Fig. 4. Letters indicate data that differed statistically significantly (α = 0.05).

FIG 4.

Inorganic carbon incorporation rate (C_inc) for individual R. pachyptila tubeworms incubated in the presence of H₂ or ΣH₂S with low or high O₂ concentrations.

Dissolved inorganic carbon incorporation by trophosome samples.

Trophosome samples from an R. pachyptila individual maintained in the presence of H₂ did not have elevated carbon fixation rates relative to that of the control (Fig. 5). Homogenized trophosome samples from this individual had similar rates of carbon fixation when incubated in the presence (0.034 ± 0.004 nmol of CO₂ min⁻¹mg of protein) or absence (0.037 ± 0.002 nmol of CO₂ min⁻¹mg of protein) of H₂. Carbon fixation was stimulated solely when ΣH₂S was added (0.067 ± nmol min⁻¹mg of protein [Fig. 5]).

FIG 5.

CO₂ fixation by homogenized trophosome samples in the absence (-) and presence of 0.16 mM H₂ or 0.11 mM H₂S.

When regressing carbon fixed against incubation time, inclusion of indicator variables for electron donors significantly reduced the residual sum of squares (F = 87; P < 0.001). Values of β₃ and β₅ from the full model (see Materials and Methods) were statistically distinguishable from zero (P = 0.003 and 0.04, respectively), indicating that ΣH₂S significantly stimulated carbon fixation, while values of β₃ and β₅ were not, indicating that the amount of carbon fixed when H₂ was added could not be distinguished from the amount fixed in its absence.

Hydrogenase activity.

Neither freshly collected R. pachyptila worms nor those that had been maintained in high-pressure aquaria with H₂ had measurable hydrogenase activity in trophosome samples (Table 1). Assay mixtures of some of the R. pachyptila samples grew slightly more turbid over the 5-min assay time course, which is reflected in the small negative values for these samples. Hydrogenase activity was apparent in Hydrogenovibrio thermophilus MA2-6; the presence of H₂ significantly increased the rate of methylene blue reduction (P < 0.001).

TABLE 1.

Hydrogenase activities of trophosome samples and cultivated bacteria

Species	Specimen conditions before harvest	Assay headspace gas	Activity (mmol of H2 min−1 mg of protein−1)a	nb
H. thermophilus MA2-6	Batch culture with H2	H2	0.39	1c
	Batch culture with H2	Ar	0.12	1
	Batch culture with H2	H2	1.48	1
	Batch culture with H2	Ar	0.00	1
R. pachyptila 1d	Freshly collected	H2	−0.001 ± 0.002	3
R. pachyptila 2	Freshly collected	H2	−0.006 ± 0.007	2
R. pachyptila 3	Freshly collected	H2	0.000 ± 0.003	2
R. pachyptila 4	Aquaria with H2	H2	−0.006 ± 0.005	2
R. pachyptila 5	Aquaria with H2	H2	−0.004 ± 0.003	2
R. pachyptila 6	Aquaria with H2	H2	−0.001 ± 0.004	2

^aValues are means ± standard deviations.

^bn, number of times activity was measured for a particular culture or individual.

^cFor the H. thermophilus MA2-6 samples, values measured in the presence of H₂ gas were statistically different (P < 0.001) from those measured in the presence of Ar.

^dR. pachyptila 1 to 6 are six different individuals.

Hydrogenase genes in the “Ca. Endoriftia persephone” genome.

The “Ca. Endoriftia persephone” genomes from R. pachyptila, Tevnia jerichonana, and Ridgeia piscesae include genes homologous to those for four types of hydrogenase (IMG gene object identifier numbers 2600441592, 2600441595, 2600436887, and 2601634699). Using the classification and E values calculated by comparing these sequences to those in HydDB, these include [FeFe] group C1 (E value < 4e⁻⁵), [FeFe] group C3 (E value < 7e⁻⁵), [NiFe] group 1e Isp type (as previously described in reference 41; E value = 0), and [NiFe] group 3b (E value = 0) (Fig. 6).

FIG 6.

Gene synteny near putative group 1e and 3b [Ni Fe] hydrogenase genes among “Ca. Endoriftia persephone.” The genes depicted include IMG gene object identifier (ID) numbers 2600441718 to 2600441727 and 2600441763 to 2600441768 (T. jerichonana), 2600436877 to 2600436887 and 2600438609 to 2600438602 (R. pachyptila vent Ph05), and 2601634007 to 2601633997 and 2601634696 to 2601634703 (R. pachyptila vent Mk28).

Abundances of hydrogenase transcripts.

Transcriptome sequencing with Novaseq resulted in an average of ∼62 million reads per sample, with ∼36 million left after trimming. Between 3 and 27% (average 15%) of these reads pseudoaligned to “Ca. Endoriftia persephone” coding DNA sequence (CDS) regions. Hydrogenase genes from type 1e and 3b were transcribed in all samples (Fig. 7; see also Table S1 in the supplemental material).

FIG 7.

Transcripts per million (tpm) encoding type 1e (left) and 3b (right) hydrogenase subunits and associated proteins for RNA extracted from trophosome samples from R. pachyptila incubated in the presence of H₂ and ΣH₂S under low- and high-O₂ conditions.

Differences in the relative abundances of transcripts encoding type 1e hydrogenase were apparent in two of the treatment comparisons: (i) when O₂ was high, these transcripts were more abundant in the presence of H₂ than in the presence of ΣH₂S, and (ii) when ΣH₂S was present and H₂ was absent, these transcripts were more abundant under low-O₂ conditions than under high-O₂ conditions. Transcript abundances for the type 3b hydrogenase were also elevated under the second set of conditions (Table 2).

TABLE 2.

Differential abundance of transcripts encoding hydrogenase subunits and regulatory proteins in “Ca. Endoriftia persephone”

Hydrogenase type	Predicted proteina	β (qval)d
		H2/ΣH2Sb		Low O2/high O2c
		High O2	Low O2	H2	ΣH2S
Type 1e	HypD	1.12 (4.0 × 10−3)			0.80 (2.1 × 10−2)
	1e maturation protein				1.28 (3.0 × 10−2)
	HyaF	2.73 (4.8 × 10−4)	1.53 (4.60 × 10−2)		2.30 (6.5 × 10−4)
	1e small subunit	3.74 (4.8 × 10−23)			3.73 (8.7 × 10−35)
	Isp1	3.66 (5.9 × 10−8)			3.34 (7.7 × 10−17)
	Isp2	3.37 (4.7 × 10−8)			3.50 (1.3 × 10−29)
	1e large subunit	3.10 (5.3 × 10−20)			3.17 (4.3 × 10−12)
Type 3b	4Fe-S				0.82 (2.0 × 10−2)
	NAD(P)H-flavin reductase				1.11 (4.5 × 10−2)
	3b small subunit				1.26 (2.4 × 10−2)

^aProteins on the 1e and 3b contig that did not show any significant differential transcript abundance are not included.

^bComparison of transcript levels in the presence of H₂ (numerator) or ΣH₂S (denominator), under either low or high O₂ concentrations.

^cComparison of transcript levels in the presence of low (numerator) or high (denominator) concentrations of O₂, when either H₂ or ΣH₂S was provided.

^dβ is the log₂ fold change relative to the treatment indicated in the denominator of the compared treatments (a/b). qval is the P adjusted value with a cutoff of 0.05.

DISCUSSION

H₂ does not appear to be a major electron donor for the R. pachyptila symbiosis. Intact R. pachyptila maintained at quasi-in situ conditions does not use significant quantities of H₂. Moreover, the presence of H₂ does not stimulate O₂ consumption or carbon fixation by intact R. pachyptila or in excised and incubated trophosome samples. Furthermore, hydrogenase activities, specifically H₂-dependent methylene blue reduction, were not measurable in trophosome samples.

Hydrogenase genes are likely to encode active hydrogenase enzymes and are transcribed and expressed.

The absence of evidence for H₂ utilization by this symbiosis is puzzling given the presence of genes encoding multiple hydrogenase enzymes in the “Ca. Endoriftia persephone” genome. The two [NiFe] hydrogenase large subunit genes include signature sequences suggesting catalytic activity of their protein products as hydrogenase. The amino acid sequence predicted from the gene encoding the [NiFe] group 1e hydrogenase large subunit has matches to the L1 and L2 signature sequences present in other group 1e members (11). The sequence predicted from the gene encoding the [NiFe] group 3b hydrogenase also has matches to L1 and L2 signature sequences from group 3b, but these matches are not exact. However, they do include the two CXXC motifs of the active site of this enzyme (11). The gene neighborhoods of these [NiFe] hydrogenase large-subunit genes provide further evidence for hydrogenase activity. Both large-subunit genes are colocated with genes encoding hydrogenase small subunits (41) (Fig. 6). For the group 1e hydrogenase genes, genes encoding accessory proteins necessary for their assembly are also nearby (41). Genes encoding the two [NiFe] hydrogenases are indeed expressed in “Ca. Endoriftia persephone” while present in hospite (2, 26–28), and the presence of transcripts of genes encoding both 1e and 3b hydrogenases is consistent with expression in hospite (Table 2 and Fig. 7). Furthermore, when differences in transcript abundance are apparent, transcripts encoding all four subunits of the type 1e hydrogenase are upregulated by similar amounts, suggesting that this enzyme is assembled as predicted, though protein abundances were not measured via proteomic or Western blot analyses.

While two potential [FeFe] hydrogenase genes from “Ca. Endoriftia persephone” are likely to share evolutionary history with group C1 and C3 [FeFe] hydrogenases, these two genes lack sequence signatures L1Fe, L2Fe, and L3Fe conserved among other [Fe Fe] hydrogenases and implicated in active-site architecture (11), raising doubt that the proteins encoded by these genes are catalytically active as hydrogenases.

Potential reasons for lack of evidence for H₂ use as a major electron donor by the R. pachyptila symbiosis.

It is possible that hydrogenase activity was present but not measureable in our assay, in which methylene blue was used as an electron acceptor. Many (42) but not all (43) hydrogenases are capable of reducing methylene blue. It is also possible that reduced methylene blue was not stable in crude cell extract prepared from trophosome samples. However, the possibility that substantial uptake hydrogenase activity was present in these samples seems unlikely, given that the presence of H₂ does not stimulate metabolic activities of intact R. pachyptila (Fig. 1 and 4) or “Ca. Endoriftia persephone” (Fig. 5).

Another reason for an absence of apparent use of H₂ as a major electron donor, despite the presence of hydrogenase-encoding genes, in “Ca. Endoriftia persephone” may be that this microorganism differentially expresses them based on its lifestyle. “Ca. Endoriftia persephone” is environmentally transmitted between generations of R. pachyptila (44, 45) and therefore spends some time as a free-living microorganism. It may be possible that this microorganism can use H₂ as a major electron donor when it is free-living but not when it is symbiotic with R. pachyptila.

It is also possible that these [NiFe] hydrogenases function primarily to maintain intracellular redox homeostasis in hospite. Both [NiFe] group 1e and 3b enzymes are reversible, capable of H₂ production (11, 46). Some group 3b hydrogenases couple the oxidation of NADPH to evolution of H₂ (11, 47) and are upregulated under hypoxia to maintain an optimal balance of reduced and oxidized cofactors, such as NAD⁺/NADH (48). Transcript levels are consistent with hydrogenase playing a role in redox homeostasis in “Ca. Endoriftia persephone.” Under high-O₂ conditions, transcript abundances were higher when H₂ was present, but transcript abundances were also elevated when ΣH₂S was the sole electron donor provided and O₂ concentrations were low (Fig. 7 and Table 2). In both cases, the symbionts may have been responding to a lack of (i) reductant or (ii) oxidant, which supports the hypothesis that these hydrogenases may play a role in redox homeostasis.

Conclusions and perspectives.

Altogether, the lack of physiological response of “Ca. Endoriftia persephone” to the provision of H₂—both in hospite and in vitro—is not consistent with the hypothesis that H₂ is an electron donor for energy metabolism. More specifically, the provision of H₂ does not lead to increased oxygen demand, nor does it stimulate carbon fixation. In contrast, previous studies have consistently shown that the provision of sulfide, for example, both increases oxygen utilization and stimulates carbon fixation. The genomic and transcriptome analyses are consistent with the presence of biochemically active hydrogenases, which may play a role in intracellular redox homeostasis. Therefore, hydrogenase activity would not likely stimulate carbon fixation, nor would it need to be present at high activity levels necessary if H₂ was a major source of electrons for autotrophy; this could explain why the presence of this gas did not measurably stimulate symbiont CO₂ fixation.

Previous omics work had strongly suggested that the ecosystem-structuring R. pachyptila symbiosis could use H₂ as an energy source/electron donor. This study casts doubt on this possibility, and also provides a cautionary tale about extrapolating from omics data to system function/physiology. This is quite timely given the growing importance of omics data, particularly for the as-of-yet-uncultivable majority of ecologically and environmentally relevant microorganisms.

MATERIALS AND METHODS

R. pachyptila collection.

R. pachyptila worms were collected during an expedition in October of 2016 on board the RV Atlantis to the hydrothermal vents at the 9°N Integrated Study Site in the East Pacific (9°50′N, 104°18′W, 2,500 m). R. pachyptila worms were collected via the human-occupied vehicle (HOV) Alvin during dives to the sites Tica and Bio 9. Live worms were stored in a thermally insulated container during submersible transit to the surface and were sorted in ice-cold seawater or dissected immediately upon recovery.

High-pressure respirometry system.

The high-pressure respirometry system (HPRS) resides in a 20-foot refrigerated intermodal shipping container. This system allows for the incubation of hydrothermal vent animals at temperatures, pressures, and water chemistries that reflect in situ conditions. This was accomplished by using Aalborg mass flow controllers to bubble gases (H₂, H₂S, CO₂, and O₂, with N₂ as a carrier gas) through 0.2-μm-filter-sterilized seawater contained within 1-m-tall gas equilibration columns to recreate the chemical conditions in situ. This gas-amended seawater was then fed to high-pressure pumps (Lewa America, Inc.) which delivered this fluid into custom-built 3-liter titanium high-pressure aquaria in which the tubeworms were kept. Pressure was maintained in the aquaria with stainless steel back-pressure/relief valves (Straval, Inc.) that allow for a flowthrough rate of ∼50 ml·min⁻¹ at a constant pressure of ∼27.5 MPa (Fig. 8).

FIG 8.

Schematic of the high-pressure respirometry system. Selected gases (e.g., H₂, H₂S, CO₂, O₂, and N₂) are bubbled into equilibration columns that feed high-pressure pumps, which, in turn, pump fluid into aquaria. Pressure is controlled via back-pressure/relief valves which allow for a flowthrough system.

Incubation conditions and respirometry.

R. pachyptila worms that were quick to withdraw into their tubes when lightly touched were selected for shipboard incubations in the HPRS. Four to six tubeworms were placed into each high-pressure aquarium and were provided with either H₂ or ΣH₂S as an electron donor, at high and low O₂ concentrations (Table 3). To measure consumption of dissolved gases, water samples were collected from the equilibration columns that supplied the aquaria (intake, or pre-aquarium sampling) and from the back-pressure valve located downstream of the aquaria (outake, or post-aquarium sampling). Dissolved H₂ was measured with a Unisense H₂ minisensor 500 (range = 0 to 800 μM; detection limit = 0.3 μM) encased in a glass flowthrough cell that was placed in-line with the intake or outtake of the aquaria. O₂ was measured using a PreSens oxygen FTC-SU-PSt3-S flowthrough cell (range = 0 to 1,400 μM). ΣH₂S (the sum of the sulfide species H₂S + HS⁻ + S²⁻) was qualitatively measured aboard ship by colorimetric analysis (49; LaMotte sulfide test kit). Subsamples were also preserved in a 2 mM zinc acetate solution and shipped to the lab, where they were stored at –80°C until analysis. For lab-based ΣH₂S quantification, the frozen samples were thawed and vortexed, and 100 μl was pipetted into a 96-well clear-bottom plate. Eight microliters of reagent A (LaMotte sulfide test kit) and 2.5 μl of reagent B were added per sample, mixed, and aliquoted into a 96-well plate ∼1 min before the absorbance was read at 670 nm with a Spectramax i3 plate reader.

TABLE 3.

Incubation conditions for R. pachyptila maintained in the high-pressure respirometry system

Treatmenta	H2 (μM)	ΣH2Sb (μM)	O2 (μM)	nc	Total biomass (g)	Biomass range (g)	Duration (h)
H2 + high O2	114 ± 5	0	272 ± 42	5 (0)	177	31–65	64
H2 + low O2	113 ± 12	0	31 ± 5	5 (1)	222	29–40	64
ΣH2S + high O2	0	148 ± 33	291 ± 11	6 (0)	48.14	2.4–16	52
ΣH2S + low O2	0	159 ± 34	32 ± 8	4 (1)	66.7	6.0–33.7	52

^aDissolved gas concentrations are the averages of measurements taken from the equilibration columns every 4 h ± standard deviations.

^bΣH₂S = [H₂S] + [HS⁻] + [S²⁻].

^cThe number of R. pachyptila worms placed in the aquarium is followed in parentheses by the number that expired during the experiment.

Upon cessation of experiments, aquaria were slowly depressurized and tubeworms were removed, weighed on a motion-compensated shipboard balance (50), and dissected in a cold room (∼4°C). Tissue samples were taken from the trophosome (symbiont containing) and the plume, vestimentum, and skin (nonsymbiont containing) for isotope incorporation analysis, RNA preservation, and enzyme assays. Samples used for trophosome incubations were processed immediately onboard the ship, as were those for transcript analysis (see below); those used for isotopic analyses and enzyme assays were freeze-clamped and maintained at –80°C until analysis.

Uptake rates for H₂ and O₂ were calculated using the formula below, where total biomass represents the total weight of the worms without their tubes in each aquarium. Intake and outtake are micromolar concentrations, total mass is in grams, and flow rate is in liters per hour.

$$\left(\frac{\text{intake}-\text{outtake}}{\text{total biomass}}\right)\times \text{flow rate}$$

Dissolved inorganic carbon incorporation by intact R. pachyptila.

To measure the incorporation of dissolved inorganic carbon (DIC, which is the sum of the species [CO₂], [HCO₃], and [CO₃²⁻]) by intact R. pachyptila in the presence of H₂ or sulfide, a 99% NaH¹³CO₃ stock solution was added to the intake tank to achieve a final isotopic composition of 2.64% ¹³C (of DIC) in the aquarium water. Fluid samples were taken via gastight syringe at the inlet and outlet of the high-pressure aquaria containing the worms, as well as the control vessels that were run without worms. These subsamples were gently filtered through a 0.2-μm cellulose acetate sterile syringe filter, and the filtrate was gently injected into a preevacuated gastight sampling vial (12-ml Exetainer; Labco Inc.). All samples were stored at 4°C and, upon return to the laboratory, were analyzed at the Yale Analytic and Stable Isotope Center (YASIC) on a Deltaplus XP stable isotope ratio mass spectrometer. Posttreatment, worms were quickly dissected and samples of trophosome, plume, and skin were weighed on a microbalance and then frozen at –80°C for later analyses. In the lab, these tissues were lyophilized with a freeze dryer (FreeZone 2.5; Labconco Inc.), acidified with 1 M HCl for ∼1 min, rinsed with distilled, deionized water, and dried in a vacuum oven (Labconco Inc.) at 50°C until weights were stable over time, indicating complete drying. Dried samples were then pulverized with a glass mortar and pestle and sent to Boston University Stable Isotope Laboratory, where the samples were weighed into tin boats with a microelectronic balance, combusted in a CN analyzer (Eurovector Inc.), separated chromatographically, and run through a GV Instruments IsoPrime isotope ratio mass spectrometer. ¹³C/¹²C isotope ratios were measured using the international standards of NBS 20 (Solenhofen Limestone), NBS 21 (spectrographic graphite), and NBS 22 (hydrocarbon oil). δ¹³C_V-PDB (per mille) was calculated using the equation below, where R is the atomic ratio (¹³C/¹²C).

$${\text{δ}}^{\text{13}}\text{C}=\left(\frac{R_{\text{sample}}}{R_{\text{standard}}}-1\right)\times \mathrm{1,000}$$

The rates of DIC incorporation into tissues were calculated from ¹³C incorporation using a modification of the method in reference 51. The atomic percentages (A%) of labeled and natural abundance samples were calculated from δ¹³C values as

$$A\mathrm{\%}=\left(\frac{R_{\text{sample}}}{R_{\text{sample}}+1}\right)\times 100$$

The percentage of ¹³C that was incorporated into biomass (¹³C_inc) was calculated as

$${}^{13}{\text{C}}_{\text{inc}}=\left(\frac{A{\mathrm{\%}}_{\text{lab}}-A{\mathrm{\%}}_{\text{nat}}}{A{\mathrm{\%}}_{\text{wat}}-A{\mathrm{\%}}_{\text{nat}}}\right)$$

modified from reference 51, where A%_lab is the atomic percentage of tissue that was incubated in the presence of ¹³C, A%_nat is the atomic percentage of tissue that was not exposed to the isotopic label, and A%_wat is the atomic percentage of DI¹³C available in the fluid to which it was exposed. The percent ¹³C_inc was multiplied by the dry weight and the percent total carbon of the sample to calculate the weight of the ¹³C incorporated (W¹³C_inc). From this weight of ¹³C incorporated, a carbon incorporation rate was calculated as described in reference 51.

Dissolved inorganic carbon incorporation by trophosome samples.

To determine whether H₂ is used by the symbionts as an electron donor for energy generation, homogenized trophosome was incubated in the presence of this gas and carbon fixation was measured radiometrically. To prepare for this assay, intact R. pachyptila worms were incubated in high-pressure respirometers in the presence of H₂ for 2 days (see above). Immediately after R. pachyptila worms were removed from the high-pressure aquaria, trophosome was excised, homogenized, and drawn into three glass syringes primed with buffered R. pachyptila saline and [¹⁴C]bicarbonate as described in reference 52. Dissolved gas concentrations in the syringes consisted of 0.18 mM O₂ and 10 mM DIC; one syringe also had 0.1 mM ΣH₂S, and one had 0.1 mM H₂. Incubations were maintained at 15°C and stirred with magnetic stir bars. Carbon fixation was assayed over 25 min at 5-min intervals. One-hundred-microliter samples were added to 100 μl of 1.2 N HCl to stop the reaction and sparged with air for 2 h to remove remaining ¹⁴CO₂. To minimize quenching, samples were dried and then resuspended in 200 μl of ScintiGest tissue solubilizer (Fisher Chemical) and 20 μl of H₂O₂. After 20 min, 3 ml of ScintiVerse BD cocktail (Fisher Chemical) was added. The following day, radioactivity was measured using a scintillation counter.

A linear regression with indicator variables was used to test whether an electron donor (H₂, ΣH₂S, or none) affected the amount of carbon fixed by the symbionts (IBM SPSS Statistics 24). An F value was calculated from the residual sum of squares from a reduced model (y = β₀ + β₁x₁, where y = carbon fixed and x₁ = time) and a full model (y = β₀ + β₁x₁ + β₂x₂ + β₃x₃ + β₄x₁·x₂ + β₅x₁·x₃), where x₂ and x₃ are indicator variables for incubations with H₂ and ΣH₂S, respectively.

Hydrogenase assay.

A spectrophotometric assay was used to measure hydrogenase activity in the H₂-oxidizing direction via methylene blue reduction (53, 54). For a positive control, Hydrogenovibrio thermophilus MA2-6 was batch cultivated under 5% H₂, 5% CO₂, and balance air, harvested via centrifugation (5,000 × g, 4°C, and 10 min), and stored at –80°C. Assay buffer (0.1 mM methylene blue, 20 mM KH₂PO₄ [pH 7] with KOH) was sparged overnight with argon in a glove bag. On the day of the assay, crude cell extracts were prepared from H. thermophilus and R. pachyptila trophosome samples by thawing them in the glove bag and disrupting them with a mortar and pestle in assay buffer. Samples (0.1 ml) and 0.9 ml of assay buffer were sealed in glass cuvettes with septa. After injecting the appropriate headspace (H₂, or argon for negative controls), cuvettes were agitated and removed from the glove bag, and methylene blue reduction was monitored at 601 nm (ε = 7,000 mol⁻¹cm⁻¹). Protein concentrations in the assays were measured with an RC DC protein assay (Bio-Rad, Inc.). For the positive control, to determine whether the presence of H₂ affected the rate of methylene blue reduction, a linear regression with indicator variables was used as described above for dissolved inorganic carbon incorporation by trophosome samples.

Bioinformatics.

Genes encoding hydrogenase and hydrogenase-associated proteins were collected from “Ca. Endoriftia persephone” genome data (R. pachyptila host [27, 28] or Ridgeia piscesae and Tevnia jerichonana hosts [55]; available through NCBI and the JGI Integrated Microbial Genomes & Microbiomes system). Hydrogenase sequences were classified using HydDB (https://services.birc.au.dk/hyddb/) (56).

Abundances of hydrogenase transcripts.

Trophosome tissue was quickly excised from three responsive tubeworms from each treatment for mRNA transcript abundance analyses. These tissues were homogenized with a stainless steel rotor-stator Tissue-Tearor (Cole Palmer) in 1 ml of TRIzol reagent and maintained at −80˚C. Total RNA was extracted using Direct-zol RNA MiniPrep (Zymo Research), following the manufacturer’s instructions. Normalized total RNA was sent to the Microbial ‘Omics Core (MOC) at the Broad Institute for library preparation and sequencing. After barcoded adaptor ligation and rRNA depletion, cDNA libraries were constructed from 0.5 to 1 μg of total RNA, using a modified RNAtag-seq protocol (57). In short, after reverse transcription, an adaptor was added to the 3′ end of the cDNA by template switching using SMARTScribe (Clontech) (58). cDNA was sequenced with a 2 × 33- to 75-bp paired-end protocol with the Illumina Novaseq 6000 platform. After sequencing, transcripts were quality assessed using FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/), overrepresented reads were removed (https://github.com/harvardinformatics/TranscriptomeAssemblyTools), and reads were quality trimmed using TrimGalore (https://github.com/FelixKrueger/TrimGalore). The abundance of transcripts from each sample was pseudoaligned against the published coding regions (CDS) of the genome for “Ca. Endoriftia persephone” (28) with kallisto (59).

Four comparisons of transcript abundances in trophosome samples were undertaken in R using sleuth (60): (i) R. pachyptila incubated under high-O₂ conditions with H₂ versus ΣH₂S, (ii) R. pachyptila incubated under low-O₂ conditions with H₂ versus ΣH₂S, (iii) R. pachyptila incubated with H₂ under high-O₂ versus low-O₂ conditions, and (iv) R. pachyptila incubated with ΣH₂S under high-O₂ versus low-O₂ conditions. These comparisons were done with the Wald test to calculate β values, using log-transformed data so that the β values represent an estimate of log₂ fold change between treatments. Values of P were adjusted for false discovery using the Benjamini-Hochberg method (61), with a significance cutoff value for P adjusted (qval) of ≤0.05.