Sample collection was performed during the NT13–15 cruise aboard the R/V Natsushima during July 2013. The study site, the Joetsu Knoll, is a well-characterized location of methane seepage offshore the Joetsu City, Niigata Prefecture, Japan (Figure 1; 37°31.1′N, 137°58.0′E, 985 mbsl). Two methane hydrates-containing sediment cores were collected during Dive 1555 of the ROV Hyper-Dolphin. Sediment was collected into the HP-Core (internal volume: ~920 mL, maximum pressure to use: 50 MPa) by abrading the internal core cylinder against an exposed vertical wall of sediment interlaced with white methane hydrates (Figure 1). The HP-Core's internal cylinder (label “1” in Figure 2b) was then immediately placed into the external cylinder (label “3” in Figure 2b) and secured by tightening (using label “2” in Figure 2b). In this manner, sediment was collected at environmentally relevant pressure and sealed into the HP-Core in situ, in order to maintain pressure throughout core recovery and onshore experimentation (see Figure 1 for in situ sampling photographs; see Figure 2a for arrangement of HP-Core on the ROV Hyper-Dolphin payload). During this deployment, we discovered that the Teflon seal on the HP-Core was compromised, most likely by small sediment grains lodged against the Teflon core liner which resulted in a partial loss of pressure (from 9.9 Mpa down to ~2 MPa) during transit of the ROV Hyper-Dolphin from seafloor to the R/V Natsushima. The core was immediately re-pressurized up to 10 MPa onboard ship by injection of 0.2 μm-filtered artificial seawater; specifically, we dissolved the following components into 1 L of MilliQ water (Merck Millipore, Massachusetts, USA): 0.14 g KH₂PO₄, 0.54 g NH₄Cl, 3.05 g MgCl₂·6H₂O, 0.11 g CaCl₂, 20.45 g NaCl, and 10 mL trace element solution (1.5 g nitrilotriacetic acid, 3.0 g MgSO₄·7H₂O, 0.59 g MnCl₂·4H₂O, 1 g NaCl, 0.1 g FeSO₄·7H₂O, 0.18 g ZnSO₄·7H₂O, 0.01 g CuSO₄·5H₂O, 0.02 g AlK(SO₄)₂·12H₂O, 0.01 g H₃BO₃, 0.007 g Na₂SeO₃·H₂O in 1 L of MilliQ H₂O). The HP-Core apparatus was stored at 4°C and 10 MPa onboard, during shipment, and upon arrival at the KCC laboratory. Besides the HP-Core, a second core was collected from adjacent hydrate-containing sediment into a traditional M-type corer (hereafter, “M-Core”). The material collected into the M-Core contained a mixture of sediment and bottom water, which by the time of recovery onboard ship had separated by density. Immediately onboard ship, subsamples of the “M-Core water” and “M-Core sediment” were frozen at −80°C for later DNA extraction and sequencing.

A unique and hence important function of the HP-Core system is the ability to supply liquid and gas substrates while maintaining the constant pressure in the chamber. In this study, the HP-Core was kept for the duration of experimentation (total 45 days) in a walk-in 4°C refrigerator in the laboratory (Figures 2c,d). Twelve days after collection from the seafloor, the HP-Core was amended with ¹³CH₄ (50 mL of 50% ¹³CH₄) and ¹⁵N₂ (50 mL of 50% ¹⁵N₂) and daily tracking of pressure, temperature, dissolved inorganic carbon concentration (DIC), and δ¹³C_DIC began for the course of a 45-day experiment in the HP incubation of seafloor microbial assemblages (Figures 3, 4). Temperature and pressure were continuously monitored (Δt = 1 s), with daily samples taken for δ¹³C_DIC. During daily sampling, pressure was kept at 10 MPa by injection of sterile artificial seawater that contained no carbon sources (see Supplementary Text). Samples for microbial community analysis were taken from the sediment water slurry at 11, 25, and 45 days (hereafter, T11, T25, and T45, respectively). Samples collected during the time course experiment at T11 and T25 consisted of water from the effluent outflow at the top of the HP-Core. This involved bleeding 6 mL of effluent from the top port, followed by filtration onto a 0.2 μm-pore sized polycarbonate membrane and freezing at −80°C. The final T45 time point consisted of sediment at the bottom of the HP-Core during the takedown of the experiment. ¹³CH₄ and ¹⁵N₂ were injected at the start of the HP incubation experiment followed by periodic additions of 100% O₂ on days 29, 30, 35, 37, 39, and 44 to stimulate aerobic methane-oxidation (Figure 4). In all cases 10 mL (at room condition) of O₂ was injected, which corresponds to the O₂ concentration in situ (~210 μmol/kg, Gamo, 2011), with the exception of the first injection on day 29, which was half this volume.

Carbon concentration and isotopic measurements were conducted on 0.2 μm-filtered effluent water samples <24 h after collection. Measurements were performed on an isotope-monitoring gas chromatography/mass spectrometry (irm-GC/MS) Thermo Finnigan Delta Plus XP isotope-ratio mass spectrometer connected to TRACE GC as previously described (Ijiri et al., 2012).

DNA was extracted from M-Core sediment and water samples using the MoBio PowerMax soil DNA isolation kit, according to manufacturer protocols (~5 g slurry/extraction). The T11, T25, and T45 (duplicate samples of T45 were extracted and sequenced) time points were extracted with the MoBio PowerSoil DNA isolation kit, according to manufacturer protocols (~0.5 g/extraction). In addition, duplicate T45 sediments were separately subjected to a simplified hot alkaline DNA extraction (Morono et al., 2014), in which sequential cell lysis is performed in heated 1M sodium hydroxide solution (Supplementary Text).

Samples were prepared for iTag-sequencing of the V4 region of the 16S rRNA gene, according to a slightly modified version of the Earth Microbiome Project's recommended protocol (see Mason et al., 2015 for protocol modifications). New England Biolabs Q5 polymerase enzyme was additionally substituted for 5-PRIME Hot Master Mix. Sequencing was performed on an Illumina MiSeq platform at Laragen, Inc. Culver City, CA, and data processing [joining paired ends, trimming sequences, chimera checking, 97% operational taxonomic unit (OTU)-picking, and taxonomic assignment] were performed as previously described (Case et al., 2015). Nonmetric multidimensional scaling (NMDS) analyses were performed in the R environment using the “vegan” package on square-root-transformed tables of relative sequence abundance (Oksanen et al., 2013; R Core Team, 2014).

In addition to sequencing of 16S rRNA genes, an assay of the monooxygenase intergenic spacer region (“MISA”) between two methane mono-oxygenase genes (pmoC and pmoA) was performed following previously described protocols (Tavormina et al., 2010; see Supplementary Text for modified primer sequences) on two samples: M-Core sediment and T45 sediment. Transformation of the MISA fragment into E. coli was performed with the 10G Elite Solo kit (Lucigen Inc., Culver City, CA). Inserts were amplified with the Lucigen Corporation GC Vector Amplification pSMART kit and separately digested with Hae III and Rsa I restriction enzymes in order to generate restriction fragment length polymorphism (RFLP) patterns. Unique inserts were sequenced at Laragen, Inc. The resulting traces were manually checked for quality, translated to amino acid sequences, aligned against pure culture and previously published pmoA fragments in MUSCLE (Edgar, 2004), and trimmed to the pmoA amino acid positions 5–49 of M. capsulatas Bath (an approach employed in Tavormina et al., 2010). These pmoA fragments, both experimental and from known organisms, were used to generate a 100-bootstrap, maximum likelihood tree in RAxML (Stamatakis, 2014).

The 16S rRNA gene sequence data have been submitted to the NCBI BioProject database under the BioProject ID of PRJNA416818. The pmoA and pmoC sequence data are available in the GenBank database with the accession numbers of MG149702-MG149775.

Of the five goals planned for the first deep-sea deployment of the HP-Coring device, all were successful, with the exception of partial depressurization during core recovery. The first goal, to develop a HP-Core deployable on the submersible payload, was achieved by deployment with the ROV Hyper-Dolphin. Furthermore, the relatively small HP-Core footprint enabled simultaneous deployment of other sampling devices during the dive [e.g., 6 push cores, a plastic tote for recovery of push cores previously deployed on the seafloor, a temperature probe and deep-sea high-pressure CO₂ injection system (Ohtomo et al., 2015), and the “M-type” corer], demonstrating that use of the HP-Coring device can be incorporated with other ROV-based scientific objectives and does not require an exclusive deployment.

Our second goal was to collect a core >10 cm in length using the HP-Core. Due to the nature of the hydrate environment at the Joetsu Knoll, we did not deploy the HP-Core like a traditional vertical push core at the seabed, we instead abraded the HP-Core against a wall of methane hydrate interspersed with sediments and bacterial mats. Abrading the HP-Core against the wall of hydrate and sediment allowed us to visualize the glass-made corer during sampling and gauge how the device was responding during the first deployment of this new technology. Using this approach enabled recovery of a significant amount of sediment, similar to the amount captured in a traditional push core.

Unfortunately, the third technical goal, to maintain in situ HP conditions through recovery onboard ship, was not met during the NT13–15 cruise. The HP-Core arrived at the sea surface having lost ~8 MPa of pressure down to ~2 MPa, believed to be the result of small sediment grains that compromised the Teflon seal where the core interfaces with the core liner. Pressure was immediately restored with filtered seawater upon recovery, and held stably during incubation at sea, transport and over the course of the 45 days-experiment in the lab, demonstrating its resilience to shipping and handling.

Goals four and five were to demonstrate the addition of liquid and gas amendments to the incubator under pressure during the HP-incubation experiment, and to be capable of extracting time-resolved output samples. Successful sampling from the outflow port for geochemical analysis was performed daily. Sampling resulted in a minor loss of internal pressure (generally <1 MPa, dependent on the user's skill level; Figure 3). Pressure was restored after sampling by pumping in fresh, sterile artificial seawater. Additionally, gas-phase amendments were injected in-line with seawater replacement throughout the incubation.

Over the course of 45 days, δ¹³C_DIC was observed to increase after ¹³CH₄ addition. The first ~30 days showed slow progressive enrichment ¹³C_DIC (Figure 4; for DIC concentration, see Supplementary Figure 1). A model of exponential increase in daily δ¹³C_DIC values fit from the data between T29 and T32 (R² = 0.97; Figure 4), matches the data from the later time interval T33–T45 very well; and this increase appears to be associated with the direct injection of O₂ beginning at T29. Stepwise rates of methane oxidation at 10 MPa were calculated by subtracting the moles of ¹³C observed between time points (t_n-t_n−1), based on the assumption that new ¹³C in the DIC pool represented newly oxidized ¹³CH₄ (Equation 1; a factor of 2 was added because the methane amendment was only 50% ¹³CH₄). The trend in methane oxidation rate, by definition, mirrors the increase in δ¹³C_DIC and shows increasing rates of methane oxidation late in the HP-incubation experiment.

Within the initial period of the experiment prior to rapid increase in methane oxidation rate (T0–T29), slow but measurable methane oxidation was observed (c.f. inset of Figure 4), with higher rates (increase in δ¹³C) over the first 10 days relative to days T11–T29.

The microbial diversity was analyzed by 16S rRNA iTag-sequencing from the background M-core, collected adjacent to the HP-core, and from different time points in the HP-Core incubation, including effluent samples from T11 and T25, and the final HP-Core sediment samples from T45 (Figures 5, 6). The M-core samples, including both sediment and overlying water, were characterized by high relative abundances of OTUs-associated with Candidate Division JS1 bacteria (20–30%), Desulfobacteraceae (6–10%), Methylococcales (2–12%), and ANME archaea (1–4%); (Figure 5; Supplementary Data). Analysis of the liquid effluent samples sampled from the pressure incubation experiment during T11 and T25 did not detect OTUs associated with ANME archaea or Methylococcales. Instead, these samples showed high relative abundance of Helicobacteraceae (9–17%), as well as other OTU's associated with Epsilon-, Delta-, and Gammaproteobacteria and Bacteroidetes (Figure 5). Sediment samples collected at the termination of the HP-core incubation (T45) showed harbored much of the same diversity of Epsilon-, Delta-, and Gammaproteobacteria observed in the HP-Core effluent samples, with the exception of the recovery of JS1 OTUs (2–4%) and a high percentage of Methylococcales. had a high relative abundance of OTU's associated with Pisirickettsiaceae (16–20%), which was a minor component of the M-Core and HP-Core effluent samples (Figure 5). The extraction method (MoBio power soil kit vs. Hot Alkaline lysis) appears to contribute a minor difference in the overall microbial 16S rRNA signature recovered from T45 sediments. When extracted with the MoBio kit, a Methylococcales-associated OTU is recovered at about two-thirds the relative abundance as recovered in samples extracted with the Hot Alkaline method. In contrast, a BD1-5-associated OTU is twice as abundant in MoBio-extracted samples as compared to Hot Alkaline-extracted samples (Figure 5). These differences are apparent in multidimensional ordination, where the T45 samples overall plot closely together, but are distinctly separate according to extraction method (Figure 6). This variation may be due to differential lysis, as has been reported in other studies (Morono et al., 2014).

Analysis of putative methanotrophic bacterial diversity was also assessed using the particulate monooxygenase-targeted MISA assay targeting both particulate methane monooxyganse (pmoA) and ammonium monooxygenase (amoA) genes (Tavormina et al., 2010). Using the MISA assay, we screened the overlying water from the M-Core and T45.1-MoBio sediment, and detected diverse pmoA sequences with a lower proportion of amoA sequences, with some overlap in MISA fragments recovered between the in situ M-Core and HP-Core after the 45-day incubation. Both samples showed a high relative abundance of a pmoA sequence putatively associated with gammaproteobacterial methanotrophs (Patterns 1 and 16 in Figure 7). Overall the pmoA sequences recovered in both samples were more similar to one another than to other pmoA genes reported from other sites or from cultured organisms (Figure 7).

During the 45-day incubation, DIC and δ¹³C_DIC data (collected and analyzed daily) suggested methane oxidation (Figure 4). In the first 11 days (T0–T11) of the experiment methane oxidation, as determined by incorporation of ¹³C into the DIC pool, appeared to be accelerating. However, it then plateaued and between T11 and T28 little methane oxidation was observed. We suspect the methane oxidation at the start of the incubation was associated with aerobic methanotrophy which ceased when O₂ was fully consumed. The theoretical amount of O₂ consumed by aerobic methanotrophy can be calculated by stoichiometric conversion using the following equation and ¹³C_DIC data to track the number of moles of methane consumed between T0 and T11:

With 4.25 μmol/kg of CH₄ consumed between T0 and T11 (calculated from data in Figure 4), corresponding oxidation of 8.50 μmol/kg of O₂ is required. This is a relatively small amount compared to known bottom water O₂ concentrations in the deep-sea (>220 μmol/kg), but it is likely that through the course of shipment of the HP-Core to KCC and static storage for 12 days at 4°C prior to ¹³CH₄ addition, oxygen levels may have been depleted in the incubation chamber relative to in situ concentrations from aerobic respiration (e.g., via sulfide oxidation or organic carbon) Respiratory processes could have continued during T0–T11, all independent of ¹³C-label and thus undetected by our geochemical measurements.

Between T11 and T28, the rate of methane oxidation was slower, with 7 μmol/kg of CH₄ oxidized during the 7 days. If we assume that during this period methane-oxidation there was a shift to anaerobic oxidation of methane coupled to sulfate (AOM; Equation 3), then only 7 μmol/kg of sulfate are stoichiometrically required:

This is well within the bounds of seawater chemistry, where sulfate is present at ~28 mmol/kg. Anaerobic conditions during this period were supported by oxidation-reduction potential (ORP) measurements. When the first ORP measurement was taken, at T29, it was at the very reduced value of −300 mV. Although we do not have ORP data to help define exactly when anaerobic conditions began during the incubation, it is clear that by T29 anaerobic conditions prevailed.

The apparent rapid switch to aerobic methanotrophy in the high-pressure incubation was triggered by the addition of oxygen at T29, stimulating an exponential rise in δ¹³C_DIC shortly after injection. Over the course of six oxygen injections between T29 and T45, 55 mL of 100% O₂ were added to the incubation chamber, corresponding to 242 μmol/kg of O₂. Conversion of δ¹³C_DIC into consumption of methane, results in an estimated consumption of 143 μmol/kg of O₂ from aerobic methanotrophy between T29 and T45, within the range of the injected O₂. Regular ORP measurements between T29 and T45 reflected the addition of oxygen but also its rapid consumption: after T30, ORP averaged −33 mV (max = 67 mV, min = −120 mV).

The hypothesis of sequential aerobic, anaerobic, and aerobic phases in the HP-Core incubation chamber are additionally supported by sequencing data of the 16S rRNA and pmoA genes (Figures 5–7). M-core samples, taken from sediments nearby the sampling location for the HP-Core at the Joetsu Knoll, revealed the presence of diverse anaerobic methanotrophs belonging to ANME-1, -2, and -3 as well as aerobic methanotrophs of the gammaproteobacterial Methylococcales order (Figure 5). The presence of aerobic methanotrophs in the M-Core, and presumably in the neighboring HP-Core at the time of collection, is further supported by recovery of pmoA genes related to gammaproteobacterial methane oxidizers (see Figure 7; Inagaki et al., 2004b; Tavormina et al., 2010). In addition, the microbial diversity observed in the HP-Core incubation at T45 suggests ingrowth of aerobic methanotrophs in the HP-incubated sediments amended with oxygen by the end of experimentation, with relatively abundant Methylococcales and loss of OTUs associated with ANME archaea (Figure 5).

In addition, the incubated T45 sediments were also enriched in other putative aerobic gammaproteobacteria (e.g., Colwelliaceae) and epsilonproteobacteria (e.g., Campylobacteraceae) OTUs than the M-core sediment. These OTUs were also observed at high relative abundance in the T11 and T25 effluent samples, suggesting they grew up during the course of the high-pressure incubation (Figure 5). Members of the Colwelliaceae are known piezophiles (Kusube et al., 2017) and in situ experiments in the deep-sea suggest they increase in abundance in response to environmental perturbation (Case et al., 2015). Similarly, many members of the epsilonproteobacteria recovered in the HP-Core incubation (Sulfurimonas and Sulfurovum spp.) are related to hydrogen or sulfide-oxidizers within the Campylobacteriales (Inagaki et al., 2003, 2004a; Campbell et al., 2006; see Supplementary Data). As oxygen was depleted in the HP-Core incubation chamber, sulfate would have been used as an electron acceptor, producing sulfide (both in AOM and other anaerobic respiratory processes). This might explain the increased abundance of epsilonproteobacteria-associated OTUs at T11 onward.

The rates of anaerobic methane oxidation calculated from our HP-incubation experiment are consistent with previously published rates of anaerobic methane oxidation (Figure 8). Methane oxidation rates are often observed to vary by many orders of magnitude, depending on the site, methane flux and electron acceptor concentrations (Rudd et al., 1974; Harrits and Hanson, 1980; Devol, 1983; Iversen et al., 1987; Reeburgh et al., 1991; De Angelis and Lilley, 1993; Ward and Kilpatrick, 1993; Hoehler et al., 1994; Joye et al., 1999; Valentine et al., 2001; Nauhaus et al., 2002; Girguis et al., 2003; Carini et al., 2005; Bowles et al., 2011; Boetius and Wenzhöfer, 2013; Timmers et al., 2015). Our ¹³C_DIC derived rates of methane oxidation in the HP-Core incubation between T11 and T29 (presumed to be primarily attributed to sulfate coupled-AOM) were low compared to measurements from other high pressure deep-sea seep incubations (Nauhaus et al., 2002; Timmers et al., 2015), and more similar to rates reported from coastal systems (Hoehler et al., 1994) or methane seep sediments with widely measured rates (Bowles et al., 2011). Rates of aerobic methanotrophy measured in the HP-Core incubation at 10 MPa were typically elevated above the majority of previously published rates from marine sediments (Figure 8). The pressure effects on aerobic methanotrophy has not been studied in detail; however, the periodic addition of oxygen to the HP-incubation system stimulated methanotrophic activity and an increase in gammaproteobacterial methanotrophs.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.