Distinct microbial communities within and on seep carbonates support long-term anaerobic oxidation of methane and divergent pMMO diversity

Abstract

At methane seeps worldwide, syntrophic anaerobic methane-oxidizing archaea and sulfate-reducing bacteria promote carbonate precipitation and rock formation, acting as methane and carbon sinks. Although maintenance of anaerobic oxidation of methane (AOM) within seep carbonates has been documented, its reactivation upon methane exposure remains uncertain. Surface-associated microbes may metabolize sulfide from AOM, maintain carbonate anoxia, contribute to carbonate dissolution, and support higher trophic levels; however, these communities are poorly described. We provide insights into microbial diversity, metabolism, activity, and resiliency within and on seep carbonates through amplicon and metagenomic sequencing, incubations, and non-canonical amino acid tagging combined with fluorescence in situ hybridization (BONCAT-FISH). Ca. Methanophaga (ANME-1) dominated the carbonate interiors in active and low activity seeps, co-occurring with Ca. Desulfaltia as main sulfate reducer, potentially a new syntrophic partner in AOM. Single-cell BONCAT-FISH revealed variability in ANME-1 activity, suggesting potential dormancy in carbonates from low activity seep sites. However, incubations with carbonates from low activity seeps (≥24 months) showed exponential AOM reactivation (~44-day doubling), suggesting these carbonates retain the potential as long-term methane sinks under dynamic seepage conditions. Surface-associated microbial communities were heterogeneous and distinct from the carbonate interior and other seep habitats. Anaerobic methane-oxidizing biofilms and sulfide-oxidizing mats were associated with carbonates with high and intermediate AOM rates potentially influencing carbonate precipitation/dissolution. Shared aerobic methanotrophs between carbonate surfaces and invertebrates indicated carbonate surfaces may represent animal epibiont reservoirs. Recovered particulate methane monooxygenases included both aerobic methanotrophs and divergent forms associated with the Methylophagaceae, suggesting a new function in this group.

Introduction

At methane seeps across the world’s oceans, anaerobic methanotrophic archaea (ANME) in syntrophic partnership with sulfate-reducing bacteria (SRB) promote carbonate precipitation by increasing alkalinity [1, 2]. These carbonates form over hundreds to thousands of years [3–5], and persist even longer [6], storing carbon long-term [7], and serving as a rare deep-sea hardground surface for diverse animal communities [8, 9]. Methane seepage is dynamic over the long carbonate lifetime [10]. Sulfate-coupled anaerobic oxidation of methane (AOM) yields little energy and ANME-SRB doubling times are on the order of months [11, 12]. ANME-SRB entombed in carbonates and associated microbial signatures have been interpreted as fossilized remnants [13, 14]. Investigations have since demonstrated that carbonates host active ANME-SRB biomass, presumably in rock pores, capable of methane removal [15, 16]. However, cell activity levels in these carbonates have not been fully resolved, and it is currently unknown if carbonates with low AOM activity can be reactivated upon reexposure to methane.

The carbonate-hosted microbial diversity and metabolic potential remains understudied, particularly compared to seep sediments. Previous studies mainly focused on the seep carbonate interior and commonly recovered ANME-1 (Methanospirareceae, synonym Methanophagaceae) [16–18]. The family ANME-1 comprises 14 recognized genera (gtdb R220), however, the lineages associated with seep carbonates require further investigation. Other taxa reported from seep carbonates include other ANME lineages (ANME-2ab, c), Atribacteria, Chloroflexi, Proteobacteria, and Desulfobacterota, including syntrophic Seep-SRB1 [15–17, 19], but the diversity of syntrophic Seep-SRBs present within seep carbonates remains to be assessed.

Surface-associated microbial communities on seafloor exposed carbonates reside at the water/carbonate interface, with potentially steep redox gradients, between oxygen in seawater, and sulfide generated from rock-hosted AOM. By consuming oxygen with sulfide or residual methane, these surface communities may impact local redox conditions and promote anoxic conditions within the rock interior [20]. Recently, experiments demonstrated that the acidity produced during aerobic methane and sulfide oxidation may dissolve seep carbonates [21, 22], highlighting the role of microorganisms in both their synthesis and dissolution. Additionally, seep carbonates host animal communities dependent on these rock-hosted chemosynthetic microorganisms [23], but focused studies on the diversity and metabolic potential of surface-associated microorganisms have been lacking.

Here we study microbes within and on deep-sea carbonates with varying AOM activity at two methane seep sites off of Southern California. Using a combination of 16S rRNA gene amplicon analysis, metagenomics, BONCAT-FISH, isotopic rate measurements, and long-term incubations, our work highlights the diversity and ecophysiological breadth of methane-oxidizing and other chemosynthetic microorganisms living within and on the surfaces of seep carbonates. We further reveal resiliency of endolithic AOM communities to long-term fluctuations in methane supply, demonstrating the ability to reactivate sulfate-coupled methane oxidation and growth after months to years of minimal activity/dormancy.

Materials and methods

Sampling

Carbonates were collected from two methane seep areas, Del Mar (1023 m depth) [24] and Santa Monica Mound (800 m depth) [25], off Southern California in May 2021 during the MBARI WF05–21 cruise aboard the R/V Western Flyer using the ROV Doc Ricketts. Six samples, rocks 1–4 (R1–4, Del Mar outcrop, 32.904249, −117.782267, 1023.1 m), Rock 7 (R7, 32.904231, −117.782251, 1023.1 m) and Rock 9 (R9, 32.904317, −117.782429, 1020.9 m), were collected from Del Mar. Further, two samples from Santa Monica Mound 800 (SMM800) with tube structure, Chimlet (33.799532, −118.646879, 804 m), and Protochimney (33.799616, −118.64678, 803 m), were included in this study. Where possible, sediment push cores and water samples adjacent to the carbonate samples were collected for 16S rRNA gene comparison. These included one water sample (DR1330-N4, 32.904248, −117.782289, 1023.2 m) and one sediment sample (DR1330-PC47, 32.904231, −117.78227, 1023.1 m) from Del Mar, and three of each at SMM800: sediment samples (DR1329-LC62, 33.799447, −118.647171, 805.2 m; DR1329-PC44, 33.799662, −118.646947, 805.3 m; DR1329-PC69, 33.798962, −118.646264, 809.8 m) and water samples (DR1329-N2, 33.79971, −118.646902, 798.5 m; DR1332-N1, 33.79941, −118.647144, 806.5 m; DR1332-N3, 33.799546, −118.646866, 800.3 m). Carbonate samples were transferred into argon flushed mylar bags, submerged in nitrogen sparged 0.2 μm filtered Niskin bottom seawater from the site, and heat sealed for transport back to the laboratory for further analysis. The in situ orientation of these carbonates was reconstructed using video footage collected during the dive. In the laboratory we subsampled the rocks with a tile saw or drill press cleaned with ethanol and Nanopure water between samples. Samples for DNA extraction were then stored at −80°C until further subsampling. Subsamples for cell counts were fixed 24 hours in 2% paraformaldehyde at 4°C, washed twice with 3x phosphate-buffered saline (PBS), and stored in 70% ethanol 30% 1x PBS at −20°C. The 10x PBS stock contained 1.37 M NaCl, 26.8 mM KCl, 0.1 M Na₂HPO₄, and 17.6 mM KH₂PO₄ in Nanopure water and was adjusted to pH 7.4 with HCl. Further sampling information and a detailed processing description are provided in the supplementary methods.

AOM rate measurements

We measured AOM rates using mono-deuterated methane (CH₃D) or ¹³C-CH₄ as substrates according to previously published methods [26, 27], and quantified δD-H₂O or ¹³C-CO₂ production, respectively, at five timepoints for experimental treatments, and three timepoints for controls. At the last timepoint (t4) after 48–143 days (further details Supplementary Table S1), we measured sulfide. The incubations with carbonate subsamples were performed in 11.7 ml or 38.5 ml serum vials with ~8 ml or 28 ml site-specific N₂ sparged and filtered seawater, and a headspace of ~1.5 ml or 8 ml at 4°C, close to in situ temperatures, under anoxic conditions. The vials were closed using 13x20 mm bromobutyl serum vial stoppers (Geo-Microbial Tech), autoclaved three times in 0.1 M NaOH prior to use. We added 100% or 50% labeled mono-deuterated methane (CH₃D), and 100% or 10% labeled ¹³C-CH₄ at 2 bars partial pressure of methane, resulting in ~3.5 mM methane in the incubations. These methane concentrations likely provide conservative estimates of methane oxidation rates, as the saturation concentration of methane at the sampled depths (800 and 1020 m) exceed 100 mM. For subsequent BONCAT-FISH we further added 200 μM HPG (L-Homopropargylglycine). In parallel, we performed an unlabeled and an autoclaved killed control for each rock (R1, R2, R3, R4, R9, Chimlet, Protochimney). The unlabeled controls contained unlabeled methane (Airgas) and no HPG. The autoclaved killed control contained 50:50 CH₃D and ¹³C-CH₄. The incubations with ¹³C-CH₄ did not yield a quantitative rate measurement for most samples, due to an unusually high manufacturer contamination of the ¹³C-CH₄ gas with ¹³C-CO₂, discovered after the fact, which lowered the sensitivity of the ¹³C-DIC assays with low AOM activity, or too high AOM activity (Chimlet). This resulted in unreliable ¹³C-CO₂ values, that were either unresolved above the ¹³C-DIC background, or outside the upper quantification range, respectively. Even though we show the results (Supplementary Table S1), we do not discuss them further.

We derived potential anaerobic methane activation rates (referred to as “AOM-CH₃D rates”) from the slope of a linear regression of δD-H₂O production over time according to CH₃D + SO₄²⁻ → HCO₃⁻ + HS⁻ + HDO, measured with a Picarro isotopic water analyzer (2140-i). For more detail about CH₃D as a tool to measure AOM rates, see a previously published discussion [26]. Using CH₃D may underestimate rates if not all four hydrogen atoms form water during methane oxidation, thus representing a conservative estimate. Still, CH₃D is potentially more sensitive than ¹³C-CH₄, because full oxidation to CO₂ is not necessary, such that methane activation is enough for a signal. We therefore report the AOM-CH₃D rates in nmol Deuterium per cm⁻³ d⁻¹. The δD-H₂O of the killed controls did not increase over time. Incubations without measurable enrichment of δD-H₂O over time are reported as rates below detection. Further details on the rate measurements are provided in the supplementary methods.

Long-term reactivation incubations

For long-term reactivation incubations we set up one incubation per low AOM rock. Remaining rock pieces from samples R1 (45 g), R3 (201 g), and R4 (287 g) were incubated in Duran bottles (250 ml, 500 ml, and 1000 ml, respectively) with sterile, anoxic artificial seawater (100 ml, 300 ml, and 400 ml, respectively, recipe Supplementary Table S2), and a headspace (~170 ml, 200 ml, and 530 ml, respectively) pressurized to 2.4 bar with unlabeled methane. The bottles were capped with black bromobutyl rubber stoppers (GL45, Ochs, Germany) and secured with open-top screw caps. Sulfide was monitored over time for which we preserved 400 μl 0.2 μm filtered medium in 400 μl 500 mM zinc acetate. Sulfide concentrations were determined using the Cline assay [28] via methylene blue absorbance at 670 nm, measured in triplicate 200 μl reactions with 20–100 μl preserved sample on a plate reader (TECAN Sunrise). At high mM sulfide concentrations, we further diluted 1:10 or 1:50 to obtain measurements in the linear range of the standard curve.

DNA extraction

For DNA analysis the frozen rocks were further subsampled into 4–7 horizons in 1–7 mm steps from surface to interior. The rock surface was scraped off with an ethanol flamed spatula, and the rock horizons were cut using a rotary tool (Dremel) with a diamond wheel and then ground to powder. Rocks and sediment were extracted with the DNeasy PowerSoil Pro Kit (Qiagen) with modifications for rocks as described elsewhere [29]. Water samples filtered onto Sterivex filters were extracted with a phenol-chloroform method. A different DNA extraction method was used for Sterivex filters to accommodate the sample type and the larger volumes necessary for their extraction. Further details are provided in the supplementary methods.

16S rRNA gene amplicon sequencing and analysis

The 16S rRNA gene (V4-V5) was amplified using 515F-Y and 926R [30] archaeal/bacterial primers with adapters (Illumina) in duplicate PCR amplifications with 33 cycles and 54°C annealing temperature (Q5 Hot Start High-Fidelity 2x Master Mix, New England Biolabs, USA). Duplicates were pooled and barcoded with Nextera XT index 2 primers (Illumina). Barcoded PCR products were combined equimolarly, purified, and sequenced on a MiSeq System (Illumina) with 15%–20% PhiX by Laragen (Culver City, CA). Further details provided in the supplementary methods.

After sequencing we removed adapters with cutadapt (v. 3.4), and inferred amplicon sequence variants (ASVs) using DADA2 (v. 1.20.0) in R (v. 4.2.2). ASVs were annotated with IDTAXA of DECIPHER (v. 2.20.0) and the SILVA_SSU_r138 database amended with in-house sequences. Decontam (v. 1.18.0) in R was used to remove contaminants. For a refined ANME-1 classification we extracted available 16S rRNA genes from gtdb representative genomes with barrnap (v0.9) [31]. Together with sequences from this study we constructed a phylogenetic tree using IQtree (v2.1.2) [32]. The SILVA database currently only distinguishes ANME-1a and ANME-1b, thereby underestimating the ANME-1 taxonomic diversity of 14 genera present in gtdb. Further, the 16S rRNA gene trees of SRB and aerobic methanotrophs were constructed with IQtree. Sequences were aligned with muscle (v. 3.8.1551).

Metagenomics sequencing and analysis

We selected three rocks, R1, R9, and Chimlet, for metagenomic sequencing of the interior and surface, and sequenced a total of six metagenomes. We prepared the library with the DNA Prep kit (Illumina) and 10–12 amplification cycles depending on input. The Keck Genomics Platform of the University of California quantity- and quality-checked the libraries with a TapeStation 4200 (Agilent), and sequenced them on a NovaSeq 6000 System, S1 flowcell, for 150 bp paired-end reads (Illumina Inc.).

Using bbduk [33] we trimmed primers and adapters. We assembled the reads five times using (i) megahit (v1.2.9) [34] coassembly, (ii) metaspades (v3.15.2) [35] individual assemblies, (iii) megahit individual assemblies, (iv) megahit coassemblies normalized with bbnorm (target = 33) [33], and (v) megahit bbnorm individual assemblies. Further details provided in supplementary methods. We binned and refined mags using metawrap (v1.3.2, concoct, maxbin2, metabat2) [36] from each assembly with reads from all samples for differential coverage. We then dereplicated the bins at 95% ANI with dRep (v2.6.2) [37]. Selected bins were manually inspected with anvio 7.1 [38], and obvious contamination was removed based on evenness of coverage. Mdmcleaner [39] was used to taxonomically classify contigs and was used in addition to anvio visualization to guide manual refining in anvio 7.1 if necessary. We classified the MAGs with gtdbtk (v2.3.2, database version r220) [40] and determined MAG coverage with coverM (v0.6.1) [41]. The mags were annotated with metabolic (v4.0) [42]. Based on checkM and checkM2 [43, 44] MAGs with > 50% completeness and < 10% redundancy were kept. MAGs that dropped below 50% completeness after refinement were kept, prioritizing lower contamination over completeness.

Phylogenetic trees of ANME and SRB MAGs were done with anvio 7.1 using the Archaea_76 and Bacteria_71 marker genes set, respectively. Reference genomes were retrieved from gtdb r214 and amended with ANME-1c genomes [45]. We updated taxonomic names to gtdb r220 where necessary.

pmoC analysis

The first set of divergent pmoC sequences was retrieved from annotating MAGs with metabolic. Only the bin_133 (Methylophagaceae) unambiguously contained the divergent pmoC based on anvio inspection. Further pmoC sequences were searched and extracted from translated predicted genes (prodigal v2.6.3) [46] from the metagenomic assemblies using diamond blastp (v2.0.6.) [47]. A pmoC database was constructed for this purpose. Reference sequences of cultivated and uncultivated microorganisms were either retrieved using NCBI blastp [48] or were extracted from MAGs from gtdb. The pmoC gene tree was constructed with IQtree. All trees were visualized with iTol (v.6) [49]. The pmoC sequences from this study are provided as a supplementary fasta file.

Cell extraction

To extract cells from the carbonate rock matrix for microscopy, we developed an extraction protocol combining a cell extraction buffer [50] with a percoll density centrifugation [51]. A detailed protocol is available on protocols.io [52].

Cell counts

Extracted cells from eight unincubated interior samples (of rocks R1, R2, R3, R4, R7, R9, Chimlet, and Protochimney) were filtered onto black, 0.2 μm pore size polycarbonate membranes (GTBP02500, Isopore, Millipore Sigma), and mounted with Citifluor containing 4′,6-diamidino-2-phenylindole (DAPI, 4.5 ng/μL). We analyzed between 13–30 field of views and 2320–11 502 cells per sample on an Elyra PS.1 SIM microscope (Zeiss, Germany) with alpha Plan-APOCHROMAT 100X/1.46 oil objective. Images with lower quality were removed prior to automatic counting with Fiji-ImageJ (2.3.0, v. 1.53q, Java 1.8.0_172). Automatic counting included Gaussian blur 1 for noise reduction, background subtraction (rolling ball, 50 pixels), and thresholding to create masks (Otsu). Particles with a minimum size of 0.04 μm² were counted after watershed transformation.

Translational activity of ANME with BONCAT-FISH

For bioorthogonal noncanonical amino acid tagging (BONCAT) and fluorescence in situ hybridization (FISH), extracted cells from selected individual HPG-incubated rock interiors (R1, R3, R9, Chimlet, Protochimney) were mounted onto slides, and HPG incorporation was visualized using a click-reaction with AF647 picolyl azide [53]. FISH was done according to standard protocols [54], using an archaeal (ARCH915 [55], dualAlexa546) and a bacterial probe mix (EUB, EUBII, and EUBIII [54], dualAlexa488). We interpret the archaeal cells in R9, Chimlet, and Protochimney as coming almost exclusively from ANME, by far the most abundant archaea based on 16S rRNA gene amplicon sequencing and metagenomics. We confirmed ANME-1 presence in R9. To do so, we mixed ANME1–350 [56] (cy3, 5’-AGTTTTCGCGCCTGATGC-3′) and a new probe, ANME1–728 (cy3, 5’-GGTCTGGTCAGACGCCTT-3′), designed in ARB using SILVA 138 database, both highly specific. Two probes targeting different regions on the 16S rRNA were combined to obtain a stronger signal. For data analysis we identified archaeal cells automatically with Fiji-ImageJ (2.3.0, v. 1.53q, Java 1.8.0_172) and inspected the archaeal cells manually for a BONCAT signal. A positive BONCAT signal was counted if a cell-shaped BONCAT-signal was present in the cy5 (BONCAT) channel. The same protocol was applied to the negative controls, consisting of cells from a non-HPG-incubated subsample of the corresponding rock. The percentage of false positive cells was determined for the respective negative control, and subtracted from the counts (Chimlet 0%, Protochimney 18%, R9 20%). Despite the generally low BONCAT (cy5) signal in the negative control, cell shapes were occasionally visible, which may be caused by unspecific autofluorescence of cells, abiotic substances, or fluorescence of unremoved dye [57]. Further details are provided in the supplementary methods.

Results

Seep carbonates with low to high AOM activity

Carbonate rocks from the Del Mar and Santa Monica Mound 800 (SMM800) methane seeps off Southern California (Fig. 1A,B [58]), were characterized with in situ observations. The Del Mar outcrop (Rocks 1–4) extended into the water column (near lower edge of oxygen minimum zone, 22 μM oxygen) and lacked active methane seepage and AOM indicators such as sulfide-oxidizing microbial mats and abundant chemosynthetic animals (Fig. 1C), suggesting low AOM activity. In the following we refer to these as low AOM rocks. Rock 9 (R9) from Del Mar was covered in white sulfide-oxidizing bacteria (Fig. 1D), suggesting active AOM and sulfide production. Chimlet and Protochimney from SMM800 are hollow carbonate tubes sealed at the top and extended into an oxygen minimum zone (8 μM oxygen). White and black microbial mats, and animals covered these rocks suggesting high AOM activity (Fig. 1E,F). Relative to Chimlet and Protochimney, we refer to R9 as having intermediate AOM activity and refer to it as intermediate AOM rock in the following. Depleted δ¹³C_carbonate values (−46.2 to −54.0‰) confirm the methane-derived origin [59] of the studied carbonates (Fig. S1).

Figure 1.

Anaerobic methane oxidation activity of methane seep carbonates from Del Mar and SMM800. In situ AOM indicators and AOM-CH₃D rate measurements characterize low to high AOM carbonate rocks. (A) Seep carbonate collection sites Del Mar (light green marker) and Santa Monica Mound 800 (SMM800, dark green marker) are located 129 km apart. Map obtained from Google Maps. (B) Biogeochemical seep carbonate setting. ROV Doc Ricketts in situ images of (C) the Del Mar outcrop, R1 and R2 originated from the top, R3 and R4 from closer to the sediment. (D) R9, from a nearby Del Mar area with sulfide oxidizing mats. (E) Chimlet and (F) Protochimney are two chemoherm-like structures and were collected from different sides of the Santa Monica Mound 800. Chimlet actively bubbled with methane upon recovery. For scale, the red laser points in the images are 29 cm apart. (G) Potential anaerobic methane activation rates (AOM-CH₃D rates, nmol D cm⁻³ d⁻¹) measured in anoxic incubations with monodeuterated methane based on: CH₃D + SO₄²⁻ ➔ HCO₃⁻ + HS⁻ + HDO. We measured δD of water over five timepoints and calculated the rate from the linear increase over time, unless stated otherwise. Error bars represent the standard error of the slope (k) calculated from the linear regression across the timepoints as described in [102]. Two subsamples of R9, R9.1 and R9.2 with different color, light grey and dark grey, respectively, were incubated for AOM-CH₃D rates. The orientation of the R9 piece dedicated for rates could not be reconstructed. At the last time point (t4) sulfide was measured and was detectable in R9.1, Chimlet top, middle, bottom, and Protochimney surface. *Deuterium above background was only detected at t4 indicating a nonlinear increase in R2 and R3., b.d. below detection, surf. surface, int. interior, btm. bottom.

We used anoxic incubations amended with monodeuterated methane (CH₃D) to measure potential anaerobic methane activation rates—used as a proxy for AOM [26]—based on increases in HDO (CH₃D + SO₄²⁻ → HCO₃⁻ + HS⁻ + HDO) over 1.6 to 4.8 months. In the following, we refer to these as AOM-CH₃D rates. The Del Mar outcrop carbonates, Rocks 1–4, had low AOM-CH₃D rates, from non-detectable to 11.6 nmol D cm⁻³ d⁻¹, without measurable sulfide production within 4+ months. Of these, R2 and R3 showed an HDO increase at the last time point (4+ months), potentially indicating stimulation of ANME-SRB activity (Fig. 1G). Based on observed color differences two Rock 9 subsamples, R9.1 (light grey) and R9.2 (dark grey) were incubated. R9.1 had an AOM-CH₃D rate of 434.2 nmol D cm⁻³ d⁻¹ and produced sulfide, confirming in situ indicators of active sulfate-coupled AOM. In comparison, R9.2 had a lower rate of 30.9 nmol D cm⁻³ d⁻¹. Differences between subsamples are expected, as carbonate rocks are naturally heterogeneous. The tubular Chimlet sample had high AOM-CH₃D rates, with the greatest rates associated with the cap (Fig. 1E,G). The outer surface sample of Protochimney also showed high AOM-CH₃D rates, while the corresponding interior piece displayed no measurable rate (Fig. 1G). This likely does not reflect total inactivity in the interior, as a second Protochimney interior piece showed sulfide production (Supplementary Table S3). Based on in situ observations and the high AOM-CH₃D rates, we refer to Chimlet and Protochimney, both from SMM800, as high AOM rocks.

Distinct seep microbial communities—heterogeneous carbonate surface and homogeneous interior

Rock subsectioning and 16S rRNA gene sequencing revealed a thin, heterogeneous surface microbial community and a more homogeneous interior community beneath it, both distinct from other seep habitats (Fig. 2A,B). The surface veneer community (0–1 mm) was different from the interior (Fig. 2A,B), often showing high relative abundances of Proteobacteria (e.g. unclassified Gammaproteobacteria, Methylococcales, Chromatiales, Thiotrichales, Beggiatoales) in contrast to the interior (Fig. 2C). The 1–2 mm section included the previously scraped surface, sometimes with surface community remnants (Fig. 2C, e.g. R1). The Chimlet and Protochimney surfaces were dominated by ANME-SRB (Halobacteriota, Desulfobacterota), like the interior at phylum level. Methanocomedenaceae and Seep-SRB1a dominated these biofilms, and on Protochimney, additionally Ca. Methanogaster and Seep-SRB2 (Fig. 2D,E). White/orange surface regions, again, showed more Proteobacteria (Fig. 2C). The interior hosted high relative abundances of Halobacteriota (primarily ANME), and variable abundances of Caldatribacteriota, Desulfobacterota, Planctomycetota, and Asgardarchaeota (Fig. 2C, additional profiles Fig. S2, median abundance ranks Fig. S3). The surface microbial community was distinct from the water column (Fig. 2B), even though they are in contact. Both rock and sediment surfaces are redox and phase transition zones, however, their communities were clearly distinct (Fig. 2B).

Figure 2.

Microbial communities from the carbonate surface to interior in context of other seep communities. 16S rRNA gene sequencing of carbonate surface scrapes (0–1 mm) and mm-scale sections. (A) Cross section of Del Mar outcrop Rock 4, with a white partially scrapeable surface layer (0–1 mm). Chimlet and Protochimney had a black biofilm instead with white and orange patches. (B) NMDS of bacterial and archaeal ASVs showing dissimilarities between rock surface and interior in context of sediment, water, and other seep communities. Each point represents one sample. (n = 103, stress = 0.18). Shapes were hand drawn for visualization purposes. (C) Archaeal and bacterial phyla in selected sectioned carbonates from surface to interior based on 16S rRNA gene sequencing. Phyla reaching ≥5% at least once are shown. Carbonates naturally varied in size and shape which led to different section sizes, even though we tried to cut them as similarly as possible. We sectioned three parts of Chimlet: Top (cap), middle, bottom (btm). (D) ANME phylogenetic groups based on gtdb taxonomy identified with a 16S rRNA phylogenetic tree (Fig. S4) to genus level where possible. (E) SRB phylogenetic groups based on a 16S rRNA phylogenetic tree (Fig. S5). SRB x-axes are scaled to the maximum abundance of the respective profile for better visibility, because SRB abundances varied substantially. ANME and SRB ASVs reaching ≥ 1% at least once were included. See full set of sectioned rock microbial communities in the supplementary material (Fig. S2, including R2, R4, R7, Chimlet top, Chimlet mid). Abbreviations: Inner surf., inner surface of Chimlet and Protochimney cavity; bl, black; ow, orange white.

We found mm-scale open veins with sediment-like material in the Del Mar outcrop, potentially inhabited by animals, with microbial communities between that of water column, sediment, and rock (“mud vein”, Fig. 2B). Mud recovered from the central cavity of Chimlet harbored a microbial community that was similar to the lithified rock (“mud chim”, Fig. 2B). R7 from Del Mar had an attached hydroid (sessile colonial invertebrate, phylum Cnidaria), whose microbiome was most similar to, but distinct from the Del Mar carbonate surface communities (Fig. 2B).

Dominant endolithic ANME-1 genera and a potential new SRB partner

Based on 16S rRNA gene amplicon sequencing and metagenomics, the ANME-1 genus Ca. Methanophaga (QENH01, ANME-1b) dominated the rock interior, and the dominant SRB MAG was Ca. Desulfaltia (Fig. 2D, 3). We recovered further ANME-1 MAGs affiliated with the genera QEXZ01 and QENJ01, that were also present in the rock interiors (Fig. 3). ANME-1 JACGMN01, identified through 16S rRNA gene amplicons (Ca. Methanoalium-2, tree Fig. S4), was mainly associated with the inner surface and deepest sections of Chimlet and Protochimney (up to 2.1% of the community), totaling to 3–4 ANME-1 genera per rock interior. Unexpectedly, low and high AOM rocks showed a high relative abundance of ANME-1 (16S rRNA gene sequencing: on average 28–35% in R1–4, and 16–26% in Protochimney and Chimlet; ANME-1 MAGs: 19% R1, 21% Chimlet, Fig. 2D, 3).

Figure 3.

MAGs and metabolic potential in and on low, intermediate, and high AOM carbonates. We sequenced the surface and the interior of R1, R9, and, Chimlet with low, intermediate, and high AOM activity, respectively, and binned MAGs from these six metagenomes to investigate taxonomy and metabolic potential. MAGs shown were selected based on relative abundance, shown in parenthesis, with a focus on Proteobacteria on the rock surface because of their higher abundance compared to the interior (Fig. 2C). A full list of MAGs is reported in the supplementary material (Supplementary Table S4). Rank-abundance plots of the top 15 MAGs (based on metagenomics) and ASVs (based on 16S rRNA gene amplicon sequencing) were generated for overlapping samples (Figs. S6, S7). Anaerobic methane oxidizing archaea dominated the rock interiors, Ca. Methanophaga in particular, and Ca. Desulfaltia was the most abundant interior SRB based on metagenomics suggesting it may be an ANME partner bacterium. In contrast, the surface communities and their metabolic potentials differed between rocks along with the AOM activity of the carbonate, methane supply strength, and oxygen concentrations. We indicated the difference in methane supply to the rocks estimated based on the in situ observations with arrow size. Further, oxygen concentrations were higher at the Del Mar (22 μM) vs. SMM800 (8 μM) seep.

In contrast to the rock interior, Chimlet’s surface mat was dominated by Methanocomedenaceae (ANME-2ab, MAG relative abundance 29%). Additionally, a few rock sections adjacent to the inner cavity walls of Chimlet and Protochimney showed the presence of Methanocomedenaceae (Fig. 2D). Members of Ca. Methanogaster (ANME-2c) were additionally detected in the Del Mar outcrop carbonate interiors and dominant in the inner cavity walls of Chimlet and Protochimney, and the Protochimney biofilm (Fig. 2D).

We recovered MAGs of known ANME partner bacteria including Seep-SRB1a (B13-G4), Seep-SRB1g (C00003106), and Seep-SRB2 (UBA3076). However, Ca. Desulfaltia (bin_051, 65% completeness) was the dominant SRB MAG in the interior (Fig. 3). Ca. Desulfaltia belongs to the family ETH-SRB1, which contains Seep-SRB1a, and the Ca. Ethanoperedens-partner genus ETH-SRB1 (Fig. 4B). Bin_051 lacked a 16S rRNA gene, but based on gtdb, Ca. Desulfaltia may belong to Seep-SRB1b [60]. Large multiheme cytochromes (oetAB) predicted to be involved in the electron transfer from ANME to SRB [61] were missing from bin_051, but were detected in two Ca. Desulfaltia gtdb MAGs (Fig. S8). Based on these results, Ca. Desulfaltia is a promising candidate for a syntrophic ANME partner bacterium, but additional evidence is required.

Figure 4.

Phylogenomic tree of ANME and Seep-SRB metagenome assembled genomes (MAGs) and reference MAGs. MAGs from this study are in bold. We report the gtdb taxonomy, previously proposed names, and historic names for reference. (A) We find four ANME-1 MAGs within three genera with highest relative abundance in the rock interior. Members of Ca. Methanocomedens and bin_006 Methanocomedenaceae, which may be part of a new genus had highest abundance in Chimlet’s black biofilm. Members of Ca. Methanomarinus were present at lower relative abundance in Chimlet’s interior. We found two Ca. Methanogaster MAGs, reaching a maximum of 0.6% relative abundance in Chimlet’s interior. (B) We found MAGs of the known ANME-partner genera Seep-SRB1a, Seep-SRB1g, and Seep-SRB2, with highest relative abundance in Chimlet’s surface, Chimlet’s interior, R1’s interior, respectively (Fig. 3). The most abundant SRB MAG in the rock interior was Ca. Desulfaltia, an uncultivated genus part of family ETH-SRB1. ETH-SRB1 further contains Seep-SRB1a, a known ANME partner, and genus ETH-SRB1, a known sulfate reducing partner bacterium of Ca. Ethanoperedens. The scale bars represent one amino acid substitution per site. Bootstrap support is indicated by black circles (> 90%) or white circles (80–89%).

Investigating the mismatch between ANME relative abundance and AOM activity with BONCAT-FISH

Low AOM rocks R1–R4 showed high DNA-based ANME relative abundances (Fig. 1, 2D, 3). Further, both low and high AOM rocks had similar cell counts (8x10⁷–7x10⁸ g⁻¹, Fig. 5A), suggesting that the active cell proportions varied. This scenario was tested with BONCAT-FISH after anoxic incubation with methane for ~50–140 days. We used an archaeal FISH probe to stain ANME, because most archaeal sequences were affiliated with ANME in the tested rocks (Fig. 2C). The higher AOM rocks R9, Protochimney, and Chimlet had a large fraction of BONCAT-positive ANME cells representing 41%, 68%, and 83%, respectively. Many of the cells displayed the commonly described ANME-1 rectangular rod shape [51] (Fig. 5B,C). Using R9, we confirmed abundance of ANME-1 using group-specific FISH probes (representative image, Fig. S9). This contrasted with the low AOM Rocks R1 and R3, where we could not identify archaeal cells by FISH, and, therefore, no BONCAT-active archaea either, which may indicate ANME cells were dormant or low activity, or active cells were too rare to be detected.

Figure 5.

Cell counts of carbonate-hosted microorganisms and BONCAT-FISH of carbonate-hosted ANME. (A) DAPI cell counts of extracted cells from the carbonate interior pre-incubation. Error bars represent the standard deviation between different fields of view. Representative BONCAT-FISH images of (B) Rock 9 and (C) Chimlet. BONCAT-FISH was used to examine the anabolic activity of endolithic ANME cells after anoxic incubation with methane and HPG for ~50–140 days. Bioorthogonal noncanonical amino acid tagging (BONCAT) measures the translational activity at the single cell level via incorporation of HPG, a methionine analog. Using fluorescence in situ hybridization (FISH) we identified archaeal cells (red, ARCH915). White circles highlight examples of archaeal cells (red) with corresponding BONCAT signal (magenta). BONCAT intensities varied substantially between cells. Arrows in (B) point to magnified representative cells for which archaeal FISH signal (red, upper image) with typical ANME-1 shape and BONCAT signal (magenta, lower image) are shown separately. R9, Protochimney, and Chimlet had 41%, 68%, and 83% BONCAT positive archaeal cells, respectively.

Reactivation of low AOM carbonates in long-term incubations

To test if carbonates with low AOM (R1–R4) have the capacity to regain AOM activity, we conducted long-term incubations (700–1000 days) under conditions mimicking renewed methane seepage. After a lag time of ~300 days, the sulfide concentration of R4 increased exponentially, suggestive of ANME-SRB growth [62, 63] with an estimated rate (r) of 0.0157 d⁻¹ and apparent doubling time of 44 days (Fig. 6A). In an additional experiment we confirmed the methane dependency of sulfide production in the reactivated R4 (Fig. S12). Similarly, adjacent R3 and R1 carbonates showed exponential sulfide production after ~600 and ~700 days, respectively (Fig. 6B,C). The growth parameters of R1 and R3 should be treated with caution because of lower R² and sampling frequency. Our results suggest AOM can be restored over months to years once methane becomes available.

Figure 6.

Exponential sulfide increase in long-term incubations of low AOM carbonates with sulfate and methane. We reactivated the initially low AOM carbonates (A) R4, (B) R3, and (C) R1, in order of reactivation, in long-term incubations mimicking methane resurge with artificial seawater (10 mM sulfate) and a methane headspace under anoxic conditions. The incubation of carbonate R2 was accidentally lost. The red dots (n = 17, n = 5, n = 5, respectively) were included in the growth rate calculation (for details see methods). Only the phase with increasing sulfide concentrations is shown. The total monitoring time was 698, 777, and 1092 days, respectively. R² and growth rate r were derived from a linear regression of log transformed concentrations (Fig. S10). The observed lag time of months to years might be explained by few initial viable ANME-SRB or a long growth preparation time. The entire measurement series, including the pre-increase phase, is reported in the supplementary material (Fig. S11). The slowing of the exponential increase at the last timepoints might point to a resource limitation or sulfide inhibition.

Diversity and metabolic potential of the surface community reflects carbonate-associated AOM activity

Proteobacteria dominated the surface of the low AOM carbonate R1. No single MAG dominated, and gammaproteobacterial bin_147 (UBA1847, Woeseiaceae) had the highest relative abundance with 1.1% (Fig. 3, rank abundance plot Fig. S6). This MAG contained genes for sulfur oxidation, methanol and methanethiol utilization, nitrite reduction, and autotrophy (form II Rubisco, phosphoribulokinase (prk)). The most abundant aerobic methanotroph, bin_129 (QPIN01, Methylomonadaceae, 0.9% rel. abundance), encoded nitrite reduction and sulfide oxidation genes. Gammaproteobacterial bin_139 (GCA-2746365, SZUA-229, 0.8% rel. abundance) contained genes for sulfide oxidation, carbon fixation (form I Rubisco, prk), and nitrite reduction. Gammaproteobacterial bin_124 (GCA-001735895, 0.6% rel. abundance) encoded genes for methanol and methanethiol utilization, carbon fixation (form I Rubisco, prk), nitrate and nitrite reduction, and sulfide oxidation. Methylotrophic autotrophy based on Rubisco has been reported previously in a member of the Beijerinckiaceae [64]. Supplementary Table S4 contains a complete list of MAGs.

The surface of intermediate AOM carbonate R9 was dominated by large sulfur-oxidizing Ca. Marithrix (bin_119, Beggiatoaceae, 25.6% rel. abundance, Fig. 3), consistent with the white in situ mat (Fig. 1). The MAG encoded nitrate, nitrite, and nitric oxide reduction genes as previously described [65], and likely utilizes sulfide generated by endolithic ANME-SRB. Like R1, R9 again hosted the aerobic methanotroph genus QPIN01 (bin_130, 1.1% rel. abundance). Further, Alphaproteobacteria bin_112 (Profundibacter, Rhodobacteraceae, 0.7% rel. abundance) encoded genes for benzoyl-CoA reduction, formaldehyde utilization, nitrate reduction to N₂, sulfide oxidation, and carbon fixation (form II Rubisco, prk).

The high AOM rock, Chimlet, from SMM800 had a distinct surface community from the Del Mar surface communities, dominated by an ANME-SRB biofilm. Chimlet’s biofilm was dominated by ANME-2 bin_006 (Methanocomedenaceae, 19.6% rel. abundance, Fig. 3), a potentially new Methanocomedenaceae genus (Fig. 4A), followed by the ANME partner bacterium Seep-SRB1a (bin_049, 10.1% rel. abundance), and bin_009 belonging to Ca. Methanocomedens (9% rel. abundance). Potential sulfur-oxidizing bacteria were present at low relative abundances, including Sedimenticolaceae (bin_120, 0.2% and bin_221 HyVt-443, 0.1% rel. abundance) and Desulfobulbaceae (bin_064, 0.5% rel. abundance).

Diverse and unique aerobic methanotrophs on carbonate rocks

16S rRNA gene amplicon sequence analysis of carbonate surfaces identified distinct aerobic methanotrophic lineages (Fig. 7). This included diverse members of Methylococcales clades that remained unclassified based on a phylogenetic tree (Fig. S13), that we refer to as uncultivated Methylomonadaceae −1, −2, −3, and −4. The Methylomonadaceae-3 subclade was recovered from several rock surfaces, including the SMM800 carbonates. IheB2–23 members were recovered from the R1 surface and, at high relative abundance, from a hydroid microbiome, attached to Del Mar R7 (Fig. 7D). IheB2–23 has previously been found on crabs [66], suggesting IheB2–23 may associate with seep invertebrates. Similarly, members of the Marine Methylotrophic Group 2 were recovered from all Del Mar carbonate surfaces, a lineage forming symbioses with sponges and seep-associated feather duster worms [67, 68]. Other aerobic methanotrophs appeared to have different habitat preferences. Methyloprofundus, a genus isolated from deep-sea sediments [69], was the main aerobic methanotroph recovered from Del Mar surface sediment, and from mud-like material in carbonates R3 and R4 (Fig. 7B,C), but had low relative abundance on carbonate surfaces. In the water column we mainly found OPU-1 and OPU-3 with a higher OPU-3 ratio at SMM800 associated with lower oxygen concentrations (Fig. 7A), consistent with earlier reports from oxygen minimum zones [70].

Figure 7.

Aerobic methane-oxidizing bacteria on the carbonate surface compared to other seep habitats based on 16S rRNA gene sequencing. The carbonate surface harbors a distinct aerobic methanotrophic community. 16S rRNA sequence of the water column (A), the sediment surface (B), the mud in open vein (C), and the rock surface (D). We first classified the 16S rRNA sequences with SILVA and selected Methylococcales sequences. The classification was refined with a 16S rRNA gene tree and reference sequences (Fig. S13). We did not detect members of Methylococcales on the Chimlet mid-section surface, and the corresponding bar is not shown.

Carbonate-associated Methylophagaceae and unbinned contigs encode divergent CuMMO genes

Metagenomic analysis revealed an unusual diversity of carbonate-associated copper membrane monooxygenase (CuMMO) genes. CuMMO genes are encoded by xmoCAB and catalyze aerobic oxidation of methane (pmoCAB), short-chain alkanes, and ammonia [71]. A member of Methylophagaceae (bin_133, genus GCA-002733105) encoded xmoCAB, which has not been reported before within Methylophagaceae (Fig. 8). Aerobic methanotrophs oxidize methane to methanol using pmoCAB, and Methylophagaceae are known methanol oxidizers [72, 73]. Therefore, methane oxidation with xmoCAB in Methylophagaceae appears likely. This is challenged by the finding that its xmoC (Fig. 8D) is divergent from aerobic methanotrophs located on the branch C-G together with ethane-oxidizing Proteobacteria and Nitrosomonas (Fig. 8C,E). Oxidation of short-chain alkanes, ammonia or other substrates are equally likely hypothetical functions. Further xmoC diversity was recovered from R1 surface assemblies and co-assemblies (unbinned contigs). A new xmoC clade clustered with xmoC from Nocardioides CF8 and Mycolicibacterium chubuense (B, Actinomycetes), both known short-chain alkane (propane/butane) oxidizers [74, 75]. Two sequences clustered with ETHIRO and Cycloclasticus (Fig. 8C, Proteobacteria), that likely oxidize ethane or propane [76], and one xmoC clustered with Halioglobus (Fig. 8C). We found one pxmC sequence (Fig. 8G), a divergent pmoC with unknown function, commonly found in aerobic methanotrophs [77]. Additional to divergent xmoC genes, we found conventional pmoC genes (> 40, Fig. 8M) clustering with methane-oxidizing Methylomonadaceae (Fig. 8, uncollapsed tree Fig. S14), and a clade (Fig. 8L) clustering with but different from USCg, an aerobic methanotroph outside the order Methylococcales. The tree is consistent with previously published pmo topologies [74, 77] broadly showing four branches: one branch with archaeal amoC (Fig. 8A), one with xmoC of Actinomycetes (Fig. 8B), one with xmoC of ethane-oxidizing Proteobacteria, Methylophagaceae from this study, Nitrosomonas, Haliea, and pxmC genes (Fig. 8C-G), and the branch with pmoC of Methylacidiphilaceae, Methylomirabilis, Alphaproteobacteria, Methylomonadaceae, Methylococcaceae, other families, as well as gammaproteobacterial amoC (Fig. 8H-N).

Figure 8.

Phylogenetic tree of CuMMO subunit C gene (xmoC) recovered from carbonate-associated MAGs and metagenomic assemblies, as well as reference sequences. The clades for ammonium (amoC) and methane CuMMO (pmoC) are well supported by experimental data. Although Mycolicibacterium (synonym Mycobacterium) has been shown to oxidize butane, and ETHIRO (unpublished, isolate lost) has been suggested to oxidize ethane, the diversity within these groups is large, and, especially for the seep sequences, the substrates represent hypotheses. The Methylophagaceae xmoC (D, blue) has an unknown substrate and CuMMO genes within Methylophagaceae have not been reported previously. Pmo refers to the genes encoding particulate methane monooxygenase, pxm to a version that some aerobic methanotrophs encode with unknown function, amo to the ammonia monooxygenase, and xmo to genes encoding CuMMO more generally. Sequences that were recovered from individual assemblies or co-assemblies, and were not part of a MAG, could not be assigned to specific taxonomic groups. Proteobacteria and Pseudomonadota refer to the same phylum. Collapsed clades are labeled with the number of sequences recovered in this study (see uncollapsed tree Fig. S14). The scale bar represents one amino acid substitution per site. Bootstrap support is indicated by black circles (> 90%) or white circles (60%–89%). Abbreviations: seqs, sequences; tent., tentatively.

Discussion

Here we demonstrate methane seep carbonate rocks host distinct surface and interior microbial communities both with previously unrecognized ecophysiology and diversity, including Ca. Desulfaltia, a potential new clade of ANME partner bacteria, and a potential new metabolic capability within Methylophagaceae containing CuMMO genes (Figs. 3, 8). We further show that the endolithic carbonate AOM community is resilient [78], maintaining viability and the capacity to serve as methane sink over periods with fluctuating methane seepage (Fig. 6). The combination of geochemical, microscopy-based cell-specific activity alongside DNA-based analyses further demonstrated the occurrence of dormant/dead ANME cells alongside active methanotrophic archaea (Fig. 5), revealing ecophysiological aspects of these rock-hosted archaea, not evident from any single method alone. We uncovered a carbonate-surface ANME-SRB biofilm exposed to low-oxygen seawater off of Southern California, similar to ANME-SRB mats reported from the euxinic Black Sea. The carbonate surfaces further featured a sulfide-oxidizing bacterial mat, and another community without dominant members, with implications for carbonate dissolution/precipitation. The surfaces further harbored a distinct aerobic methanotroph community that may represent a source of epibionts already adapted to surface growth for recruitment by seep animals (Fig. 7). Using metagenomics, we found uncharacterized divergent CuMMO genes associated with seep carbonates, further expanding the known CuMMO diversity.

Ecological insights into carbonate-hosted ANME-1 and their potential SRB partners

ANME-1 members are abundant in many seep carbonates [17], but an investigation of endolithic ANME-1 genera had been lacking. Sequences of the prevalent genus, Ca. Methanophaga, have previously been recovered from cold seep sediments and carbonates [19, 79, 80]. We recovered MAGs and 16S rRNA gene amplicons of two additional genera, QENJ01 and QEXZ01, which have previously been found in cold seep and hydrothermal vent sediments [80, 81]. Sequences of Ca. Methanoalium-2 (JACGMN01), detected primarily in chimney interiors at low relative abundance in our 16S rRNA gene amplicon survey, also occur in marine seeps and terrestrial sites [79, 80, 82]. Ca. Methanoalium MAGs contain hydrogenases unlike related ANME and Ca. Methanoalium has been proposed to be methanogenic [79].

Alongside ANME-1, Ca. Desulfaltia (Seep-SRB1b) with unknown ecology [83], was the most enriched SRB MAG inside carbonates (Fig. 3), which we hypothesize may represent a new ANME-1 syntrophic partner. This group is a sister lineage to syntrophic Seep-SRB1a (Fig. 4), and MAGs recovered from methane seeps, euxinic Black Sea waters, and groundwater [84–87] often co-occurred with ANME. The active endolithic ANME-1 in our study were primarily recovered as single cells rather than aggregates, but we cannot rule out the possibility of disaggregation of loosely associated consortia during cell separation from the rock matrix. ANME-1 have previously been observed to form tight consortia with SRB, and loosely associated aggregations with or without partner bacteria [51, 88]. What this variation means for the relationship between ANME-1 and SRB requires further study using careful sample preparation, and ideally analysis directly within the rock matrix. However, in our genome analysis of related Ca. Desulfaltia MAGs we observed encoded multiheme cytochromes (oetAB), predicted to be used in direct interspecies electron transfer with ANME [61] and further study is needed to confirm its syntrophic partnership with ANME.

Activity and viability of ANME cells in carbonates with low AOM

This study covers a large range of carbonate-associated AOM rates. Chimlet (840–3200 nmol cm⁻³ d⁻¹) had comparable rates to the highest carbonate AOM rates (440–5500 nmol cm⁻³ d⁻¹) reported from Point Dume, another Southern California seep [15], whereas the Del Mar outcrop carbonates had rates comparable to the lowest rates reported in [15] (5 nmol cm⁻³ d⁻¹). Nevertheless, differences in incubation conditions between the studies may affect the rate comparability, e.g. higher methane concentrations (1.1 mM [15] vs. 3.5 mM) could have led to comparatively higher rates under our conditions. Unexpectedly, low AOM carbonates maintained high 16S rRNA gene amplicon and metagenome-based ANME relative abundances, whereas BONCAT-FISH assays suggested ANME cells were either dormant or dead, or rare. Preservation of extracellular DNA may also explain the observed ANME-DNA in low AOM carbonates, potentially well preserved through the sheltered environment, buffered pH, and low temperature [89, 90]. However, extracellular DNA cannot completely explain the detected ANME-DNA, as the reactivation of low AOM carbonates (Fig. 6) showed that at least a few (because of the long lag time) ANME cells must have remained viable and capable of reactivation of AOM and exponential growth.

Carbonates are potential long-term methane sinks

Methane seepage fluctuates [10], and reactivation of low AOM seep carbonates with renewed methane exposure had not previously been shown. Here, we experimentally reactivated low AOM carbonates with methane, showing that carbonates may remain potential ANME-SRB habitats and methane sinks over carbonate lifespans of hundreds to thousands of years [3–6]. The apparent 44-day doubling time was similar to the fastest reported ANME-SRB doubling time of 0.7–1 months measured in sediment from a seep-periphery [91] (literature comparison, Supplementary Table S5). Presumably low initial ANME-SRB cell numbers in our low AOM carbonates and the seep periphery sediment may allow fast growth through relaxed resource or space limitations. Given that carbonates precipitate over hundreds to thousands of years with layers of different ages [4], it is possible that a successively changing ANME-SRB community has facilitated precipitation. Whether the ANME-SRB community that initially promoted carbonate formation persists within the rock or changes over time—and on what timescales—remains to be investigated.

Metabolic range and role of the carbonate surface community

Anaerobes typically do not tolerate prolonged oxygen exposure [92]. Hence, the black-colored ANME-SRB surface-biofilms on the SMM800 carbonates occurring in direct contact with the overlying oxic seawater was unexpected (Fig. 2). Site SMM800 has high methane seepage and occurs within an oxygen minimum zone (8 μM oxygen), which may explain their ability to colonize the chemoherm surfaces. Co-occurring sulfide-oxidizing bacteria may further promote anoxic conditions at the surface [20, 93]. These biofilms resemble the microbial reefs described from the euxinic Black Sea seeps with cm-scale ANME-dominated mats [59]. Recently, ANME and related methanogens were hypothesized to produce black amorphous carbon [94]. Here, the black color of the ANME-SRB biofilm dissipated over time in fixative, arguing against amorphous carbon, and for unstable pigments or iron-sulfide minerals.

Sulfide-oxidizers and aerobic methanotrophs produce acidity that can dissolve seep carbonates [21, 22]. Our in-depth characterization of surface-associated microbial communities across different seep activities suggests that the metabolic potential for dissolution is variable. Specifically, communities with less sulfide-oxidizing bacteria that were associated with low AOM carbonates, as described from carbonate colonization experiments [17], likely have a lower carbonate dissolution potential than sulfide-oxidizer dominated communities like on R9 (Fig. 1, 3). The ANME-SRB biofilms at SMM800 may continue to precipitate carbonate and protect the carbonate surface from corrosive co-occurring sulfide-oxidizers. Targeted investigations are needed to better constrain the spatial extent, environmental context, and contribution of surface microbial communities to carbonate dissolution and precipitation in the deep sea.

Microbial lineages were found to overlap between carbonate and animal surfaces. For example, the Del Mar carbonates shared the aerobic methanotrophic lineages IheB2–23 and MMG-2, and the methylotrophic group Methylophagaceae with carbonate-associated seep invertebrates, including hydroids (Fig. 7) and sea spiders (pycnogonads) [66, 95, 96]. These shared taxa point to the potential importance of the carbonate surface community in animal epibiont recruitment and exchange.

Here we recovered a Methylophagacea MAG of the genus GCA-002733105 (bin_133) encoding CuMMO genes with homology to methane, ammonia, and hydrocarbon CuMMO genes (Fig. 8). GCA-002733105 without CuMMO genes have been described, e.g. as symbionts within bathymodioline mussels supported by C1-compounds from methanotrophs [97]. We speculate GCA-002733105 members encoding CuMMO genes may have a methane-oxidizing potential, as methylotrophic capabilities are present and well documented within Methylophagaceae [98, 99]. The Methylophagaceae xmoC falls outside known methanotrophic pmoC genes, suggesting a divergent evolutionary history. Alternatively, e.g. ammonia or short-chain alkanes may be likely substrates. Future cultivation or enrichment are needed to determine taxonomy and substrate specificity of the recovered divergent xmoC genes (Fig. 8), which would yield a more complete picture of CuMMO evolution and function.

In this study, we advanced the understanding of the diversity of microorganisms, their metabolic potential, and activities within and on the surface of seep carbonates over a range of AOM activities. We identified a potential new ANME-partner, Ca. Desulfaltia, which together with further validation, might expand the known diversity of ANME-SRB partnerships. Our results emphasize that DNA sequences do not always equate microbial activity, and ecophysiological measurements allow deeper insights into dynamics and physiological states of environmental microbes. By reactivating low AOM carbonates, we showed that seep carbonates remain potential ANME-SRB habitats, even over ceasing and recurring methane seepage, acting as potential methane sinks over carbonate lifetimes of 1000s of years. Further, we revealed the carbonate surface community as a distinct seep assemblage that deserves further attention, with potential to play a role in carbonate precipitation or dissolution, as a possible reservoir for animal epibionts, and as a host for divergent CuMMO diversity. Finally, carbonate-hosted abundant and active microbes raise the question if and which other rock types in the deep sea and beyond [100, 101] may host microbial communities contributing to the global elemental cycles.

Author contributions

Conceptualization: MJM, VJO, AC; Data curation: MJM, RM, SAP; Formal analysis: MJM, RM, AC; Funding acquisition: VJO, MJM; Investigation: MJM, SAP, SAC, AKN, VJO; Methodology: MJM, VJO, SAC, AKN; Project administration: VJO, MJM; Resources: VJO; Supervision: VJO; Validation: MJM; Visualization: MJM, with input from all authors; Writing—original draft: MJM; Writing—review & editing: all authors

Conflicts of interest

The authors declare no competing interests.

Funding

This research was supported by the National Science Foundation (OCE-2048666) to VJO and the Swiss National Science Foundation Postdoctoral Fellowship (P2EZP3_195375) to MJM. This work is further supported by the U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research under Award Number DE-SC0022991 to VJO.

Data availability

All raw reads generated in this study and selected high quality MAGs have been submitted to NCBI under project number PRJNA1196099. All MAGs are available from FigShare (10.6084/m9.figshare.28232936).

Disclaimer

This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.