Introduction

Dimethylsulphide (DMS) is one of the most abundant volatile organic sulphur compounds with an estimated global production of over 300 million tons each year [1]. DMS is also the largest source of biogenic sulphur in the atmosphere, where its oxidation products aid in cloud condensation and influence the atmospheric chemistry and potentially the Earth’s climate [2].

The main precursor of DMS in the environment is dimethylsulfoniopropionate (DMSP), an abundant osmolyte (~ 109 tonnes annually) produced by marine algae, phytoplankton and plants such as Spartina and sugar cane [3]. Recent studies have shown that bacteria form significant quantities of DMSP in both oxic and anoxic coastal and marine sediments [4, 5], suggesting these ecosystems to be important environments for DMS production. Other key sources of DMS in sediments are the degradation of sulphur-containing amino acids and methoxylated aromatic compounds, reduction of dimethyl sulfoxide as well as the methylation of hydrogen sulphide and methanethiol (MT) [6, 7].

In anoxic sediments, DMS can be degraded to potent greenhouse gases methane and carbon dioxide (CO2) by methylotrophic methanogens, further highlighting the significance of DMS [8]. Cultivation-based studies on DMS-dependent methanogenesis showed that this process is carried out by certain methanogens of the genera Methanomethylovorans, Methanolobus, Methanosarcina and Methanohalophilus [9,10,11,12]. In sulphate-containing environments, sulphate-reducing bacteria (SRB) of the genera Desulfotomaculum and Desulfosarcina can also use DMS as a carbon source [13, 14].

Despite the environmental significance of DMS and its degradation products (methane and CO2), the metabolic pathways of DMS-dependent methanogenesis have received little research interest. Generally, during methylotrophic methanogenesis, the methyl group from methylated compounds is transferred to a corrinoid protein via methyltransferases (MT1). Then, the corrinoid protein is demethylated and coenzyme M (CoM) is methylated by the methylcorrinoid:CoM methyltransferases (MT2) [15, 16]. The accepted view is that specific methyltransferases are used for each methylated compound (e.g. DMS, trimethylamine, methanol) although their substrate specificities have not been studied extensively. It was shown that the genes encoding TMA- and DMA-methyltransferase (MttB and MttC) can be co-transcribed in Methanosarcina barkeri [17]. There are only a few studies on the metabolic pathways of DMS-dependent methanogenesis, which used pure cultures of Methanosarcina barkeri and Methanosarcina acetivorans, and suggest that methanogenesis from DMS is catalysed by methylthiol-CoM methyltransferase composed of two subunits (MtsA and MtsB) [16, 18]. Later, fused methylsulphide-specific corrinoid/methyltransfer proteins designated as MtsD, MtsF and MtsH were purified from M. acetivorans with major roles of MtsF and MtsH for producing methane from DMS [19]. Conversely, Fu and Metcalf (2015) showed that M. acetivorans strains require the mtsD gene to carry out DMS-dependent methanogenesis, whilst mtsF and mtsH were not critical for growth on DMS [18]. Nevertheless, the metabolic pathways of DMS-dependent methanogenesis in the environment are undocumented.

Here, we studied the microbial diversity and metabolic pathways underpinning DMS-dependent methanogenesis in anoxic sediments from the Baltic Sea. Permanently hypoxic or anoxic conditions as well as the brackish nature of the Baltic Sea sediments provide an ideal ecosystem to study anaerobic DMS degradation [20]. Our approach combining anoxic sediment incubations, amplicon sequencing, genome-centric metagenomics and metatranscriptomics provides new insight into the sediment depth profile of the methanogen diversity and key enzymes in DMS-dependent methanogenesis.

Materials and methods

Study area and sampling

The study sites were located in Himmerfjärden, Baltic Sea, Sweden (Supplementary Fig. 1). The bay has a salinity between 5 and 7‰, and consists of a series of small depositional basins with maximum water depths between 25 and 50 m that accumulated fine-grained organic-rich sediment. Organic carbon concentrations in the investigated sediments vary between 3 and 4% dry weight [21]. The depth of the sulphate-containing sediments varies between 25 and 40 cm depending on season [22]. Below this depth, sediments show high rates of methanogenesis leading to the accumulation of methane [21]. Bottom waters in the lowermost metre of the bay are oxic or hypoxic throughout the year with concentrations generally above 60 µmol L−1. However, oxygen uptake rates are high so that oxygen penetration depths are only between 0.24 and 0.63 cm [23].

Three sites were sampled using the research vessel R/V Limanda: Station H2 (N 58° 50′ 55, E 17° 47′ 42), H3 (N 58° 56′ 04, E 17° 43′ 81) and H5 (N 59° 02′ 21, E 17° 43′ 59). Duplicated sediment cores were collected using a multicorer (40 cm) and a small Rumohr-type gravity corer (140 cm). The sediment cores were transported to the Askö Laboratory of the Stockholm University Baltic Sea Centre and sliced into seven layers according to the sulphate concentrations of the sediment pore water (0 and 4.5 mM; Table 1) [21]. The sediment slices were vacuum-sealed into gas-tight bags and transported to Queen Mary University of London the next day in a cool box kept below 8 °C. Incubations were set up the same day and a portion of each sediment layer was placed at − 20 °C for DNA extraction.

Table 1 Sulphate concentrations in sediment layers from the three sampling stations. These also represent the sulphate concentrations provided in the incubations
Full size table

Incubation set-up

Triplicate incubations were prepared in an anaerobic glove box (Belle Technology, UK) for each sampling location and depth using 2.5 g of homogenised sediment and 20 mL of artificial seawater (ASW). Two sets of replicated controls were also set up. One set contained no DMS, and the other contained DMS and triple autoclaved sediment to monitor any sediment adsorption of DMS. The ASW consisted of 0.32 M NaCl, 10 mM MgSO4.7H2O, 8.8 mM NaNO3, 3.1 mM CaCl2.2H2O, 10 mM MgCl2.6H2O, 9 mM Trizma base, 0.21 mM K2HPO4.3H2O, trace elements and vitamins [24]. The sulphate concentrations of the incubations were adjusted according to the in situ sulphate concentration of the sediment pore water at each depth (Table 1). The microcosms were incubated in the dark and at 8 °C to avoid the photochemical destruction of DMS [25].

Each sample was amended with DMS as the carbon and energy source. Initially, samples were amended with 2 µmol g−1 wet sediment DMS. After the initial DMS degradation, another 2 µmol g−1 DMS were added. All subsequent additions were 4 µmol g−1 DMS. The incubations were terminated when cumulative methane concentrations became stable (between 82 and 128 days), which corresponded to the total DMS additions of 9.7–51.9 µmol g−1 DMS. After the incubation period, the supernatant and sediment were separated via centrifugation at 1000 × g for 6 min and placed at − 20 °C and − 80 °C, respectively, until further analysis.

Analytical measurements

DMS in the headspace of the incubation bottles was measured on a gas chromatograph (GC; Agilent Technologies, 6890A Series, USA) fitted with a flame photometric detector (FPD) and a J&W DB-1 column (30 m × 0.32 mm Ø; Agilent Technologies, USA). The oven temperature was 180 °C, and zero grade N2 (BOC, UK) was the carrier gas (26.7 mL min−1). FPD was run at 250 °C with H2 and air (BOC, UK) at a flow rate of 40 and 60 mL min−1, respectively. DMS standards (50 µM–10 mM) were prepared by diluting > 99% DMS (Sigma-Aldrich, USA) in distilled anoxic water previously prepared by flushing with oxygen-free N2 (BOC, UK).

Methane and CO2 were measured using GC (Agilent Technologies, USA, 6890N Series) fitted with a flame ionisation detector (FID), Porapak (Q 80/100) packed stainless steel column (1.83 m × 3.18 mm Ø; Supelco, USA) and hot-nickel catalyst which reduced CO2 to methane (Agilent Technologies, USA). The oven temperature was 30 °C, and zero grade N2 (BOC, UK) was the carrier gas (14 mL min−1). FID was run at 300 °C with H2 and air (BOC, UK) at a flow rate of 40 and 430 mL min−1, respectively. The GC was calibrated against certified gas mixture standards (100 ppm methane, 3700 ppm CO2, 100 ppm N2O, balance N2; BOC, UK). The total methane concentrations in the incubation bottles also included dissolved methane in ASW calculated using the atmospheric equilibrium solubility equation as a function of temperature, salinity and headspace concentration [26].

The total CO2 production was the sum of the CO2 concentration in the headspace and the total dissolved inorganic carbon (DIC) in the water phase. The CO2 in the headspace was measured using a gas chromatograph as above. Total DIC was measured as CO2 in the headspace after the supernatant of the slurry was fixed with 24 µL ZnCl2 (50% w/v) and acidified with 100 µL 35% HCl. An inorganic calibration series (0.1–8 mM) of Na2CO3 was used as a standard (Sigma-Aldrich, USA).

Sulphate concentrations were measured at the end of incubation, using porewater filtered through 0.2 μm syringe filters (PTFE hydrophilic; Fisher Scientific, USA). An ICS-5000 Dual Gradient RFIC Ion Chromatograph (Thermo Fisher Scientific, USA) equipped with a Dionex IonPac AS11-HC-4 μm column (2 × 250 mm) and a Dionex IonPac AG11-HC-4 μm guard column (2 × 50 mm) was used. A gradient of 1.5–22 mM KOH (Dionex EGC 500 KOH, with CR-ATC column) was applied as eluent.

DNA extraction, PCR and quantitative PCR

DNA was extracted from both the original and DMS-incubated sediment samples using the DNeasy Powersoil kit (Qiagen, NL) following the manufacturer’s instructions using FastPrep96 (MP Biomedicals, USA) at maximum speed for 40 s. The mcrA gene, which encodes the \(\alpha\)-subunit of the methylcoenzyme M reductase in all known methanogens, was amplified using the mcrIRD primer set [27]. The PCR solution contained 1 µL of DNA template, 1 µL of each primer (10 µM), 25 µL 2 × MyTaq HS Red Mix (Meridian Bioscience, USA) and 22 µL ultra-pure water. PCR conditions were 95 °C for 5 min and 39 cycles of 95 °C for 1 min, 51 °C for 1 min, 72 °C for 1 min and a final extension at 72 °C for 5 min. PCR products were cleaned using JetSeq Clean beads (1.4x; Meridian Bioscience, USA) following the manufacturer’s instructions.

Quantitative PCR (qPCR) of the mcrA gene was carried out in triplicate using the primers mlas-mod-F and mcrA-rev-R, a CFX384 Touch Real-Time PCR Detection System (Bio-Rad Laboratories, USA) and a low volume liquid handling robot for automation (Mosquito HV, SPT Labtech, UK) [28, 29]. Each reaction contained 4 ng/µL DNA, 10 µM of each primer, 2.5 µL SensiFAST SYBR (No-ROX; Meridian Bioscience, USA) and 1.8 µL ultra-pure water. The cycling conditions were 95 °C for 3 min, followed by 40 cycles of 95 °C for 15 s, 65 °C for 30 s and 72 °C for 20 s. A melt curve analysis was performed by increasing the temperature from 65 to 95 °C in 0.5 °C increments. Standard curves were produced using a serial tenfold dilution of clones containing the mcrA gene. The reaction efficiency was between 90 and 110%, and the R2 value for the standard curve was > 99%.

High-throughput sequencing and sequence analysis

For the sequencing library preparation, a second PCR was carried out to attach overhang adapters to the cleaned-up PCR products using mcrA primers containing 5′ overhang adapters (10 µM). The PCR conditions were 95 °C for 3 min, 15 cycles of 95 °C for 20 s, 55 °C for 15 s, 72 °C for 15 s and a final extension step at 72 °C for 5 min. After clean-up, the PCR products were further amplified for the addition of dual indices using 2 µL of the clean barcoded PCR products, 1 µL of each primer (5 µM), 12.5 µL 2 × Q5 Hot-start Ready mix (NEB, USA) and 8.5 µL ultra-pure water. The PCR conditions were as above. All PCR products were normalised using the SequalPrep Normalization Plate kit (Invitrogen, USA) and sequenced on a MiSeq Next Generation sequencing platform (2 × 300 bp; Illumina; USA).

The amplicon sequences were analysed using QIIME2 2021.11 on Queen Mary University of London’s Apocrita HPC facility, supported by QMUL Research-IT [30, 31]. Taxonomy was assigned to Amplicon Sequence Variants using Naive Bayes classifiers, trained using a custom mcrA database compiled using FunGenes, Python 3.10.8 and the RESCRIPt package in QIIME2 [32, 33].

Statistical analysis

All statistical analyses including the calculations of the Shannon index, permutation tests of multivariate homogeneity of group dispersions (999 permutations), ANOVA, pairwise PERMANOVA (9999 permutations) and the principal coordinate analyses (PCoA) with Bray–Curtis dissimilarity were carried out and visualised using microeco and ggplot2 in RStudio (2022.07.1) [34,35,36]. Spearman’s correlation analysis (rs) between the first three PCoA coordinates and the consumed DMS, produced methane and CO2 and sulphate concentrations was conducted using PAST 4.2 [37].

Metagenomics analysis

Paired-end (2 × 150 bp) metagenomics sequencing of the DMS-incubated sediments from 19 to 22 cm depth from the three stations was conducted on the Illumina NovaSeq 6000 platform at US Department of Energy (DOE) Joint Genome Institute (JGI). The quality of the DNA was analysed using Nanodrop One (Thermo Scientific, USA) and Qubit 2.0 Fluorometer (Invitrogen, USA). The A260/280 ratio of the samples was between 1.6 and 2.0, whereas the A260/230 ratio was between 1.8 and 2.2. The DNA concentrations were between 10 and 15 ng/µL.

In total, 154 Gb of sequencing data corresponding to 64.5 Gb from station H2, 41.7 Gb from H3 and 47.9 Gb from H5 were obtained. The data analysis was performed by JGI following the well-established JGI workflow, including assembly, feature prediction, annotation and binning (Supplementary Information) [38].

Metagenome-assembled genomes (MAGs) were recovered using MetaBAT 2.12.1 [39]. Genome completion and contamination were estimated using CheckM 1.0.12 [40]. The genome bins were assigned as high (HQ) or medium quality (MQ) according to the Minimum Information about a Metagenome-Assembled Genome (MIMAG) standards [41]. Integrated Microbial Genome (IMG) and GTDB-tk (0.2.2) databases were used to infer taxonomic affiliations [42, 43].

A list of 78 genes involved in methanogenesis (15 genes specific to methylotrophic methanogenesis) was compiled using the MetaCyc and KEGG databases (Supplementary Table 2) and quantified in the metagenomics datasets and metagenome assembled genomes (MAGs) [44, 45]. All absolute abundance counts were normalised using the CPM (copies per million) normalisation method and log-transformed. R (4.2.1) and ggplot2 were used to make a heatmap showing the log10(CPM) values for each gene [34, 36, 46].

Metatranscriptomics analysis

Metatranscriptomics were conducted on the DMS-incubated sediments (19–22 cm depth) from the three stations. Total RNA was extracted using the ZymoBIOMICS RNA miniprep Kit (Zymo Research, USA) according to the manufacturer’s instructions. The quality was checked using a Tapestation 2200 (Agilent Technologies, USA) and the absorbance was measured using a Nanodrop One (Thermo Scientific, USA). The A260/280 ratio was greater than 1.8 in all samples. The concentration of total RNA ranged between 17 and 88 ng/µL (Qubit 2.0; Invitrogen, CA, USA). DOE JGI performed the metatranscriptomics sequencing (2 × 150 bp) on the Illumina NovaSeq S4 platform and analysed the sequences following a well-established JGI-created workflow (Supplementary Information). Metatranscriptomics sequencing of the sample from station H2 was not successful. In total, 98.1 Gb of sequencing data corresponding to 44.4 Gb for the H3 and 53.7 Gb for the H5 sample were obtained.

A total of 78 methanogenesis-related genes (Supplementary Table 2) were screened within the metatranscriptomes. Pyrrolysine, an in-frame amber codon (UAG), which does not act as a stop codon during synthesis, was searched within the methyltransferase gene fragments using JGI’s Chromosome Viewer [47]. If one fragment contained pyrrolysine, the two fragments were merged. Then, the absolute abundance of the genes was calculated, normalised and log-transformed. Fragments per kilobase of transcript per million fragments mapped reads (FPKM) were calculated and a heatmap was created using ggplot2 [36, 48].

Methanolobus genome analysis for the mts genes

A total of 16 complete Methanolobus genomes were collected from NCBI and JGI IMG/MER databases [42, 49]. Using protein Basic Local Alignment Search Tool, these genomes were screened for the presence of the Mts proteins (Q48924, Q8PUA8, Q8PUA7, AAM04298.1, WP_048180685.1, AAM07726, WP_048180700.1, AAM07897.1, WP_048177073.1) downloaded from the Uniprot and NCBI databases [49, 50].

Phylogenetic analysis

The PhyloFunDB pipeline was used for the construction of the phylogenetic tree of the mcrA gene. The mcrA sequences from uncultured methanogens were manually removed and those from Methanolobus oregonensis, Methanolobus taylorii and Methanolobus tindarius were added. IQ-TREE (1.6.12) was used to create the phylogenetic tree using Methanopyrales as the out-group with 1000 bootstrap replications [51]. ModelFinder was used to find the best-fit model (mtZOA + F + G4) [52].

Phylogenetic analyses of methyltransferase and corrinoid proteins were performed in MEGA7 using the neighbour-joining method with 100 bootstrap replications and the Poisson correction method [53].

Results

Sediment depth profiles of DMS, methane and CO 2

We terminated the incubations between day 82 and 128, when cumulative methane concentrations became stable (Supplementary Fig. 2). We observed a lag phase in all incubations before methane production was detected although the DMS degradation began within the first couple of days, suggesting that SRB started to consume DMS before methanogens. We observed DMS degradation and accompanying methane and CO2 productions in all sediment incubations except for the 60–65 cm (bottom) sediment layer from the H2 station, where no methane production was observed despite DMS degradation.

The greatest DMS consumption and methane production was recorded in the incubations with the 1–2 cm sediment layer (D2) from the H3 station (Fig. 1) although these incubations had the highest initial sulphate concentration (5 mM; Supplementary Fig. 3). The DMS consumption was 48.5 ± 2 μmol g−1 wet sediment, whilst the net methane production was 62.9 ± 1.8 μmol methane g−1 wet sediment, respectively, which correspond to 86% of the theoretical methane yield (72.7 ± 3 μmol g−1 wet sediment) assuming 1 mol of DMS yields 1.5 mol of methane [54]. This indicates that ~ 41 μmol DMS g−1 wet sediment was converted to methane in these samples, suggesting that the rest of the DMS (~ 7.5 μmol DMS g−1, respectively) was degraded by SRB. Supporting this, ~ 77% of the sulphate amended to these incubations was consumed, decreasing the sulphate concentration to 9 ± 0.2 μmol g−1 wet sediment (Supplementary Fig. 3).

Fig. 1
figure 1

Average of total amount of degraded DMS, methane and CO2 per gramme of DMS-amended sediment after 82–128 days of incubation. D1 0–1 cm; D2 1–2 cm; D3 2–5 cm; D4 9–12 cm; D5 19–22 cm; D6 39–43 cm; D7 60–65 cm

Full size image

DMS degradation and net methane production were comparatively low in H2 sediment incubations, where the maximum methane production (13 ± 2.6 μmol g−1) was observed in sediments from 9 to 12 cm (D4), 19 to 22 cm (D5) and 39 to 43 cm (D6) of depth. Stoichiometrically, this corresponds to ~ 8.5 μmol DMS g−1 consumption, however, the actual concentrations of degraded DMS were 20.3 ± 2.5, 18.1 ± 2.5 and 16.6 ± 2.3 μmol DMS g−1, respectively (Fig. 1). Similarly, the highest methane production was 31.4 ± 0.4 μmol g−1 in the H5 sediment incubations of the 1–2 cm depth interval. This methane production corresponds to ~ 21 μmol DMS g−1 degradation, however, a total of 35.9 ± 0.9 μmol DMS g−1 was degraded in these incubations. These results indicate that part of the DMS was degraded via the sulphate reduction route in these incubations. Intriguingly, the sulphate concentrations in the sediments below 9 cm from all three sites increased significantly compared to the initial concentrations (Supplementary Fig. 3). This suggests that hydrogen sulphide produced as one of the end products of DMS degradation was converted to sulphate, which led to cryptic sulphur cycling in these incubations [8].

We also measured CO2 in the incubations as it is one of the metabolic end products of anaerobic DMS degradation via both methanogenesis and sulphate reduction (Fig. 1). In general, the total amount of CO2 was significantly lower than the theoretical CO2 amounts assuming only methanogenesis or sulphate reduction took place (2 mol and 0.5 mol per mol of DMS, respectively), implying that CO2 was simultaneously consumed in our incubations [13, 54].

Depth profiles of methanogen diversity and abundance

We characterised the depth profiles of methanogen diversity and abundance in our original and DMS-amended sediment samples via sequencing and quantifying the mcrA gene.

There was a statistically significant difference in methanogen diversity between the original and DMS-amended sediment samples (PERMANOVA; p < 0.01), whilst there was no difference between the control incubations without DMS and the original sediments, indicating that the shift in methanogen diversity was due to DMS addition.

All original Baltic Sea sediment samples from the surface down to the bottom of the sulphate-methane transition zone (SMTZ) at 19 cm (D5) had strong dominance of Methanolobus with 47–80% of relative abundance (Fig. 2a). Below this depth, the methanogen diversity in the original sediments becomes more varied with Methanoculleus (37–75%), unclassified Archaea (1–36%) and Candidatus Methanomethylophilus (3–42%) in addition to Methanolobus (3–37%), highlighting a shift in methanogen populations below the SMTZ.

Fig. 2
figure 2

a Relative abundance of methanogens at genus level based on mcrA sequencing. Methanolobus dominated in all original sediment down to 19 cm (D1–D4) and DMS-amended sediment except for sample H2D7 where methane production was minimal. b The mcrA gene copy numbers in original and DMS-amended sediment incubations as determined by qPCR. All reactions were set up in triplicate and the average abundance and standard error are shown. c PCoA plot of the mcrA sequences based on Bray–Curtis dissimilarity metrics. Ellipses indicate 95% confidence intervals according to treatment data. Colour indicates treatment (red untreated; green DMS-amended). Shapes indicate sampling site. D1 0–1 cm; D2 1–2 cm; D3 2–5 cm; D4 9–12 cm; D5 19–22 cm; D6 39–43 cm; D7 60–65 cm

Full size image

All DMS-amended samples, except for the H2 and H5 top and H2 bottom layers, where low or no methane production was observed, had a sharp increase in the relative abundance of Methanolobus to 61–99% regardless of the sulphate concentration in the incubations (Fig. 2a). In the H2 and H5 top sediment incubations, unclassified Methanomicrobia increased to 24 ± 3% and 22 ± 5%, respectively, although this taxon was not detected in the original sediment samples. To assess the factors influencing the methanogen diversity in the original and DMS-amended sediments, we conducted a principal coordinate analysis, which clearly separated the original and DMS-amended sediment samples (Fig. 2c). Spearman’s correlation analysis of the first principal coordinate (explaining 24.5% of the total variation in methanogen community composition) correlated positively with DMS degradation, methane and CO2 production (Supplementary Table 1; p < 0.001).

The abundance of methanogens increased significantly in DMS-amended incubations, where methane production was observed compared to the original sediment samples (Supplementary Fig. 4, p < 0.05). However, the correlation between the mcrA abundance and the methane production was not linear.

Taxonomic analysis of metagenomes from DMS-amended incubations

To gain further insight into the microbial populations degrading DMS, we conducted metagenomic sequencing of the DMS-incubated sediments from all three stations at the SMTZ (19–22 cm; D5), where both DMS-dependent methane production and sulphate reduction are likely to happen in situ.

Taxonomic classification of the metagenomes showed that Methanolobus were the dominant methanogen (69–87% of all Archaea; 34–63% of all rRNA genes) in all DMS-incubated samples, whilst the abundance of SRB was at 0.5% of all rRNA genes in all three samples. We also analysed the mcrA sequences retrieved from the assembled metagenomes, which indicated that 35% of the sequences were most closely affiliated with Methanolobus (89.6 to 98.8% similarity; Fig. 3). Furthermore, we successfully constructed 44 MAGs from the metagenomes (Supplementary Table 3). Four of these MAGs were methanogens recovered from the three stations. These medium quality MAGs have completeness ranging between 62.5 and 92.8% and contamination < 2%, but they do not contain all three rRNA genes [41]. We retrieved one mcrA sequence from the methanogen MAG (92% completeness) obtained from the H2 sample (19–22 cm; D5) and it was most similar to the mcrA sequence from Methanolobus vulcani (WP_091708234; 94.6% amino acid similarity). The phylogenetic analysis shows all mcrA sequences in the metagenome datasets and the MAG clustered together with mcrA sequences from cultured Methanolobus species (Fig. 3).

Fig. 3
figure 3

Maximum likelihood phylogenetic tree of the mcrA gene from the cultured methanogens. The tree also contains the mcrA sequences (marked in bold) from the metagenomes and the MAGs obtained within this study. ModelFinder was used to find the best-fit model for the data [52]. Bootstrap values (1000 replicates) were shown as black dots (> 80%) and grey dots (< 80%). The tree is drawn to scale, with branch lengths accounting for substitutions per site. The genus Methanopyrales was used as the outgroup

Full size image

Metabolic pathways of DMS degradation in the sediment incubations

We analysed the metabolic pathways of anaerobic DMS degradation via metagenome and metatranscriptome analyses of the samples from DMS-amended incubations with sediments at 19–22 cm (D5). We screened for 78 genes involved in methane production in the metagenomics and metatranscriptomics datasets, and the constructed methanogen MAGs (Supplementary Tables 2 and 3).

Notably, the relative expressions of the mtsA, mtsB and mtsH genes encoding for MT- and DMS-methyltransferases characterised in M. barkeri and M. acetivorans were low (< 0.01%) or even undetectable in the metatranscriptomics datasets (Fig. 4a). mtsD and mtsF had higher abundances than the other mts genes (0.3% and 0.1%, respectively); however, they were identified as Methanosarcina, Methanolobus, Methanomethylovorans and Methanococcoides with similarities between 82 and 98%. Only 0.1% and 0.055% of the mtsD sequences were affiliated with Methanolobus in H3 and H5 sediment samples, respectively, whilst mtsF was not assigned to Methanolobus in any of the samples. Similarly, the mts genes were absent or in low abundance (< 0.03%) in the metagenomes, and were affiliated with Methanomethylovorans (Fig. 4b). Furthermore, we did not find these genes in the four methanogen MAGs (Supplementary Fig. 5). However, it should be kept in mind that two of the MAGs were less than 70% complete. Surprisingly, the expression of the genes encoding for archaeal trimethylamine (TMA)- and methanol-corrinoid protein co-methyltransferases (mttB and mtaB, respectively) were dramatically high. The relative abundance of mttB was 5.8 and 5% whilst, for mtaB, it was 10.2% and 7.7% in H3 and H5 samples, respectively. Moreover, the relative expressions of the whole gene clusters encoding dimethylamine TMA- and methanol-methyltransferases (mttBC and mtaABC, respectively) were higher (3 and 3.5% and 3.5–4.2%, respectively) than that of the mts gene cluster (< 0.002%; Fig. 4). The genes encoding for TMA- and methanol-methyltransferases were also present in all the metagenomes, where mttB was the most abundantly found gene (~ 6%) involved in methylotrophic methanogenesis (Fig. 4b). This was significantly higher than all other methylotrophic methanogenesis genes searched (p < 0.001; Fig. 4b).

Fig. 4
figure 4

Heatmaps showing expression and abundance of genes involved in methylotrophic methane production. a Metatranscriptomics datasets; b metagenomics datasets. FPKM fragments per kilobase of gene per million reads, CPM copies per million reads

Full size image

The taxonomic profiling of the genes encoding for TMA and methanol methyltransferases and corrinoid proteins (MtaB, MtaC, MttB, MttC) from the metatranscriptome datasets assigned them to Methanolobus (Fig. 5). In line with this and the metagenomics sequence analysis, the entire gene clusters encoding for DMA-, TMA- and methanol-methyltransferases (mtbABC, mttABC and mtaABC, respectively) were also present in the two most complete Methanolobus MAGs (H2D5-Methanolobus and H5D5-Methanolobus; Supplementary Fig. 5).

Fig. 5
figure 5

Phylogenetic tree of (a) MT1 methyltransferase and (b) corrinoid proteins including sequences from metatranscriptomics and MAGs recovered from the metagenomics datasets. The evolutionary history was inferred using the neighbour-joining method. The optimal trees with the sum of branch length of 11.7 and 11.1 are shown for methyltransferase and corrinoid proteins, respectively. Bootstrap values (100 replicates) are shown as black (> 50) and grey (< 50%) dots. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. All ambiguous positions were removed for each sequence pair. There was a total of 637 positions for methyltransferases and 217 positions for the corrinoid proteins in the final dataset

Full size image

We also searched for genes encoding for key enzymes common to all methanogenic pathways (Supplementary Table 2). We found that all the genes in the mcrABCDG operon had a relative expression of > 1% in metatranscriptomics datasets (Supplementary Fig. 6a).

We further showed that the transcripts of several other gene clusters in central methanogenic pathway (e.g. mtrA-H, hdrA-D, mvdADG, frhABDG; Supplementary Table 2) were found at levels 0.26%, 1.75% and 0.84%, respectively. hdrA was found at strikingly high level (6.65%) compared to others, which is likely because this gene is conserved across all methanogens [55]. On the other hand, fpo and vho genes catalysing coenzyme B/coenzyme M regeneration were not transcribed in our sediment incubations. These genes were also absent in the metagenomics datasets (Supplementary Fig. 6b).

To understand whether acetoclastic and hydrogenotrophic methanogenesis pathways were active in our DMS incubations, we searched for genes specific to these methanogenesis pathways (ack, acs, coo, cdh, pta for acetoclastic and fmd, ftr, mch, mer for hydrogenotrophic methanogenesis). All genes except for acs were expressed at < 0.1%, whilst acs was expressed at 4.9% (Supplementary Fig. 7a). It should, however, be kept in mind that methylotrophic methanogens also possess the acs gene [56].

Discussion

Despite the environmental importance of DMS as a methane precursor in anoxic sediments, limited information concerning the microbial diversity and metabolism of DMS-dependent methanogenesis is available. Here, we conducted the first study on the depth profile of the microbial populations and metabolic pathways underlying DMS-dependent methanogenesis in anoxic sediments.

Our sediment incubations have shown that DMS degradation proceeds via both methanogenesis and sulphate-reduction throughout the sediment sampled at the three stations in the Baltic Sea. Higher methane yields from DMS degradation were observed in H3 and H5 stations. This may be due to higher inputs of organic carbon and nutrient from the discharge of an upstream sewage treatment plant, leading to higher rates of carbon mineralisation allowing ultimately methanogenesis to occur [21].

Multiple lines of evidence obtained from the amplicon sequencing, genome-centric metagenomics and metatranscriptomics data pointed that Methanolobus were the dominant DMS-degrading methanogens in our sediment incubations despite varying sulphate concentrations. This methanogen genus was also dominant in the original sediment samples, which suggests that halotolerant Methanolobus carry out methylotrophic methanogenesis in sulphate-bearing sediments of the Baltic Sea and potentially degrade DMS when it is available.

Methanolobus are known DMS degraders with several strains isolated from an oil well, marine, lake and estuarine sediments [10, 11]. We also recently showed Methanolobus to be the dominant DMS-degrading methanogen genus in brackish sediments from the Medway Estuary, UK [8]. Furthermore, a psychrotolerant Methanolobus strain has been isolated from a saline lake sediment in Siberia, indicating that this genus has members that can grow in low temperatures, as were measured in Baltic Sea sediments [57].

An important result of this study was the lack or very low detection (< 0.3%) of the genes encoding for methylsulphide-methyltransferases (mts) in both metagenomics and metatranscriptomics sequences retrieved from incubations, where DMS-dependent methane production was observed. On the contrary, the transcriptional profiles of genes encoding for enzymes related to TMA- and methanol-methyltransferases (mttB and mtaB) showed much higher levels of gene transcription (5.4% and 9%, respectively) in DMS-amended incubations. These highly expressed methyltransferase genes were taxonomically affiliated with Methanolobus, supporting our findings via mcrA sequencing and taxonomic analysis of the metagenomes. We searched for the mts genes in all publicly available Methanolobus genomes and found that they do not contain the mtsA, mtsB and mtsF genes, whilst seven Methanolobus genomes contain the mtsD (identity between 78 and 81%) and mtsH (identity between 66 and 69%) genes (Supplementary Table 4). This, together with our findings, suggests that isolated Methanolobus strains and those in our sediment samples do not use MtsAB when degrading DMS to methane. Nevertheless, it remains possible that certain low-abundance Methanolobus strains in Baltic Sea sediments might degrade DMS via MtsD activity. Intriguingly, previous studies have shown that the transcription of the MttB-family methyltransferases can be induced by non-cognate substrates [17]. Thus, although we cannot rule out the involvement of mtsD genes in DMS-dependent methanogenesis in these sediments, it is more likely that TMA- and methanol-methyltransferases are responsible for DMS degradation.

Our results contradict previous studies, which proposed that each methylotrophic substrate requires a specific enzyme to methylate a corrinoid protein [16, 18]. In addition, Tallant et al. (2001) showed that the methylamine-specific methylcobalamin:CoM methlytransferase, MtbA, did not catalyse the methylation of cobalamine with DMS in Methanosarcina barkeri [58]. However, a recent survey analysing the presence of genes involved in methylotrophic methanogenesis within 465 metagenomes from wetlands, ocean and hypersaline sediments showed a significantly low abundance of the mtsA compared to the mttC and mtaA that encode for TMA- and methanol-dependent methanogenesis genes, respectively [59]. Given the high concentrations of DMS and its ubiquitous precursor DMSP in the environment, it is intriguing to count low levels of mtsA in environmental metagenomes. Hence, we propose that the mtt and mta genes, encoding for TMA- and methanol-methyltransferases, are versatile methyltransferases that can catalyse the transfer of the methyl moity of DMS to a corrinoid protein. This, however, does not exclude the possibility that there are novel methylsulphide-specific methyltransferases yet to be discovered.

Conclusions

In this work, we showed that the genus Methanolobus is the key DMS-degrading methanogens in anoxic brackish sediments. Our study also provided the first evidence that DMS can be anaerobically degraded to methane via the activity of TMA and methanol methyltransferases in some Methanolobus strains. This finding challenges the accepted view that substrate-specific methyltransferases are used in methylotrophic methanogenesis. In light of the significance of this methanogenesis route in coastal and marine ecosystems, it is vital that the metabolic pathways underlying methylotrophic methanogenesis and the regulation of these pathways are unearthed.