Bacteria are important dimethylsulfoniopropionate producers in marine aphotic and high-pressure environments

Environmental parameters of deep ocean seawater and sediment

Challenger Deep seawater and surface sediment samples were taken from its entire ~11,000 depth profile (Fig. 1a and Supplementary Table 1). The clines in temperature (29.8 °C in surface waters, decreasing to ~1.0 °C below 3000 m) and pressure (0.1 MPa in surface waters to ~104 MPa at the bottom of the trench) were recorded. The waters were oxic throughout the water column and the salinity ranged between 34 and 35 Practical Salinity Units (PSUs) (Supplementary Table 1). Seawater total DMSP and DMS concentrations were similar to those in previous studies^21,26,27,28 and were positively correlated with Chl-a levels, being highest in the Chl-a maximum layer (10.51 × 10⁻³ nmol ml⁻¹ DMSP and 4.97 × 10⁻³ nmol ml⁻¹ DMS) and at lower but relatively stable levels (0.96–2.39 × 10⁻³ nmol ml⁻¹ DMSP and 0.15–1.06 × 10⁻³ nmol ml⁻¹ DMS) in the aphotic waters below 200 m (Fig. 1b, c and Supplementary Table 1). It should be noted that a small portion of this ‘background DMSP’ (<1 nM) detected through alkaline hydrolysis may arise from other organic sulfur compounds that also release DMS upon chemical lysis²⁹. Heterotrophic bacteria, photosynthetic phytoplankton (~1.5% of the total microbial community data determined by metagenomics and 16S rRNA gene amplicon analyses), picoeukaryotes, Prochlorococcus, and Synechococcus were most abundant in the surface waters with the highest seawater DMSP concentration (Supplementary Figs. 2 and 3). Bacteria were present consistently at 10⁵ cells ml⁻¹ levels throughout the water column, whereas the phototrophs were barely detected below the first 200 m (Supplementary Fig. 2). Surface water cyanobacteria likely take up DMSP³⁰ but are unlikely to be significant DMSP producers or cyclers, as few are proven to synthesize it (and at very low levels)³¹ and none contain known DMSP synthesis or catabolic genes. Dinoflagellates of the Dinophysis genus, e.g., Dinophysis acuminata that contains DSYB, has intracellular DMSP levels of 477 mM and large cells (30–120 µm)³², and were the most abundant surface water phytoplankton (up to 73% of detected phytoplankton) (Supplementary Fig. 3). These phytoplankton were likely the major contributors to the DMSP levels detected in the photic waters, although Picoeukaryotes, proposed to contain DSYB⁷, and DMSP-producing bacteria (see below) likely also contributed. However, no eukaryotic DMSP synthesis genes (DSYB or TpMMT) were detected in any metagenomes (Table 1), even from the surface waters, perhaps reflecting the need for deeper sequencing of these waters where phytoplankton are far less abundant than bacteria.

The deep ocean surface sediment DMSP concentrations (3.15–6.14 nmol g⁻¹) were two to three orders of magnitude higher than in the corresponding seawater samples per equivalent mass (ml vs. g) (Fig. 1c and Supplementary Table 1), consistent with previous observations of coastal sediments^10,33. Given the cold, dark, and high-pressure deep-sediment and -water conditions where few live phytoplankton are present (Supplementary Figs. 2 and 3), it is unlikely these phototrophs produced the observed aphotic DMSP in situ, although some is expected to arise from sinking particles, e.g., dead algae and/or fecal pellets^34,35. However, considering the high DMSP turnover rates in photic seawater samples^4,36, it is unlikely that photic-produced DMSP is the source of all aphotic DMSP. We propose that bacterial DMSP synthesis is likely an important contributor to deep-sea sediment and seawater DMSP levels. To test this hypothesis, the distribution and activity of DMSP-producing bacteria was investigated in Challenger Deep samples.

Vertical distribution of DMSP synthesis genes

Bacteria with DsyB or MmtN (and thus the potential to produce DMSP) were always or mostly present, respectively, in all seawater and sediment samples (Fig. 2 and Table 1), and their environmental DsyB and MmtN sequences clustered with ratified enzymes (Supplementary Figs. 4 and 5). Similar proportions of free-living (FL; 0.22–3 μm) and particle-associated (PA; >3 μm) bacteria, which dominated the metagenomes of both these fractions, contained DMSP biosynthesis and catabolic genes (Figs. 2a and 3a, and Table 1), indicating that size fractionation is not a foolproof method of separating DMSP-producing bacteria from phytoplankton. Bacteria with dysB were shown by qPCR and metagenomics to be relatively abundant in the surface waters (dsyB total abundance of 2.61 × 10⁵ copies L⁻¹; 0.78–0.98% of surface water bacteria) representing ~3.49–4.38 × 10³ bacteria ml⁻¹ of surface seawater. These numbers are comparable to those predicted from the ocean microbial reference gene catalog metagenomic database (OM-RGC)³⁷ in Williams et al.¹⁰, at ~4.8–9.6 × 10³ bacteria ml⁻¹. The abundance of these potential DMSP-producing bacteria initially decreased in 1000–2000 m deep seawater samples (3.46 × 10⁴ copies L⁻¹; ~0.43% bacteria at these depths), but then steadily increased with depth to reach maximal levels at 10,500 m (3.95 × 10⁶ copies L⁻¹; 4.03% of bacteria at 10,500 m), which were up to 15-fold higher than in the surface water (Fig. 2b, Table 1, and Supplementary Table 2). All detected dsyB sequences, including 37/162 metagenome assembled genomes (MAGs), were Alphaproteobacterial, mainly Rhodobacterales, Rhizobiales, and Rhodospirillales (Supplementary Data 1). At the genus level, Pseudooceanicola and Roseovarius were the most abundant potential DMSP producers at all depths, with much higher abundances (P < 0.05) in deeper waters (≥4000 m) compared to upper waters (Fig. 2a), suggesting they might be important deep water DMSP producers.

Bacteria with mmtN were always less abundant than those with dsyB in seawater metagenomes, as was also the case in coastal seawater and sediment samples in Williams et al.¹⁰, and their abundance did not obviously increase with seawater depth. However, the highest observed levels of bacteria with mmtN (~1.22% of bacteria) were found in 8000 m deep samples (Table 1). As with dsyB, the majority of mmtN homologs detected were also Alphaproteobacteria, belonging to bacterial genera known to produce DMSP: Thalassospira, Roseovarius, Labrenzia, and Novosphingobium (Fig. 2a; Williams et al.¹⁰). Furthermore, of the nine MAGs containing mmtN, eight were from Alphaproteobacteria and only one was likely from Gammaproteobacteria (Supplementary Data 1). Overall, the proportion of bacteria with the genetic potential to produce DMSP (containing dsyB and/or mmtN) was far higher (P < 0.01) in deeper waters (≥4000 m; ~2.58–5.25%) than in surface waters (~0.90–1.18%) (Fig. 2a and Table 1). Considering that the flow cytometry data of heterotrophic bacterial abundance was ~4.47 × 10⁵ and 1.86–7.56 × 10⁴ cells ml⁻¹ in the surface and deep water (4000–8000 m), respectively, this equates to similar numbers of these DMSP-producing bacteria per ml seawater at the surface and below 4000 m (up to 5.27 × 10³ and 3.99 × 10³ bacteria ml⁻¹ seawater, respectively).

Importantly, the dominant DMSP synthesis gene dsyB was shown to be transcribed in all tested seawater samples, and at the highest levels in the deeper waters (Fig. 2c), supporting the hypothesis that bacteria are important DMSP producers in the aphotic waters. Critically, these predictions from metagenomic and qPCR analyses are most likely an underestimation of bacterial DMSP-production potential, as several DMSP-producing bacteria, including many isolated from Challenger Deep samples, e.g., Marinobacter and Erythrobacter (accounting for ~0.02–2.6% and 0.02–2.8% at ≥4000 m, respectively), lack both dsyB and mmtN in their available genomes, and potentially contain novel DMSP synthesis genes and/or pathways (Supplementary Table 3; Williams et al.¹⁰).

DMSP synthesis in deep sediment

DMSP-producing bacteria with dsyB and/or mmtN were present in all deep trench sediment samples with the highest DMSP concentrations. Furthermore, there were no plastid sequences in 16S rRNA gene amplicon sequencing data from these sediments, implying bacteria as the major producers in these environments. dsyB abundances ranged from ~1 × 10³ copies g⁻¹ (in two samples) to much higher copy numbers of 0.42–1.08 × 10⁵ copies g⁻¹ in the other six sediment samples (Fig. 2b and Supplementary Table 2). Compared to dsyB, mmtN abundance was lower in sediments, with the highest value of 4.07 × 10² copies g⁻¹ found at 6980 m (Supplementary Table 2), but this gene was still detected in most sediment samples, unlike those for the water column. Lower proportions of bacteria (~0.02–0.42%) were predicted to contain dsyB and/or mmtN in trench sediments compared to the corresponding waters (>3.20%) when qPCR was used to estimate bacterial DMSP synthesis potential (Supplementary Table 2). However, qPCR normalization to the 16S rRNA gene is not as accurate as metagenomics methods due to the existence of multiple 16S rRNA gene copies in many bacteria³⁸, and the deep ocean surface sediments likely harbor far more bacteria per equivalent mass than the seawater (Fig. 2b). Indeed, dsyB and mmtN transcript abundances were far higher in all sediments than in water samples (Fig. 2c). These data indicate that DMSP-producing bacteria contribute to photic and aphotic DMSP-standing stocks, and that the significance of their contribution increases in the aphotic waters and sediments, where DMSP-producing phototrophs are far less abundant.

To identify dsyB– and mmtN-containing bacteria in the sediment, clone libraries generated from community DNA were sequenced. mmtN clones were all similar to those genera present in the seawater, mainly Roseovarius and Labrenzia sp. Sediment dsyB clones were classified into six operational taxonomic units (OTUs) at a similarity of 97%, all of which encode for proteins that cluster with functional DsyB sequences (Supplementary Fig. 6). OTU01 (57.14%) and OTU02 (19.50%) were dominant in all sediments and were homologous to Pseudooceanicola atlanticus and Salipiger profundus DsyB, respectively. OTU03, OTU04, OTU05, and OTU06 were found exclusively below 8638 m and were homologous to DsyB in Defluviimonas sp., Labrenzia aggregata, Roseivivax pacificus, and Pseudooceanicola nanhaiensis, respectively (Supplementary Fig. 6). However, the relative abundance of these potential DMSP producers, such as Pseudooceanicola, Salipiger, and Defluviimonas, was very low (<0.002%) in the bacterial sediment communities based on 16S rRNA gene amplicon sequencing. In contrast, the Gammaproteobacteria Marinobacter and Alcanivorax were the dominant bacteria in all sediment samples (up to ~81.60% and 48.10%, respectively) (Supplementary Fig. 7). Some Marinobacter isolates from this study and Williams et al.¹⁰ produce DMSP but lack dsyB and mmtN. This was also the case for Erythrobacter, which constitutes 0.06–18.30% of total bacteria in the tested sediments from 5525 m to 10,911 m (Supplementary Fig. 7). These results suggest that uncharacterized bacterial DMSP production genes and/or pathways exist and are important in these deep ocean sediments. Without knowing functional reporter genes for DMSP production in these bacteria, we are likely vastly underestimating bacterial DMSP production potential in all seawater and sediment samples.

DMSP producing isolates in seawater

Bacterial isolation experiments were performed on all water samples (0–10,400 m) and 22 of 210 isolates produced DMSP under laboratory conditions (Supplementary Table 3). As expected^6,10, most of these were Alphaproteobacteria and contained dsyB. However, several were Gammaproteobacteria and Actinobacteria, none of which gave dsyB or mmtN PCR products when tested with their respective degenerate primers¹⁰ (Supplementary Table 3), meaning they may contain novel DMSP synthesis pathways and/or genes. Seven tested DMSP-producing hadal isolates were all shown to produce DMSP when grown under physiologically relevant hydrostatic and temperature conditions (4 °C and 60 MPa) with no added methylated sulfur compounds (Supplementary Table 4). These included Pseudooceanicola, Roseovarius, Labrenzia, and Erythrobacter isolates, which represented 0.89–3.39% of hadal seawater communities (Supplementary Fig. 8), further supporting these bacteria as significant contributors to the DMSP stocks detected throughout the aphotic water column.

Vertical distribution of DMSP catabolism genes

Given that DMSP and DMS were detected throughout the Challenger Deep water column and DMSP was concentrated in the sediment, microbial samples were analyzed for their potential to catabolize DMSP. The surface water samples harbored huge (~44.43%) bacterial populations containing the genetic potential to catabolize DMSP, equivalent to ~1.98 × 10⁵ bacteria ml⁻¹ seawater. Consistent with previous studies^39,40,41, dmdA was the most abundant DMSP catabolic gene detected in all water samples (Table 1). Bacteria with the potential to demethylate DMSP (mainly SAR11, with Rhodobacterales and the SAR116 clade bacteria also detected) were most abundant in the surface waters (Fig. 3a and Supplementary Fig. 9). Surface water samples contained the highest detected levels of dmdA (~2.22 × 10⁷ copies L⁻¹), with ~35.73% and 12.73% of FL and PA bacteria, respectively, predicted to contain this gene (Fig. 3b and Table 1). In these surface FL samples, dmdA was ~4-fold higher than the sum total of DMSP cleavage genes (Table 1), suggesting DMSP demethylation is likely the dominant process in the surface waters.

The algal DMSP lyase Alma1⁴² was not detected in any trench samples, suggesting that Alma1-containing phototrophs are not major producers of DMS via this pathway in the tested photic and aphotic samples. In contrast, at least three or more of the seven bacterial DMSP lyase genes (ddd)² were detected in every water sample (Fig. 3c and Table 1). dddP was the most abundant DMSP lyase gene in the surface waters, with 3.51 × 10⁵ copies L⁻¹ detected by qPCR (Supplementary Table 2) and ~6.48% of surface ocean bacteria (~2.90 × 10⁴ bacteria ml⁻¹) predicted to contain this gene—the only DMSP lyase in >1% of bacteria at the surface. The dddK, dddW, and dddY genes were only predicted to be in 0–0.26% of the seawater bacteria (Fig. 3c and Table 1). These metagenome values are lower than predicted from the OM-RGC database, comprised largely of surface ocean bacteria⁷. The reasons for these discrepancies are likely site- and/or season-dependent.

Given only the surface waters influence the atmosphere, then dddP-containing bacteria are likely key contributors to the highest detected DMS levels at these sites (Fig. 1c), a fraction of which is transferred to the atmosphere. Seawater DddP homologs clustered into four major groups (Supplementary Fig. 10). Group I was the most abundant and closely aligned to DddP from Rhodobacteraceae and some Phyllobacteriaceae bacteria. Group II proteins closely resembled Alphaproteobacterial DddP, with SAR116 clade being the dominant form. Groups I and II were most abundant in the surface waters. Groups III and IV had multiple representatives, including Alphaproteobacteria, Gammaproteobacteria, Betaproteobacteria, and Actinobacteria (Supplementary Fig. 10), suggesting lateral gene transfer event/s⁴³. dddP was found in 43% of MAGs (69), predicted to be Alphaproteobacteria, Gammaproteobacteria, Acidimicrobiia, Bacteroidia, SAR324, Nitrososphaeria, and Anaerolineae (Supplementary Data 1).

Bacterial DMSP catabolism was also likely important in aphotic 2000–8000 m deep waters, with 14.28–36.88% of bacteria predicted to contain a DMSP catabolic gene. dmdA was still the dominant gene at these depths, predicted to be present in 5.43–26.66% of bacteria, but its relative abundance decreased with depth (Fig. 3a, b, Table 1, and Supplementary Table 2). Howard et al.⁴⁰ detected no dmdA genes in 500–4000 m deep Pacific Station Aloha seawater samples, likely due to lower sequencing depth (8.86–11.18 Mb)⁴⁴ compared to sequencing performed here (13.67–16.54 Gb) and/or the more extensive dmdA gene probe sequences used in this study (Supplementary Table 5). The abundance of Alphaproteobacterial dmdA generally decreased with water depth, whereas those dmdA sequences from Gammaproteobacteria and Actinobacteria did not vary with depth (Fig. 3a). Of 162 MAGs, 58 contained dmdA, likely from Alphaproteobacteria, Gammaproteobacteria, Acidimicrobiia, SAR324, and Nitrososphaeria (Supplementary Data 1). Interestingly, the relative abundance of bacteria with DMSP lyases significantly increased in these deeper waters (2000–8000 m), with cumulatively more ddd genes observed in metagenomes from 4000 m to the trench bottom, compared to dmdA (Table 1). DddP was still the dominant DMSP lyase in the 2000–8000 m deep waters (averaging 4.84%), but DddQ (up to 3.55%), DddL (up to 4.61%), and DddD (up to 1.61%) were better represented in these waters compared to the surface waters (Fig. 3c and Table 1). Seawater DddQ sequences were most similar to those in the Rhodobacteraceae, including Ruegeria, Leisingera, and Roseovarius (Fig. 3c). All DddL sequences were homologous to Gammaproteobacteria, represented by Marinobacter. In comparison, the DddD homologs varied through the water column, with surface waters containing Alphaproteobacterial Sagittula homologs, and Gammaproteobacterial Halomonas homologs being dominant in 8000 m samples (Fig. 3c). dddQ (8 MAGs), dddL (37 MAGs), dddD (47 MAGs), dddK (2 MAGs), and dddW (7 MAGs) were also represented in the 162 MAGs (Supplementary Data 1). Although the relative abundance and therefore the likely importance of DMSP lyase genes in these microbes increased in the 2000–8000 m deep waters compared to the surface, their absolute abundance did not necessarily increase, due to ~4-fold more bacteria being present in the surface waters, e.g., dddP copies L⁻¹ were most abundant in the surface waters based on qPCR results (Supplementary Table 2). Furthermore, 43 of 210 bacterial isolates had the ability to cleave DMSP, 29 being from the 2000–8000 m deep water samples (Supplementary Table 6).

Metagenomics showed there to be a steep decline in DMSP catabolic potential in 9600 m deep waters and below, with no ddd or dmdA gene predicted in more than 1.65% of the bacteria. Indeed, dmdA was absent in 10,400 m deep metagenomes. Despite this, qPCR data showed no correlation between dmdA and dddP absolute gene abundance and depth, other than the highest levels being in the surface waters. It is possible that there are more bacteria in the deepest waters that were not assayed by flow cytometry. Perhaps in these waters there is a stronger requirement to synthesize and store DMSP for its anti-stress properties than to catabolize it; thus, a lower proportion of bacteria in the community would have this ability.

In contrast to most seawater samples, in which dddP and dmdA were abundant, these DMSP catabolic genes were at their lowest detected levels in the deep ocean sediments that contained the highest DMSP concentrations. The DMSP lyase dddP was undetected by qPCR in all sediment samples, and only 8.30 × 10¹–2.18 × 10³dmdA copies g⁻¹ were observed in the hadal sediments (Fig. 3b and Supplementary Table 2). This could suggest that the primers are not detecting deep sediment variants of these genes, that they are scarcely present in these environments and are thus not as important in hadal sediments as they are in seawater, or that other Ddd enzymes and/or isoform enzymes exist in bacteria in these sediments. Further work measuring DMSP synthesis and catabolic process rates and/or transcript/protein abundance is required to better establish the importance of bacteria in these processes throughout the Challenger Deep water and sediment samples, and to test the hypotheses raised here.

A role for DMSP in bacterial hydrostatic pressure tolerance

As dysB abundance and transcripts increased with depth and DMSP catabolic potential was less prominent in the deepest seawater and sediment communities, the hypothesis that DMSP may help organisms to survive under deep ocean hydrostatic stress was tested. DMSP-producing bacteria, isolated from 8000 m (P. nanhaiensis ZYF240) and 9600 m deep (L. aggregata ZYF612) Mariana Trench seawater, both exhibited significantly enhanced DMSP production per colony forming unit (CFU) with increasing pressure (Fig. 2d) with no added methylated sulfur compounds. Another two isolates (Pelagibaca bermudensis J526 and Marinibacterium sp. La6) from surface seawater also showed the same result. Furthermore, DMSP-producing bacteria (wild type) could survive deep ocean pressure (60 MPa) far better than dsyB⁻ mutant strains unable to produce DMSP, with the phenotype being restored by cloned dsyB or when DMSP was provided (Fig. 2e). This provides convincing evidence in at least some marine bacteria for a new role for DMSP in protecting cells against the high hydrostatic pressures that exist in the deep ocean.

Source: Ecology - nature.com