Bacterial Taxa That Limit Sulfur Flux from the Ocean

See allHide authors and affiliations

Science  27 Oct 2006:
Vol. 314, Issue 5799, pp. 649-652
DOI: 10.1126/science.1130657


Flux of dimethylsulfide (DMS) from ocean surface waters is the predominant natural source of sulfur to the atmosphere and influences climate by aerosol formation. Marine bacterioplankton regulate sulfur flux by converting the precursor dimethylsulfoniopropionate (DMSP) either to DMS or to sulfur compounds that are not climatically active. Through the discovery of a glycine cleavage T-family protein with DMSP methyltransferase activity, marine bacterioplankton in the Roseobacter and SAR11 taxa were identified as primary mediators of DMSP demethylation to methylmercaptopropionate. One-third of surface ocean bacteria harbor a DMSP demethylase homolog and thereby route a substantial fraction of global marine primary production away from DMS formation and into the marine microbial food web.

Marine phytoplankton synthesize DMSP for use as an osmolyte (1), predator deterrent (2), and antioxidant (3). The degradation of DMSP to DMS and subsequent exchange of DMS across the ocean-atmosphere boundary is the main natural source of sulfur to the atmosphere, amounting to ∼20 Tg of sulfur annually (4). DMS-derived atmospheric sulfur affects cloud formation and the radiative properties of Earth (5). Phytoplankton are known to degrade DMSP to DMS, but efforts to predict global patterns of ocean-atmosphere DMS flux based solely on phytoplankton parameters have been unsuccessful (6). Other members of the marine plankton must therefore influence the production and emission of DMS from the surface ocean (7).

Marine bacterioplankton are known to degrade DMSP by a pathway that first converts DMSP to methylmercaptopropionate (MMPA) in a demethylation reaction, and subsequently to methanethiol (MeSH) (8) or mercaptopropionate (MPA) (fig. S1) (9). The first step of this alternative pathway is crucial to oceanic sulfur emissions, because it removes a methyl group from DMSP and eliminates DMS as a possible degradation product. Furthermore, some of the MMPA-derived sulfur is incorporated subsequently into bacterial amino acids (10) and, through trophic transfers, into the marine microbial food web. Despite the estimated 50 to 90% of DMSP that is metabolized by marine bacterioplankton through this pathway (11, 12), the taxa that mediate DMSP demethylation in ocean surface waters are unknown.

Bacteria in the marine Roseobacter clade have been shown to demethylate DMSP in culture (8). Silicibacter pomeroyi DSS-3 (13) performs both DMSP demethylation to MeSH and DMSP cleavage to DMS (10). A 20,000-member Tn5-based transposon insertion library of S. pomeroyi was screened for interruption of MeSH formation based on failure to produce a thiol from DMSP, and phenotypes of potential mutants were monitored by analysis of sulfur gas formation. A mutant unable to make MeSH yet able to produce DMS at wild-type levels had a transposon insertion in SPO1913 (Fig. 1), a gene encoding a protein in the glycine cleavage T-protein family (Pfam PF01571, Enzyme Commission number DMSP degradation to MeSH was restored by complementation of the mutant in trans with an intact SPO1913 gene (fig. S2). Enzyme assays in cell-free extracts of wild-type and mutant strains showed that SPO1913 encodes the protein responsible for the first step in MeSH formation: the demethylation of DMSP to MMPA (Table 1). This DMSP demethylase gene was designated dmdA.

Fig. 1.

Gene neighborhoods of cultured marine bacteria and selected Sargasso Sea contigs (labeled as IBEA CTG) harboring dmdA genes. Representative sequences that assembled into the Sargasso Sea contigs (i.e., with >97% identity) are indicated on Fig. 2. The P. torquis contaminant dmdA is a partial sequence on a small genome fragment. A, GntR family transcriptional regulator; B, glycine cleavage T-family protein (dmdA); C, dehydrogenase; D, glyoxalase family protein; E, aminotransferase class V; F, deoxyribodipyrimidine photolyase (phrB); G, protein of unknown function; H, acyl coenzyme A (CoA) dehydrogenase; I, acyl CoA synthase; J, hydrolase (mhpC); K, aspartate semialdehyde dehydrogenase; L, succinate dehydrogenase cytochrome b (sdhC); M, membrane protein; N, succinate dehydrogenase (sdhA); O, succinate dehydrogenase Fe-S protein; P, OsmC-like protein; Q, enoyl-CoA hydratase/isomerase.

Table 1.

DMSP demethylase activity in S. pomeroyi and E. coli strains with or without functional dmdA genes (SPO1913 or SAR11_0246) measured as MMPA formation (nmol min–1 mg protein–1) in cell-free extracts. Activity in wild-type extracts was linear with both time and amount of protein, dependent on the presence of DMSP and the coenzyme tetrahydrofolate (THF), and comparable to the rate of MeSH production by whole cells. The limit of detection was 0.02 to 0.05 nmol min–1 mg protein–1. Activityis shown ± SD.

Source of extractDMSP:THF demethylase activity
S. pomeroyi DSS-3, wild-type 0.15 ± 0.02
S. pomeroyi mutant 41-H6, Tn5 inactivation of SPO1913 0
E. coli with pABX101, recombinant SAR11_0246 0.24 ± 0.05
E. coli with pCYB1, vector alone 0

Basic Local Alignment Search Tool (BLAST) searches of genome sequences of other cultured bacteria yielded only two complete dmdA orthologs in any non-Roseobacter genome. Both were from marine bacteria in the SAR11 clade, Pelagibacter ubique HTCC1062 (14) and P. ubique HTCC1002 (15) (Fig. 1). One other partial dmdA sequence (Fig. 1) was found on a small (1.4-kb) fragment of environmental DNA contaminating the genome sequence of the sea ice bacterium Psychroflexus torquis (15); the taxonomic origin of this sequence is unknown (16).

We searched marine metagenomic libraries to determine whether dmdA-like sequences were present in natural bacterial communities. In the Sargasso Sea (11), dmdA homologs were sufficiently abundant to be harbored by about a third of bacterioplankton cells (Table 2). The Sargasso sequences formed four clades distinct from other glycine cleavage T-protein family proteins (Fig. 2). Clade A sequences clustered with DMSP demethylases from S. pomeroyi and other Roseobacters (table S1). Based on the number of clade A homologs relative to Roseobacter-like 16S rDNA sequences (13), at least 80% of Roseobacters captured in the Sargasso Sea metagenome possess a dmdA homolog. Sequences similar to clade B and clade C were not found among cultured bacteria; these sequences may be from uncultured or unsequenced marine bacterial lineages, or they may represent sequence diversity within the known dmdA-containing taxa. Clade D sequences clustered with the dmdA orthologs from P. ubique HTCC1062 and HTCC1002. Two sequence assemblies from the Sargasso Sea that contained clade D homologs showed similar gene organization and highest gene similarities to the P. ubique genomes (Fig. 1). Based on the number of clade D homologs relative to SAR11-like 16S rDNA sequences (13), only 40% of SAR11 cells demethylate DMSP; these may belong to an ecologically distinct subgroup within the taxon (17).

Fig. 2.

Minimum evolution phylogenetic tree of amino acid sequences of glycine cleavage T-protein (GcvT) family proteins, including DmdA and related aminomethyltransferases (AMT). Sequences from cultured bacteria are labeled with organism name and gene designation. Selected Sargasso Sea metagenomic library sequences are identified by sequence identification and, if applicable, a contig designation. Proteins with confirmed DMSP demethylase activity are marked with a star. Percentage of 100 bootstrap samples supporting each node are shown if >50.

Table 2.

Abundance of dmdA homologs in marine bacterioplankton metagenomic surveys. Sargasso Sea data are from surface seawater samples (Stations 1 to 7) using the unassembled shotgun library (11). Station Aloha data are grouped into photic zone (10, 70, and 130 m) and deep water (500, 770, and 4000 m) samples according to DeLong et al. (19). Sapelo Island data are from surface seawater samples (0.5 m). recA homologs were determined by BLAST analysis using the E. coli recA sequence as the query. The percentage of cells with dmdA was calculated as dmdA × 100/recA. recA is an essential single-copy gene. Mbp, mega–base pairs.

Library size (Mbp)dmdA homologsrecA homologs% of cells with dmdA
Sargasso Sea (oceanic) 1626 29 18 83 247 377 1029 37
Station Aloha (oceanic) photic zone 24.8 1 0 0 1 2 5 40
Station Aloha (oceanic) deep water 31.1 0 0 0 0 0 17 0
Sapelo Island (coastal) 15.2 9 0 0 1 10 26 38

Genes adjacent to dmdA homologs were consistent within a clade but differed across clades (Fig. 1). Previous studies have shown that MMPA can be metabolized to MeSH or MPA in seawater (fig. S1) (9, 18). Thus, whereas all DMSP demethylating taxa must have dmdA in common, a different, taxon-specific suite of genes may encode for the subsequent metabolism of MMPA (Fig. 1).

Although sequence coverage is small compared with that of the Sargasso Sea data set (Table 2), other marine metagenomic databases contain evidence of DMSP demethylase genes. Two dmdA homologs were found in photic zone samples from the Pacific Station Aloha database (19); as expected, none were in the deep water samples from this site where DMSP flux is negligible. Ten dmdA homologs were found in a metagenome from southeastern U.S. coastal water. Similar to the Sargasso Sea metagenome, the abundance of dmdA homologs in both these samples indicated that about a third of bacteria in surface ocean waters may participate in DMSP demethylation (Table 2).

The marine metagenomic surveys indicated that the majority of environmental dmdA homologs belonged to clades for which DMSP demethylase activity has not been experimentally verified (Table 2). To address this issue, the P. ubique HTCC1062 dmdA (gene SAR11_0246) was synthesized and introduced in trans into Escherichia coli. Cell-free extracts of the recombinant E. coli formed MMPA from DMSP (Table 1), confirming demethylation by a protein in the largest environmentally occurring clade (clade D).

DMSP synthesis is estimated to account for ∼1 to 10% of global marine primary production (20), consistent with previous evidence that a large fraction of active marine bacteria assimilate sulfur from DMSP (21, 22) and the wealth of DMSP demethylation genes we found in surface water bacterioplankton communities. The evidence that oceanic dmdA homologs are most similar to those from cultured SAR11 bacteria (50 to 65%, Table 2), whereas coastal homologs are most similar to those from cultured Roseobacters (90%), is consistent with known differences in the ecology and distribution of these two abundant bacterioplankton groups (13, 14). This evidence further suggests that SAR11 bacteria may dominate demethylation in the open ocean, where DMSP concentrations are low (10 to 15 nM) and relatively constant, whereas Roseobacters may dominate in phytoplankton blooms and coastal regions, where DMSP concentrations are high (up to 100 nM) and more variable. Knowledge of the kinetic and ecological diversity of bacterial DMSP demethylases represented by these major marine taxa is critical to understanding both the routing of reduced carbon and sulfur into the microbial food web and the bacterial controls on ocean-atmosphere sulfur flux with consequences to global climate regulation.

Supporting Online Material

Materials and Methods

Figs. S1 and S2

Table S1


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article