Dimethyl sulfoxide (DMSO) concentrations in deep oceanic water are higher than 1.5 nM at depths up to 1,500 m in the equatorial Pacific Ocean and never drop below 1.3 nM at depths up to 4,000 m in the Arabian Sea (Hatton et al., 1996, 1998, 1999). It is thought to be an environmentally significant compound due to the potential role it plays in the biogeochemical cycle of the climatically active trace gas dimethyl sulfide (DMS; Hatton et al., 2005). DMSO can be produced either through the transformation of DMS by both photo-oxidation and bio-oxidation routes or by direct production from marine phytoplankton (Hatton et al., 1996; Simó, 1998; Moran et al., 2012). In addition to its roles in protecting cells against photo-generated oxidants and cryogenic damage, DMSO can also be used as an alternative electron acceptor for energy conservation through microbial dissimilatory reduction

(Sunda et al., 2002; Asher et al., 2011). Although DMSO acts as the dominant organic sulfur compound in deep oceanic water, the mechanism underlying DMSO bio-reduction by bathypelagic microorganisms is unknown.

Biochemical and genetic analyses of anaerobic DMSO respiration have been performed, particularly in Escherichia coli (McCrindle et al., 2005). In E. coli, two sets of operons are involved in DMSO respiration: the dmsABC operon and the ynfEFGHI operon, which is a paralog of the former and is likely to be phenotypically silent (Lubitz and Weiner, 2003). The dmsABC operon encodes the three functional proteins DmsA (the molybdopterin cofactor-containing subunit of the DMSO reductase), DmsB (the ion-sulfur subunit), and DmsC (the NapC-like integral membrane anchor). These three subunits constitute a functional DMSO reductase that is anchored to the periplasmic side of the inner membrane by DmsC. The electron released by menaquinol (MQH₂) oxidation by DmsC is transferred via a series of [4Fe-4S] clusters in DmsB to the catalytic subunit DmsA, which reduces DMSO to DMS (Stanley et al., 2002).

Shewanella is a genus of facultative anaerobic, Gram-negative microorganisms that are widely distributed in marine and freshwater environments. The hallmark of Shewanella is their ability to utilize a broad range of terminal electron acceptors, which makes them outstanding candidates for potential applications in the bioremediation of pollutants (Nealson and Scott, 2006; Hau and Gralnick, 2007). In Shewanella species, the DMSO reduction pathway has been characterized only in Shewanella oneidensis MR-1, which was isolated from the sediment of Oneida Lake in New York (Venkateswaran et al., 1999). Two dms operons were found in the MR-1 genome, although only one of them mediated DMSO reduction under the tested conditions (Gralnick et al., 2006; Fredrickson et al., 2008).

Shewanella piezotolerans WP3 was isolated from a west Pacific deep sea sediment at a depth of 1,914 m (Wang et al., 2004; Xiao et al., 2007). Our previous study demonstrated that WP3 was able to utilize DMSO as a terminal electron acceptor for anaerobic growth (Xiao et al., 2007). However, the precise mechanism of anaerobic DMSO respiration by WP3 is still unknown. In this study, we showed that WP3 contained two dms gene clusters (type I and type II), both of which were functional; type I was essential for the ability of WP3 to thrive under in situ conditions (4°C/20 MPa) and type II was more important under other extreme conditions (i.e., 20°C/20 MPa or 4°C/0.1 MPa). The possession of two sets of DMSO respiratory systems is suggested to be an adaptive strategy for WP3 to cope with extreme deep sea environments.

DMSO and DMS are interchangeable substrates in marine environments through the bio-reduction and bio-oxidation pathways (Bürgmann et al., 2007; Muyzer and Stams, 2008; Rellinger et al., 2009; Spiese et al., 2009). As a volatile anti-greenhouse gas, DMS plays an important climatic role in cloud nucleation and hence may influence the global temperature (Charlson et al., 1987; Andreae, 1990; Curson et al., 2011). Previous study demonstrated that DMSO as a substantial sink or source for DMS (Hatton et al., 1996). The formation of DMSO would therefore lead to the removal of DMS from sea water, effectively limiting the quantity of DMS available for releasing into the atmosphere (Hatton et al., 2005). Although DMSO has been detected in abyssal ocean water (Hatton et al., 1999), few data are available on the turnover mechanisms of DMSO in deep sea environments. In this study, our results demonstrate a previously unrecognized role for biotic DMSO reduction as an important pathway of DMS production in the deep sea environment.

In the abyssal ocean, metabolic energy is harnessed from the coupling of redox reactions (Orcutt et al., 2011). Microorganisms exploit the available chemical energy by developing strategies to overcome the activation energy of reaction. The metabolic activities of microorganisms in such environment depend on the availability and speciation of electron donors and acceptors (Froelich et al., 1979). As for DMSO, the E₀′ value for its reduction (+160 mV) lies between those for fumarate/succinate (+33 mV) and nitrate/nitrite (+433 mV; Thauer et al., 1977; Wood, 1981), and thus the coupling of DMSO reduction to ATP synthesis presents no energetic difficulty. Accordingly, some culturable bacteria isolated from marine environments have also shown the ability to grow by anaerobic DMSO respiration (Jonkers et al., 1996; Süß et al., 2008). Although sulfate is the most abundant oxidized sulfur species in the dark ocean (Orcutt et al., 2011), the reduction of sulfate yields less energy than that of DMSO, and DMSO could therefore be consumed preferentially.

In Shewanella species, the DMSO reduction pathway was first characterized in S. oneidensis MR-1, which was a strain isolated from the sediment of Oneida Lake in New York (Venkateswaran et al., 1999). Two dms operons were found in the MR-1 genome; however, only one dms operon mediated DMSO reduction under the tested conditions (Gralnick et al., 2006). In contrast to the results in MR-1, both dms gene operons in WP3 were functional (Figure 3). Moreover, our results demonstrated that the presence of both operons was essential for the maximum growth of WP3 using DMSO as the sole electron acceptor (Figure 5). Type I matches the archetypal dmsEFABGH organization, which enables the DMSO reductase to localize to the outer membrane of Shewanella and allows the cells to respire DMSO more efficiently (Gralnick et al., 2006). Additionally, type I plays a dominant role in the DMSO-dependent growth of WP3 under in situ deep sea conditions (4°C/20 MPa). Type II is more similar to E. coli (dmsABC organization; Stanley et al., 2002) and may locate in the periplasmic space. The physiological analysis indicated that type II was more important for DMSO-dependent growth than type I under some stressful conditions (i.e., 20°C/20 MPa and 4°C/0.1 MPa).

To evaluate the impact of the two DMSO respiration systems on growth, we calculated the DMSO consumption of the WP3 mutants by standardizing the growth yield to the same cell density (1 OD unit ≈ 1.5 × 10⁸ cells; Supplementary Figure S3). Our data showed that WT (type I and type II together) and ΔdmsA1 (type II alone) required less DMSO than ΔdmsA2 (type I alone) to reach the same growth yield when incubated at 20°C/0.1 MPa, 4°C/0.1 MPa, or 20°C/20 MPa, indicating that type II was a more efficient DMSO respiration system at atmospheric pressure or 20°C. However, when the WP3 mutants were incubated at 4°C/20 MPa, ΔdmsA2 (type I alone) inversely required less DMSO than ΔdmsA1 (type II alone), indicating that type I was more efficient than type II under the 4°C/20 MPa in situ condition. Collectively, our results suggest that the possession of multiple systems with similar functions might be an adaptive strategy for bacteria to cope with extreme deep sea environments. Similar adaptive mechanisms can also be found in other deep sea-derived Shewanella strains. For example, Shewanella putrefaciens W3-18-1 (Sediment; under 630 m of oxic water) has been identified to contain two functional nap gene clusters (nap-α and nap-β), and the nitrate respiration in W3-18-1 starts earlier than that in MR-1 (with nap-β only) under microoxic conditions (Qiu et al., 2013). In another Shewanella strain, Shewanella violacea DSS12 (Sediment; 5,110 m), it possesses more terminal oxidases with different affinities for oxygen and less terminal reductases, implying that DSS12 has undergone respiratory adaptation to aerobiosis in the upper layer of deep sea sediments (Aono et al., 2010).

The physiological results concerning the possession of two DMSO respiration systems in WP3 are quite different from those of the two NAP systems in WP3. Although two nitrate respiration systems (nap-α and nap-β) were identified in the WP3 genome, further studies demonstrated that WP3 possessing nap-α alone was more competitive than the strain possessing both nap-α and nap-β during nitrate-dependent growth. Thus containing a single nap-α system should be the evolutionary direction for Shewanella to thrive at low temperature (Chen et al., 2011). By combining these results, we conclude that WP3 can utilize different strategies to evolve its different anaerobic respiration systems in response to deep sea environments.

The effects of pressure on the respiratory systems has been well studied in S. violacea DSS12. Although very few terminal reductases for anaerobic respiration being identified in DSS12 genome, it processes at least five sets of putative terminal oxidases for aerobic respiration (Aono et al., 2010), in which the expression of cytochrome bd (quinol oxidase) was enhanced under high pressure (Tamegai et al., 2005). In addition, the d-type terminal oxidase of DSS12 was identified to be pressure-resistant and functional under high pressure conditions (Chikuma et al., 2007). Further research demonstrated that the piezotolerance of the Shewanella terminal oxidase activity perhaps depends on the intrinsic properties of the enzymes themselves but not their membrane lipid composition (Tamegai et al., 2011). In contrast, the piezophilic property of terminal oxidases in Photobacterium profundum SS9 (amphipod; 2,500 m) seems to be different. For one thing, the cytochrome content of P. profundum SS9 was not affected by altered pressure during growth; for another, genes encoding cytochromes showed no significant change in expression under different pressures during growth of P. profundum SS9. However, the piezotolerance of P. profundum SS9 terminal oxidase activity could also be observed using cells grown under higher pressures (Tamegai et al., 2012). Collectively, piezophiles from distinct genus may choose different strategies in adaptating to high pressure.

More recently, the respective effects of temperature and pressure on Fe(III) reduction rates (FeRRs) and viability have been investigated in deep sea bacterium Shewanella profunda LT13a (sediment; 4,500 m; Picard et al., 2015). Results showed that FeRR increased linearly with temperature between 4 and 37°C, while the highest FeRR was observed between 10 and 40 MPa and then slightly decreased with increasing pressure (40–110 MPa), indicating that the respiratory chain was not immediately affected by pressure (Picard et al., 2015). Under high pressure conditions, the significantly increased energy demand for cell maintenance of S. profunda LT13a may account for its decreased viability (Picard et al., 2015). Here we observed that DMSO-dependent growth can occur when WP3 is incubated in 4°C/20 MPa conditions (Figure 5E), suggesting that the energy demand of WP3 could still be fulfilled under in situ conditions with DMSO as the sole electron acceptor.

Because the transcription profiles of these two dms gene clusters vary significantly in response to pressure or temperature changes (Figure 4), it is important to elucidate the mechanism by which dms gene clusters are regulated. Previously, a mutant lacking the fur gene (encoding the ferric uptake regulator) was found to be severely deficient in DMSO respiration in WP3 (Yang et al., 2013), indicating that Fur was the dms gene cluster regulator. Moreover, cAMP receptor protein (CRP) was also found to directly regulate the expression of the dms gene cluster in S. oneidensis MR-1 (Zhou et al., 2013). We predicted the transcriptional regulation motifs of both dms gene clusters in WP3 and found similar CRP binding motifs (data not shown). Taken together, the presence of the CRP and Fur binding motifs might be an alternative regulatory strategy for WP3 to cope with pressure or temperature changes in a subtle regulatory mechanism. Investigation into this mechanism is currently underway and will be described as part of a separate report.

We have shown that two dms operons found in S. piezotolerans WP3 are functional in DMSO respiration. Both dms gene clusters are essential for the maximum growth of WP3 under low temperature and high pressure conditions; the type I system is essential for the ability of WP3 to thrive under in situ conditions (4°C/20 MPa), whereas the type II system is more important under other stressful conditions (i.e., 4°C/0.1 MPa and 20°C/20 MPa). Based on these data, we propose that multiple copies of dms gene clusters may confer a competitive advantage to the ability of WP3 to thrive under a broader range of environmental conditions. Thus, deep sea microorganism-mediated DMSO reduction may contribute to the global DMS cycle, which has been neglected in previous studies.

LX, HJ, and XX designed the study, analyzed the data. LX and YZ participated in the experiments. LX and HJ wrote the manuscript and XX critically reviewed the manuscript. All authors read and approved the final manuscript.

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