Deep-Sea Bacterium Shewanella piezotolerans WP3 Has Two Dimethyl Sulfoxide Reductases in Distinct Subcellular Locations

ABSTRACT

Dimethyl sulfoxide (DMSO) acts as a substantial sink for dimethyl sulfide (DMS) in deep waters and is therefore considered a potential electron acceptor supporting abyssal ecosystems. Shewanella piezotolerans WP3 was isolated from west Pacific deep-sea sediments, and two functional DMSO respiratory subsystems are essential for maximum growth of WP3 under in situ conditions (4°C/20 MPa). However, the relationship between these two subsystems and the electron transport pathway underlying DMSO reduction by WP3 remain unknown. In this study, both DMSO reductases (type I and type VI) in WP3 were found to be functionally independent despite their close evolutionary relationship. Moreover, immunogold labeling of DMSO reductase subunits revealed that the type I DMSO reductase was localized on the outer leaflet of the outer membrane, whereas the type VI DMSO reductase was located within the periplasmic space. CymA, a cytoplasmic membrane-bound tetraheme c-type cytochrome, served as a preferential electron transport protein for the type I and type VI DMSO reductases, in which type VI accepted electrons from CymA in a DmsE- and DmsF-independent manner. Based on these results, we proposed a core electron transport model of DMSO reduction in the deep-sea bacterium S. piezotolerans WP3. These results collectively suggest that the possession of two sets of DMSO reductases with distinct subcellular localizations may be an adaptive strategy for WP3 to achieve maximum DMSO utilization in deep-sea environments.

IMPORTANCE As the dominant methylated sulfur compound in deep oceanic water, dimethyl sulfoxide (DMSO) has been suggested to play an important role in the marine biogeochemical cycle of the volatile anti-greenhouse gas dimethyl sulfide (DMS). Two sets of DMSO respiratory systems in the deep-sea bacterium Shewanella piezotolerans WP3 have previously been identified to mediate DMSO reduction under in situ conditions (4°C/20 MPa). Here, we report that the two DMSO reductases (type I and type VI) in WP3 have distinct subcellular localizations, in which type I DMSO reductase is localized to the exterior surface of the outer membrane and type VI DMSO reductase resides in the periplasmic space. A core electron transport model of DMSO reduction in WP3 was constructed based on genetic and physiological data. These results will contribute to a comprehensive understanding of the adaptation mechanisms of anaerobic respiratory systems in benthic microorganisms.

INTRODUCTION

Dimethyl sulfoxide (DMSO) is an abundant but poorly understood methylated sulfur compound in the marine environment (1). It is thought to be an environmentally significant compound due to the potential role that it plays in the biogeochemical cycle of the climatically active gas dimethyl sulfide (DMS) (2). DMSO can be produced through the transformation of DMS by both photooxidation and biooxidation routes or by direct production from marine phytoplankton (3–5). The formation of DMSO would therefore lead to the removal of DMS from seawater, effectively controlling DMS flux into the atmosphere (2, 5). In addition to its roles in protecting cells against photogenerated oxidants and cryogenic damage, DMSO can also be used as an alternative electron acceptor for energy conservation through microbial dissimilatory reduction (6, 7). Previous studies demonstrated that DMSO acts as a substantial sink for DMS in deep oceanic waters (8, 9). Although some culturable bacteria isolated from deep-sea environments have also shown the capacity to grow anaerobically with DMSO as the sole electron acceptor (10, 11), few data are available on the electron transport pathway underlying DMSO-induced reduction by bathypelagic microorganisms.

Electron transport pathways of anaerobic DMSO respiration have been established, particularly in Escherichia coli (12). In E. coli, the dmsABC gene cluster encodes the following three functional proteins: DmsA, a molybdopterin (MPT) cofactor-containing catalytic subunit of DMSO reductase; DmsB, an electron transfer subunit; and DmsC, a membrane anchor subunit. These three subunits constitute a functional DMSO reductase, which is anchored to the periplasmic side of the inner membrane by DmsC (13). The electron released by menaquinol oxidation by DmsC is transferred via a series of [4Fe-4S] clusters in DmsB to the active site of DmsA, where DMSO is reduced to DMS (14).

Shewanella is a genus of facultatively anaerobic, Gram-negative gammaproteobacteria widely distributed in aquatic and sedimentary systems (15). The hallmark of Shewanella is its capacity to respire a diverse array of electron acceptors, making it a potential candidate for the bioremediation of pollutants (16). In Shewanella species, the DMSO respiratory pathway has been established only in Shewanella oneidensis MR-1, a strain that was isolated from the sediments of Oneida Lake in New York (17). In contrast to E. coli, the DMSO reductase complex in MR-1 resides on the outer leaflet of the outer membrane, with DmsE as a periplasmic electron shuttle delivering electrons from the inner membrane-bound quinol dehydrogenase, CymA, to the outer membrane DMSO reductase (18, 19).

Shewanella piezotolerans WP3 was isolated from west Pacific deep-sea sediments at a water depth of 1,914 m (20). Our previous study demonstrated that two functional DMSO respiratory subsystems were responsible for the maximum growth of WP3 under in situ conditions (4°C/20 MPa) (21). However, the electron transport pathway underlying DMSO-induced reduction by WP3 remains unknown. Here, we show that the two DMSO reductases (type I and type VI) in WP3 are functionally independent despite their close evolutionary relationship. Immunogold labeling of DMSO reductase subunits revealed that the type I DMSO reductase was localized on the outer leaflet of the outer membrane, whereas the type VI DMSO reductase was located within the periplasmic space. Moreover, CymA served as a preferential electron transport protein for the type I and type VI DMSO reductases, in which type VI accepted electrons from CymA in a DmsE- and DmsF-independent manner. The possession of two sets of DMSO respiratory subsystems with distinct subcellular localizations is suggested to be an adaptive strategy for WP3 to achieve maximum DMSO utilization in deep-sea environments.

RESULTS

Type VI and type I DMSO reductases are closely evolutionarily related.

Bioinformatic analyses identified 24 dms gene clusters in 13 of 24 fully sequenced Shewanella strains. The copy number of the dms gene cluster diverged significantly among these species (Table 1). Based on the classification principle of the DMSO respiratory subsystem (16), the 24 dms gene clusters from 13 Shewanella strains were divided into 6 subsystems (Fig. 1A). Type I existed in all 13 fully sequenced Shewanella strains and matched the archetypal dmsEFABGH organization. Compared with the type I subsystem, types II and III each contained an additional gene predicted to encode a lipoprotein. No DmsH-encoding gene was found in the type IV subsystem, which instead contained a gene predicted to encode an endonuclease III-related protein. Type V was more characteristic of Shimwellia, which contains a dmsBCAG gene cluster (see Table S1 in the supplemental material). The type VI subsystem consisted of four genes (dmsABGH) and was found only in S. piezotolerans WP3. To investigate the evolutionary relationships of DMSO reductases among Shewanella strains, a phylogenetic tree containing 44 DmsA homologs (Table S1) from Shewanella and other gammaproteobacterial strains was constructed (Fig. 1B). The type VI DmsA homologs tended to cluster together with those of type I rather than those of other subsystems (II through V), suggesting close evolutionary relatedness between type VI and type I. Except for type VI, the subsystems (I through V) were located on distinct phylogenetic branches, and DmsA homologs from the same subsystem tended to group together.

TABLE 1.

The dms gene clusters are present in 13 of the 24 fully sequenced Shewanella strains

Strain	Geographical origin	Isolation site characteristics	Reference	No. of dms operon copies
S. baltica OS195	Baltic Sea (57°N, 20°E)	Seawater; anoxic basin; 140 m	54	1
S. baltica OS678	Baltic Sea (57°N, 20°E)	Seawater; low-oxygen zone; 110 m	55	1
Shewanella sp. MR-4	Black Sea	Seawater; oxic zone; 16°C; 5 m	56	1
Shewanella sp. MR-7	Black Sea	Seawater; anoxic zone; high NO3; 60 m	56	1
S. woodyi ATCC 51908	Alboran Sea (35°N, 2°W)	Detritus; 370 m	57	1
S. putrefaciens CN-32	Albuquerque, NM, USA	Subsurface; shale sandstone; 250 m	58	1
S. frigidimarina NCIMB 400	Coast of Aberdeen, United Kingdom	Seawater; North Sea	59	2
S. halifaxensis HAW-EB4	Emerald Basin, offshore Halifax Harbor, Canada	Sediment; munitions dumping area; 215 m	60	2
S. oneidensis MR-1	Oneida Lake, NY, USA	Anaerobic sediment; Mn(IV) reduction	61	2
S. pealeana ATCC 700345	Woods Hole Harbor, MA, USA	Squid nidamental gland	62	2
S. piezotolerans WP3	West Pacific Ocean (142°E, 8°N)	Sediment; 1,914 m	11	2
S. putrefaciens 200	Alberta, Canada	Crude-oil pipeline	63	2
S. sediminis HAW-EB3	Halifax Harbor, Nova Scotia, Canada	Sediment; 50 nautical miles from shore; 215 m	64	6

FIG 1.

Classification and phylogenetic analysis of DMSO respiratory subsystems in Shewanella species. (A) Classification of DMSO respiratory subsystems in Shewanella species according to a previous report (16). Name abbreviations: MR1, S. oneidensis MR-1; S200, Shewanella putrefaciens 200; CN32, S. putrefaciens CN-32; MR4, Shewanella sp. MR-4; MR7, Shewanella sp. MR-7; EB3, Shewanella sediminis HAW-EB3; EB4, Shewanella halifaxensis HAW-EB4; WP3, S. piezotolerans WP3; OS195, Shewanella baltica OS195; OS678, S. baltica OS678; NCIMB400, Shewanella frigidimarina NCIMB 400; ATCC 51908, Shewanella woodyi ATCC 51908; ATCC 700345, Shewanella pealeana ATCC 700345. (B) Phylogenetic tree of DmsA protein sequences from Shewanella and other gammaproteobacteria. We used trimethylamine N-oxide (TMAO) reductase (TorA) of Vibrio cholerae 2012EL-2176 as the outgroup. The letter N in dmsEFABGN represents a gene encoding an endonuclease III-related protein. The letter L in dmsEFABLGH represents a gene encoding a lipoprotein.

The two DMSO reductases are functionally independent.

To investigate whether the subunits from these two DMSO reductases were interchangeable, we constructed two unmarked double in-frame ΔdmsA1 ΔdmsB6 and ΔdmsA6 ΔdmsB1 deletion mutants. Physiological assays demonstrated that the two double mutants lost the ability to utilize DMSO for anaerobic growth under different conditions (Fig. 2). Moreover, transcriptional analyses revealed that deletion of the individual gene did not eliminate the expression of other genes within the same gene cluster (see Fig. S1A in the supplemental material). To confirm that the loss of DMSO-dependent growth of the ΔdmsA1 ΔdmsB6 and ΔdmsA6 ΔdmsB1 mutants can be rescued by introduction of dmsA1 and dmsA6, respectively, two complemented strains (the ΔdmsA1 ΔdmsB6-dmsA1-C and ΔdmsA6 ΔdmsB1-dmsA6-C strains [where “-C” refers to complementation]) were generated. As expected, the introduction of either dmsA1 into the ΔdmsA1 ΔdmsB6 mutant or dmsA6 into the ΔdmsA6 ΔdmsB1 mutant partially restored the ability of these double mutants to utilize DMSO for anaerobic growth (Fig. S1B and C). These results suggested that the deficiencies of DMSO-dependent growth in ΔdmsA1 ΔdmsB6 and ΔdmsA6 ΔdmsB1 mutants were attributable to the inability to form functional DMSO reductases rather than to the silencing of the expression of both dms gene clusters. In other words, functional compensation did not occur between DmsA1 and DmsA6 or between DmsB1 and DmsB6.

FIG 2.

Growth curves of WP3 mutants at different temperatures and pressures with DMSO as the sole electron acceptor. (A) 20°C and 0.1 MPa; (B) 4°C and 0.1 MPa; (C) 20°C and 20 MPa; (D) 4°C and 20 MPa. The data shown represent the results of two independent experiments, and the error bars represent standard deviations of averages of triplicate cultures.

Type I and type VI DMSO reductases have different subcellular localizations.

To further examine the subcellular localizations of both DMSO reductases in WP3, we visualized the location of the DmsB subunits (DmsB1 and DmsB6). DmsB was monitored instead of DmsA because its localization should be DmsA dependent as only DmsA has a twin-arginine translocation (Tat) signal sequence (14). Two complemented strains of the ΔdmsB1 ΔdmsB6 (ΔΔdmsB) mutant (the ΔΔdmsB-dmsB1-3HA and ΔΔdmsB-dmsB6-3HA strains) were constructed as described in Materials and Methods. Physiological assays demonstrated that both complemented strains regained the ability to utilize DMSO for anaerobic growth (see Fig. S2 in the supplemental material). To determine the subcellular localization of DmsB-HA, immunogold-labeled anti-hemagglutinin (HA) antibodies were used to probe intact cells and ultrathin sections of these two strains, and the samples were then observed by transmission electron microscopy.

Immunoelectron microscopic analysis of intact cells of these two strains grown under different conditions revealed the following: (i) under aerobic conditions, only rare gold particles were detected on the cell surface of the ΔΔdmsB-dmsB1-3HA complemented strain (see Fig. S3A in the supplemental material), whereas no gold particles were detected on the cell surface of the ΔΔdmsB-dmsB6-3HA complemented strain (Fig. S3C); and (ii) under anaerobic conditions, a number of gold particles were detected on the cell surface of the ΔΔdmsB-dmsB1-3HA complemented strain (Fig. S3B), whereas no gold particles were detected on the cell surface of the ΔΔdmsB-dmsB6-3HA complemented strain (Fig. S3D). These results implied that type I and type VI DMSO reductases had different subcellular localizations, in which the type I DMSO reductase tended to be localized extracellularly.

To further confirm their subcellular localization, ultrathin sections of the two strains grown under different conditions were analyzed by immunoelectron microscopy, which revealed the following: (i) under aerobic conditions, gold particles were scattered over the cytoplasmic matrix of both the ΔΔdmsB-dmsB1-3HA (Fig. 3A) and ΔΔdmsB-dmsB6-3HA (Fig. 3B) complemented strains, but no gold particles were located on the outer and inner membranes of these two strains; and (ii) under anaerobic conditions, gold particles were detected on the outer membrane of the ΔΔdmsB-dmsB1-3HA complemented strain (Fig. 3C and D) but within the periplasmic space of the ΔΔdmsB-dmsB6-3HA complemented strain (Fig. 3E and F), consistent with the absence of gold particles on the cell surface of the ΔΔdmsB-dmsB6-3HA complemented strain (Fig. S3D). Statistical analysis revealed that the number of gold particles per field on the outer membrane or in the periplasmic space of the ΔΔdmsB-dmsB1-3HA complemented strain was 1.13 ± 0.11 or 0.42 ± 0.09, respectively, whereas the number of gold particles per field on the outer membrane or in the periplasmic space of the ΔΔdmsB-dmsB6-3HA complemented strain was 0.15 ± 0.05 or 1.02 ± 0.07, respectively (see Fig. S4 in the supplemental material). In contrast, negative controls in which the primary antibody was omitted showed no immunogold labeling (see Fig. S5 in the supplemental material). Taken together, these results suggested that the two sets of DMSO reductases in WP3 had distinct subcellular localizations, in which the type VI DMSO reductase was located within the periplasmic space whereas the type I DMSO reductase was localized on the outer leaflet of the outer membrane.

FIG 3.

Immunoelectron microscopic analysis of DmsB-HA proteins in WP3 ultrathin sections. (A) Ultrathin section of the ΔΔdmsB-dmsB1-3HA strain grown aerobically. (B) Ultrathin section of the ΔΔdmsB-dmsB6-3HA strain grown aerobically. (C and D) Ultrathin sections of the ΔΔdmsB-dmsB1-3HA strain grown anaerobically. (E and F) Ultrathin sections of the ΔΔdmsB-dmsB6-3HA strain grown anaerobically. The 10-nm gold particles are indicated by red arrows, and inner membranes (IM) and outer membranes (OM) are indicated by black arrows. The white line in the lower left corner represents the image scale for transmission electron microscopy.

CymA is the preferential electron transport protein for DMSO reductases in WP3.

CymA, a cytoplasmic membrane-anchored tetraheme c-type cytochrome, plays a central role in anaerobic respiration by transferring electrons from the menaquinol pool to a variety of terminal reductases (18). To investigate whether the two DMSO respiratory systems of WP3 obtain electrons from CymA, the ΔcymA mutant was developed. Growth assays demonstrated that wild-type WP3 and ΔdmsA1 and ΔdmsA6 mutant strains propagated with a high growth rate when grown in a DMSO medium at 20°C and reached stationary phase within 20 h, whereas strains lacking cymA were severely deficient in DMSO-dependent growth and exhibited a relatively long lag phase (Fig. 4). In contrast, the cymA gene-complemented (ΔcymA/cymA) strain exhibited a restored ability to immediately respire DMSO for anaerobic growth. These results collectively indicated that CymA serves as a preferential electron transport protein for both the type I and type VI DMSO respiratory subsystems in WP3.

FIG 4.

Growth and corresponding DMSO consumption curves of the WP3 mutants grown at 20°C/0.1 MPa in 2216E medium using DMSO as the sole electron acceptor. The data shown represent the results of two independent experiments, and the error bars represent standard deviations of averages of triplicate cultures. ΔcymA-cymA-C, the cymA gene-complemented (ΔcymA/cymA) strain.

DmsE passes electrons to DmsA1 for DMSO reduction.

DmsE is the paralogue of MtrA, a periplasmic decaheme c-type cytochrome involved in iron oxide reduction in S. oneidensis MR-1 (18, 22). A rough model for DMSO reduction in Shewanella has been proposed in which the periplasmic DmsE mediates electron transfer from CymA to the terminal DMSO reductase (19). To investigate the role of DmsE in DMSO respiration in WP3, three in-frame deletion mutants (the ΔdmsE, ΔdmsA1 ΔdmsE, and ΔdmsA6 ΔdmsE mutant strains) were constructed. Growth assays demonstrated that all of these mutants showed growth deficiencies in anaerobic DMSO respiration of various degrees (Fig. 5A). The cell density of the ΔdmsE mutant at stationary phase was slightly lower than that of the wild-type strain but higher than those of the other mutants, including the ΔdmsA1 and ΔdmsA6 mutant strains. Interestingly, there was no apparent difference in cell density between the ΔdmsA1 ΔdmsE and ΔdmsA1 mutants, whereas the cell density of the ΔdmsA6 ΔdmsE mutant was significantly lower than that of the ΔdmsA6 mutant. In addition, culture medium was collected at different time points, filtered, and analyzed to calculate DMSO consumption by these strains (Fig. 5B). The high-performance liquid chromatography (HPLC) results showed that the amount of DMSO utilized by the ΔdmsA1 ΔdmsE mutant strain was nearly equivalent to that utilized by the ΔdmsA1 mutant strain. Compared with the ΔdmsA6 mutant, the ΔdmsA6 ΔdmsE mutant exhibited a more severe deficiency in DMSO reduction. Collectively, these results suggested that the type VI DMSO reductase accepted electrons from CymA in a DmsE-independent manner. However, the type I DMSO reductase was strongly dependent on DmsE for electron transfer.

FIG 5.

Growth and corresponding DMSO consumption curves of the WP3 mutants grown at 20°C/0.1 MPa in 2216E medium using DMSO as the sole electron acceptor. The data shown represent the results of two independent experiments, and the error bars represent standard deviations of averages of triplicate cultures.

DmsF facilitates electron transfer between the type I DMSO reductase and DmsE.

DmsF, an integral outer membrane β-barrel protein, facilitates electron transfer by forming a pore-like structure through the outer membrane to mediate direct interaction between the extracellular DMSO reductase and DmsE (23–25). To investigate the role of DmsF in DMSO respiration in WP3, three in-frame deletion mutants (the ΔdmsF, ΔdmsB1 ΔdmsF, and ΔdmsB6 ΔdmsF mutant strains) were generated. Growth assays demonstrated that the cell density of the ΔdmsF mutant at stationary phase was slightly lower than that of the ΔdmsB6 mutant but equivalent to that of the ΔdmsB1 mutant (Fig. 5C). As expected, there was no apparent difference in cell density between the ΔdmsB1 and ΔdmsB1 ΔdmsF mutants, whereas the cell density of the ΔdmsB6 ΔdmsF mutant was significantly lower than that of the ΔdmsB6 mutant. In addition, the HPLC results showed that the amount of DMSO utilized by the ΔdmsB1 ΔdmsF mutant was nearly equivalent to that utilized by the ΔdmsB1 mutant (Fig. 5D). Compared with that of the ΔdmsB6 mutant, the ΔdmsB6 ΔdmsF mutant strain exhibited a more severe deficiency in DMSO reduction. To further support the notion that DmsF is responsible for targeting the type I DMSO reductase to the outer leaflet of the outer membrane, a complemented strain of the ΔdmsB1 ΔdmsF mutant (the ΔdmsB1 ΔdmsF-dmsB1-3HA strain) was also constructed. To determine the subcellular localization of DmsB1-HA, ultrathin sections of this strain grown under anaerobic conditions were analyzed by immunoelectron microscopy (data not shown). Statistical analysis revealed that the number of gold particles per field on the outer membrane or in the periplasmic space of the ΔdmsB1 ΔdmsF-dmsB1-3HA complemented strain was 0.29 ± 0.09 or 0.32 ± 0.08, respectively, indicating that the deletion of dmsF resulted in a decreased accumulation of type I DMSO reductase to the outer membrane (Fig. S4). Collectively, these results suggested that DmsF was dispensable for electron transfer between the type VI DMSO reductase and CymA. However, DmsF played an important role in facilitating electron transfer between the type I DMSO reductase and DmsE.

DISCUSSION

Based on the data presented above and the current understanding of anaerobic DMSO respiration, we propose a core electron transport model of DMSO respiration in S. piezotolerans WP3 (Fig. 6). The dehydrogenases (DHs) located at the head end of the electron transport chain catalyze the oxidation of electron donors (XH₂), contributing to proton gradient generation and the reduction of menaquinone (26). The reduced menaquinone transfers electrons to the inner membrane-bound quinol dehydrogenase CymA, which in turn donates electrons to both the type I and type VI DMSO reductases. The type I DMSO reductase resides on the outer leaflet of the outer membrane and is unable to receive electrons from CymA directly, and the periplasmic decaheme c-type cytochrome DmsE serves as an electron shuttle delivering electrons from CymA to the outer membrane DMSO reductase. The integral membrane β-barrel protein, DmsF, may facilitate electron transfer by forming a pore-like structure through the outer membrane to mediate direct interaction between the type I DMSO reductase and DmsE (19, 27, 28). The type VI DMSO reductase is located within the periplasmic space and is capable of obtaining electrons from CymA in a DmsE- and DmsF-independent manner. In contrast, the DMSO reductase in E. coli is anchored to the periplasmic side of the inner membrane by DmsC. The electron released by menaquinol oxidation by DmsC is transferred via the [4Fe-4S] clusters in DmsB to the active site of DmsA (14, 29). In S. oneidensis MR-1, the DMSO reductase resides on the outer leaflet of the outer membrane, with DmsE as a periplasmic electron shuttle delivering electrons from the inner membrane-bound quinol dehydrogenase CymA to the outer membrane DMSO reductase (19). By combining these results, we conclude that the DMSO respiratory pathway in the deep-sea bacterium S. piezotolerans WP3 is more complex than those previously reported in other strains (12, 19).

FIG 6.

The core electron transport model of two sets of DMSO respiratory systems in the deep-sea bacterium S. piezotolerans WP3. Inner membrane-bound dehydrogenases (DH) donate electrons through the lipophilic menaquinone pool to CymA where they are passed to the type I and type VI DMSO reductases. The type I DMSO reductase resides on the outer leaflet of the outer membrane and, thus, relies on DmsE as an electron shuttle delivering electrons from CymA to the outer membrane DMSO reductase. The type VI DMSO reductase is located within the periplasmic space and is capable of obtaining electrons from CymA in a DmsE-independent manner. The trilaminar cell envelope of Gram-negative bacteria is composed of the inner membrane (IM), outer membrane (OM), and the periplasm between the IM and OM. DH, dehydrogenase; XH₂, electron donors; MQ, menaquinone; MQH₂, menaquinol.

CymA serves as a preferential electron transport protein for the type I and type VI DMSO respiratory subsystems in WP3 (Fig. 4). In the ΔdmsA1 mutant, although DmsE can accept electrons from CymA, the electrons carried by DmsE cannot be further consumed by the type I DMSO reductase due to the deficiency of the type I DMSO respiratory subsystem. However, the type VI DMSO respiratory subsystem in the ΔdmsA1 mutant can still accept electrons from CymA and function in DMSO reduction, and thus this strain can grow anaerobically in DMSO medium. In the ΔdmsA6 mutant, although the type VI DMSO respiratory subsystem has lost its function, the type I DMSO respiratory subsystem can still accept electrons from CymA, and therefore the ΔdmsA6 mutant can also grow anaerobically in the same medium. Moreover, we observed that the ΔcymA mutant exhibited no apparent growth in DMSO medium until after approximately 20 h of incubation (Fig. 4), implying that the DMSO reductases in WP3 can also accept electrons from the quinol pool in an alternative electron transfer pathway independent of CymA. In S. oneidensis MR-1, it was reported that a quinol dehydrogenase complex (SirCD) can functionally replace the c-type cytochrome, CymA, in several respiratory pathways, including fumarate, DMSO, and ferric citrate (30). Genetic analyses revealed that the expression of sirCD genes in MR-1 ΔcymA suppressor mutants is mediated by an insertion sequence (30). Using a bioinformatics approach, genes (swp4653 and swp4652) encoding SirC and SirD homologs were also identified in the WP3 genome. Transcriptional analysis demonstrated that, compared with those of the WP3 wild type (WT), the expression profiles of sirC and sirD in the ΔcymA mutant were significantly induced under the same conditions (our unpublished data). Based on these results, we propose that the SirCD complex in WP3 may play a minor role in DMSO-dependent growth in the absence of CymA. Considering that CymA is a member of the NapC/NirT family (24, 25), other subgroups, such as NapC, NrfH, and NirT, may also exert similar functions.

To evaluate the impact of dmsE deletion on growth, we calculated the DMSO consumption of the WP3 mutants by standardizing the growth yield to the same cell density (see Fig. S6 in the supplemental material). Our data showed that the ΔdmsA1 ΔdmsE mutant required the same amount of DMSO as the WP3 WT to reach the same growth under the tested conditions (20°C/0.1 MPa), further supporting the conclusion that the type VI DMSO respiratory subsystem in WP3 is DmsE independent. However, DmsE in the ΔdmsA1 mutant may dissipate electrons of the quinol pool via CymA (31), resulting in a decrease in the amount of electrons available for the type VI DMSO reductase. Although the ΔdmsA6 ΔdmsE mutant required the same amount of DMSO as the ΔdmsA6 mutant to reach the same growth (Fig. S6), the ΔdmsA6 ΔdmsE mutant showed a more severe growth deficiency in DMSO medium (Fig. 5A and B), suggesting that either other periplasmic electron shuttles functionally replace DmsE in the context of DMSO reduction or the type I DMSO reductase has some activity in the periplasm (obtaining electrons from CymA). A previous study focusing on S. oneidensis MR-1 demonstrated that, in addition to DmsE, CctA (a tetraheme c-type cytochrome) was also able to transfer electrons from CymA to the DMSO reductase, and the ΔdmsE ΔcctA double mutant grew significantly slower than either the ΔdmsE mutant or ΔcctA mutant in anaerobic DMSO medium (18). Collectively, these results suggest that functional redundancy and plasticity may be common features of Shewanella in electron transport to DMSO.

DMSO is generally thought to be a soluble organic compound that is highly permeable to biological membranes (32). Therefore, the periplasmic localization of DMSO reductases in E. coli and Rhodobacter is not surprising (14, 33). The dms gene cluster (dmsABC) in E. coli is more characteristic of the type V subsystem, which encodes three functional proteins, DmsA, DmsB, and DmsC. As it lacks genes that encode the DmsEF module and CymA, the integral membrane protein DmsC is required to anchor the DMSO reductase complex (DmsAB) to the membrane and for quinol oxidation (34). In Rhodobacter, the DMSO reductase operon consists of three genes (dorCAD), which encode three functional proteins, DorC (the NapC-like integral membrane anchor), DorD (the DMSO reductase-specific chaperone), and DorA (the periplasmic DMSO reductase). DorA proteins differ from DmsA proteins in that they do not contain iron-sulfur clusters. The Dor-type DMSO reductases are soluble and only transiently interact with the membrane-bound DorC-type cytochromes that act as electron donors for these systems (34). In Shewanella species, DMSO respiration was first characterized in S. oneidensis MR-1, in which the DMSO reductase is localized on the outer leaflet of the outer membrane (19). Here, we observed that S. piezotolerans WP3 had two sets of DMSO reductases; the type I DMSO reductase was localized on the outer leaflet of the outer membrane, whereas the type VI DMSO reductase was located within the periplasmic space. Moreover, the DMSO-dependent growth yield conferred by the presence of both DMSO reductases was higher than the growth yield conferred by either of the two DMSO reductases alone (21).

Why would two distinct DMSO respiratory subsystems confer S. piezotolerans WP3 a selective advantage for DMSO-dependent growth? Insight into this question comes from considering the typical growth environment of Shewanella. S. piezotolerans WP3 was isolated from west Pacific sediments at a depth of 1,914 m, an environment with permanently low temperature (∼2°C to ∼4°C) and high pressure (20 MPa). Theoretically, low temperature and high pressure in deep-sea environments can decrease the fluidity of lipids and thus depress the functions of biological membranes (35, 36), resulting in decreased diffusion of DMSO into the cell (37, 38). To survive and proliferate under such conditions, WP3 may have evolved DMSO respiration subsystems in which the type I DMSO reductase allows WP3 to more efficiently utilize DMSO dispersed in the extracellular environment, whereas the type VI DMSO reductase enables WP3 to utilize DMSO penetrating the periplasmic space. Consistent with this idea, it has been reported that the expression of the type I dms gene cluster at 4°C was significantly induced with increasing pressure from atmospheric pressure to high-pressure (10 and 20 MPa) conditions, whereas the expression of the type VI dms gene cluster was greatly reduced under the same conditions (21). Taken together, the two sets of DMSO reductases with distinct subcellular localizations may represent an adaptive strategy for WP3 to achieve maximum DMSO utilization in deep-sea environments.

DmsA1 shares high sequence similarity with DmsA6 (54%), and DmsB1 shares high sequence similarity with DmsB6 (56%). However, the type I and type VI DMSO reductases in WP3 displayed distinct subcellular localizations (Fig. 3; see also Fig. S3 in the supplemental material). Previous studies have demonstrated that type II secretion systems (T2SSs) are required for the proper extracellular localization of both DMSO reductase and metal reductase in Shewanella species (19, 39, 40). Based on these data, we propose that T2SS may be involved in the secretion of type I DMSO reductase rather than type VI DMSO reductase. T2SSs are molecular machines that promote the specific transport of folded periplasmic proteins in Gram-negative bacteria, but the T2SS secretion signal remains a mystery (41, 42). In general, exoproteins tend to be rich in β-strands, and this is also true of the exoproteins that are secreted by the T2SS according to currently available structures (42, 43). Further investigations into these two DMSO reductases at the sequence or structural level may provide insights into the molecular mechanism of protein sorting by T2SS.

In conclusion, we have demonstrated that the two DMSO reductases (type I and type VI) in WP3 are functionally independent despite their close evolutionary relationship. Both DMSO reductases in WP3 had distinct subcellular localizations, in which the type I DMSO reductase was localized on the outer leaflet of the outer membrane and the type VI DMSO reductase was located within the periplasmic space. CymA served as a preferential electron transport protein for the type I and type VI DMSO reductases. Based on these data, we proposed a core electron transport model of DMSO respiration in the deep-sea bacterium S. piezotolerans WP3. Moreover, possession of two sets of DMSO reductases with distinct subcellular localizations is likely to be an adaptive strategy for WP3 to achieve maximum DMSO utilization in deep-sea environments.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The bacterial strains used in this study are listed in Table 2. E. coli strain WM3064 was routinely grown in Luria broth (LB) media at 37°C with the addition of 500 μM 2,6-diaminopimelic acid (DAP) (Sigma-Aldrich, St. Louis, Mo, USA). For aerobic growth, S. piezotolerans WP3 strains were cultured in 2216E broth (44, 45) with minor modification (5 g liter⁻¹ tryptone, 1 g liter⁻¹ yeast extract, 34 g liter⁻¹ NaCl) at 20°C on a rotary shaker at 220 rpm. If necessary, chloramphenicol was added to both media (30 μg ml⁻¹ for E. coli strains and 10 μg ml⁻¹ for WP3 strains). For anaerobic growth assays, WP3 strains were cultivated in 2216E broth supplemented with 20 mM lactate and 25 mM DMSO (21). The serum bottles containing 100 ml of fresh medium were made anaerobic by flushing with nitrogen gas through the butyl rubber stopper, which was further fixed with metal seals to strip the dissolved oxygen before autoclave sterilization. To examine high-pressure growth of WP3 strains at different temperatures, each culture was grown to stationary phase in 2216E broth at 1 atm (1 atm = 0.101 MPa) and 20°C on a rotary shaker. The late-log-phase cultures were diluted into the same medium to an optical density at 600 nm (OD₆₀₀) of 0.3. Aliquots of the diluted culture (1 ml) were injected into serum bottles containing 100 ml of anaerobic medium through the butyl rubber stopper. After brief shaking, 2.5-ml disposable syringes were used to distribute the culture in 2-ml aliquots. The syringes with needles were inserted into rubber stoppers in a vinyl anaerobic airlock chamber (Coy Laboratory Products Inc., Grass Lake, MI, USA). Then, the syringes were incubated at a hydrostatic pressure of 20 MPa at 4°C or 20°C in stainless steel vessels (Nantong Feiyu Science and Technology Exploitation Co., Ltd., China) that were pressurized using water and a hydraulic pump. These systems were equipped with quick-connect fittings for rapid decompression and recompression (23, 46).

TABLE 2.

Bacterial strains and plasmids used in this study

Strain or plasmid	Description	Reference or source
E. coli
WM3064	Donor strain for conjugation; ΔdapA	65
S. piezotolerans WP3
WT	Wild-type strain	Lab stock
ΔdmsA1 mutant	dmsA1 single mutant derived from WP3	Lab stock
ΔdmsB1 mutant	dmsB1 single mutant derived from WP3	Lab stock
ΔdmsA6 mutant	dmsA6 single mutant derived from WP3	Lab stock
ΔdmsB6 mutant	dmsB6 single mutant derived from WP3	Lab stock
ΔdmsE mutant	dmsE single mutant derived from WP3	This study
ΔdmsF mutant	dmsF single mutant derived from WP3	This study
ΔcymA mutant	cymA single mutant derived from WP3	Lab stock
ΔΔdmsA mutant	dmsA1 and dmsA6 double mutant derived from WP3	This study
ΔΔdmsB mutant	dmsB1 and dmsB6 double mutant derived from WP3	Lab stock
ΔdmsA1 ΔdmsB6 mutant	dmsA1 and dmsB6 double mutant derived from WP3	This study
ΔdmsA6 ΔdmsB1 mutant	dmsA6 and dmsB1 double mutant derived from WP3	This study
ΔdmsA1 ΔdmsE mutant	dmsA1 and dmsE double mutant derived from WP3	This study
ΔdmsA6 ΔdmsE mutant	dmsA6 and dmsE double mutant derived from WP3	This study
ΔdmsB1 ΔdmsF mutant	dmsB1 and dmsF double mutant derived from WP3	This study
ΔdmsB6 ΔdmsF mutant	dmsB6 and dmsF double mutant derived from WP3	This study
ΔcymA/cymA strain	Complemented strain of ΔcymA single mutant with cymA	This study
ΔcymA-pSW2 strain	ΔcymA containing the empty pSW2 vector as negative control	This study
ΔΔdmsB-dmsB1-3HA strain	Complemented strain of ΔΔdmsB double mutant with dmsB1-3HA	This study
ΔΔdmsB-dmsB6-3HA strain	Complemented strain of ΔΔdmsB double mutant with dmsB6-3HA	This study
ΔdmsB1 ΔdmsF-dmsB1-3HA strain	Complemented strain of ΔdmsB1 ΔdmsF double mutant with dmsB1-3HA	This study
ΔdmsA1 ΔdmsB6-dmsA1-C strain	Complemented strain of ΔdmsA1 ΔdmsB6 double mutant with wild-type dmsA1	This study
ΔdmsA6 ΔdmsB1-dmsA6-C strain	Complemented strain of ΔdmsA6 ΔdmsB1 double mutant with wild-type dmsA6	This study
ΔdmsA1 ΔdmsB6-pSW2 strain	ΔdmsA1 ΔdmsB6 double mutant containing the empty pSW2 vector as negative control	This study
ΔdmsA6 ΔdmsB1-pSW2 strain	ΔdmsA6 ΔdmsB1 double mutant containing the empty pSW2 vector as negative control	This study
Plasmid
pSW2	Chloramphenicol resistance, generated from filamentous bacteriophage SW1; used for complementation	48
pSW2-dmsA1	pSW2 containing dmsA1 and the promoter region of dmsA6	Lab stock
pSW2-dmsA6	pSW2 containing dmsA6 and its own promoter region	Lab stock
pSW2- dmsB1-3HA	pSW2 containing dmsB1-3HA and the promoter region of dmsB6	This study
pSW2- dmsB6-3HA	pSW2 containing dmsB6-3HA and its own promoter region	This study
pSW2-cymA	pSW2 containing cymA and its own promoter region	This study
pRE112	Chloramphenicol resistance, suicide plasmid with sacB1 gene as a negative selection marker; used for gene deletion	Lab stock
pRE112-ΔdmsA1	pRE112 containing the PCR fragment for deleting dmsA1	Lab stock
pRE112-ΔdmsB1	pRE112 containing the PCR fragment for deleting dmsB1	Lab stock
pRE112-ΔdmsA6	pRE112 containing the PCR fragment for deleting dmsA6	Lab stock
pRE112-ΔdmsB6	pRE112 containing the PCR fragment for deleting dmsB6	Lab stock
pRE112-ΔdmsE	pRE112 containing the PCR fragment for deleting dmsE	This study
pRE112-ΔdmsF	pRE112 containing the PCR fragment for deleting dmsF	This study
pRE112-ΔcymA	pRE112 containing the PCR fragment for deleting cymA	Lab stock

Deletion mutagenesis and complementation.

In-frame deletion mutagenesis of dmsE (swp3461) was performed as previously reported (44). The primers designed to amplify PCR products for mutagenesis are summarized in Table 3. To construct the dmsE in-frame deletion mutant, two fragments flanking dmsE were amplified by PCR. Fusion PCR products were generated using the amplified PCR fragments as the templates with the primers swp3461-UF and swp3461-DR. After digestion with the restriction enzymes SmaI and KpnI, the treated fusion PCR products were ligated into the SmaI and KpnI sites of the suicide plasmid pRE112 (47), resulting in the mutagenesis vector pRE112-ΔdmsE. This vector was first introduced into E. coli WM3064 and then conjugated into the WP3 WT (wild type). Positive exconjugants were spread onto marine agar 2216E plates supplemented with 10% (wt/vol) sucrose. The chloramphenicol-sensitive and sucrose-resistant colonies were screened by PCR for the dmsE deletion. The ΔdmsA1 ΔdmsE and ΔdmsA6 ΔdmsE double mutants were constructed by introducing pRE112-ΔdmsE into the ΔdmsA1 and ΔdmsA6 mutant strains, respectively. To construct the complemented strains of the ΔdmsB1 ΔdmsB6 (ΔΔdmsB) mutant for subcellular localization experiments, the intact dmsB1 and dmsB6 gene fragments were amplified from the WP3 wild-type genomic DNA. The resulting PCR products were fused to a 3×HA epitope tag from hemagglutinin of human influenza A virus (HA tag) on the C terminus of DmsB and then inserted into the shuttle vector pSW2, which was developed from the filamentous phage SW1 of WP3 (48). The ΔΔdmsB-dmsB1-3HA and ΔΔdmsB-dmsB6-3HA strain transformants were obtained by introducing the recombinant plasmids pSW2-dmsB1-3HA and pSW2-dmsB6-3HA, respectively, into the ΔΔdmsB mutant strain via conjugation.

TABLE 3.

Primers used in this study

Primers	Primer sequence (5′-3′)a	Description
Mutation use
swp0724-UF	TTGAGCTCCGAGGTAACGCAAAATGAAAAGAC	Deleting dmsA6
swp0724-UR	CCTTTAGTAGTAGTACGCGCCAGTGCCATAAA	Deleting dmsA6
swp0724-DF	GTACTACTACTAAAGGCCGCACCCACTCA	Deleting dmsA6
swp0724-DR	TTGCATGCGAAACTGCCACCAACAAACTCATA	Deleting dmsA6
swp0725-UF	TAGAGCTCATGTCGATACGGATTACCCACTG	Deleting dmsB6
swp0725-UR	TTTGCCATATAAACGCGACGCTGAATAATACC	Deleting dmsB6
swp0725-DF	GCGTTTATATGGCAAAGGCGATACTCACC	Deleting dmsB6
swp0725-DR	AAGCATGCGCTCTGTTAATTTTTCGTCGTTCC	Deleting dmsB6
swp3459-UF	TACCCGGGTGTTATCTTGCGGGTTGCTATTGT	Deleting dmsA1
swp3459-UR	TATACCGCTGCCGTTAAGGTTCGTATCACTCC	Deleting dmsA1
swp3459-DF	TAACGGCAGCGGTATAACATTGGCATCATCAG	Deleting dmsA1
swp3459-DR	ATTCTAGAAGTAATACCGGAGTCAGCCCTTCT	Deleting dmsA1
swp3458-UF	AAGGTACCCGGCCTCTGGATGACGAATAAG	Deleting dmsB1
swp3458-UR	GTGAATGGAGTACGGCGACGGCGATGGA	Deleting dmsB1
swp3458-DF	GCCGTACTCCATTCACCGCCACCGTATTCAT	Deleting dmsB1
swp3458-DR	AAGAGCTCTGCCGTTAAGGTTCGTATCACTCC	Deleting dmsB1
swp3460-UF	AACCCCCGGGTAATGAAACCTGTGCCGAGTG	Deleting dmsF
swp3460-UR	GAGTGTCACATCATCGGCATCAACAGCAC	Deleting dmsF
swp3460-DF	CGATGATGTGACACTCGGCATGGACTAC	Deleting dmsF
swp3460-DR	TAGAGGTACCGCCTGTAACTGTTGCTGCAC	Deleting dmsF
swp3461-UF	GTACCCGGGAGTATAGGCAAACTATATGCAA	Deleting dmsE
swp3461-UR	GGCACACGGAAGAGTATTTACCTTCGCTA	Deleting dmsE
swp3461-DF	TACTCTTCCGTGTGCCGCAATTATGTCA	Deleting dmsE
swp3461-DR	GCACGGTACCTACCAGAATAGCCATCAGTCG	Deleting dmsE
swp4806-UF	AAGGTACCCCTCCTAAATCTGACCG	Deleting cymA
swp4806-UR	CAGTTAAGGGTGTTGCTCACGTTTACCCAAAG	Deleting cymA
swp4806-DF	GCAACACCCTTAACTGGGTTGTGCGGTAAGTG	Deleting cymA
swp4806-DR	AATCTAGAGAGAGAGTTAGAGCCATGTCTCAT	Deleting cymA
ChlFor	TATCACTTATTCAGGCGTAGCA	Identifying chloramphenicol resistance gene in suicide plasmid pRE112
ChlRev	CCCAACACCGGACAAAAAGGA
Complementation use
cymA-F	CGGGATCCAGACCCACCTTTTGAGGAAA	Complement cymA
cymA-R	CCGCTCGAGTCCACTTCTGCTTTGCATTG
Subcellular location
(P+3458)F	CTTACTCGAGGTTGTTTCTAGTTGTATCTC	dmsB1 amplification
(P+3458)R	TATGGGTAAACCTCAGATGGGTTTAAAATGCT	dmsB1 amplification
3458HAF	CTGAGGTTTACCCATACGATGTTCCAGATTACG	3×HA amplification
3458HAR	CACAACGCGTGAGCTCGGTATTAAGCGTAA	3×HA amplification
P0724F	CTTACTCGAGTACCTTATAAAAACAGAGAGCC	dmsB6 promoter
P0724R	TTAGTCATTTTGCGTTACCTCGACATTTA	dmsB6 promoter
swp0725F	AACGCAAAATGACTAATTTAATTCAAACAACC	dmsB6 amplification
swp0725R	TATGGGTAAACCTCAGATGGGTTTAAAATGCT	dmsB6 amplification
0725HAF	TTATTAATAGCAGAGAAGTTTACCCATACGATGTTCCAGATTACG	3×HA amplification
0725HAR	CACAACGCGTGAGCTCGGTATTAAGCGTAA	3×HA amplification
qRT-PCR use
dmsA1 For	AGGCTGTAATTCTAGCTCTGATGATG	qRT-PCR
dmsA1 Rev	AAGCACGATGACCAGGTTACCT	qRT-PCR
dmsB1 For	GAAGACATTTGTATCGGTTGTGAAA	qRT-PCR
dmsB1 Rev	GCGTTCACGGTCAATTTGC	qRT-PCR
dmsG1 For	ACCGATAATCAAGCGTTAGT	qRT-PCR
dmsG1 Rev	AAGTCACTGCTGCTATGG	qRT-PCR
dmsH1 For	GCAATGGAGTAAGTCAATTCT	qRT-PCR
dmsH1 Rev	CCGCTATGGTGTTAGTGA	qRT-PCR
dmsA6 For	GATGACAAATGTATCGGCTGTAATATG	qRT-PCR
dmsA6 Rev	TTTTTACGCTCAGTATCCATTTGC	qRT-PCR
dmsB6 For	CACTGCAGTTGGTGGGATACC	qRT-PCR
dmsB6 Rev	CGTAGCCATGGCACATTATGA	qRT-PCR
dmsG6 For	AGTCAGCACATTGAGTCA	qRT-PCR
dmsG6 Rev	TCAGCAGTTCTCTTAGTAACA	qRT-PCR
dmsH6 For	TTGCCGAAGAGGTTGTAA	qRT-PCR
dmsH6 Rev	TCATTGAGGTTGCTTCTAATAG	qRT-PCR
swp2079 For	TTAAGGCAATGGAAGCTGCAT	Reference gene
swp2079 Rev	CGTCTTTACCCGTTAATGATACGA	swp2079 primer pairs

^aThe restriction sites included in the PCR primers are underlined.

RNA extraction and qRT-PCR.

WP3 total RNA was isolated using TRIzol reagent (Molecular Research Center, Inc., Cincinnati, OH) according to the manufacturer's recommendations with modifications (49). Briefly, 2 ml of mid-log-phase cultures (2 × 10⁷ cells/ml or OD₆₀₀ of ∼0.2) was harvested by centrifugation and homogenized in 1 ml of TRIzol, and 200 μl of chloroform was added. After brief vortexing, the samples were centrifuged at 12,000 × g for 15 min at 4°C. The upper aqueous phase was transferred to a tube containing an equal volume of isopropanol (500 μl). The mixtures were thoroughly vortexed and centrifuged at 12,000 × g for 20 min at 4°C. Supernatants were discarded, and the precipitated RNA pellets were then washed twice with 1 ml of 75% ethanol. After removing residual genomic DNA, all RNA samples were resuspended in 30 μl of RNase-free water, followed by a reverse transcription reaction with the RevertAid first strand cDNA synthesis kit (Thermo Scientific, Lithuania) to obtain cDNA. Quantitative real-time PCR (qRT-PCR) assays were performed in triplicate for each sample, and the mean values and standard deviations were calculated for the relative RNA expression levels (50).

DMSO concentration analysis.

Aliquots (0.5 ml) of the culture recovered at different time points were filtered immediately through 0.22-μm Millex-GP filters (Millipore, Carrigtwohill, Ireland) and stored at −70°C prior to use. DMSO concentration analysis was performed as previously reported (21). Briefly, after 10-fold dilution with distilled water, the samples were analyzed on an Agilent 1200 series (Agilent Technologies, Santa Clara, CA, USA) high-performance liquid chromatography (HPLC) system with a diode-array detector (DAD) at 210 nm. DMSO was separated using an Aminex HPX-87H sulfonic column (Bio-Rad, Hercules, CA, USA) at 50°C with 5 mmol H₂SO₄ as the mobile phase at a flow rate of 0.5 ml min⁻¹. Commercially available DMSO (Sigma-Aldrich, St. Louis, Mo, USA) was used to generate a calibration curve to calculate the concentration of DMSO. The calibration curve was linear (R² = 1.000) over a concentration range of 0.05 to 10 mM.

Phylogenetic analysis.

DmsA protein sequences were obtained from the NCBI GenPept database using the BLASTP suite. Swp3459 (DmsA1) and Swp0724 (DmsA6) sequences were used as queries. Sequences with identities of ≥42% (E value = 0) were chosen as significantly similar. The phylogenetic tree (Fig. 1) of 44 DmsA protein sequences from Shewanella and other gammaproteobacteria was constructed using the maximum likelihood method from FastTree version 2.1.3 (JTT model, CAT approximation) (51), and the trimethylamine N-oxide (TMAO) reductase (TorA) of Vibrio cholerae 2012EL-2176 was used as the outgroup. For each data set, bootstrap values were obtained with 1,000 replicates.

Preparation of ultrathin sections of bacterial cells.

WP3 strains grown under different conditions were harvested by centrifugation and then treated with 4% (vol/vol) glutaraldehyde in phosphate-buffered saline (PBS) overnight at 4°C. Fixed cells were collected, washed three times with 0.1 M phosphate buffer (PB, pH 7.4) for 15 min, and postfixed in prechilled 1% osmic acid at 4°C for 2 h. The cells were dehydrated in a series of concentrations of alcohol (50%, 70%, 90%) for 10 min at each concentration. Subsequently, the fixed cells were permeated and embedded in epoxy resin (52). The samples were sectioned to yield 70-nm ultrathin sections with an Ultracut UC-6 microtome (Leica, Heidelberg, Germany). These sections were transferred to carbon-coated copper grids and dried under ambient conditions.

Immunogold labeling and transmission electron microscopy.

For immunogold labeling of ultrathin sections, the sections on the grids were immersed in 0.01 M phosphate-buffered saline (PBS) solution containing 0.02 M glycine to block nonspecific labeling. These sections were then incubated with mouse anti-HA monoclonal antibody (clone HA-7; Sigma-Aldrich, St. Louis, MO) for 1 h (1:500 dilutions), washed with PBS three times for 5 min, incubated with gold-conjugated anti-mouse IgG (10-nm-diameter gold nanoparticles; Sigma-Aldrich, St. Louis, MO) for 1 h (1:50 dilutions), and washed six times with PBS (5 min each wash). To increase contrast, the ultrathin sections were stained with platinum blue (TI-blue; Nisshin EM Corporation, Tokyo, Japan) and lead citrate and then observed using a Tecnai G2 Spirit BioTwin (120 kV) transmission electron microscope (TEM) (FEI Company, Eindhoven, The Netherlands). A similar procedure was applied for immunogold labeling of whole bacteria. Whole bacteria were not stained with platinum blue or lead citrate solutions to enable observation of the gold particles on the whole surface of the bacteria by TEM (19). Negative controls were performed by omitting the incubation with the primary antibody. For statistical analysis, the numbers of gold particles along the cell membrane structures were counted for 50 fields for ultrathin sections prepared from each tested WP3 strain (53).