Shewanella oneidensis uses a wide range of terminal electron acceptors for respiration. In this study, we show that the chemotactic response of S. oneidensis to anaerobic electron acceptors requires functional electron transport systems. Deletion of the genes encoding dimethyl sulphoxide and trimethylamine N-oxide reductases, or inactivation of these molybdoenzymes as well as nitrate reductase by addition of tungstate, abolished electron acceptor taxis. Moreover, addition of nigericin prevented taxis towards trimethylamine N-oxide, dimethyl sulphoxide, nitrite, nitrate and fumarate, showing that this process depends on the ΔpH component of the proton motive force. These data, together with those concerning response to metals (Bencharit and Ward, 2005), support the idea that, in S. oneidensis, taxis towards electron acceptors is governed by an energy taxis mechanism. Surprisingly, energy taxis in S. oneidensis is not mediated by the PAS-containing chemoreceptors but rather by a chemoreceptor (SO2240) containing a Cache domain. Four other chemoreceptors also play a minor role in this process. These results indicate that energy taxis can be mediated by new types of chemoreceptors.
The Shewanella species are particularly well known for their respiratory versatility. Indeed, several strains were reported to use almost 20 different terminal electron acceptors for respiration (Nealson and Scott, 2005). In addition to oxygen, S. oneidensis can respire various organic and inorganic substrates, including fumarate, nitrate, nitrite, thiosulphate, trimethylamine N-oxide (TMAO), dimethyl sulphoxide (DMSO), elemental sulphur as well as various metal electron acceptors such as cobalt, iron or manganese (Hau and Gralnick, 2007). In order to reduce these diverse electron acceptors, S. oneidensis possesses different respiratory systems. The use of fumarate as an electron acceptor requires the FccA protein (SO0970), which was therefore considered as the sole physiological fumarate reductase (Maier et al., 2003). Anaerobic respiration of DMSO involves a cluster of genes (SO1427–SO1432) in which dmsA (SO1429) encodes a DMSO reductase. This enzyme is located in the outer membrane allowing extracellular respiration of DMSO (Gralnick et al., 2006). S. oneidensis conducts respiratory nitrate ammonification using the periplasmic nitrate reductase, NapA (SO0848) and the periplasmic nitrite reductase, NrfA (SO3980) (Cruz-Garcia et al., 2007). Induced in the presence of TMAO, the torECAD operon encodes the main TMAO respiration system (Gon et al., 2002; Bordi et al., 2003; Baraquet et al., 2006). The terminal reductase TorA (SO1232) is located in the periplasm and probably receives electrons from the TorC pentahemic c-type cytochrome, as is the case in Escherichia coli (Gon et al., 2001). In addition to the genes encoding the respiratory systems described above, the S. oneidensis genome contains several other genes coding for potential respiratory systems, suggesting the ability of this bacterium to respire still unknown substrates as exogenous electron acceptors (Heidelberg et al., 2002).
Although less is known about S. oneidensis chemotaxis, S. oneidensis has been shown to respond to a wide range of anaerobic electron acceptors including nitrate, nitrite, TMAO, DMSO, fumarate, thiosulphate, Mn(III) and Fe(III) (Nealson et al., 1995; Bencharit and Ward, 2005). The genome of S. oneidensis contains a large number of chemoreceptors (at least 27 genes annoted as MCP; Heidelberg et al., 2002) and cheA-3 (SO3207) was demonstrated to be essential for behavioural responses (Li et al., 2007). It was recently shown that S. oneidensis possesses two different stator systems driving a single polar flagellum. One is sodium-dependent (PomAB) and appears to have a major role in swimming motility, while the second is driven by the proton-motive force (MotAB) and was probably acquired recently for the adaptation of the cell to a low-sodium environment (Paulick et al., 2009).
In the present study, we establish that functional anaerobic respiratory systems are required for chemotaxis towards anaerobic electron acceptors. Moreover, we show that the chemosensory signal triggering this tactic response originates from the ΔpH component of the pmf. Our findings reveal that taxis towards electron acceptors in S. oneidensis is mediated by an energy taxis mechanism. Surprisingly, we observed that the four PAS-containing MCPs of S. oneidensis are not involved in this energy taxis behaviour. We identified here one major and four minor chemoreceptors playing a role in this mechanism.
Taxis towards TMAO is mediated by an energy taxis mechanism
We first investigated the chemotactic behaviour of S. oneidensis MR-1 towards TMAO with cultures grown under aerobiosis in the presence of TMAO. We performed taxis assays with microscope slide chemotaxis chambers (see Experimental procedures; Schweinitzer et al., 2008). These slide systems allow the formation of a chemoattractant gradient within a thin channel located between two larger reservoirs. When a gradient of TMAO was created, wild-type cells accumulated in the channel, whereas in the control experiment without TMAO a smaller number of cells was observed (Fig. 1). As a negative control we used a strain deleted of the cheA-3 gene that proved to be essential for chemotactic response of S. oneidensis (Li et al., 2007). Mutant cells failed to accumulate in the channel when a gradient of TMAO was formed (Fig. 1). These results indicate that S. oneidensis is tactic towards TMAO.
To further analyse this tactic response, we then used the plug in pond method that allows simultaneously testing of several different conditions (Gauden and Armitage, 1995; Bencharit and Ward, 2005). S. oneidensis strains were grown overnight under either aerobiosis or anaerobiosis in the presence or absence of TMAO. Experiments were performed in the presence of tetracycline to allow us to distinguish chemotactic responses from growth. Chemotactic behaviours were observed after 1 h incubation of the plates in an anaerobic jar to avoid oxygen competition.
Aerobic growth of wild-type strain in the presence of TMAO resulted in cells accumulating around plugs containing TMAO (Fig. 2), confirming the tactic response observed in Fig. 1. Addition of the same amount of TMAO to the soft agar as to the plug abolished this tactic response (data not shown). Moreover, cells grown aerobically in the absence of TMAO resulted in no taxis towards TMAO (Fig. 2). As TMAO is an inducer of the TMAO reductase, these results might indicate that TMAO taxis requires the presence of the TMAO reductase. We therefore tested the ability of S. oneidensis to reduce TMAO in the growth conditions used for the plug in pond assay. The wild-type strain exhibited a strong TMAO reductase activity (2.46 ± 0.47 μmol of TMAO reduced min−1 mg−1 of protein) when the culture was grown aerobically in the presence of TMAO. The same strain grown anaerobically in the presence of TMAO exhibited a TMAO reductase activity of 3.8 ± 0.25 μmol of TMAO reduced min−1 mg−1 of protein. This result seems quite surprising at first glance for an anaerobic respiratory system; however, we recently demonstrated that TorA, the TMAO reductase of E. coli, is also expressed and functional under aerobiosis (Ansaldi et al., 2007). In order to investigate the TMAO reductase activity in S. oneidensis under aerobic conditions, we constructed a strain (MR1ΔtorA) deleted of torA, the gene encoding the TMAO reductase of S. oneidensis. This strain was then grown aerobically in the presence of TMAO. Under these conditions, the TMAO reductase activity was similar to the background level (< 0.1 μmol of TMAO reduced min−1 mg−1 of protein) and no chemotactic response towards TMAO was observed (Figs 1 and 2). When the MR1ΔtorA strain was complemented by introduction of a plasmid containing a wild-type copy of the torA gene (Bordi et al., 2003), the chemotactic response towards TMAO was restored (data not shown). Therefore, a direct correlation exists between the presence of the TMAO reductase TorA and taxis towards TMAO. The TMAO reductase is a molybdoenzyme requiring a molybdenum cofactor to be active. Certain molybdoenzymes are inhibited by tungstate, a molybdate antagonist (Kletzin and Adams, 1996). To confirm that taxis towards TMAO requires a functional TMAO reductase, cells were grown aerobically in the presence of TMAO and tungstate, and then tested for taxis towards TMAO. Under these conditions, the wild-type strain exhibited no taxis towards TMAO (Fig. 2). This result confirms that taxis towards TMAO requires a functional TMAO reductase.
We then tested the chemotactic behaviour of strains grown anaerobically (Fig. 3). The wild-type strain grown in the presence of TMAO formed a well-defined ring around the plug containing TMAO. This behavioural response was quantified by using a modified version of the plug in pond assay (see Experimental procedures). As shown in Table 1, the number of cells was twofold higher when TMAO was present in the well compared with the condition where no electron acceptor was added. Surprisingly, both the wild-type strain grown in the absence of TMAO and the MR1ΔtorA strain grown in the presence or absence of TMAO produced similar wider and fuzzier rings (Fig. 3). These results suggest that while the well-defined ring was due to the activity of the TMAO reductase TorA, the fuzzy ring might be considered due to another TMAO reductase induced under anaerobiosis. The DMSO reductase of S. oneidensis (DmsA), whose gene expression is anaerobically induced, seems a good candidate as the DMSO reductase of E. coli has been shown to weakly reduce TMAO (Simala-Grant and Weiner, 1996; Gralnick et al., 2005). In support of this hypothesis, both the wild-type and the MR1ΔtorA strains grown anaerobically in the absence of TMAO exhibited a significant TMAO reductase activity (0.7 ± 0.14 and 0.69 ± 0.02 μmol of TMAO reduced min−1 mg−1 of protein respectively) when measured on DmsA-containing membrane fractions. The absence of a ring following the anaerobic growth of the wild-type strain in the presence of tungstate further supports the possible involvement of the DmsA molybdoenzyme, or of another molybdoenzyme, in the formation of the fuzzy ring (Fig. 3). To confirm this hypothesis, we constructed and submitted a dmsA deleted strain to the plug in pond assay. The MR1ΔdmsA strain, grown in the absence of TMAO, exhibited no TMAO reductase activity above the background level (< 0.1 μmol of TMAO reduced min−1 mg−1 of protein) and failed to accumulate around plugs containing TMAO whereas the same strain grown in the presence of TMAO formed a well-defined ring as expected (Fig. 3). As a negative control, we constructed a strain deleted of both torA and dmsA. No chemotactic behaviour towards TMAO was observed whatever the growth conditions (Fig. 3). The quantitative assay shown in Table 2 confirms that the double mutant is no longer tactic towards TMAO. We therefore conclude that under anaerobiosis, taxis towards TMAO mainly involves a functional TMAO reductase when cells are grown in the presence of TMAO, and to a lesser extent a functional DMSO reductase when cells are grown in the absence of TMAO. As MR-1 and MR1ΔdmsA grown anaerobically in the presence of TMAO exhibited a similar response towards TMAO (Fig. 3), we can conclude that TorA plays a key role in TMAO taxis under both anaerobiosis and aerobiosis.
Table 1. Quantitative taxis assay of the MR-1 strain towards different electron acceptors.
The wild-type strain was grown overnight in LB medium at 30°C in anaerobic conditions in the presence of 5 mM TMAO. Culture was tested for taxis towards a well filled with LB medium containing 20 mM of electron acceptor, as indicated.
After incubation of the plates for 1 h in an anaerobic jar, the medium present in the well was recovered, diluted and plated on LB agar. Results are expressed as 107 colony-forming units (cfu) ml−1 and the standard deviations are indicated.
Table 2. Quantitative taxis assay of wild type and mutant strains towards TMAO and DMSO.
Strains were grown overnight in LB medium at 30°C in anaerobic conditions in the presence of 5 mM TMAO, except for the MR1ΔtorAΔdmsA strain, which was grown in the presence of 10 mM fumarate. Each culture was tested for taxis towards a well filled with medium containing either TMAO (20 mM), DMSO (20 mM) or no electron acceptor, as indicated. When indicated, either nigericin (50 μM), valinomycin (10 μM) or amiloride (5 mM) was added to the plates. After incubation of the plates for 1 h in an anaerobic jar, the medium present in the well was recovered, diluted and plated on LB agar. Results are expressed as 107 colony-forming units ml−1 and the standard deviations are indicated.
Together, these results clearly indicate that taxis towards TMAO requires a functional terminal reductase in S. oneidensis. The fact that trimethylamine, the product of TMAO reduction, does not promote a chemotactic response of S. oneidensis (data not shown), probably means that TMAO taxis is mediated by an energy taxis mechanism.
Taxis towards DMSO is also mediated by an energy taxis mechanism
Given that the DMSO reductase allowed taxis towards TMAO under anaerobiosis, we wondered whether the DMSO reductase, DmsA, was also involved in DMSO taxis. As expected, we observed a chemotactic response to DMSO when the wild-type strain was grown anaerobically (Fig. 4). This result was confirmed by the quantitative assay that showed that twofold more cells were attracted to DMSO compared with a medium containing no electron acceptor (Table 1). In contrast, no response to DMSO was observed when the culture was grown aerobically (Fig. 4). This result suggests the involvement of the anaerobically induced enzyme DmsA (Gralnick et al., 2005) for DMSO taxis. To confirm this hypothesis, we measured the DMSO reductase activity under different growth conditions. The activity of the wild-type strain grown anaerobically reached 0.135 ± 0.01 μmol of DMSO reduced min−1 mg−1 of protein. We observed a very low activity (< 0.01 μmol of DMSO reduced min−1 mg−1 of protein) when the same strain was grown aerobically, comparable to that of a dmsA deleted strain grown anaerobically. We then performed plug in pond assays with either the MR1ΔdmsA strain grown anaerobically or the wild-type strain grown anaerobically in the presence of tungstate. As shown in Fig. 4, neither MR1ΔdmsA, nor MR-1 grown in the presence of tungstate formed a ring of cells around plugs containing DMSO. The quantitative assay shown in Table 2 further supports the involvement of a functional DmsA enzyme in DMSO taxis. Indeed, when the dmsA gene was deleted, no taxis towards DMSO was observed. Moreover, when the dmsA gene was not expressed (ΔarcA strain; Gralnick et al., 2005), the cells were no longer attracted by DMSO. Finally, a menaquinone minus mutant (menC::mini-Tn10 strain), which was shown to be impaired in DMSO respiration (Newman and Kolter, 2000), was also unable to respond to DMSO.
These results strongly suggest that taxis towards DMSO is mediated by an energy taxis mechanism involving the molybdoenzyme DmsA.
Taxis towards other exogenous electron acceptors
As taxis towards TMAO and DMSO appear to require active electron transport systems, we wondered if the chemotactic behaviour of S. oneidensis towards other electron acceptors such as nitrate, nitrite or fumarate obeys to the same rule, always involving an energy taxis mechanism. We first observed that strain MR-1 accumulated around plugs containing nitrate, nitrite or fumarate whether the cells were previously grown in the presence or absence of oxygen (Fig. 5). Quantification indicated that about twofold more cells were recovered with these three electron acceptors compared with the condition where no electron acceptor was added (Table 1). We next monitored the nitrate, nitrite and fumarate reductase activities as described in the Experimental procedures section. As expected, cultures exhibited strong nitrate, nitrite and fumarate reductase activities under anaerobiosis. More surprising however, was the detection of nitrate, nitrite and fumarate reductase activities in samples grown aerobically (Fig. 5). This result agrees with the hypothesis that taxis towards electron acceptors requires functional respiratory systems. As the nitrate reductase NapA is a molybdoenzyme, we carried out taxis experiments with cultures grown in the presence of tungstate. As shown in Fig. 5, taxis towards nitrate was abolished in the presence of tungstate whereas taxis towards nitrite or fumarate was conserved. This is not surprising as the nitrite and the fumarate reductases (NrfA and FccA respectively) are not molybdoenzymes. As expected, we detected no nitrate reductase activity in cultures grown in the presence of tungstate whereas nitrite and fumarate reductase activities remained unchanged (Fig. 5).
Altogether, these results reveal that taxis towards anaerobic electron acceptors requires functional terminal reductases.
Taxis towards exogenous electron acceptors depends on ΔpH
The requirement for a functional terminal reductase for chemotaxis means that the pmf generated by the reductase could be the signal triggering this tactic response. The pmf consists of two components, the membrane potential (ΔΨ) and the pH difference (ΔpH). To determine the chemosensory signal involved in this process, we tested the effect of specific ΔΨ or ΔpH uncouplers on electron acceptor taxis. We used valinomycin and nigericin known to disrupt ΔΨ and ΔpH respectively (Goulbourne and Greenberg, 1981; Ingham and Armitage, 1987). As shown in Fig. 6A, taxis towards TMAO was significantly affected when nigericin was added to the cells, whereas no effect was observed in the presence of valinomycin. The data shown in Table 2 confirm the result obtained for TMAO and reveal a similar effect for DMSO. The same results were also obtained when the uncouplers were tested for taxis towards nitrite, nitrate and fumarate (Fig. 6B). Therefore, disruption of ΔpH but not of ΔΨ affects electron acceptor taxis.
Rotation of the polar flagellum of S. oneidensis was reported to be mainly dependent on the sodium-dependent PomAB stator (Thormann et al., 2004; Paulick et al., 2009). To verify that the observed behavioural response came from motility driven by a Na+ motive force, we performed the plug in pond assays in the presence of amiloride. This compound specifically inhibits the Na+-driven flagellar motor by competing with Na+ in the medium (Kojima et al., 1999). As shown in Fig. 6 and Table 2, in the presence of amiloride, taxis towards anaerobic electron acceptors was inhibited. Thus electron acceptor taxis requires a functional Na+-driven flagellar motor. The fact that chemotaxis towards yeast extract was still observed in the presence of either nigericin or valinomycin confirms that motility is pmf independent in these experimental conditions (Fig. 6B).
A strain deleted of the four PAS-containing MCPs still responds to exogenous electron acceptors
Given that taxis towards nitrate, nitrite, TMAO, DMSO and fumarate is governed by an energy taxis mechanism, we wondered which MCP of S. oneidensis is involved in this process. The energy sensing mechanism of E. coli is particularly well understood, with two MCPs, Aer and Tsr, identified as the energy transducers (Bibikov et al., 1997; Rebbapragada et al., 1997). Aer is anchored to the membrane and contains a sensory PAS domain, whereas Tsr is a classical ligand binding transmembrane receptor lacking any known oxygen or redox responsive prosthetic group. Using the BlastP program, we found four MCPs in S. oneidensis (SO0584, SO1385, SO2123 and SO3404) showing significant similarities to Aer of E. coli (expect value less than e−47). The ScanProsite and SMART programs revealed that these four Aer-like MCPs are the only MCPs in S. oneidensis possessing a PAS domain. Interestingly, SO2123 shares significant similarities with the E. coli Tsr protein (expect value of e−82), but in contrast to its E. coli homologue, SO2123 contains a PAS domain. To investigate the roles of these MCPs, we constructed a strain (MR1ΔMCP-PAS) deleted of all four MCP genes (SO0584, SO1385, SO2123 and SO3404). We then tested the swimming behaviour of MR-1 and MR1ΔMCP-PAS on soft agar plates containing various electron acceptors. Figure 7A showed that MR-1 and MR1ΔMCP-PAS formed identical swim rings in the presence of fumarate. Similar results were obtained with the other electron acceptors (Fig. 7C; data not shown). Although a role in energy taxis of one of these four MCPs cannot be rejected, another MCP must play a key role in energy taxis towards electron acceptors.
One major and four minor MCPs are involved in energy taxis towards exogenous electron acceptors
In order to identify the MCP(s) involved in energy taxis, we cloned each of the 27 MCP-encoding genes of S. oneidensis into an expression vector and introduced the resulting plasmids in an E. coli strain deleted of all its five MCPs (BT3388; Yu et al., 2002). We then searched for MCP(s) allowing restoration of E. coli BT3388 chemotaxis towards exogenous electron acceptors or other substrates (C. Jourlin-Castelli et al., in preparation). Among the 27 MCPs tested, five of them (namely SO2240, SO3282, SO3642, SO3890 and SO4454) allowed E. coli BT3388 to chemotactically respond to nitrate. To test the involvement of these five MCPs in energy taxis, we constructed S. oneidensis strains deleted of each one of them (MR1Δ2240, MR1Δ3282, MR1Δ3642, MR1Δ3890 and MR1Δ4454). We then tested the swimming behaviour of MR-1 and of the five mutants on soft agar plates containing different electron acceptors. Strains MR-1, MR1Δ3642, MR1Δ3890 and MR1Δ4454 formed identical swim rings in the presence of fumarate or nitrate, whereas strain MR1Δ3282 formed a slightly smaller one (Fig. 7C). Interestingly, strain MR1Δ2240 formed a smaller swim ring than the other tested strains (Fig. 7B and C). This result suggests that the SO2240 chemoreceptor is involved in energy taxis in S. oneidensis. The role of SO2240 was confirmed by performing a taxis assay towards TMAO with the microscope slide chemotaxis chamber (Fig. 8). However, it is noteworthy that taxis is not totally abolished in strain MR1Δ2240 on swim plate, whereas no taxis was observed for the MR1ΔcheA-3 strain as expected (Fig. 7B). We therefore wondered whether one or more of the four other MCPs could play an additional role in this process. To test this possibility, we constructed double mutant strains deleted of SO2240 and of either SO3282, SO3642, SO3890 or SO4454. Interestingly all the double mutants formed smaller swim rings than the MR1Δ2240 single mutant (Fig. 7B and C). This result suggests that the SO3282, SO3642, SO3890 and SO4454 chemoreceptors play a minor but significant role in energy taxis.
Altogether, these results indicate that at least one major (SO2240) and four minor (SO3282, SO3642, SO3890 and SO4454) MCPs are involved in energy taxis in S. oneidensis.
Different lines of evidence presented here indicate that, in S. oneidensis, taxis towards the exogenous electron acceptors TMAO, DMSO, nitrate, nitrite and fumarate is governed by the general mechanism called energy taxis. First, deletion of dmsA and torA, the genes, respectively, encoding DMSO and TMAO reductases, abolished chemotactic response towards DMSO and TMAO (Figs 1–4, Table 2). Second, cells grown in the presence of tungstate, a molybdenum antagonist, were unable to exhibit a chemotactic behaviour towards TMAO, DMSO and nitrate while the response towards nitrite and fumarate was preserved (Figs 2–5). This behaviour was expected as the TMAO, DMSO and nitrate reductases are molybdoenzymes whereas nitrite and fumarate reductases are not (Stewart, 1988; Iobbi-Nivol et al., 2001; McCrindle et al., 2005). It indicates that a functional terminal reductase is required for taxis towards TMAO, DMSO and nitrate. Third, disruption of the ΔpH by the addition of nigericin abolished electron acceptor taxis (Fig. 6 and Table 2). Therefore, this energy taxis process depends on the difference of pH generated during respiration.
In a previous study, it was proposed that the chemotactic response towards electron acceptors occurred via a metabolism-independent mechanism (Nealson et al., 1995). Our results disagree with this proposal and clearly show that S. oneidensis responds to several electron acceptors by an energy taxis mechanism. Moreover, a recent study showed that S. oneidensis could respond to soluble forms of oxidized metals such as Fe(III) or Mn(III) using energy taxis (Bencharit and Ward, 2005). We therefore propose that taxis towards many, if not all, exogenous electron acceptors occurs via an energy taxis mechanism in S. oneidensis.
In this study, we observed the synthesis of several anaerobic respiratory systems in S. oneidensis even in the presence of oxygen (Figs 1 and 5). Although the aerobic expression of these systems seems surprising, their presence probably allows the cells to rapidly move towards and use anaerobic electron acceptors as soon as oxygen is depleted. This behaviour could thus be beneficial to cells encountering changing environments.
Energy taxis is usually mediated by chemoreceptors containing a PAS domain such as Aer in E. coli, VCA0658 in Vibrio cholerae and PA1561 in Pseudomonas aeruginosa, or others lacking any known oxygen or redox responsive group such as Tsr in E. coli, TlpD in Helicobacter pylori or Tlp1 in Azospirillum brasilense (Bibikov et al., 1997; 2000; Repik et al., 2000; Greer-Phillips et al., 2004; Hong et al., 2004; Boin and Hase, 2007; Schweinitzer et al., 2008). S. oneidensis possesses four MCPs with PAS domains, one of which shares significant similarities with the E. coli Tsr protein. Deletion of these four PAS-containing MCPs resulted in identical chemotactic responses of S. oneidensis towards the different electron acceptors compared with the wild-type strain (Fig. 7A and C). Because of the possible redundancy in the substrate specificity of MCPs, this result does not necessarily mean that none of these four MCPs plays a role in taxis towards electron acceptors. Nevertheless, this result strongly suggested the involvement of at least one other MCP in this process. Indeed, by introducing each MCP encoding gene of S. oneidensis in an E. coli strain deleted of all its five MCPs, we were able to identify five MCPs (SO2240, SO3282, SO3642, SO3890 and SO4454) potentially involved in this process. Deletion of the corresponding genes in S. oneidensis confirmed this assumption and indicated that MCP SO2240 plays a major role in energy taxis while the four remaining are also involved but to a lesser extent (Fig. 7B and C). In fact, the involvement of MCPs SO3282, SO3642, SO3890 and SO4454 appeared only when MCP SO2240 was absent. These five MCPs are seemingly localized in the cytoplasmic membrane via two hydrophobic membrane-spanning regions. With the exception of SO3890, they all possess a rather large periplasmic region. Their cytoplasmic regions contain the highly conserved signalling domain common to all chemoreceptors, which is preceded by an HAMP domain involved in signal transmission from the periplasmic sensing domain to the cytoplasmic signalling domain. Interestingly, two of these five chemoreceptors contain an additional domain. The major energy taxis chemoreceptor (SO2240) contains a Cache2 domain in its periplasmic region. The Cache domains were predicted to have a role in small-molecule recognition in a wide range of proteins including animal Ca2+-channel subunits and various bacterial chemoreceptors (Anantharaman and Aravind, 2000). We can therefore wonder whether this Cache domain allows SO2240 to detect ΔpH or a signal derivating from it. Another possibility could be that SO2240, like Tsr in E. coli, detects different signals: one related to ΔpH with a still unknown region and a second with its Cache domain. Interestingly, this particular chemoreceptor seems to be highly conserved in the Shewanella species. Indeed, a chemoreceptor sharing a high degree of identity with SO2240 (55–90%) was identified in most of the Shewanella strains sequenced to date. All of these MCPs have a similar size and possess a Cache2 domain in their N-terminal region. This could suggest a general key role of these chemoreceptors in energy taxis in Shewanella. The other chemoreceptor containing an additional domain is SO3890. This MCP possesses a hemerythrin-like domain located in the C-terminal end of the protein. Hemerythrins, present in invertebrates, are well-characterized nonheme diiron O2-carrying proteins (Stenkamp, 1994). The DcrH MCP from Desulfovibrio vulgaris (Hildenborough) possesses also a hemerythrin-like domain at its C-terminal end and it was the first reported example of hemerythrin in bacteria (Xiong et al., 2000). The hemerythrin-like domain of DcrH was also shown to contain a diiron site that binds O2 and was proposed to have an O2-sensing function (Xiong et al., 2000; Isaza et al., 2006). Hemerythrin-like proteins were reported to be abundant in prokaryotes and it is striking that all Shewanella spp. possess no other hemerythrin-like proteins but the highly conserved MCP-hemerythrin (SO3890 in S. oneidensis) (French et al., 2008). Therefore, it will be very interesting to determine the potential role played by the hemerythrin-like domain in energy taxis. We found no conserved domain in the N-terminal regions of the three remaining MCPs involved in energy taxis.
Our study has demonstrated that S. oneidensis responds chemotactically to a large set of electron acceptors by energy taxis. This behaviour has previously been demonstrated in some other cases, although only for a limited number of electron acceptors. The responses towards oxygen, DMSO and nitrate in A. brasilense (Alexandre et al., 2000) and towards DMSO and oxygen in Rhodobacter sphaeroides (Gauden and Armitage, 1995) depend on energy taxis. Given the exceptional ability of S. oneidensis to respire on various substrates, this organism proves to be an appropriate model to analyse the mechanism behind energy taxis towards electron acceptors and more specifically to characterize the new types of chemoreceptors involved in this process.
Strains, medium and growth conditions
The E. coli strains CC118 λpir and 1047/pRK2013 used for conjugations were grown in LB medium at 37°C (Herrero et al., 1990). All strains of S. oneidensis used in this study are derivatives of strain MR-1. Strain MR1ΔcheA-3 is deleted of cheA-3 (SO3207) (Li et al., 2007). Strain JG90 (ΔarcA) is deleted of arcA (Gralnick et al., 2005). Strain H1 (menC:: mini-Tn10) contains an insertion in menC (Newman and Kolter, 2000). Strains MR1ΔdmsA, MR1ΔtorA and MR1ΔtorAΔdmsA are, respectively, deleted of dmsA (SO1429), torA (SO1232) or torA and dmsA. Strain MR1ΔMCP-PAS is deleted of four MCP-encoding genes of S. oneidensis (SO0584, SO1385, SO2123 and SO3404). Strains MR1Δ2240, MR1Δ3282, MR1Δ3642, MR1Δ3890 and MR1Δ4454 are, respectively, deleted of SO2240, SO3282, SO3642, SO3890 or SO4454. Starting from strain MR1Δ2240, double mutant strains were also constructed. The motility of each strain was checked and no defect was observed. S. oneidensis was grown routinely at 30°C in Luria–Bertani (LB)-rich medium. The LM medium used for taxis assays in slide chemotaxis chamber and for swim plate assays comprised 0.2 g l−1 yeast extract, 0.1 g l−1 peptone, 10 mM HEPES (pH 7.4) and 10 mM NaHCO3. For the plug in pond and enzyme assays, the growth conditions were exactly the same: cells were grown overnight in LB medium at 30°C either in aerobic or anaerobic conditions. For aerobic conditions, growth was performed on a rotary shaker at 200 r.p.m. in an Erlenmeyer flask with baffled base to improve the oxygenation of the culture (100 ml of medium in a 500 ml-containing flask). For anaerobic conditions, we used full-filled bottles. The different solutions (DMSO, TMAO, NaNO3, NaNO2, fumaric acid, tungstate) were prepared from commercially available laboratory-grade chemicals. Nigericin, valinomycin and amiloride were purchased from Sigma Chemical Co and dissolved in ethanol.
Constructions of the deletion mutants
Briefly, upstream and downstream fragments flanking the genes to be deleted were PCR-amplified using S. oneidensis MR-1 genomic DNA. The resulting amplicons of about 500 pb were fused by overlap extension PCR using complementary 24 bp sequences present on the primers. The fused PCR amplicons were digested, purified and cloned into pKNG101, which carries the sacB cassette and the strAB genes encoding the streptomycin phosphotransferase (SmR), and possesses a lambda-pir-dependent (R6K) origin of replication (Kaniga et al., 1991). The ligation products were introduced into E. coli CC118 λpir. The resulting plasmids were then transferred from E. coli CC118 λpir strains to S. oneidensis by conjugation using the E. coli strain 1047/pRK2013 as a helper. Primary integrants were selected by plating on LB medium containing streptomycin at 50 μg ml−1 and colicin A to prevent growth of E. coli strains (Cascales et al., 2007). Selection for a second homologous recombination to remove the plasmid sequence was accomplished by plating the primary integrant onto a medium containing 6% sucrose. Colonies growing in the presence of sucrose were screened for sensitivity to streptomycin and then screened for deletion of the gene of interest by PCR. All deletions included the start codon and at least one-third of the gene.
Taxis assays in slide chemotaxis chamber. S. oneidensis strains were grown either aerobically in LB medium to an OD600 of 0.5–0.6 or anaerobically in LB medium containing 5 mM TMAO to an OD600 of 0.3–0.4. Cells were then harvested by centrifugation (10 min at 3500 r.p.m.) and resuspended in LM medium containing 25 μg ml−1 tetracycline, 15 mM lactate and 0.1% Tween-20 to yield an OD600 of about 5. For this assay, we used commercial microscope slide chemotaxis chambers (μ-slide chemotaxis, Ibidi GmbH). They consist of two reservoirs (40 μl each) separated by a thin channel. The two chambers of the same system were filled with LM medium containing 25 μg ml−1 tetracycline, 15 mM lactate and 0.1% Tween-20 to avoid cell adherence to the slide's surfaces. A gradient of chemoattractant was then created in the channel by applying a solution of 200 mM chemoattractant onto one reservoir, according to the manufacturer's instructions. A 3 μl volume of cell suspension was then applied onto the opposite reservoir. The bacteria transmigrating from the second reservoir into the channel were observed using a Zeiss microscope equipped with a 40× objective and a camera. Images were taken in the channel about 90 min after addition of the cells.
Plug in pond assays. Cells were grown as previously described. If necessary, the cells were concentrated and resuspended in LB medium to reach a final density of approximately 8 × 109 cells ml−1. Subsequently, 10 ml of this preparation was mixed with 15 ml of melted 1.5% LB agar containing tetracyclin (final concentration of 25 μg ml−1). The resulting mixture was poured into Petri dishes, and wells were filled with plugs of hard agar (1.5% LB agar) containing electron acceptors. The concentration of the electron acceptors in the plug was 20 mM for nitrate, nitrite, fumarate and DMSO and 50 mM for TMAO. The plates were placed in an anaerobic jar to avoid oxygen competition and incubated for 1 h at 15°C before being photographed. Each experiment was repeated at least three times.
To quantify the tactic response of each attractant, the same protocole was used except that the concentration of tetracyclin was decreased to 2.5 μg ml−1 and that the wells were filled with LB medium, instead of hard agar plugs, containing electron acceptors at a concentration of 20 mM. After 1 h of incubation in anaerobic jar, the medium present in the well was recovered, diluted and plated on LB agar. Plates were incubated overnight and colony-forming units were counted.
For the experiments performed with inhibitors, cells were centrifuged and then resuspended in the appropriate medium for each inhibitor, to reach a final density of approximately 8 × 109 cells ml−1. For valinomycin and nigericin, the medium contained 50 mM KCl, 5 mM MgCl2 and 5 mM lactate. For amiloride, the medium contained 50 mM Tris-HCl, 50 mM NaCl, 5 mM MgCl2 and 5 mM lactate. The chemotaxis assays were performed as previously described except that cells were mixed with 1.5% agar prepared in the appropriate medium.
Swim plate assays. Swim plates (soft 0.25% LM agar, 20 mM lactate, 2 mM fumarate or 0.25 mM nitrate) were inoculated with a sterile toothpick with cells grown on plates overnight at 30°C. Inoculated plates were incubated under anaerobic conditions for 3 days at 30°C before being photographed. Each experiment was repeated at least three times.
Nitrate and nitrite reductase activities. Cells were grown as previously described and the nitrate or nitrite reductase activity monitored on intact cells by measuring the nitrite accumulation (nitrate reductase activity) or nitrite disappearance (nitrite reductase activity) with reduced methyl viologen as an electron donnor (Cruz-Garcia et al., 2007). Overnight grown cells were resuspended in 0.32 M potassium phosphate (pH 7.1), at a density of 1.5 × 109 or 5 × 109 cells ml−1. Samples of 0.8 ml were mixed with 0.1 ml of 2 mM methyl viologen and the reactions started by adding 0.1 ml of a solution containing 0.5 M NaNO3, 0.1 M Na2CO3 and 0.05 M Na2O4S2 for nitrate reductase activity or a solution containing 10 mM NaNO2, 0.1 M Na2CO3 and 0.1 M Na2O4S2 for nitrite reductase activity. The reactions were then stopped after 20 or 60 min by vortexing to oxidize the methyl viologen, and the nitrite present in the reactions measured by adding 1 ml of 0.6% sulphanilic acid solution and 1 ml of 0.6% N-1-naphthylethylenediamine solution. The experiments were performed in duplicate from three independent experiments.
TMAO, DMSO and fumarate reductase activities. Cells were grown as previously described, resuspended in 40 mM Tris-HCl (pH 7.6), washed twice and disrupted by French press. The extracts were centrifuged at 15 000 r.p.m. and the recovered supernatants centrifuged at 45 000 r.p.m. to separate the soluble and membrane fractions. TMAO, DMSO or fumarate reductase activity was measured spectrophotometrically at 37°C by following the oxidation of reduced benzyl viologen at 600 nm, coupled to the reduction of TMAO, DMSO or fumarate, as previously described (Iobbi-Nivol et al., 1996). The TMAO and fumarate reductase activities were measured on soluble fractions and the DMSO reductase activity measured on membrane fractions. Protein concentrations were measured by the technique of Lowry.
We thank G. Alexandre for critical reading of the manuscript and E. Witty for English language editing. We are grateful to M. J. Ward and K. M. Thormann for providing us the cheA-3 mutant, D. K. Newman for the ΔarcA and menC mutants, B. L. Taylor for the BT3388 strain and E. Cascales for the kind gift of colicin. We also thank G. Ball for help with conjugation experiments, A. Bernadac for assistance with microscopic experiments and G. Panis for helpful suggestions. We are indebted to two anonymous reviewers for valuable comments. This work was supported by the Centre National de la Recherche Scientifique, the Université de la Méditerranée and a PEPS grant from Institut des Sciences Biologiques (INSB-CNRS). C.B. was supported by the Conseil Général PACA and the Fondation pour la Recherche Médicale (FRM).