SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

The Gram-negative metal ion-reducing bacterium Shewanella oneidensis MR-1 is motile by means of a single polar flagellum. We identified two potential stator systems, PomAB and MotAB, each individually sufficient as a force generator to drive flagellar rotation. Physiological studies indicate that PomAB is sodium-dependent while MotAB is powered by the proton motive force. Flagellar function mainly depends on the PomAB stator; however, the presence of both stator systems under low-sodium conditions results in a faster swimming phenotype. Based on stator homology analysis we speculate that MotAB has been acquired by lateral gene transfer as a consequence of adaptation to a low-sodium environment. Expression analysis at the single cell level showed that both stator systems are expressed simultaneously. An active PomB–mCherry fusion protein effectively localized to the flagellated cell pole in 70–80% of the population independent of sodium concentrations. In contrast, polar localization of MotB–mCherry increased with decreasing sodium concentrations. In the absence of the Pom stator, MotB–mCherry localized to the flagellated cell pole independently of the sodium concentration but was rapidly displaced upon expression of PomAB. We propose that selection of the stator occurs at the level of protein localization in response to sodium concentrations.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

A widespread mechanism for bacteria to cope with changes in environmental conditions is to actively leave unfavourable environments for more suitable ones. Thus, being motile provides a survival advantage to many bacteria (Armitage, 1999). Many species are motile by means of flagella which provide an efficient form of locomotion through liquid and more viscous environments as well as on surfaces (Daniels et al., 2004). Rotation of the flagellar filaments is powered by transmembrane ion gradients, i.e. proton or sodium motive force (McCarter, 2001; Yorimitsu and Homma, 2001; Blair, 2003). The latter is more common, but not restricted, to bacteria living in habitats with high sodium levels and/or elevated pH. Coupling ion motive forces across the membrane for torque generation is conferred by two major complexes. One is the rotor-mounted switch complex composed of the proteins FliG, FliM and FliN, which is also referred to as the ‘C-ring’. The second component is a membrane-embedded complex that forms the stator (Berg, 2003). The stator system consists of several, in Escherichia coli up to 11 (Reid et al., 2006), complexes surrounding the switch complex. Each stator complex likely contains four A- and two B-subunits to form two specific ion channels (Blair and Berg, 1990; Asai et al., 1997; Yorimitsu et al., 2004). The stator protein subunits are commonly referred to as Mot (A and B, e.g. for E. coli) or Pom (A and B, e.g. for various Vibrio species). Mot/PomA possess four transmembrane domains and one cytoplasmic segment that is thought to interact with the switch complex component FliG for torque generation (Zhou et al., 1998; Asai et al., 2003). Mot/PomB has a single N-terminal membrane spanning domain that harbours an Asp residue crucial for mediating ion flow and torque generation, and a periplasmic segment with a peptidoglycan-binding motif (De Mot and Vanderleyden, 1994; Asai et al., 1997). Stator function of the complex is thought to be ensured by C-terminal anchoring of the B-subunits to the peptidoglycan layer (Berg, 2003; Blair, 2003; Kojima and Blair, 2004a). In Vibrio it was demonstrated that two auxiliary proteins, designated MotX and MotY, are essential for PomAB-mediated flagella rotation (McCarter, 1994a,b; Terashima et al., 2006).

Results from numerous studies with various bacteria have been merged into a model for the assembly of the bacterial flagellar motor–stator complex that, in the main, appears to apply to both sodium- and proton-dependent systems. Upon expression, the stator subunits form a stable pre-complex that diffuses in the inner membrane before being incorporated into the stator ring system that is surrounding the C-ring to become active ion channels (Van Way et al., 2000; Fukuoka et al., 2005; Leake et al., 2006). In E. coli, it has recently been shown that the stator ring system is a highly dynamic complex, in which the stator units are constantly exchanged with a membrane-located pool of stator units over an estimated turnover time of about 30 s (Leake et al., 2006).

In contrast to most strains of E. coli that possess one stator system that interacts with a single switch complex system, a number of species have been described that harbour more than one stator system. According to available genome data, this situation is widespread among bacteria (Toutain et al., 2005). The well-studied marine bacteria Vibrio parahaemolyticus and Vibrio alginolyticus possess two flagella systems, both powered by different stator systems. The single polar flagellum depends on a sodium-driven stator. In addition, under conditions of high viscosity lateral flagella are induced that are driven by a second stator powered by the proton motive force (Atsumi et al., 1992). Furthermore, Aeromonas species express lateral flagella in addition to the single polar system when swarming over surfaces. Both systems are thought to depend on two different stator complexes (McCarter, 2004; Merino et al., 2006). In these two organisms the number of stator systems equals the number of the flagellar systems, and each stator unit connects specifically to the corresponding switch complex.

In other bacteria, e.g. Pseudomonas aeruginosa PAO1 and Bacillus subtilis, the number of potential stator systems exceeds that of flagella systems. P. aeruginosa PAO1 harbours two different stator systems, MotAB and MotCD, which power a single polar flagellum by an unknown mechanism. Both have been demonstrated to differentially contribute to swimming motility in high viscosity environments, during swarming, and in biofilm formation (Doyle et al., 2004; Toutain et al., 2005; 2007). In B. subtilis, two separate stator systems, designated MotAB and MotPS, are functionally linked to a single peritrichous flagella system. Interestingly, MotAB is proton-driven and MotPS depends on sodium motive force (Ito et al., 2004). In these two organisms, the interaction between stator subunits and the switch complex must be effectively coordinated in order to ensure selection of the appropriate stator system according to the environmental conditions. It is as yet unknown how coordination of stator selection is achieved.

The γ-proteobacterium Shewanella oneidensis MR-1 is environmentally important due to its ability to use a wide range of alternative external electron acceptors. The organism requires flagella-mediated motility to swim towards redox-active surfaces and for community formation (Thormann et al., 2004; Bencharit and Ward, 2005). In the present study, we show that S. oneidensis MR-1 harbours two different stator systems with different ion specificity to drive a single polar flagellum. We demonstrate that both systems are simultaneously present and that stator selection most likely occurs at the level of protein localization.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Shewanella oneidensis MR-1 possesses two potential stator systems to power a single polar flagellum

Genome sequence analysis with respect to elements potentially involved in motility revealed that S. oneidensis MR-1 harbours one set of flagella genes including a single orthologue to the switch complex protein encoding gene fliG. However, two gene clusters are present that could encode two complete stator systems. One of the stator systems (SO1529/1530) exhibits homologies to the well-characterized sodium-dependent system of the Vibrio species. Accordingly, this cluster has been annotated as pomAB. Genes sharing homologies to motX and motY, required for PomAB function in Vibrio, were also annotated in S. oneidensis MR-1. The second two-gene cluster encoding putative stator components was designated motAB (SO4287/4286). Homologies of the deduced proteins are highest to a stator system found in Aeromonas hydrophila that has not yet been characterized in detail, but is hypothesized to function as a proton pump (Merino et al., 2006).

We found that S. oneidensis MR-1 displays swimming in complex media with high [Luria–Bertani (LB)] or low [Lactate Medium (LM)] nutrient concentration as well as in minimal medium (4 M). Swarming across surfaces was not observed. In order to monitor the flagellation state, cells harvested from planktonic cultures and polystyrene surfaces were characterized by flagella staining and subsequent microscopic observation. Under all conditions tested, S. oneidensis MR-1 exclusively possessed a single polar flagellum.

motAB has most likely been acquired by lateral gene transfer

In order to determine whether the two stator–one flagellar system ratio is a common feature among the genus Shewanella, we conducted a bioinformatic analysis on the genome data of 16 sequenced Shewanella strains to identify putative flagella and stator systems (Table 1). This analysis revealed that all Shewanella isolates harbour at least one set of flagella genes along with orthologues to the sodium-dependent PomAB stator system. In addition, some Shewanella isolates (e.g. S. putrefaciens CN-32 and S. sediminis HAW-EB3) possess additional flagella systems, as has recently been identified in S. piezotolerans (Wang et al., 2008). This group of Shewanella strains was also found to harbour a distinct second stator system exhibiting homologies to those of the proton-dependent Vibrio lateral flagellar (laf) system (Atsumi et al., 1992). An exception is S. baltica OS155 which appears to have two flagellar systems but only an additional MotB orthologue without a corresponding MotA partner. Among Shewanella isolates that possess a single flagellar system, two species (S. frigidimarina NCIMB 400 and S. sp. ANA-3) also harbour an additional single MotB orthologue but lack a corresponding MotA partner.

Table 1.  Putative stator and flagella systems in Shewanella isolates.
StrainNumber of putative flagella systemsNumber of putative stator systemsGene ID (coupling iona)
  • a.

    As based on homology comparisons; smf, sodium motive force; pmf, proton motive force.

S. amazonensis11Sama2432/2433 (smf)
S. baltica OS15521.5Sbal1360/1361 (smf) Sbal13946 (pmf)
S. baltica OS18511Shew1851346/1347 (smf)
S. denitrificans OS21732Sden2568/2567 (smf) Sden3632/3631 (pmf)
S. frigidimarina NCIMB 40011.5Sfri2787/2786 (smf) Sfri2374 (pmf)
S. oneidensis MR-112SO1529/1530 (smf) SO4287/4286 (pmf)
S. loihica PV-411Shew2768/67 (smf)
S. pealeana ATCC7003422Spea2987/2988 (smf) Spea0056/0057 (pmf)
S. piezotolerans WP322Swp5126/5127 (smf) Swp3615/3616 (pmf)
S. putrefaciens CN-3222Sputcn321279/1278 (smf) Sputcn323447/3448 (pmf)
S. sediminis HAW-EB322Ssed3326/3325 (smf) Ssed0050/0049 (pmf)
S. sp. ANA-311.5Shewana32898/2897 (smf) Shewana33786 (pmf)
S. sp. MR-411Shewmr42728/2727 (smf)
S. sp. MR-711Shewmr72800/2801 (smf)
S. sp. W3-18–122Sputw31812827/2828 (smf) Sputw31810491/0492 (pmf)
S. woodyi111177/1178 (smf)

Thus, with two individual complete stator systems and a single flagellar system, S. oneidensis MR-1 represents an exception among the species of Shewanella. Homology comparisons revealed that, as expected, PomAB of S. oneidensis MR-1 has orthologues in all other Shewanella species (Fig. S3A and B). In contrast, the MotAB proteins are not closely related to other Shewanella stators, but were instead found to group together with systems found in Aeromonas species (Fig. S3A and B). Moreover, both genes exhibit a significantly lower G+C content (39.04% and 36.32% for motA and motB respectively) compared with the genome average of 45.88% (Heidelberg et al., 2002). Based on these findings, we hypothesize that S. oneidensis MR-1 has acquired motAB (SO4287/4286), a potentially proton-dependent stator system, by horizontal gene transfer.

Both stators promote swimming motility in S. oneidensis MR-1

In order to determine the function of the two stator systems for motility in S. oneidensis MR-1, in-frame deletions were generated in the genes encoding the putative stators PomAB (ΔpomAB), MotAB (ΔmotAB) and in both (ΔpomABΔmotAB). In addition, a mutant was constructed lacking the three putative flagellin subunits of S. oneidensis MR-1 (SO3236–SO3238). This mutant is referred to as Δflag. None of the mutations introduced causes growth defects (data not shown). Mutants lacking the stator components PomAB, MotAB, or both were demonstrated by flagella staining and subsequent microscopic observation to be flagellated like wild-type cells. As expected, a flagellar filament was not observed in Δflag mutants (data not shown).

As assessed by light microscopy, 50–80% of wild-type and ΔmotAB cells displayed swimming in both LB and LM medium. On soft agar plates both strains were indistinguishable (Fig. 1A). In contrast, the majority (> 99%) of ΔpomAB mutants did not display swimming, and only a subpopulation of ΔpomAB mutant cells (0.1–1%) was motile (Fig. 1C). The wild-type phenotype of the deletion strains was restored upon ectopic expression of pomAB and motAB, respectively, from a plasmid (data not shown). Control strains ΔpomABΔmotAB and Δflag that were lacking both stator systems and the flagella filament, respectively, did not display swimming motility on soft agar plates or when grown in planktonic culture (Fig. 1A). This suggested that generally both systems are individually sufficient to drive flagella rotation, but the PomAB stator appears to have a more fundamental role in swimming motility. Regularly, flares of swimming cells evolved from ΔpomAB colonies on soft agar plates after prolonged incubation (Fig. 1A). Cells isolated from the fringes of such flares displayed elevated swimming ability when passed from plate to plate but rapidly lost it when repeatedly grown in planktonic cultures, a phenomenon similarly observed in B. subtilis (Ito et al., 2004). In these strains, referred to as ‘up-motile mutants’, the fraction of cells exhibiting motility was drastically increased (> 20%). Three up-motile mutants were selected at random and were shown to still harbour the deletion in pomAB. The motAB locus including the promoter region was amplified from the three isolates, and sequencing revealed that no mutation has occurred compared with the wild-type strain. Although the nature of the up-motile phenotype remains unknown, we conclude that it is not based on a switch in coupling ion specificity, because this is thought to be solely determined by the stator units. An up-motile strain was therefore included in subsequent studies on the stator physiology.

image

Figure 1. A. Swimming phenotype of S. oneidensis MR-1 mutants on a LM soft-agar plate after 24 h of incubation. Displayed are (clockwise, starting at the top): MR-1 wild type, MR-1 ΔpomAB, MR-1 Δflag, MR-1 ΔpomABΔmotAB, MR-1 ΔmotAB. B. Swimming speed of S. oneidensis MR-1 wild type (◆), ΔpomAB (up-motile) (▴) and ΔmotAB (▪) mutant cells (in LM + 5% PVP) in dependence of the sodium concentration. The error bars show the standard deviation. C. Swimming ability of wild-type and mutant cells in LM medium supplemented with 50 mM NaCl and 5% PVP.

Download figure to PowerPoint

PomAB and MotAB depend on different coupling ions

The homologies to sodium- and proton-dependent stator systems, respectively, suggested that PomAB and MotAB use different ion motive forces to drive flagellar rotation. In order to determine coupling ion specificity and contribution of the stators to swimming, we analysed swimming speed as a function of sodium concentration (Fig. 1B). Both wild type and ΔmotAB mutants displayed elevated swimming speed with increasing concentration of sodium. At low sodium levels, the wild type was significantly faster than the ΔmotAB mutant (38.1 ± 4.7 μm s−1 versus 25.1 ± 3.5 μm s−1), while at higher sodium concentrations this effect was negligible and swimming speeds of wild type and mutant was indistinguishable. In contrast, the swimming speed of ΔpomAB mutants was unaffected by sodium levels.

In a second set of experiments, inhibitors were used to determine the coupling ion specificity. In numerous studies phenamil, an amiloride analogue, has been shown to effectively and specifically inhibit sodium-dependent bacterial motors in various organisms (Atsumi et al., 1990; 1992; Kojima et al., 1999). Thus, we reasoned that PomAB of S. oneidensis MR-1 might also be inhibited upon exposure to this compound. Concentrations of up to 75 μM phenamil did not affect growth of S. oneidensis MR-1 (data not shown). Upon addition of 50 μM phenamil, the motility of the wild type was instantly reduced, but a fraction of cells (< 1%) still exhibited swimming motility. Addition of phenamil blocked swimming of a ΔmotAB mutant, while neither ΔpomAB nor ΔpomAB up-motile mutants were affected.

As a second inhibitor, the protonophor carbonylcyanide m-chlorophenylhydrazone (CCCP) was used that collapses the proton motive force. No swimming was observed upon addition of 10 μM CCCP to ΔpomAB or ΔpomAB up-motile mutants. Motility of the wild type and ΔmotAB mutant was drastically decreased upon addition of 10 μM CCCP, but a number of cells remained actively swimming at reduced speed. This drastic effect on swimming motility was unexpected, because Vibrio strains were shown to maintain sodium-dependent swimming for longer periods of time in the presence of CCCP (Atsumi et al., 1992).

From the data obtained, we conclude that pomAB and motAB encode functional stator systems that depend on different coupling ions. Sodium-dependent PomAB has a more important function, but swimming of S. oneidensis MR-1 benefits from the presence of proton-dependent MotAB under conditions of low sodium concentrations. It remained unclear how selection of the two stator systems is regulated.

pomAB and motAB are concurrently transcribed

To determine whether stator selection is regulated at the transcriptional level in response to environmental sodium concentrations, a transcriptional fusion of the mot and pom promoters to lucB was constructed in a low copy broad-host range plasmid and transferred to S. oneidensis MR-1. Promoter activity was then determined in cultures grown at different sodium concentrations by measuring luciferase activity based on LucB expression. Both promoters were found to be active in all sodium concentrations tested. Notably, an approximately 10-fold higher transcription rate from the mot promoter was measured by the lucB fusions (Fig. 2A). Activity of the mot promoter was highest at 2 mM sodium and thereafter decreased constantly. The pom promoter exhibited a similar increase in activity from 0 to 2 mM and a second peak of activity at 100 mM sodium concentration. Overall, transcriptional regulation of both promoters in response to sodium concentrations appears to occur only at a moderate level. Also the protein levels remained relatively constant; however, the level of PomB was observed to drop under conditions of low sodium (Fig. S1).

image

Figure 2. Transcription analysis of pomAB and motAB. RLU, relative light unit. A. Activity of the pom (green) and mot (yellow) promoters in dependence of the sodium concentration, as measured by luciferase activity produced from a transcriptional fusion of the corresponding promoter to lucB. The cells displayed were grown for 8 h in LM medium supplemented with the displayed amount of sodium chloride. B. Activity of the pom and mot promoters at the single cell level. Displayed are fluorescence images of the untagged wild type and cells harbouring the Ppom::gfp(ASV) and the Pmot::gfp fusion respectively. The cells were obtained from the exponential growth phase in LM medium. Corresponding differential interference contrast micrographs of the same area are displayed in the image.

Download figure to PowerPoint

To further determine whether a subpopulation of cells is expressing only one of the systems or if both are expressed simultaneously, gene expression was studied at the single cell level. Short-lived gfp(ASV) was fused to the pomAB and motAB promoters, respectively, to result in translational fusions that were integrated into the chromosome by a single homogeneous cross-over event into the promoter region of the corresponding gene. Subsequently, cells harbouring the reporter fusions were grown to exponential phase in LM medium and visualized by fluorescent microscopy. The higher promoter activity that was measured by the transcriptional reporter fusions to lucB was also reflected by an elevated level of fluorescence in cells harbouring the Pmot -gfp(ASV) fusion. We found that more than 99% of the cells exhibited a distinct fluorescence signal with either promoter fusion (Fig. 2B). Even using unstable Gfp derivatives might not fully display short-term shifts in expression as occur, e.g. during cell cycle regulation in Caulobacter (England and Gober, 2001). However, the results strongly suggest that both pomAB and motAB are simultaneously transcribed in each cell.

PomB and MotB have different localization dynamics

The promoter–reporter fusions indicated that both Pom and Mot stators are simultaneously expressed in the cells independent of the sodium concentration in the medium. To analyse how a possible functional interplay between the stators could be achieved, fusions of stator subunits to fluorescent proteins were generated. Different fusions to the cytoplasmic N-termini of PomB or MotB resulted in non-functional proteins. Therefore, mCherry was fused to the C-terminus of PomB and MotB following the putative peptidoglycan binding domain which is located in the periplasm. The pomB–mCherry and motB–mCherry fusions were individually integrated into the chromosome at the native locus by a single homologous cross-over event.

Antibodies specific to PomB and MotB were used to detect native and fusion proteins after resolution by PAGE separation and subsequent immunoblotting. Both MotB–mCherry and PomB–mCherry fusion proteins were found to be stably expressed (Fig. S1A and B). Both fusions were also found to exert no negative effects on motility as determined by light microscopy and on LM soft agar plates (data not shown). Cells harbouring pomB–mCherry and motB–mCherry fusions were grown planktonically in LM medium and localization of the fusion proteins was visualized by fluorescence microscopy (Fig. 3). As a control, mCherry was solely overexpressed from a plasmid and gave a strong and even fluorescent signal throughout the whole cell without cluster formation at any region (data not shown).

image

Figure 3. Localization of PomB–mCherry and MotB–mCherry in S. oneidensis MR-1. Displayed are ‘differential interference contrast’ and fluorescence micrographs of cells harbouring pomB–mCherry (left panel) and motB–mCherry fusions (right panel). The cells were grown in LM medium containing the indicated concentration of NaCl. The arrows mark the positions of polar localization.

Download figure to PowerPoint

PomB–mCherry localized to the cell periphery and to one cell pole in 70–80% of the cell population (Fig. 3). By costaining the flagella filament using Alexa Fluor 488 succinimidyl ester, it was confirmed that localization of PomB–mCherry occurs at the flagellated cell pole (Fig. S2), a strong indication for incorporation into the stator ring system (Fukuoka et al., 2005; Leake et al., 2006). The percentage of cells exhibiting polar localization of PomB–mCherry was not affected by the sodium concentration. However, when no sodium was added, the fluorescence intensity at the cell pole dropped, strongly suggesting a lower polar abundance of PomB–mCherry under these conditions. This was in accordance with an overall lower level of the protein under conditions of low sodium (Fig. 4B and C).

image

Figure 4. Dependence of stator localization on the sodium concentration. A. Percentage of polar localization of PomB–mCherry and MotB–mCherry stator proteins, respectively, grown in LM medium with the indicated concentrations of sodium. B. Western Blot analysis of PomB–mCherry and MotB–mCherry expression of cells grown in LM medium supplemented with sodium chloride as indicated. C. Relative fluorescence intensity at the pole of cells bearing a pomB–mCherry fusion, grown in LM media containing the displayed amount of sodium chloride.

Download figure to PowerPoint

In contrast to PomB–mCherry, localization of MotB–mCherry strongly depended on the sodium concentration in the medium. At concentrations greater than 5 mM, MotB–mCherry was predominantly found in the cell membrane with only a minor subpopulation (< 25%) containing the fusion protein localized to the flagellated cell pole (Fig. 3). However, under conditions of decreased sodium concentrations (0–2 mM), the localized fraction drastically increased to up to 70% (Fig. 4A). The protein levels of MotB–mCherry were found not to change in response to altered sodium concentrations (Fig. 4B).

The localization pattern suggested that at higher sodium concentrations two subpopulations may exist that either have the Pom or the Mot stator at the flagellated cell pole. Polar localization of both stators in more than 70% of the population at low sodium concentrations indicates that both are most likely colocalizing and, thus, might be simultaneously incorporated into the stator ring system.

Presence of PomAB influences the localization of MotAB

The gradual increase in polar localization of MotB–mCherry indicated that the sodium level influences the localization efficiency of one or both stators. PomB–mCherry was always found to be localized at the cell pole and a high level of polar localization of MotB–mCherry coincided with a decreased abundance of PomB–mCherry. We therefore hypothesized that at high sodium levels PomAB might localize to the flagellated cell pole more efficiently, excluding the MotAB stator from being incorporated into the stator ring system. If this is true, the Mot complex should be able to readily localize to the poles in the absence of PomAB. In order to test this hypothesis, a motB–mCherry fusion was created in a ΔpomAB background. As predicted, in this strain MotB–mCherry exhibited localization at the flagellated pole in more than 80% of the cells, independent of the sodium concentration (Fig. 5). We then reintroduced pomAB into this strain under control of an l-arabinose-inducible promoter on a low copy vector. Upon induction of PomAB expression, polar localization of MotB–mCherry decreased and the fluorescent signal was almost exclusively confined to the cell membrane after 100 min in 85% of the cells. In contrast, no displacement occurred in the uninduced control. During induction, the protein levels of MotB–mCherry were found to remain unchanged (Fig. 5). Displacement of MotB–mCherry from the pole was similarly observed under conditions of low sodium (data not shown). Thus, presence and abundance of PomAB to some extent influence MotAB localization.

image

Figure 5. Localization of MotAB in dependence of PomAB. A. Micrographs (differential interference contrast and fluorescence microscopy respectively) of cells harbouring MotB–mCherry in a ΔpomAB background. B. Displacement of MotB–mCherry from the cell pole by PomAB. Displayed are micrographs of cells harbouring a motB–mCherry fusion in a ΔpomAB background. The images were taken immediately and 90 min after induction of PomAB expression under control of a Para-promoter (left panel) and of a non-induced control (right panel).

Download figure to PowerPoint

Hybrid stators are non-functional

Our results strongly suggest that all four stator components, PomA, PomB, MotA and MotB, are simultaneously present in the cell. Thus, instead of being assembled to PomAB and MotAB stators, the subunits might potentially form hybrid stators, consisting of PomAMotB and MotAPomB. Former studies provide examples where such stator hybrids were artificially created and found to be active (Asai et al., 1999; Ito et al., 2005). We therefore determined whether such assembly of hybrid stators can naturally occur in S. oneidensis MR-1 and, if so, whether they result in functional complexes. For this purpose, two strains were constructed: ΔpomABΔmotB and ΔmotABΔpomB. Hence, these strains solely expressed either PomA or MotA, and were, as expected, non-motile. We then introduced low-copy vectors into these strains in which pomB–mCherry and motB–mCherry with an optimized ribosome binding site were placed under control of an l-arabinose-inducible promoter. These vectors were also introduced into the wild type and a ΔpomABΔmotAB mutant. The strains harbouring the corresponding vectors were grown in LM medium with either 100 mM or no sodium chloride added and the expression of the fusion proteins was induced by addition of l-arabinose. Functional stator formation was determined by assessing swimming motility and functional localization by light and fluorescent microscopy.

In all strain backgrounds, overexpression of MotB–mCherry led to an accumulation of the protein in the cytoplasmic membrane (Fig. 6, left panel). Occasional polar localization was only observed in the ΔpomABΔmotB strain which provides the missing cognate subunit MotA. This strain also became weakly motile. No polar localization was detected in the ΔpomABΔmotAB double mutant, the wild type, or a ΔmotABΔpomA mutant. This suggests that in S. oneidensis MR-1, MotB only interacts effectively with the corresponding MotA subunit and that this interaction is crucial for polar localization. The apparent inability to form a complex in the wild-type background indicates that functional interaction is improved by transcriptional and translational coupling, as has been described for E. coli (Wilson and Macnab, 1990; Van Way et al., 2000) and Vibrio (Yorimitsu et al., 1999).

image

Figure 6. Hybrid stator formation in S. oneidensis MR-1. Displayed are micrographs of cell expressing MotB–mCherry (left panel) and PomB–mCherry (right panel) ectopically from a plasmid. The corresponding strain background is indicated above the micrograph. Below is indicated whether or not swimming motility was observed. The arrows mark the position of stator clusters.

Download figure to PowerPoint

PomB–mCherry was also found to unspecifically accumulate in the cell membrane in a ΔpomABΔmotAB double mutant (Fig. 6, right panel). Thus, a corresponding A-subunit is also strictly required for polar localization of PomB. In a wild-type background localization could occasionally be observed, suggesting that, in contrast to MotB–mCherry, ectopically expressed PomB–mCherry successfully competes with the native PomB or MotB subunits for stator complex formation. Such interaction was observed with either MotA or PomA. Assembly of PomAPomB–mCherry yielded a functional stator complex that promoted swimming motility. In contrast, MotAPomB–mCherry was observed to localize to the cell pole but never gave rise to motile cells.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Flagella-mediated motility of Shewanella

The genus Shewanella generally has been described as being motile by means of only a single unsheathed polar flagellum (Venkateswaran et al., 1999). Our genetic analysis indicates that a single polar flagellar system is a common feature among all species of Shewanella. This system is, most likely, powered by the sodium motive force, not unexpected for a genus that has mainly been isolated from marine environments (Hau and Gralnick, 2007). However, as the genome sequence data analysis strongly suggests, flagella-mediated motility is by far more versatile within this environmentally relevant group of bacteria. A number of species were found to harbour a second flagella system, as has only recently been demonstrated for the deep-sea isolate S. piezotolerans (Wang et al., 2008). A second flagellar system probably contributes to the versatility especially of species such as S. putrefaciens that were also identified in various non-marine environments, e.g. human clinical specimens (Khashe and Janda, 1998; Holt et al., 2005).

In contrast to all other Shewanella species that have been sequenced so far, S. oneidensis MR-1 has been isolated from a freshwater environment (Venkateswaran et al., 1999). This strain shares close sequence homology to two marine species stemming from Sargasso Sea samples (Venter et al., 2004) and requires a relatively high level of sodium concentration for optimal growth (Liu et al., 2005). Therefore, it has been hypothesized that S. oneidensis MR-1 has only recently been introduced from a marine into a fresh water habitat and is in the process of adaptation to a low-sodium environment (Hau and Gralnick, 2007). It has been suggested that S. oneidensis MR-1 has the ability to acquire and chromosomally integrate foreign genes, based on integron and ISSod25-encoded integrases (Romine et al., 2008). Therefore, we speculate that the dual stator system of S. oneidensis MR-1 is a consequence of horizontal gene transfer driven by long-term motility adaptation to low-sodium conditions. The closest homologue to the motAB stator is a system found in Aeromonas and, notably, it has been shown that Shewanella and Aeromonas can be co-isolated from the same habitat (Morohoshi et al., 2005).

The role of two different stator systems in S. oneidensis MR-1

We have demonstrated that in S. oneidensis MR-1, the proton- and sodium-powered stators are individually functional for swimming motility. This requires that both stator systems interact with the same switch complex, namely FliG, which is thought to represent the interface between the static and rotating part of the flagellar motor (Lloyd et al., 1996; Berg, 2003; Yorimitsu et al., 2003; Kojima and Blair, 2004b). Proton- and sodium-dependent systems differ in charged amino acid residues that are thought to be necessary for electrostatic interactions with FliG (Zhou et al., 1998; Yorimitsu et al., 2003; Obara et al., 2008). As in the present study, it was reported that functional interactions of various stator complexes with a single switch complex commonly occur at drastically different efficiency and functionality (Gosink and Häse, 2000; Asai et al., 2003; Ito et al., 2004). These differences indicate that the switch complex is optimized to work with a particular stator system.

However, as in the present study with S. oneidensis MR-1, hypermotile mutants were frequently observed. Such up-motile phenotypes had been attributed to, e.g. a higher abundance of the stator proteins caused by a mutation leading to increased transcription (Ito et al., 2004; Terahara et al., 2006). Our expression and localization studies on the stator subunits in S. oneidensis MR-1 demonstrated that MotAB is abundant and, in absence of PomAB, readily localizes to the flagellated cell pole. Even overexpression of MotAB in a ΔpomAB background did not lead to enhanced swimming (data not shown). The up-motile phenotype was also not due to mutations in the corresponding genes. Thus, we hypothesize that in S. oneidensis MR-1, the difference in function is primarily due to lack of functional interaction or activation of the MotAB stators. Measurements of the swimming speeds revealed a significant positive impact of the Mot stator under conditions of low sodium, even though few cells remain motile in the absence of the Pom system. In a low-sodium environment, both stators localize to the flagellated pole in the majority of cells, strongly suggesting simultaneous incorporation of both systems into the stator ring complex. We speculate that incorporation and activation of the stator systems may occur cooperatively and that during this process the Mot stators benefit from the presence of the Pom system. This would explain the phenotype observed.

The rapid occurrence of hypermotile mutants, as described in diverse systems, suggests that stator–switch complex interactions are highly adaptive. Given an appropriate ecological niche, a population of cells may arise in which a different stator system takes over a more important role. Thus, the contribution of MotAB to the swimming motility of S. oneidensis MR-1 may generally be more significant in the organism's natural fresh water environment.

Regulation of stator selection

The dual stator system is not a peculiarity of S. oneidensis MR-1. In numerous species, such as P. aeruginosa and B. subtilis (Doyle et al., 2004; Ito et al., 2004; Toutain et al., 2005), selection of two or even more distinct stators has to be regulated according to the environmental situation. Among several different possible scenarios, one is that each stator system is confined to a subpopulation of cells, e.g. by strict transcriptional control in response to environmental changes. A second possibility is that different stator complexes are concurrently present in the cell and selection occurs through shifting the stator ratio in the flagellar motor complex. This could be achieved by changing the corresponding protein level of the stator systems, altering the localization efficiency of the stator complexes, or a combination of both. We have provided here the first example of stator selection at the level of functional protein localization in response to changes in environmental conditions.

Our results suggest that in S. oneidensis MR-1, both stator systems are simultaneously present in each cell. Under high-sodium conditions, Pom is localizing to the cell pole more efficiently than Mot, preventing incorporation of the second stator into the stator ring system. Polar localization of Mot gradually increases with decreasing sodium levels, while the protein levels of Pom remain unchanged. Only under conditions of low sodium, the abundance of Pom drops. Thus, it appears that the localization/interaction efficiency of one or both stators is directly or indirectly altered in response to changes in the environmental conditions. The drop in abundance of PomAB at low sodium conditions that allows a high level of Mot localization might be a consequence of elevated protein turnover due to decreased localization efficiency. Overexpression of Pom would overcome this type of regulation as has been observed in this study. Thus, regulation of Pom abundance may also be a factor for stator selection in S. oneidensis MR-1.

A remaining question is whether differences in functional localization of distinct stators may solely be attributed to direct stator–FliG interaction efficiency. For V. alginolyticus, it was recently shown that sodium binding to the stators is directly involved in assembly of the stator–motor complex (Fukuoka et al., 2009). However, it cannot be excluded that additional factors are involved. In Vibrio species, two auxiliary proteins, MotX and MotY, are crucial for sodium-dependent swimming (McCarter, 1994a,b). Both proteins are thought to be required for incorporation and stabilization of the PomAB complex at the motor (Okabe et al., 2002; 2005; Terashima et al., 2006). Proteins homologous to MotY are thought be involved in rotation of the polar flagellum in P. aeruginosa PAO1 (Doyle et al., 2004) and to be required for the function of the proton-driven lateral flagellar system of V. parahaemolyticus (Stewart and McCarter, 2003). This demonstrates that a role of such auxiliary proteins is not limited to sodium-powered systems or to bacteria harbouring more than a single flagellar system. Genes sharing homologies to motX and motY were also identified S. oneidensis MR-1. It remains to be shown if MotX and MotY actually play a role in motility in S. oneidensis MR-1, but they represent interesting candidates to mediate stator selection in this organism.

Previous studies have revealed that different bacterial stator subunits can be assembled to form functional hybrid stators. This was found to occur with stator subunits derived from single but also from different species (Asai et al., 1999; Doyle et al., 2004; Ito et al., 2005). Physiological studies on these complexes revealed that the B-subunit determines the coupling ion specificity (Asai et al., 1999; Ito et al., 2005). Thus, assembly of the A- and B-subunits to hybrid stators might also lead to a switch in coupling-ion specificity to contribute to the versatility of the flagellar motor. It has not yet been demonstrated that such hybrid stator complexes actually occur in natural systems. We showed that in S. oneidensis MR-1 such hybrids are non-functional and do not play a role in the flagellar rotation. However, it cannot be excluded that formation of hybrid stators may occur in other species.

A more likely and attractive model is that S. oneidensis MR-1 forms a hybrid motor (as opposed to hybrid stators) consisting of PomAB and MotAB stator units, using protons and sodium ions simultaneously for flagellar rotation. In such a system, the composition of the stator ring complex would be regulated dynamically in response to the environmental sodium content. The findings on the dynamics of the stator ring complex (Leake et al., 2006) strongly support the feasibility of such a dynamic hybrid-motor model, which is currently under investigation of our group. We speculate that flagellar hybrid motors may also be found in other bacteria harbouring sodium- and proton-dependent systems, such as B. subtilis or Desulfovibrio vulgaris (Heidelberg et al., 2004; Ito et al., 2004; Clark et al., 2007).

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Bacterial strains, growth conditions and media

Bacterial strains and plasmids used in this study are summarized in Table S1. E. coli strains were grown in LB medium at 37°C. For strain WM3064, 2,6-diamino-pimelic acid (DAP) was added to a concentration of 300 μM. S. oneidensis strains were grown at 30°C in LB, LM (10 mM HEPES, pH 7.3; 100 mM NaCl; 0.02% yeast extract; 0.01% peptone; 15 mM lactate), or 4 M Minimal Medium (Thormann et al., 2006). When necessary, media were solidified using 1.5% (w/v) agar and/or supplemented with 10 μg ml−1 chloramphenicol, 5 μg ml−1 gentamicin, 25 μg ml−1 kanamycin and/or 10 μg μl−1 tetracycline. For promoter induction from plasmid pBAD33, l-arabinose was added to a concentration of 0.008% (w/v) in liquid media and 0.02% (w/v) on agarose pads.

Strain and vector constructions

DNA manipulations were carried out according to standard protocols or following the manufacturer's instructions. Kits for the isolation of chromosomal DNA, isolation of plasmids and purification of polymerase chain reaction (PCR) products were purchased from Qiagen (Hilden, Germany), Sigma Aldrich GmbH (Taufkirchen, Germany), and HISS Diagnostics GmbH (Freiburg, Germany) respectively. Enzymes were purchased from New England Biolabs (Frankfurt, Germany) and Fermentas (St Leon-Rot, Germany).

In-frame deletion mutants in S. oneidensis MR-1 were constructed essentially as reported earlier, leaving only short terminal sections of the target genes (Thormann et al., 2005; 2006). For that purpose, upstream and downstream fragments of the desired gene region were amplified by PCR using the corresponding primer pairs (Table S2). After purification, the fragments were treated with the appropriate restriction enzyme and subsequently ligated to yield an in-frame fusion product which was then applied as template for a second PCR using the outer primers. The final product was isolated from an agarose gel, digested with the appropriate restriction enzymes, and ligated into the suicide vector pGP704Sac28Km. The resulting plasmid was introduced into S. oneidensis MR-1 by conjugative mating using E. coli WM3064 as a donor on LB medium containing DAP. Single-cross-over integration mutants were selected on LB plates containing kanamycin but lacking DAP. Single colonies were grown overnight in liquid LB without antibiotics and plated on LB containing 10% (w/v) sucrose to select for plasmid excision by double-cross-over events. Kanamycin-sensitive colonies were then checked for the targeted deletion by colony PCR using primers bracketing the location of the deletion, or by Southern analysis.

For complementation of a pomAB mutant, the corresponding gene region was amplified from wild-type chromosomal DNA, treated with BglII and MluI, and ligated into the broad-host range plasmid pME6031.

For the construction of pBBR-MCS5-LucB, lucB was amplified from pLucBF, introducing an optimized ribosome binding site upstream of the gene's start codon. The PCR product was digested with XhoI and KpnI and ligated into pBBR-MCS5. To prevent read-through into the vector, the T1/T2 terminator region was amplified from pBAD33 and subsequently inserted into the KpnI site downstream of the lucB gene. The promoter regions of pomAB and motAB were amplified from S. oneidensis MR-1 chromosomal DNA, and introduced PstI/BamHI sites were then used to ligate the fragment into pBBR-MCS5-LucB. As the exact position of the promoter is unknown, fragments encompassed 500 bp upstream of the pomA or motA start codons.

For the construction of integrated promoter–gfp fusions, an unstable version of gfp (gfp-ASV) was amplified from pJBA113 and cloned into pBC SK+. Translational fusions to either pomB or motB were then created by introducing the promoter regions containing the start codon into the BamHI/SphI site of pBC-gfpASV. After sequencing, the fusions were released by a BamHI/KpnI digest and ligated into the suicide vector pJP5603. The constructs were conjugated into S. oneidensis MR-1 from E. coli WM3064 and kanamycin-resistant colonies were checked for the correct insertion by PCR. For C-terminal tagging of PomB and MotB, respectively, a C-terminal fragment lacking the stop codon of the corresponding gene was amplified by PCR and cloned into the KpnI/MluI sites of pCHYC. Subsequent to sequence verification, the C-terminal mCherry fusion was released by KpnI/BamHI restriction and ligated into the suicide vector pJP5603. The construct was then mated into S. oneidensis MR-1 from E. coli WM3064 and PCR ensured correct insertion by single cross-over.

pBAD33–PomAB was constructed by amplification of the corresponding fragment from chromosomal DNA of S. oneidensis MR-1 and subsequent ligation into the KpnI/XbaI site of pBAD33. To construct pBAB33–motB–mCherry and pBAB33–pomB–mCherry, the corresponding fusion was amplified from S. oneidensis MR-1 motB–mCherry and –pomB–mCherry respectively, and an optimized ribosome binding site was introduced upstream of the start codon. The fragments were digested with KpnI and XbaI and subsequently ligated into pBAD33.

Motility assays

Rapid motility screening was carried out by spotting 3 μl of a liquid culture of the corresponding strain on plates that contained LM medium with an agar concentration of 0.25% (w/v). Although these plates are commonly referred to as ‘swarming plates’, the bacterial movement that was actually scored was flagella-mediated swimming motility. Strains to be directly compared were always spotted on the same plate. In order to enrich for the Δmot up-motile mutant, cell material from the fringe of the observed areas of elevated motility was directly transferred to another swarmer plate without interim cultivation in liquid media.

To assess swimming motility in liquid media, cell material was directly transferred from fresh overnight LB plates to LM medium to yield an OD600 of 0.2–0.4 and incubated for 30 min at room temperature. To quantify the dependence of swimming speed on different sodium concentrations, LM medium with varying concentrations of NaCl was used. KCl was added to yield an overall salt concentration of 250 mM. LM medium to which no NaCl was added was referred to as ‘low sodium’. Resuspended cells were gently sedimented by centrifugation (2000 g, 1 min) and were washed twice in LM without NaCl. After the final centrifugation step the sediment was gently resuspended in LM with the desired concentration of NaCl and 5% (w/v) polyvinylpyrrolidone (PVP-K30) to increase the viscosity and allow more reliable measurements. To assess the effect of CCCP (Sigma) and phenamil sulphonamide (Sigma), the corresponding compound was directly added from a stock solution dissolved in DMSO to yield final concentrations of 10 and 50 μM respectively. Addition of equivalent DMSO concentrations did not negatively affect swimming motility.

Microscopy was performed with an upright Zeiss Image MI (Oberkochen, Germany) equipped with a Cascade 1K camera (Visitron Systems, Puchheim, Germany) and a Zeiss Plan Neofluar 40×/1.4 objective. Ten microlitre aliquots of the culture were spotted on a microscope slide. After adding a coverslip the slide was mounted on the microscopic stage, focused at an area at the centre of the slide and one or two series of images were taken (100 frames at 20 ms each). The whole process took less than 1 min. The series were analysed using the Metamorph software (version 7.1.2.0, Molecular Devices, Ismaning, Germany).

Fluorescence microscopy

Prior to microscopy, cells were immobilized on 1% agarose-LM pads. Microscopy was performed on the system described using a Zeiss Plan Apochromat 100×/1.4 differential interference contrast objective. Image processing was carried out using Metamorph 7.1.2, Adobe Photoshop CS2, and Adobe Illustrator CS2. Quantification of polar fluorescence was carried out using Metamorph 7.1.2 software by determining the fluorescence signal strength of a defined region encompassing only the cell pole. For background subtraction, the average signal strength of 25 areas of the same size randomly distributed across the same image was calculated. When polar localization was determined, at least 250 cells per data point were observed.

Electron microscopy

Cells were allowed to attach to a glow-discharged carbon-coated grid and washed with a drop of distilled water for 1 min. Subsequently, the cells were treated with 2% uranyl acetate for 1 min and blotted to dryness. Images were taken on a JEOL TEM 1230 transmission electron microscope.

Flagella staining

Staining to visualize general flagellation by the silver impregnation method was carried out essentially as described earlier (Blenden and Goldberg, 1965). For colocalization of mCherry-tagged stator subunits with the flagellum, the flagellar filaments were visualized by fluorescence labelling using Alexa Fluor 488 carboxylic acid succinimidyl ester (Invitrogen) essentially as previously described (Turner et al., 2000). A culture of the strain to be analysed was inoculated in LB from an overnight preculture and grown to an OD600 of approximately 0.6 and motility of the cells was checked under the microscope. Subsequently, 20–30 ml of the culture was harvested by centrifugation (2000 g, 10 min), carefully washed two times in 15 ml 4 M medium at room temperature and checked for motility again. The cells were then centrifuged (2000 g, 10 min), and the sediment was carefully resuspended in 0.5 ml 4 M medium, adjusted to a pH of 7.8, and 12.5 μl staining solution (4 M medium pH 7.8 containing 0.08 mg Alexa Fluor 488) was added. The suspension was incubated for 1 h at room temperature in the dark with slow agitation. The cells were then carefully washed three times in 4 M medium (pH 7.8) and finally resuspended in 100 μl 4 M medium. Three microlitres was used for fluorescence microscopy as described. Visualization of the flagella filament was enhanced by using the ‘sharpen’ function of the Metamorph software.

Analysis of transcription levels by measuring luciferase (LucB) activities

Measurement of promoter activities using transcriptional lucB reporter fusions was carried out by a modification of a protocol described earlier (Feustel et al., 2004). Cells corresponding to 1.0–1.5 OD600 were harvested by centrifugation and resuspended in 150 μl 50 mM potassium phosphate buffer (pH of 7.0). Triplicate 45 μl aliquots were then pipetted into wells of a white 96 well polypropylene microtiter plate (Greiner, Germany). All following steps were carried out well by well using an Infinite M200 plate reader equipped with two microinjectors (Tecan, Switzerland). Fifty microlitres of 2× assay buffer (62.5 mM glycyl glycine; 25 mM MgCl2; pH 7.8) and 12.4 μl of ATP solution (100 mM) were added, mixed well, and subsequently 50 μl of d-luciferin solution (330 μM in 10 mM potassium phosphate buffer, pH 6.5) were injected. After mixing for 3 s, luminescence was determined for 5000 ms. Relative light units were obtained by dividing the measured light units by the OD600 of the corresponding culture. All measurements were done in triplicate and repeated at least twice in independent experiments. As a negative control, S. oneidensis MR-1 cells bearing the pBBR-lucB vector without inserted promoter region were used.

Immunoblot analysis

Protein lysates for Western blot analyses to check the stability of PomB–mCherry and MotB–mCherry fusion proteins were obtained from logarithmically growing LB cultures. The lysates for determining the protein levels were from overnight cultures in LM amended with 0, 2, 5, 25, 50 and 100 mM of NaCl, the overall ion concentration was adjusted to 250 mM with KCl. Cells corresponding to an OD600 of 0.25 were harvested by centrifugation, resuspended in 25 μl sample buffer (Laemmli, 1970), heated at 99°C for 5 min, and stored at −20°C. Ten microlitre of the sample was resolved by SDS-PAGE on 12% polyacrylamide gels. Subsequently, proteins were transferred to polyvinylidene difluoride membrane by semidry transfer. For detection of the proteins, polyclonal antibodies were used that were raised against internal peptide fragments of MotB (Q240 to N254; SOmotB-1) and PomB (Q116-T129; PomB-2) (Eurogentec Deutschland GmbH), respectively, in a dilution of 1:250. Secondary anti-rabbit immunoglobulin G-horseradish peroxidase antibody was used at a dilution of 1:20 000, and signals were detected using the Western Lightning Chemoluminescence Reagent (PerkinElmer LAS) followed by exposure to autoradiography film.

Phylogenetic analysis of S. oneidensis MR-1 MotAB and PomAB protein sequences

MotA (SO_4287), MotB (SO_4286), PomA (SO_1529) and PomB (SO_1530) protein sequences were used as queries in separate blast (Altschul et al., 1990) analyses against the NCBI non-redundant database (http://www.ncbi.nlm.nih.gov/) with an expect score cut-off of 1e-10. Protein sequences identified in each blast analysis were collected and aligned using the ClustalW v1.83 (gap open penalty = 10; gap extension penalty = 0.2; Gonnet protein matrix; http://www.ebi.ac.uk/clustalw; Thompson et al., 1994). Alignments were improved by manual curation and were then used to generate phylogenetic trees using the neighbour joining method (Saitou and Nei, 1987), 1000 bootstrapping replications were implemented. Trees were then visualized and manipulated using the Treedyn 198.3 (Chevenet et al., 2006).

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

We thank Michio Homma for sharing unpublished results and Martin Thanbichler for providing the pCHYC vector prior to publication as well as for help with tagging and localization of stator subunits by fluorescence microscopy. We are also grateful to Lotte Søgaard-Andersen and Penelope Higgs for critical and helpful comments on the manuscript. This work was supported by the Max-Planck-Society.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
MMI_6570_sm_Tables_S1-S2_and_Figures_S1-S3.pdf362KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.