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

Elevated levels of the second messenger cyclic dimeric GMP, c-di-GMP, promote transition of bacteria from single motile cells to surface-attached multicellular communities. Here we describe a post-translational mechanism by which c-di-GMP initiates this transition in enteric bacteria. High levels of c-di-GMP induce the counterclockwise bias in Escherichia coli flagellar rotation, which results in smooth swimming. Based on co-immunoprecipitation, two-hybrid and mutational analyses, the E. coli c-di-GMP receptor YcgR binds to the FliG subunit of the flagellum switch complex, and the YcgR–FliG interaction is strengthened by c-di-GMP. The central fragment of FliG binds to YcgR as well as to FliM, suggesting that YcgR–c-di-GMP biases flagellum rotation by altering FliG-FliM interactions. The c-di-GMP-induced smooth swimming promotes trapping of motile bacteria in semi-solid media and attachment of liquid-grown bacteria to solid surfaces, whereas c-di-GMP-dependent mechanisms not involving YcgR further facilitate surface attachment. The YcgR–FliG interaction is conserved in the enteric bacteria, and the N-terminal YcgR/PilZN domain of YcgR is required for this interaction. YcgR joins a growing list of proteins that regulate motility via the FliG subunit of the flagellum switch complex, which suggests that FliG is a common regulatory entryway that operates in parallel with the chemotaxis that utilizes the FliM-entryway.


Introduction

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

Cyclic dimeric GMP, c-di-GMP, has risen to prominence as a ubiquitous bacterial second messenger that controls lifestyle changes. Its role has been studied primarily in the Proteobacteria, where elevated c-di-GMP levels are associated with the transition from a motile planktonic state to a surface-attached state that results in multicellular communities, biofilms (reviewed in Hengge, 2009; Romling and Simm, 2009). However, c-di-GMP also regulates processes beyond this transition, e.g. virulence (Tamayo et al., 2007), cell cycle progression (Duerig et al., 2009) and long-term cell survival (Kumar and Chatterji, 2008). Several kinds of c-di-GMP receptors have recently been identified, including the so-called PilZ domain (Amikam and Galperin, 2006; Ryjenkov et al., 2006). Despite the rapid progress in understanding c-di-GMP-dependent signalling achieved in the last several years, the mechanisms through which c-di-GMP operates remain poorly understood.

In this study we deciphered a mechanism though which c-di-GMP facilitates transition of motile cells of enteric bacteria to surface-attached lifestyle. An early step in transition must involve motility inhibition. The currently known mechanisms that link c-di-GMP to motility control involve transcriptional regulation of flagellar genes (Hickman and Harwood, 2008; Krasteva et al., 2010), which is relatively slow and therefore insufficient to account for the transition. Elevated intracellular c-di-GMP levels are known to inhibit migration of Escherichia coli and Salmonella enterica serovar Typhimurium in semi-solid agar, a phenotype that can be partially reversed by inactivation of the ycgR gene (Ko and Park, 2000; Ryjenkov et al., 2006; Girgis et al., 2007) (Fig. 1). YcgR contains the PilZ domain and binds c-di-GMP (Ryjenkov et al., 2006). In a recent review, Wolfe and Visick (2008) put forth an insightful hypothesis that YcgR may inhibit motility by interfering with the flagellum switch complex. Here, we tested this hypothesis experimentally.

image

Figure 1. Swimming in semi-solid agar of E. coli MG1655 and its ΔyhjH and ΔyhjHΔycgR derivatives. Overnight cultures were concentrated to the same cell density (A600 = 5), and 3 µl was inoculated onto soft agar plates containing 0.25% agar, 1% tryptone and 0.5% NaCl (Ryjenkov et al., 2006). Images were taken after 4 h at 37°C.

Download figure to PowerPoint

The switch complex is an approximately 4-MDa structure at the base of the bacterial flagellum. It consists of three subunits, FliGMN, which are located in the cytoplasm and bound to the FliF protein embedded in the cytoplasmic membrane. The function of the switch complex is threefold: (i) flagellum assembly and integrity, (ii) energy transfer from the proton-motive force driven stator to the rotating flagellum body, and (iii) regulation of the direction of flagellum rotation (reviewed in Berg, 2003; Minamino et al., 2008).

In E. coli and related bacteria, several peritrichous flagella form a bundle. The synchronous counterclockwise (CCW) rotation of flagella in the bundle promotes smooth swimming. A change in rotation direction to clockwise (CW) results in flagellum bundle disassembly and, as a result, the cell tumbles. The ability of cells to reorient themselves is essential for chemotaxis. Cells impaired in switching are non-chemotactic and migrate poorly in semi-solid agar because they get stuck in blind alleyways. The frequency of changes in flagellum rotation direction is regulated by the chemotaxis phosphorelay system in response to attractants and repellents. CheY is the effector protein of this phosphorelay. When phosphorylated, CheY binds to the FliM subunit of the switch complex and increases the probability that the flagellar motor will rotate CW instead of CCW (Berg, 2003; Minamino et al., 2008).

In this work, we show that the c-di-GMP receptor protein YcgR from enteric bacteria binds to the switch complex, primarily via the FliG subunit, and describe how YcgR affects the motile-to-sessile transition.

Results

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

The E. coli c-di-GMP receptor YcgR interacts with the flagellum switch complex

To test the hypothesis that YcgR interacts with the flagellum switch complex, we first assayed for YcgR binding to the switch complex overexpressed in E. coli BL21(DE3), a strain that does not make its own flagella. The FliFGMN proteins are known to form a stable complex anchored in the cytoplasmic membrane by FliF, even in the absence of other flagellum components (Minamino et al., 2008). We overexpressed, in BL21(DE3), the FliFGMN proteins from S. Typhimurium, whose sequences are 89–96% identical to their E. coli counterparts. We passed the soluble crude cell extract of BL21(DE3) overexpressing the FliFGMN complex through the Co2+ column that contained immobilized YcgR–His6. The YcgR–bound proteins were eluted and assayed with the antibodies specific to the individual subunits of the switch complex. We found that three soluble components of the switch complex, FliG, FliM and FliN, were present in the eluate (Fig. 2A). This experiment suggested that YcgR might interact with the switch complex.

image

Figure 2. Interactions between E. coli YcgR and flagellum switch complex proteins. A. Pull-down of the switch complexes overexpressed in BL21(DE3) [pKOT107; pKLR2] by E. coli YcgR::His6 immobilized on the Co2+ resin. The FliGMN proteins were detected by immunobloting using FliG-, FliM- and FliN-specific antibodies (Oosawa et al., 1994). Lanes: 1, eluate from empty beads (no YcgR); 2, eluate from beads containing YcgR::His6. B. Pull-down of the switch complexes from the lysates of E. coli MG1655. Lanes: 1, empty beads; 2, YcgR::His6; 3, YcgR (118R[RIGHTWARDS ARROW]D)::His6; 4, YcgR (147S[RIGHTWARDS ARROW]A)::His6. C. Two-hybrid analysis of the interaction between YcgR and switch complex proteins. Representative plates showing growth of the two-hybrid validation strain Xl-1 Blue MRF′ that harbours plasmid pTRG expressing YcgR and plasmid pBT expressing FliG, FliM or FliN. FliG: 1, Empty vectors (negative control); 2, FliG + YcgR; 3, FliG only; 4, pBT-LGF2 + pTRG-Gal11 (positive control); 5, YcgR only. The FliM and FliN plates have the same layout as the FliG plate. D. Two-hybrid analysis of the interactions between YcgR and FliG and FliM in Xl-1 Blue MRF′ (sectors 1, 5, 6, 8, 9) and Xl-1 Blue MRF′ΔyhjH (sectors 2, 3, 4, 5, 7, 10). 1 and 2, FliG + YcgR; 3, FliN + YcgR; 4, FliG + YcgR 118R[RIGHTWARDS ARROW]D; 5, FliG + YcgR 147S[RIGHTWARDS ARROW]A; 6 and 7, FliM + YcgR; 8, FliM + YcgR 118R[RIGHTWARDS ARROW]D; 9 and 10, FliM + YcgR 147S[RIGHTWARDS ARROW]A.

Download figure to PowerPoint

To exclude the possibility that YcgR binding to the switch complex resulted from switch complex overabundance or from non-specific binding of E. coli YcgR to the heterologous, S. Typhimurium, proteins, we tested for YcgR binding to the switch complex using cell extract from the flagellated and motile strain E. coli MG1655. Again, we detected each of the three soluble components of the switch complex in the eluate (Fig. 2B), which confirms that YcgR can interact with the flagellum switch complex.

To test whether YcgR binding to the switch complex depended on c-di-GMP, we measured the amount of c-di-GMP present in the Co2+ resin-bound YcgR–His6 following extensive column washes but prior to elution, and found that c-di-GMP amount was very low, < 10% of the c-di-GMP amount required for YcgR saturation (data not shown). This observation suggests that YcgR binds to the switch complex independently of c-di-GMP. To explore this further, we tested binding to the switch complex of two YcgR mutants, 118R[RIGHTWARDS ARROW]D and 147S[RIGHTWARDS ARROW]A. We showed earlier (Ryjenkov et al., 2006) that the 118R[RIGHTWARDS ARROW]D mutant does not bind c-di-GMP in vitro and is unable to complement the ycgR mutation in vivo, whereas the 147S[RIGHTWARDS ARROW]A mutant binds c-di-GMP with somewhat higher affinity and inhibits motility in vivo better than the wild-type YcgR. Both 118R[RIGHTWARDS ARROW]D and 147S[RIGHTWARDS ARROW]A mutants were able to bind to the switch complex thus confirming the notion that YcgR binding does not require c-di-GMP (Fig. 2A and B).

FliG is the primary interacting partner of YcgR

To investigate which of the complex components, FliG, FliM or FliN, is the primary YcgR binding partner, we used the bacterial two-hybrid BacterioMatch II system (Stratagene). In this system, the YcgR protein was fused to the α-subunit of the E. coli RNA Pol, while the three proteins of the switch complex were individually fused to the phage λ CI repressor protein. An interaction between the α-subunit of RNA Pol and the λ CI repressor activates transcription of the yeast HIS3 reporter gene, which allows cell growth in the presence of a competitive inhibitor of the HIS3 enzyme. We chose this two-hybrid system because we could manipulate c-di-GMP levels in the E. coli (but not yeast) host, and because the system components have never been implicated as targets for c-di-GMP control. Using this system, we detected the interaction of YcgR with FliG but not with FliM or FliN (Fig. 2C). When we switched vectors expressing the bait and prey proteins, we observed the same interaction pattern (data not shown), which verifies that the observed protein–protein interactions are specific.

The YcgR–FliG interaction is strengthened by c-di-GMP

YcgR is activated upon binding to c-di-GMP (Ryjenkov et al., 2006). It is unclear whether the intracellular c-di-GMP pool in the E. coli host used for two-hybrid analysis, Xl-1 Blue MRF′, is sufficient to saturate the plasmid-expressed YcgR proteins. To ascertain the role of c-di-GMP in YcgR–FliG interactions, we replaced the wild-type YcgR in the two-hybrid analysis with the 147S[RIGHTWARDS ARROW]A and 118R[RIGHTWARDS ARROW]D mutants. We found that the 147S[RIGHTWARDS ARROW]A mutant, which has higher affinity to c-di-GMP compared with the wild-type YcgR, showed somewhat stronger interactions with FliG (Fig. 2D, sectors 5 and 1). The 118R[RIGHTWARDS ARROW]D mutant, incapable of c-di-GMP binding, failed to interact with FliG (not shown). These observations suggest that c-di-GMP is either required for YcgR–FliG interactions or strengthens these interactions.

To further explore the role of c-di-GMP, we used genetic means to manipulate intracellular c-di-GMP levels in the two-hybrid system host. YhjH is the major c-di-GMP phosphodiesterase that controls c-di-GMP pool available to YcgR. Inactivation of yhjH inhibits cell migration in semi-solid agar, while simultaneous inactivation of yhjH and ycgR partially rescues this phenotype (Ko and Park, 2000; Ryjenkov et al., 2006; Girgis et al., 2007) (Fig. 1). To elevate c-di-GMP levels in Xl-1 Blue MRF′, we deleted the yhjH gene. We observed that the YcgR–FliG interactions in the ΔyhjH host were strengthened, compared with the interaction observed in the wild-type strain (Fig. 2D, sectors 1 and 2). In the ΔyhjH host, no further improvement in the FliG-YcgR interaction strength was observed for the 147S[RIGHTWARDS ARROW]A mutant, compared with the wild-type host strain (not shown), while the 118R[RIGHTWARDS ARROW]D mutant still showed no interactions (Fig. 2D, sector 4). We observed no interactions between YcgR and FliM or FliN in the ΔyhjH host (Fig. 2D, sectors 7 and 3).

We also tested for the YcgR–FliG interactions in the strain Xl-1 Blue MRF′ strain where c-di-GMP levels were lowered by overexpression of the c-di-GMP phosphodiesterase YahA characterized by us earlier (Schmidt et al., 2005). YahA overexpression weakened but did not abolish YcgR–FliG interactions (data not shown). The results of the two-hybrid analysis, when taken together, show that (i) FliG is the primary target of YcgR, (ii) YcgR interacts with FliG in the absence of c-di-GMP, and (iii) these interactions are strengthened by c-di-GMP.

Interestingly, we found that the YcgR 118R[RIGHTWARDS ARROW]D mutant, which is impaired in c-di-GMP binding in vitro, showed interactions with the FliM subunit (Fig. 2D, sector 8). Furthermore, weak interactions were observed between YcgR 147S[RIGHTWARDS ARROW]A and FliM (Fig. 2D, sector 10). It is therefore possible that the c-di-GMP-free form of wild-type YcgR, which likely resembles the YcgR 118R[RIGHTWARDS ARROW]D mutant, has some affinity to FliM, which is below the detection level of the two-hybrid system. It is also feasible that c-di-GMP binding promotes YcgR contacts with the FliM subunit, a possibility entirely consistent with the analysis of the FliG-YcgR interacting surfaces presented below. Additional experimentation will be required to arrive at a detailed picture of molecular interactions between YcgR and switch complex components.

Localization of the FliG regions involved in the YcgR–FliG interactions

We also used the two-hybrid system to investigate what regions of FliG are involved in the YcgR–FliG interaction (Fig. 3A). Our choice of fragments was guided by the X-ray structure of the Thermotoga maritima FliG protein lacking the N-terminus (Fig. 3B). This protein, which is 83% identical to the E. coli FliG, has been widely used as a FliG structural model (Brown et al., 2007). We found that two regions of FliG interacted with YcgR, i.e. the N-terminus and the region between residues 190 and 296 (Fig. 3B). To verify that two different interacting surfaces on FliG are involved in YcgR interactions, we constructed glutathione S-transferase, GST, fusions to the N- (aa 1–103) and C-terminal (aa 106–331) parts of FliG and tested the ability of the purified fusion proteins to bind to YcgR in vitro. Both GST–FliG fragment fusions showed binding to YcgR in the pull-down experiment, whereas GST alone did not (Fig. 3C). Therefore, it is likely that YcgR can bind to two distinct regions of FliG.

image

Figure 3. Mapping of the FliG regions interacting with YcgR using E. coli two-hybrid system. A. Schematic representation of the FliG fragments used in two-hybrid analysis and assessment of growth of Xl-1 Blue MRF′ΔyhjH expressing YcgR (from pT-ycgR) and various FliG fragments [from plasmids pB-G series (Table 2)]. Growth: +++, strong; ++, moderate; +, weak or non-existent. B. Model of E. coli FliG based on the T. maritima FliG structure (pdb: 1LKV). Region interacting with YcgR are shown in light-grey. Predicted positions of selected E. coli FliG residues are indicated. C. Pull-down assay showing interaction between YcgR and the N-terminal, FliG-N, and C-terminal, FliG-C, fragments of FliG expressed as GST fusions. Solution of the YcgR::His6 protein was added to the glutathione resin-bound GST or GST–FliG fusions, and extensively washed. Eluted proteins were detected by Coomassie blue staining (Sambrook et al., 1989). Lanes: 1, GST + YcgR; 2, GST::FliG-C + YcgR; 3, GST::FliG-N + YcgR.

Download figure to PowerPoint

The N-terminus of FliG is believed to be buried within the fully assembled switch complex. If it were accessible, then YcgR binding would be expected to impair flagellum assembly or stability. The second region through which FliG interacts with YcgR encompasses the hydrophobic patch, aa 196–214, through which FliG binds to FliM. If YcgR were to interfere with FliG-FliM binding, this would be expected to affect flagellum rotation behaviour. Either one of the observed interactions could potentially affect energy transfer to the flagellum and therefore decrease rotation velocity. We tested these possibilities in the experiments described below.

YcgR–c-di-GMP binding to the switch complex biases flagellum rotation in the CCW direction

MG1655 and its ΔyhjH and ΔyhjHΔycgR derivatives proved to be equally motile when observed under the microscope, which indicates that YcgR is unlikely to significantly impair energy transfer to the flagellum rotor. To assess the effect of YcgR on flagella number and stability, we compared levels of the flagellin subunit FliC in the three strains and found no differences (not shown), which indicates that YcgR is unlikely to significantly affect flagella number or stability.

We then performed tethering assays, where E. coli cells were bound to a glass surface by a single flagellum, so that flagellar rotation could be measured by counter-rotation of the cell body (Silverman and Simon, 1974; Slocum and Parkinson, 1985). While we did not observe significant differences in rotation speed of the tethered cells, we observed striking differences in rotation behaviour. The wild-type cells rotated primarily in the CCW direction and reversed rotation direction to CW with a frequency of approximately 36.0 min−1, whereas the ΔyhjH cells switched much more rarely, approximately 10.8 min−1 (Table 1). This resulted in the ΔyhjH cells spending only 3.1% time in the CW rotation, compared with the 11.8% time for the wild-type. Importantly, the ycgR deletion in the ΔyhjH background fully reversed the CCW rotation bias of the ΔyhjH cells (Table 1), which suggests that YcgR binding to the switch complex increases the CCW bias. Supporting Information contains movies of typical tethered cells from the three genetic backgrounds. Based on these observations we conclude that the primary role of YcgR is to induce CCW bias in flagellum rotation in response to increased c-di-GMP levels, and thus result in smooth swimming.

Table 1.  Effects of the ΔyhjH and ΔyhjHΔycgR mutations on flagellum rotation behaviour.
StrainReversal frequency (min−1)Time in CW direction (%)
  • The results are averaged from 14–20 tethered cells per strain.

  • a.

    Values that are significantly different form those of the wide-type (P < 0.05; unpaired tow-tailed Student's t-test).

MG165536.0 ± 8.011.8 ± 2.9
MG1655ΔyhjH10.8 ± 5.8a3.1 ± 2.0a
MG1655 ΔyhjHΔycgR30.4 ± 10.611.0 ± 3.2

YcgR promotes motile-to-sessile transition in E. coli

Wolfe and Berg (1989) showed that smooth-swimming E. coli get trapped in semi-solid media because cells are unable to escape by reversing the swimming direction (Fig. 1). This mechanism might explain how YcgR promotes transition of motile cells to sessile lifestyle in semi-solid media, which likely resembles epithelial layers of animal digestive tracks, the natural environment of E. coli.

We were curious how YcgR-mediated CCW bias would affect surface attachment of cells grown in liquid. To shed light on this issue, we compared initial attachment of the wild-type, ΔyhjH and ΔyhjHΔycgR cells to glass coverslips submerged in bacterial cultures. We observed that, after a short, 2 h incubation, significantly (1.35-fold) higher numbers of the ΔyhjH mutant cells have attached to coverslips, compared with the wild-type (Fig. 4). Cells of the ΔyhjHΔycgR double mutant had attached to the coverslips somewhat worse than the ΔyhjH mutant cells (Fig. 4) but still better (1.20-fold) than those of the wild-type. These results suggest that surface attachment of liquid-grown cells is partially promoted via YcgR and partially via YcgR-independent, c-di-GMP-dependent mechanisms. The smooth-swimming cells may get to the surface faster, overcome repulsion at the medium–surface interface more readily than the wild-type cells and/or have lesser chances of migrating away from the surface. At the surface, the initial and subsequent permanent surface attachment is expected to be facilitated by c-di-GMP-dependent pili and/or exopolysaccharides (Pratt and Kolter, 1998). In conclusion, YcgR appears to promote the motile-to-sessile switch in E. coli in both semi-solid and liquid media.

image

Figure 4. Attachment of liquid-grown cells to glass coverslips. Shown are average numbers of attached cells per field of view from three independent experiments using ten randomly chosen fields per coverslip in each experiment. The asterisk ‘*’ indicates significant difference (P < 0.05).

Download figure to PowerPoint

The YcgR–FliG interaction is conserved

To investigate whether the YcgR–FliG interaction is conserved in other bacteria, we tested for the interaction between the FliG and YcgR homologues from the motile human enteropathogen Yersinia pseudotuberculosis PB1/+. We chose this bacterium for two reasons. One, the Y. pseudotuberculosis YcgR is only 41% identical to the E. coli YcgR, which is at the lower end of sequence identities among the YcgR homologues. Second, the genome of Y. pseudotuberculosis encodes two distinct sets of flagella, one of which is similar to the E. coli flagella, while the other set is similar to the lateral flagella involved in swarming on surfaces (Merino et al., 2006). The presence of two flagella sets allowed us to investigate whether YcgR controls one or both flagella types. The Y. pseudotuberculosis FliG homologue, FliG1 (YPTS_1831), predicted to be part of the swimming flagellum, is 83% identical in primary sequence to the E. coli FliG, while FliG2 (YPTS_3483), predicted to be part of the swarming flagellum, is 31% identical.

Using the two-hybrid system described above, we detected interactions between Y. pseudotuberculosis YcgR and FliG1, but not FliG2 (Fig. 5A). This result suggests that, despite only moderate sequence similarity, the YcgR proteins recognize their FliG partners belonging to flagella involved in swimming but not in surface swarming. We observed no interactions between Y. pseudotuberculosis YcgR and FliM1 or FliM2 (Fig. 5B), which is consistent with the observation that FliG is the primary interacting partner of YcgR in E. coli.

image

Figure 5. Two-hybrid analysis of interactions between YcgR and flagellar switch complex proteins from Y. pseudotuberculosis. FliG1 and FliM1 belong to the E. coli-like flagellum, and FliG2 and FliM2 belong to the lateral flagella involved in surface swarming. A. YcgR–FliG interactions. 1, YcgR only; 2, FliG2 + YcgR; 3, FliG1 + YcgR; 4, FliG2 only; 5, FliG1 only; 6, pBT-LGF2 + pTRG-Gal11 (positive control); 7, empty vectors (negative control). B. YcgR–FliM interactions. 1, Empty vectors (negative control); 2, FliM1 + YcgR; 3, YcgR only; 4, FliM2 + YcgR; 5, pBT-LGF2 + pTRG-Gal11 (positive control).

Download figure to PowerPoint

The N-terminal domain of YcgR is important for the specificity of YcgR–FliG interactions

YcgR homologues are present in representatives of the Gamma- and Betaproteobacteria. All YcgR homologues contain the N-terminal YcgR domain [Pfam: PF07317 (Finn et al., 2008)], also known as PilZN (Amikam and Galperin, 2006), which is linked to the C-terminal PilZ domain. The loop between YcgR/PilZN and PilZ constitutes the primary c-di-GMP binding site (Ryjenkov et al., 2006; Benach et al., 2007). We noticed that two-domain proteins containing C-terminal PilZ domains are distributed much more widely than YcgR proteins. We wondered how important the YcgR/PilZN domain is for specificity of YcgR–FliG interactions.

To investigate whether the non-YcgR two-domain PilZ proteins are involved in motility control via binding to flagellar switch complexes, we tested two proteins, TM0905 from T. maritima and PlzD (VCA0042) from Vibrio cholerae. Using the two-hybrid system described above, we observed no interactions between TM0905 and T. maritima FliG or FliM. Neither did we observe interactions between PlzD and V. cholerae FliG or FliM (data not shown). The latter result is consistent with the physiological observations that, while under certain conditions PlzD modestly inhibits cell motility, its effect is c-di-GMP-independent (Pratt et al., 2007). These observations suggest that the YcgR/PilZN domain is an important specificity determinant for YcgR–FliG interactions, and that PilZ domain proteins lacking this domain do not act on flagellar switch complexes. Therefore, the YcgR-based regulatory mechanism appears to be limited to the representatives of the Gamma- and Betaproteobacteria that contain YcgR homologues.

Discussion

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

The second messenger c-di-GMP has emerged as a key player in the transition of bacteria from single-cell motile state to surface-attached multicellular state. This transition involves inhibition of cell motility and surface attachment. C-di-GMP is known to exert transcriptional control over expression of flagellar genes, however the exact mechanisms remain to be discovered (Hickman and Harwood, 2008; Hengge, 2009; Krasteva et al., 2010). Here we described a new mechanism through which c-di-GMP affects flagellum-based motility at a post-translational level. Specifically, we showed that the E. coli PilZ domain c-di-GMP receptor YcgR (Ryjenkov et al., 2006) interacts with the flagellum switch complex, primarily or exclusively, with the FliG subunit. YcgR binds to the switch complex even in the absence of c-di-GMP, which allows it to monitor c-di-GMP levels at the site of flagellum, and to respond almost instantly to a burst in c-di-GMP synthesis. The YcgR–FliG interaction is conserved beyond E. coli and requires N-terminal YcgR/PilZN domains. C-di-GMP binding is known to result in large conformational changes that bring the YcgR/PilZN and PilZ domains into closer proximity (Ryjenkov et al., 2006; Benach et al., 2007; Ko et al., 2010). Apparently, these conformational changes strengthen YcgR–FliG interactions (Fig. 6).

image

Figure 6. Two signalling pathways controlling flagellum rotation in bacteria. The emerging pathway operating via FliG is depicted on the left. H-NS, FRD and YcgR from E. coli and EspE from B. subtilis use the FliG-entryway. The chemotaxis signalling pathway operating via the FliM entryway is depicted on the right. Methyl-accepting chemotaxis proteins (MCP) initiate the signalling cascade to affect phosphorylation state of output protein CheY. Phosphorylated CheY (CheY∼P) interacts with FliM.

Download figure to PowerPoint

We considered several possibilities concerning the mechanism through which the YcgR–FliG interactions might affect flagella. We did not detect differences in the amount of flagellin made or in the rotation velocity of flagella from the wild-type, ΔyhjH or ΔyhjHΔycgR strains, which suggests that flagella stability and motor efficiency are unlikely to be significantly affected. Instead, we observed that flagella of the ΔyhjH mutant acquired a strong CCW bias, a phenomenon observed earlier by Girgis et al. (2007). Importantly, we showed here that the ycgR deletion is sufficient to fully restore the wild-type ratio of CCW/CW rotation frequencies, which points to YcgR as the sole mediator of the CCW bias. Note, however, that YcgR is not solely responsible for c-di-GMP-dependent inhibition of motility in semi-solid media (Fig. 1).

We identified two regions of FliG that interact with YcgR, one of which overlaps in the site of FliG-FliM binding. Therefore, it is likely that perturbation in the FliG-FliM subunit organization causes the YcgR-induced CCW rotation bias (Fig. 6). The weak protein–protein interactions between YcgR and FliM that we observed in some mutant backgrounds (Fig. 3D) may reflect competition of YcgR and FliM for the same binding region in FliG. Additional studies will be required to arrive at a detailed understanding of molecular interactions between YcgR and switch complex components.

Interestingly, YcgR is not unique in its ability to bind to the FliG subunit of the switch complex. The first reported FliG-binding protein, which does not represent a structural flagellum component, was H-NS, the histone-like DNA-binding protein (Marykwas et al., 1996). Yet another protein, fumarate reductase (FRD), of E. coli has recently been shown to bind FliG and affect motility in a fumarate-dependent manner (Cohen-Ben-Lulu et al., 2008). FRD appears to be required for normal flagellum function because the frd gene deletion results in the cells with a decreased number of flagella, and the remaining flagella have a strong CCW rotation bias. Ability to bind to FliG appears to be independent of the enzymatic activity of FRD (Cohen-Ben-Lulu et al., 2008).

Furthermore, the putative glycosyl transferase, EspE, from Bacillus subtilis has been shown to bind to FliG (Blair et al., 2008). EspE is encoded by one of the genes of the exopolysaccharide biosynthesis gene cluster involved in biofilm formation. Therefore, the physiological function of B. subtilis EspE is most closely related to that of the proteobacterial YcgR. However, the mechanism through which EspE works is different from that of YcgR. EspE binds to the C-terminus of FliG, thus disengaging energy transfer to the flagellum rotor. Consequently, EspE has been designated a molecular clutch (Blair et al., 2008). It is peculiar that the proteobacteria evolved a mechanism distinct from the clutch, essentially for the same purpose, i.e. motile-to-sessile transition. We wondered whether different mechanisms used by the proteobacteria and Bacillus reflect different physiological stages of the transition where EspE and YcgR function.

The EspE molecular clutch is believed to ensure that in a B. subtilis biofilm, flagella stay paralyzed but can be quickly re-energized upon removal or inactivation of EspE. E. coli, on the other hand, appear to engage the YcgR–c-di-GMP mechanism at an early step in the motile-to-sessile transition, when rotating flagella are needed. It is known that the E. coli cells with paralyzed flagella attach to abiotic surfaces significantly worse than the cells containing normally functional or CCW-biased flagella (Pratt and Kolter, 1998). Our data corroborate these observations and show that YcgR is instrumental in surface attachment of liquid-grown cells (Fig. 4). The role of YcgR appears to be even more pronounced in the motile-to-sessile transition of cells in semi-solid media (Fig. 1), which more adequately represent natural environments of enteric bacteria, e.g. epithelial layers of the digestive tracts of eukaryotic hosts and fecal matter.

The results with EspE, H-NS, FRD and now YcgR, all strongly indicate that FliG acts as a common entryway through which flagellar motility is regulated (Fig. 6). This entryway is involved in the motile-to-sessile transition in such phylogenetically diverse branches of bacteria as Proteobacteria and Firmicutes (to which Bacillus belongs). The FliG-entryway appears to work in parallel with the chemotaxis signalling pathway that operates via FliM. However, from the mechanistic standpoint, the FliG-mediated flagella control would be predicted to be dominant over the FliM-mediated chemotaxis control. The YcgR-mediated CCW rotation bias is indeed dominant over chemotaxis in E. coli (Fig. 1), and the EspE-mediated flagella paralysis in B. subtilis is likely dominant. From the physiological standpoint, it is reasonable that signals controlling drastic lifestyle change, i.e. motile-to-sessile transition, would override chemotaxis signalling that might otherwise detract cells from surface attachment (E. coli) or from staying in a biofilm (B. subtilis).

What additional proteins, beyond EspE and YcgR, control the motile-to-sessile transition at the flagellum switch complex in bacteria? Is there a preference for environmental signals that affect motility via FliG versus FliM? Can the FliG-pathway account for a number of unexplained phenomena in bacterial chemotaxis (reviewed in Eisenbach, 2007)? Answers to these questions will undoubtedly shed new light on the mechanisms of bacterial adaptation, including drastic lifestyle changes.

While this manuscript was under review, two new studies (Boehm et al., 2010; Paul et al., 2010) have been published that analysed the YcgR-mediated mechanism of flagellar control. Our results are most consistent with those reported by Paul et al. (2010), who observed that YcgR interacts with the flagellar switch complex and induces CCW bias.

Experimental procedures

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

Strains, plasmids, growth conditions and molecular genetic techniques

The bacterial strains and plasmids used in this study are listed in Table 2 and the primers are listed in Table S1. E. coli was grown in Luria–Bertani (LB) medium (Sambrook et al., 1989), unless specified otherwise. E. coli Xl-1 Blue MRF′, the host strain for BacterioMatch II two-hybrid analysis, and its derivatives were grown according to the manufacturer's specifications (Stratagene). The yhjH and ycgR mutations in strains Xl-1 Blue MRF′ and MG1655 were constructed using the Datsenko and Wanner method (Datsenko and Wanner 2000). Site-directed mutagenesis was performed using a QuickChange kit (Qiagen).

Table 2.  Strains and plasmids used in this study.
Strain or plasmidRelevant genotype or descriptionSource/reference
  • a.

    American Type Culture Collection.

Strains  
 Escherichia coli  
  MG1655‘Wild-type’ATCC 700926a
  MG1655 ΔyhjHΔyhjHThis study
  MG1655 ΔyhjHΔycgRΔyhjHΔycgRThis study
  XL1-Blue MRF′Host for two-hybrid analysis Δ(mcrA)183Δ(mcrCB-hsdSMR-mrr)173 endA1 hisB supE44 thi-1 recA1 gyrA96 relA1 lac[F laqIq HIS3 aadA KmR]Stratagene
  XL1-Blue MRF′ΔyhjHΔyhjHThis study
  BL21(DE3) [pLysS]Strain used for overexpression of switch complexesInvitrogen
Plasmids  
 pKOT107pET3::FliFG (from S. Typhimurium)Oosawa et al. (1994)
 pKLR2pET3::FliMN (from S. Typhimurium)Oosawa et al. (1994)
 pET23aVector for T7-inducible expressionEMD Biosciences
 pEZ-YcgRpET23a::ycgRRyjenkov et al. (2006)
 pEZ-YcgR(118R[RIGHTWARDS ARROW]D)pET23a::ycgR118R[RIGHTWARDS ARROW]DRyjenkov et al. (2006)
 pEZ-YcgR(147S[RIGHTWARDS ARROW]A)pET23a::ycgR147S[RIGHTWARDS ARROW]ARyjenkov et al. (2006)
 pTRGVector for two-hybrid analysis, RNA Polα, TcRStratagene
 pBTVector for two-hybrid analysis, λCI, CmRStratagene
 pTRG-Gal11Positive control for two-hybrid protein interactionsStratagene
 pBT-LGF2Positive control for two-hybrid protein interactionsStratagene
 pT-YcgRpTRG::ycgRThis study
 pT-YcgR118R[RIGHTWARDS ARROW]DpTRG::ycgR118R[RIGHTWARDS ARROW]DThis study
 pT-YcgR147S[RIGHTWARDS ARROW]ApTRG::ycgR147S[RIGHTWARDS ARROW]AThis study
 pT-FliGpTRG::fliGThis study
 pT-YcgR(Yp)pTRG::YcgR from Y. pseudotuberculosisThis study
 pT-TM0905(Tm)pBT::TM0905 from T. maritimaThis study
 pT-PlzD(Vc)pBT::PlzD from V. choleraeThis study
 pB-FliGpBT::FliGThis study
 pB-FliMpBT::FliMThis study
 pB-FliNpBT::FliNThis study
 pB-YcgRpBT::YcgRThis study
 pB-FliG1(Yp)pTRG::FliG1 from Y. pseudotuberculosisThis study
 pB-FliG2(Yp)pTRG::FliG2 from Y. pseudotuberculosisThis study
 pB-FliM1(Yp)pTRG::FliM1 from Y. pseudotuberculosisThis study
 pB-FliM2(Yp)pTRG::FliM2 from Y. pseudotuberculosisThis study
 pB-FliM(Tm)pBT::FliM from T. maritimaThis study
 pB-FliG(Tm)pBT::FliG from T. maritimaThis study
 pB-FliM(Vc)pBT::FliM from V. choleraeThis study
 pB-FliG(Vc)pBT::FliG from V. choleraeThis study
 pB-G1pBT::FliG (aa 1–199)This study
 pB-G2pBT::FliG (aa 173–331)This study
 pB-G3pBT::FliG (aa 99–331)This study
 pB-G4pBT::FliG (aa 99–278)This study
 pB-G5pBT::FliG (aa 99–199)This study
 pB-G6pBT::FliG (aa 1–155)This study
 pB-G7pBT::FliG (aa 106–199)This study
 pB-G8pBT::FliG (aa 106–296)This study
 pGEX-2TKVector for overexpression of GST fusionsPharmacia
 pG-FliGNGST::FliG (aa 1–103)This study
 pG-FliGCGST::FliG (aa 106–331)This study

Bacterial two-hybrid assays

Bacterial two-hybrid assays were carried out using the BacterioMatch II system (Stratagene). The fliG, fliM and fliN genes were individually cloned in vector pBT and ycgR was cloned in pTRG to create in-frame fusions to the genes encoding the λ CI repressor and the α-subunit of E. coli RNA Pol respectively. The order of vectors was also reversed. The pairs of bait and prey plasmids were cotransformed into E. coli Xl-1 Blue MRF′ harbouring HIS3-aadA as a reporter, or the constructed ΔyhjH derivative of Xl-1 Blue MRF′. Plasmids pBT-LGF2 and pTRG-Gal11p from the BacterioMatch II kit were used as a positive control, while empty vectors, pBT and pTRG, were used as a negative control. Cotransformants were selected on permissive LB medium containing 12.5 (µg tetracycline) ml−1, 25 (µg chloramphenicol) ml−1 and subsequently streaked on the screening, M9 medium medium (Sambrook et al., 1989) containing 5 mM 3-amino-1,2,4-triazol (3-AT), a competitive inhibitor of the HIS3 enzyme.

Rotational behaviour of E. coli tethered cells

Tethering experiments were performed at room temperature, essentially as described earlier (Silverman and Simon, 1974; Slocum and Parkinson, 1985). Prior to tethering, cells of MG1655 and its derivatives were passed through a 26-gauge needle to shear cell flagella. Cells were attached to glass coverslips in the tethering buffer [10 mM potassium phosphate (pH 7.0), 0.1 M NaCl, 0.1 mM EDTA, 10 µM l-methionine, 20 mM sodium l-lactate]. The coverslips were inverted to a drop of tethering buffer on a microscope slide to which two additional coverslips had been affixed, forming an approximately 1 mm deep channel. The direction of cell rotation was recorded for 40 s. All images were captured on a Nikon E800 microscope equipped with a Hamamatsu CCD camera and processed using Image Pro-Plus 6.21 Software (MediaCybernetics, Silver Spring, MD, USA).

Cell attachment assays

Cell attachment to glass coverslips submerged in bacterial suspensions was performed essentially as described earlier (Agladze et al., 2003), except no poly-lysine coating was used. Briefly, overnight cultures were inoculated (1:100) into colonization factor antigen medium (Agladze et al., 2003) and grown at 30°C until the A600 reached 1.0. These cultures were exposed to glass coverslips for 2 h under gentle shaking. The coverslips were washed by 3× dipping into 1 l phosphate buffer saline (Sambrook et al., 1989). Micrographs of the attached cells were taken. Reported are average numbers of attached cells from three independent experiments using ten randomly chosen fields per coverslip in each experiment.

Flagellar switch complex pull-down assays

The YcgR::His6 protein and its mutant derivatives were purified as described previously (Ryjenkov et al., 2006). YcgR::His6 was immobilized on the Co2+ resin (Pierce). Plasmids for overexpression of the switch complex components, pKOT107 (FliG and FliF) and pKLR2 (FliM and FliN), were kindly provided by Michael Eisenbach and Shahid Khan (Oosawa et al., 1994). Conditions for obtaining cell extracts containing intact switch complexes were described earlier (Sagi et al., 2003). Briefly, strain BL21(DE3) was grown at 37°C in LB medium to an A600 of 0.3–0.4 and induced with 1 mM IPTG for 3 h. Cells were collected by centrifugation and resuspended in the binding buffer [100 mM NaCl, 25 mM Tris, pH 7.5, 10% glycerol, 20 mM imidazole, 1 mM protease inhibitor cocktail (Sigma)], and passed through a French pressure cell. The crude protein extracts were centrifuged at 30 000 g for 30 min.

The soluble fraction was incubated with the Co2+ resin containing immobilized YcgR::His6 at 4°C for 4 h. Following incubation, the resin was extensively washed with the same buffer containing 50 mM imidazole. The bound proteins were subsequently eluted with 500 mM imidazole. The cell extracts were separated using a 15% SDS-PAGE followed by immunoblot analysis with the rabbit antibodies against individual subunits of the switch complex [kindly provided by Tohru Minamino (Kihara et al., 1996)]. Proteins retained by the empty Co2+ resin were used as a negative control.

GST–FliG pull-down assays

The N-terminal and C-terminal fragments of FliG were expressed as GST fusions in vector pGEX-2TK (GE Healthcare Life Sciences). Recombinant proteins were purified according to manufacturer specifications and dialysed. Ten micromoles of purified proteins were immobilized on GST beads (Clontech Laboratories, CA, USA) in the GST binding buffer (25 mM Tris pH 7.5, 150 mM NaCl, 1 mM EDTA), washed with the pull-down buffer (300 mM NaCl, 50 mM NaH2PO4, 10% Glycerol, pH 7.4) and incubated with 50 micromoles of YcgR–His6 for 2 h at room temperature under gentle shaking. GST protein was used as a negative control. The beads were subsequently washed three times and protein complexes were eluted from GST beads by 20 mM reduced glutathione in 50 mM Tris buffer (pH 8.0). Proteins were boiled for 5 min with SDS sample buffer and resolved using a 15% SDS-PAGE.

Acknowledgements

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

We are grateful to Alan Wolfe and Karen Visick for the inspirational ideas, Tohru Minamino for antibodies, Michael Eisenbach and Shahid Khan for the switch complex expression plasmids, Robert Perry and Alex Bobrov for the Y. pseudotuberculosis genomic DNA, Fitnat Yildiz for the V. cholerae DNA, Daniel Wall for microscope access, Oleg Moskvin for help with photography, Daniel Wall and Alan Wolfe for critical suggestions on the manuscript, and Kurt Miller for manuscript proofreading. We thank Daniel Kearns, David Blair and Rasika Harshey for communicating their unpublished results. DNA sequencing was performed at the University of Wyoming NAEF Core facility. This work was supported by the National Science Foundation (MCB 0645876) and in part by United States Department of Agriculture Cooperative State Research Education Extension Service Grant (Project WYO-414).

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_7179_sm_Movie1.avi5813KSupporting info item
MMI_7179_sm_Movie2.avi8114KSupporting info item
MMI_7179_sm_Movie3.avi5711KSupporting info item
MMI_7179_sm_TableS1.pdf102KSupporting info item
MMI_7179_sm_Legends.pdf23KSupporting 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.