The genes involved in flagellum synthesis, motility and chemotaxis in Escherichia coli are expressed in a hierarchical fashion. At the top of the hierarchy lies the master regulator FlhDC, required for the expression of the whole set of genes. The operon flhDC is controlled by numerous regulators including H-NS, CRP, EnvZ/OmpR, QseBC and LrhA. In the present work, we report that the flhDC operon is also negatively regulated by the His-Asp phosphorelay system RcsCDB. The regulation is potentiated by the RcsB cofactor RcsA. Genetic analysis indicates that an RcsAB box, located downstream of the promoter, is required for the regulation. The binding of RcsB and RcsA to this site was demonstrated by gel retardation and DNase I protection assays. In addition, mutation analysis suggests that RcsA-specific determinants lie in the right part of the ‘RcsAB box’.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
The RcsCDB His-Asp phosphorelay system was initially identified as a positive regulator of the cps genes, involved in the biosynthesis of capsular exopolysaccharide (EPS) in Escherichia coli (Gottesman et al., 1985). This system is dispensable for growth in normal conditions, but is essential for recovery from chlorpromazine-induced stress (Conter et al., 2002). Among the approximately 30 His-Asp phosphorelay systems identified in E. coli (Mizuno, 1997), the Rcs system is unique. Indeed, all but Rcs are true two-component systems whereas Rcs is the only one involving three proteins in the phosphorelay. The regulator RcsB is activated by the transfer of a phosphate group from its cognate sensor, RcsC (Stout and Gottesman, 1990), via a histidine-containing phosphotransmitter (Hpt) domain intermediate, RcsD (previously called YojN, Takeda et al., 2001; K. E. Rudd, http:bmb.med. miami.eduEcoGeneEcoWeb). The activation of EPS synthesis by the Rcs system, in E. coli as well as in other bacterial species, requires a cofactor, RcsA. RcsB and RcsA form a heterodimer that binds to a specific site, called ‘RcsAB box’ and located between positions −100 and −70 from the transcription start point (Wehland and Bernhard, 2000). The Rcs system has also been reported to positively control the expression of cell division genes and of the osmoregulated gene osmC in E. coli (Gervais et al., 1992; . Carballes et al., 1999; Davalos-Garcia et al., 2001). In neither of these cases is regulation modulated by RcsA. The activation is direct and requires a specific sequence, the ‘RcsB box’, centred at position −44/−43 from the transcription start point. In Salmonella typhi, the Rcs system directly or indirectly represses the production of flagellin and invasion proteins (Arricau et al., 1998), and in E. coli as well as in Proteus mirabilis, it affects the swarming behaviour of the cells (Belas et al., 1998; Takeda et al., 2001). Transcriptome analyses in two-component system mutants of E. coli indicate that several of these systems, including the AtoS/AtoC system and RcsB, are involved in the regulation of flagellar synthesis (Oshima et al., 2002).
About 50 genes are involved in flagellum synthesis, motility and chemotaxis in E. coli. They are expressed in a hierarchical fashion, and have been organized into three classes according to the level in the hierarchy at which they are expressed (reviewed by Chilcott and Hughes, 2000). At the lowest level, class III contains the flagellin-encoding gene fliC and the genes involved in chemotaxis. Class II contains the σF factor-encoding gene fliA, required for the expression of the genes of Class III, the anti-σF protein-encoding gene flgM and the genes required for synthesis of the flagellum basal body. Only two genes, flhD and flhC, organized into an operon, define class I. They code for a transcriptional activator required for the expression of the genes of Class II. The formation of the flagellum basal body by the class II genes allows the cell to exclude the anti-σF protein FlgM from the cytoplasm, thus releasing σF which can then transcribe the class III genes. Thus, activation of the entire set of motility and chemotaxis genes depends on the expression of the master operon flhDC.
Several regulators of the flhDC operon have been identified. High osmolarity inhibits the operon. This effect is mediated by the protein OmpR. Two OmpR-binding sites were found in the flhDC promoter region (Shin and Park, 1995). The synthesis of the flagellum is positively regulated by the CAP–cAMP complex (Adler and Templeton, 1967; Yokota and Gots, 1970; Silverman and Simon, 1974). A CAP-binding site was found upstream of the flhDC promoter (Soutourina et al., 1999). The histone-like nucleoid-structuring H-NS protein regulates flhDC expression both positively and negatively, and H-NS-binding sites were identified upstream and downstream of the promoter region (Soutourina et al., 1999; 2002). More recently flhDC expression was reported to be regulated by quorum sensing through the two-component system QseCB (Sperandio et al., 2002) and by the LysR-type regulator LrhA (Lehnen et al., 2002). Han et al. (1999) reported that expression in E. coli of the Bordetella pertussis response regulator gene bvgA repressed transcription of flhDC, and suggested that a BvgA-like protein might also exist in E. coli. In P. mirabilis, the flhDC operon is up-regulated by four unlinked genes, umoA, umoB, umoC and umoD, encoding putative membrane or periplasmic proteins (Dufour et al., 1998).
In the present study, we report that the RcsCDB His-Asp phosphorelay system regulates the flhDC master operon. This regulation is negative and requires an RcsAB box downstream of the promoter.
RcsB negatively regulates the flhDC master operon
With the aim of identifying targets of the His-Asp phosphorelay RcsCDB system, an E. coli genomic library was screened for promoters whose expression is affected by overexpression of a His-tagged version of the response regulator RcsB (Carballes et al., 1999; see Experimental procedures). The insert of a plasmid from one clone negatively affected by RcsB contained the promoter region of the flgAMN operon. This operon is involved in the synthesis of the flagellum (Fraser et al., 1999; Nambu and Kutsukake, 2000). It belongs to the class II genes, whose expression is activated by the class I master operon flhDC. A conserved FlhDC-binding site was found 20 nucleotides upstream of the promoter −10 sequence; however, analysis of the sequence at the flgA promoter regulatory region did not reveal any putative RcsB-binding site. In addition, we performed a deletion analysis and we could not isolate a construction resistant to RcsB without affecting the FlhDC-binding site (and thereby abolishing the activity of the promoter). Taking into account also the fact that the genetic screening was not exhaustive, because none of the known targets of the Rcs system have been isolated, we considered the possibility that the effect of RcsB on the expression of flgAMN operon might be indirect, and mediated by the FlhDC transcription activator.
A transcriptional fusion between the flhDC operon promoter region and lacZ was created in the pRS550 vector and integrated into the chromosome. The DNA fragment originated from the MG1655 strain and contained a region extending from −1057 to +34 relative to the transcription start-point as determined by Soutourina et al. (1999) (Fig. 1A; λflhD4). As shown in Fig. 2A, the expression of the fusion was strongly repressed by overexpression of RcsB, leading to a repression ratio of 10.8 after 150 min (Table 1). Inspection of the sequence in the vicinity of the promoter indicated a sequence similar to the consensus of RcsB-binding sites (Carballes et al., 1999), that extends from +3 to +20 (Fig. 1B and C; site flhD). Supporting the relevance of this sequence, a fusion with a fragment extending from −1057 to +2 (λflhD1, Fig. 1A) did not display the typical repression pattern obtained with the λflhD4 construction, although a constant slight reduction of the promoter activity was still observed over the growth curve (Fig. 2B). No sequence beyond −85 was required for the activity of RcsB because a fusion extending from −85 to +34 (λflhD7, Fig. 1A) remained repressible by RcsB (Table 1). With a fusion extending from −65 to +34 (λflhD5, Fig. 1A), a typical repression pattern was also obtained, indicating that no sequence beyond −65 is required for RcsB activity (Table 1). We note that the basal activity of the promoter in this fusion was strongly reduced, probably because of the deletion of a CAP-binding site, located between positions −79 and −64, which is required for activation of the promoter by the CAP–cAMP complex (Soutourina et al., 1999). A fusion extending from −85 to +22, keeping the putative RcsB box intact (λflhD8, Fig. 1A), was also sensitive to RcsB, as shown by the 16.4-fold reduction of the activity of the fusion when RcsB was overexpressed (Table 1). Point mutations were then introduced in the putative RcsB box on the −85 to +22 DNA fragment, replacing either the conserved A at position −3 by a T or the conserved C at position +4 by a G (λflhD9 and λflhD10 respectively; Fig. 1A and C). As shown in Table 1, the activities of these fusions were only 2.7- and 2.6-fold lower when RcsB was overexpressed, as compared with the 16.4-fold effect with λflhD8 construction. Therefore, these data indicate that the flhDC operon is negatively regulated by RcsB and that this regulation requires a region containing an RcsB-binding site downstream of the promoter.
Table 1. Effects of RcsB and RcsA on flhDC::lacZ fusions.
β-Galactosidase was assayed as in Figs 1 and 3, and only the values at 150 min are given. Wild-type (wt), rcsA or rcsB strains were used. The expression of the His-tagged versions of RcsB and RcsA was induced with IPTG from the plasmids pHRcsB and pHRcsA respectively. The first and second numbers are the β-galactosidase-specific activities in the uninduced (no IPTG) and induced (with IPTG) cultures respectively. The ratio of the two values is given between brackets. ND: no data. The λflhD fusions are described in Fig. 1A.
In agreement with these results, the cell motility was reduced when RcsB was overexpressed. The magnitude of the effect was correlated with the level of expression. At 10 and 500 µM IPTG, the diameter of the motility rings for the MG1655 strain containing pHRcsB plasmid were half and one-seventh of that without IPTG respectively. Similar results were obtained with isogenic cps+ and cps– strains (MG1655 and SK1447; data not shown).
The flhDC operon is negatively regulated by the RcsB cofactor RcsA
The activation by RcsB of the cps genes is potentiated by a cofactor, RcsA. This activation requires the binding of an RcsA-RcsB heterodimer to an RcsAB box, located between −100 and −70 from the transcription start point (. Wehland and Bernhard, 2000). In order to test whether the expression of the flhDC operon is RcsA dependent, the activity of the λflhD4 fusion was monitored in the presence of a plasmid expressing a His-tagged version of RcsA from the lacP promoter (pHRcsA). As shown in Table 1, the activity of the fusion was reduced 18.8-fold when RcsA was overexpressed. Therefore, the flhDC promoter is negatively regulated by RcsA. This effect was not observed in an rcsB background (Table 1), indicating that RcsA activity requires RcsB. In contrast, the repression by RcsB was similar in wild-type and rcsA strains, in agreement with RcsA as being a dispensable RcsB cofactor (Table 1). A fusion, in which the RcsAB box was deleted, as in λflhD1, was resistant to RcsA (Table 1). The fusions with a mutated RcsAB box exhibited a strongly reduced sensitivity to RcsA (Table 1, compare λflhD9 and λflhD10 with λflhD8). These data indicated therefore that flhDC is also regulated by RcsA and that this regulation requires a site located downstream of the promoter. This site will be referred to as an RcsAB box hereafter (Fig. 1B).
RcsA-specific determinants lie in the right half of the RcsAB box
Unexpectedly, a deletion removing the right half of the RcsAB box (λflhD3, Fig. 1A and C) remained sensitive to RcsB, both in a wild-type and rcsA background (Fig. 3 and Table 1). Inspection of the sequence, however, suggested that the remaining left half of the box together with the sequence brought by the vector might have recreated a functional RcsB box (flhD3, Fig. 1C). The hybrid box differs at five positions from the wild type, but keeps the three most conserved bases at positions −4, −3 and +4 (G, A and C respectively). Interestingly, the λflhD3 fusion, which contains the hybrid RcsB box (flhD3), while still sensitive to RcsB, both in wild-type and rcsA strains, became resistant to RcsA (Fig. 3). A similar result was obtained with the λflhD6 fusion, containing also the hybrid box, but with the upstream sequences extending only to −85 (Fig. 1A and Table 1). These data therefore suggested that RcsA-specific determinants lie in the right half of the RcsAB box.
RcsB directly binds to flhDC promoter region
The ability of RcsA and RcsB to directly interact with the flhD promoter region was tested in a gel retardation assay. The experiment was performed with His-tagged versions of both wild-type RcsB and a mutant form, RcsBD56E, in which the conserved asp residue was replaced by a glu residue. This mutation makes the protein more active, probably by mimicking the phosphorylated state of the protein (Gupte et al., 1997; Davalos-Garcia et al., 2001). The RcsA protein was a MalE–RcsA protein fusion described in Kelm et al. (1997). Overproduction of each of these proteins negatively affects expression of flhD in vivo (data not shown). As shown in Fig. 4A, in the presence of equimolar amounts of MalE–RcsA and RcsBD56E, a retarded complex with the wild-type flhD4 probe started to appear at a concentration of 0.1 µM of each protein, and the amount of this complex increased with increasing concentrations of the proteins (Fig. 4A). This complex was specific as it was not observed with the flhD1 probe, in which the RcsAB box is missing. When the same experiment was performed with a mixture of MalE–RcsA and wild-type RcsB protein, eight times more proteins were required in order to obtain a retarded complex (Fig. 4B). No retarded complex was observed when 1.5 µM RcsA, RcsB or RcsBD56E were separately used with the flhD4 probe (data not shown). These results therefore suggested that flhD is directly regulated by RcsB and RcsA, and that this regulation requires a functional RcsAB box. With the hybrid probe (flhD3), in the same range of concentrations, a mixture of RcsBD56E and RcsA was unable to give a detectable complex, indicating that the probe was resistant to RcsA, in agreement with the results obtained in vivo (data not shown).
Definition of the RcsAB box in vitro
In order to confirm the identification of the RcsAB box, DNase I protection assays were conducted with purified His-tagged RcsBD56E and Mal-RcsA proteins. As shown in Fig. 5, a single protected region was observed, extending from +5 to +25 and from +2 to +22 in the template and non-template strands respectively. In agreement with the retardation assay, RcsA alone at 1.9 µM gave no protection (Fig. 5A and B, lanes 11), whereas with RcsB alone, protection was visible at 2 µM of the protein. In the presence of 1.9 µM RcsA, a protection was obtained with only 0.4 µM RcsB (Fig. 5A and B, lanes 8). The location of the protected region completely agrees with that of the RcsAB box defined in vivo by deletion and single mutations analysis.
The flhDC operon codes for a transcription regulator required to initiate the expression of genes involved in the synthesis of the flagellum and in chemotaxis. In this study, we have shown that flhDC is regulated by the His-Asp phophorelay RcsCDB system. This regulation involves the cofactor RcsA and a sequence, the ‘RcsAB box’, located downstream of the promoter. This is the first reported example of direct negative regulation of gene expression by the Rcs system.
The rcs-dependent regulation of the flhDC operon is in agreement with earlier reports and explains several observations. In mutants lacking membrane-derived oligosaccharides, the Rcs pathway is activated, resulting in increased biosynthesis of exopolysaccharide (Ebel et al., 1997). These mutants also exhibited both defective chemotaxis and a reduction in the amount of flagellin (Fiedler and Rotering, 1988). Our results suggest that both phenotypes are the consequence of the activation of the Rcs system, which by repressing the expression of the flhDC operon, reduced the expression of the genes required for the synthesis of flagellin and for chemotaxis. Takeda et al. (2001) showed that the phosphorelay leading to the activation of RcsB was deficient in ΔrcsC and ΔrcsD (ΔyojN) mutants and that both mutants displayed an exaggerated swarming behaviour. This latter phenotype, as suggested by our work, is probably the consequence of higher basal expression of the flhDC operon in the rcs mutants, leading to higher expression of genes involved in swarming. The same explanation could also apply to the recently published work of the same group, showing an up-regulation of flgC, flgG and flgI flagellar genes in an rcsB mutant (Oshima et al., 2002). Finally, Arricau et al. (1998) reported that the production of flagellin in S. typhi was negatively controlled by RcsC-RcsB. Our study suggests that this control is indirect, due to repression of the flhDC master operon by RcsB, which thus prevents expression of the flagellin biosynthesis gene, fliC. Despite the implication of more than 50 genes in the process of motility, their hierarchal organization is an efficient way of coordinating their expression by simply controlling the expression of the master operon flhDC. This is illustrated by the fact that all regulators identified so far act on flhDC expression. The importance for bacteria of controlling motility and chemotaxis in response to the environment has been discussed (see Ottemann and Miller, 1997). In order to exert this control appropriately, bacteria must have devices to monitor the surrounding conditions. Two such systems have been identified and both are His-Asp phosphorelays: the EnvZ/OmpR system (Shin and Park, 1995) and the QseCB system (Sperandio et al., 2002) responding to the osmolarity and the cell density respectively. RcsCDB is now a third example of such a system, involved in the control of motility and chemotaxis.
The various signals known to activate the Rcs pathway, either by means of mutations, overexpression of envelope-associated protein or by chemical or physical treatments, have all in common the alteration of the cell envelope. However, the molecular signal sensed by RcsC has not yet been clearly identified. Osmotic shock has been shown to activate the cps genes through the Rcs system (Sledjeski and Gottesman, 1996). In contrast, RcsCDB does not transduce the osmotic signal to at least two of its targets, osmCp1 (Davalos-Garcia et al., 2001) and flhDC (data not shown). However, the overlap of RcsAB- and OmpR-binding sites (Shin and Park, 1995; Bartlett et al., 1988; this study) and the increased basal activity of the flhDC promoter in the rcsB mutant, suggest that the Rcs system could possibly modulate the OmpR-dependent osmotic regulation. Further study will be necessary to elucidate the relationship between the Rcs system and the response to osmotic shock.
Several observations suggested that the flhDC operon might be regulated by heat shock (Adler and Templeton, 1967; Shi et al., 1992; Houry et al., 1999). In agreement, we found that the flhDC operon was negatively regulated by heat shock, but that this regulation was not mediated by the Rcs system (data not shown).
Targets regulated by the Rcs system can be classified into two types, depending on their responsiveness to the RcsB-cofactor RcsA. The exopolysaccharide synthesis genes and the flhDC genes belong to the RcsA-dependent class; the cell division fts genes and the osmoregulated osmC gene belong to the RcsA-independent class. The involvement of an additional component, RcsA, may provide the cell with a means to respond to environmental signals different from those activating the RcsC pathway. Alternatively, it may influence the threshold of activation, introducing a hierarchy of the different members of the RcsB regulon as a function of the intensity of the inducing signal. Binding of the RcsA-RcsB heterodimer to the regulatory region of the exopolysaccharide synthesis genes was reported (Kelm et al., 1997). It is, however, not clear at this point whether RcsA forms an active complex with the unphosphorylated form of RcsB or with a fraction of RcsB proteins that is phosphorylated even in the absence of a signal inducing RcsC. Our observation that, in vitro, RcsA was eightfold more active in the presence of a constitutive form of RcsB than with wild-type RcsB, suggests that RcsA is preferentially interacting with the phosphorylated form of RcsB.
For the RcsA-independent class of RcsB targets, the site required for RcsB activity (the RcsB box) is located next to the −35 sequence, centred at −43/−44. For the RcsA-dependent class, this site (the RcsAB box) is located further upstream (Wehland and Bernhard, 2000) or, as shown by this study, downstream of the promoter. Is there a mechanistic reason for having the Rcs boxes located next to the promoter for the RcsA-independent targets and further away for the others? The failure to observe binding of RcsB in gel retardation assays except when RNA polymerase (Davalos-Garcia et al., 2001) or RcsA (Kelm et al., 1997; this study) was present, suggests that RcsB might require a cofactor in order to stably bind to its site. This cofactor could be the RNA polymerase, when the site is next to −35 or RcsA when the site is further away from the promoter. At the sequence level, the distinction between an RcsB box and an RcsAB box is not straightforward, although as noted by Carballes et al. (1999), there is less conservation in the right half-site of the two boxes, suggesting that RcsA might bind to this half-site. This study in which the flhDC RcsAB box has been changed to an RcsB box with a deletion of the right half-site is in agreement with this hypothesis. Further study will be required to identify RcsA- and RcsB-specific recognition sequences.
The regulation of the flhDC promoter is the first reported case of a negative regulation by the Rcs system. Formally, two different mechanisms might account for this negative regulation. The binding of the RcsAB heterodimer might sterically prevent the accessibility of the flhDC promoter to the RNA polymerase, or alternatively might lead to premature transcription termination. Because bound RNA polymerase was reported to cover the flhD promoter region beyond the +20 position (Soutourina et al., 1999), we favour the former mechanism.
Strains, plasmids and bacteriophages
The bacterial strains, plasmids and bacteriophages used in this study are listed in Table 2. The transcriptional fusions were constructed in the cloning vector pRS550, rescued in the phage λRS45 and installed in single copy on the chromosome by lysogenization (Simons et al., 1987). We considered that a monolysogen state was achieved when corresponding colonies gave a stable blue coloration on Xgal-containing plates after several subcloning steps. The DNA fragments for cloning were generated by PCR and cloned into the BamHI–EcoRI cloning sites of pRS550.
Table 2. Bacterial strains, bacteriophage and plasmids.
To screen for RcsB targets, chromosomal DNA was extracted from the E. coli MC1061 derivative, JS219, partially digested with Sau3A and ligated to the vector pRS550. The ligation products were used to transform a JS219 strain containing a plasmid expressing a His-tagged version of RcsB from the lacP promoter (pHRcsB, previously named pFAB1, Carballes et al., 1999). Transformants were plated on Xgal-containing Luria–Bertani (LB) medium with or without IPTG, to screen for clones activated or repressed by RcsB respectively. Plasmids from such clones were purified and their inserts sequenced.
To test the effect of overexpressed RcsB or RcsA on the transcriptional fusions, cells were grown in LB broth at 37°C. Overnight cultures were diluted 100-fold, grown for five generations, then diluted 40-fold in pre-warmed medium with or without 500 µM IPTG. The cultures were sampled at intervals for assay of β-galactosidase. β-Galactosidase-specific activities are expressed in Miller units (Miller, 1992).
Each assay was done in quadriplicate. LB broth containing 0.3% agar plates were used. Plates without or with 10 and 500 µM IPTG were inoculated from exponentially growing cells and incubated for 9 h at 37°C. Wild-type MG1655 or isogenic capsule-deficient cps– strain SK1447 containing pHRcsB plasmid or vector plasmid pIM10 were tested.
Purification of proteins
His-tagged RcsB and His-tagged RcsBD56E proteins were purified as described in Carballes et al. (1999). The MalE–RcsA fusion protein was purified as described in Kelm et al. (1997) with modifications. Cells were disrupted by sonication after several cycles of freezing and thawing. The S100 fraction was loaded onto an amylose resin chromatography column (New England Biolabs). The last fractions from the elution step were more than 90% pure as judged by Coomassie blue staining of an SDS–PAGE. No further purification steps were added. Protein concentrations were determined by Bio-Rad protein assay.
The gel retardation assay was performed as described in Kelm et al. (1997). Probes were generated by PCR with 33P 5′ end-labelled primers. They extend from position −65 to +2 (flhD1) or +34 (flhD4) of the flhD promoter region (Fig. 1A) and include 25 additional bp downstream the EcoRI site of the cloning vector pRS550. Five thousand cpm per reaction were used.
DNase I footprinting
The flhD fragment was PCR-amplified from genomic DNA, and extends from −65 to +152. The 32P-end-labeled probe (5 × 104 cpm) was incubated with purified protein as indicated in corresponding figures, in a 16 µl solution containing 20 mM HEPES, pH 8, 1 mM EDTA, 7 mM MgCl2, 3 mM CaCl2, 50 mM NaCl, 7 mM β-mercaptoethanol, 10% v/v glycerol, 1 µg poly-(dI–dC/poly-(dI–dC) (Pharmacia). After 20 min of incubation, DNase I (Appligene) was added (0.625 µg ml−1 final concentration). The digestion was stopped after 3 min by adding 4 µl solution containing 1.5 M COONH4, 0.25 M EDTA and 250 µg ml−1 glycogene. After ethanol precipitation, the pellet was resuspended in 4 µl loading buffer. Digests and their corresponding sequences (performed with the same labelled primer) were analysed on 6% denaturing polyacrylamide gels.
We are grateful to F. Bernhard, for providing the pM-RcsAEC plasmid, to D. Lane, L. Poljak and A. J. Carpousis for helpful discussions on the manuscript, to M. Cashel for the strain CF6343, to J.-Y. Bouet, F. Carballes, Y. Crasnier and M. Buc for technical assistance. This work was supported in part by the Université Paul Sabatier, the French Ministère de l’Enseignement Supérieur et de la Recherche (Programme de Recherche Fondamentale en Microbiologie, Maladies Infectieuses et Parasitaires) and a grant from the Institut Universitaire de France to C.G.