The RcsAB Box

The interaction of the two transcriptional regulators RcsA and RcsB with a specific operator is a common mechanism in the activation of capsule biosynthesis in enteric bacteria. We describe RcsAB binding sites in the wza promoter of the operon for colanic acid biosynthesis in Escherichia coliK-12, in the galF promoter of the operon for K2 antigen biosynthesis in Klebsiella pneumoniae, and in thetviA (vipR) promoter of the operon for Vi antigen biosynthesis in Salmonella typhi. We further show the interaction of RcsAB with the rcsA promoters of various species, indicating that rcsA autoregulation also depends on the presence of both proteins. The compilation of all identified RcsAB binding sites revealed the conserved core sequence TaAGaatatTCctA, which we propose to be termed RcsAB box. The RcsAB box is also part ofBordetella pertussis BvgA binding sites and may represent a more distributed recognition motif within the LuxR superfamily of transcriptional regulators. The RcsAB box is essential for the induction of Rcs-regulated promoters. Site-specific mutations of conserved nucleotides in the RcsAB boxes of the E. coli wzaand rcsA promoters resulted in an exopolysaccharide-negative phenotype and in the reduction of reporter gene expression.

Encapsulation by exopolysaccharides (EPS) 1 protects bacteria against a variety of unfavorable environmental conditions. The production of EPS furthermore represents an essential factor in the virulence of bacterial pathogens (1). The biosynthesis of high molecular weight type I EPS in several enteric bacteria like Escherichia coli (2,3), Salmonella typhi (4), Klebsiella pneumoniae (5,6), and the plant pathogenic bacteria Erwinia amylovora (7)(8)(9)(10) and Pantoea stewartii (11) is controlled by the Rcs (regulation of capsule synthesis) regulatory network.
Two transcriptional regulators, RcsA and RcsB, are supposed to induce EPS biosynthesis cooperatively. The RcsB protein is highly conserved between different species with about 90% identity (10), and it represents the cytoplasmic activator of a classical bacterial two-component system. RcsB might be activated by the membrane-bound receptor RcsC (3,12) via phosphotransfer to highly conserved aspartic acid residues in the N-terminal domain of RcsB. RcsA and RcsB are both characterized by a LuxR-type C-terminal DNA binding motif, but RcsA does not contain an N-terminal phosphorylation motif. The RcsA protein is limiting for the induction of EPS biosynthesis and is virtually not detectable in the uninduced cell due to its rapid degradation by the Lon protease (13,14). The presence of RcsB is absolutely required for capsule biosynthesis, whereas an rcsA minus phenotype can be suppressed by multicopy rcsB (15). RcsA might therefore act as a coinducer of EPS biosynthesis by enhancing the DNA binding activity of RcsB. Recently, genetic evidence for an autoregulation of rcsA expression in E. coli has been reported and a DNA binding activity of RcsA at the rcsA promoter has been discussed (16).
We have previously shown that a heterodimer formed by one copy of RcsA and RcsB binds at corresponding regions approximately 500 bp upstream of the translational start sites of amsG and cpsA, the first open reading frames (ORF) in the E. amylovora ams operon for amylovoran biosynthesis, and in the P. stewartii cps operon for stewartan biosynthesis, respectively (17,18). The two operons are highly homologous, and the activation of Rcs-dependent promoters by a RcsAB heterodimer as a general mechanism remained unclear. In addition, some evidence for modified Rcs-dependent regulation mechanisms in E. coli and S. typhi has been proposed (4,16).
In this work we demonstrate that the binding of RcsAB to regulatory DNA regions is a general principle in the Rcs-mediated activation of gene expression. We present RcsAB binding sites identified in the presumed main promoters of EPS biosynthetic operons of E. coli, K. pneumoniae, and S. typhi. RcsAB further modulates the rcsA autoregulation in those species by binding to the rcsA promoters. The compilation of all identified RcsAB binding sites allows us to define the RcsAB box as a new conserved bacterial operator. The RcsAB box was analyzed in vitro and in vivo, and it was found to be essential for full promoter activity in E. coli.

MATERIALS AND METHODS
Bacterial Strains, Plasmids, Oligonucleotides, and DNA Techniques-The E. coli strains Xl1-Blue (19), BL21, C600, DH5␣ (Stratagene), and JB3034 (15) and the plasmids pQE30 (Qiagen), pMalc2 (New England Biolabs), and pfdA8 (20) were used for cloning and expression studies. Standard DNA techniques were done as described (21). DNA fragments were amplified from chromosomal DNA of strain Xl1-Blue with Vent polymerase and suitable primers. The sequences of the oligonucleotides are available upon request.
Expression and Purification of Proteins-RcsA proteins were produced with the plasmids pM-RcsA EA , pM-RcsA EC , and pM-RcsA PS (17,18) in strain BL21 as C-terminal fusions to the maltose-binding protein.
RcsB proteins were produced with the plasmids pQ-RcsB EC and pQ-RcsB EA (17) with an N-terminal poly(His) 6 tag in the strain JB3034. The proteins were purified as described (17,18). If appropriate, the purified RcsA and RcsB proteins of E. coli, E. amylovora, and P. stewartii were named according to their origin RcsA EC , RcsA EA , RcsA PS , RcsB EC , and RcsB EA , respectively. A BamHI/HindIII restriction fragment from plasmid pQHB (22) containing the coding region of Bordetella pertussis bvgA was cloned into the expression vector pMalc2, resulting in plasmid pM-bvgA. The BvgA protein was produced from * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Electrophoretic Mobility Shift Assay (EMSA)-Radioactive DNA labeling with [␣-32 P]dATP and the EMSA technique were done as described previously (17,18). The RcsAB heterodimer was obtained by mixing equimolar concentrations of the two proteins. For the reconstitution of double-stranded DNA fragments, about 1 g of each of two complementary oligonucleotides were mixed in 200 mM Tris/HCl, pH 7.5, 100 mM NaCl and incubated for 5 min at 95°C. The mixture was cooled slowly to room temperature and subsequently labeled. Phosphorylation of BvgA was obtained by incubating the protein in 50 mM Tris/HCl, pH 7.0, 20 mM MgCl 2 , 0.1 mM dithiothreitol, and 20 mM acetylphosphate for 20 min at 28°C.
Surface Plasmon Resonance (SPR) Technique-SPR measurements were performed with a BIAcore X instrument (BIAcore, Uppsala, Sweden). Biotinylated DNA (about 60 resonance units) were coupled to the streptavidin-coated sensor chip SA as recommended by the manufacturer. The experiments were carried out at a flow rate of 50 l/min. The DNA fragments and proteins were diluted in running buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 10 mM dithiothreitol, 0.1 mM EDTA). Bovine serum albumin and -DNA were added to the protein solutions to a final concentration of 200 and 8 ng/l, respectively. RcsA/RcsB mixtures of various concentrations ranging from about 47 nM to about 7.5 M were injected allowing an association time of 120 s and a dissociation time of 300 s. A reference flow cell loaded with a random DNA target of same size as the probe DNA target was used to subtract unspecific DNA/protein interactions. Regeneration of the chip surface was achieved by removing all bound proteins with a pulse of 5 l of 0.05% SDS in running buffer.
Kinetic analyses were done using the BIAevaluation 3.0 program. To determine the binding properties of the proteins, 1:1 Langmuir kinetics provided by the software were used.
Site-directed Mutagenesis-Plasmid pMW31 was constructed by cloning a 3-kb DNA fragment starting seven nucleotides upstream of the RcsAB box of the wza gene into the BglII and PstI sites of the suicide vector pfdA8 (20). In plasmid pMW29, four essential nucleotide positions in the RcsAB box TAAAGAAACTCCTA of the 3-kb fragment were modified by PCR, resulting in the sequence GAACTCAACTCCTA, where the mutated bases were underlined. The plasmids were transformed into E. coli C600 by electroporation and selected for kanamycin resistance. The correct insertion of the plasmids by homologous recombination was verified by PCR analysis of the isolated chromosomal DNA.
An approximately 1-kb DNA fragment generated by PCR and containing the complete E. coli rcsA gene starting 38 bp upstream of the RcsAB box, was cloned into the vector pBluescript KS ϩ resulting in plasmid prcsA-WT. Four essential positions of the RcsAB box TAAG-GATTATCCGA in plasmid prcsA-WT were mutated by PCR upon introduction of an EcoRI restriction site and resulting in the sequence GAATTCTTATCCGA with the mutated bases underlined.
Determination of Polysaccharides and Biochemical Assays-EPS were quantified from cells grown on cellophane-covered LB-agar for 24 h at 37°C. After harvesting, the cell number was determined and the capsules were washed off from the cells by vigorous shaking for 3 min. The cells were pelleted by ultracentrifugation, and the supernatant was dialyzed against water. The polysaccharides were determined as described (23).
The enzymatic activity of the ␤-galactosidase was determined with the o-nitrophenyl ␤-D-galactopyranoside assay after Miller (24). nucleotide positions Ϫ106 to Ϫ80. This was also observed with the 38-bp fragment Pwza 38 spanning nucleotide positions Ϫ119 to Ϫ82. This indicated that the 28-bp region spanning nucleotide positions Ϫ106 to Ϫ79 relative to the transcriptional start site of wza was essential but not sufficient for the binding of RcsAB EC (Figs. 1 and 2). An extension of the 3Ј-end of the 55-bp fragment did not further contribute to a better binding of RcsAB EC , and the extent of retardation of the 72-bp fragment Pwza 72 spanning positions Ϫ119 to Ϫ48 was comparable to that of Pwza 55 (data not shown). Incubation at 28°C compared with 37°C prior to electrophoresis increased the percent of retardation of the RcsAB EC /DNA complex about 3-fold.

Location of an RcsAB
The alignment of the 55-bp fragment with the RcsAB binding site of the E. amylovora amsG promoter (18) implicated several nucleotides in the region from Ϫ115 to Ϫ96 relative to the transcriptional start site of wza as putative targets for RcsAB ( Fig. 1). Suspected positions were further analyzed by the introduction of putative up and down mutations according to the RcsAB amsG consensus ( Table I). The replacement of the degenerated adenine at position Ϫ110 by a conserved guanine in the fragments Pwza 72 (G Ϫ109 ) and Pwza 72 (G Ϫ109 A Ϫ108 ) increased the extent of retardation by RcsAB EC (Table II). In contrast, the retardation was considerably reduced after replacing the highly conserved thymine at position Ϫ112 in fragment Pwza 72 (G Ϫ112 ) by guanine. The retardation was completely abolished in fragment Pwza 72 (C Ϫ110 T Ϫ108 C Ϫ106 ) carrying mutations in three conserved positions. A decreased percent of retardation in an EMSA was furthermore observed after the replacement of two less conserved adenines by cytosines in the fragment Pwza 72 (C Ϫ98 C Ϫ96 ). Interestingly, two mutations adjacent to the putative RcsAB consensus also reduced the extent of retardation of fragment Pwza 72 (G Ϫ91 C Ϫ90 ) by the two proteins (Table II). This gives evidence for additional DNA/protein interactions neighboring the consensus motif.
To analyze whether the wza promoter exhibits some preference for the recognition by the homologous RcsAB EC proteins, we used the 55-bp fragment of the wza promoter as a target in EMSAs with various combinations of the RcsA EA , RcsA EC , Rc-sA PS , RcsB EA , and RcsB EC proteins (Fig. 2). The wza promoter was recognized by the heterologous Rcs proteins in all combinations tested, and we could not detect significant differences in the extent of retardation of the DNA fragment. We observed no binding of RcsA EC or RcsB EC alone to the 55-bp fragment, or to the 485-bp fragment containing the complete wza promoter in concentrations up to 4.5 M. In contrast, approximately 0.2 M of the two proteins together already retarded these DNA fragments in EMSAs. The DNA fragment Pwza 48 spanning the nucleotide positions Ϫ95 to Ϫ48, including an inverted repeat sequence, was retarded neither by RcsA EC nor by RcsAB EC (Fig. 2). In addition, the mutation of four bases in the inverted repeat sequence of the fragment Pwza 72 (C Ϫ68 G Ϫ67 G Ϫ66 A Ϫ65 ) did not show any effect on the retardation by RcsAB EC (Table  II).
The binding kinetics of the RcsAB EC heterodimer at Pwza 72 was analyzed by the surface plasmon resonance technique (Fig.  3). With protein concentrations in a range of 47 nM and 7.5 M, the k a was calculated to 5.4 Ϯ 3.3 ϫ 10 4 M Ϫ1 s Ϫ1 and the k d to 1.4 Ϯ 0.4 ϫ 10 Ϫ3 s Ϫ1 , resulting in a K D of 77 Ϯ 28 nM. The K D of RcsAB EC at the E. coli wza promoter corresponds to the K D of RcsAB EA at the E. amylovora amsG promoter previously determined by the EMSA technique (18). We furthermore analyzed the DNA fragment Pwza 72 (G Ϫ112 C Ϫ109 T Ϫ108 C Ϫ107 ), containing four point mutations in highly conserved positions (Fig. 1). This fragment was not retarded by RcsAB in an EMSA (Table II). The k a was calculated to 560 Ϯ 140 M Ϫ1 s Ϫ1 and the k d to 2.4 Ϯ 1.0 ϫ 10 Ϫ2 s Ϫ1 , resulting in an approximately 10 3 -fold increased K D of 50 Ϯ 30 M.
In Vivo Analysis of the E. coli wza RcsAB Binding Site by Mutagenesis-The chromosomal merodiploids MW29 and MW31 of the E. coli K12 strain C600 were constructed after integration of the two plasmids pMW29 and pMW31 by homologous recombination. The wza promoter of strain MW31 was truncated just upstream of the RcsAB box, and strain MW29 contained additionally four point mutations in essential bases within the identified RcsAB binding site. The phenotypes of the mutants were assayed after introduction of plasmid pEA101 containing the E. amylovora rcsA gene, which resulted in the wild type strain C600 in the induction of colanic acid biosynthesis by activation of the wza promoter. The EPS production and the phenotype of the control mutant MW31 ϫ pEA101 was not altered compared with the wild type strain C600 ϫ pEA101 (Table III). Thus, the approximately 450-bp fragment upstream of wza is sufficient for full promoter activity. In contrast, the EPS production of the mutant MW29 ϫ pEA101 was drastically reduced and the mutant showed a butyrous colony type (Table III). These results demonstrate the importance of the identified RcsAB binding site for the activation of colanic acid biosynthesis, and they indicate that RcsAB might also bind in vivo to that region.
RcsA and RcsB Bind to the rcsA Promoters of E. coli, K. pneumoniae, S. typhi, and E. amylovora-The autoregulation of E. coli rcsA has been reported previously (16), and we inves- tigated whether the activation of rcsA promoters is also directed via DNA binding of RcsAB. The 277-bp PCR fragment PrcsA EC277 containing the intergenic region between the E. coli fliR and rcsA genes including the start codon of rcsA EC was clearly retarded by RcsAB EC (Fig. 4A). A putative RcsAB binding site was detected at nucleotide positions Ϫ264 to Ϫ251 relative to the translational start site of rcsA EC (Table I). Accordingly, the reconstituted 34-bp DNA fragment PrcsA EC34 spanning nucleotide positions Ϫ274 to Ϫ241 was retarded by RcsAB EC (Fig. 4A). Among the most critical positions for RcsAB binding are three conserved purines most likely represented by the sequence GGA at positions Ϫ261 to Ϫ259 in the rcsA EC promoter. The mutation of the three purines to the sequence TTC in fragment PrcsA ECM completely abolished retardation of the 277-bp fragment by RcsAB EC (Fig. 4A). This demonstrated that the proposed sequence is essential for the in vitro binding of the Rcs proteins to the rcsA EC promoter. The rcsA autoregulation appears to be dependent on the presence of both proteins, as neither RcsA nor RcsB alone in concentrations of up to 4.5 M were able to shift the 277-bp fragment of the rcsA EC promoter in EMSAs.
If the autoregulation of rcsA expression is a conserved mechanism, an RcsAB binding site should also be present in the rcsA promoters of other species. We detected putative RcsAB binding sites in the promoter regions of rcsA genes from E. amylovora, K. pneumoniae and S. typhi at locations between Ϫ321 to Ϫ 244 relative to the translational start sites (Table I). We could demonstrate an interaction of RcsAB EC with the 29-bp fragment PrcsA EA from Ϫ331 to Ϫ302 of E. amylovora rcsA, with the 29-bp fragment PrcsA KP from Ϫ239 to Ϫ267 of K. pneumoniae rcsA and with the 29-bp fragment PrcsA ST from Ϫ275 to Ϫ247 of S. typhi rcsA (Fig. 4B). An alignment of rcsA promoters of the closely related species E. coli, S. typhi and K. pneumoniae revealed several identical regions located further downstream, which might represent promoter consensus sequences (Fig. 5). The RcsAB binding sites are then located approximately 100 bp upstream of the presumed transcriptional start sites of the rcsA genes.
The compilation of the RcsAB binding sites of the four rcsA promoters and of the promoters of wza, amsG, and cpsA (Table  I) revealed several highly conserved positions within a core sequence of 14 bp, which we will further term RcsAB box.
The RcsAB Box Is Essential for rcsA Autoregulation in E. coli-The plasmids prcsA-WT, containing the cloned E. coli rcsA gene with an approximately 300-bp upstream region including the RcsAB box and prcsA-M4, containing mutations in four essential positions of the RcsAB box, were transformed into the E. coli strain DH5␣, and the bacteria were analyzed for their phenotype. The plasmid prcsA-WT with the wild type rcsA promoter increased the EPS production and resulted in a fluidal colony type. This was obviously due to an increased RcsA copy number (Table III). In contrast, the colonies remained butyrous with plasmid prcsA-M4 containing the four point mutations in the RcsAB box, and the EPS production was   dramatically decreased compared with strain DH5␣ ϫ prcsA-WT. These results show that the RcsAB box is also essential for the rcsA autoregulation in vivo in E. coli. In contrast, the introduction of plasmid prcsA-M4 in the lon minus strain SG1087 resulted in a fluidal phenotype and in an increased EPS production (Table III). Accordingly, the expression of a cpsB::lacZ fusion in the lon minus strain JB3034 clearly increased upon introduction of plasmid prcsA-M4 (Table III), but was still lower compared with that of strain JB3034 ϫ prcsA-WT. The absence of the Lon protease increases the half-life of RcsA. The background expression of rcsA with plasmid prcsA-M4 is then obviously sufficient for the activation of the colanic acid biosynthetic operon.
Identification of an RcsAB Box in the Gene Cluster for K2 Antigen Expression in K. pneumoniae, and for Vi Antigen Expression in S. typhi-The rcsA and rcsB genes have also been described from the enteric bacteria S. typhi and K. pneumoniae. If the corresponding capsular polysaccharide biosynthetic gene clusters were regulated by RcsAB, a RcsAB box should be present in the main promoters. The first ORF of the S. typhi Vi antigen cluster encodes for the putative regulator TviA (VipR) and a putative RcsAB box was found at position Ϫ322 to Ϫ309 relative to the translational start site of tviA (Table I). The 60-bp fragment PtviA spanning nucleotides Ϫ347 to Ϫ288 was retarded by RcsAB EC in EMSAs (Fig. 6). This indicates that the tviA RcsAB box is recognized in vitro by the two proteins. 16 ORFs are described for the K. pneumoniae K2 antigen biosynthesis gene cluster. An approximately 0.9-kb intergenic region containing a putative 54 -dependent promoter precedes orf3, but the DNA fragment does not contain a RcsAB box. However, a RcsAB box is located at positions Ϫ181 to Ϫ168 relative to the translational start site of orf1, encoding a GalF like protein (Table I). The retardation of the 59-bp DNA fragment PgalF spanning nucleotides Ϫ202 to Ϫ144 by RcsAB EC (Fig. 6) gives evidence that the region upstream of galF (orf1) might contain an Rcs-dependent promoter.
The Heterodimer RcsAB and the Transcriptional Regulator BvgA of Bordetella pertussis Recognize Similar DNA Sequences-Similarity searches with the programs MEME and MAST (32) revealed potential RcsAB boxes within the regulatory regions bvgA and fha of B. pertussis and Bordatella parapertussis ( Table I). The two promoters are reported to be activated by BvgA, a transcriptional regulator of the LuxR superfamily, whose DNA binding domain is homologous to that of RcsA and a Relevant genotype. b Colony type determined after 24 h of growth on LB-agar at 37°C. c After 24 h of growth on LB-agar at 37°C and estimated with the anthron assay. Means of at least three determinations. -, not applicable. d ␤-Galactosidase units estimated after Miller (24). Means of at least three determinations.
RcsB. The two 50-bp DNA fragments PbvgA BP and PbvgA BA , containing the putative RcsAB boxes of the B. pertussis and B. parapertussis bvga promoters, respectively, and the 50-bp fragment Pfha containing the putative RcsAB box of the B. pertussis fha promoter, were retarded by RcsAB in EMSAs (Fig. 7). The fragment Pfha was also retarded in vitro by the purified BvgA protein (Fig. 7), indicating that the homology of the DNA binding domains of RcsAB and BvgA is sufficient to result in the recognition of similar DNA sequences. The luxI box, a potential binding site for the LuxR protein, is not related to the RcsAB box (Table I), and it was not retarded by RcsAB in an EMSA (data not shown).

DISCUSSION
The RcsAB box has now been identified in promoters of EPS biosynthetic operons of five different species. It is always present in the promoter region preceding the first ORF, while the organization of the operons is quite variable. The E. amylovora amsG and P. stewartii cpsA genes are homologous and encode for a putative UDP-galactose lipid-carrier transferase (28). The first ORF of the E. coli colanic acid biosynthetic operon, wza, encodes for a putative outer membrane lipoprotein (33) and shows homologies to amsH, the second ORF in the ams operon. The tviA (vipR) gene is so far unique to S. typhi and encodes for a putative regulator protein (4,34). The orf1 of K. pneumoniae encodes for a GalF homologue, while galF represents the last ORF of the E. coli wza operon (33). The location of the RcsAB box is therefore correlated to the putative main promoters of EPS biosynthetic operons and not to a specific gene.
The regulation of colanic acid biosynthesis in E. coli K12 by Rcs proteins has been shown (35). A fragment of approximately 470 bp of the wza promoter was found to be sufficient for an RcsAB dependent activation (25). This is in agreement with the phenotype of our mutant MW31. The characterized RcsAB box is located just at the 5Ј-end of this essential promoter region. As observed with the amsG promoter (18), RcsA alone was not able to bind to the wza promoter. In addition, an inverted repeat sequence, previously suspected to be an RcsA binding site (16), does not seem to be essential for the in vitro binding of RcsAB. However, some beneficial effects for the RcsAB binding at the wza promoter could not be ruled out as the deletion of sequences including the repeat resulted in a reduced extent of retardation by RcsAB. A guanine, which has been shown to be important for RcsAB binding at the amsG promoter (18), is replaced by an adenine in the RcsAB box of the wza promoter. In addition, the minimal size of the RcsAB binding site at the wza promoter is obviously larger compared with the previously identified 23-bp sites in the E. amylovora amsG and P. stewartii cpsA promoters (18). The reduced binding of RcsAB to the degenerated box might therefore be stabilized by additional DNA/protein contacts. The phenotype of the mutant MW29 demonstrates an essential role of the RcsAB box for EPS production in vivo, and it is in full accordance with the results obtained by the in vitro binding studies.
The rcs regulation of EPS biosynthesis was so far demonstrated in about 10 serotypes of K. pneumoniae (5,6) and in about 20 serotypes of E. coli (3), all producing structurally different capsular polysaccharides. While the complete sequence is so far only available for the serotype K. pneumoniae K2 biosynthetic operon, the presence of a RcsAB box can be expected also in the promoters of the other operons. Elevated copies of RcsB increased EPS biosynthesis in E. coli K30, but the phenotype of rcsA and rcsB mutants indicated that both genes are not essential for low level EPS biosynthesis (3). The large noncoding upstream regions of orf3 of the K2 biosynthetic operon and of its homologue orfX of the K30 biosynthetic operon contain a putative 54 -dependent promoter, while the only detectable RcsAB box in the K2 operon is present upstream of galF (orf1). RcsAB might therefore activate the expression of galF (orf1) and possibly orf2, whereas the further downstream located genes might depend upon the regulation of additional mechanisms. The product of galF (orf1) might increase the biosynthesis of activated sugar precursors, and the homology of orf2 to the EPS related gene orf1 of Aeromonas hydrophila indicates some involvement of its gene product in K2 polysaccharide biosynthesis.
The Vi antigen, a linear homopolymer of ␣-1,4 2-deoxy-2-Nacetylgalactosamine uronic acid, is produced by all strains of S. typhi, and Salmonella paratyphi as well as from some strains of Salmonella dublin and Citrobacter freundii (36,37). The RcsBdependent biosynthesis of the Vi polysaccharide has been reported (4,38), and the rcsA and rcsB genes have been isolated from S. typhi (4). However, an involvement of RcsA in the regulation of Vi antigen biosynthesis could not be shown (4). Our findings indicate a potential molecular target for the interaction of RcsA together with RcsB upstream of tviR (vipR) at the presumed main promoter for Vi antigen biosynthesis. The interaction of the putative regulator TviA (VipR) with its own promoter, containing the RcsAB box, and with RcsB (4) has been proposed (34). It will be interesting to elucidate the in vivo role of the RcsAB box in this complex regulation mechanism.
The detection of an RcsAB box in corresponding regions in the rcsA promoters of E. coli, K. pneumoniae, S. typhi, and E. amylovora agrees with a previously proposed autoregulation mechanism of rcsA expression in E. coli (16). The self-activation of the rcsA promoter might counteract a silencing mecha-nism on rcsA expression by the histone-like protein H-NS (39), and it could be important for the rapid increase in EPS biosynthesis as a fast response on environmental stimuli. We first present evidence that the rcsA self-activation is dependent on both RcsA and RcsB. The rcsA self-activation is not confined to E. coli but might be a general mechanism in enteric bacteria. We also demonstrate that the RcsAB box, previously identified by in vitro DNA/protein interaction studies, is essential for full activity of the rcsA promoter in vivo. An inverted repeat sequence, present upstream of the RcsAB box, is dispensable at least for the in vitro binding of RcsAB. In addition, the rcsA promoter remained highly active in plasmid prcsA-WT despite the truncation of this sequence. The chromosomal rcsA expression in an E. coli lon strain was rcsB-dependent, while the RcsA protein could be detected in the same strain containing multicopy rcsA (40). The mechanism of autoregulation requires some leakiness of the rcsA promoter. Its activity might therefore be enhanced but not strictly depend on the binding of RcsAB. This is in accordance with the described cpsB induction by the high copy plasmid prcsA-M4, containing a mutated RcsAB box, in an E. coli lon strain. The leaky expression from multicopy rcsA is obviously sufficient to activate some EPS biosynthesis, when the half-life of RcsA is drastically increased by the lon mutation (14).
The consensus RcsAB box with the sequence TaAGaatatTC-ctA was determined out of 12 identified wild type RcsAB boxes from the EPS biosynthetic operons of E. coli, S. typhi, K. pneumoniae, E. amylovora, and P. stewartii, from the rcsA promoters of E. amylovora, P. stewartii, E. coli, and K. pneumoniae and from the bvga and fha promoters of B. pertussis and B. parapertussis. The RcsAB box consensus shows some variations from the consensus previously revealed by in vitro selection of the amsG promoter (18). The differences might be caused by some influences of the DNA context of the in vitro selected sequence, and also by the relative low stringency used during the in vitro selection. The RcsAB operator is centered approximately 100 to 70 bp upstream of the presumed Ϫ35 regions of the wza and rcsA promoters. This indicates that the RcsAB dimer might act as a class I activator (41) and helps in attracting the RNA polymerase or in stabilizing the transcriptional complex.
A consensus motif present in the C-terminal DNA binding domains of transcriptional regulators of the LuxR superfamily indicates some common mechanisms in protein/DNA recognition (42). While the lux box, a 20-bp repeat centered at approximately Ϫ40 bp from the luxI transcriptional start proposed to be involved in the DNA binding of LuxR from Vibrio fischeri (31,43), is not related to the RcsAB box, similar DNA sequences are recognized by RcsAB and by BvgA, a response regulator of B. pertussis virulence factor genes. The DNA binding of BvgA at the RcsAB boxes in the fha and the bvgaP1 promoters has previously been shown (44 -46). The BvgA protein is also known to bind to the ptx promoter controlling the expression of pertussis toxin, which does not contain a RcsAB box. However, BvgA binds about 10 times weaker to that promoter and evidence for a different mechanism of DNA recognition has been proposed (45)(46)(47). The BvgA binding sites at the fha and bvgaP1 promoters are located from approximately Ϫ100 to Ϫ70 upstream of the transcriptional start sites, which exactly corresponds to the location of the RcsAB box in the E. coli wza and rcsA promoters. The interaction of the ␣-subunit of the RNA polymerase with BvgA has been demonstrated (43)(44)(45), and a similar function for RcsAB might be proposed. The homology of the DNA binding domains of BvgA and RcsAB in addition to the observed recognition of similar DNA sequences could point to a common phylogenetic origin. A RcsAB boxlike sequence might therefore be recognized also by some other members of the LuxR superfamily.