Identification of an RcsA/RcsB Recognition Motif in the Promoters of Exopolysaccharide Biosynthetic Operons from Erwinia amylovora and Pantoea stewartii Subspeciesstewartii *

The regulation of capsule synthesis (Rcs) regulatory network is responsible for the induction of exopolysaccharide biosynthesis in many enterobacterial species. We have previously shown that two transcriptional regulators, RcsA and RcsB, do bind as a heterodimer to the promoter of amsG, the first reading frame in the operon for amylovoran biosynthesis in the plant pathogenic bacterium Erwinia amylovora. We now identified a 23-base pair fragment from position −555 to −533 upstream of the translational start site of amsG as sufficient for the specific binding of the Rcs proteins. In addition, we could detect an RcsA/RcsB-binding site in a corresponding region of the promoter ofcpsA, the homologous counterpart to the E. amylovora amsG gene in the operon for stewartan biosynthesis ofPantoea stewartii. The specificity and characteristic parameters of the protein-DNA interaction were analyzed by DNA retardation, protein-DNA cross-linking, and directed mutagenesis. The central core motif TRVGAAWAWTSYG of the amsG promoter was found to be most important for the specific interaction with RcsA/RcsB, as evaluated by mutational analysis and an in vitroselection approach. The wild type P. stewartii Rcs binding motif is degenerated in two positions and an up-mutation according to our consensus motif resulted in about a 5-fold increased affinity of the RcsA/RcsB proteins.

The ability to produce capsules or exopolysaccharides (EPS) 1 is characteristic for most bacterial species. General benefits of encapsulation are the prevention of desiccation, advantages in the degradation of substrates by adherence, and the binding of toxins and nutrients (1). EPS is furthermore an essential determinant for the bacterial virulence in several host-pathogen interactions (2), e.g. during infections by the plant pathogenic bacteria Erwinia amylovora and Pantoea stewartii subsp. stewartii (Ref. 3; formerly Erwinia stewartii). The dense layer of EPS is supposed to shield invading microorganisms against host defense systems like the hypersensitive response reaction. It might further prevent cell aggregation by agglutinins, and it has been reported to accelerate the wilting of infected plants by plugging xylem vessels (4).
The EPS structure is highly variable, and different types have been classified by molecular weight and structural properties (5). The biosynthesis of the high molecular weight EPS type IA in several enterobacterial species is modulated by the Rcs (regulation of capsule synthesis) regulatory network (6). Prominent examples are the regulation of colanic acid and many K antigens in Escherichia coli (7,8), Klebsiella aerogenes (9 -11), and Salmonella typhi (12). In plant pathogenic bacteria, the regulation of amylovoran synthesis in E. amylovora (13)(14)(15)(16) and stewartan synthesis in P. stewartii (17) by Rcs proteins has been reported. EPS biosynthesis is supposed to be induced after perception of external signals by membranelocated sensors like the RcsC protein. The signal might be subsequently transduced by phosphorylation of the response regulator RcsB, and the two proteins represent a typical bacterial two-component system (18). High levels of EPS biosynthesis require the coinduction by the unstable protein RcsA (7,19,20). RcsA and RcsB are grouped into the LuxR class of bacterial regulators based on the sequence of their C-terminal helix-turn-helix DNA binding motifs.
The structural genes for the biosynthesis of amylovoran in E. amylovora, stewartan in P. stewartii, and colanic acid in E. coli are clustered and regulated as operons (21)(22)(23). 2 The essential regulatory region upstream of the first open reading frame of the ams operon covers an unusually large region of about 700 bp (21). The transcriptional start site of the homologous cps operon of E. coli was mapped at 340 bp upstream of the translational start site (23). The transcriptional units of EPS operons may therefore contain large leader regions of yet unknown function. We recently reported the binding of RcsA and RcsB proteins from E. amylovora and E. coli to a putative promoter region located between Ϫ578 and Ϫ501 relative to the translational start site of amsG, the first reading frame in the ams operon (24). We could demonstrate the binding of RcsA and RcsB as a heterodimer and that RcsB, but not RcsA, is able to bind alone at higher concentrations.
In this report, we could confine the essential Rcs-binding region, and we present the recognition motif for the RcsA/RcsB dimer at the amsG promoter of E. amylovora. We could furthermore identify an RcsA/RcsB-binding site at the corresponding location in the P. stewartii cpsA promoter. Our results give evidence that the binding of RcsA/RcsB to promoters in EPS operons could be a common principle for triggering the capsule synthesis in the Rcs regulatory pathway. * 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF077292.
¶ To whom correspondence should be addressed.

MATERIALS AND METHODS
Strains, Plasmids, and Growth Conditions-The bacterial strains and plasmids used in this work are described in Table I. Bacterial cells were routinely grown in LB broth at 37°C, and ampicillin was added, if appropriate, to a final concentration of 100 g/ml.
DNA Techniques-Standard techniques such as DNA cloning, DNA analysis, and cell transformations were done as described (25). Sequencing grade plasmid DNA was isolated from strain XL1-Blue. DNA sequencing was done by the chain termination technique (26). The polymerase chain reaction (PCR) was performed with Vent polymerase after optimization for Mg 2ϩ concentration. The P. stewartii rcsA gene was amplified from chromosomal DNA of the P. stewartii wild type strain DC283 by using the following primers: RcsA PS forward, GGGG-ATCCATGCCAACGATTATTATGGATTCC; RcsA PS reverse, GGAAGC-TTCTATCTTACGTTCACGTAAATACCAG. The plasmid pM-RcsA ES was constructed by cloning the rcsA gene into the BamHI/HindIII site of the expression vector pMalc2. The fragment F 183 with the Rcsbinding site of the amsG promoter was amplified from plasmid pEA131 by PCR as described (24). The GenBank TM accession number for the P. stewartii cpsA promoter is AF077292.
Expression and Purification of Proteins-The bacteria were grown in a 10-liter fermenter at 28°C with 90% O 2 saturation and at pH 7.0. LB broth was subcultured (1:200) into fresh medium from an overnight culture, and 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside was added when the A 590 reached 0.5. The cells were incubated for additional 3 h, pelleted by centrifugation, and stored at Ϫ25°C. The RcsA proteins were expressed from plasmids pM-RcsA EA and pM-RcsA ES in strain JB3034 as a C-terminal fusion to the maltose-binding protein. The RcsB proteins were expressed from plasmids pM-RcsB EA with an N-terminal fusion to the maltose-binding protein and pQ-RcsB EA with an N-terminal poly(His) 6 -tag in the strain BL21. The Rcs proteins were purified in a two-step procedure by affinity chromatography using purified starch or metal chelate resin, followed by anionic exchange chromatography as described (24). If applicable, the purified proteins were further designated as M-Rcs proteins when fused with the maltose-binding protein and as H-Rcs proteins when fused with a poly(His) 6 -tag.
Electrophoretical Gel Mobility Shift Assay (EMSA)-DNA labeling with [␣-32 P]dATP, DNA binding assays, and separation of protein-DNA complexes from unbound DNA by native PAGE were performed as described (24). Retardation of DNA fragments by Rcs proteins was monitored by gel electrophoresis and exposure of the dried gels to x-ray films. For quantitative assays, the developed x-ray films were aligned with the corresponding dried gel, the bands of retarded and free DNA were cut out, and the amount of labeled DNA in the gel slices was quantified in a scintillation counter.
In Vitro Selection of DNA Fragments-The 23-bp fragment F 23 , representing the minimal RcsA/RcsB binding region, was fused to a restriction linker and to an approximately 0.2-kilobase pair DNA fragment from the vector pBluescript by PCR to facilitate the selection procedure. The complete sequence of the fragment F 23 was then permutated by generating six DNA pools in which three or four bases of the wild type sequence were replaced by randomized bases. The DNA pools were synthesized by PCR using the following oligonucleotides: S-A, GCAAGCTTAAATTAAGATTATTCTCAANNNNACGGCCAGTGAGC-GCGCGTAATACG; S-B, GCAAGCTTAAATTAAGATTATTCNNNNTA-TAACGGCCAGTGAGCGCGCGTAATACG; S-C, GCAAGCTTAAATTA-AGATTANNNTCAATATAACGGCCAGTGAGCGCGCGTAATACG; with SABC-rev, GCAAGCTTAAATTAAGATTA; and S-D, CGCTGCAG-CGGTATATTGAGAANNNNCTTAATTTGGCGAGTTACATGATCCC-CCATGTTG; S-E, CGCTGCAGCGGTATATTGAGAATAATNNNNAT-TTGGCGAGTTACATGATCCCCCATGTTG; S-F, CGCTGCAGCGGTA-TATTGAGAATAATCTTANNNNGGCGAGTTACATGATCCCCCATG-TTG; with SDEF-rev, CGGGATCCCACTATTCTCAGAATGACTTGG-TTG as primers, and with pBluescript KS ϩ as a template. The primers S-A until S-F contained the fragment F 23 with the randomly substituted positions and a suitable restriction site as a linker. The primers annealed to a specific region of pBluescript KS ϩ and generated with the reverse primers SABC-rev and SDEF-rev DNA fragments of about 0.2 kilobase pairs. The amplified DNA fragments were purified by agarose gel electrophoresis, and about 1 g of each DNA pool was used as target DNA in an EMSA with 1.89 M RcsA and 0.17 M RcsB at optimized conditions. The native polyacrylamide gels were stained with ethidium bromide, and the shifted DNA bands were cut out. The DNA was eluted from the gel slices by shaking overnight in 50 mM Tris-Cl (pH 7.4), 0.5 M sodium acetate, 1 mM EDTA, at 37°C, and the selected fragments were reamplified by PCR using the primers SABC-rev with SABC-for, CCGAATTCCTGCAGCCCGGG, and SDEF-rev with SDEF-for, CGCT-GCAGCGGTATATTGAGAA. The selection was repeated twice, and the DNA fragments were finally cloned into pBluescript KS ϩ and sequenced.
DNA-Protein Cross-link-Reactions were set up as described for the EMSA (24). After 20 min at 28°C, the solution was irradiated 5 cm from the ultraviolet light source for various times (as specified) with 312 nm. The reactions were cooled on ice during illumination. Oligonucleotides labeled with the photoreactive thymine analogue 5Ј-iododeoxyuracil were purchased from TIB-MolBiol/Berlin and reconstituted to doublestranded DNA. The proteins directly interacting with the template were photo-cross-linked to it and resolved by denaturing gel electrophoresis. Cross-links were carried out in combination with competitor -DNA and bovine serum albumin.

RESULTS
Location of an RcsA/RcsB-binding Site in the amsG Promoter-The binding region for the RcsA/RcsB heterodimer was located by making deletions from each terminus of the previously identified 183-bp fragment F 183 and testing the retardation of each in EMSAs. The smallest retarded fragment was 23 bp (F 23 ) (Fig. 1). Further deletions of 5 bp from either end of F 23 either reduced or completely abolished the binding of the E. amylovora RcsA/RcsB dimer (Fig. 1). The 23-bp Rcs binding region is localized from Ϫ555 to Ϫ533 upstream of the translational start site of amsG, the first reading frame of the ams operon (Fig. 2). The identified Rcs-binding site overlapped with the previously proposed promoter region of the ams operon and included the putative Ϫ35 region.
Fragment F 23 was analyzed by mutation for nucleotide positions responsible for the specific DNA-protein recognition. Sets of two or three nucleotides were mutated by nucleotide transitions, and the binding of RcsA/RcsB to the fragments was quantified in EMSAs (  were tolerated by the Rcs proteins, and no or only minor decreases in the DNA retardation were detected. These results gave evidence that a specific recognition by the RcsA/RcsB dimer might be determined by the central region of fragment F 23 , whereas the obvious essential contacts to nucleotides at positions 18 -23 might not be sequence-specific and could occur with the phosphate backbone. The protein-DNA interaction of RcsA/RcsB with nucleotides of the central region of fragment F 23 was verified by UV crosslinking. The thymine bases at nucleotide positions 8, 10, 11, and 13 were substituted by 5Ј-iododeoxyuracil and the modified fragment F 23 was incubated with RcsA/RcsB proteins at optimal DNA-binding conditions. The formation of cross-linked protein-DNA complexes was analyzed by SDS-PAGE after illumination of the DNA binding assay for 30 min at 312 nm ( Fig.  4). A cross-linked band was clearly visible and demonstrated the close contact of the Rcs proteins with at least some of the labeled thymines.
Characterization of the RcsA-RcsB-DNA Complex-We estimated the affinity of the RcsA/RcsB heterodimer for fragment F 183 with increasing concentrations of an equimolar solution of RcsA and RcsB. With this approach, the apparent binding constant of the RcsA/RcsB heterodimer was calculated at about 100 nM. The same result was obtained by using a constant amount of 30 nM RcsB or 2.5 M RcsB with increasing amounts of RcsA starting from 19 nM up to 1.5 M. The usage of different protein ratios did not change the DNA binding kinetics as the binding constant of a heterodimer to DNA should depend only on an equilibrium between the putative free RcsA/RcsB dimer and an RcsA-RcsB-DNA complex. The increase of one protein component will therefore affect the equilibrium between protein dimer and monomers in favor of the RcsA/RcsB heterodimer formation. However, the protein dimer might be rather unstable as we have not been able to detect any RcsA/ RcsB dimer formation neither in the yeast two-hybrid system nor by affinity chromatography of RcsA with the immobilized poly(His) 6 -tagged RcsB (data not shown).
Relatively high protein concentrations of about 1.7 M RcsA with 33 nM RcsB in the EMSA with fragment F 183 of the amsG promoter resulted in an additional supershifted band (Fig. 5). We investigated whether this complex included additional copies of RcsB. The 24-kDa poly(His) 6 -tagged RcsB protein in the protein-DNA complex was subsequently replaced by increasing concentrations of the 68-kDa RcsB protein modified with an N-terminal fusion of the maltose-binding protein. The resulting band pattern shown in Fig. 5 demonstrates that the supershifted protein-DNA complex obviously contains two copies of the RcsB protein. The appearance of the supershift was dependent on the concentration of RcsA and at least one additional copy of RcsA might also be included in the complex. We could not prove this possibility, as the expression of unmodified or poly(His) 6 -tagged RcsA proteins results in the formation of inclusion bodies, and only the RcsA fusion to the maltosebinding protein was produced as soluble protein. Therefore it could not be completely ruled out that the supershifted complex does contain only one copy of RcsA together with two copies of RcsB. In that case, RcsB might be able to bind to the preexisting RcsA-RcsB heterodimer complex already formed with lower concentrations of RcsA. However, the lack of a supershifted band even at high concentrations of RcsA with smaller DNA targets like the fragments F 28 or F 23 of the amsG promoter make this assumption more unlikely. We propose that the supershifted complex represents an additional RcsA/RcsB heterodimer bound to a second less specific region in fragment F 183 .
An important factor for the binding efficiency of the RcsA/ RcsB dimer is the length of the offered target DNA. We determined the retardation of different DNA fragments containing the mapped Rcs recognition site at identical assay conditions in EMSAs by using 7.6 M RcsA and 1.6 M RcsB protein. Relatively high protein concentrations had to be used to obtain substantial amounts of retarded DNA even with the smallest fragment F 23 . The largest target was fragment F 183 , and the amount of retarded DNA was calculated at 52 Ϯ 6%. The retardation of the smaller 28-and 23-bp fragments F 28 and F 23 were clearly decreased to 14.2 Ϯ 8.8 and 10.1 Ϯ 5.4%, respectively. The results demonstrate that additional and nonspecific nucleotides might stabilize the protein-DNA interactions.
Once formed, the stability of the RcsA-RcsB-DNA complex might be one major determinant for the induction of ams expression. We determined the half-life of the complex in a competition experiment with the labeled fragment F 183 as a target and about 30 nM RcsB and 550 nM RcsA, respectively. These conditions were found to be ideal for the retardation of fragment F 183 . The binding assays were first incubated at standard conditions for 10 min to receive an equilibrium between free ligands and the RcsA-RcsB-DNA complex. The assays were then supplemented at varying time intervals starting from 2 s to 1 h with about 30-fold excess of unlabeled fragment F 183 . Finally, all samples were analyzed by native PAGE, and the amounts of protein-DNA complex and unbound DNA were quantified. The results were plotted, and the half-life of the RcsA-RcsB-F 183 complex could be calculated at about 42 s. The Rcs-DNA complex can be considered to be of only low stability if compared with known half-lives of transcriptional repressors to their target DNAs. However, this result is expected for a transcriptional inducer, as the EPS biosynthesis is controlled by environmental signals and the relative low stability of the inductive Rcs-DNA complex might be essential for a fast response to changing conditions.
Determination of the RcsA/RcsB Recognition Motif in the amsG Promoter-The sequence of the identified Rcs binding region in the amsG promoter shows some palindrome elements, which could be important for the recognition by the proteins. However, the binding of a protein heterodimer does not require a palindromic binding motif. To identify nucleotide positions, which are responsible for the specificity of the RcsA/RcsB binding, we analyzed the sequence of fragment F 23 by an in vitro selection approach. The sequence of fragment F 23 was permutated by substitution of 3-or 4-bp-long stretches of nucleotides by randomized nucleotides. The resulting six probes contained random positions in different regions throughout the fragment F 23 and represented a pool of 64 and 256 DNA fragments, respectively. Each mixture was used as target DNA in EMSAs with 1.89 M RcsA and 0.17 M RcsB protein as described under "Materials and Methods." The higher protein concentrations had to be used to obtain sufficient retarded DNA for the visualization by ethidium bromide staining. The retarded DNA fragments were isolated, and the selection was repeated twice. The retarded DNA fraction after the third selection by the RcsA/RcsB proteins was quantified and compared with the retardation efficiencies of the unselected DNA mixtures (Table  II). The retardation of each DNA mixture was enhanced after the selection for at least 200%. The sequence of the nucleotide positions 9 -11 of fragment F 23 was most important for the binding of the RcsA/RcsB proteins as almost no retardation of the corresponding unselected DNA mixture was notable. This result corresponds to our previous findings in the mutation analysis and is also in agreement with the observed UV crosslink of the Rcs proteins to that DNA region. The sequence of the terminal nucleotide positions of fragment F 23 seems to be of minor importance for a specific binding of RcsA/RcsB, as already the unselected pools showed a considerable retardation in the EMSA.
The DNA fragments of the selected pools were cloned into pBluescript KS ϩ , and at least 27 clones of each pool were sequenced (Table III). Corresponding to our previous results, a stringent selection was found at the nucleotide positions 9 -11, and a purine was absolutely required for the positions 10 and 11. Only 5 of 64 possible codons were obtained after selection of The results of the in vitro selection were verified by two different approaches using the EMSA. First we quantified the retardation of representative DNA fragments isolated from the selection procedure (Table IV). The selected DNA fragments (SF) were approximately 200 bp in length and contained fragment F 23 from the amsG promoter with mutations relative to the wild type sequence. As expected, the nucleotides at both ends of fragment F 23 did not show high sequence specificity in the in vitro selection. However, the increase in DNA retardation of the optimized fragment SF (C 1 C 2 C 3 T 4 ) indicates that some selectivity does exist. Substitution of a thymine and an adenine residue in fragment SF (G 5 C 8 ) by guanine and cytosine resulted in an about 2-fold increase in retardation. Fragments SF (G 5 A 7 C 8 ) and SF (G 5 C 7 C 8 ) demonstrate the requirement for a purine in position 7. The thymine at position 6 was shown to be essential in fragment SF (G 5 A 6 C 7 C 8 ). The positions 9 -16 were already optimal in the amsG wild type sequence. The positions 9 and 10 tolerated only purines, whereas guanine was almost strictly required at position 9 and an adenine at position 10 enhanced the retardation about 2-fold. The adenine at position 11 could be replaced with pyrimidines with some decrease of the retardation efficiencies. Substitution of the thymine residue at position 16 by cytosine resulted in only about 25% residual retardation. Replacement of the thymine at position 18 with guanine in fragment SF (G 18 ) did not increase the retardation as expected from the in vitro selection. Position 17 required a pyrimidine, and an adenine considerably diminished the retardation as shown with fragments SF (C 17 G 18 ) and SF (A 17 G 18 ). Position 19 might be specific for purines as the replacement of adenine by thymine in fragment SF (G 18 T 19 ) only yielded about 25% residual retardation.
In a second approach, we analyzed the retardation of 23-bp DNA fragments reconstituted from oligonucleotides and designed according to our consensus motif (Table V). Fragment F 23 (A 1 C 2 T 4 G 5 C 8 ) contained optimized substitutions at the 5Јend, and the retardation in EMSAs was enhanced about 2-fold. This result does completely agree with our observations with fragments SF (C 1 C 2 C 3 T 4 ) and SF (G 5 C 8 ) (Table IV) first reading frame of the cps operon, was aligned with the sequence of the amsG promoter (Fig. 2). The nucleotide positions Ϫ538 to Ϫ516 correspond to the Rcs binding region of the amsG promoter, and they were analyzed as a 28-bp fragment in EMSAs with the RcsB protein of E. amylovora and the RcsA protein of E. amylovora (RcsA EA ) or P. stewartii (RcsA ES ). The fragment was clearly retarded with both protein combinations (Fig. 6), giving evidence that the binding of an RcsA/RcsB heterodimer to a region from about Ϫ510 to Ϫ540 bp upstream of the translational start site might be a common principle in the regulation of EPS biosynthesis in Erwinia. The retardation with RcsA ES compared with RcsA EA seems to be somewhat better with the fragment F 23 of the homologous cpsA promoter. Vice versa, RcsA EA seems to be more effective in the retardation of the fragment F 23 from the amsG promoter. This gives evidence that the RcsA proteins are involved in the specific DNA recognition. According to the consensus motif of the amsG promoter, the RcsA/RcsB-binding site of the cpsA promoter was degenerated in two essential positions (Table III) and did not contain palindromic elements. The replacement of the degenerated adenine by the conserved guanine in fragment F cpsA-(G9) resulted in about a 5-fold enhanced retardation in the EMSA (Table V). Despite the degenerated positions, the analyzed 28-bp fragment from the cpsA promoter showed a similar retardation compared with fragment F 23 from the amsG promoter (Table V). The degeneration might therefore be compensated by optimized nucleotides at other positions of the cpsA Rcs binding motif. DISCUSSION Genetic data gave evidence that RcsA/RcsB heterodimers play a major role in mediating transcriptional activation on positively regulated genes involved in EPS biosynthesis. The principal aim of this study was to identify essential nucleotide positions implicated in Rcs protein recognition within promot-ers of EPS biosynthetic operons. We previously described a synergistic binding of RcsA and RcsB to the promoter of amsG, the first reading frame of the E. amylovora operon for amylovoran biosynthesis. The minimal DNA fragment retaining substantial RcsA/RcsB binding has been confined to a 23-bp region now. Both gel mobility assays and protein-DNA cross-linking studies indicate the interaction of purified RcsA/RcsB proteins with that region. The RcsA/RcsB binding region covers the putative Ϫ35 consensus of the amsG promoter (21). Its relatively poor homology to E. coli 70 promoters (27) is consistent with RcsA/RcsB acting as activators for the amsG promoter. The confined minimal RcsA/RcsB-binding site is small compared with those covered from transcriptional regulators of other two-component systems. Footprinting revealed protected DNA fragments of about 80 bp for VirG and VanR (28,29). Additional sequences might also be bound by the RcsA/RcsB proteins as longer DNA fragments stabilized the protein-DNA complex more than 5-fold, but the analyzed 23-bp fragment appears to contain the nucleotide positions responsible for the RcsA/RcsB binding specificity.
The affinity of the RcsA/RcsB heterodimer to its target sequence in the amsG promoter was determined with an apparent K D value of 100 nM. It has to be considered that the determination of the K D values is based on the assumption that the activities of the protein fractions are 100%. In other well characterized bacterial two-component systems, e.g. NtrB/NtrC (30), EnvZ/OmpR (31), PhoR/PhoB (32), ComP/ComA (33), and VanS/VanR (29), phosphorylation of the response regulators increased affinity, presumably due to the formation of higher oligomeric protein states and cooperative binding (34,35). RcsB contains a well conserved phosphorylation motif, and its activity might be modulated upon phosphotransfer by RcsC (36). The percentage of phosphorylation in our RcsB preparations was unknown, but it might be rather low due to the kinetic lability of aspartyl-phosphate linkages. It is therefore possible that the K D values could be considerably increased after an increased phosphorylation of RcsB. An apparent K D value of 40 nM was reported for the transcriptional regulator VanR after increasing the percentage of phosphorylated protein with acetyl phosphate to a total of about 8%, whereas the K D value of unphosphorylated VanR was found to be about 500-fold lower (29). Both forms of VanR bind to identical DNA regions, but the phosphorylated VanR covers a larger part of DNA. Additionally, the K D value of 14 nM for the unphosphorylated form of the transcriptional regulator NtrC could be increased to a K D value of 1 nM upon phosphorylation (37). However, the affinities of the phosphorylated procaryotic transcriptional activators ComA and OmpR to their DNA targets with K D values of 1 M (33) and 1.5 M (31), respectively, are significantly lower than the affinity of RcsB to its target. The RcsA-RcsB-DNA complex shows a relative short half-life of 42 s, and there     (7), but the determined RcsA/RcsB recognition motif at the amsG promoter did not show extensive similarities to the lux box, a 20-bp palindrome similar to the LexA repressor recognition motif (38). Palindromic elements were not necessary in the recognition motif derived by in vitro selection, and they are absent in the RcsA/RcsB-binding site at the P. stewartii cpsA promoter. The in vitro selection revealed a region of about 13 bp as most important for the RcsA/RcsB binding specificity. The cross-link of 5-iododeoxyuracil-substituted thymines within this area demonstrated its close contact to the Rcs proteins. The binding mechanism of RcsA/RcsB to their target DNA seems to include a complex combination of recognition of specific bases in concert with local structural features. Proposed local single or multiple "up" mutations according to our consensus motif within the 23-bp fragment always increased the binding of RcsA/RcsB. However, the binding was decreased upon combination of up mutations from more distantly located areas, indicating that the tight binding of one Rcs protein at the 23-bp fragment requires some flexibility of the second Rcs protein.
Heterodimerization of transcriptional regulators is quite uncommon in procaryotic systems but prevalent in eucaryotic regulation mechanisms (39), and one advantage might be the recognition of additional DNA targets. RcsB, but not RcsA, is essential for the stimulation of EPS biosynthesis, and it is involved in the regulation of further RcsA-independent pathways like the stimulation of Vi polymer synthesis in Salmonella typhimurium (12) and ftsZ expression in E. coli (40). It might therefore be speculated that RcsB is responsible for the initiation of transcription, possibly by interacting with RNA polymerase, and the alternative association with specific coinducers like RcsA might enable RcsB to bind and regulate different promoters in the cell. It is yet uncertain if RcsB first binds weakly to DNA and the complex is then recognized by RcsA, or if a preformed RcsA/RcsB heterodimer binds to the      recognition site. Evidence for the latter possibility could be the appearance of the supershifted band, notable after increasing the concentration of RcsA to about 40-fold excess. The increased RcsA concentration could enhance the formation of RcsA/RcsB heterodimers, resulting in the binding of secondary sites with lower affinity. However, the supershifted band would also be consistent with the formation of a nucleoprotein complex and a secondary binding of RcsB and possibly RcsA to the already bound RcsA/RcsB heterodimer (41). The presence of a free heterodimer is further supported by the observation that overexpression of the C-terminal deleted RcsB protein inhibits EPS biosynthesis in E. coli, presumably by competition with the wild type RcsB for the RcsA protein, 3 resulting in the formation of a nonproductive heterodimer. However, the heterodimers of RcsA/RcsB might associate only weakly, as we could not detect an interaction in the yeast two-hybrid system or by affinity chromatography of RcsA with immobilized RcsB. 3 In summary, the identification of RcsA/RcsB-binding sites at corresponding locations in the main promoters of the EPS operons of E. amylovora and P. stewartii gave evidence for a conserved regulation mechanism for type IA polysaccharide biosynthesis. The description of critical nucleotides in the RcsA-RcsB-DNA complex should help to analyze further the molecular interactions in the regulation of capsule biosynthesis by Rcs proteins.