INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Ralstonia (Pseudomonas) solanacearum (42, 43) is one of the most troublesome prokaryotic phytopathogens in the world (13, 14). It infects plants via wounds or cracks at the emergence point of lateral roots (37). It subsequently spreads into the stem via the vascular system (40, 41), where populations reach >10¹⁰ cells per plant, concomitant with wilting and death of the host. R. solanacearum produces a large variety and amount of extracellular products that contribute to disease (34). One of the most important of these is the unusual extracellular polysaccharide EPS I, a large acidic polymer comprised of N-acetylgalactosamine and two derivatives thereof (25, 32). EPS I is required by R. solanacearum for efficient wilting and killing of plants, probably because it restricts water flow in the xylem (5-7). EPS I is also required for efficient and rapid colonization of the plant's vascular system (28). Production of EPS I requires the products of the 16-kb eps gene cluster (6), which appear to be transcribed from a single promoter into a large polycistronic RNA (19). DNA sequence analysis suggests that the eps operon encodes >12 polypeptides directly involved in biosynthesis and export of EPS I (19).

Because it must also survive in the soil outside of a plant host, R. solanacearum has evolved a sophisticated network for controlling expression of genes encoding production of EPS I and its other virulence factors, such as plant cell wall-degrading exoenzymes (4, 18). The network contains at least three distinct signal transduction arrays, each containing a unique two-component system (15, 26), comprised of a membrane-bound kinase sensor and a response regulator (Fig. 1). It is likely that each sensor (VsrA, VsrB, or PhcS) responds to a different environmental signal by phosphorylating its cognate response regulator (VsrD, VsrC, or PhcR), which in turn activates or represses promoters of appropriate virulence genes. While maximal transcription of eps requires all components of the network to be active, previous experiments (18) showed that the control of the eps promoter affected by both the VsrAD and PhcSR/PhcA signal transduction arrays is indirect, because the reduced eps transcription caused by inactivation of either or both of these systems can be overcome by constitutively expressed xpsR. This observation coupled with the fact that the VsrAD and PhcSR/PhcA systems control transcription of xpsR (3, 4, 10, 18) led to the proposals that VsrAD and PhcA control levels of XpsR protein in a signal-dependent manner and that XpsR acts as a signal integrator which in concert with the VsrBC system controls transcription from the eps promoter (35). How XpsR and VsrC activate transcription is unknown. While mutagenesis of the eps promoter has identified a single 20-bp region around position 70 that is essential for activation by both XpsR and VsrC (36), attempts to demonstrate binding of XpsR to the eps promoter have been unsuccessful, suggesting that like some eukaryotic transcription factors, XpsR may bind to another regulatory protein that itself is bound to the eps promoter.

View larger version (16K): [in this window] [in a new window]   FIG. 1.   Organization of the regulatory network controlling eps and other virulence genes of R. solanacearum. Signal transduction pathways through the three known two-component systems, VsrAD (18, 33), VsrBC (17, 18), and PhcSR (4), that in conjunction with phcA (2) and xpsR control transcription from the eps biosynthetic operon promoter are shown. Rectangles represent membrane-bound kinase sensors; circles represent response regulators. Arrows indicate positive transcriptional control. Closed arrowheads indicate hypothesized, signal-induced phosphoryl transfer between sensory kinase and response regulator; open arrowheads indicate positions of environmental signal input and recognition. 3-Hydroxypalmitic acid methyl ester (3-OH PAME) is proposed to be the signal perceived by the PhcSR system (4, 10); available evidence suggests that PhcR affects phcA expression indirectly. Signals perceived by VsrA and VsrB are unknown.

xpsR is the first of four genes in a 5-kb operon (8, 9, 18) whose transcription appears to be directed by a promoter found directly upstream of xpsR. Although none of the four open reading frames (ORFs) in the operon has any obvious motifs or homologs in sequence databases, their functions have been explored by analysis of site-directed mutants. The first ORF, XpsR, is a 33-kDa positive regulator of eps transcription (18); the second ORF (58 kDa) is Tek, which is processed into a major extracellular protein that is associated with EPS I (8); the third is Ert, a 44-kDa ORF (formerly called Region II or RgnII [6, 32]) which plays an ill-defined and conditional role in EPS I synthesis (9); the fourth ORF is a homolog of Tek. With the exception of XpsR, none of the ORFs appears to be absolutely required for virulence (6, 8, 9). To better understand the function of these genes and how levels of their products are transcriptionally regulated, we used different types of mutagenesis to investigate where and how PhcA and VsrD might interact with the xpsR promoter and found evidence that they use an atypical mechanism to regulate transcription from this promoter.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacteria, plasmids, and media. Descriptions of the R. solanacearum strains and plasmids used are shown in Table 1; for maps of plasmids, see Fig. 2 and 5. The Escherichia coli host strain used for most recombinant DNA manipulations was DH5 (12). Vectors used were pTZ18U/pTZ19U (22), pRK415 (20), and pRG970 (39). R. solanacearum and E. coli were grown at 30°C in 1% peptone-0.1% Casamino Acids-0.1% yeast extract with 0.5% glucose or 0.5% sucrose (BG medium) and at 37°C in LB medium (23), respectively. Antibiotics were used at 50 µg/ml for kanamycin, 100 µg/ml for ampicillin (20 µg/ml for R. solanacearum), 50 µg/ml for spectinomycin, and 25 µg/ml for tetracycline.

Construction of P_xpsR-lacZ fusions. pRLZ7 and pTZRLZ7 were constructed by inserting the 2.7-kb BamHI fragment of pCB5 (6) into BamHI-digested pRG970 (39) and pTZ18U, respectively. pRLZ3 and pTZRLZ3 were constructed by inserting the 610-bp BamHI fragment of pJH161 (18) into BamHI-digested pRG970 and pTZ18U, respectively. The primer pairs T7/RP2, RP7/RP2, RP3/RP2, RP4/RP2, RP1/RP2, and RP5/RP2 were used in PCR with pTZRLZ3 as template to generate the xpsR promoter fragments with various lengths of upstream regulatory sequences used in construction of pRLZ1-pRLZ8. One-hundred-microliter PCRs contained 1× reaction buffer (Perkin-Elmer), 1.5 mM MgCl₂, 200 µM deoxynucleoside triphosphates (dNTPs), 2.5 U of Amplitaq polymerase (Perkin-Elmer), 0.4 µM primer pair, and 100 ng of pTZRLZ3. After denaturation for 4 min at 95°C, 20 thermal cycles (1 min at 94°C, 1 min at 50°C, and 2 min at 72°C) were performed. PCR products were extracted with phenol-chloroform, precipitated with ethanol, digested with BamHI or BamHI-BglII, gel purified, ligated into BamHI-digested pRG970, and transformed into E. coli. Plasmids from transformants that were blue on LB plates with spectinomycin and 5-bromo-3-chloro-indolyl--D-galactopyranoside (X-Gal) were isolated, and orientation was confirmed by restriction enzyme analysis. Primers included T7 (5'-TAATACGACTCACTATAGGG-3'), RP1 (5'-CAGGATCCCCGGCTGTGCGACAT-3'), RP2 (5'-TAGGATCCATTGAATCCGTGCAAC-3'), RP3 (5'-CCGGATCCTACGCGTTCAGATATG-3'), RP4 (5'-GAGGATCCCAACACGCTGCTTTAC-3'), RP5 (5'-GAAGATCTATCTTTACTCTCCTTTA-3'), RP7 (5'-GAGGATCCGA TTCGTTTTTTTCTTG-3'), RP8 (5'-GAGGATCCGAATTCCGACAAATCCTGGCTAAATTGGGGAT-3'), RP9 (5'-GAGGATCCGAATTCCGACAGTACCCCCCTAAATTGG-3'), and RP10 (5'-GAGGATCCGAATTCCGACAAATCCCCCHATTGGGGATTCGTT-3'); underlining indicates altered nucleotides.

Mutagenesis of putative binding sites. To alter the PhcA binding site, two P_xpsR regions, 338 to 83 and 76 to +36, were amplified by PCR using pTZRLZ3 as template and primer pairs T7/RP6 and RP5/RP2, respectively. Since both RP5 (see above) and RP6 (5'-GAAGATCTACATCACGCCAGCTTTG-3') contain a 5' BglII site, ligation of the two BglII-digested PCR fragments together changes nucleotides 82 to 77 from TAAAAA to AGATCT. The ligation products were digested with BamHI and ligated into BamHI-digested pRG970 to give pRMB. To construct pRMB derivatives, PCR fragments generated by primer pairs RP1/RP2 and RP3/RP2 with pRMB as template were digested with BamHI and ligated into BamHI-digested pRG970 to obtain pRMB-S and pRMB-M, respectively. P_xpsR fusion plasmids with alterations in the putative VsrD binding site were generated by PCR amplification of sequences between nucleotides 338 and +36 by using a pRLZ1 template and primer pairs RP8/RP2, RP9/RP2, and RP10/RP2. Resultant PCR fragments were digested with BamHI and ligated into BamHI-digested pRG970 to obtain pRLZ15, pRLZ17, and pRLZ16, respectively.

Insertional mutagenesis of P_xpsR. pRLZ11 containing a 4-bp insertion between the VsrD and PhcA binding sites of P_xpsR was constructed in three steps: (i) the 374-bp BamHI fragment of pTZRLZ1 was inserted into BamHI-digested pUC9 to give pUCRLZ1; (ii) pUCRLZ1 was linearized with XbaI, filled-in with four dNTPs and Klenow enzyme, and recircularized to give pUCRLZ11; and (iii) the BamHI fragment of pUCRLZ11 was ligated into BamHI-digested pRG970. pRLZ12 and pRLZ13 were constructed by ligating the 114-bp SmaI fragment of pJH161 (18) with pUCRLZ1 which had been digested with XbaI and whose ends had been filled in with four dNTPs and Klenow enzyme followed by recloning of the BamHI fragments of the resultant plasmids into BamHI-digested pRG970. pRLZ12 contains a single 114-bp SmaI fragment; pRLZ13 contains two tandem 114-bp SmaI fragments.

Purification of PhcA and use in mobility shift DNA binding analyses. Five-hundred-milliliter cultures of E. coli BL21 DE3 containing the pET3d vector (38) or the phcA overexpression plasmid pET3231 (18) were grown to an A₆₀₀ of 0.3 at 37°C and then transferred to a 25°C shaker. After 10 min, isopropylthiogalactoside was added to 0.5 mM to induce overexpression of phcA and shaking was continued for 4 h. Cells were harvested, washed once with buffer A (10 mM Tris-HCl [pH 7.0], 25 mM KCl, 2 mM mercaptoethanol, 1 mM phenylmethylsulfonylfluoride), resuspended in 5 ml of buffer A, and sonicated for 3 min at 4°C. Broken cells were centrifuged at 15,000 × g for 20 min. Ammonium sulfate (1.4 g) was added to the resultant supernatant (5.5 ml), and after 1 h at 4°C, precipitated proteins were removed by centrifugation (12,000 × g, 10 min). An additional 0.32 g of ammonium sulfate was added to the supernatant, and after 1 h at 4°C, precipitated proteins were recovered by centrifugation, redissolved 1 ml of buffer A, and dialyzed extensively against buffer A. After any precipitate was discarded, the sample was applied to a 1.5-ml phosphocellulose column equilibrated with buffer A. After the column was washed with buffer A with 0.18 M KCl, PhcA was eluted with buffer A with 0.25 M KCl.

Molecular genetic techniques. The methods used for preparation, analysis, and manipulation of recombinant DNA, fragment purification, CaCl₂-mediated transformation, and electrotransformation were standard (17, 18, 21).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Two distinct upstream segments of the xpsR promoter region are required for transcriptional regulation. Previously (18), we showed that (i) transcription of xpsR is activated >10-fold by PhcA and may involve binding of PhcA to the xpsR promoter; (ii) PhcA-activated transcription of xpsR can be further increased >5-fold by the VsrAD two-component system; and (iii) a phcA mutation eliminates VsrAD activation of xpsR without affecting vsrAD expression, implying that VsrAD alone cannot activate xpsR transcription. However, where and how each protein acts to effect full transcriptional activation was not defined. Therefore, the xpsR promoter region (P_xpsR) and various lengths of upstream sequences were joined to a promoterless lacZ gene on pRG970 (41), creating a series of reporters with transcriptional fusions of xpsR to lacZ. Activation of P_xpsR by PhcA or VsrD was assessed by comparing -galactosidase levels in wild-type, phcA mutant, and vsrD mutant strains of R. solanacearum containing each reporter plasmid. Transcription of lacZ directed by a fragment with P_xpsR containing 338 bp of sequence upstream of the xpsR transcription start site (pRLZ1) was normally regulated (Fig. 2): inactivation of vsrD reduced expression of the P_xpsR::lacZ fusion 12-fold, while inactivation of phcA reduced expression 67-fold. P_xpsR expression levels in phcA mutants were not further reduced by additional inactivation of vsrD (data not shown), confirming that phcA is epistatic to vsrD. An analogous fusion plasmid with extended (>2-kb) sequences upstream of P_xpsR (pRLZ7) (Table 1) gave the same P_xpsR expression levels as pRLZ1 and showed wild-type regulation (data not shown).

View larger version (17K): [in this window] [in a new window]   FIG. 2.   Identification of PxpsR sequences required for transcriptional regulation by VsrD and PhcA. The xpsR promoter and various lengths of upstream sequences were fused to lacZ on pRG970 to generate plasmids pRLZ1 to pRLZ8. PxpsR expression (i.e., transcription directed by each PxpsR fragment) was monitored in the wild type (WT), strain AW-D5 (vsrD mutant), and strain AW1-80 (phcA mutant) of R. solanacearum by measuring LacZ activity (-galactosidase activity in Miller units) as described in Table 2, footnote a. Nucleotide numbering is relative to the transcription start site of xpsR (+1|). Striped and hatched boxes and associated sequences illustrate putative binding sites for VsrD and PhcA, respectively. The symbol + above the putative VsrD binding site dyad indicates the positions of mutations that eliminated VsrD activation of PxpsR.; N6 represents the sequence TAAATT. Underlined nucleotides in the PhcA binding site indicate the T-N11-A motif found in the binding sites of nearly all LysR-type activators (31). Asterisks indicate the positions of mutations that affected PhcA binding and activation of PxpsR (Table 3). The complete xpsR promoter region sequence is GenBank under accession no. U18136. pRLZ7 (Table 1) had PxpsR activity similar to that of pRLZ1.

Identification of critical nucleotides required for VsrD activation of P_xpsR. To test the hypothesized involvement of the palindromic sequence between nucleotides 327 and 307 in VsrD activation and more accurately define a putative binding site for VsrD, we used PCR to make three different mutations that destroyed its dyadic structure or spacing and then tested the effect on P_xpsR activation. A derivative of pRLZ1 (338 to +36 of xpsR fused to lacZ) with the CCC between 322 and 320 changed to TGG (pRLZ15) had a P_xpsR expression level that was more than sixfold lower than that of the wild type and was not activated by VsrD (Table 2). Similar results were obtained with a derivative that had the distance between two halves of the dyad shortened by 4 bp (i.e., nucleotides 319 to 316 [CTAA] were deleted; pRLZ16) (Table 2). However, when the nucleotides between positions 327 and 325 (AAT) were changed to GTA, there was no effect on P_xpsR expression or VsrD activation. In conjunction with the results of the deletion experiments described above, these results strongly suggest that the P_xpsR sequences located between 324 and 310, and more specifically the CCCC-N6-GGGG dyadic sequence (322 to 309) (Fig. 2), are critical for activation of P_xpsR by VsrD. This may be the site where VsrD binds.

View this table: [in this window] [in a new window]   TABLE 2.   Effect of site-directed mutations in the 315 palindromic region of the xpsR promoter on expression and regulation

Identification of nucleotides involved in PhcA-mediated activation of P_xpsR. A 14-bp site centered around nucleotide 77 of P_xpsR (Fig. 2) was previously proposed as a binding site for PhcA (18) because of its conformity with the consensus structure and location of binding sites for LysR-type activators that are similar to PhcA (31). Consistent with this hypothesis, pRLZ6 with nucleotides 76 to +36 of xpsR fused to lacZ, and hence lacking half of this site, showed no activation by PhcA while an xpsR promoter fragment with the putative binding site and only 40 bp of additional upstream sequence (nucleotides 117 to +36; pRLZ2) (Fig. 2) fused to lacZ, showed a sixfold-higher activation of expression by PhcA. When an xpsR promoter fragment with the putative binding site and an additional 160 bp of upstream sequence was fused to lacZ (236 to +36; pRLZ5), a twofold-higher activation of P_xpsR by PhcA was observed. While this suggests that P_xpsR sequences between nucleotides 236 and 117 can enhance or assist in PhcA activation mediated via the 77 site, they are clearly not required for PhcA to function at P_xpsR.

View this table: [in this window] [in a new window]   TABLE 3.   Effect of mutations in the 77 region of the xpsR promoter on its regulated expression and PhcA binding activity

View larger version (44K): [in this window] [in a new window]   FIG. 3.   Gel mobility shift assays of PhcA binding to 117 to +36 xpsR promoter fragments with mutations in the PhcA binding site. (A) DNA fragments containing the 117 to +36 sequences of wild-type PxpsR (WT) or PxpsR with the indicated sequence alterations were isolated, labeled with [-32P]dATP, and incubated with 0 to 20 µg of purified PhcA protein or with 20 µg of a similar protein preparation from E. coli lacking phcA (20). Reaction mixtures were electrophoresed, and the dried gels were subjected to autoradiography as described previously (30); quantitative results were obtained by phosphorimaging and are summarized in Table 3. (B) A labeled PxpsR DNA fragment lacking sequences upstream of nucleotide 77 ( > 77) was analyzed as described for panel A. The symbol > to the left of the gel indicates the position of retarded (bound) fragment.

Effect of upstream sequences on expression from P_xpsR with PhcA binding site mutations. Previous genetic analysis (18) suggested that the VsrA/VsrD two-component system by itself cannot activate P_xpsR but rather acts in conjunction with PhcA. To confirm this, we assessed the effect of PhcA binding site mutations on in vivo transcription directed by fragments with the complete xpsR promoter region (338 to +36), which included the putative VsrD binding site. Because phcA is required for P_xpsR activation (Fig. 2) and because PhcA binding site mutations reduced PhcA activation of and binding to the 117 to +36 P_xpsR fragment (Table 3 and Fig. 3), we expected there would be little or no PhcA-mediated activation of complete P_xpsR fragments with the same mutations. However, the P_xpsR-lacZ fusion constructs with mutant PhcA binding sites (pRMB and pRM7) showed at least 20-fold PhcA-mediated activation (Table 4). Deletion of sequences upstream of 286 (i.e., those containing the putative VsrD activation or binding site) from the fusion plasmids strongly reduced, but did not eliminate, P_xpsR expression (pRMB-M and pRM7-M; Table 4). This shows that even in the absence of VsrD interactions, these larger P_xpsR fragments with PhcA binding site mutations can be activated >3.5-fold by PhcA, much higher than the 0.5-fold activation observed when the same mutant xpsR promoters were harbored on the smaller (117 to +36) fragments. Thus, the sequences between 117 and 286 restored most, but not all, of the PhcA-mediated activation to xpsR promoter fragments with mutations in the PhcA binding site at nucleotide 77. Nonetheless, in all situations examined, full activation still required the presence of VsrD and its putative binding site.

View larger version (73K): [in this window] [in a new window]   FIG. 4.   Gel mobility shift assays of PhcA binding to complete xpsR promoter fragments with mutations in the 77 PhcA binding site. (A) Labeled DNA fragments containing PxpsR sequences from 338 to +36 and having the indicated sequence alterations were prepared and incubated with various amounts of purified PhcA or with 20 µg of a similar protein preparation from E. coli lacking phcA (20). (B) DNA fragments containing PxpsR sequences from 338 to 83 were labeled and either used directly as a binding substrate or first digested with XbaI (+Xba) before incubation with purified PhcA or control 20 as for panel A. Binding was detected and quantified as described in the legend to Fig. 3; results are summarized in Table 4. WT, wild type. The symbol > to the left of the gels indicates the position of retarded (bound) fragment.

Effect of increased separation between the putative VsrD and PhcA binding sites on P_xpsR expression. The above data suggest that when VsrD protein binds to a site around 315, it affects PhcA function at a site many helical turns downstream. To investigate how critical this spacing is to VsrD/PhcA-mediated activation, we inserted different lengths of spacer DNA (4 to 232 bp) into the XbaI site at 183 of wild-type P_xpsR on reporter plasmid pRLZ1 (338 to +36 of P_xpsR fused to lacZ) and assayed the effect on phcA- and vsrD-mediated activation of transcription (Fig. 5). No insertion caused a change in P_xpsR expression or regulation of greater than 60%. This effect is insignificant compared to effects of the deletion and substitution mutations described above. Thus, within these limits, the upstream location and distance of the putative VsrD binding site from the PhcA binding site are not critical for normal transcription activation and regulation of P_xpsR.

View larger version (17K): [in this window] [in a new window]   FIG. 5.   Effect of spacer insertions between the putative VsrD and PhcA binding sites on expression of PxpsR. Various lengths (4, 118, and 232 bp) of spacer DNA were inserted into the XbaI site (at 183) in PxpsR on fusion plasmid pRLZ1 (Fig. 2) to create plasmids pRLZ11 to pRLZ13. Expression directed from each PxpsR derivative was monitored in wild-type (WT), vsrD mutant (AW-D5), and phcA mutant (AW1-80) strains of R. solanacearum by measuring LacZ activity as described in Table 2, footnote a. Striped and hatched boxes represent binding sites for VsrD and PhcA, respectively.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

xpsR plays a central role in the network that regulates production of the virulence factor EPS I by R. solanacearum (Fig. 1). Previous results (18) implied that the levels of XpsR protein (in conjunction with the response regulator VsrC) determine transcription levels of the eps biosynthetic operon. In turn, levels of XpsR protein are transcriptionally controlled by two independent, signal-responsive regulatory genes: vsrD, encoding a response regulator, and phcA, encoding a global, LysR-type regulator. Increased expression or activity of PhcA, which occurs in response to a cell density signal (3-OH palmitic acid methyl ester) transduced via the PhcSR two-component system (3, 4, 10), enhances transcription of xpsR sixfold. Activation of VsrD (presumably after its phosphorylation by the VsrA sensory kinase in response to an unknown signal) further increases xpsR transcription 12-fold (Fig. 2). Thus, input from two independent sources is summed and transduced into a single output via XpsR.

Here we focused on the mechanism by which the individual signal inputs transmitted through the PhcA and VsrAD systems are summed at P_xpsR (Fig. 1). Deletion experiments showed that removal of P_xpsR sequences between nucleotides 338 and 310 eliminated VsrAD-dependent activation of P_xpsR and reduced, but did not eliminate, PhcA-mediated transcription activation. This suggested that PhcA itself can partially activate P_xpsR and that VsrD binding to the 338 to 310 region enhances PhcA-mediated activation. However, it is possible that VsrD does not directly bind here but rather controls production or activity of a regulator that does. Nonetheless, two site-directed mutations that altered the structure of a palindrome centered around 315 (Fig. 2) destroyed activation of P_xpsR by VsrD but not the activation mediated by PhcA alone. These results are consistent with the hypothesis that VsrD interacts with the CCCC-N6-GGGG dyad between nucleotides 322 and 307 of P_xpsR to enhance its activation by PhcA.

Further deletion of sequences downstream of the putative VsrD binding site reduced P_xpsR expression only 2.5-fold. However, when sequences between 117 and 76 were removed, a total loss of PhcA-mediated activation was observed. Moreover, we found that when sequences between 117 and +36 were fused to a reporter, they were sufficient to allow activation of P_xpsR by PhcA, implying that these sequences contain an independently functioning PhcA binding and activation site. A likely site centered around 77 was confirmed and further defined by analysis of the effects of mutations in the site on in vivo expression from P_xpsR and in vitro binding of PhcA to P_xpsR. Mutations in the hypothesized 77 PhcA binding site caused a dramatic reduction in the ability of PhcA to bind to and activate transcription from P_xpsR fragments lacking sequences upstream of 117. However, when the mutated PhcA binding sites were placed on complete P_xpsR fragments (338 to +36), high levels of in vivo activation and regulation by PhcA (and VsrD) were observed. Gel shift analyses showed that PhcA binding to these larger fragments was markedly enhanced, although the binding affinity of PhcA for them was still much less than that for wild-type fragments. This implied the existence of another PhcA binding site that can partially suppress the mutations in the primary 77 site. Consistent with this hypothesis, additional DNA binding studies tentatively confirmed the presence of a weak and highly cooperative PhcA binding site between nucleotides 240 and 183. In this region is found an 8-bp sequence (194 to 187; AATCPyTTA) that exactly matches one-half of the 77 PhcA binding site; the role of this sequence in PhcA activation requires confirmation by site-directed mutagenesis.

These results suggest a model where active PhcA binds to a 14-bp site centered around nucleotide 77 of P_xpsR (Fig. 2) and, as a direct result, increases P_xpsR transcription at least sixfold. Additional, but less critical, interactions of PhcA with the second site near 190 may stabilize binding and somewhat enhance activation (compare pRLZ2 and pRLZ5) (Fig. 2). Much more important for maximal activation of P_xpsR by PhcA is the interaction of signal-activated VsrD with a dyadic site much farther upstream. It is plausible that VsrD (or possibly its surrogate) binds to this dyad centered around position 315 and may increase or stabilize binding of PhcA and/or RNA polymerase. Whatever the mechanism, the observation that PhcA is a prerequisite for any significant expression of P_xpsR suggests that PhcA can be thought of as an on-off switch which in the on mode activates a low level of transcription from P_xpsR and also allows the possibility of high-level transcription. The extent to which this low-level transcription is turned up is governed by the VsrAD system acting like a rheostat or volume control.

Our experiments imply that VsrD and PhcA (and/or RNA polymerase) interact, even though they are separated by >12 helical turns. Moving the putative VsrD binding site 232 bp upstream did not dramatically affect P_xpsR transcription activation or regulation. There are only a few examples of promoters like P_xpsR which have distant and separation-insensitive cis-acting sites that bind different regulators (11). By analogy to two of these systems (NtrC and NifA [16, 27]), it is plausible that the DNA between the putative VsrD and PhcA binding sites at P_xpsR may become bent, facilitating contacts between the regulatory proteins and/or RNA polymerase. Similar to promoters recognized by NtrC and NifA, additional regulatory proteins such as IHF may assist in DNA bending at P_xpsR. Interestingly, there is a consensus IHF binding site (GATCAA-N4-CTG; 239 to 227) (16) between the putative VsrD and PhcA binding sites which requires investigation.

The marked dependence of PhcA on signal-activated VsrD for full transcription activation of P_xpsR is unusual, since most other LysR-type regulators activate transcription independent of additional activators (31). PhcA also differs from LysR-type regulators in that it may not require a coinducer and has an unusual hydrophilic tail, a 25-residue, C-terminal extension largely comprised of highly polar residues (2). PhcA also regulates many other target promoters (35), and it will be important to determine if its activation of these is also modulated by additional regulators. Our data suggest that VsrD alone cannot activate P_xpsR but rather must work through or with PhcA. In contrast, most other response regulators can directly turn on transcription of their target promoters (15). Moreover, few cases where cooperative regulation of a promoter by a response regulator and a different type of transcriptional activator have been documented (11, 15). Thus, VsrD may be a mechanistically unusual response regulator; whether its behavior at other promoters is as atypical remains to be seen.