INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Bacillus subtilis, a gram-positive bacterium, frequently encounters stress conditions in its natural environment, the soil. In order to survive, it has developed, through the course of evolution, several ways of detecting and responding to these pressures. One of the ways of accomplishing both is through the use of two-component signal transduction systems. Two-component systems are found in numerous bacteria and are composed of two main regulatory proteins, (i) the histidine-sensor kinase and (ii) the response regulator (22, 26, 27). When the histidine kinase senses a specific stimulus from the environment, it undergoes ATP-dependent autophosphorylation and then passes the phosphate to its cognate response regulator. The activated response regulator is able to elicit a response by interacting with specific proteins or by binding a specific set of promoters in order to activate or repress their transcription.

When B. subtilis is starved for phosphate, several genes of the phosphate or Pho regulon are activated or repressed through the coordination of the two-component regulatory proteins PhoP and PhoR (1, 3, 10, 11, 14, 24). These proteins have homology to DNA-binding response regulators and histidine kinases, respectively. Genetic evidence suggests that, of the three different response regulators previously shown to be involved in the Pho response, PhoP is the furthest downstream in the signaling pathway (7, 28).

Many of the Pho regulon genes and their promoters have been characterized through promoter fusion assays, in vitro transcription assays, and DNase I footprint analyses. These genes include phoA and phoB (9, 15, 17), which encode the two major vegetative alkaline phosphatases in the cell (98% of alkaline phosphatase activity); pstS, which encodes a high-affinity transport system to bring inorganic phosphate into the cell (24); and the tuaA operon, which encodes the proteins necessary for the biosynthesis of teichuronic acid, a phosphate-free anionic cell wall polymer used to replace the polyglycerol phosphate cell wall polymer, and teichoic acid, which is synthesized during phosphate-replete growth (13, 16).

Each of these genes' promoters was bound by both PhoP and PhoP~P but required PhoP~P for activation. The predominant binding region, located at approximately the same position in every promoter (21 to 60 relative to the transcription start site) contained multiple TT(A/T)ACA-like repeats separated by approximately 5 bp (16). Sequence comparisons revealed that a minimum of four of these repeats existed in each promoter, and this conserved sequence arrangement was termed the core binding region. Deletion of only one repeat from the core binding region severely reduced transcriptional activation in vivo and in vitro. Further, gel retardation assays suggested that all four repeats were required for PhoP to bind efficiently (17, 23).

Two of the stronger Pho regulon promoters, phoA and pstS, contain a secondary PhoP binding region in addition to the core binding region (17). These secondary binding sites are in the coding sequences of their genes and consist of less than four TT(A/T)ACA-like repeats, and their deletion is deleterious to promoter activation.

Promoter sequence comparisons and limited promoter deletion analyses have led to the hypothesis that directly repeated sequences in the core binding region and secondary binding sites are important features for PhoP binding and Pho regulon promoter activation, a hypothesis which can now be tested.

The study reported here focuses on phoD, encoding a phosphodiesterase-alkaline phosphatase (29, 30), whose expression is dependent on PhoP and PhoR (3). PhoD has a putative role in cell wall teichoic acid turnover during phosphate deprivation, a novel Pho regulon function in B. subtilis and possibly other gram-positive bacteria.

Several characteristics of phoD make it an interesting candidate for further study and representative of Pho regulon promoters for extensive mutagenesis. First, the phoD promoter is the strongest promoter in the Pho regulon. Second, the phoD promoter is the most tightly regulated promoter in the Pho regulon, having no promoter activity under phosphate-replete conditions or in the absence of PhoP or PhoR. Third, initial promoter studies revealed that the phoD promoter contains the elements characteristic of other Pho regulon promoters, (i) a core binding region with four TT(A/T)ACA-like repeats and (ii) a secondary binding site.

In an effort to experimentally define the PhoP consensus binding sequence and to investigate the interaction between the PhoP molecules bound to the core binding region with those bound at the secondary binding site, we mutated the phoD promoter extensively. By correlating the promoter-lacZ expression data of numerous mutant promoters with their PhoP binding data, we were able to experimentally define TT(A/T/C)ACA as the consensus binding sequence for PhoP and propose loop formation as part of the mechanism for the initiation of transcription from the stronger Pho regulon promoters.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and plasmids. B. subtilis JH642 (pheA1 trpC2 [J.A. Hoch]) was the parent strain transformed with each of the promoter-lacZ fusions. Escherichia coli MV1190 [(lac-proAB) thi supE (srl-recA)306::Tn10(Tet^r) (F':tra D36 proAB lacI^qZM15)] and CJ236 [dut ung thi relA; pCJ105 (Cm^r)], used during the oligo-directed mutagenesis of phoD, were supplied by Bio-Rad. E. coli lab strain DH5 [F 80dlacZM15 (lacZY A-argF)U169 recA1 endA1 hsdR17(r_K m_K⁺) supE44 thi-1 gyrA relA1] was used for cloning all other phoD constructions. B. subtilis MH5441 (pheA1 trpC2 amyE::phoD-lacZ Cm^r) was used as the source of RNA for primer extension.

Oligo-directed mutagenesis. Oligo-directed mutagenesis (Mutagene; Bio-Rad) was done according to the instructions of the manufacturer. Each promoter was then sequenced using dye terminator cycle sequencing ready reaction mix (Perkin-Elmer) for confirmation of the mutant construction.

Growth conditions and -galactosidase activity. The inocula for all B. subtilis cultures were grown overnight in high-phosphate defined medium containing 5 mM phosphate (24). The strains were used to inoculate low-phosphate defined medium (LPDM) (8), where optical density at 540 nm growth readings were recorded and -galactosidase specific activity levels were assayed every hour for 12 h. -Galactosidase activity was detected using the method of Ferrari et al. (4). One unit was defined as 0.33 µM of o-nitrophenol produced min¹ at 37°C. The specific activity was calculated as activity per milligram of protein. These results are expressed in terms of the fraction of activity observed compared to that of the wild-type strain (Table 1; see Fig. 3a). The final readings were taken at the point in the growth curve where activity was at its highest level before repression of the Pho response by Spo0A~P (12).

Purification of PhoP and *PhoR. Purification of PhoP and *PhoR (the cytoplasmic region of PhoR) was performed as described previously (15).

DNase I footprinting experiments. DNase I footprinting experiments were done as previously described by Liu and Hulett (15), except for the concentration of PhoP and PhoP~P used in the various experiments (see figure legends). Labeling of all probes (the phoD promoter and all mutant derivatives) on the coding strand was accomplished by digesting each plasmid construction with BamHI and filling in the sticky ends with Klenow fragment in the presence of [-³²P]dATP and [-³²P]dCTP. Subsequently, the plasmid was digested with EcoRI, and the insert was separated from the vector using a 6% polyacrylamide gel. The noncoding strands were labeled in the reverse order using [-³²P]dATP.

General methods. Transformation of E. coli was done according to the method of Hanahan (5). Transformants were selected for drug resistance and color on Luria-Bertani plates containing 150 µg of penicillin ml¹ and 120 µg of X-Gal (5-bromo-4-chloro-3-indolyl--D-galactopyranoside) ml¹. Transformation of B. subtilis was done by the two-step transformation method of Cutting and Vander Horn (2). Transformants were selected for drug resistance on Tryptose Blood Agar Base (Difco) plates containing 5 µg of chloramphenicol ml¹.

Primer extension. Strain MH5441 was grown in LPDM to stimulate the activation of phosphate starvation-inducible promoters (8). RNA was extracted, and primer extension of the phoD transcript was performed using the method described previously by Chesnut et al. (1). Two primers were used. One was complementary to the bases +63 to +78 (FMH 231), and the other was complementary to the bases +94 to +112 (FMH 232). A sequencing ladder was produced by end labeling each primer with [-³²P]dATP, annealing to pSE9, and using Sequenase (United States Biochemical) according to the instructions of the manufacturer.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Determination of the phoD transcription start site. Prior investigation of the regulation and activation of the phoD promoter in B. subtilis showed that phoD transcription was phosphate starvation inducible and regulated by the two-component signal transduction system PhoP and PhoR (3). To determine where transcription initiated at this promoter, we extracted RNA from strain MH5441 grown under low-phosphate conditions and performed primer extension using primers FMH 231 and FMH 232. The transcription initiation site was the same using both primers (Fig. 1) and was situated 26 bp upstream from the putative translation initiation codon.

View larger version (65K): [in this window] [in a new window]   FIG. 1.   Primer extension analysis of phoD. The end-labeled primer was annealed to RNA from phosphate-depleted vegetative cells grown in LPDM (lane +1) and extended with reverse transcriptase. Lanes C, G, A, and T are a sequencing ladder made by annealing the same end-labeled primer to a plasmid containing the 5' end of phoD and extending it with Sequenase (United States Biochemical). The +1 indicates the base (shown in bold print) to which the primer extension product maps.

PhoP binds to the phoD promoter. In a previous study showing that phoD was regulated by PhoP and PhoR (3), we used a 365-bp fragment containing the region of the phoD promoter extending from 321 (the numbering is based on the transcriptional start site determined above) to +44. This same promoter fragment was used as a probe to perform DNase I footprinting experiments. Both PhoP and PhoP~P were able to bind to the phoD promoter on two main regions (Fig. 2A and B). The first region, termed the core binding region, extended from 17 to 66 (coding strand), and the second region (the 5' binding region) extended from 176 to 205. The core binding region was almost fully protected by unphosphorylated PhoP at a concentration of 220 nM, whereas PhoP~P was able to fully protect it at a concentration of 55 nM. The 5' secondary binding region, which contained two TT(A/T)ACA-like repeats, was fully protected by unphosphorylated PhoP at 220 nM but required a higher concentration of PhoP~P than the core for full protection. It was not fully protected by PhoP~P until a concentration of 110 nM was used. Similar results were seen in the noncoding strand, both in the length of the binding sites and the concentrations of unphosphorylated PhoP and PhoP~P necessary for protection. On the coding strand, intermittent protection was observed between the two main binding regions when a concentration of 220 nM PhoP~P was used in the reaction. Extension of the footprinted regions, however, was not seen on the noncoding strand at these concentrations.

View larger version (67K): [in this window] [in a new window]   FIG. 2.   (A) DNase I footprint analysis of the phoD promoter bound by PhoP and PhoP~P. A 365-bp phoD promoter fragment (321 to +44) was used as the probe. F represents free lanes where no PhoP was used; G represents the G-sequencing reaction lane used as a reference. The reactions for both the coding and noncoding strands contained 0.6 µg (0.6 µM) of *PhoR. The amounts of PhoP in each of the reactions from the left to right are as follows: 20 ng (27.5 nM), 40 ng (55 nM), 80 ng (110 nM), and 160 ng (220 nM). When PhoP~P was needed, a final concentration of 4 mM ATP was added. The vertical solid lines mark the regions on the promoter which were bound by both PhoP and PhoP~P. The dashed line represents the area on the coding strand where extension of the footprint by PhoP~P occurred at intermittent places. The hypersensitive sites are indicated by arrows. (B) The phoD promoter sequence showing the PhoP and PhoP~P binding sites. The coding and noncoding sequence of the 365-bp fragment is shown. The binding sites for both PhoP and PhoP~P are represented by bold lines either above (coding) or below (noncoding) the sequence. A dashed line indicates the area between the two main binding sites which was intermittently protected when the highest concentration of PhoP~P (220 nM) was used. The arrows indicate the hypersensitive sites. The 6-bp TT(A/T)ACA-like consensus repeats (shown in bold) are underlined on both strands, and their locations in the promoter are indicated above the coding strand. The 10 and the transcriptional start site are shown in bold, underlined, and labeled. The locations of the various deletion or insertion mutations are marked for reference. The numbering is based on the transcriptional start site established by primer extension.

Point mutations in the PhoP core binding region of the phoD promoter and their effects on promoter activity. In studies of other Pho regulon promoters, TT(A/T)ACA-like repeats were protected by unphosphorylated PhoP or PhoP~P (14-17) and two sets of these TT(A/T)ACA-like repeats were defined to be the PhoP core binding unit (16). However, no mutational analysis of these repeats or the intervening sequences between them has been done to establish which bases are necessary for proper PhoP-DNA interaction in any Pho regulon promoters. Using site-directed mutagenesis, we made single-base-pair substitutions in the 3' half of the proposed core binding region of phoD. This region contains a pair of TT(A/T)ACA-like repeats, with one TT(A/T)ACA repeat centered at bp 25 and the other centered at bp 35 (the numbering is based on the transcription initiation site) with an intervening sequence between them. The sequence, TTCACAGTCGTTTAACA, is found on the coding strand of the promoter. Each base was transitionally mutated. At each position, purines were mutated to the other purine, and pyrimidines were mutated to the other pyrimidine. The strains containing these mutated phoD promoter-lacZ fusions in single copy on the chromosome were grown in LPDM, and -galactosidase activity was assayed. Our results clearly suggest that mutations of the TT(A/T)ACA-like repeats were deleterious to promoter strength (Table 1 and Fig. 3a). Most of the single point mutations in these repeats reduced promoter activity to less than 40% of wild-type levels. However, while the majority of point mutations in the TT(A/T)ACA-like repeats were deleterious, there appeared to be some flexibility within the recognition sequence, as certain mutations within these repeats were not as detrimental as others. Changing a cytosine to a thymine seemed to have the least effect of any of the point mutations, especially in the third position of the 35 repeat in which TTCACA was changed to TTTACA, bringing the wild-type sequence closer to the previously proposed consensus (16). Similar results were seen in the fifth position of the 35 and the fifth position of the 25 repeats.

View larger version (39K): [in this window] [in a new window]   FIG. 3.   (a) Point mutational analysis of the phoD promoter. The black bars represent the bases in the 3' half of the core binding region (from the 25 to the 35 consensus repeat) that were individually changed to guanines. The white bars represent all the bases which were changed from purines to purines or pyrimidines to pyrimidines. Plasmids containing these various phoD-lacZ promoter fusions in pDH32 were linearized and integrated into B. subtilis JH642 chromosome at the amyE locus as a result of a double crossover. The strains carrying these various promoter constructions were then grown in LPDM, and the promoter activity was detected every hour for a 12-h growth period. The highest level of induction attained before repression was used to calculate the specific activity of each promoter (12). The figure gives an average of three independent assays. The results are expressed in terms of the fraction of activity observed compared to the wild-type strain. (b) Consensus PhoP binding sequence based on the comparison of 34 TT(A/T)ACA-like repeats in seven different Pho regulon promoters.

Point mutations in the PhoP core binding region of the phoD promoter and their effects on PhoP binding activity. To determine if lower promoter activity was due to a reduction of PhoP's affinity to the mutant promoters, we compared the footprints of several mutant promoter constructions with the wild-type promoter. Our results show that a single point mutation in the core binding region, whether it was in the 25 or 35 consensus repeat, reduced the affinity of unphosphorylated PhoP and PhoP~P for the entire core binding region. In Fig. 4, the 25 or 35 repeats were mutated in the fourth position (both from A to G). Full protection of the core binding region in these mutant promoters was not seen until a concentration of 220 nM PhoP~P was used as opposed to 55 nM PhoP~P for the wild-type sequence. The 5' secondary binding region, however, did not appear to be affected (data not shown). In addition to DNase I footprint analysis of the promoters mutated in the TT(A/T)ACA-like consensus repeats, we also assessed changes in PhoP binding to a promoter mutated in the intervening sequence (GTCGT to GGCGT) by comparing it with the wild-type promoter (data not shown). There was a slight decrease in the ability of both unphosphorylated PhoP and PhoP~P to bind to the mutated promoter. The wild-type promoter was fully protected at a concentration of 55 nM PhoP~P, whereas the mutated promoter was only partially protected at 55 nM. However, such a difference did not appear to influence promoter activity. Taken together, these data suggest that PhoP~P must be able to recognize and bind the TT(A/T)ACA-like repeats but not the intervening sequences for full promoter activation.

View larger version (122K): [in this window] [in a new window]   FIG. 4.   DNase I footprint analysis of the wild-type phoD core binding region versus a 25 or a 35 mutant phoD core binding region using PhoP and PhoP~P. The mutant phoD promoters in pSE274b (25, TTAACA to TTAGCA) and pSE292b (35, TTCACA to TTCGCA) were used as probes, and only the coding strands are shown. The amounts of PhoP, PhoR, and ATP in each reaction were the same as those used for footprinting both the coding and noncoding strands of the wild-type phoD promoter shown in Fig. 2A. The core binding region is marked for reference by a dashed line along with the general location of the TT(A/T)ACA-like repeats.

Effect of base pair insertions in the core binding region on promoter activity and PhoP binding activity. To determine how changes in spacing between the TT(A/T)ACA-like sequences in the core binding region would affect PhoP binding affinity and promoter activation, we inserted 5 and 10 random base pairs in the intervening sequences between the 25 and 35 repeats or the 35 and 45 repeats, respectively (Fig. 2B). DNase I footprinting analysis of the promoters with the insertions between the 25 and 35 repeats revealed that both mutations were detrimental to PhoP binding affinity. Full protection of the entire core binding region was not observed with the 5-bp insertion mutant until a concentration of 220 nM PhoP~P was used (data not shown), approximately fourfold more than is required for full protection of the core binding region in the wild-type promoter. Full protection of the core binding region with the 10-bp insertion was nearly achieved at a concentration of 110 nM PhoP~P (data not shown), more than a twofold difference in binding affinity compared to the wild-type promoter. Unlike the 5-bp insertion sequence, the 10-bp insertion sequence was not protected by PhoP~P even at the highest concentrations.

View larger version (58K): [in this window] [in a new window]   FIG. 5.   DNase I footprint analysis of the phoD promoter with a 10-bp insertion between the 35 and 45 consensus repeats of the core binding region using PhoP and PhoP~P. The mutant phoD promoter in pSE344b was used as a probe. Only the coding strand was labeled. The amounts of PhoP, PhoR, and ATP in each reaction were the same as those used for footprinting both the coding and noncoding strands of the wild-type phoD promoter shown in Fig. 2A. The labels are also the same as those in Fig. 2A.

The 5' secondary binding region is necessary for promoter activation. All other Pho regulon promoters studied to date have either no secondary binding site (15, 16) or a secondary binding site in the coding region (17). The phoD promoter, however, has a secondary binding site 5' of the core binding region. To determine whether this 5' secondary binding site is necessary for full promoter activation, we made a deletion mutation, removing this site while leaving the entire sequence between it and the core binding region intact (Fig. 2B). The resulting promoter activity was only 3% of wild-type activity determined by -galactosidase assays (Table 1), indicating that the additional PhoP binding site was necessary for full promoter activity. Interestingly, the TT(A/T)ACA-like repeats in this binding site are inverted compared with those in the core binding site (Fig. 2B).

The effect of the 5' secondary PhoP binding site on PhoP binding to the core binding region. In order to investigate the role of the 5' binding site in PhoP binding to the core binding region of the promoter, the 5' deletion mutant promoter was labeled on the coding strand and used as a probe for DNase I footprinting experiments. The core binding region was bound by PhoP~P but only at a concentration of 220 nM (Fig. 6), nearly four times the amount of PhoP~P necessary for full protection of the core region in the wild-type promoter. Concurrently, we footprinted the 5' secondary binding region without the core binding region (Fig. 7). Full protection was achieved at a concentration of 220 nM PhoP~P, which is approximately double the concentration of PhoP~P necessary to bind it when the core binding region is present. These results suggest that both regions are required for PhoP to bind efficiently to this promoter.

View larger version (117K): [in this window] [in a new window]   FIG. 6.   DNase I footprint analysis of the wild-type phoD promoter versus the phoD 5' binding region deletion mutant promoter using PhoP~P. The amounts of PhoP, PhoR, and ATP in each reaction were the same as those used for footprinting both the coding and noncoding strands of the wild-type phoD promoter in Fig. 2A. The core binding region is marked for reference.

View larger version (132K): [in this window] [in a new window]   FIG. 7.   DNase I footprint analysis of the core binding region deletion mutant using PhoP~P. The amounts of PhoR and ATP in each reaction were the same as those used for footprinting both the coding and noncoding strands of the wild-type phoD promoter shown in Fig. 2A. The concentrations of PhoP used in each lane from left to right were 27.5, 55, 110, 165, 220, 275, 330, and 385 nM. Hypersensitive sites are marked with arrows, and the 5' binding region is marked with a bold line.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The 6-bp consensus for PhoP binding derived from Pho promoter sequence alignments or from experimental analysis show good agreement. Point mutations within the core showed that base pair substitutions within the consensus sequence were deleterious to transcription initiation, generally reducing promoter activity in comparison to the wild-type promoter by more than 60% (Fig. 3a). Similar mutations in the intervening sequence had little or no effect on transcription efficiency. There was some flexibility observed in the consensus repeats, as point mutations at certain positions were not as detrimental as others. A new consensus of T_29/34T_29/34(A_18/34T_10/34C_6/34)A_29/34C_27/34A_32/34 was apparent (Fig. 3b) upon adding the TT(A/T)ACA-like repeats found in phoD and tagD (14). However, the experimental data and the sequence comparison were not in agreement at the highly conserved cytosine in the fifth position. Promoter function was retained with thymine inserted in that position, a condition found only once among all the Pho promoters studied to date. Further experimental data, showing no enhancement of promoter activity when the phoD TTCACA sequence centered at bp 35 was changed to the previous consensus sequence, TT(A/T)ACA, supports a new consensus sequence, TT(A/T/C)ACA, with the realization that as more Pho regulon genes are discovered, thymine may be interchangeable with cytosine at the fifth position (Fig. 3a). Such findings would support the experimental data and would justify changing the consensus sequence at the fifth position from C to C/T.

PhoP dimers are proposed to bind to one pair of consensus repeats. PhoP is a member of the OmpR family of response regulators in which the DNA binding motif is a winged helix-turn-helix (20, 21). Members of this family are known for binding direct repeats (6, 19), and it is presumed that the single recognition helix in each DNA binding region makes contact with a specific sequence of bases in the major groove of the double helix. We have previously demonstrated that PhoP and PhoP~P are dimers in solution (15) and that both protect sequences of DNA containing direct repeats in multiples of two, suggesting that one dimer of PhoP binds to one pair of TT(A/T/C)ACA-like repeats. In this study, we demonstrated that one set of repeats is sufficient for PhoP-DNA binding (Fig. 7). Therefore, the Pho box consisting of two repeats and the core binding region consisting of a minimum of four repeats are two separate entities, the former being a functional binding unit for PhoP and/or PhoP~P and the latter being necessary to form the protein complex required for transcriptional activation of Pho regulon genes. Based on these assumptions, the core binding region of the phoD promoter, containing four repeats, should theoretically be bound by two dimers.

Pho dimer binding to the core binding region is cooperative. In direct correlation with the -galactosidase-specific activity assays, point mutations in the intervening sequence had little effect on PhoP binding affinity, suggesting that the major PhoP~P-DNA contacts were retained. In contrast, changes in either of the consensus sequences in the pair centered at 25 and 35 were detrimental to PhoP binding at the mutated binding site and importantly, at the adjacent binding site (45 and 55) demonstrating a high degree of cooperative binding within the core binding region (Fig. 4). Both unphosphorylated PhoP and PhoP~P were able to bind the mutated core region at higher concentrations, with partial footprints detected at 220 nM PhoP, a concentration of PhoP~P similar to that required to bind one pair of consensus repeats in the 5' secondary binding region (Fig. 7).

Proper spacing of the repeats within the core binding region is necessary for promoter activation but not always required for PhoP to bind efficiently. In vivo transcription assays consistently showed that both 5- and 10-bp insertions in the core binding region were deleterious to promoter activation whether the insertions were between the repeats within a single dimer binding site (35 to 25) or between two dimer binding sites (45 to 35). Either 5-bp insertion eliminated promoter activation, whereas either 10-bp insertion still allowed some promoter activity but at a severely reduced level. As promoter activity suggested, either 5-bp insertion severely reduced PhoP's affinity for the core region, requiring approximately four times the amount of PhoP~P to bind, a PhoP~P concentration similar to that required to bind the core binding region with a point mutation in the 35 or 25 repeats. This is particularly interesting because it again points out that cooperative binding is necessary to bind the core efficiently. These data clearly show that the cooperativity can be disrupted either by mutating a single dimer binding site or by changing the face of the helix of the two dimer binding sites. The binding data for the promoter with the 10-bp insertion between the 25 and 35, separating the consensus repeats within a single binding site, also correlated with promoter activity. However, the correlation between promoter activity and PhoP binding was diminished when a 10-bp insertion (between the 35 and 45) separated the two core dimer binding sites (Fig. 5). Although promoter activity was greatly reduced, PhoP~P's affinity for the core binding region was not reduced, suggesting that the two PhoP dimers were still able to make contact and bind cooperatively. This may be due to flexibility in the DNA or perhaps in PhoP itself.

PhoP binding to the secondary binding site and the core binding region is coordinated. PhoP binds to the core binding region and the 5' binding site of phoD in a coordinated fashion. Removal of either region reduces PhoP's affinity to the remaining site (Fig. 6 and 7). It is possible that a DNA loop may form between the two sites to facilitate PhoP binding to the core binding region for activation of the promoter, since deletion of the 5' binding site nearly eliminates promoter activity (Table 1). DNA loop formation had previously been proposed when analyzing the secondary binding sites in the coding regions of the phoA and pstS promoters, and it seems reasonable that this mechanism could increase the local concentration of PhoP dimers near the core binding region, ultimately strengthening the promoters (17).