Molecular Mechanism for Establishment of Signal-dependent Regulation in the PhoP/PhoQ System*

In this report, we demonstrate that H-NS is essential for establishing the Mg2+-responsive transcriptional regulation of the PhoP regulon in Salmonella. Deletion of this regulatory gene abolished the transcriptional repression of PhoP-activated genes when bacteria were grown in high environmental Mg2+, thus stimulating expression of phoP and other PhoP regulon genes. In the absence of H-NS, transcriptional activation was PhoP-dependent for those genes only activated by PhoP, but was PhoP-independent for those genes activated by both PhoP and SlyA. The H-NS protein footprints the phoP promoter in a sequence located upstream of the PhoP box; mutation of this cis-acting factor abolished transcriptional repression of the phoP gene equivalent to the phenotype exhibited in the hns mutant. Further results showed that H-NS gel shifts other PhoP regulon promoters, indicating that a PhoP-activated gene would be transcriptionally repressed via direct H-NS binding and inhibition of its activator PhoP. Furthermore, H-NS footprints a newly identified SlyA box and the reverse PhoP box in the pagC promoter, suggesting that both SlyA and PhoP compete with this regulatory protein. Therefore, H-NS should pair with SlyA and PhoP to establish a forward regulatory loop to regulate expression of pagC, and perhaps other PhoP- and SlyA-dependent genes.

In this report, we demonstrate that H-NS is essential for establishing the Mg 2؉  The PhoP/PhoQ two-component system controls several cellular functions essential for bacterial virulence in Salmonella typhimurium by responding to environmental Mg 2ϩ , pH, and host-secreted antimicrobial peptides (1)(2)(3)(4). The sensor protein PhoQ interacts with these signals and induces phosphorylation of the response regulator PhoP (5,6), which then binds to its target promoters with higher affinity in vivo (7) and stimulates transcription of PhoP-activated genes (8). A conserved hexanucleotide repeat separated by 5 nucleotides, termed PhoP box, present in the promoter region of many PhoP-regulated genes (7,9) is recognized by the PhoP protein. In addition, other regulators are also involved in activation of these PhoPdependent loci. We and other researchers have demonstrated that MarR-type regulator SlyA is required for the transcrip-tional activation of two chromosomal loci, ugtL and pagC, through a feed-forward loop (10,11). The ugtL and pagC genes are Salmonella-specific loci that are activated by PhoP (12,13) and expressed in response to low Mg 2ϩ conditions that activate the PhoP/PhoQ system (1). The PhoP protein binds to the ugtL promoter by interacting with a reverse PhoP box, and substitutions of this sequence abolished its transcription (10). Meanwhile, SlyA, whose transcription appears to be activated by PhoP (14), footprints a DNA sequence located downstream of the transcription start (ϩ1) of the ugtL promoter (10), implying that a direct protein-DNA interaction is essential for its function (10).
Contrary to the results from the in vivo expression (7), a protein-DNA interaction assay showed that the unphosphorylated PhoP protein exhibits a similar affinity to PhoP-dependent promoters in vitro when compared with the phosphorylated form (15). This raised the possibility that an unknown cellular factor is involved in establishing an in vivo transcriptional repression of PhoP-activated genes in response to high Mg 2ϩ . Indeed, another study suggested that the unphosphorylated form of the PhoP protein could mediate transcriptional activation when it was overexpressed (16), probably because PhoP could compete in the promoter region with the unknown repressor when the concentration reached a high level.
DNA-binding protein H-NS is a global transcriptional regulator that specifically silences horizontally acquired virulence genes in Salmonella (17). This process is fulfilled through interactions between H-NS and AT-rich DNA regions with low specificity (for review see Ref. 18). Interestingly, a recent report claimed that a mutation at the hns locus derived from Salmonella wild-type 14028s was lethal, but was able to be suppressed when additional deletions occurred at the phoP or rpoS locus (17). It remains unknown whether bacterial viability was affected by simultaneously deleting the 90-bp upstream and a partial coding region in the hns gene or by a cryptic mutation existing in that particular resulting strain. Nevertheless, their results implied that H-NS might interact with various PhoP-dependent promoters throughout the Salmonella genome (17).
Here we report that H-NS controls transcriptional repression of horizontally acquired PhoP-activated genes when Salmonella cells experience high Mg 2ϩ conditions. We show that a strain derived from wild-type ATCC 14028s harboring the deleted hns coding region was viable, and transcription of these PhoP-regulated genes was kept activated under both low and high Mg 2ϩ conditions. This is because the unphosphorylated PhoP is able to bind to their promoters when H-NS is absent. H-NS undergoes a two-step interaction, i.e. binding to the reg-ulator phoP promoter and the individual target promoter, which establishes the PhoP-directed, Mg 2ϩ -responsive regulation. The H-NS protein footprints an adjacent sequence located upstream of the PhoP box in the phoP promoter, and also footprints the SlyA box and PhoP box in the PhoP-and SlyA-dependent pagC promoter. We demonstrate that PhoP functions as a transcriptional activator for PhoP-activated transcription, whereas it functions as a transcriptional anti-repressor for PhoP-and SlyA-dependent transcription.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Conditions-Strains used in this study are described in Table 1. All Salmonella enterica serovar typhimurium strains are derived from the wild-type strain 14028s. Phage P22-mediated transductions in Salmonella are performed as described previously (19). Bacteria are grown at 37°C in Luria-Bertani broth or in N minimal medium, pH 7.4, supplemented with 0.1% casamino acids and 38 mM glycerol (20). MgCl 2 is added to the required concentrations. When necessary, antibiotics were added at final concentrations of 50 g/ml for ampicillin, 20 g/ml for chloramphenicol, and 50 g/ml for kanamycin. Escherichia coli DH5␣ is used as a host for the preparation of plasmid DNA.
DNA and Oligonucleotides-Chromosomal DNA was isolated using the Wizard Genomic DNA Purification Kit (Pro-mega). DNA primers used as probes, and for the construction of plasmids and strains, are described in Table 2.
Construction of Chromosomal Mutations, lac Fusions, and Epitope-tagged Proteins-PCR products were used to generate a deletion in the coding region of a gene or introduce the HA 2 epitope sequence in bacterial chromosome as described previously (21). Primers are listed as pairs for individual genes in Table 2 and plasmid pKD3 was used as the template (21). All resulting strains were confirmed using colony PCR and DNA sequencing. A lac gene was integrated behind a coding region in chromosome using plasmid pCE37 (22) into the FLP recombination target sequence generated after the Cm R cassette, which was derived from a one-step gene disruption using primers in Table 2, was removed using plasmid pCP20 (21). ⌬up-phoP::Cm R up-6 (YS11536) was constructed as follows: the Cm R cassette was introduced 36-bp upstream of the PhoP box in the phoP promoter using PCR fragments synthesized with primers 395 and 396 from pKD3 to generate ⌬up-phoP::Cm R (YS11535). Then PCR amplification was carried out using the chromosomal DNA of this strain and primers 395 and 575. This DNA product was electroporated into the wild-type strain harboring pKD46 and chloramphenicol-resistant colonies were selected. The hexamer substitution (up-6) located 12 to 6 bp upstream of the PhoP box was confirmed using colony PCR and DNA sequencing. The Cm R cassette was removed using plasmid pCP20.
Construction of Plasmids-Plasmid pYS1000 was constructed as following: plasmid pACYC184 was digested with HindIII and SalI, and the resulting 3.6-kb fragment was purified with the Gel Extraction Kit (Qiagen). This fragment was ligated with the HindIII and SalI digests of a DNA fragment amplified from the E. coli lacZ gene using primers 1 and 2 ( Table 2). Plasmid pYS1100 was constructed by using a PCR fragment containing 108 bp of the phoP promoter region, covering one promoter (P phoP-1 ) generated with primers 409 and 410, and strain 14028s chromosomal DNA, used as the template, which were digested with SalI and XhoI, and then ligated between the SalI and XhoI sites of pYS1000. Plasmids pYS1115 and pYS1244 were constructed with primers 538 and 539, and primers 775 and 776 using the Gene Tailor Site-directed Mutagenesis System (Invitrogen) with Platinum Taq Polymerase High Fidelity (Invitrogen). Plasmids pYS1118 and pYS1119 were constructed by using PCR fragments containing the 430-bp wild-type hns gene generated with primers 594 and 595, or primers 595 and 596, respectively, and strain 14028s chromosomal DNA, used as the template, which were digested with BamHI and HindIII, and then ligated between the BamHI and HindIII sites of pUHE21. All plasmids were confirmed by DNA sequencing.
Reverse Transcription-PCR (RT-PCR)-Bacterial cells were grown for 4 h in N medium supplemented with 0.01 (low) or 10 mM (high) MgCl 2 . Expression of the hns gene was induced from strains harboring pYS1118 by adding 0.2 mM IPTG under the same growth conditions. Total RNA was isolated from bacterial  Table 2 using Taq polymerase (BioLabs) and performed in a thermocycler (Bio-Rad).

Isolation of C-terminal HA-tagged H-NS (H-NS-HA)
Protein-Salmonella hns-HA strain (YS11368) harboring plasmid pUHE-hns-HA (pYS1119) was grown at 37°C with shaking to A 600 nm 0.5 in 500 ml of LB medium; then IPTG (final concentration, 0.5 mM) was added, and bacteria were incubated for another 2 h. Cells were harvested, washed with PBS once, suspended in 10 ml of PBS, and opened by sonication. The whole cell lysate was used for H-NS-HA purification by mixing with 0.3 ml of EZview Red Anti-HA affinity gel (Sigma), washed with PBS three times, and H-NS-HA was eluted with 0.25 ml of 0.1   Table 2 were incubated at room temperature for 30 min with various amounts of H-NS-HA in 15 l of an EMSA buffer consisting of 10 mM Tris-HCl, pH 8.0, 50 mM NaCl, 50 mM KCl, and 10 mM MgCl 2 . 100 pmol of unlabeled DNA was added to 32 P-labeled DNA when required. 1.8 g of anti-HA antibody (Sigma) was added to the H-NS-HA and DNA mixture when required. After addition of the DNA dye solution (40% glycerol, 0.05% bromphenol blue, 0.05% xylene cyanol), the mixture was directly subjected to 5% PAGE. Signals were detected by autoradiography.
DNase I Protection Assay-DNase I protection assays were carried out using DNA fragments amplified by PCR using Salmonella chromosomal DNA as template. Prior to the PCR, primers 470 and 514 (Table 2) were labeled with T4 polynucleotide kinase and [␥-32 P]ATP (GE Healthcare). The phoP promoter region was amplified with primers 469 and 32 P-470 and the pagC promoter region was amplified with primers 513 and 32 P-514. Approximately 25 pmol of labeled DNA and 0, 50, 100, or 200 pmol of the H-NS-HA protein were used in a 100-l reaction. DNase I digestion was carried out as described previously (23). DNase I was purchased from Invitrogen, and 0.05 units were used per reaction. Samples (3 l) were analyzed by 6% denaturing polyacrylamide electrophoresis by comparison with a DNA sequence ladder generated with the appropriate primer by using a Maxam and Gilbert AϩG reaction. The positions of radioactive DNA fragments in the gels were detected by autoradiography.
Chromatin Immunoprecipitation (ChIP) Assay-Strains harboring chromosomally encoded H-NS protein with a C-terminal HA epitope were grown in 25 ml of N medium as described above for 4 h, washed with PBS once to remove Tris (amino group), and resuspended in 25 ml of PBS. H-NS was crosslinked to promoter DNA by adding formaldehyde to 1% final concentration. ChIP assays were performed as described previously (24). Enriched promoter DNA fragments were detected by PCR using the primers listed in Table 2.
Immunoblot Analysis of HA-tagged H-NS Proteins-Bacterial suspensions for Western blot analysis were ultrasonicated. Total protein of whole cell lysates were determined using the BCA Protein Assay Kit (Pierce), and normalized to the same amount before being treated with SDS loading buffer. Expression of the hns gene was induced from strains harboring pYS1119 by adding IPTG to the required concentrations. H-NS protein was separated in 15% SDS-PAGE, transferred to nitrocellular membrane (Bio-Rad), and incubated with monoclonal anti-HA antibody. Protein signals were detected using ECL Western blotting substrate (Pierce).
␤-Galactosidase Assay-␤-Galactosidase assays were carried out in triplicate, and the activity was determined as described previously (25). Data corresponds to at least three independent assays conducted in duplicate.

RESULTS
The Regulatory Gene hns Controls Signal-responsive Repression in Horizontally Acquired PhoP-activated Loci-The unphosphorylated PhoP protein binds to target promoter DNA with a similar affinity to the phosphorylated PhoP protein in vitro (15); when overexpressed, it induces PhoP-activated genes in vivo (16). This raised the possibility that repression of PhoPactivated transcription may require other regulatory proteins in high Mg 2ϩ conditions, in which dephosphorylation of the PhoP protein is facilitated (5,6). To see if this repression process could be mediated by a universal transcriptional regulator, we constructed strains harboring a deletion mutation in chromosomal loci that were identified as global transcriptional regulators, including hns (26). We conducted RT-PCR to determine mRNA levels of the phoP transcripts in wild-type and isogenic strains carrying a mutation in those genes. When wild-type bacteria were grown in high (10 mM) and low (0.01 mM) Mg 2ϩ , expression of a PhoP-activated gene is repressed and activated, respectively (1). Therefore, the wild-type mRNA level was lower when bacteria were grown in high Mg 2ϩ than in low Mg 2ϩ (Fig. 1A). Different from the wild-type and other mutant strains we tested, the phoP mRNA level in the hns mutant grown in high Mg 2ϩ was as high as that grown in low Mg 2ϩ ( Fig.  1A and data not shown). This suggests that H-NS should be the cellular factor that mediates transcriptional repression of the phoP gene in high Mg 2ϩ conditions, thus, deletion of this locus results in derepressing phoP expression in the signal depleting condition (i.e. high Mg 2ϩ ) for the PhoP/PhoQ system. We compared transcription of those PhoP-activated genes, identified specifically from Salmonella, in wild-type and the isogenic hns mutant. The rationale for choosing these loci was because a previous report suggested that H-NS specifically silences horizontally acquired genes (17). We classified these genes as type I, which is activated by PhoP, including phoP, pcgL, pagK, and phoN (12,27,28) and type II, which is activated by PhoP and SlyA, including ugtL and pagC (10,11). Consistent with an established model (1), expression of these loci were repressed under the high Mg 2ϩ condition because the mRNA level from wild-type was reduced when bacteria were grown in high Mg 2ϩ (Fig. 1A). Mutation of the hns gene, however, abolished the transcriptional repression of both type I and II genes in this signal-depleting condition because their mRNA levels in the hns mutant grown in high Mg 2ϩ were as high as that grown in low Mg 2ϩ (Fig. 1A). The phenotype of the transcriptional activation in the hns mutant strain is solely the result of a lack of hns gene function because transcription of these genes including phoP could be restored to wild-type levels, i.e. the mRNA level was higher in low Mg 2ϩ than in high Mg 2ϩ , by a plasmid (pYS1118) containing a wild-type copy of the hns gene (Fig. 1A). Deletion of the phoP gene in hns mutant strains inhibited transcription of the type I genes under all tested conditions (Fig.  1A), suggesting that removing transcriptional repressor H-NS is not sufficient to activate their transcription. Different from the type I genes, transcriptional derepression by deleting the hns locus allows activation of the type II genes even in the absence of PhoP (Fig. 1A). PhoP appears to be unrelated to hns expression because the protein level of H-NS in a phoP mutant was similar to that in wild-type (Fig. 1A). Thus, our results demonstrate that PhoP functions as a transcriptional anti-repressor and activator for the type I genes, and as a transcriptional antirepressor for the type II genes.
Further evidence came from ␤-galactosidase assays using strains harboring lacZ fusions under control of the phoP promoter, type I promoter, and type II promoter. To compare the in vivo phoP transcription without interfering with the regulatory function of the PhoP/PhoQ system, we constructed a plas-mid (pYS1100) carrying lacZ fused to a 108-bp DNA fragment including a phoP promoter (P phoP-1 fragment from Ϫ123 to Ϫ34, Fig. 2C). This DNA could initiate a PhoP-dependent transcription from Ϫ34 (or P1 in Fig. 2C, also see Ref. 33). Meanwhile, we constructed strains harboring a chromosomal pcgL-lacZ fusion and a chromosomal pagC-lacZ fusion in addition to a set of isogenic strains with mutations in the regulatory genes. Expression of all three lacZ fusions were Mg 2ϩ -responsive and PhoP-dependent because ␤-galactosidase activity from wild- synthesized from bacteria was determined in wild-type (wt, 14028s), hns mutant (YS11708), hns mutant (YS11708) harboring pUHE21-2lac q with a wild-type copy of the hns gene (phns, pYS1118), phoP mutant (YS11590), and hns phoP mutant (YS11945). Wild-type chromosomal DNA (chr) was used as control. The constitutively transcribed rpoD gene indicated that similar amounts of total RNA were used. The protein levels of H-NS were determined in the same growth conditions from strains harboring hns-HA fusion in the chromosome or in plasmid pYS1119, whose sequence is identical to pYS1118 except a HA tag sequence at the C terminus of the hns gene (Table 1). B, P phoP1 -lacZ transcriptional fusion expressed by bacteria harboring pYS1100 was determined in wild-type (14028s), phoP mutant (YS11590), hns mutant (YS11708), hns mutant (YS11708) harboring pYS1118, and hns phoP mutant  The up-6 substitution was also introduced in the chromosome to construct the up-6 strain (YS11536). The substituted sequences are shown in parentheses. P1 and P2 are two transcription start sites. Numbering is from the A (as ϩ1) in the predicted phoP start codon (shown as uppercase letters). D, RT-PCR analysis of the mRNA level of phoP and rpoD synthesized from bacteria grown in N minimal medium, pH 7.4, with 0.01 mM (L) and 10 mM (H) Mg 2ϩ was determined in wild-type (wt, YS11535) and up-6 (YS11536) strains. E, P phoP1 -lacZ transcriptional fusion expressed by bacteria harboring pYS1100 (pwt, with wild-type sequence), pYS1115 (pup-6, with substituted UP sequence), and pYS1244 (pup-far, with substituted farther upstream sequence in C) was determined in wild-type (14028s) or phoP mutant (YS11590). Data correspond to three independent assays conducted in duplicate, and values were mean Ϯ S.D.
type strains was higher in low Mg 2ϩ than in high Mg 2ϩ , and because this activity from wild-type strains was higher than in phoP mutants in low Mg 2ϩ (Fig. 1, B-D). In addition, expression of pagC-lacZ was SlyA-dependent because ␤-galactosidase activity from wild-type was higher than in a slyA mutant (Fig.  1D). Consistent with RT-PCR results (Fig. 1A), activation of the phoP, pcgL, and pagC genes in the high Mg 2ϩ condition occurred when the hns gene was deleted because ␤-galactosidase activity from hns mutants grown in high Mg 2ϩ was greatly induced (Fig. 1, B-D). The lacZ induction in these hns mutants is solely the result of a lack of hns gene function because transcription of lacZ could be restored to wild-type levels by pYS1118, i.e. ␤-galactosidase activity was greatly reduced in high Mg 2ϩ (Fig. 1, B-D). Mutation of the phoP locus resulted in inhibition of lacZ transcription from promoters of phoP and pcgL no matter whether H-NS was present or absent (Fig. 1, B  and C). Thus, H-NS binding to a PhoP-activated promoter is to prevent unphosphorylated PhoP, which is produced in high Mg 2ϩ conditions, from initiating transcription of this gene.
On the other hand, PhoP is not essential for activation of the type II gene pagC when H-NS is absent because lacZ expression from the pagC promoter was similar in a hns mutant and a hns phoP double mutant (Fig. 1D). This indicates that PhoP functions differently in the type I and II genes. Because expression of the pagC gene is also SlyA-dependent (11), we compared the transcriptional level of the pagC-lacZ fusion from wild-type and isogenic slyA mutant strains carrying a deletion at the hns locus. lacZ expression was inhibited in a slyA mutant, but remained activated in a hns slyA mutant under all tested conditions (Fig. 1D). This result, and the above result from a hns phoP mutant, suggested that PhoP and SlyA are not required for pagC transcription when H-NS is absent. Indeed, we found that the transcriptional level of the pagC-lacZ fusion in a hns phoP slyA mutant was as high as that in the hns mutant (Fig. 1D). These experiments demonstrate that PhoP and SlyA are designed to antagonize H-NS, which represses transcription of the type II PhoP-dependent loci. Thus, the pagC promoter, perhaps ugtL as well, could be regarded as a "constitutively activated" promoter in the absence of H-NS.
The H-NS Protein Binds to a DNA Region Located Upstream of the PhoP Box in the Promoter of the phoP Gene-We conducted an EMSA using C-terminal HA-tagged H-NS protein (H-NS-HA) and a 235-bp DNA fragment including the purB-phoP intergenic region (see "Experimental Procedures"). We found that this H-NS protein could gel shift the DNA fragment ( Fig. 2A), indicating the presence of a binding site for H-NS in this DNA region. Specific H-NS binding was further identified by a supershift reaction, in which a monoclonal anti-HA antibody (Sigma) could bind to the H-NS protein and further decrease the mobility of the protein-DNA complex in a polyacrylamide gel ( Fig. 2A). Thus, we conducted DNase I footprinting assays and found that the H-NS protein protects three DNA regions in the purB-phoP intergenic region (Fig. 2B): the Ϫ112 to Ϫ100 sequence, 11 bp downstream of the purB coding region; the Ϫ86 to Ϫ74 sequence, 6 bp upstream of the PhoP box; and the Ϫ30 to Ϫ21 sequence, 4 bp downstream of the PhoP-dependent transcription start P1 (the first nucleotide of the phoP start codon as ϩ1, Fig. 2C). The Ϫ30 to Ϫ20 sequence was not present in the DNA fragment cloned in pYS1100 (see "Experimental Procedures"). This ruled out the possibility that this region is required for H-NS repression because the upstream region of P1 is sufficient to confer a Mg 2ϩ -responsive lacZ expression (Fig. 1B). To examine the Ϫ86 to Ϫ74 sequence adjacent to the PhoP box, we constructed a strain (the up-6 mutant) harboring a chromosomal substitution from actatt to tgataa (Fig. 2C) at the H-NS binding site (UP) present in the phoP promoter, and a pYS1100-derived plasmid (pup-6) carrying this hexamer substitution. Determination of phoP transcripts from the up-6 mutant indicated that mRNA levels of the phoP gene were similar in this mutant grown in low and high Mg 2ϩ conditions (Fig. 2D). Furthermore, the wild-type strain harboring plasmid pup-6 exhibited similar ␤-galactosidase activities in low and high Mg 2ϩ conditions (Fig. 2E), which was the same phenotype displayed by the hns mutant carrying the wild-type plasmid (Fig. 1B). On the other hand, lacZ expression from wild-type harboring another pYS1100-derived plasmid (pYS1244) with a heptamer substitution from ttttctt to atatgaa within the Ϫ112 to Ϫ100 sequence was similar to that from wild-type harboring a wild-type plasmid (Fig. 2E). These results indicate that binding of the H-NS protein to the UP region adjacent to the PhoP box in the phoP promoter is required for its transcriptional repression in high Mg 2ϩ .
The H-NS Protein Represses Type I PhoP-activated Genes by Binding to Their Promoters-To examine whether transcriptional repression of type I genes results from H-NS binding to their promoters, we conducted EMSA using H-NS-HA protein and PCR-generated DNA fragments carrying regulatory sequences present upstream of pcgL, pagK, and phoN coding regions (see "Experimental Procedures"). We found that the H-NS protein could gel shift all tested DNA fragments, which were supershifted by anti-HA antibodies (Fig. 3A). This result indicates that a binding site for H-NS is present in these type I promoter regions. Thus, we compared the H-NS binding ability in these promoters in a strain harboring a chromosomal hns-HA fusion with that expressed by the isogenic strain with a mutation in the regulatory gene phoP, and carried out a ChIP experiment (29) using bacteria grown in low and high Mg 2ϩ . A stronger enrichment of DNA fragments present in the phoP, pcgL, pagK, and phoN promoters was observed for the HAtagged wild-type strain grown in high Mg 2ϩ (Fig. 3B, lane d) than in low Mg 2ϩ (Fig. 3B, lane c). Furthermore, a dramatic enrichment was observed for the phoP mutant strain grown in low Mg 2ϩ , which was stronger than the wild-type strain grown in high Mg 2ϩ (Fig. 3B, lanes e and d). The ChIP assay was specific because there was no significant enrichment of H-NSbound DNA to the examined promoters when a control (i.e. wild-type strain lacking a HA-tag) strain was used (Fig. 3B,  lanes a and b). We ruled out the possibility that strong binding of the H-NS protein to these type I promoters in wild-type cells grown in high Mg 2ϩ (Fig. 3B, lane d) or in the phoP mutant strain (Fig. 3B, lane e) could be caused by high expression of the H-NS protein under these conditions (as opposed to a low level of the PhoP protein), because immunoblot analysis in cell lysates from ChIP experiments revealed that the H-NS protein level was constant in wild-type grown in both Mg 2ϩ conditions and in a phoP or slyA mutant grown in low Mg 2ϩ (Fig. 3C). This indicates that expression of H-NS is independent of PhoP and SlyA, as well as Mg 2ϩ concentration. Taken together, our observations demonstrate that the H-NS protein competes with PhoP to repress transcription directed by type I PhoP-dependent promoters.

H-NS Represses Transcription of the Type II Promoters by Binding to the PhoP Box and SlyA Box-We conducted EMSA
using H-NS-HA protein and the 250-bp DNA fragment containing regulatory motifs of the pagC promoter. 3 We found that H-NS could gel shift the pagC promoter DNA (Fig. 4A), which indicates the presence of a possible binding site for H-NS in this type II promoter region. Gel shift resulted specifically from H-NS binding because a monoclonal anti-HA antibody (Sigma) was able to further decrease the mobility of the H-NS protein and pagC DNA complex in a polyacrylamide gel (Fig. 4A). We then conducted DNase I footprinting assays and found that the H-NS protein binds to the pagC promoter throughout the DNA region from Ϫ122 to Ϫ54 (Fig. 4, B and D). We identified the pagC sequence aatatt-(N10)-attatt from Ϫ109 to Ϫ88 as the SlyA-binding site (the SlyA box) and the sequence taaatt-(N5)taaaca from Ϫ76 to Ϫ61 as the PhoP-binding site (the PhoP box, Fig. 4D) because substitutions in these sequences resulted in a complete inhibition in pagC transcription. 3 The region protected by the H-NS protein extends through the whole SlyA box and PhoP box (Fig. 4D), suggesting that binding of the H-NS protein to this promoter should compete with PhoP and the 3 W. Kong, N. Weatherspoon, and Y. Shi, submitted for publication.  PhoP-activated SlyA (14). We carried out a ChIP experiment to compare the H-NS binding ability in this type II promoter using the strain harboring the chromosomal hns-HA fusion and an isogenic strain with mutations in both phoP and slyA genes grown in low and high Mg 2ϩ . A dramatic enrichment of the DNA fragment corresponding to the pagC promoter was observed for the HA-tagged wild-type strain grown in high Mg 2ϩ (Fig. 4C, lane c), but very weak H-NS binding was observed in this strain grown in low Mg 2ϩ (Fig. 4C, lane b). Furthermore, DNA enrichment was even stronger for the phoP slyA mutant strain grown in low Mg 2ϩ (Fig. 3B, lane d). The ChIP assay was specific because there was no significant enrichment of H-NS-bound DNA to the pagC promoter when the control (i.e. wild-type strain lacking a HA-tag) was used (Fig.  4C, lane a). Thus, these results suggest that the H-NS protein establishes Mg 2ϩ -dependent transcription of the type II pagC promoter by competing with PhoP and SlyA.
The Molecular Mechanism of Mg 2ϩ -dependent Regulation Mediated by the PhoP/PhoQ System and H-NS-We titrated the H-NS protein that could mediate repression of the type I gene pcgL and the type II gene pagC using hns mutants harboring a chromosomal lacZ fusion and a plasmid (pYS1118) with an inducible hns gene. When bacteria were grown in medium without IPTG, ␤-galactosidase activity from each strain grown in low (0.01 mM) and high (10 mM) Mg 2ϩ was determined (see the actual values in lower panel of Fig. 5A) and set as 100% (Fig.  5A). In the signal-depleting condition (i.e. high Mg 2ϩ ), lacZ expression controlled by the pcgL promoter was fully repressed, like a wild-type strain, when supplemented IPTG concentrations reached 0.05 mM or higher (Fig. 5A). On the contrary, in the signal-stimulating condition (i.e. low Mg 2ϩ ), transcription remained at 78% maximum in the same IPTG concentration (Fig. 5A). This result, and a similar result in Fig. 1C, indicates that only the phosphorylated PhoP could antagonize H-NS at the pcgL promoter (Fig. 6). We determined the H-NS protein levels induced by IPTG using the hns mutant harboring a plasmid (pYS1119), which has an identical sequence to pYS1118 with a C-terminal HA tag added to the hns gene. As revealed by an immunoblot analysis, induction of the H-NS protein is proportional to the concentration of supplemented IPTG (Fig. 5B).
The SlyA amount was maintained at similar levels in low and high Mg 2ϩ (11), thus, Mg 2ϩ -responsive transcription of the pagC gene should depend primarily on the phosphorylation of PhoP. Indeed, this pagC expression was inhibited as rapidly as the pcgL gene in high Mg 2ϩ (Fig. 5A). Unexpectedly, low IPTG concentrations resulted in transcriptional repression in low Mg 2ϩ equal to high Mg 2ϩ , which was followed by a slight reduction of pagC transcription even when the IPTG concentration became higher (Fig. 5A). We therefore divided H-NSmediated transcriptional repression of the pagC gene into two stages, the first stage different from that of the pcgL gene when H-NS concentrations are low, and the second stage the same as the pcgL gene when H-NS concentrations are high. Because the H-NS protein induced in the first stage is lower than the actual amount in bacterial cytoplasm (data not shown), this Mg 2ϩindependent process implies the possibility that SlyA may facilitate binding of the unphosphorylated PhoP to the pagC promoter exhibited at the first stage. Our results demonstrate that the phosphorylated PhoP is much more resistant to H-NS repression exhibited in the second stage of pagC expression equivalent to expression of type I genes in low Mg 2ϩ (Fig. 5A).

DISCUSSION
We have identified the regulatory mechanism governing transcriptional repression of the PhoP/PhoQ two-component signaling network of S. enterica. We have established that transcriptional regulation of the PhoP regulon is controlled by the H-NS protein, which competes with PhoP protein, as well as SlyA in particular cases. This model is supported by the following data. (i) The transcriptional repression of phoP and all tested PhoP-activated genes was abolished in hns mutants when bacteria were grown under PhoP-dephosphorylated conditions (Fig. 1, A-E). (ii) The H-NS protein gel shifted all tested PhoP-dependent promoter DNA (Figs. 2A, 3A, and 4A) and footprinted the phoP promoter (Fig. 2B). (iii) The H-NS protein footprinted both the SlyA binding site and the PhoP binding site in the pagC promoter (Fig. 4B). (iv) Mutation of a H-NS binding site in the phoP promoter abolished phoP transcriptional repression to levels observed in a hns mutant (Fig. 2, D and E). (v) H-NS binding to all tested PhoP-dependent promoters was enhanced in a phoP mutant grown under PhoPphosphorylated conditions and in wild-type bacteria grown under PhoP-dephosphorylated conditions (Fig. 3B). (vi) H-NS binding to the PhoP-, SlyA-dependent pagC promoter was enhanced in a phoP slyA mutant grown under PhoPphosphorylated conditions and in wild-type bacteria grown under PhoP-dephosphorylated conditions (Fig. 4C). (vii) Inducing H-NS selectively repressed pcgL and pagC transcription in hns mutants grown under PhoP-dephosphorylated conditions (Fig. 5A).
We ruled out the possibility that the H-NS protein level could be influenced in different mutants or when bacteria were grown under all tested conditions in this study (Fig. 3C), therefore, H-NS-mediated transcriptional repression solely depends on its selective binding to the target promoters. H-NS not only binds to the promoter of the master regulator PhoP (Fig. 2, A  and B), but also binds to promoters of individual PhoP-activated genes (Figs. 3A and 4A), suggesting that H-NS and PhoP should form a counteracting regulatory pair in these horizontally acquired PhoP-activated genes. Most PhoP-activated promoters (i.e. type I promoters in this study) described to date harbor a conserved hexanucleotide repeat separated by 5 nucleotides, termed the PhoP box, located 12 bp upstream of the Ϫ10 region (9). We found one H-NS binding site (termed the UP element here) located in one helical turn upstream of the PhoP box in the phoP promoter; this spatial adjacency could explain competition between H-NS and phosphorylated PhoP when they approach these binding sites as demonstrated in this study (Fig. 3B) and illustrated in Fig. 6. This competition appears to be specific in vivo. Although H-NS could bind several DNA regions in phoP promoter in vitro (Fig. 2, B and C), only the hexamer mutation in the UP site could abolish repression of the phoP transcription in high Mg 2ϩ (Fig. 2, D and E). This feature of the phoP promoter should be applicable for other type I promoters that also possess the AT-rich sequences located upstream from the identified PhoP box (7). Although no UP consensus sequence could be found, H-NS is able to bind to these promoter DNA regions (Fig. 3, A and B).
Regulatory activity of PhoP is modulated by changing its phosphorylation level, which likely changes its conformation, and therefore, changes its in vivo affinity to target promoters when the PhoP/PhoQ system responds to environmental Mg 2ϩ (5,6). On the other hand, both unphosphorylated and phosphorylated forms of PhoP could bind to promoters with similar affinities in vitro (15). Taken together, it is likely that additional cellular factor(s) could be required for preferential binding of the phosphorylated PhoP in vivo. One example is SlyA, which mediates a feedback activation of the PhoP/PhoQ system 3 by interacting with the H-NS binding site, i.e. the UP element in the phoP promoter. Two functions of PhoP were suggested based on results in this study. (i) As a transcriptional anti-repressor. In wild-type, Mg 2ϩ -responsive transcription could be because the phosphorylated PhoP, but not unphosphorylated PhoP, was able to counteract H-NS in these promoters (Fig.  3B). (ii) As a transcriptional activator. Under high Mg 2ϩ conditions, removing H-NS in the phoP mutant could activate type II promoters (Fig. 1, A and D), but not activate type I promoters probably because an unphosphorylated PhoP is still required to turn on their expression (Fig. 1, A and C-E). Because the PhoP box is located within the Ϫ35 region of type I promoters (7), it is possible that PhoP is required to directly interact with RNA polymerase for their transcription initiation. Transcriptional activation of phoP and other PhoP-activated genes in the hns mutants can be turned on under PhoP-dephosphorylated conditions (high Mg 2ϩ ), possibly because unphosphorylated PhoP protein is able to bind to their promoter regions and function as a transcriptional activator, like the phosphorylated PhoP protein, in the absence of the competitor H-NS. FIGURE 6. Model illustrating the H-NS-dependent Mg 2؉ response of the PhoP/PhoQ system. In wild-type, the PhoP protein is phosphorylated during growth in low Mg 2ϩ . The type I PhoP-activated genes (left) are transcriptionally activated via competition of the phosphorylated PhoP, which binds with high affinity (probably facilitated by other factors) to the PhoP box (the red box) over H-NS near the UP motif (the green box), whereas type II PhoP-activated genes (green circles) are activated during growth in low Mg 2ϩ when PhoP-activated SlyA, along with the phosphorylated PhoP, bind to the promoters replacing H-NS in the SlyA box and PhoP box, and stimulating their transcription in S. typhimurium. Competition over the unphosphorylated PhoP allows the H-NS protein to occupy these chromosomal regions and repress transcription of the type I and II genes when wild-type bacteria are grown in high Mg 2ϩ conditions. In the absence of H-NS (Hns Ϫ strain), the unphosphorylated PhoP, regardless of its low DNA affinity, is able to bind to type I promoters, which facilitates transcriptional activation of these genes even in high Mg 2ϩ conditions. Meanwhile, type II promoters become independent of SlyA and PhoP in Hns Ϫ bacteria.
Type II promoters (ugtL and pagC in this study) are atypical in that they contain a PhoP box located on the opposite strand (Fig. 4D), 3 and that they require SlyA for their activation (10,11). A reverse PhoP box and SlyA box are all located farther upstream from the putative Ϫ35 region of the ugtL and pagC promoters. Contrary to the type I promoters, PhoP is not essential for transcription of these type II genes in the absence of H-NS (Fig. 1, A and D), indicating that this regulator functions solely to antagonize H-NS, i.e. as an anti-repressor. This difference is possibly derived from the involvement of the additional regulator SlyA. Although SlyA was transcriptionally activated by PhoP (10,14), protein levels of SlyA appeared to be kept at similar levels when bacteria were grown under both low and high Mg 2ϩ conditions (11). However, SlyA could only bind weakly to the pagC promoter in high Mg 2ϩ (data not shown), indicating a coordinate binding of the PhoP and SlyA proteins onto type II promoters with an unknown mechanism. H-NS binds to both the PhoP box and SlyA box present in the type II promoters although their nucleotide sequences are different (Fig. 4D). 3 Our results, therefore, provide a molecular mechanism for the feed-forward loop designed for the transcriptional activation of type II promoters, i.e. to counteract the inhibitory effect of H-NS in the upstream elements, PhoP box and SlyA box, PhoP and SlyA proteins should simultaneously bind to DNA and remove this repressor (Fig. 6). It is unknown how these distal regulatory elements enhance transcriptional initiation of the RNA polymerase complex. One possibility is that RNA polymerase requires an "entrance passage" located upstream of the Ϫ35 sequence, where, in high Mg 2ϩ , H-NS functions as a "gate-keeper" by occupying these regions and preventing polymerase complex from approaching.
Insummary,weprovideamolecularmechanismforthesignaldependent gene regulation controlled by the PhoP/PhoQ system, and perhaps by other two-component regulatory systems. Results in this study show that a type I PhoP-dependent promoter interacts with H-NS and PhoP in the following order: phosphorylated PhoP Ͼ H-NS Ͼ unphosphorylated PhoP. Therefore, H-NS antagonizes low DNA-affinity PhoP when it is unphosphorylated in high Mg 2ϩ , but not high DNA-affinity PhoP when it is phosphorylated in low Mg 2ϩ . Results in this study also show that PhoP and SlyA antagonize H-NS in type II PhoP-dependent promoters, in which the Mg 2ϩ concentration obviously determines PhoP binding to these promoters. However, it remains to be investigated which cellular signal molecule could influence SlyA binding to these promoters.