Transcription factor function and promoter architecture govern the evolution of bacterial regulons

Abstract

Evolutionary changes in ancestral regulatory circuits can bring about phenotypic differences between related organisms. Studies of regulatory circuits in eukaryotes suggest that these modifications result primarily from changes in cis-regulatory elements (as opposed to alterations in the transcription factors that act upon these sequences). It is presently unclear how the evolution of gene regulatory circuits has proceeded in bacteria, given the rampant effects of horizontal gene transfer, which has significantly altered the composition of bacterial regulons. We now demonstrate that the evolution of the regulons governed by the regulatory protein PhoP in the related human pathogens Salmonella enterica and Yersinia pestis has entailed functional changes in the PhoP protein as well as in the architecture of PhoP-dependent promoters. These changes have resulted in orthologous PhoP proteins that differ both in their ability to promote transcription and in their role as virulence regulators. We posit that these changes allow bacterial transcription factors to incorporate newly acquired genes into ancestral regulatory circuits and yet retain control of the core members of a regulon.

Gene regulatory networks undergo major modifications over time, and these modifications provide an important source of phenotypic diversity among closely related organisms (1, 2). In principle, these modifications may result from changes in cis-regulatory sequences and/or from alterations in the deployment and/or activity of the transcription factors (TFs) that act on these sequences. In eukaryotes, in which most studies of the evolution of gene regulation have been carried out, most regulatory changes involve gains and/or losses of TF binding sites (as opposed to modifications in the TF themselves (3–5), although modifications in TFs have also been reported (6–8)). The prevalence of differences in cis-regulatory sites has been ascribed to the fact that changes in the binding site for a particular TF can alter transcription of a single gene selectively without affecting the expression of other genes co-regulated by the same TF (3–5). Consistent with this notion, mutations in cis-regulatory elements underlie several morphological differences among closely related animal species (3, 5, 9, 10).

We hypothesized that the evolution of gene regulation in bacteria may differ from what has been described thus far in eukaryotes for several reasons. First, it has been suggested that TFs are more likely to evolve if they acquire new target genes (7) and, unlike closely related eukaryotic organisms, closely related bacterial species often exhibit substantial differences in gene content, so that orthologous TFs control largely distinct gene sets (11–15). Second, the position and orientation of a TF binding site within a bacterial promoter (i.e., the promoter architecture) are critical for gene transcription (16). By contrast, eukoryotic promoters are characterized by the sparse and uneven distribution of TF binding sites over large DNA regions (17).

To test our hypothesis, we examined the regulatory protein PhoP and its regulated targets in 2 related enteric bacteria: the gastroenteritis-causing Salmonella enterica serovar Typhimurium and the bubonic plague agent Yersinia pestis. The PhoP protein governs virulence and the adaptation to low-Mg²⁺ environments in these 2 bacterial pathogens (18, 19), and its activity is dictated by the Mg²⁺-responding sensor PhoQ (20). Although the Salmonella and Yersinia PhoP proteins are 79% identical at the amino acid level, most of the genes directly regulated by PhoP in S. enterica have no homologs in Y. pestis, and vice versa (Perez, et al., unpublished results), probably as a result of horizontal gene transfer.

Here, we show that (i) the Salmonella and Yersinia PhoP proteins differ in their ability to promote transcription of certain species-specific genes although they retain the capacity to transcribe ancestral members of the PhoP regulon, and (ii) the Salmonella and Yersinia genomes each harbor promoter architectures not found in the other genome, but the PhoP protein recognizes a conserved motif in these 2 organisms. Our results suggest that embedding new target genes under the control of an ancestral TF entails changes in TF function and in promoter architecture.

Results

Differential Ability of the Yersinia PhoP Protein to Express Salmonella Genes Absent from Yersinia.

To examine whether the Yersinia PhoP protein was functionally equivalent to the Salmonella PhoP protein, we used the experimental strategy schematized in supporting information (SI) Fig. S1. We began by investigating Yersinia's ability to promote transcription of the S. enterica mgtA and ugtL genes, which have no orthologs in Y. pestis and harbor distinct promoter architectures (Fig. 1A) (21–23). Y. pestis produced fluorescence levels similar to those of S. enterica when transcription of a promoterless gfp gene was driven by the promoter of the mgtA gene (Fig. 1B), which is found in most enteric species (24). By contrast, Yersinia elicited significantly lower fluorescence than Salmonella when gfp was expressed from the promoter of the Salmonella-specific ugtL gene (Fig. 1C). In principle, Yersinia's inability to promote ugtL transcription could be caused by the absence of a co-regulator(s) normally present in Salmonella. To test this possibility, we examined ugtL transcription in an engineered Salmonella strain in which the phoPQ operon was replaced by the Y. pestis phoPQ operon (Fig. 1D). The resulting strain synthesized the Yersinia PhoP protein under the same conditions and at similar levels as the Salmonella PhoP protein in the strain expressing Salmonella's phoPQ operon (Fig. 1E). There was no PhoP-promoted ugtL transcription in the strain expressing the Yersinia phoPQ operon (Fig. 1G), even though the mgtA gene was transcribed at similar levels as in the Salmonella strain harboring its own phoPQ operon (Fig. 1F).

Fig. 1.

The Salmonella and Yersinia PhoP proteins differ in their ability to promote gene transcription. (A) The Salmonella mgtA and ugtL promoters differ both in the orientation and the distance of the PhoP box to the promoter −10 region. The PhoP box is located in the place that corresponds to the promoter −35 region in the mgtA promoter but is 20 bp further upstream in the ugtL promoter. (B and C) GFP expression driven by the Salmonella mgtA and ugtL promoters in Salmonella and Yersinia. Approximately 150-nt DNA fragments corresponding to the mgtA and ugtL promoter regions (covering ≈130 nucleotides upstream and ≈20 nucleotides downstream of the transcription start sites) were cloned in front of a promoterless gfp gene in the low-copy plasmid pMS201. Organisms were grown in defined medium containing 50 μM MgSO₄, inducing conditions for the PhoP/PhoQ system. GFP expression was normalized to cell density. Shown are the ratios of the normalized GFP values between wild-type Salmonella (14028s) or Yersinia (KIM6) and their respective isogenic phoP mutant strains (EG15598 and EG14737, respectively). Values shown are mean plus SD of at least 3 independent experiments. (D) Schematic of the genomic context of the phoPQ locus in Salmonella strain EG13918 coding for the Salmonella PhoP-HA and PhoQ proteins and its derivative harboring the Yersinia phoP-HA and phoQ genes (in blue) (EG17569). (E) Expression of the Salmonella and Yersinia PhoP proteins in the Salmonella strains EG13918 and EG17569 depicted in D. Western blot analysis was performed with anti-HA and anti-RpoB antibodies (to detect the PhoP-HA and RpoB proteins, respectively) on cell extracts prepared from bacteria grown as described in B and C in medium containing 10 mM (H) or 50 μM (L) MgSO₄. (F and G) mgtA and ugtL expression in the Salmonella strains depicted in D. Cells were grown as described in B and C. Transcript levels were determined by quantitative real-time PCR and normalized to ribosomal RNA levels. Shown are the ratios of the normalized transcript levels present in the strains described in D relative to those produced by the phoPQ mutant EG15598. Values shown are mean plus SD of at least 3 independent experiments. (H–M) Single-round in vitro transcription assays with linear templates corresponding to the mgtA (H–J) and ugtL (K–M) promoters, E. coli RNA polymerase, and increasing amounts of phosphorylated Salmonella or Yersinia PhoP proteins. The upper band in H–I corresponds to treR, a PhoP-repressed transcript going in the reverse orientation (22). The upper band in K and L corresponds to a spurious transcript observed in vitro but not in vivo (21, 45). Quantification of the in vitro transcription assays is shown in J for mgtA and in M for ugtL.

We reasoned that if the observed differences in ugtL expression were caused by intrinsic properties of the PhoP proteins, it should be possible to recapitulate the observed behavior (Fig. 1 B, C, F, and G) in an in vitro transcription system where the only variable was the PhoP protein. Consistent with this notion, ugtL transcription in vitro was significantly higher with the Salmonella PhoP protein than with the Yersinia PhoP protein (Fig. 1 K–M). By contrast, the Salmonella and Yersinia PhoP proteins promoted similar levels of mgtA transcription in vitro (Fig. 1 H–J). Cumulatively, these results indicate that the function of the PhoP protein has been conserved only partially since Salmonella and Yersinia diverged from their last common ancestor > 200 million years ago (25).

The Yersinia PhoP Protein Binds to the Salmonella ugtL Promoter but Does Not Recruit RNA Polymerase.

The inability of the Yersinia PhoP protein to transcribe the ugtL gene could be caused by the absence of binding to the ugtL promoter and/or could result from a defect in a subsequent step in transcription such as recruitment of RNA polymerase. We ruled out the first possibility for 2 reasons. First, the Yersinia PhoP protein bound to both the ugtL and mgtA promoter regions in vivo just like the Salmonella PhoP protein (Fig. 2 A and B). Second, the purified PhoP proteins from Yersinia and Salmonella displayed similar affinities toward a fragment harboring the ugtL promoter in vitro (Fig. S2). As expected, the PhoP protein occupied the ugtL and mgtA promoters only when grown under low Mg²⁺-inducing conditions for the PhoP/PhoQ system (Fig. 2 A and B).

Fig. 2.

The Yersinia PhoP protein binds to the ugtL promoter but cannot recruit RNA polymerase. (A–D) Promoter occupancy by PhoP (A and B) and RNA polymerase (C and D) determined by ChIP in isogenic Salmonella strains expressing the Salmonella (EG13918) or the Yersinia (EG17569) phoPQ operon. Cells were grown in N-minimal medium containing 10 mM (H) or 10 μM (L) MgCl₂. Shown are the mean plus SD of at least 3 independent experiments.

We then tested the ability of the Salmonella and Yersinia PhoP proteins to recruit RNA polymerase to the ugtL and mgtA promoters in vivo using ChIP with an antibody directed to the β subunit of the enzyme. RNA polymerase bound to both the mgtA and ugtL promoter regions when the Salmonella strain expressing its own phoP gene was grown in the presence of the low Mg²⁺-inducing signal (Fig. 2 C and D). By contrast, RNA polymerase could be recruited to the mgtA promoter but not to the ugtL promoter in the Salmonella strain expressing the Yersinia phoP gene (Fig. 2 C and D). As expected, there was no RNA polymerase recruitment to either promoter in either strain under high Mg²⁺-repressing conditions (Fig. 2 C and D). Taken together, these findings indicate that the Yersinia PhoP protein fails to promote ugtL transcription because of inefficient RNA polymerase recruitment rather than because of defective binding to the ugtL promoter.

Promoter Architecture Dictates the Ability of the Salmonella and Yersinia PhoP Proteins to Transcribe Target Genes.

The orientation and/or location of the PhoP box seem to be responsible (at least in part) for the distinct abilities of the Salmonella and Yersinia PhoP proteins to transcribe the mgtA and ugtL genes, because the corresponding promoters differ in both of these properties (Fig. 1A). In agreement with this notion, the Salmonella strain expressing the Yersinia phoPQ operon transcribed the ancestral slyB (Fig. S3) and phoP (Fig. 1E) genes, whose promoters share the location and orientation of the PhoP box with the mgtA promoter (Fig. 3A) but not the horizontally acquired pagK, mgtC, pagC, and mig-14 genes (Fig. S3 and data not shown), the promoters of which contain the PhoP box in the same orientation as in the ugtL promoter (Fig. 3A). This finding suggested to us that the architecture of natural PhoP-activated promoters in Yersinia would resemble the Salmonella mgtA promoter architecture and not conform to the ugtL promoter structure.

Fig. 3.

Architectures of PhoP-activated promoters in Salmonella and Yersinia. (A and B) Sequences of the promoter regions of the transcripts directly activated by PhoP in Salmonella (A) and Yersinia (B). The transcription start sites are indicated in blue, the PhoP boxes are in red, and the putative −10 sequences are underlined. Boxes and arrows indicate the location and orientation of the PhoP binding sites across promoters. The Salmonella PhoP-activated transcripts have been described previously (23, 26). The Yersinia PhoP-dependent transcripts were uncovered as described in Materials and Methods (Fig. S4). PhoP binding to all promoters was determined in vivo by ChIP-chip and/or ChIP-real-time PCR and in several cases in vitro by DNase I footprinting (Fig. S4). (C) Summary of the distribution of promoter architectures and the differential ability of the Salmonella and Yersinia PhoP proteins to activate transcription. Although the mgtA-like promoter architecture is shared by several Salmonella and Yersinia PhoP-activated genes, other members of the regulon harbor promoter structures that are absent from the other organism. The Salmonella and Yersinia PhoP proteins differ in their ability to activate expression from genes of the latter group.

To determine the architecture of PhoP-activated promoters in Yersinia, we identified the transcription start site (by conducting S1 mapping or primer extension experiments) and defined the PhoP binding site (by DNase I footprinting experiments with the Yersinia PhoP protein) for 14 Y. pestis genes directly controlled by the PhoP protein (Figs. 3B and S4) (see Materials and Methods for a description of how these genes were uncovered). Analysis of the sequence, location, and orientation of the PhoP box revealed interesting similarities and differences between PhoP-activated promoters in Salmonella and Yersinia. First, a comparison with previously reported Salmonella PhoP-regulated promoters (Fig. 3A) (23, 26) revealed that the sequence of the PhoP-binding motif is similar in Salmonella and Yersinia (Fig. S5). Second, 9 of 14 Yersinia promoters contain a PhoP box in the same orientation and at the same distance from the −10 region as in the Salmonella mgtA promoter. And third, the remaining 5 Yersinia promoters also harbor the PhoP box in the same orientation as in the Salmonella mgtA promoter but further upstream: for 4 of these promoters the PhoP box is located at the same distance from the −10 region as in a group of Salmonella PhoP-activated promoters (i.e., rstA, pagP, and ompX). Therefore, none of the Yersinia PhoP-activated promoters has the PhoP box in the orientation found in the Salmonella ugtL promoter.

The promoter of the Yersinia mgtC gene exhibits an architecture not previously reported for Salmonella PhoP-activated promoters: the PhoP box is at a similar distance from the −10 region as the Salmonella ugtL promoter but is in the opposite orientation (Fig. 3 B and C). Thus, we wondered whether the Salmonella PhoP protein could promote transcription from the Yersinia mgtC promoter (the Salmonella and Yersinia mgtC genes are xenologs rather than orthologs, because they seem to have been acquired independently by the Salmonella and Yersinia lineages) (24, 27). The Salmonella PhoP protein could not promote transcription of the Yersinia mgtC gene (Fig. S6), but it elicited normal transcription levels of the Yersinia y1795 gene (data not shown), which has the Salmonella mgtA-like promoter architecture (Fig. 3 B and C). These results suggest that the ability of the PhoP protein to promote gene transcription is dependent on the orientation of the PhoP box.

The αCTD Subunit of RNA Polymerase Is Necessary for PhoP-Dependent Transcription of ugtL but Not mgtA.

The location of a TF binding site within a bacterial promoter determines the RNA polymerase subunit (e.g., σ or αCTD for the carboxy-terminal domain of the α subunit) with which an activator interacts to promote gene transcription (16). In class I promoters, the activator binds upstream of the −35 element in a promoter, and activation typically relies on interactions with the flexible αCTD. In class II promoters, the activator binds to a target that overlaps the promoter −35 element usually contacting the σ subunit of RNA polymerase (16). Because the αCTD is essential for viability (28), we used in vitro transcription assays with RNA polymerase reconstituted with either wild-type or truncated αCTD (29) to probe the requirement of αCTD for expression of PhoP-regulated promoters. The αCTD was dispensable for PhoP-dependent mgtA transcription (Fig. S7), as expected for a class II promoter and consistent with previous findings reported for the Escherichia coli mgtA promoter and the E. coli PhoP protein (22). By contrast, the αCTD was necessary for PhoP-dependent transcription of the class I ugtL promoter (Fig. S7).

A Salmonella Strain Expressing the Yersinia phoPQ Operon in Lieu of Its Own phoPQ Operon Is Attenuated for Virulence.

A functional PhoP/PhoQ system is required for virulence in both Salmonella and Yersinia (18, 19). However, a Salmonella strain expressing the Yersinia phoPQ operon was strongly attenuated for virulence, because all the mice inoculated with a dose > 1,000 higher than the LD₅₀ survived, as did those inoculated with a phoP null mutant, whereas those inoculated with a Salmonella strain harboring its own phoPQ operon died (data not shown).

Discussion

It is becoming increasingly clear that transcriptional regulatory circuits undergo major modifications over time and that these modifications provide an important source of phenotypic diversity. Although several studies have addressed the evolution of gene regulatory circuits in eukaryotes, little is known about this process in bacteria. Here, we report that the evolution of the regulons controlled by the PhoP protein in the enteric pathogens S. enterica and Y. pestis has entailed functional changes in the PhoP protein as well as in the architecture of PhoP-dependent promoters, with conservation of the DNA motif recognized by the PhoP protein. Several lines of evidence support this notion. First, the Salmonella and Yersinia PhoP proteins differ in their ability to promote transcription of subsets of PhoP-regulated genes from the other species (Figs. 1, S3, and S6). Second, Salmonella and Yersinia harbor PhoP-regulated promoter architectures not found in the other species (Fig. 3). Third, the sequence of the PhoP binding motif is similar in Salmonella and Yersinia (Fig. S5). Fourth, a Salmonella strain expressing the Yersinia phoPQ operon in lieu of its own phoPQ operon is attenuated for virulence (data not shown). Our data provide a singular example of functional differences between orthologous TFs and between the architectures of their target promoters across bacterial species.

The PhoP proteins from Salmonella and Yersinia are fully competent to express genes harboring the conserved promoter architecture driving transcription of the ancestral members of the PhoP regulon but differ in their ability to transcribe species-specific genes harboring promoter architectures not found in the other species (Figs. 1 and 3). PhoP-regulated promoters can be divided into 3 groups based on the location of the PhoP box (Fig. 3 A and B): (i) the prototypical class II promoter (16), exemplified by mgtA and the ancestral members of the regulon phoP and slyB (Perez, et al., unpublished results), in which the PhoP box is ≈12 nucleotides from the −10 region; (ii) promoters in which the PhoP box is located further upstream but in the same orientation as in the mgtA promoter (e.g., the Salmonella rstA, pagP, and ompX promoters and the Yersinia y4126, y3808, y0447, and y2563 promoters); and (iii), the typical class I promoters that contain the PhoP box even further upstream but at a variable distance from the −10 region (e.g., the Salmonella ugtL and mgtC promoters and the Yersinia mgtC promoter). The first 2 promoter architectures are present both in Salmonella and Yersinia, and the PhoP proteins from these 2 species can transcribe these promoters regardless of the species of origin (Figs. 1 B, F, H, and I, S3, and data not shown). The orientation of the PhoP box in the third group of promoters is different in Salmonella and in Yersinia (Fig. 3C). Strikingly, this difference seems to be critical for PhoP function, because the Yersinia PhoP protein is impaired in its ability to elicit transcription from the Salmonella promoters of this group, and vice versa (Figs. 1 C, G, and L, 3C, and S6). The Salmonella genes harboring the latter promoters have no close homologs outside Salmonella spp., suggesting that they were horizontally acquired by the lineage that gave rise to S. enterica.

The surface of a TF that contacts RNA polymerase is determined by the orientation and location of a TF binding site (30). For instance, the transcriptional activator cyclic APM receptor protein from E. coli interacts with αCTD through more than 1 surface, depending on the position of its binding site (31). The fact that αCTD is dispensable for transcription of PhoP-activated class II promoters (e.g., mgtA) but is necessary to transcribe the 2 groups of class I promoters (e.g., ugtL, rstA, and Yersinia mgtC) (Fig. S7 and data not shown) suggests that the Salmonella and Yersinia PhoP proteins interact with both σ and with αCTD, depending on the location of the PhoP box. The contact(s) that PhoP establishes with σ in the first group of promoters (e.g., mgtA) seems to be conserved between the 2 PhoP proteins, because Salmonella and Yersinia can promote transcription from this type of promoter. Similarly, the potential interaction(s) of PhoP with αCTD (and perhaps with σ as well) when transcribing from the second group of promoters (e.g., the Salmonella rstA, pagP, and ompX promoters and the Yersinia y4126, y3808, y0447, and y2563 promoters) is likely to be conserved between Salmonella and Yersinia because these promoters share the position and orientation of the PhoP box relative to the −10 region. By contrast, the Salmonella and Yersinia PhoP proteins seem to establish productive interactions with αCTD in different manners when transcribing the third group of promoters (e.g., the Salmonella ugtL and mgtC promoter and the Yersinia mgtC promoter) that differ in the orientation of the PhoP box. Therefore, although the known flexibility of αCTD (31) enables the PhoP protein to transcribe promoters harboring the PhoP box at a variety of positions (green box in Fig. 3A), certain promoter architectures seem to be species specific and function only with the PhoP protein corresponding to that species. Because the αCTD subunit of E. coli, S. enterica, and Y. pestis are 100% identical at the amino acid level, the functional differences between the Salmonella and Yersinia PhoP proteins cannot be ascribed to differences in the transcription machinery.

The evolution of certain transcriptional regulators in yeast has been ascribed to changes in protein–protein interactions caused by a few amino acid substitutions (1, 2, 32), whereas the reported cases in higher eukaryotes typically involve whole-domain gains or losses (6). (However, some examples in which a few amino acid substitutions seem to be responsible for functional changes have also been reported (7, 33).) The Salmonella and Yersinia PhoP proteins have 2 domains: a 120-residue N-terminal domain joined by a 5-residue linker to a 98-residue C-terminal domain (34). (The N- and C-terminal domains are 80.8% and 75.5% identical, respectively.) The C-terminal region is predicted to make contacts with RNA polymerase through unidentified residues and has a helix-turn-helix DNA binding domain. The helix-turn-helix portion is identical in Salmonella and Yersinia proteins, in agreement with the conservation of the DNA motif that PhoP recognizes (Fig. S5). Moreover, the predicted secondary structures of the entire Salmonella and Yersinia PhoP proteins are superimposable (Fig. S8), suggesting that the functional differences between these orthologous PhoP proteins result from relatively minor changes in the overall structure of the proteins altering the interaction with the transcriptional machinery.

A Salmonella strain expressing the Yersinia phoPQ operon was strongly attenuated for virulence, behaving similarly to a phoP null mutant (data not shown). Given the degree of sequence identity between the sensing domains of the Salmonella and Yersinia PhoQ proteins (35), the virulence attenuation of the Salmonella strain expressing the Yersinia phoPQ operon is unlikely to result from a defect in the sensing ability of the Yersinia PhoQ protein. Rather, it may reflect that Salmonella virulence demands the expression of horizontally acquired PhoP-dependent genes, such as mig-14 (36) and mgtC (24), whose promoters share the location and orientation of the PhoP box with the ugtL promoter (Fig. 3A) and thus are not efficiently transcribed by the Y. pestis PhoP protein (data not shown).

In sum, the evolution of a bacterial regulon entails modifications in a TF and restructuring of its target promoters, which enable transcription of horizontally acquired targets while retaining control of ancestral genes. These changes are in addition to the well-established transcriptional rewiring events whereby the promoters of conserved genes in related species differ in the presence/absence of TF binding site(s) (1, 37, 38). While there are a few reports of evolutionary changes in eukaryotic TFs contributing to the modification of gene-regulatory networks (6–8, 33), alterations in gene regulation and phenotypic differences in eukaryotes have been ascribed primarily to gains and/or losses of TF binding sites (3, 5, 10). Our results (Fig. 2) and those of others (39) indicate that, in bacteria, TF binding to a DNA sequence in vivo does not necessarily result in gene transcription, thereby highlighting the need to experimentally verify the function of identified TF binding sites.

Materials and Methods

Bacterial Strains, Plasmids, and Growth Conditions.

Bacterial strains and plasmids used in this study are listed in Table S1 and primers are listed in Table S2. All Salmonella enterica serovar Typhimurium strains were derived from wild-type strain 14028s and were grown at 37 °C in N-minimal medium (40) buffered in 50 mM Bis-Tris, pH 7.7, supplemented with 0.1% casamino acids, 38 mM glycerol, and 10 μM or 10 mM MgCl₂, unless otherwise indicated. Yersinia pestis strains were derived from wild-type strain KIM6 (41) and were grown at the optimal growth temperature of 28 °C in a modified defined medium (42), pH 7.0, supplemented with 0.1% casamino acids, 10 mM (D)-glucosamine, and 50 μM or 10 mM MgSO₄. Escherichia coli strain DH5α was used as the host for the preparation of plasmid DNA. Ampicillin and kanamycin were used at 50 μg/ml and chloramphenicol at 20 μg/ml. A detailed description of the construction of bacterial strains and plasmids as well as other molecular biology procedures is provided in the SI Text.

Identification of PhoP-Regulated Promoters in Y. pestis.

To uncover the genes directly regulated by PhoP in Y. pestis, we used a combination of expression microarray analysis (wild-type vs. phoP mutant strains) and ChIP followed by array hybridization (ChIP-chip). S1 mapping (43), primer extension (44), and DNase I footprinting (43) experiments were used to validate the microarray and ChIP-chip results and to determine the structure of the PhoP-activated promoters described in Fig. 3B. A comprehensive description of the procedures and the uncovered genes will be published elsewhere.

GFP Expression Analysis.

After overnight culture in defined medium containing 10 mM MgSO₄, the Salmonella and Yersinia strains were washed twice with Mg²⁺-free medium, inoculated (1:50 dilution) in medium containing 50 μM MgSO₄, and grown to mid-exponential phase with vigorous shaking. Fluorescence values were normalized to cell density, both of which were measured in a Victor³ 1420 Multilabel counter (Perkin Elmer).

RNA Isolation and Real-Time PCR to Determine Transcript Levels.

Salmonella cells were grown as described in the previous paragraphs for the GFP expression analysis. Total RNA was prepared using the SV Total RNA Isolation System (Promega). cDNA was synthesized using TaqMan (Applied Biosystems) and random hexamers following the manufacturer's instructions. Quantification of transcripts was performed by real-time PCR using SYBR Green PCR Master Mix (Applied Biosystems) in an ABI 7000 Sequence Detection System (Applied Biosystems). Results were normalized to the levels of 16S ribosomal RNA.