Mg2+ Sensing by the Mg2+ Sensor PhoQ of Salmonella enterica

doi:10.1016/S0022-2836(02)01268-8

J. Mol. Biol. (2003) 325, 795–807

Mg21 Sensing by the Mg21 Sensor PhoQ of Salmonella enterica Sangpen Chamnongpol, Michael Cromie and Eduardo A Groisman* Department of Molecular Microbiology, Howard Hughes Medical Institute, Washington University School of Medicine 660 S. Euclid Avenue, Campus Box 8230, St. Louis, MO 63110-1093, USA

The PhoP/PhoQ two-component regulatory system governs the adaptation to low Mg2þ environments and virulence in several Gram-negative species. During growth in low Mg2þ, the sensor PhoQ modifies the activity of the response regulator PhoP promoting gene transcription, whereas growth in high Mg2þ represses transcription of PhoP-activated genes. The PhoQ protein harbors a periplasmic domain of 146 amino acid residues that binds Mg2þ in vitro and is required for Mg2þ-mediated repression in vivo. Here, we identify periplasmic mutants of the Salmonella PhoQ protein that allow transcription of PhoP-activated genes even under high Mg2þ concentrations. When expressed in a strain harboring a PhoP variant that is phosphorylated from acetyl phosphate, some of the mutants failed to repress PhoP-promoted transcription in high Mg2þ, whereas others displayed a wild-type ability to do so. Mutant PhoQ proteins that allowed expression of PhoP-activated genes in high Mg2þ displayed a pattern of iron-mediated cleavage in vitro that was different from that displayed by wild-type PhoQ, indicative of altered Mg2þ binding. A PhoQ protein with the conserved histidine residue (H277) substituted by alanine could not promote transcription of PhoP-activated genes in low Mg2þ but could turn off expression in response to high Mg2þ. Our studies demonstrate that residues G93, W97, H120 and T156 are required for a wild-type response to Mg2þ, and suggest that Mg2þ binding to the periplasmic domain regulates several activities in the PhoQ protein. q 2003 Elsevier Science Ltd. All rights reserved

*Corresponding author

Keywords: magnesium; PhoQ; phosphatase activity; two-component system

Introduction The PhoP/PhoQ two-component system governs the adaptation to low Mg2þ environments and virulence in several Gram-negative species.1 The PhoQ protein is a sensor for extra-cytoplasmic Mg2þ and Ca2þ that controls the activity of the response regulator PhoP: growth in millimolar concentrations of Mg2þ or Ca2þ represses transcription of PhoP-activated genes whereas growth in micromolar concentrations of these divalent cations results in transcription of PhoP-activated genes.2 A Salmonella phoQ null mutant exhibits the same phenotype as that of a phoP null mutant: the inability to promote transcription of PhoPPresent address: S. Chamnongpol, Panomics Inc., 2003 E. Bayshore Road, Redwood City, CA 94063, USA. Abbreviations used: WT, wild-type. E-mail address of the corresponding author: [email protected]

activated genes.3 – 5 This suggests that the PhoP protein normally obtains its phosphoryl group from phosphorylated PhoQ, rather than from small molecular mass phosphodonors or other sensors. Moreover, it implies that PhoQ auto(trans)phosphorylation is required for transcription of PhoP-activated genes. The purified 146 amino acid residue long periplasmic domains of the PhoQ proteins from Salmonella enterica6 and Escherichia coli7 (Figure 1) specifically bind Mg2þ in vitro. A Salmonella strain expressing a PhoQ mutant harboring the T48I substitution in the periplasmic domain is less sensitive to repression by Ca2þ but responds to Mg2þ like wild-type Salmonella, consistent with the notion that the PhoQ protein has distinct binding sites for Mg2þ and Ca2þ.2,6 Bacterial membranes containing the T48I PhoQ mutant protein promoted more PhoP phosphorylation than those harboring wildtype PhoQ.8 On the other hand, the equivalent mutation in the E. coli PhoQ protein resulted in a

0022-2836/03/$ - see front matter q 2003 Elsevier Science Ltd. All rights reserved

796

Mg2þ Sensing by the Mg2þ Sensor PhoQ of Salmonella

Figure 1. Domain structure of the Salmonella PhoQ protein and amino acid sequence alignment of the periplasmic domain from the PhoQ protein from five Gram-negative species. (a) Predicted topology of the 487 amino acid PhoQ protein from S. enterica serovar Typhimurium. Numbers correspond to the amino acid residues predicted for the periplasmic domain and the conserved histidine residue in the cytoplasmic domain, which is the predicted site of phosphorylation. (b) Alignment of amino acid sequences of the predicted PhoQ periplasmic domains from S. enterica serovar Typhimurium,6 E. coli K-12,7 Pseudomonas aeruginosa,10 Erwinia carotovora27 and Providencia stuartii.28 Alignments were performed using the CLUSTAL X program.29 Shaded residues are conserved among those species and, except for the L51 and L58 residues, were mutated and tested in this study. Amino acid residues indicated by dots are conserved acidic and serine residues mutated and tested in this study. Residue T48, which when substituted by isoleucine results in a Salmonella PhoQ protein that is defective in its response to Ca2þ,6 and the acidic cluster residues, which when substituted by isosteric uncharged residues results in an E. coli PhoQ protein that is reportedly defective in its response to Mg2þ,7 are indicated by underlines.

strain that responded to both Ca2þ and Mg2þ.9 It has been suggested that a cluster of acidic residues in the periplasmic domain of the PhoQ protein is required for Mg2þ sensing in E. coli;7 yet, the acidic cluster is not found in the PhoQ protein from other bacterial species, such as Pseudomonas aeruginosa 10 (Figure 1(b)). Studies conducted in three different laboratories have established that Mg2þ promotes a phosphoPhoP phosphatase activity in the Salmonella PhoQ protein.11 – 13 However, whether Mg2þ regulates PhoQ autokinase activity has remained controversial: using membranes enriched for the PhoQ protein one group found that Mg2þ inhibited whereas PhoQ auto(trans)phosphorylation,13 another group concluded that Mg2þ had no impact on PhoQ autokinase activity.11 Because auto(trans)phosphorylation of sensor proteins is a Mg2þ-dependent reaction, the reported in vitro activities likely reflect Mg2þ acting as a signaling molecule in the periplasmic domain and as a co-factor for auto(trans)phosphorylation by the cytoplasmic domain. Here, we identify mutants in the sensing domain of the PhoQ protein that exhibit both an altered response to Mg2þ in vivo and binding to Mg2þ in vitro. We investigate the role that the conserved histidine residue in the cytoplasmic domain of the PhoQ protein plays in promoting or repressing transcription of PhoP-activated genes. Our results define residues that are important for Mg2þmediated regulation and suggest that Mg2þ binding to the periplasmic domain may regulate both

phospho-PhoP phosphatase activities in the PhoQ protein.

and

autokinase

Results An in vivo system for investigating Mg21 regulation of PhoQ-mediated responses To examine the effect of Mg2þ binding to the periplasmic domain of the PhoQ protein, we used an in vivo system that bypasses the difficulties of current in vitro assays to keep in separate compartments the periplasmic and cytoplasmic domains of the PhoQ protein, which normally experience different Mg2þ concentrations. The system consists of a set of two strains that harbor a lac transcriptional fusion to the chromosomal copy of the PhoP-activated pmrC gene. Strain EG9461 harbors a null allele of the phoQ gene and is Lac2 whether grown in low or high Mg2þ. When this strain carries plasmid pEG9050 (harboring a wild-type copy of the phoQ gene transcribed from a derivative of the lac promoter),14 pmrC transcription recapitulates the normal Mg2þ regulation displayed by a strain with a wild-type chromosomal copy of the phoQ gene: maximal expression during growth in low Mg2þ and low levels of expression during growth in high Mg2þ. The second strain, EG5931, is deleted for the phoQ gene and harbors the phoP p allele of the phoP gene, which encodes a protein that is PhoQ-independent because it can efficiently phosphorylate from acetyl phosphate.12 This strain

Mg2þ Sensing by the Mg2þ Sensor PhoQ of Salmonella

797

Figure 2. Mg2þ-dependent repression of the PhoP-activated pmrC gene requires the periplasmic domain of the PhoQ protein. (a) The ZhoQ chimera consisting of the periplasmic domain of the sensor EnvZ fused to the cytoplasmic domain of the PhoQ protein promotes high expression of the PhoP-activated pmrC gene in high Mg2þ, and (b) fails to down-regulate PhoPp-mediated gene expression in high Mg2þ. b-Galactosidase activity (Miller units) of a lac-gene fusion to the PhoP-activated pmrC gene produced by strain EG9461 (a), or by strain EG5931 (b), and harboring plasmids expressing plasmid vector (vector), wild-type phoQ (pphoQ) or zhoQ (pzhoQ). pmrC < MudJ denotes a strain harboring a transcriptional lac-gene fusion to the promoter of the pmrC gene. Data correspond to mean values of three different experiments performed in duplicate. Error bars correspond to the standard deviations (and are only shown if larger than the resolution of the Figure).

is Lacþ whether grown at low or high Mg2þ, but when it carries plasmid pEG9050, transcription of the pmrC gene recapitulates the normal Mg2þ regulation displayed by a strain with a wild-type chromosomal copy of the phoQ gene, indicating that the PhoPp protein responds normally to the PhoQ protein.12 Moreover, by expressing phoQ from a heterologous promoter, the phenotypes promoted by the mutant PhoQ proteins reflect the response to Mg2þ and not differences in PhoQ protein levels resulting from the phoPQ operon being transcriptionally autoregulated in a positive fashion.14 The PhoQ periplasmic domain is required to repress transcription in high Mg21 ZhoQ is a chimera consisting of the N-terminal periplasmic domain of the sensor EnvZ and the C-terminal cytoplasmic domain of the PhoQ protein. We investigated pmrC transcription in the phoQ and phoP p phoQ strains expressing the ZhoQ chimera and found that it remained high (i.e. , twofold change) regardless of the Mg2þ concentration used to grow the bacteria (Figure 2). (The levels of pmrC transcription in the phoP p phoQ strain expressing zhoQ were even higher than those exhibited by the strain harboring the plasmid vector.) This is in contrast to the Mg2þregulated pmrC transcription displayed by strains

expressing the wild-type PhoQ protein (Figure 2). These results are consistent with experiments in vitro showing that Mg2þ modifies the conformation of the periplasmic domain of the PhoQ protein but not that of the ZhoQ protein.2 Moreover, they indicate that the periplasmic domain of the PhoQ protein is necessary for Mg2þ to repress transcription of PhoP-activated genes. Conserved residues in the periplasmic domain of the PhoQ protein necessary for responding to Mg21 We aligned the deduced amino acid sequences of the PhoQ periplasmic domain from five Gramnegative species for which there is genetic evidence that their phoQ genes are functional, S. enterica, E. coli K-12, P. aeruginosa, Erwinia carotovora and Providencia stuartii, and identified 11 residues that are conserved across all five species. To examine the role of these residues in Mg2þ regulation in Salmonella, we examined pmrC transcription in strains expressing PhoQ proteins with amino acid substitutions in conserved residues. Mutants fell into two phenotypic classes: those that retained a wild-type response to Mg2þ (T47A, P83A, I88A, Y89A, G93A, L96A, and W97A), and those that were less sensitive to Mg2þ-promoted repression, expressing the pmrC gene even at high Mg2þ (H120A, T156A and the double mutant G93A

798

W97R) (Figure 3(a) and (b)). Western blot analysis using anti-PhoQ antibodies revealed that the steady-state levels of the mutant PhoQ proteins were similar to those produced by the wild-type PhoQ protein (data not shown), indicating that the phenotype of the various mutants is not due to differences in the levels of the PhoQ protein. We established that the Mg2þ-blind phenotype of the G93A W97R double mutant is due to both amino acid substitutions because cells expressing mutant PhoQ proteins with the single amino acid substitutions G93A, W97A or W97R did not behave like the G93A W97R double mutant (Figure 3(a)). The H120A and T156A mutants displayed similar levels of pmrC transcription as a strain harboring a wild-type copy of the PhoQ protein when grown in low Mg2þ, but they expressed higher pmrC levels than wild-type PhoQ during growth in high Mg2þ (Figure 3(b)). When examined in the phoP p phoQ strain, the three mutants fell into two phenotypic subclasses: the G93A W97R double mutant retained high levels of pmrC transcription in high Mg2þ (just like the vector-carrying strain; Figure 3(c)), indicating that it is unable to repress transcription promoted by the PhoPp protein. In contrast, mutants H120A and T156A repressed pmrC transcription similar to wild-type PhoQ during growth in high Mg2þ, indicating that they retained the ability to abolish activation of the PhoPp protein (Figure 3(c)). Similar results were obtained when the mutant PhoQ proteins were expressed from a plasmid that also expressed the PhoP protein (Figure 3(d)). Residues within an acidic cluster participate in Mg21-promoted repression The PhoQ protein harbors an acidic amino acid cluster in its periplasmic region (residues 145 – 154) that includes a stretch of five consecutive negatively charged residues, the role of which has remained unclear. On the one hand, an E. coli strain harboring a mutant PhoQ protein with isosteric substitutions in the acidic cluster exhibited limited repression by Mg2þ, though the transcription levels of the PhoP-activated gene examined were low.7 On the other hand, the acidic cluster is absent from the PhoQ proteins of Gram-negative species that respond to Mg2þ (Figure 1(b)). We determined that a Salmonella PhoQ protein deleted for residues 145 –154 conferred very low levels of pmrC transcription in low Mg2þ and failed to repress pmrC transcription in the phoP p phoQ strain in high Mg2þ (data not shown), suggesting that the deletion severely compromised PhoQ function. We evaluated PhoQ proteins harboring single amino acid substitutions in the acidic cluster. When examined in the phoQ strain, mutants D149A and D150A exhibited a decreased ability to repress pmrC transcription in high Mg2þ: the high/low Mg2þ induction ratio was 8– 11-fold compared to 70-fold for the wild-type PhoQ protein

Mg2þ Sensing by the Mg2þ Sensor PhoQ of Salmonella

(Figure 4(a) and (b)). Mutant D152A exhibited an intermediate level of induction (i.e. 18-fold) whereas mutant D151A displayed a wild-type behavior (Figure 4(b)). These four mutant proteins behaved in a different fashion in the phoP p phoQ strain: mutant D149A exhibited a wild-type response to Mg2þ whereas mutants D150A, D151A and D152A were less sensitive to Mg2þ-promoted repression: only two- to threefold versus 17-fold for the wild-type protein (Figure 4(c) and (d)). On the other hand, PhoQ proteins with the single amino acid substitutions D45A, E55A, E66A, E76A, D79A, D90A, E91A, S130A, S134A, D136A, E148A, E154A and S158A retained a wild-type response to Mg2þ both in the phoQ and phoP p phoQ strains (Table 1). These results indicate that residues within the acidic cluster contribute to a normal response to Mg2þ, but they do not appear to be as critical as conserved residues G93, W97, H120 and T156.

Mg21-blind PhoQ mutant proteins are defective for Mg21 binding in vitro An altered response to Mg2þ in vivo could be due to decreased Mg2þ binding and/or the result of conformational changes that lock the PhoQ protein in a constitutively active state without affecting Mg2þ binding. To explore the possibility that mutants G93A W97R, H120A and T156A were altered in their interaction with Mg2þ, we compared their iron-mediated cleavage pattern with that displayed by the wild-type PhoQ protein. This technique has been used to define Mg2þ-binding sites in other proteins.15 Cells expressing wild-type or mutant PhoQ proteins were incubated in the presence of iron and DTT, and the resulting products were separated by polyacrylamide gel electrophoresis and visualized by Western blotting using antiPhoQ antibodies (see Materials and Methods). Treatment of cells expressing the wild-type PhoQ protein resulted in two major cleavage products (Figure 5), the generation of which was dependent on the presence of both iron and DTT (Figure 5, lane 2 versus lanes 1 and 3). Moreover, cleavage was not observed when Mg2þ was added to the reaction (Figure 5, lane 2 versus lane 4), indicating that Mg2þ can compete with iron for the Mg2þbinding sites in the PhoQ protein and prevent cleavage of the protein. In contrast, treatment of cells expressing mutants G93A W97R, H120A and T156A lacked the two cleavage products produced by the wild-type PhoQ protein and exhibited distinct cleavage patterns (Figure 5). These results establish that PhoQ proteins G93A W97R, H120A and T156A are defective in Mg2þ binding (as opposed to being simply locked-ON mutants), and show a correlation between a deficiency to turn off transcription in response to high Mg2þ and the absence of iron-mediated cleavage products.

Mg2þ Sensing by the Mg2þ Sensor PhoQ of Salmonella

799

Figure 3. The conserved residues G93A, W97, H120 and T156 are required for Mg2þ-regulated transcription. (a) PhoQ derivatives with amino acid substitutions in conserved residues in the periplasmic domain of PhoQ that do not alter transcription of the PhoP-activated pmrC gene in response to Mg2þ. (b) PhoQ derivatives with amino acid substitutions in conserved residues in the periplasmic domain of PhoQ that allow transcription of the PhoP-activated pmrC gene in high Mg2þ. (c) The Mg2þ-blind PhoQ mutant G93A W97R (but not mutants H120A and T156A) cannot inactivate PhoPp-mediated transcription in high Mg2þ. (d) Similar profile of induction ratios of pmrC transcription in strains expressing mutant alleles of the PhoQ protein from plasmids harboring the phoQ gene (pphoQ) or both the phoP and phoQ genes (pphoP phoQ). b-Galactosidase activity (Miller units) of a lac-gene fusion to the PhoP-activated pmrC gene produced by strain EG9461 (a), (b) and (d) or by strain EG5931 (c), and harboring plasmids expressing wild-type (pphoQ and pphoP phoQ), mutated phoQs or plasmid vector (vector). pmrC < MudJ denotes a strain harboring a transcriptional lac-gene fusion to the promoter of the pmrC gene. Data correspond to mean values of four different experiments performed in duplicate. Error bars in (a)– (c) correspond to the standard deviations (and are only shown if larger than the resolution of the Figure).

800

Mg2þ Sensing by the Mg2þ Sensor PhoQ of Salmonella

Figure 4. Aspartate residues in the acidic cluster of the PhoQ periplasmic domain participate in Mg2þ-promoted repression. (a) b-Galactosidase activity (Miller units) of a lac-gene fusion to the PhoP-activated pmrC gene produced by strain EG9461 and harboring plasmids expressing plasmid vector (vector), wild-type (pphoQ), or PhoQ variants with mutations D149A, D150A, D151A and D152A. (b) Fold induction of pmrC transcription of the strains described in (a) in 10 mM Mg2þ compared to 10 mM Mg2þ. (c) b-Galactosidase activity (Miller units) of a lac-gene fusion to the PhoP-activated pmrC gene produced by strain EG5931 (b), and harboring plasmids expressing plasmid vector (vector), wild-type (pphoQ), or PhoQ variants with mutations D149A, D150A, D151A and D152A. (d) Fold repression of pmrC transcription of the strains described in (b) in 10 mM Mg2þ compared to 10 mM Mg2þ. pmrC < MudJ denotes a strain harboring a transcriptional lac-gene fusion to the promoter of the pmrC gene. Data correspond to mean values of three different experiments performed in duplicate. Error bars correspond to the standard deviations (and are only shown if larger than the resolution of the Figure).

801

Mg2þ Sensing by the Mg2þ Sensor PhoQ of Salmonella

Table 1. Effect of PhoQ and mutant PhoQ derivatives on Mg2þ-regulated transcription b-Galactosidase activity

MgCl2 (mM)

10

5

2.5

1

0.1

0.01

Fold induction (b-gal: 0.01 mM Mg2þ/ b-gal: 10 mM Mg2þ)

Vector phoQ D45A T47A E55A E66A E76A D79A P83A I88A Y89A D90A E91A G93A G93A W97R L96A W97A W97R H120A S130A S134A D136A E148A E154A T156A

1.0 11.8 4.1 14.5 1.0 25.1 14.0 20.2 1.0 1.0 21.6 22.6 11.8 26.1 222.9 27.0 1.0 21.1 60.1 18.9 36.4 53.4 23.2 21.6 138.8

0.6 21.9 9.5 24.9 4.9 47.2 17.0 33.3 2.1 4.9 67.5 37.8 33.3 50.0 167.7 50.4 4.9 48.6 67.0 32.1 76.1 76.4 37.0 41.3 118.5

3.5 45.5 10.9 48.4 6.2 79.0 31.4 63.1 5.2 6.2 62.2 73.4 53.7 97.0 172.3 88.7 11.5 98.7 103.0 62.2 98.1 116.0 83.9 78.1 124.7

5.0 101.7 23.2 126.3 6.6 181.4 123.2 186.0 6.2 29.7 218.2 212.9 208.9 226.4 220.7 201.5 51.4 137.0 168.6 211.2 312.6 239.9 167.1 199.6 186.5

5.1 281.7 149.8 315.7 50.5 628.7 461.3 414.2 72.0 146.4 282.4 789.6 658.6 657.9 322.6 489.8 47.0 164.8 373.2 702.6 356.4 652.8 490.0 655.2 364.7

5.0 454.3 268.1 640.2 89.7 929.7 612.4 835.8 193.0 268.7 425.6 795.9 996.3 878.4 438.1 721.1 69.5 192.9 365.2 763.5 616.1 803.3 698.2 971.8 665.0

5.0 38.5 65.1 44.2 89.7 37.1 43.7 41.3 193.0 268.7 19.7 35.2 84.3 33.6 2.0 26.7 69.5 9.2 6.1 40.5 16.9 15.0 30.2 45.0 4.8

b-Galactosidase activity (Miller units) of a lac-gene fusion to the PhoP-activated pmrC gene produced by strain EG9461 harboring plasmids expressing wild-type ( phoQ), mutated phoQs or plasmid vector (vector). The values represent averages of data obtained from at least four independent experiments performed in duplicate.

The putative site phosphorylation of PhoQ is dispensable for repression of PhoP-mediated transcription during growth in high Mg21 The histidine residue at position 277 (H277) in the cytoplasmic domain of the PhoQ protein is the predicted site of autophosphorylation. A strain expressing an H277A PhoQ variant failed to promote pmrC transcription in low Mg2þ (Figure 6(a)). On the other hand, this mutant retained the ability to abolish pmrC transcription in the phoP p phoQ strain (Figure 6(b)). The conserved H277 residue has been implicated in PhoQ

phospho-PhoP phosphatase activity because an H277V mutant was defective for this activity in vitro.11. However, we established that, in vivo, the H277V PhoQ protein behaved like the H277A mutant: it retained the capacity to abolish pmrC transcription in the phoP p phoQ strain in response to high Mg2þ (Figure 6(b)) and failed to promote pmrC expression in a phoQ mutant grown in low Mg2þ (data not shown). These results further support the notion that phosphorylated PhoQ is the main, possibly sole phosphodonor for the PhoP protein, and that in vivo inactivation (possibly dephosphorylation) of phospho-PhoP by the PhoQ

Figure 5. Mutant PhoQ proteins defective in the response to Mg2þ exhibit altered binding to Mg2þ. Fe2þ-mediated cleavage of PhoQ proteins expressed in E. coli cells from plasmids pUHE-phoP phoQ (wild-type), pUHE-phoP phoQ G93A W97R, pUHE-phoP phoQ H120A and pUHE-phoP phoQ T156A. Cleavage products generated in the wild-type PhoQ protein are marked by arrowheads. Generation of these products requires iron and DTT and is prevented in the presence of Mg2þ.

802

Mg2þ Sensing by the Mg2þ Sensor PhoQ of Salmonella

Figure 6. The conserved histidine residue in the PhoQ cytoplasmic domain is essential for PhoP-mediated transcription activation, but not necessary for its repression. The PhoQ protein is unable to activate PhoP-mediated transcription when the H277 residue is replaced by alanine (a) but retains the ability to repress PhoPp-dependent transcription in high Mg2þ when carried in H277A or H277V mutations (b). (c) The PhoQ H277A protein is unable to repress PhoPp-mediated transcription in the presence of Mg2þ-blind mutations, zhoQ or G93A W97R. b-Galactosidase activity (Miller units) of a lac-gene fusion to the PhoP-activated pmrC gene produced by strain EG9461 (a) or by strain EG5931 (b) and (c), and harboring plasmids expressing wild-type (pphoQ), mutated phoQs or plasmid vector (vector). pmrC
protein occurs by a mechanism other than reverse phosphotransfer to the conserved histidine residue in the PhoQ protein. We examined the behavior of PhoQ variants harboring periplasmic mutations resulting in the Mg2þ-blind phenotype (the ZhoQ chimera or the G93A W97R double mutant) and the H277A mutation in the putative site of phosphorylation. Cells expressing a PhoQ protein with the G93A W97R H277A mutations or the ZhoQ chimera with the H277A mutation were unable to promote pmrC transcription (data not shown). This indicates that the transcriptional activation exhibited by Mg2þ-blind mutants is dependent on a functional site of phosphorylation in the PhoQ protein. On the other hand, pmrC transcription was not abolished during growth in high Mg2þ when these

mutant PhoQ proteins were examined in the phoP p phoQ strain (Figure 6(c)). These results indicate that, like wild-type PhoQ, the H277A mutant abolishes PhoPp-mediated transcription only in conjunction with the proper periplasmic sensor domain.

Discussion Binding of a signal molecule to the sensing domain of a sensor protein is the first step in any signal transduction cascade. Here, we have analyzed the primary example of a sensor that responds to Mg2þ as an extracellular signal. We identified residues in the periplasmic (i.e. sensing) domain of the sensor PhoQ protein of Salmonella

Mg2þ Sensing by the Mg2þ Sensor PhoQ of Salmonella

that are required for responding to Mg2þ and determined that, in vivo, Mg2þ regulates several activities in the PhoQ protein. The periplasmic domain of the PhoQ protein is responsible for responding to Mg2þ because its replacement by the sensing domain from the sensor EnvZ resulted in a protein that promoted transcription of PhoP-activated genes even in high Mg2þ (Figure 2). We investigated the role of periplasmic residues that are conserved among the PhoQ proteins of different enteric species, and identified four (G93, W97, H120 and T156) that appear to be involved in Mg2þ recognition, because strains harboring PhoQ mutants H120A, T156A or G93A W97R were refractory to repression by Mg2þ (Figure 3). These mutants are affected in Mg2þ binding because their iron-cleavage pattern was different from that exhibited by the wild-type PhoQ protein (Figure 5). The residues identified are not adjacent to one another in the primary sequence of the 146 amino acid residue periplasmic domain of PhoQ, and it is presently unknown whether they are clustered in the threedimensional structure of the protein. The participation in Mg2þ sensing of residues that are scattered in the primary sequence of the PhoQ protein is reminiscent of divalent cation detection by the extracellular Ca2þ-sensing receptor, where a wide range of mutations throughout its extracellular domain perturbs its functions.16 Aspartic acid residues have been implicated in Mg2þ binding by proteins that use Mg2þ as a ligand or a co-factor17 – 19. We established that four aspartate residues (D149, D150, D151, and D152) in the periplasmic domain of PhoQ contribute to a wild-type response to Mg2þ because their replacement by alanine resulted in strains that were less sensitive to repression by Mg2þ (Figure 4); yet, these PhoQ variants did not exhibit the Mg2þblind phenotype of the G93A W97R, H120A and T156A mutants (Figure 3). That residues outside of the acidic stretch are involved in Mg2þ recognition is consistent with the fact that many PhoQ proteins lack an acidic stretch (Figure 1(b)) and that an E. coli strain expressing a PhoQ protein with its periplasmic domain substituted by the corresponding one from P. aeruginosa (which does not have an acidic cluster) responds normally to divalent cations. 20 The PhoQ protein requires the conserved histidine residue (H277) in its cytoplasmic domain to promote transcription of PhoP-activated genes (Figure 6). This is typical of sensor kinases of the two-component family and suggests that the PhoP protein is activated only by phosphorylated PhoQ and not by small molecular mass phosphate donors (such as acetyl phosphate) or by other sensors (at least under our in vivo assay conditions). The H277 residue is critical for the elevated levels of pmrC transcription displayed by strains expressing the G93A W97R mutant PhoQ or the ZhoQ chimera (Figure 3(b)) because cells expressing the G93A W97R H277A triple mutant or the ZhoQ

803

chimera with the H277A mutation were unable to promote pmrC transcription (Figure 6). It has been reported that Mg2þ promotes a phospho-PhoP phosphatase activity in the PhoQ protein in vitro.11,13 Our in vivo experiments support this view because transcription of pmrC was abolished in high Mg2þ in phoP p phoQ Salmonella expressing wild-type, H120A or T156A PhoQ proteins but not the G93A W97R mutant PhoQ (Figure 3(c)). Our results best fit the overlapping-site model21 where interaction between the conserved histidine of PhoQ and the conserved aspartate of PhoP share a common transition state for both PhoP phosphorylation and dephosphorylation. This model proposes that the phosphate is transferred from the phospho-histidine in the sensor to the aspartate in the regulator during phosphorylation, and that water replaces the phosphorylated histidine sidechain leading to hydrolysis during dephosphorylation. A different model for PhoQ phosphatase activity involving reverse phosphotransfer has been proposed based on the inability of the purified membrane enriched for H277V PhoQ protein to dephosphorylate phospho-PhoP.11 This in vitro phenotype might be seen only in the H277V mutant because other amino acid substitutions were not reported,11 and because the phosphatase activity of the sensor EnvZ is affected only by certain amino acid substitutions at the conserved histidine residue.21 Moreover, PhoQ mutants with H277A or H277V substitutions exhibited a wild-type ability to repress PhoPp-mediated transcription in high Mg2þ (Figure 6(b)), indicating that the conserved H277 is not required for this activity in vivo. Furthermore, the phosphatase activity by the A domain of the sensor kinase EnvZ can be altered by the cations present in the assay buffer22 and this may also be the case with the PhoQ protein. In addition to regulating PhoQ phospho-PhoP phosphatase activity, Mg2þ has been implicated in regulating PhoQ autokinase in vitro by one group13 but not by another.11 If one assumes that the steady-state levels of phospho-PhoP protein are similar in wild-type and phoP p strains, then the disparate in vivo phenotypes of our Mg2þ-blind mutants (Figure 3) support the notion that Mg2þ does regulate PhoQ autokinase and/or phosphotransfer activities. This is because the H120A and T156A mutants display wild-type ability to abolish pmrC transcription in the phoP p phoQ strain (Figure 3(c)), yet mediate higher levels of pmrC transcription than wild-type PhoQ in the phoQ strain during growth in high Mg2þ (Figure 3). In summary, our work demonstrates that the periplasmic domain of the PhoQ protein is involved in Mg2þ sensing and identifies residues that are required for a wild-type response to Mg2þ. Moreover, it suggests that the decrease in transcription of PhoP-activated genes when Salmonella experiences growth in high Mg2þ results from both an increase in phospho-PhoP phosphatase activity in the PhoQ protein and a decrease in phosphorylated PhoQ to serve as a phosphodonor to PhoP.

804

Mg2þ Sensing by the Mg2þ Sensor PhoQ of Salmonella

Table 2. Bacterial strains and plasmids Strain or plasmid

Description or phenotype

Reference or source

A. Strain S. enterica serovar Typhimurium EG9461 EG5931 E. coli DH5a

pmrC(psiD)9065
2 12 30

BL21

sup E44 Dlac U169 (f80 lacZDM15) hsd R17 recA1 endA1 gyrA96 thi-1 relA1 F2 dcm ompT hsdS gal

B. Plasmid pUHE21-2lacI q pEG9050 pEG9090 pUHE-phoQ DE145– E154 pUHE-phoQ D45A pUHE-phoQ T47A pUHE-phoQ E55A pUHE-phoQ E66A pUHE-phoQ E76A pUHE-phoQ D79A pUHE-phoQ P83A pUHE-phoQ I88A pUHE-phoQ Y89A pUHE-phoQ D90A pUHE-phoQ E91A pUHE-phoQ G93A pUHE-phoQ L96A pUHE-phoQ W97A pUHE-phoQ W97R pUHE-phoQ G93A W97R pUHE-phoQ H120A pUHE-phoQ S130A pUHE-phoQ S134A pUHE-phoQ D136A pUHE-phoQ E148A pUHE-phoQ E154A pUHE-phoQ T156A pUHE-phoQ T281R pGEX-2T pGEX-phoQ pGEX-zhoQ pGEX-phoQ H277A pGEX-phoQ H277V pGEX-zhoQ H277A pGEX-phoQ G93A W97R H277A pEG9071 pUHE-phoP phoQ G93A W97R pUHE-phoP phoQ H120A pUHE-phoP phoQ T156A

reppMBI Apr lacI q lacI q phoQ, Apr (pUHE21-2lacI q derivative) lacI q zhoQ, Apr (pUHE21-2lacI q derivative) lacI q with phoQ DE145–E154, Apr (pUHE21-2lacI q derivative) lacI q with phoQ D45A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ T47A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ E55A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ E66A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ E76A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ D79A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ P83A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ I88A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ Y89A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ D90A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ E91A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ G93A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ L96A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ W97A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ W97R, Apr (pUHE21-2lacI q derivative) lacI q with phoQ G93A W97R, Apr (pUHE21-2lacI q derivative) lacI q with phoQ H120A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ S130A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ S134A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ D136A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ E148A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ E154A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ T156A, Apr (pUHE21-2lacI q derivative) lacI q with phoQ T281R, Apr (pUHE21-2lacI q derivative) GST/tac, Apr GST/tac phoQ, Apr (pGEX-2T derivative) GST/tac zhoQ, Apr (pGEX-2T derivative) GST/tacwith phoQ H277A, Apr (pGEX-2T derivative) GST/tacwith phoQ H277V, Apr (pGEX-2T derivative) GST/tacwith zhoQ H277A, Apr (pGEX-2T derivative) GST/tacwith phoQ G93A W97R H277A, Apr (pGEX-2T derivative) lacI q phoP phoQ, Apr (pUHE21-2lacI q derivative) lacI q with phoP phoQ G93A W97R, Apr (pUHE21-2lacI q derivative) lacI q with phoP phoQ H120A, Apr (pUHE21-2lacI q derivative) lacI q with phoP phoQ T156A, Apr (pUHE21-2lacI q derivative)

14 14 2 This work This work This work This work This work This work This work This work This work This work This work This work This work This work This work This work This work This work This work This work This work This work This work This work This work Pharmacia This work This work This work This work This work This work 14 This work This work This work

Materials and Methods Bacterial strains, growth conditions and bgalactosidase assays Strains used in this study are listed in Table 2. All S. enterica serovar Typhimurium strains used in this study are derived from wild-type 14028s. E. coli DH5a was used as host for the preparation of plasmid DNA and E. coli BL21 was used as host for the iron-mediated cleavage assay. Plasmids were introduced into Salmonella by DNA electroporation using a Bio-Rad apparatus and into E. coli by chemical transformation using standard procedures. Bacteria transformed with plasmid DNA were maintained in growth media supplemented with 50 mg/ml ampicillin. E. coli were grown in Luria Broth (LB) and Salmonella strains were grown in N-minimal medium (pH 7.4)23 supplemented with

Pharmacia

0.1% (w/v) Casamino acids and 38 mM glycerol, and different concentrations of MgCl2. b-Galactosidase assays were carried out as described24 using a microtiter plate assay. A minimum of three independent assays were performed for each strain, and the results averaged for display as graphs.

Plasmid constructs Plasmids used in this study are listed in Table 2. Recombinant DNA techniques were performed according to standard protocols. All plasmid constructs were confirmed by sequence analysis of the entire phoQ gene. The levels of mutant PhoQ proteins were similar to that of the wild-type protein as determined by Western blot analysis using polyclonal antibodies raised against the cytoplasmic domain of the PhoQ protein.

805

Mg2þ Sensing by the Mg2þ Sensor PhoQ of Salmonella

Table 3. Primers used in this study Plasmid

Primer description

Primer sequence

pUHE-phoQ pGEX-phoQ pGEX-zhoQ pUHE-phoQ DE145-E154

#1042 phoQ forward #1295 phoQ reverse for pUHE21-2lacI q #1115 phoQ reverse for pGEX-2T #1091 zhoQ forward #50 ASTRECHQC

50 50 50 50 50 30 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50

pUHE-phoQ D45A pUHE-phoQ T47A pUHE-phoQ E55A pUHE-phoQ E66A pUHE-phoQ E76A pUHE-phoQ D79A pUHE-phoQ P83A pUHE-phoQ I88A pUHE-phoQ Y89A pUHE-phoQ D90A pUHE-phoQ E91A pUHE-phoQ G93A pUHE-phoQ L96A pUHE-phoQ W97A pUHE-phoQ W97R pUHE-phoQ H120A pUHE-phoQ S130A pUHE-phoQ S134A pUHE-phoQ D136A pUHE-phoQ E148A pUHE-phoQ E154A pUHE-phoQ T156A pUHE-phoQ T281R pGEX-phoQ H277A pGEX-phoQ H277V

#30 ASTRECHQN #1297 forward D45A #1298 reverse D45A #1287 forward T47A #1288 reverse T47A #1267 forward E55A #1268 reverse E55A #1269 forward E66A #1270 reverse E66A #1271 forward E76A #1272 reverse E76A #1273 forward D79A #1274 reverse D79A #1321 forward P83A #1322 reverse P83A #1323 forward I88A #1328 reverse I88A #1289 forward Y89A #1290 reverse Y89A #1275 forward D90A #1276 reverse D90A #1299 forward E91A #1300 reverse E91A #1291 forward G93A #1292 reverse G93A #1324 forward L96A #1325 reverse L96A #1326 forward W97A #1327 reverse W97A #1341 forward W97R #1342 reverse W97R #1301 forward H120A #1302 reverse H120A #1277 forward S130A #1278 reverse S130A #1279 forward S134A #1280 reverse S134A #1281 forward D136A #1282 reverse D136A #1283 forward E148A #1284 reverse E148A #1285 forward E154A #1286 reverse E154A #1303 forward T156A #1304 reverse T156A #1319 forward T281R #1320 reverse T281R #1157 forward H277A #1158 reverse H277A #1354 forward H277V #1355 reverse H277V

Single amino acid substitutions and acidic stretch deletions were introduced into the phoQ gene as described25 using pEG9050 DNA as template. The primers used are listed in Table 3 with the mutated bases underlined. In brief, two DNA fragments with overlapping ends corresponding to the phoQ gene were amplified by PCR using complementary primers. The fragments were combined for PCR amplification of the full-length gene using primer phoQ forward (#1042 with a Bam HI site) and primer phoQ reverse (#1295 with

CGC GGA TCC ATG AAT AAA TTT GCT CGC 30 CCC AAG CTT GGG TTA TTC CTC TTT CTG 30 GGA ATT CCT TCC TCT TTC TGT GTG GG 30 CGC GGA TCC GCG ATG AGG CGA ATG 30 AAA AAC TCA AAA TGA CCC ACT CGG TAG CGG AGT GGG TCA TTT TGA GTT TTT CCT GCG CGG 30 GTA AGT TTC GCG AAA ACC ACC 30 GGT GGT TTT CGC GAA ACT TAC 30 TTT GAT AAA GCC ACC TTT CGT 30 ACG AAA GGT GGC TTT ATC AAA 30 CTG CGC GGC GCT AGC AAC CTG 30 CAG GTT GCT AGC GCC GCG CAG 30 GCC AAA TGG GCC AAT AAT AAA 30 TTT ATT ATT GGC CCA TTT GGC 30 GAG CTG CCG GCC AAT CTG GAC 30 GTC CAG ATT GGC CGG CAG CTC 30 GAA AAT CTG GCA ATG CAA AGC 30 GCT TTG CAT TGC CAG ATT TTC 30 ATG CAA AGC GCT ACC ATG ACG 30 CGT CAT GGT AGC GCT TTG CAT 30 ATG ACG CTA GCT TAC GAT GAA 30 TTC ATC GTA AGC TAG CGT CAT 30 ACG CTG ATT GCC GAT GAA ACG 30 CGT TTC ATC GGC AAT CAG CGT 30 CTG ATT TAC GCT GAG ACG GGC 30 GCC CGT CTC AGC GTA AAT CAG 30 ATT TAC GAT GCA ACG GGC AAA 30 TTT GCC CGT TGC ATC GTA AAT 30 GAT GAA ACG GCC AAA TTA TTA 30 TAA TAA TTT GGC CGT TTC ATC 30 GGC AAA TTA GCA TGG ACG CAG 30 CTG CGT CCA TGC TAA TTT GCC 30 AAA TTA CTA GCG ACG CAG CGC 30 GCG CTG CGT CGC TAGTAA TTT 30 AAA TTA TTA AGG ACG CAG CGC 30 GCG CTG CGT CCT TAA TAA TTT 30 AAC GGC TTT GCT GAA ATT GAA 30 TTC AAT TTC AGC AAA GCC GTT 30 GAC GCC ACG GCC ACT CTG TTG 30 CAA CAG AGT GGC CGT GGC GTC 30 ACG CTG TTG GCC GAG GAC CAT 30 ATG GTC CTC GGC CAA CAG CGT 30 TTG AGC GAG GCC CAT TCC GCG 30 CGC GGA ATG GGC CTC GCT CAA 30 GAA GTA CGT GCA GAT GAC GAT 30 ATC GTC ATC TGC ACG TAC TTC 30 GAT GAT GCC GCC ATG ACC CAC 30 GTG GGT CAT GGC GGC ATC ATC 30 GCC GAG ATG GCC CAC TCG GTA 30 TAC CGA GTG GGC CAT CTC GGC 30 AGT TTA AAA AGG CCT CTC GCG 30 CGC GAG AGG CCT TTT TAA ACT 30 ACC GAC CTG ACG GCT AGC TTA AAA ACG 30 CGT TTT TAA GCT AGCCGT CAG GTC GGT 30 GAC CTG ACG GTGAGT TTA AAA 30 TTT TAA ACT CAC CGT CAG GTC 30

HindIII or #1115 with Eco RI sites). The resulting PCR product was digested with appropriate restriction enzymes (Bam HI and HindIII for cloning into plasmid pUHE21-2lacI q, and Bam HI and Eco RI for cloning into plasmid pGEX-2T) and ligated into a plasmid vector digested with the same restriction enzymes. Plasmid pUHE-phoQ G93A W97R was selected from the screening for plasmid pUHE-phoQ G93A. Plasmid pGEX-phoQ, which contains the wild-type phoQ gene, was constructed by ligating a Bam HI and

806

Eco RI-digested PCR product generated with primers #1042 and #1115 and pEG9050 DNA as template, into pGEX-2T DNA cut with the same restriction enzymes. Plasmid pGEX-zhoQ, which contains the 184 N-terminal amino acid residues from S. enterica EnvZ fused to 268 C-terminal amino acid residues of Salmonella PhoQ, was constructed by ligating a Bam HI and HindIII-digested PCR product generated with primers #1042 and #1295 and pEG9090 DNA as template, into pUHE21-2lacI q DNA cut with the same restriction enzymes. The resulting plasmid was digested with Bam HI and Age I, and subcloned between the Bam HI and Age I sites of pGEXphoQ to generate pGEX-zhoQ. Plasmids encoding PhoQ proteins with the H277A and ZhoQ or G93A W97R were constructed by insertion of the Bsp EI-Age I fragment from plasmid pGEX-phoQ H277A harboring an alanine substitution of the conserved histidine residue into Bsp EI and Age I-digested pGEX-zhoQ plasmid to generate plasmid pGEX-zhoQ H277A, and into Bsp EI and Age I-digested pGEX-phoQ G97A W97R plasmid to generate plasmid pGEX-phoQ G93A W97R H277A. Plasmids encoding PhoP and PhoQ proteins with the G93A W97R, H120A and T156A were constructed by subcloning of the NgoM IV-HindIII fragment from plasmid pUHE-phoQ harboring the corresponding substitutions into the NgoM IV and HindIII sites of pEG9071 to generate pUHE-phoP phoQ G93A W97R, pUHE-phoP phoQ H120A and pUHE-phoP phoQ T156A, respectively. Iron-mediated cleavage of the PhoQ protein Iron-mediated cleavage is based on the ability of iron to replace Mg2þ bound to proteins and to produce hydroxyl radicals when incubated in the presence of reducing agents such as DTT, generating compounds that can cleave a polypeptide chain in the vicinity of the bound iron. Products generated as a consequence of iron-cleavage can be identified by Western blot analysis using antibodies against the protein of interest. Ironmediated cleavage assays were performed as described15,26 using log phase cultures of E. coli BL21 cells harboring plasmids pUHE-phoP phoQ, pUHE-phoP phoQ G93A W97R, pUHE-phoP phoQ H120A or pUHEphoP phoQ T156A growing in N-minimal medium (pH 7.4) and 100 mM IPTG. Washed cells were equilibrated with Mg2þ by incubating with 0 mM or 40 mM MgCl2 on ice for ten minutes, then incubated in the presence of FeSO4 (0 mM or 250 mM) and DTT (0 mM or 20 mM) for three hours at room temperature. The cleaved products were separated on SDS-15% (w/v) polyacrylamide gels, transferred to nitrocellulose membranes by semi-dry blotting and analyzed by Western blot using polyclonal antibodies directed against the cytoplasmic and periplasmic domains of the PhoQ protein.

Acknowledgements We thank A. Bonislawski for technical help; and A. Stock, D. Buckler and an anonymous referee for valuable criticism of the manuscript. This research was supported by grant AI49561 from the NIH to E.A.G. who is an Associate Investigator of the Howard Hughes Medical Institute.

Mg2þ Sensing by the Mg2þ Sensor PhoQ of Salmonella

References 1. Groisman, E. A. (2001). The pleiotropic twocomponent regulatory system PhoP-PhoQ.. J. Bacteriol. 183, 1835– 1842. 2. Garcı´a Ve´scovi, E., Soncini, F. C. & Groisman, E. A. (1996). Mg2þ as an extracellular signal: environmental regulation of Salmonella virulence. Cell, 84, 165– 174. 3. Fields, P. I., Groisman, E. A. & Heffron, F. (1989). A Salmonella locus that controls resistance to microbicidal proteins from phagocytic cells. Science, 243, 1059– 1062. 4. Miller, S. I., Kukral, A. M. & Mekalanos, J. J. (1989). A two-component regulatory system ( phoP phoQ) controls Salmonella typhimurium virulence. Proc. Natl Acad. Sci. USA, 86, 5054– 5058. 5. Soncini, F. C., Garcı´a Ve´scovi, E., Solomon, F. & Groisman, E. A. (1996). Molecular basis of the magnesium deprivation response in Salmonella typhimurium: identification of PhoP-regulated genes. J. Bacteriol. 178, 5092– 5099. 6. Garcı´a Ve´scovi, E., Ayala, M., Di Cera, E. & Groisman, E. A. (1997). Characterization of the bacterial sensor protein PhoQ. Evidence for distinct binding sites for Mg2þ and Ca2þ. J. Biol. Chem. 272, 1440– 1443. 7. Waldburger, C. D. & Sauer, R. T. (1996). Signal detection by the PhoQ sensor-transmitter. Characterization of the sensor domain and a response-impaired mutant that identifies ligand-binding determinants. J. Biol. Chem. 271, 26630– 26636. 8. Gunn, J. S., Hohmann, E. L. & Miller, S. I. (1996). Transcriptional regulation of Salmonella virulence: a PhoQ periplasmic domain mutation results in increased net phosphotransfer to PhoP. J. Bacteriol. 178, 6369– 6373. 9. Regelman, A. G., Lesley, J. A., Mott, C., Stokes, L. & Waldburger, C. D. (2002). Mutational analysis of the Escherichia coli PhoQ sensor kinase: differences with the Salmonella enterica serovar Typhimurium PhoQ protein and in the mechanism of Mg2þ and Ca2þ sensing. J. Bacteriol. 184, 5468– 5478. 10. Macfarlane, E. L. A., Kwasnicka, A., Ochs, M. M. & Hancock, R. E. W. (1999). PhoP-PhoQ homologues in Pseudomonas aeruginosa regulate expression of the outer-membrane protein OprH and polymyxin B resistance. Mol. Microbiol. 34, 305– 316. 11. Castelli, M. E., Garcia Vescovi, E. & Soncini, F. C. (2000). The phosphatase activity is the target for Mg2þ regulation of the sensor protein PhoQ in Salmonella. J. Biol. Chem. 275, 22948– 22954. 12. Chamnongpol, S. & Groisman, E. A. (2000). Acetyl phosphate-dependent activation of a mutant PhoP response regulator that functions independently of its cognate sensor kinase. J. Mol. Biol. 300, 291– 305. 13. Montagne, M., Martel, A. & Le Moual, H. (2001). Characterization of the catalytic activities of the PhoQ histidine protein kinase of Salmonella enterica serovar Typhimurium. J. Bacteriol. 183, 1787– 1791. 14. Soncini, F. C., Garcı´a Ve´scovi, E. & Groisman, E. A. (1995). Transcriptional autoregulation of the Salmonella typhimurium phoPQ operon. J. Bacteriol. 177, 4364– 4371. 15. Godson, G. N., Schoenich, J., Sun, W. & Mustaev, A. A. (2000). Identification of the magnesium ion binding site in the catalytic center of Escherichia coli primase by iron cleavage. Biochemistry, 39, 332– 339.

Mg2þ Sensing by the Mg2þ Sensor PhoQ of Salmonella

16. Brown, E. M. & MacLeod, R. J. (2001). Extracellular calcium sensing and extracellular calcium signaling. Physiol. Rev. 81, 239– 297. 17. Ridder, I. S. & Dijkstra, B. W. (1999). Identification of the Mg2þ-binding site in the P-type ATPase and phosphatase members of the HAD (haloacid dehalogenase) superfamily by structural similarity to the response regulator protein CheY. Biochem. J. 339, 223–226. 18. Zhou, T. & Rosen, B. P. (1999). Asp45 is a Mg2þ ligand in the ArsA ATPase. J. Biol. Chem. 274, 13854– 13858. 19. Schottler, S., Wende, W., Pingoud, V. & Pingoud, A. (2000). Identification of Asp218 and Asp326 as the principal Mg2þ binding ligands of the homing endonuclease PI-SceI. Biochemistry, 39, 15895 –15900. 20. Lesley, J. A. & Waldburger, C. D. (2001). Comparison of the Pseudomonas aeruginosa and Escherichia coli PhoQ sensor domains: evidence for distinct mechanisms of signal detection. J. Biol. Chem. 276, 30827– 30833. 21. Hsing, W. & Silhavy, T. J. (1997). Function of conserved histidine-243 in phosphatase activity of EnvZ, the sensor for porin osmoregulation in Escherichia coli. J. Bacteriol. 179, 3729– 3735. 22. Zhu, Y., Qin, L., Yoshida, T. & Inouye, M. (2000). Phosphatase activity of histidine kinase EnvZ without kinase catalytic domain. Proc. Natl Acad. Sci. USA, 97, 7808– 7813. 23. Snavely, M. D., Miller, C. G. & Maguire, M. E. (1991). The mgtB Mg2þ transport locus of Salmonella typhimurium encodes a P-type ATPase. J. Biol. Chem. 266, 815– 823.

807

24. Miller, J. H. (1972). Experiments in Molecular Genetics, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. 25. Ho, S. N., Hunt, H. D., Horton, R. M., Pullen, J. K. & Pease, L. R. (1989). Site-directed mutagenesis by overlap extension using the polymerase chain reaction. Gene, 77, 51 – 59. 26. Wo¨sten, M. M., Kox, L. F., Chamnongpol, S., Soncini, F. C. & Groisman, E. A. (2000). A signal transduction system that responds to extracellular iron. Cell, 103, 113 – 125. 27. Flego, D., Marits, R., Eriksson, A. R., Koiv, V., Karlsson, M. B., Heikinheimo, R. & Palva, E. T. (2000). A two-component regulatory system, pehRpehS, controls endopolygalacturonase production and virulence in the plant pathogen Erwinia carotovora subsp. carotovora. Mol. Plant Microbe Interact. 13, 447– 455. 28. Rather, P. N., Paradise, M. R., Parojcic, M. M. & Patel, S. (1998). A regulatory cascade involving AarG, a putative sensor kinase, controls the expression of the 20 -N-acetyltransferase and an intrinsic multiple antibiotic resistance (Mar) response in Providencia stuartii. Mol. Microbiol. 28, 1345– 1353. 29. Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F. & Higgins, D. G. (1997). The CLUSTAL X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucl. Acids Res. 25, 4876– 4882. 30. Hanahan, D. (1983). Studies on transformation of Escherichia coli with plasmids. J. Mol. Biol. 166, 557 –580.

Edited by I. B. Holland (Received 14 October 2002; accepted 01 November 2002)