Characterization of the bacterial sensor protein PhoQ. Evidence for distinct binding sites for Mg2+ and Ca2+.

The PhoP/PhoQ two-component regulatory system governs several virulence properties in the Gram-negative bacterium Salmonella typhimurium. The PhoQ protein is a Mg2+ and Ca2+ sensor that modulates transcription of PhoP-regulated genes in response to the extracellular concentrations of these divalent cations. We have purified a 146-amino acid polypeptide corresponding to the periplasmic (i.e. sensing) domain of the PhoQ protein. Mg2+ altered the tryptophan intrinsic fluorescence of this polypeptide whereas Ba2+, which is unable to modulate transcription of PhoP-regulated genes, did not. Mg2+ was more effective than Ca2+ at repressing transcription of PhoP-activated genes in vivo. However, maximal repression was achieved when both cations were present. An avirulent mutant harboring a single amino acid substitution in the sensing domain of PhoQ exhibited lower affinity for Ca2+ but similar affinity for Mg2+. Cumulatively, these experiments demonstrate that Mg2+ can bind to the sensing domain of PhoQ and establish the presence of distinct binding sites for Mg2+ and Ca2+ in the PhoQ protein.

The PhoP/PhoQ two-component regulatory system governs several virulence properties in the Gram-negative bacterium Salmonella typhimurium. The PhoQ protein is a Mg 2؉ and Ca 2؉ sensor that modulates transcription of PhoP-regulated genes in response to the extracellular concentrations of these divalent cations. We have purified a 146-amino acid polypeptide corresponding to the periplasmic (i.e. sensing) domain of the PhoQ protein. Mg 2؉ altered the tryptophan intrinsic fluorescence of this polypeptide whereas Ba 2؉ , which is unable to modulate transcription of PhoP-regulated genes, did not. Mg 2؉ was more effective than Ca 2؉ at repressing transcription of PhoP-activated genes in vivo. However, maximal repression was achieved when both cations were present. An avirulent mutant harboring a single amino acid substitution in the sensing domain of PhoQ exhibited lower affinity for Ca 2؉ but similar affinity for Mg 2؉ . Cumulatively, these experiments demonstrate that Mg 2؉ can bind to the sensing domain of PhoQ and establish the presence of distinct binding sites for Mg 2؉ and Ca 2؉ in the PhoQ protein.
Two-component regulatory systems often mediate the adaptive response of bacteria to new environmental conditions (1)(2)(3)(4). These systems generally consist of a sensor protein that, in response to specific chemical or physical signals, modifies the phosphorylation state of the second component, usually a transcription factor whose affinity for DNA is modulated by phosphorylation. Sensor proteins are usually conserved in their C-terminal, cytoplasmic domain, which mediates the phosphorylation/dephosphorylation of the cognate regulatory proteins (5). On the other hand, the N-terminal, periplasmic domain of sensors is often involved in signal sensing and, therefore, confers specificity to each system.
The PhoP/PhoQ two-component system governs several virulence properties in the Gram-negative bacterium Salmonella typhimurium (6,7). The PhoQ protein is a Mg 2ϩ sensor that, in the presence of millimolar concentrations of this divalent cation, represses transcription of some 25 different PhoP-regulated loci (8,9). Several of these genes are essential for growth in low Mg 2ϩ environments, consistent with Mg 2ϩ deprivation being the regulatory signal that activates the PhoP/PhoQ system (9). Ca 2ϩ and Mn 2ϩ can replace Mg 2ϩ to repress transcription of PhoP-activated genes, but other divalent cations, including Ni 2ϩ , Cu 2ϩ , and Ba 2ϩ , have no effect (8). The regulatory role of the PhoP/PhoQ system is not limited to Salmonella pathogenesis, because several PhoP-regulated loci are not essential for virulence in mice (10,11), and phoP-hybridizing sequences have been detected in a wide variety of non-pathogenic Gram-negative species (12).
The PhoQ protein features two transmembrane regions, a long cytoplasmic tail, and a large periplasmic domain rich in acidic residues that could be involved in binding divalent cations (Ref. 13; Fig. 1). Several lines of experimental evidence suggest that the regulatory effect of Mg 2ϩ results from direct conformational changes provoked in the periplasmic domain of the PhoQ protein. First, a protein chimera in which the periplasmic domain of PhoQ was replaced by the corresponding region of the osmolarity sensor EnvZ lost the capacity to respond to Mg 2ϩ (8). Second, Mg 2ϩ modifies the trypsin susceptibility of the PhoQ protein in vitro at the same concentrations that are required to repress transcription of PhoP-activated genes in vivo (8). Finally, the changes in trypsin susceptibility in the PhoQ protein were detected in spheroplasts, which indicates that soluble components from the periplasmic space are not necessary for the Mg 2ϩ -mediated effect (8).
In this paper, we demonstrate that the sensing domain of the PhoQ protein specifically binds Mg 2ϩ . Furthermore, we establish that the PhoQ protein has distinct binding sites for Ca 2ϩ and Mg 2ϩ , and we identify a mutant PhoQ protein that is differentially altered in its response to Ca 2ϩ . Our data indicate that the PhoP/PhoQ system mediates the response to changes in the environmental levels of Ca 2ϩ and Mg 2ϩ and suggest a model in which binding of these divalent cations to the periplasmic domain of the PhoQ protein promotes a conformation that is unfavorable for the activation of the PhoP protein.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Growth Conditions-S. typhimurium strains EG9065 (psiD::MudJ) and EG9564 (psiD::MudJ pho-24) have been described (8). Escherichia coli strain EG9649 is BL21[DE3] carrying plasmid pT7-7Qp. Plasmid pT7-7Qp, which harbors the DNA region encoding the sensor domain of PhoQ under the control of the T7 10 promoter, was constructed in two steps. First, a DNA fragment comprising the region encoding the sensing domain of PhoQ (codons 45-190) was generated by the polymerase chain reaction using primers F05, 5Ј-AGGAATTCGCCATGGATAAAACCACCTTTC-3Ј, and F06 5Ј-TGCAAGCTTACATATAGGAGCGTTTTAG-3Ј, and plasmid pEG9071 (14) as template. The polymerase chain reaction product was digested with NcoI and HindIII and cloned between the NcoI and HindIII sites of plasmid pMON5907 (a generous gift of Stephen Lee (15)) to form pEG7242. The primary sequence of phoQ in plasmid pEG7242 was confirmed by DNA sequence analysis. Second, pEG7242 plasmid DNA was digested with NcoI, the 5Ј-protruding ends were filled in with the Klenow fragment of DNA polymerase and dNTPs and then digested with HindIII. The resulting 449-base pair fragment was purified and cloned between the NdeI (filled in) and HindIII sites of plasmid pT7-7 to form pT7-7Qp. Plasmid pT7-7Qp harbors an ATG initiation codon 11 base pairs downstream of the ribosome binding site in pT7-7. The ATG is followed by the DNA sequence corresponding to codons 45-190 of phoQ and by the translation termination codon TAA.
Purification of the Sensor Domain of PhoQ-The sensor domain of the S. typhimurium PhoQ protein (residues 45-190 preceded by an N-terminal Met; PhoQp) was purified from E. coli strain EG9649. Expression of PhoQp was achieved by addition of 0.7 mM isopropyl-␤-D-thiogalactopyranoside to induce the DE3-encoded T7 RNA polymerase. Cells were pelleted, resuspended in 10 mM Tris (pH 8.0), and subjected to sonication. Cell debris was removed by centrifugation, and ammonium sulfate was added (55% saturation) to the supernatant. Following centrifugation, the pellet was resuspended in 10 mM Tris (pH 8.0), and the supernatant was combined with ammonium sulfate (90% saturation), and following centrifugation, the pellet was resuspended in 10 mM Tris (pH 8.0). Both resuspended pellets were passed through an Econo-Pac 10DG desalting column (Bio-Rad) in 100 mM NaCl, 10 mM Tris (pH 8.0) and then combined. The PhoQp fragment was further purified on a Waters high performance liquid chromatography system (flow rate (0.3 ml/min) and protein monitoring at 280 nm) using several chromatographic columns. First, we used Waters Protein Pak 300 SW 7.5 mm ϫ 30-cm gel filtration column. Fractions containing PhoQp (as determined by Coomassie Blue staining of SDS-polyacrylamide gel electrophoresis) were pooled and passed through an Econo-Pac 10DG desalting column (Bio-Rad) in 20 mM NaCl, 10 mM Tris (pH 8.0). These fractions were loaded onto an ion exchange column (Waters advanced purification glass column AP-1), and PhoQp was eluted using a 20 mM to 1.0 M NaCl linear gradient in 10 mM Tris (pH 8.0). The PhoQpcontaining fractions, which eluted at 510 mM NaCl, were pooled and passed through an Econo-Pac 10DG desalting column (Bio-Rad) in 10 mM KH 2 PO 4 , 10 mM Tris (pH 6.9). These fractions were loaded onto a Bio-Rad Bio-Scale ceramic hydroxyapatite CHT-5 type I column. PhoQp was eluded with a 10 -500 mM KH 2 PO 4 linear gradient in 10 mM Tris (pH 6.9). The PhoQp-containing fractions, which eluted at 410 mM KH 2 PO 4 , were pooled and passed through an Econo-Pac 10DG desalting column (Bio-Rad) in 10 mM Tris (pH 8.0).
The identity, purity, and concentration of the fragment were analyzed by electrospray mass spectrometry, quantitative amino acid anal-ysis, and N-terminal amino acid sequencing. The predicted molecular mass for PhoQp was 17,042.2 daltons and that determined by electrospray mass spectrometry was 17,040 daltons; the extinction coefficient determined from the amino acid analysis was ⑀ 280 ϭ 1.94 cm 2 mg Ϫ1 . Electrospray mass spectrometry, quantitative amino acid analysis, and N-terminal amino acid sequencing were performed by the Protein Chemistry Laboratory of the Washington University School of Medicine.
Fluorescence and Circular Dichroism Measurements-Circular dichroism studies of the PhoQp fragment were carried out in a Jasco J600A spectropolarimeter. The PhoQp sample was analyzed at a concentration of 0.25 mg/ml in 50 mM choline chloride, 10 mM Tris (pH 7.0) at 25°C from 195 to 250 nm. Intrinsic fluorescence was measured using a PTI (Alphascan) spectrofluorometer at a constant excitation wavelength of 280 nm, and emission spectra were collected from 300 to 400 nm. These experiments were carried out with the PhoQp fragment at a concentration of 0.026 mg/ml in 10 mM Tris (pH 7.0) and varying concentrations of monovalent (choline chloride, KCl, and NaCl) and divalent salts (MgCl 2 , CaCl 2 , and BaCl 2 ). The solution containing choline chloride was titrated with aliquots of the solution containing MgCl 2 , and the change in intrinsic fluorescence was monitored as a function of Mg 2ϩ concentration. At each titration step the concentration of PhoQp and the ionic strength remained constant, thus the fluorescence changes were not due to changes in the ionic strength of the medium. A buffer blank was subtracted from the spectrum under all conditions.

RESULTS AND DISCUSSION
Expression and Purification of the Sensing Domain of the PhoQ Protein-To examine the ability of the periplasmic domain of PhoQ to bind divalent cations, we purified a polypeptide consisting of residues 45-190 of the Salmonella PhoQ protein using a combination of chromatographic techniques as described under "Experimental Procedures." Amino acid analysis and N-terminal sequencing of the purified polypeptide revealed a composition that was consistent with that predicted from the nucleotide sequence of this portion of the phoQ gene with an additional N-terminal Met, which had been incorporated into the phoQ gene in the expression vector. Furthermore, electrospray mass spectrometry revealed a mass of 17,040 daltons for the purified polypeptide, which is consistent with a predicted mass of 17,042.2 daltons.
Mg 2ϩ and Ca 2ϩ Alter the Trp Fluorescence Pattern of the Sensing Domain of PhoQ-To investigate the ability of the periplasmic domain of PhoQ to bind divalent cations, we examined the circular dichroism (CD) and fluorescence spectra of this polypeptide, which harbors four Trp residues (Fig. 1). The CD spectra of PhoQp suggested the presence of both ␣-helix and ␤-sheet structures in this polypeptide. However, no changes were detected in the CD spectra in the presence of MgCl 2 (20 mM) or NaCl (50 mM). On the other hand, the intrinsic fluorescence at 335.0 nm of PhoQp was significantly modified in the presence of MgCl 2 ( Fig. 2A). Changes could be observed with as little as 7 mM MgCl 2 . CaCl 2 could also modify the fluorescence spectra of PhoQp (data not shown) but BaCl 2 , which is unable to repress transcription of PhoP-regulated genes in vivo, hardly modified the intrinsic fluorescence of this polypeptide when tested at concentrations up to 200 mM (Fig. 2B).
These results provide a direct demonstration that the periplasmic region of PhoQ specifically binds Mg 2ϩ and Ca 2ϩ . Moreover, they are consistent with our previous findings that Mg 2ϩ could alter the trypsinization pattern of the periplasmic domain of the PhoQ protein presented in spheroplasts. Furthermore, they are in agreement with those reported by Wald-burger and Sauer (18), who found that divalent cations promoted stabilization to urea denaturation of a similar PhoQ fragment derived from E. coli. Waldburger and Sauer (18), however, could not detect changes in the fluorescence spectra of their PhoQ-derived fragment when MgCl 2 (10 mM) was added. The differences in the fluorescence results may reflect the presence of different salts in the buffers used to fix the ionic strength: 0.2 M KCl in the Waldburger and Sauer's experiments and choline chloride in those reported here. Consistent with this hypothesis, we established that higher concentrations of MgCl 2 were required to promote fluorescence changes in PhoQp when KCl was present at 100 mM (data not shown). On the other hand, choline, which is not predicted to compete with Mg 2ϩ for binding to PhoQp due to its large size, did not change the intrinsic fluorescence of PhoQp when added up to 500 mM (data not shown). Alternatively or in addition, the differences in the fluorescence results could be due to the 28 amino acid differences that exist between the 146 residues that comprise the periplasmic regions of the Salmonella and E. coli PhoQ proteins.
The PhoQ Protein Has Distinct Binding Sites for Ca 2ϩ and Mg 2ϩ -We had established previously that Ca 2ϩ could repress expression of the PhoP-activated gene psiD with half-maximal repression attained at lower concentrations than those achieved with Mg 2ϩ (8). However, the Ca 2ϩ -mediated effect could have resulted from the combined actions of Ca 2ϩ and Mg 2ϩ , since the repressing role of Ca 2ϩ had only been investigated in the presence of 8 M Mg 2ϩ .
To establish the importance of Ca 2ϩ and Mg 2ϩ in the regulation of the PhoP/PhoQ system, we examined the expression of where v ij is the velocity in the absence (0) or presence (1) of Mg 2ϩ (i) or Ca 2ϩ (j), k 1 and k 2 are the binding affinities of Mg 2ϩ and Ca 2ϩ , and x and y are the Mg 2ϩ and Ca 2ϩ concentrations. These expressions were derived from application of linkage thermodynamics (24). A global analysis of the data in Fig. 3 gives the following best fit parameter values: for the wild-type strain, k 1 , 1.4 Ϯ 0.2 10 5 M Ϫ1 ; k 2 , 4 Ϯ 1 10 3 M Ϫ1 ; v 00 ϭ 985 Ϯ 98; v 10 ϭ 106 Ϯ 11; v 01 ϭ 0; and v 11 ϭ 0; and for the pho-24 mutant, k 1 , 1.5 Ϯ 0.4 10 5 M Ϫ1 ; k 2 , 6 Ϯ 1 10 2 M Ϫ1 ; v 00 ϭ 909 Ϯ 91; vv 10 ϭ 273 Ϯ 23, v 01 ϭ 0, and v 11 ϭ 0. The possibility of Ca 2ϩ and Mg 2ϩ competing for the same site was examined and ruled out from the fit of the data with both cations present. On the other hand, we cannot presently rule out more complicated models in which Ca 2ϩ and Mg 2ϩ bind with low affinity to the Mg 2ϩ -and Ca 2ϩ -binding sites, respectively. the psiD gene in the presence of different concentrations of MgCl 2 , CaCl 2 (with no added Mg 2ϩ ), or a mixture of these two salts. Repression of psiD transcription was achieved at lower concentrations of Mg 2ϩ than Ca 2ϩ , and maximal repression was attained when both cations were present (Fig. 3A). An analysis of these data is consistent with the existence of distinct binding sites for Mg 2ϩ and Ca 2ϩ in the PhoQ protein (see legend to Fig. 3). The apparent affinities of the PhoQ protein for Mg 2ϩ and Ca 2ϩ were 1.4 Ϯ 0.2 10 5 M Ϫ1 and 4 Ϯ 1 10 3 M Ϫ1 , respectively. These two sites do not appear to interact with one another because the repressing effect in the presence of both cations was equivalent to the sum of the individual repressing effects.
A Mutant PhoQ Protein That Is Less Responsive to Ca 2ϩ -A S. typhimurium strain harboring the pho-24 allele overexpresses several PhoP-activated genes and is attenuated for mouse virulence (19). We have established previously that this mutant harbors a single amino acid substitution, Thr 48 3 Ile, in the periplasmic domain of the PhoQ protein (8). To establish the sensitivity of the mutant PhoQ protein to Mg 2ϩ and Ca 2ϩ , we examined the transcriptional activity of the psiD gene in a strain harboring a chromosomal pho-24 allele. The mutant PhoQ protein had a lower affinity for Ca 2ϩ than the wild-type (Ka Ca 2ϩ: 6 Ϯ 1 10 2 M Ϫ1 versus 4 Ϯ 1 10 3 M Ϫ1 ; Fig. 3B). On the other hand, the affinity for Mg 2ϩ was virtually identical in the wild-type and mutant PhoQ proteins (Ka Mg 2ϩ: 1.5 Ϯ 0.2 10 5 M Ϫ1 versus 1.4 Ϯ 0.2 10 5 M Ϫ1 ); yet, the transcriptional activity of the psiD gene at repressing concentrations of Mg 2ϩ was higher in the pho-24 mutant than in the wild-type strain (Fig. 3B). Cumulatively, these results indicate that Thr 48 is required for normal Ca 2ϩ sensing and support a model in which the PhoQ protein has distinct sites for Mg 2ϩ and Ca 2ϩ .
Conclusions-The PhoQ protein represents the first and only example of a receptor that senses extracellular Mg 2ϩ and Ca 2ϩ . In the PhoP/PhoQ signal transduction cascade, these cations act as primary signaling molecules rather than their familiar roles as cofactors (20) and second messengers (21). We have now provided direct evidence that Mg 2ϩ and Ca 2ϩ specifically bind to the sensing domain of the PhoQ protein. We would like to suggest that binding of divalent cations promotes a conformation in the PhoQ protein that is unfavorable for the activation of the PhoP protein, resulting in transcriptional repression of PhoP-activated genes.
The demonstration that the PhoQ protein harbors distinct sites for Mg 2ϩ and Ca 2ϩ and that these cations act independently to modulate PhoQ activity imply that both, Mg 2ϩ and Ca 2ϩ , are physiologically relevant in the control of PhoP-regu-lated loci. This is further substantiated by the virulence attenuation that results from a single amino acid substitution in the periplasmic domain of the PhoQ protein that alters its response to Ca 2ϩ . Finally, the PhoQ protein may represent a new paradigm for divalent cation-binding proteins, since it does not have an EF-hand, a motif that is common among a large number of Ca 2ϩ -binding proteins (22), and shows no sequence similarity with the Ca 2ϩ -sensing receptor from mammalian parathyroid cells (23). Solving the crystal structure of the sensing domain of the PhoQ protein may reveal features of the divalent cationbinding sites that are not apparent from homology analysis and structural predictions of its primary amino acid sequence.