The Phosphatase Activity Is the Target for Mg2+Regulation of the Sensor Protein PhoQ in Salmonella*

The PhoP/PhoQ two-component system controls the expression of essential virulence traits in the pathogenic bacterium Salmonella enterica serovar Typhimurium. Environmental deprivation of Mg2+ activates the PhoP/PhoQ signal transduction cascade, which results in an increased expression of genes necessary for survival inside the host. It was previously demonstrated that the interaction of Mg2+ with the periplasmic domain of PhoQ promotes a conformational change in the sensor protein that leads to the down-regulation of PhoP-activated genes. We have now examined the regulatory effect of Mg2+ on the putative activities of the membrane-bound PhoQ. We demonstrated that Mg2+ promotes a phospho-PhoP phosphatase activity in the sensor protein. This activity depends on the intactness of the conserved His-277, suggesting that the phosphatase active site overlaps the H box. The integrity of the N-terminal domain of PhoQ was essential for the induction of the phosphatase activity, because Mg2+ did not stimulate the release of inorganic phosphate from phospho-PhoP in a fusion protein that lacks this sensing domain. These findings reveal that the sensor PhoQ harbors a phospho-PhoP phosphatase activity, and that this phosphatase activity is the target of the extracellular Mg2+-triggered regulation of the PhoP/PhoQ system.

The PhoP/PhoQ two-component system controls the expression of essential virulence traits in the pathogenic bacterium Salmonella enterica serovar Typhimurium. Environmental deprivation of Mg 2؉ activates the PhoP/PhoQ signal transduction cascade, which results in an increased expression of genes necessary for survival inside the host. It was previously demonstrated that the interaction of Mg 2؉ with the periplasmic domain of PhoQ promotes a conformational change in the sensor protein that leads to the down-regulation of PhoP-activated genes. We have now examined the regulatory effect of Mg 2؉ on the putative activities of the membrane-bound PhoQ. We demonstrated that Mg 2؉ promotes a phospho-PhoP phosphatase activity in the sensor protein. This activity depends on the intactness of the conserved His-277, suggesting that the phosphatase active site overlaps the H box. The integrity of the N-terminal domain of PhoQ was essential for the induction of the phosphatase activity, because Mg 2؉ did not stimulate the release of inorganic phosphate from phospho-PhoP in a fusion protein that lacks this sensing domain. These findings reveal that the sensor PhoQ harbors a phospho-PhoP phosphatase activity, and that this phosphatase activity is the target of the extracellular Mg 2؉ -triggered regulation of the PhoP/PhoQ system.
Reversible protein phosphorylation is one of the most conspicuous mechanisms that regulate biological processes in cells, including modulation of enzymatic activities, proteinprotein and protein-DNA interactions, and extracellular signal transduction. In prokaryotes and in lower eukaryotes and plants (1), the most widespread and efficient sensory-response devices rely on protein phosphotransfer. They are the so-called two-component regulatory systems that enable bacteria to monitor changes in their environment and adjust their structure and physiology accordingly to survive.
The prototypical architecture of the two-component regulatory systems consists of a sensor protein and an associated effector protein (2,3). The sensor is generally a membrane protein whose C-terminal domain projects into the cytoplasm. This domain harbors a histidine autokinase activity and, in several sensors of this large family, a phosphatase activity (2, 4 -9). The cognate response-regulator is most often a transcriptional regulator protein. Its N-terminal domain harbors a conserved aspartate that accepts the phosphate from the phosphohistidyl residue of the sensor protein. This modification is propagated to the C-terminal domain affecting its DNA binding properties (2).
Two major biochemical activities play an opposite role to balance the phosphorylation status of the response regulator: the autokinase activity of the sensor that defines the phosphotransfer availability and a specific phosphatase that dephosphorylates the response regulator. By modulating these reactions, the sensor component defines the phosphorylation status of the response regulator, which directs the expression of a specific set of target genes. This modulation results in the final adaptive response of the bacteria to its primary signal. Despite a considerable amount of research on different two-component systems, carried out to understand how the signal regulates the catalytic activities of the sensor to control the phosphorylation state in the effector molecule, a clear picture of this mechanism has not yet emerged.
In Salmonella, the PhoP/PhoQ two-component system governs the adaptation to environmental Mg 2ϩ deprivation and controls the expression of essential virulence factors (10 -14). It was previously demonstrated that PhoQ is the sensor of the system that specifically recognizes extracellular Mg 2ϩ . The interaction of the sensing periplasmic domain with the divalent cation promotes a conformational change in the protein that results in the repression of the transcription of at least 20 different PhoP-activated loci (10,11,15). However, it was not clear which activity or activities that reside in PhoQ were affected by the Mg 2ϩ -induced conformational switch of the sensor.
In this work we examined how Mg 2ϩ , as the specific regulatory signal, controls the phosphorylation state of PhoP. We demonstrated that the interaction of Mg 2ϩ with PhoQ activates a phosphatase activity of the sensor protein that promotes the dephosphorylation of phospho-PhoP. Stimulation by Mg 2ϩ of this phosphatase activity required the N-terminal domain of the sensor, because no phosphatase was detected in a fusion protein lacking this domain while it retained the reversible autokinase and phosphotransfer activities. These results reveal that the sensor PhoQ harbors a phosphatase activity that acts on PhoP and that this phosphatase activity is the target of the Mg 2ϩ -triggered regulation of the PhoP/PhoQ system. * This work was supported in part by grants from the Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET; Project PIP 0849) and from Agencia Nacional de Promoción Científica y Technológica (Project 01-0000-00409) (to F. C. S.), and by a grant from the Third World Academy of Sciences, Trieste, Italy (to E. G. V.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 (16) or with the vector pUHE21-2lacI Q (16). These strains were used for the isolation of membranes harboring PhoQ or the control membranes, respectively. Escherichia coli strain PB1277 is BL21[DE3] carrying plasmid pPB1020. pPB1020 harbors a phoP His-tag fusion gene (full-length phoP fused to 6 His codons in its C terminus) under the control of the T7 10 promoter of the pT7-7 plasmid. This fusion gene was generated by polymerase chain reaction, using forward primer PhoP-NTF (5Ј-GAG-GATCCATATGATGCGCGTACTGG-3Ј) and reverse primer PhoP-H6-CTR (5Ј-TCCAAGCTTAGTGGTGGTGGTGGTGGTGGCGCAATTCAA-AAAGATATC-3Ј), and cloned between the NdeI and HindIII sites of plasmid pT7-7. The phoP His-tag fusion gene cloned into pUHE21-2lacI Q complemented a Salmonella phoP Ϫ strain to the same extent as the wild-type phoP. This shows that the addition of the His-tag to the PhoP C terminus does not affect its properties. 1 pPB1021 is a pMAL-c2 (New England Biolabs, Beverly, MA) derivative that encodes for the maltose binding protein fused to the C-terminal cytoplasmic region of PhoQ (MBP-Qc). The gene fusion was constructed by subcloning the phoQ fragment excised from pEG9050 encompassed between the StuI and the HindIII sites (which encodes from the Pro-220 to the C-terminal Glu-487 of PhoQ) into the vector pMAL-c2 between the XmnI and HindIII sites, respectively. Bacteria were grown at 37°C in Luria-Bertani (LB) 2 broth with shaking, and with the addition of 0.7 mM isopropyl-␤-D-thiogalactopyranoside (IPTG) when indicated. Ampicillin was used at a final concentration of 50 g/ml.
Genetic and Molecular Biology Techniques-Plasmid DNA was introduced into bacterial strains by electroporation using a Bio-Rad apparatus as recommended by the manufacturer. Recombinant DNA techniques were performed according to standard protocols (17). To replace His-277 with Val in PhoQ, and Asp-55 with Ala in PhoP, we used the site-directed mutagenesis protocol described by Deng et al. (18). Polymerase chain reaction-derived constructions and site-directed mutagenesis were all confirmed by DNA sequence analysis performed using the femtomole DNA sequencing system as recommended by the manufacturer (Promega).
Preparation of Membranes Enriched in PhoQ-Overnight cultures of S. typhimurium strains EG5172/pEG9050 and EG5172/pUHE21-2lacI Q were used to inoculate LB media containing 50 g/ml ampicillin. They were then grown at 37°C to logarithmic phase (OD 0.6) and induced by addition of 0.7 mM IPTG for an additional 3 h with shaking. Cells were collected, washed once with 10 mM Tris-HCl (pH 8.0), and resuspended in a solution containing 20 mM Tris-HCl (pH 8.0), 20% sucrose, 5 mM M EDTA, and 150 g/ml lysozyme. After 40-min incubation at 4°C, 20 mM MgCl 2 was added and cells were centrifuged for 20 min at 18,000 ϫ g. The pellet was resuspended in ice-cold 10 mM Tris-HCl (pH 8.0) and subjected to sonication. Nonruptured spheroplasts were removed by a brief centrifugation at 7000 ϫ g, and the membrane fraction was recovered after a 45-min centrifugation at 30,000 ϫ g. The membranes were resuspended in 10 mM Tris-HCl (pH 8.0), 2 M KCl and centrifuged 30 min at 30,000 ϫ g. The supernatant was discarded, and the pellet was resuspended in 10 mM Tris-HCl (pH 8.0), 5 mM EDTA. Finally, the membranes were washed twice with 10 mM Tris-HCl (pH 8.0) and resuspended in the same buffer at a final protein concentration of 10 mg/ml. All procedures were carried out at 4°C. Protein concentration was determined by the bicinchoninic acid assay (Bio-Rad) using bovine serum albumin as standard.
Protein Purification Protocols-The His 6 -tagged fusion protein PhoP-H6 was purified from strain PB1277. Expression of PhoP-H6 was achieved by addition of 0.7 mM IPTG to induce the DE3-encoded T7 RNA polymerase. Cells were pelleted, resuspended in sonication buffer (50 mM sodium phosphate (pH 6.0), 300 mM NaCl) and subjected to sonication. Cell debris was removed by centrifugation, and the supernatant was passed through a Ni 2ϩ -NTA-agarose affinity column equil-ibrated with 50 mM sodium phosphate (pH 6.0), 300 mM NaCl, 10% glycerol. The PhoP-H6 protein was recovered by elution with 0.5 M imidazole and dialyzed exhaustively against 25 mM Tris-HCl (pH 8.0), 50 mM KCl. The MBP-Qc fusion protein was purified from the E. coli strain TB1 (obtained from the pMAL Protein Fusion and Purification System, NEB, Inc.) transformed with plasmid pPB1021. Expression of MBP-Qc was achieved by addition of 0.3 mM IPTG to a log-phase cell culture in LB broth. After sonication and centrifugation, the soluble cellular extract containing the MBP-Qc protein was collected and subjected to amylose affinity chromatography analysis following the manufacturer's instructions (NEB, Inc.). All procedures were carried out at 4°C. The protein profile of the purified proteins was determined by SDS-polyacrylamide gel electrophoresis (PAGE).
In Vitro Phosphorylation, Dephosphorylation, and TLC Analysis-For the autokinase assay, membranes (50 g of total protein) harboring PhoQ were incubated with 50 M [␥-32 P]ATP (7500 cpm/pmol, NEN Life Science Products), in a 30-l reaction mixture containing 20 mM Tris-HCl (pH 8.0), 50 mM KCl (buffer TK). Unless stated otherwise, 1 mM MgCl 2 was added to the reaction medium. Reactions were started by addition of the membrane fraction, incubated 5 min at 37°C, and stopped by addition of 6 l of 5 ϫ SDS-PAGE sample buffer (2.5% ␤-mercaptoethanol, 9% glycerol, 10% SDS, 600 mM Tris-HCl (pH 6.8), 0.006% bromphenol blue). The amount of radiolabeled PhoQ increased proportionally to the amount of membrane protein used in the range from 0.16 to 2.6 g/l; this indicated that PhoQ was not in excess at 1.66 g/l, the membrane concentration used in the autophosphorylation assay. To remove remnant Mg 2ϩ , ATP, and free inorganic phosphate, the membranes containing autophosphorylated PhoQ were washed once with buffer TK with 5 mM EDTA, followed by two washes with buffer TK. This fraction was analyzed by TLC to confirm absence of residual ATP and inorganic phosphate and used to test the stability of phosphorylated PhoQ and for the PhoQ 3 PhoP phosphotransfer assay. The phosphotransfer assay was carried out as follows: typically, membranes (50 g of total protein) harboring phosphorylated PhoQ were coincubated in a 30-l total volume with 10 g of purified PhoP-H6 in a reaction buffer consisting of 20 mM Tris-HCl (pH 8.0) and 50 mM KCl for different periods of time. MgCl 2 was added to the reaction as indicated in each individual assay. Reactions were stopped by addition of 5 ϫ SDS-PAGE loading buffer. To obtain isolated phosphorylated PhoP-H6, 10 g of purified PhoP-H6 was added to a 30-l final volume containing membranes (50 g of total protein) harboring PhoQ, 50 M [␥-32 P]ATP (7500 cpm/pmol), and 1 mM MgCl 2 in buffer TK, and incubated for 5 min at 37°C. This reaction mixture was centrifuged at 24,000 ϫ g for 45 min at 4°C to remove the membrane fraction. The supernatant containing the phosphorylated PhoP-H6 was recovered, purified using a Ni 2ϩ -NTA agarose affinity column to eliminate remnant ATP, applied to a G-50 gel filtration column equilibrated with buffer TK to remove imidazole, and immediately used to test phosphorylated PhoP stability or for the phosphatase assay. For the phosphatase assay, phosphorylated PhoP-H6 (5 g of protein) was incubated with control membranes, membranes enriched in PhoQ or in PhoQ H277V (50 g), or with purified MBP-Qc fusion protein (5 g) in a 30-l reaction mixture containing 20 mM Tris-HCl (pH 8.0), 50 mM KCl, at 37°C, in the presence or absence of 1 mM MgCl 2 for different periods of time as indicated in each assay. The reaction was stopped by adding 5 ϫ SDS-PAGE loading buffer. All reactions were analyzed by SDS-PAGE (12% polyacrylamide), transferred to nitrocellulose, and then subjected to autoradiography. When TLC analysis was performed, reactions were stopped by addition of 1% SDS, applied to a polyethyleneimine (PEI)cellulose plate (J. T. Baker), and developed in 0.8 M LiCl, 0.8 M acetic acid as described (19). The plates were air-dried and then exposed to autoradiography films. Additionally, the stability of phospho-PhoP-H6 or phospho-PhoQ was determined by incubating each phosphoprotein in buffer TK with addition of 1 mM EDTA or 20 mM MgCl 2 , at 37°C. Aliquots were withdrawn at different time points, and the reaction was stopped by adding 5 ϫ SDS-PAGE loading buffer. The samples were analyzed by SDS-PAGE and transferred to nitrocellulose membranes, followed by autoradiography. Autoradiographies from SDS-PAGE analysis or TLC assays were densitometrically scanned to perform quantitative determinations. For the stability assays, the PhoP or PhoQ bands were cut from the gel and the incorporation of 32 P was determined using a Wallac 1209 Rackbetta liquid scintillation counter. Chemical stability assays of the phosphoproteins (alkali and acid lability) were performed as described (9).

RESULTS
PhoQ Autokinase Activity-Binding of Mg 2ϩ to the sensor protein PhoQ was identified as the primary event in the signal transduction cascade that results in the down-regulation of the PhoP-activated genes (10,15). To determine the activities of PhoQ that are affected in response to the signal, we sought to analyze sequentially the steps that are involved in the transduction mechanism.
To examine in vitro the biochemical activities of the Mg 2ϩ sensor PhoQ, this protein was expressed from the pEG9050 plasmid. Cellular fractionation was performed to isolate the membranes that harbor the sensor protein. The electrophoretic analysis of the membrane fraction revealed a major protein band (estimated to be 5-7% of the total protein by densitometric analysis) that corresponded to the predicted mobility of PhoQ. The identity of the protein was assessed also by Western blot analysis using polyclonal antibodies raised against the cytoplasmic C-terminal domain (data not shown). To avoid conformational distortions or artifacts that would result from the solubilization of the membrane protein with detergents, all experiments involving PhoQ were performed with the sensor protein in its native membrane environment. This approach ensures an accurate response to different variables in the reaction medium that mimic the environmental cues that control the system.
According to the two-component paradigm, PhoQ would first undergo autophosphorylation in a conserved histidine residue, and the phosphate would in turn be transferred to a conserved aspartate residue in the cognate transcriptional regulator PhoP. Indeed, we showed that incubation of the membranes with [␥-32 P]ATP resulted in autophosphorylation of the sensor protein. This autokinase activity was dependent on the presence of micromolar concentrations of Mg 2ϩ , because no phosphorylation was detectable in membranes devoid of the divalent cation after exhaustive washes with EDTA (Fig. 1A). When Mg 2ϩ concentrations above 200 M (demonstrated to downregulate the system (10, 11)) were used, the divalent cation had no inhibitory effect on the autokinase activity of PhoQ. This result showed that micromolar concentrations of Mg 2ϩ are required for the autophosphorylation reaction, suggesting that Mg 2ϩ -ATP is the true substrate of the autokinase. Additionally, we determined that the acid lability and alkali resistance of the phosphorylated moiety corresponded to the chemical characteristics of a phosphohistidine residue (data not shown). Moreover, the PhoQ H277V mutant protein, where the conserved His-277 is replaced by Val, was unable to undergo autophosphorylation, irrespective of the presence of Mg 2ϩ in the phosphorylation reaction (data not shown). This indicates that the phosphorylation takes place in the predicted conserved histidine residue.
Phosphotransfer from PhoQ to PhoP-We decided to investigate the next step in the activation by phosphorylation of the response regulator PhoP. To uncouple the autokinase from the phosphotransfer reaction and to analyze the effect of Mg 2ϩ exclusively on the latter, the membrane-bound PhoQ was first subjected to autophosphorylation. Immediately after, the membranes were exhaustively washed with EDTA to deplete the membranes of remnant Mg 2ϩ and ATP. The membranes harboring phospho-PhoQ were incubated with purified PhoP resulting in the phosphorylation of the response regulator. Chemical stability tests and site-specific mutagenesis identified amino acid Asp-55 of PhoP as the phosphorylated residue (not shown).
As it is shown in the time course assays performed in the presence of different concentrations of Mg 2ϩ (Figs. 2 and 3), the initial rate of phosphotransfer is favored by increasing the concentration of the divalent cation. Strikingly, Mg 2ϩ concentrations higher than 250 M stimulated the dephosphorylation of phospho-PhoP along the time course (Figs. 2 and 3B). A net loss of the protein-bound radiolabeled phosphate becomes evident when we compare the radioactivity associated to PhoQ (Figs. 2 and 3A) and to PhoP (Figs. 2 and 3B) as a function of time, for each Mg 2ϩ concentration used. Moreover, the total protein-bound phosphate decreased up to 60% when concentrations of MgCl 2 higher than 250 M were added to the phospho- transfer reaction and up to 90% at concentrations higher than 10 mM MgCl 2 (Figs. 2 and 3C). On the other hand, when the stability of each individual phosphorylated protein was assayed, phospho-PhoQ retained a 72 Ϯ 13% of the initial label over the 120-min period tested (irrespective of the MgCl 2 concentration used), whereas phospho-PhoP showed a half-life of 64 min in the presence of 1 mM EDTA and of 60 min in the presence of 20 mM MgCl 2 (Fig. 4).
The above results indicate that Mg 2ϩ neither activates an autophosphatase activity of the response regulator nor promotes PhoQ dephosphorylation and strongly suggest that a Mg 2ϩ -induced phosphatase is present and acts on phosphorylated PhoP.
PhoQ Harbors a Mg 2ϩ -controlled Phosphatase Activity-The status of the environment as detected by the sensor is reflected by the phosphorylation state of the effector molecule. The phosphorylation status of PhoP is predicted to be due to the balance of three fundamental biochemical reactions: PhoQ autophosphorylation, the phosphotransfer from PhoQ to PhoP, and the dephosphorylation of phospho-PhoP that restores the effector protein to its original state.
To examine if PhoQ was responsible for the Mg 2ϩ -induced dephosphorylation of PhoP, radiolabeled phospho-PhoP was incubated with membranes harboring unlabeled PhoQ. Surprisingly, in the absence of added Mg 2ϩ we observed dephosphorylation of PhoP with the simultaneous phosphorylation of PhoQ, indicating that a reverse phosphotransfer from PhoP to PhoQ took place (Fig. 5A). The reverse phosphotransfer was partially inhibited by EDTA, showing that a low concentration of the divalent cation favors the reaction. However, when we added 1 mM MgCl 2 to the reaction medium, dephosphorylation of phospho-PhoP took place but no phosphorylated PhoQ was detected, resulting in a net loss of the total label associated to proteins (Fig. 5A). This effect was not due to a PhoQ-unrelated phosphatase activity present in the membrane fraction, because membranes lacking the sensor protein did not show Mg 2ϩ -stimulated phosphate release from phospho-PhoP (Fig.  5A). Moreover, this result was consistent with the PhoQ-mediated dephosphorylation of PhoP triggered in the presence of high concentrations of Mg 2ϩ previously observed in the direct phosphotransfer assay (Figs. 2 and 3).
Although controlled proteolysis of native PhoQ and binding assays performed using its purified sensing domain demonstrated that extracellular Mg 2ϩ interacts with the periplasmic domain of the sensor, thereby triggering a conformational change in the protein (10,15), it was still unclear how this change was transduced in the modulation of the biochemical activities of PhoQ. To assess the role of the periplasmic sensing domain of PhoQ in the Mg 2ϩ -induced phosphatase activity, we constructed a fusion protein containing its cytoplasmic domain fused to the maltose binding protein (MBP-Qc). MBP-Qc was able to undergo autophosphorylation (not shown) and to receive the phosphate from PhoP analogously to wild-type PhoQ (Fig.  5A). However, the addition of Mg 2ϩ did not promote the dephosphorylation of PhoP. A comparative densitometric analysis of a reverse phosphotransfer time course from phospho-PhoP to either PhoQ or to MBP-Qc in the presence of 1 mM MgCl 2 is shown in Fig. 5B. A net loss of label associated to the proteins was evident only when wild-type PhoQ was used (Fig. 5C).
When we analyzed by TLC the dephosphorylation of phospho-PhoP in the presence of control membranes, membranes harboring PhoQ, or MBP-Qc, we confirmed that the detected loss of label shown in Fig. 5 corresponded to the release of inorganic phosphate (Fig. 6). This release was significantly enhanced only when wild-type PhoQ and 1 mM MgCl 2 were present in the reaction (Fig. 6A). The densitometric analysis shows the relative distribution of the total label between the phosphate bound to the proteins and the inorganic phosphate FIG. 4. Effect of Mg 2؉ on the stability of the phosphorylated PhoQ and PhoP proteins. Membranes harboring phosphorylated PhoQ depleted of Mg 2ϩ , ATP, and inorganic phosphate (0 min) were split into two equal fractions and incubated in buffer TK, with addition of 1 mM EDTA or 20 mM MgCl 2 , respectively. Aliquots were taken at 10, 30, 60, and 120 min, and 5ϫ SDS-PAGE sample buffer was added to stop the reaction. Samples were analyzed by SDS-PAGE, followed by transfer to nitrocellulose and autoradiography (PhoQ-P). An identical protocol was carried out using phospho-PhoP free of ATP and inorganic phosphate as indicated under "Experimental Procedures" (PhoP-P).
FIG. 3. Densitometric analysis of the PhoQ 3 PhoP phosphotransfer reaction. Autoradiographs from three independent assays like the one shown in Fig. 2 were densitometrically scanned, and the total amounts of phospho-PhoQ (A), phospho-PhoP (B), and protein-bound phosphate (C) were plotted. The average results are expressed as the ratio between the indicated phosphorylated protein and the total phospho-PhoQ present at the initial time of the phosphotransfer reaction (0 min). released in each sample (Fig. 6B). More than 80% of the total label was released as inorganic phosphate by membrane-bound PhoQ in the presence of MgCl 2 , whereas less than 15% of the label was detected as inorganic phosphate in the absence of the cation. On the other hand, regardless of the concentration of Mg 2ϩ used in the assay, more than 80% of the total radioactivity remained protein-bound when phospho-PhoP was incubated with either control membranes or MBP-Qc.
Cumulatively, these results demonstrate that Mg 2ϩ triggers a specific PhoP-phosphatase activity in PhoQ that renders a dephosphorylated response regulator. Moreover, they show that the N-terminal sensing domain of PhoQ plays a fundamental role of in the Mg 2ϩ -activated PhoP-phosphatase activity of PhoQ.
To assess the involvement of the conserved autophosphorylation site, His-277, in both the Mg 2ϩ -induced phosphatase activity and the reverse phosphotransfer, we used membranes harboring the mutant PhoQ H277V instead of the wild-type PhoQ. As expected, there was no reverse transfer of the label in any of the conditions tested, indicating that the conserved His-277 is the target residue that accepts the phosphate from phospho-PhoP (Fig. 5A). In addition, the mutant protein was unable to promote the dephosphorylation of PhoP even in the presence of added Mg 2ϩ . This points out that His-277 plays an essential role in the phosphatase activity of PhoQ.
Our experimental data showed that two processes are involved in the dephosphorylation of PhoP: the Mg 2ϩ -triggered phosphatase activity present in PhoQ and the reversion of the phosphotransfer between PhoP and PhoQ. To gain an insight into the latter, ADP was added to the reaction mixture. ADP stimulated the reverse phosphotransfer with subsequent dephosphorylation of PhoQ and regeneration of labeled ATP, as it is shown in the TLC analysis (Fig. 7). Formation of ATP was also achieved by addition of ADP to isolated phospho-PhoQ, and Mg 2ϩ did not stimulate the release of inorganic phosphate in this reaction (not shown). These results suggest that the PhoP phosphorylation state depends, at least partially, on the balance of a totally reversible phosphotransfer reaction between PhoQ and PhoP. Additionally, when Mg 2ϩ and ADP were simultaneously added to the reaction, there was dephosphorylation of phospho-PhoP with production of both ATP and inorganic phosphate (Fig. 7). This result shows the additive effect of the phosphatase activity and the reverse phosphotransfer taking place at the same time, implying that they are independent events. DISCUSSION According to the paradigm of the "two-component" regulatory systems, upon interaction with the environmental signal the sensor protein defines the phosphorylation state of its cognate effector protein. However, there is not a unique model that comprehensively depicts how this process is achieved, and different strategies arise from the analysis of individual systems. FIG. 5. Reverse phosphotransfer reaction and phosphatase activity associated to PhoQ. A, phospho-PhoP was incubated in the reaction buffer alone (Ϫ); in the presence of control membranes isolated from a strain containing the vector plasmid pUHE21-2 lacI Q (pUH); membranes harboring PhoQ (PhoQ); membranes harboring PhoQ H277V (Q H277-V ); or MBP-Qc (MBP-Qc), as described under "Experimental Procedures." The reactions were performed in buffer TK alone, with addition of 1 mM EDTA, or with addition of 1 mM MgCl 2 , as indicated. The reactions were allowed to proceed 10 min at 37°C and were stopped by addition of 5ϫ SDS-PAGE sample buffer. The phosphorylated proteins were separated by SDS-PAGE, transferred to nitrocellulose, and followed by autoradiography. B, phospho-PhoP was incubated with membranes harboring PhoQ or with the fusion protein MBP-Qc in the presence of 1 mM MgCl 2 as described under "Experimental Procedures." Aliquots of the reactions were withdrawn at 30 s, 1, 5, 10, and 30 min. The samples were separated by SDS-PAGE and transferred to nitrocellulose, followed by autoradiography. Quantitation of the phosphoproteins was done by densitometry. The total label on phospho-PhoP at the initiation of the reaction time (0 min) was considered 100%. Closed symbols correspond to phospho-PhoP during the incubation with membranes harboring PhoQ (squares) or MBP-Qc (circles). Open squares correspond to phospho-PhoQ and open circles correspond to phospho-MBP-Qc. C, the total amount of protein-bound phosphate during the reactions shown in B was plotted. Circles correspond to the assay of phospho-PhoP with MBP-Qc and squares to the assay of phospho-PhoP with PhoQ.
FIG. 6. Release of inorganic phosphate during the dephosphorylation of phospho-PhoP. Phospho-PhoP was incubated with control membranes (pUH), membranes harboring PhoQ (PhoQ), or MBP-Qc, without (Ϫ) or with the addition of 1 mM MgCl 2 . A, an aliquot from each reaction was spotted onto a PEI-cellulose TLC plate. Inorganic [ 32 P]phosphate was used as standard. Reactions were stopped by addition of 1% SDS, applied to a PEI-cellulose plate, and developed with 0.8 M LiCl, 0.8 M acetic acid, followed by autoradiography. B, autoradiographs from three independent assays like the one shown in A were densitometrically scanned. The average results are expressed as the percentage amount of phosphorylated protein relative to the total phospho-PhoP present at the initial time of the phosphotransfer reaction (0 min) considered as 100%. Open bars correspond to protein-bound phosphate (Protein-P); and solid bars correspond to inorganic phosphate (Pi).
For example, in systems that include OmpR/EnvZ (6,7,20), the Bacillus subtilis PhoP/PhoR (9), and CpxR/CpxA (21) the sensor protein simultaneously harbors the autokinase and the phosphatase activities. On the other hand, some response regulators exhibit an intrinsic phosphatase activity, and the sensor protein acts as a cophosphatase or enhancer of its autodephosphorylation (20,22,23). Additionally, in other systems accessory proteins either stimulate the phosphatase reaction provided by the sensor (i.e. PII proteins (8)) or act independently and dephosphorylate the phosphoaspartate in the response regulator (24 -26).
In the PhoP/PhoQ signal transduction mechanism, extracellular Mg 2ϩ controls the activity of the system acting as the specific ligand of the sensor protein PhoQ. It was demonstrated that Mg 2ϩ interacts with the periplasmic domain of PhoQ changing the conformation of the protein (10,15,27). However, it was not known how this conformational change affects the function of PhoQ, which is finally reflected in the down-regulation of the PhoP-activated genes. In this work, we demonstrate that Mg 2ϩ induces a specific phospho-PhoP phosphatase in the membrane-bound PhoQ sensor protein.
To analyze the effect of Mg 2ϩ on the putative activities of PhoQ, we first set up the conditions for the autokinase, PhoQ 3 PhoP phosphotransfer, and PhoP-phosphatase assays, using the membrane-bound sensor protein. We decided not to use detergents for the extraction of PhoQ, or to purify it as a fusion or truncated form, to avoid any modification that might alter the proper response to its specific signal. We demonstrate that PhoQ undergoes autophosphorylation in the presence of ATP and that the conserved His-277 is the target residue of this autokinase activity.
Autophosphorylated PhoQ serves as a phosphodonor of the effector protein PhoP, which becomes phosphorylated in the predicted Asp-55 residue. Mg 2ϩ is required for PhoQ autophosphorylation in micromolar concentrations. This is consistent with the Mg 2ϩ ⅐ATP chelate complex being the actual phosphodonor for the autokinase reaction. Also, the cation was required for the subsequent phosphotransfer to PhoP, as was shown for other two-component systems (28).
Interestingly, in the course of the phosphotransfer reaction, PhoP became phosphorylated and it rapidly lost the label when concentrations of MgCl 2 higher than 250 M were present in the reaction, pointing out that the divalent cation was somehow stimulating such dephosphorylation. We determined that two independent mechanisms of dephosphorylation of phospho-PhoP occur. One involves the reversion of the reaction that takes place to phosphorylate the response regulator, and the other is a specific phospho-PhoP phosphatase induced by high concentrations of Mg 2ϩ that renders the release of inorganic phosphate. When phospho-PhoP was incubated with PhoQ, a reverse phosphotransfer took place in the absence of added Mg 2ϩ . This transfer was further stimulated by ADP, showing that the PhoQ autokinase activity can be reversed, transferring the phosphate to ADP. The His-277 residue in the sensor protein was key to this mechanism, because the mutant PhoQ H277V was unable to receive the phosphate from phosphorylated PhoP. It is not clear whether this reverse phosphotransfer is a common mechanism shared by all two-component systems (see . In the PhoP/PhoQ system, the reversion of the kinase activity may operate as one of the mechanisms that control the phosphorylated state of PhoP. Because the steady state of this reaction can be altered by the ATP/ADP ratio, we can speculate that the energetic charge could be an ancillary cellular mechanism that controls this system. Although from our in vitro model we cannot assess the physiological conditions in which the reverse reaction might become relevant, it is clear that Mg 2ϩ is not promoting this dephosphorylation mechanism. Only when high concentrations of Mg 2ϩ were added simultaneously with phospho-PhoP and -PhoQ, we detected dephosphorylation of PhoP with release of inorganic phosphate. Additionally, we showed that no Mg 2ϩ -activated phospho-PhoP phosphatase was associated to membranes lacking PhoQ and that each individual phosphorylated protein (phospho-PhoP and phospho-PhoQ) was stable for more than 30 min regardless of the Mg 2ϩ concentration present in the assay. Taking also into account that PhoQ autokinase activity remained maximal and constant while increasing the MgCl 2 concentration over 200 M, we conclude that Mg 2ϩ induces a phosphatase activity in the membrane-bound PhoQ that dephosphorylates PhoP.
The fusion protein MBP-Qc retained the autokinase activity as well as its capability to accept the phosphate from phospho-PhoP, while it was unable to exert the phosphatase activity on phospho-PhoP at all Mg 2ϩ concentrations tested. This result is consistent with the loss of the capacity to be down-regulated by the signal previously demonstrated for a chimeric sensor protein harboring the N-terminal region of EnvZ fused to the cytoplasmic domain of PhoQ (10). Considering the evidence of both, it becomes clear that the interaction of Mg 2ϩ with the N-terminal sensing domain of PhoQ is essential for triggering its phospho-PhoP phosphatase activity but does not affect the autokinase activity of the sensor. Additionally, the mutant protein PhoQ H277V was ineffective as a phosphatase irrespective of the Mg 2ϩ concentration used, indicating that this activity structurally or functionally requires the intactness of the conserved His residue that is also the target for the autophosphorylation reaction.
Whereas these data are consistent with the fact that PhoQ harbors the described phosphatase activity, we cannot rule out the possibility that Mg 2ϩ induces in PhoQ a conformation that, upon interaction with phosphorylated PhoP, activates an autophosphatase in the latter. However, we favor the first hypothesis, because in the presence of Mg 2ϩ phospho-PhoP showed a half-life of approximately 60 min under nondenaturing conditions and the half-life of the regulators that exhibit autophosphatase activity ranges from seconds in CheY to a few minutes in NtrC and PhoB (31)(32)(33).
According to the data presented in this work, the PhoP/PhoQ FIG. 7. Effect of ADP on the reverse phosphotransfer reaction from phospho-PhoP to PhoQ. Phospho-PhoP was incubated with membranes harboring PhoQ without (Ϫ) or with the addition of 1 mM MgCl 2 (Mg 2ϩ ), in the presence or absence (Ϫ) of 1 mM ADP. The samples were spotted onto a PEI-cellulose TLC plate. Inorganic [ 32 P]phosphate and [␥-32 P]ATP were used as standards. Reactions were stopped by addition of 1% SDS, applied to a PEI-cellulose plate, and developed with 0.8 M LiCl, 0.8 M acetic acid, followed by autoradiography. system adjusts to the "single-regulation model" proposed by Ninfa (5). In this case the kinase is the activity that remains essentially constant while the stimuli simply elicits the phosphatase activity, and the conformational requirements for the two activities of the bifunctional state would not be mutually exclusive.
Under Mg 2ϩ -limiting conditions PhoQ would be in the kinase-dominant state. Because the system is transcriptionally autoregulated, this would increase the amount of both PhoP and PhoQ and result in the subsequent increased level of phospho-PhoP. This will, in turn, induce the expression of a set of genes whose products are known to be necessary for survival in Mg 2ϩ -limiting environments such as the Salmonella-containing vacuole inside macrophages (34). On the other hand, when the bacterium encounters an environment with a concentration of Mg 2ϩ in the millimolar range, the divalent cation would interact with the periplasmic domain of PhoQ shifting the balance toward a phosphatase-dominant state of the sensor protein. Under this condition PhoP would become dephosphorylated, and the expression of the PhoP-activated genes would be shut down. Maintenance of a basal level of expression of PhoP and PhoQ is driven by a PhoP-independent promoter in the phoPQ operon (16).
Although it is clear that more than one pathway may exist for the dephosphorylation of PhoP by PhoQ, this work explains the molecular basis of the modulation of PhoQ by Mg 2ϩ and places the regulatory key of the PhoP/PhoQ system in the control of a phosphatase activity in the sensor protein PhoQ.
Finally, from this model we envision that a detailed structural knowledge of the protein domain involved in the phosphatase activity of PhoQ and the characterization of the interaction of phospho-PhoP with the active site will provide new clues for understanding the signal transduction mechanism of two-component systems.