The H Box-harboring Domain Is Key to the Function of the Salmonella enterica PhoQ Mg2+-sensor in the Recognition of Its Partner PhoP*

In two-component signaling systems, the transduction strategy relies on a conserved His-Asp phosphoryl exchange between the sensor histidine kinase and its cognate response-regulator, and structural and functional consensus motifs are found when comparing either the diverse histidine kinases or response regulators present in a single cell. Therefore, the mechanism that guarantees the specific recognition between partners of an individual pair is essential to unequivocally generate the appropriate adaptive response. Based on sequence alignments with other histidine kinases, we dissected the Salmonella enterica Mg2+-sensor PhoQ in different subdomains and examined by in vivo and in vitro assays its interaction with the associated response regulator PhoP. This signal transduction system allows Salmonella to withstand environmental Mg2+ limitation by triggering gene expression that is vital throughout the infective cycle in the host. Using resonant mirror biosensor technology, we calculated the kinetic and equilibrium binding constants and determined that the His-phosphotransfer domain is essential for the PhoQ specific recognition and interaction with PhoP. Additionally, we show the role of this domain in the bimolecular transphosphorylation and provide evidence that this region undergoes dimerization.

Two-component systems, either orthodox or multistep phosphorelays, have been largely demonstrated to be the most widespread signal transduction mechanism in eubacteria, and a wealth of these systems have been examined by genomic homology analysis, genetic mutation screening, and partial biochemical characterization. Members of this family of proteins are also present in eukaryotes such as yeast, fungi, and plants (reviewed in Ref. 1). These systems play a central role in bacterial adaptation to extracellular conditions, acting as communication channels between the environment and the inner cellular functions.
Although there are a variety of modular arrays in the twocomponent system family of proteins, in essence, they consist of a sensor protein, typically a histidine kinase (HK) 1 responsible for the detection of a specific environmental signal that, through a His-Asp phosphotransfer mechanism, defines the activity state of its associated response regulator (RR). This RR ultimately controls the generation of an adequate adaptive response, most frequently by modulating gene expression (2,3). The growing information that proceeds from the analysis of the entire genomic sequences of different bacteria clearly shows that a single cell can harbor 30 -60 distinct His-Asp-related phosphotransfer signal transduction pairs (4 -7). In light of the complex network of homologous systems that populate a single bacterial cell, one critical and yet not well understood aspect of the mechanism is the basis for the specificity of the recognition between the HK and its cognate RR. Therefore, the characterization of the interaction between the two components is basic to understand the phosphotransfer reactions and to determine the essence for the fidelity in the recognition throughout the signaling process.
Two broad types of HK have been defined in terms of the relative location of their substrate His and surrounding residues (H box) and the ATP catalytic binding domain: class I and class II (8). In class I, the prevalent type among bacterial HK classically represented by the osmosensor EnvZ, the H box is immediately followed by the ATP-binding domain. In the vast majority of the orthodox systems in which the sensor protein belongs to class I, this protein is a transmembrane receptor (9). In the type II enzymes (whose only known member is the chemotaxis HK CheA), the H box is located in the P1 domain at the N terminus (10,11). NMR and x-ray diffraction studies generated detailed structural knowledge of a number of bacterial RR such as OmpR (12), CheY (13), NtrC (14), PhoB (15), NarL (16), and SpoOF (17). In parallel, structures of the catalytic or the His-harboring subdomains of HK that include EnvZ, PhoQ, ArcB, and CheA and a phosphorelay phosphotransferase, SpoOB, have also been elucidated (8, 18 -20). However, in terms of partner interaction, detailed structural analysis has only been provided for the interaction between CheY and CheA from Escherichia coli (21)(22)(23), and SpoOF and SpoOB from Bacillus subtilis (5,24,25), where amino acidic residues involved in the mating have been disclosed. It is worth pointing out that, as mentioned above, CheA is an atypical HK, whereas SpoOB, component of the Bacillus sporulation phosphorelay cascade, is not an HK and consists in an isolated phosphotransferase-module that lacks the catalytic ATP-binding domain, although exceptionally retaining the capacity to form a four-helix bundle superimposable onto the homologous EnvZ domain (4). Collectively, from these studies it becomes apparent that recognition determinants on both partners of a HK-RR pair are essential for the specific interaction.
In the pathogenic bacteria Salmonella enterica serovar Typhimurium (Salmonella typhimurium), the PhoP/PhoQ twocomponent system governs the adaptation to Mg 2ϩ limitedmedia (26 -28). The detection of this parameter allows Salmonella to discern extracellular versus intracellular environments and to pleiotropically control the expression of factors that determine the successful invasion and spread in the host (reviewed in Refs. 29 and 30). PhoQ is a class I HK anchored to the bacterial inner membrane by two hydrophobic membrane-spanning domains that delimit a periplasmic sensor region. High resolution structural information of PhoQ has only come out for E. coli PhoQ, where Marina et al. (19) described the crystal structure of the C-terminal catalytic region of the sensor that binds ATP. Additionally, previous work provided evidence about the Mg 2ϩ and Ca 2ϩ binding capacity of the S. typhimurium (28), E. coli (31), and Pseudomonas aeruginosa (32) PhoQ sensor domains. We recently showed that S. typhimurium PhoQ is a bifunctional sensor, demonstrating that the conformational change triggered by Mg 2ϩ upon interaction promotes a switch in the activity of the sensor from a kinase-dominant to a phosphatase-dominant state (33). This was later reinforced by results obtained by other groups (34,35). However, no studies have yet examined the molecular basis for the PhoP-PhoQ specific recognition in the transduction process cascade.
In this work we characterized the interaction between the cytoplasmic region of PhoQ and its partner PhoP using different in vivo (interference phenotypes expressing liberated sensor domains) and in vitro (affinity retention columns and resonant mirror biosensor technology) approaches. We demonstrate that the PhoQ His-phosphotransfer domain (also termed DHp, for dimerization histidine phosphotransfer, in other twocomponent systems (Ref. 10)) is responsible for the specific recognition and interaction with PhoP. Additionally, we provide evidence that this region undergoes dimerization.

EXPERIMENTAL PROCEDURES
Chemicals and Reagents-Nitrocellulose membranes were from Amersham Biosciences. [␥-32 P]ATP and [␣-32 P]ATP (3000 Ci/mmol) were obtained from PerkinElmer Life Sciences. The oligonucleotides were purchased from Bio-Synthesis, Inc. (Lewisville, TX.). Cell culture media reagents were from Difco, and chemicals were from Sigma.
S. typhimurium strain 14028s pcgG9283::MudJ (27) was transformed with pUHE-derived constructs or the vector as control, and used for the interference assays.
E. coli BL21[DE3] strain transformed with pT7-7-derived constructs or the vector as control was used for the expression of the PhoQ-derived proteins and subsequent cell fractionation (33). The soluble cell fractions were used for the affinity retention assays described below. E. coli M15/pREP4 strain was transformed with pQE-30 or pQE-31-derived constructs to express and affinity-purify the N-terminal His-tagged PhoQ-or EnvZ-derived proteins for the biosensor experiments.
To express and purify PhoP-H6 protein, E. coli BL21[DE3] carrying plasmid pPB1020 was used. 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 T710 promoter of the pT7-7 plasmid as described (33).
␤-Galactosidase Activity Assays-For the ␤-galactosidase activity assays, bacteria were grown overnight with shaking at 37°C in Luria-Bertani medium with addition of different concentrations of IPTG as indicated. Ampicillin was used at 50 g/ml final concentration in the bacterial growth media when appropriate. ␤-Galactosidase was determined in overnight cultures as described (37). Soluble or membrane protein extracts from these cultures were obtained by cell fractionation as described below, analyzed by Tricine-SDS-PAGE (16.5% polyacrylamide) (38) transferred to nitrocellulose and developed by Western blot with rabbit anti-PhoQ (C-terminal cytoplasmic domain) polyclonal antibodies.
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 (39). PCRderived constructions were confirmed by DNA sequence analysis performed using the fmol DNA sequencing system as recommended by the manufacturer (Promega Corp.).
Protein Purification and Cell Fractionation Protocols-E. coli BL21[DE3] strains transformed with pT7-7-derived plasmids were used to obtain soluble protein extracts containing overexpressed PhoQ-derived truncated proteins for the interaction assays in affinity retention columns. Briefly, overnight cultures were used to inoculate LB media containing 50 g/ml ampicillin. They were then grown at 37°C to logarithmic phase (optical density of 0.6), and induced by addition of 1.0 mM IPTG, for an additional 3 h with shaking. Cells were collected, washed once, and resuspended in a solution containing 25 mM Tris-HCl (pH 8.0), 50 mM KCl (TK buffer). Cells were subjected to sonication and centrifuged 30 min at 20,000 ϫ g, and the supernatant was collected. Protein concentration was determined by the bicinchoninic acid assay (Bio-Rad), using bovine serum albumin as standard. These supernatants were diluted to a final concentration of 2.0 mg/ml. All procedures were carried out at 4°C. The protein profile was determined by SDSpolyacrylamide gel electrophoresis (PAGE) analysis. All His 6 -tagged fusion proteins were purified as previously described (33).
Affinity Retention Columns-Soluble protein extracts obtained by cell fractionation of E. coli BL21[DE3] strains harboring the vector pT7-7 or overexpressing PhoQ LCyt , PhoQ Cyt , or PhoQ CA (400 g of total protein) were pre-incubated with purified PhoP-H6 (100 g of protein) in buffer TK, 30 min on ice. Each protein mixture was loaded onto a Ni 2ϩ -NTA resin column (0.3 ml), previously equilibrated with buffer TK and blocked with control soluble protein extract (600 g of total protein) to avoid nonspecific binding. Columns were washed with 3.0 ml of buffer TK, followed by 3.0 ml of buffer TK with 0.1 M imidazole, and finally the bound proteins were eluted with 150 l of buffer TK with 1.0 M imidazole. Control columns were performed with each individual protein fraction and treated following the same sequential steps described above. An aliquot of 20 l of each final elution fraction was analyzed by SDS-PAGE (12% polyacrylamide), transferred to nitrocellulose, and developed by Western blot with rabbit anti-PhoQ Cyt or anti-PhoP-H6 polyclonal antibodies.
All Western blots were probed with the first rabbit polyclonal antibody, as indicated, and developed by incubation with protein A conjugated with phosphatase, coupled to a chromogenic reaction using nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate as substrates.
In Vitro Phosphorylation and Phosphotransfer Assays-For the au-tokinase assay, His-tagged affinity-purified proteins (2.0 g of total protein) were incubated with 50 M [␥-32 P]ATP (7500 cpm/pmol, PerkinElmer Life Sciences), 10 min in a 30-l reaction mixture containing buffer TKM (25 mM Tris-HCl, 50 mM KCl, 1 mM MgCl 2 ). The phosphotransfer assay was carried out by co-incubating 2.0 g of PhoP and 2.0 g of PhoQ Cyt with [␥-32 P]ATP in a 30-l volume for 10 min, using the same reaction media described for the autokinase activity. For the transphosphorylation reactions, 2.0 g of each individual protein (PhoQ Cyt , PhoQ CA , PhoQ DHp , or PhoQ CytH277V ) or the indicated combination of two proteins, were incubated in a 30-l reaction mixture for 20 min, using the same reaction media described for the autokinase activity. All reactions were carried out at 37°C. The reactions were stopped by adding 5ϫ SDS-PAGE loading buffer (2.5% ␤-mercaptoethanol, 9% glycerol, 10% SDS, 600 mM Tris-HCl (pH 6.8), 0.006% bromphenol blue). All reactions were analyzed by Tricine-SDS-PAGE (16.5% polyacrylamide), transferred to nitrocellulose, and then subjected to autoradiography.
Direct Photoaffinity Labeling-The His-tagged PhoQ-derivatives PhoQ Cyt and PhoQ CA were photoaffinity-labeled with [␣-32 P]ATP using a method previously described (40). The reaction mixtures (40 l) contained each purified protein in a concentration of 0.3 g/l in buffer TKM with 5% glycerol, and 20 Ci of [␣-32 P]ATP. Reaction mixtures were incubated for 15 min on ice in the dark; after this period 20 l were removed and exposed to UV light for 5 min, on ice. UV light was generated by an ultraviolet lamp (model UVGL-55, UVP Inc.) at a distance of 5 cm from the sample. Treated and control samples were then subjected to Tricine-SDS-PAGE (16.5% polyacrylamide) followed by autoradiography.
Measurements of Equilibrium and Kinetic Constants-IAsys affinity sensor analysis experiments were conducted on the IAsys Plus apparatus (Affinity Sensors Ltd., Saxon Hill, Cambridge, UK). Purified proteins were covalently coupled to carboxymethyl dextran sensor chips (Affinity Sensors Ltd.) via free amino groups using N-hydroxysuccinimide/1-ethyl-3-(3-dimethylaminopropyl)carboxidiimide coupling chemistry at 22°C, using the protocol recommended by the manufacturer. His-tagged PhoP or His-tagged PhoQ Cyt were coupled to individual cuvettes giving a signal of 520 and 390 arc s, corresponding to 10.5 or 7.8 ng of protein, respectively. All binding reactions were carried out in buffer TKM plus 0.05% Tween 20, at 25°C, with constant stirring set up at 95%, and the data were collected at interval of 0.3 s. Ligate binding to immobilized ligand was monitored at multiple ligate concentrations, ranging 10-fold below to at least 10-fold above preliminary estimates of equilibrium dissociation constants (K D ) for each reaction. Kinetic analysis was performed using the FASTfit software (Affinity Sensors Ltd.), which provided the values of k on and k d at each ligate concentration. Second-order association rate constants (k a ) were then obtained from a linear regression plot of k on versus ligate concentration. Dissociation equilibrium constants (K D ) were then derived for each reaction from the equation: K D ϭ k d /k a . For the PhoQ Cyt -chip interaction, constants were obtained from equilibrium data. Binding curves were performed at several concentrations of PhoQ Cyt or PhoQ DHp (10fold below to at least 100-fold above preliminary estimates of K D ), and the ligate was allowed to bind for 30 min to obtain a saturating equilibrium response (Req). Bovine serum albumin (1.0 M) interaction was tested as a non-related control protein when using either PhoQ Cyt -or PhoP-chip, rendering basal level sensorgrams.

PhoQ-derived Proteins That Retain the DHp Domain
Interfere with the PhoP/PhoQ Transduction Cascade-Domain liberation has proved to be a useful in vivo tool when analyzing the functional architecture of multidomain proteins such as the histidine kinases CheA and UhpB (41)(42)(43). This approach is based on the inhibitory effect that a subcloned and overexpressed protein domain exerts when it interferes by competition with the interplay between the endogenously expressed partner proteins, provided an interaction takes place between them. Taking advantage of this approach, we used truncated proteins derived from the cytoplasmic domain of PhoQ to perform an in vivo screening for the protein subdomain that retained the capacity to interfere with the signaling cascade and thus to interact with its constituents. As is depicted in Fig. 1A, PhoQ LCyt spans the whole cytoplasmic region encompassing the putative linker (L), the H box-harboring region (DHp), and the catalytic ATP-binding domain (CA), PhoQ Cyt lacks the pu-tative L domain, PhoQ DHp contains the H box-harboring domain but lacks CA, and PhoQ CA harbors only the CA domain. All constructs were cloned into vector pUHE21-2lacI q and each resulting plasmid was introduced into the S. typhimurium 14028s pcgG9283::MudJ strain by electroporation. Protein expression was induced by addition of increasing concentrations of IPTG to the growth media as indicated in Fig. 1B. The expression levels of the reporter gene pcgG, a representative phoP-activated gene, was followed by measuring the ␤-galactosidase activity from the pcgG-lacZ transcriptional fusion. As shown in Fig. 1B,  PhoQ Cyt , and PhoQ DHp showed a 3.3-, 3.3-, 2.2-, 4.0-, and 2.6fold inhibition of pcgG expression, respectively (calculated at 0.3 mM IPTG induction), although PhoQ CA showed no effect on the reporter gene expression when compared with the strain that harbors the control vector. Cellular fractionation followed by immunodetection analysis showed that, with the exception of membrane-anchored PhoQ and PhoQ H277V , all PhoQ-derived proteins expressed partitioned into the soluble cell fraction (Fig. 1C). In the immunodetection assay PhoQ DHp appears as a faint band, although in Coomassie Blue-stained gels it showed the same intensity as the other truncated proteins, indicating that this protein did not react efficiently with the rabbit anti-PhoQ Cyt polyclonal serum used to develop the Western (this was also the case when we immunodetected a His-tagged Pho-Q DHp purified protein with this antibodies; data not shown). These results suggested that all PhoQ-derived proteins that retain the DHp domain were able to interfere with the transduction cascade that leads to gene activation. This interference effect could conceivably be attributed to different mechanisms that can simultaneously take place: (a) the dimerization or oligomerization of the endogenous PhoQ with the overexpressed proteins rendering non-functional complexes, (b) the sequestration of the chromosomally expressed PhoP by the overexpressed full-length PhoQ or its truncated derivatives, and (c) a phosphatase activity acting on phospho-PhoP. Although we cannot completely rule out the third possibility, we favor the interaction/sequestration alternative as being the dominant cause for down-regulation because: (a) when fulllength PhoQ is overexpressed pcgG inhibition is observed even when cells were grown in low Mg 2ϩ , a condition that turns down the phosphatase activity of PhoQ, and (b) when we expressed the mutant PhoQ H277V protein, which lacks the phospho-PhoP phosphatase activity (33), the inhibition of pcgG expression was almost identical to the one obtained for fulllength PhoQ (Fig. 1B).
PhoQ-derived Proteins Encompassing the DHp Domain Interact with PhoP in an Affinity Retention Assay-To concentrate in the analysis of the PhoP-PhoQ interaction, we examined the capacity of the soluble PhoQ-derived proteins to interact with PhoP. With this purpose we performed an in vitro binding assay using purified His-tagged-PhoP pre-incubated with soluble bacterial extracts containing truncated PhoQ-derived proteins ( Fig. 2A). To improve protein yields, phoQ-derived constructs were transferred and expressed under the control of the T7 10 promoter to the pT7-7 vector in E. coli BL21[DE3] and cell fractionation was performed as described under "Experimental Procedures." As is shown in Fig. 2B, the immunodetection analysis of the proteins that co-eluted with PhoP (detected also by Western blot; Fig. 3B) from Ni 2ϩ -NTA resin with imidazole showed that both PhoQ LCyt and PhoQ Cyt , but not PhoQ CA , were affinity-trapped with PhoP-H6. Controls were performed to exclude the possibility that the different truncated PhoQ-derived proteins interacted nonspecifically with the column matrix; although PhoQ Cyt and PhoQ CA alone did not show nonspecific binding to the Ni 2ϩ -NTA resin, Pho-Q LCyt showed a low nonspecific binding, which accounted for ϳ5% of the eluted protein, being the remnant 95% co-eluted specifically with PhoP-H6. PhoQ CytH277V was also tested and rendered an interaction behavior identical to the wild type PhoQ Cyt (data not shown). These results qualitatively indicated that the PhoQ derivatives that retain the H box-harboring domain exhibit the ability to interact with PhoP.
The PhoQ-derived Proteins Display Intact Functional Properties-Before performing biosensor assays, we examined whether the different PhoQ-derived proteins exhibited their putative biochemical activities. Our previous work demon-strated the autokinase and Mg 2ϩ -modulated phosphatase activities of the intact Salmonella enterica PhoQ. However, there was no biochemical demonstration that, as was shown for other sensors such as EnvZ (44), PhoR (45), or NtrB (46), PhoQ could be dissected in subdomains able to individually exert catalytic activities and reconstitute the functional protein when mixed. PhoQ-derived proteins were expressed as N-terminal His 6tagged fusions and affinity-purified using Ni 2ϩ -NTA columns, as described under "Experimental Procedures." Because the His-tagged PhoQ LCyt protein could not be efficiently expressed and purified, and in the previous experiments it had not revealed significant differences with PhoQ Cyt , it was not used in subsequent analysis. Fig. 3 shows the resultant autoradiograms from the autophosphorylation reactions of purified Pho-Q Cyt , and EnvZ Cyt (Fig. 3A), the PhoQ Cyt 3 PhoP phosphotransfer (Fig. 3B), and the transphosphorylation from PhoQ Cyt to the PhoQ DHp domain (Fig. 3C). When either PhoQ CA or PhoQ CytH277V were incubated with PhoQ DHp , phosphorylation occurred in this last domain, whereas PhoQ CA , PhoQ DHp , or PhoQ CytH277V did not autophosphorylate (Fig. 3C). Additionally, taking into account that the PhoQ CA domain harbors the conserved motifs that bind ATP leading to autophosphorylation of the sensor, its ability to bind [␣-32 P]ATP by UV cross-linking was assayed. Fig. 3D shows that PhoQ CA (and also PhoQ Cyt , used as control) retained the capacity to bind the nucleotide. Interestingly, bands that corresponded to the predicted PhoQ Cyt and EnvZ Cyt homodimers could be visualized in the autoradiograms obtained from the autokinase assay ( Fig. 3A; FIG. 2. PhoP-PhoQ interaction analysis by affinity retention chromatography. A, Coomassie Brilliant Blue-stained SDS-PAGE (12% acrylamide) showing the protein pattern that corresponds to the soluble extracts (50 g of total protein/well) from strain E. coli BL21[DE3] harboring pT7-7 (vector), pT7-7::phoQ LCyt (PhoQ LCyt ), pT7-7::phoQ Cyt (PhoQ Cyt ), or pT7-7::phoQ CA (PhoQ CA ), respectively, that were subsequently co-incubated with purified His-tagged PhoP. MWS, protein broad range molecular mass standards (New England Biolabs). B and C, protein extracts as shown in A were pre-incubated with a fixed amount of purified PhoP-H6, loaded onto Ni 2ϩ -NTA-agarose columns, and treated as described under "Experimental Procedures." The PhoP-H6 affinity-trapped proteins that co-eluted with imidazole were analyzed by SDS-PAGE (12% polyacrylamide) followed by transfer to nitrocellulose and developed using rabbit anti-PhoQ (Cterminal cytoplasmic domain) (B) or anti-PhoP polyclonal antibodies (C). The position of each protein is indicated by an arrow. the identity of the bands was confirmed by immunodetection; data not shown). Together, these results show that the PhoQderived proteins used in this work are biochemically functional.
Quantification of the Strength of the PhoP-PhoQ Interaction-To accomplish a detailed characterization of the PhoP-PhoQ interaction observed, we used resonant mirror biosensor technology that allows to quantify the strength of proteinprotein interactions. In this approach, real-time binding of a ligate to an immobilized ligand covalently attached to the dextran matrix of a sensor chip is monitored as a change in refractive index occurring at the surface of the biological layer and is directly proportional to the amount of bound protein (47). To carry out the biosensor assays, purified PhoP was first attached to the sensor chip and its interaction with the PhoQ-derived proteins was tested. Simple bimolecular interactions adequately described our data in all cases (using FASTfit program from IAsys; see "Experimental Procedures"). Table I shows association (k a ) and dissociation (k d ) rate constants, and dissociation equilibrium constant (K D ) obtained for each experiment assuming a pseudo-first-order kinetic behavior. Increasing concentrations of PhoQ Cyt were tested in the PhoP cuvette (Table  I, immobilized PhoP). From this data, a dissociation equilibrium constant (K D ) for the PhoP-PhoQ Cyt interaction was estimated to be 1.19 ϫ 10 Ϫ7 M. Interestingly, when we tested PhoQ DHp it retained the capacity to interact with PhoP, being its calculated K D ϭ 7.94 ϫ 10 Ϫ7 M, whereas PhoQ CA did not show appreciable interaction with PhoP (K D ϭ 9.59 ϫ 10 Ϫ5 M).
To assess the specificity of the obtained signals, we assayed PhoP interaction with EnvZ Cyt , the cytoplasmic region from the homologous S. typhimurium EnvZ sensor. The sensorgrams showed no detectable association of EnvZ Cyt with PhoP. In addition, PhoQ CytH277V showed kinetic binding constants similar to the ones obtained with PhoQ Cyt .
To perform counterpart experiments we immobilized Pho-Q Cyt to a new cuvette (Table I, immobilized PhoQ). In these assays the interaction with PhoP was not detected. Absence of interaction could be the result of steric hindrance of the interacting surface upon immobilization to the dextran matrix. However, we monitored the generation of the PhoQ Cyt -PhoQ Cyt complex when PhoQ Cyt was tested as ligate. The equilibrium interaction assay yielded a K D ϭ 1.62 ϫ 10 Ϫ7 M. When PhoQ DHp was used, interaction with PhoQ Cyt was also monitored, rendering a calculated K D ϭ 4.14 ϫ 10 Ϫ7 M. In these experiments dissociation of either the PhoQ Cyt -PhoQ Cyt or PhoQ Cyt -Pho-Q DHp complexes could be achieved by adding PhoP, and this dissociation was more effective than sole buffer addition (data not shown), indicating that PhoP could displace the PhoQ Cyt -PhoQ Cyt interaction. DISCUSSION In the intracellular milieu, where an estimated 40 distinct types of two-component pairs are differentially regulated by diverse input signals, crucial specific interaction instances assure the generation of an accurate response: the recognition of  3. In vitro activities of the purified PhoQ-derived truncated proteins. A, purified PhoQ Cyt or EnvZ Cyt (2.0 g of each protein, in a 30-l reaction mixture) were individually incubated to allow for the autophosphorylation reaction. Asterisk (*) shows the positions of homodimeric phosphorylated PhoQ Cyt and EnvZ Cyt proteins. B, purified PhoQ Cyt was incubated alone or together with PhoP (2.0 g of each protein, in a 30-l reaction mixture) to allow for the phosphotransfer reaction. C, purified PhoQ Cyt , PhoQ CA , or PhoQ CytH277V were incubated alone or in the presence of PhoQ DHp (2.0 g of each protein, in a 30-l reaction mixture) to allow for transphosphorylation. PhoQ DHp alone was incubated in the same reaction medium as control. Reactions were carried out for 10 min (A and B) or 20 min (C) at 37°C, in buffer TKM containing 50 M [␥-32 P]ATP (7500 cpm/pmol), as described under "Experimental Procedures." Samples were separated by Tricine-SDS-PAGE (16.5% polyacrylamide) and transferred to nitrocellulose, followed by autoradiography. D, purified PhoQ Cyt or PhoQ CA (0.3 g/l each, in a 40-l reaction mixture) were individually incubated for 15 min at 4°C, in buffer TKM containing 20 Ci of [␣-32 P]ATP. The reaction mixture was split, and half of the total volume was exposed 5 min to UV light, whereas the rest was incubated in the dark as described under "Experimental Procedures". Samples were separated by Tricine-SDS-PAGE (16.5% polyacrylamide) and transferred to nitrocellulose, followed by autoradiography. the signaling ligand by the proper sensor (in this case Mg 2ϩ binding to PhoQ), the pairing of the sensor to a unique matching response regulator (i.e. PhoQ to PhoP), and the binding of the response regulator to its specific cellular target (PhoP binding to regulatory regions in the DNA).
Previous work revealed different aspects of the Mg 2ϩ -PhoQ interaction such as the identification of the sensing dedicated domain and the characterization of the interaction with divalent cations (27,28,31,35,48). In this work we sued to identify the PhoQ domain that sustains the specific recognition of PhoP and examined the kinetic parameters that characterize the interaction.
In this work we showed by an in vivo domain liberation experiment that full-length PhoQ, the cytoplasmic derivatives PhoQ LCyt and PhoQ Cyt , and the putative dimerization phosphotransfer-harboring domain PhoQ DHp were able to down-regulate the expression of a PhoP-activated gene in inducing conditions (i.e. in Mg 2ϩ -limiting concentrations). In contrast, neither the isolated PhoQ catalytic domain (PhoQ CA ) nor EnvZ Cyt was able to exert a detectable effect, even though all the proteins tested were equally expressed. The potential involvement of the phospho-PhoP phosphatase activity in the inhibition was excluded because either overexpressed fulllength PhoQ in its kinase-dominant state or a mutant PhoQ that lacks phosphatase activity (PhoQ H277V ) showed pcgG down-regulation to the same extent as the truncated proteins. Accordingly, in vitro retention of PhoQ-derived proteins by PhoP-bound affinity columns showed that the liberated C-terminal truncated proteins (PhoQ LCyt and PhoQ Cyt , but not Pho-Q CA ) preserved the capacity, and thus the proper conformation, to effectively recognize and interact with PhoP with a binding strength that overcame subsequent column washes and allowed the elution of the complexes with imidazole. Therefore, these results pointed out that all PhoQ-derived constructs harboring the DHp region contained a relevant and specific interaction determinant.
Optical biosensor technology has proved to be an adequate biophysical tool for the analysis of both kinetic and equilibrium parameters to characterize a wide variety of biological meaningful interactions (49,50). The analysis of the biosensor assays rendered a K D ϭ 1.19 ϫ 10 Ϫ7 M for PhoP-PhoQ Cyt and a K D ϭ 7.94 ϫ 10 Ϫ7 M for PhoP-PhoQ DHp , whereas the K D for the PhoP-PhoQ CA interaction was 2.9 or 2.1 orders of magnitude higher, respectively. These results clearly demonstrated that the DHp region is essential for the interaction and that the contribution of the CA domain, if any, is low. The lack of a detectable signal when using EnvZ Cyt showed that this biophysical approach is highly sensitive to the specificity of the measured interaction. The PhoQ-PhoP calculated dissociation equilibrium constant is in agreement to those obtained for the interaction of other two-component systems partners using biosensor technology, fluorescence anisotropy and polarization, or a kinetic simulation model, estimated as 4.  (55). Here it is worth pointing out that differences in the estimated parameters could reflect structural differences among systems as well as divergences in the methodological approaches used.
Lately, efforts have been posed to elucidate how interaction is affected by the phosphorylation state of the components. Li et al. (56) working with the CheY/CheA system, and Mattison and Kenney (51) analyzing the OmpR/EnvZ interaction, found that upon phosphorylation, the response regulators reduce their affinity for the cognate kinases by 5-to at least 10-fold, respectively. In contrast, Yoshida et al. (52) showed that the EnvZ-OmpR interaction can be outcompeted with the same strength either by phosphorylated or by unphosphorylated OmpR. To explore this controversial matter, we assayed PhoQ Cyt incubated in buffer TKM with 1 mM ATP, or with 1 mM ADP, previous to the addition to the PhoP-chip. Both assays yielded essentially the same K D as the control PhoP-PhoQ Cyt (see Table  I). These results suggest that the phosphorylation state of the partners may not modify the strength of the interaction. However, additional experiments are being conducted to further analyze this observation because, even when the same conditions as those employed for phosphotransfer experiments were used, the phosphorylation status of neither PhoP nor PhoQ Cyt could be evaluated throughout the binding assay.
The PhoQ Cyt -PhoQ Cyt interaction detected either by the biosensor assay or by electrophoretic analysis, the measured Pho-Q Cyt -PhoQ DHp binding, together with the observed PhoQ Cyt3 PhoQ DHp transphosphorylation reaction, provides evidence that the PhoQ sensor dimerizes, consistent with the model that applies for other two-component systems (comprehensively reviewed by Dutta et al. (Ref. 10)). EnvZ Cyt did not interact with immobilized PhoQ Cyt , therefore showing that the bimolecular interaction between sensor molecules is also ruled by recognition specificity. Although no results were obtained with the PhoQ Cyt -chip when using PhoP as ligate, an effective dissociation of PhoQ Cyt -PhoQ Cyt or PhoQ Cyt -PhoQ DHp complexes was obtained by the subsequent addition of PhoP. These results strongly suggest that the immobilization of PhoQ Cyt to the cell matrix occurred preferentially in an orientation that rendered the PhoP-binding site sterically unavailable. Presumably, once PhoQ Cyt dimerization had taken place, added PhoP acted as a sink of free PhoQ Cyt or PhoQ DHp and the formation of these complexes displaced the primary interaction equilibrium.
In light of the interaction results examined here, it is tempting to rationalize that the in vivo interference effect observed was caused by simultaneous formation of complexes between the overexpressed proteins and the endogenous PhoQ, and sequestration of PhoP. This also suggests that interpretations of in vivo assays that use protein overexpression should be taken cautiously because they could be biased by interaction effects.
Additionally, activity assays provided evidence that relevant functional, and thus structural, features of the isolated PhoQ subdomains analyzed in this work are preserved. We also demonstrated that these subdomains are able to reconstitute distinctive PhoQ activities including bimolecular transphosphorylation as previously described for analogous systems (44,(57)(58)(59).
In conclusion, the results showed in this work reveal that DHp is the PhoQ domain that concentrates structural and functional determinants that are key to homodimerization, to specific PhoP recognition and to signal-induced phosphotransfer activities.