Histidine kinase EnvZ, a transmembrane osmotic sensor for Escherichia coli, is a bifunctional enzyme having OmpR (its cognate response regulator) kinase and phosphorylated OmpR (OmpR-P) phosphatase activities. Its cytoplasmic domain consists of domain A responsible for dimerization of EnvZ, histidine phosphotransfer and phosphatase activities, and domain B responsible for ATP binding. Here, we have constructed a number of substitution mutations at the G2 box, one of the conserved motifs in domain B, and demonstrated that they influence the phosphatase activity of EnvZ over a wide range. The effects of ADP, a cofactor for the phosphatase activity, were found to be substantially different depending upon the mutations. The effects of these mutations were also examined in vivo using a chimeric Tar–EnvZ construct (Taz1-1), and the results agreed with the in vitro data for the phosphatase and kinase activities for all mutations. Using Taz1-1 carrying the T402A mutation, three independent intragenic suppressor mutations (T235M, S269L and E276K) were isolated, and all were found in domain A. Together, the present results demonstrate for the first time that domain A and domain B are functionally co-ordinated and topologi-cally arranged in a specific manner. The G2 box may modulate the interaction between these two domains in response to extracellular osmolarity.
Bacteria rely primarily on the histidyl-aspartyl phosphorelay (HAP) signal transduction system to respond to environmental stresses (Inouye, 1996). Generally, a sensor histidine kinase and its cognate response regulator protein (also known as an aspartate receiver) constitute the two essential components for the signal transduction (Stock et al., 1990; Parkinson, 1993; Pratt and Silhavy, 1995).
EnvZ, the osmotic sensor for Escherichia coli, belongs to a large histidine kinase superfamily. It is a transmembrane protein consisting of 450 residues and forms a homodimer in the cytoplasmic membrane (Forst et al., 1987; Egger and Inouye, 1997; Park and Inouye, 1997). Using ATP, EnvZ trans-autophosphorylates its conserved His-243 residue in a dimer (Yang and Inouye, 1991; Qin et al., 2000), which is subsequently transferred to the conserved Asp-55 residue on its cognate response regulator, OmpR. EnvZ can also function as a phosphatase to dephosphorylate phosphorylated OmpR (OmpR-P) (Aiba et al., 1989; Igo et al., 1989). ATP, ADP and non-hydrolysable ATP analogues such as AMPPNP function as cofactors for the phosphatase activity to stimulate this reaction further. OmpR-P serves as a transcription regulator for ompF and ompC genes encoding major outer membrane porin proteins (Nikaido and Vaara, 1985). The kinase–phosphatase ratio of EnvZ is proposed to determine the final levels of OmpR-P inside the cell, which subsequently regulate the reciprocal expression of ompF and ompC according to environmental osmolarity changes (Yang and Inouye, 1993; Forst and Roberts, 1994; Pratt and Silhavy, 1995; Egger et al., 1997).
The cytoplasmic domain of EnvZ (EnvZc) consisting of 271 residues (residues 180–450) possesses both kinase and phosphatase activities similar to the intact EnvZ (Park and Inouye, 1997). It contains all the highly conserved regions for histidine kinases: the H, N, F, G1, G2 and G3 boxes (Parkinson and Kofoid, 1992; Dutta and Inouye, 2000). Two distinct subdomains have been identified in EnvZc, domain A (residues 223–289) and domain B (residues 290–450) (Park et al., 1998). It has been demonstrated that domain A, containing the autophosphorylation site His-243, forms a dimer and can be phosphorylated when domain B is added together with ATP, and also transfers the phosphoryl group to the conserved Asp-55 on OmpR (Park et al., 1998). Domain A alone can dephosphorylate OmpR-P in the presence of Mg2+, whereas domain B, the subdomain containing all other conserved motifs except the H box, does not have phosphatase activity by itself and can stimulate the phosphatase activity of domain A slightly only in the presence of ADP/ATP/AMPPNP (Zhu et al., 2000). Comparing the phosphatase activity of domain A with that of EnvZc, it has been proposed that the phosphatase of EnvZ is regulated by the spatial arrangement between domain B and the covalently linked domain A (Zhu et al., 2000). Nuclear magnetic resonance (NMR) structures of both subdomains have been solved (Tanaka et al., 1998; Tomomori et al., 1999). However, no structural information has yet been obtained as to how domain A and domain B are topologically arranged in an EnvZ dimer. Note that one novel ATP-binding pocket has been identified only on the NMR structure of domain B.
The conserved motifs other than the H box have been extensively characterized because of their crucial roles in ATP binding in histidine kinases (Yang and Inouye, 1993; Dutta and Inouye, 1996; Ellefson et al., 1997; Tanaka et al., 1998). Studies on the substitution mutations at the conserved Asn-347 and multiple mutations in the G1 box of EnvZ indicated that the N and G1 boxes are not involved in EnvZ phosphatase function in the absence of cofactors. On the other hand, the G2 box, which consists of the highly conserved GXGXG motif and is also known to be critical to EnvZ ATP binding, has been shown to play an important role in the phosphatase reaction (Yang and Inouye, 1993; Hsing et al., 1998). Comparison of phosphatase activities between a G1 box mutant and a G1G2 double mutant indicated that the G2 box may be involved in both kinase and phosphatase activities of EnvZ (Yang and Inouye, 1993). The T402K mutation, which is located between the first and second conserved glycine residues in the G2 box, was shown to exhibit an OmpCc OmpF− phenotype (Hsing et al., 1998). According to the NMR structure of domain B, the G2 box is on the poorly structured central loop, which extends away from the rest of the α/β sandwich fold of the molecule (Tanaka et al., 1998). A similar extended central loop structure was also shown in the X-ray structure of the CheA kinase domain, a histidine kinase homologue of EnvZ (Hsing et al., 1998; Bilwes et al., 2001). Interestingly, in CheA, a shallow gap located below the G2 box is presumed to be the binding site of the CheA P1 domain (the histidine-containing phosphotransfer domain). It was suggested that the G2 box might be able to modulate the CheA enzymatic function through the interaction with the P1 domain,.
In the present study, we demonstrated that amino acid residues in and around the G2 box are important in both EnvZ kinase and phosphatase function. Ser-400, Thr-402 and Gly-403 residues may play distinct but related roles in the phosphatase reaction. Gly-403 and Gly-405, the second and third conserved glycine residues in the G2 box, are critical to the EnvZ ATP-binding ability and the autokinase function. Altered kinase/phosphatase activ-ities resulting from different mutations led to different OmpR-P production both in vitro and in vivo. By intragenic suppressor screening, we identified three independent second-site mutations for the T402A mutation in a hybrid receptor, Taz1-1, which could suppress the Taz1-1 T402A OmpC− phenotype to restore wild-type ligand-dependent expression. All the suppressor mutations were located in domain A and exhibited ligand-independent constitutive ompC–lacZ expression phenotypes without the T402A mutation, demonstrating the functional interaction be-tween domain B and domain A. We propose that the G2 box functions as a switch to regulate EnvZ bifunctional en-zymatic activities in response to extracellular osmolarity.
Effect of G2 box mutations on the phosphatase activity of EnvZ
Several lines of evidence suggest that the G2 box plays an important role in EnvZ phosphatase activity (Yang and Inouye, 1993; Hsing et al., 1998). To clarify this notion, we constructed a series of EnvZc G2 box mutants: EnvZc G401A, G403A, G405A, T402X (X = A, S, V, K, R, D and E) and S400X (X = A, T, V, K, R, D and E). Using these proteins, we carried out the phosphatase reaction either in the absence or in the presence of ADP, a cofactor required for full phosphatase activity.
In the absence of ADP, Gly to Ala mutations on the first and third conserved glycine residues of the G2 box (G401A and G405A) showed a similar phosphatase activity to that of the wild-type EnvZc. On the other hand, the Gly to Ala mutation on the second conserved glycine residue (G403A) resulted in reduced phosphatase activity (Table 1), suggesting its possible role in the EnvZ phosphatase reaction. Substitution mutations at Ser-400 (T, A, V, K, R, D and E) showed lower phosphatase activities in all cases, indicating its important role in the phosphatase reaction (Table 1). In contrast, similar mutations at Thr-402 (S, A, V, K, R, D and E) had variable effects on the EnvZ phosphatase reaction (Table 1).
Table 1. Phosphatase rate constant k (min −1 ) values of G2 box mutants.
OmpR-P (1.67 μM) was incubated with 0.84 μM EnvZc containing various G2 box mutations in Mg2+-containing buffer at room temperature.
Aliquots were taken out at several time points and analysed by SDS–PAGE gel to quantify the amount of OmpR-P remaining. Accordingly, the half-life of OmpR-P (t1/2) was calculated for each protein to represent its phosphatase activity. The experiments were carried out at least three times to obtain the half-lives, and the results were reproducible. The value of rate constant k is determined by the following formula: k = ln2/t1/2− ln2/t1/2auto. t1/2auto is the half-life of OmpR-P alone, representing its autophosphatase activity. It equals 90 min (Zhu et al., 2000). The data represented in the table are the mean values from those independent experiments.
Interestingly, in the presence of ADP, the phosphatase activity of EnvZc G401A was dramatically stimulated, whereas ADP had no effect on EnvZc G403A and EnvZc G405A phosphatase reactions (Table 1). Both mutants retained similar activities to those in the absence of ADP, implying that these EnvZc mutants might lose ATP-binding ability. Note that ATP and ADP are assumed to bind on the same ATP-binding pocket, which is crucial for both EnvZ kinase and phosphatase activities and identified on the NMR structure of domain B of EnvZ.
ADP differentially stimulates the phosphatase activities of the EnvZ substitution mutants at Ser-400. However, all mutants had lower activities than that of the wild-type EnvZc (Table 1). Lower phosphatase activities of EnvZc S400X mutants in both the presence and the absence of ADP indicate that Ser-400 may be directly involved in the EnvZc phosphatase reaction.
In the presence of ADP, the effects of substitution mutations at Thr-402 on the phosphatase activity were much more prominent (A ≥ S, T > V > D, E > R > K). Substitution mutations with charged amino acids at this position dramatically decreased the phosphatase activity, whereas the Thr to Ser substitution kept a wild type-like reaction rate. Notably, ADP slightly inhibits the EnvZc T402K phosphatase activity in contrast to all the other substitution mutations. It is also interesting to point out that, in the presence of ADP, the EnvZc T402A mutant showed a distinctly higher phosphatase activity than that of wild-type EnvZc (Table 1).
Previously, we demonstrated that domain A by itself had Mg2+-dependent phosphatase activity (Zhu et al., 2000). The addition of domain B could slightly stimulate the domain A phosphatase activity only in the presence of ADP, whereas covalently linked domain B, as seen in EnvZc, can dramatically enhance the activity in both the presence and the absence of cofactors. Here, we also examined how the domain A phosphatase activity is affected by the addition of a separated domain B with mutation S400D, T402K or G403A in the G2 box. All these mutations did not display discernible effects on the phosphatase activity of domain A in both the presence and the absence of ADP (data not shown). Therefore, mutations on the G2 box affect EnvZ phosphatase activity only when domain B is covalently linked to domain A. This is consistent with our previous view that the spatial arrangement between domains A and B is essential for EnvZ phosphatase function (Zhu et al., 2000).
Either modulating the catalytic reaction or changing the binding affinity of OmpR/EnvZ could modify EnvZ phosphatase activity. Using a His10-OmpR/EnvZc binding assay (Qin et al., 2000), we examined the effect of mutations on the binding affinity of EnvZc for OmpR. As shown in Fig. 1, the amounts of EnvZc G401A, G403A and G405A specifically bound to His10-OmpR were almost identical to that of wild-type EnvZc. Similarly, other EnvZc mutants in the present study showed similar binding affinities for His10-OmpR to the wild-type EnvZc (data not shown).
ATP-binding ability of EnvZc G2 box mutants
It has been shown that the G2 box is critical for the nucleotide-binding ability of EnvZ (Yang and Inouye, 1993; Tanaka et al., 1998). In Table 1, we demonstrate that ADP has no effect on stimulating the phosphatase activities of EnvZc G403A and G405A, suggesting that these two residues are essential for ATP binding. We next carried out UV cross-linking experiments to examine directly the nucleotide-binding ability of EnvZc G2 box mutants.
As shown in Fig. 2, wild-type EnvZc could bind to [35S]-ATP-γS under the conditions used in the present work, whereas EnvZc N347D, a mutant known to lose ATP-binding ability (Dutta and Inouye, 1996; Tanaka et al., 1998), could not bind ATP. EnvZc G403A and G405A mutants had severely defective ATP-binding ability, whereas the G401A mutation had no effect. EnvZc S400X and T402X mutants retained the ability to bind ATP, although the amounts of bound ATP were different in various mutants. Therefore, Gly-403 and Gly-405 are the key residues in the G2 box responsible for EnvZ ATP binding.
OmpR kinase activity of EnvZc G2 box mutants
EnvZ is a bifunctional enzyme, and its kinase– phosphatase ratio could control the cellular OmpR-P concentration (Jin and Inouye, 1993; Yang et al., 1993). Mutations on the G2 box may not only affect EnvZ phosphatase activity, but also affect its OmpR kinase activity. In order to examine the overall effect of the G2 box mutations on EnvZ/OmpR phosphorylation events, different EnvZc G2 box mutants were first tested for autophosphorylation ability, and then OmpR was added to the reaction to examine the ability of OmpR phosphorylation.
With the wild-type EnvZc, the phosphoryl group of EnvZc rapidly disappeared upon addition of OmpR. Within 30 s, the radioactive band corresponding to EnvZ was reduced to <10% and almost completely disappeared 3 min after the addition of OmpR (Fig. 3A). Meanwhile, OmpR-P was not detectable throughout the reactions because of bifunctional activities of EnvZc. EnvZc G401A gave a similar result to the wild-type EnvZc (Fig. 3A). Under the same conditions, EnvZc-P or OmpR-P bands could not be detected with G403A and G405A mutants. By overexposure (18 h exposure instead of 2 h), we could observe very faint EnvZc-P bands at the 20 min auto-phosphorylation time point for both G403A and G405A mutants, whereas the intensity of the EnvZc G405A-P band was weaker than that of EnvZc G403A-P (data not shown).
When Ser-400 was substituted with other amino acids, OmpR-P became detectable (Fig. 3B and C), as expected from the fact that all the S400X substitution mutations tested had reduced phosphatase activities (Table 1). The patterns of OmpR-P accumulation agreed well with the reduced rates of the phosphatase activities of these EnvZc mutants as, in EnvZc S400T and S400A, OmpR-P could only be seen at the 30 s time point, whereas in all the other mutants, OmpR-P was detectable at all three time points (0.5, 1.5 and 3 min; Fig. 3B and C).
In the Thr-402 substitution mutations, those showing high phosphatase activities, such as T402S, T402A, and T402V (see Table 1), did not exhibit the accumulation of OmpR-P even at the 30 s time point except for T402V, in which a faint OmpR-P band could be seen at 30 s (Fig. 3D and F). With EnvZc T402K (Fig. 3D), T402D (Fig. 3E), and T402E (Fig. 3E), OmpR-P was detectable at all three time points. In the case of the T402R substitution (Fig. 3D), EnvZc was poorly autophosphorylated reflecting its poor ATP-binding ability (see Fig. 2), and OmpR-P accumulation was observed at all three time points.
ompC–lacZ expression of Taz1-1 carrying G2 box mutants
The above data showed that the kinase–phosphatase ratio of EnvZ might be altered in vitro. In order to examine the effects of the G2 box mutations in vivo, we used a hybrid receptor Taz1-1, in which the receptor domain of Tar (an aspartate chemoreceptor) is fused with the cytoplasmic signalling domain of EnvZ (Utsumi et al., 1989). It has been demonstrated that cells carrying Taz1-1 are able to respond to aspartate concentrations in the medium to mediate the expression of ompC–lacZ in E. coli RU1012 cells (Utsumi et al., 1989; Jin and Inouye, 1993).
Consistent with the in vitro data, cells with the G401A Taz1-1 exhibited aspartate-regulatable ompC expression as cells with the wild-type Taz1-1, whereas the G405A Taz1-1 could not mediate ompC expression even in the presence of aspartate (Fig. 4B). Interestingly, the G403A mutant that had impaired ATP binding (Fig. 2) showed an aspartate-regulatable ompC expression pattern. The possible reasons for this phenotype will be addressed in the Discussion.
As shown in Fig. 4C, substitution mutations at S400X (V, D, E, K and R), which lowered EnvZc phosphatase activity (Table 1) and exhibited an OmpR-P accumulation profile in the OmpR kinase assay (Fig. 3), resulted in OmpCc phenotypes. Mutations to Ala or Thr at Ser-400 showed partially regulatable phenotypes. The β-galactosidase activities of these two mutants in the absence of aspartate were higher than that of wild-type Taz1-1 in the presence of 5 mM aspartate. Adding 5 mM aspartate further increased the activities of these two mutants about 1.7-fold. Comparing the OmpR-P accumulation patterns of these two mutants with those of the wild-type Taz1-1 and the other Taz1-1 S400 mutants (Fig. 3B), one may speculate that, even in the absence of the ligand, the kinase–phosphatase ratio in the S400A and S400T mutants is already high enough to raise the cellular OmpR-P amount to stimulate ompC expression, whereas the addition of the ligand may further shift the kinase– phosphatase ratio, resulting in higher ompC expression levels.
Taz1-1 T402X mutants showed various phenotypes (Fig. 4D). Only Taz1-1 T402S exhibited a similar response to the medium aspartate concentration as the wild-type Taz1-1. Lys, Arg, Asp or Glu substitution resulted in OmpCc phenotypes. The phenotype of the Taz1-1 T402K mutant is consistent with a previous report (Hsing et al., 1998). Taz1-1 T402V could respond to aspartate concentration changes; however, it had a higher ompC expression background in a similar manner to the S400A and S400T mutants. Taz1-1 T402A cannot express ompC even in the presence of aspartate in the medium, probably because EnvZc T402A has a higher phosphatase activity than wild-type EnvZc.
Intragenic suppressors of the T402A mutation
In the intact EnvZ protein, domain B interacting with domain A might modulate EnvZ enzymatic functions, especially the phosphatase activity responding to environmental osmolarity changes (Zhu et al., 2000). The data shown above demonstrated that the G2 box plays an important role in EnvZ phosphatase activity. Amino acid residues around this motif are likely to participate in the phosphatase reaction either directly or indirectly, together with the effect of bound ADP, controlling the cellular OmpR-P concentration. In RU1012 cells, Taz1-1 T402A cannot induce β-galactosidase even in the presence of aspartate (see Fig. 4D). Taking advantage of this phenotype, we attempted to screen intragenic suppressors for this mutant to identify second-site mutations that can restore a wild type-like aspartate-inducible ompC–lacZ expression phenotype.
We first isolated Lac+ cells on lactose-MacConkey plates after mutagenesis. More than 600 Lac+ strains were then examined for their Lac+ inducibilities in response to aspartate (5 mM) added to the medium. Unlike most of the suppressors we obtained during the first screening step, which led to constitutive ompC expression in both the presence and the absence of aspartate, three different independent second-site mutations that we identified in the second screening step exhibited aspartate-inducible ompC expression, indicating the functional co-ordination between the second-site mutations and the original mutation. The effect of these second-site mutations was further confirmed by reintroducing each mutation through site-directed mutagenesis on the Taz1-1 T402A gene. The new double mutation constructs were transformed into RU1012 cells to examine β-galactosidase activities in response to different amounts of aspartate added to the medium. As shown in Fig. 5A, when 5 mM aspartate was added to the medium, all Taz1-1 T402A plus suppressor mutations T235M, S269L or E276K gave at least fourfold higher ompC expression than without aspartate. Taz1-1 mutants containing only the second-site mutations were also constructed, and in vivo phenotypes were examined. RU1012 cells harbouring any of the three constructs exhibited the OmpCc phenotype (Fig. 5B).
Interestingly, all three second-site mutations identified are positioned on domain A. According to the NMR structure of domain A, Ser-269 and Glu-276 are on helix II of domain A, whereas Thr-235 is located at the beginning of helix I (Fig. 6A) (Tomomori et al., 1999). Although different strategies may be adopted by these mutations to suppress the effect of the T402A mutation that is located in the G2 box of domain B (Fig. 6B) (Tanaka et al., 1998), the present results indicate that domain A and domain B interact with each other to exert the EnvZ function.
We also carried out complementation experiments by co-transforming Taz1-1 T402A with Taz1-1 harbouring either of the suppressor mutations. As shown in Fig. 5C, S269L and T235M suppressor mutations could complement with T402A to restore the aspartate-regulatable phenotype, whereas the E276K mutation could not. Therefore, the E276K suppressor mutation has to co-exist with T402A in the same polypeptide in order to resume regulatable ompC expression.
Previous NMR studies revealed that EnvZ domain B contains a novel ATP-binding motif, in which most conserved motifs (N, G1, F G2 and G3 boxes) in histidine kinases are structurally involved (Parkinson and Kofoid, 1992; Tanaka et al., 1998; Dutta and Inouye, 2000). Most mutations on domain B (such as G1 and N boxes) severely inhibited EnvZ kinase activity but barely affected the phosphatase activity (Jin and Inouye, 1993). In this report, we demonstrated that, in addition to its critical role in ATP binding and kinase reaction, the G2 box plays an important role in EnvZ phosphatase activity.
Among the three conserved glycine residues (GXGXG) in the G2 box, Gly to Ala mutation at the first (Gly-401) and third (Gly-405) glycine residues showed a similar phosphatase activity to that of wild-type EnvZc in the absence of ADP, suggesting that these two residues may not be directly involved in EnvZ phosphatase reaction. On the other hand, a similar mutation at the second one (Gly-403) resulted in lower phosphatase activity in both the absence and the presence of ADP, implying its important role in the phosphatase reaction. From the effect of ADP on phosphatase activities and a UV cross-linking experiment with these mutant EnvZc, it was found that Gly-403 and Gly-405 are critical to ATP binding, but Gly-401 is not. An OmpR kinase assay further demonstrated that both G403A and G405A were defective in kinase activity, whereas G401A kept a wild type-like activity. Therefore, although the conserved glycine residues at the G2 box are essential for EnvZ function (Parkinson and Kofoid, 1992; Yang and Inouye, 1993), each residue may function in a different manner. This is the first time that the roles of individual conserved glycine residues in the G2 box have been defined in the histidine kinase function.
Substitution mutations at Ser-400, which is located just upstream of the G2 box, also caused lower EnvZc phosphatase activity in both the absence and the presence of ADP. This suggests that not only the conserved Gly-403 but also amino acids in the vicinity of the G2 box may be involved in EnvZ phosphatase function. It should be noted that, when Arg-397, another residue near the G2 box, was mutated to Cys, the mutation resulted in an OmpCc phenotype in the Taz1-1/RU1012 system (unpublished data). Similar mutations at Thr-402 had various effects on the EnvZ phosphatase reaction especially in the presence of ADP, suggesting that possible conformational changes resulting from the cofactor binding may rearrange the way in which this residue participates in the EnvZ phosphatase reaction. Notably, mutations at the G2 box (S400D, T402K and G403A) in the covalently linked domain B significantly affected EnvZ phosphatase activity, whereas the same mutations in the separated domain B did not affect domain A phosphatase activity even in the presence of ADP. These results support our previous proposal that the spatial arrangement between domain A and domain B is crucial to EnvZ phosphatase function (Zhu et al., 2000).
EnvZ domain A but not domain B has been shown to be critical to EnvZ–OmpR interaction (Park et al., 1998; Tanaka et al., 1998). It is still possible that mutations on domain B may interfere directly or indirectly in the interaction between domain A and OmpR/OmpR-P to inhibit the phosphatase activity. Using a His10-OmpR/EnvZc binding assay, we found that all the EnvZc mutants in the present study show similar binding affinities for OmpR to the wild-type EnvZc. Therefore, the effects of substitution mutations at the Ser-400, Thr-402 and Gly-403 on phosphatase activity are probably caused by modulation of the EnvZ catalytic function by the mutations rather than modulation of the EnvZ–OmpR interaction.
Recently, the X-ray structure of the CheA kinase domain has been determined with different nucleotides (Bilwes et al., 2001), in which the third conserved glycine residue in the G2 box, Gly-506, was found to form a hydrogen bond to nucleotide triphosphates through the main chain nitrogen. However, in our UV cross-linking assays, both the second and the third conserved glycine residues (Gly-403 and Gly-405) in the G2 box of EnvZ were found to be essential to ATP binding. Most recently, the X-ray structure of the AMPPNP-bound kinase domain of PhoQ, another member of the histidine kinase superfamily, was also reported (Marina et al., 2001). In the PhoQ structure, Gln-442 rather than the conserved glycine residues (Gly-441, Gly-443 and Gly-445) in the G2 box was hydrogen bonded to the γ-phosphate in ATP. However, in the G2 box of EnvZc, none of the substitution mutations at Thr-402, which corresponds to Gln-442 in PhoQ, could abolish EnvZc ATP-binding ability. Nevertheless, as the G2 box is one of the most conserved motifs characterized for its role in nucleotide binding, in different intact histidine kinases, it may still interact with a bound nucleotide in a similar manner.
In vivo studies on the effects of EnvZc mutations using the hybrid Taz1-1 receptor are consistent with the in vitro data. Only Ser functionally substituted for Thr-402 in Taz1-1 to exhibit wild type-like regulatable ompC expression, indicating the importance of the hydroxyl group of Thr in EnvZ function. The Taz1-1 G403A mutant showed an unexpectedly regulatable ompC expression phenotype under the conditions used. Based on the fact that both kinase and phosphatase activities of EnvZc G403A and Taz1-1 G403A (unpublished results) have dramatically decreased compared with those of wild-type EnvZc or Taz1-1, we thought that Taz1-1 G403A might still keep an appropriate kinase–phosphatase ratio. A similar aspartate-regulatable expression pattern was observed using the pBAD33 vector ( Guzman et al., 1995 ) for Taz1-1 G403A expression in spite of the fact that the absolute β-galactosidase units were much lower (unpublished results). Note that the expression level of Taz1-1 G403A with the pBAD vector was about 100 times lower than that with the pINIII construct. This indicates that the cellular kinase contents are highly flexible to achieve aspartate-regulatable ompC–lacZ expression; therefore, the ratio of kinase–phosphatase seems to be the key factor for regulatable kinase function.
We identified three intragenic second-site mutations that could suppress the OmpC− phenotype caused by a mutation in domain B (T402A) and restore a wild type-like phenotype in vivo. This result demonstrates for the first time that two mutations in the catalytic domain of EnvZ co-ordinate with each other to fulfil normal EnvZ function. All these three second-site mutations were located at domain A (Fig. 5A) and, by themselves, exhibited OmpCc expression phenotypes. Thr-235 is positioned at the top of helix I of domain A (Fig. 5A). It could be involved in the interaction with the ADP-bound domain B (Fig. 5B), which faced the conserved His-243 residue of domain A (Fig. 5A). Meanwhile, Ser-269 is located at the bottom of a domain A four-helix bundle, the presumed area for EnvZ–OmpR interaction (Tomomori et al., 1999). Therefore, this residue may participate in modulating the binding between OmpR/OmpR-P and EnvZ. Domain B may take part indirectly in the modulation of OmpR binding through its interaction with domain A. Glu-276 is located in the middle of helix II of domain A, spatially close to the conserved His-243 residue. It has been proposed that this residue is a member of an acidic cluster in the vicinity of His-243 that is involved in the interactions with domain B and the N-terminal domain of OmpR (Tomomori et al., 1999). It is interesting to note that this residue, Glu-276, falls in the X region, which is known to be important for EnvZ phosphatase activity (Hsing et al., 1998). Taken together, no matter what strategies are adopted by these suppressor mutations to suppress the T402A mutation in domain B, the interaction between domain A and domain B has to be finely adjusted. Co-transforming Taz1-1 T402A with Taz1-1 harbouring suppressor mutation E276K could not restore aspartate-regulatable ompC expression, as seen in the Taz1-1 T402A/E276K double mutant, indicating that the E276K mutation and the T402A mutation have to co-exist in the same polypeptide to reset the proper structure to fulfil normal EnvZ function. Therefore, within an EnvZ dimer, domain B from one monomer not only interacts with helix I of domain A from the other monomer (Yang and Inouye, 1991; Qin et al., 2000) but may also interact with helix II of domain A from the same monomer.
With regard to the osmoregulation of signal transduction mediated by EnvZ/OmpR, we may speculate that, as the topological relationship between domain A and domain B is altered in an EnvZ dimer responding to extracellular signals, the local environment around the central loop including the G2 box in domain B (Fig. 5B) may change. Subsequently, the possible interactions between the residues in domain B and specific residues in domain A are rearranged to regulate the enzymatic function of EnvZ. The finely tuned EnvZ kinase– phosphatase ratio eventually determines the cellular concentration of OmpR-P, which then mediates proper a cellular response.
Strains and plasmids
Escherichia coli B strain BL21 (DE3) was used for the expression of all proteins. Strains RU1012 [MC4100 ara+, Φ(ompC-lacZ) 10–25 ΔenvZ::Km r ] ( Utsumi et al., 1989 ) and LE513 [MC4100 ara+, Φ(ompF′-lacZ) 16–23 [λp1(209)] ΔenvZ::Km r ] ( Zhu et al., 2000 ) were used for in vivo study.
pET11a-EnvZc S400X, T402X, G401A, G403A and G405A were constructed by site-directed mutagenesis using pET11a-EnvZc (Park et al., 1998) as template (X represents a number of different residues).
For the in vivo phosphatase assay, EnvZc S400X, T402X, G401A, G403A and G405A fragments were isolated from the respective pET11a vectors by NdeI–BamHI digestion. All fragments were then ligated into NdeI–BamHI-digested Taz1-1 (pYT0301, Ampr; Yang and Inouye, 1991) to replace the wild-type EnvZ fragment. These constructs were named Taz1-1 S400X, T402X, G401A, G403A and G405A respectively (note that the residue numbers are based on EnvZ not on Taz1-1). Unless indicated, all the Taz1-1 constructs in the present study contained an ampicillin-resistant gene.
All proteins were purified by 35% (NH4)2SO4 fractionation and subsequent Sephacryl S-100 H gel filtration column chromatography (Pharmacia) as described previously (Zhu et al., 2000).
In vitro phosphatase assay
OmpR-P was prepared and purified as described previously (Zhu et al., 2000). The phosphatase activity was determined as described previously (Zhu et al., 2000).
His10-OmpR/EnvZc binding assay
The His10-OmpR/EnvZc binding assay was carried out as described previously (Qin et al., 2000). The specific and non-specific binding of EnvZc protein to Ni-NTA resin was determined by adding or not His10-OmpR to the binding reaction mixture. The final eluates from these two reactions were subjected to SDS–PAGE. The gel was stained with Coomassie brilliant blue, and the amounts of proteins were estimated using a Bio-Rad Model GS-670 imaging densitometer.
UV cross-linking assay
Using various purified EnvZc proteins, a UV cross-linking assay was carried out as described previously (Tanaka et al., 1998) to examine the ATP-binding ability of different EnvZc mutants except that [35S]-ATP-γS (800 Ci mmol−1, 10 mCi ml−1; Perkin-Elmer Life Sciences) was used instead of [α-32P]-ATP.
OmpR kinase assay
EnvZc or EnvZc mutant proteins (2 μM) were incubated at room temperature in the reaction buffer containing 50 mM Tris-HCl (pH 8.0), 50 mM KCl, 5 mM MgCl2, 5 mM β-mercaptoethanol, 50 μM ATP and 0.5 μCi [γ-32P]-ATP (3000 Ci mmol−1, 10 mCi ml−1; Perkin-Elmer Life Sciences). A 10 μl aliquot was removed from the reaction mixture at 20 min, and the reaction was stopped by adding 2.5 μl of 5× SDS loading buffer. In the remaining reaction mixture, OmpR was added to a final concentration of 2 μM. Aliquots (10 μl) were removed at 0.5, 1.5 and 3.0 min, and reactions were stopped as described above. All reaction mixtures were then analysed by SDS–PAGE followed by autoradiography.
β-Galactosidase activities of RU1012 cells transformed with various Taz1-1 constructs were determined as described previously (Yang and Inouye, 1993).
Screening for intragenic suppressor
DNA from the hybrid plasmid Taz1-1 carrying the T402A mutation in the EnvZc domain was treated with in vitro mutagenesis buffer containing 100 mM potassium phosphate (pH 6.0), 1 mM EDTA and 500 mM hydroxylamine for 60 min at 65°C. Hydroxylamine was removed by a Micro Bio-Spin column with Bio-Gel P-6 in Tris buffer (Bio-Rad) according to the manufacturer's procedure. The mutagenized plasmid DNA was then transformed into RU1012 cells. Red colonies selected on lactose-MacConkey agar plates containing 5 mM aspartate and 50 μg ml−1 ampicillin were tested for the aspartate-inducible β-galactosidase expression phenotype. Plasmids were isolated from those cells that were able to induce β-galactosidase in the presence of 5 mM aspartate and retransformed into RU1012 cells. Three transformants were tested to confirm the β-galactosidase-inducible phenotype. Finally, the whole coding region of these plasmids was sequenced to identify the intragenic suppressor mutation. Plasmids that contained double mutations or only the second-site mutation were constructed by site-directed mutagenesis and then transformed into RU1012 cells for β-galactosidase assay. For co-transformation assay, the T402A mutation was introduced into pYY0401, a Taz1-1 construct with a Cmr marker (Yang and Inouye, 1991). This new Taz1-1 T402A construct and any of the three Taz1-1 constructs harbouring only the second-site mutation were sequentially transformed into RU1012 cells for β-galactosidase assay.
We thank Dr K. Inoue for stimulating discussions, and Dr L. Qin, Dr S. Phadtare, Mr S. Cai and Mr T. Yoshida for critical reading of the manuscript. This work was supported by grant GM19043 from the National Institutes of Health.