Truncation of Amino Acids 12–128 Causes Deregulation of the Phosphatase Activity of the Sensor Kinase KdpD of Escherichia coli *

The kdpFABC operon, which encodes the structural genes for the high affinity K+ transport complex KdpFABC, is regulated by the sensor kinase KdpD and the response regulator KdpE. KdpD is a bifunctional enzyme catalyzing the autophosphorylation by ATP and the dephosphorylation of the corresponding response regulator KdpE. Here, we demonstrate that the phosphatase activity of KdpD is dependent on ATP, whereas GTP, ITP, CTP, ADP, and GDP have no effect. The phosphatase activity requires only ATP binding, because nonhydrolyzable analogs (adenosine-5′-[γ-thio]triphosphate and adenosine-5′-[β,γ-imido]triphosphate) work as well. However, KdpD proteins missing amino acids 12–128 are characterized by a phosphatase activity that is independent of ATP. These proteins are still able to respond to K+ starvation, but an increase in osmolarity is no longer sensed. Comparison of different KdpD sequences reveals a conserved motif in this amino acid region that is very similar to a classical ATP-binding site (Walker A motif). Replacement of the conserved Gly37, Lys38, and Thr39residues in the consensus ATP-binding sequence results in a KdpD protein that causes a kdpFABC expression pattern comparable with that seen with KdpD proteins missing amino acids 12–128. However,in vitro phosphatase activity is comparable with that of wild-type KdpD. These results suggest that amino acids 12–128 of KdpD are important for its activity and that an additional ATP-binding site in the N-terminal region seems to be involved in modulation of the phosphatase activity.

KdpD and KdpE, two proteins that regulate the expression of the kdpFABC operon in Escherichia coli (1), are members of the large family of sensor kinase/response regulator systems (see Refs. 2 and 3 for review). The corresponding genes are organized in the kdpDE operon, which is adjacent to the kdpFABC operon that encodes the structural genes of the high affinity K ϩ transport complex KdpFABC (4,5). K ϩ is an important osmotic solute for the maintenance of turgor in bacterial cells (6), and the Kdp system can be characterized as an optional system to scavenge K ϩ from the environment. The stimulus that KdpD senses is believed to be a decrease in turgor pressure or some effect thereof (7).
Expression of kdpFABC is induced under K ϩ -limiting growth conditions (below 2 mM). In mutants lacking all other K ϩ -translocating transport systems (TrkG, TrkH, and Kup), kdpFABC is expressed in media containing 50 mM K ϩ or less. (7,8). There is no correlation of kdpFABC expression with internal K ϩ concentration when this parameter is altered by changing medium osmolarity. Therefore, neither the external nor the internal concentration of K ϩ per se seems to be sensed but what might be called the "need" for K ϩ to maintain turgor. Control by turgor is supported by the finding that a sudden increase in medium osmolarity, which reduces turgor, is able to turn on expression of the kdpFABC operon transiently (7,8). This model has been challenged by more recent findings (9,10) demonstrating that there is a difference in expression of kdp-FABC when the osmolarity of the medium is increased by a sugar or a salt. Analysis of mutant forms of KdpD that result in constitutive expression of kdpFABC independent of the K ϩ concentration of the medium but retain the ability to respond to changes in medium osmolarity led to the suggestion that KdpD senses two stimuli, decrease in turgor and K ϩ concentration (11).
The kdpD gene has been cloned, sequenced, and overexpressed (12). The gene product KdpD is an integral protein of the cytoplasmic membrane (13). Recently, KdpD was purified and reconstituted in an active form in proteoliposomes (14). This purified protein has autokinase activity. The phosphoryl group is subsequently transferred to the response regulator KdpE. Furthermore, KdpD catalyzes the dephosphorylation of purified KdpEϳP (14).
According to hydropathy analysis of the primary amino acid sequence and the analysis of a series of KdpD-alkaline phosphatase (kdpD-phoA) and KdpD-␤-galactosidase (kdpD-lacZ) fusions as well as protease susceptibility experiments (15), the following secondary structure model was established (Fig. 1). KdpD consists of a large cytoplasmic N-terminal region, four putative transmembrane domains, and an extended cytoplasmic C-terminal region. Whereas the C-terminal domain shows high similarity to transmitter domains of other sensor kinases (16), the length of the N-terminal input domain is rather unusual.
KdpD missing the four transmembrane domains became inactive in vivo and in vitro. However, proteins missing different parts of the soluble N-terminal domain could be phosphorylated in vitro, the phosphoryl group was transferred to KdpE, and they also contained phosphatase activity. The influence of the N-terminal truncations in KdpD on the transcriptional regulation of kdpFABC was tested in a trk Ϫ background. It was shown that the truncation of amino acids 12-128 influences the level of kdpFABC expression, whereas the truncation of amino acids 128 -391 exhibits expression levels that were comparable with wild-type KdpD (17).
In this communication we describe for the first time the quantification of the rate of the phosphatase activity of KdpD.
Furthermore, KdpD proteins missing amino acids 12-128 are characterized by a deregulated phosphatase activity. A conserved motif that is similar to a classical ATP-binding site (Walker A motif) can be found in this region. The strong dependence of the phosphatase activity of KdpD on ATP and alteration of the proposed nucleotide binding site by site-directed mutagenesis provide first evidence that ATP-binding modulates KdpD activity.

EXPERIMENTAL PROCEDURES
Materials-[␥-32 P]ATP was purchased from Amersham Pharmacia Biotech. All nucleotides were purchased from Sigma. NAP 10 columns to remove ATP were obtained from Amersham Pharmacia Biotech. All other materials were reagent grade and obtained from commercial sources.
Substitution of Gly 37 , Lys 38 , and Thr 39 was achieved by PCR 1 mutagenesis. The oligonucleotide primer was designed to change codons of Gly 37 , Lys 38 , and Thr 39 to codons of Ala 37 , Ala 38 , and Cys 39 . The PCR product was purified in agarose gels and digested with appropriate restriction enzymes. The DNA fragment was isolated from agarose gels and ligated to similarly treated pPV5-1 resulting in pPV5-1, G37A,K38A,T39C. Subsequently, this kdpD construct was cloned into pBAD18 (22) using XmaI and HindIII restriction sites, resulting in plasmid pBD-G37A,K38A,T39C.
DNA Sequencing-Mutations were verified by sequencing the length of the PCR-generated segment through the ligation junctions in doublestranded plasmid DNA, using the dideoxynucleotide termination method (23) and synthetic sequencing primers after alkaline denaturation of the DNA (24).
Phosphorylation and Dephosphorylation Assays-Inverted membrane vesicles containing KdpD (2 mg protein/ml) were incubated at room temperature in phosphorylation buffer, containing 50 mM Tris/ HCl, pH 7.5, 10% glycerol, 0.5 M NaCl, 10 mM MgCl 2 , and 2 mM dithiothreitol. Phosphorylation was initiated by addition of 20 M [␥-32 P]ATP (2.38 Ci/mmol). At different times, aliquots were removed and mixed with an equal volume of double concentrated SDS sample buffer (25). After incubation for 4.5 min, an equimolar amount of KdpE was added to the KdpD-containing fractions, and the incubation was continued. Further aliquots were removed at different times and mixed with SDS sample buffer as described above.
To test dephosphorylation, purified KdpE prepared as described (17) was phosphorylated in the following manner. Wild-type KdpD (4 mg/ml) was incubated in phosphorylation buffer, except that MgCl 2 was replaced with 5 mM CaCl 2, and phosphorylation was initiated with 20 M [␥-32 P]ATP (2.38 Ci/mmol). After 5 min, purified KdpE (0.1 mg/ml) was added, and the incubation was continued for 1 min. KdpD-containing membrane vesicles were removed by centrifugation. ATP was removed by gel filtration through Sephadex G25 (preincubated in 50 mM Tris/ HCl, pH 7.5, 10% glycerol, 2 mM dithiothreitol, 0.5 M NaCl). Purified KdpEϳP was used immediately. Dephosphorylation was initiated by addition of 20 mM MgCl 2 , 20 M ATP (or other nucleotides as specified), and inverted membrane vesicles containing KdpD (1 mg/ml). At different times aliquots were removed, and the reaction was stopped by addition of SDS sample buffer.
All samples were immediately subjected to SDS-polyacrylamide gel electrophoresis (PAGE) (25). Shortly before stopping SDS-PAGE an [␥-32 P]ATP standard was loaded on the gels. Gels were dried, and phosphorylation of the proteins was detected by exposure of the gels to a phosphorscreen. Phosphorylated proteins were quantified by image analysis using the PhosphorImager system of Molecular Dynamics.
Probing Signal Transduction in Vivo-In vivo signal transduction was probed with E. coli HAK006 transformed with the plasmids described. Cells were grown in TY medium (1% tryptone, 0.5% yeast extract) (11) or minimal medium (26) supplemented with NaCl and KCl as indicated. Cells were grown to midlogarithmic growth phase and harvested by centrifugation. ␤-Galactosidase activity was determined as described (27) and is given in Miller units.
Analytical Procedures-Protein was assayed by the method of Peterson (28) with bovine serum albumin as standard. Proteins were separated by SDS-PAGE (25) using 9 or 12% acrylamide gels. Immunodetection of KdpD proteins with polyclonal antibodies against KdpD was performed according to (13).

Kinase and Phosphotransfer Activities and the Influence of N-terminal Truncations in
KdpD-Inverted membrane vesicles containing equal amounts of either wild-type or truncated KdpD, as judged by Western blot analysis (data not shown), were tested for kinase activity. The time courses of autophosphorylation of different truncated KdpD proteins in comparison with wild-type KdpD is shown in Fig. 2. Autophosphorylation was tested to be linear within 0.5 min. Furthermore, transfer of the phosphoryl group to KdpE was determined. The transfer of the phosphoryl group to KdpE was very fast (within 15 s the phosphotransfer was complete). Phosphotransfer was detectable for all truncated KdpD proteins (data not shown).
Phosphatase Activity of KdpD and the Influence of N-terminal Truncations-In addition to the kinase activity KdpD catalyzes also the dephosphorylation of phosphorylated KdpE (14). KdpEϳP itself is very stable; within 2 h no major loss of the phosphoryl group was detected. Dephosphorylation of KdpEϳP was initiated by the addition of inverted membrane vesicles containing KdpD, 20 mM MgCl 2 , and 20 M ATP. Within about 5 min half of the amount of KdpEϳP was dephosphorylated (Fig. 3A). The rate of dephosphorylation of wildtype KdpD was comparable with that of the truncated forms. For wild-type KdpD only ATP was effective to stimulate the phosphatase activity; GTP, ITP, CTP, ADP, and GDP were without effect. Furthermore, nonhydrolyzable ATP analogs, ATP-␥-S and AMP-PNP, were as effective as ATP (data not shown). In the absence of ATP a very slow dephosphorylation was observed when wild-type KdpD and KdpD (⌬128 -391) 1 The  were tested (Fig. 3B). Interestingly, KdpD (⌬12-228) and KdpD (⌬12-395) are characterized by a phosphatase activity that is independent of the presence of ATP.
Influence of N-terminal Truncations in KdpD on the Regulation of kdpFABC Expression-It has been found previously that in wild-type strains kdpFABC is expressed when medium concentration of K ϩ is below 2 mM. Furthermore, an increase in medium osmolarity at constant K ϩ concentration, a maneuver that reduces turgor, caused expression of kdpFABC (7). Signal transduction mediated by truncated forms of KdpD was previously tested in E. coli deleted for other K ϩ uptake systems (TrkG and TrkH) leaving Kdp as the main K ϩ transport system. Because of the importance of the Kdp system as the major K ϩ uptake system, kdpFABC expression was never completely blocked (17). Therefore, we tested the transcriptional induction of kdpFABC with E. coli HAK006 (11) that synthesizes the constitutive K ϩ uptake systems, TrkH, TrkG, and Kup. This strain lacks the functional kdpFABC operon as well as the kdpD gene on the chromosome but contains the intact kdpE gene under the control of its own promoter. In addition, this strain harbors a kdpFABC promoter-lacZ fusion gene on the chromosome. Because the amount of regulatory proteins is very critical in signal transduction (high levels of KdpD prevent complementation of a kdpD null strain), 2 E. coli HAK006 was transformed with plasmids pBD and its derivatives. In plasmid pBD kdpD is under the control of the arabinose promoter (22). When cells were grown in the absence of an inducer (arabinose) and in the presence of the repressor glucose, the amount of KdpD produced was sufficient to complement a kdpD null strain. The truncated proteins were tested for their response to an increase in osmolarity and K ϩ limitation in comparison with wild-type KdpD (Table I). In response to an increase of osmolarity ␤-galactosidase activities of cells producing wild-type KdpD or KdpD (⌬ 128 -391) increased, whereas only basal levels of ␤-galactosidase activity were detectable in cells producing KdpD (⌬12-228) or KdpD (⌬12-395). Furthermore, K ϩ limitation resulted in high ␤-galactosidase activities when cells produced wild-type KdpD or KdpD (⌬128 -391). In the case of KdpD (⌬12-228) and KdpD (⌬12-395), ␤-galactosidase was detectable when cells were cultivated at very low K ϩ concentrations. However, ␤-galactosidase activity was much lower in comparison with that of wild-type KdpD.
Characterization of KdpD-G37A,K38A,T39C-Sequence comparison of the first 200 amino acids of so far six known KdpD sequences of different microorganisms revealed that a conserved motif exists that is similar to a classical ATP-binding site (Walker A motif) (29) (Fig. 4). In addition, we found that KdpD proteins missing these amino acids (KdpD(⌬12-395) and KdpD (⌬12-228)) are characterized by a phosphatase activity that is independent of ATP and cause deregulation of kdpFABC expression. To test whether KdpD contains a second ATPbinding site, site-directed mutagenesis was used to inactivate this site. Therefore, codons for Gly 37 , Lys 38 , and Thr 39 were replaced by codons for Ala 37 , Ala 38 , and Cys 39 in kdpD. The resulting KdpD protein was tested for kinase and phosphatase activity. The rate of autophosphorylation was comparable with that of wild-type KdpD; after addition of KdpE the phosphoryl group was rapidly transferred (data not shown). KdpD-2 K. Jung, unpublished information. G37A,K38A,T39C catalyzed the dephosphorylation of KdpEϳP. In the absence of ATP the rate of dephosphorylation was very slow, and it could be stimulated in the presence of ATP. Rates were comparable with that of wild-type KdpD or KdpD(⌬128 -391) (data not shown). Furthermore, transcriptional induction of kdpFABC was tested using E. coli strain HAK006 transformed with plasmid pBD-G37A,K38A,T39C. Inactivation of the proposed ATP-binding site impaired kdpFABC expression in response to an increase of the osmolarity of the medium (Table I). As shown before with truncated proteins missing amino acids 12-395 or 12-228, respectively, KdpD-G37A,K38A,T39C showed a modest ability to induce reporter gene expression in response to K ϩ starvation (Table I). DISCUSSION Dephosphorylation of the response regulator is well studied in various sensor kinase/response regulator systems and characterized by diverse mechanisms (16). The sensor kinase EnvZ by itself is able to enhance the rate of dephosphorylation of phospho-OmpR (31)(32)(33). In the case of NRII an auxiliary protein PII is involved (34 -36). In the chemotaxis system, dephosphorylation of CheY is mediated by a separate protein, CheZ (37), and in the process of sporulation of Bacillus subtilis different phosphatases have been identified (38).
Using purified and reconstituted KdpD we have recently shown that dephosphorylation of KdpEϳP is only dependent on the sensor kinase (14). In this communication we demonstrate for the first time that the rate of dephosphorylation can be significantly increased in the presence of ATP or nonhydrolyzable ATP analogs. In contrast, other nucleotides do not have an effect at all. The dephosphorylation activity of EnvZ is also stimulated by ATP, but in contrast to KdpD, also by ADP (31)(32)(33).
In comparison with other sensor kinases, KdpD contains an extended hydrophilic N-terminal domain of 400 amino acids exposed to the cytoplasm. This N-terminal domain is characterized by a high degree of homology between the so far known KdpD sequences from different organisms (Fig. 4). It is tempting to speculate that this domain, in addition to the four transmembrane helices, is involved in stimulus perception.
We found that the stimulation of phosphatase activity of truncated forms of KdpD missing amino acid 12-128 is no longer dependent on ATP. Furthermore, we demonstrated that amino acids 12-128 of the N-terminal domain of KdpD are important for the regulation of kdpFABC expression. KdpD(⌬12-395) and KdpD(⌬12-228) are not able to respond to an increase in osmolarity raised by the addition of NaCl. When cells synthesizing these truncated proteins were grown under K ϩ -limiting conditions, the kdpFABC operon was expressed, but the level of expression was much lower in comparison with that of wild-type KdpD. In contrast, the truncation of amino acids 128 -391 results in a KdpD protein that gave rise to the same expression level of the kdpFABC operon as wild-type KdpD.
In addition to the strong dependence of the phosphatase activity on the presence of ATP we found that amino acids 31-39 comprise a motif that is conserved among the KdpD sequences (Fig. 4) and that is very similar to a classical ATP binding site (Walker A motif) of many ATP-requiring enzymes (29). As indicated in Fig. 4, a Walker B motif can also be found. The most frequently published form of the Walker A motif is GXXXXGKT (39,40). The motif found in KdpD has one more X between the first and second Gs. There is one other ATPbinding site found in the dethiobiotin synthetase that has the same insertion. Structural analysis of this protein reveals that the dehydral angles of the conserved residues of the motif are identical to those of other proteins containing the classical motif (41). Therefore, we proposed that KdpD contains a second, regulatory ATP-binding site. To test this hypothesis, we replaced together the conserved residues Gly 37 , Lys 38 , and Thr 39 by site-directed mutagenesis. In this case kdpFABC ex-  Cells were grown to midlogarithmic growth phase in minimal medium containing different amounts of K ϩ or in TY medium in the presence of different NaCl concentrations (the basal concentration of K ϩ in this medium was determined to be 6 mM by flame photometry). ␤-Galactosidase activities were determined as described under "Experimental Procedures" and is given in Miller units (27). The data presented represent average values obtained in at least three independent experiments. pression is reduced to the same extent as it is seen with KdpD proteins missing this putative ATP-binding site. This results are in favor of the existence of a regulatory ATP-binding site in the N-terminal domain of KdpD. It is conceivable that the low level of kdpFABC expression is because of a decrease of the amount of KdpEϳP. Because in vitro no differences were seen in the kinase activities of KdpD proteins missing this site, it is suggested that ATP-binding regulates the phosphatase activity of KdpD. Indeed, we observed a deregulation of the phosphatase activity of KdpD proteins missing amino acids 12-128. However, the deregulation of the phosphatase activity in vitro was not observed with KdpD-G37A,K38A,T39C. Therefore, an interaction of the Nand the C-terminal domain may also contribute to the regulation of the phosphatase activity. An interaction of the input and transmitter domain depending on the stimulus might influence the switch between kinase and phosphatase activity in such a way that through conformational changes the accessibility of the putative regulatory ATP-binding site may vary and thereby the phosphatase activity might be tightly regulated. Although it is known that upon an osmotic upshift intracellular ATP concentration varies (42), it is too premature to draw conclusions about the interplay between ATP concentration and activity of KdpD.
There are to our knowledge only two other sensor kinases known that contain an additional nucleotide binding site (Walker A motif). In ChvG from Agrobacterium tumefaciens this putative site is located in the transmitter domain (43). The unorthodox sensor kinase BvgS from Bordetella pertussis contains such a motif in its linker region. Individual replacements of amino acid residues of this motif caused the inactivation of BvgS indicating the functional importance of this site (30).
In summary, we could show here that the phosphatase activity of KdpD can be increased in the presence of ATP or ATP analogs. Furthermore, KdpD proteins missing amino acids 12-128 are characterized by a deregulated phosphatase activity in vitro. Finally, KdpD proteins missing a conserved motif in the N-terminal domain, which is similar to a classical ATP-binding site (Walker A), either by truncation or mutagenesis diminish kdpFABC expression in response to K ϩ limitation or osmotic upshock.