The N-terminal Input Domain of the Sensor Kinase KdpD of Escherichia coli Stabilizes the Interaction between the Cognate Response Regulator KdpE and the Corresponding DNA-binding Site*

The sensor kinase/response regulator system KdpD/KdpE of Escherichia coli regulates expression of the kdpFABC operon, which encodes the high affinity K+ transport system KdpFABC. The membrane-bound sensor kinase KdpD consists of an N-terminal input domain (comprising a large cytoplasmic domain and four transmembrane domains) and a cytoplasmic C-terminal transmitter domain. Here we show that the cytoplasmic N-terminal domain of KdpD (KdpD/1-395) alone supports semi-constitutive kdpFABC expression, which becomes dependent on the extracellular K+ concentration under K+-limiting growth conditions. However, it should be noted that the non-phosphorylatable derivative KdpD/H673Q or the absence of KdpD abolishes kdpFABC expression completely. KdpD/1-395 mediated kdpFABC expression requires the corresponding response regulator KdpE with an intact phosphorylation site. Experiments with an Escherichia coli mutant unable to synthesize acetyl phosphate as well as transposon mutagenesis suggest that KdpE is phosphorylated in vivo by low molecular weight phosphodonors in the absence of the full-length sensor kinase. Various biochemical approaches provide first evidence that kdpFABC expression mediated by KdpD/1-395 is due to a stabilizing effect of this domain on the binding of KdpE∼P to its corresponding DNA-binding site. Such a stabilizing effect of a sensor kinase domain on the DNA-protein interaction of the cognate response regulator has never been observed before for any other sensor kinase. It describes a new mechanism in bacterial two-component signal transduction.


EXPERIMENTAL PROCEDURES
Materials-[␥-32 P]ATP, 125 I-labeled protein A, [␣-32 P]UTP, 32 P i , E. coli RNA polymerase, RNA guard RNase inhibitor, and RNase-free nucleotides ATP, GTP, CTP, and UTP were purchased from Amersham Biosciences. Ni 2ϩ -NTA resin, Ni 2ϩ -NTA magnetic agarose beads, the 12-tube magnet, and the Nucleotide Removal Kit were purchased from Qiagen. Goat anti-(rabbit IgG)-alkaline phosphatase was purchased from Biomol. Goat anti-(mouse IgG) horseradish peroxidase, goat anti-(rabbit IgG) horseradish peroxidase, and SuperSignal West Femto Maximum Sensitivity Substrate were obtained from Pierce. All other reagents were reagent grade and obtained from commercial sources.
In plasmid pBD and pBD3, kdpD or its derivatives were cloned into pBAD18 or pBAD33 (22), respectively, and expression is in both cases under control of the arabinose promoter (5,23). In plasmid pBD/1-395 and pBD3/1-395, three stop codons were inserted at the corresponding site of the kdpD gene, so that only the first 395 amino acids of KdpD are produced (11). Plasmid pBD/1-395/G37A,K38A,T39C 3 encodes KdpD/ 1-395, in which the ATP-binding site is inactivated by amino acid replacements (11). Plasmid pBD/H673Q encodes the KdpD derivative with the inactivated phosphorylation site (KdpD/H673Q) (5). In plasmid pPV2, the kdpDE operon is under control of the lac promoter (1). Plasmids pPV2/1-395 and pPV2/1-395/D52N were obtained by diges-tion of plasmid pPV5-3/1-395 (11) with restriction enzymes NsiI and StuI, and the purified fragment was then ligated to similarly treated vector pPV2 or pPV2/D52N (see below), respectively. To obtain a KdpE derivative with an N-terminal 10His tag, the kdpE gene was amplified by synthetic oligonucleotide primers and cloned into vector pET16b (Novagen Inc., Madison, WI) using the NdeI site, resulting in plasmid pEE, where 10His-kdpE is under control of the T7 promoter. All PCRgenerated DNA fragments were verified by sequencing through the ligation junctions in double-stranded plasmid DNA. Plasmid pSM5 encodes the kdpFABC operon. 4 Plasmid pNK2883 harboring a short version of the transposon Tn10 was used for transposon mutagenesis (24).
Oligonucleotide-specific Site-directed Mutagenesis-Introduction of a codon change corresponding to the replacement of Asp-52 against Asn in KdpE was achieved by PCR using the overlap extension method (25). The PCR product was separated by agarose gel electrophoresis, purified from the gel, digested with restriction endonucleases ClaI and XmaI, and then ligated to similar treated vector pPV2 (1) resulting in plasmid pPV2/D52N. The mutation was verified by sequencing through the ligation junctions in double-stranded plasmid DNA.
Determination of kdpFABC Expression-In vivo signal transduction was probed with strains of E. coli HAK006, RH001, and RH003 transformed with the plasmids described. Cells were grown in TY medium (1% (w/v) tryptone, 0.5% (w/v) yeast extract) supplemented with NaCl and sucrose as indicated (26) or minimal media containing the indicated K ϩ concentrations (27). Cells were grown to mid-exponential growth phase and harvested by centrifugation, and ␤-galactosidase activity given in Miller units was determined as described (28). Alternatively, expression of kdpFABC was determined in E. coli strains lacking the Trk K ϩ transport system (E. coli TKV2208, E. coli TKV2209, and E. coli RH004) by quantitative Western blot analysis of the amount of produced KdpFABC complex. These strains were grown at 37°C in phosphate-buffered minimal medium (27) containing 115 mM K ϩ until the mid-exponential phase, diluted in prewarmed medium of lower K ϩ concentration, and harvested after 30 min. Then cells were resuspended in SDS sample buffer and subjected to SDS-PAGE (29). Quantification of KdpFABC was basically performed following the protocol developed for lactose permease (30). Briefly, proteins were electroblotted to a nitrocellulose membrane. Blots were then blocked with 3% (w/v) bovine serum albumin in buffer A (10 mM Tris/HCl, pH 7.5, 0.15 M NaCl) for 1 h. Anti-KdpB antibody was added at a final dilution of 1:5000, and incubation was continued for 1 h. After washing with buffer A, 125 Ilabeled protein A was added at a final dilution of 1:5000, and incubation was continued for 1 h. After washing thoroughly, the membrane was exposed to a Storage Phosphor Screen. Known amounts of purified KdpFABC complex were used to obtain a standard curve. The amount of KdpFABC complex was then quantified using the PhosphorImager SI system by comparison to the standard curve.
Cell Fractionation and Preparation of Inverted Membrane Vesicles-E. coli strain TKR2000 transformed with plasmid pPV5-3/1-395(His6) (11) or E. coli BL21(DE3)/pLysS transformed with plasmid pEE (this study) was grown aerobically at 37°C in KML complex medium (1% (w/v) tryptone, 0.5% (w/v) yeast extract, and 1% (w/v) KCl) supplemented with ampicillin (100 g/ml). Cells were harvested at an A 600 nm of ϳ1.0 and disrupted by passage through a Ribi cell fractionator at 20,000 pounds/square inch, 5-10°C, under a constant stream of nitrogen. After removal of intact cells and cell debris (9,000 ϫ g, 10 min), membrane vesicles were collected by centrifugation at 160,000 ϫ g for 60 min. Membrane vesicles were resuspended in 50 mM Tris/HCl, pH 7.5, containing 10% (v/v) glycerol, frozen in liquid nitrogen, and stored until use at Ϫ80°C. In case of cells containing KdpE, the cytoplasm was frozen in liquid nitrogen and stored at Ϫ80°C until use.
Purification of KdpD/1-395(6His) and 10His-KdpE-Solubilization, purification, and reconstitution into E. coli phospholipids of KdpD/1-395(6His) was carried out as described before (7,11). Because KdpD/ 1-395(6His) is a membrane-attached polypeptide (11), it was solubilized by detergent and then reconstituted into proteoliposomes using the same protocol as for integral membrane proteins. However, only the outside-located KdpD/1-395(6His) mattered in these experiments. Purification of 10His-KdpE was carried out in buffer E (20 mM Tris/HCl, pH 8.0, 5% (v/v) glycerol, 80 mM NaCl, 100 mM KCl, 10 mM MgCl 2 , 2 mM ␤-mercaptoethanol, 10 mM imidazole). Protein was bound batchwise to the resin, which had been equilibrated with buffer E, by incubation of the cytoplasmic proteins and the Ni 2ϩ -NTA resin at 4°C for 30 min. The protein-resin complex was then packed into a column, and unbound protein was removed by washing with buffer E. Bound 10His-KdpE was 2 W. Epstein, unpublished data. 3 Amino acid replacements are designated as follows: the one-letter amino acid code is used followed by a number indicating the position of the residue in the wild-type KdpD or KdpE, respectively. The second letter denotes the amino acid replacement at this position. 4 K. Altendorf, unpublished data. eluted by increasing the imidazole concentration to 100 mM. Before use, 10His-KdpE was dialyzed against buffer E without KCl and imidazole. Alternatively, KdpE was purified following a method described before (12).
Transposon Mutagenesis and Selection of the Mutants-For transposon mutagenesis, E. coli HAK006 was transformed with plasmids pBD3/1-395 and pNK2883. The mutagenesis was carried out in KML complex medium by incubation of the cells at room temperature for 48 h, initiating the transposition. Then the cells were plated on KML medium supplemented with 0.1% (w/v) 5-bromo-4-chloro-3-indolyl-␤-Dgalactopyranoside (X-gal), tetracycline (12.5 g/ml), and chloramphenicol (34 g/ml). The expression of the reporter gene lacZ was used as a measure for kdpFABC expression. White colonies (no kdpFABC expression) were isolated and again tested for ␤-galactosidase activity (28). To make sure that the mutants contain only one transposon, P1 lysates were prepared (19) and transduced in E. coli HAK006/pBD3/1-395. The loss of plasmid pNK2883 was verified by a negative selection on ampicillin (100 g/ml). Then the mutants were cured from plasmid pBD3/ 1-395 by negative selection on chloramphenicol.
In Vitro Transcription Experiments-In vitro transcription experiments were done similarly as described (31) in buffer TXN (40 mM Tris/HCl, pH 8.0, 50 mM KCl, 10 mM MgCl 2 , 10 mM dithiothreitol) in a total volume of 20 l. To obtain phosphorylated KdpE, purified protein was incubated with 50 mM acetyl phosphate and 10 mM MgCl 2 at 30°C for 1 h. The PstI/SacI fragment from the plasmid pSM5, containing the kdpFABC promoter/operator region, was used as template. KdpE or KdpEϳP, respectively, was incubated with the DNA template in the presence or absence of reconstituted KdpD/1-395(6His) in a total volume of 10 l at 37°C for 10 min. (w/v) xylene cyanol). Then samples were subjected to a 6% polyacrylamide, 7 M urea gel in TBE buffer (90 mM Tris, 90 mM borate, 2 mM EDTA). After the run, the gel was dried and exposed to a Storage Phosphor Screen. The amount of produced RNA was detected using the PhosphorImager SI (Amersham Biosciences).
Synthesis of [ 32 P]Acetyl Phosphate and Phosphorylation of KdpE-Synthesis and determination of [ 32 P]acetyl phosphate was carried out as described before (32). Phosphorylation of purified KdpE (0.2 mg/ml) was carried out in buffer TNaM (50 mM Tris/HCl, pH 7.5, 20 mM MgCl 2 , 50 mM NaCl) with 40 mM [ 32 P]acetyl phosphate (8 -14 mCi/mmol). The probes were incubated at 30°C; samples were taken at indicated time points, and the reaction was stopped by addition of SDS sample buffer (29). Proteins were subjected to 12.5% SDS gels, and after the run, the gels were dried and exposed to a Storage Phosphor Screen. Phosphorylated proteins were detected and quantified using the Phosphor-Imager SI (Amersham Biosciences).
Electrophoretic Mobility Shift Assay-For mobility shift assays, double-stranded DNA fragments comprising the KdpE-binding site (6,33) were used. These were obtained by annealing of two complementary oligonucleotides. The upper strand sequence (from 5Ј to 3Ј) has the following sequence: 5Ј-CATTTTTATACTTTTTTTACACCCCGCCCG-3Ј. After dephosphorylation of the double-stranded DNA with alkaline phosphatase (10 units), phosphorylation was carried out with polynucleotide kinase (10 units) and 10 Ci of [␥-32 P]ATP (3000 Ci/mmol) at 37°C for 1 h. The DNA was then purified using the nucleotide removal kit. Binding assays were done with 100 nM DNA and 0 -4.5 M purified KdpEϳP in TNaM buffer in a final volume of 20 l. KdpD/1-395(6His) (2.5 M final concentration) in 15 l of purification buffer (50 mM Tris/HCl, pH 7.5, 10% glycerol (v/v), 0.5 M NaCl, 10 mM ␤-mercaptoethanol, 0.04% n-dodecyl-␤-D-maltoside), or just purification buffer was added. After incubation at 30°C for 15 min, 3 l of sucrose dye solution (50% (w/v) sucrose, 0.25% (w/v) bromphenol blue; 0.25% (w/v) xylene cyanol] was added, and samples were loaded onto a 5% polyacrylamide gel in TBE buffer. After the run, the gel was dried and exposed to a Storage Phosphor Screen, and the amount of KdpEϳP-bound DNA was detected using the PhosphorImager SI (Amersham Biosciences).
Co-elution of KdpD/1-395 and 10His-KdpE from Ni 2ϩ -NTA-Agarose-For co-elution experiments, Ni 2ϩ -NTA magnetic agarose beads were used. First, magnetic beads (25 l of suspension) were washed with 500 l of distilled water in a 2-ml reaction tube and subsequently collected using a "12-tube magnet." Then the beads were equilibrated with buffer E (see above), and 100 l of cytoplasmic fraction containing 10His-KdpE was added (total protein concentration 15 mg/ml). As a control, 100 l of cytoplasmic fraction containing 10His-OmpR (34) was added. The mixture was incubated under gentle movement at 4°C for 30 min. In parallel, membrane vesicles containing KdpD/1-395 were solubilized with 0.2% (w/v) n-decyl-␤-D-maltopyranoside by stirring on ice for 30 min and centrifuged at 4°C at 264,000 ϫ g in a Beckman centrifuge for 30 min. Then 1.6 ml of the supernatant (5 mg/ml total protein concentration) containing KdpD/1-395 was added to Ni 2ϩ -NTA-bound 10His-KdpE or 10His-OmpR, respectively, so that the final ratio of each response regulator to KdpD/1-395 was about 1:8. Samples were incubated under gentle movement at room temperature for 150 min before the magnetic beads were collected with the magnet. The beads were washed with 500 l of buffer E containing 0.04% (w/v) n-decyl-␤-Dmaltopyranoside. Proteins bound to the beads were then eluted with 25 l of buffer E containing 0.04% (w/v) n-decyl-␤-D-maltopyranoside and 250 mM imidazole. Beads were removed by the magnet, and the eluted proteins were detected in an immunoblot with antibodies directed against the His tag as well as against KdpD.
Analytical Procedures-Protein was assayed by a modified Lowry method (35), using bovine serum albumin as a standard.

Influence of KdpD/1-395 on the Expression of kdpFABC-
The story started with a control experiment in which cells producing the N-terminal domain of KdpD (KdpD/1-395) and KdpE were tested for kdpFABC expression. Surprisingly, high kdpFABC expression levels were detected at any K ϩ concentration tested. In detail, E. coli HAK006 containing a kdpFABC promoter/operator-lacZ fusion as well as a chromosomal copy of kdpE and producing KdpD/1-395 at very low amounts (nondetectable by a Western blot) was grown in minimal medium at different K ϩ concentrations. Under all conditions high ␤-galactosidase activities were found in the presence of KdpD/1-395 (Fig. 3A). Under K ϩ -limiting conditions (Ͻ1 mM K ϩ ), an additional stimulation of the kdpFABC expression was observed, which was 2-3-fold higher compared with the expression produced by wild-type KdpD (Fig. 3A). High osmolality imposed by addition of NaCl or sucrose had no further stimulating effect on the constitutive expression of kdpFABC caused by KdpD/1-395 (data not shown).
To obtain further support for these unexpected results, Kd-pFABC production was directly measured and quantified by Western blotting (Fig. 3B). For this we used an E. coli strain (TKV2208) that contains the intact kdpFABC operon but lacks the major K ϩ uptake system Trk. It is important to note that in a ⌬trk strain the onset of kdpFABC induction is shifted to higher K ϩ concentrations (Ͻ60 mM) (36,37). Cells containing KdpD/1-395 produced higher amounts of KdpFABC complex at any K ϩ concentration than cells producing wild-type KdpD, thereby confirming the results of the transcriptional fusion experiments. Importantly, in the absence of KdpD or in the presence of KdpD with an inactivated phosphorylation site (KdpD/H673Q), no expression of kdpFABC was observed under any condition (data not shown). Furthermore, the inactivation of the regulatory ATP-binding site within the N-terminal domain by amino acid replacements (KdpD/1-395/G37A, K38A, and T39C) had only a small effect on kdpFABC expression (10 -30% reduction, data not shown). It should also be mentioned that KdpD/1-395 containing a C-terminal His tag [KdpD/1-395(6His)] produced the same expression pattern as KdpD/1-395. In summary, a truncated KdpD protein lacking the transmitter domain and the four transmembrane domains caused semi-constitutive kdpFABC expression, whereas the complete loss of KdpD or the inactivation of the phosphorylation site His-673 in full-length KdpD prevented kdpFABC expression.

Influence of KdpE and the Phosphorylation of KdpE on the
Semi-constitutive kdpFABC Expression Caused by KdpD/1-395-To gain further insight into this unexpected phenomenon, we investigated whether the cognate response regulator KdpE was involved in this process. To test the influence of KdpE on the kdpFABC expression caused by KdpD/1-395, plasmids pPV2/1-395 and pPV2/1-395/D52N, encoding KdpD/ 1-395 and KdpE, or KdpD/1-395 and KdpE/D52N, respectively, and pBD/1-395 encoding only KdpD/1-395, were transformed in E. coli RH003 (⌬kdpDE, chromosomal kdpFABC promoter/operator-lacZ fusion). Strains were grown in minimal media containing different concentrations of K ϩ , and kdpFABC expression was determined. Importantly, in the absence of KdpE, no expression of kdpFABC was detected (data not shown). This was also the case in the presence of KdpE with an inactivated phosphorylation site Asp-52 (KdpE/D52N) (data not shown). Also, no kdpFABC expression could be detected in the alternative test system using E. coli TKV2209 (intact kd-pFABC operon, ⌬trk) transformed with the plasmids described above in the absence of KdpE or when the KdpE phosphorylation site was inactivated (data not shown). These data clearly show that the response regulator KdpE and the phosphorylation of KdpE are required for the semi-constitutive kdpFABC

FIG. 3. Influence of KdpD/1-395 on the expression of kdpFABC.
A, quantification of kdpFABC expression using the reporter gene lacZ. E. coli strain HAK006 transformed with plasmid pBD/1-395 or pBD, respectively, was grown in minimal media containing different concentrations of K ϩ , and ␤-galactosidase activity was determined using the method of Miller (28). The data represent average values of at least three independent experiments. B, quantification of KdpFABC. Cells of E. coli strain TKV2208 transformed with plasmid pBD or pBD/1-395, respectively, were exposed to different K ϩ concentrations for 30 min. Cell extracts were subjected to 10% SDS-PAGE, and the amount of produced KdpFABC was identified by Western blotting using antibodies against KdpFABC and quantified using 125 I-labeled protein A. The amounts of KdpFABC synthesized were calculated according to a standard curve obtained from known amounts of KdpFABC and represent average values of at least three independent experiments. expression caused by KdpD/1-395. Because KdpD/1-395 is not able to phosphorylate KdpE in vitro (data not shown), there must be an alternative phosphodonor for KdpE.
Alternative Phosphodonors for KdpE in the Presence of KdpD/1-395-There are at least three different possibilities for an alternative phosphodonor for KdpE in the absence of fulllength KdpD. (i) KdpE could be phosphorylated by a low molecular weight phosphodonor like acetyl phosphate. (ii) Another sensor kinase could phosphorylate KdpE (cross-talk). (iii) KdpD/1-395 could interact with the transmitter domain of another sensor kinase. The latter possibility is supported by the fact that KdpD, separated into two independent peptides (the N-terminal domain of KdpD and the C-terminal domain including the transmitter domain), supports kdpFABC expression (11). Of course, it cannot be excluded that more than one possibility accounts for the phosphorylation of KdpE.
It has been shown previously (38) that KdpE can be phosphorylated by acetyl phosphate in vitro. Therefore, it was conceivable that acetyl phosphate could also be the phosphoryl donor for KdpE in vivo. To test this, plasmid pBD/1-395 was transformed in an E. coli strain that is unable to synthesize acetyl phosphate (RH001 (HAK006 ⌬ackA ⌬pta)). Cells were exposed to different K ϩ concentrations, and kdpFABC expression was determined. As shown in Fig. 4, acetyl phosphate is not essential for the kdpFABC expression caused by KdpD/1-395. However, compared with the control the kdpFABC expression is up to 50% diminished in the ⌬ackA ⌬pta strain. The same results were observed in the alternative test system (E. coli RH004 (TKV2208 ⌬ackA ⌬pta)) by detecting the amount of produced KdpFABC (data not shown). These results support the idea that KdpE is phosphorylated at least in part by acetyl phosphate in vivo in the presence of KdpD/1-395.
To test whether another protein participates in the signaling cascade, we applied transposon mutagenesis. The screen was performed in such a way that the inactivation of the gene searched for resulted in a large reduction of kdpFABC expression. Transposon mutagenesis was carried out in E. coli HAK006/pBD3/1-395 as described under "Experimental Procedures," and the resulting clones were screened for white colonies. Four possibilities exist for a reduced or lost ␤-galactosidase activity: (i) the transposon inactivated kdpD/1-395; (ii) the transposon inactivated kdpE; (iii) the transposon inactivated the reporter gene lacZ; and (iv) the transposon inactivated the gene searched for. About 12,000 -15,000 mutants were screened for a reduced or lost ␤-galactosidase activity indicated by white colonies on indicator plates, and 11 mutants were found. To exclude that the transposon inactivated kdpD/1-395, the plasmids from these mutants were isolated, and the correct size of the kdpD/1-395 gene was checked by restriction analysis. After cloning of the chromosomal fragments harboring the transposon and sequencing the Tn10-adjacent parts, the inactivated genes were identified. Either the kdpE gene or the lacZ gene was inactivated by the transposon, and no other gene that is essential for the phosphorylation of KdpE was identified. Because of the high number of screened mutants, it is unlikely that a gene encoding an essential phosphodonor for KdpE exists. Based on these results we propose that the phosphorylation of KdpE is mediated by acetyl phosphate and other low molecular weight phosphodonors like carbamoyl phosphate, for example, in the absence of full-length KdpD.
In Vitro Transcription of kdpFABC Is Stimulated by KdpD/ 1-395-The question still remained regarding how the N-terminal domain of KdpD causes semi-constitutive kdpFABC expression. To address this, an in vitro transcription experiment with purified protein components was performed. A DNA fragment comprising 622 bp upstream of the kdpFABC start codon and the first 1518 bp of the kdpFABC operon (kdpF and a part of kdpA) served as a template, and the in vitro transcription experiments were performed as described under "Experimental Procedures." As shown in Fig. 5, there is a basal level of kdpFABC transcription in the presence of non-phosphorylated KdpE. Transcription is about 10-fold higher in the presence of phosphorylated KdpE. When reconstituted KdpD/1-395(6His) was present, transcription reached highest values, whereby slightly higher activities were detectable in the case of phosphorylated KdpE. As a control, the same experiment was performed in the presence E. coli liposomes without KdpD/1-395(6His). The presence of liposomes alone already had a stimulating effect on the in vitro transcription of kdpFAЈ, which seems to be unspecific and has to be regarded as a positive effect of the phospholipid bilayer on the transcription apparatus as described earlier (39). Presumably, because of the high background no stimulation of transcription could be detected in the case of KdpEϳP compared with KdpE when liposomes were present. In summary, these results clearly show that KdpD/1-395 has a stimulating effect on the transcription of kdpFABC in vitro.
Influence of KdpD/1-395 on the Stimulation of kdpFABC Transcription-There appear to be two different possibilities regarding how KdpD/1-395 might stimulate kdpFABC transcription: (i) KdpD/1-395 facilitates the phosphorylation of KdpE by acetyl phosphate or (ii) KdpD/1-395 stabilizes the binding of KdpEϳP to the DNA. To check the first possibility, the time-dependent phosphorylation of purified KdpE by [ 32 P]acetyl phosphate was determined in vitro in the absence and presence of purified and reconstituted KdpD/1-395(6His). As shown in Fig. 6, KdpD/1-395(6His) did not stimulate the phosphorylation of KdpE. In addition, no differences were observed regardless of whether KdpD/1-395(6His) in proteoliposomes or in solution was tested (data not shown). Therefore, these results argue against the first hypothesis.
To test the second hypothesis, we applied electrophoretic mobility shift assays. For this a double-stranded DNA fragment comprising the KdpE-binding site (6,33) was used and incubated with increasing amounts of phosphorylated KdpE. Reconstituted KdpD/1-395(6His) could not be applied in this experiment, because the proteoliposomes caused precipitation of the samples in the pockets of the native gel. For this reason, KdpD/1-395(6His) in detergent was used in this experiment and buffer containing detergent as a control. The detergent was FIG. 4. Influence of acetyl phosphate on the semi-constitutive kdpFABC expression caused by KdpD/1-395. kdpFABC expression was quantified using the reporter gene lacZ. E. coli strains HAK006 and RH001 transformed with plasmid pBD/1-395 or pBD, respectively, were grown in minimal media containing different concentrations of K ϩ , and ␤-galactosidase activity was determined using the method of Miller (28). The data represent average values of at least three independent experiments. necessary because KdpD/1-395(6His) precipitates in buffer without detergent. As shown in Fig. 7, the amount of shifted DNA is significantly increased in the presence of soluble KdpD/ 1-395(6His) compared with the control. However, an additional band corresponding to the KdpD/1-395-KdpE-DNA complex was missing indicating that there is only a weak interaction between KdpD/1-395 and KdpEϳP. Based on these results it is proposed that the N-terminal domain of KdpD (KdpD/1-395) stabilizes the binding of KdpEϳP to its corresponding DNAbinding site.
Biochemical Evidence for the Interaction between KdpD/1-395 and KdpE-To obtain direct evidence for an interaction between the N-terminal domain of KdpD (KdpD/1-395) and KdpE, a co-elution experiment with 10His-KdpE and KdpD/1-395 from Ni 2ϩ -NTA agarose magnetic beads was performed. Briefly, 10His-KdpE was first bound to the beads and then solubilized KdpD/1-395 was added in a final ratio of 1:8 (KdpE: KdpD/1-395) (see "Experimental Procedures" for details). As a control, 10His-OmpR, the response regulator of the EnvZ/ OmpR system of E. coli (34), was bound to the magnetic beads and incubated in the same manner with KdpD/1-395 as described above. Then the non-bound proteins were first washed off, and bound proteins were eluted from the beads and detected in an immunoblot with antibodies directed against the His tag as well as against KdpD. As shown in Fig. 8, the non-tagged KdpD/1-395 co-eluted with 10His-KdpE from the column (lane 1), but no co-elution of KdpD/1-395 was observed with the control protein 10His-OmpR (lane 2), demonstrating the specific interaction between the N-terminal domain of KdpD and the response regulator KdpE. DISCUSSION KdpD differs from other sensor kinases by its very large 395-amino acid-comprising, N-terminal domain, which is part of the input domain (3,11). Here we describe that in the absence of the C-terminal transmitter domain and the four transmembrane domains, the N-terminal domain of KdpD causes semi-constitutive kdpFABC expression in vivo. This observation is very astonishing, because in a kdpD null mutant or in cells producing KdpD with an inactivated phosphorylation site (KdpD/H673Q), no expression of kdpFABC was observed. However, we found that this unusual signaling requires the response regulator KdpE with an intact phosphorylation site. As expected and experimentally confirmed in this study, purified KdpD/1-395 was not able to phosphorylate KdpE. This led to the conclusion that KdpE was phosphorylated in a different way in vivo. It has been shown before that KdpE can be phosphorylated by acetyl phosphate in vitro (38). Furthermore, other response regulators can be phosphorylated by small molecules like imidazole phosphate, acetyl phosphate, carbamoyl phosphate, or phosphoramidate, whereby high concentrations FIG . 5. In vitro transcription of kdpFABC. The experiment was performed as described under "Experimental Procedures." Briefly, DNA fragments comprising the KdpE-binding site were incubated with KdpE or KdpEϳP, respectively, and RNA polymerase in the presence or absence of KdpD/1-395(6His) in proteoliposomes. Transcription was started by addition of nucleotides including [␣-32 P]UTP and carried out for 10 min. Then the reactions were stopped, and samples were loaded onto a 6% polyacrylamide/urea gel, and the amount of produced mRNA was quantified using PhosphorImaging. of these substances must be present in vitro (40 -42). To test whether KdpE is phosphorylated by acetyl phosphate in vivo, we used E. coli strains that are unable to synthesize acetyl phosphate. In these mutants, KdpD/1-395-mediated kdpFABC expression was decreased up to 50%. This suggests that KdpE is in part phosphorylated by acetyl phosphate under these circumstances. Similar results were obtained for the response regulators OmpR of E. coli and PhoP of Salmonella enterica that are phosphorylated by acetyl phosphate in the absence of their cognate sensor kinases (43,44). Moreover, for the response regulator RssB of E. coli, no cognate sensor kinase is known, and it is thought that the phosphorylation of RssB is exclusively regulated by the intracellular concentration of acetyl phosphate (45). Alternatively, KdpE could be phosphorylated by another sensor kinase, the so-called "cross-talk," or KdpD/1-395 forms a hybrid kinase with the transmitter domain of another sensor kinase, which then phosphorylates KdpE. These are obviously very often discussed theories that are difficult to prove. However, by using transposon mutagenesis no additional gene was detected. Therefore, we conclude that KdpE is phosphorylated by acetyl phosphate in the absence of an intact KdpD. However, the results clearly indicate that acetyl phosphate is not the only phosphoryl donor. It is quite conceivable that KdpE is phosphorylated by other phosphodonors like carbamoyl phosphate or phosphoramidate as well.
The question still remained how KdpD/1-395 causes semiconstitutive kdpFABC expression. KdpD/1-395 did not affect the phosphorylation of KdpE by acetyl phosphate in vitro. However, there must be a specific function of KdpD/1-395 because no kdpFABC expression was detectable in the absence of KdpD. We could clearly show that the transcription of kdp-FABC was enhanced in vitro in the presence of reconstituted KdpD/1-395. By performing such experiments, care has to be taken because transcription is enhanced per se in a lipidic environment (39). Because the amount of kdpFAЈ transcripts was highest in the presence of reconstituted KdpD/1-395, we conclude that the enhanced kdpFABC transcription in vitro is probably due to two effects: (i) the presence of KdpD/1-395, and (ii) the presence of a lipidic environment presented by the phospholipid vesicles.
There are at least two possibilities regarding how KdpD/1-395 affects kdpFABC transcription: (i) the stabilization of phosphorylated KdpE, and (ii) the stabilization of the binding of KdpEϳP to its corresponding binding site upstream of the kdpFABC-promoter/operator region. In the presence or absence of KdpD/1-395 nearly the same phosphorylation kinetics of KdpE by acetyl phosphate in vitro were observed, which argues against the first hypothesis. The mobility shift experiments provide evidence for the second hyphothesis; a much higher amount of DNA-bound KdpEϳP could be observed in the presence of KdpD/1-395 compared with the control experiment. In comparison to earlier results (18), the affinity of KdpEϳP to the DNA was relatively low in our experiments. This might be due to the presence of detergent, which was necessary to keep KdpD/1-395 in solution. These conditions might also weaken the interaction between KdpD/1-395 and KdpE, and therefore a signal for a ternary complex consisting of KdpEϳP, DNA, and KdpD/1-395 is missing. Finally, electrophoretic mobility shift assay experiments in the presence of KdpD/1-395 in proteoliposomes resulted in the formation of aggregates in the gel pockets and therefore led to misleading results. Despite these experimental difficulties, a stabilizing effect of KdpD/1-395 on the KdpEϳP-DNA interaction was clearly detectable. This is the first example for a stabilization of the interaction between a response regulator and its corresponding DNA-binding site by a domain of the cognate sensor kinase. Furthermore, it seems likely that a lipidic environment also promotes transcription efficiency, an effect that was already described earlier (39). Coincidentally, the separately produced N-terminal domain of KdpD results in a membrane-attached polypeptide, which can be released from the membrane by detergent or by washing with buffer of low ionic strength (11). Moreover, the analysis of truncated KdpD derivatives lacking different parts of the transmembrane spanning domains indicated that a certain distance between the N-and C-terminal cytoplasmic domains is required to allow signal transduction. In addition, it has been demonstrated that the N-and C-terminal domains interact with each other (10,11). Based on these results a model was established according to which the N-and C-terminal domains move toward each other under inducing (phosphorylating) conditions. It is conceivable that the movement of the cytosolic domains also includes conformational changes leading to the liberation of structures within the N-terminal domain which then promote the stabilization between KdpEϳP and the DNA (Fig. 9). This would explain why cells producing the non-phosphorylatable KdpD/H673Q prevent or cells producing the N-terminal truncated KdpD/⌬12-395 (8) diminish kdp-FABC expression. The direct interaction between the N-termi-FIG. 8. Co-elution of KdpD/1-395 and 10His-KdpE from Ni 2؉ -NTA magnetic agarose beads. First, 10His-KdpE was bound to the beads, and then solubilized membrane proteins containing "untagged" KdpD/1-395 were added. In a control experiment, 10His-OmpR (34) was used instead of 10His-KdpE. After incubation at room temperature, non-bound proteins were washed off, and bound proteins were eluted by increasing the imidazole concentration. The figure shows an immunoblot with antibodies directed against KdpD and the His tag. nal domain of KdpD and KdpE has been shown by co-elution experiments. According to the literature a quite similar interaction has been shown for the sensor kinase UhpB and the response regulator UhpA. UhpB participates not only in the phosphorylation control of UhpAϳP but also binds UhpA specifically. Therefore expression of the target gene uhpT is tightly regulated (31,46).
Despite the stabilizing effect of the N-terminal domain of KdpD on the KdpEϳP-DNA interaction, the question remains why kdpFABC expression was found to be dependent on the extracellular K ϩ concentration under K ϩ -limiting growth conditions. These effects are difficult to explain right now. It seems likely that other proteins channel additional information to KdpE under extreme K ϩ -limiting conditions. In this regard it is interesting to note that H-NS (hns) and thioredoxin reductase (trxB) somehow affect kdpFABC expression in E. coli (47). Furthermore, a very recent report (48) proposes the interaction of the input domain of KdpD and lipoproteins, namely LprJ and LprF, in Mycobacterium tuberculosis H37Rv. All these data indicate a more complex regulatory network for the regulation of kdpFABC expression than originally assumed for the KdpD/ KdpE two-component system.