Interacting specificity of a histidine kinase and its cognate response regulator: the PrrBA system of Rhodobacter sphaeroides

Two-component signal-transduction systems play major roles in cellular adaptation to various environmental conditions in prokaryotes (Grebe & Stock, 1999; West & Stock, 2001). The typical two-component system consists of a histidine kinase and a cognate response regulator. The orthologous form of the histidine kinase comprises the N-terminal sensing domain responsible for signal sensing and the C-terminal transmitter domain that is involved in ATP-dependent phosphorylation of the cognate response regulator. The transmitter domain is further divided into the dimerization domain (DD) and the catalytic ATP-binding (CA) domain (Dutta et al., 1999). The DD consists of ∼70 amino acid residues that form a secondary structure of α-helixturnα-helix (Tomomori et al., 1999). Histidine kinases exist as homodimers, in which the DDs of two identical subunits form a four-helix bundle. In the middle of the first (N-terminal) α-helix of the DD a histidine residue is conserved among all histidine kinases and the histidine residue serves as the autophosphorylation site. The CA domain exhibits a structural homology with ATPases such as Hsp90, DNA gyrase B and MutL, and contains conserved signatures of histidine kinases (N, G1, F and G2 boxes), which form the unique ATP-binding pocket (Dutta et al., 1999). This domain is responsible for the phosphotransfer reaction from ATP to the histidine residue conserved in the DD. The CA domain of one subunit is juxtaposed to the DD of the other subunit. Due to the highly flexible interdomain hinges between CA and DDs, the 3D structure of the intact transmitter domain of a histidine kinase is yet to be solved. However, the structures of each independent domain of the histidine kinase have been resolved (Tanaka et al., 1998; Tomomori et al., 1999).

The response regulator usually has a two-domain structure, with a conserved N-terminal regulatory domain and a variable C-terminal effector domain (West & Stock, 2001). The regulatory domain is also called the receiver domain and contains a conserved aspartate residue that receives the phosphoryl group from a phosphorylated histidine kinase. The phosphorylation of the regulatory domain brings about the conformational change, which leads to the activation of the effector domain that usually functions as the DNA-binding domain. Many histidine kinases are bifunctional enzymes that have both kinase and phosphatase activities acting on their cognate response regulators (Stock et al., 2000; Oh & Kaplan, 2001; Oh et al., 2004). Depending on the presence or absence of an environmental signal, the equilibrium of the kinase/phosphatase activities of a histidine kinase is regulated, and this in turn controls the cellular abundance of the phosphorylated response regulator that forms the phosphorelay couple with the histidine kinase.

The photosynthetic bacterium Rhodobacter sphaeroides can grow in various environments and has been shown to possess extensive metabolic capability (Kiley & Kaplan, 1988; Zeilstra-Ryalls et al., 1998; Oh & Kaplan, 2001). In order to adapt to various environmental conditions, R. sphaeroides possesses approximately 30 different pairs of two-component systems, excluding the CheA-CheY chemotaxis systems (). In organisms with multiple two-component systems, signal transduction through these two-component systems requires precise interactions between histidine kinases and their cognate response regulators to elicit the correct response in adaptation to a given environmental signal. Currently little is known about the mechanisms underlying the specific interaction between a histidine kinase and a cognate response regulator. Using the PrrBA and DevSR two-component systems that are involved in redox sensing in R. sphaeroides and Mycobacterium smegmatis (Eraso & Kaplan, 1994, 1995; Mayuri et al., 2002), respectively, we performed in vivo mapping of the interaction domains of the histidine kinases and their cognate response regulators by means of yeast two-hybrid assays, and revealed regions (domains) on the histidine kinases that confer recognition specificity for their cognate response regulators.

Strains, plasmids, and growth conditions.
The Escherichia coli strains, Saccharomyces cerevisiae strains and plasmids used in this study are listed in Table 1. E. coli strains were grown at 37 °C in LuriaBertani (LB) medium as described elsewhere (Sambrook et al., 1989). Ampicillin (100 µg µl¹) and kanamycin (50 µg µl¹) were added when appropriate. S. cerevisiae strains were grown at 30 °C in YPD (Clontech) or synthetic defined dropout (SD) medium (Q-Biogene) lacking adenine, histidine, leucine and tryptophan (SD/Ade/His/Leu/Trp) in the presence of different concentrations of 3-amino-1,2,4-trizole (3AT; Sigma) as described in the manufacturer's manual ().

Table 1. Strains and plasmids used in this study Abbreviations: AD, activation domain; DNA-BD, DNA-binding domain; ATP BD, ATP binding domain.

Molecular genetic techniques.
Standard protocols and manufacturer's instructions were followed for general recombinant DNA manipulations (Sambrook et al., 1989). S. cerevisiae strains were cotransformed with different pairs of two-hybrid plasmids according to the lithium acetate (LiAc)-mediated method as described elsewhere (Guthrie & Fink, 1991).

Construction of plasmids
i) pPLA and pBBC.
To construct pPLA, a 0.64 kb EcoRIBamHI fragment containing prrA was amplified with the primers PrrA(EcoRI)+ (5'-ccttgaattccgaatgacagccatgtgtc-3') and PrrA(BamHI) (5'-cttcggatcctcagcgcgggctgcgcttc-3') using the plasmid pUI1643 as the template and pfu DNA polymerase. The PCR product was restricted with EcoRI and BamHI and cloned into pGADT7 encoding the GAL4 activation domain (GAL4AD) digested with the same restriction enzymes, yielding pPLA. Most prrA derivatives were cloned into pGADT7 by the same procedure as described above to produce the fusion proteins (GAL4ADPrrA derivatives). To construct pBBC, a 0.91 kb BamHIPstI fragment containing a portion of prrB was amplified by PCR using the primers PrrBc(BamHI)+ (5'-tataggatccgcatgttcgaattcggc-3') and PrrB(PstI) (5'-atatctgcagatcaggtctggatcagg-3'), digested with BamHI and PstI, and cloned into pGBKT7, giving pBBC. Most PrrB derivatives were fused to the C terminus of the GAL4 DNA-binding domain (GAL4BD) employing the vector pGBKT7.

ii) pPLR and pBSC.
A 0.74 kb EcoRIBamHI fragment containing devR was amplified by PCR using the primers DevR(EcoRI)+ (5'-atgcgaattcaccccgcggctctgggtgcgc-3') and DevR(BamHI) (5'-atgcggatccctagttgcgccggtccagttt-3') from M. smegmatis MC2 chromosomal DNA as the template. A 0.69 kb EcoRIBamHI fragment containing a portion of devS was amplified with the primers DevSc(EcoRI)+ (5'-atgcgaattccctttcaccgcagaacagttg-3') and DevS(BamHI) (5'-atgcggatcctcagtcggggagcggcgcggt-3') by the same procedure. The DNA fragments containing devR and a portion of devS were digested with EcoRI and BamHI and cloned into pGADT7 and pGBKT7 restricted with the same restriction enzyme, yielding pPLR and pBSC, respectively.

iii) pBBSDI and pBBSDII.
A recombinant PCR method was performed to generate PrrBDevS chimeric proteins. Two rounds of PCR were carried out using pfu polymerase. The plasmids pUI1643 and pBSC were used as the templates for amplification of portions of prrB and devS, respectively. Two primary PCR reactions were performed with the primers PrrBc(BamHI)+ and BSI (5'-ctgctgcacctcggccgaacgaagctcctcggcaagctccg-3') and with BSI+ (5'-cggagcttgccgaggagcttcgttcggccgaggtgcagcag-3') and DevS(BamHI) (5'-atgcggatcctcagtcggggagcggcgcggt-3') to generate two DNA fragments each containing a 42 bp overlapping region. Two primary PCR products were used as the templates for the secondary PCR, which was performed using the primers PrrBc(BamHI)+ and DevSD(BamHI) (5'-atatggatccatcgaagatggccgtgcg-3'). The secondary PCR product was restricted with EcoRI and BamHI and cloned into pGBKT7 digested with the same restriction enzymes to yield pBBSDI. The plasmid pBBSDII was constructed in the same way as pBBSDI, except for using the primers BSII+ (5'-cgaacgtgcgcgcgggatcgtcgagctcaccagcttgatcgt-3') and BSII (5'-acgatcaagctggtgagctcgacgatcccgcgcgcacgttcg-3') in place of BSI+ and BSI.

iv) pPR and pBZD.
To construct pPR, a 0.72 kb EcoRIBamHI fragment containing ompR was amplified by PCR using the primers OmpR(EcoRI)+ (5'-atatgaattcatgcaagagaactacaag-3') and OmpR(BamHI) (5'-atatggatcctcatgctttagagccgtc-3') from E. coli chromosomal DNA as the template. The PCR product was restricted with EcoRI and BamHI and cloned into pGADT7 digested with the same restriction enzymes, yielding pPR. A 0.21 kb EcoRIBamHI fragment containing a portion of envZ was amplified with the primers EnvZDD (EcoRI)+ (5'-atatgaattcatggcggctggtgttaag-3') and EnvZDD (BamHI) (5'-atatggatccctcctgcccggtgcgcag-3') by PCR, digested with EcoRI and BamHI, and cloned into pGBKT7, giving pBZD.

Site-directed mutagenesis.
To mutate codons corresponding to Asp-19, Asp-20, Asp-63, Thr-91 and Lys-113 of PrrA to alanine, mutagenesis was performed using the Quick Change Site-directed Mutagenesis kit (Stratagene) using pULAR as a template plasmid. Synthetic oligonucleotides 33 bases long containing an alanine codon (GCC, GCT, GCG) in place of the codon of the corresponding amino acid in the middle of their sequences were used to mutagenize the prrA gene. Following the verification of mutations by DNA sequencing, a 0.48 kb EcoRIBamHI fragment containing the mutated sequence was cloned into pGADT7, giving pPLAR19, pPLAR20, pPLAR63, pPLAR91 and pPLAR113.

Analysis of in vivo proteinprotein interactions.
The S. cerevisiae strains co-transformed with both pGADT7 and pGBKT7 derivatives were grown in SD/Leu/Trp medium. The cultures were diluted to OD₆₀₀ 0.40.45 and spotted onto both solid SD/Leu/Trp medium and SD/Leu/Trp/His medium containing different concentrations of 3AT. These plates were incubated for 37 days. The crude cell extracts were prepared by three freeze/thaw cycles using liquid nitrogen. Determination of β-galactosidase activity was performed using ONPG as substrate, as described elsewhere (Schneider et al., 1996).

Immunoblotting.
SDS-PAGE and Western blotting were performed as described elsewhere (Laemmli, 1970; Mouncey & Kaplan, 1998). The protein extracts were resolved by 12.5 % (w/v) SDS-PAGE and transferred to PVDF membranes. The mouse mAbs against GAL4AD (SantaCruz) and GAL4BD (BD Biosciences) were used at dilutions of 1 : 200 and 1 : 10 000, respectively. The alkaline phosphatase-conjugated anti-mouse IgG (Sigma) was used at a 1 : 10 000 dilution for the detection of the primary antibodies. Detection was carried out by staining the membrane with 5-bromo-4-chloro-3-indolyl phosphate (BCIP) and nitroblue tetrazolium (NBT).

Detection of specific proteinprotein interactions between histidine kinases and their cognate response regulators in a yeast two-hybrid system
This study was designed to reveal which regions (domains) of the PrrA response regulator and its cognate PrrB histidine kinase confer recognition specificity for proteinprotein interaction. We first determined whether the yeast two-hybrid system could be applied to detect specific interactions between the histidine kinase and its cognate response regulator. To employ the yeast two-hybrid assay, the genes encoding the PrrA response regulator and the PrrB histidine kinase were cloned into pGADT7 (prey vector) and pGBKT7 (bait vector), respectively, to generate GAL4ADPrrA and GAL4BDPrrB fusion proteins in the yeast strain AH109. In the case of PrrB, the 3' portion of the prrB gene encoding the transmitter domain (PrrBc) lacking the N-terminal membrane-spanning domain (amino acids 1161), was cloned into the bait vector, since the PrrB transmitter domain was predicted to interact with PrrA on the basis of the in vitro phosphorylation assay (Bird et al., 1999; Comolli et al., 2002; Oh et al., 2004). The DNA fragment which was cloned into the prey vector to produce a GAL4ADPrrA fusion protein contained the 84 bp DNA sequence upstream of the prrA start codon in addition to the entire prrA gene. Since this additional DNA sequence does not contain a stop codon that lies in the same reading frame as the GAL4AD and prrA genes, the PrrA protein expressed from pPLAC (pGADT7 : : prrA) has an additional 28 amino acid residues at its N terminus. In order to assess the specificity of the interaction between the histidine kinase and its cognate response regulator using the yeast two-hybrid assay, another two-component system, the DevSR two-component system from M. smegmatis, was included. The genes for the DevS histidine kinase and DevR response regulator were cloned into the bait and prey vectors, respectively, in the same way as for the PrrBA two-component system (DevR of the GAL4ADDevR fusion protein also has an additional 28 amino acid residues at its N terminus). When proteinprotein interactions between GAL4AD and GAL4BD fusion proteins occur in the yeast strain AH109 carrying both bait and prey plasmids, the yeast strain is enabled to grow on histidine-deficient medium and synthesize β-galactosidase due to the induction of the HIS3 and lacZ reporter genes on its chromosomal DNA. It was further possible to comparatively determine the strength of the proteinprotein interactions by using various concentrations of 3AT, which serves as an inhibitor of the HIS3 product.

As shown in Fig. 1, the yeast strain (PrrA/PrrBc) expressing both the GAL4ADPrrA and the GAL4BDPrrBc fusion proteins grew on histidine-deficient medium (SD medium lacking histidine), indicating that PrrA interacts with PrrBc. Likewise, the yeast strain with the DevR/DevSc pair grew on SD medium without histidine, and its growth in the presence of 5 mM 3AT was more robust than that of the yeast strain (PrrA/PrrBc) under the same conditions. This result indicated that the proteinprotein interactions between DevR and DevSc are stronger than those between PrrA and PrrBc. The yeast strains (PrrA/DevSc and DevR/PrrBc) which expressed heterologous pairs of fusion proteins did not grow on SD medium lacking histidine. Taken together, the data presented in Fig. 1 demonstrate that the yeast two-hybrid system used in this study is applicable for the study of the specific proteinprotein interactions between a histidine kinase and its cognate response regulator.

(43K): Fig. 1. Detection of specific proteinprotein interactions between histidine kinases and their cognate response regulators by the yeast two-hybrid assay. Growth of the yeast strains transformed with both the bait and prey plasmids in the yeast two-hybrid assay is shown. The yeast strains were grown at 30 °C in SD/Leu/Trp medium. All yeast cultures were diluted to OD600 0.40.45 and spotted onto a SD/Leu/Trp plate or histidine-deficient SD plates (SD/Leu/Trp/His) containing different concentrations of 3AT (the numbers to the left indicate the concentration of 3AT). Pairs of fusion constructs are indicated in the order X/Y, where X and Y are the proteins fused to GAL4AD and GAL4BD, respectively. The level of growth is indicated with + or . The SD/Leu/Trp plate and SD/Leu/Trp/His/+3AT plates were incubated at 30 °C for 3 and 5 days, respectively.

The regulatory domain of PrrA is the region that interacts with PrrB
We next examined which domain of PrrA is responsible for its specific interaction with PrrB. The PrrA response regulator is composed of the N-terminal regulatory domain (amino acids 1131) and the C-terminal effector domain (amino acids 136184), which are connected by a linker consisting of four consecutive proline residues (Eraso & Kaplan, 1994). The regulatory domain of PrrA contains an aspartate residue (Asp-63) that receives the phosphoryl group from autophosphorylated PrrB (Comolli et al., 2002), while the effector domain functions as the DNA-binding domain (Laguri et al., 2003).

As shown in Fig. 2(A), the yeast strain (PrrAr/PrrBc) expressing both GAL4ADPrrAr (regulatory domain of PrrA) and GAL4BDPrrBc fusion proteins grew on SD medium lacking histidine. The growth of the yeast strain (PrrAr/PrrBc) on histidine-deficient medium supplemented with 3AT was more robust than that of the yeast strain (PrrA/PrrBc) under the same growth conditions. In keeping with this observation, the β-galactosidase assay revealed that a 1.9-fold higher activity of β-galactosidase was detected in cell lysates of the yeast strain (PrrAr/PrrBc) compared to the yeast strain (PrrA/PrrBc). The PrrA derivative (PrrAe), in which the regulatory domain was removed from PrrA, did not interact with PrrBc in the yeast strain (PrrAe/PrrBc). The yeast strain (P/PrrBc) harbouring the empty prey vector (P) as a negative control did not grow on histidine-deficient medium. Taken together, these results indicate that the regulatory domain of PrrA is the region in which proteinprotein interactions between PrrA and PrrB occur, and that the PrrA derivative (PrrAr) consisting of the regulatory domain alone appears to interact with PrrBc more strongly than does intact PrrA, suggesting a negative effect of the effector domain on this interaction.

(51K): Fig. 2. Mapping of the PrrA domain interacting with PrrB. (A) Schematic representation of PrrA derivatives is on the right of the figure. The numbers indicate the amino acid boundaries of the PrrA domains. The regulatory and effector domains of PrrA are connected by the linker consisting of four consecutive proline residues. The panel on the left shows the results of the yeast two-hybrid assay. Cultures were spotted in the same way as described in Fig. 1 and fusion pairs are indicated in the same order as described in Fig. 1. P (prey) refers to the absence of a protein fused to GAL4AD. The strains grown on the SD/Leu/Trp plate were incubated for 3 days and the other plates for 6 days. The numbers to the left of the growth panel indicate the specific β-galactosidase (β-gal) activities that are normalized by the value for the control strain (P/PrrBc); the numbers at the top of the growth panel show 3AT concentration. (B) Western blotting analysis for the detection of GAL4AD fusion proteins using a GAL4AD mAb. The numbers to the left indicate the molecular mass (kDa).

In order to delineate more precisely the region within the PrrA regulatory domain that is involved in proteinprotein interactions with PrrB, C-terminal deletion derivatives of PrrAr were constructed and their interactions with PrrBc were examined. The PrrAr derivatives in which 10 and 20 amino acids were removed from the C terminus of PrrAr lost their ability to interact with PrrBc, as judged by the two-hybrid assay. Interestingly, the removal of 10 and 20 amino acids from the N terminus of PrrA with the 28 extra amino acid extension at its N terminus rendered the PrrA derivatives (10PrrA and 20PrrA) unable to interact with PrrBc, as observed for the yeast strains (10PrrA/PrrBc and 20PrrA/PrrBc), although the PrrA protein itself remained intact. The PrrA derivative from which all 28 extra amino acids had been removed did not interact with PrrBc either (data not shown).

To determine whether the expression and/or stability of the GAL4ADPrrA fusion proteins in yeast cells were affected by these deletions, we performed Western blotting analysis with a mAb against GAL4AD. As shown in Fig. 2(B), all GAL4AD fusion proteins were detected in the Western blotting assay, ruling out the possibility that the inability of some PrrA derivatives to interact with PrrBc might have resulted from instability or degradation of their fusion proteins. From the results it can be inferred that the intact form or conformation of the PrrA regulatory domain is required for the correct proteinprotein interactions between PrrA and PrrB, and that it is necessary to insert a linker of the proper length between GAL4AD and PrrA when the GAL4ADPrrA fusion proteins for the yeast two-hybrid assay are constructed.

D63A and K113A substitutions in the regulatory domain of PrrA abolish proteinprotein interactions between PrrA and PrrB
The regulatory domains of response regulators have a (β/α)₅ topology with five parallel β-strands surrounded by five α-helices (Volz & Matsumura, 1991; Stock et al., 1993). The active site located in the regulatory domain of the response regulator is believed to be composed of three acidic residues, a lysine residue and a hydroxyl amino acid (threonine or serine), which are all conserved in all known response regulators (Fig. 3A). The corresponding conserved active site residues in PrrA are Asp-19, Asp-20, Asp-63, Thr-91 and Lys-113. Asp-63 has been demonstrated to be the site of phosphorylation (Comolli et al., 2002). Asp-19 and Asp-20 are predicted to be involved in coordination of a Mg²⁺ ion that is essential for phosphorylation and dephosphorylation of the response regulator (Stock et al., 1993). It is assumed that Thr-91 and Lys-113 are involved in the phosphorylation-induced conformational change of the response regulator (Lukat et al., 1991; Appleby & Bourret, 1998).

(34K): Fig. 3. Effect of replacement of conserved amino acids in the regulatory domain of PrrA on proteinprotein interactions between PrrA and PrrB. (A) Multiple alignment of the receiver domain of PrrA with those of other response regulators. The amino acid residues which are conserved in the regulatory domains of all response regulators and play a crucial role in the phosphorylation reaction are boxed. These amino acids are substituted with alanine by site-directed mutagenesis. Abbreviations: RsPrrA, PrrA of R. sphaeroides; RcRegA, RegA of R. capsulatus; EcOmpR, OmpR of E. coli; RmFixJ, FixJ of Sinorhizobium meliloti; MsDevR, DevR of M. smegmatis. The final r signifies the regulatory domain. Identical amino acids are indicated by asterisks. The similar and less similar amino acid residues are marked with : and ., respectively. (B) Yeast two-hybrid assay. All yeast strains tested carried pBBC expressing the GAL4BDPrrBc fusion protein. Pairs of fusions are indicated in the order X/Y, where X and Y are the proteins fused to GAL4AD and GAL4BD, respectively. The level of growth is indicated with + or . P refers to the absence of a protein fused to GAL4AD. The SD/Leu/Trp plate and the SD/Leu/Trp/His plates supplemented with 3AT were incubated at 30 °C for 3 and 5 days, respectively. (C) Western blotting analysis for the detection of GAL4AD fusion proteins using a GAL4AD mAb. The number to the left indicates the molecular mass (kDa).

In order to investigate the possible roles of the conserved active site residues in proteinprotein interactions between PrrA and PrrB, the conserved amino acid residues in the PrrA regulatory domain were individually changed to an alanine residue by site-directed mutagenesis, and interactions of the mutant forms of the PrrA regulatory domain with PrrBc were assessed by the yeast two-hybrid assay. As shown in Fig. 3(B), the positive-control yeast strain (PrrAr/PrrBc) grew on histidine-deficient medium, while the yeast strain (P/PrrBc) carrying the empty prey vector did not grow on histidine-deficient medium. The mutant forms (PrrArD19A, PrrArD20A and PrrArT91A) of the PrrA regulatory domain in which Asp-19, Asp-20 and Thr-91, respectively, were replaced with alanine, were shown to interact with PrrBc to the same extent as the wild-type form of the PrrA regulatory domain (PrrAr), indicating that D19A, D20A and T91A mutations of PrrAr do not affect the proteinprotein interactions between PrrAr and PrrBc to any significant degree. In contrast, replacement of Asp-63 or Lys-113 with alanine abolished proteinprotein interactions between PrrAr and PrrBc, as judged by the inability of the corresponding yeast strains (PrrArD63A/PrrBc, PrrArK113A/PrrBc) to grow on histidine-deficient medium.

To determine whether the inability of the yeast strains expressing the mutant forms (D63A and K113A) of PrrAr to grow on histidine-deficient medium was the result of instability of the corresponding GAL4ADPrrAr fusion proteins in the yeast cell, we performed Western blotting analysis with a mAb against GAL4AD. As shown in Fig. 3(C), all GAL4AD fusion proteins with molecular masses of ∼43 kDa were detected by Western blotting analysis, indicating that inability of the yeast strains (PrrArD63A/PrrBc, PrrArK113A/PrrBc) to grow on histidine-deficient medium was not the result of instability or degradation of the fusion proteins in the yeast cell.

The DD of PrrB is the region that interacts with PrrA and confers recognition specificity for interactions between PrrB and PrrA
The transmitter domain (PrrBc) of the PrrB histidine kinase, which catalyses the phosphotransfer reaction from ATP to the PrrA response regulator, is composed of two distinct domains, the DD and the CA domain (Bird et al., 1999; Comolli et al., 2002; Oh et al., 2004).

To examine the contributions of these individual domains to the PrrB interaction between the PrrB transmitter domain and PrrA, proteinprotein interactions between the different truncated PrrBc derivatives and PrrA were analysed in the yeast two-hybrid assay. As shown in Fig. 4(A), the yeast strain (PrrA/PrrBc) expressing both GAL4ADPrrA and GAL4BDPrrBc fusion proteins grew on histidine-deficient medium, whereas the negative-control strain (PrrA/B) harbouring the empty bait vector (B) did not grow on the same growth medium. The PrrBc derivatives (PrrBc320, PrrBc270 and PrrBc267) in which 142, 192 and 195 amino acids were removed from the C terminus of PrrBc, respectively, retained the intact DD, but a part or the whole of the CA domain was deleted in these PrrBc derivatives. The PrrBc derivatives containing the intact DD interacted with PrrA. On the other hand, the PrrBc derivatives (PrrBc264 and PrrBc260) whose DDs were impaired by C-terminal deletion did not interact with PrrA. PrrB DD is a PrrB derivative that consists of 70 amino acids and contains only the DD of PrrB (Fig. 4A, C). The yeast strain (PrrA/PrrB DD) expressing both GAL4ADPrrA and GAL4BDPrrB DD fusion proteins grew on histidine-deficient medium. Taken together, the results suggest that the DD of PrrB is the region in which PrrB interacts with PrrA. Interestingly, the yeast strains expressing the PrrBc derivatives that contain the intact DD and impaired (or deleted) CA domain grew better on histidine-deficient medium supplemented with 3AT than the positive-control yeast strain (PrrA/PrrBc) on the same growth medium, indicating that the PrrBc derivatives lacking the intact CA domain interact with PrrA more strongly than does PrrBc. In good agreement with this observation, β-galactosidase assay revealed that an approximately twofold higher activity of β-galactosidase was detected in cell lysates of the yeast strain (PrrA/PrrBc270) as compared to the yeast strain (PrrA/PrrBc). This result suggests that the CA domain with PrrA exerts a negative effect on the PrrB interaction. Western blotting analysis was performed with a mAb against GAL4BD to detect the PrrBc derivatives fused to GAL4BD in yeast cells (Fig. 4B). All the PrrBc derivatives fused to GAL4BD as well as the GAL4BDPrrBc fusion protein were detected immunologically at appropriate positions in the gel, indicating that the inability of the yeast strains (PrrA/PrrBc264 and PrrA/PrrBc260) to grow on histidine-deficient medium is not the result of instability of the fusion proteins in the yeast cell. The bands with molecular masses of ∼31 kDa probably resulted from non-specific cross-reactions of the antibody.

(39K): Fig. 4. Mapping of the PrrB domain interacting with PrrA. (A) Yeast two-hybrid assay. All PrrB derivatives are schematized to the right. All strains contain the pPLA plasmid encoding PrrA fused with GAL4AD in common. A B indicates the absence of a protein fused to GAL4BD. The numbers to the left side of the growth panel indicate the specific β-galactosidase (β-gal) activities that are normalized by the value for the control strain (PrrA/B). The strains were grown on the SD/Leu/Trp plate and the SD/Leu/Trp/His plates containing different concentrations of 3AT at 30 °C for 3 and 5 days, respectively. The numbers above the schematic diagrams indicate the amino acid boundaries of the PrrB derivatives. (B) Western blotting analysis for the detection of GAL4BD fusion proteins using a GAL4BD mAb. The numbers to the left indicate the molecular mass (kDa). (C) The amino acid sequence of PrrB DD. The DD of the histidine kinase is composed of the two helices α-helix 1 and α-helix 2, linked by a turn.

We next investigated whether the DD of a histidine kinase alone can confer recognition specificity for its cognate response regulator. For the yeast two-hybrid assay, the DNA fragments encoding the DDs of PrrB and DevS were cloned in the bait vector pGBKT7 to construct the GAL4BDPrrB DD and GAL4BDDevS DD fusion proteins, respectively (Fig. 5A, C). The yeast strain (PrrA/PrrB DD) expressing both GAL4ADPrrA and GAL4BDPrrB DD grew on histidine-deficient medium, while the yeast strain (DevR/PrrB DD) expressing the heterologous pair of the fusion proteins was unable to grow on the same medium (Fig. 5B). This result indicated that the DD of PrrB is sufficient in vivo to provide recognition specificity for its cognate response regulator, PrrA. In the case of the DevS histidine kinase, the DD of DevS (DevS DD) interacted with both DevR and PrrA, although interactions of DevS DD with its partner DevR appeared to be significantly stronger than those with PrrA. The control yeast strain (P/DevS DD) harbouring the empty prey vector did not grow on histidine-deficient medium (data not shown). The DevS derivative DevS DD+ATP, consisting of both the DD and CA domain, did not interact with PrrA, but gave interaction signals with its cognate response regulator, DevR. Taken together, the results shown in Fig. 5 suggest that the DD of the histidine kinase serves as the interaction region with its cognate response regulator and is sufficient to make both specific and non-specific interactions with its cognate response regulator and other related proteins. However, the presence of the CA domain masks the non-specific interactions (either directly or indirectly) and allows only for the specific interaction between a kinase and its cognate response regulator.

(37K): Fig. 5. Specific proteinprotein interactions between the DDs of histidine kinases and their cognate response regulators. (A) Schematic diagrams of the DDs of the histidine kinases used for this study. The DD is composed of the two α-helices and a turn connecting them. The numbers indicate the amino acid boundaries of the polypeptides. PrrB DD, the DD of PrrB; DevS DD, the DD of DevS; DevS DD+ATP, the truncated form of DevS (amino acids 368578) lacking its N-terminal region. H in the α-helix 1 represents the conserved histidine residue that is autophosphorylated. (B) Yeast two-hybrid assay. The strains were grown on the SD/Leu/Trp plate and the SD/Leu/Trp/His plates containing different concentrations of 3AT at 30 °C for 3 and 5 days, respectively. Pairs of fusions are indicated in the order X/Y, where X and Y are the proteins fused to GAL4AD and GAL4BD, respectively. (C) Comparison of the amino acid sequences of the DDs of PrrB and DevS. PrrB DD, the DD of R. sphaeroides PrrB; DevS DD, the DD of M. smegmatis DevS. The DDs of PrrB and DevS were inferred from the known 3D structure of EnvZ by multiple alignment analysis of the amino acid sequences.

The first α-helix and turn within the PrrB DD are important for specific interactions between PrrB and PrrA
As demonstrated above, the DD of the histidine kinase to some extent confers recognition specificity for its cognate response regulator. The DD of the histidine kinase has a secondary structure of α-helixturnα-helix (Tomomori et al., 1999). Our next question to be addressed was which portion of the DD plays a significant role in providing recognition specificity for its interaction with its cognate response regulator? To address this question, DNA fragments encoding the chimeric DDs (BS DD I and BS DD II) consisting of the α-helices with two different origins (PrrB and DevS) were generated by recombinational PCR and cloned into the bait vector. The BS DD I polypeptide is composed of the first α-helix (α-helix 1) and turn derived from PrrB and the second α-helix (α-helix 2) from DevS (Fig. 6A). Proteinprotein interactions between BS DD I and PrrA were observed in the yeast two-hybrid assay. In contrast, BS DD I did not interact with DevR (Fig. 6B). Another chimeric polypeptide BS DD II is made up of the α-helix 1 from PrrB and the α-helix 2 and turn from DevS. BS DD II did not interact with either PrrA or DevR. These results suggest that α-helix 1 and the flanking turn structure of the PrrB DD determine the recognition specificity for the PrrA response regulator.

(46K): Fig. 6. Proteinprotein interactions between PrrBDevS chimeric proteins and the response regulators PrrA and DevR. (A) The chimeric proteins of PrrB and DevS are schematized. Each chimeric protein has the DD composed of the two helices from the different origins. The numbers indicate the amino acid boundaries of each histidine kinase. (B) Yeast two-hybridassay. Pairs of fusions are indicated in the same order as described for Fig. 5. The strains were grown on the SD/Leu/Trp plate and the SD/Leu/Trp/His plates containing different concentrations of 3AT at 30 °C for 3 and 6 days, respectively. (C)Western blotting analysis for the detection of GAL4BD fusion proteins using a GAL4BD mAb.

As shown in Fig. 6(C), Western blotting analysis with a GAL4BD-specific antibody revealed that all the chimeric proteins fused to GAL4BD were normally expressed, indicating that the lack of interactions in the yeast two-hybrid assay observed for BS DD II is not due to instability of the chimeric proteins in the yeast two-hybrid system.

The EnvZ protein of E. coli is a histidine kinase that is involved in osmosensing and controls the phosphorylation state of the OmpR response regulator. The amino acid sequence of the EnvZ DD shows a high level of similarity with that of the PrrB DD, especially in the region encompassing α-helix 1 and the turn (Fig. 7A). If the α-helix 1 and turn of the DD are important for specific interactions between the DD and its cognate response regulator, the DD of EnvZ might be expected to interact not only with OmpR but also with PrrA. As shown in Fig. 7(B), the DD of EnvZ indeed interacted with PrrA as well as with its cognate response regulator, OmpR. The control yeast strain (P/EnvZ DD) harbouring the empty prey vector did not grow on histidine-deficient medium (data not shown). β-Galactosidase assays were performed on cell-free lysates of the yeast strains as a measure of the strength of the proteinprotein interactions. The interaction of the EnvZ DD with OmpR was shown to be much stronger than that with PrrA.

(25K): Fig. 7. Proteinprotein interactions between the DD of EnvZ and PrrA. (A) Comparison of the amino acid sequences of the PrrB and EnvZ DDs. Identical or conservatively substituted amino acids are indicated by asterisks or colons, respectively. RsPrrB DD, DD of R. sphaeroides PrrB; EcEnvZ DD, DD of E. coli EnvZ. (B) Interactions between the DDs of EnvZ and PrrA or its cognate response regulator, OmpR, in the yeast two-hybrid assay. The SD/Leu/Trp plate and SD/Leu/Trp/His plates containing 3AT were incubated at 30 °C for 2 and 4 days, respectively. β-Galactosidase (β-gal) activities detected in the strains are shown below the panel.

In vivo yeast two-hybrid assay is an efficient method for detecting a weak and transient proteinprotein interaction that is difficult to investigate by other in vitro methods. Yeast two-hybrid systems have been successfully employed to study the specific interaction between components of the NtrBC two-component system of Klebsiella pneumoniae (Martinez-Argudo et al., 2001, 2002) and to identify the histidine kinases which specifically interact with the DivK response regulator of Caulobacter crescentus (Ohta & Newton, 2003).

While studying proteinprotein interactions between the PrrB histidine kinase and the PrrA response regulator, we found that insertion of a linker of approximately 60 amino acids (including the linker of 32 amino acids provided by the prey vector itself) between GAL4AD and PrrA was required for the successful yeast two-hybrid assay. However, the GAL4ADDevR fusion protein without the additional linker interacted with GAL4BDDevS (data not shown), implying that the necessity of the additional linker between GAL4AD and response regulator depends on the type of response regulator.

Proteinprotein interactions between the GAL4ADPrrA and GAL4BDPrrBc fusion proteins were clearly detected in the spotting assay, while those between GAL4ADPrrBc and GAL4BDPrrA were not (data not shown). This phenomenon has been observed elsewhere for the NtrBC two-component system and explained by instability of the GAL4ADNtrB fusion protein in the yeast cell (Martinez-Argudo et al., 2002).

It was demonstrated in this study that the interaction surfaces of PrrA and PrrB reside in the regulatory domain and the DD, respectively, which is consistent with the fact that the autophosphorylated histidine in the DD must be closely juxtaposed to the conserved aspartate in the regulatory domain for the phosphotransfer reaction to ensue. Yeast two-hybrid studies on the NtrBC system have revealed that the 110 amino acid long NtrB fragment containing the DD specifically interacts with the regulatory domain of NtrC (Martinez-Argudo et al., 2002). In this study we further delineated the PrrB region with which PrrA interacts, and clearly showed that the PrrB DD consisting of only 70 amino acids interacted with PrrA and that impairment of the DD by further deletions abolished the interaction between PrrB and PrrA.

It has been suggested by structural (crystallographic and NMR titration) studies on the Spo0B-Spo0F and EnvZ-OmpR systems that the DDs of histidine kinases other than the Hpt (histidine-containing phosphotransfer)-containing CheA proteins are the sites with which their cognate response regulators interact (Tomomori et al., 1999; Zapf et al., 2000).

D63A and K113A mutations in the regulatory domain of PrrA abolished proteinprotein interactions between PrrA and PrrB, suggesting that Asp-63 and Lys-113 among the active site residues of the PrrA regulatory domain are important not only in the function of PrrA but also for proteinprotein interactions between PrrA and PrrB. Asp-63 of PrrA has been demonstrated elsewhere to be the amino acid that receives the phosphoryl group from PrrB (Comolli et al., 2002). Although there has been no report regarding the role of Lys-113 in PrrA function, it can be assumed from the earlier study on CheY that Lys-113 might be involved in a phosphorylation-induced conformational change in PrrA (Lukat et al., 1991). X-ray crystallographic studies on the D57A mutant form of CheY, which corresponds to the D63A form of PrrA, have revealed that the D57A mutation leads to a repositioning of the Lys-109 that is the counterpart of Lys-113 of PrrA and the -amino group of which is in close proximity to the Asp-57 β-carboxylate group (Sola et al., 2000). From this observation, together with our result, we assume that the precise positioning of Asp-63 and Lys-113 within the active site of PrrA is important in maintaining the correct conformation of PrrA that is required for its interaction with PrrB. The T91A mutation of PrrA did not affect proteinprotein interactions between PrrA and PrrB, which is in good agreement with the earlier finding that the T87A mutant form of CheY is phosphorylated normally by phosphorylated CheA (Appleby & Bourret, 1998), although this mutation renders CheY non-functional. Asp-19 and Asp-20 of PrrA are predicted to be involved in coordination of the catalytically essential Mg²⁺ ion together with Asp-63 (Lukat et al., 1990). Mutations of the corresponding amino acids to alanine or asparagine have been reported to convert the PhoB and OmpR response regulators into non-phosphorylatable forms (Brissette et al., 1991; Zundel et al., 1998). The D19A and D20A mutations of PrrA did not affect the proteinprotein interactions between PrrA and PrrB, indicating that the mutations did not alter PrrA conformation to an extent sufficient to abolish its interaction with PrrB.

We showed here that the recognition specificity of a histidine kinase for its cognate response regulator is conferred primarily by its DD. Using the chimeric DDs it can be inferred that α-helix 1 and its neighbouring turn structure are more important in conferring the recognition specificity on the DD. Although α-helix 1 of the DD shows some sequence conservation around the conserved histidine, the amino acid sequence of α-helix 1 and its neighbouring turn region appears to be sufficiently variable to provide the recognition specificity when the primary structures of different histidine kinases are compared (Grebe & Stock, 1999). An NMR titration experiment has suggested that OmpR interacts with the four-helix bundle of EnvZ most strongly at the lower part that corresponds to the C-terminal region of α-helix 1, turn, and N-terminal region of α-helix 2 (Tomomori et al., 1999). On the basis of the results presented here as well as those reported elsewhere (Tomomori et al., 1999; Martinez-Argudo et al., 2002; Ohta & Newton, 2003), we assume that the lower part of the four-helix bundle within the histidine kinase serves as a docking site for its cognate response regulator and that α-helix 1 and its neighbouring turn of the DD provide to some extent the recognition specificity for its response regulator.

The CA domain of the histidine kinase is known to catalyse the phosphotransfer reaction from ATP to the conserved histidine residue within the DD (Park et al., 1998). As shown in Fig. 5(B), the DevS derivative containing both the DD and CA domain exhibited a higher recognition specificity for DevR than did the DevS derivative consisting of the DD alone, which interacted also with the non-cognate response regulator PrrA. This result clearly reveals that the CA domain of the histidine kinase provides additional recognition specificity to the histidine kinase in case the DD of the histidine kinase alone does not provide sufficient recognition specificity for its interaction with the response regulator. The specificity for proteinprotein interactions between a histidine kinase and its response regulator appears to be provided not only by the DD that serves as an interaction surface, but also by the CA domain that catalyses the phosphotransfer reaction.

Deletion of the CA domain from PrrBc led to an increase in the interaction affinity of PrrBc for PrrA. This observation can be explained by the following: (i) the CA domain provides recognition specificity additional to that of the DD, which reduces the strength of proteinprotein interactions between PrrBc and PrrA, or (ii) the GAL4BDPrrBc fusion protein might phosphorylate the GAL4ADPrrA fusion protein in the yeast cell. Phosphorylation might reduce the affinity of PrrA for PrrB. Earlier, it has been reported that the affinity of the phosphorylated OmpR for EnvZ is weaker than that of the non-phosphorylated OmpR (Mattison & Kenney, 2002).

This work was supported by a Korea Research Foundation grant (KRF-2004-015-C00488) to J.-I. O. and by a USPHS grant (GM15590) to S. K.