INTRODUCTION

Chronic pulmonary infection with the bacterium Pseudomonas aeruginosa is a major factor in the poor prognosis and high mortality rate of patients with cystic fibrosis (CF)¹ (1). Most P. aeruginosa strains isolated from the CF respiratory tract overproduce an exopolysaccharide called alginate, which gives the colonies a mucoid morphology (2). This highly viscous polysaccharide plays a role in the pathogenesis of P. aeruginosa by imparting antiphagocytic properties (3) and an adherence mechanism (4). Most of the genes involved in alginate biosynthesis are in a tightly regulated operon at 34 min on the 75-min chromosome (5). High expression of the alginate biosynthetic genes requires the activation of an alternative sigma factor (²²) encoded by algT (algU) at about 68 min on the chromosome (for review, see Ref. 6). In addition, a cascade of several positive regulators are also required for high expression of alginate genes (7, 8). Two of these, AlgB and AlgR (AlgR1), belong to the superfamily of response regulators of prokaryotic two-component regulatory systems (9, 10).

Two-component regulation is a mechanism for signal transduction to control cellular adaptations in response to environmental or physiological changes (for review, see Ref. 11). Observed in many bacterial species (12, 13), as well as in yeasts (14) and plants (15), two-component systems generally include a histidine protein kinase and a cognate regulator protein. In general, the histidine protein kinase senses a specific environmental stimulus and undergoes autophosphorylation at a histidine residue present in a highly conserved carboxyl-terminal domain of the protein. This phosphate group is subsequently transferred to an aspartate residue in the amino terminus of the response regulator, resulting in a change in the activity of the response regulator that leads to an adaptive response (11, 12). Response regulators can also catalyze kinase-independent phosphorylation and dephosphorylation by low-molecular weight phosphorylated compounds (e.g. acetyl phosphate, carbamyl phosphate, etc.), which may serve to integrate environmental control with the physiological status of the cell (16).

Alginate overproduction by P. aeruginosa is generally seen in strains causing pulmonary infection of CF patients. Specific signals present in the environment of the CF lung (e.g. dehydration, high osmolarity, limiting nutrients, antibiotics) may play a role in stimulating alginate production (for review, see Ref. 17). However, the role or requirement for any particular in vivo signal in the expression of alginate genes has not been well established. The discovery of two-component response regulators (i.e. AlgB and AlgR) suggests that environmental signals may play a role in the regulation of alginate production. Moreover, inhibitors of the two-component regulatory pathway inhibit the expression of alginate biosynthetic genes (18). However, proteins in P. aeruginosa with sensor kinase activity that can phosphorylate AlgB or AlgR have not been demonstrated. A gene adjacent to algR was recently identified that encodes a protein (FimS, AlgZ) with homology to an atypical two-component sensor (19, 20), but whether it functions as a kinase of AlgR is unknown. In this study we identified a gene called kinB, located immediately downstream of algB, that encodes a protein with high similarity to typical histidine protein kinases of two component systems. Our data indicate that KinB is an inner membrane protein with histidine protein kinase activity that is capable of promoting autophosphorylation and rapid transfer of the phosphate to AlgB.

EXPERIMENTAL PROCEDURES

The P. aeruginosa strains utilized in this study were FRD1, an alginate-overproducing (Alg⁺) CF isolate and its derivative FRD444 (Alg, algB::Tn501), which contains a mercury resistance (Hg^r) transposon marker in algB (21). Escherichia coli strains HB101 and JM109 were used in routine cloning manipulations (22); BL21(DE3) was used to express His₆-tagged KinB; XL-2 Blue was used to overexpress AlgB. L broth (10.0 g of tryptone (Difco), 5.0 g of yeast extract (Difco), 5.0 g of NaCl/liter, pH 7.5) was used for the routine culture of P. aeruginosa and E. coli. A 1:1 mixture of Pseudomonas isolation agar (Difco) and L agar was used to select for P. aeruginosa following triparental matings. Selective antibiotics used for P. aeruginosa were carbenicillin at 300 µg/ml and tetracycline at 100 µg/ml; selective antibiotics used for E. coli were ampicillin at 100 µg/ml, kanamycin at 35 µg/ml, and tetracycline at 15 µg/ml. HgCl₂ was used at 18 µg/ml both for P. aeruginosa and E. coli.

Cloned DNA fragments utilized in this study are shown in Fig. 1. Most routine genetic manipulations were performed as described elsewhere (22). Plasmid DNA was isolated from E. coli using Qiagen columns and procedures (Qiagen Corp.). Genomic DNA of P. aeruginosa was prepared using a protocol previously described (21). Restriction endonucleases were purchased from Boehringer Mannheim and New England Biolabs. To isolate DNA that included sequences located downstream of algB, chromosomal DNA from FRD444 (algB::Tn501) was digested with BamHI (where Tn501 is not cut by BamHI), ligated into cosmid vector pEMR2 (23), packaged in vitro into  particles (Gigapack II cloning kit, Stratagene), and transduced into HB101. One representative clone (pDJWA10, Fig. 1) that conferred Hg^r contained approximately 15-kb DNA upstream and 10-kb DNA downstream of algB. Plasmid pDJW130 (Fig. 1) had a 0.8-kb XhoI-EcoRI fragment from pJG12 (24) cloned into vector pKS() that was used as a hybridization probe; it was digoxigenin-labeled by the polymerase chain reaction using T3 and T7 primers. This probe was used to identify a 4-kb ClaI-HindIII fragment from pDJWA10 which contained a portion of algB and the entire kinB gene (below), which was then cloned into pUC19 to form pSM67 (Fig. 1).

To prepare the DNA downstream of algB for sequence analysis, the 4-kb ClaI-HindIII fragment of pSM67 was digested with PstI or partially digested with Sau3AI, and the resulting fragments were subcloned into the PstI or the BamHI site of M13mp19 (New England Biolabs), respectively. Single-stranded DNA templates were prepared from these M13mp19 clones using a sample preparation protocol (Applied Biosystems). DNA sequencing reactions were performed with a Taq Dyedeoxy terminator cycler sequencing kit (Applied Biosystems) using a Perkin-Elmer DNA thermal cycler and run on an Applied Biosystems 373A DNA sequencer. DNA fragments were sequenced on both strands, and the sequence data obtained were aligned using SeqMan software (DNASTAR) on an Apple Macintosh computer. To verify alignment of the sequence contigs, six additional sequences were obtained by manual sequencing of pSM67 (Fig. 1) using T7 DNA polymerase version 2, the 7-deaza-dGTP sequencing kit (Life Sciences), and synthesized oligonucleotide primers: p50 (5 CGGCTGTCCTTCTCCAGGTC 3), p51 (5 CCACTACACCTCCACCGATC 3), p52 (5 CAAGCGCACGGTATCACC 3), p53 (5 GCATATCGACGCTGAGCATG 3), p54 (5 CGGTGGTGCTGGCCTGG 3), and p55 (5 CGCCATTGTCTTCCACCGC 3). Homology searches and alignments were performed with the Basic Local Alignment Search Tool (BLAST) Network Service at the National Center for Biotechnology Information, National Institutes of Health (25).

To construct a kinB-lacZ protein fusion, a 2.6-kb EcoRI fragment containing algB-kinB was cloned into pMLB1034 (26), resulting in pSM78; this was followed by the introduction of a mob site on a EcoRI fragment (27) to form pSM82 (Fig. 1). An algB-lacZ protein fusion containing the amino-terminal 379 amino acids of AlgB was constructed by cloning a 3.4-kb SmaI-DraI fragment of pMLB1034 containing lacZ into the EcoRV site of algB in pJG221 (21) to form pSM33; this was followed by the introduction of a mob site on a HindIII fragment (27) to form pSM35 (Fig. 1). Plasmids were later moved into P. aeruginosa FRD1 by triparental mating as described previously (24), which resulted in their integration into the chromosome by homologous recombination. Expression of LacZ fusion proteins in P. aeruginosa was evident by their -galactosidase activity as detected by the formation of blue colonies on L agar plates containing 5-bromo-4-chloro-3-indolyl--D-galactoside at 75 µg/ml. For immunoblot analyses of LacZ protein fusions, overnight cultures of the P. aeruginosa strains carrying lacZ fusions (FRD1::pSM82 and FRD1::pSM35) were diluted 1:50 in 100 ml of L broth with antibiotics and agitated at 37 °C to A₆₀₀ 0.7. Cells were resuspended in 10 ml of A buffer (10 mM Tris-HCl, pH 7.4, 1 mM EDTA, 1 mM EGTA, 0.2 mM dithiothreitol), passed twice through a French press (15,000 p.s.i.), and centrifuged at 10,000 × g for 10 min at 4 °C to remove unbroken cells. The supernatant obtained was used as the whole cell extract. A sample was centrifuged at 200,000 × g for 60 min at 4 °C, and the supernatant was regarded as the fraction enriched for cytoplasmic proteins; the pellet was resuspended in 1.0 ml of A buffer and regarded as the fraction enriched for membrane proteins. After determining the protein concentration by the Bradford method (28), samples were diluted in sodium dodecyl sulfate (SDS) sample buffer (60 mM Tris hydrochloride, pH 6.8, 2% SDS, 10% glycerol, 0.1 mg of bromphenol blue/ml, 5% 2-mercaptoethanol), and 30 µg of each fraction were subjected to electrophoresis on a SDS-8% polyacrylamide gel. Proteins were electrotransferred to nitrocellulose membrane, and LacZ fusion proteins were detected with rabbit anti--galactosidase polyclonal antibody (5 Prime  3 Prime, Inc., Boulder, CO; at a 1:5,000 dilution) as the primary antibody, and goat anti-rabbit horseradish peroxidase conjugate (Sigma; 1:30,000 dilution) was used as the secondary antibody. Protein bands were visualized with chemiluminescent Western blot detection reagents ECL (Amersham Corp.) and visualized on film (Kodak X-Omat AR) exposed for 2 min.

Constructions containing kinB-phoA translational fusions were based on the plasmid pSM111 (Fig. 1). Polymerase chain reaction amplification was used to generate DNA fragments starting at the MluI site in algB to sites in kinB terminating at codons for Asp-148 or Ile-211, at which the primers generated BamHI sites. These MluI-BamHI fragments were each joined to a 2.6-kb BamHI-XbaI fragment containing phoA from pPHO7 (29), and cloned into pSM111, replacing the existing MluI-XbaI fragment, to form pSM126 and pSM127, respectively. A KinB-PhoA fusion with a junction at residue F379 was constructed by using a linker to join a MluI-EcoRI restriction fragment containing "algB-kinB" to a 2.6-kb SmaI-XbaI fragment containing phoA from pPHO7; this was cloned into pSM111, replacing the existing MluI-XbaI fragment, to form pSM128 (Fig. 1). Protein fusions containing PhoA (alkaline phosphatase) were verified by Western blot analysis using rabbit anti-alkaline phosphatase (Sigma). Colonies containing PhoA fusions with alkaline phosphatase activity (i.e. localized to the periplasm) were screened for blue color on L agar containing 5-bromo-4-chloro-3-indolyl phosphate at 40 µg/ml.

To construct plasmids that overexpressed a His₆-tagged carboxyl-terminal (Gly-198 to Val-595) fragment of KinB (HC-KinB), a 1.6-kb AscI fragment of pSM67 (Fig. 1) was cloned into pNEB193 (New England Biolabs) to form pSM93; this was subsequently digested with SacI and HindIII, and the 1.6-kb fragment containing kinB was cloned in pET28.b (Novagen), resulting in pSM95 (Fig. 1). E. coli BL21(DE3) harboring pSM95 was agitated overnight at 37 °C in 100 ml of L broth with kanamycin. The cells were harvested by centrifugation and resuspended in 100 ml of L broth with kanamycin and 1 mM isopropyl -D-thiogalactopyranoside for induction of the tac promoter (Ptac). After incubation at 30 °C with aeration for 2 h, the cells were harvested and incubated in 20 ml of 20 mM Tris-HCl, pH 8.0, containing lysozyme (100 µg/ml) for 30 min at 4 °C. Following sonication, the lysate was centrifuged at 12,000 × g for 15 min, and the supernatant was filtered (0.45-µm disc filter, Millipore). HC-KinB in the cell extract was purified on a 2.5-ml His-Bind nickel column (Novagen) according to manufacturer's protocol. As estimated by SDS-PAGE and Coomassie Blue staining, HC-KinB was over 95% pure. The His₆ tag was removed from HC-KinB by digestion with 25 units/ml thrombin (Novagen) for 2 h at 22 °C to form C-KinB, which was subjected to an amino-terminal sequence analysis (Biotechnology Center, St. Jude Children's Research Hospital, Memphis, TN).

To obtain C-KinB proteins with substitutions in conserved residues, the SacI-HindIII fragment in pSM95 (Fig. 1) was cloned into pAlter1 (Promega) and mutagenized using the altered sites mutagenesis system (Promega) according to the procedure suggested by the manufacturer. The mutagenic primers used were: E257Q (5 TGCTTTCCGGCCAGCGGCGCCTG 3), H385K (5 CTGCGCGCCTCCAAGGAACTGCGCACG 3), H385Q (5 CGCGCCTCCCAGGAACTGCG CACG 3), N504Q (5 CAACCTGCTGGAACAGGCCCTGCGCCATA 3), D532E (5 CGGTGGAAGAAAATGGCGAAG 3), D532N (5 CGCGGTGGAAAACAATGGCGA 3), and G560A (5 GGCGGCGCCGCTCTCGGCCTG 3). After confirming the mutations by DNA sequence analysis, the SacI-HindIII fragment of each kinB allele was cloned into pET28.b, and mutant C-KinB proteins were expressed and purified as described above.

E. coli XL21 Blue harboring pDJW52 (Fig. 1) expresses algB under the control of Ptac as described previously (9). Cells from 400 ml of overnight culture of XL21 Blue(pDJW52) were resuspended and agitated in 400 ml of fresh L broth with ampicillin and 1 mM isopropyl -D-thiogalactopyranoside at 30 °C for 3 h. Cells were harvested and then incubated for 30 min at 4 °C in 20 ml of 20 mM Tris-HCl, pH 8.0, containing lysozyme (100 µg/ml). Following sonication, the lysate was centrifuged at 12,000 × g for 10 min, and the supernatant was centrifuged at 200,000 × g for 60 min. Proteins in the clear supernatant were precipitated with 35% ammonium sulfate (J. T. Baker Inc.), and the precipitant was resuspended and dialyzed against 15 mM BisTris propane, pH 7.0, 20 mM NaCl. A sample (20 ml containing 8.4 mg of protein) was loaded onto an AP-2 column (Waters) packed with Protein-PAK DEAE 40HR anion exchange matrix (Waters), and a linear 20-160 mM NaCl gradient in 15 mM BisTris propane, pH 7.0, was used to elute proteins from the column. AlgB eluted at 130 mM NaCl and was estimated by SDS-PAGE and Coomassie Blue staining to be >90% pure.

Autophosphorylation of C-KinB was performed at 22 °C in P buffer (50 mM Tris-HCl, pH 7.5, 50 mM KCl, 5 mM MgCl₂). C-KinB was diluted to a final concentration of 2.5 µM and distributed in 9-µl aliquots for each reaction. Each reaction was started by adding [-³²P]ATP (30 Ci/mmol, Amersham) to a final concentration of 33.3 µM and was stopped by the addition of 3 µl of 5 × SDS sample buffer. Unincorporated label was removed by passage through a 1-ml Sephadex G-25 (Pharmacia Biotech Inc.) column, and samples were electrophoresed on a SDS-10% polyacrylamide gel and examined by autoradiography. To examine the time course of C-KinB autophosphorylation, phosphorylation reactions were stopped by adding 10 µl of 200 mM sodium acetate, pH 4.0, and immediately spotting the mixture onto a phosphocellulose membrane (Beckman) pre-equilibrated with 25 mM sodium acetate, pH 4.0. The membranes were washed three times for 10 min each in 800 ml of buffer containing 25 mM sodium acetate, pH 4.0, and the radioactivity on the dried membranes was measured (TRI-CARB 2000 liquid scintillation analyzer). In studies demonstrating the transfer of phosphoryl label from C-KinB to AlgB, 13 pmol of KinB was phosphorylated for 60 min at 22 °C in a 10-µl mixture under the conditions described above. AlgB (40 pmol) was added to the mixture, and the reaction was terminated after 90 s by adding 3 µl of 5 × SDS sample buffer. The samples were passed through a 1-ml Sephadex G-25 column to remove unincorporated label and analyzed by SDS-10% PAGE, followed by autoradiography.

The nucleotide sequence data and inferred amino acid sequence reported here for kinB have been deposited in the GenBank^TM data base under accession number U97063.

RESULTS

We examined whether a gene (kinB) encoding a sensor kinase was closely linked to a known gene (algB) encoding a response regulator that controls the alginate biosynthetic operon in P. aeruginosa. Several studies on bacterial two-component regulatory systems have shown that genes encoding a response regulator and its cognate histidine protein kinase are often linked (12). A 25-kb BamHI fragment containing the DNA flanking algB (pDJWA10, Fig. 1) was obtained from genomic DNA of P. aeruginosa FRD444, a strain with an algB::Tn501 allele (21) that provided a selectable marker (mercury resistance) for the DNA in this region. A 4-kb ClaI-HindIII fragment was then subcloned from the region immediately downstream of algB (pSM67, Fig. 1). This was subjected to a sequence analysis, and the putative kinB ORF of 1,788 bp was observed in the same direction of transcription as algB (Fig. 2). The kinB ORF had a translation initiation codon that overlapped with the algB termination codon, suggesting that expression of algB and kinB may be translationally coupled. The kinB ORF predicted a polypeptide of 595 amino acids with a molecular weight of 66,078. Two hydrophobic domains at the amino terminus of KinB were observed (underlined in Fig. 2). An 11-base pair inverted repeat sequence was located 75 bp downstream of the kinB ORF that may serve as a factor-independent terminator (shown as hatched lines in Fig. 2).

A homology search showed that the KinB sequence was similar to a number of histidine protein kinases in two-component regulatory systems. Fig. 3 depicts an alignment of KinB sequences with that of PhoR, a similar sized histidine protein kinase in Bacillus subtilis (30). Overall, KinB shows 31% identity and 59% similarity with PhoR. The most conserved sequences were in four regions that are characteristic of histidine protein kinases (marked with hatched boxes in Fig. 3). The H box is proposed to be the phosphorylation domain and may also be involved in the dimerization of the kinase monomers; the N, D/F, and G boxes are proposed to form a nucleotide binding surface in the tertiary structure within the active site (31). The residues in these boxes that are believed to be critical (marked with triangles in Fig. 3) were all conserved in KinB and PhoR. In addition, both KinB and PhoR contained two hydrophobic domains that were similarly positioned in their amino termini (underlined and overlined in Fig. 3). Both hydrophobic regions of KinB were sufficient in length to form transmembrane domains, suggesting that KinB may be localized to the inner membrane, as is PhoR.

A KinB-LacZ fusion protein (encoded by pSM82, Fig. 1) was constructed to test the expression of the kinB ORF in P. aeruginosa. The KinB-LacZ was predicted to be a 157.4-kDa peptide, since the amino-terminal 379 amino acids of KinB (41.6-kDa) was fused to a lacZ derivative expressing all but the first eight amino acids of LacZ (115.8 kDa). As a control, an AlgB-LacZ fusion protein of 151.4 kDa was constructed (encoded by pSM35, Fig. 1). The plasmids containing the kinB-lacZ and algB-lacZ fusion genes were in suicide vectors, and their mobilization to P. aeruginosa FRD1 resulted in chromosomal integration at the site of DNA homology (Fig. 4A). FRD1::pSM82 and FRD1::pSM35, harboring the respective kinB-lacZ and algB-lacZ fusions in single copy, both showed -galactosidase activity, indicating that each ORF expressed a protein in P. aeruginosa. The kinB-lacZ and algB-lacZ encoded fusion proteins were also analyzed in a Western blot analysis of whole cell extracts, using polyclonal antibody specific for LacZ (Fig. 4B, lanes 1 and 4); this showed that their electrophoretic mobilities were consistent with the sizes predicted. The KinB-LacZ fusion produced in FRD1::pSM82 contained a large amino-terminal fragment of KinB that included both putative transmembrane domains. To test whether the KinB-LacZ hybrid localized to the membrane, whole cell extracts of FRD1::pSM82 were used to obtain fractions enriched for either cytoplasmic or membrane proteins. Extracts containing the AlgB-LacZ (FRD1::pSM35) were processed in parallel. Using anti-LacZ in the Western blot analysis, KinB-LacZ was detected in the membrane fraction, but not in the cytoplasmic fraction (Fig. 4B, lane 2), suggesting that KinB was indeed associated with the membrane. In contrast, the AlgB-LacZ fusion protein, which does not contain a potential transmembrane domain (9), was detected in the cytoplasmic fraction (Fig. 4B, lane 6), but not in the membrane fraction.

The two hydrophobic domains in the amino terminus (residues 13-39 and 170-190) of KinB, which may serve as transmembrane domains, were evident in the hydrophilicity plot (Fig. 5A). Thus, the region between the two putative transmembrane domains (residues 40-169) of KinB was predicted to be in the periplasmic space. To test this, KinB-PhoA fusions were constructed with junctions at residue Asp-148, Ile-211 and Phe-379 (Fig. 5B). All three KinB-PhoA fusions expressed proteins of the predicted size that were readily detected in whole cell extracts of E. coli using a Western blot analysis and antibody specific for PhoA (Fig. 5C). In that the phoA gene encodes the periplasmic enzyme alkaline phosphatase, such protein fusions are enzymatically active (PhoA⁺) only if translocated to the periplasm (32). The KinB(D148)-PhoA fusion retained the first transmembrane domain and was PhoA⁺ in E. coli (Fig. 4B), suggesting that the amino terminus of KinB between the two transmembrane domains was periplasmic. In contrast, bacteria expressing KinB(I311)-PhoA and KinB(F379)-PhoA, where fusions were downstream of the two transmembrane domains, were not enzymatically active for PhoA (Fig. 5B). This suggests that the COOH terminus of KinB was localized to the cytoplasm. Thus, the KinB amino terminus appears to be the periplasmic sensor domain and the C terminus contains the cytoplasmic histidine kinase domain.

The localization of KinB to the membrane complicated the purification of native protein for studies of its potential histidine protein kinase activity. However, all of the conserved sequences for kinase activity were present in the cytoplasmic carboxyl terminus. Thus, we tested the possibility that a carboxyl-terminal fragment of KinB may be enzymatically active, as is the case for several other membrane associated sensor kinases (33-36). DNA encoding a carboxyl-terminal fragment of KinB from Gly-198 to the end (Val-595) was cloned in frame into the His tag vector pET28.b to form pSM95 (Fig. 1). This plasmid expressed a His₆-tagged fusion protein (HC-KinB), which was purified using a nickel sulfate affinity column. To remove the His₆ sequence, the purified fusion protein was digested with thrombin, which recognizes a site between the His₆ tag and C-KinB sequence. However, an amino-terminal sequence analysis of C-KinB revealed that thrombin (which has arginine as its preferred site) also cleaved the KinB protein between residues Arg-243 and Gln-244 to generate a 39.4-kDa C-KinB polypeptide. Nevertheless, this C-KinB fragment still retained all the sequences predicted to function as a histidine protein kinase (see Fig. 3). To test this, C-KinB was incubated with ³²P-labeled ATP, and then the samples were subjected to SDS-PAGE and autoradiography. C-KinB (25 pmol) incubated with [-³²P]ATP (33 µM) showed progressive autolabeling over the 1-60-min period examined (Fig. 6A, lanes 1-6). Incubation with [-³²P]ATP at 33 or 66 µM for 40 min showed similar labeling of C-KinB (compare lanes 5 and 7), suggesting that ATP was not a limiting factor in these reactions. Accordingly, incubation with 50 pmol of C-KinB for 40 min did show increased labeling (Fig. 6A, compare lanes 5 and 8). As a control, C-KinB (25 pmol) was incubated with [-³²P]ATP (15 µM) for 60 min (Fig. 6A, lane 9), and no labeling was observed; this ruled out the possibility of nonspecific binding of ATP by C-KinB. The autoradiogram showing autolabeling of C-KinB suggested that the level of protein phosphorylation (i.e. the balance of autophosphosphorylation and dephosphorylation) was not maximum by 60 min. Thus, a quantitative time course of C-KinB autophosphorylation was performed using liquid scintillation (Fig. 6B). This showed that the maximum level of phosphorylated C-KinB under these conditions did not reach a plateau until approximately 5 h of incubation. One possible reason for this overall slow reaction was a high rate of C-KinB dephosphorylation. However, this appeared not to be the case because the phosphoryl label on C-KinB was stable after incubation with a chase of cold ATP (333 µM) for 30 min (Fig. 6C).

We tested whether autophosphorylation activity required sequences in KinB that are homologous to those of other sensor kinases. Critical residues in histidine protein kinases that were conserved in C-KinB (described above, see Fig. 3) were altered by site-directed mutagenesis of kinB. Mutant alleles of kinB were generated that expressed the following mutant HC-KinB proteins: H385K and H385Q, in which His-385 in the H box (i.e. the predicted site of phosphorylation) was changed to Lys and Gln, respectively; N504Q, where Asn-504 in the N box was mutated to Gln; D532N and D532E, in which Asp-532 of the D/F box was changed to Asn or Gln, respectively; and G560A where Gly-560 in the G box was substituted for Ala. Mutant derivatives of HC-KinB were purified in the same manner as wild-type HC-KinB and estimated to be >95% pure by SDS-PAGE. The His₆ tags on these proteins were also removed by thrombin digestion. Equivalent amounts of wild-type and mutant C-KinB derivatives, after treatment with thrombin, were examined by SDS-PAGE for relative stability of the proteins (Fig. 7A). Only C-KinB D532E (Fig. 7A, lane 7) showed any evidence of degradation beyond removal of the His₆-Arg-243 peptide (despite 27 other Arg residues, the preferred site of thrombin cleavage). When each protein (2.5 µM) was incubated with [-³²P]ATP (33 µM), the wild-type C-KinB sequence showed strong autophosphorylation activity (Fig. 7B, lane 1). However, labeling of the mutant proteins was undetectable except for the C-KinB D532N derivative in which a trace amount of phosphorylated protein was detected (Fig. 7B, lane 5). The C-KinB E257Q protein had a substitution at a nonconserved residue, and it showed autophosphorylation that was comparable with that of wild-type (Fig. 7B, lanes 8).

To determine whether AlgB-KinB may function as a two-component regulatory pair, the ability of phosphorylated C-KinB to donate a phosphate group to AlgB was examined. AlgB was overexpressed in E. coli and purified (>90%) using standard chromatographic procedures (Fig. 8A, lane 4). Purified AlgB alone was not autophosphorylated when it was incubated with [-³²P]ATP as determined by SDS-PAGE and autoradiography (Fig. 8B, lane 1). As shown above, purified C-KinB (1.3 µM) incubated with [-³²P]ATP for 60 min showed autophosphorylation (Fig. 8B, lane 2). However, when AlgB (40 pmol) was incubated for 90 s with autophosphorylated C-KinB (K*), AlgB became radiolabeled, and complete dephosphorylation of C-KinB was also observed (Fig. 8B, lane 4). Other studies of response regulators (e.g. CheY) indicate that Mg²⁺ is required for phosphorylation (37). This also appears to be the case with AlgB, since no AlgB phosphorylation was observed when the protein was preincubated with EDTA to chelate divalent cations (Fig. 8B, lane 3). In other experiments, maximum phosphotransfer from ³²P-C-KinB to AlgB was observed after only 20-40 s of incubation (data not shown). Taken together, the above results show that KinB in P. aeruginosa is a member of the sensor kinase superfamily with histidine kinase activity that can rapidly phosphorylate its cognate response regulator, AlgB.

DISCUSSION

The genes involved in alginate biosynthesis are under complex control by a cascade of regulators (6, 8). Two positive regulators of alginate production, AlgB and AlgR, affect transcriptional activation of the alginate biosynthetic operon at algD, and both have sequence similarity to the family of response regulators of two-component systems (9, 10). This suggested that the production of alginate by P. aeruginosa is influenced by environmental factors, some of which may be found in the unique environment of the CF lung (17). Prior to the recent description of FimS and its association with AlgR (19), no putative cognate sensor for AlgR had been recognized. However, FimS (also known as AlgZ) does not possess sequence similarity to typical histidine protein kinases (19, 20). The goal of this study was to identify KinB, a cognate sensor for AlgB, followed by tests for their potential interaction via phosphorylation. In that genes encoding histidine protein kinases are often closely linked to genes for their cognate response regulators (12), we examined the DNA immediately downstream of algB, and as a result kinB was discovered. KinB had a predicted molecular mass of 66 kDa and showed sequence similarity to many histidine protein kinases of two-component regulatory systems. KinB had all four conserved "boxes" characteristic of histidine protein kinases. Like many of them, KinB also had two hydrophobic domains at the amino terminus that are of sufficient length and hydrophobicity to span the inner membrane. These observations led to an analysis of a KinB-LacZ fusion protein in P. aeruginosa that suggested that KinB was indeed a membrane protein. An analysis of KinB-PhoA fusions supported the predicted membrane topology of KinB that the region between the two hydrophobic domains was in the periplasm. The COOH terminus of KinB, which contained amino acid residues conserved with other sensor kinases, was apparently localized to the cytoplasm. During appropriate in vivo conditions, the amino-terminal domain may act as an environmental sensor of some unknown factor(s) and transduce that information to the cytoplasmic domain to affect its kinase activity. It is difficult to speculate at this time just what environmental signal(s) might activate KinB, as its periplasmic domain has no significant similarity with any other known protein.

Most sensor kinases studied are capable of undergoing autophosphorylation at a conserved histidine residue in the H domain of the protein (38). Purified C-KinB was shown in this study to undergo progressive autophosphorylation when incubated with [-³²P]ATP. Interestingly, the level of autophosphorylated protein did not reach its maximum until about 5 h at room temperature in the presence of excess [-³²P]ATP. This rate is quite slow when compared with the autophosphorylation of other sensor proteins under similar conditions. These sensors include derivatives of ArcB (35) and EnvZ (39) that were deleted of their amino-terminal transmembrane domains, and they have been shown to reach maximum autophosphorylation within minutes. Since the phosphorylated form of C-KinB appeared quite stable, a high intrinsic phosphatase activity is not likely, and an explanation for the atypically slow autophosphorylation of C-KinB is not currently available. However, it is possible that the deletion of the amino terminus affected its autophosphorylation activity, even though C-KinB contained the entire kinase domain. The oligomeric state of many sensor kinases is important for their autophosphorylation activity (40-42). The periplasmic domain of some kinases facilitates dimerization when it is bound by environmental stimulatory ligands (43, 44). The rapid autophosphorylation seen in amino-truncated ArcB and EnvZ may be due to strong protein-protein interactions that remain between the monomers, which is suggested by the observed aggregation and precipitation of truncated ArcB and EnvZ with the membrane fraction when overexpressed in E. coli (35, 39). In contrast, when C-KinB was overexpressed, it remained soluble. It is currently not clear whether the native form of KinB forms a dimer or whether dimerization affects KinB autophosphorylation activity. Another explanation for the observed kinetics of C-KinB phosphorylation also relates to the soluble nature of C-KinB. When "tethered" to a membrane, as is the case for native KinB, the effective concentration of KinB may be higher than that observed with the soluble C-KinB used in these studies. In addition, the reaction condition for the C-KinB autophosphorylation assay used here may not be optimal for this protein, although similar conditions were used in the phosphorylation of truncated ArcB and EnvZ (35, 39).

Since the sequence of KinB showed high homology with other sensor kinases, substitutions of the conserved residues were made to verify that KinB is a new member of this conserved superfamily of histidine protein kinases. When the predicted histidine phosphorylation site in KinB (His-385 in the H box) was changed to either a lysine or a glutamine, autophosphorylation of C-KinB was completely lost. Moreover, mutations affecting other conserved boxes all had deleterious effects on the kinase activity, suggesting that KinB is a typical histidine protein kinase. Interestingly, while no phosphorylated protein was detected when Asp-532 in the D/F box was substituted for a glutamate, changing the same residue to an asparagine permitted some residual C-KinB autophosphorylation.

The ability of phospho-C-KinB to phosphorylate the purified response regulator AlgB was also demonstrated. When AlgB was incubated with the phosphorylated C-KinB at a molar ratio of 3 to 1, the phosphoryl group was rapidly transferred to AlgB and completed by 40 s. This rate is similar to that observed between other sensor-regulator pairs (37, 45). Also, similar to the phosphorylation of other response regulators (37, 45), AlgB phosphorylation was inhibited by EDTA, suggesting the requirement of Mg²⁺ in the phosphorylation reaction. Magnesium has been shown to bind at an aspartate-rich acid pocket within the active site of the response regulator phosphorylation domain. Binding of Mg²⁺ causes conformation changes in the response regulator, and this likely facilitates the phosphotransfer reaction between histidine protein kinases and response regulators (46-49). Previous studies with the alginate response regulator AlgR demonstrated that AlgR was capable of being phosphorylated by the well characterized histidine protein kinase CheA and by small phospho-donor molecules (50). Despite numerous attempts, AlgB could not be phosphorylated by CheA (data not shown). This suggests that phosphorylation of AlgB by C-KinB has a relatively high specificity. The possibility of AlgR phosphorylation by C-KinB, as well as the involvement of small phospho-donor molecules in AlgB phosphorylation, are currently being examined.

At least three other sensor kinase-regulator pairs have been reported in P. aeruginosa, but this is the first case that in vitro phosphorylation of the sensor and the regulator has been demonstrated in this organism. Besides AlgB-KinB, there are two other typical two-component regulatory systems: PilS-PilR are involved in the regulation of expression of type IV fimbriae (49), and PfeS-PfeR control the expression of the ferric enterobactin receptor, PfeA (51). The genes for the histidine protein kinase and the response regulator in each of these two systems are also next to each other (49, 51). The organization of pfeR-pfeS is strikingly similar to that of algB-kinB, in that the start codon for pfeS also overlaps the stop codon for pfeR (51). The three kinases, PilS, PfeS, and KinB, all have conserved residues characteristic of histidine protein kinases, but little homology beyond that. It appears likely that KinB responds to signals different from that of PilS and PfeS. Recently, another sensor-kinase pair, FimS-AlgR, has been suggested to belong to a new family of transmitter-receiver response regulators (19, 20). However, in that the predicted FimS (AlgZ) sequence lacks a conserved H box, it has been postulated that FimS may not undergo autophosphorylation, although it may still be able to transfer a phosphate group to AlgR (19). It will be of interest to determine to what extent the roles of algB-kinB system and fimS-algR system overlap in control of the virulence factors in this opportunistic pathogen.