Histidine-containing Phosphotransfer Protein-B (HptB) Regulates Swarming Motility through Partner-switching System in Pseudomonas aeruginosa PAO1 Strain

Background: This study investigates how histidine phosphotransfer protein-B (HptB) regulates Pseudomonas aeruginosa swarming. Results: HptB regulates the protein phosphatase activity of PA3346, which in turn controls the flagellar gene expression through interaction with PA3347. Conclusion: Our results reveal a partner-switching mechanism regulating the σ28-dependent motility genes. Significance: The interplay between a two-component system and σ28 has been established. The histidine-containing phosphotransfer protein-B (HptB; PA3345) is an intermediate protein involved in transferring a phosphoryl group from multiple sensor kinases to the response regulator PA3346 in Pseudomonas aeruginosa PAO1. The objective of this study was to elucidate the biological significance of the HptB-PA3346 interaction and the regulatory mechanisms thereafter. The transcription profiling analysis of an hptB knock-out mutant showed that the expression of a number of motility-related genes was altered consistent with the non-swarming phenotype observed for the mutant. Domain analysis indicated that the PA3346 C-terminal region (PA3346C) exhibits ∼30% identity with the anti-σ factor SpoIIAB of Bacillus subtilis. The presence of Ser/Thr protein kinase activity targeting an anti-σ antagonist, PA3347, at Ser-56 was confirmed in PA3346C using an in vitro phosphorelay assay. Furthermore, PA3346C and the anti-σ28 factor FlgM were found to interact with PA3347 individually both in vivo and in vitro. FlgM displaced PA3346C in binding of PA3347 and was then competitively displaced by σ28 from the PA3347-FlgM complex, forming a phosphorylation-dependent partner-switching system. The significance of PA3347 phosphorylation in linking the partner-switching system and swarming motility was established by analyzing the swarming phenotype of the PA3347 knock-out mutant and its complement strains.

The histidine-containing phosphotransfer protein-B (HptB; PA3345) is an intermediate protein involved in transferring a phosphoryl group from multiple sensor kinases to the response regulator PA3346 in Pseudomonas aeruginosa PAO1. The objective of this study was to elucidate the biological significance of the HptB-PA3346 interaction and the regulatory mechanisms thereafter. The transcription profiling analysis of an hptB knock-out mutant showed that the expression of a number of motility-related genes was altered consistent with the nonswarming phenotype observed for the mutant. Domain analysis indicated that the PA3346 C-terminal region (PA3346C) exhibits ϳ30% identity with the anti-factor SpoIIAB of Bacillus subtilis. The presence of Ser/Thr protein kinase activity targeting an anti-antagonist, PA3347, at Ser-56 was confirmed in PA3346C using an in vitro phosphorelay assay. Furthermore, PA3346C and the anti- 28 factor FlgM were found to interact with PA3347 individually both in vivo and in vitro. FlgM displaced PA3346C in binding of PA3347 and was then competitively displaced by 28 from the PA3347-FlgM complex, forming a phosphorylation-dependent partner-switching system. The significance of PA3347 phosphorylation in linking the partner-switching system and swarming motility was established by analyzing the swarming phenotype of the PA3347 knock-out mutant and its complement strains.
Two-component signal transduction systems are commonly utilized by bacteria to sense and respond to environmental alterations. A two-component system typically comprises a sensor histidine kinase and a response regulator. After receiving stimuli, the sensors undergo autophosphorylation at a conserved histidine residue before transferring the phosphate group to a response regulator either directly or through an intermediate such as a histidine-containing phosphotransfer (Hpt) 2 protein (1,2). The response regulators are typically multidomain proteins that comprise a conserved receiver and a variable effector domain. The interdomain communication between the regulatory and effector domains of a response regulator shows significant diversity. Removal of the receiver domain may inhibit or constitutively activate the effector domain (3,4). A more intricate role for response regulators was found in PhyR, which combines a receiver domain with an N-terminal domain that is extremely similar to the E subunit of RNA polymerase. Unlike other DNA-binding transcriptional regulators, PhyR acts through protein-protein interaction in a partner-switching mechanism where the E domain of PhyR binds to the antifactor NepR. As a result, the original factor EcfG is free to transcribe stress-related genes (5). Partner-switching regulatory systems typically comprise anti-anti-, anti-, and factors. Dephosphorylation of the anti-anti-factor by a Ser/Thr protein phosphatase enables its direct interaction with the anti-factor, rendering it unavailable to bind to the factor. The factor is then free to transcribe the downstream genes (6,7). Once the anti-antiis phosphorylated by the Ser/Thr protein kinase activity of the anti-factor, it dissociates from the anti-factor, which is then free to bind to the factor, consequently inhibiting the downstream gene expression. Such a signaling paradigm was first observed in Bacillus subtilis SpoIIE-SpoIIAB-SpoIIAA, which forms partner switchers when regulating sporulation-related F (8). The other example is Bordetella bronchiseptica BtrU-BtrW-BtrV, which is responsible for regulating a Type III secretion system (9).
Pseudomonas aeruginosa is a Gram-negative, motile bacterium and an opportunistic pathogen known to be the leading cause of numerous acute and chronic nosocomial infections. Our previous studies of P. aeruginosa two-component regulatory systems revealed that following activation by environmental stresses multiple sensor kinases (PA1611, PA1976, PA2824, and RetS) transfer a phosphoryl group to HptB, which then relays the signal to the response regulator PA3346. The phosphorylation on the PA3346 N-terminal receiver domain increases its Ser protein phosphatase activity, leading to dephosphorylation of the potential anti-antifactor PA3347. Although the target phosphorylation site of PA3347 was shown to be Ser-56, the corresponding Ser protein kinase and antifactors remain elusive. The isogenic hptB deletion mutant is defective in swarming, whereas the PA3346 or PA3347 mutant showed increased swarming compared with swarming and other characteristics for the involvement of motility genes in P. aeruginosa PAO1 (10 -12).
According to recent information from the Pseudomonas genome database and the P. aeruginosa Community Annotation Project (PseudoCAP), ϳ50 flagellum and chemotaxis genes are present in the P. aeruginosa PAO1 genome (13,14). The checkpoint for flagellum biogenesis is complex, requiring at least two factors, RpoN ( 54 ) (15) and FliA ( 28 ) (16). The direct interaction of FliA and the antifactor FlgM in P. aeruginosa was demonstrated using the yeast two-hybrid system, which revealed their role in regulating flagellar biogenesis using a post-translational mechanism (17).
The P. aeruginosa flagellum plays a critical role in virulence as shown by several animal models where the bacteria with flagella were more invasive than flagellum-deficient strains (18). The strain with a deletion at fliC, which codes for the major flagellin in P. aeruginosa, showed a loss of virulence in a pulmonary infection model. These results suggest that the flagellum plays a key role in the P. aeruginosa invasion of epithelial cells (19). Recently, Bordi et al. (20) demonstrated that the HptB signaling pathway is linked to the regulation of the GacA/ GacS two-component system by down-regulating the expression of rsmY, a small RNA. Because rsmY is a known regulator of the Type VI secretion system (21), HptB may also negatively regulate protein secretion.
The objective of this study was to explain the regulatory mechanism of HptB-PA3346-PA3347. The microarray analysis of an hptB mutant revealed that a number of the FliA-dependent motility genes were down-regulated, and these results were validated via real time PCR. We also demonstrate that the PA3346 C-terminal region (PA3346C) is a novel Ser/Thr protein kinase that phosphorylates and thereby regulates the activity of PA3347. Our results demonstrate that PA3347 interacts with FlgM and PA3346C in a competitive manner. Overall, these findings indicate that HptB, PA3346, PA3347, FlgM, and FliA are organized into a phosphorylation-dependent partnerswitching system that regulates bacterial swarming.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Primers-The bacterial strains and plasmids used in this study are listed in Table 1. The  primers used are listed in supplemental Table S1.
Microarray Analysis for Transcriptional Profiling of P. aeruginosa PAO1 and hptB Mutant MPA45-P. aeruginosa PAO1 and hptB mutant strain MPA45 were cultured on swarming plates for 36 h at 30°C. Bacteria were collected from the edge of swarming bacterial colonies, and the RNA was extracted using a Qiagen RNeasy Midi kit (Qiagen, German- Ap r , a fragment containing entire PA3347 in pMMB66EH 12 pMMB47S56A Ap r , a fragment containing PA3347-S56A in pMMB66EH This study pMMB47S56D Ap r , a fragment containing PA3347-S56D in pMMB66EH This study town, MD). Residual genomic DNA was removed using RQ1 DNase (Promega, Madison, WI). Approximately 10 g of the RNA from both P. aeruginosa PAO1 and the htpB knock-out mutant MPA45 was reverse transcribed into cDNA using Superscript II reverse transcriptase (Invitrogen). The resulting cDNA was purified using a Qiagen PCR purification kit and tagged with biotin using the GeneChip DNA labeling reagent (Affymetrix) according to the manufacturer's instructions. The labeled cDNA was used to hybridize the Affymetrix GeneChip. Data were normalized by Linear Models for Microarray Data (22), and differential gene expression tables were generated. Real Time Quantitative PCR Analysis-The same total RNA source for P. aeruginosa PAO1 and MPA45 used in the microarray gene expression profile was used in the validation experiment through real time quantitative PCR analysis. Approximately 1.5 g of RNA was reverse transcribed into cDNA using Superscript II reverse transcriptase (Invitrogen). For real time PCR quantification using the cDNAs of P. aeruginosa PAO1 and MPA45, primer sets specific for genes fliC, fliD, fliM, flgM, PA4915, and cheY were designed using the Primer Express software under the factory default settings. Two housekeeping genes, rpoD and proC, were used as an internal control of gene expression, and the mean of the proC and rpoD expression levels in the sample was used as the normalization factor (23). All primers were tested for the absence of nonspecific bands or primer dimer formation prior to real time PCR analysis. Real time PCR was performed with SYBR Green Master Mix and analyzed using the Fast 7500 Real-Time PCR system (Applied Biosystems, Foster City, CA). The data analyses for real time PCR were performed using the Applied Biosystems Sequence Detection System software, version 1.4. A comparative Ct method was applied to quantify the expression levels. All genes were examined in duplicate. The quantitative data of mRNA expression for P. aeruginosa PAO1 and MPA45 obtained from two independent experiments were calculated using the 2 Ϫ⌬⌬Ct method (24). Similarly, the cDNAs of P. aeruginosa PAO1 and the PA3347 deletion mutant MJL47 were used to quantify fliA and flgM using an appropriate set of primers.
Bioinformatics Analysis-Proteins resembling PA3346 were detected in the NCBI database using the protein BLAST search tool. A comparison and an alignment of these homologous protein sequences were performed using Vector NTI (Invitrogen) to identify the conserved amino acid residues. Identification of restriction sites and open reading frame predictions were also conducted using Vector NTI.
Construction of Plasmids-The restriction endonucleases and DNA-modifying enzymes were purchased from New England Biolabs (Ipswich, MA). P. aeruginosa PAO1 genomic DNA was purified with the Wizard Genomic DNA Purification kit (Promega) and used as the template for PCR amplification. The PCR product for the C-terminal region of PA3346 (from nucleotide position 1222 to 1713) and flgM was digested with restriction enzymes NdeI and NotI and ligated directionally into the expression vector pET30a (Novagen, Madison, WI) to generate plasmids pHL34 and pMBH3051, respectively. Similarly, the PCR product of fliA was digested with NdeI and HindIII and ligated into pET30a to yield the plasmid pHL40. The PCR prod-uct of PA3347 was ligated into pET100 (Invitrogen), providing the clone pMBH10047. In another set of experiments, the PCR product of flgM was digested with EcoRI and XhoI and ligated directionally into the expression vector pGEX-5X-1 (Amersham Biosciences), resulting in pMBH551. For the bimolecular fluorescence complementation (BiFC) assay, the PCR products of PA3347 and PA3347-S56A were digested with a restriction enzyme and then ligated into the pET11a-link-NGFP vector at the XhoI and BamHI restriction enzyme sites, resulting in pNB-IFC3347 and pNBIFC3347M, respectively. Similarly, the PCR products of flgM and PA3346C were digested using restriction enzymes NcoI and AatII and fused into the pMRBAD-link-CGFP vector, resulting in clones of pCBIFC3351 and pHLB-IFC46, respectively. Nucleotide sequencing was performed to verify whether the sequences were correct and in-frame in these plasmids.
Expression and Purification of Recombinant Proteins-The Escherichia coli BL21(DE3) cells were transformed with the pET-based flgM, fliA, and PA3346C clones separately, and gene expression was induced using isopropyl ␤-D-thiogalactopyranoside at a final concentration of 0.1 mM with a constant shaking rate of 150 rpm at 20°C for 16 h. The cells were collected by centrifugation, resuspended in a lysis buffer (20 mM Tris-HCl, pH 8.0, 1 mM phenylmethylsulfonyl fluoride, 500 mM NaCl), and disrupted on ice by ultrasonication. After centrifugation at 14,000 rpm at 4°C for 20 min to remove debris, the clarified supernatant was loaded onto a nickel-charged resin (Ni-Sepharose TM High Performance, Amersham Biosciences), and the proteins retained in the column were eluted using the elution buffer (50 mM sodium phosphate, 300 mM NaCl, 500 mM imidazole, pH 8.0). Similarly, glutathione S-transferase (GST)-tagged PA3347, PA3347-S56A, FlgM, and glutathione S-transferase recombinant proteins were affinity-purified on glutathione-Sepharose 4 Fast Flow resin (Amersham Biosciences) separately using the GST elution buffer (50 mM Tris-HCl, pH 8.0, 20 mM glutathione). The purified proteins were dialyzed against PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 2 mM KH 2 PO 4 , 50% glycerol, pH 7.4), and the concentration of these proteins was determined via the Bradford method using a Bio-Rad kit.
In vitro phosphorylation for mass spectrometer analysis was performed as described above except that [␥-32 P]ATP was replaced with 5 mM unlabeled ATP. The eluted sample was further resolved by SDS-PAGE. In-gel digestion was performed as described previously (25) with the following modifications: destaining was achieved by washing the gel twice with 20 mM NH 4 HCO 3 , acetonitrile (1:1 mixture) for 15 min at room temperature, and reduction was performed using 10 mM dithiothreitol in 20 mM NH 4 HCO 3 for 15 min at 56°C. Alkylation was performed using 55 mM iodoacetamide in 20 mM NH 4 HCO 3 for 20 min in darkness at room temperature. For proteolytic digestion, the gel was treated overnight at 37°C with 20 g of sequencing grade trypsin (Promega). The digested sample was sonicated for 10 min in the presence of 1% trifluoroacetic acid, and the supernatant was pooled and subjected to MALDI-TOF mass spectrometer analysis.
Kinetics of PA3347 Phosphorylation-Four different concentrations (0.25-2.0 M) of PA3347 were incubated with PA3346C (1 M) in the presence of 0.1 mM ATP and 10 mM MgSO 4 in a 30-l volume at 37°C for 7 min. During this period of time, the PA3347 phosphorylation levels were linear with regard to time. The phosphorylation assay was terminated by adding an equal volume of the SDS loading buffer and boiled for 10 min. The protein was resolved using polyacrylamide gel electrophoresis and transferred to a polyvinylidene fluoride (PVDF) membrane (Amersham Biosciences). The membrane was probed with a phosphoserine-specific mouse monoclonal antibody (Qiagen) followed by a horseradish peroxidase-conjugated secondary antibody. The band was visualized using a NovaRED TM Substrate kit (Vector Laboratories, Burlingame, CA). The phosphorylation intensity was determined by quantifying the immunoblot using ImageJ software, and kinetics analysis was performed using the GraphPad Prism software.
GST Pulldown Assay-The recombinant proteins for testing protein-protein interaction were incubated with 500 l of a 50% slurry of glutathione-agarose beads in PBS for 2 h at 4°C either in the absence or presence of 2 mM ADP and 2 mM MgSO 4 . The reaction mixture was then centrifuged at 3,000 rpm for 5 min, the precipitated glutathione beads were washed twice with PBS, the protein complex was eluted with the GST elution buffer, and the eluted fraction was concentrated using Amicon Ultra (Millipore, Billerica, MA). The proteins present in the eluted fractions were resolved on a polyacrylamide gel and then transferred onto a PVDF membrane. The membrane was incubated sequentially with a monoclonal antibody against His tag (Merck KGaA) and a horseradish peroxidase-conjugated secondary antibody (Jackson ImmunoResearch Laboratories, West Grove, PA). Finally, the target proteins were identified using a NovaRED Substrate kit (Vector Laboratories).
BiFC Assay-The BiFC assay was performed as described previously (26,27) with slight modifications. The positive control plasmid pair was pET11a-Z-NGFP and pMRBAD-Z-CGFP encoding the green fluorescence protein fused with one of the antiparallel leucine zipper tags, respectively. The negative control was the cells harboring pET11a-link-NGFP and pMRBADlink-CGFP. The interacting partners were introduced into E. coli BL21(DE3), and the transformants were cultured on Luria-Bertani agar containing 50 g/ml ampicillin, 35 g/ml kanamycin, 0.1 mM isopropyl ␤-D-thiogalactopyranoside, and 0.05% arabinose at 20°C for 72 h. The cells were examined for the presence of green fluorescence by an Olympus BX-51 epifluorescence upright microscope. The images were analyzed using a digital camera and SPOT Advanced Plus Imaging Software (version 4.6).
Antibody Production and Co-immunoprecipitation-Recombinant GST-PA3347 protein was used as an immunogen to raise rabbit polyclonal antisera for detecting both PA3347 and PA3347-S56A. The co-immunoprecipitation was performed as described previously (28,29). Protein A-Sepharose beads (125 l; Amersham Biosciences) were incubated overnight at 4°C with 50 g of rabbit polyclonal PA3347 antibody in 125 l of Pierce ImmunoPure Ag/Ab binding buffer (Thermo-Fisher Scientific Inc., Rockford, IL). For the co-immunoprecipitation of PA3347 with His 6 -FlgM, beads were incubated with 500 l of the whole cell lysates from ϳ10 9 cfu of PAO1 and 200 g of His 6 -FlgM at 4°C overnight under constant agitation. For the co-immunoprecipitation of GST-PA3347-S56A with His 6 -FlgM, beads were incubated with 500 l of the whole cell lysates of E. coli expressing GST-PA3347-S56A from ϳ10 9 cfu and 200 g of His 6 -FlgM. In both cases, whole cell lysates alone and His 6 -FlgM alone were used as negative controls. The co-immunoprecipitation of PA3347 with His 6 -PA3346C was similarly performed. The beads were washed with PBS and boiled in SDS loading dye for 10 min. Western blotting was conducted using the mouse monoclonal anti-His 6 antibody (1:2,500) as the primary antibody and the goat anti-mouse IgG antibody conjugated to HRP (1:10,000) as the secondary antibody.
Competitive Interaction Using GST Pulldown Assay-To investigate the binding activity of PA3346C versus FlgM on PA3347 in the presence of ATP or ADP, a competition assay was performed as described previously (30) with slight modifications. A constant concentration of PA3347 (25 g/ml) bound to FlgM (30 g/ml) was incubated on 100 l of a 50% GST bead slurry in six separate tubes. Following 2 h of mixing, 2 mM ADP and 2 mM MgSO 4 were added to three microcentrifuge tubes with increasing concentrations of PA3346C (0, 10, and 20 g/ml). Meanwhile, in three other tubes, 2 mM ATP was added instead of ADP and then incubated for 2 h. In a reciprocal experiment, a constant concentration of PA3347 was bound to PA3346C and incubated with increasing concentrations of FlgM (0, 15, and 30 g/ml) in the presence of either ATP or ADP. In this study, GST was used as a negative control.
Similarly, to demonstrate that increasing concentrations of FliA (0, 20, and 100 g/ml) compete to bind with FlgM (30 g/ml) in a complex comprising PA3347 (30 g/ml), a competition assay was performed. The GST beads carrying the interacting proteins were washed with PBS extensively and then eluted with the GST elution buffer. The proteins present in the eluents were resolved using SDS-polyacrylamide gel electrophoresis.
Analysis of Bacterial Swarming-A swarming assay for P. aeruginosa PAO1 and P. aeruginosa PA3347 mutant MJL47 was performed as reported previously (12). Briefly, the swarming assay medium comprises an M8 salt base (50 mM Na 2 HPO 4 , 25 mM KH 2 PO 4 , 4 mM NaCl) supplemented with 0.5% Bacto agar, 0.02% glucose, and 10 mM glutamic acid. The bacterial strains that required testing were cultured overnight, transferred to the swarming plate using sterile toothpicks, and then incubated at 37°C for 36 h. The experiment was performed three independent times.

RESULTS
Transcriptional Profiling of P. aeruginosa PAO1 and hptB Mutant-To obtain a more comprehensive picture of the physiological functions of the HptB-PA3346-PA3347 signaling system, a Pseudomonas whole genome DNA microarray was used to determine the gene expression profiles of the hptB mutant. The results were compared with those of the wild-type PAO1 strain. This study identified 142 genes with a more than 2-fold increase in expression and 235 genes showing at least a 2-fold decrease in expression. The top 20 up-and down-regulated genes are listed in supplemental Tables S2 and S3, respectively. Approximately half of the genes on the list encode a hypothetical protein. Several genes, including adhA, pldA, stp1, and pilQ that may be associated with bacterial virulence, were also noted. Among the genes up-regulated in the hptB mutant, adhA is known to be important in biofilm formation in P. aeruginosa (31). This notion is consistent with higher biofilm formation in the hptB mutant compared with P. aeruginosa (12). The gene pldA, which encodes the outer membrane phospholipase D, is known to be involved in chronic pulmonary infections (32). The stp1 product is a Ser/Thr protein phosphatase belonging to the protein phosphatase 2C family. It might regulate the Type III secretion system or alginate synthesis in P. aeruginosa (33). On the down-regulated gene list, the pilQ gene product is required for assembly and secretion of the type IV pili (34), which are cell surface organelles mediating adhesion and twitching motility. Finally, the exsC gene product acts as an anti-anti-activator for transcription of the Type III secretion system. This is in accordance with the finding that HptB positively controls the Type III secretion system in P. aeruginosa (20).

Expression of Certain Flagellum and Chemotaxis Genes Is Down-regulated in hptB Mutant-
The reduced expression genes in the hptB mutant MPA45 include several motility genes such as cheY, which functions as a switch to control the flagellar rotation direction (35), fliM, which controls flagellar rotation (36), the flagellar cap protein-encoding gene fliD, which participates in the adhesion process (37), and the flagellation protein genes fliC and flgM. To validate the microarray data, the expression levels of these motility-related genes and PA4915, which encodes a hypothetical protein involved in chemotaxis, were measured using quantitative real time PCR. Consistent with the microarray analysis results, these motility-related genes also exhibited reduced expression levels in the quantitative real time PCR assay although with a higher dynamic range of -fold change ( Table 2). The discrepancy is not uncommon and may be attributed to many factors such as a difference in data nor-malization, transcript detection (SYBR Green versus CyDyes), and selection of PCR primers and hybridization probes (38).
C-terminal Region of PA3346 Contains Kinase Domain-We have shown previously that HptB (PA3345) can transmit a phosphoryl group from a sensor kinase to PA3346 (12). As a response regulator, the N-terminal region of PA3346 has a signal receiver domain containing a conserved Asp residue. The middle portion of the protein comprises a protein phosphatase 2C domain, which can dephosphorylate PA3347 (12). Further analysis of the C-terminal region of PA3346 using the BLASTP program revealed significant similarity to several confirmed Ser/Thr protein kinases, including Bacillus stearothermophilus SpoIIAB (39), B. subtilis SpoIIAB (40), B. subtilis RsbW (7), and B. bronchiseptica BtrW (9) (Fig. 1). The amino acid residues and molecular structure of B. stearothermophilus SpoIIAB have been determined and found to be distinct from Hank's type Ser protein kinases (41) while also exhibiting significant similarities to the ATPases of the Gyrase, Hsp90, Histidine kinase, and MutL superfamily (42). Similar to SpoIIAB and RsbW, the C-terminal region of PA3346 also has N, G1, and G2 boxes responsible for /antiantagonist binding, Mg 2ϩ ion binding, and ADP/ATP binding, respectively (43, 44) (Fig. 1).
PA3346C Is Divalent Cation-dependent Ser/Thr Protein Kinase-If PA3347 is an anti-factor antagonist, it is likely to be phosphorylated, and the phosphorylation should influence its interaction with the anti-factor. Phosphorylation of PA3347 at Ser-56 using the whole cell extract of P. aeruginosa PAO1 was demonstrated previously through an in vitro phosphorylation assay (12). The presence of a kinase domain at the C-terminal region of PA3346 implies that the protein is probably the kinase in PA3347 phosphorylation. To verify this possibility, the PA3346 C-terminal region ranging from nucleotide position 1222 to 1713 was cloned and overexpressed, and the recombinant protein, designated PA3346C, was used in the in vitro phosphorylation assay with [␥-32 P]ATP as a tracer. As shown in Fig. 2, PA3346C is capable of phosphorylating PA3347 but not PA3347-S56A, the PA3347 variant with an Ala substitution at Ser-56 ( Fig. 2A). The PA3346C kinase activity was found to be dependent on the divalent cations Mg 2ϩ , Ca 2ϩ , and Mn 2ϩ but not on Ni 2ϩ , Co 2ϩ , or Zn 2ϩ (Fig. 2B). The kinetics study determined the K m (0.645 Ϯ 0.076 mM) and V max (3.144 Ϯ 0.148 mM/s) of PA3346C with PA3347 as the substrate (supplemental Fig. S1).
PA3346C Phosphorylates PA3347 at Ser-56-To further verify the phosphorylation site on PA3347, PA3347 with or without PA3346C treatment was subjected to trypsin digestion followed by MALDI-TOF mass spectrometry. The PA3347 sample treated with PA3346C and ATP yielded two peaks of 1849.997 and 1929.963 Da, respectively (supplemental Fig.  S2A), representing unphosphorylated and phosphorylated Ser-56-containing peptides (NATYLDSSALGMLLLLR) of PA3347. However, the phosphorylated Ser-56-containing peptide peak of 1929.963 Da was not observed in the PA3347-S56A site-directed mutant (supplemental Fig. S2B). Together, the results indicate that PA3347 is phosphorylated at Ser-56 by PA3346C. The result is consistent with our previous finding using the whole cell extract to phosphorylate PA3347 (12). PA3347 Interacts with PA3346C in Vitro-To function as an antiantagonist, PA3347 must interact with PA3346C to perform regulatory activity. Using the GST pulldown assay, both PA3347 and PA3347-S56A were found to form a stable complex with PA3346C but only in the presence of ADP (Fig. 3A), suggesting that ADP stabilizes PA3346C in a conformation favorable for binding with PA3347 (Fig. 3A).

TABLE 2 Comparison of expression levels of selected motility-related genes in hptB mutant determined by real time quantitative PCR (Q-PCR) and microarray
PA3347 Interacts with FlgM in Vitro-The PA3347 deletion mutant MJL47 displayed a hyperswarming phenotype, suggesting an up-regulation of motility-related gene expression (12). In P. aeruginosa PAO1, the antifactor FlgM-encoding gene (PA3351) is located immediately downstream of the hptB-PA3346-PA3347 operon (12). Because FlgM has been shown to interact with the factor FliA ( 28 ) to regulate the synthesis of flagella (17), this study examined whether PA3347 regulates the FliA-dependent flagellum activity by interacting with FlgM. As shown in Fig. 3B, the His 6 -tagged FlgM could be co-eluted with either the GST-tagged PA3347 or the PA3347-S56A proteins, indicating an interaction between the FlgM protein and PA3347 or PA3347-S56A. Binding in the absence of ATP suggests that phosphorylation of PA3347 is not required for the interaction.
PA3347 Interacts with FlgM as Well as with PA3346C in Bimolecular Fluorescence Complementation Assay-To demonstrate that interaction of PA3347 with its binding partners FlgM and PA3346C is also taking place in vivo, a set of plasmids that are stable and express the proteins upon induction in E. coli BL21(DE3) was constructed for the BiFC assay. The cells harboring positive control plasmids gave a strong green fluorescence due to GFP reassembly, whereas negative control cells showed no green fluorescence. As shown in Fig. 4, cells harboring pNBIFC3347 and pCBIFC3351, expressing the NGFP-FIGURE 1. Amino acid sequence alignment of PA3346C with other Ser/Thr protein kinases. PA3346C (from Leu-408 to Ala-571) shows a 65% similarity and 26% identity to RsbW of B. subtilis, a 50% similarity and 28% identity to BtrW of B. bronchiseptica, and a 48% similarity and 32% identity to SpoIIAB of B. subtilis. Motifs labeled N, G1, and G2, which are commonly found in histidine kinases/ATPases, are indicated above the sequences. Amino acid residues with strong similarity are only considered in consensus sequences and are shaded in gray. The identical residues are shaded in black.   cells harboring pNBIFC3347 and pHLBIFC46, which encodes PA3346C, also showed green fluorescence (Fig. 4).
Co-immunoprecipitation of PA3347 with FlgM and PA3346C Proteins in Vivo-Co-immunoprecipitation analysis showed that recombinant His 6 -FlgM and His 6 -PA3346C proteins bound to both the PA3347 expressed in PAO1 and the recombinant GST-PA3347-S56A expressed in E. coli. A control experiment demonstrated that PA3347, PA3347-S56A, the antibody against PA3347, the recombinant His 6 -FlgM, and His 6 -PA3346C proteins alone do not bind to Protein A-Sepharose beads (Fig. 5, A and B).
Expression of Both fliA and flgM Is Regulated by PA3347-To determine whether PA3347 is involved in regulating the expression of fliA and flgM, we analyzed the expression of flgM and fliA in the presence and absence of PA3347. Compared with that of the wild type, the expression of flgM and fliA was significantly increased in PA3347 mutant MJL47 by ϳ5and 13-fold, respectively, as determined by quantitative real time PCR. This is possibly because PA3347 deletion results in the partial loss of FlgM function as an antifactor, which results in an increase of FliA activity to induce fliA and flgM transcription. It is necessary to mention only a partial loss because full loss of FlgM function would bring about uncontrolled filament assembly and loss of motility.
FliA Competes for Binding to FlgM-PA3347 Complex, and FlgM Competes for Binding to PA3347-PA3346C Complex-To verify that the HptB-PA3346-PA3347 signaling system can modulate flagellum activity through the interaction of FlgM and FliA, a GST pulldown competitive assay was performed. As shown in Fig. 6, increasing concentrations of FliA competitively displaced the binding of PA3347 to FlgM, demonstrating the well known finding that FliA interacts directly with the antifactor FlgM (17).
Elucidation of whether PA3346C, FlgM, and PA3347 can form a stable complex or whether PA3346C and FlgM bind to PA3347 in a competitive manner is critical for understanding the regulatory mechanism exerted by the HptB-mediated signaling pathway. As shown in Fig. 7A, increasing concentrations of FlgM displaced the binding of PA3346C to PA3347 in the presence of ATP. However, displacement was not observed when ADP was used to replace ATP. This finding agrees with the expected partner-switching mechanism that the phosphorylation of the antiantagonist by the Ser protein kinase in the presence of ATP could lead to the dissociation of the complexes, whereas ADP can stabilize the complexes (8,9,45). By contrast, the binding of FlgM to PA3347 was not affected by increasing concentrations of PA3346C in the presence of ATP or ADP. This suggests that the binding of FlgM and PA3347 is independent of the phosphorylation state (Fig. 7B).
PA3347 Phosphorylation Is Crucial for P. aeruginosa Swarming-Consistent with a previous finding (12), deleting PA3347 increased the swarming motility, and the introduction of a plasmid that harbored a full-length PA3347 (pMMB47) into MJL47 could reverse the hyperswarming motility to the parental phenotype (Fig. 8A). More interestingly, a plasmid  The competitive binding between His 6 -FliA and His 6 -PA3347 for immobilized GST-FlgM was carried out. The pulldown of GST-FlgM (30 g/ml) was conducted in the presence of His 6 -PA3347 (30 g/ml) and increasing concentrations of His 6 -FliA. Lane 1, markers; lanes 2-4, a constant concentration of GST-FlgM bound to His 6 -PA3347 was incubated with increasing amounts of His 6 -FliA (0, 20, 100 g/ml, respectively). The proteins were resolved on an SDS-polyacrylamide gel and stained with Coomassie Blue. expressing a PA3347 with an Asp substitution at Ser-56 (pMMB47S56D) to mimic a constitutively phosphorylated Ser residue can restore the hyperswarming activity of MJL47 to that of wild-type PAO1. By contrast, the plasmid pMMB47S56A producing a non-phosphorylated variant of PA3347 cannot complement the hyperswarming phenotype of MJL47. Additionally, pMMB47 can restore the defect in the swarming phenotype of MPA45, whereas pMMB47S56A cannot (Fig. 8B). However, neither pMMB47 nor pMMB47S56A could complement the PA3346 deletion mutant MJL46 to the parental phenotype (Fig. 8C). This suggests that the absence of the PA3346 as a kinase in MJL46 prevents the phosphorylation of PA3347 produced from pMMB47 and hence formation of a stable complex with FlgM. These findings indicate that phosphorylation at Ser-56 of PA3347 plays a critical role in regulating flagellum activity.

DISCUSSION
The objective of this study was to elucidate the molecular mechanism that leads to flagellum gene regulation by the HptB-PA3346-PA3347 signaling system in P. aeruginosa PAO1. The response regulator PA3346 differs from other transcriptional regulators because it is bifunctional, possessing both phosphatase and kinase activities. PA3346C was found to be a homolog of SpoIIAB. SpoIIAB performs the dual function of a Ser protein kinase and an anti-factor (46). This study has shown that PA3346C behaves like SpoIIAB, exhibiting a Ser protein kinase activity capable of phosphorylating PA3347 and an antiantagonist binding activity.
This study also demonstrated a direct interaction between the anti-factor antagonist PA3347 and the anti-28 factor FlgM both in vitro and in vivo. The competition assay showed  that the binding of FlgM to PA3347 was not affected by excess PA3346C, indicating that FlgM and PA3346C possibly use a different structural motif to interact with PA3347. However, the FlgM-PA3347 complex can be displaced by excess FliA, suggesting that both PA3347 and FliA may interact with FlgM possibly through the same binding site. Furthermore, the role of PA3347 phosphorylation in regulating swarming motility indicates that the regulation essentially occurs through a partnerswitching mechanism. Partner-switching mechanisms for various Gram-positive bacteria have been documented. This type of regulation has been reported more recently for Gram-negative bacteria such as the Type III secretion system regulator BtrU-BtrW-BtrV in B. bronchiseptica (43). A conserved partner-switching regulatory system was also found in Chlamydia trachomatis (47). The partner-switching mechanism of P. aeruginosa may also have a widespread presence in related Gram-negative bacteria, controlling important aspects of bacterial physiology such as swarming.
According to the model established in this study (Fig. 9), HptB modifies the receiver domain of PA3346, altering the balance of phosphatase and kinase activity. In certain growth conditions, this alteration would affect flagellar gene expression through the partner-switching mechanism involving PA3347, FlgM, and FliA as demonstrated in this study. However, because swarming is a complex bacterial behavior, HptB may also affect swarming by other mechanisms. In analogy of SpoI-IAB, the kinase domain of PA3346 (PA3346C) may act as an antifactor to regulate a yet to be identified factor. This pathway may participate in cyclic di-GMP level modulation as suggested in a previous study (20,48) and therefore may explain why the PA3347 deletion mutant shows a hyperswarming phenotype. In addition, this PA3346C-mediated pathway may affect the expression of genes unrelated to flagellum biogenesis that were found in our gene profiling analysis (supplemental Tables S2 and S3). Experiments are being carried out to verify these hypotheses.
Direct observation of the leading edge of the swarming colonies using optical microscopy showed that the wild-type colony edge exhibited vigorous and rapid collective cell movements. By contrast, cells at the leading edge of the hptB deletion MPA45 colony were resting (supplemental Videos S1 and S2). The absence of collective movement in the MPA45 strain, causing a defective swarming phenotype, is probably because the flagella were absent or incompletely assembled. To test this hypothesis, flagella were stained in both the wild-type and MPA45 swarming cells and examined under a microscope. Intriguingly, although both the microarray and real time PCR analyses showed that numerous flagellum regulatory and biosynthesis genes were significantly down-regulated in the hptB mutant, no apparent difference in the flagellum morphology was observed (supplemental Fig. S3, A and B). The hptB mutant cells even swarmed normally in broth with a chemical composition identical to the swarming agar. The twitching motility of the htpB mutant was also comparable with the wild-type cells, indicating normal type IV pili. Thus, the precise defects leading to the deficient swarming activity of the hptB mutant still require investigation. A component responsible for regulating rotation or supplying energy for flagellum movement is likely defective in the hptB mutant cells as also revealed by our gene expression profiling analysis. Such gene types have been FIGURE 9. Proposed model for PA3346 -3347-FlgM partner-switching mechanism that regulates P. aeruginosa PAO1 swarming. The N-terminal region of PA3346 can dephosphorylate PA3347. PA3347 binds FlgM, an anti-factor known to negatively regulate 28 activity and affect flagellum structural biogenesis class III gene transcription. The C-terminal domain of PA3346, which is also known to act as a serine kinase, can phosphorylate PA3347, an antiantagonist factor. The absence of PA3347 or the phosphorylation state of PA3347 can activate the PA3346 C terminus (shown as the HATP domain of PA3346 in the figure) to act as an anti-factor and inactivate the yet to be identified factor in a partner-switching mechanism. REC, receiver domain; PP2C, protein phosphatase 2C domain; HATP, histidine kinase/ATPase domain.