Characterization of the AtsR Hybrid Sensor Kinase Phosphorelay Pathway and Identification of Its Response Regulator in Burkholderia cenocepacia*

Background: AtsR is a Burkholderia cenocepacia hybrid sensor kinase that negatively regulates quorum sensing. Results: The AtsR phosphorelay mechanism includes the AtsT cognate response regulator. Conclusion: AtsR regulation occurs through phosphorylation of AtsT, which is fine-tuned by the AtsR receiver domain. Significance: The AtsR/AtsT phosphorelay pathway uncovers a major global regulator of B. cenocepacia pathogenicity. AtsR is a membrane-bound hybrid sensor kinase of Burkholderia cenocepacia that negatively regulates quorum sensing and virulence factors such as biofilm production, type 6-secretion, and protease secretion. Here we elucidate the mechanism of AtsR phosphorelay by site-directed mutagenesis of predicted histidine and aspartic acid phosphoacceptor residues. We demonstrate by in vitro phosphorylation that histidine 245 and aspartic acid 536 are conserved sites of phosphorylation in AtsR, and we also identify the cytosolic response regulator AtsT (BCAM0381) as a key component of the AtsR phosphorelay pathway. Monitoring the function of AtsR and its derivatives in vivo by measuring extracellular protease activity and swarming motility confirmed the in vitro phosphorylation results. Together we find that the AtsR receiver domain plays a fine-tuning role in determining the levels of phosphotransfer from its sensor kinase domain to the AtsT response regulator.

AtsR is a membrane-bound hybrid sensor kinase of Burkholderia cenocepacia that negatively regulates quorum sensing and virulence factors such as biofilm production, type 6-secretion, and protease secretion. Here we elucidate the mechanism of AtsR phosphorelay by site-directed mutagenesis of predicted histidine and aspartic acid phosphoacceptor residues. We demonstrate by in vitro phosphorylation that histidine 245 and aspartic acid 536 are conserved sites of phosphorylation in AtsR, and we also identify the cytosolic response regulator AtsT (BCAM0381) as a key component of the AtsR phosphorelay pathway. Monitoring the function of AtsR and its derivatives in vivo by measuring extracellular protease activity and swarming motility confirmed the in vitro phosphorylation results. Together we find that the AtsR receiver domain plays a fine-tuning role in determining the levels of phosphotransfer from its sensor kinase domain to the AtsT response regulator.
Phosphorylation cascades governed by two-component signal transduction systems provide key signaling mechanisms in bacteria, Archaea, simple eukaryotes, and higher plants, allowing them to translate signals into adaptive responses (1). These regulatory pathways consist of a transmembrane protein that responds to an environmental cue leading to autophosphorylation followed by the transfer of the phosphate to a cytoplasmic response regulator (RR). 2 The sensor protein has a variable region dedicated to sensing a signal and a conserved histidine kinase (HK) domain. The RR typically consists of a conserved receiver domain (RD) and a variable effector domain that binds to DNA. Upon perceiving the signal, the HK becomes auto-phosphorylated on its conserved histidine (His) residue, and the phosphate is transferred to a conserved aspartic acid (Asp) residue on the RR. More complex phosphorelay systems require multiple phosphoryl transfer reactions involving a hybrid sensor kinase that contains an extra domain serving as a RD, a histidinephosphotransfer protein, and response regulator proteins (1,2). Because the mechanism of phosphotransfer is from His to Asp, histidine-phosphotransfer proteins shuttle phosphoryl groups between a hybrid sensor kinase and a RR, providing additional checkpoints of regulation in signaling pathways (3,4).
Burkholderia cenocepacia is a Gram-negative opportunistic pathogen belonging to the Burkholderia cepacia complex that causes severe, chronic respiratory infections in patients with cystic fibrosis and other immunocompromised conditions (5,6). Furthermore, B. cenocepacia can also be found in maize roots (7) and sewage (8) and cause banana fingertip rot (9) and onion rot (10). This bacterium is also virulent in zebrafish (11), alfalfa, nematodes, and insect infection models (12) and can survive within amoebae and macrophages (13,14). We have previously identified AtsR (Adhesion and type six secretion system Regulator), which negatively regulates the expression of quorum sensing-regulated virulence factors in B. cenocepacia (15,16). B. cenocepacia strains have two N-acylhomoserine lactone-dependent quorum-sensing systems, CepIR and CciIR, that coordinate the expression of ZmpA and ZmpB zinc metalloproteases and other virulence factors during infection (17)(18)(19)(20)(21). In the absence of atsR, expression of cepIR and cciIR is up-regulated and mediates early and increased N-acylhomoserine lactone production, suggesting that AtsR plays a role in controlling virulence gene expression by modulating the timing of quorum sensing signaling (16). AtsR also represses the expression of virulence genes by an N-acylhomoserine lactoneindependent mechanism (16). Consequently, inactivation of atsR in B. cenocepacia also leads to increased biofilm formation, adherence to polystyrene and lung epithelial cells, extracellular protease secretion, and expression of a type 6 secretion system (T6SS). The latter is exemplified by actin cytoskeletal rearrangements with the formation of characteristic "pearls on a string-like structures around infected macrophages (15,16,22).
AtsR is a predicted membrane protein with two transmembrane domains and a sensor kinase region (containing the HK and ATPase domains) attached to an RD with conserved Asp residues. AtsR lacks a DNA binding motif found in canonical response regulator proteins, which suggests that AtsR does not bind directly to the promoter regions of target genes and is likely part of a multistep signal transduction pathway. Two genes near atsR encode two putative components of the AtsR phosphorelay pathway. One is BCAM0381 (herein designated atsT), a gene co-transcribed with atsR (15). AtsT is a putative cytoplasmic transcriptional regulator containing an N-terminal helix-turn-helix domain related to domains found in repressors (23) and a receiver domain with a conserved Asp at the C terminus. The other is BCAM0378, which locates immediately upstream of atsR and encodes a hypothetical protein containing His and Asp residues that form part of a conserved motif, UPF0047 (24). BCAM0378 might be functionally necessary as a histidine-phosphotransfer protein intermediate to transfer the phosphate from AtsR to the putative response regulator AtsT.
In this work we investigated the role of critical functional residues within the individual domains of AtsR using phosphorylation assays. We identified the conserved residues His-245 and Asp-536 as phosphoacceptor sites in AtsR. Furthermore, using in vivo and in vitro approaches, we demonstrate that AtsR is a hybrid sensor kinase that regulates downstream cellular activities through direct phosphorylation of AtsT.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Growth Conditions-Bacterial strains and plasmids used in this study are listed in Table 1. Bacteria were grown in Luria broth (8) (Difco) at 37°C unless indicated otherwise. Escherichia coli cultures were supplemented as required with the following antibiotics (final concentrations): tetracycline 30 g/ml, kanamycin 30 g/ml, and trimethoprim 50 g/ml. B. cenocepacia cultures were supplemented as required with trimethoprim 100 g/ml, tetracycline 150 g/ml, ampicillin 200 g/ml, and polymyxin B 20 g/ml. General Molecular Techniques-DNA manipulations were performed as described previously (25). T4 DNA ligase (Roche Diagnostics) and Antarctic phosphatase (New England Biolabs, Pickering, Ontario, Canada) were used as recommended by the manufacturers. E. coli DH5␣ and E. coli GT115 were transformed by the calcium chloride method (26). Mobilization of complementing plasmids and mutagenesis plasmids into B. cenocepacia K56-2 was performed by triparental mating using E. coli DH5␣ carrying the helper plasmid pRK2013 (27,28). DNA amplification by PCR was performed in a Bio-Rad C1000 TM Thermal Cycler with Taq or HotStar HiFidelity DNA polymerases (Qiagen, Canada). The oligonucleotide primers are listed in Table 2. DNA sequencing was performed at the DNA sequencing Facility of York University, Toronto, Canada.
Protein Expression and Purification-For overexpression and purification of recombinant proteins, a single colony was inoculated in LB broth supplemented with 30 g/ml kanamycin and grown at 37°C. Absorbance (A 600 ) was monitored until it reached 0.6. The culture was then shifted to 30°C, and 0.2 mM isopropyl-␤-D-thiogalactopyranoside was added to induce the expression of proteins. Cultures were incubated for an additional 4 h. Samples were analyzed by SDS-PAGE and stained with Coomassie Blue to confirm protein expression. Cells were harvested by centrifugation at 8000 ϫ g for 10 min. Cell pellets were resuspended in lysis buffer containing (50 mM sodium phosphate, pH 8, 0.3 M NaCl, 10 mM imidazole, pH 8.0, 10% glycerol, 0.25% Tween 20, and 1ϫ of EDTA-free protease inhibitor mixture (Sigma)) and disrupted by One Shot cell disruptor (Thermo Scientific, Rockville, MD). After centrifugation at 27,000 ϫ g for 30 min at 4°C to remove the debris, the clarified cell lysate was loaded onto Ni 2ϩ -binding Sepharose beads and washed, and the His-tagged proteins were then eluted using increased gradient concentrations of imidazole (125-500 mM). Fractions were analyzed by SDS-PAGE and stained with Coomassie Blue to determine the integrity of the purified protein.
In Vitro Phosphorylation Assay-For autophosphorylation and phosphotransfer reactions, 5 mol of each protein was added to the phosphorylation buffer containing 50 mM Tris-HCl, pH 8.0, 50 mM KCl, 5 mM MgCl 2 , 1 mM DTT, and 5 Ci of [␥-33 P]ATP (specific activity of 3000 Ci/mmol; 3.3 M stock solution) (PerkinElmer Life Sciences) in a final volume of 10 l. The reactions were carried out at room temperature for the desired time and were terminated by adding 3ϫ sample buffer (32). Reaction products were separated by electrophoresis on 14% SDS-PAGE gels. Gels were fixed, exposed to a high resolution screen (Kodak), and analyzed using a PhosphorImager with ImageQuant software (Molecular Dynamics 5.0; Amersham Biosciences). For pulse-chase experiments, an excess of nonradioactive ATP (20 mM) was added to the reaction mixture after a 10-min preincubation with [␥-33 P]ATP. Aliquots were taken before the addition of unlabeled ATP (time zero) and at various time points after the addition of cold ATP.
Chemical Stability of Phosphorylated His-245 and Asp-536-To probe the chemical stability of the phosphorylated residues, phosphorylated AtsR and AtsR D536A proteins were prepared as described above and treated with 1 M HCl, 1 M NaOH, or left untreated for 45 min at room temperature. The HCl reaction was neutralized with 0.25 volumes of 2 M Tris, pH 8, and analyzed by 16% SDS-PAGE followed by the PhosphorImager.
Protease, Swarming Motility, and Biofilm Formation-Protease assays were performed according to Aubert et al. (16). Briefly, 18-h cultures were normalized to an A 600 of 1. The bacterial suspension (3 l) was spotted onto dialyzed brain heart infusion (D-BHI) agar plates containing 1.5% carnation milk. The plates were incubated at 37°C and examined for zones of clearing around the bacterial spots at 48 h. The protease activity was recorded by measuring the radius of the surrounding halo (from the outside of the spot to the edge of the halo). Swarming motility assays were performed as described previously (16). Three l of overnight culture, adjusted to an A 600 of 1, was spotted on a swarm plate (0.8% nutrient broth, 0.5% agar, 0.2% glucose). Plates were incubated at 37°C for 20 h, and the diameters of swarming zones were measured. Biofilm formation assays were performed as described previously (15). Assays were done in triplicate and repeated independently three times.
Macrophage Infections and T6SS Activity-Infections were performed as previously described (15) using the C57BL/6 murine bone marrow-derived macrophage cell line ANA-1. Bacteria were added to ANA-1 cells grown on glass coverslips at a multiplicity of infection of 50. Coverslips were analyzed by phase contrast microscopy after 4 h of incubation at 37°C. T6SS activity was recorded as the ability of the bacteria to induce the formation of characteristic ectopic actin nucleation around the macrophages (15,22).
Western Blot Analysis-For E. coli, overnight bacterial cultures in 5 ml of LB were diluted to an initial A 600 of 0.2 and incubated at 37°C until reaching an A 600 of 0.7. At this point, isopropyl-␤-D-thiogalactopyranoside was added to a final concentration of 0.2 mM. Cells were incubated for 4 h at 30°C and then harvested by centrifugation at 8000 ϫ g for 10 min at 4°C. The bacterial pellet was suspended in lysis buffer, and the suspension was lysed using a One-shot cell disrupter (Thermo Scientific). Cell debris were removed by centrifugation (15,000 ϫ g for 15 min at 4°C), and the clear supernatant was centrifuged at 40,000 ϫ g for 30 min at 4°C. The pellet, containing total membranes, was suspended in lysis buffer. The protein concentration was determined by the Bradford assay (Bio-Rad). Staining was performed with Coomassie Brilliant Blue. SDS-PAGE, protein transfers to nitrocellulose membranes, and immunoblots were performed as described (33). For detection of His 6 -tagged proteins, membranes were incubated with a 1:10,000 dilution of anti-His IgG2a monoclonal antibodies (Amersham Biosciences) and Alexa Fluor 680 anti-mouse IgG antibodies (Molecular Probes). For B. cenocepacia, overnight bacterial cultures in 5 ml of LB were diluted to an initial A 600 of 0.2 and incubated at 37°C for 8 h. His-tagged proteins were purified and detected by Western blot using an anti-His antibody as indicated above.

Deletion of atsT Causes the Same Phenotypes as ⌬atsR
Mutants-We reasoned that if AtsT (BCAM0381) was involved in the AtsR phosphorelay pathway, ⌬atsT and ⌬atsR mutants should have similar phenotypes. Therefore, ⌬atsT was tested in the macrophage-infection model. Phase-contrast microscopy revealed that, in contrast to K56-2, ⌬atsR and ⌬atsT noticeably induce the formation of pearls on a string-like structures around infected macrophages (Fig. 1A, arrowheads). These structures depend on T6SS-mediated rearrangements of host actin and are characteristic of an up-regulated T6SS (15,16,22). B. cenocepacia ⌬atsR⌬hcp, a T6SS-defective mutant, was used as a negative control during the infections (34), and as expected, it did not mediate the pearls on a string phenotype in infected macrophages (Fig. 1A). Similarly to ⌬atsR (15), ⌬atsT showed increased biofilm formation as evidenced by a robust ring of biofilm at the air-liquid interface (Fig. 1B). Together, these results strongly suggest that AtsT and AtsR are in the same regulatory pathway, which is likely initiated by AtsR.
Identification of Phosphoacceptor Residues within AtsR-The region of AtsR corresponding to the HK domain (amino acids 233-457) was aligned with the HK domain of two well characterized hybrid sensor kinase proteins, ArcB from E. coli and RetS from Pseudomonas aeruginosa (3,(35)(36)(37). The amino acid sequence of AtsR (606 amino acids) is 34% identical and 50% similar to that of ArcB and 26% identical and 48% similar to that of RetS. The invariant His residue that is autophosphorylated in these proteins corresponds to His-245 in AtsR. Also, the AtsR-RD (amino acids 488 -601) showed 52 and 50% similarity at the amino acid level to ArcB and RetS, respectively. This region has an invariant Asp at position 536, which corresponds to the site of phosphorylation in ArcB and RetS.
Truncated versions of AtsR spanning AtsR-HK and AtsR-RD were constructed and purified to analyze their biochemical properties ( Fig. 2A). Residues 1-200, comprising the membrane-spanning domains, were removed to facilitate protein solubility and purification. Furthermore, to assess the contribution of conserved His and Asp residues to AtsR phosphorylation, His-245 and Asp-536 were individually replaced by alanine ( Fig. 2A). These proteins were assessed for in vitro autophosphorylation using [␥-33 P]ATP. Only polypeptides containing His-245 became autophosphorylated (Fig. 2B, lanes  1 and 3). Therefore, AtsR HK containing the native His-245 is necessary and sufficient for initiation of the phosphorelay path-way as the replacement of His with alanine at position 245 abolished autophosphorylation (Fig. 2B, lanes 2 and 6). Moreover, the AtsR-RD is unable to autophosphorylate in the absence of the HK domain (Fig. 2B, lane 4). The ability of AtsR-RD to accept the phosphoryl group from the AtsR-HK was tested in a phosphotransfer assay (Fig. 2B, lanes 5-7). Two bands were obtained after incubating AtsR-HK and AtsR-RD together in a reaction with [␥-33 P]ATP (Fig. 2B, lane 5). These bands corresponded to the phosphorylated form of these proteins, indicating the transfer of phosphoryl group from the HK domain to the RD domain of AtsR. To test whether Asp-536 within AtsR-RD is the residue accepting the phosphate from His-245, AtsR-RD D536A was added to the reaction with AtsR-HK and [␥-33 P]ATP. No band corresponding to AtsR-RD D536A was detected (Fig. 2B, lane 7), clearly indicating that Asp-536 is necessary for phosphorylation.
Chemical Stability of Phosphorylated Proteins-Phosphoramidates, such as histidine or lysine phosphate, are stable in alkali conditions but extremely labile to acid. Acyl phosphates such as aspartate and glutamate, on the other hand, are labile to either acid or alkali treatment (38). To compare the chemical stability of the phosphorylation of AtsR and AtsR D536A proteins, phosphorylation reactions were subjected to either no treatment or treatment with acid or alkali. If AtsR is phosphorylated at both His-245 and Asp-536, whereas AtsR D536A is only phosphorylated at His-245, the two phosphorylated proteins should behave differently upon treatment with base. Indeed,

FIGURE 1. T6SS activity and biofilm formation of B. cenocepacia K56-2 (WT) and its mutant derivatives.
A, phase-contrast microscopy of infected ANA-1 murine macrophages to assess T6SS activity. Infections were performed at a multiplicity of infection of 50 for 4 h. White arrowheads indicate the presence of ectopic actin nucleation (pearls-on-a-string phenotype (15,22)) extending from infected macrophages, which denotes T6SS activity. B. cenocepacia K56-2 ⌬atsR⌬hcp, a T6SS-defective mutant, was used as a negative control during the infections. Experiments consisted of three independent biological repeats where similar results were obtained. B, biofilm formation by parental strains, ⌬atsR and ⌬atsT mutants. B. cenocepacia K56-2 wild-type and derivative mutants were tested for biofilm formation by crystal violet staining. Arrows indicate the ring corresponding to the biofilm formation characteristic in ⌬atsR (15) and ⌬atsT mutants. The experiment was repeated three times in triplicate, and pictures were taken after 24 h of static incubation at 37°C.
AtsR and AtsR D536A were both labile to acid (Fig. 2C, lanes 5  and 6) and relatively stable to base (lanes 3 and 4), indicating that these two proteins contain amidylphosphates. The quantitative difference between phosphorylated AtsR before and after alkali treatment in Fig. 2C suggests that at least a portion of the wild-type protein may be phosphorylated at the Asp-536. In quantitative analysis by densitometry, the phosphorylated AtsR retained 82 Ϯ 6% of the label after alkali treatment compared with the untreated control, whereas AtsR D536A retained 91 Ϯ 5% of its label. These results suggest that the majority of the phosphate in both AtsR and AtsR D536A is in the amidylphosphate form.
AtsR and AtsT Form a Cognate HK-RR Pair-We initially hypothesized that BCAM0378 could be a histidine-phosphotransfer component of the AtsR phosphorelay based on the proximity of BCAM0378 to atsR (BCAM0379) and bioinformatics data indicating the presence of highly conserved His and Asp residues within this family of proteins. To investigate the relationship between AtsR, BCAM0378, and AtsT, we performed phosphotransfer assays employing combinations of these different proteins. The histidine-phosphotransfer candidate, BCAM0378, was not phosphorylated (Fig. 3, lanes 1-3).
To rule out the possibility of a rapid phosphotransfer from AtsR to BCAM0378, we repeated the experiment at 1 and 5 min incubation times and increased the temperature to 30°C. A phosphorylated form of BCAM0378 was never detected under any condition tested (data not shown). In contrast, AtsT was  phosphorylated directly by AtsR and independently from BCAM0378 (Fig. 3, lane 4). To test whether AtsT autophosphorylates independently from AtsR, we incubated AtsT with AtsR H245A and BCAM0378 (Fig. 3, lane 3). No phosphorylated band was detected, indicating that AtsT cannot autophosphorylate in vitro without the native AtsR-phosphorylated HK domain. Together, these data suggest that AtsR acts as a conventional two-component signal transduction system that mediates signal transduction from a sensor kinase to a response regulator, and BCAM0378 is not involved in the transfer of the phosphate from AtsR to the AtsT under the conditions tested. This conclusion was supported by the lack of the pearls-on-astring phenotype in macrophages infected with a ⌬BCAM0378 mutant (Fig. 1A), strongly suggesting that BCAM0378 is not part of the AtsR regulatory pathway.
Kinetics of AtsR Phosphorylation-A pulse-chase experiment was performed to determine the stability of AtsR phosphorylation in the presence of AtsT and to determine whether phosphorylation of AtsT can take place independently from AtsR in excess of ATP. If AtsT is phosphorylated from the free nucleotide pool, then an excess of unlabeled ATP should compete with hot ATP for AtsT phosphorylation in a pulse-chase reaction. Conversely, if AtsR phosphorylates AtsT, then the label should be chased from AtsR to AtsT. AtsR was incubated with [␥-33 P]ATP for 10 min, and an excess of unlabeled ATP (20 mM) and AtsT were simultaneously added to the reaction. Labeling of AtsT coincided with immediate loss of signal from phosphorylated AtsR (Fig. 4A). A decrease of phosphorylation of nearly 50% was observed between time 0 (preincubation) and 2 min (Fig. 4C). Phosphorylated AtsR was highly stable, and the signal remained relatively strong at 60 min. We repeated this experiment and chased phosphorylated proteins for up to 80 min, and phosphorylated bands were still visible (data not shown). To control the quality of unlabeled ATP, AtsR was incubated simultaneously with both labeled and unlabeled ATP for 10 min followed by the addition of AtsT (Fig. 4B). No labeling of AtsR or AtsT was detected in the control reaction, indicating that incorporation of ␥-33 P can be inhibited by an excess of unlabeled ATP, which competes with labeled ATP in the reaction. Thus, AtsR phosphorylation at His-245 is stable, and AtsT phosphorylation is resistant to competition from excess ATP, which confirms that AtsT acquires its phosphate directly from AtsR.
The Role of Asp-536 on AtsR-AtsT Phosphorylation Kinetics-To determine the effect of Asp-536 on phosphotransfer, AtsR and AtsR D536A were incubated with [␥-33 P]ATP for 10 min in individual reactions to make a pool of phosphorylated proteins followed by the addition of AtsT to each reaction. The rate of phosphate incorporation to the RR was followed as a function of time. The D536A replacement did not abolish the ability of AtsR to autophosphorylate, but the phosphotransfer capabilities of the parental and AtsR D536A proteins were quite different. The transfer of phosphate from AtsR to AtsT took place not only slower but also at a lower rate than from AtsR D536A to AtsT. Indeed, Ͼ90% of the phosphate was transferred from AtsR D536A to AtsT after 90 s, and the amount of label gained by AtsT was almost equal to that lost by AtsR D536A (Fig. 5, A and  B). The increase in AtsT phosphorylation correlates with the disappearance of the AtsR and AtsR D536A phosphorylation signals. In contrast to native AtsR, where the phosphorylation signal was maintained from 2 min until the end of the experiment, phosphorylation of AtsT by AtsR D536A was rapidly detected 1 min after the addition of RR to the reaction, and the signal of AtsR D536A had already disappeared after 2 min under the conditions tested (Fig. 5B). Together, these results indicate that replacing Asp-536 with alanine increases and accelerates the phosphotransfer reaction to the RR. It also might be due to the fact that AtsR has two sites of phosphorylation and only His-245 transfers phosphate to AtsT, and therefore, the persistence of the signal on wild-type protein is the result of Asp-536 phosphorylation.
In Vivo Reconstitution of the AtsR Signaling Pathway-B. cenocepacia secretes two zinc metalloproteases, ZmpA and ZmpB (17)(18)(19)(20)(21). Deletion of atsR up-regulates the expression of several quorum-sensing regulated virulence factors, including swarming motility and the secretion of zinc metalloproteases, both of which are also positively regulated by the CepI/CepR quorum-sensing system (16). Moreover, whereas a ⌬cepI mutant was protease-deficient at 48 h when spotted on D-BHI milk agar plates, further deletion of atsR in ⌬cepI resulted in increased proteolytic activity, demonstrating that deletion of atsR up-regulates protease activity independently of quorum sensing (16). We took advantage of the protease and swarming motility phenotypes to investigate whether AtsR phosphorylation is required for its function in vivo. The mutant strains ⌬atsR and ⌬atsR⌬cepI were complemented at the chromosomal level with full-length atsR or atsR variants encoding either AtsR H245A , AtsR D536A , or AtsR lacking the RD (AtsR⌬RD), eliminating the putative effect of other conserved Asp residues adjacent to Asp-536. The ability of AtsR H245A , AtsR D536A , and AtsR⌬RD to suppress the phenotypes of ⌬atsR and ⌬atsR⌬cepI mutant backgrounds was first assessed using protease secretion assay (Fig.  6A). As expected, the radius of halo corresponding to casein degradation surrounding the ⌬atsR spot was bigger compared with that of the wild type within 48 h of incubation, whereas no halo was present for ⌬cepI (Fig. 6, A and B). Proteolytic activity was also detectable for ⌬atsR⌬cepI to a lesser extent compared with WT and ⌬atsR (Fig. 6, A, top row, and B). Successful complementation was achieved when a chromosomal copy of the atsR gene was restored into ⌬atsR and in ⌬atsR⌬cepI, leading to decreased protease activity to wild-type and ⌬cepI levels, respectively (Fig. 6, A, middle and bottom row, and B).
Because Asp-536 significantly increases the phosphotransfer from His-245 to AtsT in vitro, one could have expected that complementation of ⌬atsR with AtsR D536A would have a stronger inhibitory effect than atsR in vivo. However, similar results were obtained when either atsR⌬RD or atsR D536A was introduced to ⌬atsR or ⌬atsR⌬cepI, respectively. This suggests that Asp-D536 does not have a strong modulatory effect on the AtsR-AtsT phosphotransfer in vivo under the conditions tested or, alternatively, that additional components not identified in this study may be involved in modulating the AtsR-AtsT phosphorelay pathway.
In contrast, the phosphorylation status of AtsR is critical for its role in expression and/or secretion of proteases, as complementation failed when atsR H245A was introduced. These strains were also tested in a swarming motility assay. In agreement with previous results, only complementation of ⌬atsR⌬cepI strains with atsR or atsR⌬RD, but not atsR H245A , could restore swarming motility to ⌬cepI levels (Fig. 7). Protein expression was con- firmed by Western blot analysis of bacterial cell lysates prepared from K56-2 ⌬atsR and ⌬atsR⌬cepI complemented with atsR H245A , atsR D536A , atsR⌬RD, and atsR and demonstrated that the encoded AtsR variants were similarly expressed (Fig.  6C). Thus, a lack of complementation by atsR H245A was not due to a defect in protein expression. Together, these data confirm the in vitro results and suggest that His-245 is essential for the function of AtsR in vivo.
To confirm that AtsT is the cognate response regulator, which negatively controls the expression of protease activity when it is phosphorylated, atsT was deleted in ⌬atsR, ⌬atsR⌬cepI, and ⌬cepI strains, which were complemented with atsR D536A , atsR⌬RD, atsT, or atsT D208A . The resulting strains were spotted on D-BHI-milk agar plates (Fig. 8, A and B), and the proteolysis was quantified by measuring the radius of clearing around the colonies (Fig. 8C). Although complementation  of ⌬atsR and ⌬atsR⌬cepI strains with atsR D536A or atsR⌬RD reduced the proteolytic activity to WT and ⌬cepI levels (Fig. 6), further deletion of atsT bypassed the atsR D536A or atsR⌬RD complementation and resulted in an increase of proteolytic activity. Furthermore, ⌬cepI⌬atsT has the same phenotype as ⌬atsR⌬cepI, and as expected, by complementing ⌬cepI⌬atsT with atsT, proteolytic activity decreases to ⌬cepI level, whereas atsT D208A was unable to complement. These results suggest that (i) AtsT indeed acts as a negative regulator and (ii) is a direct target of AtsR contributing to the regulatory role of this protein on proteolytic activity, and (iii) Asp-208 on AtsT is required for its function.

DISCUSSION
The predicted structural features of AtsR suggested this protein could not directly control gene expression because it lacks an effector domain. We demonstrate in this study three key properties of AtsR: (i) upon autophosphorylation, AtsR transfers the phosphate to the response regulator AtsT without the participation of an intermediate histidine-phosphotransfer protein; (ii) AtsR function in vitro and in vivo depends on autophosphorylation of the His-245 residue, which is absolutely essential for initiation of signal transduction; (iii) the AtsR-RD and more specifically the Asp-536 to some extent plays a role in modulating the stability of phosphorylated AtsR.
Other studies have shown that the sequential phosphotransfer between residues within the same hybrid sensor kinase modulates the phosphotransfer to the cytosolic response regulator and the overall response by determining the specificity for the cognate RR or by regulating the autokinase activity (2, 3, 39 -42). The stability of the phosphorylated AtsR and AtsT determined by the pulse-chase kinetic experiments is remark- able as these proteins can maintain the response memory for at least 60 min, which agrees with their physiological function as negative regulators of gene expression. Comparative studies have suggested that the in vitro assay of HK autophosphorylation reflects, to a certain extent, the in vivo situation (43). For example, ArcB, NarQ, and NtrB, with a high rate of phosphotransfer but a low level of autophosphorylation, are able to respond quickly to changes in the environment thereby returning to the steady-state levels after transient activation or repression by external stresses, whereas CheA and BaeS showed high levels of self-phosphorylation even though they had a slow signal transduction rate (43). In the case of AtsR, we find that the phosphate is transferred very quickly within 2 min from AtsR to AtsT, and both proteins are capable of prolonged phosphorylation. This may indicate that unlike some other systems (44,45), AtsR might not have phosphatase activity, although further experiments are needed to validate this notion.
Based on the consistency of results obtained by in vitro and in vivo experimental approaches in this study, we propose a model ( Fig. 9) in which AtsR autophosphorylates in vivo at His-245. The phosphate is then transferred to AtsT and partly to the AtsR receiver domain on Asp-536, as stable phosphorylation of the AtsR receiver domain was detected in vitro. Based on these findings, phosphorylation of AtsT in vitro correlates with repression of B. cenocepacia gene expression in vivo. Our data strongly support the notion that AtsR phosphorylation has a significant biological relevance as a global virulence regulator modulating the expression of proteases through AtsT. Experiments are under way to identify the genes that are specifically controlled by AtsT as well as the environmental signals that trigger activation or inactivation of the AtsR/AtsT phosphorelay.