Molecular characterization of the PhoPQ-PmrD-PmrAB mediated pathway regulating polymyxin B resistance in Klebsiella pneumoniae CG43

Background The cationic peptide antibiotic polymyxin has recently been reevaluated in the treatment of severe infections caused by gram negative bacteria. Methods In this study, the genetic determinants for capsular polysaccharide level and lipopolysaccharide modification involved in polymyxin B resistance of the opportunistic pathogen Klebsiella pneumoniae were characterized. The expressional control of the genes responsible for the resistance was assessed by a LacZ reporter system. The PmrD connector-mediated regulation for the expression of pmr genes involved in polymyxin B resistance was also demonstrated by DNA EMSA, two-hybrid analysis and in vitro phosphor-transfer assay. Results Deletion of the rcsB, which encoded an activator for the production of capsular polysaccharide, had a minor effect on K. pneumoniae resistance to polymyxin B. On the other hand, deletion of ugd or pmrF gene resulted in a drastic reduction of the resistance. The polymyxin B resistance was shown to be regulated by the two-component response regulators PhoP and PmrA at low magnesium and high iron, respectively. Similar to the control identified in Salmonella, expression of pmrD in K. pneumoniae was dependent on PhoP, the activated PmrD would then bind to PmrA to prolong the phosphorylation state of the PmrA, and eventually turn on the expression of pmr for the resistance to polymyxin B. Conclusions The study reports a role of the capsular polysaccharide level and the pmr genes for K. pneumoniae resistance to polymyxin B. The PmrD connector-mediated pathway in governing the regulation of pmr expression was demonstrated. In comparison to the pmr regulation in Salmonella, PhoP in K. pneumoniae plays a major regulatory role in polymyxin B resistance.


Background
Klebsiella pneumoniae, an important nosocomial pathogen, causes a wide range of infections including pneumonia, bacteremia, urinary tract infection, and sometimes even life-threatening septic shock [1]. The emergence of multi-drug resistant K. pneumoniae has reduced the efficacy of antibiotic treatments and prompted the reevaluation of previously but not currently applied antibiotics [2,3] or a combined therapy [4]. Polymyxins, originally isolated from Bacillus polymyxa, have emerged as promising candidates for the treatment of infections [5]. As a member of antimicrobial peptides (APs), the bactericidal agent exerts its effects by interacting with the lipopolysaccharide (LPS) of gram-negative bacteria. The polycationic peptide ring on polymyxin competes for and substitutes the calcium and magnesium bridges that stabilize LPS, thus disrupting the integrity of the outer membrane leading to cell death [5,6].
The Klebsiella capsular polysaccharide (CPS), which enabled the organism to escape from complementmediated serum killing and phagocytosis [7,8], has been shown to physically hinder the binding of C3 complement [9] or polymyxin B [10]. The assembly and transport of Klebsiella CPS followed the E. coli Wzydependent pathway [11], in which mutations at wza encoding the translocon protein forming the complex responsible for CPS polymer translocation and export resulted in an inability to assemble a capsular layer on the cell surface [12]. The CPS biosynthesis in K. pneumoniae was transcriptionally regulated by the two-component system (2CS) RcsBCD [13] where the deletion of the response regulator encoding gene rcsB in K. pneumoniae caused a loss of mucoid phenotype and reduction in CPS production [14].
In Escherichia coli and Salmonella enterica serovar Typhimurium, polymyxin B resistance is achieved mainly through the expression of LPS modification enzymes, including PmrC, an aminotransferase for the decoration of the LPS with phosphoethanolamine [15] and the pmrHFIJKLM operon [16,17] (also called pbgP or arn operon [18,19]) encoding enzymes. Mutations at pmrF, which encoded a transferase for the addition of 4-aminoarabinose on bactoprenol phosphate, rendered S. enterica and Yersinia pseudotuberculosis more susceptible to polymyxin B [16,20]. The S. enterica ugd gene encodes an enzyme responsible for the supply of the amino sugar precursor L -aminoarabinose for LPS modifications and hence the Ugd activity is essential for the resistance to polymyxin B [21]. On the other hand, the E. coli ugd mutant with an impaired capsule also became highly susceptible to polymyxin B [22].
The 2CS PmrA/PmrB, consisting of the response regulator PmrA and its cognate sensor kinase PmrB, has been identified as a major regulatory system in polymyxin B resistance [23,24]. The resistance in S. enterica or E. coli has been shown to be inducible by the extracellular iron [25]. In addition to acidic pH [26], the role of ferric ions as a triggering signal for the expression of PmrA/PmrB has been demonstrated [23]. The 2CS PhoP/PhoQ which regulates the magnesium regulon [27] could also activate polymyxin B resistance under low magnesium in S. enterica, in which the PhoP/ PhoPQ-dependent control is connected by the small basic protein PmrD. The expression of pmrD could be activated by PhoP while repressed by PmrA forming a feedback loop [28,29]. The activated PmrD could then bind to the phosphorylated PmrA leading to a persistent expression of the PmrA-activated genes [30].
The PmrD encoding gene was also identified in E. coli and K. pneumoniae. However, pmrD deletion in E. coli had no effect on the bacterial susceptibility to polymyxin B [25]. Recently, the PhoP-dependent expression of pmrD has also been demonstrated in K. pneumoniae. According to the predicted semi-conserved PhoP box in the pmrD upstream region, a direct binding of PhoP to the pmrD promoter for the regulation was speculated [31].
In this study, specific deletions of genetic loci involved in CPS biosynthesis and LPS modifications were introduced into K. pneumoniae CG43, a highly virulent clinical isolate of K2 serotype [32]. Involvement of the genetic determinants in polymyxin B resistance was investigated.

Construction of specific gene-deletion mutants
Specific gene deletion was individually introduced into the chromosome of K. pneumoniae CG43S3 by allelic exchange strategy [14]. In brief, two approximately 1000-bp DNA fragments flanking both sides of the deleted region were cloned into the suicide vector pKAS46 [34]. The resulting plasmid was then mobilized from E. coli S17-1 λpir [34] to K. pneumoniae CG43S3, K. pneumoniae CG43S3ΔlacZ [35], or K. pneumoniae CG43S3ΔrcsB [14], by conjugation. The transconjugants were selected with ampicillin and kanamycin on M9 agar plates. Colonies were grown overnight in LB broth at 37°C and then spread onto an LB agar plate containing 500 μg/ml of streptomycin. The streptomycin-resistant and kanamycin-sensitive colonies were selected, and the deletion was verified by PCR and Southern analysis using gene-specific probe. The resulting K. pneumoniae mutants are listed Table 1.

Extraction and quantification of CPS
Bacterial CPS was extracted using the method described [37]. Briefly, 500 μl of overnight culture was mixed with 100 μl of 1% Zwittergent 3-14 (Sigma-Aldrich) in 100 mM citric acid (pH 2.0) and incubated at 50°C for 20 min. After centrifugation, 250 μl of the supernatant was used to precipitate CPS with 1 ml of absolute ethanol. The pellet was dissolved in 200 μl distilled water, and then 1,200 μl of 12.5 mM borax in H 2 SO 4 was added. The mixture was vigorously mixed, boiled for 5 min, cooled, and then 20 μl 0.15% 3-hydroxydiphenol (Sigma-Aldrich) was added. OD 520 was measured and the uronic acid content was determined from a standard curve of glucuronic acid and expressed as μg per 10 9 CFU.

Polymyxin B resistance assay
Polymyxin B resistance assay was performed essentially as described [10] with some modifications. In brief, the overnight-grown K. pneumoniae strains were washed twice with saline (0.85% NaCl solution, w/v) and subcultured in LB broth alone or supplemented with 1 mM FeCl 3 or with 10 mM MgCl 2 at 37°C. The log-phased (OD 600 of 0.7) bacterial culture was then washed twice and a suspension containing ca. 2.5 × 10 4 CFU/ml in LB was prepared. Then, 100 μl of the suspension was placed in each well of a 96-well micro-titer plate and 100 μl PBS or PBS-diluted polymyxin B was added to each well to final concentrations of 0, 1, 2, or 4 units/ml of polymyxin B. The plate was incubated at 37°C for 1 h with shaking. Subsequently, 100 μl of the suspension was directly plated on LB agar plates and incubated at 37°C overnight to determine the number of viable bacteria. The survival rates were expressed as colony counts divided by the number of the same culture treated with PBS and multiplied by 100. The assays were performed thrice, and the results were shown as the average ± standard deviation from triplicate samples. The survival rates at 1 and 2 units/ml ( Figure 1C) and at 2 units ( Figure 2A and Figure 3AB) of polymyxin B were shown.

Cell line, cell culture and phagocytosis assay
The mouse macrophage cell line RAW264.7 was cultivated in Dulbecco's Modified Eagle Medium (DMEM) (Gibco) supplemented with 10% fetal bovine serum (Gibco), 100 units/ml of penicillin and 100 μg/ml of streptomycin (Gibco) at 37°C under 5% CO 2 . The evaluation of bacterial phagocytosis was carried out as described with some modifications [9]. In brief, cells were washed, resuspended in DMEM containing 10% FBS, and approximately 10 6 cells per well were seeded in a 24 well tissue culture plate and incubated at 37°C for 16 h. Then 100 μl of the bacterial suspension (approximately 3 × 10 8 CFU/ml in PBS) was used to infect each well to obtain a ratio of ca. 30 bacteria per macrophage. After incubation for 2 h, the cells were washed thrice, then 1 ml of DMEM containing 100 μg/ ml of gentamycin was added and incubated for another 2 h to kill the extracellular bacteria. Cells were washed thrice, 1 ml of 0.1% Triton X-100 was added and incubated at room temperature for 10 min with gentle shaking to disrupt the cell membrane. The cell lysate was  HindIII +1095 relative to the pmrB start codon NotI +1 relative to the pmrD start codon diluted serially with PBS, plated onto LB agar plates and incubated overnight for determining viable bacteria count. The relative survival rates after phagocytosis were expressed as the colony counts of viable bacteria divided by those of the original inoculums and multiplied by 100. Three independent trials were performed, and the data shown were the average ± standard deviation from five replicas.

Construction of reporter fusion plasmid and measurement of promoter activity
The approximately 350 or 500-bp DNA fragments containing the upstream region of the K. pneumoniae pmrD or pmrHFIJKLM gene cluster were PCR-amplified with primers pmrDp01/pmrDp02 or pmrHp01/pmrHp02 (Table 2), respectively and cloned in front of a promoter-less lacZ gene of the promoter selection plasmid placZ15 [35]. The resulting plasmids, placZ15-PpmrD and placZ15-PpmrH were mobilized from E. coli S17-1 λpir to K. pneumoniae strains by conjugation. β-galactosidase activity was determined as previously described [35]. In brief, overnight cultures were washed twice with saline and subcultured in LB alone or supplemented with 10 mM MgCl 2 , 0.1 mM FeCl 3 , or 0.1 mM FeCl 3 plus 0.3 mM ferric iron scavenger deferoxamine (Sigma-Aldrich) to mid-log phase (OD 600 of 0.7). Then 100 μl of the culture was mixed with 900 μl of Z buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl, 1 mM MgSO 4 , 50 mM β-mercaptoethanol), 17 μl of 0.1% SDS, and 35 μl of chloroform and the mixture was shaken vigorously. After incubation at 30°C for 10 min, 200 μl of 4 mg/ml ONPG (o-nitrophenyl-β-D-galactopyranoside) (Sigma-Aldrich) was added. Upon the appearance of yellow color, the reaction was stopped by adding 500 μl 1 M Na 2 CO 3 . OD 420 was recorded and the β-galactosidase activity was expressed as Miller units [38]. Each sample was assayed in triplicate, and at least three independent experiments were carried out. The data shown were calculated from one representative experiment and shown as the means and standard deviation from triplicate samples.

Cloning, expression and purification of recombinant proteins
The DNA fragment of PhoP coding region was PCR amplified from the genomic DNA of K. pneumoniae CG43S3 with primers phoP05/phoP06 (  site on pET30b. Plasmids pET-PmrA and pET-PmrD (courtesy of Dr. Chinpan Chen, Academia Sinica, Taipei, Taiwan) were constructed by cloning DNA fragments PCR-amplified with KP1760-1/KP1760-2 and KP3573-1/ KP3573-2 ( Table 2) into an NdeI/XhoI site, respectively into pET29b. The resulting plasmids were transformed into E. coli BL21(DE3) (Invitrogen, USA), and the recombinant proteins were over-expressed by induction with 0.5 mM isopropyl 1-thio-β-D-galactopyranoside (IPTG) for 3 h at 37°C. The proteins were then purified from total cell lysate by affinity chromatography using His-Bind resin (Novagen, Madison, Wis). After purification, the eluent was dialyzed against 1× protein storage buffer (10 mM Tris-HCl pH 7.5, 138 mM NaCl, 2.7 mM KCl, and 10% glycerol) at 4°C overnight, followed by condensation with PEG20000, and the purity was determined by SDS-PAGE analysis.
DNA electrophoretic mobility shift assay (EMSA) EMSA was performed as previously described [14]. In brief, the DNA fragment encompassing the putative pmrD promoter region was obtained by PCR amplification and then end-labeled with [γ-32 P]ATP by T4 polynucleotide kinase. The purified His-PhoP or His-PhoP N149 protein was mixed with the DNA probe in a 50-μl reaction mixture containing 20 mM Tris-HCl pH 8.0, 50 mM KCl, 1 mM MgCl 2 , 1 mM dithiothreitol, and 7.5 mM acetyl phosphate. The mixture was incubated at room temperature for 30 min, mixed with 0.1 volume of DNA loading dye, and then loaded onto a 5% nondenaturing polyacrylamide gel containing 5% glycerol in 0.5× TBE buffer (45 mM Tris-HCl pH 8.0, 45 mM boric acid, 1.0 mM EDTA). After electrophoresis at a constant current of 20 mA at 4°C, the result was detected by autoradiography.

In vitro phosphotransfer assay
The in vitro phosphotransfer assay was performed essentially as described [30]. The phospho-PmrB C276 protein was obtained by pre-incubation of His-PmrB C276 protein (5 μM) with 40 μCi of [γ-32P]ATP in 80 μl of 1× phosphorylation buffer (10 mM Tris-HCl, pH 7.5; 138 mM NaCl; 2.7 mM KCl; 1 mM MgCl 2 ; 1 mM DTT) for 1 h at room temperature. The reaction mixture was then chilled on ice, and 5 μl of the mixture was removed and mixed with 2.5 μl of 5× SDS sample buffer as a reference sample. The phospho-PmrB C276 protein mixture (30 μl) was then mixed with equal volumes of 1× phosphorylation buffer containing either PmrA (10 μM) or PmrA with PmrD (each at 10 μM) to initiate the phosphotransfer reaction. A 10-μl aliquot was removed at specific time points, mixed with 2.5 μl of 5× SDS sample buffer to stop the reaction, and the samples were kept on ice until the performance of SDS-PAGE. After electrophoresis at 4°C, the signal was detected by autoradiography.

Kinase/phosphatase and autokinase assay
The assays were performed essentially as described [30]. The recombinant protein His-PmrB C276 (2.5 μM) was incubated with His-PmrA (5 μM) alone or with His-PmrD (5 μM) for kinase/phosphatase assay or incubated with His-PmrD (5 μM) alone for autokinase assay. The reactions were carried out in 30 μl of 1× phosphorylation buffer with 3.75 μCi [γ-32 P]ATP at room temperature and started with the addition of His-PmrB C276 . An aliquot of 10-μl was removed at specific time points, mixed with 5× SDS sample buffer to stop the reaction, and the samples were kept on ice until the performance of SDS-PAGE. After electrophoresis at 4°C, the signal was detected by autoradiography.

Statistical analysis
Student's t test was used to determine the significance of the differences between the CPS amounts and the levels of β-galactosidase activity. P values less than 0.01 were considered statistically significant.

Results
Reduced production of capsular polysaccharide had minor effect on polymyxin B resistance in K. pneumoniae K. pneumoniae CG43 is a highly encapsulated virulent strain [32]. In order to verify the role of CPS in polymyxin B resistance, the Δugd and Δwza mutants were generated by allelic exchange strategy, and their phenotype as well as the amount of CPS produced were compared with the parental strain CG43S3 and ΔrcsB mutant [14]. As shown in Figure 1A, the Δugd and Δwza mutants formed apparently smaller colonies on LB agar plate compared with the glistering colony of the parental strain CG43S3. Although the colony morphology of the ΔrcsB mutant was indistinguishable from CG43S3, the CPS-deficient phenotype was evident as assessed using sedimentation assay and the amount of K2 CPS produced ( Figure 1B). Deletion of rcsB resulted in an approximately 50% reduction of the CPS, while the Δwza mutant produced less than 20% of that of its parental strain CG43S3. The CPS biosynthesis in Δugd mutant was almost abolished, indicating an indispensible role of Ugd in CPS biosynthesis. To investigate how the CPS level was associated with polymyxin B resistance, the survival rates of the strains challenged with polymyxin B were compared. The Δugd mutant producing the lowest amount of CPS was extremely sensitive to the treatment of polymyxin B ( Figure 1C). Although the Δugd mutant was CPS-deficient, the impaired polymyxin resistance may have been largely attributed to the defect in LPS biosynthesis since the survival rates of Δwza and ΔrcsB mutants appeared to be comparable with the parental strain CG43S3. This argues against the notion that the level of polymyxin B resistance is positively correlated to the amount of CPS [10]. Nevertheless, the possibility that a higher amount of CPS was required for the resistance could not be ruled out. As shown in Figure 1D, the introduction of pRK415-RcsB [39] resulted in a significantly higher resistance to polymyxin B in both ΔrcsB mutant and its parental strain. This indicated a protective effect of large amounts of CPS in polymyxin resistance.

PmrF is involved in polymyxin B resistance and survival within macrophage
To investigate if the K. pneumoniae pmr homologues played a role in polymyxin B resistance, a pmrF deletion mutant strain and a plasmid pRK415-PmrF were generated. As shown in Figure 2A, when the strains were grown in LB medium, a low magnesium condition [40], differences in the survival rates were not apparent. When the strains were grown in LB supplemented with 1 mM FeCl 3 , an apparent deleting effect of pmrF in polymyxin B resistance was observed, and the survival rate could be restored by the introduction of pRK415-PmrF. The results indicated a role of PmrF in the polymyxin B resistance in high iron condition. In addition to the mucosa surfaces, antimicrobial peptides and proteins play important roles in the microbicidal activity of phagosome [41]. To investigate the effect of pmrF deletion in the bacterial survival within phagosome, phagocytosis assay was carried out. Since K. pneumoniae CG43S3 was highly resistant to engulfment by phagocytes in our initial experiments, the ΔrcsB mutant which produced less CPS was used as the parental strain to generate ΔpmrFΔrcsB mutant. As shown in Figure  2B, deletion of pmrF resulted in an approximately fourfold reduction in the recovery rate, which was restored after the introduction of pRK415-PmrF. This indicated an important role of pmrF not only in polymyxin B resistance but also in bacterial survival within macrophage.
Deletion effect of pmrA, pmrD or phoP on polymyxin B resistance in K. pneumoniae To investigate how PmrA, PhoP and PmrD were involved in the regulation of polymyxin B resistance in K. pneumoniae, ΔpmrA, ΔphoP and ΔpmrD mutant strains were generated. Deletion of either one of these genes resulted in a dramatic reduction of resistance to polymyxin B when the strains were grown in LB medium ( Figure 3A). The deleting effects were no longer observed when the strains grown in LB supplemented with 10 mM magnesium, implying an involvement of the PhoP-dependent regulation in LB, a low magnesium environment. Under high-iron conditions, the deletion of pmrA caused the greatest reduction in the survival rate. Introduction of pRK415-PmrA or pRK415-PhoP into the ΔpmrAΔphoP double mutant strain not only restored but also enhanced the bacterial resistance to polymyxin B (Figure 3B), which is likely due to an overexpression level of phoP or pmrA by the multicopy plasmid. Finally, whether the deletion of pmrA, phoP or pmrD affected the survival rate in phagosomes was also investigated. Interestingly, deletion of phoP resulted in most apparent effect while the pmrA deletion had less effect on the bacterial survival in macrophages. This was probably due to low iron concentration in the phagosomes [40]. The introduction of pRK415-PhoP or pRK415-PmrD could restore the recovery rates of ΔphoPΔrcsB and ΔpmrDΔrcsB, although not to the extent displayed by the parental strain. Taken together, our results indicate the presence of two independent pathways in the regulation of polymyxin B resistance and the bacterial survival within macrophage phagosomes.
Effect of pmrA, phoP or pmrD deletion on P pmrH ::lacZ or P pmrD ::lacZ activity As the functional role of the structural gene pmrF and the regulator genes phoP, pmrD and pmrA was verified, it would be of importance to investigate the regulatory network govern by PhoPQ-PmrD-PmrAB on the expression of pmr genes. Sequence analysis has revealed PhoP and PmrA box consensus in the upstream region of pmrH and PhoP box consensus in the upstream region of pmrD (Figure 4A). To investigate the interplay of PhoP, PmrA, and PmrD on the expression of pmr and pmrD genes, the reporter plasmids placZ15-PpmrH and placZ15-PpmrD were constructed and mobilized into K. pneumoniae CG43S3ΔlacZ and its derived ΔpmrAΔlacZ, ΔpmrDΔlacZ or ΔphoPΔlacZ isogenic strains, respectively. The β-galactosidase activities of K. pneumoniae transformants under different environmental conditions were determined. In the wild-type strain CG43S3ΔlacZ, the P pmrH ::lacZ activity was repressed in the presence of high magnesium but enhanced in high ferric ion ( Figure 4B). Such iron-inducible activity was abolished after the addition of iron scavenger deferoxamine. As shown in Figure 4B, deleting effect of pmrA or phoP on the activity of P pmrH ::lacZ could be observed in LB or LB supplemented with ferric iron. The negative effect of pmrD deletion was also apparent at high iron condition but was abolished after the addition of deferoxamine. The results clearly demonstrate the involvement of PmrA, PhoP and PmrD in the regulation of the expression of pmr genes, particularly in the presence of high ferric irons. As shown in Figure 4C, the P pmrD ::lacZ activity was significantly reduced in high-magnesium conditions or upon the deletion of phoP. Interestingly, the deletion of pmrA or high ferric irons had little effect on the activity of P pmrD ::lacZ. The results suggest that the expression of K. pneumoniae pmrD is regulated in a PhoP-dependent but PmrA-independent manner.

Analysis of EMSA indicates a direct binding of the recombinant PhoP to pmrD
The binding of PhoP or PmrA to P pmrH has been determined recently [31]. In order to determine whether PhoP binds directly to P pmrD , EMSA was performed. As shown in Figure 5A, binding of the recombinant His-PhoP protein to P pmrD was evident by the formation of a protein/DNA complex with a slower mobility. The binding specificity was also examined by the addition of specific DNA competitor or non-specific DNA competitor. As shown in Figure 5B, the formation of protein/ DNA complex diminished when His-PhoP N149 , in which the carboxyl-terminal helix-turn-helix domain has been truncated, was used instead of His-PhoP. The results strongly suggest the PhoP binds via its C-terminal domain to the promoter of pmrD for the activation of the pmrD expression in K. pneumoniae.

Two-hybrid analysis of the in vivo interaction between Klebsiella PmrD and PmrA
The interaction between Klebsiella PmrD and PmrA has been shown as a prerequisite for the connectormediated pathway [31]. To demonstrate in vivo interaction, a bacterial two-hybrid assay was performed. The plasmid pBT-PmrA carrying the RNAPα-PmrA coding region and the plasmid pTRG-PmrD carrying the λ-cI-PmrD coding sequence were generated. In vivo interaction between the two reporter strains allowed the binding of λ-cI to the operator region as well as the recruitment of α-RNAP for the expression of the ampR and lacZ reporter genes. The bacteria harboring the positive control plasmids pTRG-Gal11 P /pBT-LGF2 showed a more vigorous growth on the indicator plate, as reflected by the apparent colony formation when the culture was diluted serially ( Figure 6A). In contrast, the strain carrying the negative control vectors pBT/pTRG revealed impaired colony formation. As shown in Figure  6A, a similar growth pattern of the E. coli cells harboring pBT-PmrA/pTRG-PmrD to that of the positive control cells was observed indicating an interaction between the PmrD and PmrA.
The PmrD binds to PmrA to prevent dephosphorylation In S. enterica, the phosphorylation of PmrA by the cognate sensor protein PmrB has been demonstrated to enhance its affinity in binding to its target promoter. The subsequent dephosphorylation of PmrA by PmrB helped to relieve from over-activation of this system (1). In Salmonella, PmrD has been shown to be able to protect PmrA from both intrinsic and PmrB-mediated dephosphorylation (22). To verify if Klebsiella PmrD also participates in the phosphorylation, in vitro phosphotransfer assay was carried out with the recombinant proteins His-PmrA, His-PmrD and His-PmrB C276 . As shown in Figure 6B, the His-PmrA was rapidly phosphorylated upon addition of the autophosphorylated His-PmrB C276 and then gradually dephosphorylated. Addition of His-PmrD apparently prolong the phosphorylation state of the His-PmrA, which could be maintained for at least 60 min ( Figure 6B). The phosphorylated His-PmrA appeared to be very stable in the presence of the His-PmrD since the phosphorylation signal was still detectable 4 h later (data not shown). As shown in Figure 6C, the specificity of the interaction between His-PmrD and His-PmrA was also demonstrated since the phosphorylation state of His-PmrA could not be detected when incubated with the small cationic proteins RNase A or cytochrome C [30]. Similar levels of phospho-PmrB C276 were observed in the presence or absence of PmrD ( Figure 6D), suggesting the His-PmrD had no effect on the phorphorylation state of His-PmrB C276 .

Discussion
Although the amount of CPS produced by ΔrcsB mutant was more than twice of that produced by Δwza mutant, no apparent difference between the wild type strain CG43S3, Δwza mutant, and ΔrcsB mutant in polymyxin B resistance could be observed. This is different from the previous finding that K. pneumoniae CPS was an important physical barrier for the APs [10]. This discrepancy may be attributed to some of the K. pneumoniae strains used for comparison in the previous study produced extremely low level of the CPS. Nevertheless, a higher amount of CPS was protective for the bacterial resistance to polymyxin B.
On the other hand, the deletion of ugd resulted in the loss of resistance to polymyxin B. Sequence analysis of the available K. pneumoniae genome NTHU-K2044 [42], MGH78578 (http://genome.wustl.edu/) and 342 [43] revealed no PmrA [17] or PhoP box [27] in the upstream region of the manC-manB-ugd genes [44]. This implies the involvement of a regulatory mechanism different from that for S. enterica ugd, which was positively regulated by the three 2CS regulators PhoP, PmrA and RcsB [45].
Consistent with the reported findings [31], deletion of Klebsiella pmrF which encodes one of the enzymes required for synthesis and incorporation of aminoarabinose in LPS resulted in decreased resistance to polymyxin B and survival within macrophages. The pmr expression has been shown to be directly regulated by PhoP under low magnesium or by PmrA in high ferric ions, or by the connector-mediated pathway reported for Salmonella, [31]. Similar to the observations in E. coli, S. enterica [25], Yersinia pestis [46], and Pseudomonas aeruginosa [47], a positive regulatory role of PmrA and PhoP in polymyxin B resistance in K. pneumoniae was also demonstrated.
The deletion of phoP resulted in more drastic effect on the bacterial survival in macrophage than the pmrA deletion, implying a different level of control between PhoP and PmrA in K. pneumoniae resistance to phagocytosis. During phagocytosis, phagosomal maturation and phagolysosomes formation are accompanied by progressive acidification and acquisition of various hydrolases, reactive oxygen, nitrogen species, and APs [48]. Low pH and low-magnesium have been shown to be able to stimulate expression of the PhoP-activated genes [40,49]. Apart from its microbicidal activity, the APs inside phagosomes has even been reported as an inducing signal for the activation of the PhoP/PhoQ system [50]. The deletion of pmrF or phoP caused a significant reduction in intramacrophage survival of the bacterial, implying a role of the AP resistance regulation in the bacterial pathogenesis.
Until now, PmrD was only found in E. coli, Shigella flexneri, S. enterica and K. pneumoniae. Although PmrD in Klebsiella appeared to act in a way similar to the PmrD in S. enterica, they share only about 40% sequence identity. The expression of K. pneumoniae pmrD was shown to be PhoP-dependent and the regulation was achieved through a direct binding of PhoP to the putative pmrD promoter. In addition, the binding of PmrD was shown to efficiently protect the PmrA from dephosphorylation. The in vivo interaction between PmrD and PmrA demonstrated using 2-hybrid analysis further supported the presence of the connectormediated pathway in K. pneumoniae.
In summary, involvement of Klebsiella pmr in polymyxin B resistance and the regulation for the expression of pmr genes were analyzed. The regulatory network for the expression of the pmr genes is comprised of 2CS response regulators PhoP and PmrA, and the connector protein PmrD. The demonstration of PmrD in prolonging the phosphorylation state of phosphor-PmrA further confirmed the presence of a connector-mediated pathway in K. pneumoniae. The complexity in the control of pmr genes expression may provide ecological niches for K. pneumoniae in response to a variety of environmental clues; for example, in the process of infection.