YjcC, a c-di-GMP Phosphodiesterase Protein, Regulates the Oxidative Stress Response and Virulence of Klebsiella pneumoniae CG43

This study shows that the expression of yjcC, an in vivo expression (IVE) gene, and the stress response regulatory genes soxR, soxS, and rpoS are paraquat inducible in Klebsiella pneumoniae CG43. The deletion of rpoS or soxRS decreased yjcC expression, implying an RpoS- or SoxRS-dependent control. After paraquat or H2O2 treatment, the deletion of yjcC reduced bacterial survival. These effects could be complemented by introducing the ΔyjcC mutant with the YjcC-expression plasmid pJR1. The recombinant protein containing only the YjcC-EAL domain exhibited phosphodiesterase (PDE) activity; overexpression of yjcC has lower levels of cyclic di-GMP. The yjcC deletion mutant also exhibited increased reactive oxygen species (ROS) formation, oxidation damage, and oxidative stress scavenging activity. In addition, the yjcC deletion reduced capsular polysaccharide production in the bacteria, but increased the LD50 in mice, biofilm formation, and type 3 fimbriae major pilin MrkA production. Finally, a comparative transcriptome analysis showed 34 upregulated and 29 downregulated genes with the increased production of YjcC. The activated gene products include glutaredoxin I, thioredoxin, heat shock proteins, chaperone, and MrkHI, and proteins for energy metabolism (transporters, cell surface structure, and transcriptional regulation). In conclusion, the results of this study suggest that YjcC positively regulates the oxidative stress response and mouse virulence but negatively affects the biofilm formation and type 3 fimbriae expression by altering the c-di-GMP levels after receiving oxidative stress signaling inputs.

Klebsiella pneumoniae pyogenic liver abscess isolates often carry heavy capsular polysaccharides (CPS) to avoid phagocytosis or death by serum factors [22,23]. This thick and viscous structure also helps regulate the bacterial colonization and biofilm formation at the infection site [24]. Several regulators, such as RcsB, RmpA, RmpA2, KvhR, KvgA, and KvhA, help control the CPS biosynthesis by regulating the cps transcriptions in K. pneumoniae [25,26]. An increase in CPS synthesis protects K. pneumoniae from oxidative stress [27][28][29]. However, whether the modulation of c-di-GMP affects CPS synthesis remains unclear.
The expression of yjcC, an IVE gene isolated from the liver abscess isolate K. pneumoniae CG43, is inducible in the presence of 10 mM paraquat [30]. Sequence analysis of YjcC shows a signal peptide followed by 2 transmembrane domains and a CSS motif at the N-terminal region, whereas the C-terminal contains a conserved EAL domain of the PDE enzyme [31]. In addition, the encoding gene yjcC is cluster-located with soxRS genes, suggesting that it plays a role in the oxidative stress response. This study investigates whether YjcC plays a role in oxidative stress defenses and if YjcC uses PDE activity to execute its regulation.

Results
The YjcC expression is paraquat inducible, and SoxRS and RpoS dependent To confirm the previously reported paraquat-induced expression phenotype [30], the IVE DNA containing the 59 non-coding region and part of the coding sequence of yjcC was isolated from K. pneumoniae CG43S3 and cloned in front of the promoterless lacZ gene of pLacZ15 [26]. The resulting plasmid was called pP yjcC1 . The sequence analysis of P yjcC1 shows a conserved Fnr box TGTGA-N 6 -TCACA [32] centered approximately 400-bp upstream of the yjcC start codon. This process also generated recombinant plasmids pP yjcC2 and pP yjcC0 carrying truncated forms of P yjcC1 . These plasmids respectively removed the putative Fnr box and the small stem-loop sequence of the 33-bp coding region (Fig. 1A). As Fig. 1B shows, the bacteria containing pP yjcC1 exhibited the highest level of b-galactosidase activity, whereas CG43S3[pP yjcC2 ] had the lowest activity. In addition, the activity of P yjcC1 , but not P yjcC2 nor P yjcC0 , increased after added 10 mM paraquat to the culture medium. This paraquat-induced characteristic also appeared when the concentration increased to 30 mM, further enhancing the activity of P yjcC1 .
As Fig. 1C shows, the addition 30 mM paraquat to the bacterial culture significantly increased the yjcC mRNA level. Compared to the expression of the well-characterized stress response regulators SoxS, SoxR, RpoS, and Fnr, the yjcC gene expression was more responsive to paraquat than to hydrogen peroxide exposure. This study also investigates whether yjcC is subjected to regulation by SoxRS or RpoS. As Fig. 1D shows, the deletion of soxR, soxS, or rpoS reduces the yjcC expression, implying that SoxRS and RpoS play a positive role in yjcC expression.

YjcC plays a positive role in the oxidative stress response
Paraquat is a superoxide anion generator. Thus, the paraquatinducible expression suggests that YjcC plays a role in the oxidative stress response. To investigate this possibility, an yjcC deletion mutant was generated through an allelic exchange strategy. As Fig. 2A shows, the yjcC deletion mutant was more sensitive to paraquat and hydrogen peroxide when compared to the wild type bacteria K. pneumoniae CG43S3. The deletion effect could be complemented by transforming the yjcC expression plasmid pJR1 into the mutant. However, introducing the mutant pJR2, which expresses the mutant form of YjcC with the conserved E residue of the EAL domain replaced by A or pJR3 (which carries the coding region of the YjcC EAL domain), had no complementation effect. Neither of the two EAL-domain protein encoding plasmids pmrkJ and pfimK, which carry PDE activity, could complement the yjcC deletion effect. These results suggest that the stress response is YjcC dependent and both the Nterminal signaling receiving region and the EAL domain of YjcC are required and specific for an oxidative stress response.
To determine if the YjcC-EAL domain exhibits PDE activity, the recombinant expression plasmid containing the DNA coding for the EAL domain of YjcC or the AAL coding region of pJR2 was constructed and overexpresed in E. coli, and the recombinant proteins were purified. Figure 2B shows that the purified EAL domain protein exhibits PDE activity towards pNpp. This activity is lower than the level of the recombinant MrkJ [33], but considerably higher than the activity of the recombinant protein AAL yjcC . As Fig. 2C shows, the c-di-GMP level of CG43S3DyjcC[pJR1] was significantly lower than those of CG43S3[pRK415], CG43S3 DyjcC[pJR2], or CG43S3DyjcC[pJR3]. This suggests that YjcC in vivo functions as a PDE enzyme capable of reducing the intracellular c-di-GMP levels. The deletion of yjcC gene from CG43S3 increased the c-di-GMP amounts and the difference between the levels was much more apparent after the bacteria exposure to 30 mM paraquat (Fig. 2D). This also suggests that YjcC is able to degrade c-di-GMP and the catalytic activity could be enhanced by oxidative stress.
Deletion of yjcC places bacteria in an oxidative stress state As Figs. 3A and 3B show, the deletion of yjcC after treatment of H 2 O 2 or paraquat significantly raised the levels of the fluorescent probe H2DCFDA (used to monitor the formation of ROS) and carbonyl proteins. The introduction of pJR1 into CG43S3DyjcC mutant appeared to reduce the levels of ROS and the carbonyl proteins, showing that YjcC is involved in the removal of ROS or damaged molecules. Thus, this study also investigates the antioxidant activity of YjcC. As Fig. 3C shows, the deletion of yjcC reduced the oxidant scavenging activity, as assessed by the absorbance change at 517 nm for the decolorization degree of the purple color, supporting the possibility that YjcC modulates anti-oxidant activity in a certain manner. Numerous studies have shown that Fur and RpoS affect and regulate numerous SODs and catalases [7,[34][35][36][37]. Figure 3D shows that zymogel analysis and total activity measurement exhibit significant changes in the SOD or catalase activity after the deletion of fur or rpoS. However, the deletion of yjcC has no apparent influence on SOD or catalase activity, suggesting that the YjcC-dependent anti-oxidant enzyme remains to be identified.
YjcC plays a regulatory role in the virulence, CPS production, biofilm formation, and type 3 fimbriae expression YjcC, previously identified as an IVE gene product, is likely involved in infection [30]. To investigate whether YjcC is a virulence factor for the bacteria to establish infection, a mouse peritonitis model was employed. As Table 1 shows, the LD50 to Balb/c mice increased approximately 10-fold after yjcC deletion; introducing DyjcC with pJR1, but not pJR2 or pJR3, could restore the LD50. This indicates that YjcC expression at a certain stage is required for mouse infection. It is interesting to note that the DyjcC colony is smaller and less mucoid, as determined by a string test [25], than its parental strain on LB agar plate. Therefore, sedimentation analysis and glucuronic acid content measurement are carried out to determine the CPS production. As Fig. 4A shows the deletion of yjcC reduces CPS production. The CPS deficient phenotype can be fully complemented with the transformation of pJR1 into the DyjcC mutant. However, transforming the mutant with pJR2 or pJR3 partially restores glucuronic acid production.
The second messenger c-di-GMP plays an important role in bacterial biofilm formation [14][15][16]. Figure 4B shows that the biofilm formation activity of DyjcC appears to increase compared to the parental strain, whereas that transformed with pJR1 decreases biofilm formation. This can be attributed to the level changes of c-di-GMP (Fig. 2C), which indicates that the YjcC expression of pJR1 significantly reduces the c-di-GMP level, thus reducing biofilm forming activity. Moreover, type 3 fimbriae is a major determinant of biofilm formation in K. pneumoniae [38]. Therefore, this study also investigates the deletion effect on type 3 fimbriae expression. As Fig. 4C shows, the western blot hybridization with anti-MrkA antibody shows that the yjcC deletion also increased the major pilin MrkA production of type 3 fimbriae.

Effects of YjcC overexpression assessed using a transcriptome study
This study uses comparative transcriptome analysis between CG43S3[pRK415] and CG43S3[pJR1] to gain further insights into how YjcC executes its regulation. Analysis of the genome annotation of liver abscess isolate K. pneumoniae NTHU-K2044 [39] shows that the increased expression of yjcC significantly enhances the expression of 34 genes. As Table 2 shows, the YjcCactivated genes can be categorized into 12 functional groups. These include the oxidative stress response genes grxA, ybbN, dinI, priB, and stpA, which are involved in anti-oxidation [40,41] or DNA repair [41], the heat shock chaperone protein encoding genes ibpB, ibpA, htpG, and dnaK, which are generally induced in stress conditions [42], and the genes coding for chaperone ClpB and PspB to protect protein from aggregation and help maintain proton motive force (PMF) to counteract stress conditions [43,44]. Increasing the expression of yjcC also enhanced the expression of PilZ domain protein MrkH and the LuxR-type transcription factor MrkI. Conversely, 29 genes whose expressions were significantly repressed by the increase of YjcC expression include fumB, which is regulated by fnr under limited oxygen and anaerobic conditions [45]. Other YjcC negatively affected genes are metabolite transporter genes and genes coding for permease and energy metabolism involved in the synthesis of amino acids ( Table 3).
As assessed by qRT-PCR analysis, Fig. 5 shows that the mRNA level of mrkH and mrkI, respectively increased 2.87-and 3.24-fold in CG43S3[pJR1] compared to that of CG43S3[pRK415]. In contrast, the mrkA transcript levels dropped to approximately 1/3 of that of CG43S3[pRK415].

Discussion
In E. coli, all the described genes inducible by paraquat are a part of the SoxRS regulon [46][47][48]. The expression of yjcC is RpoS-dependent in E. coli [49]. Consistent with this finding, the yjcC expression in K. pneumoniae is affected by RpoS and by SoxRS at the transcriptional level. No SoxRS or RpoS binding box appears within the putative promoter sequence, suggesting the possibility of an indirect control. In E. coli, the FNR regulator controls the transition between aerobic and anaerobic growth at the transcriptional level [50]. The conserved fnr binding box present in the upstream non-coding region of yjcC implies an FNR dependent control of YjcC expression. Thus, the yjcC regulation by FNR likely occurs in poorly oxygenated environments.
The K. pneumoniae NTUH K2044 genome contains 27 genes encoding GGDEF-, EAL-, and GGDEF-EAL-domain proteins of potential DGC and PDE enzyme activity [51]. The yjcC encoding gene is also identified as member of the protein family. The family regulation specificity is determined by the sensory domain of the DGC or PDE proteins. Figure 2A shows that the PDE expression plasmids pfimK and pmrkJ, which contain the respective coding region with putative promoter, failed to complement yjcC deletion. This may be due to fimK and mrkJ genes are not induced in comparison to yjcC in the presence of paraquat. There is also the possibility that the N-terminal peptide of approximately 300 aa of YjcC plays a role in the oxidative stress response besides signal sensing. Various sensory domains can bind to small molecular signals, and through this connection, modulate the levels of c-di-GMP [52,53]. Although the signal for the CSS-motif remains unknown, we speculate that YjcC-mediated signaling sensing may occur in the periplasmic space because of the signal peptide and the transmembrane domain at the N-terminal region. The purified recombinant EAL domain of YjcC exhibits PDE activity. However, the amount of c-di-GMP in CG43S3DyjcC[pJR3] is approximately the same as that in CG43S3DyjcC[pRK415] and CG43S3DyjcC[pJR2], suggesting that the EAL domain has extremely low levels of PDE activity. The N terminal region of YjcC likely requires receiving some signal from outside, and the interaction of the sensory domain with signaling molecules activates the PDE activity. This is supported by Fig. 2D showing that the deletion of yjcC gene from CG43S3 raised the c-di-GMP levels and the influence was much more apparent after the bacteria exposure to 30 mM paraquat.
The deletion of yjcC decreases the CPS production and the virulence attenuation, indicating that YjcC is required to avoid the damage of oxidative stress. In bacteria, superoxide dismutase and catalase are common antioxidant enzymes that scavenge ROS from oxidative stress. The redox proteins (including GrxA and YbbN) required for maintaining redox status in bacteria are also protect bacteria from oxidative stress [54,55]. The deletion of yjcC has no apparent effect on the SOD and catalase activity, but appears to increase the transcription of grxA and ybbN. This suggests that YjcC is involved in regulating the redox levels in bacteria after oxidative stress. The chaperone protein ClpB stabilizes protein and suppresses the protein aggregation induced by heat or other stresses [43,56]. The phage shock proteins, PspB and PspC, act as positive regulators to transduce stress signal(s) to PspA through protein-protein interaction, maintaining the proton-motive force under extracytoplasmic stress conditions [44]. Transcriptome analysis shows that an increase in YjcC induces the expression of several heat shock proteins and chaperones, suggesting that YjcC is involved in regulating anti-stress responses. The fumarase gene fumB, which is expressed under anaerobic cell growth conditions, is regulated by Fnr and ArcA [57,58]. The result of yjcC expression reduced the fumB transcripts implies that YjcC is a component of the Fnr and ArcA regulatory pathway. Figure 2. Analysis of the deletion effects of yjcC upon exposure to oxidative stress. Diagrammatic depiction of the YjcC complementation plasmids is shown in the upper panel. The plasmid pJR2 is identical to pJR1 except that the E residue was replaced with A by site directed mutagenesis. The plasmid pJR3 carries only the EAL domain region of YjcC. Overnight cultures were collected and refreshed grown in LB until OD600 reached 0.6-0.7. 500 mM paraquat or 10 mM H 2 O 2 was then added and the cultures continued for 35 min and finally the cultures plated onto LB plates for colony formation. (B) Measurement of phosphodiesterase activity of the recombinant MrkJ protein (33), YjcC-EAL domain and YjcC-AAL domain proteins. BSA was used as a negative control. The activity is demonstrated using bis (pNpp) as substrate by the release of p-nitrophenol. (C), (D) Quantification of the c-di-GMP levels using the ELISA kit according to the manual (Wuhan EIAab Science Co., Ltd). Three independent experiments were performed for the measurement Error bars shown are standard deviations, and asterisks indicate the differences with a statistical significance, P,0.001. doi:10.1371/journal.pone.0066740.g002 In K. pneumoniae, the second messenger c-di-GMP activates type 3 fimbriae expression through MrkHI activation [18]. However, the increased synthesis of mrkH and mrkI transcripts by the overexpression of yjcC remains unclear. We have found that mrkHI expression is barely detected under LB culture (unpublished observation). Under oxidative stress pressure, the N-terminal region of YjcC may turn on the expression of MrkHI. Thus, the N-terminal region likely plays a determinant role in YjcCdependent regulation. E. coli YdeH has c-di-GMP cyclase activity [59]. The transcriptome analysis as shown in Table S3 revealed that the transcript levels of mrkA, mrkH, and mrkI in CG43[pRK415-ydeH], with c-di-GMP level of 23.1 fmole/mg 21 (data not shown), significantly increased compared to those of CG43S3[pJR1]. This also indicates that mrkHI expression is c-di-GMP level-dependent and the N-terminal part of YjcC plays a positive regulatory role in the expression of mrkHI.
Overall, these results indicate that the YjcC-mediated regulatory system is considerably more complex than expected. During infection, the transition from aerobic to microaerobic conditions or the transition from a microaerobic to oxidative stress environment, YjcC may be activated through sensory regulatory systems on the N-terminal region. Thereafter, YjcC modulates the levels of c-di-GMP to affect the expression of the downstream regulatory pathways. In conclusion, YjcC regulates the oxidative stress response, mouse virulence, CPS synthesis, biofilm formation, and type 3 fimbriae expression. This most likely occurs through the adjustment of c-di-GMP levels after receiving outside signals.

Ethics Statement
All animal experiments were performed in strict accordance with the recommendation in the Guide for the Care and Use of Laboratory Animals of the National Laboratory Animal Center (Taiwan), and the protocol was approved by the Animal Experimental Center of National Chiao Tung University (Permit number: 00990006). All surgery was performed under anesthesia, and all efforts were made to minimize suffering. Table S1 presents the bacterial strains and plasmids used in this study. E. coli and K. pneumoniae CG43 [25,26] and its derivatives were propagated at 37uC in Luria-Bertani (LB) broth. The antibiotics used include ampicillin (100 mg/mL), chloramphenicol (35 mg/mL), kanamycin (25 mg/mL), streptomycin (500 mg/mL), and tetracycline (12.5 mg/mL). Table S2 presents the primers used in this study.

Measurement of promoter activity
The promoter reporter plasmids were individually mobilized from E.coli S17-1l pir to K. pneumoniae strains by conjugation. The b-galactosidase activity was measured as described previously [26]. The bacteria were grown to the log phase in the LB medium (OD600 of 0.6-0.7) and 100 mL of the culture was mixed with 900 mL of Z buffer (60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KCl, 1 mM MgSO4, and 50 mM b-mercaptoethanol), 17 mL of 0.1% sodium dodecyl sulfate (SDS), and 35 mL of chloroform, followed by vigorous shaking. After incubation at 30uC for 10 min, 200 mL of a 4-mg/mL concentration of onitrophenyl-b-D galactopyranoside (ONPG; Sigma-Aldrich, Milwaukee, WI) was added to the mixture to initiate the reaction. When yellow coloration appeared, the reaction was stopped by adding 500 mL of 1 M Na2CO3 to the mixture. The absorbance at OD420 was recorded, and the activity was expressed as Miller units. Each sample was assayed in triplicate, and at least 3 independent experiments were conducted. The data shown were calculated from one representative experiment, and are presented as the means and standard deviations from triplicate samples.

Real-time PCR analysis
Total RNA was isolated from bacteria using High Pure RNA isolation Kit (Roche, Germany), and the residual DNA was eliminated with RNase-free DNase I (Roche, Germany). The cDNAs used for PCR were synthesized from 1.5 mg RNA using random hexamer primer from RevertAid TM H Minus First strand cDNA synthesis Kit (Fermentas, Canada). An ABI Prism 7000 Detection system was used to perform PCR following the manufacturer instructions, and the products were detected using SYBR Green RCR Master Mix (Roche, Germany). The RNA samples were normalized to the level of 23S rRNA. PCR analysis was performed in triplicate in a reaction volume of 25 mL containing 12.5 mL of SYBR Green PCR Master Mix, 300 nM of primer pairs, 9.5 mL of distilled H 2 O, and 1 mL of cDNA. Samples were heated for 10 min at 95uC and amplified for 40 cycles for 15 s at 95uC and 60 s at 60uC. Quantification was performed using the 2 2DDCt method [60].

Construction of the gene deletion mutants and complementation plasmids
Specific gene deletion was introduced into K. pneumoniae CG43 using an allelic-exchange strategy as previously described [26]. Two DNA fragments of approximately 1000-bp flanking both sides of the deleted region were cloned into pKAS46 [25]. The plasmid pKAS46 is a suicide vector containing rpsL, which allows positive selection with streptomycin for vector loss. The resulting plasmids were mobilized from E.coli S17-1l pir to K. pneumoniae CG43S3 or CG43S3DlacZ, by conjugation, respectively. The transconjugants, with the plasmid integrated into the chromosome   through homologous recombination, were selected with ampicillin and kanamycin on M9 agar plates. Several of the colonies were grown overnight in LB broth at 37uC and then spread onto an LB agar plate containing 500 mg of streptomycin/mL. Streptomycinresistant and kanamycin-sensitive colonies were selected, and the deletion was verified by PCR. Table S1 presents the resulting mutant strains. To obtain the complementation plasmids, DNA fragments containing the yjcC, and soxRS loci were PCR amplified using the primer pairs pjr1-F/pjr1-R, pjr2-F/pjr2-R, and pjr3-F/ pjr3-R (Table S2). The PCR products were cloned in pRK415 [61] to generate pJR1(pRK415-yjcC), pJR2(pRK415-yjcC-AAL), pJR3(pRK415-EAL domain only of yjcC), pmrkJ, and pfimK respectively (Table S1).

Site-directed mutagenesis
Site-directed mutagenesis was performed on the plasmid pJR1 to substitute the critical residue E with A in the EAL domain of YjcC using a QuickChange site-directed mutagenesis kit and following the manufacturer protocols (Stratagene). The resulting PCR product contained one point mutation, corresponding to the E303-to-A303 change in the active EAL site. The resulting PCR product was digested with BamHI and HindIII and ligated into BamHI/HindIII-digested plasmid pRK415-pJR2.

Paraquat and H2O2 survival assessment
One hundred microliter of bacteria grown overnight were inoculated in LB and incubated at 37uC to OD600 of 0.6-0.7. An aliquot of the bacteria was collected by centrifugation and then resuspended in 500 mM of paraquat and 10 mM H 2 O 2 respectively, and then subjected to 37uC incubation for 35 min. The colony-forming unit (CFU) of the bacteria was counted after the stress treatment, and the survival rate was determined by the CFU ratio. This study presents the representative data of at least 3 independent experiments. Every sample was assayed in triplicate, and this study presents the average activity and standard deviation.

Cloning, expression, and purification of the recombinant proteins
The coding regions of the EAL or AAL domains of yjcC were PCR amplified with the primer sets yEAL-F/yEAL-R and yAAL-F/yAAL-R (Table S2) and cloned in the NdeI/XhoI site in pET30b (Novagen, Madison, WI). This process generated pET30b-EAL or pET30b-AAL with a carboxyl-terminus His tag (RcsB-His6). The resulting plasmids pEAL (pET30b-EAL of yjcC) and pAAL (pET30b-AAL of yjcC) were individually transformed into E. coli BL21(DE3)/pLysS (Invitrogen), and the overproduction of the recombinant protein was induced by adding 0.5 mM IPTG for 4 h at 37uC. The recombinant proteins were then purified from the soluble fraction of the total cell lysate by affinity chromatography using His-Bind resin (Novagen). The purified proteins were then dialyzed against 1 TBS (Tris-buffered saline; pH 7.4) containing 10% glycerol at 4uC overnight, followed by condensation with PEG 20000. The purity was determined by SDS-PAGE.

Phosphodiesterase activity assay
The recombinant plasmids pET30b-EAL yjcC , pET30b-AAL yjcC , and pET30b-mrkJ were transformed into E. coli BL21(DE3), and the protein induced expression using 0.5 mM IPTG at 22uC for 12 h. The PDE activity was determined using the synthetic chromogenic substrate bis(p-nitrophenyl) phosphate (bis-pNPP) (Sigma-Aldrich) as previously described [62,63]. The specific PDE activity was determined using purified proteins and by measuring the release of p-nitrophenol (pNP) at 405 nm. The calculations in this study use an extinction coefficient of 1.78*104/M*cm for pnitrophenol. Control BSA without extracts was included to account for any non-enzymatic bis-pNPP hydrolysis and MrkJ, which carries PDE activity as a positive control [33].

In vivo ci-di GMP content
To measure the c-di-GMP contents, cellular extracts were prepared as described [64]. The cultured bacteria were collected and treated with formaldehyde (0.19% final concentration) and then pelleted by centrifugation. The pellet was suspended in water and heated to 95uC for 10 min before the nucleotides were extracted by 65% ethanol. The lyophilized samples were then resuspended in water and this suspension was used for c-di-GMP detection with a cyclic diguanylate ELISA kit (Wuhan EIAab Science co., Ltd). The ci-di GMP activity of crude extracts (1 mg of total protein/mL) containing WT-pRK415 vector only, pJR1, and pJR2 in DyjcC was also assayed as described above.

Determination of intracellular ROS concentration
Bacterial cultures were grown exponentially (OD600 = 0.6-0.7) and exposed to 10 mM of hydrogen peroxide or 500 mM of paraquat for 40 min. Cells were centrifuged at 13000 g, washed with 10 mM potassium phosphate (pH 7.0) buffer (Buffer A), and suspended in 500 mL of the same buffer, which contained 10 mM 29,79-dihydrodichlorofluorescein diacetate (H2DCFDA). After shaking for defined periods in darkness at room temperature, cells were centrifuged as mentioned and washed twice with 500 mL of Buffer A. Cells were suspended in 500 mL of Buffer A and disrupted by sonication. After centrifugation at 13000 g, aliquots of 100 mL supernatants were used to determine fluorescence intensity (excitation 490 nm and emission 519 nm) as described [1,2].

Oxidation of cytoplasmic proteins
Bacteria was grown to an OD600 0.6-0.7 in the presence of 10 mM hydrogen peroxide. After incubating for 30 min at 37uC, crude extracts were prepared and suspended in 500 mL of Buffer A and then disrupted by sonication. After centrifugation at 13000 g, 4 aliquots of 10 mM dinitrophenylhydrazine (DNPH) were added to the supernatant and the mixture was incubated at room temperature for 1 h with occasional stirring. Proteins were precipitated by adding one volume of 20% trichloroacetic acid (TCA) and centrifuged at 13000 g for 5 min. The precipitate was washed 3 times with a mixture of ethanol: ethyl acetate (1:1). Finally, the sediment was dissolved in 450 mL of 6M guanidine hydrochloride/dithiothreitol and carbonyl concentration was determined spectrophotometrically at 370 nm (e = 22000 M-1 cm-1) [2,3].
Evaluation of antioxidant activity with DPPH assay DPPH radical scavenging activity was estimated using the method of Liyana-Pathirana and Adedapo [65,66]. When DPPH accepts an electron donated by an antioxidant compound, the free radical is decolorized from a stable purple color to yellow which can be quantitatively measured from the changes in absorbance. A solution of 0.135 mM DPPH in methanol was prepared, and 1.0 mL of this solution was mixed with 1.0 mL of extract in methanol containing 0.02-0.1 mg of the extract. The reaction mixture was thoroughly vortexed and left in the dark at room temperature for 30 min. The absorbance of the mixture was measured with a spectrophotometer at 517 nm, with ascorbic acid and BHT as references. The scavenge ability to remove DPPH radicals was calculated using the following equation: DPPH radical scavenging activity (%) = [(Abs control2Abs sample)]/(Abs control)]6100, where Abs control is the absorbance of DPPH radical+methanol, and Abs sample is the absorbance of DPPH radical+sample extract/standard.

Determination of SOD and catalase enzyme activity
Cell-free extracts were harvested (15000 rpm, 4uC, 20 min) from the exponential phased bacteria (OD 600 = 0.7-0.8) and suspended in ice-cold potassium phosphate buffer (50 mM, pH 7). After cells were disrupted by ultra-sonication, cell debris was removed by 12000 rpm centrifugation for 10 min at 4uC, and the supernatant was collected. The extraction of total proteins was conducted on ice and the concentrations were estimated using the Bradford method [67]. Aliquots of the extracted proteins were individually loaded onto 10% native polyacrylamide gels and the proteins were separated at a constant voltage of 150 V for 2 h. The gels were then removed and stained for SOD and CAT activity using the methods of Beauchamp and Fridovich [68] and Woodbury et al. [69], respectively. The SODs were localized by soaking gels in 2.45 mM nitro blue tetrazolium for 20 min, followed by immersion in a solution of 50 mM phosphate buffer (pH 7.0), 0.028 mM riboflavin, and 0.028 M TEMED (N,N,N_,N_-tetramethylethylenediamine). The gels were then removed from the solution and exposed to light for approximately 20 min. The SOD activity produced achromatic zones in the purple gel. The expression of CAT activity was identified by soaking gels in 10 mM hydrogen peroxide for 30 min with gentle shaking and then transferring the gels to a solution of 1% ferric chloride and 1% potassium ferricyanide for 10 min. The localized-CAT produced colorless bands on a dark green background.
The SOD activity was determined using a spectrophotometer at 25uC using the xanthine oxidase-cytochrome C method [68]. The assay mixture in 0.7 mL contained 50 mM potassium phosphate (pH 7.8), 0.1 mM EDTA, 50 mM xanthine, 1.7 mU xanthine oxidase, and 10 mM cytochrome C. The reduction of cytochrome C was measured at A550. One unit (U) of SOD activity was defined as the amount of enzyme required to inhibit the reduction rate of cytochrome C by 50%. Catalase activity was also determined spectrophotometrically at 25uC by monitoring the decrease in A240 in 50 mM Tris/HCl buffer (pH 8.0) [1]. One unit (U) of activity was defined as the amount of enzyme that catalyses the oxidation of 1 mmol H 2 O 2 min 21 under assay conditions.

Mouse lethality assay
The bacterial virulence in mice was determined as described [25]. Female BALB/c mice (aged 4 to 5 wk) were obtained from the National Laboratory Animal Center and acclimatized in an animal house for 7 d. The tested bacterial strains were cultured overnight in LB medium at 37uC. Four mice in each group were injected intraperitoneally with 0.2 mL of bacterial suspension in saline in 10-fold graded doses. The LD50 values were calculated using the Reed and Muench method [70] based on the number of survivors after 14 d.

Extraction and quantification of capsular polysaccharides
The bacterial CPS was extracted using the method described [25,71]. A 500 mL sample of overnight-grown bacteria was mixed with 100 mL of 1% Zwittergent 3-14 (Sigma-Aldrich, Milwaukee, WI) in 100 mM citric acid (pH 2.0) and then incubated at 50uC for 20 min. After centrifugation, a 250 mL sample of the supernatant was transferred to a new tube, and the CPS was precipitated with 1 mL of absolute ethanol. The pellet was dried and dissolved in 200 mL of distilled water, and 1200 mL of 12.5 mM borax in H2SO4 was then added. The mixture was vigorously mixed, boiled for 5 min, and cooled before adding 20 mL of 0.15% 3-hydroxydiphenol (Sigma-Aldrich, Milwaukee, WI). The absorbance at 520 nm was measured, and the uronic acid content was determined from a standard curve of glucuronic acid and expressed as micrograms per 10 8 or 10 9 CFU.

Biofilm formation assay
The ability of bacteria to form biofilm was analyzed as described, with a minor modification. Bacteria were diluted 1/ 100 in LB broth supplemented with the appropriate antibiotics, and this dilution was inoculated into each well of a 96-well micro titre dish (Orange Scientific) and statically incubated at 37uC for 48 h. Planktonic cells were removed, and the wells were washed once with distilled water to remove unattached cells. Crystal violet (0.1% w/v; Sigma) was used to stain the attached cells for 30 min. Unattached dye was rinsed by washing 3 times with distilled water, and the stained biomass was solubilized in 1% (w/v) SDS. The absorbance was determined at 595 nm, and relative bacterial biofilm-forming activities were observed.
Western blot analysis K. pneumoniae CG43S3 and its derived mutants were grown in LB broth with agitation at 37uC. The bacterial total protein, approximately 5 mg per lane, was then subjected to western blot analysis using MrkA antiserum. Aliquots of total cellular lysates were resolved by SDS-PAGE, and the proteins were electrophoretically transferred onto a polyvinylidene difluoride (PVDF) membrane (Millipore, Billerica, MA, USA). After incubation with 5% skim milk at room temperature for 1 h, the membrane was washed 3 times with 16 PBS. The membrane was subsequently incubated with diluted anti-GAPDH (GeneTex Inc.) and anti-MrkA serum at room temperature for 1 h. After 3 washes with 16 PBS, a 5000-fold diluted alkaline phosphatase-conjugated antirabbit immunoglobulin G was added and incubated for 1 h. The blot was washed again and the bound antibodies were detected using the chromogenic reagents BCIP (5-bromo-4-chloro-3-indolyl phosphate) and NBT (Nitro blue tetrazolium).

Transcriptome analysis
Total RNA from CG43S3[pRK415] and CG43S3[pJR1] were isolated using Trizol reagent (Invitrogen), subsequently purified using RNeasy MinElute Cleanup Kit (Qiagen), and eluted in RNase-free water. The RNA samples were sequenced on Illumina's sequencing instrument. In the data analysis pipeline, we used the FASTX-Toolkit to remove or trim deep sequencing reads containing low quality bases or adaptor sequences. To estimate expression levels of genes, all the remaining reads were mapped to the Klebsiella pneumoniae NTUH-K2044 genome [39]using TopHat and determined using Cufflinks package. The genes were identified as significantly transcript abundance changed in CG43S3[pRK415] as compared to that in CG43S3[pJR1] if the log 2 fold change was greater than 2 (up and down). Finally, functional annotation tools, such as DAVID, can be used to illustrate the biological regulation role from Gene Ontology or KEGG pathway database.

Author Contributions
Conceived and designed the experiments: CJH HLP. Performed the experiments: CJH ZCW. Analyzed the data: CJH HLP. Contributed reagents/materials/analysis tools: ZCW HYH HDH. Wrote the paper: CJH HLP.