Identification of a capsular variant and characterization of capsular acetylation in Klebsiella pneumoniae PLA-associated type K57

Klebsiella pneumoniae can cause community-acquired pyogenic liver abscess (PLA). Capsular polysaccharide (CPS) is important for its virulence. Among 79 capsular (K) types discovered thus far, K57 is often associated with PLA. Here, we report the identification of a K57 variant. Cps gene locus sequencing revealed differences between the K57 reference strain 4425/51 (Ref-K57) and a variant, the PLA isolate A1142. While Ref-K57 cps contained orf13 encoding a putative acetyltransferase, the insertion of a putative transposase-encoding gene at this position was detected in A1142. This variation was detected in other K57 clinical strains. Biochemical analyses indicated that A1142 was deficient in CPS acetylation. Genetic replacement and complementation verified that orf13 was responsible for CPS acetylation. Acetylation increased CPS immunoreactivity to antiserum and enhanced K. pneumoniae induction of pro-inflammatory cytokines through JNK and MAPK signaling. While acetylation diminished the serum resistance of bacteria, it promoted adhesion to intestinal epithelial cells possibly via increasing production of type I fimbriae. In conclusion, acetylation-mediated capsular variation in K57 was observed. Capsular acetylation contributed to the variety and antigenic diversity of CPS, influenced its biological activities, and was involved in K. pneumoniae-host interactions. These findings have implications for vaccine design and pathogenicity of K. pneumoniae.

Our previous studies of 42 K. pneumoniae strains causing PLA identified those with K1 (n = 35), K2 (n = 2), K57 (n = 2), K5 (n = 1), and K54 (n = 1) capsules, as well as a new type (n = 1) 14 . Similarly, the prevalence of 50 liver abscess isolates in Southern Taiwan revealed capsular types K1, K2, K5, K20, K54, and K57, in addition to an unidentified type 16 . The chromosomal capsular polysaccharide synthesis (cps) gene cluster encodes the majority of the proteins involved in the translocation and assembly of surface polysaccharides, composed of repeated sugar subunits 17 , for the K. pneumoniae capsule. Genotyping of cps can be used to distinguish capsular types 18,19 . Information about disease-related capsular types of bacterial pathogens can contribute to diagnosis and to the development of capsule-based vaccines.
To understand pathogen-host interactions and host responses, characterization of the structures and biological activities of various capsular architectures is important. Polysaccharide modifications have been described to cause capsular variation in what were originally defined as singular capsular types in some pathogens, such as Streptococcus pneumoniae 20 and Escherichia coli K1 strains 21 . Capsular modifications also may be associated with the virulence of some bacterial strains 21,22 . Although modifications of CPS by O-acetyl and O-pyruvyl groups have been reported in a K. pneumoniae K1 PLA strain 23 , analysis of potential capsular variation and related modifications in Klebsiella is incomplete. In addition, the roles of capsular modifications in K. pneumoniae remain to be elucidated. In addition, direct links between the structural, biochemical, and genetic data for some capsular types are still lacking.
K. pneumoniae K57 is one of the PLA-associated capsular types. In this study, we discovered the presence of a capsular variant in the K57 capsular type, which was based on genetic data of the cps region and biochemical analysis of CPS modification. Our group previously published the complete sequence of the K57 cps cluster of the PLA isolate, A1142 14 . Sequencing of the cps cluster of another strain, the K57 reference strain (Ref-K57), revealed differences between the two strains at the site of a cps gene (orf13) encoding a putative acetyltransferase. Variations of this gene were also detected in other K57 clinical strains. By gene replacement and complementation analyses, we verified that orf13 is responsible for CPS acetylation, which altered K. pneumoniae K57 antigenicity, innate host response, serum resistance, and cell adhesion.

Results
Identification of differences in cps gene loci in K. pneumoniae K57 strains. Our previous study focused on the cps regions of K. pneumoniae K57 strain, which is related to PLA 14 . Thus, we sequenced and analyzed the cps gene cluster of the K57 reference strain (Ref-K57) from the Statens Serum Institute. We compared the sequence of the Ref-K57 cps with that published for the PLA isolate, A1142, another K57 strain (Table 1). We noted an obvious difference in the region between orf12 and wbaZ (Fig. 1A). Specifically, the Ref-K57 sequence included a 981-bp orf13 (DNA residues 15948-16928) in this position; the predicted orf13 gene product exhibited 38% amino acid identity (75/196) with the acyltransferase superfamily of proteins (WP_014751172). In contrast, the corresponding gene in A1142 apparently was disrupted by the insertion of a gene encoding a  Fig. 2A, upper panel); this modification was not detected in CPS isolated from A1142 ( Fig. 2A, lower panel). Using the Hestrin method to determine the degree of CPS acetylation 24,25 , Ref-K57 CPS acetylation levels were higher than those observed for A1142 CPS (Fig. 2B). The main sugar compositions of the CPS polymers from Ref-K57 and A1142 were also compared; no obvious differences were found. According to the published structure of K57 CPS 26 , the ratio of mannose (Man) and galactose (Gal) was predicted to be 2:1. The Man:Gal ratio in both Ref-K57 CPS and A1142 CPS were nearly 2:1. Thus, the CPS polymers of Ref-K57 and A1142 differed in their acetyl modification rather than in their major sugar composition.
To verify the function of the orf13 gene, we constructed gene replacement and complementation K57 strains (Fig. 2C). Specifically, we replaced the orf13 coding region (residues 15948-16928) of Ref-K57 with the transposase-interrupted orf13 from A1142 (residues 15933-17975), yielding an isogenic mutant strain that we designated Ref-K57Δ orf13::A1142orf13-tnp (the "replacement" strain). We generated chromosomal complementation of orf13 by insertion of the intact Ref-K57 orf13 gene into the intergenic region between wbaZ and gnd in the A1142 cps cluster, yielding a strain that we designated A1142::Ref-K57orf13 (the "complementation" strain). This gene replacement or complementation did not cause significant changes in CPS production compared to that in the respective parent strain, as determined by quantification of uronic acid in K57 CPS (Fig. 3A). Ref-K57 generated less CPS than A1142. The replacement strain Ref-K57Δ orf13::A1142orf13-tnp produced an amount of CPS close to that of Ref-K57, and similarly, no significant differences of CPS production were found between A1142 and A1142::Ref-K57orf13. In addition, India ink capsule staining showed no obvious differences in capsule thickness between the gene replacement or complementation strains and the respective parent strains (Fig. 3B). Acetylation enhances the immunoreactivity of K57 CPS. The surface carbohydrate comprising the K. pneumoniae capsule typically exhibits strong antigenicity. We, therefore, investigated whether acetylation affected the immunoreactivity of K57 CPS. Immunoblot analysis of equivalent amounts of CPS from each strain was undertaken using commercial anti-K57 serum, which was generated using the Ref-K57 strain by the Statens Serum Institute. As shown in Fig

CPS acetylation enhances induction of pro-inflammatory cytokines by K57 K. pneumoniae.
Microbial infection often triggers the production of host pro-inflammatory cytokines. The potential impact of K57 K. pneumoniae on host innate responses was assessed. The pro-inflammatory cytokines, tumor necrosis factor-α (TNF-α ) and interleukin-6 (IL-6), are secreted primarily by activated macrophages, mediating multiple biological effects, including activation of immune responses. A1142 stimulated TNF-α and IL-6 secretion by mouse macrophage Raw 264.7 cells as detected at ~1 and ~6 h after infection (Fig. 5A,B). Ref-K57 did not efficiently induce TNF-α and IL-6 secretion, which may be due to the lower CPS levels than that of A1142. No obvious differences were found between the Ref-K57Δ orf13::A1142orf13-tnp and Ref-K57 strains (Fig. 5A,B). Notably, the complementation strain, A1142::Ref-K57orf13, induced TNF-α and IL-6 secretion more efficiently than the A1142 parent strain. Compared to the A1142 parental strain, A1142::Ref-K57orf13 induced secretion of higher levels of TNF-α at 4 and 6 h post-infection (Fig. 5A). For IL-6 secretion, earlier induction by To investigate underlying mechanisms, we determined the effect of CPS acetylation on the activation of NF-κ B, JNK, p38 MAPK, and ERK1/2 in Raw 264.7 cell lysates by Western blotting (Fig. 5C). Consistent with the cytokine secretion data, the acetylation complementation strain induced the phosphorylation of NF-κ B, JNK and p38 MAPK more efficiently than the A1142 parental strain. Therefore, acetylation of CPS possibly increased K. pneumoniae-induced cytokine production due to enhanced activation of JNK and MAPK signaling pathways.
CPS acetylation decreases serum resistance of K. pneumoniae but promotes cell adhesion. In mice inoculated with K. pneumoniae A1142 or the complementation strain, A1142::Ref-K57orf13, in vivo competition assays revealed no significant differences in the virulence between the strains ( Fig. 6A; results of CFU viable counts shown in Figure S1). This assay reflected in vivo virulence of test strain in mice. Each test strain (LacZ-positive; blue colonies on the plates containing IPTG and X-Gal) was inoculated with the isogenic lacZ promoter deletion strain (white colonies) at a 1:1 ratio, and the numbers of bacterial colonies recovered from mice were compared. The colony numbers of the A1142 parent and A1142::Ref-K57orf13 were both similar to that of A1142Δ placZ ( Figure S1), and no significant differences between A1142 and A1142::Ref-K57orf13 in the competitive index (CI) were found (Fig. 6A). Thus, acetylation of CPS may not significantly change the in vivo virulence of A1142.
Given that the capsule protects K. pneumoniae from killing by the host sera in the immune system, we evaluated whether CPS acetylation affects the serum resistance of K. pneumoniae. In vitro serum resistance assays showed that although the A1142 strain grew in human serum (Fig. 6B), the A1142::Ref K57orf13 strain became sensitive to serum-killing, indicating that CPS acetylation decreased the serum resistance of K. pneumoniae and diminished the protective properties of the K57 capsule. Acetyl modification of CPS may reduce the ability of K. pneumoniae to escape from the host immune system. The capsule of K. pneumoniae has also been implicated in adhesion of bacteria to host cells 27 . We analyzed the effect of capsular acetylation on adhesion in a cell culture assay ( Fig. 6C; results of CFU viable counts shown in Figure S2). Notably, A1142::Ref-K57orf13 exhibited a higher rate of adhesion to human Caco-2 (intestinal epithelium-derived) cells than observed with the A1142 parental strain. Thus, CPS acetylation enhanced attachment of K. pneumoniae to host intestinal cells and may play a role in promoting its colonization and interactions with host cells.
Since India ink staining showed that the capsule thickness of A1142::Ref-K57orf13 was not significantly changed, we investigated whether fimbriae, the surface factors important for K. pneumoniae adhesion, were altered. Expression of the type I fimbriae genes, fimA and fimC, and the type III fimbriae gene, mrkD, was determined by quantitative real-time reverse-transcription PCR (RT-qPCR) (Fig. 6D). Compared to the A1142 parental strain, the expression of fimA and fimC in A1142::Ref-K57orf13 increased. No significant differences in mrkD were detected. Thus, the enhanced cell adhesion of the acetylation complementation strain could be due to higher levels of type I fimbriae. Discussion Conventional definitions of the K serotypes of encapsulated bacteria assume that each type has a specific capsular polysaccharide structure and is a genetically distinct entity. The presence of variants in a K serotype has been described in some bacteria [28][29][30] . Here, we characterized a variant of the K57 capsular serotype, a type that is associated with K. pneumoniae-induced PLA. Our work revealed that this capsular variation reflected changes in CPS acetylation. Although acetylation has been widely studied in the capsule of several pathogenic bacteria, such as the E. coli K1 strain, S. pneumoniae, and N. meningitidis [20][21][22]31,32 , this modification has been poorly characterized in Klebsiella capsular types. Our studies identified a gene whose product is responsible for the acetylation of K57 CPS and demonstrated its impact. These results have implications for K. pneumoniae vaccine design and pathogenicity.
Based on PCR genotyping of the cps cluster and immunoserotyping, the PLA isolate, A1142, was defined as a K57 capsular type 14 . However, it harbored differences in the cps gene (orf13) encoding a putative acetyltransferase. In A1142, orf13 is interrupted by insertion of a putative transposase-encoding gene and presumably inactivated. NMR, biochemical, and genetic analyses indicated that the intact orf13 of Ref-K57 mediates CPS acetylation, and A1142 is a K57 variant that is deficient in CPS acetylation. Capsular variants involving acetylation have been described in other bacterial species. For instance, in E. coli with K1 capsular polysaccharides, variation in polysaccharide O-acetylation is mediated by a phase-variable gene encoding an O-acetyltransferase 32,33 . The determinant gene, neuO, is associated with the prophage CUS-3 and not part of the cps cluster. In S. pneumoniae, the polysaccharide structures of serotypes 11A and 11E appear to be identical except for the presence of Each test strain was compared with the A1142Δ placZ mutant strain, and the ratio of LacZ-positive to LacZnegative colonies in the liver or spleen of each mouse was determined. A1142Δ placZ was fully virulent as the parental A1142. The competitive index (CI) was defined as (outputΔ placZ/output test strain )/(inputΔ placZ/ input test strain ). Each symbol represents the CI for each inoculum, with the medians shown by bars. A1142 and A1142::Ref-K57orf13, P = 0.31 for liver and P = 0.81 for spleen, Wilcoxon signed rank test. (B) Serum resistance assays. K. pneumoniae strains were incubated with human serum (75%) at 37 °C for 1 h. The numbers of recovered CFU were determined, and the survival ratio was expressed as recovered CFU/inoculum CFU. The CPS-lacking mutant, A1142Δ wzy, which is sensitive to serum killing, was analyzed for comparison. Data are presented as mean ± SEM from three independent experiments. * P < 0.05, Student's t test. (C) Adhesion of K. pneumoniae to Caco-2 cells. The adhesion rate was expressed as the proportion of the inoculum that adhered. Data are presented as mean ± SEM from three independent experiments. * P < 0.05, Student's t test. (D) Relative expression of the fimbriae genes, fimA, fimC, and mrkD, was determined by RT-qPCR. The acetylation complementation strain, A1142::Ref-K57orf13 (black bars, ◼ ), was compared to the A1142 parent strain (white bars, ◻ , set as 1). Data are presented as mean ± SEM from three independent experiments. * P < 0.05, Student's t test.
Scientific RepoRts | 6:31946 | DOI: 10.1038/srep31946 O-acetylation 21 . This modification depends on the presence of the O-acetyltransferase-encoding gene, wcjE, in the 11A cps locus. Strains of serotype 11E contain various disruptive mutations in wcjE that result in inactivation of the gene. Another example is the reversible serotype switching between S. pneumoniae 15B and 15C, a variation that has been attributed to reversible slipped-strand mutations in the wciZ gene of the cps cluster. Because this gene encodes an O-acetyltransferase, strand slippage during replication results in phase-variable O-acetylation of capsule polysaccharides 34,35 . Our studies of K. pneumoniae demonstrated that varied acetylation of CPS in K57 strains was mediated by disruptive insertion of a putative transposase-encoding gene in the cps region at the site of the gene encoding acetyltransferase, indicating another mechanism of serotype variation in bacteria.
The capsular polysaccharides of K. pneumoniae are immunogenic. Antibodies targeting the capsule have been shown to provide hosts with increased resistance to capsulated pathogens 36,37 . Modifications of the CPS may influence the properties of polysaccharides. Acetylation affects the antigenicity or immunogenicity of other bacterial polysaccharides 32,[38][39][40] , such as K1 Escherichia coli, groups W-135, Y, and C meningococci, and group B Streptococcus capsular polysaccharides. In agreement, our data demonstrated that the acetyl group influenced the immunoreactivity of K. pneumoniae K57 polysaccharides to anti-K57 antibodies. Immunoblotting of Ref-K57 CPS against anti-Ref-K57 antiserum showed that the immunoreactivity was significantly decreased by acetylation gene replacement (Ref-K57Δ orf13::A1142orf13-tnp). Therefore, the acetyl group may constitute part of the epitope of K. pneumoniae K57 CPS recognized by this antibody. CPS from A1142 displayed lower reactivity to anti-Ref-K57 antiserum, and complementation of acetylation enhanced the immunoreactivity of CPS in this strain. These results indicate that acetyl modification plays an important role in K. pneumoniae CPS antigenicity and antibody recognition, likely as the part of the polysaccharide epitope. Analysis of acetylation prevalence in K. pneumoniae clinical strains, including other PLA predominant capsular types, merits further survey. These results also suggest that polysaccharide modifications should be considered in the development of capsule-based vaccines for K. pneumoniae.
We also demonstrated the effects of CPS acetylation on innate immune responses. Notably, acetylation enhanced K57 K. pneumoniae induction of TNF-α and IL-6. This result is consistent with the previous demonstration that chemical removal of O-acetyl and O-pyruvyl groups from extracted K1 CPS attenuates the induction of pro-inflammatory cytokines 23 . In the same study, K1 CPS was shown to induce cytokine expression through activation of Toll-like receptor 4 (TLR4) and MAPK signaling pathways 23 . Our studies of K57 K. pneumoniae further revealed a role for CPS acetylation in host signal transduction. Thus, acetylation may improve CPS induction of cytokines via activating cell signaling pathways, including JNK and MAPK. We propose that acetylation possibly mediates the interactions between polysaccharides and host receptors to initiate cell signaling pathways, therefore, contributing to polysaccharide induction of cytokine expression and innate immune responses in the host.
Alternation of surface polysaccharide architecture by acetylation has been proposed to influence the virulence of several bacterial pathogens, including E. coli, N. meningitidis, and Staphylococcus aureus 21,22,31 . Epidemiological association of O-acetylated CPS and increased virulence was also reported in bacteremia-inducing E. coli K1 strains 21 . One study of S. aureus serotype 5 showed that a mutant lacking O-acetylation has reduced bacterial virulence in opsonophagocytic assays and in mice 31 . Our study of K. pneumoniae suggested that capsular acetylation contributed to pathogenicity; however, complementation of acetylation did not significantly change the in vivo virulence of the non-acetylated PLA strain A1142 in mice. Instead, acetylation influenced K. pneumoniae resistance against human serum and adhesion to human intestinal cells although with opposing effects. While acetylation reduced serum resistance of A1142, it promoted bacterial adhesion to the host cells, which may explain the unchanged in vivo virulence observed in the mouse experiments. Additionally, many other factors could also be involved in the in vivo results, i.e. an overall consequence. For example, K. pneumoniae might regulate their CPS production or acetylation levels in the host environment, or acetylation could affect other unidentified host responses. Our data imply that acetylation in K. pneumoniae K57 has different impacts on bacterial survival and adaptation. This capsular modification may play a variety of roles in K. pneumoniae pathogenicity, requiring further studies using different clinical strains and other K-type K. pneumoniae.
We demonstrated that acetylation of CPS promoted cell adhesion of K. pneumoniae. No significant changes in CPS production and the capsule thickness were found. Thus, one possibility for the better cell adhesion could be that acetylation enhanced interactions between bacterial surfaces with the host cells despite expressing similar amounts of CPS. RT-qPCR revealed that the expression of type I fimbriae genes was higher in the acetylated strain. Therefore, one explanation for increased adhesion might be due to more production of fimbriae in the acetylated CPS strain. Why and how complementation of acetylation could affect fimbriae expression is still unclear. Potential gene regulation networks or interplay between cps and fim genes should be further characterized.
Capsular typing of K. pneumoniae is important for determining its prevalence for epidemiological research. The inability to determine the types for some clinical isolates is not unusual. For example, an Australian survey using antisera for typing reported that out of 293 K. pneumoniae isolates, 88 (30%) could not be typed, and 54 (18%) had a positive reaction for more than one capsular type 41 . Klebsiella capsules can be typed using different methods, and each has some limitations that can preclude serotype determination in clinical isolates 42 . The presence of CPS variants like that described in the present study may provide another explanation. Serological diagnosis is commonly used to determine Klebsiella capsular serotypes of clinical isolates often exhibit different sensitivities. Varied acetylation of CPS may render some clinical strains less immunoreactive to antiserum. Since modifications increase the complexity of the capsule, investigators may need to employ more than one approach to determine the type accurately.
In summary, DNA sequence analysis of the cps cluster permitted identification of a candidate CPS acetyltransferase-encoding gene with distinct alleles in different K57 strains. We correlated this genetic variation with acetylation-mediated changes in K. pneumoniae PLA-related capsular serotype K57. We further showed the impacts of acetylation on K. pneumoniae CPS immunoreactivity, host innate response, serum resistance, and host cell adhesion. We suggest that acetylation contributes to the diversity and antigenic heterogeneity of the capsule and K. pneumoniae-host interactions. Our findings increase our knowledge about the variety of Klebsiella CPS and provide further insights into the role of capsular modification in bacterial pathogenesis.

Methods
Bacterial strains, cps-PCR, and cps sequencing. The Ref-K57 strain, 4425/51, was purchased from the Statens Serum Institute (Copenhagen, Denmark). Twenty-two K. pneumoniae K57 clinical isolates, including PLA and bacteremia strains, were collected from the National Taiwan University Hospital, En Chu Kong Hospital, Chang Gung Memorial Hospital, and Taipei Veterans General Hospital in Taiwan 14,19,43,44 , and were gifts from Hong Kong and Finland 14,44 . The K57 capsular type of the clinical strains was determined using wzy-targeting, PCR-based cps genotyping 14,43 . PCR analysis of orf13, the cps gene encoding a putative acetyltransferase, was performed using the following primers: K57-orf13-F (atgggtaaaaatatcaaagagcg) and K57-orf13-R (ttagaataggaaccataatcttttcc).
For cps sequencing, the cps region of the Ref-K57 strain (between the conserved genes, galF and gnd) was amplified as previously described 14 . The products were sequenced by primer walking, providing DNA sequences for the cps regions. Genes were annotated by NCBI-BLAST.
Analysis of CPS acetylation. The degree of acetylation was quantified according to a modified Hestrin method 24,25 . Extracellular polysaccharides were extracted with hot phenol as previously reported 19 and further purified by ethanol re-precipitation. Each sample was resuspended in water (0.5 mL) and combined with 1 mL of alkaline hydroxylamine (1M hydroxylamine in 1.75M NaOH), and the mixture was incubated for 30 min before addition of 0.5 mL each of 4M HCl and 0.37M FeCl 3 in 0.1M HCl. Absorbance was measured at 504 nm (A 504 ). For comparison, data are presented as A 504 normalized to 50 μ g of CPS for each sample.
To analyze potential acetylation of K. pneumoniae CPS using 1 H NMR 23 , the O-antigen-lacking mutants, Ref-K57Δ wecA and A1142Δ wecA, were constructed using an unmarked deletion method with the temperature-sensitive plasmid pKO3-Km 45 to avoid the effects of sugars from lipopolysaccharide (LPS). K. pneumoniae CPS was isolated according to the previously described methods 23 with modifications as needed. Briefly, bacteria suspended in distilled water were heated for 10 min at 100 °C to release capsular materials; CPS was precipitated with 80% (v/v) acetone at 4 °C. Dried precipitate was resuspended in 20mM Tris-HCl, pH 7.5, and treated with ribonuclease, deoxyribonuclease I, and proteinase K. The sample was then dialyzed extensively against water using an 80-kDa cutoff membrane and lyophilized. The CPS was further purified on a TSK HW-65F column, followed by dialysis and re-lyophilization. Sugar composition was determined by methanolysis and trimethylsilylation, followed by GC-MS analysis 46 .
Quantification of CPS. K57 CPS, which contains uronic acid, was quantified according to previously described methods 47,48 . Briefly, extracted samples from the equivalent amounts of bacterial overnight cultures were resuspended in 0.1 mL of water and combined with 1.2 mL of 12.5 mM tetraborate in concentrated H 2 SO 4 . After vigorous vortexing, the mixture was boiled for 5 min. After cooling, 20 μ L of 0.15% 3-hydroxydiphenol (Sigma-Aldrich, St. Louis, MO) was added. Then, the absorbance at 520 nm was measured.

Immunoblotting of CPS.
For immunoblot analysis of CPS, rabbit anti-K57 antiserum was purchased from the Statens Serum Institute. Extracted polysaccharide samples were vacuum-spotted onto a nitrocellulose membrane using a slot blot device. The membrane was overlapped with a piece of filter, and both were rinsed with Western transfer buffer containing 47.8 mM Tris, 38.6 mM glycine, 20% MeOH, and 0.037% sodium dodecyl sulfate. The membrane was dried, and non-specific sites were blocked by soaking the membrane in 1× phosphate-buffered saline with 0.5% Tween 20 (PBST) plus 5% milk for 1 h at room temperature. The membrane then was incubated with anti-K57 antiserum (1:1000 dilution) dissolved in PBST plus milk at 4 °C overnight, washed four times with PBST for 10 min each, incubated with the secondary antibody conjugated with horseradish peroxidase (goat anti-rabbit IgG-HRP, 1:10 000) for 1 h at room temperature, and washed three times with PBST for 10 min each. The ECL reagent was added for 3 min, and the membrane was exposed to X-ray film in the dark. To quantify the immunoreactivity, the resulting bands were analyzed by densitometry using Image J software (NIH, USA).

Gene replacement and complementation in K. pneumoniae. The isogenic replacement mutant of
Ref-K57 was constructed using the temperature-sensitive plasmid, pKO3-Km 43,45 . In this replacement mutant, the coding region of the putative acetyltransferase (orf13) in the cps locus of Ref-K57 was replaced by the corresponding gene (disrupted by an ORF encoding a putative transposase) from A1142 (residues 15933-17975). For chromosomal complementation of orf13 in the cps cluster of A1142, a single copy of the Ref-K57 orf13 was cloned into the intergenic region between wbaZ and gnd using the pKO3-Km vector according to a previously reported method 49 . The gene replacement or complementation was confirmed by PCR analysis and DNA sequencing.
Analysis of pro-inflammatory cytokine secretion and cell signaling. RAW 264.7 murine macrophage-like cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1× penicillin/streptomycin (PAA) and maintained at 37 °C in a humidified incubator at 95% air-5% CO 2 . Culture medium and supplements were obtained from GIBCO/BRL (Gaithersburg, MD, USA). For infection, cells (1 × 10 6 cells per well) in FBS-free DMEM medium in 24-well plates were infected with mid-log-phase (A 600 = 0.4 to 0.6) K. pneumoniae at a multiplicity of infection of 10 (MOI 10; bacteria/cells). After 1 h of infection, the cells were treated with 100 μ g/mL of gentamicin and supernatants were collected at various time points as indicated. The levels of mouse TNF-α and IL-6 in culture supernatants were measured Scientific RepoRts | 6:31946 | DOI: 10.1038/srep31946 by enzyme-linked immunosorbent assay (ELISA) kits (R&D Systems, Minneapolis, MN, USA) according to the manufacturer's instructions.
For Western blot detection of cell signaling, RAW 264.7 cells were harvested after K. pneumoniae induction (as described above). Total cellular protein extracts were obtained by lysing cells in ice cold RIPA buffer, followed by centrifugation and protein quantification using a Coomassie blue protein assay kit (Bio-Rad, Hercules, CA, USA). Equal amounts of the protein extracts were subjected to SDS-PAGE electrophoresis and subsequent electrotransfer onto a PVDF membrane. After blocking with 5% skimmed milk in PBST for 1 h at room temperature, the blots were incubated with primary antibodies overnight at 4 °C. Antibodies against phospho-ERK1/2 (Thr202/185 and Tyr204/187), phospho-p38 MAPK (Thr180/Tyr182), phospho-JNK (Thr183/Tyr185), phospho-NF-κ B p65 (Ser536) and those against respective total proteins were obtained from Cell Signaling (Beverly, MA, USA). Antibodies against β -actin were from Santa Cruz Biotechnology (Santa Cruz, CA, USA). After washes with PBST, the blots were incubated with HRP-conjugated secondary antibodies (Jackson Immuno Research Laboratories, West Grove, PA, USA), and the signals were detected using a digital imaging system (UVP, Upland, CA, USA).
In vivo competition assays. In vivo competition of K. pneumoniae in mice was analyzed as outlined in previous work 49 . Briefly, the parental A1142 or A1142::Ref-K57orf13 strains were mixed with the isogenic lacZ promoter deletion mutant (A1142Δ placZ) at a 1:1 ratio, and a per-mouse dose of 5 × 10 6 colony-forming units (CFUs) in 100 μ L of saline solution was inoculated intraperitoneally into 5-week-old BALB/c mice. Each mouse was killed 24 h post-inoculation, and the liver and spleen were removed and homogenized in 1× PBS. The number of LacZ-positive and LacZ-negative colonies on LB plates containing 1 mM IPTG/mL and 50 μ g/mL of X-Gal were counted. The competitive index (CI) was defined as (outputΔ placZ/output test strain )/(inputΔ placZ/input test strain ). Serum resistance assays. The serum resistance of the K. pneumoniae strains was analyzed as previously described 44 . Briefly, an inoculum of 2.5 × 10 4 CFU bacteria in 25 μ L was mixed with 75 μ L human serum from healthy volunteers. The mixture was incubated at 37 °C for 1 h. After serial dilution and plating, the numbers of CFU were determined. The survival ratio was calculated; values of ≧ 1 were defined as serum resistance. The experiments were repeated independently three times.
Cell adhesion assays. Adhesion assays using human enterocyte-like Caco-2 cells were performed according to previously described methods 50 . Caco-2 cells were maintained in DMEM medium supplemented with 10% heat-inactivated FBS and 1% nonessential amino acids. For adhesion assays, the cells seeded in 24-well plates (~5 × 10 5 cells per well) and were prewashed with Hanks' Balanced Salt Solution (HBSS). Mid-log-phase (A 600 = 0.4 to 0.6) K. pneumoniae in FBS-free DMEM medium were added to each well at a MOI of 50. After a 20-min incubation in a humidified 5% CO 2 atmosphere at 37 °C, the wells were washed with HBSS three times, and bacteria were released by the addition of 0.2% Triton X-100 (Sigma-Aldrich). The experiments were conducted in duplicate and repeated independently three times.
Quantitative real-time reverse-transcription PCR (RT-qPCR). K. pneumoniae gene expression was determined by RT-qPCR 49 . Total RNA from K. pneumoniae strains was isolated using the RNeasy Mini kit (Qiagen, Valencia, CA, USA) with DNase I to remove contaminating genomic DNA. A 400-ng sample of purified total RNA was reverse-transcribed and amplified by PCR with SYBR green dye (Invitrogen) in an ABI 7900 thermocycler (Applied Biosystems, Foster City, CA, USA). For each gene, the calculated threshold cycle (Ct) was normalized to the Ct of the 23S ribosomal RNA gene from the same complementary DNA sample. The relative RNA expression was calculated based on the Δ Δ Ct value. Primers used are listed in Table S1.
Statistical analysis. Comparisons of mean values were assessed by a two-tailed Student's t test using Prism 5 (Graphpad) software. P values of < 0.05 were considered significant.