Identification of Two Regulators of Virulence That Are Conserved in Klebsiella pneumoniae Classical and Hypervirulent Strains

ABSTRACT Klebsiella pneumoniae is widely recognized as a pathogen with a propensity for acquiring antibiotic resistance. It is capable of causing a range of hospital-acquired infections (urinary tract infections [UTI], pneumonia, sepsis) and community-acquired invasive infections. The genetic heterogeneity of K. pneumoniae isolates complicates our ability to understand the virulence of K. pneumoniae. Characterization of virulence factors conserved between strains as well as strain-specific factors will improve our understanding of this important pathogen. The MarR family of regulatory proteins is widely distributed in bacteria and regulates cellular processes such as antibiotic resistance and the expression of virulence factors. Klebsiella encodes numerous MarR-like proteins, and they likely contribute to the ability of K. pneumoniae to respond to and survive under a wide variety of environmental conditions, including those present in the human body. We tested loss-of-function mutations in all the marR homologues in a murine pneumonia model and found that two (kvrA and kvrB) significantly impacted the virulence of K1 and K2 capsule type hypervirulent (hv) strains and that kvrA affected the virulence of a sequence type 258 (ST258) classical strain. In the hv strains, kvrA and kvrB mutants displayed phenotypes associated with reduced capsule production, mucoviscosity, and transcription from galF and manC promoters that drive expression of capsule synthesis genes. In contrast, kvrA and kvrB mutants in the ST258 strain had no effect on capsule gene expression or capsule-related phenotypes. Thus, KvrA and KvrB affect virulence in classical and hv strains but the effect on virulence may not be exclusively due to effects on capsule production.

K lebsiella pneumoniae is a Gram-negative bacterium capable of causing a wide range of infections such as urinary tract infections (UTI), sepsis, liver abscesses, and pneumonia (1)(2)(3)(4)(5)(6). Capsule, lipopolysaccharide (LPS), adhesion factors, and siderophores frequently emerge as the primary K. pneumoniae virulence determinants, with capsule being the most extensively studied. Currently, 134 capsule types have been identified in K. pneumoniae (7). One trait strongly associated with virulence in Klebsiella is the overproduction of capsule, which contributes to a hypermucoviscosity (HMV) phenotype (5,8). HMV strains are "string test" positive and tend to be hypervirulent (hv) (8,9). Our relatively limited knowledge of conserved virulence determinants and the high diversity of surface polysaccharides pose challenges for developing vaccines and new therapeutics (10)(11)(12).
While first characterized for its role in antibiotic resistance in Escherichia coli (13,14), the MarR (multiple antibiotic resistance regulator) family of transcriptional regulators is known to regulate the expression of genes encoding proteins involved in metabolic pathways, stress responses, and virulence factors (14)(15)(16)(17)(18)(19)(20). These proteins are characterized by a winged helix-turn-helix DNA binding domain and can both positively and negatively affect gene expression (21,22). Transcriptional regulation by these proteins often results in modifications of the bacterial cell surface (23), and several MarR family members have been linked to virulence in the Enterobacteriaceae. In Salmonella, SlyA regulates Salmonella pathogenicity island-2 genes and contributes to resistance to oxidative stress, bacterial survival within macrophages, and bacterial survival in a murine model of infection (17,18,(23)(24)(25)(26). RovA, a member of the MarR/SlyA family, regulates expression of inv (an adhesion and invasion factor) in the enteric pathogens Yersinia enterocolitica and Yersinia pseudotuberculosis (27)(28)(29) as well as expression of the psa locus of Yersinia pestis, the causative agent of bubonic and pneumonic plague (30).
An in silico comparative study found the copy number of marR-like genes to range from 2 to 11 within the Enterobacteriaceae, with an average of 5.9 genes (22). That same study identified some K. pneumoniae strains that have as many as 11 marR homologues (22). We hypothesize that this high number of marR-like genes contributes to the ability of K. pneumoniae to survive in a wide variety of environments, including the human host, and that a subset of these genes regulates expression of virulence phenotypes.
In this report, we describe the contribution of the MarR family to K. pneumoniae virulence in a murine pneumonia model. Strain KPPR1S (31), a derivative of ATCC 43816, was found to contain nine marR-like genes (32). We constructed insertion disruption mutations in each of the marR homologues of strain KPPR1S and tested them in our pneumonia model. Two of these genes, designated kvrA and kvrB, affected virulence. The impact of KvrA and KvrB on virulence in this hv strain is likely due at least in part to their effect on expression of capsule genes and the HMV phenotype. Importantly, these roles were conserved in another hv K. pneumoniae strain that produces a different capsule type, and kvrA is required for full virulence of a sequence type 258 (ST258) classical strain.

RESULTS
Two marR-like genes contribute to K. pneumoniae virulence. The MarR family of transcriptional regulators has been implicated in virulence in several members of the Enterobacteriaceae (17,21,24,30,33,34). Klebsiella species contain more than the average number of marR-like genes, and we hypothesize that some members of this family are important for adaptation of K. pneumoniae in the mammalian host. Thus, we screened the genome of KPPR1S (wild-type [WT] strain) for putative marR genes and constructed loss-of-function mutants for each of the nine marR homologues identified. Growth curves (determined using optical density at 600 nm [OD 600 ] and CFU counts per milliliter) indicated that none of the mutants displayed a growth defect in vitro in Luria-Bertani (LB) medium (data not shown). To assess the effects on virulence, mice were intranasally (i.n.) infected with the WT strain or with each mutant individually and sacrificed at 48 h postinoculation (hpi) for bacterial enumeration. Two mutants, VK055_0496 and VK055_4504, displayed a decrease in bacterial burden in the lungs of infected mice compared to WT-infected mice (Fig. 1A). The spleens of mice infected with the WT had nearly 10 6 CFU/g of tissue, while the VK055_0496 mutant was barely detectable (Fig. 1A).
On the basis of our initial screen, we determined that VK055_0496 and VK055_4504 are important for infection of the lung and named these regulators KvrA and KvrB (Klebsiella virulence regulator), respectively. Further kinetics experiments using in-frame deletion mutants of kvrA (VK277) and kvrB (VK410) indicated that at 24 hpi, the ΔkvrA mutant was barely detectable in the lung whereas mice infected with the ΔkvrB mutant had~1-log-lower levels of CFU/g than the WT. By 72 hpi, the bacterial burdens of mice infected with the WT had increased several logs, but the ΔkvrA mutant was undetect-  able, and the bacterial burden of the ΔkvrB mutant remained comparable to that seen at 24 hpi (Fig. 1B). The spleens of ΔkvrA mutant-infected mice had few recoverable CFU at either 24 or 72 hpi, while the mice infected with the ΔkvrB mutant had splenic burdens that were more than 2 logs lower than the levels seen with the mice infected with the WT at 72 hpi (Fig. 1B). To test if the survival defect of the mutants was due to an inability to reach the lung, mice were infected with the WT or ΔkvrA or ΔkvrB strain and the lungs were harvested at 90 min postinoculation. The three strains displayed comparable bacterial burdens in the lungs at that early time point, suggesting that the ability of the mutants to initially reach the lungs was not impacted (Fig. 1C). Subsequent analysis indicated that the virulence defect of the ΔkvrA and ΔkvrB mutants could be attributed to the loss of KvrA and KvrB, as demonstrated by complementation of the ΔkvrA and ΔkvrB mutants (Fig. 1B). Together, these results demonstrate that KvrA and KvrB contributed to K. pneumoniae virulence in a lung model of infection. The ⌬kvrA and ⌬kvrB mutants induced an altered immune response. Infection with K. pneumoniae is known to induce a significant inflammatory response marked by lung lesions and neutrophil infiltration (35)(36)(37). Neutrophils contribute to bacterial clearance of some K. pneumoniae strains but not others (38,39). Neutrophils are considered important for clearance of ATCC 43816 (i.e., KPPR1S); however, high numbers of neutrophils also may be detrimental to host health (39). Inflammatory monocytes also are considered a primary cell type functioning in defense against K. pneumoniae (38). Consistent with the reduced number of CFU in the lung, infection of mice with the ΔkvrA and ΔkvrB mutants resulted in less inflammation than infection with the WT at 72 hpi ( Fig. 2A). To more carefully examine the histological changes following infection, the lungs from mice inoculated with the WT or ΔkvrA or ΔkvrB strain or subjected to mock inoculation (1ϫ phosphate-buffered saline [PBS]) were processed at 24 hpi for flow cytometric analysis. Infection with all three strains resulted in a significant increase in levels of neutrophils compared to the mock infection results (Fig. 2B). Although the infections with the two mutants resulted in elevated levels of neutrophils compared to the mock infection results, the increase in the levels of neutrophils was significantly lower than that seen with the WT. Among the cell populations examined, there was a 3-fold increase in the percentage of inflammatory monocytes in the mice infected with the ΔkvrA mutant compared to the mice infected with the WT (Fig. 2B). Inflammatory monocytes have been implicated in protection from K. pneumoniae (38), and the higher proportion of this cell type may have contributed to the rapid clearance of the ΔkvrA mutants. The ⌬kvrA and ⌬kvrB mutants showed increased associations with murine BMDMs. Because of the reduced bacterial burden of the ΔkvrA and ΔkvrB mutants in mice, we wanted to determine if these strains had altered interactions with murine macrophages. The ΔkvrA and ΔkvrB mutants were inoculated onto murine bone marrow-derived macrophages (BMDMs) in culture and were assessed for adherence to and internalization into these host-derived cells. A capsule mutant, manC (VK60), was included as a control as it is known to be more adherent and more readily phagocytosed than the WT (40,41). Only about 3% of the WT bacteria adhered to the BMDMs, whereas more than 35% of the manC mutant bacteria were cell associated (Fig. 3A). The ΔkvrA and ΔkvrB mutants were 4-to-6-fold more adherent than the WT (19% and 16%, respectively). A similar trend was observed when these strains were assayed for internalization by BMDM in a gentamicin protection assay. About 1% of the WT and 7% of the manC bacteria were internalized, demonstrating the antiphagocytic properties of capsule (Fig. 3B). The ΔkvrA and ΔkvrB mutants were internalized at levels of about 3% and 2%, respectively. These intermediate adherence and internalization phenotypes suggest that the ΔkvrA and ΔkvrB strains have altered interactions with host cells.
KvrA and KvrB contribute to capsule production. During mutant construction, we observed that colonies of the ΔkvrA and ΔkvrB mutants appeared to be less mucoid and less hypermucoviscous. Thus, we hypothesized that capsule production levels would be decreased in the mutants. Glucuronic acid is a key component of many different capsules, including the K2 capsule produced by KPPR1S. Measurement of uronic acid content is therefore frequently used to quantify capsule production (42,43). We determined the uronic acid concentrations in the ΔkvrA and ΔkvrB mutants along with the WT and the manC mutant. The manC mutant produced about 1.5 g uronic acid/OD 600 , which is about 25% of the level produced by the WT (Fig. 4A). The ΔkvrA and ΔkvrB mutants each produced uronic acid at only~60% of the WT levels, indicating that these mutants produce less capsule than the WT. Mucoviscosity can be measured using a sedimentation assay (44,45). HMV strains such as KPPR1S do not form tight pellets when centrifuged, and this can be quantified by measuring the OD 600 of the supernatant following low-speed centrifugation. The OD 600 of the WT was about 0.5, but the manC strain formed a tight pellet and the supernatant was essentially cleared, measuring 0.01 (Fig. 4B). The ΔkvrA and ΔkvrB mutants also formed tight pellets with cleared supernatants. These data, consistent with the reduced uronic acid levels, indicate that the ΔkvrA and ΔkvrB strains have reduced mucoviscosity.
KvrA and KvrB positively regulate capsule gene expression. Given the observations indicating that ΔkvrA and ΔkvrB have reduced hypermucoviscosity and produce less uronic acid than the WT, it is likely that expression of the capsule locus (cps locus) is reduced in these mutants. We constructed transcriptional gfp fusions to the known capsule promoters located upstream of galF, wzi, and manC (also known as cpsB) (43) and transformed these into the WT and mutant strains (Fig. 5A). The wzi promoter appeared to be affected only minimally by the loss of KvrA or KvrB. However, the levels associated with the galF and manC promoters were significantly decreased (20% to KvrA and KvrB are found in most other K. pneumoniae strains, and thus we wanted to determine if KvrA and KvrB have conserved roles in different strains. Because of the high prevalence of the K1 serotype among hv isolates associated with communityacquired liver abscess infections (5,46), we chose to test NTUH-K2044, which was isolated from a liver abscess (47). We constructed insertional mutations in kvrA (VK606) and kvrB (VK607) and performed quantitative reverse transcription-PCR (qRT-PCR) analysis on galF, wzi, and manC in both the NTUH-K2044 and KPPR1S backgrounds. In the KPPR1S strains, we observed that the levels of expression of the galF and manC genes in the ΔkvrA and ΔkvrB strains were significantly reduced, consistent with the gfp reporter data (Fig. 6A). The levels of the galF and manC genes were also significantly reduced in the NTUH-K2044 kvrA and kvrB mutants (Fig. 6B). This analysis also revealed that wzi expression was significantly reduced in the NTUH-K2044 mutants (Fig. 6B) and appeared to be slightly reduced in the KPPR1S mutants as well, although the results were not statistically significant (Fig. 6A). These data indicate that KvrA and KvrB regulate expression of several promoters in the cps locus from strains with at least two different capsule types, K1 and K2.
Both type K1 and type K2 capsules incorporate mannose, and this is true of~73% of capsule types; most of the remaining capsule types contain rhamnose, and some contain both mannose and rhamnose (48). Sequence type 258 (ST258) strains have recently been shown to be associated with many nosocomial infections caused by carbapenem-resistant K. pneumoniae, including an outbreak at the NIH (49). One such ST258 clinical isolate, KPNIH1 (49), has a cps locus containing genes that produce a rhamnose subunit, and it harbors kvrA and kvrB. MKP103, a carbapenem-sensitive derivative of KPNIH1, was constructed and was used to generate an ordered library of transposon mutants (50). The effect of kvrA (VK402) and kvrB (VK404) mutations on expression of the galF, wzi, and rmlB genes was tested in the MKP103 strain using qRT-PCR (Fig. 6C). Interestingly, although the galF and wzi promoters are virtually identical among all three strains, KvrA and KvrB did not appear to regulate these promoters in MKP103. One possible explanation is that kvrA or kvrB is not expressed in this strain; we examined this possibility by qRT-PCR and verified that both genes are indeed expressed (data not shown). Thus, these data suggest that the roles of KvrA and KvrB in regulating expression of galF and wzi expression are indirect. The rmlB promoter is quite different from the manC promoter, and KvrA and KvrB did not appear to regulate this promoter in MKP103.  Regulators of Klebsiella Virulence ® In addition to capsule regulation, we analyzed capsule production and mucoviscosity of the NTUH-2044 and MKP103 strains. As with KPPR1S, we found that uronic acid content was significantly reduced in both the NTUH-K2044 kvrA and kvrB mutants tõ 60% of the level seen with the parental strain (Fig. 7A). Similarly, both the kvrA and the kvrB mutants of NTUH-K2044 had reduced mucoviscosity (Fig. 7B). Consistent with the qRT-PCR data, KvrA and KvrB affected neither uronic acid content nor mucoviscosity in MKP103 (Fig. 7). However, MKP103 does not produce as much capsule as either KPPR1S or NTUH-K2044 and is not HMV. Together, these data suggest that the role of KvrA and KvrB in capsule production may be conserved in hv strains.
Contribution of kvrA and kvrB to virulence in NTUH-K2044 and MKP103. K. pneumoniae strains display significant genetic heterogeneity, and not all identified virulence factors are conserved between Klebsiella clinical isolates. Furthermore, the relative contributions of virulence factors to infection and pathogenesis can be dependent on the strain background (7,48,51,52). However, kvrA and kvrB homologues are conserved in K. pneumoniae strains and thus could have a conserved role in virulence. We therefore tested the kvrA and kvrB loss-of-function mutants in the NTUH-K2044 and MKP103 strains in a mouse model of pneumonia. As with the KPPR1S strain, the NTUH-K2044 kvrA mutant was strongly attenuated in the lung and did not establish infection in the spleen (Fig. 8A). Similarly to the KPPR1S kvrB mutant, the NTUH-K2044 kvrB mutant initially colonized the lung at levels that were lower than the WT levels but was present at levels similar to the WT levels by 72 hpi (Fig. 8A). In contrast to both of the WT strains and the kvrA mutant, the kvrB mutant was detectable in the spleens of most mice by 24 hpi, and by 72 hpi the burden of the kvrB mutant in the spleen had increased but was still several logs lower than the WT level (Fig. 8A). These data suggest that both KvrA and KvrB contribute to the virulence of hv K. pneumoniae strains with different types of capsule.
Classical strains such as the ST258 strains prevalent in nosocomial outbreaks of carbapenem-resistant strains do not exhibit the same level of virulence in mouse models as the hv strains (39,53). Nevertheless, when administered at a high dose (10 7 CFU), these classical isolates can initially colonize the lungs of mice but are then cleared within 48 to 72 hpi (39,53). Because kvrA and kvrB are part of the core K. pneumoniae genome, we assessed the virulence of the kvrA and kvrB mutants in MKP103. At 24 hpi, the kvrB mutant behaved similarly to the WT strain, whereas the kvrA mutant colonized the lungs of these mice at a significantly reduced level (Fig. 8B). Bacterial burdens in the spleen were very low for all three strains, but the kvrA mutant appeared to be defective for spleen colonization relative to the WT and kvrB strains (Fig. 8B). This trend for the kvrA mutant was significant (P Ͻ 0.05) in analyses of only those mice with detectable CFU. C57BL/6 mice were used in these studies, and it is possible that the kvrA mutant, as well as the kvrB mutant, would have a stronger phenotype in a host comparable to the immunocompromised patients that are typically infected with classical strains. Nevertheless, it appears that kvrA is important for the virulence of the representative hv and classical strains tested here.

DISCUSSION
The MarR family of transcriptional regulators has been implicated in regulation of virulence genes in several members of the Enterobacteriaceae (17,18,21,24,26,34,54). This work examines the contribution of the MarR family to K. pneumoniae virulence by first screening loss-of-function mutations in nine marR-like genes in a K2 hv strain (KPPR1S) using a mouse model of pneumonia. Two uncharacterized transcriptional regulators, KvrA and KvrB, were found to be important for virulence in this model. The Regulators of Klebsiella Virulence ® kvrA mutant was essentially avirulent. Inflammatory monocytes have been shown to be critical for clearance of a K. pneumoniae ST258 strain by producing tumor necrosis factor alpha (TNF-␣) that then recruits interleukin-17 (IL-17)-producing innate lymphocytes (38), and IL-17 has also been shown to control infection with ATCC 43816 (55), the strain from which KPPR1S was derived. Thus, the increased number of inflammatory monocytes observed after infection with the ΔkvrA mutant was likely contributing to the rapid clearance of this mutant. Mutations in kvrA and kvrB in a K1 hv strain (NTUH-K2044) also resulted in a virulence defect in this pneumonia model. In addition, mutations in kvrA and kvrB resulted in increased adherence to and uptake of KPPR1S by murine BMDM. These phenotypes are consistent with those of capsule-deficient strains (56), and indeed, KvrA and KvrB were found to regulate capsule gene expression from the galF and manC promoters in KPPR1S and NTUH-K2044. However, kvrA and kvrB mutations in classical ST258 strain MKP103 did not affect expression from the conserved galF promoter. Whether these effects of KvrA and KvrB on expression from the manC and galF promoters are direct or indirect is not known. The lack of an effect on galF expression in the MKP103 kvrA and kvrB mutants suggests that the effect(s) on galF expression in the hv strains may be indirect, acting through a protein present only in hv strains. The contribution of the MarR family to the virulence of K. pneumoniae may be both strain and infection site dependent. Previously, the role of the MarR family protein, PecS, was shown to be important for repressing the type 1 fimbrial gene locus in K. pneumoniae strain CG43 (57). Type 1 fimbriae have been implicated in binding to mannose-containing structures and are important for causing UTI (58). A highthroughput genetic screen for virulence factors that are important for a pneumonic infection found that the pecS mutant was attenuated only modestly in a coinfection experiment (59). The pecS ortholog (geneID VK055_4417) was targeted in our initial screen and found to be dispensable for virulence in the lung. This is consistent with a previous study where an mrk (type 3 fimbrial) mutant and an mrk fim double mutant were not attenuated in a lung model of Klebsiella infection (60) even though they were attenuated in a UTI model (61). While our study demonstrated that PecS is dispensable for KPPR1S infection in the lung, it may serve an important role in other infection sites such as the urinary tract.
The bacterial capsule serves a variety of functions, including, but not limited to, protection from the immune response of the host (56,(62)(63)(64). Capsule synthesis in K. pneumoniae is homologous to assembly of the Wzy-dependent assembly of group 1 capsules of E. coli (65). Following polymerization, the polysaccharide chains are transported across the outer membrane and are thought to be anchored to the surface via the Wzi lectin (66,67). E. coli and K. pneumoniae are among the few gammaproteobacteria that encode Wzi proteins (67); in this study, using the sedimentation assay, a wzi mutant was no longer hypermucoviscous.
Our study has revealed a role for KvrA and KvrB in regulation of sugar production for capsule synthesis in type K1 and type K2 strains as determined by a reduction in expression of the galF and manC promoters and a reduction in uronic acid levels. Furthermore, hypermucoviscosity, a property associated with hv strains (but not classical strains) and generally associated with more abundant capsule in the literature (5,9,68), was significantly reduced in both the kvrA and kvrB mutants in the hv strains. However, it is unclear whether cell-associated capsule or released extracellular polysaccharide or both are responsible for the changes in hypermucoviscosity. The difference between surface-associated and released capsule polysaccharides may reflect a difference in function. For instance, surface-associated capsule can inhibit complement deposits and thus evade the membrane attack complex, and it can block attachment to phagocytic cells (69). Free extracellular capsule may act as a decoy to sequester antimicrobial peptides produced by the host (70). Therefore, it is possible that the reduction in uronic acid and mucoviscosity that we observed may reflect a change in the ratio of free and bacterial-surface-associated saccharides. On the basis of previous studies (67), this change may be mediated by Wzi, as we observed a modest decrease in expression from the wzi promoter. However, it could also be due in part to the reduction in synthesis of precursor sugars via reduced expression from the galF and manC promoters or to altered expression of unknown genes that influence mucoviscosity.
The primary sequences of the MarR/SlyA/RovA proteins are highly conserved among different bacteria; however, the degree of similarity between different MarR-family homologues can vary substantially. For instance, while KvrA shares 84% identity with SlyA of Salmonella enterica (71) and KvrB shares 93% identity with EmrR of E. coli (72), KvrA and KvrB share only 31% identity. A consensus DNA-binding sequence has been difficult to identify for many of these family members, and regulation by this class of regulators appears to be largely mediated by derepression of H-NS transcriptional silencing (73). H-NS regulates many horizontally acquired genes, and this regulation has been shown in many instances to be counteracted by regulators of the MarR family (27,(74)(75)(76). Thus, the regulons of MarR regulators can differ between closely related bacterial species and even between strains of the same species (28).
We found that KvrA and KvrB are necessary for capsule expression, and, given the comparable capsule production defects but different disease phenotypes of kvrA and kvrB mutants, we believe that KvrA and KvrB may be regulating genes in addition to the cps locus, and efforts to define the regulons are currently under way in our laboratory along with studies to examine direct versus indirect regulation. As less than 40% of genes are conserved across different K. pneumoniae strains (52), identifying highly conserved targets affecting growth and survival of K. pneumoniae in the host is essential for overcoming strain heterogeneity as new therapeutics are developed. Current studies are under way to characterize the KvrA and KvrB regulons in order to gain a fuller understanding of how these proteins influence capsule production and virulence.

MATERIALS AND METHODS
Ethical statement. Mouse experimental procedures were conducted in accordance with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All animal studies were approved by the Institutional Animal Care and Use Committee of the University of North Carolina (UNC) at Chapel Hill. All efforts were made to minimize suffering. Animals were monitored daily following inoculation and were euthanized upon exhibiting signs of morbidity.
Construction of bacterial mutants. The primers used for the construction of mutant strains are listed in Table 2. In-frame deletion mutants of kvrA, kvrB, and marR (VK277, VK410, and VK332, respectively) were generated by allelic exchange using pKAS46 as described previously (31,78). Fragments of~500 bp upstream and downstream of the targeted gene were amplified by PCR, cloned into pKAS46, and verified by sequencing, and the resulting plasmid was introduced into KPPR1S via conjugation. Transconjugants (Rif r /Kan r ) were streaked onto LB-Strep agar plates to select for clones that had lost the vector sequences. The resulting colonies were screened for Kan sensitivity and then for loss of the targeted gene by PCR.
Insertional disruption mutants (VK279, VK322, VK323, VK324, VK325, VK280, VK327, and VK328) were constructed as follows. A DNA fragment containing an internal portion of the gene of interest was generated by PCR, cloned into pKAS46, and verified by sequencing, and the resulting plasmid was introduced in KPPR1S via conjugation. The resulting Rif r /Kan r colonies were selected for experimentation. For NTUH-K2044 0496::pKAS46 and NTUH-K2044 4504::pKAS46, transconjugants were selected on LB plates with chloramphenicol (25 g/ml) and Kan 50 . Mutants of kvrA and kvrB were identified in MKP103 from a library of transposon mutants and were kindly provided by C. Manoil (50).
Chromosomal complementation was done by allelic exchange essentially as described above for in-frame deletions. The plasmids for chromosomal complementation (pKAS46 kvrAcomp and pKAS46 kvrBcomp) were generated by amplifying a single fragment that spanned the same upstream and downstream regions as the deletion constructs using the outermost primers but that contained the full coding sequence. The insertions were cloned into pKAS46 and verified by sequencing. Following conjugation and resolution of the vector sequences, the resulting complemented strains were designated VK278 (kvrAcomp) and VK417 (kvrBcomp).
Histopathology. Groups of two or three mice were inoculated i.n. as described above. The mice were sacrificed at 72 hpi, and lungs were inflated with 1 ml of 10% neutral buffered formalin for at least 24 h. The lungs were transferred to PBS for 2 h and then immersed in 70% ethyl alcohol and embedded in paraffin. Three 5-m sections (spaced 200 m apart) per lung were stained with hematoxylin-eosin (H&E) for examination. Histology services were provided by the University of North Carolina (UNC) Center for Gastrointestinal Biology and Disease Histology Core.
Flow cytometry and antibodies. The lungs from K. pneumoniae-infected or mock-infected mice were minced into fine pieces with two surgical steel razor blades and then digested in 5.66 mg collagenase II (Gibco, Carlsbad, CA)-4 ml PBS-5% serum for 1 h at 37°C. Following digestion, the lung homogenates were triturated with an 18-gauge needle (three passages) and then passed through a 100-m-pore-size filter to generate a single-cell suspension.
BMDMs were seeded into 24-well plates at a density of 500,000 cells per well and incubated overnight at 37°C and 5% CO 2 . Bacterial cultures were grown in 2 ml LB with Rif 30 for 16 to 17 h at 37°C. Prior to the assays, the macrophages were rinsed and fresh DMEM-10% FBS was added. For the adherence assay, cells were pretreated for 1 h with 2 M cytochalasin D (Sigma, St. Louis, MO) to inhibit internalization of bacteria; this step was omitted for the internalization assays. Macrophages were then inoculated with the desired K. pneumoniae strains at a multiplicity of infection (MOI) of 50. After 1 h, the wells were gently rinsed with sterile phosphate-buffered saline (PBS) three times and then lysed with 0.5% saponin-PBS. The lysate was diluted and plated on LB agar to determine the number of CFU. Data are presented as percentages of the inoculum. For the internalization assay, cells were seeded and treated as described above, omitting the cytochalasin D treatment. Following a 1-h incubation with K. pneumoniae at an MOI of 50, the cells were rinsed with PBS three times, and then DMEM-10% FBS-200 g/ml gentamycin (Gibco) was added and the reaction mixture was incubated 30 min to kill extracellular bacteria. Following the gentamicin treatment, the cells were lysed with 0.5% saponin-PBS, diluted, and plated on LB agar to determine the number of CFU. Data are presented as percentages of the inoculum.
Construction of reporter fusions and measurement of promoter activity. DNA fragments containing the promoter region of manC, galF, and wzi from KPPR1S were amplified by PCR using Pfu Turbo and cloned into the gfp reporter plasmid pPROBE-tagless (79). The resulting plasmids, pPROBE-manC, pPROBE-galF, and pPROBE-wzi, were confirmed by restriction digest analysis and sequencing. All primers used for these clones are listed in Table 2.
To examine capsule gene expression in vitro, pPROBE-tagless, pPROBE-manC, pPROBE-galF, or pPROBE-wzi was introduced into KPPR1S, VK277, and VK410 by electroporation. These strains were grown overnight at 37°C in LB-Kan 50 , subcultured to an OD 600 of 0.2, and grown for 6 h. All strains were assayed in triplicate, and each assay was performed multiple times. Relative fluorescent units (RFU) were measured using a Synergy HT plate reader (Biotek Instruments, Winooski, VT), and the measurements were normalized to the culture OD 600 .
Mucoviscosity assay. The mucoviscosity of the capsule was determined using the sedimentation assay as previously described (31,80). Briefly, overnight cultures grown in LB were subcultured to an OD 600 of 0.2 in fresh media and grown at 37°C. After 6 h, the cultures were normalized to an OD of 1.0/ml and centrifuged for 5 min at 1,000 ϫ g. The supernatant was gently removed without disturbing the pellet for OD 600 measurement.
Extraction and quantification of capsule. Uronic acid content was extracted and quantified as previously described (31). Briefly, cultures were grown for 6 h as described above; 500 l was mixed with 100 l 1% zwittergent-100 mM citric acid, incubated for 20 min at 50°C, and centrifuged; and 300 l of supernatant was precipitated with 1.2 ml 100% ethanol. Following centrifugation, the pellet was dried and resuspended in 200 l distilled water (dH 2 O), and then 1.2 ml sodium tetraborate-concentrated H 2 SO 4 was added and the reaction mixture was subjected to vortex mixing, boiled for 5 min, and placed on ice for 10 min. Uronic acid was detected by addition of 20 l of 0.15% 3-phenylphenol (Sigma-Aldrich)-0.5% NaOH. After a 5-min incubation at room temperature, the absorbance at 520 nm was measured. The glucuronic acid content was determined from a standard curve of glucuronate lactone (Sigma-Aldrich, St. Louis, MO) and expressed as micrograms per OD 600 .
RNA extraction and qRT-PCR analysis. Overnight cultures of each K. pneumoniae strain were subcultured to an OD 600 of 0.05 and grown for 2 h in LB (with antibiotics, when necessary). A volume of 3 ml was mixed with 600 l 1% zwittergent-100 mM citric acid in 2 microcentrifuge tubes and incubated for 10 min at room temperature to facilitate pelleting (81). Total RNA was extracted with Trizol reagent (Invitrogen, Carlsbad, CA) following the manufacturer's instructions, with the following additional step. A volume of~100 l 0.1-mm-diameter silica beads (BioSpec Products, Bartlesville, OK) was added with the Trizol reagent, and the samples were placed in a Precellys 24 homogenator (Bertin Technologies) for 3 min at 5,300 rpm prior to chloroform addition to facilitate lysis. Contaminating DNA was removed with DNA-free Turbo using the manufacturer's rigorous protocol and 1-h incubations. CDNA was synthesized using an iScript cDNA synthesis kit (Bio-Rad, Hercules, CA) with 1 g RNA as the template in a 20-l reaction volume, and then the samples were diluted 10-fold with RNase-free dH 2 O. PCR was performed in a 20-l reaction volume containing 5 l cDNA, a 500 nM concentration of each primer, and 10 l SsoAdvance SYBR green Supermix (Bio-Rad) using a CFX96 RealTime system (Bio-Rad). The relative transcript levels for the target genes were normalized to the gyrB gene and calculated using the threshold cycle (2 ϪΔΔCT ) method (82). The data represent results from an average of 6 biological replicates; 3 of each were grown and processed on separate days.
Statistical analysis. Statistical analyses were performed using GraphPad Prism, version 7.0 (San Diego, CA). Mouse infection data were analyzed using the Mann-Whitney test. For all other experiments, ordinary one-way analysis of variance (ANOVA) tests (Dunnett's multiple-comparison tests) were applied.