Functional Characterization of Plasmid-Borne rmpADC Homologues in Klebsiella pneumoniae

ABSTRACT Expression of the hypermucoviscosity (HMV) phenotype and capsular polysaccharide (CPS) biosynthesis in Klebsiella pneumoniae were reported to be encoded by genes located in the chromosomal rmp locus. However, the functions of the rmp locus in the virulence plasmid remained unclear, and most of the rmp loci in clinical K. pneumoniae are plasmid carried. In this study, we investigated the functional characteristics of plasmid-borne rmp homologues in clinical hypervirulent K. pneumoniae (hvKP) strains by cloning and introducing such gene homologues into K. pneumoniae strains of different capsule types, followed by the evaluation of phenotypic changes in these strains. Acquisition of the plasmid-borne prmpADC and prmpA2D2 loci were found to result in an increase in mucoviscosity and CPS production in K1 and K2 K. pneumoniae, while only the prmpA2D2 locus contributed to phenotypic changes in the ST11/KL64 strain. Consistently, both rmpD and rmpD2 increased HMV in K1 and K2 K. pneumoniae, while only rmpD2 contributed to HMV in the ST11/KL64 strain; rmpC contributed to CPS overproduction in K1 and K2 strains but not in the ST11/KL64 strain. Furthermore, we proposed a logistic molecular basis of the HMV phenotype of K. pneumoniae on which prmpD2-mediated HMV is attributed to the increase of cell-free CPS production. Our data confirm that the rmp homologues carried by the virulence plasmid play a key role in virulence expression in K. pneumoniae, but the phenotype is highly dependent on the genetic background of the host strain and explained why most of the clinical ST11 strains carry only the prmpA2D2 locus. IMPORTANCE Klebsiella pneumoniae has become the most frequently isolated bacterial pathogen in hospital settings, with a very high mortality rate worldwide. Factors contributing to the virulence of K. pneumoniae are the overproduction of capsular polysaccharide (CPS) as well as the hypermucoviscosity (HMV) phenotype. These two phenotypes were reported to be regulated by rmpA/A2 homologues, which are often carried by virulence plasmids. Here, we determined the functional role of two plasmid-borne rmpA in mediating expression of the HMV phenotype and CPS production in K. pneumoniae. Different capsule types exhibited differences in the expression of HMV and CPS production although they harbored an identical plasmid-borne rmpA or rmpA2 locus, indicating that these virulence-related phenotypes are strongly related to the genetic background of the host strains. Our study provides a novel understanding of the regulation of virulence-related phenotypes and clinical management of K. pneumoniae infections.

K lebsiella pneumoniae is a human commensal and opportunistic pathogen that can cause severe hospital-acquired infections, especially among patients with a compromised immune system (1). In recent years, strains that exhibit a variety of drug resistance and virulence phenotypes have emerged, posing further challenges to the clinical treatment of infections caused by this important pathogen. Unlike the classical K. pneumoniae (cKP) strains, hypervirulent K. pneumoniae (hvKP) strains are less commonly associated with antibiotic resistance; such strains are considered community acquired as they cause infections in healthy hosts that manifest as pyogenic liver abscesses (2,3). Two features known to distinguish cKP from hvKP are the number of siderophore systems and the abundance of the capsular polysaccharide (CPS) that they harbor (1). Classical strains typically have one or two siderophore systems, whereas hvKP strains have three or four. In addition, hvKP strains are known to produce a very thick CPS which is associated with a hypermucoid phenotype known as hypermucoviscosity (HMV) (4). These functions are encoded by genes located in the virulence plasmid, which include the iuc (biosynthetic genes for the siderophore aerobactin), iro (biosynthetic genes for the siderophore salmochelin), and rmpA and rmpA2 (regulators of the mucoid phenotype) genes (5)(6)(7)(8).
The HMV phenotype of K. pneumoniae was previously thought to be due to the highlevel abundance of CPS in hvKP. However, recent studies demonstrated that, although expression of the HMV phenotype and CPS biosynthesis are closely related, they could be dissociable and each served distinct functions during pathogenesis (9). Walker et al. identified an rmp operon that comprised the rmpA, rmpD, and rmpC genes in strain KPPR1S, in which the rmpA gene product autoregulates the operon, rmpD confers the HMV phenotype, and rmpC promotes CPS biosynthesis (10,11). However, the rmp locus of strain KPPR1S is located in the integrative and conjugative element (ICE) ICEKp1 in the chromosome, which is not a representative of the rmp homologues present in the majority of hvKP strains (12). The rmpA and rmpA2 genes are more frequently associated with the virulence plasmid (13,14). Furthermore, the virulence plasmid has undergone evolution to generate variants that carried only one of these regulators, as well as those which carried mutated rmpA/rmpA2 genes (15,16). However, the impact of these evolution processes on the virulence level of K. pneumoniae remained unknown. In this study, we investigated the functional features of rmp homologues in clinical hvKP strains by cloning and introducing various rmp homologues into clinical K. pneumoniae strains of various capsule types and assessed the level of virulence that they encoded. Our data will provide in-depth understanding of the roles of newly evolved plasmid-borne rmp homologues in virulence expression in clinical K. pneumoniae strains and hence provide important insights into the development of effective approaches to treat hvKP infections.

RESULTS
Plasmid-borne rmpADC homologues in clinical K. pneumoniae strains. The rmp locus was mainly located in the ICEKp1 element in the chromosome of K. pneumoniae but more commonly in large virulence plasmids. It was reported that the chromosomal rmp locus of strain KPPR1S comprised the rmpA, rmpD, and rmpC genes, which played different roles in the expression of the HMV phenotype and CPS biosynthesis (10,11). We then analyzed the rmp homologues in the virulence plasmids and determined whether their functional roles resembled that of crmp, since the plasmid-borne rmp locus is much more common in clinical K. pneumoniae isolates than those located in the chromosome. There were two rmpA/rmpA2 genes in the large virulence plasmids, namely, pK2044 and pLVPK, each located in a separate genetic locus (6). The prmpA locus was found to contain the rmpA, rmpD, and rmpC genes, resembling the crmp locus of strains KPPR1S and NTU2044, with 90% sequence identity. The prmpA2 locus comprised the rmpA2 and rmpD genes, and a truncated rmpC gene, sharing 78% sequence identity with the crmpA and prmpA loci (Fig. 1). For the sake of differentiation, the rmpD homologue located downstream of the rmpA2 gene was designated rmpD2. The rmpC genes were conserved among the crmp and prmp loci. The product of the prmpC gene from plasmid p15WZ-82_Vir exhibited 88% sequence identity with the product of the crmpC gene from the chromosome of strain KPPR1S (Fig. 2a). The rmpD genes were more variable. The product RmpD2 from plasmid pVir-CR-HvKP4 exhibited 58% sequence identity with pRmpD from plasmid pLVPK and 52% sequence identity with cRmpD from the chromosome of strain KPPR1S (Fig. 2b). The plasmids p15WZ-82_Vir and pVir-CR-HvKP4 reported in our previous studies were able to mediate the expression of HMV and virulence, as well as the overproduction of CPS in K. pneumoniae strains (17,18). Plasmid p15WZ-82_Vir was found to contain both the prmpADC and the prmpA2D2 loci, whereas plasmid pVir-CR-HvKP4 carried only the prmpA2D2 locus (Fig. 1). However, the prmpA2 gene in plasmid p15WZ-82_Vir was mutated in such a way that it encoded a truncated protein. These findings prompted us to further study the role of these plasmids in mediating the onset of hypervirulence phenotypes in K. pneumoniae. It should be noted that all K1/K2 strains and an increasing number of non-K1/K2 clinical HvKP strains that harbor a virulence plasmid contain both the prmpADC and the prmpA2D2 loci.
The rmpADC homologues mediate expression of hypermucoviscosity and CPS overproduction in clinical K. pneumoniae strains. To further determine the functional role of prmpA and prmpA2, these two genetic loci were cloned and ligated to vector pCE2 and then transformed into different K. pneumoniae strains. Introduction of the virulence plasmid p15WZ-82_Vir or pVir-CR-HvKP4, as well as the prmpA or prmpA2 locus, into strain WZ1-2 could in each increase the mucoviscosity (Fig. 3a). However, none of these genetic elements caused a significant increase in CPS production in strain WZ1-2 (Fig. 3b). Furthermore, as the HMV phenotype has been primarily associated with K1 and K2 serotypes, we also included a K1 K. pneumoniae strain, KP1088, and a K2 K. pneumoniae strain, PM8, as hosts in the FIG 1 Alignment of the crmpA, prmpA, and prmpA2 loci in clinical K. pneumoniae strains. There were two rmpA/rmpA2 genes in the large virulence plasmids pK2044 and pLVPK (6). The prmpA locus was found to comprise the rmpA, rmpD, and rmpC genes, resembling the crmp locus of strains KPPR1S and NTU2044, with 90% sequence identity, whereas the prmpA2 locus contained the rmpA2, rmpD2, and truncated rmpC genes and shared 78% sequence identity with the crmpA and prmpA loci.

Functional Study of rmpADC Homologues
Microbiology Spectrum functional study of the prmpA and prmpA2 clusters. Strains KP1088 and PM8 both carried a pLVPK-like virulence plasmid with prmpA and prmpA2 clusters (Fig. 1). The original virulence plasmids harbored by the strains KP1088 and PM8 were cured, and the plasmid-cured strains KP1088PC and PM8PC were found to exhibit low mucoviscosity and low-level CPS production ( Fig. 3a and b). Both the prmpA and prmpA2 clusters were found to cause a significant increase in mucoviscosity and CPS production in strains KP1088PC and PM8PC Functional Study of rmpADC Homologues Microbiology Spectrum ( Fig. 3a and b), even though rmpC is not present in the prmpA2 cluster. Introduction of the whole plasmid p15WZ-82_Vir, however, has a rather limited effect on these two phenotypes in these three tested strains. It is worth noting that the prmpA cluster exhibited a more prominent effect in virulence expression in clinical K1 strain KP1088PC and K2 strain PM8PC, whereas the effect of the prmpA2 cluster was more apparent in KL64 strain WZ1-2.
Regulation of hypermucoviscosity and CPS overproduction in clinical K. pneumoniae strains. To further investigate the role of the prmpA and prmpA2 loci in mediating expression of the virulence-related phenotypes, the genes prmpA, prmpD, prmpC, prmpA2, and prmpD2 were individually amplified and introduced into strains WZ1-2, KP1088PC, and PM8PC, separately. The results showed that only the prmpD and prmpD2 genes were able to encode the HMV phenotype (Fig. 3c); this finding is consistent with those of previous studies (11). To rule out the possibility that the HMV phenotype is the result of a high copy number of rmpD, a low-copy-number vector, pSC101, was ligated to the target gene and transformed into strain PM8PC. As Fig. S1 in the supplemental material shows, HMV was obtained in PM8PC/rmpD-pSC101, suggesting that a high copy number of rmpD is not a necessity for HMV. Furthermore, our results indicate that prmpD caused a significant increase in mucoviscosity in the K1 strain KP1088PC and the K2 strain PM8PC, but not the ST11/KL64 strain WZ1-2. However, prmpD2 was found to cause a dramatic increase in mucoviscosity in all these three strains (Fig. 3c). These data indicate that prmpD2 is functionally better adapted in K. pneumoniae isolates of various capsule types, including ST11 carbapenemresistant K. pneumoniae (CRKP). Regulation of CPS production is much more complicated. FIG 3 Mucoviscosity assay and uronic acid assay of ST11 CRKP strain WZ1-2, K1 strain KP1088, and K2 strain PM8 carrying different virulence determinants. Mucoviscosity (a) and uronic acid production (b) of strain WZ1-2 and its variants receiving virulence plasmid p15WZ-82_Vir, pVir-CR-HvKP4, the prmpADC and prmpA2D2 loci, strain KP1088, its virulence plasmid curing variant KP1088PC, and KP1088PC receiving the prmpADC and prmpA2D2 loci, and strain PM8, its virulence plasmid curing variant PM8PC, and PM8PC receiving the prmpADC and prmpA2D2 loci. Mucoviscosity (c) and uronic acid production (d) of strains WZ1-2, KP1088PC, and PM8PC with their variants receiving prmpA, prmpD, prmpC, prmpA2, and prmpD2. Virulence plasmids p15WZ-82_Vir and pVir-CR-HvKP4 were conjugated to strain WZ1-2. wt, wild type; ns, not significant. Each value is presented as the mean, and error bar indicates SD. *, P , 0.05, **, P , 0.01, ***, P , 0.001, ns, P $ 0.05, wt, wild type strain. Previous studies showed that crmpA and crmpC, but not crmpD, regulated CPS expression in K2 strain KPPR1S (10). Although each of the prmpA, prmpD, prmpC, prmpA2, and prmpD2 genes was found to cause upregulation of CPS production in K1 strain KP1088PC, only prmpD, prmpC, and prmpD2 could encode a significant increase in CPS production in K2 strain PM8PC (Fig. 3d). No significant changes in the level of CPS production were observed after introduction of both the prmpADC and prmpA2D2 loci, or the genes prmpA, prmpD, prmpC, prmpA2, and prmpD2, into ST11/KL64 strain WZ1-2 ( Fig. 3b and d). Our data confirmed that both pRmpD and pRmpD2 in the virulence plasmid could mediate the expression of the HMV phenotype in a manner similar to cRmpD (11). Furthermore, prmpD and prmpD2 also contributed to an increase in CPS production in K1 and K2 strains. Like crmpC, prmpC caused overproduction of CPS in K1 and K2 strains but not ST11/KL64 strain WZ1-2.
Correlation of hypermucoviscosity with CPS production. Next, we tried to visualize the capsule of variants harboring different genetic elements by Indian ink staining. The nonmucoid strains, such as CRKP strain WZ1-2 and the plasmid-cured strains KP1088PC and PM8PC, exhibited a thin-layered capsule, whereas the corresponding hypermucoid variants, such as KP1088PC/pRmpD and PM8PC/pRmpD, exhibited a large halo around the cell, indicating the presence of a much thicker capsule (Fig. 4a). However, a significant increase in capsule size could not be observed in variants that produced a larger amount of CPS without exhibiting HMV, such as KP1088PC/pRmpC and PM8PC/pRmpC (Fig. 4a). Variants that exhibited both HMV and CPS overproduction, such as KP1088PC/pRmpADC and PM8PC/ pRmpADC, possessed the capsule of the largest size, suggesting that HMV and overproduced CPS resulted in the formation of a superthick capsule (Fig. 4a). However, it should be noted that this phenotype was observed only in variants that have acquired the prmpADC locus, but not the prmpA2D2 locus. One hypothesis is that the HMV phenotype is correlated with cell-associated CPS. Previous studies on Streptococcus pneumoniae demonstrated that compared to other serotypes, hypermucoid serotype 3 isolates synthesize CPS by the synthase-dependent pathway, resulting in CPS release during growth in vivo and in vitro (19). Hence, we quantified the cell-released CPS and cell-associated CPS during in vitro growth of different K types and their variants. What was consistent among all three capsule types tested was that pRmpD2 drastically increased the releases of CPS into the culture medium, causing the cell-free CPS to be the majority of total CPS (Fig. 4b). These findings suggested that HMV correlated with cell-released CPS, rather than total CPS.

DISCUSSION
In this study, we determined the functional roles of two plasmid-borne rmpA homologues in mediating the expression of the HMV phenotype and CPS production in K. pneumoniae. The tested strains included a clinical ST11/KL64 CRKP strain, a virulence plasmid-cured K1 strain, and a virulence plasmid-cured K2 strain. The prmpA locus comprised the prmpA, prmpD, and prmpC genes, resembling the crmp locus of strains KPPR1S and NTU2044. The prmpA2 locus comprised the prmpA2 and prmpD2 genes, as well as a truncated prmpC gene. Our results showed that the two gene clusters produced similar phenotypes in K1 and K2 K. pneumoniae strains, although the prmpA locus was slightly more effective in upregulating the expression of HMV and CPS production in both K1 and K2 strains. In KL64 strain WZ1-2, the prmpA2 cluster encoded a more viscous phenotype than did the prmpA cluster. We reported for the first time that different capsule types exhibited differences in the expression of HMV and CPS production even though they harbored an identical prmpA or prmpA2 locus. The finding that both the prmpADC and prmpA2D2 loci contributed to increase of the mucoviscosity and CPS production in K1 and K2 strains, but only an increase of mucoviscosity in the KL64 strain, is intriguing.
The crmpD gene was previously reported to be a key factor underlying HMV expression in K2 strain KPPR1S (11). Here, we confirm that the functional role of prmpD and prmpD2 from the virulence plasmids in K1 and K2 strains is similar to that of their chromosomal counterpart. These observations are consistent with that of a previous report on clinical non-K1/K2 CR-HvKP strains that carried a virulence plasmid without the rmpA locus (15). Both the rmpA and rmpC genes were reported to regulate CPS biosynthesis (6, 10), yet we observed the regulatory effects of prmpA and prmpA2 in the K1 strain only and observed that prmpC played a regulatory role in both K1 and K2 strains, indicating that the impact of products of these genes on the phenotypes of HMV and CPS production is strongly dependent on the genetic background of the host strains, especially the capsule type.
Furthermore, our data suggested that the HMV phenotype correlated with the release of CPS into medium during growth. A higher CPS ratio in the culture supernatant than on the bacterial surface can lead to a higher antibody-absorbing capacity by the CPS in the culture supernatant (19). Therefore, lower efficiency might occur when anti-CPS vaccines are used for the treatment of hypermucoid K. pneumoniae. On the one hand, released CPS can absorb free unbound antibodies; meanwhile, antibody-bound CPS could also be released from bacteria, resulting in ineffective immunotherapy.
One limitation of this study is that the impact of prmpADC and prmpA2D2 loci on virulence in different genetic backgrounds of K. pneumoniae strains in vivo has not been investigated. The cRmpD has been confirmed to contribute to immune evasion and virulence in a K1 K. pneumoniae strain (11), while the virulence of these components in different genetic backgrounds (K types) of K. pneumoniae remains unknown. This work calls for further studies on the investigation of these aspects. The results will provide new information on this emerging pathogen and provide insights for the development of new therapeutic strategies. Also, the molecular basis of HMV needs to be further elucidated.

MATERIALS AND METHODS
Bacterial strains, plasmids, and growth conditions. The strains and plasmids used in this work are listed in Table 1. Recombinant plasmids were introduced into Escherichia coli and K. pneumoniae strains by Functional Study of rmpADC Homologues Microbiology Spectrum heat shock at 42°C. Virulence plasmids p15WZ-82_Vir and pVir-CR-HvKP4 were introduced into K. pneumoniae strains by conjugation. E. coli and K. pneumoniae strains were grown in Luria-Bertani (LB) broth at 37°C. Kanamycin (Kan) was added to the culture medium at a concentration of 50 mg/mL appropriately.
Cloning of plasmid-borne rmp homologues. The prmpA and prmpA2 loci were amplified using genomic DNA of strains 15WZ-82 and CR-HvKP4 as the templates, respectively. Briefly, fragments of 200-bp sequences upstream of the rmpA gene and downstream of the rmpC gene were amplified and ligated to vector pCE2 by TA cloning and then transformed into E. coli strain DH5a by heat shock at 42°C. Cloning of the single gene from the prmpA and prmpA2 loci followed the same procedure as that used in the cloning of the loci. The resulting plasmids recovered from the transformants were verified by sequencing and then transformed into K. pneumoniae strains WZ1-2, KP1088PC, and PM8PC, respectively. The primers used for cloning are listed in Table S1 in the supplemental material.
Virulence plasmid curing. The virulence plasmids of K. pneumoniae strains KP1088 and PM8 were cured using sodium dodecyl sulfate (SDS) (21). Briefly, an inoculum of 100 mL overnight culture was added into LB broth containing 5%, 4%, 3%, 2%, and 1% SDS, respectively. The cultures were incubated with shaking at room temperature for 48 h and then diluted and spread on LB agar plates for selection of virulence plasmid-cured colonies. Single colonies were streaked on LB agar plates with or without 10 mg/mL potassium tellurite (Te). Colonies that could grow on the LB agar plate instead of the Te plate were further tested for the presence of the rmpA2 gene by PCR, and those which were negative for rmpA2 were regarded as plasmid-cured strains.
Mucoviscosity assay. The mucoviscosity of the test strains was determined using the sedimentation assay with slight modifications (17). Briefly, an overnight bacterial culture was normalized to an optical density (OD) of 1.0 and centrifuged for 5 min at 1,000 Â g. The supernatant was gently removed without disturbing the pellet for measurement of OD at 600 nm (OD 600 ). Results were presented as mean and standard deviation from data of three independent experiments. An unpaired two-sided Student t test was performed to analyze the statistical difference between the level of mucoviscosity of wild-type strains and that of transformants carrying different rmp genes, as well as transconjugants carrying the virulence plasmids. Analysis of data was performed by using GraphPad Prism 8 (San Diego, CA). The string test was performed by stretching bacterial colonies on a sheep blood agar plate using an inoculation loop.
CPS production. Uronic acid was extracted and quantified as described previously (17). One milliliter of the overnight culture which had been normalized to an OD of 1.0 in the mucoviscosity assay was centrifuged at 14,000 rpm for 15 min to separate the supernatant and the cell pellet. The cell pellet was resuspended in 1 mL fresh LB for quantification. The normalized culture was represented as total CPS, and the supernatant and the cell pellet were represented as cell-free and cell-associated CPS, respectively. Five hundred microliters of normalized culture (OD = 1.0), supernatant, and cell pellet was mixed with 100 mL of 1% Zwittergent 3-14 in 100 mM citric acid, respectively, followed by incubation at 50°C for 20 min. The cells were then pelleted by This study centrifugation for 10 min at 14,000 rpm, and 300 mL of the supernatant was collected and added to 1.2 mL of absolute ethanol, followed by incubation at 4°C for 20 min and then centrifugation for 5 min at 14,000 rpm. The pellet was dried and resuspended in 200 mL of distilled water, to which 1.2 mL of 12.5 mM sodium tetraborate in sulfuric acid was added, followed by incubation at 100°C for 5 min and then on ice for 10 min. A 20-mL volume of 0.15% 3-phenylphenol in 0.5% NaOH was then added. After incubation for 5 min at room temperature, the absorbance at 520 nm was measured. The glucuronic acid content of the mixture was determined by extrapolating from a standard curve of glucuronic acid and expressed as micrograms per OD unit. Results were presented as mean and standard deviation from data of three independent experiments. An unpaired two-sided Student t test was performed to analyze the statistical difference as described above.
Capsule staining. The capsule of the test strains was visualized by negative staining with Indian ink, followed by counterstaining of the bacterial cells with crystal violet (22). Briefly, a loopful of overnight bacterial culture was mixed with a single drop of Indian ink placed on a clean microscope slide. The mixture was gently spread out on the slide using a coverslip and allowed to air dry. The film was then flooded with 0.1% crystal violet for 10 s and gently rinsed with distilled water. The slide was air dried and covered with a coverslip, observed, and photographed under microscope with Â1,000 magnification. Scale bars were added by ZEN.34 (ZEN Lite), and the image of the region of interest was captured by ImageJ.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.1 MB.