Adaptive evolution of carbapenem-resistant hypervirulent Klebsiella pneumoniae in the urinary tract of a single patient

Abstract

The emergence of carbapenem-resistant hypervirulent Klebsiella pneumoniae (CR-hvKp) is a growing concern due to its high mortality and limited treatment options. Although hypermucoviscosity is crucial for CR-hvKp infection, the role of changes in bacterial mucoviscosity in the host colonization and persistence of CR-hvKp is not clearly defined. Herein, we observed a phenotypic switch of CR-hvKp from a hypermucoviscous to a hypomucoviscous state in a patient with scrotal abscess and urinary tract infection (UTI). This switch was attributed to decreased expression of rmpADC, the regulator of mucoid phenotype, caused by deletion of the upstream insertion sequence ISKpn26. Postswitching, the hypomucoid variant showed a 9.0-fold decrease in mice sepsis mortality, a >170.0-fold reduction in the ability to evade macrophage phagocytosis in vitro, and an 11.2- to 40.9-fold drop in growth rate in normal mouse serum. Conversely, it exhibited an increased residence time in the mouse urinary tract (21 vs. 6 d), as well as a 216.4-fold boost in adhesion to bladder epithelial cells and a 48.7% enhancement in biofilm production. Notably, the CR-hvKp mucoid switch was reproduced in an antibiotic-free mouse UTI model. The in vivo generation of hypomucoid variants was primarily associated with defective or low expression of rmpADC or capsule synthesis gene wcaJ, mediated by ISKpn26 insertion/deletion or base-pair insertion. The spontaneous hypomucoid variants also outcompeted hypermucoid bacteria in the mouse urinary tract. Collectively, the ISKpn26-associated mucoid switch in CR-hvKp signifies the antibiotic-independent host adaptive evolution, providing insights into the role of mucoid switch in the persistence of CR-hvKp.

Keywords: CR-hvKp; antibiotic-independent evolution; mucoid switch; persistence; urinary tract.

Fig. 1.

Characteristics of CR-hvKp strains Y8 and Y9 isolated from the urine samples of a scrotal abscess patient with UTI. (A) Schedule of antibiotic treatment and timing of Kp isolation. (B) Growth curves of Y8 and Y9. (C) MICs of 15 antibiotics against Y8 and Y9. Note: The antibiotics highlighted in the green box indicate the differences in MIC values between the two strains. (D and E) Quantitative analysis of the mucoviscosity (D) and capsular polysaccharides (E) of Y8 and Y9 over time. (F) Virulence potential of Y8 and Y9 via intraperitoneal infection in model mice (n = 10). Data in (B, D, and E) were presented as means with 95% CI of three independent experiments. P-values in (B, D, and E) were determined by unpaired t tests and those in (F) were determined by log-rank (Mantel-Cox) test, using PRISM 8.0.*P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 2.

Comparative genomics and transcriptomics analyses of CR-hvKp strains Y8 and Y9. (A) Neighbor-joining phylogenetic tree of Kp strains based on core-genome SNPs. (B) Schematic diagram of genome composition of the two CR-hvKp strains. (C) Antibiotic resistance genes and virulence genes profile of Kp strains. (D) The differences between Y8 and Y9 in four regions: three in the bacterial chromosome and one on a plasmid. (E) Two SNPs were detected between Y8 and Y9. The position and transcriptional direction of each gene were indicated by arrows. Dark gray shadowing indicates > 90% nucleotide sequence identity. (F) Volcano plot showing the gene expression of Y9 relative to Y8 based on RNA-Seq. (G) Expression levels of capsule synthesis-related genes and AcrAB efflux pump-encoding genes in Y8 and Y9 detected by 2^−ΔΔCT RT-qPCR. Data were presented as the means with 95% CI of three independent experiments. P-values determined by the unpaired t tests. *P < 0.05, **P < 0.01, ***P < 0.0001.

Fig. 3.

Analysis of the roles of six hypermucoviscosity candidate genes in CR-hvKp. (A) Schematic diagram of two methods used to verify the function of candidate genes: gain-of-function and promoter activity assays. (B and C) Effects of six candidate genes on mucoviscosity (B) and capsule production (C). Note: AT, acyltransferase-coding gene; ASAT, aspartate aminotransferase-coding gene. (D) β-galactosidase reporter assays of plasmid promoter activity. The rmpA promoter regions of Y8 and Y9 strains were fused to the promoter-less lacZ gene, yielding PrmpA_lacZ and ISKpn26+PrmpA_lacZ transcriptional fusions. The strong constitutive tet promoter served as a positive control. Data in (B–D) were presented as the mean with 95% CI of three independent experiments. P-values determined by the unpaired t tests between two groups or one-way ANOVA among multiple groups. ns, no significance; ***P < 0.001.

Fig. 4.

Relationship between hypermucoviscosity and virulence of the CR-hvKp strains. (A) Schematic diagram of the protocol for induction of peritonitis-sepsis model mice. (B) Survival rates of mice infected with hypomucoid or hypermucoid CR-hvKp. (C) Bacterial load in five main organs and blood. (D) Resistance of CR-hvKp strains to killing by NMS or heat-inactivated mouse serum (HIMS). (E) Representative histograms of anti-C3 fluorescent CR-hvKp cells. (F) Macrophage adhesion and internalization of the two strains. (G) Survival of the two strains in BMDM. Data in (D, F, and G) were presented as the mean with 95% CI of three independent experiments. P-values in (B) determined by log-rank (Mantel-Cox) test and those in (C and D) and (F and G) determined by the unpaired t tests. ns, no significance; *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 5.

Enhancement of urinary tract colonization via the loss of hypermucoviscosity of CR-hvKp. (A) Schematic diagram of the experimental protocol for mouse urinary tract colonization. (B and C) Colonization of CR-hvKp in the mouse urinary tract (n = 5). (D) Interaction between CR-hvKp and T24 human bladder cancer cells (MOI = 50). (E and F) In vitro biofilm formation (E) and autoaggregation (F) of CR-hvKp. Data in (D–F) were presented as the mean with 95% CI of three independent experiments. P-values in (B–F) determined by the unpaired t tests. ns, no significance; *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 6.

Natural mucoid phenotype switching of CR-hvKp within the mouse urinary tract. (A) The emergence of hypomucoid variants. (B) Schematic diagram of the experimental protocol for analyzing CR-hvKp evolution. (C) Comparison of numbers of SNPs in hypermucoid bacteria and spontaneous hypomucoid variants relative to the Y8 reference genome. (D) Phylogenetic network and neighbor-joining phylogenetic analysis based on the allelic profiles of core genes. (E) Type and frequency of mutations mediating mucoid phenotype switching of Y8 within the mouse urinary tract. (F) Enrichment of hypomucoid CR-hvKp in the mouse urethral environment (n = 5). Data in (A and F) were presented as the mean with 95% CI of five independent experiments. P-values in (A and C) determined by the unpaired t tests. ns, no significance; *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 7.

Eight mutation types mediating the hypermucoid to hypomucoid CR-hvKp switching within the mouse urinary tract. (A) Mutations associated with ISKpn26 deletion caused the loss or reduced expression of rmpADC, resulting in the generation of hypomucoid variants. (B and C) ISKpn26 insertion caused the inactivation of rmpA and wcaJ. (D) Inserting a base G at poly10G (276 to 285 bp) of rmpA causes a frameshift mutation, resulting in truncation of RmpA. Note: The alphanumeric combinations (A1L-E5L) in parentheses represent the strains collected from the mouse UTI model. In these labels, the initial letter indicates the mouse ID from which the strains were isolated, followed by a number representing the sequence of the strain isolated from the urine sample of that mouse, and the final letter “L” denotes a low-viscosity variant.