Klebsiella pneumoniae evolution in the gut leads to spontaneous capsule loss and decreased virulence potential

Abstract

Klebsiella pneumoniae (Kp) is an opportunistic pathogen that poses a major threat in healthcare settings. The gut is a primary Kp reservoir in hospitalized patients, and colonization is a major risk factor for Kp infection. The stability of virulence determinants such as capsule and lipopolysaccharide during gut colonization is largely unexplored. In a murine gut colonization model, we demonstrated that spontaneous capsule loss occurs rapidly but varies by Kp pathotype. A classical Kp strain and a carbapenem-resistant strain of the epidemic sequence type 258 lineage had significant levels (median of 25% and 9.5%, respectively) of capsule loss. In contrast, a hypervirulent strain did not lose capsule to a significant degree (median 0.1%), despite readily losing capsule during in vitro passage. Insertion sequences (ISs) or mutations were found disrupting capsule operon genes of all isolates and in O-antigen encoding genes in a subset of isolates. Mouse-derived acapsular isolates from two pathotypes had significant fitness defects in a murine pneumonia model. Removal of the IS in the capsule operon in a mouse-derived acapsular classical isolate restored capsule production to wild-type levels. Genomic analysis of Klebsiella rectal isolates from hospitalized patients found that 18 of 245 strains (7%) had at least one IS disrupting the capsule operon. Combined, these data indicate that Kp capsule loss can occur during gut colonization in a strain-dependent manner, not only impacting strain virulence but also potentially altering patient infection risk.

Importance: In hospitalized patients, gut colonization by the bacterial pathogen Klebsiella pneumoniae (Kp) is a major risk factor for the development of infections. The genome of Kp varies across isolates, and the presence of certain virulence genes is associated with the ability to progress from colonization to infection. Here, we identified that virulence genes encoding capsule and lipopolysaccharide, which normally protect bacteria from the immune system, are disrupted by mutations during murine gut colonization. These mutations occurred frequently in some isolates of Kp but not others, and these virulence gene mutants from the gut were defective in causing infections. An analysis of 245 human gut isolates demonstrated that this capsule loss also occurred in patients. This work highlights that mutations that decrease virulence occur during gut colonization, the propensity for these mutations differs by isolate, and that stability of virulence genes is important to consider when assessing infection risk in patients.

Keywords: Klebsiella; capsule; evolution; intestinal colonization; lipopolysaccharide; mobile genetic elements.

Fig 1

K. pneumoniae Kp4819 loses capsule during gut colonization and LB passage. Kp4819 in vivo capsule loss was determined using a murine gut colonization model (A–C) and in vitro during LB passage (D–F). Mice received 5 × 10⁶ CFU of Kp4819 via oral gavage, and 5 µL of the inoculum was also added to three tubes containing 5 mL of LB and passaged concurrently. For LB passaging, LB cultures were diluted 1:1,000 into 5 mL of fresh LB broth every 24 hours and an aliquot was plated to quantify total pale colonies. Kp4819 was quantified by culture and displayed as log₁₀ CFU/g of feces (A) or as log₁₀ CFU/mL (D). The percentage of pale colonies is displayed (B, E) and summarized as area under curve (C, F) with mean and SEM shown. For panels A–C, n = 15 across three independent experiments. For panels D–F, n = 9 across three independent experiments. For panels C and F, ****, P < 0.0001 by one-sample t-test with a hypothetical mean = 0.

Fig 2

Spontaneous insertion sequences in the capsule operon lead to acapsular Kp4819 phenotypes. Mutations in the capsule operon of six acapsular Kp4819 isolates from mouse gut colonization (Kp4819_M46-M135) or LB passaging (Kp4819_LB1) were identified by whole genome and Sanger sequencing (A). Genes in gray represent the beginning and end of the capsule operon, genes with at least one disruption are in blue, while all other capsule genes are in yellow. Capsule production of Kp4819, Kp4819 ∆rfaH, and six Kp4819 pale isolates was determined via uronic acid quantification (B). Susceptibility to active human serum after 3 hours was compared for 1 × 10⁵ CFU of Kp4819, Kp4819 ∆rfaH, and six acapsular Kp4819 isolates with the acapsular Kp4819 ∆rfaH mutant serving as a control (C). Capsule production of Kp4819_M134 isolates with the IS in the capsule operon removed via allelic exchange (M134 ∆IS1 1–3) was determined via uronic acid quantification (D). Susceptibility to active human serum after 3 hours was compared for 1 × 10⁵ CFU of Kp4819, Kp4819 ∆rfaH, and three Kp4819_M134 clones with the IS in the capsule region removed (M134 ∆IS1 1–3) via allelic exchange (E). For panels B–E, statistical significance between the Kp4819 WT and each isolate was determined by one-way analysis of variance with n ≥ 3 with Dunnett’s multiple comparisons test with ***, P < 0.001.

Fig 3

Evolved K. pneumoniae Kp4819 acapsular isolates have differing LPS structures. Insertion sequences found in the O-antigen encoding rfb operon of acapsular Kp4819 isolates were identified by whole genome sequencing (A). LPS was isolated from 1 × 10⁹ CFU of Kp strains of interest, and 10 µL of each LPS product was visualized using SDS-PAGE (B). An Escherichia coli LPS control is the first lane, followed by the CandyCane molecular weight ladder, and the Kp isolates of interest. The LPS core region is labeled a while the O-antigen region is labeled b.

Fig 4

An evolved acapsular Kp4819 isolate has fitness defects during models of infection. Kp4819_Kan was competed against the acapsular Kp4819_M134 isolate in a murine pneumonia (A) or bacteremia (B) model. In a murine pneumonia model, mice were inoculated with 1 × 10⁸ CFU of a 1:1 mixture of Kp4819_Kan and Kp4819_M134 retropharyngeally (A). To model murine bacteremia, mice were administered 1 × 10⁷ CFU of a 1:1 mixture of Kp4819_Kan and Kp4819_M134 via tail vein injection (B). The mean log₁₀ competitive index at 24 hours post-infection is displayed for both panels A and B. For panels A and B, *, P < 0.05 and ****, P < 0.0001, by one sample t-test and n ≥ 8 performed across two independent experiments.

Fig 5

K. pneumoniae pathotypes vary in spontaneous capsule loss during gut colonization. In a murine gut colonization model, mice were inoculated with 5 × 10⁶ CFU of KPPR1 (A–C) or NJST258_2 (D–F) via oral gavage. The levels of Kp in the gut were measured by fecal collection and displayed as log₁₀ CFU/g feces (A, D). The development of pale colonies was monitored for each strain during gut colonization and displayed as percentages (B, E) and as area under the curve (C, F) with mean and SEM. For panels C and F, ****, P < 0.0001 by one sample t-test with a hypothetical mean = 0. For panels A–F, n ≥ 9 across two independent experiments.

Fig 6

Anaerobic conditions inhibit spontaneous capsule loss across K. pneumoniae pathotypes. Kp4819, KPPR1, and NJST258_2 were passaged in 5 mL of LB in aerobic (A) or anaerobic conditions (B) to monitor for pale colony formation. Every 24 hours, LB cultures were diluted 1:1,000 into 5 mL of fresh LB broth, and an aliquot of each sample was plated to quantify total pale colonies. The percentage of pale colonies and (A, B) and area under the curve (C) with mean with SEM is displayed. For panel C, **, P < 0.01 and ****, P < 0.0001, by one-way analysis of variance with an n ≥ 3 and Šídák’s multiple comparisons test.