Modelling the Gastrointestinal Carriage of Klebsiella pneumoniae Infections

Abstract

Klebsiella pneumoniae is a leading cause of nosocomial and community acquired infections, making K. pneumoniae the pathogen that is associated with the second largest number of deaths attributed to any antibiotic resistant infection. K. pneumoniae colonizes the nasopharynx and the gastrointestinal tract in an asymptomatic manner without dissemination to other tissues. Importantly, gastrointestinal colonization is a requisite for infection. Our understanding of K. pneumoniae colonization is still based on interrogating mouse models in which animals are pretreated with antibiotics to disturb the colonization resistance imposed by the gut microbiome. In these models, infections disseminate to other tissues. Here, we report a murine model to allow for the study of the gastrointestinal colonization of K. pneumoniae without tissue dissemination. Hypervirulent and antibiotic resistant strains stably colonize the gastrointestinal tract of in an inbred mouse population without antibiotic treatment. The small intestine is the primary site of colonization and is followed by a transition to the colon over time, without dissemination to other tissues. Our model recapitulates the disease dynamics of the metastatic K. pneumoniae strains that are able to disseminate from the gastrointestinal tract to other sterile sites. Colonization is associated with mild to moderate histopathology, no significant inflammation, and no effect on the richness of the microbiome. Our model sums up the clinical scenario in which antibiotic treatment disturbs the colonization of K. pneumoniae and results in dissemination to other tissues. Finally, we establish that the capsule polysaccharide is necessary for the colonization of the large intestine, whereas the type VI secretion system contributes to colonization across the gastrointestinal tract. IMPORTANCE Klebsiella pneumoniae is one of the pathogens that is sweeping the world in the antibiotic resistance pandemic. Klebsiella colonizes the nasopharynx and the gut of healthy subjects in an asymptomatic manner, making gut colonization a requisite for infection. This makes it essential to understand the gastrointestinal carriage in preventing Klebsiella infections. Current research models rely on the perturbation of the gut microbiome by antibiotics, resulting in an invasive infection. Here, we report a new model of K. pneumoniae gut colonization that recapitulates key features of the asymptomatic human gastrointestinal tract colonization. In our model, there is no need to disturb the microbiota to achieve stable colonization, and there is no dissemination to other tissues. Our model sums up the clinical scenario in which antibiotic treatment triggers invasive infection. We envision that our model will be an excellent platform upon which to investigate factors enhancing colonization and invasive infections and to test therapeutics to eliminate Klebsiella asymptomatic colonization.

Keywords: Klebsiella pneumoniae; capsule polysaccharide; gut colonization; type VI secretion system.

FIG 1

K. pneumoniae colonizes the gut of immunocompetent mice without the need for antibiotic pretreatment. (A) Weight loss of mice infected with different doses of Kp52145 over 12 days. 4 to 6 mice were included in each group, except in the group infected with 3 × 10⁸ CFU, in which three mice died within 5 days postinfection. (B and C) CFU per gr of small (B) and large intestine (C) of the mice infected in panel A at 12 days postinfection. (D) The relative distribution of K. pneumoniae across the intestine sections was determined as 100% of the combined bacterial loads of the small and large intestine sections. (E and F) Bacterial loads in the small (E) and large (F) intestines of mice infected with Kp43816, SGH10, and NJST258-1. In each group, 9 to 10 mice were infected. (G) Relative distribution of K. pneumoniae strains across the intestine sections. (H) Bacterial loads in the spleen of infected mice with different K. pneumoniae strains at 12 days postinfection. Dashed lines indicate the plating detection limit. In all panels, each value is presented as the mean ± SD. **, P ≤ 0.01; ****, P ≤ 0.0001; ns, P > 0.05 for the indicated comparisons, which were determined using a one way-ANOVA with the Bonferroni correction for testing multiple comparisons.

FIG 2

Dynamics of K. pneumoniae gut colonization. (A) Bacterial loads in the different intestine sections of mice infected with Kp52145 on different days postinfection. Each value is presented as the mean ± SD. (B) Relative distribution of Kp52145 across the intestine sections. Dashed lines indicate the plating detection limit.

FIG 3

K. pneumoniae colonization does not produce severe tissue damage. (A) Hematoxylin-eosin staining of tissue at different days postinfection with Kp52145. (B) Quantification of the histopathology changes upon infection. Each dot represents a different mouse. Five slides per section of tissue, per time point and per mouse, were scored. (C and D) The cxcl1, reg3g, s100o8, reg4, and tnfa mRNA levels were assessed by qPCR in the small intestine (C) and in the colon (D) of noninfected mice (black bars) and infected mice (gray bars) at 6 (6 d) and 12 (12 d) days. 5 to 7 mice were analyzed in each group. In panel A, the images are representative of four infected mice. Each value is presented as the mean ± SD. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001 for the indicated comparisons, which were determined using a one way-ANOVA with the Bonferroni correction for testing multiple comparisons. No other comparison is significant (P > 0.05).

FIG 4

K. pneumoniae colonization has a minor effect on the bacterial gut microbiome. (A) The phyla present in the colon and small intestine of infected mice with Kp52145 were determined by 16S rRNA sequencing. (B and C) Log₂-fold change of the genera present in the colon of the infected mice at 6 (B) and 12 (C) days postinfection versus the noninfected (ni) mice. (D and E) Log₂-fold change of the genera present in the small intestine of the infected mice at 6 (D) and 12 (E) days postinfection versus the noninfected (ni) mice. In all panels, 5 to 7 mice were analyzed in each group.

FIG 5

Antibiotic treatment triggers the dissemination of K. pneumoniae from the gastrointestinal tract. (A) Six days after mice were colonized with strain Kp43816, mice were given two doses of ampicillin (i.p.) and 24 h after the last dose mice were euthanized and the bacterial loads were determined. (B) CFU per gr of small intestine, cecum, and colon of mice colonized with Kp43816 and treated with either vehicle solution (CT) or ampicillin (AMP). (C) Relative distribution of Kp43816 across the intestine sections of mice treated with either vehicle solution (CT) or ampicillin (AMP). (D) Bacterial loads in the spleen, liver, and lung of mice colonized with Kp43816 and treated with either vehicle solution (CT) or ampicillin (AMP). Dashed lines indicate the plating detection limit. In all panels, 6 mice were included in each group. Values are presented as the mean ± SD. **, P ≤ 0.01; ****, P ≤ 0.0001 for the indicated comparisons, which were determined using a Mann-Whitney U test. In panel C, **, P ≤ 0.01 for the comparisons, which were determined using a Mann-Whitney U test, of the percentage of bacteria in each of the tissue sections between mice treated with either vehicle solution (CT) or ampicillin (AMP).

FIG 6

K. pneumoniae CPS and the T6SS are required for gut colonization. (A) CFU per gr of small intestine, cecum, and colon of mice infected with Kp52145, the isogenic cps mutant (ΔmanC), and the T6SS mutant (ΔclpV).14 to 17 mice were included in each group in two independent experiments. (B) Relative distribution of Kp52145 and the isogenic cps (ΔmanC) and T6SS (ΔclpV) mutants across the intestine sections. (C) To determine the effect of the gut microbiome on the colonization by the T6SS mutant, mice were pretreated with an antibiotic cocktail that was stopped 3 days preinfection. Three days postinfection, the bacterial burdens in tissues were determined. (D) CFU per gr of small intestine, cecum, and colon of mice infected with Kp52145 and the T6SS mutant (ΔclpV) that were pretreated or not with the antibiotic cocktail (AA). 10 to 15 mice were included in each group in two independent experiments. (E) Relative distribution of Kp52145 across the intestine sections of mice pretreated or not with the antibiotic cocktail (AA). (F) CFU per gr of small intestine, cecum, and colon of E. cloacae and Klebsiella spp that were only found in mice infected with the T6SS mutant (ΔclpV) and pretreated with the antibiotic cocktail. In all panels, each value is presented as the mean ± SD. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001; ns, P > 0.05 for the indicated comparisons, which were determined using a one way-ANOVA with the Bonferroni correction for testing multiple comparisons. In panel B, *, P ≤ 0.05; **, P ≤ 0.01; ns, P > 0.05 for the comparisons, which were determined using a one way-ANOVA with the Bonferroni correction for testing multiple comparisons, of the percentage of bacteria in each of the tissue sections between the mice infected with Kp52145 and each of the mutants. In panel C, *, P ≤ 0.05; ****, P ≤ 0.0001 for the comparisons, which were determined using a Mann-Whitney U test, of the percentage of bacteria in each of the tissue sections between the mice infected with Kp52145 and those treated or not with the antibiotic cocktail (AA).

FIG 7

K. pneumoniae antagonizes other Enterobacteriaceae of the gut microbiome in a T6SS-dependent manner. (A) Bacterial killing mediated by Kp52145 and the T6SS mutant 52145-ΔclpV (ΔclpV) against E. cloacae, K. oxytoca, and K. variicola. Mock, PBS-treated prey. The number of recovered target cells following 6 h of incubation in LB is indicated. (B) Bacterial killing mediated by K. oxytoca and its isogeneic clpV mutant (K. oxytoca-ΔclpV), and by K. variicola and its clpV mutant (strain K. variicola-ΔclpV) against 52145-ΔclpV. Mock, PBS-treated prey. The number of recovered target cells following 6 h of incubation in LB is indicated. In all panels, each value is presented as the mean ± SD (n = 3). *, P ≤ 0.05; ***, P ≤ 0.001; ****, P ≤ 0.0001; ns, P > 0.05 for the indicated comparisons, which were determined using a one way-ANOVA with the Bonferroni correction for testing multiple comparisons.