Genetic requirements for Klebsiella pneumoniae-induced liver abscess in an oral infection model

Abstract

Klebsiella pneumoniae is the predominant pathogen of primary liver abscess. However, our knowledge regarding the molecular basis of how K. pneumoniae causes primary infection in the liver is limited. We established an oral infection model that recapitulated the characteristics of liver abscess and conducted a genetic screen to identify the K. pneumoniae genes required for the development of liver abscess in mice. Twenty-eight mutants with attenuated growth in liver or spleen samples out of 2,880 signature-tagged mutants that produced the wild-type capsule were identified, and genetic loci which were disrupted in these mutants were identified to encode products with roles in cellular metabolism, adhesion, transportation, gene regulation, and unknown functions. We further evaluated the virulence attenuation of these mutants in independent infection experiments and categorized them accordingly into three classes. In particular, the class I and II mutant strains exhibited significantly reduced virulence in mice, and most of these strains were not detected in extraintestinal tissues at 48 h after oral inoculation. Interestingly, the mutated loci of about one-third of the class I and II mutant strains encode proteins with regulatory functions, and the transcript abundances of many other genes identified in the same screen were markedly changed in these regulatory mutant strains, suggesting a requirement for genetic regulatory networks for translocation of K. pneumoniae across the intestinal barrier. Furthermore, our finding that preimmunization with certain class I mutant strains protected mice against challenge with the wild-type strain implied a potential application for these strains in prophylaxis against K. pneumoniae infections.

FIG. 1.

Liver abscess developed in mice with K. pneumoniae infection. Livers were retrieved from BALB/c mice which were orally inoculated with PBS or 10⁷ CFU of wild-type K. pneumoniae. As indicated with arrowheads in panel A, multiple microabscess foci developed in the infected liver by 72 hpi. Compared to what was found for the control liver retrieved from PBS-inoculated mice (B [magnification, ×200] and F [magnification, ×400]), histological examination of the livers of K. pneumoniae-infected mice showed infiltrates of inflammatory cells at 48 hpi (C [magnification, ×200] and G [magnification, ×400]) and necrosis of the liver parenchyma at 72 hpi (D [magnification, ×100 and H [magnification, ×400]). A high concentration of K. pneumoniae was detected in the infected liver at 72 hpi with anti-OmpA rabbit polyclonal sera (I) but not in the control liver (E). GFP-expressing K. pneumoniae (J [magnification, ×1,000]) was also used in the oral infection model for tracing its distribution. Liver cryosections from mice which were infected with 10⁷ CFU of GFP-expressing K. pneumoniae were prepared at 24 (K), 48 (L), or 72 (M) hpi and were examined under a fluorescein isothiocyanate view. Scale bar, 50 μm.

FIG. 2.

Dynamics of wild-type K. pneumoniae growth in different mouse tissues after an oral inoculation. A single dose of 10⁷ CFU of wild-type K. pneumoniae was used to infect BALB/c mice via the oral route. At 12, 24, 36, 48, and 72 hpi, samples from various sites in the infected mice, including proximal small intestine (PSI), distal small intestine (DSI), cecum, colon, spleen, liver, blood, and kidney samples, were retrieved and homogenized for measurements of viable bacterial counts, which are expressed as numbers of CFU g⁻¹ tissue or CFU ml⁻¹ blood. Five mice were examined at each time point. The limit of detection was approximately 50 CFU. Samples which yielded no colonies were not plotted. The bars indicate geometric means.

FIG. 3.

STM screen for KLA-attenuated mutants. (A) Schematic diagram showing the STM screen conducted in the oral infection model. A total of 2,880 mutants that produced the wild-type capsule were selected and arrayed in 60 MPs. Three 8-week-old BALB/c mice were orally administered 10⁷ CFU of bacterial culture of a particular MP pool which contains 48 uniquely tagged mutants. Livers and spleens were harvested from these mice, adequate dilutions of tissue homogenates were plated onto kanamycin-LB agar, and at least 3,000 colonies for each output were pooled for the preparation of bacterial genomic DNA. The tag mixtures carried by the mutants in each of the inocula and the output pools were PCR amplified and spotted on 48 replica blots. Each unique fluorescein-labeled tag was subsequently used as a probe for hybridization to those blots, and the mutants whose associated tags were detected in the inocula but not in the pools recovered from either the liver or the spleen were identified as KLA attenuated, indicated by circles in panel B.

FIG. 4.

Virulence attenuation in independent infection experiments. (A) Five BALB/c mice were orally administered 10⁷ CFU of a single KLA-attenuated mutant strain. The survival rate of these mice was monitored daily for 2 weeks. MDD are shown in parentheses, and both mortality rate and MDD were determined by Kaplan-Meier analysis. Strains VA05, -07, -12, -14, -15, -18, -21, and -28, which were avirulent in mice during the observation period, were regarded as class I mutants. Strains VA01, -02, -04, -08, -10, -19, -20, -22, -24, -25, -32, and -33, which resulted in mortality rates of 20% to 40% in mice, had their virulence attenuated to a statistically significant level (P < 0.05) and were therefore categorized as class II mutants. The remaining VA strains, with no significantly attenuated virulence, were grouped as class III mutants. (B) Bacterial growth in different mouse tissues. Three BALB/c mice were orally administered 10⁷ CFU of a single KLA-attenuated mutant strain. Small intestine, colon, spleen, liver, and blood samples were harvested at 48 hpi to enumerate bacterial concentrations, which were expressed as numbers of CFU g⁻¹ tissue or CFU ml⁻¹ blood. The limit of detection was approximately 50 CFU. Samples which yielded no colonies were plotted as having values of 50 CFU g⁻¹ tissue or CFU ml⁻¹ blood.

FIG. 5.

Characterization of the avirulent yet immunogenic class I mutant strains. (A) Oral administration with the class I mutants allowed mice to survive challenge with wild-type K. pneumoniae infection. Five BALB/c mice were orally administered a single dose of 10⁷ CFU of a particular class I mutant strain. The preimmunized mice, together with age-matched naïve mice, were challenged with 10⁷ CFU of wild-type K. pneumonia 6 weeks later. The survival rate of the infected mice was monitored daily for 2 weeks. MDD are shown in parentheses, and both mortality rate and MDD were determined by Kaplan-Meier analysis. Sera from all mice were collected 1 day prior to challenge with the wild-type strain and were used to detect the production of K. pneumoniae-specific antibodies. *, P < 0.05; **, P < 0.01. Statistical significance was determined by comparison of survival curves by use of the log rank test (Kaplan-Meier analysis; Prism4). (B) Cytokine production upon oral administration with the class I mutant VA28. Sera and spleens were collected at 24 hpi and were analyzed for cytokine production by an enzyme-linked immunosorbent assay. The mean production levels of cytokines (gamma interferon, tumor necrosis factor alpha, IL-12, and IL-10) in five mice per group are shown in pg per 100 μg of total proteins. (C) Specific antibodies against the outer membrane constituents and capsular polysaccharides of K. pneumoniae were detected in pooled antisera from the five mice that were preimmunized with VA28.

FIG. 6.

Role for K. pneumoniae YmdF in resistance to oxidative stress. (A) Tolerance responses of class I mutants to various stresses. Bacterial cultures of the wild type and each of the class I mutants, which were grown in LB medium at 37°C, were adjusted to a turbidity value of 1 and were then subjected to treatment with 20 mM of H₂O₂ for 2 h at 37°C. The bacterial concentration upon oxidative stress was determined using a MicroScan turbidity meter (Dade Behring, CA). Results shown are means ± standard deviations for three independent experiments. * indicates statistical significance as determined by Student's t test (P < 0.05). (B) Northern blot analysis of ymdF transcripts. Total RNAs were isolated from wild-type K. pneumoniae grown in LB, in M9-glucose, and under various stress conditions. Twenty micrograms of total RNA from each sample was subjected to Northern blotting, and a biotin-labeled ymdF RNA probe was used for the hybridization. The expression levels of ymdF upon exposure to H₂O₂ for 1, 3, 6, 9 min; treatment with 15 mM paraquat (PQ); or treatment with different concentrations of sucrose for 15 min are shown in lanes 1 to 8, and the abundances of ymdF transcripts in K. pneumoniae grown in LB or M9-glucose medium is shown in lanes 9 to 16.