Genome-Wide Screening for Enteric Colonization Factors in Carbapenem-Resistant ST258 Klebsiella pneumoniae

Abstract

A diverse, antibiotic-naive microbiota prevents highly antibiotic-resistant microbes, including carbapenem-resistant Klebsiella pneumoniae (CR-Kp), from achieving dense colonization of the intestinal lumen. Antibiotic-mediated destruction of the microbiota leads to expansion of CR-Kp in the gut, markedly increasing the risk of bacteremia in vulnerable patients. While preventing dense colonization represents a rational approach to reduce intra- and interpatient dissemination of CR-Kp, little is known about pathogen-associated factors that enable dense growth and persistence in the intestinal lumen. To identify genetic factors essential for dense colonization of the gut by CR-Kp, we constructed a highly saturated transposon mutant library with >150,000 unique mutations in an ST258 strain of CR-Kp and screened for in vitro growth and in vivo intestinal colonization in antibiotic-treated mice. Stochastic and partially reversible fluctuations in the representation of different mutations during dense colonization revealed the dynamic nature of intestinal microbial populations. We identified genes that are crucial for early and late stages of dense gut colonization and confirmed their role by testing isogenic mutants in in vivo competition assays with wild-type CR-Kp Screening of the transposon library also identified mutations that enhanced in vivo CR-Kp growth. These newly identified colonization factors may provide novel therapeutic opportunities to reduce intestinal colonization by CR-KpIMPORTANCEKlebsiella pneumoniae is a common cause of bloodstream infections in immunocompromised and hospitalized patients, and over the last 2 decades, some strains have acquired resistance to nearly all available antibiotics, including broad-spectrum carbapenems. The U.S. Centers for Disease Control and Prevention has listed carbapenem-resistant K. pneumoniae (CR-Kp) as an urgent public health threat. Dense colonization of the intestine by CR-Kp and other antibiotic-resistant bacteria is associated with an increased risk of bacteremia. Reducing the density of gut colonization by CR-Kp is likely to reduce their transmission from patient to patient in health care facilities as well as systemic infections. How CR-Kp expands and persists in the gut lumen, however, is poorly understood. Herein, we generated a highly saturated mutant library in a multidrug-resistant K. pneumoniae strain and identified genetic factors that are associated with dense gut colonization by K. pneumoniae This study sheds light on host colonization by K. pneumoniae and identifies potential colonization factors that contribute to high-density persistence of K. pneumoniae in the intestine.

Keywords: Klebsiella pneumoniae; genome-wide screening; intestinal colonization; multidrug resistance; opportunistic infections.

FIG 1

Dense colonization of the intestine by K. pneumoniae preceded K. pneumoniae bacteremia. Changes in the gut microbiota of an allo-HCT patient who developed K. pneumoniae bacteremia were characterized by 16S rRNA sequencing and qPCR of stool samples. Sample collection days are indicated relative to the day of transplant (day 0). Concurrent intravenous antibiotic treatment and detection of bacteremia are shown at the top.

FIG 2

A transposon mutant library was constructed in K. pneumoniae MH258. Distribution of transposon mutants in the MH258 genome. Track 1 (light gray, outermost) shows all coding sequences (CDSs) in the chromosomal DNA. Track 2 (dark gray) shows the CDSs in which at least one mutant was found in the library. Track 3 (black peaks, middle) shows read counts for each insertion site. Track 4 (yellow-green/purple, innermost) shows a GC skew plot using a sliding window size of 10 kb (yellow-green, above average; purple, below average). The plot was generated using DNAPlotter (49).

FIG 3

The mutant library was screened in vitro and in vivo to determine genetic factors that are specifically associated with gut colonization. (A) Experimental design of in vitro screening. To screen for defects in aerobic or anaerobic growth, the mutant libraries were sampled over 24 h as cultures reached early stationary phase and were passaged (arrowheads). (B) Experimental design of in vivo screening. Mice were treated with vancomycin (Van) and metronidazole (Met) for 3 days and inoculated with the mutant library (∼10⁸ CFU in 200 μl PBS) by oral gavage. Fecal samples were collected thereafter for 4 weeks. (C) Colonization levels of the mutant library in the inoculated mice (n = 5). Fecal samples were suspended in PBS, and serial dilutions were plated on mutant-selective plates to enumerate CFU. PBS-treated control mice (n = 3) were also evaluated to confirm the absence of cross-contamination as well as efficacy of the selective plates. Values are medians ± 95% confidence intervals. The dotted line marks the limit of detection.

FIG 4

The mutant population greatly fluctuated in the gut with temporal expansion of a subset of mutants with enhanced fitness, but not in vitro. (A to C) Graphs showing changes in the diversity of the mutant populations during aerobic growth (A), anaerobic growth (B), and gut colonization (C). Sequencing depths were normalized to the lowest value, and the percentage of total targeted TA sites that were detected in each sample was plotted. (D) t-Stochastic neighbor embedding (t-SNE) was used to reduce the dimensionality of the data and visualize the clustering of the total mutant composition during gut colonization and in vitro growth. Samples collected longitudinally from each in vitro culture or mouse are presented in color—the darker color denotes the later time point. The clusters of in vitro aerobic and anaerobic samples are marked with orange and green background circles, respectively. The cluster of day 1 in vivo samples is highlighted with a gray circle, and day 14 to 28 sample clusters for each mouse are indicated in magenta. (E) Stacked graphs of percentages of reads from the mutants that showed enhanced fitness in the gut. All the mutants representing >2% of the total reads in any samples were graphed; mutants with >10% of the total reads were colored. Insertion sites of the mutants are noted in the legend. The genes for which isogenic mutants were generated and tested for a competitive colonization are marked with asterisks.

FIG 5

Genetic factors that are linked to dense intestinal colonization and in vitro growth were revealed. Each sample (from each animal at a given time point) was analyzed using the ARTIST pipeline, and the genetic loci whose mutants persistently had reduced fitness in all the examined animals are listed here. Average log₂ fold changes in the normalized read counts for the top significant loci in intestinal colonization are shown as a heatmap on the left. The loci are color coded on the right depending on their in vivo and in vitro importance: gray, required both in vivo and in vitro; dark pink, required only in vivo from early colonization; light pink, required only in vivo at the later stages of colonization. The fold changes for the loci that were not significant in a given sample are not shown; they were treated as unchanged (white). The genes for which isogenic mutants were generated and tested for a competitive colonization are marked with asterisks.

FIG 6

Competitive colonization study with isogenic mutants confirmed significance of the genes identified by INSeq in gut colonization. Nine isogenic mutants for the selected genes were generated by replacing the entire open reading frame (ORF) with a rifampin resistance cassette and tested for colonization of antibiotic-treated mice in competition with a wild-type strain. (A) Mutants with dramatic defects in gut colonization. (B) Complementary strains of ΔtamA and ΔhemN mutants were similarly tested with a wild-type strain harboring an empty pACYC177_aadA plasmid. Closed circles (Empty), ΔtamA pACYC177_aadA (or ΔhemN pACYC177_aadA) versus WT pACYC177_aadA; open circles (complementary [Comp]), ΔtamA pTam (or ΔhemN pHemN) versus WT pACYC177_aadA. (C) Mutants with minor or no defects in gut colonization. ΔyqjA and ΔybgF mutants showed minor but significant defects, whereas the ΔcyoB mutant did not show fitness change. (D) Mutants with enhanced fitness. The ΔflhA and Δgene_2133 mutants showed enhanced fitness as predicted. The Δgene_2182 mutant showed a minor defect. Mean competitive index (CI) ± SEM for each mutant is shown on a log scale. Statistical significance by one-sample t test is shown by asterisks as follows: *, P < 0.05; **, P < 0.01; ***, P < 0.001.