Utilization of an intracellular bacterial community pathway in Klebsiella pneumoniae urinary tract infection and the effects of FimK on type 1 pilus expression

Abstract

Klebsiella pneumoniae is an important cause of urinary tract infection (UTI), but little is known about its pathogenesis in vivo. The pathogenesis of the K. pneumoniae cystitis isolate TOP52 was compared to that of the uropathogenic Escherichia coli (UPEC) isolate UTI89 in a murine cystitis model. Bladder and kidney titers of TOP52 were lower than those of UTI89 at early time points but similar at later time points. TOP52, like UTI89, formed biofilm-like intracellular bacterial communities (IBCs) within the murine bladder, albeit at significantly lower levels than UTI89. Additionally, filamentation of TOP52 was observed, a process critical for UTI89 evasion of neutrophil phagocytosis and persistence in the bladder. Thus, the IBC pathway is not specific to UPEC alone. We investigated if differences in type 1 pilus expression may explain TOP52's early defect in vivo. The type 1 pilus operon is controlled by recombinase-mediated (fimE, fimB, and fimX) phase variation of an invertible promoter element. We found that K. pneumoniae carries an extra gene of unknown function at the 3' end of its type 1 operon, fimK, and the genome lacks the recombinase fimX. A deletion mutant of fimK was constructed, and TOP52 Delta fimK had higher titers and formed more IBCs in the murine cystitis model than wild type. The loss of fimK or expression of E. coli fimX from a plasmid in TOP52 resulted in a larger phase-ON population and higher expression levels of type 1 pili and gave TOP52 the ability to form type 1-dependent biofilms. Complementation with pfimK decreased type 1 pilus expression and biofilm formation of TOP52 Delta fimK and decreased UTI89 biofilm formation. Thus, K. pneumoniae appears programmed for minimal expression of type 1 pili, which may explain, in part, why K. pneumoniae is a less prevalent etiologic agent of UTI than UPEC.

FIG. 1.

Time course of TOP52 and UTI89 bladder and kidney infections and gentamicin protection assays. Female C3H/HeN mice were inoculated with 10⁷ CFU of TOP52 or UTI89 by transurethral inoculation. For organ titers, bladders (A) and kidneys (B) were harvested at various time points postinfection, and the numbers of CFU were calculated. Titer data are combined from three independent experiments. For ex vivo gentamicin protection assays, bladders were harvested at 6 h (C) and 24 h (D) postinfection, and luminal and intracellular populations of bacteria were quantified. Short bars represent geometric means of each group, and dotted lines represent limits of detection. Significant P values, as calculated using the Mann-Whitney U test, are shown on the figure.

FIG. 2.

TOP52 progresses through an IBC pathogenic pathway. Histologic analysis of murine bladders after a 6-h infection with TOP52 (A) or UTI89 (B) revealed morphologically identical IBCs within superficial urothelial cells. Immunohistochemistry demonstrated that bacteria express type 1 pili within IBCs of both TOP52 (C) and UTI89 (D) infections. Staining was done with antibodies against type 1 pili (green) and the facet cell marker uroplakin III (red), and nuclei were visualized with Hoescht staining (blue). Urine samples from mice at 24 h postinfection with TOP52 (E) or UTI89 (F) contained long bacterial filaments. IBCs were quantified after visualization by LacZ staining and revealed that TOP52 forms significantly fewer IBCs than UTI89 at both 6 and 24 h postinoculation (G). Short bars represent geometric means, and significant P values, as calculated using the Mann-Whitney U test, are shown on the figure. Scale bar, 10 μm (A to F).

FIG. 3.

TOP52 ΔfimK has higher titers in organs and forms more IBCs in the murine cystitis model than the wild type. (A) The fim operons of E. coli and K. pneumoniae, which encode type 1 pili, have similar compositions and organizations, but the K. pneumoniae fim operon contains an extra gene of unknown function, fimK. (B) TOP52 ΔfimK had higher 6-h titers in the bladder than the wild type, and this phenotype was complemented with pfimK. (C) TOP52 ΔfimK also had consistently higher titers in the kidney at 6 h than the wild type, and this difference could be complemented back to wild-type levels. (D) Enumeration of IBCs at 6 h postinfection revealed that TOP52 ΔfimK forms more IBCs than TOP52. Titer and IBC data are combined from three independent experiments. Short bars represent geometric means of each group, and dotted lines represent limits of detection. Significant P values, as calculated using the Mann-Whitney U test, are shown in the figure.

FIG. 4.

TOP52 ΔfimK and TOP52/pfimX have higher expression levels of type 1 pili than wild type and vector controls. FimA immunoblot analysis showed that TOP52 ΔfimK makes more type 1 pili than wild-type TOP52. Expression of FimA was complemented down to wild-type levels with pfimK. TOP52/pfimX also produced a much larger FimA band than the vector control. Bacteria were normalized by OD₆₀₀, and Coomassie blue staining was used to verify similar levels of protein in each lane. Relative densitometry was calculated using ImageJ software. Phase assays, showing the orientation of the fimS promoter region of the fim operon, revealed that the wild-type TOP52 and TOP52/pBAD vector control were largely phase OFF while TOP52 ΔfimK and TOP52/pfimX had significant phase-ON populations.

FIG. 5.

TOP52 ΔfimK and TOP52/pfimX appear hyperpiliated by negative-stain EM. Negative-stain EM was performed on TOP52 cultures, and representative images are displayed. TOP52 had low to moderate piliation, and the negative control TOP52 ΔfimA-fimH appeared bald. TOP52 ΔfimK was hyperpiliated compared to the wild type, and this phenotype could be partially complemented with pfimK. TOP52/pfimX also displayed higher levels of piliation than the vector control.

FIG. 6.

The presence of fimK and fimX affects biofilm formation of TOP52 and UTI89. A 48-h biofilm assay was used to quantify biofilms produced by TOP52 strains (A) and UTI89 strains (B). TOP52 and the negative control TOP52 ΔfimA-fimH failed to form biofilms. However, TOP52 ΔfimK and TOP52/pfimX were able to form biofilms. TOP52 ΔfimK/pfimK formed significantly less biofilm than the vector control. Wild-type UTI89 formed a robust biofilm, and the negative control UTI89 ΔfimA-fimH was unable to form a biofilm. Complementation of UTI89 with TOP52 fimK resulted in significantly less biofilm formation than that produced by the UTI89/pBAD vector control. Data combine means of two independent experiments, each with duplicate plates. Error bars represent standard errors of the means.