Extended-Spectrum β-Lactamase Production Is Associated with an Increase in Cell Invasion and Expression of Fimbrial Adhesins in Klebsiella pneumoniae

Bacterial cells and culture conditions.A total of 58 ESBL-producing and 152 non-ESBL-producing K. pneumoniae strains described elsewhere were employed in the study. The strains were isolated from clinical specimens from hospitalized patients in different European countries, tested for ESBL production, serotyped with respect to O and K antigens, and analyzed for clonality as described previously (42, 43). Production of ESBLs was detected according to the recommendations of the CLSI (formerly NCCLS) as described previously (43).

Hemagglutination assays.The expression of type 1 (MSHA) and type 3 (mannose-resistant Klebsiella-like hemagglutination [MR/K-HA]) pili was examined as described previously (40). Briefly, after four passages (either serial 48-h passages in static brain heart infusion broth for the detection of MSHA or passage in nutrient broth for detection of MR/K-HA), the bacteria were allowed to grow statically for 48 h, at which time they were harvested and washed three times with phosphate-buffered saline. Fifty microliters of bacterial suspension (3 × 10⁹ CFU/ml) was mixed on porcelain plates with 50 μl erythrocyte suspension (5 × 10⁸/ml). Agglutination was observed after 3 min of gentle shaking at room temperature and after another 10 min at 4°C, and the intensity of the reaction was graded from − to +++. Positive MSHA of guinea pig erythrocytes or hemagglutination of tanned ox erythrocytes indicated expression of type 1 or type 3 MR/K-HA fimbrial adhesins, respectively, by at least 10% of the bacterial-cell population.

Detection of fimbria-encoding genes.The genes coding for type 1 (MSHA) and type 3 (MR/K-HA) pili were detected by colony hybridization using as probes the insert of pGG101 (fimAE SphI; 1.6 kb) and the insert of pFK10 (mrkBC KpnI; 2.8 kb), respectively (17, 18). For each strain, 1 microliter of overnight culture in Luria-Bertani (LB) broth was spotted onto Hybond-N nylon membranes (Amersham France, Les Ulis, France), placed on solid LB medium, incubated for 6 h at 37°C, and lysed according to the manufacturer's instructions. DNA probes were labeled with [α-³²P]dATP by using a random-primed DNA-labeling kit (Boehringer Mannheim, Meylan, France). Hybridization was performed with Rapid-hyb buffer (Amersham) as recommended by the manufacturer under high-stringency conditions. In addition, detection of the type 3-encoding gene (mrkD) was performed by PCR as described previously (21).

Cell lines and culture conditions.The human bladder epithelial cell line T24 and the ileocecal cell line HCT8 were cultivated as described previously (45). The T24 cells were cultivated in McCoy's 5A medium with 10% fetal bovine serum and the HCT8 cells in RPMI 1640 medium with 1 mM pyruvate, 2 mM glutamine, and 10% fetal calf serum and subcultivated at a ratio of 1:8 twice a week. Human A549 adenocarcinoma cells were cultivated in F-12K medium with 10% (vol/vol) fetal calf serum and 2 mM glutamine as described elsewhere (44).

Invasion assay.The ability of the bacteria to invade epithelial cells was measured as described previously (45). The cells were cultured in 24-well cell culture clusters to monolayers of approximately 7 × 10⁴ cells/well. Two × 10⁶ mid-log-phase bacteria were then added to each well (approximately 30 bacteria/epithelial cell). After centrifugation at 200 × g for 5 min, invasion was allowed to occur for 2 h at 37°C in an atmosphere of 94% air-6% CO₂. Before a second 2-h incubation (kill) period under the same conditions but with fresh medium containing 100 μg of imipenem/ml, the monolayers were washed once with Earle's balanced salt solution (EBSS). All extracellular, but not internalized, bacteria were killed by the added imipenem during the kill period. The monolayers were then washed twice with EBSS and lysed with 0.1% Triton X-100 before determination of viable counts of the released intracellular bacteria. The invasion ability was expressed as the percentage of the inoculum surviving imipenem treatment. The invasion was classified into two grades: grade I, invasion of <5% of the inoculum, and grade II, invasion of >5% of the inoculum. The distribution of the strains’ invasion and the mean relative invasion were calculated.

Adhesion assay.The adhesion of six strains that harbored the ESBL-encoding plasmid and of their plasmid-cured derivatives (see below) to epithelial cells was tested as described previously (44). Briefly, 50 μl of the bacteria (ca. 10⁸ CFU/ml) was added to wells of flat-bottom microtiter plates containing confluent monolayers of T24, HCT8, or A549 cells. After 30 min at room temperature, the nonadherent bacteria were removed by repeated washing with EBSS, and the cell monolayers were fixed with 2.5% glutaraldehyde in Hanks balanced salt solution for 5 min. The plates were then washed and incubated in blocking buffer overnight at 4°C. The bound bacteria were quantified using anti-Klebsiella serum at 1:1,000 in enzyme-linked immunosorbent assays.

Preparation of plasmid DNA and plasmid curing.To investigate the contribution of the ESBL-encoding plasmid to the ability of the bacteria to produce type 1 and type 3 pili and to adhere to and invade the epithelial cells, five strains harboring the ESBL-encoding plasmid, expressing type 1 and type 3 pili, and positive in the adhesion and invasion assays were subjected to a plasmid-curing procedure as described elsewhere (43). The plasmid-cured derivatives obtained were tested for ESBL production and for the ability to express the pathogenicity factors as described above.

Transconjugation and its effect on pilus expression.Transconjugation experiments were performed by filter mating (45) using two ESBL-producing K. pneumoniae isolates (isolates 1054 and 1147) as donors and one non-ESBL-producing K. pneumoniae isolate as a recipient strain (KPTA 29).

Transconjugants were selected on LB agar plates containing nalidixic acid (128 μg/ml) and ceftriaxone (4 μg/ml) and subjected to Vitek-2 susceptibility testing. ESBL production was confirmed as described above. Pulsed-field gel electrophoresis was used to confirm the genetic identity of the transconjugants and the recipient parental strain. The acquisition of ESBL-encoding plasmids was verified by comparative plasmid analysis of the donor and recipient strains. PCR confirmed the presence of the expected ESBL genes. Two confirmed transconjugants, T1147 and T1054, were assayed for the ability to express type 1 and type 3 pili.

To further verify the influence of ESBL-producing plasmids on type 3 fimbrial expression, a real-time reverse transcription (RT)-PCR was conducted. For this purpose, RNA was extracted from 1 ml of a 5-h culture using the NucleoBond NucleoSpin NucleoTrap kit from Machery-Nagel (Düren, Germany) according to the manufacturer's instructions. The real-time RT-PCR step was carried out with the LightCycler RNA Master Sybr green I kit (Roche Diagnostics, Mannheim, Germany) on LightCycler systems (Roche Diagnostics) with primers MrkD 2 F (5′-CCACCAACTATTCCCTCGAA-3′) and MrkD 2 R (5′-ATGGAACCCACATCGACATT-3′) (GenBank accession number AY225462). As an internal control, the 16S rRNA gene was amplified with universal primers TM1 (5′-ATGACCAGCCACACTGGAAC-3′) and TM2 (5′-CTTCCTCCCCGCTGAAAGTA-3′). A 20-μl mixture containing 1 μl of RNA and 2 μl of primers was prepared and amplified according to the manufacturer's instructions (the RNA preparations were checked for absence of DNA contamination). To compare relative gene expression in the wild-type strain and its transconjugants, the value 2[(CP_{transconjugant} − CP_{wild type})_TM − (CP_{transconjugant} − CP_{wild type})_MrkD] was calculated, where CP is the crossing point. The experiment was performed twice.

Statistical analysis.The significance of differences between groups of isolates was evaluated by Yate's corrected chi-square test. For tables showing values of n of <5 within a cell, Fisher's exact test was used. The odds ratio (OR) and 95% confidence interval (CI) were determined, and the values were taken as a gauge of association. The significance of differences between the relative invasion levels of the tested bacteria was evaluated by the unpaired t test with Welch correction.

Clonality of strains and ESBL types.After the exclusion of clonal strains, a total of 152 non-ESBL-producing and 58 ESBL-producing strains were defined. Sixteen ESBL types were identified, among which SHV-5 was the most common (23%; 12/58 isolates), followed by TEM-3 (15.6%; 9/58 isolates), SHV-4 and SHV-12 (each 10%; 6/58 isolates), and TEM-24 (8.6%; 5/58 isolates). SHV-2, -2A, -36, and -40 and TEM-5, -8, -12, -15, and -16 were detected in one strain each.

Expression of type 1 and type 3 fimbriae.The percentages of strains that did not express either type 1 or type 3 fimbrial adhesin were very low (i.e., 3.5% and 5.2% of ESBL-producing and non-ESBL-producing strains, respectively) (Table 1), suggesting that at least one of the adhesins is required for colonization irrespective of the presence of the ESBL plasmid. Non-ESBL-producing strains produced one of the fimbrial adhesins more often than ESBL producers: 50 out of the 152 non-ESBL-producing strains (32.8%) expressed either type 1 or type 3 fimbrial adhesins, and only 10 out of the 58 (17.2%) ESBL producers did so (P = 0.0268; OR = 2.353; 95% CI = 1.099 to 5.036). In contrast, the frequency of ESBL-producing strains coexpressing both type 1 and type 3 fimbrial adhesins was significantly higher than that of the non-ESBL producers (79.3% and 61.8%, respectively) (P = 0.0214; OR = 2,365; 95% CI = 1.157 to 4.834).

The magnitudes of the hemagglutination of guinea pig erythrocytes and of tanned ox erythrocytes did not differ significantly between the ESBL-producing and non-ESBL-producing strains (P > 0.05) (Fig. 1 A and B), suggesting that once the adhesin genes are turned on, the phenotypic expression of the adhesins is not affected by the presence of ESBL plasmids. Colony hybridization revealed the presence of the type 1-encoding genes in all but one strain, and only two did not harbor the type 3-encoding genes.

• Open in new tab
• Download powerpoint

FIG. 1.

MS hemagglutination of guinea pig erythrocytes (A) and MR (Klebsiella-like hemagglutination) of tanned ox erythrocytes (B) by ESBL- and non-ESBL-producing K. pneumoniae isolates. −, no hemagglutination; +, weak hemagglutination; ++, intermediate hemagglutination; +++, strong hemagglutination.

Adhesion and invasion ability.The adherence rates of ESBL- and non-ESBL-producing strains to ileocecal, bladder, and alveolar epithelial cells did not differ significantly (P > 0.05) (Fig. 2). In contrast, the frequency of isolates that efficiently invaded ileocecal epithelial cells (grade II invasion, >5% of the inoculum) was higher among ESBL producers than among non-ESBL producers (Table 2) (81%, 47/58, and 61%, 93/152, respectively; P = 0.0084; OR = 2.711; 95% CI = 1.302 to 5.643). Likewise, 27.7% (n = 16/58) of the ESBL producers invaded the bladder T24 cells efficiently, whereas only 10% (n = 15/152) of the non-ESBL producers did so (P = 0.0021; OR = 3.479; 95% CI = 1.587 to 7.627). The mean percent relative invasion of HCT8 and T24 cells by ESBL-producing strains (5.5% ± 2.8% and 3.3% ± 2.7%) was significantly higher than that of non-ESBL producers (2.9% ± 2.6% and 1.8% ± 2%) (P < 0.0001) (Fig. 3). The invasion of the A549 lung epithelial cells by both ESBL and non-ESBL producers did not exceed 1%.

• Open in new tab
• Download powerpoint

FIG. 2.

Adhesion of ESBL- and non-ESBL-producing strains to ileocecal, bladder, and alveolar epithelial cells. The error bars indicate standard deviations.

• Open in new tab
• Download powerpoint

FIG. 3.

Mean relative invasion of T24 bladder and HCT8 ileocecal epithelial cells and distribution of invasive strains among ESBL and non-ESBL producers. *, significantly higher abilities of ESBL-producing strains to invade the cells (P < 0.0001). Each open and closed box of the box-and-whisker plot indicates the mean relative invasion of a single strain.

Hemagglutination, adhesion, and invasion abilities of plasmid-cured strains.Five strains were cured of their ESBL plasmids. The lack of an ESBL-producing phenotype in all plasmid-cured derivatives was confirmed using the CLSI guidelines. The loss of the ESBL-encoding plasmids in the strains was demonstrated by agarose gel electrophoresis (data not shown). In all cured strains, the MICs to third-generation cephalosporins, e.g., cefotaxime, ceftriaxone, and ceftazidime, were <2 μg/ml. There was no difference between the five plasmid-cured derivatives and their parental strains in the abilities of the strains to express type 1 and type 3 fimbriae and to adhere to and invade tissue culture cell lines (Fig. 4).

• Open in new tab
• Download powerpoint

FIG. 4.

Invasion and adhesion abilities of five ESBL-producing strains and their corresponding plasmid-cured derivatives. The error bars indicate standard deviations.

Hemagglutination of transconjugants.The parental recipient strain KPTA 29 expressed type 1 pili (e.g., it induced MSHA of guinea pig erythrocytes), but not type 3 fimbriae (e.g., it did not induce MR hemagglutination of ox erythrocytes). Two transconjugant, T1147 and T1054, of this strain that acquired the ESBL plasmid were obtained, and both switched on to express type 3 fimbriae, as shown in a hemagglutination assay of tanned ox erythrocytes. Specific PCR indicated that both the wild-type strain KPTA 29 and its two transconjugants, T1147 and T1054, harbored the type 3-encoding genes (data not shown). Differences in type 3 fimbrial expression were further studied in the parental strain and both transconjugants by real-time RT-PCR. A 226-bp fragment of the mrkD gene was targeted. The comparative CP values between the parental strain and the transconjugants T1054 and T1147 were, respectively, 3,702 and 5,595 (average CP values), indicating that type 3 fimbrial expression was upregulated in both transconjugants, consistent with the hemagglutination results.