Uropathogenic Escherichia coli Potentiates Type 1 Pilus-Induced Apoptosis by Suppressing NF-κB

Abstract

Urinary tract infections (UTIs) are among the most common inflammatory diseases. Acute UTIs are typically caused by type 1-piliated Escherichia coli and result in urothelial apoptosis, local cytokine release, and neutrophil infiltration. To examine the urothelial apoptotic response, a human urothelial cell line was incubated with various E. coli isolates and was then characterized by flow cytometry. Uropathogenic E. coli (UPEC) induced rapid urothelial apoptosis that was strictly dependent upon interactions mediated by type 1 pili. Interestingly, nonpathogenic HB101 E. coli expressing type 1 pili induced apoptosis at approximately 50% of the level induced by UPEC, suggesting that pathogenic strains contribute to apoptosis by pilus-independent mechanisms. Consistent with this possibility, UPEC blocked activity of an NF-κB-dependent reporter in response to inflammatory stimuli, yet this effect was independent of functional type 1 pili and was not mediated by laboratory strains of E. coli. UPEC suppressed NF-κB by stabilizing IκBα, and UPEC rapidly altered cellular signaling pathways. Finally, blocking NF-κB activity increased the level of piliated HB101-induced apoptosis to the level of apoptosis induced by UPEC. These results suggest that UPEC blocks NF-κB and thereby enhances type 1 pili-induced apoptosis as a component of the uropathogenic program.

Urinary tract infections (UTIs) are among the most common infectious diseases, resulting in over 7 million clinic visits annually in the United States alone (7) and causing significant morbidity and mortality. The majority of UTIs are due to ascending infections of the urinary bladder (cystitis) by Escherichia coli that expresses type 1 pili, fibrous organelles that mediate attachment to mannosylated host cell proteins (reviewed in reference 19). The infection process results in bladder inflammation that causes symptoms such as pain and frequent or urgent voiding. Given the emergence of antibiotic-resistant strains of uropathogenic E. coli (UPEC) (A. J. Schaeffer, Editorial, Curr. Opin. Urol. 10:23–24, 2000), a more thorough understanding of pathogenic mechanisms is necessary to identify novel therapeutic targets for the treatment and prevention of UTIs.

A key feature of inflammation during UTIs is a disruption of the urothelial integrity due to the exfoliation and subsequent excretion of superficial urothelial cells (4). Studies with mice have shown that UPEC induces a rapid loss of superficial urothelial cells, and the sloughing of superficial cells is due to the induction of apoptosis (12). This apoptotic process is dependent upon the expression of intact type 1 pili, since fimH mutant UPEC strains lacking the adhesin subunit FimH that forms the pilus tip fail to induce apoptosis, yet laboratory strains of E. coli induce apoptosis if they express intact type 1 pili. As a result of the urothelial apoptotic response, clearance of a majority of UPEC from the bladder occurs during voiding of urine. Thus, urothelial apoptosis is considered to be an important host defense in response to interactions with uropathogens (12), although the relationship between pathogenicity and the level of apoptosis induced by various UPEC isolates has not yet been established.

Several signaling pathways are known to regulate apoptotic processes, but the transcription factor NF-κB lies at the nexus of both antiapoptotic and proinflammatory cascades (reviewed in references 2 and 18). In unstimulated cells, NF-κB is sequestered in the cytosol through interactions with an inhibitory subunit, IκB. Proinflammatory stimuli, such as lipopolysaccharide (LPS) and tumor necrosis factor alpha (TNF-α), activate a signaling pathway that results in phosphorylation of IκB, which in turn leads to the ubiquitination and subsequent degradation of IκB by the 26S proteasome. The liberated NF-κB then translocates to the nucleus, where it activates the transcription of both proinflammatory and antiapoptotic genes (26). Given the role of NF-κB in both inflammation and apoptosis, it is not surprising that certain gram-negative bacteria have also evolved mechanisms to modulate NF-κB activity during infection (see reference 17 for a review).

To characterize the apoptotic response in UTI pathogenesis, we incubated a human urothelial cell line with various E. coli isolates and found that UPEC bacteria induced higher levels of apoptosis than piliated nonpathogenic strains. UPEC were found to quantitatively suppress NF-κB activation and prevent IκB degradation, whereas nonpathogenic E. coli lacked these activities. These findings suggest a more complicated sequence of early events in the pathogenesis of UTIs in which UPEC targets NF-κB signaling.

MATERIALS AND METHODS

Antibodies and reagents.

Antisera against NF-κB subunit p65 (sc-109) and IκBα (sc-371) were purchased from Santa Cruz Biotech. Antiserum against phospho-Erk1 was purchased from New England Biolabs. Methyl α-d-mannopyranoside and E. coli LPS were purchased from Sigma, and recombinant human TNF-α and the proteasome inhibitor N-acetyl-Leu-Leu-Nle-CHO (ALLN) were purchased from Calbiochem.

Apoptosis assays.

TEU-2 cells were grown to approximately 70% confluence, switched to antibiotic-free medium for at least 2 h, and then incubated with various E. coli isolates. In some samples, methyl α-d-mannopyranoside was added simultaneously with the bacteria to a concentration of 25 mM. Following incubation with bacteria, the cells were washed three times with phosphate-buffered saline (PBS). The cells were then harvested by removal with trypsin, and trypsin was inactivated by the addition of medium containing 10% fetal bovine serum. The cells were recovered by and washed twice with PBS. The cells were then immediately prepared for flow cytometry by incubation in Annexin V · fluorescein isothiocyanate/propidium iodide (FITC/PI) buffer (Immunotech) on ice for 10 min according to the manufacturer's instructions. Flow cytometry was performed on a Coulter EPICS XL flow cytometer.

Flow cytometry was also used to quantify apoptosis by assessing the level of degraded chromosomal DNA. After bacterial treatment, the cells were washed three times with PBS, trypsinized, pelleted, and washed again with PBS. The cells were then resuspended in ice-cold 70% ethanol to a final cell density of 10⁶ cells/ml and stored overnight at −20°C. The following day, the cells were washed twice with PBS, and the final cell pellet was resuspended in a solution containing 1 ml of PBS, 0.1% Triton X-100, 0.1 mM EDTA (pH 7.4), RNase A (50 μg/ml), and propidium iodide (50 μg/ml; Molecular Probes). Flow cytometry was performed on a Coulter EPICS XL flow cytometer. Cell populations were gated to exclude events with minimal light scattering. In some experiments, TEU-2 cells were infected with recombinant adenoviruses at a multiplicity of infection (MOI) of 100:1 for 12 h prior to bacterial infection or TEU-2 cells were preincubated with 50 μM ALLN in dimethyl sulfoxide for 60 min and supplemented with an additional 50 μM ALLN 4 h after initial infection.

Bacterial strains.

NU14 is a clinical isolate of E. coli originally obtained from the urine of a patient with cystitis, and NU14-1 is the corresponding fimH mutant (10). HB101/pWRS1-17 is a laboratory strain of E. coli carrying a plasmid that encodes type I pili (22). 8NU is an E. coli strain isolated for this study from the urine of a patient with cystitis. All bacterial strains were propagated in Luria broth at 37°C under static conditions to promote expression of type 1 pili (3). The extent of type 1 pilus expression was determined by mannose-sensitive hemagglutination (6, 8) of guinea pig erythrocytes (Cleveland Scientific).

Cell culture.

TEU-2 urothelial cells were generated from normal human ureter. Briefly, a primary urothelial cell culture was established by enzymatic dissociation of urothelium (16), and the primary culture was infected with LSNX-16E6E7, an amphotrophic retrovirus encoding the oncoproteins E6 and E7 of human papillomavirus type 16 (5). Stable integration of the retroviral provirus was selected with G418 (200 μg/ml), and the resulting immortalized human urothelial cells were expanded. TEU-2 cells were maintained in serum-free keratinocyte medium (KSFM; GIBCO). Cells were maintained at 37°C in 5% CO₂. Ureter tissue was originally obtained from a donor in accordance with the guidelines of Northwestern University's Internal Review Board of the Office for the Protection of Research Subjects.

Luciferase assays.

TEU-2 cells grown in 24-well dishes were transiently transfected with 0.19 μg of pNFκB-luc (Stratagene) and 0.01 μg of pTK-Renilla per well using FuGENE-6 (Roche) according to the manufacturer's instructions. After 24 h, cells were incubated in antibiotic-free medium for at least 2 h and then stimulated by adding E. coli LPS, human TNF-α, and/or E. coli isolates. Cell lysates were prepared and assayed for luciferase activity using the Dual-Luciferase Reporter Assay (Promega). To normalize for transfection efficiency and extract concentration, NF-κB activity was reported as relative light units for Photinus pyralis (pNFκB-luc) divided by the relative light units for Renilla reniformus. Data points for all experimental conditions were determined in triplicate.

Immunoblotting.

For immunoblots, cell extracts were prepared in radioimmunoprecipitation assay buffer containing protease inhibitor cocktail (Sigma) and 100 μM sodium orthovanadate, fractionated on sodium dodecyl sulfate–10% polyacrylamide gels, and transferred to an Immobilon-P membrane (Millipore). IκB protein was detected by blocking the membrane with 5% milk–0.1% Tween 20 in PBS, and binding anti-IκBα serum was diluted 1:200 in PBS–1% milk–0.1% Tween 20. Phospho-Erk1/2 was detected by blocking in PBS–5% milk–0.1% Tween 20 and binding with anti-phospho-Erk in PBS–5% bovine serum albumin–0.1% Tween 20 according to the manufacturer's instructions (New England Biolabs). Bound primary antibodies were detected using secondary antibodies conjugated to horseradish peroxidase in conjunction with chemiluminescence (Amersham).

RESULTS

Urothelial apoptosis in vitro is dependent upon type 1 pili.

In a mouse model of UTI, the UPEC strain NU14 induces rapid apoptosis of urothelial cells when instilled into the bladder, and this activity is strictly dependent upon the presence of intact type 1 pili on the bacterial surface (12). To develop an in vitro model for the study of human urothelial responses to uropathogens, a urothelial cell line was established by immortalization of normal human urothelial cells using a retrovirus expressing the E6 and E7 oncoproteins of human papillomavirus type 16 (see Materials and Methods). The resultant cell line, TEU-2, retained the appearance of normal primary urothelial cells in culture, expressed the urothelial cell marker cytokeratin 13, and differentiated into transitional epithelium when placed into organotypic culture (data not shown).

To quantify the induction of urothelial apoptosis by UPEC, the E. coli isolate NU14 was incubated with TEU-2 human urothelial cells. The number of cells undergoing apoptosis as a result of exposure to E. coli was then determined using fluorescently labeled annexin V to detect the translocation of phosphatidylserine from the inner leaflet of the plasma membrane to the outer leaflet (9), an early event in the apoptosis process (Fig. 1). Whereas annexin labeled few untreated TEU-2 cells, NU14 induced an accumulation of annexin-positive cells that increased over time and was not attributable to the membrane permeability that is typically associated with necrotic cells (Fig. 1A), since these same cells did not stain positively for the membrane-impermeable dye propidium iodide. The observation that incubation of TEU-2 cells with NU14 induced annexin staining was also consistent with similar findings of NU14-induced changes in TEU-2 nuclear morphology and terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling, two alternative markers for apoptosis (data not shown).

FIG. 1.

UPEC induces apoptosis in urothelial cells in vitro. (A) TEU-2 cells were incubated with NU14 at an initial MOI of 500. The cells were then incubated in the presence or absence of 25 mM methyl α-d-mannopyranoside (Mann) for 5 h before harvest. TEU-2 cells were then harvested and processed for flow cytometry by staining with Annexin V · FITC/PI. (B) Determination of annexin-positive cell populations at various times and MOIs reveals a time-dependent increase in annexin-positive cells induced by NU14. Methyl α-d-mannopyranoside was able to block the accumulation of annexin-positive cells.

The increase in the annexin-positive cell population was not strictly dependent upon the initial MOI (Fig. 1B), since little difference was found in the number of apoptotic cells induced using initial MOIs ranging from 50 to 500 (data not shown). Most importantly, the induction of the annexin-positive population was effectively blocked by the addition of 25 mM methyl α-d-mannopyranoside, a known competitive inhibitor of interactions between type 1-piliated E. coli and eukaryotic cells mediated by the pilus FimH subunit (1, 24). These findings indicate that the UPEC strain NU14 can induce rapid apoptosis in the human urothelial cell line TEU-2 and that the induction of apoptosis is dependent upon interactions mediated by type 1 pili.

UPEC induces elevated levels of apoptosis.

The observation that NU14 can induce urothelial apoptosis in vitro that was detectable by flow cytometry permitted quantitative comparisons of the apoptotic potential of different E. coli isolates (Fig. 2A). The uropathogenic isolate NU14 induced apoptosis in approximately 79% of TEU-2 cells, but the fimH mutant strain NU14-1 did not induce apoptosis, consistent with the observations in mouse (12) as well as those results shown in Fig. 1. Likewise, a nonpathogenic isolate expressing plasmid-encoded type 1 pili, HB101/pWRS1-17, also induced apoptosis in 48% of cells, but the parental HB101 strain lacking pili failed to induce accumulation of annexin-positive cells. These findings confirm those in the mouse model, which indicated that pilus expression was sufficient to induce urothelial apoptosis in the absence of other potential E. coli virulence factors which may be expressed by UPEC strains. In our system, however, HB101/pWRS1-17 did not induce apoptosis to the same extent as the UPEC strain NU14.

FIG. 2.

UPEC induces higher levels of apoptosis than nonpathogenic E. coli. Quantification of apoptotic TEU-2 cell populations by annexin staining (A) or DNA content (B and C) indicates that various strains induced differential levels of apoptosis. (A) TEU-2 cells were incubated for 5 h with NU14 (fimH⁺), NU14-1 (fimH), HB101/pWRS1-17 (fimH⁺), or HB101 (fimH) at initial MOIs of 250. Cells were then harvested and stained with the annexin V and PI, and annexin-positive PI-negative cell populations were quantified by flow cytometry. (B) A representative histogram showing the appearance of a sub-G₁ peak following exposure of TEU-2 cells to NU14 for 7 h (inset shows forward versus side scatter and gated population). (C) Sub-G₁ populations induced by incubating TEU-2 cells with various E. coli isolates. Error bars reflect the standard deviation from the mean of independent duplicate samples.

Since the translocation of phosphatidylserine to the outer leaflet of the plasma membrane is an early event in apoptosis, it is possible that differences in annexin-positive populations induced by various E. coli isolates may not reflect differences in the eventual cellular fates. To confirm that NU14 induced apoptosis more efficiently than the piliated nonpathogen HB101/pWRS1-17, apoptosis was also assessed by examining the appearance of DNA degradation, a hallmark of apoptosis and a late apoptotic marker (Fig. 2B and C). DNA degradation was determined using propidium iodide staining followed by flow cytometry to quantify those cells whose DNA content was less than 2N (sub-G₁). Since pilus expression alone is sufficient to induce apoptosis, the levels of apoptosis induced by E. coli were also normalized to the levels of pilus expression, as determined by mannose-sensitive hemagglutination of erythrocytes. Incubation of TEU-2 cells with NU14 indeed resulted in roughly twice the number of sub-G₁ cells than did incubation with HB101/pWRS1-17 (11.3 and 5.2%, respectively). These results suggest that, although apoptosis is strictly dependent upon the expression of type 1 pili, the apoptotic potential of NU14 may be enhanced by pilus-independent mechanisms.

UPEC suppresses NF-κB activity.

Since the transcription factor NF-κB is involved in both apoptotic and inflammatory responses, we examined whether the UPEC strain NU14 could alter the NF-κB pathway. To assess NF-κB activity, TEU-2 cells were transiently transfected with a reporter construct where expression of the luciferase gene is driven by an NF-κB-dependent promoter, and a constitutive Renilla luciferase reporter was cotransfected as an internal control (Fig. 3). While stimulation of TEU-2 cells with LPS induced luciferase activity nearly sixfold relative to unstimulated cells, neither HB101 nor NU14 induced NF-κB-mediated transcription. Interestingly, stimulation of TEU-2 with LPS in the presence of NU14 cells blocked NF-κB-dependent induction of luciferase activity, whereas both HB101 and heat-killed NU14 failed to block LPS activation of NF-κB. Although neither HB101/pWRS1-17 nor the nonpathogenic isolate DH5α blocked LPS induction of NF-κB activity (not shown), the fimH variant NU14-1 did block activation of NF-κB. To eliminate the possibility that suppression of NF-κB was a unique property of the NU14 pathogenic isolate, another uropathogenic isolate was tested. The E. coli strain 8NU was obtained from the urine of a patient with bacterial cystitis and was also found to suppress NF-κB activation in response to LPS, as were other clinical isolates (unpublished data). Therefore, the suppression of NF-κB activation in response to LPS stimulation appears to be a property shared by multiple UPEC isolates.

FIG. 3.

UPEC inhibits urothelial NF-κB responsiveness to LPS. NF-κB activity was determined by cotransfection of pNFκB-luc and TK-Renilla, exposure to various E. coli isolates and/or LPS for 6 h, detection of luciferase by dual luciferase assay, and normalization of NF-κB activity to Renilla. LPS-induced NF-κB activity was inhibited by UPEC strains NU14 and 8NU but was not inhibited by heat-killed NU14 (ΔNU14). The suppression of NF-κB was not due to pili, since neither methyl α-d-mannopyranoside (Mann) nor mutation in fimH (NU14-1) altered NF-κB suppression by UPEC. Error bars reflect the standard deviation from the mean of three independent samples.

It was possible that suppression of NF-κB by UPEC was specific for the LPS stimulus, so we examined whether the UPEC-mediated suppression occurred in response to another stimulus. TNF-α is a proinflammatory cytokine that represents an alternative pathway for activating NF-κB, and NU14 bacteria were tested for their capacity to inhibit NF-κB activation by this alternative stimulus (Fig. 4). Similar to the results obtained with LPS stimulation, both live NU14 and 8NU blocked NF-κB activation in response to TNF-α, whereas heat-killed NU14 had no effect upon NF-κB activity (Fig. 4A). Taken together, these data indicate that UPEC has the capacity to suppress NF-κB-dependent transcriptional activity, and this property is not exhibited by laboratory strains of E. coli strain nor by heat-killed UPEC strains. The observations that the fimH mutant NU14-1, as well as wild type NU14 incubated in the presence of methyl α-d-mannopyranoside, retains the capacity to suppress NF-κB activation in TEU-2 cells suggest that a factor(s) other than the FimH adhesin of type 1 pili mediates NF-κB suppression by UPEC.

FIG. 4.

UPEC inhibits urothelial NF-κB response to TNF-α. Dual luciferase assays were performed to measure NF-κB activity in TEU-2 urothelial cells. (A) TNF-α (250 pg/ml) induced TEU-2 NF-κB activity that was inhibited by UPEC strains NU14 and 8NU as well as the fimH mutant NU14-1 but not inhibited by HB101 or heat-killed NU14 (ΔNU14). (B) The UPEC-mediated NF-κB suppression was not detected in conditioned media (CM). Error bars reflect the standard deviation from the mean of three independent samples.

To determine whether the suppression of NF-κB activation by NU14 was mediated by a soluble factor(s), the effects of NU14 supernatant on urothelial NF-κB activation were assessed (Fig. 4B). TEU-2 culture medium was inoculated with NU14, incubated, sterile-filtered, and then placed onto TEU-2 cells. The NU14-conditioned medium by itself activated NF-κB, and the effects of conditioned medium and TNF-α were additive. Similar results were also observed when NU14 was separated from TEU-2 cells by a semipermeable membrane (data not shown). We cannot exclude the possibility that a soluble factor(s) is produced at locally high concentrations at the E. coli-urothelial cell interface, and such a factor may have been diluted in these experiments. However, these data suggest that the suppression of NF-κB activity by NU14 requires close associations between E. coli and urothelial cells, and pilus-mediated interactions are not required since both the NU14-1 strain and NU14 in the presence of the mannose also exhibit the capacity to suppress urothelial NF-κB responses (Fig. 3 and 4A). Thus, nonspecific contacts, which nevertheless afford intimate host-pathogen interactions, are sufficient to suppress NF-κB responses.

UPEC blocks NF-κB translocation by stabilizing IκB.

To identify the mechanism by which UPEC blocked NF-κB activity, the subcellular localization of NF-κB was determined by indirect immunofluorescence, and NU14 was found to block the nuclear translocation of NF-κB in response to LPS stimulation (data not shown). Since translocation of NF-κB to the nucleus is dependent upon degradation of IκB, IκB protein levels were determined by immunoblotting (Fig. 5A). Whole-cell extracts contained IκB protein that was readily detectable in unstimulated TEU-2 cells, and IκB expression appeared unchanged when TEU-2 cells were incubated with NU14. In contrast, treatment with LPS resulted in a marked decrease in IκB protein, but coincubation with NU14 blocked the LPS-induced degradation of IκB. These data suggest that the inhibition of NF-κB activity by UPEC is mediated by stabilizing IκB, thereby preventing translocation of NF-κB to the nucleus.

FIG. 5.

UPEC stabilizes IκB and inhibits MAP kinase signaling. (A) TEU-2 cells were incubated in the presence or absence of UPEC isolate NU14 and LPS (1 μg/ml). Whole-cell extracts (15 μg) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, blotted onto an Immobilon membrane, and probed with anti-IκB serum. NU14 prevented the LPS-induced degradation of IκB. (B) TEU-2 cells starved of growth factors were preincubated with the UPEC isolate NU14 or the piliated nonpathogen HB101/pWRS1-17 (1-17). The cells were then stimulated with epidermal growth factor (EGF) (100 ng/ml) for 10 min, whole-cell lysates were prepared, and immunoblot analysis was performed using anti-phospho-Erk1/2 serum. NU14 blocked the epidermal growth factor-induced phosphorylation of Erk1/2.

Other gram-negative pathogens, such as Yersinia pestis (13, 14), can target multiple signaling pathways in host cells, so the UPEC isolate NU14 was examined for its effect on the mitogen-activated protein (MAP) kinase pathway by measuring the accumulation of phospho-Erk, an activated effector in the MAP kinase family (15, 25). TEU-2 cells starved of growth factors contained no detectable phospho-Erk, whereas epidermal growth factor (EGF) resulted in robust accumulation of phospho-Erk (Fig. 5B). While pretreatment of TEU-2 cells with HB101/pWRS1-17 did not affect the accumulation of phospho-Erk, NU14 strongly inhibited EGF induction of Erk phosphorylation. These data indicate that UPEC may alter multiple signaling pathways in host cells, including NF-κB and MAP kinase pathways.

NF-κB suppression enhances pilus-induced urothelial apoptosis.

To determine the functional consequences of NF-κB suppression upon the urothelial apoptotic response to type 1 pili, TEU-2 cells were treated with known inhibitors of NF-κB activity prior to exposure to piliated E. coli (Fig. 6). The oligopeptide ALLN is a known blocker of NF-κB signaling that functions by inhibiting the 26S proteasome (20, 23), and ALLN suppressed LPS induction of NF-κB-luciferase activity in TEU-2 cells (not shown). ALLN increased the level of apoptosis induced by HB101/pWRS1-17 to that induced by the UPEC strain NU14, although the drug itself had little effect on the level of background apoptosis in cells not treated with bacteria. ALLN also slightly increased the apoptosis induced by NU14, but this effect was likely due to nonspecific actions of the protease inhibitor.

FIG. 6.

NF-κB block enhances urothelial apoptosis induced by type 1 pili. TEU-2 cells were preincubated with the NF-κB inhibitor ALLN (25 μM) or an adenovirus encoding a superrepressor form of IκB (AdIκB) at an MOI of 100:1. Cells were then treated with the UPEC isolate NU14 or the piliated nonpathogen HB101/pWRS1-17. Apoptosis was then quantified by propidium iodide staining followed by flow cytometry to detect cells with a DNA content of less than 2N (Sub-G₁). Both inhibitors of NF-κB function increased the apoptosis induced by HB101/pWRS1-17 to levels similar to the apoptosis induced by NU14. Error bars reflect the standard deviation from the mean of two independent samples.

A nonphosphorylatable variant of IκB can also block NF-κB activation by acting as a superrepressor and preventing NF-κB translocation to the nucleus (21). TEU-2 cells were infected with an adenovirus encoding a nonphosphorylatable IκB variant (AdIκB), and NF-κB-luciferase induction in response to LPS was suppressed (data not shown). Like the proteasome inhibitor ALLN, infection with AdIκB rendered TEU-2 approximately twice as susceptible to the apoptotic effects of type 1 pili expressed by the nonpathogenic HB101/pWRS1-17. AdIκB also increased the level of apoptosis induced by NU14, and this effect is likely due to nonspecific actions of the adenovirus infection. Nonetheless, both the chemical inhibitor ALLN and the IκB superrepressor adenovirus elevated the level of apoptosis induced by the nonpathogen into a range similar to that induced by the UPEC strain. These findings suggest that the increased capacity of the UPEC strain NU14 to induce urothelial apoptosis, relative to the nonpathogen HB101/pWRS1-17, may be due at least in part to the ability of UPEC isolates to suppress NF-κB signaling. Consistent with the absence of apoptosis in response to NU14-1 (see Fig. 2), inhibition of NF-κB alone did not induce apoptosis, further supporting the notion that FimH-mediated apoptosis and NF-κB suppression are separable yet complementary events induced by UPEC.

DISCUSSION

We have shown that UPEC isolates were able to alter urothelial responses to inflammatory stimuli, yet this property was not shared by nonpathogenic laboratory strains of E. coli. UPEC was able to abrogate urothelial responses by blocking NF-κB translocation to the nucleus and by inhibiting NF-κB-dependent transcription in response to either LPS or TNF-α stimulation. These effects of UPEC were also found to be independent of the expression of intact type 1 pili on the bacterial surface, since an adhesin-deficient variant also blocked NF-κB signaling.

The suppression of NF-κB activity by UPEC was not inhibited by addition of methyl α-d-mannopyranoside, a competitive inhibitor of interactions between type 1 pili and mannosylated glycoproteins (1, 24). Likewise, the FimH mutant NU14-1 exhibited the same suppressive effects as wild-type NU14. These data demonstrate that adhesion mediated by type 1 pili is not required for NF-κB suppression, yet intimate contact between the E. coli and epithelial cells is apparently necessary, since neither bacterial conditioned media nor NU14 separated from urothelial cells by a permeable membrane suppressed NF-κB. Despite the lack of FimH-mediated binding by NU14-1, these data suggest that nonspecific interactions are sufficient in culture to permit intimate contact between UPEC and urothelial cells and presumably allow for the transport of virulence factors.

In contrast to the pilus-independent suppression of NF-κB, apoptosis of urothelial cells in response to NU14 was found to be strictly dependent upon interactions mediated by type 1 pili. These results indicate that UPEC utilizes distinct mechanisms to induce apoptosis and alters cellular signaling pathways, and the observations that laboratory E. coli expressing type 1 pili can also cause apoptosis (this study and reference 12) suggest that FimH is a major mediator of the urothelial apoptotic response induced by UPEC. Although NF-κB suppression alone is apparently not an inducer of urothelial apoptosis, UPEC suppression of NF-κB in urothelial cells may potentiate FimH-mediated apoptosis, since NF-κB activation can exert antiapoptotic effects. Indeed, we found that inhibiting NF-κB activation enhanced the apoptotic effects of piliated laboratory E. coli. These findings raise the possibility that, rather than urothelial apoptosis being strictly a host defense mechanism against E. coli infection (12), E. coli may actively enhance the apoptotic process as a component of its pathogenic program.

The type 1 pilus is thus far the best-characterized virulence factor of UPEC. The observation of pilus-independent NF-κB suppression by UPEC strongly suggests that an additional virulence factor(s) is involved in UTI pathogenesis. The ability of UPEC to block both NF-κB activation and MAP kinase signaling is reminiscent of the YopJ virulence factor of Yersinia pestis. YopJ has been shown to block NF-κB activation and MAP kinase signaling by binding and preventing the activation by the upstream MKKs and IKKβ kinases (13, 14). By inhibiting parallel pathways activated in the inflammatory response, YopJ can simultaneously suppress the expression of cytokines while promoting apoptosis during infection. Based on these similarities, UPEC may produce a virulence factor that is analogous to YopJ, and efforts to identify such a factor are under way.

The findings reported here suggest a new model for early events in the pathogenesis of UTIs. In this model, UPEC facilitates the apoptotic host defense response of urothelial cells through the proapoptotic effects of interactions with the FimH component of type 1 pili and through an unknown virulence factor(s) that potentiates the apoptotic response by suppressing NF-κB activity. As a consequence of the block of NF-κB, UPEC may also alter the expression dynamics of proinflammatory cytokines, thus leading to delayed or reduced neutrophil infiltration and associated phagocytosis of the pathogen. Indeed, we find that NU14 elicits a significantly diminished secretion of interleukin 8 from TEU-2 cells compared with a piliated K-12 strain (data not shown). The combined influence of these events may be to extend the window of opportunity for execution of other aspects of the pathogenic program. Recent evidence that E. coli can be internalized by urothelial cells and enter an intracellular vegetative cycle as a possible mechanism underlying recurrent UTIs (11) suggests that a delayed host inflammatory response may enhance the probability for uropathogens to induce internalization and thus gain shelter from host immune effector cells.

In summary, we show that UPEC has the capacity to suppress a component of the urothelial inflammatory response. This suppression is independent of type 1 pili and thus suggests the existence of an additional virulence factor(s) involved in the pathogenesis of UTIs. Finally, this study suggests a complex series of early events in the pathogenesis induced by E. coli that may enhance the potential for recurrent UTIs.