INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Enteropathogenic Escherichia coli (EPEC) associated with severe infantile diarrhea represents a major health problem among infants, particularly in developing countries (37). Research using cultured epithelial cells indicates that EPEC attaches to host cells initially in a loose manner and then consolidates attachment in a more intimate manner (17). The initial adherence phenotype, characterized in tissue culture assays as localized adherence, is associated with the production of plasmid-encoded type IV fimbriae known as bundle-forming pili (BFP) (15, 21). More intimate attachment, characterized by the development of attaching and effacing (A/E) lesions of the brush border microvilli, is encoded in a chromosomal region termed the locus of enterocyte effacement (LEE) (32). Recent studies with pediatric intestinal biopsy samples have minimized the role of BFP in host adhesion and have alternatively implicated BFP in the formation of bacterial aggregates which produce the localized adherence pattern typical of EPEC infection (25). Nevertheless, studies with volunteers who have ingested BFP-expressing and non-BFP-expressing EPEC strains have confirmed BFP as a virulence factor (5).

Attachment of EPEC to the host cell is accompanied by a number of signal transduction events, including release of inositol triphosphate and calcium, phosphorylation of myosin light chain, and activation of protein kinase C (10, 18). EPEC also synthesizes and translocates into the host cell a protein known as translocated intimin receptor (Tir), which after tyrosine phosphorylation permits intimate attachment through the bacterial protein intimin (41). Recently, we and others have reported that EPEC also induces cell death in cultured epithelial cells (2, 3, 11). Evidence of both apoptosis and necrosis has been observed. However, the bacterial structures responsible for the triggering of these cell death pathways have not been identified. In this study, we demonstrate a role for BFP in the induction of cell death, including apoptosis, in host epithelial cells.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and cultivation conditions. The characteristics of bacterial strains used in this study are listed in Table 1. The E2348/69 derivatives 31-6-1(1), JPN 15, and E2348/69(pOG127) as well as HB101pMAR7 and HB101(pCVD426) were kindly provided by J. Kaper, University of Maryland. 31-6-1(1) is a previously described mutant of E2348/69 with a TnphoA insertion in the bfpA gene of the pMAR2 (60 MDA virulence plasmid from E2348/69) plasmid (14, 15). JPN15 is an E2348/69 derivative cured of plasmid pMAR2 during passage through a volunteer (27). The plasmid pOG127 (pMAR2 plasmid with a perA::cat mutation) was transferred to strain JPN15 to generate E2348/69(pOG127). Since Per (plasmid-encoded regulator) regulates bfp expression, this strain expresses BFP at lower levels than E2348/69. CVD206 is an eae mutant of E2348/69 constructed using a suicide vector with a pir-dependent R6K replicon and the sacB gene of Bacillus subtilis (16). HB101(pMAR7) is an avirulent laboratory strain, HB101, complemented with pMAR7 plasmid (an ampicillin-resistant derivative of the EPEC adherence factor [EAF] plasmid) which contains the bfp gene (23). HB101(pCVD426) is complemented with pCVD426 generated by cloning the entire LEE region from E2348/69 into the cosmid vector pCVD551 (33). Bacteria were stored in tryptic soy broth containing 20% (vol/vol) glycerol at 70°C. Prior to use, bacteria were cultured on Trypticase soy agar with 5% defibrinated sheep blood supplemented with the appropriate antibiotics as listed in Table 1. Trypticase soy blood agar has been reported to maximize BFP expression (21). Bacterial expression of BFP was assessed by Western blotting using a polyclonal anti-BFP antiserum (generous gift of J. Giron, Universidad Autonoma de Puebla, Puebla, Mexico) as previously described (21). After overnight growth, the bacteria were harvested, washed once with antibiotic-free cell culture medium, and resuspended in the epithelial cell culture medium as indicated below.

Cell culture. The human epithelial cells HEp-2 (human laryngeal cell line), HeLa (human cervical cell line), and Caco-2 (human colonic cell line) were obtained from American Type Culture Collection, Rockville, Md. HEp-2 and HeLa cells were grown in minimum essential medium (Gibco Laboratories, Grand Island, N.Y.) supplemented with decomplemented 10% fetal calf serum (Cansera International Inc.), 0.5% L-glutamine (ICN Biomedicals Inc., Costa Mesa, Calif.), 0.1% sodium bicarbonate (ICN), and 0.1% gentamycin at 37°C in 5% CO₂. The human colonic Caco-2 cells were grown in minimum essential medium with Earl's salts (Gibco BRL) supplemented with 0.5% L-glutamine, nonessential amino acids, 10% fetal calf serum, and 0.1% gentamycin (Gibco BRL) at 37°C in 5% CO₂.

Assessment of outer leaflet levels of PS. The level of outer leaflet phosphatidylserine (PS) on cells incubated with media or bacteria was determined by flow cytometry following treatment with fluorescein isothiocyanate-conjugated annexin V (annexin V-FITC; Pharmingen International, Mississauga, Ontario, Canada) according to the method of Vermes et al. (44). Approximately 10⁶ cells (70% confluent monolayer) were infected with 10⁸ bacteria in culture medium without antibiotics for 5 to 18 h at 37°C in a CO₂ incubator. After incubation, both detached (supernatant) and adherent (trypsinized) cells were harvested and the nonadherent bacteria were removed by centrifugation (260 × g for 10 min at 4°C) after mixing with isotonic solution of 15% sucrose in phosphate-buffered saline. The pellets were then resuspended in 100 µl of binding buffer (10 mM HEPES, 140 mM NaCl, and 2.5 mM CaCl₂, pH 7.4), and thereafter 5 µl of annexin V-FITC (Pharmingen) was added to each tube. The samples were then incubated at room temperature for 20 to 30 min, mixed with 400 µl of binding buffer, screened, and supplemented with 0.5 µg of propidium iodide (PI) (Pharmingen) prior to flow cytometric analysis. An increase in outer leaflet PS, detected by annexin V-FITC, provides an indication of apoptosis and necrosis. Cells which are late apoptotic or necrotic lose membrane integrity and will stain with both annexin V-FITC and PI. Cells which retain membrane integrity, including viable and early apoptotic cells, will not take up PI. Therefore, the combined use of annexin V-FITC and PI can distinguish between early apoptotic and late apoptotic or necrotic cells (24). To assess the effect of anti-BFP on EPEC induction of cell death, E2348/69 was preincubated with equivalent concentrations of either rabbit anti-BFP or rabbit nonimmune serum for 45 min at room temperature followed by incubation with HEp-2 cells per the standard protocol. Cell death was analyzed by flow cytometry as described above.

Electron microscopic analysis of cell death. Cell death was analyzed by electron microscopy as previously described (3). Subconfluent monolayers of all three cell lines were incubated with media or bacteria (10⁸) for 5 h and were then analyzed by electron microscopy for morphological changes indicative of necrosis or apoptosis. Verotoxin 1 (VT1) treatment was used as a control for induction of apoptosis. Cells which express outer leaflet Gb₃ have been shown to be sensitive to VT-induced apoptosis (28, 43). In all cases, trypsinized cell monolayers and detached cells (supernatant) were washed twice with phosphate-buffered saline, gently overlayed with 1 ml of universal fixative (equal parts of formaldehyde and 1% glutaraldehyde), and postfixed in 2% osmium tetroxide. Dehydration was carried out in graded ethanol, followed by propylene oxide and embedding in Epon. One-micrometer-thick sections were stained with toluidine blue and lead citrate. Electron microscopic examination was conducted using a Philips 201 (N. V. Philips, Gloeilampenfabrieken, Eindhoven, The Netherlands) transmission electron microscope.

Analysis of internucleosomal fragmentation. Analysis of DNA fragmentation was carried out as previously described (3). Briefy, all trypsinized cells and those from supernatants were harvested by centrifugation and gently lysed with hypotonic lysis buffer (10 mM Tris [pH 7.4], 1 mM EDTA, 0.2% Triton X-100) on ice for 10 min. After 10 min of centrifugation at 13,700 × g and 4°C, the supernatants were mixed well with an equal volume of 1:1 phenol-chloroform and recentrifuged. After transfer, the upper phases containing DNA were incubated with 1 µg of glycogen, and a 1/10 volume of 3 M sodium acetate and 1 ml of absolute ethanol were added and incubated at 20°C overnight. The DNA was pelleted by centrifugation (14,000 × g for 20 min) at 4°C, washed once with 70% ethanol, and air dried at room temperature for 30 min. The DNA pellets were dissolved in 10 µl of Tris-EDTA buffer (10 mM Tris [pH 7.4], 1 mM EDTA), mixed with 12 µl of RNase (20 µg/ml in Tris-EDTA), and incubated at 37°C for 30 min. Samples were incubated with 3 µl of loading buffer (50 mM EDTA, 15% [wt/vol] Ficoll, 0.25% [wt/vol] bromophenol blue) at 65°C for 15 min and electrophoresed on a 1.5% agarose gel at 50 V for 90 min.

Bacterial adhesion to epithelial cells. Bacterial binding to epithelial cells was assayed as previously described (3). Approximately 10⁸ bacteria were incubated with 10⁶ cells (70 to 80% confluent monolayer) for 2 h at 37°C. After incubation with bacteria, both detached and adherent (trypsinized) cells were harvested and separated from nonadherent bacteria by centrifugation through an isotonic 15% sucrose solution. Bacterial binding was detected using a polyclonal anti-E. coli (all antigens) antibody (Virostat) and a goat anti-rabbit-FITC conjugate (Sigma) and quantified by fluorescence-activated cell sorting analysis using a Becton Dickinson FACSscan flow cytometer. All samples were analyzed with Cell Quest software. Histogram plots showing cell count versus fluorescence intensity are provided.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Effect of BFP expression on host cell morphology. Electron microscopic analysis of epithelial cell morphology indicated a correlation between BFP expression and the extent of apoptosis and necrosis. Western blot analysis confirmed expression of BFP for all bfp-positive strains and lack of expression for all bfp-negative strains (data not shown). Evidence of apoptosis included membrane blebbing, nuclear condensation, and margination. Necrotic cells, on the other hand, were typically larger and lighter, with plasma membrane lesions and mitochondrial abnormalities. For both HeLa (Fig. 1) and Caco-2 (Fig. 2) cells, all BFP-expressing strains, including E2348/69, HB101(pMAR7), and E2348/69(pOG127), induced consistently significantly higher levels of apoptosis and necrosis than those produced by incubation with the corresponding BFP-negative strains, 31-6-1(1), HB101, and JPN15, respectively (P < 0.05). Even CVD206, which expresses BFP but not intimin, triggered higher levels of cell death than did the BFP-negative strains, although the levels were somewhat lower than those expressing both intimin and BFP. Untreated cells showed less than 5% apoptosis and necrosis combined (not shown). Overnight treatment of HeLa cells with VT1 induced levels of cell death similar to those for the BFP-expressing strains. Caco-2 cells, which have been reported to express lower levels of plasma membrane VT receptor (28) and should therefore be less sensitive to VT1 treatment, showed lower levels of cell death in response to overnight VT1 treatment. HeLa cells express higher levels of membrane VT receptor (26) and were indeed more sensitive to VT-induced cell death.

View larger version (16K): [in this window] [in a new window]   FIG. 1.   Electron microscopic determination of apoptosis and necrosis of HeLa cells after 5-h incubation with various E. coli strains. Percentages were based on a count of at least 100 cells and experiments were repeated twice. Data are expressed as means ± standard deviations. Black bars, necrotic cells; white bars, apoptotic cells. Untreated cells showed less than 5% apoptosis and necrosis combined.

View larger version (16K): [in this window] [in a new window]   FIG. 2.   Electron microscopic determination of apoptosis and necrosis of Caco-2 cells after 5-h incubation with various E. coli strains. Percentages were based on a count of at least 100 cells and experiments were repeated twice. Data are expressed as means ± standard deviations. Black bars, necrotic cells; white bars, apoptotic cells.

Effect of BFP expression on host cell outer leaflet PS. Flow cytometric analysis of three cell lines infected with various E. coli strains resulted in increased outer leaflet PS levels after treatment with any of the BFP-expressing strains tested. Figure 3 shows the histogram plots of annexin V-FITC staining (to detect outer leaflet PS) for HEp-2, Caco-2, and HeLa cells treated with BFP-positive strains and the corresponding BFP-negative strains. In each case, treatment with the BFP-expressing strains [E2348/69, E2348/69(pOG127), and HB101(pMAR7)] resulted in elevated outer leaflet PS levels compared to treatment with the non-BFP-expressing strains [31-6-1(1), JPN15, and HB101]. HB101(pCVD426), which expresses the intimin adhesin but not BFP, showed PS levels identical to those for HB101 (see Fig. 7). Treatment with sterile filtered supernatants resulted in PS levels similar to those of untreated cells (not shown). Preincubation of EPEC with anti-BFP reduced the level of cell death relative to controls with either no antiserum or equivalent concentrations of rabbit nonimmune serum (Table 2). At a concentration of 0.89 mg of anti-BFP per ml, cell viability was increased to 65%, over that with no antiserum (43%) or rabbit nonimmune serum (50%).

View larger version (43K): [in this window] [in a new window]   FIG. 3.   Flow cytometric analysis of outer leaflet levels of PS in HEp-2 (a), Caco-2 (b), and HeLa (c) cells. The x axis indicates staining intensity on the cells and the y axis indicates relative cell number. Membrane PS was detected by flow cytometry using annexin-FITC. Cells were preincubated for 5 h with E2348/69 (black line) and 31-6-1(1) (dotted line) (A), E2348/69(pOG127) (black line) and JPN15 (dotted line) (B), or HB101(pMAR7) (black line) and HB101 (dotted line) (C).

Effect of BFP expression on DNA fragmentation. Agarose gel electrophoresis of cellular DNA showed evidence of internucleosomal DNA fragmentation in all cell lines infected with the BFP-expressing strains E2348/69, HB101(pMAR7), and E2348/69(pOG127). Figure 4 shows the electrophoretic result of extracted DNA from HeLa cells and clearly indicates DNA degradation after a 5-h incubation with HB101(pMAR7), E2348/69(pOG127), and E2348/69 and overnight treatment with VT1. Some DNA degradation was also observed with CVD206. No such degradation was observed when HeLa cells were untreated or incubated with HB101, JPN15, or 31-6-1(1). Similar results were achieved using Caco-2 and HEp-2 cells (not shown). These DNA patterns, although not the typical ladders generated by other apoptosis-inducing agents, are consistent with those reported for other apoptosis-inducing bacteria (11) and likely represent the mixed apoptosis-necrosis indicated by the electron microscope and flow cytometry results in this study. Furthermore, others have reported that cleavage of internucleosomal DNA in epithelial cells may be defective, resulting in less distinctive DNA fragmentation patterns (38).

View larger version (95K): [in this window] [in a new window]   FIG. 4.   Analysis of internucleosomal DNA fragmentation. Agarose gel electrophoresis of DNA extracts from HeLa cells after incubation with medium (lane 1), 200 ng of VT (24 h) (lane 2), HB101(pMAR7) (5 h) (lane 3), HB101 (5 h) (lane 4), E2348/69(pOG127) (5 h) (lane 5), JPN15 (5 h) (lane 6), CVD206 (5 h) (lane 7), 31-6-1(1) (5 h) (lane 8), E2348/69 (5 h) (lane 9), and 100-kb DNA ladder standard (lane 10).

Relative contributions of bfp and eae. In order to assess the relative contributions of bfp and eae genes to cell death, we compared treatment with wild-type EPEC E2348/69 and the eae-negative mutant and bfp-negative mutant of E2348/69 and assessed by flow cytometry the extent of apoptotic and necrotic cell death. Figure 5 shows the dot plots for four treatments of HeLa cells, comparing the intensity of annexin V-FITC staining with PI staining. Lower left quadrants reflect percentages of viable cells (nonapoptotic, nonnecrotic cells; i.e., low levels of both annexin-FITC and PI). Lower right quadrants indicate percentages of early apoptotic cells while upper quadrants (elevated PI staining) reflect cells in late apoptosis or necrosis. Treatment with E2348/69 showed increased numbers of both apoptotic and necrotic cells relative to the untreated sample. Disruption of the bfp gene in 31-6-1(1) reduced the percentage of early apoptotic cells to that of the untreated sample. The level of late apoptosis or necrosis was also diminished with the bfp-negative mutant. Disruption of the eae gene in CVD206 resulted in levels of early apoptosis similar to that for the wild type, E2348/69, but with markedly less necrosis and late apoptosis. This result suggests that BFP may be important in the induction of apoptosis while intimin may play a more significant role in the induction of necrosis. Certainly, the reduction of early apoptosis to background levels with the bfp mutant and the maintenance of wild-type levels of early apoptosis with the eae mutant support the involvement of BFP but not intimin in the induction of apoptosis. Since the upper quadrant likely includes cells which have undergone apoptosis and eventually lost membrane integrity as well as those cells which have proceeded through a necrotic pathway, it is not possible to definitively assign apoptosis and necrosis induction roles.

View larger version (40K): [in this window] [in a new window]   FIG. 5.   Flow cytometric analysis of PS exposure in HeLa cells treated with E. coli strains. Exposure of membrane PS demonstrated by dot plots showing fluorescent intensities of annexin-FITC (abscissa) and PI (ordinate). Quadrant numbers indicate percentages of cells simultaneously stained with both fluorophores, with intensities shown on respective axes. (A) Untreated cells. (B) Cells treated with E2348/69 strain (5 h). (C) Cells treated with 31-6-1(1) strain (5 h). (D) Cells treated with CVD206 strain (5 h).

Bacterial adhesion to epithelial cells. Consistent with previous reports, BFP expression correlated with adhesion to cultured epithelial cells (Fig. 6). BFP-expressing strains generally adhered better to both HeLa (shown) and Caco-2 (not shown) cells than did the non-BFP-expressing correlates. HB101(pMAR7) adhered better to HeLa cells than did HB101(pCVD426) (Fig. 6D), which adhered only slightly better than HB101, but only HB101(pMAR7) was capable of inducing significant levels of cell death (Fig. 7).

View larger version (16K): [in this window] [in a new window]   FIG. 6.   Flow cytometric detection of bacterial adhesion to HeLa cells. HeLa cells (106) were infected with 108 bacteria for 2 h. Bacterial adhesion was detected with rabbit anti-E. coli (all antigens) and FITC-goat anti-rabbit conjugate. The x axis represents the staining intensity of the cells, and the y axis represents the relative cell number. (A) JPN15 (black) and E2348/69(pOG127) (gray). (B) 31-6-1(1) (dotted), CVD206 (black), and E2348/69 (gray). (C) HB101 (black) and HB101(pMAR7) (gray). (D) HB101(pCVD426) (dotted) and HB101(pMAR7) (gray).

View larger version (24K): [in this window] [in a new window]   FIG. 7.   Flow cytometric analysis of PS exposure in HeLa cells treated with HB101 strains. Exposure of membrane PS demonstrated by dot plots showing fluorescent intensities of annexin-FITC (abscissa) and PI (ordinate). Quadrant numbers indicate percentages of cells simultaneously stained with both fluorophores, with intensities shown on respective axes. (A) Untreated cells. (B) Cells treated with HB101 (4 h). (C) Cells treated with HB101(pCVD426) (4 h). (D) Cells treated with HB101(pMAR7) (4 h).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial colonization and infection of various tissues is the result of recognition by bacterial adhesins of specific receptors on the target tissue. Once the microorganism approaches the host cell surface, it is essential for the pathogen to use its full genetic potential for attachment and the synthesis of other traits to secure a specific niche within the host that will permit its replication and survival. Accordingly, the induction of host cell death may represent a strategy developed by microorganisms to ensure their survival, infection, and invasion of the target tissue.

Induction of host cell death has been reported for several gastrointestinal pathogens, including Yersinia (34), Salmonella (35), Shigella (47, 48), Helicobacter pylori (36), and enterohemorrhagic E. coli (EHEC) (3). EPEC infection has been reported to cause pathological damage of the target tissue; however, features of apoptotic cell death were not indicated (20, 42). Evidence of EPEC-induced apoptotic and necrotic cell death has been reported for cell lines including HEp-2, T84, Caco-2, and HeLa cells (3, 11).

Our investigation of the cell death triggered by E. coli strains indicates a role for BFP in the induction of apoptosis. Flow cytometric analysis showed that the epithelial cell apoptosis was much greater with any of the BFP-expressing E. coli strains than with any of the bfp-negative strains, including the wild-type plasmid-cured EPEC strain (JPN15), the bfp-negative mutant of E2348/69 [31-6-1(1)], the nonpathogenic laboratory strain, HB101, and HB101(pCVD426). Using PS exposure as a marker of apoptosis, we tested treatment of HEp-2, Caco-2, and HeLa cells with various BPF⁺ and BFP strains. In all cases, the BFP-expressing strains induced elevated levels of apoptosis over their non-BFP-expressing correlates. Interestingly, HEp-2 and HeLa cells showed significant induction of apoptosis after incubation with HB101(pMAR7) in contrast to other BFP⁺ strains. This may be a consequence of interference by lipopolysaccharide or type 1 fimbriae expressed by wild-type E2348/69 and E2348/69(pOG127) strains but not by HB101. This is consistent with other reports that lipopolysaccharide interferes with attachment of EHEC, a related A/E pathogen, to cultured epithelial cells (6, 9).

Bacterial binding was required to induce apoptosis. Treatment with sterile filtered supernatants resulted in background levels of cell death (5%) found in untreated cells. In all cases, BFP-expressing strains induced significantly higher levels of cell death and showed higher levels of adhesion to the same cell lines. Levels of adhesion were consistent with the ability to induce apoptosis with HB101(pMAR7)- and E2348/69-infected cells showing the highest levels of both adhesion and apoptosis. The absence of the BFP adhesion appeared to have a more profound effect on induction of cell death than did the absence of other adhesins such as intimin (in strain CVD206). Preincubation with anti-BFP at concentrations which have been shown to reduce EPEC adhesion (21) was also found to inhibit EPEC induction of cell death.

Analysis of internucleosomal DNA fragmentation confirmed the induction of apoptosis by the BFP-expressing strains. Electron microscopic analysis of Caco-2 and HeLa cell lines incubated with BFP-expressing E. coli strains also shows a clear increase in both apoptotic and necrotic cells in contrast to the BFP-negative strains.

Through a double labeling experiment with annexin V-FITC and PI, we were able to distinguish between early apoptotic and late apoptotic or necrotic cells. In this analysis, HeLa cells treated with E2348/69, 31-6-1(1), and the eae-deficient CVD206 strain revealed that the eae mutant strain induced higher levels of apoptosis and less necrosis than the wild-type parent strain. This has also been observed with the T84 cell line (data not shown). Ablation of the bfp gene in strain 31-6-1(1), which still expresses intimin, reduced the level of early apoptotic cells to that of untreated cells. However, as we have previously noted, the detection of necrosis in vitro may not be physiological, since nonphagocytosed apoptotic cells may eventually lose their membrane intregrity and appear necrotic (3). Consequently, while we note differences in the relative levels of apoptosis and necrosis, we have focused primarily on the extent of apoptosis triggered by bacterial binding.

A recent study has implicated the secreted virulence factor EspF in host cell death triggered by EPEC (12). In this study, an espF mutant was attenuated in its ability to induce host cell death as measured by uptake of ethidium homodimer. Additional experiments showed that COS and HeLa cells transfected with espF appeared to undergo apoptotic cell death as determined by morphological changes. While the experiments demonstrate a role for EspF as a cell death factor, the authors conclude that EspF cannot completely account for the host cell killing by EPEC. They note that the cell killing ability of the espF mutant was reduced but not entirely lost. It should also be noted that the extent of cell death in the espF-transfected cells may not be equivalent to that of the in vivo situation due to possible overexpression of EspF compared with that produced by natural infection with EPEC. However, these studies do indicate a role for EspF in EPEC-triggered cell death. On the other hand, the evidence provided here shows clear induction of host cell death by all BFP-expressing strains, including HB101(pMAR7), and significantly less cell death with non-BFP-expressing strains, including HB101(pCVD206), than with BFP-expressing strains. These results indicate a role for BFP in mediating host cell death. It is likely that the induction of cell death, like that of the EPEC adhesion mechanism, is a multifactorial event involving several virulence factors including EspF and BFP.

Induction of apoptosis by BFP-expressing strains may be related to BFP receptor expression by the host cell. We have previously demonstrated using thin-layer chromatography and liposome assays that EPEC recognizes phosphatidylethanolamine (PE) in a specific and dose-dependent manner and that adhesion to human epithelial cells is inhibitable by anti-PE (4). We have also found that BFP expression correlates with PE recognition (28a), which when considered together with the present findings suggests that bacterial binding to host membrane PE may play a role in the induction of apoptosis. All cell lines tested in this study expressed PE. Variations in the level of plasma membrane outer leaflet PE may explain differences in the extent of apoptosis induced in different cell lines. We have shown that augmentation of outer leaflet PE levels through PE-liposome uptake by epithelial cells does increase bacterial binding, which in turn increases exposure of the apoptotic marker PS (3).

The ligation of membrane PE by BFP-expressing EPEC may trigger downstream events involving key mediators of apoptosis, including ceramide. Ceramide levels may be increased through the inhibition of ceramide acylation. A phospholipase A₂-like transacylase transfers an acyl group from PE to ceramide (1). Therefore, bacterial binding to host PE receptors may reduce the PE pool, restrict this reaction, and elevate ceramide levels. PE is also a substrate of phospholipase D, which catalyzes the formation of phosphatidic acid (PA) (29, 45) which plays a pivotal role in the balance of mitogenic and apoptotic responses. Sphingosine and sphingosine-1-phosphate increase intracellular levels of PA, thereby stimulating mitogenesis, while ceramide inhibits phospholipase D activation, thereby decreasing PA levels and enhancing the apoptotic response (22). The selective ligation of membrane PE by BFP-expressing EPEC could reduce the availability of PE for phospholipase D-mediated hydrolysis, resulting in reduced levels of cytosolic PA and disrupting the mitogen signaling cascade. Bacterial induction of apoptosis could lead to a further reduction in cytosolic exposure of PE, thereby enhancing this effect.

BFP binding to host membrane PE may also contribute to membrane events associated with apoptosis. PE is a nonbilayer phospholipid which has a high tendency to form the HII inverted micelle phase (7, 13). Therefore, PE-rich membranes tend to exert bending fluctuations (8) and therefore to promote physical membrane changes, including membrane fusion, inward membrane bending, and membrane budding. These changes may affect membrane permeability, endocytosis, cell division, and cell budding associated with the formation of apoptotic bodies. The importance of membrane PE is emphasized by findings which show that lowering the PE content diminished cell death during simulated ischemia and reperfusion (40). Therefore, events which sequester membrane PE, including bacterial binding, may contribute to apoptotic membrane changes. Further study is required to determine the mechanism by which bacterial ligation of host cell PE contributes to the induction of host cell death.

We have also reported that EHEC, another A/E gastrointestinal pathogen, binds PE and that EHEC binding to two epithelial cell lines correlates with plasma membrane outer leaflet levels of PE (4). Furthermore, EHEC, in a similar manner to EPEC, induces apoptosis in a number of epithelial cell lines, which results in increased outer leaflet PE and bacterial binding (3). The mechanism by which EHEC induces apoptosis is unknown. It may be that EHEC also expresses a BFP-like pilus which may mediate host cell PE binding and apoptosis. Interestingly, a mutation in the ler (LEE-encoded regulator) gene of EHEC resulted in the expression of an unidentified pilus and also enhanced epithelial cell adhesion (19).

In conclusion, these results clearly indicate a direct correlation between BFP expression and the induction of host cell death including apoptosis. BFP, an established EPEC virulence factor, has been previously implicated in initial host cell attachment and in bacterial autoaggregation. These findings now define another role for BFP in the pathogenesis of EPEC infection. Our earlier work has proposed that the induction of apoptosis by EPEC and EHEC provides a bacterial advantage by augmenting outer leaflet levels of the PE receptor candidate (3). Although apoptotic cells are eventually phagocytosed in vivo, receptor amplification through apoptosis can offer a temporary adhesion advantage to the bacterium. Other advantages may include enhanced access to nutrients and to the subepithelial layer. It has been suggested that apoptosis of epithelial cells triggered by Pseudomonas aeruginosa may serve as a vehicle for clearance and bacterial dissemination (39, 46). Alternatively, cell death may function as a host mechanism to limit bacterial infection. EPEC also stimulates a number of antiapoptotic pathways within the cell and it has been suggested that EPEC has developed strategies to slow host cell killing (10). It may be that the coordination of these apoptotic and antiapoptotic signaling events in the host cell by the bacterium ultimately determines the infection outcome.