INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Enteropathogenic Escherichia coli (EPEC) is a significant cause of diarrhea in young children, especially those less than 6 months of age (7, 45). Clinical symptoms are associated with intestinal attaching and effacing (A/E) lesions, which are characterized by the degeneration of microvilli, and the intimate association of bacteria with pedestal-like structures formed from host epithelial cells (3, 28, 34, 41). Donnenberg and Kaper originally proposed that EPEC colonization of intestinal epithelial surfaces occurs by a three-stage process (10). According to this multistep model, the initial binding of EPEC to intestinal epithelial cells involves their nonintimate attachment as microcolonies. Next, the bacteria induce signal transduction processes in host cells, which results in cytoskeletal rearrangement and the effacement of surface microvilli. In the final stage of attachment, the cytoskeletal components are organized to form pedestal-like structures which partially surround adherent organisms. The close proximity of bacteria to these pedestal-like structures constitutes the intimate attachment characteristic of A/E lesions. More recently, Hicks et al. proposed a modified model for EPEC colonization. This model is similar to the three-stage model of Donnenberg and Kaper except that three-dimensional microcolonies are thought to develop after, not before, intimate attachment (15).

Since adherence is an important factor in EPEC pathogenesis, considerable research has been performed to identify bacterial and eukaryotic cell structures involved in attachment. So far, two bacterial structures have been relatively well characterized. The first, bundle-forming pili (BFP), are associated with the attachment of EPEC as microcolonies to discrete sites on epithelial cells, a pattern which is referred to as localized adherence (LA) (4, 14, 36). Scanning electron micrographs of LA EPEC revealed that BFP are involved in mediating interbacterial linkages within microcolonies. Whether the BFP also function as adhesins for EPEC binding to epithelial cells remains to be resolved, however, since these structures appear to mediate bacterial binding to HEp-2 cells but not to human intestinal tissue in organ culture (14, 15, 40). A second bacterial protein involved in attachment is intimin (18, 19). This outer membrane protein is necessary for the intimate attachment of EPEC to epithelial cells and functions to focus host cell cytoskeletal components beneath adherent bacteria to form A/E lesions (31).

In contrast to what is known about bacterial structures involved in attachment, EPEC receptors on eukaryotic cells are less well characterized. Rosenshine et al. originally identified a 90-kDa tyrosine-phosphorylated protein (Hp90) in epithelial cell membranes which bound intimin (33). This protein was subsequently shown to be a secreted bacterial protein, called Tir (translocated intimin receptor), which, after translocation into the eukaryotic cell, becomes phosphorylated and serves as the receptor for intimin (6, 21). Frankel et al. reported that intimin may also bind to ₁ integrins (13).

In addition to these reports, several groups have identified oligosaccharide structures that are potential receptors for EPEC. By using various experimental approaches, N-acetylgalactosamine (35), fucosylated milk oligosaccharide sequences (5), GM₃ gangliosides (16), and the GalNAc(14)Gal portion of asialo-GM₁ and asialo-GM₂ structures (17) have all been implicated in EPEC attachment to eukaryotic cells. As well, we previously reported that EPEC bound to asialo-lactosamine sequences of N-linked glycoproteins on Chinese hamster ovary (CHO) cells (43). A possible role for O-linked glycoproteins or glycolipids in EPEC attachment to CHO cells was also suggested by data presented in our previous report.

Several reports suggest that, in general, EPEC recognize lactosyl structures on epithelial cells. However, additional carbohydrate residues (i.e., sialic acid and fucose) are frequently attached to these core structures. While our previous results with CHO cell Lec mutants indicated that EPEC do not require sialic acid in order to bind, the importance of fucose in these interactions was not addressed since CHO cells do not express certain fucosylated glycans (30). In consideration of previous results which suggested that fucosylated structures are preferentially recognized by EPEC (5), we performed binding inhibition experiments with synthetic lactosamine or fucosylated lactosamine bovine serum albumin (BSA)-glycoconjugates to investigate the role of fucose in EPEC attachment to epithelial cells. Our results demonstrated that N-acetyllactosamine (LacNAc)-BSA, followed by Lewis X (LeX)-BSA, were the most effective soluble inhibitors of EPEC E2348/69 attachment to HEp-2 cells. Furthermore, the interaction of bacteria with these specific glycoconjugates, alone, caused a decrease in the expression of BfpA, the structural subunit of BFP (9, 38), and intimin, in these organisms. These results suggest that EPEC may possess a mechanism for regulating the expression of these proteins which could contribute to the reduced ability of the bacteria to bind to HEp-2 cells.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Reagents. All glycoconjugates consisted of chemically synthesized oligosaccharide sequences (Table 1) conjugated to BSA through an 8-methoxycarbonyloctyl linker arm (25). LacNAc-BSA and Lewis Y (LeY)-BSA were generously provided by O. Hindsgaul (University of Alberta, Edmonton, Alberta, Canada). LeX-BSA was purchased from the Alberta Research Council (Edmonton, Alberta, Canada). The incorporation of ligands into BSA was determined by mass spectroscopy to be as follows (in mol/mol): LacNAc-BSA, 19:1; LeX-BSA, 26:1; and LeY-BSA, 17:1. All glycoconjugates were solubilized in phosphate-buffered saline (PBS) (5 mg/ml) and stored at 20°C prior to use. Polyclonal rabbit anti-intimin (18) and anti-BfpA (47) antibodies were kindly provided by J. B. Kaper and M. S. Donnenberg (University of Maryland School of Medicine, Baltimore), respectively. Rabbit anti-maltose-binding protein (anti-MBP) antibodies were purchased from New England Biolabs (Mississauga, Ontario, Canada).

Bacterial strains. EPEC E2348/69 (O127:H6), which is a wild-type strain isolated from an infant with diarrhea (26), was generously provided by B. B. Finlay (University of British Columbia, Vancouver, British Columbia, Canada). In each experiment, bacteria which were stored as frozen stock cultures at 70°C were grown overnight at 37°C on tryptic soy agar plates (Difco, Detroit, Mich.). An isolated bacterial colony was then inoculated into tryptic soy broth (TSB) and incubated for 16 h, without shaking, at 37°C under normal atmospheric conditions for use in experiments the following day.

Preparation of HEp-2 cell monolayers. HEp-2 cells (CCL-23) were obtained from the American Type Culture Collection (Rockville, Md.). The cells were grown at 37°C in a humidified atmosphere of 5% CO₂-95% air in minimal essential medium supplemented with 10% fetal bovine serum (FBS). Subconfluent monolayers were prepared for the binding assays by disrupting HEp-2 monolayers with a solution of 0.25% (vol/vol) tissue culture-grade trypsin in FC buffer (0.14 M NaCl, 5.0 M mM KCl, 20.0 mM Tris-HCl, 5.0 mM Tris base, 0.5 mM EDTA [pH 7.2]). After the trypsinized HEp-2 cells were suspended in fresh tissue culture medium, approximately 7.5 × 10³ cells in 150 µl of culture medium were added to individual wells, with each well containing a 6-mm-diameter removable polystyrene disk (Biomedical Workshop, University of Alberta, Edmonton, Alberta, Canada) covering its bottom, of 96-well tissue culture plates. The plates were then incubated in a CO₂ incubator until the next day.

EPEC cell binding assay. For all experiments, EPEC E2348/69 was cultured in Dulbecco modified Eagle medium (DMEM) (catalog number 23800; Gibco, Burlington, Ontario, Canada) supplemented with 44 mM NaHCO₃, 40 µM phenol red, and 25 mM glucose, which was pre-equilibrated overnight in a CO₂ incubator (42). FBS was not included in this culture medium. Prior to each experiment, 40 µl of the TSB-grown bacteria were inoculated into 4 ml of DMEM in borosilicate glass culture tubes (15 by 75 mm), which were then incubated for 1 h in a CO₂ incubator to induce the expression of EPEC virulence factors (32, 42, 44). To determine the optimum concentration of glycoconjugates for the binding inhibition experiments, 32 µl of DMEM bacterial culture (ca. 3 × 10⁶ to 5 × 10⁶ CFU) and 52 µl of DMEM (pre-equilibrated) were added to empty wells of a 96-well microtiter plate. Next, 16 µl of BSA-glycoconjugate solution (undiluted [5 mg/ml] or diluted 1:2 or 1:4 in PBS) was added to the wells to obtain final inhibitor concentrations of 0.8, 0.4, or 0.2 mg/ml, respectively. For all subsequent experiments, the volumes were adjusted such that 32 µl of DMEM bacterial culture, 56 µl of DMEM, and 12 µl of BSA-glycoconjugate (final concentration, 0.6 mg/ml) were added to wells of a 96-well plate. The plate was then incubated at 37°C in a CO₂ incubator for 30 min, after which time the entire contents from each of the wells were transferred to wells containing subconfluent HEp-2 cell monolayers from which the culture medium was first removed. After this microtiter plate was incubated for an additional 30 min at 37°C in the CO₂ incubator, the cells were washed three times with PBS, fixed with methanol for 10 min, and stained with Giemsa stain for 20 min. The polystyrene disks were removed from the microtiter plate wells, and EPEC adherence was monitored microscopically by using a ×100 objective lens. A total of 150 to 200 randomly chosen HEp-2 cells were examined, and those having attached microcolonies consisting of five or more bacteria were considered positive for LA EPEC (44).

Effect of glycoconjugate preincubation period on bacterial attachment. EPEC E2348/69 was grown in DMEM for 1 h in a CO₂ incubator as described above. Next, 32 µl of bacterial culture and 56 µl of DMEM were added to empty wells of a 96-well tissue culture plate. At this time, 12 µl of BSA-glycoconjugate (final concentration, 0.6 mg/ml per well) was added to sample wells for which the effect of preincubation with the BSA-glycoconjugate(s) was to be determined. After the plate was incubated in a CO₂ incubator for 30 min, 12 µl of BSA-glycoconjugate was added to the remaining, nonpreincubated sample wells. The contents of all wells were then immediately transferred to wells of a 96-well microtiter plate containing subconfluent HEp-2 cell monolayers grown on polystyrene disks, and the plate was returned to a CO₂, 37°C incubator for 30 min. The cells were then washed three times with PBS, fixed with methanol, and stained with Giemsa stain. The percentage of HEp-2 cells with LA EPEC was determined as described above.

Determination of BfpA and intimin expression. After EPEC E2348/69 was incubated in DMEM for 1 h to induce the expression of virulence factors, 96 µl of bacterial culture, 168 µl of DMEM, and 36 µl of BSA-glycoconjugate (final concentration, 0.6 mg/ml per well) were added to empty wells of a 48-well tissue culture plate. The plate was incubated in a CO₂ incubator for 1 h before the contents of each well were transferred to microcentrifuge tubes. The bacteria were washed once with PBS by centrifugation, and the resulting pellet was lysed in sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer containing 50 mM dithiothreitol (24). Bacterial proteins were separated by SDS-PAGE (12.5% polyacrylamide) and electrophoretically transferred to an Immobilon-P membrane (Millipore, Bedford, Mass.).

Statistical analysis. The significance of any differences in EPEC LA or in the levels of BfpA and intimin was determined by using the nonparametric Mann-Whitney U test.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Inhibition of EPEC LA by the BSA-glycoconjugates. Preliminary experiments were performed to determine the concentration of glycoconjugate(s) to be used in experiments. The results of these dose-dependent binding inhibition experiments demonstrated that LacNAc-BSA was the most effective inhibitor of EPEC E2348/69 attachment to HEp-2 cells (Fig. 1). This inhibition was concentration dependent over the range examined, i.e., 0.8, 0.4, or 0.2 mg of LacNAc-BSA per ml, resulting in approximately 87, 65, and 40% reductions, respectively, in EPEC LA to HEp-2 cells.

View larger version (18K): [in this window] [in a new window]   FIG. 1.   Optimization of BSA-glycoconjugate(s) concentration for experiments. DMEM-grown EPEC E2348/69 were preincubated with various concentrations of different BSA-glycoconjugate(s) and then added to tissue culture wells containing subconfluent monolayers of HEp-2 cells. After bacterial binding was allowed to occur, the cells were washed, fixed with methanol, and stained with Giemsa stain. HEp-2 cells with adherent microcolonies consisting of five or more bacteria were considered positive for LA EPEC. Two independent trials were performed, trial 1 () and trial 2 (), with each point representing a single determination for each sample. Bars indicate the average(s) of the data from both trials.

View larger version (22K): [in this window] [in a new window]   FIG. 2.   Inhibition of EPEC E2348/69 binding to HEp-2 cells with BSA-glycoconjugates at a final concentration of 0.6 mg/ml per well. Experiments were performed essentially as described in the legend to Fig. 1 except that bacteria were preincubated with BSA-glycoconjugates present at final concentrations of 0.6 mg/ml per well. Data points for trial 1 (), trial 2 (), and trial 3 () are indicated. Error bars for each of these points represent the range in values obtained for duplicate samples. The overlaying bar graph indicates the mean(s) of the data obtained in all three trials.

Effect of preincubation on EPEC attachment. The procedure used in the glycoconjugate inhibition binding experiments involved preincubating EPEC E2348/69 with various glycoconjugates for 30 min prior to adding the mixtures to the HEp-2 cell monolayers. This preincubation step was performed on the assumption that the multivalent BSA-glycoconjugate inhibitors required a finite amount of time to occupy bacterial adhesin receptor binding sites prior to exposing the EPEC-glycoconjugate mixtures to the HEp-2 cells. This assumption would predict that the BSA-glycoconjugate inhibitors might be less effective if this preincubation step was omitted. To test this prediction, experiments were performed in which the bacteria were either preincubated for 30 min with LacNAc-BSA before being added to the HEp-2 cells or were added simultaneously with LacNAc-BSA to the monolayers. As shown in Fig. 3, eliminating the preincubation step resulted in an approximately twofold decrease in the inhibitory activity of LacNAc-BSA, although this difference was not statistically significant (P = 0.121). Without preincubation, EPEC LA to HEp-2 cells was reduced by 25% when bacteria were incubated with LacNAc-BSA compared to BSA alone.

View larger version (21K): [in this window] [in a new window]   FIG. 3.   Effect of preincubation period on EPEC E2348/69 attachment to HEp-2 cells. The glycoconjugates (final concentration, 0.6 mg/ml per well) were either incubated with the bacteria for 30 min prior to adding the bacteria to the HEp-2 cell monolayers or added when the bacteria were added to the monolayers. The cells were then washed, fixed with methanol, and stained with Giemsa stain. HEp-2 cells considered positive for LA EPEC were determined as described earlier. The results shown are representative of those obtained in three independent experiments. The error bars represent the range in values for duplicate samples.

Effect of BSA-glycoconjugates on the expression of EPEC BfpA and intimin. The observation that the effectiveness of LacNAc-BSA as an inhibitor of EPEC E2348/69 binding to HEp-2 cells depended on the time of the preincubation step was consistent with the assumption that this inhibitor required a finite amount of time to occupy adhesin receptor binding domains. It was also possible, however, that during this preincubation period the LacNAc-BSA caused an alteration in the expression of proteins involved in EPEC adherence. To investigate this possibility, we measured the relative expression of BfpA and intimin of EPEC preincubated with different glycoconjugates in parallel with binding inhibition experiments. The results of these experiments demonstrated that the expression of BfpA and intimin was reduced when the bacteria were incubated with specific glycoconjugates (Fig. 4). Overall, BfpA and intimin levels of EPEC E2348/69 were decreased to the greatest extent by LacNAc-BSA and LeX-BSA, respectively (Fig. 5). BfpA or intimin expression by EPEC incubated with either LacNAc-BSA or LeX-BSA was significantly less than that of bacteria incubated with BSA (P = 0.037). LacNAc-BSA reduced BfpA expression significantly more than LeX-BSA (P = 0.050), whereas intimin expression was significantly reduced by LeX-BSA compared to LacNAc-BSA (P = 0.050). The effects of LeY-BSA were not included in the statistical analysis since this compound was not included in trial 1 of these experiments.

View larger version (37K): [in this window] [in a new window]   FIG. 4.   Effect of BSA-glycoconjugates on the expression of EPEC proteins associated with attachment. After EPEC E2348/69 cultured in DMEM was incubated with the BSA-glycoconjugate(s) (0.6 mg/ml, final concentration), the bacteria were washed and then lysed in sample buffer. Proteins were separated by SDS-PAGE (12.5% polyacrylamide) and transferred electrophoretically to Immobilon-P membrane. Bacterial proteins were detected with anti-intimin (A), anti-MBP (B), or anti-BfpA (C) antibodies, followed by anti-rabbit peroxidase-conjugate antibodies and analysis with the ECL detection system. The mobilities of prestained molecular size standards are indicated on the left, while proteins detected with the antibodies are indicated by arrows on the right.

View larger version (15K): [in this window] [in a new window]   FIG. 5.   Relationship between EPEC binding and BfpA or intimin levels. For each trial, the experiments investigating the effects of various glycoconjugates on bacterial binding and levels of BfpA or intimin were performed in parallel. The assay for EPEC LA to HEp-2 cells was performed as described in the legend to Fig. 2. Intimin, MBP, and BfpA were detected as described in the legend to Fig. 4. After each of the bands was analyzed by densitometry, the levels of BfpA and intimin were normalized to the level of MBP in each sample. The percentage (y axis) for each figure refers to the levels of LA, BfpA, or intimin of EPEC incubated with specific glycoconjugate(s) relative to that of bacteria cultured in the presence of BSA. Each data point represents a single determination for trial 1 (), trial 2 (×) and trial 3 (). The overlaying bar graph indicates the mean(s) of the data collected from all three trials.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Using CHO cell Lec mutants, we previously demonstrated that the expression of asialo-lactosamine sequences was sufficient for EPEC attachment to these cells (43). In the present study, we used synthetic, multivalent BSA-glycoconjugates presenting either lactosamine (LacNAc) or fucosyllactosamine structures (LeX or LeY) to determine whether fucosylated glycan sequences also had a role in these interactions (5). EPEC E2348/69 was selected for these experiments since this strain is commonly used by other research groups to investigate EPEC pathogenesis (8, 9, 26, 40). Overall, we found that LacNAc-BSA was the most effective inhibitor of bacterial binding to HEp-2 cells, followed by LeX-BSA. LeY-BSA was the least effective inhibitor in our binding assays. These results confirmed our previous findings regarding the involvement of lactosamine sequences in EPEC binding to CHO cells. Additional experiments performed to elucidate the manner in which these glycan structures inhibited binding led to the observation that incubation of EPEC E2348/69 with LacNAc-BSA and with LeX-BSA alone caused decreases in the expression of BfpA and intimin, respectively. These results are significant because they revealed that the expression of BfpA and intimin, two different proteins involved in EPEC attachment, were specifically affected by different glycoconjugates.

In previous experiments where carbohydrates were shown to inhibit bacterial attachment to epithelial cells, it was presumed that these structures were eukaryotic cell receptor analogs which acted as competitive antagonists of bacterial binding (1, 12). However, the results of our experiments, which indicated that the inhibition of EPEC E2348/69 binding by glycoconjugates correlated with a decrease in the expression of BfpA and intimin, suggested that the process of inhibition may be more complex. Recently, Knutton et al. demonstrated that EPEC downregulated intimin expression in the later stage(s) of attachment to HEp-2 cells (22). A similar decrease in BFP levels was not observed during the same time period. Our results are similar to those of Knutton et al. in that we also found that EPEC expressed lower levels of a protein(s) associated with attachment in response to environmental signals. As well, our results and those of Knutton et al. suggest that intimin and BFP, to a degree, might be differentially regulated. Unlike the results of Knutton et al., however, we found that specific glycoconjugates were able to mediate this effect independently of viable epithelial cells and affected BFP expression as well.

As of yet, we do not know the mechanism by which the glycoconjugates mediate their effect on BfpA and intimin. In consideration of our results and those of Knutton et al., an attractive model is that, in addition to the glycoconjugates possibly acting as competitive inhibitors of binding, their interaction with EPEC may stimulate bacteria to progress to later stages of the multistep colonization process. Ultimately, this progression to a later stage is accompanied by a loss of structures required for attachment, and this reduces the ability of the bacteria to bind.

An implication of the above model is that the glycoconjugates are analogs of epithelial cell receptors for EPEC. In our experiments, we used HEp-2 cells which are derived from laryngeal tissue. While results of investigations into the formation of A/E lesions with this cell line appear to parallel in vivo findings (23), recent results of Hicks et al. suggest that the initial attachment of EPEC to HEp-2 cells differs from that observed with intestinal cells (15). Specifically, Hicks et al. reported that EPEC which expressed BFP, but not intimin, was able to attach to HEp-2 cells but not to intestinal biopsy tissue. Whether the LacNAc and LeX sequences used in our experiments mimic actual intestinal cell receptors in vivo remains to be determined. However, an important aspect of our results with the soluble BSA-glycoconjugates is that it may not be necessary for these carbohydrate structures to represent the natural host cell receptors for EPEC. Although speculative, it is possible that the downregulation of BfpA and intimin caused by soluble glycoconjugates might, alone, reduce the ability of EPEC to bind to their relevant receptors in vivo, and this could form the basis for a novel therapeutic intervention in this gastrointestinal disease.

Another area for further investigation which arises from our results relates to the bacterial structure(s) the glycoconjugate(s) interact with to produce their effects. Since BFP promote the LA phenotype (9, 14, 15, 38-40), it is possible that the glycoconjugate(s) used in our experiments reduced EPEC binding to HEp-2 cells by inhibiting BFP-mediated interactions. This reduction could result from a decrease in the formation of microcolonies resulting from lower BfpA levels, as well as by blocking EPEC attachment to the surface of HEp-2 cells. It should be noted that, if this latter explanation represents their mode of action, the glycoconjugate(s) would not, as proposed by Hicks et al., be expected to prevent EPEC binding to intestinal tissue. However, in this situation, the total number of bacteria bound to these cells might still be reduced as a consequence of reduced microcolony formation. Overall, whether the effects observed in our experiments result from interaction with BFP or an as-yet-unidentified adhesin(s) remains to be resolved.

The inhibition results presented here differ from those reported previously by Cravioto et al. (5). In that study, fucosylated pentasaccharides and difucosyllactose isolated from human milk were more effective inhibitors of EPEC binding to epithelial cells than their nonfucosylated backbone structures. Significantly, in those studies, fucosyllactose trisaccharides and lactose were not inhibitory. In contrast to the results of Cravioto et al., we found that the acetamido forms of LacNAc, or LeX, conjugated to BSA were effective inhibitors. At present, the reasons for the differences between our results and those of Cravioto et al. are not known. Factors which could contribute to these differences include the number of times an organism was passaged in the laboratory (37) or the growth phase of the bacteria (2). Alternatively, these differences could be due to the nature and presentation of the carbohydrate structures used to inhibit binding. In this respect, it is possible that the different conformations of the glycoconjugates (46) or the multivalency of these structures which, because of the multipoint interactions between microbes and eukaryotic cells in general, have been shown to be more effective inhibitors (2, 20, 37) are responsible for these differences. Our laboratory is currently performing experiments with different EPEC serotypes, as well as with different fucosylated lactosyl structures as inhibitors, in order to investigate the roles of these factors in EPEC binding specificity. Regardless of the specific carbohydrate structures involved, our results provide a direction for further investigation of the regulation of EPEC attachment factors and investigation of whether similar processes are invoked upon bacterial binding to epithelial cells.

Despite our lack of understanding of the specific mechanisms by which EPEC binding to HEp-2 cells is inhibited, our results demonstrate the potential of these or similar compounds as therapeutic agents for treating EPEC infection. Our optimism is supported by recent work of other groups which reported that the oral administration of glycans reduced gastrointestinal loads of pathogenic bacteria in vivo (29, 37). In the past, it has been shown that EPEC mutants for BFP or intimin are still able to cause diarrhea in volunteers, albeit at reduced levels (11, 27). By targeting more than one virulence factor, hopefully the progression of EPEC infections can be prevented.