INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Exoenzyme S, an ADP-ribosyltransferase produced by Pseudomonas aeruginosa, contributes to the virulence of this opportunistic pathogen by causing increased tissue damage and bacterial dissemination (17, 21, 25, 37). ExoS is translocated into eukaryotic cells by the bacterially mediated type III secretory process (38). Studies of the effects of ExoS on cell function have therefore required the development of model systems which enable the bacterial translocation of ExoS yet allow the effects of ExoS to be differentiated from those of other Pseudomonas factors. The use of such model systems has linked ExoS production to complex effects on eukaryotic cell function, including inhibition of DNA synthesis, alterations in actin cytoskeletal structure, and interference with cell-matrix adherence (12, 28, 29). Alterations in multiple eukaryotic cell processes are therefore likely to contribute to the increased tissue damage associated with ExoS.

The multiple and diverse effects of ExoS on cell function can, to some extent, be explained by its multidomain structure. Residues within the amino-terminal half are required for export and the aggregation properties of ExoS, while the ADP-ribosyltransferase (ADPRT) activity of ExoS is included within the carboxy-terminal 222 amino acids of the protein (18, 38). Consistent with the multifunctional structure of ExoS, effects on cell rounding have been associated with an enzymatically inactive E381A mutant form of ExoS (12). Diversity in the cellular effects of ExoS may also relate to Ras functioning as an in vivo substrate of ExoS (24) and the ability of Ras to modulate multiple signal transduction pathways. In this regard, recent studies have linked the ADP-ribosylation of Ras by ExoS to the disruption of Ras-Raf-mediated signal transduction pathways (14, 36), providing a cellular mechanism to explain inhibitory effects of ExoS on DNA synthesis and PC12 cell differentiation (13, 29). A current understanding of cell signaling pathways also links Ras activation to the regulation of actin cytoskeletal structure, via Rac, Cdc42, and Rho (15).

Our laboratory has used a bacterial-eukaryotic coculture model system that compares the effects of ExoS-producing strain 388 and the isogenic non-ExoS-producing strain 388exoS (388S) to differentiate the effects of ExoS on eukaryotic cell function. During these studies, it became evident that ExoS could exert similar effects on DNA synthesis and morphology of different cell types (29). It was also observed that different cell lines exhibited different levels of sensitivity to ExoS. Studies described in this report further investigate the cytotoxic effects of ExoS by comparing the effects of ExoS on different human epithelial lines and examining processes underlying differential sensitivities to ExoS. Intercellular association as well as intrinsic cellular properties were found to contribute to the resistance of normal kidney (NK) epithelial cells to bacterially translocated ExoS. Conversely, sensitivity to ExoS appeared to be linked to epithelial cell growth, with rapidly growing tumor cell lines and growing subconfluent NK cells showing increased sensitivity to the effects of ExoS. The results are consistent with factors expressed by growing epithelial cells being required for the bacterial contact-dependent translocation of ExoS and the possibility that these factors become down-regulated or redistributed and less accessible as epithelial cells differentiate into confluent polarized monolayers.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and culture conditions. P. aeruginosa 388, 388S, 388exs1::Tn1, and 388popD::Tc* were all kindly provided by Dara Frank, Medical College of Wisconsin, Milwaukee, Wis. Parental strain 388 and construction of the 388S mutant have been previously described (3, 22). The 388S mutant strain fails to produce 49-kDa ExoS due to allelic exchange of the majority of the structural gene with a tetracycline gene cartridge. ExoS complementation studies of strain 388S confirmed that differences in the effects of this strain and parental strain 388 on eukaryotic cell function directly relate to ExoS production (28). Strains 388popD::Tc* (35) and 388exs1::Tn1 (38), discussed in scanning electron microscopy (SEM) studies, lack production of the type III PopD translocation and PscC secretion proteins, respectively.

Eukaryotic cell culture. Cell lines and their specific growth media are described in Table 1. All cells were maintained at 37°C in 5% CO₂-95% air, and all media and media components were purchased from Gibco-BRL (Gaithersburg, Md.). Cell lines obtained from the American Type Culture Collection (ATCC) were cultured and passaged as specified by ATCC. Primary cultures of human NK proximal tubule (NK77 and NK97612) cells were obtained from the tissue culture bank in the Department of Pathology at the Medical University of South Carolina, Charleston. NK cells were maintained as previously described (7) and cultured in vessels coated first with bovine collagen type 1 (Collaborative Biomedical, Lexington, Mass.) and then with fetal bovine serum (FBS). Culture growth medium was replaced every 2 to 3 days, and cells were passaged weekly at a subculture ratio of 1:2 or 1:3. All experiments with NK cells were performed from passages 4 to 9. CFT1 and CFT1-LCFSN cell lines were a kind gift from Gerald Pier, Harvard Medical School, Boston, Mass., and were cultured as previously described (27, 39). Medium was replaced every 2 to 3 days, and cells were passaged weekly as described for NK cells. For comparisons of NK and HT-29 cell induction of the ExoS regulon, HT-29 cells were adapted to the 50% high-glucose Dulbecco's modified Eagle's medium (DMEM)-50% Ham's F-12 (DME/F12) used to culture NK cells. For bacterial-eukaryotic cell coculture experiments, all except NK cell monolayers were detached with 0.25% trypsin-1 mM EDTA (trypsin-EDTA; Gibco-BRL), resuspended in their respective culture medium, counted, diluted to the appropriate density, and seeded in 25-cm² flasks, 35-mm-diameter dishes, or 48- or 6-well plates (Costar). NK cells were similarly detached with trypsin-EDTA, but following detachment, trypsin was neutralized with an equal volume of FBS, and cells were collected, centrifuged for 5 min at 600 × g, resuspended in growth medium, and seeded as described above on collagen-coated surfaces. All cells were allowed to grow for 2 to 3 days prior to addition of bacteria.

Examination of bacterial effects on HT-29 and NK cells. In all experiments, control and strain 388 or strain 388S cells were identically treated except that in controls no bacteria were added to the coculture medium, and in bacterially treated monolayers the indicated concentration of the respective strain was added to the coculture medium. The multiplicity of infection ranged from 10 to 100 bacteria per eukaryotic cell, depending on eukaryotic cell density.

(i) Quantification of DNA synthesis in eukaryotic cells following exposure to bacteria. Following a 3- or 4-h coculture period, bacteria were removed and monolayers were washed in medium containing gentamicin (200 µg/ml) and ciprofloxacin (100 µg/ml) (antibiotic medium) to limit further bacterial growth. Cells were then analyzed for [³H]thymidine incorporation after a 20-h pulse in antibiotic medium as previously described (29).

(ii) Immunoprecipitation, detection, and quantification of Ras proteins. Ras was immunoprecipitated from cells after coculture with bacteria, using monoclonal Y13-259 Ras antibody (ATCC), rabbit anti-rat immunoglobulin G (Sigma Chemical Co., St. Louis, Mo.), and protein G-Sepharose (Sigma), as previously described (24). Proteins were resolved on sodium dodecyl sulfate (SDS)-15% polyacrylamide gels and electroblotted, and Ras immunoblots were developed by using pan-Ras-OP22 antibody (Oncogene Research, Cambridge, Mass.) followed by peroxidase-conjugated goat anti-mouse antibody (Sigma) and visualized by enhanced chemiluminescence (Amersham Life Sciences, Arlington Heights, Ill.). Ras modification was quantified by densitometric analysis using NIH Image version 1.60 software. The areas of intensity of modified and unmodified Ras were determined, and modified Ras is expressed as a percentage of total Ras.

(iii) Examination of cell morphology. Cells were grown and cocultured with bacteria in 48-well plates. Cell morphology was assessed by phase-contrast microscopy following a 3-h coculture period, either immediately after removal of bacteria at 3 h or after the addition of antibiotic medium and further incubation for 24 h.

Quantification of bacteria associated with eukaryotic cells. Bacterial association was determined by using an adaptation of the procedure of Fleiszig et al. (9). NK and HT-29 cells were seeded at 6 × 10⁵ cells/ml (0.5 ml/well) in 48-well plates and grown until confluent. Bacteria were prepared for coculture as previously described (29) and suspended at 10⁸ CFU/ml in coculture medium; then 0.5 ml was added to each tissue culture well, and the plates were incubated for 3 h. Bacteria were removed and monolayers were washed three times with phosphate-buffered saline (PBS) to remove unassociated bacteria. Cells were detached by scraping and lysed by adding to each well 1 ml of PBS containing 0.25% Triton X-100. Lysates were thoroughly mixed; then aliquots were diluted in PBS containing 1% bovine serum albumin and plated to quantify viable bacteria. Bacterial numbers in the original inoculum were determined, and bacteria associated with cells were reported as percentage of inoculum.

Assay of ExoS activity in coculture medium. ExoS ADPRT activity was quantified by using the in vitro substrate soybean trypsin inhibitor as previously described (29) except that reactions were modified to detect lower levels of activity. For this, the final concentration of [¹⁴C]NAD was raised to 5 µM, the 14-3-3 protein cofactor (FAS) concentration was adjusted to 100 nM, and 20 µl of culture medium was added to reactions, which were allowed to proceed for 1 h. Incorporation of radiolabel was measured by liquid scintillation counting and reported as picomoles of ADP-ribose transferred to substrate per milliliter of coculture medium.

Disruption of intercellular junctions in confluent monolayer cell cultures. NK and HT-29 cells grown to confluence in 6- or 48-well plates were washed once with PBS free of calcium and then incubated with PBS containing 1 mM EDTA, or as a control PBS supplemented with 0.9 mM calcium, until all EDTA-treated cells were rounded (10 to 15 min). PBS was then removed, cells were cocultured with bacteria for 3 h, and [³H]thymidine uptake assays or Ras immunoprecipitation was performed as described above.

SEM. P. aeruginosa 388 (10⁸ CFU/ml) was cultured for 3 h with subconfluent HT-29, LNCaP, or NK cell monolayers or confluent NK cell monolayers which were grown on 12-mm-diameter glass coverslips. Bacteria were removed, and cells were washed twice with McCoy's SA-bovine serum albumin fixed in situ in 2% gluteraldehyde in 0.1 M cacodylate buffer (pH 7.4), rinsed, postfixed in 1% osmium in 0.1 M cacodylate buffer, and then dehydrated in a graded series of ethanol, and treated with hexamethyldisilane. After drying, coverslips were coated with gold and examined in a JEOL JSEM-LV5410 scanning electron microscope.

Statistical analysis. Statistical analysis was performed with StatView (Abacus Concepts, Inc.).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Effects of ExoS-producing bacteria on DNA synthesis and Ras modification in different epithelial cell lines. Differences have been observed in the sensitivity of eukaryotic cell lines to the effects of ExoS during coculture studies. To investigate cellular properties that might influence sensitivity to bacterially translocated ExoS, we compared levels of inhibition of DNA synthesis caused by ExoS-producing strain 388 or the non-ExoS-producing strain 388S in a selection of human cell lines described in Table 1. As shown in Fig. 1A, NK epithelial cells were most resistant to the effects of ExoS on DNA synthesis. CFT1-derived tracheal epithelial cells, which express mutant cystic fibrosis transductance regulator (CFTR), appeared slightly more sensitive to the effects of ExoS on DNA synthesis than NK cells. However, LCFSN cells, which express both mutant and normal CFTR, showed a sensitivity to ExoS similar to that of CFT1 cells, suggesting that expression of the mutant form of CFTR associated with cystic fibrosis did not alter sensitivity to ExoS. Tumor-derived cell lines consistently showed a greater sensitivity to ExoS production. As indicated in Fig. 1, and as previously described (36), sensitivity to ExoS did not appear to be affected by the expression of oncogenic forms of Ras by tumor cell lines.

View larger version (59K): [in this window] [in a new window]   FIG. 1.   Differential sensitivity of human cell lines to ExoS-producing bacteria. (A) Inhibition of DNA synthesis. The indicated cell lines, described in Table 1, were cocultured with 107 CFU of strain 388 or 388S (S) per ml or no bacteria for 3 or 4 h in their respective media. Bacteria were removed and replaced with medium containing 1 µCi of [3H]thymidine per ml and antibiotics to inhibit further bacterial growth. DNA synthesis was assayed after 20 h and is expressed as percent incorporation relative to non-bacterially treated control cells. The mean and SD of assays performed in triplicate are represented. (B) Ras modification. Cells were cocultured as described above, culture supernatants were removed, cells were lysed, and Ras was immunoprecipitated with Ras monoclonal Y13-259 antibody, rabbit anti-rat immunoglobulin G, and protein G-Sepharose. Immunoprecipitates were resolved by SDS-15% polyacrylamide gel electrophoresis. Ras immunoblots were developed using pan-Ras-OP22 antibody and detected by enhanced chemiluminescence. Ras (M) and Ras (U) indicates modified and unmodified Ras, respectively.

Relationship between effects of ExoS on DNA synthesis and cellular morphology. In addition to effects on DNA synthesis and Ras modification, the translocation of ExoS has been associated with cell rounding and the disruption of the actin cytoskeleton (12, 28, 29). To further understand differences in the sensitivities of epithelial cells to ExoS, cells resistant and sensitive to the effects of ExoS on DNA synthesis were compared for ExoS-associated alterations in cellular morphology. Two cell lines were selected for these studies: NK cells, representing an ExoS-resistant cell line, and HT-29 cells, representing an ExoS-sensitive cell line. HT-29 cells were chosen for these studies because their morphology and growth characteristics more closely resembled those of NK cells than other tumor-derived cell lines.

View larger version (53K): [in this window] [in a new window]   FIG. 2.   Differential effect of ExoS-producing bacteria on confluent NK and HT-29 cell morphology, DNA synthesis, and Ras modification. Confluent and subconfluent monolayers of NK and HT-29 cells were cocultured with 108 CFU of strain 388, strain 388S (S) per ml or no bacteria (0) for 3 h. Bacteria were removed, antibiotic-containing medium was added, and cells were analyzed for alterations in morphology by phase-contrast microscopy at 3 h (A) and differences in DNA synthesis and Ras modification (B), assayed as described for Fig. 1. DNA synthesis results represent the mean and SD of assays performed in triplicate. Modified and unmodified Ras [Ras(M) and Ras(U)] are labeled.

Role of bacterial contact in differential sensitivities of HT-29 and NK to the effects of ExoS. The observation that NK cells remained more resistant to ExoS than HT-29 cells when both cell types were confluent suggests that intrinsic cellular differences may contribute to their differential sensitivity to ExoS. Expression and translocation of ExoS is triggered via a poorly understood process which is thought to be initiated in vivo by the direct contact of the bacterium with a target cell but can be mimicked in vitro by Ca²⁺ ion depletion (11, 33). To determine whether differences in the sensitivity of cell lines to effects of ExoS might relate to cell surface properties that affect bacterial association or induction of the ExoS regulon, both parameters were compared in HT-29 and NK cells. When the association of bacteria with HT-29 or NK cells was examined in bacterial plating assays, no significant difference was detected in the percentage of inoculated bacteria associated with confluent HT-29 or NK monolayers following a 3-h exposure to strain 388 (0.13% ± 0.08% and 0.09 ± 0.06% [mean ± standard deviation {SD}] for HT-29 and NK cells, respectively; P = 0.77). However, since NK cells occupy a larger area and produce a more uniform monolayer than the rapidly proliferating HT-29 cells, fewer NK cells were present at confluency, which makes the efficiency of bacterial association with resistant NK cells actually greater than that of HT-29 cells (5:1 and 2:1 bacterium-to-cell ratio for NK and HT-29 cells, respectively). Association of 388S to both cell types was similar to that of strain 388. While these studies indicate that overall differences in bacterial association do not appear to contribute to the increased resistance of NK cells to the effects of ExoS, it cannot be determined from these studies whether the two cell lines differ in the manner of bacterial binding and/or cell surface localization.

View larger version (46K): [in this window] [in a new window]   FIG. 3.   Cell contact-dependent secretion of ExoS during coculture with HT-29 or NK cells. Strain 388 was grown in non-ExoS induction medium (TSB) or ExoS induction medium (TSBD-N) for 12 h, and then bacteria (108 CFU/ml) were cocultured with HT-29 or NK cells for 4 h. Coculture supernatants were assayed for ExoS ADPRT activity in reaction mixtures containing 100 µM soybean trypsin inhibitor 100 nM FAS, and 5 µM [14C]NAD in 0.2 M sodium acetate (pH 6.0). Activity is expressed as picomoles of ADP-ribose transferred per milliliter of coculture medium, and the mean and SD of analyses performed in triplicate are represented.

Role of cell junction formation in the resistance of cells to ExoS-producing bacteria. Susceptibility to P. aeruginosa infection has previously been reported to be influenced by epithelial cell polarity, with the basolateral surface of the cell being more susceptible to P. aeruginosa cytotoxicity than the apical surface (8). Confluent cultures of NK cells can mimic intact epithelial tissue by forming junctional complexes between cells, which can result in cell polarization and the separation of apical and basolateral cell surfaces (4). While HT-29 cells have also been reported to form tight junctions and become polarized under specific growth conditions (30, 34), culture conditions used in this study did not favor HT-29 cell polarization, and cells remained relatively undifferentiated and formed multilayers.

View larger version (50K): [in this window] [in a new window]   FIG. 4.   Effects of EDTA treatment on the sensitivity of NK and HT-29 cells to ExoS. NK or HT-29 cells were cultured to confluence and treated with 1 mM EDTA in PBS for ~10 min, until cells became rounded (+), or with PBS containing 0.9 mM CaCl2 (). Cells were then cocultured with 108 CFU of strain 388 or 388S (S) per ml or no bacteria (0) for 3 h. (A) NK cells were examined by phase-contrast microscopy for alterations in morphology at 3 and 24 h. (B) NK and HT-29 cells were assayed for DNA synthesis as described for Fig. 1. Results are expressed as percent inhibition of DNA synthesis of 388-treated cells relative to 388S-treated cells, with the mean and SD of assays performed in triplicate or quadruplicate represented. (C) The efficiency of Ras modification was determined by immunoblot analysis as described for Fig. 1 and quantified by densitometric analysis using NIH Image version 1.60 software. The areas of intensity of modified [(M)] and unmodified [(U)] Ras were calculated, and modified Ras is expressed as a percentage of total Ras. 0 represents Ras immunoprecipitated from non-bacterially treated cells.

Factors contributing to the sensitivity of epithelial cells to the effects of ExoS-producing bacteria. While intercellular associations were found to contribute to the resistance of NK cells to the effects of ExoS, the mechanism for the general increased sensitivity of HT-29 and other tumor cells to bacterially translocated ExoS remained unclear. A factor that differentiates HT-29 and other tumor cell lines from resistant confluent NK cells, and could potentially contribute to an increased sensitivity to ExoS, is an increased growth rate. Normal epithelial cell growth is dependent on and regulated by anchorage and cell association processes. Tumor-derived epithelial cell lines differ from normal epithelial cells in the ability to undergo anchorage-independent growth. Consistent with the possibility that factors associated with cell growth processes influence sensitivity to ExoS is the observed increased sensitivity to ExoS in (i) junctionally disrupted confluent NK cells which are in the process of reestablishing intercellular junctions and (ii) tumor cell lines, which can continue to grow in the absence contact-dependent growth signals.

View larger version (52K): [in this window] [in a new window]   FIG. 5.   Effects of ExoS on subconfluent NK and HT-29 cells. Subconfluent (sc) or confluent (c) NK or HT-29 cell monolayers were cocultured with 108 CFU of strain 388 or 388S (S) per ml or no bacteria (0) for 3 h. (A) Subconfluent NK and HT-29 cells were examined by phase-contrast microscopy for alterations in morphology at 24 h. (B) NK and HT-29 cells were assayed for DNA synthesis as described for Fig. 1. Results are expressed as percent inhibition of DNA synthesis of 388-treated cells relative to 388S-treated cells, with the mean and SD of assays performed in triplicate or quadruplicate represented. (C) The efficiency of Ras modification was determined by immunoblot analysis as described for Fig. 1 and quantified as described for Fig. 4. 0 represents Ras immunoprecipitated from non-bacterially treated cells.

View larger version (123K): [in this window] [in a new window]   FIG. 6.   Scanning electron micrographs of HT-29, LNCaP, and NK cells cocultured with ExoS-producing bacteria. Strain 388 (108 CFU/ml) was cocultured for 3 h with HT-29 or LNCaP cells or subconfluent or confluent NK cell monolayers grown on 12-mm-diameter glass coverslips. Cells were washed, fixed, then coated with gold, and examined in a JEOL JSEM-LV5410 scanning electron microscope. Representative images are shown, with the magnification of each image indicated. (A and B) HT-29 cell images show microvilli extending and then coming in closer contact with the bacterium. (C and D) LNCaP images show similar microvilli extensions to bacteria. (E and F) Subconfluent NK cell images show the basal extension of connections with the bacteria. (G and H) Confluent NK cell images show the general lack of extensions to the bacteria, with the exception of few connections between cells.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The requirement of bacterial contact for the translocation of ExoS has posed difficulties in defining ExoS-specific effects on eukaryotic cell function. The development of P. aeruginosa or Yersinia model systems, which allow the effects of ExoS to be differentiated from those of other bacterial factors, confirmed that ExoS is toxic to cultured cells and identified multiple effects on cell function (12, 28, 29). ExoS production was found to cause inhibition of DNA synthesis in both fibroblastic and epithelial cell lines (29), and studies of the HeLa epithelial cell line found ExoS to interrupt actin polymerization, resulting in cell rounding (12). Subsequent studies on the HT-29 epithelial cell line identified additional effects of ExoS on cell matrix adherence and microvillus effacement (28). While it became evident from these studies that bacterially translocated ExoS can exert similar effects on different cell types, the mechanism of the different ExoS cytotoxic effects, as well as an understanding of differential sensitivities of cell lines to the effects of ExoS, remained unknown.

When the effects of ExoS production by P. aeruginosa on different human epithelial cell lines were compared in bacterial-eukaryotic coculture studies, variations in sensitivities to ExoS were observed. Of the cell lines examined, confluent NK cells appeared most resistant to the effects of ExoS on DNA synthesis. These results are consistent with the opportunistic nature of P. aeruginosa infections and the general resistance of healthy epithelial barriers to P. aeruginosa infection. While the CFT1 cell line, which expresses the mutant CFTR, appeared more sensitive to ExoS-specific effects on DNA synthesis and cell morphology (unpublished observation) than confluent NK cells, this sensitivity was not altered by the expression of normal CFTR. This finding suggested that effects of ExoS on cell function were not enhanced by the expression of mutant CFTR protein. Tumor-derived cell lines were consistently highly sensitive to the effects of ExoS, with 10 of 10 cell lines of carcinoma, teratocarcinoma, or neuroblastoma origin examined to date showing a 50% or greater inhibition of DNA synthesis by ExoS. In all cell lines examined, effects of ExoS on DNA synthesis closely correlated with the efficiency of ADP-ribosylation of cellular Ras by ExoS.

When mechanisms of resistance to ExoS were examined in resistant NK epithelial cells and the sensitive HT-29 carcinoma cells, neither differences in the efficiency of general bacterial association nor the eukaryotic cell contact-mediated induction of the ExoS regulon appeared to account for the differences in sensitivity of the two cell lines to ExoS. In comparison, intercellular contact did contribute to the resistance of NK cells to ExoS, with EDTA-disrupted confluent NK monolayers showing increased sensitivity to the effects of ExoS on DNA synthesis and morphology. Consistent with the less efficient polarization and formation of intercellular junctions of confluent HT-29 cells, EDTA treatment did not alter the sensitivity of HT-29 cells to ExoS. These results are in agreement with previous studies of canine kidney epithelial (MDCK) cells which found differentiated MDCK cells to be resistant to cytotoxic effects of ExoS-producing P. aeruginosa (9). Collectively the data suggest that cell sensitivity to bacterially translocated ExoS requires the expression of specific eukaryotic cell factors which become altered as epithelial cells differentiate into confluent polarized monolayers. The direct association of sensitivity to ExoS with the efficiency of Ras modification also supports that intrinsic differences in NK and HT-29 cells to ExoS is, in some part, due to differences in the efficiency of internalization of enzymatically active ExoS.

The possibility that differences other than those related to polarization might exist between the two cell lines in their sensitivity to ExoS was also indicated in morphological analyses. Regardless of the degree of cell confluency, more severe cell rounding was detected in HT-29 cells exposed to strain 388 than in NK cells. The increased cell rounding observed in HT-29 cells suggests that their cytoskeletal structure is more sensitive to the effects of ExoS than that of NK cells. Transformed cells differ from normal epithelial cells in their ability to respond to nonadhesive growth factor signals, which are mediated by motile, non-actin-linked adhesion complexes (15, 23). Growth of normal epithelial cells, in comparison, maintains an anchorage-dependent growth factor requirement, mediated by stable, actin-linked adhesion junctions (2). One possible explanation for the differential sensitivity of HT-29 and NK to morphological alterations could therefore be that non-actin-linked adhesion complexes are more sensitive to the effects of ExoS than those that are actin linked. It also remains possible that more severe effects of ExoS on cell morphology are coordinated in some manner with short-term, non-ExoS-associated morphological alterations caused by strain 388S, which are apparent in coculture studies with HT-29 cells but not NK cells.

While cell confluency and intercellular junction formation contributed to the resistance of NK cells to the effects of ExoS, the molecular basis of the increased sensitivity of tumor cell lines to ExoS remains unexplained. Knowing that Ras is modified by ExoS in vivo, it was initially speculated that the increased sensitivity of tumor cells to ExoS related to their rapid proliferative rate and the corresponding increased effects of ExoS on proliferative signals mediated by activated Ras. Contrary to this premise, no biochemical or functional data which support a relationship between the activational state of Ras and the increased sensitivity to bacterially translocated ExoS have yet been obtained. GDP-bound, GTP-bound, and oncogenic Ras have been found to be modified by ExoS in an identical manner in vitro (1a, 36), and no differences have been observed in the modification of normal and oncogenic Ras in vivo (36). Also, as evident in Fig. 1, bacterially translocated ExoS showed no enhanced inhibition of DNA synthesis relative to the expression of normal or oncogenic Ras by tumor cell lines.

A possible clue as to cellular differences that might contribute to an increased sensitivity to ExoS was revealed from SEM. In examining the interaction between eukaryotic cells and P. aeruginosa, extensions were found to protrude from host cells and contact bacteria. These extensions were more numerous in subconfluent NK and tumor-derived HT-29 or LNCaP cell lines sensitive to ExoS compared to resistant confluent NK cells. In this regard, it is notable that while no quantitative differences in P. aeruginosa interaction with HT-29 or NK cells were detected in cell association assays, qualitative differences in the association of cells with bacteria became apparent by SEM. These results are consistent with studies of type III secretory processes of other bacteria which find cell surface receptors mediating type III translocation to differ from those involved in general bacterial adherence (5, 32). Studies are in progress to identify differences in cell receptor expression or adherence-mediated signaling events between ExoS-sensitive and -resistant cells to help understand the molecular basis of the enhanced sensitivity to ExoS.

Manipulation of eukaryotic cytoskeletal structures is becoming a common theme in bacterial mechanisms of pathogenesis. Among the best-characterized manipulations are the actin pedestal formation by enteropathogenic Escherichia coli (19), membrane ruffling by salmonellae and Shigella flexneri (1, 10), and the polymerized actin tail that propels Listeria monocytogenes to adjacent cells (20). While each bacterium appears to use a slightly different cytoskeletal manipulation strategy, the microvilli or filopodium-like structures attracted to P. aeruginosa most closely resemble the cellular structures associated with Legionella pneumophila (26). Although the induction of these cellular structures does not appear to be directly related to ExoS or the type III secretory system, their correlation with the effects of ExoS on cell function suggests that this structural interaction may be used by P. aeruginosa to facilitate the type III-mediated translocation of ExoS.

In conclusion, these studies provide the first direct comparison of the differential effects of bacterially translocated ExoS on human epithelial cell lines and identify cell culture conditions that can influence the cytotoxic effects of ExoS. Cellular properties affecting sensitivity to ExoS appear similar to those affecting sensitivity to P. aeruginosa, in general. Confluent normal epithelial monolayers appeared most resistant to the effects of ExoS, while subconfluent or disrupted monolayers showed increased sensitivity. Tumor cells and normal epithelial cells involved in cell growth were found to be most sensitive to the effects of ExoS, and this may relate to the ability of P. aeruginosa to utilize cellular growth processes in facilitating the translocation of ExoS.