INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Clostridium difficile is the etiologic agent of antibiotic-associated diarrhea and pseudomembranous colitis (1). Antibiotics and cytotoxic drugs disturb colonic flora, allowing overgrowth of C. difficile and production of toxins A and B. Toxin A elicits an acute inflammatory response, congestion, and necrosis when inoculated in the gut (1, 14, 36, 44). It is chemotactic and induces the release of inflammatory mediators by macrophages and neutrophils (9, 24, 31). Some studies suggested that mast cells also play an important role in the pathophysiology of toxin A (25). Thus, toxin A administered into ileal loops of rats elicited the release of inflammatory mediators such as leukotriene B4, platelet-activating factor, and rat mucosal mast cell protease II (RMCPII) (6, 26, 35). Moreover, treatment of animals with the antiallergy and antiinflammatory agent ketotifen, with the H1 histamine antagonist iodoxamide, or with histaminase reduced the inflammation and secretory responses caused by toxin A (12, 25, 34). It has been proposed that toxin A induces the secretion of inflammatory mediators from mast cells either directly, or indirectly through the release of substance P, a known activator of mucosal mast cells (7, 19, 29).

Mast cells are widely distributed in the intestinal mucosa, in skin and around blood and lymphatic vessels, and in many other tissues and organs. They can be activated to release inflammatory mediators via immunoglobulin E (IgE)-dependent and IgE-independent mechanisms (16). In IgE-independent mechanisms, mast cells can be activated by substances such as calcium ionophore, compound 48/80, substance P, and microbial products (11, 16). They can release potent mediators of inflammation and recently have been shown to play a pivotal role in host defense against bacterial infection (11, 28). The defenses in sepsis are dependent on mast cells that produce tumor necrosis factor alpha (TNF-), which in turn attracts and activates neutrophils to the site of infection (28). However, in all these studies, direct evidence of C. difficile toxin A effect on mast cells has not been described.

Thus, to investigate whether toxin A has direct effects on mast cells, we analyzed the influence of toxin A on the secretion of histamine, TNF-, and nitric oxide (NO) in vitro. We found that toxin A did not induce the release of histamine and NO, although it induced the release of small amounts of TNF-. Moreover, exposure to large doses of toxin A inhibited mast cell activation induced by IgE-dependent and IgE-independent mechanisms and also altered the mast cell cytoskeleton and induced cell death by apoptosis.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Rat PMC. Peritoneal mast cells (PMC) were from 250- to 300-g male Sprague-Dawley rats (Charles River, Canada Inc.), maintained under standard laboratory conditions with food and water ad libitum. PMC were obtained by lavage of the peritoneal cavity with HEPES-buffered Tyrode's solution (HTBS) containing 0.1% bovine serum albumin and isolated in a discontinuous gradient of sterile Percoll (Pharmacia Ltd., Uppsala, Sweden). Purity of isolated cells was checked by staining with toluidine blue (3) and observed under light microscopy. Mast cells used in all experiments were 97 to 99% pure, with viability of >96%. To study the IgE-dependent activation of mast cells, rats were infected with 3,000 third-stage larvae of Nippostrongylus brasiliensis 5 to 6 weeks before PMC isolation (2). To evaluate activation by calcium ionophore A23187 or toxin A, mast cells were obtained from naive rats. All protocols used for mast cell isolation from animals were approved by Animal Ethics Committees from University of Alberta (Canada), as well from the Mexican Institute of Social Security.

Reagents. 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), sulfanilamide, naphthylethylenediamine dihydrochloride, HEPES (sodium salt), H₃PO₄, NaNo₂, calcium ionophore A23187 (Calon), 2-mercaptoethanol, p-phenylenediamine, and E-toxate kit (Limulus amebocyte lysate, for detection of endotoxin) were from Sigma Chemical Co. (St. Louis, Mo. C. difficile toxin A free of endotoxin, as tested with the E-toxate kit, was produced and purified as previously described (45). N. brasiliensis antigen (NbAg) was obtained according to the procedure of White and Pearce (47). RPMI 1640 medium was from Gibco-BRL (Grand Island, N.Y.). Fetal bovine serum was purchased from HyClone (Logan, Utah). Monoclonal antibodies to actin were produced and purified as described by Diaz-Barriga et al. (10). Polyclonal antibodies to toxin A were produced by immunization of rabbits with purified toxin A as described previously (46).

TNF- bioassay. For the TNF- assay, 2.5 × 10⁴ naive PMC were mixed with different concentrations of toxin A (0.001, 0.01, 0.1, and 1 µg/ml) in RPMI 1640 supplemented with 10% FBS (RPMI-FBS) and incubated for 4 h at 37°C under a 5% CO₂ atmosphere. Unbound toxin A was removed by centrifugation at 600 × g, and PMC were suspended in fresh RPMI-FBS. The cell suspension was further incubated for 6 h at 37°C. The cells were then centrifuged at 900 × g, and the supernatant was tested for TNF- cytotoxicity on WEHI 164.13 cells (13). As a control, the direct effect of toxin A (1 to 100 ng/ml) on WEHI 164.13 cells was tested and shown not to be toxic; by contrast, with 1 µg of toxin A per ml, we observed evidence of toxicity. This alteration was abolished with rabbit antiserum (1:50 dilution). Under these conditions, TNF--mediated toxicity was not modified by toxin A. Antibody to toxin A did not alter the TNF--mediated toxicity of mast cells, given that antibody to TNF- fully blocked the toxicity against WEHI cells induced by supernatants from mast cells following toxin A stimulation. Mouse recombinant TNF- (Genzyme Co., Cambridge, Mass.) was used as a standard. Test samples were assayed with concentrations of TNF- from 0.8 to 100 pg/ml. WEHI 164.13 cells were incubated in 50 µl of RPMI-FBS with 50 µM 2-mercaptoethanol for 20 h at 37°C in 5% CO₂ atmosphere. Viability of cells was measured 3 h following addition of 10 µl of MTT (5 mg/ml) per well. Isopropanol-HCl (150 µl) was added to dissolve the purple formazan precipitates. The plate was read on a Vmax kinetic microplate reader (Molecular Devices Co., Menlo Park, Calif.) at 570 nm.

NO production. PMC from uninfected rats (10⁵ cells/well) were incubated for 4 h with different concentrations of toxin A (0.001, 0.01, 0.1, and 1 µg/ml) in RPMI 1640 without phenol red. After washing, they were further incubated for 72 h at 37°C. Cell supernatants were mixed with an equal volume of Griess reagent [1% sulfanilamide, 0.1% N-(1-naphthyl)ethylenediamine dihydrochloride, 2.5% H₃PO₄] and incubated for 10 min at room temperature (20). Concentration of NO₂ was determined by measuring the absorbance at 540 nm with a Vmax kinetic microplate reader (Molecular Devices). NaNO₂ was used as a standard.

Histamine release assay. PMC (2.5 × 10⁴) from N. brasiliensis-infected rats were suspended in 100 µl of HBTS containing 0.1% bovine serum albumin and incubated in the presence or absence of toxin A (0.001, 0.01, 0.1, and 1 µg/ml) for 30 min or 4 h at 37°C in a 5% CO₂ atmosphere. After incubation, cells were washed and resuspended in HBTS and challenged with 10 worm equivalents of NbAg per ml for 20 min. Isolated PMC from uninfected rats were also treated with toxin A, A23187 (2.5 µM), or medium alone under the same conditions. The supernatant and cell pellets were then separated by centrifugation at 3,000 × g. Histamine levels were measured in both supernatant and pellet fractions with a fluorometric assay (43) using a CytoFluor 2350 fluorescence spectrometer (Millipore Co., Bedford, Mass.). The percentage release of the total histamine content was calculated as described previously (4), and the spontaneous release in the absence of toxin, A23187, or NbAg was subtracted to estimate specific histamine release.

Cell morphology studies. One million naive PMC were exposed to either 1 µg of toxin A per ml or vehicle and incubated for 4 h at 37°C in a 5% CO₂ atmosphere; cells were then washed, resuspended in RPMI 1640, and further incubated for 20 h. Subsequently, cells were fixed in 2.5% glutaraldehyde for 1 h and postfixed in 1% OsO₂ in cacodylate buffer (0.3 M, pH 7.2). Cells were gradually dehydrated with ethanol and finally embedded in Epon (18). The samples were thin sectioned and observed by transmission electron microscopy (TEM) with a JEOL-X100 SX microscope (JEOL Co., Tokyo, Japan). For scanning electron microscopy (SEM), after dehydration the samples were critical point dried with CO₂ and gold coated in an ion-sputtering device. Specimens were observed with a JEOL 35C scanning electron microscope.

Cytoskeleton analysis. A total of 2.5 × 10⁴ naive PMC were incubated 37°C for 4 h with 1 µg of toxin A per ml. Cells were fixed in 3.7% paraformaldehyde in phosphate-buffered saline (PBS; pH 7.4) for 30 min at room temperature. After washing, the cells were permeabilized with 0.1% Triton X-100 in PBS for 10 min at room temperature. For visualization of cytoskeletal proteins, monoclonal antibodies directed against actin or tubulin were added to permeabilized cells and incubated for 1 h at room temperature (10). Cells were washed with PBS and then incubated with a fluorescein-conjugated sheep anti-mouse IgG antibody (1:30) (Boehringer, Mannheim, Germany) for 1 h at room temperature. Finally, after washing with PBS, coverslips were mounted with p-phenylenediamine in Gelvatol (100 µg/ml; Monsanto-Birchman Bend Plant, Birchman Bend, Mass.) and analyzed with an Olympus BH-2 fluorescence microscope (Olympus Co., Tokio, Japan).

Analysis of DNA fragmentation. A suspension of 2 × 10⁶ naive PMC was mixed with 1 µg of toxin A per ml and incubated for 4 h at 37°C. Cells were then washed with PBS (pH 7.4) and resuspended in lysis buffer (10 mM EDTA, 50 mM Tris-HCl [pH 8], 0.5% sodium dodecyl sulfate) with proteinase K (500 µg/ml) and RNase A (50 µg/ml). After incubation at 50°C for 2 h, DNA was extracted twice with an equal volume of phenol-chloroform-isoamyl alcohol (Gibco Co., Gaithersburg, Md.). DNA obtained in the aqueous fraction was precipitated overnight in 0.1 volume of sodium acetate (3 M) and 2.5 volumes of 100% ethanol at 20°C. After washing with 70% ethanol, DNA pellet was resuspended in 20 µl of TE buffer (40 mM Tris-HCl, 1 mM EDTA), and 10 µg of DNA was electrophoresed on a 1% agarose gel and stained with SYBR green (Molecular Probes, Eugene, Oreg.).

Statistical analysis. The two-tailed paired Student's t test was used for statistical evaluation of TNF- and NO studies. Results were considered significantly different when P was <0.05. For histamine release, a three-way interaction variance analysis was used. Results were considered significantly different when P was <0.01 with the Scheffe test.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

The viability of PMC from both N. brasiliensis-infected and uninfected rats was not affected by exposure of up to 4 h with doses of toxin A from 0.001 to 1 µg/ml. However, exposure to 1 µg/ml for longer periods of time (24, 48, and 72 h) decreased the viability of mast cells (17, 30, and 60%, respectively), whereas exposure to 10 µg of toxin A per ml for 4 and 24 h reduced cell viability to 12 and 26%, respectively. Accordingly, for studies of mediator secretion, toxin A doses of up to 1 µg/ml and exposure for up to 4 h were used.

Effect of C. difficile toxin A on TNF- and NO production. To determine the effects of toxin A on TNF- production, PMC from uninfected rats were incubated with different concentrations of toxin A for 4 h. TNF- in cell-free supernatants was assessed on TNF--sensitive WEHI cells. Treatment of PMC with 0.01 or 0.1 µg of toxin A per ml stimulated (P < 0.05) release of TNF- (Fig. 1A), from 14.4 pg/10⁶ in sham-treated PMC to 36.5 or 46 pg/10⁶ cells, respectively.

View larger version (12K): [in this window] [in a new window]   FIG. 1.   (A) Effects of toxin A (Tox A) on TNF production by rat PMC. Cells were incubated with toxin A for 4 h, washed, and then cultured for an additional 6 h before the supernatants were assayed for TNF- or NO. Results are the means ± SE for five experiments. Shown are effects on TNF- production (*, P < 0.05 by comparison with cells not treated with toxin A). (B) Effect on NO production. PMC were treated under same conditions as above (P > 0.05 by comparison with cells not treated with toxin A).

Effect of toxin A on histamine release from PMC stimulated by IgE- and non-IgE-dependent pathways. To investigate the effects of toxin A on IgE-dependent histamine release, PMC from N. brasiliensis-infected and uninfected rats were incubated with different concentrations of toxin (Fig. 2). Toxin A (4-h incubation, doses from 0.001 to 1 µg/ml) did not stimulate histamine secretion from normal or N. brasiliensis-sensitized mast cells above the level of spontaneous release (2.6% ± 0.5%, mean ± standard error [SE]). However, exposure to 1 µg of toxin A per ml significantly inhibited histamine release (77%, P < 0.01) from sensitized mast cells stimulated with NbAg. To analyze the inhibitory effect of toxin A on IgE-independent histamine release, naive PMC were incubated with toxin A and then treated with A23187 (2.5 µM) for 20 min. A significant inhibition of histamine release (41%, P < 0.01) was observed when PMC were treated with 1 µg of toxin A per ml for 4 h. However, treatment of PMC with up to 1 µg of toxin A per ml for 30 min did not inhibit IgE-dependent or IgE-independent histamine release (data not shown).

View larger version (11K): [in this window] [in a new window]   FIG. 2.   Effect of C. difficile toxin A on NbAg and on A23187 induction of histamine release by PMC. The cells were incubated for 4 h with different concentrations of toxin A (Tox) and further stimulated with NbAg (Ag; 10 worm equivalents/ml) or A23187 (Ion; 2.5 µM) for 20 min. A significant inhibition (*, P < 0.01) was observed on mast cells treated with 1 µg of toxin A per ml plus antigen (77%) or toxin A plus A23187 (41%). Results are the mean ± SE of five experiments with duplicates in each experiment.

Cytoskeletal changes associated with toxin A treatment on PMC. There were cytoskeletal changes in rat PMC after incubation with 1 µg of toxin A per ml for 4 h (Fig. 3). Sham-treated PMC cultured for 4 h showed a normal distribution of actin filaments (Fig. 3A). In contrast, PMC treated with 1 µg of toxin A per ml for 4 h showed cytoplasm retraction and cytoskeletal rearrangements with actin-containing dense deposits in the cytoplasm (Fig. 3B).

View larger version (80K): [in this window] [in a new window]   FIG. 3.   Effect of toxin A on PMC actin microfilaments. (A) Mast cells showed a normal distribution of actin filaments without toxin A treatment. (B) Mast cells treated with 1 µg of toxin A per ml showed cytoskeletal rearrangements with actin-forming dense deposits in the cytoplasm and cytoplasm retraction.

Morphological changes in PMC induced by C. difficile toxin A. TEM and SEM were used to examine naive PMC treated with 1 µg of toxin A per ml for 4 h (Fig. 4). Naive cells showed a regular distribution of granules with well-dispersed chromatin in the nuclei. The plasma membrane was normal, and granule contents were clearly visible (Fig. 4A). However, PMC treated with toxin A (Fig. 4C) showed a decreased number of membrane microvilli, changes in electrodense granules in the cytoplasm, and pyknotic nuclei with condensation of nuclear chromatin, all features characteristic of apoptosis. Furthermore, PMC treated with toxin A and observed with SEM (Fig. 4D) showed morphological changes and evidence of enhanced granule release compared with untreated cells (Fig. 4B).

View larger version (149K): [in this window] [in a new window]   FIG. 4.   TEM and SEM of control PMC (A and B) and PMC treated with 1 µg of toxin A per ml for 4 h (C and D). TEM showed that mast cells exposed to toxin A underwent loss of membrane microvilli and nuclear chromatin condensation (C). With SEM, some cells showed signs of alteration: surface blebbing, loss of rounding and normal granule organization, cell membrane disorganization, and changes in morphology (D).

DNA fragmentation. DNA from PMC treated with 1 µg of toxin A per ml showed a characteristic ladder pattern after agarose electrophoresis (Fig. 5, lane C). DNA from sham-treated PMC (Fig. 5, lane B) did not show DNA fragmentation.

View larger version (47K): [in this window] [in a new window]   FIG. 5.   Analysis of DNA fragmentation of PMC treated with 1 µg of toxin A per ml for 4 h. Ten micrograms of DNA was separated in a 1% agarose gel and stained with SYBR green. Lane A, molecular weight marker profile from a 123-bp ladder from Gibco-BRL; lane B, DNA from PMC incubated in RPMI 1640-5% FBS for 24 h; lane C, fragmented genomic DNA from PMC treated with toxin A for 24 h.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

In experimental enterocolitis, toxin A causes inflammation and necrosis of the intestinal wall (24, 35, 44). Both mast cells and other leukocytes have been implicated in the pathophysiology of toxin A-induced mucosal damage. To understand the mechanisms involved, several in vivo and in vitro approaches have been used. For example, the in vitro interaction of toxin A with rat peritoneal macrophages induces release of interleukin-1 and TNF-, which attract and activate neutrophils (40). Toxin A also causes release of RMCPII when injected in ileal loops of rats (6). This effect can be inhibited with a specific antagonist of substance P (34), suggesting that toxin A activates mast cells via substance P released from adjacent sensory neurons (30). Thus, macrophages, other leukocytes, and mast cells may play a role in the inflammation and necrosis associated with toxin A.

Although mast cells are an important source of several inflammatory mediators and have been implicated in toxin A-induced enterocolitis, no studies about the direct action of toxin A on isolated mast cells have been published. Thus, our study showed that low doses of toxin A induce the secretion of TNF- from freshly isolated mast cells in vitro. This is to our knowledge, the first direct report showing that a bacterial enterotoxin causes release of TNF- from mast cells. The ability of toxin A to induce release of TNF- is not limited to mast cells, as both toxins A and B cause release of TNF- by human monocytes (15).

In contrast to macrophages, mast cells store TNF- in granules and can therefore release it immediately after toxin A activation and participate in the early phase of the inflammatory reaction. TNF- is also known to cause necrosis of epithelial cells of the intestine, as well as inhibition of gastric motility (42). Thus, all the above effects suggest that mast cell-derived TNF- may play a role in the mucosal damage and fluid accumulation observed in C. difficile-associated colitis.

Although the in vitro ability of toxin A to activate inflammatory cells such as neutrophils, macrophages, and mast cells (this study) has been demonstrated, the in vivo access of the toxin to these cells is in dispute because of the large size of the toxin. However, toxin A can cause detachment and apoptosis of enterocytes (27). In this disrupted epithelium, toxin A might diffuse and interact with inflammatory cells in the lamina propria, including mast cells.

In earlier reports, there is strong evidence that histamine is involved in the C. difficile toxin A experimental colitis in rats (25, 35). However, our results in vitro indicate that toxin A was not able to directly induce the secretion of histamine or NO from mast cells. This contrasts to the in vivo degranulation observed after injection of toxin A in intestinal loops (25), suggesting that histamine is released in vivo due to indirect activation of mast cells, such as with substance P released by enteric nerves (6, 30, 34).

Our results further document that in mast cells, the mechanisms for TNF- release are different from those involved in the release of histamine. Furthermore, we found that toxin A inhibited antigen-dependent and -independent release of histamine by mast cells. To release histamine, vesicles must move and fuse with the cell membrane, a process that requires the participation of the cytoskeleton. We found that in mast cells, like in other cell types, toxin A disrupts the actin microfilaments; this alteration may reduce the capacity of mast cells to release histamine contained in granules. It was recently reported that C. difficile toxins A and B alter cytoskeleton by monoglycosylation of GTP-binding proteins Rho and Rac (17, 22, 23). These proteins have been shown to regulate the role of the cytoskeleton during secretion of inflammatory mediators in mast cells (33, 37, 38). However, the concentration of toxin used to alter the PMC cytoskeleton was high, and it is probably not achieved in vivo. Once again, degranulation of mast cells may occur in vivo by activation of neuronal circuits.

NO is also an effector molecule produced by mast cells that could be involved in the pathophysiology of enterocolitis (41). Interestingly, NO is able to potentiate mast cell cytotoxicity mediated by TNF- (5). However, toxin A did not stimulate NO synthesis by mast cells at any of the concentrations of toxin used. Thus, direct interaction of toxin A with mast cells in vitro may not induce NO production, although this does not preclude synthesis of NO by mast cells or other cells induced by indirect mechanisms such as neuropeptides. Recently, Qiu et al. (39) demonstrated that NO inhibits the enterotoxicity of toxin A on rat small intestine; if NO acts as a mechanism of defense, this might come from cells other than mast cells.

We found that prolonged exposure to toxin A caused apoptosis of PMC. This is the first report on induction of apoptosis in mast cells caused by a bacterial toxin. The induction of apoptosis by bacterial products has been documented in other cell types (8, 21, 32). In fact, toxin A caused apoptosis in human intestinal epithelial cells (27). The ability of toxin A to cause apoptosis raises the interesting possibility that Rho and Rac proteins may be involved in the mechanisms regulating program cell death. We are currently studying the mechanisms involved in toxin A-induced apoptosis in mast cells.

During the acute phase of inflammation and necrosis in the intestinal wall caused by C. difficile toxins, mast cells are active and releasing mediators such as RMCPII, platelet-activating factor, and leukotriene B4 (6, 26, 35). This finding suggests that toxin A-induced apoptosis may not occur at this stage. However, it could be that in severe cases of pseudomembranous colitis, mast cells in the lamina propria may be exposed to toxin A at doses or times that induce apoptosis; macrophages, monocytes, and neutrophils may also be affected (14, 15, 40). This would impair the ability of these cells to counteract infection and may explain in part why severe cases are so refractory to treatment and prone to recrudescence.

Our results suggest that mast cells may participate in the initial inflammatory reaction by releasing TNF- in response to toxin A. In contrast, histamine and NO may not participate when toxin A acts directly on mast cells. The disruption of the cytoskeleton caused by the toxin A may alter the ability of mast cells to secrete histamine and NO after antigen-dependent or -independent activation. In opposition to the necrotic action of toxin A in other cell types, it induces apoptosis in mast cells. Prolonged exposure to toxin A decreases function and viability of intestinal inflammatory cells, hindering their ability to counteract infection; this would probably favor severe and long-lasting cases of pseudomembranous colitis.