- Email: [email protected]
Differential effects of Clostridium difficile toxins A and B on rabbit ileum
GASTROENTEROLOGY
ALIMENTARY
1987;93:273-9
TRACT
Differential Effects of CIosfridium difficile Toxins A and B on Rabbit Ileum GEORGE
TRIADAFILOPOULOS,
CHARALABOS
POTHOULAKIS,
MICHAEL J. O’BRIEN, and J, THOMAS LAMONT Evans Memorial Department of Clinical Research and Mallory Institute of Pathology, University Hospital, Boston, Massachusetts
The pathogenesis of Clostridium difficile enterocolitis appears to involve colonization of the bowel followed by release of toxin A, an enterotoxin, and toxin B, a cytotoxin. The purpose of this study was to determine the effect of purified toxins A and B on intestinal secretion, epithelial permeability, and morphology in perfused rabbit ileal loops. Intestinal permeability after toxin exposure was assessed by blood-to-lumen clearance of [3H]mannitoI. Toxin A at doses of 5-100 pg/lO cm ileal loop caused a threefold to fivefold increase in [3H]mannitol permeability (p < 0.001) vs. equal concentrations of toxin B or buffer control. In addition, perfusate from toxin A-exposed loops contained significantly more neutrophils (p < 0.001) than toxin B or control loops. Toxin A caused severe epithelial cell necrosis with destruction of villi and polymorphonuclear infiltration Electron microscopy of mucosa subjected to a low dose of toxin revealed widespread nonspecific dilatation of endoplasmic reticulum and mitochondrial swelling. In contrast to these effects of toxin A in ileal loops, in vitro experiments with ileal explants in short-term organ culture revealed that toxin A had no effect on epithelial cell permeability, protein synthesis, release of alkaline phosphatase, or morphology. Our results show that purified toxin A but not toxin B causes severe inflammatory enteritis in rabbit ileal loops, but has no discernable effect on rabbit ileum in vitro. We speculate that Received July 9, 1986. Accepted October 14, 1986. Address requests for reprints to: J. Thomas LaMont, M.D., University Hospital, setts 02118.
75 East
Newton
Street,
Boston,
Massachu-
This work was supported by grant AM34583 from the National Institutes of Health. Part of this work was published in abstract form in GASTROENTEROLOGY 1986;90:1671. The authors thank Anne DiSorbo for preparation of the manuscript, and Pat McAndrew and Luis Bustos-Fernandez for technical assistance. 0 1987 by the American Gastroenterological Association 0016.5085/87/$3,50
toxin A may contribute significantly to intestinal damage in C. difficile-associated colitis and diarrhea. Colitis and diarrhea secondary to Clostridium diffitile infection appear to be toxin-mediated. Toxin A, an enterotoxin, causes secretion of fluid in ligated rabbit ileal loops (1) and is lethal when injected into mice (2). The cellular action of toxin A is not known, but does not appear to involve stimulation of intestinal adenylate cyclase (3). Toxin B, a cytotoxin, causes rounding of tissue culture fibroblasts, which is preceded by depolymerization of actin-containing thin microfilaments (4) and an increase in the ratio of globular to filamentous actin in fibroblasts (5). Toxin B is present in the stools of most patients with antibiotic-associated pseudomembranous colitis (6,7) and also in a large percentage of those with nonspecific colitis and diarrhea (8,9). Clinical studies of toxin A have been limited by the lack of a simple, reproducible assay, but recent studies using immunoassay suggest that toxin A is frequently found in stool filtrates of patients with C. dificile colitis (10,ll). The purpose of our study was to determine the effects of purified toxins A and B on the rabbit ileum in vivo and in vitro. We used the perfused rabbit ileal loop model to assess epithelial permeability and morphologic damage after toxin exposure. Parallel studies were performed with rabbit ileal explants in short-term organ culture. Our results suggest that toxin A is an important factor in the pathogenesis of C. dificile enteritis. Materials
and Methods
Materials Male New Zealand white rabbits weighing -2 kg were used in all studies. Animals were fasted overnight before study, but were provided with water ad libitum. Protein concentration of toxin preparations was deter-
274
TRIADAFILOPOULOS
ET AL.
mined with a protein assay kit from Bio-Rad Laboratories (Pichmond, Calif.). Alkaline phosphatase was measured calorimetrically with a kit from Sigma Diagnostics (St. Louis, MO.). L-[4,5-3H]leucine, 130 Ci/mmol, was obtained from Amersham (U.K). IJ-[~-~H (N)]mannitol, 19.1 Ci/ mmol, and 2-[1,2-3H (N)]deoxyglucose, 37.3 Ci/mmol, were obtained from New England Nuclear [Boston, Mass.). Clostridium diffcile strain 10463, a highly toxigenic strain, was kindly supplied by Dr. Nadine Sullivan of the Department of Anaerobic Microbiology, Virginia Polytechnic Institute and State University, Blacksburg, Va. Methods Toxin preparation. Toxins A and B were prepared from culture supernatants of C. dificile strain 10463. Toxin B was purified to homogeneity (single protein band at 50 kilodaltons on sodium dodecyl sulfate-polyacrylamide gel electrophoresis) as previously described by us (5). Toxin A was partially purified by ammonium sulfate precipitation and diethylaminoethyl sepharose anion exchange chromatography by the method of Sullivan et al. (2). This preparation showed a single band on 10% polyacrylamide gel under nondenaturing conditions and a major band at 229 kilodaltons by sodium dodecyl sulfatepolyacrylamide gel electrophoresis, with several faint minor bands. Toxin B biplogic activity was assayed by the rounding of IMR-90 fibroblasts exposed to serial lo-fold dilutions as previously described (4). Typical preparations of toxin B used in these studies contained between 80 and 250 pg protein/ml as measured by Bio-rad assay (12). Cell rounding was completely inhibited by preincubation with goat antitoxin B (Virginia Polytechnic Institute and State University, Blacksburg, Va.). Enterotoxin activity was assayed in the ligated rabbit ileal loop assay (13). Enterotoxin preparations used in this study exhibited cytotoxicity in the fibroblasts assay at a 10m3 to 10m4 dilution, as previously reported by others (1,2). Rabbit intestinal pe$usion studies. After catheterization of an ear vein and induction of general anesthesia by injection of pentobarbital sodium (Nembutal, 50 mg/ml, Abbott, North Chicago, Ill.), the abdomen was opened by a midline incision and IO-cm-long segments of the distal ileum were ligated with silk to form ileal loops. For each animal, three IO-cm loops were formed with at least a 5-cm distance between loops. Each loop was injected with 1 ml of either toxin A (5-100 pg per loop), phosphate-buffered saline, or toxin Q (100 pg per loop). The abdomen was closed in two layers and animals were maintained under light anesthesia for 2-6 h. The abdomen was then reopened and both renal pedicles were ligated to prevent renal excretion of [3H]mannitol (14). An intravenous injection of 100 &i [3H]mannitol was given by ear vein. Each ileal loop was then separately cannulated at its proximal and distal ends with a 19F Foley-type catheter and connected to a perfusion pump (Harvard Instruments, Cambridge, Mass.). Continuous perfusion of the three ileal loops was initiated at a rate of 0.5 ml/min with Ringer’s lactate (6 g/L NaCl, 3.1 g/L sodium lactate, 0.3 g/L KCl, 0.2 g/L CaC12, pH 7.5) as the perfusion buffer. Every 15 min thereafter, aliquots of perfusate were collected and proc-
GASTROENTEROLOGY
Vol. 93, No. 2
essed for radioactivity and neutrophil number by hemocytometry. Throughout the 4-h perfusion period the animals were anesthetized with pentobarbital intravenously, and their body temperature was maintained at 37”-38°C with a heating lamp. Rabbits were killed at the end of each perfusion study with a bolus of pentobarbital. Fullthickness samples of ileal loops were fixed and processed for light and transmission electron microscopy. Morphologic studies. One-square centimeter samples of treated and control loops were placed in 10% phosphate-buffered formalin. After fixation, 1.0 x 0.2-cm strips were processed in a VIP-200 tissue processor (Miles Laboratories, Inc., Elkhart, Ind.) and embedded on edge in paraffin. Serial sections were studied with hematoxylin and eosin. Tissue for electron microscopy was fixed in half-strength Karnovsky’s solution for 4 h, washed in phosphate buffer, postfixed in 2% osmium tetroxide, dehydrated in acetone, and embedded in epon-araldite resin. One-micrometer sections were stained with Richardson’s blue and areas were selected for thin sectioning (60-80 nm). Sections were mounted on copper grids and viewed on a Philips E.M. 300 electron microscope (Philips Electronic Instruments, Mahwah, N.J.). Organ culture studies. We next studied the effect of toxin A on rabbit intestinal explants in short-term organ culture (15). For determination of protein synthesis, rabbit ileal explants measuring -2 x 2 mm were cut and incubated in sterile organ culture dishes with 1.5 ml of Trowell’s medium (Gibco Laboratories, Grand Island, N.Y.) containing 10% fetal calf serum; penicillin, 100 U/ml; streptomycin, 100 pgiml; and [3H]leucine, 10 &i/ml. Toxin A was diluted in medium and added in a volume of 0.15 ml; medium without added toxin was used as control. Incorporation of [3H]leucine into explant proteins during a 2-h incubation was determined as previously described (16), and was expressed as disintegrations per minute per milligram of explant protein. We also measured alkaline phosphatase in explants and medium after toxin A exposure as an indicator of membrane damage from toxin (17). Enzyme was measured using pnitrophenol phosphate as substrate (18). Results were expressed as Sigma enzyme units per milligram explant protein per 2 h, as estimated by the method of Lowry et al. (19). A Sigma unit is defined as the amount of enzyme that will release 1 pmol of p-nitrophenol in an hour. We also determined the effect of toxin A on epithelial membrane permeability in ileal explants by the method of Walum and Peterson (20). Explants were incubated with glucose-free Krebs’ solution (21) containing 5 &i/ml of [3H]2-deoxyglucose. After a 1-h incubation at 37”C, explants were washed three times with Krebs’ solution containing D-glucose, 1 mgiml, and then were transferred to sterile organ culture dishes and covered with 0.5 ml of Krebs’ buffer with glucose (37°C) containing either toxin A, Triton X-100 (5% volivol), or no addition [control). Every 15 min the medium was aspirated and counted in a scintillation counter, and explants were covered with 0.5 ml of fresh medium. After 2 h explants were removed and homogenized with 1 ml of Krebs’ buffer with 5% vol/vol Triton X-100. Representative explants from each group were fixed in formalin for light microscopy. Radioactivity
August
Figure
1987
1. Effect of toxins A and B on blood-to-lumen clearance of [3H]mannitol. Ligated rabbit ileal loops were injected with either toxin A, toxin B, or control for 6 h and then perfused with buffer for 4 h. Intestinal permeability to [3H]mannitol was estimated by scintillation counting of aliquots of perfusate collected every 15 min during a 4-h period as described in Methods. Eact point represents disintegrations per minute per 0.2 ml of perfusate.
in the explants
and aspirated medium was totaled and set at loo%, and percentages of radioactivity remaining in the monolayer were calculated for each time point as described by Walum and Peterson (20).
Results In Vivo Effects of Toxins A and B on 13H]Mannitol Permeability Preliminary experiments revealed that 100 pg of toxin A reproducibly caused secretion of a hemorrhagic fluid in closed rabbit ileal loops after a 16-h incubation, as defined by a volume/length ratio of 21.0 (22). The results of a typical perfusion experiment as described in Methods are shown in Figure 1, where the effects of toxins A and B (100 pg per loop) and buffer control are compared. Toxin A caused a striking increase in [“Hlmannitol permeability compared with toxin B or buffer control, which was apparent in the first 15-min collection after a 6-h exposure to toxin. The combined results of 15 ileal loop perfusions in 5 rabbits are shown in Figure 2. In all experiments toxin A caused at least a threefold increase in blood-to-lumen clearance of [3H]mannitol with equal concentrations of toxin B or buffer control. Because toxin A elicited secretion of a hemorrhagic, purulent exudate, we examined ileal loop perfusate for the presence of neutrophils. In toxin A-exposed loops we found a significant increase (p < 0.001) in neutrophils: 131 k 46/mm3 compared with toxin B-exposed loops, 10 ? 2/mm3, or buffer-exposed loops, 14 ? 5/mm3. In additional experiments we explored the minimal dosage of toxin A required to produce an effect
Clostridium
dificile
TOXINS
275
on [3H]mannitol permeability. We observed that 5 pug of purified toxin A was sufficient to produce a threefold increase in [3H]mannitol permeability after a 2-h incubation in a closed ileal loop (1274 dpm toxin A vs. 378 dpm control). Because toxins A and B are both secreted by toxigenic strains of C. dificile, we added mixtures of toxin A (5 pg) and toxin B (10 pg) to ascertain if they had an additive effect on epithelial permeability. In five separate experiments, mixtures of toxins A (5 pg) and B (10 pg) were not more potent than toxin A (5 pg) alone as determined by blood-to-lumen clearance of [3H]mannitol (not shown). To ascertain that toxin B was not undergoing inactivation in rabbit ileal loops, we tested the cytotoxicity titer of toxin B before and after a 2-h incubation in a IO-cm rabbit ileal loop. The toxin titer of lop7 to lOpa was unchanged after incubation in a closed loop for 2 h. Intestinal Morph01 ogy of Toxin-Exposed Perfused Ileal Loops Light microscopy of tissue samples from the perfused loops (Figure 3) revealed striking morphologic damage in the toxin A-exposed loops compared with toxin B or buffer. Normal ileal morphology was observed in loops exposed to phosphate-buffered saline or toxin B. Ileal loops exposed to 10 or 25 pug of toxin A showed pathologic findings that ranged from acute inflammation and reparative hyperplasia of villi, to necrosis of villi and transmural hemorrhagic necrosis. The earliest change seen was a polymorphonuclear infiltrate, most dense toward the villus tips and within the epithelium. The lumen contained numerous polymorphs, degenerating epithelial cells with nuclear pyknosis and karyorrhexis,
0 Toxin (N=5)
Toxin B (N=5)
Buffer (N=5)
Figure 2. Combined results of the effect of toxins A and B on blood-to-lumen clearance of [3H]mannitol. Ligated ileal loops (n = 15) from 5 rabbits were injected with either toxin A (100 pg), toxin B (100 pg), or control (buffer) for 6 h and then perfused with buffer for 4 h. Intestinal permeability to [3H]mannitol was estimated by scintillation counting of perfusate. Results are expressed as mean 2 SEM disintegrations per minute per milliliter perfusate for all collections (p < 0.001 vs. control).
276
TRIADAFILOPOULOS
Figure 3. Rabbit ileal loop Figure 1. At the microscopy. Left photomicrograph
GASTROENTEROLOGY Vol. 93, No. 2
ET AL.
morphology after toxin exposure. Rabbit ileal loops were injected with toxins and perfused as described in end of the perfusion period full-thickness samples of loops were fixed in formalin and processed for light photomicrograph shows a loop injected with buffer, middle photomicrograph with toxin B (100 pg), and right with toxin A (100 pg) (magnification x150).
and goblet cells. A more severe lesion was induced by exposure to 100 pg of toxin A per loop, consisting of necrosis of villi with marked reparative hyperplasia of the flat mucosal surface. The lamina propria surrounding the residual crypts was greatly expanded by red blood cells and abundant polymorphs. The luminal exudate and dehisced epithelial cells were more abundant. This severe lesion progressed in some experiments to complete necrosis of the mucosa, which was replaced by a neutrophilic exudate, enmeshing debris, and red blood cells. The most severe lesion consisted of hemorrhagic necrosis involving the full thickness of the bowel wall. Transmission electron microscopy of ileal loop mucosa subjected to 10 pg of toxin A demonstrated extensive dilatation of the endoplasmic reticulum of columnar cells, which gave the cytoplasm a cribriform appearance (Figure 4). Many cells also showed swelling of mitochondria. Organ Culture Experiments Organ culture experiments were carried out to ascertain if toxin A had any significant effect on rabbit ileum in vitro. Previous experiments in our laboratory (16) indicated that toxin B significantly inhibited [3H]leucine incorporation into hamster cecal explants after a 2-h exposure in organ culture. However, toxin A at 20 pg/ml had no effect on incorporation of [3H]leucine into ileal explants after a 2-h incubation (Table 1). Alkaline phosphatase levels in human jejunal explants were previously
used by Falchuk et al. (17) to monitor in vitro toxicity of gliadin. We therefore measured alkaline phosphatase activity in rabbit ileal explants exposed to toxin A. No significant effect of toxin A was observed on alkaline phosphatase concentration in explants or medium after a 2-h incubation. When toxin A-exposed explants were examined by light microscopy, no significant alteration in morphology was observed in comparison with explants incubated with buffer. Because of the profound effect of toxin A on epithelial permeability in rabbit ileum in vivo we examined the effect of toxin A on leakage of [“H]2deoxyglucose from rabbit ileal explants. This technique was previously validated as a sensitive assay of plasma membrane permeability after exposure to bacterial toxins (23). As shown in Figure 5, toxin A had no effect on leakage of [3H]2-deoxyglucose from rabbit ileal explants. As expected, Triton X-100 (5% vol/vol), a known membrane active detergent, caused increased release of [“H]2-deoxyglucose compared with control, and light microscopy revealed complete mucosal autolysis.
Discussion Our results indicate that C. dificile toxin A is a potent toxin in the rabbit ileum, eliciting severe intestinal inflammation and alteration of epithelial permeability. In contrast, toxin B in comparable doses had no demonstrable effect on [3H]mannitol permeability, migration of neutrophils into the lumen, or intestinal morphology. The absence of any
August 1987
Clostridium
diffcile
TOXINS
277
Figure 4. Electron microscopy of rabbit ileal loops after toxin exposure. Rabbit ileal loops were injected with buffer (left panel) or 10 pg toxin A (right panel) as described in Figure 1. At the end of the 4-h perfusion period the tissues were fixed and processed for electron microscopy. Final magnification X 22,500.
demonstrable effect of toxin A on ileal epithelial explants in organ culture compared with the profound effects observed in vivo suggests that toxin A triggers infiltration of the ileum with neutrophils that release inflammatory mediators, causing fluid secretion, hemorrhagic necrosis, and altered membrane permeability. This mechanism is also supported by our recent observation (24) that purified toxin A, but not toxin B, is a powerful chemotactic protein for human neutrophils in vitro and also causes a prompt rise in cytosolic calcium as measured by quin 2 fluorescence. We have also recently
Table
1. Effect of Toxin
A on
Protein Svnthesis and Alkaline PhosDhatase in Ileal Exulants
Incorporation of [3H]leucine (dpmimg) Toxin A (20 I.Lgiml) Control (no addition)
observed (25) that toxin A releases serotonin and p-hexosaminidase from cultured rat basophilic leukemia cells, which resemble intestinal mucosal mast cells as regards secretagogue responsiveness (26). Our results are in essential agreement with several previous reports regarding the effects of C. dificile toxin A on the intestine. Lyerly et al. (27) showed that 50 pg of toxin A elicited a viscous hemorrhagic fluid response in rabbit ligated ileal loops, whereas the same amount of toxin B produced a weak variable response without hemorrhage. The same investigators recently reported (28) that intragastric ad-
42,089 k 7,576 43,851 !z 8,038
Secreted proteins (dpm/mg) 4,238 * 2,114 3,807 rt 1,586
Alkaline phosphatase in explants (LVmg protein z h) 14.6 2 2.2 9.9 zt 0.5
Alkaline phosphatase in medium (Uimg protein . 2 h) 4.5 f 0.5 3.7 + 0.4
Rabbit intestinal explants were incubated with toxin A or no addition (control) for 2 h in medium with [3H]leucine for estimation of protein synthesis, and without [3H]leucine for estimation of alkaline phosphatase (n = 5 for each group). Incorporation of [3H]leucine into proteins and alkaline phosphatase levels in tissue and medium were determined as described in Methods. Results are expressed as mean 2 SEM disintegrations per minute per milligram explant protein for [3H]leucine incorporation, and as mean 2 SEM alkaline phosphatase units per milligram explant protein per 2 h. No statistical differences were noted between the toxin A and control groups.
278
TRIADAFILOPOULOS
+
GASTROENTEROLOGY Vol. 93, No. 2
ET AL.
40-
.? 8 ! e
20-
I!
Trmn
I
I
1
30
I
1
60
t
0
so
1
X-100
1
120
Minutes
Figure 5. Effect of toxin A on membrane permeability in vitro. Rabbit intestinal explants were preincubated with [3H]2-deoxyglucose for 1 h, then washed and incubated with toxin A (20 &ml), 5% (volivol) Triton X-100, or no addition (control) for 15 min. Release of radioactivity from the explants in the medium was measured at 15-min intervals and percentage remaining in explants was calculated as explained in Methods. Each point represents the mean of measurements on four explants.
ministration of toxin A to hamsters and mice caused accumulation of hemorrhagic fluid in the small intestine and cecum. Toxin B alone had no effect in their model, but when toxin A and B were administered together, a synergistic effect was observed. We were unable to elicit a synergistic effect of both toxins in our model; however, our results do not exclude a synergistic effect in natural or experimental C. dificile infection, which evolves over a much longer period of time. Mitchell et al. (29) likewise reported that purified toxin A, but not partially purified toxin B, caused increased permeability and fluid accumulation in rabbit ileum and colon. Taken together with our results, it would appear that toxin A by itself is capable of causing severe hemorrhagic enteritis in acute animal models, whereas toxin B alone has no or only minimal effect under the same experimental conditions. The intriguing observation by Lyerly et al. (28) that the toxins may act synergistically, and the observation that hemorrhagic colitis is not commonly seen in humans with antibioticassociated C. dificile infection (SO), suggest that the pathophysiology of naturally acquired infection may be different from these experimental models. The cellular mechanism by which toxin A damages the intestine is not known. Although our organ culture experiments showed no effect of toxin A on protein synthesis, alkaline phosphatase, or membrane permeability, we cannot exclude a direct effect of the toxin on the enterocyte. For example, Roth-
man et al. (31) demonstrated that high concentrations (1 mg/ml) of toxin A caused leakage of intracellular K+ and inhibition of protein synthesis in HeLa cells. Moreover, several preliminary reports suggest that toxin A (32) and crude C. diffcile culture filtrate (33) cause morphologic damage to stripped rabbit ileal mucosa in Ussing chambers, increased short circuit current, and increased net chloride secretion. In summary, highly purified toxin A, but not toxin B, of C. difficile elicited an intense inflammatory response in rabbit ileal loops which was accompanied by a striking change in epithelial permeability to [3H]mannitol. In short-term organ culture, toxin A had no effect on protein synthesis, alkaline phosphatase, or membrane permeability. We speculate that inflammatory mediators released by neutrophils, mast cells, or macrophages are likely to contribute to the morphologic and functional changes reported here.
References 1. Taylor NS, Thorne GM, Bartlett JG. Comparison of two toxins produced by Clostridium dificile. Infect Immun 1981;34: 1036-43. 2. Sullivan NM, Pellet S, Wilkins TD. Purification and characterization of toxins A and B of Clostridium dificile. Infect Immun 1982;35:1032-40. 3. Lonnroth I, Lange S. Toxin A of C. dificile: production, purification and effect in mouse intestine. Acta Path01 Microbiol Immunol Stand 1983;91:395-400. 4. Wedel N, Toselli P, Pothoulakis C, et al. Ultrastructural effects of Clostridium dificile toxin B on smooth muscle cells and fibroblasts. Exp Cell Res 1983;148:413-22. 5. Pothoulakis C, Barone LM, Ely R, et al. Purification and properties of Clostridium diffcile cytotoxin B. J Biol Chem 1986;261:1316-21. 6. George RH, Symonds JM, Dimock F. Identification of Clostridium dificile as a cause of pseudomembranous colitis. Br Med J 1978;1:695. 7. Willey SH, Bartlett JG. Cultures from Clostridium dificile in stools containing a cytotoxin neutralized by Clostridium sordellii antitoxin. J Clin Microbial 1979;10:880-7. 8. Chang TW, Lauermann M, Bartlett JG. Cytotoxicity assay in antibiotic-associated colitis. J Infect Dis 1979;140:765-9. 9. George WL, Rolfe RD, Finegold SN. Clostridium dificile and its cytotoxin in feces of patients with antimicrobial agentassociated diarrhea and miscellaneous conditions. J Clin Microbial 1982;15:1049-53. im10. Lyerly DM, Sullivan NM, Wilkins TD. Enzyme-linked munoabsorbent assay for Clostridium diffcile toxin A. J Clin Microbial 1983;17:72-8. 11. Laughon BE, Viscidi RP, Gdovin SL, Yolken RI-I, Bartlett JG. Enzyme immunoassays for detection of Clostridium dificile toxins A and B in fecal specimens. J Infect Dis 1984;149: 781-8. 12. Bradford
M. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein dye binding. Anal Biochem 1976;72:248-54. 13. De SN, Chatterje DN. An experimental study on the mechanism of action of Vibrio cholerae on the intestinal mucus membrane. J Path01 Bacterial 1953;66:559-62.
Clostridium
August 1987
14. Loehry
CA, Axon ATR, Hilton PJ, Hider RC, Creamer B. Permeability of the small intestine to substances of different molecular weight. Gut 1970;11:466-70. 15. LaMont JT, Turner BS, DiBenedetto D, Handin R, Schaffer AI. Arachidonic acid stimulates mucin secretion in prairie dog gallbladder. Am J Physiol 1983;245:G92-8. 16. Pothoulakis C, Triadafilopoulos G, Clark M, Franzblau C, LaMont JT. Clostridium diffcile cytotoxin inhibits protein synthesis in fibroblasts and intestinal mucosa. Gastroenterology 1986;91:1147-53. 17. Falchuk ZM, Gebhard RL, Sessoms C, Strober W. An in vitro model of gluten-sensitive enteropathy. Effect of gliandin on intestinal epithelial cells of patients with gluten-sensitive enteropathy in organ culture. J Clin Invest 1974;53:487-500. 18. Bessey OA, Lowry OH, Brock MJ. A method for the rapid determination of alkaline phosphatase with five cubic millimeters of serum. J Biol Chem 1946:321-g. 19. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Protein measurement with the Folin phenol reagent. J Biol Chem 1951;193:265-75. 20. Walum E, Peterson A. Tritiated 2-deoxy-o-glucose as a probe for cell membrane permeability studies. Anal Biochem 1982;120:8-11. 21. LaMorte WW, Hingston SJ, Wise WE. pH dependent activity of H,- and Hz-histamine receptors in guinea pig gallbladders. J Pharmacol Exp Ther 1981;217:63040. 22. Duncan CL, Sugiyama H, Strong DH. Rabbit ileal loop response to strains of Clostridium perfringens. J Bacterial 1968;95:1560-6. 23. Thelestam M, Mollby R. Sensitive assay for detection of
24.
25.
26.
27.
28.
29.
30. 31.
32.
33.
diffcile
TOXINS
279
toxin-induced damage to the cytoplasmic membrane of human diploid fibroblasts. Infect Immun 1975;12:225-32. Pothoulakis C, Sullivan DA, Melnick AJ, Gadenne T, Meshulam T, LaMont JT. Clostridium dificile toxins A and B stimulate intracellular calcium release in human neutrophils. Clin Res 1986;34:53OA. Pothoulakis C, Theoharides T, McAndrew PE, LaMont JT. C. difficile toxin A releases mediators from mucosal-like rat basophilic leukemia (RBL) cells. Clin Res 1986;34:871A. Seldin DC, Adelman S, Austen KF, et al. Homology of the rat basophilic leukemia cell and the rat mucosa magt cell. Proc Nat1 Acad Sci (USA) 1985;82:3871-5. Lyerly DM, Lockwood PE, Richardson SH, Wilkins TD. Biological activities of toxins A and B of Clostridiuq dificile. Infect Immun 1982;35:1147-50. Lyerly DM, Saum KE, MacDonald DK, Wilkins TD. Effects of C. diffcile toxins given intragastrically to animals. Infect Immun 1985;47:349-52. Mitchell TJ, Ketley JM, Haslam SC, et al. Effect of toxin A and B of Clostridium dificile on rabbit ileum and colon. Gut 1986;27:78-85. Trnka YM, LaMont JT. Clostridium dijj?cile colitis. Adv Intern Med 1984;29:85-107. Rothman SW, Brown JE, Diecidue A, Foret DA. Differential cytotoxic effects of toxins A and B isolated from Clostridium dificile. Infect Immun 1984;46:324-31. Stephen J, Redmond SC, Mitchell TJ, et al. Clostridium difFcile enterotoxin (toxin A): new results. Biochem Sot Trans 1984;12:194-5. Hughes S, Warhurst G, Turnberg LA, Higgs NB, Gingliano LG, Drasar BS. Clostridium dificile toxin-induced intestinal secretion in rabbit ileum in vitro. Gut 1983;24:94-8.
ALIMENTARY
1987;93:273-9
TRACT
Differential Effects of CIosfridium difficile Toxins A and B on Rabbit Ileum GEORGE
TRIADAFILOPOULOS,
CHARALABOS
POTHOULAKIS,
MICHAEL J. O’BRIEN, and J, THOMAS LAMONT Evans Memorial Department of Clinical Research and Mallory Institute of Pathology, University Hospital, Boston, Massachusetts
The pathogenesis of Clostridium difficile enterocolitis appears to involve colonization of the bowel followed by release of toxin A, an enterotoxin, and toxin B, a cytotoxin. The purpose of this study was to determine the effect of purified toxins A and B on intestinal secretion, epithelial permeability, and morphology in perfused rabbit ileal loops. Intestinal permeability after toxin exposure was assessed by blood-to-lumen clearance of [3H]mannitoI. Toxin A at doses of 5-100 pg/lO cm ileal loop caused a threefold to fivefold increase in [3H]mannitol permeability (p < 0.001) vs. equal concentrations of toxin B or buffer control. In addition, perfusate from toxin A-exposed loops contained significantly more neutrophils (p < 0.001) than toxin B or control loops. Toxin A caused severe epithelial cell necrosis with destruction of villi and polymorphonuclear infiltration Electron microscopy of mucosa subjected to a low dose of toxin revealed widespread nonspecific dilatation of endoplasmic reticulum and mitochondrial swelling. In contrast to these effects of toxin A in ileal loops, in vitro experiments with ileal explants in short-term organ culture revealed that toxin A had no effect on epithelial cell permeability, protein synthesis, release of alkaline phosphatase, or morphology. Our results show that purified toxin A but not toxin B causes severe inflammatory enteritis in rabbit ileal loops, but has no discernable effect on rabbit ileum in vitro. We speculate that Received July 9, 1986. Accepted October 14, 1986. Address requests for reprints to: J. Thomas LaMont, M.D., University Hospital, setts 02118.
75 East
Newton
Street,
Boston,
Massachu-
This work was supported by grant AM34583 from the National Institutes of Health. Part of this work was published in abstract form in GASTROENTEROLOGY 1986;90:1671. The authors thank Anne DiSorbo for preparation of the manuscript, and Pat McAndrew and Luis Bustos-Fernandez for technical assistance. 0 1987 by the American Gastroenterological Association 0016.5085/87/$3,50
toxin A may contribute significantly to intestinal damage in C. difficile-associated colitis and diarrhea. Colitis and diarrhea secondary to Clostridium diffitile infection appear to be toxin-mediated. Toxin A, an enterotoxin, causes secretion of fluid in ligated rabbit ileal loops (1) and is lethal when injected into mice (2). The cellular action of toxin A is not known, but does not appear to involve stimulation of intestinal adenylate cyclase (3). Toxin B, a cytotoxin, causes rounding of tissue culture fibroblasts, which is preceded by depolymerization of actin-containing thin microfilaments (4) and an increase in the ratio of globular to filamentous actin in fibroblasts (5). Toxin B is present in the stools of most patients with antibiotic-associated pseudomembranous colitis (6,7) and also in a large percentage of those with nonspecific colitis and diarrhea (8,9). Clinical studies of toxin A have been limited by the lack of a simple, reproducible assay, but recent studies using immunoassay suggest that toxin A is frequently found in stool filtrates of patients with C. dificile colitis (10,ll). The purpose of our study was to determine the effects of purified toxins A and B on the rabbit ileum in vivo and in vitro. We used the perfused rabbit ileal loop model to assess epithelial permeability and morphologic damage after toxin exposure. Parallel studies were performed with rabbit ileal explants in short-term organ culture. Our results suggest that toxin A is an important factor in the pathogenesis of C. dificile enteritis. Materials
and Methods
Materials Male New Zealand white rabbits weighing -2 kg were used in all studies. Animals were fasted overnight before study, but were provided with water ad libitum. Protein concentration of toxin preparations was deter-
274
TRIADAFILOPOULOS
ET AL.
mined with a protein assay kit from Bio-Rad Laboratories (Pichmond, Calif.). Alkaline phosphatase was measured calorimetrically with a kit from Sigma Diagnostics (St. Louis, MO.). L-[4,5-3H]leucine, 130 Ci/mmol, was obtained from Amersham (U.K). IJ-[~-~H (N)]mannitol, 19.1 Ci/ mmol, and 2-[1,2-3H (N)]deoxyglucose, 37.3 Ci/mmol, were obtained from New England Nuclear [Boston, Mass.). Clostridium diffcile strain 10463, a highly toxigenic strain, was kindly supplied by Dr. Nadine Sullivan of the Department of Anaerobic Microbiology, Virginia Polytechnic Institute and State University, Blacksburg, Va. Methods Toxin preparation. Toxins A and B were prepared from culture supernatants of C. dificile strain 10463. Toxin B was purified to homogeneity (single protein band at 50 kilodaltons on sodium dodecyl sulfate-polyacrylamide gel electrophoresis) as previously described by us (5). Toxin A was partially purified by ammonium sulfate precipitation and diethylaminoethyl sepharose anion exchange chromatography by the method of Sullivan et al. (2). This preparation showed a single band on 10% polyacrylamide gel under nondenaturing conditions and a major band at 229 kilodaltons by sodium dodecyl sulfatepolyacrylamide gel electrophoresis, with several faint minor bands. Toxin B biplogic activity was assayed by the rounding of IMR-90 fibroblasts exposed to serial lo-fold dilutions as previously described (4). Typical preparations of toxin B used in these studies contained between 80 and 250 pg protein/ml as measured by Bio-rad assay (12). Cell rounding was completely inhibited by preincubation with goat antitoxin B (Virginia Polytechnic Institute and State University, Blacksburg, Va.). Enterotoxin activity was assayed in the ligated rabbit ileal loop assay (13). Enterotoxin preparations used in this study exhibited cytotoxicity in the fibroblasts assay at a 10m3 to 10m4 dilution, as previously reported by others (1,2). Rabbit intestinal pe$usion studies. After catheterization of an ear vein and induction of general anesthesia by injection of pentobarbital sodium (Nembutal, 50 mg/ml, Abbott, North Chicago, Ill.), the abdomen was opened by a midline incision and IO-cm-long segments of the distal ileum were ligated with silk to form ileal loops. For each animal, three IO-cm loops were formed with at least a 5-cm distance between loops. Each loop was injected with 1 ml of either toxin A (5-100 pg per loop), phosphate-buffered saline, or toxin Q (100 pg per loop). The abdomen was closed in two layers and animals were maintained under light anesthesia for 2-6 h. The abdomen was then reopened and both renal pedicles were ligated to prevent renal excretion of [3H]mannitol (14). An intravenous injection of 100 &i [3H]mannitol was given by ear vein. Each ileal loop was then separately cannulated at its proximal and distal ends with a 19F Foley-type catheter and connected to a perfusion pump (Harvard Instruments, Cambridge, Mass.). Continuous perfusion of the three ileal loops was initiated at a rate of 0.5 ml/min with Ringer’s lactate (6 g/L NaCl, 3.1 g/L sodium lactate, 0.3 g/L KCl, 0.2 g/L CaC12, pH 7.5) as the perfusion buffer. Every 15 min thereafter, aliquots of perfusate were collected and proc-
GASTROENTEROLOGY
Vol. 93, No. 2
essed for radioactivity and neutrophil number by hemocytometry. Throughout the 4-h perfusion period the animals were anesthetized with pentobarbital intravenously, and their body temperature was maintained at 37”-38°C with a heating lamp. Rabbits were killed at the end of each perfusion study with a bolus of pentobarbital. Fullthickness samples of ileal loops were fixed and processed for light and transmission electron microscopy. Morphologic studies. One-square centimeter samples of treated and control loops were placed in 10% phosphate-buffered formalin. After fixation, 1.0 x 0.2-cm strips were processed in a VIP-200 tissue processor (Miles Laboratories, Inc., Elkhart, Ind.) and embedded on edge in paraffin. Serial sections were studied with hematoxylin and eosin. Tissue for electron microscopy was fixed in half-strength Karnovsky’s solution for 4 h, washed in phosphate buffer, postfixed in 2% osmium tetroxide, dehydrated in acetone, and embedded in epon-araldite resin. One-micrometer sections were stained with Richardson’s blue and areas were selected for thin sectioning (60-80 nm). Sections were mounted on copper grids and viewed on a Philips E.M. 300 electron microscope (Philips Electronic Instruments, Mahwah, N.J.). Organ culture studies. We next studied the effect of toxin A on rabbit intestinal explants in short-term organ culture (15). For determination of protein synthesis, rabbit ileal explants measuring -2 x 2 mm were cut and incubated in sterile organ culture dishes with 1.5 ml of Trowell’s medium (Gibco Laboratories, Grand Island, N.Y.) containing 10% fetal calf serum; penicillin, 100 U/ml; streptomycin, 100 pgiml; and [3H]leucine, 10 &i/ml. Toxin A was diluted in medium and added in a volume of 0.15 ml; medium without added toxin was used as control. Incorporation of [3H]leucine into explant proteins during a 2-h incubation was determined as previously described (16), and was expressed as disintegrations per minute per milligram of explant protein. We also measured alkaline phosphatase in explants and medium after toxin A exposure as an indicator of membrane damage from toxin (17). Enzyme was measured using pnitrophenol phosphate as substrate (18). Results were expressed as Sigma enzyme units per milligram explant protein per 2 h, as estimated by the method of Lowry et al. (19). A Sigma unit is defined as the amount of enzyme that will release 1 pmol of p-nitrophenol in an hour. We also determined the effect of toxin A on epithelial membrane permeability in ileal explants by the method of Walum and Peterson (20). Explants were incubated with glucose-free Krebs’ solution (21) containing 5 &i/ml of [3H]2-deoxyglucose. After a 1-h incubation at 37”C, explants were washed three times with Krebs’ solution containing D-glucose, 1 mgiml, and then were transferred to sterile organ culture dishes and covered with 0.5 ml of Krebs’ buffer with glucose (37°C) containing either toxin A, Triton X-100 (5% volivol), or no addition [control). Every 15 min the medium was aspirated and counted in a scintillation counter, and explants were covered with 0.5 ml of fresh medium. After 2 h explants were removed and homogenized with 1 ml of Krebs’ buffer with 5% vol/vol Triton X-100. Representative explants from each group were fixed in formalin for light microscopy. Radioactivity
August
Figure
1987
1. Effect of toxins A and B on blood-to-lumen clearance of [3H]mannitol. Ligated rabbit ileal loops were injected with either toxin A, toxin B, or control for 6 h and then perfused with buffer for 4 h. Intestinal permeability to [3H]mannitol was estimated by scintillation counting of aliquots of perfusate collected every 15 min during a 4-h period as described in Methods. Eact point represents disintegrations per minute per 0.2 ml of perfusate.
in the explants
and aspirated medium was totaled and set at loo%, and percentages of radioactivity remaining in the monolayer were calculated for each time point as described by Walum and Peterson (20).
Results In Vivo Effects of Toxins A and B on 13H]Mannitol Permeability Preliminary experiments revealed that 100 pg of toxin A reproducibly caused secretion of a hemorrhagic fluid in closed rabbit ileal loops after a 16-h incubation, as defined by a volume/length ratio of 21.0 (22). The results of a typical perfusion experiment as described in Methods are shown in Figure 1, where the effects of toxins A and B (100 pg per loop) and buffer control are compared. Toxin A caused a striking increase in [“Hlmannitol permeability compared with toxin B or buffer control, which was apparent in the first 15-min collection after a 6-h exposure to toxin. The combined results of 15 ileal loop perfusions in 5 rabbits are shown in Figure 2. In all experiments toxin A caused at least a threefold increase in blood-to-lumen clearance of [3H]mannitol with equal concentrations of toxin B or buffer control. Because toxin A elicited secretion of a hemorrhagic, purulent exudate, we examined ileal loop perfusate for the presence of neutrophils. In toxin A-exposed loops we found a significant increase (p < 0.001) in neutrophils: 131 k 46/mm3 compared with toxin B-exposed loops, 10 ? 2/mm3, or buffer-exposed loops, 14 ? 5/mm3. In additional experiments we explored the minimal dosage of toxin A required to produce an effect
Clostridium
dificile
TOXINS
275
on [3H]mannitol permeability. We observed that 5 pug of purified toxin A was sufficient to produce a threefold increase in [3H]mannitol permeability after a 2-h incubation in a closed ileal loop (1274 dpm toxin A vs. 378 dpm control). Because toxins A and B are both secreted by toxigenic strains of C. dificile, we added mixtures of toxin A (5 pg) and toxin B (10 pg) to ascertain if they had an additive effect on epithelial permeability. In five separate experiments, mixtures of toxins A (5 pg) and B (10 pg) were not more potent than toxin A (5 pg) alone as determined by blood-to-lumen clearance of [3H]mannitol (not shown). To ascertain that toxin B was not undergoing inactivation in rabbit ileal loops, we tested the cytotoxicity titer of toxin B before and after a 2-h incubation in a IO-cm rabbit ileal loop. The toxin titer of lop7 to lOpa was unchanged after incubation in a closed loop for 2 h. Intestinal Morph01 ogy of Toxin-Exposed Perfused Ileal Loops Light microscopy of tissue samples from the perfused loops (Figure 3) revealed striking morphologic damage in the toxin A-exposed loops compared with toxin B or buffer. Normal ileal morphology was observed in loops exposed to phosphate-buffered saline or toxin B. Ileal loops exposed to 10 or 25 pug of toxin A showed pathologic findings that ranged from acute inflammation and reparative hyperplasia of villi, to necrosis of villi and transmural hemorrhagic necrosis. The earliest change seen was a polymorphonuclear infiltrate, most dense toward the villus tips and within the epithelium. The lumen contained numerous polymorphs, degenerating epithelial cells with nuclear pyknosis and karyorrhexis,
0 Toxin (N=5)
Toxin B (N=5)
Buffer (N=5)
Figure 2. Combined results of the effect of toxins A and B on blood-to-lumen clearance of [3H]mannitol. Ligated ileal loops (n = 15) from 5 rabbits were injected with either toxin A (100 pg), toxin B (100 pg), or control (buffer) for 6 h and then perfused with buffer for 4 h. Intestinal permeability to [3H]mannitol was estimated by scintillation counting of perfusate. Results are expressed as mean 2 SEM disintegrations per minute per milliliter perfusate for all collections (p < 0.001 vs. control).
276
TRIADAFILOPOULOS
Figure 3. Rabbit ileal loop Figure 1. At the microscopy. Left photomicrograph
GASTROENTEROLOGY Vol. 93, No. 2
ET AL.
morphology after toxin exposure. Rabbit ileal loops were injected with toxins and perfused as described in end of the perfusion period full-thickness samples of loops were fixed in formalin and processed for light photomicrograph shows a loop injected with buffer, middle photomicrograph with toxin B (100 pg), and right with toxin A (100 pg) (magnification x150).
and goblet cells. A more severe lesion was induced by exposure to 100 pg of toxin A per loop, consisting of necrosis of villi with marked reparative hyperplasia of the flat mucosal surface. The lamina propria surrounding the residual crypts was greatly expanded by red blood cells and abundant polymorphs. The luminal exudate and dehisced epithelial cells were more abundant. This severe lesion progressed in some experiments to complete necrosis of the mucosa, which was replaced by a neutrophilic exudate, enmeshing debris, and red blood cells. The most severe lesion consisted of hemorrhagic necrosis involving the full thickness of the bowel wall. Transmission electron microscopy of ileal loop mucosa subjected to 10 pg of toxin A demonstrated extensive dilatation of the endoplasmic reticulum of columnar cells, which gave the cytoplasm a cribriform appearance (Figure 4). Many cells also showed swelling of mitochondria. Organ Culture Experiments Organ culture experiments were carried out to ascertain if toxin A had any significant effect on rabbit ileum in vitro. Previous experiments in our laboratory (16) indicated that toxin B significantly inhibited [3H]leucine incorporation into hamster cecal explants after a 2-h exposure in organ culture. However, toxin A at 20 pg/ml had no effect on incorporation of [3H]leucine into ileal explants after a 2-h incubation (Table 1). Alkaline phosphatase levels in human jejunal explants were previously
used by Falchuk et al. (17) to monitor in vitro toxicity of gliadin. We therefore measured alkaline phosphatase activity in rabbit ileal explants exposed to toxin A. No significant effect of toxin A was observed on alkaline phosphatase concentration in explants or medium after a 2-h incubation. When toxin A-exposed explants were examined by light microscopy, no significant alteration in morphology was observed in comparison with explants incubated with buffer. Because of the profound effect of toxin A on epithelial permeability in rabbit ileum in vivo we examined the effect of toxin A on leakage of [“H]2deoxyglucose from rabbit ileal explants. This technique was previously validated as a sensitive assay of plasma membrane permeability after exposure to bacterial toxins (23). As shown in Figure 5, toxin A had no effect on leakage of [3H]2-deoxyglucose from rabbit ileal explants. As expected, Triton X-100 (5% vol/vol), a known membrane active detergent, caused increased release of [“H]2-deoxyglucose compared with control, and light microscopy revealed complete mucosal autolysis.
Discussion Our results indicate that C. dificile toxin A is a potent toxin in the rabbit ileum, eliciting severe intestinal inflammation and alteration of epithelial permeability. In contrast, toxin B in comparable doses had no demonstrable effect on [3H]mannitol permeability, migration of neutrophils into the lumen, or intestinal morphology. The absence of any
August 1987
Clostridium
diffcile
TOXINS
277
Figure 4. Electron microscopy of rabbit ileal loops after toxin exposure. Rabbit ileal loops were injected with buffer (left panel) or 10 pg toxin A (right panel) as described in Figure 1. At the end of the 4-h perfusion period the tissues were fixed and processed for electron microscopy. Final magnification X 22,500.
demonstrable effect of toxin A on ileal epithelial explants in organ culture compared with the profound effects observed in vivo suggests that toxin A triggers infiltration of the ileum with neutrophils that release inflammatory mediators, causing fluid secretion, hemorrhagic necrosis, and altered membrane permeability. This mechanism is also supported by our recent observation (24) that purified toxin A, but not toxin B, is a powerful chemotactic protein for human neutrophils in vitro and also causes a prompt rise in cytosolic calcium as measured by quin 2 fluorescence. We have also recently
Table
1. Effect of Toxin
A on
Protein Svnthesis and Alkaline PhosDhatase in Ileal Exulants
Incorporation of [3H]leucine (dpmimg) Toxin A (20 I.Lgiml) Control (no addition)
observed (25) that toxin A releases serotonin and p-hexosaminidase from cultured rat basophilic leukemia cells, which resemble intestinal mucosal mast cells as regards secretagogue responsiveness (26). Our results are in essential agreement with several previous reports regarding the effects of C. dificile toxin A on the intestine. Lyerly et al. (27) showed that 50 pg of toxin A elicited a viscous hemorrhagic fluid response in rabbit ligated ileal loops, whereas the same amount of toxin B produced a weak variable response without hemorrhage. The same investigators recently reported (28) that intragastric ad-
42,089 k 7,576 43,851 !z 8,038
Secreted proteins (dpm/mg) 4,238 * 2,114 3,807 rt 1,586
Alkaline phosphatase in explants (LVmg protein z h) 14.6 2 2.2 9.9 zt 0.5
Alkaline phosphatase in medium (Uimg protein . 2 h) 4.5 f 0.5 3.7 + 0.4
Rabbit intestinal explants were incubated with toxin A or no addition (control) for 2 h in medium with [3H]leucine for estimation of protein synthesis, and without [3H]leucine for estimation of alkaline phosphatase (n = 5 for each group). Incorporation of [3H]leucine into proteins and alkaline phosphatase levels in tissue and medium were determined as described in Methods. Results are expressed as mean 2 SEM disintegrations per minute per milligram explant protein for [3H]leucine incorporation, and as mean 2 SEM alkaline phosphatase units per milligram explant protein per 2 h. No statistical differences were noted between the toxin A and control groups.
278
TRIADAFILOPOULOS
+
GASTROENTEROLOGY Vol. 93, No. 2
ET AL.
40-
.? 8 ! e
20-
I!
Trmn
I
I
1
30
I
1
60
t
0
so
1
X-100
1
120
Minutes
Figure 5. Effect of toxin A on membrane permeability in vitro. Rabbit intestinal explants were preincubated with [3H]2-deoxyglucose for 1 h, then washed and incubated with toxin A (20 &ml), 5% (volivol) Triton X-100, or no addition (control) for 15 min. Release of radioactivity from the explants in the medium was measured at 15-min intervals and percentage remaining in explants was calculated as explained in Methods. Each point represents the mean of measurements on four explants.
ministration of toxin A to hamsters and mice caused accumulation of hemorrhagic fluid in the small intestine and cecum. Toxin B alone had no effect in their model, but when toxin A and B were administered together, a synergistic effect was observed. We were unable to elicit a synergistic effect of both toxins in our model; however, our results do not exclude a synergistic effect in natural or experimental C. dificile infection, which evolves over a much longer period of time. Mitchell et al. (29) likewise reported that purified toxin A, but not partially purified toxin B, caused increased permeability and fluid accumulation in rabbit ileum and colon. Taken together with our results, it would appear that toxin A by itself is capable of causing severe hemorrhagic enteritis in acute animal models, whereas toxin B alone has no or only minimal effect under the same experimental conditions. The intriguing observation by Lyerly et al. (28) that the toxins may act synergistically, and the observation that hemorrhagic colitis is not commonly seen in humans with antibioticassociated C. dificile infection (SO), suggest that the pathophysiology of naturally acquired infection may be different from these experimental models. The cellular mechanism by which toxin A damages the intestine is not known. Although our organ culture experiments showed no effect of toxin A on protein synthesis, alkaline phosphatase, or membrane permeability, we cannot exclude a direct effect of the toxin on the enterocyte. For example, Roth-
man et al. (31) demonstrated that high concentrations (1 mg/ml) of toxin A caused leakage of intracellular K+ and inhibition of protein synthesis in HeLa cells. Moreover, several preliminary reports suggest that toxin A (32) and crude C. diffcile culture filtrate (33) cause morphologic damage to stripped rabbit ileal mucosa in Ussing chambers, increased short circuit current, and increased net chloride secretion. In summary, highly purified toxin A, but not toxin B, of C. difficile elicited an intense inflammatory response in rabbit ileal loops which was accompanied by a striking change in epithelial permeability to [3H]mannitol. In short-term organ culture, toxin A had no effect on protein synthesis, alkaline phosphatase, or membrane permeability. We speculate that inflammatory mediators released by neutrophils, mast cells, or macrophages are likely to contribute to the morphologic and functional changes reported here.
References 1. Taylor NS, Thorne GM, Bartlett JG. Comparison of two toxins produced by Clostridium dificile. Infect Immun 1981;34: 1036-43. 2. Sullivan NM, Pellet S, Wilkins TD. Purification and characterization of toxins A and B of Clostridium dificile. Infect Immun 1982;35:1032-40. 3. Lonnroth I, Lange S. Toxin A of C. dificile: production, purification and effect in mouse intestine. Acta Path01 Microbiol Immunol Stand 1983;91:395-400. 4. Wedel N, Toselli P, Pothoulakis C, et al. Ultrastructural effects of Clostridium dificile toxin B on smooth muscle cells and fibroblasts. Exp Cell Res 1983;148:413-22. 5. Pothoulakis C, Barone LM, Ely R, et al. Purification and properties of Clostridium diffcile cytotoxin B. J Biol Chem 1986;261:1316-21. 6. George RH, Symonds JM, Dimock F. Identification of Clostridium dificile as a cause of pseudomembranous colitis. Br Med J 1978;1:695. 7. Willey SH, Bartlett JG. Cultures from Clostridium dificile in stools containing a cytotoxin neutralized by Clostridium sordellii antitoxin. J Clin Microbial 1979;10:880-7. 8. Chang TW, Lauermann M, Bartlett JG. Cytotoxicity assay in antibiotic-associated colitis. J Infect Dis 1979;140:765-9. 9. George WL, Rolfe RD, Finegold SN. Clostridium dificile and its cytotoxin in feces of patients with antimicrobial agentassociated diarrhea and miscellaneous conditions. J Clin Microbial 1982;15:1049-53. im10. Lyerly DM, Sullivan NM, Wilkins TD. Enzyme-linked munoabsorbent assay for Clostridium diffcile toxin A. J Clin Microbial 1983;17:72-8. 11. Laughon BE, Viscidi RP, Gdovin SL, Yolken RI-I, Bartlett JG. Enzyme immunoassays for detection of Clostridium dificile toxins A and B in fecal specimens. J Infect Dis 1984;149: 781-8. 12. Bradford
M. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein dye binding. Anal Biochem 1976;72:248-54. 13. De SN, Chatterje DN. An experimental study on the mechanism of action of Vibrio cholerae on the intestinal mucus membrane. J Path01 Bacterial 1953;66:559-62.
Clostridium
August 1987
14. Loehry
CA, Axon ATR, Hilton PJ, Hider RC, Creamer B. Permeability of the small intestine to substances of different molecular weight. Gut 1970;11:466-70. 15. LaMont JT, Turner BS, DiBenedetto D, Handin R, Schaffer AI. Arachidonic acid stimulates mucin secretion in prairie dog gallbladder. Am J Physiol 1983;245:G92-8. 16. Pothoulakis C, Triadafilopoulos G, Clark M, Franzblau C, LaMont JT. Clostridium diffcile cytotoxin inhibits protein synthesis in fibroblasts and intestinal mucosa. Gastroenterology 1986;91:1147-53. 17. Falchuk ZM, Gebhard RL, Sessoms C, Strober W. An in vitro model of gluten-sensitive enteropathy. Effect of gliandin on intestinal epithelial cells of patients with gluten-sensitive enteropathy in organ culture. J Clin Invest 1974;53:487-500. 18. Bessey OA, Lowry OH, Brock MJ. A method for the rapid determination of alkaline phosphatase with five cubic millimeters of serum. J Biol Chem 1946:321-g. 19. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Protein measurement with the Folin phenol reagent. J Biol Chem 1951;193:265-75. 20. Walum E, Peterson A. Tritiated 2-deoxy-o-glucose as a probe for cell membrane permeability studies. Anal Biochem 1982;120:8-11. 21. LaMorte WW, Hingston SJ, Wise WE. pH dependent activity of H,- and Hz-histamine receptors in guinea pig gallbladders. J Pharmacol Exp Ther 1981;217:63040. 22. Duncan CL, Sugiyama H, Strong DH. Rabbit ileal loop response to strains of Clostridium perfringens. J Bacterial 1968;95:1560-6. 23. Thelestam M, Mollby R. Sensitive assay for detection of
24.
25.
26.
27.
28.
29.
30. 31.
32.
33.
diffcile
TOXINS
279
toxin-induced damage to the cytoplasmic membrane of human diploid fibroblasts. Infect Immun 1975;12:225-32. Pothoulakis C, Sullivan DA, Melnick AJ, Gadenne T, Meshulam T, LaMont JT. Clostridium dificile toxins A and B stimulate intracellular calcium release in human neutrophils. Clin Res 1986;34:53OA. Pothoulakis C, Theoharides T, McAndrew PE, LaMont JT. C. difficile toxin A releases mediators from mucosal-like rat basophilic leukemia (RBL) cells. Clin Res 1986;34:871A. Seldin DC, Adelman S, Austen KF, et al. Homology of the rat basophilic leukemia cell and the rat mucosa magt cell. Proc Nat1 Acad Sci (USA) 1985;82:3871-5. Lyerly DM, Lockwood PE, Richardson SH, Wilkins TD. Biological activities of toxins A and B of Clostridiuq dificile. Infect Immun 1982;35:1147-50. Lyerly DM, Saum KE, MacDonald DK, Wilkins TD. Effects of C. diffcile toxins given intragastrically to animals. Infect Immun 1985;47:349-52. Mitchell TJ, Ketley JM, Haslam SC, et al. Effect of toxin A and B of Clostridium dificile on rabbit ileum and colon. Gut 1986;27:78-85. Trnka YM, LaMont JT. Clostridium dijj?cile colitis. Adv Intern Med 1984;29:85-107. Rothman SW, Brown JE, Diecidue A, Foret DA. Differential cytotoxic effects of toxins A and B isolated from Clostridium dificile. Infect Immun 1984;46:324-31. Stephen J, Redmond SC, Mitchell TJ, et al. Clostridium difFcile enterotoxin (toxin A): new results. Biochem Sot Trans 1984;12:194-5. Hughes S, Warhurst G, Turnberg LA, Higgs NB, Gingliano LG, Drasar BS. Clostridium dificile toxin-induced intestinal secretion in rabbit ileum in vitro. Gut 1983;24:94-8.