Its Disease and Toxins

Transcription

1 CLINICAL MICROBIOLOGY REVIEWS, Jan. 1988, p Vol. 1, No /88/ $02.00/0 Copyright 1988, American Society for Microbiology Clostridium difficile: Its Disease and Toxins DAVID M. LYERLY,* HOWARD C. KRIVAN, AND TRACY D. WILKINS Department of Anaerobic Microbiology, Virginia Polytechnic Institute and State University, Blacksburg, Virginia INTRODUCTION... 1 HISTORICAL PERSPECTIVES... 2 C. DIFFICILE DISEASE IN EXPERIMENTAL ANIMALS... 2 C. DIFFICILE DISEASE IN HUMANS... 3 CLINICAL DIAGNOSIS Pathology... 4 Isolation and Identification of the Organism... 4 Detection of Toxin... 5 Problems Associated with Clinical Diagnosis of the Disease... 6 TOXINS A AND B Role of Toxins A and B in PMC... 7 Factors Affecting Toxin Production... 7 Purification... 7 Physicochemical Properties... 9 Biologic Activities... 9 Immunochemical Properties Toxin Receptors Cloning of the Toxin A Gene OTHER VIRULENCE FACTORS PROPERTIES OF C. SORDELLII ENTEROTOXIN AND CYTOTOXIN...12 FUTURE DIRECTIONS ACKNOWLEDGMENTS LITERATURE CITED INTRODUCTION In the 1960s, anaerobic bacteria began to be recognized as the predominant organisms in the human large bowel and their roles as agents of disease in humans were established. Escherichia coli was removed from its false position as the most numerous organism in the human colon, and many "sterile abscesses" were found to be caused by anaerobic bacteria. Most antibiotics used at that time to sterilize the bowel were quite active against facultative bacteria but had little effect on the anaerobic population. For example, some of the more commonly used antibiotics such as kanamycin and other aminoglycosides were not active on anaerobes. As a result, pharmaceutical companies began to search for agents with improved activity against anaerobes. One of the early successes was clindamycin. This antibiotic was a derivative of a successful old-line antibiotic, lincomycin, but it was much more active against anaerobes and reached the colon via excretion in the bile. In the early 1970s, clindamycin began to be used for treating serious anaerobic infections, and it proved to be a significant improvement over previous types of therapy. Unfortunately, reports on the deaths of patients treated with clindamycin appeared as the antibiotic became widely used (90, 106, 190, 222, 226, 227). Diarrhea had been a common side effect of therapy with the parent compound, lincomycin. Patients treated with clindamycin, however, sometimes progressed beyond diarrhea to a severe inflammation of the colonic mucosa with the elaboration of pseudomembranes composed of fibrin, mu- * Corresponding author. 1 cous, necrotic epithelial cells, and leukocytes. In some patients this pseudomembrane formed a sheath over the entire colonic mucosa. Such sheaths were seen most commonly at autopsy, and the disease, pseudomembranous colitis (PMC), was thought to have a high mortality rate. The incidence of PMC varied widely between different hospitals and even between different wards in the same hospital. Some investigators reported rates as high as 10% in patients treated with clindamycin (227), but most later studies reported much lower rates (90, 190). The higher incidence in some studies probably resulted from transmission within the hospital prior to the recognition that this was a nosocomial infection. The increase in the number of cases of PMC in patients treated with clindamycin is the factor which stimulated interest in the disease, but it soon became apparent that patients treated with other antibiotics developed this disease. The early designation for the disease, "clindamycin colitis," thus was inappropriate, much to the relief of The Upjohn Company, who marketed the drug. PMC created quite a stir among investigators because it represented a "new" disease that was caused by the treatment given for other diseases and because it killed patients. PMC actually had been described in patients prior to the start of the antibiotic era, but the number of cases increased dramatically after antibiotics began to be used (86). It was speculated up until the late 1970s that the disease might be caused by Staphylococcus aureus. This organism, however, was not the cause of the numerous cases of PMC reported in the 1970s and 1980s, and there are doubts as to whether S. aureus actually has caused any cases of the disease. It has

2 2 LYERLY ET AL. now been established that the disease is almost always caused by Clostridium difficile. C. difficile produces two potent lethal toxins, an enterotoxin (toxin A) and a cytotoxin (toxin B) (12, 13, 220, 224). It is the production of these toxins by C. difficile in the gut which ultimately leads to the disease. As a result, much effort has been made to develop detection methods for these toxins and to learn how they act. In the first portion of this review, we describe some of the clinical aspects of PMC and evaluate some of the methods which have been used to detect the toxins of C. difficile in clinical specimens. The second portion of the article covers the biochemical information available on the toxins and discusses their unusual characteristics. HISTORICAL PERSPECTIVES The disease PMC has been recognized for quite some time. One of the earliest descriptions of the disease was reported in 1893 by Finney (71), who noted diphtheritic pseudomembranes and hemorrhagic diarrhea in a young woman following surgery. The number of reported cases of PMC increased dramatically, however, following the widespread use of broad-spectrum antibiotics. The causative agent of PMC, C. difficile, also has been recognized for some time. The organism was initially isolated from the stools of healthy newborn infants by Hall and O'Toole in 1935 (93). These investigators referred to the organism as Bacillus difficilis because of the difficulty they encountered in the isolation of the organism. These investigators were also the first to show that the organism is toxigenic. This observation was based on the finding that broth cultures and culture filtrates of the organism caused lesions, respiratory arrest, and death when injected into rabbits and guinea pigs. They proposed the interesting idea that the toxin was a neurotoxin because it caused convulsions in animals. C. difficile was not known to be a pathogen so the "toxin" of the organism was not studied in detail until the late 1970s, when the association of C. difficile with PMC became apparent. At that time, it was shown that stools from patients with PMC contained high levels of cytotoxic activity. Initially, it was suspected that the activity was due to viruses or perhaps mycoplasma, but tests for these agents were consistently negative. Investigators then began to suspect a toxin-producing bacterium, and the competition to identify the etiologic agent became even more intense. Results from several laboratories showed that the cytotoxic activity was neutralized by gas gangrene antiserum. Gas gangrene is a disease caused by several species of clostridia, and the antitoxin is a mixture of antisera made to crude toxin preparations from several clostridial species. When the individual antisera were tested, only the antiserum against Clostridium sordellii neutralized the cytotoxicity. This led investigators to believe that this organism might be causing PMC. C. sordellii, unfortunately, was almost never isolated from patients with the disease. Soon it was demonstrated that C. difficile was present in high numbers in patients with PMC and that the organism produces a cytotoxin which is neutralized by C. sordellii antiserum (17, 19, 20, 79, 82, , 198). We now know that C. difficile and C. sordellii produce toxins that are almost identical. Once these findings were reported, researchers began to provide evidence documenting the role of C. difficile in the pathogenesis of PMC. One of the key contributions was the development of a suitable animal model for studying the CLIN. MICROBIOL. REV. disease. It had been shown by Small in 1968 that hamsters injected with lincomycin developed severe enterocolitis and died (214), but these observations were not extended to antibiotic-associated disease in humans. In the late 1970s, it was shown that hamsters that died from cecitis after treatment with antibiotics contained high numbers of toxigenic C. difficile in their stools and that the toxin in the stools of these animals was very similar to the toxin found in stools from PMC patients (6, 18, 22, 196, 197). On the basis of these findings, the hamster model began to be used extensively in the study of the disease and is still in use today. Until 1980, it was believed that the cytotoxic factor was the only toxin produced by C. difficile. There were reports describing enterotoxic activity in fecal extracts from hamsters with antibiotic-associated cecitis (99, 197), but there was no indication that this activity and the potent cytotoxic activity represented two different toxins. Consequently, much of the early literature is rather confusing because some groups were working primarily with the cytotoxin whereas others were working mostly with the enterotoxin. Then, within a short time, Taylor et al. (224; N. S. Taylor, G. M. Thorne, and J. G. Bartlett, Clin. Res. 28:285, 1980) and Banno et al. (13) demonstrated that the cytotoxic activity could be separated from the enterotoxic activity by anionexchange chromatography. These findings, along with results from other laboratories, provided conclusive evidence that the organism produces at least two distinct toxins. The cytotoxin (i.e., the toxin detected in fecal specimens) is referred to as toxin B. The enterotoxin (i.e., the toxin responsible for the fluid response in the animal ileal loop model) is referred to as toxin A. The designations A and B refer to the elution pattern of the toxins on anion-exchange resins. Toxin A binds less tightly to the resin than toxin B and elutes before toxin B. The designations toxin D-1 and toxin D-2 have been used to denote the enterotoxin and cytotoxin, respectively, in the articles by Banno et al. (12, 13). C. DIFFICILE DISEASE IN EXPERIMENTAL ANIMALS C. difficile causes antibiotic-associated disease in a number of animal species, including hamsters, guinea pigs, and rabbits, but most of the work has been done with hamsters (1, 44, 51, 53, 68, 100, 104, 116, 141, 166, 168, 188, 192, 193, 219, 245). As with PMC in humans, the hamster disease can be initiated with a variety of antibiotics (1, 44, 68, 141, 188). If toxigenic strains of the organism are present in the environment, the animal usually develops diarrhea and dies from severe enterocolitis. The disease in hamsters is localized in the cecum (proximal colon) with some involvement of the ileum. PMC in humans, on the other hand, occurs primarily in the distal colon. Coverings of inflammatory debris consisting of erythrocytes, inflammatory cells, bacteria, and sloughed mucosal cells are sometimes present in the ceca of hamsters with the disease, and these coverings are very similar to the pseudomembranes observed in the colon of many PMC patients. The hamster model did not meet with universal acceptance; some pathologists argued that the differences between the hamster and human disease meant that there was a different cause. Fortunately, the investigators working with hamsters did not believe this. One of the most interesting observations with C. difficile infections in humans is that infants can have high numbers of toxigenic C. difficile and high levels of toxins A and B in their stools and be completely asymptomatic (see next section).

3 VOL. 1, 1988 This unusual phenomenon is also seen in infant hamsters (88, 101, 203). The reasons for this are not known, but the results obtained with this model should provide some insight into why most human infants colonized with toxigenic C. difficile do not become ill. In addition to the results obtained with the conventional hamster model described above, studies on the disease have been done in germfree mice and rats (51, 53, 166, 219, 245). Neither of these animal species is as susceptible to the disease as hamsters. Gnotobiotic animals of each species can, however, be colonized with the organism. These monoassociated animals develop some intestinal pathology, but compared with conventional hamsters, only a low percentage die from the infection. Unlike the disease in hamsters, the disease in germfree mice and rats involves the colon, and pseudomembranes are evident along the colonic wall in the infected animals. The disease in these animals develops more slowly than that in hamsters, and perhaps the chronic nature of the disease results in the production of the pseudomembranes. In this regard, it has been suggested that these animal models may be better suited than the hamster model for the study of PMC. However, these models are considerably more expensive to use and the fact that no other organisms are present makes the situation unlike PMC in humans. Gnotobiotic hamsters would be interesting to study, but unfortunately hamsters are almost impossible to raise in the germfree state. C. DIFFICILE DISEASE IN HUMANS The normal bacterial flora in the gut serves as the major barrier against colonization by pathogens. When the flora is disturbed in some manner, the host becomes susceptible to colonization or overgrowth by the pathogen. Antibiotics, the primary predisposing agent of PMC, act in this fashion by upsetting the normal flora of the bowel. The numerous reports of PMC which appeared following the use of clindamycin led investigators to believe that this antibiotic was the only one responsible for causing the disease. Actually, ampicillin and cephalosporins cause more cases than clindamycin (177, 202, 213). These antibiotics are widely prescribed, and that is why they are associated with PMC more often than other antibiotics. We now know that almost any antibiotic can cause the disease (14, 25, 32, 39, 49, 55, 56, 76, 80, 189, 213, 231) and that very rarely the disease can even occur without prior antibiotic therapy (63, 157, 170, 239). In addition, treatment with antineoplastic agents which have antibacterial activity can cause the disease (52). Most cases of PMC result from nosocomial infections. This is apparent from the increasing number of reports describing outbreaks of the disease and from the fact that C. difficile is part of the normal flora in only a low percentage of healthy adults (57, 91, 113, 115, 122, 151, , 183). The clustering of cases in hospitals and even within hospital wards is the reason why such a wide range of incidence of the disease (0.01 to 10%) has been reported (90, 190, 227). PMC cases in hospitals can be especially difficult to control because these patients have diarrhea. This mode of transmission increases the release of the organism into the environment, and the organism can be isolated from the clothing and room fixtures of the patient. In addition, the organism is transmitted by hospital personnel. Once in the environment, the organism can persist for months since it produces spores (115, 161). Efforts should be made to isolate patients and minimize cross-contamination between patients. Compromised elderly patients undergoing antibiotic C. DIFFICILE TOXINS 3 therapy are especially at risk. The organism is present in hospitals not only from patients but also because it is present as "normal" flora of the infants in many hospital nurseries. Infants, for some unknown reason, are refractory to PMC even though they carry high numbers of the organism and high levels of toxins A and B in their stools (5, 27, 30, 31, 50, 58, 89, 98, 124, 136, 152, 195, 212, 216, 221, 233, ). In fact, it has been estimated that 50% or higher of infants are colonized with toxigenic C. difficile and are asymptomatic. The rate of colonization of infants parallels the incidence of outbreaks and cross-infections within hospitals, indicating that most infants acquire the organism nosocomially (30, 176). That so many infants are colonized with toxigenic C. difficile and that so few develop the disease indicate that the manner in which they are protected is quite effective; otherwise, more infants would die from the infection. No one knows why infants are protected, but there are a number of hypotheses. The protection may well result from a combination of several factors. Colostrum contains substances (perhaps secretary antibody) which neutralize toxins A and B (114, 238), and these substances probably help to protect the infant from the toxins, but it is more complicated than this since infants who are not breast-fed do not get the disease. Fetal intestinal cells are reported to be much less sensitive to the toxins than adult intestinal cells (46), and this may contribute to the resistance. Another hypothesis is that infants may lack the toxin receptors in their intestines. The receptor to which toxin A binds (the Galal-3Gal,1-4GlcNAc trisaccharide; see subsection, "Toxin Receptors") is similar to the I,i antigens which are developmentally regulated on erythrocytes. The carbohydrate chains on infant intestinal cells may exist in an immature form which is not recognized by toxin A. The carbohydrate sequences would subsequently develop into the active receptor containing the Galal-3GalI1-4GlcNAc trisaccharide in adults. Alternatively, the receptors on the infant cells may be covered by a thicker layer of mucin than in adults, and this could prevent the toxin from binding to its receptor. Determining the exact mechanism of infant resistance is one of the more exciting research areas, but we should not assume that, because many asymptomatic infants have the toxin present, infants are not affected. Infants who are compromised by other disease or who have had intestinal surgery may be at high risk (2). Cystic fibrosis patients who are colonized with toxigenic C. difficile also are refractory to the development of PMC (169, 241, 249). It is not clear whether these patients are asymptomatic because they are colonized with nontoxigenic strains or whether the manner in which the toxins are produced and act in these patients is different from the manner in which the disease develops in other adults. It has been suggested that the intestinal environment in these patients may be similar to the environment in the infant intestine and that the manner of protection is similar in these two populations, but this has not been demonstrated. Although the role of C. difficile in PMC has been clearly established, its role in other less severe gastrointestinal illnesses is not as well defined. C. difficile and its toxins are found in patients with symptoms ranging from simple diarrhea all the way to PMC, but the role of the organism in antibiotic-associated diarrhea still is not understood. Even when the toxins are present, they may not be causing the symptoms, but currently patients with antibiotic-associated diarrhea who test positive for C. difficile toxins usually are considered to have C. difficile disease. It has been estimated that the organism may cause approximately 25% of the

4 4 LYERLY ET AL. reported cases of antibiotic-associated diarrhea and that the number of cases of C. difficile diarrhea is second only to the number caused by Campylobacterjejuni, which is believed to be the most frequent cause of bacterial diarrhea (38, 85). More clinical information is needed, however, to get a better understanding of the importance of the organism in diarrheic patients. In addition to its involvement in gastrointestinal illnesses, C. difficile can cause disease in other parts of the body, including abscesses, wound infections, osteomyelitis, pleuritis, peritonitis, septicemia, and urogenital tract infections, but these are rare (92, 133, 207). There was an initial report that C. difficile was a common isolate from urogenital infections, but this has been refuted. The organism has been isolated from neonates with necrotizing enterocolitis and from patients with inflammatory bowel disease and Crohn's disease, but it appears that the organism does not play an active role in these diseases except as a complication (29, 43, 59, 87, 94, 109, 123, 155, 215, 234). A number of methods have been used to treat patients with PMC. In many instances, just the discontinuation of the offending antibiotic is sufficient to resolve the symptoms. The most common method of treating the disease, ironically, is by the use of antibiotics. Vancomycin was one of the initial antibiotics used for the treatment of PMC, based on the findings that: (i) vancomycin was effective against grampositive bacteria and had been used for "staphylococcal colitis" (107); (ii) clinical isolates of C. difficile were susceptible in vitro to the antibiotic (42, 61, 67); (iii) vancomycin was effective against C. difficile disease in hamsters (21, 40); and (iv) very high concentrations could be achieved in the intestine because it was not absorbed. PMC patients treated with vancomycin showed prompt improvement, and this is the most common treatment used in the United States. Vancomycin is expensive, and the high doses used in the early trials have now been replaced by lower doses that are equally as effective. Even the lower doses are more expensive than some other antimicrobial agents that have been used extensively for treatment in European countries. Metronidazole is widely used in Europe for treatment of anaerobic infections and is an effective method of treatment for PMC (28, 225). Its advantage over vancomycin is its lower cost. Bacitracin also has been used successfully (60, 252). Any antibiotic that kills C. difficile should be effective if concentrations above the inhibitory concentration can be achieved in the colon. These same antibiotics can cause the disease when the concentrations in the colon are not above the inhibitory level, and they can cause the disease when therapy is stopped and the concentrations decline. Thus, vancomycin and metronidazole can cause the disease as well as cure it. Many patients treated for PMC or minor diarrhea relapse several days after they are taken off the antibiotic used for therapy. Relapse rates of 20% are reported for vancomycin and some patients "relapse" repeatedly (15, 16, 24) Ṫreatment with anion-exchange resins such as cholestyramine and cholestipol has been used with some success. The resin binds the toxins produced by the organism and minimizes the tissue damage which occurs during the disease. The alleviation of the symptoms, however, is more variable compared with the use of antibiotics, and some patients do not respond at all to this type of treatment (117, 228). Treatment of an "antibiotic-associated" disease with antibiotics obviously is not an ideal solution. Another form of treatment currently being examined is the reestablishment of the gut flora, using normal fecal flora. This approach has been successful in treating hamsters with antibiotic-associated enterocolitis and has recently been shown to be effective in treating PMC in humans (33, 34, 37, 210, 243, 246). Another approach is to give patients nontoxigenic strains of C. difficile so that the toxigenic strains cannot get established in the bowel. This has also been used successfully in hamsters and is now being tried in human patients. The yeast Saccharomyces boulardii has been used successfully to protect hamsters from clindamycin-associated enterocolitis and relapse following vancomycin treatment (64, 154, 232), but clinical trials in PMC patients have not been reported. Certain prostaglandins protect the stomach and small intestine from mucosal necrosis caused by harsh agents such as alkaline and acidic solutions and alcohol. The manner by which this protection occurs is not clearly understood, but these findings led investigators to test whether these agents would protect hamsters against the extensive intestinal mucosal damage which occurs in clindamycin-associated cecitis. The results show that prostaglandins (PG), 16,16-dimethyl-PGE2, 15(R)-15-methyl-PGE2, and 2-acetyl-2-decarboxy-15(S)-15-methyl-PGF2a, as well as castor oil and vegetable oil (which contain the prostaglandin precursor linoleic acid), prevent the disease in hamsters (201, 213). It has been suggested that the prostaglandins may affect the release of the toxins by the organism (205), but the protection also may result from the effect of the prostaglandins on the tissue. These findings are exciting since the protection observed with prostaglandins is the most dramatic protection yet observed with pharmacologic agents. CLINICAL DIAGNOSIS CLIN. MICROBIOL. REV. Pathology The most definitive diagnosis of PMC is by the endoscopic detection of pseudomembranes or microabscesses in antibiotic-treated patients with diarrhea who have C. difficile toxin in their stools. The diarrhea may be watery or bloody and may be accompanied by abdominal cramps, leucocytosis, and fever (70, 81, 228). Often the microabscesses and plaques are isolated in only certain parts of the colonic wall, with other areas remaining normal in appearance. Why certain areas are affected while other adjacent areas are not is unknown. The diagnosis of C. difficile disease in patients who develop antibiotic-associated diarrhea is considerably more difficult. In these cases, toxigenic C. difficile actually may become established in the patient as a result of the diarrhea and play no role in the disease (see subsection, "Problems Associated with Clinical Diagnosis of C. difficile Disease"). Isolation and Identification of the Organism The isolation of toxigenic C. difficile from the stool has been used for the presumptive diagnosis of PMC because patients with the disease contain high numbers (107 or greater) of C. difficile in their stools (23). The most widely used medium for the isolation of C. difficile is the cycloserine-cefoxitin-fructose-egg yolk medium developed by George et al. (83). The medium serves as a selective and differential medium for C. difficile and reportedly can detect as few as 2,000 organisms in a total count of 6 x 1010 bacteria per g (wet weight) of feces. Several modifications of the medium which affect the isolation rate of the organism have been described. For example, the incorporation of sodium taurocholate in place

5 VOL. 1, 1988 C. DIFFICILE TOXINS 5 TABLE 1. Evaluation of clinical tests for detection of C. difficile and its toxins Test Antigen detected Sensitivity Evaluation Tissue culture assay Toxin B 1 pg (50 pg/ml) The test is based on the detection of cytotoxic activity in stool specimens as noted by rounding of the tissue culture cells and neutralization of the activity by C. djficile of C. sordelifi antitoxin. The assay shows a good correlation with the disease and serves as the "gold standard" by which to measure other tests for the disease. The test is extremely sensitive, with its primary disadvantages being the assay time and expense. Counterimmunoelec- Not specific for 10 ng of antigen The test is not recommended as a clinical assay for C. difficile or its toxtrophoresis Tox+ isolates (0.5 pug/ml) ins. The C. difficile antiserum which has been used in the assay in clinical trials cross-reacts with C. sordellii and C. bifermentans and gives false-positive reactions. ELISA Toxin A or B 0.1 to 1 ng The ELISA is still in the research and development stage. The toxin A depending on (1-10 ng/ml) ELISA is close to the sensitivity needed for the test but it needs to be specific anti- shortened to <1 h to make it more suitable as a clinical test. body Latex agglutination Not specific for Not reported The Culturette Brand Rapid Latex Test for C. difficile has created much test Tox+ isolates excitement because it is rapid (ca. 3 min) and simple. The assay is not specific for Tox+ isolates; it reacts with Tox- isolates as well as C. sporogenes, P. anaerobius, and B. asaccharolyticus. of the egg yolk apparently improves the recovery rate of the organism (41, 165, 244). In addition, the concentrations of cycloserine and cefoxitin and the type of basal medium affect the recovery rate and the phenotypic properties (e.g., the intensity of the fluorescence) exhibited by the organism (130, 131). Some of the unusual metabolic products (e.g., p-cresol and isocaproic acid) produced by C. difficile have been suggested as markers for the organism (54, 84, 129, 134, 150, 171, 175, 182). The detection of these substances is not specific enough to be used as a marker for clinical diagnosis. Several microsystems, the API ZYM Micro-System, (API Laboratory Products, Ltd.), the Minitek Anaerobe II (BBL Microbiology Systems), the API An-Ident System (Analytab Products), and the RapID ANA System (Innovative Diagnostic Systems) are now available for the identification of anaerobic bacteria. The organism must be isolated initially, however, before it can be identified by these test systems. These systems have only recently become available, and their specificity for C. difficile has not been critically evaluated (4, 26, 121, 132, 153). Detection of Toxin A description of the various clinical assays described for the toxins of C. difficile is given in Table 1. The tissue culture assay has been used extensively for the detection of C. difficile toxin in stool specimens. This is based on the finding that >90% of PMC patients have cytotoxic activity in their stools (14, 19, 23, 66). The cytotoxic titer does not closely correlate with the severity of the disease, although there is some indication that the titer is higher in severe cases of PMC. In the test, dilutions of the stool specimen are added to tissue culture cells and a positive reaction is noted by rounding of the cells. Many stool specimens contain other toxins or nonspecific interfering substances which cause rounding of the cells; therefore, the specificity must be confirmed by neutralizing the cytotoxic activity with antiserum against C. difficile or C. sordelifi (62). Both toxins A and B have cytotoxic activity and are present in stool specimens from PMC patients. Toxin B, however, is a much more potent cytotoxin than toxin A and masks the weak cytotoxic activity of toxin A unless the toxins are first separated (146). In addition, the toxins are coproduced in complex media and no toxa+/toxb isolates have been identified. Therefore, the cytotoxic activity detected in the stool specimens of PMC patients is due to toxin B. The major advantage of the tissue culture assay for the detection of toxigenic C. difficile is its extreme sensitivity; 1 pg of toxin B is sufficient to cause rounding of the cells. This is the reason the test correlates well with the disease and why the test represents the "gold standard" for C. difficile toxin tests. There are several possible problems, however, with the assay as it is currently used. Toxin B, as well as toxin A, is active against a variety of mammalian cells; in fact, no cell lines have been described which are not sensitive to the toxins, although some cell lines are less sensitive than others. As a result, clinical laboratories use different cell lines, making- it difficult to compare clinical studies. In addition, the assay procedure has not been standardized. There are variations in the dilutions of stool specimens tested, the ahtiserum used, and the time of recording the results, and this inhibits the direct comparison of clinical data. Another problem is that tissue culture assays require specialized equipment and personnel, making the assay expensive. Recently, the Tox-Titer microtiter plate system (Bartels Immunodiagnostic Supplies, Inc.), which uses human foreskin cells, became available for the detection of C. difficile toxin (11, 162, 248). The system is easy to use and allows persons not equipped for standard tissue culture to utilize the assay, but it also is expensive. A number of antibody-based tests have been described. Counterimmunoelectrophoresis was the first test of this type to be used for the detection of C. difficile, primarily because it is considerably faster than tissue culture assays (1.5 h compared to overnight). However, the antiserum used in clinical trials was prepared against culture filtrates of C. difficile and the immunoprecipitin bands detected in most stool specimens do not represent the toxins. The amount of toxin necessary to give a band is in the range of 10 to 100 ng, indicating that concentrations of micrograms per gram still are needed. Most stool specimens from PMC patients do not have this concentration of toxin. Nontoxigenic strains of C. difficile, as well as strains of C. sordellii and Clostridium

6 6 LYERLY ET AL. CLIN. MICROBIOL. REV. bifermentans, react with the antiserum and give false-positive reactions in the assay (102, 103, 135, 184, 208, 242, 247). The cross-reactions are most likely due to the cell surface antigens shared by these species ( ). Some preliminary work has been done with monospecific toxin antibody in counterimmunoelectrophoresis tests, but the antibody does not provide the sensitivity needed for a clinical test (144). Several variations of microtiter plate enzyme-linked immunosorbent assays (ELISA) have been developed for the detection of toxins A and B (10, 118, 128, 144, 147, 240). These tests are based on the detection of either toxin A or B with specific toxin antibody. At the present time, these tests are still in the research and development stage. A clinical ELISA which detects either toxin A or toxin B or both would be suitable since both toxins are present in the stools of PMC patients. The toxin A ELISA, however, looks more promising for several reasons. Toxin A is a better "marker" antigen for toxigenic C. difficile since it is more stable than toxin B. In addition, it is easier to purify, making it easier to obtain toxin A antibodies. Stool specimens from patients with PMC almost always have cytotoxic titers (toxin B) of 102 or greater. Specimens with cytotoxic titers of 102 have a toxin A concentration of about 1 nglml, so the ELISA needs to be capable of detecting this concentration of toxin A. The toxin A ELISA tests which have been developed approach this level of sensitivity, but they require at least 4 to 5 h to complete. The incubation times can be shortened and direct instead of indirect procedures can be used, but these modifications decrease the sensitivity of the test. Latex agglutination tests have proved to be very popular because they are rapid and simple. For these reasons, many clinicians have been interested in the Culturette Brand Rapid Latex Test for C. difficile (Marion Scientific, Inc.). The test was initially marketed for the detection of toxin A in stool specimens and consisted of a latex reagent supposedly coated with the immunoglobulin G fraction of monospecific antiserum against toxin A. The development of the test was based on the characterization studies done by Banno et al. (12, 13) on toxin A. Our initial studies, however, showed that our highly purified toxin A preparations did not react in the test (148). The reasons for these findings were puzzling since the properties of toxin A reported by Banno et al. were almost identical to the properties we observed with toxin A. We now suspect that the toxin A used to prepare the antiserum for the commercial test contained low amounts of a contaminating protein (the "latex-reactive" antigen) which is highly antigenic. When the toxin A preparation was injected into animals, antibody against the contaminating antigen was produced. Latex reagent coated with this antiserum was most likely used to screen antigen fractions when additional antiserum was prepared, resulting in the selection of fractions containing the contaminating antigen. Each time the cycle was repeated they would have had less toxin and more of the contaminant. Nontoxigenic strains of C. difficile produce high amounts of the latex-reactive antigen and react strongly in the test. The latex-reactive antigen is quite distinct from toxin A. It is smaller in size and binds more tightly to anion-exchange gels, but, most important, it is nontoxic and shows no immunological relatedness to toxin A. In addition to detecting toxigenic and nontoxigenic isolates of C. difficile, the Culturette Brand Rapid Latex Test reacts with strains of proteolytic Clostridium botulinum and strains of Clostridium sporogenes (35, 108), Peptostreptococcus anaerobius (S. Allen, submitted for publication), and Bacteroides asaccharolyticus (35). We have found that C. sporogenes and P. anaerobius produce an antigen which cross-reacts with the latex-reactive antigen of C. difficile. The antigen from each of these species contains a subunit with an Mr of 43,000, indicating that the antigens are very similar. Analysis by immunodiffusion, however, shows that each antigen contains some unique epitopes (D. Lyerly, D. Ball, J. Toth, and T. Wilkins, submitted for publication). The function of the antigen is not known, but there is no evidence linking it with the virulence of the organism. Problems Associated with Clinical Diagnosis of the Disease The question of how to diagnose PMC seems, at first glance, to be relatively straightforward: it would seem that a positive diagnosis involves the isolation of toxigenic C. difficile or the detection of toxin in the stool of a patient treated with antibiotics. There are a number of factors, however, which complicate the clinical picture. The number of cases of PMC reported in the past several years has decreased considerably. This is because in most cases, once the patient develops antibiotic-associated diarrhea and C. difficile toxin is detected in the stool, the patient is switched to vancomycin or some other type of therapy; the patient is not allowed to progress past this stage of the disease. That the patient responds to vancomycin treatment does not confirm that C. difficile caused the disease. Cases of antibiotic-associated diarrhea from which this organism or toxins cannot be demonstrated often improve following treatment with vancomycin. In addition, vancomycin has a broad spectrum of activity, and at the high concentrations achieved in the bowel it inhibits most pathogens. Therefore, resolution of the disease by vancomycin does not prove that the disease was caused by C. difficile. The development of the diarrhea in the first place indicates that the normal flora of the intestine has been altered; as a result, the patient could be colonized with any number of pathogens, including C. difficile. C. difficile may happen to colonize the patient once the flora has been changed by the antibiotic and may not be responsible for the diarrhea. As an example, E. coli is present in stools from patients with diarrhea, but this in no way confirms that toxigenic E. coli caused the diarrhea. It should be remembered that toxin B of C. difficile is extremely potent and low numbers of toxigenic C. difficile in the gut which do not pose a problem may produce enough of the toxin to be detected in tissue culture assay. Even the isolation of toxigenic isolates of C. difficile from patients with antibiotic-associated diarrhea is not a definitive diagnosis. Toxigenic isolates vary in their toxin production over a range of at least 6 logs. This means that a "highly" toxigenic strain may produce 106 times more toxin than a "weakly" toxigenic strain. This unusual range of expression of toxin is not seen in other bacteria. The manner in which C. difficile regulates this wide variation in toxin production is not known. That C. difficile has this unusual regulation of toxin production raises questions which have to be addressed. For example, do the highly toxigenic isolates cause PMC whereas weakly toxigenic isolates cause diarrhea or are weakly toxigenic isolates not important in the disease? Another area which has caused considerable confusion includes the relative sensitivities and specificities of the various toxin assays. It has been reported that stool specimens which are cytotoxic but not enterotoxic contain toxa/toxb+ isolates. However, it should be remembered that the tissue culture assay for toxin B is extremely sensitive

7 VOL. 1, 1988 (10-12 g/ml of toxin detected) and that enterotoxic assays (10-6 g/ml of toxin detected) are 106-fold less sensitive. In addition, there are descriptions of toxa'itoxb isolates by investigators who used the Culturette Brand Rapid Latex Test. We now know that the latex test does not detect toxin A and that these isolates were nontoxigenic isolates. Questions then arise as to how to establish C. difficile as the causative agent of the diarrhea and which toxin tests should be used. Unfortunately, there are no straightforward answers. The identification of toxigenic C. difficile as the cause of the disease has to be established based on the physician's experience. The tissue culture assay for the detection of cytotoxic activity that is neutralized by C. difficile or C. sordelifi antitoxin still represents the primary toxin test and should continue to be used until more clinical data are available regarding the other tests. The clinical results with the Culturette Brand Rapid Latex Test for C. difficile look promising (110, 173, 174, 206). The test is simple and the results can be obtained within several minutes. It should be cautioned, however, that nontoxigenic isolates of C. difficile produce spores and can exist in the environment as well as toxigenic strains. Thus, nontoxigenic isolates can be spread nosocomially, and as a result "outbreaks" of nontoxigenic strains can occur which may result in local high rates of false-positive reactions. At the present time, the latex assay may be suitable as an initial screening test for the organism and positive results can be confirmed when warranted for clinical reasons. TOXINS A AND B Role of Toxins A and B in PMC The extensive tissue damage and fluid response which toxin A causes in experimental animals have led many investigators to speculate that toxin A, and not toxin B, is the toxin primarily responsible for the symptoms associated with PMC. The severity of the symptoms in antibioticassociated cecitis appears to be more closely correlated with the in vivo production of toxin A than with toxin B (36), and these findings also support the idea that toxin A causes most of the symptoms. There are several observations, however, indicating that toxin B is also important in the disease. Hamsters must be vaccinated against both toxin A and toxin B to be completely protected against the disease; hamsters vaccinated against toxin A are not completely protected (69, 137; P.-H. Kim and R. D. Rolfe, Abstr. Annu. Meet. Am. Soc. Microbiol. 1987, B222, p. 62). Like toxin A, toxin B is also highly lethal when injected into animals. In fact, the lethal dose of toxin B is approximately the same as that of toxin A when the toxins are injected intraperitoneally or intravenously (9, 145, 220). This indicates that toxin B (and toxin A for that matter) can act on target tissue outside of the intestines. We have found that if low doses of toxin A (i.e., doses too low to give a discernible response) are given intragastrically along with toxin B to hamsters, the animals die. The only pathology consists of some mucus and edema in the small intestine and hemorrhage in the lungs (146). Thus, toxin A apparently facilitates the exit of toxin B from the gut. Both of the toxins are present in stools of PMC patients, so the synergistic action seen in the oral hamster model could also occur in the human disease. These findings also raise the possibility that asymptomatic infants who have high levels of the toxins in C. DIFFICILE TOXINS 7 their intestines may be at risk if they are exposed to conditions (e.g., intestinal surgery) which result in the dissemination of the toxins into the systemic circulation. On the basis of these observations, we feel that both of the toxins are involved in the disease. Factors Affecting Toxin Production The incorporation of antibiotics such as clindamycin into the growth medium has been reported to increase the levels of cytotoxic activity in cultures of C. difficile by four- to eightfold (78, 163). This increase in toxin production does not occur with all isolates of toxigenic C. difficile, and this may explain why some investigators have not observed this stimulation in toxin production in cultures grown with clindamycin (167). The medium used for the growth of the organism also affects the amounts of toxins produced. The organism grows well in a complex medium such as brain heart infusion broth and produces high levels of both toxin A and toxin B (147, 220). Under these conditions, the production of the toxins appears to be coregulated. Highly virulent isolates produce high levels of both toxins and weakly virulent isolates produce low levels of the toxins under these conditions. Avirulent strains do not produce either of the toxins. Under "starved" conditions, however, the coproduction of the toxins does not appear to be as tightly regulated. For example, it has been reported that some isolates produce only one of the toxins when grown in media deficient in certain amino acids (96). The level at which this inhibition occurs in the cell is not known. It is unlikely that this type of inhibition of toxin A or B production occurs in the intestine of PMC patients since the organism is in an environment rich in nutrients and relatively free of competing organisms. Some clostridial toxins such as the C2 enterotoxin of C. botulinum and the enterotoxin of Clostridium perfringens have been shown to be associated with spore production (77, 251). Neither of the C. difficile toxins, however, has been shown to be associated with sporulation (111; L. D. Bobo, A. L. Delisle, B. E. Laughon, and J. G. Bartlett, Abstr. Annu. Meet. Am. Soc. Microbiol. 1986, B67, p. 35). This is based on the findings that nonsporulating mutants produce levels of toxins A and B comparable to the sporulating parental strain and that neither toxin can be extracted from spores. Like almost all other bacterial species, C. difficile has plasmids of various sizes (97, 158, 223). In C. difficile, plasmids with molecular weights ranging from 2.7 x 106 to 100 x 10' have been detected. The presence or absence of any of these plasmids has not been correlated with toxigenicity. In addition, bacteriophage specific for C. difficile and C. sordellii have been isolated and characterized (149, 209, 211), but, again, there is no evidence that the toxin genes are located on phage. Purification Toxins A and B are present in the supernatant fluids of cultures of C. difficile and can be purified from the culture filtrates. The manner by which the toxins are liberated by the cells is not clear. We have found that the concentration of the toxins in the supernatant fluid parallels the growth of the organism and that most of the toxin is released in late logarithmic or early stationary phase, suggesting that the toxins may be actively secreted by the cells. Ketley et al. (111), on the other hand, have reported that toxins A and B are released after the organism reaches stationary phase and have suggested that the toxins are released after the cells lyse.

8 8 LYERLY ET AL. CLIN. MICROBIOL. REV. TABLE 2. Purification and properties of toxins A and B of C. difficile Reference Purification method' Toxin Mr Mr subunit ph stability Heat stability Inactivation by PI proteases Banno et al. (NH4)2SO4 ppt, Sephacryl S- A 550, ,000- b Stable at 370C; Trypsin, pronase (12, 13) 300, DEAE-Sephadex A , ,000 inactivated at 600C B 450, ,000- Stable at 370C; Trypsin, pronase 500, ,000 inactivated at 600C Krivan and Thermal affinity chromatog- A 440,000- Wilkins raphy 500,000 (120) Lonnroth (NH4)2SO4 ppt, Sephacryl S- A and Lange 300, DEAE-Sephadex A- (139) 25, affinity chromatography on agarose Pothoulakis (NH4)2SO4 ppt, DEAE-Se- B 50, et al. pharose CL-6B, HPLC on (180) Mono Q Rihn et al. Ultrafiltration, fast liquid an- A 52,000 41, (200) ion-exchange chromatogra- and phy, chromatofocusing, 16,000 TSK SW-3000 Stephen et Electrophoresis, DEAE- A al. (217) Sepharose CL-6B Sullivan et Ultrafiltration or (NH4)2SO4 A 440, ,000- Stable at ph Stable at 37 C; Trypsin and chy al. (220); ppt, DEAE-Sepharose CL- 500, ,000 4 and 10 inactivated motrypsin Lyerly et 6B, ppt at ph 5.6 (toxin at 560C al. (143, A), immunoadsorption B 360, ,000- Inactivated Stable at 37 C; Trypsin and chy ) (toxin B) 470, ,000 at ph 4 inactivated motrypsin and 10 at 56 C Taylor et al. Ultrafiltration, (NH4)2SO4 A Stable at ph Inactivated at (224) ppt, Sephadex G-200, 4 and C DEAE-Sepharose CL-6B B Inactivated Inactivated at at ph 4 560C and 10 DEAE, Diethylaminoethyl; ppt, precipitation; HPLC, high-pressure liquid chromatography. b-, Not reported. The most commonly used medium for the large-scale production of the toxins is brain heart infusion broth. The organism grows well in the broth, and many investigators purify the toxins from brain heart infusion broth cultures. A modification of this approach is the brain heart infusion dialysis, sac flask in which the organism grows within a sac suspended in the broth. In the dialysis flask procedure, the low-molecular-weight nutrients in the brain heart infusion diffuse into the sac, resulting in the slow growth of the organism. The slow growth of the organism, in turn, results in the production of more toxin. In this procedure, the toxins are not contaminated with the high-molecular-weight components from the brain heart infusion. This modification was originally utilized by Sterne and Wentzel (218) for the production of botulinum toxin and works quite well for the production of C. difficile toxins. A number of purification methods have been reported for preparing toxins A and B (Table 2). Most of the purification procedures which have been described utilize an initial concentration step such as ammonium sulfate precipitation or uitrafiltration, followed by gel filtration and ion-exchange chromatography. Ion-exchange chromatography has been used extensively because it separates toxins A and B very effectively. The simplest and most gentle procedure developed thus far for the purification of toxin A is the thermal affinity method (120). The method is based on the specific binding of toxin A to the carbohydrate sequence Gala1-3Gal11-4GicNAc on bovine thyroglobulin at 4 C and the release of the toxin at 37 C (see below). Thus, complex mixtures containing toxin A can be mixed with thyroglobulin gel at 4 C to bind the toxin. The gel is then washed and the toxin is eluted at 37 C. The eluted toxin has biological and physicochemical properties similar to those of toxin A purified by conventional methods. In many of the purification studies, the purity of the toxin was either not demonstrated or analyzed at most by only one criterion. We routinely use crossed immunoelectrophoresis in conjunction with polyacrylamide gel electrophoresis to analyze our toxin preparations because each of these methods is highly resolving and picks up contaminants not detected by the other method. In addition, we have shown that the biological activity present in the preparation (e.g., the enterotoxic or cytotoxic activity) comigrates with the immunoprecipitin arc or band. This has proved to be especially important in the purification of toxin B since the toxin is extremely potent and amounts which have high levels of

9 VOL. 1, 1988 C. DIFFICILE TOXINS 9 TABLE 3. Biological activities of toxins A and B of C. difficile Tissue culture Enterotoxic dose in Enterotoxic dose in Enterotoxic dose in Vascular Reference Toxin dose rabbit ileal loop oral hamster model suckling mice Lethal dose in micea permeability (Ig) (plg) (Og) dose Banno et al. (12) A Low activity 3 b 26 ng (LD50) 1 ng B 1 pg Negative 1.5 [ig (LD5o) 6 ng Pothoulakis et al. B 5.1 fg (180) Sullivan et al. (220) A 10 ng ng (minimum dose) 1-10 ng Lyerly et al. (142, B pg Negative Negative > ng (minimum dose) 1-10 ng 145), Lima et al. (submitted for publication) Taylor et al. (224) A 40 ng 100- Negative 10 ng (LD50) 2 jig B 1 pg Negative Negative 4.4 pug (LD50) 1 jig LD50, 50% lethal dose. b-, Not reported. ' Minimum dose not reported. cytotoxic activity may not be sufficient to give a band on gels. Physicochemical Properties Toxin A purified by Banno et al. (12, 13), Taylor et al. (224), and our group (120, 143, 145, 220) exhibits similar physicochemical properties (Table 2). One of the most unusual properties of the toxin is its extremely large size. The native toxin has an Mr in the range of 400,000 to 600,000. Even under denaturing conditions, the toxin has an Mr in excess of 250,000 and does not dissociate to subunits (12, 143, 191). These findings, along with the observation that the toxin A gene is a continuous open reading frame of more than 6.1 kilobases (see below), support the hypothesis that the toxin consists of a large polypeptide with an Mr of >250,000. It has been reported that toxin A binds to agarose (139), has an estimated Mr of 52,000 (200), and contains almost 30% glycine (217). These properties have not been confirmed and the small size which was reported does not agree with the estimated toxin size determined from the toxin A deoxyribonucleic acid sequence (see below). Toxin B purified by Banno et al. (12, 13), Taylor et al. (224), and our group (143, 145, 220) also exhibits similar properties (Table 2). Like toxin A, toxin B is extremely large and has a native Mr in the range of 360,000 to 500,000. Under denaturing conditions, the toxin behaves similarly to toxin A with an Mr of at least 250,000 and does not dissociate (12, 143). Thus, it appears that toxin B may also consist of a single large polypeptide. It has been reported that toxin B consists of subunits which have an estimated Mr of 50,000 (180), but these findings have not been confirmed and do not agree with findings by other investigators (12, 143). Toxin B is less stable than toxin A and is more susceptible to ph extremes and proteases (12, 220, 224). Both of the toxins are inactivated by oxidizing agents and can be protected by the addition of a reducing agent such as dithiothreitol (145). The oxidizing agents affect a number of amino acids, so it is difficult to determine the specific amino acids which are modified. Reducing agents such as 2-mercaptoethanol and dithiothreitol and sulfhydryl-inactivating agents such as p-hydroxymercuribenzoate, N-ethylmaleimide, and iodoacetate do not inactivate the toxins, indicating that the toxins are not sensitive to reduction and that sulfhydryl groups are not involved in the binding or toxic moieties of either toxin. Both toxins contain high amounts of the amino acids Asp, Glu, and Gly, and low amounts of His and sulfur-containing amino acids (12, 143, 145). Biologic Activities Some of the biologic activities of toxins A and B are listed in Table 3. Toxin A is a potent enterotoxin with slight cytotoxic activity, whereas toxin B is an extremely potent cytotoxin. On a molar basis, toxin A is as active as cholera toxin in the rabbit ileal loop assay for enterotoxic activity. Its action, however, is quite different from cholera toxin. Cholera toxin stimulates adenylate cyclase, resulting in an increase in intracellular cyclic adenosine 5'-monophosphate. This change, in turn, eventually leads to the rice-water fluid accumulation which is seen in rabbit ileal loops injected with the toxin. There is little or no tissue damage observed with cholera toxin. Toxin A, on the other hand, causes an extensive amount of tissue damage to the gut mucosa (142, 156; A. Lima, D. Innes, Jr., K. Chadee, D. Lyerly, T. Wilkins, and R. Guerrant, Abstr. Annu. Meet. Am. Soc. Microbiol. 1987, B219, p. 61). Initially, the toxin damages the villous tips, and this is followed by the disruption of the brush border membrane. Eventually the mucosa becomes "eroded." This tissue damage apparently results in the fluid response. The hemorrhagic fluid which develops in ileal loops treated with toxin A contains high levels of albumin, indicating that the fluid results from vessel leakage and edema fluid in the area surrounding the damaged gut mucosa. Toxin A also causes a fluid response when injected into rabbit colonic loops, but the fluid is considerably less hemorrhagic than that seen in ileal loops (156). Both of the toxins cause the same type of rounding effect on tissue culture cells, and both are active against all of the mammalian cells which have been tested. This suggests the possibility that the toxins may act by the same or a similar mechanism on cells. Most of the research on the cytotoxic activity of the toxins has centered on toxin B because it is much more potent. Toxin B causes a number of nonspecific responses in mammalian cells, including the loss of intracellular potassium, decrease in protein synthesis, and decrease in synthesis of ribonucleic and deoxyribonucleic acids (72, 181, 199, 204). These probably occur as secondary responses

10 10 LYERLY ET AL. following the intoxication process. There appears to be an effect on the microfilament system of the cell since there is disorganization of the actin filaments and an increase in the intracellular amount of globular actin (180, 229, 230). This effect most likely is responsible for the "rounding" which occurs in toxin-treated cells. This rounding effect also is observed in cells treated with cytochalasin B, a fungal metabolite which acts on microfilaments. The action of cytochalasin B, however, is reversible, whereas the rounding effect caused by toxin B is not (229). Toxin B also causes polarization of the nucleus in cells, and it has been suggested that this effect is indirectly due to the disruption of the microfilament system (3). It has been reported that in tissue culture cells the mechanism of action of toxin B is very similar to that of diphtheria toxin (73-75, 230). Additional work is needed, however, to confirm these findings and to further compare the similarities of toxin B and diphtheria toxin. Both toxin A and toxin B are lethal toxins. Mice, hamsters, rats, and rhesus monkeys injected intraperitoneally or subcutaneously with small amounts of either toxin die (9, 12, 13, 145, 220, 224). There are no obvious signs in animals injected with the toxins, making it difficult to determine the manner in which the toxins act. Toxin A is also lethal when given intragastrically to hamsters (146). In this case, the animals develop intestinal pathology and diarrhea. Toxin A exerts some "chronic" type of effect on the intestine as well. This is based on the observation that hamsters given small amounts of toxin A at weekly intervals develop diarrhea and die. This effect may contribute to the severity of relapses often seen in PMC patients. Highly buffered solutions of toxin B do not cause any response when given intragastrically to hamsters unless the intestine is damaged in some manner, either by toxin A or by manipulation of the gut. If toxin B is given intragastrically along with low amounts of toxin A, there is no significant damage to the intestines and the animals die with no apparent symptoms. This is similar to what is seen in animals injected with toxin B intraperitoneally, suggesting that the toxin exits through the damaged gut mucosa and acts distally to the intestine. These results raise the possibility that any manipulation of the intestine (e.g., intestinal surgery) in patients colonized with toxigenic C. difficile may facilitate the exit of toxin B from the gut, possibly resulting in the death of the patient. This may be especially of concern in infants colonized with toxigenic C. difficile who undergo some type of intestinal trauma. Immunochemical Properties Monoclonal antibodies against the toxins have recently been developed by several laboratories and are being used to study the toxins (143, R. Rolfe, J. Szkudlarek, K. Follis, and C. Faust, Abstr. Annu. Meet. Am. Soc. Microbiol. 1986, B62, p. 34; S. Rothman, J. Brown, D. Foret, J. Gemski, M. Gentry, and P. Stricker, Abstr. Annu. Meet. Am. Soc. Microbiol. 1987, B214, p. 60). One of the monoclonal antibodies we have characterized (PCG-4 antibody) exhibits the unusual property of precipitating toxin A. The toxin contains repeating amino acid sequences (see below), and it is likely that this monoclonal antibody binds to these repeats, thereby explaining why the antibody precipitates toxin A. Another unusual property of the PCG-4 antibody is that it neutralizes the enterotoxic activity of toxin A but not the cytotoxic activity of the toxin. The antibody blocks the binding of the toxin to the Galod-3Gal,1-4GlcNAc receptor, indicating that the antibody recognizes the binding and not CLIN. MICROBIOL. REV. the active site on toxin A. The finding that the antibody neutralizes the enterotoxic, but not the cytotoxic, activity indicates that the enterotoxic and cytotoxic activities of toxin A are distinct. If this proves to be the case, it may be possible to chemically or enzymatically cleave the toxin into distinct enterotoxic and cytotoxic fragments. It is possible that the cytotoxic activity is on another subunit; if so, the subunit has not been detected on polyacrylamide gels. Further sequencing of the toxin A gene should answer this question. In addition to monoclonal antibodies which react specifically with toxin A, monoclonal antibodies which cross-react with toxins A and B have been described (143; Rolfe et al., Abstr. Annu. Meet. Am. Soc. Microbiol. 1986; Rothman et al., Abstr. Annu. Meet. Am. Soc. Microbiol. 1987). These antibodies do not neutralize the biological activities of either toxin. The findings are surprising in light of the results from previous studies showing that polyclonal antibodies and antisera against the toxins do not show significant levels of cross-reaction or cross-neutralization with the heterologous toxin (13, 138, 224). Apparently, the polyclonal antibodies used in these studies did not contain sufficient quantities of the cross-reacting antibodies. Toxin Receptors Toxin A has been shown to bind to brush border membranes from hamsters, indicating that these membranes contain the receptor needed for expression of the enterotoxic activity (119). The role of this binding component as the physiological receptor is supported by the finding that toxin A does not bind readily to brush border membranes from mice and rats and that these animals are much less sensitive to the enterotoxic activity of toxin A. The binding to the membranes is temperature dependent. At 40C, large amounts of toxin A bind to the membrane. The toxin does not bind as well at higher temperatures, but the amount of toxin A which binds to membranes at 370C is comparable to the amount of cholera toxin which binds to the GM1 ganglioside at that temperature. Toxin A also binds to rabbit erythrocytes and, when present in high concentrations (micrograms per milliliter), agglutinates the cells (119). The binding to erythrocytes exhibits the same characteristics as the receptor in hamster brush border membranes, indicating that the binding components are probably identical. The receptor in each instance involves the trisaccharide Gala1-3GalP1-4GlcNAc. The terminal a-galactose residue is very important. Toxin A does not bind to erythrocytes treated with either ot-galactosidase, which enzymatically cleaves the terminal a-galactose, or with the lectin from Bandeirea simplicifolia, which binds specifically to terminal a-galactose (172). The entire trisaccharide sequence is needed for recognition by the toxin. Hemagglutination by toxin A cannot be inhibited by simple sugars such as galactose, by methyl-a-d-galactoside, by stachyose and raffinose, both of which contain Galal-6Gal linkages, or by the disaccharide Galal-3Gal. The specificity of the binding is further demonstrated by the observation that toxin A does not bind to Salmonella milwaukee, which expresses nonreducing Galal-3Gal41-3GalNAc on its lipopolysaccharide (140), or to Galal- 3GalP1-4Glc or Galal-4GalI1-4GlcNAc (48). Bovine thyroglobulin is the only glycoconjugate which has been shown to competitively inhibit the binding of toxin A to rabbit erythrocytes. Thyroglobulin also contains the trisaccharide Ga1oL-3Gal1P1-4GlcNAc and has been used to purify the

11 VOL. VO. 1 a,198c.difficile 1988 TOXINS 11 toxin by thermal affinity chromatography (see subsection, "Purification"). One difference we have observed in the binding of toxin A that to rabbit erythrocytes and brush border membranes is the low amount of binding of toxin A which occurs at 370C with membranes does not occur with erythrocytes. Essentially all of the toxin can be recovered from the erythrocytes. This probably reflects the unusual binding portion of toxin A and the manner in which the receptor is expressed on brush border membranes and rabbit erythrocytes. The binding portion of toxin A consists of a number of repeats. This is based on the sequence data obtained from a portion of the toxin A gene which codes for a nontoxic, hemagglutinating fragment of the toxin (see next subsection). On erythrocytes, the Galal-3Gal31-4GlcNAc trisaccharide is present singly or as branched structures containing two of the trisaccharide units (47, 65, 95). In this instance, the toxin binds to the trisaccharide by one or perhaps two of the numerous repeats which it contains. There are probably more of the trisaccharide structures on hamster brush border membranes, and this increased number and clustering of these structures most likely results in binding through more of the repeats. Thus, this increased receptor density results in a tighter binding of the toxin to the cell. As a result, the toxin cannot be completely dissociated from the membrane receptors at higher temperatures. The multiple repeated binding sites on toxin A and the apparent need for a high density of receptors for binding at body temperature suggest the interesting possibility that the toxin has evolved in such a way as to discriminate between the receptor on the cell surface and free receptors (e.g., soluble receptors and receptors on mucin). If the toxin had a single high-affinity binding site, soluble receptors could bind and inactivate the toxin. Patients with diarrhea have large membranes in amounts of mucin and partially digested cell their colons, indicating that the toxin would have to discriminate between these receptors and receptors clustered on the cell surface. By binding only to areas of high receptor density at body temperature, the toxin is able to do this. There is very little information on the receptor(s) needed for the expression of the cytotoxic activity of toxins A and B. In both instances, the toxins bind to the cell and cluster in coated pits, indicating that receptor-mediated endocytosis is involved in the intoxication process (230; V. Kushnaryov, personal communication). The nature of the receptor involved in the binding, however, is not known. It has been reported that tissue culture cells exposed to cholesterol, gangliosides, lecithin, and phospholipase C are still highly sensitive to the toxin, suggesting that the receptor is not a lipid (229). Wheat germ agglutinin, concanavalin A, and ricin have been reported to delay or reduce the cytotoxic effect, suggesting that the receptor on tissue culture cells may be a carbohydrate (229; N. J. Levy and A. B. Onderdonk, Program Abstr. 21st Intersci. Conf. Antimicrob. Agents Chemother. abstr. no. 719, 1981). Toxin B also has been reported to bind to human erythrocyte ghosts and to rabbit erythrocytes (45, 230), but we have been unable to confirm these findings. In any event, the toxin receptor(s) involved in the cytotoxic activity will most likely prove to be some type of ubiquitous molecule since both toxins are extremely active against such a wide variety of cell types. Cloning of the Toxin A Gene A 2.1-kilobase segment of the toxin A gene has recently been cloned and sequenced (S. B. Price, C. J. Phelps, T. D. BC CIA B BBC IABC CI ABBC IABB CI AB CI AB F - _m- - _-_-- _ Pst I 4 FIG. 1. Diagram of the 2.1-kilobase toxin A gene insert. The insert represents a PstI restriction digest fragment of deoxyribonucleic acid from C. difficile VPI At least five large blocks of repeats (1, 2, 3, 4, and 5) are evident. Each of these blocks starts with the consensus sequence designated I consisting of about 90 base pairs. The I sequence is followed by three, four, or five smaller repeats (A, B, C), each about 60 base pairs. (Reprinted with permission from J. Johnson, Department of Anaerobic Microbiology, Virginia Polytechnic Institute and State University, Blacksburg). Wilkins, and J. L. Johnson, Cuff. Microbiol., in press; C. H. Dove, S. B. Price, C. J. Phelps, and J. L. Johnson, Abstr. Annu. Meet. Am. Soc. Microbiol. 1987, D88, p. 86). The toxin A fragment coded for by the gene fragment reacts with polyclonal and monoclonal antibodies against toxin A and shows a reaction of partial immunological identity with native toxin A. In addition, it possesses the hemagglutinating activity, but not the enterotoxic, cytotoxic, or lethal activities, associated with toxin A. Thus, either the portion of the gene coding for the toxic activities is not encoded within this gene fragment or the gene product has not assumed the correct conformation for the expression of the toxic activity. The structure of the cloned gene fragment is very complex. There are a number of blocks of repeating sequences. Within each of these larger blocks, there are four or five repeats which show decreasing levels of homology (Fig. 1). These smaller repeats consist of about 60 base pairs; thus, each of the repeats codes for a segment of about 20 amino acids. The function of these repeating segments is not known, but their presence helps to explain some of the unusual properties of toxin A. For example, for a monoclonal antibody to precipitate an antigen, the epitope recognized by the antibody must be present in multiple copies (either multiple subunits or multiple internal repeats within a polypeptide). Therefore, these repeats in toxin A most likely represent the epitopes recognized by the PCG-4 monoclonal antibody that precipitates the toxin. Molecules which exhibit hemagglutinating activity must also be multivalent, and it is likely that the repeats on toxin A are involved in the recognition of the Galal-3GalI1-4GlcNAc trisaccharide receptor on rabbit erythrocytes and bovine thyroglobulin. The 2.1-kilobase fragment is located at the 3' end of the gene, indicating that the binding portion of the toxin is at the carboxylterminus. More than 4 kilobases of the gene have been sequenced upstream of this fragment (J. Johnson, personal communication). This entire 6.1-kilobase segment has an open reading frame, indicating that the toxin A polypeptide has an Mr in excess of 250,000. OTHER VIRULENCE FACTORS A motility-altering factor which causes altered motor activity in the intestine has been described in culture filtrates of C. difficile (105). This factor does not cause secretion and tissue damage, and it appears to be distinct from toxins A and B. Nontoxigenic strains which are very similar to highly toxigenic strains with the exception of toxin production do not cause disease in hamsters, suggesting that either the role of the motility-altering factor in disease is minor or it is only

12 12 LYERLY ET AL. produced by toxigenic strains. Additional studies are needed to confirm the production of this factor and to learn more about its relevance to the disease. In addition to the hemorrhagic enterotoxic activity caused by toxin A, a second nonhemorrhagic enterotoxic activity has been described (12). This activity has an Mr of about 200,000 and is not stable. These findings have not been confirmed, and it is not known whether this activity is due to a fragment of toxin A or whether it represents a distinct activity. PROPERTIES OF C. SORDELLII ENTEROTOXIN AND CYTOTOXIN C. sordeffli is not a common inhabitant of the human intestine and is not recognized as an important human pathogen. In animals, however, the organism causes enteritis and enterotoxemia (7, 8, 178, 194). It has been known for some time that this organism produces two toxins which are very similar to toxins A and B of C. difficile. In fact, the identification of C. difficile as the agent of PMC came about because investigators used C. sordeliji antitoxin to neutralize the cytotoxic activity in stool specimens from PMC patients. Like toxins A and B, the enterotoxin and cytotoxin of C. sordellii can be separated by ion-exchange chromatography (164, 250). The cytotoxin of C. sordellii has been purified by diethylaminoethyl-trisacryl, gel filtration on Ultrogel AcA34, and hydroxyapatite adsorption chromatography and has an Mr of about 250,000 and a pi of 4.5 (179). The cytotoxin has a 50% lethal dose in mice of about 3 ng, indicating that it is much more potent than toxin B. The 50% tissue culture infective dose of the cytotoxin, however, is about 16 ng, indicating the toxin is much less active than toxin B in tissue culture assays. The enterotoxin of C. sordellii has been purified by ultrafiltration and immunoaffinity chromatography, using a monoclonal antibody against toxin A (R. D. Martinez and T. D. Wilkins, Abstr. Annu. Meet. Am. Soc. Microbiol. 1987, B213, p. 60). The toxin has an Mr of about 500,000 and a pi of 5.8. It shows partial immunological identity to toxin A and, like toxin A, exhibits hemagglutinating and cytotoxic activities. It is not known whether the genes for the enterotoxin and cytotoxin of C. sordellii are located on the chromosome or are extrachromosomal. It is interesting that both of the toxins are produced by strains of C. sordellii and that they share many properties with toxins A and B. This suggests that the toxin genes of C. difficile and C. sordellii evolved from common ancestral genes. CLIN. MICROBIOL. REV. FUTURE DIRECTIONS Almost 10 years have elapsed since the discovery of C. difficile as the agent of PMC. During this time, we have learned much about the organism and its toxins, but there are many questions which remain to be answered. In the clinical aspects of the disease, we still do not have a clear picture of the role of this organism in antibiotic-associated diarrhea. There is some evidence that C. perfringens causes clinical illnesses similar to that caused by C. difficile, and we suspect that other clostridia play a role in these types of diseases. Therefore, it is important that studies on the epidemiology of the disease continue and that clinicians be aware that other clostridia may be involved. There are other clinical areas which need to be addressed. We have known for several years that many infants colonized with toxigenic C. difficile are refractory to the disease, but we still do not know why this occurs. Research in this area may give us clues on better ways to treat the patient, especially those persons who relapse repeatedly with PMC. The toxins produced by C. difficile evidently lead to the progression of the disease, but the manner in which this occurs appears to be much more complex than just the direct toxic action on the gut mucosa. The initial reaction appears to involve the binding of toxin A to its receptor (possibly the Galal-3GalI1-4GlcNAc trisaccharide glycoconjugate) in the brush border membrane, but the mechanism by which the toxin kills mucosal cells and causes the enterotoxic response is not known. The extensive damage which occurs from the action of toxin A and the intense inflammatory infiltrate probably results in the dissemination of both toxins A and B into the systemic circulation. Toxin A which is no longer enterotoxic (i.e., neutralized with monoclonal antibody that neutralizes the enterotoxic activity) is still lethal when injected. This raises the intriguing possibility that toxin A acts by a different mechanism once it leaves the intestine. Toxin B also acts on tissues outside of the intestine, but the receptor for the toxin and the target tissues have not been identified. Much work remains to be done on the structures of toxins A and B. It has been suggested that the enterotoxic and cytotoxic activities of toxin A reside in small distinct subunits complexed with a large hemagglutinin (230). The large size of the toxin A gene indicates, however, that these activities reside in a single large polypeptide rather than in individual subunits. Additional studies are needed to examine these possibilities and to determine which structural features are shared by toxins A and B. Perhaps these studies will give us some insight on their cytotoxic and lethal activities. Not only are toxins A and B important, but it is also important to learn how the organism becomes established in the intestine and whether the organism produces other virulence factors which are involved in the disease. For example, it is important to determine whether the labile, nonhemorrhagic enterotoxic activity which has been described is due to a fragment of toxin A or represents a distinct toxic activity. We hope that answers to these questions will direct us to better ways of recognizing and treating the disease. ACKNOWLEDGMENTS Much of the work presented in this article was supported by Public Health Service grant Al from the National Institutes of Health and State Support grant from the Commonwealth of Virginia. We thank Roger Van Tassell and Robert Carman for their comments and suggestions. LITERATURE CITED 1. Abrams, G. D., M. Allo, G. D. Rifkin, R. Fekety, and J. Silva, Jr Mucosal damage mediated by clostridial toxin in experimental clindamycin-associated colitis. Gut 21: Adler, S. P., T. Chandrika, and W. F. Berman Clostridium difficile associated with pseudomembranous colitis. Occurrence in a 12-week-old infant without prior antibiotic therapy. Am. J. Dis. Child. 135: Ahlgren, T., I. Florin, C. Jarstrand, and M. Thelestam Loss of surface fibronectin from human lung fibroblasts exposed to cytotoxin from Clostridium difficile. Infect. Immun. 39: Al-Jumaili, I. J., and A. J. Bint Simple method of isolation and presumptive identification of Clostridium difficile.

13 VOL. 1, 1988 Zentralbl. Bakteriol. Parasitenkd. Infektionskr. Hyg. Abt 1 Orig. 250: Al-Jumaili, I. J., M. Shibley, A. H. Lishmann, and C. 0. Record Incidence and origin of Clostridium difficile in neonates. J. Clin. Microbiol. 19: Allo, M., J. Silva, Jr., R. Fekety, G. D. Riflin, and H. Waskin Prevention of clindamycin-induced colitis in hamsters by Clostridium sordellii antitoxin. Gastroenterology 76: Al Mashat, R. R., and D. J. Taylor Bacteria in enteric lesions of cattle. Vet. Rec. 112: Al Mashat, R. R., and D. J. Taylor Clostridium sordellii in enteritis in an adult sheep. Vet. Rec. 112: Arnon, S. S., D. C. Mills, P. A. Day, R. V. Henrickson, N. M. Sullivan, and T. D. Wilkins Rapid death of infant rhesus monkeys injected with Clostridium difficile toxins A and B: physiologic and pathologic basis. J. Pediatr. 101: Aronsson, B., M. Granstrom, R. Mollby, and C. E. Nord Enzyme immunoassay for detection of Clostridium difficile toxins A and B in patients with antibiotic-associated diarrhoea and colitis. Eur. J. Clin. Microbiol. 4: Bachamkin, I., L. Lotz-Nolan, and D. Skalina Evaluation of a commercial cytotoxicity assay for detection of Clostridium difficile toxin. J. Clin. Microbiol. 23: Banno, Y., T. Kobayashi, H. Kono, K. Watanabe, K. Ueno, and Y. Nozawa Biochemical characterization and biologic actions of two toxins (D-1 and D-2) from Clostridium difficile. Rev. Infect. Dis. 6:S11-S Banno, Y., T. Kobayashi, K. Watanabe, K. Ueno, and Y. Nozawa Two toxins (D-1 and D-2) of Clostridium difficile causing antibiotic-associated colitis: purification and some characterization. Biochem. Int. 2: Bartlett, J. G Antibiotic-associated pseudomembranous colitis. Rev. Infect. Dis. 1: Bartlett, J. G Treatment of antibiotic-associated pseudomembranous colitis. Rev. Infect. Dis. 6:S235-S Bartlett, J. G Treatment of Clostridium difficile colitis. Gastroenterology 89: Bartlett, J. G., T. W. Chang, M. Gurwith, S. L. Gorbach, and A. B..Onderdonk Antibiotic-associated pseudomembranbus colitis due to toxin-producing clostridia. N. Engl. J. Med. 298: Bartlett, J. G., T. W. Chang, N. Moon, and A. B. Onderdonk Antibiotic-induced lethal enterocolitis in hamsters: studies with eleven agents and evidence to support the pathogenic role of toxin-producing clostridia. Am. J. Vet. Res. 39: Bartlett, J. G., T. W. Chang, N. S. Taylor, and A. B. Onderdonk Colitis induced by Clostridium difficile. Rev. Infect. Dis. 1: Bartlett, J. G., N. Moon, T. W. Chang, N. Taylor, and A. B. Onderdonk Role of Clostridium difficile in antibioticassociated pseudomembranous colitis. Gastroenterology 75: Bartlett, J. G., A. B. Onderdonk, and R. L. Cisneros Clindamycin-associated colitis in hamsters: protection with vancomycin. Gastroenterology 73: Bartlett, J. G., A. B. Onderdonk, R. L. Cisneros, and D. L. Kasper Clindamycin-associated colitis due to a toxinproducing species of Clostridium in hamsters. J. Infect. Dis. 136: Bartlett, J. G., N. S. Taylor, T. Chang, and J. Dzink Clinical and laboratory observations in Clostridium difficile colitis. Am. J. Clin. Nutr. 33: Bartlett, J. G., F. J. Tedesco, S. Shull, B. Lowe, and T. Chang Symptomatic relapse after oral vancomycin therapy of antibiotic-associated pseudomembranous colitis. Gastroenterology 78: Bartlett, J. G., S. H. Willey, T. W. Chang, and B. Lowe Cephalosporin-associated pseudomembranous colitis due to Clostridium difficile. J. Am. Med. Assoc. 242: Bate, G Comparison of Minitek Anaerobe Il, API An-Identg, and RapID ANAO Systems for identification of Clostridium difficile. Am. J. Clin. Pathol. 85: C. DIFFICILE TOXINS Boenning, D. A., G. C. Fleisher, J. M. Campos, C. W. Hulkower, and R. W. Quinlan Clostridium difficile in a pediatric outpatient population. Pediatr. Infect. Dis. 1: Bolton, R. P., and M. A. Culshaw Faecal metronidazole concentrations during oral and intravenous therapy for antibiotic associated colitis due to Clostridium difficile. Gut 27: Bolton, R. P., R. J. Sherriff, and A. E. Read Clostridium difficile associated diarrhoea: a role in inflammatory bowel disease? Lancet i: Bolton, R. P., S. K. Tait, P. R. F. Dear, and M. S. Losowsky Asymptomatic neonatal colonisation by Clostridium difficile. Arch. Dis. Child. 59: Borriello, S. P Clostridium difficile and its toxin in the gastrointestinal tract in health and disease. Res. Clin. Forums 1: Borriello, S. P Antibiotic-associated diarrhoea and colitis. Martinus Nijhoff Publishers, Boston. 33. Borriello, S. P., and F. E. Barclay Colonization resistance to Clostridium difficile infection. Microecol. Ther. 14: Borriello, S. P., and F. E. Barclay An in-vitro model of colonisation resistance to Clostridium difficile infection. J. Med. Microbiol. 21: Borriello, S. P., F. E. Barclay, P. J. Reed, A. R. Welch, J. D. Brown, and D. W. Burdon Analysis of latex agglutination test for Clostridium difficile toxin A (D-1) and differentiation between C. difficile toxins A and B and latex reactive protein. J. Clin. Pathol. 40: Borriello, S. P., F. E. Barclay, A. R. Welch, J. M. Ketley, T. J. Mitchell, J. Stephen, and G. E. Griffin Host and microbial determinants of the spectrum of Clostridium difficile mediated gastrointestinal disorders. Microecol. Ther. 15: Bowden, T. A., A. R. Mansberger, and L. E. Lykins Pseudomembranous enterocolitis: mechanism of restoring floral homeostasis. Am. Surg. 47: Brettle, R. P., and E. Wallace Clostridium difficileassociated diarrhoea. J. Infect. 8: Brook, I Isolation of toxin producing Clostridium difficile from two children with oxacillin- and dicloxacillin-associated diarrhea. Pediatrics 65: Browne, R. A., R. Fekety, Jr., J. Silva, Jr., D. I. Boyd, C. 0. Work, and G. D. Abrams The protective effect of vancomycin on clindamycin-induced colitis in hamsters. Johns Hopkins Med. J. 141: Buggy, B. P., K. H. Wilson, and R. Fekety Comparison of methods for recovery of Clostridium difficile from an environmental surface. J. Clin. Microbiol. 18: Burdon, D. W., J. D. Brown, D. J. Youngs, Y. Arabi, N. Shinagawa, J. Alexander-Williams, and M. R. B. Keighley Antibiotic susceptibility of Clostridium difficile. J. Antimicrob. Chemother. 5: Cashore, W. J., G. Peter, M. Lavermann, B. S. Stonestreet, and W. Oh Clostridial colonization and clostridial toxin in neonatal necrotizing enterocolitis. J. Pediatr. 98: Chang, T., J. G. Bartlett, S. L. Gorbach, and A. B. Onderdoihk Clindamycin-induced enterocolitis in hamsters as a model of pseudomembranous colitis in patients. Infect. Immun. 20: Chang, T.-W., S. L. Gorbach, and J. G. Bartlett Inhibition of binding of Clostridium difficile toxin by steroids. J. Infect. Dis. 142: Chang, T.-W., N. M. Sullivan, and T. D. Wilkins Insusceptibility of fetal intestinal mucosa and fetal cells to Clostridium difficile toxins. Acta Pharmacol. Sin. 5: Chien, J. L., S. C. Li, and Y. T. Li Major glycosphingolipids of bovine-erythrocyte membranes, p In R. E. Harmon (ed.), Cell surface carbohydrate chemistry. Academic Press, Inc., New York. 48. Clark, G. F., H. C. Krivan, T. D. Wilkins, and D. F. Smith Toxin A from Clostridium difficile binds to rabbit eryth-

14 14 LYERLY ET AL. rocyte glycolipids with terminal Galal-3Galp1-4G0cNAc sequences. Arch. Biochem. Biophys. 257: Colardyn, F., G. Verschraegen, G. Claeys, and D. Vogelaers Clostridium dijsile-associated diarrhoea during treatment with imipenem. Eur. J. Clin. Microbiol. 3: Collignon, A., J. Cotte-Laffitte, A. M. Quero, and J. C. Torlotin Clostridium difficile and its cytotoxin in stools of hospitalized infants. Influence of antibiotic treatment. Pathol. Biol. 34: Corthier, G., F. Dubos, and P. Raibaud Modulation of cytotoxin production by Clostridium difficile in the intestinal tracts of gnotobiotic mice inoculated with various human intestinal bacteria. Appl. Environ. Microbiol. 49: Cudmore, M. A., J. Silva, R. Fekety, M. K. Liepman, and K.-H. Kim Clostridium difficile colitis associated with cancer chemotherapy. Arch. Intern. Med. 142: Czuprynski, C. J., W. J. Johnson, E. Balish, and T. Wilkins Pseudomembtranous colitis in Clostridium difficile-monoassociated rats. Infect. Immun. 39: D'Ari, L., and H. A. Barker p-cresol formation by cell-free extracts of Clostridium difficile. Arch. Microbiol. 143: Decoux, G. M., aqd C. Devroede Acute colitis related to spiramycin. Lancet ii: Deery, H. G., P. G. Jones, C. A. Kauffman, R. Fekety, J. Silva, T. L. Sell, and R. B. 5upena Effect of therapy with latamoxef (moxalactam) on carriage of Clostridium difficile. J. Antimicrob. Chemother. 13: Delmee, M., G. Bulliard, and G. Simon Application of a technique for serogrouping Clostridium difficile in an outbreak of antibiotic-associated diarrhea. J. Infect. 13: Donta, S. T., and M. G. Meyers Clostridium difficile toxin in asymptomatic neonates. J. Pediatr. 100: Dorman, S. A., E. Liggoria, W. C. Winn, Jr., and W. L. Beeken Isolation of Clostridium difficile from patients with inactive Crohn's disease. Gastroenterology 82: Dudley, M. N., J. C. McLaughlin, G. Carrington, J. Frick, C. H. Nightingale, and R. Quintiliani Oral bacitracin vs vancomycin therapy for Clostridium difficile-induced diarrhea. Arch. Intern. Med. 146: Dzink, J.,'and J. G. Bartlett In vitro susceptibility of Clostridium difficile isolates from patients with antibioticassociated diarrhea or colitis. Antimicrob. Agents Chemother. 17: Ehrich, M., R. L. Van Tassell, J. M. Libby, and T. D. Wilkins Production of Clostridium difficile antitoxin. Infect. Immun. 28: Ellis, M. E., B. M. Watson, P. J. Milewski, and G. Jones Clostridium difficile colitis unassociated with antibiotic therapy. Br. J. Surg. 70: Elmer, G. W., and L. V. McFarland Suppression by Saccharomyces boulardii of toxigenic Clostridium difficile overgrowth after vancomycin treatment in hamsters. Antimicrob. Agents Chemother. 31: Eto, T., Y. Ichikawa, K. Nishimura, S. Ando, and T. Yamakawa Chemistry of lipid of the posthemolytic residue or stroma of erythrocytes. XVI. Occurrence of ceramide pentasaccharide in the membrane of erythrocytes and reticulocytes of rabbit. J. Biochem. 64: Fekety, R Prevention and treatment of antibiotic-associated colitis, p In D. Schlessinger (ed.), Microbiology American Society for Microbiology, Washington, D.C. 67. Fekety, R., J. Silva, J. Armstrong, M. Allo, R. Browne, J. Ebright, R. Lusk, G. Rifkin, and R. Toshniwal Treatment of antibiotic-associated enterocolitis with vancomycin. Rev. Infect. Dis. 3:S273-S Fekety, R., J. Silva, R. Toshniwal, M. Allo, J. Armstrong, R. Browne, J. Ebright, and G. Rifkin Antibiotic-associated colitis: effects of antibiotics on Clostridium difficile and the disease in hamsters. Rev. Infect. Dis. 1: Fernie, D. S., R. 0. Thomson, I. Batty, and P. D. Walker Active and passive immunization to protect against antibiotic CLIN. MICROBIOL. REV. associated caecitis in hamsters. Dev. Biol. Stand. 53: Finegold, S. M Clinical considerations in the diagnosis of antimicrobial agent-associated gastroenteritis. Diagn. Microbiol. Infect. Dis. 4:87S-91S. 71. Finney, J. M Gastro-enterostomy for cicatrizing ulcer of the pylorus. Johns Hopkins Hosp. Bull. 11: Florin, I., and M. Thelestam Intoxication of cultured human lung fibroblasts with Clostridium difficile toxin. Infect. Immun. 33: Florin, I., and M. Thelestam Polyphosphate-mediated protection from cellular intoxication with Clostridium difficile toxin B. Biochim. Biophys. Acta 805: Florin, I., and M. Thelestam ADP-ribosylation in cultured cells treated with Clostridium difficile toxin B. Biochem. Biophys. Res. Commun. 139: Florin, I., and M. Thelestam Lysosomal involvement in cellular intoxication with Clostridium difficile toxin B. Microb. Pathogen. 1: Fournier, G., J. Orgiazzi, B. Lenoir, and M. Dechavanne Pseudomembranous colitis probably due to rifampicin. Lancet i: Frieben, W. R., and C. L. Duncan Homology between enterotoxin protein and spore structural protein in Clostridium perfringens type A. Eur. J. Biochem. 39: George, R. H., M. Johnson, D. Youngs, and D. W. Burdon Induction of Clostridium difficile toxin by antibiotics, p In J. D. Nelson and C. Grassi (ed.), Current chemotherapy and infectious diseases, vol. 2. American Society for Microbiology, Washington, D.C. 79. George, R. H., J. M. Symonds, F. Dimock, J. D. Brown, Y. Arabi, N. Shinagawa, M. R. B. Keighley, J. Alexander-Williams, and D. W. Burdon Identification of Clostridium difficile as a cause of pseudomembranous colitis. Br. Med. J. 1: George, W. L Antimicrobial agent-associated colitis and diarrhea. West. J. Med. 133: George, W. L Antimicrobial agent-associated colitis and diarrhea: historical background and clinical aspects. Rev. Infect. Dis. 6:S208-S George, W. L., E. J. C. Goldstein, V. L. Sutter, S. L. Ludwig, and S. M. Finegold Etiology of antimicrobial-agentassociated colitis. Lancet i: George, W. L., V. L. Sutter, D. Citron, and S. M. Finegold Selective and differential medium for isolation of Clostridium difficile. J. Clin. Microbiol. 9: Gianfrilli, P., A. Pantosti, and I. Luzzi Evaluation of gas-liquid chromatography for the rapid diagnosis of Clostridium difficile associated disease. J. Clin. Pathol. 38: Gilligan, P. H., L. R. McCarthy, and V. M. Genta Relative frequency of Clostridium difficile in patients with diarrheal disease. J. Clin. Microbiol. 14: Goulston, S. J. M., and V. J. McGovern Pseudomembranous colitis. Gut 6: Greenfield, C., J. R. Aguilar Ramirez, R. E. Pounder, T. Williams, M. Danvers, S. R. Marper, and P. Noone Clostridium difficile and inflammatory bowel disease. Gut 24: Griffin, G. E., A. McDougall, S. A. Ash, and S. P. Borriello Age related susceptibility of hamsters to clindamycin induced Clostridium difficile caecitis. Microecol. Ther. 14: Gurwith, M. J., C. Langston, and D. M. Citron Toxinproducing bacteria in infants. Am. J. Dis. Child. 135: Gurwith, M. J., H. R. Rabin, and K. Love Diarrhea associated with clindamycin and ampicillin therapy: preliminary results of a cooperative study. J. Infect. Dis. 135:S104- S Hachler, H., and J. Wust Reexamination by bacteriophage typing of Clostridium difficile strains isolated during a nosocomial outbreak. J. Clin. Microbiol. 20: Hafiz, S., R. S. Morton, M. G. McEntegart, and S. A. Waitkins Clostridium difficile in the urogenital tract of males and

15 VOL. 1, 1988 females. Lancet i: Hall, I. C., and E. O'Toole Intestinal flora in new-born infants. Am. J. Dis. Child. 49: Han, V. K. M., H. Sayed, G. W. Chance, D. G. Brabyn, and W. A. Shaheed An outbreak of Clostridiulm difficile necrotizing enterocolitis: a case for oral vancomycin therapy. Pediatrics 71: Handfland, P., H. Egge, U. Dabrowski, S. Kuhn, D. Roelcke, and J. Dabrowski Isolation and characterization of an I-active ceramide decasaccharide from rabbit erythrocyte membranes. Biochemistry 20: Haslam, S. C., J. M. Ketley, T. J. Mitchell, J. Stephen, D. W. Burdon, and D. C. A. Candy Growth of Clostridium difficile and production of toxins A and B in complex and defined media. J. Med. Microbiol. 21: Hayter, P. M., and J. W. Dale Detection of plasmids in clinical isolates of Clostridium difficile. Microbios Lett. 27: Holst, E., I. Helin, and P. A. Mardh Recovery of Clostridium difficile from children. Scand. J. Infect. Dis. 13: Humphrey, C. D., C. W. Condon, J. R. Cantey, and F. E. Pittman Partial purification of a toxin found in hamsters with antibiotic-associated colitis. Gastroenterology 76: Humphrey, C. D., W. B. Lushbaugh, C. W. Condon, J. C. Pittman, and F. E. Pittman Light and electron microscopic studies of antibiotic associated colitis in the hamster. Gut 20: laconis, J. P., and R. D. Rolfe Clostridium difficileassociated ileocecitis in clindamycin-treated infant hamsters. Curr. Microbiol. 13: Jarvis, W., 0. Nunez-Monteil, F. Thompson, V. Dowell, M. Towns, G. Morris, and E. Hill Comparison of bacterial isolation, cytotoxicity assay, and counterimmunoelectrophoresis for the detection of Clostridium difficile and its toxin. J. Infect. Dis. 147: Jarvis, W., 0. Nunez-Monteil, F. Thompson, V. Dowell, M. Towns, G. Morris, W. Vogler, E. Winton, and E. Hill False-positive counterimmunoelectrophoresis tests for Clostridiium difficile: the role of Clostridiurm bifermentans and Clostridiurn sordellii. J. Infect. Dis. 148: Jones, R. L., W. S. Adney, and R. K. Shideler Isolation of Clostridium difficile and detection of cytotoxin in the feces of diarrheic foals in the absence of antimicrobial treatment. J. Clin. Microbiol. 25: Justus, P. G., J. L. Martin, D. A. Goldberg, N. S. Taylor, J. G. Bartlett, R. W. Alexander, and J. R. Mathias Myoelectric effects of Clostridinm difficile: motility-altering factors distinct from its cytotoxin and enterotoxin in rabbits. Gastroenterology 83: Kabins, S. A., and T. J. Spira Outbreak of clindamycinassociated colitis. Ann. Intern. Med. 83: Kahn, M. Y., and W. H. Hall Staphylococcal enterocolitis-treatment with oral vancomycin. Ann. Intern. Med. 65: Kamiya, S., S. Nakamura, K. Yamakawa, and S. Nishida Evaluation of a commercially available latex immunoagglutination test kit for detection of Clostridium difficile D-1 toxin. Microbiol. Immunol. 30: Keighley, M. R. B., D. Young, M. Johnson, R. N. Allen, and D. W. Burdon Clostridium difficile toxin in acute diarrhoea complicating inflammatory bowel disease. Gut 23: Kelly, M. T., S. G. Champagne, C. H. Sherlock, M. A. Noble, H. J. Freeman, and J. A. Smith Commercial latex agglutination test for detection of Clostridium diffiule-associated diarrhea. J. Clin. Microbiol. 25: Ketley, J. M., S. C. Haslam, T. J. Mitchell, J. Stephen, D. C. A. Candy, and D. W. Burdon Production and release of toxins A and B by Clostridiurm difficile. J. Med. Microbiol. 18: Ketley, J. M., T. J. Mitchell, S. C. Haslam, J. Stephen, D. C. A. C. DIFFICILE TOXINS 15 Candy, and D. W. Burdon Sporogenesis and toxin A production by Clostridiun difficile. J. Med. Microbiol. 22: Kim, K., H. L. Dupont, and L. K. Pickering Outbreaks of diarrhea associated with Clostridium difficile and its toxin in day-care centers: evidence of person-to-person spread. J. Pediatr. 102: Kim, K., L. K. Pickering, H. L. Dupont, N. Sullivan, and T. Wilkins In vitro and in vivo neutralizing activity of human clostrum and milk against purified toxins A and B of Clostridium difficile. J. Infect. Dis. 150: Kim, K.-H., R. Fekety, D. H. Batts, D. Brown, M. Cudmore, J. Silva, Jr., and D. Waters Isolation of Clostridium difficile from the environment and contacts of patients with antibiotic-associated colitis. J. Infect. Dis. 143: Knoop, F. C Clindamycin-associated enterocolitis in guinea pigs: evidence for a bacterial toxin. Infect. Immun. 23: Kreutzer, E. W., and F. D. Milligan Treatment of antibiotic-associated pseudomembranous colitis with cholestyramine resins. Johns Hopkins Med. J. 143: Krishnan, C Detection of Clostridiurm difficile toxins by enzyme immunoassay. J. Hyg. 96: Krivan, H. C., G. F. Clark, D. F. Smith, and T. D. Wilkins Cell surface binding site for Clostridium difficile enterotoxin: evidence for a glycoconjugate containing the sequence Galcl-3Galp1-4GlcNAc. Infect. Immun. 53: Krivan, H. C., and T. D. Wilkins Purification of Clostridium difficile toxin A by affinity chromatography on immobilized thyroglobulin. Infect. Immun. 55: Kudhair, A., M. Delmee, and G. Wauters Rapid identification of Clostridinm difficile by determination of leucine arylamidase activity. Eur. J. Clin. Microbiol. 5: Kupier, E. J., J. H. Oudbier, W. N. H. M. Stuifbergen, A. Jansz, and H. C. Zanen Application of whole-cell DNA restriction endonuclease profiles to the epidemiology of Clostridiurm difficile-induced diarrhea. J. Clin. Microbiol. 25: LaMont, J. T., and Y. M. Trnka Therapeutic implications of Clostridimn difijcile toxin during relapse of chronic inflammatory bowel disease. Lancet i: Larson, H. E., F. E. Barclay, P. Honour, and I. D. Hill Epidemiology of Clostridiulm difficile in infants. J. Infect. Dis. 146: Larson, H. E., P. Honour, A. B. Price, and S. P. Borriello Clostridinm difficile and the etiology of pseudomembranous colitis. Lancet i: Larson, H. E., J. V. Parry, A. B. Price, D. R. Davies, J. Dolby, and D. A. J. Tyrrell Undescribed toxin in pseudomembranous colitis. Br. Med. J. 1: Larson, H. E., and A. B. Price Pseudomembranous colitis: presence of clostridial toxin. Lancet. ii: Laughon, B. E., R. P. Viscidi, S. L. Gdovin, R. H. Yolken, and J. G. Bartlett Enzyme immunoassays for detection of Clostridinm difficile toxins A and B in fecal specimens. J. Infect. Dis. 149: Levett, P. N Detection of Clostridiulm dificile in faeces by direct gas liquid chromatography. J. Clin. Pathol. 37: Levett, P. N Effect of antibiotic concentration in a selective medium on the isolation of Clostridiurm difficile from faecal specimens. J. Clin. Pathol. 38: Levett, P. N Effect of basal medium upon fluorescence of Clostridium dfficile. Lett. Appl. Microbiol. 1: Levett, P. N Identification of Clostridium difficile using the API ZYM system. Eur. J. Clin. Microbiol. 4: Levett, P. N Clostridium difficile in habitats other than the human gastrointestinal tract. J. Infect. 12: Levett, P. N., and K. D. Phillips Gas chromatographic identification of Clostridiuln difficile and detection of cytotoxin from a modified selective medium. J. Clin. Pathol. 38: Levine, H. G., M. Kennedy, and J. T. LaMont Counterimmunoelectrophoresis versus cytotoxicity assay for the de-

16 16 LYERLY ET AL. tection of Clostridium difficile toxin. J. Infect. Dis. 145: Libby, J. M., S. T. Donta, and T. D. Wilkins Clostridium difficile toxin A in infants. J. Infect. Dis. 148: Libby, J. M., B. S. Jortner, and T. D. Wilkins Effects of the two toxins of Clostridium difficile in antibiotic-associated cecitis in hamsters. Infect. Immun. 36: Libby, J. M., and T. D. Wilkins Production of antitoxins to two toxins of Clostridium difficile and immunological comparison of the toxins by cross-neutralization studies. Infect. Immun. 35: Lonnroth, I., and S. Lange Toxin A of Clostridium difficile: production, purification and effect in the mouse intestine. Acta Pathol. Microbiol. Immunol. Scand. 91: Luderitz, O., D. A. R. Simmons, and 0. Westphal The immunochemistry of Salmonella chemotype VI 0-antigens. Biochem. J. 97: Lusk, R. H., R. Fekety, J. Silva, R. A. Browne, D. H. Ringler, and G. D. Abrams Clindamycin-induced enterocolitis in hamsters. J. Infect. Dis. 137: Lyerly, D. M., D. E. Lockwood, S. H. Richardson, and T. D. Wilkins Biological activities of toxins A and B of Clostridium difficile. Infect. Immun. 35: Lyerly, D. M., C. J. Phelps, J. Toth, and T. D. Wilkins Characterization of toxins A and B of Clostridium difficile with monoclonal antibodies. Infect. Immun. 54: Lyerly, D. M., C. J. Phelps, and T. D. Wilkins Monoclonal and specific polyclonal antibodies for immunoassay of Clostridium difficile toxin A. J. Clin. Microbiol. 21: Lyerly, D. M., M. D. Roberts, C. J. Phelps, and T. D. Wilkins Properties of toxins A and B of Clostridium difficile. FEMS Microbiol. Lett. 33: Lyerly, D. M., K. E. Saum, D. MacDonald, and T. D. Wilkins Effect of toxins A and B given intragastrically to animals. Infect. Immun. 47: Lyerly, D. M., N. M. Sullivan, and T. D. Wilkins Enzyme-linked immunosorbent assay for Clostridium difficile toxin A. J. Clin. Microbiol. 17: Lyerly, D. M., and T. D. Wilkins Commercial latex test for Clostridium difficile toxin A does not detect toxin A. J. Clin. Microbiol. 23: Mahoney, D. E., P. D. Bell, and K. B. Easterbrook Two bacteriophages of Clostridium difficile. J. Clin. Microbiol. 21: Makin, T Rapid identification of Clostridium difficile by direct detection of volatile organic acids from primary isolation media. J. Clin. Pathol. 37: Malamou-Ladas, H,, S. O'Farrell, J. Q. Nash, and S. Tabaqchali Isolation of Clostridium difficile from patients and the environment of hospital wards. J. Clin. Pathol. 36: Mardh, P.-A., I. Helin, I. Colleen, M. Oberg, and E. Hoist Clostridium difficile toxin in faecal specimens of healthy children and children with diarrhoea. Acta Pediatr. Scand. 71: Marler, L., S. Allen, and J. Siders Rapid enzymatic characterization of clinically encountered anaerobic bacteria with the API ZYM system. Eur. J. Clin. Microbiol. 3: Massot, J., 0. Sanchez, R. Couchy, J. Astoin, and A. L. Parodi Bacterio-pharmacological activity of Saccharomyces boulardii in clindamycin-induced colitis in the hamster. Arzneim. Forsch. 34: Meyers, S., L. Mayer, E. Bottone, E. Desmond, and H. D. Janowitz Occurrence of Clostridium difficile toxin during the course of inflammatory bowel disease. Gastroenterology 80: Mitchell, T. J., J. M. Ketley, S. C. Haslam, J. Stephen, D. W. Burdon, D. C. A. Candy, and R. Daniel Effect of toxin A and B of Clostridium difficile on rabbit ileum and colon. Gut 27: Moskovitz, M., and J. G. Bartlett Recurrent pseudomembranous colitis unassociated with prior antibiotic therapy. Arch. Intern. Med. 141: Muldrow, L. L., E. R. Archibald, 0. L. Nunez-Montiel, and R. J. Sheehy Survey of the extrachromosomal gene pool CLIN. MICROBIOL. REV. of Clostridium difficile. J. Clin. Microbiol. 16: Mulligan, M. E Epidemiology of Clostridium difficileinduced intestinal disease. Rev. Infect. Dis. 6:S222-S Mulligan, M. E., S. Halebian, R. Y. Y. Kwok, W. C. Cheng, S. M. Finegold, C. R. Anselmo, D. N. Gerding, and L. R. Peterson Bacterial agglutination and polyacrylamide gel electrophoresis for typing Clostridium difficile. J. Infect. Dis. 153: Mulligan, M. E., R. D. Rolfe, S. M. Finegold, and W. L. George Contamination of a hospital environment by Clostridium difficile. Curr. Microbiol. 3: Nachamkin, I., L. Lotz-Nolan, and D. Skalina Evaluation of a commercial cytotoxicity assay for detection of Clostridium difficile toxin. J. Clin. Microbiol. 23: Nakamura, S., M. Mikawa, N. Tanabe, K. Yamakawa, and S. Nishida Effect of clindamycin on cytotoxin production of Clostridium difficile. Microbiol. Immunol. 26: Nakamura, S., N. Tanabe, K. Yamakawa, and S. Nishida Cytotoxin production by Clostridium sordellii strains. Microbiol. Immunol. 27: O'Farrell, S., M. Wilks, J. Q. Nash, and S. Tabaqchali A selective enrichment broth for the isolation of Clostridium difficile. J. Clin. Pathol. 37: Onderdonk, A. B., R. L. Cisneros, and J. G. Bartlett Clostridium diffidile in gnotobiotic mice. Infect. Immun. 28: Onderdonk, A. B., B. R. Lowe, and J. G. Bartlett Effect of environmental stress on Clostridium difficile toxin levels during continuous cultivation. Appl. Environ. Microbiol. 38: Orchard, J. L., R. Fekety, and J. R. Smith Antibioticassociated colitis due to Clostridium difficile in a Kodiak bear. Am. J. Vet. Res. 44: Peach, S. L., S. P. Borriello, H. Gayna, F. E. Barclay, and A. R. Welch Asymptomatic carriage of Clostridium difficile in patients with cystic fibrosis. J. Clin. Pathol. 39: Peikin, S. R., J. Galdibini, and J. G. Bartlett Role of Clostridium difficile in a case of non-antibiotic-associated pseudomembranous colitis. Gastroenterology 79: Pepersack, F., M. Labbe, C. Nonhoff, and E. Schoutens Use of gas-liquid chromatography as a screening test for toxigenic Clostridium difficile in diarrhoeal stools. J. Clin. Pathol. 36: Peters, B. P., and I. J. Goldstein The use of fluoresceinconjugated Bandeirea simplicifolia B4-isolectin as a histochemical reagent for the detection of al-d-galactopyranosyl groups. Exp. Cell Res. 120: Peterson, L. R., J. J. Holter, C. J. Shanholtzer, C. R. Garrett, and D. N. Gerding Detection of Clostridium difficile toxin A (enterotoxin) and B (cytotoxin) in clinical specimens. Am. J. Clin. Pathol. 86: Peterson, L. R., J. J. Holter, C. J. Shanholtzer, C. R. Garrett, and D. N. Gerding Detection of C. difficile toxin A (enterotoxin). Am. J. Clin. Pathol. 87: Phillips, K. D., and P. A. Rogers Rapid detection and presumptive identification of Clostridium difficile by p-cresol production on a selective medium. J. Clin. Pathol. 34: Phua, T. J., T. R. Rogers, and A. P. Pallett Prospective study of Clostridium difficile colonization and para-cresol detection in the stools of babies on a special care unit. J. Hyg. 93: Pierce, P. F., Jr., R. Wilson, J. Silva, Jr., V. F. Garagusi, G. D. Riflin, R. Fekety, 0. Nunez-Montiel, V. R. Dowell, Jr., and J. M. Hughes Antibiotic-associated pseudomembranous colitis: an epidemiologic investigation of a cluster of cases. J. Infect. Dis. 145: Popoff, M. R Bacteriological examination in enterotoxaemia of sheep and lamb. Vet. Rec. 114: Popoff, M. R Purification and characterization of Clostridium sordellii lethal toxin and cross-reactivity with Clostridium difficile cytotoxin. Infect. Immun. 55: Pothoulakis, C., L. M. Barone, R. Ely, B. Faris, M. E. Clark, C. Franzblau, and J. T. LaMont Purification and prop-

17 VOL. 1, 1988 erties of Clostridium difficile toxin B. J. Biol. Chem. 261: Pothoulakis, C., G. Triadafilopoulos, M. Clark, C. Franzblau, and J. T. LaMont Clostridium difficile cytotoxin inhibits protein synthesis in fibroblasts and intestinal mucosa. Gastroenterology 91: Potvliege, C., M. Labbe, and E. Youassowsky Gas liquid chromatography as screening test for Clostridium difficile. Lancet ii: Poxton, I. R., B. Aronsson, R. Mollby, C. E. Nord, and J. G. Collee Immunochemical fingerprinting of Clostridium difficile strains isolated from an outbreak of antibiotic-associated colitis and diarrhoea. J. Med. Microbiol. 17: Poxton, I. R., and M. D. Byrne Detection of Clostridium difficile toxin by counterimmunoelectrophoresis: a note of caution. J. Clin. Microbiol. 14: Poxton, I. R., and M. D. Byrne Immunological analysis of the EDTA-soluble antigens of Clostridium difficile and related species. J. Gen. Microbiol. 122: Poxton, I. R., and M. D. Byrne Demonstration of shared antigens in the genus Clostridium by an enzyme-linked immunosorbent assay. J. Med. Microbiol. 17: Poxton, I. R., and T. D. I. Cartmill Immunochemistry of the cell-surface carbohydrate antigens of Clostridium difficile. J. Gen. Microbiol. 128: Price, A. B., H. E. Larson, and J. Crow Morphology of experimental antibiotic-associated enterocolitis in the hamster: a model for human pseudomembranous colitis and antibioticassociated diarrhoea. Gut 20: Prigogine, T., C. Potvliege, A. Burette, T. Verbeet, and J. Schmerber Pseudomembranous colitis and rifampicin. Chest 80: Ramirez-Ronda, C. H Incidence of clindamycin-associated colitis. Comments and corrections. Ann. Intern. Med. 81: Rautenberg, P., F. Stender, and U. Ullman Detection of toxin A by immunoblotting. Zentralbl. Bakteriol. Parasitenkd. Infektionskr. Hyg. Abt. 1 Orig. 261: Rehg, J. E Cecal toxin(s) from guinea pigs with clindamycin-associated colitis, neutralized by Clostridium sordellii antitoxin. Infect. Immun. 27: Rehg, J. E., and S. P. Pakes Implication of Clostridium difficile and Clostridium perfringens iota toxins in experimental lincomycin-associated colitis of rabbits. Lab. Anim. Sci. 32: Richards, S. M Clostridium sordellii in lambs. Vet. Rec. 111: Rietra, P. J. G. M., K. W. Slaterus, H. C. Zanen, and S. G. M. Meuwissen Clostridial toxins in faeces of healthy infants. Lancet ii: Rifkin, G. D., R. Fekety, and J. Silva Neutralization by Clostridium sordellii antitoxin of toxins implicated in clindamycin-induced cecitis in the hamster. Gastroenterology 75: Rifkin, G. D., J. Silva, Jr., and F. R. Fekety Gastrointestinal and systemic toxicity of fecal extracts from hamsters with clindamycin-induced colitis. Gastroenterology 74: Rifkin, G. D., J. Silva, Jr., F. R. Fekety, and R. B. Sack Antibiotic-induced colitis. Implication of a toxin neutralised by Clostridium sordellii antitoxin. Lancet ii: Rihn, B., G. Beck, H. Monteil, F. Lecuf, and R. Girardot Effect of the cytotoxin of Clost'ridium difficile on cultured hepatoma cells. Biol. Cell 53: Rihn, B., J. M. Scheftel, R. Girardot, and H. Monteil A new purification procedure for Clostridium difficile enterotoxin. Biochem. Biophys. Res. Commun. 124: Robert, A., G. L. Bundy, S. 0. Field, J. E. Nezamis, J. P. David, A. J. Hanchar, C. Lancaster, and M. J. Ruwart Prevention of cecitis in hamsters by certain prostaglandins. Prostaglandins 29: Rocca, J. M., A. S. Peiterse, R. Rowland, R. Hecker, and G. E. Rich Clostridium difficile colitis. Aust. N. Z. J. Med. 14: C. DIFFICILE TOXINS Rolfe, R. D., and J. P. laconis Intestinal colonization of infant hamsters with Clostridium difficile. Infect. Immun. 42: Rothman, S. W., and J. E. Brown Inhibition of membrane functions in intact HeLa cells by Clostridium difficile cytotoxic culture filtrates. Curr. Microbiol. 6: Ruwart, M. J., H. E. Renis, A. L. Erlandson, R. M. DeHaan, J. E. Nezamis, C. Lancaster, and J. P. Davis ,16- Dimethyl PGE2 inhibits toxin release from Clostridium difficile, p In B. Samuelsson (ed.), Advances in prostaglandins & thromboxane research, vol. 12. Raven Press, New York Ryan, R. W Considerations in the laboratory diagnosis of antibiotic-associated gastroenteritis. Diagn. Microbiol. Infect. Dis. 4:79S-86S Saginur, R., R. Fogel, L. Begin, B. Cohen, and J. Mendelson Splenic abscess due to Clostridium difficile. J. Infect. Dis. 147: Sands, M., M. Yungbluth, and H. M. Sommers The non-value of counterimmunoelectrophoresis for the direct rapid detection of Clostridium difficile toxin in stool filtrates. Am. J. Clin. Pathol. 79: Schallehn, G Clostridium sordellii bacteriophage active on Clostridium difficile. Lett. Apple. Microbiol. 1: Schwan, A., S. Sjolin, U. Trottestam, and B. Aronsson Relapsing Clostridium difficile enterocolitis cured by rectal infusion of normal faeces. Scand. J. Infect. Dis. 16: Sell, T. L., D. R. Schaberg, and F. R. Fekety Bacteriophage and bacteriocin typing scheme for Clostridium difficile. J. Clin. Microbiol. 17: Sherertz, R. J., and F. A. Sarubbi The prevalence of Clostridium difficile in a nursery population: a comparison between patients with necrotizing enterocolitis and a asymptomatic group. J. Pediatr. 100: Silva, J., R. Fekety, C. Werk, J. Ebright, M. Cudmore, D. Batts, C. Syrjamaki, and J. Lukens Inciting and etiologic agents of colitis. Rev. Infect. Dis. 6:S214-S Small, J. D Fatal enterocolitis in hamsters given lincomycin hydrochloride. Lab. Anim. Care 18: Smith, M. F., S. P. Borriello, G. S. Clayden, and M. W. Casewell Clinical and bacteriological findings in necrotizing enterocolitis: a controlled study. J. Infect. Dis. 7: Stark, P. L., A. Lee, and B. D. Parsonage Colonization of the large bowel by Clostridium difficile in healthy infants: quantitative study. Infect. Immun. 35: Stephen, J., S. C. Redmond, T. J. Mitchell, J. Ketley, D. C. A. Candy, D. W. Burdon, and R. Daniel Clostridium difficile enterotoxin (toxin A): new results. Biochem. Soc. Trans. 12: Sterne, M., and L. M. Wentzel A new method for the large-scale production of high-titre botulinum formol-toxoid types C and D. J. Immunol. 65: Sugiyama, T., M. Mukai, R. Yamashita, and K. Sunakawa Experimental models of Clostridium difficile enterocolitis in gnotobiotic mice, p In Germfree research: microflora control and its application to the biomedical sciences. Alan R. Liss, Inc., New York Sullivan, N. M., S. Pellett, and T. D. Wilkins Purification and characterization of toxins A and B of Clostridium difficile. Infect. Immun. 35: Svedhem, A., B. Kaijser, and I. MacDowall Intestinal occurence of Campylobacter fetus subspecies jejuni and Clostridium difficile in children in Sweden. Eur. J. Clin. Microbiol. 1: Swartzberg, J. E., R. M. Maresca, and J. S. Remington Clinical study of gastrointestinal complications associated with clindamycin therapy. J. Infect. Dis. 135:S99-S Swindlehurst, J., T. G. Tachovsky, and M. Knoppers Antibiotics are necessary for plasmid isolation from Clostridium difficile, p In D. L. Lark (ed.), Protein-carbohydrate interactions in biological systems. The molecular biology of microbial pathogenesis. Academic Press, Inc., New York Taylor, N. S., G. M. Thorne, and J. G. Bartlett Com-

18 18 LYERLY ET AL. parison of two toxins produced by Clostridium difficile. Infect. Immun. 34: Teasley, D. G., D. N. Gerding, M. M. Olson, L. R. Peterson, R. L. Begbard, M. J. Schwartz, and J. T. Lee, Jr Prospective randomized trial of metronidazole versus vancomycin for Clostridium difiile-associated diarrhoea and colitis. Lancet ii: Tedesco, F. J Clindamycin-associated colitis. Review of the clinical spectrum of 47 cases. Digest. Dis. 21: Tedesco, F. J., R. W. Barton, and D. H. Alpers Clindamycin-associated colitis: a prospective study. Ann. Intern. Med. 81: Tedesco, F. J., J. Napier, W. Gamble, T. W. Chang, and J. G. Bartlett Therapy of antibiotic-associated pseudomembranous colitis. J. Clin. Gastroenterol. 1: Thelestam, M., and M. Bronnegard Interaction of cytopathogenic toxin from Clostridium difficile with cells in tissue culture. Scand. J. Infect. Dis. 22: Thelestam, M., and I. Florin Cytopathogenic action of Clostridium difficile toxins. J. Toxicol. Toxin Rev. 3: Toffler, R. B., E. G. Pingoud, and M. I. Burrell Acute colitis related to penicillin derivatives. Lancet ii: Toothaker, R. D., and G. W. Elmer Prevention of clindamycin-induced mortality in hamsters by Saccharomyces boulardii. Antimicrob. Agents Chemother. 26: Torres, J. F., R. Cedillo, J. Sanchez, C. Dillman, S. Giono, and 0. Munoz Prevalence of Clostridium difficile and its cytotoxin in infants in Mexico. J. Clin. Microbiol. 20: Trnka, Y. M., and J. T. LaMont Association of Clostridium difficile toxin with symptomatic relapse of chronic inflammatory bowel disease. Gastroenterology 80: Vesikari, T., E. Isolauri, M. Maki, and P. Gronroos Clostridium difficile in young children. Acta Pediatr. Scand. 73: Viscidi, R., S. Wiley, and J. G. Bartlett Isolation rates and toxigenic potential of Clostridium difficile isolates from various patient populations. Gastroenterology 81: Viscidi, R. P., and J. G. Bartlett Antibiotic-associated pseudomembranous colitis in children. Pediatrics 67: Wada, N., N. Nishida, S. Iwaki, H. Ohi, T. Miyawaki, N. Taniguchi, and S. Migita Neutralizing activity against Clostridium difficile toxin in the supernatants of cultured colostral cells. Infect. Immun. 29: Wald, A., H. Mendelow, and J. B. Bartlett Nonantibiotic associated pseudomembranous colitis due to toxin producing clostridia. Ann. Intern. Med. 92: Walker, R. C., P. J. Ruane, J. E. Rosenblatt, D. M. Lyerly, CLIN. MICROBIOL. REV. C. A. Gleaves, T. F. Smith, P. F. Pierce, Jr., and T. D. Wilkins Comparison of culture, cytotoxicity assays, and enzymelinked immunosorbent assay for toxin A and toxin B in the diagnosis of Clostridium difficile-related enteric disease. Diagn. Microbiol. 5: Welkon, C. J., S. S. Long, C. M. Thompson, Jr., and P. H. Gilligan Clostridium difficile in patients with cystic fibrosis. Am. J. Dis. Child. 139: West, S. E. H., and T. D. Wilkins Problems associated with counterimmunoelectrophoresis assays for detecting Clostridium difficile toxin. J. Clin. Microbiol. 15: Wilson, K. H., and R. Freter Suppression of Clostridium difficile by isolates from indigenous hamster flora. Microecol. Ther. 15: Wilson, K. H., M. J. Kennedy, and F. R. Fekety Use of sodium taurocholate to enhance spore recovery on a medium selective for Clostridium difficile. J. Clin. Microbiol. 15: Wilson, K. H., J. N. Sheagren, R. Freter, L. Weatherbee, and D. Lyerly Gnotobiotic models for study of the microbial ecology of Clostridium difficile and Escherichia coli. J. Infect. Dis. 153: Wilson, K. H., J. Silva, and F. R. Fekety Suppression of Clostridium difficile by normal hamster cecal flora and prevention of antibiotic-associated cecitis. Infect. Immun. 34: Wu, T. C., and J. C. Fung Evaluation of the usefulness of counterimmunoelectrophoresis for diagnosis of Clostridium difiile-associated colitis in clinical specimens. J. Clin. Microbiol. 17: Wu, T. C., and S. M. Gersch Evaluation of a commercial kit for the routine detection of Clostridium difficile cytotoxin by tissue culture. J. Clin. Microbiol. 23: Wu, T. C., V. P. McCarthy, and V. J. Gill Isolation rate and toxigenic potential of Clostridium difficile isolates from patients with cystic fibrosis. J. Infect. Dis. 148: Yamakawa, K., S. Nakamura, and S. Nishida Separation of two cytotoxins of Clostridium sordellii strains. Microbiol. Immunol. 29: Yamakawa, K., S. Nishida, and S. Nakamura C2 toxicity in extract of Clostridium botulinum type C spores. Infect. Immun. 41: Young, G. P., P. B. Ward, N. Bayley, D. Gordon, G. Higgins, J. A. Trapani, M. I. McDonald, J. Labrooy, and R. Hecker Antibiotic-associated colitis due to Clostridium difficile: double-blind comparison of vancomycin with bacitracin. Gastroenterology 89: