Interaction of Clostridium difficile and Escherichia coli with Microfloras in Continuous-Flow Cultures and Gnotobiotic Mice

Transcription

1 INFECTION AND IMMUNITY, Nov. 1986, p /86/ $02.00/0 Copyright C) 1986, American Society for Microbiology Vol. 54, No. 2 Interaction of Clostridium difficile and Escherichia coli with Microfloras in Continuous-Flow Cultures and Gnotobiotic Mice KENNETH H. WILSON'* AND ROLF FRETER2 Infectious Diseases Section and Medical Service, Veterans Administration Medical Center,'* and Department of Medicine and Department of Microbiology and Immunology, University of Michigan,2 Ann Arbor, Michigan Received 16 December 1985/Accepted 15 August 1986 We studied the interactions between the entire cecal flora of hamsters and the pathogens Clostridium difficile and Escherichia coli in gnotobiotic mice and in a continuous-flow (CF) culture system in which the growth medium consisted of an extract of fecal pellets from germfree mice. CF cultures and germfree mice were colonized first with C. difficile and E. coli and then with the cecal flora of hamsters. Both in vivo and in vitro hamster flora markedly suppressed the potential pathogens. Contents of CF cultures inoculated with hamster flora were introduced into gnotobiotic mice previously colonized with C. difficile and E. coli. These mice were compared with mice given homogenates of hamster ceca. In both groups, the C. difficile population decreased by a factor of more than 106 and the E. coli population decreased by a factor of 104 to 105. CF culture contents also reduced the size of the dilated germfree mouse cecum to normal. When veal infusion broth was used as a medium, contents of CF cultures colonized with hamster flora failed to eliminate C. dificile from mice. Thus, the extract of fecal pellets appeared to contain a substance important for sustained colonization by important components of the cecal flora. We also studied the ability of collections of isolates to suppress the potential pathogens in both gnotobiotic mice and CF cultures. A total of 150 isolates obtained from predominant hamster flora at the ecologic climax stage (C flora) suppressed C. difficile and E. coli to 10 and 1 to 3%, respectively, of the population sizes attained in monoassociated mice. A total of 67 isolates obtained during ecologic succession combined with a C flora consisting of 100 isolates suppressed the potential pathogens to 0.3 and 0.03% of their original levels, respectively. Similar degrees of suppression occurred in CF cultures, further indicating that anaerobic CF cultures are promising models for investigation of the microbial ecology of C. difficile. A prerequisite for the development of Clostridium difficile colitis is the presence of a sufficiently large population of the causative agent. The size of the C. difficile population, in turn, is determined by the interactions between this pathogen and the rest of the microflora. When introduced into germfree mice, C. difficile rapidly establishes a stable population of around 109 CFU per cecum (19, 28); when indigenous mouse or hamster flora is subsequently introduced into the mice, the pathogen is then suppressed to undetectable levels (27, 28). The indigenous flora thus modulates the size of the C. difficile population over a 108-fold range. The present study was undertaken to establish the feasibility of studying this interaction by using a continuous-flow (CF) culture system as an in vitro model of the interactions between C. difficile and the indigenous flora and by using synthetic microfloras to suppress C. difficile in gnotobiotic animals. The former approach could be used to study the mechanisms by which the flora suppresses C. difficile, the latter to establish which components of the flora exert this effect. The mechanisms by which the colonic microflora normally antagonizes C. difficile are clearly of interest, but these control mechanisms are difficult to study in vivo, where precise manipulation of the physicochemical environment is not possible. On the other hand, it is well known that bacterial interactions which do not occur in the colon are often seen in in vitro systems used to study the microbial ecology of the colon (10, 13, 15) and that in vitro systems often fail to reproduce interactions that do occur in vivo (10). Uncritical use of such systems has been a major impediment * Corresponding author. 354 to understanding the microbial ecology of the colon (7). Several studies have indicated that CF cultures may be used as models of the colonic ecosystem (8, 16, 22) and are useful in allowing in vitro studies of the mechanisms of bacterial interactions (6). In the present study, we performed experiments to test the ability of our CF culture system to reproduce interactions between C. difficile and the colonic flora of hamsters. We compared bacterial interactions in gnotobiotic mice with those in CF cultures. Suppression of both Escherichia coli and C. difficile by complete cecal flora (taken directly from animals without intervening culturing) as well as by collections of isolates from native hamster flora was used to make this comparison. In addition, we tested the ability of CF cultures to allow propagation of all the species important for the suppression of C. difficile in vivo. The colonic flora of adult humans appears to consist of 400 to 500 species of bacteria (12), and rodent flora is probably equally complex (8). Because of this complexity, it has been difficult in the past to study the interactions between C. difficile and a synthetic microflora consisting of pure cultures of isolates. Synthetic floras must themselves be very complex to simulate the functions of natural floras (5). The microflora of the colon in all species studied is dominated by large numbers of very fastidious anaerobes, all of which are exceedingly difficult to cultivate. It is well established that the predominant anaerobic flora plays a major role in the suppression of other potentially pathogenic organisms (5). The present study was initially designed to determine the ability of isolates from the predominant anaerobic flora of the hamster cecum to suppress C. difficile. However, large numbers of isolates from the predominant anaerobic flora alone did not have a striking effect. Implantation of the

2 VOL. 54, 1986 C. DIFFICILE, E. COLI, AND SYNTHETIC MICROFLORAS 355 predominant anaerobic flora in nature occurs by an elaborate process of ecologic succession. During this process, the composition of the microflora continuously changes as one group of microbes becomes Rumerically dominant, only to be supplanted by a new group of organisms, which is in turn supplanted, and so on (14, 20, 21). This process (ecologic succession) occurs as a relatively orderly progression through stages and ultimately culminates in the establishment of the climax-stage ecosystem, dominated by many species of anaerobes. It seemed reasonable to explore the possibility that simulating this process in an experimental situation might be a useful way to obtain a flora able to suppress C. difficile. Syrian hamsters are the best-studied model of C. difficile colitis. Since a germfree hamster has never been developed, we used germfree mice for gnotobiotic studies involving hamster flora. Although it is clear that if two species of host are dissimilar enough the colonic flora of one will not function normally in the other (23), hamsters and mice appear to be closely enough related that crossing species lines does not present this problem (28). The use of mice also has the advantage that they are relatively resistant to C. difficile toxin (19, 28) and, therefore, an inflammatory reaction of the colonic mucosa is not a variable which confounds the investigation of bacterial interactions. In all of the studies described below, we used E. coli as well as C. difficile because E. coli is the best-studied organism in experiments with synthetic microfloras (5) and in experiments evaluating CF culture systems (6, 8). Thus, the behavior of E. coli in the present studies allows a comparison with the results of previous work. MATERIALS AND METHODS Animals. Syrian hamsters (70 to 100 g) were obtained from Charles River Breeding Laboratories, Inc., Wilmington, Mass. Germfree CD-1 mice were originally obtained from Charles River Breeding Laboratories and then maintained and bred in germfree Texler-type polyvinyl germfree isolators. Conventional BALB mice were obtained from William Murphy of the Department of Microbiology and Immunology, University of Michigan, Ann Arbor. Bacterial isolates. The strain of C. difficile used in these studies, 49A, was originally isolated from a hamster with cecitis. It is toxigenic and causes colitis in hamsters but not in mice (28). E. coli C25 is a streptomycin-resistant mutant of an isolate from human feces and has been described in several previous studies (5, 7, 8). Isolates from hamster flora were obtained by sacrificing 70- to 100-g animals (Charles River Breeding Laboratories) with CO2 anesthesia and taking them intact into an anaerobic chamber, where ceca were homogenized in 30 ml of prereduced Trypticase soy broth (TSB) (BBL Microbiology Systems, Cockeysville, Md.). Serial 10-fold dilutions were plated on modified Bryant medium 10 (4) and Aranki-Freter medium (1) and incubated inside the chamber for 5 days at 37 C in plastic bags containing palladium-coated aluminum pellets. Microscopic counts of cecal homogenates were performed in a Petroff- Hausser chamber. Colonies were picked only if the viable count was greater than 40% of the microscopic count. Isolates were picked at random from the primary isolation medium to slants of modified Bryant medium 10 and subcultured repeatedly until heavy growth was obtained. The growth was then washed off the slants with TSB; all isolates were pooled, and 0.05% dithiothreitol was added. The suspension was then placed in a glass vial, sealed anaerobically, and sterilized externally with 2% peracetic acid in the lock of a Trexler-type polyvinyl germfree isolator. The contents of the vial were then administered to gnotobiotic mice. A collection of isolates from the cecal flora dominating the climax-stage ecosystem will be referred to as a C flora. In some experiments, isolates from hamster flora were obtained during ecologic succession and will be referred to as an S flora. The S flora was obtained by inoculating a CF culture with hamster cecal contents. The CF culture was then cultured quantitatively on modified Bryant medium 10 and on Aranki-Freter medium at 2, 6, 10, and 12 days after inoculation of hamster cecal contents into the system. Approximately 20 colonies were picked from each medium at each time point in an attempt to sample all types of colony morphology. Isolates were passed in subcultures as described above. Homogenates of hamster ceca were prepared by taking C02-anesthetized hamsters into the anaerobic chamber, homogenizing the ceca in 30 ml of prereduced TSB, and then adding 0.05% dithiothreitol. Inoculation of mice. Overnight broth cultures of C. difficile and E. coli were administered to germfree CD-1 mice. Mice were kept in sterile Trexler-type polyvinyl isolators with access through a stainless steel lock sterilized with 2% peracetic acid. Animals were inoculated intragastrically through a steel feeding needle. To buffer gastric acid prior to inoculation with isolates from hamster flora, we gave mice already associated with E. coli and C. difficile 0.5 ml of a sterile solution of 7.5% sodium bicarbonate equilibrated with CO2. Then, each animal received 0.5 ml of the bacterial suspension intragastrically and 1.0 ml rectally. Mice were inoculated with bacterial suspensions three times over a 2-week period, and the floras were then allowed to equilibrate for 3 weeks before mice were sacrificed by cervical dislocation for quantitation of C. difficile, E. coli, and cecal size. Quantitative cultures. Population sizes of C. difficile were measured by sacrificing mice, taking them into the anaerobic chamber, homogenizing the ceca in 30 ml of prereduced TSB, and plating serial dilutions on cefoxitin-cycloserinefructose agar (9) containing 0.1% sodium taurocholate (25). Dilution tubes were then taken from the anaerobic chamber, and dilutions were inoculated onto deoxycholate agar containing 100,ug of streptomycin per ml for E. coli counts. CF cultures. The CF culture system has been described previously (8). Briefly, it consisted of a culture vessel containing 7 ml of broth medium fed by means of a peristaltic pump at a continuous dilution rate of /h. The culture vessels and media were kept in an anaerobic chamber heated to 37 C. The culture medium used in the present studies was designed to simulate the "natural" substrate of the cecal flora as closely as possible. Fecal pellets from germfree mice were washed free of hair and food particles. An equal volume of a balanced salt solution was added to loosely packed, washed fecal pellets, and the pellets were homogenized in a Waring blender. The balanced salt solution was a modification of one described previously (11) and contained, per liter, anhydrous CaCl2 (0.2 g), anhydrous MgSO4 (0.2 g), K2HPO4 (1.0 g), NaHCO3 (3.75 g), and NaCl (6.35 g). The homogenate was then centrifuged for 30 min at 2,300 x g, and an equal volume of distilled water was added to the supernatant. NaHCO3 (10 g) was then added per liter, and the resulting medium was autoclaved for 40 min at 121 C and allowed to equilibrate with the gas mixture within the anaerobic chamber (85% N2, 10% H2, 5% C02) for 48 h prior

3 356 WILSON AND FRETER TABLE 1. Flora in CF culture (no. Population sizes of E. coli and C. difficile in anaerobic CF cultures containing various floras Mean log1o CFU/ml ± SEM (range) of: of cultures) E. coli C. difficile C. difficile alone (4) 8.3 ± 0.3a ( ) C. difficile and E. coli 9.5 ± 0.08 ( ) a ( ) (9) C. difficile and E. coli 5.0 ± 1.2 (0-8.1) 2.7 ± 0.6 (0-5.5) with hamster flora (9)b a P < by the Student t test. bthree of these CF cultures did not contain E. coli. to use. CF cultures were inoculated first with C. difficile and E. coli and then either with isolates from cecal flora or with a small piece of cecal wall and adherent cecal contents. Inoculation was performed three times in a 2-week period. After inoculation with whole flora or isolates, CF cultures were allowed to equilibrate for at least 2 weeks. To associate gnotobiotic mice with contents of CF cultures, we diluted the contents by a factor of 10 with prereduced TSB, added 0.05% dithiothreitol, sealed the suspension anaerobically in a glass vial, and administered it to mice as described for bacterial isolates. Statistical analysis. Population sizes were converted to log1o values prior to statistical testing because log1o values are distributed normally and most parametric statistical tests assume that data are normally distributed (2). The Student t test was used for comparisons. RESULTS CF cultures. Three CF cultures containing enriched veal infusion broth as a medium were inoculated with the cecal flora of hamsters and allowed to equilibrate. The contents of these CF cultures were then administered to mice that had been monoassociated with C. difficile. The C. difficile population was suppressed from to (loglo CFU per cecum ± standard error of the mean for 12 mice). Mice associated with the flora from the CF cultures harbored less than 10% of the C. difficile population observed in the ceca of monoassociated mice. This degree of suppression was far lower, however, than that caused by association with an entire cecal flora, i.e., with homogenates of hamster ceca which, as shown earlier, caused a total elimination of C. difficile from the cecum (28). The clear implication was that species or subspecies of bacteria important for the suppression of C. difficile had failed to colonize these CF cultures. For this reason, further experiments were performed with a habitat-simulating culture medium consisting of extracts of fecal pellets from germfree mice. In fecal pellet extract medium, C. difficile growing alone attained a population size of 8.3 ± 0.3 (mean log1o CFU per milliliter ± standard error of the mean); the population size attained in the presence of E. coli ( ) was lower (Table 1; t = 3.73, P < 0.005). When CF cultures were inoculated with hamster flora, both E. coli and C. difficile were suppressed to low levels (Table 1). The degree of suppression of E. coli was variable, ranging from undetectable levels (<10 CFU/ml) to CFU/ml, and the range of C. difficile populations was also wide. However, in the case of C. difficile, despite this large variation, the largest population in the presence of hamster flora (log 5.5) was still only around 0.1% of the mean level attained in pure cultures (log INFECT. IMMUN. 8.3). Gram-stained smears of CF cultures harboring hamster flora showed a predominance of gram-negative and gramvariable rods and were very similar to gram-stained smears of hamster cecal contents. Two experiments were performed to compare gnotobiotic mice conventionalized by whole hamster flora with gnotobiotic mice conventionalized by CF culture contents (Table 2). Germfree mice were colonized with C. difficile and E. coli prior to receiving CF culture contents or cecal homogenates. The degree of suppression produced by CF culture contents was practically identical to that produced by whole hamster flora. CF culture contents also decreased the sizes of the dilated germfree mouse ceca to the same degree as did whole hamster flora. Suppression by synthetic microfloras. The mean recovery rate of viable bacteria from the colonic flora was 54% of the microscopic count. With the exception of some isolates obtained during the early stages of ecologic succession, all isolates, including all those in the C floras, were anaerobic. Gram-stained smears of cecal contents of mice associated with the synthetic floras appeared very similar to those of cecal contents from conventional mice or hamsters and were dominated by tapered, palisading gram-negative rods. E. coli alone suppressed C. difficile to 10% of the level jt attained in monoassociated mice (Table 3). It mnade little difference whether a C flora was introduced directly into gnotobiotic mice or was introduced after propagation in a CF culture; the suppression of C. difficile and E. coli was of the same order of magnitude in either case. The C floras suppressed E. coli to 0.8 to 3.2% of the level it attained in monoassociated mice or in mice associated with both C. difficile and E. coli. However, the C floras suppressed C. difficile only slightly, again to 10% of the level it achieved in monoassociated mice. When association with an S flora preceded the introduction of a C flora, 99.7% of the C. difficile population and 99.97% of the E. coli population were eliminated from mice. When data from the C-flora groups TABLE 2. Comparison of CF cultures colonized by cecal flora from hamsters, gnotobiotic mice associated with CF culture contents, and gnotobiotic mice associated with hamster floraa Cecal size (% Mean log1o CFU/cecum or ml of mice) wt ± SEM) E. coli C. difficile Experimental group of total body ± SEM of: Expt(no. 1 Mice associated with 5.2 ± ± 0.1 <7.6 ± 0.1 C. difficile and E. coli (6) 2 CF culture A Undetectable 1.0 CF culture-associated 2.6 ± ± 0.3 <2.5 mice (6)b Homogenate-associated 2.7 ± ± 0.3 <2.5 mice (6)c 3 CF culture B Not doned 2.2 CF culture-associated e 4.8 ± 0.2 <2.5 mice (8)b Homogenate-associated 1.8 Q.1e 5.2 ± 0.3 <2.5 mice (8)c a Extracts of germfree fecal pellets were used as the CF culture medium. b Mice associated with contents of CF cultures containing hamster cecal flora. c Mice associated with anaerobically prepared homogenates of hamster ceca. d The original CF culture was not inoculated with E. coli, but gnotobiotic mice were associated with E. coli prior to receiving CF culture contents. e P < 0.05 by the Student t test.

4 VOL. 54, 1986 C. DIFFICILE, E. COLI, AND SYNTHETIC MICROFLORAS 357 TABLE 3. Effects of isolates and whole hamster flora on population sizes of E. coli and C. difficile in CF cultures and gnotobiotic mice % Population size (actual loglo CFU/ml or cecum + SEM) ofa: Expt E. coli in: C. difficile in: CF cultures Mice CF cultures Mice Monoassociated Diassociatedb 100 ( ) 100 (10.1 ± 0.1) 10 (7.4 ± 0.1) 10 (7.6 ± 0.1) C flora-1c 0.6 (7.3) 0.8 (8.0 ± 0.1) 1.6 (6.5) 10 (7.6 ± 0.1) C flora-2d 1.3 (7.6) 3.2 (8.6 ± 0.2) 0.3 (5.8) 10 (7.6 ± 0.05) S flora + C florae (4.2) 0.03 (6.5 ± 0.1) 0.03 (4.9) 0.3 (6.1 ± 0.1) Whole floraf (5.0) (5.2 ± 0.3) (2.7) < a Results are expressed as a percentage of the population size that the organism attained when colonizing alone. Except for diassociated CF cultures (n = 9), each result for CF cultures represents 1 CF culture; the number of mice varied with the experiment, as noted below. b Germfree mice (n = 6) were associated with C. difficile and E. coli. c 150 C-flora isolates were inoculated into a CF culture, and the contents were used to inoculate mice (n = 8) colonized with E. coli and C. d difficile. A separate collection of 149 C-flora isolates were inoculated simultaneously into CF cultures and gnotobiotic mice (n = 4). e 67 S-flora isolates were inoculated into CF cultures and were followed 4 days later by 100 C-flora isolates. CF culture contents were then inoculated into mice (n = 6) colonized with E. coli and C. difficile. f Gnotobiotic mice (n = 14) colonized with C. difficile and E. coli were inoculated with homogenates of whole hamster flora. were pooled and population levels of E. coli and C. difficile in the C-flora groups were compared with levels in the S-flora + C-flora group, P was less than for both comparisons (for E. coli, T = 7.530; for C. difficile, T = 9.383). The most complex flora, whole hamster flora, had the greatest suppressive effect. In general, results in CF cultures were parallel to those in gnotobiotic mice. Exceptions to this rule included a tendency for synthetic floras to suppress C. difficile more in CF cultures than in gnotobiotic mice. DISCUSSION Recently, several studies have shown that the colonic microflora appears to remain stable when introduced into various CF culture systems (8, 16, 22). However, since the colonic flora consists of hundreds of species of anaerobic bacteria which are difficult to isolate and difficult to identify, it is not easy to determine how similar a flora in a CF culture is to an original colonic flora. Hamster flora presents a special problem in that the predominant anaerobes have not been characterized previously and, if past experience with mouse flora is any indication (8, 24), most components of this flora belong to no known taxa. The present study stops short of a detailed cataloging of large numbers of isolates from this complex, uncharacterized microflora. Nonetheless, it supplies strong evidence that the CF culture system described provides a useful model of the interactions of C. difficile with the cecal flora of hamsters. Normalization of the dilated germfree rodent cecum is a function of the cecal flora, as is suppression of E. coli to the low population size at which it is usually present in a complete flora. Total normalization of these parameters has been accomplished only by administering intestinal contents without intervening culturing. Thus, the fact that CF culture contents had essentially the same effect on these parameters as cecal homogenates had indicates that the flora transferred in the CF culture contents was probably about as complex as hamster cecal flora. Hamster flora has the same effect on these variables as mouse flora does (28), indicating that for the purposes of these experiments the two floras are interchangeable. Our results also indicated that all organisms important for the suppression of C. difficile indeed colonized the CF cultures. By 2 weeks after inoculation of cecal contents into the CF cultures, the original inoculum had been diluted by a factor of Thus, if organisms important for the suppression of C. difficile had not actually colonized the CF culture vessel, they would have been washed out many times over. Some component(s) of the fecal pellets from germfree animals appeared to be essential for colonization of a CF culture vessel by certain components of the colonic flora. This is another instance of the value of habitat-simulating media in the study of ecology. Given the stability of the human colonic flora in the face of alterations in diet (17), it is not surprising that some host factor(s) is possibly important in determining the composition of the colonic microflora. Identification of such a factor(s) could shed light on the fundamental aspects of the interrelationship between the host and its microflora. The experiments with synthetic microfloras also supported the use of our CF culture system as a model of the microbial ecology of C. difficile. The overall similarity of results obtained in vitro to those obtained in vivo is obvious (Table 3). In gnotobiotic animals, 150 isolates from predominant hamster flora reproducibly eliminated 90% of the C. difficile population. However, the remaining population was around 4 x 107 CFU per cecum, and previous studies (26, 29) have shown that colitis can occur in hamsters when the population size of C. difficile strain 49A exceeds 107. In an attempt to diversify the types of isolates present in a synthetic flora, we obtained 67 isolates from a CF culture recently inoculated with hamster flora. These isolates dominated the flora while bacterial populations had not yet reached a state of equilibrium. When these isolates were combined with 100 isolates from predominant hamster flora, C. difficile was suppressed to 106 CFU per cecum. This result represents an improvement over previous approaches. For instance, when nontoxigenic C. difficile was used to antagonize a toxigenic strain (3, 26), suppression occurred only when the nontoxigenic strain was established first. In the present study, the pathogen was established first and then was suppressed to nonpathogenic levels by a synthetic microflora. Our data show that a synthetic microflora can be devised to suppress C. difficile below pathogenic levels, but it is not simply a matter of randomly collecting a large number of isolates from the predominant colonic flora. The complexity of these experiments and the fact that they met with only partial success in suppressing C. difficile suggest that an empiric approach has serious limitations. Further knowledge must be gained about the composition of the cecal floras of

5 358 WILSON AND FRETER rodents and about mechanisms for suppression before a more analytical approach can be taken. ACKNOWLEDGMENTS This work was supported by the Veterans Administration and by Public Health Service grant RO1-AM33967 from the National Institutes of Health. We acknowledge the technical assistance of Patricia Permoad, Mayurika Patel, and Brenda Franklin and the secretarial assistance of Peggy Wood and Lisa Aja. LITERATURE CITED 1. Aranki, A., S. A. Syed, E. B. Kenny, and R. Freter Isolation of anaerobic bacteria from human gingiva and mouse cecum by means of a simplified glove box procedure. Appl. Microbiol. 17: Best, W. R On the logarithmic transformation of intestinal bacterial counts. Am. J. Clin. Nutr. 23: Borriello, S. P., and F. E. Barclay Protection of hamsters against Clostridium difficile ileocaecitis by prior colonization with non-pathogenic strains. J. Med. Microbiol. 19: Eller, C., M. R. Crabill, and M. P. Bryant Anaerobic roll tube media for nonselective enumeration and isolation of bacteria in human feces. Appl. Environ. Microbiol. 22: Freter, R., and G. D. Abrams Function of various intestinal bacteria in converting germfree mice to the normal state. Infect. Immun. 6: Freter, R., H. Brickner, M. Botney, D. Cleven, and A. Aranki Mechanisms that control bacterial populations in continuous-flow culture models of mouse large intestinal flora. Infect. Immun. 39: Freter, R., H. Brickner, J. Fekete, M. M. Vickerman, and K. E. Carey Survival and implantation of E. coli in the intestinal tract. Infect. Immun. 39: Freter, R., E. Stauffer, D. Cleven, L. V. Holdeman, and W. E. C. Moore Continuous-flow cultures as in vitro models of the ecology of large intestinal flora. Infect. Immun. 39: 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: Hentges, D., and R. Freter In vivo and in vitro antagonism of intestinal bacteria against Shigella flexneri. Correlation between various tests. J. Infect. Dis. 110: Holdeman, L. V., E. P. Cato, and W. E. C. Moore (ed.) Anaerobe laboratory manual, 4th ed. Virginia Polytechnic Institute and State University, Blacksburg. 12. Holdeman, L. V., I. J. Good, and W. E. C. Moore Human fecal flora: variation in bacterial composition within individuals and a possible effect of emotional stress. Appl. Environ. Microbiol. 31: Ikari, N. S., D. M. Kenton, and V. M. Young Interactions in the germfree mouse intestine of colicinogenic and colicinsensitive microorganisms. Proc. Soc. Exp. Biol. Med. INFECT. IMMUN. 130: Lee, A., and E. Gemmell Changes in the mouse intestinal microflora during weaning: role of volatile fatty acids. Infect. Immun. 5: Maier, B. R., and D. J. Hentges Experimental Shigella infections in animals. Infect. Immun. 6: Miller, T. L., and M. J. Wolin Fermentation by the human large intestine microbial community in an in vitro semicontinuous culture system. Appl. Environ. Microbiol. 42: Moore, W. E. C., E. P. Cato, I. J. Goode, and C. V. Holdeman The effect of diet on the human fecal flora, p In W. R. Bruce (ed.), Gastrointestinal cancer: endogenous factors. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 18. Nakamura, S., S. Hakashio, T. Inamatsu, N. Nishida, N. Taniguchi, and S. Nishida Toxigenicity of Clostridium difficile isolates from patients and health adults. Microbiol. Immunol. 24: Onderdonk, A. B., R. L. Cisneros, and J. G. Bartlett Clostridium difficile in gnotobiotic mice. Infect. Immun. 28: Rolfe, R., and J. P. laconis Intestinal colonization of infant hamsters with Clostridium difficile. Infect. Immun. 42: Schaedler, R. W., R. Dubos, and R. Costello The development of the bacterial flora in the gastrointestinal tract of mice. J. Exp. Med. 122: Veilleux, B. G., and I. Rowland Stimulation of the rat intestinal ecosystem using a two-stage continuous culture system. J. Gen. Microbiol. 123: Weinack, 0. M., G. H. Soeembos, C. F. Smyser, and A. S. Soerjadi Reciprocal competitive exclusion of Salmonella and Escherichia coli by native intestinal microflora of the chicken and turkey. Avian Dis. 26: Wilkins, T. D Microbiological considerations in interpretation of data obtained with experimental animals, p In W. R. Bruce (ed.), Gastrointestinal cancer: endogenous factors. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 25. 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., and J. N. Sheagren Antagonism of toxigenic Clostridium difficile by nontoxigenic C. difficile. J. Infect. Dis. 147: Wilson, K. H., J. N. Sheagren, and R. Freter Population dynamics of ingested Clostridium difficile in the gastrointestinal tract of the Syrian hamster. J. Infect. Dis. 151: 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 E. 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: