Screening Medium for Detecting Escherichia coli 0157:H7 Associated with Hemorrhagic Colitis

Transcription

1 JOURNAL OF CLINICAL MICROBIOLOGY, OCt. 1985, p /85/ $02.00/0 Copyright C 1985, American Society for Microbiology Vol. 22, No. 4 H7 Antiserum-Sorbitol Fermentation Medium: a Single Tube Screening Medium for Detecting Escherichia coli 0157:H7 Associated with Hemorrhagic Colitis J. J. FARMER III* AND BETTY R. DAVIS Enteric Bacteriology Section, Center for Infectious Diseases, Centers for Disease Control, Atlanta, Georgia Received 29 April 1985/Accepted 3 July 1985 Escherichia coli serotype 0157:H7 has been isolated from outbreaks and sporadic cases of hemorrhagic colitis. There is convincing evidence that it can cause this diarrheal disease. Because of the interest in hemorrhagic colitis, it has become desirable to detect this particular strain in human feces, which usually contains many other strains of E. coli. Two characteristics of the incriminated E. coli 0157:H7 strain have made its isolation and identification easier. It does not ferment D-sorbitol rapidly, in contrast to about 95% of other E. coli strains. In addition, the strain has H antigen 7, but only about 10% of other E. coli strains have this particular antigen. To screen for E. coli 0157:H7 we devised H7 antiserum-sorbitol fermentation medium (18 g of enteric fermentation base, 10 g of D-sorbitol, 4 g of agar, 10 ml of Andrade indicator, 989 ml of water; all ingredients were mixed, autoclaved, and cooled; 1 ml of E. coli H7 antiserum was then added). Colonies to be screened were inoculated into this medium. Strains of E. coli 0157:H7 gave a characteristic pattern; they did not ferment sorbitol and were immobilized in the semisolid medium because of the reaction of their flagella with the flagella antiserum. Almost all other strains of E. coli gave a different pattern; they fermented sorbitol or were not immobilized by the H7 serum or both. Strains which were presumptive positives (sorbitol negative, H7 positive) were then tested in E. coli 0157 serum by slide or tube agglutination. The number of strains which were presumptive positive by H7-sorbitol medium but then were not found to be 0157 was less than 1%. A second approach has been helpful in deciding which colonies,to screen in H7-sorbitol medium. MacConkeysorbitol agar (22.2 g of MacConkey agar base [which contains no sugar], 10 g of D-sorbitol, 1,000 ml of water) was designed as a plating medium. Stools were plated on MacConkey agar to estimate the number of E. coli colonies and also plated on MacConkey-sorbitol agar to estimate the number of sorbitol-negative colonies of E. coli. These two approaches have proved useful for isolating and identifying E. coli 0157:H7 from human feces and from feces of animals infected in the laboratory with this strain. The results suggest that media may be formulated in a similar fashion for detecting other specific strains of E. coli. In 1982, there were two outbreaks of an unusual form of bloody diarrhea called hemorrhagic colitis (10). This disease was characterized by the sudden onset of severe abdominal pain with cramps, followed within 24 h by watery diarrhea, which subsequently became bloody. The stools were described by some of the patients as "all blood and no stool." Recovery was usually within a few days without complication or specific therapy. Originally, the stool specimens were negative for all the usual enteric pathogens, but eventually a strain of Escherichia coli 0157:H7 was implicated as the causative agent in both outbreaks (10, 12). This rare E. coli strain was also isolated from raw hamburger which was the incriminated source in each outbreak (12), and was shown to cause diarrhea in the infant rabbit animal model for diarrhea (4). E. coli 0157:H7 also produces a toxin which is cytotoxic to the Vero cell culture line (5, 7). This toxin is very similar or identical to the "Shiga toxin" which is produced by Shigella dysenteriae 1, other strains of the genus Shigella, and E. coli (7). Toxin production in these strains is associated with lysogeny by one of two closely related bacteriophages (8). The production of this shiga-like toxin may be an important aspect of pathogenesis (8). Interest in E. coli 0157:H7 increased when it was also found in sporadic cases of hemorrhagic colitis and in cases of * Corresponding author. 620 hemolytic uremic syndrome (9). Experiments are being done in animals to study the pathogenic mechanisms of these E. coli strains. In both of these diseases and in animal experiments, it would be helpful to detect E. coli 0157:H7 in the presence of other E. coli strains which normally inhabit the intestine. The purpose of this study was to devise a simple procedure to screen for E. coli 0157:H7. We took advantage of two of its properties that are absent in most other strains of E. coli: its flagellar antigen H7 and its inability to ferment D-sorbitol after overnight incubation. In this paper, we describe a new screening medium for E. coli 0157:H7 and show how this general approach can be used to devise other single-tube media for detecting specific strains of E. coli in epidemic investigations or special studies. In addition, we describe a plating medium with added sorbitol that is useful in detecting colonies of E. coli 0157:H7 in the presence of other E. coli colonies. MATERIALS AND METHODS General procedures. Unless exceptions are given, the following statements hold throughout this paper: the temperature of incubation was 36 ± 1 C; commercial media were used whenever possible (the terms "from individual ingredients" or "was made with" appear if a commercial medium was not used); media were sterilized in an autoclave

2 VOL. 22, 1985 H7 ANTISERUM-SORBITOL FERMENTATION MEDIUM 621 FIG. 1. At left, reaction in H7-sorbitol medium of the epidemic strain of E. coli 0157:H7; the strain is immobilized. At right, a strain of E. coli which does not have H7 antigen: it is not immobilized. at 121 C for 15 min; and refrigeration was at a temperature of C. Strains. Biochemical reactions (3) were tested on 64 different cultures of E. (0/li 0157:H7 from stock cultures, outbreaks, and sporadic cases. The tube test for D-sorbitol fermentation was incubated for 7 days before it was discarded as negative. E. coli 0157:H7 A8187-M5 (Michigan outbreak) was used in all the animal experiments. Strains of E. coli from control subjects without diarrhea were also used. The distribution of different H (flagella) antigens among E. coli strains was based on 573 strains (388 motile; 185 nonmotile) which had been serotyped at the Enteric Bacteriology Section, Centers for Disease Control, Atlanta, Ga., between 1972 and Duplicate isolates of the same strain were excluded before these data were tabulated. The data for the fermentation reactions of E. (coli were based on over 1,000 strains which had been tested between 1972 and 1982 (some strains were not tested for each one of the sugars). A number of other strains were isolated from adult Rhesus monkeys and infant rabbits (5 to 21 days old) with and without diarrhea. Sera. All sera for determining the 0 and H antigens of E. coli were obtained from the Biological Products Division, Centers for Disease Control. They had been produced in rabbits by standard methods (2) and were preserved with an equal volume of glycerol (these sera were defined to be a 1:2 dilution because of the added glycerol). The individual sera used were E. (oli 0157 serum and E. coli H7 serum. Agglutination in E. coli 0157 serum. A colony was touched with a sterile loop and used to heavily inoculate 3 ml of Trypticase soy broth. This was grown for 5 to 6 h, during which time the culture became very turbid and went into the stationary phase of growth. The broth culture was then heated for 1 h at 1000C (in live steam in an Arnold sterilizer or in a beaker of boiling water). The heated suspension was then diluted by adding 5 volumes of formalinized saline (5 g of NaCl, 5 ml of commercial Formalin, 995 ml of water) to one volume of heated culture. This is referred to as the heated antigen. For the actual agglutination reaction, the following were mixed in a disposable glass tube (100 by 13 mm): 0.15 ml of heated antigen, 0.75 ml of formalinized saline, and 0.1 ml of a 1:100 dilution of E. c oli 0157 serum. This mixture was incubated in a 48 C water bath overnight and then read for visible agglutination when compared with a control tube (0.15 ml of heated antigen, 0.85 ml of formalinized saline). Slide agglutination. Slide agglutination was used to test colonies directly from plating media such as MacConkey agar, MacConkey-sorbitol agar, sheep blood agar, and Trypticase soy agar. A wooden applicator stick was used to touch the colony and remove as much of the growth as possible. The cells sticking to the bottom of the stick were then suspended in a small drop (about ml) of saline. An equal-sized drop of E. coli 0157 antiserum was added, the drops were mixed with applicator stick, and the slide was rocked back and forth for 1 min. The epidemic strain of E. coli 0157:H7 agglutinated rapidly and completely with E. coli 0157 serum. H antigen determination by immobilization. Immobilization for H antigen determination was described many years ago by Orskov (as quoted in reference 6) but has not been used very often in serotyping E. c(li. This technique is based on the fact that a motile strain of E. coli will migrate through a column of semisolid agar (0.4% agar) but will be immobilized (Fig. 1) if an antiserum to its flagella (H antiserum) is incorporated into the medium. Titration of E. coli H7 serum by H immobilization. The titration was done in Enteric Fermentation Base (Difco Laboratories, Detroit, Mich.) with 0.4% agar added. The medium was made, and 6.3 ml was dispensed into glass tubes (13 by 100 mm). These were autoclaved and cooled to 45 C, and 0.1-ml portions of various dilutions of H7 serum were added to give final serum dilutions of 1:64 to 1:65,536. The tubes were mixed and then placed in the refrigerator for 1 h to harden. They were removed from the refrigerator and allowed to warm to room temperature. A straight needle was used to touch the growth of E. coli 0157:H7 and to inoculate the center of each tube to a depth of about 0.3 cm. The tubes were then incubated overnight and compared with a control tube which contained no serum (Fig. 1). For each tube, we recorded the distance the culture migrated from the inoculation stab line. The tubes with the highest concentrations of serum strongly prevented motility; in contrast, the culture migrated to the bottom of the control tube without serum (Fig. 2). MacConkey-sorbitol agar. The epidemic strain of E. coli 0157:H7 fermented lactose but not D-sorbitol after overnight incubation. Thus, on lactose-containing media such as Mac- Conkey agar, the epidemic strain is indistinguishable from most fecal strains of E. (oli which were also lactose positive. However, there is a commercial medium, MacConkey agar base (Difco), which is identical to MacConkey agar except that it does not contain lactose. E. coli 0157:H7 does not ferment sorbitol (at 24 h), but about 95% of E. (0li strains do. Thus, MacConkey agar base plus 1% D-sorbitol (Mac- Conkey-sorbitol agar) was formulated to detect sorbitolnegative colonies of the epidemic strain in the presence of sorbitol-positive colonies of other E. c oli strains. Mac- Conkey-sorbitol agar was prepared by adding the following ingredients to a 2-liter flask: 22.2 g of MacConkey agar base,

3 622 FARMER AND DAVIS J. CLIN. MICROBIOL. 10 g of D-sorbitol (Sigma Chemical Co., St. Louis, Mo.), and 1,000 ml of water. A magnetic stirring bar (2 in. [5.08 cm]) was also added. The medium was autoclaved at 121 C for 20 min (to prevent boiling over, the autoclave door was opened about 3 min after the buzzer sounded), placed on a magnetic stirrer to mix the melted agar, cooled to 45 to 50 C, and then poured into plastic disposable petri dishes (25 ml each). After the dishes had cooled overnight, they were dried (laminar flow cabinet), placed in sterile bags, sealed, and then stored in the refrigerator until used. H7 antiserum-sorbitol fermentation medium (H7-sorbitol medium). This medium was designed as a single-tube screening test for the strain of E. coli 0157:H7. It contains D-sorbitol and E. coli H7 antiserum. In H7-sorbitol medium, the strain of 0157:H7 does not ferment sorbitol within 24 h and is immobilized; most other strains of E. coli are not immobilized because they have a flagella antigen other than H7 and they ferment sorbitol. H7-sorbitol medium contained 18 g of enteric fermentation base, which contains 10 g of peptone, 3 g of meat extract, and 5 g of NaCl. These were added to a 2-liter flask, and 100 ml of distilled water was added to make a wet paste. An additional 890 ml of water was added, a 2-inch magnetic stirring bar was put into the flask, and the mixture was placed on a magnetic stirrer until all the ingredients had dissolved (about 1 min); 4 g of agar, 10 g of D-sorbitol, and 10 ml of Andrade indicator (0.2 g of acid fuchsin, 100 ml of distilled water, 16 ml of 1 N NaOH) were then added. The medium was autoclaved at 121 C for 20 min, stirred on the magnetic stirrer about 1 min to evenly distribute the melted agar, and then cooled to 45 C for about 30 min in a 45 C water bath. One milliliter of glyc antiserum was added. The final medium can be dispensed 1:64 1:128 1:256 1:512 1:4096 1:8192 1:16,384 1:32,768 FIG. 2. Titration of H7 antiserum by H numbers represent the final dilution of H7 (control) contained no antiserum. The cross-h where E. coli 0157:H7 migrated away frc inoculation. 3 ml) of a plastic tissue culture dish (such as Linbro no from Flow Laboratories, McLean, Va., or Costar no from Costar, Cambridge, Mass.). The medium was allowed to solidify (about 1 h) and then was stored in sterile plastic bags at 4 C until used. When many colonies were to be screened, we preferred to use the 24-well plates, but we have also dispensed the medium into sterile glass tubes (13 by 100 mm) at 3 ml per tube. The medium works best when the surface is thoroughly dry. This can be accomplished in three ways: (i) by using it after storage at 4 C for 2 to 3 weeks, (ii) by drying it under a stream of sterile air in a vertical laminar flow safety cabinet for 2 to 3 h, or (iii) by keeping the tubes in a dry 36 C incubator for several days. Use of H7-sorbitol screening medium. A colony to be screened as a possible E. coli 0157:H7 was touched with a thin wire needle which was stabbed about 5 mm into the center of dry H7-sorbitol medium (in either a 24-well disposable plastic tray or a test tube). The medium was incubated at 36 C for 18 to 24 h and read for sorbitol fermentation and immobilization by the H7 serum. A reference strain of E. coli 0157:H7, which does not ferment sorbitol and contains H7 antigen, was included in each run along with a negative control strain of E. coli. Sorbitol fermentation was indicated by a change in the indicator from colorless to dark pink or red, often with the appearance of gas bubbles. Immobilization in H7 serum was indicated by the failure of the test strain to migrate from the stab line (see Fig. 1 and 2). RESULTS erinated E. coli H7 D-Sorbitol fermentation by E. coli 0157:H7. A total of 64 isolates were studied for their biochemical properties. Only into several different containers for the innmobilization test. one strain of E. coli 0157:H7 fermented sorbitol after About 2.7 ml was dispensed into each of 24 wells (capacity, overnight incubation. The strain was from a Canadian case of hemorrhagic colitis, and its history in regard to exposure to sorbitol (which selects sorbitol-positive mutants) was unknown. All the other cultures of E. coli 0157:H7 required 3 or more days of incubation to become sorbitol positive. Dm Sorbitol fermentation in E. coli 0157:H7 did occur upon further incubation. Strain A8187-M5 was streaked onto MacConkey-sorbitol agar; after about 7 days, sorbitolpositive (pink) daughter colonies appeared (Fig. 3) on sorbitol-negative (colorless) colonies. Tubes of sorbitol fermentation broth were incubated with growth from these papillae, and sorbitol was fermented with acid and gas production at 48 h. The experiment above is an example of artificial selection; however, 63 of 64 naturally occurring cultures of 1:1024 1:2048 E. coli 0157:H7 were sorbitol negative after 2 days of incubation. Other fermentation reactions of E. coli. Table 1 gives the fermentation reaction of our collection of E. coli strains for 20 carbohydrates and polyhydroxyl alcohols commonly used for identification. These data can be used in helping to decide on a carbohydrate(s) to include in single-tube screening media for detecting other specific strains of E. coli in epidemics or special studies. E. coli 0157 serum. E. coli 0157 serum seems to be very specific for 0157 strains. When 0 antigen 157 was originally added to the antigenic schema for E. ccli, a serum was prepared against its reference strain (strain A2, which has a 1:65,536 CONTROL different H antigen). This serum had a homologous titer immobilization. The (tube agglutination) of 1:2,560. The serum did not react with serum; the last tube any of the other standard 0 antigen strains except for atched areas indicate The 0157 serum had a titer of 1:1,280 against this heterologm the stab line of ous antigen. However, E. ccli 0116 is a very rare type; no isolate has been seen in our laboratory since 0116 serum has

4 VOL. 22, 1985 H7 ANTISERUM-SORBITOL FERMENTATION MEDIUM 623 FIG. 3. Delayed sorbitol fermentation of E. coli 0157:H7. Sorbitol-positive papillae (daughter colonies) are present on the colonies which were originally sorbitol negative. The six arrows point to sorbitol-positive daughter colonies of various sizes. been in use. Thus, the cross-reaction with E. coli 0116 should not be a problem. E. coli 0157 can also be detected by slide agglutination with 0157 serum. The 0157 serum had a titer of 1:8 by slide agglutination (2+ agglutination at 1 min; 4+ agglutination at 3 min) against E. coli 0157:H7. Immobilization titration. Figure 2 shows the results of the titration of E. coli H7 serum against the strain of E. coli 0157:H7. The titer was defined to be 1:4,096 (the endpoint was defined as almost complete immobilization). Based on these results, the in use dilution of H7 serum was set at 1:2,000 (1 ml of glycerinated serum per 1,000 ml of medium). Figure 1 shows the immobilization of E. coli 0157:H7 at different concentrations of H7 serum. TABLE 1. Fermentation reactions of E. coli strains which may be useful in designing media to detect a particular strain Sugar or polyhydroxyl % of E. voli alcohol strains positive" D-Adonitol... L-Arabinose D-Arabitol... 8 Cellobiose... 2 Dulcitol Erythritol... 1 Lactose Maltose D-Mannitol D-Mannose Melibiose Mucic acid ca-methyl-d-glucoside... 1 Raffinose L-Rhammose Salicin D-Sorbitol Sucrose Trehalose D-Xylose "Percentage of strains positive at 1 or 2 days at 36 C; the vast majority are positive after 24 h of incubation. Distribution of H antigens among strains of E. coli. Table 2 gives the distribution of E. coli cultures into each of the 53 H antigens currently recognized in the serotyping schema. Certain H antigens were common, but others were rare or not observed. H7 was the most common, representing 14.2% TABLE 2. H antigen Distribution of H antigens among 388 motile strains of E. coli No. (9%) of H No. antigen (91) of Hantigen ~~strains'" nie strains"~ (10.3) (2.1) (4.4) (2.1) (0) (5.4) (6.4) (0.3) (5.4) (2.6) (6.9) (0.5) (14.2) (1.8) (2.8) (5.7) (3.9) (0.3) (4.1) (1.5) 13..._... b (1.0) (0.3) (0.8) (2.3) (0.5) (0.3) (2.3) (4.9) (0.8) (1.0) (0.3) (7.7) _... h (0.3) (0) (2.1) (0.3) (1.8) (0.5) (1.0) Percent of 388 motile strains. bh antigens 13 and 22 have been deleted from the antigenic schema.

5 624 FARMER AND DAVIS of the motile cultures and 9.6% of the total number of strains serotyped, which included 185 nonmotile cultures. Specificity of the H7-sorbitol medium in detecting the epidemic strain. During the search for E. coli 0157:H7, over 300 colonies from various animal experiments were picked to H7-sorbitol medium. There were only three colonies which were found to be false-positive (sorbitol negative and immobilized, but not confirmed to be 0157:H7). These were easily shown not to be E. coli 0157:H7 by tube agglutination in 0157 serum but are counted as false-positive results in evaluating H7-sorbitol as a screening medium. In all instances, the strain producing the false-positive result was a nonfermenter which was nonmotile and had been picked as a sorbitol-negative colony from MacConkey-sorbitol agar. Sorbitol-negative, nontmotile strains of E. coli were not detected in the animal experiments. DISCUSSION Much of the early work in enteric bacteriology was directed toward isolating enteric pathogens. The objective was to detect Salmonella and Shigella strains in the presence of normal (aerobic) stool flora, which are predominately E. coli. The most popular approach has been to incorporate lactose as the differential sugar in enteric plating media. Most Salmonella and Shigella strains do not ferment lactose, but E. coli does. Lactose-negative colonies are considered suspect and tested further as possible Salmonella or Shigella strains. In the last 40 years, strains of E. coli have been shown to cause diarrhea by several different pathogenic mechanisms. The majority of these pathogenic strains ferment lactose and thus cannot be differentiated from nonpathogenic strains of E. coli which are also lactose positive. Thus, it has been difficult to distinguish specific pathogenic strains of E. coli. In our study, we wanted to detect E. coli 0157:H7 in the presence of other E. coli strains. The strain of E. coli associated with hemorrhagic colitis had two properties that made it easy to differentiate from other E. coli strains. It did not ferment D-sorbitol within 2 days, whereas about 95% of other E. coli did. It had H7 antigen, but over 90% of other E. coli strains are either nonmotile or have a different H antigen. Since sorbitol fermentation and the presence of the H7 antigen can be determined in a single tube, the screening method is very specific. This was true for all the animal experiments. In diarrheal stools from humans, there are more false-positive results (Joy Wells, personal communication) which must be ruled out by agglutination in 0157 antiserum. The falsepositive results in the screening medium are easily detected with 0157 antiserum in the confirmatory step. The singletube screening method was particularly useful for screening a large number of colonies. This same approach can also be applied in other studies when a specific strain of E. coli is being sought. One or more sugars can be incorporated into a screening medium based on the fermentation reactions of a particular strain. These can be selected by comparing the reactions of the strain with those of most E. coli given in Table 1. Once the H antigen of the strain is known, an H antiserum can also be included in a semisolid to immobilize only strains with that H antigen. The specificity of this approach can be estimated from Table 1, which gives the distribution of H antigens in a collection of E. coli strains. The incorporation of D-sorbitol instead of lactose into MacConkey agar was also helpful in detecting E. coli 0157:H7 when it was mixed with other E. coli. In the animal J. CLIN. MICROBIOL. studies (4), many MacConkey agar plates had only lactosepositive colonies. However, MacConkey-sorbitol agar contained sorbitol-negative colonies among the red sorbitolpositive colonies. The MacConkey-sorbitol plate indicated the most likely colonies to test in H7-sorbitol medium. Plates of MacConkey-sorbitol should be picked after 24 h of incubation to avoid the selection of sorbitol-positive mutants of E. coli 0157:H7. E. coli 0157:H7 has been definitely associated with hemorrhagic colitis in both outbreaks and sporadic cases, and it appears that this strain causes this diarrheal disease (1, 4, 5, 7, 9-12). Stool specimens from patients who present with symptoms typical of the illness or with frank blood in their stool can be screened for the presence of E. coli 0157:H7. Specimens from less severe diarrhea can also be screened. Sorbitol-negative strains of E. coli can be considered as suspect for 0157:H7 but must be confirmed with antisera to 0157 and H7. We have also had success with testing colonies directly from plates of MacConkey agar with 0157 serum in the slide agglutination test. Those colonies which agglutinate must then be serologically confirmed after growth on a noninhibitory medium to eliminate false-positive results which can occur with this method. Both antisera for these procedures are now commercially available; E. coli sera 0157 and H7 can be purchased from R. A. Wilson, Pennsylvania State University, 105 Henning Building, University Park, PA 16802, or from Difco Laboratories (contact A. E. Bunner). An alternative is to freeze a stool specimen taken in the acute phase of illness and submit it to a reference laboratory. Plating on MacConkey-sorbitol agar and screening in H7-sorbitol medium has been particularly useful in specifically identifying E. coli 0157:H7. Future studies will be required to determine the frequency of E. coli 0157:H7 in the United States and in other countries. We hope that MacConkey-sorbitol agar and H7-antiserumsorbitol fermentation medium will assist in the isolation and identification of this strain. LITERATURE CITED 1. Centers for Disease Control Isolation of E. coli 0157:H7 from sporadic cases of hemorrhagic colitis-united States. Morbid. Mortal. Weekly Rep. 31: Edwards, P. R., and W. H. Ewing Identification of Enterobacteriaceae, 3rd ed. Burgess Publishing Co., Minneapolis. 3. Farmer, J. J., III, M. A. Asbury, F. W. Hickman, D. J. Brenner, and the Enterobacteriaceae Study Group Enterobacter sakazakii: a new species of "Enterobacteriaceae" isolated from clinical specimens. Int. J. Syst. Bacteriol. 30: Farmer, J. J., III, M. E. Potter, L. W. Riley, T. J. Barrett, P. A. Blake, C. A. Bopp, M. L. Cohen, A. Kaufmann, G. K. Morris, R. S. Remis, B. M. Thomason, and J. G. Wells Animal models to study Escherichia coli 0157:H7 isolated from patients with hemorrhagic colitis. Lancet i: Johnson, W. M., H. Lior, and G. S. Bezanson Cytotoxin Escherichia coli 0157:H7 associated with hemorrhagic colitis in Canada. Lancet i: Kauffman, F Enterobacteriaceae. 2nd Munksgaard, Copenhagen. 7. O'Brien, A. D., T. A. Lively, and M. S. Chen Escherichia coli 0157:H7 strains associated with hemorrhagic colitis in the United States produce a Shigella dysenteriae 1 (Shiga)-like cytotoxin. Lancet i: O'Brien, A. D., J. W. Newland, S. F. Miller, R. K. Holmes, H. W. Smith, and S. M. Formal Shiga-like toxinconverting phages from Escherichia coli strains that cause hemorrhagic colitis or infantile diarrhea. Science 226: Remis, R. S., K. L. MacDonald, L. W. Riley, N. D. Puhr, J. G.

6 VOL H7 ANTISERUM-SORBITOL FERMENTATION MEDIUM 625 Wells, B. R. Davis, P. A. Blake, and M. L. Cohen Sporadic cases of hemorrhagic colitis associated with Escherichia coli 0157:H7. Ann. Intern. Med. 101: Riley, L. W., R. S. Remis, S. D. Helgerson, H. B. McGee, J. G. Wells, B. R. Davis, R. J. Hebert, E. S. Olcott, L. M. Johnson, N. T. Hargrett, P. A. Blake, and M. L. Cohen Outbreaks of hemorrhagic colitis associated with a rare E. coli serotype. N. Engl. J. Med. 308: Stewart, P. J., and W. Desormeauz Hemorrhagic colitis in a home for the aged-ontario. Can. Dis. Weekly Rep. 9: Wells, J. G., B. R. Davis, I. K. Wachsmuth, L. W. Riley, R. S. Remis, R. Sokolow, and G. K. Morris Laboratory investigation of hemorrhagic colitis outbreaks associated with a rare Escherichia coli serotype. J. Clin. Microbiol. 18: