Screening for Enteropathogenic Escherichia coli in Infants with

Transcription

1 INFECTION AND IMMUNITY, Jan. 1991, p /91/ $02.00/0 Copyright 1991, American Society for Microbiology Vol. 59, No. 1 Screening for Enteropathogenic Escherichia coli in Infants with Diarrhea by the Fluorescent-Actin Staining Test STUART KNUTTON,1* ALAN D. PHILLIPS,2 HENRY R. SMITH,3 ROGER J. GROSS,3 ROBERT SHAW,' PETER WATSON,2 AND ELIZABETH PRICE2 Institute of Child Health, University of Birmingham, Birmingham B16 8ET,1 Queen Elizabeth Hospital for Children, London E2 8PS,2 and Division of Enteric Pathogens, Central Public Health Laboratory, Colindale, London NW9 5HT,3 England Received 22 June 1990/Accepted 18 October 1990 The attaching effacing (AE) adherence property is now recognized as an important virulence characteristic of enteropathogenic Escherichia coli (EPEC). The fluorescent-actin staining (FAS) test (S. Knutton, T. Baldwin, P. H. Williams, and A. S. McNeish, Infect. Immun. 57: , 1989), which is diagnostic for the AE lesions produced by EPEC (and Vero cytotoxin-producing E. colt), has provided an additional tool with which to investigate this important class of enteric pathogens. In this study, we screened for the AE adherence property in two groups of E. coli isolated from infants with diarrhea by using the FAS test and compared the results with those from O:H serotyping, localized adhesion to HEp-2 cells (LA), and the EPEC adherence factor (EAF) probe. Only 16 of 41 (39%) E. coil strains previously diagnosed as EPEC by 0 antigen serogrouping were FAS test positive, and of these only 12 belonged to recognized EPEC O:H serotypes; 9 strains which did belong to EPEC O:H serotypes were FAS test negative. Of a second group of 297 untyped E. coli, 7 (2.3%) were FAS test positive, and of these only 2 belonged to EPEC serogroups; 5 belonged to serogroups not regarded as EPEC serogroups or were nontypeable. Of the 23 FAS-test-positive strains identified, 10 were EAF probe positive and showed good LA; 13 were EAF probe negative and showed a quantitatively distinctly poor LA. EAF-positive and EAF-negative strains, however, showed equally good adhesion to human small intestinal mucosa. None of the FAS-test-positive E. coli hybridized with probes for Vero toxins 1 or 2. We conclude that the FAS test is diagnostic not only for classical EPEC and Vero cytotoxin-producing E. coli but also for EPEC strains which are not currently being diagnosed because they belong to serotypes not generally regarded as EPEC serotypes. Enteropathogenic Escherichia coli (EPEC) strains remain an important cause of diarrheal disease in many developing parts of the world and of sporadic cases of infant diarrhea in developed countries. EPEC strains are currently defined as diarrheic E. coli belonging to serogroups epidemiologically incriminated as pathogens but whose pathogenic mechanisms have not been proven to be related to either heatstable or heat-labile enterotoxins or to epithelial cell invasion (4). Within EPEC 0 serogroups, certain O:H combinations have been found to be associated with diarrheal outbreaks. Serotyping with 0 and H antisera remains the standard diagnostic test for EPEC, but because O:H serotyping is both time-consuming and expensive and generally can only be carried out in reference laboratories, routine diagnostic laboratories in the United Kingdom generally rely solely on tests for the 0 serogroup. Many E. coli selected by this screening procedure have not been subsequently identifiable as EPEC O:H serotypes (5). There is thus a need for simpler and more accurate diagnostic tests for EPEC. In the absence of recognized enterotoxins (12), attempts to develop improved diagnostic tests for EPEC strains have centered on their adherence properties; volunteer challenge studies have shown the importance of adhesion in EPEC pathogenicity (14). The initial observation of Cravioto et al. (3), that most EPEC outbreak strains adhere to cultured HEp-2 cells whereas non-epec strains rarely adhere, led to the development of HEp-2 cell adhesion as an assay for virulence among diarrheic E. coli strains (17). Two distinct * Corresponding author. 365 patterns of adhesion have been described among EPEC isolates (21): localized adhesion (LA), which is characterized by the formation of bacterial microcolonies, and diffuse adhesion (DA), in which bacteria cover the cell uniformly. A third pattern of adherence, termed aggregative adhesion (AA), has recently been described in connection with other E. coli isolates (20). In one EPEC strain, a 60-MDa EPEC adherence factor (EAF) plasmid was found to be necessary for full expression of LA (1). Nataro et al. (19) cloned a 1-kb segment of DNA from this EAF plasmid and showed it to be a highly sensitive and specific DNA probe for detecting the EAF property. However, not all EPEC strains have been reported to adhere to HEp-2 cells or to possess EAF plasmids (13, 19). EPEC serogroups 0119 and 0125 were shown by electron microscopy of intestinal biopsies from children with persistent diarrhea to produce a characteristic brush border membrane lesion in the small intestinal mucosa consisting of intimate attachment of bacterial microcolonies, localized destruction of brush border microvilli, and pedestal formation of the apical enterocyte plasma membrane beneath attached bacteria (23, 27). An animal model of this type of adhesion was developed by Moon et al. (18), who coined the term "attaching effacing" (AE) E. coli for E. coli able to cause this characteristic brush border membrane lesion. More recent animal studies have shown the AE adherence property to be an important virulence characteristic of EPEC (25). We developed an in vitro human small intestine model of EPEC adhesion and showed the AE adherence property to be a characteristic of EPEC belonging to all of

2 366 KNUTTON ET AL. the major EPEC 0 serogroups (11). The AE adherence property has also been found to be a characteristic of Vero cytotoxin-producing E. coli (VTEC) (also called enterohemorrhagic E. coli), which colonize and cause an AE lesion in the large bowel mucosa (26). Ultrastructural studies showed that an accumulation of cytoplasmic fibrils, subsequently identified as cytoskeletal actin, beneath intimately attached bacteria was an additional feature of the AE lesion (10). On the basis of our observation that EPEC strains produce an identical AE lesion when they adhere to some tissue culture cells (9), we developed a specific fluorescent-actin staining (FAS) test, which is diagnostic for the AE adherence property and thus for EPEC and VTEC (10). The FAS test was initially validated by using a wellcharacterized collection of AE EPEC and VTEC strains (10). We have now used the FAS test to identify potential EPEC pathogens in infants with diarrheal disease and compared the results with results from serotyping, AE adhesion to human intestinal mucosa, LA to HEp-2 cells, and EAF and VT gene probes. MATERIALS AND METHODS Bacterial strains. Patients seen at, or admitted to, Queen Elizabeth Hospital for Children, London, with diarrheal symptoms routinely have stool microbiology examinations performed by standard methods. These were performed as follows. (i) For parasites, a wet preparation was stained with iodine for Giardia lamblia and other parasites, and a stool smear was stained by the modified Ziehl-Neelson assay for Cryptosporidium species. (ii) For bacteria, Salmonella and Shigella species were grown on Difco SS plates and typed by using Central Public Health Laboratory (Colindale, London) agglutinating antisera. Campylobacter isolation was performed at 42 C on agar plates containing trimethoprim, vancomycin, and colistin. (iii) For viruses, negative-staining electron microscopy was used to detect stool viruses. Two groups of E. coli were examined. The first group (typed E. coli) consisted of 41 E. coli isolates diagnosed as EPEC during 1988 by 0 antigen serogrouping. The second group (untyped E. coli) consisted of 297 untyped E. coli cultures obtained during the period of August to December In order to reduce the amount of work involved in the initial screening of this group, FAS tests were performed with mixed cultures consisting of six separate colonies from a MacConkey agar isolation plate, rather than testing six individual colony picks. Only FAS-test-positive cultures of this group were examined further, in which case single FAS-test-positive colonies were selected. For FAS testing and adhesion studies, bacteria were grown aerobically for 18 h at 37 C in Mueller-Hinton broth. Serogrouping. E. coli isolates grown on MacConkey agar were serogrouped by slide agglutination by using commercial (Central Public Health Laboratory) polyvalent and then appropriate monovalent "OK" antisera. If a positive result was found, the reaction mixture was titrated against boiled strains to endpoint dilution to exclude a nonspecific reaction. The antisera recognized serogroups 018ac, 026, 044, 055, 0111, 0114, 0119, 0125, 0126, 0127, 0128, 0142, and Serotyping. 0 and H antigens were determined by standard tube agglutination methods (8). The following are the E. coli serotypes generally considered to be EPEC serotypes (3, 22), which we have used as the basis for this study: 018ab:H- (H14), 018ac:H7, 026:H- (H1l), 044:H18 INFECT. IMMUN. (H34), 055:H- (H6 and H7), 086:H34, O111:H- (H2, H12, and H21), 0114:H2, 0119:H- (H6), 0125:H21, 0126:H- (H2 and H27), 0127:H- (H6), 0126ab:H2 (H7), 0142:H6, and 0158:H23. AE adherence property. The FAS test, which is diagnostic for the AE adherence property of E. coli, is based on visualization of the high concentrations of actin present in the AE lesion by using a fluorescent phallotoxin that binds specifically to polymerized actin. Each E. coli isolate was examined in the FAS test by using 3- and 6-h incubations of bacteria with HEp-2 and human embryonic lung cells as previously described (10). The FAS test preparations were also examined by phase-contrast microscopy, and the pattern of HEp-2 cell adhesion (localized, diffuse, or aggregative [20]) was recorded. Quantitation of LA was performed in 6-h assays by selecting microscope fields at random, counting 500 cells, and recording the number of cells with adherent microcolonies (more than five bacteria). An adhesion index was defined as the percentage of cells with adherent microcolonies. The 41 typed E. coli strains and the FAS-test-positive, untyped E. coli were tested for their ability to adhere to human intestinal mucosa and produce an AE lesion; it was not feasible to test all 297 untyped E. coli strains because availability of human intestinal mucosa was limited. Normal duodenal mucosal biopsies obtained with informed consent from adults were maintained in organ culture and infected with E. coli isolates for 8 h, as previously described (11). At the end of the incubation period, biopsies were washed, processed, and examined by scanning electron microscopy. Quantitation of mucosal colonization was performed by taking five low-power micrographs (50Ox) selected at random, and the surface area colonized by AE bacteria was determined morphometrically. An adhesion index was defined as the percentage of the mucosal surface colonized by AE bacteria. EAF and VT probe tests. The EAF probe was that described by Nataro et al. (19) and consisted of a 1-kb BamHI-SalI restriction enzyme fragment isolated from plasmid pmar22, which was cloned from plasmid pmar2. The VT1 probe was a HincII fragment of 0.75 kb from a VT1- encoding phage carried by E. coli 026:H11 (serotype H19), and the VT2 probe was a 0.85-kb AvaI-PstI fragment from a VT2-encoding phage carried by E. coli E32511 (serotype 0157:H-) (28). The probe fragments, both of which are derived from structural genes, were labeled with deoxyadenosine-5-[35s]thiotriphosphate by the random primer method. Strains were tested by colony hybridization as described by Maniatis et al. (16). RESULTS Only 34 of the 41 typed E. coli strains were confirmed as belonging to EPEC 0 serogroups when full O:H serotyping was performed. The serotypes of the 34 strains and the results of the various tests performed with the strains are shown in Table 1. Strains 1 to 20 belong to recognized EPEC O:H serotypes; strains 21 to 25 belong to EPEC 0 serogroups (but not classical O:H serotypes) and are FAS test positive (FAS+); strains 26 to 34 belong to EPEC 0 serogroups and are FAS-. Of the 34 strains, 16 (47%) were FAS+ and the same 16 strains adhered to human intestinal mucosa and produced an AE lesion demonstrable by scanning electron microscopy (Fig. 1). Of the 16 FAS+ E. coli, 11 belong to recognized EPEC O:H serotypes (026:H-, 055:H-, 055:H7, 0114:

3 VOL. 59, 1991 SCREENING FOR ENTEROPATHOGENIC E. COLI 367 TABLE 1. Characteristics of typed E. coli strains' Strain O:H serotype Classical EPEC FAS AE intestinal HEp-2 cell EAF serotype test lesion adhesion probe 1. KH1/8 0114:H (LA/G) + 2. AH16/3 0125ac:H (LA/G) + 3. AJ8/6 0127ab:H (LA/G) + 4. CH26/1 0127ab:H (LA/G) + 5. CH7/6 026:H (LA/P) - 6. Y129/3 026:H (LA/P) - 7. AR2/5 055:H (LA/P) - 8. WT25/1 055:H (LA/P) - 9. BO1/9 055:H (LA/P) OS26/9 055:H (LA/P) BA31/3 0119:H (LA/P) LC9/6 Olllab:H (AA) 13. H015/6 0126:H (AA) WH8/ :H (AA) D026/4 044:H KS22/6 044:H WM22/1 055:H PK19/4 0126:H WJ19/ :H CS7/ :H AK11/5 0114:H (LA/G) WA24/9 0128ab:H (LA/G) BF23/5 Olllab:H (LA/P) 24. AN26/7 0127ab:H (LA/P) 25. UD19/8 0128ab:H (LA/P) 26. AN11/3 055:H OR20/6 055:H BT19/10 055:H WM22/9 0114:H WT28/9 0114:H OK29/ :H W024/9 0126:H BU11/4 0126:H WL28/ ab:H a All 34 strains were VT probe negative. H2, 0119:H-, 0125:H21, 0127:H-, and 0127:H6), 3 belong to EPEC serogroups but are nonmotile (0114:H- and 0128:H-), and 3 belong to O:H serotypes not generally regarded as classical EPEC by virtue of their H antigens (O111:H25, 0127:H40, and 0128:H6). Nine FAS- E. coli strains belong to recognized EPEC serotypes (044:H18, 055:H7, O111:H21, 0126:H-, and 0126:H27). LA to HEp-2 cells and EAF detection by the EAF DNA probe are properties considered diagnostic for EPEC (13). We therefore compared these properties with the FAS test results. The FAS+ strains all exhibited LA, but we detected two quantitatively distinct levels of LA, which we have termed good LA (LA/G) and poor LA (LA/P). LA/G is the LA pattern described by others for many classical EPEC strains in 3-h adhesion assays in which one sees large microcolonies of adherent bacteria, with generally >50% cells having microcolonies (Fig. 2A). For illustrative purposes, we used fluorescence micrographs to show LA patterns of adhesion because of their greater clarity. This is justified because each intense spot of actin fluorescence seen in the FAS test is produced by a single AE bacterium (10). In 3-h assays, LA/P was generally very weak and featured very small microcolonies of AE bacteria on <1% of the cells (Fig. 2B). Indeed, in many cases, 3-h assays did not fulfill our criteria of LA (i.e., five bacteria per microcolony) and only individual AE bacteria or small aggregates of less than five bacteria were seen (Fig. 2C). We therefore extended the assay time to 6 h, after which time the LA/P pattern was still poor but somewhat more pronounced. In 6-h assays, localized microcolonies were apparent although <10% of cells generally had adherent microcolonies of AE bacteria and the microcolonies were generally much smaller than was the case for LA/G (Fig. 2D). The LA/G and LA/P patterns correlated with presence and absence of EAF, since the 6 FAS+ strains which showed the LA/G pattern were all EAF+ and the 10 strains which showed the LA/P pattern were all EAF-. The 18 FAS- strains were all LA-, although 3 strains, all belonging to EPEC serotypes (Olllac:H21, 0126:H-, and 0126:H27), were HEp-2 cell adherent and showed an AA pattern of cell adhesion. None of the 34 strains hybridized with the VT probes. The typed E. coli group had been preselected on the basis of 0 serogrouping. Since we and others (7) have suggested that there may be EPEC strains which fall outside the currently accepted list of EPEC serotypes, we examined a second group of strains which consisted of 297 untyped E. coli strains from patients with diarrheal symptoms. Of the 297 strains, 7 were FAS+ (Table 2). Subsequent routine 0 serogrouping of the 297 strains showed that 10 belonged to EPEC 0 serogroups but that only 2 of the 10 (Table 2, no. 35 and 36) were FAS+. Since FAS tests were performed initially with pools of six colonies, single FAS+ colonies were selected prior to O:H serotyping and further testing. O:H serotyping confirmed that just two of the seven FAS+

4 368 KNUTTON ET AL. INFECT. IMMUN. FIG. 1. Scanning electron micrograph of cultured human small intestinal mucosa showing the characteristic AE brush border membrane lesion produced by all the FAS+ strains. Magnification, x2,000. strains belonged to EPEC serogroups and that only one strain belonged to an EPEC serotype; the remaining five belonged to 0 serogroups not regarded as EPEC (02 and 021) or were 0 antigen untypeable (Table 2). All seven FAS+ strains showed AE adhesion to human intestinal mucosa and LA, but, as was the case with the typed E. coli group of strains, these LA strains could also be clearly divided into LA/G and LA/P strains, and again, LA/G strains were EAF+ and LA/P strains were EAF-. Since on initial testing it appeared that LA/G and LA/P strains showed equally good adhesion to human intestinal mucosa, we attempted to quantitate adhesion of the seven FAS+ strains both to HEp-2 cells and to human intestinal mucosa. HEp-2 cell adhesion indices of the four LA/G EAF+ strains ranged from 59 to 96, whereas the adhesion indices for the three LA/P EAF- strains ranged from 4 to 14. Human intestinal mucosa adhesion indices of the seven strains ranged from 5 to 41, although in this case there was no correlation between the adhesion index and EAF positivity; EAF- strains adhered just as efficiently as EAF+ strains (Table 2). None of the seven strains hybridized with the VT probes. Some patient details relating to the seven FAS+ untyped E. coli strains are also included in Table 2. Males and females were infected, with an age range of 3 to 16 months. Of the seven cases, four had persistent diarrhea (i.e., lasting for more than 14 days), and in only one case (no. 37) was another putative pathogen (Aeromonas caviae) detected. All cases with persistent diarrhea were EAF+; the remainder were EAF-. Although the number of FAS+ E. coli found in this part of the study is small (n = 7), it was similar to the number of cases of Salmonella (n = 6), Campylobacter (n = 8), and G. lamblia (n = 4) infections found over the same time period. Also identified among the 297 strains were 4 strains showing the AA pattern of HEp-2 cell adhesion and 6 showing a DA pattern. DISCUSSION Twenty-three different E. coli strains possessing the AE adherence property were identified in this study by using the FAS test. None hybridized with the VT probes, and all adhered to small intestinal mucosa, which suggests that all 23 strains are potential EPEC pathogens. On the basis of serotype and adherence properties, this study has identified four types of putative AE EPEC pathogens (Table 3). However, volunteer challenge studies (14) and/or casecontrol epidemiological studies (2, 7, 15) have only confirmed an association with diarrhea in the case of EAF+ EPEC strains which belong to classical serotypes. Of the 23 FAS+ E. coli strains identified, only 5 were EAF+, belonged to classical serotypes, and can therefore be confirmed as EPEC pathogens. Five strains were EAF+ and belonged to nonclassical serotypes, and 13 strains were EAF-. Although EAF- EPEC have been shown to cause mild diarrhea in adult volunteers (14), case-control epidemiological studies have not confirmed an association with diarrhea of EAF- EPEC (7, 15). It is probably premature, therefore, to conclude that all the FAS+ strains identified in this study are EPEC pathogens, although we believe that they probably are. FAS+, EAF- E. coli strains could be (i) EAF+ EPEC strains which have lost EAF adhesin or regulatory genes, (ii) EPEC strains which possess adherence factors distinct from EAF, or (iii) nonpathogenic E. coli strains which happen to possess the AE property but lack other essential EPEC

5 VOL. 59, 1991 SCREENING FOR ENTEROPATHOGENIC E. COLI 369 FIG. 2. Fluorescence micrographs of FAS test preparations showing typical 3-h LA/G adherence patterns. Magnification, x300. virulence determinants. EAF promotes adhesion of EPEC to human intestinal mucosa and, fortuitously, to some tissue culture cells lines. EAF+ EPEC strains which have been cured of the EAF plasmid are still FAS+ (10) but colonize the gut only poorly and adhere poorly both to HEp-2 cells (9) and to human intestinal mucosa (11) compared with the parent strain. This is not the case with the FAS+, EAFstrains identified in this study. Although the EAF- strains showed poor LA to HEp-2 cells, they adhered to human intestinal mucosa as efficiently as the EAF+ strains and TABLE 2. 5 (A), 3-h LA/P (B and C), and 6-h LA/P (D) AE much more efficiently than we previously found in the case of an EPEC strain cured of the EAF plasmid (11). For similar reasons, we find it difficult to believe that the EAF-, FAS+ strains would be nonpathogenic, since they were isolated from patients with diarrhea, they possess the AE adherence property, and they are able to colonize human intestinal mucosa, albeit in vitro, as efficiently as EAF+ EPEC. This leaves us with the most likely conclusion that FAS+, EAF- E. coli are probably EPEC pathogens and that they probably possess adherence factors distinct from, but perhaps similar Characteristics of FAS-test-positive untyped E. coli strains E. coli characteristics' Patient details Strain Classical Intestinal HEp-2 cell O:H serotype C adhesion adhesion EAF otpe serotype inexprobe index Pattern Index Sex Age (mo) (duration Symptom(s) [daysl)' 35. IR3/8 0127ab:H LA/G 96 + M 3 D (>14) 36. AR9/8 0128ab:H LA/P 5 - M 13 D (10) 37. MD25/7 02:H45-41 LA/G 72 + M 3 D (>14) 38. FT8/8 O?:H1-16 LA/G 59 + F 16 D,V (14) 39. BC9/8 O?:H1-5 LA/G 87 + F 4 D,V (>14) 40. WK15/11 021:H3-17 LA/P 4 - F 12 D (7) 41. YA21/9 O?:H8-18 LA/P 14 - M 16 D,V (4) a All 7 E. coli strains were VT probe negative. b D, Diarrhea; V, vomiting.

6 370 KNUTTON ET AL. TABLE 3. Types of putative AE EPEC pathogens FAS EAF HEp-2 Association Type Serotype test probe adhesion with diarrheaa 1 Classical + + LA/G + 2 Classical + - LA/P - 3 Nonclassical + + LA/G - 4 Nonclassical + - LA/P NA a Case-control epidemiological studies. NA, Not assessed. to, EAF. This may be reflected in the fact that both EAF+ and EAF- AE E. coli strains show LA patterns of adhesion. For the following reasons, we do not find it surprising that case-control epidemiological studies have not yet shown an association between diarrhea and EAF- EPEC: (i) because the AE property of EAF- isolates has not previously been assessed, and (ii) in South America, where the major studies have been performed (7, 15), three serogroups (055, 0111, and 0119) were found to account for a high percentage (73 to 90%) of EPEC isolates, and these were all EAF+ serogroups. This may not be the case in other parts of the world. In this study, for example, the FAS+ 055 serogroup E. coli isolates from diarrhea cases were all EAF-. Now that FAS+, EAF- E. coli have been isolated from patients with diarrhea and tests for the AE adherence property are available, case-control epidemiological studies in which both the serotype and the AE and EAF properties are examined need to be carried out in order to assess the possible role of FAS+, EAF- EPEC in infant diarrhea. Serotyping remains the standard diagnostic test for EPEC. O serogrouping is the routine diagnostic test for EPEC in the United Kingdom. In this study, the 34 typed E. coli were diagnosed as EPEC by serogrouping, but only 20 of the 34 (59%) belonged to EPEC serotypes. Thus, as has been found previously (5), 0 serogrouping alone appears to identify a significant number of false-positive EPEC. Only 11 of the 20 strains belonging to EPEC serotypes possessed the AE property. If the nine strains belonging to EPEC serotypes which lack the AE property are EPEC pathogens, they must either have lost the AE property or cause diarrhea by a mechanism different from that of AE EPEC. Certainly EPEC, as currently defined, can include E. coli pathogens that cause diarrhea by more than one mechanism. There is evidence that the FAS- 055:H7 strain (Table 1, no. 17) has probably lost the AE property because this strain was isolated from the brother of a patient from whom a FAS+ strain of the same serotype was isolated (no. 8). It may be significant, however, that three of the remaining nine FASstrains which belonged to EPEC serotypes showed the AA pattern of HEp-2 cell adhesion, a pattern now associated with the enteroadherent, aggregative class of diarrheic E. coli (13). Only two of the seven FAS+ untyped E. coli group would have been diagnosed as EPEC by serogrouping, and only one would have been diagnosed by serotyping; the remaining five would not have been diagnosed. Although the numbers are low, this study suggests that in the United Kingdom EPEC strains belonging to serotypes not generally regarded as EPEC serotypes could be just as common as classical EPEC. Diagnosis of EPEC will always be problematic until there is a definition of EPEC that is based not on serology but on specific virulence properties. EAF detected with the EAF DNA probe has been used as a diagnostic test for EPEC (13, 15), but this study has shown that EAF may not be a INFECT. IMMUN. virulence factor for all EPEC strains. Most classical EPEC strains do, on the other hand, possess the important AE property, and so this is a candidate virulence property which could form the basis for a mechanistic definition of EPEC. The FAS test, however, would probably be unsuitable for routine EPEC diagnosis because of the hazardous nature of the fluorescent phallotoxin reagent involved. A DNA probe with 100% sensitivity and 98% specificity in detecting E. coli with the AE property has been recently described (8a). Such a probe would be more practical for routine EPEC diagnosis. In this study, as in a previous one (10), FAS test positivity was found to correlate with LA. Six DA and 8 AA strains identified in this study were all FAS-. Although the role of strains showing DA and AA is still uncertain (6), this study has confirmed that such strains are not AE EPEC. A significant observation from this study was the identification of two quantitatively distinct LA patterns of HEp-2 cell adhesion produced by the EAF+ and EAF- strains. The two patterns are fundamentally the same and quite distinct from DA and AA patterns; hence, we feel justified in retaining the term LA to describe the new LA/P pattern. In order to detect the LA/P pattern, we now recommend careful examination of 6-h adhesion assays. Of course, to be certain of detecting the pattern, we recommend the use of a 6-h FAS test. Although the EAF- strains identified in this study could be distinguished from EAF+ strains by their low percentage of adhesion in HEp-2 assays, there may be other EAF- E. coli strains which cannot be distinguished from EAF+ E. coli on this basis (24). In the 3-h assays used by many workers, and without the benefit of the FAS test to assess the AE property of bacteria, we believe most FAS+, EAF- E. coli would probably be incorrectly assessed as nonadherent or as giving a weak DA pattern. This may explain why the LA/P pattern has not been previously described. It is now clear that different types of AE E. coli exist and that the FAS test is useful in screening for these different types. Of particular interest now will be types 2, 3, and 4 (Table 3) and their possible role in infant diarrhea. ACKNOWLEDGMENTS This work was supported by the Wellcome Trust (S.K. and R.S.). We thank Glynne Jones for providing HEp-2 cells, Roy Cockel for providing human intestinal tissue, and Urmilla Patel, Pauline Stow, and Lesley Gould for help in collecting E. coli strains. REFERENCES 1. Baldini, M. M., J. B. Kaper, M. M. Levine, D. C. A. Candy, and H. W. Moon Plasmid-mediated adhesion in enteropathogenic Escherichia coli. J. Pediatr. Gastroenterol. Nutr. 2: Chatkaeomorakot, A., P. Echeverrhia, D. N. Taylor, K. A. Bettleheim, N. R. Blacklow, 0. Sethabutr, J. Seriwatana, and J. B. Kaper HeLa cell adherent Escherichia coli in children with diarrhea in Thailand. J. Infect. Dis. 156: Cravioto, A., R. J. Gross, S. M. Scotland, and B. Rowe An adhesive factor found in strains of Escherichia coli belonging to traditional infantile enteropathogenic serotypes. Curr. Microbiol. 3: Edelman, R., and M. M. Levine Summary of a workshop on enteropathogenic Escherichia coli. J. Infect. Dis. 147: Farmer, J. J., III, B. R. Davis, W. B. Cherry, D. J. Brenner, V. R. Dowell, Jr., and A. Balows "Enteropathogenic serotypes" of Escherichia coli which really are not. J. Pediatr. 90: Gomes, T. I. T., P. A. Blake, and L. R. Trabulsi Prevalence of Escherichia coli strains with localized, diffuse, and aggregative adherence to HeLa cells in infants with diarrhea

7 VOL. 59, 1991 and matched controls. J. Clin. Microbiol. 27: Gomes, T. I. T., M. A. M. Vieira, I. K. Wachsmuth, P. A. Blake, and L. R. Trabulsi Serotype specific prevalence of Escherichia coli strains with EPEC adherence factor genes in infants with and without diarrhea in Sao Paulo, Brazil. J. Infect. Dis. 160: Gross, R. J., and B. Rowe Serotyping of Escherichia coli, p In M. Sussman (ed.), The virulence of Escherichia coli: reviews and methods. Academic Press, Inc. (London), Ltd., London. 8a.Jerse, A. E., J. Yu, B. D. Tall, K. G. Gicquelais, and J. B. Kaper Abstr. Annu. Meet. Am. Soc. Microbiol. 1990, B-158, p Knutton, S., M. M. Baldini, J. B. Kaper, and A. S. McNeish Role of plasmid-encoded adherence factors in adhesion of enteropathogenic Escherichia coli to HEp-2 cells. Infect. Immun. 55: Knutton, S., T. Baldwin, P. H. Williams, and A. S. McNeish Actin accumulation at sites of bacterial adhesion to tissue culture cells: basis of a new diagnostic test for enteropathogenic and enterohemorrhagic Escherichia coli. Infect. Immun. 57: Knutton, S., D. R. Lloyd, and A. S. McNeish Adhesion of enteropathogenic Escherichia coli to human intestinal enterocytes and cultured human intestinal mucosa. Infect. Immun. 55: Law, D Virulence factors of enteropathogenic Escherichia coli. J. Med. Microbiol. 26: Levine, M. M Escherichia coli that cause diarrhea: enterotoxigenic, enteropathogenic, enteroinvasive, enterohemorrhagic and enteroadherent. J. Infect. Dis. 155: Levine, M. M., J. P. Nataro, H. Karch, M. M. Baldini, J. B. Kaper, R. E. Black, M. L. Clements, and A. D. O'Brien The diarrheal response to some classic serotypes of enteropathogenic Escherichia coli is dependant on a plasmid encoding an enteroadherence factor. J. Infect. Dis. 152: Levine, M. M., V. Prado, R. Robins-Browne, J. B. Kaper, S. L. Moseley, K. Giequelais, J. P. Nataro, P. Vial, and B. Tall Use of DNA probes and HEp-2 cell adherence to detect diarrheagenic Escherichia coli. J. Infect. Dis. 158: Maniatis, T., E. F. Fritsch, and J. Sambrook Molecular cloning: a laboratory manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 17. Mathewson, J. J., and A. Cravioto HEp-2 cell adherence SCREENING FOR ENTEROPATHOGENIC E. COLI 371 as an assay for virulence among diarrheagenic Escherichia coli. J. Infect. Dis. 159: Moon, H. W., S. C. Whipp, R. A. Argenzio, M. M. Levine, and R. A. Gianella Attaching and effacing activities of rabbit and human enteropathogenic Escherichia coli in pig and rabbit intestines. Infect. Immun. 41: Nataro, J. P., M. M. Baldini, J. B. Kaper, R. E. Black, N. Bravo, and M. M. Levine Detection of an adherence factor of enteropathogenic Escherichia coli with a DNA probe. J. Infect. Dis. 152: Nataro, J. P., J. B. Kaper, R. Robins-Browne, V. Prado, and M. M. Levine Patterns of diarrheagenic Escherichia coli to HEp-2 cells. Paediatr. Infect. Dis. 6: Nataro, J. P., I. C. A. Scaletsky, J. B. Kaper, M. M. Levine, and L. R. Trabulsi Plasmid-mediated factors conferring diffuse and localized adherence of enteropathogenic Escherichia coli. Infect. Immun. 48: Robins-Browne, R. M Traditional enteropathogenic Escherichia coli of infantile diarrhea. Rev. Infect. Dis. 9: Rothbaum, R., A. J. McAdams, R. A. Gianella, and J. C. Partin A clinicopathologic study of enterocyte-adherent Escherichia coli: a cause of protracted diarrhea in infants. Gastroenterology 83: Scotland, S. M., G. A. Wilishaw, H. R. Smith, R. J. Gross, and B. Rowe Adhesion to cells in culture and plasmid profiles of enteropathogenic Escherichia coli isolated from outbreaks and sporadic cases of infant diarrhea. J. Infect. 19: Tzipori, S., R. Gibson, and J. Montanaro Nature and distribution of mucosal lesions associated with enteropathogenic and enterohemorrhagic Escherichia coli in piglets and the role of plasmid-mediated factors. Infect. Immun. 57: Tzipori, S., I. K. Wachsmuth, C. Chapman, R. Birner, J. Brittingham, C. Jackson, and J. Hogg The pathogenesis of haemorrhagic colitis caused by Escherichia coli 0157:H7 in gnotobiotic piglets. J. Infect. Dis. 154: Ulshen, M. H., and J. L. Rollo Pathogenesis of Escherichia coli gastroenteritis in man-another mechanism. N. Engl. J. Med. 302: Willshaw, G. A., H. R. Smith, S. M. Scotland, A. M. Field, and B. Rowe Heterogeneity of Escherichia coli phages encoding Vero cytotoxins: comparison of cloned sequences determining VT1 and VT2 and development of specific gene probes. J. Gen. Microbiol. 133: