APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Oct. 1997, p. 4127–4131 0099-2240/97/$04.0010 Copyright © 1997, American Society for Microbiology
Vol. 63, No. 10
A Simple Filtration Technique To Detect Enterohemorrhagic Escherichia coli O157:H7 and Its Toxins in Beef by Multiplex PCR KASTHURI VENKATESWARAN,1* YURI KAMIJOH,1 EIJI OHASHI,1 2 AND HISAO NAKANISHI Central Research Laboratory, Nippon Suisan Kaisha, Ltd., Hachioji City, Tokyo 192,1 and Public Health Research Institute of Kobe, Minatojima-Nakamachi, Kobe 650,2 Japan Received 9 June 1997/Accepted 16 July 1997
Primers, specific for a unique base substitution in uidA of Escherichia coli O157:H7, were coupled with oligonucleotides for the shiga-like toxin I (SLT-I) and SLT-II genes in a multiplex PCR assay. A minimum of 102 CFU per PCR (10 ml) was necessary to amplify E. coli O157:H7-specific bands by multiplex PCR. Food particles as well as various unknown metabolic by-products of bacteria inhibited the PCR, but a simple two-step filtration procedure eliminated this inhibition. To reliably generate PCR products, an E. coli inoculum of 103 CFU g of food slurry21 in a nonspecific medium was required with 6 h of enrichment at 37°C. However, when the food homogenate was incubated overnight, E. coli O157:H7 at an initial inoculum of even 1 CFU g21 was detected. allele in O157:H7 strains. The third and fourth sets of primers were directed to the conserved regions within the genes encoding SLT-I and SLT-II, respectively (14). Multiplex PCR studies have largely concentrated on the identification of bacterial strains or toxins with DNA extracted from pure cultures (2, 4, 14). PCR amplification of enrichment broth was successful in foods only when an immunomagnetic separation step for E. coli O157:H7 was performed (15), but PCR methods for detection of bacteria directly from food samples need at least partially purified DNA (12, 15, 27). The potential of the multiplex PCR assay as a rapid diagnostic tool for screening O157:H7 isolates directly from beef is described in this work. The assay requires a preenrichment step followed by a two-step filtration procedure and multiplex PCR-based amplification of four sets of target DNAs. Since no DNA extraction step is required, this method has greater convenience and speed than many established procedures (12, 15). The time needed for detection, including sample preparation, PCR, and agarose gel electrophoresis, is less than 24 h. This method is therefore more rapid and sensitive than verotoxin assay (20), enzyme-linked immunosorbent assay (ELISA) (23), DNA hybridization (8), and direct-sample PCR (15) techniques. Bacterial isolates. Table 1 includes a list of all strains tested in this study. These strains were purchased from the American Type Culture Collection (Rockville, Maryland) or received as gifts from the National Children’s Medical Research Center (Tokyo, Japan) and the National Institute of Health—Japan (Tokyo). Biochemical, serological, and immunological characterization of toxigenic strains of E. coli. The U.S. Food and Drug Administration procedure was used to verify the identity of all E. coli isolates (24). Isolates were streaked onto Levine’s eosine methylene blue (EMB) agar (Nissui, Tokyo, Japan), Fluorocult violet-red bile agar (VRB-MUG; Merck, Darmstadt, Germany), and Fluorocult E. coli O157:H7 agar (Merck). Colonies that were sorbitol negative (green coloration) in Fluorocult E. coli O157:H7 agar were presumptively identified as E. coli O157:H7. All strains were examined to determine their characteristics on EMB agar, for gas production at 44.5°C in EC medium (Nissui), and for b-glucuronidase activity (24).
Most Escherichia coli strains are harmless commensals in the human gut. However, some strains, such as E. coli O157:H7, can cause severe food-borne disease and are referred to as enterohemorrhagic E. coli (EHEC) (12, 18, 19). The EHEC strains of serotype O157:H7 cause hemorrhagic colitis, which may develop into life-threatening hemolytic-uremic syndrome. EHEC strains produce toxins known as verotoxins or Shigalike toxins (Shiga-like toxin I [SLT-I] and SLT-II) because of their similarity to the toxins produced by Shigella dysenteriae (13, 16). A large food-poisoning outbreak in 1993, traced to the consumption of undercooked hamburgers contaminated with O157:H7, infected over 700 persons in four states of the United States, resulting in 51 cases of hemolytic-uremic syndrome and four fatalities (5). A waterborne O157:H7 outbreak involving 200 children in Saitama Prefecture, Japan, resulting in 2 fatalities was reported. Recently, more than 8,000 children (including seven fatalities), in 43 of the 47 total Japanese prefectures, were shown to excrete O157:H7 after consuming the midday school meal. One of the major challenges in diagnosing such problems is identifying the strain of E. coli responsible. Bioassays and conventional methods are commonly used to differentiate toxigenic from nontoxic E. coli strains (20, 22), but there are limitations with these methods. Recently, the multiplex PCR has been used for the detection of various toxin-producing E. coli (4, 12, 17) strains. In the multiplex PCR method, two or more primer sets are used to simultaneously amplify multiple target sequences. Cebula et al. (4) developed an assay that simultaneously identifies isolates of O157:H7 and the types of SLT they encode. The first set of primers is specific for the uidA gene, which encodes b-glucuronidase in E. coli (2). Although O157:H7 isolates fail to exhibit b-glucuronidase activity, they do carry the uidA gene (9, 10). Exploiting the uniqueness of a 92-base change in the uidA gene, Cebula et al. (4) designed a second set of primers in a mismatch amplification mutation assay format (6) to preferentially amplify the uidA * Corresponding author. Mailing address: Center for Great Lakes Studies, University of Wisconsin—Milwaukee, 600 E. Greenfield Ave., Milwaukee, WI 53204. Phone: (414) 382-1712. Fax: (414) 382-1705. E-mail:
[email protected]. 4127
4128
VENKATESWARAN ET AL.
APPL. ENVIRON. MICROBIOL.
TABLE 1. Characteristics of E. coli O157:H7 and other toxigenic strains Fluorescence in: a
Microbe
Strain
E. coli S. dysenteriae E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli E. coli S. sonnei
ATCC 25922T ATCC 13313T NCMRC-N NCMRC-TA PHRIK 94-4 PHRIK 94-39 PHRIK 94-52 PHRIK 95-7 PHRIK 95-113 ATCC 43886 ATCC 35401 ATCC 35150 ATCC 43889 ATCC 43890 ATCC 43894 ATCC 43888 ATCC 43895 NIHJ 1 NIHJ F1 NIHJ 3-2 NIHJ 910001 NIHJ 91003 NIHJ 23-95 NIHJ 7 NIHJ 91005 NIHJ 545-94 NIHJ 840020 NIHJ 840021 NIHJ 790081 NIHJ 790083 NIHJ 251 NIHJ 285 NIHJ 269 NIHJ 276 NIHJ 14 NIHJ 188 NIHJ 348 NIHJ 381 ATCC 2930 T
Serotype
Toxin productionb
Diarrheagenic E. coli classification or sourcec
SLT-I O157:H7 O157:H7 O157:H7 O157:H7 O157:H7 O157:H7 O157:H7 O25:K98 O78:H11 O157:H7 O157:H7 O157:H7 O157:H7 O157:H7 O157:H7
O44:H18
O167:H9
O26:H11 O26:H11 O111:H1 O111:HO157:H7 O157:H7 O157:H7 O157:H7
SLT-I, SLT-II SLT-I, SLT-II SLT-I, SLT-II SLT-II SLT-I, SLT-II SLT-I, SLT-II SLT-I, SLT-II LT LT, STs SLT-I, SLT-II SLT-II SLT-I, SLT-II SLT-I, SLT-II SLT-II SLT-I, SLT-II LT ST1 ST2 VT1 VT2h SLT-II invE VT2a eae
SLT-I SLT-I SLT-I SLT-II SLT-II SLT-I, SLT-II SLT-I
EHEC EHEC EHEC EHEC EHEC EHEC EHEC ETEC ETEC EHEC EHEC EHEC EHEC ETEC EHEC ETEC ETEC ETEC VTEC VTEC EPEC EIEC VTEC EPEC Urinary tract infection Urinary tract infection Veterinary infection Veterinary infection EPEC EPEC EPEC EPEC EHEC EHEC EHEC EHEC
Sorbitol utilization
Serological characterization for O157:H7
1 2
1 2
2e 2
2 2 2 2 2 2 2 1 1 2 2 2 2 2 2 2 1 1 2 2 1 2 1 1 2 1 2 1 1 1 2 2 2 2 2 2 2
2 2 2 2 2 2 2 1 1 2 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2
1 1 1 1 1 1 1 2 2 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 1 1 1 1 2
Glittering in EMB agar
Gas in EC broth (44.5°C)
VRB-MUG agar
E. coli O157:H7 agar
1d 2
1 2
1 2
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 1 1 1 1 1 1 2 2 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 1 2 1 1 2 2 2 1 1 2 2 2 1 1 1 1 1
2 2 2 2 2 2 2 2 1 2 2 2 2 2 2 2 1 1 1 1 1 1 1 1 2 2 1 1 1 1 1 1 2 2 2 2 1
a ATCC, American Type Culture Collection; NCMRC, National Children’s Medical Research Center; PHRIK, Public Health Research Institute, Kobe, Japan; NIHJ, National Institute of Health—Japan. b Toxigenic properties were investigated by ELISA titers, animal model, cell culture, and PCR methods. ST, heat-stable enterotoxin; VT, verotoxin; eae, gene for attachment to and effacement of the microvilli of enterocytes; invE, gene that encodes outer membrane protein involved in invasiveness. c Diarrheagenic E. coli classifications are as described by Levine (18). For some samples, the source of the bacterial strain is given. Abbreviations: ETEC, enterotoxigenic E. coli; VTEC, verotoxigenic E. coli; EPEC, enteropathogenic E. coli; EIEC, enteroinvasive E. coli. d 1, positive reaction. e 2, negative reaction.
The b-glucuronidase activity was monitored by using Bactident E. coli (Merck) as recommended by the manufacturer. Biochemical characteristics of the bacterial isolates were determined by using the EB-20 system (Nissui) and the Biolog (Hayward, Calif.) identification system as described earlier (24, 25). Serotyping of the O somatic antigen (O157) and H flagellar antigen (H7) was carried out with commercially available kits (Denka Seiken antisera; Tokyo, Japan). The ELISA for the verotoxin production of all the strains was performed with the reversed passive latex agglutination kit (Denka Seiken) as recommended by the manufacturer. Morphological and biochemical characteristics of various toxigenic strains of E. coli and Shigella strains used in this study are tabulated (Table 1). All bacterial strains tested except four isolates of E. coli and S. dysenteriae ATCC 13313 showed the typical metallic-sheen colonies on EMB agar plates. Gas was
produced from lactose at 44.5°C by all O157:H7 strains but not by nine of the other E. coli strains. Sorbitol was not utilized by E. coli O157:H7, whereas other toxigenic strains utilized this carbohydrate. Neither S. dysenteriae nor Shigella sonnei utilized sorbitol, but both exhibited green coloration on Fluorocult E. coli O157:H7 agar. Cross-reaction of the somatic antigen O157 and the flagellar antigen H7 was not seen between O157 and O25, O26, O78, or O111, nor was it seen between H7, H11, and H2. An immunological test for the production of verotoxins by the various strains showed that SLT-II was produced by many strains and that SLT-I was not produced at a titer higher than that of SLT-II (data not shown). Groupwise variation in the production of verotoxins revealed that O157:H7 strains produced SLT-II predominantly over SLT-I. However, this group could not be differentiated based solely on its serological and immunological characteristics.
VOL. 63, 1997
DETECTION OF E. COLI O157:H7 IN BEEF BY MULTIPLEX PCR
The absence of sorbitol fermentation by O157:H7 is a phenotypic characteristic used to isolate this organism from clinical and food specimens. Though it is a useful test, confirmation with O157 and H7 antisera is still required since other bacteria share this serotype and a number of O157:H7 strains ferment sorbitol (11). Antibodies to the O157 antigen are also used in many assays to detect O157:H7 in clinical and food samples. These tests are not specific and provide no information about the toxin types produced by the isolates, since the O157 antigen is present on other E. coli species (1, 21). Additionally, anti-O157 serum often cross-reacts with Citrobacter freundii, Escherichia hermanii, and other bacteria (3). Analyses of food products with anti-O157 serum have recognized O157 isolates that neither produced SLT nor were of the H7 serotype (26). We have confirmed in the present study a previous report (7) that the O157:H7 serogroup does not exhibit b-glucuronidase activity. In addition to O157:H7 strains, one heat-labile toxin (LT)-producing E. coli (ATCC 43886) had no fluorescence on VRB-MUG agar containing lactose. However, this strain showed fluorescence on Fluorocult E. coli O157:H7 agar, in which sorbitol is used as carbon source. Such lactose suppression of b-glucuronidase activity in E. coli has been previously reported (7, 24). Therefore, differentiation of E. coli O157:H7 from other toxigenic and nontoxigenic strains on the basis of biochemical characteristics is dubious. Specificity of multiplex PCR primers. Bacterial strains were grown on Trypticase soy agar (Nissui) plates overnight at 37°C. The bacterial cells in a well-isolated colony were resuspended in sterile phosphate-buffered saline (PBS) (0.1 M; pH 7.0) to a concentration of 105 CFU ml21 and used as the DNA template for PCR. Oligonucleotide primers were based on previously published nucleotide sequence data for uidA, O157:H7-specific, SLT-I, and SLT-II genes (4) and synthesized with an oligonucleotide synthesizer (Beckman Instruments, Inc., Fullerton, Calif.) as recommended by the manufacturer. All bacterial DNA amplification was carried out in a final volume of 100 ml. Ten microliters from a 105 CFU ml21 bacterial suspension prepared in PBS, or food homogenate as described below, was added to the PCR mixture, and amplification conditions were employed as detailed elsewhere (24). Following DNA amplification, 15-ml aliquots from each PCR mix were analyzed by submarine gel electrophoresis on 2% agarose (Nusieve GTG; FMC Bioproducts, Rockland, Maine) gels. Amplicons corresponding to the uidA gene (147 bp), the uidA gene specific for the E. coli O157:H7 serovar (252 bp), the SLT-I gene (348 bp), and the SLT-II gene (584 bp) were generated with suitable PCR primers from toxigenic and nontoxigenic E. coli strains isolated from various outbreaks (Fig. 1). The uidA gene was observed in all 37 E. coli strains and both Shigella strains tested. E. coli strains with the O157:H7 serotype were distinguished simultaneously with the SLT type known to be produced by these strains. The amplicons generated approximated the sizes predicted on the basis of the selected primer sets. As anticipated, only the uidA gene (147 bp) was amplified from wild-type E. coli, whereas the expected toxin gene-specific products, but not O157:H7-specific products (252 bp), were amplified from the SLT-producing nonO157:H7 serotypes examined. PCR results indicated that ATCC strains 43890, 43894, and 43895 were positive for an SLT-I-homologous sequence (348 bp) and that ATCC strains 43894 and 43895 were positive for SLT-II (548 bp). Several of the SLT-producing, non-O157:H7 serotypes of E. coli examined produced only toxin gene-specific amplification products. The type of SLT identified by the multiplex assay correlated well with the Vero cell toxicity data for these isolates (data not
4129
FIG. 1. Agarose gel electrophoresis of amplicons generated by multiplex PCR from E. coli strains isolated from various outbreaks. Lane 1, type strain ATCC 25922; lane 2, O157:H7 strain producing SLT-I; lane 3, O157:H7 strain producing SLT-I and SLT-II; lanes 4 and 5, O157:H7 strains producing SLT-II; lane 6, virulent strain not producing SLT-I or SLT-II; lanes 7 to 9, strains other than O157 serovar producing SLT-I; lanes 10 to 14, strains isolated from urinary tract and veterinary infections; lane M, 100-bp marker. Numbers to the left of the gel are molecular sizes (base pairs).
shown). Among the non-E. coli strains tested, only S. dysenteriae yielded the SLT-I amplicon. This result is not unexpected, since the Shiga toxin of S. dysenteriae type 1 is almost identical to the SLT-I of O157:H7. Although the multiplex PCR assay does not discriminate between S. dysenteriae type 1 and nonO157:H7 EHEC serotypes that produce only SLT-I, the uidAspecific primers for the O157:H7 serotype readily distinguished S. dysenteriae type 1 species from O157:H7 isolates. Detection of E. coli O157:H7 in beef by multiplex PCR and removal of PCR-inhibitory products from foods. A 25-g sample of sliced beef was homogenized for 1 min with a homogenizer (model SH-001; Elmex, Tokyo, Japan) in 225 ml of Trypticase soy broth (TSB) (Nissui) to produce a beef homogenate for all experiments. E. coli ATCC 25922 (type strain) and ATCC 43895 (O157:H7) were grown in TSB overnight at 37°C and serially diluted with beef homogenate as the diluent to get final concentrations ranging from 0 to 109 CFU g21. These artificially inoculated beef homogenate microcosms (10 ml each) were incubated at 37°C with shaking (140 rpm; rotary shaking). After incubation, subsamples (1 ml) were drawn at intervals of 0, 6, and 18 h, centrifuged (4°C; 10,000 3 g for 10 min), and resuspended in sterile PBS. A 10-ml sample suspension was used as the template for the PCR assay in the absence of DNA extraction. Appropriate controls, such as buffer alone, PCR mixture with no DNA, and uninoculated food samples, were run as negative controls to check for the possibility of contamination. Total viable counts were enumerated in Trypticase soy agar. The typical E. coli and sorbitol-negative O157:H7 strains were counted in Fluorocult E. coli O157:H7 agar, when necessary. The sensitivity of the PCR assay for detecting artificially inoculated E. coli O157:H7 43895 in beef is presented in Table 2. When whole bacterial cells were resuspended in sterile PBS, 102 CFU of E. coli in a 10-ml PCR mixture was needed to amplify all four amplicons. Amplicons specific for E. coli or toxigenic genes were not amplified for the uninoculated food sample (Fig. 2, lanes F). At time zero, 105 CFU g of beef homogenate21 failed to yield the desired PCR products. Although a faint band at 147 bp (E. coli specific) was noticed after 18 h of enrichment in TSB, the amplicons generated by toxigenic strain- and O157:H7-specific probes were not distinct. This indicates that food particles and/or other unknown metabolic by-products are inhibitory to the PCR. To remove any PCR-inhibitory substances from food, subsamples drawn from the microcosms were subjected to twostep filtration. A 400-ml sample of a sterile-PBS-washed sample was passed through a 5-mm-diameter Ultrafree filter tube (cat-
4130
VENKATESWARAN ET AL.
APPL. ENVIRON. MICROBIOL.
TABLE 2. Sensitivity of filtration method in the amplification of various bands specific to E. coli O157:H7 serovar and its toxins PCR band(s) amplified when the initial inoculum level was in the range of (CFU g of food slurry21)a:
Preenrichment incubation in TSB (h)
105
104
Untreated food slurry
0 6 18
None E, O, I, II E, O
None E, O, I, II E, O
None E, O, I E, O
None None E, O
None None E, O
None None E, O
Beef slurry (5-mm pore size filtered)
0 6 18
None E, O, I, II O, I, II
None E, O, I, II O, I, II
None E, O, I, II None
None None II
None None None
None None II
Beef slurry (5-mm pore size filtered; retained on 0.2-mm-pore-size filters)
0 6 18
None E, O, I, II E, O, I, II
None E, O, I, II E, O, I, II
None E, O, I, II E, O, I, II
None None E, O, I, II
None None E, O, I, II
None None E, O, I, II
0 6 18
105/10 107/10 108/10
104/10 105/10 108/107
103/10 104/102 108/108
102/10 103/102 107/108
10/10 102/103 107/108
1/10 10/103 106/108
Sample or parameter
Populationb
a b
103
102
101
1
Abbreviations: E, E. coli band (147 bp); O, E. coli O157:H7 band (252 bp); I, SLT-I band (348 bp); II, SLT-II band (584 bp). The numerator and denominator denote E. coli O157:H7 serovar and food microflora populations (CFU per gram), respectively.
alog no. UFC3 0GV; Millipore, Bedford, Mass.) and centrifuged (10,000 3 g, 4°C, 10 min). The 5-mm-diameter-tube filtrate was then passed through a 0.2-mm-diameter Ultrafree centrifuge tube and centrifuged (10,000 3 g, 4°C, 10 min) to remove bacteria. The material trapped on the 0.2-mm-poresize filter was then resuspended in 400 ml of sterile PBS, and a 10-ml sample was used as a template for the PCR assay. Particulate matter, such as food particles, was removed by passing the samples through 5-mm-pore-size Ultrafree membrane filters, and dissolved materials were removed by simple centrifugation. Bacterial cells were subsequently captured on 0.2-mm-pore-size Ultrafree membrane filters. The PCR product bands were not clearly visible when samples were passed through 5-mm-pore-size filters only (Table 2). When beef samples were cultured in TSB for 6 h (at 37°C), it was possible to detect initial sample inocula of 102 CFU of E. coli 25922 (Fig. 2A, lane 3) and 103 CFU E. coli ATCC 43895 (O157:H7) (Fig. 2A, lane 14). When 18-h, TSB-grown samples were filtered through 5-mm-followed by 0.2-mm-pore-size filters, both strains, E. coli ATCC 25922 (Fig. 2B, lane 6) and ATCC 43895 (Fig. 2B, lane 17), were detected at an initial sample inoculum of 1 CFU g21. Filtration greatly improved the sensitivity of the assay, such that it was possible to detect low numbers (102 to 103) of E. coli cells per gram after 6 h of enrichment and even 1 CFU g21 after 18 h of enrichment. After 18 h of incubation and the two-step filtration, the PCR products became much more distinct and low-cell-number bands appeared. Absence of PCR product from 18-h, TSB-grown food homogenate clearly shows that metabolic by-products remaining in the filtrate after 5mm-pore-size filtration posed a problem for PCR amplification. The sensitivity of PCR amplification was greatly enhanced once dissolved metabolites were eliminated from samples by 0.2-mm-pore-size filtration. Influence of competitive food microflora on the sensitivity of multiplex PCR. The results presented here indicate that the E. coli O157:H7 detection was enhanced when the beef homogenate was subjected to filtration. Although 8 to 10% of the bacterial population was removed along with particulate matter, the majority of the bacterial cells were collected after 5-mm-pore-size filtration. This was evident at any preenrich-
ment incubation level. When 6-h, TSB-enriched beef homogenate was passed through 5-mm-pore-size filters, a minimum of 104 CFU of E. coli O157:H7 cells per g was required to amplify all four bands (faint bands), or the ratio of E. coli O157:H7 to other, competitive microflora in the sample had to be in the range of 102/1 (Table 2). However, a combination of 18 h of enrichment in TSB and a simple two-step filtration treatment generated all four bands clearly, even when E. coli O157:H7 and other food microflora were present at a ratio of 1/102
FIG. 2. Multiplex PCR amplicons from treated and untreated beef samples cultivated for 6 h (A) or 18 h (B). Amplicons from artificially inoculated E. coli ATCC 25922 (lanes 1 to 7) and ATCC 43895 (O157:H7) (lanes 11 to 17) are shown. Results at time zero: lanes 1 and 11, 105 CFU g21; lanes 2 and 12, 104 CFU g21, and so on. Lane F, E. coli-free beef control; lane M, 100-bp marker. Numbers to the left of the panels are molecular sizes (base pairs).
DETECTION OF E. COLI O157:H7 IN BEEF BY MULTIPLEX PCR
VOL. 63, 1997
(Table 2). However, successful PCR amplification after an overnight incubation was due to not only the proliferation of the target organism but the removal of PCR-inhibitory substances as well. The development of molecular methodologies to detect pathogenic microorganisms in food, clinical, and environmental samples has led to improved patient diagnosis and the more precise determination of public health risks associated with food consumption and environmental exposure. The standard bioassays used for identification of pathogenic E. coli, such as assays for cytopathic effects on Vero cells and rabbit ileal loop assays, are difficult to adapt for the screening of large numbers of E. coli isolates. Bioassays are labor-intensive and costly, and the results can take several weeks. The application of molecular approaches in food microbiology offers a more efficient alternative to these labor-intensive methods. The multiplex PCR method described here is a highly effective means for specifically detecting and characterizing EHEC organisms directly from food. The major advantage of this protocol over existing assays is that it can identify the types of SLT encoded by the strain and simultaneously discriminate other SLT-producing E. coli strains from O157:H7, the predominant serotype implicated in disease. We are thankful to M. Satake for encouragement; T. Takeda, National Children’s Medical Research Center, and K. Tamura, National Institute of Health—Japan, for the supply of bacterial strains; K. Nealson and D. Moser for critically reading the manuscript; and K. Hanai, T. Kurusu, and A. Murakoshi for their technical assistance. REFERENCES 1. Aleksic, S., H. Karch, and J. Bockemuhl. 1992. A biotyping scheme for Shiga-like (Vero) toxin-producing Escherichia coli O157 and a list of serological cross-reactions between O157 and other gram-negative bacteria. Zbl. Bakteriol. 276:221–230. 2. Bej, A. K., S. C. McCarty, and R. M. Atlas. 1991. Detection of coliform bacteria and Escherichia coli by multiplex polymerase chain reaction: comparison with defined substrate and plating methods for water quality monitoring. Appl. Environ. Microbiol. 57:2429–2432. 3. Bettelheim, K. A., H. Evangelidis, J. L. Pearce, E. Sowers, and N. A. Strockbine. 1993. Isolation of a Citrobacter freundii strain which carries the Escherichia coli O157 antigen. J. Clin. Microbiol. 31:760–761. 4. Cebula, T. A., W. L. Payne, and P. Feng. 1995. Simultaneous identification of strains of Escherichia coli serotype O157:H7 and their Shiga-like toxin type by mismatch amplification mutation assay-multiplex PCR. J. Clin. Microbiol. 33:248–250. 5. Centers for Disease Control and Prevention. 1993. Update: multistate outbreak of Escherichia coli O157:H7 infections from hamburgers—western United States, 1992–1993. Morbid. Mortal. Weekly Rep. 42:258–263. 6. Cha, R. S., H. Zarbl, P. Keohavong, and W. G. Thilly. 1992. Mismatch amplification mutation assay (MAMA): application to the c-H-ras gene. PCR Methods Applic. 2:14–20. 7. Chang, G. W., J. Brill, and R. Lum. 1989. Proportion of b-D-glucuronidasenegative Escherichia coli in human fecal samples. Appl. Environ. Microbiol. 55:335–339. 8. Feng, P. 1993. Identification of Escherichia coli O157:H7 by DNA probe specific for an allele of uidA gene. Mol. Cell. Probes 7:151–154.
4131
9. Feng, P., and K. A. Lampel. 1994. Genetic analysis of uidA gene expression in enterohemorrhagic Escherichia coli serotype O157:H7. Microbiology 140: 2101–2107. 10. Feng, P., R. Lum, and G. W. Chang. 1991. Identification of uidA gene sequences in b-D-glucuronidase-negative Escherichia coli. Appl. Environ. Microbiol. 57:320–323. 11. Fratamico, P. M., R. L. Buchanan, and P. H. Cooke. 1993. Virulence of an Escherichia coli O157:H7 sorbitol-positive mutant. Appl. Environ. Microbiol. 59:4245–4252. 12. Gannon, V. P. J., R. K. King, J. Y. Kim, and E. J. G. Thomas. 1992. Rapid and sensitive method for detection of Shiga-like toxin-producing Escherichia coli in ground beef using the polymerase chain reaction. Appl. Environ. Microbiol. 58:3809–3815. 13. Griffin, P. M., and R. V. Tanke. 1991. The epidemiology of infections caused by Escherichia coli O157:H7, other enterohemorrhagic Escherichia coli, and the associated hemolytic uremic syndrome. Epidemiol. Rev. 13:60–98. 14. Jackson, M. P., R. J. Neill, A. D. O’Brien, R. K. Holmes, and J. W. Newland. 1987. Nucleotide sequence analysis and comparison of the structural genes for Shiga-like toxin I and Shiga-like toxin II encoded by bacteriophages from Escherichia coli 933. FEMS Microbiol. Lett. 44:109–114. 15. Jinneman, K. C., P. A. Trost, W. E. Hill, S. D. Weagant, J. L. Bryant, C. A. Kaysner, and M. M. Wekell. 1995. Comparison of template preparation methods from foods for amplification of Escherichia coli O157 Shiga-like toxins type I and II DNA by multiplex polymerase chain reaction. J. Food Prot. 58:722–726. 16. Karmali, M. A. 1989. Infection by verotoxin-producing Escherichia coli. Clin. Microbiol. Rev. 2:15–38. 17. Lang, A. L., Y.-L. Tsai, C. L. Mayer, K. C. Patton, and C. J. Palmer. 1994. Multiplex PCR for detection of the heat-labile toxin gene and Shiga-like toxin I and II genes in Escherichia coli isolated from natural waters. Appl. Environ. Microbiol. 60:3145–3149. 18. Levine, M. M. 1987. Escherichia coli that cause diarrhea: enterotoxigenic, enteropathogenic, enteroinvasive, enterohemorrhagic, and enteroadherent. J. Infect. Dis. 155:377–389. 19. O’Brien, A. D., and R. K. Holmes. 1987. Shiga and Shiga-like toxins. Microbiol. Rev. 51:206–220. 20. Padhye, N. V., and M. P. Doyle. 1992. Escherichia coli O157:H7: epidemiology, pathogenesis, and methods for detection in food. J. Food Prot. 55:555– 565. 21. Rice, E. W., E. G. Sowers, C. H. Johnson, M. E. Dunnigan, N. A. Strockbine, and S. C. Edberg. 1992. Serological cross-reactions between Escherichia coli O157 and other species of the genus Escherichia. J. Clin. Microbiol. 30:1315– 1316. 22. Saliba, L. J., and R. Helmer. 1990. Health risks associated with the pollution of coastal bathing waters. World Health Stat. Q. 43:177–187. 23. Todd, E. C. D., R. A. Szabo, P. Peterkin, A. N. Sharpe, L. Parrington, D. Bundle, M. A. J. Gidney, and M. B. Perry. 1988. Rapid hydrophobic grid membrane filter–enzyme-labeled antibody procedure for identification and enumeration of Escherichia coli O157 in foods. Appl. Environ. Microbiol. 54:2536–2540. 24. Venkateswaran, K., A. Murakoshi, and M. Satake. 1996. Comparison of commercially available kits with standard methods for the detection of coliforms and Escherichia coli in foods. Appl. Environ. Microbiol. 62:2236– 2243. 25. Venkateswaran, K., A. Shimada, A. Maruyama, T. Higashihara, H. Sakou, and T. Maruyama. 1993. Microbial characteristics of Palau Jellyfish Lake. Can. J. Microbiol. 39:506–512. 26. Willshaw, G. A., H. R. Smith, D. Roberts, J. Thiriwell, T. Cheasty, and B. Rowe. 1993. Examination of raw beef products for the presence of Vero cytotoxin producing Escherichia coli, particularly those of serogroup O157. J. Appl. Bacteriol. 75:420–426. 27. Witham, P. K., C. T. Yamashiro, K. J. Livak, and C. A. Batt. 1996. A PCR-based assay for the detection of Escherichia coli Shiga-like toxin genes in ground beef. Appl. Environ. Microbiol. 62:1347–1353.
