Application of Multiplex PCR for Detection of Non-O157 Verocytotoxin-Producing Escherichia coli in Bloody Stools: Identification of Serogroups O26 and O111

Multiple serotypes of verocytotoxin (Shiga-like toxin)-producing Escherichia coli (VTEC) have been isolated from humans, but E. coli serotype O157:H7 has been the most common (11, 14). The spectrum of disease associated with E. coli O157:H7 has been well described, but the role of non-O157 VTEC strains in causing human enteric disease has been less well defined (11, 14). Recent reports have implicated the emerging importance of multiple non-O157 VTEC serotypes as a cause of hemorrhagic colitis and hemolytic-uremic syndrome (2, 3, 17, 26), with strains belonging to serogroups O26 and O111 being the most commonly associated with human disease (6, 11, 17, 23). Failure to ferment sorbitol and serologic agglutination with specific O157 and H7 antisera allow for the rapid detection of VTEC O157:H7. Current methods for the detection of non-O157 VTEC depend primarily on the presence of verocytotoxin (VT) and strain isolation and then on serotyping. Although slide agglutination testing with enteropathogenicE. coli serogroup O26- and O111-specific antisera will result in cross-reactions with the respective VTEC serogroups (23), most clinical laboratories do not stock such antisera nor do they look for VTEC serogroups other than O157. VT or the genes that encode it can be detected by using tissue cell assays and PCR or hybridization detection of VT genes (14, 21, 24). The detection of serum antibodies to either VT or the lipopolysaccharide of the infecting strain has been available only in research settings (6, 19, 22). Commercial enzyme immunoassays (EIA) to detect VT or O157 lipopolysaccharide directly in stools may be useful especially for large-scale testing (7, 18). These kits are specific only for the genes targeted, so that the O157 EIA does not detect non-O157 VTEC strains (7). Our efforts have focused on the development of a multiplex PCR assay to detect the more common VTEC serogroups, O26, O111, and O157, directly in stool specimens. We describe here the application of this assay for detection of VTEC serotype O26:H11 directly in bloody fecal specimens from a patient with severe hemorrhagic colitis.

A previously healthy 16-year-old female was admitted to hospital with an abrupt onset of bloody diarrhea, abdominal cramps, nausea, and vomiting after a 2-day history of fever, chills, and malaise. She did not recall eating anything unusual or partially cooked foods. No other family members or friends were ill, and she had been on no medications. She had no exposures to animals or any travel history. On examination, she had generalized abdominal tenderness, with guarding in the lower quadrants bilaterally. Laboratory investigations revealed normal hemoglobin, leukocyte, and platelet counts and renal and liver function tests. Shigella spp., Salmonella spp.,Campylobacter spp., and Yersinia spp. were not detected by culture in the visibly bloody stool specimens. Stool specimens were also negative for Clostridium difficiletoxin. There was a heavy growth of sorbitol-negative colonies on sorbitol MacConkey agar (SMAC), but slide agglutination testing with Bacto E. coli O Antiserum O157 (Difco Laboratories, Detroit, Mich.) was negative. The persistence of bloody stools prompted in-house testing for non-O157 VTEC isolates with a multiplex PCR assay directly applied to another two fecal specimens and to colony sweeps of their 24-h-old SMAC plates.

PCR performed directly on stool specimens and colony sweeps were positive for VT1 and the eaeA gene of E. coli O26:H11 but negative for VT2 and the eaeAgenes of E. coli serotypes O157:H7 and O111:H8 (Fig.1). PCR performed on multiple isolated colonies also confirmed the absence of a coinfection with strains of serogroups O111 and O157. Once the patient’s attending physician was notified of the results, considerations for further investigations such as endoscopy and laparotomy were cancelled. The patient recovered over the next several days without any complications. The sorbitol-negative isolate was identified as E. coli by using the Vitek GNI card (BioMerieux Vitek Inc., Hazelwood, Mo.) and was subsequently confirmed to be a VTEC strain of serotype O26:H11 by the Ontario Ministry of Health Reference Laboratory.

• Open in new tab
• Download powerpoint

Fig. 1.

Multiplex PCR amplification of VTEC serotypes O26:H11, O111:H8, and O157:H7 directly in clinical and seeded stool specimens. Lane 1, 123-bp DNA size markers (Gibco BRL); lanes 2 to 4, stool specimens from patient that were VT1 andeaeA_O26 positive (lane 2), VT1 positive andeaeA_O111 negative (lane 3), and VT1 positive andeaeA_O157 negative (lane 4); lane 5, seeded stool specimen with a strain of E. coli O111:H8 that was VT1 and eaeA_O111 positive; lane 6, clinical stool specimen with a strain of E. coli O157:H7 that was VT1, VT2, and eaeA_O157 positive; lane 7, clinical specimen negative for VTEC and seeded with E. coli ATCC 25922; lanes 8 to 10, positive controls with E. coliO26:H11, O111:H8, and O157:H7 strains, respectively; lane 11, PCR assay reagent control (no template DNA).

For the multiplex PCR assay, fecal specimen DNA was extracted by using the XTRAX DNA extraction kit (Gull Laboratories Inc., Salt Lake City, Utah). Colony sweeps and isolated colonies from agar plates were each suspended in 100 μl of 1× PCR buffer (Perkin-Elmer Cetus, Norwalk, Conn.) and boiled for 10 min prior to use as template DNA. The multiplex PCR assay was performed as described by Paton and Paton (20) but with 5 pmol each of VT1 primers, 2.5 pmol of VT2 primers, and 20 pmol each of the eaeA gene primers (either eaeA_O157:H7,eaeA_O111:H8, oreaeA_O26:H11 primers) (Gibco-BRL). Primers (Table 1) were designed to detect VT1 and VT2 gene sequences (9), and specificeaeA gene sequences of E. coli serotypes O157:H7 (16), O111:H8 (16), and O26:H11 (12). Each PCR amplification assay included positive controls from boiled bacterial suspensions of E. coliO157:H7 strain CL8 (positive for VT1, VT2, and eaeA),E. coli O111:H8 strain RC541 (positive for VT1 andeaeA), and E. coli O26:H11 strain RC559 (positive for VT1 and eaeA); a negative control fromE. coli ATCC 25922; and a reagent control (no template DNA). Thermocycling conditions with a GeneAmp 9600 thermocycler (Perkin-Elmer Cetus) were as follows: 94°C for 40 s, 56°C for 40 s, and 72°C for 40 s, for 25 cycles, ending with a 3-min extension at 72°C. PCR products were visualized on a 1% agarose gel after staining with ethidium bromide.

The gene sequence variations in the 3′ end of eaeA gene homologues of VTEC serotypes O157:H7 and O111:H8 allowed the development of primers with sufficient serotype specificity to be used in multiplex PCR assays (16). Hii et al. sequenced the 3′ end of the E. coli O26:H11 eaeA gene and demonstrated preliminary data for detecting E. coliO26:H11 strains (12). The specificity of theeaeA_O26:H11,eaeA_O111:H8, andeaeA_O157:H7 primers were tested on 199 human and cattle VTEC isolates of diverse serotypes (Table2). These isolates were known to possess VT genes and the highly conserved central region of the eaeA gene (16). Of the 30 different VTEC serotypes, serotypes O5:NM, O69:H11, O98:NM, O118:NM, and O118:H16 were detected by PCR by using theeaeA_O26:H11 primers in addition to strains of serogroup O26. TheeaeA_O111:H8 primers detected strains of serotypes O76:H25 and O156:H25 in addition to O111:H8 and O111:NM strains. The eaeA_O157:H7 primers detected strains of serotypes O145:H8 and O145:NM in addition to O157:H7 strains. Although there was cross-detection of serotypes when these primers were used, the most frequent VTEC serotypes associated with bloody diarrhea in cattle include serotypes O5:NM, O26:H11, O103:H2, O111:H8, O111:H11, O111:NM, and O118:H16 (5). Thus, theeaeA_O26 andeaeA_O111 primers would be useful in screening fecal samples from calves with hemorrhagic colitis since most of these serotypes except for O103:H2 and O111:H11 would be detected. In humans, these cross-reactions may also not be problematic since VTEC strains of serotypes O69:H11, O76:H25; O118:NM, O145:H8, and O156:H25 have not been implicated in causing human disease (10, 14, 27). Strains of serotype O5 have been associated with human disease only infrequently (11, 14, 27). Therefore, eaeA_O26 andeaeA_O111 primers would be useful in screening human fecal samples, even though they may also detect VTEC strains of a few other serotypes that are rarely associated with human disease.

Recognition of the frequency and significance of non-O157 VTEC-associated infections has been hampered by the lack of specific diagnostic tools for rapid detection in the clinical laboratory (3, 26). Culture methods do not readily detect non-O157 VTEC strains since not all VTEC strains are sorbitol negative (15, 26). Direct fecal VT cytotoxicity assays are more sensitive, while the testing of individual isolates results in lower sensitivity unless large numbers are screened (21, 24). Detection of VT genes alone may not suffice, since it is not yet known which VT-producing bacteria are truly pathogenic in humans (26). There is also the potential for strains to lose their VT genes upon serial passage or subculture (13).

Various PCR-based strategies to look for E. coliO157:H7 and other VTEC serogroups have been developed (8, 16, 20, 21). Paton et al. initially incubated feces in an overnight broth culture, prior to PCR amplification, which resulted in a delay in detection (20, 21). Most recently, they first used a two-step multiplex PCR assay to detect VT1, VT2, eaeA, andhlyA (enterohemolysin) gene sequences and then a second multiplex assay to detect the O antigen-encoding regions ofE. coli O157 (rfb_O157) and O111 (rfb_O111) (20). Their genericeaeA primers detected a conserved region of theeaeA gene in both enteropathogenic E. coliand VTEC, and their hlyA primers were also not serotype specific (25). The combination of primers to detect VT1, VT2, and the rfb_O157 andrfb_O111 regions might be more serotype specific; however, the amplified fragments ofrfb_O157 and VT2 were too similar in size to be easily discernible by gel visualization. Unlike these previous studies, in which PCR was performed on crude DNA extracts from primary fecal broth cultures, our multiplex PCR assay detected VT genes and provided specific serotype identification directly in fecal samples within 2 to 3 h of specimen receipt.

To our knowledge, this report illustrates the first description of serotype O26 primers used in a multiplex PCR assay directly on stool specimens to provide an etiologic diagnosis of hemorrhagic colitis. Many cases of non-O157 VTEC-associated disease are being diagnosed only fortuitously (1, 15, 23). The development of methods for rapid and accurate diagnosis will enhance studies on the epidemiology of VTEC-associated disease, prevent the use of unnecessary antibiotics and investigative procedures, allow the use of novel treatment modalities (2, 4, 26) which may prevent the development of hemolytic-uremic syndrome, and reduce public health risks by identifying potential sources more quickly. VTEC, especially non-O157:H7 VTEC, must be considered a possible cause of bloody diarrhea, regardless of whether the organism isolated from SMAC plates is sorbitol fermenting or non-sorbitol fermenting (2, 23, 26). Although we used a multiplex PCR assay, clinical laboratories may consider the use of commercially available EIA assays as an initial rapid screen for VTEC. Clinical laboratories should be vigilant of the potential for non-O157 VTEC to cause enteric disease.

• Copyright © 1998 American Society for Microbiology