Comparison of Multiplex Immunochemical and Molecular Serotyping Methods for Shiga Toxin–Producing Escherichia coli

Introduction

Food safety concerns have been on the rise due to the increased number of human illness outbreaks caused by Shiga toxin–producing Escherichia coli (STEC)–contaminated foods. These human illnesses (Paton et al., 2000; Mathusa et al., 2010) include mild-to-bloody diarrhea, abdominal cramps, vomiting, strokes, hemorrhagic colitis, hemolytic uremic syndrome, and thrombotic thrombocytopenic purpura. Severe cases can end with chronic renal failure, chronic nervous system deficiencies, and death (Brooks et al., 2005; Mathusa et al., 2010). Although O157 is the best-known serogroup for STEC infection, in 2013 the largest STEC outbreak in the United States was traced to serogroup O121 in frozen food (CDC, 2013). In 2012, the largest U.S. STEC outbreak was O157 in spinach salad, but the next two largest were traced to O26 in clover sprouts (CDC, 2012a) and O145, for which no source of contamination was identified (CDC, 2012b).

According to the Centers for Disease Control and Prevention (Scallan et al., 2011), 48 million cases of foodborne illness occur annually in the United States, but only 9.4 million are attributed to major pathogens. The others are caused by unspecified agents. In order to understand how to prevent the 80% of foodborne illnesses caused by unspecified agents (CDC, 2011), we need to better identify pathogenic microbial organisms. Improvements in detection methods are key to understanding etiology. A valid assay for detection and characterization of non-O157 STEC is likely to generate a significant impact on management of a major portion of the 38.4 million cases of unattributed disease annually in the United States. In this study, conventional methods of serotyping using antisera and two newly developed multiplex polymerase chain reaction (PCR)– and antibody-based microbead assays using the Luminex technology are compared using 161 STEC isolates. These isolates were obtained from the fecal samples of cattle in feedlots, on irrigated pastures, on the range, and in dairy farms in California.

Materials and Methods

The STEC isolates used in this study were previously isolated from fecal samples of cattle from dairy farms, feedlots, and cow–calf operations grazing on irrigated pastures or rangeland forages in California (Bollinger, 2008). These isolates were typed for the O antigen through the use of conventional serotyping (Bollinger, 2008) using slide agglutination with rabbit antisera kits (Denka Seiken Co.).

The Luminex assays used in this study were previously described (Lin et al., 2011; Clotilde et al., 2013). The 7-plex Luminex immunoassay (Clotilde et al., 2013) is able to identify strains belonging to serogroups O26, O45, O103, O111, O121, O145, and O157. Briefly, this immunoassay was conducted in black, 96-well round-bottom polystyrene microplates (Corning Costar, Lowell, MA), according to Luminex standard immunoassay protocols. All incubations were for 1 h at room temperature, protected from light, and shaken at 600 rpm. A 100-μL aliquot of each test sample was combined with 5000 microbeads of each of the 7 different assay specificities mixed in a single microplate well. The microbeads were washed from unbound reaction components using the Bio-Plex Pro Wash Station (Bio-Rad Laboratories, Inc., Hercules, CA). Samples were washed three times with phosphate-buffered saline (PBS)–T (PBS with 0.1% Tween20). Microbeads were then resuspended in 100 μL of 4 μg/mL biotinylated detector antibodies, and the resultant mixture was incubated for 1 h at room temperature on a rocking platform. The mixture was then washed 3 times as described earlier, resuspended in 100 μL of 4 μg/mL streptavidin labeled with R-phycoerythrin, and the resultant mixture was incubated, washed, and resuspended in 100 μL of PBS. The samples were analyzed on a Luminex 100 flow analyzer with Bio-Plex Manager 5.0 standard software (Bio-Rad). Data were acquired for 120 s. Background fluorescence was measured using samples containing all assay reagents except the analyte(s) of interest (i.e., blanks).

The 10-plex PCR-based Luminex assay (Lin et al., 2011) is able to identify all of the same 7 serogroups as the immunoassay (O26, O45, O103, O111, O121, O145, and O157), plus O91, O113, and O128. PCR reactions and microbead hybridization were performed as described (Lin et al., 2011). PCR reactions were performed in an Applied Biosystems Verti thermal cycler (Life Technologies, Grand Island, NY). Microbead hybridization was performed according to the standard Luminex protocol in a 96-well reaction plate (Bio-Rad). Briefly, microbeads were diluted to a concentration of 150 per μL for each bead region in 1.5×TMAC solution (4.5 M tetramethylammonium chloride) (TMAC) solution (Sigma-Aldrich, St. Louis, MO), 0.15% Sarkosyl (Sigma-Aldrich), 75 mM Tris-HCl pH 8.0 (Sigma-Aldrich), 6 mM EDTA pH 8.0 (Sigma-Aldrich). Five thousand microbeads of each bead region were used per analysis. Five microliters of PCR reaction was mixed with microbeads in Tris EDTA buffer pH 8.0 (Sigma-Aldrich) in a total volume of 50 μL. Reaction mixtures were denatured at 94°C for 15 min and then hybridized at 52°C for 30 min in a thermal cycler. Microbeads were collected on a LifeSep 96F magnetic separation unit (Dexter Magnetic Technologies, Elk Grove Village, IL), and supernatants were removed. Microbeads were resuspended in 75 μL of 4 μg/mL Streptavidin labeled with R-Phycoerythrin (SAPE; Invitrogen) in 1.0×TMAC (3 M TMAC [Sigma-Aldrich], 0.1% Sarkosyl [Sigma-Aldrich], 50 mM Tris–HCl pH 8.0 [Sigma-Aldrich], 4 mM EDTA [Sigma-Aldrich] with 0.1% bovine serum albumin [Sigma-Aldrich]). Microbeads were analyzed on Bio-Plex 200 with Bio-Plex Manager 6.0 software with the following settings: 100 beads per region, 120-s timeout, 50-μL sample volume, platform heater 52°C, and DD gate 5000-25000 with high RP1 target turned on. Signal-to-background ratios >5.0 were considered positive for that analyte, with background fluorescence calculated for samples containing all components except for template DNA.

The isolates that had conflicting serogroups (i.e., belonging to one serogroup according to the Denka Seiken serotyping kit, but showing up as a different one by the 10-plex PCR-based Luminex assay) were subjected to additional confirmatory serotyping. Where O157 was suspected, the serogroup was confirmed using the Remel O157 latex test (Thermo Scientific, Lenexa, KS). Otherwise, we performed confirmation using O serogroup-specific antisera (O26, O45, O91, O103, O111, O113, O121, O128, O145, and O157) from MiraVista/Statens Serum Institut (Indianapolis, IN) by slide agglutination.

Results

We compared the results of serotyping 161 unique field isolates of STEC using three different techniques: conventional latex agglutination, a Luminex PCR assay, and a Luminex immunoassay (Table 1). For 138 strains (86%), the results of all 3 assays agreed; in 23 cases (14%) we observed different results. Of these 23 strains, 11 were untypeable by Denka Seiken, and 11 were apparently mistyped by Denka Seiken. In these 23 cases of assay disagreement, we attempted to confirm serotyping using a fourth test, which was antibody based. One strain (#85) was unable to be serotyped by our Luminex immunoassay or by a fourth test, although the Luminex PCR test confirmed the Denka Seiken result as serotype O157.

Eleven strains were untypeable (OUT) using conventional methods of serotyping with antisera, but were typed successfully using both Luminex assays. In 9 of these 11 cases, the 2 Luminex results agreed with each other, and for 8 of these 9 strains the serotype was further confirmed as O157 by Remel O157 latex agglutination. The 9^th strain (#120) was O45 in both Luminex assays. The other two strains that were untypeable with the Denka Seiken serotyping kit (#25 and #32) were also untypeable with the Luminex immunoassay, but were identified as O111 via Luminex PCR and confirmed with the MiraVista/Statens Serum Institut assay.

A different subset contained 11 strains that were apparently mistyped by the Denka Seiken serotyping kit. Of these 11 strains, 7 were untypeable by our Luminex immunoassay, but were typed as O128 via both the Luminex PCR assay and the Statens confirmatory test. An 8^th strain (#70) typed as O121 via Luminex immunoassay, but typed as both O121 and O128 by both the Luminex PCR assay and the Statens confirmatory test. Two strains typed as O125 (#86) and O158 (#105) by Denka Seiken were typed as O157 in both Luminex assays and in the Remel test. The 11^th strain (#88) was typed as O137 by Denka Seiken but as O111 by both of the Luminex assays as well as the Statens test.

Another way of analyzing the results is a comparison of the two Luminex tests. Our Luminex PCR assay serotyped all 161 strains, while 10 strains (6%) were untypeable by our Luminex immunoassay. In 9 of these 10 cases, confirmation was provided by the Statens assay. As described above the 10^th case (#85) was typeable only by the Luminex PCR test. Strains that were typed by the Luminex PCR assay but untypeable by the Luminex immunoassay may have failed to express their genes for production of the O-antigen at levels high enough for immunoassay detection.

Discussion

Development of ways to rapidly serotype STEC shed and carried by cattle is extremely important, as certain STEC serogroups are now considered adulterants in beef products destined for human consumption. The implementation of rapid and accurate serotyping methods to monitor the STEC prevalence in those operations may inform producers how to modify cattle handling and beef processing, in order to significantly decrease STEC prevalence on cattle hides, minimizing microbial contamination of carcasses at beef-processing plants (Ridell et al., 1993; Bell, 1997; McEvoy et al., 2000). Numerous efforts have been devoted to evaluate the significance of STEC on hides, and ways to decrease the carriage of STEC by feedlot cattle, because cattle hides can play a major role in carcass contamination at slaughter (Bacon et al., 2000; Elder et al., 2000; Barkocy-Gallagher et al., 2003; Bollinger et al., 2008; Ennis et al., 2012).

As a result of the first two major outbreaks of human STEC illnesses attributed to consumption of beef, specifically hamburgers contaminated with E. coli O157 (Riley et al., 1983), safety concerns with beef have gained attention from producers, consumers, and governmental agencies. For this reason, most U.S. studies have focused on prevalence of E. coli O157 in cattle (Barkocy-Gallagher et al., 2003; Hancock et al., 1994). However, prevalence of non-O157 STEC has been shown to be much higher than O157 in cattle (Bettelheim, 2001).

Over 435 different STEC serotypes have been recovered from cattle, and over 470 STEC serotypes have been isolated from humans (Mathusa et al., 2010). Notably, there is overlap among STEC serotypes carried by cattle and humans (Mathusa et al., 2010), although cattle remain asymptomatic of STEC colonization. Even though E. coli O157:H7 is the most commonly isolated serotype in outbreaks, O26 is the second most isolated serogroup in reported outbreak-related STEC cases in Asia, Europe, and the United States (Mathusa et al., 2010). The six non-O157 STEC most commonly isolated in human illnesses outbreaks in the United States are the following: O26, O45, O103, O111, O121, and O145 (Brooks et al., 2005).

We found our PCR-based Luminex assay outperformed conventional serotyping methods using antisera, as it was able to serotype 22 additional strains. It also outperformed our antibody-based Luminex assay, serotyping an additional 10 strains. A total of 22 STEC isolates gave 1 result by the conventional serotyping method using antisera, and a different result with both Luminex assays, as well as confirmatory assays. Only one strain was untypeable via Luminex immunoassay and typed as O157 by conventional methods of serotyping using antisera and the Luminex genetic assay. In addition to reliable data, the Luminex platforms provide simultaneous testing for multiple serogroups using a single sample. Thus, less sample material is needed for this accurate and rapid fluorescent magnetic microbead testing format. These advantages are somewhat outweighed by the high cost of Luminex instrumentation and reagents. Laboratories with high throughput can justify this cost by calculating the total number of assays performed in these multiplex assays. Future work will be necessary to keep multiplex assays up to date with evolving regulatory requirements, which may include additional serogroups.