Microbead-Based Immunoassay for Simultaneous Detection of Shiga Toxins and Isolation of Escherichia Coli O157 in Foods

373

Journal of Food Protection, Vol. 74, No. 3, 2011, Pages 373–379 doi:10.4315/0362-028X.JFP-10-344

Microbead-Based Immunoassay for Simultaneous Detection of Shiga Toxins and Isolation of Escherichia coli O157 in Foods

LAURIE M. CLOTILDE,1 CLAY BERNARD IV,1 GARY L. HARTMAN,2 DAVID K. LAU,2 AND J. MARK CARTER1*

1Western Regional Research Center, Agricultural Research Service, U.S. Department of Agriculture, Albany, California 94710; and 2

U.S. Food and Drug Administration, Alameda, California 94502, USA Downloaded from http://meridian.allenpress.com/jfp/article-pdf/74/3/373/1685091/0362-028x_jfp-10-344.pdf by guest on 30 September 2021

MS 10-344: Received 9 August 2010/Accepted 12 November 2010

ABSTRACT Shiga toxin–producing Escherichia coli (STEC) is a significant foodborne pathogen with great economic consequences. There has been an increased food safety concern with this organism since outbreaks of human illnesses caused by this pathogen were first reported in 1982. Therefore, developing a reliable, sensitive, and rapid assay capable of detecting E. coli O157 and the main toxins produced by STEC (i.e., Shiga toxins 1 [Stx1] and 2 [Stx2]) will directly benefit regulatory agencies by minimizing analysis time. Here, we use Luminex technology to detect multiple analytes in a single 50-ml sample. Using commercially available monoclonal antibodies coupled to carboxylated magnetic microbeads, we developed an immunoassay capable of simultaneously serotyping E. coli O157 and detecting Stx1 and/or Stx2. The specificity and sensitivity of this immunoassay was tested against a collection of 34 E. coli isolates belonging to various O serogroups phenotypically different for Stx. The results were compared with microplate sandwich enzyme-linked immunosorbent assay (ELISA), and no cross-reactivity was observed for any of the monoclonal antibodies used. An increased sensitivity up to 1,000 times was observed in the microbead-based immunoassay when compared with the microplate sandwich ELISA. The results indicate that Luminex technology has the potential to simultaneously detect multiple targets without loss of specificity and/or sensitivity. A blind experiment was conducted with 48 samples of ground beef, lettuce, and milk spiked with #2 CFU/g E. coli. All the samples were correctly identified, with no false positives or false negatives. This microbead-based immunoassay could be extended to simultaneously detect additional foodborne pathogens and their toxic markers.

Shiga toxin–producing Escherichia coli (STEC) com- In recent years, research has turned toward molecular prises significant foodborne pathogens with high economic technologies to meet the need for rapid and multiplexed impact. In the United States, recent annual cost estimates for detection. For example, two-dimensional microarrays have acute care ranged from $1 to $2 billion, based on the the ability to simultaneously detect hundreds of specific assumption that E. coli O157 caused 73,000 illnesses and 61 probes, but suspension microbead arrays such as Luminex deaths, and non-O157 STEC caused more than 36,000 can offer more flexibility and cost-effectiveness (16). illnesses (34, 41). Although beef products, especially in the Briefly, Luminex technology is a multiplexed microbead- ground form, remain the main source of infection (47), other based sensing system capable of analyzing (theoretically) up documented sources include green leafy vegetables (8, 9), to 100 different binding reactions per sample (14, 15). This and dairy products (10, 11) contaminated with STEC- technology utilizes microbeads internally stained with infected cattle feces (48). The pathogenesis of human different ratios of red and infrared fluorophores to create infections with STEC depends on the production of AB5 100 unique spectral addresses (27). Assays using this toxins called Shiga toxins (Stx1 and Stx2) (36, 39). STEC (5, technology can be designed in a sandwich format, wherein 32, 52) have often been linked to outbreaks of human antigens are captured by monoclonal antibodies (mAb) that illnesses, and most of these are traced to consumption of E. are covalently immobilized on the surface of the micro- coli O157–contaminated foods (7, 11, 51). Currently, there beads. The bound analytes are detected with secondary mAb are no commercially available kits capable of detecting E. labeled with biotin and a phycoerythrin reporter. The coli O157 and its two main virulence factors (i.e., Stx1 and Luminex analyzer instrument is then used to quantify the Stx2) simultaneously in foods. Thus, this study focused on amount of bound analyte and to qualify the mixture of this specific STEC serotype as well as the main virulence unique microbeads, allowing multiplexed interrogation. The factors. specificity of the mAb used allows differentiation of closely related analytes, as evidenced earlier by the serotyping of bacteria (2) and viruses (24). A microbead-based immuno- * Author for correspondence. Tel: 510-559-6053; Fax: 510-559-6429; assay presents several advantages (26, 37, 38), which E-mail: [email protected]. include (i) multiplex format, (ii) reduced cost of analysis as 374 CLOTILDE ET AL. J. Food Prot., Vol. 74, No. 3 the degree of multiplexing increases, (iii) rapid data TABLE 1. Shiga toxin–producing Escherichia colia acquisition, and (iv) excellent sensitivity and specificity. b b Strain no. Serotype Stx1 Stx2 In addition, the large surface area of the microbeads allows coupling of up to 100,000 capture mAb per microbead, thus 1 O26:H11 22 allowing high density of capture mAb, with increased 2 O26 22 antigen binding and enhanced sensitivity of the immuno- 3 O26 z 2 z assay (33). 4 O45:H2 2 5 O45:H61 22 In this article, we describe a microbead-based immu- 6 O45 z 2 noassay for the simultaneous detection of E. coli O157, 7 O45 z 2 Stx1, and/or Stx2. We first compared the specificity and 8 O45 2 z sensitivity of Luminex technology with that of the standard 9 O91 z 2 sandwich enzyme-linked immunosorbent assay (ELISA). 10 O91 2 z Then, the test procedure was evaluated with ground beef, 11 O91 22 Downloaded from http://meridian.allenpress.com/jfp/article-pdf/74/3/373/1685091/0362-028x_jfp-10-344.pdf by guest on 30 September 2021 lettuce, and milk samples spiked with individual serogroups 12 O103:H8 22 13 O103:H38 22 of E. coli expressing various combinations of Stx1 and Stx2. 14 O103 z 2 MATERIALS AND METHODS 15 O111:H8 z 2 16 O111:H21 22 mAb and analytes. The mAb used were anti-Stx1 (Sifin 17 O111:H21 22 GmbH, Berlin, Germany), anti-Stx2 (Sifin GmbH), and anti–E. coli 18 O121:H19 2 z O157 lipopolysaccharide (LPS; OEM Concepts, Saco, ME). When 19 O121:NMc 2 z we prescreened via capture ELISA, we identified only one good- 20 O121:NM 2 z quality mAb against Stx1, which happened to be specific for the B 21 O121 2 z subunit (data not shown). As the tertiary structure of Stx protein 22 O121 22 includes five copies of the B subunit, the anti-Stx1B mAb works for 23 O128:H2 22 both capture and detection. This same strategy applies to the anti– 24 O128:H7 22 E. coli O157 LPS mAb. Even after partial lysis by polymyxin, 25 O128:H21 22 bacteria exhibit many copies of surface-accessible LPS, so the 26 O145:H28 22 same mAb can be used for both capture and detection. For Stx2, 27 O145:NM 2 z two different mAb were used, one specific to the A subunit for 28 O145 2 z detection and one specific to the B subunit for capture. The anti- 29 O157:H7 22 Stx2 mAb used had been previously tested against Stx2 and 30 O157:H7 2 z its variants (55). The results showed it recognized Stx2c, Stx2d, 31 O157:H7 zz Stx2d-activatable, and Stx2e, but not Stx2f, which is only rarely 32 O157:H7 zz associated with human disease (18, 45). Detector mAb were 33 O157:H7 zz biotinylated with the EZ-Link Sulfo-NHS-Long Chain-Biotin kit 34 O157:H7 zz (Pierce, Rockford, IL), according to the manufacturer’s instruc- a tions. With regard to analytes, #2 CFU/ml or g of each E. coli Drs. Robert Mandrell and Beatriz Quinõnes (USDA-ARS, strain tested were grown in 50 ml of brain heart infusion broth Produce Safety Microbiology Research Unit, Albany, CA) (BHIB; Oxoid, Ltd., Basingstoke, UK) or spiked into 50 g of food. provided these E. coli strains. Unless otherwise specified, only Samples were then incubated overnight (15 h) in a shaking the serogroup of those strains were identified by serological identification. incubator (100 rpm) at 37uC. After enrichment, the samples were b serially diluted to determine the sensitivity of the immunoassay. The Stx genotypes were previously determined by PCR (data not Bacterial counts for each sample were established by plate count shown) by Drs. Robert Mandrell and Beatriz Quinõnes, and the on tryptic soy agar (Oxoid, Ltd.). The 50-ml samples were then phenotypes were assessed by a verocytotoxicity assay (data treated with 2.5 ml of 50,000 U/ml polymyxin B?sulfate (Axxora, shown in Table 1). c The strain was not motile for the H antigen, i.e. ‘‘NM.’’ L.L.C., San Diego, CA) for 30 min in a shaking incubator (100 rpm) at 37uC. Highly concentrated (i.e., undiluted) samples were used to establish the point at which binding sites on the 0.1% Tween 20 (PBS-T) was used for washes. Reagent-grade microbeads were fully saturated and to explore immunoassay chemicals (i.e., 2,29-azino-bis(3-ethylbenzthiazoline-6-sulfonic ac- cross-reactivity and nonspecific binding. Samples were tested by id) [ABTS], H2O2, KCl, Na3 citrate, NaCl, and Tween 20) were both conventional microplate sandwich ELISA and by microbead- purchased from Sigma Chemical Co. (St Louis, MO), immuno- based immunoassay. All analytes used are listed in Table 1 and globulin (Ig) G–free, protease-free bovine serum albumin from were obtained from Drs. Robert Mandrell and Beatriz Quinõnes Jackson ImmunoResearch (West Grove, PA), streptavidin labeled (U.S. Department of Agriculture, Agricultural Research Service, with R-phycoerythrin from Invitrogen (Carlsbad, CA), and Produce Safety Microbiology Research Unit, Albany, CA). streptavidin–horseradish peroxidase conjugate from Zymed (San Appropriate biosafety level 2 precautions were used when handling Francisco, CA). pathogenic E. coli organisms. Covalent coupling of mAb to carboxylated magnetic Buffers and reagents. The reagents used in both microplate microbeads. Three different sets of spectrally unique carboxylated sandwich ELISA and microbead-based immunoassay were pre- magnetic microbeads (6.5 mm in diameter; Luminex Corp., Austin, pared in phosphate-buffered saline (PBS) with a pH of 7.4, 0.02% TX) were coated with capture mAb. For Stx1 (bead region no. 33) Tween 20, and 1% bovine serum albumin (PBS-TB). PBS and and Stx2 (bead region no. 34), mAb was coupled at 5 mg/ml, and J. Food Prot., Vol. 74, No. 3 SHIGA TOXINS AND E. COLI O157 375 for E. coli O157 LPS (bead region no. 35), mAb was coupled at 10 mg/ml, according to the manufacturer’s instructions provided with the Bio-Plex Amine Coupling Kit (Bio-Rad, Hercules, CA). The amount of mAb coupled to the microbeads was optimized, based on performance of the mAb in preliminary capture ELISA (data not shown), and calculated with the surface area of the microbeads and the surface area of a microplate well. Briefly, each mAb was covalently coupled to a unique set of microbeads (1.25 | 106 beads per 100-ml coupling reaction) in a two-step carbodiimide-coupling protocol that utilized 50 mg/ml N-hydroxy- sulfosuccinimide sodium salt (Pierce Chemicals, Rockford, IL) and 50 mg/ml 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydro- chloride (Pierce Chemicals). After each reaction, the concentration of the microbead suspension was determined by a hemacytometer, and the coupled microbeads stored at 2 to 8uC in the dark. Each Downloaded from http://meridian.allenpress.com/jfp/article-pdf/74/3/373/1685091/0362-028x_jfp-10-344.pdf by guest on 30 September 2021 coupling reaction was confirmed with phycoerythrin-labeled goat anti-mouse IgG antibodies (Invitrogen) at a concentration of 4 mg/ml.

Microbead-based immunoassay protocol. Microbead-based FIGURE 1. Summary of assay procedures. immunoassays were conducted in black, 96-well round bottom polystyrene microplates (Corning Costar, Lowell, MA), according and detection mAb concentrations were determined by serial to Luminex protocols (www.luminexcorp.com). All incubations dilutions. For statistical analysis, the LOD was defined as the were for 1 h at room temperature, protected from light, and shaken lowest concentration tested at which the average absorbance was at 600 rpm. In brief, a 100-ml aliquot of each test sample was higher than that of the absorbance of the negative control plus three combined with 5,000 microbeads of each of the three different standard deviations. specificities: Stx1, Stx2, and E. coli O157 LPS in a single microplate well. The microbeads were washed from unbound Blind food samples. Ground beef, romaine lettuce, and milk reaction components with the Bio-Plex Pro Wash Station (Bio- were purchased from a local supermarket. A total of 50 g of food Rad). Samples were washed three times with PBS-T. Microbeads sample was mixed with 450 ml of BHIB in a sterile Mason jar (Ball were then resuspended in 100 mlof4mg/ml biotinylated detector Corp., Broomfield, CO). The mixtures were blended for 2 min with mAb, and the resultant mixture was incubated. The mixture was an Oster 12-speed blender (Jarden Corp, Rye, NY) on the then washed three times as described earlier, resuspended in 100 ml ‘‘Liquefy’’ setting. Fifty-milliliter aliquots (ground beef [n ~ of 4 mg/ml streptavidin labeled with R-phycoerythrin, and the 18], lettuce [n ~ 14], or milk [n ~ 16]) were spiked with #2 resultant mixture was incubated, washed, and resuspended in 100 ml CFU/g of blinded E. coli strains (Table 1) in 50-ml conical tubes of PBS. The samples were analyzed by Bio-Plex Manager 5.0 (BD, Franklin Lakes, NJ). The samples were spiked after blending standard software (Bio-Rad) with the Luminex 100 flow analyzer. for biosafety reasons, to minimize aerosolization of live pathogens. Data were acquired for 120 s. For statistical analysis, the lower The blind food samples were enriched overnight (15 h) in BHIB limit of detection (LOD) was defined as the lowest concentration for milk, or in combination with 20 mg/liter novobiocin for lettuce, tested at which the average median fluorescent intensity (MFI) was or 20 mg/liter novobiocin, 2.5 mg/liter potassium tellurite, 0.05 mg/ higher than that of the MFI of the negative control plus three liter cefixime, and 40 mg/liter vancomycin for ground beef. The standard deviations. The background fluorescence (i.e., blank rationale for using these selective enrichments is explained in the samples) was measured on samples containing all assay reagents ‘‘Results and Discussion’’ section of this article. Samples were except the analyte(s) of interest. In fact, every microbead has an then incubated overnight at 37uC in a shaking incubator (100 rpm). inherent fluorescence because of the two fluorochromes embedded Each sample was treated with 2.5 ml of 50,000 U/ml polymyxin in each microbead. B?sulfate for 30 min at 37uC in a shaking incubator (100 rpm). Five thousand magnetic fluorescent microbeads labeled with the Sandwich ELISA assay protocol. One hundred microliters different capture mAb were mixed with each sample (52.5 ml), of a 1-mg/ml solution in PBS (pH 7.4) of capture mAb was added and the resultant mixture was incubated for 1 h at 37uCina to each well of a 96-well plate (Nunc, Rochester, NY) and shaking incubator (100 rpm). The microbeads were recovered from incubated overnight at 4uC. After removing the capture antibody the 52.5-ml mixture with a LifeSep 50 SX magnetic separator solution, the plates were then blocked with 200 ml of PBS-TB for (Dexter Magnetic Technologies, Elk Grove Village, IL), washed 1 h at room temperature. After three washes with PBS-T, 100 mlof twice with PBS-T, and resuspended in 1.5 ml of PBS-T. The analyte solutions was added to the plate and incubated for 1 h at microbeads were then concentrated with a Dynal DynaMag-2 room temperature. After three washes with PBS-T, 100 mlofa magnetic separator (Invitrogen), washed twice with PBS-T, and 0.5-mg/ml solution in PBS pH 7.4 of detector mAb was added to resuspended in 300 ml of PBS. The assay was then run in triplicate each well and incubated for 1 h at room temperature. After three as described above in the ‘‘Microbead-based immunoassay washes with PBS-T, 100 ml of a 1/5,000 dilution of streptavidin– protocol’’ paragraph in a 96-well microplate. A summary of the horseradish peroxidase conjugate in PBS-TB was added to each assay procedures is shown in Figure 1. well and incubated for 1 h at room temperature. After three washes with PBS-T, 100 ml of ABTS–horseradish peroxidase substrate RESULTS AND DISCUSSION solution was added to each well to detect bound, conjugated mAb. After 15 min of incubation at room temperature, absorbance was Specificity. The specificity of the mAb was tested in read at 405 nm with the Wallac Victor 2 Multilabel Counter both microplate sandwich ELISA and microbead-based (PerkinElmer, Inc., Waltham, MA). The optimal antigen, capture, immunoassay, using positive and negative controls for Stx1, 376 CLOTILDE ET AL. J. Food Prot., Vol. 74, No. 3

TABLE 2. Comparison of LOD for standard sandwich ELISA ‘‘Materials and Methods.’’ In a recent review by Hussein and microbead-based immunoassay and Bollinger (22), the use of BHIB and various antibiotics LOD (CFU/100 ml) was explained. In short, BHIB supports production of higher Stx levels, with an improved detection rate from 78 Microbead-based to 95% when enriched in BHIB instead of sorbitol Standard sandwich ELISA immunoassay MacConkey agar (4). Because commercial milk is pasteur- Analyte Intact Polymyxin Intact Polymyxin ized, minimum background bacteria are present; thus only BHIB was used, as recommended by others (17, 42). 7 7 5 4 Stx1 10 10 10 10 Lettuce usually contains slightly higher numbers of 6 5 5 5 Stx2 10 10 10 10 background bacteria; therefore, BHIB supplemented with 3 3 0 0 O157 LPS 10 10 10 10 20 mg/liter novobiocin was used. Novobiocin is commonly used to supplement enrichment broths for STEC (50), as STEC are known to survive in the presence of this antibiotic

Stx2, and O157 LPS, as presented in Table 1. None of the Downloaded from http://meridian.allenpress.com/jfp/article-pdf/74/3/373/1685091/0362-028x_jfp-10-344.pdf by guest on 30 September 2021 Stx mAb tested cross-reacted with any of the 14 non–Stx- (49). In fact, novobiocin is toxic to most gram-positive bacteria and has poor activity against most gram-negative producing E. coli, nor did Stx1 mAb cross-react with STEC bacteria (29, 40). Inclusion of novobiocin in enrichment that produced only Stx2, or vice versa. Likewise, E. coli O157 LPS mAb did not cross-react with any of the 28 non- media at 20 mg/liter is recommended for detection of E. coli O157 E. coli strains. Non-O157 STEC that we tested O157 (1) and other STEC (6, 21) from foods. Finally, raw belonged to the O26, O45, O91, O103, O111, O121, O128, ground beef contains many background bacteria; thus, and O145 serogroups, which make up 74.2% of all non- BHIB supplemented with four antibiotics was used. O157 STEC-related outbreaks of human illnesses in the Antibiotics such as the ones used in this study (tellurite, United States (10). cefixime, novobiocin, and vancomycin) are often added alone or in combination to enrichment media to enhance Sensitivity. Our multiplex assay is constructed around STEC detection by inhibiting growth of background antibody-based detection of the bacterial biomarker antigens bacteria. In our previous study (23), both E. coli O157 O157 LPS, Stx1, and Stx2. But to make it more practically and non-O157 STEC were able to survive and grow in useful for food safety testing, we validated the assay in food BHIB containing novobiocin (20 mg/liter) as well as three by using live bacteria rather than purified antigens. Thus, we other antibiotics (tellurite, cefixime, and vancomycin at 2.5, report the assay’s sensitivity in terms of CFU, rather than 0.05, and 40 mg/liter, respectively). In this study, nanograms of LPS or Stx. As described in ‘‘Materials and preparation of spiked samples played an important role Methods,’’ the LOD was determined with serial dilutions of in the detection signal; ergo, the detection signal an overnight culture of E. coli O157:H7 (strain no. 34) that was optimized by supplementing the enrichment broth had been grown in 50 ml of BHIB. The LOD for E. coli in with various antibiotics to inhibit background bacteria. the three component assays (based on detection of Stx1, Yokoyama et al. (53) demonstrated that treating E. coli Stx2, or O157 LPS) were established as the number of with polymyxin enhanced the release of Stx proteins. bacteria at which the MFI was equal to the background Eighteen ground beef, 14 lettuce, and 16 milk samples intensity plus three standard deviations (Table 2). When were blindly spiked with #2 CFU/g randomly selected E. detecting the O157 LPS analyte, the LOD was 1 CFU/100 ml coli strains (Table 1) and enriched overnight (15 h). The E. coli in the microbead-based immunoassay, while the first sample (blank) shows the baseline for each bead type. 5 4 LOD for Stx1 and Stx2 were 10 and 10 CFU (respectively) Detection of E. coli strains positive for Stx1, Stx2, and/or of E. coli per 100 ml. Compared with the standard O157 LPS in ground beef, lettuce, and milk is shown on microplate sandwich ELISA, our microbead-based immu- Figures 2, 3, and 4. In the 18 ground beef samples tested noassay was 100, 10, and 1,000 times more sensitive for E. (Fig. 2), 9 (blank and nos. 1, 11, 12, 13, 16, 17, 23, and 26) coli detection, based on Stx1, Stx2, and O157 LPS, were negative for all three analytes, 3 (nos. 3, 9, and 14) respectively. When using E. coli cells treated with were positive for Stx1 only, 2 (nos. 10 and 28) were positive polymyxin, the microbead-based immunoassay was 1,000, for Stx2 only, and 4 (nos. 31, 32, 33, and 34) were positive equivalent, and 1,000 times more sensitive for Stx1, Stx2, for all three analytes. In the 14 lettuce samples tested and O157 LPS, respectively, than was the standard (Fig. 3), 7 (blank and nos. 2, 11, 12, 17, 24, and 26) were sandwich ELISA. Although different strains of STEC (and negative for all three analytes, 3 (nos. 3, 9, and 14) were even different strains of O157:H7) are likely to exhibit positive for Stx1 only, 1 (no. 27) was positive for Stx2 only, somewhat different levels of Stx expression, and thus and 3 (nos. 31, 32, and 34) were positive for all three different LOD in our assay, Luminex technology clearly analytes. In the 16 milk samples tested (Fig. 4), 9 (blank and provides enhanced sensitivity over the microplate format. nos. 1, 2, 5, 11, 13, 16, 22, and 24) were negative for all three analytes, 1 (no. 9) was positive for Stx1 only, 5 Blind food samples. The performance of the mi- (nos. 10, 18, 19, 21, and 27) were positive for Stx2 only, and crobead-based immunoassay was evaluated with ground 1 (no. 33) was positive for all three analytes. These beef, lettuce, and milk samples spiked with #2 CFU/g E. phenotypic characteristics were consistent with the ones coli. All food samples were diluted in BHIB (1:10 ratio) listed in Table 1. Of the 48 blind samples analyzed, all were supplemented or not with antibiotics, as described in correctly identified and yielded no false positives and/or J. Food Prot., Vol. 74, No. 3 SHIGA TOXINS AND E. COLI O157 377

FIGURE 4. A triplex microbead-based immunoassay to detect Downloaded from http://meridian.allenpress.com/jfp/article-pdf/74/3/373/1685091/0362-028x_jfp-10-344.pdf by guest on 30 September 2021 FIGURE 2. A triplex microbead-based immunoassay to detect Stx1, Stx2, and O157 LPS in milk spiked with #2 CFU/g. The Stx1, Stx2, and O157 LPS in ground beef spiked with #2 CFU/g. relative levels of Stx1, Stx2, and O157 LPS were calculated by The relative levels of Stx1, Stx2, and O157 LPS were calculated by assuming the highest MFI signal as 100%. assuming the highest MFI signal as 100%. false negatives. Based on these results, the microbead-based inexpensive, but quite time-consuming (17). PCR is faster immunoassay appears to be a good tool for screening and can detect the presence of genes encoding Stx1, Stx2, samples from the food supply chain. Moreover, because the and E. coli O157 (17) in mixed cultures. However, although magnetic microbeads concentrated and isolated E. coli PCR provides results within 3 h, it does not indicate whether O157, microplate wells containing these organisms (sample the detected Stx genes are actually expressed. Furthermore, nos. 29 through 34) could be picked and streaked on solid PCR methods require considerable sample preparation prior agar to obtain an isolate for further characterization (data not to analysis, reagents can be expensive, and multiplexing of shown). This feature is very attractive to regulatory PCR based assays is challenging to develop and optimize. agencies, as an isolated E. coli organism is required prior Assays based on antibodies offer some advantages over to taking legal action and issuing a mandatory recall of the PCR and culture. Immunochemical methods provide contaminated product. excellent sensitivity and specificity. They are also generally E. coli O157 is known to cause most of the STEC more robust to matrix effects, compared with assays based outbreaks of human illnesses in the United States (7), while on nucleic acid detection. More importantly, immunochem- various non-O157 E. coli strains producing Stx1 and/or Stx2 ical assays are readily adaptable to field-deployable devices are also known to be pathogenic to humans (25, 52). Thus, a and advanced assay platforms. For example, in addition to critical need exists for a multiplexed immunoassay for these Luminex (33), antibodies have been successfully applied in a fiber optic fluorometer (28), a capillary-based flow three analytes (Stx1, Stx2, and O157 LPS). This study clearly demonstrates the effectiveness of a microbead immunosensor (35), lateral flow devices (54), and a immunoassay based on Luminex technology for simulta- compact biochip detection system (46). Luminex technology provides many advantages over neous detection of Stx1, Stx2, and E. coli O157 LPS in various food matrices. conventional sandwich ELISA. First, the Luminex measures Standard microbiological methods, such as culturing numerous beads per sample (i.e., typically 100), whereas for and streaking on selective agar plates, are simple and the conventional format, three microplate wells per sample are measured. The high number of replicates in the Luminex format provides higher precision than does the microplate format (13). Second, the Luminex format is much faster, achieving maximum sensitivity with a 3-h analysis time (with 1-h incubations for each step, as reported by McBride et al. (33)), while a conventional microplate sandwich ELISA requires about 16 h. Other detection methods (i.e., culture, biochemical testing, ELISA, and nucleic acid amplification) can be time-consuming and require several tests to detect Stx1, Stx2, and/or O157. Third, E. coli O157 and the main virulence factors of E. coli (Stx1 and Stx2) can be detected simultaneously from the same sample without loss in sensitivity. This reduces sample preparation time and requires less sample material, and fewer reagents, and FIGURE 3. A triplex microbead-based immunoassay to detect consumable supplies (44). Fourth, Luminex technology is Stx1, Stx2, and O157 LPS in lettuce spiked with #2 CFU/g. The attractive to regulatory agencies. Indeed, The U.S. Food relative levels of Stx1, Stx2, and O157 LPS were calculated by Safety and Inspection Service, the U.S. Food and Drug assuming the highest MFI signal as 100%. Administration, the Centers for Disease Control and 378 CLOTILDE ET AL. J. Food Prot., Vol. 74, No. 3

Prevention, and many state health departments have already ACKNOWLEDGMENTS invested in this assay technology and instrumentation. Fifth, The work presented in this article was supported by the U.S. Luminex is one of the few multiplex devices allowing for Department of Agriculture, Agricultural Research Service (USDA-ARS), testing of complex food matrices while maintaining high National Program 108, under Current Research Information System 5325- degree of sensitivity (20, 30). Finally, we have found that by 42000-043-00. We thank Drs. Robert Mandrell and Beatriz Quin˜ones using the microbeads, a positive sample may be streaked (USDA-ARS, Produce Safety Microbiology Research Unit, Albany, CA) for providing STEC strains. We thank Drs. Larry Stanker, David Brandon, directly onto bacterial media to obtain an isolate for further and Robert Hnasko (USDA-ARS, Foodborne Contaminants Research Unit, characterization (data not shown). Albany CA) for technical advice. We thank Henry Lau, Lillian Hsu, Other studies (12, 33, 43) have supported the Myriam Lonhienne, Robert Gee, and You Jin Lee for technical assistance. advantages of Luminex technology for multiplex assay We thank the Bio-Rad BioPlex team for repairing our Luminex instrument. requirements. For example, McBride et al. (33) developed a multiplexed microbead-based immunoassay to simulta- REFERENCES neously detect simulants of biological warfare agents 1. AOAC International. 2000. Official methods of analysis, 17th ed. Downloaded from http://meridian.allenpress.com/jfp/article-pdf/74/3/373/1685091/0362-028x_jfp-10-344.pdf by guest on 30 September 2021 (Bacillus globigii, Erwinia herbicola, ovalbumin, and AOAC International, Gaithersburg, MD. bacteriophage MS2). Compared with conventional ELISA 2. Bash, M. C., F. Lynn, N. F. Concepcion, J. W. Tappero, G. M. data for the detection of antibodies to different foot-and- Carlone, and C. E. Frasch. 2000. Genetic and immunologic characterization of a novel serotype 4, 15 strain of Neisseria mouth disease nonstructural proteins, microbead-based meningitides. FEMS Immunol. Microbiol. 29:169–176. assays were shown to be a more sensitive and efficient 3. Bergervoet, J. H. W., J. Peters, J. R. C. M. van Beckhoven, G. W. van method, providing both time and cost savings to the den Bovenkamp, J. W. Jacobson, and J. M. van der Wolf. 2008. laboratory (12). Compared with a glass-slide antibody Multiplex microsphere immuno-detection of potato virus Y, X, and microarray, the Luminex microbead array format also PLRV. J. Virol. Methods 149:63–68. provided an improved dynamic range and LOD for the 4. Beutin, L., S. Zimmermann, and K. Gleier. 2002. Evaluation of the VTEC-Screen ‘‘Seiken’’ test for detection of different types of Shiga detection of bacterial and viral proteins (43). toxin (verotoxin)–producing Escherichia coli (STEC) in human stool Food matrices play a major role in assay performance samples. Diagn. Microbiol. Infect. Dis. 42:1–8. (19, 31). In fact, food components may mask or even 5. Bonnet, R., B. Souweine, G. Gauthier, C. Rich, V. Livrelli, expose antibody-binding epitopes, therefore decreasing or J. Sirot, B. Joly, and C. Forestier. 1998. Non-O157:H7 Stx2- increasing the detection of the analyte. These food producing Escherichia coli strains associated with sporadic cases of hemolytic uremic syndrome in adults. J. Clin. Microbiol. 36:1777– components can also nonspecifically bind to an antibody, 1780. leading to false elevated signals. Our utilization of 6. Brooks, H. J., B. D. Mollison, K. A. Bettelheim, K. Matejka, K. A. magnetic microbeads facilitates separation from complex Paterson, and V. K. Ward. 2001. Occurrence and virulence factors of food matrices and allows for automated washing. Others non-O157 Shiga toxin–producing Escherichia coli in retail meat in showed that the use of magnetic microbeads instead of the Dunedin, New Zealand. Lett. Appl. Microbiol. 32:118–122. older, nonmagnetic variety significantly improves signal- 7. Centers for Disease Control and Prevention. 2003. E. coli O157 outbreak summaries. Available at: http://www.cdc.gov/foodborne to-noise ratio (3). Overall, there was no nonspecific outbreaks/ecoli_sum.htm. Accessed 4 August 2004. binding observed for any of the analytes tested, indicating 8. Centers for Disease Control and Prevention. 2006. Update on multi- a high degree of mAb specificity. In addition, the state outbreak of E. coli O157:H7 infections from fresh spinach, sensitivity of the microbead-based immunoassay was October 6, 2006. Available at: http://www.cdc.gov/foodborne/ greater than the sensitivity of conventional sandwich ecolispinach/100606.htm. Accessed 21 May 2010. ELISA was. This is due to an overall greater signal-to- 9. Centers for Disease Control and Prevention. 2006. Ongoing multistate outbreak of Escherichia coli serotype O157:H7 infections noise ratio, as well as the capability of the microbead associated with consumption of fresh spinach—United States, suspension to query the entire 50-ml sample volume, September 2006. Morb. Mortal. Wkly. Rep. 55:1045–1046. compared with only 100 ml for the microplate. 10. Centers for Disease Control and Prevention. 2006. Bacterial Our newly developed microbead-based immunoassay is foodborne and diarrheal disease national case surveillance. Annual capable of detecting Stx , Stx , and E. coli O157 LPS report, 2005. Available at: http://www.cdc.gov/foodborneoutbreaks/. 1 2 Accessed 2 April 2009. simultaneously, not only in growth media (BHIB), but also 11. Centers for Disease Control and Prevention. 2009. Surveillance for within complex food matrices such as ground beef, lettuce, foodborne disease outbreaks—United States, 2006. Morb. Mortal. and milk. With the ability of Luminex technology to detect up Wkly. Rep. 58:609–615. to 100 analytes in a single well, antibodies specific to 12. Clavijo, A., K. Hole, M. Li, and B. Collignon. 2006. Simultaneous additional foodborne pathogens (e.g., non-O157 STEC, detection of antibodies to foot-and-mouth disease non-structural Listeria monocytogenes, and Salmonella enterica serovar proteins 3ABC, 3D, 3A, and 3B by a multiplexed Luminex assay to differentiate infected from vaccinated cattle. Vaccine 24:1693–1704. Typhimurium) and their corresponding toxins and virulence 13. Dasso, J., J. Lee, H. Bach, and R. G. Mage. 2002. A comparison of proteins could be easily added to our current immunoassay. ELISA and flow microsphere-based assays for quantification of We have begun work on such an expansion of this project and immunoglobulins. J. Immunol. Methods 263:23–33. look forward to introducing a Luminex tool for screening 14. Diaz, M. R., T. Boekhout, B. Theelen, M. Bovers, F. J. Caban˜es, and food as well as clinical samples. As a further improvement to J. W. Fell. 2006. Microcoding and flow cytometry as a high- this technology we will also validate our immunoassay in the throughput fungal identification system for Malassezia species. J. Med. Microbiol. 55:1197–1209. new mobile Luminex platform (MagPix), thus realizing the 15. Diaz, M. R., and J. W. Fell. 2005. Use of a suspension array for rapid crucial migration from a bench-research top application to a identification of the varieties and genotypes of the Cryptococcus field-deployable commercial system. neoformans species complex. J. Clin. Microbiol. 43:3662–3672. J. Food Prot., Vol. 74, No. 3 SHIGA TOXINS AND E. COLI O157 379

16. Dunbar, S. A. 2006. Applications of Luminex xMAPTM technology 37. Nolan, J. P., and F. F. Mandy. 2001. Suspension array technology: for rapid, high-throughput multiplexed nucleic acid detection. Clin. new tools for gene and protein analysis. Cell. Mol. Biol. 47:1241– Chim. Acta 363:71–82. 1256. 17. Feng, P., and S. D. Weagant. 2009. BAM: diarrheagenic Escherichia 38. Nolan, J. P., and L. A. Sklar. 2002. Suspension array technology: coli. Available at: http://www.fda.gov/Food/ScienceResearch/Laboratory evolution of the flat-array paradigm. Trends Biotechnol. 20:9–12. Methods/BacteriologicalAnalyticalManualBAM/UCM070080. Accessed 39. O’Brien, A. D., and G. D. LaVeck. 1983. Purification and 24 June 2010. characterization of a Shigella dysenteriae 1–like toxin produced by 18. Friedrich, A. W., M. Bielaszewska, W. Zhang, M. Pulz, T. Kuczius, Escherichia coli. Infect. Immun. 40:675–683. A. Ammon, and H. Karch. 2002. Escherichia coli harboring Shiga 40. Pao, S., D. Patel, A. Kalantari, J. P. Tritschler, S. Wildeus, and B. L. toxin 2 gene variants: frequency and association with clinical Sayre. 2005. Detection of Salmonella strains and Escherichia coli symptoms. J. Infect. Dis. 185:74–84. O157:H7 in feces of small ruminants and their isolation with various 19. Garber, E. A. E., R. M. Eppley, M. E. Stack, M. A. McLaughlin, and media. Appl. Environ. Microbiol. 71:2158–2161. D. L. Park. 2005. Feasibility of immunodiagnostic devices for the 41. Paton, J. C., and A. W. Paton. 2000. Shiga toxigenic Escherichia coli detection of ricin, amanitin, and T-2 toxin in food. J. Food Prot. 68: infections. Sci. Med. May/June:28–37. 1294–1301. 42. Rantsiou, K., and L. Cocolin. 2008. Detection and quantification of

20. Gooding, J. J. 2006. Biosensor technology for detecting biological Shiga-like toxin–producing Escherichia coli and determination of Downloaded from http://meridian.allenpress.com/jfp/article-pdf/74/3/373/1685091/0362-028x_jfp-10-344.pdf by guest on 30 September 2021 warfare agents: recent progress and future trends. Anal. Chim. Acta their vitality in food. Abstract presented at Food Micro 2008 559:137–151. Conference, Aberdeen, Scotland, 1 to 4 September 2008. 21. Hara-Kudo, Y., H. Konuma, H. Nakagawa, and S. Kumagai. 2000. 43. Rao, R. S., S. R. Visuri, M. T. McBride, J. S. Albala, D. L. Matthews, Escherichia coli O26 detection from foods using an enrichment and M. A. Coleman. 2004. Comparison of multiplexed techniques for procedure and an immunomagnetic separation method. Lett. Appl. detection of bacterial and viral proteins. J. Proteome Res. 3:736–742. Microbiol. 30:151–154. 44. Seidel, M., and R. Niessner. 2008. Automated analytical microarrays: 22. Hussein, H. S., and L. M. Bollinger. 2008. Influence of selective a critical review. Anal. Bioanal. Chem. 391:1521–1544. media on successful detection of Shiga toxin–producing Escherichia 45. Sonntag, A. K., E. Zenner, H. Karch, and M. Bielaszewska. 2005.

coli in food, fecal, and environmental samples. Foodborne Pathog. Pigeons as a possible reservoir of Shiga toxin2f–producing Esche- Dis. 5:227–244. richia coli pathogenic to humans. Berl. Muench. Tieraerztl. 23. Hussein, H. S., L. M. Bollinger, and M. R. Hall. 2008. Growth and Wochenschr. 118:464–470. enrichment medium for detection and isolation of Shiga toxin– 46. Stratis-Cullum, D. N., G. D. Griffin, J. Mobley, A. A. Vass, and T. V. producing Escherichia coli in cattle feces. J. Food Prot. 71:927–933. Dinh. 2003. A miniature biochip system for detection of aerosolized 24. Jensen, T. H., A. Jensen, A. M. Szilvay, and J. Kjems. 1997. Probing Bacillus globigii spores. Anal. Chem. 75:275–280. the structure of HIV-1 rev by protein footprinting of multiple 47. U.S. Department of Agriculture, Animal and Plant Health Inspection monoclonal antibody-binding sites. FEBS Lett. 414:50–54. Service, Veterinary Services. 1997. An update: Escherichia coli 25. Karmali, M. A., M. Petric, C. Lim, P. C. Fleming, G. S. Arbus, and O157:H7 in humans and cattle. U.S. Department of Agriculture, H. Lior. 1985. The association between idiopathic hemolytic-uremic Animal and Plant Health Inspection Service, Veterinary Service, syndrome and infection by verotoxin-producing Escherichia coli. J. Centers for Epidemiology and Animal Health, Fort Collins, CO. Infect. Dis. 151:775–782. 48. Valcour, J. E., P. Michel, S. A. McEwen, and J. B. Wilson. 2002. 26. Kellar, K. L. 2003. Applications of multiplexed microsphere-based Associations between indicators of livestock farming intensity and assays to studies of infectious disease. J. Clin. Ligand Assay 26:76– incidence of human Shiga toxin–producing Escherichia coli 86. infection. Emerg. Infect. Dis. 8:252–257. 27. Kettman, J. R., T. Davies, D. Chandler, K. G. Oliver, and R. J. 49. Vimont, A., M. L. Delignette-Muller, and C. Vernozy-Rozand. 2007. Fulton. 1998. Classification and properties of 64 multiplexed Supplementation of enrichment broths by novobiocin for detecting microsphere sets. Cytometry 33:234–243. Shiga toxin–producing Escherichia coli from food: a controversial 28. King, K. D., J. M. Vanniere, J. L. Leblanc, K. E. Bullock, and G. P. use. Lett. Appl. Microbiol. 44:326–331. Anderson. 2000. Automated fiber optic biosensor for multiplexed 50. Vimont, A., C. Vernozy-Rozand, M. P. Montet, C. Lazizzera, C. immunoassays. Environ. Sci. Technol. 34:2845–2850. Bavai, and M. L. Delignette-Muller. 2006. Modeling and predicting 29. Kirby, W. M., D. G. Hudson, and W. D. Noyes. 1956. Clinical and the simultaneous growth of Escherichia coli O157:H7 and ground laboratory studies of novobiocin, a new antibiotic. AMA Arch. Intern. beef background microflora for various enrichment protocols. Appl. Med. 98:1–7. Environ. Microbiol. 72:261–268. 30. Ligler, F. S., K. E. Sapsford, J. P. Golden, L. C. Shriver-lake, C. R. 51. Wendel, A. M., D. H. Johnson, U. Sharapov, J. Grant, J. R. Archer, T. Taitt, M. A. Dyer, S. Barone, and C. J. Myatt. 2007. The array Monson, C. Koschmann, and J. P. Davis. 2009. Multistate outbreak biosensor: portable, automated systems. Anal. Sci. 23:5–10. of Escherichia coli O157:H7 infection associated with consumption 31. Lindstrom, M., and H. Korkeala. 2006. Laboratory diagnostics of of packaged spinach, August–September 2006: the Wisconsin botulism. Clin. Microbiol. Rev. 19:298–314. investigation. Clin. Infect. Dis. 48:1079–1086. 32. Lior, H. 1994. Classification of Escherichia coli, p. 31–72. In C. L. 52. Willshaw, G. A., S. M. Scotland, and B. Rowe. 1997. Vero- Gyles (ed.), Escherichia coli in domestic animals and humans. CAB cytotoxin–producing Escherichia coli. In M. Sussman (ed.), Esche- International, Wallingford, UK. richia coli: mechanisms of virulence. Cambridge University Press, 33. McBride, M., S. Gammon, M. Pitesky, T. W. O’Brien, T. Smith, J. Cambridge. Aldrich, R. G. Langlois, B. Colston, and K. S. Venkateswaran. 2003. 53. Yokoyama, K., T. Horii., T. Yamashino, S. Hashikawa, S. Barua, T. Multiplexed liquid arrays for simultaneous detection of simulants of Hasegawa, H. Watanabe, and M. Ohta. 2000. Production of Shiga biological warfare agents. Anal. Chem. 75:1924–1930. toxin by Escherichia coli measured with reference to the membrane 34. Mead, P. S., L. Slutsker, V. Dietz, L. F. McCaig, J. S. Bresee, C. vesicle-associated toxins. FEMS Microbiol. Lett. 192:139–144. Shapiro, P. M. Griffin, and R. V. Tauxe. 1999. Food-related illness 54. Zhang, M. Z., M. Z. Wang, Z. L. Chen, J. H. Fang, M. M. Fang, J. and death in the United States. Emerg. Infect. Dis. 5:607–625. Liu, and X. P. Yu. 2009. Development of a colloidal gold-based 35. Narang, U., P. R. Gauger, A. W. Kusterbeck, and F. S. Ligler. 1998. lateral-flow immunoassay for the rapid simultaneous detection of Multianalyte detection using a capillary-based flow immunosensor. clenbuterol and ractopamine in swine urine. Anal. Bioanal. Chem. Anal. Biochem. 255:13–19. 395:2591–2599. 36. Noda, M., T. Yutsudo, N. Nakabayashi, T. Hirayama, and Y. Takeda. 55. Zhang, W., M. Bielaszewska, M. Pulz, K. Becker, A. W. Friedrich, 1987. Purification and some properties of Shiga-like toxin from H. Karch, and T. Kuczius. 2008. New immuno-PCR assay for Escherichia coli O157:H7 that is immunologically identical to Shiga detection of low concentrations of Shiga toxin 2 and its variants. J. toxin. Microb. Pathog. 2:339–349. Clin. Microbiol. 46:1292–1297.