Model GLH-01 homogenizer with a 10 mm × 115 mm, saw tooth generator probe (Omni International, Kennesaw, GA, USA) was generally used at ca. 20,000 rpm (#8 setting of Omni GLH external Speed Control SC115). Initial studies utilized the lighter duty Omni model TH-01 homogenizer, but this model proved unable to maintain speed with some samples.

Ground beef marked “90% lean” was purchased at a local supermarket and used within 24 h of purchase. Samples were kept on ice during all procedures prior to application of sample to assay wells. Four-gram samples were weighed into 50 mL polypropylene conical centrifuge tubes and spiked with a small volume (generally 8 µL) of ricin solution. Each sample was thoroughly mixed using a plastic spatula, and then 8 mL of extraction buffer were added (phosphate-buffered saline [PBS]-100 mM galactose). Samples were homogenized for 30 s at 20,000 rpm, and pieces of beef were then dislodged from the homogenizer probe and returned to the homogenate using a spatula. The sample was homogenized for an additional 30 s at the same speed. Between samples, the probe was cleaned by two washes with water at 30,000 rpm. Deionized water was used for all washes and buffers in this study.

Ricin and RCA-1 were obtained from Vector Laboratories (Burlingame, CA, USA). For preparation of crude ricin (CR), castor seeds were weighed and ground thoroughly in mortar and pestle, in PBS (10 mL/g). The homogenate was centrifuged at 3000 g for 5 min in a fixed angle rotor, and the aqueous supernatant, below the oil layer, was collected. This procedure was repeated 3 additional times. The protein content was determined by assay with bicinchoninic acid [24], and ricin and RCA-1 were estimated as 2.4 mg/mL and 3.0 mg/mL, respectively, by ELISA [11]. The extract was diluted to 1 mg/mL ricin and used to spike ground beef samples.

Colorimetric ELISAs were performed on Immulon^® 4HBX plates (Dynex, Chantilly, VA, USA), coated as described previously [11]. Briefly, wells were coated with proteins at 5 μg/mL in PBS, excess “sticky” sites were blocked with 10 mg/mL BSA in PBS-0.05% Tween^®-20 (BPT). Coated, rinsed plates were treated with 2% sucrose, then dried at 37 °C and stored desiccated at 4 °C for up to 6 months. For ECL assays, 96-well standard, uncoated plates were obtained from Meso Scale Discovery ([MSD], Gaithersburg, MD, USA; Cat. No. L15XA-3). Details for antibody coating are given below.

Monoclonal antibodies were prepared using isolated ricin A and B chains as immunogens in BALB/c mice. MAbs were purified, characterized, and biotinylated, as described previously [11]. Tris(2,2’-bipyridyl)ruthenium(II) (Ru[bpy]₃)-conjugates of antibodies were prepared using the N-hydroxysuccinimide ester (MSD, Cat. No. R91AN-1) and spin columns for buffer exchange and conjugate purification, per manufacturer’s protocol (Sulfo-Tag^® labeling kit, MSD, Cat. No. R91CN-1). MAbs are designated by the corresponding hybridoma clone numbers, prefixed by designation of the conjugate, for example b-1795 for biotinylated mAb (b-mAb) 1795 and Ru-1443 for Ru(bpy)₃-conjugated mAb 1443.

Assay wells contained 100 μL of standards, controls, or samples, using BPT containing 100 mM galactose as diluent. Samples were generally assayed neat or as dilutions of 1:2, 1:5, or 1:10 by addition 100, 50, 20 or 10 μL of sample to assay wells containing 0, 50, 80, or 90 μL, respectively, of BPT-galactose. After samples and standards were applied, plates were sealed and incubated 1 h, with shaking. Wells were emptied and rinsed by manual pipetting of wastes into 0.5% sodium hypochlorite solution for inactivation of toxin, and wells were rinsed an addditional 4 times with water. Biotinylated detection antibody was then added (100 μL at 100 ng/mL in BPT containing 100 mM galactose, to minimize nonspecific binding and binding to agglutinins via their carbohydrate-binding sites). After incubation with shaking for 1 h, wells were washed 4 times with water. Horseradish peroxidase (HRP)-conjugated streptavidin (Invitrogen, South San Francisco, CA, USA) was applied (1:5000, 100 μL/well) and incubated 30 min, with shaking. Following water washes, the assay was developed by adding tetramethylbenzidine substrate solution (TMB, K-Blue, Neogen, Lexington, KY, USA), 100 μL/well. The reaction was stopped after 30 min by the addition of 100 µL/well 0.3 N HCl. Absorbance was read at 450 nm, with subtraction of the absorbance at 650 nm, using Model M2 plate reader using SoftMax^® Pro 5.3 software (Molecular Devices, Sunnyvale, CA, USA).

In these assays, all steps were conducted with a total volume of 30 μL, except for application of biotinylated mAbs which employed 50 μL. Dilutions were prepared as for the ELISA described above, with scaling for volume. ECL wells were washed 4 times between steps with PBS-0.05% Tween-20 (PBST). Incubations were 60 min for application of samples and standards, 30 min for other steps.

In sandwich assays employing biotinylated detection antibodies, the secondary reagent was streptavidin conjugated with Ru(II) tris-bipyridine 4-methylsulfonate (MSD Sulfo-Tag—Streptavidin, Cat. No. R32AD-5), 30 µL/well at 0.5 µg/mL diluted in BPT-100 mM galactose. For assays utilizing direct detection of analyte, Ru(bpy)₃-conjugated mAbs were used (50 μL/well, 0.2–1 μg/mL). Plates were again washed 4 times, as above, tapped to empty, then 150 µL of tripropylamine solution (MSD Read Buffer with Surfactant, Cat. No. R92TC-2, diluted 1:4 with water) were added. After dispersing any bubbles that formed during pipetting, plates were read immediately on a 2400 Sector 2400 Imager, with Discovery Workbench v3.0 software (MSD).

Data are expressed as mean ± standard deviation (sd, n = 3), unless otherwise indicated. Analyses of ground beef were done at least 5 times, with results shown from one typical experiment. Limit of detection (LOD) was computed as the analyte concentration at which the lower one-sided 95% confidence interval (CI) equaled the blank + 3 sd. Confidence and prediction intervals were computed using SlideWrite^® v6 (Advanced Graphics Software, Carlsbad, CA, USA). Assay curves were fitted to a 4-parameter logistic model using either SlideWrite or SoftMax Pro.

For directly coated ECL plates, a variety of conditions were tested. Figure 1 shows results obtained using 2 different antibody pairs. As expected, signal increased at higher coating concentrations and higher b-mAb concentrations. Luminescence was 300–400% greater with mAb coated at 1 μg/mL instead of 0.25 μg/mL, but increased less than 100% when the plate coating concentration was increased to 4 μg/mL. The use of 100 μg/mL b-mAb increased the ECL signal by about 200% compared to the results obtained at 25 μg/mL. Standard conditions selected for subsequent studies were 2 µg/mL for coating the capture mAbs and 100 ng/mL for biotinylated detection mAbs. The ability of these antibody pairs to discriminate between ricin and RCA-1 is illustrated in Figure 2. The 1797/b-1443 pair afforded 2 orders of magnitude selectivity in the detection of ricin over RCA-1 at the highest level tested (100 ng/mL), and greater selectivity was shown at lower concentrations of analytes.

To shorten assay time, it could be advantageous to add b-mAb and Ru(bpy)₃-streptavidin simultaneously, rather than performing separate incubation and wash steps for the two reagents. Figure 3 illustrates the results of comparing procedures. Simultaneous reagent addition raised the background ECL approximately 5-fold, apparently due to nonspecific binding of the b-mAb-streptavidin complex to the capture mAb layer. Although the higher background only influenced the ECL response significantly below 1 ng/mL ricin, all further assays employed sequential addition.

An alternative assay format used directly labeled Ru(bpy)₃-mAbs, instead of indirect detection of biotinylated mAbs using streptavidin conjugate. The titration of two labeled mAbs is illustrated in Figure 4(a). The useful assay range (>10⁴ luminescence units) corresponded to labeled mAbs at approximately 200 ng/mL. Another assay parameter that was investigated was the performance of fresh versus dried assay plates. As shown in Figure 4(b), assays conducted on plates freshly coated with mAb were indistinguishable from those conducted on dried mAb-coated plates. Because the two coating protocols produced similar results, dried plates were used in subsequent experiments. This facilitated work flow and enabled preparation of batches of coated plates, for most reproducibility.

Ground beef was spiked with pure or crude ricin and homogenized in PBS-galactose (2 mL/g). The slurry was diluted for application to mAb-coated assay wells.

Figure 5 illustrates the recovery of pure ricin from ground beef, determined by ELISA as well as ECL analysis. Recoveries varied from 30 to 60% at different spike levels, but were more consistent and generally higher by ECL analysis. The 1 ng/g spike could not be determined quantitatively by ELISA, but was readily determined by ECL. The limit of detection was also computed for each assay (Figure 6): 0.5 ng/g for ECL and 1.5 ng/g for ELISA.

The analysis of ground beef spiked with crude ricin is illustrated in Figure 7.