Anal Bioanal Chem DOI 10.1007/s00216-014-7934-1

RESEARCH PAPER

Development and validation of an ELISA kit for the detection of ricin toxins from biological specimens and environmental samples Hsiao Ying Chen & Hung Tran & Ling Yann Foo & Tracey Wenhui Sew & Weng Keong Loke

Received: 12 March 2014 / Revised: 21 May 2014 / Accepted: 28 May 2014 # Springer-Verlag Berlin Heidelberg 2014

Abstract Ricin is a toxin that can be easily extracted from seeds of Ricinus communis plants. Ricin is considered to be a major bio-threat as it can be freely and easily acquired in large quantities. A deliberate release of such toxin in civilian populations would very likely overwhelm existing public health systems, resulting in public fear and social unrest. There is currently no commercially available or FDA-approved prophylaxis such as vaccines, or therapeutic antitoxins or antidotes, available for ricin intoxication. Patient treatment is typically supportive care based on symptoms, often designed to reinforce the body’s natural response. This paper describes the development and validation of a robust ELISA test kit, which can be used to screen for ricin in biological specimens such as whole blood and faeces. Faecal specimens are shown in this study to have better diagnostic sensitivity and a wider diagnostic window compared to whole blood. From these results, it is concluded that faeces is the most suitable clinical specimen for diagnosis of ricin poisoning via the oral route. The ELISA test kit can also detect ricin in environmental samples. An advantage of this ELISA kit over other commercial off-the-shelf (COTS) detection kits currently on the market that are developed to screen environmental samples only is its ability to diagnose ricin poisoning from clinical specimens as well as detect ricin from environmental samples. Published in the topical collection Analysis of Chemicals Relevant to the Chemical Weapons Convention with guest editors Marc-Michael Blum and R. V. S. Murty Mamidanna. H. Y. Chen (*) : L. Y. Foo : T. W. Sew : W. K. Loke DSO National Laboratories, 20 Science Park Drive, 118230 Singapore, Singapore e-mail: [email protected] H. Tran (*) Land Division, Defence Science and Technology Organisation (DSTO), 506 Lorimer Street, Fishermans Bend, Melbourne, Victoria 3207, Australia e-mail: [email protected]

Keywords Ricinus communis . Ricin . ELISA . Intoxication . Oral gavage . Diagnostics . Rat

Introduction Ricin, a potent toxin, originates from seeds of Ricinus communis plants. These seeds, commonly known as castor beans, are used worldwide to produce castor oil for its many industrial applications. The toxin is found in castor meal, a byproduct of castor oil production, which can be easily extracted using established methods [1]. The R. communis plant is widely distributed around the world and easily accessible as it is commonly found in the wild and often used as an ornamental plant in domestic gardens. Its accessibility and high toxicity make many believe that ricin poses a high risk to national security and public health. As a result, ricin is classified as a category B agent by the Centers for Disease Control and Prevention (CDC) in the USA [2] and a Schedule 1 agent of the Chemical Weapons Convention by the Organisation for the Prohibition of Chemical Weapons (OPCW). Ricin belongs to the group of type 2 ribosome-inactivating proteins (RIPs) [3, 4]. It is a heterodimeric glycoprotein consisting of two polypeptides, A-chain (approximately 30 kDa) and B-chain (approximately 32 kDa), also known as RTA and RTB, respectively. At the cellular level, RTB is responsible for the binding and internalisation of ricin in cells. In order to do this, as RTB is a lectin, it binds specifically to certain eukaryotic cell surface receptors, such as glycoproteins and glycolipids containing beta-1,4-linked galactose residues, to facilitate the entire internalisation process [5–7]. Once inside the cells, RTA exerts its toxic effects at the molecular level by inhibiting protein synthesis, which ultimately leads to cell death [8, 9]. The process involves catalytically inactivating the ribosome by removing an adenine from position 4324 of the 28S ribosomal RNA (rRNA) in the 60S

H.Y. Chen et al.

ribosomal subunit [10]. The depurinated rRNA is then unable to bind protein elongation factor-2, which leads to cessation of protein synthesis [10]. Animal studies have suggested that the fate of ricin after intoxication is largely route-specific. Fodstad et al. [11] had injected ricin (labelled with 125I) intravenously into mice and detected 125I-labelled ricin in its intact form distributed in the liver, blood, lungs, spleen, kidneys and heart. The examination of urine samples withdrawn from the bladder with a syringe revealed that ricin excreted in urine was in the form of a low molecular weight degradation product. Intravenously administered ricin did not enter the gastrointestinal tract and, therefore, was not detected in faeces by Fodstad et al. Working on an oral ingestion model, Roy et al. [12] found that ricin excreted in stool was biologically active in cytotoxicity assay, providing evidence that intact ricin is present in faeces. The ricin LD50 (median lethal dose that is lethal to 50 % of a population) for oral ingestion in the mouse model is about 30 mg/kg [10]. Symptoms of ricin intoxication after ingestion occur 4 to 6 h but may take as long as 10 h [13–15]. Initial symptoms are non-specific and may include the following: nausea, abdominal pain and vomiting, followed by diarrhoea, cramps, dilation of pupils, fever, dehydration, anuria (absence of or defective urine production), sore throat, headache, hypotension, heartburn, internal bleeding of the stomach and intestines and failure of the liver, spleen and kidneys. Death can occur when the circulatory system collapses after three or more days. There is currently no commercially available or FDA-approved prophylaxis such as vaccines, or therapeutic antitoxins or antidotes, available for ricin intoxication. There is also no specific or specialised treatment for patients exposed to ricin. Patient treatment is typically supportive care based on symptoms, often designed to reinforce the body’s natural response. This, however, does not imply that early diagnosis of oral ricin poisoning is unimportant as failure to do so can lead to moribund outcomes. On another level, early and rapid diagnosis of ricin poisoning and/or detection of ricin release is the only means by which to limit the impact of such bioterrorism act if one is to occur. As mentioned earlier, ricin has characteristics that make it a potential bio-terror agent. Consequently, it is considered a major bio-threat. It can be freely and easily acquired in large amounts, simply extracted and purified to increase its toxicity and potency levels. A deliberate release of such toxin in a civilian population would very likely overwhelm existing public health systems, resulting in public fear and social unrest. Current commercial off-the-shelf (COTS) ricin detection kits are intended solely for environmental testing. To date, no ricin detection kit on the market is developed and validated purely for clinical applications. This paper aims to address this shortfall by outlining the development and validation of a quick and simple sandwich enzyme-linked immunosorbent assay (ELISA) test kit for the diagnosis and detection of trace

levels of ricin toxin in both clinical specimens and environmental samples.

Materials and methods Antibodies, toxins and lectins Monoclonal antibody (Mab) RA999 was purchased from Meridian Life Science (USA). The hybridoma cell line for Mab 7G12 was kindly provided by the US Army Medical Research Institute of Infectious Diseases (USAMRIID) and produced at the University of Western Australia using methods described previously [9]. The rabbit polyclonal antibody (Pab), nRC1, was raised against the ricin toxoid (Toxin Technology, USA) as immunogen. Two New Zealand white rabbits were used. They were each immunised with 1 mg of immunogen mixed with Freund’s complete adjuvant (Sigma, USA). Three boosters of 1 mg immunogen in Freund’s incomplete adjuvant followed, 14 days apart. The immunogen was injected subcutaneously into four different sites, 0.5 ml per site, in each rabbit. The rabbits were terminally bled when the titre of the antibodies had plateaued. The polyclonal antibodies were affinity purified on Pierce Protein A/G agarose (Thermo Scientific, USA). This antibody was biotinylated using the EZ-Link Sulfo-NHS-LC-biotin kit (Thermo Scientific, USA). Purified ricin was either purchased from commercial suppliers (Toxin Technology, USA) or produced in-house from seeds of known cultivars in Singapore using the method described by Hegde et al. [16]. Castor beans used in this animal study were purchased from Sandeman Seeds (France). Purified lectins together with abrin and R. communis agglutinin were purchased from either Sigma-Aldrich (Australia) or Calbiochem (USA). This panel includes R. communis agglutinin RCA 120 , abrin from Abrus precatorius (jequirity bean) and lectins from Viscum album (European mistletoe), Arachis hypogaea (peanut), Canavalia ensiformis (Jack bean), Triticum vulgaris (wheat), Tetragonolobus purpureas (Lotus tetragonolobus, winged or asparagus pea), Lens culinaris (lentil), Phaseolus vulgaris (French bean), Sambucus nigra (European elderberry), Phytolacca americana (American pokewood) and Saponaria officinalis (soapwort). Production and assay procedures of the ELISA kit All components of the ELISA kit were contract manufactured in an ISO 13485-certified facility. The ELISA kit was packaged with all components required to successfully perform a 32-well microplate assay. These components include an antibody-coated microtitre plate, phosphate-buffered saline (PBS) as non-reactive control (NRC), recombinant ricin A-

Development and validation of an ELISA kit to detect ricin toxins

chain as reactive control (RC), sample treatment buffer (STB), 10× wash buffer (PBS containing 0.05 % Tween 20), biotinylated nRC1 anti-ricin Pab, high-sensitivity streptavidinconjugated horseradish peroxidase (Strep-HRP conjugates, Thermo Scientific, USA), HRP diluent (PBS containing stabiliser and bovine serum albumin, BSA), substrate 3,3′,5,5′tetramethylbenzidine (TMB) solution (BioFX TMB Slow Kinetic One Component HRP Microwell Substrate, SurModics, USA) and stop solution of 2 M sulphuric acid. Antibody-coated ELISA plates were produced where each well was coated with 150 ng of 7G12 Mab and 100 ng of RA999 Mab in 0.1 M sodium carbonate, pH 9.6, and incubated overnight at room temperature. This was followed by 1 h blocking using 150 μl of PBS containing 4 % (w/v) BSA at room temperature. The plates were washed and subsequently stabilised using a commercial stabiliser before drying overnight. Each completed ELISA plate, vacuum-sealed in an aluminium pouch with 1.0 g of desiccant, was packaged together with the above reagent components. All incubation volumes were 100 μl unless specified. Volumes of 25 μl of sample solution (whole blood, processed faeces or environmental samples) were pre-incubated with 50 μl of STB for 10 min at room temperature. All assay samples were tested in replicates. At the same time, quadruplicates of RC and NRC were also prepared, treated and incubated in the same way. These RC and NRC served as internal assay controls to ensure assay result validity. For the assay to be valid, the average RC must have an absorbance reading greater than 0.6 and NRC must have an absorbance reading less than 0.1 optical units at 450 nm. The resultant samples or assay control mixtures (75 μl) were added to each ELISA well containing 25 μl (approximately 200 ng) of biotinylated nRC1 Pab dissolved in PBS containing 4 % (w/v) BSA. The ELISA plate was then sealed with an adhesive plastic cover and incubated for an hour at 37 °C with shaking at 600 rpm. Each well of the ELISA plate was then washed six times with 250 μl wash buffer. This was followed by 10 min of incubation of 1:75 dilution of Strep-HRP conjugates at room temperature. ELISA plate wells were once again washed six times with 250 μl wash buffer. The substrate TMB solution was then added to each well and the plate incubated at room temperature for 10 min in the dark for colour development. The ELISA reaction was stopped by adding 50 μl of stop solution to each well. The absorbance of each well was read at 450 nm on a spectrophotometer (SpectraMax 250, Molecular Devices, USA). An individual well result is considered positive when the absorbance signal is twice the average absorbance of all NRC wells of the performed assay. This value of twice the average absorbance of all NRC wells on an ELISA plate is known as the cut-off value, COV, to determine if a sample is considered to be positive or negative.

Validation: precision study Assay precision measurements determine the variation of results from replicates of different sets of concentrations, operators, days and/or batches of reagents. Our precision study was conducted using the referenced methods of EP05A2 by the Clinical and Laboratory Standards Institute (CLSI) [17] which involved testing over five different days (day 1, day 5, day 10, day 15 and day 22) with three different operators and batches of the ELISA kits. On each testing day, all three operators performed precision study assays on plates from production batch no. 2. On the same day, operator 1 also assayed additional plates from production batch nos. 1 and 3. Assay plates designed with 32 wells were used in this study. PBS containing 4 % (w/v) BSA was spiked with purified ricin and prepared fresh on the day of testing. The ricin concentrations tested were 0.00, 0.50, 1.25 and 5.00 ng/ml (six replicates for each concentration group). Together with four replicates of each assay NRC and RC (provided in the ricin ELISA kit), these were assayed simultaneously on the same plate. The ELISA kit intra-assay variations (within-run variation amongst replicate wells of each tested ELISA plate) were determined independent of testing day, operator or ELISA plate batches. The coefficients of variation (CV) of the replicates in each group of ricin-spiked concentrations (0.00, 0.50, 1.25 and 5.00 ng/ml) and assay controls (NRC an RC) were calculated. Each tested ELISA plate would generate six CVs. The average of these six CVs would be the reported withinrun precisions for that particular assay run. To investigate the complexity of between-run precisions, a mixed-effects ANOVA model was employed to analyse the contributing factors to the observed variability of the precision study results. The model assigned each ricin-spiked concentration (0.00, 0.50, 1.25 and 5.00 ng/ml) together with the assay control (NRC and RC) as standards, a total of six standards. The model then applied these standards as a fixed effect to the contribution of observed result variations in the precision study. Different ELISA production batches, operators and test days were assigned as random effects as they can be randomly drawn from the entire population (as an example, another operator could be using another batch of ELISA kit this time or sometime in the future). Similarly, interactions between test day and batch, test day and operator, as well as batch and operator, were also assumed to be random effects. All the above-listed fixed and random effects can be represented in the following equation: yijkmn ¼ μ þ β i þ b j1 þ bk2 þ bma þ bjk4 þ bjm5 þ bkm6 þ εijkmn

i j k

1, 2, 3, 4, 5, 6 (standards) 1,…, 5 (day) 1, 2, 3 (batch)

H.Y. Chen et al.

m n βi bj1 bk2 bma bjk4 bjm5 bkm6

1, 2, 3 (operator) 1,…, 4 or 6 (replications) Effect of standard (fixed effect) Effect of day (random effect) Effect of batch (random effect) Effect of operator (random effect) Interaction between day and batch (random) Interaction between day and operator (random) Interaction between batch and operator (random)



b j1 eN 0; σ2b1 ; bk2 eN 0; σ2k2 ; bma eN 0; σ2ma ; bjk4 eN 0; σ2b4 ;
bjms eN 0; σ2b6 By using the mixed-effects ANOVA model approach, a variance component analysis was performed to determine the percentage of the total variance attributable to each assigned random effect or simply known as factor in this model. Factors under investigation in this model were effects of (1) betweenday testing, (2) between-production-batch testing, (3) between-operator testing, (4) interactions of between-day and between-batch testing, (5) interactions of between-day and between-operator testing and (6) interactions of between-batch and between-operator testing. R 3.0.1 [18] and the lme4 [19] statistical packages were used to perform the linear mixed-effects analysis of the relationship between the measured optical units and the six standards of this precision study. Validation: robustness study The robustness of the ELISA kit was also investigated to determine the assay tolerance range. Deliberate variations to four chosen assay parameters were introduced, and the results were compared to those of non-deviated assays. The deviated parameters of interest were pipetting errors in sample volumes (25±5 μl) and Strep-HRP volumes (54±2 μl), incubation temperature (37±2 °C) and incubation time (60±10 and 10 ±1 min). Similar to the precision study, the 32-well plate layout was used for this study. Human whole blood spiked with ricin of concentrations 0.00, 0.50, 1.25 and 5.00 ng/ml (six replicates per concentration group) and four replicates of each assay control (NRC and RC) were assayed simultaneously. The significance of the results was analysed by the twofactor ANOVA statistical method, which assessed the relationship between deviated and non-deviated assay conditions, as well as the associated different ricin concentration levels tested. Validation: analytical sensitivity and cross-reactivity Validation of the ELISA kit to determine the assay’s analytical limit of detection (LOD) was conducted with PBS as well as

ricin-spiked human faeces (n=5) and human whole blood (n= 13). Analytical LOD is the lowest concentration level that can be determined to be statistically different from a blank (95 % confidence). The levels of cross-reactivity were assessed using ricin from different cultivars (HPLC purified grade), commercial-grade abrin, R. communis agglutinin and a panel of lectins. All materials were diluted to the same concentration and treated the same way for testing and comparison. Validation: human specimens and environmental air samples Human whole blood specimens from Singapore, USA and Australia populations were tested. Blood specimens from Singapore were collected through donations with consent, and those of USA populations were purchased (BioChemed Services, USA) while blood specimens from the Australia population were obtained from the Australian Red Cross (through a Material Transfer Agreement). Whole blood specimens collected, without any processing, were tested. Human faeces from the Singapore population were also collected through donations with consent. Faeces were weighed and mixed thoroughly in two volumes of PBS. This solubilised homogeneous slurry was then centrifuged (15 min at 2,360×g) to clarify the resulting supernatant for testing. Absorbance results greater than the COV (twice the average absorbance of all NRC wells) of the performed assay were considered positive. In Australia, aerosol samples were collected using a horizontal wet wall cyclone. The cyclone is a largevolume air sampler that collects ambient aerosol particles at the rate of ~800 l of air per minute and concentrates these particles into a liquid. The environmental samples used in this study were collected at several different sites around Australia. Each environmental sample was generated by collection of aerosol particles over a 24-h period and concentrated down to a solution (~10 ml) consisting of 0.01 % (v/v) Tween 80 in water. These collection tubes were then clarified by centrifugation (15 min at 3,034×g) whereby the supernatant was used for testing. Air sampling in Singapore was carried out using the C100 Collector integrated into an IBAC (FLIR Systems Inc., USA) or the Coriolis® Delta (Bertin Technologies, France). C100 Collector sampling was undertaken at 150 l/min for at least 12 h. The collected samples were washed into a sample collection vial using rinse fluid provided by the manufacturer. Coriolis® Delta sampling was carried out at a flow rate of 300 l/min for 30 min. Air was drawn through an inlet into a collection buffer. A sample volume of 10–15 ml was collected in each operation at different Singapore sites. These collected air samples, in liquid form, were tested without centrifugation. Absorbance results greater than the COV (twice the average absorbance of all NRC wells) of the performed assay were considered positive.

Development and validation of an ELISA kit to detect ricin toxins

Animal studies: crude ricin preparation Castor beans were dried in a vacuum oven at 30 °C until constant weight was achieved. They were then homogenised in a blender with two volumes of PBS. The resulting slurry was shaken for 48 h at 600 rpm on a rotary shaker. The seed meal was removed by centrifugation at 4,000×g for 30 min at 4 °C. The supernatant was then filtered sequentially through Whatman grade 3 filter paper, followed by 5 and 0.45 μm Supelco filtration units (Sigma-Aldrich, USA) to remove insoluble solids and microbes prior to storage at 4 °C. The protein content of this crude ricin preparation was determined by the modified Lowry assay (BioRad, USA). A small amount of this crude ricin preparation was fractionated, 5 and 10 μg per lane, using sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE at 10 %). Purified ricin was also run on the same SDS-PAGE at 0.5, 1.0, 2.0 and 4.0 μg per lane. By using the intensities of the purified ricin bands on the Coomassie blue-stained gel, a calibration curve was constructed and the amount of ricin in the crude ricin preparation was estimated to be approximately 13 % of the total protein. The amount of crude ricin preparation given to each animal for testing was based on this ricin content percentage. Animal studies: intoxication procedure Ricin intoxication studies in rats were carried out in Singapore. Prior to intoxication, background or baseline specimens (blood, faeces and urine) from each rat under investigation were analysed as negative controls for the study. Male Wister rats (250–300 g) received a single bolus of crude ricin preparation equivalent to a ricin dose of 14 mg/kg (n=4), 28 mg/kg (16 rats in total for this group; however, sampling numbers varied to maximise different time points, with most time points at n=4 or n=8) or 56 mg/kg (n=3) by oral gavage. The oral gavage was carried out using a Tom CatTM catheter tubing (Covidien, USA) fitted to a hypodermic syringe containing the crude ricin preparation. The tubing was gently introduced into the oesophagus of the rat. After the challenge, the rats were housed in ventilated metabolic cages for observations. The cages were changed every 24 h to prevent crosscontamination of faeces between days. The rats were maintained under a 12-h light-dark cycle with food and water provided ad libitum. All procedures were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC). Animal studies: faecal, blood and urine collections Falcon tubes (50 ml) were attached to the metabolic cages in order to collect the faeces for analysis. The collected faeces were processed in the same manner as the human faeces. The processed supernatant was stored at 4 °C and assayed within

24 h. Blood was collected via tail bleed with local anaesthesia. Rat tails were cleansed of contaminating faeces with lots of warm water before the bleed. Approximately 350 μl of blood was collected in EDTA tubes at each time interval, in line with animal ethics guidelines. Again, blood specimens were stored at 4 °C and assayed within 24 h. Urine collection using the attached Falcon tube was not ideal as it would crosscontaminate with faeces that may contain ricin. Consequently, urine specimens were collected from rats using spot collection method [20] where a plastic beaker was placed directly under the penis and the rats would be tickled to produce the urine sample.

Results and discussions Validation: precision and robustness studies Precision is often defined as the closeness of agreement between test results obtained under stipulated identical conditions. A series of precision experiments were conducted on five different days to determine within-run and between-run precisions [17, 21], often referred to as intra- and inter-assay variations. Within-run precisions, expressed as CV percentages, are tabulated in Table 1. As expected, there were minimal within-run result variations obtained for this ELISA kit as shown by the low CV values ranging from 2.9 to 6.0 %. These low CV values can be attributed to the effectiveness of stringent quality control during the ELISA kit production stage, ensuring well-to-well consistency. There were three different operators with three different production batches of the ELISA kit to be tested on five different days, totaling 25 assays in the precision study. Consistently, low within-run CV values were obtained for each of these 25 assays performed. This suggests that each ELISA kit can be run confidently with minimal within-run or well-to-well results variation. Ideally, within-run precision results can be dissected down to individual ricin-spiked concentration and assay control to determine the results variation amongst these groups. Such method would have generated 150 CV values compared to the current summarised 25 CV values, as displayed in Table 1. Therefore, the approach taken to average out the CV values amongst the groups of ricin-spiked concentration and assay control, looking at the overall within-run variations amongst these groups, is considered to be more appropriate. The mixed-effects ANOVA model results have clearly indicated that both assigned fixed effects, measured optical absorbance units and the standards in the precision study were statistically significant with low p values, as shown in Table 2. The model has also identified how random effect factors (day, operator, batch and the interactions amongst them) may or may not contribute to the variability of results obtained and,

H.Y. Chen et al. Table 1 ELISA kit precision study: within-run precision results (CV expressed as percentages) Day

Operator 1

Operator 2

Operator 3

Batch 1

Batch 2

Batch 3

Batch 2

Batch 2

1

3.6

3.4

5.5

3.0

3.8

5 10 15 22

4.3 3.5 3.3 4.6

5.5 3.9 6.0 4.1

4.1 5.5 3.9 5.0

3.9 3.6 4.7 2.9

3.0 4.5 4.2 3.5

ultimately, the precision of the ELISA kit. The percentage of the total variation attributable to each factor is presented in Table 3. A breakdown of all assigned random effect factors such as different testing days, operators, production batches of the ELISA kit as well as the interactions amongst these factors have shown that these factors did not contribute significantly to the variability of the precision study results. Factors with random effects such as different ELISA production batches, testing days and operators contributed less than 1.0 % of the variability to the observed results in the precision study. In the model, operator-related factors attributed less than 0.5 % and ELISA kit production batch had 0.0 % contributed to the observed results variability. Consequently, the ELISA kit has been shown to produce consistent results amongst different operators, production batches and testing days. The majority (>99 %) of the observed variability was attributed to other residual factors beyond the control of the ELISA kit. Assayto-assay variations, not to be confused with batch-to-batch or between-batch variations, are believed to be the main residual factor. As an example, during the precision study, ricin-spiked concentration samples were prepared daily. Daily fluctuation of these ricin-spiked concentrations will be amplified in the results, contributing to the observed variability. It is these residual factors, not those manufactured, provided or originated by the ELISA kit, which can arise from assay to assay, contributing significantly to the variability of results. To further investigate, analyses using all data (from all tested days, all operators and all production batches) obtained in the precision study to calculate the CV values of the six standards (ricin-spiked concentrations plus NRC and RC) are tabulated in Table 4. Indeed, CV values for all ricin-spiked concentration samples were higher than expected (>10 %). For the ricinTable 2 ELISA kit precision study: ANOVA table for the mixed-effects model—both fixed effects are statistically significant at the α=0.05 significance level

Optical units Standard

Estimate

Standard error

t statistic

p value

−0.27 0.327

0.0515 0.0125

−5.237 26.107

1). a Ricin detected in faeces from intoxicated rats at the 14-mg/kg dose. b

Ricin detected in faeces from intoxicated rats at the 28-mg/kg dose. * and ^ indicate only three out of the four rats produced faeces at the stipulated time period

degradation products, could potentially be useful in extending the diagnostic window further. An adopted spot urine method [20] was used to collect urine free from stool contamination. Metabolites from urine collected were fractionated on SDSPAGE and transferred onto a membrane for Western blotting to determine if the current ELISA antibodies can detect these excreted ricin metabolites. Unfortunately, only one of the antibodies used in the ELISA system can detect a 15-kDa peptide, believed to be a ricin metabolite. The appearance of this 15-kDa peptide was rather erratic, varying degrees from rat to rat (data not shown). The lack of reproducibility and the fact that not all current antibodies can detect this ricin

metabolite have raised questions into the suitability of detecting ricin in urine for this ELISA system. Therefore, it was concluded, for the current ELISA system, to not utilise urine specimens for the diagnosis of ricin intoxication. It is clear from our studies that faeces are the more suitable clinical specimens to use than whole blood when detecting the presence of ricin via oral route poisoning. Currently, there is no approved antidote for ricin poisoning and treatment remains essentially supportive in nature. However, this does not mean that early diagnosis of oral ricin poisoning is unimportant as failure to differentiate ricin poisoning from normal food poisoning could lead to moribund outcomes when the

H.Y. Chen et al.

patient goes into shock at home between 48 and 72 h post ricin ingestion and is not managed adequately in a hospital setting. Early diagnosis of ricin poisoning would ensure that such patients are admitted into hospital to allow for supportive medical care in a timely manner. Besides, developments in diagnostic capabilities and medical countermeasures for ricin poisoning are occurring concurrently. Current absence of an approved or experimental antidote for ricin poisoning should not be assumed to lead to similar absence in the future. As medical care is always given on the basis of medical evidence of disease, development of a competent diagnostic kit for ricin poisoning will always be essential and should not be dictated by the presence or absence of medical countermeasure solutions. Human cases of ricin poisoning are rare and often are not followed up by diagnostic examination of clinical specimens because, currently, no diagnostic kits exist on the market that can do such testing. This ELISA test kit was developed for these rare cases of ricin poisoning incidents, mass casualties due to bioterrorism attacks or infrequent accidents. The low incidence rate of ricin poisoning has resulted in a lack of commercial interest; however, national laboratories and agencies cannot overlook the impact of such incidence to national security and the demand it places on our current health system. The application of this ELISA test kit can also be extended for use in postmortem forensic confirmation cases. Unique applications in areas of clinical diagnosis and forensic investigations are very strong drivers in support of this ELISA test kit’s practical relevance, which differentiate it from current COTS ELISA kits tailored for environmental samples only, most of which have not been validated for clinical specimen testing.

Conclusion The ELISA kit described in this paper aims to fulfil an essential and critical gap in the market of COTS ricin detection and that is in its clinical application for testing and diagnosis of ricin poisoning using specimens such as blood and faeces. Ricin intoxication from accidents as results of acquiring and/or purifying the toxin for bioterrorism acts or cases of mass casualties due to bioterrorism attacks, although rare, will undoubtedly overwhelm the current health system and create public fears and social unrest. Therefore, it cannot be overlooked and dismissed as a rare chance event. The described ELISA kit enables clinicians to order a quick and accurate test to rule out or confirm cases of ricin poisoning in a matter of 1.5 h. The results presented so far, although based on an animal model, are believed to be applicable in diagnosing ricin intoxication via the oral route in humans. The data have suggested that the developed and validated ELISA kit is reproducible and robust with minimal to no cross-reactivity

with other closely related lectins. Our studies have concluded that detection of ricin in faecal specimens provides a wider diagnostic window (as early as 4 h to at least 5 days after ricin intoxication) and delivers better assay output signals resulting in improved diagnostic sensitivity and greater OD/COV values (at least ten in asymptomatic ricin dose). Clearly, ricin detection in faeces is dose- and time-dependent for oral route poisoning, and faeces is the ideal clinical specimen, not whole blood. Acknowledgments The authors wish to thank Dr. Seng Kok Yong from DSO National Laboratories for analysing the precision study results using the mixed-effects ANOVA model. The authors also wish to acknowledge Lim Chau Wen Kevin and Ng Siew Lai from DSO for their supporting roles in performing assays and animal care and Dr. Mick Alderton from DSTO for his general guidance and review of this paper. Lastly, the authors wish to thank USAMRIID for providing the anti-ricin hybridoma cell lines. This work was jointly supported by grants from the Future Systems and Technology Directorate and Armed Forces Medical Corps of the Singapore Ministry of Defence and the Defence Science and Technology Organisation of the Australian Department of Defence.

References 1. Tavasolian B, Kharrazy P (1978) Extraction and partial purification of ricin from Ricinus communis L. Pahlavi Med J 9(1):21–26 2. Shea D, Gottron F (2004) Ricin: technical background and potential role in terrorism. CRS report for Congress (received through the CRS web) 3. Knight B (1979) Ricin—a potent homicidal poison. Br Med J 1(6159):350–351 4. Lord JM, Roberts LM, Robertus JD (1994) Ricin: structure, mode of action, and some current applications. FASEB J 8(2):201–208 5. Olsnes S (2004) The history of ricin, abrin and related toxins. Toxicon 44(4):361–370 6. Sandvig K, van Deurs B (2000) Entry of ricin and Shiga toxin into cells: molecular mechanisms and medical perspectives. Embo J 19(22):5943–5950 7. Day PJ, Pinheiro TJ, Roberts LM, Lord JM (2002) Binding of ricin A-chain to negatively charged phospholipid vesicles leads to protein structural changes and destabilizes the lipid bilayer. Biochemistry 41(8):2836–2843 8. Lord MJ, Jolliffe NA, Marsden CJ, Pateman CSC, Smith DC, Spooner RA, Watson PD, Roberts LM (2003) Ricin: mechanisms of cytotoxicity. Toxicol Rev 22(1):53–64 9. Dertzbaugh MT, Rossi CA, Paddle BM, Hale M, Poretski M, Alderton MR (2005) Monoclonal antibodies to ricin: in vitro inhibition of toxicity and utility as diagnostic reagents. Hybridoma 24(5): 236–243 10. Audi J, Belson M, Patel M, Schier J, Osterloh J (2005) Ricin poisoning: a comprehensive review. Jama 294(18):2342–2351 11. Fodstad O, Olsnes S, Pihl A (1976) Toxicity, distribution and elimination of the cancerostatic lectins abrin and ricin after parenteral injection into mice. Br J Cancer 34(4):418–425 12. Roy CJ, Song K, Sivasubramani SK, Gardner DJ, Pincus SH (2012) Animal models of ricin toxicosis. In., vol. 357: 243–257 13. Furbee B, Wermuth M (1997) Life-threatening plant poisoning. Crit Care Clin 13(4):849–888 14. Reed RP (1998) Castor oil seed poisoning: a concern for children. Med J Aust 168(8):423–424

Development and validation of an ELISA kit to detect ricin toxins 15. Challoner KR, McCarron MM (1990) Castor bean intoxication. Ann Emerg Med 19(10):1177–1183 16. Hegde R, Maiti TK, Podder SK (1991) Purification and characterization of three toxins and two agglutinins from Abrus precatorius seed by using lactamyl-sepharose affinity chromatography. Anal Biochem 194(1):101–109 17. Tholen D, Kallner A, Kennedy J, Krouwer J, Meier K (2004) Evaluation of precision performance of quantitative measurement methods; approved guideline—second edition. Clinical and Laboratory Standards Institute, CLSI document EP05-A2 (CLSIEP05-A2) 18. Team RC (2013) R: a language and environment for statistical computing. R Foundation for Statistical Computing. Vienna, Austria 19. Bates DM, Maechler M, Bolker B, Walker S (2014) Ime4: linear mixedeffects models using Eigen and S4 classes. R package version 1.1-6 20. Kurien BT, Everds NE, Scofield RH (2004) Experimental animal urine collection: a review. Lab Anim 38(4):333–361 21. Chesher D (2008) Evaluating assay precision. Clin Biochem Rev/ Aust Assoc Clin Biochem 29(Suppl 1):S23–S26 22. Griffiths GD, Knight SJ, Holley JL, Thullier P (2013) Evaluation by ELISA of ricin concentration in fluids and tissues after exposure to aerosolised ricin, and evaluation of an immunochromatographic test for field diagnosis. J Clin Toxicol 3:162. doi:10.4172/2161-0495. 1000162 23. Garber EAE, Eppley RM, Stack ME, McLaughlin MA, Park DL (2005) Feasibility of immunodiagnostic devices for the

24.

25.

26. 27.

28.

29.

30.

31.

detection of ricin, amanitin, and T-2 toxin in food. J Food Prot 68(6):1294–1301 Men J, Lang L, Wang C, Wu J, Zhao Y, Jia PY, Wei W, Wang Y (2010) Detection of residual toxin in tissues of ricin-poisoned mice by sandwich enzyme-linked immunosorbent assay and immunoprecipitation. Anal Biochem 401(2):211–216 Brandon DL (2011) Detection of ricin contamination in ground beef by electrochemiluminescence immunosorbent assay. Toxins 3(4): 398–408 Drickamer K (1988) Two distinct classes of carbohydrate-recognition domains in animal lectins. J Biol Chem 263(20):9557–9560 Roberts LM, Lamb FI, Pappin DJC, Loard JM (1985) The primary sequence of Ricinus communis agglutinin. Comparison with ricin. J Biol Chem 260(29):15682–15686 Hegde R, Podder SK (1998) Evolution of tetrameric lectin Ricinus communis agglutinin from two variant groups of ricin toxin dimers. Eur J Biochem 254(3):596–601 Ramsden CS, Drayson MT, Bell EB (1989) The toxicity, distribution and excretion of ricin holotoxin in rats. Toxicology 55(1–2):161–171 Cook DL, David J, Griffiths GD (2006) Retrospective identification of ricin in animal tissues following administration by pulmonary and oral routes. Toxicology 223(1–2):61–70 He X, McMahon S, Henderson 2nd TD, Griffey SM, Cheng LW (2010) Ricin toxicokinetics and its sensitive detection in mouse sera or feces using immuno-PCR. PloS One 5(9)