Anal. Chem. 2007, 79, 659-665
Detection of Functional Ricin by Immunoaffinity and Liquid Chromatography-Tandem Mass Spectrometry F. Becher,*,† E. Duriez,† H. Volland,† J. C. Tabet,‡ and E. Ezan†
CEA, Service de Pharmacologie et d’Immunologie, 91191 Gif-sur-Yvette, France and LCSOB UMR 7613 CNRS, Universite´ Pierre et Marie Curie, Paris, France
The toxin ricin is a biological weapon that may be used for bioterrorist purposes. As a member of the group of ribosome-inactivating proteins (RIPs), ricin has an Achain possessing N-glycosidase activity which irreversibly inhibits protein synthesis. In this paper, we demonstrate that provided appropriate sample preparation is used, this enzymatic activity can be exploited for functional ricin detection with sensitivity similar to the best ELISA and specificity allowing application to environmental samples. Ricin is first captured by a monoclonal antibody directed against the B chain and immobilized on magnetic beads. Detection is then realized by determination of the adenine released by the A chain from an RNA template using liquid chromatography coupled to tandem mass spectrometry. The immunoaffinity step combined with the enzymatic activity detection leads to a specific assay for the entire functional ricin with a lower limit of detection of 0.1 ng/ mL (1.56 pM) after concentration of the toxin from a 500 µL sample size. The variability of the assay was 10%. Finally, the method was applied successfully to milk and tap or bottled water samples. Development of sensitive, specific, and fast methods for detection of biological toxins is of absolute necessity due to the potential threat of bioterrorist attacks since biological toxins may be used as components of biological weapons.1 Ricin is a highly toxic lectin originating from the castor bean plant Ricinus communis. The relatively easy production of ricin by extraction from the bean, its wide availability, and the lack of specific treatment of ricin poisoning make it a potential weapon for military or terrorist use.1,2 Ricin is a type 2 ribosome-inactivating protein (RIP) consisting of a catalytic chain (A chain, ≈32 kDa) linked to a lectinic B chain (≈34 kDa) by a disulfide bridge.3 RIPs have N-glycosidase activity which irreversibly inactivates ribosomes by specifically depurinating the first adenosine in the GAGA nucleotide sequence from * To whom correspondence should be addressed. Phone: 33-1-69-08-13-15. Fax: 33-1-69-08-59-07. E-mail:
[email protected]. † CEA. ‡ Universite´ Pierre et Marie Curie. (1) Bigalke, H.; Rummel, A. Toxicology 2005, 214, 210-220. (2) Audi, J.; Belson, M.; Patel, M.; Schier, J.; Osterloh, J. JAMA 2005, 294, 2342-51. (3) Olsnes, S. Toxicon 2004, 44, 361-70. 10.1021/ac061498b CCC: $37.00 Published on Web 12/09/2006
© 2007 American Chemical Society
a highly conserved loop, i.e., the sarcin-ricin loop, present in the 28S subunit.4,5 Both chains of ricin are necessary for the toxic action: the B chain (RTB) mediates the cellular binding and entry of the A chain (RTA) which possesses the N-glycosidase activity and inhibits protein synthesis.3 The toxicity of ricin measured in terms of the LD50 in mice is on the order of 2 µg/kg body weight after intraperitoneal injection and between 3 and 30 µg/kg after inhalation or oral absorption.2,6 The enzymatic activity of ricin has been previously exploited for development of detection methods by measuring the release of adenine after reaction with ribosome or small RNA or DNA substrates. Detection has been performed by conversion of the released adenine to fluorescent ethenoadenine,7,8 detection of labeled [3H]-adenine by radioactivity,9 or measurement of adenine by liquid chromatography coupled with mass spectrometry.10 Although the limits of detection of these methods were not clearly stated, it can be estimated that sensitivity is low and ranged between 10 and 100 ng/mL. These assays also suffer from a lack of specificity for ricin because any other RIP may also release adenine by the same process.11 Moreover, adenine naturally present in the matrix would interfere with the assay. As a consequence, none of these assays has been applied to environmental samples. More sensitive and specific methods for detection of ricin rely on immunoassay. ELISA generally shows high sensitivity with reported detection limits between 0.1 and 80 ng/mL.12-14 Recently, a more sensitive immuno-polymerase chain reaction assay was (4) Amukele, T. K.; Roday, S.; Schramm, V. L. Biochemistry 2005, 44, 441625. (5) Strirpe, F. Toxicon 2004, 44, 371-383. (6) Fredriksson, S. A.; Hulst, A. G.; Artursson, E.; de Jong, A. L.; Nilsson, C.; Van Baar, B. L. Anal. Chem. 2005, 77, 1545-55. (7) McCann, W. P.; Hall, L. M.; Siler, W.; Barton, N.; Whitley, R. J. Antimicrob. Agents Chemother. 1985, 28, 265-273. (8) Zamboni, M.; Brigotti, M.; Rambelli, F.; Montanaro, L.; Sperti, S. Biochem. J. 1989, 259, 639-643. (9) Brigotti, M.; Barbieri, L.; Valbonesi, P.; Stirpe, F.; Montanaro, L.; Sperti, S. Nucleic Acids Res. 1998, 26, 4306-4307. (10) Hines, H. B.; Brueggemann, E. E.; Hale, M. L. Anal. Biochem. 2004, 330, 119-122. (11) Endo, Y.; Gluck, A. Nucleic. Acids Res. Symp. Ser. 1990, 22, 21-22. (12) Griffiths, G. D.; Newman, H.; Gee, D. J. J. Forensic Sci. Soc. 1986, 26, 349358. (13) Garber, E. A.; Eppley, R. M.; Stack, M. E.; McLaughlin, M. A.; Park, D. L. J. Food. Prot. 2005, 68, 1294-1301. (14) Poli, M. A.; Rivera, V. R.; Hewetson, J. F.; Meril, G. A. Toxicon 1994, 32, 1371-1377.
Analytical Chemistry, Vol. 79, No. 2, January 15, 2007 659
reported with a detection limit of 10 fg/mL.15 However, the major limitation of these assays is the detection of both functional and nonfunctional ricin. Measuring the functional toxin specifically is important because only active toxin poses a threat to human health and life. Other alternatives, such as mass spectrometric methods, can generally achieve rapid, sensitive, and highly specific characterization of proteins. A recently published method uses both MALDITOF and LC/MS/MS.6 MALDI-TOF was used to screen for ricin peptides after trypsination, and LC/MS detected the peptides in the multiple reactions monitoring mode (MRM). No information on sensitivity was given, but it can be hypothesized that the limit of detection was between 100 and 1000 ng/mL, as described for analysis of other proteins or toxins with similar technology.16-18 These methods appear very specific but less sensitive, and like immunoassays, they do not specifically measure functional ricin. Clearly, there is a need for specific detection of active ricin reaching the sensitivity of the best ELISA, i.e., at least 0.1 ng/ mL. Immunorecognition of antibodies combined with mass spectrometric detection is considered as one of the most specific analytical approaches.19 This strategy has been applied to the detection of peptides/proteins in biological samples20 and recently to botulinum toxin.21 Here we report work on such an experimental combination for ricin detection. Using appropriate sample preparation and optimized detection based on the N-glycosidase enzymatic activity, we demonstrate that specific detection of the functional entire ricin at a level of around 0.1 ng/mL is possible and applicable to environmental matrices. EXPERIMENTAL SECTION Chemicals and Reagents. Ricin was provided by Dr. B. Beaumelle of the Departement Biologie-Sante´, Universite´ Montpellier, France. It was extracted from castor beans as previously described.22 Antibodies directed against the ricin B chain were provided by the Laboratoire d’Etudes et de Recherches en Immunoanalyse (CEA, Saclay). Sodium acetate, Hepes, BSA, 15N2-1,3-adenine (internal standard), and adenine were from Sigma-Aldrich (St. Louis, MO). Ultrapure water was from a Milli-Qplus 185 purifier (Millipore, France), gradient-grade methanol from Merck (Darmstadt, Germany), HPLC-quality acetonitrile from SDS (Peypin, France), and analytical formic acid from Sigma-Aldrich (Sigma Chemical Co., St Louis, Mo.). Nanosep 10K Omega filters were from MWG (Val de Fontenay, France). Eppendorf 1.5 mL safe lock tubes (Eppendorf, Germany) were used in sample handling. Nitrogen HP45 was from Air Liquide (Paris, France). (15) Lubelli, C.; Chatgilialoglu, A.; Bolognesi, A.; Strocchi, P.; Colombatti, M.; Stirpe, F. Anal. Biochem. 2006, 355, 102-109. (16) Lin, S.; Shaler, T. A.; Becker, C. H. Anal. Chem. 2006, 78, 5762-5767. (17) Van Baar, B. L.; Hulst, A. G.; Wils, E. R. Toxicon 1999, 37, 85-108. (18) Van Baar, B. L.; Hulst, A. G.; Roberts, B.; Wils, E. R. Anal. Biochem. 2002, 301, 278-89. (19) Tsikas, D. J. Biochem. Biophys. Methods 2001, 49, 705-31. (20) Kippen, A. D.; Cerini, F.; Vadas, L.; Stocklin, R.; Vu, L.; Offord, R. E.; Rose, K. J. Biol. Chem. 1997, 272, 12513-22. (21) Kalb, S. R.; Goodnough, M. C.; Malizio, C. J.; Pirkle, J. L.; Barr, J. R. Anal. Chem. 2005, 77, 6140-6146. (22) Beaumelle, B.; Alami, M.; Hopkins, C. R. J. Biol. Chem. 1993, 268, 23661-9.
660
Analytical Chemistry, Vol. 79, No. 2, January 15, 2007
Safety Considerations. Ricin is a highly toxic protein. Even though low toxin concentrations, i.e., below 1 µg/mL, were needed for this work, it must be handled with strict adherence to safety rules for the handling of toxic substances. All contact with the substance should be avoided. Preparation of Standard and Quality Control (QC) Samples. Two stock standard solutions containing 0.5 mg/mL adenine (one for the calibration standard and one for the QC samples) were prepared by dissolving 10 mg in 20 mL of a 70/30 mixture of water and methanol. One solution containing 0.1 mg/mL of internal standard (I.S.) was prepared in the same mixture. The stock solutions were stored at -20 °C. Preparation of Magnetic Beads for Immunoaffinity Extraction. Magnetic beads (100 µL suspension) with covalently bound protein G (Dynabeads Protein G, Dynal Biotech, Oslo, Norway) were washed three times with 0.5 mL of Na-acetate buffer at pH 5 following the protocol provided by the supplier. For the separation of the beads from solution, a Magnetic Particle Concentrator for Microcentrifuge Tubes (Dynal MPC-S, Dynal Biotech, Oslo, Norway) was employed. After removal of washing solution, the beads were then incubated for 1 h at room temperature with gentle shaking with ricin antibody diluted to 100 µg/ mL in Na-acetate buffer pH 5. After washing (three times with 0.5 mL Na-acetate pH 5), the magnetic beads are resuspended in 100 µL of Hepes-buffer pH 7.4 with 1 mg/mL BSA and transferred to the samples for immunoaffinity precipitation. Sample Processing. Five microliters of magnetic beads with captured IgG prepared as described was added to 500 µL of ricin diluted in Hepes solution containing 1 mg/mL BSA. Samples were incubated for 2 h at 37 °C with gentle shaking. After washing (three times with Hepes-buffer, pH 7.4) to remove weak nonspecific binding, 20 µL of the depurination buffer (ammonium acetate 10 mM, pH 4) was added. Forty microliters of RNA substrate was added (3.55 nmol) in ammonium acetate buffer. After 4 or 24 h of incubation, 3 µL of 1000 ng/mL internal standard and 10 µL of 0.2 M ammonia was added. Samples were filtered in the Nanosep 10K at 14 000 g for 10 min before injection of 40 µL in the chromatographic system for adenine quantification. Chromatographic Conditions for Adenine. Chromatography was performed on an Atlantis C18 column (150 × 2.1 mm i.d., 5 mm particle size, 80 Å porosity) from Waters, Saint Quentin en Yvelines, France. Various C18 analytical columns, i.e., Supelcogel ODP-50 (Supelco, St. Quentin-Fallavier, France), Xterra MS (Waters, Saint Quentin en Yvelines, France), Uptisphere 3 Bio P2 (Interchim, Montlucon, France), Zorbax Stable Bond, and Zorbax Extend (Agilent Technology, Les Ulis, France), were evaluated. The mobile phase A was 100% 1 mM ammonium acetate pH 5.3. Mobile phase B was 50% acetonitrile-50% 1 mM ammonium acetate pH 5.3. The gradient was as follows: from t ) 0 to 0.5 min, 90% A, followed by t ) 0.5 to 2 min by a linear gradient from 90% A to 0% A, 0% A was maintained for 1 min, then from t ) 3 to 3.2 by a linear gradient from 0% A to 90% A. The column was then set at 90% A. The time lapse between two cycles was 12 min. The gradient was delivered at a flow rate of 0.3 mL/min. The mobile phase was introduced into the atmospheric pressure source of the mass spectrometer for detection and quantification of adenine and I.S., except at the beginning of the analysis since it was diverted to waste and replaced with acetonitrile/water (50/
Figure 1. Schematic representation of the assay for functional ricin.
50; v/v) solvent delivered at the same flow rate. Column temperature was 40 °C, and injection volume was 40 µL. Under these source conditions, the retention time was 2.9 min for adenine and I.S. Mass Spectrometry Conditions. The triple-quadrupole mass spectrometer was an API3000 (Applied Biosystems) coupled to the HPLC column via an electrospray interface. Nitrogen was used as nebulizing (1.23 L min-1) and curtain gas (1.25 L min-1). The atmospheric pressure source was operated in the electrospray positive ion mode, and the molecular species produced were selected for low-energy collision processes for MS/MS experiments (multiple-reaction monitoring, MRM). Collision-induced dissociations (CID) were achieved with nitrogen. The selected ion transitions monitored were m/z 136 f 119 and m/z 138 f 120 for adenine and I.S., respectively. The dwell time for each CID transition was 0.5 s. The interchannel delay was 30 ms. The total scan time was 1.06 s. For maximum sensitivity of adenine, the mass spectrometer parameters were optimized as follows: capillary voltage 4.5 kV, declustering potential 51 V (66 V for I.S.), collision energy Elab ) 30 eV for analyte and I.S., source temperature 550 °C. RESULTS AND DISCUSSION The first step of the assay consists of a specific capture of ricin by its B chain (Figure 1), performed on magnetic beads coated with antibodies. Then, the captured ricin is incubated with a 14 mer RNA substrate containing the GAGA sequence and mimicking the natural RNA ribosomal substrate of the toxin. Finally, detection is based upon the LC/MS/MS analysis of adenine released by the toxin A chain. It should be noted that another method exploiting ricin enzymatic activity was recently reported.23 It differs in the detection mode, which is based upon the MALDI-TOF analysis of the depurinated substrate. As mass spectrometry is (23) Kakb, S. R.; Woolfitt, A. R.; Barr, J. R. 54th ASMS Conference on Mass Spectrometry, 2006, Session MP06, Poster 135.
generally more sensitive for small molecules, such as adenine, than for oligonucleotides due to adduct ions,24 we believe that our strategy should lead to more efficient ricin detection. Optimization of each step of the assay is presented and discussed below. Development of the LC/MS/MS Method for Adenine Detection. High-performance liquid chromatography coupled with mass spectrometry (LC/MS) was selected as it provides rapid, sensitive, and selective detection of small molecules without the need for chemical modification of the analyte. Our initial experiments were based on a previously published method10 using 0.1% formic acid as mobile phase and a Zorbax C18 column. However, the limit of detection (LOD) for adenine of this method was 2.4 ng/mL, which resulted in a low ricin LOD around 100 ng/mL. As the ricin detection limit relies primarily on the sensitivity for adenine, particular attention was devoted to the LC/MS optimization. Dissociation of Adenine. The mass spectrometer was operated in the positive ion mode. The most abundant ion was observed at m/z 136, which corresponds to MH+. The recorded CID spectrum of the m/z 136 ion displays the abundant product ion at m/z 119 (Figure 2A). This ion was assigned to elimination of ammonia. It was previously shown at high collision energy that this loss derives either from the exocyclic N-6 or from the ring N-1 nitrogen.25 Although it lowers signal intensity, the MRM detection mode, i.e., 136 f 119, was selected as it leads to a decrease in background noise and increased sensitivity. 15N2-1,3-adenine was used as internal standard. The spectrum of the m/z 138 ions is shown in Figure 2B and displays the principal product ions at m/z 120 and 121, corresponding to loss of labeled ammonia from the 15N-1 nitrogen and unlabeled ammonia from the exocyclic N-6 nitrogen, (24) Meng, Z.; Simmons-Willis, T. A.; Limbach, P. A. Biomol. Eng. 2004, 21, 1-13. (25) Nelson, C.C.; McCloskey, A. J. Am. Chem. Soc. 1992, 114, 3661-3668.
Analytical Chemistry, Vol. 79, No. 2, January 15, 2007
661
Figure 2. ESI-MS/MS spectra of (A) m/z 136 of unlabeled adenine added to a mixture of mobile phase A and B (50/50) and (B) m/z 138 of 15N -1,3-adenine added to a mixture of mobile phase A and B (E 2 lab ) 30 eV).
respectively. The product ion at m/z 120 was more abundant and selected for the MRM transition. The ion at m/z 109 observed in Figure 2A can be attributed to a loss of HCN which is released from the N-1 and/or N-3 position, i.e., loss of HC-2N-3 and/or HC-2N-1, as confirmed by the labeled adenine. Liquid Chromatography. Use of 0.1% formic acid or 0.1% acetic acid in the mobile phase resulted in adenine elution close to the void volume, even at low percentages of acetonitrile. Poor peak efficiency was also observed (Figure 3A). In order to make the uncharged species predominant during the chromatographic separation and hence increase the retention time, a mobile phase buffered above the pKa of adenine was tested, i.e., at pH ) 5.3. Moreover, increasing retention on column should allow addition of a higher percentage of organic modifier to the mobile phase, which is generally better for electrospray ionization/desorption conditions. Various analytical columns were evaluated in parallel in terms of chromatographic peak efficiency. The best result was obtained with the Atlantis C18 column showing the highest number of theoretical plate, i.e., 3150. This column was selected 662 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007
for the rest of this work. Figure 3B shows the adenine peak obtained with these conditions using acetonitrile increased to 50% in the mobile phase B. Following this optimization, the lower limit of detection for adenine was decreased to 50 pg/mL, which should allow a 50-fold increase in ricin detection sensitivity compared to previous work.10 Enzymatic Reaction. Ricin catalyzes the depurination of the first adenosine in the GAGA nucleotide sequence from the 28S ribosomal subunit. The ricin A chain also catalyzes the hydrolysis of short synthetic oligonucleotides provided that a base-paired stem and a GAGA tetraloop are present.11 In our experiments, a 14-mer substrate was used since it was previously reported to be the optimal length, showing the highest catalytic turnover when compared to stem-loop structures ranging from 10- to 18-mer.26 Ricin is known to depurinate both RNA and DNA fragments.27 A DNA substrate is easier to handle as it avoids the need to work in an RNAase-free environment, unlike an RNA substrate which (26) Chen, X.; Link, T. M.; Schramm, V. L. Biochemistry 1998, 37, 11605-11613. (27) Amukele, T. K.; Schramm, V. L. Biochemistry 2004, 43, 4913-22.
Figure 3. Influence of mobile phase pH on adenine retention by C18 columns. Chromatographic separation of adenine (A) on Zorbax RP C18 with 0.1% formic acid in the mobile phase and (B) on Atlantis C18 with 1 mM ammonium acetate in the mobile phase at pH ) 5.3. A gradient mobile phase was used as described in the Experimental Section.
can be damaged by the presence of contaminating RNAase. However, we observed a 5-fold higher catalytic turnover with the RNA substrate, whose concentration was optimized to 89 µM (10 times the Km value of ricin, i.e., 8.1 µM26). The buffer for in vitro depurination was selected from among various possibilities described in previous reports.9,10,26 Sodium acetate was finally chosen to reduce any possible interference during electrospray ionization. The optimum incubation time for maximum sensitivity was found to be 24 h. This delays the results of the assay, which can be prejudicial in the case of bioterrorist attack. However, a preliminary result can be given after 6 h, i.e., 4 h of incubation and two additional hours for sample processing, with a 3-fold lower sensitivity and confirmed 20 h later if more sensitivity is needed. Immunocapture of Ricin. Any other RIP present in an environmental sample can depurinate the GAGA sequence and lead to a false-positive result. Another possible source of interference is that RNA-degrading enzymes (RNAase) may be present in these samples. Both may degrade the GAGA sequence prior to the ricin reaction, which would led to a lowered measured toxin activity or even a false-negative result. Another point is the endogenous presence of free adenine in various food samples, e.g., in milk, which can be difficult to differentiate from the adenine released by ricin. For these reasons, we developed a specific extraction of the toxin in order to detect ricin specifically in complex samples. One of the most specific means of separating an analyte from a biological matrix is an immunoaffinity extraction with specific antibodies. Our strategy was to use monoclonal antibodies directed against the ricin B chain (RTB) to capture ricin on a solid support, e.g., beads, and perform the depurination without elution of ricin from the beads. This strategy has two possible advantages: (i) The first is that antibodies directed against the ricin B chain are
not able to neutralize the enzymatic activity. This could be the case with antibodies directed against the A chain as the latter possesses the catalytic site. Such problems, together with reduced assay sensitivity, have been encountered in botulinum toxin detection.28 (ii) The second advantage is that the method is specific to the entire toxin since the assay response requires that ricin possess both the B chain (antibody binding) and the A chain (enzymatic activity). The entire toxin can therefore be differentiated from the ricin A chain alone, which is far less toxic29 due to decreased cell adherence and internalization.30 Finally, use of magnetic beads provides a simple procedure for binding the antibody in a first step and then separating the captured analyte from the biological matrix using an external magnet. Moreover, magnetic beads allow concentration of the analyte from large sample volumes during the immunocapture. Beads can be resuspended in a lower dilution volume than the initial sample volume. Using a reasonable 500 µL sample volume, ricin concentrated on the antibody-coated beads was resuspended in 20 µL of depurination buffer. Figure 4 depicts the adenine peak obtained from 1 ng/mL of a spiked ricin sample analyzed with or without immunocapture. This result shows an approximately 20fold increase in sensitivity due to the concentration of the toxin from a 500 µL sample volume by the magnetic beads. Assay Sensitivity and Precision. To demonstrate that the release of adenine is related to the ricin concentration, a calibration curve was generated between 0.2 and 100 ng/mL ricin (Figure 5). The value at 0.2 ng/mL was estimated to be the lower limit of (28) Kalb, S. R.; Moura, H.; Boyer, A. E.; McWilliams, L. G.; Pirkle, J. L.; Barr, J. R. Anal. Biochem. 2006, 351, 84-92. (29) Fu, T.; Burbage, C.; Tagge, E. P.; Brothers, T.; Willingham, M. C.; Frankel, A. E. Int. J. Immunopharmac. 1996, 18, 685-692. (30) Simmons, B. M.; Stahl, P. D.; Russell, J. H. J. Biol. Chem. 1986, 17, 79121920.
Analytical Chemistry, Vol. 79, No. 2, January 15, 2007
663
Figure 4. Adenine signal from a sample spiked with 1 ng/mL ricin (A) without and (B) after immunoconcentration. Chromatographic separation was performed with a gradient mobile phase on Atlantis C18 as described in the Experimental Section. Table 1. Measured Adenine Concentration after Addition of 20 ng/mL Ricin to Three Environmental Matrices
Figure 5. Calibration curve of ricin in Hepes buffer from 0.2 to 100 ng/mL. Calibration was done by establishing a nonlinear regression with no weighting of the adenine/I.S. peak area ratio versus ricin concentration.
quantification as this value leads to a signal-to-noise ratio for the adenine peak equal of roughly 10. The extrapolated lower limit of detection was 0.1 ng/mL based on a signal-to-noise ratio of the adenine peak of at least 3. The nonlinear profile of the calibration curve was explained by saturation of binding sites on the antibodies for 100 ng/mL ricin. The extraction recovery of ricin at this concentration was determined as only 48% via measurement of adenine released by the toxin captured on the beads and in the supernatant. Optimization of the amount of beads added to the sample or utilization of a more concentrated antibody solution to coat the beads should improve recovery. Assay precision was measured by preparing and analyzing five different spiked ricin samples at 20 ng/mL. The CV was 10%, and this value shows that the method has an acceptable precision despite the three steps required, i.e., immunocapture of ricin, depurination of the RNA substrate, and LC/MS/MS analysis of adenine. Application to Spiked Environmental Samples. The newly developed method was applied to environmental matrices to 664 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007
origin of samples
measured adenine concentration (ng/mL), n ) 2
tap water bottled water milk
201 169 112
demonstrate its usefulness. Poisoning by ricin may occur via ingestion, inhalation of an aerosol, or parenteral injection. Drinking water is one of the possible ways of deliberate dissemination of ricin by the oral route.2 A more complex matrix, i.e., milk, was also chosen. Five hundred microliters of each was spiked with 20 ng/mL ricin and analyzed. The results are presented in Table 1. Adenine originating from the RNA depurination by ricin was detectable in all three spiked samples (tap and bottled water and milk) with a measurable signal well above the limit of detection of the assay. The adenine signal was significantly lower in the spiked milk sample. The most probable explanation is interference during the extraction of ricin from the matrix, i.e., at the immunocapture step. This point should be clarified in the future validation of the method in various environmental matrices. The negative control samples did not show an adenine signal originating from the matrix. CONCLUSION The presence of ricin in the environment in the event of a suspected bioterrorist incident should be confirmed using several analytical methods. Immunoassays are highly sensitive, and mass spectrometric methods are very specific but less sensitive. However, none can distinguish between functional and nonfunctional ricin. Our approach based on immunocapture by anti B
chain antibodies coupled to mass spectrometry determination of the release of adenine by the A chain allows sensitive and specific determination of the entire active toxin. This is the first method capable of specifically detecting functional ricin with a sensitivity similar to that of enzyme immunoassay and easily applicable to environmental samples. The assay requires 26 h, which may appear long in the event of a bioterrorism incident. However, a preliminary response can be given after 6 h with 3-fold lower sensitivity but still below the ng/mL scale. Means for improving assay speed and/or sensitivity must be considered. The first possibility is to work with a larger sample volume. Then, the catalytic efficiency of the depurination reaction by ricin could be increased by developing a modified RNA substrate. For instance,
introduction of a 2′-deoxyribonucleoside at the second position of the GAGA sequence, i.e., GdAGA, was shown to increase the catalytic constant.31 A last approach would be to optimize the sensitivity of adenine detection using improved analytical technologies, such as the use of nanoLC/MS. We demonstrated the feasibility of the approach using immunocapture and mass spectrometry. The method should now be validated in a separate study to establish more precisely the sensitivity and precision and also confirm the specificity when applied to environmental matrices or even clinical samples.
(31) Amukele, T. K.; Schramm, V. L. Biochemistry 2004, 43, 4913-4922.
AC061498B
Received for review August 11, 2006. Accepted October 30, 2006.
Analytical Chemistry, Vol. 79, No. 2, January 15, 2007
665