Detection of Intact Ricin in Crude and Purified Extracts from Castor

Jan 21, 2009 - Human Protection and Performance Division, Defence Science and Technology Organisation, 506 Lorimer Street, Fishermans Bend, Victoria ...
0 downloads 0 Views 156KB Size
Download PDF
Anal. Chem. 2009, 81, 1529–1535

Detection of Intact Ricin in Crude and Purified Extracts from Castor Beans Using Matrix-Assisted Laser Desorption Ionization Mass Spectrometry Craig S. Brinkworth,* Eloise J. Pigott, and David J. Bourne Human Protection and Performance Division, Defence Science and Technology Organisation, 506 Lorimer Street, Fishermans Bend, Victoria 3207, Australia Ricin is a highly toxic protein from the seeds of the castor bean plant. Crude extracts from castor beans are toxic by several routes, and there is international concern about the use of these extracts by terrorist organizations. Lethality in aerosolized form has spurred the development of methods for the rapid detection of this protein from air samples that is critical in determining the illicit use of this material. Matrix-assisted laser desorption ionization (MALDI) mass measurement with an automated laser firing sequence was used to detect intact ricin from solutions containing less than 4 µg/mL of ricin in the presence of other endogenous seed proteins. This sensitivity was attained with the addition of 0.01% Tween 80 to the extracts that greatly enhanced the ricin signal. Importantly, this treatment substantially reduces the interference from the castor bean seed storage proteins. Commonly the ricin signal can be completely obscured by the oligomers of seed storage proteins, and this treatment reveals the ricin molecular ion, allowing the analyst to make a judgment as to the ricin content of the extract. This method provides for sensitive and rapid identification of intact ricin from aqueous samples with little sample preparation and is amenable to automatic acquisition. Aerosolized biological warfare agents are considered to pose a threat because of their indiscriminate nature and the widespread panic that could occur after a dissemination event. To warn against spread of these biological agents, air monitoring is commonly undertaken to extract these agents for detection.1 Ricin is a toxin produced by the castor bean plant. It gained notoriety as the suspected poison used to murder an Ukranian secret service agent administered via injection by an umbrella tip.2 However, ricin can also be lethal when administered via inhalation of the aerosolized protein. Analytical tests are required to interrogate extracted air samples to determine the presence of ricin. By far the most common methods of detection for ricin are immuno based assays because of their sensitivity and utility under aqueous conditions.3 This technology has been adapted and

optimized for the detection of ricin in the field.4 These assays use antibodies that bind to sites on the ricin molecule and in many cases use fluorescence to detect this bound antibody. However, in practice these assays are not always unambiguous5 because of false positives from cross reactivity of other molecules with similar binding motifs. Thus it would be advantageous to use additional methods utilizing orthogonal technologies to confirm the presence of ricin. The use of two (or more) disparate technologies is a requirement of some regulatory bodies for a positive identification of ricin. In addition, the ricin toxicity depends in part on the maintenance of the native structure, particularly the disulfide bond between the A and B-chains.6 Therefore it is important that methodologies are developed that analyze for the intact protein. Other methods, utilizing various technologies, have been reported for the detection of ricin.3 Mass spectrometry is one of these other technologies that have also been used to analyze ricin and investigate component proteins and peptides from the castor bean extracts. Commonly this analysis uses either of two approaches; detection of intact ricin or using LC/MS/MS to identify peptides generated by proteolytic digestion of ricin.7,8 Fredriksson and coworkers reported a matrix-assisted laser desorption ionization (MALDI) mass spectrum of intact crude ricin but no detection limit was reported.8 This is an ideal complementary technology to confirm the presence of ricin from field air extraction samples after positive results from immunoassays. One problem with using mass spectrometry as a confirmatory method is the relative lack of sensitivity of this technique compared to biological assays. To alleviate this problem we have developed a robust method of sample preparation that has more comparable detection limits. Importantly this methodology analyzes the intact ricin molecule unlike the other mass spectrometric analysis which uses proteolytic digestion of the ricin and detection of some of the resultant peptides. Samples containing an analyte that possessed both biological activity and the appropriate molecular weight (MALDI mass spectrometry) could be tentatively identified as the intact ricin toxin. In addition, protein MALDI mass

* To whom correspondence should be addressed. E-mail: craig.brinkworth@ dsto.defence.gov.au. Phone: + 61 3 9626 8421. Fax: + 61 3 9626 8342. (1) Campbell, J.; Francesconi, S.; Boyd, J.; Worth, L.; Moshier, T. Mil. Med. 1999, 164, 541–542. (2) Crompton, R.; Gall, D. Med. Leg. J. 1980, 48, 51–62. (3) Ler, S. G.; Lee, F. K.; Gopalakrishnakone, P. J. Chromatogr., A 2006, 1133, 1–12. 10.1021/ac802240f CCC: $40.75 Published 2009 by the American Chemical Society Published on Web 01/21/2009

(4) Petrovick, M. S.; Harper, J. D.; Nargi, F. E.; Schwoebel, E. D.; Hennessy, M. C.; Rider, T. H.; Hollis, M. A. Lincoln Lab. J. 2007, 17, 63–84. (5) Tate, J.; Ward, G. Clin. Biochem. Rev. 2004, 25, 105–120. (6) Lord, M. J.; Jolliffe, N. A.; Marsden, C. J.; Pateman, C. S.; Smith, D. C.; Spooner, R. A.; Watson, P. D.; Roberts, L. M. Toxicol. Rev. 2003, 22, 53– 64. (7) Ostin, A.; Bergstrom, T.; Fredriksson, S. A.; Nilsson, C. Anal. Chem. 2007, 79, 6271–6278. (8) Fredriksson, S. A.; Hulst, A. G.; Artursson, E.; de Jong, A. L.; Nilsson, C.; van Baar, B. L. Anal. Chem. 2005, 77, 1545–1555.

Analytical Chemistry, Vol. 81, No. 4, February 15, 2009

1529

spectrometry is quick and relatively simple to perform compared to the LC/MS/MS methodologies. Recently, we published results into the enhanced sensitivity that is observed for Bovine Serum Albumin (BSA) when biological detergents such as Tween 80 are present in the solution.9 This is particularly topical given that one of the potential aqueous buffers used in the collection of wet extracted air samples consists of a solution of 0.01% Tween 80. This same solution could also be easily added to dry air extraction samples. In this paper we present a sensitive, fast, and robust analytical method that provides complementary information to the biological assays for the detection of ricin toxin. EXPERIMENTAL SECTION Material. HPLC grade water and acetonitrile were from Merck (Darmstadt, Germany). Tween 80, Tween 20, Triton X-100, Triton X-114, purified ricin from Ricinus communis (castor bean) (now discontinued), and trifluoroacetic acid (TFA) were from Sigma (St. Louis, MO). Ricin is a very toxic protein and therefore requires appropriate safety measures. All containers containing more than a lethal dose were handled in a level 2 biosafety cabinet equipped with HEPA filters. Dilute working solutions containing less than a lethal dose were all handled in a fumecupboard. R-Cyano-4hydroxycinammic acid (HCCA), sinapinic acid (SA) and 2,5dihyroxybenzoic acid/2-hydroxy-5-methoxybenzoic acid (SDHB) were from Bruker Daltonik Gmblt (Leipzig, Germany) and were in excess of 99.6% pure. The mass spectra were acquired using a MTP 384 Ground Steel MALDI Target and an autoflex II TOF/ TOF from Bruker Daltonik Gmblt (Leipzig, Germany). Comparisons Between Detergents using Ricin with HCCA as the Matrix. Stock solutions of detergent were prepared in water containing 2% (v/v) of Tween 20, Tween 80, Triton X-100, and Triton X-114; 0.2%, 0.02%, and 0.002% solutions of each of the detergents were prepared by serial dilution of the stock solutions. A working solution of purified ricin (250 fmol/µL) was prepared in water by dilution of a ricin stock solution (3.4 mg/mL). The MALDI spots were prepared by mixing the detergent solution and ricin working solution in a 1:1 ratio for a total volume of 2 µL. This mixture was added to matrix solution (sat. HCCA in acetonitrile: 0.1% TFA, 33:67) in a 1:1 ratio for a total volume of 4 µL. The combined solution was mixed, spun, and 1 µL of the combined solution applied to the target and allowed to air-dry. For comparison the same procedure was performed using water instead of a detergent solution. MALDI mass spectra were acquired on a Bruker autoflex II TOF-TOF mass spectrometer using an automated laser firing sequence. Sensitivity Tests Using 0.01% Tween 80, Water, and Purified Ricin. A ricin solution (4 pmol/µL) in water was prepared from a stock solution of ricin (3.4 mg/µL). This solution was used to prepare two further diluted ricin solutions of 1 pmol/ µL in 0.01% Tween 80 and 1 pmol/µL in water. These solutions were subsequently serial diluted in 0.01% Tween 80 and in water to produce solutions of concentration 500 fmol/µL, 250 fmol/µL, 125 fmol/µL, 62.5 fmol/µL, and 31.25 fmol/µL. The spots were prepared by mixing ricin containing solution (2 µL) with matrix solution (sat. HCCA in acetonitrile: 0.1% TFA, 33:67) (2 µL). The (9) Brinkworth, C. S.; Bourne, D. J. Eur. J. Mass Spectrom. 2007, 13, 311– 319.

1530

Analytical Chemistry, Vol. 81, No. 4, February 15, 2009

combined mixture was briefly spun, and 1 µL of the mixture applied to the target and allowed to air-dry. MALDI mass spectra were acquired on a Bruker autoflex II TOF-TOF mass spectrometer using an automated laser firing sequence. Only the mass spectra from the following ricin concentrations are presented in the paper: 250 fmol/µL, 125 fmol/µL, 62.5 fmol/µL, 31.25 fmol/ µL in 0.01% Tween 80 and 1 pmol/µL, 500 fmol/µL, 250 fmol/µL in water. Production of the Aqueous Extract from Castor Bean Cultivars. Aqueous extracts containing the peptide/protein component of seeds of the castor bean plants (R. communis) were provided by DSTL (Porton Down, U.K.). The extracts were produced using the following protocol: the seeds were crushed and the mixture dissolved in acetone. This was repeated several times to ensure the removal of the castor oil. The wet mash was dried, and the solid suspended in 2% acetic acid and allowed to incubate at room temperature for several hours. The mixture was filtered producing a clear aqueous extract containing the peptide/ protein component of the seeds including ricin. For this work, the sanguineus and zanzibariensis strains were used. Molecular 30 kDa Cut off Fractions of the Crude Cultivars. An aqueous castor bean extract from the sanguineus cultivar was produced as described above. Aqueous extract (10 mL) was subjected to a 30 kDa Amicon Ultra Centrifugal Filter Device from Millipore (Billerica, MA, U.S.A) according to the manufacturer’s instructions to yield a filtrate containing components 20000 counts; B (italic), 10000-20000 counts; C, 1000-10000 counts; and NOP, no observable peak. The signal/noise ratio of the peak is indicated by the number in the box.

spots used to determine the effect of different concentrations and detergents on the formation of the oligomer in the gas phase were prepared by mixing 4 pmol/µL sanguineus 30 kDa cut off aqueous crude extract (1 µL) with one of the detergent solutions (1 µL), and then HCCA matrix solution (2 µL) was added. The solution was mixed, and 1 µL applied to a MALDI target and allowed to air-dry. Acquisition of Mass Spectral Data. MALDI mass spectra were acquired on an autoflex II TOF-TOF mass spectrometer from Bruker Daltonik (Leipzig, Germany). The spectra were acquired in linear mode using the following settings: Laser Power 30% (compared to 25-30% for a standard typtic digest), Laser Frequency: 25 Hz, Ion Source 1:20.00 kV, Ion Source 2: 17.50 kV, Lens: 8.50 kV, Pulsed Ion Extraction: 350 ns, Matrix Suppression: Deflection (600 m/z), m/z Detection: 8000-800000, Range: Medium, Sample Rate: 0.5 GS/s, Detector Gain: 19.0× (Turbo: On), High Mass Accelerator: 9.65 kV, Electronic Gain: Enhanced and Real Time Smooth: Off. An automated laser firing sequence was used to obtain the data that involved the accumulation of 3000 shots (100 shots per position) using a spiral pattern originating in the center of the spot. The shots were summed, and the resulting data smoothed using the following parameters: Algorithm, Savitzky Golay; Width, 100 m/z; and Cycles, 3. RESULTS AND DISCUSSION Detection Limits of Intact Ricin Only. Ricin was dissolved in the following detergents: Tween 20, Tween 80, Triton X-100, and Triton X-114 at final concentrations of 0.1%, 0.01%, and 0.001%. The final concentration of ricin in each solution was 125 fmol/ µL. These detergents were chosen because they had been shown to enhance the signal intensity of intact BSA MALDI mass spectrometry in previous work.9 The MALDI spots were prepared as described in the experimental section. The results for the +1 and +2 charge states for each solution are summarized in Table 1. With the exception of 0.001% Tween 80, all the other detergent solutions produced an observable signal for the +2 charge state of intact ricin. None of the 0.001% solutions of any of the detergents produced a signal for the +1 charge state. 0. 01% Tween 20, Tween 80, Trition X-100, and Triton X-114 produced the most intense signal intensity and highest S/N for the +1 and +2 charge state for each of the concentrations used. Tween 80 produced the most enhanced signal intensity of the four detergents at a concentration

of 0.01%, producing a signal for the +2 charge state with an absolute intensity above 20000 counts and a S/N of 398:1. The corresponding +1 charge state signal was also the most intense of the detergent concentrations (>10 000 counts and S/N of 120: 1). Tween 20 produced the next most intense peak followed by Triton X-100 and Triton X-114. All the signals representing the +1 charge states in the other detergents were comparable. At 0.1% of detergent concentration, the signal intensity decreased as compared to the 0.01% detergent spectra. Also, the S/N of the +2 charge state in each of the 0.1% spectra was about 80:1. Similarly the signals representing the +1 charge states in these spectra were less than those in the 0.01% counterparts. The signal intensity decreased further when the samples are prepared with 0.001% with the signal representing the +2 charge state in each of the spectra having an absolute intensity of no more than 10000 counts and a S/N of less than 50:1. No signal was observed for the +1 charge states at this concentration. In addition, no signal was observed representing any of the charge states of ricin when the spot was prepared in the absence of detergent. As previously stated this was also the case when 0.001% Tween 80 was used. These results are consistent with the trends that were observed with our experiments with BSA, demonstrating an optimal detergent concentration. Similar experiments were carried out using the matrixes sinapinic acid and SDHB (data not shown), but neither of these matrixes produced the enhancement in sensitivity of HCCA. For ricin, 0.01% Tween 80 produced the most significant signal enhancement for 62.5 fmol of ricin spotting with HCCA. To ascertain the effectiveness of inclusion of 0.01% Tween 80 in the analyte solution on ricin sensitivity, serial dilutions of ricin in 0.01% Tween 80 were prepared in the following concentrations: 250 fmol/µL, 125 fmol/µL, 62.5 fmol/µL, and 31.25 fmol/µL. For comparison purposes serial dilutions of ricin in water were also prepared in the following concentrations: 1 pmol/µL, 500 fmol/ µL, and 250 fmol/µL. The spots were prepared by mixing 2 µL of each of the ricin solution with 2 µL of the sat. HCCA matrix solution and spotting 1 µL onto the target and allowing to air-dry. The spectra obtained are shown in Figure 1. The spectra on the left were acquired from the solutions of ricin in 0.01% Tween 80 while those on the right are from solutions prepared from solutions of ricin in water. What is apparent from this figure is the difference in the achievable sensitivity using the two different solutionss0.01% Tween 80 and water. This is demonstrated by the fact that no observable signals for any of the charge states of ricin are observable at 125 fmol of spotted ricin in water while there is still an observable peak for the +2 charge state of ricin, albeit a weak one, for 15.63 fmol of spotted ricin in 0.01% Tween 80. At 31.25 fmol of spotted ricin in 0.01% Tween 80 the +1, +2, and +3 are observable, and the spectrum is not that dissimilar to the spectrum generated from 250 fmol of ricin in water. The enhanced sensitivity achieved by the addition of 0.01% Tween 80 to the analyte occurs because of its influence on the solubility of the protein during the drying process.9 The long alkyl chains dispersed with hydrophilic moieties increases the solubility of the ricin such that as the drop evaporates it remains in solution longer. This increased time in solution results in a more uniform distribution of the analyte rather than in sweet spots that are typical with larger proteins. The great advantage of spots with Analytical Chemistry, Vol. 81, No. 4, February 15, 2009

1531

Figure 1. Comparison between the MALDI mass spectra of intact ricin (62.5 fmol) prepared using 0.01% Tween80 and water spotted with HCCA. A 1:1 ratio of analyte including detergent/water: matrix was used for a total of 4 µL. Ricin prepared in 0.01% Tween 80 is shown on the left and that prepared in water is on the right. The charge states and the amounts of total ricin spotted are indicated on the figure.

these properties is the resilience of these spots to automated laser firing sequences. This advantage manifests itself in two ways: First, once the spots are prepared by the operator the mass spectra of each spot can be measured automatically saving time and effort, and second, this automation assists in making the spectra more uniform and robust. This second facet is vital if the method is to be used to determine the presence of ricin from field samples in the context of counter-terror operations. Detection of Intact Ricin in the Presence of Seed Storage Proteins. The illicit production of ricin involves a crude protein extraction from the castor bean seeds. This extraction involves the removal of the organic components of the seeds, including the castor oil, leaving an aqueous fraction containing ricin and other proteins/peptides. It has been reported that ricin accounts for about 1-5% of the total mass of the seeds.10 The presence of ricin in such abundance means that there is no need for further purification to make it a potentially useful terrorist target and threat. As a result, the expectation is that these other proteins would also be present in any field samples collected. One class of protein that are also produced are the seed storage proteins with molecular weights of around ∼11000 Da. Any mass spectrometric method that allowed for the identification of intact ricin in the presence of these proteins would be advantageous. The question is whether the method that we have developed for the detection of pure intact ricin is suitable for more realistic samples. The mass spectra of the MALDI intact mass measurement of an aqueous crude extract of a castor bean plant spotted with (10) Robertus, J. D. Cancer Treat. Res. 1988, 37, 11–24.

1532

Analytical Chemistry, Vol. 81, No. 4, February 15, 2009

Figure 2. Comparison between the MALDI mass spectra of intact crude aqueous zanzibariensis cultivar prepared using HCCA, SA, and SDHB as the matrix. A 1:1 ratio of analyte:matrix solution was used in the spot preparation. The oligomers resulting from the seed storage proteins (SSPs) are labeled on the figure.

HCCA, SA, and SDHB are shown in Figure 2. The spectra using HCCA and SA were acquired using an automated laser firing sequence, but the one with SDHB had to be acquired manually because of the properties adopted by the crystal during the drying processsthe spots consists of a series of “sweet” spots. The spectra acquired using HCCA and SA as the matrix are dominated by the singularly charged oligomers of the seed storage proteins, that is, signals at 11500 (monomer), 23000 (dimer), 34500 (trimer), 46000 (tetramer), 57500 (pentamer), and 69000 (hexamer) m/z. The peaks representing each oligomer decrease in intensity from the monomers but even the hexamers and larger can still be very intense or contribute substantially to the background. Also, the intensity of the HCCA mass spectrum is greater than the SA mass spectrum which is consistent with the trend that we have observed throughout these studies. This seed storage ionization pattern swamps the signals representing ricin making it impossible to detect the molecule. This pattern was reduced when SDHB was used as the matrix as demonstrated by the dimer, trimer, and tetramer peaks all being substantially less intense than in the previous spectra in relation to the monomer. However, the spectra overall produced peaks that were less intense than the other two matrixes (∼40000 counts compared to ∼120000 counts for SA and ∼140000 for HCCA) and less resolved. In addition, even with the reduction of the oligomer formation, it was still impossible to see the ricin peak. Even if the ricin peak had been observed this method of spectral acquisition would have been too labor intensive because of the need to locate sweet spots within the MALDI spots to generate the spectra. The presence of ricin in the sample was confirmed by 1D gel electrophoresis of the samples that had been

spotted. If the oligomer formation could be reduced for the matrix HCCA then it may be possible to detect the peak for ricin. To investigate this further a crude aqueous extract of castor bean was treated with a 30 kDa cut off molecular weight filter. This produced a sample with the ricin removed but still containing the seed storage proteins. The protein concentration of this solution was calculated to be 80 pmol/µL and was used in a series of experiments. Preliminary work had suggested that the addition of biological detergents such as Tween 80 to the analyte solution may reduce the formation of the oligomer in the mass spectrometer. Tween 20, Tween 80, Triton X-100, and Triton X-114 at concentrations of 1%, 0.1%, 0.01%, and 0.001% were tested to determine their effect on the oligomer formation when HCCA was used as the matrix (Supporting Information, Figure S-1). At concentrations of 0.01% and 0.001% none of the detergents had an effect on reducing the formation of the oligomers reliably. Conversely, at 0.1% all of the detergents substantially reduced the formation of the higher order oligomers, with most spectra only having peaks representing the monomer and dimer and occasionally a very small peak for the trimer. We believe that this is because the detergent solvates the individual monomers by hydrophobic and hydrophilic interactions between the detergent and the protein. The hydrophobic interactions occur between the long alkyl chains of the detergent and the hydrophobic residues of ricin while the hydrophilic interactions occur between the polar oxygen atoms scattered along the alkyl chains of the detergents and the polar amino acid in ricin. However, the peaks in these spectra were less intense and had poorer S/N than the previous spectra acquired at lower concentrations of detergents; this is consistent with our work using pure ricin. For any method to be useful for rapid detection of ricin it would need to be both sensitive and reduce the formation of the oligomers. In addition, at the higher 0.1% concentration, the preparation of the spots became more difficult as the reduced surface tension often meant that adjacent spots often ran into each other even for spots only containing a combined volume of 1 µL. Therefore, while 0.1% detergent reduced oligomer formation, these experimental limitations precluded the utilization of any of the 0.1% detergent solutions as suitable in a standard preparative technique for the robust analysis of samples potentially containing ricin, particularly using an automated laser firing sequence. Finally, as expected, none of the 1% detergent solutions produced interpretable mass spectra. Oligomer formation is also heavily dependent on analyte concentration. Dilutions of the 80 pmol/µL seed storage protein were prepared in water in the following concentrations: 20, 10, and 2.5 pmol/µL to determine its effect on oligomer formation. Briefly, 1 µL of 20, 10, or 2 pmol/µL of seed storage protein was mixed with 0.02% Tween 80, and the resulting mixture added to 2 µL of HCCA matrix solution. Two microliters were deposited on the target and allowed to air-dry. The resulting spectra are summarized in Figure 3. At 10 pmol in 0.01% Tween 80 the oligomer formation is substantially reduced with no sign of the pentamer or hexamer. Further reductions in the amount spotted resulted in further decreases in the formation of the higher order oligomers. At 1.25 pmol of spotted seed storage protein in the 0.01% Tween 80 the monomer is the dominant peak with only a

Figure 3. Comparison between the MALDI mass spectra of purified seed storage proteins from a castor bean extract at various amounts of spotted material prepared with HCCA as the matrix. A 1:1 ratio of analyte:matrix solution was used in the spot preparation. The oligomers resulting from the seed storage proteins (SSPs) and amounts are labeled on the figure. The asterisk indicates the peak that could be potentially ricin.

very small peak observed for the dimer. This result is most likely due to a combination of the reduced analyte concentration and the solvating properties of Tween 80 previously discussed. Interestingly, in the 10 and 5 pmol mass spectra, there was a peak at about 62 kDa (indicated by an asterisk). Coincidently, this is the same m/z ratio as purified ricin shown in Figure 1 suggesting that the cut off filters may be letting some of the ricin through and providing further evidence that this method is acceptable for ricin detection. These results suggest that the most useful way of reducing the formation of oligomers is to reduce the total amount of seed storage proteins spotted. Serendipitously, it appears that using this method has indicated that there may still be ricin present in the molecular cut off fractions that was not observable at higher concentrations of the seed storage proteins. Finally, we present our results into the efficiency of the process for the detection of ricin in the presence of the seed storage proteins. Studies suggest that ricin makes up 1-5% of the total weight of the bean.10 This makes it one of the most abundant proteins in an aqueous extract from a castor bean seed. In the worse case scenario, that would mean that there would be equal portions of seed storage proteins and ricin. To evaluate the situation of equal amounts of both ricin and seed storage proteins, both were mixed in 0.01% Tween 80 and the mixture spotted as previously described. Figure 4 shows the resulting intact protein MALDI mass spectra obtained from 1 µL spots containing a total of 500 fmol and 31 fmol of each analyte. At 500 fmol, the + 1 Analytical Chemistry, Vol. 81, No. 4, February 15, 2009

1533

Figure 4. Intact MALDI mass spectra of equal amounts (500 and 31 fmol) of seed storage proteins and purified ricin in 0.01% Tween 80 using HCCA as the matrix. The peaks are labeled on the top spectrum.

charge state of the seed storage proteins at around ∼11 kDa are easily observable, but none of the other singly charged oligomers are present as expected. Also the + 1, + 2, + 3, and + 4 charge states of ricin are easily identifiable too. The resolution of this spectrum is sufficient to enable differentiation of the singly charged trimer and dimer of the seed storage protein from the + 2 and +3 charge state of ricin, respectively. This same scenario is also repeated at 31 fmol of both analytes and with a reasonable S/N for the peaks involved. These spectra clearly demonstrate that it is possible to identify ricin in the presence of seed storage proteins if the concentration of the seed storage protein is sufficiently diluted. Does the dilution of a crude extract really enable identification of the protein ricin? To address this question the protein content of a crude zanzibariensis cultivar extract was determined using the GE Healthcare Quant Kit. The total protein concentration was determined to be 1.9 µg/µL. The SDS-PAGE separation of the crude zanzibariensis cultivar used in this study was recently reported.11 The gel suggests that for this particular extract the ricin constitutes about 40% of the total protein content. Therefore the ricin concentration of the extract used is 0.76 µg/µL or 1.2 pmol/µL (assuming the MW of ricin is 62 kDa). An aliquot of crude zanzibariensis cultivar was diluted in a 1:1 ratio with 0.02% Tween 80 solution producing a solution of the crude cultivar in 0.01% Tween 80. This solution was subsequently serial diluted seven times producing solutions that were 4, 8, 16, 32, 64, and 128 times more dilute than the original crude zanzibariensis extract. The MALDI spots were prepared by mixing 1 µL of the cultivar solution with 1 µL of sat. HCCA matrix solution and (11) Tran, H.; Leong, C.; Loke, W. K.; Dogovski, C.; Liu, C. Q. Toxicon 2008, 52, 582–588.

1534

Analytical Chemistry, Vol. 81, No. 4, February 15, 2009

Figure 5. Effect of serial dilution of a crude cultivar in 0.01% Tween 80 upon the observed mass spectra. The spots were prepared by mixing equal portions of the cultivar solution and HCCA matrix solution. Dilutions of the crude cultivar are shown on the figure.

depositing 1 µL onto the target. Figure 5 shows the spectra from the 2, 8, 32, and 128 times dilutions of the crude zanzibarensis cultivar. The two times diluted spot is dominated by the singly charged oligomers and has an almost identical appearance to the spectra in Figure 2. A small peak is observable at the m/z ratio of ricin, but it would be difficult to categorically identify this peak as ricin. At 8 times diluted the oligomer formation is already substantially reduced with almost no peak representing the pentamer or hexamer and that for the tetramer being much reduced in intensity. The increasing simplicity of the mass spectrum in the m/z range greater than 40000 Da means that the peak at m/z ratio ∼62000 Da representing ricin is much more easily identified. At 32 times dilution, the +2 charge state of ricin is now observable in addition to the +1 charge state because of the substantial reduction in the intensity of the peaks representing the singly charged dimer and trimer. Finally at a dilution of 128 times the original solution only the peak representing the singly charged monomer of the seed storage proteins is observed. In this spectrum the +1 and +2 charge states representing ricin are well resolved and can be easily identified. Therefore based upon the initial concentration of ricin in the crude extract (1.2 pmol/ µL) this spectrum represents 48 fmol of spotted ricin. Intact MALDI mass spectrometry shows great promise as a quick and robust method for the identification of ricin from infield air extraction samples. The samples can be spotted as delivered containing the aqueous buffer 0.01% Tween 80 commonly used for wet air extraction samples or the buffer added in the case of dry air extraction samples. Also, the only sample

preparation required is the addition of buffer if required and the dilution of the sample. Also, because each sample is deposited on discrete spots on the target there is no danger of cross contamination or carry-over between analyses. The method is sensitive and robust. In the case of wet air extraction samples gains in the overall detection limit could be achieved by preconcentration of the protein present in the sample prior to spotting. CONCLUSION MALDI mass measurement can be used to detect the biological warfare agent ricin as an intact molecule in both crude and purified castor bean extracts. This is achievable by ensuring the correct dilution of the seed storage proteins in 0.01% Tween 80 to suppress oligomer formation in the gas phase and enhance the sensitivity of the experiment. Using these measures it is possible to detect ricin with confidence down to 31 fmol. This translates into an actual concentration of 4 µg/mL for a wet extraction field sample. This has applications in the analysis of aqueous samples from field wet air sampling stations for the detection of this toxin and would complement the other immunoassay techniques currently employed to detect this molecule. It could easily be adapted to samples from dry extraction air sampling. The method is robust

and quick because the 0.01% Tween 80 can be added to the aqueous buffers before sampling and the only significant sample preparation prior to spot preparation is dilution. Also the acquisition of the data can be achieved through an automated laser firing sequence. ACKNOWLEDGMENT We thank DSTL (Porton Down, U.K.) for the crude aqueous extracts of the sanguineus and zanzibariensis castor bean seeds and also for seeds of the sanguineus castor bean plant. We thank Christina Athanassiou from DSTO (Fishermans Bend, VIC, Australia) for the 30 kDa molecular cut off fractions of crude aqueous sanguineus castor bean extract. SUPPORTING INFORMATION AVAILABLE Further details are given in Figure S-1. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review December 2, 2008.

October

22,

2008.

Accepted

AC802240F

Analytical Chemistry, Vol. 81, No. 4, February 15, 2009

1535