The Genoa 2005 Outbreak. Determination of Putative Palytoxin in

A new method for sensitive, specific, and direct determination of palytoxin is proposed herein. It is based on combination of reversed-phase liquid ...
0 downloads 0 Views 356KB Size
Anal. Chem. 2006, 78, 6153-6159

The Genoa 2005 Outbreak. Determination of Putative Palytoxin in Mediterranean Ostreopsis ovata by a New Liquid Chromatography Tandem Mass Spectrometry Method Patrizia Ciminiello,† Carmela Dell′Aversano,† Ernesto Fattorusso,*,† Martino Forino,† G. Silvana Magno,† Luciana Tartaglione,† Claudio Grillo,‡ and Nunzia Melchiorre‡

Downloaded by UNIV OF NEBRASKA-LINCOLN on August 23, 2015 | http://pubs.acs.org Publication Date (Web): August 5, 2006 | doi: 10.1021/ac060250j

Dipartimento di Chimica delle Sostanze Naturali, Universita` degli Studi di Napoli Federico II, via D. Montesano 49, 80131, Napoli, Italy, and Dipartimento La Spezia, Agenzia Regionale per la Protezione dell’Ambiente Ligure (ARPAL), via Fontevivo 21, 19125, La Spezia, Italy

A new method for sensitive, specific, and direct determination of palytoxin is proposed herein. It is based on combination of reversed-phase liquid chromatography with mass spectrometry (LC-MS). The new method was set up on a turbo ion spray-triple quadrupole MS instrument operating in selected ion monitoring (SIM) and multiple reaction monitoring (MRM) acquisition modes (positive ions). The minimum detection levels for matrixfree toxin on column were thus estimated from the data to be 200 and 125 pg in SIM and MRM modes, respectively. Spiking experiments before and after extraction allowed us to assess limits of detection and quantitation for palytoxin in matrix, accuracy, and intraday and interday reproducibility of the method. The developed method was decisive for the analysis of a plankton sample collected along Genoa coasts in July 2005 when respiratory illness in people exposed to marine aerosols occurred. It is suggested that putative palytoxin was the causative agent responsible for patients’ symptoms and demonstrated for the first time the presence of such a toxin in Italian waters. During summer 2005, symptoms of rhinorrea, cough, fever, bronchoconstriction with mild dyspnea, and wheezes were observed in about 200 people exposed to marine aerosols by recreational or working activities on the beach and promenade of Genoa, Italy (Figure 1). Conjunctivitis was also observed in some cases, and 20 people required extended hospitalization. During the event, weather conditions were quite stable, the seawater appeared clear and barely circulating and its temperature was ∼25 °C. At the same time, an environmental suffering involving mostly epibenthos, both sessile (cirripeds, bivalves, gastropods) and mobile (echinoderms, cephalopods, little fish), was observed. A careful look at the marine plankton brought to the light that an unusual proliferation of the tropical microalga Ostreopsis ovata (Figure 2) occurred along the investigated coastal * Corresponding author. E-mail: fattoru@unina.it. Phone: (+39) 081-678503. Fax: (+39) 081-678552. † Universita` degli Studi di Napoli Federico II. ‡ Agenzia Regionale per la Protezione dell'Ambiente Ligure (ARPAL). 10.1021/ac060250j CCC: $33.50 Published on Web 08/05/2006

© 2006 American Chemical Society

Figure 1. Map of Italy indicating the location of toxic outbreak and sampling site.

Figure 2. Appearance of O. ovata from the Gulf of Genoa seen by inverted microscope (250×).

areas during the toxic outbreak. Some Ostreopsis strains are regarded as the producing organisms of palytoxin (Figure 3) and its analogues, a complex polyether macromolecule and one of the most potent marine toxins so far known.1-5 Thus, the need arose (1) Moore, R. E.; Scheuer P. J. Science 1971, 172, 495-498. (2) Taniyama, S.; Arakawa, O.; Terada, M.; Nishio, S.; Takatani, T.; Mahamud, Y.; Noguchi, T. Toxicon 2003, 42, 29-33.

Analytical Chemistry, Vol. 78, No. 17, September 1, 2006 6153

Downloaded by UNIV OF NEBRASKA-LINCOLN on August 23, 2015 | http://pubs.acs.org Publication Date (Web): August 5, 2006 | doi: 10.1021/ac060250j

Figure 3. Structure of palytoxin.

to set up a method for detection of palytoxin to investigate whether the O. ovata blooming during the Genoa endemic disease was producing the toxin. Detection and quantitation of palytoxin in biological samples can be accomplished by both analytical means and biological assays.6 However, it is often necessary to use a combination of methods to confirm the presence of the toxin. The simplest way to detect palytoxin is the animal toxicity assay, which consists of injecting intraperitoneally into a mouse a contaminated sample and measuring the death time. Lethal potency is expressed as mouse units (MU) where 1 MU is the amount of toxin that kills a 20-g mouse in 4 h. Based on the reported LD50 value for palytoxin of 450 ng/kg, 1 MU is presumed to be 9 ng of palytoxin.7 Although the mouse responds to the injected palytoxin by exhibiting several characteristic symptoms prior to death, namely, sudden jerks and convulsions, the mouse bioassay is not able to unequivocally individuate the nature of the causative agent. Alternative assays take advantage of palytoxin functional properties and include in vitro cytotoxicity,8 delayed hemolysis,9 and antibody-based hemolysis neutralization10 tests. All these assays are extremely sensitive, but in most regulatory situations, positive results require further confirmation by instrumental methods. (3) Usami, M.; Satake, M.; Ishida, S.; Inoue, A.; Kan, Y.; Yasumoto, T. J. Am. Chem. Soc. 1995, 117, 5389-5390. (4) Ukena, T.; Satake, M.; Usami, M.; Oshima, Y.; Naoki, H.; Fujita, T.; Kan, Y.; Yasumoto, T. Biosci., Biotechnol. Biochem. 2001, 65, 2585-2588. (5) Ukena, T.; Satake, M.; Usami, M.; Oshima, Y.; Fujita, T.; Naoki, H.; Yasumoto, T. Rapid Commun. Mass Spectrom. 2002, 16, 2387-2393. (6) Tan, C. H.; Lau, C. O. In Seafood and Freshwater Toxins: Pharmacology, Physiology and Detection; Botana, L. M., Ed; Marcel Dekker Inc.: New York, 2000; pp 533-547. (7) Onuma, Y.; Satake, M.; Ukena, T.; Roux, J.; Chanteau, S.; Rasolofonirina, N.; Ratsimaloto, M.; Naoki, H.; Yasumoto, T. Toxicon 1999, 37, 55-65. (8) Yasumoto, T.; Fukui, M.; Sasaki, K.; Sugiyama, K. J. AOAC Int.. 1995, 78, 574-582. (9) Habermann, E.; Ahnert-Hilger, G.; Chatwal, G. S.; Beress, L. Biochim. Biophys. Acta 1981, 649, 481-486. (10) Bignami, G. S. Toxicon 1993, 31, 817-820.

6154

Analytical Chemistry, Vol. 78, No. 17, September 1, 2006

Liquid chromatography with electrospray ionization-tandem mass spectrometric detection (LC-ESI-MS/MS) has great potential for rapid, sensitive, and unambiguous identification of palytoxin in contaminated material. In fact, the capability of ESI to produce multiply charged molecules under mild conditions has accessed detection of a high MW compound such as palytoxin (C129H223N3O54) by extending the mass range for m/z-limited mass spectrometers. In this paper, we present a new approach to the analysis of palytoxin based on reversed-phase LC-ESI-MS/MS. The systematic investigation of MS and chromatographic parameters leading to an optimized analytical technique and its application to the analysis of the Mediterranean O. ovata is reported herein. The obtained results point to putative palytoxin as the causative agent of the Genoa outbreak. EXPERIMENTAL SECTION Chemicals. All organic solvents were of distilled-in-glass grade (Carlo Erba, Milan, Italy). Water was distilled and passed through a MilliQ water purification system (Millipore Ltd., Bedford, MA). Formic acid (95-97%, laboratory grade) and ammonium formate (AR grade) were purchased from Sigma Aldrich (Steinheim, Germany). Acetic acid (laboratory grade) was purchased from Carlo Erba (Milan, Italy). An analytical standard of palytoxin was purchased from Wako Chemicals GmbH (Neuss, Germany). Collection and Identification of the Plankton. The harmful algae bloom occurred in July 2005 along the rocky and indented coast of Genoa, Italy, which presents pebbly seabed colonized by macroalgae. The following representative samples were collected by operators of the Regional Environmental Protection Agency (ARPAL): (i) a seawater sample, collected at 50 cm below the water surface; (ii) a concentrated plankton sample, obtained by several horizontal net hauls using Apstein plankton net (20-µm mesh); (iii) samples of macroalgae (Rhodophyta, Clorophyta,

Downloaded by UNIV OF NEBRASKA-LINCOLN on August 23, 2015 | http://pubs.acs.org Publication Date (Web): August 5, 2006 | doi: 10.1021/ac060250j

Phaeophyta) covered by a plankton film, collected, and closed underwater in sampling bags (PBI type). The seawater sample was used to determine the qualitative and quantitative composition of phytoplankton. The phytocenoses were characterized after treatment with Lugol solution and sedimentation in Uthermo¨hl tubes.11Analysis on the whole sedimentation chamber with inverted microscope showed O. ovata to be the major species (1.8 × 106 cell L-1) present in the sample, together with a few cells of diatoms (Coscinodiscus spp.) and other potentially toxic dinoflagellates, namely, Coolia monotis, Prorocentrum lima, and Amphidinium sp. The concentrated plankton sample was preserved alive for further investigations and an aliquot was shipped to the Department of Natural Product Chemistry, University of Naples “Federico II”, for chemical analysis. The macroalgal samples were transferred to hermetic sealed vessels, and the sampling bags were rinsed with filtered (0.22µm mesh) seawater collected at the sampling site. The rinsing water was added to the vessels shaking for 5 min, and the cell suspension was filtered through a plankton net (20-µm mesh). The procedure was repeated several times in order to obtain a plankton pellet, which was suspended again in filtered seawater. An aliquot of such a sample was treated with Lugol solution; the epiphytic dinoflagellates were counted by inverted microscope and were identified as O. ovata (∼4 × 106 cells). A seawater sample was collected at the same sampling sites of the toxic outbreak in a period when no Ostreopsis spp. was observed (May 2006). Analysis with inverted microscope showed that the sample contained 3.1 × 108 cells L-1, mainly diatoms and dinoflagellates. Among diatoms, the major genera were Chaetoceros (C. brevis, C. compressus, C. costatus, etc.), Hemiaulus (H. hauckii, H. sinensis), and Pseudo-nitzschia (P. delicatissima, P. seriata). Potentially toxic dinoflagellates such as Alexandrium spp. and Dinophysis spp. were present. Aliquots of this sample, each containing 3.1 × 106 cells, were used in spiking experiments for determination of matrix effect, accuracy, and reproducibility of the developed method. Extraction. The concentrated plankton sample was centrifuged using a fixed-angle rotor (rmax 9.5 cm) at 5000 rpm for 20 min in order to separate pellet from seawater. The pellet was sonicated with 4 mL of a methanol/water (1:1, v/v) solution for 3 min in pulse mode, while cooling in an ice bath. The mixture was centrifuged at 5500 rpm for 30 min, the supernatant was decanted, and the pellet was washed twice with 3 mL of methanol/water (1:1, v/v). The extracts were combined, and the volume was adjusted to 10 mL with extracting solvent and 50 µL of formic acid in order to have a methanol/water (1:1, v/v) with 0.5% formic acid solution. The obtained mixture was analyzed directly by LCMS (5 µL injected). The seawater (500 mL) was added of an equal volume of methanol and extracted twice with 1-L aliquots of dichloromethane. The dichloromethane layers were evaporated to dryness, and the residue was dissolved in methanol (2 mL) and analyzed directly by LC-MS (5 µL injected). The aqueous layer was extracted twice with 500 mL of butanol. The butanol layer was evaporated to dryness, dissolved in 4 mL of methanol/ water (1:1, v/v) with 0.5% formic acid, and analyzed directly by LC-MS (5 µL injected). (11) Utermo ¨hl, H. Mitt. Int. Ver. Theor. Angew. Limnol. 1958, 9, 1-38.

Mouse Bioassay. Mouse lethality of both pellet and butanol extracts was tested by injecting intraperitoneally the hundredth part of each sample suspended in 1 mL of 1% Tween 60 solution. Female mice CD1 Swiss weighing 14-17 g were used. Liquid Chromatography-Mass Spectrometry. Mass spectrometric experiments were performed using a PE-Sciex (Concorde, ON, Canada) API-2000 triple-quadrupole mass spectrometer equipped with a Turbospray (TSI) source, coupled to an Agilent (Palo Alto, CA) model 1100 LC, which included a solvent reservoir, in-line degasser, binary pump, and refrigerated autosampler. Two different columns, namely, a 3-µm Hypersil C8 BDS (50 × 2.00 mm) and a 5-µm Gemini C18 (150 × 2.00 mm) column (Phenomenex, Torrance, CA) were employed. The following key parameters were investigated: (i) type of organic modifier (methanol or acetonitrile); (ii) buffer and acid character (ammonium formate, formic acid, or acetic acid); and (iii) acid concentration (10, 15, 20, 25, 30, 40, 50 mM). The final conditions recommended for routine operation are the following: the 5-µm Gemini C18 (150 × 2.00 mm) column maintained at room temperature and eluted at 0.2 mL min-1 with water (eluent A) and 95% acetonitrile/water (eluent B), both eluents containing 30 mM acetic acid; a gradient elution was required (20-100% B over 10 min and hold 4 min). MS detection was carried out in positive ion mode at unit resolution using a mass peak width of 0.7 ( 0.1 Th (FWHM). Selected ion monitoring (SIM) experiments were carried out using a turbogas temperature of 300 °C, an ion spray voltage of 5500 V, a declustering potential (DP) of 8 V, a focusing potential of 350 V, and an entrance potential of 11 V. A collision energy of 50 eV and a cell exit potential of 10 V were used in multiple reaction monitoring (MRM) experiments. The bicharged ion at m/z 1340 and the tri-charged ion at m/z 912 were monitored in SIM experiments (500 ms, dwell time). The following transitions m/z 1340 f 327 (DP ) 8 V) and m/z 912 f 327 (DP ) 50 V) were monitored in MRM experiments. A dwell time of 500 ms for each transition was used. The presence of okadaic acid (OA), dinophysistoxin-1 (DTX1), 13-desmethyl-C spirolide, azaspiracids (AZA-1, AZA-2, and AZA3), yessotoxin (YTX), and brevetoxins (PbTx-2 and PbTx-9) was investigated in the pellet extract and in the dichloromethane extract of seawater using the LC-MS method developed by Quilliam et al.12 The presence of paralytic shellfish poisoning (PSP) toxins and domoic acid was investigated in the pellet extract and in the butanol extract of seawater by hydrophilic interaction liquid chromatography-mass spectrometry (HILIC-MS).13,14 Dose Response and Minimum Detection Level. By using the above LC and MS conditions, the linearity of dose-responses was examined by injecting standard solutions of palytoxin in methanol/water (1:1, v/v) with 0.5% formic acid at four dose levels (2.7, 0.9, 0.3, and 0.1 µg mL-1). Calibration curves for matrix-free standards were generated in positive SIM (m/z 1340; dwell time, (12) Quilliam, M. A.; Hess, P.; Dell’Aversano, C. In Mycotoxins and phycotoxins in perspective at the turn of the millennium; deKoe, W. J., Samson, R. A., Van Egmond, H. P., Gilbert, J., Sabino, M., Eds.; deKoe, W. J.: Wageningen, The Netherlands, 2001; pp 383-391. (13) Dell’Aversano, C.; Hess, P.; Quilliam, M. A. J. Chromatogr., A. 2005, 1081, 190-201. (14) Ciminiello, P.; Dell′Aversano, C.; Fattorusso, E.; Forino, M.; Magno, G. S.; Tartaglione, L.; Quilliam, M. A.; Tubaro, A.; Poletti. R. Rapid Commun. Mass Spectrom. 2005, 19, 2030-2038.

Analytical Chemistry, Vol. 78, No. 17, September 1, 2006

6155

Downloaded by UNIV OF NEBRASKA-LINCOLN on August 23, 2015 | http://pubs.acs.org Publication Date (Web): August 5, 2006 | doi: 10.1021/ac060250j

Figure 4. Full-scan mass spectrum of palytoxin in positive ion mode, obtained by direct infusion of the toxin standard in a mixture of acetonitrile/ water (1:1, v/v) with 2 mM ammonium formate and 50 mM formic acid.

1000 ms) and MRM (m/z 1340 f 327; dwell time, 1000 ms) modes. To evaluate matrix effect, spiking experiments post-extraction were carried out. An aliquot of the seawater sample collected in May 2006 was centrifuged in order to separate pellet from seawater, which were extracted as reported above. After confirming the lack of palytoxin in the extracts by applying the LC-MS method, palytoxin-free pellet and butanol extracts were used to prepare matrix-matched standards at four levels of palytoxin (2.7, 0.9, 0.3, and 0.1 µg mL-1). Calibration curves for matrix-matched standards were generated in positive MRM (m/z 1340 f 327; dwell time, 1000 ms) mode. The average of triplicate measurements was used for plotting. Peak areas were used to express peak intensity. The minimum limits of detection (LOD) and quantitation (LOQ) for matrix-free and matrix-matched palytoxin were thus estimated from the data. 6156

Analytical Chemistry, Vol. 78, No. 17, September 1, 2006

Evaluation of Accuracy and Reproducibility. Ten aliquots of the palytoxin-free seawater sample were centrifuged, thus obtaining 10 pellet and 10 seawater samples, respectively. On the basis of the minimum quantitation level for matrix-matched palytoxin acquired in the dose-response experiment, each pellet and seawater sample was spiked with appropriate amounts of palytoxin (9 µg). Successively, five spiked pellet samples and five spiked seawater samples were separately extracted and analyzed in triplicate in the same day. Results were used to evaluate the intraday reproducibility in terms of relative standard deviation (RSD). To test interday reproducibility, five spiked pellet samples and five spiked seawater samples were separately extracted and analyzed in triplicate over five different validation days. Results were expressed in terms of RSD. In all the above experiments, LC-MS analyses were carried out in positive MRM mode (m/z 1340 f 327; dwell time, 1000 ms) and palytoxin was quantitatively

Downloaded by UNIV OF NEBRASKA-LINCOLN on August 23, 2015 | http://pubs.acs.org Publication Date (Web): August 5, 2006 | doi: 10.1021/ac060250j

Figure 5. LC-MS analyses of a 1 µg/mL standard solution of palytoxin in positive ion mode. SIM (a) was performed for ions at m/z 1340 [M + 2H]2+ and m/z 912 [M + 3H + 3H2O]3+. MRM experiment (b) was performed by selecting ion transitions consistent with the fragmentation behavior of palytoxin. Chromatographic conditions are reported in the Experimental Section.

determined in spiked samples by direct comparison to matrixmatched standard solutions of palytoxin at similar concentrations injected under the same experimental conditions. On the basis of the obtained results, accuracy of the method was assessed in terms of recovery (extraction efficiency), expressed by [(mean observed concentration)/(spiked concentration)] × 100. RESULTS AND DISCUSSION Optimization of Mass Spectrometry. The limited data concerning the ESI ionization behavior of palytoxin6 required a long and complex optimization work to enhance ESI-MS parameters. Figure 4 shows a full-scan mass spectrum (FS-MS) obtained by direct infusion of the toxin standard in a mixture of acetonitrile/ water (1:1, v/v) with 2 mM ammonium formate and 50 mM formic acid. In the ionization conditions used, the FS-MS spectrum was dominated by the presence of the following: (i) a major cluster of ions in the mass range m/z 1240-1370; (ii) a cluster of ions in the range m/z 900-920; and (iii) a dominant ion at m/z 327. Palytoxin has a molecular mass of 2680 Da. Thus, the ions at m/z 1368, 1360, 1352, and 1340 could be assigned to the bicharged ions [M + K + NH4]2+, [M + K + H]2+, [M + Na + H]2+, and [M + 2H]2+, respectively. Multiple losses of water molecules (110 H2O) from the [M + 2H]2+ ion produced the ions at m/z 1332, 1323, 1314, 1305, 1296, 1287, 1278, 1269, 1260, and 1251. The ions at m/z 912, 906, and 900 could be assigned to the tricharged [M

+ 3H + 3H2O]3+, [M + 3H + 2H2O]3+, and [M + 3H + H2O]3+ ions, respectively. The ion at m/z 327 dominated the mass spectrum and could arise from the cleavage between carbons 8 and 9 of the toxin molecule15 and the additional loss of a molecule of water, which occurred in the ionization source under the conditions used. Decrease in declustering potential value allowed reduction of the fragment ion in favor of the tricharged and bicharged ions. Moreover, the use of an acidified mixture of acetonitrile/water (1:1, v/v), as infusion solvent of the toxin standard, allowed reduction in the presence of adduct ions in favor of the protonated bicharged and tricharged ions. The ions at m/z 1340, [M + 2H]2+, and m/z 912, [M + 3H + 3H2O]3+, were chosen for the SIM experiments. All source parameters (declustering potential, focusing potential, entrance potential) were further optimized by repeated flow injections of palytoxin standard via a Rheodyne valve injector. A temperature of 300 °C at the turbo ion spray source allowed us to achieve best sensitivity. The bicharged ion at m/z 1340 and the tricharged ion at m/z 912 were selected as precursor ions for MS/MS product ion scan experiments. The fragmentation pattern produced in the product ion spectra contained an intense ion at m/z 327, whatever the precursor ion used. Thus, the following transitions Q1 f Q3 at m/z values of 1340 f 327 and 912 f 327 were selected for MRM experiments, which permit better selectivity and better signal-tonoise ratios than SIM. The collision energy, the cell exit potential, and the collision gas settings were optimized to achieve best sensitivity. Optimization of Chromatography. To set up chromatographic conditions for determination of palytoxin, we first considered the LC-MS method by Lenoir et al.16 It was developed on a nano-ESI-QTOF mass spectrometer using a Hypersil ODS C18 column eluted with a mixture of water acidified to pH 2.5 with trifluoroacetic acid and pure acetonitrile. We tested the reported conditions on our LC-MS system, namely, a TSI-triple quadrupole MS instrument, but the intensity of the chromatographic peak was extremely poor for the transition at the m/z value of 1340 f 327, while no signal could be acquired for the transition at m/z 912 f 327. A second attempt to chromatographically resolve palytoxin was made by using the system proposed by Quilliam et al.;12 such a technique has been successfully employed for the comprehensive analysis of a wide range of phycotoxins and is proposed as a “universal” toxin analysis method. It uses a reversed-phase Hypersil C8 BDS column and a mobile phase containing ammonium formate and formic acid. In these conditions, most of phycotoxins are simultaneously separated using gradient elution with a high degree of sensitivity and selectivity. Thus, it appeared desirable to also extend the applicability of such a method to determination of palytoxin. Unfortunately, under the above chromatographic conditions, palytoxin response was very poor: the chromatographic peak was broad and showed significant tailing. A number of eluting systems were then tested on the Hypersil C8 BDS column. Particular attention was paid to character and (15) Uemura, D.; Hirata, Y.; Iwashita, T.; Naoki, H. Tetrahedron 1985, 41, 10071017. (16) Lenoir, S.; Ten-Hage, L.; Turquet, J.; Quod, J. P.; Bernard, C.; Hennion, M. C. J. Phycol. 2004, 40, 1042-1051.

Analytical Chemistry, Vol. 78, No. 17, September 1, 2006

6157

Downloaded by UNIV OF NEBRASKA-LINCOLN on August 23, 2015 | http://pubs.acs.org Publication Date (Web): August 5, 2006 | doi: 10.1021/ac060250j

Figure 6. Calibration curves for palytoxin dissolved in methanol/water (1:1, v/v) with 0.5% formic acid (matrix-free standard) and spiked in pellet and butanol extracts (matrix-matched standards). Each point is the mean of three replicate analyses in positive MRM mode (m/z 1340 f 327). LOD (S/N ) 3) and LOQ (S/N ) 10) for matrix-free and matrix-matched standards are reported.

Figure 7. LC-MS analyses in MRM positive ion mode of pellet (a) and butanol (b) extracts from a concentrated plankton sample collected along Genoa coasts in late July 2005 and dominated by O. ovata. The presence of putative palytoxin is demonstrated in both samples. Chromatographic conditions are reported in the Experimental Section.

percentage of organic modifier and buffer. Both acetonitrile and methanol were tested as possible organic modifiers. Acetonitrile was preferred since it provided a sharper peak than methanol and resulted in a shorter retention time for palytoxin. The use of an acidic mobile phase (pH 2.5-3.2) without addition of ammonium salts was preferred since it allowed us to improve signal intensity in positive ion mode. Both formic acid and acetic acid were tested at different concentrations in the range 10-50 mM. The best results in the signal-to-noise ratio for all the transitions monitored were achieved by using 20 mM formic acid. A different stationary phase, namely, a Gemini C18 column, was also considered and its performance compared to that of the Hypersil column. Optimization experiments regarding organic modifier and acid character and percentage were performed similarly as reported above. The Gemini C18 column provided better results than Hypersil C8 BDS in terms of both sensitivity and peak shape when using water as eluent A and acetonitrile/water (95:5, v/v) as eluent B, both containing 30 mM acetic acid. A gradient elution (20100% B over 10 min and hold 4 min) allowed determination of palytoxin in a 15-min analysis. SIM and MRM chromatograms obtained under the conditions for a 1 µg/mL standard solution of palytoxin are shown in Figure 5. Linearity of the Dose-Responses. The palytoxin standard was analyzed in triplicate at four concentration levels (2.7, 0.9, 0.3, and 0.1 µg mL-1) in both SIM (m/z 1340) and MRM (m/z 1340 f 327) modes. The minimum detection levels for matrixfree toxin on column were thus estimated from the data to be 6158 Analytical Chemistry, Vol. 78, No. 17, September 1, 2006

200 and 125 pg in SIM and MRM modes, respectively. Correlation coefficients of >0.9998 for both experiments indicated a high degree of linearity of the plots within the tested concentration range. To investigate the matrix effect, a seawater sample collected along Genoa coasts in a period when no Ostreopsis spp. was blooming in the sea was employed. The sample was extracted as reported in the Experimental Section, and the lack of palytoxin in both pellet and butanol extracts was assessed by applying the LC-MS method. The obtained pellet and butanol extracts were spiked at four dose levels (2.7, 0.9, 0.3, and 0.1 µg mL-1) with palytoxin. The possibility of ionization suppression or enhancement for palytoxin was evaluated by comparing the MRM response of palytoxin in pellet and butanol extracts to the MRM response of the analyte dissolved in pure methanol/water (1:1, v/v) with 0.5% formic acid, at the same concentrations. Calibration curves for matrix-free and matrix-matched standards were generated in positive MRM mode and are shown together with LOD (S/N ) 3) and LOQ (S/N ) 10) in Figure 6. Good linearity was observed in all cases. However, the slope of the curves for matrixmatched standards indicated a slight enhancement effect (6-8%) of the signal at low palytoxin concentration (0.1-0.3 µg mL-1) and a more significant suppression effect (14-20%) at palytoxin levels higher than 0.9 µg mL-1. This result indicates that the matrix effect over the tested concentration range is analyte concentration-dependent, which suggests that matrix-matched standards should be used for accurate quantitation.

Downloaded by UNIV OF NEBRASKA-LINCOLN on August 23, 2015 | http://pubs.acs.org Publication Date (Web): August 5, 2006 | doi: 10.1021/ac060250j

Evaluation of Accuracy and Intraday and Interday Reproducibility. To evaluate recovery (extraction efficiency), and intraday/interday reproducibility of the method, 10 aliquots of the palytoxin-free seawater sample were centrifuged in order to separate pellet from seawater. Each pellet and seawater sample was spiked with 9 µg of palytoxin before extraction. To evaluate the intraday reproducibility, five spiked pellet samples and five spiked seawater samples were separately extracted and analyzed in triplicate by LC-MS (MRM mode) in the same day. RSD values for such experiments were 3.2 and 8.5% for pellet and butanol extracts, respectively. Interday repeatability was tested using five spiked pellet samples and five spiked seawater samples, which were separately extracted and analyzed in triplicate over five different validation days. Good interday repeatability was indicated by RSD values of 4.2 and 9.6% for pellet and butanol extracts, respectively. The above spiking experiments also allowed us to evaluate the accuracy of the extraction method. Recoveries were estimated to be 91-98% for pellet extracts and 73-82% for butanol extracts with RSD values of