Determination of domoic acid in seawater and phytoplankton by liquid chromatography–tandem mass spectrometry

Journal of Chromatography A, 1163 (2007) 169–176

Determination of domoic acid in seawater and phytoplankton by liquid chromatography–tandem mass spectrometry Zhihong Wang ∗ , Kristen L. King, John S. Ramsdell, Gregory J. Doucette Marine Biotoxins Program, Center for Coastal Environmental Health & Biomolecular Research, NOAA/National Ocean Service, 219 Fort Johnson Road, Charleston, SC 29412, USA Received 7 February 2007; received in revised form 8 June 2007; accepted 18 June 2007 Available online 30 June 2007

Abstract Domoic acid (DA) is an algal neurotoxin produced by diatoms primarily of the genus Pseudo-nitzschia and is responsible for the human intoxication syndrome known as amnesic shellfish poisoning. A method has been developed to determine DA in seawater and phytoplankton matrices by liquid chromatography–tandem mass spectrometry for both quantitation and confirmation purposes. Sample extraction and clean-up was achieved on a C18 solid-phase extraction (SPE) cartridge. An acidic condition is critical for retaining hydrophilic DA on the cartridge. Direct injection of SPE eluate for analysis is recommended in order to avoid loss of DA by drying with heat prior to resuspension and injection. DA was quantified using the fragments produced from the protonated DA ion through multiple reaction monitoring (MRM). Recoveries exceeded 90% for all seawater samples spiked with DA and approximated 98% of toxin standard added to cultured phytoplankton material. Acceptable reproducibility (ca. 5% or less) was obtained for all intra-day and inter-day samples. The detection limit was 30 pg/ml level with a 20 ␮l injection volume, which demonstrated the value of this method for not only confirming DA production by minimally toxic phytoplankton species, but also for investigating the potentially important role of dissolved DA in marine food webs. Published by Elsevier B.V. Keywords: Domoic acid; Solid-phase extraction; LC–MS/MS; Seawater; Phytoplankton

1. Introduction Domoic acid (DA) (Fig. 1) is a neurotoxin produced by several species of phytoplankton, predominantly of the diatom genus Pseudo-nitzschia. This toxin enters food webs through feeding interactions and can accumulate in higher trophic levels [1,2]. Human consumption of DA-contaminated shellfish causes amnesic shellfish poisoning (ASP). DA has been shown to result in illness and deaths in humans and marine mammals [3,4]. A regulatory action level of 20 ␮g DA/g shellfish tissue is employed world wide for the harvesting and consumption of shellfish resources to protect human health. The phytoplankton species responsible for the first documented outbreak of ASP, which occurred in eastern Canada during 1987, was Pseudo-nitzschia multiseries (formerly known as Nitzschia pungens f. multiseries) [5]. Pseudo-nitzschia

∗

Corresponding author. Tel.: +1 843 762 8991; fax: +1 843 762 8737. E-mail address: [email protected] (Z. Wang).

0021-9673/$ – see front matter. Published by Elsevier B.V. doi:10.1016/j.chroma.2007.06.054

species are distributed widely throughout coastal waters of the world’s oceans, with some taxa showing DA production while others do not. The toxicity of Pseudo-nitzschia strains can vary considerably both between and within species [6,7]. Thus far, in addition to the rhodophycean macroalga observed originally to produce DA, diatoms from two other genera (Amphora, Nitzschia) have been reported as sources of this toxin [5]. Seawater and phytoplankton samples generally contain low levels of domoic acid. For field seawater samples, a high level of DA observed in surface water from the Cardigan River estuary in Canada was ca. 12 ng/ml [8] and other reported DA concentrations are generally less [9]. Levels of DA in laboratory Pseudo-nitzschia cultures have been reported to range from non-detectable to ca. 2500 ng/ml in the whole culture [7,8]. Several analytical methods are currently available for the quantitative determination of DA in seawater and in general they employ HPLC with fluorescence detection (FLD) [8,10,11]. The standard method is pre-column derivatization with 9-fluorenylmethylchloroformate (FMOC-Cl) followed by LC–FLD, yielding a detection limit of 15 pg/ml with a 20 ␮l

170

Z. Wang et al. / J. Chromatogr. A 1163 (2007) 169–176

Fig. 1. Chemical structure of domoic acid.

injection volume [8]. However, this technique has several disadvantages related to use of the FMOC-Cl reagent: significant interference due to reagent artifacts and dependence of derivatization efficiency on buffer concentration and sodium chloride level. In order to overcome the disadvantages associated with the FMOC-Cl reagent, derivatization with 6-aminoquinolylN-hydroxysuccinimidyl carbamate followed by LC–FLD was implemented with a detection limit of 0.001 ng DA on column [10]. Another HPLC–FLD approach is based on derivatization of the toxin with 4-fluoro-7-nitro-2,1,3-benzoxadiazole with a detection limit better than 1 ng/ml [11]. DA is identified based on the coincidence of LC retention time of the suspected chromatographic peaks with those of DA standard peaks; however, the suspected toxin peaks may represent compounds other than DA. An unambiguous method such as LC–mass spectrometry (LC–MS) or NMR must be used to confirm the presence of DA, especially for newly suspected source organisms [5,6]. LC–MS with electrospray ionization (ESI) interface has been demonstrated as the primary technology for the determination and unequivocal confirmation of DA or its isomers in shellfish [7,12,13]. In this paper, an LC–MS technique was developed for the determination and unequivocal confirmation of DA or its isomers in seawater and marine phytoplankton. It is known that an ESI interface on the mass spectrometer is susceptible to salt effects. Direct injection of high-salt samples often causes poor spray performance, decreases in instrumental sensitivity and dynamic range from ionization suppression in the ion source, and contamination of the mass spectrometer through the accumulation of salts from samples [14–17]. Seawater and marine phytoplankton samples contain a high concentration of salts. Published sample preparation methods for the quantitative extraction of DA from natural seawater and removal of salt interference for subsequent analysis are limited [18]. In reported HPLC–FLD methods, natural seawater samples are centrifuged or filtered to remove particles before derivatization with fluorophores in buffer and injection onto the HPLC [8,10,11]. Chan et al. [18] reported using sol–gel amorphous titania (TiO2 ) as a solid phase sorbent to pre-concentrate DA from seawater before derivatization for an HPLC–FLD method: amorphous titania sorbent was suspended in 30 ml buffered seawater to extract DA and adsorbed toxin then isolated from seawater by centrifugation prior to release with 0.5 ml of 0.1 M sodium borated buffer. Samples with high (50–200 mM) ionic strength usually require dilution prior to introduction to LC or eluting salt components to waste with

a switching valve before a mass spectrometer’s ion source to reduce the negative effect on the mass spectrometer from salts used as buffers in samples [17]. This buffer with a very high ionic strength (i.e., 0.1 M) used for desorption of DA from titania sorbent limits its direct injection to LC for MS analysis. Other sample clean-up methods for DA include using strong-anionexchange solid-phase extraction of DA from unsalted seafood samples [19] and direct elution of DA from C18 reversed-phase sorbent from shellfish extracts loaded with less than 15% acetonitrile/water without SPE pretreatment and SPE washing steps [20]. Reversed-phase SPE is a common way to remove salts, since the more hydrophobic the analyte, the stronger it binds to reversed-phase sorbents. DA has four chargeable groups: three carboxyl groups and one amino group (Fig. 1). The pKa values of DA were first reported as 2.10, 3.72, and 4.97 for the carboxyl groups and 9.82 for the amino group in 1958 [21] and those pKa values were widely cited [8,18,22,23]. Another set of pKa values were reported in 1992 with 1.85, 4.47, and 4.75 for the carboxyl groups and 10.60 for the amino group. The latter values were based on an NMR method and thought to provide accurate pKa values, while the pKa data published earlier were not supported by method and accuracy statements [24]. Based on the above pKa data, DA can possess charge states from −3 to 1 depending on pH. The pH of seawater and marine phytoplankton samples is about 8. Under these conditions, DA is hydrophilic because all carboxyl groups and amino group are charged, which makes it difficult for extraction by reversed-phase sorbents. Herein, we have developed a method to extract DA and remove salts from seawater and marine phytoplankton samples using a C18 reversed-phase sorbent, yielding samples that are directly compatible with LC–MS analysis. Conditions affecting the stability of DA were also investigated. 2. Experimental 2.1. Chemicals CRM-DA-e, domoic acid certified reference standard (99.4 ␮g/ml) was purchased from the National Research Council, Halifax, Canada. Acetonitrile, methanol, and water were HPLC grade (Burdick & Jackson, Muskegon, MI, USA). Acetic acid (≥99.7%, ACS reagent) was obtained from Aldrich (Milwaukee, WI, USA). Formic acid (Guaranteed Reagent, EMD brand, minimum 98%) was purchased from VWR International (West Chester, PA, USA). 2.2. Sample extraction and clean-up Natural seawater samples (salinity ca. 35) were filtered using a Durapore membrane (0.22 ␮m pore size; Millipore, Billerica, MA, USA). A 1.5 ml aliquot of filtered seawater sample was placed in a 13 mm × 100 mm disposable screw-cap glass test tube and acidified with 0.5 ml of 2% formic acid in 5% methanol/water (2:5:93, v/v/v) to yield 0.5% formic acid in the sample. After vortex mixing, the sample was desalted and

Z. Wang et al. / J. Chromatogr. A 1163 (2007) 169–176

extracted for DA by SPE using Bond Elut C18 LRC 10 ml columns (500 mg, endcapped, acid-washed silica, and 17% carbon loading; Varian, Harbor City, CA, USA). SPE columns were conditioned with 10 ml of methanol followed by 10 ml of HPLC grade water. The seawater sample solution was passed slowly through the SPE column drop by drop (flow rate ca. 1 ml/min) using a vacuum manifold, followed with 4 ml of 0.5% aqueous formic acid (0.5:99.5, v/v) as a rinse of the sample tube and the SPE column. The SPE column was washed further with 4 ml of 0.15% (0.15: 99.85, v/v) aqueous formic acid. DA adsorbed to the cartridge was eluted into a 5 ml glass centrifuge tube (Pyrex Brand, 13 mm × 135 mm, with graduations from 0 to 5 ml in 0.1 ml increments and pennyhead stoppers; Fisher Scientific, Suwanee, GA, USA) with 3 ml (slightly less than 3 ml from SPE to avoid volume exceeding 3.0 ml) of 50% of methanol/water (50:50, v/v). Additional 50% methanol/water was added carefully to the eluate centrifuge tube using a pasteur pipet (disposable, borosilicate; VWR, West Chester, PA, USA) to reach 3.00 ml line, as the eluant volume from the SPE column was about 0.1 or 0.2 ml less than 3.00 ml. The centrifuge tube was closed using the stopper and shaken to mix the solution to homogeneity. The solution in the centrifuge tube was then transferred to an LC vial for LC–MS analysis. The Pseudo-nitzschia culture samples were filtered through a Durapore membrane (0.45 ␮m pore size) to separate the cells from the medium. For the culture medium, a 1.5 ml aliquot was pipetted into a 13 mm × 100 mm screw-cap glass test tube for extraction of dissolved DA and clean-up as described above. The filter and accompanying cell pellet were placed in a 15 ml plastic conical centrifuge tube and 2.5 ml of filtered seawater were added to cover the filter, followed by sonication (1/8 inch micro tip probe, Branson Sonifier 450; Danbury, CT, USA) to disrupt the cells and release their DA into the seawater. After sonication, additional filtered seawater was added to the lysate to provide a total sample volume equivalent to the amount of culture filtered. A 1.5 ml of aliquot of the cell lysate solution was pipetted into a 13 mm × 100 mm screw-cap glass test tube for extraction of cell-derived (i.e., particulate) DA and clean-up. A whole culture sample (i.e., cells contained in their seawater growth medium) was also sonicated in a 50 ml conical centrifuge tube to disrupt the cells and release their DA into the seawater. A 1.5 ml aliquot of sonicated whole culture was pipetted into a 13 mm × 100 mm screw-cap glass test tube for extraction of total DA and cleanup. The extraction and clean-up of dissolved, particulate, and total DA in medium, cells, and whole culture with SPE, respectively, followed the same procedure as that outlined above for seawater. Two clones of Pseudo-nitzschia multiseries were used for development of the SPE method: clone CLN-19 (minimally toxic) and clone CLN-20 (highly toxic). Both clones were isolated by S. Bates (Department of Fisheries & Oceans Canada) in 2001 from a cross of clones CL-45 and CL-47, each originating from New London Bay, PEI, Canada. Cultures of each clone were grown at 20 ◦ C in f/2 + Si medium at an irradiance level of 90 ␮E/m2 /s provided on a 16:8 light:dark cycle [25]. Culture A was from clone 20 (36,800 cells/ml) and culture B from clone 19 (64 000 cell/ml).

171

2.3. SPE method development 2.3.1. Breakthrough volume of washing solvents and elution volume of eluting solvents DA (1 ␮g) dissolved in 2 ml SPE washing solvent was loaded on the conditioned cartridge. The cartridge was washed with the washing solvent (solvents tested including different amounts of formic acid or acetic acid in methanol/water or in water). The adsorbed toxin was eluted with 3 ml eluting solvents (solvents tested including different amounts of methanol or acetonitrile in water). Immediately after the loading of DA, the cartridge eluant was collected every 5 ml for the loading and washing solvents for a total of 15 ml, and an initial 1.5 ml followed by 3 × 0.5 ml for the eluting solvent. The cartridge eluate was analyzed directly by LC–MS/MS. 2.3.2. Determination of recoveries The determination of DA recovery from seawater was performed by adding DA followed by 1.5 ml of filtered seawater to a 13 mm × 100 mm disposable screw-cap glass test tube at five toxin concentration levels (0.1, 1, 10, 100, 1000 ng/ml seawater), each in triplicate. The extraction and clean-up of DA was performed according to the procedure described above. The eluant from filtered seawater (no DA present) treated by SPE according to the procedure described in Section 2.2 was used as a solvent to prepare DA calibration solutions. The determination of DA recovery from phytoplankton samples was performed by the following standard addition method: known amounts of DA were added to a 13 mm × 100 mm disposable screw-cap glass test tube followed by 1.5 ml of sonicated phytoplankton culture. The spiked culture was then extracted and cleaned up using the protocol outlined above. 2.3.3. DA standards prepared in solvents other than SPE eluate from DA-free seawater The DA calibration standards were prepared using the SPE eluate of filtered seawater. DA standards were also made in solvents with different percentage of acetonitrile or methanol in water for comparison of the effect of solvent matrix on DA determination by LC–MS. 2.3.4. Stability of DA during the drying procedure A 2 ml aliquot of DA solution taken from the 3 ml SPE eluate was transferred to a disposable 13 mm × 100 mm screw-cap glass test tube and evaporated using a turbo evaporator under nitrogen gas (Zymark, Hopkinton, MA, USA). The dried residue was re-solubilized by sonicating and vortex mixing in 2 ml of the same solvent used to elute the DA from SPE cartridges. The solution was analyzed by LC–MS and the results were compared to those of SPE eluates prepared without a drying step. 2.4. Liquid chromatography–mass spectrometry (LC–MS) LC–MS was performed on an HP1100 LC system (Agilent Technologies, Palo Alto, CA, USA) coupled to an Applied Biosystems/MDS Sciex API 4000 triple quadruple mass spectrometer equipped with a Turbo VTM source (Applied

172

Z. Wang et al. / J. Chromatogr. A 1163 (2007) 169–176

Biosystems, Foster City, CA, USA). LC separations were performed on a Luna C18(2) column (150 mm × 2 mm, 5 ␮m; Phenomenex, Torrance, CA, USA). Mobile phase consisted of water (A) and acetonitrile (B) in a binary system, with 0.1% formic acid as an additive. For confirmation of DA presence and determination of the relative amount of DA at each elution step of SPE method development, the LC was operated under isocratic conditions: 30% B for 3.5 min; the injection volume was 5 ␮l, with a flow rate 0.2 ml/min, and a column oven temperature of 40 ◦ C. For quantitative determination of DA in samples, the LC was operated using a gradient elution: 2 min of 5% B, linear gradient to 40% B at 15 min, 95% B at 17 min, hold for 5 min, then return to initial conditions at 23 min and hold for 5 min before next injection; the injection volume was 20 ␮l, with a flow rate of 0.2 ml/min, and a column oven temperature of 40 ◦ C. The detection of DA by mass spectrometer was achieved by multiple reaction monitoring (MRM) with Turboionspray interface in positive ion mode. Three MRM transitions from the protonated DA ion were monitored: m/z 312 → 266, m/z 312 → 248, and m/z 312 → 193. The ion spray (IS) voltage was 5 kV and the declustering potential (DP) was 69 V. The collision energy (CE) was 23 eV for the precursor/product pairs m/z 312 → 266 and m/z 312 → 248, and 26 eV for m/z 312 → 193. Nitrogen was used as the nebulizer gas (GS1, 45), turbo gas (GS2, 50 at 520 ◦ C), curtain gas (CUR, 20), and collision gas (CAD, 4). Quantitation was achieved using an external standard method due to unavailability of isotopically labeled DA as internal standards. DA calibration standard solutions included the following concentration levels: 0.05, 0.2, 0.5, 2, 5, 20, 50, 100, 200 ng/ml. 3. Results and discussion 3.1. Development of solid phase extraction procedure C18 reversed-phase sorbent is commonly used for desalting aqueous matrices. Domoic acid is hydrophilic due to its four chargeable groups (three carboxyl groups and one amino group). In order to retain DA on a C18 sorbent, acidification of the samples prior to loading and washing of the SPE under acidic conditions were used to partially protonate the carboxyl groups and enhance the hydrophobicity of the DA molecule. Without acidification of the samples, DA eluted out of the C18 SPE at low percentage of organic component (less than 15% acetonitrile [20]) immediately after the application of the samples. Formic acid and acetic acid were tested for their effect on DA extraction by C18 SPE (see Section 2.3.1 above). Low percentage of organic solvent (5%, 12% and 30% methanol) was also added to the washing solvent to test the breakthrough volume of washing solvent. A 1 ␮g amount of DA was retained on the SPE sorbent after washing (5 ml × 2) with either 2% or 1% formic acid in 5% methanol/water solvent, which included 2 ml of the same solvent used for loading DA on SPE cartridges. DA began to elute at the third 5 ml solvent wash with less than 1% (by peak area) DA detected. Two DA carboxyl groups are protonated in either 2% or 1% formic acid. A little more protonation of the third DA carboxyl group (next to the amino group with

the lowest pKa ) should occur in 2% formic acid solvent than 1% formic acid solvent; however, slightly more DA (less than 0.3% of total DA) was detected in the third 5 ml wash of the SPE column with 2% formic acid in 5% methanol/water, which may be due to the increase in organic solvent level by increasing the quantity of acid. DA eluted out of cartridge at the second 5 ml wash of the cartridge when methanol amount in the washing solvent increased to 12%; DA eluted with less than 5% at the first 5 ml wash of the cartridge and with majority of DA at the second 5 ml wash when methanol amount in the washing solvent increased to 30%. Breakthrough volume should be higher when no methanol is added to the washing solvent. When 2% formic acid was replaced by 2% acetic acid, no DA or a trace amount (less than 0.5%) was detected in the first 5 ml wash of the SPE cartridge, but DA started to elute with the second 5 ml solvent wash. Even though two DA carboxyl groups with higher pKa values were protonated in 2% acetic acid, the organic solvent component was also increased due to the greater carbon chain length of acetic acid compared to formic acid and thus the breakthrough volume was less for acetic acid. DA was not detected in the 15 ml solvent wash of the SPE column after further lowering the formic acid amount in the wash solvent to 0.5% and 0.1% in water. Water was inserted as one of the wash steps following the loading of DA under acidic conditions, yielding DA levels in the water fractions from non-detectable to above 50%. Therefore, acidic conditions are required to reproducibly retain DA on the C18 sorbent. Protonation of carboxyl groups of DA molecules is an equilibrium process in aqueous solution. Thus, trace amounts of DA (less than 1% of total DA) may elute from the SPE cartridge and become detectable during the washing step with increased loading of DA onto the cartridge. This loss will minimally affect the quantitation of DA, since the recovery of the toxin from SPE columns is well above 90% (see details below). The salinity and pH of natural open ocean seawater are about 35 and 8, respectively. Therefore, acidification of samples is required before loading onto a C18 SPE cartridge. The recovery of DA was not affected by the amount of formic acid in the loading sample nor the wash solvent at levels ranging from 0.1 to 2%. Seawater samples of 1 ml spiked with 0.15 ␮g of DA and acidified with formic acid at 2%, 0.5%, and 0.1% in 2.5% methanol, were loaded onto the SPE cartridges, followed by washing with 7 ml of 1%, 0.5%, and 0.1% formic acid in water, respectively. DA was eluted in 3 ml of 50% acetonitrile/water and then injected directly into the LC–MS for analysis, yielding recoveries all above 90%. Interestingly, it has been reported that DA shows decomposition in acidic shellfish extracts [26]. We did not observe decomposition or low recovery of DA based on LC–MS analysis repeated several weeks after the initial tests. Considering the potentially wide range of individual sample compositions, loading material onto the SPE cartridge in 0.5% formic acid, washing with 0.5%, and then 0.15% formic acid were adopted as standard procedures, thereby avoiding possible negative effects from low pH and assuring that the SPE eluate was suitable for DA bioassays. Elution of DA from C18 SPE columns was achieved by increasing the percentage of organic solvent without addition

Z. Wang et al. / J. Chromatogr. A 1163 (2007) 169–176

of acid. Greater than 90% of the total DA was eluted from the cartridge in 2 ml of 50% methanol/water or acetonitrile/water with a trace amount eluted after 2 ml of eluting solvent. Thus, the volume of DA elution solvent for SPE was set conservatively at 3 ml, resulting in a two-fold dilution of the original 1.5 ml sample. Reversed-phase polymeric sorbents (Strata X, 500 mg; Phenomenex) were also tested for extraction of DA from seawater, but did not show better extraction efficiency than C18 reversedphase sorbents for this toxin; therefore, C18 reversed-phase sorbent was used throughout this study. 3.2. Stability of DA during heating In order to concentrate the 3 ml of SPE eluate into a smaller volume to lower the method detection limit and re-dissolve DA in the same solvent (10%, acetonitrile/water) used for the NRC certified standard, the eluate was evaporated with nitrogen gas in a turbo evaporator. Recovery of DA from seawater was not reproducible with the drying of the SPE eluate. For recovery tests with the drying step included, a set of 15 seawater samples spiked with DA at five concentrations (triplicate), 70% of samples showed recovery above 90%, whereas substantially less than 90% of the toxin was recovered from the remaining samples. Similar experiments were repeated a second time, with all 15 samples showing recoveries between 50% and 80%. Therefore, the stability of DA during heating was examined (see Section 2.3.4). Six SPE eluates were tested at the same time: from 10% to 25% of the starting DA was lost during the drying step as compared to direct injection of the eluate into LC–MS. The loss of DA during solvent evaporation did not always occur; however, losses as high as 50% were observed. LC–MS running at full scan mode was not successful in identifying DA decomposition product(s) caused by heating. Reducing the drying temperature while increasing drying time did not eliminate the occasional, yet significant loss of DA. All DA spike experiments with seawater produced recoveries higher than 90% when injecting SPE eluate directly into the LC for analysis. As a result, turbo evaporation/heating cannot be recommended for concentrating DA in SPE eluates for the purpose of quantitation. 3.3. Mass spectrometry conditions Three fragments (m/z 193, m/z 248, and m/z 266) of the protonated DA ion at m/z 312 were selected for analysis by the MRM method. The major fragment at m/z 266 represents loss of HCOOH, the fragment at m/z 248 represents loss of HCOOH and water, and the fragment at m/z 193 represents loss of HCOOH and C2 H3 O2 N from the protonated DA ion. The MRM transition of m/z 312 → 266 was used for quantitation purposes, while the other two MRM transitions were used for further confirmation of DA presence. These two confirmation MRM transitions can also be used for quantitation when significant interference occurs for the MRM transition of m/z 312 → 266. Nonetheless, significant interference for the DA MRM transitions monitored in seawater and phytoplankton samples have yet to be observed, although severe interference for one or two DA MRM transitions

173

have been observed for body fluid samples taken from marine mammals (data not shown). For monitoring the three MRM transitions, compound-dependent parameters such as declustering potential and collision energy were adjusted using syringe pump infusion to achieve maximum intensity. Source/gas parameters were adjusted by flow injection and set to balance between maximum intensity and minimum contamination. The sensitivity of DA detection increased with the temperature of the turbo gas (TEM), since DA was eluted at high aqueous composition. The temperature of the turbo gas at a given setting varies slightly with the individual heater pair associated with the instrument. High temperature seems to increase the mass flux into the MS; however, the TEM was not set to the maximum permitted for the instrument in order to maintain MS signal stability and reduce contamination. The reproducibility of peak areas for repeat injections of one DA-spiked seawater sample (100 ng/ml) with SPE clean-up using the LC gradient method was 1.2% relative standard deviation (n = 6). 3.4. LC–MS chromatograms of DA To date, 10 isomers of DA have been reported [23]. The NRC certified DA reference standard used herein contains a mixture of DA and several of its isomers (epi-domoic acid, isodomoic acid A, D, and E). Isodomoic acid A, D, and E account for about 1.3% of the total concentration of DA and its isomers. Epi-DA, the C5 –diastereomer, can be formed spontaneously by the gradual isomerization of DA in solution (see specifications for NRC certified reference material). In this paper, all data represent the sum of DA and epi-DA. DA standard solutions prepared in various solvents showed different LC retention times and peak distributions with the same LC conditions (Fig. 2). The peak distributions and retention times were also affected by injection volume under isocratic conditions with 30% of B (data not shown). The LC mobile phase acetonitrile/water/formic acid used in our experiments is the common mobile phase for DA detection [12,26]. Employing LC gradient conditions (see Section 2.4), DA doublet peaks became significant with increasing acetonitrile in the DA injection solution, while the DA peak showed no discernable difference with the amount of methanol in the injection solution, compared to the NRC DA certified solution in 10% acetonitrile/water. DA isomers alone seem not to be able to account for the DA peak distribution observed in LC–MS chromatograms. In Fig. 2C, the ratio of the two major DA peaks with the injection of DA solution in 80% acetonitrile/water solvent does not match any concentration ratio of the DA and its isomer in the original NRC certified reference solution. Since DA is much less soluble in acetonitrile than methanol [19], DA might form clusters in solution that affected their LC peak distribution. Lack of individual standards of DA and its isomers makes it difficult to examine ratios among MRM transitions and factors affecting DA LC chromatographic resolution for differentiating DA and its isomers. The DA calibration standards were prepared using the SPE eluate of filtered seawater in order to have peak retention time and distribution match those of unknown samples in a seawater matrix. Methanol/water eluant was used to elute DA from SPE in order to have the DA peak distribution match

174

Z. Wang et al. / J. Chromatogr. A 1163 (2007) 169–176

Fig. 2. LC peak distribution of DA standard prepared in different solvents (gradient elution as described in Section 2.4): (A) 10% acetonitrile/water, (B) 60% acetonitrile/water, (C) 80% acetonitrile/water, (D) 50% methanol/water, (E) 80% methanol/water, (F) SPE eluate from DA blank seawater. DA peak eluted at about 12 min.

those commonly reported in previous publications [1,13,20,27]. The relative standard deviation of the DA peak area at same DA concentration in different solvents (10%, 60%, and 80% acetontile/water, 50% and 80% methanol/water, and eluate from C18 SPE of filtered seawater) was less than 6% for all three MRM transitions measured, which demonstrated that the DA peak areas did not show obvious changes in reproducibility with respect to the solvent used to prepare these standards. For routine DA analysis by LC-MS, either 10% acetonitrile/water or 50% methanol/water (the solvent used to extract DA from shellfish) solvents may be used to prepare the DA calibration standards for unknown sample analysis. An LC gradient elution with a high amount of organic component following DA elution was used and proved to be an effective column wash for removing carry-over from algal samples and maintaining analyte signal stability. A large injection volume (20 ␮l) could be used without peak band-broadening with 95% water as the initial component in the LC elution scheme. The sensitivity of the LC gradient method was enhanced slightly

with DA eluting at a high percentage of organic component as compared to the more common isocratic elution with 10% of B. 3.5. Analysis of DA in seawater and phytoplankton The accuracy of the method was evaluated for DA recovery efficiency by spiking DA-free natural seawater at five concentration levels each in triplicate, followed by SPE extraction of all samples. The mean recovery ranged from 95 to 104% and the precision determined as relative standard deviation (RSD) ranged from 1 to 6%, as summarized in Table 1. Inter-day precision and accuracy were determined by spiking DA-free natural seawater at a concentration of 100 ng/ml of DA on different days over a period of several months. The average recovery was 99% and relative standard deviation was 2% (n = 10). The calibration standard for the method was linear from 0.05 ng/ml up to about 400 ng/ml, with an r2 value of 0.9997 or better. The limit of detection (LOD) was 0.03 ng/ml of DA in seawater, representing a signal:noise ratio of 3. The limit of quantitation (LOQ)

Z. Wang et al. / J. Chromatogr. A 1163 (2007) 169–176 Table 1 Recovery of DA spiked in seawater (n = 3) Spiked concentration (ng/ml)

Measured concentration (ng/ml)

0.1 1 10 100 1000

0.098 0.98 9.6 99 1028

± ± ± ± ±

0.004 0.05 0.5 3 9

RSD (%)

Recovery (%)

4 6 5 3 1

98 98 96 99 103

was 0.1 ng/ml of DA in seawater at a signal:noise ratio of 10 or slightly higher. Two clones of Pseudo-nitzschia multiseries were used to examine the DA SPE extraction and clean-up method for phytoplankton. Dissolved DA in culture filtrate, particulate DA in cells, and total DA in whole cultures were analyzed separately as described in Section 2.2. Addition of DA standard (100 ng/ml) to sonicated whole culture was also used to evaluate the accuracy of the method. The results are summarized in Table 2. The two P. multiseries clones tested showed a distinct difference in DA production levels. Product ion spectra of the protonated DA ion (m/z 312) for these two cultures matched those published [7,13], further confirming their production of DA. The total DA obtained by summation of the dissolved and particulate toxin agreed very closely with that measured in the sonicated whole culture, showing a minimal difference of 3% for culture A and 4% for culture B. The calculated total DA in the whole culture, including the addition of 100 ng/ml of DA (i.e., the sum of DA measured in the whole culture without standard addition versus with addition of 100 ng/ml DA standard), also matched that measured directly with a difference of 1% for culture A and 2% for culture B; thus, the estimated recoveries were 98% of the DA added to both cultures A and B. The volume of filtrate or sonicated whole culture that can be applied to the SPE cartridge for clean-up is limited by the breakthrough volume, efficiency of desalting, and the tolerance of the mass spectrometer to salt. Signal stability was not maintained when washing of the SPE was insufficient to remove the salts; Intensity of the DA peak gradually decreased with the accumulation of salts at the surface of ion source. Analyte signal was lost immediately after several direct injections of samples containing a large amount of salts which affects the interface between the LC and the first scanning quadrupole. At severe condition, contamination was observed to penetrate into the first scanning quadrupole. The total volume of seawater and washing solvent that could be applied to the cartridge was 15 ml when samples and washing solvent was acidified with 2% formic acid in 5% methanol (Section 3.1). About 8 ml of washing solvent was required to remove the salt

175

from the cartridge with 1.5 ml of seawater sample loaded, in order to maintain the reproducibility of peak areas for repeat injections of a DA seawater sample for LC–MS analysis using the method described. However, cells collected on a filter can be disrupted in a volume of seawater much smaller than the original phytoplankton sample volume, which accommodates the analysis of particulate DA representing sample volumes far greater than the 1.5 ml recommended for this method. A switching valve before a MS to divert the salt components from LC eluant to waste may not be sufficient to remove the salt for maintaining MS signal stability as the ionic strength of seawater is much higher than 200 mM. However, with the use of a switching valve, the volume of a seawater or phytoplankton sample applied to the SPE may be increased and the volume of washing solvent may be reduced, thus the detection limit of the method may decrease. Most LC–MS instruments do not have switching valves installed. No studies have been conducted to determine if the detection limit is lowered through the use of a switching valve to remove the residue salt after SPE clean-up. The two confirmatory MRM transitions m/z 312 → 248 and m/z 312 → 193 are used to improve the specificity of the method and increase the confidence of DA identification when large numbers of samples are screened for DA confirmation and quantitation. A common method employed to identify the erroneous signals involves comparing the peak area ratios of the confirmatory MRM transitions for the unknown with those for the standards. The observed average ratio of the peak area between MRM transition m/z 312 → 266 and m/z 312 → 248 was 4.65, the ratio between the MRM transition m/z 312 → 266 and MRM transition m/z 312 → 193 was 5.53. The phytoplankton samples exhibited a lower ratio of 4.28 versus 4.65 for the m/z 312 → 248 transition. The observed ratios for both confirmatory MRM transitions in the two clones of Pseudo-nitzschia multiseries tested were within ± 20% of the average value. The wide distribution of the ratios was due largely to the difficulty in accurately measuring peak areas of the two confirmatory MRM transitions at DA concentrations below 0.2 ng/ml. The DA ion MRM ratios also showed fluctuation (within 10%) especially in long-term scale, which may be due to variation in instrumental status (e.g., change in lens voltages associated with instrumental calibration and in the level of nitrogen supply for interface and collision gas). When acetic acid was used instead of formic acid as the mobile phase additive, more fluctuation was observed for DA ion MRM ratios over extended time periods; the conversion of DA ions from liquid phase to gas phase at higher pH of the mobile phase with acetic acid additive may be more susceptible to fluctuation of LC-MS interface condition. Using formic acid as the additive yielded less matrix problems and higher analyte recovery in complicated sample matrices (e.g., animal fluids;

Table 2 Analysis of DA in algal culture Sample

Culture A Culture B

Measured concentration (ng/ml), mean ± SD, n = 3 Filtrate

Cell

Whole culture

Whole culture with DA 100 ng/ml added

12.2 ± 0.1 1.24 ± 0.09

107 ± 2 1.11 ± 0.08

123 ± 3 2.27 ± 0.08

221 ± 2 101 ± 9

176

Z. Wang et al. / J. Chromatogr. A 1163 (2007) 169–176

data not shown); therefore, formic acid was adopted as the additive in the LC mobile phase for DA analysis. Since no significant interference was observed for seawater and phytoplankton samples, the calculated DA concentration from MRM transition m/z 312 → 266 was used as the reported DA concentration and the confirmatory transition ion ratios for the unknown were set to ±20% of those for the DA standards run in the same batch. When the confirmatory transition ion ratios for the unknown samples exceed ±20% of ratios for the standards, MRM transitions other than m/z 312 → 266 should be used for DA quantitation or more MRM transitions other than the three transitions used herein the paper should be added for more accurate DA quantitation. 4. Conclusions A C18 reversed phase, solid phase extraction method has been developed for extraction and clean-up of domoic acid in seawater and phytoplankton sample matrices compatible with LC–MS analysis for the purposes of both highly sensitive quantitation as well as unambiguous confirmation of the presence of DA or its isomers. An acidic condition is the critical feature required to retain this hydrophilic toxin on C18 SPE sorbent for extraction and clean-up. Drying the SPE eluate in a turbo evaporator with heating appears to cause inconsistent and potentially substantial losses of DA. The protocol combined solid phase extraction for sample preparation and LC–MS/MS for analyte determination, resulting in DA detection and confirmation at pg/ml levels. Such a sensitive method will serve as an important tool for investigating the unknown, but potentially important role played by dissolved DA in marine food webs [2]. Moreover, for phytoplankton species that are unknown producers of DA, the presence of this toxin at low concentrations can be confirmed further by the full scan product ion spectra after positive results are obtained using this protocol. Acknowledgements The authors wish to express their thanks to Dr. S. Bates (Department of Fisheries & Oceans Canada, Moncton, NB, Canada) for providing the clones of Pseudo-nitzschia multiseries clones CLN-19 and CLN-20. Disclaimer: This publication does not constitute an endorsement of any commercial product or intend to be an opinion beyond scientific or other results obtained by the National Oceanic and Atmospheric Administration (NOAA). No reference shall be made to NOAA, or this publication furnished by NOAA, to any advertising or sales promotion which would indicate or imply that NOAA recommends or endorses any pro-

prietary product mentioned herein, or which has as its purpose an interest to cause the advertised product to be used or purchased because of this publication. References [1] P.R. Costa, R. Rosa, A. Duarte-Silva, V. Brotas, M.A.M. Sampayo, Aquat. Toxicol. 74 (2005) 82. [2] G.J. Doucette, I. Maneiro, I. Riveiro, C. Svensen, in: E. Gran´eli, J.T. Turner (Eds.), Ecology of Harmful Algae, Springer-Verlag, Heidelberg, 2006, p. 283. [3] T.M. Perl, L. Bedard, T. Kosatsky, J.C. Hockin, E.C. Todd, R.S. Remis, N. Engl. J. Med. 322 (1990) 1775. [4] C.A. Scholin, F. Gulland, G.J. Doucette, S. Benson, M. Busman, F.P. Chavez, J. Cordaro, R. Delong, A.D. Vogelaere, J. Harvey, M. Haulena, K. Lefebvre, T. Lipscomb, S. Loscutoff, L.J. Lowenstine, R. Martin III, P.E. Miller, W.A. Mclellan, P.D.R. Moeller, C.L. Powell, T. Rowles, P. Silvagni, M. Silver, T. Spraker, V. Trainer, F.M. Van Dolah, Nature 403 (2000) 80. [5] S.S. Bates, J. Phycol. 36 (2000) 978. [6] C.K. Cusack, S.S. Bates, M.A. Quilliam, J.W. Patching, R. Raine, J. Phycol. 38 (2002) 1106. [7] P. Ciminiello, C. Dell’Aversano, E. Fattorusso, M. Forino, G.S. Magno, L. Tartaglione, M.A. Quilliam, A. Tubaro, R. Poletti, Rapid Commun. Mass Spectrom. 19 (2005) 2030. [8] R. Pocklington, J.E. Milley, S.S. Bates, C.J. Bird, A.S.W. de Freitas, M.A. Quilliam, Int. J. Environ. Anal. Chem. 38 (1990) 351. [9] L.B. Busse, E.L. Venrick, R. Antrobus, P.E. Miller, V. Vigilant, M.W. Silver, C. Mengelt, L. Mydlarz, B.B. Prezelin, Harmful Algae 5 (2006) 91. [10] T. Sun, W.H. Wong, J. Agric. Food Chem. 47 (1999) 4678. [11] K.J. James, M. Gillman, M. Lehane, A. Gago-Martinez, J. Chromatogr. A 871 (2000) 1. [12] J.F. Lawrence, B.P. Lau, C. Cleroux, D. Lewis, J. Chromatogr. A 659 (1994) 119. [13] A. Furey, M. Lehane, M. Gillman, P. Femandez-Puente, K.J. James, J. Chromatogr. A 938 (2001) 167. ˚ Gustavsson, J. Samskog, K.E. Markides, B. Langstr¨om, J. Chromatogr. [14] S.A. A 937 (2001) 41. [15] S. Canarelli, I. Fisch, R. Freitag, J. Chromatogr. A 948 (2002) 139. [16] P. Kebarle, L. Tang, Anal. Chem. 65 (1993) 972A. [17] D.T. Rossi, in: D.T. Rossi, M.W. Sinz (Eds.), Mass Spectrometry in Drug Discovery, Marcel Dekker, New York and Basel, 2002, p. 172. [18] I.O.M. Chan, V.W.H. Tsang, K.K. Chu, S.K. Leung, M.H.W. Lam, T.C. Lau, P.K.S. Lam, R.S.S. Wu, Anal. Chim. Acta 583 (2007) 111. [19] M.A. Quilliam, M. Xie, W.R. Hardstaff, J. AOAC Int. 78 (1995) 543. [20] P.T. Holland, P. Mcnabb, A.I. Selwood, T. Neil, J. AOAC Int. 86 (2003) 1095. [21] T. Takemoto, K. Daigo, Chem. Pharm. Bull. 6 (1958) 578. [22] N. Pi˜neiro, J.M. Le˜ao, A. Gago Mart´ınez, J.A. Ridr´ıguez V´azquez, J. Chromatogr. A 847 (1999) 223. [23] B. Jeffery, T. Barlow, K. Moizer, S. Paul, C. Boyle, Food Chem. Toxicol. 42 (2004) 545. [24] J.A. Walter, D.M. Leek, M. Falk, Can. J. Chem. 70 (1992) 1156. [25] R.R.L. Guillard, in: W.L. Smith, M.H. Chanley (Eds.), Culture of Marine Invertebrate Animals, Plenum Press, New York, 1975, p. 29. [26] P. Vale, M.A.M. Sampayo, Harmful Algae 1 (2002) 127. [27] A. L´opz-Rivera, B.A. Su´arez-Isla, P.P. Eilers, C.G. Beaudry, S. Hall, M. Fern´andez Amandi, A. Furey, K.J. James, Anal. Bioanal. Chem. 381 (2005) 1540.