Simple and rapid determination of anatoxin-a in lake water and fish muscle tissue by liquid-chromatography–tandem mass spectrometry

Journal of Chromatography A, 1122 (2006) 180–185

Simple and rapid determination of anatoxin-a in lake water and fish muscle tissue by liquid-chromatography–tandem mass spectrometry Sara Bogialli a , Milena Bruno b , Roberta Curini a , Antonio Di Corcia a,∗ , Aldo Lagan`a a a

b

Dipartimento di Chimica, Universit`a “La Sapienza”, Piazza Aldo Moro 5, 00185 Roma, Italy Dipartimento d’Igiene Ambientale, Istituto Superiore di Sanit`a, Viale Regina Elena 299, 00161 Roma, Italy Received 17 January 2006; received in revised form 20 April 2006; accepted 25 April 2006 Available online 12 May 2006

Abstract Anatoxin-a (AN) is a powerful neurotoxin that can be produced by cyanobacteria in eutrophic freshwaters. Consequently, AN can contaminate lakes, rivers and basins destined for drinking water and aquaculture. Two simple, specific and sensitive procedures for determining AN in lake water and fish muscle tissue are presented. Both analytical protocols are based on liquid-chromatography (LC)–tandem mass spectrometry (MS) with electrospray ionization. MS data were acquired in the multi reaction monitoring mode by selecting four precursor to product ion transitions. After filtration, AN in lake water was analyzed by directly injecting 0.5 ml of the aqueous sample in the LC column. Analysis of AN in fish muscle tissue involved the matrix solid-phase dispersion technique. The analyte was extracted from tissue by 4 ml of water acidified to pH 2 and heated at 80 ◦ C. After acidification and filtration, 0.2 ml of the aqueous extract was injected in the LC column. Analyte recovery ranged between 71 and 79% and was not substantially affected by both the analyte concentration and the type of fish. Phenylalanine is an essential amino acid invariably present in any animal tissue. Like AN, this amino acid produces a pseudo molecular ion at m/z 166, it has a very similar fragmentation pattern and LC retention time. This method is able to prevent identifying phenylalanine for AN as the latter compound is eluted more than 1 min before the former one and the two compounds have remarkably different relative ion signal intensities. On the basis of a signal-to-noise ratio of 10, limits of quantification of AN in water and fish fillet were estimated to be 13 ng/l and 0.5 ng/g, respectively. © 2006 Elsevier B.V. All rights reserved. Keywords: Cyanobacteria; Anatoxin-a; Lake water; Fish tissue; Liquid-chromatography–tandem mass spectrometry

1. Introduction Heavy blooms of cyanobacteria (blue–green algae) are one of the consequences of the worldwide trend for increasing eutrophication in many waters. This phenomenon is thought to result from increased exogenous nutrient loading. As a consequence, the frequency of cyanobacterial blooms in freshwater has dramatically increased throughout the world [1]. It is estimated that 50% of cyanobacterial blooms are toxic, producing hepatotoxins, neurotoxins and lipopolysaccharide endotoxins [2]. Anatoxin-a (AN) was the first cyanotoxin to be structurally elucidated and can be produced by Anabaena flos-aquae, Aphanizomenon flos-aquae and Oscillatoria sp. Very recently, Phormidium favosum has been identified as a new AN pro-

∗

Corresponding author. Tel.: +39 06 49913752; fax: +39 06 49913680. E-mail address: [email protected] (A. Di Corcia).

0021-9673/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2006.04.064

ducing species [2]. AN has a high toxicity (LD50 i.p. mouse 0.2 mg/kg) and it is a potent nicotinic agonist that acts as a postsynaptic, depolarizing, neuromuscular blocking agent. AN may contaminate freshwaters and has been associated with a number of animal fatalities, including cattle and dogs [3–6]. Several analytical protocols have been proposed for analyzing AN in real water samples. These methods involve liquid–liquid extraction [7], solid-phase extraction (SPE) cartridges filled with C-18 bonded silica [8,9] or a weak cation exchanger [10,11], disk type SPE after pH adjustment of the water sample to 10 [12,13], solid-phase microextraction [14] for isolating AN from water. Some of these works propose derivatization procedures to make AN suitable for gas chromatography with electron capture detection [7] fluorescence detection [10,14], or increase retention of the highly polar AN on a liquid-chromatography (LC) column [12]. One work made use of a photo diode array for detecting AN in lake water [9]. Only three works employed detectors potentially capable of ensuring

S. Bogialli et al. / J. Chromatogr. A 1122 (2006) 180–185

unequivocal analyte identification, i.e. tandem mass spectrometry (MS) [8], ion-trap MS [11] and time-of-flight MS [13]. Bioaccumulation of cyanotoxins by aquatic animals, including fishes and mollusks, is well documented [15–19]. In these studies, the authors outlined risk for human health associated to consumption of cyanotoxin-contaminated fishes. Therefore, in addition to direct ingestion of water contaminated by cyanotoxins, another route of exposure of humans to these toxins is the consumption of aquatic animals, which have ingested cyanobacteria and accumulated their toxins. Cyanotoxins are rarely ingested by man in such large amounts to reach an acute lethal dose, but damage provoked by chronic effects is rather probable if exposure is frequent and prolonged. In spite of the fact that AN is a hydrophilic compound, Gugger et al. [20] showed this toxin was present in liver of dogs dead after drinking AN-contaminated water. To our knowledge, only one paper dealing with analysis of AN in fish muscle is quoted in the literature [21]. The analytical procedure described in this work involves, however, a laborious and time consuming sample treatment. The so-called matrix solid-phase dispersion (MSPD) technique is attracting more and more attention for extracting contaminants from solid biological matrices [22]. A fine dispersion of the biological matrix onto a solid support such as silica, alumina, diatomaceous earth, C-18-bonded silica and other sorbents, is easily obtained by blending the sample and the sorbent with a mortar and pestle. After blending, this material is packed into a mini-column and analytes are eluted by a suitable extractant. The abrasive action of the sorbent during blending has been demonstrated to disrupt the gross architecture of the matrix [22], so that a tight and quasi-homogeneous layer of the matrix components is formed on the sorbent surface. Over conventional sample treatment procedures, MSPD offers distinct advantages in that: (i) the analytical protocol is drastically simplified and shortened; (ii) the possibility of emulsion formation is eliminated; (iii) consumption of toxic, flammable and expensive solvents is substantially reduced; (iv) last but not least, the extraction efficiency of the analytes is enhanced as the entire sample is exposed to the extractant. Recently, we have proposed a rapid and simple method for determining sulfonamide antibacterials in fish tissue [23]. The uniqueness of this method is that it is based on analyte extraction from the matrix dispersed on sand by hot water. After little manipulation, a large volume of the aqueous extract was analyzed by LC–single quadrupole MS. Besides being highly specific, the LC–tandem MS technique with acquisition in the multi reaction monitoring (MRM) mode offers very sensitive detection. In fact, preliminary experiments showed that AN could be quantified at levels well lower than 0.1 ␮g/l by simply injecting it from an aqueous solution into a LC–MS/MS apparatus. This work proposes two simple and sensitive LC–tandem MS-based methods for analyzing AN in lake water and fish muscle tissue. Analysis of AN in lake water is performed by direct injection of a relatively large sample volume into the LC–MS/MS apparatus, while analysis of AN in fish involves extraction by the MSPD technique with water as extractant.

181

2. Experimental 2.1. Chemicals AN was purchased from ICN Biomedical, Aurora, OH. 1,9diaminononane and phenylalanine were from Sigma–Aldrich, Milwaukee, WI. The former compound was used as internal standard (IS). A standard solution of AN was prepared by dissolving it in water to obtain 250 ␮g/ml concentration. After preparation, this solution was stored at −18 ◦ C in the dark to minimize analyte degradation. It was freshly prepared every month. Two AN working standard solutions were prepared by further diluting with water to obtain analyte concentrations of 5 and 0.5 ␮g/ml. A 1 mg/ml solution of the IS was prepared by dissolving 1,9diaminononane in methanol. This solution was further diluted with 10 mmol/l formic acid–acidified water to obtain IS concentration of 5 ␮g/ml. A 0.1 mg/ml solution of phenylalanine was prepared by dissolving it in methanol. Sand (Crystobalite, 40–200 mesh size) was provided by Fluka AG, Buchs, Switzerland. Acetonitrile of RS gradient grade and formic acid were obtained from Carlo Erba, Milano, Italy. 2.2. Water samples Water samples were collected at three lakes (Albano, Nemi, Vico) in the area of Rome using 2.5 l Ruttner bottles and stored at −18 ◦ C until analysis. Preliminary analyses showed they were AN-free. One hundred milliliters of lake water was spiked with variable volumes of the analyte working standard solution and a constant volume of 100 ␮l of the IS solution. After filtration with a 125 mm diameter Black Ribbon 589 paper filter (Schleicher & Schuell, Legnano, Italy), 500 ␮l of the sample was injected into the LC column. 2.3. Fish samples Trout, Mullet, Perch and Goldfish samples were from retail markets. Preliminary analyses showed they were analyte-free. These fish species were chosen considering they live in sites, such as lakes or basins designed for aquaculture, which, being prone to eutrophication, may contain algal toxins. 2.4. Extraction apparatus for anatoxin-a in fish The design of the home-made extraction apparatus used in this work was very similar to that shown in a previous paper [24], with the exception that the analyte-containing water leaving the extraction cell was collected in a calibrated glass tube instead of a sorbent cartridge. An 8.1 cm × 8.3 mm i.d. stainless-steel column was used as extraction cell. Prior to blending with sand, fish fillet samples were finely diced with scissors. For recovery studies, 1 g of tissue was put in a porcelain mortar and spiked with variable volumes of working standard solution, taking care of uniformly spreading it on the sample. An intimate contact between analyte and fish tissue was obtained by pounding with the pestle for some minutes.

182

S. Bogialli et al. / J. Chromatogr. A 1122 (2006) 180–185

Table 1 Selected reaction monitoring conditions for detecting anatoxin-a in water and fish Compound

Transition, m/z

Anatoxin-a

166 > 149 166 > 131 166 > 107 166 > 91

DANa (IS)b

159 > 142

a b

Product ion, m/z ]+

[M+H-NH3 [M+H-NH3 H2 O]+ [M+H-NH3 COCHCH3 ]+ [C7 H7 ]+ [M+H-NH3 ]+

Cone Voltage, V

Collision energy, eV

25

15

28

12

DAN = 1,9-diaminononane. IS = internal standard.

Then, 1 h was allowed for equilibration storing the mortar at 4 ◦ C. Thereafter, 5 g of sand were added to the mortar and the mixture was blended with the pestle for about 10 min by moderately heating the mortar with a hairdryer, until an apparently dry material was obtained. This material was then packed into the extraction cell, taking care to tap the tube to avoid loose packing of the particles. Any void space remaining after packing the solid material was filled with sand. A stainless steel frit (2 ␮m pore size) and a polyethylene (20 ␮m pore size) frits were located, respectively above and below the mixture. The tube was then put into the oven and heated at 80 ◦ C for 5 min. Water acidified to pH 2 (HCl) was then passed through the cell at 1-ml/min flow rate to extract the analytes and 4 ml of it was collected in a tube. After extraction, 10 ␮l of the IS solution was added, then the extract was acidified to pH 3.1 with 10 mol/l formic acid and filtered through a regenerated cellulose filter (0.2 ␮m pore size, 25-mm diameter, Alltech, Sedriano, Italy) to prolong the life of the guard column. By doing so, the guard column was replaced with a new one after more than 150 injections of fish extracts. Two hundred microliters of the final extract were injected into the LC column. 2.5. LC–MS/MS analysis The liquid chromatograph consisted of a Waters pump (Model 600 E, Milford, USA) with a 500 ␮l-injection loop, an Alltima HP 5-␮m C-18 guard column (7.5 mm × 4.6 mm i.d.) and an analytical (250 mm × 4.6 mm i.d.) columns (Alltech) thermostated at 35 ◦ C. The LC apparatus was interfaced to a bench-top triple-quadrupole mass spectrometer (Model Micromass QUATTRO MICRO API, Waters) by an electrospray ion (ESI) source. Mobile phase component A was 10 mmol/l formic acid in acetonitrile and component B was aqueous 10 mmol/l formic acid. At 1.0 ml/min, the mobile phase gradient profile was as follows (t in min): t0 , A = 6%; t7 , A = 20%; t8 , A = 100%; t10 , A = 100%; t11 , A = 6%; t20 , A = 6%. Retention time of AN varied ≤0.5% over 2 weeks. A diverter valve led 400 ␮l/min of the LC column effluent into the ion source that was operated in the positive ion mode. High-purity nitrogen was used as drying and curtain gases; high-purity argon was used as collision gas. Nebulizer gas was set at 650 l/h while the cone gas at 50 l/h; the probe and desolvatation temperatures were maintained, respectively at 120 and 350 ◦ C. The settings for the gas

pressure in the collision cell was set at 3 mbar. Capillary voltage was 3000 V, extractor voltage was 2 V. Cone voltage, collision energy and other transmission parameters were optimized for each analyte (see Table 1). Mass axis calibration of each massresolving quadrupole Q1 and Q3 was performed by infusion of a sodium and cesium iodide solution at 10 ␮l/min. Unit mass resolution was set and maintained in each mass-resolving quadrupole by keeping a full width at half maximum of approximately 0.7 amu. All the source and instrument parameters for monitoring AN were optimized by standard solutions of 5 ␮g/ml infused at 10 ␮l/min by a syringe pump. The MRM mode was used for quantitation by selecting one fragmentation reaction for the IS and four fragmentation reactions for AN (Table 1). 2.6. Quantitation Recovery of AN in fish tissue samples at any given concentration was assessed by summing the ion current profiles relative to the transitions considered, normalizing them to the peak area of the IS, and comparing these ratios to those obtained by injecting a blank sample extract to which AN was added post-extraction. We followed this procedure to obviate matrix effects that weakened the analyte ion signal intensity. On the contrary, no matrix effect was observed when injecting AN from lake water samples. The mass spectrometry data handling system used was the “Mass Lab” software from Waters. 3. Results and discussion 3.1. General remarks Phenylalanine is an essential amino acid that is acquired by animals through their diet and thus is invariably present in any animal tissue. AN and phenylalanine produce the same pseudomolecular ion at m/z 166, have very similar fragmentation patterns and similar capacity ratios on a reversed-phase LC column. Recently, there has been concern about possible misidentification of AN in biological matrices due to the presence of phenylalanine [20,25]. Fig. 1 shows two MRM LC–MS/MS chromatograms resulting from injections of an intact perch fillet extract and one spiked with AN at 5 ng/g. The large peak at 7.3 min retention time was for endogenous phenylalanine, as confirmed by adding authentic phenylalanine to the extract. It

S. Bogialli et al. / J. Chromatogr. A 1122 (2006) 180–185

183

Fig. 1. MRM LC–MS/MS chromatograms resulting from injection of (A) an uncontaminated extract of a perch fillet and (B, C, D, E) an extract contaminated by 5 ng/g anatoxin-a. On the peak tops, their absolute areas and ion signal relative abundances (%) generated by decomposition reactions of anatoxin-a (retention time = 6.1 min) and endogenous phenylalanine (retention time = 7.3 min) are reported.

can be seen that the amino acid displayed ion signals for all of the precursor to product ion transitions selected for detecting AN. Nevertheless, AN could be differentiated from phenylalanine in that it was eluted well before the amino acid. Also, relative ion signal intensities generated by the individual decomposition reaction of AN were different from those relative to phenylalanine (see again Fig. 1). The ESI process generates gas-phase ions from AN with an extremely high efficiency. Therefore, AN in water was analyzed without any enrichment step by directly injecting 0.5 ml of filtered lake water into the LC column. By doing so, AN could be simply and rapidly quantified at levels well below 0.1 ␮g/l (see below). This procedure was routinely employed in a project financially supported by the Region Lazio and devoted to assess the contamination levels of AN and other cyanotoxins in some lakes of the Region. Over 6 months of direct analyses of lake

water samples, we did not note any failure of the LC–MS/MS apparatus. 3.2. Recovery studies of anatoxin-a in fish tissue When not explicitly mentioned, recovery studies were conducted by spiking trout fillet samples with AN at 25 ng/g and analyzing. Following conditions reported elsewhere [23], initial extraction experiments were performed by using pure water heated at 80 ◦ C as extractant. Experiments performed in duplicate showed that pure water gave poor extraction yield of AN (recovery = 38%). This unsatisfying result was traced to the presence on the siliceous material (sand) supporting the biological matrix of silanols able to strongly bind compounds bearing amino groups, such as AN. Protonation of the amino groups by extracting with heated water acidified to pH 2 increased remark-

184

S. Bogialli et al. / J. Chromatogr. A 1122 (2006) 180–185

ably recovery of AN (recovery = 74%). Attempts to extract larger amounts of AN by further decreasing the pH of water failed probably due to analyte decomposition. By progressively increasing the temperature, water becomes more and more effective in extracting organic compounds from solid matrices as result of its decrease in polarity [26]. On the other hand, a risk inherent to the use of hot water as extractant is that it could decompose those compounds that are thermolabile and/or prone to hydrolytic attack. With the aim of enhancing recovery of AN in fish tissue obtained by extracting it with acidified water at 80 ◦ C, we performed two recovery experiments where the extraction temperature was in one case lower, i.e. 70 ◦ C, and in the second case higher than 80 ◦ C, i.e. 100 ◦ C. At each temperature, four extractions were carried out. Recovery data showed that the best result was just that obtained by extracting at 80 ◦ C (recovery, RSD = 77%, 7%). Probably, the lower extraction yield observed at 100 ◦ C was due to some decomposition of AN. According to our previous work [23], initial recovery experiments were performed by passing through the extraction cell 4 ml of water heated at 80 ◦ C. Anyway, additional recovery studies of AN in fish tissue were conducted by both increasing and decreasing the extractant volume mentioned above. The experiment performed by using a larger water volume had the obvious purpose of ascertaining if a larger water volume succeeded to extract larger amounts of AN, while the rationale behind the use of a lower extractant volume was that of extracting the maximum amount of AN with a minimum extractant volume. Since this method does not include any concentration step of

the extract, the extractant volume influences directly the sensitivity of the method. At any extractant volume considered, duplicate experiments were performed. Results showed that the use of only 3 ml of extractant gave an unacceptable low recovery (52%) of AN, while no significant increase of the extraction yield was observed by extracting with more than 4 ml of the extractant. In order to assess that the extraction yield of AN from fish muscle tissue was not affected by the concentration of the analyte we analyzed samples of trout fillets that were contaminated by AN at three different concentrations, i.e. 5, 25 and 100 ng/g. Four measurements performed at each analyte concentration showed that recovery of AN ranged between 71 and 78% with RSD not larger than 9% and it was not significantly affected by the AN concentration in fish tissue. We evaluated if the type of fish could affect recovery of AN. For this purpose, we spiked Trout, Perch, Mullet and Goldfish fillet samples with 25 ng/g of the targeted compound and analyzed. Results from quadruplicate experiments for each type of fish showed that recovery varied between 71 and 79%. To check that the extraction efficiency of AN in fish fillet was not dependent on the particular type of fish, mean accuracy data were compared among them by using the one-way analysis of variance (ANOVA) test at the P = 0.05 significance level. Statistical analysis gave a F4,16 value of 2.079 and was lower than the critical value (3.729), showing that the extraction method was not influenced by the type of fish. This result indicates that this method could be employed for analyzing AN in fishes other than those considered in this study.

Fig. 2. MRM LC–MS/MS chromatogram resulting from direct injection of 0.5 ml of a filtered lake water sample spiked with 50 ng/l anatoxin-a (retention time = 6.15). AN, anatoxin-a; IS, internal standard (1,9-diaminononane).

S. Bogialli et al. / J. Chromatogr. A 1122 (2006) 180–185

3.3. Limits of detection (LODs) and quantification (LOQs) of anatoxin-a in fresh water and fish muscle tissue LOQs of the method were estimated from two MRM LC–MS–MS chromatograms resulting from analyses of a lake water sample spiked with 50 ng/l AN (Fig. 2) and an extract of trout muscle tissue spiked with the target compound at 2 ng/g level (not shown here), respectively. After extracting the sum of the ion currents of the transitions selected for AN, the resulting trace was smoothed twice by applying the mean smoothing method. Thereafter, the peak height-to-averaged background noise ratio was measured. The background noise estimate was based on the peak-to-peak baseline near the analyte peak. LOQs were then calculated on the basis of a minimal accepted value of the signal-to-noise ratio (S/N) of 10. Based on this definition, LOQ of AN was 13 ng/l in water and 0.5 ng/g in fish fillet. When performing detection with a MS/MS arrangement, the most important condition to be satisfied for affirming the presence of a targeted compound is that at least two precursor to product ion transitions give signals distinguishable from the background ion current. Accordingly, a definition of LOD (S/N = 3) was adopted, considering the transition that, between the best two ones, had the worst S/N value namely 166 > 131. On this basis, LOD of AN was 8 ng/l in water and 0.2 ng/g in fish fillet. 3.4. Linear dynamic range Under the instrumental conditions reported in Section 2, the linear dynamic range of the ESI/MS/MS detector was estimated for AN. Analyte amounts varying from 0. 02 to 200 ng and a constant amount of 2.5 ng of the internal standard were injected from suitably prepared standard solutions into the LC column. At each analyte amount, three replicate measurements were made. Signal against amount-injected curves were then constructed by averaging the peak area of AN and relating this area to that of the internal standard. Results showed that ion signals of AN were linearly correlated with injected amounts up to 200 ng, with R2 of 0.9975. 4. Conclusion This work has shown that anatoxin-a in lake water can be simply and rapidly analyzed at concentration slightly higher than 10 ng/l by direct injection of a filtered sample in a LC–tandem MS apparatus. Also, it has again shown that the environmentally

185

friendly and inexpensive water, besides being an effective extractant for anatoxin-a in fish tissue, produces sufficiently clean extracts requiring little manipulation before final analysis by LC–tandem MS. Besides being highly specific, the ESI/MS/MS detector provides sensitivity for analyzing anatoxin-a in fish fillets at levels lower than 1 ng/g. References [1] S.G. Bell, G.A. Codd, Agricultural Chemicals and the Environment, Royal Society of Chemistry, 1996, p. 109. [2] K. Sivonen, G. Jones, in: I. Chorus, J. Bartram (Eds.), Toxic cyanobacterial in water—A Guide to their Public Health Consequence, Monitoring and Management, Worlds health Organization, E & FN Spon, Routledge, London, 1999, p. 41. [3] O.M. Skulberg, G.A. Codd, W.W. Carmichael, Ambio 13 (1984) 244. [4] W.W. Carmichael, J. Appl. Bacteriol. 72 (1992) 445. [5] C. Edwards, K.A. Beattie, C.M. Scrimgeour, G.A. Codd, Toxicon 30 (1992) 1165. [6] K. Sivonen, K. Himberg, R. Luukkainen, S.I. Niemela, G.K. Poon, G.A. Codd, Toxic. Ass. 4 (1989) 339. [7] H. Fromme, A. K¨ohler, R. Krause, D. F¨uhrling, Environ. Toxicol. 15 (2000) 120. [8] J. Pietsch, S. Fichtner, L. Imhof, W. Schmidt, H.-J. Brauch, Chromatographia 54 (2001) 339. [9] A. Ballot, L. Krienitz, K. Kotut, C. Wiegand, S. Pflugmacher, Harmful Algae 4 (2005) 139. [10] J.K. James, I.R. Sherlock, M.A. Stack, Toxicon 35 (1997) 963. [11] A. Furey, J. Crowley, M. Lehane, K.J. James, Rapid Commun. Mass Spectrom. 17 (2003) 583. [12] M. Takino, S. Daishima, K. Yamaguchi, J. Chromatogr. A 862 (1999) 191. [13] M. Maizels, W.L. Budde, Anal. Chem. 76 (2004) 1342. [14] A. Namera, A. So, J. Pawliszyn, J. Chromatogr. A 963 (2002) 295. [15] A. Amorim, V.M. Vasconcelos, Toxicon 37 (1999) 1041. [16] F. Tencala, D. Dietrich, C. Schlatter, Aquat. Toxicol. 30 (1994) 215. [17] L. Thostrup, K. Christoffersen, Arch. Hydrobiol. 145 (1999) 447. [18] V.M. Vasconcelos, Aquat. Toxicol. 32 (1995) 227. [19] D.E. Williams, M. Craig, S.C. Dawe, M.L. Kent, R.J. Andersen, C.F.B. Holmes, Toxicon 35 (1997) 985. [20] M. Gugger, S. Lenoir, C. Berger, A. Ledreux, J.-C. Druart, J.-F. Humbert, C. Guette, C. Bernard, Toxicon 45 (2005) 919. [21] V. Hormaz´abal, Ø. Østensvik, B. Underdal, O.M. Skulberg, J. Liq. Chromatogr. Related Technol. 23 (2000) 185. [22] S.A. Barker, J. Chromatogr. A 885 (2000) 115. [23] S. Bogialli, R. Curini, A. Di Corcia, M. Nazzari, R. Samperi, Anal. Chem. 75 (2003) 1798. [24] C. Crescenzi, G. D’Ascenzo, A. Di Corcia, M. Nazzari, S. Marchese, R. Samperi, Anal. Chem. 71 (1999) 2157. [25] A. Furey, J. Crowley, B. Hamilton, M. Lehane, K.J. James, J. Chromatogr. A 1082 (2005) 91. [26] S.B. Hawthorne, Y. Yang, D.J. Miller, Anal. Chem. 66 (1994) 2912.