Liquid Chromatography−Tandem Mass Spectrometry Application, for

Anal. Chem. 2007, 79, 3436-3447

Liquid Chromatography-Tandem Mass Spectrometry Application, for the Determination of Extracellular Hepatotoxins in Irish Lake and Drinking Waters Orla Allis,† Justine Dauphard,† Brett Hamilton,† Aine Ni Shuilleabhain,‡ Mary Lehane,†,§ Kevin J. James,† and Ambrose Furey*,†

PROTEOBIO, Mass Spectrometry Centre for Proteomics and Biotoxin Research, Cork Institute of Technology, Bishopstown, Cork, Ireland, Northern Regional Fisheries Board, Corlesmore, Ballinagh, Co. Cavan, Ireland, and Department of Applied Sciences, Limerick Institute of Technology, Moylish Park, Limerick, Ireland

A novel method for the determination of hepatotoxins; microcystins (MCs), and nodularin (Nod) in lake water and domestic chlorinated tap water has been developed using liquid chromatography hyphenated with electrospray ionization triple quadrupole mass spectrometry (LCESI-MS/MS). Optimization of the mass spectrometer parameters and mobile-phase composition was performed to maximize the sensitivity and reproducibility of the method. Detection of the hepatotoxins was carried out using multiple reaction monitoring experiments, thus improving the selectivity of the method. A total ion chromatogram and a precursor ion scan on ion m/z 135 was also applied to all samples to detect unknown microcystins or microcystins for which there are no standards available. A comprehensive validation of the LC-ESI-MS/ MS method was completed that took into account matrix effects, specificity, linearity, accuracy, and precision. Good linear calibrations were obtained for MC-LR (1200 µg/L; R2 ) 0.9994) in spiked lake and tap water samples (1-50 µg/L; R2 ) 0.9974). Acceptable interday repeatability was achieved for MC-LR in lake water with RSD values (n ) 9) ranging from 9.9 (10 µg/L) to 5.1% (100 µg/L). Excellent limits of detection (LOD) and limits of quantitation (LOQ) were achieved with spiked MCs and Nod samples; LOD ) 0.27 µg/L and LOQ ) 0.90 µg/L for MC-LR in the “normal linear range” and LOD ) 0.08 µg/L and LOQ ) 0.25 µg/L in the “low linear range” in both lake and chlorinated tap water. Similar results were obtained for a suite of microcystins and nodularin. This sensitive and rapid method does not require any sample preconcentration, including the elimination of solid-phase extraction (SPE) for the effective screening of hepatotoxins in water below the 1 µg/L WHO provisional guideline limit for MC-LR. Furthermore, SPE techniques are timeconsuming, nonreproducible at trace levels, and offer poor recoveries with chlorinated water. The application of this LC-ESI-MS/MS method for routine screening of hepatotoxins in lake and chlorinated tap water (average Cl2 ) 0.23 mg/L) is achieved and this study represents 3436 Analytical Chemistry, Vol. 79, No. 9, May 1, 2007

the first direct method for the screening of hepatotoxins in chlorinated tap water. Cyanobacteria (blue-green algae) are very important freshwater phytoplanktonic organisms due to their ability to potentially produce hepatotoxins and neurotoxins.1 These potent toxins produced by various cyanobacteria genera represent a serious health and environmental threat to water bodies, in particular those that supply drinking water treatment plants. The most common class of hepatotoxins are the microcystins, of which to date there are over 70 isolated and structurally elucidated.1,2 Perhaps the most common microcystin producers are members of the Microcystis spp. genera; in particular Microcystis aeruginosa. However, Planktothrix is also a common producer of microcystins, especially in Central Europe.3 Microcystins are cyclic heptapeptides with the following general structure: cyclo(D-alanine-X-DMeAsp-Z-Adda-D-glutamate-Mdha). These large compounds contain two variable L-amino acids at the X and Z positions, which give rise to the naming of the compounds; for example, microcystin-LR (MC-LR) contains the amino acids; leucine (L) and arginine (R) (Figure 1a, Table 1). Nodularin (Nod) is a cyclic pentapeptide and, like the microcystins, adversely affects hepatocytes by binding to the protein phosphatases 1 and 2A, thus inhibiting normal cell regulation4,5 (Figure 1b, Table 1). At high toxin concentrations, the disruption of the hepatocyte structure leads to a complete collapse of liver function followed by death; at subacute levels, the microcystins are potent tumor promoters.6,7 * To whom correspondence should be addressed. Phone: (+353)-21-4326701. E-mail: [email protected]. Fax: (+353)-21-4345191. † Cork Institute of Technology. ‡ Northern Regional Fisheries Board. § Limerick Institute of Technology. (1) Sivonen, K.; Jones, G. Cyanobacterial toxins; E and FN Spon: London, 1999. (2) Spoof, L.; Vesterkvist, P.; Lindholm, T.; Meriluoto, J. J. Chromatogr., A 2003, 1020, 105-119. (3) Ernst, B.; Hoeger, S. J.; O’Brien, E.; Dietrich, D. R. Aquat. Toxicol. 2006, 79, 31-40. (4) Yoshizawa, S. R.; Matsushima, M. F.; Watanabe, M. F.; Harada, K.-I.; Ichihara, A.; Carmichael, W. W.; Fujiki, H. Cancer Res. Clin. Oncol. 1990, 116, 609-614. (5) Dawson, R. M. Toxicon 1998, 36, 953-962. (6) Falconer, I. R.; Buckley, T. H. Med. J. Aust. 1989, 150, 351. 10.1021/ac062088q CCC: $37.00

© 2007 American Chemical Society Published on Web 04/03/2007

Figure 1. General structure of (a) microcystins and (b) nodularin. Table 1. General Structure of Microcystins (Figure 1a) and Nodularin (Figure 1b) microcystins

molecular weight

MC-LR MC-RR MC-YR MC-LW MC-LF MC-LA Nodularin

994 1037 1044 1024 985 909 824

X

Z

leucine arginine tyrosine leucine leucine leucine

arginine arginine arginine tryptophan phenylalinine alanine arginine

There are a number of documented public health incidences associated with microcystin toxin poisoning, in particular via drinking water8-10 and also through hemodialysis water. In 1996, more than 50 patients died from microcystin intoxication in a hemodialysis center in Caruaru, Brazil.11-13 These human fatalities in Caruaru stress the need for a sensitive and rapid method for the detection of hepatotoxins in both lake and treated tap waters. In 1998, the World Health Organization (WHO) set up a provisional guideline limit of 1 µg/L for MC-LR in drinking water.9 Following these guidelines Spain has set a drinking water standard for total microcystins of 1.0 µg/L.14 This low guideline limit was the impetus to develop a rapid and sensitive method for the detection of this toxin and other commonly occurring microcystins (7) Sherlock, I. R.; James, K. J.; Caudwell, F. B.; MacKintosh, C. Nat. Toxins 1997, 5, 247-254. (8) Ueno, Y.; Nagata, S.; Tsutsumi, T.; Hasegawa, A.; Watanabe, M. F.; Park, H. D.; Chen, G. C.; Chen, G.; Yu, S. Z. Carcinogenesis 1996, 17, 13171321. (9) WHO. Addendum to volume two, Health criteria and other supporting information, Guidelines for drinking water quality, 2nd ed.; WGO: Geneva, 1998. (10) Zhou, L.; Yu, H.; Chen, K. Biomed. Environ. Sci. 2002, 15, 166-171. (11) Jochimsen, E.; Carmichael, W. W.; An, J.; Cardo, D.; Cookson, D.; Holmes, C.; Antunes, M. de C.; de Melo Filho, D.; Lyra, T.; Barreto, V.; Azevedo, S.; Jarvis, W. N. Engl. J. Med. 1998, 338, 873-878. (12) Pouria, S.; de Andrade, A.; Barbosa, J.; Cavalcanti, R. L.; Barreto, V. T.; Ward, C. J.; Preiser, W.; Poon, G. K.; Neild, G. H.; Codd, G.A. Lancet 1998, 352, 21-26. (13) Yuan, M.; Carmichael, W. W.; Hilborn, E. D. Toxicon 2006, 48, 627-640. (14) BOE Real Decreto 140/2003, BOE no. 45, 2003.

in a range of water types: lake water and chlorinated tap water. Hepatotoxins are commonly monitored in complex biological matrixes using either biological methods of detection, enzymelinked immunosorbent assays (ELISAs) and protein phosphatase inhibition assays.15,16 or physicochemical methods of detection. The latter methods of detection are necessary for the quantitation and identification of individual toxins. At present, the routine analysis of aqueous samples for hepatotoxins commonly requires a solid-phase extraction (SPE)17 or solvent extraction step followed by either one or a combination of the subsequent methods: HPLCUV using diode array detection, CE-MS, FAB-MS, LC-MS, and LC-MS/MS.2,17-35 More recent LC-ESI-MS methods include a method for the detection of MC-LR in surface water36 this method utilized doubly protonated [M + 2H]2+ MC-LR thereby achieving greater sensitivity when compared to the typical [M + H]+ utilized for monitoring MC-LR. Meriluoto et al.37 introduced a high-throughput LC-ESIMS method for the analysis of 10 microcystins and nodularins; this method enabled very fast analysis times of 2.8 min/sample, with detection limits of 50-100 pg/injection. Ruangyuttikarn et al.38 published an LC-ESI-MS method using SIM, with a high limit of detection (LOD; 100 µg/L), and an improved LC-MS/MS ion trap method was reported by Ortea et al.18 giving an LOD of 0.1 ng on column (MC-LR). LC-ESI-MS/MS performed in multiple reaction monitoring (MRM) mode facilitated the determination of four microcystin variants: MC-LR, MC-RR, MC-LW, and MCLF with low limits of detection29 employing the LC-ESI-MS/MS (15) Rivasseau, C.; Hennion, M. Anal. Chim. Acta 1999, 399, 75-87. (16) An, J.; Carmichael, W. W. Toxicon 1994, 32, 1495-1507. (17) Lawton, L. A.; Edwards, C.; Codd, G. A. Analyst 1994, 119, 1525-1530. (18) Ortea, P. M.; Allis, O.; Healy, B. M.; Lehane, M.; Ni Shuilleabhain, A.; Furey, A.; James, K. J. Chemosphere 2004, 55, 1395-1402. (19) Park, H.; Namikoshi, M.; Brittain, S. M.; Carmichael, W. W.; Murphy, T. Toxicon 2001, 39, 855-862. (20) Poon, G. K.; Griggs, L. J.; Edwards, C.; Beattie, K. A.; Codd, G. A. J. Chromatogr., A 1993, 628, 215-233. (21) Edwards, C.; Lawton, L. A.; Beattie, K. A.; Codd, C. A. Rapid Commun. Mass Spectrom. 1993, 7, 714-721. (22) Namikoshi, M.; Rinehart, K. L.; Sakai, R. J. Org. Chem. 1990, 55, 61356139. (23) Yuan, M.; Namikoshi, M.; Otsuki, A.; Watanabe, M. F.; Rinehart, K. L. J. Am. Soc. Mass Spectrom. 1999, 10, 1138-1151. (24) Kondo, F.; Ikai, Y.; Oka, H.; Ishikawa, N.; Watanabe, M. F.; Watanabe, M.; Harada, K.-I.; Suzuki, M. Toxicon 1992, 30, 227-237. (25) Bateman, K. P.; Thibault, P.; Douglas, D. J.; White, R. L. J. Chromatogr., A 1995, 712, 253-268. (26) Zweigenbaum, J. A.; Henion, J. D.; Beattie, K. A.; Codd, G. A.; Poon, G. K. J. Pharm. Biomed. 2000, 23, 723-733. (27) Spoof, L.; Karlsson, K.; Meriluoto, J. J. Chromatogr., A 2001, 909, 225236. (28) Spoof, L.; Meriluoto, J. J. Chromatogr., A 2002, 947, 237-245. (29) Cong, L.; Huang, B.; Chen, Q.; Lu, B.; Zhang, J.; Ren, Y. Anal. Chim. Acta 2006, 569, 157-168. (30) McElhiney, J.; Lawton, L. A. Toxicol. Appl. Pharm. 2005, 203, 219-230. (31) Hummert, C.; Reichelt, M.; Weiss, Liebert, H.-P.; Luckas, B. Chemosphere 2001, 44, 1581-1588. (32) Brittain, S.; Mohamed, Z. A.; Wang, J.; Lehmann, V. K. B.; Carmichael, W. W.; Rinehart, K. L. Toxicon 2000, 38, 1759-1771. (33) Ramanan, S.; Tang, J.; Velayudhan, A. J. Chromatogr., A 2000, 883, 103112. (34) Rivasseau, C.; Martins, S.; Hennion, M. J. Chromatogr., A 1998, 799, 115169. (35) Feng, X. G.; Ding, Z.; Wei, T.; Yuan, C. W.; Fu, D. G. Biomed. Environ. Sci. 2006, 19, 225-231. (36) Zhang, L.; Ping, X.; Yang, Z. Talanta 2004, 62, 191-198. (37) Meriluoto, J.; Karlsson, K.; Spoof, L. Chromatographia 2004, 59, 291-298. (38) Ruangyuttikarn, W.; Miksik, I.; Pekkoh, J.; Peerapornpisal, Y.; Deyl, Z. J. Chromatogr., B 2004, 800, 315-319.

Analytical Chemistry, Vol. 79, No. 9, May 1, 2007

3437

methodology previously developed by Zhang et al.36 A lengthy solid-phase microextraction procedure followed by microbore-LCESI-QTOF-MS was used by Zhao et al.39 in the determination of MC-LR and MC-RR. Low limits of detection, 1.6 (MC-LR) and 0.6 pg (MC-RR), were obtained using this method. Although good sensitivity was achieved with all/most of the aforementioned LCESI-MS methods, SPE or some sample extraction/concentration technique is required prior to sample analysis to achieve below the WHO provisional guideline limit of 1 µg/L for MC-LR. Extraction of hepatotoxins using SPE is tedious and timeconsuming and gives variable rates of recovery, which may be due to differences in sample matrix composition and SPE batchto-batch variation.40-43 Efficient extraction of microcystins is also problematic because of the differing polarities of members of this toxin group; therefore, using a generic SPE method for all microcystins is impossible.40 Lawton et al.17 reported recovery problems with MC-LW, especially from tap water, and Moollan encountered serious problems with this procedure in the analysis of chlorinated waters.44 SPE is also an expensive method if large screening programs are initiated, as cartridges are disposed of after each extraction; however, this expense can be greatly reduced if SPE cartridges are made in-house. This paper presents a novel approach that eliminates the need for conventional SPE cartridges and solvent extraction prior to the LC-MS/MS analysis of microcystins in water. It involves employing a reversed-phase guard cartridge system as the analytical column. The advantage of this cartridge system is that the short cartridge with closely packed and small stationary phase (∼3 µm) provides narrow chromatography peaks (width 95%), MC-LF (>95%), and nodularin (96%) from Alexis (39) Zhao, Y.-Y.; Hrudey, S.; Li, X.-Y. J. Chromatogr. Sci. 2006, 44, 359-365. (40) Furey, A.; Muniz-Ortea, P.; Crowley, J.; Allis, O.; Hamilton, B.; Diaz Sierra, M.; Lehane, M.; James, K. J. Florida 2002; Intergovernmental Oceanographic Commission of UNESCO, Paris, 2004; pp 258-260. (41) Aranda-Rodriguez, R.; Kubwabo, C.; Benoit, F. M. Toxicon 2003, 42, 587599. (42) Mhadhbi, H.; Ben-Rejeb, S.; Cle´roux, C.; Martel, A.; Delahaut, P. Talanta 2006, 70, 225-235. (43) Maizels, M.; Budde, W. L. Anal. Chem. 2004, 76, 1342-1351. (44) Moollan, R. W.; Rae, B.; Verbeek, A. Analyst 1996, 121, 233-238. (45) EC Off. J. Eur. Commun. 2002, L221, 8.

3438

Analytical Chemistry, Vol. 79, No. 9, May 1, 2007

Biochemicals (Alpha Technologies, Dublin, Ireland) and MC-LA (g95%) and MC-YR (g95%) from Calbiochem (Nottingham, U.K.). MC-LR (95%), trifluoroacetic acid (TFA), formic acid, nonafluoropentanoic acid (NFPA), heptafluorobutanoic acid (HFBA), pentafluoropentanoic acid (PFPA), ammonium acetate, sodium acetate, and Riedel-de Hae¨n Vario H DPD free chlorine reagent were purchased from Sigma-Aldrich (Dublin, Ireland). Chlorine standard, 25 mg/L, and DPD free chlorine reagent were purchased from Hach Lange Ltd. (Dublin, Ireland). All solvents including HPLC grade acetonitrile, methanol, and water were purchased from Labscan (Dublin, Ireland). Free Chlorine in Chlorinated Tap Water. Chlorine standards (0.01-0.5 mg/L) were prepared in LC grade water. Chlorinated tap water samples were collected from different locations in Cork City, Ireland, to obtain the average free Cl2 level. Using the N,Ndiethyl-p-phenylenediamine (DPD) colorimetric method, standards, and samples were analyzed at 530 nm on a single-beam UV-vis spectrophotometer (UV-1201, Shimadzu). Two different DPD reagents (10-mL sample) were investigated, Hach Lange and Riedel-de Hae¨n. These sealed powder sachets were dispensed into standards, samples, and LC grade water blanks prior to analysis. This method of analysis was completed for the chlorinated tap water samples 12, 36, and 60 h after collection of samples. The pH of each sample was analyzed using a portable pH meter by Hanna instruments (Leighton Buzzard, U.K), calibrated using buffers 4, 7, and 10. Chlorination of MC-LR (Degradation Assessment). Experiments were carried out on an Agilent HP 1100 series HPLC system Agilent (Palo Alto, CA) with a UV-diode array detector. A Luna C18 column (150 mm L × 1.0 mm i.d., 3 µm) purchased from Phenomenex (Macclesfield, Cheshire, U.K.) was used at 40 °C with an injection volume of 50 µL. Gradient reversed-phase chromatography was performed using water with 0.05% TFA (buffer A) and acetonitrile with 0.05% TFA (buffer B). The gradient program using a flow rate of 0.07 mL/min, started at 70% A and was held steady for 5 min; the composition was then reduced to 30% A and held for 4 min, and 0% A and held for another 4 min. The gradient was then increased to 30% A and held for 17 min, and finally, the gradient was increased to 70% A for 20 min. HPLC water was spiked with MC-LR (10 mg/L) and with a number of low range chlorine standards (25 mg/L), giving final concentrations of 0.1, 0.2, 0.5, 1.0, and 2.0 mg/L chlorine and 10 mg/L MC-LR. A MC-LR standard (10 mg/L) and the spiked chlorinated MC-LR samples were then analyzed by LC-UV-diode array detection, and the absorbance was measured at the λmax 238 nm. The aforementioned MC-LR standards and spiked chlorinated MC-LR samples were analyzed over 2 days, with a sample number equal to three (n ) 3) each day. Sample and Standard Preparation. Lake water samples, blank lake water, and chlorinated tap water were all filtered through Whatman filters (0.45-µm nylon filters) and bath sonicated (Ultrasonic LC 20 H) for 10 min prior to sample analysis. Lake water cyanobacterial samples were filtered through Whatman filters (0.45-µm nylon filters) prior to LC-MS/MS analysis to remove all cyanobacterial cells, as the objective was to analyze extracellular microcystin levels in water only. Standard addition experiments were performed on water samples by spiking MCLR standard (0, 10, 20, 25, 30, and 35 µg/L in methanol) into

six amber vials (1.5 mL). These solutions were evaporated to dryness under nitrogen (99% purity) and reconstituted with the test sample (1 mL). All standard solutions for method validation were prepared by spiking MCs and Nod into amber vials, evaporating them to dryness, and reconstituting them in the relevant water type. Intrumentation and Analytical Conditions (LC-MS/MS). All LC-MS/MS experiments were carried out on an Agilent HP 1100 series LC system Agilent hyphenated with an API3000 triplestage quadrupole mass spectrometer (Applied Biosystems, Warrington, U.K.) using a turbo-assisted ion spray source (Sciex, Toronto, Canada) at 425 °C. A SecurityGuard guard cartridge system with a C18 ODS octadecyl cartridge (4.0 mm L × 2.0 mm i.d.) purchased from Phenomenex was used as an analytical chromatography column at 30 °C and at a flow rate of 0.2 mL/ min. Gradient reversed-phase chromatography was performed using water with 1.0 mM ammonium acetate (buffer A) and methanol with 1.0 mM ammonium acetate (buffer B). The gradient program started at 10% A, and this was held steady for 3 min; the composition was rapidly raised to 100% B and held for a further 7 min. An equilibration time of 10 min was also needed. The sample was loaded onto the cartridge at a very low organic composition, thus effectively preconcentrating the sample onto the cartridge prior to starting the gradient program, which elutes the hepatotoxins off very quickly. MRM experiments were performed using low resolution in order to obtain the highest sensitivity, and all the Q1/Q3 pairs had a dwell time of 100 ms with an overall scan time of 3.36 s. The optimized conditions are shown in Table 2, with four different MRM transitions (Q1/Q3 pairs) obtained for each hepatotoxin together with optimum collision energies for each transistion. The most intense Q1/Q3 paired response was used for quantitative purposes during this study, for example, MC-LR (Q1/Q3 ) m/z 995.5/135), MC-YR (Q1/Q3 ) m/z 1045.5/135), MC-RR (Q1/Q3 ) m/z 519.8/135), MC-LA (Q1/Q3 ) m/z 910.5/776), MC-LW (Q1/Q3 ) m/z 1025.5/135), MC-LF (Q1/Q3 ) m/z 986.5/135), and Nod (Q1/Q3 ) m/z 825.5/135). All microcystins investigated utilized the characteristic MC product ion at m/z 135 [C9H11O]+ for quantitation with the exception of MC-LA, which used the ion at m/z 776 [M + 2H - C9H11O]+ as was previously reported.18,66 This Q1/Q3 pair of m/z 910.5/776.0 for MC-LA gave the best ion response and was therefore chosen for quantitation. A total ion chromatogram (TIC) and a precursor ion scan on ion m/z 135 (a portion of the common ADDA fragment characteristic of most microcystins and nodularins, Figure 1) was also integrated into the LC-MS/MS methods to detect unknown microcystins or microcystins for which there are no standards. The source dependent parameters used were ionspray voltage (IS) 5000 V, nebulizer gas 10, curtain gas 12, and collision gas 4. Throughout each analysis, a valco switching valve was used to divert the eluent flow to waste for 0.8 min after sample injection, upon completion of the analysis, and during the subsequent equilibration time to limit the flow of eluent reaching the mass spectrometer. Analyst version 1.3.2 software was used to acquire and process the data. Nanospray Hybrid Quadrupole Time-of-Flight (QqTOF) Mass Spectrometry. A quadrupole/time-of-flight mass spectrometer (Qstar, Applied Biosystems, Foster City, CA) with a nanospray (ES023C, Protana ) ion source was used to produce collision

Table 2. Optimized LC-MS/MS Parametersa MRM transitions microcystin analyte

MS parameters

Q1 mass Q3 mass CE DP FP ion ratio: (m/z) (m/z) (V) (V) (V) 50 µg/L

MC-LR

[M + H]+

MC-RR

[M + 2H]2+

MC-YR

[M + H]+

MC-LW

[M + H]+

MC-LF

[M + H]+

MC-LA

[M + H]+

nodularin

[M + H]+

995.6 995.6 995.6 995.6 519.8 519.8 519.8 519.8 1045.5 1045.5 1045.5 1045.5 1025.5 1025.5 1025.5 1025.5 986.5 986.5 986.5 986.5 910.5 910.5 910.5 910.5 825.5 825.5 825.5 825.5

135.0 127.0 107.0 155.0 135.0 127.0 174.0 200.0 135.0 107.0 127.0 136.0 135.0 127.0 213.0 375.0 135.0 213.0 852.0 478.0 776.0 402.0 374.0 759.0 135.0 227.0 163.0 107.0

105 110 115 90 40 70 50 45 110 130 130 125 95 125 75 50 110 65 25 35 30 35 45 30 75 70 65 95

80 80 80 80 65 65 65 65 80 80 80 80 90 90 90 90 100 100 100 100 110 110 110 110 75 75 75 75

200 200 200 200 150 150 150 150 200 200 200 200 200 200 200 200 250 250 250 250 350 350 350 350 400 400 400 400

1.00 0.51 0.50 0.30 1.00 0.41 0.16 0.19 1.00 0.46 0.45 0.39 1.00 0.56 0.44 0.39 1.00 0.61 0.49 0.54 1.00 0.59 0.59 0.43 1.00 0.50 0.34 0.39

a Collision energy (CE), declustering potential (DP), and focusing potential (FP) for each precursor/product ion pair of the seven hepatotoxins monitored and the ion ratio of each precursor/product ion pair at 50 µg/L.

ionization dissociation (CID) spectra in positive mode. Microcystin standards (MC-LR, MC-RR, MC-YR, MC-LW, MC-LF, MC-LA, Nod) were dissolved in methanol (0.01 µg/10 µL). CID spectra were obtained from the [M + H]+ ions of MC-LR, m/z 995; MCYR, m/z 1045; MC-LW, m/z 1025; MC-LF, m/z 986; MC-LA, m/z 910; Nod, m/z 825; and [M + 2H]2+ ions of MC-RR, m/z 5.19. The solution to be analyzed (5-10 µL) was loaded into a nanospray capillary (Long NanoES spray capillary, ES381, Protana) using a 0.5-10-µL Eppendorf GELoader tip and a TOMY PMC-060 Capsulefuge. Once opened, an ion source voltage (IS) of -1200 V was applied to the tip of the nanospray capillary to generate a spray. While scanning in the range m/z 50-1400, the voltages used were as follows: declustering potential (DP) 90 V, focusing potential (FP) 205 V, declustering potential 2 (DP2) 15 V, and collision energy (CE) 75. The accumulation time was 1 s, Pulsar frequency was 6.991 kHz and pulse duration 13 ms. Mobile-Phase Study. The mobile-phase study involved the comparison between methanol and acetonitrile as organic components of the mobile phase. A number of different modifiers, TFA, formic acid, ammonium acetate, sodium acetate, NFPA, HFBA, and PFPA, were also compared for the enhancement of instrument response, namely, chromatographic separation and MS ionization efficiency. Chromatography studies were carried out using the various mobile-phase compositions with a standard mixture of MC-LR, MC-RR, MC-YR, MC-LW, MC-LF, MC-LA, and Nodularin (100 µg/L, n ) 7). Both ionization and chromatographic effects of the mobile-phase composition were investigated during this study. Analytical Chemistry, Vol. 79, No. 9, May 1, 2007

3439

Postcolumn Infusion Study To Examine Ion Suppression/ Enhancement Effects. Postcolumn infusion experiments were conducted to determine ion suppression/enhancement effects using the approach described by Bonfiglio et al.46 The experimental setup used in these experiments involved the infusion of a standard into a blank matrix solution. A T-junction was placed between the LC system and the MS source (after the MS flow splitter), and the standard mixture, monitored by MRM, was introduced at 50 µL/min into the LC eluent. The standard mixture (100 µg/L) was placed in a 1-mL glass gastight syringe (Hamilton, Birmingham, U.K.) and delivered with a syringe pump (Harvard Apparatus, Holliston, MA). A blank matrix (LC grade water) was injected (10 µL) via the autosampler into the LC system. The response of the standard mixture was monitored continuously to produce a profile of the matrix effect. This was repeated several times (n ) 5) to ensure the effect was reproducible. The blank matrix (LC grade water) was then changed, for comparison and assessment purposes, to two other typical environmental sample matrixes, lake water, and chlorinated tap water. These matrixes were tested in the same manner as the LC grade water. RESULTS AND DISCUSSION Cyanobacterial blooms have been known to be toxic to birds, animals, and humans for many years.47-49 To ensure the health and safety of humans and animals, both wild and domestic, adequate monitoring programs must be initiated to prevent the consumption of water contaminated by cyanobacterial toxins. However, this can only be achieved through the development of sensitive and selective analytical methodologies that allow the rapid analysis of various types of water samples without the need for timely and unreliable preconcentration and cleanup steps. From previous studies the only microcystins prevalent in Irish waters are MC-LR, MC-RR, MC-YR, and MC-LA;7,18,40 therefore, it is prudent for the protection of livestock and human health, to develop under this Environmental Protection Agency-supported project a rapid and dependable method to determine these microcystins in lake and drinking tap water samples. Chromatographic Optimization Using the Cartridge System. When performing the analysis of compounds like microcystins, the use of reversed-phase chromatography is logical but the suite of known microcystins encompasses quite a large variation in polarity. As a consequence, chromatographic run times and column equilibration times can be quite long, especially when full chromatographic resolution is required. When using massresolved analyses, such as MRM experiments on a triple quadrupole (QqQ) mass spectrometer, the requirement for peak resolution is less important. What is important however is the elution of the analyte in a narrow peak. Earlier analyses using conventional analytical columns required long run times, even when using very sharp gradients.2,18 Monolithic columns reduce run times but have the disadvantage of high flow rates, which is limiting for electrospray; however, mobile-phase splitting alleviates this problem. The methodology reported here employs a reversed(46) Bonfiglio, R.; King, R.; Olah, T.; Merkle, K. Rapid Commun. Mass Spectrom. 1999, 13, 1175-1185. (47) Codd, G. A.; Bell, S. G. J. Water Pollut. Control 1985, 34, 225-232. (48) Carmichael, W. W. In The Water Environment: Algal Toxins and Health; Carmichael, W. W., Ed.; Environment Science Research 20; Plenum Press: New York, 1981; pp 1-13. (49) Carmichael, W. W. Sci. Am. 1994, 270, 64-72.

3440 Analytical Chemistry, Vol. 79, No. 9, May 1, 2007

Figure 2. Graphical representation of the optimization of the organic solvent (HPLC grade acetonitrile and methanol) and modifier (trifluroacetic acid (TFA), ammonium acetate (NH4Ac), nonafluoropentanoic acid (NFPA), heptafluorobutanoic acid (HFBA), and pentafluoropentanoic acid (PFPA)) for mobile-phase selection. This figure illustrates the data for MC-LR standard (100 µg/L) with a sample number equal to 7 (n ) 7).

phase guard cartridge system as the analytical column. This cartridge system was designed to act as a guard column prior to analytical chromatography separation of analytes on an analytical column. However, on examination, it was found that this C18 cartridge independently gave adequate chromatographic separation to a mixture of microcystin toxins. Mobile-Phase Study. Methanol and acetonitrile were investigated in order to determine the best solvent, and five different modifiers, 0.05% trifluroacetic acid, 1 mM ammonium acetate, 0.05% NFPA, 0.05% HFBA, and 0.05% PFPA, were assessed to ascertain the best mobile-phase modifier. Chromatographically all scenarios performed well (data not shown); therefore, the choice of mobile phase was made based upon observed sensitivity of the microcystins during analysis. Figure 2 highlights the effect of different solvents and mobile-phase modifiers upon the response of MC-LR using the LC-MS/MS methodology developed during this study. It is quite clear that methanol facilitated a much greater response, invariably due to better ionization, than acetonitrile. Additionally, Figure 2 indicates that ammonium acetate improves the response of MC-LR with respect to the other modifiers used in this study. Method Validation. In August 2002, the European Commission (EC) devised a set of guidelines pertaining to the performance of analytical methods and the interpretation of results.45 These guidelines were originally established for the identification and quantification of organic residues and contaminants in animals and in fresh meat, but have recently been found to be useful for the confirmation of pesticides in water samples50 and antibiotics in surface waters.51 The EC criteria are based on the application of identification points (IPs). These points are allocated for the use of different molecular spectrometric techniques or a combination of techniques used for the identification of a compound. The report indicates that a minimum of four IPs is necessary for the identification of a compound and that any MS technique or combination of techniques can be used to achieve this. The commission decision document details the number of identification points that can be earned” by the detection of a precursor/product (50) Hernandez, F.; Ibanez, M.; Sancho, J. V.; Pozo, O. J. Anal. Chem. 2004, 76, 4349-4357. (51) Pozo, O. J.; Guerrero, C.; Sancho, J. V.; Iba´n ˜ez, M.; Pitarch, E.; Hogendoorn, E.; Herna´ndez, F. J. Chromatogr., A 2006, 1103, 83-93.

Figure 3. Sample matrix ion-suppression/enhancement assessment: postcolumn infusion of MC-LR (100 µg/L) in the presence of nontoxic water matrixes; (A) filtered lake water and (B) filtered chlorinated tap water. The gray shaded area represents the typical retention time of MC-LR.

ion; this is also dependent on the technique used.45 For example, the number of IPs earned using low-resolution mass spectrometry and obtaining two precursor ions, each with one product ion, is five. Overall, these criteria show that adequate confirmation can be made in low-resolution instruments (e.g., QqQ) acquiring two MS/MS transitions and measuring the ratio between them. This increase in the number of MS transitions monitored greatly helps in confirming the correct identity of compounds. The overall advantage of using the IP system is that the confirmation of identity can be achieved by an internationally accepted protocol. These EU guidelines can also be applied to developed LC-MS/ MS methods for the analysis of hepatotoxins in water samples to ensure that correct and accurate toxin identification is obtained at the provisional regulatory limit of 1 µg/L (MC-LR). (a) Matrix Effects. “Matrix effect” is a term that is used to describe any change in the mass spectrometry response of an analyte that is a result of the specific matrix of the sample being assayed. Matrix effects can lead to either a reduced response (ion suppression) or an increased response (ion enhancement) of the

mass spectrometry system.52 These effects can severely compromise quantitative analysis of environmental samples using LCESI-MS. Three different types of experiments were completed in order to investigate any matrix effects in this LC-ESI-MS method: postcolumn infusion, standard addition, and percentage recoveries. Postcolumn infusion is one of the best techniques used to observe any matrix effects. King et al.52 investigated ionization suppression in electrospray ionization and concluded that ionization suppression is the result of high concentrations of nonvolatile materials present in the spray with the analyte. These nonvolatile solutes are typically salts such as sulfates and phosphates, but King and co-workers suggested that these solutes can also include the analyte. Therefore, an infusion experiment was completed using this LC-MS/MS method. A series of postcolumn infusion experiments using standard microcystins and nodularin were undertaken. Essentially, the responses of the infused toxins were monitored during the analysis of blank LC grade water, toxin(52) King, R.; Bonfiglio, R.; Fernandez-Metzler, C.; Miller-Stein, C.; Olah, T. J. Am. Soc. Mass Spectrom. 2000, 11, 942-950.

Analytical Chemistry, Vol. 79, No. 9, May 1, 2007

3441

free lake water, and toxin-free chlorinated tap water. Figure 3 shows that none of the matrixes analyzed had any effect on the response, either positive (ion enhancement) or negative (ion suppression), of the toxins that were infused postcolumn. In terms of MC analysis, the most complex matrix encountered are those from cyanobacterial cells; it was not within the scope of this study to include these samples, and our primary objective was the assessment of the safety of water delivered through the taps directly into peoples’ homes. This methodology has been developed under a funded EPA project to detect dissolved (extracellular) microcystins in water samples devoid of all cyanobacterial cells. A standard addition experiment was performed on an Irish lake water sample, Town Lake, Killeshandra, Co. Cavan. This sample was previously analyzed using the optimized method and was found to contain 20.27 µg/L MC-LR. This standard addition experiment was repeated five times (n ) 5), and from the extrapolated calibration plot (Figure 4), the concentration of MCLR was found to be 20.95 µg/L. This standard addition method of analysis also confirms that no matrix effects transpire using real lake water samples. Percentage recovery experiments were also performed by spiking known concentrations of seven different hepatotoxins into two different matrixes: lake water and chlorinated tap water. The actual concentrations were calculated from a calibration curve of hepatotoxin standards in LC grade water. From Table 3, it is evident that the percentage recovery of each toxin in lake water was excellent, with g98.4% recovery obtained for MC-LR over the concentration range (0.5-200 µg/L). The percentage recovery of the hepatotoxins in chlorinated tap water was also excellent, with g90% obtained for MC-LR over the wide range 0.5-200 µg/ L. The use of a high injection volume (100 µL) had no negative impact on percentage recoveries both in the lake and in the chlorinated tap water. These excellent percentage recoveries for lake water and chlorinated tap water corroborate with the results from the postcolumn infusion and standard addition experiments,

Table 3. Matrix Effects: Percentage Recovery of Seven Hepatotoxins (MC-LR, MC-RR, MC-YR, MC-LW, MC-LF, MC-LA, and Nodularin) Spiked Into Two Different Matrixes, Lake Water and Chlorinated Tap Watera lake water microcystin analyte MC-LR

linear range low normal

MC-RR

low normal

MC-YR

low normal

MC-LF

normal

MC-LW

normal

MC-LA

normal

nodularin

low normal

chlorinated tap water

spiked conc recovery RSD recovery RSD (µg/L) (%) (%) (%) (%) 0.50 1.00 1.50 50.00 100.00 200.00 0.50 1.00 1.50 50.00 100.00 200.00 0.50 1.00 1.50 50.00 100.00 200.00 50.00 100.00 200.00 50.00 100.00 200.00 50.00 100.00 200.00 0.50 1.00 1.50 50.00 100.00 200.00

99.6 98.4 99.0 99.8 98.9 98.7 100.0 99.8 99.2 100.0 99.5 98.8 99.7 98.4 100.0 99.8 99.5 93.5 99.3 100.0 99.9 92.8 99.6 93.6 90.7 99.7 99.4 99.6 100.0 99.5 99.4 93.8 99.8

8.9 9.4 7.6 9.7 10.0 8.6 10.0 6.3 3.9 7.8 3.7 2.9 9.5 4.7 2.1 8.8 7.4 2.3 10.0 6.6 4.2 4.0 6.2 6.4 6.5 4.2 8.9 4.3 8.5 5.9 5.2 3.8 7.5

99.6 89.9 98.1 99.2 99.9 100.0 99.0 95.2 98.5 99.8 99.6 100.0 83.8 88.6 89.4 100.0 97.0 93.3 90.3 90.3 93.6 96.5 90.6 86.9 92.5 86.9 84.9 99.1 99.8 99.4 99.0 91.9 99.8

5.2 7.7 2.7 5.9 4.0 7.6 9.5 5.2 4.7 6.3 4.5 3.3 7.8 5.8 4.4 7.7 5.5 6.7 14.0 12.0 8.5 10.0 9.5 5.1 4.6 11.0 8.0 6.9 7.6 8.1 8.9 4.0 5.7

a The spiked concentrations were 0.50, 1.00, 1.50, 50.0, 100, and 200 µg/L and the RSDs are also shown (n ) 5).

that conventional water treatment processes, i.e., coagulation, sedimentation, and filtration, are not adequate for the complete or satisfactory removal of dissolved microcystins.53-55 Powdered activated carbon has also been studied as a technique to remove microcystins; however, Cook and Newcombe found that some microcystin analogues (MC-LA) are not readily removed using activated carbon while others (MC-RR, MC-YR) are very efficiently removed.56 Oxidation studies using ozone,57,58 chlorine,59,60 and UV photolysis in the presence of TiO261 have shown that MC-LR is readily oxidized to nontoxic degradation products under Figure 4. Standard addition calibration plot for a lake water sample (Town Lake, Killeshandra, Co. Cavan) for the determination of MCLR concentration (20.95 µg/L).

confirming that matrix effects do not pose a problem with this LC-MS/MS method of analysis for hepatotoxins. (b) Assessment of Free Chlorine and Microcystin Degradation in Water. Microcystins are said to be removed from natural waters using the typical chemicals (oxidants and disinfectants) applied to community waterworks. However, it is clear 3442 Analytical Chemistry, Vol. 79, No. 9, May 1, 2007

(53) Hoffman, J. R. H. Water SA 1976, 2, 58-60. (54) Himberg, K.; Keijola, A.-M.; Hiisvirta, L.; Pyysalo, H.; Sivonen, K. Water Res. 1989, 23, 979-984. (55) Ho, L.; Onstad, G.; von Gruten, U.; Rinck-Pfeiffer, S.; Craig, K.; Newcombe, G. Water Res. 2006, 40, 1200-1209. (56) Cook, D.; Newcombe, G. Water Sci. Technol.: Water Supply 2002, 2, 201207. (57) Rositano, J.; Nicholson, B. C.; Pieronne, P. Ozone Sci. Technol. 1998, 20, 223-238. (58) Rositano, J.; Newcombe, G.; Nicholson, B. C.; Sztajnbok, P. Water Res. 2001, 20, 223-238. (59) Nicholson, B. C.; Rositano, J.; Burch, M. D. Water. Res. 1994, 28, 12971303.

appropriate conditions. As chlorine is the principle preoxidation treatment and final disinfectant used worldwide, a number of significant studies have been carried out on the degradation of microcystins using chlorine.59,62-65 Nicholson et al.59 determined that if a chlorine residual of at least 0.5 mg/L was present, oxidation of MC-LR occurred after 30-min contact time at a pH below 8. Tsuji et al.63 reported that, at a pH of 7.2, the removal of MC-LR after a 60-min contact time was 35% at a free chlorine dose of 0.7 mg/L, 72% at a dose of 1.4 mg/L, and 100% at a dose of 2.8 mg/L. A concentration of 2.8 mg/L free chlorine for 30 min effectively removed 99% of MC-LR, which concurrent with the previous study by Nicholson et al.,59 indicated that MC-LR readily decomposes by chlorination and the decomposition depends on the free chlorine dose with no noxious products being detected from the chlorination process. The significance of pH was further reinforced through a detailed kinetic study by Acero et al.,64 who reported that the degradation of MC-LR is favored at low pHs. However, monochloramine, which can be formed during chlorination of ammonium-containing waters, is not capable of oxidizing microcystins. For this study, the free chlorine level was determined in chlorinated tap water over 60 h. Both the Hach Lange and Riedelde Hae¨n DPD reagents gave compariable results ((1%). The average pH value for the chlorinated tap water samples prior to analysis was pH 7.50 at 11 °C, and the average free chlorine levels detected throughout this study were 0.23 ( 0.003 mg/L from water taken directly from the tap, which decreased to 0.10 mg/L after 60 h. To compare the effect of free Cl2 on the oxidation of MC-LR with previously published data,59,63-65 an in-house study was initiated to determine the rate of decomposition of MC-LR in the presence of different concentrations of Cl2. Chlorine spiked (0.1 and 0.2 mg/L) into LC water containing 10 mg/L MC-LR showed no reduction in the peak area of MC-LR (n ) 3) over 48 h. A 10% reduction in MC-LR response was observed on addition of 0.5 mg/L chlorine (n ) 3) over 48 h. A 45% reduction in MCLR response was observed on addition of 1.0 mg/L chlorine (n ) 3) over 48 h. A 60% reduction in MC-LR response was observed on addition of 2.0 mg/L chlorine (n ) 3) over 48 h. These data show the necessity to screen chlorinated tap water for microcystin levels in Ireland, as both these data and those from published studies59,60,63-65 have shown that complete oxidation of the microcystins does not rapidly occur at average free chlorine levels of 0.23 mg/L in tap water. (c) Specificity. A key factor in any analytical procedure is its ability to demonstrate specificity for a particular analyte. It is often necessary to have a combination of two or more analytical procedures in order to achieve the required level of discrimination. (60) Nicholson, B. C.; Rositano, J. In Workshop on cyanobacteria (blue-green algae) and their toxins, Brisbane, Australia, 1997; Australian Water and Wastewater Association: Sydney, Australia. 1997. (61) Lawton, L. A.; Robertson, P. K. J.; Cornish, B. J. P. A.; Jaspars, M. Environ. Sci. Technol. 1999, 33, 771-775. (62) Bull, R. J.; Birnbaum, L. S.; Cantor, K. P.; Rose, J. B.; Butterworth, B. E.; Pegram, R.; Tuomisto, J. }. Appl. Toxicol. 1995, 28, 155-166. (63) Tsuji, K.; Watanuki, T.; Kondo, F.; Watanabe, M. F.; Nakazawa, H.; Suzuki, M.; Uchida, H.; Harada, K.-I. Toxicon 1997, 35, 1033-1041. (64) Acero, J. L.; Rodriguez, E.; Meriluoto, J. Water Res. 2005, 39, 1628-1638. (65) Shi, H.-X.; Qu, J.-H.; Wang, A.-M.; Ge, J.-T. Chemosphere 2005, 60, 326333. (66) Yuan, M.; Namikosh, i. M.; Otsuki, A.; Rinehart, K. L.; Sivonen, K.; Watanabe, M. F. J. Mass Spectrom. 1999, 34, 33-43.

Identification tests should be able to discriminate between compounds of closely related structures that are likely to be present. In this method, both retention time stability and precursor/product ion pairs were utilized for specificity purposes. The chromatography demonstrated by using the rapid gradient in this methodology does not allow for complete chromatographic resolution of each of the seven hepatotoxins. However, identification by retention time can be used as a complimentary method in demonstrating specificity for each hepatotoxin because of the high selectivity of MRM. Stability in retention time for each hepatotoxin in both lake and tap water over a 3 day period with five repeat injections each day was excellent. The EC criteria have recommended that the retention time for a compound should not shift by more than 5%.45 The retention time variation of the hepatotoxins in lake and chlorinated tap water is well within this limit with a retention time shift of