Analysis of individual and total microcystins in surface water by on-line preconcentration and desalting coupled to liquid chromatography tandem mass spectrometry

• Published in: Journal of Chromatography A 2017-09, Vol.1516, p.9-20
• Main Authors: Munoz, Gabriel, Vo Duy, Sung, Roy-Lachapelle, Audrey, Husk, Barry, Sauvé, Sébastien
• Format: Article
• Language: English
• Subjects: Cyanotoxins; Lemieux oxidation; Matrix effect assessment; Method validation; On-line enrichment and desalting; Surface water; Total microcystins via MMPB
• ISSN: 0021-9673
• DOI: 10.1016/j.chroma.2017.07.096

•Individual and total microcystins were screened in surface water samples from Québec, Canada.•Lemieux oxidation-based samples were analyzed by on-line SPE – UHPLC-MS/MS in just 8min.•In view of the matrix complexity, the on-line desalting was a key step to ensure analytical method robustness.•Quantification performances were examined for various surface water matrixes.•Matrix effects were efficiently compensated using 4-PB or D3-MMPB standardization. A fast and high-throughput method is proposed for the determination of total microcystins (ΣMC) in environmental surface waters. After a 1-h Lemieux-von Rudloff oxidation step to yield the 2-methyl-3-methoxy-4-phenylbutyric acid (MMPB) moiety, samples were quenched, filtered, and directly analyzed. This was achieved via solid phase extraction (SPE) coupled on-line to ultra-high performance liquid chromatography electrospray ionization triple stage quadrupole mass spectrometry. The choice of on-line SPE settings was conducted using experimental designs. Given the matrix complexity of oxidation extracts, the on-line desalting step was found to be a critical parameter to ensure suitable method robustness. The on-line sample loading volume was 5mL, and the wash volume applied for on-line desalting was 3mL. Instrumental analysis was performed in just 8min. The method limit of quantification was 0.5ngL−1 ΣMC (i.e. 2000 times lower than the current World Health Organization − WHO drinking water guideline). Excellent determination coefficients were observed for matrix-free and matrix-based calibration curves alike, and the linearity range tested spanned∼4 orders of magnitude. Accuracy and intermediate precision did not depend on the spike level and proved satisfactory (in the range of 93–110% and 3–6%, respectively). A thorough assessment of instrumental matrix effects was conducted by comparing standard additions curves in several lake and river oxidation extracts with the matrix-free reference. Regardless of the internal standard used (4-PB or D3-MMPB), instrumental matrix effects were efficiently compensated. The matrix effect that may occur at the earlier sample preparation stage was evaluated separately. While the oxidation step was generally not complete (yield ∼65%), the conversion rates of MCs into MMPB remained within a consistent range of values regardless of matrix type. No significant back-pressure was observed upon consecutive injections of oxidation-based samples, while the instrumental sensitivity rema