Validation d'une méthode d'analyse des microcystines par SPE/CLHP/UV-DAD

Validation of analysis method of microcystins by SPE/HPLC/UV-DAD

¹ Ecole d'application du service de santé des armées, 1, place Alphonse Laveran, 75230 Paris Cedex 05
² Centre de recherche d'expertise et de contrôle des eaux de Paris, 144, avenue Paul Vaillant Couturier, 75014 Paris

Résumé

Une des conséquences de l'eutrophisation des plans d'eau, amplifiée par les impacts anthropiques, est le développement en période estivale de " fleurs d'eau " ou " blooms " de cyanobactéries ou algues bleues-vertes. Les cyanobactéries apparaissent dans les lacs, rivières, étangs, et réservoirs d'eau douce du monde entier ; elles produisent des toxines dangereuses pour les échelons supérieurs de la pyramide trophique : les microcystines, heptapeptides monocycliques dont la cible est la cellule hépatique.

L'intérêt croissant suscité par la présence de toxines de cyanobactéries dans l'eau, plus spécialement les microcystines, explique le besoin d'une technique analytique. Le Code de la Santé Publique introduit la recherche de microcystine-LR dans les eaux destinées à la consommation humaine.

Ce travail a permis de mettre au point une méthode d'extraction en phase solide des microcystines -LR et -RR avant de les séparer et de les quantifier par chromatographie liquide haute performance (CLHP) couplée à un détecteur à barrette de diodes. Les différents essais réalisés ont permis de déterminer les conditions d'analyse optimales et de caractériser les performances de la méthode, linéaire dans un domaine de concentration de 0,05 à 4 µg.l^-1, spécifique, répétable, reproductible, et permettant d'atteindre une limite de quantification 0,05 µg.l^-1.

Abstract

One of the consequences of eutrophisation of lakes, increased by human activities is the development in summer of cyanobacteria blooms or blue-green algae. Cyanobacteria appear in lakes, rivers, and water supply all over the world. They produce dangerous toxins for the top level of the trophic pyramid: the microcystins heptapeptides whom aim is the hepatic cell.

The increased interest in presence of these toxins explains the necessity of an analytical technique. The Code of Public Health introduces the research of microcystin-LR in drinking water.

This work develops a method of extraction of microcystins -LR and -RR, before separating them and quantifying by HPLC/UV-DAD. The tests realized determine optimum analytical conditions. The pre-treatment of the sample has been tested: the results of a step of acidification are not satisfiying. The pre-treatment consists on addition of 100 µl of sodium thiosulfat and 5 ml of methanol. The extraction has also been optimised ; the protocol of solid phase extraction (SPE) is : conditioning by 10 ml of methanol and 10 ml water, deposit of sample at a flow inferior at 10 ml/min, dryness of cartridge by aspiration of air during 30 minutes and elution by 5 ml of methanol acidified by 1 % trifluoroacetic acid. A step of cleaning by 5 ml of the mixing methanol/water 80/20 (v/v) has not been retained. The elution has been tested; we obtained better results with methanol acidified by 1 % of trifluoroacetic acid. After concentration, the separation is realised by HPLC on C18 column with an elution gradient. Different essays have been realised to determine the mobile phase : it is Na₂HP0₄, 12 H₂O, acidified at pH 2,5 by perchloric acid. The identification is realised by a DAD between 200 and 350 nm and quantification in effected at 238 nm, maximum of absorption of microcystins.

The method performances are characterized according to the norm AFNOR T 90-210. Linearity is tested on work domain de travail: 0,05 - 4 µg.l^-1 by a test of adequation to the linear mode, a Fisher test for a risk of 1 %. The response is linear from 0,05 to 4 µg.^-1; the method is specifie, repeatable and reproducible. This method reach limits of quantification of 0,05 µg.l^-1.

The regression linear model is acceptable for the microcystins -LR and -RR and the calibration domain can be validated, the model error being negligible. The validity of linear model is verified. The chosen limit of quantification: 0,05 µg.l^-1 is verified by justness critter and fidelity critter. The limit of detection is 0,017 µg.l^-1. The specificity of the method is tested on three different matrixes: ultra-pure water, tap water and filtered raw water. The repeatability is stable on the work domain: 0,05 - 4 µg.l^-1. The method is just and the internal reproducibility is verified. The developed method is validated.