In vitro assessment of modes of toxic action of pharmaceuticals in aquatic life.

An ecotoxicological test battery based on a mode-of-action approach was designed and applied to the hazard identification and classification of modes of action of six pharmaceuticals (carbamazepine, diclofenac, ethinyl estradiol, ibuprofen, propranolol, and sulfamethoxazole). The rationale behind the design of the battery was to cover the relevant interactions that a compound may have with biological targets. It is thus not comprehensive but contains representative examples of each category of mode of toxic action including nonspecific, specific, and reactive toxicity. The test battery consists of one test system for nonspecific toxicity (baseline toxicity or narcosis), two test systems for specific effects, and two test systems for reactive toxicity. The baseline toxicity was quantified with the Kinspec test, which detects membrane leakage via measurements of membrane potential. This test system may also be used to detect the specific effects on energy transduction, although this was not relevant to any compound investigated in this study. As examples of specific receptor-mediated toxicity, we chose the yeast estrogen screen (YES) as a specific test for estrogenicity, and the inhibition of chlorophyll fluorescence in algae to assess specific effects on photosynthesis. Reactive modes of action were assessed indirectly by measuring the relevance of cellular defense systems. Differences in growth inhibition curves between a mutant of Escherichia coli that could not synthesize glutathione and its parent strain indicate the relevance of conjugation with glutathione as a defense mechanism, which is an indirect indicator of protein damage. DNA damage was assessed by comparing the growth inhibition in a strain that lacks various DNA repair systems with that in its competent parent strain. Most compounds acted merely as baseline toxicants in all test systems. As expected, ethinylestradiol was the only compound showing estrogenic activity. Propranolol was baseline-toxic in all test systems exceptforthe photosynthesis inhibition assay, where it surprisingly showed a 100-fold excess toxicity over the predicted baseline effect. The exact mode of toxic action could not be confirmed, but additional chlorophyll fluorescence induction experiments excluded the possibility of direct interference with photosynthesis through photosystem II inhibition. Mixture experiments were performed as a diagnostic tool to analyze the mode of toxic action. Compounds with the same mode of toxic action showed the expected concentration addition. In the photosynthesis inhibition assay, agreement between experimental results and prediction was best for two-stage predictions considering the assigned modes of action. In a two-stage prediction, concentration addition was used as a model to predict the mixture effect of the baseline toxicants followed by their independent action as a single component combined with the specifically acting compound propranolol and the reference compound diuron. A comparison with acute toxicity data for algae, daphnia, and fish showed generally good agreement for the nonspecifically acting compounds but also that the proposed test battery offered better diagnostic value in the case of the specifically acting compounds.