ABSTRACT
In silico methodologies can be used in the discovery of new drugs for measuring toxicity, predicting effects of substances not yet analyzed by in vivo methodologies. The ADMET Predictor® software (absorption, distribution, metabolism, elimination, and toxicity [ADMET]) was used in this work to predict toxic effects of microcystin variants MC-LR, MC-YR, MC-RR, and MC-HarHar. In the case of rodents, predictive results for all analyzed variants indicated carcinogenic potential. The predictive model of respiratory sensitivity in this group differentiated microcystins into 2 categories: sensitizer (MC-LR and -YR) and non-sensitizer (MC-HarHar and -RR). Predictive results for humans indicated that MC-LR and -RR are phospholipidosis inducers; on the other hand, MC-LR showed the highest predictive value of permeability in rabbit cornea and probability of crossing lipoprotein barriers (MC-LR>-YR>-HarHar>-RR). Considering bioavailable fractions, microcystins are more likely to cause biological effects in rats than humans, showing significant differences between models. The results of ADMET predictions add valuable information on microcystin toxicity, especially in the case of variants not yet studied experimentally.
