RESUMEN
The lack of predictivity of animal's models has increased the failure rate of drug candidates. Thus, the reversion of this scenario using preliminary in vitro assays and metabolism prediction can reduce the unnecessary use of animals, as well as predict toxic effects at preclinical and clinical stages. The present study aimed to evaluate safety of four biologically active molecules (RN104, RI78, ICH, PCH) with potential therapeutic applications synthesized in our laboratory. Initially, we used MTT cytotoxicity against A549, H9C2, HepG2, LLC-PK1 and NEURO-2 cell lines. RN104 showed the lowest cytotoxicity and further studies were conducted with it. The neutral red (NR) test was performed according to OECD-129 and then acute toxicity test (OECD-423). According to NR results we administered at 300 mg/kg on animals; however, no toxic effect was observed, while 2,000 mg/kg resulted in the death of one animal per group. After, metabolism prediction studies, performed using both ligand-based and structure-based, suggests three potential metabolites. In silico results suggested that potential metabolites could be fast eliminated and, then, this could be an explanation for lower observed toxicity in in vivo experiments. The results showed limitations of the NR as a predictor of the initial dose for the acute toxicity study, which may be related to metabolism. Therefore, the combination of theoretical and experimental studies is relevant to a general understanding of new molecule's toxicity.