Synthesis and biological evaluation against Cruzipain of new compounds based on 2D Virtual Screening

CECILIA SAIZ, CHIARA PIZZO, ALAN TALEVI, CAROLINA L. BELLERA, L. GAVERNET, LUIS BRUNO BLANCH, JUAN J. CAZZULO, AGUSTINA CHIDICHIMO, EDUARDO MANTA, PETER WIPF, GRACIELA MAHLER

Trieste, Italia

Conferencia; 2nd Conference on Drug Development for the Third World: From Computational Molecular Biology to Experimental Approaches; 2009

International Centre for Theoretical Physics

Chagas disease is endemic throughout much of Central America, and South America where an estimated 8 to 11 million people are infected. Unfortunately, despite the impressive advances in understanding the biology of T. cruzi, the only drugs currently available against this organism are those that were already registered 21 years ago, nifurtimox and benznidazole. These compounds are active only in the acute state of Chagas disease and the side effects of both can be severe. The growing knowledge of the basic biology of Trypanosome cruzi facilitates the development of new, rationally developed approaches to Chagas disease-specific chemotherapy. The cathepsin L-like cysteine protease termed cruzipain, is responsible for the major proteolytic activity of all stages of the parasite life cycle and it is an interesting target for the development of potential therapeutics for the treatment of the disease. Cruzipain is differentially expressed in the four main stages of the parasite’s biological cycle, and it is essential for the survival of T. cruzi within host cells. Virtual screening (VS) methodology has proven to be very useful in the discovery and development of new drugs for neglected diseases. The present work describes the selection and synthesis of novel potential cruzipain inhibitors based on computational tools. A Virtual Screening (VS) strategy was applied over on 537,503 chemical structures from ZINC 5 database to select new cruzipain inhibitors. The methodology included a 2D QSAR approach based on application of linear discriminant analysis (LDA) and Multiple Linear Regression (MLR) on constitutional and topological descriptors from Dragon (Milano Chemometrics, 2003) , to derive five discriminant functions (DF1-5) and two QSAR models (Q1 and Q2) for the prediction of inhibitory activity on cruzipain. The models were internally validated through leave-group-out and randomization tests and externally validated with independent sets of compounds. Descriptor ranges methodology was used for Applicability Domain estimation. Selected compounds were also ranked according to the number of neighbors in the training set with Tanimoto similarity coefficient (based on atom pairs) > 0.5, > 0.7 and > 0.8. Some of the compounds selected by VS were synthesized or purchased and their inhibitory activity on cruzain was assayed. One of the selected compounds showed a weak inhibition on the enzyme and in vitro tests against epimastigotes of T. cruzi,