Drug repositioning for the treatment of Chagas diseases. Application of machine learning methods for the discovery of new drugs.

ALBERCA, LUCAS NICOLÁS; RUIZ, DANIELA; MORALES, JUAN FRANCISCO; SBARAGLINI, MARÍA LAURA; CARRILLO, CAROLINA; TALEVI, ALAN

San Diego

Exposici�n; SLAS 2018 International Conference and Exhibition; 2018

Society for Laboratory Automation and Screening (SLAS)

Drug repositioning (or drug repurposing) involves finding new therapeutic indications for approved, discontinued or experimental drugs. This strategy has considerable advantages over the search for de novo drugs since the cost and time required for the approval of an innovative medication are greatly reduced. It is a particularly useful strategy to search new treatments for neglected diseases such as Chagas disease, an infectious disease caused by Trypanosoma cruzi that affects more than 6 millons people in Latin America. In this work, we have developed in silico models capable of recognizing inhibitors of N-myristoyl transferase (NMT), that catalyzes the transfer of myristate from myristoyl coenzyme A to the amino terminal glycine residue of the protein target. This enzyme has been genetically and biochemically validated as a molecular target for T. cruzi, although scarcely exploited to the moment for the search of drug candidates against Chagas disease. Through an extensive literature search we have compiled a database comprising compounds previously tested against trypanosomatid NMT (including NMT from species of the Trypanosoma genre). Using this database, we have inferred and validated 1000 computational linear classifiers capable of discriminating between ?inhibitors? and ?non-inhibitors? of these molecular targets. The models were generated by machine learning methods using the R environment. The best individual models were combined by different ensemble learning approaches. The best ensembles were applied in the virtual screening of Drug Bank 4.0 and Sweetlead databases, which compile drugs already approved by international regulatory agencies. After submitting the resulting hits to applicability domain assessment and using accessibility as additional selection criterion, 3 drugs were purchased for in vitro evaluation: Dicyclomine (for the treatment of irritable bowel syndrome), quinestrol (used in hormone replacement therapy) and danazol (for the treatment of endometriosis and fibrocystic breast disease). These 3 drugs showed a strong trypanocidal activity on T. cruzi epimastigotes. These results reflect the ability of computer guided drug repositioning to identify novel trypanocidal compounds with a relatively small investment of time and resources.