Claudiu Lungu
A series of compounds with known experimental IC50 (nM) targeting phosphodiesterase10A (PDE10A) was studied. Recently PDE10A was proposing as a colon cancer drug target. A QSAR model was build using the compounds having as target variable their inhibitory effect on PDE10A expressed as IC50 (nM). A multiple correlation technique was used in order to select the appropriate descriptors for building a regression model. Descriptors used were functional group base descriptors and some centrality descriptors. The regression model was build using artificial neural network regression (ANN). A model with, r2=0.9769, and a standard error deviation of 0.41 was build. Model was used to predict IC50 (nM) for a series of screening resulted compound. Template used for screening was established by generating a hypothesis using the common pharmacophore. The pharmacopohore hypothesis was build using functional groups displacement criteria. Hypothesis resulted was used for virtual screening. Compounds resulted were classified using a score. Best 17 compounds where chosen (when score decreased with 1/3 of best value). A comparison between best PDE10A inhibitor cited in the literature, best dataset compound with inhibitory effect on PDE10A and best compound resulted after screening with best IC50 predicted were analyzed. All three structures were analyzed in complex with PDE10A. Poses were generated using docking. Results demonstrated the importance of Pi-Pi bounds with Phe 696 as being crucial in PDE10A inhibition. The conclusion is sustained by both QSAR model and the common pharmacophore hypothesis.
PDFShare this article
Chemical Sciences Journal received 912 citations as per Google Scholar report
