Machine learning-assisted in silico discovery of PDE10A Inhibitors: Integration of QSAR modeling, docking and MD simulations

Interest in phosphodiesterase 10A (PDE10A) inhibitors has been steadily increasing, particularly for their potential in the treatment of schizophrenia. While medicinal chemists have made considerable efforts to design effective inhibitors with minimal side effects, none have been approved for neurodegenerative disorders. This may be due to current research gaps in this field. In this study, we used an In silico approach to evaluate 69 novel pyrimidine derivatives as PDE10A inhibitors, using a QSAR approach. The results of this study showed an R 2 = 0.9097, CCCtr = 0.9527, R 2 Yscr = 0.1022 R2ext = 0.8353. The best generated models encompass five important variables, including ATSC8m, ATSC3e, MATS3s, SHCsats, and WPATH. The parameters including the mass, Sanderson electronegativities, I-state, Weiner path number and saturated carbon played an important role in designing new lead compounds. Based on the above data, we have designed 10 compounds and predicted their activity as PDE10A inhibitors; the highest-ranking compounds were further studied using molecular docking protocols with PDE10A protein sequence reported in the protein data bank (PDB ID 2OVY). The best 6 compounds showing a good inhibitory profile with strong binding interactions in the active site of 2OVY were additionally studied using predictive models of ADMET and DFT, and the results showed an increased stability in the drug complexes due to a larger HOMO-LUMO gap. MD simulations over 100ns further validated the stability of the best complexes (A2, A4 and A9) via RMSD, RMSF, Rg and SASA analyses. Notably, these compounds maintained stable interactions with key active-site residues such as Phe283, Phe254, and Ile246 throughout the simulation, reinforcing their binding stability. Based on these findings, compounds A2, A4 and A9 are proposed as promising leads for further in vitro and in vivo validation as PDE10A inhibitors for schizophrenia treatment. • A validated QSAR model was developed to predict PDE10A inhibitory activity of novel pyrimidine derivatives. • Five key molecular descriptors were identified as major contributors to activity prediction. • Designed compounds demonstrated strong binding affinity and favorable interactions in molecular docking studies. • ADMET, DFT, and molecular dynamics simulations confirmed their drug-like potential and structural stability. • The integrated in silico strategy supports the identification of promising leads for schizophrenia therapy.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.