In Silico Drug Repurposing Framework Predicts Repaglinide, Agomelatine and Protokylol as TRPV1 Modulators with Analgesic Activity

Pain is one of the most common symptoms experienced by patients. The use of current analgesics is limited by low efficacy and important side effects. Transient receptor potential vanilloid-1 (TRPV1) is a non-selective cation channel, activated by capsaicin, heat, low pH or pro-inflammatory agents. Since TRPV1 is a potential target for the development of novel analgesics due to its distribution and function, we aimed to develop an in silico drug repositioning framework to predict potential TRPV1 ligands among approved drugs as candidates for treating various types of pain. Structures of known TRPV1 agonists and antagonists were retrieved from ChEMBL databases and three datasets were established: agonists, antagonists and inactive molecules (pIC50 or pEC50 < 5 M). Structures of candidates for repurposing were retrieved from the DrugBank database. The curated active/inactive datasets were used to build and validate ligand-based predictive models using Bemis-Murcko structural scaffolds, plain ring systems, flexophore similarities and molecular descriptors. Further, molecular docking studies were performed on both active and inactive conformations of the TRPV1 channel to predict the binding affinities of repurposing candidates. Variables obtained from calculated scaffold-based activity scores, molecular descriptors criteria and molecular docking were used to build a multi-class neural network as an integrated machine learning algorithm to predict TRPV1 antagonists and agonists. The proposed predictive model had a higher accuracy for classifying TRPV1 agonists than antagonists, the ROC AUC values being 0.980 for predicting agonists, 0.972 for antagonists and 0.952 for inactive molecules. After screening the approved drugs with the validated algorithm, repaglinide (antidiabetic) and agomelatine (antidepressant) emerged as potential TRPV1 antagonists, and protokylol (bronchodilator) as an agonist. Further studies are required to confirm the predicted activity on TRPV1 and to assess the candidates' efficacy in alleviating pain.

Automated summary

This study aimed to develop a computational framework to predict potential TRPV1 ligands among approved drugs for treating different types of pain. The researchers built predictive models and validated them using molecular docking and neural network algorithms. After screening and validation, certain approved drugs were identified as potential TRPV1 agonists or antagonists. However, further research is needed to confirm the predicted activity and evaluate the efficacy of these candidate drugs in alleviating pain.