Pinpointing potent hits for cancer immunotherapy targeting the TIGIT/PVR pathway using the XGBoost model, centroid-based virtual screening, and MD simulation

Abstract

T cell immunoreceptor with immunoglobulin and ITIM domain (TIGIT) is one of the most promising targets for cancer immunotherapy. The combination of TIGIT and poliovirus receptor (PVR), which is highly expressed on the tumor surface, inhibits the killing of tumor cells by immune cells. Although antibody blocking the PVR/TIGIT immune checkpoint has shown encouraging anti-tumor effects, small molecules targeting TIGIT to block PVR/TIGIT have not yet been studied. In this study, diverse computational approaches were employed to identify potential inhibitors of this therapeutic targets. First, virtual alanine scanning was used to identify hotspot residues of TIGIT that were effective inhibitory sites. Second, the Extreme Gradient Boosting (XGBoost) classification model and the RO4 rule were used to initially exclude negative compounds. Then, centroid-based virtual screening was used in combination with absorption, distribution, metabolism, excretion, and toxicity (ADMET) prediction to identify the four most promising candidate molecules. Molecular dynamics simulation trajectory analysis showed stable dynamic behavior of the candidate molecules and proteins. Molecular mechanics and Poisson-Boltzmann surface area (MMPBSA) calculations showed that MCULE-5939418698 had the lowest binding free energy (-39.79 kcal/mol). Binding-conformation and energy-decomposition analyses indicated significant involvement of residues L47, Q53, V54 and N58 in inhibitor binding. Principal component analysis and free energy landscape analysis further demonstrated that the binding of MCULE-4861917955 made the system thermodynamically more favorable. Thus, we screened potential inhibitors targeting TIGIT and provide a fresh pipeline for future drug screening research.