Application of SNP2SIM to Venetoclax Resistant Variants in BCL2
Mentor: Dr. Matthew McCoy, Innovation Center for Biomedical Informatics (ICBI), Department of Oncology, Georgetown University Medical Center.
Date/Time: August 24, 2021 at 2:00pm
Abstract: BCL-2, a member of a class of pro-survival proteins, binds and restrains members of a family of pro-apoptotic proteins. Its pro-survival functionality is exerted via the interaction of BCL-2 Homology 3 (BH3) motifs of pro-apoptotic proteins with a complementary groove on its surface. Understanding of this pathway prompted the emergence of cancer therapies such as Venetoclax which target the BH3-binding groove of BCL-2. Treatment of patients with such compounds has revealed the emergence of several drug-selected variants of BCL-2. One such variant, G101V, was determined to reduce the affinity of BCL-2 to Venetoclax whilst simultaneously maintaining affinity for pro-apoptotic proteins and therefore conferring drug resistance.
Previous coarse-grain binding simulations of BCL-2 variants have detected that the most well-known variant of BCL-2, G101V, is not strongly resistant to Venetoclax compared to other known variants. This suggests that there may be an alternative drug resistance mechanism. The objective of this project was to utilize the existing SNP2SIM workflow to simulate the all-atom molecular dynamics of BCL-2 wildtype, BCL-2 variant G101V, as well as additional BCL-2 variants, to generate protein structure scaffolds and to integrate these results with the current Venetoclax resistance curation for BCL-2. Specifically, this project focused on testing various clustering methods and dimensionality reduction methods to generate protein scaffolds representative of the conformational dynamics of the BH3-binding groove for each variant. All atom molecular dynamics of protein structure were simulated using NAMD (Nanoscale Molecular Dynamics) and VMD (Visual Molecular Dynamics). These new structures will be used as the scaffolding in new coarse-grained binding simulations using AutoDock Vina to assess if the detection of drug resistance can be improved.