Unraveling structural requirements of amino-pyrimidine T790M/L858R double mutant EGFR inhibitors: 2D and 3D QSAR study

A systematic pipeline of protein structure selection for computer-aided drug discovery: A case study on T790M/L858R mutant EGFR structures

Virtual screening (VS) is a routine method to evaluate chemical libraries for lead identification. Therefore, the selection of appropriate protein structures for VS is an essential prerequisite to identify true actives during docking. But the presence of several crystal structures of the same protein makes it difficult to select one or few structures rationally for screening. Therefore, a computational prioritization protocol has been developed for shortlisting crystal structures that identify true active molecules with better efficiency. As identification of small-molecule inhibitors is an important clinical requirement for the T790M/L858R (TMLR) EGFR mutant, it has been selected as a case study. The approach involves cross-docking of 21 co-crystal ligands with all the structures of the same protein to select structures that dock non-native ligands with lower RMSD. The cross docking performance was then correlated with ligand similarity and binding-site conformational similarity. Eventually, structures were shortlisted by integrating cross-docking performance, and ligand and binding-site similarity. Thereafter, binding pose metadynamics was employed to identify structures having stable co-crystal ligands in their respective binding pockets. Finally, different enrichment metrics like BEDROC, RIE, AUAC, and EF1% were evaluated leading to the identification of five TMLR structures (5HCX, 5CAN, 5CAP, 5CAS, and 5CAO). These structures docked a number of non-native ligands with low RMSD, contain structurally dissimilar ligands, have conformationally dissimilar binding sites, harbor stable co-crystal ligands, and also identify true actives early. The present approach can be implemented for shortlisting protein targets of any other important therapeutic kinases.

Computational identification of natural product inhibitors against EGFR double mutant (T790M/L858R) by integrating ADMET, machine learning, molecular docking and a dynamics approach

Double mutated epidermal growth factor receptor is a clinically important target for addressing drug resistance in lung cancer treatment. Therefore, discovering new inhibitors against the T790M/L858R (TMLR) resistant mutation is ongoing globally. In the present study, nearly 150 000 molecules from various natural product libraries were screened by employing different ligand and structure-based techniques. Initially, the library was filtered to identify drug-like molecules, which were subjected to a machine learning based classification model to identify molecules with a higher probability of having anti-cancer activity. Simultaneously, rules for constrained docking were derived from three-dimensional protein-ligand complexes and thereafter, constrained docking was undertaken, followed by HYDE binding affinity assessment. As a result, three molecules that resemble interactions similar to the co-crystallized complex were selected and subjected to 100 ns molecular dynamics simulation for stability analysis. The interaction analysis for the 100 ns simulation period showed that the leads exhibit the conserved hydrogen bond interaction with Gln791 and Met793 as in the co-crystal ligand. Also, the study indicated that Y-shaped molecules are preferred in the binding pocket as it enables them to occupy both pockets. The MMGBSA binding energy calculations revealed that the molecules have comparable binding energy to the native ligand. The present study has enabled the identification of a few ADMET adherent leads from natural products that exhibit the potential to inhibit the double mutated drug-resistant EGFR.

TMLRpred: A machine learning classification model to distinguish reversible EGFR double mutant inhibitors

The EGFR is a clinically important therapeutic drug target in lung cancer. The first-generation tyrosine kinase inhibitors used in clinics are effective against L858R-mutated EGFR. However, relapse of the disease due to the presence of resistant mutation (T790M) makes these inhibitors ineffective. This has necessitated the need to identify new potent EGFR inhibitors against the resistant double mutants. Therefore, various machine learning techniques ((instance-based learner (IBK), naïve Bayesian (NB), sequential minimal optimization (SMO), and random forest (RF)) were employed to develop twelve classification models on three different datasets (high, moderate, and weakly active inhibitors). The models were validated using fivefold cross-validation and independent validation datasets. It was observed that the random forest-based models showed best performance. Also, functional groups, PubChem fingerprints, and substructure of highly active inhibitors were compared to inactive to identify structural features which are important for activity. To promote open-source drug discovery, a tool has been developed, which incorporates the best performing models and allows users to predict the potential of chemical molecules as anti-TMLR inhibitor. It is expected that the machine learning classification models developed in this study will pave way for identifying novel inhibitors against the resistant EGFR double mutants.