We will employ a comprehensive computational protocol to enable the discovery and optimization of novel lead compounds for melanin-concentrating hormone receptor 1 (MCHR1). Since there is no existing crystal structure, we will execute a workflow for target analysis, specifically focusing on target validation and identification. This step involves analyzing structures generated through homology modeling, utilizing tools like AlphaFold or Schrödinger software. The aim is to select an appropriate structure that will be used in the subsequent virtual screening campaign. Protein structures will be prepared with Maestro´s Protein Preparation Wizard and protonation states of the titratable residues will be assigned using the PROPKA tool. Long-run (~ 5 µs) molecular dynamics (MD) simulations will be performed followed by clustering of trajectories to obtain representative structures of the receptor. Following, we will perform control calculations on our protein models to confirm that the prepared protein structure, and our preferred docking parameters can properly differentiate known actives from inactive molecules. We will compare ~1600 known active binders of MCHR1 reported in the literature against property-matched decoys. These decoys have similar physical properties as the actives but different topologies that would be expected to reduce binding significantly. We will prepare a library of 1600 actives and 16000 decoys (ratio of about 1:10 actives:decoys) for control calculations. The performance of binding pockets of the prepared protein targets can be assessed by analyzing receiver-operator characteristic (ROC) curves. The ROC curve quantifies the true positive rate as a function of the false positive rate. These control calculations are critical for this first step because, in our docking campaign with hundreds of millions to billions of molecules, only molecules ranked around the top 0.1% are thoroughly evaluated. By leveraging our library (a carefully curated library of the drug-like molecules and macromolecules obtained from the Enamine database), we will generate a targeted sub-library for designing MCHR1 antagonists. This sub-library will include only PAINS-free molecules (to reduce the risk of confronting false positives) with good ADMET properties (e.g., molecules with favourable cell permeability, acceptable aqueous solubility, and elimination of compounds with potential toxicophores). The initial stages of our computational pipeline will involve ligand-based virtual screening using pharmacophoric models along with the utilization of machine learning (ML) models. Different supervised learning models (random forest (RF), neural networks (NN), support vector machine (SVM), and logistic regression (LR)) will be used, optimized, and evaluated for their ability to predict active and inactive compounds with the best performing models used to screen our database. The ML models will be developed using the experimental data extracted from the literature, specifically focusing on ligand binding affinities associated with MCHR1. The ML models will incorporate these ligands to predict both the activity of compounds and their binding affinities. Common pharmacophore hypotheses will be created using a diverse set of known MCHR1 ligands. This ligand-based virtual screening approach will be applied to assess our curated library. Following this, a structure-based virtual screening will be employed, molecular docking will be done using the standard precision (SP) Glide docking score. The top 1000 docked compounds will be re-docked using extra-precision (XP) Glide docking score including strain energy corrections. The top hits from the virtual screening will be clustered into groups based on their scaffold similarity. Next, all-atom MD simulations will be conducted to study the relative stability of the receptor–ligand interactions of the top hits. Finally, to prioritize compounds for in vitro testing, we will utilize more rigorous and expensive alchemical free energy perturbation (FEP) calculations (following the protocol reported by Jiang et al, DOI: 10.1021/acs.jcim.9b00362) to accurately predict the potency of the selected top binders. This comprehensive in silico protocol is expected to provide potent and novel candidates of MCHR1 that can be further experimentally validated and optimized through biochemical assays.
