CReM-guided design of ketamine–dextromethorphan conjugates targeting the GluN2B NMDA receptor amino-terminal domain: an integrated computational study

Abstract

Anxiety disorders remain incompletely controlled in many patients, motivating exploration of alternative glutamatergic strategies. A computational workflow was used to generate and prioritize ketamine–dextromethorphan conjugates against a template-derived human GluN2B NMDA receptor amino-terminal-domain model corresponding to residues 32–393 of the 5EWM construct. Twenty-nine conjugates were generated using the rule-based CReM framework and evaluated by ADME prediction, docking, molecular dynamics, MM/GBSA, pharmacophore mapping, density functional theory, and toxicity prediction. Conjugate 25 showed the strongest docking score (− 9.2 kcal/mol) compared with ketamine (− 6.3 kcal/mol) and dextromethorphan (− 5.9 kcal/mol). Although predicted as BBB permeant, conjugate 25 also showed high molecular weight, high lipophilicity, poor aqueous solubility, P-glycoprotein substrate status, and one Lipinski violation. During the 500 ns simulation, the receptor sampled distinct conformational regimes, whereas protein-fitted ligand RMSD remained predominantly ~ 0.12–0.20 nm. PC1 and PC2 captured 64.57% of conformational variance, and the free-energy landscape identified three accessible minima. MM/GBSA indicated favorable effective interaction energetics dominated by van der Waals contributions. Pharmacophore and single-point DFT analyses were interpreted descriptively rather than as affinity measures. Toxicity prediction identified neurotoxicity, respiratory toxicity, immunotoxicity, clinical-toxicity, CYP2C9, and CYP2D6 liability signals. Overall, conjugate 25 is a computationally prioritized GluN2B-domain lead requiring experimental pharmacokinetic, toxicological, and receptor-level validation.