The team utilized a multistage in silico workflow to screen a library of nearly 30,000 compounds, ultimately prioritizing Molecule_001344 for its ability to bind the receptor without relying on the canonical ionic anchor typically required by classical antagonists. This finding provides a structural basis for developing subtype-selective inhibitors that may avoid off-target cholinergic effects associated with traditional muscarinic receptor drugs.
The study addresses a significant therapeutic challenge in prostate cancer management. While androgen deprivation therapy is initially effective, resistance frequently leads to castration-resistant prostate cancer, with a subset of tumors undergoing lineage plasticity to become NEPC. CHRM4 signaling has been linked to this neuroendocrine differentiation, particularly through its interaction with nerve growth factor and the activation of AKT/MYCN pathways. However, the high sequence identity of the orthosteric binding site across the muscarinic receptor family (M1–M5) complicates the design of selective inhibitors. Classical antagonists like tiotropium rely on a conserved Asp112 residue for high affinity, making it difficult to target CHRM4 specifically without affecting other subtypes.
To overcome this structural challenge, the researchers applied a rigorous virtual screening pipeline to a GPCR-A-focused ChemDiv library containing 29,488 compounds. After de-duplication and filtering for pan-assay interference compounds, 26,498 unique structures remained for analysis. The workflow integrated structure-based virtual screening, induced-fit docking, and MM/GBSA binding energy calculations. This hierarchical approach allowed the team to refine poses while accounting for local receptor flexibility, a crucial step in identifying stable binding modes that might be missed by rigid docking protocols alone.
Unlike the reference ligand tiotropium, which forms a salt bridge with Asp112, Molecule_001344 engaged in hydrogen bonding with Asn417 and Ser116, reinforced by π-cation interactions with Tyr113. The stability of this pose was supported by favorable MM/GBSA binding energy estimates, with Molecule_001344 yielding a ΔG_bind of -75.44 kcal/mol, the most favorable among the top-ranked hits.
To assess the dynamic stability of the predicted binding mode, the team performed three independent 100 ns molecular dynamics simulations for each system, including apo CHRM4, the tiotropium-bound complex, and the Molecule_001344-bound complex. The simulations were conducted in a membrane-embedded environment to mimic physiological conditions. Analysis of the trajectories showed that Molecule_001344 exhibited the lowest inter-replicate dispersion in protein RMSD, with an overall value of 3.27 ± 0.09 Å. This consistency indicated that the complex maintained a stable conformational state across all three independent simulations, unlike Molecule_006995, which showed greater trajectory dependence and variability.
Principal component analysis and free-energy landscape analysis further differentiated the dynamic behavior of the selected compounds. Molecule_001344 demonstrated reproducible localization within a single dominant low-energy basin, with all three replicate trajectories remaining committed to this state throughout the simulation period. In contrast, the apo receptor sampled the broadest region of the conformational space, while Molecule_006995 populated multiple distinct low-energy regions. The persistence of the Asn417-Ser116 hydrogen-bond network in Molecule_001344 was confirmed by high occupancy rates, with the Asn417 bond maintained at 99.00% and the Ser116 bond at 93.67% across the simulations. This reproducibility supports the polar-clamp motif as a characteristic feature of the compound’s binding mode.
Pharmacokinetic and toxicological predictions were also evaluated to assess the drug-likeness of the prioritized compounds. Molecule_001344 satisfied Lipinski’s rule of five without violations and showed no PAINS alerts. It exhibited a favorable predicted bioavailability score of 0.55 and high gastrointestinal absorption. However, the compound was predicted to cross the blood-brain barrier, which may be a liability for a non-CNS therapeutic indication. Additionally, Molecule_001344 was identified as a potential P-glycoprotein substrate, which could influence its intestinal absorption and tissue distribution. These predictions highlight specific liabilities that would need to be addressed during further optimization.
Quantum-chemical calculations provided additional insights into the electronic structure of the leading hits. Molecule_001344 displayed a HOMO-LUMO gap of 0.22 eV, which was slightly larger than that of Molecule_006995 (0.19 eV) but comparable to tiotropium (0.20 eV). The compound showed the highest hardness and lowest softness among the hits, indicating a lower tendency for charge redistribution. These electronic properties complement the structure-based findings, suggesting that Molecule_001344 possesses a stable electronic regime suitable for receptor binding. The frontier orbital maps revealed that the HOMO density was distributed over the aromatic core, while the LUMO extended into the heteroatom-rich regions, consistent with the predicted binding interactions.
Cross-subtype selectivity profiling was performed to evaluate the potential for off-target effects. Molecule_001344 was cross-docked into the crystal structures of CHRM1, CHRM2, CHRM3, and CHRM5. MM/GBSA rescoring revealed a relative energetic preference for CHRM4, with binding free energy values ranging from -50.46 to -56.93 kcal/mol for the other subtypes, compared to -75.44 kcal/mol for CHRM4. This energetic margin of 18–25 kcal/mol suggests a potential for subtype selectivity, although the authors note that these computational estimates cannot be directly translated into experimental selectivity ratios. The structural comparison indicated that Molecule_001344 did not engage the equivalent serine residues in the other subtypes, which may contribute to its preferential binding to CHRM4.
The study acknowledges several limitations, primarily that the findings are computational hypotheses that require experimental validation. The virtual screening was conducted on a focused library, sampling only a limited region of chemical space. The MM/GBSA values were interpreted comparatively rather than as absolute binding free energies, and the 100 ns simulation timescale may not capture slower GPCR conformational transitions. Furthermore, equivalent replicate MD simulations were not performed for the other muscarinic subtypes, leaving the contribution of the dynamic binding mode to subtype preference unresolved. Experimental binding and functional assays will be necessary to confirm the predicted activity and selectivity of Molecule_001344.
Despite these limitations, the study provides a valuable structural framework for the development of CHRM4-targeted therapies. By identifying a non-canonical binding mode that bypasses the conserved Asp112 anchor, the researchers have opened a new avenue for designing selective inhibitors. Molecule_001344 emerges as a promising lead compound with a distinct orthosteric recognition pattern, supported by favorable binding energetics, dynamic stability, and drug-like properties. Further experimental evaluation of its binding affinity, functional activity, and effects on CHRM4-associated signaling in neuroendocrine prostate cancer models will be essential to determine its therapeutic potential.
This hierarchical filtering strategy was designed to isolate chemotypes with stable binding modes and favorable interaction energetics. The docking scores were used for relative ranking rather than as quantitative estimates of binding free energy, with more negative values indicating more favorable predicted binding within the receptor pocket. This approach allowed the researchers to prioritize compounds that exhibited specific interaction patterns with key binding-site residues, such as Tyr113 and Asn417, which were recurrent features in the top-ranked poses.
To complement the structure-based screening, the team developed an AutoQSAR model using experimental pIC50 data from the ChEMBL database. The selected kernel partial least-squares model achieved a ranking score of 0.8396, with an R2 value of 0.8352 and a root mean square error of 0.3567 pIC50 units. Molecule_001344 received the highest predicted pIC50 value of 6.739 among the hit compounds, compared to 8.212 for the reference ligand tiotropium. However, the low Tanimoto similarity coefficients between the screened compounds and the training set indicate that these predictions involve a degree of extrapolation beyond the chemical space represented in the training data. This ligand-based assessment provided an independent estimate of bioactivity that supported the structural prioritization of the leading candidates.
Subject of Research: Exploring CHRM4 orthosteric recognition: computational identification of a non-canonical Asn417-Ser116 polar-clamp binding mode.
Article Title: Exploring CHRM4 orthosteric recognition: computational identification of a non-canonical Asn417-Ser116 polar-clamp binding mode.
Article References: Kayode, I. O., Huang, K.-H., & Liu, Y.-N. (2026). Exploring CHRM4 orthosteric recognition: computational identification of a non-canonical Asn417-Ser116 polar-clamp binding mode.. Results in Chemistry, 32, Article 104039. https://doi.org/10.1016/j.rechem.2026.104039
Image Credits: AI Generated
DOI: 10.1016/j.rechem.2026.104039
Keywords: Exploring, CHRM4, orthosteric, recognition, computational, identification, non-canonical, Asn417-Ser116, polar-clamp, binding, mode, scientific research
News Source: Nathaniel Bowman. (November 10, 2026). Computational Study Identifies Novel CHRM4 Binding Mode for Prostate Cancer. Scienmag.
