Correlating Selectivity Estimates and Mapping Pharmacophores Across PDE2, PDE5, PDE6, PDE10, and PDE11: An In-Silico Study of cGMP-Pathway Off-Target Liability

Abstract

Introduction: Marketed phosphodiesterase-5 (PDE5) inhibitors are clinically limited by off-target inhibition of the related isoforms PDE6 and PDE11, manifesting as visual disturbance respectively, while PDE2 and PDE10 share cross-reactive chemotypes relevant to cGMP-directed CNS and cardiovascular drug discovery. Aim: To construct in-silico ligand-based pharmacophore models for PDE2, PDE5, PDE6, PDE10, and PDE11, and re-evaluate inter-isoform selectivity using paired bioactivity data. Methodology: ChEMBL bioactivity records for each isoform were clustered by Tanimoto similarity (Butina algorithm, cut-off 0.55) and diversity-sampled by the MaxMin method. Representative 3D conformers were generated (ETKDG/MMFF94) and superimposed by flexible Open3DAlign; consensus pharmacophore features were identified using RDKit and reported as percentage feature coverage. Selectivity was assessed from paired-isoform compounds, correlating selectivity ratios by Pearson coefficient. Key Findings: PDE10 pharmacophore generation succeeded only for a rigid, congeneric quinoline-pyrazine series (seven features, 94-100% coverage), despite having the largest dataset (6,994 measurements), indicating alignment success depends on scaffold rigidity rather than data volume. PDE11 gave the most feature-rich model (nine features, 100% coverage). Corrected selectivity analysis showed sildenafil is 24-fold selective for PDE5 over PDE6 and 2,300-fold over PDE11, whereas tadalafil is 335-fold selective over PDE6 but only 8.3-fold over PDE11, correlating with each drug's side-effect profile. PDE6- and PDE11-selectivity correlated weakly (r = 0.50), showing neither can substitute for the other in candidate triage. Physicochemical descriptors (cLogP, TPSA, fraction-sp3) discriminated PDE6-sparing from cross-reactive compounds. Conclusion: These structural and selectivity insights establish a rational, in-silico framework for designing isoform-selective cGMP-pathway inhibitors with reduced off-target liability.