Pharmacophore Modelling of Phospho-di-esterase modulators - Ligand based In-Silico designing of Potential NCE’s

Abstract

Phosphodiesterases (PDEs) 1, 3, 4, 7, 8, and 9 regulate cyclic nucleotide (cAMP/cGMP) signalling implicated in inflammatory and mucus-hypersecretory airway disease, yet isoform-selective inhibitor design is hindered by high catalytic-domain homology and limited pharmacophore data. Objective: To generate ligand-based pharmacophore models for all six isoforms using a fully in-silico workflow, and to identify shared and isoform-specific features governing potency and selectivity. Methods: Bioactivity data (up to 6,595 measurements per isoform) were retrieved from ChEMBL and supplemented with catalog reference inhibitors. Compounds were clustered by ECFP4/Tanimoto similarity (Butina algorithm, threshold 0.55), and diversity-representative training sets were selected via MaxMin sampling. 3D conformers (ETKDG/MMFF94) were superimposed using flexible Open3DAlign, and consensus features were extracted with RDKit and expressed as percentage coverage across the aligned set. Results: Six- to nine-point pharmacophores (83-100% coverage) were resolved for each isoform, comprising acceptor, donor, aromatic, and hydrophobic centres, plus a positively ionizable feature unique to PDE7. Alignment success correlated with training-set rigidity rather than potency: a potent but flexible PDE8 chemotype (7.4 rotatable bonds) failed to converge, whereas a rigid, less potent series (3.8 rotatable bonds) succeeded; the PDE4 catechol-ether motif likewise showed no consensus despite sub-nanomolar activity. Best potencies ranged from 32 pM (PDE4) to low nanomolar (PDE7, PDE8). PDE3A/PDE3B, PDE4B/PDE4D, and PDE9/PDE2 overlaps were identified as unresolved selectivity liabilities. Conclusion: The resulting models provide validated 3D screening queries for virtual screening of natural and synthetic inhibitors, supporting ongoing designing isoform-selective cGMP-pathway inhibitors with reduced off-target liability.