Objective: The objective of this study was to perform docking-based analysis of bile acid binding on the protein complex of channels and to derive neural network that predicts the influence of bile acids and their synthetic analogues on the activity of BK(Ca) channels in smooth muscle cells based on descriptors for bile acids and their synthetic analogues and on their already published activities using patch-clamp techniques.
Materials and methods: Ligands for molecular docking were optimized using computer routine for minimization of energy by using the force field MMFF94 via Chem3D 15.0 and ligands and protein channel complex were prepared in AutoDockTools 1.5.6. AutoDock Vina 4.0 software was used for blind docking; processing and verification of the obtained results was performed via Discovery Studio 4.0. Neural network was derived using descriptors for bile acids and their synthetic analogues and their already published activities on calcium-activated K+ channels in smooth muscle cells (ChemDraw Professional 15.0, Dragon 6 software).
Results: Molecular docking was performed for: lithocholic acid, deoxycholic acid, 5β-cholanoic acid, 3β-hydroxi-5β-cholanoic acid, henodeoxycholic acid, ursocholic acid and α-muricholic acid. Neural network model Multiple layer perceptron is derived, having 0.9259 training performances and 0.3673 test performances, training error 0.0073 and test error 0,1607. Model was tested for henodeoxycholic, ursocholic and α-muricholic acid, and internal validation of the model is performed.
Conclusions: Molecular docking suggested that the pharmacophore for maximizing the activity of BK(Ca) channels in the steroid skeleton of bile acids is the C3 quasi-axial α-OH group and the C24 carboxyl function. Derived neural network model successfully predicted activities of tested bile acids on Ca2+ activated K+ channels in smooth muscle cells.