Machine learning based approach to develop potential dual inhibitors of Neprilysin and Dipeptidyl peptidase-4: Implication in the treatment of type 2 diabetes

Neutral endopeptidase (NEP or neprilysin) is a key enzyme associated with the metabolic inactivation of numerous bioactive natriuretic peptides (NP). Among these, bradykinin, endothelin, angiotensin II, amyloid β protein, substance P and glucagon-like peptide 1 (GLP-1) are the key NPs that affect the heart, kidney, and other organs. Among these, GLP-1, an essential simulator of insulin secretion, is frequently found to be impaired or downregulated in the case of diabetes, particularly type 2 diabetes. The level of GLP-1 is also diminished by another serine protease, Dipeptidyl peptidase-4 (DPP-4). Utilizing this concept of NEP and DPP-4 in orchestrating the degradation of the GLP-1 level and consequently affecting the type 2 diabetes outcome, we developed dual NEP/DPP-4 inhibitors that could offer an alternative regimen for treating type 2 diabetes.

In the present study, we have developed a machine learning (ML) based prediction model considering the pharmacophoric features of both NEP and DPP-4. The model was tested against an array of in-house developed and newly designed and synthesized NCEs as training sets with the reported inhibitors against the enzyme. The work was further validated and standardized with molecular docking and dynamics studies and corroborated with numerous biological studies. The cumulative analysis yielded 1a and 1f as the best lead molecules with potent dual inhibition of NEP and DPP-4 with anti-diabetic potential.