Targeting Proteins Involved in Neurodegenerative Diseases and Cancer: From Traditional Structure-Based Approach to Artificial Intelligence Models

Computer-aided drug design (CADD) and machine learning (ML) techniques are transforming drug discovery, enabling the efficient identification of small molecules with therapeutic potential. Neurological conditions, which affect 43% of the global population and represent the leading cause of ill health and disability worldwide, alongside cancer, with over 2 million new cases and 611,720 deaths expected in the USA in 2024, underscore the urgent need for effective treatments. [1, 2] This dissertation addresses such a challenge by applying CADD and ML methodologies to the discovery of ligands targeting proteins implicated in neurological disorders and tumors: Sigma receptors (SRs) and Tyrosinase (TYR). The Introduction section provides detailed insights into the methodologies employed throughout the study. In Case study 1: targeting Tyrosinase enzyme, a therapeutic target associated with melanoma and Parkinson's disease, efforts were focused on the identication of new inhibitors targeting two distinct forms of tyrosinase: Agaricus Bisporus (AbTYR) and human Tyrosinase (hTYR). CADD techniques, spaning from docking studies, MM-GBSA calculations and molecular dynamics simulations, were employed in retro-analyses to rationalize experimental data. These approaches offered valuable insights into ligand-receptor interactions and revealed key structural determinants of binding, advancing our understanding of tyrosinase modulation. In Case study 2: targeting Sigma Receptors, Sigma Receptors (SRs) are investigated as versatile targets in drug discovery, since the several implications in tumor and neurological conditions. The study encompassess a range of methodologies, including molecular docking and dynamics for retrospective analyses, as well as an in-depth structure-based approach to create a pharmacophore model for Sigma 1 receptor (S1R) compounds. This work, conducted during my 10-month industry internship at Net4Science, led to the identification of promising therapeutic candidates. Furthermore, during a six-month exchange program at the University of Vienna, a machine learning (ML) approach was utilized to predict active and selective compounds targeting SRs, enhancing the potential for precision in drug discovery. ML algorithms, trained on structural and physicochemical data, achieved high accuracy in identifying SRs compounds. The findings presented in this research highlight the effective contribution of CADD and ML in modern drug discovery, offering novel methodologies and insights into the design and optimization of compounds targeting SRs and TYRs.