Search for chemical probes against cancer therapeutic target proteins: association of artificial intelligence with biophysical and biochemical results from a library of chemical fragments

• Grant number: 24/13316-0
• Support Opportunities: Scholarships in Brazil - Support Program for Fixating Young Doctors
• Start date: September 01, 2024
• End date: January 31, 2025
• Field of knowledge: Health Sciences - Pharmacy
• Agreement: CNPq
• Principal Investigator: Katlin Brauer Massirer
• Grantee: Rafael Junqueira Borges
• Host Institution: Centro de Biologia Molecular e Engenharia Genética (CBMEG). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Associated research grant: 24/02215-9 - Search for chemical probes against cancer therapeutic target proteins: association of artificial intelligence with biophysical and biochemical results from a library of chemical fragments, AP.R

Abstract

The Projeto Temático "INCT 2014: Open Access Center for Medicinal Chemistry", to which this Auxílio de Fixação de Jovens Doutores (FJD) is associated, aims to bring together scientists from different areas of biology and chemistry to connect results from genomics to medicine, agriculture and industry and offer methodologies and reagents for studying specific proteins in open access to the community and for finding chemical probes and inhibitors. This objective is combined with the attempt to reduce the discrepant concentration of studies on a few proteins and expand the therapeutic targets of medicines. This process can be accelerated with the combination of experimental and computational approaches, such as the development of new scanning technologies with biophysical, biochemical, and computational characterization methodologies using a library of commercially available compounds. In 2022, the Center for Medicinal Chemistry (CQMED) at UNICAMP, host laboratory of this project FJD, established a drug discovery platform based on chemical fragments. The advantage of this approach is the restricted size and low chemical complexity of the fragments, which leads to binding in difficult-to-reach regions, such as allosteric sites and hot spots. The reduced molecular size leads to a low affinity binding, which must subsequently be increased with the growth of the molecule to acquire potency and selectivity. The implemented techniques at CQMED are weak-affinity chromatography coupled with mass spectrometry, unprecedented in Brazil, and thermal shift assay capable of quickly screening a library of thousands of fragments and generating quantitative results of protein fragments, complemented with enzymatic assays informative regarding function. In order to reveal the binding sites, the best hits can be subjected to cross-linking followed by bottom-up mass spectrometry (MS) or explore exchange of deuterium and hydrogen. Crystallography is an alternative as it offers high resolution data of the complex, but it requires crystallization with an estimated success rate of 30%. The objective of this Auxílio FJD is to accelerate the search for chemical probes against proteins through the development of an algorithm that integrates the data from these techniques in fragment screening with computational methods, including artificial intelligence (AI), that predict binding sites and fragment growth possibilities. We will use currently available methodologies, such as mathematical models of quantitative structure-activity relationships, hot spot search, docking, molecular dynamics, free energy calculations, fragment growth, de novo design, synthetic tractability (synthesizability) of new compounds, AI and SEQUENCE SLIDER. The latter was a program that the grant candidate developed, which integrates information from structural data (crystallography and MS), phylogenetic relationships, interaction energy and AI to model sequence hypotheses and score them with probabilities, being quickly useful for the objective of ligand ranking and growth and selectivity analysis. We will use biophysical data on fragments and inhibitors against HSP90 available in the PDB and provided by collaborators for training and calibration of the algorithm. We will apply this predictor against cancer therapeutic target proteins, ASCC3 and BICC1, on experimental data obtained at this INCT, and we will validate the binding sites of the fragments and new predicted ligands with site-directed mutation and biophysical and biochemical assays. The predictor will overcome the barrier of the impossibility of obtaining structural data on complexes, which was the case for ASCC3 and BICC1, or even using this information to find more potent and selective inhibitors. With this project, we intend to offer a experimental and computational methodology to the Medicinal Chemistry community that accelerate drug discovery, particularly against ASCC3 and BICC1.