Developing a Computer-Aided Drug Design Approach to Discovery Lead-Like PMM2 Enzyme Activators for Congenital Disordes of Glycosylation (CDG) Therapy

• Autor(a) principal: Magrinho, Savador da Costa Macedo
• Data de Publicação: 2023
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
• Texto Completo: http://hdl.handle.net/10362/154165
• topic: N-glycosylation; PMM2-CDG; CDG therapies; Computer-Aided Drug Design; Drug repurposing; Pharmacological chaperoning; Domínio/Área Científica::Engenharia e Tecnologia::Outras Engenharias e Tecnologias

The Congenital Disorders of Glycosylation (CDG) comprise a group of rare metabolic diseases caused by inherited defects in cellular glycosylation pathways, ultimately leading to a wide range of multisystem disease phenotypes. The most common sub-type, PMM2-CDG, with >1000 reported cases worldwide, is characterised by genetic defects in the coding gene for phosphomannomutase 2 (PMM2), which catalyses the interconversion of mannose 6-P to mannose 1-P, a substrate for posttranslational N-glycosylation. While no effective treatment is yet available, previous functional characterisation studies in patient-observed mutations have suggested the possibility of designing pharmaco-chaperone (PC) therapies to stabilize PMM2 structure and partially rescue its residual activity. Using an experimentally-validated compound dataset containing over 10.000 drug-like molecules whose interaction with human PMM2 enzyme, namely thermal stability and IC50 (half maximal inhibitory concentration), had been previously assayed, two machine learning strategies were developed to unveil possible PCs for hPMM2 activation: 1) A quantitative structure-activity relationship (QSAR) classification model to predict the interaction of submitted molecules with PMM2 and 2) a QSAR regression model to estimate an IC50 value for such ligand-protein interactions. These QSAR models later served as computational tools to perform a virtual drug screen to search for and select hits with the desired PC profile. Compounds yielding interesting results were submitted to docking studies to further explore possible protein-ligand interactions and the complex’s overall affinity. Promising hits included quinolone derivatives, phenylpropanoic acid derivatives, and vitamins, amongst others. Some of these compounds were acquired and underwent in vitro experiments in patient-derived fibroblast cell lines to further analyse their effect on mutant PMM2 activity from cell extracts using different molecular biology techniques. The applications of different biomarkers for this disease model were identified and studied. This experimental work allowed to elaborate on the therapeutic potential of selected hits and provided notions on defining the models’ predictive capabilities for such complex structure-biological activity relationships.