Uso de assinaturas estruturais para proposta de mutações em enzimas β-glicosidase usadas na produção de biocombustíveis

• Ano de defesa: 2019
• Autor(a) principal: Diego César Batista Mariano
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; Brasil; ICB - INSTITUTO DE CIÊNCIAS BIOLOGICAS; ICX - DEPARTAMENTO DE CIÊNCIA DA COMPUTAÇÃO; Programa de Pós-Graduação em Bioinformatica; UFMG
• Palavras-chave em Português: biocombustíveis; β-glicosidases; assinaturas estruturais; mutações; SSV; Bioinformática; Biocombustíveis; Glicosídeo Hidrolases; Mutação
• Link de acesso: http://hdl.handle.net/1843/32078

β-glucosidases (EC 3.2.1.21) are key enzymes in the second-generation biofuel production. They act synergically with endoglucanases and exoglucanases in the conversion of biomass to fermentable sugars. However, most known β-glucosidases are highly inhibited by high glucose concentrations. Hence, the search for mutations that improve the activity of non-tolerant β-glucosidases has great importance to the industry. In this thesis, I present a systematic review of the literature to collect information about glucose-tolerant β-glucosidases and to construct a database, called BETAGDB. In addition, important residues for the activity and glucose-tolerance were characterized in the catalytic pocket. Finally, I proposed a method based on the difference of variation of structural signatures to propose mutations in enzymes, called Structural Signature Variation (SSV). SSV uses graph modeling to create a structural signature that identifies glucose-tolerant β-glucosidases. The SSV method was evaluated in three case studies: (i) 27 mutations described in the literature were manually classified as beneficial or not. The classification was then reproduced using SSV. The method obtained an accuracy of 0.74 and a precision of 0.89; (ii) 18 beneficial mutations were proposed for the non-tolerant β-glucosidase Bgl1B. Experimental results of three mutations corroborate the outcomes obtained by the second case study and demonstrate that SSV is an effective method for the proposal of mutations in β-glucosidases; and (iii) SSV was compared with SVM to verify whether the Euclidean distance, metric used by SSV for comparison of signatures, was effective. It was also compared with BioGPS, a method that uses fingerprints to propose mutations based on three-dimensional structures. SSV obtained values of precision and specificity superior to SVM. In comparison to BioGPS, SSV was able to correctly predict five in seven bench-validated mutations inserted amidase activity into a lipase. The results obtained in this thesis may aid in the production of mutant β-glucosidase enzymes capable of enhancing the production of second-generation biofuels. The SSV method can be extended to other enzymes and can also be used together to other strategies and tools to propose more efficient mutations. SSV is available at <http://bioinfo.dcc.ufmg.br/ssv>.