IN SILICO DESIGN OF DE NOVO GLYCOSIDE HYDROLASES FOR SECOND-GENERATION ETHANOL PRODUCTION

Ethanol is a low-carbon renewable fuel produced via biomass fermentation. While global production relies primarily on first-generation (1G) processes using crops like sugarcane and corn, second-generation (2G) ethanol converts lignocellulosic biomass. By utilizing 1G agricultural residues, 2G production increases ethanol yields per unit area without expanding land use. Ethanol as a fuel is well-established in Brazil, where the light-vehicle fleet is widely adapted to its use as high-blend or pure. However, scaling 2G output remains limited by the high costs of breaking down complex biomass. Because cellulose is the primary constituent, highly efficient cellulases tailored for industrial conditions are required. To tackle this challenge, we designed de novo glycoside hydrolase models targeting cellulose.

We modeled a theozyme mimicking the hydrolysis transition state, comprising a tetrasaccharide substrate, a catalytic water, and two acetic acid molecules representing active-site Glutamate or Aspartate residues, at the DFT level using ORCA 6. Protein backbones capable of accommodating this catalytic geometry were generated using RFdiffusion, followed by sequence design with LigandMPNN. To filter candidates, conformational ensembles of the reactant bound in the active site were generated via PLACER combined with AlphaFold3 structural and ligand prediction. Preliminary results show agreement between designed models and AlphaFold predictions, placing the substrate within the binding pocket. Moving forward, quantum chemical filters, such as Condensed Fukui Functions, will be introduced to assess pocket nucleophilicity and electrophilicity. As future perspectives, molecular dynamics simulations and hybrid QM/MM calculations are expected to be carried out prior to experimental validation.