Integrating Genome-Scale Metabolic Modeling and Experimental Analysis of Xylitol and Arabitol Production by Candida tropicalis from Sugarcane Hemicellulosic Hydrolysate

Agro-industrial residue valorization through biotechnological processes promotes sustainable bioprocesses. Xylitol and arabitol are polyols of industrial interest associated with pentose metabolism. This study integrated genome-scale metabolic modeling and experimental assessment to investigate metabolic fluxes related to their formation by Candida tropicalis and evaluate production from sugarcane hemicellulosic hydrolysate. The genomes of three C. tropicalis strains were analyzed: Karbala-1-B1 (GCA_052815155.1), ZD-3 (GCA_022984785.1), and Y3313 (GCA_023623415.1). Coding regions were predicted using AUGUSTUS through the Galaxy platform, and protein sequences were compared by BLASTp against the reference proteome of strain MYA-3404. The homology results refined the gene–protein–reaction (GPR) associations of the genome-scale metabolic model iYS854, implemented in COBRApy. The network included the reduction of L-arabinose to L-arabitol and polyol excretion reactions. Flux Balance Analysis (FBA) estimated specific growth rates of 1.3602 h^-1 on glucose, 1.1416 h^-1 on xylose, and 0.7534 h^-1 on glycerol. Under a xylose uptake constraint of 10 mmol gDW^-1 h^-1, the predicted maximum xylitol production flux was 10.00 mmol gDW^-1 h^-1. Under growth constraints, this flux decreased to 6.85 and 2.74 mmol gDW^-1 h^-1, considering minimum biomass requirements of 75% and 90%, respectively, highlighting the effect of metabolic flux allocation between growth and product formation. Experimental validation involved two stages: a defined medium containing sugars corresponding to those in the hemicellulosic hydrolysate and sugarcane hydrolysate. In the defined medium, xylitol reached 5.41 g L^-1, while arabitol had a mean concentration of 1.62 g L^-1. In the hydrolysate, containing 30 g L^-1 xylose and 5.81 g L^-1 arabinose, xylitol and arabitol concentrations reached 13.50 and 1.77 g L^-1, respectively. Xylose consumption reached 99.59% and arabinose consumption 74.52%, with a xylitol yield of 0.45 g g^-1. The integration of modeling and experimentation made it possible to relate the predicted behavior of the metabolic network to the fermentative performance of C. tropicalis. Computationally, the progressive reduction in the maximum xylitol production flux, from 10.00 to 2.74 mmol gDW^-1 h^-1 under increasingly stringent growth constraints, demonstrated the effect of growth constraints on the maximum predicted xylitol production flux, while the inclusion of the L-arabinose reduction reaction allowed the conversion of this pentose into L-arabitol to be represented in the metabolic model. Experimentally, this relationship was corroborated by the production of xylitol and arabitol in both the defined medium and the hydrolysate, accompanied by the consumption of both pentoses, reinforcing the biological relevance of the metabolic capabilities investigated. Overall, the results highlight genome-scale metabolic modeling as a tool for interpreting metabolic flux distribution and, when combined with experimental assessment, guiding the development of bioprocesses for polyol production from agro-industrial residues.