Development of a Hybrid Gene Regulatory Network–Bioprocess Model to Enhance the Prediction of Bioethanol Production by Saccharomyces cerevisiae

Abstract

Albeit bioprocess kinetics are typically predicted using empirical and unstructured models, a gap still exists in connecting bioprocess performance to the molecular events that control the efficiency of bioethanol production using Saccharomyces cerevisiae. Herein, a hybrid genetic regulatory network (GRN)–bioprocess model is proposed, predicting transcription from important genes (HXK2, PDC5, and ADH1) involved in the glucose-sensing, glycolysis, and bioethanol production processes, associating glucose consumption as well as biomass and bioethanol production rates to the regulatory events that control bioprocess kinetics. Parameter estimation and validation of the hybrid GRN–bioprocess model were conducted in batch trials supplemented with varying glucose contents via quantification of gene transcription levels. Calculation of the normalized root mean square error (NRMSE) confirmed that the hybrid model developed could accurately predict bioethanol production as opposed to the Monod model. NRMSE values ranged between 0.57–0.91 and 0.60–10.09 for the hybrid and Monod models, respectively, indicating enhanced performance of the novel approach proposed, which improved biomass concentration prediction by 89.4%, glucose concentration prediction by 16.2%, and bioethanol concentration prediction by 60.7% in the experiment conducted using an initial glucose content of 40 g L–1. The hybrid model introduced nonconstant transcription-dependent biomass and product yields as novel functions regulated by the plethora of interactions between the regulatory molecules of the pathways involved, offering substantially improved bioprocess prediction. The proposed framework advances our comprehension of the dynamic properties of bioethanol fermentation via consideration of complex cellular mechanisms controlling the synthesis of rate-limiting enzymes, which are typically ignored by the empirical/unstructured models often applied, providing a systematic understanding of bioethanol manufacture.