Evaluation of Genome-Scale Model Reconstruction Strategies for Lentilactobacillus kefiri DH5 and Deciphering Its Metabolic Network

Background/Objectives: Genome-scale metabolic models (GSM) are key tools for predicting microbial physiology, yet species within the genus Lentilactobacillus remain largely unexplored. Lentilactobacillus kefiri DH5 is an obligately heterofermentative lactic acid bacterium with unique redox metabolism, but no curated GSM model exists for this species. This study aimed to generate the first GSM model for L. kefiri DH5, evaluate multiple reconstruction tools, and characterize metabolic features underlying its heterofermentative metabolism. Methods: Draft GSM models were generated from the L. kefiri DH5 genome annotation using five reconstruction tools. For each tool, gap-filling was performed on a CDM, followed by quality assessment using the MEMOTE. Manual curation was performed using the COBRApy library. Results: Among the five reconstructions, the KBase-derived draft demonstrated the highest quality and production potential for metabolites characteristic of heterofermentative fermentation. During manual curation of this model, reaction directions in central carbon metabolism and amino acid pathways were corrected. Analysis further identified an alternative NADH-regenerating glucose shunt via D-gluconate, supported by omics data and enzyme promiscuity considerations. Incorporation of this pathway resolved the redox imbalance and allowed the model to reproduce metabolic exchange profiles characteristic of obligate heterofermenters. Conclusions: We developed the first manually curated genome-scale model of L. kefiri DH5 and showed that the choice of reconstruction tool substantially affects model quality and predictive power. We also proposed an alternative glucose assimilation shunt via gluconolactone, which resolved the redox imbalance in the model and enabled representation of the heterofermentative metabolism.