On-line Prediction Of Products Concentrations In Glutamate Fermentation Using Metabolic Network Model And Linear Programming

An accurate and effective mathematical model is a must for implementing control andoptimization of fermentation processes. The metabolic reaction network (flux) model (MRmodel) and technique appeared in the early 1990s. After that, MR model has beenpenetrating and expanding into many fermentation application areas. MR is applicable tofermentation progress of the common products which's metabolic ways has been known. Inour study , a metabolic network model was developed to on-line predict process state in theglutamate fermentation.Before using the MR model, the covering scope and on-line measurable reactions of theMR model should be checked and pre-validated so as to calculate the unmeasurablesubstance concentration or production patterns. basic patterns and characteristic of cell growthand product formation in glutamate fermentation showed that glutamate fermentation is atypical non-growth associated fermentation process ,Lactate was main product except mainproduct — glutamate . Glutamate production phase was relatively long and more importantfrom the standpoint of process control and optimization. So a simplified MR model onlyconsidered the metabolic reactions occurred in the production phase was developed. Themodel only contained consumption reactions of glucose and O2 ,production reactions ofglutamate,lactate and CO2 and energy maintain reactions , of which only two reaction rates(CER and OUR) were on-line measurable. The model contained 20 metabolic equations and25 reaction rates. The model was un-determined system and unknown rates was solved by lineprogramming and the minimum NADH production. was the objective functionA metabolic network model combining with the LP optimization was developed foron-line predicting glutamate concentration and formation patterns under different DO levels.Compared with the traditional unstructured models, The results indicated the power andadvantages of the MR model in terms of general prediction abilities, prediction accuracy, andclear biochemical interpreting of data. The proposed MR model potentially supplies analternative way for on-line control and optimization of fermentation processes.To validate on-line prediction abilities of MR above , metabolic flux distribution ofglutamic acid bacteria at different time when setting DO at 10%,30% and 50% was developedby all metabolic reaction rates solved with added off-line glucose calculated consumption rateand off-line glutamate and lactate calculated production rates with more simplified MRmodel and the least squares method. The results showed that metabolic flux of glutamatewas low and metabolic flux of TCA was high at 24h when setting DO at 50% which wasaccorded with on-line prediction results . At the same time , metabolic flux of pyruvate andα-ketoglutarate which was the key node impacting glutamate production was analysed. Theresults showed the flux diversion of pyruvate to the lactate when setting low DO was higherthan high DO and flux diversion of α-ketoglutarate to the succinate when setting high DOwas higher than low DO. At last ,the relation of the energy diversion and product formationwas studied. The result showed the metabolic flux of NADH and ATP increased evidently andglutamate production rate decreased when setting DO at 50%.