Modeling And Substrate Feed Optimization For The Vitamin C Precursor-2-keto-L-gulonic Acid Fermentation Process

Vitamin C is an essential and exogenous nutrient for humans and a few otherprimates, which is used widely in pharmaceutical preparations, beverage and foodindustry, animal feeds and cosmetic products. This versatile applicability leads to arapidly increasing demand for vitamin C. The current global market of vitamin C isgreat and increasing with an annual growth rate estimated at3-4%. The two-stepprocess is the main industrial production method for vitamin C synthesis, whichinvolves two fermentation steps, in the first step of which D-sorbitol is converted toL-sorbose, and in the second one L-sorbose is converted to2-KGA. In this dissertation,improving the yield and productivity of the second step of the two-step process isstudied by optimizing the fermentation conditions and modeling the fermentationprocess. The2-KGA fermentation process is mathematically modeled and the modelsare validated by experimental data. Furthermore, based on the model, the controlstrategy for the2-KGA fermentation process is suggested for improving the2-KGAproductivity. The following contents are included in this dissertation.According to the process guide, the cultivation conditions in2-KGAfermentation are further improved, such as the initial concentration of MgSO4·7H2O,the initial concentration of L-sorbose, the temperature and pH. Single factorexperiments are carried out to examine the effect of the different initial concentrationsof MgSO4·7H2O and L-sorbose to the yield and productivity of2-KGA fermentationprocess. According to the single factor experiments, the effect of temperature and pHon2-KGA biosynthesis shows that multi-stage pH control strategy can shorten2-KGAfermentation period, and maintaining the temperature at a high level in the late phaseof2-KGA fermentation is helpful to enhance the ultimate concentration of2-KGA,which applies solely to the termination time of2-KGA fermentation.An unstructured model is developed for2-KGA fermentation process. It isdifficult to model the2-KGA mixed culture process. Following the principle of gradualimprovement, we firstly focus on constructing a unstructured model for2-KGAfermentation process. Gluconobacter oxydans (G. oxydans) and Bacillus megaterium (B. megaterium) are included in this mixed cultivation, and the interaction betweenthem is complicated. Although much effort has been devoted to investigating theirinteraction, it remains unclear. The growth of G. oxydans or B. megaterium iscustomarily modeled in Monod-type equation in the publications. However, theMonod-type equation or Monod model is deficient in describing growth characteristicsof the mixed culture of G. oxydans and B. megaterium, such as all cells of B.megaterium being extinct or forming spores ultimately, the growth of G. oxydans beingenhanced when B. megaterium began to autolyze, and G. oxydans growing slowly afterB. megaterium extinguished. Therefore, G. oxydans and B. megaterium can beclassified as two species which coexist in a ecosystem, and their mixed culture growthcan be modeled according to the population theory. Then the unstructured model of2-KGA fermentation process is proposed in consideration of the models of substrateuptake and product formation. Experimental results demonstrate that the unstructuredmodel is able to describe2-KGA fermentation process with reasonable accuracy.According to the values of identified model parameters, the type of interaction betweenG. oxydans and B. megaterium is concluded to be predation, where G. oxydans is apredator, and B. megaterium is a prey which will extinguish finally.Based on the unstructured model of2-KGA fermentation process, a structuredmodel is further developed for2-KGA fermentation process. In this fermentationprocess, only G. oxydans is responsible for2-KGA producing, and B. megaterium actsas the concomitant strain to stimulate the growth or2-KGA producing of G. oxydans.Therefore, modeling the metabolic pathway of L-sorbose channeled into the cytoplasmof G. oxydans is necessary for developing a structured model of2-KGA fermentationprocess. Many efforts have been devoted to studying the biosynthesis pathway of2-KGA from L-sorbose or other sugars by G. oxydans or other strains. The completegenome sequence of G. oxydans is also accomplished. But the metabolic pathway ofL-sorbose in G. oxydans’ cell remains unknown. Based on the metabolic pathway ofsugars in Gluconobacter species (sp.), and according to the general glucose metabolicpathway and characteristics in mixed culture fermentation of2-KGA from L-sorbose, asimplified metabolic pathway is proposed for G. oxydans. According to the metabolicpathway of G. oxydans, the stoichiometric balance equations are built for L-sorbosemetabolism in G. oxydans. Taking account of the growth lag of G. oxydans afterinoculation, a regulator model is introduced to describe the lag phases in the course ofcultivation. The effect of B. megaterium to G. oxydans is ascribed to that of the deathrate of B. megaterium to the specific L-sorbose uptake rate of G. oxydans. Thus, a macrokinetic model is developed for G. oxydans. The structured model of2-KGAfermentation process is consequently proposed when the macrokinetic model iscombined with a bioreactor model. The specific substrate uptake rate and the specificgrowth rate obtained from the macrokinetic model are then coupled into a bioreactormodel such that the relationship between the substrate feeding rate and the main statevariables, such as the medium volume, the biomass concentrations, the substrateconcentration and the product concentration is set up. The structured model mayprovide a theoretical basis on the practice of optimizing2-KGA fermentation processby operating the external variables. A series of experiments are carried out for modelvalidation. The testing results indicate that the L-sorbose and2-KGA concentration canbe well described by the proposed structured model. The simulation data of G. oxydansand B. megaterium is consistent with the results of the light microscopic examinationsin this cultivation. Their growth characteristics are also in agreement with the cell massdata in the publications.Based on the structured model, a optimized feeding strategy is proposed forimproving the productivity of2-KGA fermentation process. Firstly, the feasibilitystudy of improving the2-KGA fermentation efficiency is performed by a series of feedbatch experiments. With respect to the batch fermentation, feed batch experiments areperformed to evaluate the effect of different feeding culture, such as pulsed feedingculture, constant feed rate culture and linearly increasing feed rate culture. The resultdemonstrates that the yield or productivity of the feed batch experiment is alwaysbetter than that of the batch experiment. The start time-point of feeding is alsoinvestigated in fed-batch cultures with constant feed rate in order to determine theoptimum start time of feeding. Then, based on the structured model of2-KGAfermentation process, a feeding strategy is designed and carried out by controlling thespecific growth rate of G. oxydans in the mixed culture. The experimental resultsindicate that the optimized feeding strategy can ultimately improve the productivity of2-KGA fermentation process by maximizing the potential of G. oxydans convertingL-sorbose to2-KGA. The test results also suggest that the optimized feeding strategy isa feeding process with an increasing feeding rate in this process.