Modeling And Experiments Of Fermentation For Human Interleukin-11 With Recombinant E.coli

Recombinant human interleukin-11(rhIL-11) is still the only therapy drug for thrombocytopenia. Although the biological properties and clinical behaviors of rhIL-11 have been studied extensively, its bioprocess development data and modeling have rarely been reported.In this work, based on former models in the literature, a new three-compartment model was designed, which was characterized by the introduction of the hypothesis of product feed-back inhibition into the kinetics of product accumulation. The model was able to simultaneously simulate dynamic changes of biomass concentration (x), the single substrate concentration (S) and the concentrations of the three compartments in the model (XA, XG and XP).The parameter estimation problem was converted into an optimization process, i.e. a course searching for a group of estimated valued which led to the minimization of simulation errors. Genetic algorithm was adopted as the searching tool for its global searching ability. In the algorithm, binary coding was chosen; the initial population number was set to be 100; roulette wheel selection was utilized; single point crossover with the possibility of 0.9 and single point mutation with the possibility of 0.05 was used.Experimental data collected from both flask and fementor cultivation were used for parameter estimation of the new model. With the optimized values of parameters, the model gave a fine simulation of experimental data.It was the first time to combine simple structured model with genetic algorithm for the modeling of fermentation. Such combination should be applicable for the modeling of other foreign protein production during the microbial fermentation process.