Scale down of recombinant protein production: a comparative study of scaling performance

A large bioreactor is heterogeneous with respect to concentration gradients of substrates fed to the reactor such as oxygen and growth limiting carbon source. Gradient formation will highly depend on the fluid dynamics and ma ss transfer capacity of the reactor, especially in the area in which the su bstrate is added. In this study, some production-scale (12 m(3) bioreactor) conditions of a recombinant Escherichia coli process were imitated on a la boratory scale. From the large-scale cultivations, it was shown that locall y high concentration of the limiting substrate fed to the process, in this case glucose, existed at the level of the feedpoint. The large-scale proces s was scaled down from: (i) mixing time experiments performed in the large- scale bioreactor in order to identify and describe the oscillating environm ent and (ii) identification of two distinct glucose concentration zones in the reactor. An important parameter obtained from mixing time experiments w as the residence time in the feed zone of about 10 seconds. The size of the feed zone was estimated to 10%. Based on these observations the scale-down reactor with two compartments was designed. It was composed of one stirred tank reactor and an aerated plug flow reactor, in which the effect of osci llating glucose concentration on biomass yield and acetate formation was st udied. Results from these experiments indicated that the lower biomass yiel d and higher acetate formation obtained on a large scale compared to homoge neous small-scale cultivations were not directly caused by the cell respons e to the glucose oscillation. This was concluded since no acetate was accum ulated during scale-down experiments. An explanation for the differences in results between the two reactor scales may be a secondary effect of high g lucose concentration resulting in an increased glucose metabolism causing a n oxygen consumption rate locally exceeding the transfer rate. The results from pulse response experiments and glucose concentration measurements, at different locations in the reactor, showed a great consistency for the two feeding/pulse positions used in the large-scale bioreactor. Furthermore, me asured periodicity from mixing data agrees well with expected circulation t imes for each impeller volume. Conclusions are drawn concerning the design of the scale-down reactor.