Evaluation of the effects and interactions of gas blending and feeding strategy on a Fab' fermentation process by Escherichia coli
A method to improve production of Fab' fragments using gas blending and pH-stat feeding strategy in a fermentation process with Escherichia coli has been developed. Regime analysis together with design of experiments (DoE) has been used to evaluate the effect of gas blending and feeding strategy on a Fab' fermentation process. The economic implications of the fermentation strategy have also been considered. Batch-feeding fermentations carried out at 20 L and 450 L scale indicated that cascade control was not sufficient to maintain a constant DOT level throughout the fermentation. DOT levels dropped to zero during induction phase at both scales. Regime analysis was performed based on experimental KLa determination. KLa values of -400 h"1 were observed at 20 L and 450 L scale. Comparison of time of oxygen consumption (tQc) and time of oxygen transfer (t0i) suggested that 02 limitation is present in this fermentation process and that this worsens as the scale increases. A gas blending system at 20 L scale was proposed to address this problem. A factorial 22 experimental design was executed to evaluate independently the effects and interactions of two main engineering factors (related to oxygen transfer in the broth) on Fab' titre: DOT level and agitation rate. Gas blending was successful in maintaining constant levels of DOT at 20 L scale. Fab' production was increased by 77 % at of agitation rate of 500 rpm independent of the DOT level, compared to operation at the same scale but without gas blending. High levels of product localisation in the periplasm of 84 - 93% were also obtained. Furthermore, based on t0c and t0T, it could be suggested that no oxygen limitation is likely to occur in the fermentations performed with gas blending, regardless of the agitation rate. Batch-fed fermentations either with or without gas blending carried out at 20 L scale indicated the presence of glycerol oscillations. A pH-stat feeding strategy in a gas blending system was implemented to address this problem. Results showed that a 2-fold increase in the production of Fab' at 20 L scale, compared to a fermentation operated in a pulsed fed- batch mode could be achieved. A consistently high level (>90%) of product localisation in the periplasm was also achieved and no negative impact on product recovery was observed. Finally, a preliminary economic analysis was performed to estimate the production costs in £/mg of product. The effects of gas blending and feeding strategy at 20 L scale on the final product cost were evaluated by comparing: batch-fed non gas-blending, batch-fed gas blending, and pH-stat gas blending fermentations. A fermentation production cost of £1.45/mg of Fab' was estimated for a pH-stat gas blending system. This resulted in cost savings of up to 75% when compared to the production cost of £5.80/mg of Fab' in the batch-fed system without gas blending. The results obtained in this work provide an impetus for further studies to evaluate the potential of gas blending and pH-stat feeding strategy for the industrial production of Fab' antibody fragments. The use of statistical design of experiments together with regime analysis was found to be a very useful tool to gain a better understanding of this system.