Summary: | This thesis illustrates the need for biologies manufacturing processes to be considered as a whole, as the operation of the upstream process can greatly alter the performance of downstream steps. It develops methods to allow this to be done at minimal cost and with the minimum requirement for process material. The industrial bioprocess for antibody fragment (Fab9) production in E. coli is used as the case study. Two variations of the antibody production process were examined. In the first lactose was used both to induce Fab' production and to provide the post-induction energy source. Cell growth was controlled by phosphate limitation. The second process used isopropyl p-D-l-thiogalactopyranoside (IPTG) for Fab' induction, with a glycerol feed providing energy post-induction. Both variations produced high cell density broths (ca. 20% solids w/v), and industrially relevant quantities of Fab' (ca. 1 g/L). However, the IPTG-induced fermentation was chosen for detailed study as it was the most controllable and reproducible of the two. The fermentation was investigated together with the initial unit operations of broth centrifugation, heat/chemical Fab9 extraction, and spheroplast centrifugation. Methods were developed to allow all aspects of the two centrifugation steps in the initial process to be mimicked at small-scale. Loss of cell integrity upon entry and exit to the industrial centrifuge was mimicked through the use of a rotating disc and capillary jet device respectively. Clarification was mimicked through the adaptation of an existing method to high cell density feeds ( > 10% w/v) by the addition of a dilution step and a mathematical correction based on the Richardson-Zaki equation. Dewatering of the solids cake was predicted to within 8% at small-scale using a novel method that maintains industrial spin speed, spin time, and height of the sediment cake at millilitre-scale. All small-scale methods developed were verified at pilot-scale. The heat/chemical Fab' extraction step was mimicked in microfuge tubes. Outputs, in terms of % total protein and Fab' release, were shown to be within 6% of pilot- scale results in all cases tested. Any differences between pilot- and small-scale performances were shown not to be statistically significant. Product quality, as assessed by SDS-PAGE, was shown to be very similar to pilot-scale. As the periplasmic extraction was identified as the step most detrimental to overall yield of the initial DSP the microfuge method was used in a 2-factor statistical design of experiments study to analyse the robustness of the extraction procedure. The process was then considered as a whole. Post-induction glycerol feed rate was varied and the impacts upon subsequent processing analysed using both the methods developed and pilot-scale data. This demonstrated the usefulness and accuracy of the small-scale methods developed, and also showed that optimizing the feed rate for maximum Fab' production during fermentation did not result in optimum Fab9 production after initial DSP. Instead a higher feed rate should be used to produce cells with weaker periplasmic membranes to facilitate more effective Fab9 release upon heat/chemical extraction. The thesis concludes with a discussion of the business relevance of the small-scale methods developed. Their general applicability for candidate screening, process development and process understanding is suggested, and the regulatory issues relevant to their use are explored.
|