ABSTRACT

Bioprocess design engineers face a set of challenges unlike those faced by process designers in most other industries. Perhaps the most important of these is that process designers cannot interact with the “machinery” of bioproduction directly, and so the quality attributes of the product are manipulated primarily by selection of the production cell line and by the optimization of the process and conditions under which those cells are cultured (Anderson and Reilly 2004; Butler 2005; Chun et al. 2003; Dowd et al. 1999; Farid 2007; Jain and Kumar 2008; Kunkel et al. 2000). A second challenge is posed by the fact that the “machinery” is complex in that a large number of process parameters, for example, pH, dissolved oxygen (DO), temperature (T), and nutrient levels can influence cellular processes, thereby impacting the quality attributes of the protein produced (Byrne et al. 2007). This multivariate optimization of a multivariate parameter set is clearly a more demanding task than that of simple optimization toward a single objective. Furthermore, there are, as yet, no rigorous models to explain the impact of process design choices on the quality

12.1 Introduction .................................................................................................. 223 12.2 Challenge One: Controlled Cell Culture in a Microbioreactor ....................225 12.3 Challenge Two: Adding Automation for Throughput ................................... 229 12.4 Challenge Three: Evaluating the Model and Interpreting the Data ............. 232 12.5 Challenge Four: Assembling a Complete Solution .......................................234 12.6 Application and Next Steps .......................................................................... 236 Acknowledgments .................................................................................................. 237 References .............................................................................................................. 237

attributes of the resulting bioproducts, so most of this process understanding must be developed empirically.