Abstract
Cell culture process development requires the screening of large numbers of cell lines and process conditions. The development of miniature bioreactor systems has increased the throughput of such studies; however, there are limitations with their use. One important constraint is the limited number of offline samples that can be taken compared to those taken for monitoring cultures in large-scale bioreactors. The small volume of miniature bioreactor cultures (15 mL) is incompatible with the large sample volume (600 µL) required for bioanalysers routinely used. Spectroscopy technologies may be used to resolve this limitation. The purpose of this study was to compare the use of NIR, Raman, and 2D-fluorescence to measure multiple analytes simultaneously in volumes suitable for daily monitoring of a miniature bioreactor system. A novel design-of-experiment approach is described that utilizes previously analyzed cell culture supernatant to assess metabolite concentrations under various conditions while providing optimal coverage of the desired design space. Multivariate data analysis techniques were used to develop predictive models. Model performance was compared to determine which technology is more suitable for this application. 2D-fluorescence could more accurately measure ammonium concentration (RMSECV 0.031 g L-1 ) than Raman and NIR. Raman spectroscopy, however, was more robust at measuring lactate and glucose concentrations (RMSECV 1.11 and 0.92 g L-1 , respectively) than the other two techniques. The findings suggest that Raman spectroscopy is more suited for this application than NIR and 2D-fluorescence. The implementation of Raman spectroscopy increases at-line measuring capabilities, enabling daily monitoring of key cell culture components within miniature bioreactor cultures. © 2017 American Institute of Chemical Engineers Biotechnol. Prog., 33:337-346, 2017.