Evaluation of scaling parameters towards an improved process development strategy for CHO cell perfusion cultures-step-wise scale-up from 15 mL to 5 L

Abstract

Process development towards the production of difficult-to express proteins in perfusion cultivation necessitates the implementation of scale-down models to manage the high number of experiments which have to be conducted in Quality by Design (QbD) initiatives. Successive simplification of the process mode accompanied by the application of single-use equipment provides possibilities to mimic more complex processes in smaller scales in less time, with less work and less cost intensive conditions. The reverse up-scaling and transfer of process conditions and set-points from simple approximated cultivation systems to the final, more complex production method requires the application of appropriate scalable culture parameters. In this work, scaling parameters were evaluated to systematically transfer culture criteria between three bioreactor systems and cultivation modes towards an improved process development strategy. The capability of simpler small-scale systems to approximate ATF-perfusion processes was assessed. Chinese Hamster Ovary (CHO) cells expressing a recombinant protein were cultured in 15 mL Advanced Microscale Bioreactor (ambrTM) system in pseudo-continuous mode, 1 L DASGIP multiparallel bioreactor system in continuous mode, and 5 L Sartorius B-DCU system in Alternating Tangential Flow (ATF)-perfusion mode. Main critical process parameters (CPP) were analyzed, i.e. cell specific metabolic rates and productivity of the cell cultures, compared and transferred based on the evaluated scaling parameters. Different aspects of the particular processes and their impact on cell culture were assessed in order to build up an improved process understanding. Process parameters were found to be sufficiently transferable when scaled by the cell specific perfusion rate (CSPR). Bench-top bioreactors operated in continuous and perfusion culture modes showed coinciding metabolic rates but were distinct from ambrTM pseudo-continuous cultures where a stable offset over a wide range of CSPR was characterized. Cell specific protein productivity (qP) was similar between DASGIP and ambrTM cultures but different from ATF-perfusion cultures. Aside from an imprecise measurement method, a partial retention of recombinant protein in the ATF-module was determined to have contributed to this observation. Moreover, deviating pCO2 was identified as a possible impact factor on qP. The differences were attributed to the physical characteristics, i.e. mass transfer, and different scales of the bioreactors rather than the different operation modes. Additionally, the biomass measure which is used for the calculation of cell specific rates was identified to play a crucial role when CPP should be transferred with the presented scaling-parameters. In general, transferability was feasible since similar trends were observed for metabolic rates and recombinant protein related data.