Generating stable, high-producing cell lines for recombinant protein production requires an understanding of the potential limitations in the cellular machinery for protein expression. In order to increase our understanding of what makes a stable high producer, we have generated a panel of 17 recombinant monoclonal antibody expressing Chinese hamster ovary subclones (CHO-mAb) with specific productivities ranging between 3 and 75 pg cell À1 day À1 using the dihydrofolate reductase (dhfr) expression system and compared the molecular features of these high-and lowproducer clones. The relative heavy chain (HC) and light chain (LC) transgene copy numbers and mRNA levels were determined using real-time quantitative PCR (RT qPCR). We observed that not only higher transgene copy numbers and mRNA levels of both HC and LC were characteristic for the high-producer clones as compared to the low-producer clones but also a more favorable HC to LC transgene copy numbers ratio. By studying the long-term stability of the CHO-mAb subclones in the absence of methotrexate (MTX) selective pressure over 36 passages we observed a 35-92% decrease in volumetric productivity, primarily caused by a significant decrease in HC and LC mRNA levels with little change in the transgene copy numbers. Using Southern blot hybridization we analyzed the HC and LC transgene integration patterns in the host chromosome and their changes in course of gene amplification and long-term culturing. We observed that MTX-induced gene amplification caused chromosomal rearrangements resulting in clonal variability in regards to growth, productivity, and stability. No further obvious DNA rearrangements occurred during long-term culturing in the absence of MTX, indicating that other mechanisms were responsible for the decreased transcription efficiency. Our results implicate that the amplified transgene sequences were arranged in tandem repeats potentially triggering repeat-induced gene silencing. We hypothesize that the decline in transgene mRNA levels upon long-term culturing without MTX was mainly caused by transgene silencing consequently leading to a loss in mAb productivity. The exact molecular mechanisms causing production instability are not yet fully understood. The herein described extensive characterization studies could help understand the limitations to high-level, stable recombinant protein production and find ways to improving and accelerating the process for high-producer cell line generation and selection.

Current industry practices for large-scale mammalian cell cultures typically employ a standard platform fed-batch process with fixed volume bolus feeding. Although widely used, these processes are unable to respond to actual nutrient consumption demands from the culture, which can result in accumulation of by-products and depletion of certain nutrients. This work demonstrates the application of a fully automated cell culture control, monitoring, and data processing system to achieve significant productivity improvement via dynamic feeding and media optimization. Two distinct feeding algorithms were used to dynamically alter feed rates. The first method is based upon on-line capacitance measurements where cultures were fed based on growth and nutrient consumption rates estimated from integrated capacitance. The second method is based upon automated glucose measurements obtained from the Nova Bioprofile FLEX® autosampler where cultures were fed to maintain a target glucose level which in turn maintained other nutrients based on a stoichiometric ratio. All of the calculations were done automatically through in-house integration with a Delta V process control system. Through both media and feed strategy optimization, a titer increase from the original platform titer of 5 to 6.3 g/L was achieved for cell line A, and a substantial titer increase of 4 to over 9 g/L was achieved for cell line B with comparable product quality. Glucose was found to be the best feed indicator, but not all cell lines benefited from dynamic feeding and optimized feed media was critical to process improvement. Our work demonstrated that dynamic feeding has the ability to automatically adjust feed rates according to culture behavior, and that the advantage can be best realized during early and rapid process development stages where different cell lines or large changes in culture conditions might lead to dramatically different nutrient demands.

No abstract