Perfusion cultures of human tumor cells: a scalable production platform for oncolytic adenoviral vectors.

This study demonstrates the novel use of the HeLaS3 human tumor cell line for propagating ONYX-411, a recombinant oncolytic adenoviral vector. HeLaS3 cells enabled high levels of vector production without the risk of generating vector recombinants, which is possible with HEK293 cells. The development of a high-cell-density perfusion process using ATF technology yielded production levels as high as 6 x 10(11) vp/mL, which was approximately sevenfold greater than the titers achieved in fed-batch bioreactors. Several experiments were performed at the bench (15 L) and pilot (70 L) scales to demonstrate the robust and scalable nature of this industrially relevant technology.

Changes in the overall extent of protein glycosylation by Chinese hamster ovary cells over the course of batch culture.

Glycosylation of recombinant proteins can change during the culture of animal cells. Since lipid-linked oligosaccharides (LLOs) are the carbohydrate donors in N-linked glycosylation, their availability is postulated to influence the extent of glycosylation. To test this hypothesis, relative LLO and glycosylation levels in Chinese hamster ovary (CHO) cells were monitored over the course of batch culture for corresponding changes. Radiolabelling studies reveal that throughout the length of culture, intracellular LLO levels remained within a 2-fold range and overall CHO protein glycosylation varied by less than 30%. The implications of these results and hypotheses to explain the findings are discussed.

Glycosylation by Chinese hamster ovary cells in dolichol phosphate-supplemented cultures.

N-linked glycosylation often imparts important properties to protein therapeutics. An essential step in this intracellular process is the transfer of oligosaccharide from dolichol monophosphate (Dol-P) to a potential glycosylation site. Variability in the success rate of this reaction affects the extent of protein glycosylation. The critical role of Dol-P suggests that its availability may influence the extent of glycosylation by limiting the pool of lipid-linked oligosaccharides (LLOs), the glycosyl donor. To test this hypothesis, the impact of Dol-P supplementation on protein glycosylation in Chinese hamster ovary (CHO) cells was investigated. Although exogenous Dol-P was incorporated by CHO cells and processed into LLOs in a dose-dependent manner, Dol-P supplementation had no marked effects on LLO or overall cellular glycosylation levels. While concentrations of exogenous Dol-P exceeding 100 microg/ml were detrimental to CHO cell viability, maximum non-toxic supplemental doses of Dol-P had no significant impact on the glycosylation of recombinant interferon- gamma produced by batch cultures of CHO cells. These results show that glycosylation in CHO cells cannot be readily enhanced by Dol-P feeding under normal culture conditions.

A GFP-based screen for growth-arrested, recombinant protein-producing cells.

The growth of anchorage-dependent Chinese hamster ovary (CHO) cells is arrested upon serum deprivation; however, a portion of these cells remain viable for extended time periods in serum-free culture. This work presents a strategy to both rapidly generate a heterogeneous population of CHO cells as well as to select for subpopulations that remain robust and continue to produce recombinant protein when their growth is arrested. Stable expression of recombinant proteins in mammalian cells is often a tedious and time-consuming process because only a small percentage of transfected cells will express sufficient quantities of protein. To overcome the limitations associated with standard transformation and selection methods, bicistronic retroviral expression technology was used. First, bicistronic retroviral constructs encoding for both interferon gamma (IFN-gamma), the model therapeutic protein, and green fluorescent protein (GFP), the quantitative selectable marker, were generated. Next, recombinant retroviruses were obtained from transient transfection of a helper-cell line and were used to infect susceptible CHO cells. Cells with the bicistronic expression module stably integrated into their genome fluoresce green and could thereby be easily isolated by fluorescence-activated cell sorting. Upon subjecting successfully infected cells to serum withdrawal, significant declines in cell viability and GFP expression occurred. After imposing this selection pressure on the cells for 8 days, GFP producers were isolated from the survivors by fluorescence-activated cell sorting and expanded. To evaluate the effectiveness of the screening process, the selected cells were exposed to a second round of serum deprivation. Unlike the original cell population from which it was derived, the subpopulation remained robust and continued to stably express both GFP and IFN-gamma throughout the extended period of serum-free culture. Within 2 weeks, cells selected for recombinant protein production under serum-free conditions were successfully generated and isolated.