This article describes the step‐wise approach undertaken to select a serum‐free medium (SFM) for the efficient production of a recombinant adenoviral vectors expressing β‐galactosidase (Ad5 CMV‐LacZ), in the complementing human embryonic kidney 293S cells. In the first step, a 293S‐derived transfectoma, secreting a soluble epidermal growth factor receptor sEGFr (D2‐22), was used to estimate the potential of selected serum‐free formulations to support the production of a recombinant protein as compared to serum‐containing medium. Assays showed that only one among six commercial serum‐free formulations could support both sEGFr production and cell growth in static or suspension culture. In commercially available calcium‐containing serum‐free formulations, the cell aggregates reached up to 3 mm in diameter. In the second step, 293S cells were gradually adapted to a low‐calcium version of the selected medium (LC‐SFM). Cells were cloned, then screened according to their ability to grow at a rate and an extent comparable to parental cells in serum‐containing medium (standard) as single cells or small aggregates. The 293SF‐3F6 clone, first adapted to and then cloned in the selected serum‐free medium, was selected for further experiments. Bioreactor run performed with the 293SF‐3F6 clone showed similar growth curve as in the shake‐flask controls. In the final step, the recombinant viral vector productivity of the 293S cells and the 293SF‐3F6 clone was tested. The 293SF‐3F6 cells infected by Ad5 CMV‐LacZ in 3 L‐scale bioreactor maintained the specific productivities of both β‐galactosidase and adenoviral vector equivalent to the shake‐flask controls in suspension culture. Results from this study clearly demonstrate that the 293SF‐3F6 cell line thus selected may be used either for establishing stable transfected cell line or for the production of adenoviral vectors required for gene therapy studies. © 1998 John Wiley & Sons, Inc. Biotechnol Bioeng 59: 567–575, 1998.

Human 293S cells, a cell line adapted to suspension culture, were grown to 5 x 10(6) cells/mL in batch with calcium-free DMEM. These cells, infected with new constructions of adenovirus vectors, yielded as much as 10 to 20% recombinant protein with respect to the total cellular protein content. Until recently, high specific productivity of recombinant protein was limited to low cell density infected cultures of no more than 5 x 10(5) cells/mL. In this paper, we show with a model protein, Protein Tyrosine Phosphatase 1C, how product yield can be maintained at high cell densities of 2 x 10(6) cells/mL by a medium replacement strategy. This allows the production of as much as 90 mg/L of active recombinant protein per culture volume. Analysis of key limiting/inhibiting medium components showed that glucose addition along with pH control can yield the same productivity as a medium replacement strategy at high cell density in calcium-free DMEM. Finally, the above results were reproduced in 3L bioreactor suspension culture thereby establishing the scalability of this expression system. The process we developed is used routinely with the same success for the production of various recombinant proteins and viruses.

Human 293S cells, a cellline adapted to suspension culture, were grown to 5 x 10 6 cells/mL in batch with calcium-free DMEM. These cells, infected with new constructions of adenovirus vectors, yielded as much as 10 to 20% recombinant protein with respect to the total cellular protein content. Until recently, high specific productivity of recombinant protein was Iimited to low cell density infected cultures of no more than 5 xl 0 5 cells/mL. In this paper, we show with a model protein, Protein Tyrosine Phosphatase 1 C, how high product yield can be maintained at high cell densities of 2 xl 0 6 cells/mL by a medium replacement strategy. This allows the production of as much as 90 mglL of active recombinant protein per culture volume. Analysis of key Iimiting/inhibiting medium components showed that glucose addition along with pH control can yield the same productivity as a medium replacement strategy at high cell density in calcium-free DMEM. Finally, the above results were reproduced in 3L bioreactor suspension culture thereby establishing the scalability of this expression system. The process we developed is used routinely with the same success for the production of various recombinant proteins and viruses.Abbreviations: CFDMEM -calcium-free DMEM; CS -bovine calf serum; hpi -hours post-infection; J+ -enriched Joklik medium; MLP -major late promoter; MOl -multiplicity of infection (# of infectious viral particIe/cell); q -specific consumption rate (mole/cell.h); pfu -plaque forming unit (# of infectious viral particIe); Y -yield (/LgIE6 cells or mole/cell)

The human adenovirus/293S cell expression system is used for the production of either recombinant protein or adenovirus vectors for use in gene therapy. In this work, the production of protein tyrosine phosphatase (PTP1C) was used as a model for the scale‐up of both applications. Maximum specific production of 30 to 45 μg of active protein/106 cells was maintained upon infection with adenovirus vectors at cell densities between 2 × 106 to 3 × 106 cells/mL in a 3.5‐L bioreactor. This was achieved by resuspending the culture in fresh medium at infection time. The pH was kept at 7.0 throughout the experiment and, at 24 h postinfection, glucose and essential amino acids were added. Attempts to replace the complete change of medium at the time of infection with nutrient supplementation of the used medium led to lower production levels, suggesting that protein expression was limited not by the absence of a key nutrient but by inhibitory factors. Two potentially inhibitory factors were investigated: lactic acid accumulation and increased osmolarity. Medium acidification such as that which would be brought about by lactic acid accumulation was shown to depress PTP1C production. The lactate molecule itself decreased the cell viability when added in concentrations of 20 mM or more. But the specific productivity was affected at higher lactate concentrations of 40 mM or more. Additions of glucose, amino acids, and NaHCO3 used to control pH, led to increases in osmolarity. Osmolarities above 400 mOsm lowered cell density. However, specific production was not significantly affected below 500 mOsm. But, at 500 mOsm, PTP1C production peak was shifted from 48 to 72 hpi. Because of the cell loss, this per cell yield increase did not translate into higher volumetric production. When glucose concentrations was kept at 5 mM by fed‐batch addition, lactate production and increases in osmolarity were reduced. In shake flasks, this method permitted maximum production with cells resuspended either in fresh or spent medium at infection. This fed‐batch process was implemented successfully at the 3.5‐L scale. Fed‐batch with glucose may provide a means to increase infected‐cell density beyond 3 × 106 cells/mL.