Establishment of Higher Passage PER.C6 Cells for Adenovirus Manufacture


  • Marina Berdichevsky,

    1. Fermentation and Cell Culture, Bioprocess R&D, Merck & Co., Inc., P.O. Box 4, West Point, Pennsylvania 19486
    Search for more papers by this author
  • Marie-Pierre Gentile,

    1. Fermentation and Cell Culture, Bioprocess R&D, Merck & Co., Inc., P.O. Box 4, West Point, Pennsylvania 19486
    Search for more papers by this author
  • Benjamin Hughes,

    1. Fermentation and Cell Culture, Bioprocess R&D, Merck & Co., Inc., P.O. Box 4, West Point, Pennsylvania 19486
    Current affiliation:
    1. Biopharmaceutical Development, GlaxoSmithKline, King of Prussia, PA 19406
    Search for more papers by this author
  • Peter Meis,

    1. Fermentation and Cell Culture, Bioprocess R&D, Merck & Co., Inc., P.O. Box 4, West Point, Pennsylvania 19486
    Current affiliation:
    1. University of California, Berkeley, Dept. of Chemical Engineering, Berkeley, CA 94720
    Search for more papers by this author
  • Joseph Peltier,

    1. Fermentation and Cell Culture, Bioprocess R&D, Merck & Co., Inc., P.O. Box 4, West Point, Pennsylvania 19486
    Current affiliation:
    1. Ohio State University School of Medicine, Columbus, OH 43210
    Search for more papers by this author
  • Ilse Blumentals,

    1. Fermentation and Cell Culture, Bioprocess R&D, Merck & Co., Inc., P.O. Box 4, West Point, Pennsylvania 19486
    Current affiliation:
    1. Australian Institute for Bioengineering and Nanotechnology, University of Queensland, Brisbane, Queensland 4072, Australia
    Search for more papers by this author
  • John Auniņš,

    1. Fermentation and Cell Culture, Bioprocess R&D, Merck & Co., Inc., P.O. Box 4, West Point, Pennsylvania 19486
    Search for more papers by this author
  • Nedim Emil Altaras

    Corresponding author
    1. Fermentation and Cell Culture, Bioprocess R&D, Merck & Co., Inc., P.O. Box 4, West Point, Pennsylvania 19486
    • Fermentation and Cell Culture, Bioprocess R&D, Merck & Co., Inc., P.O. Box 4, West Point, Pennsylvania 19486. Ph: (215) 652–0592. Fax: (215) 652–8691
    Search for more papers by this author


PER.C6 cells, an industrially relevant cell line for adenovirus manufacture, were extensively passaged in serum-free suspension cell culture to better adapt them to process conditions. The changes in cell physiology that occurred during this passaging were characterized by investigating cell growth, cell size, metabolism, and cultivation of replication-deficient adenovirus. The changes in cell physiology occurred gradually as the population doubling level, the number of times the cell population had doubled, increased. Higher passage PER.C6 (HP PER.C6) proliferated at a specific growth rate of 0.043 h-1, 2-fold faster than lower passage PER.C6, and were capable of proliferation from lower inoculation cell densities. HP PER.C6 cell volume was 16% greater, and cellular yields on glucose, lactate, oxygen, and amino acids were greater as well. In batch cultures, HP PER.C6 cells volumetrically produced 3-fold more adenovirus, confirmed with three different constructs. The increase in productivity was also seen on a cell-specific basis. Although HP PER.C6 were more sensitive to the “cell density effect”, requiring lower infection cell densities for optimal specific productivity, they proliferated more after infection than lower passage PER.C6, increasing the number of cells available for virus production. The extensive passaging established HP PER.C6 cells with several desirable attributes for adenovirus manufacture.


The demand for large-scale production of adenoviral (Ad) vectors has grown as they have become the most widely used delivery vector in clinical trials, with over 240 trials completed worldwide over the past 2 decades (1). Adenoviruses have become the vector of choice because of their ability to replicate at high titers in complementary cell lines and to effectively deliver therapies such as vaccines, cancer therapeutics, and therapeutic genes (13). Complementing E1-deleted adenovirus with a cell line expressing the E1 gene can produce replication-deficient adenovirus, and two examples of recombinant cell lines that have been employed for this purpose are human embryonic kidney cells (HEK 293) transformed with Ad type 5 (Ad5) DNA fragments (4) and human embryonic retinoblasts (PER.C6) transformed with Ad5 E1A and E1B-encoding sequences (5). Recombination between adenovirus and HEK 293 cells is capable of generating replication-competent adenovirus (RCA), an undesirable product. The advantage of using PER.C6 cells is the absence of overlapping sequences with Ad5 vectors, minimizing the risk of producing RCA (1) and making PER.C6 cells an industrially relevant cell line.

Similar to HEK 293 cells, PER.C6 cells can maintain a high specific virus productivity at infection cell densities that are significantly lower than the peak cell densities obtained during uninfected cell growth, otherwise known as the “cell density effect”. It has been proposed that this effect is due to nutrient limitation, accumulation of an inhibiting byproduct, lowered intracellular pH (pHi), and an altered distribution of cells in the cell cycle (2, 68).

Several studies have looked into improving production of adenovirus in PER.C6 cells in stirred-tank bioreactors. Both Xie et al. (9) and Maranga et al. (10) found that cell metabolism increases after infection. To minimize nutrient limitation, Maranga et al. (10) performed medium exchanges on uninfected and infected cultures. The medium exchanges were able to increase the peak cell density of uninfected cultures from 3 to 9 × 106 cells/mL as well as to increase the virus titer at 1 and 3 × 106 cells/mL when compared to control cultures without medium exchanges. However, medium exchanges did not extend the range of high specific virus productivity, and one of the conclusions that the authors made was that, beside environmental factors, virus production is dependent on cell physiology at the time of infection. Zhang et al. (8) studied the effect of cell cycle distribution on specific Ad5 productivity and also have concluded that the physiological state of the cells at the time of infection is an important parameter.

The studies described in this article investigated how PER.C6 cell behavior, physiology, and virus production changed as a result of extensive passaging and adaptation to the cell culture environment, leading to the production of “higher passage” PER.C6 cells. The cells were characterized at different population doubling levels, the number of times the population had doubled, on the basis of cell growth, size, metabolism, and virus productivity.

Materials and Methods

Cell Growth. Human transformed embryonic retinoblasts (PER.C6; 5) were obtained from Crucell (Leiden, Netherlands), adapted to suspension culture under serum-free conditions, and extensively passaged in the laboratory. To distinguish higher passage cells from lower passage cells, higher passage cells were referred to as HP PER.C6 (higher passage PER.C6). The cells were routinely cultured in the same proprietary serum-free medium supplemented with 6 mM l-glutamine (Mediatech Inc., Herndon, VA; SAFC Biosciences, Lenexa, KS).

The following vessels were used to culture HP PER.C6 and PER.C6 cells: 125−500 mL shake flasks (Corning, Corning, NY) with a working volume of 50−200 mL; 2 L Wave bioreactor cellbags (Wave Biotech, Bridgewater, NJ) with a working volume of 1 L; 3 L stirred-tank bioreactors (Sartorius, Goettingen, Germany) with a working volume of 1.8−2.0 L; or custom 300 and 1000 L stirred-tank bioreactors with working volumes of 250 and 800 L, respectively.

The shake flasks were gassed with 5% CO2 in air at inoculation and were agitated at 100 RPM. The Wave bioreactors were rocked at 20 RPM and 8° and were continually gassed with 7.5% CO2 in air (the CO2 concentration was determined for each system individually to optimize cell growth and virus production). Both 3 and 300 L stirred-tank bioreactors were maintained with dissolved oxygen (DO) and pH controlled at 50% air saturation and 7.15, respectively (9). All culture vessels were maintained at 36.5 °C. Samples were removed for cell enumeration, metabolite measurement, and cell fixation.

Cell Banking. During the continuous passaging of the cells, portions of the cultures were used to generate cell banks. The purpose of the cell banks was to preserve the cells at various passages and corresponding population doubling levels (PDLs), enabling direct side-by-side comparisons of the cells' properties. Briefly, the cells were concentrated either by filtration or centrifugation, resuspended in medium containing dimethyl sulfoxide (Sigma, St. Louis, MO), and cryogenically frozen.

The PDL of the Crucell PER.C6 research Working Cell Bank was arbitrarily assigned to zero (11). As the cells continued to be passaged, their PDL increased, and in this work cells with PDL ≥ 300 are considered as HP PER.C6 and cells with PDL ≤ 100 are considered as simply PER.C6.

Adenovirus Production. First-generation Ad5 and Ad6 vectors (E1-deleted) expressing HIV-1 gag, pol, nef, and green fluorescent protein (GFP) transgenes were used. Cells were infected in stirred-tank bioreactors at cell densities of 0.2−1.0 × 106 viable cells/mL with a multiplicity of infection (MOI) of 35 adenovirus particles per cell (Ad vp/cell). The infected cells were then harvested at a time that has been previously determined to maximize virus recovery of each construct (ranging from 47 to 78 h), using an approach described by Kamen and Henry (2). To determine virus concentration, samples were treated with 0.1% Triton X-100 (Fisher, Pittsburgh, PA) for 1 h at 18−21 °C with continuous mixing, followed by centrifugation to remove cellular debris. The resulting supernatant was collected as the Triton X-100 lysed (TL) samples to determine the total virus concentration. Samples were analyzed immediately or stored at −70 °C without additives for later analysis. Storage of the samples at −70 °C did not impact measurement of the virus concentration (data not shown).

Analytical Methods. Cell concentration and viability were determined with a Cedex Automated Cell Culture Analyzer (Innovatis AG, Bielefeld, Germany) that utilizes the trypan blue exclusion method and image analysis software. Cell size was determined with a CASY 1 Cell Counter (Reutlingen, Germany) that measures particle volume by detecting differences in electrical resistance. Cell size was also determined by analyzing fixed cells using electron microscopy. On-line glucose and lactate measurements were performed with a YSI2700 unit (YSI, Yellow Springs, OH) withdrawing cell-free samples through 0.22 μm in situ FISP probe (Flownamics Analytical Instruments, Madison, WI) connected to a bioreactor. To determine amino acid concentrations, samples were clarified via centrifugation to remove cellular debris, frozen at −70 °C, and analyzed via HPLC.

Adenovirus particle (Ad vp) concentrations were measured by anion exchange chromatography (12, 13) using the TL samples to represent the virus concentration produced. The relative standard deviation (RSD) associated with this method is 10%. Reported values of volumetric productivity are normalized to the lowest productivity within a group.

Theoretical Calculations. The doubling time (Td) was determined using the following equation:

equation image(1)

where μ is the cell-specific growth rate; μ was computed using a standard growth equation and a semilog regression with the experimental data:

equation image(2)

where XV0 is the initial viable cell concentration. The specific nutrient consumption and metabolite production rates (qs) were determined using:

equation image(3)

where S is the nutrient or metabolite concentration. According to eq 3, a positive qs value means that the metabolite is net produced, whereas a negative value means that the metabolite is net consumed. Assuming that qs is constant during exponential growth, eq 3 can be integrated to yield:

equation image(4)

where S0 is the initial substrate concentration.

To determine qs values for both nutrients and metabolites, the integral of viable cells (IVC), defined by

equation image(5)

was computed using the trapezoidal rule, and the qs values were obtained through linear regression of eq 4 to experimental data. Only data from exponential growth phase were used to calculate Td and qs values. To determine the rate of glutamine consumption accurately, eq 3 was modified to include the first-order chemical degradation of glutamine to ammonium:

equation image(6)

where kdeg is the first-order glutamine degradation constant, previously determined to be 0.0023 h−1 under the conditions used in this work (14). Equation 6 can be integrated to yield

equation image(7)

where Gln0 is the initial glutamine concentration. Gln dt and XV dt were both computed using the trapezoidal rule.

The specific oxygen consumption rate (qO2) was computed using linear regression of experimental data collected during the exponential growth phase of the culture and the relationship

equation image(8)

where OUR is the oxygen uptake rate, calculated using

equation image(9)

and the static method to determine kLa.

Cellular yields on glucose, lactate, oxygen, and amino acids were computed to account for differences in growth rates between HP PER.C6 and PER.C6 cultures, allowing for a direct comparison of their metabolism. Cellular yield is defined by

equation image(10)

and determined using linear regression or by

equation image(11)

Metabolic parameters of HP PER.C6 cells were normalized to PER.C6 cells to account for volumetric differences between these cells. The normalization was performed by assuming that the cells are spherical and by calculating the ratio R of HP PER.C6 to PER.C6 cell diameters (Dx):

equation image(12)

The normalized metabolic parameters were then computed by either multiplying or dividing by R, depending on the units of the parameter.

Cell-specific virus production was calculated using two methods: by dividing the measured virus concentration in the TL sample (in units of Ad vp/mL) by the cell density at infection or by the peak cell density during infection (both in units of cells/mL). Reported values of cell-specific productivity are normalized to the lowest value within a group.

Results and Discussion

Cell Growth. PER.C6 and HP PER.C6 cells were repeatedly cultured in various vessels and working volumes. Examples of cell growth in 250 L working volumes are illustrated in Figure 1. The cells exhibited exponential cell growth until the peak viable cell densities were reached. Viabilities remained above 90% throughout the exponential growth phase (data not shown) and declined only after reaching peak cell density, which was 3.0 × 106 cells/mL for both cell sources in the culture medium used for this work. The PER.C6 doubling time was consistent with values reported previously during generation of PER.C6 clones (15).

Figure Figure 1..

Viable cell concentrations of PER.C6 and HP PER.C6 cells in 250 L working volume as a function of time. The least-squares regression during exponential phase and calculated specific growth rates are shown. In parentheses is the standard deviation of the least-squares regression. Error bars represent the standard deviation of duplicate samples analyzed on the Cedex.

The change in cell growth rates and corresponding doubling times (shown in Figure 1) between PER.C6 cultures of various PDLs appeared to be monotonic. Seven PER.C6 cell banks ranging in PDL from 38 to 405 were expanded side-by-side for five passages in duplicate and exhibited growth rates that increased with PDL (Figure 2).

Figure Figure 2..

Average specific growth rates of six PER.C6 cell banks expanded through five passages (Passage 1, the vial thaw, was not included in calculating the specific growth rates because cell growth is significantly slower during this passage than any other). Error bars represent standard deviations of duplicate cultures expanded from Passages 2 through 5.

The improvement in cell growth was most likely due to the cells continually adapting to their cell culture environment. Chinese Hamster Ovary (CHO) cells may require up to 100 passages to adapt to suspension growth in a serum-free environment (16, 17) and, for the purpose of comparison, the PER.C6 cultures used in this work spanned approximately 90 passages. The original pool of PER.C6 cells was adapted to suspension culture during three passages (14 days), and it is likely that the cells required much longer than this short period to fully adapt to suspension culture in serum-free medium. Similar to PER.C6, CHO cells have been shown to experience a continuous increase in cell growth rate as PDL increased by approximately 300, following serum removal (18). HEK 293 cells in monolayer culture have also been shown to grow faster with increasing passage number, and the authors suggest that the Ad E1 gene integrated into the cells may alter cell growth through continued expression (19). Furthermore, PER.C6 cells are aneuploid (11) and therefore likely subject to mutation and adaptation, making an alternate explanation that the faster-growing cells diluted out the slower-growing cells unlikely.

HP PER.C6 cells were also able to withstand low inoculation cell densities (ICD). The use of lower ICDs became a practical necessity to maintain a passaging schedule that supported exponential cell growth, while compensating for the faster growth of HP PER.C6 cells. The data in Figure 1 is an example of this. PER.C6 cells were routinely inoculated at ≥0.1 × 106 cells/mL, whereas HP PER.C6 cells were routinely inoculated at ≥0.04 × 106 cells/mL, without impact on cell growth. To determine the lower limit on ICDs, shake flasks were inoculated with exponentially growing PER.C6 and HP PER.C6 cells for two passages using a range of ICDs: 0.01, 0.02, 0.05, and 0.1 × 106 cells/mL. Doubling times from the first passage are illustrated in Figure 3.

Figure Figure 3..

Doubling times of PER.C6 and HP PER.C6 cultures inoculated at various inoculation cell densities (ICDs) in shake flasks. Error bars represent the standard deviation of the least-squares regression.

At 0.05 and 0.1 × 106 cells/mL, cell growth was consistent during both passages of both cell types and was similar to historical data: Td ≈ 30 h for PER.C6 and Td ≈ 20 h for HP PER.C6. At an ICD of 0.02 × 106 cells/mL, PER.C6 cell growth slowed during the first passage, and nearly stopped during the second passage, while HP PER.C6 cells continued to grow consistently. At an ICD of 0.01 × 106 cells/mL, PER.C6 cell growth nearly stopped (Td > 100 h). HP PER.C6 cells grew as expected during the first passage but slowed during the second passage (Td = 86 h).

The importance of ICD is well-documented in literature. Lower ICDs have been shown to result in slower cell growth of bone marrow and neural stem cells (20, 21) and Sf9 insect cells (22). At lower ICDs the bone marrow stem cells were more sensitive to the presence of a co-culture of stromal cells. Sf9 cells could be inoculated at 0.03 × 106 cells/mL without increasing their doubling time but reproducible growth curves could not be obtained using ICDs below 0.02 × 106 cells/mL. Such examples indicate that cells depend on the presence of other cells for survival, and a decreased dependence is indicative of the cells' increased robustness. HP PER.C6 have been used in multiple large-scale manufacturing batches and their performance has been consistent with this finding; HP PER.C6 have been more robust than PER.C6 based on consistency of cell growth and viability. The increase in growth rate is also a desirable attribute because it permits a shorter production process where less time is required to expand the cells to the desired volume and cell density.

Cell Size. The CASY 1 Cell Counter was used to measure cell size on different days of exponential cell growth during batch cultures of PER.C6 and HP PER.C6 cells. HP PER.C6 cells were bigger than PER.C6 cells: average diameters measured during exponential cell growth were 18.5 (± 0.4, n = 9) μm and 17.7 (± 0.3, n = 11) μm, respectively. A mean difference of 0.6 μm had a p value of 0.033 (>95% probability). Using this value and eq 12, the ratio R of HP PER.C6 cell volume to PER.C6 cell volume was determined to be 1.16; i.e. HP PER.C6 cell volume was 16% greater than PER.C6 cell volume. Representative examples of the size distributions in PER.C6 and HP PER.C6 cultures during similar stages of exponential cell growth (1.0 and 0.8 × 106 cells/mL, respectively) are illustrated in Figure 4.

Figure Figure 4..

Size distributions in PER.C6 and HP PER.C6 culture samples obtained during exponential growth (cell densities were 1.0 and 0.8 × 106 cells/mL, respectively) in shake flasks and measured using the CASY instrument. The smaller peaks correspond to nuclei released from non-viable cells and the larger peaks correspond to viable cells.

The Cedex instrument, which employs image analysis, also consistently measured a larger diameter for HP PER.C6 cells (data not shown). Additionally, fixed cells were analyzed using electron microscopy, and PER.C6 cells were found to have an average diameter of 14.0 μm, whereas HP PER.C6 cells had an average diameter of 16.5 μm.

The increased size of HP PER.C6 cells was likely related to their increased cell growth, but the relationship between mammalian cell size and proliferation is not well-understood (23) and appears to depend on the cell line. Generally, a cell grows in size and mass as it proceeds through the cell cycle. There is more evidence that cell size at least partially controls cell proliferation than vice versa, through size checkpoints. The molecular mechanisms of these checkpoints have not been elucidated, but one proposal is that protein synthesis rates control cell size (24). Changing the cell size through chemical means has been shown to control proliferation rates of rat glioma cells (25), while blocking progression through the cell cycle did not prevent the cells from increasing their mass and volume (26).

The increased cell size and growth rate of HP PER.C6 cells supported the conclusions made in these studies. An increased size could lead to a faster growth rate because bigger cells would reach the size checkpoints for cell cycle progression sooner than smaller cells. The increased growth rate of HP PER.C6 cells could be a cumulative result of adaptation and increased cell size; however, what caused the cell size to increase with PDL is unknown. Measuring the cell cycle phase distribution and comparing it between PER.C6 and HP PER.C6 cells could provide useful insights; however, PER.C6 aneuploidy has so far made such measurements difficult and inconclusive.

Cell Metabolism. HP PER.C6 glucose and oxygen consumption, lactate production (Table 1), and amino acid metabolism (Table 2) are compared to those previously reported using PER.C6 cells (14). Cell-specific metabolic rates are reported before and after normalizing HP PER.C6 metabolic parameters to the smaller PER.C6 cells. The normalization was necessary because the increased HP PER.C6 cell size is likely accompanied by increased cell mass (27), supported by increased nutrient consumption and metabolite production.

Table Table 1.. Main Metabolic Parametersa for PER.C6 and HP PER.C6 Cultures
 PER.C6bHP PER.C6HP PER.C6c normalized to PER.C6 cell volume
  1. a As defined by eq 3, negative qs values indicate the nutrient is consumed, and positive values indicate that the metabolite is produced. Shown in parentheses is the standard deviation associated with least-squares regression.b Reported by Maranga and Goochee, 2006 (14).c Since HP PER.C6 cells were determined to be larger in volume than PER.C6 cells, the metabolic rates and yields of HP PER.C6 were normalized to those of PER.C6 cells by dividing or multiplying by the ratio R defined by eq 12.d The standard deviation is not provided because these values were calculated using eq 11 in the Materials and Methods section.

qGlc (pmol/cell hr)−0.128 (±0.005)−0.199 (±0.003)−0.172
qLac (pmol/cell hr)0.180 (±0.004)0.282 (±0.006)0.243
qO2(pmol/cell hr)−0.225 (±0.003)−0.1743 (±0.0008)−0.150
YX/Glc (1012cells/mol)0.171 (±0.009)0.204 (±0.002)0.236
YX/Lac (1012 cells/mol)0.121d0.156 (±0.003)0.181
YX/O2 (1012 cells/mol)0.0969d0.25b0.29
YO2/Glc (mol/mol)1.76 (±0.05)0.876b0.876
Table Table 2.. Amino Acid Consumption/productiona Rates and Yield Ratios for PER.C6 and HP PER.C6 Cultures
amino acidPER.C6b (fmol/cell h)HP PER.C6 (fmol/cell h)ratio of yieldsc (YHP PER.C6/YPER.C6)
  1. a As defined by eq 3, negative qs values indicate the nutrient is consumed, and positive values indicate that the metabolite is produced. Shown in parentheses is the standard deviation associated with least-squares regression.b Reported by Maranga and Goochee, 2006 (14).c These ratios were determined by first using eq 11 to calculate the PER.C6 and HP PER.C6 yields on each amino acid, then by using eq 12 to normalize HP PER.C6 yields to PER.C6 yields, and last by dividing the normalized HP PER.C6 yield by the PER.C6 yield.d The qGln was computed taking into account the spontaneous glutamine chemical degradation as explained in Materials and Methods.

Ala5.4 (±0.2)5.8 (±0.1)2.2
Arg−4.6 (±0.7)−7.4 (±0.6)1.5
Asn−0.34 (±0.04)−2.6 (±0.2)0.3
Asp−2.8 (±0.2)−6.36 (±0.05)1.0
Cys−0.7 (±0.09)−2.3 (±0.3)0.7
Glnd−1.7 (±2)−47.2 (±0.2)0.9
Glu2.9 (±0.5)3.9 (±0.7)1.7
Gly2.0 (±0.2)4.6 (±0.4)1.0
His−1.0 (±0.1)−2.46 (±0.04)0.9
Ile−4.0 (±0.7)−9.4 (±0.2)1.0
Leu−6.1 (±0.9)−14.3 (±0.3)1.0
Lys−3.7 (±0.9)−9.9 (±0.3)0.9
Met−1.7 (±0.5)−3.9 (±0.3)1.0
Phe−1.5 (±0.2)−4.2 (±0.2)0.8
Pro3.5 (±0.4)7.5 (±0.6)1.1
Ser−9.9 (±0.9)−2.8 (±1)0.8
Thr−2.3 (±0.2)−5.7 (±0.3)0.9
Trp−0.60 (±0.05)−1.0 (±0.1)1.4
Tyr−1.3 (±0.1)−2.9 (±0.3)1.0
Val−4.2 (±0.7)−10.88 (±0.09)0.9

HP PER.C6 cells consumed nutrients and produced metabolites faster than PER.C6 cells, even after accounting for the increase in volume; however, calculating cellular yields revealed that HP PER.C6 produced more biomass per quantity of glucose and oxygen consumed and lactate produced. Interestingly, PER.C6 consumed 1.8 mol of oxygen for every mole of glucose, whereas HP PER.C6 consumed only 0.9 mol of oxygen for every mole of glucose. Together with increased cellular yield, this difference implies that HP PER.C6 utilized glucose more toward increasing biomass and less toward oxidative metabolism.

For most amino acids, HP PER.C6 cellular yields were approximately equivalent or greater than PER.C6 yields after accounting for volumetric differences. Asparagine yield appeared to be significantly lower, but this phenomenon could be attributed to a decrease in asparagine synthesis. Asparagine is a non-essential amino acid that is both consumed and produced. If PER.C6 and HP PER.C6 had similar asparagine requirements but HP PER.C6 produced less of this amino acid, the net consumption of asparagine would appear greater in HP PER.C6 cells.

Adenovirus Production. Volumetric virus production was measured using anion exchange chromatography for three different adenovirus constructs produced in large-scale batches, representing both Ad5 and Ad6 serotypes (Figure 5). The differences in large-scale productivity were representative of differences observed at multiple smaller scales and in various production vessels. Although the volumetric productivity varied between the constructs, HP PER.C6 cells consistently produced 3-fold more of each construct than PER.C6 cells. An increased volumetric productivity translates into smaller quantities of raw materials and a shorter production process since the required final volume of cells can be reduced to produce the same desired amount of adenovirus product.

Figure Figure 5..

Volumetric productivity of three different adenovirus constructs in PER.C6 and HP PER.C6 cells. The cultures were infected in large-scale (≥250 L working volume) stirred-tank bioreactors at the optimum cell density for volumetric production: 0.8 × 106 PER.C6 cells/mL and 0.4 × 106 HP PER.C6 cells/mL. Multiplicity of infection was 35 Ad vp/cell. Optimal harvest times had been previously determined for each construct, were not different between the two cell types (data not shown), and ranged from 47 to 78 h post-infection. Anion exchange chromatography was used to determine titers and its RSD is 10%.

To ensure a change in productivity did not affect infectivity, the particle to infectivity ratios were determined. The same purified Ad5 construct produced in either cell type had a particle to infectivity ratio of 1.5 Ad vp/IU.

This improvement in productivity was also seen on a pre-infection cell-specific basis (Figure 6). When PER.C6 and HP PER.C6 cells were infected at the same cell density of 0.4 × 106 cells/mL, HP PER.C6 cells produced 6- to 7-fold more virus per cell; however PER.C6 cells were capable of maintaining a maximal specific yield over a wider range of infection cell densities (also referred to as the “cell density effect”) than HP PER.C6 cells. PER.C6 cells exhibited a consistent productivity (based on the pre-infection cell density) of 1.63 (±0.08) between 0.3 and 0.9 × 106 cells/mL, and 1.0 at a pre-infection cell density of 1.0 × 106 cells/mL. HP PER.C6 cells had a sharply decreasing productivity with increasing infection cell density.

Figure Figure 6..

Cell-specific productivities at different infection cell densities in PER.C6 and HP PER.C6 cells in 3 L stirred-tank bioreactors. The pre-infection cell density was the basis for calculating cell-specific productivity in this case. Multiplicity of infection was 35 Ad vp/cell and harvest times were 67 ± 5 h post infection.

In order to maximize volumetric productivity, PER.C6 cultures were infected at the highest cell density within the range of consistent per-cell productivity: ∼0.8 × 106 cells/mL. In the case of HP PER.C6 cells, the optimal infection cell density (that maximized volumetric productivity) was ∼0.4 × 106 cells/mL. Since at 0.4 × 106 cells/mL, the difference in per-cell productivity was 6- to 7-fold but the PER.C6 infection cell density was twice as great, the net result was the expected 3-fold difference in volumetric productivity.

The cells' behavior after infection helped explain how HP PER.C6 cells were able to produce 6- to 7-fold more virus on a per-cell basis when using the pre-infection cell density as basis for the calculation (Figure 7). They continued to grow during at least the first 15 h of infection, such that the peak cell density was approximately 100% greater than the pre-infection cell density, higher than has been previously reported (28). Therefore, a greater number of cells was available for virus production. In contrast, PER.C6 cell density increased approximately 30% during infection, consistent with earlier findings (28). When the peak cell density was used to calculate per-cell productivity instead of the pre-infection cell density (Figure 8), the difference was approximately 3-fold. This is a more accurate but less frequently used method of estimating the volumetric productivity.

Figure Figure 7..

Viable cell concentrations of PER.C6 and HP PER.C6 cells before and after infection with adenovirus. Error bars represent standard deviation of duplicate samples analyzed on the Cedex.

Figure Figure 8..

Side-by-side comparison of adenovirus productivity at various PDLs of PER.C6 cells in 3 L stirred-tank bioreactors. The peak cell density during infection was used as the basis for productivity calculations. Infection cell densities ranged from 0.20 to 0.52 × 106 cells/mL to obtain comparable results between PER.C6 and HP PER.C6 cells and to minimize the cell density effect in HP PER.C6 cells. Multiplicity of infection was 35 Ad vp/cell and harvest times were 67 ± 5 h post-infection. Error bars represent standard deviations of productivity in duplicate cultures.

Cell growth arrest occurs after adenovirus infection through inhibition of protein and DNA syntheses (29, 30). Because of their faster cell growth, HP PER.C6 cells were able to complete more replication cycles than PER.C6 cells before cell cycle arrest occurred, leading to the observed cell growth after infection. HP PER.C6 were infected with adenovirus expressing green fluorescent protein (GFP) and analyzed using FACS to monitor the kinetics of the infection. Despite cell growth during infection, the MOI and harvest times used in this work allowed 100% infection of the cells (data not shown).

The effect of cell density on virus production has been previously documented but not fully explained, although it has been proposed that the effect results from depletion of key nutrients or accumulation of inhibitory byproducts at higher cell densities (2). Ferreira et al. (7) studied the cell density effect in Ad5-producing 293 cells and showed that at higher infection cell densities, the intracellular pH (pHi) decreased more during infection. He hypothesized that lower pH activated DNase II, which, in turn, led to apoptosis and degradation of Ad5 DNA.

HP PER.C6 could have appeared to be more sensitive than PER.C6 to the cell density effect potentially because of the significant cell growth during infection. The increase in cell number likely led to the same cell density effects that PER.C6 experienced at higher infection cell densities. The HP PER.C6 peak cell density during infection was 0.8 × 106 cells/mL (Figure 7), which was also the maximum PER.C6 cell density that did not reduce the specific productivity.

Beside environmental factors, the physiological state of the cells could also have impacted specific productivity. Zhang et al. (8) showed that cell cycle distribution was dependent on the cell density; specifically, the percentage of cells in the S phase, which is when synthesis of viral DNA occurs, decreased at higher cell densities. In HEK 293 cells, cell-specific adenovirus production increased with passage number and was accompanied by an increase in the percentage of cells in the S phase (19). Measuring PER.C6 cell cycle distribution would be useful for this purpose, but as stated above, such measurements have so far been difficult and inconclusive because of the cells' aneuploidy.

After infecting six PER.C6 cell sources, the increase in per-cell productivity appeared to be gradual with respect to PDL (Figure 8). One hypothesis to explain this behavior proposed that E1 expression increased from PER.C6 to HP PER.C6 cells, thereby increasing the rate of adenovirus replication. E1 protein expression and E1 DNA copy number were assayed for both PER.C6 and HP PER.C6 cells using Western Blot and QPCR, respectively, and neither was found to be significantly different between the two cell types (data not shown). Another hypothesis was that the increased productivity was related to the state of the cells, analogous to the relationship between cell size and cell proliferation. Lloyd et al. (31) sorted CHO cells on the basis of cell size using centrifugal elutriation and found that production of selected proteins was dependent on the cell size, not the cell cycle phase. Specifically, protein production increased with cell size. Virus replication involves protein synthesis as well, but production of viral particles and production of proteins, such as monoclonal antibodies, are very different processes. Gilbert et al. (32) showed that while the expression of a marker gene such as GFP is indicative of viral protein expression, it is not indicative of adenovirus productivity. Further work is required to elucidate how cell size, cell cycle distribution, DNA replication, and assembly of viral particles are related in PER.C6 cells.


Higher passage PER.C6 cells have been established to support adenovirus production. After extensive passaging (that likely furthered adaptation of the cells to suspension culture in serum-free medium), the cells were shown to possess several desirable attributes for manufacturing. Their growth rate and robustness increased, enabling a shorter, more consistent production process. The cell size increased as well and potentially contributed to the increased growth rate through regulation of the cell cycle. Metabolic changes were evidenced by increased cellular yields on glucose and oxygen consumption, lactate production, and both consumption and production of amino acids.

HP PER.C6 cells were shown to have a greater capacity for adenovirus production than PER.C6 cells, volumetrically producing 3-fold more adenovirus (which was consistent among three different constructs). HP PER.C6 were more sensitive to the cell density effect and to maintain a high cell-specific productivity required a lower infection cell density than PER.C6 cells. However, HP PER.C6 were capable of proliferating post-infection, and when the increase in cell number was accounted for, HP PER.C6 produced 3-fold more adenovirus than PER.C6 on a cell-specific basis. The difference in productivity could not be explained by changes in E1 expression and DNA copy number between PER.C6 and HP PER.C6 cells and could be related to cell cycle regulation imposed by the increased cell size of HP PER.C6.

PER.C6 cells better adapted to suspension culture growth and adenovirus production have been established, and work is ongoing exploring the mechanisms responsible for the physiological differences between PER.C6 and HP PER.C6 cells.


The authors thank Charles F. Goochee for scientific guidance, Franklin Lu for data on the cell density effect; Sheetal Pai for data on infection with GFP-containing adenovirus; and Jayanthi Wolf for data on E1 expression and DNA copy number data. The authors also thank Christian Metallo, Liangzhi Xie and many other contributors within Fermentation and Cell Culture and the Merck Research Laboratories who have provided innumerable insights, ideas, and discussions during the course of this work.