Biochemical Engineering Journal 39 (2008) 297304

Analysis of cell-cycle-dependent production of tissue plasminogen activator analogue (pamiteplase) by CHO cells
Masami Yokota a,b, , Yasunori Tanji b
a

Fermentation and Biotechnology Labs, Technology Section, Astellas Pharma Inc., 156 Nakagawara, Kiyosu-shi, Aichi 452-0915, Japan b Department of Bioengineering, Tokyo Institute of Technology, Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, 4259 J2-15 Nagatsuta, Midori-ku, Yokohama 226-8501, Japan Received 19 July 2007; received in revised form 25 September 2007; accepted 29 September 2007

Abstract The tissue plasminogen activator analogue (pamiteplase) gene was introduced into CHO cells, and several methotorexate (MTX) resistant clones were isolated. In a medium turnover culture of the recombinant CHO cells under varied serum concentrations, the specic growth rate of cells and the specic production rate of pamiteplase showed positive dependencies on serum concentrations; however, the relationship between them was not linear. To investigate the relationship between cell growth and pamiteplase production during the cell cycle, synchronous cultures of MTX-resistant clones were grown and mitotic selection was applied using various concentrations of growth factor (Daigos GF21). The production rate of pamiteplase showed an immediate response to the differences of GF21 concentrations during their cell cycle transition. The production rate peaked at G2/M phase at a lower level with clone PM200 and at a higher level with its derivative clone PM400, which are resistant to 200 and 400 nM MTX concentration, respectively. Clone PM400 had two peaks of pamiteplase production, a smaller one at early S phase and a larger one at G2/M phase, and the heights of these peaks were proportional to the concentration of GF21. Pamiteplase production varied more dramatically than cell cycle time depending on the medium conditions, and each parameter was affected independently. The cell cycle time became saturated at a lower concentration of GF21; however, the pamiteplase production rate kept increasing even at higher concentrations. 2007 Elsevier B.V. All rights reserved.
Keywords: Cell cycle; CHO; Recombinant protein production; Specic production rate; Cell cycle time; Mitosis; t-PA; Modelling

1. Introduction The production of human therapeutic proteins through recombinant animal cell culture is one of the fastest-growing areas of the pharmaceutical industry. It is well known that animal cells have three cell cycle phases, namely, G1, S, and G2/M. Understanding the relationship between the cell cycle and protein expression is critical when attempting to optimize the medium and culture conditions for commercial production of recombinant proteins in animal cell culture. In the development of a production process, it is very important to understand the relationship between cell growth rate and target protein production rate. For positively growth-associated production, a cell culture process that promotes cell proliferation

Corresponding author at: Fermentation and Biotechnology Labs, Technology Section, Astellas Pharma Inc., 156 Nakagawara, Kiyosu-shi, Aichi 452-0915, Japan. Tel.: +81 52 400 8343; fax: +81 52 401 4405. E-mail address: masami.yokota@jp.astellas.com (M. Yokota).

is necessary to attain high protein production levels. However, for negatively growth-associated production, which several studies have reported for monoclonal antibody (MAb) production, it is necessary to develop procedures which retain cells in culture vessels and maintain high cell viability despite a low growth rate. It is important to study cell-cycle-dependent characteristics in order to understand the relationship between cell growth and target protein production. Although many previous cell cycle studies have been done using both endogenous and foreign genes [1,8,11,12,16,18], the relationship between the cell cycle and protein production under various medium conditions is still unclear. In order to get basic information for process development, we investigated whether the recombinant CHO cells transformed with a tissue plasminogen activator (t-PA) analogue (pamiteplase) [17,26] would have growth-associated or nongrowth-associated production characteristics. Additionally, we evaluated the effect of serum concentration on either the growth rate or pamiteplase production rate. The relationship between the growth rate and the production rate over the entire culture

1369-703X/$ see front matter 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.bej.2007.09.016

298

M. Yokota, Y. Tanji / Biochemical Engineering Journal 39 (2008) 297304

was found to be non-linear. Investigating the kinetic characteristics of cell cycle behind these phenomena was considered to be useful to optimize the pamiteplase production process. Cell-cycle-dependent target protein productivity could vary with the cell line, the nature of the recombinant gene expressed, or the promoter/enhancer used to generate product expression. Most studies have focused on the relationship between the specic production rate and production phase of the target protein. However, it is also important to remember that overall process productivity might depend on the specic production rate and cell cycle time. There have been no studies investigating the effect of the medium on both specic production rate and cell cycle time simultaneously. The use of expression plasmids amplied by the presence of the gene for dihydrofolate reductase (DHFR) [in the presence of methotorexate (MTX)] is one popular way of creating and maintaining a high-number of gene copies in DHFR-decient host cells. There have been no studies on differences in either specic protein production rate prole or cell cycle time between the parent and the derivative DHFR-amplied recombinant CHO cells. In this study, the pamiteplase gene was introduced into CHO cells, and several clones were isolated during the process of making stepwise increases in MTX concentration. Investigating the differences in cell cycle characteristics between the parent cells and the derivative cells will give useful information to select competent cells suitable for more efcient protein production. The aim of this study is to determine the cell-cycle-dependent recombinant protein production prole and subsequently determine how media conditions affect both this prole and the cell cycle time. In addition, a comparison was made between the cell-cycle-dependent pamiteplase production rate proles and cell cycle times of the parent cells and their derivative clones, which are resistant to increases in MTX concentration. 2. Materials and methods 2.1. Cells and cell culture methods Two transformed cell lines, PM200 and PM400, were used for synchronization studies. CHO DHFR-negative cells [10] were transfected with a plasmid (pKSV10) carrying pamiteplase gene expressed from the SV40 early promoter and mouse DHFR gene expressed from the ALMP promoter. The DHFR gene was used for gene amplication under selective pressure of up to 400 nM MTX. The amplied sequences were integrated into the CHO genome. Both cell lines contained an insert for pamiteplase which was co-amplied with the DHFR gene in the presence of MTX. Pamiteplase is a modied recombinant t-PA that contains a nger domain, a growth factor domain, a kringle-2 domain, and a serine protease domain, as well as a point mutation at the kringle-2-serine protease linkage site (del 92173, 275 Arg Glu) [26]. Cell lines have been selected in a stepwise manner for resistance to MTX [20]. The PM200 cell line is stably resistant to 200 nM MTX. The PM400 cell line was cloned from cells that had been passaged several times from clone PM200 while being adapted to medium containing

400 nM MTX. Cells were maintained as monolayer cultures in 75-cm2 T-asks containing 20 mL of () minimum essential medium (MEM) supplemented with 10% fetal bovine serum (FBS) (Gibco, Grand Island, NY), and the T-asks were incubated in a humidied 5% CO2 incubator at 37 C. The viable cell concentration was estimated using an automated cell-counter (Vi-CELL XR 2.03, Beckman Coulter) and the trypan-blue exclusion method. 2.2. Medium turnover cultivation method of clone PM200 Seed cells were maintained as monolayer cultures in 75-cm2 T-asks containing 20 mL of () MEM supplemented with 10% FBS. Seed cells were inoculated in 60-mm dishes at a concentration of 1.5 105 cells/mL containing 5 mL of () MEM supplemented with 0.3, 1, 3, and 10% FBS. The dishes were incubated in the humidied 5% CO2 incubator at 37 C. Fresh medium was put in the plates every day. Before the assay, the spent medium was drained, and 2 mL fresh medium was added, after which the plates were incubated for 23 h in the humidied 5% CO2 incubator at 37 C. The supernatant was then harvested and assayed by ELISA to determine the pamiteplase concentration, and the cells were trypsinized and counted using a cell counter (SYSMEX Corporation). Two dishes of each serum concentration were assayed every day so that the reported concentrations of cells and pamiteplase were the average of two dishes. 2.3. Synchronization by mitotic cell selection Clone PM400 cells were allowed to grow to approximately 50% conuency in 75-cm2 T-asks, at which time they were subjected to mitotic selection. This was performed according to the method of Mariani et al. [1] with the following modications: at 50% conuence, the medium was drained and 5 mL of fresh medium was added. The ask was then tapped ve times on each side. The medium containing the dislodged mitotic cells was collected in a centrifuge tube (Falcon) and placed on ice to stop cell cycle progression, and the T-ask was re-fed with fresh medium and placed in the humidied 5% CO2 incubator at 37 C. This procedure was repeated successively three times at intervals of approximately 1 h to collect enough mitotic cells for the cell cycle studies. The medium containing mitotic cells was centrifuged and the cells were resuspended in medium for the synchronization study. The cells were then spread onto 35-mm dishes (Falcon) at a density of about 3 105 cells per dish in 1 mL of the medium. GIT100, GIT30, and GIT15 media were prepared from GIT medium (Nihon Pharmaceutical Co. Ltd.) and Daigos T medium (Nihon Pharmaceutical Co. Ltd.) for synchronization studies. T medium is a chemically dened basal medium. GIT medium is a mixture of T medium and Daigos GF21 solution (WAKO Chemicals) that contains serum factors. In this paper, GIT100, GIT30, and GIT15 indicate a mixture of GIT medium and T medium in ratios of 100:0, 30:70, and 15:85, respectively. Dishes containing mitotic cells were incubated in a humidied 5% CO2 incubator at 37 C.

M. Yokota, Y. Tanji / Biochemical Engineering Journal 39 (2008) 297304

299

A dish was used periodically to determine the viable cell densities and pamiteplase concentrations. The culture supernatants were aliquoted and kept frozen at 80 C for later analysis. Dishes were trypsinized and cell suspension was prepared by pipetting. The cell suspension was applied to a cell analyzer, Vi-CELL XR (Beckman), mixed with trypan-blue solution, and cell density, viability and cell size were measured automatically. The size of a single cell was represented as the average of two diameters measured in vertical and horizontal directions with two-dimensional photo-images of cells. Fifty photo-images were taken by ow-cell system for each sample. Each photoimage contained 512 viable cells depending the cell density. The circularity of cells was calculated by dividing the shorter diameter by the longer diameter of each cell. The values of circularity for the recombinant CHO cells were approximately 0.93 during the synchronized culture, which means that recombinant CHO cells had near perfectly round shape after trypsinization. 2.4. Pamiteplase quantication The enzymatic activity of the pamiteplase secreted into the culture medium was determined by a chromogenic assay. The pamiteplase was incubated in 200 L of 0.3 mM CH3 SO2 -(d)HHT-Gly-Arg-pNA, AcOH (Pentapharma) in a microtitration plate and the change in absorbance at 405 nm was measured (n = 3). Absorbance was converted to pamiteplase concentration (mg/L) by interpolation from the standard curve. t-PA ELISA was used to quantify the t-PA in aliquots of supernatants from the medium turnover culture grown in medium containing serum. The ELISA plates were incubated for 1 h at room temperature (RT) with a coating buffer (200 L, 50 mM phosphate buffer, pH 7.5) containing a goat anti-t-PA IgG antibody. Each plate was incubated a second time for 30 min (RT) with coating buffer containing 0.5% bovine serum albumin. The supernatant samples were diluted with coating buffer, loaded into ELISA plates, and incubated for 1 h (RT). Afterwards, a peroxidase-conjugated anti-t-PA antibody was added, and the sample was incubated for another hour (RT). After washing, the peroxidase substrate was added and the plate sat for 30 min before the absorbance was read at 405 nm. 2.5. Curve tting and non-linear regression analysis Curve tting and non-linear regression analyses were performed using SigmaPlot scientic graphing software (SYSTAT Software Inc.). 3. Results and discussion 3.1. Medium turnover culture with various FBS concentrations The specic growth rate of the cells increased with increasing concentration of FBS (Fig. 1(A)). Each culture showed exponential growth at a rate dependent on the FBS concentration, which means that the declining culture conditions between daily medium changes had no effect on cell growth.
Fig. 1. (A) Growth of clone PM200 in medium turnover culture. Clone PM200 was cultured with medium containing 0.3, 1, 3, and 10% FBS. Cell density is the mean of two dishes. The curve was tted using y = axb . Solid circle: FBS 0.3%; solid triangle: FBS 1%; solid square: FBS 3%; open circle: FBS 10%. (B) Specic production rate of pamiteplase in medium turnover culture specic production rates were measured by replacing the medium with 2 mL of fresh medium, followed by 23 h of incubation at 37 C. The rates were calculated by dividing the pamiteplase concentration change in the supernatant by the cell density. Specic production rate is the mean of two dishes. Solid circle: FBS 0.3%; solid triangle: FBS 1%; solid square: FBS 3%; open circle: FBS 10%. (C) Relationship between specic growth rate and specic production rate. Specic growth rates were calculated by tting the curve in Fig. 1. Solid circle: day 0; solid triangle: day 1; solid square: day 2; open circle: day 3; open triangle: day 4; open square: day 5.

The specic production rate of pamiteplase behaved differently, however. While it also increased with increasing concentrations of FBS at rst, it decreased over time (Fig. 1(B)) in every culture, and the rate of decrease became more rapid

300

M. Yokota, Y. Tanji / Biochemical Engineering Journal 39 (2008) 297304

Fig. 2. Cell growth, pamiteplase concentration and cell size in synchronized culture. Cell growth, pamiteplase concentration, and cell size in batch cultures synchronized using mitotic selection; (A) clone PM200 with GIT100 medium; (B) clone PM400 with GIT15 medium; (C) clone PM400 with GIT30 medium; (D) clone PM400 with GIT100 medium. Solid circle: cell density; open diamond: pamiteplase concentration; open triangle: cell size; dashed line: curve tted for cell density.

at higher concentrations of FBS. Since the pamiteplase production rate was dependent on FBS concentration the synthesis of recombinant protein is thought to be stimulated by several factors present in the FBS but declining culture conditions between daily medium changes are thought to be responsible for the decrease in production rate. The relationship between the specic cell growth rate and the specic pamiteplase production rate seems to be positively correlated, however, non-linear correlations are observed during the early stages of the cultures (Fig. 1(C)). These non-linear correlations do not agree with a previous report [25], which showed a linear correlation between the specic growth rate and the specic production rate of IFN-, at least during the logarithmic growth phase, with various concentrations of FBS. Our results indicate that a non-linear relationship between the specic growth rate and the specic protein production rate could occur under various medium conditions. Other studies have shown that the expression of recombinant t-PA is growthassociated and investigated only the cell cycle-dependency of t-PA production [18]. Another study reported that endogenous DHFR mRNA synthesis increased when the media was changed 7 h before synchronizing the culture [13]. However, there have been no studies investigating how changing the medium conditions just before synchronizing a culture inuences the cell cycle or the recombinant protein production rate.

3.2. Synchronized growth The number of cells in each cell line increased approximately twofold between 14 and 20 h after mitotic selection (Fig. 2), which indicates that almost all cells went through G1, S, and G2/M phases. The growth curve was smoothed by determining the best-t regression-line between the observed points [5]. The curve tted for the cell density and time in Fig. 2 was based on the following logistic function: x = a 1 + e(t c)/b +d (1)

where x is the cell density (cells/mL); t is the time (h) from synchronization; and a, b, c, and d are constants. The constants a, b, c, and d were obtained using the SigmaPlot curve tting program (Table 1). c is the cell cycle time for each clone, which ranged from 16.6 to 18.2 h. The mean cell size at 2 h was smaller than that at 0 h (Fig. 2), which can be explained by the presence of large cells at 0 h (Fig. 3). The cell size distribution at 0 h and the presence of small cells in pairs observed in video images (not shown) indicate that mitotic selection allows for the selection of three types of cells: large cells that are just about to begin mitotic division, paired small cells undergoing mitotic division, and single small daughter cells just after division, which are all weakly attached to the surface of the dishes. Two hours after the beginning of

M. Yokota, Y. Tanji / Biochemical Engineering Journal 39 (2008) 297304 Table 1 Constants of sigmoid formulation PM200 (GIT100) a (cells/mL) b (h) c (h) d (cells/mL) 2.49 105 0.850 18.2 3.58 105 PM400 (GIT15) 2.34 105 0.534 17.9 2.59 105 PM400 (GIT30) 2.10 105 0.738 16.6 2.94 105

301

PM400 (GIT100) 2.48 105 1.29 16.8 2.93 105

synchronous cultures, the large cells disappeared and all cells had a small size, that is, the large cells harvested from mitotic selection completed mitosis within 2 h. The cell size increased continuously from G1 to G2/M phases in each case (Fig. 3). A basic property of these cells in culture is that their volume increases approximately twofold in the transition from the G1 to the G2/M phases. As expected, the volumes of the G2/M cells were approximately twice that of those in G1 phase. A direct relationship between the cell volume and the cell cycle has been reported in several cases [2,4,13,19,21,23]. Clone PM400 showed a smaller and narrower distribution of cell size than clone PM200 (Fig. 3). The cell size of clone PM400 decreased along with the decrease in GF21 concentration. This is in contrast to the phenomenon observed with recombinant

CHO-producing IFN-, which experienced an increase in size when grown in a medium with a low serum concentration [25]. The cell cycle time of clone PM400 was shorter than that of the parental clone, PM200 (Table 1). In general, a higher resistance to MTX leads to a lower specic growth rate [15,25], but in this case, the relationship is inverse. Since a wide distribution has been reported for the relationship between MTX resistance and specic production rate [3], it should be possible to select a clone that has both a higher specic growth rate and a higher production rate in the course of increasing MTX resistance. Clone PM400 had the same cell cycle time when cultured with GIT100 and GIT30, but a longer time with GIT15. The cell cycle time for clone PM400 with GIT15 was the same as that for clone PM200 with GIT100 (Fig. 2).

Fig. 3. Distribution of cell size in synchronized culture. 3D and contour graphs of the distribution of cell size in batch cultures synchronized using mitotic selection; (A) clone PM200 with GIT100 medium; (B) clone PM400 with GIT15 medium; (C) clone PM400 with GIT30 medium; (D) clone PM400 with GIT100 medium.

302

M. Yokota, Y. Tanji / Biochemical Engineering Journal 39 (2008) 297304

3.3. Specic productivity The continuous increase in cell size (Fig. 3) means that the cell cycle traversed in a typical way [2,4,19,23,24]. Both recombinant clones, PM200 and PM400, secreted pamiteplase during every cell cycle phase, but the host cells did not (data not shown). The pamiteplase concentration in the culture medium increased continuously from G1 phase to G2/M phase, and the nal concentration differed between the clones and GF21 concentrations (Fig. 2). The increase in pamiteplase concentration was not constant. The protein production rate was calculated from the differences in pamiteplase concentration between consecutive sample times as well as the cell density calculated from the smoothed curve ttings (Fig. 4). The specic growth rate () during the G2/M phase of the synchronized culture was calculated using the following equations: = 1 dx x dt ct b 1 + exp ct b
2

(2) (3)

dx a = exp b dt

The widths of distribution of the specic growth rate in the G2/M phase narrowed with the decrease in GF21 concentration, which indicates better synchronization. According to studies conducted under similar conditions [14,16], the duration of G1, S, and G2/M should be approximately 5, 8, and 5 h, respectively, with the production phase possibly occurring during the S to G2/M phases for both clones under every culture condition. In this study, the highest peak of the specic production rate occurred during the G2/M phase under every culture condition (Fig. 4). A low-specic production rate peak occurred during the early S phase of clone PM400. This peak was clearly observable in the GIT100 and GIT15 media cultures, but such a peak was not clearly discernable for clone PM200. Specic pamiteplase production rate of both clones began to increase at S phase and peaked at G2/M phase, which means that the amplication process does not affect the regulation of the pamiteplase gene expression. This result is coincident with the study of cell cycle regulation of endogenous CHO DHFR gene expression in both non-amplied and amplied cell lines [13]. The specic pamiteplase production rate of clone PM400 was obviously larger than that of clone PM200 at early S and late SG2/M phases. Clone PM400 is considered to be superior to clone PM200 for two reasons, namely the specic production rate and the cell cycle time as mentioned in Section 3.2. A previous report using similar materials [18] reported the synthesis of recombinant DHFR under control of the ALMP promoter during all cell cycle phases, but particularly during the G1 and early S phases. In contrast, the magnitude of the secretion rate of recombinant t-PA under control of the SV40 early promoter measured by 2 h incubation of cells which were collected (using ow cytometry) during each cell cycle phase was as follows: late-S > early S > G2/M > G1. This previous study depended on samples taken over a rather long sampling inter-

Fig. 4. Proles of specic growth and production rates in synchronized culture. Specic growth and production rates in batch cultures synchronized using mitotic selection; (A) clone PM200 with GIT100 medium; (B) clone PM400 with GIT15 medium; (C) clone PM400 with GIT30 medium; (D) clone PM400 with GIT100 medium; solid circle: specic production rate; dashed line: specic growth rate.

M. Yokota, Y. Tanji / Biochemical Engineering Journal 39 (2008) 297304

303

val, so the details of the specic production rate prole of t-PA were not clear. In addition, the relative production rate of t-PA (late-S/G1) in the current study is higher than that in the previous study [18], particularly with clone PM400. Several studies have reported that the expression phases of endogenous protein and recombinant protein depend upon promoters [16]. It has been reported that SV40 early promoter and CMV promoter perform strongly during the S phase, and ALMP promoter is strong mainly during the G1 phase [9,18]. The current study differs from previous studies using recombinant cells in that there is a two-phase expression prole that occurs during the early S to G2/M phases. This resembles the expression pattern of amplied endogenous DHFR activity that increases strongly at the early S and the G2/M phases in CHO cells [13]. The specic production rate responded to the change in GF21 concentration relatively quickly (Fig. 4). It is notable that the specic production rate of clone PM400 was lower with GIT30 than GIT100, but the cell cycle time was almost the same under both conditions (Table 1 and Fig. 4). In the case of GIT15, in addition to a decrease in the specic production rate, an increase in cell cycle time, the same as seen with clone PM200 with GIT100, was observed. This increase in cell cycle time is disadvantageous for the production of recombinant protein. These observations indicate that, in a batch culture, a decline in medium conditions will cause the height of specic production rate peaks to decrease, particularly during the S and G2/M phases at rst. Later, as medium conditions decline further, the cell cycle time will lengthen, resulting in a decrease in the production rate. This phenomenon partially agrees with known medium turnover culture data (Fig. 1(A)(C)). A deeper understanding of the phenomena in medium turnover culture will require a more detailed synchronized culture study using other materials, i.e. metabolites exhibiting inhibition on growth or production. When attempting to develop an improved medium for higher protein productivity, namely by increasing the nutrient or growth factor concentrations, it is important to note that cell cycle time reached the minimum at GF30 (saturation point), but the specic production rate showed a tendency to increase up to GF100. Consequently, in the case of recombinant protein production in which recombinant protein expression depends on the S or G2/M phase, medium development could improve the specic production rate to a large degree. This is a very attractive option since cell cycle time can be shortened only so much, and medium conditions have an immediate and large inuence on the specic production rate. A previous study reported a linear relationship between the proportion of cells in the G1 phase and the specic IFN- production rate in static cultures supplemented with or without l--phosphatidic acid (PA) [11]. In contrast, this study suggests that recombinant protein production is inuenced not only by the proportions of each cell cycle phase, but also by medium conditions. In fact, these variations might inuence productivity greatly. This idea is consistent with that asserted in another study [5]. Both specic production rate and cell cycle time should be taken into consideration when designing medium components for optimal production.

3.4. Cell cycle model It is useful to construct and use a cell cycle model to optimize the cell culture process. Several cell cycle models were proposed [6,22]. A cell cycle model for antibody production has been proposed which assumes that the mRNA synthesis rate and the translation rate of the target protein are constant, and the mRNA synthesis rate is zero at M phase and translation rate is zero at M or G2/M phase [7]. Based on this model, decreasing growth rate would lead to accumulation of mRNA, and result in an increase in specic production rate. This model is useful for non-growth-associated production, such as antibody production. However, distinct growth-associated production is demonstrated in this study of pamiteplase production. We have shown that pamiteplase can be produced in every cell cycle phase with a small peak at early S and a large peak at late S and G2/M phases. Those peaks vary dramatically under varied medium conditions. Another study revealed that the endogenous DHFR mRNA level in CHO cells varies during a cell cycle and is strongly inuenced by the medium conditions [13]. Based on these ndings, the cell cycle model of pamiteplase synthesis would show an increasing rate of pamiteplase mRNA synthesis and translation from early S to G2/M, which is strongly inuenced by the growth factor concentrations. In this report, growth factor concentration was chosen as a varied condition of the medium. Since accumulation of metabolites, inhibiting factors for cell growth and protein synthesis, would occur in batch culture, a cell cycle model should consider those effects. One of the possible reasons for the observed decrease in the pamiteplase production rate in the late phase of the medium turnover culture is the effect of increased metabolites accompanied with increased cell density. Northern hybridization analysis of synchronized cultures using not only growth factor but also inhibiting factor such as lactate or ammonia will be useful for construction of a cell cycle model for optimization of this cell culture process. 4. Conclusion In a medium turnover culture study using recombinant CHO cells expressing the t-PA analogue, pamiteplase, the specic growth rate was constant at a rate proportional to the serum concentrations. A non-linear relationship was observed between the cell growth rate and protein production rate. This phenomenon is attributable to the synchronization of cultures. Recombinant clone PM200 (resistant to 200 nM MTX) and its derivative clone PM400 (resistant to 400 nM MTX) experienced their production phases mainly from the early S to G2/M phases, that is, the cells show growth-associated production in the batch culture during that time and the expression of pamiteplase genes in both clones are cell-cycle regulated equally. Clone PM400 had two distinct production peaks, a small one during the early S phase and a large one during late S to G2/M phases. The heights of both peaks were clearly proportional to the GF21 concentrations during synchronized culture. The specic production rate and cell cycle time responded to medium conditions independently. When the GF21 concentration in the medium was decreased somewhat, only the specic production rate decreased. However, further reduction

304

M. Yokota, Y. Tanji / Biochemical Engineering Journal 39 (2008) 297304 [11] H. Yamaji, K. Sakai, T. Joho, E. Izumoto, H. Fukuda, Cell cycle analysis of Chinese hamster ovary cells stimulated by phosphatatidic acid in serumfree culture, J. Biosci. Bioeng. 98 (6) (2004) 487489. [12] J. Aggeler, L.N. Kapp, S.C.G. Tseng, Z. Werb, Regulation of protein secretion in Chinese hamster ovary cells by cell cycle position and cell density, Exp. Cell Res. 139 (1982) 275283. [13] J.N. Feder, Y.G. Assaraf, L.C. Seamer, R.T. Shimke, The pattern of dihydrofolate reductase expression through the cell cycle in rodent and human cultured cells, J. Biol. Chem. 264 (1989) 2058320590. [14] M. Al-Rubeai, A.N. Emery, S. Chalder, D.C. Jan, Specic monoclonal antibody productivity and the cell cycle-comparisons of batch, continuous and perfusion cultures, Cytotechnology 9 (1992) 8597. [15] M.B. Gu, J.A. Kern, P. Todd, D.S. Kompala, Effect of amplication of dhfr and lac Z genes on growth and -galactosidase expression in suspension cultures of recombinant CHO cells, Cytotechnology 9 (1992) 237245. [16] M.B. Gu, P. Todd, D.S. Kompala, Foreign gene expression (galactosidase) during the cell cycle phases in recombinant CHO cells, Biotechnol. Bioeng. 42 (1993) 11131123. [17] M. Katou, Y. Suzuki, I. Miyamoto, T. Watanabe, K. Mori, H. Arakawa, H. Gushima, Biochemical and pharmacokinetic properties of YM866, a novel brinolytic agent, Thromb. Haemost. 65 (1991) 1193. [18] M. Kubbies, H. Stockinger, Cell cycle-dependent DHFR and t-PA production in cotransfected, MTX-amplied CHO cells revealed by dual-laser ow cytometry, Exp. Cell Res. 188 (1990) 267271. [19] O.T. Ramirez, R. Mutharasan, Cell cycle- and growth phase-dependent variations in size distribution, antibody productivity, and oxygen demand in hybridoma cultures, Biotechnol. Bioeng. 36 (1990) 839848. [20] R.J. Kaufman, R.T. Shimke, Amplication and loss of dihydrofolate reductase genes in a Chinese hamster ovary cell line, Mol. Cell. Biol. 1 (1981) 10691076. [21] R. Zippel, E. Martegani, M. Vanoni, G. Mazzini, L. Alberghina, Cell cycle analysis in a human cell line (EUE cells), Cytometry 2 (1982) 426430. [22] S.J. Kromenaker, F. Srienc, Cell-cycle-dependent protein accumulation by producer and nonproducer murine hybridoma cell lines: a population analysis, Biotechnol. Bioeng. 38 (1991) 665677. [23] T.H.A.W. Splinter, M. Reiss, Separation of lymphoid-line cells according to volume and density, Exp. Cell Res. 89 (1974) 343351. [24] T. Seewoster, J. Lehmann, Cell size distribution as a parameter for the predetermination of exponential growth during repeated batch cultivation of CHO cells, Biotechnol. Bioeng. 55 (1997) 793797. [25] V. Leelavacharamas, A.N. Emery, M. Al-Rubeai, Growth and interferon production in batch culture of CHO cells, Cytotechnology 15 (1994) 6571. [26] Y. Kawauchi, T. Morinaga, M. Yokota, A. Kinoshita, K. Kawamura, Y. Suzuki, M. Takayama, K. Furuichi, H. Gushima, Gene construction and large scale production of a novel brinolytic agent (YM866) in CHO cells, Thromb. Haemost. 65 (1991) 1193.

of the GF21 concentration prolonged cell cycle time, decreased the specic production rate, and lowered cell cycle production peaks. These ndings partly explain the non-linear relationship between the specic growth rate and the specic production rate observed in medium turnover cultures. They also suggest that the development of a better culture process and medium for improving the recombinant protein productivity of cells with similar production characteristics depends mainly on stimulating the potential of a specic production rate and preventing declining medium conditions while maintaining a minimum cell cycle time. References
[1] B.D. Mariani, D.L. Slate, R.T. Shimke, S phase-specic synthesis of dihydroate reductase in Chinese hamster ovary cells, Proc. Natl. Acad. Sci. U.S.A. 78 (1981) 49854989. [2] C.B. Thompson, I. Scher, M.E. Schaefer, T. Lindsten, F.D. Finkelman, J.J. Mond, Size-dependent B lymphocyte subpopulations: relationship of cell volume to surface phenotype, cell cycle, proliferative response, and requirements for antibody production to TNP-Ficoll and TNP-BA, J. Immunol. 133 (1984) 23332342. [3] C.H. Fann, F. Guirgis, G. Chen, M.S. Lao, J.M. Piret, Limitations to the amplication and stability of human tissue-type plasminogen activator expression by Chinese hamster ovary cells, Biotechnol. Bioeng. 69 (2000) 204212. [4] C.M.L. West, P.C. Keng, R.M. Sutherland, Growth phase related variation in the radiation sensitivity of human colon adenocarcinoma cells, Int. J. Radiat. Oncol. Biol. Phys. 14 (1988) 12131219. [5] D.R. Lloyd, V. Leelavatcharamas, A.N. Emery, M. Al-Rubeai, The role of the cell cycle in determining gene expression and productivity in CHO cells, Cytotechnology 30 (1999) 4957. [6] E. Suzuki, D.F. Ollis, Cell cycle model for antibody production, Biotechnol. Bioeng. 34 (1989) 13981402. [7] E. Suzuki, D.F. Ollis, Enhanced antibody production at slowed growth rates: experimental demonstration and a simple structured model, Biotechnol. Prog. 6 (1990) 231236. [8] F.M. Scott, V.S. de Serrano, F.J. Castellino, Appearance of plasminogen activator activity during a synchronous cycle of a rat adenocarcinoma cell line, PA-III, Exp. Cell Res. 169 (1987) 3946. [9] F.W.F. Lee, C.B. Elias, P. Todd, D.S. Kompala, Engineering Chinese hamster ovary (CHO) cells to achieve an inverse growth-associated production of a foreign protein, -galactosidase, Cytotechnology 28 (1998) 7380. [10] G. Urlaub, L.A. Chasin, Isolation of Chinese hamster cell mutants decient in dihydrofolate reductase activity, Proc. Natl. Acad. Sci. U.S.A. 77 (1980) 42164220.