Optimization of a Roller Bottle Process for the Production of Recombinant Erythropoietin E. I. TSAO, M. A. BOHN, D. R. OMSTEAD, A N D M. J. MUNSTER Department of Bioprocess Development R. W. Johnson Pharmaceutical Research Institute Raritan, New Jersey 08869
INTRODUCTION Erythropoietin (EPO) is a glycoprotein hormone that regulates red blood cell production by controlling both the proliferation and differentiation of progenitor erythroid cells.' This hormone, produced mainly in the kidneys in humans, is accepted as a major physiological regulator of erythroid differentiatiom2 Renal failure in humans often results in anemia with decreased levels of EP0.3 This observation has led to the recombinant expression and subsequent therapeutic use of E P O for patients with chronic renal failure. The EPO gene has been cloned into a number of mammalian cell lines and amplification of the gene has allowed its commercial exploitation as a b i o p h a r m a c e ~ t i c a l .EPO ~ ~ has subsequently been approved for use in many countries worldwide. In addition to use in renal failure, EPO may have many more therapeutic applications where anemia is associated, for example, AIDS, surgery, and chemotherapy. Human EPO has a molecular weight of approximately 30,000 and is composed of 166 amino acids.* Some 30-50% of its mass comprises carbohydrate8 and, although these moieties are not essential for biological activity, they are important in preventing premature removal of circulating levels of the molecule by the liver.' We have studied the production of recombinant human E P O by Chinese Hamster Ovary (CHO) cells in a roller bottle-based system. Although other technologies for the production of secreted mammalian cell proteins are available, this technology has been chosen for the following reasons: (1) a roller bottle process relies on very simple technology; ( 2 ) a roller bottle process is scalable with the use of automated bottle handling equipment; (3) roller bottle technology has been extensively used in the vaccine industry and in the preparation of materials for research and clinical trials;'@I5 (4) contamination of one or more roller bottles does not necessarily result in the failure of an entire batch of product; (5) the product characteristics remain consistent throughout the development phases and are consistent with those produced at the full manufacturing scale; 127
128
ANNALS NEW YORK ACADEMY OF SCIENCES
(6) the possibility of altered product properties for expressed products exists16 when anchorage-dependent cell lines are adapted to grow in suspension as may be required in process scaleup; (7) a shift to serum-free medium in a production phase is easily accomplished with no prior adaptation to that medium being necessary; this is of particular importance to efficient recovery of product; (8) mass transfer efficiency is good;17 (9) direct monitoring of the cells is po~sible.'~
We wished to assess roller bottle technology as a means to produce EPO on a laboratory scale and to further assess its suitability for large-scale manufacturing. In order to do this, we needed to optimize the laboratory-scale process; we present our findings herein. MATERIALS AND METHODS
Cell Line A Chinese Hamster Ovary (CHO) cell line4expressing recombinant human EPO was used in this study. The EPO gene was isolated from a human fetal liver genomic library and cloned into a suitable vector prior to being transformed into the CHO cell line. Methotrexate was used as a selective agent and for amplification of the gene product.
Culture Media DMEM/F12 supplemented with fetal bovine serum (FBS) was used in the development of inoculum and in the roller bottle growth phase. DMEM/F12 without FBS was used in the roller bottle production phases. Production Process The development of the roller bottle inoculum was initiated from a single ampoule. Following growth of the cells to near confluency in a T-flask, the cells were sequentially transferred to a series of spinner flasks of increasing scale until the desired cell number was obtained. Roller bottles were then inoculated and cells were allowed to grow to near confluency (usually 4 days). Spent medium was then removed by decanting and the monolayers were washed with phosphate buffered saline. Serum-free medium was added and the bottles were then incubated further for 7 days. EPO was synthesized and accumulated into the medium during this incubation. This step was repeated a second time in the process, generating two harvests of EPO-containing medium (H1 and H2). EPO Analysis EPO concentrations were determined by an ELISA method (I. Ghobrial, personal communication). In this method, EPO is sandwiched between an anti-EPO
TSAO et a[.: ERYTHROPOIETIN
129
monoclonal antibody and a rabbit anti-EPO polyclonal antibody. Goat anti-rabbit IgG(fc) horseradish peroxidase conjugate is bound to the EPO-antibodies sandwich and a chromogenic substrate solution (ortho-phenylene diamine) is used to develop a color reaction. Purified E P O is used as a standard. The working detection range of the assay is 1 to 20 pg/mL. RESULTS AND DISCUSSION Cell Growth and EPO Production Kinetics
FIGURE 1 illustrates the growth of the CHO cells during the process. In a typical run, population doubling times of the cells are approximately 24 hours in the growth
'
Harvest 2
'
Harvest 1
\
1E07
0
5
Medium-Shif
10
15
20
25
Time (day) FIGURE 1. Growth curve of the CHO-rHuEPO cell line in roller bottles.
phase and 7 days in the production phases. The maximum cell density achieved is approximately 5.0 x 108 cells/bottle. The kinetics of E P O formation are illustrated in FIGURE 2. The EPO titer in H2 phase is consistently higher than the titer in H1 phase for two reasons. First, although the cells are switched to serum-free medium with no prior adaptation, the cells continue to proliferate and reach a cell density at H2 that is two times higher than that at H1. Second, there is a lag in initiation of E P O formation during the first 4 days of H1. Presumably, this lag can b e attributed to the adjustment of the cells to serum-free production medium. No significant lag during the H2 phase is seen following the addition of fresh serum-free medium. The optimal incubation time was determined to be 7 days for each harvest period. Although incubation periods greater than 8 days resulted in higher E P O titer,
ANNALS NEW YORK ACADEMY OF SCIENCES
130
ao C
.-0 4-
0
60 C 0,
0
C 0
40
0
a
w
al > .+
20
0 al w
0
0
2
4
6
a
10
12
Time (day) FIGURE 2. EPO production kinetics in H1 and H2 phases.
significant cell detachment in H2 always occurred. This is undesirable in that it makes subsequent recovery of EPO difficult. Low pH, depletion of essential nutrients (e.g., oxygen, glucose, glutamine, and other amino acids) from the medium, and limited surface area are believed to contribute to this cell detachment. Effects of Incubation Temperature The effect of temperature on EPO production was studied and is shown in FIGURE3. As temperature was increased from 32 to 38 "C, there was a concomitant increase in H1 titer. H2 titer, on the other hand, reached a maximum at 37 "C with no additional increase at 38 "C. In general, incubation at 38 "Cincreased the growth rate (determined microscopically as a reduced time to reach near confluence) of the cell line and resulted in an increased cell density at both the beginning and end of H1. However, due to high cell density (44% higher than that at 37 "C) and rapid medium acidification, severe cell detachment in the latter part of the H2 phase was noted when the cells were incubated at 38 "C. In fact, the viable cell density in the 38 "C bottles at H2 was only 20% of that observed in the 37 "C bottles. As a consequence, the final titer achieved in the H2 phase is approximately the same as that in the H1 phase. Although incubation at 38 "C offers higher total EPO production (when H1 and H2 are combined), the concomitant increase in the amounts of cells and debris detached from the surface presents an offsetting problem in subsequent downstream clarification steps. Incubation at the extremes (32 and 42 "C) of the tested temperature ranges results in drastically reduced EPO titers. In addition, there were no noticeable differences in cell morphology over the 32-37 "C range. Conversely, incubation at 42 "C effected a sharp decline in viability without further cell growth.
TSAO ef al.: ERYTHROPOIETIN
131
Effects of Rotation Rate When rotation rate was varied from 0.2 to 1.0 rpm while incubation temperature was held at 37 "C, the previously described pattern was observed (FIGURE4): namely, E P O accumulation in H2 was twofold to threefold greater than that in H1. Furthermore, E P O titers were consistent throughout the 0.2-1.0 rpm range. However, when the rotation rate was increased further to 2.0 rpm, a different pattern was exhibited. Relative to lower rpm, the E P O titer at 2.0 rpm was significantly increased in H I , whereas the H 2 titer was significantly reduced. These differences are thought to result from improved mixing and aeration in H1 and from subsequent reduced nutrient availability in H2, respectively. In turn, these factors lead to accelerated growth and E P O production in H I and to cell detachment in H2. Overall, we found that, at 2.0 rpm, the specific cellular productivity was 40% lower in the H2 phase, presumably due to the stress of high rotation speed. In addition, it should be noted that no adverse effects on cell growth or E P O production were observed during a 90-minute cessation in the rotation during either growth or production phases.
Effects of Carbon Dioxide Gassing
To control the p H of the bicarbonate-buffered medium, the roller bottles were gassed with 10% C02/balance air at each process step. The extent of this gassing had a significant effect on both the cell density and E P O titer achieved (FIGURES5a and 5b). In the growth phase (FIGURE5a), longer gassing time resulted in higher cell density. This effect may be pH-mediated as the medium p H values of the bottles gassed for 0 and 120 seconds were 7.27 and 6.86 at the end of the growth phase,
37OC'
36OC U
0
20
0
a W
15
a,
.-+>
10
a, CE
5
0 -
0
FIGURE 3. Effect of incubation temperature on EPO production.
ANNALS NEW YORK ACADEMY OF SCIENCES
132
40
c 35 0
.+
1
0.2 RPM
0.5 RPM 1 .o RPM'
zc
*
+ 30 al 0 S
Control
2.0 RPM
25
0 0 20
0
n W
15
al
> .+
10
CE
5
0 al
0 1 2 3 4 FIGURE4. Effect of rotation speed on EPO production.
respectively. In the production phases, longer gassing (and consequently lower medium pH) correlated with higher EPO yield in the pH range of 6.6 to 7.2 (FIGURE 5b). These higher titers resulted from higher cell densities rather than from higher specific cell productivities. Effects of Inoculum Density
The effect of initial cell density in the roller bottle growth phase on cell growth and productivity was also examined. The results are presented in TABLE 1. Roller bottles inoculated with high initial cell densities showed a rapid decline of medium pH and dissolved oxygen level. It was also observed that, in general, higher inoculum levels resulted in higher H1 and H2 titers. However, at the highest inoculum density investigated (2.4 x lo7 cells/bottle), cell detachment became a problem in the later production phase due to low medium pH and depletion of nutrients. In this case, medium pH was very low (6.45) at H2 and the dissolved oxygen level was close to zero (data not shown). Furthermore, higher inoculum densities also increased the amounts of suspended cells and debris. Effects of Inoculum Generation Number
In some bacterial processes, it has been observed that the physiological state of the inoculum is critical to the overall process yield.I8We examined whether, for these EPO-producing cells, their performance in roller bottles might be directly affected by the physiological state of the cells produced during an extended inoculum development. We also examined the genetic stability of cultures by monitoring their growth
TSAO et al.: ERYTHROPOIETIN
2*5E08
133
3
0
30
15
60
120
Gassing Time (sec) FIGURE 5. Effect of gassing on (a) cell density and (b) EPO production.
TABLE1. Effect of Initial Cell Density on Cell Growth and EPO Production
Cell density at mediumshift (cells/RB) Cell density at H2 (cells/RB) Relative EPO titer at H1 (I%!mL) Relative EPO titer at H2 (w/mL) "RB = roller bottle.
0.8 x 107
Initial Cell Density (cells/RB)" 1.2 x 107 1.6 x 107 2.0 x 107
2.4 x 107
9.5 x lo7
1.6 x lo8
3.4 x loR
3.8 x lo*
5.2 x loH 6.5 x lo8
2.6 x lo8
2.9 x lo8
7.1 x 10* cellsdetached
36.8
35.4
48.6
51.6
73.4
50.0
82.4
83.4
> 100
> 100
ANNALS NEW YOEW ACADEMY OF SCIENCES
rn U Media-Shift
2
0
6
4
8
10
12
14
18
16
20
lnoculum Generation Number FIGURE 6. Effect of inoculum generation number on cell density.
characteristics and EPO-producing potential throughout the course of inoculum development. FIGURES 6 and 7 show the effects of the inoculum generation number on cell densities and EPO titer. In both cases, cell density and EPO titer were found to be independent of generation number in the range we examined. d"
s
45
1
2
40
I
.-0 +
c
fz 35 1 w 0
c 30 0
7
0
2
4
6
8
10
12
14
16
18
lnoculum Generation Number FlGURE 7. Effect of inoculum generation number on EPO production.
20
TSAO et al.: ERYTHROPOIETIN
135
CONCLUSIONS We have characterized the effects of various process parameters, including growth and production kinetics, incubation temperature, rotation rate, C 0 2 gassing, inoculum density, and inoculum generation number, on both the growth of the recombinant CHO cells and the production of EPO in roller bottles. These studies have allowed us to optimize a laboratory-scale roller bottle process. Our approach has been to examine a range of possible set points for each process variable and then to select the optimal conditions. These optimal conditions and ranges are not always those that resulted in the highest titer because process reliability and reproducibility are more important in practice. In many of the studies discussed, high titer was correlated with extreme detachment of the monolayers. In practice, the resulting debris severely hinders clarification of the EPO-containing medium prior to concentration. This ultimately results in a disproportionate decrease in the overall purification yield; this is undesirable. Nevertheless, the knowledge of the sensitivity of each process parameter has proved to be immensely helpful in dealing with various process disturbances.
ACKNOWLEDGMENTS We thank Kelly Frame and Gregory Price for providing cells used in the inoculum generation number experiments and John Gardner for his editorial comments.
REFERENCES
1. SPIVAK, J. L. 1989. Blood Rev. 3: 130-135. S. B. & L. 0 . JACOBSON. 1970. Erythropoietin and the Regulation of Erythropoie2. KRANTZ, sis. University of Chicago Press. Chicago. A. J., J. CARO,0 . MILLER& R. SILVER. 1980. Ann. Clin. Lab. Sci. L O 250-257. 3. ERSLEV, R. SMALLING, J. C. EGRIE,K. K. CHEN, 4. LIN,F. K., S. SUGGS,C. H. LIN, J. K. BROWNE, G. M. Fox, F. MARTIN, Z. STABINSKY, S. M. BRADRAWI, P. H. LAI& E. GOLDWASSER. 1985. Proc. Natl. Acad. Sci. U.S.A. 82: 7580-7584. J. S., K. L. BERKNER, R. V. LEBO& J. W. ADAMSON. 1986. Proc. Natl. Acad. Sci. 5. POWELL, U.S.A. 83: 6465-6469. INC.1985. Patent W 0 85/02610. 6. KIRIN-AMGEN, INC.1987. Patent US 4,667,016. 7. KIRIN-AMGEN, T. W. STRICKLAND & D. A. YPHANTIS. 1987. Biochemistry 8. DAVIS,J. M., T. ARAKAWA, 2 6 2633-2638. W. A. & R. H. PAINTER.1972. Can. J. Biochem. 5 0 909-917. 9. LUKOWSKY, G. F. 1985. Animal Cell Biotechnology. Vol. 1. R. E. Spier & J. B. Griffiths, Eds.: 10. PANINA, 21 1-242. Academic Press. New York/London. M. P., J. A. GEORGIADES, G. J. STANTON, F. DIANZANI & H. M. JOHNSON. 11. LANGFORD, 1979. Infect. Immun. 26(1): 3641. 12. LIEBHABER, H., T. PAJOT& J. T. RIORAN. 1969. Proc. SOC.Exp. Biol. Med. 130(1): 12-14. W., V. JAEGER,D. HERBST,H. HAUSER& J. HOPPE.1989. Eur. J. Biochem. 13. EICHNER, 185(1): 135-140. 1986. J. Immunol. 14. MUUL,L. M., E. P. DIRECTOR,C. L. HYATT& S. A. ROSENBERG. Methods 88(2): 265-275. 15. LEE,L. F. 1971. Avian Dis. 15(3): 565-571. J. B. & R. E. SPIER.1980. Arch. Virol. 63: 1-9. 16. CLARKE,
136
ANNALS NEW YORK ACADEMY OF SCIENCES
17. GRIFFITHS, J. B. 1988. Animal Cell Biotechnology. Vol. 3. R. E. Spier & J. B. Griffiths, Eds.: 179-220. Academic Press. New York/London. 18. HUNT, G. R. & R. W. STIEBER. 1986. Manual of Industrial Microbiology and Biotechnology. A. L. Demain & N. A. Solomon, Eds.: 3240. Amer. SOC.Microbiol. Washington, District of Columbia.
INTRODUCTION Erythropoietin (EPO) is a glycoprotein hormone that regulates red blood cell production by controlling both the proliferation and differentiation of progenitor erythroid cells.' This hormone, produced mainly in the kidneys in humans, is accepted as a major physiological regulator of erythroid differentiatiom2 Renal failure in humans often results in anemia with decreased levels of EP0.3 This observation has led to the recombinant expression and subsequent therapeutic use of E P O for patients with chronic renal failure. The EPO gene has been cloned into a number of mammalian cell lines and amplification of the gene has allowed its commercial exploitation as a b i o p h a r m a c e ~ t i c a l .EPO ~ ~ has subsequently been approved for use in many countries worldwide. In addition to use in renal failure, EPO may have many more therapeutic applications where anemia is associated, for example, AIDS, surgery, and chemotherapy. Human EPO has a molecular weight of approximately 30,000 and is composed of 166 amino acids.* Some 30-50% of its mass comprises carbohydrate8 and, although these moieties are not essential for biological activity, they are important in preventing premature removal of circulating levels of the molecule by the liver.' We have studied the production of recombinant human E P O by Chinese Hamster Ovary (CHO) cells in a roller bottle-based system. Although other technologies for the production of secreted mammalian cell proteins are available, this technology has been chosen for the following reasons: (1) a roller bottle process relies on very simple technology; ( 2 ) a roller bottle process is scalable with the use of automated bottle handling equipment; (3) roller bottle technology has been extensively used in the vaccine industry and in the preparation of materials for research and clinical trials;'@I5 (4) contamination of one or more roller bottles does not necessarily result in the failure of an entire batch of product; (5) the product characteristics remain consistent throughout the development phases and are consistent with those produced at the full manufacturing scale; 127
128
ANNALS NEW YORK ACADEMY OF SCIENCES
(6) the possibility of altered product properties for expressed products exists16 when anchorage-dependent cell lines are adapted to grow in suspension as may be required in process scaleup; (7) a shift to serum-free medium in a production phase is easily accomplished with no prior adaptation to that medium being necessary; this is of particular importance to efficient recovery of product; (8) mass transfer efficiency is good;17 (9) direct monitoring of the cells is po~sible.'~
We wished to assess roller bottle technology as a means to produce EPO on a laboratory scale and to further assess its suitability for large-scale manufacturing. In order to do this, we needed to optimize the laboratory-scale process; we present our findings herein. MATERIALS AND METHODS
Cell Line A Chinese Hamster Ovary (CHO) cell line4expressing recombinant human EPO was used in this study. The EPO gene was isolated from a human fetal liver genomic library and cloned into a suitable vector prior to being transformed into the CHO cell line. Methotrexate was used as a selective agent and for amplification of the gene product.
Culture Media DMEM/F12 supplemented with fetal bovine serum (FBS) was used in the development of inoculum and in the roller bottle growth phase. DMEM/F12 without FBS was used in the roller bottle production phases. Production Process The development of the roller bottle inoculum was initiated from a single ampoule. Following growth of the cells to near confluency in a T-flask, the cells were sequentially transferred to a series of spinner flasks of increasing scale until the desired cell number was obtained. Roller bottles were then inoculated and cells were allowed to grow to near confluency (usually 4 days). Spent medium was then removed by decanting and the monolayers were washed with phosphate buffered saline. Serum-free medium was added and the bottles were then incubated further for 7 days. EPO was synthesized and accumulated into the medium during this incubation. This step was repeated a second time in the process, generating two harvests of EPO-containing medium (H1 and H2). EPO Analysis EPO concentrations were determined by an ELISA method (I. Ghobrial, personal communication). In this method, EPO is sandwiched between an anti-EPO
TSAO et a[.: ERYTHROPOIETIN
129
monoclonal antibody and a rabbit anti-EPO polyclonal antibody. Goat anti-rabbit IgG(fc) horseradish peroxidase conjugate is bound to the EPO-antibodies sandwich and a chromogenic substrate solution (ortho-phenylene diamine) is used to develop a color reaction. Purified E P O is used as a standard. The working detection range of the assay is 1 to 20 pg/mL. RESULTS AND DISCUSSION Cell Growth and EPO Production Kinetics
FIGURE 1 illustrates the growth of the CHO cells during the process. In a typical run, population doubling times of the cells are approximately 24 hours in the growth
'
Harvest 2
'
Harvest 1
\
1E07
0
5
Medium-Shif
10
15
20
25
Time (day) FIGURE 1. Growth curve of the CHO-rHuEPO cell line in roller bottles.
phase and 7 days in the production phases. The maximum cell density achieved is approximately 5.0 x 108 cells/bottle. The kinetics of E P O formation are illustrated in FIGURE 2. The EPO titer in H2 phase is consistently higher than the titer in H1 phase for two reasons. First, although the cells are switched to serum-free medium with no prior adaptation, the cells continue to proliferate and reach a cell density at H2 that is two times higher than that at H1. Second, there is a lag in initiation of E P O formation during the first 4 days of H1. Presumably, this lag can b e attributed to the adjustment of the cells to serum-free production medium. No significant lag during the H2 phase is seen following the addition of fresh serum-free medium. The optimal incubation time was determined to be 7 days for each harvest period. Although incubation periods greater than 8 days resulted in higher E P O titer,
ANNALS NEW YORK ACADEMY OF SCIENCES
130
ao C
.-0 4-
0
60 C 0,
0
C 0
40
0
a
w
al > .+
20
0 al w
0
0
2
4
6
a
10
12
Time (day) FIGURE 2. EPO production kinetics in H1 and H2 phases.
significant cell detachment in H2 always occurred. This is undesirable in that it makes subsequent recovery of EPO difficult. Low pH, depletion of essential nutrients (e.g., oxygen, glucose, glutamine, and other amino acids) from the medium, and limited surface area are believed to contribute to this cell detachment. Effects of Incubation Temperature The effect of temperature on EPO production was studied and is shown in FIGURE3. As temperature was increased from 32 to 38 "C, there was a concomitant increase in H1 titer. H2 titer, on the other hand, reached a maximum at 37 "C with no additional increase at 38 "C. In general, incubation at 38 "Cincreased the growth rate (determined microscopically as a reduced time to reach near confluence) of the cell line and resulted in an increased cell density at both the beginning and end of H1. However, due to high cell density (44% higher than that at 37 "C) and rapid medium acidification, severe cell detachment in the latter part of the H2 phase was noted when the cells were incubated at 38 "C. In fact, the viable cell density in the 38 "C bottles at H2 was only 20% of that observed in the 37 "C bottles. As a consequence, the final titer achieved in the H2 phase is approximately the same as that in the H1 phase. Although incubation at 38 "C offers higher total EPO production (when H1 and H2 are combined), the concomitant increase in the amounts of cells and debris detached from the surface presents an offsetting problem in subsequent downstream clarification steps. Incubation at the extremes (32 and 42 "C) of the tested temperature ranges results in drastically reduced EPO titers. In addition, there were no noticeable differences in cell morphology over the 32-37 "C range. Conversely, incubation at 42 "C effected a sharp decline in viability without further cell growth.
TSAO ef al.: ERYTHROPOIETIN
131
Effects of Rotation Rate When rotation rate was varied from 0.2 to 1.0 rpm while incubation temperature was held at 37 "C, the previously described pattern was observed (FIGURE4): namely, E P O accumulation in H2 was twofold to threefold greater than that in H1. Furthermore, E P O titers were consistent throughout the 0.2-1.0 rpm range. However, when the rotation rate was increased further to 2.0 rpm, a different pattern was exhibited. Relative to lower rpm, the E P O titer at 2.0 rpm was significantly increased in H I , whereas the H 2 titer was significantly reduced. These differences are thought to result from improved mixing and aeration in H1 and from subsequent reduced nutrient availability in H2, respectively. In turn, these factors lead to accelerated growth and E P O production in H I and to cell detachment in H2. Overall, we found that, at 2.0 rpm, the specific cellular productivity was 40% lower in the H2 phase, presumably due to the stress of high rotation speed. In addition, it should be noted that no adverse effects on cell growth or E P O production were observed during a 90-minute cessation in the rotation during either growth or production phases.
Effects of Carbon Dioxide Gassing
To control the p H of the bicarbonate-buffered medium, the roller bottles were gassed with 10% C02/balance air at each process step. The extent of this gassing had a significant effect on both the cell density and E P O titer achieved (FIGURES5a and 5b). In the growth phase (FIGURE5a), longer gassing time resulted in higher cell density. This effect may be pH-mediated as the medium p H values of the bottles gassed for 0 and 120 seconds were 7.27 and 6.86 at the end of the growth phase,
37OC'
36OC U
0
20
0
a W
15
a,
.-+>
10
a, CE
5
0 -
0
FIGURE 3. Effect of incubation temperature on EPO production.
ANNALS NEW YORK ACADEMY OF SCIENCES
132
40
c 35 0
.+
1
0.2 RPM
0.5 RPM 1 .o RPM'
zc
*
+ 30 al 0 S
Control
2.0 RPM
25
0 0 20
0
n W
15
al
> .+
10
CE
5
0 al
0 1 2 3 4 FIGURE4. Effect of rotation speed on EPO production.
respectively. In the production phases, longer gassing (and consequently lower medium pH) correlated with higher EPO yield in the pH range of 6.6 to 7.2 (FIGURE 5b). These higher titers resulted from higher cell densities rather than from higher specific cell productivities. Effects of Inoculum Density
The effect of initial cell density in the roller bottle growth phase on cell growth and productivity was also examined. The results are presented in TABLE 1. Roller bottles inoculated with high initial cell densities showed a rapid decline of medium pH and dissolved oxygen level. It was also observed that, in general, higher inoculum levels resulted in higher H1 and H2 titers. However, at the highest inoculum density investigated (2.4 x lo7 cells/bottle), cell detachment became a problem in the later production phase due to low medium pH and depletion of nutrients. In this case, medium pH was very low (6.45) at H2 and the dissolved oxygen level was close to zero (data not shown). Furthermore, higher inoculum densities also increased the amounts of suspended cells and debris. Effects of Inoculum Generation Number
In some bacterial processes, it has been observed that the physiological state of the inoculum is critical to the overall process yield.I8We examined whether, for these EPO-producing cells, their performance in roller bottles might be directly affected by the physiological state of the cells produced during an extended inoculum development. We also examined the genetic stability of cultures by monitoring their growth
TSAO et al.: ERYTHROPOIETIN
2*5E08
133
3
0
30
15
60
120
Gassing Time (sec) FIGURE 5. Effect of gassing on (a) cell density and (b) EPO production.
TABLE1. Effect of Initial Cell Density on Cell Growth and EPO Production
Cell density at mediumshift (cells/RB) Cell density at H2 (cells/RB) Relative EPO titer at H1 (I%!mL) Relative EPO titer at H2 (w/mL) "RB = roller bottle.
0.8 x 107
Initial Cell Density (cells/RB)" 1.2 x 107 1.6 x 107 2.0 x 107
2.4 x 107
9.5 x lo7
1.6 x lo8
3.4 x loR
3.8 x lo*
5.2 x loH 6.5 x lo8
2.6 x lo8
2.9 x lo8
7.1 x 10* cellsdetached
36.8
35.4
48.6
51.6
73.4
50.0
82.4
83.4
> 100
> 100
ANNALS NEW YOEW ACADEMY OF SCIENCES
rn U Media-Shift
2
0
6
4
8
10
12
14
18
16
20
lnoculum Generation Number FIGURE 6. Effect of inoculum generation number on cell density.
characteristics and EPO-producing potential throughout the course of inoculum development. FIGURES 6 and 7 show the effects of the inoculum generation number on cell densities and EPO titer. In both cases, cell density and EPO titer were found to be independent of generation number in the range we examined. d"
s
45
1
2
40
I
.-0 +
c
fz 35 1 w 0
c 30 0
7
0
2
4
6
8
10
12
14
16
18
lnoculum Generation Number FlGURE 7. Effect of inoculum generation number on EPO production.
20
TSAO et al.: ERYTHROPOIETIN
135
CONCLUSIONS We have characterized the effects of various process parameters, including growth and production kinetics, incubation temperature, rotation rate, C 0 2 gassing, inoculum density, and inoculum generation number, on both the growth of the recombinant CHO cells and the production of EPO in roller bottles. These studies have allowed us to optimize a laboratory-scale roller bottle process. Our approach has been to examine a range of possible set points for each process variable and then to select the optimal conditions. These optimal conditions and ranges are not always those that resulted in the highest titer because process reliability and reproducibility are more important in practice. In many of the studies discussed, high titer was correlated with extreme detachment of the monolayers. In practice, the resulting debris severely hinders clarification of the EPO-containing medium prior to concentration. This ultimately results in a disproportionate decrease in the overall purification yield; this is undesirable. Nevertheless, the knowledge of the sensitivity of each process parameter has proved to be immensely helpful in dealing with various process disturbances.
ACKNOWLEDGMENTS We thank Kelly Frame and Gregory Price for providing cells used in the inoculum generation number experiments and John Gardner for his editorial comments.
REFERENCES
1. SPIVAK, J. L. 1989. Blood Rev. 3: 130-135. S. B. & L. 0 . JACOBSON. 1970. Erythropoietin and the Regulation of Erythropoie2. KRANTZ, sis. University of Chicago Press. Chicago. A. J., J. CARO,0 . MILLER& R. SILVER. 1980. Ann. Clin. Lab. Sci. L O 250-257. 3. ERSLEV, R. SMALLING, J. C. EGRIE,K. K. CHEN, 4. LIN,F. K., S. SUGGS,C. H. LIN, J. K. BROWNE, G. M. Fox, F. MARTIN, Z. STABINSKY, S. M. BRADRAWI, P. H. LAI& E. GOLDWASSER. 1985. Proc. Natl. Acad. Sci. U.S.A. 82: 7580-7584. J. S., K. L. BERKNER, R. V. LEBO& J. W. ADAMSON. 1986. Proc. Natl. Acad. Sci. 5. POWELL, U.S.A. 83: 6465-6469. INC.1985. Patent W 0 85/02610. 6. KIRIN-AMGEN, INC.1987. Patent US 4,667,016. 7. KIRIN-AMGEN, T. W. STRICKLAND & D. A. YPHANTIS. 1987. Biochemistry 8. DAVIS,J. M., T. ARAKAWA, 2 6 2633-2638. W. A. & R. H. PAINTER.1972. Can. J. Biochem. 5 0 909-917. 9. LUKOWSKY, G. F. 1985. Animal Cell Biotechnology. Vol. 1. R. E. Spier & J. B. Griffiths, Eds.: 10. PANINA, 21 1-242. Academic Press. New York/London. M. P., J. A. GEORGIADES, G. J. STANTON, F. DIANZANI & H. M. JOHNSON. 11. LANGFORD, 1979. Infect. Immun. 26(1): 3641. 12. LIEBHABER, H., T. PAJOT& J. T. RIORAN. 1969. Proc. SOC.Exp. Biol. Med. 130(1): 12-14. W., V. JAEGER,D. HERBST,H. HAUSER& J. HOPPE.1989. Eur. J. Biochem. 13. EICHNER, 185(1): 135-140. 1986. J. Immunol. 14. MUUL,L. M., E. P. DIRECTOR,C. L. HYATT& S. A. ROSENBERG. Methods 88(2): 265-275. 15. LEE,L. F. 1971. Avian Dis. 15(3): 565-571. J. B. & R. E. SPIER.1980. Arch. Virol. 63: 1-9. 16. CLARKE,
136
ANNALS NEW YORK ACADEMY OF SCIENCES
17. GRIFFITHS, J. B. 1988. Animal Cell Biotechnology. Vol. 3. R. E. Spier & J. B. Griffiths, Eds.: 179-220. Academic Press. New York/London. 18. HUNT, G. R. & R. W. STIEBER. 1986. Manual of Industrial Microbiology and Biotechnology. A. L. Demain & N. A. Solomon, Eds.: 3240. Amer. SOC.Microbiol. Washington, District of Columbia.
