A High-Yielding, Generic Fed-Batch Cell Culture Process for Production of Recombinant Antibodies
Paul W. Sauer,1 John E. Burky,1 Mark C. Wesson,1 Heather D. Sternard,1 Limin Qu2
1
Cell Culture Development, Protein Design Labs, 34801 Campus Drive, Fremont, California 94555; telephone: 510-574-1549; fax: 510-574-1500; e-mail: 2 Process Engineering, Protein Design Labs, 3955 Annapolis Lane, Plymouth, Minnesota 55447
Received 12 January 1999; accepted 24 September 1999
Abstract: A fed-batch cell culture process was developed that has general applicability to all evaluated Sp2/0 (n = 8) and NS0 (n = 1) antibody-producing cell lines. The two key elements of this generic process were a protein-free concentrated feed medium, and a robust, metabolically responsive feeding strategy bad on the off-line measurement of gluco. The fed-batch process was shown to perform equivalently at the 15 L development scale and 750 L manufacturing scale. Compared to batch cultures, the fed-batch process yielded a 4.3 fold increa in the average integral of viable cell concentration and a 1.7 fold increa in average specific antibody production rate, equi
valent to a 7.6 fold increa in average final antibody concentration. The highest producing cell line reached a peak viable cell concentration of 1.0 × 107 cell mL−1 and a final antibody concentration of 750 mg L−1 in a 10 day process. For all lines evaluated, reducing bioreactor pH t point from 7.2 to 7.0 resulted in an additional 2.4 fold increa in average final antibody concentration. The optimized fed-batch process consistently yielded a volumetric productivity exceeding 50 mg L−1 day−1. This generic, high-yielding fed-batch process significantly decread development time, and incread manufacturing efficiency, thereby facilitating the clinical evaluation of numerous recombinant antibodies. © 2000 John Wiley &
Sons, Inc. Biotechnol Bioeng 67: 585–597, 2000.
Keywords: generic fed-batch process; concentrated protein-free feed medium; cell culture; gluco metabolism; pH; humanized antibody
INTRODUCTION The u of monoclonal antibodies as therapeutics holds great promi for treating many human dias, including autoimmune and inflammatory conditions, cancers, and viral dias. However, murine monoclonal antibodies are verely restricted as therapeutics due to their immunogenicity and short half-life in humans. Recombinant antibodies, such as chimeric and humanized antibodies, have supe-
Correspondence to: Paul W. Sauer
rior immunogenic and pharmacokinetic properties relative to their murine counterparts. Hence, there has been a steady ri in the number of recombinant antibodies undergoing clinical evaluation over the last decade, with veral gaining regulatory approval. Fed-batch culture is a frequently ud method for the industrial manufacture of cell culture bad recombinant therapeutics. A common objective of many fed-batch strategies is to maximize final product concentration, which is the mathematical product of the specific production rate and the integral of viable cell concentration. It follows that increas in either of the variables will increa final product concentration. Extensive reviews of strategies aimed at increasing the integral of viable cell concentration and the specific production rate have been published (Bibila and Robinson, 1995; Xie and Wang, 1997). The integral of viable cell concentration can be incread by extending culture duration and/or by increasing peak viable cell concentration. Many fed-batch strategies increa the integral of viable cell concentration by utilizing feed solutions bad on total or partial concentrates of basal medium to replenish consumed nutrients (Bibila et al., 1994; Fike et al., 1993; Luan et al., 1987; Zhou et al., 1995). Problems often associated with incread cell concentration or incread culture duration are elevated levels of ammonia and/or lactate, both of which have been shown to be toxic to veral ma
mmalian cell lines (Chang et al., 1995; Hasll et al., 1991). Specific strategies propod to address this problem of metabolite accumulation in batch and fed-batch cell culture process include: maintaining low residual gluco and/or glutamine concentrations (Glacken et al., 1986; Ljunggren and Haggstrom, 1994; Xie and Wang, 1996; Zhou et al., 1997b); removing low molecular weight components using dialysis membranes (Adamson et al., 1983; Portner et al., 1996); lecting cell lines that can grow in glutamine-free medium (Birch et al., 1994; Griffiths, 1973); utilizing the glutamine syntheta expression system (Beb-
© 2000 John Wiley & Sons, Inc.
bington et al., 1992; Brown et al., 1992); and replacing glutamine with glutamine-bad dipeptides (Christie and Butler, 1994). Significant effort has also been expended trying to increa the specific production rate of recombinant mammalian cells. The maximum specific antibody production rate in hybridoma cells has been reported to be approximately 80 pg cell−1 day−1 (Savinell et al., 1989). Relative to this propod maximum, high rates of specific productivity in recombinant cells have been achieved by linking heavy- and light-chain cDNA to different amplifiable and lectable markers (Wood et al., 1990), and by performing quential transfections using different lection markers (Four et al., 1992). Our fed-batch strategy was shaped by factors common to industrial biotechnol
ogy. Clinical efficacy has historically been unpredictable, resulting in a low percentage of drug candidates reaching the market. It is of strategic importance, therefore, that a steady stream of drug candidates be supplied to the clinic, necessitating a compresd product development cycle. The substantial variability obrved between recombinant cell lines often translates, however, into a time and/or labor intensive development effort. Our objective, therefore, was to rapidly develop a robust and scalable fed-batch process, that would significantly increa volumetric productivity relative to batch process, and would be applicable to all antibody producing cell lines.
chain expression plasmids were constructed by inrting DNA encoding the VH region into the XbaI site of pVg1, pVg2, or pVg4 depending on the desired isotype. The lightand heavy-chain plasmids (20 g each) were linearized by an appropriate restriction enzyme and co-transfected into 1 × 107 Sp2/0 cells by electroporation (Bio-Rad, Richmond, CA; GenePulr). Transfectants were grown in 96-well plates in DMEM with 10% FBS for 2 days and then placed under lection for gpt expression using 1 g mL−1 mycophenolic acid. NS0 cells followed the same protocol except for the u of a single expression plasmid containing both light and heavy chain. After transfection, procedures ud to lect production cell lines varied widely. Variables included (1) number of subclonings, (2) type of media ud during subcloning, (3) prence or abnce of rum during subcloning, (4) u of ampli
fication markers, and (5) method ud to quantify cellline productivity. After lecting a production cell line, a master cell bank (MCB) and a working cell bank (WCB) were made. Fedbatch data for all cell lines were generated using cells expanded from a WCB vial, except for cell line F data, which were generated using cells from an MCB vial. All cell banks were confirmed to be free of mycoplasma. Basal Medium The BM-1 basal medium was a fully defined rum-free medium containing only one protein (1 mg L−1 transferrin). BM-1 was a 1:1 mixture of Iscove’s Modification of Dulbecco’s Medium (IMDM) and Ham’s F-12, supplemented with mangane sulfate, sodium metasilicate, ammonium molybdate, ammonium metavanadate, nickel chloride, stannous chloride, ethanolamine, -mercaptoethanol and pluronic F-68. The BM-2 basal medium was a powder formulation of Gibco Hybridoma SFM (Life Technologies, Bethesda, MD; Cat. #12045-076) without phenol red. BM-2 was a fully defined rum-free medium containing two proteins (10 mg L−1 transferrin and 10 mg L−1 insulin). The BM-3 basal medium was identical to BM-2 except that the 10 mg L−1 transferrin was replaced by 5 mg L−1 sodium iron(III) EDTA (Baker, Phillipsburg, NJ; L699-07). Feed Medium The FM-1 feed medium was a partial concentrate of the BM-1 basal medium, developed in a manner similar to that described by Bibila et al. (1994). FM-1 did not contain
MATERIALS AND METHODS Cell-Line Information Data from nine recombinant cell lines (designate
d A–I) are prented in this article (Table I). Of the nine, one was an NS0-derived cell line (C), and the remaining eight were Sp2/0-derived cell lines. Sp2/0-Ag14 cells (Shulman et al., 1978) were obtained from the American Type Culture Collection (CRL-1581). NS0 cells (Galfre and Milstein, 1981) were obtained from the European Collection of Animal Cell Cultures (#8511050). Each recombinant cell line produced either a human or humanized antibody of the IgG1, IgG2, or IgG4 isotype. Prior to transfection, Sp2/0 cells were cultured in Dulbecco’s Modification of Eagle’s Medium (DMEM) with 10% FBS at 37°C in the prence of 7.5% CO2. Light-chain expression plasmids were constructed by inrting DNA encoding the Vk region into the XbaI site of pVk. Heavy-
Table I.
Cell line summary. A Sp2/0 Human IgG1 B Sp2/0 Humanized IgG1 C NS0 Humanized IgG1 D Sp2/0 Humanized IgG2 E Sp2/0 Humanized IgG4 F Sp2/0 Humanized IgG4 G Sp2/0 Humanized IgG1 H Sp2/0 Humanized IgG2 I Sp2/0 Human IgG1
Cell line Cell line derivation Antibody type Antibody isotype
586
BIOTECHNOLOGY AND BIOENGINEERING, VOL. 67, NO. 5, MARCH 5, 2000
sodium chloride, sodium bicarbonate, sodium phosphate monobasic, HEPES, or alanine. Most of the FM-1 components were formulated at 10× the concentration found in BM-1. Amino acid concentrations were stoichiometrically balanced, in a manner similar to that communicated by Zhou et al. (1997b). Gluco concentration was 29 g L−1. FM-1 was supplied by Life Technologies as a liquid concentrate (Jayme et al., 1993, 1996). Osmolality of FM-1 was 500 ± 20 mOsm kg−1. The FM-2 feed medium was identical to FM-1 except that 10 mg L−1 transferrin was replaced by 10 mg L−1 sodium iron(III) EDTA. FM-2 was supplied by Life Technologies in powdered form. Osmolality of FM-2 was 370 ± 20 mOsm kg−1. The difference in osmolality between FM-1 and FM-2 reflects differences in the amounts of NaOH and HCl ud to prepare the respective media. Bioreactor Equipment The cell culture development system consisted of four identically configured 3 L glass bioreactors (Applikon, Foster City, CA) and two identically configured 15 L stainless steel bioreactors (B. Braun Biotech, Allentown, PA; Biostat EC). The 3 L and 15 L bioreactors had 2 L and 10 L working volumes, respectively. Process control t points for both the 3 L and 15 L bioreactors were maintained using a Digital Control Unit (B. Braun Biotech). Each 3 L bioreactor was configured with a standard 60 mm diameter marine impeller; each 15 L bioreactor was configured with a custom 120 mm diameter, 45° pitch-blade impeller. The pH was measured using a gel-filled electrode (Mettler-Toledo, Wilmington, MA; DPAS); the dissolved oxygen (DO) concentration was measured with a pola
rographic electrode (Mettler-Toledo). The pH was adjusted using CO2 gas and 1M Na2CO3. Each bioreactor was configured with two mass-flow controllers (Sierra Instruments, Monterey, CA), one for air and one for oxygen. The gas flow rate was held constant while the ratio of air and oxygen was adjusted to maintain the DO t point. The controller t points were: pH, 7.20 (except as indicated); DO, 30% air saturation; temperature, 37°C; agitation, 200 rpm (3 L bioreactor), and 100 rpm (15 L bioreactor); gas-flow rate, 100 sccm (3 L bioreactor), and 200 sccm (15 L bioreactor). Inoculum Preparation A frozen vial was thawed and subquently expanded for 1 to 2 weeks using a quence of T-flasks (Falcon, Franklin Lake, NJ), roller bottles (Falcon) and spinner flasks (Bellco Biotech, Vineland, NJ) prior to inoculating the bioreactor. Cells were maintained in the exponential growth pha throughout the expansion process. Cells were cultured at 37°C in a 7.5% CO2 atmosphere using a reach-in incubator (Forma Scientific, Marietta, OH). Fed-Batch Bioreactor Protocol Basal medium was filtered into the bioreactors using a 0.2 m-rated membrane (Millipore Corp., Bedford, MA; Mil-
lipak-20). The 3 L and 15 L bioreactors were batched with 1.0 L and 4.0 L of basal medium, respectively. Feed medium was filtered into sterile glass bottles using a 0.2 mrated filter (Millipore; Millipak-20, Bedford, MA). The volume of feed medium for the 3 L and 15 L bioreactors was 1.0 L an
d 5.0 L, respectively. Feed medium was placed on a balance (Sartorious, Gottingen, Germany; IC64), stored at ¨ 4–8 °C, and aptically connected to its respective bioreactor. The post-inoculation viable cell concentration was 0.2–0.3 × 109 cell L−1. Inoculum volume for the 3 L and 15 L bioreactors was 0.20 L and 1.0 L, respectively. For the initial 2 days of culture, bioreactor sampling was performed once daily; each day thereafter, sampling occurred twice daily with an approximate sampling interval of 8 hours. Analytical methods were performed immediately after bioreactor sampling. The initial addition of feed medium was performed 2 days after inoculation. The feeding protocol was bad on maintaining a desired post-feed gluco concentration. Specifically, the volume of feed medium (Vfeed) to be added was calculated using Equation (1), where glctar was the post-feed target gluco concentration, glc was the culture’s gluco concentration prior to feeding, glcfeed was the feed medium’s gluco concentration, and V was the culture volume prior to addition. Post-feed target gluco concentrations were typically incread over the cour of a culture. Initial target values ranged from 3.5 to 4.5 g L−1; final target values ranged from 4.5 to 6.5 g L−1 (e.g., e Fig. 3a). Vfeed = glctar − glc glcfeed − glctar V (1)
After the bioreactor was sampled and the gluco assay performed, the volume of feed medium to be added (if required) was calculated. Addition of feed medium was initiated within 1 hour after biore
actor sampling. Feed medium was pumped into the bioreactor using a simple control loop. The control loop turned on a peristaltic pump (Watson Marlow, Wilmington, MA; 101UR) until the feed medium balance weight equaled the ur-defined t point. The elapd time of feed medium addition was less than 30 minutes. By the end of the culture the entire volume of feed medium had been added, unless otherwi stated. The culture was harvested when either the viability of the culture dropped below 20%, or the viable cell concentration dropped below 2.0 × 109 cell L−1. Analytical Methods The pH, pCO2, and pO2 were measured using a blood-gas analyzer (Radiometer; ABL-5). If the off-line pH measurement deviated from the on-line pH measurement by more than 0.04 pH units, the on-line pH was adjusted to equal that of the off-line measurement. Ammonia concentration was quantified using a colorimetric assay (Kodak, Rochester, NY; Biolyzer). Gluco, lactate, and glutamine concentrations were measured enzymatically (YSI, Yellow Springs, OH; 2700 SELECT). Osmolality was measured using the freezing-point depression method (Precision Systems,
SAUER ET AL.: GENERIC FED-BATCH PROCESS
587
Natick, NJ; Osmette). Viable cell concentration and percent viability were determined using a hemacytometer (Hausr Scientific, Horsham, PA; Neubauer) and the trypan blue dye exclusion method. In addition, a 1 mL sample was filtered with a 0.2- m rated membrane (Millipore; Millex-GV) and retained at 2–8 °C for quantification of antibody concentration and, if desired, amino acid analysis. All instrumentation was calibrated and operated according to manufacturer’s instructions. Determination of Antibody Concentration Antibody concentration was determined using a protein A affinity column (PerSeptive Biosystems, Framingham, MA; Pores) and an HPLC system consisting of a pump (PerkinElmer, Oak Brook, IL; 410), an autosampler (Perkin-Elmer; LC2000) and a UV-detector (Perkin-Elmer; LC235). Effluent was monitored at 280 nm. Two solvents (A and B) were made, each containing 50 mM potassium phosphate, 150 mM potassium chloride and 5% isopropanol. The pH of solvent A was adjusted to 7.3 ± 0.1 and the pH of solvent B was adjusted to 1.7 ± 0.1. The column was equilibrated using a mixture of 70% A and 30% B at a flow rate of 1.5 mL min−1. After 50 L of sample was injected, a 1 minute linear gradient to 40% A and 60% B was run and held at that point for another 4 minutes before returning to the equilibration conditions. Frozen calibrators were ud to determine assay variability. The intra-assay coefficient of variation was less than 2%. The inter-assay variability was less than 10%. Amino Acid Analysis Amino acid analysis was performed using a highperformance liquid chromatography system (Hewlett Packard, Corvallis, OR; 1050) using a pre-column OPAderivitization method (Godel et al., 1992). Equations
The integral of viable cell concentration was obtained by dividing Equation (4) by the final culture volume.
tfinal tinitial
XvVdt
Vfinal
(5)
Average Specific Gluco Consumption Rate
The average specific gluco consumption rate qave was glc calculated using Equation (6), where glct was the culture’s residual gluco concentration at time t, glcfeed was the feed medium’s gluco concentration, and Vfeed was the volume of feed medium added to the culture. The average specific gluco consumption rate was obtained by dividing the cumulative gluco consumed by the integral of viable cells; this expression was evaluated at the end of the culture. qave = glc glctinitial Vtinitial − glctfinal Vtfinal +
tfinal tinitial
Vfeed glcfeed (6)
XvVdt
The average specific glutamine consumption rate was calculated in the same manner as qave. glc
Average Specific Antibody Production Rate
The average specific antibody production rate qave was calAb culated using Equation (7), where Abt was the antibody concentration at time t. The average specific antibodyproduction rate was obtained by dividing the cumulative antibody produced by the integral of viable cells; this expression was evaluated at the end of the culture. qave = Ab Abtfinal Vtfinal − Abtinitial Vtinitial
tfinal tinitial
(7)
XvVdt
The average specific production rates of lactate and ammonia were calculated in the same manner as qave. Ab
Integral of Viable Cell Concentration
The integral of viable cells for each discrete time interval was obtained using Equations (2) and (3), where (XvV)1 and (XvV)2 are the total number of viable cells at times t1 and t2, respectively. (XvV)ave is the average total number of viable cells for the time interval t2– t1.
做模型
t2 t1
Average Specific Oxygen Uptake Rate
The oxygen uptake rate (OUR) was calculated using the stationary method according to Equation (8), where kLa was the volumetric oxygen transfer coefficient, C* was the DO concentration in equilibrium with the gas pha, and CL was the culture’s dissolved oxygen concentration. OUR = kLa C* − CL (8)
XvVdt = XvV
ave t2
− t1
2
(2) (3)
XvV 1 + XvV XvV ave = 2
The integral of viable cells was estimated by summing all of the discrete time interval values at t (t1 + t 2) ÷ 2:
色影网站
强上妹妹tfinal tinitial
XvVdt =
t
XvV
ave t2
− t1
(4)
kLa was measured using the static gassing-out method as described by Equation (9). kLa was empirically determined using BM-1 basal medium at veral different bioreactorworking volumes. OUR accounted for changes in culture volume. CL = C* 1 − e−kLa t (9)
588
BIOTECHNOLOGY AND BIOENGINEERING, VOL. 67, NO. 5, MARCH 5, 2000
The average specific oxygen uptake rate qave was calcuOUR lated using Equation (10), where OUave was the cumulative oxygen consumed during time interval t2 – t1. The cumulative oxygen consumed was calculated using Equation (11) where all of the discrete time interval values were summed.
ave qOUR =
OUave
tfinal tinitial
(10)
科目二后视镜
XvVdt t2 − t1 (11)
OUave =
OURt2 Vt2 + OURt1 Vt1 2
RESULTS AND DISCUSSION Fed-Batch Development A batch culture using cell line A provided a performance baline (Fig. 1). The culture lasted 5 days and attained a final antibody concentration of 70 mg L−1. A number of the amino acids, including glutamine, were depleted from the culture medium (data not shown), suggesting that the cells stopped replicating due to nutrient limitation. To replace consumed nutrients, the FM-1 feed medium was developed,
Figure 2. Feeding protocol and process robustness. A controlled experiment was performed using four 3 L bioreactors, cell line A, BM-1 basal medium, and FM-1 feed medium. The experimental variable was the postfeed target gluco concentration (2, 4, 6, or 8 g L−1), as shown in Fig. 2a. The feeding protocol stipulated that a sufficient volume of feed medium be added such that the post-feed gluco concentration equal a specified target value. Figure 2a and 2b contain both measured pre-feed values and calculated post-feed values. To enhance the graphical clarity of Figure 2a and 2b, only the post-feed values are reprented by symbols. Figure 2c and 2d contain only measured pre-f
eed values.
天堂鸟植物along with a simple feeding protocol, that linked the addition of feed medium to the culture’s gluco concentration. Feeding Protocol and Process Robustness
Figure 1. Batch culture. Cell line A was cultured using BM-1 basal medium in a 15 L bioreactor. Figure 1a, b, c prent the viable cell concentration, % viability and antibody concentration over time, respectively.
An optimal feeding regimen would reproducibly maximize product while minimizing process failure. It was hypoth-
1372年SAUER ET AL.: GENERIC FED-BATCH PROCESS吉林高考分数线
589
esized that linking the feeding protocol to a metabolic indicator (i.e., gluco) could lead to the overfeeding or underfeeding of a culture. To test this hypothesis, an experiment was performed examining veral post-feed target gluco concentrations (Fig. 2). Using four 3 L bioreactors, cell line A was cultured in BM-1 basal medium and fed FM-1 feed medium. A relatively wide range of pos
t-feed target gluco concentrations (2, 4, 6, and 8 g L−1) was investigated (Fig. 2a). The different values resulted in different volumetric feed rates (Fig. 2b), which in turn affected cell growth (Fig. 2c). A post-feed target gluco concentration of 2 g L−1 supported cell growth comparable to the batch culture data prented previously (Fig. 1a). The viable cell concentration of the 4 g L−1 culture temporarily declined after day 4 (Fig. 2c), indicating that the volumetric feed rate was approaching the lower boundary of the desired t of possible feeding rates. The 6 and 8 g L−1 post-feed target gluco concentrations resulted in very similar viablecell concentrations (Fig. 2c). However, the antibody concentration of the two cultures differed significantly (Fig. 2d). The 6 g L−1 culture reached a final antibody concentration of 240 mg L−1, while the 8 g L−1 culture reached a final antibody concentration of 180 mg L−1. This difference in final antibody concentration was primarily a function of the difference in the specific antibody production rate (Table II). The 4 g L−1 and the 6 g L−1 cultures had equivalent antibody production through day 7 (Fig. 2d). At day 7, the 6 g L−1 culture expired, reaching a final antibody concentration of 240 mg L−1, while the 4 g L−1 culture continued 1 additional day, reaching a final antibody concentration of 260 mg L−1, the highest concentration achieved. This slight increa in final antibody concentration was the result of a slight increa in the integral of viable cell concentration (Table II). The results show that it was possible to: (1) significantly underfeed a culture (2 g L−1 culture), resulting in performance esntially equival
ent to a batch culture; (2) overfeed a culture (8 g L−1 culture), resulting in reduced antibody production; (3) slightly underfeed a culture (4 g L−1 culture), resulting in a mid-run decrea in viable cell concentration. The relatively flat respon of antibody production to a relatively wide range of intermediate post-feed target values (i.e., 4 and 6 g L−1) demonstrated the robustness of the fed-batch process. The slight underfeeding of the 4 g L−1 culture, combined with the slightly lower final antibody concentration of the 6 g L−1 culture, resulted in a final feeding protocol with variable post-feed target gluco concentrations, increasing from 4.0 to 5.5 g L−1 over the duration of the culture.
Reproducibility of the Fed-Batch Process Five concutive cell line A fed-batch runs were performed at the 15 L bioreactor scale to establish process consistency (Fig. 3). The process exhibited a high degree of reproducibility, evidenced by the low variability of the data. The fed-batch process yielded a final antibody concentration of 220 ± 40 mg L−1, a threefold improvement relative to the batch process final antibody concentration of 70 mg L. Six months elapd from the initial conception of the fed-batch process to the final transfer of technology to manufacturing. This rapid development of the fed-batch process confirms the strategic value of using partial concentrates of basal medium as a starting point for developing feed medium, as communicated by Bibila et al. (1994).
Scaleability of the Fed-Batch Process Five concutive cell line A fed-batch runs were performed at t
he 750 L bioreactor scale. The 750 L scale fed-batch data were compared to the 15 L scale fed-batch data (Table III). No substantial differences were obrved between scales with regard to both magnitude and variation of data. For example, the final antibody concentration was 220 ± 40 mg L−1 at the 15 L scale and 240 ± 40 mg L−1 at the 750 L scale. The variation of all derived data at both scales was less than 25%. Product quality, as determined by standard gel and chromatographic analytic techniques, met internal quality control lot relea and protein characterization specifications.
General Applicability of the Fed-Batch Process The fed-batch process was applied to a number of additional cell lines. The same process developed using cell line A was ud, except that the BM-1 basal medium was replaced with BM-2 basal medium. This change was made to standardize on a single basal medium that could be ud from cell line development through the production process (the BM-1 basal medium could not support clonal growth in the abnce of rum). Four cell lines (cell lines B–E) were subjected to the fed-batch protocol at the 15 L bioreactor scale using the BM-2 basal medium. The data from the four cell lines are prented along with cell line A fed-batch data (Table IV). A batch run was also performed for each cell
色卡纸
Table II. Feeding protocol and process robustness. Figure 2 legend contains experimental details. P
ost-feed target gluco concentration Final culture volume (L) Final antibody concentration (mg L−1) Integral of viable cell concentration (109 cell day L−1) Specific antibody production rate (mg 109 cell−1 day−1) 2.0 g L−1 1.1 70 7 10 4.0 g L−1 2.2 260 21 13 6.0 g L−1 2.2 240 19 13 8.0 g L−1 2.2 180 17 10
590
BIOTECHNOLOGY AND BIOENGINEERING, VOL. 67, NO. 5, MARCH 5, 2000
