Cytotechnology 7: 25-32, 1991. 9 1991 KluwerAcademic Publishers. Printed in the Netherlands.
Optimization of cell culture conditions for G-CSF (granulocyte-colony stimulating factor) production by genetically engineered Namalwa KJM-1 cells Shinji Hosoi, Kazunari Murosumi, Katsutoshi Sasaki, Mitsuo Satoh, Hiromasa Miyaji, Mamoru Hasegawa, Seiga Itoh, Tatsuya Tamaoki and Seiji Sato 1
Tokyo Research Laboratories, Kyowa Hakko Kogyo Co. Ltd., 3-6-6 Asahi-machi, Machida-shi, Tokyo 194, Japan Received 29 March 1991; acceptedin revisedform 23 July 1991
Key words: G-CSF, high density culture, Namalwa KJM-1 cells, optimization, productivity, serum-free medium
Abstract An expression vector for G-CSF, pASLB3-3, was constructed and introduced into Namalwa KJM-1 cells (Hosoi et al., 1988), and cells resistant to 100 nM of methotrexate (MTX) were obtained. Among them, the highest producer, clone SC57, was selected and the productivity of this clone was further characterized. The maximal production of G-CSF was at the most 1.8 ~tg/ml/day using a 25 cm 2 tissue culture flask, even though the cell number was above 7 • 105 cells/ml. The limiting factors at high density were analyzed as the deficiency of nutrients, such as glucose, cysteine and serine, and pH control. The depression of specific G-CSF productivity per celt under the batch culture conditions was overcome by using a perfusion culture system, BiofermenterT M (Sato, 1983) with modifications of nutrients supplementation by a dialysis membrane and/or dissolved oxygen (DO) supplementation by microsilicone fibers. ITPSGF medium was modified to elevate concentrations of amino acids and glucose by 2.0- and 2.5-times, respectively. Under the control of pH at 7.4 and DO at 4 ppm, the specific G - C S F productivity was not depressed even at high cell density (above 1 • 107 cells/ml), and the amount of G-CSF reached 41 ].tg/ml. These results indicated the possibility of finding the optimum culture conditions for the production of recombinant proteins by Namalwa KJM-1 cells.
Abbreviations: ABTS - 2,2'-Azino-di-(3-ethylbenzothiazoline)-6-sulfonic acid; BSA - Bovine Serum Albumin; BSA-PBS - Phosphate-buffered Saline without Ca 2+ and Mg 2§ containing Bovine Serum Albumin; dhfr - Dihydrofolate Reductase; DO - Dissolved Oxygen; G-CSF - Granulocyte Colonystimulating Factor; HEPES - 4-(2-Hydroxyethyl)-l-piperazineethansulfonic Acid; IFN - Interferon; MTX - Methotrexate; P B S ( - ) - Phosphate-buffered saline without Ca 2+ and Mg2+; Tween-PBS Phosphate-buffered saline without Ca 2+ and Mg 2+ containing 0.05% of Tween 20 1 Present address: Research Laboratoiies, KyowaMedix Co. Ltd., 600-1 Minami-ishiki, Nagaizumi-cho, Sunto-gun, Shizuoka411, Japan.
26 Introduction
High density culture is useful for obtaining a large quantity of cellular products and can reduce the cost of the medium. In addition, to clarify the differences in physiological conditions between cells at low and at high density was very important in devising a new system for high density culture. We devised a unique culture technique and system for suspension culture, Biofermenteff M, by analyzing the limiting factors of high density culture (Sato et al., 1983). Biofermenter T M consists of a suspension vessel containing a cone-type cell sedimentation column as cell separator and serving as well for axis of impeller. We previously reported the growth and maintenance of Namalwa KJM-1 cells, a serumindependent subline of Namalwa cells (B lymphoblastoid cells), in suspension culture at high density in chemically defined serum- and albumin-free ITPSGF medium (Hosoi et aL, 1988). Several genes of cytokines were successfully expressed in this cell line and the products were detected in the culture media (Miyaji et al., 1990a,b,c). A method of dihydrofolate reductase (dhfr) gene coamplification was available for efficient expression of foreign genes in this cell line (Miyaji et al., 1990c). We had shown that the expression level of beta-interferon (B-IFN) in this celt line was augmented approximately proportionally to the increase of the cell density in perfusion culture (Miyaji et al., 1990a). Thus, Namalwa KJM-1 is a useful host cell for the production of recombinant products. Molecular cloning and expression of human granulocyte colony-stimulating factor (G-CSF), which stimulates the production of granulocytes almost exclusively from committed precursor cells in soft agar medium, have been reported (Nagata et al., 1986). The production of recombinant G-CSF in E. coli (Komatsu, 1987) and Chinese hamster ovary (CHO) cells (Oheda et al., 1988) have been reported. In this paper, we show the possibility of finding the optimum culture conditions for the production of G-CSF as an example of recombinant
proteins produced by genetically engineered Namalwa KJM-1 cells.
Materials and methods Cells and culture medium
Namalwa ceils, a human lymphoblastoid cell line, were provided from Mr. F. Klein (Frederick Cancer Research Center, Frederick, Maryland, USA). They were adapted to serum- and albumin-free RPMI-1640 medium supplemented with 4-(2hydroxyethyl) - 1 - piperazineethanesulfonic acid (HEPES) (10 mM), L-glutamine (4 mM), penicillin (25 U/ml), streptomycin (25 gg/ml), insulin (3 gg/ml), transferrin (5 gg/ml), sodium pyruvate (5 mM), sodium selenite (125 nM), galactose (1 mg/mi) and Pluronic F68 (1 mg/ml); we called this ITPSGF medium (Hosoi et al., 1988).
Chemicals and immunological reagents
RPMI-1640 was purchased from Nissui Seiyaku Co. Ltd., Tokyo. Pluronic F68 (Pepol B-188) from Toho Chiba Chemicals Co. Ltd., Chiba, Japan, MTX from Sigma Chemical Co., St. Louis, MO, ABTS from Nakarai, Kyoto, Japan, Peroxidase-conjugated anti-rabbit immunoglobuIin from D A K O Japan Co., Ltd., Kyoto, Japan, biotin-conjugated anti-rabbit immunoglobulin G and alkaline phosphatase-conjugated avidin from R&D systems, Inc., McKinley Place N.E., MN were used. Rabbit anti-human G-CSF (anti-GCSF) polyclonal antibody and Mouse anti-G-CSF monoclonal antibody KM341 (Yoshida and Shithara, 1990) were obtained from our laboratories. Other chemicals were obtained as described previously (Hosoi et al., 1988; Miyaji et al., 1990a; Satoh et al., 1990).
D N A manipulations
All enzymes were purchased from Takara Shuzo, Kyoto, Japan. Plasmid DNA preparation, DNA
27 fragment purification, E. coli DNA polymerase I reaction, ligation and the introduction of plasmid DNA into E. coli were done as described by Nishi et al. (1984).
Construction of an expression vector f o r G-CSF An expression plasmid pASLB3-3 (Fig. 1) consists of the following five D N A fragments: i) a 0.45-kilobase pairs (kb) SalI-AatII fragment containing a part of hG-CSF cDNA from pCSF3-3 (Kuga et al., 1990); ii) a 0.2-kb AatII-DdeI (blunt) fragment containing a part of hG-CSF cDNA from pCfTA1 (Komatsu, 1987); iii) an 8.7-kb XhoI-SmaI fragment containing an ampicillin-resistance gene, a G418-resistance gene, and a dhfr transcription unit from pSE1B d2-4
(Miyaji, 1990c); iv) a 0.3-kb XhoI-BglI fragment containing the SV40 early promoter from pAGE107 (Miyaji et al., 1990a). The BglI site was converted to a BanII site by insertion of synthetic DNA in the following sequence (A); v ) a 0.3-kb BanII-Sau3AI fragment containing a segment and part of the U5 sequence of the HTLV- 1 LTR from pATK-03 (Seiki, 1983). The Sau3AI (BarnHI) site was converted to a SalI site by insertion of a synthetic DNA in the following sequence (n). Sau3AI
SalI
(A) 5' CGGGCT 3' (B) 5'-G GATCCCCGGTC GACC 3' GGAGC 5'
Electroporation A somatic hybridizer SSH-1 (Shimadzu Seisakusyo, Kyoto, Japan) and an SSH-C13 chamber (distance of electrode; 2 mm) were used details of the procedures were previously described by Miyaji et al. (1990a).
XhoI
Selection of MTX-resistant subclones
pASLB3-3
/i/ dill
ClaIZ
G418-resistant transformants were plated into 96well plates at 4 x 104 cells per well in 0.2 ml of RPMI-1640 medium containing 10% FCS, 0.3 mg/ml G418 and 50 nM MTX. Drug-resistant subclones were obtained two to five weeks after starting selection. 100 nM MTX-resistant subclones were obtained from 50 nM MTX-resistant subclones by the same procedure, except for increasing concentrations of MTX.
Fig. 1. Construction of an expression vector pASLB3-3. The following abbreviations were used: G418/Km, the Tn5-derived G418 resistance gene; Ap, ampicillin resistance gene; P1, pBR322 P1 promoter: Ptk, the Herpes simplex virus thymidine kinase promoter; PSE, the SV40 early promote Atk, the polyadenylation signal from the Herpes simplex virus thymidine kinase gene; ASE, the polyadenylation signal from the SV40 early gene; AB G, the polyadenylation signal from the rabbit B-globin gene; Sp.l G, Splicing signal from the rabbit B-globin gene; R+U5', the R segment and part the U5 sequence of the HTLV-1 LTR; dhfr, murine dihydrofolate reductase gene.
Measurement of G-CSF sandwich-type ELISA a) sandwich-type ELISA. G-CSF was measured by sandwich-type ELISA. A 96-well ELISA plate (Flow Laboratories, Inc., McLean, USA.) was coated with 50 [.tl of purified rabbit polyclonal anti-G-CSF antibody (10 ~tg/ml) at 4~ overnight. After removal of the antibody solution, the
28 residual protein-binding sites on the plates were blocked with 100 pl of P B S ( - ) containing 1% o f B S A (1% BSA-PBS). T h e n 50 ~tl of a sample containing G - C S F were added to the wells a n d incubated overnight at 4 ~ After washing with P B S ( - ) containing 0.05% o f T w e e n 20 (TweenPBS) three times, they were incubated with antibody solution (10 ~g/ml o f purified m o u s e monoclonal antibody KM341 in 1% B S A - P B S ) at r o o m temperature for 1 h. After washing three times with T w e e n - P B S , they were incubated with a 200-fold diluted peroxidase-conjugated antirabbit irnmunoglobulin for 2 h at r o o m temperature. After rinsing three times with T w e e n - P B S , the final reaction was visualized by incubating with A B T S and 0.01% o f hydrogen peroxide. The absorbance at 415 n m was measured. G - C S F purified from S C 5 7 conditioned m e d i u m was used as a standard. All experiments were done in duplicate from each conditioned medium.
Results
Production of G-CSF in Namalwa KJM-1 cells An expression vector for G-CSF, pASLB3-3, was Constructed and introduced into Namalwa KJM-1 cells by electroporation. The highest G - C S F producer among 100 n M M T X resistant clones, clone SC57, was selected, and the productivity of this clone was further characterized. Using a 25 cm 2 tissue culture flask, the maximal production of G CSF was at most 1.8 ~tg/ml/ day, even though the cell number was above 7 • 105 cells/ml (Fig. 2). This limitation of productivity m a y be solved b y use o f a perfusion culture system. 2.5 E - 2.0
m
u
b) densitometry. S D S - P A G E was run according tO the m e t h o d o f Laemmli (1970) with reduction using 2-mercaptoethanol. Samples were loaded on a r e a d y - m a d e gradient slab gel ( 1 0 - 2 0 % polyacrylamide, Daiich Pure Chemical, Tokyo). Proteins on the gel were visualized with Coomassie Brilliant Blue (CBB) R-250. G - C S F contents were calculated by absorbance o f each sample at 560 n m using a densitometer CS-930 (Shimadzu Seisakusyo, K y o t o , Japan) versus purified G - C S F derived from SC57 as a standard.
Amino acid and glucose analysis A m i n o acid concentrations were determined by High Performance Amino Acid Analyzer JLC200 ( J E O L , T o k y o , Japan). Glucose concentration was assayed with an enzymatic analyzer GL-101 (Mitsubishi Kasei, T o k y o , Japan).
CO 0
6)
1.5
o (0
v
0
JQ
1.0
c~ =::L
0.5
(n
E 14.
G-CSF
Q;
o !
2
r
4 Culture
I
!
1
6 8 10 period (days)
[
0.0
12
Fig, 2. Production of G-CSF by SC57 (t00 nM MTX-resistant clones) 10 ml of cells were inoculated at 5 x 105 cells/ml into the 25 em2 tissue culture flask and cultured at 37~ in 5% carbon dioxide atmosphere. Cell counts and medium changes were done at the indicated time. Determinations were made from duplicate cultures, The conditioned media were assayed for G-CSF by sandwich-typeELISA.
High density culture of SC57 SC57 (2 • 106 cells/ml) was inoculated in I T P S G F m e d i u m in a perfusion culture system, Biofermenter T M with a dialysis membrane. The production level was augmented with an increase in cell density (Fig. 3). SC57 proliferated to 1.6 • 107 cells/ml and the amount o f G - C S F reached 14
29 107 .
106 16
Eo,,
14 12
E
10
-~
~ =::z
12 10
v
'd
~' 107l
8
u_
6
0
4 o
---III-106|
~ 0
2
9
~ ,
4 Culture
~
G-CSF ,
t
6 8 period
,
(5
-2 j
,
~ ,
10 12 (days)
106.
0
"1"
"r"
EJ
I!
14
2
4
6
8
10
6 4 2 0
E It. O3
o
C u l t u r e p e r i o d (days)
0 4
Fig. 3. High density culture of SC57 using a perfusion culture system, Biofermenter, with a dialysis membrane. Cells were inoculated at a density of 2 x 106 cells/ml into a perfusion culture vessel (Net culture volume was 200 ml) with a dialysis membrane (Spectrum Medical Industries Inc., Terminal annex, LA; Working area: 78.5 cm 2) and cultured at 37~ for 13 days with agitation of 40 rpm. The perfusion rate was gradually shifted up stepwise in proportion to cell density from 1.25 vol/day to 5 vol/day. DO and pH were not controlled. Cell counts were done every day. Each point represents the mean for duplicate.
gg/ml. The specific G-CSF productivities were 4.0 pg/106 cells/day on day 2, 2.9 gg/106 cells/ day on day 7, and 2.4 gg/106 cells/day on day 12.
Effects of pH control Compared with the physiological conditions in the flask, those in perfusion culture system were significantly different. Unless otherwise noted, we assessed these differences from the viewpoint of productivity by semi-continuous culture method centrifuged every day (net culture vol. was 500 ml). DO,-lactate and ammonia, have no effect on the G-CSF productivity of SC57 (data not shown). Accurate control of pH by the addition of 7.5% sodium bicarbonate affected the productivity (Fig. 4). Higher productivity was observed under the pH controlled at 7.4 rather than at 7.0. The specific G-CSF productivity on day 5 was 2.3 ~g/106 cells/day at pH 7.4 and 1.2 ~g/106 cells/
Fig. 4. Effect of control pH on productivity. SC57 cells were inoculated into 500 ml of spinner flask with micro-silicone fibers (Nagayanagi Kogyo Co., Ltd., Tokyo) at the density of about 2 x 106 cells/ml and cultured at 37~ for 9 days with agitation of 40 rpm. pH was measured by an in situ pH probe (Ingold, Urdorf, Switzerland) and controlled varied less than 0.1 pH unit by the addition of 7.5% NaHCO 3 solution. DO was measured by an in situ DO probe (Oriental Electric Co. Ltd., Tokyo) and controlled at 4 ppm by flow rate of air through micro-silicone fibers. Cell counts, samplings and medium changes were done everyday. Each point represents the mean for the duplicate.
day at pH 7.0. In these cases, the cell densities were above 3 x 106 cells/ml.
Effects of nutrients SC57 cells were inoculated into 500 ml spinner flask with micro-silicone fibers at the density of 4 • 106 cells/ml. After cultivation for 24 hours, the conditioned medium was recovered and amino acid contents were analyzed by High Performance Amino Acid Analyzer JLC-200. Amino acid contents of basal RPMI-1640 medium was compared. From the results of this experiment, the deficiency of cysteine (initial concentration; 50 mgB, after cultivation; not detected), serine (initial concentration; 30 mg/l, after cultivation; 1.1 rag/1 ), and glucose (initial concentration; 2 g/l, after cultivation; 0.59 g/1 ) was observed. Addition of 3 g/1 of glucose to the ITPSGF medium did not affect the productivity but did affect the growth rate at high densities above 4 x 106 cells/ml (data not shown).
30 10 8
10 7
a.a. rich m e d i u m 40 -~
E _~
_=~
_=o 30
10
40
In
E
~ ,.:
E~
r 20
50
~ v ,,
co o,. r -- +
r
30 .s
1 0 7.
E
2O
r
10 6 0
1
2
3
4
5
6
(5
7
c u l t u r e period (days)
10
r 10 6
,
U n d e r the semi-continuous batch culture conditions, SC57 was cultured in the I T P S G F m e d i u m with 0.1 m M of cysteine and serine and 3 g/1 o f glucose added with the p H controlled at 7.4 for 6 days (Fig. 5). T h e specific G - C S F productivity o f SC57 was enhanced to 5.7 Ixg/i06 cells/ d a y at p H 7.4 (day 3), which was 1.8 times higher than that o f the control, 3.2 gg/106 cells/day.
High density culture of SC57 using a perfusion culture system Using a perfusion culture system, B i o f e r m e n t e r T M with a dialysis m e m b r a n e and micro-silicone fibers (net culture vol. was 250 ml), SC57 was cultured for 18 days (Fig. 6). At the beginning of the culture, n o r m a l I T P S G F m e d i u m was used, but after day 16, a special m e d i u m in which a m i n o acid and glucose concentration was increased 2 and 2.5 times respectively, was used. T h e specific G - C S F productivity was 2.1 g g / 106 cells/day on day 3, 2.0 ktg/106 cells/day on day 7, 2.6 ~tg/106 cells/day on day 11, 2.7 ktg/106 cells/day on d a y 15, 3.4 gg/106 cells/day on day 17 (the a m o u n t o f G - C S F reached 41 I.tg/ml).
i
2
Fig. 5. Effects of nutrient. SC57 cells were inoculated into 500 ml of spinner flask with micro-silicone fibers at the density of about 4 x 106 cells/ml and cultured at 37~ for 6 days with agitation of 40 rpm. 3 mg/ml of glucose and 0.1 mM of serine and cysteine were added to ITPSGF medium. The controlled pH was 7.4 varied less than 0.1 pH unit. Other conditions were identical to Fig. 4.
n (/3 O
,
I
4
,
i
9
6
Culture
i
8
.
!
,
i
,
1
,
i
,
i
,
0
101214161820 period (days)
Fig. 6. G-CSF production under the optimum conditions. High density culture of SC57 was performed by a perfusion culture system, BiofermenterTM with a dialysis membrane and microsilicone fibers. Net culture volume was 250 ml. SC57 cells were inoculated at the density of 2.5 x 106 cells/ml and cultured at 37~ for 18 days with agitation of 40 rpm. The perfusion rate was set to about 1.0 vol/day and the dialysis rate was shifted up stepwise in proportion to cell density from 1.0 vol/day to 4.0 vol/day. Measurement and control of pH and DO were done under the identical conditions in Fig. 4. Measurement of G-CSF contents were performed by densitometry. Other conditions were identical to Fig. 4.
Discussion H i g h density culture m a k e s it possible to use a smaller vessel to obtain a required a m o u n t o f target proteins. G e n e technology m a y be applied further to simplify the cellular requirements for nutrients, growth factors, and regulatory factors and to increase productivity. H o w e v e r , the physiological conditions of production phase were s o m e w h a t different f r o m those o f the growth phase. Thus, attention to the physiological and b i o c h e m i c a l balance of the growth m e d i u m is needed for the technology of maintaining the specific productivity o f the cells at high level for long culture periods. T o achieve high-level expression o f r e c o m binant proteins, gene amplification using dhfr is a very effective tool. F o r this purpose, Chinese h a m s t e r o v a r y ( C H O ) ceils deficient in dhfr h a v e been successfully used as the host cell line (Kaufm a n and Sharp, 1982; K a u f m a n et al. 1985). But
31 several cell lines that contain active endogenous dhfr genes can be used for gene amplification with dhfr (Dorai and Moore, 1987). We previously showed that the expression level of B-IFN secreted by a recombinant Namalwa KJM-1 was augmented approximately proportionally with increase in cell density in perfusion culture (Miyaji e t al., 1990a) and that a method of dhfr gene coamplification was available in this cell line (Miyaji e t a L , 1990c). An expression vector for G-CSF, pASLB33,was constructed and introduced into Namalwa KJM-1 ceils (Hosoi e t al. , 1988), and cells resistant to 100 nM of methotrexate (MTX) were obtained. Among them, the highest producer, clone SC57, was selected and the productivity of this clone was further characterized. The specific G-CSF productivity of SC57 was depressed to less than 2.5 /.tg/106 cells/day when the cell number was above 7 x 105 cells/ml (Fig. 2). However, the production of G-CSF might be correlated to the cell density, because G-CSF accumulated in large quantities in a perfusion culture system (Fig. 3). These results showed that the specific productivity w a s affected by the physiological conditions, i.e. pH, DO, and the concentrations of nutrients and metabolites. Analyses of productivity in high density culture showed that the limiting factors were the deficiency of nutrients, such as glucose, cysteine, and serine (Fig. 5). The importance of choosing the appropriate medium composition to avoid nutrient limitation, to obtain acceptable high densities in reactor systems, and to maintain high production rate of a target product has been reported for Ltk and BHK cells (Wagner e t al., 1988). Our results showed that the growth medium should be different from a production medium in the amino acid consumption. The specific productivity was depressed at a lower than optimum pH (Fig. 4). It is reported that the pH control results in a significant improvement in the WN production by recombinant CHO cells (Smiley e t al., 1989). Our results showed that the range of controlled pH which gave higher productivity was limited more than that of recombinant E . coli.
The depression of productivity was solved by means of the Biofermenter with micro-silicone fibers and a dialysis membrane in a modified medium in which concentrations of amino acids and glucose were elevated 2 and 2.5 times, respectively (Fig. 6). As a result, the specific productivity was not decreased even at a high cell density, and the amount of G-CSF reached 41 gg/ml. This increase in specific productivity before day !5 was due to shifts in the dialysis rate and that of day 17 was due to a change of medium. These results showed that the high density culture of SC57 was optimized without decreasing of specific productivity. In this paper, we have shown the way to find the optimum culture conditions for the production of G-CSF as an example of recombinant proteins produced by clone SC57, genetically engineered Namalwa KJM-1 cells. Optimum conditions may be different between clones and host cell lines, but it is possible to find the optimum conditions for each cell as shown in this report.
Acknowledgements Our most sincere appreciation is due to the Ministry of Intemational Trade and Industry of Japan, and NEDO (New Energy and Industrial Technology Development Organization) for their financial support and excellent advice. We also thank Ms. Sachiko Shitara and Ms. Sachiko Yokokawa for their excellent technical assistance.
References 1. Dorai H and MooreGP (1987) The effectof dihydrofolate reductase-mediatedgene amplificationon the expressionof transfectedimmunoglobulingenes.J. Immunol.139:42324241. 2. HosoiS, Mioh H, AnzaiC, Sato S and FujiyoshiN (1988) Establishmentof Namalwacell lines which grow continuously in glutamine-freemedium. Cytotechnology1: 151158. 3. KaufmanRJ and Sharp PA (1982) Amplificationand expression of sequences cotransfectedwith a modular dihydrofolate reductase complementaryDNA gene. J. Mol. Biol. 159: 601-621.
32 4. Kaufman RJ, Wasley LC, Spiliotes AJ, Gossels SD, Latt SA, Larsen GR and Kay RM (1985) Coamplification and coexpression of human tissue-type plasminogen activator and murine dihydrofolate reductase sequences in Chinese hamster ovary cells. Moh Cell. Biol. 5: 1750-1759. 5. Komatsu Y, Matsumoto T, Kuga T, Nishi T, Sekine S, Saito A, Okabe M, Morimoto M, Itoh S, Okabe T and Takaku F (1987) Cloning of granulocyte colony-stimulatirig factor cDNA from human macrophages and its expression in Escherichia coli. Jpn. J. Cancer Res. 78: 11791181. 6. Kuga T, Komatsu Y, Mizukami T, Sato M and Itoh S (1990) Effect of N-terminal deletion of signal peptide on the secretion of human granulocyte colony-stimulating factor in mammalian cells. Biotechnoh Lett. 12: 87-92. 7. Laemmli UK (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature. 227: 680-685. 8. Miyaji H, Mizukami T, Hosoi S, Sato S, Fujiyoshi N and Itoh S (1990a) Expression of human beta-interferon in Namalwa cells which were adapted to serum-free medium. Cytotechnology 3: 133-140. 9. Miyaji H, Harada N, Mizukami T, Sato S, Fujiyoshi N and Itoh S (1990b) Expression of human lymphotoxin in Namalwa ceils which were adapted to serum-free medium. Cytotechnology 4: 39--43. 10. Miyaji H, Harada N, Mizukami T, Sato S, Fujiyoshi N and Itoh S (1990c) Efficient expression of human beta-interferon in Namalwa cells adapted to serum-free medium by a dhfr gene coamplification method. Cytotechnology 4: 173180. 11. Nagata S, Tsuchiya M, Asano S, Kaziro Y, Yamazaki T, Yamamoto O, Hirata Y, Kubota N, Oheda M, Nomura H and Ono M (1986) Molecular cloning and expression of cDNA for human granulocyte colony-stimulating factor. Nature 319: 415--418.
12. Nishi T, Saito A, Oka T, Itoh S, Tagaoka C and Taniguchi T (1984) Construction of plasmid expression vectors carry{ng the Escherichia coli tryptophan promoter. Agric. Biol. Chem. 48: 669-675. 13. Oheda M, Hase S, Ono M and Ikenaka T (1988) Structure of the sugar chains of recombinant human granulocytecolony-stimulating factor produced by chinese hamster ovary cells. J. Biochem. 103: 544-546. 14. Sato S, Kawamura K and Fujiyoshi N (1983) Animal cell cultivation for production of biological substances with a novel perfusion culture apparatus. J. Tissue Culture Methods 8: 167-171. 15. Satoh M, Hosoi S and Sato S. CHO (1990) (Chinese Hamster Ovary) cells spontaneously secret a Cysteine Endopeptidase. In Vitro Cell. Dev. Biol. 26:1101-1104. 16. Seiki M, Hattori S, Hirayama Y and Yoshida M (1983) Human adult T-cell leukemia virus: Complete nucleotide sequence of the provirus genome integrated in leukemia cell DNA. Proc. Natl. Acad. Sci. USA. 80: 3618-3622. 17. Smiley AL, Hu W-S and Wang DIC (1989) Production of Human Immune Interferon by recombinant Mammalian cells cultivated on Microcarriers. Biotechnol. Bioeng. 33: 1182-1190. 18. Wagner R, Ryll T, Krafft H and Lehmann J (1988) Variation of amino acid concentration in the medium of HU-IFN and HU IL-2 producing cell line. Cytotechnology 1: 145150. 19. Yoshida H and Shitara S (1989) Generation and Characterization of Monoclonal antibodies to Recombinant Human Granulocyte-colony Stimulating Factor (G-CSF) and Its Mutein. Agric. Biol. Chem. 53: 1095-1101.
Address for offprints: Shinji Hosoi, Tokyo Research Laboratories, Kyowa Hakko Kogyo Co. Ltd., 3-6-6 Asahi-machi, Machida-shi, Tokyo 194, Japan
Optimization of cell culture conditions for G-CSF (granulocyte-colony stimulating factor) production by genetically engineered Namalwa KJM-1 cells Shinji Hosoi, Kazunari Murosumi, Katsutoshi Sasaki, Mitsuo Satoh, Hiromasa Miyaji, Mamoru Hasegawa, Seiga Itoh, Tatsuya Tamaoki and Seiji Sato 1
Tokyo Research Laboratories, Kyowa Hakko Kogyo Co. Ltd., 3-6-6 Asahi-machi, Machida-shi, Tokyo 194, Japan Received 29 March 1991; acceptedin revisedform 23 July 1991
Key words: G-CSF, high density culture, Namalwa KJM-1 cells, optimization, productivity, serum-free medium
Abstract An expression vector for G-CSF, pASLB3-3, was constructed and introduced into Namalwa KJM-1 cells (Hosoi et al., 1988), and cells resistant to 100 nM of methotrexate (MTX) were obtained. Among them, the highest producer, clone SC57, was selected and the productivity of this clone was further characterized. The maximal production of G-CSF was at the most 1.8 ~tg/ml/day using a 25 cm 2 tissue culture flask, even though the cell number was above 7 • 105 cells/ml. The limiting factors at high density were analyzed as the deficiency of nutrients, such as glucose, cysteine and serine, and pH control. The depression of specific G-CSF productivity per celt under the batch culture conditions was overcome by using a perfusion culture system, BiofermenterT M (Sato, 1983) with modifications of nutrients supplementation by a dialysis membrane and/or dissolved oxygen (DO) supplementation by microsilicone fibers. ITPSGF medium was modified to elevate concentrations of amino acids and glucose by 2.0- and 2.5-times, respectively. Under the control of pH at 7.4 and DO at 4 ppm, the specific G - C S F productivity was not depressed even at high cell density (above 1 • 107 cells/ml), and the amount of G-CSF reached 41 ].tg/ml. These results indicated the possibility of finding the optimum culture conditions for the production of recombinant proteins by Namalwa KJM-1 cells.
Abbreviations: ABTS - 2,2'-Azino-di-(3-ethylbenzothiazoline)-6-sulfonic acid; BSA - Bovine Serum Albumin; BSA-PBS - Phosphate-buffered Saline without Ca 2+ and Mg 2§ containing Bovine Serum Albumin; dhfr - Dihydrofolate Reductase; DO - Dissolved Oxygen; G-CSF - Granulocyte Colonystimulating Factor; HEPES - 4-(2-Hydroxyethyl)-l-piperazineethansulfonic Acid; IFN - Interferon; MTX - Methotrexate; P B S ( - ) - Phosphate-buffered saline without Ca 2+ and Mg2+; Tween-PBS Phosphate-buffered saline without Ca 2+ and Mg 2+ containing 0.05% of Tween 20 1 Present address: Research Laboratoiies, KyowaMedix Co. Ltd., 600-1 Minami-ishiki, Nagaizumi-cho, Sunto-gun, Shizuoka411, Japan.
26 Introduction
High density culture is useful for obtaining a large quantity of cellular products and can reduce the cost of the medium. In addition, to clarify the differences in physiological conditions between cells at low and at high density was very important in devising a new system for high density culture. We devised a unique culture technique and system for suspension culture, Biofermenteff M, by analyzing the limiting factors of high density culture (Sato et al., 1983). Biofermenter T M consists of a suspension vessel containing a cone-type cell sedimentation column as cell separator and serving as well for axis of impeller. We previously reported the growth and maintenance of Namalwa KJM-1 cells, a serumindependent subline of Namalwa cells (B lymphoblastoid cells), in suspension culture at high density in chemically defined serum- and albumin-free ITPSGF medium (Hosoi et aL, 1988). Several genes of cytokines were successfully expressed in this cell line and the products were detected in the culture media (Miyaji et al., 1990a,b,c). A method of dihydrofolate reductase (dhfr) gene coamplification was available for efficient expression of foreign genes in this cell line (Miyaji et al., 1990c). We had shown that the expression level of beta-interferon (B-IFN) in this celt line was augmented approximately proportionally to the increase of the cell density in perfusion culture (Miyaji et al., 1990a). Thus, Namalwa KJM-1 is a useful host cell for the production of recombinant products. Molecular cloning and expression of human granulocyte colony-stimulating factor (G-CSF), which stimulates the production of granulocytes almost exclusively from committed precursor cells in soft agar medium, have been reported (Nagata et al., 1986). The production of recombinant G-CSF in E. coli (Komatsu, 1987) and Chinese hamster ovary (CHO) cells (Oheda et al., 1988) have been reported. In this paper, we show the possibility of finding the optimum culture conditions for the production of G-CSF as an example of recombinant
proteins produced by genetically engineered Namalwa KJM-1 cells.
Materials and methods Cells and culture medium
Namalwa ceils, a human lymphoblastoid cell line, were provided from Mr. F. Klein (Frederick Cancer Research Center, Frederick, Maryland, USA). They were adapted to serum- and albumin-free RPMI-1640 medium supplemented with 4-(2hydroxyethyl) - 1 - piperazineethanesulfonic acid (HEPES) (10 mM), L-glutamine (4 mM), penicillin (25 U/ml), streptomycin (25 gg/ml), insulin (3 gg/ml), transferrin (5 gg/ml), sodium pyruvate (5 mM), sodium selenite (125 nM), galactose (1 mg/mi) and Pluronic F68 (1 mg/ml); we called this ITPSGF medium (Hosoi et al., 1988).
Chemicals and immunological reagents
RPMI-1640 was purchased from Nissui Seiyaku Co. Ltd., Tokyo. Pluronic F68 (Pepol B-188) from Toho Chiba Chemicals Co. Ltd., Chiba, Japan, MTX from Sigma Chemical Co., St. Louis, MO, ABTS from Nakarai, Kyoto, Japan, Peroxidase-conjugated anti-rabbit immunoglobuIin from D A K O Japan Co., Ltd., Kyoto, Japan, biotin-conjugated anti-rabbit immunoglobulin G and alkaline phosphatase-conjugated avidin from R&D systems, Inc., McKinley Place N.E., MN were used. Rabbit anti-human G-CSF (anti-GCSF) polyclonal antibody and Mouse anti-G-CSF monoclonal antibody KM341 (Yoshida and Shithara, 1990) were obtained from our laboratories. Other chemicals were obtained as described previously (Hosoi et al., 1988; Miyaji et al., 1990a; Satoh et al., 1990).
D N A manipulations
All enzymes were purchased from Takara Shuzo, Kyoto, Japan. Plasmid DNA preparation, DNA
27 fragment purification, E. coli DNA polymerase I reaction, ligation and the introduction of plasmid DNA into E. coli were done as described by Nishi et al. (1984).
Construction of an expression vector f o r G-CSF An expression plasmid pASLB3-3 (Fig. 1) consists of the following five D N A fragments: i) a 0.45-kilobase pairs (kb) SalI-AatII fragment containing a part of hG-CSF cDNA from pCSF3-3 (Kuga et al., 1990); ii) a 0.2-kb AatII-DdeI (blunt) fragment containing a part of hG-CSF cDNA from pCfTA1 (Komatsu, 1987); iii) an 8.7-kb XhoI-SmaI fragment containing an ampicillin-resistance gene, a G418-resistance gene, and a dhfr transcription unit from pSE1B d2-4
(Miyaji, 1990c); iv) a 0.3-kb XhoI-BglI fragment containing the SV40 early promoter from pAGE107 (Miyaji et al., 1990a). The BglI site was converted to a BanII site by insertion of synthetic DNA in the following sequence (A); v ) a 0.3-kb BanII-Sau3AI fragment containing a segment and part of the U5 sequence of the HTLV- 1 LTR from pATK-03 (Seiki, 1983). The Sau3AI (BarnHI) site was converted to a SalI site by insertion of a synthetic DNA in the following sequence (n). Sau3AI
SalI
(A) 5' CGGGCT 3' (B) 5'-G GATCCCCGGTC GACC 3' GGAGC 5'
Electroporation A somatic hybridizer SSH-1 (Shimadzu Seisakusyo, Kyoto, Japan) and an SSH-C13 chamber (distance of electrode; 2 mm) were used details of the procedures were previously described by Miyaji et al. (1990a).
XhoI
Selection of MTX-resistant subclones
pASLB3-3
/i/ dill
ClaIZ
G418-resistant transformants were plated into 96well plates at 4 x 104 cells per well in 0.2 ml of RPMI-1640 medium containing 10% FCS, 0.3 mg/ml G418 and 50 nM MTX. Drug-resistant subclones were obtained two to five weeks after starting selection. 100 nM MTX-resistant subclones were obtained from 50 nM MTX-resistant subclones by the same procedure, except for increasing concentrations of MTX.
Fig. 1. Construction of an expression vector pASLB3-3. The following abbreviations were used: G418/Km, the Tn5-derived G418 resistance gene; Ap, ampicillin resistance gene; P1, pBR322 P1 promoter: Ptk, the Herpes simplex virus thymidine kinase promoter; PSE, the SV40 early promote Atk, the polyadenylation signal from the Herpes simplex virus thymidine kinase gene; ASE, the polyadenylation signal from the SV40 early gene; AB G, the polyadenylation signal from the rabbit B-globin gene; Sp.l G, Splicing signal from the rabbit B-globin gene; R+U5', the R segment and part the U5 sequence of the HTLV-1 LTR; dhfr, murine dihydrofolate reductase gene.
Measurement of G-CSF sandwich-type ELISA a) sandwich-type ELISA. G-CSF was measured by sandwich-type ELISA. A 96-well ELISA plate (Flow Laboratories, Inc., McLean, USA.) was coated with 50 [.tl of purified rabbit polyclonal anti-G-CSF antibody (10 ~tg/ml) at 4~ overnight. After removal of the antibody solution, the
28 residual protein-binding sites on the plates were blocked with 100 pl of P B S ( - ) containing 1% o f B S A (1% BSA-PBS). T h e n 50 ~tl of a sample containing G - C S F were added to the wells a n d incubated overnight at 4 ~ After washing with P B S ( - ) containing 0.05% o f T w e e n 20 (TweenPBS) three times, they were incubated with antibody solution (10 ~g/ml o f purified m o u s e monoclonal antibody KM341 in 1% B S A - P B S ) at r o o m temperature for 1 h. After washing three times with T w e e n - P B S , they were incubated with a 200-fold diluted peroxidase-conjugated antirabbit irnmunoglobulin for 2 h at r o o m temperature. After rinsing three times with T w e e n - P B S , the final reaction was visualized by incubating with A B T S and 0.01% o f hydrogen peroxide. The absorbance at 415 n m was measured. G - C S F purified from S C 5 7 conditioned m e d i u m was used as a standard. All experiments were done in duplicate from each conditioned medium.
Results
Production of G-CSF in Namalwa KJM-1 cells An expression vector for G-CSF, pASLB3-3, was Constructed and introduced into Namalwa KJM-1 cells by electroporation. The highest G - C S F producer among 100 n M M T X resistant clones, clone SC57, was selected, and the productivity of this clone was further characterized. Using a 25 cm 2 tissue culture flask, the maximal production of G CSF was at most 1.8 ~tg/ml/ day, even though the cell number was above 7 • 105 cells/ml (Fig. 2). This limitation of productivity m a y be solved b y use o f a perfusion culture system. 2.5 E - 2.0
m
u
b) densitometry. S D S - P A G E was run according tO the m e t h o d o f Laemmli (1970) with reduction using 2-mercaptoethanol. Samples were loaded on a r e a d y - m a d e gradient slab gel ( 1 0 - 2 0 % polyacrylamide, Daiich Pure Chemical, Tokyo). Proteins on the gel were visualized with Coomassie Brilliant Blue (CBB) R-250. G - C S F contents were calculated by absorbance o f each sample at 560 n m using a densitometer CS-930 (Shimadzu Seisakusyo, K y o t o , Japan) versus purified G - C S F derived from SC57 as a standard.
Amino acid and glucose analysis A m i n o acid concentrations were determined by High Performance Amino Acid Analyzer JLC200 ( J E O L , T o k y o , Japan). Glucose concentration was assayed with an enzymatic analyzer GL-101 (Mitsubishi Kasei, T o k y o , Japan).
CO 0
6)
1.5
o (0
v
0
JQ
1.0
c~ =::L
0.5
(n
E 14.
G-CSF
Q;
o !
2
r
4 Culture
I
!
1
6 8 10 period (days)
[
0.0
12
Fig, 2. Production of G-CSF by SC57 (t00 nM MTX-resistant clones) 10 ml of cells were inoculated at 5 x 105 cells/ml into the 25 em2 tissue culture flask and cultured at 37~ in 5% carbon dioxide atmosphere. Cell counts and medium changes were done at the indicated time. Determinations were made from duplicate cultures, The conditioned media were assayed for G-CSF by sandwich-typeELISA.
High density culture of SC57 SC57 (2 • 106 cells/ml) was inoculated in I T P S G F m e d i u m in a perfusion culture system, Biofermenter T M with a dialysis membrane. The production level was augmented with an increase in cell density (Fig. 3). SC57 proliferated to 1.6 • 107 cells/ml and the amount o f G - C S F reached 14
29 107 .
106 16
Eo,,
14 12
E
10
-~
~ =::z
12 10
v
'd
~' 107l
8
u_
6
0
4 o
---III-106|
~ 0
2
9
~ ,
4 Culture
~
G-CSF ,
t
6 8 period
,
(5
-2 j
,
~ ,
10 12 (days)
106.
0
"1"
"r"
EJ
I!
14
2
4
6
8
10
6 4 2 0
E It. O3
o
C u l t u r e p e r i o d (days)
0 4
Fig. 3. High density culture of SC57 using a perfusion culture system, Biofermenter, with a dialysis membrane. Cells were inoculated at a density of 2 x 106 cells/ml into a perfusion culture vessel (Net culture volume was 200 ml) with a dialysis membrane (Spectrum Medical Industries Inc., Terminal annex, LA; Working area: 78.5 cm 2) and cultured at 37~ for 13 days with agitation of 40 rpm. The perfusion rate was gradually shifted up stepwise in proportion to cell density from 1.25 vol/day to 5 vol/day. DO and pH were not controlled. Cell counts were done every day. Each point represents the mean for duplicate.
gg/ml. The specific G-CSF productivities were 4.0 pg/106 cells/day on day 2, 2.9 gg/106 cells/ day on day 7, and 2.4 gg/106 cells/day on day 12.
Effects of pH control Compared with the physiological conditions in the flask, those in perfusion culture system were significantly different. Unless otherwise noted, we assessed these differences from the viewpoint of productivity by semi-continuous culture method centrifuged every day (net culture vol. was 500 ml). DO,-lactate and ammonia, have no effect on the G-CSF productivity of SC57 (data not shown). Accurate control of pH by the addition of 7.5% sodium bicarbonate affected the productivity (Fig. 4). Higher productivity was observed under the pH controlled at 7.4 rather than at 7.0. The specific G-CSF productivity on day 5 was 2.3 ~g/106 cells/day at pH 7.4 and 1.2 ~g/106 cells/
Fig. 4. Effect of control pH on productivity. SC57 cells were inoculated into 500 ml of spinner flask with micro-silicone fibers (Nagayanagi Kogyo Co., Ltd., Tokyo) at the density of about 2 x 106 cells/ml and cultured at 37~ for 9 days with agitation of 40 rpm. pH was measured by an in situ pH probe (Ingold, Urdorf, Switzerland) and controlled varied less than 0.1 pH unit by the addition of 7.5% NaHCO 3 solution. DO was measured by an in situ DO probe (Oriental Electric Co. Ltd., Tokyo) and controlled at 4 ppm by flow rate of air through micro-silicone fibers. Cell counts, samplings and medium changes were done everyday. Each point represents the mean for the duplicate.
day at pH 7.0. In these cases, the cell densities were above 3 x 106 cells/ml.
Effects of nutrients SC57 cells were inoculated into 500 ml spinner flask with micro-silicone fibers at the density of 4 • 106 cells/ml. After cultivation for 24 hours, the conditioned medium was recovered and amino acid contents were analyzed by High Performance Amino Acid Analyzer JLC-200. Amino acid contents of basal RPMI-1640 medium was compared. From the results of this experiment, the deficiency of cysteine (initial concentration; 50 mgB, after cultivation; not detected), serine (initial concentration; 30 mg/l, after cultivation; 1.1 rag/1 ), and glucose (initial concentration; 2 g/l, after cultivation; 0.59 g/1 ) was observed. Addition of 3 g/1 of glucose to the ITPSGF medium did not affect the productivity but did affect the growth rate at high densities above 4 x 106 cells/ml (data not shown).
30 10 8
10 7
a.a. rich m e d i u m 40 -~
E _~
_=~
_=o 30
10
40
In
E
~ ,.:
E~
r 20
50
~ v ,,
co o,. r -- +
r
30 .s
1 0 7.
E
2O
r
10 6 0
1
2
3
4
5
6
(5
7
c u l t u r e period (days)
10
r 10 6
,
U n d e r the semi-continuous batch culture conditions, SC57 was cultured in the I T P S G F m e d i u m with 0.1 m M of cysteine and serine and 3 g/1 o f glucose added with the p H controlled at 7.4 for 6 days (Fig. 5). T h e specific G - C S F productivity o f SC57 was enhanced to 5.7 Ixg/i06 cells/ d a y at p H 7.4 (day 3), which was 1.8 times higher than that o f the control, 3.2 gg/106 cells/day.
High density culture of SC57 using a perfusion culture system Using a perfusion culture system, B i o f e r m e n t e r T M with a dialysis m e m b r a n e and micro-silicone fibers (net culture vol. was 250 ml), SC57 was cultured for 18 days (Fig. 6). At the beginning of the culture, n o r m a l I T P S G F m e d i u m was used, but after day 16, a special m e d i u m in which a m i n o acid and glucose concentration was increased 2 and 2.5 times respectively, was used. T h e specific G - C S F productivity was 2.1 g g / 106 cells/day on day 3, 2.0 ktg/106 cells/day on day 7, 2.6 ~tg/106 cells/day on day 11, 2.7 ktg/106 cells/day on d a y 15, 3.4 gg/106 cells/day on day 17 (the a m o u n t o f G - C S F reached 41 I.tg/ml).
i
2
Fig. 5. Effects of nutrient. SC57 cells were inoculated into 500 ml of spinner flask with micro-silicone fibers at the density of about 4 x 106 cells/ml and cultured at 37~ for 6 days with agitation of 40 rpm. 3 mg/ml of glucose and 0.1 mM of serine and cysteine were added to ITPSGF medium. The controlled pH was 7.4 varied less than 0.1 pH unit. Other conditions were identical to Fig. 4.
n (/3 O
,
I
4
,
i
9
6
Culture
i
8
.
!
,
i
,
1
,
i
,
i
,
0
101214161820 period (days)
Fig. 6. G-CSF production under the optimum conditions. High density culture of SC57 was performed by a perfusion culture system, BiofermenterTM with a dialysis membrane and microsilicone fibers. Net culture volume was 250 ml. SC57 cells were inoculated at the density of 2.5 x 106 cells/ml and cultured at 37~ for 18 days with agitation of 40 rpm. The perfusion rate was set to about 1.0 vol/day and the dialysis rate was shifted up stepwise in proportion to cell density from 1.0 vol/day to 4.0 vol/day. Measurement and control of pH and DO were done under the identical conditions in Fig. 4. Measurement of G-CSF contents were performed by densitometry. Other conditions were identical to Fig. 4.
Discussion H i g h density culture m a k e s it possible to use a smaller vessel to obtain a required a m o u n t o f target proteins. G e n e technology m a y be applied further to simplify the cellular requirements for nutrients, growth factors, and regulatory factors and to increase productivity. H o w e v e r , the physiological conditions of production phase were s o m e w h a t different f r o m those o f the growth phase. Thus, attention to the physiological and b i o c h e m i c a l balance of the growth m e d i u m is needed for the technology of maintaining the specific productivity o f the cells at high level for long culture periods. T o achieve high-level expression o f r e c o m binant proteins, gene amplification using dhfr is a very effective tool. F o r this purpose, Chinese h a m s t e r o v a r y ( C H O ) ceils deficient in dhfr h a v e been successfully used as the host cell line (Kaufm a n and Sharp, 1982; K a u f m a n et al. 1985). But
31 several cell lines that contain active endogenous dhfr genes can be used for gene amplification with dhfr (Dorai and Moore, 1987). We previously showed that the expression level of B-IFN secreted by a recombinant Namalwa KJM-1 was augmented approximately proportionally with increase in cell density in perfusion culture (Miyaji e t al., 1990a) and that a method of dhfr gene coamplification was available in this cell line (Miyaji e t a L , 1990c). An expression vector for G-CSF, pASLB33,was constructed and introduced into Namalwa KJM-1 ceils (Hosoi e t al. , 1988), and cells resistant to 100 nM of methotrexate (MTX) were obtained. Among them, the highest producer, clone SC57, was selected and the productivity of this clone was further characterized. The specific G-CSF productivity of SC57 was depressed to less than 2.5 /.tg/106 cells/day when the cell number was above 7 x 105 cells/ml (Fig. 2). However, the production of G-CSF might be correlated to the cell density, because G-CSF accumulated in large quantities in a perfusion culture system (Fig. 3). These results showed that the specific productivity w a s affected by the physiological conditions, i.e. pH, DO, and the concentrations of nutrients and metabolites. Analyses of productivity in high density culture showed that the limiting factors were the deficiency of nutrients, such as glucose, cysteine, and serine (Fig. 5). The importance of choosing the appropriate medium composition to avoid nutrient limitation, to obtain acceptable high densities in reactor systems, and to maintain high production rate of a target product has been reported for Ltk and BHK cells (Wagner e t al., 1988). Our results showed that the growth medium should be different from a production medium in the amino acid consumption. The specific productivity was depressed at a lower than optimum pH (Fig. 4). It is reported that the pH control results in a significant improvement in the WN production by recombinant CHO cells (Smiley e t al., 1989). Our results showed that the range of controlled pH which gave higher productivity was limited more than that of recombinant E . coli.
The depression of productivity was solved by means of the Biofermenter with micro-silicone fibers and a dialysis membrane in a modified medium in which concentrations of amino acids and glucose were elevated 2 and 2.5 times, respectively (Fig. 6). As a result, the specific productivity was not decreased even at a high cell density, and the amount of G-CSF reached 41 gg/ml. This increase in specific productivity before day !5 was due to shifts in the dialysis rate and that of day 17 was due to a change of medium. These results showed that the high density culture of SC57 was optimized without decreasing of specific productivity. In this paper, we have shown the way to find the optimum culture conditions for the production of G-CSF as an example of recombinant proteins produced by clone SC57, genetically engineered Namalwa KJM-1 cells. Optimum conditions may be different between clones and host cell lines, but it is possible to find the optimum conditions for each cell as shown in this report.
Acknowledgements Our most sincere appreciation is due to the Ministry of Intemational Trade and Industry of Japan, and NEDO (New Energy and Industrial Technology Development Organization) for their financial support and excellent advice. We also thank Ms. Sachiko Shitara and Ms. Sachiko Yokokawa for their excellent technical assistance.
References 1. Dorai H and MooreGP (1987) The effectof dihydrofolate reductase-mediatedgene amplificationon the expressionof transfectedimmunoglobulingenes.J. Immunol.139:42324241. 2. HosoiS, Mioh H, AnzaiC, Sato S and FujiyoshiN (1988) Establishmentof Namalwacell lines which grow continuously in glutamine-freemedium. Cytotechnology1: 151158. 3. KaufmanRJ and Sharp PA (1982) Amplificationand expression of sequences cotransfectedwith a modular dihydrofolate reductase complementaryDNA gene. J. Mol. Biol. 159: 601-621.
32 4. Kaufman RJ, Wasley LC, Spiliotes AJ, Gossels SD, Latt SA, Larsen GR and Kay RM (1985) Coamplification and coexpression of human tissue-type plasminogen activator and murine dihydrofolate reductase sequences in Chinese hamster ovary cells. Moh Cell. Biol. 5: 1750-1759. 5. Komatsu Y, Matsumoto T, Kuga T, Nishi T, Sekine S, Saito A, Okabe M, Morimoto M, Itoh S, Okabe T and Takaku F (1987) Cloning of granulocyte colony-stimulatirig factor cDNA from human macrophages and its expression in Escherichia coli. Jpn. J. Cancer Res. 78: 11791181. 6. Kuga T, Komatsu Y, Mizukami T, Sato M and Itoh S (1990) Effect of N-terminal deletion of signal peptide on the secretion of human granulocyte colony-stimulating factor in mammalian cells. Biotechnoh Lett. 12: 87-92. 7. Laemmli UK (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature. 227: 680-685. 8. Miyaji H, Mizukami T, Hosoi S, Sato S, Fujiyoshi N and Itoh S (1990a) Expression of human beta-interferon in Namalwa cells which were adapted to serum-free medium. Cytotechnology 3: 133-140. 9. Miyaji H, Harada N, Mizukami T, Sato S, Fujiyoshi N and Itoh S (1990b) Expression of human lymphotoxin in Namalwa ceils which were adapted to serum-free medium. Cytotechnology 4: 39--43. 10. Miyaji H, Harada N, Mizukami T, Sato S, Fujiyoshi N and Itoh S (1990c) Efficient expression of human beta-interferon in Namalwa cells adapted to serum-free medium by a dhfr gene coamplification method. Cytotechnology 4: 173180. 11. Nagata S, Tsuchiya M, Asano S, Kaziro Y, Yamazaki T, Yamamoto O, Hirata Y, Kubota N, Oheda M, Nomura H and Ono M (1986) Molecular cloning and expression of cDNA for human granulocyte colony-stimulating factor. Nature 319: 415--418.
12. Nishi T, Saito A, Oka T, Itoh S, Tagaoka C and Taniguchi T (1984) Construction of plasmid expression vectors carry{ng the Escherichia coli tryptophan promoter. Agric. Biol. Chem. 48: 669-675. 13. Oheda M, Hase S, Ono M and Ikenaka T (1988) Structure of the sugar chains of recombinant human granulocytecolony-stimulating factor produced by chinese hamster ovary cells. J. Biochem. 103: 544-546. 14. Sato S, Kawamura K and Fujiyoshi N (1983) Animal cell cultivation for production of biological substances with a novel perfusion culture apparatus. J. Tissue Culture Methods 8: 167-171. 15. Satoh M, Hosoi S and Sato S. CHO (1990) (Chinese Hamster Ovary) cells spontaneously secret a Cysteine Endopeptidase. In Vitro Cell. Dev. Biol. 26:1101-1104. 16. Seiki M, Hattori S, Hirayama Y and Yoshida M (1983) Human adult T-cell leukemia virus: Complete nucleotide sequence of the provirus genome integrated in leukemia cell DNA. Proc. Natl. Acad. Sci. USA. 80: 3618-3622. 17. Smiley AL, Hu W-S and Wang DIC (1989) Production of Human Immune Interferon by recombinant Mammalian cells cultivated on Microcarriers. Biotechnol. Bioeng. 33: 1182-1190. 18. Wagner R, Ryll T, Krafft H and Lehmann J (1988) Variation of amino acid concentration in the medium of HU-IFN and HU IL-2 producing cell line. Cytotechnology 1: 145150. 19. Yoshida H and Shitara S (1989) Generation and Characterization of Monoclonal antibodies to Recombinant Human Granulocyte-colony Stimulating Factor (G-CSF) and Its Mutein. Agric. Biol. Chem. 53: 1095-1101.
Address for offprints: Shinji Hosoi, Tokyo Research Laboratories, Kyowa Hakko Kogyo Co. Ltd., 3-6-6 Asahi-machi, Machida-shi, Tokyo 194, Japan
