shake flasks for the production of recombinant proteins ... shake flasks can successfully support the growth of in- .... scaling up and productivity of BEVS. To test ...
Appl Microbiol Biotechnol(1992) 37:74-78
Applied Microbiology Biotechnology © Springer-Verlag 1992
Optimization of protein-production by the baculovirus expression vector system in shake flasks Ronen Neutra, Ben-Zion Levi, and Yuval Shoham Department of Food Engineeringand Biotechnology,The Technion, Technion City, Haifa 32000 Israel Received 26 July 1991/Accepted 19 November 1991
Summary. Shake flasks were successfully employed for the cultivation of Spodoptera frugiperda (Sf-9) insect cells and for the production of fl-galactosidase, a recombinant model protein, utilizing the baculovirus expression vector system. The culture doubling time and maximal cell density were 20 h and 5 x 106 cells/ml respectively. The optimal liquid volumes for flasks rotating at 100 rpm were 25-40% of the flask total volume. Enzyme production (about 600 mg/l) was best at a multiplicity of infection of between 1 and 20 and at a cell density at time of infection of 0.7 x 10 6 cells/ml. At a rotation speed of 100 rpm, Pluronic F-68 had no effect on growth and enzyme production.
Introduction The baculovirus expression vector system (BEVS) is widely applicable as an alternative to prokaryotic or other eukaryotic systems for the expression of heterologous proteins (Luckow and Summers 1988). A variety of recombinant proteins produced by the system was recently reviewed by Luckow (1990). These proteins are usually produced by replacing the viral polyhedrin gene with a foreign gene of interest. The polyhedrin structural gene is a non-essential late gene that is highly expressed in infected insect cells (25-75% of total cell protein) but is not essential for viral propagation in tissue-cultured insect cells (Smith et al. 1983). Post-transcriptional and post-translational modifications such as: RNA-splicing, glycosylation, phosphorylation, assembly of multimeric proteins and signal recognition have been observed in BEVS (Pennock and Miller 1984). Moreover, in some cases a high level of expression of a certain recombinant proteins was achieved only by BEVS (Levi and Ozato 1990). Yields of recombinant protein produced via BEVS, although varying widely, can exceed 600 mg/1 (Maiorella and Harano 1988). The level of protein production may be
Offprint requests to: Y. Shoham
influenced by various parameters such as: viral infection, cell culture and physical conditions. Effective production of baculoviruses and recombinant proteins can be achieved using insect larvae, but this process is labour intensive, time consuming and difficult to automate (Shapiro 1986). Another way of producing the viral or recombinant proteins is to use suspension cultures of insect cells. Batch-wise and semi-continuous cultures were used to produce the Autoorapha californica nuclear polyhedrosis virus (AcNPV) in insect cells (Weiss and Vaughn 1986). Recently a continuous production of the baculovirus in a cascade of insect cell reactors was also demonstrated (Kompier et al. 1988; van Lier et al. 1990). Most of the work with insect cells in suspension culture currently utilizes mammalian-cell bioreactors or spinner flasks. This kind of equipment is expensive and is not common in every laboratory. In principle, shake flasks can be an excellent solution for growing insect cells since these cells can readily grow in suspension without the need of a CO2 atmosphere. However, there is currently no systematic study reporting optimal growth condition and production of recombinant proteins with BEVS in shake flasks. The use of simple shake flasks for the production of recombinant proteins via BEVS would make this procedure more accessible to laboratories that lack standard mammalian culture equipment. In this communication we demonstrate that shake flasks can successfully support the growth of insect cells. Optimization of the the production of fl-galactosidase, a model recombinant protein, was studied utilizing BEVS in shake flasks.
Materials and methods Cells and media. Spodoptera frugiperda (Sf-9) cells (ATCC) were maintained in TNM-FH medium (Summers and Smith 1987). TNM-FH mediumwas prepared from Grace's powdered medium (Gibco, Grand Island, NY, USA), Yeastolate and lactoalbumin hydrolysate(Difco, Detroit, Mich., USA) and sterilized by filtration (0.2-~tm filter dual membrane, Sigma, St. Louis, Mo., USA). The mediumwas supplementedwith 10% heat-inactivated(30 min
75 at 56° C) foetal calf serum (FCS) (Biological Industries, Kibbutz Bet-Haemek, Israel), 50 txg/ml of gentamicin sulphate and 2.5 p.g/ ml of fungizone (Biological Industries). The pH of the medium was 6.2+0.1. Cultures were grown as a monolayer in stationary 80-cm tissue culture flasks or 100-mm tissue culture dishes in a 28 ° C incubator. Cell passage was performed by flushing the monolayer off the surface of the flask or dish with a policeman. Growth in shake flasks (model no. 4980, Pyrex, Coming Glass Works, Corning, NY, USA) was carded out in a 27° C room on an Orbit Shaker (Lab-Line, Melrose Park, IL, USA) at 100 rpm.
Virus stock and infection. Recombinant vires was produced by cotransfecting Sf-9 cells with Auto#rapha californica nuclear polyhedrosis virus (AcMNPV) and plasmid pAc360-fl-gal (kindly obtained from Dr. Summers, Texas A&M University, TX, USA) using the calcium-phosphate precipitation technique (Summers and Smith 1987) with modifications (Levi and Ozato 1990). Recombinant viruses containing the lacZ gene were detected and purified either by end-point dilution assay (three rounds) using o-nitrophenyl-fl-D-galactopyranoside (ONPG) as a substrate or by plaque assay (two rounds) using 5-bromo-4-chloro-3-indolyl-fl-D-galactopyronoside (X-gal) in soft agar plates. The virus stock titre was determined to be approximately 7 × 108 plaque-forming units/ml using the end-point dilution assay (Summers and Smith 1987). Infection in shake flasks was made by introducing viruses from the viral stock to the desired multiplicity of infection (MOI). Cell growth and enzyme measurement. Cell growth and enzyme production were monitored daily by removing 0.5-ml samples from each flask. Cell number was measured immediately using an improved Neubauer Brightline (Herenz, Hamburg, FRG) haemocytometer. The results were the average of two independent counts of at least 200 cells each. Viability was determined using the trypan blue dye (0.04%, Sigma, St. Louis, Mo., USA) exclusion method, fl-Galactosidase activity was assayed by the procedure of Miller (1972) using ONPG as the substrate. The samples were treated by three cycles of freezing and thawing followed by centrifugation at 13 000 O for 5 min to remove cell debris. Ten microlitres of the diluted samples (1 : 100 in phosphate-buffered saline) were then mixed with 0.99 ml Z buffer (60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KC1, 1 mM MgSO4, 50 mM fl-mercaptoethanol) and 0.2 ml ONPG (4 mg/ml). Following incubation for 30 min at 30 ° C the reactions were terminated by the addition of 0.5 ml of 0.5 M Na2CO3. The absorbance at 420 nm was measured and the number of moles of ONPG hydrolysed was calculated assuming a molar extinction coefficient for nitrophenyl of 4860 at pH 10.0 (Schlief and Wensink 1981). One unit of activity is defined as the amount of enzyme that hydrolyses 1 Ixmol ONPG.min-a. The oxygen transfer rate (OTR) in shake flasks was measured using the sulphite method (Wang et al. 1979).
Results
Table 1. Optimal volume of media for growth of Spodopterafruoi-
perda (Sf-9) cells in different-sized shake flasks Shake flask volume (ml)
Optimal liquid volume (ml)
Flask volume (%)
Maximal cell density (cells. ml- l)a
125 250 500
40 90 120-150
32 36 24-42
2.9 x 106 3.9 × 10 6 2.9 x 106
a Growth was carded out at 100 rpm and 28°C 60
50
~ ~
o~
30
~
~
~o
N ~0
0 0
50
100
Culture Time
150
200
[Hours]
Fig. 1. Seeding density effect on Spodopterafruoiperda (Sf-9) cell growth kinetics in 125 ml-shake flasks. Seeding densities were (in cells.ml-1): rq, 5 x 104; ill, 0.8 × ]05; O, ] × 106; O, 2.2 x 106
(8 ml) o f m e d i u m gave p o o r growth. W h e n 60 or 80 ml m e d i u m was used at 1 0 0 r p m , the mixing was not strong e n o u g h to prevent cells f r o m sedimenting to the b o t t o m o f the flask. Thus, g r o w t h was inhibited due to p o o r mixing. With 8 ml m e d i u m , no g r o w t h was detected a n d cell viability d r o p p e d p r e s u m a b l y due to excess shear forces. Similar results were o b t a i n e d with different sizes o f shake flasks (Table 1). The optimal v o l u m e o f m e d i u m at 100 r p m was b e t w e e n 25-40% o f the flask volume, in w h i c h cell densities r e a c h e d 3 5 × 106 cells/ml.
Growth kinetics o f Sf-9 cells in shake flasks Seeding density effect on cell growth Preliminary g r o w t h experiments with Sf-9 cells indicated that the cells can successfully g r o w in shake flasks. To optimize the v o l u m e o f liquid that can be used for g r o w i n g Sf-9 cells in shake flasks, different volumes o f m e d i u m were used u n d e r the s t a n d a r d g r o w t h conditions (28 ° C; 100 r p m ; T N M - F H m e d i u m containing 10% F C S a n d antibiotics). The best results in 125-ml shake flasks were o b t a i n e d with 40 ml med i u m w h e r e cell density r e a c h e d 3 x 1 0 6 c e l l s . m l -~ with a d o u b l i n g time o f 20 h. With 20 ml m e d i u m the g r o w t h rate a n d cell density were s o m e w h a t lower. Either larger v o l u m e s (60, 80 ml) or a smaller v o l u m e
The initial cell density has i m p o r t a n t implications for scaling u p a n d productivity o f BEVS. To test the seeding density effect on cell growth, shake flasks (125 ml) were seeded with various c o n c e n t r a t i o n s (0.12-2.2 x 106 cells, m l - 1 ) o f Sf-9 cells. Optimal g r o w t h rate a n d final cell density were o b t a i n e d with cultures seeded at 0.82.2 x 10 6 cells.m1-1 (Fig. 1). Cultures seeded at 5 x 105 cells, m l - 1 r e a c h e d a density o f a b o u t 4 x 106 whereas, w h e n the seeding density was 0.05 x 106 c e l l s - m l - 1 the cultures grew p o o r l y reaching a cell density o f only 1 x 106 cells.m1-1.
76 6O 50
I
'
'
''
'
I
"
'
• '"'
T
5O ~" ~
~,~
,~
40
40
. ~
~
30
"OE
o
_~ ~~ ~
~0
30
20, m
10
> t
0
E
~
I
i
10
|
B
B
E c 200 .=.
200
2 ~oo
O
100
0 50
100
150
Culture Time [hours]
0
50 Culture Time
100 [hours]
Fig. 2A, B. Effect of multiplicity of infection (MOI) on Sf-9 cell growth and on fl-galactosidase production. Cell were seeded at 2 x 106 cells, ml-1 and inoculated at an MOI of: ~,, 0 (no virus); A, 0.1; O, 0.5; ©, 1; B, 5; rq, 20. Cell growth (A) and fl-galactosidase activity (B) were monitored daily
Fig. 3. The effect of infection time on Sf-9 cell growth and on fl-galactosidase production. Cells were seeded at 4 x 105 cells, m1-1 in 125-ml shake flasks and infected with the recombinant virus at: &, no infection; rq, 0; A, 24; O, 48; O, 72h. Cell growth (A) and fl-galactosidase activity (B) were monitored daily
The effect of MOI on fl-galactosidase production
cell densities of 1-5 x 106 cells.m1-1 is reported with suspension cell culture (Summers and Smith 1987). The optimal time of infection for fl-galactosidase production in shake flasks was determined as follows: Sf-9 cells were grown in 125-ml shake flasks at an initial cell density of 4 x 105 cells.m1-1, and each day duplicate flasks were infected with the recombinant virus at an MOI of 1. As expected, growth was inversely correlated to the time of infection (Fig. 3A). The earlier the infection, the lower was the extent of growth. Infection after 24 h at a cell density of 7.5 x 105 cells.m1-1 (early exponential growth phase) resulted in the highest enzyme activity (239 units .m1-1, Fig. 3B), whereas infection at 72 h (late exponential growth phase) resulted in only residual activity (11 units, m l - 1, Fig. 3B).
MOI values of 10 or greater is known to give a fairly synchronous infection (Summers and Smith 1987). Since it is desirable to use a minimal amount of virus for infection, we tested the effect of different MOI values on growth kinetics and fl-galactosidase production. Sf-9 cells were grown to mid-exponential phase in 125ml shake flasks. Infection was carried out at different MOI values (0.1 to 20) and cell densities as well as flgalactosidase production were monitored daily (Fig. 2). An inverse correlation between MOI and growth yield was found. Increase in cell number was recorded with a MOI smaller than 1, whereas at a MOI larger than 1, a decrease in cell viability was detected from day one post-infection. The highest fl-galactosidase activity was at 100 to 120 h post-infection with an MOI of 5.
Effect of Pluronic F-68 on cell 9rowth and protein production Time of infection Time of infection is an important parameter for obtaining high yields of recombinant proteins. Infection at
Pluronic F-68 is a non-ionic block copolymer of poly(oxypropylene) and poly(oxyethylene) with an average molecular mass of 8400 Da. Pluronic F-68 was
77 shown to protect insect cells (Murhammer and Goochee 1988, 1990) and hybridoma cells (Gardner et al. 1990) from the adverse effects of sparging and stirring in small-scale bioreactors. Under our standard growth conditions (40 ml in 125-ml shake flasks, 100 rpm) Pluronic F-68 (0.05% and 0.5%) had no effect on cell growth and enzyme production. However, at a higher rotation speed (150rpm) Pluronic F-68 (0.05%) improved growth (data not shown).
Discussion
In this study we optimized several parameters for the production of a model recombinant protein using BEVS in shake flasks. We first considered the optimal volume of medium in the shake flask that will support growth. Liquid volumes of 25-45% of the flask volume was found to be optimal for Sf-9 cell growth. The amount of liquid in the flask has several implications for the cells e.g. mixing, shear forces and overall oxygen transfer. For a given rotation speed the volume of the liquid will determine both mixing and shear forces. The larger the volume, the lower the mixing and shear forces. At 100 rpm, we found that liquid volumes that exceeded 50% of the total flask volume resulted in the sedimentation and accumulation of cells at the bottom of the flasks. At liquid volumes lower than 10% of the total flask volume, cell growth was severely inhibited, presumably as a result of shear forces. Animal cells are known to be extremely sensitive to shear forces. Tramper et al. (1986) investigated the effect of shear on the viability of insect cells and showed that at 14 N m -2 cell viability decreased. Similar effects were found by Schurch et al. (1988) for CRL-8018 hybidoma cells. Our data suggest that mixing rather than oxygen transfer is the main limitation for working with large volumes in shake flasks. We did not observe significant differences in cell growth (extent and doubling time) when utilizing different amounts of media in the the range 15-40% of total volume. The OTR measurements (see Materials and methods) in 125-ml shake flasks containing 40ml water gave values of about 10 mmol. 1-1. h - 1. The specific oxygen uptake of animal cells falls in the range of 0.05-0.5 mmol. 1-1. h-1 per 106 cells (Prokop and Rosenberg 1989). For insect cells a specific oxygen uptake rate of 0.17 m m o l . l - l . h -1 per 106 cells was obtained (Maiorella and Harano 1988). Therefore it is clear that even at cell densities of 5 × 106 cells, ml-1 oxgen transfer should not be the limiting factor in shake flasks. Caron et al. (1990) and Maiorella and Harano (1988) reported that in spinner flasks (100-3000ml) oxygen transfer may limit the growth of insect cells. In their studies they noticed that reducing the culture volume improve the final cell density. In our hands, the doubling time (usually 18-24 h) and the extent of growth of insect cells in shake flasks (2-5x 106 cells.ml-1), were comparable to those obtained with spinner flasks or bioreactors (Wu et al.
1989). It should be noted, however, that cell growth is highly dependent on factors such as: routine cell passage, culture age, and the age of the medium. These factors can influence dramatically the reproducibility of the results. The results of the effect of seeding density on cell growth in shake flasks indicate that an inoculum of 820% was most suitable. At a low seeding density (
