Kinetics, Experimental and Simulation Studies of Chinese Hamster Ovary Cell Growth in a Packed-Bed Bioreactor
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
World Applied Sciences Journal 15 (11): 1568-1575, 2011
ISSN 1818-4952
© IDOSI Publications, 2011
Kinetics, Experimental and Simulation Studies of
Chinese Hamster Ovary Cell Growth in a Packed-Bed Bioreactor
Mehdi Shakibaie, 1Fatemeh Tabandeh, 2Ali Reza Zomorodipour,
1,4
3
Hossein Mohammad-Beigi, 4Sirus Ebrahimi and 1Hasan Habib-Ghomi
Department of Industrial and Environmental Biotechnology,
1
National Institute of Genetic Engineering and Biotechnology, Tehran 14178-63171, Iran
2
Department of Molecular Genetics,
National Institute of Genetic Engineering and Biotechnology, Tehran 14178-63171, Iran
3
Department of Chemical and Petroleum Engineering,
Sharif University of Technology, Tehran 11365-11155, Iran
4
Department of Chemical Engineering,
Sahand University of Technology, Tabriz, 53317-11111, Iran
Abstract: To figure out the relationship between nutrient deprivation and cell growth, simulation can be
effective. In the present work, the growth of Chinese Hamster Ovary (CHO) cells was simulated based on
stoichiometric and kinetics calculations. The simulation results were compared to the experimental data from
a packed-bed bioreactor and a good agreement was observed. The kinetic parameters (KGlc, KAmm, KLact and µmax)
were optimized by a genetic algorithm method using stoichiometric parameters (YX/Glc, YX/Amm, and YLac/Glc). The
stoichiometric and kinetic parameters were used in the simulation to study the growth of CHO cells. The
concentrations of the toxic by-products, ammonium and lactate, were measured at different values of dissolved
oxygen, 30% and 50% of saturated air. While dissolved oxygen was maintained at 30% of saturation, the
maximum that ammonium and lactate concentrations reached were 0.099 and 2.1 g/l, respectively. At dissolved
oxygen level of 50% air saturation, the maximum ammonium and lactate concentrations decreased to the values
of 0.082 and 1.8 g/l, respectively. The cell density of 22×108 cells/l was achieved after 120 h of cultivation at
dissolved oxygen of 50% in a packed-bed bioreactor.
Key words: Packed-bed bioreactor CHO cells Stoichiometry Kinetics Simulation
INTRODUCTION In mammalian cell cultures, cell growth and
productivity can be improved by optimizing cell culture
Development and optimization of cell culture process conditions. The lack of available nutrients, especially
is necessary to meet the demand for mammalian cells. The glucose has been cited as a potential inducer of apoptosis
scale-up and optimization of mammalian cell cultures is in cell cultures [9-11]. Accumulation of toxic by-products
influenced by many factors [1-3]. Since the animal cells such as ammonium and lactate are the main barriers to
have low productivity, a large volume of culture media is longevity of batch cultures. High levels of ammonium and
required to produce therapeutic proteins. Packed-bed lactate inhibit the CHO cell growth and productivity
bioreactors can meet the criteria as high cell density and [12-16].
productivity can be attained under low-volume conditions A mathematical model has been developed to predict
[4,5]. High cell density cultures have been developed with the CHO cell growth according to the concentration of
the aim of improving the volumetric productivity as it is main nutrients and by-products [17]. Asymmetric logistic
proportional to both the final cell density and specific equations have been used to predict nutrient deprivation
productivity, i.e. the amount of product formed per unit and monoclonal antibody production in hybridoma T-
cell mass per unit time [6-8]. flask cultures [18]. Liu et al., studied the bioreactor
Corresponding Author: Fatemeh Tabandeh, Department of Industrial and Environmental Biotechnology,
National Institute of Genetic Engineering and Biotechnology, Tehran 14178-63171, Iran.
Tel: +98 2144580359, Fax: +98 2144580399.
1568World Appl. Sci. J., 15 (11): 1568-1575, 2011
behavior by using kinetic models which enabled the In order to study the effect of dissolved oxygen (DO) on
estimation for any given set of operating conditions. A toxic by-products formation, DO content was adjusted to
study was carried out on using modeling techniques to 30% and 50% of saturated air as a manipulated variable by
determine an optimal time point for the addition of the bioreactor control program. The polyester disks (134
effective additives to promote the productivity in the CHO g) purchased from New Brunswick Scientific Co., were
cell cultures [19]. Oxygen solubility in large-scale used as microcarriers for the attachment of CHO cells
industrial bioreactors was known as one of the key present in the packed-bed bioreactor. The polyesters were
parameters. The demand for oxygen is different depending removed from the bioreactor and the CHO cells were
on cell and from the supply point, the critical parameter is separated by EDTA-Trypsin solution. The exposure time
the volumetric transfer coefficient [20,21]. for cell removal is 5 min at 37°C. After cells were removed
The present work studied how the CHO cell growth from surfaces, serum was added to the culture. Harvested
is influenced by the concentration of main nutrients. CHO cells were centrifuged at 1000 rpm for 5 min at 4°C.
cells are widely used in biopharmaceutical industry as
they can tolerate environmental stress which made them Sample Analysis: Cell viability was determined by the
known as stable hosts. The comparison of a simulation Trypan blue exclusion method. Both grids of a neubauer
prediction with actual behavior usually leads to an haemocytometer slide were loaded with the cell
increased understanding of the process. Oxygen supply, suspension and microscopic cell counts were performed
the key parameter for industrial-scale bioreactors, was on four large squares of each grid.
measured in a packed-bed bioreactor. We implemented the Glucose, ammonium and lactate concentrations were
genetic algorithm and simulator to predict the cell density enzymatically measured by glucose, ammonium and
and nutrient exhausting using logistic equations. Through lactate assay kits (ChemEnzyme Co., Iran).
that, the stoichiometric and kinetic parameters were
measured using both experimental data and genetic
Kinetic Model of CHO Cultivation: For batch cultivation
algorithm. These parameters were inserted into the
process, the mass balance equations for CHO cell growth,
simulation model to study the effect of nutrient
substrate consumption and product formation are as
deprivation and by-product formation on the cell growth
follows:
in a packed-bed bioreactor.
dX (1)
MATERIALS AND METHODS = rx
dt
Cell Line and Culture Media: The CHO cells were dCGlc
= − rGlc (2)
obtained from the National Institute of Genetic dt
Engineering and Biotechnology (NIGEB). The CHO cells
were cultured in the basal medium consisting of DMEM
dC Lac (3)
(Gibco) and Ham’s F12 (Gibco) at a volume ratio of 1:1 and = rLac
dt
10% fetal bovine serum (Gibco) in a humidified incubator
dC Amm (4)
at 37°C and 5% CO2. Each instance was put into a T-75 = ramm
dt
flask. Each set of T-flasks were inoculated from a common
seed culture and 12 ml of the basal media.
Where X is the CHO concentration (108 cells/l), CGlc is the
Inoculum Preparation: Cells were removed from the concentration of glucose (g/l) and CAmm and CLact are
surface of T-75 flasks using the EDTA-trypsin solution ammonium and lactate concentrations (g/l), respectively.
(Gibco) and inoculated into the microcarrier culture Experimental data showed that product inhibitors are
system. The initial cell density for inoculums was about of importance for the cultivation of CHO [16]. In this
4×108 cells/l. The CHO cells must be in the growth phase study, the cell growth rate, rx, is given by:
to shorten the lag phase.
CGlc (5)
rx = max X
(CGlc + KGlc ).K I
Bioreactor: The batch bioreactor experiment was
performed in a 4 l Celligen bioreactor (New Brunswick
C C (6)
Scientific, USA) with a working volume of 3.8l at KI =( Lact + 1)( Amm + 1)
K Lac K Amm
37°C, pH of 7.2 and with an agitation rate of 50-100 rpm.
1569World Appl. Sci. J., 15 (11): 1568-1575, 2011
Where µmax is the maximum specific growth rate (h 1), KGlc Simulation: The simulation model was carried out by a
is the Monod constant for glucose and KLact and KAmm are simulator computer program. The initial concentrations of
the parameters used to describe product inhibitors. KI is glucose, lactate and ammonium were 3.7, 0.006, 0.001 g/l
the total inhibitory constant. and the initial cell density considered for driving the
The substrate consumption and product formation simulation was 4×108 cells/l.
rates can be calculated by:
RESULTS AND DISCUSSION
1 (7)
rGlc = − rx
YX / Glc
It is difficult to measure the cell density in a packed-
rLac = −YLac / Glc rGlc (8) bed bioreactor during the cultivation since the cells have
been immobilized onto microcarriers. Thus, the data on
1 (9) yield coefficients from the T-75 flask experiment were
rammo = rx
YX / Amm
applied to the simulation model for bioreactor. Table 1
provides a summary of the yield coefficients from the T-75
Where YX/Glc is the glucose yield coefficient (108 cells/g), flask monolayer cultivation.
YLac/Glc is the lactate to glucose yield coefficients (g/g) and The kinetic parameters were estimated using the
YX/Amm is the yield of ammonium based on cell growth (108 genetic algorithm. The optimized parameters (KGlc, KAmm,
cells/g). KLact and µmax) are shown in Table 2. Figure 1 depicts the
According to the above descriptions, there are three computed profiles based on genetic algorithm for cell,
parameters (YX/Glc, YLac/Glc and YX/Amm) estimated from the glucose, lactate and ammonium under such an optimal
experimental data and four parameters determined by model. In order to characterize the quality of the
minimizing Eq 10 using optimization method (genetic prediction model, the residual standard deviation (RSD)
algorithm). The objective of optimization is to minimize the was determined [23].
objective function, :
RSD
RSD(%) = (14)
n X(i) − Xe(i ) CGlc(i) − CeGlc(i) Clac(i) − CeLac(i) CAmm(i) − CeAmm(i) y
=∑ + + +
Xei CeGlc(i) CeLac(i) CeAmm(i) Where RSD ∑ ( y − x ) , xi is an experimental value and yi
2
i=1 1 n
= (%) i i
n
(10) is a predicted value, y is the average of experimental
i
values and n is the number of experimental points. The
where Xe, CeGlc, CeLac and CeAmm are the measured RSD (%) values are 3.4, 8.3, 8.2 and 6.9 for glucose,
concentrations of cell, glucose, lactate and ammonium at lactate, ammonium and biomass concentrations,
a given sampling time (i). X, CGlc, CLac and CAmm are the respectively. All the RSD (%) values were below 10%
concentrations computed by the model at a given indicating that the prediction model terms are significant
sampling time (i). [22].
Determination of Kla by a Dynamic Method: Table 1: Kinetic and stoichiometric parameters of CHO cells in T-75 flasks
Volumetric mass transfer coefficient (kLa) was Parameter Unit Value
calculated by a dynamic method using the following YX/Glc 108cells/g 4.2
equations: YX/Lac 108cells/g 8.9
Yx/Amm 108cells/g 230
dCL (11) YLac/Glc g/g 0.5
= OTR − OUR
dt
OTR k L a (CL* − CL )
= (12) Table 2: Kinetic parameters for the proposed simulation from genetic
algorithm
OUR = qo2 X (13) Parameter Unit Value
KiLact g/l 1.968
Where CL* is the maximum oxygen concentration in the KiAmm g/l 0.173
liquid, CL is the oxygen concentration in the liquid, OUR µmax 1/h 0.090
is the oxygen uptake rate and qO2 is the specific oxygen KGlc g/l 3.457
respiration rate.
1570World Appl. Sci. J., 15 (11): 1568-1575, 2011
(a) (b)
(c) (d)
Fig. 1: Experimental (closed diamond) and simulated data (represented by line) for (a) CHO cells density (b) Glucose
consumption, (c) Ammonium formation, (d) Lactate formation.
(a) (b)
(c)
Fig. 2: Profile concentrations in packed-bed bioreactor at different amount of dissolved oxygen. (a) Glucose
consumption, (b) Ammonium formation, (c) Lactate formation, (closed square) DO 30%, (closed triangle) DO 50%.
1571World Appl. Sci. J., 15 (11): 1568-1575, 2011
(a) (b)
(c)
Fig. 3: Illustration of the ability of simulation to fit experimental data. Experimental data for all experiments are compared
with simulation: (a) Glucose consumption, (b) Ammonium formation, (c) Lactate formation at different of dissolved
oxygen. (closed square) DO 30%, (closed triangle) DO 50%.
These optimized parameters were used in the The profile of glucose consumption in the
simulation. Glutamine is not essential for CHO cells packed-bed bioreactor is shown in (Fig. 2a). The
having the biosynthetic glutamine synthetase. The concentration of glucose decreased to the values of 0.3
glutamine catalyzes the following reaction [23]: g/l at 50% DO and 0.21 g/l at 30% DO by the end of
cultivation, whereas a reduction to 0.56 g/l was
L-glutamate + NH4+ +ATP _ L-glutamine + ADP + Pi observed using the simulation program after 120 h
+ H+ cultivation. The simulation results show a relatively
good agreement with the experimental data in terms of
Thus, glutamine was not taken as the limiting glucose consumption (Fig. 3a). No significant difference
substrate in the simulation. The CHO cell density reached was observed in glucose utilization when DO was
22×108 cells/l in the packed-bed bioreactor after 120 h switched from 30% to 50%. There are lots of biochemical
cultivation. This value was 26×108 cells /l based on the reactions in the metabolic pathways consuming cell
simulation results for the same culturing period. Without energy to mitigate the adverse effects of by-products
taking into account the negative effect of by-products in formation. For instance, the mechanism of alanine
Monod equation, the cell density was computed as 30×108 secretion to the medium by CHO cells reduces
cells/l at 80 h of cultivation based on the simulation. ammonium toxicity. The reaction that occurs between
This result indicated that the accumulation of by-products glutamate and pyruvate leads to the formation of alanine
had a great impact on cell density. It may be due to the and decreases the interacellular pyruvate level [24]. The
reduced culture longevity because of the formation of other reaction that needs ATP is the conversion of
toxic by-products. glutamate to glutamine which also consumes ammonium.
1572World Appl. Sci. J., 15 (11): 1568-1575, 2011
Fig. 4: Metabolic pathways in CHO cells. Gln is glutamine; Glu is glutamate; Pyr is pyruvate; KG is ketoglutarate; Mal
is malate; Glc is glucose; Glc-6P is glucose-6-phosphate; Lac is lactate; Cit is citrate; OAX is oxaloacetate; Ac-
CoA is acetyl-CoA.
Since the simulation model excluded all these reactions data and simulation results were comparable in terms of
happening in CHO cells by means of glucose and ATP lactate concentration, whereas for ammonium
consumption, Slightly higher glucose consumption concentration a deviation observed in simulation results
observed in the bioreactor compare to the cell growth f rom the experimental data especially at lower DO
model might be due to the fact that all these reactions (Fig. 3a,b,c).
happening in CHO cells by means of glucose and ATP High oxygen concentration usually has adverse
consumption were excluded from the simulation model. effects on mammalian cell growth. However, since oxygen
The concentrations of ammonium and lactate were limitations within the microcarriers is critical, for
measured as the main by-products to provide more immobilized cells high levels of oxygen is nontoxic and
information on cell behavior. Fig 2b shows the data on can improve the performance of fixed-bed bioreactors
ammonium formation from experimental runs in the packed [21,22]. To determine the effect of dissolved oxygen on
bed bioreactor. Ammonium concentration was below the by-products formation, the packed-bed bioreactor with
inhibitory level. The inhibitory concentrations that have common seed was adjusted at two different values of DO.
been reported from previous studies for ammonium and The 20% reduction in DO led to an increase in ammonium
lactate are about 10 and 18 mM, respectively. The and lactate production from 0.08 and 1.9 at 30% DO to
ammonium concentration was 0.08 g/l at 50% DO and 0.0999 and 2.1 g/l at 50% DO, respectively. Major
0.099 g/l at 30% DO after 120 h cultivation in the packed- metabolic pathways in CHO cells including those of by-
bed bioreactor. The ammonium concentration reported for products formation and glucose consumption are shown
simulation program was 0.088 g/l after the same period of in Fig 4. Glucose and glutamine are the major sources of
time. Lactate concentration reached 1.9 g/l at 50% DO and carbon and energy in animal cell cultures. Increase in
2.1 g/l at 30% DO in the packed-bed bioreactor, whereas lactate formation at lower DO is because lactate forms
the computed value using the simulation program was 2.4 under anaerobic conditions, i.e. oxygen limitations. More
g/l (Fig 2c). Since detoxification of lactate occurs at low energy produced through oxidative pathway (38 mol ATP)
glucose concentrations [16], CHO cells can consume compare to oxygen limitation conditions (2 mol ATP),
lactate as an alternative carbon source. The experimental which is the reason behind higher glucose consumption
1573World Appl. Sci. J., 15 (11): 1568-1575, 2011
at high level of DO. Furthermore, the increased level of 5. Chen, J.P. and C.T. Lin, 2006. Dynamic Seeding and
ammonium might be related to the mechanism through Perfusion Culture of Hepatocytes with
which glutamine consumes to furnish the energy deficit Galactosylated Vegetable Sponge in Packed-Bed
(Fig. 4). Bioreactor. J. Biosci. Bioeng., 102: 41-45.
The volumetric mass transfer coefficient (kLa) is 6. Oh, D.J., S. Kyo Choi and H. Nam Changt, 1994.
difficult to predict, although it can be measured and the High-Density Continuous Cultures of Hybridoma
value reported for 120 h of cultivation was 8.006 (1/h). The Cells in a Depth Filter Perfusion System. Biotechnol.
specific respiration rate was calculated by equation (13). Bioeng., 44: 895-901.
According to the cell density this amount was 0.4 mmol 7. Chen, Z., D. Lütkemeyer, K. Iding and J. Lehmann,
O2/l-h/106cells/ml. 2001. High-density culture of recombinant Chinese
In summary, this research was focused on hamster ovary cells producing prothrombin in
mathematical calculations for the prediction of CHO cell protein-free medium. Biotechnol. Lett., 23: 767-770.
growth in a packed-bed bioreactor. To our knowledge, no 8. Liu, H., X.M. Liu, S.C. Li, B.C. Wu, L. Ye, Q.W. Wang
empirical models have yet been developed to simulate the and Z.L. Chen, 2009. A high-yield and scaleable
experimental data of CHO cells in a packed-bed bioreactor. adenovirus vector production process based on high
In this study, the kinetic parameters were optimized by density perfusion culture of HEK 293 cells as
genetic algorithm. Then the CHO cell growth was suspended aggregates. J. Biosci. Bioeng.,
simulated based on Monod kinetics which showed a good 107: 524-529.
agreement with the laboratory results, hence it can be 9. Takuma, S., C. Hirashima and J.M. Piret, 2007.
used to predict the CHO cell growth and determine the Dependence on Glucose Limitation of the pCO2
values of important factors expressing nutrient Influences on CHO Cell Growth, Metabolism and IgG
Production. Biotechnol. Bioeng., 97: 1479-1488.
consumption and some metabolites production. Improved
10. Kim, D.Y., J. Chul Lee, H. Nam Chang and D. Jae Oh,
understanding of the process leads to the promotion of
2005. Effects of supplementation of various medium
cell density and improvement of productivity which has
components on Chinese hamster ovary cell cultures
a great impact on the industrial production of recombinant
producing recombinant antibody. Cytotechnology,
proteins by mammalian cell cultures.
47: 37-49.
11. Kurano, N., C. Leist, F. Messi, S. Kurano and
ACKNOWLEDGEMENTS
A. Fiechter, 1990. Growth behavior of Chinese
hamster ovary cells in a compact loop bioreactor.
This work was financially supported in part by
2. Effects of medium components and waste
International Center of Genetic Engineering and
products. Biotechnol., 15: 113-128.
Biotechnology (ICGEB, Italy). 12. Yang, M. and M. Butler, 2000. Effects of Ammonia on
CHO Cell Growth, Erythropoietin Production and
REFERENCES Glycosylation. Biotechnol. Bioeng., 68: 370-380.
13. Hansen, A.H. and C. Emborg, 1994. Influence of
1. Warum, F.M., 2004. The CHO cell is widely used in Ammonium on Growth, Metabolism and Productivity
production of biopharmaceuticals because they are of a Continuous Suspension Chinese Hamster Ovary
known as stable hosts, which can easily grow under Cell. Biotechnol. Progr., 70: 121-124.
environmental stress. Nat. Biotechnol., 11: 1393-1398. 14. Lu, S., X. Sun and Y. Zhang, 2005. Insight into
2. Matsunaga, N., K. Kano, Y. Maki and T. Dobashi, metabolism of CHO cells at low glucose
2009. Culture scale-up studies as seen from the concentration on the basis of the determination of
viewpoint of oxygen supply and dissolved carbon intracellular metabolites. Process Biochemistry,
dioxide stripping. J. Biosci. Bioeng, 107: 412-418. 40: 1917-1921.
3. Matsunaga, N., K. Kano, Y. Maki and T. Dobashi, 15. Fang, Z., Y. Xiao-Ping, S. Xiang-Ming and
2009. Estimation of dissolved carbon dioxide Z. Yuan-Xing, 2006. Reduction of Toxic Effect of
stripping in a large bioreactor using model medium. Ammonia on Metabolism of Recombinant CHO-GS
J. Biosci. Bioeng., 107: 419-424. Cell. Chin. J. Biotech., 22: 94-100.
4. Meuwly, F., U.V. Stockar and A. Kadouri, 2004. 16. Altamirano, C., A. Illanes, S. Becerra, J.J. Cairo and
Optimization of the medium perfusion rate in a F. Godia, 2006. Considerations on the lactate
packed-bed bioreactor charged with CHO cells. consumption by CHO cells in the presence of
Cytotechnology, 46: 37-47. galactose. J. Biotechnol., 125: 547-556.
1574World Appl. Sci. J., 15 (11): 1568-1575, 2011
17. Simon, L. and N.M. Karim, 2001. Model predictive 23. Dyring, C., A.H. Hansen and C. Emborg, 1994.
control of apoptosis in mammalian cell culture. Observation on the influence of glutamine,
European Symposium on Computer Aided process asparagines and peptone on growth and t-pA
engineering, 11: 761-766. production of Chinese Hamster Ovary (CHO) cells.
18. Acosta, M.L., A. Sanchez, F. Garcia, A. Contreras Cytotechnology, 16: 37-42.
and E. Molina, 2007. Analysis of kinetic, 24. Chena, P. and S.W. Harcum, 2005. Effects of amino
stoichiometry and regulation of glucose and acid additions on ammonium stressed CHO cells.
glutamine metabolism in hybridoma batch Biotechnol., 117: 277-286.
cultures using logistic equations. Cytotechnology,
54: 189-200. Nomenclature
Amm Ammonium
19. Liu, C.H. and L.H. Chen, 0000. Enhanced recombinant Glc Glucose
M-CSF production in CHO cells by glycerol addition: Lac Lactate
model and validation. Cytotechnology, 54: 89-96. CGlc Glucose concentration (g/l)
CLact Lactate concentration (g/l)
20. Heidemann, R., D. Lütkemeyer, H. Büntemeyer and CAmm Ammonium concentration (g/l)
J. Lehmann, 1998. Effects of dissolved oxygen levels CL* Maximum oxygen concentration in the liquid (mg/l)
and the role of extra-and intracellular amino acid CL Oxygen concentration in the liquid (mg/l)
KLa Volumetric mass transfer coefficient (min 1)
concentrations upon the metabolism of mammalian KGlc Glucose Monod constant (g/l)
cell lines during batch and continuous cultures. KiLact Inhibition constant for lactate (g/l)
Cytotechnology, 26: 185-197. KiAmm Inhibition constant for ammonium (g/l)
qO2 Specific oxygen consumption rate
21. Fassnacht, D. and R. Portner, 1999. rGlc Glucose consumption rate
Experimental and theoretical considerations on rLac Lactate formation rate
oxygen supply for animal cell growth in fixed-bed rAmm Ammonium formation rate
YX/Glc Glucose yield coefficient (108cells/g)
reactors. J. Biotechnol., 72: 169-184. YX/Lac Lactate yield coefficient (108cells/g)
22. Rivera, E.C., A.C. Costa, D.I.P. Atala, F. Maugeri, YX/Amm Ammonium yield coefficient (108cells/g)
M.R. Wolf Maciel and R.M. Filho, 2006. Evaluation of YX/Amm Ammonium yield coefficient (108cells/g)
YLac/Glc Glucose yield coefficient (g/g)
optimization techniques for parameter estimation: YAmm/Glc Glucose yield coefficient (g/g)
Application to ethanol fermentation considering Greek letter
the effect of temperature. Process Biochemistry, µmax Maximum specific growth rate (1/h)
41: 1682-1687.
1575You can also read
