Biotechnol. J. 2011, 6, 1497–1503 DOI 10.1002/biot.201000273 www.biotechnology-journal.

com

Technical Report

Adaptation of the “Dynamic Method” for measuring the specific

respiration rate in oxygen transfer systems through diffusion
membrane

Marilena Martins Pamboukian1, Carlos Augusto Pereira2, Elisabeth de Fatima Pires Augusto3 and Aldo Tonso1
1 Departamento de Engenharia Química, Escola Politécnica, USP, Brazil
2 Laboratório de Imunologia Viral, Instituto Butantan, Brazil
3 Laboratório de Biotecnologia Industrial, Instituto de Pesquisas Tecnológicas – IPT, Brazil

Monitoring the specific respiration rate (QO2) is a valuable tool to evaluate cell growth and physi- Received 30 July 2010
ology. However, for low QO2 values the accuracy may depend on the measurement methodology, Revised 23 December 2010
as it is the case in animal cell culture. The widely used “Dynamic Method” imposes serious diffi- Accepted 26 April 2011
culties concerning oxygen transfer cancellation, especially through membrane oxygenation. This
paper presents an improved procedure to this method, through an automated control of the gas
inlet composition that can minimize the residual oxygen transfer driving force during the QO2
measurement phase. The improved technique was applied to animal cell cultivation, particularly
three recombinant S2 (Drosophila melanogaster) insect cell lines grown in a membrane aeration
bioreactor. The average measurements of the proposed method reached 98% of stationary liquid
phase balance method, taken as a reference, compared to 21% when the traditional method was
used. Furthermore, this methodology does not require knowledge of the volumetric transfer coef-
ficient kLa, which may vary during growth.

Keywords: Animal cell culture · Biochemical engineering · Diffusion membrane · Oxygen transfer · Specific respiration rate

1 Introduction lar respiration are among the most interesting,

showing excellent correlation with cell growth, and
Adequate monitoring and control techniques can yet reporting on the metabolic cell needs through a
certainly help to ensure bioprocess robustness and non-invasive technique [3–6]. In a bioprocess, oxy-
maximum productivity and income [1, 2]. The high gen transfer from the gas phase to the medium oc-
value-added products generated exclusively by curs simultaneously to the consumption of the dis-
such a system also justify the adoption of advanced solved oxygen (DO) by the cells. Therefore, a mass
(and expensive) monitoring and control methods in balance of oxygen in the liquid phase can be de-
this industry. scribed as:
Within the numerous process variables that can
dC
be monitored and controlled, those related to cellu- = OTR – OUR = k La ⋅(CS − C) − QO2 ⋅ X (1)
dt
where C (g O2/L) is the DO concentration in the
Correspondence: Professor Aldo Tonso, Caixa Postal 61548 – medium, OTR (g O2/L.h) is the oxygen transfer rate,
CEP 05424-970 São Paulo, SP, Brazil OUR (g O2/L·h) is the oxygen uptake rate, kLa (h–1)
E-mail: [email protected]
is the volumetric oxygen transfer coefficient, CS
Abbreviations: DO, dissolved oxygen; OTR, oxygen transfer rate; OUR, oxy-
(g O2/L) is the concentration of DOin equilibrium
gen uptake rate; PDM, proposed dynamic method; QO2, specific respiration with the gas phase, QO2 (mg O2/cell·h) is the specif-
rate; SLPB, stationary liquid phase balance; TDM, traditional dynamic ic respiration rate, and X (cell/mL) is the viable cell
method concentration.

© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1497

Biotechnology Biotechnol. J. 2011, 6, 1497–1503
Journal

According to Henry’s law, the concentration of The application of the dynamic method to mem-
DO relates to the oxygen fraction in the gas as: brane aeration bioreactors presents an additional
pO2 P ⋅ yO2 complication compared to gas sparging system: be-
cS = = (2) sides surface transfer that is not cancelled, the
H H
residual gas in the membrane after halting the flow
where H (L.atm/gO2) is the Henry’s law, pO2 (atm) is creates a positive gradient between this gas and the
the partial pressure of oxygen in the gas phase, P DO in the liquid, which produces an extra transfer.
(atm) is the total pressure and yO2 is the mole frac- This latter transfer can be significant since the tub-
tion of oxygen in the gas phase. ing surface can be many times higher than head
QO2 depends on the estimation of OUR and cell space surface.
concentration, the latter obtained on-line through This paper proposes a modification to the “Dy-
specific sensors (turbidimeter), or off-line through namic Method“ to reduce this O2 transfer. The re-
different techniques applied to samples taken from sults obtained with this new proposed dynamic
the bioreactor (e.g., for cell count, gravimetric, tur- method are compared to the traditional procedure
bidity). Two methods have been widely exploited to and that proposed by Kamen et al. [6]. These tech-
determine OUR in bioprocesses [7]: the gas balance niques were applied to three different recombinant
and the dynamic method. The gas balance is based S2 (Drosophila melanogaster) insect cell strains
on the measurement of the difference between grown in a membrane aeration bioreactor.
oxygen mole flow rates in the bioreactor inlet and
outlet gases. The dynamic method requires a total
cancellation of the oxygen transfer to the bioreac- 2 Materials and methods
tor for a short time interval, during which the re-
duction of the DO concentration, due exclusively to 2.1 Cells and culture medium
cell respiration, is monitored [8]. OUR is deter-
mined as the slope of the curve C = f (t) (Eq. 1). Three experiments were performed using three
Gas balance method allows on-line OUR moni- genetically modified cell lines derived from D.
toring, but, on the other hand, has disadvantages melanogaster – S2 (Invitrogen, USA): Run 1 was
that may hinder its application. This is a technique carried out with a cell strain (S2AcRVGP) constitu-
that requires additional equipment; more impor- tively expressing the rabies virus glycoprotein
tant is that small differences between inlet and out- (RVGP) [10]; Run 2 was carried out with a cell lin-
let gas composition may result in large inaccuracies eage [S2MtEGFP] expressing the enhanced green
as in animal cell cultivation that presents low oxy- fluorescent protein (EGFP) after induction with
gen consumption rates. CuSO4 [11]; Run 3 was carried out with a cell strain
In turn, the dynamic method requires a com- (S2AcHBsAgHy) constitutively synthesizing hepa-
plete cancellation of oxygen, which is usually ac- titis B surface antigens (HBsAg) [12].
complished, in sparged transfer systems, by stop- Cells were cultivated in batch mode, using
ping gas inlet. In effect, this does not cancel super- serum-free medium SF900 II® (Invitrogen, USA),
ficial O2 transfer at the bioreactor headspace, and were inoculated with 5 × 105 cells/mL [13].
which is not significant for high O2 transfer condi- In Run 2, 700 µM CuSO4 was added after
tions, but may result in an important error in OUR 2.5 days to induce protein synthesis, and in Run 3,
measuring for low O2 transfer situations. Neverthe- 5 mM sodium butyrate was added after 5 days to
less, in most animal cell cultures, sparging is not enhance protein expression.
commonly used, since bubbles can easily damage Total cell density was determined with a hema-
cells due to their lack of wall. The degree of sus- cytometer and the viability by the trypan blue ex-
ceptibility is strain dependent and insect cells are clusion method.
among the most susceptible. To minimize these
harmful effects, it is common to use bubble-free 2.2 System description
aeration systems; the most successful alternative is
the diffusion of gases through membranes. These The tests were carried out in a 2-L bench bioreac-
membranes are made of materials with high gas tor Discovery 100 (Inceltech, France) with 1-L
permeability such as silicone or teflon, leading to working volume, with temperature controlled at
an acceptable transfer rate between the gas inside 28°C and agitation at 100 rpm.
the membrane and the liquid phase in the bioreac- The oxygen supply was made through silicone
tor. Several meters of tubing can be installed inside membrane with the following characteristics:
the reactor, resulting in a large area of gas ex- Silastic® silicone tubing (Dow Corning Corpora-
change [9]. tion, USA) with internal diameter of 2 mm, outer di-

1498 © 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Biotechnol. J. 2011, 6, 1497–1503 www.biotechnology-journal.com

The TDM method was originally described by

Bandyopadhyay et al. [8], and its standard steps
were: (a) gas flow is interrupted, (b) DO decrease is
monitored; (c) the same initial gas flow condition is
re-established.
With the SLPB method, since the gas flow inlet
and mean pressure in the membrane were kept
constant during the experiments, the kLa system
value is approximately constant. Thus, it is possible
to calculate OUR by oxygen mass balance in liquid
phase using the strategy described by Kamen et al.
[6], which is valid for constant DO concentration.
Equation (1) becomes:

Figure 1. Scheme of the oxygen transfer control system through a diffu- OUR = kLa · (CS – C) (3)
sion membrane. (1) Oxygen mass flow controller, (2) mixed gas mass
flow controller, (3) control system, (4) headspace valve, (5) DO probe,
(6) exhaust valve (back pressure regulator), (7) diffusion membrane. Considering kLa, C and H constants and known
throughout the cultivation, OUR can be calculated
in real time based on CS (Eq. 5). In our system,
ameter of 3.2 mm and 8.3 m length (interface area kLa was determined as described by Bartholomew
of 0.052 m2). et al. [14] in the absence of cells, in water, under the
To maintain DO at 40% air saturation, a polaro- same experimental conditions: geometry of the
graphic sensor InPro 6110 (Mettler Toledo, bioreactor, gas flow rate and mean membrane pres-
Switzerland) was used to inform a LabVIEW (Na- sure. An average value of 1.18 ± 0.06 h-1 was ob-
tional Instruments, USA) control routine of any de- tained.
viation of C. In order to re-establish the desired C The PDM methodology applies the same princi-
value in liquid, the routine determined a new CS ple of the dynamic method described above, but
(Eq. 1) and the corresponding yO2 (Eq. 2) that was uses a different strategy to cancel oxygen transfer.
implemented through two mass flow controllers Instead of zeroing gas flow, a gas mixture with a yO2
(MFC) 5850EM (Brooks Instruments, Holland) that is in equilibrium with the O2 concentration of
used to blend O2 and N2 gas lines (Fig. 1). Total gas the liquid is circulated through the membrane,
flow rate was kept constant at 52 mL/min. seeking CS = C, and then cancelling the transfer
In the proposed configuration, oxygen transfer driving force during the OUR measurement. For
varies with oxygen partial pressure (pO2) within the this, the control system sends continuously a signal
silicone tubing, which depends on the total pres- to the MFC to adjust O2 flow so that the gas mixture
sure inside the membrane (P) and the mass frac- has the desired composition, i.e., Cs = C. Also, to en-
tion of oxygen in the aeration gas (yO2). As the total sure cancellation of surface transfer, valve 4 (Fig. 1)
flow rate is kept constant, the total pressure is also is opened, allowing the same gas mixture flow si-
constant. The total pressure of the system is ap- multaneously to the headspace.
proximately equal to the mean logarithmic pres-
sure at the silicone tube inlet and outlet [9]. The
outlet pressure is regulated by a valve (10BP Back 3 Results and discussion
Pressure Regulator; Fairchild, USA), which can be
adjusted from 0.03 to 2.0 bar. Growth profiles before induction (Run 2) and pro-
tein expression enhancement (Run 3) were very
2.3 Measurements of specific oxygen consumption similar for the three runs. As an illustration, Fig. 2
rate presents results of Run 2.Table 1 shows data for the
three experiments.
QO2 was calculated based on OUR and cell concen- In Fig. 2A, it can be seen that the DO was pre-
tration in each sample (Eq. 1). Three methods for cisely controlled. The peaks observed relate to DO
measuring OUR were used: (i) the traditional dy- concentration changes, and result from the appli-
namic method (TDM), (ii) continuous determina- cation of TDM and PDM. As expected, QO2 (Fig. 2A)
tion by stationary liquid phase balance (SLPB), and and specific growth rate (data not shown) had
(iii) determination using the proposed dynamic very similar profiles, showing a plateau of maxi-
method (PDM), with cancellation of the oxygen mum values, up to the CuSO4 induction (3.25 days).
transfer driving force. After this, QO2 diminished drastically, although

© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1499

Biotechnology Biotechnol. J. 2011, 6, 1497–1503
Journal

Figure 2. (A) Profile of cell concentration, DO and QO2 as a function of

time for Run 2 (S2MtEGFP cells). (B) DO measurements using TDM and
PDM at two time points during culture.

growth was still significant, indicating a stronger The explanation for this difference between the
effect of induction on oxygen uptake. Values of TDM and PDM derives from the fact that the TDM
4.16 × 10–10 mg O2/cell.h for wild S2 cells [13], and is unable to completely cancel the O2 transfer
11.8 × 10–10 to 23.7 × 10–10 mg O2/cell.h for modified through diffusion membranes. The PDM cancels
cells [15] has been reported, and are very similar to the transfer using a flow of gas through the mem-
those shown in this paper. Those QO2 values are in- brane and headspace, the composition of which is
ferior to that described in literature for other insect in equilibrium with the DO in the liquid, zeroing
cells: 48.0 × 1010 mg O2/cell.h [4, 13, 15, 16] or for the driving force.
mammalian cell lines (CHO, HeLa and NS0): Figure 2B clearly shows the differences be-
68.0 × 10–10 to 131.2 × 10–10 mg O2/cell.h [16–18]. tween the DO concentration profiles in the two
The QO2 measurements obtained using TDM methods, and explains the differences in QO2 val-
and PDM present very different results, with PDM ues. In the TDM, the membrane retains the gas cir-
always showing higher values than those of TDM. culation just before procedure initialization and
Since it is impossible to apply TDM and PDM si- maintains the transfer of oxygen to the system dur-
multaneously, individual results were compared ing the test. In this case, the DO concentration de-
with those of a third method (SLPB), to evaluate caying due to cellular consumption is always small-
their accuracy. er than expected (Fig. 2B), since a residual oxygen
Table 1 presents the ratio between the PDM and transfer exists, contributing to its increase. The
SLPB, as well as between TDM and SLPB, calculat- variation observed results from the balance be-
ed for all available data, and their average, SD and tween consumption and residual transfer (Eq. 1).
CV(%) (100 × SD/average) values. The average of Depending on the composition of gas trapped in
PDM/SLPB ratio was 98%, whereas that of TDM/ the system, differences can be very significant.
SLPB was just 21%. These average values are sta- To illustrate the errors involved in the TDM, we
tistically different (t-test, for α = 0.05) and attest estimated the oxygen transfer through the mem-
that the proposed method gives results that are in brane at the beginning of the method, when gas
most cases much closer to SLPB. flow rate is interrupted.

1500 © 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

 Table 1. Data from several samples of the three runsa)

Comparison: SLPB to PDM Comparison: SLPB to TDM

Time X QO2 PDM/SLPB Cs residual Time X QO2 TDM/SLPB Cs residual
(days) (106cell/mL) (10–9mg O2/cell·h) (mg O2/L) (days) (106cell/mL) (10–9mg O2/cell·h) (mg O2/L)
SLPB PDM ratio SLPB TDM ratio
3.79 2.41 1.14 0.88 0.77 10.0 2.87 0.94 4.43 0.41 0.09 6.3
4.79 7.68 1.53 1.71 1.12 17.6 4.19 4.16 1.29 0.50 0.39 12.6
Biotechnol. J. 2011, 6, 1497–1503

5.95 15.60 1.54 1.80 1.17 25.1 5.20 10.68 1.18 0.14 0.12 21.3
6.77 22.35 1.32 1.40 1.06 26.4 6.29 19.60 1.24 0.11 0.09 27.6
7.23 24.86 1.10 1.15 1.05 26.4 7.79 26.60 1.01 0.06 0.06 26.4
8.25 24.72 1.12 1.21 1.08 23.9

Run 1 – S2AcGPV
9.2 19.30 1.40 1.54 1.10 25.1
2.2 3.36 2.79 2.97 1.06 11.3 4.13 8.46 1.99 0.11 0.06 17.6
2.87 4.90 2.74 2.30 0.84 13.8 5.89 17.80 0.61 0.05 0.08 11.3

© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

3.25 6.00 2.47 2.07 0.84 15.1

Run 2 –
4.68 10.15 2.17 1.50 0.69 21.3

S2MtEGFP
5.34 13.80 0.98 0.37 0.38 10.0
7.03 4.10 3.26 3.78 1.16 11.3 4.95 1.20 6.05 2.00 0.33 8.8
8.87 10.40 1.98 2.13 1.08 16.3 6.03 2.60 4.44 1.23 0.28 11.3
10.89 16.60 1.14 1.56 1.37 20.1 7.97 6.60 1.84 0.78 0.42 13.8

Run 3 –
9.91 14.20 1.23 0.46 0.37 18.8

S2AcHBsAg
average 0.98 average 0.21
SD 0.24 SD 0.15
CV 24.6 CV 71.7
a) X: viable cell concentration, CS residual: CS value just before starting the dynamic method.
www.biotechnology-journal.com

1501
Biotechnology Biotechnol. J. 2011, 6, 1497–1503
Journal

OTRresidual = kLa . (CS residual – C) (4) area. In contrast, the addition of a surfactant such
as Pluronic reduced the kLa up to 35% [21–23].
Where: OTRresidual is the oxygen transfer rate; Hence changes in the composition of the culture
CS residual is the oxygen saturation concentration in medium must change kLa, making constant kLa a
equilibrium with the gas inside the membrane (see hypothesis difficult to sustain. Moreover, there is
Table 1); C is the oxygen concentration in the cul- evidence that the presence of cells as suspended
ture medium, controlled at 40% of saturation with solids interfere with the measurement of kLa [24].
air. Thus, for Run 2, at 3.25 days (Fig. 2B; Table 1), Therefore, the assumption of a constant kLa, re-
we obtain: quired for the SLPB methodology, may compromise
the measurements.
OTRresidual = 1.01 × (17.6 – 0.4 × 6.99) = In this sense, the PDM, a more laborious tech-
15.0 mg O2/L.h nique that requires continual gas formulation,
adapts more easily to the common variations that
where kLa = 1.01 h–1 calculated through this PDM occur throughout the culture as a function of me-
(considered more precise) and C = 40% of oxygen tabolism and cell growth.Thus, it constitutes an ex-
solubility in water at 28°C, according to U.S. Geo- cellent choice for measuring specific respiration
logical Survey Techniques of Water-Resources In- rate in oxygen transfer through membranes or un-
vestigations (available at http://pubs.water.usgs. der low transfer conditions, when surface transfer
gov/twri9A6). contribution may significantly interfere in the
OTR calculated from the TDM values (Fig. 2B; TDM.
Table 1; 4.13 days; no oxygen accumulation) gives:

OTRTDM = QO2 × X = 0.11 × 10-9 × 8.46 × 106= 4 Concluding remarks

0.93 mg O2/L.h.
This study evaluated a new methodology for meas-
Therefore, residual OTR is 16 times greater than uring the QO2, and compares it to two other meth-
TDM measurement, which represents a significant ods (TDM and SLPB). The proposed method is
error.These discrepancies varied from 2 to 16 times based on the ability to completely eliminate oxygen
for the conditions in Table 1.This data dispersion is transfer during the test, so that the variation of oxy-
due to the great variation of CS values at the begin- gen concentration can be exclusively associated to
ning of the TDM assay. The higher the CS value, the cell respiration. This ability to completely cancel
greater the undesired O2 transfer. This variability the transfer is of great importance in low transfer
explains the higher CV values observed for the rate systems and in membrane aeration bioreactors
TDM/SLPB ratio, as compared to the PDM/SLPB (typical situations of animal cells cultures), but also
ratio, corroborating that the PDM method is more in some microbial systems characterized by low
precise. consumption of oxygen.
Comparing the reference methodology (SLPB) The data shown indicate that the PDM results in
to the new methodology (PDM) shows that SLPB is QO2 values are significantly higher (up to 20 times)
a relatively simpler method than the PDM, since it than those obtained by the traditional method. The
depends only on the measured composition of the proposed methodology has the advantage of not
inlet gas and prior knowledge of the kLa value, depending on changes in medium composition or
whereas the PDM requires the formulation of gas rheology.
circulating in the membrane and at the headspace
in real time, to completely eliminate possible O2
transfer during the assay. This work was partially supported by grants from the
Indeed, kLa characterization in bioreactors (in Brazilian agencies FAPESP and CNPq. Marilena M.
water or in culture media), and in the absence of Pamboukian had a scholarship from CNPq. Carlos A.
cells, is quite easy and can be done by many meth- Pereira is recipient of CNPq 1A research fellowship.
ods [14, 19, 20]. However, medium composition
clearly affects the kLa value. For example, by intro- The authors have declared no conflict of interest.
ducing ions in the water up to a limiting concentra-
tion of 10 g/L NaCl, Van’t Riet [19] showed that kLa
can increase by a factor of 100% compared to dis-
tilled or tap water measurement. The presence of
ions strongly changes the coalescent characteristic
of pure water, increasing drastically the interfacial

1502 © 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Biotechnol. J. 2011, 6, 1497–1503 www.biotechnology-journal.com

5 References cells as assessed by EGFP expression. Cytotechnology 2007,

54, 15–24.
[12] Jorge, S., Santos, A., Spina, A., Pereira, C., Expression of the
[1] Konstantinov, K., Chuppa, S., Sajan, E., Tsai, Y. et al., Real-
hepatitis B virus surface antigen in Drosophila S2 cells. Cy-
time biomass-concentration monitoring in animal-cell cul-
totechnology 2008, 57, 51–59.
tures. Trends Biotechnol. 1994, 12, 324–333.
[13] Pamboukian, M. M., Jorge, S. A. C., Santos, M. G., Yokomizo,
[2] Lecina, M., Soley, A., Gràcia, J., Espunya, E. et al., Application
A. Y. et al., Insect cells respiratory activity in bioreactor. Cy-
of on-line OUR measurements to detect actions points to
totechnology 2008, 57, 37–44.
improve baculovirus-insect cell cultures in bioreactors. J.
[14] Bartholomew, W. H., Karow, E. O., Sfat, M. R., Wilhem, R. H.,
Biotechnol. 2006, 125, 385–394.
Oxygen transfer and agitation in submerged fermentations
[3] Palomares, L., Lopez, S., Ramirez, O., Utilization of oxygen
– Mass transfer of oxygen in submerged fermentation of
uptake rate to assess the role of glucose and glutamine in
Streptomyces griseus. Ind. Eng. Chem. 1950, 42, 1801–1809.
the metabolism of infected insect cell cultures. Biochem.
[15] Batista, F. R. X., Moraes, A. M., Buntemeyer, H., Noll, T., In-
Eng. J. 2004, 19, 87–93.
fluence of culture conditions on recombinant Drosophila
[4] Palomares, L., Ramirez, O., The effect of dissolved oxygen
melanogaster S2 cells producing rabies virus G glycoprotein
tension and the utility of oxygen uptake rate in insect cell
cultivated in animal protein-free medium. Biologicals 2009,
culture. Cytotechnology 1996, 22, 225–237.
37, 108–118.
[5] Higareda, A., Possani, L., Ramírez, O., The use of culture re-
[16] Gray, D. R., Chen, S., Howarth, W., Inlow, D. et al., CO2 in
dox potential and oxygen uptake rate for assessing glucose
large-scale and high-density CHO cell perfusion culture.
and glutamine depletion in hybridoma cultures. Biotechnol.
Cytotechnology 1996, 22, 65–78.
Bioeng. 1997, 56, 555–563.
[17] Shuler, M. L., Kargi, F., Bioprocess Engineering Basic Con-
[6] Kamen, A., Bédard, C.,Tom, R., Perret, S. et al., On-line mon-
cepts, Prentice-Hall PTR, Upper Saddle 2002.
itoring of respiration in recombinant-baculovirus infected
[18] Yoon, S., Konstantinov, K. B., Continuous, real-time moni-
and uninfected insect cell bioreactor cultures. Biotechnol.
toring of the oxygen uptake rate (OUR) in animal-cell biore-
Bioeng. 1996, 50, 36–48.
actors. Biotechnol. Bioeng. 1994, 44, 983–990.
[7] Eyer, K., Oeggerli, A., Heinzle, E., On-line gas analysis in an-
[19] Van’t Riet, K., Review of measuring methods and results in
imal cell cultivation: II. Methods for oxygen uptake rate es-
nonviscous gas-liquid mass transfer in stirred vessels. Ind.
timation and its application to controlled feeding of gluta-
Eng. Chem. Process Des. Dev. 1979, 18, 357–364.
mine. Biotechnol. Bioeng. 1995, 45, 54–62.
[20] Ruchti, G., Dunn, I. J., Bourne, J. R., Comparison of dynamic
[8] Bandyopadhyay, B., Humphrey, A. E., Taguchi, H., Dynamic
oxygen-electrode methods for the measurement of kLa.
measurement of the volumetric oxygen transfer coefficient
Biotechnol. Bioeng. 1981, 23, 277–290.
in fermentation systems. Biotechnol. Bioeng. 1967, 9, 533–
[21] Metz, B., Kossen, N. W. P., van Suijdam, J. C., The rheology of
544.
mold suspensions. Adv. Biochem. Eng. 1979, 11, 103–156.
[9] Qi, H., Goudar, C., Michaels, J., Henzler, H. et al., Experi-
[22] Morão, A., Maia, C., Fonseca, M., Vasconcelos, J. et al., Effect
mental and theoretical analysis of tubular membrane aera-
of antifoam addition on gas-liquid mass transfer in stirred
tion for Mammalian cell bioreactors. Biotechnol. Prog. 2003,
fermenters. Bioprocess Eng. 1999, 20, 165–172.
19, 1183–1189.
[23] Lavery, M., Nienow, A., Oxygen transfer in animal cell cul-
[10] Yokomizo, A., Jorge, S., Astray, R., Fernandes, I. et al., Rabies
ture medium. Biotechnol. Bioeng. 1987, 30, 368–373.
virus glycoprotein expression in Drosophila S2 cells. I.
[24] Ju, L. K., Sundararajan, A., The effects of cells on oxygen
Functional recombinant protein in stable co-transfected
transfer in bioreactors. Chem. Eng. J. 1995, 56, B15–B21.
cell line. Biotechnol. J. 2007, 2, 102–109.
[11] Santos, M., Jorge, S., Brillet, K., Pereira, C., Improving het-
erologous protein expression in transfected Drosophila S2

© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1503