The words you are searching are inside this book. To get more targeted content, please make full-text search by clicking here.
Discover the best professional documents and content resources in AnyFlip Document Base.
Search
Published by soedito, 2017-08-25 02:52:40

12_ANIMAL CELL TECHNOLOGY_707

12_ANIMAL CELL TECHNOLOGY_707

319

TABLE 2. A comparison of statistical results for the mean number of dead cells contained in 1 ml cell
suspension

With the CHO cell line used in our experiments, it was not possible to omit the staining
step and use only morphological characteristics of the cells for the determination of cell
viability. In contrast to the results reported [2], we did not find any significant
difference in size, shape or density of viable and non-viable cells.
The results show that DIA is superior to the manual hemocytometer method, and an
important additional advantage is the possibility of automation.
For automation of viability determination in fermentation processes, an online image
processing system, developed for the characterization of morphology of mycelial
microorganisms [3], can be used. Sampling, dilution, image acquisition and image
analysis are done automatically under the control of a PC. Figure 2 shows the
experimental set-up of the online DIA system.

4. Acknowledgements
The work was supported by the German Federal Ministry of Science and Technology
(BMFT) and the Dr. Karl Thomae GmbH. The authors would like to thank Prof.
Pfizenmaier for kindly providing the CHO cells and the possibility to perform the
experiments at the Institut für Zellbiologie und Immunologie, Universität Stuttgart.

5. References

[1] Konstantinov, K. et al. (1994) Real-time biomass-concentration monitoring in animal-cell
cultures, TIBTECH 12, 324-333

[2] Frame, K. K. and Hu, W.-S. (1989) Cell Volume Measurement as an Estimation of Mammalian
Cell Biomass, Biotechnol. Bioeng. 36, 191-197

[3] Treskatis, S.-K. (1995) Wachstum, Morphologic und Antibiotikaproduktion von Streptomycten
in Submerskultur , PhD thesis, Universität Tübingen, Germany

THE VIABLE CELL MONITOR: A DIELECTRIC SPECTROSCOPE FOR

GROWTH AND METABOLIC STUDIES OF ANIMAL CELLS ON
MACROPOROUS BEADS

Y. GUAN and R.B. KEMP
Institute of Biological Sciences
University of Wales, Penglais, Aberyshvyth, SY23 3DA, UK

Abstract. One of the problems in using macroporous carriers to grow animal cells in
culture has been to assess biomass on-line. The possibility that dielectric spectroscopy
could be the solution was explored using a Viable Cell Monitor (VCM) optimised for
animal cells. The instrument was validated using suspension-adapted CHO cells and

was shown to measure the volume fraction of viable cells. It was then applied to

Cytopore 1 carrier cultures over 7-day periods, the results indicating that the technique
was a reliable complement to metabolic studies.

1. Introduction

The biotechnological exploitation of animal cells to produce monoclonal antibodies
and recombinant proteins in culture has focused attention on the need to provide a
precisely controlled environment for their growth. This was recognised more than a
decade ago [1] but there is still a paucity of on-line biosensors to monitor culture
variables [2], let alone use them to control the process. This is true even of the
fundamental measurement of biomass where the only two methods available are optical
density using laser light [3] and dielectric spectroscopy [4]. Whereas the former gives
the total number (living and dead) of cells, the latter identifies only the living cells.
Both sets of information are valuable in assessing percentage viability but viable cell
number is of primary importance in metabolic studies and those relating to the
efficiency of target protein production.

Commercial dielectric spectroscopes designed to monitor the biomass of
microorganisms have been available for several years [4, 5] and have been employed
with limited success to measure the number of viable animal cells in culture [6]. It is
only recently, however, that one optimised for that purpose has come on the market
with the name of Viable Cell Monitor (VCM). The first aim of the present study was to
evaluate the instrument using Chinese hamster ovary (CHO) cells growing in
suspension.

321
O.-W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 321-328.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

322

The productivity per unit volume of animal cells in simple batch culture is poor
because the cell number concentration is comparatively low at the end

Various devices have been used to increase the numbers, including fed-batch
and perfusion cultures, but one of the most promising ways is to allow the cells to stick
to macroporous carriers [7]. A difficult problem with this method, however, is to
ascertain the number and viability of the cells because many of them are buried deep
within the bead. There was preliminary evidence that the dielectric spectroscopy could
fulfil this role [8] and. thus, the second aim of the study was to see if the VCM was
appropriate for this task. Earlier it had been shown that CHO cells could be grown in
macroporous carriers for 3 weeks |9| and so these were used for this work.

2. Theoretical

A detailed theoretical treatment of the theory for measuring biomass by dielectric
spectroscopy is available elsewhere [5] but. as a simplified account, the electric field
set up by the 4 electrode pins of the VCM probe in the biorcactor creates a force field
which pushes ions in opposite directions until stopped by intact cell plasma
membranes [10]. Thus, a charge separation (polarisation) is set up at the poles which is
measured as capacitance in Farads (F) by a phase shift between the outer electrodes
(current) and the inner ones (voltage). The more the cells, the greater the amount of
plasma membrane and the more the capacitance Dead cells and debris do not have
intact and/or functioning plasma membranes and so do not contribute to capacitance.
It is obviously necessary to reverse the field periodically and the rate of change in
direction is measured by its frequency as Hertz (Hz). This has a profound effect on
capacitance since the ions need a finite time to move up to and polarise the plasma
membranes. The higher the frequency, the less time there is for polarisation and the
smaller the value for capacitance. The curve of capacitance against frequency is called
the The most appropriate frequency for animal cells has been found to be
0.5 Hz and the VCM is fixed at this point. The culture medium has a residual

capacitance which must be deducted from the value for the cell suspension to

give the capacitance increment of the As the biomass

(cell number) concentration increases, so does the size of Capacitance of the cell
suspension has a complex relationship to the conductivity of it. If the latter is constant
during growth or only changes slightly, then no allowance for it needs to be made but
there can be quite significant changes during long-term culture at high cell densities.
For this reason, the VCM can be calibrated for a range of conductivity values simply by
adding salt to a test sample of culture medium. Without this cross-talk calibration, an

increase of from an initial value of can produce a shift in

capacitance from 0 pF to 0.1 pF.

323

3. Experimental

The cells were CHO 320 which produce recombinant interferon when grown in a
defined medium based on RPMI 1640 [11] with BSA free of fatty acids [12]
For some experiments the cells were grown on Cytopore 1 macroporous carrier beads

(Pharmacia) suspended at a mass concentration of The Applicon
bioreator system (Applicon Ltd., Tewksbury. Glous.. UK) controlled the bicarbonate-

buffered pH at temperature at and the dissolved oxygen
concentration at 55% saturation [12]. It also received the analogue capacitance and

conductivity signals from the VCM (Aber Instruments Ltd.. Aberystwyth. SY23 3AH,

UK) operating at 0.5 MHz. All the digitised signals were logged by Applicon BioXpert
and its moving average facility was used over 1 h to smooth the capacitance signal.
Off-line measurements [12] were made of cell density by Coulter counter Model D.
(Coulter Ltd.. Luton, UK) and of cell size by Skatron Argus 100 flow cytometer
(Skatron Ltd., POB 34, Newmarket, UK). The latter was calibrated with a range of

highly monodisperse latex beads (Dyno Particles AS, POB 160, Lillestrom, Norway).

Since these always underestimate cell size. 400 cells were photographed and measured
for size distribution [12], The concentrations of glucose and lactate were measured,

after deproteinisation, using respectively. Sigma kits 635 and 826. Protein was assayed

by the BioRad DC protein assay kit.

4. Results and Discussion

When CHO 320 cells were cultured in suspension for 140 h. the capacitance curve
gave the characteristic bell-shape (Figure 1A). depicting three phases; an increase for
the first 80 h, followed by a plateau and then by a decline for the final 40 h. Statistical
analysis gave a standard error for the curve of 0.085 pF. Calibrating capacitance in
terms the viable cell number concentration gave the following semi-empirical
correlation:

(1)

where is the increment of viable cell number concentration and that of
capacitance. There was considerable data scatter, but this was reduced by using the
moving average smoothing technique over 1 h (see Figure 1A). As a result, the lowest

measurable cell density was It was also calculated that 1 pF was

equivalent to approximately
It was necessary to establish what exactly capacitance measures in terms of

biomass. Since animal cells take up a gross spherical shape in suspension, the
dielectric principle of measurement is [15,12].

(2)

324

where K is a constant, Nv is cell number concentration, V cell is the average individual
cell volume and is volume fraction of the cell entity. Eq. (2) indicates that the

capacitance measurements could relate to individual cell size and/or viable cell density
and/or the volume fraction of viable cells. Estimates were made of all three variables
and the results are shown in Figure 1B. In the first 100 h. the average cell volume
varied by ca. 21%. In relation to Eq. (2). this finding implies that using the VCM
signal to represent viable cell density would give a 20% relative error when
compared to less than 10% for the volume fraction This difference is illustrated in
Figure 1B which clearly shows that capacitance was better reflected in the volume
fraction than in the viable cell density for the first 100 h. Cell counts were also
unreliable at low number densities. This is common for Coulter counting and is due to
sampling errors. After 100 h. there was a decline that correlated poorly with the viable
cell volume fraction but may have been due to apoptotic cells, which are known to be
smaller than living cells [13].

When the cells were grown on Cytopore 1 beads for long periods in batch culture,
the medium had to be renewed by partial replacement after
allowing the beads to settle for a short while by stopping the agitator. Thus, the growth
as represented by the smoothed capacitance trace, was punctuated by sharp peaks
(Figure 2A). Nevertheless, there was considerable growth over the 168-h period

representing It is not possible to estimate the accuracy of this
number because a direct measurement cannot be performed, but a similar increase with
time was shown for the off-line protein measurement. Protein concentration in itself
cannot be an accurate measurement of viable cells because (i) dead ones would
accumulate in the carrier pores and (ii) BSA and other proteins can be trapped in the
beads. However, the increase could be indicative of cell growth. The changing
capacitance signal could not have been due to altered conductivity because there was

very little change in it over the period (see Figure 2A).

325
Some indication of the metabolic rate of the cells could be gauged from the
consumption of glucose (Figure 2A). With growth, the total glucose used by the cell
population increased so that, with replacement taking place at regular intervals, the
maximum amount of substrate available to the cells gradually decreased with time.
One of the characteristics of cells in culture is that they produce considerable amounts
of lactate. This can be seen in Figure 2B as a virtual mirror image of glucose
consumption. Of course, some of it is removed as medium is replaced to give the saw-
tooth appearance. Lactate was excreted as a result of glycolysis and the oxidation of
glutamine [11,12] but this was not due to anaerobic conditions, the oxygen saturation
being maintained at 55%. Rather, it seems to be caused by the poorly designed medium
which does not contain a sufficient amount of the required biosynthetic precursors
[12,14], The cell has to resort to making some of the necessary components by
oxidoreductive catabolic processes, leading to the formation of lactate as a by-product
under fully aerobic conditions. Its excretion causes the pH of the culture medium to
fall, necessitating the introduction of NaOH to neutralise it. The on-line record of this

titration is probably a fairly accurate reflection of cell growth (Figure 2B), better than
the protein estimation.

Although a great deal more needs to be done to authenticate the use of dielectric
spectroscopy to measure biomass in beads, it is probable that it will become the
standard method of assessment in due course. More confidence in its application could
be derived from studies on whether it measures in proportion the number of cells in
free-standing aggregates. Once validated for animal cell systems, the VCM will
become a valued tool in metabolic studies because it makes these other measurements
specific to size, in this case volume. This is already being done for the estimation of
overall metabolic rate by combining the VCM with flow microcalorimetry to give heat
flux [12]. a variable which can be used on-line to control the growth of animal cells in
suspension culture.

326

5. Acknowledgements

The authors are thankful to the Nuffield Foundation (NUF-URB97) and the Wellcome Trust (VS/3/97) for
vacation scholarships to Dr R.B. Kemp which enabled Miss Rachel Ryan and Mr Stuart Martin, respectively, to
assist in part of these studies. He gratefully acknowledges that the research was supported by BBSRC grants,
2/3680 and 2/TO7389.

6. References

1. Cartwright, T. (1994) Animal Cells as Bioreactors. Cambridge University Press, Cambridge, UK.
2. Zhou, W. and Mulchandani. A. (1995) Recent advances in bioprocess monitoring and control. American

Chemical Society Symposium Series 613, 88–98.
3. Zhou. W.C. and Hu. W.S. (1994) On-line characterisation of a hybridoma cell-culture process.

Biotechnology and Bioengineering 44, 170–177.
4. Harris. C.M., Todd, R.W., Bungard. S.J., Lovitt. R.W., Morris, J.G. and Kell. D.B. (1987) The dielectric

permittivity of microbial suspensions at radio-frequencies: a novel method for the real time estimation of
microbial bilmass. Enzyme and Microbial Technology 9, 181–186.
5. Davey, C.L. and Kell. D.B. (1995) The low-frequency dielectric properties of biological cells, in D. Walz. H.
Berg and G. Milazzo (eds.), Bioelectrochemistry of Cells and Tissues, Birkhauser Verlag, Basel,
Swizerland, pp. 159–207.
6. Cerckel. I., Garcia. A.. Degouys. V., Dubois. D., Fabry. L. and Miller, A.O.A. (1993) Dielectric
spectroscopy of mammalian cells 1. Evaluation of the biomass of HeLa- and CHO cells in suspension by low-
frequency dielectric spectroscopy. Cytotechnology 13, 185–193.
7. Blüml, G.. Reiter. M., Gaida. T.. Schmatz. C., Assadian, A.. Strutzenberger, K. and Katinger, H. (1994)
Development of a new type of macroporous carrier, in R.E. Spier, J.B. Griffiths and W. Berthold (eds.),
Animal Cell Technology, Butterworth-Heinemann, Newton, MA, USA, pp.267–269.
8. Degouys, V., Cerckel. I., Garcia. A.. Harfield. J., Dubois, D., Fabry. L. and Miller, A.O.A. (1993) Dielectric
spectroscopy of mammalian cells 2. Simultaneous in-situ evaluation by aperture impedance pulse
spectroscopy and low-frequency dielectric spectroscopy of the biomass of HTC cells on Cytodex 3.
Cytotechno/ogy 13, 195–202.
9. Reiter. M., Borth. N., Blüml. G., Wimmer. K., Harant. H., Zach. N., Gaida. T., Schmatz, C. and Katinger, N.
(1992) Flow cytometry and two-dimensional eletrophoresis (2-DE) for system evaluation of long term
continuous perfused animal cell cultures in macroporous beads. Cytotechnology 9, 247–253.
10. Davey, C.L., Guan. Y., Kemp. R.B. and Kell. D.B. (1997) Real-time monitoring of the biomass content of
animal cell cultures using dielectric spectroscopy, in K. Funatsu, Y. Shirai and T. Matsushita (eds.). Animal
Cell Technology. Vol. 8, Kluwer Academic Publishers. Dordrecht, The Netherlands, pp.61–65.
11. Hayter. P.M., Curling. E.M.A., Baines. A.J., Jenkins, N., Salmon. I., Strange, P.G. and Bull, A.T. (1991)
Chinese hamster ovary cell growth and interferon production kinetics in stirred batch culture. Applied
Microbiology and Biotechnology 34, 559–564.
12. Guan. Y . Evans. P.M. and Kemp. R.B. (1998) Specific heat flow rate: .An on-line monitor and potential
control variable of specific metabolic rate in animal cell culture that combines microcalorimetry with
dielectric spectroscopy. Biotechnology and Bioengineenng (in press).
13. Wyllie, A.H. and Durall, E. (1992) Cell death, in J.O’D. McGee. P.G.Isaacson and N.A. Wright Oxford
Book of Pathology, Vol. 1. Oxford University Press. Oxford. UK. Pp. 141–157.
14. Kemp, R.B. and Y. Guan (1998) Probing the metabolism of genetically-engineered mammalian cells by heat
flux. Thermochimica Acta (in press).

Discussion 327
Godia:
Kemp: The capacitance measurement you make, is it not at a constant
frequency? You also mention the importance of changes in
Konstantinov: conductivity in the medium and that you correct them by on-line
conductivity measurements in parallel to your electrical
Kemp: measurements.
Al-Rubeai: This is done at a set frequency of 0.5 MHz. This is not necessarily
Kemp: ideal for every mammalian cell, and the system can be changed to
other frequencies. For conductivity measurements the instrument
is calibrated with a range of conductivities by adding salt. The
instrument will automatically adjust for changes in conductivity
because conductivity will affect the capacitance measurement.
You stated that there are two basic methods for measuring cell
density: the optical and the capacitance. I believe they are not
conflicting, but complementary, because the optical method
measures total cell density and the capacitance measures viable cell
density, so you can combine both signals to measure viability. Can
you comment on this possibility, and would you need 2 sensors to
do this?
I think it is a good point. We have not tried it because we cannot
afford to buy the additional sensors. I think that it is an excellent
way of going forward.
Putting additional sensors into a culture gets complicated. Why is
it necessary to have on-line measurement of cell number or
viability? A one off sample gives you opportunities for flow
cytometry, which measures viability very simply.
We want to measure the specific rate of heat production, and the
combination of heat flow and biomass measurements will allow you
to control the culture. The advantage of heat flow is that it is an
instant rate, you do not have to measure concentrations of things.
So when the cells start to decline you can immediately adjust the
conditions.

328 A comment on Al-Rubeai’s question: if you are doing a batch
Ozturk: culture, you do not need to measure cell density because you have
to live with what you have. However, perfusion of fed-batch
Wurm: systems, would benefit from this especially in high density culture
Kemp: where conditions change fast.

Do the size and number of gas particles in your reactor affect the
electronics?

No, it is a sparged system and there is no problem because bubbles
do not charge up.

MEASUREMENTS OF CHANGES IN CELL SIZE DISTRIBUTION TO
MONITOR BACULOVIRUS INFECTION OF INSECT CELLS.

ERNESTO CHICO1,2 and VOLKER JÄGER1.
1Gesellschaft für Biotechnologische Forschung mbH
Mascheroder Weg 1, D-38124 Braunschweig, Germany.
2Center of Molecular Immunology
POBox 16040, Habana 11600, Cuba.

Abstract
We have found that the cell size distributions measured by a CASYTM cell counter can
be used to follow the infection process of various insect cell lines. After infection, cell
size deviates from the typical distribution of exponentially growing cells, shifting to an
increased amount of cells with bigger cell diameters. This deviation has proved to be
dependent on the MOI as well as time post-infection. A method is proposed to
estimate the degree of infection of a population of insect cells based on the cell size
distribution (CSD). The potential of using this method for measuring the ratio of

infected to non-infected cells is discussed.

Experimental conditions
Cell size distributions were measured using a CasyTM cell counter with a measuring
capillary of 150 µ m. No more than 5000 cells per single measurement were counted to
avoid particle clumping inside the capillary. Most of the experiments were carried out
with the BTI-Tn-5Bl-4 (High Five™) cell line grown in suspension using fortified Ex-
Cell 401 serum-free medium. A recombinant AcNPV was used for infection.

Main experimental observations
During some stages of the baculovirus infection, there is a mixed population of
infected and non-infected cells in the culture. However, there is a lack of a simple
experimental method to distinguish between infected and non-infected cells, which can
be reliably used for a numeric estimation of the degree of infection (DOI) of the
culture, refered as the fraction of infected cells in the population.
Cell size distribution (CSD) of insect cells, measured by a CASY™ cell counter, do
not change significantly during the batch growth curve (data not shown). On the
contrary, during the infection process the CSDs of insect cells deviate from the typical
distribution of exponentially growing cells, shifting to an increased amount of cells
with bigger cell diameters. This deviation has proved to be dependent on the
multiplicity of infection (MOI) as well as on time post-infection as shown in Figs. 1
and 2.
Our experiments prove that cell size clearly increases with the time post infection and
with increasing MOI, becoming a potential early qualitative indicator of a successful
infection process. This qualitative indicator of infection could help in decisions

329

O.-W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 329-331.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

330
concerning virus transfection and propagation as well as during recombinant protein
production.

From qualitative monitoring to numeric estimation of the degree of infection
Based on our experimental data as well as on the statistical nature of the CSD
measurements a calculation method is proposed for numeric estimating of the degree of
infection in insect cell populations using its CSD.

331

Principles of the method:

• A population of insect cells
after infection can be regarded
as a mixture of infected and
uninfected (growing) cells.

• The fraction of cells in the
population, which can not be
statistically described as
uninfected cells, will be
considered as infected cells.

This method has been calibrated using samples of known degree of infection, resulting
in good linearity and reproducibility. However, interference could be expected from
other events affecting cell size (formation of cell aggregates, oxygen limitation,etc.)

Conclusions
From the changes in the cell size distribution of insect cells after baculovirus infection,
it was possible to develop a method of monitoring the degree of infection of insect cell
populations. Since the measurement of cell size distribution is a reproducible and rapid
technique, the detection of changes in cell size after infection represents a more simple
approach to evaluate the success of baculovirus infection in comparison with other
techniques such as PCR, plaque assay or detection of protein production. The
approach for numeric estimation of the number of infected cells presented in this work
is simple, reliable, and has a lot of potential applications in process design and control,
mathematical modelling and provides a tool for better understanding of the kinetics of
baculovirus infection.

Acknowledgement

We want to thank Schärfe System GmbH for providing financial support allowing E. Chico to attend ihe
15th ESACT Meeting.

DEAD CELL ESTIMATION - A COMPARISON OF DIFFERENT METHODS

A. FALKENHAIN , TH. LORENZ*,U. BEHRENDT*, J. LEHMANN
Cell Culture Technology, P. O. Box 100131, University of Bielefeld,
Germany
*Boehringer Mannheim GmbH, Nonnenwald 2, 82377 Penzberg,
Germany

1. Introduction

The success of a mammalian cell cultivation depends on the ratio of viable and dead
cells - the viability. The viability is routinely measured by the trypan blue exclusion
method [1]. In short cultivation processes the density of viable cells normally is high.
Therefore, the trypan blue exclusion method gives an exact value for the viable as well

as for the dead cells. For longer cultivation processes e.g. fed batch or dialysis
fermentations the trypan blue exclusion only shows a snapshot of the status of the
culture: cells which died previously might have lysed and cannot be determined

microscopically. Therefore, further methods were evaluated to determine the total
amount of cells which had been present during the cultivation process at a given time:
conductive electronical cell count (CASY 1®; Schärfe System GmbH, Reutlingen,
Germany) [2], biochemical determination of lactate dehydrogenase (LDH) [3] and
biochemical determination of DNA [4].

2. Material and methods

Murine hybridoma have been cultivated in RPMI based media in 1000 1-bioreactors [5,
6]. The cell densities were determined using the Trypan blue (Cat. No. 15250-061, Life
Technologies GmbH, Eggenstein, Germany) exclusion method and the CASY l® - Cell
Counter and Analyzer System (Schärfe System GmbH, Reutlingen, Germany).The
principle of this method is the change of the conductivity along an aperture during the
flow of a cell containing liquid. A pulse area analysis is performed. The result of a
measurement is a size distribution curve. A 150 µm capillary was used to detect
particles between 3.4 and 30 µm in diameter. The LDH activity was measured using a
test combination (Cat. No. 139 084, Boehringer Mannheim GmbH, Germany) which
has been adapted for the determination of LDH on a Hitachi 705 analyser. The DNA
concentration was measured by staining with bisbenzimide (Cat. No. B 1782, Sigma-
Aldrich Chemie GmbH, Germany). The DNA-bisbenzimide complex is detected with a
fluorescence photometer.

333
O.-W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 333-336.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

334
3. Results and discussion

For the quantification of dead cells by analysis of LDH activity and DNA concentration
their intracellular content had to be determined. The intracellular LDH activity is
instable during a cultivation and has to be determined for each sample of a fermentation
process. The intracellular DNA concentration is regarded to be constant and has to be
measured only once for a specific cell line.
The number of dead cells was calculated as follows:

Number of dead cells = c (LDH in supernatant) / c (LDH per cell)
Number of dead cells = c (DNA in supernatant) / c (DNA per cell)
where c means concentration. The size distribution curve of the CASY® method was
classified for three types of particles:
3.4 to 5 µm cell debris
5 to 10 µm dead cells and
10 to 30 µm viable cells.
Distribution curves for samples of a cultivation taken at different fermentation times are
shown in Figure 1A and 1B. During a course of fermentation the amount of cell debris
( 3 . 4 - 5 µm) as well as the amount of dead cells (5 - 10 µ m) increases significantly. For
both values a linear correlation versus the number of dead cells estimated by the LDH
method was observed. Since dead cells disintegrate in an unknown number of particles
it is difficult to assign the counts (3.4 - 10 µm) to a number of dead cells. Therefore the
counts in the range of 5 - 10 µm were used during the following considerations.

335
Figure 2 shows the time course of dead cells during a fermentation process determined
by the four different methods.

The dead cells measured by the trypan blue exclusion method show an inconsistent
course. The number is very low due to the fast disintegration of dead cells and depends
on the operator. The DNA concentration gives a higher number of dead cells during the
fermentation. During the last third of the process the concentration is constant which
might be due to an equilibrium between the release of DNA from the cells and the
degradation of DNA in the culture supernatant. The increase of the DNA concentration
is observed later than the increase of the LDH activity. This can be explained by the
higher stability of the cells' nucleus in comparison to the cells themselves. An increase
of the LDH activity during fermentation can already be measured after about 40 hours.
This shows the high sensitivity of this parameter. Since the enzyme is highly stable
(about 10 days in RPMI based medium) [3] cells which just have lysed as well as cells
which lysed previously are included. The density of the dead cells determined by CASY
1® shows a similar course in comparison to the LDH method. Slightly higher values
were obtained.

336

4. Acknowledgement

Many thanks to ESACT which offered a bursary.

5. References

1) Cook, J.A., Mitchell, J . B . (1989) Viability measurements in mammalian systems, ANALYTICAL
BIOCHEMISTRY, 179, 1-7

2) Winkelmeyer, P., Glauner, B., Lindl, T. (1993) Quantification of cytotoxicity by cell volume and cell
proliferation, ATIA, 21, 269-280

3) Goergen, J.L., Marc, A., Engasser, J. M. (1993), Determination of cell lysis and death kinetics in
continuous hybridoma cultures from measurement of lactate dehydrogenase release,
CYTOTECHNOLOGY, 11, 189-195

4) Loontiens, F . G . , McLaughlin, L. W., Diekmann, S., Clegg, R. M. (1991) Binding of Hoechst 33528 and
4', 6-Diamidino-phenylindole to self-complementary decadeoxynucleotides with modified exocyclic
base substituents, BIOCHEM1STRY, 30, 182-189

5) Comer, M. J., Kearns, M. J., Wahl, J., Munster, M., Lorenz, Th., Szperalski, B., Koch, S., Behrendt, U.,
Brunner, H. (1990) Industrial production of monoclonal antibodies and therapeutic proteins by dialysis
fermentation, CYTOTECHNOLOGY , 3, 295-299

6) Behrendt, U, Koch, S., Gooch, D. D., Steegmans; U., Comer, M. J. (1994) Mass spectrometry: A tool
for on-line monitoring of animal cell cultures, CYTOTECHNOLOGY, 14, 157-165

FED-BATCH CULTURE DEVELOPMENT
BASED ON BIOMASS MONITORING

Eric DE BUYL1, Alan MAXWELL2 and Luc FABRY1
1. SmilhKline Beecham Biologicals s. a.,

89 rue de l’Institut, B - 1330 Rixensart
2. University of Paisley, High Street, Paisley PA12BE, Scotland

INTRODUCTION

It is known that glucose level in the culture medium is an important parameter
influencing mammalian cells physiology (for inst., see ref.l); growth, lactate
accumulation, recombinant protein production and glycosylation are among the
parameters likely to be affected.

The aim of this work was to investigate the effect of glucose level on the growth
parameters of a recombinant CHO cell line expressing a viral glycoprotein (glucose
consumption, recombinant protein production); the development of a feeding policy
was also tested; the principle of this policy was based on the knowledge of the specific
consumption rate in various growth conditions and on the on-line measurement of the
biomass with an optical density sensor.

MATERIAL AND METHODS

Batch, fed-batch and continous culture were realised at the 1,5L scale (B.Braun
Biostat B bioreactors); the growth conditions were: proprietary protein-free medium,
temperature 33°C (see ref.3 for a discussion on growth at sub-physiological
temperatures), initial pH of 7,4, maintained above 7,2 through head-space aeration,
dissolved oxygen 35% of air saturation through O2 sparging. During fed-batches,
glucose level was maintained by glucose addition through manual adjustment of a
peristaltic pump setting, based on off-line glucose concentration measurement (YSI
glucose/lactate analyser). Seed culture were produced in disposable shake flasks (500
ml total volume, 150 ml of culture, 100 RPM), the PDL at the seeding being identical
for each bioreactor scale.

Fed-batches were also realised at the 10L scale, in a bioréactor (MBR) equipped with
a Wedgewood optical density sensor. Glucose level was maintained in the same way,
except during automated feeding trial, where the setting of the pump was based on the
optical probe reading through a calibration curve of glucose consumption vs. OD

337
O.-W. Merlen et al. (eds.), New Developments and New Applications in Animal Cell Technology, 337-342.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

338

The evaluation of the glucose effect was based on the determination of kinetic
paranatrdea;miqnelatcµe, grss/p1:e0cq6icsf,eislclpsle.ahccoiftuairct efgoplrurocqodpsu)ecatincoodnnosrufamttehpe(teirxoepnlerreaasstseee;doqfpi,npsrmpoetMecai/f1sie0c6 rec.protein production
cells.hour for qs and qlac
in the growth medium;
cathepsin L was chosen as a marker because it was demonstrated previously for other

recCHO lines that its release in the growth medium was strongly affected by growth

conditions, even for equally high viabilities during the culture (always higher than

90% during all the cultures).

The specific growth rate was estimated by fitting of the equation dX/dt = µX
(SigmaPlot software) to the data of cell count in function of time.

The recombinant protein was analysed by Elisa and Western Blotting (Pharmacia
Phast system), with detection either with a Mab or with various lectins, to obtain a
preliminary assessment of the quality of its glycosylation (kit Boehringer n° 1210
238).

RESULTS AND DISCUSSION

1. Glucose level effect on cell metabolism

The biomass, glucose, lactate and recombinant protein concentration profile of three
relevant 10L bioreactor trials are given in fig. 1.

339
The specific glucose consumption rate is decreasing during fed-batches (fig.3 a to d);
the mean value of is higher for higher glucose concentrations (fig. 2); however, the
value of is not very reproducible (trial E16 vs. trial E17 for instance); the effect of
glucose concentration is less clear on (contrary to what has been observed by others
(ref. 1); these tendencies are also observed during chemostat trials (fig. 2): is
reduced two fold and five fold during glucose-limited chemostat (residual glucose
0,19 mM), but is hardly affected by glucose limitation.

340

2. Development of an automated feeding policy

The linearity of the link between cell count and the reading of the Wedgewood probe
is well established (fig.4).

The specific glucose consumption rate is steadily decreasing during the cultures
(fig.3a to 3d).
A calibration curve, with plotted vs. optical density, was established on the basis of
the results obtained during trial E12/E13 (fig 5). However, it can be concluded that the
evolution of is not reproducible enough to be used in a robust automated glucose
feeding system: the glucose level during the automated feeding trial (trials El6 and
El7) increases indeed very quickly above the set point of 1.1 mM. The arrows
indicates the point where manual adjustment of glucose feed is resumed (fig. 1)

341
A slight tendency toward a decrease in protease release with decrease in glucose
concentration could be concluded from fed-batches results; this tendency is not

confirmed by chemostat results (fig. 6).

Although it gives only a limited insight into the glycosylation pattern of the protein,
the blots (fig.7) revealed with MAA (Maackia amurensis) lectin suggests that only in
the conditions of severe glucose limitation is the glycosylation affected (main band
detected being splitted in two sub-bands); the blots for all fed-batches are similar to

the one for chemostat at 8.5 mM residual glucose (results not shown).
Since the protein of interest here is heavily glycosylated, the determination of its
heterogeneity as described in ref.2 (densitometry of WB) is probably not possible
(detection with other lectins gives a smear, presumably because of an heterogeneous

glycosylation).

342

CONCLUSION
The glucose concentration in the growth medium influences specific glucose
consumption rate; however, a glucose limitation well below 1.1 mM is needed to
modify the specific growth rate or the lactate production rate; the limitation of lactate

production could be interesting to facilitate scaling up of the production, as far as pH
control is concerned.

The specific recombinant protein production rate is not affected by glucose limitation;
however, a detailed analysis of the glycosylation should be carried out on sample
produced under glucose limitation; a modification of the glycosylation in these
conditions is indeed detected by a simple lectin blot.
The extracellular protease release is hardly affected by the glucose level for this cell
line.
The growth is adequately described by fitting the cell counts with the equation dX/dt =

µX (exponential growth); again, between 1.1 mM and 16 mM, specific growth rate is
not affected by glucose concentration.
The cell count is well predicted by the reading of the Wedgewood probe; the specific
glucose consumption rate, however, is not reproducible enough to be used as part of a
glucose regulation loop; this lack of reproducibility is probably partly due to slight,
unnoticed, modifications of seed culture generation; a clear tendency of diminution of

with glucose concentration can be detected in the trials carried out at various
glucose concentration from 1.1 to 16 mM; however, this variation is not very
reproducible in our conditions, leading to the failure of the feeding policy tested.

KEYWORDS

rec CHO cell line, fed-batch, chemostat on-line measurement, optical density, specific
glucose consumption, specific recombinant protein production.

REFERENCES

1. Fieder, J., Schorn, P., Bux, R. and Noé, W., Cytotechnology 14 (suppl. l, 2.15), 1994
Increase of productivity in recombinant CHO-cells by enhanced glucose level

2. Hayter, P. et al., Biotech.Bioeng. 39, pp.327-335, 1992
Glucose-limited chemostat culture of CHO cells producing recombinant hu-INF

3. Kretzmer, G. et al., Proceedings of the 14th Esact meeting, p. 319, 1997
Temperature - A factor influencing cell behaviour

ON-LINE MONITORING OF PROTEIN AND SUBSTRATE/PRODUCT
CONCENTRATION IN MAMMALIAN CELL CULTIVATION PROCESS

H.Lübben, J. Hagedorn and T. Scheper
Institut für Technische Chemie, 30167 Hannover, Germany
Introduction
The characterisation and optimisation of mammalian cell culture process requires an
on-line monitoring of product as well as low molecular weight concentration. For this
purpose an automated analyzer based on the principle of flow injection analysis (FIA)
was developed. It is an modular and flexible three channel set-up containing pumps,
different valves, oxygen electrodes with amplifiers, a fluorescence detector and space
for buffer reservoirs. The system can be used to monitor protein concentration via
immunoassay as well as substrat and product concentration by enzymatic reaction.
Heterogenous immunoassay
For the heterogenous immunoassay product depending antibodies or protein G are
immobilized on a polymer loaded in a flow-through cartridge.

After sample injection all target proteins bind to the antibodies and will be eluted after a
washing step. The protein fluorescence is monitored at 280 nm (exc.) and 340 nm
(em.), so that labeling of the analyte is not necessary. The advantages of this set up are:
short analysis time in the range of 6-8 minutes for each cycle, no sample dilution is
necessary. Furthermore the immuno components can be used several times, which is
very economically due to the high costs of specific antibodies.

343
O.-W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 343-346.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

344

Enzymatic Analysis
For the analysis of low molecular weight metabolites, enzymes are immobilized on a
epoxy activated support material. The detection is carried out by amperometric
measurement of oxygen consumption. This method allows simultanous determination of
different analytes such as glucose, maltose, L-amino acids, saccharose and L-glutamine.

345
Sampling

To obtain a representative, cell free sampling from the reactor for FIA analysis, a
tubular in-situ filtration probe was applied during cultivation process (ESIP, Eppendorf-
Netheler-Hinz, Germany). The sampling module is compatible with 19 mm or 25 mm
standard port of a bioreactor. For the module a polypropylene micro filtration
membrane with a pore size between 0,2 and 0,6 µm is used, providing a sterile barriere
that cannot be penetrated by any microorganism. The module contains a small dead
volume of less than 2 ml, a large filter surface of nearly 40 qcm and new membrane
sealings made of PTFE ferrules well, known in HPLC technique. Continuous and
discontinous performance with flow rates up to 2 ml/min are possible. The sampling
device can be sterilised together with the culture medium inside the reactor.
Experimental data
The major advantages of both techniques are their high selectivity, the short analysis
time (6-8 or 3 minutes) and that the immobilized components can be used several times.
Product concentration:

The comparison of HIA showed a good correlation to ELISA data with a deviation
below 5 %. A higher accuracy was achieved (CV < 5%) using a triplicate sample
determination. The HIA detection in this case was performed off-line. A higher
frequency seemed to be not appropriate regarding the low specific production rates of
animal cells. Using the ESIP sampling module the detection frequency can be increased
up to every 6 minutes. In order to preserve the activity of the immobilised antibodies a
frequency every 30 minutes is recommended.

346

Substrate concentration:
As a main substrate in all mammalian cell culture processes glucose was chosen as a
target for on-line monitoring and as a modell system.
The comparison of on-line and offline data (YSI 2700 analyser) showed sufficient
correlation of the concentration values. See Figure 5.

The interruptions and interferences pointed out as (1) (2) and (3) were caused by air
bubbles on the amperometric electrode or by blocking of the tubing after microbial
contamination.
Aknowledgements
The authors would like to thank the European Commission Directorat General XII for
supporting the participation in this meeting and the presentation of our research.
Reference

Hilmer, J.-M., Scheper, Th.: A new Version of an In situ Sampling System for Bioprocess Analysis. Acta
Biotechnol.16 (1996) 2-3

PROCESS CONTROL AND ON-LINE FEEDING STRATEGIES FOR
FED-BATCH AND DIALYSIS CULTURES OF HYBRIDOMA CELLS

J. O. SCHWABE, R. PÖRTNER

Technische Universität Hamburg-Harburg, und

Bioverfahrenstechnik, Denickestr. 15, D-21071 Hamburg, Germany

1. Abstract

The concept of the fed-batch strategy was to minimise the formation of inhibiting
metabolites and to increase the yield of monoclonal antibodies by carefully supplying
substrates. A process control system based on fieldbus technology was used for
monitoring and control. External program routines were implemented to control

dissolved oxygen (DO) and to calculate the oxygen uptake rate (OUR) and cumulative

oxygen consumption (COC) simultaneously. Concentrated feed solution was supplied
according to the off-line estimated stoichiometric relation between oxygen and glucose
consumption (GC). The developed feeding strategy initiated feeding automatically when
the OUR decreased because of substrate limitation. Hybridoma cells were cultivated in
batch and fed-batch cultures in laboratory scale using media containing an iron-rich,
protein-free supplement. The antibody concentration increased three- to four-fold
compared to the conventional batch culture when applying this strategy. It was not
possible to avoid inhibition by metabolic products such as lactate and ammonium during
fed-batch phase. This was accomplished by using dialysis and ‘nutrient-split’ feeding in
a membrane dialysis reactor [Pörtner et al., 1997] and resulted in a ten-fold increase of
the antibody concentration compared to the batch.

2. Cell Line and Culture Conditions

The hybridoma cell line IV Fl9.23 is producing monoclonal IgG1 antibodies (MAb)
against penicillin-G-amidase for the application in an affinity chromatography.
Cultivation was carried out in a 1:1 mixture of Iscove’s MDM and Ham’s F12. The
media was completed with and 1 % (v/w) of a protein-free, iron-rich supplement
and Dolníková, 1991]. During the fed-batch phase and for ‘nutrient-split’ feeding a 10-
fold concentrated, salt-free nutrient solution containing amino acids and vitamins with
50 mmol l-1 glucose and concentrated glutamine solution (200 mmol l-1, Life
Technologies were supplied separately. Dialysis cultivation was carried out in a
membrane dialysis reactor (Bioengineering AG, CH) described by Pörtner et al. (1994).

347

O.-W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 347-349.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

348
3. Process Control System
The process control (Fig. 1) was based on a field bus technology under a Unix-
compatible real-time environment. Control platform (X-Control, Direct Digital Control)
and data management (X-Manager) were working independently for process safety
reasons.

4. On-line Characterisation using OUR
On-line characterisation is a basic approach to describe metabolism and physiological
state of the cells on-line. A process control system stores and controls measurable data.
An user-interface enables the operator to change process parameters and to provide data
from off-line analysis. Non-measurable variables, e.g. the oxygen uptake rate OUR and
cumulative oxygen consumption COC, were calculated to describe the current
metabolism. Based on these variables the feed flow was calculated on-line. Metabolic
rate ratios and stoichiometric coefficients were calculated from off-line analysis. The
stoichiometric coefficient described the ratio of oxygen and glucose consumption.
This parameter was not constant during cultivation and was therefore regularly updated
from off-line glucose analysis.

349

5. Results

5.1. FED-BATCH STRATEGY

The feeding strategy was based on the coupling of on-line and off-line measurements

and enabled the fed-batch operation without expensive on-line substrate analysis. The

cell growth was characterised on-line using the oxygen uptake rate OUR and the feed
was supplied based on the stoichiometric relation of glucose and oxygen uptake.

The viable cell concentration and oxygen uptake rate OUR of batch, fed-batch and
dialysis fed-batch are shown in Figure 2. The cell concentration in fed-batch increased
1.5-fold and the antibody concentration about 3-fold compared to the conventional
batch. The fed-batch strategy was not able to avoid inhibition by accumulated
metabolites after 80 h cultivation time.

5.2. DIALYSIS FED-BATCH

The removal of inhibiting metabolites such as ammonia and lactate from the cultivation
chamber using a dialysis membrane resulted in a 10-fold
increase in the viable cell and antibody concentration compared to batch cultivation.
Substrate utilisation was very efficient due to nutrient split feeding strategy. The
concentrated feed was supplied to the culture chamber and buffer solution was used as
dialysate in the dialysis chamber.

6. References

and Dolníková, J. 1991. Hybridoma growth and monoclonal antibody production in iron-rich
protein-free medium: Effect of nutrient concentration. Cytotechnology 7: 33-38
Lüdemann, I., Pötner, R. and Märkl, H. (1995) ‘Nutrient-Split’ feeding strategy for dialysis cultures of
hybridoma cells. Proceedings 8th JAACT meeting, Iizuka, Japan
Pörtner, R., Bohmann, A., Lüdemann, I. and Märkl, H. (1994) Estimation of specific glucose uptake rates in
cultures of hybridoma cells. Journal of Biotchnology 34: 237-246

METABOLIC NETWORK ANALYSIS OF A HYBRIDOMA CELL LINE
USING MASS BALANCES AND C14-LABELLED GLUCOSE AND
GLUTAMINE

P.-A. RUFFIEUX, I.W. MARISON, U. VON STOCKAR
Swiss Federal Institut of Technology
LGCB - 1015 Lausanne, Switzerland

1. Introduction

Continuous cultures of a hybridoma cell line (Zac3) were performed with two C14
labelled substrates: glucose and glutamine. A simplified network was used to describe
the metabolism of the cells [1], see figure 1. The measurement of the main species
consumed or produced by the cells allows one to solve the network and determine the
rate of the 20 reactions.
The use of labelled glucose and glutamine allows determination of very small fluxes
with high accuracy. Thus it was possible to determine the flux of carbon to biomass and
also the ratio of substrate which is degraded into carbon dioxide from glucose and
glutamine. These results could be confirmed from those obtained by the network, for
example the CO2 production rate.

2. Materials and Methods

A hybridoma cell line (ZacJ) was grown into a CSTR of 1.5 liter working volume.

When the stationary phase is reached, the standard medium is changed to labelled

medium, with exactly the same composition, the only difference is that a part of the

glucose or glutamine is radioactive. The labelled atoms of carbon are followed in the

different phases: the biomass is collected and the activity measured, the exiting the

reactor is trapped into NaOH, and the activity of the medium is determined after liquid

chromatography, which separate the different metabolites (glutamine, lactate,

alanine.etc). This method is describe elsewhere[2].

The specific rate of consumption or production of all the major components is

determined according to the following equations: (1) IN-OUT + PRODUCTION =
ACCUMULATION;

Where : F=Flow rate of species i
X=biomass concentration consumption rate of
Production rate of

351

O. - W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 351-353.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

352
The specific oxygen uptake rate is determined with a balance around the liquid phase.
With a constant D.O. and a determined experimentally, it is possible to calculate the
oxygen transfer rate from a knowledge of the molar fraction of oxygen in the gas
phase[3].

3. Results
The concentration of the main metabolites was determined by off-line analysis and the
specific rates of consumption/production are summarised in table 1:

353

The use of glucose and glutamine enables

determination of the fate of the C labelled atoms [2]. This
supplementary information allows confirmation of the

model used for the network and to determine some fluxes

which are difficult to measure :

production rate, due to the use of buffer

to control pH.
• Carbon assimilation into biomass, not possible to
determine without tracers.

With the stoichiometric coefficients of the reactions in the
network, a matrix A is defined and Ar=q. With r being a
vector of the flux through the biochemical reaction and q
the measured rates of change of extracellular metabolites.
By solving this system, the rates r1 to r20 are obtained

measured =
calculated =
Thus 92% of the 14C in glucose goes to lactate.
In the network, this value could be approximate by the
calculation of r 5 / r 4 = 89%

4. Conclusions

This simple metabolic network gives a good idea of the main fluxes into the cell,

especially for the catabolic reactions. This calculation requires only 7 rates of

consumption.
The use of labelled glucose and glutamine enables the metabolic routes to be

followed with great accuracy, especially production rate. Normally this rate is

difficult to measure due to buffer system used to control pH. Therefore it

was possible to validate the results of the network.

To use completely the information provide by the C-labelled experiments, it is

necessary to write equations, which take into account the position of each C-atom in all
reactions (Mapping equations). However, such networks are very complicated and it is

difficult to perform the required measurements, since this would involve the use of

substrate with the C-labelled at different positions of the structure, together with

analysis of the position of the label in the product.

5. Reference

1. Zupke C. et. al. Intracellular flux analysis in hybridomas using mass balances and in vitro 13C NMR
Biotechnol. Bioeng.(1994) Vol. 45 Pp. 292-303
2. Ruffieux P-A. et. al. Development of carbon balances for continuous animal cell culture using 14-C
labelled glucose, AOnnimlinael cGealls-tAecnhanloylsoisgyin, (1997), Pp. 725-729
3. Oeggerli. A. et al. Animal-Cell Cultivation .1. Control of Dissolved-Oxygen and
pH. Biotechnol. Bioeng., (1995) Vol 45. Pp. 42-53.

DYNAMIC MEDIUM OPTIMIZATION BY ON-LINE HEAT FLUX
MEASUREMENT AND A STOICHIOMETRIC MODEL IN MAMMALIAN
CELL CULTURE

Y. GUAN and R. B. KEMP
Institute of Biological Sciences, The University of Wales
Aberystwyth, SY23 3DA, UK

Abstract. Metabolic flow rate was detected on-line and in real time by a Heat Flux
Probe for animal cell culture. For a model system, CHO320 cells producing
recombinant human Interferon-γ (IFN-γ ), this probe identified different metabolic
states during cell growth. These were then related to dynamic changes in glucose and
glutamine consumption, which would allow optimization of the feeding medium.
Based on heat flux measurement and the collected data from a stoichiometric model,
dynamic optimization of the two substrates in a feeding medium would then be

possible by automatically triggering two peristaltic pumps.

1. Introduction

To increase the production duration and the culturing density of animal cells, an
important factor is to feed the cells with the correctly formulated medium. Except for
continuous culture, cellular metabolic activity varies during the culturing time and
therefore the composition of the medium should be changed to meet the dynamically
altered optimal medium requirement [ 1.2].

To achieve this aim. a direct measurement of cellular metabolic activity is
necessary. Cells produce heat as an integral part of their metabolism and its flow rate
thus is a measure of total metabolic rate. Heat flow rate is measured on-line by
calorimetry and its quantity per amount of cell mass is known as heat flux. Among
other methods dielectric spectroscopy was chosen to measure cell mass in real time.

2. Experimental

CHO320 cells were grown in an Applikon 3-L bioreactor system with BioXpert
software for data acquisition and controlled medium feeding. Measurement of heat
flow rate was realized by an ex situ flow microcalorimeter and that of viable cell
biomass by an in situ dielectric Viable Cell Monitor [3,4].

3. Results and Discussion

The data depicted in the figure are used to justify, by a thermochemical approach, the
following growth equation:

355

O.-W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 355-357.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

356
The growth reaction is easily be characterized by its set of stoichiometric coefficients,

An important concept in the present work on the stoichiometric growth equation is that
the set (matrix) of stoichiometric coefficients has a one-to-one corresponding
relationship to the metabolic status of the entire cell population [5]. If the metabolic
status is expressed by the metabolic rate, this means that:

(3)

where is advancement of
eq. (1) and is the meta-

bolic rate in terms of the
growth equation. Eq. (3) clear-
ly delivers the experimental
strategy for optimizing cell
growth and the production of
target proteins, i.e. the on-line
diagnostics of cellular meta-
bolic requirement can be
achieved by on-line mea-
surement of heat flux. For em-
pirically divided time intervals
in growth, the data are
incorporated into the equation
to obtain stoichiometric coe-
fficients, see table 1 below.

The results show that the heat flux probe is an early and sensitive indicator of the
changing metabolic state of cells. Since the production of the target protein deteriorates
with these changes, action must be taken to retard the metabolic decline. Analysis of
stoichiometric coefficients at different time intervals showed the changing cellular
requirement for the relative amounts of major catabolites.

357

For the present purpose, metabolic states are considered to be directly associated
with the cellular metabolic rate whereas external physiological conditions are the
extracellular environmental conditions in which the cells grow. Although different,
they decrease in the same direction for a deleterious physiological condition, which
thus will negatively affect the metabolic state, reducing the metabolic flux either for
the cell growth or for the target protein production, or both. This implies it is possible
to detect worsening physiological conditions using heat flux and then decide when to
add fresh medium. The classical approach to this problem is fed-batch culture. The
difficulties here are to know when to add more nutrient and in what quantities. The use
of on-line heat flux measurement can solve these by at least the following: (a) timing of
feeding — as the heat flux curve decreases to a certain extent, medium is automatically
fed by a control loop; (b) quantity of feeding — the basic approach would be to feed the
cells with the correct relative amounts of the major substrates. For instance, to
maintain the maximal growth of CHO320 cells, glucose and glutamine should be fed
by a molar ratio of 2.7 to 1 in the fast growing period (Table 1).

A more refined approach would be to feed the cells continuously at a rate
determined by heat flux and dependent on the consumption rate of major substrates in
their correct molar ratio. The same approach would be possible for the same cells
grown on macroporous beads. Furthermore, since the increase in the proportion of

unwanted heterogeneous proteins is also a reflection of worsening physiological
conditions, it is likely that heat flux can also be used to monitor the quality of the
products, typically glycoproteins

The heat probe in this work is a robust, reliable and novel biosensor for on-line use
in cell culture. It detects total metabolic rate and changes in it. which are related to
alterations in substrate consumption rates leading to their depletion. The stoichiometric
approach has shown the actual substrate requirement of CHO320 cells for growth. In
the fed-batch method advocated here, the cells are given substrates in the molar ratio
determined by the stoichiometric coefficients. In recombinant cells, this would sustain
production of foreign proteins and should lead to much needed improvements in
quality.

Acknowledgments: Y.G. was the recipient of a generous bursary from ESACT. The project is funded by the
Biotechnology and Biological Sciences Research Committee (UK) under grants 2/3680 and 2/TO3789.

4. References

1.Kemp, R.B., Evans, P.M. and Guan, Y. (1997) An enthalpy balance approach to the study of metabolic activity
in mammalian cells. .J. Thermal Anal., 49, 755-770.

2.Kemp. R.B. and Guan, Y., (1997) Heat flux and the calorimetric-respirometric ratio as measures of catabolic
flux in mammalian cells, Thermochim. Acta 300, 199-211.

3.Guan, Y., Lloyd, P.C., Evans, P.M. and Kemp, R.B. (1997) A modified continuous flow microcalorimeter for
measuring heat dissipation by mammalian cells in batch culture, J. Thermal Anal., 49, 785-794.

4.Guan, Y., Evans, P.M., Kemp, R.B. (1998) Specific heat flow rate: An on-line monitor and potential control
variable of specific metabolic rate in animal cell culture that combines microcalorimetry with dielectric
spectroscopy, Biotechnol. Bioeng. (in press).

5.Guan, Y., Kemp, R.B., (1996) Medium design with the aid of heat flux measurement in mammalian cell
culture, in BioThermoKinetics of the living cell, eds. H.V. Westerhoff, J.L.Snoep, F.E.Sluse, J.E.Wijker and
B.N. Knolodenko, pp. 387-397, BTK, Amsterdam.

MODELING OF GLYCOPROTEIN PRODUCTION BY CHINESE HAMSTER
OVARY CELLS FOR PROCESS MONITORING AND CONTROL

J. STELLING*, R. K. BIENER*, J. HAAS†, G. OSWALD†,
D. SCHULLER† W. NOE† and E. D. GILLES*
* Institut für Systemdynamik und Regelungstechnik, 70550 Stuttgart, FRG
†Dr. Karl Thomae GmbH, 88397 Biberach, FRG

Abstract

Substantial improvements of animal cell culture process performance can be achieved
by model-based methods of process optimization and control. Here, a structured mathe-
matical model describing the dynamics of Chinese hamster ovary (CHO) cell growth and
recombinant glycoprotein production has been developed. It covers the main aspects of
cell metabolism as well as aspects linked to cell cycle regulation. Model predictions fit
well to experimental data obtained from fed-batch cultures and allow for insight into the
complex behaviour of CHO cells in culture. Model-based improvement of feeding trajec-
tories is shown. Further applications for process monitoring and control are discussed.

1 Introduction

In animal cell culture technology, substantial improvements of process performance can
be achieved by model-based methods of rational medium design, feeding strategy opti-
mization and on-line monitoring and control. When applied to fed-batch processes, these
methods can enhance growth and production efficiency by reducing growth limitations
due to depletion of essential nutrients (e.g. glucose and amino acids) and / or accumulation
of toxic by-products (e.g. ammonia and lactate). [1,2]
Whereas most research in this area is focused on monoclonal antibody production by
hybridoma cell lines [1,2], here a structured mathematical model describing the dynamics
of CHO cell growth and recombinant glycoprotein production has been developed. As an
exemplary application, the model-based improvement of feeding trajectories in laboratory
scale fed-batch fermentations is shown.

2 Model structure

The CHO model is composed of two main modules: A cell cycle part including the
subpopulations of phase cells (producers), phase cells (non-producers) and
dead cells is used to describe population dynamics. Phase transitions are assumed to be

359

O.-W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 359-361.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

360
governed by the accumulation of cell-cycle phase specific cyclins and by cell size. The
mathematical formulation results in the two-dimensional population balance

(1)
with f: number density function, cell volume, intracellular cyclin concentration,

velocities, source/sink, subpopulation index and time.
The complementary metabolic part has been designed as a cybernetic model. For a close
representation of cell physiology, the model includes the main metabolic pathways (gly-
colysis and PPW, citrate cycle, amino acid metabolism), metabolite production, formation
of cellular components (proteins, lipids, nucleotides) and product, energy metabolism
(ATP, NADH) and growth control by essential amino acids / toxic by-products [1].

3 Simulation results
A series of CHO fed-batch fermentations has been used for parameter estimation. As
shown in Fig. 1, simulation results agree well with experimentally observed time series.

Analysis of model simulations furthermore allows for the identification of growth-limiting
steps, e.g. amino acid uptake rates and ATP generation as well as the plausible prediction
of unmeasured variables, e.g. share of subpopulations and average cell size (Fig. 2).

361

4 Application: Feed design

The model-based optimization of the feed composition lead to a feed medium including
lower glutamine concentration and higher levels of specific essential amino acids.

As a result - compared to a standard fed-batch fermentation - significantly higher cell

yields and an increase in final product concentration have been achieved (Fig. 3).

5 Concluding remarks

The mathematical model presented is able to give a good description of CHO growth
and production dynamics. Compared to purely metabolic models [1,2] or other combined
cell-cycle / metabolic approaches [3], it provides a closer, modular representation of many
aspects of cell biology. Model-based feed design serves only as one component of an inte-
grated approach to process improvement. Further steps include complete off-line process
optimization, on-line monitoring and (model-predictive) control [1].

6 Acknowledgements

This work was supported by the State Government of Baden-Württemberg (FRG) and Dr.
Karl Thomae GmbH (Biberach, FRG).

7 References

[1] Biener, R.K. et al. (1996) Model-based control of animal cell cultures, in M.J.T. Carrondo,
B. Griffiths and J.L.P . Moreira (eds.), Animal cell technology: from vaccines to genetic
medicine, Kluwer Academic Publishers, pp. 639-645

[2] Xie, L. and Wang, D.I.C. (1996) High cell density and high monoclonal antibody production
through medium design and rational control in a bioreactor,Biotechnol. Bioeng. 51, 725-729

[3] Martens, D.E. et al. (1995) A combined cell-cycle and metabolic model for the growth of
hybridoma cells in steady-state continuous culture, Biotechnol. Bioeng. 48, 49-65

THE TEMPERATURE EFFECT IN MAMMALIAN CELL CULTURE
AN ARRHENIUS INTERPRETTION

Gerlinde Kretzmer1, Torsten Buch1, Konstantin Konstantinov2, David
Naveh2 1Institut für Technische Chemie, University Hannover,
Germany 2Bayer Corporation, Pharmaceutical Division, Berkeley,
USA

Introduction

Recently cultivation temperature has become of interest for process characterisation and
optimisation for research and manufacturing.
All chemical and biochemical reaction are known to be temperature dependent.
Therefore, one expects reactions rates going down with reducing the cultivation
temperature. Results show that the reduction in rates is different for different reactions.
The primary metabolism is influenced by reducing the temperature, while the secondary
metabolism especially the protein production is not effected or even improved. Cell
growth rate as a bulk of reactions taking place is slowed down with temperature
reduction in most cases. The extend to which the reduction will occur depends on the
specific reaction studied.
All reactions follow the thermodynamical rules and Arrhenius was the first to set up a
mathematical equation to explain the relationship between chemical reaction rates and
temperature.

Material and Method
Data from Batch cultures of different cell lines were used for the calculations: adherent
BHK 21, suspension rBHK, rCHO I, rCHO II, hybridoma. The cultivations of the
adherent BHK were carried out in serum containing medium, the other in serum free
medium.
Data from continuous cultivation of rBHK (FVIII) were used.

For many reactions the rate expression can be written as a product of a temperature-
dependent term and a composition term, or

ri = f1 (temperature) * f2(composition)

= k*f2(temperature)

363

O.-W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 363-366.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

364

For such reactions the temperature-dependent term, the reaction rate constant, has been
found in practical all cases to be well represented by Arrhenius´law:

This was found true also for enzyme-activity-temperature relationship.
For a small temperature range Arrhenius´law is also applicable for microbial
cultivations.
For example the cell growth can be described as follows:

This can be plotted as an Arrhenius plot:

This equals

Similar equations can be set up for metabolic reactions and for product formation.
With quasi-linear regression of the following form

the activation energy and the frequency factor can be calculated.

Results and discussion
Growth rate (Fig. 1)
All cell lines show a decrease in growth rate with decreasing temperature. The non
recombinant BHK and the hybridoma show a steady decrease whereas the recombinant
cell lines show stable or better growth rate between 37 and 33°C.

365

Arrhenius plot growth rate (Fig. 2)
All cell lines show two different slopes in their Arrhenius plots. The plot of the
recombinant cell lines and the hybridoma have an area with a steep slope and one with
a flat slope. The non recombinant BHK cell line has two areas with nearly the same
slope, separated by a plateau.
Arrhenius plot growth rate and oxygen uptake rate of rBHK (Fig. 3)
Again a two stage relationship occurs for the growth rate as seen with other
recombinant cell lines. The relationship between ln (OUR) and 1/T is linear.

366

Arrhenius plot glucose consumption rate and productivity of rBHK (Fig. 4)
A linear relationship of ln (Glucose) and 1/T occurs. Specific glucose consumption rate
is decreasing with decreasing temperature. The relationship between ln(Productivity)
and 1/T is also linear but the slope is reverse. The specific productivity is increasing
with decreasing temperature

Conclusions
The Arrhenius plot of the growth reaction is only in very small temperature range
linear. For all the recombinant cell lines the activation energy at higher temperature
differs from that a lower temperature. In between there is a temperature range where
the activation energy does not change with temperature. There is independent from
the activation energy.
The kinetic reactions of glucose consumption, lactate production and oxygen uptake
follow Arrhenius´ law linearly. The activation energy of those reaction lies in the range
of 80 to 140 Ws/mol.
The Arrhenius plot of the production reaction shows a negative slope at lower
temperature. This would lead to a negative activation energy which is thermodynamical
senseless.
The product formation is a very complex system build up of many single reaction steps.
From the Arrhenius plot one can assume that at higher temperature one of the reactions
must be inhibited . This occurs also by the cell growth: If the temperature is high
enough the growth decreases because of the denaturation of proteins.

Summary
Cell growth and the primary metabolic reactions can be represented by Arrhenius´ law.
That means Arrhenius´ law can be used for a process model in a small temperature
range. For the product formation this simple approach is not sophisticated. The kinetic
approach for the production has to be studied furthermore.

INTEGRATED PROCESSES
AND

SCALE-UP

DIELECTROPHORETIC FORCES CAN BE EXPLOITED TO INCREASE
THE EFFICIENCY OF ANIMAL CELL PERFUSION CULTURES

N. KALOGERAKIS1, A. DOCOSLIS2 AND L. A. BEHIE3
1Laboratory of Biochemical Engineering & Environmental Biotechnology
Technical University of Crete, Chania 73100, Greece
2Bioengineering Laboratory, Dept. of Chemical Engineering
State University of New York, Buffalo, NY 14260 USA
3Pharmaceutical Production Research Facility (PPRF), Faculty of

Engineering The University of Calgary, Calgary, Alberta, Canada T2N 1N4

1. ABSTRACT

Dielectrophoresis is a well established and effective means for the manipulation of viable
cells. Various applications have been found, ranging from electrofusion, to individual cell
manipulation, and to differential separation from cell mixtures. Its effectiveness,
however, greatly depends upon the utilization of very low electrical conductivity media.
High conductivity media, as in the case with cell culture media result only in negative
dielectrophoresis (i.e., induction of weaker repulsive forces) and excessive medium
heating. Recently, a dielectrophoresis-based cell separation device (DEP-filter) has been
developed for perfusion cultures that successfully overcomes these problems and
provides a very high degree of viable cell separation while most of the nonviable cells are
removed from the bioreactor by the effluent stream. The latter results in high viabilities
throughout the culture period and minimization of lysed cell proteases in the bioreactor.
However, an important question that remains to be answered is whether we have any
adverse effects by exposing the cultured cells to high frequency dielectrophoretic fields
for extended periods of time. A special chamber was constructed to quantitate the effect
under several operational conditions. Cell growth, glucose, lactate and monoclonal
antibody production data suggest that there is no appreciable effect and hence, operation
over long periods of time of our DEP-filter should not have any adverse effect on the
cultured cells.

2. INTRODUCTION
Perfusion cultures represent the most cost effective mode of operation for the large scale

production of therapeutic proteins, such as HBAg, tPA, erythropoietin, monoclonal
antibodies, etc. The main characteristic of a perfusion culture is the retention of the cells in
the bioreactor, i.e. no cells are removed by the effluent stream which results in high cell
densities, increased volumetric productivity and reduced downstream purification
requirements. A suitable cell retention device, located in the effluent stream of the
bioreactor, is an important part of a cell perfusion bioreactor system. Existing cell
retention systems are based on sedimentation, centrifugation or conventional filtration
techniques and are subject to significant limitations. A recent device3, 17 that employs

ultrasonic resonance fields to transiently aggregate animal cells and are intermittently
settled back in the bioreactor has also several limitations. The main disadvantage being its
inability to selectively separate viable from nonviable cells and second the high power

369
O.-W. Merten et al. (eds.), New Developments and New Applications in Animal Cell Technology, 369-375.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.

370

requirements that can cause excessive medium heating with possible detrimental effects to
cellular viability.

As an aelxteprlnoaittaivtieotnootfhde ieexleiscttirnogphdoevreictiecs,(DwEe Ph)avfoercdeesv4e. lDopieeldecatcroepllhroerteesnitsioisntdheevmicoetiboanseodf
on the
neutral particles (e.g., spheres, bubbles, biopolymers, biological cells, etc.) under the
influence of a divergent electric field. Numerous techniques have been developed using
dielectrophoresis that include electroporation,37 electrofusion,1,14,18 electroinjection,13
measurements of the dielectric properties of the cell,6,7,10 as well as cell separation in low
conductivity media.2,8,1112,15 We have recently shown that there is a frequency range
where selective separation of the viable from non-viable cells is achievable in high
conductivity media typically found in cell culture systems4. As Fuhr et al.5 has also
shown using anchorage dependent mouse fibroblasts, cell separation in highly conductive
culture media can only be achieved under negative DEP, using high frequency AC fields.
Their results were also encouraging in that no adverse effects due to continuous exposure
of the cells to high frequency electric fields were detected. They found that high field
frequencies (above 10 MHz) had no significant effect on cell growth.

In this present work, we examine whether there are any adverse effects by exposing the
cultured cells to high frequency dielectrophoretic fields for extended periods of time. A
special chamber was constructed to quantitate the effect under several operational
conditions. Cell growth, glucose, lactate and monoclonal antibody production data are
presented here that have been obtained from cultured cells exposed to high frequency
dielectrophoretic fields over long periods of time. Finally, the inherent capability of the
DEP-filter for preferential removal of apoptotic bodies is also discussed.

3. MATERIALS AND METHODS
The murine lymphocyte hybridoma HFN 7.1 producing an IgG antibody reactive with
human fibronectin (ATTC: CRL-1606) was used in these experiments. The cell culture
medium was DMEM (Dulbecco's Modified Eagle's Medium, Sigma) supplemented with
10% Fetal Bovine Serum (Gibco/BRL). In addition CHO/dhfr- cells (ATCC: CRL-9096)
were grown in EMDM (Iscove's modified Dulbecco's medium) supplemented with 4 mM
L-glutamine, 0.1 mM hypoxanthine and 0.01 mM thymidine and adjusted to contain 1.5
g/L sodium bicarbonate and 10% FBS after adaptation to grow in suspension.. The cells
were grown in suspension in a suitably modified (open bottom) polycarbonate vial at
37°C in a 5% and humidity controlled incubator. The bottomless vial was glued onto
an oxidized silicon wafer where a grid of micro-electrodes were deposited. In this case the
silicon was not etched through in the areas between adjacent electrodes as the case is for
the construction of the DEP filter. We simply wanted to grow the cells in the vicinity of
the electrodes and quantify any adverse effects from the exposure to the high frequency
electrical fields. The analysis of glucose and lactate was done by an YSI-2700 analyzer,
the aminoacids by HPLC (HP1090), the MAb by ELISA and cell viability by trypan blue.

4. THEORY

The net time-averaged induced dielectrophoretic force for AC electric fields is directly
proportional to the cell volume, the real part oEf t2he(rmmsedviaulmuep),erim.e.i,ttivity, and the
gradient of the electric field intensity squared,

371
where r is the cell radius and is the intensity of the electric field. The frequency
dependence of the induced DEP force is given by the dimensionless Clausius-Mossotti
function,

where the underlined parameters denote complex quantities. The complex permittivity of
the surrounding medium is given by

where and is the angular frequency of the applied field. According to the
single-shell model,9,10 the cell can be represented as an ohmic, spherical particle,
enclosed by a thin insulating shell (cellular membrane). The transmembrane conductance
and surface conductivity are assumed to be negligible since mammalian cells do not have a
cell wall and their membrane thickness is at least three orders of magnitude smaller than
their cell radius15,16. If, furthermore, the cell cytoplasm is modeled as a linear ohmic
dielectric fluid (i.e., there are no dielectric losses) of permittivity and conductivity
the resulting effective cell permittivity is given by9

where is the area-specific membrane capacitance and are time
constants. As shown in Equation (1), the real part of is directly related to
The sensitivity of to dcieslclussiszeedanedlsveawrihoeurse4o.thHeraveliencgtriecsptiamraamteedtetrhseovuenr a wide
have
frequency range been known
parameters in through a series of levitation experiments, one can plot
versus and determine the frequency range where we can have preferential cell retention
of viable cells ninontvhieabbiloerceealclstobryththroeuegfhflutheentascttrieoanmo4f. negative dielectrophoretic forces and
withdrawl of

5. DESCRIPTION OF THE DEP-FILTRATION SYSTEM

The DEP-filter, shown in Figure la, can be briefly described as a grid of micro-electrodes
deposited on a silicon substrate, with the silicon etched through in the areas between
adjacent electrodes. The filter was made using photolithography and silicon
microfabrication dteetcahilnsiqcuanesb. eTfhoeunedlecetlrsoedwehseraer4e. made of gold deposited on a chromium
layer. Structural The DEP-filter is mounted at the bottom
of a housing device and the whole unit is submerged in the cell suspension. A schematic
diagram with details of the whole system is given in Figure 1. A schematic of the forces
acting on viable and nonviable cells as well as their expected trajectories in the vicinity of
the DEP-filter is shown in Figure 2.

Our first results have shown that, at low flow rates, the retention of the viable cells can be
very high (up to 98%) whereas it is extremely low, often below 15%, for nonviable cells
regardless of the operating conditions. The latter is an indication that cell debris does not
experience any dielectrophoresis. It was also shown that the retention of viable cells is
subject to many factors, such oafstfhreeqeuffelnuceyntofsttrheeamap4p. lied electric field, voltage across the
filter electrodes and flow rate

372

6. EXPERIMENTAL RESULTS & DISCUSSION

A series of experiments were designed to examine whether cells staying in the vicinity of
the DEP-filter electrodes suffer from adverse effects. In a typical perfusion culture due to
bulk agitation, the cells are expected to arrive and stay near the DEP electrodes only for a
short period of time. To emulate this, we applied a high frequency (10 MHz) electric field
(25 V pk to pk) intermittently. However, we were unable to see any adverse effect.
Hence, it was decided to expose the cells continuously and examine if any changes in
their growth characteristics and metabolism can be observed. In figure 3, the results from
three subcultures of HFN 7.1 cells are shown. Actually the last subculture was monitored
throughout the stationary phase. Comparison of viable cell density, viability, glucose
uptake, lactate production as well as monoclonal antibody production data between the
DEP-culture and the control suggests that there are no adverse effects. Similar results
were obtain for 15 aminoacids measured by HPLC analysis (data not shown). These
experiments are quite conservative as in an actual perfusion culture, the cumulative
exposure of the cells is expected to be much less. Similar results shown in Figure 4 are
obtained from another series of experiments using CHO cells adapted to grow in
suspension.

373

374

7. REFERENCES

1. Abidor, I.G. and Sowers, A.E. 1992. J. 61: 1557-1569.
2. Archer, G.P., Render, M.C., Betts, W.B. and Sancho, M. 1993. Microbios 76: 237-244.
3. Doblhoff-Dier, O., Gaida, T., Katinger, H., Burger, W., Gröschl, M. and Benes, E. 1994.

Biotechnol. Prog. 10: 428-432.
4. Docoslis, A., Kalogerakis, N., Behie, L.A., Kaler, K.V.I.S. 1997. Biotechnol. Bioeng. 54: 239-250.
5. Fuhr, G. Glasser, H., Müller, T. and Schnelle, T. 1994. Biochim. Biophys. Acta 1201: 353-360.
6. Gascoyne, P.R.C., Becker, F.F. and Wang, X.-B. 1995. Bioelectrochem. Bioenerget. 36: 115-125.
7. Gimsa, J., Marszalek, P., Loewe, U. and Tsong, T.Y. 1991. Biophys. J. 60: 749-760.
8. Huang, Y., Wang, X.B., Tame, J.A. and Pethig, R. 1993. J. Phys. D: Appl. Phys. 26: 1528-1535.
9. Kaler, K.V.I.S. and Jones, T.B. 1990. Biophys. J. 57: 173-182.
10. Kaler, K.V.I.S., Xie, J.P., Jones, T.B. and Paul, R. 1992. Biophys. J. 63, 58-69.

11. Krishna, G.G., Anwar, A.K.W., Mohan, D.R. and Ahmad, A. 1989. J. Biomed. Eng. 11: 375-380.
12. Markx, G.H., Talary, M.S. and Pethig, R. 1994. J. Biotechnol. 32: 29-37.
13. Neil, G.A. and Zimmermann, U. 1993. Methods in Enzymology 221: 339-361.
14. Neumann, E., Sowers, A.E. and Jordan, C.A. 1989. Electroporation and Electrofusion in Cell

Biology. Plenum Press, New York-London.
15. Pohl, H.A. 1977. In: Methods of cell separation (N. Catsimpoolas, ed.), Plenum Press, New York,

vol.1, 67-169.
16. Sukhorukov, V.L., Arnold, W.M. and Zimmermann, U. 1993. J. Membrane Biol. 132: 27-40
17. Trampler, F., Sonderhoff, S.A., Pui, P.W.S., Kilburn, D.G. and Piret, J.M. 1994. Bio/Technol.

12: 281-284.
18. Zimmermann, U. 1986. Rev. Physiol. Biochem. Pharmacol. 105:176-256.

Discussion 375
Al-Rubeai:
Kalogerakis: Have you done any work on the effect of cell density on cell
retention?
Aunins:
Kalogerakis: I only showed low cell densities (1 million cells/ml) as our
prototype broke - it is very sensitive which is why we are moving
Reuveny: away from it. There are problems, eg gas bubbles, so we incline
Kalogerakis: the device by 40°, and pearl chains form, but we can overcome this
problem with agitation.
Your perfusion rate is going to be a constrained variable, so I was
wondering if you have explored differences in field strength?
Also, what are the practical limits for nano fabrication for the area?

It is superficial velocity, and not perfusion rate, which limits the
operation. The open area was 1.5 cm by 2 cm for a 1.5 1 fermenter
working at 2 volumes per day. So it is not bulky and can be used
in a modular battery.

Do you look for the small ions in your medium?

If you mean current going through the medium, yes, this is a
serious problem. This is why we coat the electrodes with silicon
dioxide, which is an insulator. An electrical field does exist and
this poses problems in terms of manufacturing from the
microfabrication point of view.


Click to View FlipBook Version