Embed Size (px)
Citation preview
Serum-Free Large-Scale TransientTransfection of CHO Cells
Madiha Derouazi, Philippe Girard, Frederic Van Tilborgh, Keyvan Iglesias,Natalie Muller, Martin Bertschinger, Florian M. Wurm
Laboratory of Cellular Biotechnology, IGBB, Faculty of Life Science, SwissFederal Institute of Technology, 1015 Lausanne, Switzerland;fax: 41 21 693 61 40; e-mail: [email protected]
Received 3 December 2003; accepted 23 March 2004
Published online 26 July 2004 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/bit.20161
Abstract: To date, methods for large-scale transient geneexpression (TGE) in cultivated mammalian cells havefocused on two transfection vehicles: polyethylenimine(PEI) and calcium phosphate (CaPi). Both have been shownto result in high transfection efficiencies at scales beyond10 L. Unfortunately, both approaches yield higher levels ofrecombinant protein (r-protein) in the presence of serumthan in its absence. Since serum is a major cost factorand an obstacle to protein purification, our goal was to de-velop a large-scale TGE process for Chinese hamster ovary(CHO) cells in the absence of serum. CHO-DG44 cells werecultivated and transfected in a chemically defined mediumusing linear 25 kDa PEI as a transfection vehicle. Param-eters that were optimized included the DNA amount, theDNA-to-PEI ratio, the timing and solution conditions forcomplex formation, the transfection medium, and the celldensity at the time of transfection. The highest levels ofr-protein expression were observed when cultures at a den-sity of 2.0 � 106 cells/ml were transfected with 2.5 Ag/mlDNA in RPMI 1640 medium containing 25 mM HEPES atpH 7.1. The transfection complex was formed at a DNA:PEIratio of 1:2 (w/w) in 150mMNaClwith a 10-min incubationat room temperature prior to addition to the culture. Theprocedure was scaled up for a 20-L bioreactor, yieldingexpression levels of 10 mg/l for an intracellular protein and8 mg/l for a secreted antibody. B 2004 Wiley Periodicals, Inc.
Keywords: polyethylenimine; CHO; transient geneexpression
INTRODUCTION
Recombinant proteins (r-proteins) are gaining importance
in medicine as more of them obtain approval for use in the
prevention and treatment of disease. It is preferable to
express r-proteins in cultivated mammalian cells when
requiring correct protein folding and posttranslational
processing. The most rapid and least expensive method
for this is transient gene expression (TGE) (Wurm and
Bernard, 1999). With this approach it has been possible to
produce milligram to gram quantities of an r-protein within
days or weeks after the initiation of the project. TGE in-
volves the viral or nonviral delivery of a recombinant gene(s)
into cultured cells followed by recovery and purification of
the r-protein after a few days (Wurm and Bernard, 1999). For
nonviral gene delivery (transfection), the gene is first cloned
into an appropriate mammalian expression vector and the
cells are exposed to the prepared DNA-vehicle complexes.
To date, the successful transfection of mammalian cells
has been accomplished with DNA-containing complexes
formed with calcium phosphate, polycations (polyplexes),
and liposomes (lipoplexes) (Thomas and Klibanov, 2003).
Human embryonic kidney cells (HEK 293) are among
the most frequently used hosts for TGE. They are readily
transfected with a broad variety of methods yielding up to
20 mg/l of r-protein (Pham et al., 2003). For large-scale
applications, two different transfection methods have been
used: calcium phosphate-DNA coprecipitation (CaPi) (Gi-
rard et al., 2002; Meissner et al., 2001) and polyplex for-
mation with PEI (Durocher et al., 2002; Schlaeger and
Christensen, 1999). The transfection of HEK 293 cells with
PEI has been scaled up to 14 L (Pham et al., 2003) and that
with CaPi to beyond 100 L (Girard et al., 2002). Both
methods are effective in the presence or absence of serum,
but higher r-protein yields have been observed in the pres-
ence of serum. Since serum is a predominant cost factor for
large-scale TGE and often makes downstream processing
of the r-protein more difficult, one of our goals was to de-
velop a serum-free process. Second, as HEK 293 cells are
of human origin and only a small number of biopharma-
ceuticals produced in them have gained approval from
regulatory authorities, we wanted to develop a method for
TGE in CHO cells.
CHO cells are extensively used in the biotechnology
industry to produce stable cell lines for r-protein expres-
sion. However, the development of a stable cell line is a
costly and time-consuming process, and this investment
may be lost if the r-protein is not approved for clinical use.
For an integrated approach to the production of an r-protein
for clinical studies, it would be ideal to use transiently
expressed material for preclinical and early clinical trials
(Phase I), since this is a faster and less expensive approach
relative to production from a stable cell line. However, for
B 2004 Wiley Periodicals, Inc.
Correspondence to: F. M. Wurm
this strategy to be effective the transient and stable material
must originate from the same source.
Unfortunately, methods for nonviral gene delivery into
CHO cells are not as well developed as those for HEK
293 cells, and the large-scale transfection of CHO cells
under serum-free conditions has not been reported. Trans-
fection of adherent CHO cells using CaPi is serum- and cell
cycle-dependent and requires an osmotic shock to achieve a
high efficiency of gene delivery (Batard et al., 2001;
Grosjean et al., 2002). For suspension-adapted CHO cells,
preliminary experiments using the CaPi method have
revealed a low transfection efficiency that is enhanced
about 4-fold following an osmotic shock with glycerol
or dimethylsufoxide (unpubl. data). However, this proce-
dure is not practical for scale-up since it requires two
complete medium exchanges. Here we demonstrate the
efficient expression of r-proteins in suspension-adapted
CHO cells using PEI-mediated gene delivery in serum-free
conditions. The method proved to be scalable in a 20-L
stirred bioreactor.
MATERIALS AND METHODS
Cell Line
Suspension-adapted CHO DG44 cells, deficient in dihy-
drofolate reductase activity (Urlaub et al., 1983, 1986),
were cultured in serum-free ProCHO5 CDM medium (Bio-
Whittaker, Walkersville, MD) supplemented with 0.68 g/l
hypoxanthine and 0.194 g/l thymidine (Sigma Chemical,
St. Louis, MO). The cells were maintained in agitated cul-
tivation systems in a 5% CO2 atmosphere with a relative
humidity of 95%. The cells were passaged every 3–4 days
at a seeding density of 2.5 � 105 cells/ml.
Plasmids
The vector pEGFP-N1 expressing the enhanced green
fluorescent protein (GFP) was purchased from ClonTech
(Palo Alto, CA). As previously described, the human anti-
Rhesus D IgG light and heavy chain genes (Miescher et al.,
2000; Zahn-Zabal et al., 2001) were separately cloned into
pEAK8 (Edge Biosystems, Gaithersburg, MD) to produce
pLH1 and pLH2, respectively, or into pMYKEF-I (kindly
provided by Dr. Y.S. Kim, Korean Research Institute of
Bioscience and Biotechnology) (Kim et al., 2002) to produce
pKML and pKMH, respectively. Plasmid DNA was purified
on a NucleobondAX anion exchange column (Macherey-
Nagel, Duren, Germany) according to the manufacturer’s
protocol and stored at a concentration of 1 mg/ml in TE
buffer (10 mM Tris-HCl, 1 mM EDTA, pH 7.4).
Transfection Agents
Stock solutions of linear 25 kDa PEI (Polysciences,
Eppenheim, Germany) and JetPEI (PolyPlus Transfection,
Illkirch, France) and branched PEIs with molecular weights
of 0.423, 0.6–0.8, 1.2, 1.8–2, and 10–25 kDa (Sigma
Chemical) were prepared in water at a final concentration
of 1 mg/ml. The pH was adjusted to 7.0 with HCl. The
solutions were sterilized using a 0.22 Am filter and stored
frozen at –80jC. JetPEI was stored at 4jC.
Transfection in Agitated 12-Well Plates
One day prior to transfection, cells were seeded at 1 �106 cells/ml in ProCHO5 CDM medium. On the day of
transfection the cells were washed once with either a
DMEM/F12 based medium (Applichem, Darmstadt, Ger-
many) or RPMI 1640 medium (Applichem) supplemented
with hypoxanthine and thymidine as described above
and then suspended in the same medium at 1–2 �106 cells/ml as indicated in the text. Cells (1 ml) were
then dispensed into 12-well plates (TPP, Wohlen, Switzer-
land) for transfection.
Stock solutions of DNA and PEI were diluted separately
in sterile 150 mM NaCl or 278 mM glucose, as indicated.
The final volume of each solution was equivalent to 5% of
the volume of the culture to be transfected. The PEI
solution was then added to the diluted DNA and the
mixture was incubated at room temperature for 10 min
unless otherwise noted. Finally, 100 Al of the DNA-PEI
mix was added to each well and the plates were incubated
with agitation (200 rpm) for 5 h at 37jC in an atmosphere
with 5% CO2 and 95% humidity (Girard et al., 2001).
Unless otherwise stated, the DNA concentration in the
transfection medium before dilution was kept constant at
2.5 Ag/ml. The cells were then diluted with 1 ml of
ProCHO5 CDM medium and the incubation was continued
for 3 days.
Transfection in Bioreactors
CHO DG44 cells were seeded at a density of 2 � 106 cells/
ml in either a 3-L bioreactor (Applikon, Schiedam, The
Netherlands) or a 20-L bioreactor (BioEngineering, Wald,
Switzerland) in RPMI 1640 medium. Transfection fol-
lowed immediately after seeding with a DNA-PEI mix that
was prepared as described above. At 5 h posttransfection
the cultures were diluted with one volume of ProCHO5
CDM medium. Sampling was performed at the times in-
dicated. The cultures were agitated at 150 rpm during
transfection and at 200 rpm after dilution of the culture.
Dissolved oxygen was maintained at 20% by sparging air
into the culture and the pH was maintained at 7.1 using
NaOH and CO2. The culture was fed with glucose in order
to keep the level at 4 g/l, and the sodium bicarbonate con-
centration was maintained at 10 mM.
Protein Quantification
For cultures in 12-well plates, reporter protein expression
was quantified on day 3 posttransfection. For GFP analysis,
538 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 87, NO. 4, AUGUST 20, 2004
the cells in each well were lysed by addition of 1 ml of 1%
Triton X-100 (Sigma Chemical) in TE. After incubation at
room temperature for 1 h, the fluorescence was measured
using a Cytofluor Series 4000 plate-reader (PerSeptive Bio-
systems, Farmingham, MA). GFP was excited at 485 nm
with a bandwidth of 20 nm and the emission fluorescence
was measured at 530 nm with a bandwidth of 25 nm. The
background fluorescence from cultures transfected in the
absence of plasmid was subtracted from each value to
obtain relative fluorescence units (RFU). IgG concentration
in the culture medium after centrifugation was determined
by sandwich ELISA using a goat antihuman kappa light
chain antibody for capture and alkaline phosphatase-
conjugated goat antihuman IgG (BioSource, Lucerne,
Switzerland) for detection.
For the analysis of transfections in bioreactors, 2 ml
of culture was dispensed per well in a 12-well plate and
1 ml of 1% Triton X-100 in TE was added. After 1 h
of incubation at room temperature, the GFP was measured
as described above. The IgG titer was determined as des-
cribed above.
RESULTS
Selection of a Delivery Vehicle for the Serum-FreeTransfection of CHO Cells in Suspension
Several different commercially available branched and
linear PEIs ranging in molecular weight from 0.4–25 kDa
were tested for their suitability as DNA delivery vehicles
for the transfection of CHO DG44 cells adapted to serum-
free suspension growth (Table I). Cells were distributed
into 12-well plates at a density of 1 � 106 cells/ml in
DMEM/F12 medium and transfected with pEGFP-N1 using
different DNA:PEI (w:w) ratios ranging from 1:1 to 1:4.
Transfection with either of the two linear PEIs (JetPEI and
25 kDa) resulted in measurable levels of GFP expression
for DNA:PEI ratios of 1:2 and higher. The highest level of
GFP expression was observed in cultures transfected with
linear 25 kDa PEI (Table I). Importantly, the cells remained
in suspension following transfection with either of these
two PEIs (Table I). For most of the transfections with
branched PEIs, the cells rapidly became adherent despite
agitation and the absence of serum (Table I). Among the
branched PEIs, only the one with a molecular weight of
10–25 kDa promoted gene transfer (Table I). However, the
level of GFP expression was 3-fold lower than that
observed for the transfection with linear 25 kDa PEI and
only a small fraction of the transfected cells remained in
suspension. Based on these results, further optimization
studies were carried out with linear 25 kDa PEI.
In addition to the experiments described above, suspen-
sion-adapted CHO DG44 cells were also transfected with
different commercially available liposomes. Transfection
with Lipofectamine 2000 (Invitrogen, San Diego, CA) and
DOTAP (Roche Diagnostic, Rotkreuz, Switzerland)
yielded good GFP expression levels (Table II). However,
the cost of these reagents precluded their use in large-scale
transfections of suspension cultures. Finally, transfections
with polybrene and polylysine (Sigma Chemical) resulted
in very low GFP expression levels (Table II).
Small-Scale Optimization of Transfection With Linear25 kDa PEI
The transfections for the optimization studies described
below were performed with suspension-adapted CHO cells
in agitated 12-well plates. Several parameters that are
critical to the formation of DNA-PEI complexes, including
the amount of DNA, the DNA:PEI ratio, the solution
conditions, and the length of the incubation period were
evaluated. The culture medium for transfection and the cell
density at the time of transfection were also investigated.
DNA and PEI were initially mixed in either 150 mM
NaCl or 278 mM glucose, since both of these solutions had
previously been reported to support polyplex formation
(Kircheis et al., 2001). Various amounts of PEI were mixed
in one of these two solutions with a constant amount of
DNA yielding a final DNA concentration of 2.5 Ag/ml in
the culture medium. Cells were then transfected and the
level of GFP expression was determined 3 days later. For
Table I. Transfection of suspended CHO DG44 cells with different
polyethylenimines.
PEI [kDa] Chemical structure GFP expressiona Cell conditionb
0.423 branched � adherent
0.6– 0.8 branched � adherent
1.2 branched � adherent
1.8– 2 branched � adherent
10– 25 branched + adherent
JetPEI linear ++ suspension
25 linear +++ suspension
aCells at 1 � 106 cells/ml were transfected in DMEM/F12 with pEGFP-
N1. Level of GFP expression: high (+++), medium (++), low (+), and not
detected (�).bAfter transfection.
Table II. Transfection of suspended CHO DG44 cells with different
chemical agents.
Supplier Name Compound GFP Expressiona
Invitrogen Lipofectamine 2000 liposome +++
Roche DOTAP liposome ++
Roche DOSPER liposome +
GTS GenePorter liposome +
Promega TransFast liposome �Sigma Polybrene polymer +
Sigma Polylysine polymer +
aCells at 1 � 106 cells/ml were transfected in DMEM/F12 with pEGFP-
N1. Level of GFP expression: high (+++), medium (++), low (+), and not
detected (�).
DEROUAZI ET AL.: TRANSIENT TRANSFECTION OF CHO CELLS 539
complex formation in either NaCl or glucose, the DNA:PEI
ratio of 1:2 (w:w) resulted in the highest level of GFP
expression (Fig. 1). At this ratio, however, GFP expression
was 20% higher following complex formation in NaCl
as compared to glucose (Fig. 1). For transfections with
DNA:PEI ratios less than a 1:2 ratio, very little GFP
expression was observed, regardless of the solution used
for complex formation (Fig. 1). Next, the effect of pH
on complex formation was tested by varying the pH of
buffered 150 mM NaCl from 5 to 6.5. The transfections
were done with DNA:PEI ratios of 1:2 and 1:3 and a final
DNA concentration in the culture medium of 2.5 Ag/ml.
No significant differences were observed in GFP expres-
sion under the conditions tested (data not shown). Based on
these results, further optimization studies were performed
with DNA-PEI complexes formed in 150 mM NaCl at
pH 5.5.
The low levels of GFP expression observed at DNA:PEI
ratios of 2:1 and 1:1 may have resulted from insufficient
incorporation of all the DNA into polyplexes. To inves-
tigate this possibility, complex formation was evaluated by
determining the quantity of PEI needed to condense 1 Ag of
plasmid. Various quantities of linear 25 kDa PEI were
added to plasmid DNA. After a 10-min incubation at room
temperature, the solutions were centrifuged for 2 min at
16,000g and the DNA concentration of each supernatant
was measured spectrophotometrically. Incorporation of
100% of the DNA into complexes was observed following
addition of 0.3 Ag or more of PEI to 1 Ag DNA (Fig. 2).
Independent of the plasmid DNA, the transition from 100%
free DNA to 100% complexed DNA always showed the
same pattern (data not shown). A slight variation in the
amount of PEI required for complete DNA condensation
was observed depending on the DNA quality (data not
shown). This result demonstrated that at the DNA:PEI
ratios used in the experiment shown in Figure 1, all of the
DNA was probably incorporated into polyplexes. There-
fore, other factors must have been responsible for the low
levels of GFP expression following transfections with
DNA:PEI ratios of 2:1 and 1:1.
The transfections described above were performed in a
modified DMEM/F12 medium that we routinely use for the
transfection of HEK 293 cells with CaPi (Girard et al.,
2001). Transfections were attempted in other media
including RPMI 1640 since it has been shown to support
the serum-free transfection of mammalian cells when using
PEI as a delivery vehicle (Schlaeger and Christensen, 1999).
Using two different DNA:PEI ratios (1:2 and 1:3), a higher
level of GFP expression was observed following trans-
fections in RPMI buffered with 25 mM HEPES (pH 7.1) as
compared to cells transfected in DMEM/F12-based media
(data now shown). For this reason, further optimization
experiments were performed in RPMI 1640.
In the experiments described above, the PEI and DNA
were mixed and then incubated for 10 min at room tem-
perature prior to addition to the cells, as previously reported
Figure 1. Influence of the composition of the solution for DNA-PEI
complex formation at various DNA:PEI ratios on transfection efficiency.
CHO cells at density of 5 � 105 cells/ml in DMEM/F12 medium were
transfected with pEGFP-N1 using different DNA:PEI ratios (w:w) as
indicated. The DNA-PEI complexes were formed in either 278 mM
glucose or 150 mM NaCl. The final DNA concentration in the culture
medium was 2.5 Ag/ml. GFP expression was measured on day 3 post-
transfection. Each bar represents the average of three transfections.
Relative fluorescence units (RFU).
Figure 2. Condensation of DNA by PEI. Varying amounts of linear
25 kDa PEI were added to 1 Ag plasmid DNA in 150 mM NaCl and
incubated for 10 min at room temperature. After centrifugation the
concentration of DNA remaining in the supernatant was measured
spectrophotometrically at a wavelength of 260 nm. Each point represents
the average of three independent samples.
540 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 87, NO. 4, AUGUST 20, 2004
(Boussif et al., 1995). To determine if this is the optimal
time for formation of the DNA-PEI complex, the time of
incubation after mixing was varied from 5–20 min. At
a DNA:PEI ratio of 1:2, the 5- and 10-min incubations
yielded GFP levels that were 15% higher than those ob-
served following a 15- or 20-min incubation (Fig. 3). Less
variability in GFP levels was observed for the different
incubation times when a DNA:PEI ratio of 1:3 was used
(Fig. 3). Following this experiment, the DNA-PEI mixture
was allowed to incubate 10 min prior to transfection.
Finally, the cell density at the time of transfection
was investigated. Cultures at a density of either 1 or 2 �106 cells/ml were transfected with pEGFP-N1 using various
DNA:PEI ratios. The final DNA concentrations in the
medium ranged from 1.5–5.0 Ag/ml. For most conditions
tested, the GFP expression level was higher when cells
were transfected at the higher cell density (Fig. 4). For
transfections at a density of 2 � 106 cells/ml, the highest
levels of GFP expression were observed at DNA concen-
trations of 2.0–2.5 Ag/ml and at a DNA:PEI ratio of 1:2
(Fig. 4). These conditions also yielded the highest GFP ex-
pression levels when the cell density was 1 � 106 cells/ml
(Fig. 4).
Overall, the optimal conditions for the PEI-mediated
transfection of suspension-adapted CHO cells were found
to be a DNA:PEI ratio of 1:2 with a final DNA con-
centration in the culture medium of 2.5 Ag/ml. The DNA
and PEI were mixed in the presence of 150 mM NaCl and
allowed to incubate for 10 min prior to addition to the
culture. The CHO cells were grown in ProCHO5 CDM
medium and then transferred at a cell density of 2 �106 cells/ml to RPMI 1640 containing 25 mM HEPES at
pH 7.1. At 5 h posttransfection, the cells were diluted with
one volume of ProCHO5 CDM medium. These conditions
were the basis for the large-scale transfections in bio-
reactors described below.
Serum-Free Transfection of CHO Cells in Bioreactors
Initially, a 500-ml culture of suspension-adapted CHO cells
at a density of 2 � 106 cells/ml was transfected with
pEGFP-N1 in a 3-L bioreactor. The DNA-PEI complex was
formed using the optimized conditions described above.
GFP expression was detectable as early as 4 h posttrans-
fection (data now shown), and by 7 h posttransfection many
fluorescent cells were observed (Fig. 5). GFP fluorescence
increased with time, with the highest level observed at
5 days posttransfection (Fig. 5). Subsequently, the GFP
level declined. The maximum level of fluorescence was
estimated to correspond to 10 mg/l of intracellular GFP
(Girard, 2001). The transfected culture continued to grow as
single cells until 4 days posttransfection, when aggregates
began to appear (Fig. 5). An arrest of cell growth and a
decrease in cell viability were observed 5 days posttrans-
fection (data not shown).
For the second set of experiments in bioreactors, the cells
were transfected with pLH1 and pLH2 encoding the light
and heavy chain IgG genes, respectively. Two 500-ml
cultures in 3-L bioreactors were transfected in parallel at
Figure 4. Influence of DNA concentration, DNA:PEI ratio, and cell
density on transfection efficiency. CHO cells at a density of either 1 � 106
or 2 � 106 cells/ml in RPMI 1640 medium were transfected with varying
amounts of pEGFP-N1 and varying DNA:PEI ratios. GFP expression was
measured on day 3 posttransfection. Each bar represents the average of
three transfections.
Figure 3. Effect of DNA-PEI complex incubation time on transfection
efficiency. CHO cells at a density of 2 � 106 cells/ml in RPMI 1640
medium were transfected with pEGFP-N1 at different DNA:PEI ratios in
150 mM NaCl with a final DNA concentration in the medium of 2.5 Ag/ml.
The DNA-PEI solution was allowed to incubate at room temperature for
various times as indicated. GFP expression was measured on day 3
posttransfection. Each bar represents the average of three transfections.
DEROUAZI ET AL.: TRANSIENT TRANSFECTION OF CHO CELLS 541
DNA:PEI ratios of 1:2 and 1:3. To have a visual and rapid
control of the transfection, pEGFP-N1 DNA (2% of the
total DNA) was added to the transfection mix. The culture
transfected at a DNA:PEI ratio of 1:2 reached a maximum
cell density of 4.5 � 106 cells/ml, which was 1.5 times
higher than the density observed in the parallel culture
transfected at a DNA:PEI ratio of 1:3 (data not shown).
This difference correlated well with a difference observed
between the two cultures in IgG titer. A 1:2 DNA:PEI ratio
resulted in a 30% higher titer than the culture transfected
with a DNA:PEI ratio of 1:3 (data not shown).
The transfection method was scalable to a 20-L
bioreactor, as shown in Figure 6. A culture with a volume
of 6 L was transfected with pLH1 and pLH2 using the
optimized conditions described above. At 5 h posttransfec-
tion the cultures were diluted with one volume of ProCHO5
CDM medium to give final volumes of 1 and 13 L. The IgG
titers for the two cultures (in 3-L and 20-L bioreactors)
were approximately the same at 6 days posttransfection
(Fig. 6). At this point the run in the 20-L bioreactor was
discontinued. The titer in the 3-L bioreactor reached about
8 mg/l at 10 days posttransfection (Fig. 6).
Figure 6. Transfection of CHO cells in 3-L and 20-L bioreactors. CHO cells at a density of 2 � 106 cells/ml in RPMI 1640 medium were transfected at a
DNA:PEI ratio of 1:2 using LH1 and pLH2 in a 3:7 ratio (98% of the total DNA). The remaining 2% of the DNA was pEGFP-N1. The transfection volumes
were 500 ml for the 3-L bioreactor and 6 l for the 20-L bioreactor. Five hours posttransfection the cultures were diluted with one volume of ProCHO CDM
medium. A, lgG concentration was determined by ELISA as described in Materials and Methods. B, Cell density and viability were determined by the
Trypan blue exclusion method.
Figure 5. Transfection of CHO cells in a 3-L bioreactor. CHO cells at a density of 2 � 106 cells/ml in 500 ml RPMI 1640 were transfected with pEGFP-
N1 at a DNA:PEI ratio of 1:2 with a final DNA concentration of 2.5 Ag/ml in the culture medium. At 5 h posttransfection the culture was diluted with one
volume of ProCHO5 CDM medium. GFP expression was measured as described in Materials and Methods at the times indicated. Relative fluorescence
units (RFU).
542 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 87, NO. 4, AUGUST 20, 2004
A summary of the results from several transfections in
bioreactors ranging in scale from 1–13 liters is shown in
Table III, where the IgG titers at 3 days posttransfection
are compared to results from optimized transfections at the
1-ml scale in 12-well plates. Some of the transfections
noted in Table III were performed with pKML and pKMH
encoding the light and heavy IgG genes, respectively, under
the control of the murine cytomegalovirus (CMV) imme-
diate early promoter (Kim et al., 2002). In contrast, the
light and heavy chain IgG genes are controlled by the
elongation factor 1 alpha (EF-1a) promoter in pLH1 and
pLH2. Overall, the IgG titers for the transfections in bio-
reactors were as high as or higher than those observed in
the small-scale transfections (Table III). These results
demonstrate the scalability of the optimized transfection
protocol described here.
DISCUSSION
Large-scale TGE is currently under development in many
laboratories as a process to rapidly produce milligram to
gram quantities of r-proteins in mammalian cells (Duro-
cher et al., 2002; Girard et al., 2002; Meissner et al., 2001;
Pham et al., 2003; Schlaeger and Christensen, 1999; Schlae-
ger et al., 2003). Here we report the optimization of a low-
cost transient transfection method for suspension-adapted
CHO cells under serum-free conditions using the cationic
polymer PEI. The process achieved yields up to 8 mg/l of
antibody in a 20-L bioreactor. The transfection of sus-
pension cultures of CHO cells in serum-free conditions has
also been achieved using a peptide-based transfection agent
(Ro 1539). Under theses conditions, a yield of about 7 mg/l
of secreted alkaline phosphatase at the 15-ml scale was re-
ported (Schlaeger et al., 2003).
Of the different PEIs tested, the linear ones were better
than the branched ones for promoting gene transfer under
serum-free conditions in CHO cells. With the branched
PEIs, reporter protein expression was only observed with
the one having the highest molecular weight (10–25 kDa),
as previously reported (Godbey et al., 1999a). Furthermore,
the use of branched PEIs induced cell adherence in agitated
12-well plates in the absence of serum. However, the
reason for this effect on suspension cells is not known. The
linear 25 kDa PEI was chosen for optimization studies
since it yielded the highest levels of r-protein expression
while preventing aggregation and adherence of the cells in
suspension. The most important parameters for optimum
r-protein expression following PEI-mediated transfection
were the DNA amount, the DNA:PEI ratio, and the cell den-
sity at the time of transfection. Parameters that were less
critical included the incubation time of the PEI with the DNA
prior to transfection and the pH of the NaCl solution used for
PEI and DNA dilution.
The three critical parameters identified in this study may
be interrelated and may depend on the physical and
chemical properties of PEI and DNA-PEI complexes. We
observed complete DNA condensation at a DNA:PEI ratio
(1:0.3) that corresponds to an N/P ratio (PEI nitrogen to
DNA phosphate) of 2. Nearly complete DNA condensation
has also been found with a branched 25 kDa PEI at an N/P
ratio of 3 (Kunath et al., 2003). In our hands, however,
successful gene transfer was only observed with N/P ratios
of 6 or more, and the optimal N/P ratio for the transfection
of CHO cells was found to be 13. Likewise, an optimal N/P
ratio from 9 to 13 has previously been shown for the trans-
fection of various cell lines with PEI (Boussif et al.,
1995). The excess amount of PEI needed to promote gene
transfer relative to the amount needed to condense a
quantity of DNA may indicate the importance of positive
surface charge effects on the behavior of the transfection-
active particle. For example, the overall charge of the com-
plex may be important for release from the endosome
following its uptake by the cell. PEI is thought to inhibit the
acidification of endosomes by binding protons, leading to
an influx of chloride ions into the endosome. This results in
osmotic swelling and eventual rupture of the organelle
(‘‘proton sponge effect’’) and release of the complex into
the cytoplasm (Boussif et al., 1995; Kichler et al., 2001).
Alternatively, the surface charge of the DNA-PEI complex
may play a role in its nuclear transport, since it has been
shown that the nucleus is accessible to DNA–PEI com-
plexes without the breakdown of the nuclear membrane
during mitosis (Brunner et al., 2002; Godbey et al., 1999a;
Pollard et al., 1998). In contrast, CaPi- and liposome-
mediated transfections are known to be cell cycle-dependent
(Brunner et al., 2000; Grosjean et al., 2002). Third, the ex-
cess of PEI may be important in determining the size and/
or shape of the transfection-active particle. Particle size
Table III. PEI transfection at different scale.
Culture systema Plasmids IgG titer [mg/l] Culture volume Sampling time
12-well plates pKML/pKMH 2 2 ml day 3
12-well plates pLH1/pLH2 4 2 ml day 3
Spinner pLH1/pLH2 5.4 100 ml day 3
BR 3 L pLH1/pLH2 6.1 1.2 L day 3
BR 5 L pKML/pKMH 2.3 4.5 L day 3
BR 20 L pLH1/pLH2 5 13 L day 3
BR 20 L pKML/pKMH 2 13 L day 2
aBioreactor (BR). The IgG titer in the 5 L bioreactor may have been reduced due to inefficient
aeration and stripping.
DEROUAZI ET AL.: TRANSIENT TRANSFECTION OF CHO CELLS 543
has been shown to be influenced by the DNA:PEI ratio,
with the size decreasing as the ratio of PEI to DNA is
increased (Dunlap et al., 1997). In turn, smaller particles
have been shown to be less efficient for transfection
(Kircheis et al., 2001). Finally, PEI has been shown to be
cytotoxic to endothelial cells and L929 fibroblasts (God-
bey et al., 2001; Kunath et al., 2003). This may be due to
damage to the plasma membrane following exposure to
PEI (Choksakulnimitr et al., 1995). In support of these
findings, we consistently observed higher r-protein yields
following transfections with a DNA:PEI ratio of 1:2 than
with a ratio of 1:3 or higher (cf. Figs. 1, 3, and 4). These
results may be explained by an increase in cell death
following transfection with the higher amounts of PEI.
Each of these properties of PEI and DNA-PEI complexes
may therefore pose constraints on the critical parameters
for PEI-mediated transfection of cells in serum-free
medium.
The amount of plasmid DNA taken up by CHO cells
following transfection with PEI was not determined. It has
been shown, however, that after CaPi transfection of CHO
cells about 5% of the plasmid DNA is taken up by about
half the cells in the population. This corresponds to an
uptake of about 10,000–50,000 copies of plasmid per cell
(Batard et al., 2001). Although an osmotic shock following
transfection substantially increases r-protein expression,
this step does not result in higher plasmid uptake by the
cells (Batard et al., 2001). Following CaPi-mediated trans-
fection with the GFP gene, GFP-positive cells were de-
tected as early as 6–8 h posttransfection (Grosjean, 2003).
Interestingly, most, if not all, of these cells underwent cell
division prior to expressing GFP (Grosjean, 2003). For
CHO cells transfected with pEGFP-N1 in the presence of
PEI, GFP-positive cells were observed at 4 h posttransfec-
tion. Considering that the GFP needs a maturation time of
60 min before becoming fluorescent (Tsien, 1998), delivery
of the plasmid DNA into the nucleus must have occurred
less than 3 h after transfection. However, we do not know if
GFP-positive cells underwent mitosis between transfection
and the onset of GFP expression. Similarly, Godbey et al.
(1999b) found DNA-PEI complexes in cells at 2–3 h post-
transfection and in nuclei at 3.5–4.5 h posttransfection.
These results suggest that the mechanisms of gene transfer
by CaPi and PEI are distinct. Additional experiments are
currently under way in our laboratory to measure the rate
and level of plasmid uptake resulting from PEI-mediated
transfection of CHO cells.
In conclusion, we optimized TGE in suspension cultures
of CHO cells using linear 25 kDa PEI in serum-free
medium. The optimal transfection conditions were found to
be a cell density of 2 � 106 cells/ml in RPMI medium with
a DNA:PEI ratio of 1:2 and a final DNA concentration in
the culture medium of 2.5 Ag/ml. The cells were diluted 5 h
posttransfection with one volume of seed train medium.
The method was found to be scalable to a 20-L bioreactor.
Transfection using liner 25 kDa PEI is cost-efficient and
suitable for serum-free operations. It offers high trans-
fection efficiency, simple operation with only one medium
exchange procedure, and rapid DNA transfer into the
nucleus without the need for an osmotic shock. These
features make PEI a more suitable transfection vehicle than
liposomes or CaPi for large-scale TGE with CHO cells.
We thank Dr. David Hacker for critically reading the manuscript and
Dr. Yeon-Soo Kim for providing the pMYK/EF-I plasmid.
References
Batard P, Jordan M, Wurm F. 2001. Transfer of high copy number plasmid
into mammalian cells by calcium phosphate transfection. Gene 270:
61–68.
Boussif O, Lezoualc’h F, Zanta MA, Mergny MD, Scherman D, Demeneix
B, Behr JP. 1995. A versatile vector for gene and oligonucleotide
transfer into cells in culture and in vivo: polyethylenimine. Proc Natl
Acad Sci USA 92:7297– 7301.
Brunner S, Sauer T, Carotta S, Cotten M, Saltik M, Wagner E. 2000. Cell
cycle dependence of gene transfer by liploplex, polyplex and recom-
binant adenovirus. Gene Ther 7:401– 407.
Brunner S, Furtbauer E, Sauer T, Kursa M, Wagner E. 2002. Overcoming
the nuclear barrier: cell cycle independent nonviral gene transfer
with linear polyethylenimine or electroporation. Mol Ther 5:80 –86.
Choksakulnimitr S, Masuda S, Tokuda H, Takakura Y, Hashida M. 1995.
In-vitro cytotoxicity of macromolecules in different cell-culture sys-
tems. J Control Release 34:233– 241.
Dunlap DD, Maggi A, Soria MR, Monaco L. 1997. Nanoscopic structure of
DNA condensed for gene delivery. Nucleic Acids Res 25:3095– 3101.
Durocher Y, Perret S, Kamen A. 2002. High-level and high-throughput
recombinant protein production by transient transfection of suspen-
sion-growing human 293-EBNA1 cells. Nucleic Acids Res 30:E9–9.
Girard P. 2001. Rapid protein expression in mammalian cells: large-scale
transient transfection. Ph.D. Thesis, EPFL, Lausanne, Switzerland.
Girard P, Jordan M, Tsao M, Wurm FM. 2001. Small scale bioreactor
system for process development and optimization. Biochem Eng J
7:117– 119.
Girard P, Derouazi M, Baumgartner G, Bourgeois M, Jordan M, Jacko B,
Wurm F. 2002. 100-Liter transient transfection. Cytotechnology 38:
15–21.
Godbey WT, Wu KK, Mikos AG. 1999a. Size matters: molecular weight
affects the efficiency of poly(ethylenimine) as a gene delivery vehicle.
J Biomed Mater Res 45:268– 275.
Godbey WT, Wu KK, Mikos AG. 1999b. Tracking the intracellular path of
poly(ethylenimine)/DNA complexes for gene delivery. Proc Natl
Acad Sci USA 96:5177– 5181.
Godbey WT, Wu KK, Mikos AG. 2001. Poly(ethylenimine)-mediated
gene delivery affects endothelial cell function and viability. Bio-
materials 22:471– 480.
Grosjean F. 2003. Correlation between cell cycle and calcium phos-
phate transient transfection of CHO cells. PhD Thesis, EPFL, Lau-
sanne, Switzerland.
Grosjean F, Batard P, Jordan M, Wurm FM. 2002. Correlation between
CHO cell cycle and transfection efficiency using Ca/PO4. Cytotech-
nology 38:57– 62.
Kichler A, Leborgne C, Coeytaux E, Danos O. 2001. Polyethylenimine-
mediated gene delivery: a mechanistic study. J Gene Med 3:135– 144.
Kim SY, Lee JH, Shin HS, Kang HJ, Kim YS. 2002. The human
elongation factor 1 alpha (EF-1 alpha) first intron highly enhances
expression of foreign genes from the murine cytomegalovirus pro-
moter. J Biotechnol 93:183– 187.
Kircheis R, Wightman L, Wagner E. 2001. Design and gene delivery
activity of modified polyethylenimines. Adv Drug Deliv Rev 53:
341–358.
544 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 87, NO. 4, AUGUST 20, 2004
Kunath K, von Harpe A, Fischer D, Petersen H, Bickel U, Voigt K, Kissel
T. 2003. Low-molecular-weight polyethylenimine as a non-viral vec-
tor for DNA delivery: comparison of physicochemical properties,
transfection efficiency and in vivo distribution with high-molecular-
weight polyethylenimine. J Control Release 89:113– 125.
Meissner P, Pick H, Kulangara A, Chatellard P, Friedrich K, Wurm FM.
2001. Transient gene expression: recombinant protein production with
suspension-adapted HEK293-EBNA cells. Biotechnol Bioeng 75:
197– 203.
Miescher S, Zahn-Zabal M, De Jesus M, Moudry R, Fisch I, Vogel M,
Kobr M, Imboden MA, Kragten E, Bichler J, Mermod N, Stadler BM,
Amstutz H, Wurm F. 2000. CHO expression of a novel human re-
combinant IgG1 anti-RhD antibody isolated by phage display. Br J
Haematol 111:157 – 166.
Pham PL, Perret S, Doan HC, Cass B, St-Laurent G, Kamen A, Durocher
Y. 2003. Large-scale transient transfection of serum-free suspension-
growing HEK293 EBNA1 cells: peptone additives improve cell growth
and transfection efficiency. Biotechnol Bioeng 84:332– 342.
Pollard H, Remy JS, Loussouarn G, Demolombe S, Behr JP, Escande D.
1998. Polyethylenimine but not cationic lipids promotes trans-
gene delivery to the nucleus in mammalian cells. J Biol Chem 273:
7507– 7511.
Schlaeger EJ, Christensen K. 1999. Transient gene expression in mam-
malian cells grown in serum-free suspension culture. Cytotechnology
30:71– 83.
Schlaeger EJ, Kitas EA, Dorn A. 2003. SEAP expression in transiently
transfected mammalian cells grown in serum-free suspension culture.
Cytotechnology 42:47– 55.
Thomas M, Klibanov AM. 2003. Non-viral gene therapy: polycation-
mediated DNA delivery. Appl Microbiol Biotechnol 62:27– 34.
Tsien RY. 1998. The green fluorescent protein. Annu Rev Biochem 67:
509–544.
Urlaub G, Kas E, Carothers AM, Chasin LA. 1983. Deletion of the diploid
dihydrofolate reductase locus from cultured mammalian cells. Cell
33:405– 412.
Urlaub G, Mitchell PJ, Kas E, Chasin LA, Funanage VL, Myoda TT,
Hamlin J. 1986. Effect of gamma rays at the dihydrofolate reduc-
tase locus: deletions and inversions. Somat Cell Mol Genet 12:
555–566.
Wurm F, Bernard A. 1999. Large-scale transient expression in mamma-
lian cells for recombinant protein production. Curr Opin Biotech
10:156– 159.
Zahn-Zabal M, Kobr M, Girod PA, Imhof M, Chatellard P, de Jesus M,
Wurm F, Mermod N. 2001. Development of stable cell lines for
production or regulated expression using matrix attachment regions.
J Biotechnol 87:29–42.
DEROUAZI ET AL.: TRANSIENT TRANSFECTION OF CHO CELLS 545
