Methods 55 (2011) 44–51

Contents lists available at ScienceDirect

Methods
journal homepage: www.elsevier.com/locate/ymeth

A simplified polyethylenimine-mediated transfection process for large-scale

and high-throughput applications
Céline Raymond a,c, Roseanne Tom b, Sylvie Perret b, Pascal Moussouami a, Denis L’Abbé b,
Gilles St-Laurent b, Yves Durocher a,b,c,⇑
a
Département de Biochimie, Faculté de Médecine, Université de Montréal, QC, Canada
b
Biotechnology Research Institute, National Research Council of Canada, 6100 Royalmount Avenue, Montreal, QC, Canada H4P 2R2
c
NSERC Strategic Network for the Large Scale Production of Single-Type Glycoform Monoclonal Antibodies (MabNet), Canada

a r t i c l e i n f o a b s t r a c t

Article history: Transient gene expression in mammalian cells is a valuable alternative to stable cell lines for the rapid
Available online 24 April 2011 production of large amounts of recombinant proteins. While the establishment of stable cell lines takes
2–6 months, milligram amounts of protein can be obtained within a week following transfection. The
Keywords: polycation polyethylenimine (PEI) is one of the most utilized reagents for small- to large-scale transfec-
Transfection tions as it is simple to use and, when combined with optimized expression vectors and cell lines, provides
Suspension culture high transfection efficiency and titers. As with most transfection reagents, PEI-mediated transfection
Polyethylenimine
involves the formation of nanoparticles (polyplexes) which are obtained by its mixing with plasmid
Serum-free
HEK293
DNA. A short incubation period that allows polyplexes to reach their optimal size is performed prior to
EBNA1 their addition to the culture. As the quality of polyplexes directly impacts transfection efficiency and pro-
ductivity, their formation complicates scalability and automation of the process, especially when per-
formed in large-scale bioreactors or small-scale high-throughput formats. To avoid variations in
transfection efficiency and productivity that arise from polyplexes formation step, we have optimized
the conditions for their creation directly in the culture by the consecutive addition of DNA and PEI. This
simplified approach is directly transferable from suspension cultures grown in 6-well plates to shaker
flasks and 5-L WAVE bioreactors. As it minimizes the number of steps and does not require an incubation
period for polyplex formation, it is also suitable for automation using static cultures in 96-well plates.
This ‘‘direct’’ transfection method thus provides a robust platform for both high-throughput expression
and large-scale production of recombinant proteins.
Crown Copyright Ó 2011 Published by Elsevier Inc. All rights reserved.

1. Introduction lishment of stable cell lines as it shortens the time required to ob-
tain the target protein in sufficient amount from several months to
The rising of protein–protein interaction studies as well as the a few weeks only [7–12]. The capacity to quickly express recombi-
constantly increasing demand of proteins in drug discovery pro- nant proteins in mammalian cells by transfection is thus of great
grams have brought the need for high-throughput protein expres- benefit not only for large-scale applications but also for high-
sion and automation to the foreground [1–5]. The need for throughput expression for functional screening and crystallization
recombinant proteins with post-translational modifications has purposes [3,11–15].
fostered the development of mammalian cell-based expression The human embryonic kidney 293 cell line (HEK293) and its
systems rather than prokaryotes and lower eukaryotes [6]. Tran- derivatives, such as the HEK293-EBNA1 that stably expresses the
sient gene expression in mammalian cells is widely used for the Epstein-Barr virus Nuclear Antigen 1 (293E) or the simian virus
fast production of recombinant proteins, mainly for pre-clinical 40 Large-T antigen (293T), are the most commonly used cell lines
in vitro and in vivo applications. Over the last 10 years, major for transient gene expression [10–12]. Indeed, HEK293 cells allow
improvements of the technique leading to better yields have made proper folding and relevant post-translational modifications of the
transient gene expression a competitive alternative to the estab- proteins. They are also easy to grow in serum-free suspension cul-
ture and highly transfectable with most gene transfer agents [11].
Polyethylenimine (PEI) is a widely used transfection agent for
⇑ Corresponding author at: Biotechnology Research Institute, National Research
Council of Canada, 6100 Royalmount Avenue, Montreal, QC, Canada H4P 2R2. Fax:
small to large-scale protein expression as it is highly efficient
+1 514 496 6785. and cost-effective [10–12]. PEI is a cationic polymer which binds
E-mail address: yves.durocher@cnrc-nrc.gc.ca (Y. Durocher). to and condensates DNA into small cationic nanoparticles, thus

1046-2023/$ - see front matter Crown Copyright Ó 2011 Published by Elsevier Inc. All rights reserved.
doi:10.1016/j.ymeth.2011.04.002
 C. Raymond et al. / Methods 55 (2011) 44–51 45

facilitating DNA cell uptake, most likely via their interaction with meyer flasks shaken at 120 rpm in a humidified incubator at 37 °C
cell-surface heparan sulfate proteoglycans [16,17]. Linear and with 5% CO2 (standard conditions).
branched 25 kDa PEIs are the most widely used polymers, the lin-
ear PEI being more efficient and probably less toxic for the trans- 2.2. Plasmids
fection of HEK293 cells grown in suspension cultures [7,18,19].
The current transfection procedure consists in the formation of Secreted human placental alkaline phosphatase (SEAP) and
PEI–DNA complexes (polyplexes) prior to their addition to the cell green fluorescent protein (GFP) were used as reporter genes in this
culture (indirect method). Typically, PEI and DNA are mixed to- study. The pTT vectors encoding SEAP or GFP were essentially as
gether and then incubated at room temperature for 10–15 min to previously described [7]. The chimeric monoclonal antibody
allow complexation before their addition to the cells. This incuba- against the colorectal carcinoma TAG72 antigen consists of the var-
tion time allows polyplexes to grow up to the most suitable size for iable heavy and light-chain domains of the murine monoclonal
a maximum transfection efficiency, which significantly decreases IgG1 antibody B72.3 coupled to human gamma 4 and kappa con-
when incubation time exceeds 30 min [20]. The preparation pro- stant regions [25,26]. The heavy and light chains (HC and LC) were
cedure (order of addition of DNA and PEI, DNA concentration, addi- each cloned between the EcoRI and BamHI sites of the pTT5 vector
tion and mixing speed) also influences the size and polydispersity [27,28]. Plasmids were amplified in Escherichia coli (DH5a) grown
of the polyplexes [21–23]. However, polyplexes formation be- overnight in CircleGrow medium (MP Biomedicals, Solon, OH) sup-
comes difficult to control when working with large-scale produc- plemented with 50 lg/mL ampicillin and purified using MAXIprep
tions, as issues such as mixing parameters and time required to or QIAprep spin Miniprep columns (Qiagen, Mississauga, ON).
transfer the polyplexes into the bioreactor become significant. This Endotoxins in selected plasmid preps were determined by the
DNA–PEI mixing step and the following incubation time are also LAL method using an EndosafeÒ-PTS™ spectrophotometer and
complicating automation when using high-throughput robotic FDA-licensed cartridges (Charles River Laboratories, Wilmington,
systems. MA).
An alternative technique to the indirect method is the consecu-
tive addition of DNA then PEI directly to the cell culture, allowing
2.3. Cell counts
the formation of polyplexes in situ (direct method) [19]. The pres-
ent study aims to evaluate the direct transfection method as an
Cell density and viability were determined using either a hema-
alternative technique for high-throughput, medium- and large-
cytometer (Hausser Scientific, Horshaw, PA) after erythrosine B
scale protein production purposes. It has been optimized with
staining, or the Cedex Innovatis’ automated cell counter Cedex
293-6E cells (a HEK293 cell line expressing a truncated version
Analyzer (Roche, Laval, Qc) based on the trypan blue exclusion
of EBNA1) cultured in 96-well plates, 6-well and 6-deep well agi-
method.
tated plates, 125-mL shaker flasks and 5L WAVE bioreactor, trans-
fected with plasmids encoding green fluorescent protein (GFP),
human secreted alkaline phosphatase (SEAP) or a chimeric 2.4. Transfection
IgG1/4 monoclonal antibody using linear or branched 25 kDa PEI.
The direct transfection process is performed in serum-free medium Linear 25 kDa polyethylenimine (LPEI) was obtained from Poly-
with cells grown in suspension (except for the 96-well plate for- sciences (Warrington, PA), and branched 25 kDa PEI (BPEI) from
mat) and does not require a medium exchange at any time. We Aldrich (Milwaukee, WI). Stock solutions (1 mg/mL) were pre-
show here that our optimized direct method provides yields com- pared in ultrapure water, sterilized by filtration (0.2 lm), aliquoted
parable to the indirect method, greatly simplifies process scale-up and stored at 20 °C. Cells were diluted 2 days before transfection
and is amenable to automation, thus providing a robust and simple in fresh medium at 0.40–0.45  106 cells/mL. No centrifugation step
platform for protein expression at any scales. was necessary to eliminate spent medium as the latter has no
impact on transfection efficiency or productivity (data not shown).
Cells were transfected with viability greater than 98% at densities
2. Materials and methods between 1.5 and 2.0  106 cells/mL, except for the 96-well plates
for which cell density was reduced to 0.2  106 cells/mL the day
2.1. Cell lines of transfection, as cells were grown in static instead of suspension
culture. DNA and PEI were separately diluted in complete serum-
The human embryonic kidney 293 cell line stably expressing a free F17 medium in sterile tubes. Transfection reagents volume
truncated Epstein-Barr virus Nuclear Antigen-1 (293-6E) was was generally 10% of the final culture volume. The final DNA con-
grown in suspension in a serum-free F17 medium (Invitrogen, centration was 1 lg per mL of culture, unless indicated otherwise.
Carlsbad, CA) supplemented with 0.1% Pluronic-F68, 25 lg/mL Direct method: the DNA:PEI ratio was 1:3 (w:w). Plasmid DNA
geneticin G418 and 4 mM glutamine [24]. Cultures were main- was directly added to the cells and the suspension was allowed
tained between 0.1 and 2.0  106 cells/mL in 125 or 250 mL Erlen- to incubate under agitation for 5 min at 37 °C before the addition

Table 1
Direct transfection parameters from 96-well plates to 5L WAVE batches.

96-well plate 6-well plate 6 deep-well ThinCert™ plate 125 mL shake flask 10 L WAVE Cellbag™
Working volume 200 lL 2 mL 6 mL 25 mL 5 L
Seeding cell concentration (106cell/mL) 0.2 1.5-2.0 1.5-2.0 1.5-2.0 1.5-2.0
Cell suspension initial volume 100 lL 1.8 mL 5.0-5.4 mL 22.5 mL 4.8 L
Final DNA concentration 1 lg/mL 1 lg/mL 1 lg/mL 1 lg/mL 0.75 lg/mL
DNA:LPEI ratio w:w 1:3 1:3 1:3 1:3 1:3
DNA solution volume 20 lL 100 lL 500-300 lL 1.25 mL 100 mL
PEI solution volume 80 lL 100 lL 500-300 lL 1.25 mL 100 mL
Agitation speed No (adherent) 120 rpm 140 rpm 120 rpm 25 rpm, 8°; 28 rpm,
9,5° 24 hpt
46 C. Raymond et al. / Methods 55 (2011) 44–51

of PEI. Indirect method: DNA:PEI ratio was 1:2.5 (w:w). PEI was 3. Results
added dropwise to DNA. The mixture was vortexed for 4 s and
incubated 15 min at room temperature. Polyplexes were then 3.1. Optimization of transfection parameters for the direct method
added to the cells.
In all cases, cells were fed 24 h post-transfection with TN1 pep- We first determined the optimal parameters for the direct
tone to a final concentration of 0.5% (w/v) to enhance productivity transfection of 293-6E cells grown in agitated 6-well plates in ser-
[29]. Details specific to each scale and experiment are described in um-free F17 medium. The pTT vector expressing the human se-
Table 1. creted placental alkaline phosphatase (SEAP) was used as it
allows rapid and quantitative measurement of the product in the
culture medium 5 days post-transfection using a simple and robust
2.4.1. Transfection in 5 L WAVE Bioreactor colorimetric assay [7,30]. The optimal DNA concentration and
Scalability to 5 L bioreactors was assessed with the chimeric DNA:PEI ratio were first determined. Various concentrations of
B72.3 antibody. The culture medium (3.5 L) was transferred and plasmid DNA were added to the culture followed by 5 min incuba-
acclimatized for one hour at 37 °C in a 10L WAVE Cellbag™ (GE tion under agitation to allow for a uniform mixing. Then various
Healthcare, Piscataway, NJ) inflated at 0.1 bar and aerated at concentrations of PEI were added to the culture and the plates re-
100 ccm with 95% air + 5% CO2. The rocking speed and angle were turned to the incubator for 5 days. Each set of experiment was
18 rpm and 8° to minimize foaming. Cells grown in shaker flasks independently repeated three times. SEAP activity in the culture
were transferred in the WAVE bag (final volume of 4.8 L) to obtain medium was then measured using pNPP as a substrate and ex-
an initial density of 0.40–0.45  106 cells/mL. The rocking speed pressed as relative value (in %). Fig. 1 (panels A and B) shows that
and angle were then set at 25 rpm, 8° for the growth phase. Two the optimal plasmid DNA concentration lies between 0.5 and 1 lg
days later, cells were co-transfected with the LC and HC plasmids per mL of culture with 2.5–3.0 lg/mL LPEI, a slightly higher SEAP
at a 6:4 ratio (w:w). Plasmids were added to the culture at a final productivity being obtained at DNA and LPEI concentrations of 1
concentration of 0.75 lg/mL with a DNA:LPEI (w:w) ratio of 1:3 and 3 lg/mL, respectively. Maximal absolute SEAP titers were of
(direct method) or 1:2.5 (indirect method). Direct transfection 62, 67 and 75 mg/L for DNA:LPEI concentrations of 0.5:2.5,
was achieved by adding 100 mL of DNA followed 5 min later by 0.75:2.5 and 1:3 lg/mL, respectively (averages of 3 independent
100 mL of LPEI. The transfer line was rinsed with 50 mL of medium experiments). For the branched 25 kDa PEI (BPEI), maximum SEAP
after each addition. For indirect transfection, 200 mL of polyplexes productivity was observed with 1.5 lg/mL of DNA with 2.5–3.0 lg/
were prepared and incubated for 15 min before being added to the mL BPEI. Maximal absolute SEAP titer was of 65 mg/L at a ratio of
cells, followed by a rinse of the transfer line with 50 mL of medium. 1.5:3. As previously observed using the indirect method, LPEI was
Twenty-four hours post-transfection, the cultures were fed with more efficient than BPEI at transfecting HEK293 cells [7,18] and
TN1 (0.5% w/v final). Then the rocking speed was increased to increasing the concentrations of plasmid and PEI (Fig. 1B) did not
28 rpm and the angle to 9.5°. Culture samples were collected every increase productivities [31]. A DNA:LPEI ratio of 1:3 with 1
day for 7 days post-transfection. lg/mL of plasmid DNA was thus chosen for the subsequent
experiments.
Since in the direct method plasmid DNA is directly added to the
2.5. Products analyses
culture, a minimum incubation time may be needed to allow its
adequate mixing prior to addition of PEI in order to obtain optimal
2.5.1. GFP
transfection efficiency. In addition, the presence of nucleases re-
GFP was analyzed 48 h post-transfection (hpt) by flow cytome-
leased by the cells may reduce productivity by degrading the plas-
try using a BD LSRII cytometer (BD Biosciences, Mississauga, ON).
mid DNA. Thus the impact of incubation time with plasmid DNA
For that purpose, 200 lL were taken out from the cultures in 6-well
prior to LPEI addition was evaluated. As shown in Fig. 2A, an incu-
plates (500 lL from cultures of 5 mL and more). Only viable cells
bation time of 5–15 min with the DNA prior to PEI addition yielded
were analyzed for the expression of GFP. Remaining cultures were
a maximum of 65–70% of GFP-expressing cells, as measured by
kept in the incubator for subsequent SEAP analysis.
flow cytometry 2 days post-transfection. Similarly, the same incu-
bation time allowed to reach maximal SEAP expression as mea-
2.5.2. SEAP sured 5 days post-transfection (Fig. 2B). Longer incubation time
Unless otherwise indicated, SEAP was assessed at day 5 post- yielded only slightly lower titers, suggesting that plasmid DNA is
transfection (dpt). SEAP transfected cells were centrifuged and not significantly degraded by cell-derived nucleases that may
the supernatant was diluted in water (typically 1/1000). For the accumulate in the conditioned medium.
96-well plate scale, the culture medium was sampled without cen-
trifugation step and diluted in water (1/1000). Fifty microliter of
diluted supernatant was transferred to a 96-well plate and mixed 3.2. Comparison of direct and indirect transfection methods
with an equal volume of SEAP assay solution containing 20 mM
paranitrophenyl phosphate (pNPP), 1 mM MgCl2, and 1 M dietha- 3.2.1. Comparison in ThinCert™ plates and 125 mL shake flasks
nolamine. Absorbance was read at 405 nm at 1 min intervals for The direct transfection approach was then compared to the
5 min. commonly used indirect method (i.e. formation of polyplexes prior
to their addition to the culture) using the optimal parameters
determined for both methods. For the direct transfection, DNA
2.5.3. Chimeric B72.3 antibody and PEI concentration used were 1 and 3 lg/mL with a DNA pre-
Concentration of the chimeric B72.3 antibody from direct and incubation of 5 min prior to PEI addition. For the indirect transfec-
indirect methods in WAVE bioreactor were determined by tion, previously determined optimal parameters (not shown) were
protein-A HPLC using a 800 lL POROSÒ 20 micron Protein A ID Car- 1 lg/mL of DNA, and 2.5 lg/mL of PEI (DNA:PEI ratio of 1:2.5).
tridge (Applied BioSystems, Foster City, CA) according to the man- Transfections conditions were directly scaled up from 2 mL cul-
ufacturer’s recommendations. The antibody present in the culture tures in 6-well plates to 6 mL cultures in ThinCert™ 6-deep well
medium was also visualized following reducing and non-reducing plates and 25 mL cultures in 125 mL shaker flasks. SEAP expression
SDS–PAGE stained by Coomassie. was measured 5 days post-transfection for ThinCert™ plates
 C. Raymond et al. / Methods 55 (2011) 44–51 47

A 120

Relative SEAP activity (%)

DNA 0.5 ug/ml
100 DNA 0.75 ug/ml
DNA 1 ug/ml
80

60

40

20

0
0.5 1 1.5 2 2.5 3 3.5 4
LPEI (ug/ml)

B 120
Relative SEAP activity (%)

DNA=1ug/ml
100 DNA=2ug/ml
DNA=3ug/ml
80

60

40

20

0
1.5 2 2.5 3 3.5 4 4.5 5
LPEI (ug/ml)

C
120
Relative SEAP activity (%)

DNA 0.5 ug/ml

DNA 0.75 ug/ml
100 DNA 1 ug/ml
DNA 1.5 ug/ml
80

60

40

20

0
0.5 1 1.5 2 2.5 3 3.5 4
BPEI (ug/ml)

Fig. 1. Effect of DNA and PEI concentrations on SEAP production following direct transfection. Cells were grown in agitated 6-well plates at 1.5  106 cells/mL in 1.8 mL of
complete culture medium. Plasmid DNA and PEI solutions were each diluted in 100 lL of culture medium at various concentrations. Following DNA addition, the plates were
returned to agitation in the incubator for 5 min, then PEI was added to the cultures. Tryptone N1 was added 24 hpt at 0.5% (w:v) and SEAP measured in the culture medium
5 days post-transfection (dpt) as indicated in Section 2. Error bars are standard deviation from three independent experiments.

(Fig. 3A) or followed daily over 8 days post-transfection for the 3.2.2. Comparison in a WAVE bioreactor
shaker flasks (Fig. 3B). With both the direct and indirect methods, To demonstrate scalability of the process, a comparison of the
SEAP accumulation in the culture medium was very similar. For the direct and indirect methods was also done at the 5L-scale using a
ThinCert™ plates, slightly higher SEAP activity was obtained (70 vs WAVE bioreactor with the chimeric B72.3 monoclonal antibody.
61 mg/L) when DNA and PEI solution volumes represented 16% vs Expression of the monoclonal antibody was achieved by co-trans-
10% (1 vs 0.6 mL) of the final culture volume, respectively, and both fection of two plasmids, one encoding the light chain (LC) and the
conditions yielded higher titers compared to the direct method other one the heavy chain (HC). The optimal LC:HC ratio for achiev-
(58 mg/L). Surprisingly, transfection efficiencies (% GFP-positive ing highest Mab expression was assessed in 6-well plates and
cells) were higher in the indirect (53%) vs direct method (39– found to be 3:2 using 0.75 lg/mL DNA and 2.25 lg/mL PEI (DNA:-
43%). For the shaker flask experiments, a steady SEAP increase PEI ratio of 1:3, not shown). This optimal LC:HC ratio is slightly dif-
was observed for up to day 4 post-transfection and then reached ferent to the 2:3 ratio previously found for the same antibody
a plateau at day 5. Maximum titers found at day 5 post-transfec- transiently expressed in CHO cells [32]. These optimal conditions
tion were of 75 and 88 mg/L for the direct and indirect transfection, were used to transfect two 5 L cell cultures in 10 L Cellbag™ using
respectively. the direct and indirect methods (see Section 2 for culture and
48 C. Raymond et al. / Methods 55 (2011) 44–51

A 80 A
%GFP-positive cells

80

GFP (%)
60 70
60
40 50

SEAP (mg/L) or
40
20 30
20
0
10
0 10 20 30 40
0
Indirect Direct Direct
B 160
(2x300uL) (2x500uL)
SEAP (mg/L)

120 Transfection method

80 B 120
40 100

SEAP (mg/L)
80
0
0 10 20 30 40 60
Incubation time (min) 40

Fig. 2. Impact of cells incubation time with DNA before LPEI addition on 20
transfection efficiency. Cells (1.5  106/mL) were plated in 6-well plates (1.8 mL
per well). DNA and PEI solutions were prepared to get a DNA:PEI ratio of 1:3 in the
0
0 1 2 3 4 5 6 7 8 9
culture. 100 lL of plasmid DNA solution was added to the wells then 100 lL of LPEI
solution was added at various time points after DNA addition. Plates were returned Days post-transfection
to agitation in the incubator after each transfection, at 120 rpm under standard
conditions. GFP was measured by flow cytometry 48 hpt (panel A) while SEAP Fig. 3. Comparison of SEAP production in 6-deep well ThinCert™ plates or 125-mL
activity was determined 5 dpt (panel B). Each point represents the average SEAP shaker flasks using the direct or indirect methods. (A) Cells in 6-deep well
values from transfections done in triplicate ±SD. ThinCert™ (Greiner Bio-One) plates were transfected as follows: direct method – 5.4
or 5.0 mL of cultures at densities of 1.5–2.0  106 cells/mL were added to each well.
Then 300 or 500 lL of plasmid DNA (95% SEAP plasmid and 5% GFP plasmid) were
transfection procedure). The cultures were sampled daily for up to added to the wells containing 5.4 or 5.0 mL of culture, respectively. After 5 min of
7 days post-transfection and antibody titers measured by protein- incubation at 37 °C, 300 or 500 lL of LPEI solution were added, for a final volume of
6 mL/well in all cases. GFP (black bars) was monitored by flow cytometry 48 hpt
A HPLC using an in-house purified and quantified chimeric B72.3
while SEAP activity (gray bars) was measured 5 dpt; indirect method – 5.4 mL of
standard (Fig. 4A). Supernatants from 5 days post-transfection cultures at densities of 1.5–2.0  106 cells/mL were added to each well. 600 lL of
were also analyzed by reducing and non-reducing Coomassie- polyplexes were then added for a final volume of 6 mL/well. (B) 125 mL Erlenmeyer
stained SDS–PAGE (Fig. 4B). Maximum antibody titers were shaker flasks containing 22.5 mL of culture at densities between 1.5 and
2.0  106 cells/mL were transfected using 25 lg of plasmid DNA (95% SEAP plasmid
reached on day 6 and corresponded to 116 and 103 mg/L for the
and 5% GFP plasmid) in 1.25 mL of medium followed 5 min later by the addition of
cultures transfected with the direct and the indirect methods, 75 lg of LPEI in 1.25 mL of medium (direct method; open squares) or 2.5 mL of
respectively. This shows that the direct method is a viable and sim- polyplexes made of 25 lg of DNA and 62.5 lg of LPEI (indirect method; open
ple alternative to the traditional polyplexes transfection approach triangles). GFP was monitored by flow cytometry 48 hpt while SEAP activity was
for large-scale transfections. measured daily until 8 dpt. Data are mean values of 3 independent experiments
±SD.

3.3. High-throughput transfection and impact of culture mode

columns were all saturated as their expected capacity are 20 lg of
3.3.1. Miniprep plasmid DNA production DNA according to the manufacturer. The plasmid quality was very
We studied here the feasibility and reproducibility of the direct good as indicated by the average A260/280 ratio of 1.89. This sug-
transfection process in 96-well plates. We paid special attention to gests that plasmid DNA concentration determination prior to
make the protocol easily amenable to automation. The first param- transfection is not necessary as long as column saturation is
eter that we addressed was suitability and reproducibility of mini- reached, thus greatly simplifying automation of the process.
prep plasmid DNA for efficient transfection of HEK293 cells. One
critical aspect was to show that plasmid yield is constant not only 3.3.2. Transfection in 96-well plates with miniprep DNA
between minipreps done in parallel, but also between minipreps The transfection conditions in 96-well plates were optimized in
done at different times. To that purpose, 24 E. coli colonies trans- a final culture volume of 200 lL. The optimal DNA:LPEI ratio and
formed with the pTT/SEAP plasmid were picked to inoculate concentration was found to be 1:3 and 200 ng of plasmid per well,
2 mL of culture medium per well in a 24-deep well plate. The respectively (data not shown). The impact of cell number per well
cultures were then grown overnight at 37 °C. The following day, on volumetric yield was also tested. Volumetric SEAP yield in-
24 pTT/SEAP plasmid minipreps were performed using a commer- creased steadily from 2  104 to 4  104 cells/well (e.g. 1  105 to
cial kit (Qiagen). A second set of 24 minipreps was performed again 2  105 cells/mL) and then slightly declined at 5  104 cells/well
as a completely independent experiment, for a total of 48 mini- (Fig. 5A). Subsequently, 72 wells of a 96-well plate were transfec-
preps. The plasmid DNA was eluted from the columns using ted with 24 pTT-SEAP plasmid minipreps in triplicates using the di-
200 lL of TE buffer and quantified by absorbance at 260 and rect method (Fig. 5B). The average SEAP titer measured 5 days
280 nm using a Nanodrop™ spectrophotometer. The first mini- post-transfection was 17.2 ± 2.7 mg/L. This SEAP productivity ob-
preps series yielded an average of 26.6 ± 1.0 lg while the second served in Fig. 5A and B was about four-times lower than that ob-
series gave 28.1 ± 1.6 lg (Table 2). This suggests that the minipreps tained with transfection in agitated 6-well plates. To verify that
 C. Raymond et al. / Methods 55 (2011) 44–51 49

method (Fig. 6). Cells transfected in non-agitated 6-well plates se-

A
creted 25.4 ± 4.4 mg/L of SEAP while agitated cultures produced
Chimeric B72.3 antibody (mg/L)
140
Direct 117.1 ± 16.5 mg/L (average of two independent experiments done
120 Indirect in triplicates). This clearly shows that suspension-growing cells al-
100 low reaching titers that are 4–5 times higher compared to adherent
80
cells. This difference likely result from the higher optimal cell den-
sity at time of transfection of suspension vs adherent cells (1.5–
60 2.0  106 cells/mL vs 0.36  106 cells/mL, respectively). The lower
40 SEAP yield measured in the 96-well plate format (17.2 mg/L, panel
B) compared to the non-agitated 6-well plate (25.4 mg/L, panel D)
20 may also be explained by the lower optimal cell density found for
0 the former culture format.
0 1 2 3 4 5 6 7 8
Days post-transfection
4. Discussion

B D I D I With the continual increase in demand for high-throughput and

high-level protein expression, robust and simple processes provid-
ing predictable protein yields are needed. The robustness, versatil-
200
ity and relative simplicity of PEI-mediated transient expression
116 systems concomitant with the use of performing cell lines and epi-
97 somal expression vectors [10–12] have made protein expression in
66 mammalian cells a valuable alternative to bacterial systems. For a
given cell line, the efficiency of PEI-mediated transfection depends
55
on many parameters such as the culture medium in which the
transfection occurs, the mode of cell culture (i.e. adherent vs sus-
37 pension), the cell density at time of transfection, the DNA concen-
31 tration in the culture medium and the DNA:PEI ratio. The
formation of polyplexes by mixing together plasmid DNA and a
22 polycation is a dynamic process involving particle formation,
14 growth and aggregation. As the size and polydispersity of
polyplexes are important parameters for efficient gene transfer
-DTT +DTT
[20,33], it is a significant source of variation in transfection effi-
Fig. 4. Chimeric B72.3 monoclonal antibody productions in 5 L WAVE bioreactors. ciencies, especially when up-scaling or down-scaling culture vol-
following direct or indirect transfections. 4.8 L of culture in a 10 L Cellbag™ were umes [34]. The preparation of polyplexes, which consists of
transfected with the direct or indirect method as described in Section 2. The mixing DNA with a cationic polymer such as PEI, is typically done
cultures were sampled daily for up to 8 days post-transfection. (A) Antibody titers
via pipetting and as such, is restricted to relatively small volumes.
were determined by protein-A HPLC against a purified chimeric B72.3 standard.
(B) 20 lL of culture medium sampled at day 5 was analyzed by SDS–PAGE under When performing large-scale (multi-liters) transfections, mixing of
reducing or non-reducing condition. D: direct transfection; I: indirect transfection. DNA with PEI is often done by manual swirling in glass bottles, a
process that is difficult to control and which imposes a limitation
on the volume that can be handled. In addition, high-throughput
PEI-mediated transfection in 96-well plates is challenging as it in-
Table 2
Yield and purity of miniprep plasmid DNA. volves very small volumes that are not easily amenable to efficient
and reproducible mixing. To overcome these challenges we have
Plasmid 1st series 2nd series Average of 48
optimized parameters for the direct transfection approach, which
miniprep (average ± SD) (average ± SD) minipreps (±SD)
consists of the addition of plasmid DNA and PEI directly to the cul-
Concentration 133.2 ± 5.0 140.5 ± 8.4 136.8 ± 7.8
ture. We have shown that this approach provides recombinant pro-
(lg/mL)
Yield (lg) 26.6 ± 1.0 28.1 ± 1.6 27.4 ± 1.6 tein titers similar to the usual indirect method based on the
A260/A280 1.93 ± 0.01 1.86 ± 0.02 1.89 ± 0.04 addition of pre-made polyplexes to the culture. As the direct trans-
fection method implies less pipetting and mixing steps, it is also
well suitable for automation in 96-well plates. Even though trans-
fection of adherent cells generates 4- to 5-fold less product com-
this lower yield was not due to contaminants present in the mini-
pared to transfection of suspension cells, a likely result of the 4-
preps DNA, we compared productivity of suspension cells in 6-well
to 5-fold lower optimal transfection density of adherent vs suspen-
plates using miniprep and maxiprep plasmid DNA (Fig. 5C). The
sion cultures, transfection in 96-well plate allows the high
average SEAP productivity and transfection efficiency obtained
throughput parallel evaluation of multiple variants for expression
with minipreps DNA was of 112.1 ± 2.3 mg/L and 78.9 ± 0.6% vs
screening.
123.3 ± 11.7 mg/L and 82.0 ± 4.6% for maxipreps. This demon-
The direct transfection approach has also been described re-
strates that the use of miniprep DNA provides volumetric yields
cently by Backliwal et al. [35]. In their study, the authors per-
only slightly lower to those obtained with maxiprep DNA. This
formed the direct transfection using a high cell density
might be due to the significantly higher level of endotoxins mea-
suspension (20  106 cells/mL) that was obtained following centri-
sured in those DNA preparations (see Fig. 5 legend for details).
fugation of a seed train. The cell suspension was then diluted with
fresh medium to 1  106 cells/mL four hours post-transfection.
3.3.3. Transfection of adherent vs suspension cultures While this approach was shown to provide twice as much titers
The impact of the culture mode (adherent vs suspension) on compared to their direct method, it is difficult to perform at
SEAP productivity was evaluated using the direct transfection large-scale as it does involve a centrifugation step.
50 C. Raymond et al. / Methods 55 (2011) 44–51

A 20

15

SEAP (mg/L)
10

0
20 30 40 50

Cell number/well (x10E04/mL)

B 25

20
SEAP (mg/L)

15

10

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
miniprep number

C 160
GFP (%)

140
120
100
SEAP (mg/L) or

80
60
40
20
0
mini #1 mini #2 mini #3 maxi #1 maxi #2 maxi #3

D mini#1 mini#2 mini#3 maxi#1 maxi#2 maxi#3

kb
10.0
8.0
6.0
5.0
4.0
3.0
2.0
1.5
1.0

0.5

Fig. 5. Transfection in 96-well plates. (A) the optimal cell number per well was first evaluated. Twenty microliter of plasmid DNA were first added to the wells. Then, 100 lL
of cells at various densities were plated and mixed with DNA by one up and down pipetting. After 5 min at room temperature, 80 lL of LPEI was added to each well. The large
volume of PEI solution was increased to 40% of the final culture volume (200 lL/well) to enhance mixing of DNA and PEI with the cells. After transfection, cells were returned
to the incubator under standard conditions, without agitation. (B) SEAP titers from 72 transfections performed in 96-well plates using 24 pTT/SEAP plasmid minipreps.
Twenty microliter of plasmid DNA were first added to the wells, then 100 lL of cells at 0.4  106cells/mL were added followed by 80 lL of LPEI as described above. (C) SEAP
titers 5 dpt (gray bars) and % GFP-positive cells 2 dpt (black bars) obtained from transfection suspension cultures in 6-well plates using three independent minipreps (mini) or
maxipreps (maxi) DNA. (D) to evaluate the relative quality of miniprep vs maxiprep DNA, 6 ll of the three minipreps (corresponding to about 850 ng) shown in panel C was
loaded side-by-side with 1 lg each of the three maxiprep on an agarose gel and stained with ethidium bromide. The endotoxin levels as determined by the LAL method was
very high in the miniprep DNA (>75000 EU/mg) compared to maxiprep DNA (average of 2500 EU/mg).
 C. Raymond et al. / Methods 55 (2011) 44–51 51

140 costs for the culture vessels are reduced by a factor of 4. Com-
120 bined with the significantly increased volumetric productivity ob-
tained with suspension cells, these factors strongly illustrate the
SEAP (mg/L)

100
benefits of suspension over static cultures for recombinant protein
80 production by transfection.
60 In conclusion, the direct transfection of 293 cells in serum-free
suspension cultures offers a simple, robust and scaleable process
40
for small-scale high-throughput and large-scale protein expression
20 with volumetric yields comparable to the commonly used indirect
0 method. As the process can use miniprep- or maxiprep-quality
Adherent Suspension DNA without impacting transfection efficiency or productivity
Culture condition and does not require a medium exchange at any time, it greatly
facilitates automation as well as large-scale operations.
Fig. 6. Transfection of adherent vs suspension cultures. Productivity of adherent vs
suspension cells was compared in 6-well plates. Cells (1.8 mL) were respectively Acknowledgments
plated at 4  105 or 1.5–2  106 cells/mL for the adherent or suspension cultures.
Plasmid DNA (2 lg) in 100 lL of culture medium was then added, followed by the
addition of 100 lL of LPEI (6 lg). SEAP activity was measure 5 dpt. Data show the C. Raymond is supported by the Natural Sciences and Engineer-
average of 2 independent experiments done in triplicates ±SD. ing Research Council of Canada (NSERC) MabNet Strategic Net-
work. This is a NRC publication number 52782.
We have shown here that the direct transfection method can be
applied to 6-well or 6-deep well plates, 125 mL shaker flasks and to References
large-scale multi-liter transfection in bioreactors using exactly the
[1] C. Bieniossek et al., Nat. Methods 6 (6) (2009) 447–450.
same conditions. As ThinCert™ 6 deep-well plates can be stacked [2] S. Trowitzsch et al., J. Struct. Biol. 172 (1) (2010) 45–54.
easily and can accommodate up to 8 mL of culture/well, we found [3] J. Koehn, I. Hunt, Methods Mol. Biol. 498 (2009) 1–18.
them convenient for producing significant quantities of proteins in [4] J.E. Nettleship et al., J. Struct. Biol. 172 (1) (2010) 55–65.
[5] H. Lin et al., Science 320 (5877) (2008) 807–811.
a high-throughput manner. The method can even be adapted to 24 [6] Y. Durocher, M. Butler, Curr. Opin. Biotechnol. 20 (6) (2009) 700–707.
deep-well plates since transfection of 293-EBNA1 cells grown in [7] Y. Durocher, S. Perret, A. Kamen, Nucleic Acids Res. 30 (2) (2002) E9.
these plates was shown to be efficient [36,37]. Moreover, we have [8] C. Liu et al., Mol. Biotechnol. 39 (2) (2008) 141–153.
[9] P.L. Pham et al., Biotechnol. Bioeng. 84 (3) (2003) 332–342.
shown that plasmid DNA prepared using silica-based commercial [10] L. Baldi et al., Biotechnol. Lett. 29 (5) (2007) 677–684.
miniprep kits is fully compatible with PEI-mediated transfection, [11] P.L. Pham, A. Kamen, Y. Durocher, Mol. Biotechnol. 34 (2) (2006) 225–237.
giving expression levels only slightly lower to those obtained with [12] S. Geisse, Protein Expr. Purif. 64 (2) (2009) 99–107.
[13] J.E. Lee, M.L. Fusco, E. Ollmann Saphire, Nat. Protoc. 4 (4) (2009) 592–604.
anion-exchange purified plasmids. The quantity of plasmid DNA [14] J.E. Nettleship et al., Protein Expr. Purif. 62 (1) (2008) 83–89.
generated with miniprep kits (25 lg) is sufficient to transfect [15] T. Battle et al., Comb. Chem. High Throughput Screen 9 (9) (2006) 639–649.
up to 25 mL of cell culture, which is enough for small-scale affinity [16] S. Paris, A. Burlacu, Y. Durocher, J. Biol. Chem. 283 (12) (2008) 7697–7704.
[17] I.A. Khalil et al., Pharmacol. Rev. 58 (1) (2006) 32–45.
purification and crystallization trials [13]. For that purpose, the
[18] Y. Durocher et al., Recombinant protein production with prokaryotic and
addition of the glycosyltransferase inhibitor kifunensine can be eukaryotic cells. A comparative view on host physiology, Recombinant Protein
used to generate glycoproteins with high mannose-type glycosyla- Production by Transient Transfection of Suspension-Growing Cells, Kluwer
academic publishers, Dordrecht, The Netherlands, 2001.
tion that are more susceptible to endoglycosidases treatment and
[19] E.J. Schlaeger, K. Christensen, Cytotechnology 30 (1–3) (1999) 71–83.
to crystallization [38,39]. Preliminary results have shown that this [20] X. Han et al., Cytotechnology (2009). PMID:19649718.
transfection method can also be applied to suspension CHO cells [21] S.C. De Smedt, J. Demeester, W.E. Hennink, Pharm. Res. 17 (2) (2000) 113–126.
grown in serum-free medium (data not shown). [22] R. Kircheis, L. Wightman, E. Wagner, Adv. Drug Deliv. Rev. 53 (3) (2001) 341–
358.
Still today, many high-throughput transient expression plat- [23] O. Boussif et al., Proc. Natl. Acad. Sci. USA 92 (16) (1995) 7297–7301.
forms use adherent 293 cells [13,38,40]. We have shown here that [24] R. Tom, L. Bisson, Y. Durocher, Cold Spring Harbor Protocols 2008 (4) (2008).
optimal transfection conditions differ significantly between adher- pdb.prot4976.
[25] D.J. King et al., Biochem. J. 281 (Pt 2) (1992) 317–323.
ent and suspension cultures, and that suspension cultures provide [26] A. Thor et al., Cancer Res. 46 (6) (1986) 3118–3124.
significantly higher volumetric yields vs adherent ones. Also, cell [27] C. Shi et al., Biochemistry 44 (48) (2005) 15705–15714.
density measurement at time of transfection, a critical parameter [28] J. Zhang et al., Protein Expr. Purif. 65 (1) (2009) 77–82.
[29] P.L. Pham et al., Biotechnol. Bioeng. 90 (3) (2005) 332–344.
for optimal gene transfer and expression, can only be qualitatively [30] Y. Durocher et al., Anal. Biochem. 284 (2) (2000) 316–326.
estimated when growing adherent cultures. As for scaling-up pro- [31] E. Carpentier et al., J. Biotechnol. 128 (2) (2007) 268–280.
duction with adherent cells, it requires costly culture flasks such as [32] S. Schlatter et al., Biotechnol. Prog. 21 (1) (2005) 122–133.
[33] J.Y. Cherng et al., Eur. J. Pharm. Biopharm. 47 (3) (1999) 215–224.
cell factories (e.g. CellStack™ system) which involve extensive
[34] J.C. Kasper et al., Eur. J. Pharm. Biopharm. 77 (2011) 182–185.
manipulations and complicate cell harvesting for intracellular or [35] G. Backliwal et al., Biotechnol. Bioeng. 99 (3) (2008) 721–727.
membrane-associated products. A single 6000 cm2 cell factory [36] S.D. Chapple et al., BMC. Biotechnol. 6 (1) (2006) 49.
[37] A. Davies et al., Protein Expr. Purif. 42 (1) (2005) 111–121.
(10 layers of 600 cm2) that typically uses 2 L of culture medium
[38] V.T. Chang et al., Structure 15 (3) (2007) 267–273.
provides up to 1  109 cells at time of transfection, compared to [39] J.E. Nettleship, N. Rahman-Huq, R.J. Owens, Methods Mol. Biol. 498 (2009)
up to 4  109 cells with suspension cultures. As 2 L of suspension 245–263.
cells can be easily grown in two 3 L Erlenmeyer shaker flasks, the [40] N.S. Berrow et al., Nucleic Acids Res. 35 (6) (2007) e45.