Limiting factors governing protein expression following polyethylenimine-mediated gene transfer in HEK293-EBNA1 cells

Journal of Biotechnology 128 (2007) 268–280

Limiting factors governing protein expression following polyethylenimine-mediated gene transfer in HEK293-EBNA1 cells Eric Carpentier, S´ebastien Paris, Amine A. Kamen, Yves Durocher ∗ Animal Cell Technology Group, Biotechnology Research Institute, National Research Council Canada, 6100 Royalmount Avenue, Montreal, Quebec, Canada H4P 2R2 Received 12 July 2006; received in revised form 2 October 2006; accepted 16 October 2006

Abstract Transient gene expression in mammalian cells is intensively used for the rapid generation of recombinant proteins for biochemical, biophysical and pre-clinical studies. Still, the principles behind DNA transfer to the cells and the cellular cascade of events that ultimately dictate protein expression levels are not fully understood. Using polyethylenimine (PEI) mediated transfection of HEK293-EBNA1 cells, we sought to determine the most critical parameters that drive and limit recombinant protein production. Our results showed that a maximum of 65,000 plasmid copies/cell can be recovered in total extracts at 1 day post-transfection. Analyses performed after cell sorting revealed equal amounts of plasmid DNA in GFP-positive and -negative populations. However, nuclear plasmid content was three-fold higher in GFP-positive cells (1850 copies) than in GFP-negative cells (550 copies). The fact that significant amounts of plasmid DNA are found in the nucleus of GFP-negative cells suggests that its transcriptional competency is impaired. Interestingly, transfecting cells using a wide range of plasmid quantities at the optimal DNA:PEI ratio did not significantly affect the number of expressing cells. Thus, it appears that successful transgene expression is more likely to depend on a cellular “competent” state than to the quantity of plasmid DNA delivered per cell. Moreover, Northern blot analysis and SEAP/GFP measurement following plasmid titration experiments showed that transcriptional and translational processes are operating near to saturation under optimal transfection conditions. Overall, our results suggest that events that regulate nuclear translocation of plasmid DNA and its transcriptional competency as well as translational/post-translational limitations represent major bottlenecks in the success of a PEI-mediated protein production. © 2006 Elsevier B.V. All rights reserved. Keywords: Transient gene expression; Polyethylenimine; HEK293 cells; pTT vector; Protein production

Abbreviations: hpt, hours post-transfection; PEI, polyethylenimine; r-protein, recombinant protein; GFP, green fluorescent protein; BFP, blue fluorescent protein; SEAP, secreted alkaline phosphatase; BCS, bovine calf serum; stDNA, salmon testes DNA; EBNA1, Epstein–Barr nuclear antigen 1; HSFM, hybridoma serum-free medium; HEK293, human embryonic kidney 293 cells; 293E, HEK293-EBNA1 cells; pNPP, para-nitrophenyl phosphate; BSA, bovine serum albumin; DIG, digoxigenin ∗ Corresponding author. Tel.: +1 514 496 6192; fax: +1 514 496 6785. E-mail addresses: [email protected], [email protected] (Y. Durocher). 0168-1656/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.jbiotec.2006.10.014

E. Carpentier et al. / Journal of Biotechnology 128 (2007) 268–280

1. Introduction Delivery of exogenous DNA in animal cells is a widely used process in biological sciences for the expression of functional recombinant proteins (rproteins). The technique usually involves the use of plasmid DNA, either alone (electroporation) or complexed with natural or synthetic carrier molecules such as cationic lipids or polymers (Wiethoff and Middaugh, 2003). With the development of economic and efficient gene transfer processes, large-scale transfection of mammalian cells has gained major interest for the very fast production of r-proteins for pre-clinical, biochemical and biophysical studies (Baldi et al., 2005; Derouazi et al., 2004; Durocher et al., 2002; Girard et al., 2002; Jordan et al., 1998; Pham et al., 2003, 2005; Schlaeger and Christensen, 1999; Schlaeger et al., 2003; Shi et al., 2005). Such process allows for the generation of milligram to gram amounts of recombinant protein in a very short period of time (usually 1–3 weeks), which contrasts with the classical approach of stable cell lines establishment that often takes several months (for reviews, see: Pham et al., 2006; Wurm and Bernard, 1999). Although usually more expensive than prokaryotic expression systems, mammalian cells are well suited for the biomanufacturing of human proteins as they normally insure proper post-translational modifications that are often essential for many of the most promising therapeutic proteins. The versatility and ease of transfection also allow for simultaneous screening of different cell lines and plasmid constructs for the optimization of r-protein yield, function and quality. Polyethylenimines (PEIs) are cationic polymers widely used in the industry for over 30 years in various applications, such as water purification, mineral extraction, shampoo formulation, etc. PEI was first reported as an efficient transfection reagent by Boussif et al. (1995) and is now often described as one of the most promising non-viral vector for gene therapy. Schlaeger and Christensen (1999) have first demonstrated PEI’s usefulness for large-scale transfection and it has since gained increasing attention from the scientific community; PEI is a cost-effective transfection reagent that has been reported many times as one of the most efficient cationic compound for in vitro delivery of plasmid DNA into mammalian cells.

269

PEI efficiently condenses DNA and it is well acknowledged that PEI/DNA complexes (polyplexes) are taken up by the cell via interactions with heparan sulphate proteoglycans expressed at the cell surface (Kopatz et al., 2004) followed by endocytosis into acidified endosomal compartments (Godbey et al., 1999). Since PEI has a substantial buffering capacity, it is believed that polyplexes are released into the cytoplasm following endolysosome rupture induced by the socalled “proton sponge” effect (Boussif et al., 1995). The polyplexes must then find their way to the nucleus where transcription of plasmid DNA can take place. How nuclear translocation of polyplexes occurs is not well characterized. As reported for other transfection reagents, it is believed to be one of the major obstacles to high transfection efficiency (Bieber et al., 2002; Brunner et al., 2002; Escriou et al., 2001; Pollard et al., 1998; Zabner et al., 1995). For large-scale applications, increasing both the number of cells expressing the transgene (i.e. transfection efficiency) and cell’s specific productivity are direct ways to increase volumetric protein production. Most studies on PEI-mediated transfection are focused on designing better in vivo delivery systems for therapeutic applications, and as such, less efforts are dedicated toward optimization of r-protein production. Thus, the aim of this study was to deepen our understanding of the transfection process and to identify cellular bottlenecks that limit r-protein production following PEI-mediated gene transfer in HEK293EBNA1 cells. We looked at the fate and distribution of plasmid DNA in the culture and monitored the impact of plasmid dosage on DNA transfer efficiency, mRNA and protein synthesis.

2. Materials and methods 2.1. Chemicals The 25 kDa linear PEI was obtained from Polysciences (Warrington, PA). The polymer was solubilized in HCl-acidified water (1 mg/ml), neutralized with NaOH, sterilized by filtration (0.22 ␮m) and kept at −20 ◦ C. Salmon testes DNA (stDNA) was obtained from Sigma (Oakville, Ontario). TN1 peptone was obtained from OrganoTechnie SA (Teknisciences, Canada). Plasmids were purified using the Maxiprep

270

E. Carpentier et al. / Journal of Biotechnology 128 (2007) 268–280

kit from Qiagen (Valencia, CA). The dimeric cyanine dye YOYO-1 was obtained from Molecular Probes (Eugene, OR).

restriction site. YOYO-1 labelling of plasmid DNA was performed as described by the manufacturer. 2.3. Nuclei isolation

2.2. Cell culture, transfection and plasmids The human embryonic kidney 293 cell line stably expressing EBNA1 (293E) was maintained as a suspension culture in low-calcium HSFM supplemented with 1% bovine calf serum (BCS), 50 ␮g/ml geneticin, 0.1% pluronic F-68 (Sigma, Oakville, Ontario) and 10 mM HEPES (Durocher et al., 2002). Cells were routinely diluted every two or three (weekends) days at 0.2 × 106 /ml (0.07 × 106 /ml on Fridays) in 20 ml total volume contained in a 125 ml plastic Erlenmeyer flask (Corning). Cells doubling time was calculated to be 18 h and cultures were kept for a maximum period of three months, before thawing another vial. For transfection, cells were centrifuged and resuspended in fresh medium at a density of 0.5 × 106 cells/ml. The cell suspension was distributed in 6-well plates (2 ml/well) or kept in suspension in Erlenmeyer flasks (120 rpm). DNA mixes (total of 1 ␮g/ml of culture to be transfected, unless specified otherwise) were diluted in 5% of the initial culture volume in fresh serum-free medium. An equivalent volume of diluted PEI (2 ␮g of PEI per ␮g of DNA) was added to the DNA. The mixture was immediately vortexed and incubated for 15 min at room temperature prior to its addition to the cell culture. When required, cultures were fed 24 h post-transfection (hpt) with 0.5% TN1 peptone (w/v, final; Pham et al., 2005) using a concentrated solution (5%, w/v in HSFM), or with plain HSFM for controls. The pTT/SEAP and pTT/GFP vectors have been described previously (Durocher et al., 2002). The pTT/hCD4-GFP vector was constructed by PCR amplification of the extracellular and transmembrane domain (bases 1-1275) of human CD4 (ATCC: MGC-34241) with forward (5 TCAGAATTCACCATGAACCGGGGAGTCCCT-3 ) and reverse (5 -TGATCTAGATCGGTGCCGGCACCTGAC-3 ) primers. The PCR product was digested with EcoRI and XbaI and cloned in-frame with GFP in the pTT/GFP vector digested with the same enzymes. The pTT plasmid encoding the blue fluorescent protein (pTT/BFP) was constructed by subcloning of the sgBFPTM gene (Q-Biogen, Montreal, QC) from AdCMV5-BFP in the pTT vector using the BamHI

Cells (1 × 107 ) were first rinsed with PBS, pelleted, and resuspended in 500 ␮l of ice-cold cell lysis buffer (20 mM HEPES-KOH pH 7.9, 3 mM MgCl2 , 10 mM NaCl, 0.1% (w/v) Thesit, 10% glycerol, 0.2 mM EDTA). After 10 min on ice with frequent agitation, the nuclei were pelleted (500 × g for 5 min at 4 ◦ C) and washed twice with 2 ml of 20 mM HEPES-KOH pH 7.9, 0.2 mM EDTA, 20% glycerol prior to DNA extraction with DNAzol reagent (Invitrogen) using the manufacturer’s protocol. 2.4. Cell sorting Cells were transfected with the pTT/hCD4-GFP plasmid and the cells expressing the fusion protein were magnetically isolated using the MACSelect system (Miltenyi Biotec, Gladbach, Germany). Briefly, the transfected cells (1 × 108 ) were harvested at 48 hpt and resuspended in 3.2 ml of ice-cold PBE (PBS supplemented with 0.5% (w/v) BSA and 5 mM EDTA). The cells were then magnetically labelled on ice for 15 min by adding 500 ␮l of anti-hCD4 coated MACSelect 4 magnetic micro beads (Miltenyi Biotec). The volume was then adjusted to 5 ml with PBE and the cells were passed on a column attached to a strong magnet for magnetic separation. The cells passing through the column were harvested and constituted the negative population. After three PBE washes (4 ml each), the positive cells were eluted by removing the column from the magnet and by passing 5 ml of PBE into the column. The eluted positive cells were passed once again on the column to increase separation efficiency, which was monitored by flow cytometry analysis. 2.5. SEAP and GFP analyses Determination of SEAP activity was performed essentially as previously described (Durocher et al., 2000). GFP was analyzed by flow cytometry using an EPICS Profile II (Coulter, Hialeah, FL) equipped with a 15-mW argon-ion laser and only the viable cell population was considered. GFP and SEAP analyses were performed 48 or 72 hpt, respectively, before pro-

E. Carpentier et al. / Journal of Biotechnology 128 (2007) 268–280

tein accumulation shows saturation signs (Durocher et al., 2002). Results were only considered when SEAPmediated para-nitrophenyl phosphate (pNPP) hydrolysis rates were over 7 absorbance units/min for 50 ␮l of supernatant, corresponding to about 7.7 ␮g of SEAP protein/ml (in-house determination), or when at least 25% GFP positive cells were obtained. 2.6. Northern and Southern analyses For Northern analyses, total RNA extractions, electrophoretic separation, transfer and probing were performed essentially as described by the manufacturers. Briefly, total RNA was extracted using Trizol (Invitrogen) and quantified by spectrophotometry. Twenty micrograms of total RNA were resolved on a formaldehyde/agarose gel and transferred on a positively charged nylon membrane (Roche, Laval, QC). For Southern analyses, DNA was extracted from whole cells or nuclear preparations using DNAzol (Invitrogen) as described by the manufacturer. After isolation, DNA was treated with RNAse A (Sigma–Aldrich) (10 ␮g/ml), re-precipitated and quantified spectrophotometrically at 260/280 nm. Between 1 and 5 ␮g of DNA were digested overnight with XmnI (New England Biolabs, Beverly, MA) and resolved on a 0.8% agarose gel followed by alkaline transfer (0.4N NaOH, 1 M NaCl) to a nylon membrane. Known amounts of XmnI-linearized pTT/GFP were also loaded as a standard for quantitative analyses. For dot blot experiments, DNA was diluted in 0.4N NaOH, 1 M NaCl and directly applied onto the nylon membrane using a Minifold I filtration manifold (Schleicher & Schuell, Keene, NH). Standards were diluted in stDNA for a total of 0.5 ␮g/well. Genomic DNA content was considered to be 7.1 ␮g/106 cells, as previously estimated for human cells (Ausubel et al., 1990). PEI did not interfere with the blotting process as PEI-DNA complexes or free DNA standards gave comparable results (data not shown). SEAP and GFP probes were synthesized using the DIG DNA labelling kit from Roche following the manufacturer’s protocol. For both Southern and Northern analyses, the membranes were blocked and probed in DIG easy HybTM hybridization buffer and revealed using the DIG luminescent detection kit for nucleic acids (Roche). The light emission was captured using a Kodak Digital Science Image Station 440 cf and immunoreactive bands were quantified using the Kodak

271

Digital Science 1D image analysis software version 3.0 (Eastman Kodak Company). 3. Results 3.1. Plasmid uptake, nuclear translocation and protein expression We first directed our efforts in determining the fate and distribution of the plasmid DNA after its addition to the cells. A plasmid uptake kinetic study was performed by harvesting pTT/GFP transfected cells at different time points and analyzing their total plasmid content by Southern blotting (Fig. 1). Plasmid DNA was found associated to the cells after only 15 min post-transfection and maximal association, on a per cell basis, occurred around 26 hpt, before steadily decreas-

Fig. 1. Plasmid uptake kinetic. Cells were transfected using the pTT/GFP plasmid and harvested at different time points. Total DNA was extracted and subjected to dot-blot analysis. Known amounts of pTT/GFP, diluted in stDNA, were used as standard. Top panel: dot-blot result using 0.125 or 0.5 ␮g of total DNA extract; bottom: densitometric analysis of upper panel. The results are representative of two independent experiments: squares, per 106 cells; circles, per ml of culture.

272

E. Carpentier et al. / Journal of Biotechnology 128 (2007) 268–280

Fig. 2. PEI/DNA distribution analysis using YOYO-1 labelled plasmid. Cells were transfected with YOYO-1 labelled pTT/BFP plasmid and pictures were taken at different time points using a fluorescent microscope with blue (BFP) and green (plasmid DNA) filters. The fluorescence images are superimposed to the visible picture.

ing thereafter. Considering the total DNA content of a human cell to be 7.1 pg (Ausubel et al., 1990) and the amount of DNA loaded on the membrane, the maximal plasmid uptake was estimated to be around 6.5 × 104 copies/cell. Interestingly, while specific plasmid uptake reached a maximum at 26 hpt (6.5 × 104 plasmids/cell), volumetric capture, i.e. considering cell density (data not shown) was maximal at 48 hpt (480 ng per ml of culture), corresponding to 48% of the initial amount added. We then looked at the distribution of the transfected DNA by using the pTT/BFP plasmid labelled with the green fluorescent dye YOYO-1 for transfection. Pictures were taken at various time points to visualize the fluorescently labelled DNA and simultaneously detect expression of the blue fluorescent protein. Results showed that after 2 hpt, most of the cells (if not all) had captured the polyplexes, although only 25% of the cells were shown later to express the transgene (Fig. 2). The presence of labelled plasmids associated with cells was still detectable after 96 hpt. Interestingly, we observed no qualitative difference in the amount of associated fluorescent polyplexes between BFP-positives and -negatives cells up to 96 hpt. To better quantify the amount of plasmid DNA taken up by expressing and non-expressing cells, transfections were carried out with the pTT/hCD4-GFP plas-

mid and cells were magnetically sorted in CD4-GFP positive and negative populations 48 h after transfection. The CD4-GFP fusion allows for both sorting and visualization of the cells expressing the transgene. As seen by fluorescent microscopy and flow cytometry analyses (Fig. 3A, upper and lower panels, respectively), the CD4-GFP positive population was enriched ∼1.5 times (from 52% to 73%), while a nearly completely CD4-GFP negative population (0.6% GFP-positive cells) was obtained. Southern analyses of these populations showed comparable amounts of plasmid (about 2 × 104 copies per cell for this experiment) in the total extract of the three preparations (Fig. 3B), which corroborate the observation made with the YOYO-1 labelled pTT/BFP experiment. When the nuclear fraction of these populations was analysed, the plasmid content was estimated at 1500, 550 and 1850 copies/cell for total, CD4-GFP negative and CD4GFP enriched populations, respectively. This showed that a 42% enrichment in the number of CD4-GFP positive cells resulted in only ∼23% increase in nuclear plasmid copies. The amounts of plasmid found in the total fraction of unsorted cells varied from 1 × 104 to 4.5 × 104 /cell between experiments (n = 3) and the plasmid copy number found in the nuclear fractions was systematically 5–10% of that measured in the total extracts (Table 1). In this experiment, considering the

E. Carpentier et al. / Journal of Biotechnology 128 (2007) 268–280

273

Fig. 3. Plasmid distribution in expressing and non-expressing cells. The cells were transfected using the pTT/hCD4-GFP plasmid and sorted using the MACSelectTM system 48 hpt. (A) Fluorescence microscopy observation of the sorted populations (top panel) and flow cytometry analysis of the same population (bottom panel). (B) Southern analysis of total or nuclear DNA extracts of the unsorted and sorted populations. About 0.5 or 5 ␮g of total or nuclear extracts were loaded, respectively. Known amounts of pTT/GFP plasmid diluted in stDNA were used as standard. U, unsorted; N, negative; P, positive. The results are representative of three independent experiments.

initial plasmid load of 2.7 × 105 copies/cell, and the cell density at 48 hpt (1.3 × 106 /ml; data not shown), about 20% of the added DNA was found to be associated with cells while only 1–2% was found in the nuclear fraction. Interestingly, a significant amount of plasmid was always found in the nuclear fraction of the CD4-GFP negative population. Table 1 Total and nuclear plasmid copies found per cell at 48 hpt in three independent experiments Experiment #

Total plasmid

Nuclear plasmid

Nuclear/total (%)

1 2 3

20,000 46,300 11,600

1500 2400 980

7.5 5.2 8.4

3.2. Protein expression versus plasmid amounts To identify and better understand the cellular limitations to protein overexpression, we evaluated the impact of gene dosage on protein production. Our preliminary results showed that the best approach to reliably deliver different amounts of plasmid per cell was to use stuffer DNA to replace part of the coding plasmid as compared to using different amounts of PEI:DNA complexes (data not shown). To level the differences in the final yields obtained between independent experiments, mainly caused by variations in transfection efficiencies, the results were normalized as relative values to the optimal transfection condition (being 100% of coding plasmid) obtained in each individual experiment. When performing titration

274

E. Carpentier et al. / Journal of Biotechnology 128 (2007) 268–280

ing that limitations occur somewhere in the processes leading to protein synthesis. Flow cytometry analyses performed 48 hpt showed that the percentage of GFP-positive cells was relatively constant regardless of the amount of pTT/GFP used, although a slight drop (∼10%) was observed when using less than 50% of pTT/GFP plasmid in the mixture (Fig. 4B). These results suggest that while increasing the percentage of coding plasmid in the transfection mixture leads to an increase in both the amount of plasmids delivered per cell and the quantity of protein produced, only a sub-population of the cells do express the transgene, independently of the amount of coding plasmid used for transfection. 3.3. Protein expression versus mRNA accumulation

Fig. 4. Protein expression following plasmid titration. (A) SEAP (squares) and GFP (circles) expression levels following transfection with increasing amounts of coding plasmid. The different plasmid DNA amounts were added to the cells (from 2% to 100% or 20 ng/ml to 1 ␮g/ml) by diluting the coding plasmids in non coding DNA (stDNA) to keep DNA and PEI concentrations constant (1 and 2 ␮g/ml, respectively). The results are expressed as relative values of the protein level obtained with the standard transfection condition (100% coding plasmid). (B) Transfection efficiency (% of GFP expressing cells) of the samples described above. Results were normalized against values obtained using 100% of coding plasmid. Analyzes were performed 48 hpt (GFP) or 72 hpt (SEAP). Error bars represent the standard deviation of three independent experiments.

experiments with the pTT/SEAP and pTT/GFP vectors, protein expression was not found to be linearly correlated with the amount of plasmid added to the cells (Fig. 4A). Indeed, about 70% of maximal SEAP and 50% GFP expression were obtained with only 10% of coding vector. Beyond 10%, the accumulation rate of both reporter proteins drastically decreased, suggest-

In an attempt to better characterize the apparent saturation observed in Fig. 4A, we looked at the SEAP mRNA accumulation levels. The transfected cells were harvested 72 hpt, followed by total RNA isolation and Northern blot analysis (Fig. 5A). A sharp increase in SEAP mRNA level was observed when using between 2% and 10% of coding plasmid, followed by a lower but steady accumulation beyond 10% without reaching a plateau. Interestingly, about 50% of the maximal mRNA level was obtained when using only 10% of plasmid DNA (Fig. 5B). These data suggest that the reduced protein accumulation rate observed beyond 10% of coding plasmid could results in part by a reduction in transcription efficiency (mRNA produced per plasmid DNA). Moreover, the relationship between SEAP expression and SEAP mRNA levels (Fig. 5C) also suggests a translational/post-translational limitation at higher plasmid copy numbers. 3.4. Peptone feeding and protein/mRNA accumulation Peptone feeding has recently been reported to significantly increase the expression of r-protein upon PEI-mediated transient gene expression (Pham et al., 2003, 2005). The effect of adding such peptones on mRNA expression level using varying amounts of coding plasmid DNA was thus evaluated. Our results confirm that r-protein yield is increased by peptone feeding and interestingly, the mRNA level was also found

E. Carpentier et al. / Journal of Biotechnology 128 (2007) 268–280

275

Fig. 6. Effect of peptones on SEAP activity and mRNA using increasing amounts of pTT/SEAP plasmid (from 1% to 100%) while keeping total DNA constant by compensating with stDNA. Feeding was carried out 24 hpt and the analyses were performed 72 hpt. Panel A: mRNA signal (top panel) and the overall RNA load (bottom panel) for each samples. Twenty ␮g of total RNA were loaded per sample. Panel B: SEAP activities (non-fed, open triangles; fed, close triangles) and mRNA densitometric signal intensity values (non-fed, open squares; fed, close squares) are expressed relative to that obtained when using 100% pTT/SEAP plasmid.

3.5. Polyplex amounts and cell growth

Fig. 5. SEAP mRNA and protein expression levels upon plasmid titration. The cells were transfected with increasing amounts of pTT/SEAP plasmid (from 1% to 100%) while keeping DNA and PEI concentrations constant by adding stDNA. (A) Northern blotting result (top panel) and total RNA load (bottom panel). Twenty ␮g of total RNA was loaded for each sample. (B) Densitometric analysis (squares) of the Northern blot experiments and corresponding SEAP activity values (circles) are taken from three independent experiments. (C) SEAP activity plotted against mRNA level. Both SEAP activities and mRNA levels are expressed as relative to value obtained using 100% coding vector to allow inter-experiments comparison.

to increase for all pTT/SEAP amounts used (Fig. 6A and B). With 100% of coding plasmid in the transfection mixture, the net increase in the protein production yield at 72 hpt was about 40–50% while mRNA content increased by about 20–25%.

Since no apparent plateau could be reached for mRNA synthesis when increasing the amount of coding plasmid in the transfection mixture (see Fig. 5A and B), we looked at the effect of adding more polyplexes (using only coding plasmid) on SEAP mRNA contents, SEAP protein levels and cell densities. Cells were transfected with 1.5- and 2-fold the standard amount of DNA:PEI (1:2) complexes (Fig. 7). The results show that cell-specific SEAP mRNA level was increased by more than 50% using 1.5 or 2 times the standard polyplex amount, although SEAP production was reduced by ∼20% and ∼40%, respectively. This drop in protein expression was associated with a proportional ∼25% and ∼35% reduction in cell densities, respectively. This drop in cell density was not associated with decreased cell viability as determined by trypan blue exclusion assay (data not shown), suggesting a cytostatic effect caused by the higher concentration of polyplexes. We performed the same experiment using only stDNA in the transfection mixture and observed the same growth inhibition effect (data not shown), suggesting that the

276

E. Carpentier et al. / Journal of Biotechnology 128 (2007) 268–280

Fig. 7. Effect of polyplex amount on viable cell count, mRNA level and protein production. Cells were transfected with 100%, 150% or 200% polyplex amounts using the pTT/SEAP plasmid. Cell number (squares), SEAP activity (circles) and mRNA level (triangles) were measured 72 hpt. Values of all parameters were normalized against the value obtained with cells transfected using 100% polyplex.

polyplex load per se, and not the plasmid load, is responsible for the reduced cell growth and SEAP production.

4. Discussion When using large-scale transfection of mammalian cells for the production of r-proteins, the final yield will not only depend on transfection efficiency, but also on the use of an appropriate gene expression cassette, the number of transcriptionally competent plasmid copies delivered to the nucleus and the use of an optimized nutrient source, the latter promoting cell growth and bringing sufficient precursors and energy sources for mRNA and protein synthesis. This study was undertaken to determine potential bottlenecks that could limit r-protein production following PEI-mediated gene transfer in 293E cells. More specifically, we looked at plasmid DNA uptake, mRNA synthesis and protein production following transfection with increasing levels of pTT plasmid encoding SEAP or GFP reporter proteins. We first showed that from the total amount of plasmid DNA added for transfection, almost half could be recovered with the cellular fraction at 48 hpt and a maximum of 6.5 × 104 plasmids/cell was found at 26 h post-transfection. The remaining

plasmid DNA could have been lost by binding to the culture vessel, degraded or maintained in solution in the culture medium. Kichler et al. (2001) also conducted similar experiments using PEI-mediated transfection and reported about 5 × 104 copies of plasmid DNA per cell after only 7 hpt. Since plasmid DNA is protected against nucleases once condensed by PEI (Godbey et al., 2000), the cells were not treated with DNAse prior to DNA extraction. Although the cells were extensively washed, it is still possible that part of the quantified DNA was not internalized and remained at the cell surface. However, confocal microscopy observation of cells transfected with a fluorescent-labelled plasmid showed that most if not all of the plasmid DNA was found inside the cells 24 hpt (Paris et al., in preparation). In addition, the cell-associated fluorescent polyplexes were found equally distributed throughout the cellular population, although only a fraction of the cells was shown later to express the transgene, an observation already been made by others (Escriou et al., 1998a; Escriou et al., 1998b; Fasbender et al., 1997; James and Giorgio, 2000; Remy-Kristensen et al., 2001; Tseng et al., 1997). To help determine if cellular and/or nuclear uptake of plasmid DNA could be limiting for the successful expression of the transgene, we magnetically sorted cells transfected with the pTT/hCD4-GFP plasmid into negative and positive sub-populations. Interestingly, Southern analyses of the sorted cells revealed similar amounts of plasmid DNA associated with the total extracts of these populations, supporting the observations made with the fluorescent-labelled plasmid. On average, 1 × 104 to 4.5 × 104 copies of plasmid DNA were found per cell after 48 hpt. Wiethoff and Middaugh (2003) reported, using cationic lipids as transfection vehicle, that 1 × 106 plasmid copies/cell were initially needed for optimal transfection, and that 1 × 102 to 1 × 104 copies were required in the nucleus for detectable transgene expression. When analyzing the nuclear fractions of the sorted populations, a three-fold higher plasmid copy number was found in CD4-GFP positive cells compared to the negative population (1850 copies/cells versus 550 copies/cells). Again, although the isolated nuclei were extensively washed, the exact location of the nuclear-associated plasmid cannot be ascertained and does not preclude that a fraction could be bound to the cytoplasmic face of the nuclear envelope. Even if a relatively high variability in absolute numbers was obtained between

E. Carpentier et al. / Journal of Biotechnology 128 (2007) 268–280

experiments, nuclear plasmid levels were always found to be 10–20 times lower than those found in total extracts, suggesting that nuclear translocation is a nonefficient process. Indeed, Pollard et al. (1998) showed that only 1% of plasmids translocated to the nucleus after cytoplasmic microinjection of PEI-DNA complexes. Godbey et al. (1999) reported that when using double-labelled PEI/DNA complexes, most of the DNA that reached the nucleus was still found associated with PEI. This may represent an additional bottleneck to expression as plasmid DNA has to dissociate from PEI in order to be transcriptionally competent. To that respect, some studies suggest that only a fraction of the nuclear PEI-delivered plasmid actually undergo transcription (Honore et al., 2005; Schaffer et al., 2000; Tachibana et al., 2004). Transient gene expression is thus dependent on both the efficiency of DNA delivery to the nucleus and its expression (Luo and Saltzman, 2000). The relatively high amount of nuclear plasmid found in the non-expressing population thus supports the concept of transcriptional competency of plasmid DNA. We performed DNA titration experiments and found that the same amount of expressing cells could be obtained over a 50-fold range of pTT/GFP vector in the transfection mixture. However, the expression levels did not correlate linearly to plasmid DNA amounts as SEAP and GFP expression reached ∼70% and 50% of maximal levels, respectively, with only 10% of pTT plasmid in the transfection mixture. Those experiments were performed by diluting the pTT plasmid in stuffer DNA to keep DNA and PEI amounts constant for all transfections. Attempts to vary the amount of polyplexes added to the cells in titration experiments resulted in very poor transfection efficiencies and protein yields at low polyplex concentrations (data not shown). This is in agreement with a recent work by Kichler et al. (2005), who suggested that a threshold amount of transfection reagents with endosomebuffering capacities, such as PEIs and lipospermines, is required to yield significant reporter gene expression. The fact that the use of a wide range of plasmid DNA quantities (from 20 ng to 1 ␮g) does not significantly impact the amount of transfected cells but rather modulate cell-specific gene expression level suggests that a successful transfection is not only driven by the amount of plasmid delivered to the cell. This led us to speculate that gene expression also depends on cer-

277

tain cellular characteristics or physiological traits that allow the polyplexes to be efficiently translocated in the nucleus and transcribed. This physiological competence state could be related to the cell cycle, as already suggested (Brunner et al., 2000, 2002; Mannisto et al., 2005; Tait et al., 2004) and should be further investigated. In addition, the fact that transfection efficiencies can vary significantly from one experiment to the other even though the same transfection parameters and culture conditions are rigorously applied at all time stress the importance of a better understanding and control of cell physiology for improving transfection outputs. The hyperbolic-like relation between gene dosage and protein expression obtained with both reporter proteins suggests saturation(s) in the cellular processes leading to protein expression. We thus looked at transcription efficiency by analyzing the SEAP mRNA content of cells transfected with increasing pTT/SEAP plasmid titers. The dose–response curve showed a ‘biphasic’ mRNA accumulation rate, having a break point at around 10% of the optimal plasmid content, as also observed with SEAP protein accumulation. This suggests that protein production could be limited in part by a transcriptional overload when using more than 10% of plasmid. Still, mRNA accumulation did not show complete saturation, as it steadily increased between 10% and 100% of pTT/SEAP plasmid. When looking at the relationship between SEAP expression and SEAP mRNA levels, the results suggest that translational/post-translational limitations also occur at higher mRNA levels. Adding more polyplexes to the cells increased cell-specific SEAP mRNA by up to 50% although protein production decreased. This again suggests that the translational/post-translational processes might saturate before transcription does when using standard plasmid copy amount. The observed reduction in SEAP production when adding more polyplexes could arise from the observed cytostatic effect (this study) or their toxicity as reported by others (Clamme et al., 2003; Lecocq et al., 2000; Thomas et al., 2005; Thomas and Klibanov, 2003), suggesting that optimal protein production requires the correct balance between delivering enough complexes to the nucleus without causing too much adverse effects on the cells. The hypothesis of translational limitation is reinforced by the fact that the addition of peptones 24 hpt, which most likely supply cells with nutrients, increased protein production by more than 50% at

278

E. Carpentier et al. / Journal of Biotechnology 128 (2007) 268–280

72 hpt. Interestingly, even if this limitation can be partially alleviated by feeding, the same relationship between gene copy number, protein and mRNA levels was obtained when transfecting with varying plasmid amounts. These observations suggest that production could probably be further increased without the need of delivering more gene copies per expressing cells. Peptone addition did not seem to prevent PEI cytostatic effect, as combining feeding with an increased amount of polyplex also resulted in a lower cell count and SEAP production, when compared to feeding cells transfected with an optimal amount of polyplexes (data not shown). Amino acids’ availability is known to modify the expression of specific target genes at the level of transcription (Fafournoux et al., 2000; Jousse et al., 2004; Lindsley and Rutter, 2004; Proud, 2002). Cells have several sensory systems that detect energy and metabolic status and adjust fluxes accordingly through modulation of metabolic pathways (Proud, 2002). Although the underlying processes are just being uncovered in mammals, amino acids by themselves can play, in concert with hormones, an important role in the control of gene expression. An example of these processes is the protein kinase TOR (target of rapamycin) that is activated by nutrients, growth factors and energy metabolism and is known to increase cell growth and proliferation via the regulation of protein synthesis (Lindsley and Rutter, 2004). The improved SEAP production upon peptone feeding could be somehow related to the inactivation of a repressor of mRNA translation mediated by TOR and/or other nutrient sensing pathways and need to be investigated. The high levels of secreted protein expression, driven by virus-derived promoters, could also have a substantial effect of the endoplasmic reticulum (ER) integrity and subsequently, on protein expression levels. Indeed, two known mechanisms are involved in protein synthesis repression upon ER challenge, namely the unfolded protein response (UPR) and the ER overload response (EOR). Those mechanisms work toward restoring ER integrity and, upon failure, they can lead to cell death (Hay and Sonenberg, 2004). It could be interesting to see if those mechanisms do play a role in the translational/post-translational limitations observed in this study. Moreover, protein expression levels are hard to predict and are greatly influenced by the nature of the gene and its product. Those analyses

will be extended to additional proteins to see if their production dynamics diverge from what was observed using SEAP. This may help determine if transfection conditions should be optimized for each target protein in order to ensure their maximal expression. In light of these results, we conclude that nuclear translocation is an inefficient process, as up to 48% of added plasmids were found associated with total cell extracts after the first day of the transfection, while only 1–2% are found later in the nuclear fraction at 48 hpt. Moreover, significant amounts of plasmid can be recovered in the nuclear fraction of cells that do not detectably express the transgene. Decreasing the quantity of plasmid used for transfection decreased cell-specific productivity without significantly affecting the number of expressing cells. We thus suggest that another barrier for gene expression, using PEImediated transfection, arise from a cellular physiological state that limit nuclear translocation of plasmid DNA and/or its transcription. Production of r-proteins by PEI-mediated gene transfer would also greatly benefit from cellular engineering, medium development and feeding strategies aiming at alleviating putative transcriptional and translational limitations. Overcoming the nuclear translocation and transcriptional barriers, by using more efficient gene transfer reagents or cell culture conditions, would most certainly increase the number of expressing cells and, at the same time, improve the overall volumetric yields. Acknowledgements We would like to thank Lucie Bourget and Eric Massicotte for their support and insightful comments on flow cytometry analyses. References Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D.D., Seidman, J.D., Struhl, K., 1990. Current Protocols in Molecular Biology. John Wiley & Sons Inc., New York, NY. Baldi, L., Muller, N., Picasso, S., Jacquet, R., Girard, P., Thanh, H.P., Derow, E., Wurm, F.M., 2005. Transient gene expression in suspension HEK-293 cells: application to large-scale protein production. Biotechnol. Prog. 21, 148–153. Bieber, T., Meissner, W., Kostin, S., Niemann, A., Elsasser, H.P., 2002. Intracellular route and transcriptional competence of polyethylenimine-DNA complexes. J. Control Release 82, 441–454.

E. Carpentier et al. / Journal of Biotechnology 128 (2007) 268–280 Boussif, O., Lezoualc’h, F., Zanta, M.A., Mergny, M.D., Scherman, D., Demeneix, B., Behr, J.P., 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., 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. Brunner, S., Sauer, T., Carotta, S., Cotten, M., Saltik, M., Wagner, E., 2000. Cell cycle dependence of gene transfer by lipoplex, polyplex and recombinant adenovirus. Gene Ther. 7, 401–407. Clamme, J.P., Krishnamoorthy, G., Mely, Y., 2003. Intracellular dynamics of the gene delivery vehicle polyethylenimine during transfection: investigation by two-photon fluorescence correlation spectroscopy. Biochim. Biophys. Acta 1617, 52–61. Derouazi, M., Girard, P., Van, T.F., Iglesias, K., Muller, N., Bertschinger, M., Wurm, F.M., 2004. Serum-free large-scale transient transfection of CHO cells. Biotechnol. Bioeng. 87, 537–545. Durocher, Y., Perret, S., Kamen, A., 2002. High-level and highthroughput recombinant protein production by transient transfection of suspension-growing human 293-EBNA1 cells. Nucl. Acids Res. 30, E9. Durocher, Y., Perret, S., Thibaudeau, E., Gaumond, M.H., Kamen, A., Stocco, R., Abramovitz, M., 2000. A reporter gene assay for high-throughput screening of G-protein-coupled receptors stably or transiently expressed in HEK293 EBNA cells grown in suspension culture. Anal. Biochem. 284, 316–326. Escriou, V., Carriere, M., Bussone, F., Wils, P., Scherman, D., 2001. Critical assessment of the nuclear import of plasmid during cationic lipid-mediated gene transfer. J. Gene Med. 3, 179–187. Escriou, V., Ciolina, C., Helbling-Leclerc, A., Wils, P., Scherman, D., 1998a. Cationic lipid-mediated gene transfer: analysis of cellular uptake and nuclear import of plasmid DNA. Cell Biol. Toxicol. 14, 95–104. Escriou, V., Ciolina, C., Lacroix, F., Byk, G., Scherman, D., Wils, P., 1998b. Cationic lipid-mediated gene transfer: effect of serum on cellular uptake and intracellular fate of lipopolyamine/DNA complexes. Biochim. Biophys. Acta 1368, 276–288. Fafournoux, P., Bruhat, A., Jousse, C., 2000. Amino acid regulation of gene expression. Biochem. J. 351, 1–12. Fasbender, A., Marshall, J., Moninger, T.O., Grunst, T., Cheng, S., Welsh, M.J., 1997. Effect of co-lipids in enhancing cationic lipid-mediated gene transfer in vitro and in vivo. Gene Ther. 4, 716–725. Girard, P., Derouazi, M., Baumgartner, G., Bourgeois, M., Jordan, M., Jacko, B., Wurm, F.M., 2002. 100-liter transient transfection. Cytotechnology 38, 15–21. Godbey, W.T., Barry, M.A., Saggau, P., Wu, K.K., Mikos, A.G., 2000. Poly(ethylenimine)-mediated transfection: a new paradigm for gene delivery. J. Biomed. Mater. Res. 51, 321–328. Godbey, W.T., Wu, K.K., Mikos, A.G., 1999. Tracking the intracellular path of poly(ethylenimine)/DNA complexes for gene delivery. Proc. Natl. Acad. Sci. USA 96, 5177–5181. Hay, N., Sonenberg, N., 2004. Upstream and downstream of mTOR. Genes Dev. 18, 1926–1945.

279

Honore, I., Grosse, S., Frison, N., Favatier, F., Monsigny, M., Fajac, I., 2005. Transcription of plasmid DNA: influence of plasmid DNA/polyethylenimine complex formation. J. Control Release 107, 537–546. James, M.B., Giorgio, T.D., 2000. Nuclear-associated plasmid, but not cell-associated plasmid, is correlated with transgene expression in cultured mammalian cells. Mol. Ther. 1, 339–346. Jordan, M., K¨ohne, C., Wurm, F.M., 1998. Calcium-phosphate mediated DNA transfer into HEK-293 cells in suspension: control of physicochemical parameters allows transfection in stirred media. Cytotechnology 26, 39–47. Jousse, C., Averous, J., Bruhat, A., Carraro, V., Mordier, S., Fafournoux, P., 2004. Amino acids as regulators of gene expression: molecular mechanisms. Biochem. Biophys. Res. Commun. 313, 447–452. Kichler, A., Leborgne, C., Coeytaux, E., Danos, O., 2001. Polyethylenimine-mediated gene delivery: a mechanistic study. J. Gene Med. 3, 135–144. Kichler, A., Leborgne, C., Danos, O., 2005. Dilution of reporter gene with stuffer DNA does not alter the transfection efficiency of polyethylenimines. J. Gene Med. 7, 1459–1467. Kopatz, I., Remy, J.S., Behr, J.P., 2004. A model for non-viral gene delivery: through syndecan adhesion molecules and powered by actin. J. Gene Med. 6, 769–776. Lecocq, M., Wattiaux-De, C.S., Laurent, N., Wattiaux, R., Jadot, M., 2000. Uptake and intracellular fate of polyethylenimine in vivo. Biochem. Biophys. Res. Commun. 278, 414–418. Lindsley, J.E., Rutter, J., 2004. Nutrient sensing and metabolic decisions. Comp. Biochem. Physiol. B: Biochem. Mol. Biol. 139, 543–559. Luo, D., Saltzman, W.M., 2000. Synthetic DNA delivery systems. Nat. Biotechnol. 18, 33–37. Mannisto, M., Ronkko, S., Matto, M., Honkakoski, P., Hyttinen, M., Pelkonen, J., Urtti, A., 2005. The role of cell cycle on polyplexmediated gene transfer into a retinal pigment epithelial cell line. J. Gene Med. 7, 466–476. Pham, P.L., Kamen, A., Durocher, Y., 2006. Large-Scale Transfection of Mammalian Cells for the Fast Production of Recombinant Protein. Mol. Biotechnol. 34, in press. Pham, P.L., Perret, S., Cass, B., Carpentier, E., St-Laurent, G., Bisson, L., Kamen, A., Durocher, Y., 2005. Transient gene expression in HEK293 cells: peptone addition posttransfection improves recombinant protein synthesis. Biotechnol. Bioeng. 90, 332– 344. Pham, P.L., Perret, S., Doan, H.C., 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, J.S., Loussouarn, G., Demolombe, S., Behr, J.P., Escande, D., 1998. Polyethylenimine but not cationic lipids promotes transgene delivery to the nucleus in mammalian cells. J. Biol. Chem. 273, 7507–7511. Proud, C.G., 2002. Regulation of mammalian translation factors by nutrients. Eur. J. Biochem. 269, 5338–5349. Remy-Kristensen, A., Clamme, J.P., Vuilleumier, C., Kuhry, J.G., Mely, Y., 2001. Role of endocytosis in the transfection of L929

280

E. Carpentier et al. / Journal of Biotechnology 128 (2007) 268–280

fibroblasts by polyethylenimine/DNA complexes. Biochim. Biophys. Acta 1514, 21–32. Schaffer, D.V., Fidelman, N.A., Dan, N., Lauffenburger, D.A., 2000. Vector unpacking as a potential barrier for receptor-mediated polyplex gene delivery. Biotechnol. Bioeng. 67, 598–606. Schlaeger, E.-J., Christensen, K., 1999. Transient gene expression in mammalian cells grown in serum-free suspension culture. Cytotechnology 30, 71–83. Schlaeger, E.-J., Kitas, E.A., Dorn, A., 2003. SEAP expression in transiently transfected mammalian cells grown in serum-free suspension culture. Cytotechnology 42, 47–55. Shi, C., Shin, Y.O., Hanson, J., Cass, B., Loewen, M.C., Durocher, Y., 2005. Purification and characterization of a recombinant Gprotein-coupled receptor, Saccharomyces cerevisiae Ste2p, transiently expressed in HEK293 EBNA1 cells. Biochemistry 44, 15705–15714. Tachibana, R., Ide, N., Shinohara, Y., Harashima, H., Hunt, C.A., Kiwada, H., 2004. An assessment of relative transcriptional availability from nonviral vectors. Int. J. Pharm. 270, 315–321. Tait, A.S., Brown, C.J., Galbraith, D.J., Hines, M.J., Hoare, M., Birch, J.R., James, D.C., 2004. Transient production of recombinant proteins by Chinese hamster ovary cells using

polyethyleneimine/DNA complexes in combination with microtubule disrupting anti-mitotic agents. Biotechnol. Bioeng. 88, 707–721. Thomas, M., Ge, Q., Lu, J.J., Chen, J., Klibanov, A.M., 2005. Crosslinked small polyethylenimines: while still nontoxic, deliver DNA efficiently to mammalian cells in vitro and in vivo. Pharm. Res. 22, 373–380. Thomas, M., Klibanov, A.M., 2003. Non-viral gene therapy: polycation-mediated DNA delivery. Appl. Microbiol. Biotechnol. 62, 27–34. Tseng, W.C., Haselton, F.R., Giorgio, T.D., 1997. Transfection by cationic liposomes using simultaneous single cell measurements of plasmid delivery and transgene expression. J. Biol. Chem. 272, 25641–25647. Wiethoff, C.M., Middaugh, C.R., 2003. Barriers to nonviral gene delivery. J. Pharm. Sci. 92, 203–217. Wurm, F., Bernard, A., 1999. Large-scale transient expression in mammalian cells for recombinant protein production. Curr. Opin. Biotechnol. 10, 156–159. Zabner, J., Fasbender, A.J., Moninger, T., Poellinger, K.A., Welsh, M.J., 1995. Cellular and molecular barriers to gene transfer by a cationic lipid. J. Biol. Chem. 270, 18997–19007.