Comparative gene transfer between cationic and thiourea lipoplexes

THE JOURNAL OF GENE MEDICINE RESEARCH ARTICLE J Gene Med 2010; 12: 45–54. Published online 24 November 2009 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/jgm.1417 Comparative gene transfer between cationic and thiourea lipoplexes Marie Breton1 Jeanne Leblond2 Johanne Seguin1 Patrick Midoux3 Daniel Scherman1 Jean Herscovici4 Chantal Pichon3 Nathalie Mignet1 * Abstract 1 Methods The MTT test was used for cytotoxicity assessment. Transfection efficiency was determined by luciferase read-out. Permeation to propidium iodide and enhanced green fluorescent protein was evaluated by flow cytometry. Kinetics of internalization and DNA release were monitored by confocal microscopy with labelled DNA. Endocytosis inhibitors were used to study the mechanisms of lipoplex internalization. Inserm, U640, CNRS, UMR8151, Unité de Pharmacologie Chimique et Génétique, Université Paris Descartes, Faculté de Pharmacie, Paris, France 2 Canada Research Chair in Drug Delivery, Faculty of Pharmacy, University of Montreal, Montreal, Quebec, Canada 3 Centre de Biophysique Moléculaire CNRS UPR 4301, Université d’Orléans, INSERM, Orléans, France 4 Inserm, U640, CNRS, UMR8151, Unité de Pharmacologie Chimique et Génétique, Ecole Nationale Supérieure de Chimie de Paris, Paris, France *Correspondence to: Nathalie Mignet, Université Paris Descartes, 4 avenue de l’observatoire, 75006, Paris, France. E-mail: nathalie.mignet@parisdescartes.fr Background We have previously developed lipopolythiourea lipids as neutral DNA condensing agents for systemic gene delivery. Optimization of the lipopolythiourea structure led to efficient transfecting agents. To further evaluate these lipids, we investigated the internalization process of the thiourea lipoplexes and their intracellular mechanism of transfection versus that of cationic lipoplexes. Results Although thiourea/DNA complexes exhibit an almost similar level of transfection compared to that of cationic complexes, the thiourea lipoplexes were shown to be six-fold less internalized. Complexes were able to permeabilize the cytoplasmic membrane to 30 kDa molecules. Finally, DNA was shown to be released in less than 10 min in the cellular cytoplasm versus 30 min for cationic lipoplexes. Conclusions Despite a weaker internalization compared to cationic lipids, the thiourea lipoplexes were able to transfect cells at a similar level as a result of its greater ability to destabilize the cytoplasmic membrane and release DNA Copyright  2009 John Wiley & Sons, Ltd. Keywords cationic lipoplexes; endocytosis; internalization; membrane binding; neutral lipoplexes; transfection Introduction Received: 25 May 2009 Revised: 5 October 2009 Accepted: 20 October 2009 Copyright  2009 John Wiley & Sons, Ltd. Synthetic systems for DNA delivery mostly involve ionic interactions. Cationic lipids and polymers have been developed to favour strong interaction with DNA phosphates anionic moieties and to form stable structures in which DNA is protected from nuclease degradation. This cationic charge also promotes the cell interaction and uptake [1]. There is conclusive evidence that both complexes made from lipids (lipoplexes) and complexes made from polymers (polyplexes) enter cells via endocytosis [2,3]. However, this endocytosis can use several distinct pathways such as clathrin-mediated endocytosis, caveolaemediated endocytosis or macropinocytosis. Many studies have shown that clathrin-mediated endocytosis is implied in the endocytosis of lipoplexes [4] 46 but the role of the other pathways remains poorly defined. Polyplexes might also use more than a single pathway such as clathrin-mediated endocytosis and macropinocytosis [5] or caveolae-mediated endocytosis [6,7]. It is of major importance to elucidate the internalization process and, in particular, to identify whether the complexes need to escape endosomal structures. In such a case, destabilizing fusogenic lipids or peptides should be introduced in the complexes to increase endosomal escape. Some years ago, we chose to develop original lipids able to compact DNA through hydrogen binding. We hypothesized that uncharged lipids would provide an easier DNA liberation into the cells compared to cationic lipids, and would limit unspecific protein interactions for in vivo delivery. We used thiourea functions because of their ability to form strong hydrogen bonds with ionic species such as phosphates [8] and introduced three of these functions on a lipid to form a lipopolythiourea [9]. Subsequently, the synthesis of numerous lipopolythioureas of various lipid lengths bearing different spacers and head moieties has allowed us to define an optimized hydrophobic/hydrophilic balance providing easy formulating lipids that are able to transfect cells efficiently [10,11]. Although these complexes are able to achieve transfection levels similar to those of cationic lipoplexes, their drastic chemical difference with cationic entities requires further investigation on their mechanism of action. In the present study, we investigated the internalization and intracellular trafficking of thiourea lipoplexes and compared them with cationic complexes. Because thioure a lipid/DNA complexes are globally uncharged at physiological pH, we suspected that their uptake would be less efficient than that of polyamine based lipoplexes. Moreover, as thioureas are suspected to interact with the phosphates of DNA by hydrogen bonds, we expected them to release DNA more efficiently than in the case of ionic-mediated complexes. To evaluate these two potential properties that would allow thiourea lipoplexes to lead to transfection, we thus compared the thiourea lipid DDSTU (DD for didecyl, S for the serine based linker, TU for its thiourea moiety) and the lipopolyamine DMAPAP (di-miristyl aminopropyl aminopropyl) based complexes (Figure 1) in terms of cell internalization, and the kinetics of DNA release, cytotoxicity and transfection. M. Breton et al. Materials and methods The cationic lipid whose name according to the nomenclature is 2-{3-[bis-(3-amino-propyl)-aino]-propylamino}-Nditetradecylcarbamoyl methyl-acetamide or RPR209120, which we termed DMAPAP, was described in the supporting information to the study by Thompson et al. [12]. DDSTU was synthesized according to previously reported procedures [10]. All chemicals were purchased from Sigma Aldrich (St Louis, MO, USA) unless otherwise stated. Dioleoylphosphatidylethanolamine (DOPE) was purchased from Avanti Polar lipids, Inc. (Alabaster, AL, USA). Cell lines B16 murine cells were grown as monolayers incubated at 37 ◦ C in a humidified atmosphere of 5% CO2 into Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Invitrogen, Cergy Pontoise, France) supplemented with L-glutamine (29.2 mg/ml), penicillin (50 units/ml), streptomycin (50 units/ml) and 10% foetal bovine serum (FBS). Human epithelial ovarian carcinoma (HeLa) cell lines (CCL2; ATCC, Rockville MD, USA) and HeLa-EGFP cells expressing constitutively the enhanced green fluorescent protein (EGFP) were routinely grown as monolayers incubated at 37 ◦ C in a humidified atmosphere of 5% CO2 . HeLa-EGFP cell clones were obtained by transfection of HeLa cells with pEGFP (pEGFPemd-cmv, 4.797 kb; Packard, Meriden, CT, USA) and selection under geneticine. The clone was isolated by cell sorting (FACSVantage; Becton Dickinson, Grenoble, France). Cells were maintained by regular passage, respectively, in DMEM (Gibco, Invitrogen, France) supplemented with 10% heat-inactivated FBS together with 100 units/ml penicillin and 50 units/ml streptomycin each (Gibco, Invitrogen). Cells were free from mycoplasma as demonstrated by bis-benzimidazole staining [13]. The Eahy 926 is a stable, easily maintained endothelial cell line derived from the fusion of human umbilical vein endothelial cells and A549 cells. Eahy 926 cells were grown as monolayers incubated at 37 ◦ C in a humidified atmosphere of 5% CO2 into DMEM (Gibco, Invitrogen, France) supplemented with L-glutamine (29.2 mg/ml), penicillin (50 units/ml), streptomycin (50 units/ml) and 10% FBS. Figure 1. Structures of the lipids used in the present study: the polythiourea DDSTU (DD for didecyl, S for the serine based linker, TU for its thiourea moiety) and the lipopolyamine DMAPAP (di-miristyl aminopropyl aminopropyl) Copyright  2009 John Wiley & Sons, Ltd. J Gene Med 2010; 12: 45–54. DOI: 10.1002/jgm 47 Gene transfer by neutral and cationic lipoplexes J774 is a murine macrophage cell line established from a tumour that arose in a female BALB/c mouse. J774 cells were grown as monolayers incubated at 37 ◦ C in a humidified atmosphere of 5% CO2 into DMEM (Gibco, Invitrogen) supplemented with L-glutamine (29.2 mg/ml), penicillin (50 units/ml), streptomycin (50 units/ml) and 10% FBS. Liposome preparation The DDSTU was suspended via an ethanolic injection protocol. DMAPAP/DOPE (1 : 1) was prepared via the same method. Lipids (and colipids if necessary) were dissolved in ethanol and were added dropwise to ten volumes of water under vigorous agitation. The mixture was stirred overnight and then evaporated under reduced pressure at room temperature to obtain a fairly concentrated solution of liposomes. As an example, DDSTU (3 mg, 4 µmol) was dissolved in 300 µl of ethanol. This solution was added dropwise into 3 ml of stirred filtered water. The mixture was stirred overnight, and then evaporated under reduced pressure at room temperature to obtain a clear suspension of DDSTU at approximately 10 mM concentration. Plasmid preparation Plasmid pVax2 was used for all experiments. pVax2 is a derivative of the commercial plasmid pVax1 (Gibco, Invitrogen, France), which was digested with the restriction enzymes HincII and BamHI to excise the promoter. The plasmid was then blunted with the Klenow fragment, dephosphorylated with alkaline phosphatase, pCMVbeta (Clontech, Palo Alto, CA, USA), and was digested with EcoR1 and BamHI to excise the cytomegalovirus (CMV) promoter. The CMV promoter was blunted with Klenow enzyme and ligated into the blunted pVax1 to give pVax2. The plasmid pXL3031 was digested with EcoRI and BamHI and then treated with the Klenow fragment to produce a blunted fragment containing the luciferase cDNA. This fragment was ligated into pVax2 after EcoRV digestion and phosphatase alkaline dephosphorylation to give the pVax21-Luc. [14] DNA/lipid complex preparation Plasmid (100 µl, 0.02 g/l in H2 O) was added dropwise with constant vortexing to various amounts of thiourea or cationic liposomes (in 100 µl of H2 O). TU/P indicates the ratio in nanomoles of thiourea function (two per lipid) versus nanomoles of DNA phosphates. buffer) after the addition of 5 µl of bromophenol blue. The gel was run at 80 V/cm. DNA was revealed with ethidium bromide and visualized under ultraviolet light. Size measurement Particle diameter was determined by dynamic light scattering on a Zeta Sizer NanoSeries Malvern (Malvern Instruments, Vénissieux, France). The concentration of the samples was approximately 0.1 mM in H2 O. Phase transition of the lipoplexes as a function of the pH A 2.5 mM stock solution of Nile Red was prepared in ethanol. Lipoplexes of DDSTU at ratio TU/P= 40 and DMAPAP at ratio N/P= 4 were prepared at pH 6.7 in a 5 mM MES/HEPES/sodium acetate buffer and 10 µM Nile Red solution. The lipoplexes were then freeze/thawed five times. The Nile Red emission maximum was determined at different pH values, using a protocol in which the pH was first lowered step by step to acidic pH (around 3), then raised to pH 6.7, and subsequently increased step by step from pH 6.7 to approximately pH 8.5. Nile Red fluorescence was measured on a Varian spectrofluorimeter (Varian Inc., Palo Alto, CA, USA) at 25 ◦ C. The excitation wavelength was set at 550 nm and the fluorescence emission was recorded from 550 nm to 700 nm at 5-nm intervals. The wavelength of the maximal emission (λmax emission) of Nile Red was calculated using a four parameters log-normal fit. Cytotoxicity Murine B16 melanoma cells were grown as described above. Exponentially growing B16 cells were plated onto 96-well plates at 5000 cells per well in 100 µl of culture medium. Twenty-four hours after plating, 100 µl of medium with the DDSTU lipid or the DMAPAP lipid was added at different concentrations to the wells (in triplicate) containing the cells and incubated for 48 h at 37 ◦ C and 5% CO2 . After 48 h, cell viability was assayed using the MTT test [15] and absorbance was read at 562 nm in a microplate reader (BioKinetics Reader, EL340; BioTek Inc., Winooski, VT, USA). Appropriate controls with DMEM only and MTT were run to substract background absorbance. The results are presented as percentage of controls cells. The concentration of the lipid that inhibited cell viability by 50% (inhibitory concentration for 50% of cells, or IC50 ) was determined using the GraphPad Prism software (GraphPad Software Inc., San Diego, CA, USA). Results are presented as the mean ± SEM of six independent experiments each run in triplicate. Gel retardation experiments Transfection method Compaction of DNA was verified by loading 15 µl of the samples (0.1 µg DNA) on an agarose gel (0.8% in a TAE The murine melanoma cell line B16 was cultured in DMEM containing 10% (v/v) fetal bovine serum Copyright  2009 John Wiley & Sons, Ltd. J Gene Med 2010; 12: 45–54. DOI: 10.1002/jgm 48 and 100 µg/ml penicillin/streptomycin (GibcoBRL, Life Technology, Merelbeke, Belgium) with 5% CO2 at 37 ◦ C. One day before transfection, cells were treated with trypsin and deposited into 24-wells plates (45 000 cells/well) and incubated 24 h at 37 ◦ C. Fifty microlitres of DDSTU/DNA or DMAPAP/DNA (corresponding to 0.5 µg of DNA) complexes were loaded on each well and the plates were incubated at 37 ◦ C for 24 h in 1 ml of DMEM + 10% FBS. Then the cells were washed twice with PBS and treated with 200 µl of a passive lysis buffer (Promega, Madison, WI, USA). After 15 min, the cells were centrifuged for 5 min at 14000 g. Ten microlitres of supernatant and 10 µl of iodoacetamide were deposed on a 96-well plate which was incubated at 37 ◦ C for 1 h. Protein quantification was performed with the Bio-Rad assay kit (Hercules, CA, USA) and reported to the bovine serum albumin taken as a reference curve. Luciferase activity was quantified using the Luciferase quantification assay kit (Promega). On 10 µl of the lysed cells, 50 µl of the luciferin substrate was injected via an injector and the absorbance was read immediately at 563 nm on a Wallac Victor 21 420 Multilabel Counter (Perkin Elmer, Waltham, MA, USA). The results are presented as the mean ± SEM of the experiments, each run in triplicate. To test the effect of endocytosis inhibitors, 1 h prior transfection, the medium was supplemented by DMEM + 10% FBS + methyl-β-cyclodextrine (MβCD) 10 mM or DMEM + 10% FBS + 5-(N-ethyl-N-isopropyl)amirolide (EIPA) 100 µM or DMEM + 10% FBS + sucrose 0.45M. The medium was then rinsed and replaced by the lipoplexes diluted in DMEM + 10% FBS. The results are presented as the mean ± SEM of the experiments each run in triplicate. Binding and internalization studies Two days before experiments, HeLa cells were seeded at 1 × 105 cells/wells in 24-well plastic culture plates and the transfection was performed with lipoplexes prepared extemporaneously. Plasmid DNA was labelled with the Label IT Cy5 nucleic acid labelling (MIRUS, Madison, WI, USA) at a 1 : 2 reagent/pDNA weight ratio. The Label IT Labelling reaction was carried out according to manufacturer’s instructions. For the binding experiments, cells were first incubated for 30 min at 4 ◦ C prior transfection. Cells were then incubated during indicated times with thiourea lipoplexes at a ratio TU/P = 40 or cationic lipoplexes at a ratio N/P = 4 made with 2 µg of Cy5-labelled plasmid at either 4 ◦ C or at 37 ◦ C. For the fluorescence measurement, cells were scraped gently from the culture wells and extensively washed with ice-cold PBS. After incubation at 37 ◦ C, cells were cooled and extensively washed with ice-cold PBS before harvested by trypsin treatment. The cell pellets were resuspended in PBS containing 15.2 mM NaF and 0.2% phenoxyethanol. The fluorescence intensity of the cell suspension was recorded with a LSR flow cytometer (Becton Dickinson, Sunnyvale, CA, USA). The fluorescence of Cy5-labelled plasmid was recorded at Copyright  2009 John Wiley & Sons, Ltd. M. Breton et al. λexc of 633 nm and 10 000 events were counted in each sample. For all data shown, the mean autofluorescence intensity (background fluorescence) of control cells has been subtracted from the mean fluorescence intensity of treated cells. Data are the mean of three separate experiments performed in duplicate. Permeabilization experiments Trypsinized EGFP-HeLa cells were washed and suspended (107 cells/ml) in PBS. Samples were composed of 50 µl of this cell suspension in 400 µl of sheath fluid buffer. DDSTU lipoplexes at ratio TU/P = 40 or DMAPAP lipoplexes at ratio N/P = 4 were added to the samples and incubated for indicated times at 37 ◦ C. Just before measurement, 10 µl of propidium iodide (PI) (5 µg/ml) was added to the sample. The cell fluorescence was immediately recorded by flow cytometry with FACSort (Becton Dickinson) in the FL-3 (λem > 650 nm) channel. Note that the percentage of PI positive cells found for control cells has been subtracted from that of treated cells. The fluorescence of EGFP was measured using 488 nm excitation wavelength (λexc ) and 520 nm of emission wavelength (λem ) in the FL-1 channel. The results were expressed in terms of the percentage of EGFP positive cells and PI positive cells. Ten thousand events were recorded from each sample. Internalization process of lipoplexes HeLa cells were incubated at 37 ◦ C with 0.5% PERhodamine (Avanti Polar Lipids, Inc.) labelled liposomes complexed with a fluorescein-labelled plasmid. The plasmid was labelled with Label IT fluorescein nucleic acid labelling (Mirus, Madison, WI, USA) at a 1 : 2 reagent/pDNA weight ratio according to the manufacturer’s instructions. The complexes of DDSTU were prepared at a TU/P = 40 ratio and the DMAPAP complexes at a N/P = 4 ratio. The intracellular fate of lipids and DNA was followed by confocal microscopy, with a Zeiss Axiovert 200 M microscope coupled to a Zeiss LSM 510 scanning device (Carl Zeiss, Oberkochen, Germany). The inverted microscope was equipped with a Plan-Apochromat × 63 objective (numerical aperture = 1.4) and with a temperature-controlled stage. To evidence presence of DNA in the nucleus, 1 µM DRAQ5, a far-red fluorescent DNA dye (Biostatut Limited, Shepshed, UK), was added to the medium to stain the cell nuclei blue when we observed green spots in the nucleus. Images were recorded with the software LSM Image Browser (Carl Zeiss) and were calculated with the public-domain software ImageJ (http://rsbweb.nih.gov/ij/). Each image was represented by 512 × 512 pixels of 0.28 × 0.28 µm2 each, and recorded with a line mode to reduce background noise (average on two scanning images). The acquisition of each image was performed with the confocal laser scanning microscope’s meta mode selecting specific domains of the emission spectrum. J Gene Med 2010; 12: 45–54. DOI: 10.1002/jgm 49 Gene transfer by neutral and cationic lipoplexes Figure 2. Physico-chemical parameters of lipoplexes. (A) Agarose gel electrophoresis of DNA complexed with DDSTU at TU/P = 40 and DMAPAP at N/P = 4 and lipoplexes size (nm). (B) Lipid phase transition determination. The maximal emission wavelength of Nile Red inserted in DDSTU or DMAPAP lipoplexes measured as a function of pH Results Physico-chemical characterization Because the characterization of DDSTU and DMAPAP lipids has been studied previously [9,12], in the present study, we report only data corresponding to lipoplexes formulations used for the cellular studies. As shown by the agarose gel electrophoresis (Figure 2A), the absence of DNA migration indicated that it was well compacted at a TU/P ratio of 40 for DDSTU and an N/P ratio of 4 for DMAPAP. Under these formulations, both lipoplexes exhibited a size of approximately 200 nm (Figure 2A). Next, the lipid phase transition of each type of lipoplexes was assessed as a function of pH by using the maximal emission wavelength of Nile Red incorporated in lipoplexes as described for SAINT complexes [16]. Figure 2B shows that a red shift of the Nile Red emission wavelength in cationic complexes occurred when the pH decreased, whereas no change was observed with thiourea complexes. These data show that the thiourea lipid did not undergo a lipidic phase transition at acidic pH as could be expected for uncharged lipids. Conversely, the red shift observed at a pH around 6 with cationic complexes indicates an environmental change of the fluorophore. This is related to a phase transition of DOPE present in cationic lipoplexes. Indeed, DOPE is known to undergo a phase transition at this pH upon the protonation of the phosphate moiety [17]. Transfection efficiency After physico-chemical characterization of lipoplexes, we assessed their transfection efficiency on a B16 murine melanoma cell line, a J774 murine macrophage cell line and two human cell lines, HeLa (epithelial ovarian carcinoma cells) and Eahy 926 (endothelial cells). Gene transfection performed on different cell lines indicated that the DDSTU complexes were able to transfect cells Copyright  2009 John Wiley & Sons, Ltd. in the presence of 10% serum. We could demonstrate a dose dependence of the transfection level on B16, Eahy 926 and HeLa cells. The luciferase level on the B16 cell line was lower than that obtained with cationic lipoplexes under the same conditions of non-aggregating complexes (N/P = 8, p < 0.05; Figure 3). The luciferase level was slightly lower for neutral lipoplexes compared to cationic lipoplexes in the case of Eahy 926 cells. However, the number of transfected cells was quite similar (approximately 50–60%). For HeLa cells and J774, comprising macrophage cells that are hard to transfect with cationic lipids [18], the transfection level mediated by thiourea and cationic lipoplexes was not significantly different (Figure 3). In the absence of serum, similar results were obtained, except for macrophages in which enhanced transfection efficiency was observed with the two types of lipoplexes (not shown). The best transfection results were globally obtained at a TU/P ratio of 20 or 40. The TU/P ratio of 40 was chosen for the subsequent experiments. The cytotoxicity of lipoplexes was assessed by a MTT colourimetric assay on B16 cells: IC50 of DDSTU and DMAPAP lipoplexes was 29 µM and 49 µM, respectively. Influence of temperature on the amount of cell-associated plasmid The kinetics of cell-associated DNA after incubation of HeLa cells with Cy5-labelled plasmid complexed by thiourea or cationic lipoplexes was evaluated at 4 ◦ C and 37 ◦ C by flow cytometry (Figure 4). The data obtained show that, at 4 ◦ C, the cellular binding of DNA complexed with thiourea liposomes was five- to six-fold lower than that of DNA complexed with cationic liposomes. At 37 ◦ C, the fluorescence intensity of cells incubated with both lipoplexes was two- to three-fold higher than that measured at 4 ◦ C. With respect to the binding at 4 ◦ C, the amount of DNA complexed with cationic liposomes J Gene Med 2010; 12: 45–54. DOI: 10.1002/jgm 50 M. Breton et al. Figure 3. Transfection efficiency of lipoplexes. Luciferase expression obtained after 24 h of transfection of B16, Eahy 926, J774 and HeLa cells with DDSTU/DNA complexes at different TU/P ratios in the presence of serum. Data are expressed as CPS/µg of protein content. The cytotoxicity of DDSTU and DMAPAP lipoplexes was assessed by a MTT colorimetric assay. Statistics were calculated with an unpaired t-test with Welch’s correction (∗ p < 0.05; no sign means that data are not statistically different) was 4.5- to six-fold higher than that of DNA complexed with thiourea liposomes. The kinetics of cell-associated DNA indicate that the interaction of thiourea lipoplexes with cells appears to be faster than that of cationic ones. This suggests that differences exist with respect to cellular association between these two types of lipoplexes. Permeabilizing activity of lipoplexes We have previously observed that the transfection efficiency, at low TU/P ratios, was significantly improved by the addition of liposomal lipopolythioureas (data not shown). Moreover, thiourea lipoplexes do not tend to Copyright  2009 John Wiley & Sons, Ltd. undergo a phase change at lower pH. We were interested in whether this lipid could destabilize the cell membrane. Thus, we first studied the permeabilizing activity of thiourea lipoplexes to small molecules such as PI. Cells were incubated with lipoplexes and, at the indicated times, PI was added to the cell suspension, and then the cell fluorescence-associated to PI was immediately recorded. As shown in Figure 5, 100% of cells were permeabilized (PI-positive cells) after 15 min of incubation with DDSTU lipoplexes whereas 40% of cells were permeabilized with DMAPAP lipoplexes. In line with the DNA binding at 4 ◦ C, the kinetics of permeabilization activity of thiourea was faster than that of cationic lipoplexes. J Gene Med 2010; 12: 45–54. DOI: 10.1002/jgm Gene transfer by neutral and cationic lipoplexes 51 Figure 4. Kinetics of cell-associated Cy5-labelled DNA complexed with DDSTU or DMAPAP liposomes (dashed line) or 37 ◦ C (full line). HeLa cells were incubated at either 4 ◦ C (dashed line) or at 37 ◦ C (full line) for the indicated times with 2 µg of Cy5-labelled pDNA complexed with DDSTU lipoplexes (TU/P = 40) or DMAPAP (N/P = 4). The mean cell fluorescence intensity (a.u., arbitrary unit) was measured by flow cytometry. Data are the mean of three separate experiments performed in duplicate Figure 5. Fluorescence intensity of PI labelling of HeLa cells and of EGFP release from EGFP-HeLa cells induced by DDSTU lipoplexes (
) and DMAPAP lipoplexes (). Values are the mean of three independent experiments performed in duplicate We further evaluated whether membrane permeabilization induced by DDSTU lipoplexes could allow the passage of a small protein such as EGFP. For this purpose, a HeLa cell clone stably transfected with a plasmid encoding EGFP (EGFP-HeLa cells) was used [19]. The incubation of cells with DDSTU lipoplexes led to a rapid decrease of the number of EGFP-positive cells (Figure 5). Fifty percent and 20% of cells were EFGPpositive after 5 and 30 min of incubation, respectively. These data are indicative of a high permeabilizing activity of these lipoplexes. By contrast, approximately 90% of cells were still EGFP positive after 30 min of incubation with cationic lipoplexes, indicating that these lipoplexes did not exhibit membrane permeabilizing activity. Overall, these results demonstrate a strong membrane destabilization activity of the DDSTU lipoplexes that permits the passage of a low molecular weight molecule such as PI and small protein of 27 kDa such as EGFP. Internalization process and real-time intracellular trafficking of lipoplexes Different internalization paths have been suggested for DNA complexes, such as actin-mediated endocytosis [20], caveolae-mediated endocytosis [6] or macropinocytosis Copyright  2009 John Wiley & Sons, Ltd. [5]. Cells could take up thiourea lipoplexes by one of these pathways. To investigate the efficient uptake route of these lipoplexes, transfection experiments were conducted in the absence and presence of different inhibitors of the endocytosis pathways (Figure 6). Methylβ-cylodextrine [2], sucrose [21] and EIPA [22] inhibit the caveolae-mediated endocytosis, clathrin-mediated endocytosis and macropinocytosis, respectively. The transfection efficiency of DDSTU complexes was reduced to approximately the same level (ten-fold) by these inhibitors. By contrast, the influence of these inhibitors on the transfection efficiency of DMAPAP complexes was not similar. Indeed, sucrose and EIPA reduced the transfection level by 100-fold, whereas the effect of methyl-βcyclodextrine was comparable to that obtained for transfection by DDSTU complexes. These results indicate that different mechanisms are involved in DDSTU complex internalization and transfection. For DMAPAP complexes, internalization and transfection use the clathrin-mediated endocytosis and macropinocytosis, whereas caveolaemediated endocytosis appears to be involved to a lesser extent. Time lapse experiments were performed to follow the internalization process of the two types of lipoplexes. Studies were conducted on live cells with rhodamine-labelled liposomes and fluorescein-labelled J Gene Med 2010; 12: 45–54. DOI: 10.1002/jgm 52 M. Breton et al. Figure 6. Influence of different endocytosis inhibitors on the transfection efficiency of lipoplexes. Luciferase expression of cells transfected with DDSTU/DNA (TU/P = 40) and DMAPAP/DNA (N/P = 4) lipoplexes in the absence (control) or presence of different endocytosis inhibitors. Statistics were calculated with an unpaired t-test with Welch’s correction (∗∗ p < 0.01, ∗∗∗ p < 0.005; no sign means data are not statistically different) plasmid. Various optical sections of cells were recorded for each time period. The images obtained correspond to mid optical sections aiming to maximize the observation of events in the cytoplasm, as well as those close to the nuclei. It could be observed that DDSTU complexes were strongly bound to the plasma membrane, as indicated by the typical staining of the cell periphery (Figure 7). Up to 10 min of incubation, most of plasmid was still complexed with DDSTU liposomes as indicated by yellow coloured spots corresponding to the merge of red-labelled liposomes and green-labelled plasmids. From 16 min of incubation, a higher number of green spots could be observed corresponding to plasmids without or with low amounts of liposomes beneath the plasma membrane and throughout the cytoplasm. After 20 min, some of these spots were found close to the nuclei and they were clearly found inside the nucleus after 30 min. Concerning cationic lipoplexes, most of plasmids was complexed with DMAPAP when bound to the plasma membrane (yellow spots). In several cells, the number of green spots was lower than that observed with DDSTU and they were not localized close to the nucleus until after 30 min of incubation. A clear presence of plasmids inside the nucleus was observed only after at least 60 min. Consistently in the flow cytometry analysis, it has to be noted that the overall number of DMAPAP lipoplexes inside each cell was higher (at least five-fold more) than that of DDSTU lipoplexes. These observations validate the different behaviour of these two types of vectors and indicate that, in the thiourea-based lipoplexes, plasmids can be more quickly dissociated. Copyright  2009 John Wiley & Sons, Ltd. Discussion Some years ago, we developed a neutral polythiourea lipid for gene delivery aiming to avoid nonspecific elimination of the complexes from the blood. We expected these compounds to interact less efficiently with cells than the polyamine lipids because of their lack of cationic charges. The optimization of the structure of these lipids [23] provided us with lipoplexes exhibiting high transfection efficiency on various cell lines in the presence of serum. In the present study, we investigated the transfection process mediated by thiourea lipoplexes and compared it with the one mediated by cationic complexes. After characterizing the complexes, we studied their kinetics of cell internalization and their membrane permeabilizing activity, as well as DNA release into the cells. It should be emphasized that, although thiourea and cationic lipoplexes gave a similar level of transfection, internalization of the thiourea complexes was measured to be six-fold less efficient than for cationic ones. Given the non-ionic nature of the thiourea lipoplexes, this low internalization could be expected. Therefore, the similar transfection efficiency means that either DDSTU complexes were internalized via a different process compared to DMAPAP complexes or that they could release DNA more efficiently. To evaluate the first point, we looked at the effect induced by the thiourea lipoplexes on the cellular membrane. Indeed, because the complexes were not subject to lipid phase change as a function of the pH, we wondered how the thiourea lipoplexes interacted with the membranes and released DNA. These lipoplexes obviously interact rapidly with the plasma membrane. We followed J Gene Med 2010; 12: 45–54. DOI: 10.1002/jgm Gene transfer by neutral and cationic lipoplexes 53 Figure 7. Real-time confocal microscopy of lipoplex uptake. HeLa cells were incubated with fluorescein (green)-labelled plasmid DNA complexed either with rhodamine (red)-labelled thiourea lipoplexes (A) or cationic lipoplexes (B). DRAQ5 was added to the medium to stain the cell nuclei (blue). Images were recorded on live cells as described in the Materials and methods. The images shown correspond to representative images of analysed cells. Zoomed images correspond to a magnification of the area delineated by the white rectangle on the images recorded after 22 and 26 min of incubation with DDSTU lipoplexes (A) and after 32 and 60 min of incubation with DMAPAP lipoplexes (B). Scale bars: (A) 12 µm; (B) 15 µm the permeabilization effect of the two types of lipoplexes. By contrast to cationic complexes, we demonstrated that thiourea lipoplexes exhibit a membrane permeabilizing activity allowing the passage of small molecules, such as PI, and also small proteins, such as EGFP. A mechanism based on detergent membrane destabilization might then be suggested to explain both membrane interaction and DNA release by thiourea lipoplexes. [24] The intracellular fate of thiourea and cationic lipoplexes was also evaluated. Using different endocytosis inhibitors, we could show that the internalization of both types of complexes involves several pathways. Inhibitors are obviously not completely specific and can induce some Copyright  2009 John Wiley & Sons, Ltd. secondary effects that can impair the interpretation of our transfection results; however, there is an apparent trend involving multiple paths for DDSTU complexes and mostly clathrin-mediated endocytosis and macropinocytosis for DMAPAP complexes. Moreover, this result is not surprising. Indeed, an extensive review of uptake pathways in nonviral gene delivery leads to the conclusion that the contribution of each endocytic path is not yet well understood and depends on the nature and characteristics of the gene vectors [25]. To evaluate the second point, we followed DNA release from the complexes at 37 ◦ C and observed that a higher amount of free plasmid could be found after 10 min J Gene Med 2010; 12: 45–54. DOI: 10.1002/jgm 54 of incubation with thiourea complexes compared to incubation with the cationic lipoplexes. Thiourea lipids have less affinity for DNA than cationic lipids, as shown by gel retardation experiments. Indeed, for the same amount of plasmid, five-fold more thiourea lipids than cationic lipids are needed to fully compact DNA. The interaction between thiourea lipid and DNA is assumed to occur via hydrogen bond interaction between the thiourea functions of the lipids and the phosphate functions of DNA [26]. This interaction is less efficient than ionic interactions because more lipid is required to interact with DNA, even though the supramolecular structure tends to stabilize the assembly, and DNA was shown to be protected from serum degradation in these structures [9]. Hence, intracellular plasmid release from thiourea lipoplexes might be facilitated by weaker interactions between the lipid and plasmid DNA compared to the ionic bonding involved in cationic lipoplexes. This could explain why thiourea complexes could transfect efficiently despite a lower internalization rate compared to cationic lipoplexes. Conclusions Finally, our data demonstrate two striking differences between thiourea and cationic lipoplexes with respect to their permeabilizing activity on the plasma membrane and the release of DNA inside the cells. Indeed, DDSTU complexes were shown to permeabilize the membrane to small proteins. Flow cytometry and real-time confocal microscopy experiments also allowed us to conclude that, despite being taken up to a lesser extent by cells, DDSTU lipoplexes delivered DNA more rapidly in the nucleus compared to cationic lipoplexes. Overall, these experiments show that the internalization pathway and subsequent intracellular trafficking are of the utmost importance for efficient gene delivery. Acknowledgements This work was conducted in Paris, France. The authors state that no potential financial or personal conflicts exist concerning this manuscript. References 1. Wasungu L, Hoekstra D. Cationic lipids, lipoplexes and intracellular delivery of genes. J Control Release 2006; 116: 255–264. 2. Zuhorn IS, Kalicharan R, Hoekstra D. Lipoplex-mediated transfection of mammalian cells occurs through the cholesteroldependent clathrin-mediated pathway of endocytosis. J Biol Chem 2002; 277: 18021–18028. 3. Hoekstra D, Rejman J, Wasungu L, et al. Gene delivery by cationic lipids: in and out of an endosome. Biochem Soc Trans 2007; 35: 68–71. 4. Rejman J, Conese M, Hoekstra D. Gene transfer by means of lipo- and polyplexes: role of clathrin and caveolae-mediated endocytosis. J Liposome Res 2006; 16: 237–247. Copyright  2009 John Wiley & Sons, Ltd. M. Breton et al. 5. Khalil IA, Kogure K, Futaki S, et al. High density of octaarginine stimulates macropinocytosis leading to efficient intracellular trafficking for gene expression. J Biol Chem 2006; 281: 3544–3551. 6. Rejman J, Bragonzi A, Conese M. Role of clathrin- and caveolaemediated endocytosis in gene transfer mediated by lipo- and polyplexes. Mol Ther 2005; 12: 468–474. 7. Buhlmann P, Nishizawa S, Xiao KP, et al. Strong hydrogen bondmediated complexation of H2 PO4 − by neutral bis-thiourea hosts. Tetrahedron 1997; 53: 1647–1654. 8. Tranchant I, Mignet N, Crozat E, et al. DNA complexing lipopolythiourea. Bioconjug Chem 2004; 15: 1342–1348. 9. Leblond J, Mignet N, Largeau C, et al. Lipopolythioureas: a new non-cationic system for gene transfer. Bioconjug Chem 2007; 18: 484–493. 10. Leblond J, Mignet N, Largeau C, et al. Lipopolythiourea transfecting agents: Lysine thiourea derivatives. Bioconjug Chem 2008; 19: 306–314. 11. Sakurai F, Inoue R, Nishino Y, et al. Effect of DNA/liposome mixing ratio on the physicochemical characteristics, cellular uptake and intracellular trafficking of plasmid DNA/cationic liposome complexes and subsequent gene expression. J Control Release 2000; 66: 255–269. 12. Thompson B, Mignet N, Hofland H, et al. Neutral postgrafted colloidal particles for gene delivery. Bioconjug Chem 2005; 16: 608–614. 13. Chen TR. Insitu detection of mycoplasma contamination in cellcultures by fluorescent hoechst-33258 stain. Exp Cell Res 1977; 104: 255–262. 14. Escriou V, Carriere M, Bussone F, et al. Critical assessment of the nuclear import of plasmid during cationic lipid-mediated gene transfer. J Gene Med 2001; 3: 179–187. 15. Scudiero DA, Shoemaker RH, Paull KD, et al. Evaluation of a soluble tetrazolium formazan assay for cell-growth and drug sensitivity in culture using human and other tumor-cell lines. Cancer Res 1988; 48: 4827–4833. 16. Wasungu L, Stuart MCA, Scarzello M, et al. Lipoplexes formed from sugar-based gemini surfactants undergo a lamellar-tomicellar phase transition at acidic pH. Evidence for a noninverted membrane-destabilizing hexagonal phase of lipoplexes. Biochim Biophys Acta Biomembranes 2006; 1758: 1677–1684. 17. Koltover I, Salditt T, Radler JO, et al. An inverted hexagonal phase of cationic liposome-DNA complexes related to DNA release and delivery. Science 1998; 281: 78–81. 18. Mounkes LC, Zhong W, Cipres-Palacin G, et al. Proteoglycans mediate cationic liposome-DNA complex-based gene delivery in vitro and in vivo. J Biol Chem 1998; 273: 26164–26170. 19. Swenson ES, Price JG, Brazelton T, et al. Limitations of green fluorescent protein as a cell lineage marker. Stem Cells 2007; 25: 2593–2600. 20. Bausinger R, von Gersdorff K, Braeckmans K, et al. The transport of nanosized gene carriers unraveled by live-cell imaging. Angew Chem Int Ed 2006; 45: 1568–1572. 21. Sahay G, Batrakova EV, Kabanov AV. Different internalization pathways of polymeric micelles and unimers and their effects on vesicular transport. Bioconjug Chem 2008; 19: 2023–2029. 22. Nakase I, Niwa M, Takeuchi T, et al. Cellular uptake of arginine-rich peptides: roles for macropinocytosis and actin rearrangement. Mol Ther 2004; 10: 1011–1022. 23. Leblond J, Mignet N, Leseurre L, et al. Design, synthesis, and evaluation of enhanced DNA binding new lipopolythioureas. Bioconjug Chem 2006; 17: 1200–1208. 24. Sato H, Felix JB. Peptide–membrane interactions and mechanisms of membrane destruction by amphipathic alpha-helical antimicrobial peptides. Biochim Biophys Acta Biomembranes 2006; 1758: 1245–1256. 25. Khalil IA, Kogure K, Akita H, et al. Uptake pathways and subsequent intracellular trafficking in nonviral gene delivery. Pharmacol Rev 2006; 58: 32–45. 26. Jubian V, Dixon RP, Hamilton AD. Molecular recognition and catalysis – acceleration of phosphodiester cleavage by a simple hydrogen-bonding receptor. J Am Chem Soc 1992; 114: 1120–1121. J Gene Med 2010; 12: 45–54. DOI: 10.1002/jgm