¹AI Virtanen Institute, University of Kuopio, Kuopio, Finland

²Department of Pharmaceutics, University of Kuopio, Kuopio, Finland

³School of Pharmacy, University of California, San Francisco, California, USA

⁴Department of Medicine, University of Kuopio, Kuopio, Finland

^aCorrespondence: S Ylä-Herttuala, AI Virtanen Institute, University of Kuopio, PO Box 1627, FIN-70211 Kuopio, Finland

Different lipids and cationic polymers were tested in vitro for their ability to transfect rabbit aortic smooth muscle cells and human endothelial cells with lacZ marker gene. Toxicity of the complexes was evaluated with MTT assay. Selected plasmid-polymer complexes with different charge ratios were then tested for in vivo gene transfer efficiency using adventitial gene transfer by placing a silastic gene delivery reservoir (collar) around the carotid artery. Transfection efficiency was determined by X-gal staining 3 days after the gene transfer. Based on in vitro experiments, fractured polyamidoamine dendrimers and polyethylenimines (PEI) were selected for in vivo experiments. Fractured dendrimers (generation 6, ± charge ratio of 3) had the highest in vivo gene transfer efficiency (4.4% ± 1.7). PEI with mol- ecular size of 25 kDa (± charge ratio 4) was also effective (2.8% ± 1.8) in this model. PEI of 800 kDa showed a constant but modest gene transfer efficiency (1.8% ± 0.1) with all charge ratios. A low level gene transfer was also detected with naked DNA (0.5% ± 0.3). No signs of inflammation were seen in any of the study groups. We show here that in vitro cell culture experiments can be used to identify efficient in vivo gene transfer methods for arterial gene therapy, but the charge ratios for each complex must be optimized in vivo. It is concluded that fractured dendrimer and PEI are efficient gene delivery vehicles and can be used for arterial gene therapy via adventitial gene delivery route.

fractured dendrimers; polyethylenimine; nonviral gene transfer; collar model; carotid artery

Introduction

A number of different techniques have been used to deliver genes to vascular cells.^{1,2,3,4,5,6,7,8,9,10,11} To date, viral systems have been more efficient than nonviral vehicles in vascular gene transfer.^10,12,13 However, the use of viral vectors in gene therapy has raised questions concerning their immunogenicity, oncogenic properties and unknown long-term effects.¹⁴ Such safety concerns have not been linked to nonviral delivery of plasmid DNA. Furthermore, plasmids are easier to prepare and purify in large quantities than viral vectors. Efficient arterial gene transfer would be a beneficial tool for the treatment of vascular diseases, such as postangioplasty restenosis, postbypass atherosclerosis, peripheral atherosclerotic disease, stenosis of vascular prosthesis anastomoses and thrombus formation.^8,13 Thus, there is a clear need to develop efficient nonviral gene transfer techniques for cardiovascular applications in vivo. In most conditions only a temporary expression of the transfected gene will be required to achieve a beneficial biological effect.^1,7,8

Cationic polymers have been widely used in gene transfer experiments in vitro and in vivo.^{15,16,17,18,19,20,21} The purpose of the present study was to investigate the gene transfer efficiency of different liposomes and cationic polymers in vitro to smooth muscle cells (SMC) and endothelial cells (ECV 304) and to evaluate the most promising vehicles for in vivo gene delivery to rabbit carotid artery. Based on in vitro studies, fractured polyamidoamine dendrimers (generation 6) and PEI of 25 kDa and 800 kDa were chosen for in vivo study. Since behavior of cationic polymer-DNA complexes in vitro is dependent on the charge ratio,¹⁸ different ± charge ratios were tested in an in vivo model. To date, very little is known about the optimal charge ratios for various cationic polymer-plasmid DNA complexes for arterial gene transfer.

Our results show that both fractured dendrimers and PEI-25 mediate efficient arterial transfer of -galactosidase plasmid at ± charge ratios of 3 and 4, respectively. Whereas PEI-25 showed toxic effects with high ± charge ratios in vitro and possibly also in vivo, the fractured dendrimers did not show any signs of toxicity at ± charge ratio of 3.

Results

In vitro transfection efficiency

Gene transfer efficiencies of different lipid formulations and cationic polymers in vitro are summarized in Table 1. Fractured dendrimers had the highest transfection efficiency of 9.1% and 3.6% in SMC and ECV 304 cells, respectively. Transfection was most efficient with the highest ± charge ratio 6. PEI-25 was also found to be effective in both cells types. The most effective lipid was DOGS which transfected 3.5% (± charge ratio 4) and 2.7% (± charge ratio 8) of SMC and 0.3% of ECV 304 cells (± charge ratio 8).

In vitro toxicity

The survival of SMC varied from 64.0% to 96.9% (Table 1). No obvious correlation was seen between lipid concentration and cell survival. Cells transfected with fractured dendrimer complexes had a high survival rate of 95.5% at ± charge ratio 1.5 which decreased to 90.5% when ± charge ratio was raised to 6. The effect of charge ratio on toxicity was more evident with PEI-25 where the survival rate decreased from 85.1% (± charge ratio 4.5) to 69.6% (± charge ratio 9). PEI-800 also showed toxicity at ± charge ratio of 9 with a survival rate of 64%.

ECV 304 cells tended to have a lower survival rate than SMC. ECV 304 cells transfected with lipid-plasmid complexes had typically a higher survival rate than cells transfected with cationic polymer-plasmid complexes (Table 1).

In vivo transfection efficiency

Results of in vivo gene transfer to rabbit carotid arteries are shown in Figure 1. The highest transfection efficiency as measured by -galactosidase activity was achieved with fractured dendrimer at ± charge ratio of 3 where 4.4 ± 1.7% of the cells in the adventitia and media showed -galactosidase activity. Higher or lower ± charge ratios did not improve the transfection efficiency (Figure 1). PEI-25 had an optimal ± charge ratio of 4 with a transfection efficiency of 2.8 ± 1.8%. PEI-800 showed a constant 1-2% gene transfer efficiency with all charge ratios studied. Naked DNA was capable of transfecting 0.5 ± 0.3% of the cells in the arterial wall (Figure 1). In addition, fractured dendrimers and PEI at ± charge ratio 1 and fractured dendrimers at ± charge ratio 0.5 were tested. Transfection efficiency with these complexes was at the same level as complexes with ± charge ratios 1.5-2 (data not shown).

Representative histology of the arteries after the gene transfer is shown in Figure 2. After extravascular gene delivery most of the -galactosidase activity was seen in the adventitial layer and in medial SMC. No signs of inflammation were seen in any of the study groups. Macrophage-specific immunostaining was similar in all study groups, with PEI-25 being the only exception at ± charge ratio 8, where the percentage of macrophages of all arterial cells exceeded 10% (Figure 1b). The arterial structure and endothelium remained intact throughout the experiments in all study groups (Figure 2).

Discussion

We have previously shown that adventitial transfer of VEGF plasmid-liposome into rabbit carotid arteries reduces intimal thickening and that adventitial gene transfer can have effects on the endothelium.¹³ Adventitial delivery of plasmids and oligonucleotides has also been reported in a number of other applications.^8,10,22 In human atherosclerotic arteries, 5% transfection efficiency via the intravascular route has been recently achieved with adenoviruses.²³ We show here that the transfection efficiency with fractured dendrimers and PEI is clearly higher than that achieved previously with DOTMA/ DOPE liposomes.¹⁰ Thus, it is conceivable that gene transfer with cationic polymer-plasmid complexes will become competitive with adenoviruses for in vivo gene delivery applications. Relatively high gene transfer efficiency observed with naked DNA is likely due to the effective in vivo gene transfer method utilized in the present study where an arterial collar serves as a physical gene transfer reservoir for the gene delivery solution.

Both fractured dendrimers and PEI form small toroidal complexes with DNA and exhibit considerable buffer and swelling capacity at low endosomal pH.^16,18,20 These factors have been suggested as crucial for the high in vitro transfection efficiency of fractured dendrimers and PEI as compared to polylysines.^16,18 We found that in vitro cell culture experiments can be used to identify efficient in vivo gene transfer vehicles (fractured dendrimer and PEI-25), but the transfection efficiencies in vitro and in vivo at different ± charge ratios did not correlate with each other. Thus, charge ratios for each complex must be optimized in vivo. The size of the complexes varies with ± charge ratio¹⁸ which may affect the interaction of the cationic polymer-plasmid complexes with anionic matrix compounds in the arterial wall. Intact dendrimers have also been used for cardiac gene transfer.²⁴ In vivo toxicity and immunogenicity of intact dendrimers are likely to be minimal.²⁵ In our experiments fractured dendrimers did not show any signs of inflammation or toxicity (Figure 2).

Systemic administration of plasmid carriers leads to interactions with plasma proteins and accumulation of genes into undesired locations, mostly in the lung and liver.^8,26,27 This leads to poor delivery to the desired target tissues. Local gene transfer with a small amount of plasmid DNA should reduce systemic spread of plasmids. Intravascular gene delivery through atherosclerotic lesions and lipid-rich atheroma may also be low.^6,8,23 Furthermore, human lesions are frequently rich in connective tissue and contain only a limited number of transfectable cells in the intimal part of the lesion. To circumvent these problems adventitial gene delivery can be performed with a silastic or biodegradable collar,¹³ biodegradable gel,⁵ or direct injection into adventitia.²² It is concluded that local gene transfer with cationic polymer-plasmid complexes provides an efficient way for the treatment of arterial diseases during vascular surgery, such as prosthesis and anastomosis operations and by-pass surgery.

Materials and methods

Reagents

N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium methylsulfate (DOTAP) and dioleoylphosphatidylethanolamine (DOPE) were supplied by Avanti Polar-lipids (Alabaster, AL, USA) Dioctadecylamido-glycylspermine (DOGS) and dipalmitoylphosphatidylethanolamyl-spermine (DPPES) were kindly supplied by Dr Jean-Serge Remy, Louis Pasteur University, Strasbourg, France. Lipofectin, a 1:1 (wt/wt) liposome formulation of the cationic lipid N-[1-(2,3,-dioleyloxy)propyl]-N,N,N-trimethylammonium chloride (DOTMA) and DOPE was purchased from Gibco/BRL (Gaithersburg, MD, USA). Fractured PAMAM dendrimers (generation 6) were synthesized as described previously.^15,16 PEI²⁰ of molecular weights 25 kDa, PEI-25, (Aldrich, Milwaukee, WI, USA), 50 kDa (Sigma, St Louis, MO, USA), 750 kDa (Aldrich), and 800 kDa, PEI-800, (Fluka, Buchs, Switzerland) were used. Poly-l-lysine MW 260 kDa, cell culture media, fetal calf serum, penicillin-streptomycin and MTT solution were supplied by Sigma. X-gal was obtained from MBI Fermentas (Vilnius, Lithuania) and OCT compound from Miles Scientific (Naperville, IL, USA). N-[2-Hydroxyethyl]piperazine-N'-[2-ethanesulfonic acid] (HEPES) and 2-[N-Morpholino]ethanesulfonic acid (MES) were supplied by Sigma.

In vitro transfection efficiency assay

Rabbit aortic RAASMC and human ECV 304 cells were plated at a density of 50 000 and 100 000 cells per plate (6 cm, Sarstedt, Leicester, USA), respectively. Cells were incubated for 24 h before transfection. -Galactosidase (lacZ) expression plasmid¹⁰ (2 g per plate) was complexed with lipids or cationic polymers to give ± charge ratios of 1-9 (see below). The complexes (total volume 170 l, 50 mm MES, 50 nm Hepes, 75 mm NaCl, pH 7.2) were allowed to stand at room temperature for 20 min, and added dropwise on to the cells. The cells were incubated for 5 h in serum-free medium (DME-H21) and washed twice with PBS before addition of the growth medium consisting of DME-H21 with 10% fetal calf serum and antibiotics (100 units/ml of penicillin and 100 g/ml of streptomycin). After 2 days cells were fixed with 3.7% formaldehyde for 15 min and washed twice with PBS. X-gal staining solution (1 mg/ml, 2 mm MgCl₂, 5 mm K₃Fe(CN)₆, 5 mm K₄Fe(CN)₆, 1 ´ PBS)) was added on to the cells and incubated for 1 h. Blue X-gal-positive cells were counted and the transfection efficiency was expressed as the percentage of positive cells of the total number of cells.

In vitro toxicity assay

SMC or ECV 304 cells were plated in 96-well plates at density of 20 000 cells per well in 100 l of growth medium consisting of DME-H21 with 10% fetal calf serum and antibiotics (100 units/ml of penicillin and 100 g/ml of streptomycin). Cells were incubated for 24 h at 37°C. Growth medium was aspirated and 150 l serum-free medium added. DNA-carrier complexes were prepared by adding a solution of DNA in MES-Hepes to an equal volume of carrier in water in 96-well trays. Complex (50 l) containing 0.6 g of DNA and varying proportions of carrier were added to the serum-free medium. Cells were incubated at 37°C for 5 h before the complex and serum-free media were aspirated and 150 l of growth medium was added. Cells were incubated for 2 days after which the growth medium was replaced with serum-free medium (150 l). MTT-solution (3-(4,5-dimethylthiazolyl-2)-2,5-diphenyl tetrazolium bromide, 10 l, 5 mg/ml) was then added and cells were incubated for 2 h. SDS/DMF buffer (200 g/ml SDS, 50% DMF, pH 4.7) was added and incubated for 19 h. Absorbance was measured at 570 nm.¹⁶ Survival percentage was calculated as compared to mock-treated cells (100% survival).

Preparation of plasmid-carrier complexes for in vivo transfection

For in vivo studies pSV--galactosidase expression plasmid¹⁰ was produced using Qiagen Giga kit (Qiagen, Hilden, Germany). Purified plasmid was analyzed for the absence of microbial contaminants and lipopolysaccharide (Limulus assay; Sigma). In each transfection 25 g of plasmid DNA was used. Sixth generation fractured PAMAM dendrimers¹⁶ and PEI of molecular weights 25 kDa and 800 kDa²⁰ were used. Apparent ± charge ratio of the cationic polymer to plasmid was calculated on the following basis: 1 g of DNA contains 3.1 nmol phosphate anionic charges (assuming a mean molecular weight of 325 for a nucleotide sodium salt); 1 l of PEI (0.45 g/l) contains 10 nmol of amine cationic charges; sixth generation fractured dendrimer has 3.4 mol primary amines/mg. Cationic polymer was added to Ringer solution (volume 515-560 l, depending on the ± charge ratio) and 25 g (1 g/l) DNA was slowly added (total volume of solution 600 l). The solution was allowed to stand at room temperature for 30 min and was used for gene transfer within 2 h.

Experimental animals

Thirty-four New Zealand White male rabbits of 2.1-3.5 kg were used. Fentanyl-fluanisone (0.3 ml/kg, s.c.; Janssen Pharmaceutica, Beerse, Belgium) and midazolam (1.5 mg/kg, i.m.; Roche, Basel, Switzerland) were used for anesthesia. The left carotid artery was exposed using midline neck incision. The artery was carefully separated from the surrounding tissue and a 3 cm silastic collar (MediGene Oy, Kuopio, Finland) was positioned around the artery.¹⁰ Rabbits were re-anesthetized for gene transfer, which was performed 5 days after the installation of the collar.¹⁰ The collar was opened and filled with 600 l of the gene transfer solution. Rabbits were killed 3 days after the gene transfer and arteries were removed for histological analyses. All animal procedures were approved by the Animal Care and Use Committee, University of Kuopio, Finland.

Histological analysis

Collared arteries were divided into three equal parts: the proximal third was immersion-fixed in 4% paraformaldehyde/15% sucrose (pH 7.4) for 4 h, rinsed in 15% sucrose (pH 7.4) overnight and embedded in paraffin. The medial third was directly embedded in OCT compound and processed for frozen sections. In some rabbits part of the medial third was stored at -70°C for later analyses. The distal third was immersion-fixed in 4% paraformaldehyde/phosphate-buffered saline (pH 7.4) for 30 min, rinsed 24 h in phosphate buffer (pH 7.2) and embedded in OCT compound.²⁸

Ten randomly selected frozen sections (10 m) from each rabbit were stained with X-gal for 18 h to identify -galactosidase-positive cells. Gene transfer efficiency was calculated independently by two observers (MPT and MOH) as a percentage of X-gal-positive cells of all arterial cells.¹⁰ Mean values of the results are reported. Smooth muscle cells were identified using HHF-35 immunostaining (1:500 dilution, Enzo Diagnostics, Farmingdale, USA). Macrophages and endothelium were identified using RAM-11 (1:50 dilution; Dako, Hamburg, Germany) and CD-31 (1:50 dilution; Dako), respectively.¹⁰ Hematoxylin was used as a nuclear stain. Paraffin sections were used for immunostainings and nuclear stainings.

Acknowledgements

This study was supported by grants from the Technical Development Centre of Finland, Finnish Foundation for Cardiovascular Research, Academy of Finland and Sigrid Juselius Foundation. We thank Ms Mervi Nieminen for excellent technical help and Ms Marja Poikolainen for preparing the manuscript.

1 Nabel EG, Nabel GJ. Complex models for the study of gene function in cardiovascular biology. Annu Rev Physiol 1994; 56: 741-761, MEDLINE

2 Simons M et al. Antisense c-myc oligonucleotides inhibit arterial smooth muscle cell accumulation in vivo. Nature 1992; 359: 67-70, MEDLINE

3 Isner JM et al. Clinical evidence of angiogenesis after arterial gene transfer of phVEGF165 in a patient with ischaemic limb. Lancet 1996; 348: 370-374, Article MEDLINE

4 Gibbons GH, Dzau VJ. Molecular therapies for vascular diseases. Science 1996; 272: 689-693, MEDLINE

5 Stephan DJ et al. A new cationic liposome DNA complex enhances the efficiency of arterial gene transfer in vivo. Hum Gene Ther 1996; 7: 1803-1812, MEDLINE

6 Feldman LJ, Isner JM. Gene therapy for vulnerable plaque. J Am Coll Cardiol 1995; 26: 826-835, MEDLINE

7 Finkel T, Epstein SE. Gene therapy for vascular disease. FASEB J 1995; 9: 843-851, MEDLINE

8 Ylä-Herttuala S. Vascular gene transfer. Curr Opin Lipid 1997; 8: 72-76,

9 Ylä-Herttuala S et al. Transfer of 15-lipoxygenase gene into rabbit iliac arteries results in the appearance of oxidation-specific lipid-protein adducts characteristic of oxidized low density lipoprotein. J Clin Invest 1995; 95: 2692-2698, MEDLINE

10 Laitinen M et al. Gene transfer into the carotid artery using an adventitial collar: comparison of the effectiveness of the plasmid-liposome complexes, retroviruses, pseudotyped retroviruses, and adenoviruses. Hum Gene Ther 1997; 8: 1645-1650, MEDLINE

11 Keogh A-C et al. High efficiency reporter gene transfection of vascular tissue in vitro and in vivo using a cationic lipid-DNA complex. Gene Therapy 1997; 4: 162-171, MEDLINE

12 Barr E et al. Efficient catheter-mediated gene transfer into the heart using replication-defective adenovirus. Gene Therapy 1994; 1: 51-58, MEDLINE

13 Laitinen M et al. VEGF gene transfer reduces intimal thickening via increased production of nitric oxide in carotid arteries. Hum Gene Ther 1997; 8: 1737-1744, MEDLINE

14 Morgan RA, Anderson WF. Human gene therapy. Annu Rev Biochem 1993; 62: 191-217, MEDLINE

15 Haensler J, Szoka FC Jr. Functionalization of polyamidoamine cascade polymers and their use for efficient transfection of cells in culture. Bioconjug Chem 1993; 4: 372-379, MEDLINE

16 Tang MX, Redeman CT, Szoka FC Jr. In vitro gene delivery by degraded polyamidoamine dendrimers. Bioconjug Chem 1996; 7: 703-714, Article MEDLINE

17 Kukowska-Latallo JF et al. Efficient transfer of genetic material into mammalian cells using Starburst polyamidoamine dendrimers. Proc Natl Acad Sci USA 1996; 93: 4897-4902, MEDLINE

18 Tang MX, Szoka FC. The influence of polymer structure on the interactions of cationic polymers with DNA and morphology of the resulting complexes. Gene Therapy 1997; 4: 823-832, MEDLINE

19 Baker A et al. Polyethylenimine (PEI) is a simple, inexpensive and effective reagent for condensing and linking plasmid DNA to adenovirus for gene delivery. Gene Therapy 1997; 4: 773-782, MEDLINE

20 Boussif O et al. A versatile vector for gene and oligonucleotide transfer into cells in culture and in vivo: polyethylenimine. Proc Natl Acad Sci USA 1995; 92: 7297-7301, MEDLINE

21 Meunier-Durmort C et al. Adenovirus enhancement of polyethylenimine-mediated transfer of regulated genes in differentiated cells. Gene Therapy 1997; 4: 808-814, MEDLINE

22 Rios CD et al. Adenovirus-mediated gene transfer to normal and atherosclerotic arteries. Arterioscler Thromb Vasc Biol 1995; 15: 2241-2245, MEDLINE

23 Laitinen M et al. Adenovirus-mediated gene transfer to lower limb artery of patients with chronic critical leg ischaemia. Hum Gene Ther 1998; 9: 1481-1486, MEDLINE

24 Qui L et al. Efficient transfer of genes into murine cardiac grafts by starburst polyamidoamine dendrimers. Hum Gene Ther 1998; 9: 553-560, MEDLINE

25 Roberts JC, Bhalgat MK, Zera RT. Preliminary biological evaluation of polyamidoamine (PAMAM) Starburst dendrimers. J Biomed Mat Res 1996; 30: 53-65,

26 Plank C, Mechtler K, Szoka FC Jr, Wagner E. Activation of the complement system by synthetic DNA complexes: a potential barrier for intravenous gene delivery. Hum Gene Ther 1996; 7: 1437-1446, MEDLINE

27 Zhu N, Liggit D, Liu Y, Debs R. Systemic gene expression after intravenous DNA delivery into adult mice. Science 1993; 261: 209-211, MEDLINE

28 Ylä-Herttuala S et al. Colocalization of 15-lipoxygenase mRNA and protein with epitopes of oxidized low density lipoprotein in macrophage-rich areas of atherosclerotic lesions. Proc Natl Acad Sci USA 1990; 87: 6959-6963, MEDLINE

Figure 1 Gene transfer to rabbit carotid arteries. Transfection efficiency of fractured dendrimers (a), PEI-25 (b) and PEI-800 (c) with different ± charge ratios are presented on the left axis. Percentages of RAM-11-positive cells (macrophages) of arterial cells are presented on the right axis. Transfection efficiency was evaluated 3 days after the gene transfer. Values (mean ± s.e.) are calculated from 10 randomly selected sections evaluated by two independent observers. For comparison, naked DNA is included in each group.

Figure 2 Histology of transfected arteries 3 days after the gene transfer. (a-d) X-gal stained frozen sections from arteries transfected with (a) fractured dendrimer/pSV--galactosidase complex at ± charge ratio of 3; (b) polyethyleneimine-25 kDa/pSV--galactosidase complex at ± charge ratio of 4; (c) polyethyleneimine-800 kDa/pSV--galactosidase complex at ± charge ratio of 4; (d) naked pSV--galactosidase plasmid. 25 g of plasmid DNA was used in each transfection. Arrows indicate X-gal-positive cells; l, lumen, m, media, a, adventitia. (e-h) RAM-11 immunostaining (1:50 dilution) for macrophages; (e) fractured dendrimer/pSV--galactosidase complex at ± charge ratio of 3; (f) polyethyleneimine-25 kDa/pSV--galactosidase complex at ± charge ratio of 4; (g) polyethyleneimine-800 kDa/pSV--galactosidase complex at ± charge ratio of 4; (h) naked pSV--galactosidase plasmid. Arrows indicate RAM-11-positive cells. (i and j) HHF-35 immunostaining (1:500 dilution) from a typical collared artery. Arrow 1, endothelium; arrow 2, internal elastic lamina; arrow 3, external elastic lamina. (k) CD-31 immunostaining (1:50 dilution) for endothelium. Arrow indicates endothelial cells. (l) Hematoxylin-stained section. Magnifications: a-h, ´25; i, ´10; j, ´100; k, ´25; and l, ´10.