¹Department of Clinical Pharmacology, Groningen University Institute for Drug Exploration (GUIDE), University of Groningen, The Netherlands

²Department of Cardiology, University of Groningen, The Netherlands

³Department of Medical Microbiology, Molecular Virology Section, University of Groningen, The Netherlands

⁴Department of Therapeutic Gene Modulation, University of Groningen, The Netherlands

Correspondence to: A J M Roks, Department of Clinical Pharmacology, University of Groningen, A Deusinglaan 1, 9713 AV, Groningen, The Netherlands

Previously, we demonstrated that recombinant Semliki Forest virus (SFV) vector rapidly and selectively transfers genes into cultured vascular smooth muscle cells (VSMC), leaving endothelial cells (EC) unaffected. From this, we hypothesized that recombinant SFV in vivo only transfers genes into the media of balloon-injured but not intact vessel, that gene expression in VSMC is fast, and that the specificity of SFV for VSMC is caused by specific binding sites. To address these hypotheses, we studied the time course of in vivo SFV-LacZ and Ad-LacZ expression in balloon-injured rat aorta. In addition, the fusion characteristics of fluorescent pyrene-labeled SFV were explored in cultured VSMC and EC. In intact aorta, no LacZ expression was found in the intima or media at 24 h. In contrast, in denuded aorta, LacZ expression was detected in as early as 12 h after incubation. LacZ expression was predominantly present in the media. Ad-LacZ expression started after 12 h, but was predominantly present in the adventitia. Ad-LacZ expression in the media started after 72 h. In vitro transfection with SFV showed that fusion was higher and, moreover, saturable in VSMC as compared with EC, indicating the presence of specific SFV binding sites on VSMC, but not EC. From this we conclude that in vivo selectivity of SFV in balloon-injured vessels is based on the removal of the endothelium, which results in accessibility of VSMC in the media that carry specific binding sites for the SFV vector.

Gene Therapy 2001 9, 95-101. DOI: 10.1038/sj/gt/3301632

Semliki Forest virus; adenovirus; vascular smooth muscle cells; endothelial cells; rat aorta; balloon injury; gene therapy

Introduction

Gene delivery by recombinant virus is considered as a possible intervention strategy to prevent restenosis following angioplasty. Several recombinant viruses, eg adenovirus, have been used to reduce neointima formation in animal models for restenosis.¹ However, the viral vectors currently employed possess features that may ultimately limit their clinical use, including non-selectivity for cell type, late onset of gene expression, and safety problems for the host.² Therefore, the search for new vectors seems mandatory.

Limited cell type selectivity of viral vectors is a major issue, since transfection of non-target tissue may be potentially harmful. Recently, we discovered that recombinant Semliki Forest virus (SFV) produced in BHK21 cells displays a high selectivity for vascular smooth muscle cells (VSMC) over endothelial cells (EC) in cell culture.³ Furthermore, the transgene expression of SFV vectors is extremely rapid.^3,4,5 Others have shown that SFV containing mutated envelope proteins are safe vectors for in vivo gene delivery.^6,7 Therefore, SFV may be attractive as a vector for antiproliferative therapy on VSMC to reduce post-angioplasty restenosis. However, the features of SFV with respect to in vivo transfection of vascular cells have not been tested yet. Based on the previous in vitro study, we speculate that SFV represents a vector for selective and rapid gene delivery into VSMC balloon-injured blood vessels. Furthermore, the mechanism for the observed selectivity of the SFV for VSMC is still unknown. We speculate that the selectivity depends on the presence of specific binding sites for SFV on VSMC, yet absent on EC.

In the present study, we assessed the cell-type selectivity and time-course of SFV expression in vivo, using the reporter gene LacZ, in balloon-injured and intact rat aorta. The time-course of expression was compared with that of an adenoviral vector. To address the issue of mechanism for selectivity, we studied the binding and fusion characteristics of SFV in cultured VSMC and EC.

Results

In vitro titration of recombinant SFV and adenovirus

In order to compare recombinant SFV with adenovirus in in vivo experiments, the titer of both vectors was first assessed in cultured rat aortic smooth muscle cells (A7r5). Stock solutions of SFV-LacZ (10⁹ IU/ml, as assessed in the BHK21 producer cell line, which parallels 10¹⁰ virus particles (VP)/ml) and Ad-LacZ (10¹¹ p.f.u./ml, as assessed in HEK 293 producer cells, which parallels 1.25 ´ 10¹² VP/ml) were diluted in 10-fold dilution steps and added to A7r5 cells. The number of LacZ-positive cells was counted after 24 and 48 h expression and the titer was calculated as described in the Materials and methods. For SFV-LacZ the titer in A7r5 was maximal at 24 h expression, and represents 10% of the virus particles (Table 1). After 48 h, the SFV titer was similar to that at 24 h. In contrast, the Ad-LacZ titer was still increasing after 48 h, but represented only 0.08% of the number of virus particles. Therefore, SFV-LacZ is relatively faster and more efficient than Ad-LacZ and the titer that is used in vivo had to be adjusted accordingly. Hence, for in vivo SFV-LacZ and Ad-LacZ experiments, 50 l of virus suspension with a A7r5-based titer (24 h and 48 h point for SFV-LacZ and Ad-LacZ, respectively) of 10⁹ IU/ml was used per aortic segment. For SFV-LacZ this titer equals the titer as assessed in BHK 21 producer cells. For Ad-LacZ, this titer parallels a HEK 293 cell-assessed titer of 5 ´ 10¹⁰ p.f.u./ml. The amount of Ad-LacZ virus applied per segment thus represents 2.5 ´ 10⁹ p.f.u. per segment, a quantity that is commonly used to perform experimental gene therapy on arteries.^8,9

In vivo gene transfer

To evaluate the efficiency of SFV-mediated gene delivery into balloon-injured vessels in vivo, SFV-LacZ (50 l of 10⁹ IU/ml) was locally administered to intact and partly denuded rat abdominal aortic segments using a dwelling technique, administering the virus for 10 min. The denuded segments were treated with SFV-LacZ immediately after balloon injury. After 3, 6, 12, 24, 48 and 72 h (for intact aorta only 24 h, where expression was optimal in balloon-injured vessels), LacZ expression was evaluated.

In balloon-injured aortic segments LacZ expression was detected starting from 12 h after vector administration (Table 2). At this time-point, LacZ expression was exclusively present in the border region of the media and the adventitia, and deep into the adventitia. LacZ expression was maximal after 24 h. After 48 and 72 h, the presence of LacZ staining decreased, although the intensity of staining seemed maximal at 48 h. In general, LacZ was predominantly detected in the media in 66% of the positive sections (Figure 1). LacZ staining was also present in the adventitia in 40% of the positive sections. Expression in the adventitia was limited to a single or a few cells in the media of nourishing vessels, or fibrous and fat tissue. Notably, the intima of these nourishing vessels was not positive for LacZ.

When SFV-LacZ was administered to intact aorta, LacZ staining was absent from the media or intima Figure 1. In 12 ± 1% of the sections small areas of blue staining were seen in the adventitia. In addition, no LacZ expression was found in denuded aorta of rats incubated with recombinant replication-defective SFV without a transgene (SFVcontrol, n = 63 sections).

Ad-LacZ was administered to denuded aorta in the same way as SFV-LacZ (50 l of a 10⁹ IU/ml solution (A7r5 titer)). Expression was studied after 12, 24, 48 and 72 h (Table 2). LacZ expression could be detected as soon as 12 h after Ad-LacZ treatment, and the percentage of positive section did not seem to depend on the time-course. There was a striking difference with SFV-LacZ with respect to the location of LacZ expression. Ad-LacZ expression was predominantly present in the adventitia (92% of all positive sections). In contrast with SFV-LacZ, Ad-LacZ transfected the intima (14% of all positive sections). As expression time increased, the location of LacZ appeared to progress from the outer layers of the adventitia towards the border between the adventitia and the media, but only after 72 h LacZ was detected within the media. LacZ expression in the media was relatively low (8% of all positive sections).

Binding and fusion properties of SFV in cultured EC and VSMC

To test the molecular basis for the specificity of SFV for VSMC versus EC, binding and fusion characteristics of pyrene-labeled SFV were determined in cultured VSMC and EC. When increasing amounts of virus were added to VSMCs, a substantial fraction of the virus (10-25% depending on the amount of virus added) was found to associate with the cells during a period of 1 h (Figure 2a). Part of the cell-associated virus appeared to be fused with a cellular target membrane, as calculated on the basis of dilution of the pyrene fluorophore Figure 2a. The amount of virus fused with VSMCs clearly saturated as virus concentrations increased, consistent with the notion that a limited number of virus particles can fuse from within endosomes in a certain period of time.

In sharp contrast with VSMC, virus binding to EC was limited Figure 2b. Furthermore, and perhaps even more importantly, a negligible fraction of the EC-associated virus appeared to be fused. This strongly indicates that EC lack the receptor for proper binding of SFV and internalization of the virus into endosomes to allow fusion of the virus with the endosomal membrane.

Time-course of SFV-LacZ expression in cultured VSMC

The onset of SFV transgene expression after virus administration is an important issue because it will determine whether recombinant SFV can be applied to interfere with target mechanisms activated at an early or late stage after balloon injury. Therefore, we explored which factors influence the time-course of SFV expression. Virus concentration and metabolic state of the target cells are putative factors that influence expression time of SFV. To explore the relative contribution of both factors, time course of expression of SFV-LacZ was studied at different MOI in quiescent and proliferating A7r5 cells. In both quiescent and proliferating cells an increase of MOI led to increased transfection efficiency and reduced time to half-maximal expression (Figure 3). Although the presence of 10% FCS lowered the maximum transfection efficiency by 20% at MOI 10, the time to half-maximal expression was not influenced by the proliferative state of the cells Figure 3.

During virus production, -galactosidase is formed that may be trapped in the SFV particle and carried into the host cell during fusion, which leads to the false impression of rapid LacZ expression. To investigate whether this happens, A7r5 cells were incubated for 30 min with SFV-LacZ at MOI 10⁴. After removal of the virus, the cells were grown for 6 h either in the presence or absence of the protein synthesis inhibitor cycloheximide (1 M). None of the cycloheximide-treated cells expressed LacZ whilst 100% of the vehicle-treated cells were positive for LacZ (Figure 4).

Discussion

From data of a previous in vitro study showing strong selectivity of SFV for vascular smooth muscle cells over endothelial cells,³ we hypothesized that SFV could be a rapid and selective viral vector for in vivo gene transfer into balloon-injured vessels. In the present paper we show that (1) SFV is a vector selective for in vivo gene transfer into the lamina media of balloon-injured rat aorta; (2) expression in balloon-injured aorta is detected starting from 12 h after a 10-min virus incubation period; and (3) its selectivity for VSMC over EC is caused by the presence of specific binding sites in VSMC that are absent in EC. The selectivity and rapid expression may represent favorable properties for clinical use of SFV as a vector in balloon-injured vessels.

The in vivo selectivity and the in vitro binding/fusion data are in line with our previous work.³ There we demonstrated that SFV-LacZ expression is low in cultured human and rat EC compared with VSMC and to EC transfected with adenovirus-lacZ. Now we show that in VSMC, SFV most likely displays specific, as well as nonspecific binding, and that only SFV bound to specific binding sites fuses. In EC, only nonspecific binding of SFV seems to exist and hence, only a very low amount of SFV fuses. This explains why VSMC were and EC were not transfected. Absence of binding to EC may also explain why SFV did not transfect the intima of intact aorta in the present study. Furthermore, the endothelium most likely formed a barrier for SFV to penetrate into the media of the intact vessels. As a consequence, transfection of VSMC in the media was prevented in intact aorta.

The absence of transfection of rat aortic EC is in apparent contradiction with other observations indicating that wild-type SFV does infect brain EC in BALB/c mice, thereby causing blood-brain barrier damage.¹⁰ This contradiction may be based on the difference in EC type, the difference in virus strain (avirulent mutant SFV-A7 derived from wild-type versus recombinant SFV3-Helper2), and/or the difference in host species (mouse versus rat/human). A possible difference in susceptibility of brain and somatic EC for infection with SFV would explain why wild-type SFV typically is a neurotropic virus.¹¹ Hence, wild-type SFV may have adapted to brain endothelial cells rather than peripheral endothelial cells. On the other hand, the SFV that infected brain EC was grown in a BALB/c mouse brain cell line (MBA-1), whilst the SFV used in this study was grown in baby hamster kidney cells (BHK 21). Therefore, the difference in transfection of brain and aortic EC may also be caused by adaptation of virus strain to the host cell. If true, this ultimately may provide a method for targeting of the SFV vector. However, it cannot be excluded that differences in host species are involved. Wild-type SFV causes allergic encephalomyelitis in mice,¹⁰ whilst in humans it is considered avirulent.¹¹

In experiments in both intact and denuded aorta, LacZ expression was detected in the adventitia. Most likely, this expression is due to leakage of virus via the nourishing vessels in the adventitia. When LacZ was expressed in these small vessels, strikingly, cells in all tissue layers except for the intima were transfected. Leakage of virus through small vessels has been observed for adenovirus, both in the present as well as previous studies in the pig coronary artery (K Walsh, St Elizabeth Medical Center, Boston, personal communication). Our results suggest that these small vessels are not as strongly sealed for passage of SFV as large vessels such as the aorta itself. Less likely, but still possible, staining in the adventitia may have resulted from accidental spilling of SFV during withdrawal of the catheter.

Apart from cell type selectivity, other important features of the SFV vector revealed in this study relate to its efficacy of cell entry and rapidity of transgene expression. An incubation period of only 10 min in denuded aorta with SFV was sufficient to induce significant cell entry and LacZ expression. Furthermore, LacZ expression was detected as early as 12 h after virus administration. The present in vitro data suggest that virus concentration is of major importance, in contrast to the proliferative state of the cells. Therefore, transgene expression in vivo could be enhanced by increasing SFV titers in the vessel wall. A clinically relevant method to accomplish this is to use intramural injecting balloons to overcome anatomic barriers and deliver SFV straight into the vessel wall.¹² Alternatively, transgene expression levels can be improved employing translational enhancers.⁴

We suggested that SFV may have favorable properties compared with other viruses, such as adenovirus. More specifically, SFV was selective for, and more rapidly expressed in, cultured VSMC as compared with Ad-LacZ.³ Indeed, the present study shows that in vivo SFV-LacZ is more specific and more rapidly expressed in the media of denuded vessels. This feature of SFV-LacZ may be particularly interesting for gene transfer after stenting where neointima formation due to proliferation and migration of myofibroblasts and VSMC is the main mechanism leading to restenosis.¹³ Stent placement represents more than 80% of all percutaneous transluminal coronary angioplasty procedures in western countries, and it is therefore of particular interest to develop in-stent gene transfer techniques that employ recombinant SFV as a vector.

Faster expression in the media and absence of transgene expression in the intima of SFV-LacZ may both be attributed to the higher specificity for non-endothelial cells of SFV-LacZ as compared with Ad-LacZ. Previous studies confirm that Ad-LacZ is not selective.^8,14 To demonstrate that non-selectivity of adenovirus can be prevented, Kim et al⁹ showed that adenovirus could be transcriptionally targeted with the use of the VSMC-specific SM22 promoter. However, as adenovirus entrance in EC is not prevented by using a transcriptionally targeted vector, the SM22-driven adenoviral vector may still be cytopathic to EC.¹⁵ In the present study, we show that SFV does not enter EC, which explains why SFV was not toxic to EC in vitro.³ This feature may improve re-endothelialization and avoid a cytopathic immune response against endothelial cells in vivo. Future in vivo studies will have to confirm whether the specificity of SFV for VSMC beneficially influences anti-restenotic treatment as compared with other viral vectors.

In the present study, faster expression of SFV-LacZ in VSMC as compared with Ad-LacZ is observed both in vivo and in vitro. Furthermore, the in vitro data presented in Table 1 suggest that in SFV-LacZ stock solutions, a larger part of the virus particles present is able to cause transgene expression in the first 48 h after virus administration than in Ad-LacZ stocks. This suggests that less viral protein has to be administered when SFV is used than in case of adenoviral vectors, if the goal is to rapidly transfer genes into VSMC. A decreased number of virus particles may be advantageous as to limit the chance for immunosuppression of virus expression in vivo. Another difference with adenovirus is that SFV expression is transient within the first 72 h. Transient expression of SFV is known to be caused by virus-induced apoptosis. Also adenovirus causes apoptosis in VSMC, but in a slower rate than SFV, which may explain the more stable expression of Ad-LacZ in the first 72 h as compared with SFV-LacZ.³ Cytotoxicity may have a beneficial effect on restenosis, but may also limit the applicability of SFV when long-lasting gene expression is desired. The production of recombinant SFV with an inhibited cytotoxicity, as accomplished for the closely related Sindbis virus, may be feasible,^16,17 and it is certainly worthwhile to construct and test these vectors in future studies.

In conclusion, this study shows that SFV represent a promising vector system for rapid and selective gene transfer into VSMC of balloon-injured aorta. Its selectivity is probably based on the presence of specific virus binding sites on VSMC that are not present on endothelial cells. Thus, transfection by SFV is prevented in intact vessels. Given the selectivity and rapidity of SFV-mediated gene expression in VSMC, and the possibilities to enhance transgene expression, SFV has the potential to become applicable as a gene delivery system for prevention of restenosis in balloon-injured vessels. Further studies are required to optimize SFV for in vivo use.

Materials and methods

Cell culture

Rat aortic smooth muscle cells (A7r5) and human vascular endothelial cells (EC-RF24) were cultured as described before.³ Assessment of binding and fusion of SFV was performed using cells seeded on coverslips and grown to confluence.

Production of recombinant Semliki Forest virus and adenovirus

SFV3-LacZ was produced using pSFV3-Helper2, to obtain conditionally infectious replication-defective particles.⁶ Titers were assessed using BHK21 cells as described previously.⁵ Ad-LacZ was prepared and titers were assessed as described previously.¹⁸

Animals

Approval of animal experiments was given by the institutional Animal Care and Use Committee. Male Wistar rats (Harlan, Zeist, The Netherlands) weighing 350 to 400 g were housed in a light- and temperature-controlled facility, and fed with standard rat chow (Hope Farms, Woerden, The Netherlands) and tap water. Each treatment group consisted of four rats unless indicated otherwise. During all surgical procedures the rats were anesthetized with isoflurane/O₂/N₂O.

Balloon-injury and administration of virus in abdominal aorta

Angioplasty balloon catheters (Cheetah catheter, Medtronic, Minneapolis, USA) with a 2.5-mm balloon were inserted via the femoral artery and positioned in the abdominal aorta. Then, the balloon was inflated to a pressure of 4 atm, deflated and passed forward and backward four times for 1 cm, repeating the complete procedure four times. This leads to a partial denudation of the intima, as confirmed by microscopic investigation. After balloon-injury the femoral artery was sutured to restore the circulation. Intact and balloon-injured aortic segments were incubated with virus using a dwelling technique. The aortic segment was prepared free from the vena cava over a length of 1.5 to 2.0 cm, starting between the renal arteries and proceeding towards the bifurcation to the femoral arteries, and branches were closed. Two microvascular clips were positioned at each end of the segment. The segment was then punctured and carefully rinsed with saline using a syringe with a small needle. A polyethylene catheter, pre-filled with virus solution, was inserted via the puncture site and sealed with a 6-0 silk suture. Virus solution was introduced via the catheter and pressure applied such that the aorta obtained a diameter of 2 ± 0.2 mm. An incubation period of 10 min was used. After these 10 min, the virus and catheter were withdrawn from the segment. The aorta was sutured with a 9-0 suture thread, and blood flow was restored by removing the clips.

Detection of reporter gene expression

For collection of aortic segments, rats were anesthetized, heparinized and the treated aortic segments were dissected. For detection of -galactosidase activity, the segments were washed in cold phosphate buffered saline (PBS), fixed in cold 1.25% glutaric aldehyde (10 min), rinsed with phosphate buffer, and incubated for 6 h with X-gal substrate solution. The X-gal substrate contained 1 mg/ml 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-gal) (Eurogentec, Seraing, Belgium), 5 mM K₃Fe(CN)₆ (Merck Eurolab, Amsterdam, The Netherlands), 5 mM K₄Fe(CN)₆ (Merck Eurolab), 2 mM MgCl₂/6H₂O (Merck) and 10% N,N-dimethylformamide (Sigma, Amsterdam, The Netherlands), in phosphate buffer pH 7.5. After staining, the segments were washed with distilled water and stored at 4°C in 4% paraformaldehyde. Before sectioning, the segments were washed with distilled water, dehydrated with alcohol, treated with xylene, stained with eosine and embedded in paraffin. Three-m sections were cut from random locations in each segment. The position of the blue, LacZ-positive cells in eosin-stained sections was determined with the help of parallel sections that were additionally treated with a Verhoeff-von Giesson staining to locate the lamina elastica interna. The percentage of LacZ-positive section was calculated for each individual rat. From these individual percentages, the overall transfection efficiency was calculated in percentage positive sections ± standard error of the means (s.e.m.) per rat.

Binding and fusion characteristics of SFV in cultured cells

To measure binding and fusion of SFV, we used a previously described assay based on measurement of pyrene phospholipid monomers and eximers, which represent fused and non-fused SFV, respectively.¹⁹ In short, pyrene-labeled SFV was produced in BHK-21 cells. Coverslips with A7r5 or EC-RF24 were incubated with 1 to 10 nmol of pyrene-labeled SFV (approximately 10⁹ to 10¹⁰ IU) in serum-free medium at 37°C for 1 h. After this, unbound SFV was washed out with ice-cold Hanks buffer. The cells were isolated from the coverslips and the amount of pyrene monomers and eximers was measured by fluorescence at 397 nm and 480 nm, respectively, after excitation at 345 nm. Then, all pyrene molecules were converted to monomers by addition of C12E8 (Nikko Chemicals, Tokyo, Japan) and the amount was measured in order to assess the total amount of pyrene associated (= total bound) with the cells.

SFV-LacZ expression in cultured VSMC

A7r5 cells were seeded at a density of 2.5 ´ 10³ cells/well and allowed to attach for 4 h in DMEM + 10% FCS. Then, the cells were washed with PBS and the medium was replaced with DMEM + 10% or DMEM + 0.1% FCS for 20 h to obtain cells in a proliferating or quiescent state respectively.²⁰ Then medium was changed to serum-free DMEM and SFV-LacZ was added at 0.1 to 10⁴ MOI (1 h) and transfection efficiency was studied after 1, 3, 6, 12 and 24 h. The half-maximal expression time of each curve was assessed by computerized curve fitting (SigmaPlot 5.0; Jandel Scientific, Chicago, IL, USA).

To exclude the possibility of intraparticle transfer of -galactosidase, A7r5 rat aortic smooth muscle cells were incubated for 30 min with SFV-LacZ at MOI 10⁴. Thereafter, the cells were washed with PBS, and cultured in DMEM + 10% FCS in the presence of either 1 M cycloheximide or vehicle for 6 h. After 6 h at 37°C, 1 M cycloheximide was added to the vehicle-treated cells as well, and all cells were stained.

Statistical analysis

Gene transfer efficiencies were compared by Students' t test. Times to half-maximal expression were tested through linear regression analyses after logarithmic transformation of the MOI. All analyses were performed with SPSS 10.0 software (Jandel).

Acknowledgements

We are very grateful to Hans Bartels, Alexandra Beuving, Martin Houwertjes, Gera Kamps, Bianca Meijeringh, Egbert Scholtens, Lucia van der Veen, and Jetti van Wijk, for their technical support. We thank Dr A Suurmeijer of the Department of Pathology, University Hospital Groningen for his useful suggestions. Further, we thank Medtronic-Bakken Research Maastricht, Inex Pharmaceuticals (grant to J Wilschut), and the Netherlands Heart Foundation (ICIN grant No. 295-0019 to Dr RA Tio) for their financial support.

1 Feldman L.J., Tahlil O., Steg G. Perspectives of arterial gene therapy for the prevention of restenosis. Cardiovasc Res 1996; 32: 194, 194. MEDLINE

2 Feldman L.J., Steg G. Optimal techniques for arterial gene transfer. Cardiovasc Res 1997; 35: 391, 391. MEDLINE

3 Roks A.J.M.et al. . Vectors based on Semliki Forest virus for rapid and efficient gene transfer into non-endothelial cardiovascular cells: comparison to adenovirus. Cardiovasc Res 1997; 35: 498, 498. MEDLINE

4 Sjöberg E.M., Suomalainen M., Garoff H. A significantly improved Semliki Forest virus expression system based on translation enhancer segments from the viral capsid gene. Bio/technol 1994; 12: 1127, 1127.

5 Liljeström P., Garoff H. A new generation of animal cell expression vectors based on the Semliki Forest virus replicon. Bio/technol 1991; 9: 1356, 1356.

6 Berglund P.et al. . Semliki Forest virus expression system: production of conditionally infectious recombinant particles. Bio/technol 1993; 11: 916, 916.

7 Smerdou C., Liljeström P. Two-helper RNA system for production of recombinant Semliki Forest virus particles. J Virol 1993; 73: 1535, 1535.

8 Schulick A.H.et al. . Endothelium-specific in vivo gene transfer. Circ Res 1995; 77: 475, 475. MEDLINE

9 Kim S.et al. . Transcriptional targeting of replication-defective adenovirus transgene expression to smooth muscle cells in vivo. J Clin Invest 1997; 100: 1006, 1006. MEDLINE

10 Soiluhänninen M.et al. . Semliki forest virus infects mouse brain endothelial cells and causes blood-brain barrier damage. J Virol 1994; 68: 6291, 6291. MEDLINE

11 Atkins G.J., Sheahan B.J., Liljeström P. The molecular pathogenesis of Semliki Forest virus: a model virus made useful? J Gen Virol 1999; 80: 2287, 2287. MEDLINE

12 Varenne O.et al. . Percutaneous adenoviral gene transfer into porcine coronary arteries: is catheter-based gene delivery adapted to coronary circulation? Hum Gene Ther 1999; 10: 1105, 1105. Article MEDLINE

13 Bult H. Restenosis: a challenge for pharmacology. Trends Pharmacol Sci 2000; 21: 274, 274. MEDLINE

14 Lee S.W.et al. . In vivo adenoviral vector-mediated gene transfer into balloon-injured rat carotid arteries. Circ Res 1993; 73: 797, 797. MEDLINE

15 Kafri T.et al. . Cellular immune response to adenoviral vector infected cells does not require de novo viral gene expression: implications for gene therapy. Proc Natl Acad Sci USA 1998; 95: 11377, 11377. Article MEDLINE

16 Dryga S.A., Dryga O.A., Schlesinger S. Identification of mutations in a Sindbis virus variant able to establish persistant infection in BHK cells: the importance of a mutation in the nsP2 gene. Virology 1997; 228: 74, 74. Article MEDLINE

17 Scallan M.F., Allsopp T.E., Fazakerley J.K. Bcl-2 acts early to restrict Semliki Forest virus replication and delays virus-induced programmed cell death. J Virol 1997; 71: 1583, 1583. MEDLINE

18 Haisma H.J.et al. . Targeting of adenoviral vectors through a bispecific single-chain antibody. Cancer Gene Ther 2000; 7: 901, 901. MEDLINE

19 Stegmann T.et al. . Evaluation of viral membrane fusion assays. Comparison of the octadecylrhodamine dequenching assay with the pyrene excimer assay. Biochemistry 1993; 32: 11130, 11130.

20 Filipeanu C.M., Henning R.H., de Zeeuw D., Nelemans A. Intracellular angiotensin II and cell growth of vascular smooth muscle cells. Br J Pharmacol 2001; 132: 1590, 1590. MEDLINE

Figure 1 Expression of SFV-LacZ in rat abdominal aorta 24 h after administration (10-min incubation) of 50 l SFV (10⁹ IU/ml) in intact aorta (a), or after balloon-injury (b). Black arrowheads, LacZ expression in the media; L, lumen.

Figure 2 Binding and fusion characteristics of fluorescent pyrene-labeled SFV in A7r5 vascular smooth muscle cells (a) and EC-RF24 endothelial cells (b). The amount of bound and fused virus was measured 1 h after addition of SFV.

Figure 3 In vitro time-course of SFV-LacZ expression in quiescent (a) and proliferating (b) VSMC (A7r5). MOI, multiplicity of infection (No. IU of SFV/No. cells). n = 4 per point.

Figure 4 (a) Absence of LacZ expression in cultured VSMC (A7r5) when 1 mol/l chloorheximide was added after a 30-min incubation period with SFV-LacZ (10⁹ IU/ml). (b) Presence of LacZ expression when chloroheximide was added after the 6-h expression period. Magnification, ´200.

Table1 Titration of SFV-LacZ and Ad-LacZ in cultured rat aortic smooth muscle cells (A7r5 cells)

Table2 In vivo gene delivery of LacZ reporter into balloon-injured rat abdominal aorta with the use of SFV-LacZ and Ad-LacZ