Endothelium-Specific In Vivo Gene Transfer
Abstract Targeted expression of genetic material within the vascular endothelium is potentially a powerful tool for the investigation of endothelial cell (EC) biology. We developed, optimized, and characterized an efficient somatic transgenic model of EC-specific gene transfer. Rat carotid arteries were infused with adenovirus expressing a β-galactosidase (β-gal) gene. The level and cell-type specificity of recombinant gene expression were measured by assaying β-gal activity in vessel extracts and by counting transduced cells in histological sections. Toxicity was evaluated by counting total ECs (3 days) and by measuring neointimal formation (14 days). Effects of transduction on the proliferation of vascular cells were measured with bromodeoxyuridine and [3H]thymidine. Maximum recombinant gene expression resulted from infusion of 1×1010 to 1×1011 plaque-forming units (pfu) per milliliter; ≈35% of luminal ECs were transduced. A high degree of EC specificity (90% to 98% of total transduced cells) was maintained over this range of virus concentrations. More highly concentrated virus resulted in loss of β-gal expression and a large decrease in luminal EC number (97% decrease, P<.001). Gene transfer at 4×1010 pfu/mL was efficient, preserved EC integrity, and caused minimal neointimal formation. After gene transfer, there were early (3-day) increases in both EC and smooth muscle cell proliferation. At 14 days, only EC proliferation remained elevated (18% versus 1.4% in vehicle-infused arteries, P=.005). This animal model permits efficient highly EC-specific gene transfer. Vascular toxicity is minimal, although the EC proliferative index is elevated. This model will be useful in experiments that elucidate the biological role of EC gene products and define pathways of EC gene regulation and signal transduction in vivo.
The vascular endothelium plays a major role in the regulation of important biological processes, including hemostasis,1 vasomotion,2 3 and the initiation of atherosclerosis.4 The molecular mechanisms through which the endothelium participates in these diverse biological processes are, however, incompletely understood.
Transfer and expression of genetic material within the vascular endothelium in vivo represents a powerful tool for the investigation of EC biology. Overexpression of specific genes within the endothelium could permit elucidation of the biological roles of their protein products. In addition, in vivo introduction of genetic regulatory sequences spliced to reporter genes would allow the definition of DNA sequences involved in the regulation of EC gene expression after physiological stimulation. This general strategy has been accomplished in other tissues in vivo5 6 but never in endothelium. Moreover, EC-specific expression of either dominant negative or constitutively active components of signal transduction pathways7 could help clarify pathways of endothelial gene regulation in vivo.
The execution of informative in vivo experiments using targeted gene expression in endothelium is dependent on the development of an animal model system of reproducible, efficient, EC-specific gene transfer. With such a system, the level of expression of transferred genes would be easily quantified, and the biological consequences of recombinant gene expression would be confidently and specifically attributed to alterations in EC physiology. Previous work has suggested that adenovirus-mediated gene transfer might be suitably efficient and sufficiently EC specific to permit the execution of biological experiments predicated on EC-specific gene transfer8 9 10 ; however, this has never been demonstrated quantitatively. In addition, the question of whether adenovirus-mediated gene transfer per se perturbs the phenotype of a normal artery (a critical issue in the design and interpretation of endothelial gene transfer experiments) has not been addressed.
In the present study, we describe the development of an animal model system of somatic EC-specific gene transfer using an adenovirus vector. The system is optimized to allow the use of lower concentrations of adenovirus, and effects of virus exposure and gene transfer on vessel phenotype are defined.
Materials and Methods
Viruses and Plasmids
Av1LacZ4, a recombinant replication-defective adenovirus, contains a nuclear targeted Escherichia coli β-gal gene under the control of the Rous sarcoma virus long terminal repeat promoter. The construction of this vector (initially provided by Bruce Trapnell, Genetic Therapy Inc) has been described in detail.11 Exposure of a cell transduced with this vector to X-Gal results in a histochemical reaction that stains the nucleus blue. To construct AdSV40Luc, an expression cassette containing the simian virus 40 early promoter driving expression of a firefly luciferase gene, was ligated into the Bgl II site of pAdBglII (a plasmid containing the adenovirus type 5 sequences 0 to 1 and 9.2 to 16.1 map units; kindly provided by Dr Blake Roessler, University of Michigan). The resulting plasmid was linearized by digestion with Nhe I and cotransfected into 293 cells along with the large Cla I fragment of DNA prepared from type 5 adenovirus strain dl327.12 AdHS-1.2, which contains a cDNA for biologically active hirudisin,13 was constructed by methodology essentially identical to that used to construct Av1LacZ4 and AdSV40Luc. Recombinant virus was plaque-purified and propagated on 293 cells. Concentrated stocks of viruses were prepared by cesium banding and dialyzed against 10 mmol/L Tris-HCl (pH 7.5) and 1 mmol/L MgCl2 containing 10% glycerol. Aliquots of 10 to 50 μL of virus stock were stored over liquid nitrogen until use. The titers of concentrated virus stocks (1 to 2×1011 pfu/mL for Av1LacZ4, 5×1010 pfu/mL for AdSV40Luc, and 5×1011 pfu/mL for AdHS-1.2) were determined by plaque titration on 293 cells. The plasmid pCMVLuc,14 containing the human cytomegalovirus immediate-early promoter driving expression of a firefly luciferase gene, was used in cointernalization studies (see below).
All animal procedures were approved by the Animal Care and Use Committee of the National Heart, Lung, and Blood Institute. Adult male Sprague-Dawley rats (Taconic Farms, Germantown, NY) weighing 350 to 450 g were used. Rats were anesthetized with intraperitoneal injections of ketamine (Fort Dodge Laboratories, Inc) and xylazine (Miles Inc) in doses of 100 and 10 mg/kg, respectively. After surgery, animals were allowed water ad libitum but were not allowed food until 8 to 12 hours later in order to minimize the risk of aspiration.
In Vivo Gene Transfer Into Vessel Segments
To perform in vivo arterial gene transfer, the left carotid system was surgically exposed, proximal and distal control was obtained, and an arteriotomy was made on the external carotid artery. A 1-cm segment of common carotid artery was isolated. A 24-gauge polytetrafluoroethylene catheter was introduced through the external carotid arteriotomy, and the isolated vessel segment was flushed with ≈1 mL medium 199 (Biofluids).
Aliquots of adenovirus were thawed, placed on wet ice, and used within 2 hours. Medium 199 containing rat serum albumin (1 mg/mL, Sigma Chemical Co) was used as needed to dilute virus stocks. For each rat, 50 μL of adenovirus-containing or control solution were drawn up into a 24-gauge polytetrafluoroethylene catheter mounted on a tuberculin syringe. The catheter was inserted into the external carotid arteriotomy and secured with a silk tie, and the solution was infused into the isolated carotid segment, resulting in distension of the common carotid artery. The solution was allowed to dwell in the vessel segment for 20 minutes, during which time the carotid segment remained distended. The solution was then withdrawn into the catheter, the catheter was removed, and external carotid artery was ligated. Blood flow was reestablished through the common and internal carotid arteries.
Fixation and Removal of Vessel Segments
Animals were killed by overdosing with pentobarbital (≈600 mg/kg IP). At the time of death, none of the animals had clinical evidence of wound infection, and all common carotid arteries were patent. For experiments involving histological examination of vessels, perfusion fixation was carried out 3, 7, or 14 days after gene transfer. After retrograde cannulation of the abdominal aorta and division of the left jugular vein, the arterial tree was cleared of blood by perfusion with PBS (pH 7.4) at a pressure of 100 mm Hg. After clearing, vessels were perfusion-fixed in situ with 2% formaldehyde (Mallinckrodt) and 0.2% glutaraldehyde (Sigma) in PBS (pH 7.2). Fixative was perfused through the arterial cannula at 100 mm Hg for 5 minutes. The perfusion-fixed vessels were then excised and incubated in fixative for an additional 2 hours. Other arteries were not perfusion-fixed but were removed and immediately frozen for subsequent measurement of recombinant β-gal or luciferase activity.
Evaluation of Recombinant β-gal and Luciferase Gene Expression
Recombinant β-gal expression was evaluated 3 days after gene transfer by two methods: (1) histochemical staining of excised fixed arteries with X-Gal and subsequent identification of blue-stained nuclei, both grossly and on histological sections, and (2) detergent extraction of excised arteries and measurement of β-gal activity in the tissue extracts. At 7 and 14 days after gene transfer, β-gal expression was evaluated by histochemical staining alone. For histochemical staining, vessels were washed in PBS and incubated in X-Gal for 2 hours, as described previously.11 The arteries were then divided into segments of ≈2 to 4 mm in length, and these segments (2 to 5 per vessel) were embedded in paraffin blocks. Sections (5 μm thick) were cut from the blocks and counterstained with nuclear fast red. For Av1LacZ4-exposed vessels, both total and X-Gal–stained ECs (identified by characteristic morphology, luminal location, and specific staining; see below) per histological cross section were counted. The number of sections counted per vessel ranged from 4 to 12. Initially, cells were counted in 8 to 12 randomly spaced sections per vessel; however, the variability of both total and X-Gal–stained cells between sections was so low (data not shown) that in later experiments only 4 sections per vessel were counted. The vessel cross sections were always spaced a minimum of 50 μm apart to avoid counting the same cells twice. From the data obtained from individual cross sections, the mean transduced cell number per section was calculated for each artery. A mean percentage of transduced ECs per vessel section was also calculated for each artery by dividing the mean number of transduced ECs per section by the mean number of total ECs per section.
Recombinant β-gal or luciferase activity in vessel extracts was measured after mincing each vessel into small pieces in lysis buffer (200 μL per vessel; 0.2% Triton X-100 in 100 mmol/L potassium phosphate [pH 7.8] for β-gal measurement, and 1% Triton X-100, 100 mmol/L potassium phosphate, 1 mmol/L dithiothreitol, and 2 mmol/L EDTA [pH 7.8] for luciferase measurement). Minced pieces were further homogenized in a Polytron (Brinkman), and aliquots of vessel lysate were assayed for β-gal or luciferase activity. β-gal activity was measured with the substrate 3-(4-methoxyspirol[1,2-dioxetane-3,2′-tricyclo-[220.127.116.11,7]decan]-4-yl)phenyl-β-d-galactopyranoside (Galacto-Light, Tropix). Luciferase activity was measured with the substrate 2-(6′-hydroxy-2′-benzothiazolyl)-Δ2-thiazoline-4-carboxylic acid (d-luciferin, Analytical Luminescence Laboratory). Light emission for both assays was measured with a Monolight 2010 luminometer (Analytical Luminescence Laboratory). For Av1LacZ4, standard curves were generated by using purified E coli β-gal (specific activity, 300 U/mg; Boehringer Mannheim). Luciferase activity in AdSV40Luc-transduced cells was quantified in RLU. All samples were assayed in duplicate. Total protein in vessel extracts was measured with the BCA protein assay (Pierce), with bovine serum albumin used as a standard.
Evaluation of the EC Specificity of Gene Transfer
The EC specificity of in vivo gene transfer was assessed by examination of histological sections taken from animals killed 3 days after gene transfer. The same histological sections used to calculate the number and percentage of transduced ECs (see above) were examined for the presence of blue-stained nuclei in the vascular media. Medial cells were identified as smooth muscle cells by characteristic spindle morphology and by immunohistochemical staining of companion sections by use of an antibody to smooth muscle cell actin (Sigma Chemical) at a dilution of 1:10 000. For each vessel, the mean number of smooth muscle cells with blue nuclei per histological section was calculated. EC specificity of transduction was then determined for each vessel by the following formula: mean X-Gal–positive (ie, having blue nuclei) ECs per section/(mean X-Gal–positive ECs per section+mean X-Gal–positive smooth muscle cells per section)×100.
Evaluation of EC Number in Vessel Sections
To evaluate the potential acute toxicity of the Av1LacZ4 vector, we counted total luminal ECs present in histological sections of control and virus-infused vessels harvested 3 days after gene transfer. Vessel sections used for this analysis included those described above (taken from arteries exposed to Av1LacZ4) as well as identically prepared sections taken from arteries exposed to control solutions (see below). Total luminal ECs were again counted in a minimum of four randomly chosen microscopic cross sections per vessel, separated by at least 50 μm. To confirm the identification of luminal cells as ECs, additional vessel cross sections were cut from the same paraffin blocks used to provide sections for cell counting and were processed for immunohistochemical detection of von Willebrand factor. Staining was performed by use of an antibody to von Willebrand factor (Dako Corp) at a dilution of 1:3200. Bound antibody was detected with a secondary antibody conjugated to horseradish peroxidase.
Two groups of control vessels were used for the biochemical and histological analyses described above. One group was infused with medium 199 containing 1 mg/mL rat serum albumin, the medium used to dilute virus stocks and hereafter referred to as “viral diluent.” This group served as the control for studies involving the evaluation of recombinant β-gal gene expression, both in tissue extracts and in X-Gal–stained cross sections. An additional group of control vessels was used for studies involving analysis of potential acute (3 days after gene transfer) endothelial loss after exposure to Av1LacZ4. This second series of control vessels was exposed to 10 mmol/L Tris-HCl (pH 7.5) and 1 mmol/L MgCl2 containing 10% glycerol, the vehicle used in storing virus stocks and hereafter referred to as “virus storage buffer.” By controlling for both the virus storage buffer and diluent, any differences in EC number between virus-treated and control vessels could be attributed to exposure to adenovirus.
Scanning Electron Microscopy
Carotid arteries were perfusion-fixed and stained with X-Gal as described above. Vessels were then divided in half longitudinally, dehydrated, critical point–dried with liquid carbon dioxide, coated with gold palladium in a plasma coater, and visualized with a Jeol JSM35 scanning electron microscope.15
Evaluation of EC and Smooth Muscle Cell Proliferation
We investigated the effect of adenovirus exposure on EC and smooth muscle cell proliferation by using the proliferation markers BrdU and [3H]thymidine. EC proliferation was assessed 3, 7, and 14 days after transduction (with BrdU); medial smooth muscle cell proliferation was assessed 3 and 14 days after transduction (with BrdU). In addition, the ability of the transduced EC population to undergo mitosis 3 days after transduction was investigated by combined X-Gal staining and [3H]thymidine autoradiography. At 17, 9, and 1 hour before death, animals were given either BrdU by subcutaneous administration (30 mg/kg body wt per dose, Sigma) or [3H]thymidine by intraperitoneal injection (50 mg/kg body wt per dose, New England Nuclear). Animals were killed, and carotid vessels were excised, X-Gal–stained, processed, and embedded in paraffin as described above. In addition to the left carotid artery, the right carotid artery (as a control for baseline EC mitosis in the animal) and a segment of proximal duodenum (as a positive control for cell mitosis) were harvested from each animal and processed in the same manner as the left carotid artery. Because of a concern (for animals treated with BrdU) that the nuclear localized X-Gal staining might interfere with binding or detection of anti-BrdU antibody, vessel (and positive-control duodenal) sections of these animals were immersed in xylene overnight to extract the blue reaction product. Anti-BrdU staining was then performed by using a specific antibody (Sigma) at a dilution of 1:100. Bound antibody was detected with a secondary antibody conjugated to horseradish peroxidase. The BrdU proliferative index of ECs was calculated for each artery from two cross sections per vessel from the following formula: mean BrdU-positive ECs per cross section/mean total ECs per cross section×100.
The rate of medial smooth muscle cell proliferation (as measured by the BrdU index) was determined on the same histological sections used to assess EC proliferation. Smooth muscle cell proliferation was measured at 3 and 14 days after gene transfer. The BrdU index of medial smooth muscle cells in unmanipulated right carotid arteries, harvested from the same animals and processed identically to the left sided vessels, was also determined and used as the baseline mitotic rate of smooth muscle cells in carotid artery media. The BrdU index of smooth muscle cells was calculated for each artery from one cross section per vessel from the following formula: BrdU positive smooth muscle cells per cross section/total smooth muscle cells per cross section×100.
The ability of adenovirus-transduced ECs to undergo mitosis at 3 days after gene transfer was evaluated with [3H]thymidine instead of BrdU, since X-Gal staining was less likely to interfere with autoradiography than with anti-BrdU immunohistochemistry. Sections of 5 μm thickness were dipped in Kodak NTB-2 emulsion, stored at 4°C for 3 weeks, and developed with Kodak D 19 developer. Slides were then counterstained with nuclear fast red. ECs with at least 10 overlying silver grains were identified “[3H]thymidine positive” and therefore as proliferative. The proliferation index of ECs was calculated for each vessel from the mean of two cross sections (the variability in proliferative indices between two sections taken from a particular vessel was typically ≤25%). In addition to the overall EC proliferative index, [3H]thymidine indices were also calculated individually for both the transduced (X-Gal–positive) and untransduced EC populations.
Heat Inactivation of Av1LacZ4
To achieve low temperature heat inactivation,16 aliquots of virus were incubated at 45°C for 15 minutes, cooled at 4°C for 10 minutes, and then used immediately. Cos-7, a simian virus 40–transformed African green monkey kidney cell line, and 293 cells were grown in improved MEM supplemented with 10% fetal bovine serum, 1 mmol/L glutamine, 100 U/mL penicillin, 100 μg/mL streptomycin, and 0.25 μg/mL amphotericin (all from Biofluids). Experiments in Cos-7 and 293 cells were carried out in 24- and 96-well tissue culture plates, respectively.
Experiments to determine the effect of heat inactivation on the ability of Av1LacZ4 to mediate endosomal lysis and plasmid-mediated transfection were carried out in Cos-7 cells. Cos-7 cells (1×105 cells per well) were washed with Opti-MEM1 (GIBCO/BRL) and then further incubated in the same medium (225 μL per well) for 2 hours. pCMVLuc (5.0 μg diluted to a volume of 20 μL with Opti-MEM1) was added to each well, followed immediately by the addition of Av1LacZ4 (2.5×107 pfu diluted to 25 μL in Opti-MEM1). After 2 hours at 37°C, an additional 240 μL of Opti-MEM1 was added to each well. Twenty-two hours later, the medium was changed to improved MEM containing 10% calf serum and 1 mmol/L glutamine. After an additional 24-hour incubation, the cells were washed twice with PBS, lysed, and assayed for luciferase activity as described above. Experiments to determine the effect of heat inactivation on the ability of Av1LacZ4 to transfer a functional β-gal gene were carried out in 293 cells. 293 cells (3×104 cells per well) were infected with Av1LacZ4 at mutiplicities of infection of 10, 100, and 1000 in a volume of 100 μL. Twenty-four hours later, the cells were washed with PBS, fixed, and stained with X-Gal as described above.
For each group of vessels receiving the same treatment in individual experiments, mean values were calculated from the values for the individual vessels constituting the group (eg, total ECs, total and percentage X-Gal–positive ECs, X-Gal–positive smooth muscle cells, BrdU-positive ECs, BrdU-positive and total smooth muscle cells, and [3H]thymidine-positive ECs). Means of experimental groups were compared with means of control groups, and differences were evaluated for significance by Student’s unpaired t test. Results are reported as mean±SD unless indicated otherwise. Not all possible pairwise tests were performed; differences were considered to be significant at P<.05.
In Vivo Gene Transfer and Expression of β-gal
Seventy-one rat carotid arteries were incubated in vivo with solutions containing the Av1LacZ4 vector. Fifty-two vessels were fixed, stained with X-Gal, and then sectioned and used for histological studies; 17 vessels were homogenized and used for β-gal activity studies; and 2 vessels were fixed and then processed en bloc for scanning electron microscopy to evaluate intimal integrity. Several of the vessels that were histologically sectioned were used for more than one assay; eg, recombinant gene expression was measured (by counting blue nuclei) and proliferative rate was determined (by counting BrdU-stained cells) on sections from the same vessel. For this reason, the total n value for assays on Av1LacZ4-infused vessels was >71. Twenty-four vessels were infused with control solutions, as detailed further below.
Efficiency and Specificity of EC Gene Transfer
To determine both the efficiency of EC gene transfer and the percentage of transduced cells that were ECs, 20 vessels were transduced at concentrations of Av1LacZ4 ranging from 1×109 to 2×1011 pfu/mL, fixed, stained with X-Gal, and sectioned. Vessels were exposed to virus at concentrations of 1×109 pfu/mL (n=2), 2×109 pfu/mL (n=3), 1×1010 pfu/mL (n=2), 2×1010 pfu/mL (n=3), 4×1010 pfu/mL (n=2), 5×1010 pfu/mL (n=2), 1×1011 pfu/mL (n=3), and 2×1011 pfu/mL (n=3). Controls (n=3) received the viral diluent only.
Three days after transduction, arteries were harvested, stained with X-Gal, and sectioned. The vast majority of X-Gal–positive cells seen in tissue sections were identified as ECs (see “Materials and Methods”). The number of transduced ECs was determined by counting X-Gal–positive ECs in cross sections of vessels (Fig 1⇓). Recombinant gene expression was found in those vessels exposed to concentrations of Av1LacZ4 from 2×109 to 1×1011 pfu/mL. Control vessels and vessels exposed to 1×109 pfu/mL Av1LacZ4 showed no X-Gal–stained cells; vessels exposed to 2×109 pfu/mL Av1LacZ4 showed only very few stained cells (mean, three per section; range, one to six). Infusion of Av1LacZ4 in the range of 1×1010 to 1×1011 pfu/mL resulted in recombinant gene expression within 42±20 (range, 19 to 91) ECs per cross section. More specifically, the mean number of X-Gal–stained ECs at 1×1010 pfu/mL was 31 (range, 29 to 32) per cross section. At 2×1010 pfu/mL, 31 (range, 19 to 38) X-Gal–stained ECs per section were present; at 4×1010 pfu/mL, 50 (range, 36 to 64) X-Gal–stained ECs per section were present; at 5×1010 pfu/mL, 31 (range, 25 to 37) X-Gal–stained ECs per section were present; and at 1×1011 pfu/mL, 62 (range, 36 to 91) X-Gal–stained ECs per section were present. This range of total transduced ECs per section represents recombinant gene expression in 34±15% of total luminal ECs (mean±SD of all vessels transduced with Av1LacZ4 concentrations ranging from 1×1010 to 1×1011 pfu/mL, n=12). Strikingly, exposure to Av1LacZ4 at the highest concentration (2×1011 pfu/mL, undiluted virus stock) resulted in complete absence of X-Gal staining both macroscopically and in histological vessel cross sections (n=3). This unanticipated finding was supported by a concomitant lack of β-gal activity in detergent extracts of two additional vessels transduced at 2×1011 pfu/mL (see below).
Although the vast majority of X-Gal–positive cells in vessels exposed to Av1LacZ4 were identified as ECs (Fig 2A⇓), occasional X-Gal–positive cells were identified as smooth muscle cells (see “Materials and Methods”; Fig 2B⇓). The relation between virus concentration and EC specificity of transduction was determined by counting total transduced cells as well as transduced ECs in sections taken from vessels transduced with viral solutions containing 2×109 to 1×1011 pfu/mL (Fig 3⇓). With exposure to Av1LacZ4 at 2×109 pfu/mL, 90% of X-Gal–stained cells were ECs (n=3; range, 75% to 100%). At 1×1010 pfu/mL, 96% of X-Gal–stained cells were ECs (n=2; range, 95% to 96%); at 2×1010 pfu/mL, 95% of X-Gal–stained cells were ECs (n=3; range, 92% to 97%); at 4×1010 pfu/mL, 95% of X-Gal–stained cells were ECs (n=2; range, 94% to 96%); at 5×1010 pfu/mL, 98% of X-Gal–stained cells were ECs (n=2; range, 98% to 99%); and at 1×1011 pfu/mL, 95% of X-Gal–stained cells were ECs (n=3; range, 90% to 100%). Thus, a very high degree of EC specificity (means, 90% to 98%) was maintained throughout the entire range of virus concentrations at which gene transfer occurred. No adventitial staining was observed in any of the vessels.
Because we found a relatively constant level of EC transduction over vector concentrations ranging from 1×1010 to 1×1011 pfu/mL and because our results showed a loss of recombinant gene expression at the highest vector concentration (2×1011 pfu/mL), we performed additional experiments to aid in the definition of ideal vector concentration. Arteries were exposed to Av1LacZ4 at concentrations of 1×109 pfu/mL (n=2), 2×109 pfu/mL (n=3), 4×109 pfu/mL (n=3), 1×1010 pfu/mL (n=2), 4×1010 pfu/mL (n=2), 6×1010 pfu/mL (n=1), 1×1011 pfu/mL (n=2), and 2×1011 pfu/mL (n=2). Controls (n=3) received viral diluent. Three days after transduction, arteries were harvested, and β-gal activity was measured from tissue extracts (Fig 4⇓). Extracts of vessels exposed to 1×109 and 2×109 pfu/mL had mean β-gal activities of 1.3 (range, 1.1 to 1.4) μU/μg total vessel protein and 0.8 (range, 0.5 to 1.2) μU/μg, respectively. These values were virtually identical to those of control vessels (0.8 [range, 0.4 to 1.3] μU/μg total vessel protein). Incubation of the arterial segments with higher concentrations of Av1LacZ4 in the range of 4×109 to 1×1011 pfu/mL resulted in an increase in β-gal activity. Mean β-gal activity at 4×109 pfu/mL was 7.9 (range, 6.4 to 10.4) μU/μg total vessel protein; at 1×1010 pfu/mL, activity was 12.1 (range, 11.5 to 12.7) μU/μg total vessel protein; at 4×1010 pfu/mL, activity was 8.4 (range, 7.4 to 9.3) μU/μg total vessel protein; at 6×1010 pfu/mL, activity was 21.1 μU/μg total vessel protein (n=1); and at 1×1011 pfu/mL, activity was 12.5 (range, 8.5 to 16.5) μU/μg total vessel protein. Exposure to Av1LacZ4 at a concentration of 2×1011 pfu/mL resulted in a dramatic decrease in β-gal activity in tissue extracts (1.6 [range, 1.5 to 1.7] μU/μg total vessel protein), approximately equivalent to control values. When combined with the data from X-Gal–stained vessels (Fig 1⇑), these results indicate that maximal levels of both transduced ECs and recombinant gene expression are achieved at virus concentrations (5×109 to 5×1010 pfu/mL) significantly below those maximally attainable. Therefore, we performed most of the remaining studies at a virus concentration of 4×1010 pfu/mL. The loss of recombinant gene expression with undiluted virus (2×1011 pfu/mL), as seen both in vessel extracts and with X-Gal staining (Figs 1⇑ and 4⇓), was likely due to acute toxicity of the viral infusate, as we have reported previously for injured vessels,17 and as will be discussed later.
Duration of Recombinant β-gal Gene Expression
To define further the ability of this gene transfer system to permit the study of recombinant gene expression in normal uninjured vessels, we performed a series of time-course studies. Our goals were first to examine the duration of recombinant gene expression and then to define the effects (if any) of the gene transfer protocol on the arterial phenotype during the time period of maximal recombinant gene expression. We began by studying the duration of recombinant gene expression in 20 rat carotid arteries transduced with Av1LacZ4 at a concentration of 4×1010 pfu/mL. Arteries were harvested at 3 days (n=10), 7 days (n=6), and 14 days (n=4) after gene transfer, X-Gal–stained, and sectioned. Control vessels (n=5 at each time point) were infused with viral diluent and were processed identically to the virus-infused vessels. Three days after gene transfer, 45±26 transduced ECs per histological section were present (range, 20 to 98 stained cells per section; mean, 34% of total ECs; Fig 5⇓). At 7 days, 47±10 transduced ECs per section were present (range, 37 to 66 stained cells per section; mean, 39% of total ECs); at 14 days, 20±23 transduced ECs were present (range, 4 to 53 stained cells per section; mean, 17% of total ECs). Thus, significant recombinant gene expression was detectable for up to 2 weeks after gene transfer but at a level reduced at 14 days when compared with that found at 3 to 7 days. No X-Gal staining was observed in control vessels.
Vessel Morphology After Adenovirus Gene Transfer
We next assessed whether gene transfer into uninjured arteries resulted in alterations in vessel morphology. Vessel morphology was evaluated histologically by assessing endothelial integrity, by evaluating the presence of acute medial necrosis and acute inflammatory cell infiltrates, and by examining vessels for the development of a neointima and the presence of chronic inflammatory infiltrates at a later time point (14 days).
We assessed the acute effects of gene transfer on endothelial integrity 3 days after gene transfer by using the same histological sections of rat carotid arteries that were used for the study of dose responsiveness and EC specificity (Figs 1⇑ and 3⇑, n=19). Therefore, the concentrations of Av1LacZ4 ranged from 1×109 to 2×1011 pfu/mL. Both viral diluent (n=3) and virus storage buffer (n=3) were included as controls.
Light microscopic analysis of histological sections revealed that vessels infused with either control solution and vessels infused with Av1LacZ4-containing solutions (from 1×109 to 1×1011 pfu/mL) had an intact luminal endothelial layer (Fig 6A⇓). In contrast, infusion of Av1LacZ4 at a concentration of 2×1011 pfu/mL resulted in virtually complete denudation of the endothelium (Fig 6B⇓).
The relation between virus concentration and EC number was addressed in a more quantitative manner by counting ECs in stained tissue sections. After infusion of Av1LacZ4 at a concentration of 2×1011 pfu/mL, EC number was significantly decreased to 3% of control values (3±4 cells per section versus 139±12 cells with infusion of virus storage buffer control, P<.001, Fig 7⇓). Infusion of Av1LacZ4 at concentrations below 2×1011 pfu/mL did not affect the number of luminal ECs.
To achieve a more complete assessment of the effect of gene transfer (at submaximal virus concentration) on the integrity of the luminal endothelium, we performed scanning electron microscopy of two vessels that had been exposed to 4×1010 pfu/mL Av1LacZ4. Imaging of the entire luminal surface established that the integrity of the luminal endothelium was not compromised 3 days after gene transfer (Fig 8⇓).
Because we previously observed medial necrosis and neutrophilic medial infiltrates in balloon-injured carotid arteries transduced at high virus concentrations,17 we examined sections of vessels transduced with Av1LacZ4 from 1×109 to 2×1011 pfu/mL for these features. No evidence of medial necrosis or inflammatory infiltrates was found in the examination of hematoxylin and eosin–stained sections taken at 3, 7, or 14 days after gene transfer.
We also examined histological sections of vessels of animals killed 14 days after gene transfer for additional evidence of vessel injury. Evidence for vessel injury was obtained by examination for the presence of a neointima (defined as one or more cell layers between the internal elastic lamina and the luminal endothelium). Arteries were harvested from seven rats exposed to Av1LacZ4 at 4×1010 pfu/mL and five rats exposed to viral diluent only. Only small areas of neointima were present in five of seven vessels exposed to Av1LacZ4 (15 of 28 total cross sections contained neointima); neointima was absent in the other two vessels. Among the diluent-exposed vessels, two of five vessels (5 of 20 total cross sections) contained neointima. When present, the neointima generally encompassed <20% of the luminal circumference and in all cases was limited to one to two cell layers in thickness (Fig 9⇓). Thus, little evidence of vessel injury could be found 14 days after virion infusion. We conclude from these morphological studies that significant vessel toxicity, manifested by endothelial denudation, results from exposure to highly concentrated virus. At lower virus concentrations, there is only very slight evidence of vascular injury at times up to 14 days after gene transfer.
To determine if the loss of ECs with exposure to virus stock was specific for Av1LacZ4, we investigated the effects of AdHS-1.2 infusion. After infusion of undiluted AdHS-1.2 stock (5×1011 pfu/mL, n=6), EC number was significantly decreased to 37% of control values (51±21 cells per section versus 139±12 cells with infusion of virus storage buffer control, P<.001). Infusion of AdHS-1.2 at 1×1011 pfu/mL resulted in a smaller decrease in luminal ECs (114±11 cells per section, n=2, P=.10).
Infusion of Heat-Inactivated Adenovirus
Heat inactivation of adenovirus at 45°C blocks endosome disruption but not virus binding or internalization; the result is a failure to achieve recombinant gene expression in cells transduced with heat-inactivated adenovirus vectors.16 Aliquots of Av1LacZ4 were heat-inactivated and tested both for their ability to enhance plasmid-mediated gene transfer and their ability to transfer a functional β-gal gene. Heat inactivation resulted in a 270-fold decrease in the ability of Av1LacZ4 to enhance transfer of a luciferase-expressing plasmid (pCMVLuc) into Cos-7 cells and a 4 log decrease in the efficiency of β-gal gene transfer into 293 cells (data not shown). We infused three arteries with aliquots of heat-inactivated Av1LacZ4 at a concentration of 2×1011 pfu/mL. Infusion resulted in no detectable β-gal expression (by X-Gal stain) yet a significant decrease in EC number compared with arteries infused with virus storage buffer control (2±1 versus 139±12 cells per section, P<.001).
EC and Smooth Muscle Cell Proliferation After Exposure to Av1LacZ4
To characterize further the effect of gene transfer on the arterial phenotype, we measured the proliferative rate of both ECs and smooth muscle cells in transduced arterial segments. Twenty-one rat carotid arteries were exposed to Av1LacZ4 at a concentration of 4×1010 pfu/mL and were harvested at 3 (n=8), 7 (n=6), and 14 (n=7) days after transduction. For each of the three time points, we also examined five control vessels that were infused with viral diluent alone. Vessels exposed to Av1LacZ4 and harvested 3 days after gene transfer had an EC BrdU proliferation index of 18.1±7.3%, nearly 10-fold greater than that measured in the right carotid arteries (2±1.1%, P<.001, n=5). However, this elevated proliferation index was not significantly different from that calculated for left carotid arteries receiving diluent only (13.5±8.1%, P=.31, Fig 10⇓). Seven days after gene transfer, the EC BrdU index for virus-treated vessels was 14.5±13.1% versus 3.5±1.5% for diluent-treated vessels. This 4-fold difference between the indices was not significant (P=.12). Fourteen days after gene transfer, the EC BrdU index of Av1LacZ4-exposed vessels remained elevated at 17.7±9.8%, a significant 13-fold increase over the value of 1.4±1.6% for diluent-treated vessels (P=.005). Therefore, adenovirus transduction resulted in a significant increase in EC proliferation at 14 days.
The same anti-BrdU–stained cross sections from arteries exposed to Av1LacZ4 and harvested 3 (n=8) and 14 (n=7) days after gene transfer were used to analyze the effect of the gene transfer protocol on medial smooth muscle cell proliferation. Three days after gene transfer, the proliferative index of medial smooth muscle cells in virus-exposed vessels was essentially equal to that of vessels infused with viral diluent (9.1±8.0% versus 7.4±4.6%, P=.68). Fourteen days after gene transfer, the BrdU index of medial smooth muscle cells in virus-exposed vessels decreased but remained essentially identical to the index of diluent-exposed vessels (0.9±0.8% versus 0.6±0.6%, P=.40). The proliferative index of medial smooth muscle cells in the unmanipulated right carotid arteries was 0.3±0.3% at 3 days and 0.4±0.4% at 14 days. Therefore, medial smooth muscle cell proliferative rates increased early after the surgical procedure but were not specifically affected by gene transfer.
Proliferation of the Transduced EC Population
Because the EC proliferative index was increased in transduced vessels, we sought to determine whether the proliferating cells were specifically transduced or untransduced cells. Combined [3H]thymidine autoradiography and X-Gal staining was performed on sections of four rat carotid arteries that had been exposed to Av1LacZ4 at a concentration of 4×1010 pfu/mL and harvested 3 days after transduction (Fig 11⇓). The [3H]thymidine index for the X-Gal–positive (transduced) population of ECs was 11.3±3.4%, essentially identical to that of the X-Gal–negative (untransduced) EC population (15.3±4.7%, P=.21). Therefore, the transduced EC population includes mitotically active cells, with no difference in mitotic activity specifically attributable to gene transfer.
In Vivo Gene Transfer and Expression of Luciferase
We considered the possibility that certain of our results, such as the plateau of recombinant gene expression and the range of virus concentration over which this plateau occurred, might be features that were unique to the Av1LacZ4 virus. Therefore, we carried out additional experiments with the AdSV40Luc vector, which has both a different promoter and a different transgene than Av1LacZ4. Seventeen rat carotid arteries were exposed to AdSV40Luc, then removed after 3 days, and assayed for recombinant luciferase activity. The arteries were exposed to diluted AdSV40Luc at concentrations of 8×107 pfu/mL (n=2), 4×108 pfu/mL (n=3), 2×109 pfu/mL (n=3), and 1×1010 pfu/mL (n=4) and to undiluted virus stock at 5×1010 pfu/mL (n=5). Controls (n=3) received viral diluent. Three days after gene transfer, arteries were harvested, and luciferase activity was quantified in tissue extracts (Fig 12⇓). Luciferase activity was undetectable in extracts of vessels exposed to control solution (<1 RLU/μg total vessel protein). Extracts of arteries exposed to AdSV40Luc at 8×107 and 4×108 pfu/mL had low levels of luciferase activity, mean values of 300 (range, 120 to 470) and 220 (range, 49 to 520) RLU/μg total vessel protein, respectively. Exposure of vessels to higher concentrations of AdSV40Luc resulted in an increase in luciferase activity. Mean luciferase activity at 2×1010 pfu/mL was 3100 (range, 1500 to 5100) RLU/μg total vessel protein; at 1×1010 pfu/mL, activity was 5300 (range, 1400 to 15000) RLU/μg total vessel protein; and at 5×1010 pfu/mL, activity was 3600 (range, 720 to 9300) RLU/μg total vessel protein. Thus, a plateau of recombinant gene expression as virus concentration approaches ≈1010 pfu/mL appears to be a constant feature of EC-specific gene transfer in the rat carotid artery. Whether this same concentration of virus achieves a plateau of recombinant gene expression in other species will require specific experimentation.
The present study demonstrates the validation of a model of in vivo EC-specific arterial gene transfer with adenovirus vectors. Specifically, the major findings are as follows: (1) Recombinant gene expression is ≈95% EC specific over a wide range of virus concentrations. (2) Approximately 35% of luminal ECs express the recombinant gene. (3) The level of recombinant gene expression is easily quantified in vessel extracts. (4) Peak levels of gene expression can be achieved at submaximal virus concentrations, permitting conservation of virus stocks and avoiding acute toxicity. (5) At submaximal virus concentrations, the phenotype of a transduced rat carotid artery is minimally altered by gene transfer, with increased EC (but not smooth muscle cell) proliferation, only slight neointimal formation, and no evidence of EC loss or inflammatory infiltrates. (6) Toxicity resulting in luminal EC loss occurs only with exposure to very high concentrations of adenovirus.
Several groups have reported animal models of high-efficiency adenovirus vector–mediated gene transfer into peripheral arteries. Lemarchand et al8 described what appeared to be largely endothelial gene transfer in sheep carotid arteries, but the extent of EC specificity was not reported. Moreover, the level of gene transfer was not quantified or optimized, and no data were presented concerning the phenotype of the transduced arteries. In two previous studies from our laboratory, adenovirus-mediated gene transfer into sheep carotid arteries resulted in significant recombinant gene expression in both the luminal endothelium and the adventitia, with access of virions to the adventitia through branches of vasa vasorum that bypass the vascular media.9 10 Interestingly, Guzman et al,18 using an uninjured rat carotid artery model similar to that in the present study, found the endothelium to be largely resistant to adenovirus-mediated gene transfer. Therefore, despite the existence of several previous studies of adenovirus-mediated gene transfer into normal arteries,8 9 10 18 19 20 the present quantitative description and characterization of a model of highly EC-specific gene transfer is novel.
The EC specificity that we observed likely results from the properties of both the adenovirus and the rat carotid artery. Our group9 and others19 20 have found that the intact endothelium of elastic arteries is a barrier to the penetration of adenovirus vectors. If the endothelium is damaged, smooth muscle cell transduction occurs.9 20 Moreover, in arterial segments with branches, high levels of gene transfer are found in the adventitia.9 It is notable that the rat common carotid artery has no branches21 and that the endothelium remains intact despite manipulation of the vessel at the time of gene transfer (Figs 2⇑ and 8⇑). As a result, surgical isolation of the common carotid artery followed by infusion of adenovirus vectors results in highly EC-specific gene transfer. Rat carotid smooth muscle cells are clearly susceptible to high-efficiency in vivo gene transfer with the Av1LacZ4 vector if the endothelium is removed.17 The rare transduced smooth muscle cell seen in the present study (Fig 2B⇑) likely results from a very low level of nondenuding endothelial injury caused by surgical manipulation.
We found significant vascular toxicity at high virus titers, manifested by endothelial denudation and loss of recombinant gene expression. This observation is similar to our findings in a model of gene transfer into balloon-injured rat arteries17 and again suggests that the ability of adenovirus vectors to achieve high levels of recombinant gene expression is limited by associated toxicity. The finding of significant EC loss after infusion of both Av1LacZ4 and AdHS-1.2 implies that toxicity is a general effect of highly concentrated adenovirus. That heat-inactivated Av1LacZ4 also caused significant cell loss despite loss of an ability to mediate endosomal lysis implicates adenovirus binding and/or internalization in the observed toxicity. Relevant to this observation, adenovirus penton-base protein is known to cause adherent cells in culture to “round up” and detach,22 an effect mediated through an RGD amino acid sequence in the penton base that is also found in a number of extracellular matrix molecules. Normally, the RGD sequence of adhesion molecules mediates cell-matrix adhesion via binding to cell surface integrins.23 It is possible that RGD-mediated cell detachment (as a result of competitive displacement of adhesion molecules by adenovirus) is causing the EC loss that we observed; however, further experiments will be necessary to confirm this. We also considered the possibility that something copurifying with the adenovirus might be the toxic agent. However, the requirement that this toxic agent be of equivalent buoyant density to adenovirus and nondialyzable argues against this possibility. Finally, the existence of a plateau of recombinant gene expression at virus concentrations at which toxicity was not found indicates that more “efficient” (ie, fewer virions, same expression) gene transfer can be achieved at submaximal virus concentrations. Virus delivery at concentrations along this plateau will minimize virus-related vascular toxicity and will permit conservation of adenovirus stocks.
To define the biological substrate on which we may study recombinant gene expression in ECs with this animal model, we investigated several aspects of the phenotype of the transduced arteries. Both the endothelium and the media were intact and free from inflammatory infiltrates 3 days after gene transfer. However, the proliferative indices of both transduced and untransduced ECs as well as that of underlying medial smooth muscle cells were somewhat elevated. This increase in proliferation at 3 days is apparently a consequence of the surgical procedure rather than of gene transfer per se (Fig 10⇑). At 14 days after gene transfer, inflammatory infiltrates are still absent, smooth muscle cell proliferation returns to baseline, but EC proliferation remains elevated, at this time as a direct result of the gene transfer protocol. Thus, at both 3 and 14 days, although recombinant gene expression is not studied in a uniformly quiescent normal artery, the large majority (82%) of ECs are not proliferating. The modestly elevated endothelial proliferative rate at 14 days (a time at which recombinant gene expression is declining) may be due to immune-mediated rejection and replacement of transduced cells by neighboring untransduced cells, as described elsewhere for adenovirus-transduced hepatocytes.24 Use of second generation adenovirus vectors might permit a further decrease in EC proliferation at 14 days and would appear to be particularly useful if prolongation of recombinant gene expression were desired.25 26 Prolongation of gene expression was not, however, an objective of the present study; as shown by others in both the myocardium5 and peripheral nervous system,6 prolonged recombinant gene expression is not necessary to support the utility of a somatic gene transfer system to study mechanisms of gene regulation in vivo.
Despite the potential limitation related to increased cellular proliferation and the small amount of gene transfer into smooth muscle cells of the vascular media (representing ≈5% of total transduced cells), this animal model of EC-specific gene transfer is likely to be useful for the study of in vivo EC biology. Cell-type–specific targeting of recombinant gene expression, primarily through germ-line transmission of transgenes in mice, has yielded significant biological insights elsewhere in the cardiovascular system.27 Extensive efforts to use germ-line targeting to achieve EC-specific expression in adult animals28 29 have not yet been successful. Even if specific germ-line targeting of endothelium is achieved, the present somatic gene-transfer system will retain significant advantages. First, the present system is site specific, with the potential to study local rather than systemic effects of gene expression and the possibility to use other vessels in the same animal as controls. Second, the system offers the possibility of testing individual (or multiple) genetic constructs without generating separate lines of transgenic animals.
In summary, we have described, characterized, and optimized an animal model of EC-specific recombinant gene expression. Future experiments will define the utility of this system in the study of mechanisms of endothelial gene regulation30 and signal transduction in vivo.
Selected Abbreviations and Acronyms
|RLU||=||relative light units|
Dr Schulick was funded by the Pharmacology Research Associates Training program of the National Institute of General Medical Sciences, National Institutes of Health. We thank Dr Jeffrey J. Rade for providing the AdHS-1.2 vector and Eleonora Dorfman for help in the preparation of virus stocks.
- Received December 23, 1994.
- Accepted June 6, 1995.
- © 1995 American Heart Association, Inc.
Gimbrone MA. Vascular Endothelium in Hemostasis and Thrombosis. Edinburgh, Scotland: Churchill Livingstone; 1986.
Yanagisawa M, Kurihara H, Kimura S, Tomobe Y, Kobayashi M, Mitsui Y, Yazaki Y, Goto K, Masaki T. A novel potent vasoconstrictor peptide produced by vascular endothelial cells. Nature. 1994;332:411-415.
Ignarro LJ, Buga GM, Wood KS, Byrns RE, Chaudhuri G. Endothelium-derived relaxing factor produced and released from artery and vein is nitric oxide. Proc Natl Acad Sci U S A. 1994;84:9265-9269.
Aoyagi T, Izumo T. Mapping of the pressure response element of the c-fos gene by direct DNA injection into beating hearts. J Biol Chem. 1993;268:27176-27179.
Bessereau JL, Stratford-Perricaudet LD, Piette J, Le Poupon C, Changeux JP. In vivo and in vitro analysis of electrical activity-dependent expression of muscle acetylcholine receptor genes using adenovirus. Proc Natl Acad Sci U S A. 1994;91:1304-1308.
Abdellatif M, MacLellan WR, Schneider MD. p21 Ras as a governor of global gene expression. J Biol Chem. 1994;269:15423-15426.
Lemarchand P, Jones M, Yamada I, Crystal RG. In vivo gene transfer and expression in normal uninjured blood vessels using replication-deficient recombinant adenovirus vectors. Circ Res. 1993;72:1132-1138.
Rome JJ, Shayani V, Flugelman MY, Newman KD, Farb A, Virmani R, Dichek DA. Anatomic barriers influence the distribution of in vivo gene transfer into the arterial wall. Arterioscler Thromb. 1994;14:148-161.
Lee SW, Trapnell BC, Rade JJ, Virmani R, Dichek DA. In vivo adenoviral vector–mediated gene transfer into balloon-injured rat carotid arteries. Circ Res. 1993;73:797-807.
Graham FL, Prevec L. Manipulation of adenovirus vectors. In: Murray EJ, ed. Methods in Molecular Biology: Gene Transfer and Expression Protocols. Clifton, NJ: The Humana Press Inc; 1991;7:109-128.
Knapp A, Degenhardt T, Dodt J. Hirudisins: hirudin-derived thrombin inhibitors with disintegrin activity. J Biol Chem. 1992;267:24230-24234.
Seth P, Rosenfeld M, Higginbotham J, Crystal RG. Mechanism of enhancement of DNA expression consequent to cointernalization of a replication-deficient adenovirus and unmodified plasmid DNA. J Virol. 1994;68:933-940.
Schulick AH, Newman KD, Virmani R, Dichek DA. In vivo gene transfer into injured carotid arteries: optimization and evaluation of acute toxicity. Circulation. 1995;91:2407-2414.
Guzman RJ, Lemarchand P, Crystal RG, Epstein SE, Finkel T. Efficient and selective adenovirus-mediated gene transfer into vascular neointima. Circulation. 1993;88:2838-2848.
Willard JE, Landau C, Glamann DB, Burns D, Jessen ME, Pirwitz MJ, Gerard RD, Meidell RS. Genetic modification of the vessel wall: comparison of surgical and catheter-based techniques for delivery of recombinant adenovirus. Circulation. 1994;89:2190-2197.
Steg PG, Feldman LJ, Scoazec JY, Tahlil O, Barry JJ, Boulechfar S, Ragot T, Isner JM, Perricaudet M. Arterial gene transfer to rabbit endothelial and smooth muscle cells using percutaneous delivery of an adenoviral vector. Circulation. 1994;90:1648-1656.
Bai M, Harfe B, Freimuth P. Mutations that alter an arg-gly-asp (RGD) sequence in the adenovirus type 2 penton base protein abolish its cell-rounding activity and delay virus reproduction in flat cells. J Virol. 1993;67:5198-5205.
Yang Y, Nunes FA, Berencsi K, Furth EE, Gonczol E, Wilson JM. Cellular immunity to viral antigens limits E1-deleted adenoviruses for gene therapy. Proc Natl Acad Sci U S A. 1994;91:4407-4411.
Engelhardt JF, Ye X, Doranz B, Wilson JM. Ablation of E2A in recombinant adenoviruses improves transgene persistence and decreases inflammatory response in mouse liver. Proc Natl Acad Sci U S A. 1994;91:6196-6200.
Hunter JJ, Zhu H, Lee KJ, Kubalak S, Chien KR. Targeting gene expression to specific cardiovascular cell types in transgenic mice. Hypertension. 1993;22:608-617.
Hilkert RJ, de la Monte S, Lee M, Yun JS, Wagner TE, Quertermous T. Transgene expression directed by the endothelin-1 promoter is cell-type-specific but integration-site dependent. J Vasc Med Biol. 1993;4:138-147.
Harats D, Kurihara H, Belloni P, Oakley H, Ziober A, Ackley D, Cain G, Kurihara Y, Lawn R, Sigal E. Targeting gene expression to the vascular wall in transgenic mice using the murine preproendothelin-1 promoter. J Clin Invest. 1995;95:1335-1344.
Dong G, Dichek DA. Regulated gene expression in endothelial cells following adenoviral gene transfer. Circulation. 1994;90(suppl I, pt 2):I-140. Abstract.