- Email: [email protected]
Injury induced neointima formation and its inhibition by retrovirus-mediated transfer of nitride oxide synthase gene in an in-vitro human saphenous vein culture model
Atherosclerosis 161 (2002) 113– 122 www.elsevier.com/locate/atherosclerosis
Injury induced neointima formation and its inhibition by retrovirus-mediated transfer of nitride oxide synthase gene in an in-vitro human saphenous vein culture model Hong Yu a,*, S. Ram Kumar a, Lili Tang a, Thomas T. Terramani a, Vincent L. Rowe a, Ying Wang a, Rahul A. Nathwani a, Fred A. Weaver a, Darwin Eton b a
Department of Surgery, Vascular Di6ision, Keck School of Medicine of the Uni6ersity of Southern California, 2025 Zonal A6enue, RMR 505, Los Angeles, CA 90089, USA b Department of Surgery, Uni6ersity of Miami, Miami, FL 33156, USA Received 9 January 2001; received in revised form 31 January 2001; accepted 10 July 2001
Abstract Human saphenous veins were cultured to characterize neointima formation and feasibility of gene transfer to inhibit the intimal proliferative response to injury. Mechanical injury was introduced by abrading the luminal surface of the vein patch with a sterile cotton bud. Both injured and non-injured vein patches were cultured and transduced with retroviral vectors carrying marker or therapeutic genes. After a 14-day culture, the thickness of the intimal layer of non-injured vein patches reached 90 9 28 mm at the edge and 61 922 mm at the center (n=29) from the original 22 9 12 mm at harvest (n =6, P= 0.02). Mechanical injury to the intimal surface prior to culture resulted in an exaggerated proliferative response. The intimal thickness of injured vein patches increased from 3.4 91 mm right after injury to 128 9 23 mm (n= 12, P B 0.001) at the edge after 14-day culture. Genes were transduced efficiently into a luminal layer of cultured veins using a pseudotyped murine leukemia viral vector. Transduction of gene encoding nitric oxide synthase resulted in reduction of neointima formation to 33 9 7 mm (n = 12) at the edge after 14-day culture compared to 90 mm (PB0.01) seen in untransduced non-injured vein patches. Marker gene transduction did not alter intimal proliferative response or its immunohistochemical profile. The data suggest that cultured vein can be used as a model for studying the effects of injury to blood vessels and to evaluate the effects of candidate therapeutic genes. © 2002 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Vein culture; Neointima; Gene transfer; Nitride oxide synthase; Endothelial cell; Retroviral vector; Injury
1. Introduction Intimal response to injury continues to be an unresolved obstacle to long term graft patency. Organ culture offers a simplified in vitro system to study the responses of vascular tissues in vivo while maintaining phenotypic and three-dimensional architecture allowing spatial relationships of cell types that are potentially interactive during tissue injury and growth regulation [1]. Numerous ex vivo vascular tissue culture systems * Corresponding author. Tel.: + 1-323-442-3792; fax: +1-323-4423164. E-mail address: [email protected] (H. Yu).
utilizing either arteries or veins have been evaluated in the last two decades as in vitro models to study vascular responses to various external stimuli including injury [2–4]. Veins appear to be the more attractive targets because they are more readily available and thin enough to permit diffusion of nutrients and metabolites. Culture of human saphenous vein was first described in 1990 by Soyombo et al. [5], where she suggested that it could be used as an in vitro model to study neointima formation in vein grafts. Varty et al. showed that the neointima developed in cultured vein segments is similar to the intimal hyperplasia in stenosed vein grafts [6]. Few have studied the effect of injury on neointima formation [7]. Gene transfer has
0021-9150/02/$ - see front matter © 2002 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 0 2 1 - 9 1 5 0 ( 0 1 ) 0 0 6 2 5 - 6
1 114
H. Yu et al. / Atherosclerosis 161 (2002) 113–122
been reported in cultured arteries using adenoviral vectors [8,9] and the transfer of the human TIMP-1 gene has been shown to inhibit neointimal formation in a cultured vein [10]. In this study, an in vitro injured vein culture model was established by introducing mechanical injury to the cultured vein patches, that may be used to mimic intimal proliferation of injured blood vessels in vivo and to evaluate the effect of gene transfer on neointimal formation. Since nitric oxide synthase (NOS) has been shown to inhibit neointima formation in vivo [11,12] through increasing the nitric oxide production, the effect of NOS gene transduction into the cultured vein was studied. We found that the transfer of NOS gene can inhibit the neointima formation in the cultured vein, which demonstrates that the vein culture model can be used to study the effect of gene transfer on inhibition of neointima.
2. Materials and methods
2.1. Culture of human saphenous 6ein Excreta human saphenous veins (total 35 samples) were obtained from patients undergoing vascular or coronary artery bypass surgery at the University of Southern California Hospitals, which was approved by the Institutional Review Board (IRB No. 969042). Harvested veins were gently flushed with heparinized saline and peri-adventitial tissue was debrided in the operating room. Meticulous handling reduced unnecessary trauma. Details of saphenous vein culture are as previously described [13]. The veins were pinned down to hardened Sylgard medium on the culture plate surface. Mechanical injury was induced by abrading the luminal surface of the vein patch ten times with a sterile cotton bud.
2.2. Histological analysis To perform differential staining of the elastic lamina, vein patches were fixed overnight in 10% formaldehyde, embedded in paraffin and cut into 4 mm tissue sections. Slides were de-paraffinized and stained using the standard Verhoeff’s elastic staining (American MasterTech Scientific, CA) procedure. Arbitrarily, 10% surface of the vein patch in outer area was defined as the edge and 90% of the surface in the inner area as the center. Intimal thickness was measured from the elastic layer at ten random points over the center and both edges of the patches. Linemorphometric analyses were performed on computerassisted images at 40× magnification using commercial software OPTIMAS (Optimas Corporation, WA).
For immunohistochemical staining of paraffin sections, de-paraffinized sections (4 mm) of vein patches were antigen retrieved by pressure-cooking the slides in Antigen Retrieval Citra 10× solution (BioG Genex) in a microwave oven at high power for 15 min followed by 40% power setting for 15 min. The whole staining was performed at room temperature. After washing with PBS and 20 min incubation with 20% horse serum (Ventana Medical Systems, AZ) to block nonspecific reaction, sections were incubated overnight with various mouse anti-human IgG primary antibody (1:500 dilution, Dako Laboratories, CA), as detailed in Section 3. The antibody was washed off with PBS and sections incubated for 45 min with horse anti-mouse IgG biotinylated secondary antibody (1:20 dilution, Vector Laboratories, CA) followed with PBS washing. After blocking the endogenous peroxidase activity by 20-min incubation with 3% hydrogen peroxide (Sigma, MO), the sections were then incubated with the avidin–biotinylated peroxidase complex (Vector Laboratories) for 30 min and rinsed with PBS. Bound antibody was detected by adding 3-3%-diaminobenzidine substrate (Ventana Medical Systems) for 20 min. The sections were washed with tap water, counterstained with Meyer’s hematoxylin (Sigma) for 1 min, dehydrated in graded alcohol solutions followed by xylene and mounted using Permount medium. Scanning and transmission electron microscopy was performed as described [14].
2.3. Retro6iral transduction VSV-G pseudotyped MuLV (VSV-G/MuLV) vectors were generated as described previously [15]. Vector plasmids pG1nBgSvNa, pLtSN [16] and pLCNSN [17] carrying genes encoding for nuclear-localized galactosidase (-Gal), tissue type plasminogen activator (tPA) and human endothelium-specific constitutive nitric oxide synthase (NOS), respectively, were provided by Genetic Therapy Inc./Novartis, Dr David Dichek (University of California at San Francisco), and Dr A.W.Clowes (University of Washington), respectively. The viral titers were 3–10×106 cfu/ml. Vein patches were cultured 7 days prior to transduction as described [15] with VSV-G/MuLV vectors mentioned above. The transduced vein was continually cultured for an additional 7 days. To study the effect of duration of transduction, whole vein segments cultured for 7 days were immersed in viral supernatants for 15, 30, 60 and 120 min, after which they were rinsed with PBS and cultured for another 7 days. They were then stained with X-Gal [15]. The number of stained cells at the edge and center of the paraffin sections was counted over five random fields by an observer blinded to the procedure.
H. Yu et al. / Atherosclerosis 161 (2002) 113–122 Table 1 Intimal proliferation in cultured vein patches Vein patch
Thickness of the intimal layer (mm)
P value* (n)
Edge
Center
Non-cultured Uninjured Injured
22 911 3.4 9 1.0
20 911 3.9 90.5
0.26 (6) 0.78 (4)
Cultured Uninjured Injured
90 918 128 923
70 914 90 924
0.021 (29) 0.00006 (12)
0.01
0.03
P value†
* Comparison between edge and center of the same vein patch. Comparison between corresponding sites on injured and uninjured patches. Excreta human saphenous vein patches were cultured for 14 days after scraping the intimal layer with a cotton swab (injured) and studied against control patches from the same vein cultured without scraping (uninjured). Non-cultured patches belong to a different pool of patients. The thickness of the intimal layer was measured after differential elastic staining of paraffin sections at 40× magnification. Values are expressed as mean 9S.E. of mean and the number of veins studied are shown in the bracket (n). Variables were compared using Student’s t-test. †
115
seen in the intimal layer of human saphenous vein segments prior to culture (Fig. 1a). After 14 days culture in a dish, the surface of vein patches was covered with fibrous cells resembling smooth muscle cells (Fig. 1b). A new layer covering the intimal surface in the cultured vein patches was approximately three times thicker (61–80 mm, Fig. 1f,g) at all sites than non-cultured patches (21 mm) (Fig. 1(e); Table 1). To mimic mechanical injury during surgery, the luminal surface of vein patches was scraped with a cotton swab. The induced injury resulted in loss of endothelial covering revealed by electron microscopy (Fig. 1d), associated with a reduction in the thickness of the intimal surface from 21.1 to 3.64 mm (Table 1). However, on culture, injured vein patches had a greater intimal proliferative response. Cells could be seen sprouting out of the patches after 3 days in culture and a thicker intimal layer, 128 mm at the edge and 90 mm at the center, was observed (n= 12, Table 1, Fig. 1h). While the thickness of the intima on non-cultured vein patches was similar at all points on the patch, the intimal layer of the cultured patch was 32% (uninjured) and 42% (injured) thicker at the edge (Fig. 1f) than at the center (Fig. 1g, PB0.03, n=12, Table 1).
2.4. Determination of tPA concentration After transduction with VSV-G/MuLV/LtSN, vein patches were cultured for an additional 7 days and allowed to continue in culture in serum-free William’s medium for a further 48 h, by then the supernatants were collected. Control vein patches were either transduced with lacZ gene or not transduced at all. tPA production by vein patches at harvest was determined by immersing injured or uninjured vein segments directly in serum-free medium for 2 days. Quantitation of tPA antigen and fibrinolytic enzyme activity was performed with TintElize and Chromolize tPA assay kits from Biopool (Ventura, CA).
2.5. Statistical analyses All values are expressed as mean9S.E. of mean. 2 Statistics were used to compare proportions and frequency distribution of variables among the different groups. Mean values for continuous variables were compared with analysis of variance and paired twotailed Student’s t-tests. Significance was attributed to a P value B 0.05. 3. Results
3.1. Vein culture and intimal proliferation The cobblestone architecture of endothelial cells was
3.2. Characterization of the cells in the new intimal layer Immunohistochemical analysis of cultured injured and uninjured vein patches showed similar monoclonal antibody staining characteristics (Table 2). The cells in the newly formed intima and in the media of cultured veins stained positive for the smooth muscle components a-actin and calponin (Fig. 2a,b). However, while the new layer stained negative for the muscle protein marker caldesmon, the medial layer was weakly positive (Fig. 2c); similarly, while the newly formed intima stained weakly positive with anti-myosin heavy chain antibody, the medial layer stained strongly positive (Fig. 2d). Prior to culture, the intimal layer of uninjured veins stained positively for endothelial cell specific markers, such as vWF (Fig. 2e) and anti-EC antibody, but after culture the newly formed intimal layer showed no staining (Fig. 2f).
3.3. Gene transfer into cultured 6ein After 7 days in culture, vein segments could be transduced with high efficiency using the VSVG pseudotyped MuLV vector carrying different genes. Transduction increased progressively with increasing duration of viral admixture and was highest at 2 h of exposure to viral supernatant (Fig. 3a). Longer durations of transduction did not alter transduction efficiency significantly (data not shown).
116
H. Yu et al. / Atherosclerosis 161 (2002) 113–122
H. Yu et al. / Atherosclerosis 161 (2002) 113–122
117
Table 2 Characterization of cells in the newly formed layer of cultured veins by immunohistochemistry Tissue
Uninjured Injured Transduced
Muscle marker antigens
Fibroblast antigen
EC marker antigens
a-Actin
Calponin
Caldesmon
Myosin heavy chain
5B5
vWF
Anti-EC
+++ +++ +++
+++ +++ +++
− − −
+ + −
++ ++ ++
− − −
− − −
Human saphenous vein explants were cultured for 14 days and paraffin sections stained for specific cell type markers. The staining pattern of vein patches cultured after scraping the luminal surface with a cotton swab (injured) and patches transduced with the lacZ gene using the VSV-G pseudotyped MuLV vector (transduced) were compared with that of control patches (uninjured). VWF, Von-Willebrand factor; EC, endothelial cell; −, negative staining; +, weakly positive staining; ++, positive staining; +++, strongly positive staining.
Transduction was seen at both luminal and adventitial surfaces of the patches. However, no penetration into deeper layers was observed (Fig. 3b). Transduction was greater at the edge (Fig. 3c) than at the center at all durations of transduction (Fig. 3d). There were 479 9 and 13 9 3 transduced cells per high power field (cells/ hpv, 200× ) at the edge and center, respectively, on the cultured patch transduced with viral vector for 2 h (n = 10, P= 0.0015). Injured vein patches could be transduced with significantly higher transduction efficiencies (Fig. 3(a), upper panel) than uninjured controls (Fig. 3(a), lower panel) at both the edge (1179 15 cells/hpv) and center (4692 cells/hpv, n = 3, PB 0.001). The immunohistochemical staining pattern on the newly formed layer of vein patches transduced with the VSV-G/MuLV vector carrying the lacZ gene was similar to that of untransduced injured and uninjured patches (Table 2).
3.4. tPA production by 6ein patches tPA production from the vein patches was studied at different stages— at harvest, after the induction of injury, after culture and after transduction (Table 3). Induction of injury to the luminal surface of harvested veins caused a fall in tPA production of vein patch (P =0.04). After 14 days in culture, tPA production from both uninjured and injured vein patches fell to negligible amounts. Transduction of tPA gene to cultured vein patches led to a 28- and 12.5-fold increase in
Table 3 Production of tPA by vein patches Vein patch
Non-cultured Cultured
Cultured/lacZ Cultured/tPA
tPA antigen (ng/ml/cm 2) Uninjured 10.7 9 0.1 Injured 4.8 90.9
0.6 90.2 1.0 90.1
0.7 90.2 0.8 9 0.1
7.9 9 1.8 27.3 9 5.9
tPA acti6ity (IU/ml/cm 2) Uninjured 24.6 91.7 Injured 10.5 9 1.0
0.3 90.2 0.3 90.1
0.3 9 0.2 0.049 0.01
9.3 92.0 33.8 910.0
tPA production and activity from freshly harvested human saphenous veins (non-cultured) were studied without (uninjured) or after (injured) induction of injury. After culture for 7 days, vein patches were transduced with VSV-G/MuLV/G1nBgSvNa (Cultured/lacZ) or VSV-G/MuLV/LtSN (Cultured/tPA) vector and the culture continued for another 7 days. Untransduced vein patches cultured for 14 days (cultured) were studied as controls. The tPA production in the collected cultured media was measured and normalized to the area of vein patch. Values are expressed as mean 9S.E. of five samples.
tPA production in injured and uninjured vein patches, respectively, with a concomitant increase in tPA activity. LacZ co-transduction did not affect tPA production or activity.
3.5. Effect of gene transduction on neointima formation The thickness of the intima layer of vein segment transduced with genes encoding ecNOS (Fig. 4c) was significantly (PB0.01) reduced compared to that of
Fig. 1. Human saphenous vein culture. (a) Cobblestone architecture of endothelial cells was shown by scanning electron micrograph on uncultured human saphenous vein explants. (b) The intimal layer of native veins was replaced by sheets of fibrous cells after a 14-day culture. (c) Transmission electron micrograph show endothelial lining of the vessel wall (arrow), N indicates the EC nuclear. (d) Loss of endothelial cells after mechanical injury to the intimal surface. The thickness of the intimal layer was shown by Verhoeff’s Von Geison staining of vein sections of uncultured (e), 14-day cultured at edge (f), 14-day cultured at center (g), injured and cultured at edge (h). The internal elastic lamina was shown by arrows. The scale bar in (e) applies to all pictures in lower panel. (L, luminal surface; I, intimal layer; M, medial layer of the vein patch.) Fig. 2. Immunohistochemical staining pattern of cultured vein patches. Paraffin sections of human saphenous vein explants cultured for 14 days were stained with antibodies for specific cell-type markers: a-actin (a), calponin (b), caldesmon (c), myosin heavy chain (d), vWF (e,f). The medial (arrow) and newly formed layers (arrowhead) of vein patches stained strongly positive for the muscle markers a-actin (a) and calponin (b). The medial layer stained weakly positive for caldesmon, while the intimal layer, negative (c). The medial layer stained strongly positive for myosin heavy chain, while the intimal layer was only weakly positive (d). While the luminal surface of non-cultured patches stained positive for the anti-vWF marker directed against endothelial cells (e), the cultured vein patches did not (f).
118
H. Yu et al. / Atherosclerosis 161 (2002) 113–122
Fig. 3. Transduction to the cultured veins. Human saphenous vein explants were cultured for 7 days, transduced with the VSV-G pseudotyped MuLV vector carrying the lacZ gene and cultured for another 7 days. Transduction efficiency increased with increasing durations of transduction (shown in minutes in (a)) and was greater in the vein segment scraped with a cotton swab prior to culture (upper panel in (a)) than the unscraped segment (lower panel in (a)). Paraffin sections of the transduced vein patches stained with X-Gal showed transduced cells (blue) at both the luminal (L) and adventitial (A) surfaces of the vein (b), but no transduction in the medial layer. There were more transduced cells at the edge of a cultured vein patch (c) than the center (d) through en face view of transduced vein by microscopy (40 × ). Blue dots represent transduced cell nuclei stained with X-Gal. Fig. 4. Effect of NOS transduction on neointima formation. Sections of (a) uncultured, (b) 14-day cultured, (c) NOS gene transduced and cultured vein segments were stained with Verhoeff’s Von Geison staining. The dark filled arrows indicate the internal elastic lamina. The line arrows indicate the new layer intima that was significantly reduced after NOS transduction (c) compared with the untransduced (b). (d) Immunohistochemical staining for ecNOS in the cultured vein patch transduced with NOS gene. Dark brown layer at the luminal surface of the vein indicates NOS positive. The scale bar in (a) applies to a –c panels; the scale in (d) is doubled.
H. Yu et al. / Atherosclerosis 161 (2002) 113–122 Table 4 Neointima formation and inhibition in an in vitro cultured vein model Position*
Edge Center
Intimal thickness (mm)† b-Gal‡
Uncultured
Cultured
21 911 (6) 20 9 11 (6)
999 34 (17) 95935 (6) 80 916 (17) 74 9 21 (6)
NOS§ 33 9 7 (12) 26 9 4 (12)
* Position signifies the site on the vein, edge or center, where the intimal thickness was measured. † Intimal thickness was measured from the stained elastic lamina of veins which were uncultured, cultured for 14 days, cultured for 7 days and then transduced with genes encoding for b-Gal, NOS or aG1 and cultured for 7 more days. All values are expressed as mean 9 S.E. of mean with the number of vein samples in parentheses. ‡ P = 0.80 and 0.70 for edge and center, respectively compared to the correspondent cultured. b-Gal transduction did not reduce intimal proliferation showing a lack of effect of transduction itself. § PB0.01 for both edge and center compared to the correspondent cultured. NOS transduction significantly reduced neointima formation.
untransduced (Fig. 4b), while there was no significant difference on the intima thickness between untransduced vein and that transduced with lacZ gene (Table 4). The ecNOS immunoreactivity was concentrated on the luminal surface of the cultured vein patch (Fig. 4d).
4. Discussion Understanding the mechanism of intimal proliferation in injured blood vessels is critical to achieving prolonged patency of re-vascularized vessels. In this study, we used an in vitro vein culture model that mimics intimal proliferation of injured blood vessels in vivo, to evaluate the effect of gene transfer on neointimal formation. When vein patches were cultured for 14 days, proliferating cells grew over the intimal surface of the patches (Fig. 1b), eventually organizing into a new layer covering the intima (Fig. 1f– h). The intimal layer of cultured vein patches was significantly thicker at the edge than at the center (Table 1, Fig. 1f,h). This, coupled with the fact that there was considerable spillage of cells from the patches on to the culture dish, suggests that cellular proliferation is more marked at the edge of a vein patch than at the center. This difference in proliferative response could in part be due to better nutrition to the edge of the patch; however, we hypothesized that the injury caused to the edges by cutting and pinning may also stimulate proliferation. The response of vein patches to the induction of injury supports this hypothesis. Mechanical injury to the intima of vein patches denuded the intimal layer leading to loss of endothelial lining (Fig. 1d), reduced thickness of intimal layer
119
(Table 1) and loss of positive staining with endothelial cell-specific markers. In culture, cells could be seen growing out of the cut edges of injured vein patches several days earlier than in uninjured vein segments, which agrees with Soyombo et al. who reported that injured veins had greater proliferation response [18]. The increased proliferation caused by injury was further demonstrated from the greater incorporation of BrdU, a thymidine analog, at the edge of a cultured vein patch than at the center and in injured patches than uninjured ones. Further, transduction efficiency, which was greater at the edge than at the center and greater on an injured patch than an uninjured patch, also correlated well with the differences in proliferation induced by injury, as the retroviral transduction requires proliferating cells for gene transfer. Corresponding to the greater proliferation, the injured vein also had a thicker intimal layer than that of the uninjured after 14-day culture. The increase in the thickness of the intimal layer of these vein patches was 38-fold at the edge and 23-fold at the center, as against the 4- and 3-fold increase seen at corresponding sites on uninjured patches (Table 1). This is contrary to Soyombo et al., who did not observe the difference on intimal layer thickness between the injured and uninjured vein [18]. The discrepancy could be due to the degree of induced injury. In Soyombo’s study, the injury was from surgical preparation that was relatively minor. In this study, the injury was more severe, involving the scraping of the luminal surface with a cotton bud, which resulted in total loss of the endothelium layer. Some previous reports suggest a central role of the endothelial layer on the arterial wall mass and intimal hyperplastic response [19–21]. Neointima formation was promoted not only by an intact endothelial covering [20], but also in a paracrine manner in a saphenous vein culture model [22]. However, in vivo studies of arterial injury indicate that total endothelial denudation and subsequent exposure of sub-endothelial layers stimulates arterial lesion growth [23]. Re-endothelialization of vessel walls restricts the extensive neointimal proliferation that follows luminal injury [24]. Similarly, in vitro studies of endothelial and smooth muscle cell co-cultures showed that endothelial cells markedly inhibit the growth characteristics of smooth muscle cells [25,26]. Our model concurs with the latter opinion— mechanical injury to the intimal surface causes loss of endothelial lining leading to an aggressive proliferative response. Further, mechanisms of vascular injury that lead to medial cell damage have been shown to cause significantly greater smooth muscle replication than those that lead to mere endothelial denudation [23], an effect mediated by smooth muscle mutagenic factors, such as basic-fibroblast growth factor [27]. Correspondingly, the cellular proliferative response observed in our
120
H. Yu et al. / Atherosclerosis 161 (2002) 113–122
model in uninjured vein patches is significantly lower than that of injured vein patches, even though uninjured patches also lose their endothelial covering on culture. The endothelial layer in uninjured veins, for whatever duration of time it is intact, could also inhibit SMC proliferation. As greater cellular proliferation occurs at the injured edges of a cultured patch, it is likely that the new layer seen in uninjured vein patches is formed largely by the migration of dividing cells from the edges. To better characterize the cells forming the intima, we studied their immunohistochemical properties with cell-specific markers. What the specific characteristics of smooth muscle cells are is largely a debated and unsettled argument. Several studies have shown that the smooth muscle cell can exist in a multitude of phenotypically heterogeneous forms that bear little resemblance to the ‘typical’ medial smooth muscle cell [28] and that these cells tend to assume an altered phenotype when grown in vitro [29]. In our model, the intimal layer of intact veins stained positively for endothelial cell-specific markers, such as anti-EC antibody or vWF (Fig. 2g) prior to culture; however, even uninjured patches stained negatively for them after culture (Fig. 2h), probably because endothelial cells are over-grown by smooth muscle cells [4]. Morphologically, cells that grew out of cultured vein patches had a spindle shape resembling smooth muscle cells. The cells that formed the new layer were similar to the medial cells in immunostaining, except that they stained negative for caldesmon and only weakly positive for myosin heavy chain, indicating some phenotypic alteration. Several studies have shown that the neointima of injured blood vessels are formed by collections of smooth muscle cells that contain certain smooth muscle markers but lack others, apart from expressing some non-muscle variant markers as well [30,31]. Frid et al. [32] showed calponin-positive and caldesmon-negative cells in the human aorta, similar to the profile of the cells that formed the intimal layer in our model. Hence, it is likely that an altered phenotype of smooth muscle cells, akin to cells observed in some intimal lesions in vivo, forms the new layer of our cultured veins. Most studies of gene transfer to vascular tissues were focused on using adeno, or adeno-associated viral or other non-viral vectors [8– 10]. However, these vectors have their own disadvantages, e.g. unstable gene expression, cytopathicity and immunogenicity. In contrast, retroviral vector has less side effect, causing less pathological response when applied to human. However, retroviral vectors have had limited usage in vascular gene therapy applications due to their poor transduction efficiency both in vitro and in vivo. We have demonstrated that VSV-G pseudotyped MuLV vectors have significantly improved transduction efficiency [33] without pseudo-transduction [15]. Since the
retroviral vector can only transduce the proliferating cells, the transduction efficiency was low when the vein was at initial culture period (1–5 days) because cells were not in proliferative stage. In this study, the induced injury to the vein will cause more proliferative response, which will make the VSV-G viral vector suitable for efficient gene transfer. We chose to perform transduction after 7 days of culture because the cells start to proliferate rapidly and a new layer of intima was not significant (data not shown). Retroviral tPA gene transfer was used to study the feasibility of up-regulating synthesis of therapeutic proteins. tPA production and activity fell with injury to vein patches at harvest as a consequence of loss of the endothelial layer (Table 3). For similar reasons, culture of uninjured vein patches also resulted in loss of tPA production. However, transduction of tPA gene significantly increased tPA production and activity from both injured and uninjured veins. The increase was 2.3-fold higher in injured than in uninjured patches, which could be a result of more proliferation and higher transduction in the injured patches. Transduced injured vein patches produced even higher tPA than vein patches at harvest (P= 0.03). This ability to up-regulate tPA secretion when viewed in the context of re-vascularized vessels could imply greater resistance to thrombosis and thereby, improved patency rates of injured vessels and bypass grafts. We used this vein culture model to study the effect of gene transfer on neointima formation. Transduction of NOS gene resulted in the reduction of neointima formation while lacZ gene transfer had no effect on neointima formation (Table 4). NOS plays an important role in synthesis of nitric oxide, a potent vasodilator and inhibitor of SMC proliferation and migration [34,35]. Gene transfer of NOS into EC using Sendai virus/liposome in vivo [12], retroviral vectors in vitro [17] and adenoviral vectors [36] has been successful in reducing SMC hyperplasia. By augmenting NOS synthesis in vein patches using retroviral gene transfer, we have shown significant reduction in intimal thickness, supporting prior studies. Retrovirus-mediated gene transfer has been suggested to cause phenotypic changes and alterations in the proliferation of target vascular cells [37]. However, we did not detect any difference in the cell marker profile of either cultured cells or cells on a new layer of cultured vein patches after transduction with the VSVG pseudotyped MuLV vector carrying the lacZ gene (Table 2). Similarly, the thickness of the intimal layer of vein patches transduced with the lacZ gene was similar to that of untransduced patches. Hence, retroviral mediated transduction itself did not alter either the phenotype of proliferating cells or their proliferative response in cultured vein patches. However, no transduction was observed in vein patches that were transduced within 5
H. Yu et al. / Atherosclerosis 161 (2002) 113–122
days of culture (data not shown), as retroviral vectors require proliferating cells for gene transfer. This is a potential limitation to the direct application of retroviral vectors in vivo. Our study demonstrates the utility of an in vitro human saphenous vein culture model in studying the effects of injury to blood vessels and inhibition of neointima formation through transduction of cultured veins with therapeutically favorable genes. Future studies using this model to examine the effects of genes aimed at containing intimal proliferation are bound to be of significant clinical impact. We believe this vein culture model will serve as an in vitro model for studying the effect of injury to blood vessels and refining the role of gene therapy in the management of vascular disease. Acknowledgements This work was supported in part by grants from the American Heart Association, Pacific Vascular Research Foundation, May R. Wright Foundation, Culpepper Foundation and James H. Zumberge Fund. The authors thank Donald Cramer, PhD for critical reading of the manuscript. References [1] Minuth WW, Kloth S, Aigner J, Sittinger M, Rockl W. Approach to an organo-typical environment for cultured cells and tissues. Biotechniques 1996;20:498 – 501. [2] Fingerle J, Kraft T. The induction of smooth muscle cell proliferation in vitro using an organ culture system. Int Angiol 1987;6:65 – 72. [3] Merrick AF, Shewring LD, Cunningham SA, Gustafsson K, Fabre JW. Organ culture of arteries for experimental studies of vascular endothelium in situ. Transpl Immunol 1997;5:3 – 9. [4] Slomp J, Gittenberger deGroot AC, van Munsteren JC, Huysmans HA, van Bockel JH, van Hinsbergh VW, et al. Nature and origin of the neointima in whole vessel wall organ culture of the human saphenous vein. Virch Arch 1996;428:59 – 67. [5] Soyombo AA, Angelini GD, Bryan AJ, Jasani B, Newby AC. Intimal proliferation in an organ culture of human saphenous vein. Am J Pathol 1990;137:1401 –10. [6] Varty K, Porter K, Bell PR, London NJ. Vein morphology and bypass graft stenosis. Br J Surg 1996;83:1375 – 9. [7] Soyombo AA, Angelini GD, Newby AC. Neointima formation is promoted by surgical preparation and inhibited by cyclic nucleotides in human saphenous vein organ cultures. J Thorac Cardiovasc Surg 1995;109:2 –12. [8] Yao A, Wang DH. Heterogeneity of adenovirus-mediated gene transfer in cultured thoracic aorta and renal artery of rats. Hypertension 1995;26:1046 –50. [9] Rekhter MD, Simari RD, Work CW, Nabel GJ, Nabel EG, Gordon D. Gene transfer into normal and atherosclerotic human blood vessels. Circ Res 1998;82:1243 – 52 see comments. [10] George SJ, Johnson JL, Angelini GD, Newby AC, Baker AH. Adenovirus-mediated gene transfer of the human TIMP-1 gene inhibits smooth muscle cell migration and neointimal formation in human saphenous vein. Hum Gene Ther 1998;9:867 – 77.
121
[11] Janssens S, Flaherty D, Nong Z, Varenne O, van Pelt N, Haustermans C, et al. Human endothelial nitric oxide synthase gene transfer inhibits vascular smooth muscle cell proliferation and neointima formation after balloon injury in rats. Circulation 1998;97:1274 – 81. [12] von der Leyen HE, Gibbons GH, Morishita R, Lewis NP, Zhang L, Nakajima M, et al. Gene therapy inhibiting neointimal vascular lesion: in vivo transfer of endothelial cell nitric oxide synthase gene. Proc Natl Acad Sci USA 1995;92:1137 –41. [13] Eton D, Terramani TT, Wang Y, Takahashi AM, Nigro JJ, Tang L, et al. Genetic engineering of stent grafts with a highly efficient pseudotyped retroviral vector. J Vasc Surg 1999;29:863 – 73. [14] Holt CM, Francis SE, Rogers S, Gadsdon PA, Taylor T, Clelland C, et al. Intimal proliferation in an organ culture of human internal mammary artery. Cardiovasc Res 1992;26:1189 – 94. [15] Yu H, Eton D, Wang Y, Kumar S, Tang L, Terramani T, et al. High efficiency in vitro gene transfer into vascular tissues using a pseudotyped retroviral vector without pseudotransduction. Gene Ther 1999;6:1876 – 83. [16] Dichek DA, Nussbaum O, Degen SJ, Anderson WF. Enhancement of the fibrinolytic activity of sheep endothelial cells by retroviral vector-mediated gene transfer. Blood 1991;77:533 –41. [17] Chen L, Daum G, Forough R, Clowes M, Walter U, Clowes AW. Overexpression of human endothelial nitric oxide synthase in rat vascular smooth muscle cells and in balloon-injured carotid artery. Circ Res 1998;82:862 – 70. [18] Soyombo AA, Angelini GD, Bryan AJ, Newby AC. Surgical preparation induces injury and promotes smooth muscle cell proliferation in a culture of human saphenous vein. Cardiovasc Res 1993;27:1961 – 7. [19] Daley SJ, Gotlieb AI. Fibroblast growth factor receptor-1 expression is associated with neointimal formation in vitro. Am J Pathol 1996;148:1193 – 202. [20] Koo EW, Gotlieb AI. Endothelial stimulation of intimal cell proliferation in a porcine aortic organ culture. Am J Pathol 1989;134:497 – 503. [21] De Mey JG, Schiffers PM. Effects of the endothelium on growth responses in arteries. J Cardiovasc Pharmacol 1993;21(1):S22 –5. [22] Allen KE, Varty K, Jones L, Sayers RD, Bell PR, London NJ. Human venous endothelium can promote intimal hyperplasia in a paracrine manner. J Vasc Surg 1994;19:577 – 84. [23] Fingerle J, Au YP, Clowes AW, Reidy MA. Intimal lesion formation in rat carotid arteries after endothelial denudation in absence of medial injury. Arteriosclerosis 1990;10:1082 –7. [24] Schwartz SMR, Deblois D, Giachelli CM. Relevance of smooth muscle replication and development to vascular disease. In: Schwartz SM, Mecham RP, editors. The Vascular Smooth Muscle Cell. San Diego: Academic Press, 1995:81 – 121. [25] Fillinger MF, O’Connor SE, Wagner RJ, Cronenwett JL. The effect of endothelial cell coculture on smooth muscle cell proliferation. J Vasc Surg 1993;17:1058 – 67 Discussion, see pp. 1067 – 1058. [26] Powell RJ, Cronenwett JL, Fillinger MF, Wagner RJ, Sampson LN. Endothelial cell modulation of smooth muscle cell morphology and organizational growth pattern. Ann Vasc Surg 1996;10:4 – 10. [27] Reidy AM. Regulation of arterial smooth muscle growth. In: Schwartz SM, Mecham RP, editors. The Vascular Smooth Muscle Cell. San Diego: Academic Press, 1995:271 – 95. [28] Gittenberger-de Groot AC, DeRuiter MC, Bergwerff M, Poelmann RE. Smooth muscle cell origin and its relation to heterogeneity in development and disease. Arterioscler Thromb Vasc Biol 1999;19:1589 – 94. [29] Schwartz SM, Campbell GR, Campbell JH. Replication of smooth muscle cells in vascular disease. Circ Res 1986;58:427 – 44.
122
H. Yu et al. / Atherosclerosis 161 (2002) 113–122
[30] Glukhova MA, Shekhonin BV, Kruth H, Koteliansky VE. Expression of cytokeratin 8 in human aortic smooth muscle cells. Am J Physiol 1991;261:72 –7. [31] Babaev VR, Bobryshev YV, Stenina OV, Tararak EM, Gabbiani G. Heterogeneity of smooth muscle cells in atheromatous plaque of human aorta. Am J Pathol 1990;136:1031 – 42. [32] Frid MG, Shekhonin BV, Koteliansky VE, Glukhova MA. Phenotypic changes of human smooth muscle cells during development: late expression of heavy caldesmon and calponin. Dev Biol 1992;153:185 – 93. [33] Friedmann T, Yee JK. Pseudotyped retroviral vectors for studies of human gene therapy. Nat Med 1995;1:275 – 7.
[34] Sarkar R, Meinberg EG, Stanley JC, Gordon D, Webb RC. Nitric oxide reversibly inhibits the migration of cultured vascular smooth muscle cells. Circ Res 1996;78:225 – 30. [35] Moncada S. Nitric oxide in the vasculature: physiology and pathophysiology. Ann NY Acad Sci 1997;811:60 – 7 Discussion, see pp. 67 – 69. [36] Cable DGT, Schaff HV, Pompili VJ. Recombinant endothelial nitric oxide synthase-transduced human saphenous veins: gene therapy to augment nitric oxide production in bypass conduits. Circulation 1997;96(2):173 – 8. [37] Inaba M, Toninelli E, Vanmeter G, Bender JR, Conte MS. Retroviral gene transfer: effects on endothelial cell phenotype. J Surg Res 1998;78:31 – 6.
Injury induced neointima formation and its inhibition by retrovirus-mediated transfer of nitride oxide synthase gene in an in-vitro human saphenous vein culture model Hong Yu a,*, S. Ram Kumar a, Lili Tang a, Thomas T. Terramani a, Vincent L. Rowe a, Ying Wang a, Rahul A. Nathwani a, Fred A. Weaver a, Darwin Eton b a
Department of Surgery, Vascular Di6ision, Keck School of Medicine of the Uni6ersity of Southern California, 2025 Zonal A6enue, RMR 505, Los Angeles, CA 90089, USA b Department of Surgery, Uni6ersity of Miami, Miami, FL 33156, USA Received 9 January 2001; received in revised form 31 January 2001; accepted 10 July 2001
Abstract Human saphenous veins were cultured to characterize neointima formation and feasibility of gene transfer to inhibit the intimal proliferative response to injury. Mechanical injury was introduced by abrading the luminal surface of the vein patch with a sterile cotton bud. Both injured and non-injured vein patches were cultured and transduced with retroviral vectors carrying marker or therapeutic genes. After a 14-day culture, the thickness of the intimal layer of non-injured vein patches reached 90 9 28 mm at the edge and 61 922 mm at the center (n=29) from the original 22 9 12 mm at harvest (n =6, P= 0.02). Mechanical injury to the intimal surface prior to culture resulted in an exaggerated proliferative response. The intimal thickness of injured vein patches increased from 3.4 91 mm right after injury to 128 9 23 mm (n= 12, P B 0.001) at the edge after 14-day culture. Genes were transduced efficiently into a luminal layer of cultured veins using a pseudotyped murine leukemia viral vector. Transduction of gene encoding nitric oxide synthase resulted in reduction of neointima formation to 33 9 7 mm (n = 12) at the edge after 14-day culture compared to 90 mm (PB0.01) seen in untransduced non-injured vein patches. Marker gene transduction did not alter intimal proliferative response or its immunohistochemical profile. The data suggest that cultured vein can be used as a model for studying the effects of injury to blood vessels and to evaluate the effects of candidate therapeutic genes. © 2002 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Vein culture; Neointima; Gene transfer; Nitride oxide synthase; Endothelial cell; Retroviral vector; Injury
1. Introduction Intimal response to injury continues to be an unresolved obstacle to long term graft patency. Organ culture offers a simplified in vitro system to study the responses of vascular tissues in vivo while maintaining phenotypic and three-dimensional architecture allowing spatial relationships of cell types that are potentially interactive during tissue injury and growth regulation [1]. Numerous ex vivo vascular tissue culture systems * Corresponding author. Tel.: + 1-323-442-3792; fax: +1-323-4423164. E-mail address: [email protected] (H. Yu).
utilizing either arteries or veins have been evaluated in the last two decades as in vitro models to study vascular responses to various external stimuli including injury [2–4]. Veins appear to be the more attractive targets because they are more readily available and thin enough to permit diffusion of nutrients and metabolites. Culture of human saphenous vein was first described in 1990 by Soyombo et al. [5], where she suggested that it could be used as an in vitro model to study neointima formation in vein grafts. Varty et al. showed that the neointima developed in cultured vein segments is similar to the intimal hyperplasia in stenosed vein grafts [6]. Few have studied the effect of injury on neointima formation [7]. Gene transfer has
0021-9150/02/$ - see front matter © 2002 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 0 2 1 - 9 1 5 0 ( 0 1 ) 0 0 6 2 5 - 6
1 114
H. Yu et al. / Atherosclerosis 161 (2002) 113–122
been reported in cultured arteries using adenoviral vectors [8,9] and the transfer of the human TIMP-1 gene has been shown to inhibit neointimal formation in a cultured vein [10]. In this study, an in vitro injured vein culture model was established by introducing mechanical injury to the cultured vein patches, that may be used to mimic intimal proliferation of injured blood vessels in vivo and to evaluate the effect of gene transfer on neointimal formation. Since nitric oxide synthase (NOS) has been shown to inhibit neointima formation in vivo [11,12] through increasing the nitric oxide production, the effect of NOS gene transduction into the cultured vein was studied. We found that the transfer of NOS gene can inhibit the neointima formation in the cultured vein, which demonstrates that the vein culture model can be used to study the effect of gene transfer on inhibition of neointima.
2. Materials and methods
2.1. Culture of human saphenous 6ein Excreta human saphenous veins (total 35 samples) were obtained from patients undergoing vascular or coronary artery bypass surgery at the University of Southern California Hospitals, which was approved by the Institutional Review Board (IRB No. 969042). Harvested veins were gently flushed with heparinized saline and peri-adventitial tissue was debrided in the operating room. Meticulous handling reduced unnecessary trauma. Details of saphenous vein culture are as previously described [13]. The veins were pinned down to hardened Sylgard medium on the culture plate surface. Mechanical injury was induced by abrading the luminal surface of the vein patch ten times with a sterile cotton bud.
2.2. Histological analysis To perform differential staining of the elastic lamina, vein patches were fixed overnight in 10% formaldehyde, embedded in paraffin and cut into 4 mm tissue sections. Slides were de-paraffinized and stained using the standard Verhoeff’s elastic staining (American MasterTech Scientific, CA) procedure. Arbitrarily, 10% surface of the vein patch in outer area was defined as the edge and 90% of the surface in the inner area as the center. Intimal thickness was measured from the elastic layer at ten random points over the center and both edges of the patches. Linemorphometric analyses were performed on computerassisted images at 40× magnification using commercial software OPTIMAS (Optimas Corporation, WA).
For immunohistochemical staining of paraffin sections, de-paraffinized sections (4 mm) of vein patches were antigen retrieved by pressure-cooking the slides in Antigen Retrieval Citra 10× solution (BioG Genex) in a microwave oven at high power for 15 min followed by 40% power setting for 15 min. The whole staining was performed at room temperature. After washing with PBS and 20 min incubation with 20% horse serum (Ventana Medical Systems, AZ) to block nonspecific reaction, sections were incubated overnight with various mouse anti-human IgG primary antibody (1:500 dilution, Dako Laboratories, CA), as detailed in Section 3. The antibody was washed off with PBS and sections incubated for 45 min with horse anti-mouse IgG biotinylated secondary antibody (1:20 dilution, Vector Laboratories, CA) followed with PBS washing. After blocking the endogenous peroxidase activity by 20-min incubation with 3% hydrogen peroxide (Sigma, MO), the sections were then incubated with the avidin–biotinylated peroxidase complex (Vector Laboratories) for 30 min and rinsed with PBS. Bound antibody was detected by adding 3-3%-diaminobenzidine substrate (Ventana Medical Systems) for 20 min. The sections were washed with tap water, counterstained with Meyer’s hematoxylin (Sigma) for 1 min, dehydrated in graded alcohol solutions followed by xylene and mounted using Permount medium. Scanning and transmission electron microscopy was performed as described [14].
2.3. Retro6iral transduction VSV-G pseudotyped MuLV (VSV-G/MuLV) vectors were generated as described previously [15]. Vector plasmids pG1nBgSvNa, pLtSN [16] and pLCNSN [17] carrying genes encoding for nuclear-localized galactosidase (-Gal), tissue type plasminogen activator (tPA) and human endothelium-specific constitutive nitric oxide synthase (NOS), respectively, were provided by Genetic Therapy Inc./Novartis, Dr David Dichek (University of California at San Francisco), and Dr A.W.Clowes (University of Washington), respectively. The viral titers were 3–10×106 cfu/ml. Vein patches were cultured 7 days prior to transduction as described [15] with VSV-G/MuLV vectors mentioned above. The transduced vein was continually cultured for an additional 7 days. To study the effect of duration of transduction, whole vein segments cultured for 7 days were immersed in viral supernatants for 15, 30, 60 and 120 min, after which they were rinsed with PBS and cultured for another 7 days. They were then stained with X-Gal [15]. The number of stained cells at the edge and center of the paraffin sections was counted over five random fields by an observer blinded to the procedure.
H. Yu et al. / Atherosclerosis 161 (2002) 113–122 Table 1 Intimal proliferation in cultured vein patches Vein patch
Thickness of the intimal layer (mm)
P value* (n)
Edge
Center
Non-cultured Uninjured Injured
22 911 3.4 9 1.0
20 911 3.9 90.5
0.26 (6) 0.78 (4)
Cultured Uninjured Injured
90 918 128 923
70 914 90 924
0.021 (29) 0.00006 (12)
0.01
0.03
P value†
* Comparison between edge and center of the same vein patch. Comparison between corresponding sites on injured and uninjured patches. Excreta human saphenous vein patches were cultured for 14 days after scraping the intimal layer with a cotton swab (injured) and studied against control patches from the same vein cultured without scraping (uninjured). Non-cultured patches belong to a different pool of patients. The thickness of the intimal layer was measured after differential elastic staining of paraffin sections at 40× magnification. Values are expressed as mean 9S.E. of mean and the number of veins studied are shown in the bracket (n). Variables were compared using Student’s t-test. †
115
seen in the intimal layer of human saphenous vein segments prior to culture (Fig. 1a). After 14 days culture in a dish, the surface of vein patches was covered with fibrous cells resembling smooth muscle cells (Fig. 1b). A new layer covering the intimal surface in the cultured vein patches was approximately three times thicker (61–80 mm, Fig. 1f,g) at all sites than non-cultured patches (21 mm) (Fig. 1(e); Table 1). To mimic mechanical injury during surgery, the luminal surface of vein patches was scraped with a cotton swab. The induced injury resulted in loss of endothelial covering revealed by electron microscopy (Fig. 1d), associated with a reduction in the thickness of the intimal surface from 21.1 to 3.64 mm (Table 1). However, on culture, injured vein patches had a greater intimal proliferative response. Cells could be seen sprouting out of the patches after 3 days in culture and a thicker intimal layer, 128 mm at the edge and 90 mm at the center, was observed (n= 12, Table 1, Fig. 1h). While the thickness of the intima on non-cultured vein patches was similar at all points on the patch, the intimal layer of the cultured patch was 32% (uninjured) and 42% (injured) thicker at the edge (Fig. 1f) than at the center (Fig. 1g, PB0.03, n=12, Table 1).
2.4. Determination of tPA concentration After transduction with VSV-G/MuLV/LtSN, vein patches were cultured for an additional 7 days and allowed to continue in culture in serum-free William’s medium for a further 48 h, by then the supernatants were collected. Control vein patches were either transduced with lacZ gene or not transduced at all. tPA production by vein patches at harvest was determined by immersing injured or uninjured vein segments directly in serum-free medium for 2 days. Quantitation of tPA antigen and fibrinolytic enzyme activity was performed with TintElize and Chromolize tPA assay kits from Biopool (Ventura, CA).
2.5. Statistical analyses All values are expressed as mean9S.E. of mean. 2 Statistics were used to compare proportions and frequency distribution of variables among the different groups. Mean values for continuous variables were compared with analysis of variance and paired twotailed Student’s t-tests. Significance was attributed to a P value B 0.05. 3. Results
3.1. Vein culture and intimal proliferation The cobblestone architecture of endothelial cells was
3.2. Characterization of the cells in the new intimal layer Immunohistochemical analysis of cultured injured and uninjured vein patches showed similar monoclonal antibody staining characteristics (Table 2). The cells in the newly formed intima and in the media of cultured veins stained positive for the smooth muscle components a-actin and calponin (Fig. 2a,b). However, while the new layer stained negative for the muscle protein marker caldesmon, the medial layer was weakly positive (Fig. 2c); similarly, while the newly formed intima stained weakly positive with anti-myosin heavy chain antibody, the medial layer stained strongly positive (Fig. 2d). Prior to culture, the intimal layer of uninjured veins stained positively for endothelial cell specific markers, such as vWF (Fig. 2e) and anti-EC antibody, but after culture the newly formed intimal layer showed no staining (Fig. 2f).
3.3. Gene transfer into cultured 6ein After 7 days in culture, vein segments could be transduced with high efficiency using the VSVG pseudotyped MuLV vector carrying different genes. Transduction increased progressively with increasing duration of viral admixture and was highest at 2 h of exposure to viral supernatant (Fig. 3a). Longer durations of transduction did not alter transduction efficiency significantly (data not shown).
116
H. Yu et al. / Atherosclerosis 161 (2002) 113–122
H. Yu et al. / Atherosclerosis 161 (2002) 113–122
117
Table 2 Characterization of cells in the newly formed layer of cultured veins by immunohistochemistry Tissue
Uninjured Injured Transduced
Muscle marker antigens
Fibroblast antigen
EC marker antigens
a-Actin
Calponin
Caldesmon
Myosin heavy chain
5B5
vWF
Anti-EC
+++ +++ +++
+++ +++ +++
− − −
+ + −
++ ++ ++
− − −
− − −
Human saphenous vein explants were cultured for 14 days and paraffin sections stained for specific cell type markers. The staining pattern of vein patches cultured after scraping the luminal surface with a cotton swab (injured) and patches transduced with the lacZ gene using the VSV-G pseudotyped MuLV vector (transduced) were compared with that of control patches (uninjured). VWF, Von-Willebrand factor; EC, endothelial cell; −, negative staining; +, weakly positive staining; ++, positive staining; +++, strongly positive staining.
Transduction was seen at both luminal and adventitial surfaces of the patches. However, no penetration into deeper layers was observed (Fig. 3b). Transduction was greater at the edge (Fig. 3c) than at the center at all durations of transduction (Fig. 3d). There were 479 9 and 13 9 3 transduced cells per high power field (cells/ hpv, 200× ) at the edge and center, respectively, on the cultured patch transduced with viral vector for 2 h (n = 10, P= 0.0015). Injured vein patches could be transduced with significantly higher transduction efficiencies (Fig. 3(a), upper panel) than uninjured controls (Fig. 3(a), lower panel) at both the edge (1179 15 cells/hpv) and center (4692 cells/hpv, n = 3, PB 0.001). The immunohistochemical staining pattern on the newly formed layer of vein patches transduced with the VSV-G/MuLV vector carrying the lacZ gene was similar to that of untransduced injured and uninjured patches (Table 2).
3.4. tPA production by 6ein patches tPA production from the vein patches was studied at different stages— at harvest, after the induction of injury, after culture and after transduction (Table 3). Induction of injury to the luminal surface of harvested veins caused a fall in tPA production of vein patch (P =0.04). After 14 days in culture, tPA production from both uninjured and injured vein patches fell to negligible amounts. Transduction of tPA gene to cultured vein patches led to a 28- and 12.5-fold increase in
Table 3 Production of tPA by vein patches Vein patch
Non-cultured Cultured
Cultured/lacZ Cultured/tPA
tPA antigen (ng/ml/cm 2) Uninjured 10.7 9 0.1 Injured 4.8 90.9
0.6 90.2 1.0 90.1
0.7 90.2 0.8 9 0.1
7.9 9 1.8 27.3 9 5.9
tPA acti6ity (IU/ml/cm 2) Uninjured 24.6 91.7 Injured 10.5 9 1.0
0.3 90.2 0.3 90.1
0.3 9 0.2 0.049 0.01
9.3 92.0 33.8 910.0
tPA production and activity from freshly harvested human saphenous veins (non-cultured) were studied without (uninjured) or after (injured) induction of injury. After culture for 7 days, vein patches were transduced with VSV-G/MuLV/G1nBgSvNa (Cultured/lacZ) or VSV-G/MuLV/LtSN (Cultured/tPA) vector and the culture continued for another 7 days. Untransduced vein patches cultured for 14 days (cultured) were studied as controls. The tPA production in the collected cultured media was measured and normalized to the area of vein patch. Values are expressed as mean 9S.E. of five samples.
tPA production in injured and uninjured vein patches, respectively, with a concomitant increase in tPA activity. LacZ co-transduction did not affect tPA production or activity.
3.5. Effect of gene transduction on neointima formation The thickness of the intima layer of vein segment transduced with genes encoding ecNOS (Fig. 4c) was significantly (PB0.01) reduced compared to that of
Fig. 1. Human saphenous vein culture. (a) Cobblestone architecture of endothelial cells was shown by scanning electron micrograph on uncultured human saphenous vein explants. (b) The intimal layer of native veins was replaced by sheets of fibrous cells after a 14-day culture. (c) Transmission electron micrograph show endothelial lining of the vessel wall (arrow), N indicates the EC nuclear. (d) Loss of endothelial cells after mechanical injury to the intimal surface. The thickness of the intimal layer was shown by Verhoeff’s Von Geison staining of vein sections of uncultured (e), 14-day cultured at edge (f), 14-day cultured at center (g), injured and cultured at edge (h). The internal elastic lamina was shown by arrows. The scale bar in (e) applies to all pictures in lower panel. (L, luminal surface; I, intimal layer; M, medial layer of the vein patch.) Fig. 2. Immunohistochemical staining pattern of cultured vein patches. Paraffin sections of human saphenous vein explants cultured for 14 days were stained with antibodies for specific cell-type markers: a-actin (a), calponin (b), caldesmon (c), myosin heavy chain (d), vWF (e,f). The medial (arrow) and newly formed layers (arrowhead) of vein patches stained strongly positive for the muscle markers a-actin (a) and calponin (b). The medial layer stained weakly positive for caldesmon, while the intimal layer, negative (c). The medial layer stained strongly positive for myosin heavy chain, while the intimal layer was only weakly positive (d). While the luminal surface of non-cultured patches stained positive for the anti-vWF marker directed against endothelial cells (e), the cultured vein patches did not (f).
118
H. Yu et al. / Atherosclerosis 161 (2002) 113–122
Fig. 3. Transduction to the cultured veins. Human saphenous vein explants were cultured for 7 days, transduced with the VSV-G pseudotyped MuLV vector carrying the lacZ gene and cultured for another 7 days. Transduction efficiency increased with increasing durations of transduction (shown in minutes in (a)) and was greater in the vein segment scraped with a cotton swab prior to culture (upper panel in (a)) than the unscraped segment (lower panel in (a)). Paraffin sections of the transduced vein patches stained with X-Gal showed transduced cells (blue) at both the luminal (L) and adventitial (A) surfaces of the vein (b), but no transduction in the medial layer. There were more transduced cells at the edge of a cultured vein patch (c) than the center (d) through en face view of transduced vein by microscopy (40 × ). Blue dots represent transduced cell nuclei stained with X-Gal. Fig. 4. Effect of NOS transduction on neointima formation. Sections of (a) uncultured, (b) 14-day cultured, (c) NOS gene transduced and cultured vein segments were stained with Verhoeff’s Von Geison staining. The dark filled arrows indicate the internal elastic lamina. The line arrows indicate the new layer intima that was significantly reduced after NOS transduction (c) compared with the untransduced (b). (d) Immunohistochemical staining for ecNOS in the cultured vein patch transduced with NOS gene. Dark brown layer at the luminal surface of the vein indicates NOS positive. The scale bar in (a) applies to a –c panels; the scale in (d) is doubled.
H. Yu et al. / Atherosclerosis 161 (2002) 113–122 Table 4 Neointima formation and inhibition in an in vitro cultured vein model Position*
Edge Center
Intimal thickness (mm)† b-Gal‡
Uncultured
Cultured
21 911 (6) 20 9 11 (6)
999 34 (17) 95935 (6) 80 916 (17) 74 9 21 (6)
NOS§ 33 9 7 (12) 26 9 4 (12)
* Position signifies the site on the vein, edge or center, where the intimal thickness was measured. † Intimal thickness was measured from the stained elastic lamina of veins which were uncultured, cultured for 14 days, cultured for 7 days and then transduced with genes encoding for b-Gal, NOS or aG1 and cultured for 7 more days. All values are expressed as mean 9 S.E. of mean with the number of vein samples in parentheses. ‡ P = 0.80 and 0.70 for edge and center, respectively compared to the correspondent cultured. b-Gal transduction did not reduce intimal proliferation showing a lack of effect of transduction itself. § PB0.01 for both edge and center compared to the correspondent cultured. NOS transduction significantly reduced neointima formation.
untransduced (Fig. 4b), while there was no significant difference on the intima thickness between untransduced vein and that transduced with lacZ gene (Table 4). The ecNOS immunoreactivity was concentrated on the luminal surface of the cultured vein patch (Fig. 4d).
4. Discussion Understanding the mechanism of intimal proliferation in injured blood vessels is critical to achieving prolonged patency of re-vascularized vessels. In this study, we used an in vitro vein culture model that mimics intimal proliferation of injured blood vessels in vivo, to evaluate the effect of gene transfer on neointimal formation. When vein patches were cultured for 14 days, proliferating cells grew over the intimal surface of the patches (Fig. 1b), eventually organizing into a new layer covering the intima (Fig. 1f– h). The intimal layer of cultured vein patches was significantly thicker at the edge than at the center (Table 1, Fig. 1f,h). This, coupled with the fact that there was considerable spillage of cells from the patches on to the culture dish, suggests that cellular proliferation is more marked at the edge of a vein patch than at the center. This difference in proliferative response could in part be due to better nutrition to the edge of the patch; however, we hypothesized that the injury caused to the edges by cutting and pinning may also stimulate proliferation. The response of vein patches to the induction of injury supports this hypothesis. Mechanical injury to the intima of vein patches denuded the intimal layer leading to loss of endothelial lining (Fig. 1d), reduced thickness of intimal layer
119
(Table 1) and loss of positive staining with endothelial cell-specific markers. In culture, cells could be seen growing out of the cut edges of injured vein patches several days earlier than in uninjured vein segments, which agrees with Soyombo et al. who reported that injured veins had greater proliferation response [18]. The increased proliferation caused by injury was further demonstrated from the greater incorporation of BrdU, a thymidine analog, at the edge of a cultured vein patch than at the center and in injured patches than uninjured ones. Further, transduction efficiency, which was greater at the edge than at the center and greater on an injured patch than an uninjured patch, also correlated well with the differences in proliferation induced by injury, as the retroviral transduction requires proliferating cells for gene transfer. Corresponding to the greater proliferation, the injured vein also had a thicker intimal layer than that of the uninjured after 14-day culture. The increase in the thickness of the intimal layer of these vein patches was 38-fold at the edge and 23-fold at the center, as against the 4- and 3-fold increase seen at corresponding sites on uninjured patches (Table 1). This is contrary to Soyombo et al., who did not observe the difference on intimal layer thickness between the injured and uninjured vein [18]. The discrepancy could be due to the degree of induced injury. In Soyombo’s study, the injury was from surgical preparation that was relatively minor. In this study, the injury was more severe, involving the scraping of the luminal surface with a cotton bud, which resulted in total loss of the endothelium layer. Some previous reports suggest a central role of the endothelial layer on the arterial wall mass and intimal hyperplastic response [19–21]. Neointima formation was promoted not only by an intact endothelial covering [20], but also in a paracrine manner in a saphenous vein culture model [22]. However, in vivo studies of arterial injury indicate that total endothelial denudation and subsequent exposure of sub-endothelial layers stimulates arterial lesion growth [23]. Re-endothelialization of vessel walls restricts the extensive neointimal proliferation that follows luminal injury [24]. Similarly, in vitro studies of endothelial and smooth muscle cell co-cultures showed that endothelial cells markedly inhibit the growth characteristics of smooth muscle cells [25,26]. Our model concurs with the latter opinion— mechanical injury to the intimal surface causes loss of endothelial lining leading to an aggressive proliferative response. Further, mechanisms of vascular injury that lead to medial cell damage have been shown to cause significantly greater smooth muscle replication than those that lead to mere endothelial denudation [23], an effect mediated by smooth muscle mutagenic factors, such as basic-fibroblast growth factor [27]. Correspondingly, the cellular proliferative response observed in our
120
H. Yu et al. / Atherosclerosis 161 (2002) 113–122
model in uninjured vein patches is significantly lower than that of injured vein patches, even though uninjured patches also lose their endothelial covering on culture. The endothelial layer in uninjured veins, for whatever duration of time it is intact, could also inhibit SMC proliferation. As greater cellular proliferation occurs at the injured edges of a cultured patch, it is likely that the new layer seen in uninjured vein patches is formed largely by the migration of dividing cells from the edges. To better characterize the cells forming the intima, we studied their immunohistochemical properties with cell-specific markers. What the specific characteristics of smooth muscle cells are is largely a debated and unsettled argument. Several studies have shown that the smooth muscle cell can exist in a multitude of phenotypically heterogeneous forms that bear little resemblance to the ‘typical’ medial smooth muscle cell [28] and that these cells tend to assume an altered phenotype when grown in vitro [29]. In our model, the intimal layer of intact veins stained positively for endothelial cell-specific markers, such as anti-EC antibody or vWF (Fig. 2g) prior to culture; however, even uninjured patches stained negatively for them after culture (Fig. 2h), probably because endothelial cells are over-grown by smooth muscle cells [4]. Morphologically, cells that grew out of cultured vein patches had a spindle shape resembling smooth muscle cells. The cells that formed the new layer were similar to the medial cells in immunostaining, except that they stained negative for caldesmon and only weakly positive for myosin heavy chain, indicating some phenotypic alteration. Several studies have shown that the neointima of injured blood vessels are formed by collections of smooth muscle cells that contain certain smooth muscle markers but lack others, apart from expressing some non-muscle variant markers as well [30,31]. Frid et al. [32] showed calponin-positive and caldesmon-negative cells in the human aorta, similar to the profile of the cells that formed the intimal layer in our model. Hence, it is likely that an altered phenotype of smooth muscle cells, akin to cells observed in some intimal lesions in vivo, forms the new layer of our cultured veins. Most studies of gene transfer to vascular tissues were focused on using adeno, or adeno-associated viral or other non-viral vectors [8– 10]. However, these vectors have their own disadvantages, e.g. unstable gene expression, cytopathicity and immunogenicity. In contrast, retroviral vector has less side effect, causing less pathological response when applied to human. However, retroviral vectors have had limited usage in vascular gene therapy applications due to their poor transduction efficiency both in vitro and in vivo. We have demonstrated that VSV-G pseudotyped MuLV vectors have significantly improved transduction efficiency [33] without pseudo-transduction [15]. Since the
retroviral vector can only transduce the proliferating cells, the transduction efficiency was low when the vein was at initial culture period (1–5 days) because cells were not in proliferative stage. In this study, the induced injury to the vein will cause more proliferative response, which will make the VSV-G viral vector suitable for efficient gene transfer. We chose to perform transduction after 7 days of culture because the cells start to proliferate rapidly and a new layer of intima was not significant (data not shown). Retroviral tPA gene transfer was used to study the feasibility of up-regulating synthesis of therapeutic proteins. tPA production and activity fell with injury to vein patches at harvest as a consequence of loss of the endothelial layer (Table 3). For similar reasons, culture of uninjured vein patches also resulted in loss of tPA production. However, transduction of tPA gene significantly increased tPA production and activity from both injured and uninjured veins. The increase was 2.3-fold higher in injured than in uninjured patches, which could be a result of more proliferation and higher transduction in the injured patches. Transduced injured vein patches produced even higher tPA than vein patches at harvest (P= 0.03). This ability to up-regulate tPA secretion when viewed in the context of re-vascularized vessels could imply greater resistance to thrombosis and thereby, improved patency rates of injured vessels and bypass grafts. We used this vein culture model to study the effect of gene transfer on neointima formation. Transduction of NOS gene resulted in the reduction of neointima formation while lacZ gene transfer had no effect on neointima formation (Table 4). NOS plays an important role in synthesis of nitric oxide, a potent vasodilator and inhibitor of SMC proliferation and migration [34,35]. Gene transfer of NOS into EC using Sendai virus/liposome in vivo [12], retroviral vectors in vitro [17] and adenoviral vectors [36] has been successful in reducing SMC hyperplasia. By augmenting NOS synthesis in vein patches using retroviral gene transfer, we have shown significant reduction in intimal thickness, supporting prior studies. Retrovirus-mediated gene transfer has been suggested to cause phenotypic changes and alterations in the proliferation of target vascular cells [37]. However, we did not detect any difference in the cell marker profile of either cultured cells or cells on a new layer of cultured vein patches after transduction with the VSVG pseudotyped MuLV vector carrying the lacZ gene (Table 2). Similarly, the thickness of the intimal layer of vein patches transduced with the lacZ gene was similar to that of untransduced patches. Hence, retroviral mediated transduction itself did not alter either the phenotype of proliferating cells or their proliferative response in cultured vein patches. However, no transduction was observed in vein patches that were transduced within 5
H. Yu et al. / Atherosclerosis 161 (2002) 113–122
days of culture (data not shown), as retroviral vectors require proliferating cells for gene transfer. This is a potential limitation to the direct application of retroviral vectors in vivo. Our study demonstrates the utility of an in vitro human saphenous vein culture model in studying the effects of injury to blood vessels and inhibition of neointima formation through transduction of cultured veins with therapeutically favorable genes. Future studies using this model to examine the effects of genes aimed at containing intimal proliferation are bound to be of significant clinical impact. We believe this vein culture model will serve as an in vitro model for studying the effect of injury to blood vessels and refining the role of gene therapy in the management of vascular disease. Acknowledgements This work was supported in part by grants from the American Heart Association, Pacific Vascular Research Foundation, May R. Wright Foundation, Culpepper Foundation and James H. Zumberge Fund. The authors thank Donald Cramer, PhD for critical reading of the manuscript. References [1] Minuth WW, Kloth S, Aigner J, Sittinger M, Rockl W. Approach to an organo-typical environment for cultured cells and tissues. Biotechniques 1996;20:498 – 501. [2] Fingerle J, Kraft T. The induction of smooth muscle cell proliferation in vitro using an organ culture system. Int Angiol 1987;6:65 – 72. [3] Merrick AF, Shewring LD, Cunningham SA, Gustafsson K, Fabre JW. Organ culture of arteries for experimental studies of vascular endothelium in situ. Transpl Immunol 1997;5:3 – 9. [4] Slomp J, Gittenberger deGroot AC, van Munsteren JC, Huysmans HA, van Bockel JH, van Hinsbergh VW, et al. Nature and origin of the neointima in whole vessel wall organ culture of the human saphenous vein. Virch Arch 1996;428:59 – 67. [5] Soyombo AA, Angelini GD, Bryan AJ, Jasani B, Newby AC. Intimal proliferation in an organ culture of human saphenous vein. Am J Pathol 1990;137:1401 –10. [6] Varty K, Porter K, Bell PR, London NJ. Vein morphology and bypass graft stenosis. Br J Surg 1996;83:1375 – 9. [7] Soyombo AA, Angelini GD, Newby AC. Neointima formation is promoted by surgical preparation and inhibited by cyclic nucleotides in human saphenous vein organ cultures. J Thorac Cardiovasc Surg 1995;109:2 –12. [8] Yao A, Wang DH. Heterogeneity of adenovirus-mediated gene transfer in cultured thoracic aorta and renal artery of rats. Hypertension 1995;26:1046 –50. [9] Rekhter MD, Simari RD, Work CW, Nabel GJ, Nabel EG, Gordon D. Gene transfer into normal and atherosclerotic human blood vessels. Circ Res 1998;82:1243 – 52 see comments. [10] George SJ, Johnson JL, Angelini GD, Newby AC, Baker AH. Adenovirus-mediated gene transfer of the human TIMP-1 gene inhibits smooth muscle cell migration and neointimal formation in human saphenous vein. Hum Gene Ther 1998;9:867 – 77.
121
[11] Janssens S, Flaherty D, Nong Z, Varenne O, van Pelt N, Haustermans C, et al. Human endothelial nitric oxide synthase gene transfer inhibits vascular smooth muscle cell proliferation and neointima formation after balloon injury in rats. Circulation 1998;97:1274 – 81. [12] von der Leyen HE, Gibbons GH, Morishita R, Lewis NP, Zhang L, Nakajima M, et al. Gene therapy inhibiting neointimal vascular lesion: in vivo transfer of endothelial cell nitric oxide synthase gene. Proc Natl Acad Sci USA 1995;92:1137 –41. [13] Eton D, Terramani TT, Wang Y, Takahashi AM, Nigro JJ, Tang L, et al. Genetic engineering of stent grafts with a highly efficient pseudotyped retroviral vector. J Vasc Surg 1999;29:863 – 73. [14] Holt CM, Francis SE, Rogers S, Gadsdon PA, Taylor T, Clelland C, et al. Intimal proliferation in an organ culture of human internal mammary artery. Cardiovasc Res 1992;26:1189 – 94. [15] Yu H, Eton D, Wang Y, Kumar S, Tang L, Terramani T, et al. High efficiency in vitro gene transfer into vascular tissues using a pseudotyped retroviral vector without pseudotransduction. Gene Ther 1999;6:1876 – 83. [16] Dichek DA, Nussbaum O, Degen SJ, Anderson WF. Enhancement of the fibrinolytic activity of sheep endothelial cells by retroviral vector-mediated gene transfer. Blood 1991;77:533 –41. [17] Chen L, Daum G, Forough R, Clowes M, Walter U, Clowes AW. Overexpression of human endothelial nitric oxide synthase in rat vascular smooth muscle cells and in balloon-injured carotid artery. Circ Res 1998;82:862 – 70. [18] Soyombo AA, Angelini GD, Bryan AJ, Newby AC. Surgical preparation induces injury and promotes smooth muscle cell proliferation in a culture of human saphenous vein. Cardiovasc Res 1993;27:1961 – 7. [19] Daley SJ, Gotlieb AI. Fibroblast growth factor receptor-1 expression is associated with neointimal formation in vitro. Am J Pathol 1996;148:1193 – 202. [20] Koo EW, Gotlieb AI. Endothelial stimulation of intimal cell proliferation in a porcine aortic organ culture. Am J Pathol 1989;134:497 – 503. [21] De Mey JG, Schiffers PM. Effects of the endothelium on growth responses in arteries. J Cardiovasc Pharmacol 1993;21(1):S22 –5. [22] Allen KE, Varty K, Jones L, Sayers RD, Bell PR, London NJ. Human venous endothelium can promote intimal hyperplasia in a paracrine manner. J Vasc Surg 1994;19:577 – 84. [23] Fingerle J, Au YP, Clowes AW, Reidy MA. Intimal lesion formation in rat carotid arteries after endothelial denudation in absence of medial injury. Arteriosclerosis 1990;10:1082 –7. [24] Schwartz SMR, Deblois D, Giachelli CM. Relevance of smooth muscle replication and development to vascular disease. In: Schwartz SM, Mecham RP, editors. The Vascular Smooth Muscle Cell. San Diego: Academic Press, 1995:81 – 121. [25] Fillinger MF, O’Connor SE, Wagner RJ, Cronenwett JL. The effect of endothelial cell coculture on smooth muscle cell proliferation. J Vasc Surg 1993;17:1058 – 67 Discussion, see pp. 1067 – 1058. [26] Powell RJ, Cronenwett JL, Fillinger MF, Wagner RJ, Sampson LN. Endothelial cell modulation of smooth muscle cell morphology and organizational growth pattern. Ann Vasc Surg 1996;10:4 – 10. [27] Reidy AM. Regulation of arterial smooth muscle growth. In: Schwartz SM, Mecham RP, editors. The Vascular Smooth Muscle Cell. San Diego: Academic Press, 1995:271 – 95. [28] Gittenberger-de Groot AC, DeRuiter MC, Bergwerff M, Poelmann RE. Smooth muscle cell origin and its relation to heterogeneity in development and disease. Arterioscler Thromb Vasc Biol 1999;19:1589 – 94. [29] Schwartz SM, Campbell GR, Campbell JH. Replication of smooth muscle cells in vascular disease. Circ Res 1986;58:427 – 44.
122
H. Yu et al. / Atherosclerosis 161 (2002) 113–122
[30] Glukhova MA, Shekhonin BV, Kruth H, Koteliansky VE. Expression of cytokeratin 8 in human aortic smooth muscle cells. Am J Physiol 1991;261:72 –7. [31] Babaev VR, Bobryshev YV, Stenina OV, Tararak EM, Gabbiani G. Heterogeneity of smooth muscle cells in atheromatous plaque of human aorta. Am J Pathol 1990;136:1031 – 42. [32] Frid MG, Shekhonin BV, Koteliansky VE, Glukhova MA. Phenotypic changes of human smooth muscle cells during development: late expression of heavy caldesmon and calponin. Dev Biol 1992;153:185 – 93. [33] Friedmann T, Yee JK. Pseudotyped retroviral vectors for studies of human gene therapy. Nat Med 1995;1:275 – 7.
[34] Sarkar R, Meinberg EG, Stanley JC, Gordon D, Webb RC. Nitric oxide reversibly inhibits the migration of cultured vascular smooth muscle cells. Circ Res 1996;78:225 – 30. [35] Moncada S. Nitric oxide in the vasculature: physiology and pathophysiology. Ann NY Acad Sci 1997;811:60 – 7 Discussion, see pp. 67 – 69. [36] Cable DGT, Schaff HV, Pompili VJ. Recombinant endothelial nitric oxide synthase-transduced human saphenous veins: gene therapy to augment nitric oxide production in bypass conduits. Circulation 1997;96(2):173 – 8. [37] Inaba M, Toninelli E, Vanmeter G, Bender JR, Conte MS. Retroviral gene transfer: effects on endothelial cell phenotype. J Surg Res 1998;78:31 – 6.