Antisense to Transforming Growth Factor-β1 Facilitates the Apoptosis of Macrophages in Rat Vein GraftsHeaton N.S. · Wolff R.A. · Malinowski R.L. · Hullett D.A. · Hoch J.R.
W.S. Middleton VA Hospital and the University of Wisconsin School of Medicine and Public Health, Madison, Wisc., USA
Background: The success of peripheral vein grafts is limited by intimal hyperplasia. Transforming growth factor (TGF)-β1 has effects on cell proliferation, apoptosis and extracellular matrix synthesis. We have previously observed positive changes in vessel healing with antisense to TGF-β1. Methods: Adenovirus was used to transduce rat femoral artery vein grafts with antisense to TGF-β1 (Ad-AST) or the sequence encoding the bioactive form of TGF-β1 (Ad-BAT). Grafts were harvested at 1, 2, 4 and 12 weeks and formalin fixed for immunohistochemical studies of the cell markers proliferating cellular nuclear antigen (proliferation) and active caspase 3 (apoptosis). In situ DNA fragmentation assays were also performed to confirm active caspase 3 results. Results: Ad-AST treatment significantly (p = 0.05) increased apoptosis of macrophages inside the internal elastic lamina. In addition, Ad-AST treatment significantly increased the cellularity of the graft at early time points and reduced it at later time points (p = 0.01). Conclusion: The low levels of TGF-β1 in Ad-AST treatment promote apoptosis of macrophages and provide an environment that is more conducive to the proliferation or infiltration of cells that contribute to healthy vessels.
Copyright © 2008 S. Karger AG, Basel
Chronic arterial occlusion is a common condition in aging patients which leads to intermittent claudication, rest pain and tissue loss. Peripheral vein grafting is the preferred method for bypassing arterial occlusion, and the use of autogenous vein has been shown to improve patency results compared with artificial materials . However, even with the use of autogenous vein, this grafting procedure has had reported failure rates of 50% and higher [2, 3]. This high failure rate is due to excessive remodeling of the vein to cope with arterial pressure leading to the development of intimal hyperplasia (IH) .
IH is the major cause of vein graft failure and is characterized by the proliferation of cells and the accumulation of extracellular matrix. This process results in the progressive reduction of the open lumen area. No single cause of IH has been identified, but it has been suggested that the process begins as a healing response that evolves into an unregulated cycle of cellular proliferation and extracellular matrix deposition.
Cytokines and chemokines, which act as cellular regulators, have long been implicated in the formation of IH. Therefore, they are logical targets for gene therapy to attempt to control the healing of the graft and prevent eventual failure . The explanted living tissue used in autogenous vein grafts provides a target for gene therapy ex vivo which avoids the risks associated with systemically exposing the patient to the treatment . This form of gene therapy is practical and hence has a high likelihood of translation into clinical practice if it proves efficacious .
Transforming growth factor (TGF)-β1 is a cytokine that has powerful effects on inflammation and wound healing as well as on normal growth and development . It is known to regulate cell growth, apoptosis, cell differentiation and extracellular matrix synthesis . Increased TGF-β1 levels have been measured in human atherosclerotic plaque, restenotic arteries and other fibrotic diseases, such as systemic sclerosis and liver cirrhosis [10, 11]. The specific effects of this cytokine can vary depending on concentration as well as on cell type with the added complication that other signaling pathways can be initiated by TGF-β1. In light of these facts, TGF-β1 involvement in fibroproliferative processes is poorly understood as a whole .
Several cell types have been implicated in the healing of vessel trauma and in the development of IH. Macrophages are a key cell type that are protected from apoptosis by TGF-β1 , as well as being major factors in the development of IH  and important in recruiting additional cells to the area of the trauma . Myofibroblasts are another cell type important in the inflammation response that are stimulated to deposit extracellular matrix in response to TGF-β1 [16, 17]. Mature smooth muscle cells and endothelial cells, which are indicative of a healthy and mature vessel, undergo apoptosis in response to high levels of TGF-β1 [18, 19]. Here, we consider the possibility that TGF-β1 is a key factor in tipping the scale from a positive and desired mode of healing to an unhealthy fibroproliferative one.
Although TGF-β1 expression is detectable by PCR in cDNA from epigastric vein, in our rat model of vein grafting, 4 h after surgery, there is a many-fold increase in the amount of TGF-β1 mRNA . The mRNA expression peaks at 2 weeks with a gradual decline through 4 and 12 weeks, but even at 12 weeks, the TGF-β1 mRNA expression is still greatly elevated over the expression that was seen in the unoperated epigastric vein. TGF-β1 protein is found in two forms, an inactivated form associated with the TGF-β1 latency protein and an active form cleaved from the latency protein. In the natural course of maturation of the grafted vein, both these forms peak at 4 weeks after surgery . In our previous work, we showed that adenovirus-mediated antisense to TGF-β1 RNA treatment was able to knock down TGF-β1 mRNA production and protein levels in rat femoral artery vein grafts and reduce IH formation . Throughout the 12 weeks of our studies, the antisense treatment results in an approximately 90% reduction in the expression of the latent form and a 60–75% reduction in the bioactive form of TGF-β1. We have also shown that reducing TGF-β1 levels reduces the amount of collagen and modifies the cellular composition of the graft . We hypothesize that the positive changes in vessel healing previously observed can be at least partially attributed to the proliferation and apoptosis of cell types linked to desired and undesired routes of vein healing, respectively.
Materials and Methods
Rat epigastric vein-to-femoral artery interposition grafts were placed and harvested in retired breeder male Lewis rats, as previously described . Grafts were harvested at 1, 2, 4 and 12 weeks after placement. Patency rates of harvested grafts were greater than 95% with non-patent grafts due to technical errors during the grafting process. All animal care complied with the Principles of Laboratory Animals (National Society for Medical Research) and the Guide for the Care and Use of Laboratory Animals (National Academy Press). Experimental procedures were approved locally by the Research Animal Resource Center at the University of Wisconsin-Madison.
For the adenoviral gene therapy, 8 mm of freshly harvested, cannulated and irrigated epigastric vein was placed in an adenoviral solution (107 PFU/ml) for 20 min ex vivo. The 3 adenoviral treatments were: (1) infection with adenovirus expressing the mRNA antisense strand to TGF-β1 (Ad-AST), (2) infection with adenovirus that did not express a gene (Ad-CMVpLpA), and (3) infection with adenovirus expressing the mRNA sense orientation of the TGF-β1 gene (Ad-BAT). The Ad-AST construct expressed the TGF-β1 protein coding region in the antisense orientation (coding region with 130 bp of the 5′ untranslated region, 1,303 bp total). The Ad-BAT construct contained a modified TGF-β1 sequence so that the bioactive form of the TGF-β1 protein was expressed and did not require posttranslational modification . Each construct used the cytomegalovirus promoter.
At the appropriate time point, each vein was pressure fixed with buffered formalin in vivo, removed from the animal and then placed in additional formalin overnight for additional fixing of the tissue. The fixed tissue was then embedded in paraffin with the distal anastomosis of the graft away from the cutting side of the block.
The vein tissue was sectioned starting 3 mm from and cutting toward the distal anastomosis. Data from four 5-μm sections 800 μm apart were averaged per vein due to the eccentric nature of IH formation. Antigen retrieval (1 h at 96°C in 1× TE) was performed before blocking nonspecific antibody binding with 2% bovine serum albumin in phosphate-buffered saline. Sudan Black (0.3% in 70% ethanol; Sigma) was used to block the natural autofluorescence of the tissue.
Monoclonal antibodies diluted in 0.2% bovine serum albumin in phosphate-buffered saline were applied to the tissue: proliferating cellular nuclear antigen (PCNA, 2 μg/ml, murine IgG2A, Invitrogen P8825) and active caspase 3 (1 μg/ml, rabbit IgG, R&D Systems AF835). The PCNA antibody bound the nuclei of cells that were actively dividing, and the caspase antibody bound the active form of the protein which is an integral part of the apoptotic cascade. Fluorescent secondary antibodies were applied to the tissue: Alexa fluor-568 (5 μg/ml, goat anti-mouse IgG2A, Molecular Probes A21134) and Alexa fluor-488 (5 μg/ml, goat anti-rabbit, Molecular Probes A11034). Tissue was mounted in Prolong Gold with DAPI (Invitrogen S36939) to help prevent fluorescent tag degradation under UV light. Total cellularity of the graft was determined by DAPI staining.
Apoptosis data from the active caspase 3 stain was verified qualitatively using an in situ apoptosis detection method that enzymatically labeled nicked chromosomal DNA (Trevigen). DNA fragmentation is a well-established marker of apoptosis and a direct downstream effect of the caspase-3-activated DNase . The in situ DNA fragmentation assay also shows that the active caspase 3 results cannot be attributed to an artifact of the immunohistochemical process.
Images of the vein cross-sections were captured by using a Nikon (Tokyo, Japan) Eclipse E600 microscope and Olympus (Melville, N.Y., USA) DP70 color camera. Areas in the image that stained red, green or blue were measured using MetaMorph image analysis software (Universal Imaging, Downingtown, Pa., USA). For each stained color, a range of pixel intensities specific to the stain were defined so that no unstained cells would be measured. Once the range was determined for each color, the parameters were saved in the program and used to determine the colored areas with no further input from the operator. On each image, the internal elastic lamina was manually traced by the operator to define the boundary between the neointima and the media of the vein. Since the media and adventitia merge as the vessel remodels, these layers were grouped as the media/adventitia.
Due to the eccentricity of the vein remodeling process, the 4 images per vein were averaged to give representative values for the entire vein. For each time point, an average of all grafts (n ≥ 5) was calculated. Rat assignment to each treatment group was randomized and the staining of the tissue sections was similarly randomized to insure that statistically significant differences observed were due to the experimental treatment.
The statistical analysis was performed using SAS statistical software (SAS Institute, Inc., Cary, N.C., USA). p values ≤0.05 were considered significant. The 3 treatments were compared across time using an analysis of variance (ANOVA) model. Data were log transformed before p values were calculated to better meet the assumptions of the ANOVA. The general linear models procedure was used to test the main effects (time and treatment) and their possible interactions. If there was a significant interaction effect, the treatments were compared at each time point. If there was no significant interaction between time and treatment, the main effects (time and treatment) were analyzed. Values are expressed as the mean ± standard error of the mean.
As previously reported , manipulating TGF-β1 levels in the vein changes the types and ratios of cells present. To determine which specific types of cells were proliferating and undergoing apoptosis, images from the current study and those used to determine cell type previously were compared. Any areas that consistently appeared to overlap either proliferation or apoptosis markers with specific cell types across a treatment were noted. Additional stains for specific cell type and either PCNA or caspase were then performed on slides representative of the treatment groups to confirm proliferation or apoptosis of the specific cell types. Cell-specific markers were: actin smooth muscle-1 (ASM-1, 1.25 μg/ml, Sigma F3777), h-caldesmon (15 μg/ml, Sigma C4562) and macrophage (5 μg/ml, Serotech MCA341R, clone ED1 against the rat homologue of CD68). ASM-1 antibody was conjugated FITC. For h-caldesmon and ED1, Alexa fluor-568 IgG1 was applied as the secondary antibody at 5 μg/ml.
There was a significant IH area/cell number difference between the treatments (determined by the DAPI stain and published previously). To account for those differences, we divided the total area staining positive for PCNA and caspase by the total area of nuclei (fig. 1). After controlling for variation in the cellularity of the grafts, statistical significance was seen at 4 and 12 weeks for PCNA and at 2, 4 and 12 weeks for caspase. Specifically, the Ad-AST treatment reduced proliferation at 4 weeks but increased it at 12 weeks. Ad-AST increased active caspase 3 activity at 4 and 12 weeks. In contrast, Ad-BAT increased active caspase 3 activity at 2 weeks (with a trend toward increased caspase activity at 1 week).
|Fig. 1. Statistically significant differences between treatments remain after controlling for cellularity of grafts. Ad-AST treatment decreased proliferation at 4 weeks and increased proliferation at 12 weeks. In addition, Ad-AST treatment increased active caspase 3 at 4 and 12 weeks. Values are the ratios of PCNA and active caspase 3, respectively, divided by the total area of nuclei. All values are means ± SEM. p values represent differences between treatment groups across all time points. a p ≤ 0.05 versus Ad-CMVpLpA; b p ≤ 0.01 versus Ad-CMVpLpA; c p ≤ 0.05 versus Ad-BAT; d p ≤ 0.01 versus Ad-BAT.|
To further dissect the differences between treatments, measurements were made using the internal elastic lamina to separate proliferation/apoptosis in the neointima from proliferation/apoptosis in the adventitial/medial layer (table 1). Both Ad-AST and Ad-BAT treatments had the most significant effects in the neointimal layer.
|Table 1. Vein graft cross-sectional area staining positive for PCNA, caspase and nuclei (DAPI)|
At 1 week, in the neointima, the Ad-AST treatment group had significantly higher PCNA and active caspase 3 than the Ad-CMVpLpA control group. However, the Ad-AST also had approximately 16-fold more cells than Ad-CMVpLpA (table 1, nuclei). In the medial/adventitial layer, at the same time point, there was a trend toward increased PCNA and active caspase and an approximately 2.5-fold increase in cells with Ad-AST treatment. Vein grafts treated with Ad-BAT showed only an increase in cell number in the neointima at this time point.
At 2 weeks, in the neointima, active caspase 3 peaked in the Ad-BAT group and was significantly higher than in the empty vector group.
At 4 weeks, in the neointima, both caspase and PCNA were lower in Ad-AST than in Ad-BAT. Also, Ad-AST was lower in PCNA than in Ad-CMVpLpA. At this time point, there was a significant reduction in PCNA in the medial/adventitial layer by Ad-AST.
At 12 weeks, in the neointima, PCNA and caspase were again higher in Ad-AST than in both Ad-BAT and Ad-CMVpLpA.
As opposed to the complex trends seen with proliferation and apoptosis, the amount of cells in the graft followed a simpler progression over time (table 1, nuclei). In the medial/adventitial layer, the Ad-CMVpLpA control and Ad-BAT treatment started with low cellularity at the 1-week time point. Both showed a steady increase until they peaked at 4 weeks and then dropped slightly at 12 weeks. Overall, the Ad-AST treatment caused higher cellularity levels for the first 2 weeks and then caused lower cellularity levels at the 4- and 12-week time points. In the medial/adventitial layer, at every time point, the Ad-AST treatment had significantly different (higher or lower) levels of cells present in the graft compared with the Ad-BAT treatment. In the neointimal layer, the 4- and 12- week time points showed Ad-AST with a lower level of cells than Ad-BAT.
In addition, in both the medial/adventitial and the neointimal layer, at 1, 4 and 12 weeks, Ad-AST had a significantly different (higher or lower) level of cells present in the graft compared with the Ad-CMVpLpA control.
When stained for caspase and PCNA at 1, 2, 4 and 12 weeks, no location or cell type difference emerged between the treatments except for macrophage apoptosis at 4 and 12 weeks in the Ad-AST-treated grafts. At 4 and 12 weeks, cell types were no longer randomly dispersed throughout the graft and were mostly confined to one region of the vessel. At these time points, cells expressing CD68 (macrophages) were located at or near the internal elastic lamina (on the lumen side) in all 3 treatments. Most of the macrophages contained active caspase 3 in the Ad-AST treatment while few if any macrophages were undergoing apoptosis in the Ad-CMVpLpA or Ad-BAT treatments (fig. 2, top row).
|Fig. 2. Antisense to TGF-β1 promotes apoptosis of macrophages but myofibroblasts have little apoptotic activity. Magnification of representative 12-week images ×100, with insets ×400. Top row: cells expressing CD68 (macrophages) stain red and cells expressing active caspase 3 stain green. Colocalization of the 2 colors appears as yellow. Almost all of the CD68-positive cells in the Ad-AST treatment are undergoing apoptosis which contrasts with little apoptosis of CD68-positive cells in the other 2 treatments. Bottom row: cells expressing ASM-1 (myofibroblasts) stain red and cells expressing active caspase 3 stain green. There are no significant differences in the colocalization of the myofibroblast cell marker (ASM-1) and active caspase 3 between treatments. N = Neointima; A = adventitial/medial layer.|
Cells expressing both ASM-1 and h-caldesmon are defined as mature smooth muscle cells, while cells expressing only ASM-1 are defined as myofibroblasts or synthetic muscle cells [25, 26]. In the Ad-AST treatment, there was a clear layer of mature smooth muscle cells in the adventitial/medial layer directly outside the internal elastic lamina, as shown previously ; however, these cells showed little colocalization with apoptosis or proliferation markers at any time point in any treatment.
Myofibroblasts are the dominant cell type in the neointima in all treatments; however, despite the number of these cells being significantly reduced in the Ad-AST treatment, there was no costaining with apoptosis markers (fig. 2, bottom row). No proliferation of myofibroblasts was seen in the Ad-CMVpLpA or Ad-BAT treatment either, indicating that Ad-AST must be affecting the transformation of fibroblasts into myofibroblasts or migration of myofibroblasts from outside the vein graft wall.
Apoptotic cells with fragmented DNA in the vein graft tissue sections were labeled via enzymatic reaction. This method was used to qualitatively determine that the significantly different levels of active caspase 3 activity previously described could be replicated when looking at a different step in the apoptotic process. DNA fragmentation assays showed the same trends of apoptotic activity between treatments as caspase, with fragmented DNA in the same locations and in the same relative amounts (fig. 3).
|Fig. 3. In situ detection of DNA fragmentation corresponds to active caspase 3 activity. Magnification of 4-week images representative of each treatment group ×400. The column on the left is active caspase 3 (green) with DAPI counterstain (blue). The column on the right is in situ detection of DNA fragmentation (red) and DAPI counterstain (blue). Top row: Ad-AST treatment; middle row: Ad-CMVpLpA treatment; bottom row: Ad-BAT treatment. N = Neointima; A = adventitial/medial layer.|
The effects of TGF-β1 are partially realized by changes in proliferation and apoptosis of cells present in the vein tissue. The increase in TGF-β1 levels in a vein after the trauma of the grafting process is thought to be a component of the healing response in the vascular system . While TGF-β1 is important for vessel healing, its continued presence in the tissue leads to IH formation and vascular disease .
TGF-β1 is a cytokine that has multiple effects on cell proliferation and death depending on its concentration  and the cell type that it is acting upon . When the antisense TGF-β1 treatment is applied to the graft, we see that there are significantly more cells in the vessel at 1 and 2 weeks. This early increase in cell number is likely due to the Ad-AST treatment reducing TGF-β1 and therefore reducing the apoptosis of endothelial cells, increasing the release of paracrine factors by these cells to stimulate infiltration of cells from the bloodstream.
The antisense TGF-β1 treatment speeds the sequence of events involved in healing (whereas the treatment with bioactive TGF-β1 delays progressing through this sequence of events). For the Ad-AST treatment, the major increase in cell number occurred prior to the 1-week time point (the vein used for the graft has fewer cells per cross-section). Further observation shows that in the control (Ad-CMVpLpA), the largest fold increase in cell number occurred between 1 and 2 weeks, and in the Ad-BAT (overexpression of TGF-β1) treatment, this increase is delayed until between 2 and 4 weeks. However, after this initial delay, there are significantly more cells in the Ad-BAT and Ad-CMVpLpA treatments than in the Ad-AST treatment indicating that the cells that have now infiltrated these grafts are expressing the paracrine factors necessary to result in further cell infiltration and proliferation.
Trauma to an artery due to grafting causes TGF-β1 levels to increase dramatically . The Ad-BAT and Ad-CMVpLpA treatments achieve high levels of TGF-β1 early, and this high level is maintained over time. The high level of TGF-β1 initially inhibits apoptosis of cell types that secrete extracellular matrix (myofibroblasts) which helps to heal the vein graft, but which eventually becomes responsible for IH . The Ad-AST treatment keeps TGF-β1 levels low through the early stages of the healing response, reducing the infiltration and maturation of myofibroblasts.
In this study, we did not observe significant changes in cell proliferation, death or survival with Ad-BAT treatment. The rat epigastric vein utilized in our vein graft model is a relatively thin-walled vessel. In this respect, it is perhaps analogous to the human cephalic vein, which has a thinner medial layer than the human saphenous vein. Human cephalic vein grafts achieve long-term patency rates that are inferior to those of saphenous vein grafts with many graft failures resulting from marked IH. We believe our model is analogous to this extreme clinical situation with large amounts of TGF-β1 produced in the control grafted vein . Given this high background of TGF-β1 production, additional TGF-β1 may fail to drive a significant change for many of our measures. For example, Ad-BAT treatment has minimal effect on IH in this model. However, there have been statistically significant increases in the mRNA for tissue inhibitor of metalloproteinase 1 and in the mRNA for collagen . Interestingly, these effects were more pronounced at 1 and 2 weeks than at 4 and 12 weeks. With adenovirus vectors, protein expression at 2 weeks is characteristically 50% of that seen 4 days after transfection. Therefore, a diminishing treatment effect is expected for the Ad-BAT treatment with time.
This contrasts with the Ad-AST treatment which uses the same adenovirus vector. Significant differences in proliferation and apoptosis between treatments were not seen until 4 weeks after grafting. These differences between treatments were stronger still at 12 weeks. Thus, the early reduction in TGF-β1 levels is able to mediate changes that were propagated even after the antisense mRNA was no longer being expressed. We attribute this extended effect to an early resolution of chronic inflammation.
We have observed neutrophils infiltrating the cell wall, but macrophages are the major infiltrating cell type. At 1 week, macrophage can be more than 50% of the cells found in the vessel wall. Vein grafts treated with adenovirus designed to increase TGF-β1 have had approximately twice as many macrophages within the intima at 4 and 12 weeks as control veins. This treatment has also resulted in an additional smaller increase in macrophage within the media/adventitia. Treatment with Ad-AST did not change macrophage numbers in the neointima compared with controls. However, this treatment did reduce macrophage numbers in the media/adventitia to less than 50% of those seen in the controls at 4 and 12 weeks . In these studies, we have shown that Ad-AST treatment leads to increased macrophage apoptosis. In all treatments, macrophages are present on the internal elastic lamina at 4 and 12 weeks. In the Ad-AST treatment, almost every macrophage on the internal elastic lamina is undergoing apoptosis while very few macrophages are undergoing apoptosis in the Ad-CMVpLpA and Ad-BAT treatments. Taking into consideration that macrophages migrate from the lumen to the adventitia by passing through the internal elastic lamina, the apoptosis of macrophage at the internal elastic lamina explains the previous observation of fewer macrophages reaching the media/adventitia.
After controlling for differences in cell numbers, the only significant difference in proliferation was in the Ad-AST treatment. At the 12-week time point, proliferation was increased by the Ad-AST treatment, but not as much as apoptosis was increased by this treatment. Proliferation markers did not colocalize exclusively with the cell type markers used in this study. However, we do observe an increase over time in the number of mature smooth muscle cells in the medial/adventitial layer of Ad-AST-treated vein grafts. This layer of smooth muscle cells may be proliferating between the 4- and 12-week time points when data were not collected, or smooth muscle cells may be migrating into the graft.
The Ad-AST treatment has been shown to reduce IH formation and cause the vessel to take on an artery-like phenotype. We have presented data that suggest these changes are at least partially mediated by causing the cell types responsible for the deposition of IH and other negative healing responses to undergo apoptosis (macrophages) or reducing their infiltration (myofibroblasts). In contrast, the Ad-BAT and Ad-CMVpLpA treatments showed little apoptosis of macrophages or myofibroblasts which allowed for unchecked extracellular matrix synthesis and the development of an extensive neointima.
This material is based on work supported by the Office of Research and Development, Biomedical Laboratory Research and Development Service, Department of Veterans Affairs. Glen Leverson, PhD, of the Department of Surgery Biostatistics Office assisted in the statistical analysis. Kristin Jakubowski, BS, provided valuable insight during the preparation of the manuscript. Ad-CMVpLpA was provided by the University of Michigan Medical Vector Core.
Dr. Randal Wolff
W.S. Middleton VA Hospital
2500 Overlook Terrace, G-wing Rm G9
Madison, WI 53705 (USA)
Tel. +1 608 265 9143, Fax +1 608 265 9144, E-Mail email@example.com
Received: August 10, 2007
Accepted after revision: December 4, 2007
Published online: March 20, 2008
Number of Print Pages : 10
Number of Figures : 3, Number of Tables : 1, Number of References : 31
Journal of Vascular Research (Incorporating 'International Journal of Microcirculation')
Vol. 45, No. 5, Year 2008 (Cover Date: August 2008)
Journal Editor: Pohl U. (Munich), Meininger G.A. (Columbia, Mo.)
ISSN: 1018–1172 (Print), eISSN: 1423–0135 (Online)
For additional information: http://www.karger.com/JVR