Distinct roles of E2F proteins in vascular smooth muscle cell proliferation and intimal hyperplasia

Abstract

Intimal hyperplasia (IH) and restenosis limit the long-term utility of bypass surgery and angioplasty due to pathological proliferation and migration of vascular smooth muscle cells (VSMCs) into the intima of treated vessels. Consequently, much attention has been focused on developing inhibitory agents that reduce this pathogenic process. The E2F transcription factors are key cell cycle regulators that play important roles in modulating cell proliferation and cell fate. Nonselective E2F inhibitors have thus been extensively evaluated for this purpose. Surprisingly, these E2F inhibitors have failed to reduce IH. These findings prompted us to evaluate the roles of different E2Fs during IH to determine how selective targeting of E2F isoforms impacts VSMC proliferation. Importantly, we show that E2F3 promotes proliferation of VSMCs leading to increased IH, whereas E2F4 inhibits this pathological response. Furthermore, we use RNA probes to show that selective inhibition of E2F3, not global inhibition of E2F activity, significantly reduces VSMC proliferation and limits IH in murine bypass grafts.

Cardiovascular disease is the leading cause of death in developed countries and claims almost 40% of all deaths in the United States. Coronary bypass surgery and angioplasty are two major therapeutic modalities for treating cardiovascular disease. Unfortunately, pathological intimal hyperplasia (IH) ensues after these procedures largely due to the proliferation of vascular smooth muscle cells (VSMCs) in the media and their migration into the intima of the treated vessel (1–3). Such proliferation is induced by a number of growth-stimulatory signals that are activated by vascular injury and leads to high long-term failure rates of bypass surgery and angioplasty for treatment of cardiovascular disease (4–12). These failures often cause death and disability and may require repeated treatment by surgery or angioplasty. Accordingly, development of molecular strategies that effectively inhibit such pathogenic cellular processes has been the focus of much research and many clinical trials over the past 20 years (13).

E2F is a family of structurally related DNA binding proteins whose activity is regulated by the retinoblastoma tumor suppressor protein Rb and its related family members, p107 and p130 (14). In mammalian cells, at least eight members of the E2F family of proteins exist, and these proteins possess both distinct and overlapping roles in proliferation, apoptosis, and development (15–21). In general, the mammalian E2F family can be divided into subclasses based upon shared functional properties and sequence homologies. For instance, E2F1, E2F2, and E2F3 are strong transcriptional activators whose accumulation is tightly regulated and in most cell types correlates with increased cell proliferation (22, 23). In contrast, E2F4 and E2F5 are not regulated by cell growth and are involved in the repression of growth-promoting E2F responsive genes (24). Given that E2F proteins are key regulators of the cell cycle and that very little is known about the functional roles of specific E2Fs in the vasculature, here we evaluated the roles of E2F3, an activator E2F, and E2F4, a repressor E2F, in IH by using E2F-deficient mice. Our results indicate that lack of E2F3 reduces IH, whereas lack of E2F4 accelerates this pathological process. These findings prompted us to hypothesize that therapeutic agents that selectively target the growth-promoting E2Fs, E2F1, and E2F3 would be more effective at limiting this pathological process than nonselective E2F inhibitors that also inhibit the activity of the growth-suppressing E2Fs, such as E2F4. To test this hypothesis, we designed and evaluated RNA-probes that selectively inhibit the different E2F proteins. Such selective RNA agents were tested for their ability to inhibit proliferation of VSMCs in vitro and in vivo and their ability to reduce IH in a mouse bypass graft model.

Results

Role of E2F3 and E2F4 in VSMCs.

To determine the predominant E2Fs present in murine VSMC we isolated and cultured VSMC from the aorta of WT mice (Fig. 1). By manipulation of culture conditions we were able to arrest the cells in G₀ of the cell cycle (quiescence) and then stimulate them to enter the cell cycle. This progression was monitored by flow cytometric analysis of DNA content [supporting information (SI) Fig. 6]. To identify which E2F family members are present in these cells, we prepared nuclear extracts from quiescent and stimulated cultures and performed electrophoretic mobility shift assays (EMSA) (Fig. 1A). In quiescent VSMCs (0 h), complexes of p130 with E2F4 and p130 with E2F5 predominate. Once the cells enter the cell cycle (21 h), these complexes are replaced by free forms of E2F3 (isoforms a and b) and E2F4 as the major E2F species (bands supershifted with E2F-specific antibodies are denoted by * in Fig. 1A). These data suggested that E2F3 and E2F4 are expressed in VSMCs and that it would be pertinent to examine E2F3^−/− mice (lacking both E2F3a and E2F3b) as a paradigm for loss of an activator E2F and E2F4^−/− mice as an example of loss of a repressor E2F in a mouse model of IH.

Fig. 1.

Lack of E2F3 reduces IH, whereas lack of E2F4 can accelerate IH. (A) Nuclear extracts were prepared from G₀ and serum-stimulated VSMC cultures. EMSA analysis of E2F complexes was performed. MEF nuclear extracts were used as controls to identify the various E2F complexes. ∗, Indicates bands supershifted with E2F-specific antibodies. pRb, p130, and p107 are pocket proteins that are complexed with different E2Fs in cells. (B) Representstive photomicrographs of stained sections of uninjured or injured carotid arteries of WT, E2F3^−/−, and E2F4^−/− mice 4 weeks after injury. (C) Intimal/Medial Ratio 4 weeks after injury to the vessels. Data comparing WT (n = 10), E2F3^−/− (n = 9), and E2F4^−/− (7) mice are presented as mean ± SEM and demonstrate ∗, P < 0.001 (WT vs. E2F3^−/−) and †, P < 0.01 (WT vs. E2F4^−/−, respectively).

We examined the consequences of E2F3 and E2F4 deficiency in response to carotid artery vascular injury (25, 26). The left panels of Fig. 1B illustrate representative images of the right uninjured common carotid arteries from WT littermates and either E2F3^−/− mice or E2F4^−/− mice (Fig. 1B, Uninjured). Four weeks after injury to the left common carotid arteries, the carotid arteries of WT mice (n = 10) demonstrate significant IH (Fig. 1B, Injured). In contrast, neointima formation was greatly reduced in the carotid arteries of E2F3^−/− mice (n = 9) (Fig. 1B, Injured) resulting in a significant decrease in the intimal/medial area ratio in E2F3^−/− arteries compared with arteries of WT littermates (Fig. 1C). Importantly, a significant increase in the thickness of the vascular smooth muscle layer of the intima was observed in the E2F4-deficient (n = 7) carotid arteries compared with WT littermates (n = 10) (Fig. 1B, Injured). This increase corresponded to approximately a 2-fold increase in the intimal/medial ratio of arteries from E2F4^−/− mice compared with the arteries of WT littermates. These data indicate that lack of E2F4 activity hastens the development of neointima formation in vivo and support a concept where highly specific therapy to suppress activator E2Fs and/or augment repressor E2Fs will limit IH. Thus, we next evaluated whether it would be advantageous to develop inhibitors specific for the growth-promoting E2Fs, particularly E2F3 to attempt to limit this pathological process in WT animals where both E2F3 and E2F4 are present.

Designing and Evaluating siRNAs Against E2F1 and E2F3.

As a strategy to develop more potent and selective inhibitors of the human and murine growth-promoting E2Fs (E2F1 and E2F3), we made use of siRNA technology. Mouse and human sequences were aligned and regions of identity were considered for siRNA targeting. Several siRNAs for each E2F target were designed and chosen for further analysis.

The inhibitory effects of the E2F-specific siRNAs were determined by Western blot analysis of the target E2F protein levels in VSMCs after transfection of the various siRNAs. Transient delivery of siRNAs against E2F1, (F1-2, F1-3, F1-4, F1-5) (Fig. 2A Left) or E2F3, (F3–2, F3–5, F3–6) (Fig. 2A Right) into vena cava VSMC cells specifically reduced the expression of E2F1 and E2F3 proteins. The siRNA-transfected VSMCs with reduced E2F1 protein or E2F3 protein showed significantly decreased proliferation (measured by ³H-thymidine incorporation) in culture (Fig. 2B). The siRNA-mediated effects on proliferation were specific to the E2F targeted and could be partially reversed by co-transfection of a gene encoding either a modified E2F1 or a modified E2F3 transcript that was not degraded by the respective siRNAs (F1-3 and F3-6 respectively) (Rescue) (Fig. 2B). The siRNA effect was greater in E2F4+/+ vena cava VSMCs compared with vena cava VSMCs derived from E2F4^−/− littermates (≈90% vs. ≈20% reduction in proliferation respectively) (Fig. 2C). This observation is consistent with our findings that reveal the opposing roles of E2F3 and E2F4 in the development of restenosis following arterial damage in E2F deficient mice (Fig. 1 B and C).

Fig. 2.

siRNA-mediated depletion of different E2Fs can reduce growth of VSMCs in vitro. (A) Mouse vena cava VSMCs transfected with either a nonspecific control siRNA (siSCR) or siRNAs to either: E2F1 (F1-2, F1-3, F1-4, F1-5) (Left) or E2F3 (F3-2, F3-5, F3-6) (Right). The siRNAs were transfected either individually or together (siE2F3 pool, siE2F1 pool). Western blot analysis was performed on extracts of transfected cells. (B) Mouse vena cava VSMCs transfected with either siSCR or siRNAs to either E2F1 (F1-3, F1-4, F1-5) (Left) or E2F3 (F3-2, F3-5, F3-6) (Right) (n = 3; P < 0.05). siRNAs were either transfected alone or along with a rescue construct (F1-3 Rescue and F3-6 Rescue, respectively). Cells were synchronized at the G1/S boundary by addition of 0.5 mM hydroxy urea (HU). Cells were stimulated to reenter the cell cycle by addition of serum rich media containing ³H-thymidine and lacking HU. Cells were analyzed for ³H-thymidine incorporation as a measure of cell proliferation. (C) VSMCs from vena cavae of WT or E2F4^−/− mice transfected and analyzed as described in B (n = 3; P < 0.001 for siSCR vs. siRNAs; P < 0.01 siRNAs vs. siRNAs+rescue).

The above findings support the hypothesis that “selective” E2F antagonists, such as siRNAs against E2F1 and E2F3, may prove to be more effective at inhibiting VSMC proliferation in vivo than nonselective inhibitors of the entire E2F family.

Development and Evaluation of an RNA Aptamer Inhibitor Specific to E2F3.

To test this hypothesis, we designed a decoy-like inhibitor specific for E2F3 that we compared with the well-established, global inhibitor of E2F activity; an E2F DNA decoy (27, 28). The E2F DNA decoy consists of a double-stranded DNA sequence that mimics the consensus binding sequence of E2F and thus competitively inhibits the DNA binding activity of the entire family of E2F proteins. To develop a selective E2F3 decoy inhibitor, we screened an aptamer library containing ≈10¹⁴ different RNA sequences for aptamers that bound specifically to the human E2F3 protein using SELEX (29–31) (SI Materials and Methods). Like the E2F DNA decoy inhibitor, RNA aptamers should inhibit E2F activity by binding to the E2F3 protein and preventing it from binding DNA. After ten rounds of selection, the resulting aptamers were cloned, sequenced, and organized into families based on sequence similarity. The largest family consisted of aptamers containing the sequence 5′-ACCCNACCCACACTG-3′, where n = T, C, CCC, G, or nothing (Fig. 3A). Most members of this family bound E2F3 with K_d values of 5 nM or less, with aptamer 8-2 binding the tightest with a K_d of 0.2 nM. Importantly, based on in vitro binding data, aptamer 8-2 had a much higher affinity for bacterially purified E2F3 than for the other E2F family members and thus this RNA probe was chosen for further analysis (data not shown).

Fig. 3.

Binding of the 8-2 RNA aptamer and E2F DNA Decoy to individual E2F family members. (A) RNA aptamers with high affinity to E2F3 were isolated from a complex RNA library using SELEX. Aptamer name is listed on the left, K_d for binding to E2F3 is listed on the right, and the number of times a sequence was cloned is shown in parentheses. Aptamer 8-2 (highlighted in red) was further characterized. Aptamer 8-2mut, with the conserved region (underline) of aptamer 8-2 reversed, was used as a negative control. (B) Plots of aptamer 8-2 binding to E2Fs based on PhosphorImager analysis of the EMSA data shown in B. K_is of E2F Decoy and aptamer 8-2 binding to E2Fs derived from binding curves (Lower). (C) Effect of 8-2, 8-2mut, E2F Decoy, and SCR Decoy on mouse VSMC cellular proliferation. The effect of the RNA aptamer and E2F Decoy on cellular proliferation was compared with that of the E2F siRNAs, using ³H-thymidine incorporation analysis as a measure of cellular proliferation.

EMSA was used to assess the binding of aptamer 8-2 to the various E2F proteins expressed in mammalian cells (SI Fig. 7). Mouse embryo fibroblasts (MEFs) were infected with adenovirus expressing the different E2F proteins. Nuclear extracts were then generated from the infected cells. A ³²P-labeled DNA probe containing the E2F binding sites from the DHFR promoter was used to assess binding of the E2F proteins (32). To assess binding efficiency and selectivity the intensities of the E2F-probe complexes in the EMSA assay were quantitated by using a PhosphorImager to calculate the respective K_i measurements (Fig. 3B). Aptamer 8-2 displayed preferential binding to E2F3, with some binding to E2F1 as well. The K_i of aptamer 8-2 for E2F3 was the lowest, at 0.1 nM. The K_i for E2F1 (≈1 nM) was approximately 10-fold higher than for E2F3, whereas the K_i values for E2F2, E2F4, and E2F5 were approximately 1,000-fold higher than that for E2F3. In contrast, the E2F DNA decoy competes with all of the E2Fs for DNA Probe binding within a 4- to 5-fold range demonstrating that RNA aptamer 8-2 is a highly selective binder of the activator E2Fs and the E2F DNA decoy is not (Fig. 3B). A mutant version of aptamer 8-2, 8-2mut, which displays no in vitro binding to E2F3 was used as a negative control in this assay. This mutant RNA aptamer was unable to compete for probe binding with any of the E2F proteins at the lower concentrations (<100 nM) of oligo used (data not shown). Nonspecific binding for all of the RNA oligos was observed at concentrations >100 nM. Next the inhibitory activity of aptamer 8-2 was tested in a cell proliferation assay. Delivery of the 8-2 aptamer to VSMCs in culture reduced proliferation as compared with cells transfected with the mutant control aptamer (Fig. 3C).

Delivery of Oligonucleotide Probes into Venous Grafts in Vivo.

To test whether selective inhibition of E2F activity is more effective than nonselective inhibition, we compared the effects of the selective E2F inhibitors (E2F1/3 siRNAs and aptamer 8-2) with those of the nonselective E2F inhibitor (E2F decoy) on the development of IH in a well-established mouse model of venous bypass grafting (33, 34) (Fig. 4A). Given that one of the main obstacles to effective utilization of oligonucleotide-based therapeutics in vivo is delivery, we evaluated the efficiency of uptake (SI Fig. 8B) of the RNA/DNA oligos probes by the inferior vena cava (IVC). Analysis of uptake revealed that between 50% and 70% of the full-length labeled oligonucleotide probes were transported into the venous grafts at 37°C. In contrast, <15% of the input siRNA was associated with the venous grafts incubated on ice (SI Fig. 8B). Together these data suggest that the oligonucleotides are efficiently delivered into murine IVCs.

Fig. 4.

Uptake of oligos in venous grafts in vivo. (A) Schematic of experimental approach for assessing delivery and activity of DNA or RNA oligos in grafted vessels. (B) Extracts from venous grafts (n = 3) treated with either scrambled siRNA (siSCR) or siRNAs to E2F1 and E2F3 (F1-3/F3-6) were resolved on SDS/PAGE and assessed by immunoblotting. Extracts from MEFs of WT and the various E2F knockout mice were used as markers for various E2F proteins.

We next evaluated the ability of the E2F siRNAs to silence their target genes. Specifically, we demonstrated by Western blot analysis that ex vivo delivery of siRNAs against E2F1 and E2F3 (F1-3 and F3-6 in combination) into mouse venous grafts reduced E2F1 and E2F3 protein expression by ≈48% and 42% respectively as measured with Image J software, 48 h after the grafts had been implanted in mice. Importantly, F1-3 and F3-6 had no effect on the expression of other E2F family members (see E2F2 and E2F4 westerns) (Fig. 4B). Together, these results indicate that ex vivo delivery of the selective E2F siRNA inhibitors results in specific inhibition of the targeted E2Fs in the bypass grafts of mice.

Reduced IH in Mouse Venous Bypass Grafts Correlates with Reduced VSMC Proliferation in the Intima of Venous Grafts.

We determined the effect of the selective inhibitors on the development of IH in our mouse model of venous bypass grafting (33, 34) (Fig. 5). Briefly, inferior vena cava to carotid artery vein graft procedures were performed on animals that received either no treatment (Gel control) (n = 11), negative control scrambled siRNA (siSCR) (n = 13), E2F siRNA (siE2F) (n = 13), aptamer 8-2 (n = 13), aptamer 8-2mut (n = 10), control DNA decoy (SCR Decoy) (n = 12), orE2F Decoy (n = 13). Four weeks postprocedure the grafts were harvested, fixed, sectioned, and stained for analysis. Analysis of the graft sections confirmed that IH had developed in control animals (Gel control), in animals treated with either SCR Decoy, E2F Decoy, or 8-2mut aptamer, as well as in animals treated with scrambled siRNA (siSCR) (Fig. 5). In contrast, treatment of venous grafts with either aptamer 8-2 or with the siRNAs against both E2F1 and E2F3 (F1-3 and F3-6 respectively) (siE2F) resulted in a significant decrease in IH when compared with control samples (Fig. 5). The siRNAs against the E2Fs reduced the Intimal-to-Medial Ratio by ≈42% and ≈57% when compared with siSCR control and Gel control groups respectively, P < 0.0001 (Fig. 5B Right). Of note, the E2F siRNAs had no significant effect on medial area (data not shown). Similarly, the selective E2F3 inhibitory aptamer, 8-2, reduced the Intimal-to-Medial Ratio by 52% whereas the E2F Decoy was much less effective further supporting the notion that selective E2F inhibitors may prove to be more efficacious at limiting IH in vivo (Fig. 5 C and D). The decrease in IH for grafts treated with the siRNA inhibitors against the growth-promoting E2Fs (siE2F) is associated with a ≈3-fold reduction in VSMC proliferation in the intima of the venous grafts as assessed by bromodeoxyuridine (BrDU) incorporation (Fig. 5E).

Fig. 5.

Selective RNA inhibitors reduce IH in venous bypass grafts. (A) Photomicrographs of murine vein-graft 28 days after implantation treated with either pluronic gel alone (Gel control) or gel containing a nonspecific scrambled siRNA (siSCR) or siRNAs against E2F1 and E2F3 (F1-3/F3-6) (Upper). Medial (M) and Intimal (I) layer (Lower). (B) The intimal/medial ratio was plotted as % Inhibition (Lower). Each bar represents an average measurement from Gel control (n = 11), siSCR- (n = 13), and siE2F- (n = 13) treated mice. Intimal/Medial Ratio is reduced by ≈42% after treatment with the siRNAs against the E2Fs compared with control siRNA (siSCR); ∗, P < 0.0001. (C) Photomicrographs from murine vein-graft treated with a binding defective 8-2 mutant RNA aptamer (8-2mut) (n = 10), 8-2 RNA aptamer (n = 13), nonspecific scrambled DNA Decoy oligo (SCR Decoy) (n = 12), and the E2F DNA Decoy (E2F Decoy) (n = 13) (Upper). (D) Scatter plots and bar graphs as in B. ∗, P < 0.0001 compared with 8-2mut. †, P < 0.01 compared with E2F Decoy. (E) Proliferation of VSMCs treated with either nonspecific scrambled siRNA (siSCR) (n = 6) or siRNAs against E2F1 and E2F3 (siE2F) (n = 6) (Upper). Proliferation measured by BrDU-positve VSMCs within the intima of vein grafts, using immunohistochemical analysis.

In summary, our data indicate that the growth-promoting E2Fs (e.g., E2F3) promote IH, whereas the growth-suppressing E2Fs (e.g., E2F4) inhibit IH in mice. These data also indicate that the selective inhibition of growth-promoting E2Fs significantly reduces the development of IH in a mouse model of venous bypass grafting.

Discussion

Here we evaluate the roles of activator (E2F1 and E2F3) and repressor (E2F4) E2Fs during IH. We found that the lack of E2F3 impedes the development of IH, whereas the lack of E2F4 enhances neointima formation in a mouse model of vascular injury (Fig. 1 B and C). This observation led us to the development of RNA probes (siRNA and aptamers) that act as selective inhibitors of the activator E2Fs and we show that these inhibitors can be effectively delivered to vein grafts ex vivo for therapeutic purposes. Specifically, we show that short-term, local delivery of either siRNAs targeting the growth-promoting E2Fs (E2F1 and E2F3) or of an RNA aptamer (8-2) targeting E2F3, results in reduced proliferation of VSMCs in the intima (Fig. 5E) and reduced IH (Fig. 5 A–D) after vein bypass grafting in the mouse when compared with delivery of a nonselective E2F inhibitor (Fig. 5 C and D). Imporatantly, these results indicate that a therapeutic approach that specifically blocks only the proliferative functions of the E2Fs will be most effective for limiting restenosis in the clinic. Moreover, they indicate that inhibitory agents that do not distinguish between the various E2F family members, for example ones that inhibit both E2F3 and E2F4 function (e.g., E2F Decoy), will likely be suboptimal agents for controlling VSMC proliferation and IH.

Consistent with this interpretation of our results, two large randomized Phase III studies recently demonstrated that a nonselective E2F inhibitor, an E2F DNA decoy, did not significantly impact IH and graft failure (35–37). Thus, one explanation for the lack of clinical efficacy of the E2F DNA decoy, which bears the consensus E2F DNA binding site for all of the E2Fs, is that this reagent inhibits the activity of both growth-stimulating E2Fs and growth-repressing E2Fs. Therefore, it is not surprising that its administration may result in a phenotype similar to the one we observe in E2F4^−/− VSMCs treated with E2F3 siRNAs in culture (Fig. 2C); the E2F3 siRNAs are much less effective inhibitors of cell proliferation in VSMCs derived from vena cava of E2F4-deficient mice than WT mice (Fig. 2C). This result suggests that E2F3 and E2F4 play opposing roles in VSMC proliferation and is consistent with our observation that mice lacking E2F4 (a growth-arresting E2F) exhibit increased IH following arterial damage, whereas mice lacking E2F3 (a growth-promoting E2F) show reduced IH compared with WT control mice (Fig. 1 B and C).

Although technical challenges are still associated with the therapeutic application of RNA oligonucleotides, such as specificity, cost of synthesis, delivery, and stability, siRNAs and RNA aptamers are among the fastest developing therapeutic approaches for gene and protein inhibition, respectively (38, 39). What is more encouraging, however, is that in the therapeutic setting of bypass surgery, many of these hurdles appear to be surmountable. For example, the likelihood of the siRNAs having nonspecific toxicity because of nonspecific effects on other mRNAs is greatly reduced because the siRNAs are directly and transiently delivered to bypass grafts ex vivo, which should in turn greatly reduce the potential for systemic toxicity. Moreover, the delivery of these reagents to grafts ex vivo will also substantially reduce the quantity required for treatment and thus reduce the cost of their use in this clinical setting. In addition, currently intensive work is also being performed to further increase oligonucleotide stability and facilitate cellular delivery/tissue bioavailability (38–40). Thus, the clinical utility of these selective E2F RNA inhibitors can be evaluated in the setting of cardiovascular surgery in the near future.

Prevention of restenosis remains a major clinical issue both in the setting of bypass graft surgery and percutaneous coronary intervention. The Prevent IV trial demonstrated that in a double-blind phase III clinical study of over 3,000 patients, ≈46% of patients experience vein graft failure within 18 months (defined as ≥75% stenosis) (36). Studies evaluating drug-eluting stents for reducing restinosis have revealed significant limitations to this approach (41). In particular, both the efficacy of drug eluting stents and their safety due to their potential to increase thrombosis have been called into question (42). Thus therapeutic strategies for ongoing restenosis that are both safe and effective are greatly needed.

Although these preclinical studies in animal models are encouraging, before these reagents can be translated into the clinic it will be important to optimize their therapeutic efficacy and safety profiles. To this end, we have recently shown that we can generate safer and more effective siRNA-based therapies by targeting siRNAs to a specific subset of cells with aptamers that bind to receptors expressed on the surface of target cells. Similarly, the E2F siRNAs and 8-2 aptamer could be targeted to VSMCs by appending them to an RNA aptamer directed against proteins expressed on the surface these cells. In addition, the therapeutic efficacy of these reagents could be enhanced by engineering an aptamer-siRNA chimera composed of the E2F siRNAs in combination with the 8-2 aptamer. Because siRNAs inhibit at the mRNA level and the aptamers at the protein level the use of this chimeric E2F inhibitor may result in sysnergistic inhibition of E2F activity.

In summary, this study provides evidence that different E2F family members have opposing activities in the clinically relevant setting of vascular injury and thus provide a valuable mechanistic insight that can be used to generate more effective therapies for limiting graft failure in patients. Although E2F-functional diversity has been extensively documented for some cell types (e.g., fibroblasts), little is known about the role of different E2F proteins in the vasculature. Our results also highlight the potential utility of RNA probes for dissecting biological processes such as IH. Most importantly, our data suggest that in vivo inhibition of the entire family of E2F proteins will not be therapeutically effective, whereas inhibition of select E2Fs still promises to be an effective way to block pathogenic cellular proliferation after vascular damage.

Materials and Methods

Cell Culture.

VSMCs from thoracic aortas and vena cavae of WT, E2F3^−/−, and E2F4^−/− mice were obtained and cultured as described in refs. 42 and 43.

In Vivo Arterial Injury Model.

Mechanically induced endothelial denudation was performed on the carotid artery of mice as described in ref. 25. Arterial sections were stained with hemotoxylin-eosin and a modified Verhoeff/vanGieson/Masson's stain. Morphometric analysis was performed with National Institutes of Health Image software. Intimal-to-medial ratio is defined by (area of intima)/(total area); where total area = area of intima + area of media.

Nuclear Extracts and Western Blot Analyses.

VSMCs from vena cava of WT mice were transfected twice, using Superfect (Qiagen, Valencia, CA) with 1 μM scr siRNA (control), siRNA against E2F3 (F3-2 alone, F3-5 alone, F3-6 alone, or a combination of F3-1, F3-5, F3-6 (siE2F3 pool), or siRNA against E2F1 (F1-5 alone, F1-4 alone, F1-3 alone, F1-2 alone, or a combination of F1-5,-4,-3,-2 (siE2F1 pool). Cells were allowed to recover for 24 h and assayed for E2F1 or E2F3 protein expression levels. Nuclear Extracts of vena cava VSMCs were resolved on SDS/PAGE and proteins subsequently transferred onto PVDF membrane for immunoblotting. The following primary antibodies were used for immunoblotting: anti-E2F3a (Santa Cruz Biotechnology, Santa Cruz, CA; catalog no. SC-879), anti-E2F1 (SC-251), anti-E2F2 (SC-633), and anti-E2F4 (SC-1082).

siRNAs and E2F Decoy Oligonucleotides.

The siRNA target sequences are as follows: F1-3, aagaucucccuuaagagcaaa and F3–6, aagacuucauguguaguugau. Oligonucleotides used to make the 14-mer Decoy and the scr control were described in ref. 28.

DNA Synthesis Assay.

VSMCs from vena cava of WT or E2F4^−/− mice were seeded in 60-mm dishes at 50% confluency and transfected twice with either 1 μM scrambled siRNA (siSCR), siRNA against E2F1 (F1-3,-4, or -5), siRNA against E2F3 (F3-2, -5, or -6), or F1-3 plus 4 μg of pCDNA3-HAE2F1, or F3-6 plus 4 μg of pCDNA3-HAE2F3amut (Rescue) for 24 h, using Superfect (Qiagen). Cells were transfected with 200 pmol of either 8-2, 8-2mut aptamer, E2F Decoy, or SCR Decoy. DNA Synthesis was assessed as described in ref. 40.

siRNA Up-Take.

The vena cavae from 3 mice per condition were incubated in DMEM containing either a total of 2.5 nmol siSCR and trace amounts (100,000 cpm) of end-labeled ³²P-siSCR, 2.5 nmol of 8-2 aptamer and trace amounts of end-labeled ³²P-8-2, or 2.5 nmol of E2F decoy and trace amounts of end-labeled ³²P-E2F decoy. The oligos were incubated either at 37°C or on ice for 30 min. The vessels were then washed before quantitating uptake of oligos into the vessels. Uptake was determined by measuring ³²P using a Scintillation Counter. The percentage of uptake was measured by dividing the amount of ³²P within the vessels by the input (100,000 cpm) ³²P-labeled oligonucleotide × 100.

Bypass Graft Model.

The venous bypass graft model was performed as described in refs. 33 and 34 (see SI Materials and Methods for details). Morphometric analysis of tissue sections from bypass grafts was performed by using images of ×40 magnification. Perimeter and area measurements for the lumen, neointima, and media were performed by plainimetry, using ImageTool (Version 3.0, University of Texas Health Science Center, San Antonio, TX). Neointima was identified by the criss-cross, random-appearing orientation of smooth muscle cells and by the primarily red color imparted by the prevalence of VSMC cytoplasm and relative absence of collagen. Media was recognized by the circular orientation of VSMCs and the primarily green color imparted by collagen. The measurements were used to create concentric circles to calculate area and perimeter equivalents for each section. The radii of these circles were used to calculate the average thickness of each graft layer.

In Vivo Proliferation.

After venous bypass and treatment with either siSCR (n = 6) or siE2F (n = 6) siRNAs, mice were fed Bromodeoxyuridine (BrDU) (Sigma, St. Louis, MO; 1 mg/ml BrDU in drinking water) in drinking water for four weeks. Treated mice were killed, and the grafts were formalin-fixed and processed for immunohistochemistry with an anti-BrDU antibody (1:100) (DakoCytomation, Glostrup, Denmark).

Statistics.

Statistical analysis was conducted by using a one-way ANOVA. A P ≤ 0.05 was considered a significant difference. In addition, two-tailed unpaired t tests were conducted to compare each treatment group with every other.

Abbreviations

IH

intimal hyperplasia

VSMC

vascular smooth muscle cell

EMSA

electrophoretic mobility shift assay

DHFR

dihydrofolate reductase

MEF

mouse embryo fibroblast

SELEX

systemic evolution of ligands by exponential enrichment.