Synthetic Retinoid Am80 Suppresses Smooth Muscle Phenotypic Modulation and In-Stent Neointima Formation by Inhibiting KLF5
http://www.100md.com
循环研究杂志 2005年第11期
the Department of Cardiovascular Medicine (K.F, A.I., Y.O., H.I., G.N., T.S., K.M., R.N.) and Nano Bioengineering Education Program (I.M.), Graduate School of Medicine, University of Tokyo, School of Biomedical Science, Tokyo Medical and Dental University (H.K.)
Research Foundation Itsuu Laboratory (K.S.), Tokyo, Japan.
Abstract
Modulation of smooth muscle cell (SMC) phenotype plays a central role in neointima formation. We recently demonstrated that Am80, a synthetic retinoic acid receptor -specific agonist, inhibits the activity of the transcription factor KLF5, which is essential for neointima formation after vascular injury. In the present study, we aimed to further analyze the mechanism by which Am80 inhibits KLF5 and the effects of inhibiting KLF5 on SMCs and vascular lesion formation, as well as to evaluate potential of Am80 for use in the prevention of in-stent neointima formation. We found that Am80 inhibited both the expression and transcriptional function of KLF5. Of particular interest was our finding that KLF5 forms a transcriptionally active complex with unliganded RAR/RXR heterodimer on the PDGF-A promoter; Am80 disrupts this complex, thereby inhibiting KLF5-dependent transcriptional activation. Knocking down KLF5 using small interfering RNA suppressed serum-induced downregulation of SMC differentiation marker gene expression in cultured SMCs, and haploinsufficiency of KLF5 in mice attenuated phenotypic modulation of SMCs after vascular injury, indicating that KLF5 plays a key role in the control of SMC phenotype. Am80 augmented expression of the SMC differentiation marker genes in culture and within the vessel walls, and oral administration of Am80 significantly inhibited in-stent neointima formation in a rabbit stent-placement model. Taken together, these results demonstrate that KLF5 plays an important role in the control of SMC phenotype after vascular injury and suggest the feasibility of using Am80, delivered systemically and/or with a drug eluting stent, to prevent in-stent neointima formation.
Key Words: stent smooth muscle restenosis retinoid phenotypic modulation
Introduction
The neointima that forms after percutaneous coronary intervention (PCI) is mainly composed of smooth muscle cells (SMCs) and the extracellular matrix they produce.1 In response to external stresses such as mechanochemical stress, the effects of various humoral factors, and direct interaction with inflammatory cells, SMCs undergo phenotypic modulation. These phenotypically modulated SMCs proliferate, migrate, and produce various paracrine factors, extracellular matrix, and matrix proteases, thereby promoting the remodeling of the vascular wall.2,3 Consequently, SMCs are considered to be one of the most important targets of therapies aimed at preventing restenosis.
We recently showed that the Kre筽pel-like transcription factor KLF5 mediates stress-induced vascular remodeling, and that haploinsufficiency of KLF5 in KLF5+/eC mice much reduces neointima formation in vascular injury models.4 Moreover, we identified a novel synthetic retinoid, Am80, to be a potent inhibitor of KLF5.4 Am80 is a retinoic acid receptor (RAR) -specific agonist that has been safely used to treat acute promyelocytic leukemia. As expected from its in vitro inhibitory effects on KLF5, Am80 inhibited neointima formation in a mouse model in which the femoral artery was injured by cuff placement.4 It has not yet been directly determined, however, whether Am80 inhibits neointima formation after stenting. Such an analysis is important because the mechanisms underlying neointima formation likely differ between injury models5 and stenting is the most frequently performed PCI. It also remains unclear how inhibiting KLF5 with Am80 leads to suppression of neointima formation.
In the present study, we analyzed the mechanism by which Am80 inhibits KLF5 and the effects of inhibiting KLF5 on SMCs and vascular lesion formation, and then evaluated the potential of Am80 for use in the prevention of in-stent neointima formation. Our results demonstrate that KLF5 plays a key role in the control of SMC phenotype and that the KLF5 inhibitor Am80 is a novel modulator of SMC function. They also establish the feasibility of using Am80 to prevent in-stent restenosis.
Materials and Methods
Rabbit Stent Placement
Twenty-two male New Zealand White rabbits (Oriental Yeast Co, Ltd, Tokyo, Japan) weighing 3.5±0.3 kg were randomly assigned to treatment with Am80 or placebo. The Am80 was administrated orally at a dose of 1 mg · kgeC1 · deC1 beginning 1 day before stenting and continuing for 28 days thereafter. All animals received aspirin (40 mg/d) daily during the same period. Under general anesthesia, the iliac arteries were visualized using intravascular ultrasound imaging, after which they were injured by placement of a 2.5 mmx16 mm JOSTENT (JOMED) stent (6 to 8 atm, 30-second balloon inflation) under fluoroscopic guidance. Taking into consideration the vascular diameters, the inflation pressure of the stent balloons was determined so that the balloons were dilated to the standard stent-to-artery ratio (1.0:1&1.1:1).10 The final diameters of the lumens were then measured to ensure the appropriate stent-to-artery ratio and to rule out stent malapposition. All animal procedures strictly adhered to the guidelines for animal experiments of the University of Tokyo. An enhanced Materials and Methods section is available online at http://circres.ahajournals.org.
Results
KLF5 Plays a Role in Controlling SMC Phenotype
Modulation of the SMC phenotype is invariably observed in vascular lesions, and the phenotypically modulated SMCs are known to actively contribute to vascular disease development and progression. Our earlier observation that KLF5+/eC mice exhibit much reduced neointima formation after vascular injury4 strongly suggests that KLF5 is involved in controlling SMC phenotype. To test that idea, KLF5 was knocked-down in cultured SMCs using small interfering RNA (siRNA). SMCs were transfected with either a KLF5-siRNA expression vector or a negative control vector and then cultured in the serum-free, defined medium for 48 hours, after which the cells were stimulated with fetal bovine serum for 24 hour. As previously reported,6 serum stimulation led to downregulation of SM -actin and SM-MHC expression in cells transfected with the negative control vector, whereas expression of KLF5 and its target gene, PDGF-A, was upregulated (Figure 1A through 1D). Transfection of KLF5-siRNA vector resulted in reduced basal expression of KLF5 and PDGF-A and inhibition of their serum-induced upregulation (Figure 1A and 1B). In addition, knocking down KLF5 inhibited the serum-induced downregulation of SM -actin and SM-MHC (Figure 1C and 1D). KLF5 thus appears to be required for serum-induced modulation of SMC differentiation marker gene expression.
To further analyze the role of KLF5 in phenotypic modulation, the common carotid arteries of KLF5+/eC and wild-type mice were ligated, after which neointimal growth and SMC phenotype were analyzed. Histomorphometry of vessels 7 days after ligation showed that intima/media area ratios were significantly reduced in KLF5+/eC mice (supplemental Figure I). As previously reported,7 their was minimal expression of KLF5 in uninjured carotid arteries from wild-type animals. Basal expression of KLF5 in KLF5+/eC mice tended to be lower than in wild-type, but the difference was not statistically significant (Figure 1E). On day 7 after injury, however, KLF5 expression was markedly increased (12-fold over basal) within the arteries of wild-type mice and was also upregulated in KLF5+/eC animals, but to a significantly lesser degree (Figure 1E). Likewise, upregulation of PDGF-A was greatly reduced in KLF5+/eC animals (Figure 1F). In uninjured arteries, there were no significant differences in the expression of SM -actin and SM-MHC in wild-type and KLF5+/eC mice (Figure 1G and 1H). The downregulation of SM -actin and SM-MHC expression seen in the injured arteries of wild-type animals, however, was significantly attenuated in KLF5+/eC animals, suggesting that haploinsufficiency of KLF5 curtailed phenotypic modulation in response to the injury.
Am80 Inhibits KLF5 Expression and Transcriptional Activity in SMCs
We previously showed that KLF5 binds to RAR and that Am80 inhibits the transcriptional activity of KLF5.4 The molecular mechanism(s) by which Am80 inhibits KLF5 is not yet clear, however. To address that question, we first tested the possibility that Am80 inhibits KLF5 expression in SMCs. As shown in Figure 2A, Am80 suppressed KLF5 mRNA expression at concentrations as low as 0.1 eol/L. Likewise, all-trans retinoic acid (atRA) also inhibited KLF5, but was an order of magnitude less potent; 0.1 eol/L atRA did not have significant effect on KLF5 levels. In addition, we found that Am80 suppressed the KLF5 promoter activity normally induced by serum stimulation (Figure 2B). Those data suggest that Am80 suppresses KLF5 expression at least in part by inhibiting its transcription, although other post-transcriptional mechanism might also play a role.
We then examined the extent to which Am80 also inhibited the functionality of KLF5 in cultured SMCs overexpressing exogenous KLF5. The level of overexpressed Flag-tagged KLF5 protein was >10-fold higher (data not shown) than that of the endogenous protein, and Am80 had no effect on the levels of Flag-tagged KLF5 protein generated from the CAG-promoter driven construct. Moreover, our reporter analysis showed that, in the absence of Am80, this overexpression of KLF5 resulted in a 5.5-fold increase in PDGF-A promoter activity (Figure 2C). Despite having no effect on KLF5 expression levels under these conditions, Am80 significantly inhibited both KLF5-dependent activation of the PDGF-A promoter (Figure 2C). As expected from the inhibition of PDGF-A promoter, Am80 inhibited PDGF-A mRNA expression in SMCs (Figure 2D). atRA also inhibited PDGF-A expression, though to a lesser extent. Taken together, these findings indicate that Am80 inhibits KLF5-dependent PDGF-A transactivation by inhibiting both the expression and function of KLF5.
KLF5, RAR, and RXR Form a Transcriptional Complex on the PDGF-A Promoter That Is Disrupted by Am80
To further explore the mechanism by which Am80 inhibits KLF5-mediated gene transactivation, we assessed the function of RARs in PDGF-A transcription. Combinations of constructs expressing KLF5, RAR, and retinoic X receptor (RXR) were cotransfected into cells along with the PDGF-A promoter reporter. Whereas overexpression of KLF5 increased the reporter activity 5.5-fold over that seen with the reporter alone, overexpression of RAR and/or RXR had little effect on PDGF-A reporter activity (Figure 3A). Furthermore, coexpression of KLF5 and RAR led to a significantly greater activation of the reporter than was seen with KLF5 alone (9.5 vs 5.5-fold increase in activity), though coexpression of KLF5 with RXR did not. Am80 concentration-dependently inhibited the reporter activity induced by overexpression of KLF5, RAR, and RXR.
On the basis of the PDGF-A reporter analysis, we hypothesized that unliganded RAR serves as a coactivator in KLF5-dependent PDGF-A transcription and that Am80 disrupts the complex formed by KLF5 and RAR. To test that idea, Flag-tagged KLF5, RAR, and RXR were overexpressed in NIH3T3 cells incubated with or without Am80 or atRA (1 eol/L), and the cell lysates were subjected to coimmunoprecipitation. As shown in Figure 3B, KLF5 was detected in precipitates pulled down with either anti-RAR or anti-RXR antibody, whereas RAR and RXR were detected in precipitates pulled down with anti-Flag antibody (Figure 3C and 3D). RAR and RXR were also detected in precipitates pulled down with antibody against their respective counterparts, indicating that heterodimerization took place between RAR and RXR (Figure 3C and 3D). Given that KLF5 is known to physically interact with RAR,4 these data suggest that either KLF5 interacts with both RAR and RXR, or it directly interacts only with RAR within the RAR/RXR heterodimer. To resolve the nature of the interaction of KLF5 with RAR and RXR, we overexpressed Flag-KLF5 and RXR without RAR. Under those conditions, KLF5 was not detected in precipitates pulled down with anti-RXR antibody, nor was RXR detected in precipitates pulled down with anti-Flag antibody, indicating that KLF5 does not directly bind to RXR (Figure 3E). Treating cells with Am80 significantly reduced levels of KLF5 in precipitates pulled down with anti-RAR or anti-RXR antibody (Figure 3B). Likewise, Am80 reduced levels of coimmunoprecipitated RAR and RXR pulled down with Flag-KLF5 (Figure 3C and 3D). On the other hand, Am80 did not affect coimmunoprecipitation of RAR and RXR (Figure 3C and 3D). Apparently, Am80 inhibits the interaction between KLF5 and the RAR/RXR heterodimer. Interestingly, atRA at the same concentration (1 eol/L) did not effectively inhibit the interactions between KLF5 and RAR/RXR in these experiments.
We next performed chromatin immunoprecipitation (ChIP) assays to further confirm the involvement of the RAR/RXR heterodimer in the activation of the PDGF-A promoter by KLF5. Using real-time polymerase chain reaction (PCR), we analyzed the abundance of the sequences of interest in immunoprecipitated chromatin samples (Figure 3F and 3G). SMCs were incubated with or without Am80 for 2 hours and then fixed with formaldehyde, after which the cross-linked chromatin samples were subjected to immunoprecipitation with anti-KLF5, anti-RAR, or anti-RXR antibody.8 The results showed that KLF5, RAR, and RXR all bound to the endogenous PDGF-A promoter in SMCs and that Am80 markedly reduced the abundance of the PDGF-A promoter sequence in the immunoprecipitates pulled down with anti-KLF5, anti-RAR, or anti-RXR antibody. Taken together, the results suggest that KLF5 and unliganded RAR/RXR bound to and activated the promoter as components of a transcriptional complex and that Am80 disrupted the interaction between KLF5 and RAR/RXR, thereby inhibiting their binding to the promoter and thus PDGF-A transcription.
Am80 Affects the Differentiation State of SMCs
That Am80 inhibits KLF5 activity suggest it might also affect SMC phenotype. As expected, adding Am80 to cultured SMCs upregulated expression of the SMC differentiation marker genes SM-MHC and SM -actin (Figure 4A and 4B), suggesting that Am80 promotes differentiation in phenotypically modulated cultured SMCs. Moreover, Western blot analyses confirmed that Am80 induced expression of SM-HMC protein at concentrations as low as 0.01 eol/L (Figure 4C). Although atRA also induced expression of SM -actin and SM-MHC at a higher concentration than Am80, the extent of expression were less than was induced by Am80.
Am80 Inhibits Proliferation and Migration of SMCs
That phenotypic modulation is a prerequisite for proliferation and migration of differentiated SMCs3 suggested to us that inhibition of phenotypic modulation by Am80 might suppress the pathophysiological proliferation and migration of SMCs seen in vascular diseases, such as in-stent neointima formation. Consistent with that idea, we found that Am80 dose-dependently inhibited both mitogen-stimulated DNA synthesis and cell proliferation in cultured human coronary arterial SMCs and rat aortic SMCs (Figure 5A through 5D). SMC proliferation was significantly inhibited at a concentration of 0.1 eol/L, and the same concentration also significantly inhibited fetal bovine serum-induced SMC migration (supplemental Figure II). We also examined the effect of Am80 on proliferation of endothelial cells because re-endothelialization is important for inhibition of thrombosis and for stabilization of the remodeling process, but found that Am80 has no effect on DNA synthesis or proliferation in cultured human umbilical vein endothelial cells (HUVECs) (Figure 5E and 5F), suggesting that Am80 selectively inhibits SMC proliferation within the vascular wall.
Am80 Reduces In-Stent Neointima Formation
We then tested whether orally administered Am80 would reduce in-stent neointima formation after placing metal coronary stents in the iliac arteries of New Zealand White rabbits. We initially selected the appropriate dosage of Am80 by administering a series dosages to rabbits and determining the relationship between the oral intake and plasma concentration (supplemental Table IA). A dose of 1 mg · kgeC1 · deC1 was selected so that the plasma Am80 concentration of the rabbits that underwent stent placement reached between 0.1 to 1 eol/L (supplemental Table IB). To compare the injury caused by stent placement in rabbits receiving Am80 or placebo, we used the injury score originally described for pigs.9 Using that scale, we found that the injuries caused by stent placement were similar in both the Am80 and placebo groups (0.15±0.36 versus 0.15±0.36); a few rabbits in each group showed laceration of the internal elastic lamina, but no injuries with higher scores were detected. When the stented arteries were harvested 28 days later and subjected to histological analysis, however, we found that the Am80 group showed significantly less neointima formation than the placebo group (Figure 6A through 6F).
Histomorphometry of vessels from 7 animals in each group showed that although the medial areas were similar, the intimal area, intima/media area ratios, and percent stenosis were significantly (P<0.05) reduced in the Am80 group, which is indicative of the diminished neointimal growth (Figure 6G through 6J). It is also noteworthy that during this experiment, no rabbits showed any signs of adverse effects of Am80 on body weight, appetite, blood chemistry, or major organ histology.
When we evaluated endothelialization of in-stent lesions, we found that despite the marked inhibition of neointimal growth, high magnification light microscopy revealed that almost the entire circumference of the lumens of the injured arteries on day 28 was completely covered by ECs in both groups (data not shown).
Am80 Inhibits Phenotypic Modulation of SMCs In Vivo
To investigate the mechanism by which Am80 reduced in-stent neointima formation, we first analyzed the effect of Am80 on SMC growth and apoptosis within the vascular walls. Because SMC proliferation is more pronounced during the first week after stent placement than at later times and shows more modulated phenotypes,10 the stent-related lesions were analyzed on day 7 after stenting. We found that in control animals, fibrin and cellular deposits had already formed adjacent to the stent struts by this time, as has been reported previously,10 but such lesions were barely detectable in Am80-treated animals, and the intimal area was significantly reduced (supplemental Figure III). Thus, Am80 clearly inhibited lesion formation during the first week after stent placement. We also analyzed cell proliferation by staining the tissue with proliferating cell nuclear antigen (PCNA) (Figure 7). Within the intimal lesions of the control animals, 19% of cells were PCNA-positive. Within the media, PCNA-positive SMCs were often found in the areas beneath the intimal lesions, and 7.2% of cells showed PCNA positivity in the all medial areas. By contrast, only 0.7% of cells were PCNA-positive in the media of the Am80-treated animals, indicating that Am80 markedly inhibited proliferation of medial SMCs (Figure 7E).
The early inhibition of intimal lesion formation suggests Am80 might also interfere with other processes involved in early lesion formation, such as platelet aggregation, thrombus formation, and monocyte infiltration.1 On the other hand, Am80 apparently had no effect on apoptotic processes at this time point, as very few terminal deoxynucleotidyl transferase biotin-dUTP nick end labeling (TUNEL)-positive apoptotic cells were found in sections obtained from either group, and the number of TUNEL-positive cells was unaffected by Am80 (supplemental Figure IV).
Consistent with earlier findings,11 we also found that on day 7, expression of SM -actin protein was diminished within the stented vascular walls of control animals (supplemental Figure V). In clear contrast, many more SMCs were positive for SM -actin in Am80-treated animals. At the same time, the level of SM -actin mRNA expression was much lower in the stented vascular walls of control animals than Am80-treated animals (Figure 8A). In similar fashion, Am80 also inhibited downregulation of SM-MHC gene expression (Figure 8B and 8C). Expression of total SM-MHC transcripts, which included both the SM1 and SM2 isoforms, was significantly stronger in the Am80-treated group on day 7, as was expression of the SM2 isoform, a marker for mature SMCs. On the other hand, expression of SMemb, a marker for phenotypically modulated SMCs that was induced after stenting in control animals, was markedly reduced in Am80-treated animals (Figure 8D).
By day 28, Am80 treatment led to further increases in expression of SM-MHC and SM -actin (Figure 8A through 8C), suggesting acceleration of SMC differentiation. Relative levels of SM1 protein in the neointima as compared with those in the media were higher in the Am80 group than in the placebo group on day 28, as were levels of SM -actin protein, suggesting that Am80 treatment also induced the SMC differentiation marker genes at the protein level (supplemental Figure VE through VH). These patterns of expression indicate that Am80 inhibits phenotypic modulation of SMCs and promotes SMC differentiation after stenting in vivo. As expected, expression of KLF5 and its target gene PDGF-A was markedly inhibited in Am80-treated animals (Figure 8E and 8F), suggesting that in vivo, Am80 suppresses phenotypic modulation at least in part by inhibiting KLF5, just as it does in vitro. In contrast, expression of myocardin, a coactivator of serum response factor, was unaffected by Am80 treatment.
Discussion
In the present study, we have shown that KLF5 is important for control of SMC phenotype in vitro and after vascular injury in vivo. We also showed that the synthetic retinoid Am80 inhibits KLF5 by interfering with both its expression and its interaction with RAR. In this way, Am80 affected the phenotype, proliferation, and migration of cultured SMCs and inhibited neointima formation after stent placement in vivo. These results establish KLF5 as a novel molecular target for therapeutic intervention in the treatment of vascular disease and provide evidence of the feasibility of using its antagonist Am80 to prevent neointima formation and in-stent restenosis.
The phenotypic modulation of SMCs requires the coordinate regulation of a number of genes, which means that SMC phenotype is very likely governed by the activity of a transcription factor network.2,3 Our previous studies suggest that KLF5 is a component of just such a transcriptional network,4,7 making it a potentially useful target for therapeutic intervention in vascular disease. To assess the role of KLF5 in the control of the SMC phenotype, we analyzed the effects of knocking down KLF5. In cultured SMCs, KLF5 knockdown attenuated downregulation of SM-MHC and SM -actin by serum (Figure 1), which mimics phenotypic modulation in terms of SMC marker gene expression.6 Similarly, SMCs in the injured carotid arteries of KLF5+/eC mice exhibited a more differentiated phenotype than those in wild-type animals, judging from the expression of SMC differentiation marker genes (Figure 1).
So how does KLF5 control SMC phenotype One possibility is that KLF5 directly controls expression of SMC marker genes; it might positively regulate SMemb on one hand and negatively regulate SM -actin and SM-MHC on the other. Another possibility is that KLF5 controls expression of paracrine factors, which in turn controls the phenotype and differentiation state of SMCs.12 Indeed, we previously identified PDGF-A as one of the downstream targets of KLF54 and, as expected, PDGF-A expression was diminished in KLF5-siRNA transfected SMCs, in the vascular lesions in KLF5+/eC mice (Figure 1F), and in the stented vascular walls of rabbits given Am80 (Figure 8F). It is also likely that KLF5 controls other unidentified genes,13 which may cumulatively affect SMC phenotype.
The results of the present study also provide novel insight into the actions of RARs in SMCs. Reporter (Figure 3A) and ChIP (Figure 3F and 3G) analyses showed that RAR/RXR is involved in the transcriptional control of PDGF-A. Because the proximal promoter region of PDGF-A does not contain a canonical retinoic acid response element (RARE), it is likely that RAR/RXR binds to the promoter via interaction with KLF5. Consistent with that idea is our finding that KLF5 directly interacts with RAR (Figure 3); of particular interest to us is the finding that Am80 inhibited that interaction as well as KLF5-dependent transactivation of the PDGF-A promoter. This suggests that unliganded RAR/RXR heterodimer and KLF5 form a transcriptionally active complex on the PDGF-A promoter, and that by disrupting formation of that complex, Am80 inhibits the activity of KLF5. Penderies et al14 similarly reported that unliganded RAR enhances Smad-dependent transcription and that this effect was inhibited by a RAR agonist, which disrupted the interaction between RAR and Smads. Together with those data, our findings suggest a novel mode of RAR action in which unliganded RARs serve as coactivators for other transcription factors. In this mode, RAR agonists may inhibit coactivation by RARs, which contrasts with the classical RARE-dependent transactivation induced by RAR agonists. Notably, earlier studies have also revealed yet another RARE-independent mode of RAR action in which RARs transrepress other transcription factors, such as activator protein-1.15 Given these novel actions, one needs to consider a broader range of genes as potential targets of RARs. Interestingly, a number of possible retinoid-regulated genes have been identified in SMCs, many of which do not have canonical RARE in their transcriptional regulatory regions.16 It is very likely that some of these genes are regulated by RARs functioning as coactivators and/or corepressors. Moreover, results of the present study suggest that KLF5 is one of the key mediators of retinoid actions in SMCs.
We also examined the effects of Am80 on SMC proliferation, migration, and apoptosis as they relate to neointima formation and in-stent restenosis.1,5 Am80 clearly inhibited proliferation of medial SMCs 7 days after stent placement (Figure 7). Although higher concentrations of Am80 induced apoptosis among cultured SMCs, there was no apparent increase in the numbers of apoptotic cells within the vascular walls 7 days after stent placement (supplemental Figure IV). Further study will be required to fully assess the importance of apoptosis in the in vivo effects of Am80, but the fact that higher concentrations of Am80 were required to induce apoptosis than to inhibit proliferation among cultured SMCs suggests that its antiproliferative action is central to Am80’s overall inhibition of neointima formation. In addition, Am80’s inhibition of neointima formation on day 7 after stent placement suggests that it affects nonmuscle cells as well as SMCs. This is because the early lesion is known to be formed by various cell types, including macrophages, lymphocytes, and platelets. Interestingly, in that regard, atRA has been shown to promote fibrinolysis and to inhibit thrombosis, platelet aggregation, and inflammation.17
Previous studies have shown that atRA reduces neointima formation in balloon injury models,16,18 making it an attractive candidate for local delivery by drug-eluting stent. Unfortunately, however, its instability with respect to heat, light, oxidation, and sterilization markedly diminishes its utility. The higher stability and selectivity of Am80 would be expected to enhance its utility and broaden its applicability to other ailments in which abnormal cellular differentiation and chronic inflammation play important roles. The dramatic results recently obtained with drug-eluting stents using sirolimus and paclitaxel have stimulated significant enthusiasm among interventional cardiologists. There are still potential problems, however. For example, the strong antiproliferative effects of these agents may delay the healing of in-stent lesions, causing thrombosis.19 The unique biological activities of Am80 might provide a means to overcome these problems by inhibiting tissue remodeling and stabilizing the vascular wall.
Acknowledgments
We gratefully acknowledge Eriko Magoshi, Noriko Yamanaka, and Michiko Hayashi for their excellent technical assistance and Kazuya Shibaki, Koji Osaka, Hideo Fumiyama, and Sachie Hayashi for their great support. This study was supported in part by a Grant-in-Aid from the Ministry of Education, Culture, Sports, Science, and Technology, Japan, a research grant from National Institute of Biomedical Innovation, Japan (to R.N.), and research grants from Takeda Science Foundation, Sankyo Foundation of Life Science, Mitsubishi Pharma Research Foundation, and Tokyo Biochemical Research Foundation (to I.M.).
References
Schwartz RS, Henry TD. Pathophysiology of coronary artery restenosis. Rev Cardiovasc Med. 2002; 3: S4eCS9.
Manabe I, Nagai R. Regulation of smooth muscle phenotype. Curr Atheroscler Rep. 2003; 5: 214eC222.
Owens GK, Kumar MS, Wamhoff BR. Molecular regulation of vascular smooth muscle cell differentiation in development and disease. Physiol Rev. 2004; 84: 767eC801.
Shindo T, Manabe I, Fukushima Y, Tobe K, Aizawa K, Miyamoto S, Kawai-Kowase K, Moriyama N, Imai Y, Kawakami H, Nishimatsu H, Ishikawa T, Suzuki T, Morita H, Maemura K, Sata M, Hirata Y, Komukai M, Kagechika H, Kadowaki T, Kurabayashi M, Nagai R. Kre筽pel-like zinc-finger transcription factor KLF5/BTEB2 is a target for angiotensin II signaling and an essential regulator of cardiovascular remodeling. Nat Med. 2002; 8: 856eC863.
Indolfi C, Mongiardo A, Curcio A, Torella D. Molecular mechanisms of in-stent restenosis and approach to therapy with eluting stents. Trends Cardiovasc Med. 2003; 13: 142eC148.
Blank RS, Thompson MM, Owens GK. Cell cycle versus density dependence of smooth muscle alpha actin expression in cultured rat aortic smooth muscle cells. J Cell Biol. 1988; 107: 299eC306.
Watanabe N, Kurabayashi M, Shimomura Y, Kawai-Kowase K, Hoshino Y, Manabe I, Watanabe M, Aikawa M, Kuro-o M, Suzuki T, Yazaki Y, Nagai R. BTEB2, a Kre筽pel-like transcription factor, regulates expression of the SMemb/Nonmuscle myosin heavy chain B (SMemb/NMHC-B) gene. Circ Res. 1999; 85: 182eC191.
Manabe I, Owens GK. CArG elements control smooth muscle subtype-specific expression of smooth muscle myosin in vivo. J Clin Invest. 2001; 107: 823eC834.
Schwartz RS, Huber KC, Murphy JG, Edwards WD, Camrud AR, Vlietstra RE, Holmes DR. Restenosis and the proportional neointimal response to coronary artery injury. J Am Coll Cardiol. 1992; 19: 267eC274.
Kollum M, Kaiser S, Kinscherf R, Metz J, Kubler W, Hehrlein C. Apoptosis after stent implantation compared with balloon angioplasty in rabbits. Arterioscler Thromb Vasc Biol. 1997; 17: 2383eC2388.
Christen T, Verin V, Bochaton-Piallat M-L, Popowski Y, Ramaekers F, Debruyne P, Camenzind E, van Eys G, Gabbiani G. Mechanisms of neointima formation and remodeling in the porcine coronary artery. Circulation. 2001; 103: 882eC888.
Raines EW. PDGF and cardiovascular disease. Cytokine Growth Factor Rev. 2004; 15: 237eC254.
Oishi Y, Manabe I, Tobe K, Tsushima K, Shindo T, Fujiu K, Nishimura G, Maemura K, Yamauchi T, Kubota N, Suzuki R, Kitamura T, Akira S, Kadowaki T, Nagai R. Kre筽pel-like transcription factor KLF5 is a key regulator of adipocyte differentiation. Cell Metab. 2005; 1: 27eC39.
Pendaries V, Verrecchia F, Michel S, Mauviel A. Retinoic acid receptors interfere with the TGF-b/Smad signaling pathway in a ligand-specific manner. Oncogene. 2003; 22: 8212eC8220.
Bastien J, Rochette-Egly C. Nuclear retinoid receptors and the transcription of retinoid-target genes. Gene. 2004; 328: 1eC16.
Miano JM, Berk BC. Retinoids: versatile biological response modifiers of vascular smooth muscle phenotype. Circ Res. 2000; 87: 355eC362.
Barstad RM, Hamers MJ, Stephens RW, Sakariassen KS. Retinoic acid reduces induction of monocyte tissue factor and tissue factor/factor VIIa-dependent arterial thrombus formation. Blood. 1995; 86: 212eC218.
Wiegman PJ, Barry WL, McPherson JA, McNamara CA, Gimple LW, Sanders JM, Bishop GG, Powers ER, Ragosta M, Owens GK, Sarembock IJ. all-trans-retinoic acid limits restenosis after balloon angioplasty in the focally atherosclerotic rabbit. Arterioscler Thromb Vasc Biol. 2000; 20: 89eC95.
Guagliumi G, Farb A, Musumeci G, Valsecchi O, Tespili M, Motta T, Virmani R. Sirolimus-eluting stent implanted in human coronary artery for 16 months. Circulation. 2003; 107: 1340eC1341.(Katsuhito Fujiu, Ichiro M)
Research Foundation Itsuu Laboratory (K.S.), Tokyo, Japan.
Abstract
Modulation of smooth muscle cell (SMC) phenotype plays a central role in neointima formation. We recently demonstrated that Am80, a synthetic retinoic acid receptor -specific agonist, inhibits the activity of the transcription factor KLF5, which is essential for neointima formation after vascular injury. In the present study, we aimed to further analyze the mechanism by which Am80 inhibits KLF5 and the effects of inhibiting KLF5 on SMCs and vascular lesion formation, as well as to evaluate potential of Am80 for use in the prevention of in-stent neointima formation. We found that Am80 inhibited both the expression and transcriptional function of KLF5. Of particular interest was our finding that KLF5 forms a transcriptionally active complex with unliganded RAR/RXR heterodimer on the PDGF-A promoter; Am80 disrupts this complex, thereby inhibiting KLF5-dependent transcriptional activation. Knocking down KLF5 using small interfering RNA suppressed serum-induced downregulation of SMC differentiation marker gene expression in cultured SMCs, and haploinsufficiency of KLF5 in mice attenuated phenotypic modulation of SMCs after vascular injury, indicating that KLF5 plays a key role in the control of SMC phenotype. Am80 augmented expression of the SMC differentiation marker genes in culture and within the vessel walls, and oral administration of Am80 significantly inhibited in-stent neointima formation in a rabbit stent-placement model. Taken together, these results demonstrate that KLF5 plays an important role in the control of SMC phenotype after vascular injury and suggest the feasibility of using Am80, delivered systemically and/or with a drug eluting stent, to prevent in-stent neointima formation.
Key Words: stent smooth muscle restenosis retinoid phenotypic modulation
Introduction
The neointima that forms after percutaneous coronary intervention (PCI) is mainly composed of smooth muscle cells (SMCs) and the extracellular matrix they produce.1 In response to external stresses such as mechanochemical stress, the effects of various humoral factors, and direct interaction with inflammatory cells, SMCs undergo phenotypic modulation. These phenotypically modulated SMCs proliferate, migrate, and produce various paracrine factors, extracellular matrix, and matrix proteases, thereby promoting the remodeling of the vascular wall.2,3 Consequently, SMCs are considered to be one of the most important targets of therapies aimed at preventing restenosis.
We recently showed that the Kre筽pel-like transcription factor KLF5 mediates stress-induced vascular remodeling, and that haploinsufficiency of KLF5 in KLF5+/eC mice much reduces neointima formation in vascular injury models.4 Moreover, we identified a novel synthetic retinoid, Am80, to be a potent inhibitor of KLF5.4 Am80 is a retinoic acid receptor (RAR) -specific agonist that has been safely used to treat acute promyelocytic leukemia. As expected from its in vitro inhibitory effects on KLF5, Am80 inhibited neointima formation in a mouse model in which the femoral artery was injured by cuff placement.4 It has not yet been directly determined, however, whether Am80 inhibits neointima formation after stenting. Such an analysis is important because the mechanisms underlying neointima formation likely differ between injury models5 and stenting is the most frequently performed PCI. It also remains unclear how inhibiting KLF5 with Am80 leads to suppression of neointima formation.
In the present study, we analyzed the mechanism by which Am80 inhibits KLF5 and the effects of inhibiting KLF5 on SMCs and vascular lesion formation, and then evaluated the potential of Am80 for use in the prevention of in-stent neointima formation. Our results demonstrate that KLF5 plays a key role in the control of SMC phenotype and that the KLF5 inhibitor Am80 is a novel modulator of SMC function. They also establish the feasibility of using Am80 to prevent in-stent restenosis.
Materials and Methods
Rabbit Stent Placement
Twenty-two male New Zealand White rabbits (Oriental Yeast Co, Ltd, Tokyo, Japan) weighing 3.5±0.3 kg were randomly assigned to treatment with Am80 or placebo. The Am80 was administrated orally at a dose of 1 mg · kgeC1 · deC1 beginning 1 day before stenting and continuing for 28 days thereafter. All animals received aspirin (40 mg/d) daily during the same period. Under general anesthesia, the iliac arteries were visualized using intravascular ultrasound imaging, after which they were injured by placement of a 2.5 mmx16 mm JOSTENT (JOMED) stent (6 to 8 atm, 30-second balloon inflation) under fluoroscopic guidance. Taking into consideration the vascular diameters, the inflation pressure of the stent balloons was determined so that the balloons were dilated to the standard stent-to-artery ratio (1.0:1&1.1:1).10 The final diameters of the lumens were then measured to ensure the appropriate stent-to-artery ratio and to rule out stent malapposition. All animal procedures strictly adhered to the guidelines for animal experiments of the University of Tokyo. An enhanced Materials and Methods section is available online at http://circres.ahajournals.org.
Results
KLF5 Plays a Role in Controlling SMC Phenotype
Modulation of the SMC phenotype is invariably observed in vascular lesions, and the phenotypically modulated SMCs are known to actively contribute to vascular disease development and progression. Our earlier observation that KLF5+/eC mice exhibit much reduced neointima formation after vascular injury4 strongly suggests that KLF5 is involved in controlling SMC phenotype. To test that idea, KLF5 was knocked-down in cultured SMCs using small interfering RNA (siRNA). SMCs were transfected with either a KLF5-siRNA expression vector or a negative control vector and then cultured in the serum-free, defined medium for 48 hours, after which the cells were stimulated with fetal bovine serum for 24 hour. As previously reported,6 serum stimulation led to downregulation of SM -actin and SM-MHC expression in cells transfected with the negative control vector, whereas expression of KLF5 and its target gene, PDGF-A, was upregulated (Figure 1A through 1D). Transfection of KLF5-siRNA vector resulted in reduced basal expression of KLF5 and PDGF-A and inhibition of their serum-induced upregulation (Figure 1A and 1B). In addition, knocking down KLF5 inhibited the serum-induced downregulation of SM -actin and SM-MHC (Figure 1C and 1D). KLF5 thus appears to be required for serum-induced modulation of SMC differentiation marker gene expression.
To further analyze the role of KLF5 in phenotypic modulation, the common carotid arteries of KLF5+/eC and wild-type mice were ligated, after which neointimal growth and SMC phenotype were analyzed. Histomorphometry of vessels 7 days after ligation showed that intima/media area ratios were significantly reduced in KLF5+/eC mice (supplemental Figure I). As previously reported,7 their was minimal expression of KLF5 in uninjured carotid arteries from wild-type animals. Basal expression of KLF5 in KLF5+/eC mice tended to be lower than in wild-type, but the difference was not statistically significant (Figure 1E). On day 7 after injury, however, KLF5 expression was markedly increased (12-fold over basal) within the arteries of wild-type mice and was also upregulated in KLF5+/eC animals, but to a significantly lesser degree (Figure 1E). Likewise, upregulation of PDGF-A was greatly reduced in KLF5+/eC animals (Figure 1F). In uninjured arteries, there were no significant differences in the expression of SM -actin and SM-MHC in wild-type and KLF5+/eC mice (Figure 1G and 1H). The downregulation of SM -actin and SM-MHC expression seen in the injured arteries of wild-type animals, however, was significantly attenuated in KLF5+/eC animals, suggesting that haploinsufficiency of KLF5 curtailed phenotypic modulation in response to the injury.
Am80 Inhibits KLF5 Expression and Transcriptional Activity in SMCs
We previously showed that KLF5 binds to RAR and that Am80 inhibits the transcriptional activity of KLF5.4 The molecular mechanism(s) by which Am80 inhibits KLF5 is not yet clear, however. To address that question, we first tested the possibility that Am80 inhibits KLF5 expression in SMCs. As shown in Figure 2A, Am80 suppressed KLF5 mRNA expression at concentrations as low as 0.1 eol/L. Likewise, all-trans retinoic acid (atRA) also inhibited KLF5, but was an order of magnitude less potent; 0.1 eol/L atRA did not have significant effect on KLF5 levels. In addition, we found that Am80 suppressed the KLF5 promoter activity normally induced by serum stimulation (Figure 2B). Those data suggest that Am80 suppresses KLF5 expression at least in part by inhibiting its transcription, although other post-transcriptional mechanism might also play a role.
We then examined the extent to which Am80 also inhibited the functionality of KLF5 in cultured SMCs overexpressing exogenous KLF5. The level of overexpressed Flag-tagged KLF5 protein was >10-fold higher (data not shown) than that of the endogenous protein, and Am80 had no effect on the levels of Flag-tagged KLF5 protein generated from the CAG-promoter driven construct. Moreover, our reporter analysis showed that, in the absence of Am80, this overexpression of KLF5 resulted in a 5.5-fold increase in PDGF-A promoter activity (Figure 2C). Despite having no effect on KLF5 expression levels under these conditions, Am80 significantly inhibited both KLF5-dependent activation of the PDGF-A promoter (Figure 2C). As expected from the inhibition of PDGF-A promoter, Am80 inhibited PDGF-A mRNA expression in SMCs (Figure 2D). atRA also inhibited PDGF-A expression, though to a lesser extent. Taken together, these findings indicate that Am80 inhibits KLF5-dependent PDGF-A transactivation by inhibiting both the expression and function of KLF5.
KLF5, RAR, and RXR Form a Transcriptional Complex on the PDGF-A Promoter That Is Disrupted by Am80
To further explore the mechanism by which Am80 inhibits KLF5-mediated gene transactivation, we assessed the function of RARs in PDGF-A transcription. Combinations of constructs expressing KLF5, RAR, and retinoic X receptor (RXR) were cotransfected into cells along with the PDGF-A promoter reporter. Whereas overexpression of KLF5 increased the reporter activity 5.5-fold over that seen with the reporter alone, overexpression of RAR and/or RXR had little effect on PDGF-A reporter activity (Figure 3A). Furthermore, coexpression of KLF5 and RAR led to a significantly greater activation of the reporter than was seen with KLF5 alone (9.5 vs 5.5-fold increase in activity), though coexpression of KLF5 with RXR did not. Am80 concentration-dependently inhibited the reporter activity induced by overexpression of KLF5, RAR, and RXR.
On the basis of the PDGF-A reporter analysis, we hypothesized that unliganded RAR serves as a coactivator in KLF5-dependent PDGF-A transcription and that Am80 disrupts the complex formed by KLF5 and RAR. To test that idea, Flag-tagged KLF5, RAR, and RXR were overexpressed in NIH3T3 cells incubated with or without Am80 or atRA (1 eol/L), and the cell lysates were subjected to coimmunoprecipitation. As shown in Figure 3B, KLF5 was detected in precipitates pulled down with either anti-RAR or anti-RXR antibody, whereas RAR and RXR were detected in precipitates pulled down with anti-Flag antibody (Figure 3C and 3D). RAR and RXR were also detected in precipitates pulled down with antibody against their respective counterparts, indicating that heterodimerization took place between RAR and RXR (Figure 3C and 3D). Given that KLF5 is known to physically interact with RAR,4 these data suggest that either KLF5 interacts with both RAR and RXR, or it directly interacts only with RAR within the RAR/RXR heterodimer. To resolve the nature of the interaction of KLF5 with RAR and RXR, we overexpressed Flag-KLF5 and RXR without RAR. Under those conditions, KLF5 was not detected in precipitates pulled down with anti-RXR antibody, nor was RXR detected in precipitates pulled down with anti-Flag antibody, indicating that KLF5 does not directly bind to RXR (Figure 3E). Treating cells with Am80 significantly reduced levels of KLF5 in precipitates pulled down with anti-RAR or anti-RXR antibody (Figure 3B). Likewise, Am80 reduced levels of coimmunoprecipitated RAR and RXR pulled down with Flag-KLF5 (Figure 3C and 3D). On the other hand, Am80 did not affect coimmunoprecipitation of RAR and RXR (Figure 3C and 3D). Apparently, Am80 inhibits the interaction between KLF5 and the RAR/RXR heterodimer. Interestingly, atRA at the same concentration (1 eol/L) did not effectively inhibit the interactions between KLF5 and RAR/RXR in these experiments.
We next performed chromatin immunoprecipitation (ChIP) assays to further confirm the involvement of the RAR/RXR heterodimer in the activation of the PDGF-A promoter by KLF5. Using real-time polymerase chain reaction (PCR), we analyzed the abundance of the sequences of interest in immunoprecipitated chromatin samples (Figure 3F and 3G). SMCs were incubated with or without Am80 for 2 hours and then fixed with formaldehyde, after which the cross-linked chromatin samples were subjected to immunoprecipitation with anti-KLF5, anti-RAR, or anti-RXR antibody.8 The results showed that KLF5, RAR, and RXR all bound to the endogenous PDGF-A promoter in SMCs and that Am80 markedly reduced the abundance of the PDGF-A promoter sequence in the immunoprecipitates pulled down with anti-KLF5, anti-RAR, or anti-RXR antibody. Taken together, the results suggest that KLF5 and unliganded RAR/RXR bound to and activated the promoter as components of a transcriptional complex and that Am80 disrupted the interaction between KLF5 and RAR/RXR, thereby inhibiting their binding to the promoter and thus PDGF-A transcription.
Am80 Affects the Differentiation State of SMCs
That Am80 inhibits KLF5 activity suggest it might also affect SMC phenotype. As expected, adding Am80 to cultured SMCs upregulated expression of the SMC differentiation marker genes SM-MHC and SM -actin (Figure 4A and 4B), suggesting that Am80 promotes differentiation in phenotypically modulated cultured SMCs. Moreover, Western blot analyses confirmed that Am80 induced expression of SM-HMC protein at concentrations as low as 0.01 eol/L (Figure 4C). Although atRA also induced expression of SM -actin and SM-MHC at a higher concentration than Am80, the extent of expression were less than was induced by Am80.
Am80 Inhibits Proliferation and Migration of SMCs
That phenotypic modulation is a prerequisite for proliferation and migration of differentiated SMCs3 suggested to us that inhibition of phenotypic modulation by Am80 might suppress the pathophysiological proliferation and migration of SMCs seen in vascular diseases, such as in-stent neointima formation. Consistent with that idea, we found that Am80 dose-dependently inhibited both mitogen-stimulated DNA synthesis and cell proliferation in cultured human coronary arterial SMCs and rat aortic SMCs (Figure 5A through 5D). SMC proliferation was significantly inhibited at a concentration of 0.1 eol/L, and the same concentration also significantly inhibited fetal bovine serum-induced SMC migration (supplemental Figure II). We also examined the effect of Am80 on proliferation of endothelial cells because re-endothelialization is important for inhibition of thrombosis and for stabilization of the remodeling process, but found that Am80 has no effect on DNA synthesis or proliferation in cultured human umbilical vein endothelial cells (HUVECs) (Figure 5E and 5F), suggesting that Am80 selectively inhibits SMC proliferation within the vascular wall.
Am80 Reduces In-Stent Neointima Formation
We then tested whether orally administered Am80 would reduce in-stent neointima formation after placing metal coronary stents in the iliac arteries of New Zealand White rabbits. We initially selected the appropriate dosage of Am80 by administering a series dosages to rabbits and determining the relationship between the oral intake and plasma concentration (supplemental Table IA). A dose of 1 mg · kgeC1 · deC1 was selected so that the plasma Am80 concentration of the rabbits that underwent stent placement reached between 0.1 to 1 eol/L (supplemental Table IB). To compare the injury caused by stent placement in rabbits receiving Am80 or placebo, we used the injury score originally described for pigs.9 Using that scale, we found that the injuries caused by stent placement were similar in both the Am80 and placebo groups (0.15±0.36 versus 0.15±0.36); a few rabbits in each group showed laceration of the internal elastic lamina, but no injuries with higher scores were detected. When the stented arteries were harvested 28 days later and subjected to histological analysis, however, we found that the Am80 group showed significantly less neointima formation than the placebo group (Figure 6A through 6F).
Histomorphometry of vessels from 7 animals in each group showed that although the medial areas were similar, the intimal area, intima/media area ratios, and percent stenosis were significantly (P<0.05) reduced in the Am80 group, which is indicative of the diminished neointimal growth (Figure 6G through 6J). It is also noteworthy that during this experiment, no rabbits showed any signs of adverse effects of Am80 on body weight, appetite, blood chemistry, or major organ histology.
When we evaluated endothelialization of in-stent lesions, we found that despite the marked inhibition of neointimal growth, high magnification light microscopy revealed that almost the entire circumference of the lumens of the injured arteries on day 28 was completely covered by ECs in both groups (data not shown).
Am80 Inhibits Phenotypic Modulation of SMCs In Vivo
To investigate the mechanism by which Am80 reduced in-stent neointima formation, we first analyzed the effect of Am80 on SMC growth and apoptosis within the vascular walls. Because SMC proliferation is more pronounced during the first week after stent placement than at later times and shows more modulated phenotypes,10 the stent-related lesions were analyzed on day 7 after stenting. We found that in control animals, fibrin and cellular deposits had already formed adjacent to the stent struts by this time, as has been reported previously,10 but such lesions were barely detectable in Am80-treated animals, and the intimal area was significantly reduced (supplemental Figure III). Thus, Am80 clearly inhibited lesion formation during the first week after stent placement. We also analyzed cell proliferation by staining the tissue with proliferating cell nuclear antigen (PCNA) (Figure 7). Within the intimal lesions of the control animals, 19% of cells were PCNA-positive. Within the media, PCNA-positive SMCs were often found in the areas beneath the intimal lesions, and 7.2% of cells showed PCNA positivity in the all medial areas. By contrast, only 0.7% of cells were PCNA-positive in the media of the Am80-treated animals, indicating that Am80 markedly inhibited proliferation of medial SMCs (Figure 7E).
The early inhibition of intimal lesion formation suggests Am80 might also interfere with other processes involved in early lesion formation, such as platelet aggregation, thrombus formation, and monocyte infiltration.1 On the other hand, Am80 apparently had no effect on apoptotic processes at this time point, as very few terminal deoxynucleotidyl transferase biotin-dUTP nick end labeling (TUNEL)-positive apoptotic cells were found in sections obtained from either group, and the number of TUNEL-positive cells was unaffected by Am80 (supplemental Figure IV).
Consistent with earlier findings,11 we also found that on day 7, expression of SM -actin protein was diminished within the stented vascular walls of control animals (supplemental Figure V). In clear contrast, many more SMCs were positive for SM -actin in Am80-treated animals. At the same time, the level of SM -actin mRNA expression was much lower in the stented vascular walls of control animals than Am80-treated animals (Figure 8A). In similar fashion, Am80 also inhibited downregulation of SM-MHC gene expression (Figure 8B and 8C). Expression of total SM-MHC transcripts, which included both the SM1 and SM2 isoforms, was significantly stronger in the Am80-treated group on day 7, as was expression of the SM2 isoform, a marker for mature SMCs. On the other hand, expression of SMemb, a marker for phenotypically modulated SMCs that was induced after stenting in control animals, was markedly reduced in Am80-treated animals (Figure 8D).
By day 28, Am80 treatment led to further increases in expression of SM-MHC and SM -actin (Figure 8A through 8C), suggesting acceleration of SMC differentiation. Relative levels of SM1 protein in the neointima as compared with those in the media were higher in the Am80 group than in the placebo group on day 28, as were levels of SM -actin protein, suggesting that Am80 treatment also induced the SMC differentiation marker genes at the protein level (supplemental Figure VE through VH). These patterns of expression indicate that Am80 inhibits phenotypic modulation of SMCs and promotes SMC differentiation after stenting in vivo. As expected, expression of KLF5 and its target gene PDGF-A was markedly inhibited in Am80-treated animals (Figure 8E and 8F), suggesting that in vivo, Am80 suppresses phenotypic modulation at least in part by inhibiting KLF5, just as it does in vitro. In contrast, expression of myocardin, a coactivator of serum response factor, was unaffected by Am80 treatment.
Discussion
In the present study, we have shown that KLF5 is important for control of SMC phenotype in vitro and after vascular injury in vivo. We also showed that the synthetic retinoid Am80 inhibits KLF5 by interfering with both its expression and its interaction with RAR. In this way, Am80 affected the phenotype, proliferation, and migration of cultured SMCs and inhibited neointima formation after stent placement in vivo. These results establish KLF5 as a novel molecular target for therapeutic intervention in the treatment of vascular disease and provide evidence of the feasibility of using its antagonist Am80 to prevent neointima formation and in-stent restenosis.
The phenotypic modulation of SMCs requires the coordinate regulation of a number of genes, which means that SMC phenotype is very likely governed by the activity of a transcription factor network.2,3 Our previous studies suggest that KLF5 is a component of just such a transcriptional network,4,7 making it a potentially useful target for therapeutic intervention in vascular disease. To assess the role of KLF5 in the control of the SMC phenotype, we analyzed the effects of knocking down KLF5. In cultured SMCs, KLF5 knockdown attenuated downregulation of SM-MHC and SM -actin by serum (Figure 1), which mimics phenotypic modulation in terms of SMC marker gene expression.6 Similarly, SMCs in the injured carotid arteries of KLF5+/eC mice exhibited a more differentiated phenotype than those in wild-type animals, judging from the expression of SMC differentiation marker genes (Figure 1).
So how does KLF5 control SMC phenotype One possibility is that KLF5 directly controls expression of SMC marker genes; it might positively regulate SMemb on one hand and negatively regulate SM -actin and SM-MHC on the other. Another possibility is that KLF5 controls expression of paracrine factors, which in turn controls the phenotype and differentiation state of SMCs.12 Indeed, we previously identified PDGF-A as one of the downstream targets of KLF54 and, as expected, PDGF-A expression was diminished in KLF5-siRNA transfected SMCs, in the vascular lesions in KLF5+/eC mice (Figure 1F), and in the stented vascular walls of rabbits given Am80 (Figure 8F). It is also likely that KLF5 controls other unidentified genes,13 which may cumulatively affect SMC phenotype.
The results of the present study also provide novel insight into the actions of RARs in SMCs. Reporter (Figure 3A) and ChIP (Figure 3F and 3G) analyses showed that RAR/RXR is involved in the transcriptional control of PDGF-A. Because the proximal promoter region of PDGF-A does not contain a canonical retinoic acid response element (RARE), it is likely that RAR/RXR binds to the promoter via interaction with KLF5. Consistent with that idea is our finding that KLF5 directly interacts with RAR (Figure 3); of particular interest to us is the finding that Am80 inhibited that interaction as well as KLF5-dependent transactivation of the PDGF-A promoter. This suggests that unliganded RAR/RXR heterodimer and KLF5 form a transcriptionally active complex on the PDGF-A promoter, and that by disrupting formation of that complex, Am80 inhibits the activity of KLF5. Penderies et al14 similarly reported that unliganded RAR enhances Smad-dependent transcription and that this effect was inhibited by a RAR agonist, which disrupted the interaction between RAR and Smads. Together with those data, our findings suggest a novel mode of RAR action in which unliganded RARs serve as coactivators for other transcription factors. In this mode, RAR agonists may inhibit coactivation by RARs, which contrasts with the classical RARE-dependent transactivation induced by RAR agonists. Notably, earlier studies have also revealed yet another RARE-independent mode of RAR action in which RARs transrepress other transcription factors, such as activator protein-1.15 Given these novel actions, one needs to consider a broader range of genes as potential targets of RARs. Interestingly, a number of possible retinoid-regulated genes have been identified in SMCs, many of which do not have canonical RARE in their transcriptional regulatory regions.16 It is very likely that some of these genes are regulated by RARs functioning as coactivators and/or corepressors. Moreover, results of the present study suggest that KLF5 is one of the key mediators of retinoid actions in SMCs.
We also examined the effects of Am80 on SMC proliferation, migration, and apoptosis as they relate to neointima formation and in-stent restenosis.1,5 Am80 clearly inhibited proliferation of medial SMCs 7 days after stent placement (Figure 7). Although higher concentrations of Am80 induced apoptosis among cultured SMCs, there was no apparent increase in the numbers of apoptotic cells within the vascular walls 7 days after stent placement (supplemental Figure IV). Further study will be required to fully assess the importance of apoptosis in the in vivo effects of Am80, but the fact that higher concentrations of Am80 were required to induce apoptosis than to inhibit proliferation among cultured SMCs suggests that its antiproliferative action is central to Am80’s overall inhibition of neointima formation. In addition, Am80’s inhibition of neointima formation on day 7 after stent placement suggests that it affects nonmuscle cells as well as SMCs. This is because the early lesion is known to be formed by various cell types, including macrophages, lymphocytes, and platelets. Interestingly, in that regard, atRA has been shown to promote fibrinolysis and to inhibit thrombosis, platelet aggregation, and inflammation.17
Previous studies have shown that atRA reduces neointima formation in balloon injury models,16,18 making it an attractive candidate for local delivery by drug-eluting stent. Unfortunately, however, its instability with respect to heat, light, oxidation, and sterilization markedly diminishes its utility. The higher stability and selectivity of Am80 would be expected to enhance its utility and broaden its applicability to other ailments in which abnormal cellular differentiation and chronic inflammation play important roles. The dramatic results recently obtained with drug-eluting stents using sirolimus and paclitaxel have stimulated significant enthusiasm among interventional cardiologists. There are still potential problems, however. For example, the strong antiproliferative effects of these agents may delay the healing of in-stent lesions, causing thrombosis.19 The unique biological activities of Am80 might provide a means to overcome these problems by inhibiting tissue remodeling and stabilizing the vascular wall.
Acknowledgments
We gratefully acknowledge Eriko Magoshi, Noriko Yamanaka, and Michiko Hayashi for their excellent technical assistance and Kazuya Shibaki, Koji Osaka, Hideo Fumiyama, and Sachie Hayashi for their great support. This study was supported in part by a Grant-in-Aid from the Ministry of Education, Culture, Sports, Science, and Technology, Japan, a research grant from National Institute of Biomedical Innovation, Japan (to R.N.), and research grants from Takeda Science Foundation, Sankyo Foundation of Life Science, Mitsubishi Pharma Research Foundation, and Tokyo Biochemical Research Foundation (to I.M.).
References
Schwartz RS, Henry TD. Pathophysiology of coronary artery restenosis. Rev Cardiovasc Med. 2002; 3: S4eCS9.
Manabe I, Nagai R. Regulation of smooth muscle phenotype. Curr Atheroscler Rep. 2003; 5: 214eC222.
Owens GK, Kumar MS, Wamhoff BR. Molecular regulation of vascular smooth muscle cell differentiation in development and disease. Physiol Rev. 2004; 84: 767eC801.
Shindo T, Manabe I, Fukushima Y, Tobe K, Aizawa K, Miyamoto S, Kawai-Kowase K, Moriyama N, Imai Y, Kawakami H, Nishimatsu H, Ishikawa T, Suzuki T, Morita H, Maemura K, Sata M, Hirata Y, Komukai M, Kagechika H, Kadowaki T, Kurabayashi M, Nagai R. Kre筽pel-like zinc-finger transcription factor KLF5/BTEB2 is a target for angiotensin II signaling and an essential regulator of cardiovascular remodeling. Nat Med. 2002; 8: 856eC863.
Indolfi C, Mongiardo A, Curcio A, Torella D. Molecular mechanisms of in-stent restenosis and approach to therapy with eluting stents. Trends Cardiovasc Med. 2003; 13: 142eC148.
Blank RS, Thompson MM, Owens GK. Cell cycle versus density dependence of smooth muscle alpha actin expression in cultured rat aortic smooth muscle cells. J Cell Biol. 1988; 107: 299eC306.
Watanabe N, Kurabayashi M, Shimomura Y, Kawai-Kowase K, Hoshino Y, Manabe I, Watanabe M, Aikawa M, Kuro-o M, Suzuki T, Yazaki Y, Nagai R. BTEB2, a Kre筽pel-like transcription factor, regulates expression of the SMemb/Nonmuscle myosin heavy chain B (SMemb/NMHC-B) gene. Circ Res. 1999; 85: 182eC191.
Manabe I, Owens GK. CArG elements control smooth muscle subtype-specific expression of smooth muscle myosin in vivo. J Clin Invest. 2001; 107: 823eC834.
Schwartz RS, Huber KC, Murphy JG, Edwards WD, Camrud AR, Vlietstra RE, Holmes DR. Restenosis and the proportional neointimal response to coronary artery injury. J Am Coll Cardiol. 1992; 19: 267eC274.
Kollum M, Kaiser S, Kinscherf R, Metz J, Kubler W, Hehrlein C. Apoptosis after stent implantation compared with balloon angioplasty in rabbits. Arterioscler Thromb Vasc Biol. 1997; 17: 2383eC2388.
Christen T, Verin V, Bochaton-Piallat M-L, Popowski Y, Ramaekers F, Debruyne P, Camenzind E, van Eys G, Gabbiani G. Mechanisms of neointima formation and remodeling in the porcine coronary artery. Circulation. 2001; 103: 882eC888.
Raines EW. PDGF and cardiovascular disease. Cytokine Growth Factor Rev. 2004; 15: 237eC254.
Oishi Y, Manabe I, Tobe K, Tsushima K, Shindo T, Fujiu K, Nishimura G, Maemura K, Yamauchi T, Kubota N, Suzuki R, Kitamura T, Akira S, Kadowaki T, Nagai R. Kre筽pel-like transcription factor KLF5 is a key regulator of adipocyte differentiation. Cell Metab. 2005; 1: 27eC39.
Pendaries V, Verrecchia F, Michel S, Mauviel A. Retinoic acid receptors interfere with the TGF-b/Smad signaling pathway in a ligand-specific manner. Oncogene. 2003; 22: 8212eC8220.
Bastien J, Rochette-Egly C. Nuclear retinoid receptors and the transcription of retinoid-target genes. Gene. 2004; 328: 1eC16.
Miano JM, Berk BC. Retinoids: versatile biological response modifiers of vascular smooth muscle phenotype. Circ Res. 2000; 87: 355eC362.
Barstad RM, Hamers MJ, Stephens RW, Sakariassen KS. Retinoic acid reduces induction of monocyte tissue factor and tissue factor/factor VIIa-dependent arterial thrombus formation. Blood. 1995; 86: 212eC218.
Wiegman PJ, Barry WL, McPherson JA, McNamara CA, Gimple LW, Sanders JM, Bishop GG, Powers ER, Ragosta M, Owens GK, Sarembock IJ. all-trans-retinoic acid limits restenosis after balloon angioplasty in the focally atherosclerotic rabbit. Arterioscler Thromb Vasc Biol. 2000; 20: 89eC95.
Guagliumi G, Farb A, Musumeci G, Valsecchi O, Tespili M, Motta T, Virmani R. Sirolimus-eluting stent implanted in human coronary artery for 16 months. Circulation. 2003; 107: 1340eC1341.(Katsuhito Fujiu, Ichiro M)