Conditional Deletion of Krüppel-Like Factor 4 Delays Downregulation of Smooth Muscle Cell Differentiation Markers but Accelerates Neointimal Formation Following Vascular Injury

Phenotypic switching of smooth muscle cells (SMCs) plays a key role in the development of vascular diseases such as atherosclerosis and restenosis after percutaneous coronary interventions.¹ Differentiated SMCs in adult vessels express a unique repertoire of contractile proteins (ie, SMC differentiation markers) and exhibit an extremely low rate of proliferation. However, in association with vascular injury or disease, SMCs modulate their phenotype and participate in the formation of neointima by markedly decreasing expression of SMC differentiation markers and increasing proliferation, migration, and synthesis of extracellular matrix proteins. Therefore, elucidation of the molecular mechanisms controlling SMC phenotypic switching is likely to provide important insights toward understanding of the development of vascular disease.

A hallmark of SMC phenotypic switching is the coordinate downregulation of expression of SMC differentiation markers including smooth muscle (SM) α-actin, SM22α, and SM–myosin heavy chain (SM-MHC).¹ The promoter/enhancer regions of these SMC differentiation makers contain common cis elements, including multiple CC(A/T-rich)₆GG (CArG) elements and a transforming growth factor (TGF)-β control element.¹ Previously, we identified Krüppel-like factor 4 (Klf4) (formerly known as gut-enriched Krüppel-like factor or GKLF) as a binding factor for the TGF-β control element located within the SM22α promoter using a yeast 1-hybrid screen.² Of interest, Klf4 behaved as a potent repressor of SMC differentiation markers including SM α-actin, SM22α, and SM-MHC in cultured SMCs through multiple mechanisms including: (1) direct Klf4 binding to the TGF-β control element within the promoter/enhancer regions of SMC differentiation markers; (2) interaction with serum response factor (SRF), a trans-binding factor of CArG elements; and (3) repression of expression of an SRF coactivator, myocardin.^3,4 Results from our previous studies also showed that Klf4 contributed to phenotypic switching of SMCs in vitro in response to platelet-derived growth factor (PDGF)-BB and oxidized phospholipids.^4,5 Indeed, PDGF-BB and oxidized phospholipids activated Klf4 expression, and small interfering (si)RNA-induced knockdown of Klf4 attenuated PDGF-BB– and oxidized phospholipid–induced downregulation of SM α-actin and/or SM-MHC expression in cultured SMCs. Moreover, expression of Klf4 mRNA was transiently induced in rat carotid arteries after balloon injury.⁴ The preceding results provide evidence that Klf4 is a potent transcriptional repressor of SMC differentiation markers in cultured SMCs and suggest a potential role for Klf4 in control of SMC phenotypic switching in vivo. However, as yet, no studies have directly assessed the function of Klf4 following vascular injury in vivo.

Klf4 regulates cellular proliferation in multiple cell types. Overexpression of Klf4 inhibited DNA synthesis and reduced cellular growth in colon cancer cells and fibroblasts.^6–8 Klf4-induced growth suppression was caused by cell cycle arrest at the G₁/S boundary, and there is evidence showing that this is the result of transcriptional activation of the cyclin-dependent kinase inhibitor p21^WAF1/Cip1.^9,10 In response to DNA damage, p53 and Klf4 were induced, and these proteins synergistically activated p21^WAF1/Cip1 gene transcription.⁹ Of interest, induction of Klf4 following DNA damage was abolished in p53-null cells.⁹ Moreover, siRNA-induced knockdown of Klf4 attenuated DNA damage–induced expression of p21^WAF1/Cip1, as well as the accumulation of cells at the G₁ phase.¹¹ These results indicate that Klf4 is essential in mediating p53-induced G₁/S cell cycle arrest as a consequence of DNA damage. Consistent with these findings, Klf4 expression is significantly decreased in multiple human cancers including colon cancer and gastric cancer.^8,12 In addition, loss of Klf4 in mouse gastric epithelia resulted in increased proliferation and altered differentiation of the gastric epithelia.¹² Results of recent studies also demonstrated that Klf4 induces p21^WAF1/Cip1 expression and reduces cellular growth in cultured SMCs.¹³ Taken together, these studies provide compelling evidence that Klf4 inhibits cellular proliferation in multiple cell types. However, elevated Klf4 levels have been reported in mammary carcinoma and squamous cell carcinoma of the oropharynx.^14,15 Indeed, Rowland et al¹⁶ found that Klf4 had a potential to switch from a growth-inhibiting tumor suppressor to a growth-promoting oncogene in response to changes in the cellular context. As such, the functions of Klf4 in vivo are poorly defined.

Klf4-null mice are born at the expected Mendelian ratio but die shortly after birth because of loss of skin barrier function,¹⁷ precluding the analysis of Klf4 function in adult animals. To overcome this limitation, we have derived conditional Klf4-deficient mice by breeding tamoxifen-inducible Cre recombinase–expressing (ERT-Cre) mice¹⁸ with mice carrying a loxP allele of Klf4 (Klf4^loxP mice)¹⁹ and analyzed their phenotype in response to vascular injury. We show that Klf4 deficiency results in a delay in suppression of SMC differentiation markers and enhanced neointimal formation following vascular injury. We analyze the molecular mechanisms of Klf4 action and demonstrate that Klf4 induces p21^WAF/Cip1 expression through direct binding of Klf4 and enhanced binding of p53 to the promoter/enhancer regions of the p21^WAF/Cip1 gene in cultured SMCs, as well as in injured carotid arteries in vivo.

Materials and Methods

Animal protocols were approved by the University of Virginia Animal Care and Use Committee. ERT-Cre mice¹⁸ and Klf4^loxP mice¹⁹ were bred to generate tamoxifen-inducible Klf4-deficient mice.

An expanded Materials and Methods section is available in the online data supplement at http://circres.ahajournals.org.

Results

Klf4 Expression Was Induced in the Medial Layer of Vessels After Vascular Injury

We derived tamoxifen-inducible Klf4-deficient mice by breeding ERT-Cre mice with Klf4^loxP mice. Seven- to 8-week-old ERT-Cre⁺/Klf4^loxP/loxP mice were injected with 4-hydroxytamoxifen (4-OHT) 10 times IP over a 14-day period to activate Cre recombinase (Figure 1A), because 1 or 3 times of 4-OHT injection did not elicit sufficient recombination of the Klf4^loxP allele (data not shown). ERT-Cre⁻/Klf4^loxP/loxP mice were injected with 4-OHT and used as controls. After 4-OHT injection, a highly efficient recombination of the Klf4^loxP allele was observed in the aorta of ERT-Cre⁺/Klf4^loxP/loxP (Klf4 knockout [KO]) mice, whereas no recombination was seen in ERT-Cre⁻/Klf4^loxP/loxP (control) mice (Figure 1B). The Klf4 gene consists of 4 exons, and exon 1 only encodes a first methionine codon. Recombination of the Klf4^loxP allele deletes exons 2 and 3 and causes a frameshift mutation in exon 4, which abolishes Klf4 function completely.¹⁹ In support of this, expression of Klf4 protein and Klf4 mRNA was markedly reduced in the lung and the colon of Klf4 KO mice, respectively (Figure 1C and 1D). However, there were no differences in visible appearance (data not shown) and body weight (23.0±1.3 g in control mice versus 22.2±1.2 g in Klf4 KO mice, n=10, P=0.65) between Klf4 KO and control mice after 4-OHT injection, which is in contrast to defects in skin barrier function observed in conventional Klf4-null mice, which die shortly after birth.¹⁷ We determined changes in Klf4 expression patterns in response to carotid ligation injury in wild-type mice. Klf4 mRNA expression was rapidly induced at day 1 after injury, whereas SM α-actin mRNA expression was decreased at day 3 and day 7 (Figure 1E). Injury-induced changes in Klf4 protein levels were also examined by immunohistochemistry. In agreement with previous studies,^2,20 Klf4 was expressed in endothelial cells but not in SMCs of uninjured vessels in control mice (Figure 1F). Three days after injury, Klf4 expression was induced in the medial SMC layer of carotid arteries of control mice. In contrast, Klf4 induction was undetectable in Klf4 KO mice. These results clearly indicate that Klf4 is induced in the SMC layer of carotid arteries of control mice following vascular injury and that tamoxifen-inducible Klf4-deficient mice are suitable model to test the effects of Klf4 deletion on the vascular phenotype after injury.

• Download figure
• Download PowerPoint

Klf4 Ablation Caused a Delay in Repression of SMC Differentiation Markers After Vascular Injury

We examined whether deletion of the Klf4 gene affects downregulation of SMC differentiation markers following vascular injury. As shown in Figure 2A and Figure I in the online data supplement, expression of both SM α-actin and SM22α was markedly decreased in the medial layer of carotid arteries in control mice at day 3 and day 7 after ligation injury. In Klf4 KO mice, expression of these SMC differentiation markers was decreased at day 7 after injury. However, repression of both SM α-actin and SM22α was attenuated in carotid arteries of Klf4 KO mice as compared with control mice at day 3 following injury. The results indicate that ablation of the Klf4 gene causes a transient delay in repression of SMC differentiation markers in response to vascular injury.

• Download figure
• Download PowerPoint

Binding of Klf4 to the promoter regions of SMC differentiation marker genes was examined in injured carotid arteries of wild-type mice. Of major significance, in vivo chromatin immunoprecipitation (ChIP) assays showed that Klf4 bound to the TGF-β control element-containing promoter regions of the SM α-actin gene and the SM22α gene in carotid arteries at day 3 after injury, whereas no Klf4 binding was detected in uninjured arteries (Figure 2B). Consistent with the kinetics of Klf4 expression in response to carotid ligation injury (Figure 1E and 1F), Klf4 binding was not seen in carotid arteries at day 7 after injury. Association of Klf4 with the intronic 1 region of the SM22α gene, which contains no Klf4 binding sites, was undetectable in both injured and uninjured carotid arteries. These results provide evidence that Klf4 is rapidly induced in SMCs of carotid arteries following vascular injury and that it represses expression of SMC differentiation markers by binding to the promoter regions of these genes at early time points following injury.

Conditional Klf4 Mutant Mice Exhibited Enhanced Neointimal Formation Following Vascular Injury

Next, we performed morphometric analyses in carotid arteries of Klf4 KO and control mice. Of major interest, Klf4 KO mice exhibited enhanced formation of neointima as compared with control mice (Figure 3). Neointimal area was significantly larger in Klf4 KO mice than control mice by 2.8-fold at day 7, 3.0-fold at day 14, and 3.1-fold at day 21 after injury (Figure 3C). Medial areas of injured vessels were increased as compared with uninjured vessels in both Klf4 KO and control mice but not significantly different from one another although Klf4 KO mice exhibited a modest increase (Figure 3B). Medial areas of uninjured carotid arteries were not different between Klf4 KO mice and control mice at any time point. These results indicate that, in addition to regulating SMC differentiation marker expression following vascular injury, Klf4 also plays a key role in inhibiting neointimal formation.

• Download figure
• Download PowerPoint

Klf4 Reduced the Proliferation Rate of SMCs Without Affecting Apoptosis

To determine the mechanisms of enhanced neointimal formation in Klf4 KO mice, proliferation and apoptotic rates in the media and the intima were examined by 5-bromodeoxyuridine (BrdUrd) and TUNEL staining, respectively. BrdUrd staining revealed that medial cells in injured carotid arteries of Klf4 KO mice exhibited enhanced proliferation as compared with control mice (15.3±2.5% versus 9.7±1.9% at day 7, 13.9±0.8% versus 5.4±2.1% at day 14, 4.7±1.0% versus 1.6±0.4% at day 21), although the fraction of BrdUrd-positive cells in the intima was not different between Klf4 KO and control mice (Figure 4A and 4C). The apoptotic rate in the media and the intima was not different between Klf4 KO mice and control mice, although it was increased in injured vessels as compared with uninjured vessels in both groups (Figure 4B and 4D). We also tested whether Klf4 ablation alters the accumulation of macrophages and T lymphocytes in carotid arteries after vascular injury by Mac2 and CD3-ε immunostaining, respectively (supplemental Figure II). Although previous studies showed the induction of Klf4 expression in cultured macrophages in response to inflammatory cytokines,²¹ accumulation of macrophages as well as T lymphocytes was not different between Klf4 KO and control mice. These results suggest that enhanced neointimal formation in Klf4 KO mice following vascular injury is caused by increased proliferation of medial cells, rather than an altered apoptotic rate. This is likely to be caused, at least in part, by enhanced SMC proliferation, because SMCs are a major cell type in the media and the recruitment of monocytes/macrophages and T lymphocytes to the site of vascular injury was unaltered in Klf4 KO mice.

• Download figure
• Download PowerPoint

To determine the role of Klf4 in regulating SMC growth, SMCs were isolated from thoracic aorta of Klf4^loxP/loxP mice and infected with adenovirus-expressing Cre recombinase or empty adenovirus to generate Klf4-null SMCs and control SMCs, respectively. Recombination of the Klf4^loxP allele in cultured Klf4-null SMCs was confirmed by PCR (Figure 5A). Cellular growth was assessed by cell counting over a time course of 96 hours. We observed no differences in cell number between Klf4-null and control SMCs (Figure 5B and 5C), indicating that basal expression of Klf4 is low as seen in medial SMCs in vivo (Figure 1F) and does not contribute to SMC growth under these culture conditions. However, adenovirus-mediated overexpression of Klf4 markedly reduced cell number in both Klf4-null and control SMCs, demonstrating its potent growth repressing properties for SMCs. Consistent with these results, BrdUrd-incorporation assays showed that adenovirus-mediated expression of Klf4 reduced the proliferation rate of cultured SMCs (Figure 5D). Overexpression of Klf4 did not elicit any morphological changes indicative of apoptosis in nuclei of SMCs (Figure 5E) and did not induce expression of apoptosis markers including cleaved caspase 3, Bax, and Puma (Figure 5F). Taken together, results provide evidence that induction of Klf4 expression suppresses cellular proliferation in SMCs.

• Download figure
• Download PowerPoint

Klf4 Induced p21^WAF1/Cip1 Expression in Concert With p53

Results of previous studies in multiple cell culture systems have shown that Klf4 is a mediator of p53-induced activation of p21^WAF1/Cip1 gene transcription in response to DNA damage and that p21^WAF1/Cip1 is a key factor leading to cell cycle arrest at the G₁/S transition.^9–11 We tested whether p21^WAF1/Cip1 contributes to Klf4-induced suppression of SMC proliferation. First, injury-induced changes in expression of p21^WAF1/Cip1 were examined in the medial SMC layer of carotid arteries in Klf4 KO and control mice. Expression of p21^WAF1/Cip1 was increased at day 3 and day 7 after injury in control mice but not in Klf4 KO mice (Figure 6A). Second, results in cultured mouse SMCs showed that adenovirus-mediated overexpression of Klf4 induced p21^WAF1/Cip1 expression, whereas p53 expression was unaltered (Figure 6B). Third, adenovirus-mediated overexpression of Klf4 did not reduce cellular proliferation in p21^WAF1/Cip1 KO mouse SMCs (Figure 6C). These results suggest that Klf4-induced suppression of SMC proliferation is mediated in part by activation of p21^WAF1/Cip1.

• Download figure
• Download PowerPoint

Previous studies showed that Klf4 activates p21^WAF1/Cip1 gene transcription through the proximal Klf4 binding site in HEK293 cells and that p53 induced p21^WAF1/Cip1 gene transcription via the p53 binding site and the proximal Klf4 binding site.^9,22 The consensus Klf4 binding site is 5′-(G/A)(G/A)GG(C/T)G(C/T)-3′, whereas the consensus element for p53 is 2 copies of a 10-bp motif, 5′-(G/A)(G/A)(G/A)C(A/T)(A/T)G(C/T)(C/T)(C/T)-3′, separated by multiple nucleotides. Inspection of the mouse p21^WAF1/Cip1 promoter/enhancer sequence revealed a novel Klf4 consensus binding site at −2611/−2605 bp, as well as the proximal Klf4 binding site (−176/−170 bp) and the p53 binding site (−2869/−2838 bp). We tested the involvement of these cis elements for Klf4-induced activation of p21^WAF1/Cip1 gene transcription in cultured aortic SMCs. Klf4 induced transcriptional activity of the p21^WAF1/Cip1 promoter/enhancer–luciferase construct by 2.2-fold, but the induction was reduced by mutation of either 1 of these cis elements (Figure 7A). Simultaneous mutation of all 3 of these cis elements completely abolished the response. Moreover, results of siRNA-induced knockdown experiments showed that suppression of p53 blocked Klf4-induced p21^WAF1/Cip1 activation in rat aortic SMCs (Figure 7B and supplemental Figure III). These results suggest that, in addition to 2 Klf4 binding sites, p53 and its binding site are required for Klf4-induced activation of the p21^WAF1/Cip1 gene.

• Download figure
• Download PowerPoint

ChIP assays were carried out to determine whether Klf4 alters the association of Klf4 and p53 with the p21^WAF1/Cip1 promoter/enhancer in SMCs. Adenovirus-mediated expression of Klf4 induced the binding of Klf4 to the promoter/enhancer regions of the p21^WAF1/Cip1 gene that contain the proximal as well as the distal Klf4 binding sites in cultured mouse SMCs (Figure 7C). Klf4 overexpression did not induce Klf4 binding to the c-fos promoter in which the consensus Klf4 binding site is absent (data not shown). Of interest, Klf4 induced the binding of p53 to the p53 binding site within the p21^WAF1/Cip1 promoter/enhancer. Finally, association of Klf4 and p53 with the p21^WAF1/Cip1 promoter/enhancer was examined in injured carotid arteries of wild-type mice by in vivo ChIP assays. Three days after injury, at which time Klf4 expression was induced in the SMC layer (Figure 1E and 1F), binding of both Klf4 and p53 to the p21^WAF1/Cip1 promoter/enhancer was increased within injured carotid arteries (Figure 7D). Taken together, these results provide novel evidence that Klf4 induces p21^WAF1/Cip1 gene transcription through Klf4 binding to the proximal and the distal Klf4 binding sites, as well as the recruitment of p53 to the p53 binding site within the p21^WAF1/Cip1 promoter/enhancer.

Discussion

In the present studies, we found that injury-induced suppression of SMC differentiation markers is delayed, but neointimal formation accelerated in the carotid arteries of tamoxifen-inducible conditional Klf4-deficient mice in vivo. We also showed that enhanced neointimal formation in conditional Klf4-deficient mice was caused by increased cellular proliferation in the media rather than an altered apoptotic rate. In addition, we demonstrated that Klf4 represses SMC proliferation by inducing p21^WAF1/Cip1 gene transcription through enhanced binding of Klf4 and p53 to the p21^WAF1/Cip1 promoter/enhancer. As such, results of the present studies provide the first in vivo example, to our knowledge, showing that deletion of a single transcription factor alone is sufficient to delay injury-induced downregulation of SMC differentiation markers and that a single transcription factor can function as a SMC growth repressor, as well as a repressor of SMC differentiation.

Although we discovered that injury-induced downregulation of SMC differentiation markers was delayed in SMCs and injury-induced cellular proliferation was accelerated in medial SMCs in tamoxifen-inducible conditional Klf4-deficient mice, it is possible that loss of Klf4 in other cell-types may also contribute to the phenotype observed in these mice. For example, although we showed that accumulation of macrophages and T lymphocytes was unaltered in Klf4 KO mice, we cannot rule out the possibility that loss of Klf4 may alter the function of these cells and thereby alter SMC phenotypic switching, growth, and lesion formation. Indeed, Feinberg et al²¹ reported that Klf4 was induced in cultured macrophages in response to interferon-γ, lipopolysaccharide, and tumor necrosis factor-α and that Klf4 partially mediated effects of these factors on the induction of inducible NO synthase by binding to the promoter/enhancer region of this gene. In addition, Hamik et al²⁰ showed that Klf4 was expressed in vascular endothelial cells in vivo and that Klf4 induced expression of antiinflammatory and antithrombotic factors including endothelial NO synthase and thrombomodulin, whereas knockdown of Klf4 led to enhancement of tumor necrosis factor-α–induced expression of vascular cell adhesion molecule-1 and tissue factor in cultured endothelial cells. Based on the results of these studies, enhanced neointimal formation in conditional Klf4-deficient mice may be caused by the combinatorial effects of multiple cell types including SMCs, macrophages, and endothelial cells. However, Klf4 function in macrophages was proinflammatory, and thus its loss in macrophages would be expected to reduce inflammation and neointimal formation, neither of which was observed in our conditional Klf4-deficient mice. The loss of antiinflammatory functions of Klf4 in endothelial cells could contribute to enhanced neointimal formation. However, of critical importance, results of in vivo ChIP assays in injured carotid arteries showed enhanced Klf4 binding to the promoter regions of SMC differentiation marker genes, as well as the p21^WAF1/Cip1 gene, thus strongly suggesting that Klf4 directly regulates expression of differentiation markers and p21^WAF1/Cip1 in SMCs. However, analysis of cell type–specific KO of the Klf4 gene will be required to clearly define the cell autonomous functions of Klf4 in vascular lesion development.

We have discovered that Klf4 functions as a suppressor of SMC differentiation as well as an inhibitor of SMC proliferation in response to vascular injury. These results are highly intriguing in that they provide the first example of a single transcriptional regulator that can function both as a SMC growth repressor and a repressor of SMC differentiation, a marked contrast to the belief that cellular proliferation is inversely related to cellular differentiation. Indeed, there are now several examples that differentiation and proliferation are not necessarily mutually exclusive processes in SMCs. First, during late embryogenesis and postnatal development, SMCs exhibit an extremely high rate of proliferation, yet at the same time, they undergo the most rapid rate of induction of expression of SMC differentiation markers in the rat and chicken aorta.^1,23 Second, PDGF-BB–induced suppression of SMC differentiation markers has been shown to be independent of its mitogenic effect.²⁴ Indeed, daily pulsing treatment with PDGF-BB in postconfluent culture of SMCs caused a sustained decrease in expression of SMC differentiation markers in spite of the absence of sustained mitosis. Third, phenotypically modulated SMCs within advanced human atherosclerotic lesions typically exhibit very low rates of proliferation.²⁵ Taken together, these results support a model wherein SMC growth and differentiation are not mutually exclusive.

Results of the present studies showed that Klf4 induced p21^WAF1/Cip1 gene expression through a novel distal Klf4 binding site located within close proximity of the p53 binding site, in addition to the proximal Klf4 binding site. We also found that Klf4-induced p21^WAF1/Cip1 expression was mediated, in part, by increased p53 binding to the p53 binding site. The increased p53 binding to the p21^WAF1/Cip1 promoter/enhancer was caused by the recruitment of p53, because Klf4 did not induce p53 expression. These results are supported by previous studies demonstrating a physical interaction between the zinc finger region of Klf4 and the N-terminal domain of p53 by coimmunoprecipitation assays.⁹ Our model is that vascular injury induces expression of Klf4, which counteracts against the mitogenic stimuli of SMCs by inducing p21^WAF1/Cip1 expression in concert with p53. However, in conditional Klf4-deficient mice, neointimal formation is accelerated because of the lack of p21^WAF1/Cip1 induction. Consistent with this model, overexpression of p21^WAF1/Cip1 inhibited neointimal formation following vascular injury, whereas p21^WAF1/Cip1 ablation enhanced neointimal formation in the carotid injury model.²⁶ Moreover, overexpression of p53 decreased neointimal formation following carotid injury, whereas loss of p53 accelerated formation of neointima in a vein bypass graft model, a femoral injury model, and in ApoE KO mice.²⁶ However, results of recent studies by Wassmann et al¹³ differed from ours regarding the mechanisms whereby Klf4 suppresses SMC proliferation, because those authors showed that Klf4 induced p53 expression in cultured SMCs. Although the reasons for this discrepancy are unknown, we found no differences in the apoptotic rate between conditional Klf4 mutant mice and control mice, suggesting the selective recruitment of p53 to the p21^WAF1/Cip1 promoter/enhancer rather than a simple induction of p53 expression. Recently, p53 has been shown to be subjected to multiple types of posttranslational modifications, including acetylation, methylation, phosphorylation, and ubiquitylation.²⁷ It will be interesting to determine whether Klf4 regulates posttranslational modifications of p53 that selectively enhance its binding to the p21^WAF1/Cip1 promoter/enhancer.

In summary, we provide evidence that Klf4 plays a critical role in regulation of SMC phenotypic switching and neointimal formation in response to vascular injury. Indeed, Klf4 is an early suppressor of SMC differentiation markers and an inhibitor of SMC proliferation by inducing p21^WAF1/Cip1 expression in concert with p53. Further studies are needed to determine the role of Klf4 in the development of vascular diseases, including atherosclerosis in experimental animal models, as well as in humans.

Footnotes