Krüppel-like Factor 5 Shows Proliferation-specific Roles in Vascular Remodeling, Direct Stimulation of Cell Growth, and Inhibition of Apoptosis

Abstract

Krüppel-like factor 5 (KLF5), originally isolated as a regulator of phenotypic modulation of vascular smooth muscle cells, induces pathological cell growth and is expressed in the neointima. Although induction of KLF5 up-regulates growth factors like platelet-derived growth factor-A chain, how KLF5 actually contributes to vascular remodeling, notably its direct effects on cell proliferation, had been poorly clarified. To investigate the effects of KLF5 on neointimal formation, we at first performed adenoviral overexpression of KLF5 to rats subjected to carotid balloon injury. Neointimal formation and proliferating cell nuclear antigen-positive rate were significantly increased at 14 days after injury in the KLF5-treated animals. At the cellular level, overexpression of KLF5 also resulted in markedly increased cell proliferation and cell cycle progression. As a molecular mechanism, we showed that KLF5 directly bound to the promoter and up-regulated gene expression of cyclin D1, as well as showing specific transactivation of cyclins and cyclin-dependent kinase inhibitors in cardiovascular cells. Conversely, knockdown of KLF5 by RNA interference specifically down-regulated cyclin D1 and impaired vascular smooth muscle cell proliferation. Furthermore, KLF5 attenuated cleavage of caspase-3 under conditions of apoptotic stimulation. Moreover, KLF5-administered animals exhibited a significant decrease in terminal deoxynucleotidyltransferase-mediated dUTP nick end-labeling-positive cells in the medial layer, suggesting inhibition of apoptosis in the early phase after denudation. These findings collectively suggest that KLF5 plays a central role in cardiovascular pathologies through direct and specific stimulation of cell growth as well as inhibition of apoptosis.

Krüppel-like factor 5 (KLF5,4 also referred to as BTEB2 in earlier studies but now understood to be a truncated form of KLF5) is a zinc finger transcription factor that belongs to a family known as the Sp/KLF factors, which are comprised of ∼20 members in mammals (1, 2). Each has individually important biological functions on cell proliferation, apoptosis, development, and oncogenic processes, among others, as shown by numerous studies (3, 4). KLF5 is most commonly accepted to be an oncogenic factor. Its constitutive expression in fibroblasts results in increased cell colony formation, and oncogenic properties of H-Ras and Wnt-1 are mediated by KLF5 (5-7). In clinical tumors, the grade of malignancy has been shown to be associated with the expression level of KLF5. In prostate and breast cancers, however, KLF5 is thought to be a tumor suppressor; thus, this factor shows context-dependent properties (5, 6, 8-11).

We first identified this factor as a transcription factor that regulates expression of the embryonic form of smooth muscle myosin heavy chain (SMemb/NMHC-B), which is selectively expressed in the proliferative dedifferentiated smooth muscle phenotype (12). Vascular remodeling including coronary artery disease is characterized by atherosclerotic plaque formation in which proliferating smooth muscle cells and neointimal cells encroach upon the vascular lumen (13). Further apoptosis of vascular cells affects remodeling by inducing regression of neointimal hyperplasia after vascular injury (e.g. restenosis) (14, 15).

KLF5 functions to mediate cardiovascular remodeling in response to external stress (e.g. angiotensin II) by regulating atherosclerosis, angiogenesis, and cardiac hypertrophy (16). It is expressed in proliferating smooth muscle cells of coronary artery lesions, namely proliferative stellate-shaped cells, and expression of this factor in lesions is clinically associated with restenosis and cardiac allograft vasculopathy (17, 18). KLF5 expression is therefore associated with proliferating smooth muscle cells in the cardiovasculature (19).

However, it had yet to be shown how KLF5 directly induces cell proliferation and regulates apoptosis and how KLF5 mediates proliferative activity in vascular lesions and cells. In the present study, we have addressed the direct actions of KLF5 on cell growth and apoptosis in the vasculature and its constituting cells.

EXPERIMENTAL PROCEDURES

Preparation of Plasmid Constructs and Recombinant Adenoviruses—pCAG-KLF5 and adenoviral KLF5 and Sp1 constructs were constructed as described previously (16). The human cyclin D1 promoter constructs were from Dr. D. Nagata (20).

Rat Arterial Injury Model, Adenoviral Gene Delivery, and Immunohistochemistry—Male Sprague-Drawley rats (weighing 400-450 g) were anesthetized with chloral hydrate (Wako, 370 mg/kg of body weight, intraperitoneally). Balloon denudation of the left common carotid artery was performed as described previously (8). Immediately after injury, 50 μl (2 × 10⁹ plaque-forming unit/ml) of adenoviral construct (empty, KLF5 or Sp1) was infused and incubated for 30 min. 14 days after operation, rats were euthanized with a lethal dose of anesthetic. Serial cross sections (6 μm) were obtained from each carotid artery sample and stained with Elastica Van Gieson for morphometric analysis or prepared for immunohistochemistry with anti-KLF5 rat monoclonal antibody (KM1785, 1:4000) or anti-proliferating cell nuclear antigen (PCNA) antibody (Dako, 1:1000). All experimental protocols complied with the guide-lines for animal experiments of the University of Tokyo.

Cell Preparation and Culture—Rat aortic smooth muscle cells (VSMCs) were isolated from the thoracic aorta as described previously and cultured in Dulbecco's modified Eagle's medium/F12 medium (Invitrogen) supplemented with 10% fetal calf serum (21). Cells between 7 and 12 passages were used in experiments. C2/2 rabbit vascular smooth muscle cells and 10T1/2 mouse myofibroblasts were maintained in Dulbecco's modified Eagle's medium (Sigma) supplemented with 10% fetal calf serum, 100 μg of streptomycin/ml, and 100 units of penicillin G/ml.

Vascular Smooth Muscle Cell Proliferation Assay—2 × 10⁵ rat VSMCs were at first infected with each adenoviral construct (100 multiplicities of infection) for 12 h, reseeded in duplicate in 6-well plates, and then starved from serum for 24 h. After serum stimulation, cells were washed with phosphate-buffered saline, trypsinized, and then counted using a cell counter (Beckman-Coulter). Experiments were done in triplicate.

Cell Cycle Analysis—Cells were seeded into 6-well plates at a density of 1 × 10⁵/well. After serum starvation and transfection of adenoviral construct, cells were stimulated with 10% fetal calf serum. 24 h later, cells were trypsinized, washed with phosphate-buffered saline, and then fixed with 70% ethanol at -20 °C. The day before analysis, cells were resuspended in propidium iodide buffer (50 μg/ml propidium iodide, 0.1% Triton-X, 0.1 mm EDTA, 0.05 mg/ml RNase). Cells were transferred to a Falcon 33654 tube and then subjected to flow cytometry (FACSCalibur, BD Biosciences). Cell cycle distribution was calculated using the ModFit software (Verity Software House).

Western Blot Analysis—Western blotting of proteins extracted from VSMCs or rat carotid artery specimens was performed using anti-KLF5 rat monoclonal antibody (KM1785, 1:4000), anti-cyclin D1 mouse monoclonal antibody (BD Biosciences Pharmingen, 1:500), anti-Sp1 rabbit polyclonal antibody (Santa Cruz Biotechnology, 1:500), anti-p21 mouse monoclonal antibody (Santa Cruz Biotechnology, 1:200), anti-p27 mouse monoclonal antibody (BD Transduction Laboratories, 1:2000), anti-cyclin E mouse monoclonal antibody (Santa Cruz Biotechnology, 1:500), anti-cyclin A rabbit polyclonal antibody (Upstate Biotechnology, 1:500), and anti-cleaved caspase 3 rabbit polyclonal antibody (Cell Signaling, 1:500). Protein loading was controlled by probing the same membrane with anti-glyceraldehyde-3-phosphate dehydrogenase mouse monoclonal antibody (Upstate Biotechnology, 1:10000) or by anti-β-actin mouse monoclonal antibody (AbCam; 1:1000).

Semiquantitative Reverse Transcriptase (RT)-PCR—Total RNA was purified from VSMCs and subjected to RT-PCR as described previously (22). The PCR primers were: rat KLF4, 5′-CTGGCGAGTCTGACATGGCTGTCAG-3′ and 5′-CGCCACTCTCCAGGTCTGTGCCAAC-3′; rat myocardin, 5′-CCAAACCAAAGGTGAAGAAGCTC-3′ and 5′-TGTCTTAACTCTGACACCTTGAG-3′; rat KLF6, 5′-TATCTTCAGGATGAGCCCTGCTAC-3′ and 5′-AGACTTCACCAATGGGATCAGAGG-3′; rat p21, 5′-AGTATGCCGTCGTCTGTTCG-3′ and 5′-GAGTGCAAGACAGCGACAAG-3′; and rat p27, 5′-CAGCTTGCCCGAGTTCTA-3′ and 5′-TGGGGAACCGTCTGAAAC-3′.

Co-transfection Reporter Assay—Co-transfection reporter assay was performed as described previously using C2/2 cells (5 × 10⁴ cells/well in 24-well plate) transfected with 200 ng of pGL2-human cyclin D1 promoter deletion constructs and 200 ng of pCAG-KLF5 (20). Assays were done in duplicate, and expression levels of KLF5 were confirmed by immunoblotting with anti-KLF5 antibody.

Electrophoretic Mobility Shift Assay (EMSA)—EMSAs were performed using fluorescein isothiocyanate-labeled double-stranded, synthetic oligonucleotides. Overlapping Sp1-binding sites located upstream (Sp1-1 or Sp1-2) were analyzed. Nucleotides sequences of the sense strand were as follows: Sp1-1wt2wt, 5′-GGCGCCCGCGCCCCCCTCCCCCTGC-3′; Sp1-1mut2mut, 5′-GGCGCaatttCCCCCTaattaTGC-3′; Sp1-1mut2wt, 5′-GGCGCaatttCCCCCTCCCCCTGC-3′; Sp1-1wt2mut, 5′-GGCGCCCGCGCCCCCTaattaGC-3′. The underlined letters depict nucleotide substitutions to induce mutations. Reaction mixtures contained 50 ng of recombinant KLF5-DNA-binding domain protein and 50 ng of probe.

Chromatin Immunoprecipitation Assays—Chromatin immunoprecipitation assays were performed as described previously with rat VSMCs cultured in 10-cm dishes stimulated with 10% fetal calf serum or kept deprived from serum after starvation for 48 h (23). Chromatin samples were immunoprecipitated with anti-KLF5 rabbit polyclonal antibody (raised against a synthetic peptide) or preimmune serum as control. PCR was performed with the following primers: rat cyclin D1 promoter, 5′-CGGCGATTTGCATATCTACGAAGG-3′ and 5′-AAGCCGGGCAGAGAAAAAGGAG-3′.

Construction of Small Hairpin Interfering RNA for KLF5—We searched target sequences for KLF5 RNA interference by an online siRNA sequence selector (Clontech) and used the following sequence, 5′-GGTCCAGACAAGATGTGAA-3′. Designed oligonucleotides were then annealed and ligated into pSIREN, a human pU6 promoter plasmid (Clontech). The ligated construct was verified for KLF5 knockdown and amplified for transfection.

Terminal dUTP Nick-end Labeling (TUNEL) Staining and Apoptotic Assays—5 × 10⁴ rat VSMCs were transfected with adenoviral constructs and then treated with 2 μmol/liter anisomysin (Sigma). Apoptotic cells were detected using the in situ cell death detection kit (Takara) and counterstained with propidium iodide. Balloon injured rat carotid arteries were harvested 48 h after balloon injury and stained for apoptotic nuclei using the in situ cell death detection kit (Takara). Specimens were treated with TUNEL reaction mixture and then diaminobenzidine-stained.

Statistical Analysis—Data are shown as mean ± S.D. Differences between two groups were analyzed using the Student's t test. The threshold of significance was taken as p < 0.05.

RESULTS

KLF5 Promotes Neointimal Formation and Cell Proliferation—KLF5 expression is induced in the aortic wall by vascular injury in rats and in rabbits and has been suggested to contribute to the remodeling process. However, how it promotes vascular remodeling has not been clarified. We thus overexpressed KLF5 in rats subjected to carotid balloon injury by adenoviral transfer to address the direct effects of KLF5 on vascular remodeling. Under conditions in which overexpression of KLF5 in rat carotid arteries was confirmed by immunohistochemical staining and Western blotting (Fig. 1, A and B), histological examination showed that neointimal formation was markedly increased by adenoviral transfer of KLF5 at 2 weeks after balloon injury (Fig. 1C). Morphometric analysis (n = 10 each) confirmed significantly increased intima/media ratio (I/M) in the KLF5 group (I/M: 1.40 ± 0.23) as compared with Sp1 or empty groups. Sp1, a similar zinc finger transcription factor, was used as control. There was no significant difference in neointimal formation between the empty group (I/M: 0.66 ± 0.26) or the Sp1 group (I/M: 0.70 ± 0.23). PCNA immunostaining showed a significantly higher positive rate in the intimal and medial layers of the KLF5 group (52 ± 12%) as compared with those in the empty group (16 ± 3.7%) or in the Sp1 group (24 ± 3.1%) (Fig. 1D).

FIGURE 1.

Effects of KLF5 on neointimal formation. KLF5 was overexpressed by administration of adenoviral (Adv) construct to rats subjected to carotid balloon injury. A, immunohistochemical (IHC) staining for KLF5 in rat carotid arteries 14 days after injury and adenoviral transfection. Brown-stained nuclei are KLF5-positive cells. Counterstaining was done with methyl green (scale bar, 20 μm). B, Western blot analysis using protein extracted from balloon injured rat carotid arteries (7 days after injury). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a protein loading control. C, representative figures of rat carotid arteries 14 days after injury and transfection with adenoviral constructs. Elastica Van Gieson staining is shown. Neointimal formation was then calculated as the intima/media ratio based on morphometric analysis (n = 10 each, scale bar, 100 μm, ^**, p < 0.01). D, immunohistochemical analysis using anti-PCNA antibody and counterstained with hematoxylin-eosin. The PCNA-positive rate was calculated as the ratio of PCNA-positive nuclei/hematoxylin-positive nuclei (scale bar, 20 μm, ^**, p < 0.01).

KLF5 Stimulates SMC Proliferation and Accelerates Cell Cycle Progression—Based on the stimulatory effect of KLF5 on neointimal formation in the animal model, we next examined the effects of KLF5 overexpression on cell proliferation using primary cultured rat VSMCs. KLF5 induction resulted in increased cell proliferation after serum stimulation as compared with those of Sp1 or empty groups (Fig. 2A). Cell cycle analysis showed markedly increased S-phase bulging after KLF5 transfection both in myofibroblasts and in VSMCs, which are the two major cellular components of vascular remodeling (Fig. 2B) (24).

FIGURE 2.

Effects of KLF5 on SMC proliferation and cell cycle progression. A, cell count of rat VSMCs transfected with adenoviral KLF5 and other constructs. Experiments were done in triplicate, and the error bar denotes S.D. (^**, p < 0.01). B, flow cytometric analysis on cell cycle distribution. C, effects on expression levels of cyclins, cyclin-dependent kinase inhibitors, and other cellular factors analyzed by Western blot (C) and semiquantitative RT-PCR (D). Lanes 1-3 are under serum starvation and lanes 4-6 are 24 h after serum stimulation. Note that higher expression levels of cyclin D1 were seen in the KLF5 group, especially 24 h after stimulation. Amounts of loaded protein or RNA were confirmed with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) blotting or 18 S ribosomal RNA transcripts. Adv, adenoviral.

To further investigate the effects of KLF5 on cell cycle progression, expression levels of cyclins, cyclin-dependent kinase inhibitors (CKIs), and other cellular factors were analyzed by Western blot (Fig. 2C) and RT-PCR analysis (Fig. 2D). Although serum starvation alone resulted in up-regulation of expression levels of cyclin D1 in the Sp1 group, expression levels were markedly augmented in the KLF5 group at 24 h after serum stimulation. The expression level of cyclin E was higher in the Sp1 group, whereas neither cyclin A nor p27 showed significant difference in expression levels after serum stimulation. In contrast, the expression level of p21 in the KLF5 group was higher than that of the empty group, which was much lower as compared with that induced in the Sp1 group. These different effects between KLF5 and Sp1 on the cyclins and CKIs may explain the different effects among these factors found in cell growth and cell cycle distribution studies. Under the same conditions, we also investigated the transcription levels of other KLFs, namely KLF4, which has also been implicated to be involved in regulating SMC growth, and myocardin, as myocardin has been shown to be a growth inhibitor of SMCs (25). KLF5 overexpression did not influence the expression levels of KLF4; however, it suppressed those of myocardin.

KLF5 Is Directly Recruited to the Cyclin D1 Promoter in VSMCs—Previous studies have shown KLF5 to regulate cyclin D1 in fibroblasts or cancer cell lines (6, 10). To address the transactivation activity of KLF5 on cyclin D1 in cardiovascular cells, we transfected deletion mutants of the human cyclin D1 promoter reporter into C2/2 cells aiming to map cis-elements that are required for transcriptional activity of the cyclin D1 promoter. Schematic illustrations of the cyclin D1 promoter-reporter deletion constructs are shown (Fig. 3, A, B, and D). Deletion of the putative AP1-binding sites, two NFκB-binding sites, and STAT-binding sites did not affect luciferase reporter activity. pGL2/-161/Sp1, which encodes multiple Sp1-binding sites, an activating transcription factor/cAMP-response element-binding protein-binding site, and an NFκB-binding site, had almost the same effect on promoter activity as pGL2/-1719, which encoded the full-length cyclin D1 promoter. Interestingly, when the Sp1-binding sites were deleted (pGL2/-95/ΔSp1), promoter activity decreased to one-half of that of the full-length cyclin D1 promoter. Further deletion of the putative activating transcription factor/cAMP-response element-binding protein-binding site and the NFκB-binding site (pGL2/-23/ΔNFκB3) resulted in reduction of promoter activity almost to the same level as the pGL2-basic vector (Fig. 3C). These results suggested that cis-regulatory elements involved in KLF5-induced transcriptional activation of cyclin D1 lie between -161 bp and -95 bp of the Sp1-binding sites, which is reasonable given that KLF5 and all other Sp/KLF members commonly bind GC-rich sites/Sp1-binding sites (1).

FIGURE 3.

Functional analysis of KLF5-binding sites in transcriptional regulation. A, schematic illustration of the cyclin D1 promoter-reporter deletion constructs that were used. Diagrams depict locations of cis-elements in the cyclin D1 promoter. CREB, cAMP-response element-binding protein; LUC, luciferase. B, schematic illustration of the cyclin D1 promoter-reporter deletion constructs that were used. wt, wild type. C, co-transfection reporter assay to map the KLF5-responsive region of the cyclin D1 promoter. Assays were done in duplicate, and data are shown as average. D, mutation alanalysis of Sp1-binding sites in the cyclin D1 promoter. Diagrams show mutated Sp1-binding site reporter constructs. Asterisks indicate the positions at which the point mutation (mut) was introduced. E, the horizontal axis shows the relative reporter activities. Assays were done in duplicate, and data are shown as average. F, EMSAs were performed using fluorescein isothiocyanate-labeled probes encoding the upstream Sp1-binding sites (Sp1-1 and Sp1-2). 100× molar excess of non-labeled double-stranded oligonucleotide encoding wild-type sites (lane 3) or mutant sites (lane 4) were used for competition experiments. Mutations to the Sp1-1 and Sp1-2 sites were introduced separately into the probe as shown in lanes 5 and 6 because of their overlapping nature. Nucleotide sequences of wild-type and mutated Sp1-binding sites are shown. DBD, KLF5-DNA binding domain. G, recruitment of KLF5 to the cyclin D1 promoter in vivo. Input samples represent 0.5% of total DNA, whereas immunoprecipitations (IP) include 5% of the resuspended DNA as shown. Note that direct recruitment of KLF5 to the cyclin D1 promoter was only seen under serum stimulation conditions (lane 6). The error bar denotes S.D.

To further examine the functions of the Sp1-binding sites, analysis using point mutations to each of the sites was done. Insertion of a mutation to the second Sp1-binding site (Sp1-2) resulted in significant reduction in reporter activity as compared with the wild-type construct. We also saw a marginal elevation in reporter activity with mutation of the first Sp1-binding site (Sp1-1). Effects were not seen with mutations in the two downstream Sp1-binding sites (Fig. 3E). EMSA experiments confirmed that KLF5 preferentially binds the Sp1-binding site, Sp1-2. Under conditions in which KLF5 bound a DNA probe containing the first two Sp1-binding sites (Fig. 3F, lane 2) in a sequence-specific manner, as shown by the fact that wild-type competitor could compete out this band (lane 3) but a competitor containing mutant sequences for both Sp1-1 and Sp1-2 could not (lane 4), specific mutations against Sp1-1 and Sp1-2 were tested. These results showed that a mutant probe for Sp1-2 (lane 6) could not compete as well as a mutant for Sp1-1 (lane 5), thus suggesting that KLF5 shows preferential binding to Sp1-2. Collectively, the reporter assay and EMSA show that the actions of KLF5 on the cyclin D1 promoter are mediated by the Sp1-2 site.

To further investigate the recruitment of KLF5 to the rat cyclin D1 promoter in vivo, a chromatin immunoprecipitation assay was done. PCR analysis was performed to amplify the region of the rat cyclin D1 promoter from -213 bp to -69 bp that contains the Sp1-binding motifs. In vivo binding of KLF5 to the cyclin D1 promoter was clearly induced in response to serum stimulation (Fig. 3G), which suggested that this interaction was induced by external growth stimulation. This finding is consistent with past studies that have shown that induction of KLF5 activates the PDGF-A chain promoter, which in turn promotes proliferative change through phenotypic modulation mediated by such para/autocrines on VSMCs (16, 22).

KLF5 Knockdown Attenuates Cell Proliferation and Cell Cycle Progression—To further investigate the role of KLF5 on cell proliferation in vascular cells, knockdown of KLF5 was carried out using plasmid-based siRNAs. Under conditions of specific knockdown of KLF5 in rat VSMCs, marked suppression of cyclin D1 expression was seen, whereas cyclin A expression was not affected (Fig. 4A). On analysis of cell growth, KLF5 knockdown resulted in significant suppression of VSMC proliferation after serum stimulation (Fig. 4B). To better characterize the inhibitory effect of KLF5 knockdown on cell proliferation, cell cycle distribution was analyzed (Fig. 4C). 38% of cells were in the S-phase in the empty group, 30% were in the si-luciferase (siLuc) group, and 18% were in the siKLF5 group. Therefore, S-phase transition was markedly decreased by RNA interference against KLF5.

FIGURE 4.

Effects of KLF5 knockdown on cyclin D1 expression, cell proliferation and cell cycle progression. A, expression levels of KLF5 and other cellular factors by RNA interference. Expression levels of KLF5, cyclin D1, and cyclin A are shown as bar graphs after densitometric analysis. B, the actual number of cells was counted in triplicate, and the average was shown as line plots (^**, p < 0.01). C, staining with 7-amino-actinomysin D followed by analysis with flow cytometer and cell cycle distribution. The error bar denotes S.D.

KLF5 Expression Attenuates VSMC Apoptosis—Cell proliferation is inextricably linked to apoptosis. To investigate the contribution of apoptotic properties of KLF5 on cell proliferation, we examined the effects on caspase-3 cleavage by Western blot using rat VSMCs before and after serum stimulation. Cleavage of caspase-3 was seen in the empty and Sp1 groups under serum deprivation and in the Sp1 group after stimulation. In contrast, cleavage was not seen either under serum deprivation or after stimulation in the KLF5 group (Fig. 5A). Even under cytotoxic conditions using anisomycin, KLF5 exhibited protective effects against apoptosis (Fig. 5B).

FIGURE 5.

Effects of KLF5 on SMC apoptosis. A, uncleaved and cleaved levels of apoptotic caspase-3 analyzed by Western blot. Note that cleaved caspase-3 was specifically not seen in the KLF5 group under serum deprivation. Amounts of loaded protein were confirmed with β-actin blotting. Adv, adenoviral. B, the effects of KLF5 on apoptosis under anisomycin treatment (2 μmol/liter). The right lower insets show the same distribution of cells in each group counterstained with propidium iodide. The right bar graph shows TUNEL-positive apoptotic cell rate (^**, p < 0.01). The error bar denotes S.D. C and D, the effects of KLF5 overexpression on VSMC apoptosis in the later (14 days) (C) or in the early (48 h) (D) phase in rat carotid arteries after balloon injury and adenoviral transfection as assessed by TUNEL staining. Brown-stained nuclei were TUNEL-positive and were slightly counterstained with methyl green. Intestine samples were stained as positive control. The TUNEL-positive rate was calculated as the ratio of TUNEL-positive nuclei/methyl green-positive nuclei (scale bar, 20 mm, ^**, p < 0.01). The error bar denotes S.D.

Concurrently, we analyzed apoptosis in the injured arteries using TUNEL staining. At 2 weeks after injury, there were no apoptotic cells seen in the neointimal and medial layers in either group (Fig. 5C). However, arteries administered with KLF5 showed significantly decreased apoptotic cells in the medial layer 48 h after injury (Fig. 5D). This indicates that the anti-apoptotic effect by KLF5 was exerted in the early phase after vascular injury.

DISCUSSION

Direct Promotion of VSMC Proliferation by KLF5 via Transactivation of Cyclin D1 Transcription—In the present study, KLF5 overexpression resulted in promotion of cell proliferation, and gene silencing of KLF5 with RNA interference showed marked suppression of cyclin D1 expression and decreased cell growth. These results suggest that KLF5 acts to induce cell proliferation in cardiovascular cells, VSMCs, and myofibroblasts. As cell growth is coupled with cell cycle progression, we focused on understanding the mechanisms whereby KLF5 regulates cyclin D1, a key initiator of the cell cycle engine. To date, a direct relationship between KLF5 and cyclin D1 had not yet been shown. Our studies for the first time showed direct activation of cyclin D1 by KLF5 with reporter assays and chromatin immunoprecipitation and also that a putative binding motif lies around the Sp1-binding sites in the cyclin D1 promoter region. Thus, mechanistically, induction of KLF5 leads to direct transactivation of cyclin D1 expression, which subsequently stimulates cell proliferation.

Transcriptional mechanisms at the chromatin level are another area of interest. Past studies suggest that the coactivator/acetylase p300 is involved in transcriptional activation by KLF5 (8, 26). For regulation of transcription of the cyclin D1 promoter, it appeared that p300 at least in part is one of the chromatin cofactors that are involved. RNA interference studies showed that knockdown of p300 attenuated acetylated histone levels (as a measure of transcriptional activation), and conversely, that knockdown of KLF5 also attenuated acetylated histone levels as well as promoter recruitment of p300 to a promoter region that mediates KLF5 actions, thus suggesting cooperative actions of KLF5 and p300 in transcriptional activation of this gene promoter.5

Significance of KLF5 in Induction of CDKs and CKIs—Cell proliferation is coupled with cell cycle progression and subsequent mitosis (27). Further, cell cycle progression is mediated by cyclins (e.g. cyclin D1, cyclin E, and cyclin A) (28) and inhibited by CKIs (e.g. p21 and p27) (29). Initiation of S-phase transition is in particular promoted by cyclin D1 (30). In cardiovascular cells, induced and forced-expressed KLF5 showed direct recruitment onto the cyclin D1 promoter coupled with markedly increased transactivation of cyclin D1 in contrast to relatively low transactivation of p21, which coordinately resulted in marked cell proliferation. In general, p21 induction suppresses CDK/cyclin complex kinase activity and subsequently induces cell growth arrest (31). However, it has also been reported that proper cell cycle progression requires coordinated assembly of p21 and CDK/cyclin complex formation in addition to repression of p21, which results in decreased cell growth (29, 32-34). Surprisingly, in our experiment, overexpression of Sp1 in VSMCs resulted in lower induction of cyclin D1 and much higher induction of p21, which subsequently resulted in negative regulation of cardiovascular cell proliferation. KLF6 up-regulation by Sp1 may have further reinforced the expression of p21 in an additive manner (35). This result was consistent with observations on caspase-3 cleavage. KLF5 overexpression did not, however, reduce the expression levels of p27 at the mRNA or protein levels. This may be due at least in part to context-dependent regulation of p27 expression in VSMCs. Although both PDGF-BB and angiotensin II are mitogenic stimuli, PDGF-BB attenuates p27 expression, but angiotensin II has no effect on expression, thus suggesting the possibility of differential regulation of p27 under cell growth conditions (36). The different results on cell growth between Sp/KLF families may be due to the different proportion of transactivation effects on CDKs and CKIs. It was also surprising that overexpression of these factors showed opposite effects, one showing proliferative effects, whereas the other showed apoptotic effects, although these two transcription factors belong to the same Sp/KLF family that has similar DNA binding activities (1).

Contribution of KLF5 to Vascular Remodeling, Inhibition of Apoptosis, and Direct Promotion of VSMC Proliferation—The role of KLF5 in vascular pathology was first recognized upon its identification as a regulatory factor of phenotypic modulation that directly controls expression of SMC marker genes and paracrine factors such as PDGF-A chain and transforming growth factor-β in vitro and in vivo (16, 19, 22). KLF5 is presently understood to exhibit growth promoting effects in vascular remodeling through these para/autocrines (37).

In general, chronological progression of vascular remodeling in response to external injury consists of several phases of cellular activities as exemplified by apoptosis in initial phases and proliferation in later phases. Concerning early apoptotic effects, apoptotic SMC death is seen at early time points after injury, and higher apoptotic rates correlate with decreased neointimal formation or restenosis after angioplasty (38-40). Significantly decreased apoptosis, as seen in the medial layer of the KLF5 group, recognized as early as 48 h after injury, may have led to increased neointimal formation. In addition to the anti-apoptotic property shown in the present in vitro experiments, KLF5 heterozygous knock-out mice in fact show increased apoptosis after vascular wire injury, thus confirming a regulatory role of KLF5 in vascular pathological apoptosis (41). Concerning late proliferative effects, overexpression of KLF5 significantly increased neointimal formation through direct stimulatory effects on vascular remodeling. KLF5 also showed proliferative effects on SMCs both in vitro and in vivo in the present study (SMC proliferation assay, PCNA labeling, etc). TUNEL staining, however, did not show any apoptotic cells at 14 days after injury in any group, indicating that the stimulatory effect on vascular remodeling in the late phase was due mainly to activation of cell growth.

Taken together, KLF5 regulates both apoptosis in the early phase and proliferation in the late phase, thus indicating that this factor acts as a bimodal and likely central regulator of proliferative cardiovascular pathologies.