Augmentation of Therapeutic Angiogenesis Using Genetically Modified Human Endothelial Progenitor Cells with Altered Glycogen Synthase Kinase-3 (cid:1) Activity*

Previously we reported that inhibition of glycogen synthase kinase-3 (cid:1) (GSK3 (cid:1) ), a key regulator in many intracellular signaling pathways, enhances the survival and migration of vascular endothelial cells. Here we investigated the effect of inhibition of GSK3 (cid:1) activity on the angiogenic function of endothelial progenitor cell (EPC) and demonstrated a new therapeutic angiogenesis strategy using genetically modified EPC. As we previously reported, two biologically distinct types of EPC, spindle-shaped “early EPC” and cobblestone-shaped “late EPC” could be cultivated from human peripheral blood. Catalytically inactive GSK3 (cid:1) gene was transduced into both EPC. Inhibition of GSK3 (cid:1) signaling pathway (GSK3 (cid:1) -KM), where lysine resi-dues at positions 85 and 86 were mutated to methionine and alanine, respectively, and control adenoviral vector expressing green fluorescent protein (GFP) was used for gene delivery (36, 37). EPC cultured for 10 days were transduced with multiplicity of infection 50 of adenovirus-GSK3 (cid:1) -KM (AdGSK-KM) and control adenovirus-GFP (AdGFP) in EGM-2 medium with 2% fetal bovine serum for 12 h. Effective gene transfer was confirmed by more than 90% of green fluorescence-positive control EPC transduced with AdGFP, and cells were assayed or harvested. Western Blot Analysis— Western blot analysis was performed to con- firm that the transduction of AdGSK-KM into EPC could lead to the sufficient expression of inactive phosphorylated GSK3 (cid:1) (36, 37). Early EPC and late EPC were transduced with multiplicity of infection 50 of AdGFP or AdGSK-KM for 12 h and washed in phosphate-buffered saline (PBS) and harvested by scraping in lysis buffer. After determi-nation of concentration with a protein assay kit (Pierce), 30 (cid:3) g of protein was separated by SDS-PAGE and transferred to a polyvinyli-dene difluoride membrane (Millipore Corp., Bedford, MA). The mem- brane was blocked with PBS containing 0.3% Tween 20 (T-PBS) and 3% dry milk and incubated with primary antibody overnight at 4 °C. Then membrane was washed three times with T-PBS and reblocked and incubated with secondary antibody fo r 1 h atroom temperature. ECL Plus (Amersham Biosciences) was used for detection. To reprobe the membrane, it was treated with Restore Western blot stripping buffer (Pierce). The primary antibodies used with anti-phospho-GSK3 (cid:1) antibody (1:750 dilution; Cell Signaling Technology), anti-total GSK3 antibody (1:750 dilution; Santa Cruz Biotechnology),

Previously we reported that inhibition of glycogen synthase kinase-3␤ (GSK3␤), a key regulator in many intracellular signaling pathways, enhances the survival and migration of vascular endothelial cells. Here we investigated the effect of inhibition of GSK3␤ activity on the angiogenic function of endothelial progenitor cell (EPC) and demonstrated a new therapeutic angiogenesis strategy using genetically modified EPC. As we previously reported, two biologically distinct types of EPC, spindle-shaped "early EPC" and cobblestone-shaped "late EPC" could be cultivated from human peripheral blood. Catalytically inactive GSK3␤ gene was transduced into both EPC. Inhibition of GSK3␤ signaling pathway led to increased nuclear translocation of ␤-catenin and increased secretion of angiogenic cytokines (vascular endothelial growth factor and interleukin-8). It enhanced the survival and proliferation of early EPC, whereas it promoted the survival and differentiation of late EPC. Transplantation of either of these genetically modified EPC into the ischemic hind limb model of athymic nude mouse significantly improved blood flow, limb salvage, and tissue capillary density compared with nontransduced EPC. Inhibition of GSK3␤ signaling of either of these genetically modified EPC augmented the in vitro and in vivo angiogenic potency of these cell populations. These data provide evidence that GSK3␤ has a key role in the angiogenic properties of EPC. Furthermore, the genetic modification of EPC to alter this signaling step can improve the efficacy of cell-based therapeutic vasculogenesis.
Previous studies have identified that normal adults have a small amount of circulating endothelial progenitor cells (EPC) 1 in peripheral blood (1,2). In response to cytokine stimulation, these cells are mobilized from bone marrow and home to the ischemic tissue and contribute to new vessel formation (3,4). According to these characteristics, EPC have been investigated as an agent for therapeutic angiogenesis (5)(6)(7). These studies have shown that administration of EPC to animals with limb or myocardial ischemia can enhance neovascularization, salvage tissue, and improve myocardial function. Human clinical trials using autologous progenitor cells also showed improved myocardial viability and blood flow (8).
However, there are several limitations in the therapeutic application of EPC. First, the amount of available EPC required for therapeutic angiogenesis is limited, so large amounts of peripheral blood are required (4, 9 -11). Furthermore, the isolated circulating cell population that contributes to postnatal neovascularization is heterogeneous and displays variable morphological growth characteristics. In this regard, EPC have been described either as spindle-shaped cells with limited proliferation capacity (1,2,5,12,13), or as cobblestoneshaped cells that are morphologically similar to mature EC cells and can be grown exponentially for more than 60 days (14 -17). Therefore, the genetic modification of EPC might overcome some of these limitations by altering the cell phenotype to promote the efficiency of cell-based therapeutic angiogenesis.
Recently, efforts have been made to dissect the regulatory pathways that control the phenotypes of stem cell populations. Many of these studies have focused on the role of the canonical Wnt/␤-catenin signaling pathway. For example, it has been shown that Wnt signaling promotes the self-renewal of hematopoietic side population cells, neuronal precursor cells, and embryonic stem cells (18 -21). Other studies have examined the phosphoinositide 3-kinase/Akt-signaling pathway. For example, it has been shown that elevated Akt signaling can improve the survival of transplanted mesenchymal stem cells (22) and that phosphoinositide 3-kinase/Akt-signaling is a determinant of side population cell phenotype through regulation of the Bcrp1 transporter expression (23).
Glycogen synthase kinase-3␤ (GSK3␤) is a serine/threonine kinase that is under the control of both Wnt (24,25) and phosphoinositide 3-kinase/Akt (26,27) signaling pathways, which are described above. Recently, it has been shown that a pharmacological inhibitor of GSK3 promotes the self-renewal of embryonic stem cells in vitro (20). GSK3␤ controls several downstream transcription factors that are crucial in cell survival and function including ␤-catenin (28), heat shock factor-1 (29), cAMP-response element-binding protein (30), AP-1 (31), Myc (32), NFAT (33), CCAAT/enhancer-binding protein ␣ (34), and cyclin D1 (35). GSK3␤ has also been shown to be involved in the regulation of angiogenesis through its ability to modulate vascular endothelial cell migration and survival (36). Therefore, we reasoned that it might be possible to genetically manipulate GSK3␤ activity in EPC with viral vectors to alter their vasculogenic properties in vivo.

EXPERIMENTAL PROCEDURES
Isolation and Culture of EPC from Human Peripheral Blood-All human projects in this study were approved by the institutional review board of the Seoul National University Hospital. Two kinds of EPC with different biological characteristics, spindle-shaped "early EPC" and cobblestone-shaped "late EPC," were cultured as previously described (16). In brief, peripheral blood mononuclear cells were isolated from human volunteers and resuspended in the EGM-2 BulletKit system (Clonetics) consisting of endothelial basal medium, 5% fetal bovine serum, human epidermal growth factor, VEGF, human basic fibroblast growth factor, insulin-like growth factor 1, ascorbic acid, and heparin. 1 ϫ 10 7 mononuclear cells/well were seeded on a 2% gelatin (Sigma)-coated 6-well plate and incubated in 5% CO 2 incubator at 37°C. Under daily observation, the first medium change was done about 6 days after plating. Thereafter, media were changed every 3 days.
Gene Transfer to Cell-Replication-defective adenoviral vector expressing catalytically inactive GSK3␤ (GSK3␤-KM), where lysine residues at positions 85 and 86 were mutated to methionine and alanine, respectively, and control adenoviral vector expressing green fluorescent protein (GFP) was used for gene delivery (36,37). EPC cultured for 10 days were transduced with multiplicity of infection 50 of adenovirus-GSK3␤-KM (AdGSK-KM) and control adenovirus-GFP (AdGFP) in EGM-2 medium with 2% fetal bovine serum for 12 h. Effective gene transfer was confirmed by more than 90% of green fluorescence-positive control EPC transduced with AdGFP, and cells were assayed or harvested.
Western Blot Analysis-Western blot analysis was performed to confirm that the transduction of AdGSK-KM into EPC could lead to the sufficient expression of inactive phosphorylated GSK3␤ (36,37). Early EPC and late EPC were transduced with multiplicity of infection 50 of AdGFP or AdGSK-KM for 12 h and washed in phosphate-buffered saline (PBS) and harvested by scraping in lysis buffer. After determination of concentration with a protein assay kit (Pierce), 30 g of protein was separated by SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Millipore Corp., Bedford, MA). The membrane was blocked with PBS containing 0.3% Tween 20 (T-PBS) and 3% dry milk and incubated with primary antibody overnight at 4°C. Then membrane was washed three times with T-PBS and reblocked and incubated with secondary antibody for 1 h at room temperature. ECL Plus (Amersham Biosciences) was used for detection. To reprobe the membrane, it was treated with Restore Western blot stripping buffer (Pierce). The primary antibodies used with anti-phospho-GSK3␤ antibody (1:750 dilution; Cell Signaling Technology), anti-total GSK3 antibody (1:750 dilution; Santa Cruz Biotechnology), anti-␣-tubulin antibody (1:4000 dilution; Oncogene), and anti-phospho-␤-catenin (1: 500 dilution; Cell Signaling Technology). The secondary antibodies were anti-rabbit IgG/horseradish peroxidase conjugate (1:2500 dilution; Promega). Densitometric quantitative analysis of the immunoblot result was performed using TINA 2.01 (Raytest).
Analysis of ␤-Catenin Nuclear Translocation-The effect of GSK3␤ inhibition in canonical Wnt pathway was investigated by examining the subcellular localization of ␤-catenin. Early EPC and late EPC were washed in PBS, fixed with methanol for 30 min at Ϫ20°C, and washed three times with ice-chilled T-PBS. Anti-␤-catenin antibody and antimouse IgG/phycoerythrin were used as primary and secondary antibodies, respectively (1:1000, 1:2500 dilution; BD Biosciences). Nuclei were considered ␤-catenin-positive if staining intensity was equal to or greater than that seen in the cytoplasm (38).
Measurement of Cytokine Secreted from Cells-For 3 days, 1 ϫ 10 6 cells were incubated with growth factor-free EBM-2 medium, and then the supernatant was harvested. Cytokine concentration was measured with an enzyme-linked immunosorbent assay kit (VEGF, Quantikine; IL-8, R & D Systems) (16).
Cell Proliferation and Survival Assay-Effect of GSK3␤-KM gene transfer to EPC survival and proliferation was assessed by WST-1 assay. Briefly, 1 ϫ 10 4 cells were seeded to each well of 96-well plates and incubated for 48 h. For evaluation of cell survival and cell proliferation, 200 l of serum-free EBM-2 culture medium and EGM-2 with 5% fetal bovine serum were used, respectively. Then 20 l of cell proliferation assay reagent WST-1 (Roche Applied Science) was added to each well and incubated for 4 h. Absorbance of 440 nm was measured by an enzyme-linked immunosorbent assay reader.
Matrigel Network Formation Assay-A Matrigel network formation assay was performed to assess the effect of the GSK3␤-KM gene on the ability of EPC to integrate into vascular structures. Both early EPC and late EPC were transduced with AdGFP or AdGSK-KM. Four-chamber slides (Nalgene) were coated with Matrigel (Becton Dickinson Labware), and 2 ϫ 10 4 cells in 500 l of EGM-2 medium were added to each chamber. After 24 h, the endothelial cell network formed by EPC was evaluated from images of four representative fields per chamber by computer-assisted analysis using Image-Pro Plus (Media Cybernetics).
Hind Limb Ischemia Model and Transplantation of Genetically Modified EPC-All procedures were approved by the Experimental Animal Committee of Clinical Research Institute, Seoul National University Hospital. Female athymic nude mice (n ϭ 61; Jackson Laboratory), 7-8 weeks old and weighing 17-20 g, were anesthetized with intraperitoneal injection of a combination of anesthetics (ketamine (50 mg/kg) and xylazine (20 mg/kg); Bayer Korea), and an ipsilateral femoral artery was surgically removed as previously described (5,39). EPC transduced with AdGFP or AdGSK-KM were washed gently five times with PBS, harvested by brief trypsinization and forceful pipetting, and resuspended with growth factor-free EBM-2 medium. Because a too high dose of EPC may salvage most limbs and conceal the effect of AdGSK-KM on the angiogenic potential of EPC (5,40), the dose of injected cells was empirically determined to be 2 ϫ 10 5 cells/mouse through pilot experiments. As an administrative route for EPC, we chose direct cell injection into systemic arterial circulation to maximize delivery into ischemic skeletal muscle and reduce potential retainment by lung, liver, or spleen. One day after surgery, 2 ϫ 10 5 of EPC in 100 l of EBM-2 medium were injected into the left ventricular cavity of mice (5). A syringe needle of 30 gauge was advanced via percutaneous subxyphoid access, and brief blood regurgitation was confirmed before and after cell injection to ensure intracardiac space injection. Mice received 1) AdGFP-early EPC (n ϭ 17), 2) AdGSK-KM-early EPC (n ϭ 16), or 3) AdGFP-late EPC (n ϭ 18), or 4) AdGSK-KM-late EPC (n ϭ 10).
Evaluation of in Vivo Angiogenesis Effect of EPC-A laser Doppler perfusion imager (Moor Instruments), which maps tissue blood flow by the shift in the laser light frequency, was used for serial noninvasive physiological evaluation of neovascularization (5). Each mouse was followed by serial recording of surface blood flow of hind limb on days 0, 3, 7, 14, and 21. After laser Doppler perfusion imager scanning, the digital color-coded images were analyzed to quantify blood flow of the area from the knee joint to toe, and mean values of perfusion were calculated. To avoid data variations due to ambient light and temperature, hind limb perfusion was expressed as the ratio of ischemic to nonischemic limb. Photographs of the limb were also recorded and visually analyzed as "limb salvage" (completely normal status without a sign of ischemia), "foot necrosis" (necrosis of toe or below knee), or "limb loss" (necrosis or loss of tissue above knee).
Tissue capillary density was determined as a histological evaluation of neovascularization. At day 21, mice were euthanized by an overdose of anesthetics, and the lower calf muscles of the ischemic limb were harvested and snap frozen in liquid nitrogen-chilled isopentane. Tissues were embedded in OCT compound (Sakura Finetek), and multiple slices were prepared in 10-m thickness. Slices were fixed with 1% paraformaldehyde for 5 min, washed briefly with PBS, and stained with TRITC-conjugated murine EC-specific Bandeiraea simplicifolia lectin 1 (Sigma) and fluorescein isothiocyanate-conjugated human EC-specific Ulex europaeus lectin 1 (Sigma), each diluted 10 g/ml in PBS for 30 min at room temperature. Five random fields were selected from each slice, and fluorescent vascular ECs were counted to assess density of capillary composed of murine and/or human cell.
Statistical Analysis-All data are presented as mean Ϯ S.E. Statistical significance was evaluated by means of Student's t test or analysis of variance. The incidence of limb salvage was evaluated by 2 . Provability values of two-tailed p Ͻ 0.05 were considered to denote statistical significance. with GSK3␤-KM on the EPC survival and proliferation was evaluated using WST-1 assay in serum-free or growth factor-rich conditions, respectively. In serum-free conditions, the survival of EPC was enhanced when transduced with AdGSK-KM compared with AdGFP (early EPC, 78.1 Ϯ 23.6%, p Ͻ 0.001; late EPC, 66.7 Ϯ 25.5%, p Ͻ 0.001, respectively). In complete media with serum and growth factors, the proliferation of early EPC was also enhanced when transduced with AdGSK-KM compared with AdGFP (35.7 Ϯ 19.0% increase, p ϭ 0.003), but the proliferation of EPC was not significantly enhanced (p Ͼ 0.05). *, p Ͻ 0.05.

Modulation of GSK3␤ Signaling Pathway by Dominant Negative GSK3␤ Transduction into EPC-Early and late EPC cultures were transduced with adenoviral vectors expressing GFP (AdGFP) or a catalytically inactive form of GSK3␤ (AdGSK-KM)
. Under the culture conditions, greater than 90% of the cells expressed GFP reporter gene (Fig. 1A). Western blot analysis revealed that the transduction of AdGSK-KM into EPC could lead to increased expression of inactive phosphorylated GSK3␤. Both early EPC and late EPC transduced with AdGSK-KM showed a marked increase of total GSK protein (early EPC, 2.5-fold; late EPC, 2.6-fold) and phosphorylated GSK3␤ (early EPC, 2.2-fold; late EPC, 7.5-fold) compared with control AdGFP-transduced EPC. Inactivated GSK3␤ signaling pathway in EPC transduced with AdGSK-KM was identified by the decrease of phosphorylated ␤-catenin (early EPC, 0.4-fold; late EPC, 0.7-fold) (Fig. 1, B and C).
To investigate further the modulation of GSK3␤ signaling pathway by AdGSK-KM from biochemical perspective, nuclear translocation of ␤-catenin, a marker for Wnt pathway activation, was investigated at the subcellular level by direct ␤-catenin immunofluorescence staining. EPC transduced with AdGSK-KM showed profound nuclear translocation of ␤-catenin, and it was more significant in early EPC than late EPC (89.5 Ϯ 5.2 versus 61.3 Ϯ 9.1%, p Ͻ 0.05). EPC transduced with control AdGFP showed no nuclear ␤-catenin (Fig. 2A).
The effect of GSK3␤ inactivation on the secretion of angiogenic cytokines from EPC, which is regarded as one of the major angiogenic mechanisms of EPC, was also investigated (10,16). The concentrations of VEGF and IL-8, which is under the regulation of the GSK3␤ signaling pathway (41,42), were significantly higher in the supernatant of early EPC than in late EPC. In early EPC, transduction of AdGSK-KM led to significantly increased secretion of VEGF and IL-8 (VEGF, 132.3 Ϯ 9.3 versus 247.5 Ϯ 26.1 pg/10 5 cells, p ϭ 0.002; IL-8, 401.7 Ϯ 21.6 versus 516.7 Ϯ 11.7 pg/10 6 cells, p ϭ 0.001). In late EPC also, transduction of AdGSK-KM led to a significant increase of IL-8 secretion (265.7 Ϯ 23.4 versus 334.4 Ϯ 10.1 pg/10 6 cells, p ϭ 0.01), although it did not affect the amount of VEGF that was barely secreted from late EPC (Fig. 2B).
Effect of AdGSK-KM on the in Vitro Angiogenesis Model-A Matrigel network formation assay was performed to investigate the ability of EPC to form endothelial cell networks reminiscent of tubules. Both AdGFP-and AdGSK-KM-transduced early EPC in Matrigel showed little evidence of endothelial network formation. In contrast, late EPC formed an endothelial network reminiscent of tubules, and the network formation was more evident when late EPC was transduced with AdGSK-KM than with control AdGFP (network length: 2.04 Ϯ 0.15 versus 1.11 Ϯ 0.72 mm, respectively, p Ͻ 0.05; network circle number: 7.3 Ϯ 3.1 versus 2.4 Ϯ 1.6/mm 2 , respectively, p Ͻ 0.05) (Fig. 3A).

Enhanced Angiogenesis by Inactivation of GSK3␤
Effect of AdGSK-KM on EPC Proliferation and Survival-Transduction of GSK3␤-KM significantly increased survival of both early and late EPC exposed to serum-free condition (early EPC, 1.8-fold; late EPC, 1.7-fold). Transduction of GSK3␤-KM also increased proliferation of early EPC (1.4-fold) cultured in medium with growth factors but did not affect the proliferation of late EPC cultures in the same condition (Fig. 3B).
Effect of AdGSK-KM on the in Vivo Angiogenesis Model-The in vivo angiogenic effect of GSK3␤-KM-transduced EPC in a murine model of hind limb ischemia was evaluated. Serial evaluation of hind limb blood perfusion by laser Doppler perfusion imager revealed significant improvement in vascular perfusion at 21 days using either early or late EPC transduced with GSK3␤-KM compared with control EPC transduced with GFP or media (Fig. 4, A and B). Capillary densities were also evaluated in tissue sections retrieved at day 21 from lower calf muscle of ischemic hind limb. We confirmed that the endothe-lial morphology of histological murine skeletal muscle specimen is appropriate, by various staining methods including the classic histological method (hematoxylin-eosin), using biochemical characteristics (alkaline phosphatase), and the immunologic method using surface marker (CD31) (Fig. 5A). Higher capillary densities were observed in mice that received either early or late EPC transduced with GSK3␤-KM, corroborating the vascular perfusion measured by laser Doppler perfusion imager (early EPC, 2.0-fold increase; late EPC, 1.6-fold increase) (Fig. 5B). The incorporation of human EPC into the skeletal capillary system was assessed using immunofluorescence staining with human-specific fluorescein isothiocyanateconjugated U. europaeus lectin 1 at day 7, the time at which the proliferation of endothelial cell is maximal (39). A greater number of human-derived EPC were detected in mice given early or late EPC transduced with GSK3␤-KM (early EPC, 5.4-fold increase; late EPC, 2.2-fold increase) (Fig. 5C). FIG. 5. Administration of EPC transduced with AdGSK-KM increased capillary density of ischemic skeletal muscle. A, the endothelial morphology of a histological murine skeletal muscle specimen was identified using various staining of transverse-sectioned muscle with the classic histological method (hematoxylin-eosin), using biochemical characteristics (alkaline phosphatase), and the immunologic method using surface marker (CD31). All images showed similar morphological characteristics, dots scattered on a background. Staining using mouse endothelial cellspecific B. simplicifolia lectin 1 was performed in a tilted section of skeletal muscle, showing the longitudinal aspect of vascular endothelial cells. Incorporation of human EPC, which was marked with human endothelial cell-specific U. europaeus lectin 1 (UEA-1), into sites of neovascularization, was identified in the merged image. B, representative photographs of histological evaluation of neovascularization in ischemic tissue at day 21. Capillary densities of mice given EPC transduced with AdGSK-KM were markedly increased compared with control Some athymic nude mice develop extensive necrosis of ischemic hind limb and eventually lose their limbs by autoamputation (5,39). Serial examinations of mice revealed that the administration of EPC that were genetically modified with AdGSK-KM resulted in a significantly lower rate of limb loss and higher rate of limb salvage compared with control EPC (Fig. 6, A and B). More than half of the mice receiving AdGFPearly EPC lost limbs (58.8%, 10 of 17), whereas mice that received early EPC genetically modified with AdGSK-KM lost limbs only in 18.8% of cases (3 of 16) (p Ͻ 0.05). More limb salvage was also evident in the AdGSK-KM-early EPC group (50.0% (8 of 16)) than in the AdGFP-early EPC group (17.6% (3 of 17)), although this was statistically marginal (p ϭ 0.07). Analysis of limb status from mice groups that received late EPC also showed a lower rate of limb loss (0% (0 of 10) versus 50% (9 of 18), p Ͻ 0.01) and a higher rate of complete limb salvage (70% (7 of 10) versus 27.8% (5 of 18), p Ͻ 0.05) in the AdGSK-KM-late EPC group compared with the AdGFP-late EPC group. DISCUSSION In this study, EPC were isolated from human peripheral blood and transduced with catalytically inactive GSK3␤ (GSK3␤-KM). Collectively, the genetically modified EPC displayed enhanced proliferation, survival, and differentiation in vitro. These cells also displayed augmented vasculogenic potential in vivo, suggesting that GSK3␤ is a regulator of angiogenic function of EPC.
In our previous study, we demonstrated that two biologically distinct types of EPC emerge sequentially from a single source of adult peripheral blood, although the in vivo vasculogenic potential was not appreciably different between these two populations (16). In the current study, we observed that transduction of GSK3␤-KM into early and late EPC significantly enhance the in vivo vasculogenic potential of the early and late EPC to comparable degrees. The modulation of GSK3␤ signaling pathway by transduction of catalytically inactive GSK3␤ gene, GSK3␤-KM, was proved at the subcellular level by the increase of phosphorylated GSK3␤ and nuclear translocation of ␤-catenin. However, we could observe differences in the effect of GSK3␤-KM on the in vitro angiogenic function of the two EPC populations. These in vitro results might shed light on the possible mechanisms for the increased in vivo vasculogenic potential of GSK3␤-KM-transduced EPC.
In our study, the survival of both early and late EPC was improved by transduction of GSK3␤-KM. This can be speculated from the prosurvival effect of GSK3␤-KM on mature endothelial cells (36), because EPC is a vascular endotheliumcommitted stem cell (1,7). However, transduction of GSK3␤-KM enhanced the proliferation of early EPC, whereas the proliferation of late EPC was not significantly influenced. The weak effect of GSK3␤-KM on late EPC can be explained by the relatively smaller amount of ␤-catenin nuclear translocation in late EPC than early EPC, which was shown in immunoblot analysis and immunofluorescence staining. In addition, the proliferation of late EPC may not be significantly influenced by transduction of GSK3␤-KM, because late EPC exhibits a high exponential growth rate innately, contrary to the limited proliferative capacity of early EPC (14,16,17). The high growth rate might also lead to the less nuclear translocation of ␤-catenin observed in late EPC, because ␤-catenin subcellular localization can be affected by cell density (43).
Early EPC contributes to neovascularization mainly by secreting the angiogenic cytokines that help recruit resident mature vascular endothelial cells and induce their proliferation and survival, whereas late EPC enhances neovascularization by providing a sufficient number of endothelial cells based on their high proliferation potency (10,16). Transduction of GSK3␤-KM enhanced secretion of VEGF in early EPC. GSK3␤ phosphorylates the HIF-1␣ oxygen-dependent degradation domain. Therefore, the inactivation of GSK3␤ by GSK3␤-KM increases VEGF probably through HIF-1␣ accumulation, which increases VEGF transcription (41). Transduction of GSK3␤-KM also enhanced secretion of IL-8 in early and late EPC. This could be explained by up-regulation of IL-8 by ␤-catenin, which is downstream of GSK3␤ (42). Conversely, GSK3␤-KM promoted the differentiation of late EPC, but early EPC did not exhibit this property. These data suggest that early and late EPC may be subject to different but overlapping regulatory controls and that the effect of GSK3␤ signaling on phenotype may be influenced by the status of EPC differentiation.

Enhanced Angiogenesis by Inactivation of GSK3␤
single patient (40). To overcome this limitation, EPC could either be extensively amplified in vitro or be genetically modified to improve their vasculogenic properties. Previous studies have genetically modified EPC with VEGF or telomerase reverse transcriptase genes to enhance their vasculogenic function (40,44). Here, we targeted the GSK3␤ signaling step. Under conditions of these assays, control EPC salvaged ϳ20 -30% of the ischemic limbs. However, when EPC were modified with GSK3␤-KM, salvage improved to ϳ50 -70%. Furthermore, in these experiments, we used smaller numbers of EPC than in our previous study (16), suggesting further that genetic modification with GSK3␤-KM can overcome the limitation of EPC number.
In conclusion, our data show that vasculogenic potentials of both spindle-shaped early EPC and cobblestone-shaped late EPC can be augmented by transduction of catalytically inactive GSK3␤ gene. Diminished GSK3␤ signaling was shown to increase the proliferative potential of early EPC, promote the differentiation potential of late EPC, and exert an antiapoptotic action on both. In vivo, GSK3␤-KM-transduced EPC exhibited a more potent vasculogenic response than control EPC. These data show that GSK3␤ signaling is an important determinant of EPC behavior, and they support the hypothesis that the administration of biologically modified EPC may be a strategy for increasing the potency of EPC for therapeutic vasculogenesis.