OUP user menu

★ editor's choice ★

Inhibition of endothelial progenitor cell glycogen synthase kinase-3β results in attenuated neointima formation and enhanced re-endothelialization after arterial injury

Benjamin Hibbert , Xiaoli Ma , Ali Pourdjabbar , Erik Holm , Katey Rayner , Yong-Xiang Chen , Jiangfeng Sun , Lionel Filion , Edward R. O'Brien
DOI: http://dx.doi.org/10.1093/cvr/cvp156 16-23 First published online: 19 May 2009


Aims Endothelial progenitor cells (EPCs) are circulating pluripotent vascular cells capable of enhancing re-endothelialization and diminishing neointima formation following arterial injury. Glycogen synthase kinase (GSK)-3β is a protein kinase that has been implicated in the regulation of progenitor cell biology. We hypothesized that EPC abundance and function could be enhanced with the use of an inhibitor of GSK-3β (GSKi), thereby resulting in improved arterial repair.

Methods and results Human EPCs were expanded ex vivo, treated with a specific GSKi, and then assessed for both yield and functional characteristics by in vitro assays for adherence, apoptosis, and survival. In vivo functionality of treated human EPCs was assessed in immune-tolerant mice subjected to femoral artery wire injury. Re-endothelialization was assessed at 72 h and neointima formation at 7 and 14 days following injury. GSKi treatment resulted in an improvement in the yield of EPCs and a reduction in apoptosis in cells derived from both healthy controls and patients with coronary artery disease. Treatment also increased vascular endothelial growth factor secretion, up-regulated expression of mRNA for the α-4 integrin subunit, and improved adhesion, an effect which could be abrogated with an α-4 integrin blocking antibody. EPCs without or with ex vivo GSKi treatment enhanced re-endothelialization 72 h following injury as well as reduced neointima formation at 7 days (e.g. endothelial coverage: 7.2 ± 1.7% vs. 70.7 ± 5.8% vs. 87.2 ± 4.1%; intima to media ratios: 1.05 ± 0.19 vs. 0.39 ± 0.08 vs. 0.14 ± 0.02; P < 0.05 for all comparisons), an effect that was persistent at 14 days.

Conclusion GSKi improves the functional profile of EPCs and is associated with improved re-endothelialization and reduced neointima formation following injury.

  • Endothelial progenitor cells
  • Glycogen synthase kinase
  • Neointima
  • Re-endothelialization
  • Vascular endothelial growth factor
  • Integrin

1. Introduction

Injury to the vascular endothelium is thought to be the initiating process in development of atherosclerotic disease1 and has also been implicated in the pathogenesis of other vascular disorders such as transplant vasculopathy and neointimal hyperplasia following balloon angioplasty.2 Data from observational clinical studies have noted an association between circulating numbers of endothelial progenitor cells (EPCs) and both the presence of coronary artery disease (CAD)3 and the likelihood of future cardiovascular events.4 These findings, as well as evidence in both animal5 and clinical studies6 of improved outcomes through delivery of isolated populations of progenitor cells have reinforced the importance of re-establishing endothelial integrity following injury as a key step in the prevention of vascular disease.

Circulating progenitor cell populations—specifically EPCs—play a key role in arterial repair following injury.7 For example, statins,8 granulocyte colony stimulating factor,9 and direct transplantation of ex vivo cultured EPCs5 improve cell-mediated repair and ultimately reduce neointima formation in animal models of arterial injury. Moreover, stents that effectively ‘capture’ EPCs by means of CD34 antibodies10 loaded on stent struts demonstrate enhanced re-endothelialization and safety in preliminary human trials.11 While transplantation of a patient's progenitor cells is a conceptually attractive strategy for enhancing vessel re-endothelialization, it is important to note that EPCs from patients with CAD are typically low in abundance and show attenuated functional properties compared with cells from healthy controls.12 Ultimately, pharmacological regulation of these endogenous cell populations represents the most likely manner in which EPC biology will be exploited for clinical benefit.

Glycogen synthase kinase (GSK)-3β is a serine/threonine protein kinase known to negatively regulate Wnt signalling through phosphorylation of the nuclear transcription factor β-catenin and hence, direct its degradation.13 Recently, GSK inhibition has been shown to stimulate progenitor and haematopoietic stem cell capacity in vivo through modulation of Wnt, Hedgehog and Notch signalling.14 Wnt signalling is also known to play a key role in the mobilization of vascular progenitor cells and enhancement of neovascularization, as Aicher et al.15 demonstrated using the Wnt signalling antagonist Dickkopf-1. However, to date little is known about the effects of GSK signalling on EPC function in CAD patients or intima development following arterial injury. The purpose of the current study was to determine whether a specific inhibitor of GSK (GSKi) could augment the in vitro yield and functions of human EPCs as well as improve arterial repair in vivo.

2. Methods

2.1 EPC culture and cell labelling

Human EPCs derived from normal controls and patients with established CAD were cultured in the usual manner.1618 Patients with CAD had to have established disease as defined by previous MI, flow limiting stenoses requiring stent insertion or surgical revascularization. All patients were on medical therapy at time of EPC collection including maximally titrated statin, aspirin, beta-blocker, and ACE-Inhibitor or angiotensin receptor blocker therapy. Blood was collected by venipuncture and anticoagulated with EDTA. Peripheral blood mononuclear cells (5 × 106) isolated by ficoll centrifugation were cultured in EGM-2 media (Cambrex) and plated on fibronectin-coated plates. Inhibition of GSK was achieved using a specific GSK inhibitor N-(4-methoxybenzyl)-N′-(5-nitro-1,3-thiazol-2-yl) urea (referred to as GSKi, Calbiochem) or LiCl (Sigma) at the indicated doses. Confirmation of effective GSK inhibition was achieved by western blot. Briefly, nuclear protein was isolated by differential centrifugation. β-Catenin nuclear levels were assessed using an anti-β-catenin (Cell Signaling Technologies) antibody diluted 1:1000. For complete protocol, see Supplementary material online at cardiovascres.oxfordjournals.org.

Adherent cells were maintained for 7 days prior to enumeration. For the task of enumeration, EPCs were defined as cells dually positive for AcLDL uptake and ulex europeus agglutinin I (UEAI) binding. DiI-AcLDL (2.5 µg/ml, Molecular Probes) was incubated with cultured EPCs for 1 h in a cell incubator. Subsequently, cells were washed and fixed with Cytofix Buffer (BD) and incubated with FITC-UEAI (5 µg/mL, Sigma) for 30 min. Plates of cells were again washed and incubated with a DAPI nuclear counterstain before a coverslip was applied to the well and double positive cells were counted in six random high-power fields (×200 magnification).

2.2 EPC survival, apoptosis, VEGF secretion, and adherence

Seven day old EPCs were used for all experiments unless otherwise indicated. For the cell survival assay, six high-power fields were enumerated for each individual. Subsequently, cells were washed and the media changed every 4 days after which cells were again enumerated. Data are expressed as a percentage of initial cells present on day 7. For apoptosis studies, cells from patients and healthy controls were cultured then lifted and recoated at a density of 2 × 106 mature EPCs per well. The cells were then allowed to incubate for an additional 4 days prior to being lifted with EDTA supplementation of the media and gentle agitation. Cells were pelleted and resuspended in hank's balanced salt solution (HBSS). Prior to analysis by flow cytometry 10 µL of propidium iodide was added to the cell suspension. EPCs were identified by uptake of AcLDL-alexa488 (Invitrogen) and UEA-1 FITC labelling in separate experiments. Inability to exclude propidium iodide identified apoptotic cells. A total of 10 000 events were analysed and the data expressed as a percentage of total EPCs being apoptotic. All experiments were conducted on a Beckman Coulter Cytomics FC 500 cytometer.

The secretion of vascular endothelial growth factor (VEGF) by EPCs was measured using a VEGF ELISA kit (R&D Systems) using the manufacturer's provided protocol. Briefly, EPCs were plated in equal numbers and incubated in VEGF-free EGM-2 for 24 h. Subsequently, 200 µL of the culture supernatant was added to a 96-well plate coated with anti-human VEGF antibody. After 2 h of incubation, the conjugated secondary antibody was added and allowed to incubate for another 2 h. Substrate solution was added and the wells interrogated for absorption at 450 nm using a Bio-Rad micro-plate reader.

For adherence studies, EPCs were detached from their fibronectin coated plates by incubation with 0.5 mmol EDTA. Cells were pelleted, resuspended in EGM-2 and enumerated. Subsequently, 100 000 EPCs were plated in 24-well fibronectin coated plates and incubated for 30 min before being washed three times with HBSS and labelling the adherent cells labelled so that six random high-power fields could be enumerated. Blocking experiments were performed using a specific antibody to the α-4 integrin subunit (MAB16983, Millipore) at a concentration of 10 μg/mL and allowing cells to incubate for 2 h with the antibody.

Q-PCR was performed on the Light Cycler Q-PCR System (Roche) and the data were analysed using the accompanying software package. Total RNA was isolated from EPCs using Trizol (Invitrogen) and RT performed using standard techniques. Amplicons were cloned into the pGEM-T vector, sequenced, and isolated using the PhasePrep BAC DNA kit (Sigma). The plasmids were then linearized, purified, and diluted to generate standard curves for Q-PCR analysis. Primers and probes were designed using the PrimerQuest software and are available in Supplemental materials online at cardiovascres.oxfordjournals.org. PCR was performed with an annealing temperature of 56° for all primer combinations. The Q-PCR reagents utilized were the QuantiTect SYBR PCR system and the QuantiTect Probe PCR system (Qiagen). All primer combinations were confirmed to have a single amplicon on agarose gel and SYBR green PCR was utilized for quantification of alpha integrins 1–5 and GAPDH. Confirmation of α-4 integrin mRNA upregulation was done using a probe-specific Q-PCR technique. The α-4fwd and α-4rev primers were used in conjunction with a 5′ 6-FAM labelled and 3′ TAMRA modified α-4probe. For these experiments, Qiagen QuantiTect Probe PCR system was utilized.

2.3 CD-1 nude mouse femoral artery wire injury model

CD-1 nude mice were acquired from Charles River Laboratories and acclimatized in our facilities for 2–6 weeks prior to surgeries. To assess the ability of ex vivo treated EPCs to enhance arterial repair, the femoral artery of the mice was injured by insertion of a 32 gauge blunt needle (Strategic Applications Inc.) to induce neointima formation as previously described.19 Following arterial incision, the needle was introduced into the lumen, advanced proximally, and passed five times in order to denude endothelium and mechanically stretch the vessel. Subsequently, 5 × 105 of control-EPCs (C-EPCs), GSKi-treated EPCs (G-EPCs) or vehicle (n = 6 per group) were injected and the artery ligated proximal to the incision site. Arterial injuries were performed proximal to major branches and the femoral artery was ligated distal to these branches. Hence, blood flow was maintained in the main arterial segment due to run-off via the intact branch vessels. None of the animals exhibited signs of ischaemia to their hind limbs and all animals had full use of the limb immediately post-surgery. General anaesthesia was achieved during anaesthesia using isoflurane.

For assessment of re-endothelialization, mice were sacrificed at 24 (n = 2), 48 (n = 2), and 72 h (n = 6) post-injury. Mice sacrificed at 24 and 48 h had enface preparation and staining with DAPI and were examined by fluorescent microscopy. To permit quantification of re-endothelialization, mice were perfused with 0.5% Evans blue, then perfusion-fixed with formaldehyde until clear of dye. For assessment of neointimal formation, mice were sacrificed at 7 and 14 days (n = 6 per time point). Femoral arteries were fixed in buffered formalin then dehydrated with ethanol. Arteries were mounted in paraffin blocks and sectioned in 5 µm sections. Sections were haematoxylin and eosin-stained and analysis performed using a computer-assisted digital imaging system (Image-Pro Plus, Media Cybernetics).

2.4 Ethics and statistics

Animal procedures followed the University of Ottawa Animal Care Committee and the Canadian Council on Animal Care guidelines. All protocols involving human donors were approved by the Ottawa Heart Institute Research Ethics Committee. These studies conform with the Declaration of Helsinki for the use of human tissue and animal experiments conformed with the Guide for the Care and Use of Laboratory Animals. For statistical procedures, a P-value less than 0.05 was considered significant. Analyses were performed using the Sigmastat 3.5 package. Two-way comparisons were performed with a Student's t-test and multiple comparisons using one-way ANOVAs with Holm–Sidak post hoc test. Data are expressed as mean ± standard error of the mean.

3. Results

3.1 GSK-3β inhibition improves EPC yield

Given the paucity of EPCs found in patients with CAD, we first tested the effects of GSKi treatment on EPC yield in vitro after 7 days. Outgrowth of EPCs was reproducibly achieved in all subjects and the phenotype confirmed using the commonly accepted parameters of AcLDL uptake and UEA-1 binding (Figure 1A and B).17 Media was supplemented with either 104 nM (IC50, ×1) or 208 nM (×2) GSKi. In order to confirm the specificity of the GSKi, nuclear β-catenin levels were assessed by western blotting. Under control (untreated) conditions GSK-3β phosphorylates β-catenin thereby targeting it for degradation. However, with GSKi treatment, the degradation of β-catenin is interrupted thereby resulting increased nuclear levels of the protein (Figure 1C). When compared with control treatment media supplemented with GSKi at both ×1 and ×2 concentrations led to a progressive increase in the yield of EPCs derived from healthy controls (48.5 ± 4.6 vs. 68.6 ± 6.6 vs. 75.7 ± 6.8, respectively) and more than quadrupled the yield of CAD EPCs (14.0 ± 8.0 vs. 41.7 ± 17.0 vs. 58.5 ± 15.8; Figure 1D). To confirm that the observed effects on EPC yield were specific to GSK-3β inhibition, EPCs were cultured with varying doses of a second GSK-3β inhibitor, lithium chloride (LiCl; Supplementary material online, Figure S1). Similar to GSKi, LiCl augmented the yield of EPCs derived from both healthy controls and CAD patients in a dose-dependent manner. Combination treatment with GSKi and LiCl did not result in further improvement in EPC yield (Figure 1E), thereby suggesting that the increase in EPC number was achieved through a specific inhibitory effect on the GSK-3β isoform that was maximal with one or the other treatment and not altered in a synergistic manner due to an off-target effect of one of the drugs.

Figure 1

GSK-3β inhibitor improves attenuated levels of EPCs from patients with CAD. (A and B) High and low power magnification of EPCs at 7 days labelled with DAPI (blue), AcLDL-DiI (red), UEA-1-FITC (green), and merged image used for enumeration. (C) Western blot of nuclear fraction from EPCs cultured in control (C) media and media supplemented with GSKi (×2) showing increased levels of β-catenin. (D) Comparison of EPC yields in both healthy controls and patients with CAD when treated with control (C) media, the 104 nm (×1) and 208 nm GSKi (×2), n = 6. (E) Dual treatment of EPCs with both GSKi and LiCl does not synergistically improve EPC yields, n = 6. * or # denotes statistical significance, P < 0.05.

Given the known role of GSK in the regulation of apoptosis, we hypothesized that, in part, the improved EPC yields with GSKi may be linked to lower levels of apoptosis.20 To test this hypothesis, we cultured EPCs for 7 days, replated them at equal densities then treated them with either control EGM-2 media or media supplemented with ×2 GSKi. EPCs from CAD patients had a faster rate of attrition than EPCs from healthy control subjects (Figure 2A). GSKi treatment markedly improved EPC survival for both cell populations. In fact, EPCs from CAD patients survived to the same extent as EPCs from control subjects when treated with GSKi. At baseline, EPCs from CAD patients demonstrated a higher rate of apoptosis compared with EPCs from controls (Figure 2B). In both groups, ×2 GSKi caused a reduction in the percentage of apoptotic cells—a result which may explain the improved survival observed in GSKi-treated cells.

Figure 2

GSKi enhances EPC survival and adherence. (A) EPC survival is impaired in patients with CAD. Treatment with GSKi in both healthy controls and patients with CAD significantly improves long-term viability, n = 6. (B) GSKi decreases apoptosis in both EPCs derived from healthy controls and patients with CAD. (C) VEGF secretion by EPCs is enhanced by GSKi in a dose-dependant manner. (D) GSKi improves adhesive properties of EPCs derived from both healthy controls and CAD patients, n = 12. (E) GSKi treatment upregulates mRNA of the α-4 integrin isoform as measured by Q-PCR, n = 6. (F) Introduction of a specific α-4 integrin subunit blocking antibody (α-4 Ab) demonstrates the reversibility of improved EPC adhesion and implicates α-4 in EPC adhesion, n = 6. * or # represents significant differences P < 0.05.

3.2 GSK-3β inhibition improves the functional profile of EPCs

Clinical studies suggest that not only the abundance of EPCs but also the functional capacity of these cells is important in determining the in vivo biological effect.21 Hence, we tested the ability of GSKi to affect the secretion of a relevant endothelial growth factor, VEGF. Equal numbers of EPCs were incubated for 24 h in VEGF free media before VEGF levels were measured in the media by ELISA. There was a dose-dependent effect of GSKi on VEGF secretion by EPCs, with the ×2 dose of GSKi resulting in a more than ×6 increase over control conditions (Figure 2C).

Previously, our group and others reported that EPCs are incorporated into healing arteries post-injury (e.g. after stent implantation).16,22,23 As the steps that lead to progenitor cell homing and adhesion into the vessel wall are incompletely understood, we began by characterizing the adhesive characteristics of EPCs to fibronectin. Adherence to fibronectin was similar for control and CAD EPCs. However, treatment with GSKi resulted in an approximate four-fold increase cell adherence for both populations of cells (Figure 2D). Next, we determined whether GSKi treatment upregulated the expression of one or more of α-integrin subunits as this integrin family of cell surface receptors plays a key role in vascular cell adhesion. We cloned α-integrin subunits 1 through 5 as well as the GAPDH gene from human PBMCs and performed SYBR green Q-PCR on mRNA from control, GSKi ×1, and GSKi ×2 treated EPC samples. Levels of mRNA for the α-1, α-2, α-3, and α-5 integrin subunits did not change with GSKi treatment (data not shown). In contrast, the α-4 integrin subunit was upregulated with both GSKi concentrations (Figure 2E). To confirm that the observed increase in adhesion was mediated via α-4 regulation, we tested the ability of an α-4 integrin subunit blocking antibody to impair EPC adhesion in a fibronectin assay (Figure 2F). As observed in the previous experiments, addition of GSKi resulted in an increased number of adherent cells. However, this effect was completely abrogated by pretreatment of the cultured cells with the α-4 integrin subunit blocking antibody—thereby suggesting that the observed increase in EPC adhesion with GSKi treatment is mediated via upregulated expression of the α-4 integrin subunit.

3.3 GSK-3β inhibition enhances EPC-mediated arterial repair

Werner et al.5 first described the ability of ex vivo cultured EPCs to mediate arterial repair in vivo. We hypothesized that the enhanced ex vivo function achieved by GSK inhibition would translate into improved in vivo function. To test this hypothesis vehicle, C-EPCs or G-EPCs were systemically injected into immune compromised mice subjected to femoral artery injury. Notably, cells from CAD patients were used for these experiments, despite the fact that at baseline these cells are less abundant and functionally deficient compared with cells from healthy control subjects. Re-endothelialization was assessed by enface examination at 24, 48, and 72 h post-blunt needle injury. At 24 h post-injury, there was near complete disruption of the endothelium with only few cells overlying the media (Figure 3A). By 48 h post-injury re-endothelialization was evident, particularly in the cell treatment groups. Thus, a 72 h time point was selected for quantitative assessment of the re-endothelialization area of arteries perfused with Evan's blue solution (Figure 3B). Treatment with C-EPCs alone resulted in a 10-fold increase of re-endothelialization area, an effect that was further enhanced by EPCs pretreated with GSKi (e.g. vehicle: 7.2 ± 1.7%, C-EPCs: 70.7 ± 5.8%, and G-EPCs: 87.2 ± 4.1%; P < 0.05, Figure 3C). These findings suggest a marked enhancement of endothelial regeneration by EPC transplantation—a benefit which was further augmented through abrogation of GSK-3β/β-catenin signalling.

Figure 3

Rapid re-endothelialization of injured arteries is promoted by EPC transfusion and further enhanced by GSKi pretreatment. (A) Wire injury results in significant denudation of endothelium by 24 h exposing underlying smooth muscle cells (red arrows). EPC infusion results in nearly complete re-endothelialization by 48 h when compared with vehicle-treated arteries. (B) Intact and enface isolated femoral arteries. Mice were perfused with Evans blue prior to dissection. Intact endothelium excludes the blue dye. (C) Percentage of arterial re-endothelialization as assessed by Evans blue perfused arteries, n = 5 for vehicle and n = 6 for C-EPC and G-EPC groups. * or # denotes statistical significance, P < 0.05 or less.

To assess neointima formation in these mice, euthanasia was carried out 7 and 14 days after blunt injury (Figure 4A). When compared with sham (uninjured) arteries, all of the injured arteries in each groups was nearly three times larger due to the uniform (mechanical) dilatation that resulted from the intra-luminal passage of the blunt needle (Figure 4B). Assessment of intimal area was adjusted for media area and expressed as the intima to media ratio (I:M). At 7 days, the I:M was attenuated by C-EPCs and reduced to less than one-sixth of controls by infusion with G-EPCs (e.g. vehicle: 1.05 ± 0.19, C-EPCs: 0.39 ± 0.08, G-EPCs: 0.14 ± 0.02; P < 0.05 for all comparisons; Figure 4C). Intimal lesions showed a marked increase in cellularity in both groups of mice that received EPC infusions, but the intima of the vehicle-treated mice consisted primarily of extracellular matrix. The comparison of the I:Ms 14 days post-injury was similar to that performed on the data recorded at 7 days post-injury (e.g. vehicle: 1.20 ± 0.23, C-EPCs: 0.39 ± 0.07, and G-EPCs: 0.13 ± 0.03; P < 0.05 for all comparisons).

Figure 4

Mouse femoral artery wire injury model of neointima formation. (A) Wire injury results in significant intima formation at both 7 and 14 days. Representative 7 days cross-sections at low magnification and 14 days high-power cross-sections are shown with haematoxylin and eosin staining. Media (M), neointima (N), and lumen (L) are labelled for reference. Arrows indicate the internal elastic lamina. (B) Wire injury results in significant increase in artery volume compared with sham. No significant differences exist between the vehicle group, the control-EPCs (C-EPC) and the GSKi-treated EPC (G-EPC) groups, n = 6. (C) Transplantation of C-EPCs reduces both 7 and 14 days intima to media ratios (I:M) an effect further enhanced by pretreatment of EPCs with GSKi. No significant differences in intima/media ratios were seen between 7 and 14 days. * or # denotes statistical significance, P < 0.05 or less.

4. Discussion

Relative to healthy controls, patients with CAD have a paucity of circulating EPCs—cells that are now thought to be of central importance for the maintenance of vessel wall homeostasis.3,21 Hence, it is postulated that increasing EPC number and function may facilitate arterial repair—particularly the restoration of the endothelium. However, it is now evident that the functional properties of EPCs are of at least equal importance to their abundance. In this study, we describe a pharmacological strategy of inhibiting GSK that results in an expansion and functional enhancement of EPCs in vitro as well as amelioration of vascular healing in vivo. Originally, GSK was identified as a kinase that phosphorylates glycogen synthase; however, subsequent studies identified a broader range of substrates. Indeed, when GSK-3β is inactivated β-catenin levels rise. β-Catenin is the principal mediator of Wnts, a family of secreted glycoproteins that regulate a multitude of cell processes, including haematopoiesis and stem cell function.

In the current study, we begin by showing in vitro that GSKi enhances survival and reduces apoptosis of EPCs. As well, GSKi upregulates expression of the α-4 integrin subunit mRNA and enhances EPC adhesion in a manner that can be inhibited by means of an α-4 blocking antibody. In vivo, we note that GSKi-treated human EPCs attenuates neointima formation and enhances re-endothelialization of injured femoral arteries in immune-tolerant mice. Although the reduction in the I:M may be due to a direct intimal effect, it is more likely that enhanced re-endothelialization was instrumental in producing this result. Nonetheless, this study is the first to demonstrate the importance of GSK signalling in EPC-mediated arterial repair while also highlighting functional capacity of the cells as an important mediator of effect. Perhaps of greater importance is the fact that GSKi-mediated augmentation of EPC number and function is possible in feeble cells harvested from patients with CAD—a hurdle that previously had not been adequately addressed or overcome.

Rapid restoration of the endothelium and its role as a protective barrier and/or regulator of the local milieu (e.g. fibrinolysis) is of critical importance in vascular repair. Although the re-endothelialized area increased dramatically with infusion of just the C-EPCs (e.g. ∼10-fold compared with vehicle treatment), there was an important additional enhancement in re-endothelialization area with G-EPC infusion (e.g. from 70.7 to 87.2%). Indeed, in a remarkably short period (72 h), the arteries were almost completely re-endothelialized with GSKi-treated EPCs. Perhaps if we had lowered the number of cells infused to less than 5 × 105 or if the calibre of the arteries were larger (e.g. larger animal or human coronary artery), the degree of re-endothelialization with the C-EPCs would have been smaller—and there might have been a higher incremental difference in the re-endothelialized area with the G-EPCs. Clinically, those patients in acute need of repaving their endothelium (e.g. acute coronary syndrome patients) are also the same individuals with remarkably low EPC counts; hence, their need for even a modest augmentation in potent EPC number and/or function will be dear—particularly in the first 24–72 h after an acute ischaemic event. Finally, it is attractive to speculate as to how GSKi may be modulating re-endothelialization. Certainly, GSKi-mediated prevention of EPC apoptosis and/or increased VEGF secretion may play key roles in the regrowth of the endothelium in these arteries. Previously, Kim et al.24 demonstrated that transduction of a non-phosphorylatable, constitutively active mutant of GSK is protective against EC apoptosis and enhances endothelial cell migration towards VEGF. As well, Choi et al.25 demonstrated an increase in VEGF secretion in EPCs transfected with a GSK-3β dominant-negative mutant. Our data also suggests that GSKi facilitates EPC adhesion via an up-regulation of the α-4 integrin subunit. It is important to point out that the potentially important role of the α-4 integrin in enhancing arterial repair seen in the current study is somewhat at odds with our previous publication that demonstrated a reduction in stent neointimal formation, when the α-4 integrin subunit was blocked using a systemically administered antibody.26 However, the apparent difference in outcome for these two studies may be the result of the cells that were targeted. In the current study, the effect is limited to EPCs treated ex vivo before being infused in vivo, whereas in our previous study all cells, including monocytes and neutrophils were affected by the systemically delivered anti-α-4 blocking antibody.

Our study is not without potential limitations. First, although ex vivo GSKi treatment results in robust and durable reductions in lesion formation following arterial injury, the clinical feasibility of this strategy may be limited as transplantation of ex vivo manipulated cells can be labour intensive and impractical. Secondly, in the current study, it is impossible to definitively ascribe all of the beneficial vascular healing effects to the infusion of G-EPCs as it is conceivable that the observed increase in re-endothelialization might be due to paracrine effects of GSKi-treated EPCs. For example, could these infused cells adhere at or near the site of re-endothelialization where they then secrete pro-endothelial cytokines, such as VEGF, and thereby attract adjacent (endogenous) endothelial cells that then repopulate the endothelium? In fact, while robust re-endothelialization is observed with G-CSF treatments that mobilize EPCs, there is little evidence to support the direct contribution of (tagged) bone marrow-derived EPCs to endothelial reconstitution.27 Finally, inhibition of GSK is becoming an attractive target for vascular healing with evidence suggesting benefits in myocardial neovascularization following infarction in rats, in addition to our data implicating GSK signalling in arterial repair following injury.28 However, given the known roles of GSK signalling in both energy metabolism and oncogenic transformation, careful animal studies of any systemically based therapies must be undertaken before clinical studies can be embarked upon.

In summary, our studies demonstrate that GSK-3β signalling is a key modulator of progenitor cell-mediated vessel wall homeostasis following mechanical injury—a novel observation with significant potential for therapeutic development. Specifically, inhibition of this enzyme enhances the yield of EPCs in vitro and promotes EPC-mediated arterial healing in vivo. Further studies with systemic and device-based delivery systems targeting the GSK-3β signalling pathway to reduce intima formation are ongoing.

5. Supplementary material

Supplementary material is available at Cardiovascular Research online.

Conflicts of interest: none declared.


This work was supported by the Interventional Cardiology Group at the University of Ottawa Heart Institute. E.O.B. is supported by a grant (UOP #36383) that is jointly funded by the Canadian Institutes of Health Research (CIHR) and Medtronic, and holds a Research Chair (URC #57093) that is jointly funded by the CIHR and Medtronic.


View Abstract