Glycogen Synthase Kinase-3β Inhibition Augments Diabetic Endothelial Progenitor Cell Abundance and Functionality via Cathepsin B: A Novel Therapeutic Opportunity for Arterial Repair

  1. Edward R. O’Brien1,5
  1. 1Division of Cardiology, University of Ottawa Heart Institute, Ottawa, Ontario, Canada
  2. 2Department of Biochemistry, Microbiology and Immunology, University of Ottawa, Ottawa, Ontario, Canada
  3. 3Ottawa Hospital Research Institute, Sprott Stem Cell Centre and Regenerative Medicine Program, The Ottawa Hospital, Ottawa, Ontario, Canada
  4. 4Department of Cellular and Molecular Medicine, University of Ottawa, Ottawa, Ontario, Canada
  5. 5Division of Cardiology, Libin Cardiovascular Institute of Alberta, Calgary, Alberta, Canada
  1. Corresponding author: Edward R. O’Brien, ed.obrien{at}


Progenitor cell therapy is hindered in patients with diabetes mellitus (DM) due to cellular senescence. Glycogen synthase kinase-3β (GSK3β) activity is increased in DM, potentially exacerbating impaired cell-based therapies. Thus, we aimed to determine if and how GSK3β inhibitors (GSKi) can improve therapeutic efficacy of endothelial progenitor cells (EPC) from patients with DM. Patients with DM had fewer EPCs and increased rates of apoptosis. DM EPCs also exhibited higher levels of GSK3β activity resulting in increased levels of phosphorylated β-catenin. Proteomic profiling of DM EPCs treated with GSKi identified 37 nonredundant, differentially regulated proteins. Cathepsin B (cathB) was subsequently confirmed to be differentially regulated and showed 40% less baseline activity in DM EPCs, an effect reversed by GSKi treatment. Finally, in vivo efficacy of cell-based therapy was assessed in a xenotransplant femoral wire injury mouse model. Administration of DM EPCs reduced the intima-to-media ratio, an effect that was further augmented when DM EPCs were pretreated with GSKi yet absent when cathB was antagonized. In DM, increased basal GSK3β activity contributes to accelerated EPC cellular senescence, an effect reversed by small molecule antagonism of GSK3β, which enhanced cell-based therapy after vascular injury.


Use of endothelial progenitor cell (EPC) populations for cell-based therapies is beneficial in a host of cardiovascular conditions, including peripheral vascular disease, pulmonary arterial hypertension, and myocardial infarction. However, the administration of autologous EPCs in patients with established disease is often hindered due to attenuated cellular yield and biologic activity. Accordingly, a myriad of strategies have been used with the goal of improving EPC cellular yield, survival, and function, including overexpression of endothelial nitric oxide synthase (1) or administration of small molecule inhibitors of glycogen synthase kinase-3β (GSK3β) (2).

GSK3β is a ubiquitously expressed serine/threonine protein kinase negatively regulated by Wnt signaling. Notably, GSK3β is regulated by various modalities, including phosphorylation, substrate priming, retention in inhibitory complexes, and subcellular compartmentalization (38). Accordingly, GSK3β levels do not necessarily reflect actual physiologic activity. Under basal conditions, GKS3β phosphorylates β-catenin (β-cat) resulting in proteasomal degradation of this nuclear transcription factor. Pharmacologic GSK3β inhibition or Wnt3a stimulation promotes hematopoietic stem cell self-renewal and repopulation of cell lineages in vitro and in vivo (9). In EPCs, GSK3β inhibition by transfection of a dominant negative mutant or by small molecule inhibition improves the therapeutic capacity of cells, thereby augmenting angiogenesis in ischemia models and improving arterial repair after vascular injury (2,10). This is of particular interest given that GSK3β expression and activity are dysregulated in patients with diabetes mellitus (DM) (11) a population in whom EPC function is attenuated (12).

Hence, we sought to explore the potential for pharmacologic inhibition of GSK3β to improve arterial repair in patients with DM. Herein we demonstrate that inhibition of GSK3β in DM EPCs abrogates apoptosis and improves EPC yields in vitro. Moreover, using a proteomic approach, we identify and confirm the differential regulation of candidate proteins for the observed benefits of GSK3β inhibition. Among the identified proteins, increased cathepsin B (cathB) activity is demonstrated to be essential for reductions in EPC apoptosis and necessary for increased efficacy in a model of cell-based therapy. These findings suggest that inhibition of GSK3β is an important strategy for improving autologous cell-based therapy in patients with DM and acts through a novel mechanism involving increased activity of cathB.

Research Design and Methods

Cell Isolation

EPCs were isolated as previously described (2,10,13). Cells were plated on human fibronectin (Sigma-Aldrich) coated six-well plates at a density of 5.0 × 106 peripheral blood mononuclear cells per well in endothelial growth media-2 (EGM2; Lonza). After 4 days in culture, nonadherent cells were removed and plates washed with PBS. All experiments performed with day 4–7 cells, with samples from individual donors representing a single replicate. For enumeration, EPCs were incubated with 1,1’-dioctadecyl-3,3,3′,3′-tetramethlyiodocarbocyanine-acetylated LDL (acLDL, 2.5 μg/mL; Invitrogen), followed by fluorescein isothiocyanate-conjugated Ulex europaeus agglutinin-1 (5 μg/mL; Sigma-Aldrich) and then counterstained with DAPI.

GSK3β Inhibitors

GSK-3β small molecule and peptide inhibitors (GSKi) were assayed for efficacy in increasing cell yield and blocking phosphorylation of β-cat. All inhibitors were diluted in DMSO to a final concentration of 0.1% or sterile PBS if water-soluble. Specifically, AR-A014418 (Sigma-Aldrich), CHIR98014 (Cedarlane), (22,3E)-6-bromoindirubin-3-oxime (Calbiochem), GSK peptide inhibitor (Calbiochem), and LiCl (20 nmol/L; Sigma-Aldrich) were tested for in vitro efficacy.

Apoptosis Assay

EPCs were maintained under basal culture conditions or serum-starved for 24 h as indicated. Nonadherent cells were removed by washing with PBS. Subsequently, adherent cells were lifted by gentle agitation with 1 mmol/L EDTA. Cells were counted, and 1 × 105 EPCs were stained with Annexin V-fluorescein isothiocyanate and propidium iodide, according to the manufacturer’s instructions (Becton, Dickinson and Company). All flow studies were performed on a Beckman Coulter Cytomics FC 500 cytometer. Early apoptotic cells were defined as AnnexinV+/propidium iodide.

Protein Lysate Preparation

Cells were centrifuged at 220g for 5 min at room temperature. Cell pellets were put on ice, and cell lysis buffer (7 mol/L urea [w/v], 2 mol/L thiourea [w/v], 4% CHAPS [w/v], and 1% dithiothreitol [DTT; w/v]) was added. Cell lysates were vortexed and kept at room temperature for 30 min to enable protein solubilization. Cell lysates were then sonicated, vortexed, and centrifuged at 14,000g for 15 min at room temperature. The supernatant was transferred, and protein quantification was realized with the 2-D Quant Kit (GE Healthcare).

Two-Dimensional PAGE

The total proteins (30 μg) were passively rehydrated overnight and applied to immobilized pH gradient strips (11 cm, pH 4–7; Bio-Rad). Isoelectric focusing was done using the Agilent fractionator (Agilent) in the in-gel mode. Each focused strip was subsequently equilibrated in 4.0 mL equilibration buffer I (6 mol/L urea [w/v]; 50 mmol/L Tris-Cl, pH 8.8; 2% SDS [w/v]; 30% glycerol [v/v]; bromophenol blue [trace]; 1% DTT [w/v]) for 15 min with gentle agitation, followed by the equilibration buffer II (equilibration solution I with DTT replaced by 2.5% iodoacetamide [w/v]) for 15 min with gentle agitation. The two-dimensional (2-D) separation was performed on a 10% SDS-PAGE gel in an Ettan DALTsix Electrophoresis Unit (GE Healthcare) at 10 mA per gel at 25°C for ∼18 h. Two technical replicates were done independently for each biological sample.

Image Acquisition and 2-D Gel Analyses

The same scanning conditions were used for each gel. The scanned gels were analyzed for gel-to-gel matching using PDQuest 2-D analysis software (advanced version 8.0; Bio-Rad), according to the protocol provided by the developer. Each matched protein spot was assigned a unique sample spot protein number. For gel comparison, a statistical approach was applied for determining statistically differentially regulated proteins using the PDQuest software. The Student t test was performed with 95% significance level to determine which proteins were statistically differentially regulated between the healthy cells and the patient cells nontreated and the patient cells nontreated and treated with GSKi. A minimum of 1.5-fold change was considered for the upregulated proteins and 0.67-fold for downregulated proteins. Protein spots with differential expression patterns on 2-D gel electrophoresis maps were excised with the automated spot excision robot, the EXQuest spot cutter (Bio-Rad).

Protein Identification

Liquid chromatography-mass spectrometry (LC-MS) analysis was performed at the Ottawa Hospital Research Institute Proteomics Core Facility (Ottawa, ON, Canada). Peptides were loaded onto a peptide trap (Agilent) for 5 min at 15 µL/min using a Dionex UltiMate 3000 RSLC nano high-performance LC. Peptides were eluted over a 20-min gradient of 3–45% acetonitrile with 0.1% formic acid at 0.3 µL/min onto a 10-cm analytical column (New Objective Picofrit self-packed with Zorbax C18) and sprayed directly into a LTQ Orbitrap XL hybrid MS using a nanospray source (Thermo Scientific, Waltham, MA). MS data were acquired in a data-dependent fashion, with MS scans acquired in the Fourier transform cell while tandem MS (MS/MS) scans were acquired in the ion trap module.

MS/MS spectra were matched against a custom database (2011_07_human_con) consisting of human sequences from SwissProt (2011_07 version of uniprot_sprot.fasta.gz from concatenated with a database of common contaminants (downloaded from on 9 June 2011) using MASCOT 2.3.01 software (Matrix Science, London, U.K.) with MS tolerance of ±5 ppm and MS/MS tolerance of 0.6 Da. Oxidation of methionine, carbamidomethylation of cysteine, deamidation, protein N-terminal acetylation, conversion of peptide N-terminal Glu or Gln to Pyro-Glu, and phosphorylation of serine or threonine were allowed as potential modifications.

Quantitative PCR

Total RNA was isolated using Trizol (Invitrogen) and purified using RNeasy minikits. Subsequently, RNA was quantified using a NanoDrop 1000 (Thermo Scientific), and real-time PCR was performed using Omniscript kit as directed (Qiagen). All real-time PCR experiments were performed using the SYBR Green Jumpstart Taq Ready Mix (Sigma-Aldrich) on a LightCycler 480 (Roche) and analyzed with accompanying software according to the Pfaffl method (14).

Western Blots

Western blots were performed using standard techniques. Briefly, protein was isolated in radioimmunoprecipitation assay buffer using a ratio of 50 μL/1 million cells. The sample was then allowed to incubate on ice for 30 min, followed by centrifugation. The supernatant was assayed using a standard bicinchoninic acid assay (Thermo Scientific). Protein was then separated on 10% acrylamide gels and transferred to polyvinylidene fluoride membranes using iBlot as directed (Invitrogen). After transfer, the membrane was blocked for 1 h with 5% skim milk in Tris-buffered saline-Tween 20 at room temperature. Primary antibodies were incubated overnight at 4ºC. Primary antibodies were plasminogen activator inhibitor-2 (PAI-2, 8:1000; AP6562c, Abgent), β-actin (1:100000; Sigma-Aldrich), gelsolin (1:1000; ab11081, Abcam), GDP dissociation inhibitor-2 (1:2000; ab49193, Abcam), small calcium-binding mitochondrial carrier-1 (SCaMC-1) protein (1:500; sc-133987, Santa Cruz Biotechnology), and CatB (1:10000; ab58802, Abcam). Membranes were then washed and incubated with biotinylated secondary antibodies (Santa Cruz Biotechnology) for 1 h, then visualized using ECL Plus (Amersham Biosciences).

CathB Activity Assay

CathB activity was assayed in day 5 EPCs using a standardized CatB fluorometric assay kit as directed (Abcam). Briefly, EPCs were washed with PBS, lifted with EDTA, and 5 × 106 cells collected by centrifugation. Cells were lysed by incubation with cell lysis buffer, pelleted, and 50 μL was transferred to a 96-well plate. Subsequently, 2 μL of 10 mmol/L cathB substrate labeled with amino-4-trifluoromethyl coumarin was added, and samples were incubated for 2 h. Plates were read on the SynergyMx microplate reader (BioTek).

Vascular Endothelial Growth Factor Secretion Assay

EPCs were cultured using standard techniques to day 4. Cells were subsequently lifted, counted, and replated in 96-well plates at equivalent densities in vascular endothelial growth factor (VEGF)-free media with treatment as indicated. After 24 h, media was removed and assayed for VEGF levels using a standard VEGF ELISA kit (R&D Systems) according to the manufacturer’s protocol.

Human Umbilical Vein Endothelial Cells Adhesion Assay

Human umbilical vein endothelial cells (HUVEC) were cultured to confluence in 96-well plates, then treated with 10 ng/mL tumor necrosis factor-α for 6 h to activate cells. EPCs (106) were cultured in with 5 μmol/L calcein acetomethoxy for 30 min, lifted with EDTA, pelleted, and resuspended in EGM-2. Subsequently, 4 × 104 cells were plated on the activated HUVECs and allowed to adhere for 1 h. Plates were read on the SynergyMx microplate reader, washed three times with PBS, and reread. Adherence was expressed as the percentage of fluorescence retained after washing.

Cell Invasion Assay

Day 5 EPCs were treated as indicated. A modified Boyden chamber and a 12-μm Nuclepore filter (Becton Dickinson) with a matrigel matrix (Becton Dickinson) was used to assay EPC invasiveness. Briefly, 5 × 106 EPCs were placed in the upper chamber with endothelial basal medium and with EGM-2 in the lower chamber. Cells were permitted to migrate for 24 h at 37ºC. Cells were counterstained with DAPI and counted in six random high-power fields.

CD-1 Nude Femoral Artery Wire Injury Model

CD-1 nude athymic male mice were purchased from Charles River Laboratories and permitted to acclimatize for 2 to 6 weeks before surgery. Mice underwent femoral artery wire injury as previously described (15). Subsequently, 2 × 105 EPCs from patients with DM were infused into the adjacent vein using a blunt needle. None of the animals exhibited ischemia in the hind limb. Mice were recovered and were killed at 14 days for tissue analysis.

Tissue Processing

Mice underwent perfusion fixation with buffered formalin at time they were killed. Arteries were fixed for 24 h in formalin and dehydrated in ethanol. Arteries were mounted in paraffin blocks and sectioned in 5-μm sections until 100 μm from the branch vessel site. These sections were stained with hematoxylin and eosin, and analysis was performed using a computer-assisted digital imaging system (Image-Pro Plus; Media Cybernetics).

Ethics and Statistics

All protocols involving human donors were approved by the Ottawa Heart Institute Research Ethics Committee, with participants providing written informed consent. These studies conformed with the Declaration of Helsinki for the use of human tissue. Animal experimental protocols were approved by the University of Ottawa Animal Care Committee and adhered to the Canadian Council on Animal Care guidelines.

Data are expressed as mean ± SEM. Statistical significance was determined for P < 0.05. Pairwise comparisons were performed using a paired Student t test, with multiple comparisons performed with a one- or two-way ANOVA, with Holm-Šídák post hoc testing as appropriate.


DM Accelerates Apoptosis in EPCs Through Increased GSK3β Activity

EPCs isolated from human subjects were cultured for 7 days and then characterized using immunolabeling for Ulex europeus agglutinin-1 and acetylated-LDL uptake (Fig. 1A). The baseline characteristics of the human subjects are presented in Supplementary Table 1. Samples derived from patients with DM yielded fewer EPCs than those derived from healthy controls (n = 12, 14.9 ± 4.6 vs. 38.5 ± 6.7 cells per high-power field, P < 0.01; Fig. 1B). Several GSKis were supplemented in increasing concentrations to identify the optimal inhibitor and concentration. CHIR98014 at a concentration of 1 μmol/L yielded optimal EPC yields and significantly greater inhibition of GSK3β (Supplementary Fig. 1B and C). Subsequently, CHIR98014 was used as the preferential GSKi in all experiments.

Figure 1

Increased GSK3β signaling in DM EPC reduces in vitro cell yield. A: EPCs at 7 days labeled with fluorescein isothiocyanate-conjugated Ulex europaeus agglutinin-1 (left panel, green), AcLDL-DiI (red) and DAPI (blue, middle panel), and merged images (right panel). B: Light microscopy images of cells from patients with DM and healthy control patients under basal conditions or treated with GSKi. C: GSKi treatment increases EPC yield in healthy control patients and in patients with DM. HPF, high-power field. D: GSKi treatment attenuates apoptosis under basal and serum-starved conditions. E: EPCs from patients with DM have higher levels of phosphorylated (phospho)–β-catenin. GSKi treatment in both groups markedly reduced these levels.

Notably, supplementation of the culture media with GSKi resulted in ∼300% increases in the yields of EPCs in DM and in healthy controls (P < 0.01; Fig. 1C). Under basal conditions, the apoptosis index at 96 h was higher in patients with DM as measured by annexin V and propidium iodide double labeling (9.2 ± 0.9 vs. 7.3 ± 0.9, P = 0.02; Fig. 1D) an effect attenuated through GSK3β inhibition. As expected, serum starvation, used to reproduce cell stress after therapeutic transplantation, resulted in marked increases in the apoptosis index in DM and healthy cells, an effect abrogated with GSKi treatment to near basal levels (Fig. 1D). Importantly, higher levels of phosphorylated β-cat), the product of active GSK3β activity, in EPCs derived from DM patients (0.55 ± 0.08 vs. 0.42 ± 0.04, P = 0.04) were markedly reduced in both cohorts of cells after GSKi treatment (P < 0.01; Fig. 1E). As expected, basal levels of GSK3β were increased in DM patients, with no change after GSKi treatment (Supplementary Fig. 2). These findings demonstrate that increased basal activity of GSK3β in EPCs from patients with DM results in accelerated apoptosis in vitro, an effect abrogated by use of isoform-specific small molecule inhibitors.

Proteomic Profiling of EPCs in DM

To ascertain mechanistic insight into the beneficial effects of GSKi on EPCs, analyses were performed of the proteome of EPCs from patients with DM, from DM patients treated with GSKi, and from healthy control subjects (n = 3 for each). Isolation protocols were scaled up to yield sufficient cellular yields for the analyses. Differential yields between healthy control subjects and DM patients were maintained in scaled-up protocols, as were the effects of GSKi (Supplementary Table 2). After 2-D gel electrophoresis and digital image analysis, 242 unique protein spots were identified. Differentially regulated candidate targets were identified if a 2.0-fold upregulation or 0.5-fold downregulation of spot intensity was identified between the groups (Supplementary Fig 3). A total of 37 nonredundant proteins met these criteria for significant differential expression (P < 0.05). These spots were excised from the Sypro-Ruby–stained gels and underwent in-gel trypsinization. The peptide mixtures were analyzed by LC-MS/MS analysis, and the results of the MS identification are presented in Table 1.

Table 1

Proteomics analysis

Western blot analysis of three target proteins of interest was performed. Specifically, cathB upregulation (1.9 ± 0.06 vs. 5.8 ± 2.0), gelsolin downregulation (5.6 ± 1.5 vs. 1.9 ± 0.9), and PAI-2 upregulation (2.1 ± 0.7 vs. 5.1 ± 0.5) was confirmed in EPCs from patients with DM (P < 0.05 for all comparisons; Fig. 2B). Because β-cat acts as a transcription factor, we hypothesized that regulation of protein levels seen by GSKi were most likely transcriptional in nature. Indeed, quantitative PCR of mRNA isolated from DM EPCs under basal and GSKi-treated conditions revealed a 4.6 ± 1.2-fold increase in cathB (P < 0.01; Fig. 2C), a 0.5 ± 0.4-fold reduction in gelsolin (P < 0.05), and a 10.4 ± 5.0-fold increase in PAI-2 (P < 0.01). Neither SCaMC-1 nor GDP dissociation inhibitor-2 appeared differentially regulated at the mRNA or protein level. These findings confirmed the validity of the proteomics results and identified several targets known to regulate apoptosis and to be expressed in EPCs.

Figure 2

Proteomic analyses of EPCs from patients with DM with and without GSK3β inhibition. A: Sample 2-D gels used in proteomics analysis. Western blot (B) and quantitative analysis (C) of cathB, gelsolin, and PAI-2 (n = 6). D: Quantitative PCR analysis of candidate genes identified by proteomics analysis. Fold change represents GSKi sample compared with control sample in EPCs derived from patients with DM (n = 6). GDI-2, GDP dissociation inhibitor-2; MW, molecular weight.

CathB Is Required for GSKi-Mediated Reductions in Apoptosis

To ascertain the role of cathB activity in EPC dysfunction in patients with DM, we assayed enzyme activity levels. As predicted by mRNA and protein levels, EPCs from patients with DM had 40% less measurable activity compared with healthy control cells (5,587.3 ± 1,455.1 vs. 8,251.2 ± 771.9 relative fluorescence units, P = 0.03; Fig. 3A). Treatment with GSKi raised measurable activity fourfold in both groups (P < 0.01). To ascertain whether cathB activity was required for changes in phenotype associated with GSKi, we used CA074, a specific inhibitor of cathB (12). Supplementation of the culture media with CA074 abrogated a GSKi-induced dose-dependent increase in cathB activity at all levels assayed (Fig. 3B). In turn, functional assays were performed to ascertain the necessity of intact cathB activity for phenotypic changes in apoptosis rates, VEGF secretion, and endothelial adhesion by EPCs cultured from patients with DM. Notably, the effect of GSKi on apoptosis could be attenuated when cells were cultured in the presence of the cathB inhibitor CA074 (Fig. 3C). Blockade of cathB activity demonstrated no effect on VEGF secretion or binding of EPCs to activated HUVECs (Fig. 3D and E). Instead, only EPC invasive capacity, as assessed by matrigel invasion, was also cathB-dependent, because this parameter increased threefold with GSKi treatment (P < 0.01; Fig. 3F). These findings demonstrate that EPCs derived from patients with DM have intrinsically lower cathB activity and that induction of expression by GSKi attenuates higher levels of apoptosis and improves invasiveness in vitro.

Figure 3

CathB is required for the GSKi-mediated reduction in the rate of apoptosis in DM EPCs. A: CathB activity is reduced at baseline in DM EPCs and increased by treatment with GSKi (n = 6). B: Dose-dependent increase in cathB activity is achieved with GSKi, and the effect is abrogated by treatment with CA074, the cathB inhibitor (n = 12). C: The basal rate of apoptosis is reduced by 60% in EPCs treated with GSKi. This effect is lost with cathB inhibition (n = 12). D: Improvements in VEGF secretion achieved with GSKi occur independently of cathB activity (n = 12). E: Increased EPC adhesion achieved with GSKi occurs independently of cathB activity (n = 12). F: Improvement in EPC invasion achieved with GSKi is dependent of cathB activity (n = 6). RFUs, relative fluorescence units.

Improvements in EPC-Based Cell Therapy Mediated by GSK3β Inhibition Require CathB Activity

To investigate the effect of DM on reparative capacity of EPCs, an increasing number of cells from healthy control subjects and patients with DM were administered in a xenotransplant femoral artery wire injury model. The intima-to-media (IM) ratio was assessed at 14 days. DM and cell dose both significantly altered the therapeutic effect as assessed by two-way ANOVA, with EPCs from patients with DM exhibiting lower therapeutic benefit at 2.5 × 104 and 1.0 × 105 cells (P < 0.05 for all dose comparisons; Fig. 4A). To test the therapeutic necessity of cathB upregulation to DM EPC–mediated arterial healing, we administered cells from DM patients with GSKi and CA074 (Fig. 4B). Administration of cells alone reduced the IM ratio 40% (1.37 ± 0.15 vs. 0.81 ± 0.12, P < 0.01; Fig. 4C) at 14 days. Pretreatment of cells with GSKi resulted in further reduction in the IM ratio compared with untreated EPCs (0.50 ± 0.07, P = 0.02), an effect that was lost when cells were pretreated with GSKi and CA074 (0.77 ± 0.11, P = 0.03 vs. EPC + GSKi). Of note, effects were observed in each individual subject (Fig. 4D). These findings confirm that upregulation of cathB by GSK3β inhibition partly explains the therapeutic enhancement of EPC-based therapy in patients with DM.

Figure 4

CathB is required for improvement in DM EPC–mediated arterial repair achieved with GSKi. A: Changes in neointima formation using increasing doses of EPCs from healthy age-matched control subjects and patients with DM (n = 6). B: Representative 14-day cross-sections at low magnification with magnified regions of neointima highlighted. Arrows indicate the internal elastic lamina. C: GSKi treatment of DM EPCs results in important reductions in neointima formation, an effect lost with cathB inhibition (n = 6). D: Changes in neointima formation of individual patients according to treatment group. All patients demonstrated improvement in arterial homeostasis with GSKi treatment (n = 6). I-to-M, intima-to-media.


Identifying cell-enhancement strategies, be it genetic modification (1) or small molecule antagonism (2), is essential to improve the therapeutic efficacy of cell-based therapies. Indeed, transplanted cells in a wide variety of models demonstrate poor engraftment, with high rates of cell attrition. Herein, we highlight important differences in GSK3β signaling as a factor for enhanced EPC senescence in DM resulting in accelerated rates of apoptosis. Moreover, using a proteomics approach, we identified upregulation of cathB as protective for reductions in basal and stress-induced apoptosis. Finally, in a xenotransplant model, we confirm that cathB activity is required for GSKi-induced improvements in EPC mediated arterial repair.

Patients with DM have increased rates of cardiovascular disease and markedly higher rates of in-stent restenosis after revascularization. This, in part, is owing to attenuated EPC function in patients and fewer circulating cells (16). Multiple mechanisms of EPC dysfunction have been identified, including endothelial nitric oxide synthase uncoupling (17), increased reactive oxygen species, and the effects of advanced glycation end products (18). As well, glucose is known to impair the activity of the phosphatidylinositide 3-kinase/Akt pathway, a regulator of GSK3β signaling, and has been implicated in EPC differentiation by forkhead box class O1 transcription factors (19). GSK3β is known to be highly upregulated in a number of tissues in DM, with our data now confirming increased phosphorylated β-cat, the end product of GSK3β, in EPCs derived from patients with DM. Thus, dysregulation of β-cat signaling represents a new target for cell enhancement of EPCs in patients with DM.

To date, three studies suggest a beneficial effect of GSK3β antagonism in EPC-based therapy (2,10,20). The current report is the first to use an unbiased proteomic approach to identify differentially regulated proteins in EPCs in order to identify a potential mechanism of benefit. Using this technique, we identified cathB, a protein with known roles as both a pro- and antiapoptotic factor. Similar to observations in several cell lines (12,21), we noted cathB downregulation in DM-derived EPCs resulted in enhanced apoptosis, an effect rescued with GSKi therapy. Furthermore, low cathB activity has been identified as being linked to progression of diabetic nephropathy, ostensibly due to insufficient fibronectin degradation (22). Analogous to our model, studies using bone marrow–derived EPCs have demonstrated promotion of angiogenesis in regions of glomerular lesions representing a potential therapeutic target (23). Our data now compliment these findings, highlighting the role of low cathB activity in DM and, in the case of cell-based therapy, a mechanism to improve therapeutic effect through GSK3β inhibition.

The mechanism by which GSK3β regulates cathB remains to be elucidated. Although others have noted GSK3β-mediated relocation of cathB from lysosomes to cytosol as a means by which it modulates apoptosis (21,2426), the current study supports transcriptional modulation of cathB as the most likely mechanism through which increased activity was achieved. Indeed, GSK3β is known to inhibit nuclear factor (NF)-κβ in quiescent cells resulting in increased apoptosis, an effect that depends on β-cat (27). Previous reports have noted an upregulation of cathB by NF-κβ, which has a binding site in the cathB promoter (28). Supporting this concept are recent reports showing that exogenous recombinant heat shock protein 27 treatment improves EPC function in vivo (29) and increases NF-κβ signaling (30). Nonetheless, although it remains attractive to hypothesize that GSK3 effects on NF-κβ are responsible for cathB regulation, definitive studies in EPCs are needed.

This study is not without limitations. First, the cell types being used for therapeutic effect continue to broaden, and we cannot be certain that the mechanisms described in the current study apply to other cell populations. However, the current experiments were performed in primary cells commonly used in clinical studies and were derived from patients with DM.

Second, although there are clear improvements in apoptosis and invasiveness, we do not demonstrate increased cell retention in our in vivo model. However, it is well documented in numerous animal models of cell-based therapy that cellular retention is a rare event, whereas the paracrine effect of cell therapy may be an important early beneficial mechanisms (31).

Finally, although numerous specific inhibitors for GSK3β were tested, we cannot entirely rule out that nonspecific GSK3α inhibition may have partly contributed to the observed biological effect. However, the relative contribution of GSK3α in DM-induced EPC dysfunction is unlikely to be significant but remains to be elucidated.

Despite these limitations, our study is the first to highlight cathB regulation by GSK3β as a potential cell-enhancement strategy for patients with DM, and our unbiased proteomic approach highlights potential future targets, such as PAI-2, for future investigation.

In conclusion, inhibition of GSK3β activity in EPCs from patients with DM results in upregulation of cathB expression and activity. Increased cathB activity improves EPC invasiveness, reduces apoptosis, and ameliorates the therapeutic effect of cell-based therapy. Small molecule antagonism of GSK3β is a cell enhancement strategy for patients with DM.

Article Information

Funding. The Canadian Institute for Health Research and Medtronic collectively provide E.R.O. with a peer-reviewed research chair (URC #57093, IGO 94418) and an operating grant.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. B.H. and J.R.L. planned, designed, and carried out experiments, analyzed data, and wrote the manuscript. X.M., T.Se., J.E.R., T.Si., and Y.-X.C. carried out experiments and reviewed the manuscript. D.S. and E.R.O. conceived experiments and reviewed the data and manuscript. E.R.O. is the guarantor of this work and as such had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. This work was presented at the 62nd Annual Scientific Session of the American College of Cardiology, San Francisco, CA, 9–11 March 2013.


  • Received June 23, 2013.
  • Accepted November 21, 2013.

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. See for details.


| Table of Contents