The Peroxisome Proliferator–Activated Receptor-γ Agonist Pioglitazone Increases Number and Function of Endothelial Progenitor Cells in Patients With Coronary Artery Disease and Normal Glucose Tolerance

The study was designed as a prospective, randomized, double-blind clinical trial on patients with stable coronary artery disease and normal glucose tolerance. The trial was conducted with approval of the local ethics committee (Ethikvotum 130/05) and the German Federal Institute for Drugs and Medical Devices (Bfarm no. 4030855; Eudora-CT no. 2005-003939-42). Informed consent was obtained from all subjects.

All patients had angiographically documented and clinically stable coronary artery disease. Medication was not changed during the course of the study. Exclusion criteria included functional New York Heart Association class III or IV heart failure, left ventricular dysfunction measured as left ventricular ejection fraction <40%, elevated serum creatinine level, active liver disease, alanine transaminase levels of ≥2.5 times the upper limit of normal, regular use of nonsteroidal anti-inflammatory drugs, psychiatric diseases, and pregnancy.

Fifty-four eligible patients with normal fasting glucose underwent an oral glucose tolerance test (Dextro O.G-T.; Roche). Eighteen patients showed impaired glucose tolerance, and 36 patients with normal glucose tolerance were enrolled in the study. They were randomized to a 30-day treatment with pioglitazone (45 mg) or matching placebo in addition to an optimized cardiovascular medication. Randomization was performed by the pharmacy of the Universitätsklinikum des Saarlandes independently from the investigators. Venous blood samples were taken on the day of enrolment in the study and after 30 days of placebo/TZD therapy. Adiponectin serum levels were quantified by the human adiponectin radio immunoassay (Linco). High-sensitivity C-reactive protein (hsCRP) was measured by a turbidimetric latex test (CRP Dynamik; Biomed).

Fluorescence-activated cell sorter (FACS) analyses were used to characterize mononuclear cells (MNCs) as previously described (20). MNCs were selected using Ficoll density gradient centrifugation (Biocoll Separating Solution; Biochrom) from 20 ml human blood drawn in sodium citrate and resuspended in 1 ml endothelial cell basal medium (EBM; CellSystems) with supplements (1 μg/ml hydrocortisone, 3 μg/ml bovine brain extract, 30 μg/ml gentamicin, 50 μg/ml amphotericin B, 10 μg/ml human endothelial growth factor, and 20% FCS). One hundred microliters was diluted in FACS buffer (PBS, 0.1% bovine albumin, and 15 units/l aprotinine) and immediately used for assessment of isotype controls and CD34-fluorescein isothiocyanate (CD34-FITC; Becton Dickinson)/goat anti-human KDR (R&D Systems) with the conjugated secondary antibody anti-goat streptavidin-phycoerythrin (PE) (DAKO). All antibody incubations were kept on ice in the dark. Isotype identical antibodies (IgG 2_a-FITC and IgG 2_a-PE; Becton Dickinson) served as controls in every experiment, and unspecific binding was blocked by addition of human serum. Cells were washed twice in FACS buffer and fixed with 2% paraformaldehyde. Each measurement was performed in two separate tubes by assessment of 10⁵ MNCs in the lymphocyte gate using the Becton Dickinson FACSCalibur and Cell Quest Pro software. The intra-assay variability (coefficient of variation) was similar between groups and assays: FACS analysis of CD34⁺/KDR⁺ EPCs, placebo 28% and TZD 29%; 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine–labeled acetylated LDL (DiLDL)⁺/lectin⁺ EPCs, placebo 29% and TZD 30%. The interassay correlation between CD34⁺/KDR⁺ EPCs and DiLDL⁺/lectin⁺ EPCs was R² = 0.54 (P < 0.01).

MNCs were isolated by Ficoll density gradient centrifugation, the cell pellet was resuspended in 5 ml EBM (Cell Systems), and 4 × 10⁶ mononuclear cells were cultured on fibronectin-coated dishes in EBM (CellSystems) (18,35). After 4 days in culture, adherent cells were incubated with DiLDL (2.4 μg/ml; CellSystems) and stained using FITC-labeled Ulex europaeus agglutinin I (lectin, 10 μg/ml) (Sigma). All measurements were performed in triplicates. Double-positive cells were counted manually by two independent observers blinded to the study (Nikon Eclipse E600; magnification ×10). Expression of PPARγ in human EPCs was examined by RT-PCR: PPAR1 forward, 5′-CTGGCCTCCTTGATGAATAA; PPAR1 reverse, 5′-GGCGGTCTCCACTGAGAATA; PPAR2 forward, 5′-AGGGCGATCTTGACAGGAAAG; PPAR2 reverse, 5′-CCCATCATTAAGGAATTCATGTCA. To test the effects of pioglitazone in vitro, EPCs isolated and cultured from healthy volunteers were treated with pioglitazone (10 μmol/l; Takeda Pharmaceuticals), the PPARγ antagonist GW9662 (1 μmol/l per l; Alexis Biochemicals), or adiponectin (20 μg/ml; Alexis Biochemicals) for 24 h.

Modified Boyden chambers were used to assess the migratory capacity of EPCs (25,35). On day 1, 4 × 10⁶ MNCs were plated on noncoated six-well culture dishes and cultured for 4 days. Culture medium was removed, and cells were harvested by trypsination and centrifugation. The pellet was resuspended in 300 μl EBM and counted. Cells (1 × 10⁵) in 500 μl EBM without supplements were placed in the upper chamber (HTS Fluoroblock, 8 μm pore size, in triplicate; BD Biosciences), and the chamber was placed in a 24-well plate containing EBM without supplements and 100 ng/ml SDF-1 (R&D Systems). After 24 h, the filter was carefully removed, washed, fixed, and incubated with labeled DiLDL as described above. SDF-1–stimulated migratory capacity was then quantified by counting the migrated EPCs on the lower surface of the filter using fluorescence microscopy (magnification ×40, in triplicate).

Colony-forming units (CFUs) were counted to examine the clonal expansion capacity of EPCs. After density gradient centrifugation, 5 × 10⁶ MNCs were plated in a fibronectin-coated six-well plate for 48 h (in triplicate). Then, the nonadherent cells in the supernatant were centrifuged and resuspended, and 10⁶ cells were plated into 24-well plates. After 5 more days in culture, EPC colonies (defined as clusters of more than 15 cells) were counted under a light microscope. Values are expressed as absolute colony number per well.

NADPH oxidase activity was measured by a lucigenin-enhanced chemiluminescence assay in buffer B containing 50 mmol/l phosphate (pH 7.0), 1 mmol/l EGTA, protease inhibitors (Complete; Roche), 150 mmol/l sucrose, 0.005 mmol/l lucigenin, and 0.1 mmol/l NADPH, as described previously (36). EPCs were lysed after washing with PBS in ice-cold buffer B, lacking lucigenin and substrate. Basal and phorbol myristate acetate (PMA)-stimulated NADPH oxidase activity was determined in 100 μg aliquots of the sample measured over 10 min in quadruplicates using NADPH as substrate in a scintillation counter (Berthold Lumat LB 9501) in 1-min intervals.

Results are represented as mean ± SE. Statistical analysis was performed using the GraphPad Prism software (version 3.02). ANOVA and Student's t test were applied where applicable. Post hoc comparisons were made by using the Bonferroni test. Values of P < 0.05 were considered statistically significant.

The baseline characteristics of the patients are summarized in Table 1 and did not differ between groups. There was no difference in medication between groups. The medication was not changed during the course of the study. Serum glucose and lipid concentrations were not different between groups and were not affected by treatment with pioglitazone. TZD treatment significantly reduced insulin concentrations and the homeostasis model assessment (HOMA) index (Table 1). None of the smokers stopped smoking during the course of the trial. Nine TZD and 10 placebo patients underwent coronary angioplasty or stenting (one drug-eluting stent in each group). The effect of TZD on EPC parameters did not differ between percutaneous coronary intervention (PCI) and non-PCI patients. Six individuals in the placebo group and four in the TZD group met the current International Diabetes Federation criteria for metabolic syndrome.

Adiponectin is a peptide hormone regulator produced in adipocytes, which is known to be increased by TZD treatment (1,37). The mean adiponectin level in the placebo group was 12.2 ± 1.2 μg/ml before and 12.2 ± 1.4 μg/ml after the study. In contrast, the adiponectin serum concentrations of the TZD group showed an increase to 322% after 4 weeks (11.4 ± 1.1 vs. 36.8 ± 2.1 μg/ml; P < 0.001) (Fig. 1A). Importantly, in every individual TZD patient, but no patient of the placebo group, the adiponectin level increased by more than twofold.

hsCRP serum concentration is a marker of vascular inflammation and independent predictor of vascular events. Treatment with the PPARγ agonist reduced hsCRP levels from 3.9 ± 0.7 to 1.69 ± 0.3 mg/l (P < 0.05; Fig. 1B), whereas placebo treatment had no effect (3.8 ± 0.7 vs. 2.9 ± 0.7 mg/l).

CD34⁺/KDR⁺ EPCs have been shown to predict cardiovascular events (20). Figure 2A depicts a representative FACS scan for CD34-FITC/KDR-PE in peripheral blood. Treatment with TZD for 30 days resulted in upregulation of CD34⁺/KDR⁺ EPC numbers to 142% (55 ± 5 vs. 78 ± 6/10⁵ cells; P < 0.01; Fig. 2B). In the placebo group, the number of CD34⁺/KDR⁺ EPCs did not change.

Quantification of DiLDL uptake and lectin staining was used as an established second method to quantify and characterize EPCs (Fig. 3A) (11). Placebo treatment had no effect on the number of DiLDL⁺/lectin⁺ double-positive cells per microscopic field (122 ± 5 before treatment vs. 103 ± 4 after treatment; P > 0.05). In contrast, DiLDL⁺/lectin⁺ EPCs increased from 86 ± 4 to 154 ± 5 cells/microscopic field after TZD treatment (P < 0.05).

Migration is an important functional property of EPCs that can be regulated independent of EPC numbers (11,25). After 1 month of therapy, there was no change in placebo controls, whereas TZD treatment resulted in an increase of EPC migration to 146 ± 9% per EPC number (P < 0.05; Fig. 4).

The ability to clonally expand and to create colonies in an endothelial-specific medium is a key functional feature of EPCs (14). There was no significant change in CFUs in patients treated with placebo (99 ± 9 CFUs at baseline vs. 119 ± 8 after 1 month; P > 0.05). After treatment with TZD, however, the development of CFUs per number of EPCs was upregulated to 172 ± 12%; P < 0.001; Fig. 5A).

The NADPH oxidase is an important source of endothelial superoxide anions (38). Reactive oxygen species impair the function of both mature and immature endothelial cells (16,39,40). Treatment of cultured EPCs from healthy volunteers (n = 4) with pioglitazone (10 μmol/l, 24 h) decreased basal NADPH oxidase activity from 40 ± 3.29 to 16 ± 8.34 relative light units (RLU)/μg. Similarly, TZD was able to prevent the PMA-induced increase of NADPH oxidase activity (from 72 ± 5.14 to 44 ± 5.45 RLU/μg) (Fig. 5B).

The RT-PCR analysis showed that PPARγ is expressed in human EPCs. To test whether upregulation of EPCs is mediated via PPARγ, the EPCs were treated with pioglitazone (10 μmol/l 24 h) alone and in the presence of the PPARγ antagonist GW9662 (1 μmol/l, 24 h). TZD treatment increased EPC numbers to 189 ± 15% (P < 0.05). Figure 6A shows that the increase of EPCs in the presence of TZD is completely inhibited by the PPARγ antagonist (n = 4; P < 0.05). In vitro treatment with pioglitazone (10 μmol/l, 24 h) increased the migratory capacity per number of EPCs to 158 ± 10%; the effect was completely prevented by cotreatment with GW9662 (1 μmol/l, 24 h). The upregulation of EPC numbers and the improvement of migratory function by TZD were mimicked by treatment of the cultured EPCs with adiponectin at a concentration similar to that observed in patients after TZD treatment (20 μg/ml, 24 h) (Fig. 6B).

The main finding of this prospective, randomized trial is the demonstration of a novel vascular effect of TZDs in normoglycemic individuals with coronary artery disease. Treatment with pioglitazone increased the number and the function of circulating EPCs. The effects of TZD on EPCs occurred on top of the treatment with aspirin, β-blockers, and inhibitors of the renin-angiotensin system and, importantly, in the presence of statin treatment.

To evaluate the adherence to the study medication, serum adiponectin concentrations were measured. Adiponectin has been shown to be increased by TZD treatment mediated by an effect on adipocytes contributing to improved insulin sensitivity (1,37). Adiponectin levels at least doubled in all of the pioglitazone-treated individuals, but no placebo-administered patient exhibited a similar increase. These findings suggest a very good compliance. To test whether adiponectin contributes to the regulation of EPCs by TZD, cultured human EPCs were exposed to adiponectin at concentrations similar to those observed in the patients taking TZD. Adiponectin potently increased both EPC numbers and function, suggesting that adiponectin contributes as a mediator to the effects of TZD on EPCs.

In agreement with previous observations, TZD treatment significantly lowered the established marker of vascular inflammation hsCRP (8,41). Our data show that this effect can be observed in normoglycemic patients with coronary artery disease after 30 days of treatment. These data are in agreement with recent data showing that treatment of isolated EPCs with recombinant CRP inhibits EPC differentiation, survival, and function and caused a concentration-dependent increase in reactive oxygen species production and apoptosis. Interestingly, the PPARγ agonist rosiglitazone was able to inhibit the negative effects of CRP on EPC biology (42,43). Most of these effects, however, were observed at a CRP concentration of 15 μg/ml, which is significantly higher than in our study population with mean hsCRP values of ≤5 μg/ml.

Several markers have been used to identify EPCs. Here, CD34⁺/KDR⁺ EPCs were quantified because they have been shown to predict cardiovascular outcomes in patients with coronary artery disease. The recent EPCAD (Endothelial Progenitor Cells Coronary Artery Disease) study showed an association between this population of EPCs and death from cardiovascular causes that is independent of the severity of coronary artery disease, cardiovascular risk factors, and medication known to influence cardiovascular outcomes (20). In addition to FACS analysis, we used a second, independent method of EPC characterization by culturing mononuclear cells and selection of EPCs by endothelial growth factors, adhesion to fibronectin, the ability to uptake LDL, and staining for lectin (11,14). As a third established method, we quantitated EPC CFUs, which have been shown to predict endothelial function in humans (14). All three methods of EPC quantification showed a robust increase of EPC numbers in patients taking the PPARγ agonist.

In addition to EPC numbers, functional properties of EPCs have been shown to determine cardiovascular disease (14,20). Recent data suggest that EPC function may be impaired by cardiovascular risk factors independent of the EPC number (26,27). In mice and in cultured human EPCs, pioglitazone prevents apoptotic cell death of EPCs by a mechanism involving phosphatidylinositol 3-kinase (34). Intracellular reactive oxygen species, such as superoxide and H₂O₂, impair the function of mature and immature endothelial cells, e.g., by promoting cell death (16,39,40). Of the several sources within vascular cells, the multi-subunit NADPH oxidase is a predominant contributor of endothelial superoxide free radicals (38). Here, cell culture experiments show that TZD treatment reduces basal and PMA-stimulated NADPH oxidase activity in EPCs, pinpointing a mechanism by which TZD treatment reduces EPC apoptosis and increases EPC numbers. Two widely studied functional characteristics of EPCs are their potential to migrate and their capacity to replicate. These features are likely to be of great importance for the improvement of endothelial function, neoangiogenesis, and inhibition of atherogenesis (44). The data show that TZD treatment increased both the migratory and the colony forming capacity per number of EPCs in patients with normal glucose tolerance, suggesting that the biological effect of TZD on EPCs may be significantly greater than the extent of the increase of EPC numbers. These effects are mediated by PPARγ because GW9662 reversed the effects of pioglitazone on EPC numbers and migration.

Cardiovascular disease accounts for ∼70% of mortality in patients with diabetes. Prospective studies show that compared with their nondiabetic counterparts, the relative risk of cardiovascular mortality for men with diabetes is two to three and for women with diabetes is three to four. Several large trials have demonstrated that optimal control of blood pressure and LDL cholesterol level can substantially reduce excess cardiovascular risk in patients with diabetes (45,46). Metabolic control alone is not sufficient for the prevention of cardiovascular events in patients with diabetes (47). However, even with optimal control of the potent cardiovascular risk factors blood pressure and LDL cholesterol, incremental risk for cardiovascular events remains high compared with individuals without diabetes (45,46). Beneficial effects of TZDs on vascular inflammation and function seem to be possibly independent of glucose lowering and have been demonstrated in nondiabetic, healthy individuals (2,4,48,49). EPCs are independent predictors of endothelial function, cardiovascular events, and cardiovascular mortality (14,20,21). Patients with diabetes are characterized by an impairment of EPC function that is observed independently of other cardiovascular risk factors (22,25,27). On the basis of our study, it is interesting to speculate that patients with diabetes may benefit from PPARγ agonists in addition to insulin sensitization by upregulation of EPCs. This needs to be confirmed in further trials.

Here, we show that treatment with pioglitazone powerfully improves number and function of EPCs in patients with vascular disease and normal glucose tolerance. Serum glucose and lipid concentrations were not affected by treatment with pioglitazone. TZD treatment reduced insulin concentrations and the HOMA index, suggesting that TZDs exert metabolic effects that occur in normoglycemic individuals and that do not result in significant changes of serum glucose (such as regulation of adiponectin and insulin). The effects of TZDs in mice and in isolated cultured EPCs suggest that some of these effects are effective on the level of progenitor cells. Until now, TZDs are only approved and used for individuals with elevated serum glucose levels. These clinical data viewed together with preclinical studies therefore evoke the provocative hypothesis that these agents may be also beneficial for patients with vascular disease despite normal glucose tolerance, such as normoglycemic individuals with stable coronary artery disease.

The design of the study has limitations. It does not allow answering the question of whether TZD therapy confers a prognostic benefit to nondiabetic coronary artery disease patients reducing cardiovascular events at long-term follow-up. Furthermore, subgroups of patients that may especially benefit from TZD treatment and the kinetics and time course of the effects of TZD on EPCs are not known. Finally, the intracellular signaling pathways mediating the effect of TZD on number and function of EPCs remain to be elucidated. These issues need to be addressed in further studies. However, this is a proof-of-concept study providing the first evidence for a new mechanism of TZD action and setting the stage for large outcome trials using risk factor stratification by EPC numbers and function.

In summary, EPCs have emerged as a new dimension of vascular biology. Increasing evidence suggests that bone marrow–derived adult stem cells significantly contribute to vascular and cardiac function. Improvement of EPC function may represent a novel and relevant protective mechanism of TZDs that could potentially benefit patients with vascular diseases in the absence of manifest diabetes.

U.L. has received support from the Deutsche Forschungsgemeinschaft. This work has received support from Universität des Saarlandes. The Klinik für Innere Medizin III, Universitätsklinikum des Saarlandes, has received an unrestricted grant from Takeda (Aachen, Germany).

We thank Simone Jäger and Ellen Becker for their excellent technical assistance.