Priming of late endothelial progenitor cells with erythropoietin before transplantation requires the CD131 receptor subunit and enhances their angiogenic potential


Youssef Bennis, Laboratoire de Pharmacodynamie, UMR INSERM 1076, Faculté de Pharmacie de Marseille, 27, Boulevard Jean Moulin, 13385 Marseille, CEDEX 5, France.
Tel.:+33 4 91 83 56 41; fax: +33 4 91 25 50 36.


Summary.  Background:  Endothelial colony-forming cells (ECFCs) are promising candidates for cell therapy of ischemic diseases. Erythropoietin (EPO) is a cytokine that promotes angiogenesis after ischemic injury. EPO receptors (EPORs) classically include two EPOR subunits, but may also associate with the β-common chain (CD131) in a newly identified receptor involved in EPO cytoprotective effects.

Objective:  The aim was to take advantage of the proangiogenic properties of EPO to enhance ECFC graft efficiency. We postulated that priming ECFCs by adding epoietin α in culture medium prior to experiments might increase their angiogenic properties. We also explored the role of the CD131 subunit in EPO priming of ECFCs.

Methods and Results:  By western blotting on cord blood ECFC lysates, we showed that EPOR and CD131 expression increased significantly after EPO priming. These proteins coimmunoprecipitated and colocalized, suggesting that they are covalently bound in ECFCs. EPO at 5 IU mL−1 significantly stimulated proliferation, wound healing, migration and tube formation of ECFCs. EPO priming also increased ECFC resistance to H2O2-induced apoptosis and survival in vivo. Similarly, in vivo studies showed that, as compared with non-primed ECFC injection, 5 IU mL−1 EPO-primed ECFCs, injected intravenously 24 h after hindlimb ischemia in athymic nude mice, increased the ischemic/non-ischemic ratios of hindlimb blood flow and capillary density. These effects were all prevented by CD131 small interfering RNA transfection, and involved the phosphoinositide 3-kinase–Akt pathway.

Conclusion:  These results highlight the potential role of EPO-primed ECFCs for cell-based therapy in hindlimb ischemia, and underline the critical role of CD131 as an EPO coreceptor.


The prognosis of patients with severe peripheral artery disease remains poor when there is no indication for revascularization therapies, and amputation is necessary in more than one-third of patients at the stage of critical limb ischemia [1]. Hence, new cell-based therapies are being developed to stimulate revascularization of ischemic tissues. Recent clinical trials have illustrated the potential of autologous bone marrow (BM)-derived or peripheral blood-derived stem cell preparations to improve blood flow in patients with critical limb ischemia [2–4]. Although clinical improvements were observed, results have been variable, and efficacy remains to be proved. However, the BM preparations currently being used contain mainly hematopoietic and mesenchymal stem cells, whereas endothelial progenitor cells (EPCs) are only slightly represented. This has prompted the development of an alternative cell therapy product that could be enriched in progenitor cells displaying vasculogenic properties and isolated from a less invasive source. In cord blood, endothelial colony-forming cells (ECFCs) have been characterized as a homogeneous cell population of non-hematopoietic origin, obtained after long-term culture [5,6]. They are considered to be relevant EPCs because of their specific vasculogenic activity in vivo [7]. ECFCs are considerably enriched in cord blood as compared with adult peripheral blood, and cord blood ECFCs show a higher clonogenic potential, related to cord blood cell immaturity [8]. These potentialities have been explored by the use of mouse models of hindlimb ischemia, where ECFCs promote recovery of tissue perfusion, angiogenesis, and muscular regeneration [7,9]. Although promising, ECFC-based strategies have several limitations, with respect to the low proportion of engrafted cells observed in preclinical models, the time required for isolation and expansion of a sufficient number of cells from blood, and the possible impairment of clonogenic and functional properties of circulating EPCs in patients presenting with cardiovascular risk factors. Therefore, specific cell ‘enhancement strategies’ aimed at improving the efficacy of cell-based proangiogenic therapies have been explored, and among them is ex vivo priming of the cell candidate before transplantation by proangiogenic factors [10,11].

Erythropoietin (EPO) is recognized as the main regulator of erythropoiesis, but recent studies have suggested the existence of non-hematopoietic effects of EPO [12]. In adults, EPO contributes to physiologic and pathologic angiogenesis, and the body’s innate response to tissue injury and, especially, to ischemic injury [12–14]. The tissue-protective effects of EPO are likely, at least in part, to be carried out through induction of angiogenesis and antiapoptotic properties [14]. Accordingly, EPO induces a proangiogenic response in cultivated mature endothelial cells, and mobilizes BM-derived early EPCs and recruits them to ischemic tissue, where they stimulate reparative angiogenesis [15,16]. Also, antiapoptotic effects of EPO were reported in vitro in nerve cell cultures subjected to excitotoxic stress and in vivo after experimental cerebral ischemia [14]. In this context, we hypothesized that priming of ECFCs with EPO before transplantation could enhance their angiogenic potential after hindlimb ischemia. In this way, Zhang et al. [17] showed improved cardiac function in animals subjected to myocardial infarction and injected with EPO-primed BM stromal cells. Similarly, Oda et al. [18] reported an enhancement of the angiogenic potential of human mononuclear BM cells cultured in the presence of EPO and then transplanted in a mouse hindlimb ischemia model and in patients with chronic lower limb ischemia. At the molecular level, the tissue-protective effects of EPO seem to be mediated by the association of the known EPO receptor (EPOR) subunit involved in the hematopoietic effects of EPO with the CD131 subunit, also called β-common chain receptor, which is a constitutive part of cytokine receptors such as interleukin (IL)-3, IL-5, and granulocyte–macrophage colony-stimulating factor (GM-CSF) [19,20]. This hypothesis is supported by the fact that CD131 knockout mice are unable to respond to EPO’s tissue-protective effects [19]. On this basis, using RNA interference targeting CD131, we explored the role of CD131 in the priming effect of EPO on ECFC activity. This, this study was aims at evaluating: (i) the effect of ex vivo priming of cord blood-derived ECFCs by EPO on the functional properties of the cells; (ii) the angiogenic potential of EPO-primed ECFCs transplanted in a model of hindlimb ischemia in nude mice; and (iii) the impact of CD131 silencing on the priming effect of EPO.

Materials and methods

ECFC preparation and EPO priming

Human umbilical cord blood samples (30–50 mL) from healthy donors were collected, in compliance with French legislation, and ECFCs were isolated and characterized as previously described [21,22]. EPO priming consisted of a 24-h incubation with human recombinant EPO (10 000 IU mL−1 epoietin α, Eprex; Janssen-Cilag, Issy-les-Moulineaux, France) diluted in EBM-2 (Clonetics Bio, Emerainville, France) supplemented with 0.5% fetal bovine serum (FBS) to a final concentration of 1.5–10 IU mL−1. After extensive washing, cells were used in subsequent assays. In blocking experiments, antibody against the extracellular domain of the EPOR subunit (H194, 1 : 200; Santa Cruz Biotechnology, Santa Cruz, CA, USA) was added during EPO incubation. Cells maintained in 0.5% FBS/EBM-2 were used as control cells. In several experiments, cells maintained in EGM-2 (EBM-2 enriched with growth factors and vitaminsp Clonetics-Bio) were used as positive controls.


ECFC lysates (1 mg protein mL–1) were incubated overnight at 4 °C with protein G Sepharose (Amersham Biosciences, Piscataway, NJ, USA) previously coupled either to a rabbit anti-CD131 antibody (K17, 1 : 200; Santa Cruz Biotechnology), a mixture of two rabbit anti-EPOR antibodies (M20 and H194, 1 : 200; Santa Cruz Biotechnology), or a rabbit isotype control antibody (1 : 200; Santa Cruz Biotechnology). The protein G Sepharose beads were washed with low-detergent lysis buffer, and western blotting was performed.

Membrane protein extraction

Membrane extracts were prepared with the Qproteome Plasma Membrane Protein Kit (Qiagen, Valencia, CA, USA) according to the manufacturer’s protocol. Briefly, sequential addition of different buffers to ECFC pellets followed by incubation with a ligand specific for membrane molecules and precipitation with magnetic beads that bind to the ligand resulted in the isolation of the membrane compartment. After washing, plasma membrane proteins were eluted under native conditions, and the ligand remained bound to the beads. Fractions were then subjected to western blotting.

Western blotting

ECFC lysates or protein G Sepharose-immunoprecipitated beads were applied to the NuPAGE system (Invitrogen, Carlsbad, CA, USA), loaded for electrophoresis on a 4–12% gradient polyacrylamide gel, and transferred onto nitrocellulose membranes. Membranes were then blocked and probed overnight at 4 °C with primary antibody. Membranes were washed and incubated with anti-rabbit or anti-mouse peroxidase-conjugated secondary antibody (1 : 10 000; Sigma-Aldrich, Lyon, France). Antibody detection was accomplished with the ECL Plus system (Amersham Biosciences), and protein bands were quantified with a gel image analysis system. The primary antibodies used were rabbit anti-CD131 (K17, 1 : 200; Santa Cruz Biotechnology), rabbit anti-EPOR (M20, 1 : 200; Santa Cruz Biotechnology), mouse anti-Akt (1 : 500; Invitrogen), rabbit anti-S473 phospho-Akt (pAkt, 1 : 500; Invitrogen), and mouse anti-human β-actin (1 : 5000; Sigma-Aldrich). Band densities were represented as an index of EPOR or CD131 to β-actin or pAkt corrected by β-actin to Akt corrected by β-actin.

EPOR and CD131 immunostaining

Confluent ECFCs, seeded on coverslips, were fixed with a 1 : 1 methanol and acetone mixture at − 20 °C for 10 min and blocked with phosphate-buffered saline (PBS)/FBS 10% (v/v) before EPOR and CD131 staining, respectively, with rabbit antibody against CD131 (K17, 1 : 100; Santa Cruz Biotech, Heidelberg, Germany) or mouse monoclonal anti-EPOR fluorescein-conjugated antibody (1 : 100; R&D Systems, Lille, France) overnight in blocking solution at 4 °C in PBS. Cells were washed, and secondary anti-rabbit Alexa Fluor 568-conjugated antibody (1 : 200; Sigma-Aldrich, Saint Quentin, France) was added. Images were taken with a fluorescence microscope (Leica, Bron, France). 4′,6-Diamidino-2-phenylindole (1:1000; Sigma-Aldrich) was used to visualize nuclei.

Proliferation, migration and tube formation assays

ECFC proliferation was measured by fluorescence labeling and a bromodeoxyuridine (BrdU) incorporation assay. The migration capacity of ECFCs was evaluated by both analysis of stromal cell-derived factor-1 (SDF-1)-induced chemotaxis in a modified Boyden chamber and wound healing experiments. The in vitro angiogenic activity of ECFCs was tested with a capillary-like tube formation assay in Matrigel. All of these experiments are described in detail in Data S1.

Survival assays

The capacity of EPO priming to protect ECFCs against H2O2-induced cytotoxicity was analyzed with cell death ELISA and lactate dehydrogenase (LDH) release assays, as described in Data S1.

The survival of ECFCs was also evaluated in vivo with bioluminescence studies. For this, ECFCs were infected with pCignalLenti-CMV-luc vector (Cignal Lentireporter; SA Biosciences, Courtaboeuf, France), according to the manufacturer’s protocol. The transduction was stabilized by the addition of puromycin during the culture, and bioluminescence emission was evaluated with a Photon Imager Imaging System (Biospace Lab, Paris, France). A total of 1 × 105 cells were resuspended in 400 μL of ice-cold Phenol Red-free Matrigel (BD Bioscience, Le Pont de Claix, France), and implanted on the back of an 8-week-old male NMRI nude mouse (Harlan, Gannat, France) by subcutaneous injection with a 25-gauge needle. Implants of Matrigel with EPO-pretreated ECFCs or ECFCS were injected, respectively, on the right and left side of the mouse’s back (n = 5). Bioluminescence emission was collected at days 0, 1, 3 and 6 after implantation with a Photon Imager Imaging System (Biospace Lab) and intraplug luciferin (Promega) administration. The collected data were analyzed with m3 vision (Biospace Lab), and results were expressed as a mean percentage of the bioluminescence emission (photons s–1 cm–2 sr–1) of the contralateral control Matrigel plug. Results are reported as mean ± standard deviation (SD).

Vascular endothelial growth factor (VEGF) assays

VEGF concentrations in supernatants of ECFCs treated or not treated with 10 IU mL−1 EPO were determined with a commercially available ELISA kit (Quantikine Human VEGF; R&D Systems), according to the manufacturer’s instructions. Values were normalized to cell number (pg 10–5 cells), and are expressed as mean ± SD.

Small interfering RNA (siRNA) transfection protocol

The transfection was performed with a Nucleofector electroporation system, following the manufacturer’s protocol for human umbilical vein endothelial cells (HUVECs) (Amaxa). Briefly, 2 × 106 ECFCs were suspended in 100 μL of HUVEC Nucleofector Solution. The cell suspension was transfected with 20 μL of 10 μm (i.e. 2.5 μg) CD131 siRNA or control siRNA (Santa Cruz Biotechnology). To determine the optimal conditions, CD131 extinction was analyzed by western blotting after incubation with 5, 10 and 20 μL of 10 μm siRNA for 48, 72 or 96 h. Maximal CD131 extinction was observed at 72 h with 20 μL of siRNA (data not shown). There was no significant inhibition of CD131 expression in control siRNA-transfected ECFCs (Fig. 4A). All functional assays with CD131 siRNA were then performed 72 h after transfection with 20 μL of 10 μm siRNA.

Mouse model of hindlimb ischemia

This study was approved by the local ethics committee, and was conducted according to the the recommendations of the Helsinki Declaration. Unilateral hindlimb ischemia was performed after femoral artery ligation as previously described [10]. Twenty-four hours later, mice were randomly allocated to receive intravenous injections of PBS, 105 ECFCs or 105 ECFCs incubated for 24 h with 5 IU mL−1 EPO (EPO-ECFCs). For siRNA silencing experiments, two additional groups of mice were studied: those injected with ECFCs with siRNA silencing CD131 (siRNACD131 ECFCs) and those injected with siRNACD131 ECFCs primed with EPO. Hindlimb ischemic damage was quantified on day 14, as previously described [23]. The score was calculated for each mouse as follows: 0, no necrosis; 1, necrosis of one toe; 2, necrosis of two or more toes; 3, foot necrosis; 4, leg necrosis; and 5, autoamputation of the entire leg. Laser Doppler perfusion imaging (Perimed, Craponne, France) was used to assess revascularization from day 1 to day 28 after surgery. Perfusion results are expressed as a ratio of ischemic to non-ischemic limb blood flow. Fourteen days after cell transplantation or PBS injection, capillary density was assessed by immunohistochemistry on the gastrocnemius muscle, as previously described [10]. Capillaries and muscle fibers were counted in six different fields of each section. The capillary density of each muscle was expressed as the ratio of capillary number to muscular fiber number, in order to take into account muscle trophicity.

Statistical analysis

Data are expressed as means ± SD of at least three independent experiments. Significant differences were identified with anova followed by the Student–Newman–Keuls multiple comparisons test. Clinical necrosis score was analyzed with the Kruskal–Wallis test followed by the Mann–Whitney U–test. Statistical analysis was performed with SigmaStat 2.03 (SPSS, Chicago, IL, USA). A P-value of < 0.05 was considered to be statistically significant.


EPOR and CD131 are expressed on ECFCs and upregulated by EPO priming

Western blot experiments revealed an EPOR-identified 59-kDa protein in whole cell lysates of control ECFCs, EPO-ECFCs, and UT-7 cells (Data S1). Expression levels of EPOR and CD131 in total cell lysates (Fig. 1C,F) were significantly increased in EPO-ECFCs (% of ECFCs: 227% ± 59% and 152% ± 19%, respectively, n = 3, < 0.05), whereas no significant increase in expression levels was measured in plasma membrane fractions (% of ECFCs: 127% ± 26% and 103% ± 14%, respectively, n = 3, not significant [NS]) (Fig. 1B,E). Immunostaining showed that EPOR and CD131 colocalized in ECFCs and EPO-ECFCs (Fig. 2A). Association of the two subunits in both ECFCs and EPO-ECFCs was confirmed with coimmunoprecipitation experiments (Fig. 2B).

Figure 1.

 Expression of erythropoietin (EPO) receptor (EPOR) and CD131 subunits in endothelial colony-forming cells (ECFCs). Representative entire immunoblots from individual samples showed expression of the 59-kDa EPOR subunit with the M20 EPOR antibody (A) and of the 130-kDa CD131 subunit with the K17 CD131 antibody (B) in both the plasma membrane (PM) and intracellular (IC) fractions. Densitometric analysis of signals indicated that expression levels of EPOR and CD131 in the PM fraction ([B] and [E], respectively) were not significantly modified after EPO priming (5 IU mL−1, 24 h), whereas they were significantly increased in total cell lysates ([C] and [F], respectively). Data are corrected by β-actin densities (not shown), and expressed as percentage of ECFCs (means ± standard deviation, n = 3). *< 0.05 vs. ECFCs. WB, western blotting.

Figure 2.

 Association of erythropoietin (EPO) receptor (EPOR) and CD131 subunits. Immunochemistry and coimmunoprecipitation assays indicated that EPOR and CD131 could be bound in endothelial colony-forming cells (ECFCs). (A) EPOR and CD131 were detected with a fluorescein-conjugated mouse monoclonal EPOR antibody (green) and a rabbit polyclonal CD131 antibody recognized by an Alexa Fluor-568-conjugated secondary anti-rabbit antibody (red). White arrows show colocalization of EPOR and CD131 in ECFCs and ECFCs incubated for 24 h with EPO (EPO-ECFCs). (B) Western blotting with the CD131 antibody showed that CD131 immunoprecipitated with EPOR antibody-bound G-Sepharose beads, suggesting that EPOR and CD131 could be covalently associated in ECFC and EPO-ECFC lysates. IP, immunoprecipitation; WB, western blotting.

EPO increases ECFC proliferation and survival in vitro

EPO concentration-dependently increased the number of fluorescent ECFCs in cultures. The effect was significantly higher 24, 40 and 48 h after priming with 5 IU mL−1 EPO than in control ECFCs (% of control ECFCs: 196% ± 7% vs. 158% ± 7%, 227% ± 9% vs. 179% ± 6%, and 243% ± 8% vs. 197% ± 7%, respectively, < 0.01) (Fig. 3A). EPO concentration-dependently increased BrdU incorporation, with a maximal response with 5 IU mL−1 (% of control ECFCs: 138% ± 14%, < 0.01) (Fig. 3B). This effect was abolished in the presence of EPOR antibody (% of control ECFCs: 107% ± 12%, NS) (Fig. 3B).

Figure 3.

 Erythropoietin (EPO) increased endothelial colony-forming cell (ECFC) proliferation and survival after oxidative stress in vitro. EPO priming dose-dependently increased proliferation of ECFCs as measured by fluorescence counting (A) and bromodeoxyuridine (BrdU) incorporation (B). Data are expressed as percentage of control (means ± standard deviation [SD], n = 3–4). H2O2-induced lactate dehydrogenase (LDH) release (C) and DNA fragmentation (D) of ECFCs in culture were reduced by EPO priming. Data are expressed as perentage of H2O2 control (means ± SD, n = 4). EPO effects were maximal with 5 IU mL−1 EPO, and were significantly prevented in the presence of EPOR antibody (Ab). Growth factor-supplemented medium (EGM-2) was used as a positive control. EPO 1.5, ECFCs incubated with 1.5 IU.mL−1 EPO; EPO 5, ECFCs incubated with 5 IU.mL−1 EPO; EPO 10, ECFCs incubated with 10 IU.mL−1 EPO; EPO 5 + Ab, ECFCs incubated with 5 IU.mL−1 EPO and EPOR antibody. *< 0.05 vs. control; #< 0.05 vs. EPO 5.

After EPO priming, LDH release from H2O2-exposed ECFCs was concentration-dependently reduced as compared with non-primed control cells, with a higher effect being observed with 5 IU mL−1 EPO (% of H2O2 control ECFCs: 68% ± 6%, < 0.001) (Fig. 3C). A similar reduction was observed when cells were cultured in the presence of growth factor-containing medium and used as a positive control, before H2O2 challenge. The protective effect of 10 IU mL−1 EPO was less marked, although significant (% of H2O2 control ECFCs: 77% ± 3%, P < 0.01), but at 1.5 IU mL−1 the impact of EPO was not significant. The reduction in H2O2-induced LDH release by ECFCs was abolished when cells were primed with EPO in the presence of EPOR antibody (% of H2O2 control ECFCs: 81% ± 6%, NS) (Fig. 3C,D). Similar results were obtained with measurement of DNA fragmentation. Priming of ECFCs by EPO dose-dependently reduced H2O2-induced DNA fragmentation (% of H2O2 control ECFCs: 71% ± 2%, < 0.001, and 78% ± 2%, < 0.01, respectively, with 5 and 10 IU mL−1).

Finally, we tested ECFC survival in vivo when cells in Matrigel were subcutaneously transplanted into mice (Fig. 4A). No difference in total bioluminescence emission was observed between EPO-ECFCs and ECFCs before cell inclusion in Matrigel (data not shown). We observed a significant increase in total bioluminescence emission 3 and 6 days after implantation of EPO-ECFC-containing Matrigel as compared with ECFC-containing Matrigel (Fig. 4B) (% of day 0: day 3, 8.3% vs. 6.7%, and day 6, 1.04% vs. 0.7%, respectively, n = 5, < 0.05). No statistical difference was observed on day 1 and day 8 (% of day 0: day 1, 643% vs. 625%, and day 8, 0.99% vs. 0.52%, n = 5, < 0.05).

Figure 4.

 Erythropoietin (EPO) enhanced endothelial colony-forming cell (ECFC) survival in the Matrigel plug assay. (A) Representative in vivo bioluminescence imaging of ECFCs in transplanted Matrigel. (B). Box plot with median (horizontal line), interquartile range (box) and minimum and maximum values (whiskers) represents bioluminescence emission at various time points. Values are quantified as photons s–1 cm–2 sr–1, and are expressed as percentage of day (D)0. We observed a significant increase in bioluminescence emission in the ECFCs incubated for 24 h with EPO (EPO-ECFCs) (white box) on D3 and D6 as compared with the ECFC group (gray box) (n = 5 or n = 6 for each group). *< 0.05 (Wilcoxon test: ECFCs vs. EPO-ECFCs).

EPO increases ECFC angiogenic properties in vitro

EPO-ECFCs displayed higher migratory activity than control ECFCs, filling the gap created by the scratch within 8 h in wound healing assays (Fig. 5Aa–d). This effect was concentration-dependent, and the maximal effect was observed with 5 IU mL−1 (% of control ECFCs: 148% ± 12%, < 0.01) (Fig. 5B). Priming of ECFCs with EPO also enhanced SDF-1-induced ECFC migration measured in the Boyden chamber assay, as compared with control ECFCs (Fig. 5C; representative micrographs are shown in Fig. 5Ae,f). The maximal effect was observed with 5 IU mL−1 EPO (% of control ECFCs: 158% ± 17%, < 0.05), similarlt to what was found when cells were cultured in positive control medium. The EPO-induced increase in migratory features was prevented by EPOR antibody (% of control ECFCs: 112% ± 12%, NS, and 118 %± 26%, NS) (Fig. 5B,C).

Figure 5.

 Erythropoietin (EPO) priming increased endothelial colony-forming cell (ECFC) angiogenic properties in vitro. EPO priming increased wound recovery as compared with control ECFCs (control, t0 h [Aa] and t8 h [Ac]; EPO, t0 h [Ab] and t8 h [Ad]). EPO priming increased migration in transwell chamber ([Ae] control; [Af] EPO) and improved tube formation in Matrigel ([Ag] control; [Ah] EPO). EPO effects were significant and maximal with 5 IU mL−1 EPO for wound healing (B) and migration assays (C) and with 10 IU mL−1 EPO for tube formation in Matrigel (D). These effects were significantly reduced in the presence of EPO receptor antibody (Ab) (B–D). Data are expressed as percentage of control (means ± SD, n = 4–5). No significant modification of the vascular endothelial growth factor (VEGF) concentration in the culture medium occurred after EPO priming of ECFCs (E). Growth factor-supplemented medium (EGM-2) was used as a positive control. EPO 1.5, ECFCs incubated with 1.5 IU.mL−1 EPO; EPO 5, ECFCs incubated with 5 IU.mL−1 EPO; EPO 10, ECFCs incubated with 10 IU.mL−1 EPO; EPO 5 + Ab, ECFCs incubated with 5 IU.mL−1 EPO and EPOR antibody. *< 0.05 vs. control; #< 0.05 vs. EPO 5; $< 0.05 vs. EPO 10.

Treatment with EPO concentration-dependently promoted tube formation by ECFCs, as attested by the significant increase in network length relative to controls (118% ± 3%, < 0.01, and 142% ± 12%, < 0.01, for EPO used at 5 and 10 IU mL−1, respectively) (Fig. 5D). The capillary network formed with 10 IU mL−1 was similar to that observed with cells cultured in positive control medium. No significant increase was observed when ECFCs were treated with 10 IU mL−1 EPO in the presence of EPOR antibody (Fig. 5D).

As shown in Fig. 5E, no significant change in VEGF levels could be detected in the medium of ECFCs treated with doses of EPO ranging from 1.5 to 10 IU mL−1 (25.8 ± 2.6 pg mL−1, 23.9 ± 1.9 pg mL−1, and 24.5 ± 2.3 pg mL−1, respectively for 1.5, 5 and 10 IU mL−1) as compared with control ECFCs (25.8 ± 2.6 pg mL−1). Consistently, EPO failed to increase VEGF mRNA levels in ECFCs (data not shown).

RNA interference targeting CD131 prevents the effects of EPO priming on ECFCs in vitro

CD131 protein remained strongly inhibited by the specific CD131 siRNA (siRNACD131 ECFCs) for 72 h following transfection in ECFCs as compared with control siRNA (siRNAc ECFCs) (% of control ECFCs: 33% ± 9%, < 0.01) (Fig. 6).

Figure 6.

 Transfection of CD131 small interfering RNA (siRNA) in endothelial colony-forming cells (ECFCs) significantly reduced CD131 expression. Western blotting performed 72 h after transfection of siRNA in ECFCs revealed that expression of CD131 was significantly decreased in ECFCs transfected with CD131 siRNA (siRNACD131 ECFCs) as compared with ECFCs transfected with control siRNA (siRNAc ECFCs). The densitometric data are corrected by β-actin levels and expressed as percentage of siRNAc ECFCs (means ± standard deviation, n = 4). *< 0.05 vs. siRNAc ECFCs. A representative immunoblot analysis is shown in the upper panel.

As compared with siRNAc ECFCs, inhibition of CD131 expression by CD131 siRNA significantly prevented the effect of 5 IU mL−1 EPO on ECFC proliferation (siRNACD131 EPO-ECFCs vs. siRNAc EPO-ECFCs), as demonstrated by measurement of BrdU incorporation (% of control ECFCs: siRNACD131 EPO-ECFCs, 105% ± 10% vs. siRNAc EPO-ECFCs, 143% ± 22%, < 0.05) (Fig. 7A) and on ECFC survival assessed by H2O2-induced LDH release (% of H2O2 control ECFCs: siRNACD131 EPO-ECFCs, 94% ± 23% vs. siRNAc EPO-ECFCs, 72% ± 17%, < 0.05) (Fig. 7B) or DNA fragmentation (% H2O2 of control ECFCs: siRNACD131 EPO-ECFCs, 95% ± 22% vs. siRNAc EPO-ECFCs, 74% ± 16%, < 0.05) (Fig. 7C).

Figure 7.

 RNA interference targeting CD131 reversed the proliferative, antiapoptotic and angiogenic effects of erythropoietin (EPO) priming on endothelial colony-forming cells (ECFCs) in vitro. Enhancement of bromodeoxyuridine (BrdU) incorporation after EPO priming of ECFCs with control small interfering RNA (siRNA) (siRNAc ECFCs) (shaded bars) was prevented in ECFCs with siRNA silencing CD131 (siRNACD131 ECFCs) (black bars) (A). Data are expressed as percentage of control (means ± SD, n = 4). EPO protective effects against H2O2 in siRNAC ECFC (shaded bars) were also abolished in siRNACD131 ECFCs (black bars) (B and C). Data are expressed as percentage of H2O2 control (means ± standard deviation [SD], n = 4). Enhancement of area recovery (D), transwell migration (E) and tube formation in Matrigel (F) after EPO priming of siRNAc ECFCs (shaded bars) was also prevented in siRNACD131 ECFCs (black bars). Growth factor-supplemented medium (EGM-2) was used as a positive control. Data are expressed as percentage of control (means ± SD, n = 4–5). EPO 1.5, ECFCs incubated with 1.5 IU.mL−1 EPO; EPO 5, ECFCs incubated with 5 IU.mL−1 EPO; EPO 10, ECFCs incubated with 10 IU.mL−1 EPO; EPO 5 + Ab, ECFCs incubated with 5 IU.mL−1 EPO and EPOR antibody. *< 0.05 vs. control; #< 0.05 vs. EPO 5; $< 0.05 vs. EPO 10. Ab, antibody; LDH, lactate dehydrogenase.

Similarly, silencing of CD131 significantly abrogated: (i) the effect of 5 IU mL−1 EPO on ECFC migration measured in the wound healing assay (% of control ECFCs: siRNACD131 EPO-ECFCs, 118% ± 34% vs. siRNAc EPO-ECFCs, 171% ± 45%, < 0.05) (Fig. 7D) and in the modified Boyden chamber assay (% of control ECFCs: siRNACD131 EPO-ECFCs, 124% ± 12% vs. siRNAc EPO-ECFCs, 154% ± 28%, < 0.05) (Fig. 7E); and (ii) the effect of 5 IU mL−1 EPO (% of control ECFCs: siRNACD131 EPO-ECFCs, 120% ± 18% vs. siRNAc EPO-ECFCs, 139% ± 39%, < 0.05) and 10 IU mL−1 EPO (% of control ECFCs: siRNACD131 EPO-ECFCs, 112% ± 27% vs. siRNAc EPO-ECFCs, 175% ± 46%, < 0.05) on cell differentiation into vascular tubes in Matrigel (Fig. 7F).

EPO priming enhances ECFC angiogenic potential in vivo in a murine model of hindlimb ischemia, and CD131 siRNA silencing impairs EPO-induced angiogenesis

On day 14 postsurgery, the clinical necrosis score was significantly lower in the siRNAc EPO-ECFC group than in the PBS group (respectively, 0.55 ± 0.12 vs. 1.9 ± 0.06, < 0.05), whereas the clinical necrosis score was not significantly different between the siRNAc ECFC and PBS groups (respectively, 1.4 ± 0.18 vs. 1.9 ± 0.06, NS) (Fig. 8A). The clinical necrosis score was significantly higher in the siRNACD131 EPO-ECFC group than in the siRNAc EPO-ECFC group (1.00 ± 0.18 vs. 0.55 ± 0.12, < 0.05) (Fig. 8A).

Figure 8.

 Erythropoietin (EPO) priming increased endothelial colony-forming cell (ECFC) angiogenic properties in vivo. The clinical necrosis score at day 14 was lower in the siRNAc EPO-ECFC group than in the siRNAc ECFC group and phosphate-buffered saline (PBS) group (A). Data are expressed as mean ± standard deviation (SD) (n = 5–7 mice for each condition); *< 0.05 vs. PBS; #< 0.05 vs. 5 IU mL−1 EPO. Representative hindlimb blood flow measured by laser Doppler perfusion imaging (LDPI) at day 14 in the PBS (Ba), siRNAc ECFC (Bb), siRNAc EPO-ECFC (Bc), siRNACD131 ECFC (Bd) and siRNACD131 EPO-ECFC (Be) groups. Hindlimb reperfusion expressed as ischemic/non-ischemic hindlimb perfusion ratio (Bf) was significantly higher in the siRNAc EPO-ECFC group than in the siRNAc ECFC (days 3, 7, and 14), siRNACD131 ECFC (days 3 and 14) and siRNACD131 EPO-ECFC (days 3 and 14) groups. Data are expressed as mean ± SD (n = 5–7 mice for each condition). Comparisons of LDPI values among the five groups were made by one-way anova followed by the Student–Newman–Keuls multiple comparison procedure; *< 0.05 for siRNAc EPO-ECFCs vs. siRNAc ECFCs; #< 0.05 for siRNAc EPO-ECFCs vs. siRNACD131 ECFCs; $< 0.05 for siRNAc EPO-ECFCs vs. siRNACD131 EPO-ECFCs. Representative muscle vessel immunostaining with antibody against PECAM-1 of the PBS (Ca), siRNAc ECFC (Cb), siRNAc EPO-ECFC (Cc), siRNACD131 ECFC (Cd) and siRNACD131 EPO-ECFC (Ce) groups. Quantification showed that ischemic/non-ischemic capillary density was significantly higher in the siRNAc EPO-ECFC group than in the PBS, siRNAc ECFC and siRNACD131 ECFC groups (Cf). Data are expressed as mean ± SD (n = 5–7 mice for each condition); *< 0.05 vs. PBS; #< 0.05 vs. siRNAc EPO-ECFCs. EPO-ECFCs, ECFCs incubated for 24 h with EPO; siRNAc ECFCs, ECFCs with control siRNA; siRNACD31 ECFCs, ECFCs with siRNA silencing CD131.

Recovery of blood flow in the ischemic hindlimb was significantly increased in the siRNAc ECFC, siRNACD131 ECFC, siRNAc EPO-ECFC and siRNACD131 EPO-ECFC groups as compared with the PBS group from day 3 to day 28 (< 0.05) (Fig. 8B). It was also significantly higher the siRNAc EPO-ECFC group than in the siRNAc ECFC group on day 3 (58.0% ± 6% vs. 42.5% ± 6.2%, < 0.05), day 7 (66.8% ± 7% vs. 50.8% ± 9%, < 0.05), and day 14 (84.8% ± 20% vs. 53.4% ± 17%, < 0.05) (Fig. 8B). Moreover, recovery of blood flow was significantly higher in the siRNAc EPO-ECFC group on day 3 than in the siRNACD131 ECFC group (43.2% ± 9%, < 0.05) and siRNACD131 EPO-ECFC group (41.5% ± 6%, < 0.05), and was higher on day 14 than in the siRNACD131 ECFC group (62.0% ± 11%, < 0.05) and the siRNACD131 EPO-ECFC group (61.0% ± 20%, < 0.05) (Fig. 8B). However, on day 28, there were no statistically significant differences in recovery of blood flow between the siRNAc ECFC, siRNAc EPO-ECFC, siRNACD131 ECFC and siRNACD131 EPO-ECFC groups (89.0% ± 22%, 102% ± 16%, 79.0% ± 20%, and 90.5% ± 18%, respectively).

Fourteen days postsurgery, immunochemical examination of muscle sections from mice treated with siRNAc ECFCs showed a higher density of capillaries than in mice treated with PBS (% of non-ischemic limb: 0.95% ± 0.17% vs. 0.69% ± 0.04%, < 0.05) (Fig. 8C). Moreover, the increase in capillary density was significantly higher in the siRNAc EPO-ECFC group than in the siRNAc ECFC group (% of non-ischemic limb: 1.25% ± 0.08% vs. 0.95% ± 0.17%, < 0.05) (Fig. 8C). The effect was abrogated in the siRNACD131 EPO-ECFC group as compared with the siRNAc EPO-ECFC group (0.96 ± 0.08, < 0.05) (Fig. 8C).

Clinical necrosis score, blood flow recovery and capillary density in the siRNAc ECFC group were not significantly different from those in the control ECFC group (data not shown).

EPO-induced upregulation of the phosphoinositide 3-kinase (PI3K)–Akt pathway required the CD131 subunit

Western blot experiments indicated that the Ser473 pAkt/Akt ratio increased after EPO priming of ECFCs in a concentration-dependent manner. A significant effect was observed at 5 and 10 IU mL−1 EPO as compared with control ECFCs (% of control ECFCs: 230% ± 58% and 395% ± 78% vs. 100% ± 20.4%, respectively, < 0.05) (Fig. 9A). The increase in the Ser473 pAkt/Akt ratio was reduced after 5 IU mL−1 EPO priming of siRNACD131 ECFCs as compared with siRNAc ECFCs (% of siRNAc ECFCs: 96% ± 59% vs. 319% ± 49%, respectively, < 0.01) (Fig. 9B). The increase in the Ser473 pAkt/Akt ratio was also reduced after 5 IU mL−1 EPO priming of ECFCs in the presence of LY294002, a PI3K inhibitor, as compared with EPO-ECFCs (% of ECFCs: 132% ± 18% vs. 263% ± 35%, respectively, < 0.01) (Fig. 9C). Moreover, EPO priming of ECFCs in the presence of LY294002 before transplantation in ischemic mice prevented the EPO-induced improvement in hindlimb reperfusion 14 days after surgery (58% ± 3% vs. 72% ± 2%, respectively, < 0.01) (Fig. 9D).

Figure 9.

 Erythropoietin (EPO)-induced phosphorylation of Akt is dependent on phosphoinositide 3-kinase (PI3K) and requires CD131. Immunoblots from individual samples showed 60-kDa pAkt, 60-kDa Akt and 42-kDa β-actin bands with a rabbit polyclonal antibody against Akt phosphorylated on Ser473, a mouse mAb against Akt, and a mouse mAb against β-actin (top of [A], [B], and [C]). Densitometric analysis of signals indicated that the pAkt/Akt ratio was concentration-dependently increased in ECFCs primed with 5 IU mL−1 EPO (EPO 5-ECFCs) (A, bottom) and that the 5 IU mL−1 EPO-induced increase in the pAkt/Akt ratio was abolished in ECFCs transfected with small interfering RNA (siRNA)CD131 (siRNACD131 ECFCs) (B, bottom). Moreover, EPO priming of ECFCs with LY294002 (LY), a PI3K inhibitor, prevented the EPO-induced increase in the pAkt/Akt ratio (C, bottom) and the improvement in 14-day postischemic hindlimb perfusion recovery after EPO-ECFC transplantation (D). Data are expressed as percentage of control (means ± standard deviation, n = 3). EPO 1.5, ECFCs incubated with 1.5 IU.mL−1 EPO; EPO 5, ECFCs incubated with 5 IU.mL−1 EPO; EPO 10, ECFCs incubated with 10 IU.mL−1 EPO; EPO 5 + Ab, ECFCs incubated with 5 IU.mL−1 EPO and EPOR antibody. *< 0.05 vs. control (A and C) or siRNAc Control-ECFCs (B) or ECFCs (D). #< 0.05 vs. EPO 5 (A and C) or siRNAc EPO 5-ECFCs (B) or EPO 5-ECFCs (D).


To our knowledge, this is the first study documenting the role of EPO as a priming agent for ECFC-based cell therapy. Indeed, we demonstrated that EPO enhanced ECFC proliferation, stress-induced survival and angiogenic properties in vitro. In addition, priming ECFCs with EPO before transplantation in a mouse model of hindlimb ischemia increased the improvement in revascularization. Interestingly, we found that siRNA CD131 silencing abolished the positive effects of EPO, indicating that the CD131 subunit is required.

The impact of EPO on ECFCs, the non-hematopoietic-derived subpopulation of EPCs, remains unknown, despite their increasingly recognized specific and major contribution to adult vasculogenesis [24]. Our in vitro experiments indicated that EPO enhanced proliferation of ECFCs. The effect was maximal for EPO at a concentration of 5 IU mL−1, and it decreased at higher concentrations. A similar bell-shaped concentration–response curve has been observed for the proliferative effect of EPO on circulating myeloid EPCs from healthy subjects [25] and on cultured mature endothelial cells [15]. EPO also enhanced ECFC migration in vitro, as shown by the increased ability of EPO-stimulated ECFCs to heal a wound and to migrate along an SDF-1 gradient. Recruitment of ECFCs involving SDF-1 and CXCR-4 interaction has been shown to be critical for the angiogenic activity of EPCs in ischemic models [9,26,27]. In addition, Brunner et al. [27] reported enhanced SDF-1-mediated homing of BM stromal cells into ischemic myocardium in the presence of EPO. Thus, our results suggest that EPO could increase SDF-1-mediated recruitment of ECFCs. We also showed that EPO stimulated ECFC differentiation into vascular tubes in Matrigel. This effect concentration-dependently increased up to 10 IU mL−1 EPO. A similar linear concentration–response curve has been described for cocultivated peripheral myeloid EPCs and HUVECs treated with recombinant human EPO [28]. Moreover, enhanced ECFC functions were observed in vitro after EPO removal, suggesting that this cytokine has a sustained effect on ECFCs that may be relevant as an ‘add-on’ strategy in cell-based therapy. One of the current challenges is to identify strategies for overcoming the limitations of EPC-based therapeutic angiogenesis. Indeed, impairment of EPC activity in various clinical conditions, such as diabetes or coronary artery disease, may result in poor treatment efficacy [29,30]. Such alterations are mainly documented for the myeloid EPC population. However, regarding ECFCs, difficulties arise from the limited supply of ex vivo expanded cells from peripheral blood, variability in neovascularization-promoting capacity, or unfavorable changes during expansion process. Therefore, one possibility is to reduce the required number of efficient cells by using ECFCs that have been stimulated before infusion.

In that context, we reported that EPO-primed ECFCs were more potent for therapeutic revascularization in a model of hindlimb ischemia. Blood flow recovery measured by laser Doppler was significantly increased in ischemic hindlimbs from day 3 to day 14 after femoral artery ligature, and was associated with an increase in capillary density in ischemic tissues on day 14. This effect may have prevented hindlimb necrosis, as we observed a reduced clinical necrosis score in EPO-ECFC-transplanted mice. However, on day 28, statistical multiple comparisons of blood flow recovery ratios showed that there was no longer a significant difference between EPO-ECFC-transplanted mice and ECFC-transplanted mice. It should be noted that, on day 28, ECFC transplantation permitted the recovery of ∼90% of the blood flow ratio as compared with 50% in PBS-treated mice. Similar results have been previously obtained with pharmacologic treatment [31]. Therefore, it could be suggested that this maximal effect on blood flow recovery was obtained earlier when ECFCs were primed with EPO. Nevertheless, the main therapeutic objective in critical limb ischemia is to obtain, as soon as possible, tissue reperfusion to allow limb salvage, limit infectious risk, and maintain mobility [32]. The long-term effect is therefore of less importance.

Priming of progenitor cells with p38 inhibitors or endothelial nitric oxide synthase enhancers has been used to augment cell homing, integration and revascularization after induction of ischemia in various tissues [33]. Similar strategies using EPCs pretreated with either activators of the EphB4–ephrin-B2 system [10] or with SDF-1 [11] were successfully applied in hindlimb ischemia mouse models, but no one used the EPO angiogenic properties to prime ECFCs before transplantation in ischemic models. EPO was shown to increase the efficiency of therapeutic angiogenesis by BM stromal cells when coinjected in a model of rat hindlimb ischemia [34] and in a model of myocardial infarction [17]. However, a recent demonstration that exogenous EPO administration induced excessive smooth muscle cell-rich lesions in a carotid artery injury model calls for caution when EPO is used in clinical scenarios associated with high atherosclerosis risk [35]. ECFC priming with EPO ex vivo, rather than coinjection, may be an interesting way to circumvent the unwanted and potential damaging effect of EPO in vivo.

We report here, for the first time, that ECFCs expressed EPOR and that EPO increased this expression, which agrees with the findings of Beleslin-Cokic et al. [36], showing an increase in EPOR mRNA after EPO treatment of endothelial cells. However, western blots performed after isolation of plasma membrane fractions from cell lysates revealed that the increase in EPOR expression was not statistically significant in plasma membrane fractions, suggesting an EPO-induced increase of EPOR synthesis instead of EPOR homing to the membrane. Further studies should evaluate EPOR expression in the ECFC plasma membrane after longer periods of EPO priming. The specificity of the available EPOR antibodies was questioned by Elliott et al. [37], who concluded that the M20 antibody (Santa Cruz) was suitable for detection of the 59-kDa EPOR by immunoblotting. We observed that a 59-kDa protein was detected with the use of the M20 antibody in ECFC lysates and in UT-7 cell lysates, which are known to express EPOR [38]. We confirmed the presence of this protein, which was identified as the EPOR subunit, by fluorescence microscopy, with a second different anti-EPOR mouse mAb for specific immunostaining. In this study, we also observed that ECFCs expressed the CD131 subunit and that EPO increased this CD131 expression. Upregulation of EPOR and CD131 may thus potentiate the effects in vivo of endogenous EPO on transplanted ECFCs. Moreover, it has been reported that CD131 could bind the EPOR subunit and form a heterotrimeric receptor [39]. Such an association may also exist in ECFCs, as we showed that EPOR coimmunoprecipitated and colocalized. It seems that, as when CD131 is shared by IL-3, IL-5 or GM-CSF receptors, according to the presence of the specific ligand [40], the association of CD131 with EPOR may be potentiated by EPO. Additionally, we highlighted a central role of the CD131 subunit in EPO priming, as CD131 siRNA silencing abolished all EPO effects in vitro and in vivo. Brines and Cerami [41] suggested that EPO might have CD131-dependent tissue-protective effects in the injured nervous system and heart. CD131-dependent EPO effects have also been reported by Imamura et al. [42–44], who showed an increase in peritubular capillary endothelial cell number in a kidney ischemia/reperfusion injury model by using a carbamylated EPO (CEPO) as a synthetic variant of EPO that binds specifically to the CD131 subunit. In our work, we also observed that EPO effects were prevented in vitro by the use of EPOR antibody, suggesting that they required the availability of both CD131 and EPOR subunits. The mechanism by which EPO and its interaction with EPOR subunits potentiate ECFC activity in vivo remains unclear. Besides the observation that ECFCs promote angiogenic behavior in vitro, we point out, for the first time, the capacity of EPO priming to strengthen ECFC resistance to oxidative stress in vitro. This antiapoptotic effect is in line with previous vascular and neuronal system reports involving several transduction pathways, including the PI3K–Akt and extracellular signal-related kinase 1/2 pathways and the nuclear trafficking of FOXO3a, downstream of the activation of the EPO heterotrimeric receptor [14]. Here, we showed that EPO priming of ECFCs concentration-dependently increased the phosphorylation of Akt, and that this effect was blocked in the presence of LY29400, which is a PI3K inhibitor. Interestingly, we observed that CD131 silencing reversed the pAkt/Akt ratio, thus suggesting that the EPO-induced Ser473 phosphorylation of Akt required the CD131 subunit. Previous studies demonstrated that the PI3K–Akt pathway was markedly upregulated in CEPO-treated kidneys, thus preventing the ischemia/reperfusion injury-induced tubular epithelial apoptosis [42,45]. These effects observed with the use of CEPO suggest that the CD131-mediated EPO-induced activation of the PI3K–Akt pathway may upregulate antiapoptotic genes, such as those of the Bcl-2 family [46].

In our work, the in vitro resistance to oxidative stress afforded by EPO was associated with an enhancement of cell survival in vivo in the nude mouse bearing ECFCs seeded in Matrigel. Accordingly, it could be suggested that this increase in ECFC survival induced by EPO may represent a major way to improve the efficiency of cell therapy in ischemic diseases [47] that may involve the PI3K–Akt pathway, as we showed a loss of EPO benefits on perfusion recovery when priming ECFCs in the presence of LY294002.

Finally, we observed that VEGF mRNA expression (RT-PCR, data not shown) and protein secretion were not modified upon EPO exposure, indicating that promotion of ECFC activity did not involve VEGF upregulation. VEGF-independent effects of EPO have been previously reported, such as its proliferation-promoting effect on tumoral endothelial cells [48] in vitro or its direct effects on endothelial sprouting [49]. Interestingly, Sautina et al. [50] demonstrated that EPO-induced production of nitric oxide was independent of VEGF, whereas it involved an interaction between VEGFR-2 and CD131. By contrast, in vivo, stimulation of ischemia-induced neovascularization by EPO is commonly thought to occur through upregulation of the VEGF–VEGF receptor system [51], and several findings argue in favor of EPO-induced stimulation of VEGF secretion by various tissues [52,53]. It may be that the involvement of VEGF in EPO-induced angiogenesis is cell-dependent, as Westenbrink et al. [49] showed, in rats with heart failure, that EPO fosters VEGF expression predominantly in cardiomyocytes but not in endothelial cells, which, in turn, stimulates myocardial endothelial proliferation and incorporation of EPCs into vessels.

In conclusion, these results are the first evidence that EPO priming of ECFCs may provide an effective, safe and innovative therapeutic strategy to improve overall ECFC graft efficiency in ischemic diseases. This effect requires the CD131 subunit and EPOR, and results from an increase in ECFC survival and improved angiogenic properties of ECFCs.


Y. Bennis, G. Sarlon-bartoli, B. Guillet, L. Lucas, L. Pellegrini, L. Velly, M. Blot-Chabaud, F. Dignat-Georges, F. Sabatier, P. Pisano: wrote the paper; D. G. Françoise and P. Pascale: revised the intellectual content and approved the version to be published.


We thank K. Fallague, P. Stellmann and S. Pons for technical assistance. We thank P. Nguyen for kindly providing UT-7 cells. This work was supported by research funding from Institut National de la Santé et de la Recherche Médicale.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.