Outgrowth Endothelial Cells: Characterization and Their Potential for Reversing Ischemic Retinopathy

Abstract

Purpose.: Endothelial progenitor cells (EPCs) have potential for promoting vascular repair and revascularization of ischemic retina. However, the highly heterogeneous nature of these cells causes confusion when assessing their biological functions. The purpose of this study was to provide a comprehensive comparison between the two main EPC subtypes, early EPCs (eEPCs) and outgrowth endothelial cells (OECs), and to establish the potential of OECs as a novel cell therapy for ischemic retinopathy.

Methods.: Two types of human blood-derived EPCs were isolated and compared using immunophenotyping and multiple in vitro functional assays to assess interaction with retinal capillary endothelial cells and angiogenic activity. OECs were delivered intravitreally in a mouse model of ischemic retinopathy, and flat mounted retinas were examined using confocal microscopy.

Results.: These data indicate that eEPCs are hematopoietic cells with minimal proliferative capacity that lack tube-forming capacity. By contrast, OECs are committed to an endothelial lineage and have significant proliferative and de novo tubulogenic potential. Furthermore, only OECs are able to closely interact with endothelial cells through adherens and tight junctions and to integrate into retinal vascular networks in vitro. The authors subsequently chose OECs to test a novel cell therapy approach for ischemic retinopathy. Using a murine model of retinal ischemia, they demonstrated that OECs directly incorporate into the resident vasculature, significantly decreasing avascular areas, concomitantly increasing normovascular areas, and preventing pathologic preretinal neovascularization.

Conclusions.: As a distinct EPC population, OECs have potential as therapeutic cells to vascularize the ischemic retina.

Endothelial progenitor cells (EPCs) constitute a minor cell population in peripheral blood, but they have been recognized as having a major role in vascular repair of ischemic tissues. ¹ Evidence suggests that the introduction of autologous or allogeneic EPCs into patients could have major therapeutic benefits. ² EPCs incorporate into vessels during limb ischemia, ^3,4 myocardial infarct, ⁵ and tumor expansion. ⁶ However, contradictory reports indicate that bone marrow-derived circulating endothelial precursors do not always incorporate into adult vasculature. ^7,8 Furthermore, preclinical and clinical studies evaluating the therapeutic potential of EPCs have generated inconsistent outcomes. ⁹ This discrepancy is mostly due to the heterogeneous nature of EPCs and the range of different isolation methods used. ^10,11 EPCs lack a uniform definition and there is a pressing need to clearly define more homogeneous EPC subpopulations to understand their contributions to angiogenic processes. 

Historically, EPCs have been isolated using cell sorting by surface phenotype selection or in vitro cell culture. Cell sorting of EPCs from bone marrow, umbilical cord blood, or peripheral blood is heavily dependent on the surface markers used. In addition, because there is still debate about the most appropriate markers that define an EPC, ambiguity is inevitable. ¹² Among the diverse markers cited in clinical trials, CD34 is the most frequently used for freshly isolated EPCs.² Cell culture is an alternative approach to obtain homogeneous EPCs, enables expansion of cell numbers, and is based on cell adhesion to specific substrates in specialized media. Using this in vitro approach, at least two distinct types of EPCs with different angiogenic properties have been identified: early EPCs (eEPCs) and outgrowth endothelial cells (OECs). ^{13 –15}  

Ischemic retinopathies are leading causes of blindness and a potential disease target for EPC-based therapy. Only relatively recently has the new therapeutic concept of cytotherapy been applied to the ischemic retina. Following the finding that adult hematopoietic stem cells contribute significantly to neovascularization in the ischemic retina, ¹⁶ it was demonstrated that intravitreal transplantation of myeloid progenitors accelerates retinal vascular repair in a model of ischemic retinopathy, ¹⁷ though transplanted cells differentiated not into endothelial cells but microglia. Moreover, using four different models of ischemic vascular damage, Caballero et al. ¹⁸ demonstrated that healthy but not diabetic CD34⁺ cells attached to and assimilated into the retinal vasculature. In this study, there was partial incorporation of injected cells ranging from ∼2% for diabetic cells to ∼31% for healthy cells. These studies show the potential of EPC therapy for ischemic retinopathies, but they also highlight the need for a more thorough characterization of EPC subsets so that the precise fate and usefulness of delivered cells can be determined without the potential to evoke unwanted responses. This is especially important in the context of a complex milieu such as diabetes, which is known to alter EPC phenotype. For example, Sca-1⁺ bone marrow-derived EPCs transplanted into diabetic mice convert to a proinflammatory and an antiangiogenic phenotype and exacerbate limb ischemia. ¹⁹ Therefore, transplanting a cell type carrying monocytic markers into the diabetic retina could enhance differentiation toward an inflammatory cell fate and, therefore, potentially promote retinopathic progression. ²⁰  

In the present study, we provided full phenotypic and functional analysis of two distinct endothelial progenitor subsets: eEPCs and OECs, ²¹ also referred to as endothelial colony-forming cells. ^22,23 Only OECs interacted with retinal capillary endothelium by directly incorporating into the endothelial tube-networks. Moreover, when OECs were injected into ischemic retinas, they contributed significantly to vascular repair, thereby reducing the stimulus for pathologic preretinal neovascularization. 

Methods

EPC Isolation and In Vitro Culture

Fresh human peripheral blood was obtained under full ethical approval from healthy volunteers at the Northern Ireland Blood Transfusion Service (Belfast, United Kingdom). Mononuclear cells (MNCs) were obtained by density gradient fractionation, resuspended in complete medium (EGM-2 MV; Lonza Ltd., Slough, UK) supplemented with 10% FBS and plated on 5-cm Petri dishes precoated with fibronectin (Sigma, St. Louis, MO) at a density of 2 × 10⁶ cells/mL to obtain eEPCs. eEPCs at days 7 to 9 were used for experiments. To obtain OECs, MNCs were resuspended in complete medium (EGM-2 MV; Lonza Ltd.) supplemented with 10% FBS and seeded on 24-well culture plates precoated with rat tail collagen type 1 (BD Biosciences, Bedford, UK) at a density of 1 × 10⁷ cells/mL. Population doubling was carefully monitored, and OECs at early passages were used for all experiments. Frozen human cord blood mononuclear cells were purchased (Allcells; Stem Cell Technologies, Emeryville, CA) and plated under the same conditions described previously. 

Immunocytochemistry

Cells were fixed in 4% paraformaldehyde for 20 minutes at room temperature. After blocking with 5% goat serum and 0.1% Tween20 in PBS for 1 hour at room temperature, cells were incubated with a primary antibody overnight at 4°C. After washing with PBS, cells were incubated with appropriate secondary antibodies for 1 hour at room temperature and observed under a confocal fluorescence scanning microscope (Nikon, Tokyo, Japan). The primary antihuman antibodies used were monoclonal antibodies for CD31 (DakoCytomation, Glostrup, Denmark), smooth muscle actin (Dako), Vimentin (Dako), and monoclonal antibody (Stro-1; Chemicon International, Herts, UK). Polyclonal rabbit antibodies used were against VEGFR2 (Santa Cruz Biotechnology, Santa Cruz, CA), von Willebrand factor (Sigma), pan-cadherin (Cell Signaling Technology, Beverly, MA), ZO-1 (Zymed Laboratories, San Francisco, CA), and β-catenin (Cell Signaling). Polyclonal goat anti-CD133 antibody (Santa Cruz Biotechnology) was also used. Respective anti-rabbit/anti-mouse Alexa Fluor 350, 488 and 568 IgGs (Molecular Probes, Invitrogen, Paisley, UK) were used as secondary antibodies. 

Flow Cytometry

For staining, 1 × 10⁶ cells were used per sample. Cells were filtered using cell strainers with a 35-μm nylon mesh (BD Falcon; BD Biosciences, Franklin Lakes, NJ), resuspended in 100 μL FACS buffer, and incubated with respective antibodies for 45 minutes at 4°C. Antibodies used were against CD31, CD45, CD14, CD133, CD117 (eBioscience, San Diego, CA), CD34 (Miltenyi Biotec, Bergisch Gladbach, Germany), CD105 (Chemicon International), and CD146 (BD Biosciences). After staining, cells were washed in PBS and finally resuspended in 500 μL FACS buffer for analysis using a flow cytometer (FACSCalibur; Becton Dickinson, Franklin Lakes, NJ) with analysis software (CellQuest and FlowJo; Becton Dickinson). At least 20,000 events were acquired for each sample. Respective isotype controls were used to determine accurate settings for data analysis. 

In Vitro Coculture Assay

Primary retinal microvascular endothelial cells (RMECs) were freshly isolated from bovine eyes. Human dermal microvascular endothelial cells (DMECs) were purchased from PromoCell GmbH (Heidelberg, Germany). Cells were labeled with a fluorescent membrane labeling kit (PKH [Sigma]; QTracker [Invitrogen]) before experiments according to manufacturer's guidelines. DMECs were mixed in a 1:1 ratio with eEPCs/OECs before plating onto collagen-coated coverslips. Forty-eight hours later, when monolayers were confluent, cells were fixed in methanol at −20°C for 15 minutes before they were stained with anti–pan-cadherin (Cell Signaling, Beverly, MA), anti–β-catenin (Cell Signaling), and anti–ZO-1 (Zymed). 

Tubulogenesis Assay

Cells were labeled using a fluorescence membrane labeling kit (PKH; Sigma) according to the manufacturer's protocol. RMECs (2 × 10⁵) were mixed with 5 × 10⁴ OECs/eEPCs and resuspended in growth factor–reduced basement membrane matrix (Matrigel; BD Biosciences). Fifty-microliter aliquots were spotted onto four-well plates (Nunc, Roskilde, Denmark). After basement membrane matrix (Matrigel; BD Biosciences) polymerization, spots were covered in endothelial cell basal medium[b]-2 (EBM-2). After 24 to 72 hours, wells were assessed for the presence of tubelike structures and imaged using a confocal scanning microscope (Nikon). 

Oxygen-Induced Retinopathy Model

All experiments were performed in conformity to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the UK Home Office Regulations. Oxygen-induced retinopathy was induced in C57/BL6 wild-type mice as previously described. ²⁴ Briefly, postnatal day (P) 7 newborn mice and their nursing dams were exposed to 75% oxygen (Pro-Ox 110 Chamber Controller; Biospherix, Redfield, NY) for 5 days. At P5 they were transferred back to room air. At P12, six mice received a 1-μL intravitreal injection containing 1 × 10⁵ adult peripheral blood-derived OECs from an early passage that had previously been labeled (Qtracker 655; Invitrogen). Phenol red-free DMEM without growth factors and serum was used as vehicle and injected in the left eye of each pup as a control. All pups were euthanatized at P15 with sodium pentobarbital and eyes fixed in 4% paraformaldehyde. Retinal flat mounts were stained with isolectin B4 (Sigma) and streptavidin-AlexaFlour488 (Invitrogen), human-specific CD31 and CD105 antibodies (Dako), and cell death labeling kit (TUNEL; Roche, Burgess Hill, UK). Stained retinas were visualized and imaged using a confocal microscope. Area quantification was performed using ImageJ software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html). Three-dimensional image reconstruction was performed (EZ-C1 FreeViewer and Volocity software; PerkinElmer, Wellesley, MA). 

Statistical Analysis

All data are expressed as mean ± SEM. Statistical significance was evaluated by one-way ANOVA with Tukey-Kramer's posttest (InStat version 3.06 for Windows; GraphPad Software, San Diego, CA). 

Results

Characterization of Two Distinct EPC Subpopulations Derived from Human Peripheral Blood

Mononuclear cells taken from adult human peripheral blood were plated on two different substrates according to previously described methods ¹⁴ and were cultured with EBM-2 medium that contained hEGF, VEGF, hFGF-B, and R³-IGF-1 and were supplemented with 10% FBS. Cells that attached to fibronectin appeared after 5 to 7 days; most were spindle-shaped but some were spherical (Fig. 1A). These represented the eEPC population. By contrast, cells that attached to collagen I and appeared after 3 to 4 weeks were endothelial-like in appearance and formed a cobblestone-shaped cell monolayer (Fig. 1B). These corresponded to the late EPC subpopulation, also known as OECs. 

Figure 1.

 

Morphology and growth of endothelial progenitors in vitro. (A) eEPCs appeared after 7 days as spindle-shaped cells. (B) OECs appeared after 3 weeks as cobblestone-shaped cells. (C) Growth curves for the ex vivo expansion of OECs derived from human cord blood (CB) and peripheral blood (PB). (D) Representative karyotype for OECs at early passage in culture. Scale bars, 100 μm (A, B).

eEPCs showed low proliferative potential, and cell numbers did not increase after 1 week in culture. By 4 weeks, eEPCs detached from the substrate and decreased in number. OECs had high proliferative potential reaching 28 population doublings (PDs) in 40 days (Fig. 1C), with a doubling time of approximately 34 hours. Moreover, OECs that were single-cell cloned demonstrated a remarkable potential for clonogenic expansion. OECs were isolated from human cord blood with a higher efficiency, appeared earlier in culture (∼12 days), had even higher proliferative potential than peripheral blood-derived cells, and reached 68 PDs in 80 days (Fig. 1C) with a shorter doubling time of 28 hours. Adult peripheral blood-derived OECs from six different donors consistently showed great proliferative potential (20–34 PDs) but then entered a cell aging program characterized by changes in cell morphology, cell cycle arrest, and other senescent features that are under further investigation. FBS supplementation was required for the expansion of OECs. When OECs were cultured in 2.5% FBS medium, they only reached 15 PDs in 25 days, indicating that low serum decreases OEC proliferative potential by 46% compared with 10% FBS-supplemented medium. It was reported that long-term in vitro expansion of EPCs can induce genome instability and karyotype aberrations²⁵; therefore, early and late passages of OECs from eight donors were karyotype-assessed as described in Supplementary Methods. Most OECs had a diploid number of completely normal chromosomes (Fig. 1D); however, a proportion of cells had clonal chromosomal abnormalities in 4 of 8 samples. Interestingly most of them involved translocation/deletion defects in chromosome 14 (Supplementary Fig. S1). OECs were frozen for storage without affecting their phenotype, growth, or original karyotype. 

eEPCs and OECs both displayed the classical EPC phenotype. They bound isolectin, endocytosed DiI-Ac-LDL, and expressed CD31 (Fig. 2A). However, given that macrophages and dendritic cells can also exhibit these same characteristics, ²⁶ we assessed the expression of other endothelial marker proteins such as von Willebrand factor (vWF), which was found to be expressed only on OECs (Fig. 2B). OECs also expressed higher VEGFR-2 levels than eEPCs (Fig. 2B), which was confirmed by Western blot analysis as detailed in Supplementary Methods (data not shown). Furthermore, OECs formed cell monolayers with intercellular junctions characterized by the immunostaining for ZO-1 and β-catenin (Fig. 2B). Importantly, OECs did not show expression for CD133 or monoclonal antibody (Stro-1; Chemicon International; Fig. 2B), indicating they do not represent a subpopulation of hematopoietic or mesenchymal stem cells, respectively. 

Figure 2.

 

Phenotypic characterization of two types of endothelial progenitors. (A) Both eEPCs and OECs bind lectin, endocytose acLDL, and express CD31. Nuclei are stained blue with DAPI. (B) OECs express the endothelial markers von Willebrand factor (vWF) and VEGFR2, form intercellular junctions through ZO-1 and β-catenin, and do not express CD133 or Stro-1. Nuclei stained blue or red with DAPI or PI, respectively. (C) Cell surface immunophenotype of eEPCs, OECs, and DMECs. Green: histograms assessing endothelial markers. Red: hematopoietic markers. Blue: progenitor markers. Gray histograms: respective isotype controls. The percentage of positive cells appears in the top right of each panel. Scale bars, 50 μm.

Cell surface immunophenotyping of eEPCs and OECs by flow cytometry was performed while using human DMECs as a comparator cell type (Fig. 2C). There was no difference in CD31 expression between eEPCs and OECs, corroborating previous results from immunocytochemistry (Fig. 2A); however, endothelial markers CD105 and CD146 were high in OECs but low or absent in eEPCs. Hematopoietic markers CD45 and CD14 were present only in eEPCs representing 99% and 77% of the total number of cells, respectively, demonstrating a hematopoietic origin for eEPCs. On the contrary, OECs did not show any expression of hematopoietic markers but primarily expressed CD105 (26%) and CD146 (99%), which closely matched DMEC expression of CD105 (36%) and CD146 (86%). This suggests that OECs are committed to the endothelial lineage. Interestingly, progenitor cell markers such as CD34 and CD117 were higher in OECs than in eEPCs or DMECs, indicating that OECs exhibit a more immature/primitive phenotype. This was also true for the progenitor cell marker CD133, which, though negative by immunocytochemistry, was determined by flow cytometry to be weakly expressed on 3% of OECs. 

To confirm that OECs are not derived from mesenchymal progenitor cells, we assessed the expression of α-smooth muscle actin (α-SMA) by immunocytochemistry (Supplementary Fig. S2A) and Western blot analysis (Supplementary Fig. S2B), as previously described. ²⁷ Pericytes were used as positive controls for α-SMA. Both subtypes of EPCs lacked α-SMA expression. This finding corroborates other differences between EPCs and mesenchymal stem/progenitor cells (cell morphology, culture substrate requirements, monoclonal antibody [Stro-1; Chemicon International] and CD31 expression) and strengthens the assertion that although MSCs and EPCs might be related during development, in adults they represent two dissimilar cell types. 

OECs Closely Interact with Mature Endothelial Cells by Forming Adherens and Tight Junctions

Endothelial cell function requires integrity of the monolayer through the generation of intercellular junctions. To test whether these two types of EPCs can integrate with mature endothelial cells and cooperate in the formation of a cell monolayer, eEPCs and OECs were labeled green using a membrane dye and were cocultured with red-labeled retinal microvascular endothelial cells (RMECs). Two days after plating, OECs formed a confluent, uniform monolayer, integrating side by side with RMECs (Fig. 3B). eEPCs, on the other hand, did not contribute to monolayer formation, which remained nonconfluent (Fig. 3A). Moreover, eEPCs showed a loose connection to RMECs, seeming to be randomly lying on top of mature endothelial cells. In fact, after 72 hours, the number of eEPCs decreased, and many appeared free floating in the medium. The same coculture experiment was performed using DMECs; consistently, only the OECs, but not the eEPCs, cooperated with the mature endothelial cells in forming a confluent cell monolayer (Fig. 3C). To confirm that the monolayer formed by DMECs and OECs had the characteristics of an endothelial barrier, immunostaining for intercellular junctions was performed. OECs closely interacted with mature endothelial cells by forming adherens and tight junctions, as demonstrated by the uniform expression of cadherin (Fig. 3D), β-catenin (Fig. 3E), and ZO-1 (Fig. 3F) throughout the entire cell monolayer. Triple staining using prelabeled cells was not possible because of technical difficulties as the fixation quenched the membrane labeling. For this reason, quantum dots nanotechnology (Qdot; Invitrogen) was used for labeling the cells that were later stained for β-catenin. In agreement with previous results, DMECs (pink Qdots) established adherens junctions with OECs (green Qdots) through β-catenin (Supplementary Fig. S3). 

Figure 3.

 

OECs integrate into mature endothelial cell monolayers and establish adherens and tight junctions. (A) Nonconfluent monolayer formed by red-labeled RMECs cocultured with green-labeled eEPCs. (B) Fully confluent monolayer formed by red-labeled RMECs and green-labeled OECs. (C) Confluent monolayer formed by red-labeled DMECs and green-labeled OECs used for cell junction immunostaining in (D–F). Uniform distribution of cadherin (D), β-catenin (E), and ZO-1 (F) on cell junctions formed between DMECs and OECs. Insets: negative controls. Scale bars, 50 μm.

OECs Fully Incorporate into a Retinal Microvascular Tube Network

eEPCs and OECs were cultured in basement membrane matrix (Matrigel; BD Biosciences) to assess their capacity to form tubes because de novo tube forming potential in 3D culture systems is a distinctive feature of endothelial cells. eEPCs did not form any tubes and remained as small, spherical-shaped cells throughout the whole experiment (Fig. 4A), whereas OECs showed marked potential for de novo tube formation (Fig. 4B). Only OECs formed intracellular vacuolar compartments that fused to form intercellular luminal spaces (Fig. 4E). This OEC-tube forming capacity was comparable to RMECs. Indeed, when cocultured, red-labeled OECs readily incorporated into the tubes formed by the green-labeled RMECs (Figure 4D, Movie S1), even when plated at a lower ratio (1:4). eEPCs cocultured with RMECs at a 1:4 ratio did not incorporate into the tubes formed by mature endothelial cells, and few attached to the periphery of the tubes (Figure 4C, Movie S2). These findings demonstrate that eEPCs themselves do not have the capacity to form tubes and do not integrate into a network formed by retinal endothelial cells. By contrast, quantification of tube formation determined that OECs contributed to the network by forming tubes in unison with RMECs (Figs. 5A, 5B). A significant increase in the tube area was observed when comparing the cocultures with RMECs alone (P < 0.05). Quantification of red and green tubes indicated that this increase was principally due to OEC-derived tubes (Fig. 5C). These data demonstrate a specific role for OECs in promoting angiogenesis by contributing to the network formation as building blocks for vascular tube formation. 

Figure 4.

 

OECs have de novo tubulogenic potential and form vascular networks in unison with RMECs (A) eEPCs do not have intrinsic tubulogenic potential. (B) OECs have de novo tube-forming potential. (C) eEPCs labeled in green do not incorporate into the vascular network formed by RMECs labeled in red. (D) OECs labeled in green form tubes that integrate into the RMEC vascular network labeled in red. (E) OEC-based endothelial tubes have intracellular vacuoles indicating lumen formation. Scale bars, 200 μm (A, B); 100 μm (C, D).

Figure 5.

 

OECs contribute to retinal microvascular network formation in vitro. (A) Green-labeled RMECs in basement membrane matrix form tubes. (B) Addition of red-labeled OECs significantly increased tube formation by the formation of OEC-derived tubes that incorporated into the green-labeled RMEC network. (C) Quantification of tube areas. *P < 0.05 compared with control (n = 4). Scale bars, 400 μm.

OECs Contribute to Vascular Repair in a Model of Ischemic Retinopathy

Vascular engraftment in vivo has been described as the most rigorous test of true potential for putative EPCs. ¹¹ Therefore, to determine a response to ischemia and potential for participating in preretinal and intraretinal neovascularization, we used a murine model of ischemic retinopathy. This model demonstrates a reproducible retinal vasodegeneration (P7–12), followed by an acute hypoxia phase (P12–15) that finally leads to preretinal neovascularization (P15–18). Based on in vitro findings, OECs were preferred to eEPCs for the in vivo studies. Labeled (Qdot; Invitrogen) and unlabeled OECs were delivered into the vitreous of each mouse at P12. Intravitreal injection was chosen as the administration route because it has been demonstrated that after intravenous infusion of EPCs, most cells accumulate in the liver and spleen ²⁸ ; therefore, local administration had the potential to maximize homing to the sites in need of vascular repair. OECs were injected into right eyes while vehicle was injected into left eyes. 

After 72 hours, flat mounted retinas were stained with isolectin, TUNEL, human-specific endothelial antibodies, and CD68. Confocal microscopy revealed that OECs integrated into the intraretinal microvasculature as single cells forming part of host vessels (Fig. 6A) and as vessels growing into ischemic areas (Fig. 6B). Injected OECs distinctly targeted the ischemic central retina (Supplementary Figs. S4A, S4B) and were clearly nonrandomly distributed along the microvascular network (Supplementary Figs. S4C, S4D). Injected OECs remained fully viable 3 days after delivery, as demonstrated by lack of TUNEL-positive staining (Supplementary Fig. S5). Evaluation of the phagocyte marker CD68 indicated no close association or interaction between OECs and retinal CD68⁺ microglia/macrophages that appeared with distinct morphology (Supplementary Fig. S6) and located more deeply in the retinal tissue. OEC distribution along the superficial retinal vascular network and acquisition of endothelial markers was clearly demonstrated with CD105 and CD31 immunostaining (Fig. 6C, Movies S3, S4). These human-specific antibodies reacted only with injected human OECs and not with resident murine vasculature that was stained green with isolectin. 

Figure 6.

 

OECs integrate into the ischemic retinal vasculature in vivo. (A) OECs labeled in red with quantum dots and injected into ischemic retinas subsequently incorporated into the resident vasculature (stained green with isolectin). (B) OECs labeled in red with quantum dots form tubes that assist in retinal vascular remodeling after ischemic insult. (C) OECs incorporated into resident ischemic vasculature expressed endothelial markers CD105 and CD31 and were located in the superficial vascular plexus. Scale bars, 200 μm (A–C).

Further assessment of avascular, neovascular, and normovascular areas indicated OECs intravitreal injection significantly decreased the retinal avascular area by 40% compared with controls (P < 0.001; Figs. 7A, 7B) and significantly increased the normovascular area by 31% (P < 0.001; Fig. 7C). Although some OECs were found integrated into preretinal pathologic neovascularization (Supplementary Fig. S7), quantification revealed that OEC injection significantly decreased pathologic neovascularization by 58% compared with vehicle-injected eyes (P < 0.01; Fig. 7C). These results demonstrate that OECs directly contribute to vascular repair of ischemic retina. 

Figure 7.

 

OECs contribute to vascular repair of ischemic retina. (A, B) Representative flat mounted retinas of C57BL/6 mice injected with vehicle or OECs, respectively. Lectin staining (green) identifies retinal vasculature. Avascular regions are surrounded by a yellow line. Insets, white: avascular (ischemic) areas. Scale bars, 1 mm. (C) Quantification of avascular, neovascular, and normovascular areas. ***P < 0.001 and **P < 0.01 comparing vehicle-treated and OEC-treated eyes.

Discussion

This study has identified considerable differences between eEPCs and OECs as the two main EPC types isolated in vitro. These differences are based on several factors. In terms of morphology and growth, eEPCs appeared after 1 week as spindle-shaped cells with low proliferative potential, and OECs appeared after 4 weeks as cobblestone-shaped cells with high proliferative potential. The immunophenotype was also distinct, with only eEPCs expressing hematopoietic markers whereas OECs highly expressed endothelial markers. OECs and eEPCs demonstrated different de novo tubulogenesis potential with only the former showing tubule-forming capacity. Consistently, only OECs directly incorporate into the tube network formed by retinal endothelial cells. These findings are in agreement with recent reports on these two subsets of EPCs. ^29,30 Taken together, the evidence suggests that among these two EPC types, it is the OECs that distinctly play a direct role contributing to angiogenesis as structural units. 

There is confusion in the EPC field about cell nomenclature and lineage. However, a rigorous EPC definition states that endothelial progenitors are cells capable of acquiring an endothelial phenotype described not only by cell surface markers but mainly by the capacity to form a tubule network in vitro and the potential to directly incorporate into resident vasculature in vivo. ¹¹ This strict definition is appropriate only for OECs because eEPCs do not fulfill these basic requirements. Consistent with findings of published studies and the outcomes of the current investigation, eEPCs are clearly hematopoietic cells of monocytic lineage. 

By contrast, OECs can be regarded as “true” endothelial progenitors that appear to be programmed to differentiate into endothelium. Indeed, a notable finding in this study was that OECs fully integrated into uniform monolayers with mature endothelial cells by forming adherens and tight junctions. This is important because endothelial progenitors are expected to repair damaged vessels by closely interacting with resident capillary endothelium and reestablishing a functional endothelial barrier. It remains uncertain whether OECs could be capable of differentiation into other cell types, but this is unlikely. Although not investigated in the present study, it is acknowledged that a mixed transplantation of eEPCs and OECs could have a synergistic effect on neovascularization. ³¹  

When comparing OECs from cord blood and adult peripheral blood, it was determined that cord blood–derived OECs have higher proliferative potential, as indicated by their longer life spans in culture and shorter doubling times. These findings are consistent with previous studies that also show cord blood OECs form a vascular network that remains stable for a longer time ³² and has a slightly more angiogenic phenotype ³³ than their adult counterparts. It is also important to make a distinction between OECs and mature endothelial cells. We have demonstrated that OECs have higher expression of progenitor cell markers than DMECs. Furthermore, other reports indicate that OECs have higher proliferation potential, are more sensitive to angiogenic factors, ³⁴ and have a delayed senescence ³⁵ and a stronger mitogenic paracrine effect ³⁶ than mature endothelial cells. Collectively, this evidence suggests that OECs are not circulating endothelial cells sloughed from the vascular wall. Moreover, regardless of source, OECs have certain remarkable characteristics that make them a potential candidate for cell therapies. First, they can be isolated from patients' own blood allowing autologous therapy. Second, they can be grown in vitro, and their cell numbers can be greatly amplified. Third, they are liable to beneficial genetic modification “ex vivo.” Fourth, they do not spontaneously transform, and their growth is limited by a senescence program; hence, they are safe for use in cell transplantations. Fifth, purified cells can be efficiently cryopreserved for storage in patient cell banks. Sixth, they have a relatively stable karyotype, though our data suggest that chromosomal screening would have to be conducted before clinical use for cytotherapy. 

Recent studies ^22,23 have focused on defining the in vitro angiogenic potential of OECs, but few have tested these cells in vivo. This is the first time OECs have been tested as a cell replacement therapy in the ischemic retina. Intravitreous injection of OECs into ischemic retinas resulted in transplanted cell integration into the resident vasculature. Importantly, the OECs demonstrated functional benefits such as a decrease in ischemia and an enhancement of normal retinal vasculature, and this shows the potential benefits of a cytotherapy using well-defined and characterized EPCs to treat retinal ischemia. This preclinical study validates our hypothesis that an EPC cytotherapy is feasible for ischemic retinopathy and underscores the translational potential of such a therapy into the clinical arena. 

We present evidence that OECs injected into ischemic retinas could be readily identified as integrated within the resident vasculature 72 hours after intravitreal injection. This relatively short time was chosen because the oxygen-induced retinopathy model reproduces an acute ischemic pathology; however, it was recently demonstrated that OECs injected into the systemic circulation of NOD/SCID mice are able to lodge and survive in nine different vascular beds for up to 7 months after injection without inducing any thrombosis or infarcts. ³⁷ Taken together, OECs are a novel and exciting prospect for vascular stem cell therapy for ischemic retinopathies, especially because they show clear reparative and vessel formation properties while having limited replicative potential, thereby reducing the neoplastic risk associated with other stem cell therapies. 

Supplementary Materials

Supplementary Methods - (PDF) 

Supplementary Figures - (PDF) 

Movie S1 - 1.2 MB (QuickTime Movie) - Confocal z-stack of in vitro 3D co-culture system of RMECs in red and eEPCs in green. 

Movie S2 - 1.2 MB (QuickTime Movie) - Confocal z-stack of in vitro 3D co-culture system of RMECs in red and OECs in green. 

Movie S3 - 8.6 MB (QuickTime Movie) - Confocal z-stack of flat mounted retina identifies OECs with a CD105 human specific monoclonal antibody (in red) contributing to vascular repair of the superficial vascular tree within the ischaemic retina (vessels labeled green with isolectin). 

Movie S4 - 6.6 MB (QuickTime Movie) - Confocal z-stack of flat mounted retina identifies OECs with a CD31 human specific monoclonal antibody (in red) contributing to vascular repair of the superficial vascular tree within the ischaemic retina (vessels labeled green with isolectin). 

Footnotes

 Supported by grants from the Juvenile Diabetes Research Foundation (JDRF), Fight for Sight (UK), The Medical Research Council (MRC), and The Guide Dogs for the Blind Association.

Footnotes

 Disclosure: R.J. Medina, None; C.L. O'Neill, None; M.W. Humphreys, None; T.A. Gardiner, None; A.W. Stitt, None

The authors thank Judith Briggs for expert assistance in preparing karyotypes, Pauline Linton for constant support isolating RMECs from bovine retinas, Liza Colhoun for valuable help with confocal microscopy, and Stuart Logan for assistance with the PerkinElmer software (Volocity).