|Home | About | Journals | Submit | Contact Us | Français|
Several studies demonstrate that hematopoietic tissues are a source of endothelial progenitor cells (EPCs), which contribute to newly formed blood vessels during tissue repair in adults. However, it is not clear which cell type in these hematopoietic tissues gives rise to EPCs.
To identity the origin of endothelial progenitors within the hematopoietic hierarchy, and assess their in vivo revascularization potential.
Using a single cell sorting approach and in vitro multi-lineage differentiation assays, here we show that individual CD34+CD45+CD133+CD38+ cells from cord blood uniquely have the ability to differentiate into T and B lymphoid, myeloid, and endothelial cells. The latter were characterized by the expression of VE-cadherin, KDR, vWF, eNOs, the lack of CD45, CD133 and c-fms. Unexpectedly when transplanted into hind-limb ischemic NOD-scid IL2Rgammanull mice, freshly-isolated CD34+CD45+CD133+CD38+ cells maintained their hematopoietic identity and were rarely found to integrate into host blood vessels. Nevertheless, they significantly improve perfusion, most likely through a paracrine mechanism. On the other hand, endothelial cells derived in vitro from this fraction, are able to form vessels in vivo in both Matrigel plug and hind-limb ischemia transplantation assays.
These findings indicate that the CD34+CD45+CD133+CD38+ cell fraction contains a common progenitor for the hematopoietic and vascular lineages, and may represent a valuable cell source for therapeutic applications.
It has been hypothesized that endothelial progenitor cells (EPCs) reside in adult bone marrow (BM) and are mobilized into peripheral circulation by cytokines or tissue ischemia 1, 2. Transplantation of either culture-expanded EPCs or freshly isolated cells from adult hematopoietic sources results in enhanced blood-flow 3 and improved function of ischemic tissues 4–10. However, engraftment levels vary significantly from laboratory to laboratory. This variability could be due to the heterogeneity of vascular precursor populations identified in hematopoietic tissues as well as differences in experimental design. To date, controversy still exists with respect to the identification and the origin of these precursors in hematopoietic tissues. For instance, a number of cell types obtained using different strategies have been referred to as EPCs, including differentiated endothelial cells with more limited proliferation ability 11–13, and cells associated with the myelo-monocytic lineage 14–16.
Although several investigators still use the whole mononuclear cell fraction to study postnatal revascularization 14, 17, a number of surface markers have been shown to be useful to identify endothelial progenitors in hematopoietic tissues. EPCs were first isolated from peripheral blood using antibodies to VEGFR-2 (KDR) or CD34 4. While expression of KDR in hematopoietic tissues is controversial, CD34 has found common use as a marker for isolating EPCs 3, 7, 9. Although it is clear that CD34 purification enriches for EPCs, CD34 by itself is not a particularly good marker since it is also expressed in HSCs 18, multiple hematopoietic progenitor cells 19 as well as mature circulating endothelial cells (ECs) 20. The hematopoietic stem cell marker CD133 21 has been suggested to provide better enrichment for endothelial progenitors since it is expressed on EPCs but down-regulated in mature endothelial cells 22. Consistently, a number of studies support the premise that the CD133+ cell fraction is enriched for EPCs 23, and provide evidence for superior perfusion following their transplantation in animal models of ischemia 24, 25. However, because CD133 is also present in early hematopoietic progenitors 21, phenotypic distinction between hematopoietic and endothelial progenitors is not possible. To make things more complicated, two recent studies argue that CD133+ cells do not possess the ability to generate endothelial cells 26, 27. Therefore to date, the origin of endothelial progenitors within the hematopoietic hierarchy remains controversial.
This study was designed to assess whether hematopoietic progenitors, in particular CD133+, have the ability to differentiate into endothelial cells. To rule out the existence of independent progenitors for hematopoietic and endothelial lineages within the CD133 compartment, we tested for the presence of a common precursor at the single cell level. In this way, one can characterize both the single starting cell and its specific progeny. By using single cell sorting assays in combination with replating studies in multiple cord blood samples, here we show the existence of a cell endowed with the ability to differentiate towards the endothelial, myeloid, and lymphoid lineages.
Samples of umbilical cord blood from healthy term newborns were obtained according to procedures approved by the institutional review boards of UT Southwestern Medical Center and University of Minnesota. Mononuclear cells (MNCs) were isolated by Ficoll-Hypaque density-gradient centrifugation. In most cases, prior to FACS sorting, cells were pre-enriched for CD34 using MACS magnetic beads. Detailed purification approach is described in the Online Data Supplement.
Single CD34+CD45+CD133+CD38+ cells were seeded in proliferation medium containing VEGF, SCF, and TPO. After one week, resulting colonies were divided into four sub-fractions, and then sub-cultured in specific conditions to assess their ability to differentiate into endothelial, myeloid, T and B lymphoid lineages. Differentiation and characterization procedures are described in detail in the Online Data Supplement.
NOD-scid IL2Rgammanull mice (Jackson laboratories) were used as recipients. Animal care and all procedures were performed according to University of Minnesota Institutional Animal Care & Usage Guidelines. Detailed information regarding these procedures as well as analyses of engraftment and Laser Doppler Perfusion Imaging are described in the Online Data Supplement.
We began by sub-fractionating the CD34+CD133+ fraction from umbilical cord blood (UCB) based on CD38 and CD45 expression. As observed in Figure 1A (and online Figure I), the large majority of CD34+CD45+CD133+ cells co-express CD38. The ability of this early hematopoietic cell population to generate endothelial precursors in vitro was assessed side-by-side with more committed hematopoietic cell types, including CD33+ (myeloid progenitors) (Figure 1B), and CD14+ (monocytes/macrophages) (Figure 1C). CD33+ and CD14+ cells adhered to the plastic dish when cultured under endothelial conditions, displayed a spindle endothelial-like phenotype by week 2, but failed to expand under these culture conditions (Figure 1D). On the other hand, the CD34+CD45+CD133+CD38+ cell fraction proliferated considerably and eventually gave rise to an endothelial-like cell population, which emerged after 4–6 weeks in culture (Figure 1D and 1E), that resembled HUVECs (Figure 1F). The endothelial phenotype of these cells was confirmed by the expression of VE-cadherin, vWF, KDR, and to a lesser extent eNOs (Figure 1H). These endothelial markers were first detected by week 3, and were abundant by week 6 (Figure 1H), while hematopoietic markers were significantly down-regulated by this time (Figure 1G).
To dissect the multi-lineage differentiation potential of this cell fraction, clonal analysis was performed by FACS single cell sorting (Figure 2). Clones derived from CD34+CD45+CD133+CD38+ single cells (85 from a total of 1024 wells, cloning efficiency of 8.2%) were expanded for one week, and each divided into 4 fractions for endothelial, myeloid as well as T and B lymphoid cell growth (Figure 2).
Our results showed that 49.5% of obtained clones (42 from a total of 85 clones) derived from the CD34+CD45+CD133+CD38+ population were able to produce all 4 lineages (Figure 3A), as confirmed by detailed characterization of these clones. Several myeloid colonies were detected following the replating of one-fourth of the cells from each clone in complete hematopoietic methylcellulose medium (Figure 3B). To determine the capacity of these clones to differentiate into T and B lymphocytes in vitro, the second and third parts of these colonies were cultured on OP9-DL1 28 and MS-5 29 stromal cells, respectively. The presence of B lymphocytes was demonstrated by expression of CD19 and IgM after culture on MS-5 for 45 days (Figure 3Cl). Lineage-specific differentiation was further confirmed by a PCR assay for DJ and VDJ recombination at the immunoglobulin heavy chain locus (Figure 3C). Likewise, double positive CD4+CD8+ T lymphocytes were observed after 45 days in culture on OP9-DL1 (Figure 3D). To assess whether these clonal-derived lymphocytes were capable of undergoing T-cell activation, day 45 expanding CD4+CD8+ T lymphocytes were stimulated with IL-2, anti-CD3 and anti-CD28 antibodies for 5 days 28. At this time, cells became CD27+CD69+, a phenotype characteristic of activated T-cells, while before stimulation these markers were absent (Figure 3D).
Consistent with the initial results (Figure 1), CD34+CD45+CD133+CD38+-derived clones yielded a monolayer of proliferating cells in the presence of endothelial cell growth medium (resulting from one fourth of original clone; Figure 2). The endothelial nature of these cells expanded for 4–5 weeks was confirmed by FACS analysis which revealed the expression of CD146, CD144 (VE-cadherin), and KDR, as well as the lack of CD45 and CD14 (Figure 4A; a total of 42 endothelial cultures were analyzed and all displayed similar results). The majority of these cells co-expressed KDR and CD146 (Figure 4B), confirming their endothelial phenotype. Further immunostaining analysis demonstrated co-expression of vWF and VE-cadherin in expanded endothelial cells (Figure 4C). When cultured in Matrigel, these clonal-derived monolayer cells were capable of generating extensive capillary-like structures (Figure 4D). Additionally, we performed quantitative gene expression analysis to further confirm the endothelial, and non-hematopoietic, nature of these clones. Accordingly, high levels of KDR, vWF, VE-cadherin and eNOs were detected in CD34+CD45+CD133+CD38+-derived endothelial clones, similarly to control HUVECs (Figure 4E). On the other hand, hematopoietic markers, including CD45, CD133 and c-fms, the receptor for macrophage colony-stimulating factor, were absent (Figure 4F), confirming the non-hematopoietic nature of these cells. Since CD41 and Runx-1 have been associated with the emergence of hematopoietic cells from endothelial cells 30–33 during development, a process known as hemogenic endothelium, we also investigated the expression levels for these genes in CD34+CD45+CD133+CD38+-derived endothelial clones and their freshly isolated hematopoietic counterparts. CD34+CD45+CD133+CD38+ cells, before EC differentiation, were found to express much higher levels of CD41 and Runx-1 (Online Figure II) than their respective endothelial cell clones. We also followed the expansion potential of CD34+CD45+CD133+CD38+-derived endothelial progenitor clones (a total of 20 clones). These cells were amenable to expansion, reaching their growth peak at about 50 days, at which point, senescence occurred (Online Figure II).
To examine whether transplantation of CD34+CD45+CD133+CD38+ cells would increase vascularization in the ischemic hind-limb of NOD-scid IL2Rgammanull mice that had been subjected to femoral artery ligation, freshly purified CD34+CD45+CD133+CD38+ cells (3×105) were locally transplanted into the ischemic thigh muscle area immediately after surgery at three different injection points. Control groups consisted of mice receiving the same number of MNCs or simply PBS. Serial analyses with Laser Doppler Perfusion Imaging (LDPI) revealed superior vascularization in mice that had been treated with CD34+CD45+CD133+CD38+ cells (Figure 5A). Accordingly, the ratio of the ischemic/non-ischemic hind-limb blood flow in the CD34+CD45+CD133+CD38+-transplanted group was increased when compared with PBS- or MNC-treated mice at post-operative days 21 and 28 (Figure 5B).
One month after the transplantation, mice were analyzed for the presence of human cells by immunofluorescence staining. We performed double staining with human and mouse specific anti-CD31 antibodies in order to best discriminate recipient- vs. donor-derived vessels, as well as to identify chimeric vessels. As expected, all control mice injected with PBS (n=7) stained only with mouse CD31, similar to the control muscle staining (Online Figure III). In the group of mice that had received MNCs, we were able to detect human cells in 4 out of 7 mice; however most of these mice presented only a few CD31+ cells in 1 or 2 areas of the injected muscle (Online Figure IV), a finding that supports the lack of perfusion improvement in mice transplanted with MNCs (Figure 5A and 5B). Although human CD31+ cells were easily detected in most of the mice transplanted with CD34+CD45+CD133+CD38+ cells (7 out of 8), we found only one chimeric vessel (one chimeric vessel in 1 out of 7 engrafted mice) (Figure 5C). The majority of engrafted human CD31+ cells were found surrounding the recipient vessels (Figure 5D), a pattern observed in all transplanted mice. We found that the majority of the human CD31+ cells surrounding the recipient vessels maintained expression of CD45+ (Figure 5E). These results indicate that the endothelial differentiation of this progenitor is compromised in the xenogeneic environment, and that contribution of human cells to improved perfusion was not due to donor-derived vasculogenesis, but possibly to a paracrine mechanism.
To assess whether the ischemic model does not favor the in vivo differentiation of CD34+CD45+CD133+CD38+ cells into endothelial cells, we differentiated them first in vitro into endothelial cells (Figure 6A and 6B), and injected these into mice using the Matrigel plug assay. Two weeks after implantation, plugs were removed and analyzed by immunofluorescence staining for human and mouse CD31. In the case of freshly isolated cells, we observed again the presence of human CD31+ cells surrounding the mouse-derived vessels or alternatively, as clusters of hematopoietic cells (Online Figure V). In contrast, plugs that were injected with CD34+CD45+CD133+CD38+-derived endothelial cells presented both donor- and recipient-derived vessels (Figure 6C and 6D), as indicated by staining with human and mouse specific anti-CD31 antibodies (Figure 6E and 6G, and 6F, respectively). These newly formed vessels were functional as evidenced by the presence of erythrocytes in the lumen (Figure 6C-G).
Based on these findings, we investigated the ability of CD34+CD45+CD133+CD38+-derived endothelial cells to contribute to new vessels in the ischemic hind-limb model. To confirm the functionality of human-derived endothelial cells, biotinylated tomato lectin, which specifically binds to perfused endothelial cells, was injected intravenously into transplanted mice just prior to sacrifice. As observed in Figure 7, functional human-derived blood vessels were integrated into the host circulatory system, as demonstrated by co-staining for Lectin and CD31. These results confirm the endothelial nature of the endothelial cells that were differentiated from the common progenitor, the CD34+CD45+CD133+CD38+ cell fraction.
The identification of endothelial progenitors in the adult circulation by Asahara and colleagues 4 has challenged the dogma that vasculogenesis is restricted to embryogenesis. Adult endothelial progenitors are found in association with the hematopoietic compartment 1, 2. However whether endothelial and hematopoietic progenitors have a common origin or represent independent lineages within the bone marrow remains a topic of debate. Interestingly, a study involving the analysis of blood and bone marrow samples obtained from chronic myeloid leukemia (CML) revealed the presence of BCR/ABL, the oncoprotein encoded by the Philadelphia chromosome, in both hematopoietic and endothelial cells 34. This is particularly relevant since detection of the Philadelphia chromosome in all blood lineages of individuals with CML has been considered as direct evidence for the existence of the HSC in humans 35. Stronger evidence for an association between these two lineages in postnatal life was provided by Grant and colleagues 36. By transplanting individual Sca-1+c-Kit+Lin− BM HSCs from GFP transgenic mice into irradiated recipients, these authors demonstrated that upon hematopoietic engraftment, donor-derived GFP+ cells contributed to the endothelial lineage when mice were subjected to retinal ischemia 36. Similar results were observed following the transplantation of cord blood CD34+ cells into NOD/scid mice. These results were corroborated by another study demonstrating that infusion of individual Sca-1+c-Kit+Lin− BM HSCs resulted in broad incorporation of donor-derived cells into the endothelial compartment even in the absence of additional vascular injury, which was not a result of cell fusion 37. Further studies by this group suggest that endothelial engraftment resulted from the stem cell pool or from a common myeloid progenitor but not from a lymphoid progenitor 38. Although all these findings support the hypothesis that endothelial progenitors arise within the hematopoietic compartment, it is still unclear which cell type in the hematopoietic hierarchy gives rise to the endothelial lineage.
By using a combination of surface markers, single cell sorting assays, and subsequent replating studies, here we demonstrate for the first time, the existence of an adult common precursor for endothelial, myeloid, and lymphoid progenitors in human cord blood. Unexpectedly, this bipotent precursor resides within the CD34+CD45+CD133+CD38+ fraction, downstream of CD34+CD38− cells, suggesting that a multipotent progenitor, not the hematopoietic stem cell, is the point of divergence of the endothelial and hematopoietic lineages.
Despite the evident ability of CD34+CD45+CD133+CD38+ cells to differentiate into hematopoietic and endothelial cells in vitro, hematopoietic differentiation predominates in ischemic transplantation. This could be attributed to two major issues: i) hemangioblasts may not be endowed with in vivo vascularization potential since so far, this ability has not been investigated in ES- or embryo-derived hemangioblasts; ii) an ischemic xenogeneic mouse environment is not appropriate or sufficient to induce the differentiation of this human precursor into endothelial cells.
Since our goal here was to investigate whether hematopoietic cells, in particular CD133+, have the ability to differentiate into endothelial cells, we did not address the existence of other angiogenic cells outside of the hematopoietic compartment, as suggested by other investigators 26, 27, 39. It might be the case that these endothelial progenitors are endowed with robust in vivo revascularization potential, although this has yet to be determined in animal models of ischemia. Mouse-to-mouse transplantation experiments involving the analyses of mice that had been subjected to unilateral femoral artery occlusion following the engraftment of BM cells isolated from transgenic mice expressing enhanced green fluorescent protein (GFP) revealed that donor GFP+ cells fail to incorporate into the adult growing vasculature, but were detectable around growing collateral arteries 40. A similar outcome was obtained following the transplantation of hematopoietic stem cells isolated from GFP transgenic mice directly into ischemic myocardium of wild-type mice 41. Interestingly here we observed a similar engraftment pattern following the transplantation of freshly isolated CD34+CD45+CD133+CD38+ cells into hind-limb ischemic mice. In these engrafted mice, human CD31+CD45+ cells were found in large quantities surrounding recipient’s vasculature, suggesting that the improved perfusion observed in transplanted mice may be due mostly due to a paracrine pro-angiogenic function of these cells. This would be in agreement with recent studies involving the transplantation of human CD133+ cells 42, 43 or EPCs 44, 45, which indicate that these cell types secrete angiogenic factors 40, 44.
On the other hand, when CD34+CD45+CD133+CD38+ cells are allowed to differentiate into endothelial cells in vitro, and then injected into mice using the Matrigel plug assay or the ischemia model, these cells give rise to functional human-derived and chimeric mouse-human blood vessels. Taken together our data reveals that the CD34+CD45+CD133+CD38+ cell fraction, which per se may have an angiogenic effect in vivo, is endowed with the ability to differentiate in vitro into myeloid, lymphoid, and endothelial lineages. The endothelial cells generated from this common precursor are able to participate in new blood vessel formation, providing a rationale for the use of this cell population, as well as their progenitor, for therapeutic applications in ischemia conditions.
Recent investigations have highlighted the potential of EPCs for therapeutic applications in vascular medicine. Hematopoietic tissues from adults have been the primary cell source in these laboratory studies and early clinical trials. Most previous investigations have involved the infusion of heterogeneous populations of cells. To date, it is not clear which cell type within these hematopoietic tissues is responsible for vascular regeneration. We demonstrate here through clonal analysis that at least one source of endothelial progenitor cells in human cord blood can be traced back to a common progenitor of lymphoid and myeloid hematopoietic cells, a cell marked by the antigens, CD34, CD45, CD133, and CD38. While injection of this freshly-isolated progenitor cell into ischemic hind-limbs of immunodeficient mice results in improved recovery from ischemia, engrafted cells do not adopt an endothelial fate directly in this in vivo environment, but rather adopt a pro-angiogenic hematopoietic fate. However when this progenitor is first differentiated in vitro into endothelial cells, these cells generate functional human-derived and chimeric mouse-human blood vessels. These findings provide insights into the identity of EPCs and their positioning with regard to the hematopoietic hierarchy, giving scientific rationale for future clinical applications in patients with ischemic vascular diseases.
We are grateful to J.C. Zuniga-Pflucker and M. Andreeff for the OP9-DL1 and MS5 stromal cell lines, respectively. We thank A. Filareto and M. Iacovino for technical guidance, and Michael Kyba for critical reading of the manuscript.
SOURCES OF FUNDING
This work was supported by a scientist development grant (to RCRP) from the American Heart Association. The Ob-Gyn Tissue Procurement Facility at the UT Southwestern Medical Center is funded by NIH grant HD011149.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.