Endothelial progenitor cells (EPC) were identified by Asahara et al.1, 2 as CD34+VEGFR2+ mononuclear cells that were able to contribute to neovascularization in sites of ischemia. This finding changed the paradigm that vasculogenesis was restricted to development of blood vessels in the embryo, and initiated the notion of using EPC to promote therapeutic angiogenesis (reviewed3, 4). Evidence collected in animal models demonstrated that EPC can effectively restore impaired vascularization in damaged ischemic tissues,5-7 and recently, human clinical trials have validated the ability of EPC to improve critical limb ischemia and treat cardiovascular diseases.8-10 Because the levels of circulating EPC decline with age, and are reduced in several diseases,11-14 including liver fibrosis and inflammatory bowel disease (IBD), it is possible that the impairment in mobilization and/or increased EPC senescence might lead to perpetuation of the pathologic state.15 Also, a key factor in IBD progression is the development of abnormal or inadequate vasculogenesis, the presence of immature vessels with altered pericyte support, and an increased responsiveness of the dysfunctional microvasculature to growth factors.16 Furthermore, intestinal microvascular and endothelial cell dysfunction lead to persistent tissue hypoperfusion/ischemia, contributing to the persistence of chronic inflammation.17, 18 Likewise, it has been demonstrated, in several different injury models, that both circulating and transplanted EPC promote vascularization and play a role in liver regeneration. This contribution is not by direct differentiation to hepatocytes, but by reducing fibrosis and creating a microenvironment conducive to hepatic proliferation and differentiation.19-24 Therefore, promoting vasculogenesis may be an essential step for fostering organ repair, not only through contribution to the resident endothelial cell pool, but also through the production of factors that induce tissue recovery. The use of colony-forming assays to further characterize EPC demonstrated that this population contained both hematopoietic stem cell (HSC)-derived cells with myeloid progenitor activity, which could differentiate into macrophages but were unable to form secondary EPC colonies, and endothelial colony-forming cells (ECFC), with robust proliferative potential and vessel-forming activity in vivo.25, 26 Putative sources of ECFC for cell therapy include peripheral blood (PB), unfractionated bone marrow (BM), and cord blood (CB). Although autologous adult sources offer the advantage of immune compatibility, CB, because of its much higher ECFC content25, 27 and its ability to generate highly proliferative and expandable ECFC colonies constitutes an ideal source of allogeneic cells for use in cell therapies. To test the basic biologic ability of ECFC to home to and contribute to different phenotypes, the model/method used should provide all necessary stimuli/support to induce the cells in question into all putative lineages. During fetal life, in the absence of injury/apoptosis, or selective pressure, the developmental signaling arising from proliferating cells and supportive microenvironmental niches is able to promote the robust induction of engrafted cells to tissue-specific cell phenotypes.28 Using a fetal sheep transplantation model, we previously demonstrated the ability of human HSC and clonally derived marrow stromal cells (BMSC) to contribute to hepatogenesis, and showed that the route of administration influenced the levels of donor-derived hepatocytes and their pattern of distribution throughout the parenchyma of the recipient's liver.29, 30 We also demonstrated, using this model, that subpopulations of BMSC are able to repopulate the intestinal stem cell compartment and generate mature intestinal cells at higher levels than unselected cultures.31 Here we investigated the innate ability of human CB-derived ECFC to home, upon transplantation, to the hepatic and intestinal parenchyma, and their intrinsic ability to integrate and contribute to the cytoarchitecture of the intestine and liver in the absence of injury.
Although the vasculogenic potential of circulating and cord blood (CB)-derived endothelial colony-forming cells (ECFC) has been demonstrated in vitro and in vivo, little is known about the inherent biologic ability of these cells to home to different organs and contribute to tissue-specific cell populations. Here we used a fetal sheep model of in utero transplantation to investigate and compare the intrinsic ability of human CB-derived ECFC to migrate to the liver and to the intestine, and to define ECFC's intrinsic ability to integrate and contribute to the cytoarchitecture of these same organs. ECFCs were transplanted by an intraperitoneal or intrahepatic route (IH) into fetal sheep at concentrations ranging from 1.1-2.6 × 106 cells/fetus. Recipients were evaluated at 85 days posttransplant for donor (human) cells using flow cytometry and confocal microscopy. We found that, regardless of the route of injection, and despite the IH delivery of ECFC, the overall liver engraftment was low, but a significant percentage of cells were located in the perivascular regions and retained the expression of hallmark endothelial makers. By contrast, ECFC migrated preferentially to the intestinal crypt region and contributed significantly to the myofibroblast population. Furthermore, ECFC expressing CD133 and CD117 lodged in areas where endogenous cells expressed those same phenotypes. Conclusion: ECFC inherently constitute a potential source of cells for the treatment of intestinal diseases, but strategies to increase the numbers of ECFC persisting within the hepatic parenchyma are needed in order to enhance ECFC therapeutic potential for this organ. (HEPATOLOGY 2012;56:1086–1096)
Materials and Methods
ECFC Culture and Characterization.
ECFCs were kindly provided by Dr. Yoder.25, 27 ECFCs were isolated from human umbilical cord blood (20-100 mL) using Histopaque 1077, washed 3 times with EBM-2 medium (Cambrex, Walkersville, MD) with 10% fetal-bovine serum (FBS; Hyclone, Logan, UT), 2% penicillin/streptomycin (Invitrogen), and 0.25 μg/mL amphotericin B (Invitrogen) (complete EGM-2 medium). Cells were seeded onto tissue culture plates precoated with type-1 rat-tail collagen (BD Biosciences, Bedford, MA) at 37°C, 5% CO2, in a humidified incubator. Nonadherent cells and debris were removed after 24 hours and complete EGM-2 medium was added. Medium was changed daily for 7 days and then every other day until first passaging. Cells were expanded and transduced using a murine stem cell virus (MSCV)-derived retroviral vector encoding DsRed. Flow cytometric analysis prior to transplant demonstrated that ECFC expressed CD31, CD105, CD144, CD146, and were negative for CD45. Low levels of CD34, CD117, and CD133 were also expressed.25
The DsRed+ECFC were injected into fetal sheep at a concentration of 0.5-2.6 × 107/kg by either intraperitoneal (IP; n = 9) or intrahepatic (IH; n = 4) administration at 59-65 gestational days. Animals transplanted IP received: 5.25 × 105 cells (n = 1); 1.3 × 106 cells (n = 4); or 2.6 × 106 (n = 4). Animals transplanted IH received 1.58 × 106 cells. All surgical procedures were performed at the University of Nevada, Reno, and approved by the Institutional Animal Care and Use Committee (IACUC). The animals were cared for in accordance with the U.S. Public Health Service PHS Policy of the Humane Care and Use of Laboratory Animals. At 143-145 days of gestation, PB, liver, and small intestine were collected and analyzed by flow cytometry and confocal microscopy.
Tissue Preparation and Analysis.
Tissues were washed with ice-cold phosphate-buffered saline (PBS) (Gibco, Aukland, CA), fixed in ice-cold PBS containing 4% paraformaldehyde, cryopreserved in 2 parts 20% sucrose, 1 part OCT compound (TissueTek, Torrance, CA), and frozen in isopentane cooled in liquid nitrogen. Cryoblocks were sectioned in 8-μm-thick sections using a Leica Minotome and adhered to Superfrost/plus slides (Fisher Scientific, Santa Cruz, CA). Slides were washed with PBS and blocked with 10% normal goat serum (NGS) (Atlanta Biologicals, Lawrenceville, GA) in PBS and incubated overnight with the following antihuman primary antibodies: CD31, CD34, Von Willebrand factor (vWF), cytokeratin-20* (Biogenex, San Ramon, CA); Factor VIII (Affinity Biologicals, Ontario, Canada); vimentin*, smooth muscle actin (SMA*) (Sigma, St Louis, MO); CD117*, chromogranin A (Abcam Cambridge, MA); CD133* (Novusbio, Littleton, CO); DsRed (ClonTech, Mountain View, CA). *Denotes crossreactivity with ovine cells. Slides were washed in 2% NGS in PBS and incubated for 30 minutes with the secondary antibody (Molecular Probes, Eugene, OR). The nuclei were counterstained with DAPI (Biogenex).
Imaging of Tissue Sections.
Immunofluorescence microscopy imaging was performed using an Olympus Fluoview 1000 Confocal System for both confocal and single plane focus images. Imaging employed 40× and 60× objectives, in combination with an argon laser to fluoresce both DAPI and Alex-Fluor 488 secondary antibodies, and diode-based lasers to image Alexa-Fluor 594 and 647, along with DsRed using filters for 594 nm and 647 nm wavelengths. Images were placed in panels using Adobe Photoshop 7.
Data Analysis and Statistics.
An average of ≥14,000 total cells were counted per tissue section and the numbers of human cells in each section determined by DsRed as a percentage of the total number of cells. The mean engraftment percentages were calculated by averaging the results. Statistical analysis was performed using a combination of analysis of variance (ANOVA) and either two-tailed Student's t tests or Mann-Whitney U tests, depending on the results of the variance analysis. The StatPro (Palisade, Ithaca, NY) and Sigma Plot v. 11 (Systat, San Jose, CA) statistical packages were used for all statistical analyses/tests.
In Situ Hybridization.
Paraffin blocks were sectioned and prepared as described.28 After rehydration, antigen retrieval was performed twice for 10 minutes each at 94°C in 1× Target Retrieval Solution (Dako, Carpinteria, CA). The slides were cooled to room temperature and digested using 20 μg/mL proteinase K (Invitrogen, Carlsbad, CA) for 30 minutes for the human sections and 45 minutes for the sheep sections. All slides were then prehybridized for 5 minutes at 85°C in 50% Di-Formamide (Sigma) and 2× SSC, and hybridized with either a human- or a sheep-specific probe as described.28, 30
Assessment of Donor-Derived ECFC in Circulation.
PB of transplanted animals was analyzed by flow cytometry with monoclonal antihuman: CD105, CD45, CD146 and CD13/APN-FITC and CD45-PerCP antibodies (BD Biosciences). Flow cytometric analysis was performed with a FACScan (BD Biosciences) using an age-matched, nontransplanted animal as a control.
Transplanted ECFC Preferentially Home to the Intestinal Crypt Region, Whereas in the Liver ECFC Lodge in the Perivascular Areas.
Human ECFC in the intestine or liver parenchyma were detected by DsRed fluorescence and confirmed with an antibody specific for the DsRed protein. Following IP injection (n = 9), the percentage of donor-derived ECFC detected in the liver was 0.11 ± 0.02% (Fig. 1A-C), and the higher cell dose did not result in increased levels of donor-derived ECFC. In animals transplanted IH, despite the local delivery of cells, the proportion of human ECFC found within the hepatic parenchyma was 0.16 ± 0.07%, not significantly different from that observed in animals transplanted IP. DsRed protein expression and fluorescence in situ hybridization with human- and sheep-specific probes were used to confirm the presence of human cells in the chimeric tissues (Fig. 1D,E).
In both IP- and IH-injected animals, 98.70 ± 0.92% and 91.69 ± 3.12%, respectively (P < 0.05) of ECFC lodged preferentially in perivascular areas (Fig. 1B,C,F). Perivascular areas were defined by morphology as those within a distance of two cell bodies from the vasculature.
ECFC engrafted within the intestine of the transplanted animals at significantly higher numbers than seen in the liver. In animals injected IP, the overall percentage of intestinal engraftment was 9.77 ± 1.2%, and in animals receiving an IH injection it was 8.64 ± 1.47%, demonstrating that ECFC intestinal homing was conserved across both IH and IP injection routes, and that ECFC favored the intestine as a homing site.
Within the intestine, 81.89 ± 2.21% of the engrafted ECFC, regardless of the administration route, located preferentially to the intestinal crypt region (CPT) and the villi of the intestine (Fig. 2A,B). Although the vast majority engrafted in and around the CPT, the villi region contained 13.63 ± 1.66% of the engrafted cells. In animals transplanted IP, the cell dose did not have a statistically significant effect on engraftment, because in the animals receiving 2.6 × 107/kg (n = 4) the overall ECFC engraftment was 10.89 ± 1.92%, compared to 8.64 ± 1.47% in animals receiving 1.36 × 107/kg (n = 4). Similarly, the contribution to the interstitial CPT area was 26.71 ± 3.75% in the 2.6 × 107/kg transplanted animals and 23.3 ± 3.35% in the group receiving half this cell dose. A comparison between the percentage of overall intestinal engraftment versus CPT interstitial tissue contribution of the transplanted ECFC, per animal, can be seen in Fig. 2C.
Donor-Derived ECFC Circulate in the PB of Both IP- and IH-Transplanted Animals.
To determine if transplanted cells were present in circulation, flow cytometric analysis of peripheral blood was performed on both IH- and IP-injected animals using human-specific antibodies for CD105, CD146, and APN/CD13. In IH-transplanted animals the percentages of CD13, CD105, and CD146 were 3.06 ± 0.5%, 2.3 ± 0.8%, and 1.27 ± 0.14%, respectively, whereas IP-transplanted animals had 2.6 ± 0.61% CD13, 0.78 ± 0.18% CD105, and 1.12 ± 0.41% CD146-positive circulating cells.
ECFC Engraft the Liver in Perivascular Areas and Continue to Express Endothelial Markers.
To confirm that the observed red fluorescence was indeed due to expression of the vector-encoded DsRed by the transplanted cells, immunofluorescent labeling with an anti-DsRed antibody was performed and the results are depicted in Fig. 1B,C.
To determine whether liver-engrafted ECFC maintained an endothelial lineage, we investigated whether these cells continued to express the phenotypic markers exhibited prior to transplant. As can be seen in Fig. 3A-F, all ECFC, regardless of administration route, continued to express CD31 (IH: 98.45 ± 0.47%; IP: 91.25 ± 2.85%; P < 0.05) and vWF (IH:97.73 ± 1.45%; IP:88.8 ± 2.3%; P < 0.01). Within the liver, ECFC also contributed to cells expressing factor VIII:c antihemophilic globulin (FVIII:C), with 59.35 ± 6.86% of IH-injected ECFC and 42.38 ± 7.39% of IP-injected ECFC expressing this protein (Fig. 4A-C). CD34 expression was lower than that seen with other endothelial cell markers and was observed in 39.64 ± 3.79% of DsRed-positive cells following IH injection and in 38.25 ± 6.36% following IP injection. Representative images of DsRed-positive donor cells in conjunction with anti-CD34 immunofluorescence show the colocalization of CD34 with DsRed (Fig. 4D-F).
In the Intestine ECFC Express SMA and Vimentin and Localize with Phenotypically Similar Endogenous Cells.
Independent of injection route or cell dose, 25.37 ± 1.30% of engrafted DsRed+ ECFC coexpressed SMA and vimentin. Moreover, these DsRed+/SMA+/vimentin+ cells constituted 28.51 ± 1.06% of the entire vimentin+/SMA+ population within the small intestine (Fig. 5A-F; Supporting Fig. 1).
Furthermore, 9.46 ± 0.69% of donor-derived DsRed+ cells expressed CD117 and localized to interstitial areas harboring the endogenous CD117+ population (Fig. 6A-C). In addition, a small number of donor-derived cells (<1%) expressed the enteroendocrine marker, chromogranin A. (Fig. 6D-F). Although transplanted cells did not contribute to the intestinal epithelial cell layer (Fig. 7A-C), 91.88 ± 1.82%, of donor-derived cells expressed the progenitor cell marker CD133, and could also be found in zones where the recipient's endogenous CD133+ cells were localized (Fig. 7D-F).
EPC and ECFC, whether isolated from BM, PB, or CB, represent a promising tool for developing novel cell therapies.20, 32, 33 Both EPC and ECFC have already proven successful in vasculogenic settings, and represent a promising cell therapy in regenerative medicine, but their potential application to other organ systems has yet to be investigated.32-38 CB has served as a safe, “ready-to-use” source of hematopoietic stem/progenitor cells for HSC transplantation. In addition to HSC, CB also contains ECFC, in higher frequency (≈15-fold) and with 100-fold greater replating potential than adult PB.25 Furthermore, CB-derived ECFC have greater telomerase activity, and in vivo assays show the ability of these cells to originate a greater density of vessels than the ECFC from adult blood samples.3 Therefore, we used a sheep in utero transplantation model to investigate and compare the intrinsic ability of human CB-derived ECFC to migrate to the liver and intestine, and to define ECFC's intrinsic ability to integrate and contribute to the cytoarchitecture of these organs in the absence of injury. We and others have previously shown, using this same animal model, that human cells transplanted IP migrate to and engraft multiple tissues, and that the route of administration and homing molecules influence the levels and pattern of distribution of donor-derived cells.28, 39-41 Here, we show that IP transplantation of human ECFC into fetal sheep resulted in preferential engraftment of these cells into the developing intestine, such that 89-fold higher levels of donor cells were present in the intestine than the liver. Furthermore, fusion between donor and recipient cells was not apparent, as demonstrated by in situ hybridization using human- and sheep-specific probes. In animals that were transplanted IH, the cells still preferentially migrated to the intestine with 54-fold more transplanted cells localizing to this organ than the liver, demonstrating that, despite the local delivery of cells, ECFC egress the liver to populate the gut. In agreement with these results, and in contrast to what we have shown when MSCs were directly injected into the liver,30 donor-derived ECFCs were detected in circulation in IH-transplanted animals at levels similar to those of animals transplanted IP, showing that one of the routes of ECFC trafficking to the intestine is through the circulation. Furthermore, circulating donor-derived cells, in addition to displaying phenotypic markers of endothelial cells, expressed the metalloprotease CD13/APN, a critical regulator of endothelial morphogenesis during angiogenesis.42 The preferential migration of ECFC to the intestine is most likely due to the expression of EphB2 by these cells, and agrees with our previous data showing that clonally derived BMSC that express high levels of EphB2 have a 4 to 5-fold increase in levels of intestinal engraftment when compared to EphB2low BMSC.31 ECFC located to perivascular regions of the liver regardless of the injection route; however, higher percentages of cells were found in these areas after IP transplantation. This could be explained by the route that IP-transplanted cells follow through the intestinal lymphatics, reaching the liver through the portal vein. By contrast, a significantly higher percentage of IH-transplanted cells retained an endothelial phenotype as determined by CD31, vWF, and FVIII positivity,43 showing that the administration route influences the localization, and might have an effect on the type of cell that is selected to repopulate the organ. In addition, whereas other studies using this and other models have described the ability of HSC and MSC to contribute to mature hepatocytes,29, 30, 44, 45 in the present study transplantation of ECFC did not result in generation of mature hepatocytes (data not shown). In addition, we did not see direct incorporation of ECFC into the vasculature. This is in agreement with other studies that showed that no long-lasting in vivo vessel formation by ECFC occurs unless MSC or other cells such as 10T1/2 are coimplanted simultaneously.46 Therefore, depending on the needs of the patient, either independent or combined ECFC and MSC therapies may be warranted to achieve an effective therapeutic result. Also, an adequate number of cells must reach an affected organ to produce a beneficial effect, and here we show that the intrinsic ability of ECFC to migrate to the liver was low. This may be due to the model system used; perhaps in the presence of liver damage, more ECFC would be recruited to the site of injury.22 Nonetheless, even the low percentage of donor-derived ECFC within the hepatic parenchyma already represents a substantial contribution to the endothelial cell pool, given the liver's large size and its densely packed cellular structure. Still, the development of approaches to promote higher levels of ECFC engraftment within the liver would be of interest to increase the therapeutic potential of these cells.
In contrast to the liver outcome, these studies demonstrated that ECFC displayed an enhanced ability to migrate to the intestine and preferentially lodge within the CPT. Because the numbers of EPC are decreased in the peripheral blood of IBD patients,18 and intestinal microvascular and endothelial cell dysfunction can lead to persistent tissue hypoperfusion and ischemia, contributing to chronic inflammation,18, 47 our present findings suggest that ECFC transplantation may constitute a tool for improving the vascular abnormalities found in IBD. Myofibroblasts, another intestinal cell population affected during IBD, play an important role in inflammation and fibrosis, and have recently been shown to help maintain intestinal mucosal tolerance.48, 49 Here we show that, following transplantation, ECFC contribute significantly to a population that is vimentin+/SMA+, with almost 30% of the small intestine vimentin+/SMA+ population being donor-derived, demonstrating that ECFC localize to the proper anatomical region and display the phenotype of a putative myofibroblast. Although in humans and in animal models, controversy still exists regarding the origin of myofibroblasts, it has been shown that vascular pericytes, and BM-derived stem cells, all contribute to the myofibroblastic population.49, 50 Furthermore, Medici et al.51 have shown that endothelial cells undergo endothelial-mesenchymal transition, and acquire a stem cell-like phenotype upon treatment with ALK2 ligands such as transforming growth factor beta2 (TGF-β2) or BMP4. Here, we also demonstrated that a large percentage of donor-derived cells that localized around the CPT expressed CD133. Because a small percentage of the transplanted ECFC already expressed CD133,25 it is difficult to ascertain whether there was upregulation of this marker after engraftment, or whether there was a selective pressure toward CD133+ ECFC engrafting the intestine. Although the role of CD133+ cells during intestinal development has not been reported, because animals were transplanted during the fetal period, it is possible that during this time higher percentages of CD133+ cells are present within the intestine. Of note, these cells colocalized in areas where host CD133+ cells resided, and these cells were not part of the mature epithelia. Almost 10% of ECFC continued to express CD117 and migrated to areas in the intestine that contained recipient's CD117+ cells. In the intestine, CD117 is a marker of the interstitial cell of Cajal (ICC),52 which is responsible for the generation and propagation of electrical slow waves underlying rhythmic contractile activity in the intestine. Alteration of the normal functioning of these cells likely plays a role in the pathogenesis of various diseases.53,54 Because ECFC expressed CD117 prior to transplant, and due to limitations of the model, it is not possible to determine whether ECFC can contribute to or function as the ICC population of the intestine, but because they localized with the endogenous CD117+ population, it would be interesting to explore ECFC contribution to ICC in models which lack these cells.
In conclusion, our studies have established, in a noninjury model, that human ECFC can lodge within the liver and contribute to the vasculature, and that strategies to increase the number of cells persisting within the hepatic parenchyma may be necessary to increase ECFC therapeutic potential. By contrast, the significant levels of ECFC engraftment in the intestine and the supporting niche formed by the donor-derived ECFC strongly suggest that these cells could constitute a potential resource for the treatment of IBD.