Human Cord Blood Progenitors Sustain Thymic T-Cell Development and a Novel Form of Angiogenesis

TRANSPLANTATION of hematopoietic stem cells is currently the treatment for reconstitution of the immune system in immunodeficiency diseases1,2 or after myeloablative chemotherapy.3,4 Clinical experience5,6 and animal models7-11 indicate that both the proliferative and differentiative capacity of the transfused stem cells are critical for restoration of host immune competence, including a functional mature B- and T-cell repertoire.

Human umbilical cord blood (CB) has been recently identified as a rich source of multipotent hematopoietic progenitors.12-15 As compared with adult bone marrow (BM), CB progenitors have been shown to have a higher proliferative and self-renewal potential,16-19 suggesting a higher capacity for reconstitution of hematopoiesis. Colony-forming unit (CFU) assays demonstrated that CB hematopoietic stem cells differentiate along granulocyte, erythrocyte, monocyte, and megakaryocyte lineages.20-23 Transfusion of human CB in immunodeficient severe combined immunodeficiency (SCID) mice demonstrated repopulation of the BM with clonogenic progenitors, which support development of erythroid, myeloid, and B-cell lineages.24-26 Although CB is a mixture of primitive and more committed hematopoietic progenitors, only the most primitive CD34⁺CD38⁻ cell population allows for long-term multilineage engraftment in nonobese diabetic (NOD)/SCID mice.27 Whether CD34⁺CD38⁻ CB stem cells can also reconstitute T-cell lymphopoiesis in vivo is presently unknown. Understanding the T-cell developmental potential of CB progenitors is important to determine whether transplanted patients will recover T-cell–dependent primary antibody responses and immunity against environmental pathogens.

Human CB CD34⁺ cells injected into transplanted fetal thymus or fetal thymic organ culture can give rise to mature thymocytes.28-30 Although these studies suggest the presence of T-cell progenitors, they do not establish whether such progenitors exist in the CB as terminally committed cells or more primitive elements. Moreover, these studies cannot address the question of whether T-cell progenitors traffic from the BM to the thymic compartment. To address these issues, we developed an in vivo model of continuous human T-cell lymphopoiesis. In this model, the BM compartment of NOD/SCID mice is reconstituted with T- and B-cell–depleted human CB, followed by transplantation of human fetal thymus under the kidney capsule. Implicit to this model is the requirement of primitive human CB progenitors to engraft in the BM compartment and sustain T-cell lymphopoiesis by regulated homing of their progeny to the thymus. Furthermore, a functional vascular connection between the mouse tissue and the human thymic graft must develop to mediate progenitor homing and the exit of mature T cells to the periphery. Therefore, we extended our studies to investigate the development of this critical vascular interface at the site of thymic transplants.

Here we report that CB-derived hematopoietic stem cells, engrafted in the BM of NOD/SCID mice, generated T-cell progenitors capable of homing to and developing within a human thymic graft transplanted at a distant site. Human thymopoiesis led to the exit of mature T cells to peripheral lymphoid compartments. These results demonstrate that human CB stem cell transplantation supports thymic T-cell development. Surprisingly, BM reconstitution with CB also led to engraftment of CB-derived progenitors of the endothelial lineage, which were recruited to form new blood vessels extending from the thymic implant into the surrounding mouse tissue. This finding shows that human CB may also represent a unique reservoir of endothelial progenitors and suggests a novel mechanism of neovascularization involving the participation of circulating endothelial stem cells.

NOD/LtSz-SCID mice (NOD/SCID) were obtained from our colony (original founders a kind gift of Dr L. Shultz, Bar Harbor Laboratories, Bar Harbor, ME). The colony was derived by cesarean section under specific pathogen-free conditions. Mice were maintained on an irradiated sterile diet and autoclaved acidified water.

Sterile human CB was obtained from normal deliveries at the Mary Birch Hospital for Women (San Diego, CA). Samples were processed within 24 hours. Mononuclear cells were first separated on a Ficoll gradient. CB cells were depleted of mature T and B cells by negative selection using an immunomagnetic separation technique. Cells were with anti-CD3 monoclonal antibody (MoAb) (OKT3; American Type Culture Collection [ATCC], Bethesda, MD) and anti-CD19 MoAb (HD37, Dako, Carpinteria, CA) for 30 minutes at 4°C followed sequentially by a biotin-conjugated goat anti-mouse IgG (Caltag, Burlingame, CA), fluorescein isothiocyanate (FITC)-conjugated streptavidin (GIBCO, Gaithersburg, MD) and biotin-magnetic beads (Miltenyi Biotec, Sunnyvale, CA). The cells were applied to a VarioMacs magnetic column (Miltenyi Biotec, a kind gift of Dr Kevin Mills) and nonadherent cells collected. This method consistently yielded a cell purity > 95% as determined by double-color immunostaining using an R-phycoerythrin (RPE)-conjugated anti-human T-cell receptor (TCR)α/β (BMA031; Immunotech, Westbrook, ME) and an FITC-conjugated anti-CD20 MoAbs (B-Ly1; Dako). Cell preparations contained less than 5% CD20-positive contaminants, but no detectable TCRα/β-positive cells. The proportion of CD34⁺ progenitors in each CB sample was finally assessed with an FITC-conjugated anti-CD45 MoAb (Becton Dickinson, Bedford, MA) and a biotin-conjugated anti-CD34 MoAb (QBEnd10, Immunotech) followed by Cy-Chrome-streptavidin (Pharmingen, San Diego, CA). CD34⁺ cells represented 2% to 5% of total CB cells after T- and B-cell depletion.

Fetal thymic tissue (18 to 24 weeks of gestation) was obtained from ABR (Alameda, CA). To deplete the tissue of endogenous thymocytes, the thymi were minced in ≈1 mm × 1 mm pieces and cultured in Iscove medium containing 10% fetal calf serum (FCS), penicillin/streptomycin, and 1.35 mmol/L deoxyguanosine (dGuo) (Sigma, St Louis, MO) for 5 to 6 days.

NOD/SCID mice (6 to 8 weeks old) were sublethally irradiated (300 cGy) and injected in the tail vein with T- and B-cell–depleted CB cells. The number of injected cells was adjusted to deliver 100 to 200 × 10³ CD34⁺ progenitors per mouse. After 4 to 6 weeks, the mice were transplanted under the kidney capsule with the human thymic fragments (2 to 3 pieces/mouse). At 2 and 4 months postthymic transplant, the mice were euthanized. Bone marrow cells were recovered by flushing femurs and tibia with RPMI-10% FCS. The thymic implants and spleens were excised and either embedded for histologic analysis or cells were prepared by straining through a 100-μm stainless steel mesh. Peripheral blood lymphocytes (PBLs) were obtained by cardiac puncture and separated from red blood cells on a Ficoll-Paque gradient (Pharmacia, Uppsala, Sweden).

BM, thymic, splenic, and PBL cells were incubated for 15 minutes at 4°C in the presence of polyclonal mouse IgG (50 μg mL⁻¹) and anti-mouse Fc MoAb (2.4G2; Pharmingen). Lymphocytes were then stained using the following antibodies: FITC anti-mouse H2k^d (SF1-1.1, Pharmingen); RPE anti-human HLA-A,B,C (G46-2.6; Pharmingen); an FITC anti-human CD45 MoAb (2D1; Becton Dickinson), an RPE anti-human CD4 MoAb (13B8.2; Immunotech); a Cy-Chrome anti-human CD8 (RPA-T8; Pharmingen), or an FITC anti-human CD8 (B9.11; Immunotech); an RPE anti-human TCR TCRα/β (BMA031; Immunotech) and FITC, RPE, or CyChrome species and isotype matching control antibodies. In preliminary experiments, only minimal cross-reactivity to mouse determinants was observed for the MoAbs, whereas the anti-human CD45 and HLA-A,B,C MoAbs showed no cross-reactivity. Therefore, these antibodies were used to gate for human lymphocytes. The samples were analyzed with a FACScan (Becton Dickinson). In some experiments, single- and double-positive thymocytes were stained with FITC anti-human CD8 and RPE anti-human CD4 MoAbs and purified by cell sorting (FACStar Plus; Becton Dickinson).

Ten-micrometer cryostat tissue sections were fixed in phosphate-buffered saline (PBS) 2% paraformaldehyde (PFA) for 10 minutes at room temperature. Sections were permeabilized in PBS/0.1% Triton-X 100 for 5 minutes, blocked first in PBS/50 mmol/L glycine and then in PBS/10% goat or donkey serum for 1 to 2 hours. For immunohistochemistry, sections were treated for 5 minutes with a 0.05% H₂O₂ solution for saturation of the endogenous peroxidase activity and then probed with a human-specific anti-CD45 MoAb (T29/33; Dako) followed by a biotinylated human-adsorbed goat anti-mouse IgG Fab₂ (Caltag) and Peroxidase-streptavidin (GIBCO).

For the detection of human lymphoid and endothelial cells, thymic sections were probed with anti-CD45 MoAb and Lissamine-Rhodamine (LSRC) anti-mouse IgG (Jackson Labs, West Grove, PA) or a combination of a mouse anti-CD34 MoAb (8G12; Becton Dickinson) and a goat anti–PECAM-1 (platelet endothelial adhesion molecule-1) (polyclonal IgG M-20; Santa Cruz Biotechnology, Santa Cruz, CA) followed by an LSRC anti-mouse IgG Fab₂ and an FITC anti-goat IgG Fab₂ (Jackson Labs). In some experiments, an anti-human–specific von Willebrand Factor (vWF) MoAb (Takara, Shiga, Japan; clone vW1-2) was used. The anti–HLA-A2 antibody (clone MI2.1) was the kind gift of Dr H. Kaneshima (Systemix, Palo Alto, CA). The secondary reagents were preadsorbed to eliminate cross-species reactivity. In some experiments, NOD/SCID mice reconstituted with T- and B-depleted CB cells or human umbilical vein endothelial cells (HUVECs) labeled with the vital dye DiI (Molecular Probes, Eugene, OR) were used as hosts for thymic transplants. In these mice (n = 4), blood vessels developed at the interface with the thymic transplants and comprising endothelial cells of CB origin (ie, DiI⁺) were quantified by morphometric analysis. For this purpose, a total of 160 optical fields (324.8 μm × 216.5 μm) were scored for the presence of PECAM-1⁺ vascular profiles containing DiI⁺ cells.

For simultaneous detection of human CD45⁺ and apoptotic cells, air-dried thymic sections were fixed in PBS 2% PFA for 10 minutes at room temperature. Sections were incubated with digoxigenin-labeled deoxyuridine triphosphate (dUTP) and terminal deoxynucleotidyl transferase (TdT) enzyme (Oncor, Gaithersburg, MD) for 1 hour at 37°C. After blocking the enzymatic reaction, the sections were incubated with an FITC sheep antidigoxigenin antiserum (Oncor) for 30 minutes at room temperature. The sections were blocked overnight in PBS/10% donkey serum and sequentially probed with the anti-CD45 MoAb and an LSRC donkey anti-mouse IgG preabsorbed with sheep IgG (Jackson Labs). Sections were analyzed on a Zeiss Axiovert microscope with a scanning laser confocal attachment (MRC 1024; BioRad, Hercules, CA).

To identify male BM-derived T cells in the thymic grafts, total or sorted thymocytes were lysed in 10 mmol/L Tris-HCl, 1.5 mmol/L MgCl₂, 50 mmol/L KCl, 0.5% Tween 20, 100 μg/mL⁻¹ Proteinase K (Boehringer Mannheim, Indianapolis, IN), pH 8.3, for 45 minutes at 56°C. After inactivation of Proteinase K by heating the samples at 95°C for 10 minutes, an aliquot of each sample (DNA from 10,000 cells) was used for PCR amplification of human Y chromosome sequences. The Y chromosome–specific probes were: forward 5′-TGGGCTGGAATGGAAAGGAATCGAAAC-3′ and reverse 5′-TCCATTCGATTCCATTTTTTTCGAGAA-3′.31 PCR was performed for 25 cycles at 95°C for 1 minute, 65°C for 1 minute, and 72°C for 1 minute, and a 10-minute final extension at 72°C. PCR products were separated on 1.2% agarose gels and visualized by ethidium bromide.

To investigate the presence of male BM-derived stromal cells at the site of thymic engraftment, sections were first stained with an anti-human CD34 MoAb to localize human vessels. Based on this localization, samples were then scraped from consecutive sections using a 30 G needle under microscopic visualization.32 Samples were collected at 3 sites: within the thymic graft, at the interface of the transplant with the mouse kidney, and within the kidney parenchyma distant from the graft. The scraped tissue was placed in 50 μL of lysing buffer, DNA was extracted, and used as template for PCR amplification of human Y chromosome-specific sequences.

To simultaneously identify cells bearing human Y chromosome and CD4, CD8, CD34, or PECAM-1 in tissue sections, we adapted the protocol described by Gerritsen et al33 for 2-color FISH. Briefly, 10-μm cryostat sections were fixed in acetone for 7 minutes at −20°C. The sections were blocked for 1 hour at room temperature in PBS/10% donkey serum and then probed sequentially with anti-CD4/CD8 MoAbs (OKT4, OKT8), or anti-CD34 MoAb (8G12) followed by LSRC or CY5 donkey anti-mouse IgG absorbed to sheep IgG. In some experiments, sections were probed with anti–PECAM-1 followed by a biotin donkey anti-goat IgG and Cy5 streptavidin. In this case, free binding sites of the secondary reagent were blocked with sheep IgGs (50 μg/mL⁻¹). Sections were fixed for 10 minutes at room temperature in PBS/1% PFA and dehydrated through graded ethanol solutions (70%, 80%, and 100%). Thirty microliters of hybridization mixture (Hybrisol V, Oncor) containing the digoxigenin-labeled Y chromosome-specific DYZ1 and DYZ2 probes (Oncor) was then applied to the sections. The slides were sealed with a glass coverslip, heated to 80°C for 10 minutes, and hybridized for 16 hours in a humidified chamber at 37°C. After washing in 1X sodium citrate buffer (SSC) for 5 minutes at 72°C, the sections were incubated in the presence of FITC sheep antidigoxigenin for 30 minutes at room temperature. Sections were washed 3 times and analyzed by confocal microscopy.

To study the capability of CB-derived BM progenitors to sustain human thymopoiesis in vivo, NOD/SCID chimeras were generated by reconstitution of the murine BM compartment with human CB depleted of mature T and B cells, followed by transplantation of human fetal thymus under the kidney capsule. Control mice received only thymic transplants. We refer to these two groups of animals as CB/Thy-NOD/SCID and Thy-NOD/SCID chimeras, respectively. At 4 and 6 months post-CB reconstitution, the BM and thymocytes from the thymic transplants were harvested and analyzed by flow cytometry for the presence of human cells expressing the leukocyte common antigen CD45.

Figure 1a shows human CD45⁺cells detected in the BM and thymic grafts of CB/Thy-NOD/SCID chimeras at 4 months post-CB reconstitution. As many as 80% and 95% of the lymphoid cells populating the BM and thymic grafts, respectively, were human. At 6 months, a decrease in the proportion of human cells was observed in the BM compartment (Fig 1b, %CD45⁺ cells: 7.7 ± 2.8 at 6 months v 51.3 ± 6.5 at 4 months, mean ± standard error of mean [SEM], n = 11). However, overall, the percentage of human CD45⁺ cells in the thymic transplants remained unchanged (Fig 1b; %CD45⁺ cells: 55.8 ± 13 at 6 months v 57.3 ± 10.4 at 4 months, mean ± SEM, n = 11). Indeed, by 6 months, the thymic grafts had grown considerably from the original size (eg, ≈2 mm³) at the time of implantation to ≈250 to 500 mm³. Moreover, the comparison of thymic cell numbers at 4 and 6 months (Fig 1c and d) demonstrates an increased number of human cells. In contrast to the significant growth of thymic tissue in CB/Thy-NOD/SCID chimeras, the grafts of control Thy-NOD/SCIDs displayed significantly lower cellularity (CD45⁺ cells: 3.1 ± 0.8 × 10⁶v 12.1 ± 4.3 × 10⁶; mean ± SEM; n = 11, P = .02).

Three-color immunostaining for human CD45, CD4, and CD8 showed that the thymic grafts 4 months postthymic transplant (eg, 6 months post-CB reconstitution) comprised thymocytes at both double- and single-positive stages of maturation (Fig2a through d). Notably, most thymocytes in the grafts of CB/Thy-NOD/SCIDs were human (Fig 2a). Overall, they comprised a significantly higher percentage of CD4 single-positive cells than thymocytes from Thy-NOD/SCIDs (Fig 2b, d, and e; %CD4⁺: 17.9 ± 3.3 v 6.8 ± 1.6, mean ± SEM, P < .01, n = 11). Conversely, in the absence of human BM, fewer human CD45⁺ cells were detected in the thymic grafts and the distribution of thymic subsets was abnormally skewed toward a CD8 single-positive phenotype (Fig 2c through e).

Further phenotypic analysis confirmed that human CD4 and CD8 single-positive thymocytes of either CB reconstituted or control mice expressed high levels of TCR α/β, HLA class I, and CD69 (not shown). This phenotype is consistent with that of terminally differentiated thymocytes, further evidence for a full thymic T-cell development program. However, thymocytes from Thy-NOD/SCID mice demonstrated a significantly lower viability as determined by light scatter analysis (eg, % viable TCR^high cells = 59 ± 7.7 in Thy-NOD/SCID v 81 ± 4.8 in CB/Thy-NOD/SCID mice [n = 6], mean ± SEM, P < .05) suggesting that development in these control grafts is incomplete and many cells undergo cell death. Finally, flow cytometry of lymphocytes from the NOD/SCID mouse thymi showed the presence of less than 2% human CD45⁺cells, indicating that human T cells developed preferentially in the human thymic microenvironment.

To study the morphology of thymic grafts, sections from human fetal thymi before transplantation and after engraftment in vivo were stained for human CD45 and/or mouse major histocompatibility complex (MHC) class I antigens. In agreement with previous reports,34human fetal thymi cultured in the presence of deoxyguanosine were composed mainly of an epithelial network, almost completely devoid of human CD45⁺ cells (not shown). After engraftment in mice reconstituted with human BM, the thymic transplants displayed a defined cortex and medulla densely repopulated by human CD45⁺ cells (Fig 3a). Few mouse cells were observed within the grafts (Fig 3b, red fluorescence, arrow). In contrast, grafts from Thy-NOD/SCIDs showed a profoundly perturbed architecture, with no clear demarcation of cortical and medullary regions (Fig 3d). Consistent with the lower cellularity of these transplants, the grafts showed large areas depleted of lymphocytes or harboring only dispersed human CD45⁺ cells. Clusters of mouse cells were often observed within these grafts (Fig3e, red fluorescence).

To investigate the occurrence of apoptosis in the developing thymic grafts, sections of thymic transplants were stained using a TdT-mediated dUTP nick-end labeling (TUNEL) assay. Thymic transplants from CB/Thy-NOD/SCID chimeras showed few apoptotic cells, mostly scattered in the cortex (Fig 3c, green fluorescence). In contrast, thymic grafts from the control Thy-NOD/SCIDs showed numerous apoptotic cells arranged in rosette-like clusters throughout the sections (Fig3f, green fluorescence) or grouped in foci of more than 50 cells around Hassals corpuscles (not shown). A rosette pattern of apoptotic cells has previously been described in pediatric thymi.35Two-color immunofluorescence to detect fragmented DNA and human CD45 demonstrated that the apoptotic cells were human (not shown). Morphometric analysis on more than 30 microscopic fields from sections cut at 100-μm intervals demonstrated that grafts from control Thy-NOD/SCIDs harbored a higher number of apoptotic events than those from mice reconstituted with human CB (number apoptotic cells per field = 0.99 ± 0.2 in Thy-NOD/SCIDs, n = 3 v 0.3 ± 0.1 in CB/Thy-NOD/SCIDs, n = 3, mean ± standard deviation [SD], P = .02; size of each field = 249 × 217 μm). This result supports the hypothesis that a high number of thymocytes developing in the grafts of the Thy-NOD/SCID mice die in the thymus and is consistent with the increased proportion of nonviable thymocytes shown by our fluorescence-activated cell sorting (FACS) analysis.

Taken together, the data indicate that on reconstitution of a human BM hematopoietic compartment by CB cells, human thymic grafts transplanted at a distant site maintain their thymopoietic capability. The data suggest that CB-derived stem cells generated thymocyte progenitors capable of homing to and developing within the human thymic grafts.

The fact that some human CD45⁺ cells were detected in thymic grafts of control Thy-NOD/SCID mice indicated that some lymphoid cells of thymic donor origin had survived the dGuo treatment and may have contributed to the cellularity of the thymic grafts in the CB/Thy-NOD/SCIDs. To unequivocally demonstrate the presence of CB donor-derived thymocyte progenitors in the grafts of CB/Thy-NOD/SCIDs, chimeras were generated with male human CB cells and transplantation of female fetal thymi. At 4 months postthymic transplant, the CD4/CD8 double- and single-positive subsets from the thymic grafts were sorted and used for PCR amplification of Y chromosome DNA-specific sequences. These experiments showed that thymic transplants from CB/Thy-NOD/SCID mice harbored immature double-positive as well as mature single-positive thymocytes of CB origin (Fig 4). Thus, CB-derived thymocyte progenitors indeed home to the thymic grafts and develop through intermediates to the most mature single-positive stages. Parallel experiments using NOD/SCID mice transplanted with CB and thymi mismatched for HLA-A2, confirmed that as many as 95% of thymocytes repopulating the thymic transplants were of CB origin, as determined by FACS analysis for CD3, HLA-A2, CD4, and CD8.

To assess whether reconstitution of human thymopoiesis in the NOD/SCID chimeras led to repopulation of the periphery by mature T cells, spleens and peripheral blood were screened for the presence of human CD45⁺ TCR α/β⁺ T cells by immunohistochemistry and/or flow cytometry. Human CD45⁺cells filling the periarteriolar T-cell–dependent areas of the spleen were detected in mice reconstituted with human CB (Fig 5a), but not in control mice (Fig 5b). Flow cytometry of splenocytes demonstrated that human CD45⁺ cells represented 7.34% ± 1.6% (mean ± SEM, n = 5) of the total lymphoid cells populating the spleen of CB/Thy-NOD/SCIDs. Immunofluorescence of either splenocytes or PBL for human CD45 and TCR α/β or human CD45, CD4, and CD8 further demonstrated that the majority of circulating human lymphocytes in these mice were, in fact, T cells with mature CD4 and CD8 phenotypes (Fig 5c; %TCR α/β⁺ cells within the CD45 subset = 62.3 ± 7.8, n = 9). In contrast, less than 0.5% human T cells were detected in the spleen or peripheral blood of control Thy-NOD/SCID mice.

To directly demonstrate that circulating T cells were BM-derived and therefore of CB origin, splenic sections from the chimeras reconstituted with male CB and female thymi were analyzed by FISH for the presence of cells expressing the human Y chromosome and the T-cell markers CD4 and CD8 (Fig 5d). Numerous cells coexpressing male Y chromosome (green fluorescence) and CD4/CD8 markers (red fluorescence) were observed. Notably, not all of the cells expressing the CD4/CD8 markers displayed a positive signal for the Y chromosome. This result may reflect a limitation of this technique to detect chromosomes on a single focal plane imposed by the tissue sectioning. Similar results were obtained in 2 other chimeras. Unlike SCID mice reconstituted with T- and B-cell–depleted CB followed by transplantation with thymic tissue, we never detected circulating mature T cells in SCID mice without thymic transplants, regardless of whether they were reconstituted with either whole CB (n = 50)36 or purified CD34⁺ cells (n = 20) (B.E.T., unpublished observations). Altogether, these results indicate that colonization of the thymic grafts by CB-derived thymocyte progenitors generated a mature T-cell progeny capable of exit and repopulation of peripheral immune compartments.

Revascularization plays an important role in the engraftment of transplants. In the case of thymic grafts, the newly formed vasculature is also a primary component regulating the homing of progenitors and possibly release of mature T cells from this organ.37Because these processes of lymphocyte trafficking may be restricted by species-specific adhesive interactions on the endothelium,38 we next investigated the extent to which host and/or donor endothelium contributed to the revascularization of the thymic transplants. For this purpose, sections were stained with an anti-human CD34 MoAb and an anti–PECAM-1 polyclonal antibody. In experiments testing the species specificity of these antibodies, we found that the anti–PECAM-1 antibody cross-reacts with human and mouse endothelium, whereas the anti-CD34 MoAb is human-specific. Figure 6 shows a series of microscopic fields from a thymic graft of a CB/Thy-NOD/SCID chimera. Numerous human blood vessels identified by the coexpression of CD34 and PECAM-1 (Fig 6a and b; coexpression marked by yellow) were observed in the subcapsular and interlobular connective tissue (Fig6a, arrows). Many human vessels were also identified infiltrating the mouse kidney parenchyma directly adjacent to the graft (Fig 6b). To determine the distance human vessels could be detected within the surrounding mouse tissue, we analyzed a series of consecutive microscopic fields (Fig 6d). Human vessels were detectable within a range of ≈700 μm from the edge of the thymic graft. Thymic transplants of control Thy-NOD/SCID mice demonstrated a similar human vascular network, indicating that these vessels may form from endothelial elements contained within the transplant. This human vascular component provides a mechanism by which human marrow-derived T-cell progenitors could be effectively targeted to human thymic grafts at a distant site in our xenogeneic model.

Previous studies have shown that an adherent fraction of human CB cells can be induced to differentiate in vitro into endothelial cells.39 This observation suggested that human CB may contain circulating endothelial cells or endothelial progenitors. To test whether CB-derived cells of the endothelial lineage had engrafted in our NOD/SCID chimeras and participated in the revascularization of the thymic transplants, frozen sections from female thymi engrafted into mice reconstituted with male CB were screened for the presence of male cells at the interface of the mouse kidney with the thymic grafts. This anatomically discrete region contained human CD34⁺blood vessels (Fig 6), but lacked detectable human CD45⁺lymphoid cells as assessed by confocal microscopy (not shown). The presence of human Y chromosome–positive cells at this site was first investigated by a PCR technique, which allows amplification of gene sequences from tissue sections.31 Tissue samples from thymic graft/kidney interfaces were dissected from frozen sections under microscopic visualization. DNA was extracted and used for PCR of human Y chromosome–specific sequence (Fig 7). DNA from regions within the thymic grafts containing thymocytes of BM origin (eg, male) served as positive controls. In addition, to exclude the possibility of detecting Y chromosome–positive cells of marrow origin circulating in the mouse blood stream, DNA from the mouse kidney distant from the grafts was also used. Figure 7a documents the tissue sampling sites from a representative tissue section. Human Y chromosome DNA was detected at the interface of the thymic grafts with the mouse kidney, as well as within the thymic grafts, but not within the mouse kidney at sites distant from the graft (Fig 7b). The data indicate that CB-derived cells are resident at the interface of the mouse kidney with the thymic implant in a region in which no CD45⁺ human cells (eg, leukocytes) are detected.

To determine whether these cells at the graft interface comprised endothelial cells, we used the in situ fluorescence hybridization assay described above to identify cells coexpressing the endothelial markers CD34, PECAM-1, or vWF and human Y chromosome. In 3 of 5 grafts, this analysis showed the presence of stromal elements coexpressing CD34 or PECAM-1 and human Y chromosome (Fig 7c through e) or human vWF and human Y chromosome (Fig 8a and b). CD34⁺ cells bearing the Y chromosome were identified at the interface of the grafts with the mouse kidney (Fig 8a, arrows) and at a subcapsular location (Fig 8b). Figure 7e shows the interface of a graft with the mouse kidney stained for PECAM-1 (red fluorescence) and Y chromosome (green fluorescence). Cells coexpressing the 2 markers can be observed within a blood vessel (arrowheads), indicating the presence of CB-derived endothelial cells. Other Y chromosome⁺ cells negative for PECAM-1 (arrows) are also evident in this region. They appear as large fibroblast-like cells by light microscopy. Y chromosome⁺ endothelial or stromal cells were not observed within the thymic parenchyma (not shown). The presence of blood vessels of cord origin was further demonstrated by studying SCID chimeras transplanted with CB and thymic tissue mismatched for HLA-A2. Figure 8c and d shows microscopic fields at the thymic-kidney interface of 1 of such chimeras stained with an anti-human–specific vWF antibody (red fluorescence) and an anti–HLA-A2 specific MoAb (green fluorescence) identifying cells of thymic origin. As can be observed, blood vessels of both CB origin (eg, positive for vWF, but negative for HLA-A2, arrowheads) and thymic origin (eg, positive for both vWF and HLA-A2, arrows) are present at this site.

These results indicate that unlike the human vessels described in the control Thy-NOD/SCIDs, which must be of thymic graft origin, recruitment of CB-derived cells of the endothelial lineage occurred in the CB/Thy-NOD/SCIDs. Quantitative histology using T-and B-depleted CB cells labeled with a fluorescent vital dye indicated that 40% to 60% of PECAM-1⁺ vascular profiles identified at the interface with the thymic grafts comprised cells of CB origin. We considered whether the formation of CB-derived vessels could be attributed to mature endothelial cells, which might contaminate our CB preparations. Thus, T- and B-depleted CB samples (n = 3) were analyzed by flow cytometry for cells expressing the mature endothelial markers PECAM-1, E-selectin, and vWF. These experiments showed that nonmyeloid (eg, CD11b⁻) PECAM-1⁺ cells, possibly comprising mature endothelial cells, represented less than 0.5% of T- and B-depleted CB. Purification of this subset by positive selection on a magnetic column demonstrated that only 5.7% of CD11b⁻/PECAM-1⁺ cells expressed vWF (ie, 0.02% of T- and B-cell–depleted CB cells) and could therefore be unequivocally identified as mature endothelial cells. Thus, at most, ≈3,000 vWF⁺ cells were injected per mouse in our reconstitution experiments. Further experiments in which as many as 2 × 10⁶ HUVEC labeled with a vital dye were injected in our NOD/SCID model demonstrated that mature endothelial cells do not efficiently contribute to the development of blood vessels at the site of thymic transplantation (manuscript in preparation). Taken together, these experiments indicate that CB-derived endothelial cells present in the vessels between the mouse kidney and the transplants were derived from circulating progenitors trafficking into this site of surgical injury and subsequent healing.

In these studies, we demonstrate that CB hematopoietic progenitors, engrafted in the BM compartment of immunodeficient hosts, sustain human T-cell lymphopoiesis by generating circulating progenitors capable of homing to and developing within a human thymus. Furthermore, we provide evidence that the in vivo developmental program of CB stem cells is not limited to blood cell lineages, but also extends to the generation of endothelial cells, as well as stromal elements, which contribute to the revascularization of transplants and wound healing. This result discloses a novel role of CB-derived stem cells in angiogenesis.

Immunodeficient SCID and NOD/SCID mice are valuable in vivo models to study human hematopoiesis.40-43 In particular, studies have shown that intravenous injection of human BM or whole CB in these mice results in the engraftment of primitive hematopoietic stem cells of myeloid, erythroid, and B-cell lymphoid lineages.24-27 We show that injection of CB rigorously depleted of mature T and B cells consistently results in high levels of engraftment of human cells in the BM compartment. This observation indicates that mature T and/or B cells from CB are not required for initial engraftment of human stem cells.44,45 

Bone marrow reconstitution was demonstrated up to 6 months post-CB injection, indicating long-term in vivo self-renewal of the engrafted stem cells. However, by 6 months, exhaustion of the human BM component was evident in most of our chimeras. Several factors may explain this observation. First, it may reflect a limitation of the murine environment to support human hematopoiesis. Second, this result may indicate a propensity of primitive CB stem cells to proliferate at high rate in vitro,15,18 but for a limited number of cycles in vivo. Third, the purification procedure may deplete the CB of the most primitive progenitors or this population does not efficiently home to the BM compartment. Late graft failure after initial engraftment has been associated with T-cell–depleted allogeneic BM transplantation (BMT) in human patients.46 To date, only 1 report has investigated the repopulation capability of purified human CD34⁺ progenitors in NOD/SCID mice.26 While engraftment of CD34⁺ progenitors could be demonstrated up to 4 months postreconstitution, further experiments testing for the persistence of primitive hematopoietic progenitors by serial BM transplants showed much lower levels of BM engraftment in secondary as compared with primary recipients after 6 months (eg, 6% to 10% human cells in secondary recipients v 51% in primary recipients). These results suggest a decline in the frequency and/or engrafting capability of primitive progenitors after reconstitution with purified hematopoietic progenitors from CB.

A key result in our experiments is that engraftment of CB-derived BM progenitors supported human T lymphopoiesis in animals transplanted with human thymic grafts. In the presence of BM input, the thymic transplants increased in cellularity, preserved a relatively normal architecture, and supported thymocyte survival and development. Moreover, human thymopoiesis led to the production of mature T cells of CB origin that populated the periphery. These results represent the first evidence that CB stem cells comprise progenitors of the T-cell lineage capable of supporting continuous thymopoiesis in vivo from circulating elements. Recent studies have demonstrated that primitive hematopoietic progenitors from CB (eg, CD34⁺CD38⁻) are required for long-term engraftment of human cells in the NOD/SCID BM.27 In contrast, more committed CD34⁺CD38⁺ progenitors appear not to engraft, as they are not detected 4 weeks after transplantation.27 This suggests that circulating thymic progenitors, which colonize the thymic grafts in our model, arise from primitive progenitors in the BM. The requirement of a thymic tissue for the generation of mature circulating T cells in our chimeras is consistent with ours as well other previous reports showing a lack of T-cell lymphopoiesis in SCID mice transplanted with purified CD34⁺ CB cells,36,47 BM, or fetal liver hemopoietic progenitors.48 Our observation has important implications for the use of CB in clinical transplantation. For example, it indicates that CB stem cells may successfully reconstitute pediatric recipients with naive thymus-derived T cells and reestablish a normal T-cell repertoire after ablative radiotherapy or chemotherapy. In light of our results and other transplant models,48,49CB reconstitution in combination with thymus transplantation may also be envisaged as a strategy to regenerate thymic-derived T-cell lymphopoiesis in aging, human immunodeficiency virus (HIV)-induced immunodeficiency, or autoimmune diseases.

T-cell lymphopoiesis in our model was dependent on the successful trafficking of circulating BM-derived progenitors to the human thymic graft. The contribution of human vessels to the revascularization of the thymic grafts in our chimeras may have played an important role in this process. Indeed, evidence that human hematopoietic cells can only establish weak adhesive interaction with the mouse endothelium has been previously reported.38 In contrast with the successful progenitor cell homing observed in our model, previous in vivo studies of human thymopoiesis using fragments of human fetal liver and fetal thymus showed that progenitors from the fetal liver grafts transplanted under the kidney capsule of SCID mice fail to reconstitute thymopoiesis in thymic tissue engrafted under the capsule of the opposite kidney.40 These data indicated a defect of progenitor homing in that model.

We show that CB-derived cells contribute to the generation of new blood vessels at sites of graft implantation and wound healing. This result provides evidence for the existence of a previously unrecognized source of endothelial progenitors within human CB. That such cells have characteristics of progenitors is supported by their ability to engraft and survive in vivo, while retaining the ability to home and develop into blood vessels at sites of wound healing. These data provide the first documentation that the developmental potential of CB stem cells in vivo is not restricted to hematopoietic lineages. The coexistence of hematopoietic and endothelial progenitors in CB may not be fortuitous. Much evidence supports a developmental and/or functional relationship between endothelial cells and hematopoietic precursors during embryogenesis. Thus, formation of blood vessels progresses simultaneously with hematopoiesis in the blood islands of the yolk sac; the center of these islands form primitive blood cells and the outer cell layers develop into endothelial cells.49,50 Peptides secreted by the developing thymic epithelium such as thymosin α₁ may promote differentiation of immature thymocyte precursors51 and function as angiogenic factors for endothelial cells.52 The umbilical cord itself may represent a unique site of hematopoiesis and endothelial cell development. Hence, in the mouse embryo, the endothelium of the proximal portion of umbilical arteries display an unusual rounded morphology with clusters of CD34⁺ cells bound to the luminal surface.53 This pattern is reminiscent of that described for the paraaortic hematopoietic sites in the avian embryo.54 It suggests that the umbilical cord may comprise sites in which endothelial and hematopoietic progenitors coexist and be released into the embryo’s circulation.

The formation of new vessels from circulating progenitors contrasts with the current view that neovascularization of tissues in adult life occurs only by angiogenesis—the local proliferation of preexisting endothelial cells.55 The evidence we present for a role of circulating progenitors in neovascularization rather implies mechanisms of endothelial differentiation similar to those occurring during vasculogenesis, the process by which preendothelial cells condense to form new vascular channels in the embryo.55 Interestingly, a similar mechanism of vessel formation from circulating endothelial progenitors was recently shown in a model of limb ischemia. In this study, immunodeficient mice, injected with human CD34⁺cells purified from peripheral blood of adults, develop new human blood vessels in the ischemic tissue.56 The development of BM-derived endothelial cells has been also documented in dogs transplanted with allogeneic BM.57 Thus, the presence of circulating endothelial progenitors and their unique mode of vessel formation appears not to be limited to fetal life. Indeed, our studies suggest that reparative/inflammatory processes such as those occurring at sites of surgical injury and wound healing may provide stimuli for the mobilization, homing, and/or differentiation of such endothelial progenitors. Thus, the physiologic role of stem cells documented to circulate in the CB of newborns and peripheral blood of adults may not be restricted to hematopoiesis, but also extend to healing processes such as those following birth trauma, accidental tissue injury, and surgery.

We thank the medical staff of the Mary Birch Hospital for the procurement of the CB and Dr Alberto Hayek for instructing us on murine microsurgery. We also thank Drs Zaverio Ruggeri and Thomas Edgington for critical reading of this manuscript. This is TSRI manuscript 11495-MEM.