Functional selectin ligands mediating human CD34⁺ cell interactions with bone marrow endothelium are enhanced postnatally

Hematopoietic progenitor cells (HPCs) can home to the bone marrow (BM) after a simple intravenous injection, but the adhesive mechanisms mediating the initial interactions of human HPCs with the BM endothelium have not been evaluated in vivo. Using fluorescence intravital microscopy and homing assays in NOD/SCID mice, we show that endothelial selectins are necessary for human adult CD34⁺ cell homing, since rolling on BM endothelium and retention in the BM compartment are drastically reduced (>90%) in endothelial selectin–deficient NOD/SCID mice. Comparative analyses of CD34⁺ cells collected from adults and from cord blood (CB) reveal that neonatal cells display reduced rolling fractions compared with adult CD34⁺ cells obtained from peripheral blood or BM, suggesting abnormal selectin ligand function on neonatal progenitors. Flow cytometric and intravital microscopy studies suggest that this defect results from nonfunctional P-selectin ligand on a subset (∼30%) of neonatal CD34⁺ cells. Further analyses indicate that P-selectin glycoprotein ligand-1 (PSGL-1) is expressed in a nonfunctional form among neonatal CD34⁺ cells that do not bind P-selectin and that this subset is enriched in primitive CD34⁺CD38^lo/– progenitors. These results underscore the potential to improve homing of CB CD34⁺ cells to the BM by manipulation of selectins and their ligands.

Hematopoietic progenitor cells (HPCs) can be transplanted to the bone marrow (BM) of another individual following a simple intravenous infusion. The ability of HPCs to home to the BM was first demonstrated a few decades ago when Jacobson and colleagues showed that shielding of the spleen allowed mice to recover from a lethal dose of radiation (1). This study, along with others (2, 3), established that pluripotent stem cells present in the spleen or infused exogenously could repopulate distant hematopoietic organs. This feature is the basis for the clinical use of BM transplantation in treatment of hematologic malignancies (4). However, despite an advanced knowledge of the mechanisms that allow the migration of mature leukocytes into inflamed tissues, the mechanisms of human HPC homing to the BM are still poorly understood.

The migration of leukocytes to sites of inflammation is initiated by labile but critical adhesive interactions (attachment and rolling) that are largely mediated by the selectin family and its glycoconjugated ligands. Two members of the selectin family, P- and E-selectins, are expressed on endothelial cells (5, 6). P-selectin is stored in granules of endothelial cells and platelets and is rapidly translocated to the cell surface after stimulation with various secretagogues. E-selectin expression is induced by endotoxin or inflammatory cytokines. Although the E-selectin gene is silent in cultured endothelial cells, low levels of E-selectin are expressed in most tissues in vivo and regulate leukocyte homeostasis together with endothelial P-selectin (7). Mice lacking both P- and E-selectins display severe leukocytosis and can develop spontaneous skin infections (7, 8). The main ligand for P-selectin, P-selectin glycoprotein ligand-1 (PSGL-1), is a disulfide-bonded homodimeric mucin-like glycoprotein expressed on leukocytes (9), platelets (10), and CD34⁺ cells (11, 12). PSGL-1 requires specific posttranslational modifications in order to be functional. These include sialylation and fucosylation of O-linked sugars, as well as sulfation of tyrosine residues present in the N-terminus of the protein (13). While P- and E-selectin are critical for myeloid cell rolling in venules of the systemic circulation, the leukocyte selectin (L-selectin) mediates lymphocyte rolling in specialized venules of secondary lymphoid organs. The close interaction between leukocytes and endothelial cells during the rolling step allows the activation of leukocyte integrins triggered by chemokines present on the surface of the endothelium. This leads to the firm adhesion and migration of leukocytes through the vessel wall into the extravascular space (14). Whether a similar paradigm is applicable for immature hematopoietic precursors is still unclear. However, studies in mice suggest that the recruitment of HPCs into the BM requires multiple adhesion pathways, including endothelial selectins and VCAM-1 (15–18).

The study of primitive human hematopoiesis in vivo has been facilitated by the generation of NOD/LtSz-scid/scid mice (hereafter referred to as NOD/SCID mice) that have multiple defects in innate and adaptive immunologic functions (19). The BM of NOD/SCID mice can be repopulated by human HPCs (CD34⁺ cells) after sublethal doses of radiation (20). Clinically, human HPCs are obtained from three different sources: BM, mobilized peripheral blood (mPB), and cord blood (CB). CB-derived progenitors represent an especially promising source of human hematopoietic stem cells because they are widely available and their use is associated with reduced graft-versus-host disease (21). However, transplantation using CB cells has been restricted mostly to children due to the limited number of cells available in the placenta (22).

Here, we describe the molecular mechanisms mediating the initial adhesive interactions of human HPCs in the BM microcirculation of NOD/SCID mice using fluorescence intravital microscopy. We show that human CD34⁺ cell rolling on BM endothelium and extravasation into the BM compartment are completely dependent upon endothelial selectins. In addition, we demonstrate that the initial interactions of CB CD34⁺ cells are reduced compared with those derived from adult sources. We show that a large subset (∼30%) of neonatal CD34⁺ cells, enriched in primitive CD34⁺CD38^lo/– cells, does not bind P-selectin, and that this inability to bind P-selectin originates from incomplete posttranslational modifications of PSGL-1. These results may have implications for therapies using CB progenitor/stem cells.

Human CD34⁺ cells roll and arrest in NOD/SCID BM microvessels. The thin cortex of murine cranial bones allows one to observe the behavior of fluorescently labeled cells in the BM of living animals using epifluorescence intravital microscopy (17). Because of the protection afforded by the surrounding bone, the BM microvasculature can be observed without direct injury from the surgical preparation. The bulk of cranial BM is concentrated around the bone sutures of the calvaria. However, individual BM microvessels are not clearly visible in most areas by fluorescence intravital microscopy except for a section in the middle of the parietal bone (Figure 1a, box), which displays a well-defined vascular network composed of sinusoids (Figure 1b, “S”, honeycomb appearance) draining into at least two collecting venules that run perpendicular or parallel to the sagittal sinus (Figure 1b, Cv, and Figure 1, e–g). The blood flow in the midparietal area of the NOD/SCID BM is centripetal, draining into the sagittal sinus. To evaluate the distribution of active hematopoiesis in the calvaria, BM nucleated cells from EGFP-Tg mice were transplanted into sublethally irradiated (300 cGy) NOD/SCID recipient mice. Fourteen days after transplantation, EGFP-expressing hematopoietic cells had heavily repopulated the parasagittal areas of the parietal bone (including the aforementioned midparietal section; Figure 1, c and d), but also areas in the frontal and occipital bones near the coronal and lambdoidal sutures, respectively (not shown). These results indicate that the midparietal area carries active hematopoiesis and that the distribution of hematopoietic sites in the cranium of NOD/SCID animals is similar to that described in C57BL/6 animal studies using rhodamine 6G uptake (17).

Figure 1

Intravital microscopy of the parietal BM in NOD/SCID mice. (a) Low-power view (2× objective) of the murine parietal bone. The sagittal suture (SS), which joins both parietal bones, appears dark due to the blood circulating in the underlying sagittal sinus. The midparietal area (box) viewed at higher magnification (10× objective) in (b) represents the area of the parietal bone typically used for the IVM recordings. This area comprises several collecting venules (Cv) and sinusoids (S). (c and d) Engraftment of the NOD/SCID skull microvasculature with BM nucleated cells from EGFP-Tg mice. Sublethally irradiated NOD/SCID mice were transplanted with BM nucleated cells from EGFP-Tg mice. (c) Two weeks after injection, donor cells have engrafted the parasagittal regions and also the midparietal area (box) shown at higher magnification in (d). Note that the microvessels are masked by green fluorescent hematopoietic cells. (e–g) Human CD34⁺ cell rolling in a NOD/SCID collecting venule. CFSE-labeled CD34⁺ cells purified from mPB samples were injected via the carotid artery and visualized (10× objective) in the midparietal area of a NOD/SCID mouse. (e) A free-flowing CD34⁺ cell (t = 0 s) in the central Cv (f) tethers to the vessel wall (0.30 s), and (g) rolls during the following 0.53 seconds. The white arrow marks the initial site of interaction. Supplemental data of a video depicting the BM intravital microscopy model can be downloaded at http://www.jci.org/cgi/content/full/110/4/559/DC1. Bars in a and c, 500 μm; bars in b, d, and g, 100 μm.

Having established that the midparietal vascular network in NOD/SCID mice encompasses BM tissue, we next set out to evaluate the behavior of human HPCs in the BM microcirculation. CD34⁺ cells from G-CSF–mobilized healthy donors or from patients with hematologic malignancies in remission were isolated. Purified CD34⁺ cells were fluorescently labeled and injected via the carotid artery of NOD/SCID mice prepared for intravital microscopy. Labeled cells can be individually tracked in BM microvessels (Figure 1, e–g) and the percentages of rolling and arrested cells determined by careful analysis of video recordings. To comparatively evaluate CD34⁺ cell interactions in sinusoids and collecting venules, we initially studied the numbers of rolling and arrested HPCs in each vessel type separately. Although there was a trend toward a larger CD34⁺ cell rolling fraction in sinusoids compared with collecting venules, the difference was not significant (Figure 2; P = 0.14). Therefore, the numbers of rolling and arrested cells in both types of microvessels were pooled in subsequent analyses. Interestingly, most cell arrests in collecting venules (92 of 147, or 63% of arrested cells) occurred within two vessel diameters of vascular junctions, suggesting hotspots for the recruitment of HPCs in the BM. Although the average diameters and centerline velocities (Vmax) were lower in sinusoids than in collecting venules (Table 1, P < 0.05), the shear rates were similar between these two groups of microvessels. The mean percentage of rolling human CD34⁺ cells (∼22%), similar to that described for a murine cell line and fetal liver progenitor cells in the BM of C57BL/6 mice (17), indicates that human CD34⁺ cells can efficiently interact with NOD/SCID BM microvessels.

Figure 2

Human CD34⁺ cell rolling and arrest in collecting venules and sinusoids of the BM in NOD/SCID mice. CFSE-labeled human CD34⁺ cells purified from mPB were injected into NOD/SCID mice, and the fractions of cells rolling or arrested in collecting venules or sinusoids were determined by IVM analysis of videotapes. Bars represent mean ± SEM. The differences between Cv and sinusoids for the number of rolling and arrested cells are not significant (P = 0.14 and 0.6, respectively). n = 7 different human donors and the same number of mice.

Table 1

Hemodynamic characteristics of BM microvessels

P- and E-selectins are required for human CD34⁺ cell rolling in BM microvessels and homing to the BM compartment. Previous studies have suggested important roles for P- and E-selectins and VCAM-1 in mouse progenitor homing to the BM, and that each pathway (VCAM-1/β1 integrin and selectin/mucin ligands) contributes equally to recruitment of mouse HPCs (15–18). To evaluate the role of endothelial selectins and VCAM-1 in the initial interactions of human CD34⁺ cells with BM microvessels, we backcrossed the P- and E-selectin mutations into the NOD/SCID background and used an mAb (MK 2.7) to inhibit VCAM-1 function. CD34⁺ cells were purified from MNCs prepared from mPB samples and fluorescently labeled with CFSE. Mice were preinjected with either an anti–VCAM-1 Ab (MK 2.7) or rat IgG control prior to the injection of CD34⁺ cells. Anti–VCAM-1 administration reduced CD34⁺ cell rolling by about 33% (P = 0.05) in the BM microvasculature of wild-type NOD/SCID mice, compared with IgG injection or no injection (Figure 3). Strikingly, CD34⁺ cell rolling was drastically reduced (∼95% reduction) in BM microvessels of P/E^–/– NOD/SCID mice preinjected with either IgG or anti–VCAM-1 (Figure 3). These differences were not due to alterations in hemodynamics of the BM microcirculation since there was no difference in the shear rate between the two groups of mice (Table 1). These results indicate that, in contrast to murine progenitors (16, 17), the initial interactions of human CD34⁺ cells are largely dependent on endothelial selectins that are constitutively expressed in the BM tissue and that VCAM-1 plays a partial role in this activity.

Figure 3

Role of P- and E-selectins and VCAM-1 in the initial interactions of human CD34⁺ cells with BM microvessels. Fluorescently labeled human CD34⁺ cells purified from mPB were injected into P/E^–/– or wild-type mice backcrossed into the NOD/SCID background. Mice were injected via carotid catheter with control rat IgG (or no Ab in two experiments) and then with labeled CD34⁺ cells. After 15 minutes of recording, fluorescent cells were completely cleared from the circulation. The same mouse was then injected with anti–VCAM-1 and a second bolus of fluorescent CD34⁺ cells. Horizontal bars represent mean values. Each dot represents a single experiment before (filled squares) or after anti–VCAM-1 injection (open circles). *P = 0.05 by paired Student t test; **P = 0.003 by unpaired Student t test compared with IgG group in wild-type animals.

To determine whether the defect in the initial interactions of human CD34⁺ cells in P/E^–/– NOD/SCID mice prevents extravasation into the BM compartment (i.e., homing), unsorted MNCs from mPB were injected into wild-type or P/E^–/– NOD/SCID mice. Homed human cells were detected by FACS analysis using species-specific Ab’s against CD34 and CD45 (Figure 4, a and b). Homing of CD34⁺/CD45⁺ human cells was reduced by approximately 90% in P/E^–/– mice compared with wild-type animals (Figure 4, c–e). Of note, CD45⁺CD34^– lymphocytes also lodged in the BM of NOD/SCID mice 2 hours after their injection (Figure 4c, lower region), suggesting that donor lymphocytes also rapidly home to the recipient BM after transplantation. However, the numbers of lymphocytes were only modestly reduced (39%, P = 0.02) in the BM of P/E^–/– animals (Figure 4d). Taken together, our intravital observations and the results from the short-term homing assays indicate that endothelial selectins are necessary for human CD34⁺ cell rolling on BM microvessels and homing into the BM compartment of NOD/SCID mice.

Figure 4

Role of P- and E-selectins in the homing of human CD34⁺ cells to the BM of NOD/SCID mice. Sublethally irradiated NOD/SCID or P/E^–/– NOD/SCID mice were injected intravenously with 2 × 10⁷ human MNCs from mPB samples. Cells were allowed to home for 2 hours, and the femoral BM was harvested. (a) Double staining of the transplanted human MNCs for human CD34 and CD45. (b) BM cells recovered from an uninjected control mouse stained for human CD34 and CD45. (c) Wild-type NOD/SCID mouse transplanted with human MNCs. (d) P/E^–/– NOD/SCID mouse transplanted with human MNCs. (e) Number of human CD34⁺ cells homed per femur in transplanted wild-type NOD/SCID and P/E^–/– NOD/SCID mice. The percentage of gated CD34⁺/CD45⁺ human cells (upper region in the scattergrams in c–d) in the recipient BM was determined by FACS analysis, and the number of homed cells per femur was calculated based on the femoral cellularity. The number of homed human CD34⁺ cells is drastically reduced in P/E^–/– NOD/SCID mice. n = 6 mice injected with MNCs from two different human donors. *P = 0.0008.

Defective interactions of CB-derived CD34⁺ cells with BM microvessels. Human HPCs can be routinely harvested for clinical transplantation from three different sources, including the BM, mPB, and CB. To evaluate differential homing mechanisms of human HPCs, we isolated CD34⁺ cells from fresh samples obtained from these three sources. Fluorescently labeled CD34⁺ cells were injected via a carotid artery catheter, and their interactions in the contralateral midparietal BM were recorded. Detailed analyses of videotapes revealed that the mean percentages of rolling and arrested cells in NOD/SCID BM microvessels were similar between mPB- and BM-derived CD34⁺ cells (Figure 5). However, the interactions of CB-derived CD34⁺ cells were significantly altered (∼50% and ∼60% reduction for rolling and arrested cells, respectively) compared with those derived from adult sources (Figure 5). Hemodynamic parameters were similar among the three groups (Table 1). These results suggest a defect in the capacity of neonate-derived CD34⁺ cells to interact with the endothelium of BM microvessels.

Figure 5

Comparative analysis of the interactions of CD34⁺ cells with BM microvessels in NOD/SCID mice. Fluorescently labeled CD34⁺ cells from steady-state human BM, mPB, or CB were injected into NOD/SCID mice, and the numbers of cells interacting (rolling or arrested) with the BM microvasculature were determined by analysis of video recordings from fluorescence intravital microscopy experiments. n = 7–10 mice and human donors. *P < 0.02 compared with BM and mPB. **P = 0.03.

Neonate-derived CD34⁺ cell binding to soluble P-selectin, but not soluble E-selectin, is impaired. Since endothelial selectins are required for rolling of CD34⁺ cells (Figure 3), the lower rolling numbers of CB CD34⁺ cells suggested a reduced expression or function of selectin ligands on neonatal progenitor cells. To evaluate this possibility, we tested the ability of CB- and mPB-derived CD34⁺ cells to bind P-selectin or E-selectin in a fluid-phase assay. Mononuclear cells were isolated from fresh mPB and CB samples and stained with anti-CD34 and the human P-selectin–IgG chimera in the presence and absence of divalent cations. Eight individual donor pair samples were thus prepared in parallel and evaluated by FACS analysis. As shown in Figure 6a, the geometric mean of fluorescence of P-selectin–IgG binding among CD34⁺ cells was 59% lower in CB than in mPB (P = 0.008). Binding was specific since it was abrogated in divalent cation–chelated samples (Figure 6, a and b). Analysis of flow cytometry scattergrams revealed that while the vast majority (90% ± 1%) of mPB CD34⁺ cells bound P-selectin–IgG, approximately one-third of CB-derived CD34⁺ cells (34% ± 3%) did not bind soluble P-selectin (Figure 6b), suggesting a defect in expression or function of P-selectin ligand on a subpopulation of CB CD34⁺ cells. This finding was not exclusive to the human P-selectin chimeric construct since similar results were obtained with murine P-selectin–IgM (data not shown). To evaluate the function of E-selectin ligands, we analyzed the binding of a murine E-selectin–IgM chimera to purified adult and neonatal CD34⁺ cells. In contrast to P-selectin binding, only 60–70% of CD34⁺ cells bound E-selectin–IgM, but there was no significant difference in binding between CB and mPB (68% ± 2% and 62% ± 6% for CB and mPB CD34⁺ cells, respectively; n = 6, P = 0.33). These data indicate that function or expression of P-selectin ligands, but not E-selectin ligands, is significantly altered on neonate-derived CD34⁺ cells.

Figure 6

Binding of soluble P-selectin and PSGL-1 expression on CB and mPB CD34⁺ cells. (a) MNCs obtained from CB or mPB samples were stained for CD34 and P-selectin–IgG binding in the presence or absence of 5 mM EDTA. Data represent geometric mean of fluorescence ± SEM of P-selectin–IgG binding on gated CD34⁺ cells. n = 8 separate experiments from different donors. *P = 0.008. (b) Scattergram profiles of CB and mPB MNCs stained for CD34⁺ cells and P-selectin–IgG in the presence or absence of 5 mM EDTA. In the presence of divalent cations, virtually all mPB CD34⁺ cells bind P-selectin–IgG, whereas two distinct subpopulations are present among CB CD34⁺ cells: one that binds P-selectin–IgG normally (∼70%) and one that does not bind soluble P-selectin (∼30%). (c) Expression of PSGL-1 in CB and mPB CD34⁺ cells. (d) PSGL-1 is the sole P-selectin ligand on CB and mPB CD34⁺ cells. MNCs were preincubated with 2 μg of function-blocking (KPL1) or non-blocking (PSL-275) anti–PSGL-1 before staining to detect P-selectin–IgG chimera binding and CD34. The numbers indicate the percentage of CD34⁺ cells in each quadrant. The scattergrams represent one of at least five replicates.

Because PSGL-1 has been reported to be the main functional receptor for P-selectin in BM-derived CD34⁺ cells (24), we analyzed its expression. Using two different mAb’s against PSGL-1 (PSL-275 and KPL1), we found that most adult and neonatal CD34⁺ cells expressed this antigen (94% ± 2% in CB and 97% ± 1% in mPB, n = 5) (Figure 6c). To investigate whether PSGL-1 is the main functional P-selectin ligand on mPB and CB CD34⁺ cells, MNCs were stained with anti-CD34 Ab and incubated with either function-blocking (KPL1) or non-blocking (PSL-275) mAb against PSGL-1 prior to evaluating soluble P-selectin binding. Preincubation of cells with KPL1, but not with PSL-275, almost completely inhibited P-selectin–IgG binding in both CB- and mPB-derived CD34⁺ cells (Figure 6d). The normal expression of PSGL-1 glycoprotein on CB CD34⁺ cells, together with the evidence of abnormal function in a subset of CD34⁺ cells, suggests defective posttranslational modifications of the PSGL-1 protein in a subpopulation of CB CD34⁺ cells.

The PSGL-1/P-selectin pathway plays a major role in the initial interactions of CD34⁺ cells with BM microvessels. To assess whether PSGL-1 mediates CD34⁺ cell rolling in vivo, we treated fluorescently labeled mPB CD34⁺ cells with anti–PSGL-1 or control IgG and evaluated their behavior in the BM microcirculation after intracarotid injection. As shown in Figure 7a, the number of rolling CD34⁺ cells was significantly lower (∼56% reduction) in the group treated with anti–PSGL-1. To further test whether the reduced rolling capacity of CB-derived CD34⁺ cells might be due to defective PSGL-1, we tested the individual roles of PSGL-1, P-selectin, and E-selectin using E^–/– mice backcrossed in the NOD/SCID background and inhibitory Ab’s against P-selectin and PSGL-1. We found that the reduction observed earlier (Figure 3) in CB CD34⁺ cell rolling compared with cells derived from mPB was similar in wild-type NOD/SCID and E^–/– NOD/SCID mice (Figure 7b). However, inhibition of P-selectin function in both wild-type and E^–/– NOD/SCID mice greatly reduced (by 74% and 68%, respectively) the number of rolling CB CD34⁺ cells in the BM microvasculature (Figure 7b). In addition, blockage of PSGL-1 function on CB CD34⁺ cells inhibited their interactions in both wild-type and E-selectin–deficient mice (68% and 55% reduction, respectively). Inhibition using an anti–E-selectin Ab (clone 9A9) produced similar results to E-selectin deficiency (n = 2, data not shown). These data indicate that the PSGL-1/P-selectin pathway plays a major role in initial interactions and strongly suggest that the reduced ability of CB CD34⁺ cells to interact with the BM microvessels is due to altered P-selectin binding.

Figure 7

Role of P-selectin and PSGL-1 in CD34⁺ cell interactions with BM microvessels. Fluorescently labeled CD34⁺ cells from mPB (a) or CB (b) were preincubated with anti–PSGL-1 (black bars) or isotype-matched control Ab and injected into wild-type or E^–/– NOD/SCID mice that had been treated with anti–P-selectin (white bars) or isotype-matched control Ab (gray bars). The number of cells interacting with the BM microvasculature was determined by analysis of video recordings from fluorescence intravital microscopy experiments. *P < 0.02 compared with control Ab; **P = 0.06 compared with control Ab; ^#P < 0.02 compared with mPB control Ab group shown in a.

Neonatal CD34⁺ cells defective in P-selectin binding are enriched in primitive CD34⁺CD38^lo/– cells. CD34⁺ cells represent a heterogeneous population of progenitors with various degrees of hematopoietic maturation. The lack or dim expression of CD38 in human CD34⁺ cells is considered to be a surrogate marker of their primitive nature (25). To further characterize the population of CB-derived CD34⁺ cells that do not bind P-selectin–IgG, we analyzed P-selectin chimera binding and CD38 expression among CD34⁺ cells. As shown in Figure 8, CD34⁺ cells that did not bind P-selectin–IgG (Figure 8a, region in red) contained 23% ± 3% CD38^lo/– cells, compared with only 9% ± 1% CD34⁺ cells that bound soluble P-selectin (Figure 8b, region in green; n = 8 different donors, P = 0.0006). Thus, CD34⁺ progenitors that do not bind P-selectin are enriched in primitive HPCs.

Figure 8

Reduced binding of soluble P-selectin on primitive neonatal HPCs (CD34⁺CD38^lo/–). MNCs obtained from CB samples were stained for CD34 (using PE) and CD38 (using FITC) expression, and P-selectin–IgG chimera binding (using Cy5). (a) The gated regions represent CD34⁺ cells that bind (R2, green) or do not bind (R1, red) P-selectin–IgG. (b) CD38 expression on cells from R1 and R2 regions. The majority of cells that do not bind P-selectin–IgG are primitive CD34⁺CD38^lo/– cells. Shown is one representative of eight replicates.

Engraftment of NOD/SCID mice by CB CD34⁺ cells is not compromised. To determine whether the defect of CB CD34⁺ cell rolling might influence their engraftment into NOD/SCID mice, CD34⁺ cells were purified from CB and healthy mPB donors and transplanted into sublethally irradiated (350–375 cGy) NOD/SCID animals or NOD/SCID mice deficient in β2 microglobulin. One month after transplantation, mice were sacrificed to assess the engraftment levels of human-derived progenitors. As shown in Figure 9, there was a trend toward higher engraftment levels in the CB group compared with the mPB group, as measured by the frequency of human CD45⁺ cells and human progenitors in the BM compartment. These data are consistent with those of Wang et al., who found a higher level of engraftment of CB MNCs than of either BM or mPB MNCs (26), and suggest that CB CD34⁺ cells can compensate for their reduced initial interactions with the BM endothelium.

Figure 9

Engraftment of CB CD34⁺ cells in NOD/SCID mouse BM is not compromised. CD34⁺ cells isolated from CB or healthy mPB donors were transplanted into sublethally irradiated (350–375 cGy) NOD/SCID animals or NOD/SCID mice deficient in β2 microglobulin (NOD/SCID-β2m^null). Four weeks after transplantation, mice were sacrificed to assess the levels of human-derived CD45⁺ cells (a) or colony-forming progenitors (b) in their BM. There is a trend toward higher engraftment of CB-derived CD34⁺ cells (P = 0.07 for CD45⁺ and P = 0.1 for CFC content). Each circle in a represents an individual mouse.

The BM microvasculature differs from that of other tissues in several aspects. For example, the BM compartment does not contain capillaries but is endowed with a rich sinusoidal network. The fenestrated endothelium lining BM sinusoids promotes the release of billions of blood cells each day. Yet hematopoietic progenitors and mature blood cells (for example, B lymphocytes) likely reenter the BM under homeostatic conditions, and such trafficking is critical during hematopoietic development and BM transplantation. Because collecting venules mediate the recruitment of mature leukocytes during inflammation, we initially analyzed whether human HPCs interacted preferentially with sinusoids or collecting venules. We found that the two vessel types recruited circulating progenitors equally well. It is interesting that a high frequency of cell arrest was mapped within two vessel diameters of venular junctions, suggesting preferential endothelial areas for the recruitment of CD34⁺ cells. It is possible that these hotspots reflect differential endothelial adhesion molecule or chemokine expression in areas of nonlaminar blood flow. For example, alterations in shear stress can upregulate ICAM-1 and VCAM-1 expression (27). Although the numbers of rolling and arrested HPCs were similar between sinusoids and collecting venules, the majority of HPCs may in fact use sinusoids for BM homing, since the overall surface area of BM sinusoids is likely larger (28).

Strikingly, human CD34⁺ cell rolling was virtually absent in BM microvessels lacking P- and E-selectins. Endothelial selectins play a critical function in the recruitment of mature leukocytes (7, 8), but only a partial role (50% reduction) for P- and E-selectins was found in murine progenitor cell rolling (17) and homing (16) in the BM. Near-complete inhibition of HPC homing was achieved by blocking VCAM-1 function in selectin-deficient mice in both assay systems. In contrast to the studies with murine progenitors, we found that both rolling of human CD34⁺ cells on BM microvessels and homing to the NOD/SCID mouse BM were almost completely dependent on the expression of P- and E-selectin. Even if VCAM-1 inhibition significantly reduced the number of rolling cells, our results suggest that VCAM-1 cannot compensate when endothelial selectins are absent.

It is important to note that although the human α4 integrin (VLA-4) and human PSGL-1 have been shown to bind mouse VCAM-1 and mouse P-selectin, respectively, we cannot exclude the possibility that a difference in the affinities of human adhesion molecules for their respective murine counterreceptors may have influenced the results in this xenogeneic model. Our data are in agreement with those of Peled and colleagues, who found that a density of immobilized P- or E-selectin lower than the density of immobilized VCAM-1 could mediate human CD34⁺ cell tethering under physiological flow in vitro (29), suggesting that human CD34⁺ cells preferentially interact with endothelial selectins. However, blockade of the VLA-4 integrin (the ligand of VCAM-1) significantly inhibited the migration of human CD34⁺ cells into fetal sheep BM (30).

It is likely that VLA-4 integrin, a versatile receptor that can mediate both cell-cell (via VCAM-1) and cell-matrix (via fibronectin) interactions (31), operates at each step involved in HPC homing, including rolling, arrest, and migration. The modest effect of VCAM-1 inhibition in our studies argues that VLA-4 may act primarily at the latter steps of the homing cascade (arrest and migration), in which fibronectin plays important roles. Because selectin interactions with their ligands produce slow rolling and can induce signaling, it is also conceivable that the interactions of HPCs with endothelial selectins may enhance VLA-4’s affinity by increasing the contact of HPCs with immobilized chemokines on the endothelial surface, or perhaps by selectin ligand–mediated “inside-out” signaling (29, 32).

Blockade of adhesion molecules involved in the interaction of cells with extracellular matrix molecules (ECMs) can also alter progenitor behavior in vivo. For example, homing to the BM of β1 integrin–deficient murine progenitors is severely compromised (33), and blockade of either one of the two major β1 integrins (α4 and α5) on human CB progenitors was shown to inhibit migration through ECM and engraftment in NOD/SCID mice (34). These results, together with those of our colleagues, highlight the importance of two critical and largely nonoverlapping pathways for human HPC homing in which the initial interactions are mediated by endothelial selectins, while β1 integrins are largely involved in the subsequent steps.

Since human hematopoietic progenitors can be obtained from three different sources for transplantation, we sought to comparatively evaluate whether the hematopoietic milieu from which CD34⁺ cells were harvested might influence their behavior in vivo. Using our intravital microscopy system, we found that the number of rolling neonate-derived CD34⁺ cells in the BM of NOD/SCID mice was significantly reduced compared with those of adult CD34⁺ cells obtained either from steady-state BM or mPB. Because endothelial selectins mediate HPC rolling (Figure 3), we assessed the function of selectin ligands on CD34⁺ cells. Binding of soluble P-selectin was reduced in a subset of neonatal CD34⁺ cells compared with binding in HPCs derived from adult subjects.

A defect in the posttranslational processing of PSGL-1 protein was suggested by two observations: PSGL-1 is expressed on all CB CD34⁺ cells (Figure 6c), but a subset of these cells do not bind P-selectin; and PSGL-1 is the sole P-selectin ligand on CD34⁺ cells obtained from either steady-state BM (24), neonatal blood (ref. 29 and this study), or mPB (Figure 6d). Our results suggest that the reduced interactions of CB-derived CD34⁺ cells are due to surface expression of nonfunctional PSGL-1, since the PSGL-1/P-selectin pathway plays a major role in CD34⁺ cell rolling in vivo. Although the molecular defect responsible for defective PSGL-1 function in the subset of CB-derived CD34⁺ cells remains to be defined, specific posttranslational modifications in the PSGL-1 polypeptide have been well described for mature leukocytes, including sialylation, fucosylation, and tyrosine sulfation (13).

Interestingly, Mariscalco and colleagues found that neonatal neutrophil rolling on monolayers of Chinese hamster ovary (CHO) cells transfected with P-selectin was reduced compared with that of adult neutrophils (35). Further studies are needed to characterize PSGL-1’s defect(s) in both mature and immature neonatal cells. It is also noteworthy that P-selectin was shown to be absent in endothelial cells of newborn rat mesentery and decreased in newborn human mesentery (36). It is thus tempting to speculate that the P-selectin/PSGL-1 adhesion pathway is not fully functional during fetal life and may become operative only in the postnatal period.

Since the fraction of neonatal CD34⁺ cells that does not bind P-selectin is enriched in primitive CD34⁺CD38^lo/– cells, our findings may have important implications for clinical BM transplantation. Neonatal blood has been used as a source of HPCs in more than 1,500 cases of transplantation thus far (37). Although CB-derived progenitors offer several advantages over those harvested from adult sources, including wide availability, ease of harvest, and lower incidence of graft-versus-host disease, the limited number of cells recovered per unit has restricted transplantation mostly to individuals with low body weights (i.e., children) (22). Significant delays in platelet and myeloid engraftment have been reported in CB cell transplantation (22, 38). It is unclear, however, whether this is due to low numbers of transplanted HPCs or a defect in homing. Our results indicate that although homing of CB CD34⁺ cells is impaired, their engraftment is at least as high as that of adult CD34⁺ cells. Thus, these data raise the possibility that the defect in the initial interactions of CB CD34⁺ cells in the BM microcirculation may be compensated for by other “youthful” characteristics of neonatal cells. For example, CB progenitors have a greater proliferation and expansion capacity (26, 39) and a greater proportion of immature CD34⁺CD38^– cells than do their adult counterparts (40). In support of the present studies, while our manuscript was under review, Szilvassy et al. reported on a discrepancy between homing and engraftment of fetal liver progenitor cells (reduced homing and higher engraftment of fetal HPCs compared with BM- and mPB-derived HPCs) (41). Greater insights into the mechanisms and regulation of PSGL-1’s posttranslational modifications in neonatal cells may thus allow us to optimize the initial interactions of CB-derived stem cells with endothelial cells, and to enhance their lodgement in the BM. Our results indeed underscore the enormous potential of CB cells as a source of hematopoietic stem cells.

View Supplemental Data

We thank Steve Fruchtman, Luis Isola, and the Stem Cell Lab staff for help in procuring human adult hematopoietic samples, and the Labor and Delivery team at Mount Sinai Medical Center for assistance in obtaining neonatal blood. We thank Robert S. Schaub, John Lowe, Barry Wolitzky, and Masaru Okabe for providing reagents. We also thank Seunghee Kim-Schulze for help in the initial experiments and Yoshio Katayama for helpful discussions. These studies were funded in part by a scholarship from the American Society of Hematology and the National Institutes of Health (R01 DK-56638) to P.S. Frenette.

Conflict of interest: No conflict of interest has been declared.

Nonstandard abbreviations used: hematopoietic progenitor cell (HPC); bone marrow (BM); P-selectin glycoprotein ligand-1 (PSGL-1); mobilized peripheral blood (mPB); cord blood (CB); phycoerythrin (PE); low-density mononuclear cell (MNC); red blood cell (RBC); carboxyfluorescein succinimidyl ester (CFSE); P- and E-selectin null (P/E^–/–); enhanced green fluorescent protein (EGFP); transgenic (Tg); intravital microscopy (IVM).

1. Jacobson, LO, Marks, EK, Robson, MJ, Gaston, E, Zirkle, RE. The effect of spleen protection on mortality following X-radiation. J Lab Clin Med 1949. 34:1538-1543.
2. Lorenz, E, Uphoff, D, Reid, TR, Shelton, E. Modification of irradiation injury in mice and guinea pigs by bone marrow injections. J Natl Cancer Inst 1951. 12:197-201.
   View this article via: PubMed Google Scholar
3. Barnes, DWH, Corp, MJ, Loutit, JF, Neal, FE. Treatment of murine leukaemia with x-rays and homologous bone marrow. Preliminary communication. Br Med J 1956. 2:626-627.
4. Thomas, ED. History, current results, and research in marrow transplantation. Perspect Biol Med 1995. 38:230-237.
   View this article via: PubMed Google Scholar
5. Frenette, PS, Wagner, DD. Adhesion molecules—Part II: Blood vessels and blood cells. N Engl J Med 1996. 335:43-45.
   View this article via: PubMed CrossRef Google Scholar
6. Kansas, GS. Selectins and their ligands: current concepts and controversies. Blood 1996. 88:3259-3287.
   View this article via: PubMed Google Scholar
7. Frenette, PS, Mayadas, TN, Rayburn, H, Hynes, RO, Wagner, DD. Susceptibility to infection and altered hematopoiesis in mice deficient in both P- and E-selectins. Cell 1996. 84:563-574.
   View this article via: PubMed CrossRef Google Scholar
8. Bullard, DC, et al. Infectious susceptibility and severe deficiency of leukocyte rolling and recruitment in E-selectin and P-selectin double mutant mice. J Exp Med 1996. 183:2329-2336.
   View this article via: PubMed CrossRef Google Scholar
9. Moore, KL, et al. Identification of a specific glycoprotein ligand for P-selectin (CD62) on myeloid cells. J Cell Biol 1992. 118:445-456.
   View this article via: PubMed CrossRef Google Scholar
10. Frenette, PS, et al. P-Selectin glycoprotein ligand 1 (PSGL-1) is expressed on platelets and can mediate platelet-endothelial interactions in vivo. J Exp Med 2000. 191:1413-1422.
    View this article via: PubMed CrossRef Google Scholar
11. Zannettino, AC, et al. Primitive human hematopoietic progenitors adhere to P-selectin (CD62P). Blood 1995. 85:3466-3477.
    View this article via: PubMed Google Scholar
12. Tracey, JB, Rinder, HM. Characterization of the P-selectin ligand on human hematopoietic progenitors. Exp Hematol 1996. 24:1494-1500.
    View this article via: PubMed Google Scholar
13. Moore, KL. Structure and function of P-selectin glycoprotein ligand-1. Leuk Lymphoma 1998. 29:1-15.
    View this article via: PubMed Google Scholar
14. Springer, TA. Traffic signals on endothelium for lymphocyte recirculation and leukocyte emigration. Annu Rev Physiol 1995. 57:827-872.
    View this article via: PubMed CrossRef Google Scholar
15. Papayannopoulou, T, Craddock, C, Nakamoto, B, Priestley, GV, Wolf, NS. The VLA4/VCAM-1 adhesion pathway defines contrasting mechanisms of lodgement of transplanted murine hemopoietic progenitors between bone marrow and spleen. Proc Natl Acad Sci USA 1995. 92:9647-9651.
    View this article via: PubMed CrossRef Google Scholar
16. Frenette, PS, Subbarao, S, Mazo, IB, von Andrian, UH, Wagner, DD. Endothelial selectins and vascular cell adhesion molecule-1 promote hematopoietic progenitor homing to bone marrow. Proc Natl Acad Sci USA 1998. 95:14423-14428.
    View this article via: PubMed CrossRef Google Scholar
17. Mazo, IB, et al. Hematopoietic progenitor cell rolling in bone marrow microvessels: parallel contributions by endothelial selectins and VCAM-1. J Exp Med 1998. 188:465-474.
    View this article via: PubMed CrossRef Google Scholar
18. Vermeulen, M, et al. Role of adhesion molecules in the homing and mobilization of murine hematopoietic stem and progenitor cells. Blood 1998. 92:894-900.
    View this article via: PubMed Google Scholar
19. Shultz, LD, et al. Multiple defects in innate and adaptative immunologic function in NOD/LtSz-scid mice. J Immunol 1995. 154:180-191.
    View this article via: PubMed Google Scholar
20. Dick, JE, Bhatia, M, Gan, O, Kapp, U, Wang, JC. Assay of human stem cells by repopulation of NOD/SCID mice. Stem Cells 1997. 15(Suppl. 1):199-203; discussion 204–207.
    View this article via: PubMed Google Scholar
21. Rocha, V, et al. Comparison of outcomes of unrelated bone marrow and umbilical cord blood transplants in children with acute leukemia. Blood 2001. 97:2962-2971.
    View this article via: PubMed CrossRef Google Scholar
22. Rubinstein, P, et al. Outcomes among 562 recipients of placental-blood transplants from unrelated donors. N Engl J Med 1998. 339:1565-1577.
    View this article via: PubMed CrossRef Google Scholar
23. Maly, P, et al. The alpha(1, 3)fucosyltransferase Fuc-TVII controls leukocyte trafficking through an essential role in L-, E-, and P-selectin ligand biosynthesis. Cell 1996. 86:643-653.
    View this article via: PubMed CrossRef Google Scholar
24. Levesque, JP, et al. PSGL-1-mediated adhesion of human hematopoietic progenitors to P-selectin results in suppression of hematopoiesis. Immunity 1999. 11:369-378.
    View this article via: PubMed CrossRef Google Scholar
25. Terstappen, LW, Huang, S, Safford, M, Lansdorp, PM, Loken, MR. Sequential generations of hematopoietic colonies derived from single nonlineage-committed CD34⁺CD38- progenitor cells. Blood 1991. 77:1218-1227.
    View this article via: PubMed Google Scholar
26. Wang, JC, Doedens, M, Dick, JE. Primitive human hematopoietic cells are enriched in cord blood compared with adult bone marrow or mobilized peripheral blood as measured by the quantitative in vivo SCID-repopulating cell assay. Blood 1997. 89:3919-3924.
    View this article via: PubMed Google Scholar
27. Walpola, PL, Gotlieb, AI, Cybulsky, MI, Langille, BL. Expression of ICAM-1 and VCAM-1 and monocyte adherence in arteries exposed to altered shear stress. Arterioscler Thromb Vasc Biol 1995. 15:2-10.
    View this article via: PubMed Google Scholar
28. Lichtman, MA. The ultrastructure of the hemopoietic environment of the marrow: a review. Exp Hematol 1981. 9:391-410.
    View this article via: PubMed Google Scholar
29. Peled, A, et al. The chemokine SDF-1 stimulates integrin-mediated arrest of CD34⁺ cells on vascular endothelium under shear flow. J Clin Invest 1999. 104:1199-1211.
    View this article via: JCI PubMed CrossRef Google Scholar
30. Zanjani, ED, Flake, AW, Almeida-Porada, G, Tran, N, Papayannopoulou, T. Homing of human cells in the fetal sheep model: modulation by antibodies activating or inhibiting very late activation antigen-4-dependent function. Blood 1999. 94:2515-2522.
    View this article via: PubMed Google Scholar
31. Masumoto, A, Hemler, ME. Multiple activation states of VLA-4. Mechanistic differences between adhesion to CS1/fibronectin and to vascular cell adhesion molecule-1. J Biol Chem 1993. 268:228-234.
    View this article via: PubMed Google Scholar
32. Lorenzon, P, et al. Endothelial cell E- and P-selectin and vascular cell adhesion molecule-1 function as signaling receptors. J Cell Biol 1998. 142:1381-1391.
    View this article via: PubMed CrossRef Google Scholar
33. Potocnik, AJ, Brakebusch, C, Fassler, R. Fetal and adult hematopoietic stem cells require beta1 integrin function for colonizing fetal liver, spleen, and bone marrow. Immunity 2000. 12:653-663.
    View this article via: PubMed CrossRef Google Scholar
34. Peled, A, et al. The chemokine SDF-1 activates the integrins LFA-1, VLA-4, and VLA-5 on immature human CD34(⁺) cells: role in transendothelial/stromal migration and engraftment of NOD/SCID mice. Blood 2000. 95:3289-3296.
    View this article via: PubMed Google Scholar
35. Mariscalco, MM, Tcharmtchi, MH, Smith, CW. P-Selectin support of neonatal neutrophil adherence under flow: contribution of L-selectin, LFA-1, and ligand(s) for P-selectin. Blood 1998. 91:4776-4785.
    View this article via: PubMed Google Scholar
36. Lorant, DE, Li, W, Tabatabaei, N, Garver, MK, Albertine, KH. P-selectin expression by endothelial cells is decreased in neonatal rats and human premature infants. Blood 1999. 94:600-609.
    View this article via: PubMed Google Scholar
37. Gluckman, E. Hematopoietic stem-cell transplants using umbilical-cord blood. N Engl J Med 2001. 344:1860-1861.
    View this article via: PubMed CrossRef Google Scholar
38. Barker, JN, et al. Survival after transplantation of unrelated donor umbilical cord blood is comparable to that of human leukocyte antigen-matched unrelated donor bone marrow: results of a matched-pair analysis. Blood 2001. 97:2957-2961.
    View this article via: PubMed CrossRef Google Scholar
39. Mayani, H, Lansdorp, PM. Biology of human umbilical cord blood-derived hematopoietic stem/progenitor cells. Stem Cells 1998. 16:153-165.
    View this article via: PubMed Google Scholar
40. Cardoso, AA, et al. Release from quiescence of CD34⁺ CD38- human umbilical cord blood cells reveals their potentiality to engraft adults. Proc Natl Acad Sci USA 1993. 90:8707-8711.
    View this article via: PubMed CrossRef Google Scholar
41. Szilvassy, SJ, Meyerrose, TE, Ragland, PL, Grimes, B. Differential homing and engraftment properties of hematopoietic progenitor cells from murine bone marrow, mobilized peripheral blood, and fetal liver. Blood 2001. 98:2108-2115.
    View this article via: PubMed CrossRef Google Scholar