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

During development, hematopoietic stem cells migrate from the yolk sac and the aorta gonad mesonephros (AGM) region into the fetal liver and then into the fetal bone marrow. Stem cells continuously produce all mature blood cells and are defined in functional repopulation assays based on their ability to home to the bone marrow microenvironment and to durably repopulate transplanted recipients with myeloid and lymphoid cells.1-4 We and others developed functional in vivo assays for primitive human SCID (severe combined immunodeficiency) repopulating cells (SRCs), which provide a means to measure the multilineage engraftment properties of human stem cells.5-7 Determining the role of chemokines and adhesion molecules in human stem cell migration and engraftment will help to identify the mechanisms that govern stem cell development and advance clinical stem cell transplantation.

Recently we discovered that human stem cell engraftment of nonobese (NOD)/SCID mice is dependent on the expression of the chemokine SDF-1 and its receptor CXCR4.8 However, the possible role of SDF-1 in mediating activation of different adhesion molecules, which are expressed on repopulating human stem cells during this process, is currently unknown. Stem cell homing and engraftment is presumably a multistep process that shares some common features with the migration of leukocytes to inflammatory sites and homing of lymphocytes into lymph nodes.9-11 First, the transplanted human cells, which migrate through the blood circulation, must interact with the bone marrow vascular endothelial cells. This results in rolling on endothelial (E) and platelet (P) selectins, which is followed by firm shear resistant adhesion to the vessel wall.12-14 These interactions are mediated through the coordinated action of adhesive molecules and activation processes triggered specifically by chemokines such as SDF-1 and vascular ligands, eg, intercellular adhesion molecule–1 (ICAM-1) and vascular cellular adhesion molecule–1 (VCAM-1).9-14 Following arrest on the bone marrow microvasculature, stem cells extravasate through the endothelium and into the hematopoietic compartment. The β2 integrin, lymphocyte function associated antigen–1 (LFA-1), is involved in the spontaneous transendothelial migration of immature human CD34⁺ cells in vitro.15,16 Furthermore, SDF-1 activation of LFA-1 that is present on human CXCR4⁺ T lymphocytes led to firm shear resistant adhesion to endothelial ICAM-1.17 The major β1 integrins, very late activation antigen–4 (VLA-4), and VLA-5 have been implicated in the adhesive interactions of human, primate, and mouse stem cells with the bone marrow extracellular matrix (ECM) and stromal cells.18-22 

In the present study, we investigated both specific and overlapping adhesive functions of LFA-1, VLA-4, and VLA-5 in interactions between immature human CD34⁺/CXCR4⁺ cells and the vascular endothelial, ECM, and stromal elements of the bone marrow. The role of the major stem cell chemokine SDF-1 in either triggering adhesiveness of these integrins or in directing human SRCs/stem cells to migrate across the different bone marrow compartments, which express ligands to these integrins, was investigated. We tested the ability of CD34⁺/CXCR4⁺ cells, which migrate in vitro across human endothelial and stromal cells toward SDF-1, to engraft and repopulate NOD/SCID mice in vivo through the combined activities of these integrins.

The following reagents were used in this study: human ICAM-1 (gift of Dr L. Klickstein, Brigham and Women's Hospital, Boston, MA); recombinant soluble human VCAM-123 and monoclonal antibody (mAb) to VCAM-1 (gift of Dr R. Lobb, Biogen, Cambridge, MA); human SDF-1 and macrophage inflammatory protein–1α (MIP-1α) (R&D Systems, Minneapolis, MN); bovine serum albumin (BSA, fraction V), calcium- and magnesium-free Hank's balanced salt solution (HBSS), ethylenediaminetetraacetic acid (EDTA), and 4-(2-Hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) (Sigma, St Louis, MO); human serum albumin (HSA) (Calbiochem, La Jolla, CA); human fibronectin (FN) (Chemicon, Temecula, CA); and collagen type I (laminin) (Cellagen; ICN Pharmaceuticals, CA).

Human cord blood cells were obtained from full-term deliveries after informed consent and were used in accordance with approved procedures by the human experimentation and ethics committees of the Weizmann Institute, Rehovot, Israel. Cord blood samples were diluted in phosphate-buffered saline (PBS) and supplemented with 1% fetal calf serum (FCS) (Bet Haemek, Israel). Low-density mononuclear cells (MNCs) were collected after standard separation on Ficoll-Paque (Pharmacia Biotech, Uppsala, Sweden) and washed with PBS. Human umbilical vascular endothelial cells (HUVECs) were isolated from umbilical cord veins according to the method of Jaffe et al.24 Human adherent stromal cells were prepared and grown as previously described25 from leftover bone marrow cells for allogeneic transplantation. Mouse stromal cell lines were grown as previously described.26 

Enrichment of human CD34⁺ cells was performed with a magnetic bead separation kit (mini-MACS, Miltneyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer's instructions. The purity of the enriched CD34⁺ cells was 60%-85% when cells were passed over 1 column and more than 98% when the cells were passed over 2 columns.

Flow cytometry analysis was done as previously described.8 Bone marrow cells from mice that received transplantations were flushed and resuspended in staining buffer (PBS, 0.1% BSA, and 0.02% sodium azide) after red blood cells were lysed with ammonium chloride, and 10⁵ cells were incubated with 10 μ/mL purified antimouse CD16/CD32 Fc receptor (FcR) (PharMingen, San Diego, CA) and 1% human plasma for 20 minutes at 4°C. Cells were stained with human-specific direct-labeled antibodies and incubated for 30 minutes on ice. Isotype control antibodies (Becton Dickinson, San Jose, CA) were used to exclude false-positive cells. Murine bone marrow cells from nontransplanted mice were used a negative control, and human cells were used as a positive control. Dead cells were gated out by staining with propedium iodide (Sigma). Immature human cells from engrafted mice were identified by double-staining with anti-CD34 fluorescein isothiocyanate (FITC) (Becton Dickinson) and anti-CD38 phycoerythrin (PE) (Coulter Electronics, Miami, FL) and the presence of pre–B-lymphoid cells by anti-CD45 FITC (IQP, The Netherlands) and anti-CD19 PE (Coulter Electronics).

The levels of CXCR4 expression on human CD34⁺ cells were detected with anti-CXCR4 PE (PharMingen) together with anti-CD34 FITC. The levels of immature cells in the transwell migration assay were analyzed by staining with anti-CD34 FITC and anti-CD38 PE. After staining, cells were washed twice in the same buffer and analyzed by fluorescence-activated cell sorter (FACS) (FACSort and CellQuest software, Becton Dickinson). When antibodies to VCAM-1 and ICAM-1 (Novocastra Laboratories, Newcastle, England) were used, they were detected by using secondary FITC-conjugated F(ab′)2 fragment goat antimouse immunoglobulin G (IgG) (Jackson, West Grove, PA).

Semisolid progenitor cultures specific for human colonies were performed as previously described.5 In brief, 200 × 10³ bone marrow cells per mL from transplanted mice were plated in 0.9% methylcellulose (Sigma), 15% FCS, 15% human plasma, 50 ng/mL stem cell factor, 5 ng/mL interleukin-3 (IL-3), 5 ng/mL granulocyte-macrophage–colony stimulating factor (GM-CSF) (R&D Systems), and 2 U/mL erythropoietin (EPO) (Ortho Biotech, Don Mills, Ontario, Canada). The cultures were incubated at 37°C in a humidified atmosphere containing 5% carbon dioxide (CO₂) and were scored 14 days later.

NOD/SCID mice (NOD/LtSzPrKdc^scid/PrKd^scid) were bred and maintained under defined flora conditions in individually ventilated (high-efficiency particle-arresting filtered air) sterile microisolator cages (Techniplast, Italy) at the Weizmann Institute. All the experiments were approved by the animal care committee of the Weizmann Institute. The 8-week-old mice were irradiated with a sublethal dose of 375 cGy (67 cG/min) from a cobalt source prior to transplantation. Human CD34⁺ enriched cells (2 × 10⁵ cells per mouse; purity, 60%-98%), were injected into the tail vain of irradiated mice in 0.5 mL Roswell Park Memorial Institute medium (RPMI 1640) with 10% FCS. For in vivo blocking experiments, the cells were first preincubated for 30 minutes on ice with 5 μg/mL antihuman CD34 antibodies as a negative control (Becton Dickenson) or the following mAbs (Serotec, Oxford, England): anti-β1 (MCA1188), VLA-4 (MCA697), VLA-5 (MCA1187), or anti–LFA-1 (MCA1149). The cells were washed and injected into mice, which were killed after 30-45 days. The cells from the femur, tibia, and pelvis bones were flushed with a syringe.

Chemotaxis experiments were assayed (Costar, Cambridge, MA) by using transwells (diameter, 6.5 mm; pore, 5 μm) as previously described.8 We added 100 μL chemotaxis buffer (RPMI 1640 with 1% FCS) containing 2 × 10⁵ CD34⁺cells to the upper chamber, and 0.6 mL chemotaxis buffer with or without 125 ng/mL SDF-1 or MIP-1α to the bottom chamber. After 4 hours, migrating (bottom chamber) and nonmigrating (upper chamber) cells were counted for 30 seconds using a FACSort. Endothelial or bone marrow stromal cells (2 × 10⁴ cells per well) were seeded into the upper chamber, precoated with 25 μg/mL FN for 1 hour at room temperature, and grown for 48 hours. In some experiments, 10⁶ CD34⁺ cells per mL were preincubated (30 minutes at 4°C) in 100 μL migration buffer containing either 5 μg control isotype-matching mAb (Becton Dickinson) or a murine mAb specific either to α4 or to LFA-1 (Serotec). The cells were washed before transendothelial or transstromal migration. Endothelial and stromal cells were activated by overnight incubation with 2 ng/mL TNF-α (R&D Systems), which was washed before the migration.

PBS (500 μL) containing 20 μg/mL human FN or 2.5% BSA as control was placed in 24-well plates (Falcon, Becton Dickinson) and incubated overnight at 4°C. Wells were washed with PBS, blocked with 1000 μL 2.5% BSA in PBS, and incubated for 1 hour at room temperature. Plates were then washed 3 times with adhesion medium (RPMI 1640 supplemented with 0.2% BSA). CD34⁺ enriched cells (6 × 10⁴ cells) in 200 μL adhesion medium were added to the precoated wells. The cells were allowed to adhere for 30 minutes at 37°C in a humidified atmosphere containing 5% CO₂ and then washed 4 times with prewarmed adhesion medium to remove nonadherent cells. The adherent cells were collected with medium containing 0.01% EDTA and by gentle shaking with vortex. The cells were counted and assessed for colony forming cells (CFCs).

Laminar flow assays were performed as previously described.27 ICAM-1 and soluble VCAM-1 were coated at 10 μg/mL in the presence of 2 μg/mL HSA carrier on polystyrene plates (Becton Dickinson). The plates were washed 3 times with PBS and blocked with 20 μg/mL HSA in PBS for 2 hours at room temperature. Alternatively, washed plates were coated with 10 μg/mL SDF-1 in PBS for 30 minutes at room temperature before being blocked with HSA. The plates were assembled as the lower wall of a parallel wall flow chamber and mounted on the stage of an inverted microscope. Cord blood 2 × 10⁶ CD34⁺ cells per mL (purity, at least 98%) were suspended in binding buffer, perfused into the chamber, and allowed to settle on the substrate-coated chamber wall for 1 minute at 37°C. Flow was initiated and increased in 2-fold to 2.5-fold increments every 5 seconds, thereby generating controlled shear stresses on the wall. Cells were visualized with an inverted phase-contrast Diaphot Microscope (magnification objective × 20) (Nikon, Tokyo, Japan) and photographed with a long integration LIS-700 CCD video camera (Applitech, Holon, Israel) connected to an AG-6730 S-VHS video recorder (Panasonic, Osaka, Japan). The number of adherent cells resisting detachment by the elevated shear forces was determined after each interval by analysis of videotaped cell images and was expressed as the percent of originally settled cells. To test the effects of phorbol 12-myristate 13-acetate (PMA), cells were suspended in binding medium containing 100 ng/mL PMA (Sigma) seconds before being perfused into the chamber. All adhesion experiments were performed at least 3 times on multiple test fields.

Migration assays in 3-D ECM-like gels were performed as previously described.28 Purified (at least 98%) cord blood CD34⁺ cells were suspended in a 10-μL drop consisting of 1.8 μg/mL collagen type I (6 μg/mL laminin) and 2.5 μg/mL FN in RPMI 1640. A second drop without cells was placed 1.5 mm from the first drop. An SDF-1 depot was created in a third drop, which was supplemented with 500 ng/mL SDF-1 and placed 1.5 mm downstream from the second drop and 3-5 mm from the first drop. Once the drops began to polymerize, they were gently connected with a fine needle to form a continuous 3-D gel, and cell migration within this gel was tracked by time-lapse video microscopy. Cell images were visualized and videotaped on an AG-6730 S-VHS time-lapse video recorder at 25 frames per minute. The proportions of polarized, nonmotile, randomly migrating, and directionally migrating cells within the entire population of cells in the field were determined within 60-90 minutes of tracking. The role of β1 integrins was examined by preincubating 10⁶CD34⁺ cells per mL for 20 minutes at 4°C in a 200 μL RPMI 1640 mixture containing 1% BSA and either 5 μg control isotype-matching mAbs or specific murine mAbs to either the α4, α5, or β1 integrins. Subsequently, the cells were washed and added to the 3-D gels.

Immature human cord blood CD34⁺/CXCR4⁺ cells express the major integrins LFA-1, VLA-4, and VLA-5 on their surface (Figure 1A and 1B). The adhesiveness of integrins to endothelial ligands was measured in integrin-dependent adhesion assays using a parallel-plate flow chamber, which simulates blood flow. Purified cord blood CD34⁺ cells treated briefly with PMA or left untreated were allowed to bind for 1 minute to immobilized ICAM-1 and VCAM-1 in stasis. Alternatively, the cells were briefly allowed to interact with ICAM-1 and VCAM-1 coimmobilized with SDF-1. The cells were then subjected to incremented shear flow that generated increasingly detached forces on the adherent cells. SDF-1 rapidly activated the firm shear-resistant adhesion of human CD34⁺/CXCR4⁺ cells to immobilized ICAM-1 and VCAM-1 (Figure 1C and 1D). Chemokine-mediated activation was almost as powerful as activation with the nonphysiological agonist PMA, and the activation was specific because it was totally inhibited in the presence of the integrin inhibitor EDTA (Figure 1C and 1D).

Following their firm adhesion to the vessel wall, stem cells extravasate through the endothelial lining of the vasculature. We further studied the role that SDF-1 plays in regulating the extravasation of immature human CD34⁺/CXCR4⁺cells through endothelial cells. This was accomplished through transendothelial migration assays using 2 types of TNF-α–activated endothelial cells: HUVECs or the murine bone marrow–derived endothelial cell line, MBA-2.1.24,26 Upon activation with TNF-α in vivo, both cell types significantly up-regulated ICAM-1 and VCAM-1 surface expression, mimicking the constitutive expression of these ligands by bone marrow endothelium (Figure2A). CXCR4-dependent migration of cord blood CD34⁺ cells occurred across activated HUVECs toward a chemotactic gradient of SDF-1 in a transwell assay. This migration was partially blocked by antibodies to LFA-1, but it was not inhibited by antibodies to VLA-4 or VLA-5, and no additive effect was observed when all the antibodies were used together (Figure 2B). Surprisingly, antibodies to VLA-4 partially inhibited transendothelial migration by CD34⁺/CXCR4⁺ cells only when bone marrow–derived MBA-2.1 endothelial cells were used. We found no effect when antibodies to VLA-5 were used; however, an additive effect was observed when antibodies to LFA-1, VLA-4, and VLA-5 were mixed together (Figure 2B). These results may suggest that stem cell migration through endothelium depends on the origin of the endothelial cells and the levels of VCAM-1 expression (Figure 2B).

Next, the cell surface phenotype of migrating and nonmigrating CD34⁺ cells (designated M and NM, respectively, in Figure2) was characterized. SDF-1 preferentially induced transmigration of primitive CD34⁺/CD38^−/low/CXCR4⁺cells across the endothelium (Figure 2C). Notably, cells migrating through the endothelial cells gave significantly higher levels of engraftment compared with nonmigrating cells collected from the upper chamber (Figure 2D). Bone marrow cells from mice engrafted with migrating cells gave rise to high levels of both myeloid CFCs (Figure2E I) as well as lymphoid CD19⁺ pre–B cells (Figure 2E II). Mice transplanted with nonmigrating cells did not contain human progenitor CFCs (Figure 2E I). We therefore conclude that SDF-1 can selectively induce the transendothelial migration of the human SRC/stem cell subset within the CD34⁺ progenitor cell population.

Following extravasation through the vascular endothelium, stem cells encounter the bone marrow ECM barriers. In contrast to SDF-1, other chemokines, such as MIP-1α,29-31 are poor chemoattractants of human CD34⁺ progenitor cells (Figure3A). However, both SDF-1 and MIP-1α can activate the binding of CD34⁺ cells to the ECM protein FN (Figure 3B). Furthermore, both chemokines preferentially induce the binding of immature multilineage CFCs to FN (Figure 3C).

FN is part of the basal lamina underlining the endothelial cells and is also present in the bone marrow extravascular space. To unravel the mechanism by which SDF-1 regulates the interactions between cord blood CD34⁺/CXCR4⁺ cells and FN, we studied the contribution of the major integrins VLA-4 and VLA-5 to binding of these cells to FN. Adhesion of CD34⁺ cells to FN was mainly dependent on VLA-5. Activation of the cells with SDF-1 or MIP-1α caused enhanced dependency on VLA-5, which suggests that in this case, VLA-5 but not VLA-4 is the major FN receptor (Figure 3D).

During bone marrow transplantation, stem cells need to pass through the basal lamina, which is composed of the ECM proteins laminin, collagen, and FN. SDF-1–mediated interactions between migrating human cord blood CD34⁺/CXCR4⁺ cells and the ECM were studied in vitro by monitoring the migratory properties of these cells through a 3-D ECM-like gel that was reconstituted with a meshwork of laminin, collagen, and FN. This novel system allows for close examination of the random and directional migration of cells toward a newly generated chemoattractant source in real time.28 Most CD34⁺ cells embedded in this gel remained spherical and failed to polarize or migrate in the absence of SDF-1 (Figure4A). However, upon introduction of an SDF-1 gradient, 40% of the cells polarized in a time-dependent manner (Figure 4A). As many as 30% of the cells migrated toward a gradient of SDF-1 (Figure 4B). The percentage of polarized and migrating CD34⁺ cells in a 3-D ECM-like gel correlated with the levels of CXCR4 expression on cord blood CD34⁺ cells (48% positive cells) (Figure 1A) and with the frequency of 20%-30% of cells migrating toward a gradient of SDF-1 (Figure 2B). In contrast to the dominant role of VLA-5 in facilitating the static adhesion to FN, SDF-1–induced polarization and directional migration of CD34⁺/CXCR4⁺ cells in 3-D ECM-like gels was dependent on both VLA-4 and VLA-5 integrins (Figure 4A and 4B). Thus, weak VLA-4 adhesion/deadhesion interactions with FN during migration across 3-D ECM-like gels are as important as interactions between VLA-5 and FN. Antibodies to LFA-1 had no effect on SDF-1– induced polarization and migration of the cells (Figure 4 A and 4B).

We demonstrated above that SDF-1 activates the major integrins LFA-1, VLA-4, and VLA-5 in a specific manner during SRC/stem cell migration across endothelial cells and ECM. To determine the in vivo role of the major integrins LFA-1, VLA-4, and VLA-5 during the engraftment process of human SRCs/stem cells, enriched CD34⁺ cord blood cells were pretreated with antibodies against each of the above integrins separately or as a control with anti-CD34 and anti–VLA-6 antibodies before transplantation. Neutralizing antibodies to β1, VLA-4, or VLA-5 blocked murine bone marrow engraftment by human CD34⁺cells, whereas control anti-CD34 or anti–VLA-6 antibodies did not. This demonstrated a major role for these integrins in the engraftment process (Figure 4C and 4D). Neutralizing antibodies to LFA-1 significantly decreased the levels of engraftment (Figure 4C). Thus anti–LFA-1 antibodies caused partial inhibition of engraftment, and anti–VLA-4 or anti–VLA-5 antibodies mediated near-complete inhibition of human SRC/stem cell engraftment. Taken together, this data strongly suggest that LFA-1, VLA-4, and VLA-5 function in a sequential or consecutive manner, which is mediated by SDF-1, to support the recruitment of stem cells into the bone marrow.

Once stem cells have crossed the ECM barrier, they then interact with the bone marrow mesenchymal microenvironment. The ability of SDF-1 to induce this crucial step in stem cell trafficking within the bone marrow stromal microenvironment was tested by examining the migration of immature CB CD34⁺/CXCR4⁺ cells through bone marrow–derived murine stromal cells (adipocyte 14F1.1 and osteoblast MBA-15.1 cell lines) or primary human stromal cells.25,26In contrast to endothelial cells, stromal cells express only low levels of ICAM-1 or VCAM-1, and the expression of these adhesion molecules was not affected by TNF-α (data not shown). Similar levels of CD34⁺/CXCR4⁺ cells migrated through endothelial and stromal cells in response to SDF-1 (Figure 2B and Figure5A). However, the migration of CD34⁺ cells through the stromal layer is dependent on both integrins VLA-4 and VLA-5 but not on LFA-1 (Figure 5A). As observed in transendothelial migration induced by SDF-1, this chemokine preferentially induces transmigration across stromal cells of primitive CD34⁺/CD38^−/low/CXCR4⁺ cells (Figure 2B and data not shown). We further tested the ability of CD34⁺/CXCR4⁺ cells that migrated through the different stromal cells to engraft irradiated NOD/SCID mice. Cells migrating through the stroma layers gave significantly higher levels of engraftment compared with nonmigrating cells collected from the upper chamber (Figure 5B). Moreover, multilineage differentiation of engrafted SRCs/stem cells, which included lymphoid CD19⁺pre–B cells (Figure 5C I), primitive CD34⁺/CD38^−/low cells (Figure 5C II), and myeloid CFCs (data not shown), was achieved only by migrating cells. We therefore conclude that SDF-1 preferentially induces the transstromal migration of human SRCs/stem cells.

In the present study we demonstrate that the major integrins VLA-4, VLA-5, and LFA-1 are present in immature human CD34⁺ cells and that they are functionally activated by SDF-1. In addition, neutralizing antibodies to LFA-1 significantly reduced the levels of engraftment, and neutralizing antibodies to either VLA-4 or VLA-5 fully blocked the engraftment of the murine bone marrow by human SRCs/stem cells. These results, together with previous results which demonstrated that engraftment of human SRCs/stem cells is dependent on SDF-1/CXCR4, suggest that activation of the major integrins VLA-4, VLA-5, and LFA-1 on SRCs/stem cells by SDF-1 is essential for the multistep process of migration and engraftment.

In vitro studies have shown that spontaneous transendothelial migration of human CD34⁺ cells through HUVECs is dependent on LFA-1.15,16 Furthermore, anti–LFA-1 blocking antibodies completely prevented the IL-8–induced mobilization of murine hematopoietic stem cells.32 We demonstrate here that SDF-1–dependent transendothelial migration is partially dependent on LFA-1 (Figure 2B). These results suggest a role for LFA-1 in the process of homing and engraftment of hematopoietic stem cells. However, murine hematopoietic stem cells with radioprotective capacity can be both positively or negatively stained for LFA-1.33Moreover, it has been reported that CD34⁺ LFA-1 cells express LFA-1 within 24 hours in culture or during treatment with various cytokines.34 We have shown that the levels of ICAM-1 and VCAM-1 on endothelial cells differ between bone marrow and HUVECs (Figure 2A). In another study we found that SDF-1 is expressed by human bone marrow endothelial cells.35 This result was confirmed by others who also demonstrated that murine lung endothelial cells do not express SDF-1, which suggests that homing to the bone marrow is initiated by endothelial-bound SDF-1.36 

In the present study we discovered that LFA-1 is involved in the SDF-1–dependent transendothelial migration of CD34⁺/CXCR⁺ cells and in the engraftment of human SRCs/stem cells. Concomitantly, SDF-1 is capable of activating shear-resistant adhesion of CD34⁺/CXCR⁺ cells to ICAM-1. These results suggest that during the process of engraftment, LFA-1/ICAM-1 interactions are needed. The type of interactions between repopulating SRCs/stem cells and the recipient adherent cells appears to be determined by the repertoire of adhesion molecules expressed by the given endothelial or stromal cell. We found that antibodies to VLA-4 can partially block SDF-1–induced migration of CD34⁺/CXCR4⁺ cells through an endothelial cell layer formed by bone marrow–derived cell line MBA-2.1 but not through HUVECs. This fact and recently published results suggest a role for VLA-4/VCAM-1 in the extravasation process.37 Moreover, there is the possibility that the levels of VCAM-1 expressed by endothelial cells determine its usage by repopulating CD34⁺/CXCR4⁺ cells.

The involvement of SDF-1 in the rapid physiological shift from rolling behavior on endothelial cells to firm ICAM-1/LFA-1–dependent arrest of human CD4⁺ T lymphocytes suggests a similar mechanism of action for SDF-1 in the control of migrating CD34⁺/CXCR⁺ cells.38,39 Indeed, in this study, SDF-1 stimulated an integrin-mediated arrest of immature human CD34⁺ cells on vascular endothelium under shear flow.35 Interestingly, transendothelial migration of human CD34⁺ cells to a gradient of SDF-1 was also demonstrated to be dependent on E selectin.40 In contrast to transendothelial migration, transstromal migration of CD34⁺cells in response to SDF-1 is dependent on VLA-4 and VLA-5 but not LFA-1 (Figure 5). However, the role of SDF-1 and integrins in the extravasation process under shear flow and in the microenviromental localization of stem cells is still obscure.

Migration of CD34⁺/CXCR4⁺ cells through the ECM is dependent on both VLA-4 and VLA-5 (Figure 4). VLA-4, which binds to FN and VCAM-1, as well as VLA-5, which binds to the arginine-glycine-aspartic acid amino acid sequence (RGD) of FN, are believed to be important mediators of direct interactions between stem cells and the bone marrow microenvironment.18-20 Blocking antibodies against VLA-4 and VLA-5 inhibit the formation of day 12 spleen colonies by murine stem cells.41 Moreover, blocking antibodies to VLA-4 were recently shown to inhibit the homing of human CD34⁺ cells into the bone marrow of fetal sheep.42 VLA-5 is expressed by mouse stem cells and human long-term SCID repopulating cells and can also mediate their adhesion to FN.22 The role that such interactions play in the process of human stem cell repopulation, as well as the exact site at which such interactions occur, is unknown. In our studies, static adhesion of CB CD34⁺ cells and CFCs to FN is mainly mediated by VLA-5 but not VLA-4 (Figure 4). However, when fetal liver or bone marrow CD34⁺ cells were used, both integrins were equally important for the adhesion to FN.43In contrast to static adhesion of cord blood CD34⁺ cells, SDF-1–induced polarization and directional migration of CD34⁺/CXCR4⁺ cells in 3-D ECM-like gels is dependent on both VLA-4 and VLA-5. Thus, the process of migration is different from adhesion, and weak adhesion-deadhesion interactions between VLA-4 and its FN ligands are essential for the migration processes but not for firm adhesion interactions. It is therefore possible that VLA-5 is activated more in the initial stages of adhesion to the ECM, whereas both integrins are needed for the movement of cells through the ECM.

Our data suggest that during development and in clinical bone marrow transplantation, migration of human stem cells to and within the bone marrow microenvironment is mediated by SDF-1. Interestingly, the bone marrow of mice transplanted with fetal liver cells from CXCR4-deficient murine fetuses was engrafted. However, the levels of donor-derived multilineage hematopoiesis in the bone marrow of these mice was significantly reduced.44,45 Furthermore, CXCR4-null fetal liver cells recovered from the bone marrow of primary transplanted mice failed to repopulate secondary recipients.45 These results demonstrate that long-term repopulation by pluripotent stem cells capable of homing to the bone marrow and high levels of multilineage engraftment of primary and secondary transplanted recipients are absent in CXCR4-null fetal liver cells. CXCR4-null fetal liver cells can home to the bone marrow independently of CXCR4, but fail to proliferate and differentiate in the absence of CXCR4.

Broxmeyer et al46 have created a transgenic SDF-1 mouse and have found that SDF-1 is a survival factor for murine stem cells. In another study SDF-1 was also found to be a survival factor for human CD34⁺ cells.47 The small population of human SCID repopulating stem cells and primitive human stem cells capable of repopulating preimmune sheep were lately shown to be vascular endothelial growth factor receptor 2 (VEGFR2/KDR).48Studies done by Peichev et al49 and confirmed by us (data not shown) have shown that all the human hematopoietic CD34⁺ KDR⁺ cells are also CXCR4⁺. Furthermore, CXCR4 was also shown to be essential for the homing and engraftment of NOD/SCID mice by malignant human pre–B-acute lymphoblastic leukemia (ALL) cells, and its expression levels correlated with the NOD/SCID mice repopulating potential of human acute myeloid leukemia (AML) cells.50,51 

We and others have found that SDF-1, like ICAM-1 and VCAM-1, is specifically expressed on human bone marrow endothelial cells.35,36 We have also shown that SDF-1 can convert the rolling of CD34⁺ cells on endothelial cells into arrest.35 In addition, total body irradiation or conditioning with cytotoxic drugs, which are needed for stem cell transplantation, also significantly increases the levels of SDF-1 produced by bone marrow stromal cells.52 Lastly, homing of human CD34⁺ SRCs/stem and progenitor cells to the murine bone marrow or spleen is dependent on SDF-1 and CXCR4.53 It is evident, therefore, that CXCR4/SDF-1 interactions promote the homing and engraftment of human stem and progenitor cells. Alternative mechanisms of homing and engraftment by pluripotent stem cells may also exist. However, they are secondary to the SDF-1/CXCR4 pathway because CXCR4-null fetal liver cells lack true stem cell properties, and both SDF-1–null and CXCR4-null embryos have impaired bone marrow hematopoieis.

Based on our studies, we suggest the following scenario for the homing of human SRCs/stem cells to the bone marrow. Transplanted human CD34⁺/CD38^−/low/CXCR4⁺SRCs/stem cells that express LFA-1, VLA-4, VLA-5, and E and P selectin ligands54 reach the bone marrow and are recruited to specific vascular sites that constitutively express E/P selectin, ICAM-1, and VCAM-1. Upon activation with endothelium expressing or presenting SDF-1, LFA-1 and VLA-4 are activated on rolling stem cells to support their firm adhesion to the vessel wall. In response to SDF-1, the arrested human stem cells extravasate into the bone marrow ECM compartment (diapedesis) using LFA-1 and VLA-4. In the extravascular space, by using VLA-4 and VLA-5 for movement across FN, the stem cells polarize and migrate through the basal lamina toward local gradients of SDF-1, which are produced by specialized stromal cells, and orient themselves through the different elements of the bone marrow microenvironment and into the “stem cell niches” (Figure6).

In summary, we further identify repopulating human SRCs/stem cells as functionally expressing the integrins LFA-1, VLA-4, and VLA-5 both for migratory and adhesion processes triggered by endothelial- or stromal-associated SDF-1. Furthermore, our in vitro and in vivo data suggest a chemokine-dependent differential role for these major integrins in the multi-step process of migration, engraftment, and retention of human SRCs/stem cells in the murine bone marrow microenvironment.