Guanine nucleotide exchange factor Vav1 regulates perivascular homing and bone marrow retention of hematopoietic stem and progenitor cells

Abstract

Engraftment and maintenance of hematopoietic stem and progenitor cells (HSPC) depend on their ability to respond to extracellular signals from the bone marrow microenvironment, but the critical intracellular pathways integrating these signals remain poorly understood. Furthermore, recent studies provide contradictory evidence of the roles of vascular versus osteoblastic niche components in HSPC function. To address these questions and to dissect the complex upstream regulation of Rac GTPase activity in HSPC, we investigated the role of the hematopoietic-specific guanine nucleotide exchange factor Vav1 in HSPC localization and engraftment. Using intravital microscopy assays, we demonstrated that transplanted Vav1^−/− HSPC showed impaired early localization near nestin⁺ perivascular mesenchymal stem cells; only 6.25% of Vav1^−/− HSPC versus 45.8% of wild-type HSPC were located less than 30 μm from a nestin⁺ cell. Abnormal perivascular localization correlated with decreased retention of Vav1^−/− HSPC in the bone marrow (44–60% reduction at 48 h posttransplant, compared with wild-type) and a very significant defect in short- and long-term engraftment in competitive and noncompetitive repopulation assays (<1.5% chimerism of Vav1^−/− cells vs. 53–63% for wild-type cells). The engraftment defect of Vav1^−/− HSPC was not related to alterations in proliferation, survival, or integrin-mediated adhesion. However, Vav1^−/− HSPC showed impaired responses to SDF1α, including reduced in vitro migration in time-lapse microscopy assays, decreased circadian and pharmacologically induced mobilization in vivo, and dysregulated Rac/Cdc42 activation. These data suggest that Vav1 activity is required specifically for SDF1α-dependent perivascular homing of HSPC and suggest a critical role for this localization in retention and subsequent engraftment.

Hematopoietic stem cells (HSC) are defined by their ability to self-renew and the potential to generate all mature hematopoietic lineages, both during homeostasis and after transplantation. Maintenance of these functions critically depends on the interaction of HSC with one or several specialized microenvironments (so-called niches) in the bone marrow (BM) (1). Different cell populations have been proposed to form niches for HSC, including osteoblasts (2, 3), endothelial cells (4), stromal-derived factor-1 alpha (SDF1α)-expressing reticular cells (5), and nestin⁺ mesenchymal stem cells (MSC) (6). These components appear to be spatially related in the BM (5–7), but it is not clear whether they constitute a number of functionally distinct HSC niches or contribute to a complex multicellular hematopoietic microenvironment, and, specifically, there is conflicting evidence on the relative roles of osteoblastic versus vascular components of the niche in regulating hematopoietic stem and progenitor cell (HSPC) functions. Extracellular cues proposed to play a decisive role in hematopoietic stem/progenitor cell interaction with and retention in the niche include stem cell factor (SCF), concentration gradients of the chemoattractant CXCL12/SDF1α, and β₁ integrin-mediated adhesion to extracellular matrix and VCAM1⁺ endothelium (8–12).

The crucial intracellular pathways triggered by these signals are less well characterized, and how they are coordinated to regulate HSPC localization relative to different components of the niche is not known. Although previous work has provided insight into the role of Rho GTPases Rac1, Rac2 (13, 14), and Cdc42 (15) in HSPC trafficking and engraftment (reviewed in ref. 16), cells deficient in these GTPases show multiple proliferative, survival, and cytoskeletal defects, which precluded a more precise dissection of the key pathways regulating HSPC trafficking and localization. Thus, Rac-deficient HSPC show reduced chemotaxis to SDF1α and β₁ integrin-mediated adhesion in vitro but also defects in SCF-mediated proliferation and survival (14), consistent with the integration of multiple receptors and signaling pathways. In vivo, Rac1 is required for HSPC engraftment and endosteal localization in the BM, whereas deletion of both Rac1 and Rac2 not only impairs engraftment but causes massive HSPC mobilization (13, 14). Dissecting this complex regulation involves defining the functional interactions between GTPases and specific guanine nucleotide exchange factors (GEFs) that activate them (17), and this remains a major challenge.

Here we focus on the Vav subfamily of GEFs, which consists of three related proteins: Vav1 (hematopoietic-specific) and Vav2 and Vav3 (more broadly expressed) (18). Vav proteins have nonredundant functions in lymphopoiesis related to their role in immunoreceptor signaling. Thus, Vav1^−/− mice show defective T and B-1 cell development (19–21), whereas mutation of all three Vav genes completely impairs both the B- and T-lymphoid lineages (22). In neutrophils, Vav proteins are required for sustained β₂ integrin-mediated adhesion, intravascular endothelial crawling, and superoxide production (23–25). Despite the well-characterized function of Vav proteins in immune processes, virtually nothing is known about the role of these GEFs in primitive hematopoietic cells. We hypothesized that the hematopoietic-specific GEF Vav1 may regulate HSPC engraftment and retention by mediating responses to a subset of microenvironmental signals.

Therefore, we investigated the effect of genetic deletion of Vav1 on HSPC homing, localization, and retention in the BM microenvironment, and the role of Vav1 in mediating HSPC responses to chemokine and adhesion ligands. Our results suggest that Vav1 activity is required specifically for SDF1α-dependent perivascular homing of HSPC and suggest a critical role for this localization in BM retention and subsequent engraftment.

Results

Dysregulated Rho GTPase Activation in Vav1^−/− Hematopoietic Progenitors.

We reasoned that, if Vav1 is a biologically relevant and nonredundant GEF in HSPC, deletion of Vav1 in this cellular compartment would induce alterations in Rho GTPase activation. In addition, Vav1 itself would be expected to be activated by factors present in the BM microenvironment. Vav1 was rapidly phosphorylated in WT hematopoietic progenitors in response to in vitro stimulation by SCF/SDF1α (Fig. 1A Upper). Whereas Vav1^−/− cells did not show any agonist-induced tyrosine phosphorylation, a cross-reactive band was observed in nonstimulated Vav1^−/− cells. As Vav1 protein was absent from Vav1^−/− immunoprecipitates (Fig. 1A Lower), this suggested a compensatory, abnormal increase in the activation of another Vav protein. Indeed, increased baseline phosphorylation of Vav3 was detected with a phosphospecific antibody in Vav1^−/− progenitors (Fig. 1B).

Fig. 1.

Dysregulated Rho GTPase activation in Vav1^−/− hematopoietic progenitors. (A) Vav1 activation as demonstrated by immunoprecipitation (IP) with a Vav1-specific antibody, followed by detection with phosphotyrosine (P-Tyr) antibody. WT or Vav1^−/− lineage-depleted cells were starved in 1% FBS for 6 h and stimulated with SCF + SDF1α (100 ng/mL each) for the indicated time points. WB, Western blot. (B) Vav3 phosphorylation in WT and Vav1^−/− progenitors, detected with a phosphospecific antibody. (C) Levels of active (GTP-bound) and total Rac and Cdc42 in WT or Vav1^−/− lineage-depleted cells. Cells were starved and stimulated as in A. GTP-bound Rho GTPases were precipitated with agarose-conjugated PAK1-p21-binding domain (PBD) and detected by Western blot. (D) Activation status of different signaling pathways in WT and Vav1^−/− progenitors, analyzed by Western blot with phosphospecific antibodies. Lineage-depleted cells were either freshly purified (B, baseline activation) or starved in 1% FBS for 6 h and then stimulated with SCF + SDF1α (100 ng/mL each) for the indicated time points. Arrows indicate phospho-Vav3 (B) and the different isoforms of phospho-PAK and phospho-JNK (D).

We observed two distinct abnormalities in hematopoietic progenitors isolated from the BM of Vav1^−/− mice. First, the baseline levels of GTP-bound (active) Rac and Cdc42 were unexpectedly elevated, which is consistent with the dysregulated activation of either or both Vav2 and Vav3. The total level of Rac and Cdc42 proteins was also increased, compared with WT progenitors (Fig. 1C). Second, in the absence of Vav1, both Rac and Cdc42 were unresponsive to in vitro stimulation by the combination of SCF and SDF1α, known critical regulators of HSPC engraftment and activators of Rac and Cdc42 GTPases in HSPC (13–15) (Fig. 1C). To determine whether these changes were physiologically relevant, we studied known downstream targets of Rac which we have previously demonstrated to be involved in HSPC homing and retention (13, 14, 26, 27). Vav1^−/− progenitors showed increased baseline activation of the Rac/Cdc42 effector p21-activated kinase (PAK) and reduced activation of PAK after SCF/SDF1α stimulation (Fig. 1D). There were no measurable changes in other kinase signaling pathways, including ERK, JNK, and AKT (Fig. 1D). These results suggested that Vav1 may function as a critical signaling molecule regulating HSPC functions.

Abnormal Responses to SDF1α in Vav1^−/− HSPC.

The BM of Vav1^−/− mice contained normal numbers of immunophenotypically defined long-term HSC, modestly increased frequency of short-term HSC, and slightly reduced numbers of common myeloid progenitors and granulocyte-monocyte progenitors as defined by surface markers (Fig. S1A). However, the frequency of functionally defined myeloid progenitors in the BM, as determined by in vitro growth in methylcellulose, was similar to that of WT mice; we did not observe any alterations in the number of colony-forming cells (CFC) obtained in the presence of SCF or granulocyte-colony-stimulating factor (G-CSF) alone or multiple cytokines, or in the frequency of high-proliferation potential colony-forming cells (HPP-CFC), which represent more primitive progenitors (Fig. S1B). Moreover, the frequency of cobblestone area-forming cells (CAFC) in Vav1^−/− BM was virtually identical to that of WT mice both at early (reflecting mature progenitors) and late (representing more primitive HSPC) time points (Fig. S1C). Consistent with these data, analysis of the cell cycle (Fig. S1D) and apoptosis (Fig. S1E) of immunophenotypically defined HSPC subsets in the BM of Vav1^−/− mice did not reveal any alterations.

We have previously shown that deficiency of the Vav target Rac is associated with massive mobilization of HSPC from the BM. However, analysis of the number of progenitors in peripheral blood of Vav1^−/− mice unexpectedly revealed significantly decreased numbers of circulating CFC and absence of circadian variation (Fig. 2A). Because circadian fluctuations in the numbers of circulating HSPC inversely correlate with periodic changes in SDF1α levels in the BM microenvironment (28), these data suggested an abnormal response of Vav1^−/− HSPC to SDF1α. To test this hypothesis, we treated Vav1^−/− and WT mice with the SDF1α receptor (CXCR4) antagonist AMD3100, known to induce HSPC mobilization (29). AMD3100 mobilized only 55% of the number of Vav1^−/− progenitors that the same treatment induced in WT mice (739 vs. 407 CFC/mL, considering untreated mice as baseline; Fig. 2B). G-CSF-induced progenitor mobilization, which is mediated at least in part by SDF1α down-regulation (30) or proteolytic degradation (31) in the BM, was also severely impaired in Vav1^−/− mice (Fig. 2C). Thus, in the absence of Vav1, HSPC are relatively insensitive to physiologic or induced changes in SDF1α levels or activity in the BM in vivo.

Fig. 2.

Abnormal SDF1α responses in Vav1^−/− HSPC. (A) Day–night variation in the number of CFC in peripheral blood of WT and Vav1^−/− mice. Blood was collected 4 or 16 h after the onset of light. Data represent mean ± SD, n = 4–6; *P < 0.05 (t test). NS, nonsignificant. (B) Numbers of CFC in peripheral blood of WT and Vav1^−/− mice, either untreated or treated with 5 mg/kg of AMD3100 for 1 h. Data represent mean ± SD, n = 6–7; *P < 0.05 (t test). (C) Numbers of CFC in peripheral blood of WT and Vav1^−/− mice, either untreated or treated with G-CSF at a daily dose of 200 μg/kg for 6 d. Data represent mean ± SD, n = 3; *P < 0.05, **P < 0.005 (t test). The graph shows one of two experiments that yielded similar results. (D–F) In vitro chemotaxis of WT or Vav1^−/− LSK cells on a fibronectin-coated coverslip in the presence of an SDF1α gradient, determined by time-lapse microscopy. (D) Paths followed by individual cells in 1 h. Representative fields (325 × 325 μm) from one out of three similar experiments are shown. (E) Net path length migrated in 1 h (mean ± SEM, n = 36–38 cells per genotype) and (F) speed of migration (mean ± SEM, n = 14–41). For E and F, one representative experiment of a total of three is shown. **P < 0.005 (t test).

SDF1α acts both as a chemoattractant, being the only chemokine known to be active on HSPC (32), and as a signal that increases β₁ integrin affinity for ligand binding in different cell types, including hematopoietic progenitors (“inside-out signaling”) (33, 34). To dissect the requirement of Vav1 in SDF1α-mediated HSPC functions, we studied the role of Vav1 in each of these processes in vitro. Vav1^−/− lin⁻ Sca1⁺ ckit⁺ (LSK) cells observed by time-lapse microscopy showed significantly abnormal migratory response to an SDF1α gradient (Fig. 2 D–F and Movies S1 and S2). Compared with WT LSK cells, the distance migrated by Vav1^−/− cells in 1 h (Fig. 2E) and their speed (Fig. 2F) were significantly decreased. Absence of Vav1 did not affect the directionality of cell migration. On the other hand, Vav1 was not required for baseline or SDF1α-induced β₁ integrin-mediated adhesion, measured either as static adhesion to fibronectin- or VCAM1-coated plates (Fig. S2A) or as the resistance of adherent cells to detachment from an immobilized β₁ integrin ligand under conditions of shear stress (Fig. S2B). SDF1α stimulation induced β₁ integrin-mediated firm adhesion of WT and Vav1^−/− progenitors to a similar extent, suggesting that deletion of Vav1 does not affect SDF1α-mediated inside-out β₁ integrin signaling (Fig. S2B). Surface expression of β₁ integrin subunits α₄ and α₅ was similar in WT and Vav1^−/− LSK or lin⁻ ckit⁺ Sca1⁻ cells (Fig. S2C). Together, these data suggest that Vav1 mediates chemotactic but not adhesive responses to SDF1α in HSPC.

Engraftment Defect of Vav1^−/− HSPC.

SDF1α/CXCR4 signaling is required for engraftment and retention of HSPC in the BM after transplantation (8, 9, 35). To further study the role of Vav1 in SDF1α responses in vivo, we determined whether Vav1^−/− HSPC were able to engraft and reconstitute hematopoiesis. In initial repopulation assays, WT and Vav1^−/− BM cells were transduced with retroviral vectors encoding two different fluorescent proteins and transplanted into isogenic recipients. In this experimental setting, Vav1^−/− cells displayed dramatically reduced engraftment ability (Fig. S3). This experiment was also performed with the reverse color combination, with similar results. To eliminate possible effects of in vitro manipulation on engraftment of Vav1^−/− cells, we performed competitive repopulation assays with freshly isolated, unmanipulated cells. Vav1^−/− cells did not contribute significantly to donor chimerism in peripheral blood at any time point after transplant (Fig. 3A), and were absent from BM and spleen 4 mo posttransplant (Fig. 3B). All three lineages examined (myeloid, B-lymphoid, and T-lymphoid) demonstrated defective repopulation in mice transplanted with Vav1^−/− BM cells (Tables S1 and S2). In contrast, WT BM cells showed stable chimerism at a level consistent with the input (53.5–63.2%), arguing against the possibility that the different genetic background of donor and recipient mice (C57BL/10J versus C57BL/6J) may be responsible for the engraftment failure of Vav1^−/− cells.

Fig. 3.

Engraftment defect of Vav1^−/− BM cells. (A) Peripheral blood chimerism of lethally irradiated B6.SJL (CD45.1) mice transplanted with 3 × 10⁶ WT or Vav1^−/− (CD45.2) BM cells and an equal number of WT CD45.1 BM cells as competitors. (B) CD45.2 chimerism in BM and spleen for the recipient mice in A, killed 4 mo posttransplant. Data in A and B are expressed as the percentage of CD45.2⁺ cells in the nucleated fractions measured by flow cytometry. Mean ± SD, n = 5, *P < 0.005 (t test). Note: Chimerism for Vav1^−/− cells in BM and spleen was less than 2%. (C) Survival of lethally irradiated B6.SJL mice transplanted with 3 × 10⁶ WT or Vav1^−/− (CD45.2) BM cells in the absence of competitor cells; n = 10 mice per genotype. (D) Leukocyte counts in peripheral blood of the recipient mice in C surviving at each time point. Mean ± SD, n = 6–10, *P < 0.005 (t test).

Vav1^−/− BM cells also performed poorly in noncompetitive repopulation assays: A fraction of mice (4/10, 40%) transplanted with Vav1^−/− BM in the absence of competitor cells died within the first month posttransplant, whereas all animals (10/10) receiving WT cells survived (Fig. 3C). Surviving mice engrafted with Vav1^−/− BM showed significantly delayed hematopoietic reconstitution, compared with mice receiving WT cells (Fig. 3D). Taken together, these data demonstrate a significant defect in the capacity of Vav1^−/− HSPC to engraft in transplant reconstitution assays, despite the normal progenitor content and even an elevated frequency of phenotypically defined, short-term HSC in the BM of Vav1^−/− mice. This phenotype is observed at all time points analyzed, suggesting a defect common to both short- and long-term repopulating cells.

Vav1^−/− HSPC Show Abnormal Homing Toward Nestin⁺ MSC.

To gain further mechanistic insight into this engraftment defect, we studied the role of Vav1 in HSPC homing and microlocalization in the BM. Vav1^−/− HSPC showed normal homing efficiency as determined by CFC assay at 16 h (Fig. 4A) and by intravital microscopy of 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindodicarbocyanine perchlorate (DiD)-labeled LSK cells 1 h posttransplant (Fig. 4B). Therefore, Vav1 is not required for initial homing of HSPC into the BM cavity. Because SDF1α is expressed at highest levels by perivascular nestin⁺ MSC (6) and at lower levels by other cell populations including osteoblasts (5, 36, 37), we directly visualized the homing of DiD-labeled WT or Vav1^−/− LSK cells to defined regions of the calvarium BM using intravital microscopy (7). The use of two transgenic reporter mouse strains, Nestin-GFP and Col2.3-GFP, allowed the study of HSPC localization with respect to perivascular nestin⁺ MSC and osteoblastic cells, respectively.

Fig. 4.

Bone marrow homing and localization of Vav1^−/− HSPC. (A) Homing efficiency of WT and Vav1^−/− progenitors (CFC) to the BM of C57BL/10J recipients 16 h after transplant. Data represent mean ± SD, n = 5. (B) Number of DiD⁺ LSK cells detected by intravital microscopy in a 4 × 6 mm region of the calvarium of Col2.3-GFP mice 1 h after transplant, for an input of 30,000 LSK cells (mean ± SD, n = 3). (C–E) Distances (in μm) of transplanted LSK cells to nestin⁺ cells (C), osteoblastic cells (D), and endosteal surface (E) in the calvarium of Nestin-GFP (C) or Col2.3-GFP (D and E) mice, determined 1 h after transplant. Data represent measurements for individual LSK cells; horizontal lines represent the mean. C represents pooled data from three mice per genotype (n = 24–32); D and E represent data from one representative mouse per genotype, out of three analyzed (n = 32–34; total number of events analyzed, n = 77–98). *P = 0.005 (t test).

Compared with WT HSPC, Vav1^−/− HSPC localized significantly farther from nestin⁺ cells 1 h after transplant. Only 6.25% of Vav1^−/− HSPC, compared with 45.8% of WT HSPC, were located less than 30 μm from a nestin⁺ cell (n = 24–32 HSPC analyzed). In addition, 25% of Vav1^−/− HSPC versus only 8.3% of WT HSPC were found in microscopic fields (330 × 330 μm) lacking nestin⁺ cells. Considering only WT and Vav1^−/− HSPC with nestin⁺ cells in their proximity, the mean distance of HSPC to the nearest nestin⁺ cell was significantly shorter for WT versus Vav1^−/− HSPC (WT, 39.54 ± 6.91 μm; Vav1^−/−, 65.24 ± 5.22 μm; n = 22–24; P = 0.005) (Fig. 4C and Fig. S4). In contrast, the distribution of Vav1^−/− HSPC relative to osteoblasts (Fig. 4D and Fig. S4) and endosteal surface (Fig. 4E) 1 h after transplant was not significantly different from that of WT HSPC (n = 77–98 HSPC analyzed).

Impaired Retention of Vav1^−/− HSPC in the BM Niche.

To assess retention of transplanted HSPC in the BM, we examined by intravital microscopy the number and localization of DiD-labeled LSK cells present in the calvarium BM cavity at 48 h posttransplant. Contrary to the earlier time point, the total number of Vav1^−/− LSK cells in the BM cavity at 48 h was significantly decreased, compared with WT cells in both Nestin-GFP and Col2.3-GFP recipients (60% and 44% reduction, respectively; P < 0.05 for both strains, for a total of three to five recipient mice analyzed per strain and genotype; Fig. 5A), suggesting impaired retention. At this time point, there was not any significant difference in the localization of retained Vav1^−/− HSPC with respect to nestin⁺ cells, osteoblasts, or endosteal surface, compared with WT HSPC (Fig. 5 B–D). We attempted to determine whether this reduction in the number of donor cells in the BM correlated with their increase in circulation; however, the numbers of circulating donor cells detected at this time point (<0.25% donor chimerism in peripheral blood 48 h after transplant of 1.5 × 10⁵ lin⁻ ckit⁺ cells, for both genotypes) was too low to allow any conclusions.

Fig. 5.

Reduced retention of Vav1^−/− HSPC in the BM. (A) Number of DiD⁺ LSK cells detected by intravital microscopy in a 4 × 6 mm region of the calvarium of Nestin-GFP and Col2.3-GFP mice 48 h after transplant (mean ± SD, n = 4–5) for an input of 30,000 LSK cells. *P < 0.05 (t test). (B–D) Distances (in μm) of transplanted LSK cells to nestin⁺ cells (B), osteoblastic cells (C), and endosteal surface (D) in the calvarium of Nestin-GFP (B) or Col2.3-GFP (C and D) mice, determined 48 h after transplant. Data represent measurements for individual LSK cells; horizontal lines represent the mean. B represents pooled data from three mice per genotype (n = 29–71); C and D represent data from one representative mouse per genotype, out of three analyzed (n = 38–59; total number of events analyzed, n = 102–139).

Although Vav1^−/− HSPC did not demonstrate any alteration in proliferation or apoptosis (Fig. S1D and E) in primary mice during homeostatic hematopoiesis, it was still possible that defects in the proliferation or survival of these cells following transplantation—rather than reduced retention—could account for the reduced numbers of Vav1^−/− HSPC observed at the 48 h imaging time point. To determine the proliferation status of transplanted hematopoietic progenitors shortly after the initial BM homing, we injected WT or Vav1^−/− lin⁻ ckit⁺ cells into lethally irradiated recipients, which were pulsed with BrdU 48 h after transplant. There was no significant difference in the fraction of donor-derived (CD45.2⁺) BM cells incorporating BrdU between the two genotypes (45.8 ± 6.10% vs. 50.5 ± 5.8%, WT vs. Vav1^−/−, n = 5–6 mice per genotype; Fig. S5A). An independent measure of HSPC proliferation after transplant was obtained from in vivo imaging experiments, in which the presence of clusters of DiD⁺ cells with a reduction in dye intensity can be taken as an indication of proliferation of single transplanted cells (7). The fraction of proliferating LSK cells, as determined by the percentage of cell clusters 48 h after transplant, was similar for WT and Vav1^−/− cells (38.0 ± 12.4% vs. 42.9 ± 14.2%, WT vs. Vav1^−/−, n = 4–5 mice per genotype; Fig. S5B), consistent with the BrdU-labeling studies. Likewise, the fraction of lin⁻ ckit⁺ cells that were apoptotic (Annexin V⁺) 48 h after transplant was not significantly different between the two genotypes (Fig. S5C). These data are consistent with normal ERK, JNK, and AKT activation previously noted (Fig. 1B) and support the hypothesis that the failure of Vav1^−/− cells to engraft is due to reduced retention in the BM. Taken together, these data imply that Vav1 is required in HSPC for a normal response to SDF1α leading to Rac/Cdc42 activation, perivascular localization, retention in the BM niche, and engraftment.

Discussion

Previous work has demonstrated that Rho GTPases function as central molecules critical for the interaction of HSPC with the BM hematopoietic microenvironment. Inactivation of Rac1 alone impairs HSPC engraftment, whereas deletion of both Rac1 and Rac2 (13, 14) or Cdc42 (15) leads to massive egress of HSPC from the BM. Rac and Cdc42 are activated by multiple receptors via a large family of Rho GEFs, of which ∼70 members have been described so far (17). Thus, genetic studies examining the phenotype of Rho GTPase knockouts, even when performed in a lineage-specific fashion, likely demonstrate the combinatorial effects of multiple receptors and downstream signaling pathways. Indeed, one of the challenges in understanding the specific roles of chemokine and adhesion receptors in HSPC function is that many downstream GEFs may activate multiple Rho GTPases in an apparently redundant and cell- or agonist-specific fashion.

Here we have demonstrated a role for Vav1 as a key functional regulator of Rac/Cdc42 in HSPC, as shown by the fact that deletion of this single GEF completely alters Rho GTPase activation patterns in hematopoietic progenitors. Vav1^−/− progenitors show defective activation of Rac and Cdc42 and their effector PAK in response to SDF1α and SCF, agonists previously shown to activate Rac in HSPC and critical in HSPC engraftment. Unexpectedly, however, Vav1^−/− progenitors also showed baseline hyperactivation of Rac and Cdc42, probably due to compensatory hyperactivation of Vav2/Vav3.

Although Vav- and Rac-deficient HSPC exhibit similar engraftment defects in repopulation assays, their phenotypes are in most other respects remarkably different, allowing a more precise dissection of the critical pathways regulating HSPC trafficking and interaction with the hematopoietic microenvironment. Whereas Rac-deficient HSPC show alterations in multiple cellular processes involved in engraftment and retention (SCF-induced proliferation, chemotaxis, homing, survival, and β₁ integrin-mediated adhesion), the engraftment failure of Vav1^−/− HSPC appears to be specifically associated with a much more restricted defect: abnormal response to SDF1α leading to abnormal localization and reduced retention, underscoring the key role of this distinct pathway in HSPC engraftment. Interestingly, the lack of postengraftment retention is not related to the absence of any immunophenotypically or in vitro defined HSPC population in Vav1^−/− mice, suggesting that signaling pathways are qualitatively or quantitatively differentially used in posttransplant versus ontogenic migration and expansion of HSPC. On the other hand, Vav1^−/− and Rac1^−/−;Rac2^−/− mice unexpectedly show opposite mobilization phenotypes, with decreased versus highly increased numbers of circulating progenitors, respectively. The findings that both Vav1^−/− and Rac1^−/− HSPC, which have normal β₁ integrin-mediated adhesion, do not show increased mobilization, whereas Rac2^−/− and Rac1^−/−;Rac2^−/− HSPC mobilize in inverse correlation to their decreased binding to fibronectin, imply that HSPC mobilization may be more dependent on loss of integrin-mediated adhesion than on migration per se, at least as measured by in vitro assays.

These data also suggest that different Rho GEFs play selective roles downstream of different extracellular signals, thus controlling distinct cellular processes. Thus, inactivation of other GEFs expressed in HSPC leads to phenotypes strikingly different from Vav1^−/−. Similar to combined genetic deletion of Rac1 and Rac2, pharmacologic inhibition of Tiam1/Trio-mediated Rac activation by NSC23766 induces HSPC mobilization (13), suggesting that these GEFs may be required for integrin-mediated Rac activation and HSPC adhesion. Conversely, deletion of Vav1, which is not targeted by NSC23766 (38), does not affect adhesion nor induce mobilization but leads to defective SDF1α responses.

Vav1 is required for appropriate HSPC responses to SDF1α under both physiologic conditions and in the posttransplant hematopoietic microenvironment. Thus, Vav1^−/− HSPC showed impaired in vitro chemotaxis; reduced mobilization induced by AMD3100 or G-CSF (both of which depend on CXCR4/SDF1α inactivation); absence of physiologic, circadian mobilization [related to oscillations in SDF1α levels in the BM (28)]; and reduced homing to the proximity of SDF1α-expressing nestin⁺ MSC, recently implicated as perivascular cells that contribute to the HSC niche and are required for HSC maintenance (6). These observations delineate a specific requirement for Vav1 in the proposed role of the CXCR4/SDF1α axis in the regulation of HSPC trafficking and engraftment. Although initial data showed that homing of HSPC is blocked by anti-CXCR4 antibodies (39), subsequent studies found that Cxcr4^−/− progenitors home normally (8), which is consistent with our findings in Vav1^−/− cells. Taken together, these studies suggest that this pathway is dispensable for the initial BM homing. However, CXCR4 signaling is required for the engraftment and retention of human HSPC in non-obese diabetic (NOD)/SCID mice (9), and murine Cxcr4^−/− HSPC show reduced retention in the BM microenvironment after transplantation (8, 35), which parallels our observations with Vav1^−/− HSPC. Together, these data support the idea that the initial bone marrow homing and subsequent retention are two components of the engraftment process with distinct biochemical regulation.

Vav1^−/− HSPC appear to retain partial responsiveness to an SDF1α gradient, being able to migrate short distances and being mobilized at lower levels by AMD3100. Vav2 and Vav3, which have been shown to play both unique and overlapping roles in lymphocytes, may play partially compensatory roles in some of the migratory and adhesive functions of HSPC, which may allow the retention of Vav1^−/− HSPC under steady-state conditions but may not be sufficient under the more stringent conditions of a BM transplant. Vav3 (and possibly Vav2) is hyperactivated in unstimulated Vav1^−/− progenitors (Fig. 1B).

The defect in early localization of Vav1^−/− HSPC near nestin⁺ MSC correlates with impaired BM retention and engraftment. In contrast, correct positioning of Vav1^−/− HSPC with respect to osteoblastic cells or the matrix of the endosteal surface was not sufficient for functional engraftment. These data do not rule out a role for the osteoblastic space in long-term HSPC maintenance; however, they support the idea that localization in the proximity of a perivascular mesenchymal cell is required in the initial stages of HSPC homing and is important for their early retention. Nestin⁺ MSC are major producers of SDF1α in the BM (6), and we have shown that Vav1 is required for appropriate responses of HSPC to SDF1α, consistent with the abnormal localization of Vav1^−/− HSPC in the perivascular, SDF1α-abundant space. Thus, we describe a signaling pathway differentially regulating HSPC localization with respect to perivascular versus osteoblastic niche components.

Together, our data imply that the hematopoietic-specific GEF Vav1 is essential for the correct response of HSPC to SDF1α in the BM microenvironment, leading to Rac/Cdc42 activation, perivascular localization, retention in the BM niche, and subsequent engraftment, and define a critical biochemical pathway in the regulation of Rho GTPase activity in HSPC.

Methods

Mice.

All procedures involving mice followed Children's Hospital Boston Institutional Animal Care and Use Committee guidelines. Vav1^−/− mice have been previously reported (20) and were backcrossed into a C57BL/10J (CD45.2) background. Age- (8- to 16-wk-old) and sex-matched C57BL/10J mice (Jackson Laboratory) were used as WT controls. B6.SJL (CD45.1) mice (Jackson Laboratory) were used as recipients in some transplant assays. All recipient mice were lethally irradiated (¹³⁷Cs source, 11.5-Gy whole-body irradiation) before BM transplant.

Transplants, Progenitor Assays, and Flow Cytometry.

Hematopoietic repopulation ability was assessed in both competitive and noncompetitive assays (see SI Methods for details). Progenitor content in BM and peripheral blood was determined by methylcellulose CFC assays, as previously described (13, 14). The frequency of circulating HSPC was determined under different conditions (diurnal vs. nocturnal, AMD3100, G-CSF). In vivo HSPC proliferation and apoptosis were determined by BrdU incorporation and Annexin V binding assays (BD Biosciences), respectively, either in primary, unmanipulated mice or in transplanted lin⁻ ckit⁺ cells 48 h posttransplant. Biochemical assays, immunophenotyping, and progenitor homing assays were performed essentially as described elsewhere (13, 14).

Chemotaxis and Adhesion Assays.

In vitro migration in response to SDF1α was measured by time-lapse microscopy in Dunn chemotaxis chambers, as described (14) (see also SI Methods). Static adhesion assays were performed as previously described (13, 14). Firm adhesion and the effect of SDF1α on adhesion were measured with a computer-operated flow-chamber device (Mirus Nanopump; Cellix; SI Methods).

Intravital Microscopy.

HSPC microlocalization in the BM was analyzed by intravital microscopy using previously described equipment and procedures (7). The use of two transgenic reporter mouse strains, nestin-GFP and Col2.3-GFP, allowed the study of HSPC localization with respect to perivascular nestin⁺ MSC and osteoblastic cells, respectively. Cells were imaged at two time points posttransplant, 1 h (to determine the initial homing and localization) and 48 h (to assess short-term retention and the effect of proliferation as determined by clusters of DiD⁺ cells with generally reduced dye intensity).