The Wave2 scaffold Hem-1 is required for transition of fetal liver hematopoiesis to bone marrow

The transition of hematopoiesis from the fetal liver (FL) to the bone marrow (BM) is incompletely characterized. We demonstrate that the Wiskott–Aldrich syndrome verprolin-homologous protein (WAVE) complex 2 is required for this transition, as complex degradation via deletion of its scaffold Hem-1 causes the premature exhaustion of neonatal BM hematopoietic stem cells (HSCs). This exhaustion of BM HSC is due to the failure of BM engraftment of Hem-1−/− FL HSCs, causing early death. The Hem-1−/− FL HSC engraftment defect is not due to the lack of the canonical function of the WAVE2 complex, the regulation of actin polymerization, because FL HSCs from Hem-1−/− mice exhibit no defects in chemotaxis, BM homing, or adhesion. Rather, the failure of Hem-1−/− FL HSC engraftment in the marrow is due to the loss of c-Abl survival signaling from degradation of the WAVE2 complex. However, c-Abl activity is dispensable for the engraftment of adult BM HSCs into the BM. These findings reveal a novel function of the WAVE2 complex and define a mechanism for FL HSC fitness in the embryonic BM niche. Hematopoietic stem cells (HSCs) migrate from the fetal liver to the bone marrow (BM) during embryogenesis. Here the authors show that the WAVE2 complex scaffold Hem1 is required for engraftment of HSCs in BM, not through its canonical role regulating actin polymerization, but through c-Abl survival signaling.

We examined the role of the WAVE2 complex scaffold Hem-1 in the migration of FL HSC to the BM. Deletion of Hem-1 resulted in degradation of the WAVE2 complex [21][22][23][24] , but surprisingly the migration of FL HSC to the fetal BM was not altered. Rather, after arriving in the fetal marrow niche, Hem-1 −/− FL HSC underwent apoptosis. Within 6-8 weeks Hem-1 −/− mice underwent marrow fibrosis and hematopoietic failure, and subsequently died. Neither FL nor young marrow Hem-1 −/− HSC could engraft irradiated wild-type (wt) adult mice. Without the WAVE2 complex present, Abi-1 was not present to activate downstream c-Abl signaling. Reconstituting c-Abl expression rescued Hem-1 −/− HSC survival in the adult marrow niche. This study implies that FL HSCs are not fit for BM occupancy until after a niche survival signal mediated by the WAVE complex through c-Abl. This defines a novel WAVE complex function, survival signaling, and sheds light on the regulation of the transition of hematopoiesis from the FL to the marrow during development.

Results
Premature death and hematopoietic defects in Hem-1 −/− mice. We hypothesized that the hematopoietic-specific WAVE complex scaffold Hem-1 is important for FL HSC transition to the BM. In the present study, Hem-1 was constitutively deleted in a murine model to assess fetal HSC development and migration (Supplementary Fig. 1a-d). Constitutive deletion permitted study of whether Hem-1 was essential for the development of any other organ system outside the hematopoietic system. In addition, it ensured that all HSCs had the gene deleted, and therefore a small number of HSC escaping conditional deletion could not skew the study. Intercrosses of Hem-1 +/− mice produced Hem-1 +/+ , Hem-1 +/− , and Hem-1 −/− E14.5 fetuses at the expected Mendelian ratio (Fig. 1a). However, Hem-1 −/− mice exhibited growth retardation and died prematurely after birth, with an average life expectancy of 6 weeks (Fig. 1b, c). These abnormalities were associated with a dramatic defect in BM hematopoiesis, including a significant reduction in the number of total BM nucleated cells (BMCs), BM phenotypic HSCs and hematopoietic progenitor cells (HPCs), and BM cobblestone area-forming cells (CAFCs) in 5-week-old Hem-1 −/− compared with littermate Hem-1 +/+ mice of the same age ( Fig. 1d-h). In addition, Hem-1 −/− mice developed a myelofibrosis-like disease with BM fibrosis, as demonstrated by an increase in reticulin staining; extra-medullary hematopoiesis, neutrophilia, and lymphopenia (Supplementary Fig. 2a-d). Heterozygote Hem-1 +/− mice developed normally, similar to Hem-1 +/+ mice, and showed none of the abnormalities observed in Hem-1 −/− mice.
Hem-1 −/− FL HSCs can migrate to the BM. FL HSCs transition to the BM starting around E16.5-17.5, and continues briefly after birth 1-3 . This transition requires significant cell migration and adherence. Therefore, we next examined whether Hem-1 deletion leads to defects in FL HSC actin polymerization, migration, adherence, and homing to the BM. Unexpectedly, HSC-enriched Lin − /Sca-1 + /Kit + (LSK) E14.5 Hem-1 −/− FLCs showed no defects in F-actin polymerization, actin capping, and migration in response to the HSC chemokine stromal derived factor-1 alpha (SDF-1α), compared to littermate Hem-1 +/+ equivalent cells (Fig. 3a, b). Hem-1 −/− FL Lin − cells could adhere to fibronectin equally as well as the cells from Hem-1 +/+ littermates (Fig. 3b). In addition, E14.5 Hem-1 −/− FL LSK cells expressed levels of HSC adhesion and BM homing components (CXCR4, VLA-4, VLA-5, Tie2) equal to or greater than E14.5 Hem-1 +/+ FL LSK cells ( Supplementary Fig. 4). In contrast, neutrophils from Hem-1 −/− mice are defective in fMLP-stimulated F-actin polymerization, actin capping and migration, and adhesion to fibronectin as the cells from Hem-1 mutant mice reported previously (Supplementary Fig. 5) 21 . Furthermore, we found that inhibition of CDC42 with a specific inhibitor, CASIN, suppressed both E14.5 Hem-1 +/+ and Hem-1 −/− FL LSK cell adhesion and migration in vitro (Fig. 3b). These findings suggest that unlike neutrophils and other hematopoietic cells in adult mice, FL HSCs can migrate and home to the BM independent of the WAVE complex, perhaps via the CDC42 Wiskott-Aldrich syndrome protein (WASP) pathway. This is consistent with the observation that HSCs from WASP-deficient mice had decreased BM homing capability in association with a defect in adhesion to collagen 31 .
Hem-1 −/− FL HSCs cannot survive in the BM. We then measured the ability of Hem-1 −/− E14.5 CD45.2 FL LSK cells to   e E14.5 fetal liver hematopoietic stem and progenitor cells subsets are not different between Hem-1 −/− and littermate Hem-1 +/+ mice (n = 10). f E14.5 FL cobblestone area-forming cells (CAFCs) are not different between Hem-1 −/− and Hem-1 +/+ mice (n = 3). g Five-week Hem-1 −/− BM exhibits decreased hematopoietic stem and progenitor cell subsets (n = 5, **p < 0.01, ***p < 0.001, Student's t test). h Five-week Hem-1 −/− BM CAFCs are reduced compared to littermate Hem-1 +/+ mice (n = 3, *p < 0.05, **p < 0.01, Poisson statistics). Error bars represent the mean ± SD survive and proliferate after migration to the BM compared to littermate equivalent Hem-1 +/+ cells (Fig. 3e). We found that Hem-1 −/− E14.5 CD45.2 FL LSK had both a higher rate of apoptosis and a decreased rate of cycling cells compared to equivalent cells from Hem-1 +/+ littermate controls. This suggests that HSC-enriched LSK cells from the E14.5 Hem-1 −/− FL are just as capable of migrating to the BM and homing to the osteoblast niche as their normal counterparts, but cannot survive and proliferate once there 31 . Thus, Hem-1 deletion does not impair FL to BM hematopoietic cell homing or adherence to the niche, suggesting that the WAVE2 complex has a distinct function in FL HSPCs besides regulating cell migration and adhesion by mediating survival and expansion after migration from the FL to the BM. This is consistent with the observation that knockdown of WAVE2 had no significant effect on HSC migration to the BM but prevented HSCs from expanding in the BM 17 .
However, the mechanism by which the WAVE complex regulates HSC expansion in the BM was unknown, and thus was studied further. First, we performed a detailed analysis of HSCs in the BM from E18.5 through postnatal day (PD) 1, PD3 and PD7 in both Hem-1 −/− and Hem-1 +/+ fetuses and neonates. As expected, we found that the number of HSCs in the BM was low in E18.5 Hem-1 +/+ fetuses and then increased rapidly from PD1 to PD7 in Hem-1 +/+ neonates (Fig. 4a). By contrast, Hem-1 −/− neonates exhibited a moderate increase in the number of BM HSCs on PD3 and then an abrupt decrease on PD7 to a level similar to that in E18.5 fetuses (Fig. 4a). We then measured the cell cycle distribution and apoptotic fraction of Hem-1 −/− BM HSPCs compared with those of Hem-1 +/+ littermate equivalent cells on E18.5, PD1, PD3 and PD7 using Ki-67/7-aminoactinomycin-D (7-AAD) and Annexin V staining, respectively. The cell cycle distribution was not significantly different between the cells from Hem-1 +/+ and Hem-1 −/− littermates until PD7 (Fig. 4b).
Interestingly, a high fraction of BM HSCs from both E18.5 Hem-1 −/− and Hem-1 +/+ fetuses were apoptotic (Fig. 4c), indicating that FL HSCs arriving at the BM were not fit for their new environment. However, in Hem-1 +/+ BM, the apoptotic fraction fell rapidly over the ensuing days, whereas in Hem-1 −/− BM the apoptotic fraction remained at a high level. This implies that there is an intrinsic survival signal the incoming FL HSCs require, but lack when Hem-1 is deleted.
Hem-1 −/− FL HSCs lack c-Abl survival signaling. We therefore sought to identify the FL HSC intrinsic survival signal required for engraftment after HSC transition to the BM. In addition to regulating actin polymerization, which mediates cell migration and adherence, the WAVE complex can also promulgate c-Abl signaling via c-Abl's interaction with Abi-1 [25][26][27][28][29][30] . Deletion of c-Abl results in defects in embryonic hematopoiesis resembling those seen here 33,34 , and constitutive activation of c-Abl via the t(9;22) increases HSC survival and proliferation and generates chronic myeloid leukemia (CML) [35][36][37] . Therefore, we investigated whether the defect in the Hem-1 −/− FL HSC transitioning to BM hematopoiesis was due to a lack of c-Abl survival signaling.
Without Hem-1 as the assembly scaffold, the other components of the WAVE2 complexes are reported to degrade 21,23,24 . As expected, deletion of the WAVE scaffold Hem-1 here also resulted in degradation of the WAVE2 components Abi-1, Abi-2, WAVE2 and Sra-1 in FL Lin − cells (Fig. 5a). As a result of the loss of its interacting partner Abi-1, the expression of c-Abl protein was significantly reduced in Lin − cells from Hem-1 −/− FL Lin − cells ( Fig. 5a and Supplementary Fig. 7). Similar findings were also observed in PD3 BM Lin − cells. In E14.5 Hem-1 −/− FL LSK cells, the loss of c-Abl did not affect the phosphorylation of Crkl, Jak2, Stat3, Stat5, Erk, Akt, and S6, downstream effectors of HSC survival and hematopoiesis (Fig. 5b, c and Supplementary  Fig. 8) [35][36][37] . However, by PD3, Hem-1 −/− BM LSK cells exhibited a significant reduction in the phosphorylation of all of these signaling components ( Fig. 5d and Supplementary Fig. 8). These results suggest that: (1) the c-Abl protein relies on the WAVE2 complex for stability in FL HSPCs [25][26][27][28][29][30] , and (2)  Time after transplantation (months) To determine whether the hematopoietic transition from the FL to the BM is dependent on c-Abl signaling for HSC survival, we incubated E14.5 FL and adult BM Lin − cells from Hem-1 +/+ mice with Imatinib (c-Abl kinase inhibitor) [35][36][37] , and then analyzed FL and BM LSK cell proliferation and apoptosis ( Fig. 6a-d). Imatinib resulted in decreased cultured Hem-1 +/+ FL LSK cell proliferation and increased apoptosis, but not in BM LSK cells. Significantly, the inhibition of c-Abl activity with Imatinib reduced the long-term engraftment of competitively transplanted Hem-1 +/+ FL HSCs (Fig. 6e- (Fig. 6j, k). These findings demonstrate that FL HSCs depend on c-Abl for survival and engraftment in the BM, and that Hem-1 −/− FL HSCs lack this signal.

Discussion
In this study, we constitutively deleted Hem-1 in mice to originally assess two questions: First, was Hem-1 function specific for hematopoietic cells, or were other organs affected by its deletion? Second, were Hem-1 and Hem-2 interchangeable, or did they have specific functions? In the initial characterization of the Hem-1 −/− mice we found that all other organ systems developed normally, and functioned into adulthood. The only defects were in the hematopoietic system. Thus, Hem-1 function was indeed specific for blood cells, and was not required for the normal development and function of any other organ systems. Second, consistent with the previous report by Park et al. 21 that a Hem-1 point mutation affected blood cell development, we found multiple hematopoietic defects in the Hem-1 −/− mice. Thus, Hem-2 cannot replace Hem-1 function in hematopoiesis.
The Hem-1 −/− FL HSC were essentially normal in numbers and proliferative function, but after migration to the marrow osteoblast niche, marrow HSC became rapidly depleted, resulting in a myelofibrosis-like phenotype, with anemia, myeloid metaplasia, and marrow reticulin. The Hem-1 −/− mice lacked other WAVE2 components, consistent with earlier findings in flies, amoeba, and mice, that WAVE complexes are obligate heteropentamers, probably to prevent unregulated actin polymerization, and cell paralysis [22][23][24][25] . Crystal structures of HEM-2 in the WAVE complex structure indicate that it likely attaches to the membrane and serves as a scaffold for the other WAVE components to assemble 16 . Thus, without the scaffold and the protection of the assembled heteropentamer, it is not surprising that the individual WAVE protein components are prone to destruction.
c-Abl-null mice also exhibit neonatal lethality, and if they survive they become runting, splenic, and marrow atrophy, with lymphopenia and increased susceptibility to infections, similar to the Hem-1 −/− mice here 33,34 . Consistent with previous reports, when Abi-1 is degraded in the Hem-1 −/− cells c-Abl fails to signal [25][26][27][28][29] , resulting in decreased phosphorylation of downstream c-Abl targets, such as Crkl, Stat3, and S6, both in FL and marrow stem/progenitor cells 35,36 . Depleting or inhibiting c-Abl inhibit wt FL HSC engraftment capability, but they do not harm engraftment of adult marrow HSC 35,36 . Restoring c-Abl expression to the FL Hem-1 −/− HSC rescues their marrow engraftment. Thus, there is a WAVE2/c-Abl survival signal required at the FL HSC to marrow transition that confers fitness to the HSC for marrow hematopoiesis, similar to thymic T cell selection.
Since the WAVE2 complex is not needed for FL HSC to migrate to the marrow osteoblast niche, there must be alternative pathways that can mediate appropriate actin polymerization for FL HSC migration and adherence. WASP complexes are a potential candidate to mediate an alternative pathway for actin polymerization besides WAVE utilized by the FL HSC to reach the marrow 38 . After activation by rac or cdc42, WASP also activates Arp2/3, and stimulates actin polymerization 39,40 . However, Rac1/2 and cdc42 deletion prevents FL HSC from reaching the marrow for transition of hematopoiesis, consistent with their importance in upstream regulation of actin polymerization, epistatically above both WAVE and WASP 39,40 . This is consistent with our data demonstrating that the cdc42 inhibition blocked FL HSC migration and adhesion.
The data here also have implications for the molecular pathogenesis of CML. The 9;22 chromosomal translocation in CML fuses BCR to c-Abl and results in HSC immortality, but these HSCs are addicted to c-Abl for survival 35,36 . This implies that the CML HSCs more closely resembles the FL HSCs than the marrow HSCs. It is therefore possible that WAVE2 might be important for BCR-Abl signaling, and WAVE2 could be an additional target for therapy in CML. In addition, this finding suggests that pregnant CML patients should not be treated with a c-Abl inhibitor at late gestation. It might inhibit the survival of fetal HSC after migration from the FL to the marrow, and result in a decreased HSC population into adulthood. Exposure to imatinib during pregnancy results in an increased incidence of fetal malformations 41 , but there is little data on blood counts as children of such pregnancies mature 41 . Such children could have a higher risk of marrow aplasia or myelofibrosis in adulthood.
In summary, the scaffold upon which the WAVE2 complex assembles in hematopoietic cells, Hem-1, is required for the engraftment of FL HSC in the marrow during embryonic development. Surprisingly, the engraftment failure of Hem-1 −/− FL HSC is not due to decreased actin polymerization resulting in poor migration or adherence to the marrow niche. Rather, this engraftment requires c-Abl signaling, which is lost when its partner Abi-1 is degraded with the rest of the WAVE complex when Hem-1 is deleted. After the FL HSC has engrafted the marrow, the c-Abl survival signal is no longer needed, indicating that FL HSCs are specifically for and modified by the marrow microenvironment.

Methods
Model generation. For the generation of Hem-1-deleted mice, a mouse Hem-1 genomic clone was obtained from the mouse 129s6/Sv BAC library 42 . A lacZ-neo cassette, in which the neomycin phosphotransferase gene is linked to the lacZ gene placed between the independent ribosomal entry sequences and an SV40 polyadenylation signal, replaced a sequence covering the Hem-1 coding sequence of the first exon. The gene-targeting construct was electroporated into embryonic stem (ES) cells, and the cells were selected with neomycin. Recombinant ES cell clones were then injected into C57BL/6 mouse blastocysts, and Hem-1 +/− mice were Peripheral blood counts and organ histology. Blood was obtained through retroorbital bleeding and transferred to ethylenediaminetetraacetic acid (EDTA)-coated tubes [43][44][45] . Peripheral blood cell numbers were determined using a Vet Abc Hematological analyzer (Scil Animal Care, Gurnee, IL, USA). Tibiae, spleen, and  lung from 5-week-old Hem-1 +/+ or Hem-1 −/− mice were fixed in 4% paraformaldehyde (Fisher Scientific, Pittsburgh, PA, USA) for 24 h. Tibiae were decalcified in 14% EDTA for 7 days. The bones were then embedded in paraffin and 5-μm longitudinal sections were obtained. After de-paraffinization and rehydration, the sections were processed for staining with hematoxylin and eosin for the histologic assessment of organ morphology, silver for reticulin, or myeloperoxidase for myeloid cells.
Isolation of BM and FL hematopoietic cell subsets. The femora and tibiae were harvested from mice immediately after they were euthanized with CO 2 [43][44][45] . BM cells were flushed from the bones into Hank's balanced salt solution (HBSS) containing 2% fetal calf serum (FCS) using a 21-gauge needle and syringe. FLCs were obtained from E14.5 embryos. The cells were centrifuged through Histopaque 1083 (Sigma, St. Louis, MO, USA) to isolate mononuclear cells (MNCs). For the isolation of Lin − cells, MNCs from the BM or FL were incubated with biotinconjugated rat antibodies specific for murine CD3e, CD11b, CD45R/B220, Ter-119, and Gr-1 (CD11b antibody was excluded in FL MNCs). The labeled mature lymphoid and myeloid cells were depleted twice by incubation with goat anti-rat IgG paramagnetic beads (Thermo Fisher Scientific, Waltham, MA, USA) at a bead: cell ratio of 4:1. Cells binding to the paramagnetic beads were removed with a magnetic field. The negatively isolated Lin − cells were washed twice with 2% FCS/ HBSS and re-suspended in complete medium (RPMI-1640 medium supplemented with 10% FCS, 2 mM L-glutamine, 10 µM HEPES buffer, and 100 U/ml penicillin and streptomycin) at 1 × 10 6 cells/ml. LSK cells (Lin − /Sca-1 + /c-kit + cells) were sorted with an Aria II cell sorter (BD Biosciences, San Jose, CA, USA) after the Lin − cells were pre-incubated with anti-CD16/32 antibody to block the Fcγ receptors and then stained with anti-Sca-1-PE and c-Kit-APC antibodies. Dead cells were excluded by gating out the cells stained with propidium iodide (PI).

Analysis of the frequencies of hematopoietic cell subsets by flow cytometry.
MNCs from BM or FL were pre-incubated with biotin-conjugated anti-CD3e, anti-CD45R/B220, anti-Gr-1, anti-CD11b, and anti-Ter-119 antibodies, with anti-CD16/32 antibody to block the Fcγ receptors (CD11b antibody was excluded in FL MNCs) [43][44][45] . The cells were then stained with streptavidin-fluorescein isothiocyanate (FITC) and anti-Sca-1-PE-Cy7, c-Kit-APC-Cy7, CD150-APC, and CD48-Pacific Blue antibodies. The frequencies of HPCs (Lin − /Sca-1 − /c-kit + ), LSK cells (Lin − /Sca-1 + /c-kit + ), and HSCs (CD150 + /CD48 − /LSK) were analyzed using an Aria II cell sorter. For each sample, approximately 5 × 10 5 to 1 × 10 6 cells were acquired and the data were analyzed using the BD FACSDiva 6.0 software (BD Biosciences) and the FlowJo software (FlowJo, Ashland, OR, USA) . The number of the different hematopoietic cell populations in each mouse was calculated by multiplying the total number of MNCs harvested from each mouse or embryo with the frequencies of each population of MNCs. The information for all antibodies used in the staining is provided in Supplementary Table 1. CAFC assay. Stromal cell feeder layers were prepared by seeding 10 3 /well FBMD-1 stromal cells in each well of flat-bottom 96-well plates (Falcon, Lincoln Park, NJ, USA) 44,45 . One week later, BMCs re-suspended in CAFC medium (Iscove's modified Dulbecco's medium (MDM) supplemented with 20% horse serum, 10 −5 M hydrocortisone, 10 −5 M 2-mercaptoethanol, 100 U/ml penicillin, and 100 μg/ml streptomycin) were overlaid on these stromal layers in six threefold serial dilutions. Twenty wells were plated for each dilution to allow limiting dilution analysis of the precursor cells forming hematopoietic clones under the stromal layer. The cultures were fed weekly by changing one-half of the media. The frequencies of CAFCs were determined at weekly intervals. Wells were scored positive if at least one darkphase hematopoietic clone (containing five or more cells) was observed. The frequency of CAFCs was then calculated using Poisson statistics as described previously 3,4 .
Ionizing irradiation. Male CD45.1 mice at 8 to 10 weeks of age were exposed to a lethal dose (9.5 Gy) of total body irradiation in a J.L. Shepherd Model Mark I 137 Cesium γ-irradiator (J.L. Shepherd, Glendale, CA, USA) at a dose rate of 1.080 Gy/min for BMT preconditioning. The mice were irradiated on a rotating platform.
Timed mating and non-competitive SCT. Two Hem-1 +/− females in estrous were put together with one Hem-1 +/− male and checked for vaginal plugs the next morning, designated as 0.5 days post-coitum (dpc) [43][44][45] . At 14.5 dpc (E14.5), the FLs were dissected from the fetuses, and single-cell suspensions were prepared in HBSS. A total of 5 × 10 5 FLCs pooled from three to four Hem-1 +/+ or Hem-1 −/− embryos were transplanted into lethally irradiated CD45.1 recipient mice. The recipient mice were monitored up to 30 days after transplantation, and the number of dead/moribund mice was recorded on a daily basis. The genotypes were determined using a small amount of brain tissues from the fetuses prior to the transplant, prepared, and genotyped as described above.
Competitive repopulation assay. A total of 5 × 10 5 FLCs were pooled from three to four E14.5 Hem-1 +/+ and Hem-1 −/− embryos [43][44][45] . The cells were mixed with 5 × 10 5 competitive BMCs pooled from three CD45.1 mice and then transplanted into lethally irradiated CD45.1 recipients via retro-orbital injection of the venous sinus. Donor cell engraftment was determined at 1, 2, 3, and 4 months after transplantation by immunostaining of the cells in the recipient peripheral blood with FITC-conjugated anti-CD45.2; phycoeythrin (PE)-conjugated anti-B220, CD11b, and Gr-1; and APC-conjugated anti-B220 and CD3e antibodies and analyzed by flow cytometry. Four months after transplantation, the BMCs from the recipient mice were used to analyze the engraftment ability of the donor cells. Donor-derived HPCs and LSK cells from the BM were further evaluated using an Aria II flow cytometer. The information for all antibodies used in the staining is provided in Supplementary Table 1.
Homing assay. Thirty thousand FL LSK cells from E14.5 Hem-1 +/+ and Hem-1 −/− embryos were stained with 5 μM CFSE-mixed isomers (Thermo Fisher Scientific, Waltham, MA, USA) for 10 min according to the manufacturer's instructions 32,46 . The stained cells were re-suspended in 100 μl IMDM medium containing 30,000 cells and retro-orbitally transplanted into lethally irradiated male CD45.1 recipient mice. The mice were euthanized at 16 and 48 h after transplantation. The BMCs were harvested and analyzed by flow cytometry using an LSRII flow cytometer (BD Bioscience) for the presence of CFSE + cells in PInegative cells. The absolute numbers of CFSE + cells in the BM from femora and tibiae were calculated by multiplying the total numbers of BMCs by the percentage of CFSE + cells.
Intra-vital microscopy. E14.5 FL LSK cell micro-localization in the BM, including the three-dimensional distance to the nearest col2-3-EGFP osteoblast and endosteum, was analyzed and measured by intra-vital microscopy using previously described equipment and procedures 6,32,46 . Thirty thousand LSK cells were labeled with DiD (Thermo Fisher Scientific, Waltham, MA, USA) and injected into nonirradiated Col2.3-GFP transgenic mice. A defined region of the calvarium BM cavity (4 × 6 mm 2 ) was scanned using a confocal/two-photon hybrid microscope as previously described, allowing the visualization of DiD + LSK cells and GFP + cells. The cells were imaged at two time points post transplant, 16 and 48 h. To evaluate the distance between the transplanted FL LSK cells and the osteoblastic and endosteal surfaces, we computed values as described previously 6 . The shortest distance from a transplanted cell to a Col2.3-GFP osteoblast or the bone endosteal surface was determined in three dimensions using the Pythagorean theorem. Three recipient mice were analyzed per time point and donor genotype. DiD-labeled cells were identified and distinguished from auto-fluorescent cells using two confocal images at 633 nm (650-760 nm detection) and at 532 nm (560-640 nm detection). A 330 0.9NA water-immersion objective (Lomo) was used for all imaging. For three-dimensional analysis of the BM cavity, Z-stacks were acquired at 1-3 mm steps. A PCI-based image capture board (Snapper24, Active Silicon) was used to acquire up to three channels simultaneously using iPhoton32 software that was developed in-house running under Mac OS X.
Apoptosis assay. BMCs were incubated with anti-CD16/32 at 4°C for 15 min to block the Fc-γ receptors and then stained with antibodies against various cellsurface markers in the dark 44,45 . After Annexin V staining with a kit from BD Pharmingen (San Diego, CA, USA) according to the manufacturer's instructions, the apoptotic cells within different hematopoietic cell populations were analyzed with an Aria II flow cytometer.
Migration assays. In vitro migration in response to SDF-1α was analyzed as previously described 47 . Lin − cells (1 × 10 5 ) from FL of E14.5 Hem-1 +/+ and Hem-1 −/− embryos were pre-incubated with vehicle or 10 µM CASIN (Sigma) for 2 h to inhibit the activity of CDC42. They were plated in the upper well of a 24-well Transwell chamber separated with a filter containing 5.0 μm pore size (Corning, Corning, NY, USA) in IMDM medium with 2% bovine serum albumin (BSA). After a 4-h incubation against an SDF-1 gradient (100 ng/ml) in the lower chamber, all cells that migrated through the filter were collected. These cells were then stained with c-Kit-APC and Sca-1-PE and analyzed using an LSRII flow cytometer (BD Bioscience) for the percentages of LSK cells within PI-negative cells. The numbers of migrating LSK cells were calculated by multiplying the total number of migrating Lin − cells by the percentage of migrating LSK cells.
Adhesion assays. To determine the ability of hematopoietic cells to adhere to a substrate in vitro, 10,000 FL LSK cells from E14.5 Hem-1 +/+ and Hem-1 −/− embryos in Stemspan medium with 10 ng/ml thrombopoietin (TPO) and 10 ng/ml stem cell factor (SCF) were plated in triplicate on 24-well non-tissue culture-treated plates previously coated with fibronectin (CH-296, 20 μg/ml; Clontech Laboratories, Mountain View, CA, USA) 47 . The cells were incubated for 1 h at 37°C, after which the supernatant was removed, and the wells were washed once with phosphate-buffered saline (PBS) to remove non-adherent cells. The number of adherent cells was counted under a light microscope. Neutrophil F-actin adhesion, migration, and polarization assays. Neutrophil isolation was performed as follows: BM cells (2 × 10 7 ) from 5-week-old Hem-1 +/+ and Hem-1 −/− mice were suspended in 3 ml HBSS-EDTA buffer and transferred on the top of pre-prepared Percoll gradient separation solution containing 78% Percoll (1.11 g/ml), 69% Percoll (1.09 g/ml), and 52% Percoll (1.083 g/ml) 20,47 . Cells were separated by centrifugation at 1500 x g for 30 min. The neutrophil layer was collected from the 78/68% Percoll interface and red blood cells were lysed with 0.83% NH 4 Cl solution. Neutrophils were counted and used for testing the capacities of adhesion, migration, and polarization.
For the F-actin polymerization and capping assay, neutrophils (2 × 10 5 ) from 5week-old Hem-1 +/+ and Hem-1 −/− mice were serum-starved in PBS and stimulated with 10 nM fMLP (Sigma) for 120 s. The cells were then fixed with 4% paraformaldehyde for 15 min and permeabilized with 0.1% Triton X-100 (Sigma) for 15 min. After blocking in 2% BSA, the cells were stained with FITC-conjugated phalloidin (Thermofisher Scientific) and DAPI (Sigma), and mounted for fluorescence imaging analysis on a Zeiss fluorescence microscope equipped with a ×40 oil-immersion objective lens, a AxioCam MRm camera, and the AxioVision Rel. 4.8 software (Jena, Germany). The number of cells with F-actin capping was counted in 10 different fields under a ×40 oil-immersion objective lens. Percentages of cells with F-actin capping cells were presented. Images shown are representatives of more than 100 cells examined for each genotype.
For the neutrophil adhesion assay, neutrophils (2 × 10 5 ) isolated from 5-week old Hem-1 +/+ and Hem-1 −/− mice were plated in triplicate on 24-well non-tissue culture-treated plates previously coated with fibronectin (CH-296, 20 μg/mL; Clontech Laboratories). Cells were incubated for 2 h at 37°C, after which the unattached cells were removed by washing with PBS. The numbers of adherent neutrophils were counted under light microscope.
For the neutrophil migration assay, neutrophils (1 × 10 5 ) isolated from 5-week old Hem-1 +/+ and Hem-1 −/− mice were plated in the upper well of a 24-well transwell chamber separated with a filter containing 5.0 μm pore size (Corning, Corning, NY, USA) in DMDM medium with 2% BSA. After 4-h incubation against an SDF-1α gradient (100 ng/ml) at the lower chamber, cells that migrated through the filter were collected and counted under light microscopy. In addition, neutrophil migration was also measured using a 96-well chamber with polycarbonate filters (Neuro Probe, Inc., Gaithersburg, MD, USA) in response to fMLP stimulation. Briefly, 29 μL fMLP (10 nM) was added to the bottom wells of the 96-well plate. Polycarbonate filters with a 3.0 μm pore size were placed between the lower plate and upper plate of the chamber. Fifty microliters of medium containing neutrophils (1 × 10 5 ) were added to the top plate. The chamber was incubated for 1 h at 37°C and 5% CO 2 . Non-migrating cells on the top of the filter were removed by gentle scraping. The number of migrated cells in the lower plate was counted under three different 200× fields and mean values were presented.
Intracellular phospho-flow cytometric analysis. One million cells from E14.5 FLs or PD3 BMs were stained with various antibodies, fixed, and permeabilized using the Fixation/Permeabilization Solution from BD Pharmingen (San Diego, CA, USA) 44,45 . The cells were subsequently stained with anti-p-Crkl, p-JAK2, p-STAT3, p-STAT5, p-p42/44 ERK, p-Akt ser473, and p-S6 antibodies and then analyzed using a flow cytometer. The information for all antibodies used in the staining is provided in Supplementary Table 1.
Imatinib administration. Imatinib (Selleckchem.com, Houston, TX, USA) was resuspended in Dulbecco's phosphate-buffered saline at 10 mmol/L. Lin − cells (3 × 10 5 ) from E14.5 FLs or 5-week-old BMs were incubated with 5 μmol/L imatinib or vehicle. Forty-eight hours later, p-Crkl, apoptosis, and the cell cycle were analyzed. For transplantation, 5 × 10 4 Lin − cells from E14.5 FLs or 5-week-old BM were incubated with 5.0 μmol/L imatinib or vehicle for 1 h at 37°C. The treated cells along with 5 × 10 5 recipient-derived BMCs were transplanted into lethally irradiated recipient mice that were treated with imatinib (100 mg/kg) for 48 h before and after transplantation. At 1, 2, 3, and 4 months after transplantation, donor cell engraftment in the peripheral blood and BM was determined as described above. The 20-µL PCR reaction was prepared as follows: 10 µL of 2× SYBR Green PCR master mix, 0.4 µL of 10 µM of the appropriate forward and reverse primers, 7.6 µL RNase-free water, and 2 µL cDNA template. A negative control (no DNA template) was also performed for each master mix prepared. The qPCR was performed with 10 min at 95°C, followed by 40 cycles of 15 s at 95°C and 1 min at 60°C. A dissociation melting curve was generated using temperatures from 60°C to 95°C. An arbitrary unit was calculated by the comparative CT method according to the CT values of the internal control. Mouse hypoxanthine guanine phosphoribosyl transferase was used as a constitutively expressed internal reference for mouse mRNA. The data are representative of two independent experiments performed in triplicate. The sequences for all the primers used in the qRT-PCR assays are shown in Supplementary Table 2.
Transduction of constitutive c-Abl in hematopoietic cells. The pEGFP-c-Abl plasmid (generous gift from Dr. Zhi-min Yuan, Harvard T.H. Chan School of Public Health) was digested with BamHI and XbaI. The c-Abl fragment was inserted into the BglΙΙ and HpaI sites of the pLent-GFP-vector to generate the pLent-GFP-c-Abl vector 48 . To up-regulate c-Abl expression in hematopoietic cells, FL Lin − cells from E14.5 Hem-1 +/+ and Hem-1 −/− embryos were maintained in Stemspan medium with TPO (10 ng/ml) and SCF (10 ng/ml) and infected twice with viral particles containing the pLent-GFP-c-Abl or pLent-GFP-vector under centrifugation (900 × g) at 32°C for 30 min. Two days later, the percentages of GFP + cells within the infected cells were measured with an LSRII flow cytometer and transplanted into 9.5 Gy lethally irradiated CD45.1 recipient mice along with 5 × 10 5 BMCs from recipient mice. The engraftment ability of the GFP + cells in the peripheral blood and BM was analyzed 4 months post transplantation.
Statistical analysis. The data exhibited normal variation. No data sets were excluded from the analysis. Past experimentation was used to predetermine sample size 44,45 . The experiments were not randomized, except for the in vivo animal studies with mice as described. The investigators were not blinded to allocation during experiments and outcome assessment. The data were analyzed by analysis of variance (ANOVA). Differences among the group means were analyzed by Student-Newman-Keuls multiple comparisons test after one-way or two-way ANOVA. For experiments in which only single experimental and control groups were used, the group differences were examined by an unpaired Student's t test. The frequencies of CAFC were analyzed by using Poisson statistics. The differences in the distribution of the cell cycle phases were determined by χ 2 test. Survival curves were constructed using the Kaplan-Meier method and compared using the log-rank test. Differences were considered significant at p < 0.05. All analyses were performed with GraphPad Prism from the GraphPad software.