ARAP3 Functions in Hematopoietic Stem Cells

ARAP3 is a GTPase-activating protein (GAP) that inactivates Arf6 and RhoA small GTPases. ARAP3 deficiency in mice causes a sprouting angiogenic defect resulting in embryonic lethality by E11. Mice with an ARAP3 R302,303A mutation (Arap3KI/KI) that prevents activation by phosphoinositide-3-kinase (PI3K) have a similar angiogenic phenotype, although some animals survive to adulthood. Here, we report that hematopoietic stem cells (HSCs) from rare adult Arap3KI/KI bone marrow are compromised in their ability to reconstitute recipient mice and to self-renew. To elucidate the potential cell-autonomous and non-cell-autonomous roles of ARAP3 in hematopoiesis, we conditionally deleted Arap3 in hematopoietic cells and in several cell types within the HSC niche. Excision of Arap3 in hematopoietic cells using Vav1-Cre does not alter the ability of ARAP3-deficient progenitor cells to proliferate and differentiate in vitro or ARAP3-deficient HSCs to provide multi-lineage reconstitution and to undergo self-renewal in vivo. Thus, our data suggest that ARAP3 does not play a cell-autonomous role in HSPCs. Deletion of Arap3 in osteoblasts and mesenchymal stromal cells using Prx1-Cre resulted in no discernable phenotypes in hematopoietic development or HSC homeostasis in adult mice. In contrast, deletion of Arap3 using vascular endothelial cadherin (VEC or Cdh5)-driven Cre resulted in embryonic lethality, however HSCs from surviving adult mice were largely normal. Reverse transplantations into VEC-driven Arap3 conditional knockout mice revealed no discernable difference in HSC frequencies or function in comparison to control mice. Taken together, our investigation suggests that despite a critical role for ARAP3 in embryonic vascular development, its loss in endothelial cells minimally impacts HSCs in adult bone marrow.


Introduction
Hematopoietic stem cells (HSCs) are the critical source of all blood cells. Their potential for self-renewal and multi-lineage repopulation sustains the rapid turnover of the blood system throughout life. The first HSC arises from the hemogenic endothelium in the Aorta-Gonad-Mesonephros (AGM) region of the embryo and subsequently colonizes the fetal liver [26]. In the adult mouse, HSCs reside in complex bone marrow (BM) niches that are not mutually exclusive. Extensive research has shown that HSC perivascular and osteoblastic niches are comprised of endothelial cells, mesenchymal stromal cells, osteoblasts, sympathetic nerves and non-myelinating Schwann cells [1][2][3].
ARAP3 is a dual Arf and Rho GTPase-activating protein that was first identified in porcine leukocytes for its ability to bind to phosphatidylinositol (3,4,5)triphosphate (PIP 3 ) [22]. ARAP3 contains two distinct GAP domains that accelerate the rate of GTP hydrolysis to attenuate Arf6 and RhoA signaling [23,24]. Previous in vitro studies found that either exogenous ARAP3 expression in epithelial cells or RNAi-mediated ARAP3 depletion in endothelial cells disrupts F-actin or lamellipodia formation, respectively, resulting in a cell rounding phenotype and failure to spread [25,26]. This implies that ARAP3 controls Arf6 and RhoA in a tightly regulated fashion, and that maintaining precise regulation of ARAP3 activity is crucial to actin organization in the cell. RhoA has been characterized in vivo to regulate migration and chemotaxis of mature hematopoietic cells [27,28], as well as HSPC engraftment, multi-lineage repopulation and cell survival [9,14,15], while the role of Arf6 in hematopoiesis is largely unknown.
In mice, ARAP3 is most highly expressed in the endothelium and bone marrow, and has been found to be critical to vascular development [29,30]. Germline deletion or Tie2-Cre-mediated deletion of Arap3 in mice leads to embryonic lethality by E11 due to defects in sprouting angiogenesis of the endothelium [29]. Since HSCs arise from the hemogenic endothelium during embryonic development around E10.5 [31], and give rise to all subsequent hematopoietic cells in the fetal liver and in the adult BM, this genetic model precludes further studies of ARAP3 function in definitive hematopoiesis and HSC function. Conditional Arap3 deletion in neutrophils has been shown to alter their adhesion-dependent functions [32,33], but the role of ARAP3 in HSPCs has yet to be defined.
ARAP3 is a phosphoinositide 3-OH kinase (PI3K)-and Rap-regulated GAP that is recruited to the plasma membrane in a PIP 3 -dependent fashion. PI3Kdependent activation of ARAP3 involves binding of its two most N-terminal pleckstrin homology (PH) domains to PIP 3, a lipid second messenger generated downstream of PI3K. This drives recruitment of ARAP3 to the plasma membrane to facilitate interaction with its GTPase substrates [34]. PIP 3 binding is ablated when a tandem arginine to alanine mutation is introduced at residues R302,R303 in the first PH domain of ARAP3, preventing ARAP3 activation and recruitment to the plasma membrane [29]. Arap3 R302,303A/R302,303A knock-in mutant mice (here referred to as KI/KI) phenocopy Arap3 null mice, suggesting an essential role for PI3K-dependent activation of ARAP3 [29].
In this study, we first investigate ARAP3 function in adult hematopoiesis using KI/KI 2 mice, since about 2% of KI/KI mice are viable [29,33]. We report that KI/ KI HSCs are compromised in their ability to repopulate and self-renew in serial transplantation assays. To elucidate potential cell-autonomous and non-cellautonomous roles for ARAP3 in HSC function, we selectively delete Arap3 in the hematopoietic compartment, and in endothelial and stromal cells of the HSC niche, respectively, using Cre driven by suitable promoters (Vav1, Prx1, Cdh5/VEcadherin). Using these genetic models, we report that ARAP3 does not play a major role in regulating HSPC functions.

Arap3 R302,303A mutation impairs HSC functions
To study whether ARAP3 function affects adult hematopoiesis and HSC function, we studied KI/KI mice expressing mutant ARAP3 R302,303A. This point mutation interferes with the ability of ARAP3 to bind PIP 3 and its subsequent activation by PI3K [29,33]. Most KI/KI mutant mice die embryonically at E11 [29], but a small subset (,2%) was viable and fertile when the expected birth ratio was 25% (Table 1). By 8-12 weeks of age, these mice were indistinguishable from their littermate controls in gross appearance as well as by phenotypic characterization of their peripheral blood (Fig. 1A). KI/KI BM also showed normal progenitor cell numbers as determined by colony-forming cell (CFC) assays (Fig. 1B), and normal HSC frequencies as determined by flow cytometry using SLAM family surface markers, CD48 2 CD150 + LSK (Lin 2 Sca1 + cKit + ) [35] (Fig. 1C).
To study the function of these mutant HSPCs, purified LSK cells from KI/KI mutant mice or control mice were injected with competitor bone marrow cells into each irradiated recipient mouse. Reconstitution in individual recipient mice was followed every 4 weeks post-transplant. We found that KI/KI LSKs displayed a significantly lower donor chimerism in all lineages of the peripheral blood from recipient mice (Figs. 1D and 1E). Donor-derived cells in the BM, LSK, and SLAM LSK (CD48 2 CD150 + LSK) compartments were significantly lower in mice transplanted with KI/KI cells, in comparison to control cells (Fig. 1F). Four months after the primary transplant, BM cells were harvested and 2610 6 total BM cells were injected into secondary irradiated recipient mice. Tertiary transplants were performed similarly. Peripheral blood and BM HSC reconstitution after each transplant was analyzed by flow cytometry. We found that the defects in reconstitution of KI/KI cells were exacerbated upon serial transplantations, indicating compromised HSC self-renewal (Figs. 1G-1I). Interestingly, a myeloid bias in the multi-lineage reconstitution of serially transplanted recipients arose within KI/KI reconstituted mice (Figs. 1J and 1K), reminiscent of aged HSCs [36][37][38].

ARAP3 is dispensable for steady-state hematopoiesis
Our data from the knock-in mice prompted us to further elucidate the role of ARAP3 in regulating HSC function using ARAP3 conditional knockout (CKO) mice. To study if ARAP3 plays a cell-autonomous role in hematopoietic cells, we crossed Arap3 flox/flox (f/f) mice with mice expressing a Vav1 promoter-driven Cre. Vav1 expression begins around E11.5, and is fully turned on and expressed in greater than 99% of hematopoietic cells by E13.5, thereby excising floxed alleles in most, if not all, fetal liver and adult hematopoietic cells [39].
To measure the deletion efficiency at the DNA level, we cultured 1.5610 4 unfractionated BM cells from either Arap3 flox/flox ;Vav1-Cre tg (f/f;Vav) CKO or f/f control mice in semi-solid methylcellulose cultures. Each progenitor cell gives rise to an individual colony, from which DNA was isolated to measure Arap3 deletion efficiency on a clonal basis. Clonal analysis by polymerase-chain reaction (PCR) genotyping showed near 100% Vav1-Cre-mediated excision of Arap3 (S1A Fig.). We also measured Arap3 deletion efficiency at the transcript level using quantitative real-time PCR (qRT-PCR). f/f;Vav showed a greater than 95% deletion of Arap3 transcripts in the BM when compared to f/f control mice (S1C Fig.). In contrast, the transcript levels of the other ARAP family members, Arap1 and Arap2, remained unchanged in the BM (S1D Fig.). To ensure a more complete deletion of ARAP3 in hematopoietic cells, we generated Arap3 flox/2 ; Vav1-Cre tg (f/2;Vav) mice. These mice showed 100% deletion in all f/2;Vav mice we examined at both DNA and RNA levels ( Figs. 2A and 2B). We found that both

ARAP3 does not play a cell-autonomous role in HSCs
We next examined if ARAP3 affects primitive hematopoietic compartments. We found that f/2;Vav mice had a normal distribution of phenotypic HSPCs within the LSK compartment, as determined by SLAM markers (Fig. 2E). Furthermore, Arap3 deficiency in BM cells did not affect hematopoietic progenitor cell proliferation or differentiation in CFC assays ( Fig. 2F and S2C Fig.). Of note, when purified f/f;Vav LSKs were plated, they exhibited abilities to form various types of colonies with normal frequencies (S2D Fig.) and morphology (not shown). In contrast, Arap3 2/2 neutrophils from the CKO mice exhibited enhanced polyRGD-induced adhesion (S2E Fig.), as previously published [32]. These data together suggest the cell-intrinsic functions of ARAP3 are limited to more differentiated myeloid cells, rather than in the immature HSPC populations. We next assessed whether ARAP3 plays a role in HSC function in vivo using competitive bone marrow transplantation (BMT) assays. LSK cells sorted from f/ f;Vav mice or f/f littermate controls were transplanted into lethally-irradiated recipient mice and peripheral blood reconstitution was evaluated. There were no significant differences in multi-lineage reconstitution (S2H Fig.) or donor chimerism, either in the total blood cell population (S2F Fig.) or in myeloid/Tcell/B-cell lineages after primary BMT (S2G Fig.). Secondary transplants also showed normal multi-lineage reconstitution (S2I-S2K Figs.). Furthermore, serial transplantation of f/2;Vav LSKs showed no discernable differences in donor contribution (Figs. 2G and 2I) or multi-lineage reconstitution (Figs. 2H and 2J) when compared to littermate Arap3 flox/2 (f/2) or f/f control donors. Together, our data firmly established that ARAP3 does not cell-autonomously impact HSC homeostasis or function.  We next investigated whether ARAP3 plays a role in the HSC niche, acting noncell-autonomously to regulate HSCs. To study this, we deleted Arap3 using Prx1-Cre, shown to induce excision in nearly all osteoblasts and 95% of perivascular stromal cells but not in endothelial cells [2,40]. Arap3 flox/2 ;Prx1-Cre tg (f/2;Prx1) mice were born alive in expected Mendelian ratios. Hematopoietic development of f/2;Prx1 mice appeared normal as determined by CBC and lineage distribution in hematopoietic tissues, as well as by progenitor numbers and function in CFC assays (data not shown). f/2;Prx1 mice also exhibited normal frequencies of phenotypic HSCs and progenitors as characterized by flow cytometric analysis (Fig. 3A). Using competitive BMT assays, we found that f/2;Prx1 LSKs repopulated comparably to control f/2 LSK cells (Fig. 3B). These data indicate that ARAP3 in Prx1-expressing niche cells is not a significant regulator of steadystate hematopoiesis or HSC homeostasis.

ARAP3 expression in endothelial cells is important for embryonic development but not adult HSC functions
Arap3 deletion in endothelial cells using Tie2-Cre resulted in embryonic lethality due to a cell-autonomous angiogenesis defect, though only with more robust excision on the f/2, but not f/f, background [29]. Since Tie2-Cre also shows some expression in stromal cells [41,42], we utilized the vascular endothelial cadherin (VEC or Cdh5) promoter-driven Cre (VEC-Cre) that is found to be more endothelial cell-specific than Tie2-Cre [43,44]. Arap3 flox/flox ;VEC-Cre tg (f/f;VEC) mice were born at expected Mendelian ratios (Table 2).
To obtain a more complete deletion of Arap3 in endothelial cells, we generated Arap3 flox/2 ;VEC-Cre tg (f/2;VEC) mice. These CKO mice were born at a significantly reduced ratio ( Table 2). The surviving CKO mice appeared grossly normal, and showed an approximate 88% excision efficiency (Fig. 4A), ranging from 80% to 96%, compared to an average 80% excision rate in f/f;VEC mice (S1B Fig.). By qRT-PCR, an approximate 9% Arap3 transcripts remained in f/2;VEC BM ( Fig. 4B), in contrast to 15% in f/f;VEC BM (S1C Fig.). In agreement with previously published data, this indicates that ARAP3 function in endothelial cells is essential for embryonic development.
Both f/f;VEC and surviving f/2;VEC adult mice had similar peripheral blood composition to their control littermates (S3A Fig. and Fig. 4C) and normal lineage distribution in hematopoietic tissues (S3B Fig. and Fig. 4D). These mice also exhibited a normal distribution of immature HSPC populations, with expected numbers of CD48 2 CD150 + LSK cells (Fig. 4E). Hematopoietic progenitor numbers were normal as determined by CFC assays (Fig. 4F and S3C Fig.). We next examined HSC functions in vivo using competitive BMT assays.    To assess whether ARAP3 expression in BM endothelial cells is important to support HSC functions, we performed reverse transplantation of wild-type bone marrow cells into f/f and f/2 control (SJL:CTL) or f/f;VEC and f/2;VEC CKO (SJL:VEC-CKO) recipient mice. HSC homeostasis or engraftment was not altered, and multi-lineage reconstitution of the hematopoietic compartment was unaffected by Arap3 deletion in the host endothelial HSC niche ( Fig. 4K and data not shown). Taken together, our investigation suggests that in spite of a critical role for ARAP3 during embryonic vascular development [29,30], loss of ARAP3 in endothelial cells minimally impacts HSCs in the adult BM.

Discussion
In the present study, we utilize KI/KI mice and generate three Arap3 CKO mouse models to study ARAP3 functions in hematopoiesis of the adult mouse. We show that KI/KI mice exhibit defective HSC functions upon transplantation, indicating that PI3K-mediated ARAP3 function is important for HSCs. To elucidate the potential cell-autonomous and non-cell-autonomous roles of ARAP3 in HSCs, we conditionally deleted Arap3 in hematopoietic cells and in several cell types in the HSC niche. Arap3 null HSCs from f/2;Vav mice were functionally competent to repopulate the HSC pool and self-renew upon serial transplantation, revealing ARAP3 is not required in regulating HSC homeostasis or function in a cellautonomous manner. Furthermore, ablation of ARAP3 in perivascular stromal cells, osteoblastic cells, and endothelial cells of the HSC niche did not alter HSC function or maintenance in f/2;Prx1 and f/2;VEC mice. Reverse transplantation experiments strongly suggest that ARAP3 expression in the BM endothelial niche is not required to support HSC functions; however, future investigation is needed to assess the potential role of ARAP3 in BM mesenchymal and osteoblastic niches.
Our study demonstrates that f/2;VEC mice displayed partial embryonic lethality, indicating a critical role for ARAP3 in endothelial cells during embryonic development but not in the adult BM niche for HSCs.
ARAP3 was first purified from porcine leukocytes for its PIP 3 binding ability and is highly expressed in hematopoietic tissues, however our studies show that ARAP3 is not a critical cell-intrinsic regulator of hematopoiesis. ARAP3 does not play a cell-autonomous role in regulating HSC homeostasis or function. However, our observation does not rule out a cell-autonomous role for ARAP3 under stress conditions. This phenomenon has been seen in genetic studies of other proteins, such as SIRT1 [45,46]. It is possible that a physiological challenge to the mice is necessary to elicit a specific function for ARAP3 in hematopoiesis. In our model, Arap3 is excised in hematopoietic cells during early development using Vav1-Cre, and compensatory mechanisms may account for normal hematopoiesis in adult mice. It would be interesting to study whether acute loss of ARAP3 in hematopoietic cells of the adult mouse, such as with an inducible Cre system [47], would reveal a role for ARAP3 in hematopoiesis. Additionally, ARAP3 is part of a dual-GAP family that also includes ARAP1 and ARAP2. Although each family member targets different G-protein substrates and has different active functional domains [48][49][50], these three proteins may have overlapping and redundant roles in hematopoiesis, and may work in conjunction to regulate HSPC function. Thus, genetic studies of mice deficient in Arap1 or Arap2, or with combination deletions of multiple ARAP proteins would clarify whether there is a significant role for the ARAP family of proteins in hematopoiesis.
We find that deletion of Arap3 in stromal and osteoblastic cells using Prx1-Cre does not affect steady-state hematopoiesis or HSC homeostasis. Furthermore, deletion of Arap3 in endothelial cells using VEC-Cre does not impact its ability to support wild-type HSCs. However, high efficiency of Arap3 deletion in f/2;VEC mice results in embryonic lethality, suggesting ARAP3 plays an important role in developing endothelial cells, in agreement with previously published data [29,30]. Since ARAP3 deficiency in embryonic endothelial cells disrupts angiogenic sprouting and vascular structure, this developmental-related dysregulation may indirectly affect HSC emergence, development, and maintenance in the adult mouse. However, our studies do not exclude the possibility that compensatory mechanisms may be upregulated in surviving f/2;VEC mice sometime between HSC emergence and adulthood. One approach to answer this question would be to induce the excision of Arap3 in endothelial cells when ARAP3 is no longer required for vasculature development. Future investigations are warranted to examine the emergence of definitive hematopoietic progenitors and HSCs from the endothelium in the AGM region of Arap3 null embryos and further our understanding of the role of ARAP3 in endothelial cells.
Our results demonstrate that the ARAP3 R302,303A mutation (KI/KI) that disrupts ARAP3 recruitment to the plasma membrane or activation by PI3K markedly impairs HSC function, while loss of ARAP3 does not. This could be due to compensatory mechanisms or changes in gene expression and signaling pathways of rare surviving KI/KI mice. Another possibility for compromised HSCs in KI/KI mice might be that the ARAP3 R302,303A mutant acts in a dominant negative manner to prevent translocation of other interacting players [51][52][53][54][55] to the plasma membrane, which may affect HSC function. For example, ARAP3 can bind the phosphatase SHIP2, a negative regulator of PI3K signaling [53,54,56], as well as CIN85 and Odin, both shown to be involved in receptor endocytosis and motility [51,55,[57][58][59]. Although ARAP3 has not been shown to act as a scaffold for the plasma membrane translocation of any interacting proteins, and the in vivo relevance of these interactions remains to be tested, the fact that we see a cell-intrinsic defect with KI/KI HSCs but not with f/2;Vav HSCs in the BMT assays speaks in favor of this possibility. Lastly, it is possible that ARAP3 plays an important role in a cell type other than hematopoietic, endothelial, stromal, or osteoblastic cells to impact HSC function. One way of determining whether the phenotype of KI/KI HSCs is due to a cell-autonomous effect of the R302,303A mutant in HSCs would be to generate a Vav1-Cre-driven conditional KI/KI mouse to genetically study this question.
RhoA has been characterized in vivo to regulate migration and chemotaxis of mature hematopoietic cells [27,28], as well as HSPC engraftment, multi-lineage repopulation and cell survival [9,14,15]. ARAP3 is not the only GTPaseactivating protein that targets RhoA, and our data suggest it is not a major regulator of RhoA activation in HSPCs. Other GAPs, such as p190B RhoGAP, play important roles in mediating HSPC function through its inactivating activity on RhoA [20,21]. Due to the large number of regulators for RhoA, it is likely that each acts in its own individual temporal-and spatial-specific manner. The possibility of redundancies between the various GAPs also exists [60], such that changing the dynamic by ablating one GAP is not enough to alter the process of normal hematopoiesis. It would be interesting to investigate whether deletion of multiple GAPs in hematopoietic cells would result in greater deficiencies than migration or engraftment alone.
As a dual GTPase-activating protein, ARAP3 targets Arf6 as well as RhoA. Arf6 has mostly been studied in non-hematopoietic cells with regard to its role in membrane trafficking and the cell actin cytoskeleton [61][62][63]. Like RhoA, it is actively involved in cell migration, adhesion, proliferation and cytokinesis [64][65][66]. However, the potential role of Arf6 in HSPCs and hematopoiesis has not been well established. One study showed that the decrease of active Arf6-GTP in platelets is critical to the activation of Rho GTPases that is necessary for cytoskeletal rearrangements preceding full platelet function [67]. It is important to further investigate and understand the role of Arf6 in hematopoiesis, particularly in HSCs.
ARAP3 has been implicated in the regulation and progression of several human diseases, including defense against bacterial infection, diabetes and gastric carcinoma, by capitalizing on the ability of ARAP3 to manipulate vesicle internalization and cell invasion [68][69][70]. Dysregulation of Rho family GTPases and their regulators have also been correlated with human blood disorders and tumorigenesis [71][72][73][74][75]. While aberrant expression of ARAP3 has not yet been found in blood disorders, its ability to regulate the actin cytoskeleton makes it a potential target for the dysregulation of homeostatic cell functions. Thus, continued study of ARAP3 in normal and abnormal hematopoiesis will be important to elucidate a more comprehensive understanding of its role in the blood system.

Generation of Arap3 transgenic mice
Arap3 flox/flox and Arap3 KI/KI mice were generated as described previously [29]. Vav1-Cre mice were originally generated by Dr. Thomas Graf [39] and backcrossed to C57Bl/6J background for 8 generations. VEC-Cre mice were kindly provided by Dr. Nancy Speck [44] and backcrossed to C57Bl/6J background for 8 generations. These two strains of mice were crossed to generate Arap3 flox/+ ;Cre tg mice, which were then crossed to Arap3 flox/flox mice to generate Arap3 flox/flox ;Cre tg conditional knockout mice. Initial studies presented in S1-S3 Figs of f/f;Vav and f/ f;VEC mice were done in mixed Bl6/129 background, while studies in main figures were later performed on a pure C57Bl/6J background following backcrossing for 8 generations. Arap3 flox/+ mice on the pure C57Bl/6J background were crossed with CMV-Cre mice on the C57Bl/6J background (The Jackson Laboratory) to generate Arap3 +/2 mice. These mice were used to generate Arap3 flox/2 ;Vav1-Cre tg mice, Arap3 flox/2 ;VEC-Cre tg and Arap3 flox/2 ;Prx1-Cre tg conditional knockout mice (all on a pure Bl6 background) that will ensure a more complete deletion efficiency. Prx1-Cre mice on a C57Bl/6J background were purchased from The Jackson Laboratory.
The animal studies were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Institutional Animal Care and Use Committee of the Children's Hospital of Philadelphia.
For quantitative real-time PCR, total RNA was isolated from FACS-sorted bone marrow cells, CFC assays or hematopoietic tissues using Trizol Reagent (Invitrogen Life Technologies) followed by isolation with the RNeasy Mini kit (Qiagen). cDNAs were produced using BioRad iScript kit and qRT-PCR reactions were performed on an Applied Biosystems 7900HT real-time PCR system using Sybr-Green detection with the following primers:

Complete blood counts
Peripheral blood was collected by retro-orbital bleeding into capillary blood collection tubes with EDTA (BD). CBC analysis was performed using the mouse setting on a HemaVet 950 machine (Drew Scientific).

Cell sorting and flow cytometry
Cells from either peripheral blood or hematopoietic tissues were lysed of red blood cells, then stained with surface markers on ice and washed in PBS with 2% bovine calf serum. Surface markers used to identify cell populations are: CD3e-PE for T-cells, CD19-APC for B-cells, Gr1-PE and Mac1-APC for myeloid cells (eBiosciences), and propidium iodide for viability. This method was also used for peripheral blood analysis of transplanted mice with the addition of CD45.1-PE-Cy7 and CD45.2-FITC antibodies. Flow cytometry was performed on a FACS Canto analyzer (BD).
For cell sorting, bone marrow cells were first depleted of lineage-positive cells using biotinylated lineage cocktail and streptavidin-coupled Dynabeads (Invitrogen) per manufacturer's protocol. Lineage-negative (Lin 2 ) cells were then stained for LSKs as described above and sorted on a FACS Aria Cell Sorter (BD). All analysis of FACS data was performed using FlowJo software (TreeStar).

Colony-forming cell (CFC) assays
Unfractionated bone marrow cells or sorted LSK cells were plated at a concentration of 15,000 or 200 cells per plate, respectively, in duplicate using semisolid methylcellulose (Methocult M3434, StemCell Technologies) containing SCF, IL-3, IL-6 and erythropoietin. Cells were incubated at 37˚C, 5% CO 2 with high humidity, and colonies were enumerated after 10-12 days in culture.

Competitive bone marrow transplantation (BMT), serial BMTs, and reverse BMTs
Bone marrow cells (CD45.2 + ) were harvested and sorted for LSK surface markers as described above. 500-1000 LSK cells were mixed with 350,000 competitor bone marrow cells (CD45.1 + ) and injected into each lethally-irradiated (a split dose of 10Gy, 137 Cs source) [76] F1 recipient mouse (CD45.1 + CD45.2 + ). Donor-derived reconstitution in the periphery was measured by flow cytometry every 4 weeks post-transplant. At 12-16 weeks, recipient mice were sacrificed and analyzed for donor HSPCs, as described above with the addition of CD45.1-PE-Cy7 and CD45.2-FITC antibodies. For secondary transplantation, 2610 6 unfractionated bone marrow cells from pooled primary recipient mice were transplanted into lethally irradiated F1 recipients. Reconstitution was again measured every 4 weeks post-transplant and data of secondary transplant endpoints are shown where mentioned. Tertiary transplants were performed similarly. In reverse BMTs, 1610 6 total bone marrow cells from CD45.1 + wild-type donor SJL mice were transplanted into lethally irradiated f/f and f/2 control or f/f;VEC and f/2;VEC (CD45.2 + ) CKO recipients. Reconstitution and the percentage of CD45.1 + LSK and SLAM LSK cells were measured 8 weeks after transplantation.