Rasa3 Controls Megakaryocyte Rap1 Activation, Integrin Signaling and Differentiation into Proplatelet

Rasa3 is a GTPase activating protein of the GAP1 family which targets Ras and Rap1. Ubiquitous Rasa3 catalytic inactivation in mouse results in early embryonic lethality. Here, we show that Rasa3 catalytic inactivation in mouse hematopoietic cells results in a lethal syndrome characterized by severe defects during megakaryopoiesis, thrombocytopenia and a predisposition to develop preleukemia. The main objective of this study was to define the cellular and the molecular mechanisms of terminal megakaryopoiesis alterations. We found that Rasa3 catalytic inactivation altered megakaryocyte development, adherence, migration, actin cytoskeleton organization and differentiation into proplatelet forming megakaryocytes. These megakaryocyte alterations were associated with an increased active Rap1 level and a constitutive integrin activation. Thus, these mice presented a severe thrombocytopenia, bleeding and anemia associated with an increased percentage of megakaryocytes in the bone marrow, bone marrow fibrosis, extramedular hematopoiesis, splenomegaly and premature death. Altogether, our results indicate that Rasa3 catalytic activity controls Rap1 activation and integrin signaling during megakaryocyte differentiation in mouse.


Introduction
Ras families GTPase-activating proteins (GAP), like Ras GAPs, Rho GAPs and Arf GAPs, are tumor suppressors as the loss of their GAP activity allows uncontrolled Ras, Rho and Arf activities and promotes cancer. Rasa3 (or GAP1 IP4BP , R-Ras GAP) is a member of the Ras GAP1 subfamily with Rasa2 (or GAP1 m ), Rasa4 (or Capri) and Rasal (or Rasal1) [1][2][3][4][5]. This Ras GAP subfamily is known to function as dual GAP for Ras an Rap-GTPases [6,7]. Rasa3 protein structure is characterized by a conserved basic domain structure comprising two N-terminal tandem C2 domains, a central GAP domain and a C-terminal pleckstrin homology (PH) domain that is associated with a Bruton's tyrosine kinase (Btk) motif [8]. Binding of the latter domain to phosphoinositides determines Rasa3 targeting to the cytosolic leaflet of the plasma membrane where it inactivates Ras and Rap1 [9][10][11]. Down-regulation of Rasal and Rasa4 induces cellular transformation in vitro [12,13], and Rasal is downregulated in multiple human tumors by epigenetic silencing [14]. Rasa4 inactivation in mouse leads to impaired macrophages Fcc receptor-mediated phagocytosis and oxidative burst, as well as to increased bacterial infection [15]. No clear definition of Rasa2 function in vivo is currently available. Mutant mice expressing a catalytically-inactive Rasa3 protein have been reported to die at mid embryonic life [16]. Indeed, removal of exons 11 and 12 of the mouse Rasa3 gene, 2 exons which are essential for the Ras GAP activity, leads to the expression of a 88 amino acidstruncated but catalytically inactive Rasa3 protein [16]. Phenotypically, Rasa3 mutant embryos present massive subcutaneous and intraparenchymal hemorrhages probably consecutive to abnormal adherens junctions between capillary endothelial cells [16]. Multiple roles for Ras and Rap1, the Rasa3 targets, have been defined in hematopoietic cells: these proteins control cellular proliferation, differentiation, migration and adhesion. In particular, Rap1 has been implicated in the maturation of megakaryocytes and the pathogenesis of chronic myelogenous leukemia [17]. Here, we found that catalytic inactivation of Rasa3 specifically in the hematopoietic system results in a lethal syndrome characterized by major alterations during megakaryopoiesis. These alterations were associated with increased active Rap1 level and constitutive integrin activation in megakaryocytes, a phenotype quite different clinically, biologically and mechanistically from that of recently published mice with a spontaneous missense mutation between the two N-terminal tandem C2 domains of Rasa3 [18].

Results
The SCID-Rasa3 model In order to study the specific effects of a catalytically-inactive Rasa3 mutant protein on the hematopoietic system and to circumvent the early embryonic lethality reported in Rasa3 2/2 mice, we used irradiated Severe Combined Immune Deficient (SCID) mice reconstituted with E12.5 liver cells derived from Rasa3 +/+ , Rasa3 +/2 or Rasa3 2/2 embryos. SCID mice were first analyzed 6 weeks after irradiation/reconstitution: all Rasa3 genotypes were able to reconstitute the lymphoid compartment in irradiated SCID mice since no significant difference was detected between SCID-Rasa3 +/+ , SCID-Rasa3 +/2 and SCID-Rasa3 2/2 mice in total numbers of splenic T and B cells (Table  S1). No significant difference was observed in red blood cell, blood platelet and bone marrow megakaryocyte counts as well as spleen weight between SCID-Rasa3 +/+ and SCID-Rasa3 2/2 mice at this stage (Table S1).
Decreased survival, hemorrhages and splenomegaly in SCID-Rasa3 2/2 mice More than 80% of SCID mice reconstituted with Rasa3 2/2 cells died within 14 months after reconstitution while, at the same time, about 95% of SCID-Rasa3 +/+ and SCID-Rasa3 +/2 mice were still alive (Fig. 1a). Pathological analysis revealed that 85% of SCID-Rasa3 2/2 mice presented with thoracic and/or peritoneal hemorrhages (data not shown) and that more than 80% had a splenomegaly (Fig. 1b). Our results below present the analysis of a total of 24 moribund SCID-Rasa3 2/2 mice. Among these 24 mice, 20 had a megakaryocytic dysplasia associated with a severe thrombocytopenia, and the remaining 4 developed a preleukemia. The main objective of this study was to define the cellular and the molecular mechanisms of the megakaryocytic dysplasia.
Collectively, our results indicate that the loss of Rasa3 catalytic activity in 20/24 SCID-Rasa3 2/2 mice leads to megakaryocyte alterations, to thrombocytopenia, hemorrages and a regenerative anemia.
Altered megakaryocyte adhesion, motility and capacity to differentiate in proplatelet forming megakaryocytes in SCID-Rasa3 2/2 mice Bone marrow was isolated from SCID-Rasa3 mice 2 months after irradiation/reconstitution and cultured under a confocal microscope. Despite a ,2-fold increased percentage of megakaryocytes in the SCID-Rasa3 2/2 bone marrow, there was a trend for a decreased number of megakaryocytes released from

Author Summary
Megakaryocytes are the bone marrow cellular precursors of circulating blood platelets and give rise to nascent platelets by forming branching filaments called proplatelets. Terminal differentiation of round megakaryocytes into branched proplatelet forming megakaryocytes is a complex cytoskeletal-driven process which is affected in rare human familial thrombocytopenias. Interactions of megakaryocytes with extracellular matrix proteins are essential in this process since constitutive megakaryocyte integrin activity caused by specific mutations in ITGA2B or ITGB3 genes encoding for extracellular matrix protein receptors may result in abnormal adherent megakaryocytes, defect in proplatelet formation and thrombocytopenia. Here, we show that Rasa3, a GTPase activating protein of the GAP1 family, controls Rap1 activation and integrin signaling during megakaryocyte differentiation. We found that Rasa3 catalytic inactivation in mice altered megakaryocyte development, adherence, migration, actin cytoskeleton organization and differentiation into proplatelet. Thus, these mice presented a severe thrombocytopenia, bleeding and anemia.
Collectively, these results indicate that Rasa3 2/2 FLC abnormally develop into mature megakaryocytes, and that Rasa3 2/2 megakaryocytes derived from FLC culture have an altered actin cytoskeleton organization associated with an abnormal adherent phenotype, a reduced motility and an absence of normal terminal differentiation in proplatelets. Interestingly, this Rasa3 2/2 megakaryocyte phenotype (i.e. defect in proplatelet formation, dotted actin cytoskeletal pattern with reduced stress fibers and abnormal adherent megakaryocytes) resembles that of rare thrombocytopenic patients with a constitutive aIIbb3 integrin activity caused by specific mutations in ITGA2B or ITGB3 genes [20][21][22].
Integrin activation triggers megakaryocyte adhesion to immobilized integrin ligands like collagen-I or fibrinogen and an outside-in signaling, resulting in the reorganization of the actin filaments and the modification of the cell shape [23]. Megakaryocytes from FLC-Rasa3 +/+ cultured on day 3 adhered to collagen-I-and fibrinogen-coated plates, but nearly not to Poly-D-Lysine-coated plates, as expected (Fig. 5b). Adherence to immobilized collagen-I and fibrinogen resulted in cell spreading reaching diameters over 50 mm in a limited number of Rasa3 +/+ megakaryocytes, as described (Fig. 5c, red dots, and Fig. 5d) (24).
Adherence to Poly-D-Lysine-, collagen-I-and fibrinogen-coated plates was significantly higher in Rasa3 2/2 than in Rasa3 +/+ megakaryocytes (Fig. 5b). The percentage of megakaryocytes with a diameter over 50 mm was significantly increased in the Rasa3 2/2 culture, as compared with the Rasa3 +/+ culture (Fig. 5d). Outside-in integrin activation triggers the binding of the cytoskeletal protein talin to membrane integrins [24]. In association with their abnormal adhesion properties, Poly-D-Lysine adherent Rasa3 2/2 megakaryocyte recruited more talin to their membrane, as compared with Rasa3 +/+ megakaryocytes ( Fig. 5e) Collectively, our results indicate that Rasa3 2/2 megakaryocytes have a constitutively activated inside-out aIIbb3 integrin signaling associated with major alterations in outside-in integrin signaling leading to cell adherence and spreading independently of integrin ligands.
Altogether, these results indicate that the absence of Rasa3 increases Rap1 activation, and that Rap1 rather than Ras is probably responsible for the abnormal Rasa3 2/2 adherent megakaryocyte phenotype. Thus, increased Rap1 activation in the absence of Rasa3 leads to constitutive activation of integrins and increased outside-in signaling.
Altogether, these results indicate that Rasa3 +/2 platelets present adhesion and activation defects in resting conditions, suggesting  The percentage of Rasa3 2/2 adherent megakaryocytes with a diameter over 50 mm was significantly increased, as compared with Rasa3 +/+ megakaryocytes (mean 6 SEM of 3 independent experiments). E. Rasa3 +/+ and Rasa3 2/2 PDL-adherent megakaryocytes were stained with phalloidin-TRICT (actin, red), CD41-APC (magenta) and Talin-FITC (green). Confocal images were obtained from that a similar pathological mechanism is present both in megakaryocytes and platelets.
Preleukemia in 4/24 SCID-Rasa3 2/2 mice In the ,20% (4/24) remaining SCID-Rasa3 2/2 mice, a very different phenotype was observed: a massive and homogeneous cellular infiltration was detected in the bone marrow and spleen, suggestive of a leukemia (Fig. 7a and data not shown). Adult naïve SCID mice intraperitoneally injected with 10 7 splenocytes isolated from these SCID-Rasa3 2/2 mice did not develop a similar proliferative disorder within 4 months after injection, suggesting the presence of a preleukemia rather than a leukemia in these 4 SCID-Rasa3 2/2 mice (data not shown). No fibrosis was detected in the bone marrow of these 4 SCID-Rasa3 2/2 mice. Flow cytometry analysis with a panel of antibodies revealed that cells massively infiltrating the bone marrow and the spleen were positive for CD117/c-Kit, CD38 and Sca-1, and negative for all other cell surface markers tested, including B220, CD3, MAR-1, Gr1, Mac1, Ter119, CD71, CD4, CD34 and F4.80 (Fig. 7b, 7c and data not shown). As expected, the percentage of B220 + , CD3 + , Gr1 int Mac1 + , Ter119 + CD71 + , CD41 + and F4.80 + cells was significantly decreased in the bone marrow and the spleen of these 4 mice (data not shown). These 4 mice had a reduced survival (survival range: 6-11 months after SCID mice irradiation/ reconstitution) and a splenomegaly (spleen weight range: 0.185-1.062 g).
Collectively, these results indicate that about 20% of SCID-Rasa3 2/2 mice develops a preleukemia with a massive infiltration of bone marrow and spleen with CD117 + Sca-1 + CD38 + cells, probably leading to bone marrow failure and premature death. They also suggest that Rasa3 is a potential tumor suppressor gene, acting may be on Ras, as proposed by Blanc et al. [18]. However, the level of active, GTP-bound Ras was similar in CD117 + /c-Kit + hematopoietic stem cells derived from Rasa3 +/+ and Rasa3 2/2 FLC cultures (Fig. 7d).

Discussion
Using a Rasa3 catalytic mutant in FLC and irradiated/ reconstituted SCID models, we show here that Rasa3 catalytic activity controls megakaryocyte development and differentiation into proplatelet forming megakaryocytes. In the irradiated/ reconstituted SCID model, these megakaryocyte alterations are associated with thrombocytopenia, bleeding, regenerative anemia and decreased survival, as well as with bone marrow fibrosis, extramedular hematopoiesis and splenomegaly.
An increased percentage of mature megakaryocytes with an abnormal morphology was detected in bone marrow cells from irradiated/reconstituted SCID mice when Rasa3 catalytic activity was inactivated. This increased percentage was associated with a slightly decreased percentage of progenitors with megakaryocyte potential, suggestive of a megakaryopoisis alteration. An obvious megakaryopoiesis alteration was also detected in Rasa3 2/2 FLC culture, where the number of CFU for immature megakaryocyte was significantly decreased and associated with the presence of numerous mature megakaryocytes. Ploidy in these Rasa3 2/2 abnormal megakaryocytes was also slightly altered. On the contrary to active Ras level, level of active GTP-bound Rap1 was significantly increased in Rasa3 2/2 megakaryocytes. Interestingly, the small GTPase Rap1 is both a Rasa3 substrate and a well known regulator of integrin signaling in megakaryocytes and platelets [2,[25][26][27][28][29]. Both inside-out and outside-in integrin signaling are controlled by Rap1, including aIIbb3 signaling. Thus, the increased active GTP-bound Rap1 level detected in Rasa3 2/2 megakaryocytes represents a plausible molecular mechanism linking Rasa3 to integrin signaling and the altered megakaryocyte development and differentiation. Indeed, altered inside-out and outside-in integrin signaling in Rasa3 2/2 mega- Figure 7. SCID-Rasa3 2/2 mice develop a CD117 + CD38 + Sca-1 + cell preleukemia. A. Representative images of a hematoxylin/eosin-stained section of a femur from a SCID-Rasa3 +/+ mouse and one of the four SCID-Rasa3 2/2 mice with a homogeneous cellular infiltration of the bone marrow and the spleen (upper panels: magnification: 620; lower panels: magnification: 6100). B. CD117 + splenocyte percentages in SCID-Rasa3 +/+ (n = 10) and in the four SCID-Rasa3 2/2 mice with a preleukemia. Statistics (unpaired t test): ***: P,0.001. C. Representative flow cytometry analysis of bone marrow cells from a SCID-Rasa3 +/+ mouse (left histogram) and one of the four SCID-Rasa3 2/2 (right histogram) mice with a preleukemia, using a CD117 antibody. The histograms show the CD117 fluorescence intensity and the relative number of cells (events). D. Fetal liver cells from Rasa3 +/+ and Rasa3 2/2 E12.5 embryos were stained with a CD117 antibody and analyzed for active GTP-bound Ras level by immunofluorescence using GST-Raf1-RBD and a FITC-conjugated mAb against GST. The graph represents the intensity of active GTP-bound Ras staining, expressed in arbitrary units (A. U.), in Rasa3 +/+ and Rasa3 2/2 CD117 + HSC. Mean 6 SEM are presented. doi:10.1371/journal.pgen.1004420.g007 karyocytes probably results in the adherence and motility defects that we observed in this study. These defects may secondarily lead to an abnormal distribution of megakaryocytes between osteoblastic and vascular niches and to altered megakaryopoiesis. Constitutive activation of integrin signaling in Rasa3 2/2 megakaryocytes is associated with alteration in actin cytoskeleton organization, including a lack of stress fiber assembly, in talin recruitment to the plasma membrane and in cell adherence and spreading that occurred independently of integrin ligands. These alterations probably prevent terminal differentiation of Rasa3 2/2 megakaryocytes since megakaryocyte aIIbb3 and b1 integrins are known to control proplatelet production and platelet release [30][31][32]. Moreover, stress fiber assembly is known to require optimal b1 integrin activation, a process also regulated by aIIbb3 integrin [33,34]. In future work, it will be important to analyze Rasa3 2/2 platelets, since integrins play also important roles in these cells. Our preliminary studies indicate that unstimulated Rasa3 +/2 platelets have altered adhesion to BSA-coated plates and activation, as compared with Rasa3 +/+ platelets, thus mimicking defect of Rasa3 2/2 megakaryocytes.
Interestingly, constitutive aIIbb3 integrin activation in human megakaryocytes mimics most of the Rasa3 2/2 megakaryocyte phenotypical traits. Indeed, in rare thrombocytopenic patients with activating mutations in ITGA2B or ITGB3 genes, megakaryocyte spreading on fibrinogen is abnormal, with 50% of spread cells showing a disordered actin distribution where focal adhesion points are more evident than stress fibers [21]. Sustained and substrate-independent activation of the outside-in aIIbb3 signaling was detected in megakaryocytes of these patients, leading to severely impaired proplatelet formation and congenital thrombocytopenia [20][21][22]. It is noteworthy here that these patients do not develop the entire Rasa3 2/2 phenotype, like megakaryocytosis and bone marrow fibrosis, and its consequences. This discrepancy suggests that Rasa3 has additional function beside the control of integrin signaling, and/or that the enzyme has slightly different roles in man and mouse.
In a recent report, Peters and collaborators have described a new spontaneous mutant mouse with a missense mutation in the Rasa3 protein [18]. The G125V Rasa3 Scat mutation causes mislocalization of the protein to the cytosol and phenotypical traits that are clinically and biologically most often different from SCID-Rasa3 2/2 and Rasa3 2/2 phenotypes (Table S5). Indeed, Rasa3 Scat/Scat mice have a cyclic phenotype of crisis-remission with a first embryonic to P9 wave of lethality -which affect ,60% of the mutant mice -followed by a second wave of lethality at P30 (affecting 94% of the first crisis survivors). This unexplained cyclic phenotype is fully transferable via hematopoietic stem cells injection into SCID or RAG 2/2 mice, ruling out the possibility that expression of the mutant Rasa3 Scat protein outside the hematopoietic system is responsible for the different phenotype [18]. Another notable difference between Rasa3 Scat/Scat and SCID-Rasa3 2/2 mice is the presence of a delayed erythropoiesis in the former mice. By contrast, in SCID-Rasa3 2/2 mice, many hallmarks of regenerative anemia are present. It is noteworthy that no bone marrow fibrosis nor extramedullar hematopoiesis have been reported in the Rasa3 Scat/Scat model, and no mechanism was presented to explain the severe Rasa3 Scat/Scat thrombocytopenia. Finally, no evidence for predisposition to oncogenesis was observed in Rasa3 Scat/Scat mice, but the very small numbers of homozygous mice that survive the second crisis period (,6% of Scat/Scat newborns) may explain this difference and preclude more extensive analysis. The cause of the major differences between the Rasa3 Scat/Scat and SCID-Rasa3 2/2 phenotypes is currently not known, but may be due to the different mutation present in the Rasa3 protein -affecting protein localization and enzymatic activity, respectively -and/or to the different genetic background of the two models. Indeed, relocalization of the Rasa3 Scat/Scat protein from the membrane to the cytosol may eventually create a new function in this cell compartment and lead to phenotypic alterations that are not present in mice expressing a catalytically-inactive and truncated Rasa3 protein.
About 20% of SCID-Rasa3 2/2 mice develop a preleukemia characterized by a massive infiltration of bone marrow and spleen with CD117 + Sca-1 + CD38 + cells, a phenotype very similar to acute myeloid leukemia in man. The exact mechanism of this preleukemia was not defined in this work, but active GTP-bound Ras level was similar in Rasa3 +/+ and Rasa3 2/2 fetal liver CD117 + hematopoietic stem cells. However, our studies in the human K562 leukemic cell line which overexpresses Rasa3 suggest that Rasa3 is a probable negative regulator of proliferation in these cells (Fig. S5). Alternatively, it has been reported that b1 and b3 integrin signaling regulates the balance among hematopoietic stem cell self-renewal, differentiation and quiescence in the osteoblastic niche [35,36]. Furthermore, b1 and b3 integrins can regulate stem cell functions via direct or indirect participation in cellular signaling [37], providing a potential mechanism to explain the predisposition to preleukemia in a minor percentage of SCID-Rasa3 2/2 mice.
In conclusion, our results demonstrate that mice with a catalytic inactivation of Rasa3 protein in the hematopoietic system develop a lethal syndrome characterized by defects during megakaryocyte development and differentiation, and leading to a severe thrombocytopenia. This syndrome is associated with Rap1 and integrin signaling alterations and a predisposition to develop preleukemia.

Ethics statement
All animal studies were authorized by the Animal Care Use and Review Committee of the Université de Liège and of the Université Libre de Bruxelles.

Mice
Rasa3 2/2 mice with Rasa3 exons 11 and 12 replaced by a neomycin resistance cassette express a catalytically-inactive Rasa3 truncated protein [16]. These mice were analyzed on a hybrid 129/SvJ6C57BL/6J genetic background. C.B.-17 SCID mice were purchased from Charles River, Belgium. All mice were bred in a specific pathogen free facility at the GIGA-Research Centre. The Rasa3 genotype was determined by PCR as previously described [16]. For reconstitution, 4-6 week-old C.B.-17 SCID mice were irradiated (200 rad) and a total homogenate of E12.5 fetal liver cells (FLC) obtained from Rasa3 embryos was intravenously injected. SCID-Rasa3 2/2 mice were killed and analyzed either when moribund (ie presenting a severely reduced mobility and/or feeding incompatible with a more than 2 days survival) or 14 months after irradiation/reconstitution.

Fetal liver cells (FLC) isolation and megakaryocyte differentiation
Individual liver was recovered from E12.5 embryo and single cell suspension was prepared by passage through a 23-gauge needle. Recovered cells were cultured in DMEM (Gibco) supplemented with 10% heat-inactivated FBS, 2 mM L-Glutamine, 50 U/mL Penicillin, 50 ng/mL streptomycin, 0.1 mM nonessential amino acids and 50 ng/ml of recombinant mouse TPO for megakaryocyte differentiation (PreProtech).

Bone marrow explants analysis
Bone marrow from SCID-Rasa3 +/+ and SCID-Rasa3 2/2 femurs were flushed with PBS. The marrow was cut in 1 mm transverse sections and placed in an incubation chamber containing complete DMEM medium. Chamber was maintained at 37uC for 6 h. Megakaryocytes at the periphery of the explant were observed under a confocal microscope (Nikon A1R, 206 objective). Each experiment was performed in duplicates. One transversal section was used to determine by flow cytometry the number of CD41 + cells present in the explant. Images were acquired sequentially at 10 min intervals and processed with NISsoftware and ImageJ. Three mice from each genotype were analyzed.
Inside-out aIIbb3 integrin and outside-in integrins signaling in megakaryocytes FLC from Rasa3 +/+ and Rasa3 2/2 embryos were cultured in the presence of TPO as described above. On day 3, recovered cells were enriched for mature megakaryocytes on a 1.5-3% bovine serum albumin (BSA) gradient under gravity for 45 min at room temperature. The percentage of mature megakaryoctes in the enriched population was always over 70%. Cells were resuspended in Tyrode's buffer containing 1 mM CaCl 2 and 1 mM MgCl 2 for 3 h. For inside-out integrin signaling, cells were incubated for 30 min at room temperature with FITC-fibrinogen (250 mg/ml) and 100 ng/ml TPO, 1 mM MnCl 2 or nothing, in the presence or absence of 10 mM EDTA. After a 10-fold dilution with PBS containing 1 mg/ml propidium iodide, fibrinogen binding was quantified by flow cytometry [22]. Specific fibrinogen binding was defined as binding that was inhibited by 10 mM EDTA. To compare independent experiments, specific fibrinogen binding was expressed as a percent of maximal binding obtained in the presence of 1 mM MnCl 2 , an activator of integrins. For outside-in integrin signaling, coverslides were coated with murine fibrinogen (100 mg/ml), collagen-I (35 mg/ml) or Poly-D-Lysine (PDL, 15 mg/ml) for 1 h at room temperature, blocked with denatured BSA (5 mg/ml) for 30 min and washed with PBS before use. Cells (25610 3 ) were incubated for 18 h on the indicated substrate and non adherent cells were removed. Adherent cells were fixed in 10% formalin, permeabilized with 0.2% Triton X-100 in PBS and stained as described below. Cells were then analyzed by confocal microscopy and ImageJ Software. For Rap1 inhibitor studies, purified mature megakaryocytes were cultured over PDL coatedplates as in outside-in experiments in the presence of 3 mM GGTI-298 (Sigma) or DMSO as control. Adherent cells were fixed in 10% formalin, permeabilized with 0.2% Triton X-100 in PBS and stained as described above. Cells were then analyzed by confocal microscopy and ImageJ Software.

Flow cytometry analysis and antibodies
A single-cell suspension of femur bone marrow was prepared by flushing the bones with PBS followed by gentle disaggregation through Pasteur pipette. Cells were released from spleen by gentle disruption with a piston of syringe. Spleen cells were treated with ACK buffer to lyse erythrocytes and washed once with PBS. Cells were incubated with 2.4G2 to saturate Fcc receptors II and IIIa before staining with primary and secondary antibodies in PBS containing 0.1% FBS and 0.1% NaF for 20 min, and washed with the same solution before flow cytometric analysis on a FC 500 (Beckman Coulter). Cell counts were determined by adding fluorospheres (Flow-Count Fluorospheres, Beckman Coulter) to the cell suspension, as described by the manufacturer. The following anti-mouse biotinylated or fluorochrome-conjugated antibodies were obtained from BD Pharmingen: anti-CD3e, anti-CD71, anti-CD41 and anti-CD117. Anti-B220, anti-F4/80, anti-IgM, anti-Mac1, anti-Sca-1, anti-CD34, anti-CD38, anti-Ter119, anti-CD41 and anti-Gr1, as well as streptavidinecychrome 5 were obtained from eBioscience. Anti-FceRIa (Mar-1) was obtained from O. Leo's laboratory (Université Libre de Bruxelles, Belgium). JON/A antibody was obtained from Emfred Analytics. Fetal liver cell were analyzed on a FACS CantoII (Beckman Coulter). For hematopoietic stem and megakaryocyte progenitor cells staining, anti-mouse biotinylated or fluorochromeconjugated antibodies specific for Ter-119, Gr1, Mac1, CD4, CD8, CD5, IL7Ra, B220 and c-Kit (CD117) were used to define the c-Kit + Lin 2 cell population [19,38]. Then, anti-Sca-1, anti-CD34 and anti-Flk2/Flt3 were used to define the hematopoietic stem cells, whereas anti-Sca-1, anti-FcRcII/III and anti-CD150 were used to define the megakaryocytes progenitor cells (all antibodies were from eBioscience, except anti-Flk2, from BD Pharmingen and anti-CD150, from BioLegend). Streptavidin phycoerythrin-Texas Red was from Invitrogen. Debris, aggregates and propidium iodide-positive dead cells were first excluded. Cells were analyzed using an LSRII flow cytometer (Becton Dickinson). Data were analyzed with FlowJosoftware (Tree Star, Ashland, OR).

Ploidy assay
Fetal liver cells were stained for CD41 as described above and fixed with 5% formalin for 15 min. Cells were permeabilized in PBS containing 0.25% Tx-100 for 5 min at 4uC. DNA was stained with DAPI for 20 min and DNA content in CD41 + cells was determined by flow cytometry.

Histology
Spleen and liver were fixed in paraformaldehyde 4% and embedded in paraffin following standard procedures. Femurs were fixed in paraformaldehyde 3.7%, decalcified in 0.5M EDTA pH 8 for one week and then processed as spleen and liver. Serially cut 5mm-thick sections were stained with hematoxylin/eosin or Sirius Red (for femur) according to standard protocols.

Immunohistochemistry of spleen and femur
Spleen was processed as described and sections were stained with an anti-B220 antibody [39]. Femur sections were stained with a rabbit polyclonal anti-von Willebrand Factor (vWF) antibody from Dako. For quantification of megakaryocytes in osteoblastic and vascular niches, the whole diaphysis of three consecutive femur sections was scanned with a conventional microscope (206 objective) for vWF + cells, as described [40]. Megakaryocytes in the osteoblastic niche were calculated as the number of megakaryocytes in contact with the endosteal border. Megakaryocytes in the vascular niche were calculated as the number of megakaryocyte per vessel border. Osteoblastic and vascular borders were calculated with ImageJ software. Results are means 6 SEM of 3 mice per genotype.

Blood analysis
Platelet counts were determined with Unopette (Becton Dickinson). Red cells, total white cells, lymphocytes, neutrophils, eosinophils, basophils, hemoglobin, hematocrit and red cell volume were quantified with a Cell Dyn 3500 analyzer (Abott Diagnostic). Serum erythropoietin and thrombopoietin levels were determined with ELISA mouse EPO and mouse TPO Quantikine kits (R&D Systems). Blood smears were stained with Giemsa's, methylene blue and Romanowsky's solutions.

Immunofluorescence and confocal microscope analysis
Immunofluorescence studies using conventional and confocal microscopes were performed on total FLC cultured in the presence of TPO, on purified mature megakaryocytes and on FL hematopoietic stem cells. Cells were fixed in 5% formalin for 15 min, washed, permeabilized with 0.2% Tx-100 in PBS containing 2% of FBS for 15 min and incubated 1 h at room temperature with APC-conjugated anti-CD41 (MW Reg30, eBioscience) for megakaryocyte or CD117 (BD Pharmingen) for HSC. Active, GTP-bound Rap1 or Ras immunofluorescence was detected using GST-RalGDS-RBD or GST-Raf1-RBD, respectively, and a FITC-conjugated mAb against GST (Santa Cruz) as described [41] Negative controls included the omission of GST-RalGDS-RBD/GST-Raf1-RBD, the substitution of GST-RalGDS-RBD/GST-Raf1-RBD with GST and the substitution of the anti-GST antibody with an irrelevant FITC-conjugated mouse IgG. After several washes, phalloidin-TRICT (Sigma) and DAPI (Sigma) were added for 20 min in PBS. After 3 washes in PBS, samples were mounted in ProLong (Invitrogen) for observation under a confocal microscope (NikonA1R) and/or an epifluorescence microscope (Nikon Eclipse 90i). For active Rap1 or Ras images, z-sections of 0.150 microns were acquired from megakaryocytes or HSC. Pseudocolor scale was used to depicture the intensity of active Rap1 or Ras staining along the cell membrane. ImageJ was used to quantify the intensity of active Rap1 or Ras staining on each cell. All images were acquired and analyzed in the same conditions.
For immunofluorescence studies of adherent megakaryocyte, cells were fixed with 10% formalin for 15 min, washed, permeabilized with 0.2% Tx-100 in PBS containing 2% of FBS for 15 min and incubated 1 h at room temperature with the indicated primary and secondary antibodies. After several washes, phalloidin-TRICT (Sigma) and DAPI (Sigma) were added for 20 min in PBS. After 3 washes in PBS, samples were mounted in ProLong (Invitrogen) for observation under a confocal microscope (NikonA1R). The following antibodies were used: APC-conjugated anti-CD41 (MW Reg30, eBioscience), anti-Rap1 (Millipore), anti-Talin-FITC and anti-rabbit-alexa 488.

CFU-Mk assay
A collagen-based system (MegaCult-C, StemCell Technologies, Inc.) was used for the colony assay. Briefly, 1.25610 5 freshly isolated fetal liver cells were resuspended in IMDM completed with recombinant mouse TPO (50 ng/ml), IL-3 (20 ng/ml) and IL-6 (10 ng/ml), followed by addition of cold collagen. Suspension was dispensed into 2 wells of a four chamber slide (Millipore) for duplicates. Cultures were kept at 37uC in a 5% CO2 atmosphere for 3 days. The collagen matrix was then fixed in a methanolacetone solution (1:3), at room temperature for 20 min for colony fixation. Slides were then allowed to air dry for 15 min and stained for Acetylcholinesterase. For scoring, acetylcholinesterase-positive colonies with 3 or more immature megakaryocytes of about 10 mm of diameter were scored as CFU-Mk. Mature megakaryocytes averaged approximately 30 mm in diameter.

Platelet adherence assay
In order to test the adhesion of unstimulated platelets to BSAcoated surface, 3.5610 6 platelets in 300 ml of tyrode's buffer were added to each well of a 8 chambers slide (Millipore) and incubated for 45 min in a CO 2 incubator at 37uC. Adherent platelets were washed twice with PBS, fixed with 10% formalin, and stained with phalloidin-TRICT.

Flow cytometry analyses of platelet activation
Washed platelets were stimulated or not with ADP (25 mM) or collagen-related peptide (CRP) (1 mg/ml),under non-stirring conditions. After 15 minutes of activation, saturating concentrations of FITC-conjugated CD62 anti-P-selectin and PE-conjugated JON/A antibodies were added to the platelets, and incubations were continued for additional 15 minutes in the dark. Samples were fixed before the analysis with a FACS Calibur flow cytometer (BD Biosciences).

Platelet aggregation analysis
Light transmission was recorded during platelet aggregation induced by ADP (50 mM) in the presence of 2 mM CaCl 2 on a Chrono-Log Lumi-Aggregometer (Havertown, PA).

Proliferation assay on a Rasa3-inducible K562 leukemic cell line
The Rasa3-tet-ON-inducible K562 cell line was generated by GEnTarget Inc. Briefly, Rasa3 expression and TetR repressor lentiviruses were generated and cotransduced in K562 cell by the company. K562 mutant cell line (K562-Rasa3) was cultured in IMDM supplemented with 10% heat-inactivated FBS, 2 mM L-Glutamine, 50 U/mL Penicillin, 50 ng/mL streptomycin, 0.1 mM nonessential amino acids, 10 mg/ml blasticidin and 1 mg/ml puromycin. Treatment of K562-Rasa3 cells with tetracycline (2 mg/ml) induced Rasa3 expression from the lentiviral constructs after 48 h. For the proliferation assay, 4610 5 cells per ml were cultured in the absence or presence of tetracycline for 12 days. At the indicated days, number of alive cells was counted with a hemocytometer. Death cells were excluded by trypan blue staining. Rasa3 expression was confirmed by western blot. Two independent experiments were performed in duplicates-triplicates.

Supporting Information
Table S1 Total numbers of T and B cells were determined in the spleen of SCID-Rasa3 +/+ , SCID-Rasa3 +/2 and SCID-Rasa3 2/2 mice 6 weeks after irradiation/reconstitution by flow cytometry on the basis of 145-2C11 and B220 expression. A trend for higher B220 + B cell number was observed in SCID-Rasa3 2/2 mice as compared with SCID-Rasa3 +/+ mice, but the difference did not reach statistical significance (P = 0.053, unpaired t test). Red blood cell, blood platelet and bone marrow megakaryocyte counts as well as spleen weight were also analyzed 6 weeks after irradiation/reconstitution. No significant difference was observed between SCID-Rasa3 +/+ and SCID-Rasa3 2/2 mice. Megakaryocyte counts per field of view were obtained with a 620 objective, 3 fields per mouse, 5 SCID-Rasa3 +/+ and 4 SCID-Rasa3 2/2 mice. (DOC)

Table S2
Bone marrow cells were isolated from SCID-Rasa3 +/+ and SCID-Rasa3 2/2 mice 2 months after irradiation/reconstitution, incubated with antibodies directed against cell surface markers and analyzed by flow cytometry for the percentage of cells within the bone marrow cells or within a subpopulation of bone marrow cells defined by specific markers. (DOC)

Table S3
Age-matched SCID-Rasa3 +/+ and moribund SCID-Rasa3 2/2 mice were analyzed for their total number of nucleated splenocytes and, after flow cytometry with relevant antibodies, for their percentages (%) and cell numbers (n) of splenic mature (macrophages, T and B cells) and immature (megakaryocytes, myeloid cells, hematopoietic progenitors and erythroblasts) cells. Results indicate that in SCID-Rasa3 2/2 mice, total number of nucleated splenocytes as well as percentage and number of immature splenic cells are significantly increased, consistent with a markedly increased hematopoiesis in the spleen of these mice, as compared with SCID-Rasa3 +/+ mice. By contrast, the percentage of mature cells is decreased in the spleen of SCID-Rasa3 2/2 mice, as compared with SCID-Rasa3 +/+ mice, although their number is increased, a probable consequence of the increased hematopoiesis in this organ. (DOC)