mTORC1 activity is essential for erythropoiesis and B cell lineage commitment

Mechanistic target of rapamycin (mTOR) is a serine/threonine protein kinase that mediates phosphoinositide-3-kinase (PI3K)/AKT signalling. This pathway is involved in a plethora of cellular functions including protein and lipid synthesis, cell migration, cell proliferation and apoptosis. In this study, we proposed to delineate the role of mTORC1 in haemopoietic lineage commitment using knock out (KO) mouse and cell line models. Mx1-cre and Vav-cre expression systems were used to specifically target Raptorfl/fl (mTORC1), either in all tissues upon poly(I:C) inoculation, or specifically in haemopoietic stem cells, respectively. Assessment of the role of mTORC1 during the early stages of development in Vav-cre+Raptorfl/fl mice, revealed that these mice do not survive post birth due to aberrations in erythropoiesis resulting from an arrest in development at the megakaryocyte-erythrocyte progenitor stage. Furthermore, Raptor-deficient mice exhibited a block in B cell lineage commitment. The essential role of Raptor (mTORC1) in erythrocyte and B lineage commitment was confirmed in adult Mx1-cre+Raptorfl/fl mice upon cre-recombinase induction. These studies were supported by results showing that the expression of key lineage commitment regulators, GATA1, GATA2 and PAX5 were dysregulated in the absence of mTORC1-mediated signals. The regulatory role of mTOR during erythropoiesis was confirmed in vitro by demonstrating a reduction of K562 cell differentiation towards RBCs in the presence of established mTOR inhibitors. While mTORC1 plays a fundamental role in promoting RBC development, we showed that mTORC2 has an opposing role, as Rictor-deficient progenitor cells exhibited an elevation in RBC colony formation ex vivo. Collectively, our data demonstrate a critical role played by mTORC1 in regulating the haemopoietic cell lineage commitment.


Results mTORC1 plays a critical role in B cell and RBC lineage development in vivo. Genotyping analyses
of Vav-cre +/− x Raptor wt/fl breedings revealed that the mice exhibiting the Vav-Raptor KO (Vav-cre −/+ Raptor fl/ fl ) did not reach weaning age (4 wk), while being present at normal ratios during gestation, suggesting that Vavcre −/+ Raptor fl/fl mice died perinatally ( Table 1). Analysis of embryos (E13-17) revealed a significant downregulation of Raptor expression in FL isolated from Vav-Raptor KO mice as expected, and the embryos were pale during gestation compared to Vav-cre −/− Raptor fl/fl (Vav-Raptor control) (Fig. 1A,B, & data not shown). Induction of Raptor excision with poly(I:C) in adult Mx1-Raptor cKO mice revealed a significant downregulation in Raptor expression in the BM and spleen (Fig. 1C) together with an increase in splenic weight and cellularity, and BM cellularity in Mx1-cre + Raptor fl/fl mice (Mx1-Raptor cKO) compared to Mx1-cre − Raptor fl/fl mice (Mx1-Raptor control) (Fig. 1D-G).
Analysis for the presence of selected haemopoietic lineages in these mouse models revealed a small but significant decrease in the percentage of Ter119 + erythroid populations in Vav-Raptor KO FL at E15 compared to Vav-Raptor control ( Fig. 2A & Suppl Fig. 1A). Additionally, there was a trend in decrease in the percentage of Ter119 + erythroid populations in the BM, coupled with a significant increase in percentage of Ter119 +  Table 1. Deletion of Raptor in the haemopoietic lineages leads to death prior to weaning. The expected mendelian ratios and actual genotyped ratios of Vav-cre +/− Raptor wt/fl mouse matings are shown at weaning (4 wk; n = 26) and at embryonic day 13 (E13; n = 49). Chi test values and statistical analyses are shown (**p ≤ 0.001). p value calculated by the Chi Test which determines p value based on degrees of freedom.
population in the spleen likely due to increased extramedullary haemopoiesis in Mx1-Raptor cKO mice compared to Mx1-Raptor control (  (Fig. 2F). These data suggest a vital role for mTORC1 in B cell development. Both the KO models showed a concomitant increase in the percentage of CD11b + Gr1 − immature myeloid cells, together with a significant decrease in the percentage of CD11b + Gr1 + mature myeloid cells (Fig. 2G,H,J; Suppl. Fig. 1). These data were coupled with a significant increase in the number of CD11b + Gr1 − myeloid cells in the Mx1-cKO BM (Fig. 2I), while the mature myeloid cell numbers were unaltered in the Mx1-cKO BM and spleen (Fig. 2K). These data suggest that myeloid lineage maturation is deregulated in the absence of Raptor expression. As it is well established that mice can survive in the absence of white blood cell populations 22 , the absence of adult Vav-Raptor KO mice is likely due to the critical role of mTORC1 in erythropoiesis which results in the decrease in RBC populations and the pallor of the mice.

mTORC1 plays a critical role in B and RBC lineage commitment in vivo.
To gain a deeper understanding of the stage at which Raptor-deficiency blocks lineage commitment/development we carried out flow cytometric analysis on the haemopoietic progenitor populations. Supporting the lack of CD19 + B cells, a significant reduction in the percentage of pre-proB and pro-B cells was noted in FL and BM isolated from Vav-Raptor KO and Mx1-Raptor cKO mice respectively, compared with control mice (Fig. 3A,B; Suppl. Fig. 2). This finding www.nature.com/scientificreports www.nature.com/scientificreports/ was coupled with a significant elevation in the Lin − Sca-1 + CD117 hi (LSK) population, suggesting a block in B cell development prior to lineage commitment. Of note, no significant differences in LSK and progenitor B cell populations were observed between VavCre +/− Raptor wt/wt and VavCre − Raptor fl/fl mice, indicating that the phenotypic changes were due to the excision of Raptor (Fig. 3A). Analysis of the proportion of myeloid progenitors in E15 www.nature.com/scientificreports www.nature.com/scientificreports/ FL isolated from Vav-Raptor KO mice revealed no significant changes in the Sca-1 lo CD117 hi , CMP and GMP populations compared to the control mice ( Fig. 3C; Suppl. Fig. 3A). An elevation in the percentage of MEPs in E13 and E15 FLs isolated from Vav-Raptor KO mice suggested a developmental arrest in erythropoiesis at the MEP stage in the absence of Raptor expression at the HSC stage in vivo (Fig. 3D). Interestingly, adult mice with an induced Raptor-deficiency exhibited a significant decrease in Sca-1 lo CD117 hi population and MEP population suggesting a block in erythropoiesis prior to the MEP stage in Mx1-Raptor cKO BM (Fig. 3E). Additionally, a significant decrease in GMP population was noted, suggesting that differentiation of CMPs towards GMP and MEPs is deregulated in the BM in Mx1-Raptor cre cKO mice. The difference in progenitor populations between the FL and BM with Raptor-deficiency highlights the potential for mTORC1 to play similar but not identical roles in lineage maintenance at different stages of development and ontogeny.
In support of our phenotypic analyses showing a block in B cell and RBC development, analysis of key genes responsible for enabling differential lineage commitment revealed a significant downregulation in the expression of B cell specific TFs Ebf1 and Pax5 in FLs isolated from Vav-Raptor KO mice, while Cebpα levels were unaltered (Fig. 3F,G). Assessing erythroid lineage commitment factors, Pu.1, β-Globin and Gata1 levels were significantly

Inhibition of mTORC1 blocks RBC differentiation in vitro.
To further assess the role of mTORC1 in RBC differentiation in vitro, we made use of the human BCR-ABL + erythroleukaemia cell line K562, that differentiates towards a RBC-like lineage when exposed to stress [23][24][25] . Stress was induced by replacing glucose in complete media with galactose (Gal-media) 26 . A significant increase in the percentage of CD71 + GlyA + cells, an elevation of the erythroid marker CD71 surface expression and a reduction in cellular granularity was observed (Fig. 4A,B, Suppl. Fig. 4A) in K562 cells cultured in Gal-media, coupled with a significant increase in gene expression of β-GLOBIN, GATA1 and GATA2 (Fig. 4C). Treatment of K562 cells with either rapamycin (allosteric mTORC1 inhibitor) or AZD8055 (dual mTORC1/2 inhibitor) reduced erythroid differentiation in vitro, as indicated by a significant decrease in the generation of erythroid cells and a reduction in gene expression of erythroid markers, (Fig. 4B,C), indicating mTORC1 inhibition blocks RBC differentiation in vitro.
To assess erythroid colony formation capacity of HPCs in the absence of mTORC1 activity, HPCs were isolated from BM of Mx1-Raptor control or cKO mice and CFC assays were performed. Mx1-Raptor cKO mice lacked CFC activity, as indicated by the lack of colony-forming unit-erythroid (CFU-E), blast-forming unit-erythroid (BFU-E) or granulocyte-erythroid-megakaryocyte-macrophage (GEMM) colonies in the absence of Raptor expression, compared to cre − controls (Fig. 5A,B). Interestingly, myeloid progenitor CFC assays performed in Vav-cre − (Vav-Rictor control) and cre + Rictor fl/fl (Vav-Rictor KO) HPCs to assess the role of mTORC2 in early myeloid/erythroid colony formation demonstrated a significant increase in CFU-E colonies and a trend in increase in GEMM colonies in Vav-Rictor KO HPCs (Fig. 5C,D), suggesting a suppressive role of mTORC2 in erythropoiesis. Supporting these data, flow cytometric analysis of the CFCs generated showed a decrease in percentage of Ter119 + cells in Mx1-Raptor cKO HPCs, while an increase was noted in Vav-Rictor KO HPCs (Suppl. Fig. 5).

Discussion
Our data support previously published work demonstrating a critical role of mTORC1 during B cell and erythroid development in two mouse models of Raptor-deletion in vivo, and further demonstrate the importance of mTORC1 function in in vitro RBC differentiation models: CFC assays and human cell line K562. www.nature.com/scientificreports www.nature.com/scientificreports/ Examining the role of mTORC1 in normal erythropoiesis at the HSC stage, in both fetal and adult mice, revealed subtle differences at the two developmental stages. Raptor-deficient embryos were paler than Raptor-control mice, characteristic of mice with RBC deficiency as noted in EPO/EPO-R mice 27 . Knight et al., demonstrated high mTORC1 activity in proE cells (reticulocytes) characterised by elevated phosphorylation of the downstream target S6 thereby identifying a critical role for mTORC1 in RBC development. Additionally, they showed a link between iron availability and haemoglobin (Hb) synthesis in RBC development with mTORC1 signalling, as RBCs have decreased mTORC1 signalling during iron deficiency (ID) in vitro and in vivo 19 . We observed a significant decrease in Ter119 + erythrocytes, with an arrest at the MEP stage shown by a significant increase in the MEP population at E15, together with an increase in Sca-1 lo CD117 hi cells in the FL of Vav-Raptor KO. This is consistent with published data demonstrating Vav-Raptor KO FL exhibit microcytic anaemia with www.nature.com/scientificreports www.nature.com/scientificreports/ decreased Hb, exhibiting increased proE cells and decreased EryC populations. Conversely, constitutive activation of mTORC1 (Vav-TSC1 KO) leads to macrocytic anaemia with larger reticulocytes and RBCs with elevated Hb levels. Despite increased Hb levels in RBCs, there was a decrease in RBC output in the BM (decreased EryB), leading to stress erythropoiesis in the spleen (elevated EryA population) with an overall decrease in Hb 19 . A steady decline in progenitor populations occurred over time, with no significant difference in MEP populations between the controls and Vav-Raptor KO FL at E17 (data not shown), diminishing any significant differences observed in FL at E13/E15. This is likely due to a transition in haemopoiesis from the FL to the BM, which begins before birth at E16.5 28 . The block in development of the erythroid lineage was supported by a significant downregulation in Gata1, β-Globin, Klf1 and Klf2 expression with Raptor-deficiency at the HSC stage, suggesting mTORC1-mediated signalling may aid in the regulation of GATA switching to enable erythropoiesis.
In contrast, induced Raptor-deficiency in adult mice (Mx1-Raptor cKO) is not lethal, and these mice displayed a significant decrease in Sca-1 lo CD117 hi and MEP populations in the BM. This may be due to the lineage potential of progenitors differing between fetal and adult mice, as FL HSCs possess a higher proliferative and metabolic capacity (increased oxidative phosphorylation) than BM-derived HSCs which could affect Sca-1 lo CD117 hi and MEP primitive populations 29 . Nevertheless, there is a trend in decrease in Ter119 + erythroid population in the BM of adult mice with Raptor-deficiency, as noted in the Vav-cre model, suggesting a disruption in erythropoiesis in the BM. Indeed, Guo et al., demonstrated that Mx1-Raptor cKO mice have reduced RBCs and Hb in the BM due to increased apoptosis of RBCs. Additionally, erythrocyte development was blocked at the proE stage 30 confirming our results that mTORC1 blocks RBC development. Mx1-Raptor cKO mice exhibited splenomegaly, with the spleen showing a significant increase in Ter119 + population, which could suggest a possible compensatory mechanism leading to the promotion of erythropoiesis due to a lack of erythrocytes, suggesting extramedullary haemopoiesis in the spleen, as has been shown previously 31 . Alternatively, as the spleen is known to be a main site for erythrophagocytosis, it is possible that the function of these erythrocytes is compromised and are accumulated in the spleen for phagocytosis 30 . We observe splenomegaly and an increase in red pulp. As the red pulp is one of the major sites of erythrocyte destruction, it is possible that an accumulation of erythrocytes in the spleen indicates erythrocyte depletion 32 . The observed differences between the models may be due to the mechanism of Raptor-excision. While both models result in the deletion of Raptor in all haemopoietic lineages, the Mx1-Raptor cKO requires inoculation with poly(I:C), through induction of type 1 interferons, which may disrupt haemopoietic cell survival. Published studies indicate that the impact of poly(I:C) treatment on haemopoietic cells is transient, lasting for up to 48 hr 33 . Here, phenotypic analyses were carried out 5 wk after the last poly(I:C) inoculation, by which time the key poly(I:C) induced effect should be Raptor-excision. While this timeline may enable the outgrowth of undeleted alleles in Mx1-Raptor cKO mice, the optimisation of poly(I:C) induced Raptor-excision leading up to this study reduced the likelihood of this occurring in the timeframe of these experiments, and Raptor expression was consistently downregulated up to 8 wk post-poly(I:C) inoculation.
Using established mTOR inhibitors to block RBC differentiation in the K562 cell line provided support for the results presented from our mouse models, indicating that erythroid development is dependent on mTORC1 activity. Interestingly, a recent report shows that K562 cells express a distinct mTOR-containing complex, mTORC3 (mTOR associated with ETV7) that lacks other mTORC1/2 containing proteins, and is associated with rapamycin-resistance when upregulated in human cancer cells 34 . Our finding that K562 differentiation is inhibited by rapamycin treatment suggests that mTORC3 does not play a major role in this process. Supporting this, Ohyashiki et al. have previously shown that iron regulates mTORC1 signalling in RBCs as iron chelators greatly reduced mTORC1 signalling in K562 cells 35 . Overexpression of eIF4E and sustained mTORC1 expression in erythroid progenitor cell line I/11 also improves differentiation 36,37 . An additional role of mTORC1 in initiation of translation at early stages in RBC development has been reported, demonstrating the regulation of mitochondrial biogenesis from HSPCs to proerythroblasts by mTORC1 38 . This study shows a unique requirement for mTORC1 in RBC development. These data, together with our findings suggest that mTORC1, iron metabolism and erythropoiesis have an inter-dependent regulatory link.
Assessment of erythroid colony formation capacity revealed that Raptor-deficient BM lacked colony formation capacity. Consistent with this, Mx1-Raptor cKO cells have previously been shown to lack the ability to form CFU-E and BFU-E colonies 30 . Indeed, shRNA targeting S6K1 in murine BM cells significantly decreased CFU-E and BFU-E colony formation capacity demonstrating the importance of mTORC1-S6K axis is mediating erythropoiesis 39 . ER-Raptor cKO LSK cells also lack colony formation capacity which is rescued by the retroviral transduction of Raptor suggesting the importance of mTORC1 in haemopoiesis 40 . Our assessment of Rictor-deficient BM colony formation capacity revealed that Vav-Rictor KO BM formed a significantly higher number of erythroid colonies than their controls. This suggests a possible feedback system for erythroid regulation wherein mTORC1 drives erythroid formation, whereas mTORC2 suppresses erythropoiesis. Our study compared the Mx1-Raptor cKO and Vav-Rictor KO models. Vav-mediated KO models generate a constitutive KO, which can result in signalling pathway adaptation 41 . Indeed, crosstalk between ERK-MAPK and mTOR signalling pathways has recently been demonstrated, with ERK activity regulating mTORC1 activation to limit the promotion of HSC cycling in favour of quiescence. HSCs derived from MEK1-cKO mice exhibit exhaustion due to increased mTORC1-mediated ROS production resulting in increased mitochondrial damage 42 . While each model has a different timing for gene deletion, the cells placed in CFC assays were removed from Mx1-Raptor mice 5 wk after inoculation with poly(I:C) in vivo, thus inducing optimal Raptor excision.
Of note, publications also suggest that mTORC1 negatively regulates erythropoiesis, via heme-regulated eIF2α kinase (HRI) and activation transcription factor 4 (ATF4), in a mouse model of ID. HRI is activated in response to heme deficiency, thereby stimulating ATF4 to regulate Hb levels via the translation of stress response genes. In mutant mice that lack phosphorylated eIF2α (eIF2αP) in the RBC lineage (eAA) 43 , mTORC1 activity is elevated in erythroid precursors. This study demonstrated that eIF2αP and ATF4 of HRI-integrated stress response are required in ID to repress mTORC1 signalling and mitigate ineffective erythropoiesis. As such, pharmacological www.nature.com/scientificreports www.nature.com/scientificreports/ inhibition of mTORC1 significantly improved RBC production and differentiation, and increased Hb levels in the blood 43 . Furthermore, constitutive activation of mTORC1 in haemopoietic lineages exhibited macrocytic hyperchromic anaemia with splenomegaly and ineffective erythropoiesis 19 , similar to that seen in HRI −/− mice. However, although ATF4 is required to inhibit mTORC1 signalling, ATF4 −/− mice develop microcytic hypochromic anaemia, unlike HRI −/− mice 43 . These studies suggest that mTORC1 activity plays an inhibitory role in erythropoiesis during ID anaemia. Furthermore, Foxo3 −/− mice (model with similar genotypic profile to β-thalassemia), exhibit increased phosphorylation of mTOR targets including AKT, S6, 4EBP1, suggesting aberrant mTOR signalling in diseased conditions. Treatment of Foxo3 −/− mice with rapamycin increased RBC counts and Hb levels, with increased cell cycling in immature erythrocytes compared to untreated mice 39 . Therefore, in contrast to normal erythropoiesis, where mTORC1 plays a fundamental role in RBC generation, diseased conditions alter mTORC1 signalling to disrupt erythropoiesis, which is overcome by mTORC1 inhibitors.
The early block in B cell lineage commitment and development in the FL and BM of both Raptor-deficient models, mirrored that described previously in Mx1-Raptor cKO mice 31 . The observed increase in LSK populations, which normally give rise to pre-proB cells, in both BM and E15 FL suggests a block at the LSK stage during B cell development with Raptor-deficiency at the fetal and in adult stages of development. These findings were supported by the observed downregulation in Ebf1 and Pax5, TFs essential for B cell lineage commitment. A similar trend has been observed in the BM of mTOR-cKO mice, and the spleen of Mx1-Raptor cKO mice, with the majority of LSK cells in S phase suggesting increased cell cycling due to perturbations in mTORC1 signalling 29,30 . Iwata et al., demonstrate a block in B cell development at the pre-B cell stage within Mb1-Raptor cKO adult mice, which is not rescued by the introduction of an anti-apoptotic Bcl XL transgene suggesting that this block, caused by the lack of mTORC1, is independent of BCL XL 44 . Our model, which excises Raptor at the HSC stage, demonstrates a requirement for mTORC1 signalling for B cell lineage commitment. In ER-Raptor cKO BM there is an increase in LSK cells together with a decrease in phosphorylation of 4EBP1 and S6 and decrease in mature B cells 40 indicating that the role of mTORC1 in B lymphopoiesis is not model specific. Indeed, Zeng et al. highlight the importance of mTORC1 specifically in B cell development and proliferation, but not survival, in an IL7R-mTORC1-Myc dependent and STAT5-independent manner, as the loss of mTORC1 leads to a block at the proB cell stage in CD2-mTOR and CD2-Raptor cKO models 45 . Furthermore, Keating et al. demonstrated a critical role for mTORC1 in B cell class switching and anti-viral responses, with Rosa26-Raptor KO mice having reduced germinal centre formation in response to influenza infection, a finding that was replicated with rapamycin treatment. Interestingly, rapamycin treatment prevented B cell class-switching, yielding antibodies that mediated heterosubtypic protection, by targeting more conserved regions of hemagglutinin, a surface protein expressed on viruses 46 . H&E staining of spleens isolated from Mx1-Raptor cKO mice lack a defined architecture and displayed a reduction in marginal zones and germinal centres (data not shown).
In agreement with previously published data, similar roles of mTORC1 in myeloid lineage development were observed in both fetal and adult mice with a significant increase in CD11b + Gr1 − populations and a concomitant decrease in CD11b + Gr1 + populations in both Vav-Raptor KO FL and Mx1-Raptor cKO BM 30,31 . This result has been recapitulated in ER-cre model suggesting that the role of mTORC1 in myeloid development is not model specific 40 . We observed a decrease in GMP population leading to a decrease in mature myeloid population with Raptor-deficiency in Mx1-Raptor cKO mice. Indeed, a decrease in CMPs has previously been reported in Mx1-Raptor cKO mice 30 supporting the decrease in mature myeloid lineage with Raptor-deficiency. However, the increase in CD11b + population in both models also suggests that even though there is a block in early myeloid populations, it does not disrupt the ability of the progenitors to develop into immature myeloid cells. ER-Raptor cKO mice displayed an increase in GMP and MEP populations with a block in CD11b + Gr1 + mature myeloid populations with no changes in pS6, alluding to the phenotypic differences between KO models. The lack of change in pS6 expression in myeloid cells (with a significant decrease in pS6 in B cells) suggests a differential regulation of mTORC1 activity in different haemopoietic lineages 40 . Nonetheless, mTORC1 plays a role in the maturation of myeloid cells in both adult and fetal models. While previous studies support a cell autonomous effect of Raptor-deletion in the development of haemopoietic lineages 30,31 , mixed BM chimeras should be performed to confirm our results, as myeloid lineage expansion has the potential to inhibit erythroid and lymphoid lineages 47,48 . In addition, deletion of Raptor in HSCs has been shown to affect osteoclast generation, which in turn would impact the BM microenvironment and subsequent haemopoiesis 49 .
Taken together, we performed analyses in two different mTORC1 KO mouse models at different stages of development in vivo and two distinct RBC differentiation assays in vitro to tease out the role of mTORC1 in haemopoietic lineage development. We confirm the fundamental role of mTORC1 in haemopoiesis, during RBC and B cell lineage commitment both in fetal and adult mice (Fig. 5E).

Materials and Methods
Mice. www.nature.com/scientificreports www.nature.com/scientificreports/ lineage cells in a single cell suspension were isolated from BM, spleen and blood of transgenic mice, and filtered through a 70 µm nylon mesh (Fisher Scientific, Leicestershire, UK). The cells collected from the BM and spleen were enriched for haemopoietic lineage cells by density centrifugation using Lympholyte-Mammal (Cedarlane, Canada), centrifuging the cell suspension for 20 min at 625 g at RT. Thereafter the cells were washed in PBS, centrifuged at 500 g for 10 min at RT and counted.
Flow cytometry. Samples derived from the organs were prepared for flow cytometry as described previously 51 . All antibodies were purchased from BD Biosciences (Oxford, UK), except Gr-1 (Clone RB6-8C5; eBiosciences) and Ter119 (Clone TER-119; BioLegend). K562 cells with/without treatment were harvested and 2 × 10 5 cells were stained with GlyA (Clone GAR2) and CD71 (Clone AC102) from Miltenyi Biotec. All samples were suspended in 100 µl PBS and acquired using the FACSDiva software package (BD Biosciences) on the FACSCantoII flow cytometer. The resultant data were analysed using FlowJo software (Tree Star Inc., OR).
qRT-PCR. RNA was extracted from fresh cells by following the RNAeasy Qiagen Kit protocol. cDNA was then made from the RNA using standard protocols provided by Invitrogen. Alternatively, CD45 + cells were sorted into the PCR tubes (300 cells/tube) containing the one-step PCR master mix of 5 µl (Cells direct 2x reaction (2.8 µl), 0.2x Primer mix (1.4 µl), RNAase out (0.056 µl), Superscript III RT/Platinum Taq Mix (0.112 µl), TE Buffer (0.672 µl) (part of SuperScript ™ III Platinum ™ One-Step qRT-PCR Kit, Invitrogen)). Samples were run on a PCR with the following conditions: 50 °C for 15 min, 95 °C for 2 min followed by 20 cycles of 95 °C for 15 sec and 65 °C for 4 min. The samples were then held at 4 °C. After the completion of the PCR, the samples were diluted 10x with TE buffer (45 µl). qPCR was carried out by using 300 nM of forward and reverse primer for each gene. All reactions were performed in technical triplicates and at least three biological replicates using the 7900HT Fast Real-Time PCR system (Applied Biosystems, Warrington, UK), programmed to complete 40 cycles as follows: 50 °C for 20 min, 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 sec, and 60 °C for 1 min. The primers used are listed in Supplementary Table 1. The q/RT-PCR follows the MIQE guidelines 52 .
Colony forming cell (CFC) assay. Bones from transgenic mice were crushed and enriched for haemopoietic lineages as described above. Haemopoietic progenitor cells (HPCs) were isolated using the MACS mouse CD117 MicroBeads according to the manufacturer's protocol (Miltenyi Biotec, Surrey, UK). Up to 2 × 10 4 cells were then plated in MethoCult TM M3334 erythroid colony promoting methylcellulose or MethoCult TM GF M3434, in 6 cm dishes as per the manufacturer's protocol (Stemcell Technologies, Grenoble, France). The dishes were monitored and colonies were counted.
Statistics. The Chi Test was used to assess whether there were statistical differences between the different genotypes generated from crossing Vav-cre −/+ Raptor fl/fl mice. Chi Test calculates deviations by chance if the expected ratios are known. Medelian ratios were the expected ratios, which were compared to the genotypes observed at weaning stage of mice. The statistical significance, if any, is calculated using the Chi Test taking into account the degrees of freedom 53 . Statistical analyses were carried out between the data sets using GraphPad Prism 6 Software (San Diego, California, USA). Statistical analyses carried out were unpaired t tests, or for data with multiple comparisons one way ANOVA, where p ≤ 0.05 is considered significant. * ≤ 0.05; ** ≤ 0.001; *** ≤ 0.0001; **** ≤ 0.00001.