|Home | About | Journals | Submit | Contact Us | Français|
Pten deletion from adult mouse hematopoietic cells activates the PI3-kinase pathway, inducing hematopoietic stem cell (HSC) proliferation, HSC depletion, and leukemogenesis. Pten is also mutated in human leukemias, but rarely in early childhood leukemias. We hypothesized that this reflects developmental changes in PI3-kinase pathway regulation. Here we show that Rictor deletion prevents leukemogenesis and HSC depletion after Pten deletion in adult mice, implicating mTORC2 activation in these processes. However, Rictor deletion had little effect on the function of normal HSCs. Moreover, Pten deletion from neonatal HSCs did not activate the PI3-kinase pathway or promote HSC proliferation, HSC depletion, or leukemogenesis. Pten is therefore required in adult, but not neonatal, HSCs to negatively regulate mTORC2 signaling. This demonstrates that some critical tumor suppressor mechanisms in adult cells are not required by neonatal cells. Developmental changes in key signaling pathways therefore confer temporal changes upon stem cell self-renewal and tumor suppressor mechanisms.
Hematopoietic stem cells (HSCs) persist throughout life by undergoing self-renewing divisions regulated by networks of proto-oncogenes and tumor suppressors (He et al., 2009). Mechanisms that promote HSC self-renewal are commonly over-activated or ectopically activated in leukemia cells (Reya et al., 2001). These self-renewal mechanisms are co-opted by leukemia cells irrespective of whether they arise from HSCs or from restricted progenitors (Krivtsov et al., 2006).
HSC properties and self-renewal mechanisms change over time. The transcription factor Sox17 is required for the maintenance of fetal and neonatal, but not adult, HSCs (Kim et al., 2007). Conversely, the epigenetic/transcriptional regulators Bmi-1, Tel/Etv6 and Gfi-1 regulate the maintenance of postnatal, but not fetal, HSCs (Lessard and Sauvageau, 2003; Park et al., 2003; Hock et al., 2004a; Hock et al., 2004b). The temporal changes in HSC self-renewal mechanisms raise the question of whether the driver mutations that are competent to cause leukemia also change over time.
Hematopoietic malignancies occur throughout life and the incidence of specific driver mutations varies with patient age, even within the same subtype of leukemia (Downing and Shannon, 2002). Myeloid leukemias in young children are often driven by translocations involving transcription factors (Downing and Shannon, 2002). In contrast, myeloid leukemias in older children and adults are much more likely to have mutations that activate the PI3-kinase pathway, including Bcr-Abl and Flt3 internal tandem duplications (Meshinchi et al., 2001; Zwaan et al., 2003; Martelli et al., 2006; Kharas et al., 2008; Kharas et al., 2010). These observations raise the question of whether age-dependent changes in leukemia driver mutations occur because of developmental changes in signaling mechanisms that regulate self-renewal.
The Pten tumor suppressor negatively regulates PI3-kinase pathway signaling and is commonly inactivated in many cancers (Di Cristofano and Pandolfi, 2000). Pten loss-of-function mutations occur frequently in T-precursor ALL (T-ALL) and Pten is inactivated in some acute myeloid leukemias (AML) by mutations, promoter methylation, or transcriptional repression (Dahia et al., 1999; Gutierrez et al., 2009; Larson Gedman et al., 2009; Yoshimi et al., 2011). While B-ALL usually presents in young children and rarely involves Pten deletion (Mullighan et al., 2011) pediatric T-ALL exhibits an older mean age of presentation (~9-10 years of age) (Smith et al., 1999; Clappier et al., 2007; Karrman et al., 2009) and commonly involves Pten deletion or other mutations that activate PI3-kinase pathway signaling (Gutierrez et al., 2009).
One possibility is that PI3-kinase pathway regulation changes over time in hematopoietic cells. Pten deletion from hematopoietic cells in adult mice leads to the development of T-ALL (Yilmaz et al., 2006; Zhang et al., 2006; Guo et al., 2008; Lee et al., 2010). Pten deletion from adult mouse HSCs causes these cells to go into cycle and to be depleted by a tumor suppressor response induced by hyper-activation of mTOR (Yilmaz et al., 2006; Lee et al., 2010). Both leukemogenesis and HSC depletion after Pten deletion can be rescued by treating mice with rapamycin (Yilmaz et al., 2006). This suggests that these phenotypes are mediated by RAPTOR/mTORC1 activation; however, rapamycin can also indirectly inhibit RICTOR/mTORC2 signaling (Sarbassov et al., 2006) and mTORC1 activation is influenced by mTORC2 signaling (Guertin and Sabatini, 2007; Guertin et al., 2009). Consequently, the relative contribution of mTORC1 and mTORC2 to these phenotypes is uncertain. It also remains unclear whether Pten deletion drives HSCs into cycle by activating physiological mechanisms that induce HSC division or by activating mechanisms that do not normally regulate HSC division.
To address these questions we evaluated PTEN function, RICTOR/mTORC2 function, and PI3-kinase pathway activation in fetal, neonatal and adult HSCs. AKT phosphorylation did not detectably increase in HSCs that divided under physiological conditions. Consistent with this, Rictor deletion had little effect on HSC frequency or function. However, Rictor deletion largely rescued the increased AKT phosphorylation, HSC depletion, and leukemogenesis observed after Pten deletion from adult HSCs. This suggests that Pten deletion promotes HSC proliferation by activating signaling pathways that have a limited role in HSC self-renewal under physiological conditions. The observation that Pten deletion provides no selective advantage to neonatal hematopoietic cells may explain why Pten mutations are rare in leukemias that affect young children. Temporal changes in the signaling mechanisms that regulate self-renewal lead to temporal changes in leukemogenic mechanisms.
HSCs divide rapidly during fetal and neonatal development (Morrison et al., 1995) and after cyclophosphamide/G-CSF treatment (Morrison et al., 1997). We compared the cell cycle status of HSCs (CD150+CD48−Lin−Sca-1+c-kit+) isolated from embryonic day (E)14.5 fetal mice, postnatal day (P)14 neonatal mice, 8 week old adult mice, and mice that were treated with cyclophosphamide followed by two daily doses of G-CSF (Fig. 1A, B; Table S1). We also administered poly-inosine:poly-cytosine (pIpC) to Mx1-Cre; Ptenfl/fl mice 6 weeks after birth and isolated HSCs for cell cycle analysis at 8 weeks after birth. The fraction of HSCs in S/G2/M phases of the cell cycle decreased significantly from E14.5 to P14 to 8 weeks after birth (Fig. 1B). The frequency of HSCs in S/G2/M phases of the cell cycle after cyclophosphamide/G-CSF treatment was significantly greater than after Pten deletion (Fig. 1B).
To assess whether AKT activation correlates with the frequency of HSCs/MPPs in cycle we compared AKT phosphorylation between CD48−Lineage−Sca-1+c-Kit+ (CD48−LSK) HSCs/MPPs (Kiel et al., 2005; Kim et al., 2006) from E14.5 fetal liver, P14 bone marrow, 8-week old adult wild-type bone marrow, and 8-week old bone marrow after Pten deletion from Mx1-Cre; Ptenfl/fl mice. The cell cycle distribution in the HSC/MPP population was similar to that of highly purified HSCs (Fig. 1C), consistent with prior studies (Foudi et al., 2009). As expected, AKT and S6 were more highly phosphorylated in Pten-deficient adult HSCs/MPPs as compared to wild-type adult HSCs/MPPs (Fig. 1D). However, phosphorylated AKT levels were lower in fetal and neonatal HSCs/MPPs compared to normal adult HSCs/MPPs (Fig. 1D), even though fetal and neonatal HSCs/MPPs were more mitotically active (Fig. 1C). Levels of phosphorylated S6 were higher in fetal HSCs/MPPs but lower in neonatal HSCs/MPPs as compared to adult HSCs/MPPs (Fig. 1D). Phosphorylated AMPK and GSK3ß levels did not significantly differ between fetal, neonatal, and adult HSCs/MPPs. Fetal HSCs/MPPs in G0/G1 versus S/G2/M phases of the cell cycle exhibited similar levels of AKT phosphorylation (Fig. 1E).
We next evaluated AKT and S6 phosphorylation in sorted HSCs/MPPs from the bone marrow of untreated or cyclophosphamide/G-CSF-treated 8 week-old adult mice. After cyclophosphamide and 2 days of G-CSF, HSCs/MPPs did not exhibit increased AKT or S6 phosphorylation (Fig. 1F) despite the high percentage of cells in S/G2/M (Fig. 1B, C). Granulocyte-monocyte progenitors (Lineage−c-kit+Sca-1−CD34+CD16/32+ cells; GMPs) from the same mice did have higher levels of phosphorylated AKT and S6 after cyclophosphamide/G-CSF treatment (Fig. 1F). CD48−LSK cells were also harvested from the spleens of mice treated with cyclophosphamide and 4 days of G-CSF. HSCs/MPPs from these mice still did not exhibit increased AKT phosphorylation relative to normal adult HSCs, though an increase in 4EBP phosphorylation was observed (Fig. 1G). Rapidly dividing HSCs therefore exhibit little or no increase in AKT phosphorylation relative to quiescent adult HSCs.
To determine whether Pten regulates neonatal HSCs, we induced Cre in Mx1-Cre; Ptenfl/fl mice by administering a single subcutaneous dose of pIpC 2 days after birth, resulting in complete Pten deletion in HSCs (Fig. S1A). Bromo-deoxyuridine (BrdU) was then administered for 24 hours before the mice were sacrificed (Fig. 2A, B). The overall proliferation of unfractionated bone marrow cells at 2, 3, 4, and 5 weeks after birth was not affected by Pten deletion (Fig. 2A). The overall rate of BrdU incorporation into control CD150+CD48−Lin−Sca-1+c-Kit+ HSCs declined from 2 (25±8.5%) to 5 (10±6.8%) weeks after birth as HSCs made the transition to a more slowly dividing adult phenotype (Fig. 2B). Pten deletion had no effect on BrdU incorporation into HSCs at 2 weeks after birth, but Pten-deficient HSCs showed a trend toward increased BrdU incorporation relative to control HSCs at 3 weeks after birth and the effect was statistically significant at 4 and 5 weeks after birth (Fig. 2B).
To determine whether Pten deletion mobilizes neonatal HSCs we administered pIpC to 2 day-old neonatal mice or to 6-week old adult mice, then measured the number of HSCs per spleen at 2, 4, and 8 weeks after birth (Fig. 2C). We observed a >40-fold increase in the number of HSCs in the spleens of 8 week-old adult mice (two weeks after Pten deletion; Fig. 2C). In contrast, Pten deletion had no effect on the number of HSCs in the spleen of 2 week-old mice and only modestly (but significantly) increased the number of HSCs in the spleens of 4 week-old mice (Fig. 2C). We also did not observe any effect of Pten deletion on the overall proliferation of splenocytes until 4 weeks after birth (Fig. 2D). Pten deletion therefore does not lead to increased HSC proliferation or mobilization until around 4 weeks after birth.
We compared the reconstituting capacity of HSCs from neonatal and adult mice after Pten deletion at 2 days or 6 weeks after birth. We transplanted 10 CD45.2+ CD150+CD48−Lin− Sca-1+c-kit+ HSCs from donor mice of each age and genotype into irradiated CD45.1+ mice along with 300,000 wild-type CD45.1+ bone marrow cells. Control adult (Fig. 2E, F) and neonatal (Fig. 2G, H) HSCs gave long-term multilineage reconstitution by myeloid and lymphoid cells in all recipient mice. Pten deficient adult HSCs gave only transient multilineage reconstitution that lasted less than 16 weeks after transplantation in all recipient mice (Fig. 2E, F). Pten deficient neonatal HSCs gave long-term multilineage reconstitution in most recipients that lasted at least 16 weeks after transplantation, but none of these recipients were multilineage reconstituted after 24 weeks (Fig. 2G, H). Pten-deficient neonatal HSCs therefore retained the capacity to give long-term multilineage reconstitution, suggesting that they did not initially depend upon Pten. The ultimate depletion of Pten-deficient neonatal HSCs 16 to 24 weeks after transplantation presumably reflects their maturation into adult HSCs as fetal HSCs acquire an adult phenotype after transplantation into adult mice (Fig. S1B, C).
To test whether Pten deletion causes neoplastic proliferation in neonatal mice we treated littermate control and Mx1-Cre; Ptenfl/fl mice with pIpC at 2 days or 6 weeks after birth. Two weeks after pIpC treatment we observed a 3-fold increase in spleen cellularity and weight in adult but not in neonatal Pten-deleted mice (Fig. 2I, J). We also observed histological signs of myeloproliferative disorder in adult, but not in neonatal, spleen after Pten deletion. Two weeks after Pten deletion in adult mice, spleens exhibited a substantial expansion of the red pulp and extramedullary hematopoiesis consistent with myeloproliferative disorder (Fig 2K, L). Two weeks after Pten deletion in neonatal mice, spleens had qualitatively fewer follicles than spleens from littermate controls (data not shown), consistent with the reduction in B-cells in neonatal Pten deficient mice (Fig. 4J, K). Otherwise, the histological appearance of neonatal Pten-deficient and control spleens was indistinguishable (Fig. 2M, N). Pten is therefore not required to prevent myeloproliferative disorder in neonatal mice.
When Pten is deleted from young adult mice they succumb to T-ALL by 12 weeks after Cre induction (Yilmaz et al., 2006; Lee et al., 2010). This latency period makes it difficult to test whether Pten deletion is leukemogenic in neonatal mice because they would mature into adults before the onset of leukemia; therefore, to test whether Pten is a tumor suppressor in neonatal mice we deleted Pten from p53-deficient mice. In adult mice this greatly accelerates leukemogenesis (Lee et al., 2010). This allowed us to test whether neonatal mice are immediately competent to develop leukemia after Pten deletion or whether an adult developmental context is required. We administered pIpC to Mx1-Cre; Ptenfl/fl; p53−/− mice and littermate controls (that lacked Mx1-Cre) at 2 days or 6 weeks after birth. By 20 days after pIpC treatment, all adult Mx1-Cre; Ptenfl/fl; p53−/− mice died with signs of leukemia (Fig. 3A). Control p53−/− or p53+/− adult mice that retained Pten all survived for at least 100 days with no evidence of hematopoietic neoplasms (Fig. 3A, B). All moribund adult Mx1-Cre; Ptenfl/fl; p53−/− mice had large mediastinal masses and infiltrative precursor T-ALL (Fig. 3E-H). Adult Mx1-Cre; Ptenfl/fl; p53−/− mice therefore developed T-ALL and died within 20 days of pIpC.
At 21 days after pIpC treatment, when all of the adult Mx1-Cre; Ptenfl/fl; p53−/− mice were dead, none of the neonatal Mx1-Cre; Ptenfl/fl; p53−/− mice that had been treated with pIpC at 2 days of age showed any signs of illness (Fig. 3A). These mice did not die until 44 to 60 days after pIpC treatment, when they were 6 to 8 week-old adults (Fig. 3A). Indeed, the Mx1-Cre; Ptenfl/fl; p53−/− mice died at the same time after birth irrespective of whether pIpC was administered at 2 days or 6 weeks after birth (Fig. 3C). When neonatal Mx1-Cre; Ptenfl/fl; p53−/− mice were sacrificed at 17 days after pIpC treatment (a time point at which all adult Mx1-Cre; Ptenfl/fl; p53−/− mice showed clear signs of T-ALL) none of the neonatally-treated mice showed any signs of illness or leukemia: they had normal thymus and spleen histology (Fig. 3I-L); however, when these mice became ill approximately 50 days after pIpC treatment they had large mediastinal masses and infiltrative T-ALL (Fig. 3M-P). The observation that Mx1-Cre; Ptenfl/fl; p53−/− mice succumbed to leukemia at the same time, irrespective of whether Pten was deleted 2 days or 6 weeks after birth (Fig. 3C), demonstrates that an adult developmental context is required for leukemia after Pten deletion. We performed similar experiments in mice with a p53 heterozygous background and obtained similar results (Fig. 3B, D).
To test whether Pten is required in neonatal HSCs to regulate PI3-kinase pathway signaling we administered pIpC to Mx1-Cre; Ptenfl/fl mice or littermate controls at 2 days or 6 weeks after birth. We isolated CD48−LSK cells two weeks later and performed Western blots. The total levels of PTEN, AKT, S6 and GSK3β were similar in neonatal and adult HSCs/MPPs from control mice (Fig. 4A, B). In both the neonatal and adult HSCs/MPPs we observed a complete loss of PTEN protein from Mx1-Cre; Ptenfl/fl cells (Fig. 4A, B). In adult HSCs/MPPs, Pten deletion increased the phosphorylation of AKT, S6 and GSK3β, but not MAPK or AMPK (Fig. 4A, B). In neonatal HSCs/MPPs, Pten deletion had little or no affect on AKT, S6 and GSK3β phosphorylation (Fig. 4A, B). Pten deletion did not increase AKT phosphorylation in HSCs/MPPs until 3 to 4 weeks after birth (Fig. 4C), the same time we began to observe increased HSC proliferation after Pten deletion (Fig. 2B).
To determine whether neonatal HSCs/MPPs are capable of activating the PI3-kinase pathway, we incubated neonatal or adult HSCs/MPPs from control and Mx1-Cre; Ptenfl/fl mice in culture with 2% fetal bovine serum at 37°C for 30 minutes. In contrast to uncultured neonatal Pten deficient HSCs/MPPs, phosphorylated AKT was present at comparable levels in neonatal versus adult Pten deficient HSCs/MPPs (Fig. 4D). Neonatal HSCs are therefore capable of phosphorylating AKT when exposed to serum in culture and PTEN negatively regulates this process. Several PI3K pathway proteins, including AKT, S6, RICTOR, mTOR, RAPTOR and SHIP1, are expressed at similar levels in neonatal and adult HSCs (Fig. 4A, B, E). Phosphorylation of mTOR at Serine 2481, which reflects mTORC2 function (Copp et al., 2009), was also similar in neonatal and adult HSCs/MPPs (Fig. 4F). A number of other gene products known to regulate the PI3-kinase pathway were expressed at similar levels in neonatal and adult HSCs by quantitative RT-PCR (Fig. S2A). Neonatal HSCs/MPPs are therefore capable of activating the PI3-kinase pathway, but PTEN is not required in neonatal HSCs/MPPs in vivo to negatively regulate pathway activation.
Neonatal restricted progenitors (CD48+LSK cells) and unfractionated bone marrow cells did exhibit increased AKT phosphorylation following Pten deletion (Fig. 4G). Pten deletion also increased AKT and S6 phosphorylation in lymphoid primed multipotent progenitors (Lineage−c-kit+Sca-1+CD34+FLT3+ cells; LMPPs) and GMPs from neonatal and adult mice (Fig. 4H). In contrast, adult CD4+CD8+ thymocytes had much higher levels of AKT and S6 phosphorylation after Pten deletion than neonatal CD4+CD8+ thymocytes (Fig. 4I). As in prior studies (Yilmaz et al., 2006; Zhang et al., 2006), we observed an increase in myelopoiesis and a decrease in B lymphopoiesis in adult bone marrow after Pten deletion. We also observed an increase in myelopoiesis and a decrease in B lymphopoiesis in the bone marrow and spleen of neonatal mice after Pten deletion (Fig. 4J, K). These data demonstrate that many restricted hematopoietic progenitors in neonatal mice depend upon PTEN to regulate cellular signaling and function, just as observed in adult mice, but that HSCs and T lineage progenitors undergo a temporal change in their dependence upon PTEN between the neonatal and adult stages.
To further explore the differences in PI3-kinase pathway regulation between neonatal and adult HSCs we compared the gene expression profiles of P14 neonatal HSCs and 8 week-old adult HSCs from wild-type mice. We identified 14 genes that were significantly more highly expressed in neonatal HSCs, and 21 genes that were significantly more highly expressed in adult HSCs (Table S2; fold change>2.0). Ten of these genes could potentially regulate PI3-kinase pathway signaling based on prior studies. The differential expression of 8 of these genes was confirmed by quantitative RT-PCR (Fig. S2B, Table S3). For example, Bank1 was expressed more highly in neonatal relative to adult HSCs (Fig. S2B) and is known to negatively regulate AKT phosphorylation (Aiba et al., 2006). This provides candidate genes that could potentially explain the difference in PTEN dependence between neonatal and adult HSCs but gene-targeted mice will have to be generated to functionally test these hypotheses.
We next assessed the mechanisms by which Pten deletion promotes HSC depletion and leukemogenesis in adult mice. Pten deletion can activate both mTORC1 and mTORC2 (Guertin and Sabatini, 2007); however, mTORC1 promotes proliferation by phosphorylating S6kinase and 4EBP1 while mTORC2 phosphorylates AKT at Ser473 (Guertin et al., 2006; Guertin and Sabatini, 2007). Since increased AKT phosphorylation at Ser473 distinguished Pten deficient HSCs from HSCs that were dividing under physiological conditions (Fig. 1C) we tested whether the effects of Pten deficiency were mediated by increased mTORC2 activation. We generated a floxed allele that allowed us to conditionally delete Rictor (Fig. S3), a key component of the mTORC2 complex (Guertin et al., 2006), and we generated Mx1-Cre; Rictorfl/fl mice. We administered pIpC 6 weeks after birth and observed an almost complete loss of RICTOR protein from Mx1-Cre; Rictorfl/fl LSK cells, but mTOR was still expressed (Fig. 4L). We confirmed that both Rictor alleles were recombined in all HSCs tested (Fig. S4A).
To determine whether Rictor deletion attenuates AKT phosphorylation in HSCs/MPPs, we administered pIpC to control, Mx1-Cre; Ptenfl/fl, Mx1-Cre; Rictorfl/fl and Mx1-Cre; Ptenfl/fl; Rictorfl/fl mice 6 weeks after birth. As expected, Pten deletion increased AKT phosphorylation at Ser473 (Fig. 4M). Rictor deletion substantially reduced AKT Ser473 phosphorylation in HSCs/MPPs, to levels lower than observed in control HSCs/MPPs, even when Pten and Rictor were simultaneously deleted (Fig. 4M). This result is consistent with the expectation that Rictor deletion reduces mTORC2 function.
To further evaluate RICTOR regulation of the PI3-kinase pathway, we harvested LSK cells from 8 week old control, Mx1-Cre; Ptenfl/fl, Mx1-Cre; Rictorfl/fl and Mx1-Cre; Ptenfl/fl; Rictorfl/fl mice, 2 weeks after pIpC treatment. These cells were incubated for 30 minutes in Iscove’s medium + 2% fetal bovine serum before being harvested for western blotting. In parallel, control and Pten deficient LSK cells were incubated with 250 nM Torin1, an active site inhibitor of mTOR that inhibits both mTORC1 and mTORC2 kinase activity (Thoreen et al., 2009). Pten deletion increased AKT phosphorylation, and Rictor deletion or Torin1 treatment eliminated AKT phosphorylation, both in wild-type and Pten-deficient LSK cells (Fig. 4N). We also observed reduced AKT phosphorylation at Thr308 in Mx1-Cre; Ptenfl/fl; Rictorfl/fl LSK cells relative to Mx1-Cre; Ptenfl/fl LSK cells, and Torin1 treatment completely blocked AKT phosphorylation at Thr308 even though this site is not thought to be an mTORC1 or mTORC2 substrate (Fig. 4N). A similar decrease in Thr308 phosphorylation was previously observed in Rictor deficient prostate epithelium (Guertin et al., 2009). These results confirm that Rictor deletion reduces mTORC2 function and AKT phosphorylation in primitive hematopoietic progenitors.
To test if RICTOR is required by hematopoietic cells we analyzed Vav-Cre; Rictorfl/fl mice and Mx1-Cre; Rictorfl/fl mice. Vav-Cre expresses in HSCs during early embryonic development (de Boer et al., 2003). Vav-Cre; Rictorfl/fl mice were born at Mendelian ratios and they matured to adulthood without any gross abnormalities (data not shown). We did not observe significant differences in bone marrow cellularity, spleen cellularity, spleen weight or thymus weight between Vav-Cre; Rictorfl/fl neonatal mice and littermate controls 14 days after birth (Fig. S4C, D). The frequency and absolute number of HSCs in the bone marrow of Vav-Cre; Rictorfl/fl neonatal mice was slightly but significantly elevated relative to littermate controls (Fig. 5A; Fig. S4E). We also deleted Rictor two days after birth by administering a single dose of pIpC to Mx1-Cre; Rictorfl/fl mice to achieve complete recombination of the Rictorfl alleles (Fig. S4B). At P14 we did not observe significant differences in the frequency or absolute number of bone marrow HSCs between Mx1-Cre; Rictorfl/fl mice and littermate controls (Fig. 5A; Fig S4E). Rictor is therefore not necessary for the formation of HSCs or for fetal/neonatal hematopoiesis.
To determine whether Rictor regulates the proliferation of neonatal HSCs we evaluated the cell cycle status of HSCs isolated at P14 from Vav-Cre; Rictorfl/fl mice and Mx1-Cre; Rictorfl/fl mice that received pIpC at P2. We did not observe significant differences between control and Rictor deficient HSCs in the frequency of HSCs in S/G2/M phases of the cell cycle in either genetic background (Fig. 5B). We also observed no significant difference in the BrdU incorporation rate of Rictor deficient and control HSCs (Fig. 5C). Rictor is therefore not required to regulate the maintenance or cell cycle of neonatal HSCs.
To determine whether Rictor regulates adult hematopoiesis or HSC function, we administered pIpC to adult Mx1-Cre; Rictorfl/fl mice and littermate controls 6 weeks after birth. These mice were observed for 18-24 weeks following pIpC treatment and they did not develop any overt signs of illness (data not shown). After the observation period, we detected a small, but significant, decrease in white blood cell counts in Rictor deleted mice (Fig. S4F), but red blood cell counts (Fig. S4G), platelet counts (Fig. S4H), and thymus cellularity (Fig. S4I) were not significantly affected by Rictor deletion. We observed a trend toward reduced spleen cellularity in multiple experiments but the effects were not statistically significant (Fig. S4l). We also observed no significant difference in bone marrow cellularity, HSC frequency, or absolute HSC numbers (Fig. 5D-F). We did observe a significant reduction in the frequency of MPPs (CD150−CD48−LSK) after Rictor deletion (Fig. 5E, F). Rictor is therefore not required for the maintenance of adult HSCs but RICTOR may regulate MPPs.
Rictor deficient bone marrow cells gave long-term multilineage reconstitution of all recipient mice for at least 16 weeks after transplantation (Fig. 5G-K). We observed significantly reduced levels of donor lymphocyte reconstitution in recipients of Rictor deficient bone marrow cells over the first 12 weeks after transplantation, but this difference was not significant by 16 weeks after transplantation (Fig. 5I, J). We did not observe any differences in the frequency of donor HSCs or MPPs in recipients of Rictor-deficient or control cells 16 weeks after transplantation (Fig. 5L). We also did not observe significant differences in the reconstituting capacity of Rictor deficient and control bone marrow cells in secondary recipients (Fig. 5M, N). Rictor is therefore not necessary for HSC maintenance or function in normal adult mice, though its absence may lead to modest deficits in lymphoid reconstitution.
We administered pIpC to control, Mx1-Cre; Ptenfl/fl, Mx1-Cre; Rictorfl/fl, and Mx1-Cre; Ptenfl/fl; Rictorfl/fl mice 6 weeks after birth and analyzed HSCs two weeks later. Since rapamycin rescues the proliferation of Pten deficient HSCs (Yilmaz et al., 2006), we also administered rapamycin to some Mx1-Cre; Ptenfl/fl mice and controls for one week prior to analysis. Rapamycin did not appear to affect mTORC2 function in primitive hematopoietic progenitors as it did not affect AKT phosphorylation at Ser473 in Pten deficient LSK cells in vivo (7 days at 4mg/kg/day; Fig. 6A), or after 30 minutes of culture in rapamycin-containing medium (100nm; Fig. 6B). In contrast, Rictor deletion may modestly reduce mTORC1 activation as phosphorylated S6 levels were modestly reduced in Rictor-deficient as compared to control LSK cells and phosphorylated S6 levels were reduced as a fraction of total S6 levels in Mx1-Cre; Ptenfl/fl; Rictorfl/fl LSK cells relative to Mx1-Cre; Ptenfl/fl LSK cells (Fig. 4N). A decline in mTORC1 signaling after Rictor deletion may be expected as mTORC2 signaling increases AKT signaling and therefore mTORC1 activation (Guertin and Sabatini, 2007; Guertin et al., 2009). In primitive hematopoietic progenitors rapamycin thus appears to reduce mTORC1 function without affecting mTORC2 while Rictor deletion appears to eliminate mTORC2 function while modestly reducing mTORC1 function.
Pten deletion significantly increased the fraction of HSCs in S/G2/M phases of the cell cycle and rapamycin rescued this effect (Fig. 6C). Rictor deletion by itself did not significantly affect the frequency of HSCs in S/G2/M phases of the cell cycle but did significantly reduce the fraction of Pten deficient HSCs in S/G2/M phases of the cell cycle (Fig. 6C). Treating Pten/Rictor compound mutant mice with rapamycin did not further reduce HSC proliferation. Rictor deletion also significantly reduced the mobilization of Pten deficient HSCs but rapamycin had no effect (Fig. 6D). Taken together this suggests that mTORC2 mediates the effects of Pten deletion on HSC function through mTORC1-dependent and mTORC1-independent mechanisms, consistent with studies in other cell types (Guertin and Sabatini, 2007).
Rapamycin treatment, but not Rictor deletion, reduced the mobilization of wild-type HSCs from the bone marrow to the spleen after cyclophosphamide/G-CSF treatment (Fig. 6E, F). This suggests that mTORC1, not mTORC2, is the major mediator of HSC proliferation under physiological conditions in wild-type mice. This conclusion is consistent with the paucity of S473 phosphorylation of AKT (an mTORC2 site; (Guertin et al., 2006)) in fetal, neonatal, and cyclophosphamide/G-CSF mobilized HSCs (Fig. 1D-G) and with the lack of effect of Rictor deletion on HSC frequency or proliferation (Fig. 5). Pten deletion therefore induces the proliferation and mobilization of HSCs through mTORC2 dependent mechanisms that do not mimic physiological HSC self-renewal mechanisms.
To test if Rictor is necessary for the depletion of Pten-deficient HSCs, we administered pIpC to 6 week-old control, Mx1-Cre; Ptenfl/fl, Mx1-Cre; Rictorfl/fl and Mx1-Cre; Ptenfl/fl; Rictorfl/fl mice, then isolated HSCs 2 weeks later. We transplanted 10 CD45.2+ HSCs from donor mice into irradiated CD45.1+ mice along with 300,000 wild-type CD45.1+ bone marrow cells. As expected, control HSCs gave high levels of long-term multilineage reconstitution in all recipient mice and Pten-deficient HSCs gave only transient multilineage reconstitution in all recipient mice (Fig. 6G-K). Rictor deficient HSCs gave long-term multilineage reconstitution in 3 of 12 recipient mice. Most of the remaining mice had long-term myeloid reconstitution, suggesting that HSC activity was maintained in these mice, but they did not necessarily have long-term lymphoid reconstitution (Fig. 6G-K). In contrast to Pten-deficient HSCs, Pten; Rictor compound mutant HSCs gave long-term multilineage reconstitution in almost all recipient mice with levels of donor myeloid and T cell reconstitution that were at least as high as from control HSCs (Fig. 6G-K). Pten; Rictor compound mutant HSCs gave much lower levels of B lineage reconstitution than control HSCs (Fig. 6I), suggesting that Rictor deficiency largely rescued the function of Pten deficient HSCs but did not rescue the function of Pten deficient B lineage progenitors.
Secondary recipients of Rictor-deficient cells exhibited long-term multilineage reconstitution in most mice, but levels of donor cells were significantly lower than from control cells (Fig. 6L-P). Secondary recipients of Mx1-Cre; Ptenfl/fl; Rictorfl/fl cells also exhibited long-term multilineage reconstitution in most recipient mice, with levels of donor cell reconstitution that were intermediate between control and Rictor-deficient cells (Fig. 6L-P). Rictor deficiency therefore substantially rescued HSC function after Pten deletion, enabling Pten deficient HSCs to give long-term multilineage reconstitution of primary and secondary recipient mice.
To determine whether Rictor is necessary for the development of myeloproliferative disorder and leukemia following Pten deletion, we administered pIpC to control, Mx1-Cre; Ptenfl/fl, Mx1-Cre; Rictorfl/fl and Mx1-Cre; Ptenfl/fl; Rictorfl/fl mice 6 weeks after birth. We administered rapamycin for 7 days to some of these mice. We harvested spleens two weeks after the initial pIpC treatment. Spleens from Mx1-Cre; Ptenfl/fl mice were significantly more cellular than spleens of other genotypes, as expected (Fig. 7A). Spleens from Mx1-Cre; Ptenfl/fl; Rictorfl/fl and rapamycin treated Mx1-Cre; Ptenfl/fl mice were also significantly more cellular than control spleens, but they were significantly less cellular than spleens from Mx1-Cre; Ptenfl/fl mice (Fig. 7A). Rapamycin in Mx1-Cre; Ptenfl/fl; Rictorfl/fl mice further reduced spleen cellularity, rendering it similar to control spleens (Fig. 7A). Rictor deletion and rapamycin treatment therefore additively reduce the severity of myeloproliferative disorder after Pten deletion, suggesting that mTORC1 and mTORC2 both contribute to this phenotype (Fig. 7A). Nonetheless, extramedullary hematopoiesis was still histologically evident in the spleens of rapamycin-treated Mx1-Cre; Ptenfl/fl; Rictorfl/fl mice, suggesting that rapamycin treatment and Rictor deletion did not completely eliminate the myeloproliferative disorder (Fig. S5).
To determine whether Rictor is necessary for progression to acute leukemia following Pten deletion, we administered pIpC to control, Mx1-Cre; Ptenfl/fl, Mx1-Cre; Ptenfl/+; Rictorfl/fl, Mx1-Cre; Ptenfl/fl; Rictorfl/+ and Mx1-Cre; Ptenfl/fl; Rictorfl/fl mice 6 weeks after birth. All control and Mx1-Cre; Ptenfl/+; Rictorfl/fl mice survived for the 115 day duration of the experiment (n=7-11 mice/treatment; Fig. 7B) with no signs of leukemia (Fig. 7F, J). All Mx1-Cre; Ptenfl/fl mice died by 77 days after pIpC treatment (n=8, range 14-77 days; Fig. 7B) with T-ALL (Fig. 7G, K). All but one Mx1-Cre; Ptenfl/fl; Rictorfl/+ mouse died by 80 days after pIpC treatment (n=6, range 26-115 days; Fig. 7B) with T-ALL (data not shown). Nine of 12 Mx1-Cre; Ptenfl/fl; Rictorfl/fl mice survived for 115 days (N=12, range 19-115 days; Fig. 7B) with myeloproliferative disorder in the spleen (Fig. 7I) but no evidence of T-ALL in the spleen or thymus (Fig. 7I, M). Spleens from Mx1-Cre; Ptenfl/fl; Rictorfl/fl mice were larger than spleens from control or Mx1-Cre; Ptenfl/+; Rictorfl/fl mice (Fig. 7C, E). In contrast, thymuses from Mx1-Cre; Ptenfl/fl; Rictorfl/fl mice were not enlarged relative to control or Mx1-Cre; Ptenfl/+; Rictorfl/fl mice (Fig. 7D, E). The Mx1-Cre; Ptenfl/fl; Rictorfl/fl mice that died in this experiment did have T-ALL. However, genotyping of genomic DNA from one T-ALL revealed that the floxed Rictor allele had not been deleted in this clone of leukemia cells (data not shown). Rictor deletion thus significantly reduced the incidence of leukemia in mice after Pten deletion. PTEN is therefore required in adult, but not neonatal, hematopoietic cells to suppress leukemia by reducing mTORC2 activation.
The PI3-kinase pathway regulates adult HSC proliferation and leukemogenesis. Akt1/2 deletion increases adult HSC quiescence (Juntilla et al., 2010). Increased activation of the PI3-kinase pathway by Pten deletion, over-expression of Akt1, or Tsc2 deletion, promotes adult HSC proliferation and HSC depletion (Yilmaz et al., 2006; Zhang et al., 2006; Chen et al., 2008; Gan et al., 2008; Kharas et al., 2010; Lee et al., 2010). However, fetal and neonatal HSCs/MPPs did not exhibit increased AKT phosphorylation relative to adult HSCs/MPPs even though more fetal/neonatal HSCs/MPPs were in cycle (Fig. 1B-E). This suggests that while increased PI3-kinase pathway activation by Pten deletion can drive adult HSCs into cycle, sustained AKT phosphorylation is not observed in HSCs that divide frequently under physiological conditions. Consistent with this, Rictor deletion reduced AKT phosphorylation (Fig. 4M, N) but had only modest effects on the maintenance and function of HSCs (Fig. 5). Nonetheless, Rictor was required for increased HSC proliferation (Fig. 6C), HSC depletion (Fig. 6G-K) and leukemogenesis (Fig. 7) following Pten deletion. PTEN is therefore required in adult, but not neonatal, hematopoietic cells to maintain HSCs and to suppress leukemia by reducing mTORC2 activation. Rictor deletion also reduces the incidence of prostate cancer after Pten deletion, while having little effect on wild-type prostate epithelium (Guertin et al., 2009).
Our data indicate that mTORC1 and mTORC2 have distinct functions in dividing HSCs. While Rictor-deficient HSCs gave long-term multilineage reconstitution of irradiated mice (Fig. 5), Kalaitzidis et al. found that Raptor-deficient HSCs were unable to reconstitute irradiated mice (see companion manuscript). Fetal HSCs and cyclophosphamide/G-CSF mobilized HSCs exhibited increased mTORC1 but not mTORC2 activation relative to quiescent adult HSCs (Fig 1D, G). Rapamycin treatment, but not Rictor deletion, reduced HSC mobilization following cyclophosphamide/G-CSF treatment (Fig. 6F). These results demonstrate that mTORC1 is more important than mTORC2 for the proliferation of HSCs under physiological conditions.
In contrast, both mTORC1 and mTORC2 were activated by Pten deletion (Fig. 1D). Rictor deletion, but not rapamycin treatment, reduced HSC mobilization following Pten deletion (Fig. 6D). Rapamycin treatment and Rictor deletion both reduced HSC proliferation after Pten deletion (Fig. 6C) and additively reduced the severity of myeloproliferative disorder following Pten deletion (Fig. 7A). These results demonstrate that mTORC2 is a major mediator of the effects of Pten deletion on hematopoietic cells, but that mTORC2 likely signals through mTORC1–dependent and mTORC1-independent pathways. Therefore both mTORC1 and mTORC2 contribute to the hematopoietic phenotypes observed after Pten deletion, consistent with our previously published results (Yilmaz et al., 2006; Lee et al., 2010) and the results in the companion manuscript from Kalaitzidis et al.
Our data indicate that temporal changes in the regulation of the PI3-kinase pathway lead to temporal changes in the mechanisms that regulate HSC function and leukemogenesis. This likely explains why conditional deletion of Pten from fetal hematopoietic cells using VE-Cadherin-Cre does not lead to the development of leukemia until around 8 weeks after birth (Guo et al., 2008). It is not yet clear why Pten deletion increases PI3-kinase pathway activation in adult, but not neonatal, HSCs. Neonatal HSCs were capable of activating the PI3-kinase pathway when cultured (Fig. 4D), but exhibited a cell-autonomous difference relative to adult HSCs in their dependence upon PTEN in transplantation assays (Fig. 2E-H). We have identified a number of gene products that are differentially expressed between neonatal and adult HSCs and that could potentially influence PI3-kinase pathway signaling (Fig. S2). However, much more work will be required to assess their functions in neonatal and adult HSCs.
Since neonatal HSCs and other hematopoietic cells can harbor mutations in Pten and p53 without transforming into leukemia until adulthood, our data suggest that mutated cells may persist for years in children before a change in developmental context renders these mutations competent to induce leukemia. Prolonged persistence of cells with leukemogenic mutations has been documented in human pediatric leukemia patients (Wiemels et al., 1999a; Wiemels et al., 1999b). In the past, this latency has been attributed to the time necessary to accrue secondary mutations. However, our data suggest an additional explanation: that some of these mutations were not competent to cause leukemia until the blood cells matured into a susceptible developmental context. This could explain why T-ALL presents more commonly in older rather than younger children. Our findings raise the possibility of reprogramming therapies that are analogous to differentiation therapies. Reprogramming of adult hematopoietic cells to a fetal or neonatal identity (Kim et al., 2007; He et al., 2011) could blunt the effects of leukemogenic mutations that require an adult developmental context to induce neoplastic proliferation.
Ptenfl/fl mice, Mx1-Cre mice and p53−/− mice have been previously described (Jacks et al., 1994; Kuhn et al., 1995; Groszer et al., 2006). These mice were all backcrossed for at least 8 generations onto a C57BL/6Ka-Thy-1.1 (CD45.2) background. See Supplementary Methods for a description of the generation of Rictorfl allele. Expression of Mx1-Cre was induced by a single subcutaneous injection of pIpC (Amersham) in neonatal mice at P2 or by three intraperitoneal injections over six days in adult mice beginning 6 weeks after birth. pIpC dose was determined empirically for each lot so that >95% of HSC colonies exhibited complete recombination of the Ptenfl/fl and Rictorfl/fl alleles (Figs. S1, S4). Neonatal mice received 1-2 μg/dose, depending on pIpC lot. Adult mice received 10 μg/dose. All mice were housed in the Unit for Laboratory Animal Medicine at the University of Michigan (UM), where these studies were initiated, or at the Animal Resource Center at UT-Southwestern Medical Center, where the studies were completed. All animal procedures were approved by the UM and UT Southwestern Committees on the Use and Care of Animals.
Bone marrow cells and splenocytes were obtained, stained, and analyzed by flow cytometry as previously described (Lee et al., 2010; Nakada et al., 2011). Antibodies and detailed methods are in supplementary methods (Table S1).
BrdU (Sigma; diluted to 5-10 mg/ml in PBS) was administered by IP injections (100 mg/kg/dose) given every 8 hours beginning 24 hours prior to bone marrow harvest. HSCs were stained and enriched by c-kit selection as described in the Supplementary Methods, and BrdU incorporation was measured by flow cytometry using the APC BrdU Flow Kit (BD Biosciences). For cell cycle analysis, HSCs were stained with antibodies against CD150, CD48, lineage markers, c-kit and Sca-1. c-kit+ cells were enriched by selection with paramagnetic beads (Miltenyi Biotec, Auburn, CA). The enriched cells were fixed and permeablized with cytofix/cytoperm buffer (BD Biosciences) then washed in staining medium, and stained with DAPI (20 μg/ml). HSCs were analyzed by flow cytometry. For analysis of AKT phosphorylation in HSCs/MPPs in different phases of the cell cycle (Fig. 1E), Vybrant DyeCycle Violet Stain (Invitrogen) was used to analyze DNA content while preserving cell viability. A detailed protocol is provided in supplementary methods.
Twenty or thirty thousand CD48−LSK cells were sorted and then resorted (to ensure purity) into Trichoracetic acid (TCA), and the volume was adjusted to a final concentration of 10% TCA. LSK cells were sometimes incubated at 37°C in Iscove’s medium plus 2% fetal bovine serum for 30 minutes prior to protein extraction in 10% TCA. Extracts were incubated on ice for at least 15 minutes and centrifuged at 16,100×g at 4°C for 10 minutes. Precipitates were washed in acetone twice and dried. The pellets were solubilized in 9M urea, 2% Triton X-100, 1% DTT. LDS loading buffer (Invitrogen) was added and the pellet was heated at 70°C for 10 minutes. Samples were separated on Bis-Tris polyacrylamide gels (Invitrogen) and transferred to PVDF membrane (Millipore). Antibodies are listed in Supplementary Methods (Table S1). Western blotting was performed according to the protocol from Cell Signaling Technologies, and blots were developed with the SuperSignal West Femto chemiluminescence kit (Thermo Scientific). Blots were stripped (1% SDS, 25 mM glycine pH 2) prior to re-probing.
Competitive reconstitution experiments were performed as previously described (Lee et al., 2010; Nakada et al., 2011). A complete description of long-term reconstitution assays is provided in the supplementary methods.
After administering pIpC to mice as described above, the mice were monitored daily and morbid mice were euthanized. Spleen and thymus tissues were fixed in buffered 10% formalin and embedded in paraffin for sectioning and hematoxylin and eosin (H&E) staining. Bone marrow cytospin specimens were Wright-Giemsa stained (Sigma). J.A.M., a pediatric hematologist, reviewed all slides and bone marrow specimens.
This work was supported by the National Institute on Aging (R37 AG024945), the Howard Hughes Medical Institute, and the Cancer Prevention and Research Institute of Texas. J.A.M. was supported by a grant from the Pediatric Scientist Development Program of the Association of Medical School Pediatric Department Chairs and the National Institute of Child Health and Human Development. Torin1 was a gift of Nathaniel Gray (Harvard University). J.A.M. performed all experiments and participated in the design and interpretation of all experiments. T.I. and K.L.G. generated the Rictorfl allele. D.N. backcrossed the Rictorfl allele onto a C57BL/6Ka-Thy-1.1 background and assisted with reconstitution experiments. J.Y.L. assisted with the analysis of reconstitution experiments. S.J.M. participated in the design and interpretation of all experiments and wrote the paper with J.A.M.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.