|Home | About | Journals | Submit | Contact Us | Français|
Pten negatively regulates the phosphatidylinositol 3-kinase (PI3K) pathway and is required to maintain quiescent adult hematopoietic stem cells (HSCs). Pten has been proposed to regulate HSCs cell autonomously and non-cell autonomously, but the relative importance of each mechanism has not been directly tested. Furthermore, the cytokines that activate the PI3K pathway upstream of Pten are not well defined. We sought to clarify whether Pten cell autonomously or non-cell autonomously regulates HSC mobilization. We also tested whether Pten deficiency affects the HSC response to granulocyte colony-stimulating factor (G-CSF) and interferon-α (IFNα) since these cytokines induce HSC mobilization or proliferation, respectively. We show that Pten regulates HSC mobilization and expansion in the spleen primarily via cell-autonomous mechanisms. Pten-deficient HSCs do not require G-CSF to mobilize, although they are hyper-sensitized to even low doses of exogenous G-CSF. Pten-deficient HSCs are similarly sensitized to IFNα. Pten therefore modulates the HSC response to inflammatory cytokines.
The balance between hematopoietic stem cell (HSC) proliferation and quiescence is tightly regulated (He et al., 2009). Adult HSCs are usually quiescent, but their proliferation rates increase upon exposure to cytokines (Baldridge et al., 2010, Essers et al., 2009, Morrison et al., 1997, Schuettpelz et al., 2014), bacterial infections (Rodriguez et al., 2009), and mutations that hyper-activate mitogenic pathways such as the phosphatidylinositol 3-kinase (PI3K) and Ras pathways (Kharas et al., 2010, Van Meter et al., 2007, Yilmaz et al., 2006, Zhang et al., 2006). Transient proliferation maintains the HSC pool and supports hematopoiesis, but sustained proliferation impairs HSC function and reduces long-term self-renewal capacity (Baldridge et al., 2010, Essers et al., 2009, Kharas et al., 2010, Rodriguez et al., 2009, Schuettpelz et al., 2014, Van Meter et al., 2007, Yilmaz et al., 2006, Zhang et al., 2006). Adult HSCs therefore require mechanisms to promote quiescence.
The PI3K pathway regulates cell metabolism, survival, and proliferation (Luo et al., 2003), and it arbitrates the balance between adult HSC quiescence and proliferation (Lee et al., 2010, Magee et al., 2012, Yilmaz et al., 2006, Zhang et al., 2006). PI3K signal transduction increases in mouse HSCs after conditional Pten deletion (Lee et al., 2010, Magee et al., 2012, Yilmaz et al., 2006, Zhang et al., 2006). Pten-deficient HSCs proliferate at increased rates and mobilize to extramedullary organs such as the spleen. They are functionally impaired and only transiently reconstitute irradiated mice (Lee et al., 2010, Magee et al., 2012, Yilmaz et al., 2006, Zhang et al., 2006). Similar phenotypes have been observed with other PI3K-pathway-activating mutations, including Tsc1 deletion, Itpkb deletion, or constitutive AKT expression (Gan et al., 2008, Kharas et al., 2010, Siegemund et al., 2015). The effects of Pten deletion on HSC proliferation and function are mediated by the kinase mammalian target of rapamycin (mTOR) via two complexes, mTORC1 and mTORC2 (Kalaitzidis et al., 2012, Lee et al., 2010, Magee et al., 2012, Yilmaz et al., 2006). mTOR impairs self-renewal by inducing p16 and p53 expression (Lee et al., 2010) and by inducing aberrantly high rates of protein synthesis (Signer et al., 2014). Together, these findings show that Pten is crucial for inhibiting the PI3K/mTOR pathway in HSCs, and sustained activation of the pathway compromises HSC function.
Pten has been proposed to regulate HSCs cell autonomously and non-cell autonomously (Tesio et al., 2013, Yilmaz et al., 2006). Several studies have noted that Pten-deficient HSCs have elevated PI3K pathway activity, consistent with a cell-autonomous function (Kalaitzidis et al., 2012, Lee et al., 2010, Magee et al., 2012, Signer et al., 2014). However, Pten-deficient myeloid cells have recently been shown to express high levels of granulocyte colony-stimulating factor (G-CSF) and other pro-inflammatory cytokines that can mobilize HSCs and deplete the bone marrow HSC pool (Tesio et al., 2013). Csf3-deficient mice (null for G-CSF) had reduced HSC mobilization following conditional Pten deletion, suggesting that Pten expression in myeloid cells non-cell autonomously regulates HSCs by suppressing G-CSF production (Tesio et al., 2013). These cell-autonomous and non-cell-autonomous mechanisms are not mutually exclusive, but their relative importance to HSC regulation has not been directly tested. Furthermore, a primary conclusion by Tesio et al. was that Pten lacks any HSC-autonomous function, in contrast to prior conclusions (Yilmaz et al., 2006, Zhang et al., 2006).
Since Pten regulates many cellular processes that are thought to be relevant for HSC self-renewal—including PI3K signal transduction, metabolism, protein synthesis, and proliferation—we sought to clarify whether it functions primarily via cell-autonomous or non-cell-autonomous mechanisms. Furthermore, we tested whether Pten interacts with two cytokines that are known to induce HSC proliferation and mobilization: G-CSF and interferon-α (IFNα). We found that Pten regulates HSC mobilization and expansion in the spleen primarily via cell-autonomous mechanisms, although non-cell-autonomous effects were also observed. In our hands, Csf3 deletion did not impair Pten-deficient HSC mobilization, in contrast to prior findings (Tesio et al., 2013). However, Pten-deficient HSCs were hyper-sensitized to G-CSF such that even low doses of the cytokine caused a marked increase in spleen HSCs. Higher G-CSF doses almost completely depleted Pten-deficient HSCs from the bone marrow without further expanding the spleen HSC pool. Pten-deficient HSCs were also sensitized to mobilizing effects of IFNα. Both G-CSF and IFNα hyper-activated the PI3K pathway in Pten-deficient HSCs. Together, our data show that Pten regulates HSCs by acting downstream rather than upstream of inflammatory cytokines, and it modulates the HSC response to G-CSF and IFNα.
To test whether Pten cell autonomously or non-cell autonomously regulates HSC mobilization and expansion, we transplanted 300,000 Cre− control or Ptenf/f;Mx1-Cre bone marrow cells (CD45.2) and 300,000 wild-type competitor cells (CD45.1) into irradiated CD45.1 recipient mice (Figure 1A). Six weeks after the transplants we administered poly-inosine:poly-cytosine (pIpC) (3 doses, 10 μg/dose every other day) to delete Pten. We measured bone marrow and spleen CD45.2+ and CD45.1+ HSC (CD150+CD48−Lineage−Sca1+c-Kit+) frequencies 14 days after Cre induction. The bone marrow of recipient mice had similar numbers of donor CD45.2+ and competitor CD45.1+ HSCs irrespective of genotype (Figure 1B). In contrast, the spleens of Ptenf/f;Mx1-Cre recipients had a ~50-fold increase in donor HSCs after Pten deletion, relative to control recipients, while competitor spleen HSCs increased by only ~3-fold (Figure 1C). Pten-deficient myeloid cell (CD11b+Gr1+) frequencies increased in the spleens of recipient mice, but competitor myeloid cell frequencies did not change (Figure 1D). Pten therefore regulates expansion of the spleen HSC and myeloid populations primarily via cell-autonomous mechanisms, with only minor contributions from non-cell-autonomous mechanisms.
Pten has been shown to inhibit G-CSF expression in myeloid cells (Tesio et al., 2013), but it is not clear whether myeloid Pten expression is actually critical for suppressing HSC mobilization. To test this, we deleted Pten in myeloid cells with Lysozyme M-Cre (Lyz-Cre), and we measured bone marrow and spleen HSC frequencies at 8–10 weeks after birth. As positive controls we measured bone marrow and spleen HSCs in pIpC-treated Ptenf/f;Mx1-Cre mice (which lack Pten in all hematopoietic cells) and G-CSF-treated wild-type mice (3.8 μg/mouse twice daily for 7 days). Ptenf/f;Lyz-Cre mice had only modest increases in spleen weights, spleen HSC frequencies, and spleen HSC numbers relative to controls. They had no change in bone marrow HSC frequencies despite efficient, myeloid-specific Pten deletion (Figures 1E–1G and S1). In contrast, Ptenf/f;Mx1-Cre and G-CSF-treated mice had much larger increases in spleen HSC numbers and a concomitant reduction in bone marrow HSCs (Figures 1F and 1G). These findings again show that the non-cell-autonomous effects of Pten deletion on HSCs are small when compared with the cell-autonomous effects.
G-CSF has been proposed as a key mediator of Pten-deficient HSC mobilization, but systemic G-CSF levels have not been previously measured in Pten-deleted mice. To better understand how G-CSF levels change following Pten deletion, we gave pIpC to 6-week-old control and Ptenf/f;Mx1-Cre mice (3 doses, 10 μg/dose every other day), and we measured serum G-CSF levels 14 days later. As a positive control, we administered lipopolysaccharide (LPS) to wild-type mice as previously described (Boettcher et al., 2014). LPS has been shown to induce G-CSF expression in endothelial cells (Boettcher et al., 2014, Burberry et al., 2014). We found that Ptenf/f;Mx1-Cre mice had significantly elevated G-CSF levels relative to controls, but the levels were much lower than those observed in LPS-treated mice (Figure 2A). This raised the question of whether low G-CSF levels are necessary and/or sufficient to mobilize Pten-deficient HSCs.
To test whether G-CSF is necessary to mobilize Pten-deficient HSCs, we administered pIpC to: (1) Csf3+/−, (2) Csf3−/−, (3) Ptenf/f;Csf3+/−;Mx1-Cre, and (4) Ptenf/f;Csf3−/−;Mx1-Cre littermates. After 14 days we measured bone marrow and spleen myeloid cell frequencies and HSC numbers. As expected, Ptenf/f;Csf3−/−;Mx1-Cre (G-CSF null) mice had significantly lower bone marrow and spleen myeloid cell frequencies as compared with Ptenf/f;Csf3+/−;Mx1-Cre (G-CSF heterozygous) mice (Figures 2B and S2A), consistent with the established role for G-CSF in myelopoiesis. In contrast, Ptenf/f;Csf3+/−;Mx1-Cre and Ptenf/f;Csf3−/−;Mx1-Cre had similar bone marrow and spleen HSC numbers following Pten deletion, and spleen HSC numbers were greatly expanded in both genotypes of mice relative to Cre− controls (Figures 2C and S2B). Thus, G-CSF is not required to mobilize Pten-deficient HSCs to the spleen.
We next tested whether Pten-deficient HSCs are hyper-sensitive to the mobilizing effects of G-CSF, even though G-CSF is not absolutely required for mobilization. We deleted Pten from 6-week-old mice, and 2 weeks later we began a 7 day course of vehicle, low-dose (0.5 μg/day) or high-dose (5 μg/day) G-CSF. We then measured bone marrow and spleen HSC numbers. Low-dose G-CSF treatment caused a dramatic increase in spleen HSC numbers in Pten-deleted mice (Figure 2D). High-dose G-CSF treatment almost completely depleted the bone marrow HSC pool in Pten-deleted mice without further increasing spleen HSC numbers (Figures 2D and 2E). These effects of G-CSF were far more severe than those observed in wild-type mice (Figures 2D and 2E). Interestingly, Rictor deletion did not impair HSC mobilization at either G-CSF dose (Figures 2D and 2E). Thus, Pten modulates the HSC response to G-CSF, but the PI3K pathway (or at least mTORC2) is not required for physiologic HSC mobilization.
G-CSF induces physiologic HSC mobilization by binding the G-CSF receptor on monocyte-macrophage lineage cells that, in turn, modulate stromal levels of CXCL12 (Christopher et al., 2011, Day et al., 2015, Liu et al., 2000). We therefore tested whether Pten-deficient myeloid cells could indirectly hyper-sensitize HSCs to G-CSF. We administered vehicle, low-, and high-dose G-CSF to Ptenf/f;Lyz-Cre adult mice for 7 days, and we measured bone marrow and spleen HSC numbers. HSCs in Ptenf/f;Lyz-Cre mice were only modestly sensitized to G-CSF as evidenced by a small increase in spleen HSC numbers after high-dose treatment, relative to wild-type controls (Figure 2D). This change contrasted with the much greater effects of low- and high-dose G-CSF on Ptenf/f;Mx1-Cre HSCs (Figures 2D and 2E). These findings suggest that: (1) low concentrations of G-CSF can stimulate HSC mobilization and extramedullary expansion, (2) these effects are normally suppressed by Pten, and (3) the mechanisms are independent of PI3K pathway regulation in myeloid cells.
Pten-deficient HSCs proliferate in the spleen more than G-CSF-mobilized, wild-type HSCs (Figures S2C and S2D). The marked expansion of Pten-deficient spleen HSCs that we observed after G-CSF treatment could reflect increased extramedullary proliferation as opposed to increased mobilization. To better assess mobilization without the confounding effects of extramedullary proliferation, we splenectomized 5-week-old wild-type and Ptenf/f;Mx1-Cre mice. We deleted Pten 1 week later and administered vehicle or low-dose G-CSF for 5 days beginning at 8 weeks after birth (2 weeks after pIpC). We then measured peripheral blood HSC numbers by flow cytometry. Pten deletion and G-CSF treatment both independently and synergistically increased peripheral blood HSC numbers (Figure 2F). Pten deletion also synergistically enhanced HSC mobilization after treatment with the CXCR4 inhibitor AMD3100 (Figure 2G). Thus, Pten suppresses HSC mobilization in response to both low-dose G-CSF treatment and CXCR4 inhibition.
To better understand why Pten-deficient HSCs are hyper-sensitive to G-CSF, we analyzed PI3K pathway activity in wild-type and Pten-deficient HSCs and multipotent progenitors (HSC/MPPs; CD48−LSK) after 2 days of treatment with vehicle, low-, or high-dose G-CSF. We assessed phosphorylation of AKT (an mTORC2 target) and S6 (an mTORC1 target) by western blotting. In wild-type HSC/MPPs, G-CSF had no effect on AKT or S6 phosphorylation at either dose (Figure 2H). In Pten-deficient HSC/MPPs, low-dose G-CSF increased AKT phosphorylation relative to untreated Pten-deficient HSC/MPPs, and high-dose G-CSF markedly increased S6 phosphorylation (Figure 2H). Thus, G-CSF hyper-activates the PI3K/mTOR pathway in Pten-deficient HSCs but not wild-type HSCs.
We tested whether Pten modulates the expression of surface proteins that are known to regulate HSC homing including CXCR4, VLA-4, and CD44 (Avigdor et al., 2004, Rettig et al., 2012). We also characterized expression of MT1-MMP, a matrix metalloproteinase that is expressed in G-CSF-stimulated HSCs in an mTOR-dependent manner and that promotes mobilization (Shirvaikar et al., 2010, Vagima et al., 2009). Whereas G-CSF treatment caused a modest increase in CXCR4 and CD44 expression as determined by flow cytometry (Figure 2J), Pten deletion had no effect on CXCR4, VLA-4, or CD44 expression (Figure 2I). Likewise, MT1-MMP expression was elevated in G-CSF-mobilized HSCs, consistent with prior studies (Shirvaikar et al., 2010, Vagima et al., 2009), but it was not changed in Pten-deficient HSCs (Figure 2K).
Our data suggest that G-CSF mobilizes and expands Pten-deficient spleen HSCs through mechanisms that are distinct from its normal physiologic activity. HSCs likely require Pten to keep low levels of G-CSF and other inflammatory cytokines from hyper-activating the PI3K/mTOR pathway. This observation is consistent with prior data showing that mobilization and extramedullary expansion of Pten-deficient HSCs requires mTORC2 (Magee et al., 2012), while physiologic mobilization does not (Figure 2D).
In light of the observed interaction between Pten and G-CSF, we tested whether Pten also modulates the HSC response to IFNα. This hypothesis was based on prior data showing that IFNα induces AKT phosphorylation in less-pure progenitor populations (Essers et al., 2009) and recognition of the fact that interferon signaling, like Pten deletion, drives HSCs into cycle and depletes the HSC pool (Baldridge et al., 2010, Walter et al., 2015). Furthermore, Ptenf/f;Mx1-Cre mice (which require pIpC-induced IFNα signaling to delete the Pten allele) have been shown to have more severe HSC phenotypes than tamoxifen-inducible Ptenf/f;Scl-CreER mice (Tesio et al., 2013). This raised the question of whether pIpC treatment contributes to the phenotypes observed in Ptenf/f;Mx1-Cre mice.
We first tested whether pIpC activates the PI3K pathway in wild-type and Pten-deficient HSCs. We gave three doses of pIpC (10 μg/dose) to 6-week-old wild-type and Ptenf/f;Mx1-Cre mice to delete Pten. Two weeks later, we treated the mice with vehicle or a single additional 10 μg dose of pIpC to re-induce IFNα. We isolated HSC/MPPs 24 hr later and performed western blots to assess PI3K pathway activity. HSC/MPPs from wild-type mice that received the additional pIpC dose (hereafter called pIpC-treated) had high AKT and S6 phosphorylation relative to controls that did not receive the additional dose (Figure 3A). S6 phosphorylation was even higher in pIpC-treated, Pten-deficient HSC/MPPs (Figure 3A). STAT1, a known IFNα target, was also more highly phosphorylated in pIpC-treated HSC/MPPs, consistent with prior studies (Essers et al., 2009). These data show that IFNα induces PI3K/mTOR signaling in HSCs and that its effect on mTORC1/S6K signaling is greater in Pten-deficient HSCs than in wild-type HSCs.
We next devised a strategy to delete Pten without pIpC so that we could characterize the interaction between Pten and IFNα independently of Cre induction. We isolated HSCs from E16.5 Ptenf/f mice and treated half of the cells with Tat-Cre for 30 min ex vivo to delete Pten (Figure 3B). We transplanted control or Tat-Cre-treated HSCs into irradiated recipients (350 HSCs and 300,000 competitor bone marrow cells per recipient). Unlike adult HSCs, Pten-deficient fetal HSCs were able to reconstitute, consistent with previously published findings (Magee et al., 2012). Six weeks after the transplants we administered pIpC to half of the mice (three doses, 10 μg/dose every other day). Two weeks later we measured spleen cellularity and HSC frequencies. pIpC caused a significant increase in spleen cellularity and spleen HSC frequency in recipients of Tat-Cre-treated HSCs, but bone marrow HSC frequencies did not differ between the treatment groups (Figures 3C and 3D). Spleen HSC numbers were significantly increased in recipients of Tat-Cre-treated HSCs, and this value further increased after pIpC treatment (Figure 3E). Thus, Pten modulates the HSC response to IFNα, much like it modulates the response to G-CSF.
We have conducted experiments to clarify whether Pten regulates HSCs cell autonomously or non-cell autonomously. Our data confirm that Pten regulates HSCs via both cell-autonomous and non-cell-autonomous mechanisms (Figure 1). The data also show that the cell-autonomous effects of Pten deletion are significantly greater than the non-cell-autonomous effects (Figure 1). While our analyses of Ptenf/f;Csf3−/−;Mx1-Cre mice do not support an obligate role for G-CSF in HSC mobilization (Figure 2C), we have found that Pten-deficient HSCs are hyper-sensitive to the mobilizing and depleting effects of G-CSF (Figures 2D and 2E). This hyper-sensitivity does not appear to reflect enhancement of normal physiologic mobilization mechanisms. Rather, G-CSF enhances PI3K/mTOR pathway activation in HSCs that lack Pten (Figure 2H). This could occur through direct engagement of the G-CSF receptor or through indirect activation of other cytokines in the microenvironment. Pten-deficient HSCs also hyper-activate mTORC1/S6K and mobilize in response to IFNα (Figure 3). Together, these data suggest that Pten modulates the HSC response to inflammatory cytokines, and pIpC contributes to the HSC proliferation, mobilization, and self-renewal phenotypes that have been widely described using Ptenf/f;Mx1-Cre mice (Kalaitzidis et al., 2012, Lee et al., 2010, Magee et al., 2012, Signer et al., 2014, Yilmaz et al., 2006, Zhang et al., 2006).
Our data potentially link two major modes of HSC regulation—the PI3K pathway and pro-inflammatory cytokines—as common causes of HSC depletion during illness or aging. The consequences of PI3K pathway activation in HSCs have been extensively studied (Gan et al., 2008, Kharas et al., 2010, Lee et al., 2010, Magee et al., 2012, Siegemund et al., 2015, Signer et al., 2014), but the signals that activate the pathway—either in native or stressed conditions—have not been well characterized. Likewise, inflammatory signals have drawn scrutiny for their putative role in bone marrow failure, HSC aging, and pre-leukemic clonal evolution (Baldridge et al., 2010, Baldridge et al., 2011, Essers et al., 2009, Walter et al., 2015), but the downstream mechanisms by which these signals deplete HSCs have not been fully resolved. Our data suggest a model in which inflammatory cytokines hyper-activate the PI3K pathway in HSCs leading to increased protein synthesis and tumor-suppressor expression, which ultimately depletes the HSC pool. Ongoing experiments will test whether mTORC1 or mTORC2 inactivation can preserve the function of cytokine-stimulated HSCs. If so, mTOR inhibitors may have a role in sustaining the HSC pool in patients with inflammation and otherwise tenuous HSC function (e.g., bone marrow transplant patients with complicating graft-versus-host disease or infections).
The Ptenf/f, Rictorf/f, Mx1-Cre, and Lyz-Cre strains have been described previously and are available from The Jackson Laboratory (Clausen et al., 1999, Groszer et al., 2006, Kuhn et al., 1995, Magee et al., 2012). Expression of Mx1-Cre was induced by three intraperitoneal injections of pIpC (GE Life Sciences; 10 μg/dose) over 5 days beginning 6 weeks after birth. Recombinant human G-CSF (Amgen) was given subcutaneously at the doses noted in the text and figure legends. AMD3100 (Genzyme) was given subcutaneously at 5 mg/kg 2 hr before euthanizing the mice, as previously described (Devine et al., 2008). All mice were housed in the Department for Comparative Medicine at Washington University. All animal procedures were approved by the Washington University Committees on the Use and Care of Animals.
Serum G-CSF levels were measured with the mouse G-CSF Quantikine ELISA Kit (R&D Systems). For positive controls, mice were injected with 35 μg of LPS at 72 and 24 hr prior to sacrifice, as described previously (Boettcher et al., 2014).
Fetal Ptenf/f HSCs were sorted into 0.1 ml of Iscove’s modified Dulbecco’s medium with 2% fetal bovine serum and incubated with 30 units of Tat-Cre (Millipore) for 30 min at 37°C. A complete description of the optimization protocol is provided in the Supplemental Experimental Procedures (Figure S3).
S.N.P., A.S.C., J.V., and J.A.M. performed all experiments except the analysis of serum G-CSF levels and cell-cycle assays (performed by D.A.M. and L.G.S.) and analyses of Pten/Csf3 compound mutant mice (performed by R.A.J.S.). S.N.P., A.S.C., R.A.J.S., L.G.S., and J.A.M. designed experiments and interpreted the data. J.A.M. directed the study and wrote the manuscript.
This work was supported by grants from the Children's Discovery Institute of Washington University and St. Louis Children's Hospital (L.G.S. and J.A.M.), the St. Baldrick's Foundation (J.A.M.), and the Department of Defense (CA130124, J.A.M.). A.S.C. is supported by a training grant to the Department of Pediatrics (5T32HD043010-12). J.A.M. and L.G.S. are scholars of the Child Health Research Center for Excellence in Developmental Biology at Washington University (K12-HD076224). Analyses of Pten/Csf3 compound mutant mice were conducted in Sean Morrison's laboratory. We thank John Dipersio and Julie Ritchey for providing AMD3100 and for performing the splenectomies.
Published: May 12, 2016
Supplemental Information includes Supplemental Experimental Procedures and three figures and can be found with this article online at http://dx.doi.org/10.1016/j.stemcr.2016.04.008.