Nucleophosmin Mediates Mammalian Target of Rapamycin–Dependent Actin Cytoskeleton Dynamics and Proliferation in Neurofibromin-Deficient Astrocytes

Neurofibromatosis type 1 (NF1) is one of the most common autosomal dominant tumor predisposition syndromes, affecting approximately 1 in 3,000 people worldwide. Individuals with NF1 develop tumors involving both the peripheral and central nervous systems. Neurofibromas, the benign Schwann cell tumor for which the disorder is named, are seen in 85% to 95% of adults with NF1, whereas in children the most common tumor is an astrocytic neoplasm (glioma) involving the optic pathway (1). Although NF1-associated optic pathway gliomas (OPGs) typically exhibit low proliferative indices, they can be locally invasive and infiltrate into normal brain structures to result in significant morbidity.

The observation that individuals with NF1 are predisposed to develop astrocytic tumors suggests that the NF1 gene product neurofibromin is a critical regulator of astrocyte growth. In this regard, neurofibromin functions as a Ras GTPase-activating protein (GAP) to negatively regulate the Ras proto-oncogene by accelerating the conversion of Ras from a GTP-bound active form to a GDP-bound inactive form (2). Consistent with the role of neurofibromin as a Ras-GAP, loss of neurofibromin expression in human NF1-associated tumors as well as in cells derived from Nf1-deficient mice results in hyperactivation of Ras and downstream Ras targets, including mitogen-activated protein kinase, RAF, and Akt (3–6).

To study the role of neurofibromin in astrocyte growth regulation and glioma formation, our laboratory has developed several murine models of NF1-associated neurologic disease by conditionally inactivating Nf1 expression in brain astrocytes (7–9). Studies of these genetically engineered mouse models and derivative astrocytes have revealed that Nf1^−/− astrocytes have several unique biological properties. First, Nf1^−/− astrocytes exhibit increased cell proliferation both in vitro and in vivo (8, 10). This increase in astrocyte proliferation requires Ras-dependent hyperactivation of the mammalian target of rapamycin (mTOR), a serine/threonine protein kinase that regulates ribosomal biogenesis, protein translation, and cell growth/proliferation. In support of this finding, Nf1^−/− fibroblasts and malignant peripheral nerve sheath tumors derived from NF1 patients exhibit Ras-dependent mTOR pathway hyperactivation (11), indicating that mTOR is an important downstream effector of neurofibromin signaling. Second, Nf1^−/− primary astrocytes exhibit a 2-fold increase in cell motility that is associated with hyperactivation of the small GTPase Rac1 (9). Similarly, Nf1^−/− Schwann cells have enhanced chemokinetic and chemotactic migration compared with wild-type controls (12). Third, Nf1^−/− astrocytes show an 8-fold increase in protein synthesis rates associated with increased expression of ribosomal proteins, including the nuclear-cytoplasmic shuttling protein nucleophosmin (NPM; ref. 10). Consistent with a role for neurofibromin in regulating protein synthesis and ribosomal biogenesis, the increased protein translation rate in Nf1^−/− astrocytes is completely rescued by treatment with the mTOR inhibitor rapamycin.

In this study, we sought to define the molecular signaling events that underlie the increases in cell proliferation and cell motility characteristic of Nf1^−/− astrocytes. In addition to regulating astrocyte proliferation, we find that mTOR regulates actin stress fiber formation and astrocyte motility, indicating that mTOR is an important regulator of actin cytoskeleton dynamics in astrocytes. We show that Rac1 functions downstream of mTOR to regulate actin stress fiber formation, astrocyte motility, and astrocyte proliferation. We identify NPM as a critical target of mTOR signaling both in vitro and in vivo and show that mTOR/Rac1 signaling is required for the increased expression of NPM in Nf1^−/− astrocytes in vitro and in vivo. Moreover, we show that NPM regulates Nf1^−/− astrocyte actin stress fiber formation as well as Nf1^−/− astrocyte motility and proliferation in vitro. Together, these data identify a unique, mTOR-dependent signaling pathway that regulates actin cytoskeleton dynamics and cell proliferation in astrocytes and suggest that targeting these signaling intermediates may be useful for treating disorders characterized by mTOR pathway hyperactivation.

Mice. Nf1^f/f mice were generously provided by Dr. Luis Parada (University of Texas Southwestern, Dallas, TX) and maintained as continuous breeding colonies by intercrossing (13). To conditionally inactivate Nf1 in astrocytes and to generate the Nf1 optic glioma model (Nf1^f/mut;GFAP-Cre mice), Nf1^f/f mice were crossed with mice expressing Cre recombinase under the control of the human 2.2-kb glial fibrillary acidic protein (GFAP) promoter as described previously (7, 8). Mice were used in accordance with established animal studies protocols at the Washington University School of Medicine.

Primary astrocyte cultures. Murine neocortical astrocyte cultures were generated from postnatal day 2 Nf1^f/f pups as described previously (9) and maintained in astrocyte growth medium (DMEM containing 10% fetal bovine serum and antibiotics). This method generates cultures that consist of >97% GFAP-immunoreactive astrocytes (Supplementary Fig. S1A). To inactivate the Nf1 gene, Nf1^f/f astrocytes (passage 1) were treated with adenovirus (Ad5) encoding Cre recombinase (University of Iowa Gene Transfer Vector Core, Iowa City, IA). Wild-type astrocytes were treated identically with an adenovirus encoding LacZ. Four days after Cre adenovirus treatment, neurofibromin could no longer be detected by Western blot (data not shown; ref. 9). Astrocytes were analyzed at passage 2 (12–15 days after dissection).

Pharmacologic inhibitors. Rapamycin (Sigma) was used at a concentration of 10 nmol/L. Leptomycin B (Calbiochem) was used at a concentration of 20 ng/mL. All pharmacologic treatments were for 16 to 18 h unless otherwise indicated.

Cell motility. Transwell Boyden chambers with 0.8-μm membranes (Costar) were used to measure astrocyte motility (9). Matrigel (BD Biosciences) was added to the inside of the Transwell and allowed to solidify at room temperature. Astrocytes were plated in triplicate Transwells on the opposite side of the membrane and allowed to adhere for 1 h. The Transwells were rinsed in PBS and maintained in astrocyte growth medium. The astrocytes were allowed to migrate toward the Matrigel for 48 h followed by fixation of the membranes in 100% methanol for 30 min at −20°C. The membranes were stained with hematoxylin, rinsed, and dried before mounting on a glass slide for counting. For each membrane (three per condition), three representative areas were counted at ×40 magnification. Each experiment was repeated thrice with identical results.

Cell proliferation. Astrocytes were plated (60,000 cells per well) in 24-well dishes on glass coverslips and allowed to adhere for 24 h. Astrocytes were rinsed twice in serum-free DMEM and maintained in serum-free DMEM for 24 h before exposure to bromodeoxyuridine (BrdU; 10 μg/mL; Sigma) or [³H]thymidine (1 μCi/μL) for 16 h. Astrocytes exposed to BrdU were fixed for immunocytochemistry using antibodies against BrdU (1:200; Abcam) and GFAP (1:200; Sigma). [³H]thymidine incorporation was determined by scintillation counting as described previously (10).

Immunocytochemistry. Astrocytes (passage 2) were plated in triplicate on glass coverslips in 24-well plates (60,000 cells per well) in astrocyte growth medium. After 24 h, astrocytes were rinsed twice and incubated in serum-free medium for 18 to 24 h. Pharmacologic inhibitors were added as indicated. Astrocytes were fixed in 3.7% formaldehyde, permeabilized with 0.1% Triton X-100, and blocked in 10% goat serum/1% bovine serum albumin (BSA) in PBS at 37°C. To identify focal adhesions, astrocytes were incubated with a vinculin monoclonal antibody (1:2,500; Sigma) in 1% BSA in PBS overnight at 4°C. Alexa Fluor 568–conjugated anti-mouse IgG secondary antibody (1:1,000; Molecular Probes) was used for detection. To identify the actin cytoskeleton, astrocytes were incubated with BODIPY-phalloidin or phalloidin-Alexa Fluor 568 (Molecular Probes) for 20 min in the dark. GFAP staining was done using a rabbit polyclonal antibody (1:500; Abcam). Nuclei were counterstained with 4′,6-diamidino-2-phenylindole. Representative photomicrographs were obtained using a fluorescence microscope (Nikon Eclipse TE300 inverted microscope) equipped with a digital camera (Optronics).

Western blotting. Astrocytes were serum starved overnight, harvested by scraping in PBS, and lysed in NP40 lysis buffer [0.5% NP40, 150 mmol/L NaCl, 50 mmol/L Tris (pH 7.6), 1 mmol/L DTT] with protease inhibitors. Western blotting was done as described previously (14). All antibodies are from Cell Signaling Technology and used at a 1:1,000 dilution unless otherwise noted. Primary antibodies included Rac1 (clone 23A8; 1:500; Upstate Biotechnology), phosphorylated S6, S6, phosphorylated elongation factor 4E-binding protein 1 (4EBP1), 4EBP1, phosphorylated Akt, Akt, phosphorylated protein kinase Cα (PKCα), PKCα (1:500; Santa Cruz Biotechnology), mTOR, tubulin (Sigma), and NPM (Zymed). Horseradish peroxidase–conjugated secondary antibodies were purchased from Cell Signaling Technology and detection was accomplished by enhanced chemiluminescence (Pierce). Equal protein loading was confirmed with either nonphosphorylated mTOR pathway antibodies or tubulin (data not shown) with identical results.

Rac1 activation assay. GTP-bound Rac1 was measured using the Rac activation kit from Upstate Biotechnology according to the manufacturer's instructions. Briefly, astrocytes were lysed and incubated with PAK1-PBD–conjugated agarose beads. An aliquot of the lysate was saved for Western blotting to ensure equal protein loading. Beads were washed in lysis buffer containing protease inhibitors, boiled in Laemmli buffer, and separated on SDS-PAGE gels for Western blotting.

Retroviral constructs. The mTOR small interfering RNA (siRNA) was generated using the siXpress Human H1 PCR Vector System (Mirus; ref. 14). Briefly, double-stranded siRNA expression cassettes were generated by PCR using H1 promoter and gene-specific primers. Several gene-specific sequences were tested for their ability to block mTOR protein expression. In these experiments, we used a mTOR siRNA corresponding to nucleotides 139 to 152 of the Mus musculus mTOR gene (National Center for Biotechnology Information accession no. NM_02009; sequence 5′-CATCTAGCAACGTGAGCGTCCTGC-3′). A control expression cassette encoded the H1 promoter alone. The PCR expression cassettes were cloned into pCR2.1 (Invitrogen), digested with EcoRI, and cloned into the MSCV.IRES.GFP retrovirus.

Viral transduction. MSCV.IRES.GFP retroviruses were transfected into 293T packing cells with ψ helper DNA using LipofectAMINE 2000 (Invitrogen) as described previously (14). Astrocytes were passaged 7 to 8 days after dissection, treated with adenovirus (LacZ or Cre), and then transduced with filtered supernatant from the 293T cells collected over a 36-h period. Transduced astrocytes were evaluated by fluorescence microscopy (Nikon Eclipse TE300 inverted microscope). Typically, >90% of astrocytes exhibited green fluorescent protein expression, indicating efficient viral transduction. Each experiment was done two to three times with identical results.

In vivo rapamycin treatment and immunohistochemistry. Rapamycin (Calbiochem) was prepared in a vehicle containing 5.2% polyethylene glycol 400/5.2% Tween-80. Two-month-old Nf1^f/mut;GFAP-Cre mice were treated with daily injections (5 days/wk) of rapamycin (5 mg/kg) or vehicle alone in 100 μL total volume. After 2 weeks of treatment, the mice were perfused transcardially with 4% paraformaldehyde, and the optic nerves were dissected for paraffin embedding and sectioning. Slides were deparaffinized in xylene and subjected to microwave antigen retrieval. After washing and blocking steps, brain sections were incubated overnight with phosphorylated S6 (1:200) or NPM antibodies (1:1,000) followed by incubation with biotinylated secondary antibodies (1:200) at room temperature for 1 to 2 h. Immunoreactivity was visualized with the Vectastain avidin-biotin complex system and 3,3′-diaminobenzidine (Vector Laboratories). All sections were photographed with a digital camera (Optronics) attached to an inverted microscope (Nikon).

Increased cell proliferation and cell motility as well as decreased actin stress fiber formation in Nf1^−/− astrocytes are mediated by mTOR. Previous studies in our laboratory have shown that Nf1^−/− astrocytes exhibit a 2-fold increase in cell proliferation that is restored to wild-type levels by treatment with the mTOR inhibitor rapamycin (Fig. 1A; ref. 10). Under these conditions, there was no statistically significant effect of rapamycin treatment on wild-type astrocyte proliferation. In addition, Nf1^−/− astrocytes exhibit a 2-fold increase in cell motility and altered cell spreading, two processes that require proper organization of the actin cytoskeleton (8, 9, 15). Recently, mTOR has also been shown to modulate actin cytoskeleton dynamics (16, 17). Based on these observations, we sought to determine whether the increase in cell motility observed in Nf1^−/− astrocytes was also mediated by mTOR. Using a Boyden chamber assay, the increase in Nf1^−/− astrocyte motility was ameliorated by treatment with 10 nmol/L rapamycin (Fig. 1B). As before, no effect of rapamycin on wild-type astrocytes was seen. These data indicate that both increased astrocyte proliferation and motility are mediated by mTOR-dependent signaling in Nf1^−/− astrocytes.

To extend our findings to other cytoskeleton properties, we next examined stress fiber formation in Nf1^−/− astrocytes. Actin stress fibers are bundles of filamentous actin that extend through the cytoplasm and interact with membrane integrins at specialized regions of the cell membrane called focal adhesions to anchor the cell to the extracellular matrix (18). Actin stress fibers have been used as a model system to study actin cytoskeleton dynamics. To directly visualize the actin cytoskeleton, we used fluorophore-labeled phalloidin, a fungal toxin that specifically binds actin. Whereas the majority (80%) of wild-type astrocytes exhibited numerous actin stress fibers, only 20% of Nf1^−/− astrocytes had stress fibers (Fig. 1C). In addition, we performed immunocytochemistry for vinculin, a protein localized to focal adhesions (19). Consistent with a decrease in actin stress fibers, vinculin-positive, membrane-localized focal adhesions were nearly absent in Nf1^−/− astrocytes, whereas wild-type astrocytes had numerous focal adhesions (Fig. 1C; Supplementary Fig. S1B).

This stress fiber assay provided us with a tractable system to examine the effects of rapamycin on actin cytoskeleton dynamics as a function of time of treatment. In these studies, we treated wild-type and Nf1^−/− astrocytes with a low concentration of rapamycin (10 nmol/L) for various lengths of time (1–24 h). We observed a dramatic increase in actin stress fibers and vinculin-positive focal adhesions in Nf1^−/− astrocytes even with 1 h of rapamycin treatment (Fig. 1C). In contrast, rapamycin had no effect on stress fiber formation or focal adhesions in wild-type astrocytes.

To confirm that the effect of rapamycin on actin stress fiber formation was mediated by mTOR, we introduced a mTOR-specific siRNA into wild-type and Nf1^−/− astrocytes. Knockdown of mTOR expression was confirmed by Western blot (Fig. 1D). Similar to astrocytes treated with rapamycin, genetic inhibition of mTOR in Nf1^−/− astrocytes restored actin stress fiber formation (Fig. 1D). Together, these data indicate that mTOR regulates actin cytoskeleton dynamics and cell proliferation in Nf1^−/− astrocytes.

mTOR-dependent regulation of actin stress fiber formation occurs independently of 4EBP1 and S6 kinase. Because mTOR signaling underlies the hyperproliferation, increased cell motility, decreased actin stress fiber formation (Fig. 1), and increased protein translation (10) in Nf1^−/− astrocytes, we next sought to determine which mTOR downstream effectors are responsible for mediating these biological phenotypes. The best-studied mTOR effectors are ribosomal S6 kinase (S6K) and 4EBP1, proteins that regulate rates of protein translation. First, we examined the activation status of these effectors using phosphorylation-specific antibodies. Although the target of S6K, ribosomal S6, is robustly hyperactivated in Nf1^−/− astrocytes, there was no change in 4EBP1 phosphorylation at Thr³⁷/Thr⁴⁶ (Fig. 2A), two sites required for 4EBP1 inactivation and translation initiation (20). These data suggest that 4EBP1 is not an important mediator of the mTOR-dependent abnormalities in cell proliferation and actin cytoskeleton dynamics in Nf1^−/− astrocytes.

To directly examine the effect of S6K on actin stress fibers, we overexpressed human S6K in wild-type astrocytes. Although S6K overexpression in wild-type astrocytes resulted in increased S6 activation (see Fig. 5A), we observed no changes in actin stress fiber formation (Fig. 2B). Similarly, rapamycin had no effect on actin cytoskeleton organization in S6K-expressing astrocytes. Taken together, these data suggest that rapamycin-sensitive, mTOR-dependent regulation of the astrocyte actin cytoskeleton occurs independently of the mTOR effectors 4EBP1 and S6K.

Rac1, but not PKCα, is hyperactivated in Nf1^−/− astrocytes. Recently, PKCα has been implicated in mTOR-dependent regulation of the actin cytoskeleton (17). PKCα is regulated by phosphorylation at Thr⁵⁰⁰, Thr⁶⁴¹, and Ser⁶⁶⁰ and exhibits no activity without phosphorylation of all sites (21). Using a commercially available antibody specific for PKCα phosphorylated at Thr⁶⁴¹, we observed no changes in the PKCα phosphorylation in Nf1^−/− astrocytes compared with wild-type astrocytes (Fig. 3A). Furthermore, PKCα activation was not sensitive to rapamycin treatment in astrocytes.

The small GTPase Rac1 has also been shown to act downstream of mTOR and modulate actin stress fiber formation (16). Previous studies from our laboratory have shown that Nf1^−/− astrocytes have increased levels of active, GTP-bound Rac1 (9). To determine whether Rac1 hyperactivation was mediated by mTOR signaling in astrocytes, we treated wild-type and Nf1^−/− astrocytes with rapamycin (10 nmol/L) and assayed Rac1 activation. In these experiments, we found that rapamycin blocked Rac1 hyperactivation in Nf1^−/− astrocytes (Fig. 3B), indicating that Rac1 acts downstream of mTOR.

To determine whether Rac1 activation is required for the cytoskeleton abnormalities observed in Nf1^−/− astrocytes, we inhibited Rac1 signaling using a mutant form of Rac1 (Rac1^N17) that functions in a dominant-negative fashion to inhibit the activity of endogenous Rac1. We found that the expression of Rac1^N17 in Nf1^−/−, but not wild-type, astrocytes resulted in a dramatic increase in actin stress fiber formation (Fig. 3C). Similarly, Rac1^N17 expression also restored the increased motility observed in Nf1^−/− astrocytes to wild-type levels (Fig. 3C). Taken together, these data indicate that increased Rac1 activity resulting from rapamycin-sensitive mTOR hyperactivation underlies the actin cytoskeleton abnormalities observed in Nf1^−/− astrocytes.

Rac1 has also been shown to regulate cell proliferation in a variety of cell types (22, 23), and Rac1^−/− mouse embryonic fibroblasts (MEF) exhibit both impaired migration and cell proliferation (24). To determine whether Rac1 regulates cell proliferation in Nf1^−/− astrocytes, we expressed Rac1^N17 in wild-type and Nf1^−/− astrocytes and examined BrdU incorporation. Although expression of dominant-negative Rac1 had little effect on wild-type astrocytes, it restored proliferation of Nf1^−/− astrocytes to wild-type levels (Fig. 3D). Taken together, these data indicate that Rac1 is an important regulator of mTOR-dependent actin stress fiber formation, cell motility, and proliferation in astrocytes.

Expression of NPM is regulated by mTOR in Nf1^−/− astrocytes. Previous studies in our laboratory showed that NPM was one of several proteins that are highly expressed in Nf1^−/− astrocytes (10). In addition, NPM is overexpressed in a variety of neoplasms, including prostate, ovarian, and colon cancers. Moreover, Npm-deficient fibroblasts exhibit decreased proliferation (25). NPM is a nuclear export protein that interacts with nucleolar components of newly synthesized ribosomes to facilitate their transport from the nucleolus/nucleus to the cytoplasm and increase the rates of protein synthesis (26). Expression of a NPM mutant that is unable to shuttle from the nucleus to the cytoplasm blocks ribosomal export (26). In addition, treatment with rapamycin inhibits NPM protein expression in MEFs (27). In Nf1^−/− astrocytes, we also found that mTOR inhibition by rapamycin dramatically reduced NPM expression in Nf1^−/− astrocytes with no effect on wild-type astrocytes (Fig. 4A).

To determine whether mTOR also regulates NPM expression in vivo, we used a mouse model of NF1-associated optic glioma (Nf1^f/mut;GFAP-Cre; ref. 7). For these experiments, we treated 2-month-old Nf1^f/mut;GFAP-Cre with either rapamycin (5 mg/kg) or vehicle for 2 weeks (10 doses; n = 5 mice/group). We found that rapamycin treatment dramatically reduced mTOR pathway activation as evidenced by phosphorylated S6 immunohistochemistry (Fig. 4B). Moreover, NPM expression was also decreased by rapamycin treatment. Together, these data show that mTOR is a critical regulator of NPM expression in astrocytes both in vitro and in vivo.

Because S6K/S6 and Rac1 were the mTOR effectors activated in Nf1^−/− astrocytes, we next sought to determine whether either of these mTOR targets regulated NPM expression. When S6K was overexpressed in wild-type astrocytes, there was no change in NPM expression despite an increase in S6 phosphorylation (Fig. 5A). This is in stark contrast to NPM regulation in primary MEFs where S6K is a potent inducer of NPM protein expression (27) and highlights the potential differences in mTOR signaling mechanisms between fibroblasts and astrocytes. However, when Rac1^N17 was expressed in Nf1^−/− astrocytes, NPM expression was attenuated (Fig. 5B). No change in S6 or Akt (data not shown) activity was observed following Rac1^N17 expression in Nf1^−/− or wild-type astrocytes, again showing that Rac1 functions downstream of mTOR. To provide additional support for Rac1-mediated NPM regulation, we expressed a constitutively active Rac1 molecule (Rac1^V12) in wild-type astrocytes to mimic Rac1 hyperactivation resulting from Nf1 loss. In these experiments, Rac1^V12 induced increased NPM expression similar to Nf1^−/− astrocytes (Fig. 5C). No change in S6 activation was observed. Collectively, these results suggest that NPM functions downstream of mTOR and Rac1 via an S6K-independent mechanism.

NPM mediates alterations in actin cytoskeleton dynamics and cell proliferation in Nf1^−/− astrocytes. To determine whether NPM regulates actin stress fiber formation in Nf1^−/− astrocytes, we used two complementary approaches to disrupt NPM function. One of the major roles of NPM is to mobilize ribosomes from the nucleolus to the cytosol (26). To disrupt this process, we expressed a mutant form of NPM [NPM double leucine mutant (NPMdL)] that prevents NPM shuttling from the nucleolus to the cytoplasm. Previous studies from our laboratory have shown that this mutant behaves like a true dominant-negative protein, forming hetero-oligomers with endogenous NPM and blocking endogenous NPM from shuttling from the nucleus to the cytoplasm (26). Expression of the NPMdL mutant rescued actin stress fiber formation in Nf1^−/− astrocytes but had no effect on wild-type astrocytes (Fig. 6A). Second, we took advantage of the observation that NPM export from the nucleolus/nucleus is dependent on its interaction with the nuclear export protein CRM1 (26). Leptomycin B is an antibiotic derived from Streptomyces that binds to CRM1 to specifically block CRM1-dependent nuclear export (28). When Nf1^−/− astrocytes were treated with 20 ng/mL leptomycin B, actin stress fiber formation was rescued (Fig. 6A). As before, no effect of leptomycin B treatment on wild-type astrocytes was observed. Next, we directly assessed the effect of increased NPM expression on actin stress fiber formation in wild-type astrocytes. In these experiments, increased expression of His-tagged NPM resulted in decreased actin stress fiber formation in wild-type astrocytes, which was rescued by leptomycin B treatment (Fig. 6B), similar to Nf1^−/− astrocytes. Together, these data show that increased expression of functionally mobile NPM protein mediates the decrease in actin stress fibers characteristic of Nf1^−/− astrocytes.

To extend these observations, we next examined the effect of inhibiting NPM function on Nf1^−/− astrocyte motility and proliferation. Similar to what we observed for rapamycin treatment and Rac1^N17 expression, expression of the NPMdL mutant restored cell motility (Fig. 6C) and proliferation (Fig. 6D) to wild-type levels. Together, these data indicate that NPM is a critical mTOR effector important for regulating actin cytoskeleton dynamics and cell proliferation in Nf1^−/− astrocytes.

mTOR has been shown to integrate signals from a variety of extracellular inputs, including growth factors, amino acids, glucose, ATP, and oxygen. In addition, mTOR-dependent signaling modulates numerous cellular properties, including cell proliferation, cell motility, and protein translation. We had previously found that Nf1^−/− astrocytes exhibit defects in each of these properties, suggesting that Nf1-deficient astrocytes would represent a good model system to examine the mechanisms underlying mTOR regulation of cell proliferation and motility. In this study, we use both pharmacologic and genetic approaches to show that the Nf1 gene product neurofibromin regulates actin stress fiber formation, astrocyte motility, and astrocyte proliferation in a mTOR-dependent fashion. Using Nf1^−/− astrocytes, we further show that neurofibromin regulation of these important biological properties is mediated by Rac1-dependent modulation of NPM expression and function.

The ability of mTOR to regulate these diverse biological properties is not unique to neurofibromin. Studies on two other genetic disorders that predispose individuals to glioma formation have revealed roles for mTOR pathway regulation of cell proliferation and motility. Inactivation of the tumor suppressor gene PTEN, a negative regulator of phosphatidylinositol 3-kinase signaling, causes Lhermitte-Duclos disease (29), and PTEN inactivation is the one of the most common genetic changes observed in human gliomas (30). When PTEN is inactivated in astrocytes, astrocyte proliferation is increased in vitro and in vivo (31). Moreover, in a murine glioma model, PTEN inactivation promotes astrocyte invasion in vitro and in vivo (32). Similarly, tuberous sclerosis complex (TSC) is caused by mutations in one of two genes, TSC1 or TSC2, which negatively regulate mTOR signaling (33). Inactivation of either TSC1 or TSC2 results in mTOR pathway hyperactivation in vitro and in vivo (34, 35), and inactivation of the TSC1/TSC2 complex promotes increased cell proliferation (36–38). Moreover, loss of TSC1 results in dramatic increases in NPM protein expression and NPM-dependent increases in ribosomal biogenesis (27). The TSC1/TSC2 gene products also regulate actin cytoskeleton dynamics. TSC1/TSC2 inactivation results in a decrease in actin stress fibers (39), whereas TSC2 reexpression in TSC2^−/− human cells promotes increased cell adhesion and decreased cell migration (40). In addition, TSC1 inactivation results in decreased focal adhesions, whereas TSC1 overexpression promotes an increase in actin stress fibers and associated focal adhesions (41). Consistent with these findings, we have inactivated PTEN and TSC1 in primary astrocytes, and like Nf1 inactivation, these astrocytes exhibit fewer actin stress fibers that can be rescued by short-term rapamycin treatment.⁴

mTOR signals through two distinct protein complexes: one that is sensitive to rapamycin and one that is insensitive to rapamycin-mediated inhibition (16, 17). Based on our observations that actin stress fiber formation in Nf1^−/− astrocytes was rescued by short-term rapamycin treatment, we first examined mTOR signaling intermediates that are sensitive to rapamycin, 4EBP1, and S6K. However, our data show that 4EBP1 activation is not altered by Nf1 inactivation and overexpression of S6K does not alter actin stress fiber formation in astrocytes. These results suggest that neurofibromin regulation of mTOR signaling operates in a rapamycin-sensitive manner involving effectors other than S6K and 4EBP1. Consistent with our findings in astrocytes, S6K1/S6K2 inactivation in mice impairs viability but does not block cell proliferation (42). Moreover, cell cycle progression in S6K1/S6K2^−/− mice remained sensitive to rapamycin. In addition, PTEN inactivation in neurons results in a mTOR-dependent increase in neuronal soma size independently of S6K1/S6K2 signaling (43).

In Nf1^−/− astrocytes, we found that another mTOR target, the small GTPase Rac1, acts downstream of mTOR to regulate actin stress fiber formation, cell motility, and cell proliferation. We provide several lines of evidence that the small GTPase Rac1 acts downstream of mTOR to regulate mTOR-dependent cellular processes. First, Rac1 hyperactivation in Nf1^−/− astrocytes is robustly blocked by rapamycin treatment. Second, genetic inhibition of Rac1 in Nf1^−/− astrocytes using the Rac1^N17 mutant fails to block phosphorylation of the mTOR target ribosomal S6. Third, although mTOR can initiate a positive feedback loop to activate the upstream kinase Akt, we observed no changes in Akt activation following either short-term or long-term rapamycin treatment in Nf1^−/− astrocytes.⁴ These data indicate that Rac1 activation in Nf1^−/− astrocytes is the direct result of mTOR and not compensatory Akt activation. Finally, we show that genetic inhibition of Rac1 rescues actin stress fiber formation in Nf1^−/− astrocytes and restores Nf1^−/− astrocyte motility and proliferation to wild-type levels. Together, these results indicate that Rac1 functions as a mTOR pathway signaling intermediate that regulates actin cytoskeleton dynamics and cell proliferation in Nf1^−/− astrocytes.

One clue as to how mTOR/Rac1 hyperactivation might regulate astrocyte proliferation and motility derived from the observation that neurofibromin also regulates protein translation in a rapamycin-dependent mechanism (10). Protein translation requires the synthesis of new ribosomes, a process that requires both rRNA transcription and processing as well as active shuttling of ribosomal proteins and RNAs from the nucleus to the cytoplasm. A key regulator of ribosome nuclear export is NPM, which functions as a nuclear-cytoplasmic shuttling protein. Recent studies have shown that NPM directly interacts with ribosomal subunits to facilitate their nuclear export (26). In this fashion, NPM promotes an increase in the cytoplasmic pool of actively translating ribosomes and increases protein synthesis rates. We show that NPM is overexpressed in Nf1^−/− astrocytes in vitro and in a Nf1 murine optic glioma model in vivo. This is consistent with previous studies showing that NPM expression is induced by mitogenic signals (44, 45). Furthermore, we show that NPM expression is regulated by mTOR, as mTOR inhibition by rapamycin inhibits NPM expression both in vitro and in vivo. This supports previous studies showing that NPM is highly expressed in TSC1^−/− fibroblasts and is blocked by rapamycin treatment (27). This increase in NPM expression is due to increased translation of existing NPM transcripts, as rapamycin treatment or TSC1 reexpression suppresses NPM translation (27).

Our data now show that the ability of NPM to shuttle between the nucleus and cytoplasm is a critical step in NPM-dependent regulation of actin cytoskeleton dynamics and proliferation, as inhibition of NPM shuttling using both genetic (NPMdL mutant) and pharmacologic (leptomycin B) inhibitors rescues these phenotypes in Nf1^−/− astrocytes. The exact mechanism by which NPM regulates cell motility and proliferation in astrocytes is currently unclear. It is possible that NPM-mediated increases in cytosolic ribosomes could lead to the production of specific proteins involved in cell motility and proliferation. In this regard, NPM associates with mature ribosomes and actively translating polysomes in the cytosol (26), suggesting that it has cytosolic functions at the polysome. At the polysome, NPM could regulate the translation of specific mRNA transcripts by the polysome machinery or assist in this process by supplying more cytosolic ribosomes available for translation. Support for this notion derives from a previous study that showed that mitogenic signaling by Akt or Ras has distinct effects on the recruitment of specific mRNA transcripts to polysomes (46). When Akt or Ras signaling was blocked by acute pharmacologic inhibition, there was little change in mRNA transcription but a dramatic change in the profiles of polysome-associated mRNAs. Future studies examining the profiles of actively translated proteins in wild-type and Nf1^−/− astrocytes may elucidate the mechanism underlying NPM regulation of actin stress fiber formation, motility, and proliferation in astrocytes.

Together, these studies indicate that neurofibromin/mTOR signaling regulates actin cytoskeleton dynamics and cell proliferation in astrocytes and identify Rac1 and NPM as components of the mTOR signaling pathway. This study suggests that therapies that specifically target these signaling molecules may prove useful for the treatment of NF1 and related disorders characterized by mTOR pathway activation. As gliomas frequently exhibit mTOR pathway activation (47) and NPM is highly expressed in human NF1-associated gliomas,⁵

Grant support: Department of Defense and National Cancer Institute (D.H. Gutmann) and Pew Scholars Program in Biomedical Sciences and Department of Defense grant W81XWH0610129 (J.D. Weber).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Ryan Emnett, Nisha Wadhwani, and Diane Ma for their excellent technical assistance.