Selective Inhibition of Growth of Tuberous Sclerosis Complex 2–Null Cells by Atorvastatin Is Associated with Impaired Rheb and Rho GTPase Function and Reduced mTOR/S6 Kinase Activity

Tuberous sclerosis complex (TSC) is a tumor suppressor gene syndrome characterized by multiple tumors of the brain, kidney, heart, and skin (1, 2). Mutations in two genes, TSC1 and TSC2, are causally linked to the development of TSC. Our current understanding of the tumor suppressor effect of hamartin and tuberin, the protein products of TSC1 and TSC2, respectively, relates to the GTPase-activating protein function of tuberin toward the small GTPase Ras homologue enriched in brain (Rheb; refs. 1–4). Rheb is a major regulator of mammalian target of rapamycin complex 1 (mTORC1), which controls a pathway that mediates protein synthesis and cell growth (5). Thus, in cells lacking functional hamartin or tuberin, elevated levels of active Rheb (GTP-Rheb) lead to constitutive activation of mTOR resulting in phosphorylation of downstream p70 S6 kinase (S6K), S6, and 4EBP1 (6–8) to increase protein translation and cell growth. Recent attention has focused on mTORC1 as a therapeutic target to address the uncontrolled cell and tumor growth that occurs in TSC and TSC-related conditions, including lymphangioleiomyomatosis (LAM), a proliferative smooth muscle disorder affecting the lung (9, 10). Rapamycin inhibits the constitutively activated forms of S6K and S6 and 4EBP1 that are thought to contribute to TSC2-null cell growth. Concerns over long-term effects of rapamycin as an immunosuppressant, and agent that potentially stimulates feedback signaling leading to oncogenic Akt (11–13), have led to a search for additional agents that target the Rheb-mTOR-S6K-S6 pathway.

Rheb belongs to the Ras family of G proteins (5). During their synthesis, Ras family G proteins undergo post-translational modifications resulting in the attachment of isoprenoid lipid moieties (14, 15). This isoprenylation step is necessary for the biological activity of small G proteins and enables them to associate with relevant membranes for downstream signaling (14, 15). Ras-related G proteins are typically farnesylated (15). In contrast, Rho G proteins are typically geranylgeranylated, whereas RhoB undergoes both geranylgeranylation and farnesylation (14). Small G proteins are GTPases that cycle between active GTP-bound and inactive GDP-bound forms, and upstream signals trigger activation of G proteins and propagation of their downstream responses. The role of RhoA, which is geranylgeranylated, and its effector, Rho kinase (ROCK), in controlling smooth muscle cell contractility, nonmuscle cell cytoskeletal organization, cell cycle progression, and apoptosis has been extensively studied (16). In contrast, less is known about Rheb function and properties. Studies suggest that Rheb, which is constitutively activated in tuberin-null states, is exclusively farnesylated and that, importantly, the farnesylation of Rheb is critical for mTORC1 kinase activity toward S6K (17–19).

The statins are a family of pharmacologic agents that are best known for their cholesterol lowering effect (20). Statins are specific inhibitors of 3-hydroxy-3-methylglutaryl CoA (HMG-CoA) reductase, a rate-limiting step in the biosynthesis of cholesterol (21). Isoprenoids geranylgeranylpyrophosphate (GGPP) and farnesylpyrophosphate (FPP), which lie downstream of mevalonate, are metabolic products of the HMG-CoA pathway (20, 21). As a result of HMG-CoA reductase inhibition, the statins also inhibit the synthesis of downstream isoprenoids. The process of isoprenylation (farnesylation and/or geranylgeranylation) of Ras/Rho family proteins is required for G protein activity. At least some of the pleiotropic benefits of statins that are independent of the cholesterol lowering effect are thought to involve interference with the normal synthesis of isoprenoids, thereby impairing Ras/Rho G protein function (reviewed in ref. 22).

Atorvastatin (also known as Lipitor), a third-generation synthetic statin that is widely used clinically, is a member of the lipophilic class of statins. Based on the observations that loss of tuberin leads to constitutively activated Rheb, and that tuberin-null cells additionally contain elevated levels of GTP-RhoA (23), we investigated whether atorvastatin is a potential pharmacologic inhibitor of tuberin-null cell growth. Here, we describe significant and selective mitigation of tuberin-null cultured cell growth by atorvastatin. Furthermore, we show that atorvastatin blocks Rheb isoprenylation and downstream phosphorylation of mTOR-S6K-S6 (rapamycin-sensitive pathway), in addition to blocking RhoA activity (rapamycin insensitive) in tuberin-null cells, both of which are thought to contribute to the cell growth advantage observed in tuberin-deficient states.

Reagents and inhibitors. Atorvastatin calcium in powder form was obtained from Pfizer, Inc. and dissolved in ethanol as recommended. Inhibitors used were the Rho kinase (ROCK) inhibitor Y27632 (Calbiochem), the farnesyl transferase inhibitor (FTI-277), and geranylgeranyl transferase inhibitor (GGTI-298; Sigma). Mevalonate (100 μmol/L), FPP (10 μmol/L), GGPP (10 μmol/L), ubiquinone (30 μmol/L), and squalene (100 μmol/L) were from Sigma. Unless otherwise stated, treatments were for 24 to 48 h at the doses indicated in the text.

Cell culture. For descriptive purposes, tuberin-null cells are denoted Tsc2^−/− and tuberin-expressing cells are denoted Tsc2^+/+. ELT-3 cells are uterine-derived leiomyoma tumor cells cultured from the Eker rat that was obtained from and characterized by Dr. Cheryl Walker (M. D. Anderson Cancer Center, Houston, TX) and reexpression of the Tsc2 gene in ELT-3 cells is described elsewhere (24). Mouse embryonic fibroblasts (MEF) were obtained from Tsc2^+/+ and Tsc2^−/− littermate controls and immortalized by knockout of the p53 gene (25). MEFs and human embryonic kidney cells (HEK293; American Type Culture Collection) were grown in DMEM containing 10% serum.

Cell growth assays. Cell growth was assessed by Coulter counting and [³H]thymidine incorporation. For Coulter counting, cells were seeded at a density of 0.5 × 10⁴ to 1 × 10⁴ on 12-well plates and grown overnight. Serum was withdrawn for 72 h (ELT-3) and 24 h (MEFs). Following serum withdrawal, cells were counted in a Model Z1 Coulter Counter (Beckman Coulter) to obtain day 0 values. Treatments with serum and/or atorvastatin were at the doses indicated in the text and cells were counted again after specific times. Cell counts were expressed as fold growth compared with day 0. The incorporation of [³H]thymidine was done as described previously (24). All experiments were done in duplicate and repeated at least twice for reproducibility.

Cell transfection. Wild-type (WT) Rheb, farnesylation mutant Rheb (C181S Rheb), and corresponding empty vector were gifts from Dr. Joseph Avruch (Massachusetts General Hospital, Boston, MA), Dr. Andrew Tee (University of Dundee, Dundee, United Kingdom), and Dr. Richard Lamb (Institute of Cancer Research, London, United Kingdom). HEK293 cells were grown to 60% confluence on 10-cm dishes and transfected with plasmids using LipofectAMINE 2000 (Invitrogen).

Immunoblot analysis. Preparation of cell lysates and immunoblot analysis was done as described (24). Primary antibodies used were against phosphorylated p42/44 mitogen-activated protein kinase (MAPK), p42/44 MAPK, phosphorylated Akt (Ser⁴⁷³), Akt, phosphorylated S6, S6, phosphorylated S6K, S6K, phosphorylated mTOR, mTOR, caspase-3 (Cell Signaling Technology), tuberin, Rheb, and RhoA (Santa Cruz Biotechnology). Secondary horseradish peroxidase–conjugated antibody (Cell Signaling Technology) was used with chemiluminescent signal detection by enhanced chemiluminescence (Cell Signaling Technology). Semiquantitative densitometric analysis was done by ImageQuant software (Molecular Dynamics). On images where weak or no bands were observed, densitometry was done on overexposed immunoblots. Data were expressed as fold activation relative to control.

Immunoprecipitation. Immunoprecipitation of mTOR was done using a protocol published by Kim et al. (26).

GTP-RhoA pull-down assay. Cellular GTP-RhoA was affinity purified using glutathione S-transferase-rhotekin Rho binding domain agarose beads (Upstate/Millipore) following the manufacturer's recommendation.

Subcellular fractionation. The protocol described by Sterpetti et al. (27) was used.

In vivo Rheb guanine nucleotide binding. A modified form of the method described in ref. 28 was used. ELT-3 cells were grown to 50% confluence, and serum was withdrawn and treated with atorvastatin or vehicle for 72 h. Cells were washed, incubated with phosphate-free DMEM for 1 h, and labeled with 0.25 mCi/mL of [³²P]orthophosphate (Amersham) for 4 h. Cells were harvested and lysates were adjusted to 0.5 mol/L NaCl before immunoprecipitation for 1 h with protein G-Sepharose beads that were preincubated overnight with 20 μg of anti-Rheb (Santa Cruz Biotechnology). Nucleotides were eluted a 68°C and resolved by TLC. [³²P]GTP and [³²P]GDP were quantified by phosphorimager and ImageQuant software. The percentage GTP-bound Rheb was calculated and normalized for moles of phosphate (27).

Animals. All procedures were approved by the Children's Hospital Animal Care and Use Committee. Twelve-week-old male and female mice heterozygous for Tsc2 (29) were treated with atorvastatin given at 100 mg/kg/d by gavage for 2 weeks. Controls were treated with vehicle control (0.5% methyl cellulose) at the same volume. Tissues were homogenized by Dounce homogenizer and lysates were prepared as described above.

Statistical analysis. Results are expressed as mean ± SD. Statistical analysis using a nonparametric Kruskal-Wallis ANOVA with Dunn's multiple comparison test for follow-up analysis was done using GraphPad Prism 3.0 software.

Atorvastatin inhibits the growth of Tsc2^−/− cells. Studies have shown that members of the statin family can inhibit cell growth (28). We tested the potential inhibitory effect of atorvastatin on ELT-3 cells and MEFs and compared Tsc2^+/+ and Tsc2^−/− cells (Fig. 1). Cultured cells were treated with 10% serum in the presence of vehicle or atorvastatin and the growth curves were examined. Dose-dependent growth inhibition of both Tsc2^+/+ and Tsc2^−/− cells was observed in response to atorvastatin (Fig. 1A). Moreover, increased sensitivity of Tsc2^−/− cells relative to Tsc2^+/+ cells with statistically significant inhibition of cell counts at 1 μmol/L atorvastatin was observed. The IC₅₀ for Tsc2^−/− cells was lower than that for Tsc2^+/+ cells [Tsc2^+/+ MEFs (1.71 μmol/L) versus Tsc2^−/− MEFs (0.69 μmol/L); Tsc2^+/+ ELT-3 (2.75 μmol/L) versus Tsc2^−/− ELT-3 (0.89 μmol/L)]. Similarly, significant inhibition of serum-induced incorporation of [³H]thymidine was observed in Tsc2^−/− cells compared with Tsc2^+/+ cells, indicating that the increased sensitivity of Tsc2^−/− cell growth to atorvastatin is paralleled by inhibition of DNA synthesis.

We have previously shown that the growth advantage conferred by the loss of tuberin in ELT-3 cells is most apparent under conditions of serum withdrawal (30) and that Tsc2^−/− ELT-3 cell growth is also estrogen responsive (24). Figure 1B shows that continued growth of Tsc2^−/− ELT-3 cells (and MEFs; Supplementary Data) in the absence of serum is significantly inhibited by atorvastatin without any apparent effect on Tsc2^+/+ cell growth. Under identical serum-free conditions, significant inhibition of [³H]thymidine incorporation was also observed in Tsc2^−/− ELT-3 cells. Similarly, atorvastatin inhibited the estrogen-stimulated growth in Tsc2^−/− cells with no effect on Tsc2^+/+ cells (Fig. 1C). The observed reduction in cell number by higher doses of atorvastatin of Tsc2^−/− cells below control cell levels in the absence of serum (Fig. 2A) suggested that atorvastatin may induce cell death. Concordantly, atorvastatin also resulted in increased cleaved poly(ADP-ribose) polymerase (PARP; Fig. 1D), a substrate for caspase-3, the activation of which is a terminal event for apoptosis, an observation that was more apparent in Tsc2^−/− MEFs compared with Tsc2^−/− ELT-3 cells.

Inhibition of Tsc2^−/− cell growth by atorvastatin is reversed by FPP and GGPP isoprenoids. The observed antiproliferative effects of atorvastatin could be due to targeting any of the downstream products of HMG-CoA reductase (outlined in schema in Supplementary Data), including the isoprenoids FPP and GGPP, and/or other byproducts, such as ubiquinone and squalene. To examine the targets downstream of HMG-CoA that mediate the inhibitory effect of atorvastatin, we tested cell growth in the presence and absence of these downstream products. Atorvastatin-induced inhibition of Tsc2^−/− cell growth was reversed by mevalonate supplement, a compound that lies at the apex of the cholesterol biosynthesis cascade (20). Furthermore, significant reversal of atorvastatin-induced growth inhibition was observed in response to GGPP and FPP. In contrast, no reversal was observed by ubiquinone or squalene supplement. These data suggested that atorvastatin exerts its effects on Tsc2^−/− cells by inhibiting protein prenylation. Consistent with this observation, we found that, under conditions of serum withdrawal, either a farnesyl transferase inhibitor (FTI-277) or a geranylgeranyl transferase inhibitor (GGTI-298) selectively and dose dependently inhibits Tsc2^−/− cell growth, and the combination has an additive effect that approached the inhibition seen with atorvastatin (shown in Supplementary Data). In contrast, no growth of Tsc2^+/+ cells was observed with resulting minimal effects of farnesyl transferase inhibitor and geranylgeranyl transferase inhibitor.

Atorvastatin inhibits the prenylation of Rheb. Farnesylation of Rheb is important for Tsc2^−/− cell growth (17). We tested the effect of atorvastatin on Rheb prenylation by immunoblot (Fig. 2A). A shift in Rheb to a slower migrating form that corresponds to the unprenylated form was observed in both Tsc2^+/+ and Tsc2^−/− ELT-3 cells, indicating that atorvastatin treatment interferes with Rheb prenylation.

Atorvastatin inhibits the downstream phosphorylation of mTOR-S6K-S6. In tuberin-deficient states, constitutively active Rheb leads to downstream constitutive phosphorylation of mTOR-S6K-S6 to regulate cell size and protein synthesis (18). The effect of atorvastatin on these downstream mediators of cell growth in tuberin-null states was next examined. We observed significant inhibition of mTOR phosphorylation using an antibody directed at the autophosphorylation site of mTOR, a finding that was replicated in MEFs but best observed in the ELT-3 cell line (Fig. 2B). Furthermore, reduced association between endogenous mTOR and Rheb in atorvastatin-treated Tsc2^+/+ and Tsc2^−/− cells was observed, a process that was reversed by FPP supplement.

Immunoblot analysis for phosphorylated forms of S6K and S6 showed significant dose-dependent inhibition of the constitutive phosphorylation of S6K and S6 by atorvastatin in Tsc2^−/− ELT-3 cells (Fig. 2C) with parallel data replicated in Tsc2^−/− MEFs (data not shown). In contrast, no inhibition of phosphorylated p42/44 extracellular signal-regulated kinase (ERK) 1/2 MAPK or phosphorylated Akt was observed in Tsc2^−/− cells nor was there inhibition of phosphorylated S6K or phosphorylated S6 in Tsc2^+/+ cells. It should be noted in the Fig. 2C graphs that, although the Tsc2^+/+ cell phosphorylated S6K and phosphorylated S6 bands were not visible, they were set to the same starting fold value as for Tsc2^−/− cells for ease of comparing any change. Interestingly, in ELT-3 cells, a slight reduction in Akt phosphorylation was seen in Tsc2^+/+ cells in response to atorvastatin. Under identical conditions to the rescue studies described above, atorvastatin-induced inhibition of S6 and S6K phosphorylation was reversed by FPP, whereas no rescue was observed by GGPP, ubiquinone, or squalene (Fig. 2D).

Inhibition of Rheb prenylation is critical for the growth-inhibitory effect of atorvastatin. In tuberin-deficient states, levels of GTP-Rheb are increased, leading to the downstream activation of mTORC1-S6K (3, 4). We measured the levels of endogenous GTP-Rheb in ELT-3 cells in response to atorvastatin. As expected, GTP-Rheb levels in Tsc2^−/− cells were higher compared with Tsc2^+/+ cells; moreover, this was reduced by atorvastatin (Fig. 3A). Using ectopically expressed WT Rheb in HEK293 cells, we observed ∼60% GTP-Rheb levels, in keeping with other reports (18, 31), which were reduced by atorvastatin treatment. In contrast, the levels of GTP loading of a mutant form of Rheb that is resistant to farnesylation, C181S Rheb, were lower than that of WT Rheb. We next assayed the growth of HEK293 cells overexpressing WT or C181S Rheb (Fig. 3B) and the response to atorvastatin. WT Rheb expression was associated with a modest but significant growth advantage that was significantly and dose dependently inhibited by atorvastatin compared with control cells. In contrast, cells expressing C181S Rheb were resistant to the growth-inhibitory effects of atorvastatin.

Atorvastatin leads to altered Rheb membrane association in Tsc2^−/− cells. We investigated the membrane localization of endogenous Rheb in response to atorvastatin by subcellular fractionation of serum-deprived ELT-3 cells (Fig. 3C). Under serum-free conditions, the majority of Rheb was cytosolic in Tsc2^+/+ cells, and this was unchanged by atorvastatin treatment. Increased amounts of membrane-associated Rheb, as reflected by a reduction in the S100 to P100 ratio, were found in Tsc2^−/− cells compared with Tsc2^+/+ cells (Tsc2^+/+, 2.05 + 0.18; Tsc2^−/−, 1.59 ± 0.20). Treatment of Tsc2^−/− cells with atorvastatin led to a considerable reduction in the amount of membrane-bound Rheb, as reflected in an increase in the S100 to P100 ratio (Tsc2^−/− control, 1.59 ± 0.20; atorvastatin, 3.25 ± 0.28). The majority of mTOR was membrane associated and its localization was unchanged by atorvastatin. These results suggest that atorvastatin inhibits the membrane localization of Rheb in Tsc2-null cells.

Atorvastatin inhibits in vivo phosphorylation of S6. Loss of Tsc2 expression is associated with constitutive phosphorylation of S6. We investigated the in vivo effect of atorvastatin on Tsc2-heterozygous mice. All mice tolerated atorvastatin or vehicle with no apparent side effects. Immunoblot analysis of tissue lysates showed reduction in the level of phosphorylated S6 in lung and kidney (Fig. 3D).

Atorvastatin inhibits the activity of RhoA in Tsc2^−/− cells. Our Fig. 2A data suggest that atorvastatin inhibits the growth of Tsc2^−/− cells in a manner that is partially GGPP dependent. Rheb is normally farnesylated and its signaling is impaired when it is farnesylation deficient (17, 19), suggesting that other geranylgeranylated proteins, such as Rho GTPase, are targeted by atorvastatin. Given the role of Rho in cell morphology, we examined cell phenotype in the presence and absence of atorvastatin. Alteration in cell phenotype was only observed in Tsc2^−/− cells treated with atorvastatin with significant rounding of cells noted (Fig. 4A). Furthermore, significant reversal of atorvastatin-induced phenotype was observed by mevalonate or GGPP supplement, suggesting that atorvastatin selectively exerts its morphologic effect by inhibiting geranylgeranylation in Tsc2^−/− cells. The Y27632 ROCK inhibitor led to a modest reduction in Tsc2^−/− cell growth, supporting a role for Rho in Tsc2^−/− cell growth. Furthermore, Tsc2^−/− ELT-3 cells contained increased levels of active GTP-RhoA compared with Tsc2^+/+ cells. Interestingly, the elevated levels of GTP-RhoA in Tsc2^−/− cells were not inhibited by rapamycin (Fig. 4B). In contrast, atorvastatin effectively reduced the elevated levels of GTP-RhoA in Tsc2^−/− cells, and this was reversed by GGPP and not FPP. Tsc2^−/− cells also contained increased levels of membrane-associated RhoA compared with Tsc2^+/+ cells (Tsc2^+/+, 1.96 + 0.56; Tsc2^−/−, 1.40 ± 0.09; Fig. 4C). Furthermore, treatment with atorvastatin led to a strong reduction in membrane-associated RhoA in Tsc2^−/− cells as reflected by a 5.7-fold increase in the S100 to P100 ratio (Tsc2^−/− control, 1.4 + 0.09; atorvastatin, 8.06 ± 3.15) compared with a 2.1-fold increase in the S100 to P100 ratio in Tsc2^+/+ cells (Tsc2^+/+ control, 1.96 ± 0.56; atorvastatin, 4.16 + 1.27). These data suggest that RhoA contributes to the growth advantage of tuberin-null states and that atorvastatin can mitigate this rapamycin-insensitive pathway.

In this study, we show for the first time that a statin, such as atorvastatin, selectively inhibits the growth of Tsc2^−/− cells compared with Tsc2^+/+ cells. Moreover, we show that atorvastatin can impair Rheb GTPase activity and function, which has not been shown before. Furthermore, we show that atorvastatin attenuates Rheb-mTOR complex formation and mTOR-S6K activity in Tsc2^−/− cells. The growth inhibition is associated with reduction of Rheb function and signaling as well as reduced RhoA function. The findings of growth inhibition and altered Rheb/Rho function in two unrelated Tsc2^−/− cell types, MEFs and the rat ELT-3 smooth muscle line, indicate that these are dependent on tuberin deficiency rather than a chance event. The antiproliferative activity of statins is associated with the lipophilic statins, such as atorvastatin, used here (28). Significant inhibition of cellular proliferation was achieved at lower doses of atorvastatin in Tsc2^−/− cells compared with Tsc2^+/+ cells, the latter requiring a 5- to 10-fold higher dose of atorvastatin for the same reduction, evident with and without serum. Serum concentrations of statins vary from 10 to 200 nmol/L when prescribed as anticholesterol agents in humans. The antiproliferative effect as low as 100 nmol/L noted here in the absence of serum, and 1,000 nmol/L with serum, suggests a possible overlap with anticholesterol serum levels, albeit at the lower range, although the short-term (24–48 h) administration of atorvastatin in cell culture used here is difficult to compare with oral statin administration over months/years. Moreover, recent published human studies (32) show that higher statin doses are relatively well tolerated, presenting the possibility that the effective dose in culture may be near clinically relevant levels. In keeping with the selective antiproliferative effect of atorvastatin in Tsc2^−/− cells, lower concentrations of atorvastatin (1 μmol/L) were also required for the induction of cleaved PARP compared with Tsc2^+/+ cells, suggesting that at least some of the inhibitory effect of atorvastatin on Tsc2^−/− cells seems to involve a selective apoptotic response as measured by cleaved PARP. Although some difference in levels of sensitivity of Tsc2^−/− MEFs and ELT-3 cells to atorvastatin-induced apoptosis was observed, the inhibitory effect was comparable. The effect of statins on cell cycle and apoptosis seems independent of each other and both can occur in the same cell line at different concentrations (33). Our data suggest that statins are efficacious inhibitors of cell growth independent of the balance between proliferation and apoptosis in tuberin-null cells.

Unlike tuberin-expressing cells, tuberin-deficient cells show a growth response to estrogen in the absence of serum (24), and our finding that atorvastatin inhibits this response is intriguing. Recent data suggest that Rheb is involved in the proliferative response to estrogen (34), and statins inhibit the growth of breast cancer cells in response to estradiol (35–37). Thus, the mechanisms involved in estrogen-induced Tsc2^−/− cell growth and the action of statins, whether overlapping or not with estrogen-responsive breast cancer cells, warrant further investigation, especially in light of the fact that LAM, a TSC2 deficiency syndrome, occurs mainly in women.

Our finding that mevalonate blocks the inhibitory effect of atorvastatin on Tsc2^−/− cell growth indicates that the effects are HMG-CoA reductase pathway dependent rather than independent. Mevalonate is the precursor of the isoprenoids FPP and GGPP, and the finding that FPP or GGPP, though not squalene or ubiquinone, results in significant reversion of inhibition indicates that isoprenylation is targeted. These findings are in keeping with the literature showing that tumor cells are often more sensitive than their normal counterparts to statin-mediated growth inhibition via isoprenoid-mediated suppression (28). Complete reversion of atorvastatin effects in Tsc2^−/− cells seemed to require both isoprenoids, suggesting that inhibition of both farnesylation and geranylgeranylation contributes to the antigrowth effects of atorvastatin in Tsc2^−/− cells. This was independently confirmed by the use of farnesyl transferase inhibitor and geranylgeranyl transferase inhibitors, both of which were inhibitory toward Tsc2^−/− cells, with an additive effect. These results, and the knowledge that Rheb and Rho, which are prenylated, are implicated in Tsc2^−/− cell growth lead us to conclude that both GTPases are statin targets.

In support of the concept that atorvastatin targets Rheb, our data identified the slower mobility form of Rheb corresponding to unprenylated Rheb after statin treatment. Its presence in both Tsc2^+/+ and Tsc2^−/− cells and the comparable inhibition of Rheb-mTOR interaction in both Tsc2^+/+ and Tsc2^−/− cells suggest that atorvastatin is pharmacologically active in both cell types. Yet, the selective inhibition by atorvastatin of cell growth only occurs in Tsc2^−/− cells. Concordantly, a selective effect of atorvastatin on the kinase activity of mTOR as measured by S6K/S6 phosphorylation was also found in Tsc2^−/− but not Tsc2^+/+ cells. We interpret these results within the context of reports showing that, whereas Rheb-mTOR binding per se is unaffected by GTP-Rheb, the kinase activity of mTOR toward S6K is GTP-Rheb dependent (18). Thus, we propose that reduced Rheb prenylation by atorvastatin selectively affects Tsc2^−/− cells because Rheb is in its constitutively GTP-bound active form in TSC2-null states. This conclusion is also supported by the selective statin response on GTP-Rheb levels in Tsc2^−/− ELT-3 cells and in cells overexpressing Rheb where 60% of Rheb is GTP bound (18).

It could be argued that the reduced levels of Rheb bound to mTOR in atorvastatin-treated cells are due to reduced total Rheb levels. However, we did not observe a reduction in total Rheb levels by statin treatment. On the contrary, inhibition of isoprenylation of G proteins by statin frequently leads to decreased degradation resulting in the accumulation of inactive total G protein in the cytosol (38), an effect that was most evident for Rheb in Fig. 2A. The finding that atorvastatin targets Rheb function is further supported by the Rheb overexpression and localization studies. We show that statin inhibits Rheb-induced proliferation, whereas the reversal by C181S Rheb indicates the statin effect on Rheb prenylation is important in mediating growth inhibition. Reports have shown that, when overexpressed, a significant amount of Rheb lies at or near plasma or organelle membranes (39–42). Our study is the first to examine the cellular localization of endogenous Rheb, although the fractionation method used here does not distinguish organelle versus plasma membrane. The finding that both Tsc2^+/+ and Tsc2^−/− cells contain cytosolic endogenous Rheb and a relatively low GTP-Rheb fraction may be explained by the condition of serum deprivation. However, Tsc2^−/− cells contain an increased fraction of membrane-associated Rheb, in keeping with the increased levels of GTP-Rheb in these cells. The reduction of membrane-associated Rheb fraction in Tsc2^−/− cells by atorvastatin agrees with the growth-inhibitory property of statins. Although the majority of mTOR appears in the membrane fraction, this may be partly influenced by its large size.

In addition to its selective effect on cell growth, atorvastatin treatment also selectively leads to rounding of Tsc2^−/− but not Tsc2^+/+ cells (Fig. 4A). It is well recognized that the rounding response of cells, such as fibroblasts, is induced by Rho inactivation (43), as is statin-induced cell rounding (44), and the reversal of rounding by GGPP is concordant with Rho as a statin target in Tsc2^−/− cells. The finding of increased levels of active RhoA in Tsc2^−/− cells that are rapamycin insensitive agrees with Goncharova et al. (23) and supports a role for RhoA in Tsc2^−/− growth. Interestingly, atorvastatin inhibits the elevated GTP-RhoA levels in Tsc2^−/− cells, and this is reversed by GGPP. The reduced RhoA membrane localization by statin treatment in both cell types is concordant with other reports of statin effects on Rho localization (38), although the additional inhibitory effect on Rho-GTP loading by atorvastatin is selective for Tsc2^−/− cells. RhoA may regulate Janus-activated kinase-signal transducers and activators of transcriptions (45), the latter that may be dysregulated in tuberin-null cells (46), and the mechanism of RhoA-mediated Tsc2^−/− cell growth warrants additional investigation.

Our results support the model (Fig. 4D) that atorvastatin treatment of Tsc2^−/− cells attenuates Rheb and RhoA function, with concordant inhibition of the dual pathways of tuberin-null cell growth: the rapamycin-sensitive Rheb-mTORC1-S6K-S6 pathway and the rapamycin-insensitive activation of RhoA (23, 29, 47). Our study did not detect inhibition of p42/44 ERK MAPK phosphorylation by atorvastatin, suggesting that the statin effects on Tsc2^−/− cells are not mediated via Ras. Although potential atorvastatin effects on other prenylated proteins cannot be absolutely ruled out, the unique roles of Rheb and Rho in TSC2-null cell growth, based on their altered function due to TSC2 mutation, are well recognized by genetic model systems and mammalian cell studies (18, 40), thus particularly implicating them in the differential statin effect on TSC2-null cells. The growth-inhibitory effects of statins have led to their consideration as potential pharmacologic agents for the treatment of cancer and are currently being evaluated in clinical trials (28, 48). Our finding that short-term atorvastatin administration to Tsc2-heterozygous mice attenuates the high levels of phosphorylated S6 found in these mice in lung and kidney, two sites involved in LAM and TSC, suggests that statins may be efficacious in vivo. Furthermore, their relative long-term safety profile supports advocating their potential use as an alternative or adjunct to rapamycin therapy, particularly in light of recent studies linking long-term rapamycin treatment with Akt induction and associated oncogenesis (49). Lastly, further studies of statin effects may provide additional insight into tuberin-null cell growth, leading to potential new therapeutic targets for TSC and LAM.

Grant support: NIH grants K08HL74113 (G.A. Finlay), HL032723 (B.L. Fanburg), and NS31535 (D.J. Kwiatkowski); The LAM Foundation Established Investigator Award (G.A. Finlay); and Pfizer ARA and American Heart Association (D. Toksoz).

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 Drs. Joseph Avruch, Andrew Tee, and Richard Lamb for the WT and farnesylation mutant Rheb expression vectors and for their expert advice of evaluating the GTP loading of Rheb and Dr. Cheryl Walker for providing the tuberin-null ELT-3 cell line.