Redox Regulates Mammalian Target of Rapamycin Complex 1 (mTORC1) Activity by Modulating the TSC1/TSC2-Rheb GTPase Pathway^*

Reagents, Antibodies, and Plasmids

PAO, British anti-Lewisite (BAL, also known as 2,3-dimercapto-1-propanol), diamide, 2-deoxyglucose (2-DG), U0126, and rapamycin were obtained from Sigma-Aldrich. HBSS was obtained from Invitrogen. No-Glucose Dulbecco's modified Eagle's medium (No-Glucose DMEM, Invitrogen) was used for glucose starvation treatment. [³²P]Orthophosphate was obtained from MP Biomedicals. Mouse LAMP2 (H4B4) and rat LAMP2 antibodies (GL2A7) were obtained from the Developmental Studies Hybridoma Bank at the University of Iowa and Abcam, respectively. HA and Myc antibodies were from Covance. AMPKα antibody was from Bethyl Laboratories. The p18 antibody was described previously (11). Other all antibodies used in this study were obtained from Cell Signaling. HA-S6K, Myc-Rheb, HA-TSC1, HA-TSC2, HA-RagA(QL), Myc-RagC(SN), HA-RagA(TN), and HA-RagC(QL) expression plasmids were described previously (7, 29). Lipofectamine (Invitrogen) was used for plasmid transfection according to the product manual.

Cell Culture and Treatments

HEK293T, HeLa, and mouse embryonic fibroblast cell lines (wild-type, TSC1^−/−, TSC2^−/−, p18^−/−, and p18^rev cells) were cultured in high glucose DMEM (Invitrogen) containing 10% fetal bovine serum (FBS, HyClone Laboratories) and penicillin/streptomycin (Invitrogen) (11). For amino acid stimulation assays, cells were treated with HBSS for 1 h and then cultured in low glucose DMEM for another 15 min. For cysteine oxidant or reducing agent treatment, cells were treated with HBSS for 1 h and then incubated in the reagent solution for 15 min under the following conditions: 5 μm PAO for HEK293T cells, 0.1 μm PAO for MEF cells, 250 μm diamide, and 0.5 mm BAL in HBSS. For the amino acid or glucose starvation treatments, cells were incubated in HBSS or No-Glucose DMEM for 1 h, respectively. 2-DG was added directly to the cell medium to a final concentration of 50 mm, and cells were incubated for 1 h. For rapamycin treatment, cells were preincubated with 20 nm rapamycin for 15 min before oxidant or reducing agent treatment.

Immunoblot and Immunoprecipitation

Cells were lysed on ice for 5 min in ice-cold lysis buffer (40 mm HEPES (ph 7.5), 120 mm NaCl, 1 mm EDTA, 10 mm pyrophosphate, 10 mm glycerophosphate, 50 mm NaF, 1.5 mm Na₃VO₄, 0.3% CHAPS, and a mixture of protease inhibitors (Roche Applied Science)). After centrifugation at 13,000 × g for 15 min, the supernatant was mixed with 5× SDS sample buffer and boiled for 5 min. The samples were subjected to SDS-PAGE and immunoblotted with the indicated antibodies. For immunoprecipitations, lysates were incubated with 1 μg of the indicated antibody for 2 h and precipitated with protein G/L-Sepharose.

In Vivo GTPase Assay

In vivo GTPase assays were performed as reported previously (17, 37). Briefly, cells were washed once with phosphate-free DMEM and incubated with 0.5 mCi/ml [³²P]orthophosphate for 4 h. Cells were then lysed with lysis buffer (50 mm Tris, pH 7.5, 100 mm NaCl, 10 mm MgCl₂, 1 mm DTT, 0.5% Nonidet P-40, and protease inhibitors). Tagged small GTPase protein was immunoprecipitated with Myc or HA antibody. Immunoprecipitates were washed three times each with washing buffer I (50 mm Tris, pH 8.0, 0.5% Triton X-100, 500 mm NaCl, 5 mm MgCl₂, 1 mm DTT) and buffer II (50 mm Tris, pH 8.0, 0.1% Triton X-100, 100 mm NaCl, 5 mm MgCl₂, 1 mm DTT). The GTPase-bound nucleotides were eluted with elution buffer (2 mm EDTA, 0.2% SDS, 1 mm GDP, 1 mm GTP) at 68 °C for 15 min. Eluted nucleotides were separated on polyethyleneimine cellulose plates (Baker-Flex) in 0.75 m KH₂PO₄, pH 3.4.

siRNA Transfections

Effectene (Qiagen) was used to transfect 0.8 million HeLa cells in 6-cm dishes with siRNA oligonucleotides to TSC1 (SMARTpool, Dharmacon). 72 h after transfection, the cells were rinsed once with cold phosphate-buffered saline, lysed in ice-cold lysis buffer, and analyzed by immunoblotting as above.

Immunofluorescence Staining

Cells were fixed for 15 min with 4% paraformaldehyde. The fixed cells were rinsed three times with PBS and then permeabilized with 0.2% Triton X-100 in PBS for 10 min and blocked with 2% BSA-TBS for 30 min at room temperature. After rinsing three times with TBS containing 0.05% Tween (TBS-T), samples were incubated with primary antibodies in PBS for 1 h at room temperature, rinsed three times with TBS-T, incubated with secondary antibodies for 1 h at room temperature, washed two times with TBS-T, and rinsed two times with water. Samples were mounted on microscope slides using Prolong Gold anti-fade reagent (Invitrogen). Antibodies were used at the following dilutions: mTOR (1:400), RagC (1:200), LAMP2 (1:200), Alexa Fluor 488 anti-rat IgG (1:500), Alexa Fluor 594 anti-rabbit IgG (1:500), Alexa Fluor 488 anti-rabbit IgG (1:500), and Alexa Fluor 594 anti-mouse IgG (1:500). Images were obtained by the Fluoview 300 confocal microscope (Olympus) with 60× objective lenses. Observations were carried out three times with different samples prepared independently, and representative images were processed in the same manner using Adobe Photoshop CS3.

Statistical Analysis

All data were analyzed by analysis of variance with Scheffe's post hoc tests. Asterisks in Figs. 2 and 7 represent statistical significance (p value <0.05).

mTORC1, but Not mTORC2, Is Activated by Cysteine Oxidant

PAO is a cell-permeable oxidizing reagent that cross-links vicinal thiol groups and has been widely used as a cysteine oxidant (38–40). In contrast, BAL is a reducing agent that efficiently reverses oxidation reactions induced by PAO (41, 42). As reported previously, PAO treatment induced the phosphorylation of endogenous S6K (Thr-389), a substrate of mTORC1, in HEK293T cells (Fig. 1A) (34). BAL completely blocked the stimulatory effect of PAO on S6K phosphorylation, confirming that it is the oxidizing activity of PAO that is responsible for S6K phosphorylation. Interestingly, PAO decreased the phosphorylation of the AKT hydrophobic site (the site catalyzed by mTORC2), suggesting that PAO specifically activates mTORC1, but not mTORC2, and that cysteine oxidants may not directly enhance mTOR kinase activity. Furthermore, PAO induces the phosphorylation of S6K, the site catalyzed by mTORC1, in a manner independent of AKT activation. Amino acids are known to be required for insulin-mediated stimulation of TORC1 (43), but amino acids are not required for PAO to phosphorylate S6K (Fig. 1A). Pretreatment with 2-deoxyglucose, a compound that causes energy starvation by blocking glucose utilization, however, diminished PAO-induced S6K phosphorylation. Furthermore, in the absence of glucose, S6K activation by PAO was consistently compromised (Fig. 1A). These data suggest that glucose starvation may inhibit a step in the mTORC1 pathway downstream or parallel to PAO signaling. To validate the fact that PAO stimulates S6K phosphorylation via mTORC1 activation, we examined the effect of PAO on the phosphorylation of 4EBP1, another important TORC1 substrate under amino acid starvation conditions (44, 45). Consistent with the effect of PAO on S6K phosphorylation under amino acid starvation conditions, PAO treatment also induced 4EBP1 phosphorylation (Fig. 1B). To investigate the mechanism by which mTORC1 activity is enhanced by PAO, we also examined the activities of major upstream kinases such as AKT, ERK1/2, and AMPK under the same conditions. Consistent with the result shown in Fig. 1A, the phosphorylation of AKT, a major upstream positive regulator of mTORC1, is suppressed by PAO treatment under amino acid starvation conditions (Fig. 1B). We also found that the phosphorylation of both ERK and AMPK was enhanced by PAO treatment (Fig. 1B). Previous studies have determined that AMPK is a negative regulator for the mTORC1 pathway. Thus, the result suggests that PAO also acts downstream of AMPK to overcome its inhibitory effect on the mTORC1 activation. It has been reported that ERK stimulates mTORC1 activity by inhibiting the TSC2 tumor suppressor. To determine whether PAO-induced mTORC1 activation requires ERK1/2 activities, we examined the effect of the MEK inhibitor (U0126) on PAO-induced S6K phosphorylation. Interestingly, the MEK inhibitor had no effect on PAO-induced S6K phosphorylation, although it completely abolished PAO-induced Thr/Tyr phosphorylation on ERK1/2, the sites required for ERK1/2 activity (Fig. 1C). These results indicate that PAO does not require ERK1/2 activity to induce mTORC1 activation. Together, these data suggest that PAO does not require these regulatory kinases to stimulate mTORC1 activity.

PAO Stimulates mTORC1 in a Manner Independent of Rag Small GTPases

Cysteine oxidants are able to stimulate mTORC1 under amino acid starvation conditions. This raises the possibility that increased redox potential may activate Rag small GTPases because Rags are specific modulators for mTORC1 localization and because active Rags are able to induce mTORC1 activity under conditions of amino acid starvation (7, 8). To test this hypothesis, we examined the effect of a dominant negative form of the Rag complex (RagA(TN)·RagC(QL)) on PAO-induced S6K phosphorylation. Because transfection efficiency in HEK293T cells is ∼50% in our system, HA-tagged S6K1 (HA-S6K) was co-transfected with RagA(TN)·RagC(QL), and its phosphorylation was measured. For this aim, we first tested the effects of PAO and BAL on transfected HA-S6K. The results showed that transfected HA-S6K behaved similarly to the endogenous protein (Fig. 2A). PAO treatment enhanced the phosphorylation of HA-S6K, whereas BAL abolished PAO-induced S6K phosphorylation. If PAO enhances S6K phosphorylation by inhibiting a phosphatase, then rapamycin would not cause a dramatic decrease of S6K Thr-389 phosphorylation. In fact, as expected, rapamycin completely blocked S6K Thr-389 phosphorylation stimulated by PAO, indicating that PAO stimulates S6K phosphorylation by acting upstream of mTORC1. Consistent with previous observations, overexpression of dominant negative Rag components significantly reduced amino acid-induced S6K phosphorylation (Fig. 2B) (7, 8). However, similar levels of RagA(TN)·RagC(QL) expression failed to reduce PAO-induced S6K phosphorylation, suggesting that PAO may induce mTORC1 activation in a manner independent of Rag GTPases (Fig. 2C).

PAO-induced mTORC1 Activation Is Independent of Rag-Ragulator System

Translocation of mTORC1 to lysosomes is a critical step for amino acid-induced mTORC1 activation, where Ragulator (MP1·p14·p18 complex) plays a crucial role in anchoring Rag small GTPases on the lysosomal membrane (11, 12, 46). Sancak et al. (12) showed that amino acid stimulation failed to induce S6K phosphorylation and mTOR translocation to the lysosome in Ragulator-deficient cells. To further investigate the role of Rag in PAO-induced mTORC1 activation, we tested whether PAO stimulates S6K phosphorylation in Ragulator-deficient cells. p18^−/− cells and p18^−/− cells reconstituted with wild-type p18 (p18^rev) cells were used as Ragulator-deficient and control cells, respectively (12). Consistent with previous observations, amino acid stimulation failed to induce S6K phosphorylation in p18^−/− cells. However, PAO was able to stimulate S6K phosphorylation in p18^−/− cells, further confirming a dispensable role of the Rag-Ragulator in PAO-induced mTORC1 activation (Fig. 3A). We also examined the effect of PAO and amino acid stimulation on mTOR and RagC localization in p18^−/− and p18^rev cells. As reported previously, mTOR was diffusely expressed around the nucleus without co-localizing with LAMP2 (late endosome or lysosome marker) in both p18^−/− and p18^rev cells under amino acid starvation conditions (Fig. 3B, upper panels) (12). Upon amino acid stimulation, mTOR perfectly co-localized with LAMP2 in p18^rev cells but not in p18^−/− cells (Fig. 3B, middle panels). However, we found that PAO failed to stimulate mTOR-LAMP2 co-localization in p18^rev cells (Fig. 3B, bottom panels), indicating that PAO-induced mTORC1 activation is not associated with mTORC1 localization at the lysosome. RagC was always co-localized with LAMP2 in p18^rev cells but was expressed diffusely throughout the cytoplasm in p18^−/− cells regardless of the presence or absence of amino acids or PAO stimulation (Fig. 3C). Consistently, similar observations were seen in HEK293T cells (Fig. 3D).

To further examine the relationship between redox potential and Rag-induced mTORC1 activation, we tested the effect of BAL treatment on S6K phosphorylation induced by the active Rag complex (RagA(QL)·RagC(SN)). Consistent with results shown in Figs. 1A and 2A, the levels of S6K phosphorylation were significantly reduced by BAL treatment under growth conditions (Fig. 4, lane 1 versus lane 5). However, BAL-induced reduction of S6K phosphorylation was attenuated when the active Rag complex was expressed (Fig. 4, lane 2 versus lane 6 and lane 4 versus lane 8). Together, these results support our hypothesis that cysteine oxidants induce mTORC1 activation in a manner independent of Rag-Ragulator function.

TSC1 and TSC2 Are Required for Redox Regulation of mTORC1

A question that arises from the above observations is how mTORC1 activity is up-regulated by cysteine oxidants under conditions of amino acid depletion in a Rag GTPase-independent manner. To address this question, we investigated the role of TSC-Rheb signaling in redox-dependent mTORC1 regulation. We first monitored the effects of PAO and BAL on S6K phosphorylation under conditions of Rheb or TSC1·TSC2 overexpression (Fig. 5A). As expected, Rheb overexpression caused a strong enhancement of S6K phosphorylation. Interestingly, we found that Rheb-induced S6K phosphorylation was insensitive to both PAO and BAL treatment (Fig. 5A). Moreover, PAO-induced S6K phosphorylation was diminished by overexpression of TSC1·TSC2. These data suggest that PAO and BAL may act upstream of TSC1·TSC2 and Rheb.

To further test the function of the TSC complex in redox-induced mTORC1 regulation, we examined the effects of PAO and BAL in TSC1^−/− and TSC1^+/+ MEF cells. Consistent with previous observations, TSC1^−/− cells showed a higher level of S6K phosphorylation than TSC1^+/+ cells. As seen in HEK293T cells, PAO strongly enhanced S6K phosphorylation, whereas BAL abolished it in TSC1^+/+ cells (Fig. 5B). However, we found that enhanced S6K phosphorylation in TSC1^−/− cells was resistant to both PAO and BAL treatments but completely inhibited by rapamycin treatment (Fig. 5B). We also performed similar experiments in TSC2^−/− cells. Results identical to those in TSC1^−/− cells were also observed in TSC2^−/− cells (Fig. 5C). PAO did not further increase S6K phosphorylation in the TSC2^−/− cells, whereas it enhanced S6K phosphorylation in the control TSC2^+/+ cells. Consistently, BAL did not inhibit S6K phosphorylation in the TSC2^−/− cells, whereas it inhibited S6K phosphorylation in the TSC2^+/+ cells, further supporting a role of TSC1·TSC2 in the redox regulation of mTORC1.

The functional importance of the TSC complex in redox regulation was further examined in cells with transient TSC1 knockdown (Fig. 5D). Expression of TSC1 was efficiently down-regulated by siRNA targeting TSC1 in HeLa cells and confirmed by anti-TSC1 immunoblotting. As expected, TSC1 knockdown simultaneously decreased TSC2 protein levels because TSC2 is stabilized in a complex with TSC1. In control siRNA-treated cells, PAO and BAL treatments resulted in an increase and a decrease of S6K phosphorylation, respectively. Importantly, knockdown of TSC1 in HeLa cells significantly diminished the effect of PAO or BAL on S6K phosphorylation. Together, these data suggest a model in which redox acts upstream of TSC1·TSC2 to regulate the mTORC1 pathway.

Effects of Redox on the Integrity of mTOR Complexes

It has been reported that nutrients, such as amino acids, and cysteine oxidants affect the association between mTOR and Raptor under specific lysis conditions (9, 34, 47). We also observed that PAO treatment reduced the affinity between mTOR and Raptor and that this effect was attenuated by BAL (Fig. 6A). In contrast, PAO and/or BAL had no effect on the interaction between mTOR and Rictor, supporting a specific role of redox regulation in mTORC1. Subsequently, we examined the effect of Rheb on the interaction between mTOR and Raptor. Overexpression of Rheb also reduced the interaction between endogenous mTOR and Raptor with a concomitant increase in S6K phosphorylation (Fig. 6B). PAO treatment further diminished the interaction between mTOR and Raptor. These results suggest that PAO may activate Rheb, thereby decreasing the affinity between mTOR and Raptor.

PAO Activates Rheb

The activity of Rheb in vivo can be assessed by measuring the ratio of GTP- to GDP-bound forms (17, 48). To test our assumption that the cysteine oxidant modulates the TSC-Rheb pathway, we examined the effects of PAO and/or BAL on the Rheb GTP/GDP ratio. We found that PAO treatment significantly increased Rheb in the GTP-bound state by ∼2-fold (Fig. 7A). BAL treatment alone did not decrease the overexpressed Rheb GTP level but did effectively attenuate the stimulating effect of PAO. This is consistent with the observation that BAL treatment fails to inhibit enhanced S6K phosphorylation under the condition of Rheb overexpression (Fig. 5A). The effect of PAO on guanine nucleotide loading was specific to Rheb as PAO did not affect the Ras GTP/GDP ratio (Fig. 7B). We also examined the effect of diamide, another cysteine oxidant. Diamide similarly enhanced the level of GTP-bound Rheb, and this effect was blocked by BAL (Fig. 7C). Therefore, our data indicate that PAO treatment specifically induces Rheb activity.