TSC1/TSC2 and Rheb have different effects on TORC1 and TORC2 activity

Abstract

Target of rapamycin (TOR) plays a central role in cell growth regulation by integrating signals from growth factors, nutrients, and cellular energy levels. TOR forms two distinct physical and functional complexes, termed TOR complex 1 (TORC1) and TOR complex 2 (TORC2). TORC1, which is sensitive to rapamycin, regulates translation and cell growth, whereas TORC2, which is insensitive to rapamycin, regulates cell morphology and cell growth. The Ras homology enriched in brain (Rheb) small GTPase is known to be a key upstream activator of TORC1, although the mechanism of Rheb in TORC1 activation remains to be determined. However, the function of Rheb in the TORC2 regulation has not been elucidated. By measuring Akt and S6K phosphorylation as a functional assay for TORC1 and -2, here, we report that dRheb has an inhibitory effect on dTORC2 activity in Drosophila S2 cells. This negative effect of dRheb on dTORC2 is possibly due to a feedback mechanism involving dTORC1 and dS6K. We also observed that Rheb does not activate TORC2 in human embryonic kidney 293 cells, although it potently stimulates TORC1. Furthermore, tuberous sclerosis complex 1 (TSC1) and TSC2, which are negative regulators of Rheb, have negative and positive effects on TORC1 and -2, respectively. Our observations suggest that TSC1/2 and Rheb have different effects on the activity of TORC1 and -2, further supporting the complexity of TOR regulation.

Target of rapamycin (TOR) is a protein kinase that belongs to a phosphatidylinositol kinase-related kinase family and is highly conserved in all eukaryotes (1). The TOR signaling pathway has a wide array of functions to regulate cell growth, proliferation, apoptosis, and autophagy by integrating multiple signals from growth factors, energy, and nutrients. Emerging evidence suggests that TOR exists in two distinct physical and functional complexes, termed TOR complex 1 (TORC1) and TORC2. TORC1 and -2 were initially identified in the budding yeast Saccharomyces cerevisiae, which has two TOR genes (2). Recently, TORC1 and -2 have also been characterized in mammalian and Drosophila cells that have a single TOR gene, termed mTOR and dTOR, respectively (3–6). Both of the mammalian TOR complexes contain mTOR and mammalian homologue of LST8 (mLST8)/Gβ-like (GβL) (2, 7). However, regulatory associated protein of mTOR (Raptor) and rapamycin-insensitive companion of mTOR (Rictor)/mAVO3 are exclusively associated with TORC1 and -2, respectively (5, 6). Importantly, only TORC1, but not TORC2, is sensitive to inhibition by rapamycin (2, 5). The prevailing view is that TORC1 mainly regulates cell growth and protein synthesis by phosphorylating two key translational regulators eukaryote initiation factor 4E-binding protein (4EBP1) and S6K1 (8).

The function of TORC2 is much less well characterized. However, it has been reported that TORC2 regulates the actin cytoskeleton by modulating protein kinase C and Rho-family small GTPases in a rapamycin-insensitive manner (5, 6, 9, 10). One of the exciting findings about TORC2 is that it directly phosphorylates Akt/PKB at the hydrophobic motif site Ser-473, which is the phosphorylation site of a presumed PDK2, an enzyme not identified until recently (11). Akt/PKB plays essential roles in many aspects of cellular regulation, including cell growth and cell survival (12). The demonstration that TORC2 has PDK2 activity places TORC2 at the center stage of cellular signaling and multiple cellular processes. TORC2 does not phosphorylate the wild-type S6K1 under normal physiological conditions. But, interestingly, TORC2 can phosphorylate a mutant S6K1 that has a deletion of the C-terminal region (Fig. 1A) (13). Both Akt and S6K1 belong to the AGC kinase family. Many AGC kinases, such as Akt, do not contain the C-terminal extension present in S6K1. Therefore, it is speculated that TORC2 may also phosphorylate other members of the AGC kinases, such as PKC and SGK (5, 14).

Fig. 1.

Rheb and TSC1/2 show opposite effects on TORC1 and -2 in Drosophila S2 cells. (A) Schematic illustration of S6K1 and Akt as functional readouts of TORC1 and -2, respectively. S6K1 was used as a functional readout of TORC1 activity, whereas S6K1 3A/ΔC and Akt were used as functional indicators of TORC2 activity. Phosphospecific antibodies recognizing Ser-473 of Akt and Thr-389 of S6K1 were used to detect Akt and S6K1 activity, respectively. (B) dRheb RNAi promotes dAkt phosphorylation and inhibits dS6K phosphorylation. Drosophila S2 cells were treated with 4 μg of dRheb dsRNA as indicated. Insulin stimulation (600 ng/ml insulin for 30 min) is indicated. The phosphorylation of dAkt and dS6K were detected by the phosphospecific antibodies recognizing mammalian Akt Ser-473 and S6K1 Thr-389, respectively. To avoid confusion, we kept the labeling as p-dAkt(S505) and p-dS6K(T398) for Drosophila proteins and p-Akt(S473) and p-S6K1(T389) for mammalian proteins in all figures, although the same phosphospecific antibodies were used. dS6K protein was detected by Drosophila anti-dS6K antibody, and dAkt protein was monitored by mammalian anti-Akt antibody. The two Akt isoforms generated by alternative splicing are indicated by arrows. The phospho-Akt antibody also detected a nonspecific band between the two Akt splicing forms. It should be noted that the phospho-Akt Western blot of the samples treated with insulin was exposed for a much shorter time than the corresponding Western blot of samples without insulin treatment. (C) Knockdown of dTSC1 or -2 by RNAi decreased dAkt phosphorylation but enhanced dS6K phosphorylation. Experiments were performed similarly to those described in B.

It has been well established that TOR activity is positively regulated by the phosphatidylinositol 3-kinase (PI3K) signaling in higher eukaryotes, such as Drosophila and mammals (8, 15). Furthermore, recent studies have identified that the tumor suppressors tuberous sclerosis complex 1 (TSC1) and TSC2 function as key negative regulators of TOR in both Drosophila and mammals (16, 17). Mutation of either the TSC1 or TSC2 gene results in TSC (17). TSC1 and -2 form a physical complex that is important for their physiological functions (18). Phosphorylation of S6K1 and eukaryote initiation factor 4E-binding protein is highly elevated in TSC tumor cells, indicating constitutive activation of TORC1 in TSC1^−/− or TSC2^−/− cells (8, 17). These results demonstrate that TSC1 and -2 negatively regulate TORC1 activity. The connection between TOR and PI3K is revealed by the observation that Akt, which is activated by the PI3K pathway, phosphorylates and inhibits TSC2 (19–21). Furthermore, the understanding of the molecular function of TSC1/2 in regulating TORC1 was significantly advanced when Ras homology enriched in brain (Rheb) was identified as a key component acting downstream of TSC1/2 and upstream of TOR (22–27). TSC2 inhibits Rheb activity by functioning as a GTPase-activating protein toward Rheb. In addition, Rheb stimulates phosphorylation of S6K1 and eukaryote initiation factor 4E-binding protein in a TOR-dependent manner. Rheb may directly bind to and stimulate TORC1 function (28). Therefore, a signaling pathway of PI3K-Akt-TSC2-Rheb-TORC1 has been proposed. Interestingly, active S6K1, in turn, suppresses the PI3K-Akt pathway by inactivating insulin receptor substrate (IRS) (29, 30). S6K1 has been shown to directly phosphorylate IRS1 and, thereby, attenuates PI3K activation in response to growth factors, such as insulin. Therefore, S6K1 is a key component in the feedback regulation of PI3K by TORC1.

The function of TSC1/2 and Rheb in TORC1 regulation is well established. However, no conclusion has been reached as to whether and how TSC1/2 and Rheb regulate TORC2. Here, we show that, in Drosophila S2 cells, knockdown of dRheb expression by RNA interference (RNAi) ablates the phosphorylation of dS6K1 but enhances the phosphorylation of dAkt. Knockdown of other components of the TORC1 signaling pathway, including dRaptor and dS6K1, also increases dAkt phosphorylation. In contrast, knockdown of dTSC1 or -2 enhances dS6K1 phosphorylation but inhibits dAkt phosphorylation, suggesting that dRheb positively regulates dTORC1 while negatively affecting dTORC2. Consistent with these observations, S6K1 phosphorylation is elevated, whereas Akt phosphorylation is inhibited in TSC1^−/− and TSC2 ^−/− mammalian cells. Furthermore, Rheb activates TORC1 kinase activity but not TORC2, as determined by in vitro kinase assays. Together, our data demonstrate that TSC1/2 and Rheb have different effects on TORC1 and -2.

Results

dRheb, dTSC1, and dTSC2 Have Opposite Effects on the Phosphorylation of dAkt and dS6K.

TORC1 and -2 phosphorylate different substrates and have distinct cellular functions. However, TORC1 and -2 share two common components, TOR and LST8/GβL. TORC1 has been demonstrated to be activated by Rheb in both Drosophila and mammalian cells (22–25, 31). However, it is not clear whether Rheb directly activates TORC1 or requires intermediate molecules. To determine whether Rheb activates TORC2, we used RNAi in cultured Drosophila S2 cells to knock down Drosophila Rheb (dRheb) expression. TORC2 activity is examined by measuring the phosphorylation of the hydrophobic motif site, Ser-505, of Drosophila Akt (dAkt), which is a direct physiological substrate of TORC2 (Fig. 1A) (11). In parallel, TORC1 activity is measured by the phosphorylation of the hydrophobic motif site, Thr-398, of Drosophila S6K (dS6K). Knockdown of dRheb expression with dsRNAs promoted the phosphorylation of dAkt and inhibited the phosphorylation of dS6K (Fig. 1B). In addition, knockdown of dRheb enhanced insulin-stimulated dAkt phosphorylation but blocked dS6K phosphorylation (Fig. 1B). These data show that dRheb displays opposite effects on dTORC1 and -2. dRheb plays a positive role in the phosphorylation of the dTORC1 substrate dS6K and a negative role in the phosphorylation of the TORC2 substrate dAkt.

As expected, knockdown of dTSC1 or -2 significantly increased dS6K phosphorylation. Interestingly, knockdown of either dTSC1 or -2 expression decreased dAkt phosphorylation, especially in response to insulin stimulation (Fig. 1C), indicating that dTSC1 and -2 stimulate dTORC2. The above observations are consistent with data obtained with dRheb knockdown, therefore further supporting the notion that dRheb activates dTORC1 and inhibits dTORC2. However, the above data cannot distinguish whether dRheb inhibits dTORC2 directly or indirectly.

TORC2 Mediates the Inhibitory Effect of dRheb on dAkt Phosphorylation.

We examined the effects of components in the TOR signaling pathway on phosphorylation of dAkt and dS6K. Knockdown of dS6K increased dAkt phosphorylation; however, knockdown of dAkt did not significantly inhibit dS6K phosphorylation (Fig. 2A). These results are consistent with two reports that dAkt does not play a major role in dS6K phosphorylation in Drosophila cells (11, 32). However, Lizcano et al. (33) reported that knockdown of dAkt inhibited the phosphorylation of dS6K. The reasons for this discrepancy are not obvious; therefore, further studies are needed to clarify the function of dAkt in dTORC1 regulation in Drosophila S2 cells.

Fig. 2.

Knockdown of the TOR signaling pathway components indicates that dRheb inhibits dTORC2 in vivo. (A) Effects of knockdown of TOR signaling pathway components on phosphorylation of dAkt and dS6K. Drosophila S2 were treated with individual dsRNA as indicated. Where indicated, cells were treated with 600 ng/ml insulin for 30 min. The phosphorylation of dS6K and dAkt, as well as the protein levels, were detected by the antibodies described in Fig. 1B. (B) The enhancement of dAkt phosphorylation by Rheb RNAi depends on TORC2 components. Drosophila S2 cells were treated with Rheb dsRNA and other components of the TOR pathway as indicated. Phosphorylation of dS6K and dAkt, as well as the protein levels, were detected by the antibodies described in Fig. 1B. (C) The effect of dRaptor RNAi on dAkt phosphorylation depends on TORC2. Drosophila S2 cells were cultured and treated similarly to that described in A. Four micrograms of each dsRNA was used individually or in combination with dRaptor. The phosphorylation of dS6K and dAkt and the protein levels were detected by the antibodies described in Fig. 1B. (D) Chico RNAi inhibits TORC2 activity. Drosophila S2 cells were cultured in 12-well plates and treated for 4 days with indicated dsRNA(s). Four micrograms of dsRNA was used for each target gene and added to each well on days 1 and 3. Two Chico dsRNAs, targeting different coding regions of the Chico gene, were used to determine the RNAi effect. dAkt phosphorylation and dS6K phosphorylation were determined by pAkt (Ser-473) antibody and pS6K1 (Thr-389) antibody, respectively. Protein levels were monitored by anti-Akt and anti-dS6K antibodies. (E) Amino acids have opposite effects on the phosphorylation of dAkt and dS6K. S2 cells were starved for amino acids for 30 min and restimulated with amino acids for 10 min as indicated. The amino acid-absent medium was made based on the published Schneider’s Drosophila medium (GIBCO BRL) recipe, by removing the amino acids and yeastolate. Phosphorylation and protein levels of dAkt and dS6K were determined. RNAi of dRheb and dRas were indicated as controls. The Drosophila dAkt phospho-antibody, pdAkt(Ser-505), was used in the Western blot of E. The Drosophila dAkt phosphoantibody does not detect the nonspecific band between the two dAkt isoforms. Western blotting with the Drosophila pdAkt(Ser-505) antibody produced results similar to those obtained by an anti-mammalian pAkt(Ser-473) antibody.

As expected, knockdown of dRaptor decreased dS6K phosphorylation but increased dAkt phosphorylation (Fig. 2A). These results are consistent with the current model that Raptor is a key component present in only TORC1 (3–6). Furthermore, knockdown of dRictor decreased dAkt phosphorylation and reproducibly caused a moderate increase in dS6K phosphorylation. Moreover, knockdown of dTOR decreased phosphorylation of both dAkt and dS6K, consistent with the role of dTOR in both dTORC1 and -2. It should be noted that knockdown of dRictor and dTOR caused only partial reductions in dAkt phosphorylation, although the results are readily reproducible (Fig. 2A). These results suggest the presence of an additional source of PDK2 activity in S2 cells or an incomplete knockdown by RNAi experiments. Consistent with the later possibility, knockdown of dTOR did not completely eliminate phosphorylation of dS6K (Fig. 2A), whereas rapamycin completely eliminated dS6K phosphorylation under the same conditions (data not shown). We also noticed that dLST8 RNAi did not inhibit dS6K phosphorylation, whereas mLST8 RNAi inhibits S6K1 phosphorylation in human embryonic kidney 293T (HEK293T) cells (7). A reasonable explanation could be that dLST8 is not as potent for dTORC1 function in Drosophila S2 cells as it is in HEK293T cells. Knockdown of dPTEN increased phosphorylation of both dAkt and dS6K (Fig. 2A), supporting the fact that both TORC1 and -2 are stimulated by PI3K signaling.

PDK1 phosphorylates the activation loop of S6K1 and Akt (34, 35). It has been reported that knockout of PDK1 eliminates phosphorylation of S6K1 on both the activation loop Thr-229 and the hydrophobic motif site Thr-389 in mammalian cells (36), suggesting that PDK1 activity is required for TORC1 activity or that the phosphorylation of the activation loop by PDK1 is necessary for Thr-389 phosphorylation by TORC1. In contrast, PDK1 knockouts eliminate only the activation loop Thr-308 phosphorylation in Akt but not the hydrophobic motif site Ser-473 phosphorylation, indicating that the PDK1-dependent phosphorylation of activation loop of Akt is not required for phosphorylation of Ser-473 by PDK2 (37). Consistent with the mammalian model, we observed that knockdown of dPDK1 inhibited TORC1-dependent dS6K phosphorylation but did not inhibit dAkt phosphorylation (Fig. 2A). In fact, knockdown of dPDK1 increased TORC2-dependent dAkt phosphorylation. Together, our data suggest that TORC1 and -2 are regulated differently, and the two TOR complexes may negatively affect each other.

To further test the relationship between Rheb and TORC components in dS6K and dAkt phosphorylation, we used a combination of different RNAis. Knockdown of dRheb elevated dAkt phosphorylation. The dRheb RNAis-induced dAkt phosphorylation was blocked by a simultaneous knockdown of any of the TORC2 components, including dTOR, dLST8, or dRictor (Fig. 2B). In contrast, knockdown of dRaptor, which is only present in TORC1, or dS6K, which is a downstream target of TORC1, had little effect on dAkt phosphorylation induced by Rheb RNAi. These results show that dAkt phosphorylation induced by dRheb knockdown depends on TORC2. As expected, down-regulation of dTOR and dRaptor further decreased dS6K phosphorylation in the presence of dRheb RNAi. These observations support our model that dRheb has opposite effects on TORC1 and -2. However, the above data do not distinguish whether dRheb inhibits TOR2 directly or indirectly. We performed a similar combinatory RNAi of dRaptor and other TORC components. Knockdown of dRaptor increased dAkt phosphorylation, and this effect was blocked by knockdown of dLST8, dTOR, and dRictor (Fig. 2C), supporting the importance of TORC2 in dAkt phosphorylation.

A possible mechanism for Rheb to inhibit TORC2 is the feedback inhibition of IRS by S6K1. Rheb activates TORC1, which then activates S6K1, which, in turn, phosphorylates and inhibits IRS. We examined the effect of the Drosophila IRS gene, Chico. When Chico dsRNA was combined with dRheb, dRaptor, dS6K, dPDK1, or dPTEN, the phosphorylation of dAkt was inhibited (Fig. 2D). Our results obtained in Drosophila cells suggest that dRheb regulates TORC1 and -2 oppositely. It is worth noting that knockdown of either dRheb or dRaptor produced a stronger dAkt phosphorylation than the knockdown of dS6K. These data indicate that a negative feedback inhibition by dS6K cannot be the sole factor mediating the inhibitory effect of dRheb on dAkt (Fig. 2D). dRheb may affect the balance between the two TOR complexes, thereby exerting its positive effect on dTORC1 and negative effect on dTORC2.

TORC1 is known to be regulated by nutrients, such as amino acids (8, 38). We observed that removal of amino acids induced a dramatic dephosphorylation of dS6K (Fig. 2E). Interestingly, amino acid starvation increased dAkt phosphorylation. Addition of amino acids stimulated dS6K and, at the same time, reversed the nutrient starvation-induced dAkt phosphorylation. It is worth noting that the effects of amino acids on dS6K and dAkt can be readily observed within 10 minutes. Rapamycin effectively blocked the effects of amino acids on the phosphorylation of both dS6K and dAkt (Fig. 5, which is published as supporting information on the PNAS web site). These results demonstrate that TORC1 and -2 are regulated differently by amino acids.

Rheb Stimulates Phosphorylation of the TORC1 Substrate S6K1 but Not the TORC2 Substrates Akt and S6K1 3A/ΔC.

To examine the regulation of the TOR complexes by Rheb in mammalian cells, we used a S6K1 mutant containing both a TOS motif mutation and a C-terminal deletion (S6K1 3A/ΔC) (13, 39). Mutation of the TOS motif renders this S6K1 unphosphorylatable by TORC1 (Fig. 1A). As a result, phosphorylation of S6K1 3A/ΔC on Thr-389 would be resistant to rapamycin inhibition. Interestingly, S6K1 3A/ΔC can still be phosphorylated by TORC2 as a result of the deletion of the C-terminal extension. We used the phosphorylation of S6K1 as the indicator of TORC1 activity and the phosphorylation of S6K1 3A/ΔC and Akt as the indicators of TORC2 activity (Fig. 1A). As expected, rapamycin inhibited phosphorylation of S6K1 but not Akt or S6K1 3A/ΔC in HEK293 cells (Fig. 3A). Rheb stimulated S6K1 phosphorylation (Fig. 3A), and the stimulatory effect of Rheb on S6K1 was completely inhibited by rapamycin. Rheb, however, also slightly activated S6K1 3A/ΔC. We suspected that the modest increase in phosphorylation of S6K1 3A/ΔC was due to a slight responsiveness of S6K1 3A/ΔC to TORC1. Therefore, we treated the cells with rapamycin to exclude the TORC1 effect. Indeed, the phosphorylation of S6K1 3A/ΔC reduced to the basal level after rapamycin treatment. In agreement with the results obtained with S6K1 3A/ΔC, Akt phosphorylation was not stimulated by Rheb (Fig. 3A). In addition, LY294002, a PI3K inhibitor, inhibited the phosphorylation of wild-type S6K1, S6K1 3A/ΔC, and Akt (Fig. 3A). These results are consistent with the notion that both TORC1 and -2 are regulated by the PI3K pathway. The above data show that Rheb activates TORC1 but not TORC2 in vivo, as determined by the phosphorylation of S6K1, S6K1 3A/ΔC, and Akt. A noticeable difference from the Drosophila S2 results is that Rheb does not inhibit TORC2 in HEK293 cells (Fig. 3A).

Fig. 3.

Regulation of S6K1, Akt, and S6K1 3A/ΔC phosphorylation by Rheb and TSC1/2. (A) Rheb stimulates only the phosphorylation of S6K1 but not Akt or the rapamycin-resistant phosphorylation of S6K1 3A/ΔC. HA-S6K1, GST-Akt, and HA-S6K1 3A/ΔC were transfected into HEK293 cells in the presence or absence of MYC-Rheb (50 ng). The phosphorylation of S6K1 and S6K1 3A/ΔC were detected by the phospho-specific antibody recognizing S6K1 Thr-389, whereas the phosphorylation of Akt was detected by the phospho-Akt antibody, pAkt (Ser-473). Protein levels were determined by using anti-HA (S6K1 and S6K1 3A/ΔC), anti-GST (transfected GST-Akt), anti-Akt (endogenous Akt), and anti-MYC (Rheb) antibodies. Cells were treated with either 50 μM LY294002 or 25 nM rapamycin for 30 min as indicated. (B) TSC1/2 inhibit only S6K1 phosphorylation but not S6K1 3A/ΔC. HEK293 cells were transfected with HA-S6K1 and HA-S6K1 3A/ΔC. MYC-Rheb (25 ng) or MYC-TSC1 (250 ng)/HA-TSC2 (250 ng) was cotransfected where indicated. The phosphorylation was detected by pS6K1 (Thr-389) antibody, and the protein level was determined by either anti-HA or anti-MYC antibody.

We further tested the effect of TSC1/2. Expression of TSC1/2 inhibited phosphorylation of the cotransfected S6K1 but not S6K1 3A/ΔC (Fig. 3B). In fact, TSC1/2 slightly increased the phosphorylation of S6K1 3A/ΔC, indicating that TSC1/2 regulate TORC1 and -2 oppositely. To further demonstrate the differential regulation of TORC1 and -2, the effects of various stimulation conditions on the phosphorylation of TORC1 and -2 substrates were examined. We found that phosphorylation of S6K1 3A/ΔC and Akt are similarly regulated by various extracellular stimuli. Consistent with our earlier results, phosphorylation of S6K1 was regulated differently from Akt and S6K1 3A/ΔC (Fig. 6, which is published as supporting information on the PNAS web site). For example, energy starvation induced by 2-deoxy-glucose strongly inhibited phosphorylation of S6K1 but had little effect on the phosphorylation of S6K1 3A/ΔC and Akt. Collectively, our data indicate that TORC1 and -2 are regulated differently by TSC1/2 and intracellular energy levels.

In TSC mutant mouse embryonic fibroblast (MEF) cells, both Rheb and TORC1 are constitutively active, as determined by the high S6K1 phosphorylation (Fig. 4A and B). Both TSC1^−/− and TSC2^−/− MEF cells have higher basal S6K1 phosphorylation and lower Akt phosphorylation than the wild-type MEF cells. These results suggest that TSC1/2 negatively regulates TORC1 and positively regulates TORC2. It has been reported that constitutively active S6K1 causes an inhibition of IRS (29, 30). The inhibition of IRS by S6K1 leads to a decreased response to insulin stimulation and, therefore, decreases TORC2 activity. Prolonged rapamycin treatment, which inhibits S6K1 activity, should restore the insulin response in the TSC^−/− MEF cells. As expected, rapamycin treatment dramatically enhanced insulin-stimulated Akt phosphorylation in TSC^−/− MEF cells to an extent similar to the TSC wild-type MEF cells (Fig. 4 A and B). These results support the idea that the S6K1-mediated feedback inhibition contributes to the inhibition of TORC2 by Rheb. However, rapamycin could not enhance the basal Akt phosphorylation in either the TSC1^−/− or the TSC2^−/− MEFs to the extent as that observed in the TSC wild-type cells, even though rapamycin completely blocked S6K1 phosphorylation (Fig. 4 A and B). These results suggest that the feedback inhibition by S6K1 only partly contributes to the effect of TSC1/2 on Akt phosphorylation and TORC2 activity, indicating that TSC1/2 may also regulate TORC2 in a more direct manner. Consistently, down-regulation of dRheb and dRaptor produced a more dramatic increase of dAkt phosphorylation than down-regulation of dS6K (Fig. 2D).

Fig. 4.

TSC and Rheb regulate TORC1 and -2 differently both in vivo and in vitro. (A) TSC1 knockout increases S6K1 phosphorylation and decreases Akt phosphorylation. MEF cells were cultured in six-well plates for 48 h before analysis. Cells were treated with rapamycin (25 nM) for 3 or 24 h where indicated. Insulin (600 ng/ml) was added for 30 min as indicated. The phosphorylation of Akt and S6K1 was detected by pAkt (Ser-473) and pS6K1 (Thr-389) antibody, respectively. Endogenous protein levels were detected by anti-Akt and Anti-S6K1 antibodies. (B) TSC2 knockout increases S6K1 phosphorylation and inhibits Akt phosphorylation. Experiments were similar to those described in A, except that TSC2^−/− cells were used. (C) Rheb promotes TORC1 activity but has no effect on TORC2 activity. HEK293 cells were cultured in 10-cm plates and transfected with either TORC1 components (MYC-mTOR, HA-Raptor and MYC-mLST8/GβL) or TORC2 components (MYC-mTOR, HA-Rictor/mAVO3 and MYC-mLST8/GβL). MYC-Rheb (200 ng) was cotransfected where indicated. TORC1 and -2 were immunoprecipitated by anti-HA (HA-Raptor or HA-Rictor/mAVO3) antibody. GST-Akt and GST-S6K1, purified from HEK293 cells, were used as substrates for the in vitro kinase assays. The phosphorylation of GST-Akt and GST-S6K1 was detected by pAkt (Ser-473) and pS6K1 (Thr-389) antibody, respectively. Coimmunoprecipitated mTOR was detected by the anti-mTOR antibody, whereas the protein levels of HA-Rictor and HA-Raptor were detected by anti-HA antibody. The amount of substrates (GST-Akt and GST-S6K1) was determined by Coomassie blue staining as indicated.

Rheb has been shown to stimulate S6K1 phosphorylation (an indirect in vivo assay for TORC1 activation) in transfected cells, but an increase in TORC1 kinase activity by Rheb has not been demonstrated in vitro. We wanted to more directly test the effect of Rheb on the kinase activity of TORC1 and -2. We measured TORC1 and -2 kinase activity in vitro using purified GST-S6K1 and GST-Akt as substrates. Hypophosphorylated GST-S6K1 was purified from rapamycin-treated HEK293 cells that had been transfected with a GST-S6K1 expression vector. Similarly, hypophosphorylated GST-Akt was purified from transfected HEK293 cells that had been treated with LY294002. TORC1 was immunoprecipitated by Raptor, whereas TORC2 was immunoprecipitated by Rictor/mAVO3. We found that TORC1 isolated from transfected HEK293 cells phosphorylated GST-S6K1 on Thr-389 in vitro but could not phosphorylate recombinant GST-Akt on Ser-473 (Fig. 4C). The phosphorylation of GST-S6K1 was dramatically enhanced when TORC1 was isolated from Rheb-coexpressed cells, although the mTOR levels were equivalent in the immunoprecipitates. Therefore, we presented in vitro biochemical evidence that Rheb transfection, indeed, stimulated TORC1 kinase activity. It is worth noting that Rheb coexpression caused a mobility shift of Raptor (Fig. 4C), which is due to phosphorylation (data not shown). Future study will be required to determine whether the increased phosphorylation of Raptor by Rheb contributes to TORC1 activation. Consistent with a previous report (11), TORC2 immunoprecipitated from HEK293 cells phosphorylated Ser-473 of GST-Akt in vitro but could not phosphorylate Thr-389 of the recombinant GST-S6K1 (Fig. 4C). Importantly, Rheb coexpression did not increase the ability of TORC2 to phosphorylate GST-Akt in vitro. These results clearly demonstrate that Rheb activates TORC1 but not TORC2.

Discussion

The recent identification of TORC1 and -2 significantly increased the importance of TOR in cell growth regulation and expanded the complexity of the TOR signaling network. Although the two TOR complexes share common components, they display distinct cellular functions and phosphorylate different downstream substrates. Studies have demonstrated that TSC1 and -2 suppress TORC1 activity by inhibiting Rheb (22–25, 31). It has been established that Rheb is a key upstream activator of TORC1; however, whether Rheb regulates TORC2 in a manner similar to TORC1 had not been answered.

Our data show that Rheb does not activate TORC2, although Rheb is a potent activator of TORC1 in mammalian cells. Moreover, dRheb displays an inhibitory effect on dTORC2 in Drosophila S2 cells. Consistent with this result, the TSC^−/− MEF cells, which have elevated activities of both Rheb and TORC1, show a low Akt phosphorylation on S473, indicating a decreased TORC2 activity. In other words, TSC1/2 inhibit TORC1 but display a positive effect on TORC2. Together with the direct regulation of Rheb by TSC1/2, it is consistent that Rheb activates TORC1 and inhibits TORC2 in MEF cells. However, the inhibitory effect of Rheb toward TORC2 may not be direct. We propose that the negative effect of dRheb on dTORC2 is, in large part, due to a feedback inhibitory loop involving dS6K in the Drosophila S2. It is worth noting that overexpression of Rheb did not cause a significant inhibition of Akt phosphorylation in HEK293 cells. In these cells, PTEN activity is relatively low, and TORC2 activity is high. Therefore, the effect of overexpressed Rheb on TORC2 may be overridden by the low PTEN activity in HEK293 cells. Consistent with the hypothesis that Rheb has an inhibitory effect on TORC2, TSC1/2 have a positive effect on the phosphorylation of Akt, as determined in both the knockout and overexpression cells (29, 30). Our data support the argument that Rheb activates TORC1 in a more direct manner, possibly by a direct interaction. However, we do not suggest that Rheb directly inhibits TORC2. On the contrary, the inhibitory effect of Rheb on TORC2 may be rather indirect, certainly including a feedback inhibition involving TORC1 and downstream targets.

We observed that the TORC1-specific component Raptor has an inhibitory effect on TORC2, whereas the TORC2-specific component Rictor, reciprocally, has an inhibitory effect on TORC1. A possible molecular basis for these mutually inhibitory effects is that Raptor and Rictor may compete for common components of the TOR complexes, such as mTOR and mLST8. Rheb may activate TORC1 and inhibit TORC2 by increasing the ratio of TORC1 to TORC2 in cells. In addition to the negative feedback regulation, the above model suggests another mechanism by which Rheb inhibits TORC2. In conclusion, our data from both Drosophila S2 and mammalian cells support a model in which the effect of TSC1/2 and Rheb on TORC1 and -2 are different.

Materials and Methods

Drosophila RNAi.

Drosophila RNAi experiments were performed as described in ref. 40, with minor modification (see Supporting Materials and Methods and Table 1, which are published as supporting information on the PNAS web site). Drosophila S2 cells were cultured in 12-well plates for 4 days, with a starting density of 2 × 10⁵ cells per well. On days 1 and 3, 4 μg of dsRNA targeting the gene of interest was added directly to the cells. Cells were lysed at the end of day 4, with 150 μl per well of the mild lysis buffer (10 mM Tris·HCl, pH 7.5/100 mM NaCl/1% Nonidet P-40/50 mM NaF/2 mM EDTA/1 mM DTT/1 mM PMSF/10 μg/ml leupeptin/10 μg/ml aprotinin). Cell lysates were analyzed by Western blot.

Immunoprecipitations and Kinase Assay.

HEK293 cells were cultured in 10-cm plates. When the cell density reached 3 × 10⁶ cells per plate, cells were transfected with either TORC1 components [MYC-mTOR, hemagglutinin (HA)-Raptor, and MYC-mLST8/GβL] or TORC2 components (MYC-mTOR, HA-Rictor/mAVO3, and MYC-mLST8/GβL); MYC-Rheb (200 ng) was cotransfected where indicated. Cells were lysed in 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS) lysis buffer as described in ref. 3. To immunoprecipitate TORC1 or -2, 1 μg of anti-HA (HA-Raptor or HA-Rictor/mAVO3) antibody was added to each of the cellular lysates and incubated at 4°C for 90 min. Then 20 μl of protein G Sepharose slurry (50%) was added to the lysates and incubated for another hour. Immunoprecipitates were washed four times in the lysis buffer. At the last wash, the beads were divided equally for parallel experiments. For kinase assay, the immunoprecipitate was first washed once with kinase buffer and then incubated in 15 μl of kinase assay reaction mix at 37°C for 30 min. Kinase assay reactions were designed as reported in refs. 3 and 11. To stop the reaction, 5 μl of 4× SDS sample buffer was added to each reaction, which was then boiled for 5 min.

Abbreviations

GβL

Gβ-like

HEK

human embryonic kidney

IRS

insulin receptor substrate

MEF

mouse embryonic fibroblast

mLST8

mammalian homologue of LST8

mTOR

mammalian TOR

PI3K

phosphatidylinositol 3-kinase

Raptor

regulatory associated protein of TOR

Rheb

Ras homology enriched in brain

Rictor

rapamycin-insensitive companion of TOR

RNAi

RNA interference

TOR

target of rapamycin

TORC

TOR complex

TSC

tuberous sclerosis complex

HA

hemagglutinin.