Isolation of Hyperactive Mutants of Mammalian Target of Rapamycin*

The mammalian target of rapamycin (mTOR) is a Ser/Thr kinase that plays essential roles in the regulation of a wide array of growth-related processes such as protein synthesis, cell sizing, and autophagy. mTOR forms two functionally distinct complexes, termed the mTOR complex 1 (mTORC1) and 2 (mTORC2); only the former of which is inhibited by rapamycin. Based on the similarity between the cellular responses caused by rapamycin treatment and by nutrient starvation, it has been widely accepted that modulation in the mTORC1 activity in response to nutrient status directs these cellular responses, although direct evidence has been scarce. Here we report isolation of hyperactive mutants of mTOR. The isolated mTOR mutants exhibited enhanced kinase activity in vitro and rendered cells refractory to the dephosphorylation of the mTORC1 substrates upon amino acid starvation. Cells expressing the hyperactive mTOR mutant displayed larger cell size in a normal growing condition and were resistant to cell size reduction and autophagy induction in an amino acid-starved condition. These results indicate that the activity of mTORC1 actually directs these cellular processes in response to nutrient status and confirm the biological functions of mTORC1, which had been proposed solely from loss-of-function analyses using rapamycin and (molecular)genetic techniques. Additionally, the hyperactive mTOR mutant did not induce cellular transformation of NIH/3T3 cells, suggesting that concomitant activation of additional pathways is required for tumorigenesis. This hyperactive mTOR mutant will be a valuable tool for establishing physiological consequences of mTOR activation in cells as well as in organisms.

insights. mTORC1 has been implicated in the regulation of a wide array of growth-related processes such as protein translation, cell sizing, ribosomal biogenesis, and autophagy. Although direct targets of mTORC1 leading to those cellular processes are poorly identified, the phosphorylation of ribosomal S6 kinase 1 (S6K1) and eIF4E-binding protein 1 (4EBP1) by mTORC1 coordinately leads to activation of protein translation and an increase in cell size (19). In contrast, mTOR antagonizes autophagy, the bulk degradation process of intracellular components, to provide energy and nutrient sources. Although yeast TORC1 controls the early activation step of autophagy through the phosphorylation of Atg13 (20), a similar regulatory mechanism in mammalian cells has not been elucidated. mTORC2 was previously sought PDK2, which directly phosphorylates Ser-473 in the hydrophobic motif of Akt1 (21), and is involved in phosphatidylinositol-3-phosphate kinase-Akt signaling. In addition, mTORC2 regulates the organization of actin cytoskeleton in mammals (9) as in yeast (22), although detailed mechanisms remain unclear.
As described above, the physiological roles of mTOR have been defined from loss-of-function analyses both pharmacologically as well as (molecular)genetically. Because cellular responses caused by mTORC1 inactivation are similar to those induced by growth factors or nutrient deprivation, mTORC1 activity is believed to be regulated by signals from these factors. However, it has been elusive whether and how amino acid deprivation causes down-regulation of mTORC1 activity, which in turn induces these cellular responses. Toward elucidating whether modulation in the mTORC1 activity directs these cellular responses and understanding the roles of mTOR in cellular and physiological contexts, most important clues will be available from analysis using gain-of-function mutants of mTOR. Furthermore, although several lines of evidence suggest that some types of cancer are accompanied by hyperactivation of mTOR (23), it remains unclear whether mTOR hyperactivation is a primary cause. The gain-of-function analysis would also provide an understanding of how mTOR hyperactivation links to tumorigenesis. In this report we have developed a yeast screening system for the isolation of hyperactive mutants of mTOR. This system allowed us to identify mTOR mutants with enhanced kinase activities. We have also investigated the effects of the isolated mTOR mutants on the cellular processes that have been associated with the mTOR activity.
Screening for Hyperactive mTOR-The TOR2-mTOR chimera construct was designed based on a previous report (24). To construct the TOR2N⌬-pRS426 plasmid (pYO189), which contains the Tor2 coding region (amino acid residues 1-1689), the SmaI-BamHI fragment from pJK4 (25) was inserted into the SmaI-BamHI sites of pRS426 (26). A DNA fragment corresponding to the C-terminal one-third of mTOR (amino acid residues 1721-2549) flanked by additional sequences homologous to TOR2 at the 5Ј-terminal region and to the vector at the 3Ј-terminal region for homologous recombination was obtained by PCR with pcDNA1-HA-mTOR (27) as a template using primers mTOR0422-F (TGAAGCAATTAATTAATTT-CACATCTAGAATGGCTCATGATTTAGGTTTGGATCC-Ggtgcagaccatgcagcagcag; capital letters indicate homologous sequence to TOR2) and mTOR0422-R (CAAAAGCTGGAGC-TCCACCGCGGTGGCGGCCGCTCTAGAACTAGTGGAT-CCtctagatgtggcttcacttata; capital letters indicate homologous sequence to the vector). The PCR fragment and pYO189 linearized with BamHI were co-transformed into a wild-type yeast strain, and the resultant gap-repaired TOR2-mTOR-pRS426 plasmid, pYO203, was recovered from uracil-prototroph transformants. To generate the acceptor plasmid pYO207, the region corresponding to the kinase domain of mTOR (residues 2186 -2422) in the pYO203 was replaced by a fragment consisting of HindIII and XhoI recognition sites using two-step overlap extension PCR. First PCR reactions were separately performed with pYO203 as a template using primers oYO112 (ctcgagcggaagcttGAGGAAAACAAACTCGTGCC) and oYO101 (GCTATGACCATGATTACGCC) or oYO111 (aagcttccgctcg-agTATGACCCTCTGCTGAATTG) and oYO113 (ccatgatggtgagtgaagag). The first PCR products were used as overlapping templates for the second PCR reaction using primers oYO101 and oYO113. The second PCR products were digested with EcoNI and SpeI and inserted between the EcoNI and SpeI sites of pYO203.
The DNA region corresponding to the mTOR kinase domain (amino acid residues 2161-2451) was randomly mutagenized by error-prone PCR using GeneMorph® II random mutagenesis kit (Stratagene, La Jolla, CA) with primers GapMutF (gcaagtcatcacatccaagc) and GapMutR (ctgcagaataggagtctgtc) according to the manufacturer's instruction. The amplified DNA fragment was purified by phenol extraction followed by ethanol precipitation and was introduced directly into the lst8 ts strain TM478 (MATa ura3-52 lst8 ts ) together with pYO207 digested with HindIII and XhoI. The molar ratio of the mutagenized PCR product to the acceptor plasmid was 5 to 1. Transformation was conducted essentially as described by Gietz and Woods (28). Transformed yeast strains were plated on synthetic complete without uracil (SC-U) plates and incubated at 25°C. Sixteen hours later, culture temperature was shifted to 37°C, and plates were incubated for 7 days. Candidate plasmids were isolated from the colonies grown at 37°C.
Mammalian Cell Culture and Transfection-HeLa and HEK293 cells were maintained in DMEM supplemented with 10% fetal bovine serum (FBS) under 5% CO 2 atmosphere. Transient transfection was performed with LipofectAmine2000 (Invitrogen) according to the manufacturer's instructions. For amino acid deprivation, cells were washed once with DMEM deprived of amino acids (gifted from Ajinomoto Co.) and incubated in the same medium. For rapamycin treatment, cells were incubated in DMEM containing 100 nM rapamycin. For transient transfection with puromycin selection experiments, 24 h after transfection cells were trypsinized and re-seeded in DMEM ϩ 10% FBS containing 1.5 g/ml puromycin and grown for 24 h.
Cell Lysate Preparation, Immunoprecipitation, and in Vitro Kinase Assay-After 48 h of transfection, cells were washed with ice-cold phosphate-buffered saline, rapidly frozen on liquid N 2 , and then lysed in lysis buffer (20 mM Tris-HCl (pH 7.5), 1 mM EDTA, 100 mM NaCl, 1% Triton X-100, 50 mM NaF, 10 mM ␤-glycerophosphate, 1 mM phenylmethylsulfonyl fluoride, 4 g/ml aprotinin, 4 g/ml leupeptin). The lysate was cleared by centrifugation at 15,000 ϫ g for 5 min. The protein concentration of lysate was determined using Bio-Rad DC protein assay kit (Bio-Rad). Cell lysate was subjected to SDS-PAGE and transferred to Immobilon polyvinylidene difluoride membrane (Millipore) for immunoblotting.
Cell Size Determination-HeLa cells were transfected with pIRES-puro3-FLAG-mTOR or pIRES-puro3-FLAG-mTOR SL1ϩIT . Twenty-four hours after transfection the cells were trypsinized and split to re-plate into two dishes and cultured in DMEM ϩ 10% FBS containing 1.5 g/ml puromycin. After 24 h of selection, the cells were transferred into DMEM ϩ 10% FBS or modified DMEM containing an amino acid mixture at 25% in DMEM and cultured for another 24 h. Then the cells were trypsinized, suspended in ice-cold phosphate-buffered saline containing 1% FBS, and washed once with ice-cold phosphate-buffered saline. The cells were diluted with Isoton 2 (Beckman Coulter, Fullerton, CA) and subjected to a Coulter counter (Multisizer 3, Beckman Coulter). The experiments were repeated 5 times, and at least 10,000 cells were analyzed at each experiment.
Transformation Assay-NIH/3T3 cells were transfected with empty vector or expression plasmids for wild-type FLAG-mTOR, FLAG-mTOR SL1ϩIT , or an oncogenic form of RAS (H-RAS 12V ) by calcium phosphate coprecipitation. Transfected cells were trypsinized and reseeded in DMEM ϩ 10% FBS with or without 1 mg/ml G418. Focus formation efficiency was measured by counting foci, and transfection efficiency was evaluated by counting G418-resitant colonies.

RESULTS
Screening for Hyperactive mTOR Mutants-We have previously obtained active mutants of Tor2 that exhibit increased in vitro kinase activity because of a point mutation within the kinase domain. 4 Importantly, these mutants were also able to partially suppress the temperature-sensitive growth of an lst8 ts strain, TM478. To obtain mTOR mutants with enhanced kinase activity, we utilized the finding that the kinase domain of mTOR is interchangeable with that of Tor2 in S. cerevisiae (24) (Fig. 1A). To this end, Tor2-mTOR chimera mutants were screened for those that could rescue growth of TM478 at the restrictive temperature by virtue of enhanced activity of their mTOR kinase domain. The screening scheme is summarized in Fig. 1B. A cDNA fragment encompassing the mTOR kinase domain was randomly mutagenized through error-prone PCR, and the resultant product was introduced into TM478 together with the acceptor plasmid, which lacks the region corresponding to the kinase domain of Tor2-mTOR. By screening about 4 million transformants, we obtained three clones of TOR2-mTOR SL (suppression of lst8 ts ) mutants (TOR2-mTOR SL1 , TOR2-mTOR SL2 , and TOR2-mTOR SL3 ) that partially suppressed temperature-sensitive growth of the TM478 (data not shown). Each mutant has the following mutations: three amino acid substitutions in Tor2-mTOR SL1 (V2198A, L2216H, and L2260P; the number of the position is based on wild-type mTOR), two in Tor2-mTOR SL2 (A2290V, K2440R), and one in Tor2-mTOR SL3 (L2302Q). Because Tor2-mTOR SL1 most potently suppressed temperature sensitivity of TM478, we focused on this mutant in the following experiments.

mTOR Mutants Exhibit Enhanced Kinase Activity in Vitro-
To investigate whether three amino acid substitutions in Tor2-mTOR SL1 indeed enhanced kinase activity of mTOR, we conducted an in vitro kinase assay of the mTOR mutants using GST-4EBP1 as a substrate. The FLAG-tagged mTOR mutant carrying three mutations (V2198A, L2216H, and L2260P; FLAG-mTOR SL1 ) was transiently expressed in HeLa cells and then purified by immunoprecipitation using anti-FLAG beads in the mild lysis condition, which preserves the integrity of the mTOR complexes (6). Immunopurified FLAG-mTOR SL1 phosphorylated Thr-37/46 of GST-4EBP1, which are the sites phosphorylated by mTORC1, stronger than wild-type FLAG-mTOR ( Fig. 2, lanes 2 and 3), indicating that FLAG-mTOR SL1 possessed enhanced kinase activity.
Previously, we also isolated two Tor2 mutants with a point mutation within the FRB domain (I1957T or A1960V) that confer weak rapamycin resistance on S. cerevisiae. 5 Interestingly, Tor1 with a mutation at the homologous site (I1954V or A1957V), was recently identified as conferring caffeine resistance on S. cerevisiae and displaying enhanced kinase activity (29). Thus, we attempted to measure kinase activity of the mTOR mutants carrying the homologous mutation (I2017T or A2020V). FLAG-mTOR I2017T and FLAG-mTOR A2020V mutants exhibited higher activity than wild-type FLAG-mTOR or FLAG-mTOR SL1 (Fig. 2, lanes 2 and 4, and data not shown). Remarkably, introduction of both SL1 and I2017T mutations into FLAG-mTOR (FLAG-mTOR SL1ϩIT ) further increased the kinase activity (Fig. 2, lane 5). These results indicate that mutations in the kinase domain (V2198A, L2216H, and L2260P) and in the FRB domain (I2017T or A2020V) enhance the kinase activity of mTOR independently and, thus, had an additive effect on the kinase activity of mTOR. 5 T. Umeda and T. Maeda, unpublished information. The dotted line in each structure indicates the junction for creating the Tor2-mTOR chimera. The resultant Tor2-mTOR chimera with 2518 amino acid (a.a.) residues is composed of the N-terminal two-thirds of Tor2 segment (residues 1-1689, shaded bar) and the C-terminal one-third of mTOR segment (residues 1721-2549, open bar). B, the scheme of screening for the Tor2-mTOR chimera mutant with enhanced kinase activity. The DNA fragments corresponding to the kinase domain of mTOR was randomly mutagenized with error-prone PCR and introduced together with the acceptor plasmid into the lst8 ts strain (TM478). TOR2-mTOR plasmids harboring random mutations within the kinase domain were constructed via in vivo homologous recombination in yeast cells. The transformants that grew under nonpermissive temperature were selected as candidates. The mTOR Mutants Hyperactivate the mTORC1 Pathway-Next, we investigated the effect of expressing these mTOR mutants on the mTORC1 pathway in mammalian cells. We co-expressed HA-S6K1 as a reporter for the mTORC1 activity together with various FLAG-mTOR mutants in HeLa cells. The phosphorylation status of HA-S6K1 during amino acid starvation was examined using an antibody specific to phospho-Thr-389 of S6K1. Phosphorylation levels of Thr-389 was rapidly decreased in response to amino acid starvation in cells expressing wild-type FLAG-mTOR (Fig. 3A, lanes 1-4). In contrast, in cells expressing FLAG-mTOR mutants, phosphorylation of Thr-389 under the amino acid-starved condition stayed higher. The resistance to dephosphorylation correlated well with the strength of in vitro kinase activity of the mTOR mutants (Fig.  3A, lanes 5-8, 9 -12, and 13-16). Especially, in cells expressing FLAG-mTOR SL1ϩIT , which displayed the highest kinase activity in vitro, phosphorylation of HA-S6K1 was maintained at a considerable level even after 3 h of amino acid starvation (Fig.  3B, lanes 13-16). These results indicate that the mTOR mutants activate the mTORC1 pathway under the amino acidstarved condition. However, because HA-S6K1 was gradually dephosphorylated even in cells expressing FLAG-mTOR mutants as the starvation period became longer, these mTOR mutants are unlikely to activate the mTORC1 pathway constitutively. These mTOR mutants are also sensitive to rapamycin. Rapamycin treatment induced dephosphorylation of HA-S6K1 both in cells expressing FLAG-mTOR mutants and wildtype FLAG-mTOR despite the higher initial phosphorylation levels in the former cells ( Fig. 3C and data not shown).
The effect of expressing the mTOR mutants on endogenous mTORC1 substrates was also investigated. We used pIRES-puro3-FLAG-mTOR, in which both FLAG-mTOR and puromycin-Nacetyltransferase were translated bicistronically from a single mRNA using an internal ribosomal entry site (IRES). Thus, cells transiently expressing FLAG-mTOR could be selected by puromycin. HeLa cells were transfected with pIRES-puro3-FLAG-mTOR or its mutant variants and grown in a puromycin-containing medium. Selected transfectants were then incubated in an amino acid-deprived medium. During amino acid starvation, the phosphorylation levels of endogenous S6K and 4EBP1 were maintained higher in FLAG-mTOR mutants expressing cells as compared with that in wild-type FLAG-mTOR-expressing cells (Fig. 3D, lanes 1-4 versus 5 -8, 9 -12, and 13-16). Especially in FLAG-mTOR SL1ϩIT -expressing cells, phospho-Thr-37/46 of 4EBP1 appeared to be retained nearly completely 60 min after amino acid deprivation (Fig. 3D, lanes  13-16). Furthermore, even in the amino acid-repleted condition (time 0), phosphorylation level of S6K was higher in FLAG-mTOR SL1ϩIT expressing cells than that in wild-type FLAG-mTOR-expressing cells (Fig. 3D, lanes 1 and 13). Taken together, our mTOR mutants could maintain phosphorylation of endogenous mTORC1 substrates as well as exogenously expressed HA-S6K1 under the amino acid-starved condition.
Most recently, Tamanoi and co-workers (30) have reported mTOR mutants that exhibit constitutive activation under nutrient-starved conditions. One of these mutants, carrying the E2419K mutation in the kinase domain, exhibited remarkable activity under the amino acid-starved condition in vivo and in vitro. To compare the mutant with our mutants, we introduced  NOVEMBER 14, 2008 • VOLUME 283 • NUMBER 46

JOURNAL OF BIOLOGICAL CHEMISTRY 31865
E2419K amino acid substitution into FLAG-mTOR (FLAG-mTOR E2419K ) and examined the activity toward the phosphorylation of co-expressed HA-S6K1. As reported previously, in FLAG-mTOR E2419K expressing cells, HA-S6K1 showed resistance to dephosphorylation under the amino acid-starved condition. However, in our assay condition, the phosphorylation level of HA-S6K1 was gradually decreased during amino acid starvation, similar to our mutants (Fig. 3E, lanes 13-15), and thus, mTOR E2419K was by no means constitutively active. In addition, like our FLAG-mTOR SL1ϩIT , FLAG-mTOR EKϩIT , which possessed both E2419K and I2017T mutations, also exhibited higher activity than the original mutants (Fig. 3E,  lanes 16 -18). Importantly, FLAG-mTOR SL1ϩIT showed higher activity toward the phosphorylation of HA-S6K1 than FLAG-mTOR E2419K or FLAG-mTOR EKϩIT . Thus, mTOR SL1ϩIT would be more desirable to investigate the impact of mTORC1 hyperactivation on cellular functions.

Hyperactive mTOR Exhibits Modestly Enhanced Kinase Activity toward Akt in Vitro, Whereas It
Exerts a Limited Effect on the mTORC2 Pathway in Cells-Because mTOR functions as mTORC2 to regulate actin organization and phosphorylation of Akt1 Ser-473, we examined whether mTOR SL1ϩIT could also hyperactivate the mTORC2 pathway. To this end we measured the phosphorylation level of Akt1 at Ser-473 as an indicator of mTORC2 activity. First, we measured in vitro kinase activity toward Akt1. Immunoprecipitated FLAG-mTOR SL1ϩIT phosphorylated Ser-473 of Akt1 stronger, although modestly, than wild-type FLAG-mTOR (Fig. 4A, lanes 2 and 4). Next, we expressed FLAG-mTOR SL1ϩIT in HeLa cells and assessed phosphorylation of endogenous Akt under a serum-starved condition. Unexpectedly, in FLAG-mTOR SL1ϩITexpressing cells, the phosphorylation of endogenous Akt was hardly retained under the starved condition (Fig. 4B, lanes 1-3 versus 4 -6). In contrast to Akt, the phosphorylation of endogenous S6K stayed higher in FLAG-mTOR SL1ϩIT -expressing cells as compared with that in wildtype FLAG-mTOR-expressing cells (Fig. 4B, lanes 1-3 versus 4 -6). Thus, the effect of hyperactive mTOR is likely limited to the mTORC1 pathway, although hyperactive mTOR exhibited enhanced kinase activity toward Ser-473 of Akt1 in vitro. One possible explanation for the different effects of the active mutant on mTORC1 and mTORC2 activities is it may arise from different assemblies of two mTOR complexes of this mutant. To address this possibility, complex formation of FLAG-mTOR SL1ϩIT was compared with that of wild-type FLAG-mTOR. Immunoprecipitates of FLAG-mTOR SL1ϩIT and wild-type FLAG-mTOR contained similar amount of Raptor, Rictor, and mLst8 (Fig. 4C), indicating that the disparity in activation of mTORC1 and mTORC2 by the active mutant cannot be attributed to different effects on assembly of two mTOR complexes. Another possible explanation for the limited activation of Akt is that even if mTORC2 activity is enhanced, decreased plasma membrane phosphatidylinositol 1,4,5-trisphosphate caused by serum starvation renders Akt unavailable for phosphorylation by mTORC2. Consistently, when Akt1 was overexpressed, under the condition where restriction in   Fig. 2 and under "Experimental Procedures." As a negative control, kinase reaction was performed with anti-FLAG beads only (Buffer). The phosphorylation level of Akt1 was examined by using anti-phospho-Akt(Ser-473) antibody. IP, immunoprecipitation. B, the effect of the active mTOR mutant on endogenous Akt. HeLa cells were transfected with pIRES-puro3-FLAG-mTOR (WT) or pIRES-puro3-FLAG-mTOR SL1ϩIT (SL1ϩIT) and selected by puromycin as described in Fig. 3D. After 24 h, puromycin-resistant cells were transferred into fresh growth medium for 4 h and incubated in serum-free DMEM (SerumϪ) as indicated times. Cells were lysed with lysis buffer, and lysate was subjected to immunoblotting using the indicated antibodies. C, the complex formation of the active mTOR mutant. HeLa cells were transfected with the expression plasmids for wild-type FLAG-mTOR (WT) or FLAG-mTOR SL1ϩIT (SL1ϩIT). After 48 h immunoprecipitation was performed using anti-FLAG beads, and co-precipitated proteins were subjected to immunoblotting using the indicated antibodies. D, the effect of the active mTOR mutant on overexpressed Akt1-HA. HeLa cells were co-transfected with the expression plasmids for Akt1-HA and wild-type FLAG-mTOR (WT) or FLAG-mTOR SL1ϩIT (SL1ϩIT). After 48 h, cells were transferred into fresh growth medium for 4 h and incubated in serum-free DMEM (SerumϪ) at the indicated times. Cells were lysed with lysis buffer, and lysate was subjected to immunoblotting using the indicated antibodies.
availability of Akt1 for phosphorylation could be relieved, its phosphorylation was significantly sustained in cells expressing the active mutant even after serum starvation (Fig. 4D).
Expression of the Hyperactive mTOR SL1ϩIT Mutant Increases Cell Size-The observation that the mTOR SL1ϩIT mutant hyperactivates the mTORC1 pathway prompted us to evaluate the roles of mTOR in the regulation of cellular functions using gain-of-function analyses. First, we examined the cell size of hyperactive mTOR-expressing cells under nutrient-repleted conditions. HeLa cells were transfected with pIRES-puro3-FLAG-mTOR SL1ϩIT or pIRES-puro3-FLAG-mTOR and selected by puromycin for 24 h. Then, the transfected cells were further incubated in DMEM with 10% FBS without puromycin for an additional 24 h and subjected to cell size determination. Cell size distribution of FLAG-mTOR SL1ϩIT -expressing cells was shifted rightward compared with that of wild-type FLAG-mTOR-expressing cells, and the mean cell diameter was larger in FLAG-mTOR SL1ϩIT -expressing cells (Fig. 5). We further investigated the effect of mTOR hyperactivation under the amino acid-starved condition. A modified DMEM containing amino acids mixture at 25% relative to that in complete DMEM was used as the amino acid-starved condition. Amino acid star-vation caused significant reduction in cell size in wild-type FLAG-mTOR-expressing cells (Fig. 5). In FLAG-mTOR SL1ϩITexpressing cells, however, cell size reduction was not as drastic as in wild-type FLAG-mTOR-expressing cells (Fig. 5), suggesting that cell size reduction in the starvation condition is dependent on inactivation of mTOR. These results demonstrated that mTOR is a central regulator of cell size.
Expression of the Hyperactive mTOR SL1ϩIT Mutant Suppresses Starvation-induced Autophagy-Autophagy is induced in response to rapamycin treatment, suggesting that mTORC1 negatively regulates autophagy induction (31). Autophagy is also induced in response to starvation, although direct evidence that starvation induces autophagy through inactivating mTORC1 remains to be obtained. In addition, the possibilities of mTOR-independent pathways in starvation-induced autophagy were also reported (32,33). To clarify the involvement of mTOR in starvation-induced autophagy, we examined the effect of hyperactivation of mTOR on autophagy induction. To monitor the degree of autophagy, we devised LC3 immunoblotting (34). LC3 exists as the cytosolic form LC3-I and the membrane-bound form LC3-II, and LC3-I is converted to LC3-II as autophagy proceeds. Because the amount of LC3-II is closely correlated with the number of autophagosome, it is a good indicator for autophagy (35). In this assay, we used HEK293 cells instead of HeLa cells because the amount of LC3-II can be more easily evaluated in HEK293 cells. Because LC3-II itself is also degraded by autophagy, amino acid starvation was performed with or without lysosomal protease inhibitors (pepstatin A and E64-d) to assess autophagy flux more accurately (34). LC3-I was detected very weakly in this experimental setting, probably because LC3-I is less immunoreactive to anti-LC3 antibody than LC3-II in most cases (34). Nevertheless, in wild-type FLAG-mTOR-expressing cells, accumulation of LC3-II was clearly observed during amino acid starvation in the presence of inhibitors (Fig. 6, lanes 7-9). In HEK293 cells were transfected with pIRES-puro3-FLAG-mTOR (WT) or pIRES-puro3-FLAG-mTOR SL1ϩIT (SL1ϩIT) and transfected cells were selected as in Fig.  3D. Puromycin-resistant cells were transferred into amino acid-deprived DMEM (AAϪ) in the presence or absence of protease inhibitors (10 g/ml pepstatin A and 10 g/ml E64-d) and incubated for indicated times. The amount of endogenous LC3-I and LC3-II was examined by immunoblotting using anti-LC3 antibody. contrast, in FLAG-mTOR SL1ϩIT -expressing cells, accumulation of LC3-II was less obvious (Fig. 6, lanes 10 -12). Thus, mTOR hyperactivation under the starvation condition rendered cells resistant to autophagy induction, demonstrating that autophagy was induced by starvation at least partly through mTOR inactivation.
Hyperactive mTOR Does Not Induce Cellular Transformation in NIH/3T3 Cells-To determine whether mTOR hyperactivation could contribute to tumorigenesis, we performed a transformation assay using NIH/3T3 cells. NIH/3T3 cells were transfected with expression plasmids for wild-type FLAG-mTOR and FLAG-mTOR SL1ϩIT . We also transfected the cells with an empty vector and an expression plasmid for the oncogenic mutant of RAS (H-RAS 12V ) as a negative and a positive control, respectively. The number of foci of transformed cells in un-selected cultures was counted, and the transfection efficiency for each expression plasmid was determined by counting G418-resistant colonies. The plasmid expressing H-RAS 12V induced a reasonable number of transformed foci (Table 1). In contrast, the plasmid expressing either wild-type FLAG-mTOR or FLAG-mTOR SL1ϩIT failed to induce conspicuous foci above the level found in the cells transfected with the empty vector. Because transfection efficiencies were in a similar range among the expression plasmids, this result suggests that hyperactive mTOR itself is not transforming to NIH/3T3 cells.

DISCUSSION
In this report we have isolated hyperactive mutants of mTOR (mTOR SL1 , mTOR SL2 , and mTOR SL3 ) through screening for Tor2-mTOR chimera mutants that suppress temperature-sensitive growth of an lst8 ts strain. We also revealed that I2017T and A2020V substitutions in the FRB domain enhance mTOR kinase activity, whose homologous mutations were previously isolated from independent screening by Powers et al. in Tor1 (29) and by our group in Tor2. 5 Remarkably, we successfully obtained a potent activating mutant of mTOR by the combinatorial introduction of SL1 and I2017T mutations.
Interestingly, yeast mutants with homologous mutations are able to overcome TORC1 inactivation caused by a wide array of means. Mutations in Tor1 or Tor2 at sites homologous to Ile-2017 and Ala-2020 of mTOR confer S. cerevisiae cells resistance to both caffeine and rapamycin, and that to Val-2198, one of the mutation sites in the mTOR SL1 mutant, also corresponds to the site in a Tor2 mutant allowing Rhb1-independent growth in Schizosaccharomyces pombe (30). We propose that enhanced kinase activity of TOR contributes in common to overcome TORC1 inactivation by divergent causes, although direct evidence has to be demonstrated.
What, then, is the mechanism underlying the enhanced kinase activity in our mTOR mutants? In the case of mTOR SL1 , we found that two mutations (V2198A and L2216H) are important for the enhanced kinase activity (data not shown). Interestingly, these two mutations were located on the same ␣-helix when the three-dimensional structure of the mTOR kinase domain was modeled based on the similarity to phosphatidylinositol-3-phosphate kinase ␥ (not shown). This ␣-helix corresponds to the ␣C helix conserved in the structures of kinase domains, which is known to be an important mediator of conformational changes in a kinase catalytic center (36). In the fibroblast growth factor receptor 2 (FGFR2) kinase, a naturally occurring activating mutation (K526E) located on ␣C leads to kinase activation by disrupting the "molecular brake" in the kinase hinge region as revealed by a structural analysis (37). Thus, the mutations located on the ␣-helix of mTOR corresponding to ␣C might contribute to the enhanced kinase activity in a similar manner. A few activating mutations on the same putative ␣-helix were also isolated in Tor2 of S. pombe, confirming the importance of this region on the kinase activity of TOR (30). Although it also resides in the kinase domain, the E2419K mutation seems to activate mTOR in a different way. The E2419K mutation is just adjacent to the repressor domain and, thus, was proposed to relief repression of the kinase activity mediated by this domain (30). In contrast to these mutations in the kinase domain, the I2017T mutation is in the FRB domain. Reinke et al. reported that a mutation at the homologous site in Tor1 increased Kog1 binding (29). However, in our coprecipitation experiment, the interaction between Raptor and FLAG-mTOR I2017T or FLAG-mTOR SL1ϩIT was similar to wild-type FLAG-mTOR ( Fig. 4C and data not shown). Furthermore, we found that the kinase activity of FLAG-mTOR I2017T was higher than wild-type FLAG-mTOR even if Raptor was removed from the mTOR complex by washing immunoprecipitated mTOR with a Triton X-100-containing buffer (data not shown). Because Ile-2017 is located within the binding region to FKBP38 (38), it is possible that the association with FKBP38 might be changed in this mTOR mutant. Further investigation is needed to clarify the mechanism by which the I2017T mutation causes the enhanced kinase activity.
Irrespective of the mode of action, our hyperactive mTOR mutants would be a useful tool for investigating the direct biological function of mTOR because of its potent ability to activate the mTORC1 pathway as evidenced by the prolonged phosphorylation of S6K and 4EBP1 under the amino acidstarved condition (Fig. 3). We re-evaluated the involvement of mTOR activity in previously attributed cellular functions from gain-of-function aspects. In hyperactive mTOR-expressing cells, cell size was enlarged, which is consistent with a previous report that overexpression of S6K1 or eIF4E resulted in cell size enlargement (39). Notably, hyperactivation of the mTORC1 pathway partially interfered with cell size reduction caused by amino acid starvation, suggesting that inactivation of mTORC1 is a requisite for cell size reduction. These observations confirmed that the cell controls its own size in response to environmental nutrient availability through the mTORC1 activation and inactivation and further emphasized the importance of mTOR as the primary regulator of cell size.
It was uncertain whether starvation-induced autophagy is promoted through mTOR so far. Rapamycin treatment induced autophagy in COS-7 cells (40), whereas leucine-deprivation and rapamycin treatment has an additive effect on autophagy induction in C2C12 cells (32). Furthermore, Kanazawa et al. (33) reported that rapamycin had no effect on the suppression of autophagy by amino acid addition in isolated hepatocytes. Our result from the gain-of-function approach argues that inactivation of mTOR is essential for autophagy induced by amino acid starvation. Slight accumulation of LC3-II observed in hyperactive mTOR expressing cells might suggest that another pathway also controls autophagy in parallel with the mTOR pathway, although it might also be due to slight inactivation of mTORC1 activity in hyperactive mTOR expressing cells. mTOR activation has been implicated in tumorigenesis because rapamycin effectively prevents cellular transformation caused by hyperactivation of the phosphatidylinositol-3-phosphate kinase-Akt pathway (41). Overexpression of eIF4E, which acts as a downstream effector of mTORC1 and plays a key role in mRNA translation, also caused cellular transformation (42). Furthermore, the tumor suppressors TSC1/2 and LKB1 negatively regulate mTORC1 activity (43)(44)(45)(46). These observations suggest that mTORC1 up-regulation is sufficient to induce cellular transformation by itself. However, our observation that the hyperactive mTOR mutant did not induce transformation of NIH/3T3 cells argues that concomitant activation of mTOR with additional pathways such as the Akt-dependent survival pathway rather than the sole activation of mTOR is required for cellular transformation. One possible reason for this is the feedback inhibition of the phosphatidylinositol-3-phosphate kinase-Akt pathway by hyperactivation of mTORC1. In fact, it has been previously demonstrated that hyperactivation of mTORC1 in the TSC-related tumors showed only a modest phenotype but resulted in more severe tumor if reactivation of the phosphatidylinositol-3-phosphate kinase-Akt pathway occurs by additional mutation in PTEN (47).
Unexpectedly, the hyperactive mTOR mutant was not able to fully activate the mTORC2 pathway. In hyperactive mTOR mutant expressing cells, endogenous Akt was dephosphorylated under the serum-starved condition in a similar manner as in wild-type mTOR-expressing cells, although in vitro kinase activity toward Akt1 was increased in hyperactive mTOR compared with wild-type mTOR. In contrast, endogenous S6K was resistant to the dephosphorylation under the same conditions, indicating that mTORC1 and mTORC2 substrates are differently regulated and the mechanisms to dephosphorylate Akt1 are able to overcome enhanced kinase activity of mTORC2. Recently, Sabatini and co-workers demonstrated that in vitro kinase activity of mTORC2 was enhanced by insulin treatment, suggesting that the modification of kinase activity might control the Akt phosphorylation (21). In contrast, earlier reports indicated that phosphorylation of Ser-473 of Akt1 was dominantly controlled by the membrane-translocation of Akt1 through the production of phosphatidylinositol 3,4,5-trisphosphate, and thus, the kinase responsible for the phosphorylation of Ser-473 of Akt1 appears to be localized at the plasma membrane and constitutively active (48). Our result is consistent with the latter observation. Because Akt1 dissociates from the plasma membrane by a decrease in phosphatidylinositol 1,4,5-trisphosphate during serum starvation regardless of the kinase activity of mTORC2, Ser-473 of Akt1 could, thus, be dephosphorylated in the hyperactive mTOR-expressing cells. It remains possible, however, that hyperactive mTOR did not have enough activity to maintain the phosphorylation of Akt under serum-starved conditions, because we observed that the enhancement of the kinase activity of hyperactive mTOR toward Akt1 was much smaller than that toward GST-4EBP1. It is also possible that alteration of kinase activity of mTORC2 could selectively regulate the phosphorylation of substrates other than Akt. The effect of hyperactive mTOR on the other mTORC2 substrates including PKC␣ (8) is to be examined in a future study.
In conclusion, we obtained a hyperactive mTOR mutant that is able to activate the mTORC1 pathway more potently than the activated mTOR mutants previously reported (Fig. 3E) (30). This mutant proved to be a useful tool for re-evaluation of mTOR function from the gain-of-function approaches as well as to probe pathway epistasis and identify novel effectors of the mTOR pathway. In addition, we are now generating transgenic animals that express the hyperactive mTOR. The analyses of the animals will provide crucial knowledge on the role of mTOR in the nutrient response of the whole organism as well as that in tumorigenesis.