Nutrient-dependent Multimerization of the Mammalian Target of Rapamycin through the N-terminal HEAT Repeat Region*

The mammalian target of rapamycin (mTOR) plays a pivotal role in the regulation of cell growth in response to a variety of signals such as nutrients and growth factors. mTOR forms two distinct complexes in vivo. mTORC1 (mTOR complex 1) is rapamycin-sensitive and regulates the rate of protein synthesis in part by phosphorylating two well established effectors, S6K1 (p70 ribosomal S6 kinase 1) and 4E-BP1 (eukaryotic initiation factor 4E-binding protein 1). mTORC2 is rapamycin-insensitive and likely regulates actin organization and activates Akt/protein kinase B. Here, we show that mTOR forms a multimer via its N-terminal HEAT repeat region in mammalian cells. mTOR multimerization is promoted by amino acid sufficiency, although the state of multimerization does not directly correlate with the phosphorylation state of S6K1. mTOR multimerization was insensitive to rapamycin treatment but hindered by butanol treatment, which inhibits phosphatidic acid production by phospholipase D. We also found that mTOR forms a multimer in both mTORC1 and mTORC2. In addition, Saccharomyces cerevisiae TOR proteins Tor1p and Tor2p also exist as homomultimers. These results suggest that TOR multimerization is a conserved mechanism for TOR functioning.

The target of rapamycin (TOR) 2 is a highly conserved Ser/ Thr protein kinase that regulates cell growth from yeast to mammals (1,2). This protein was originally identified by the analysis of mutants that show resistance to rapamycin in Sac-charomyces cerevisiae (3). Although two highly homologous genes (TOR1 and TOR2) exist in yeast, only one TOR gene is known in higher eukaryotes (1,2). Mammalian TOR (mTOR) is a large protein with a molecular mass of 289 kDa and belongs to the phosphatidylinositol-3-phosphate kinase-related kinase (PIKK) family, which includes ATM, ATR, and DNA-dependent protein kinase (1). PIKK family proteins have characteristic domains: at their N-terminal halves, tandem HEAT (Huntingtin, elongation factor 3, A subunit of protein phosphatase 2A, and TOR1) repeats (4); and at their C-terminal regions, FAT, kinase catalytic, and FATC domains (5). In the case of mTOR, there are ϳ20 tandem HEAT repeats, which are believed to mediate protein-protein interaction. Indeed, several proteins that interact with this region have been identified (6 -9). Between FAT and kinase catalytic domains, mTOR has the FRB domain, to which rapamycin, a specific inhibitor of mTOR, binds in complex with FKBP12 (FK506-binding protein 12) (1).
To date, both in yeast and mammalian cells, TOR is known to form two distinct complexes (TORC1 and TORC2), each of which has different roles (10 -13). In mammalian cells, mTORC1, which consists of mTOR, mLST8 (also named G␤L), and raptor, is sensitive to rapamycin and regulates the phosphorylation of two well established effectors, S6K1 (p70 ribosomal S6 kinase 1) and 4E-BP1 (eukaryotic initiation factor 4E-binding protein 1), leading to the promotion of protein synthesis (7,14,15). In the other complex known as mTORC2, mTOR is associated with mLST8 and rictor (also named mAVO3). This complex is insensitive to rapamycin and appears to regulate actin organization (12,13). In addition, it was reported recently that mTORC2 is responsible for the phosphorylation of Akt/ protein kinase B at Ser 473 (16). Although rapamycin has been used as a specific inhibitor of mTOR activity, it has been revealed that rapamycin binds to mTOR in mTORC1 but not in mTORC2 (12,13), suggesting that mTOR may contribute to more cellular functions than previously thought.
The activity of mTORC1 is regulated by a variety of signals such as amino acid availability, growth factors, and energy status. Recently, it was revealed that tuberous sclerosis tumor suppressor proteins TSC1 and TSC2 act as upstream negative regulators of mTORC1 in signaling from growth factors and energy status (1,2). TSC2 exerts its negative regulation on mTOR through activity as a GTPase-activating protein on the small GTP-binding protein Rheb (17)(18)(19), which directly interacts with and positively regulates mTOR (20). Insulin stimulation leads to Akt/protein kinase B activation, which is thought to inactivate the GTPase-activating protein activity of TSC2 by undefined mechanisms. Under energy starvation conditions, AMP-activated protein kinase is activated and then enhances TSC2 activity (21). On the other hand, the mechanism by which amino acid availability leads to mTOR activation is less understood. Although the regulation of mTOR by amino acid availability is unlikely to require TSC2 (22), Rheb overexpression could prevent S6K1 dephosphorylation caused by amino acid withdrawal (17)(18)(19), suggesting that amino acid availability may lead to the modulation of Rheb independently of the TSC1/2 complex. Recent studies also indicated the involvement of human VPS34 in amino acid-mediated signaling to mTOR independently of the TSC-Rheb pathway (23,24). In contrast to mTORC1, no signal leading to mTORC2 activation has been identified.
Multimerization is a well documented general mechanism for the activation of protein kinases. ATM, a member of the PIKK family, forms a dimer, and dimerization is important for its regulation (25). In this study, we show that mTOR exists as a multimer via its N-terminal HEAT repeat region in mammalian cells and that this multimerization is promoted by amino acid sufficiency but hindered by butanol treatment, which inhibits phosphatidic acid (PA) production by phospholipase D (26). The state of mTOR multimerization does, however, not directly correlate with the phosphorylation state of S6K1. mTOR is likely to multimerize within both mTORC1 and mTORC2. We also show that S. cerevisiae TOR proteins Tor1p and Tor2p also exist as homomultimers. These results indicate that TOR multimerization is conserved through evolution and might be important for TOR functioning.
Mammalian Cell Culture and Transfection-COS-7, HEK293T, and HeLa cells were grown at 37°C in Dulbecco's modified Eagle's medium (DMEM) with 1000 or 4500 mg/liter glucose (Sigma) supplemented with 10% fetal bovine serum under 5% CO 2 atmosphere. Transient transfection of the indicated plasmids was performed with PolyFect reagent (Qiagen, Hilden, Germany) according to the manufacturer's instructions. In these assays, the transfection efficiencies were estimated to 30 -40%, and the total expression levels of recombinant mTOR proteins were usually Ͻ3-fold compared with those of endogenous mTOR. For amino acid starvation, cells were washed once with Dulbecco's phosphate-buffered saline (D-PBS) containing 100 mg/liter each CaCl 2 and MgCl 2 and 1000 or 4500 mg/liter glucose (according to DMEM used for cell cultures) and incubated in the same saline for the indicated times. In experiments for the re-addition of amino acids, an amino acid mixture was added at the following final concentrations: L-Arg, 84 mg/liter; L-Cys, 48 mg/liter; L-His, 84 mg/liter; L-Ile, 105 mg/liter; L-Leu, 105 mg/liter; L-Lys, 145 mg/liter; L-Met, 30 mg/liter; L-Phe, 66 mg/liter; L-Thr, 95 mg/liter; L-Trp, 20 mg/liter; L-Tyr, 72 mg/liter; L-Val, 94 mg/liter; and L-Gln, 584 mg/liter.
Cell Lysate Preparation and Immunoprecipitation-After 24 -48 h of transfection, cells grown on 6-well plates were washed once with ice-cold D-PBS and lysed with 0.3 ml of lysis buffer (40 mM HEPES-NaOH (pH 7.5), 120 mM NaCl, 1 mM EDTA, 0.3% CHAPS, 50 mM NaF, 10 mM ␤-glycerophosphate, 10 g/ml aprotinin, and 10 g/ml leupeptin). In the experiment shown in Fig. 1A, 1% Triton X-100 was substituted for 0.3% CHAPS. After centrifugation at 13,000 ϫ g for 10 min, the supernatant was incubated with anti-FLAG M2 beads for 2 h. Immune complexes captured on the beads were washed three times with lysis buffer and resolved by SDS-PAGE. For sequential immunoprecipitation, cell lysate was prepared from a 100-mm dish using 1.0 ml of lysis buffer as described above. Cleared cell lysates were subjected to immunoprecipitation with anti-FLAG M2 beads for 2 h; the beads were washed three times with lysis buffer; and immune complexes were eluted with lysis buffer containing the 3X FLAG peptide (150 ng/ml). The elute was further immunoprecipitated with anti-Myc beads (Sigma) for 2 h, and the beads were washed four times with lysis buffer.
Gel Filtration Chromatography-About 4.0 ϫ 10 7 HEK293T or HeLa cells were lysed with 0.3 ml of lysis buffer on ice for 20 min. After centrifugation at 13,000 ϫ g for 10 min, the supernatant was passed through a 0.22-m filter (Millipore Corp., Bedford, MA). About 1.5-2.0 mg of proteins in Ͻ0.3-ml volume were loaded onto a Superose 6 HR 10/30 column (Amersham Biosciences) pre-equilibrated with lysis buffer. The proteins were eluted at 0.2 ml/min, and 0.5-ml fractions were collected. Sixteen microliters of each fraction were then analyzed by immunoblotting with the indicated antibodies.
Yeast Cell Culture and Immunoprecipitation-Yeast strains were grown to A 600 ϭ 1.0 at 30°C in YPD medium (yeast extract/peptone/dextrose). For nitrogen or carbon starvation, cells were washed once with water and cultured in synthetic dextrose medium without ammonium sulfate and supplemented with leucine and uracil or in synthetic complete medium without glucose, respectively. For other treatments, each reagent was added directly to YPD medium. Cells were harvested by centrifugation at 1500 ϫ g for 5 min and disrupted in yeast lysis buffer (D-PBS, 10% glycerol, 0.5% Tween 20, 50 mM NaF, 10 mM ␤-glycerophosphate, 1.5 mM Na 3 VO 4 , 40 g/ml aprotinin, 20 g/ml leupeptin, 10 g/ml pepstatin A, and 1 mM phenylmethylsulfonyl fluoride) using ceramics beads and a Multi-Beads Shocker (Yasui Kikai Corp., Osaka) 10 times for each 30-s pulse. After centrifugation at 13,000 ϫ g for 10 min, the supernatants were incubated with anti-FLAG M2 beads for 2 h at 4°C. Immune complexes were washed three times with yeast lysis buffer and resolved by SDS-PAGE.

RESULTS
mTOR Exists as a Multimer-To examine whether mTOR exists as a multimer, Myc epitope-tagged mTOR and FLAG epitope-tagged mTOR were transiently coexpressed in mammalian cells. As shown in Fig. 1A, Myc-mTOR was co-immunoprecipitated with FLAG-mTOR, indicating that mTOR forms a multimer in the cell. The endogenous mTOR-associated proteins raptor, rictor, and mLST8 (7,(12)(13)(14)(15) were also co-immunoprecipitated with FLAG-mTOR (data not shown). Similar to mTOR-raptor interaction (7,14), mTOR-mTOR interaction was also disrupted when we used lysis buffer containing Triton X-100 as a detergent (Fig. 1A). To determine the region involved in the multimerization of mTOR, a series of truncation mutants of mTOR with a FLAG epitope tag were coexpressed with Myc-mTOR in mammalian cells, and their ability to interact with Myc-mTOR was examined by co-immunoprecipitation analysis (Fig. 1, B and C). A FLAG-tagged N-terminal region (fragment 1-1484) of mTOR was able to  1C). On the other hand, shorter deletion mutants of the N-terminal region (fragments 1-1380, 1-670, and 575-1380) still inter-acted with recombinant full-length mTOR (Fig. 1C). Reciprocal immunoprecipitation with Myc-mTOR showed similar results (Fig. 1D). Another PIKK protein containing the HEAT repeat region, ATR (31), did not interact with Myc-mTOR (Fig. 1E), indicating that the interaction is specific. These results suggest that mTOR is able to multimerize through its N-terminal HEAT repeat region. mTOR Multimerization Is Affected by Amino Acid Sufficiency and Butanol Treatment-We next investigated whether mTOR multimerization is affected by amino acid starvation. Cells transiently transfected with expression plasmids for both Myc-mTOR and FLAG-mTOR were first maintained in DMEM containing 10% serum, and then the medium was replaced with D-PBS for 150 min before harvesting. The amount of Myc-mTOR co-immunoprecipitated with FLAG-mTOR was greatly reduced by this treatment ( Fig.  2A, lanes 1 and 6). The addition of amino acids to D-PBS restored the interaction (lane 2). These results suggest that mTOR multimerization is modulated by extracellular amino acid sufficiency. Insulin is another known input leading to S6K1 phosphorylation at Thr 389 through the activation of mTORC1 signaling, although this action of insulin requires extracellular amino acids (32,33). Consistent with the action of insulin for S6K1 phosphorylation, insulin alone did not enhance mTOR multimerization (lane 3). Rapamycin treatment was found to have no effect on mTOR multimerization at the concentration inhibiting S6K1 phosphorylation (lanes 4 and 5).
Similar but more marked changes were observed using N-terminal fragment 1-1380, for which multimerization was considerably more robust compared with full-length mTOR (Fig. 2B). Kinase-dead mTOR (27) also responded to amino acid sufficiency (Fig. 2C). Thus, the N-terminal HEAT repeat region is sufficient for modulating multimerization in response to amino acid sufficiency, and the kinase activity of mTOR is not required for the modulation. Both FLAG-tagged fragments 1-670 and 575-1380 were able to interact with Myc-tagged fragment 1-1380, but these interactions were less sensitive to amino acid FIGURE 1. mTOR forms a multimer via the HEAT repeat domain. A, COS-7 cells were cotransfected with the expression plasmids for Myc-mTOR and FLAG-mTOR. After 24 h, the cells were lysed with lysis buffer containing 0.3% CHAPS or 1% Triton X-100. FLAG-mTOR was immunoprecipitated (IP), and the immune complexes were analyzed by immunoblotting using the indicated antibodies. B, shown is a schematic representation of domains in mTOR. The regions of mTOR fragments used in C and D are shown below. C, COS-7 cells were transfected with the expression plasmids for FLAG-tagged mTOR fragments and Myc-tagged full-length mTOR. After the recovery of cell lysates, the interaction between FLAG-tagged fragments and Myc-mTOR was analyzed as described for A. D, HEK293T cells were transfected with the expression plasmids for Myc-tagged mTOR fragments and FLAG-tagged full-length mTOR. After the recovery of cell lysates, the interaction between Myc-tagged fragments and FLAG-mTOR was analyzed as described for A, except that the antibodies for immunoprecipitation and immunoblotting were switched. E, COS-7 cells were transfected with the expression plasmid for Myc-mTOR or FLAG-ATR alone or together and analyzed as described for A. The results from control immunoprecipitation with cell lysates that had been transiently transfected with Myc-mTOR and FLAG-mTOR are also shown.

mTOR Multimerization via the HEAT Repeat Region
withdrawal (Fig. 2D). Full-length N-terminal HEAT repeats are therefore likely to be important for fully mediating the modulation of mTOR multimerization in response to amino acid sufficiency.
To identify other conditions affecting mTOR multimerization, we tested several treatments that were already reported to reduce S6K1 and 4E-BP1 phosphorylation (18, 21, 26, 34). The addition of 2-deoxyglucose, which leads to AMPactivated protein kinase activation, did not significantly affect mTOR multimerization (Fig. 3A, lane 5). On the other hand, the addition of 1-butanol significantly and that of 2-butanol weakly reduced mTOR multimerization (Fig. 3A, lanes 6 and 7). The different effects of 1-and 2-butanol on mTOR multimerization appear to be physiologically relevant because 1-butanol more strongly inhibits PA production compared with 2-butanol (26). Again, similar results were observed for mTOR N-terminal fragment 1-1380 (Fig. 3B). Although PA has been shown to bind directly to the FRB domain of mTOR and to activate mTOR signaling (26), our results suggest the possibility that PA also exerts a regulatory effect on mTOR through other mechanisms.
The State of mTOR Multimerization Does Not Directly Correlate with the Phosphorylation State of S6K1-To address the correlation between mTOR multimerization and S6K1 phosphorylation at Thr 389 , we examined the time courses of both events in response to changes in amino acid sufficiency. Phosphorylation of S6K1 at Thr 389 has been extensively used as a readout of mTORC1 activity both in vivo and in vitro (7,32). Amino acid withdrawal gradually decreased Myc-mTOR co-immunoprecipitation with FLAG-mTOR over 90 min, whereas S6K1 phosphorylation at Thr 389 was almost completely diminished after 30 min (Fig. 4A). Conversely, mTOR multimerization increased up to 60 min after changing the medium from D-PBS to DMEM containing 10% serum, although S6K1 phosphorylation occurred as quickly as in 5 min and peaked after 30 min (Fig. 4B). Thus, S6K1 phosphorylation by amino acid stimulation occurs prior to the induction of mTOR multimerization. In addition, using the coumermycin-GyrB (gyrase B)-induced dimerization system (35,36), we artificially induced dimerization of GyrB-mTOR fusion protein independently of amino acid availability. However, dimerization of GyrB-mTOR did not stimulate Thr 389 phosphorylation of S6K1 under amino acid-deprived conditions (Fig. 4C). Taken together, these results suggest that the state of mTOR multimerization does not directly correlate  1, 4, and 7), treated with amino acids in D-PBS (lanes 2, 5, and 8), or transferred into DMEM containing 10% serum (lanes 3, 6, and 9) for 60 min before harvesting. Immunoprecipitation assays were performed as described for A. SEPTEMBER 29, 2006 • VOLUME 281 • NUMBER 39 with the Thr 389 phosphorylation state of S6K1, although amino acid sufficiency promotes both events.

mTOR Multimerization via the HEAT Repeat Region
Multimeric mTOR Exists in Both mTORC1 and mTORC2-To examine whether mTOR in mTORC1 or mTORC2 is in a multimeric state, we performed two sequential immunoprecipitations with cell lysates from COS-7 cells transiently expressing FLAG-mTOR with and without Myc-mTOR. First, FLAG-mTOR was immunoprecipitated using anti-FLAG M2 beads, and then the immunoprecipitated complex released from the anti-FLAG M2 beads was further immunoprecipitated with anti-Myc beads. After these sequential immunoprecipitations, endogenous raptor and rictor were still recovered with Myc-mTOR (Fig. 5A), suggesting the formation of ternary complexes composed of FLAG-mTOR, Myc-mTOR, and raptor or rictor. Similar results were also observed in HEK293T cells (data not shown). Thus, mTOR appears to form a multimer in both mTORC1 and mTORC2.
We also examined the effect of raptor overexpression on mTOR multimerization. The overexpression of Xpress-raptor together with Myc-mTOR and FLAG-mTOR had only a marginal effect on the multimerization of mTOR under amino acidreplete conditions (Fig. 5B, lanes 2, 4, and 6). This result indicates that raptor itself does not likely mediate the formation of multimeric mTOR. On the other hand, the overexpression of Xpress-raptor, which would increase the ratio of the raptorassociated mTOR fraction to the rictor-associated one, led to mTOR multimerization being less sensitive to amino acid starvation in a dose-dependent manner (lanes 1, 3, and 5), suggesting that mTOR multimerization in the raptor-associated fraction, viz. mTORC1, might be insensitive to amino acid sufficiency.
Gel Filtration Analysis of mTOR-To investigate the change in the multimerization of endogenous mTOR in response to amino acid sufficiency and to determine which of the two complexes (mTORC1 or mTORC2) contributes to the change, we analyzed lysates from two representative cell lines by gel filtration chromatography: HEK293T and HeLa cells, which are abundant in mTORC1 and mTORC2, respectively (7,12). Fig. 6A shows the elution profiles of endogenous mTOR, raptor, and rictor in HEK293T cells. Endogenous mTOR was distributed mainly into three peaks: fractions 4 and 5 (Ͼ2 MDa; peak A), fractions 11-13 (ϳ1 MDa; peak B), and fractions 17 and 18 (ϳ0.5 MDa; peak C). Raptor distributed to peak B, and rictor distributed to peaks A and B as well as to some lower ranges (fractions 14 and 15), indicating that peak A contained mainly mTORC2, peak B contained mTORC1 and mTORC2, and peak C contained neither. As shown in Fig. 6B, the elution profile of endogenous mTOR in HeLa cells was quite different from that in HEK293T cells; in HeLa cells, a much larger amount of mTOR distributed to peak A and only a marginal amount to peak C. Raptor again distributed only to peak B, and rictor distributed mainly to peak A. The difference is consistent with previous observations that HEK293T cells contain more mTORC1 compared with HeLa cells and that HeLa cells have more mTORC2 compared with HEK293T cells (7,12).
To determine which of these peaks corresponds to a multimeric form, we used lysates from cells transiently expressing both Myc-mTOR and FLAG-mTOR. The elution profiles of both recombinant mTOR proteins were similar to that of endogenous mTOR (Fig. 6C). Co-immunoprecipitation analysis with three peak fractions revealed that multimeric mTOR was distributed mainly in peak A, to a lesser extent in peak B, and barely in peak C (Fig. 6D).
In response to amino acid starvation, the relative amount of mTOR in peak A consistently decreased in lysates from HeLa cells; the distribution of mTOR in peak A in comparison with peak B was larger in DMEM than in PBS (Fig. 6B). Rictor was also slightly shifted into lower molecular mass fractions in PBS compared with DMEM (Fig. 6B). These changes were not as evident in lysates from HEK293T cells (Fig. 6A). These results suggest that it is mTORC2 that is subjected to modulation in multimerization by amino acid sufficiency, which is consistent with the above observation indicating that mTOR multimerization in mTORC1 is insensitive to amino acid sufficiency. We also noticed increases in the relative amounts of mTOR at fractions 16 -18 and of raptor at fractions 18 -21 in DMEM for both HEK293T and HeLa cells lysates, although the relevance to mTOR multimerization remains unclear at present.
TOR Multimerization Is Conserved in Yeast-To investigate whether TOR multimerization is conserved through evolution, we constructed S. cerevisiae diploid strains expressing differently tagged TOR proteins at the N termini from each allele. Upon immunoprecipitation with anti-FLAG M2 beads, an association between HA 3 -Tor1p and FLAG 3 -Tor1p was detected (Fig. 7A, lane 2), and Tor2p-Tor2p interaction was similarly detected (Fig. 7B, lane 2). On the other hand, as reported previously (10,11), no interaction was observed between Tor1p and Tor2p (Fig. 7, A and B,  lanes 1). Thus, TOR multimerization is also conserved in yeast, and multimerization is restricted to homomultimerization, i.e. Tor1p-Tor1p or Tor2p-Tor2p. As Tor1p is able to form TORC1 but not TORC2 in yeast (10,11), the TOR protein in TORC1 is also multimeric in yeast.
We next examined whether TOR multimerization in yeast is also affected by changing culture conditions. As in mammalian cells, nutrient starvation has been demonstrated to inhibit TORC1 signaling in yeast (37); however, nitrogen or carbon starvation did not affect either Tor1p-Tor1p or Tor2p-Tor2p interaction (Fig. 7, A and B, lanes 3 and 4). Rapamycin treatment also did not change multimerization, as in mammalian cells (Fig. 7, A and B, lanes 5). We also tested the effects of 1-and 2-butanol on yeast TOR multimerization, although PA involvement in TOR signaling in yeast has not been demonstrated. These treatments did not show significant effects on either Tor1p-Tor1p or Tor2p-  After washing, the immune complex was eluted with the 3X FLAG peptide. The eluates were then immunoprecipitated with anti-Myc beads. Cell lysates, eluates with 3X FLAG peptides, and immunoprecipitates with anti-Myc beads were analyzed by immunoblotting with the indicated antibodies. B, COS-7 cells were transiently transfected with the expression plasmids for Myc-mTOR and FLAG-mTOR alone (lanes 1 and 2) or together with increasing amounts of the expression plasmid for Xpress-raptor (60 ng (lanes 3 and 4) and 300 ng (lanes 5 and 6)). After 40 h, the cells were starved of amino acids in D-PBS for 90 min and then left in D-PBS (lanes 1, 3, and 5) or transferred into D-PBS containing amino acids (lanes 2, 4, and 6) for 60 min before lysis. The cell lysates were immunoprecipitated with anti-FLAG M2 beads, and the immunoprecipitates were analyzed by immunoblotting using anti-Myc, anti-FLAG, and anti-Xpress antibodies.

mTOR Multimerization via the HEAT Repeat Region
Tor2p interaction (Fig. 7, A and B,  lanes 6 and 7), although such treatments substantially affected mTOR multimerization. Thus, in contrast to mTOR, we observed no change in TOR multimerization in yeast.

DISCUSSION
We have demonstrated that mTOR forms a multimer in vivo via the N-terminal HEAT repeat domain. mTOR multimerization occurs in both mTORC1 and mTORC2 (Fig. 5A). Our study has also revealed that TOR multimerization is conserved in yeast Tor1p and Tor2p. Because Tor1p exists solely in TORC1 (10,11,40), this observation indicates that the multimerization occurs in TORC1 (Fig.  7A). Consistent with our results, it was reported recently that Tor2p in yeast and TOR in Drosophila also exist as multimers in vivo and that TORC2 is in an oligomeric state in yeast (38,39).
Thus, multimerization of TOR is evolutionally conserved from yeast to mammals, occurs in both TORC1 and TORC2, and therefore appears to be important for TOR functioning. Gel The positions corresponding to three major peaks of mTOR are also indicated. B, HeLa cells were treated, and the cell lysate was analyzed as described for A. C, HEK293T cells were transiently transfected with expression plasmids for Myc-mTOR and FLAG-mTOR and then cultured in DMEM containing 10% serum. After 40 h, the cell lysate was prepared and analyzed as described for A. D, each fraction obtained as described for C was subjected to immunoprecipitation (IP) with anti-FLAG M2 beads, and the immunoprecipitates were analyzed by immunoblotting using anti-Myc and anti-FLAG antibodies. Equal volumes of fractions 4 and 5 (Fr. 4ϩ5) were mixed and used for immunoprecipitation.
In response to amino acid sufficiency, only a few changes in the formation of protein complexes containing mTOR have been reported thus far (7,41). We have demonstrated here, for the first time to our knowledge, that mTOR multimerization is sensitive to amino acid deprivation as well as to butanol treatment. As these two conditions have already been reported to reduce S6K1 and 4E-BP1 phosphorylation (1,7,14,26), it is conceivable that mTOR multimerization regulates mTORC1 activity or mTOR functioning in general. However, careful comparison between the time courses of mTOR multimerization and S6K1 phosphorylation in response to changes in amino acid sufficiency revealed that changes in the latter precede changes in the former. This difference in time courses may be due to S6K1 phosphorylation being regulated both positively by a protein kinase, possibly mTOR itself, and negatively by a protein phosphatase, both of which are under reciprocal control by mTORC1, as suggested previously (32,42); and therefore, the phosphorylation state of S6K1 may not be proportional to the activity of mTORC1. More conceivably, mTOR multimerization may not directly regulate mTORC1 activity, but a common nutrient-sensitive mechanism may regulate both mTOR multimerization and mTORC1 activity independently. As mTOR multimerization in mTORC1 appeared to be refractory to amino acid deprivation (Fig. 5B), amino acid-mediated change in mTOR multimerization may occur solely in mTORC2. Consistent with this idea, formation of a complex of Ͼ2 MDa (peak A) containing mTOR and rictor was sensitive to amino acid sufficiency in HeLa cells (Fig. 6D). If this is the case, the output regulated by mTOR multimerization appears to differ from Akt phosphorylation at Ser 473 , which was recently identified as a target of mTORC2 (16), because it is not modulated by amino acid sufficiency (data not shown) (33). The observation that artificially dimerized GyrB-mTOR did not promote the phosphorylation of Akt at Ser 473 is consistent with this idea (data not shown). Instead, it is possible that mTOR multimerization might play a role in the regulation of a different subset of mTORC2 outputs. In line with this argument, it has been reported that amino acid sufficiency affects the activation of some mTORC2 outputs such as RhoA and Rac1 (13).
At present, it remains to be determined whether mTOR-mTOR interaction is direct or not. Raptor will not mediate the multimerization of mTOR because mTOR-raptor interaction is reduced by rapamycin treatment (data not shown) (7,43), whereas mTOR-mTOR interaction is not. Moreover, butanol treatment reduced mTOR-mTOR interaction but not mTORraptor interaction (Fig. 3 and data not shown). The overexpression of Xpress-raptor consistently had only a marginal effect on the multimerization of mTOR under amino acidreplete conditions (Fig. 5B).
Unlike mTOR, the multimerization of TOR proteins in yeast was not affected by nutrient deprivation. This may be because TOR multimerization is regulated only in TORC2, as men-tioned above for mTORC2, and the tested conditions affect only TORC1. Otherwise, the regulation of mTOR multimerization may be an evolutionally new mechanism.
In conclusion, we found that TOR exists as a multimer in vivo. As the state of multimeric mTOR is sensitive to amino acid sufficiency, elucidation of the role and regulatory mechanism of multimerization will provide an important clue to understanding mTOR signaling.