Thr2446 is a novel mammalian target of rapamycin (mTOR) phosphorylation site regulated by nutrient status.

The mammalian target of rapamycin (mTOR) is a key regulator of protein translation. Signaling via mTOR is increased by growth factors but decreased during nutrient deprivation. Previous studies have identified Ser2448 as a nutrient-regulated phosphorylation site located in the mTOR catalytic domain, insulin stimulates Ser2448 phosphorylation via protein kinase B (PKB), while Ser2448 phosphorylation is attenuated with amino acid starvation. Here we have identified Thr2446 as a novel nutrient-regulated phosphorylation site on mTOR. Thr2446 becomes phosphorylated when CHO-IR cells are nutrient-deprived, but phosphorylation is reduced by insulin stimulation. Nutrient deprivation activates AMP-activated protein kinase (AMPK). To test whether this could be involved in regulating phoshorylation of mTOR, we treated cultured murine myotubes with 5'-aminoimidazole-4-carboxamide ribonucleoside (AICAR) or dinitrophenol (DNP). Both treatments activated AMPK and also caused a concomitant increase in phosphorylation of Thr2446 and a parallel decrease in insulin's ability to phosphorylate p70 S6 kinase. In vitro kinase assays using peptides based on the sequence in amino acids 2440-2551 of mTOR found that PKB and AMPK are capable of phosphorylating sites in this region. However, phosphorylation by PKB is restricted when Thr2446 is mutated to an acidic residue mimicking phosphorylation. Conversely, AMP-kinase-induced phosphorylation is reduced when Ser2448 is phosphorylated. These data suggest differential phosphorylation Thr2446 and Ser2448 could act as a switch mechanism to integrate signals from nutrient status and growth factors to control the regulation of protein translation.

The rate of protein translation is dynamically regulated, being up-regulated at times of cellular growth and attenuated when cells lack nutrients or amino acids (1). Rapamycin attenuates the rate of protein translation and the identification of its cellular target, the mammalian target of rapamycin (mTOR), 1 has greatly increased our understanding of pathways controlling protein translation (2). Activated mTOR stimulates translation by inducing phosphorylation of the eIF4E binding protein (4EBP1) (3)(4)(5)(6)(7)(8)(9) and stimulating phosphorylation and activation of p70 S6 kinase (4, 9 -11). Nutrient and amino acid deprivation inactivate p70 S6 kinase and promote the dephosphorylation of 4EBP1 via mechanisms that are not fully understood (1,(12)(13)(14)(15)(16)(17)(18)(19). mTOR was first identified in yeast (20), and subsequently a single isoform has been cloned in mammals (21)(22)(23). The mammalian isoform is a 240-kDa protein, which contains a kinase domain bearing a striking homology to the PI 3-kinase family of lipid kinases. However, mTOR functions only as a protein kinase (24 -26), and evidence has been presented that mTOR can directly phosphorylate 4EBP1 and p70 S6 kinase (24,27,28). However, the target sequence in both these molecules is very different giving rise to suggestions that the real mechanism for mTOR-mediated changes in phosphorylation may be via an inhibition of protein phosphatase 2A (PP2A) (29).
Rapamycin complexes with the cytosolic receptor FK506binding protein , and this complex binds to a distinct region of mTOR upstream of the catalytic domain (22,30). The finding that a specific monoclonal antibody binding to the extreme C terminus region of mTOR (27) and that deletion of aa 2430 -2450 (31) both increased the basal protein kinase activity of mTOR suggested this C-terminal region is of regulatory importance. Evidence has been presented that kinases, including PKB, can phosphorylate this region of mTOR and that such phosphorylation is likely to have a regulatory role (32)(33)(34). We and others (17,31,35,36) subsequently identified Ser 2448 as the PKB-mediated phosphorylation site. The phosphorylation of Ser 2448 on mTOR is directly related to amino acid and nutrient status (17,35). More recently it has been suggested that this nutrient effect on mTOR might act through AMPK as activation of this kinase is associated with reduced signaling through mTOR and reduced phosphorylation of Ser 2448 (36,37). This suggests that phosphorylation of Ser 2448 might act as a switch, controlling the activity and function of mTOR and that AMPK may play a role in this process.
We hypothesized this regulation might involve other phosphorylation events in the vicinity of Ser 2448 . The region surrounding Ser 2448 is highly conserved in all mammalian sequences for mTOR and contains a number of ser and thr residues that might be regulatory phosphorylation sites as determined by Scansite (38). Here we identified Thr 2446 as a novel phosphorylation site in mTOR, and we showed that phosphorylation is activated by amino acid deprivation or activation of AMPK while it is attenuated by growth factor stimulation.
Further evidence is provided that phosphorylation at Thr 2446 and Ser 2448 are mutually exclusive indicating these sites may act as switches that integrate the counteracting signals of growth factors and nutrient deprivation.

EXPERIMENTAL PROCEDURES
Materials and Reagents-Phospho-AMPK, total AMPK, and phospho-p70 S6 kinase antibodies were purchased from Cell Signaling Technologies (Beverly, MA). Total p70 S6 kinase and PKC-antibodies and protein G-agarose was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Phospho-ERK and total ERK was purchased from New England Biolabs (Beverly, MA). PKB antibodies were kindly provided by Dr Dario Alessi, University of Dundee. Rapamycin, PD98059, and okadaic acid were obtained from Calbiochem. AMPK was purified from liver as described previously (39). All other reagents were obtained from Sigma unless stated otherwise. Peptides were generated by the Wolfson Insitute for Biomedical Research (University College London) and were based on the sequence of aa 2440 -2551 in mTOR with addition of an N-terminal lysine to improve binding to P81 paper. The peptides used are described in Fig. 5.
mTOR Antibodies-The sheep phospho-mTOR Ser 2448 antibody (phospho-Ser 2448 ) was as described previously (35). Polyclonal total mTOR antibody was produced in rabbit as described in Withers et al. (40) using a glutathione S-transferase fusion protein corresponding to the region between amino acids 668 and 939 of the mTOR sequence. Phospho-mTOR Thr 2446 antiisera was raised in rabbit using the phosphopeptide CSRTRT(P)DSYS, corresponding to amino acids Ser 2442 -Ser 2450 of the human mTOR sequence with an N-terminal cysteine added. The antibody was affinity-purified through SulfoLink resin (Pierce) columns. First the antibody was run through a column coupled to the dephosphopeptide, and the unbound was then run-through a second column coupled to the phosphopeptide, and phosphospecific antibodies were eluted.

Inverse Relationship between Phosphorylation of Thr 2446 and
Ser 2448 on mTOR-Here we have developed an affinity-purified antibody that recognizes phospho-Ser 2446 in mTOR. The antibody identifies a band that corresponds to the molecular weight recognized by antibodies that recognize total mTOR, and the reactivity of this band with the phospho-antibody band is blocked by co-incubation with the phosphopeptide that was used to make the antibody (Fig. 1), while the dephosphopeptide has no effect on immunoreactivity of the phospho-antibody (data not shown). While total levels of mTOR did not change with nutrient deprivation, we find that immunoreactivity with the Thr 2446 phospho-antibody increases when cells are deprived of nutrient (Fig. 1). Furthermore we find that the rapid increase in reactivity to the phospho-Thr 2446 antibody during nutrient deprivation inversely parallels the decrease in reactivity to the phospho-Ser 2448 antibody under the same starvation conditions ( Fig. 2A). The reverse was observed when CHO-IR cells were stimulated with insulin with a decrease in reactivity with the phospho-Thr 2446 antibody observed with prolonged stimulation with insulin corresponding to an increase in reactivity to the phospho-Ser 2448 antibody under the same conditions (Fig. 2B). Insulin-mediated activation of Ser 2448 phosphorylation requires activation of PI 3-kinase and PKB (35). Here we find that treatment with the PI 3-kinase inhibitor wortmannin, which blocks the activation of PKB, blocks the serum-induced phosphorylation on Ser 2448 but has no effect on nutrient induced increases in Thr 2446 phosphorylation (Fig. 3).
Effect of AMPK Activators on mTOR Phosphorylation-AMPK is activated when ATP/AMP ratios fall (42), and the phosphorylation of AMPK was increased following aa withdrawal in CHO-IR cells (data not shown). Therefore to determine the physiological significance of nutrient status in regulating mTOR Thr 2446 phosphorylation, mouse H2K muscle cells were treated with dinitrophenol, a metabolic poison that rapidly decreases cellular ATP level, or with AICAR riboside, a cell-permeable compound that in some cell types, including the H2Ks, is degraded to ZMP, which in turn is capable of activating AMPK. Activity (Fig. 4) of AMPK were increased by both AICAR and DNP in these cells. Both treatments also cause an acute increase in Thr 2446 phosphorylation, even in the presence FIG. 1. Phosphorylation of mTOR on Thr 2446 increases with nutrient deprivation. CHO-IR cells were grown in Ham's F-12 medium containing 10% fetal calf serum and where indicated were starved in Dulbecco's-PBS for 60 min. Cell lysates were separated by SDS-PAGE and Western blotted as indicated with either affinity-purified mTOR phospho-Thr 2446 antibody, the phospho-Thr 2446 -mTOR antibody blocked with phosphopeptide, or an antisera that recognized total levels of mTOR.
Thr 2446 , a Nutrient Status-regulated mTOR Phosphorylation Site of insulin. This is matched by an attenuation of insulin stimulation of p70 S6 kinase phosphorylation (Fig. 4).
Phosphorylation of mTOR Regulatory Region-In an effort to identify kinases capable of phosphorylating this region, an in vitro kinase assay approach was taken using a range of peptides based on aa 2440 -2551 of mTOR (Fig. 5). p70 S6 kinase phosphorylated all the peptides with efficiency greater than or equal to the phosphorylation of the wild type peptide (data not shown) suggesting that this kinase was phosphorylating various sites in the peptide and that its action did not depend on the status of the Thr 2446 site. However, phosphorylation of the T2446E peptide by PKB and AMPK was very low compared with phosphorylation of wild type peptides. PKB phosphorylated the T2446A peptide equally as well as it phosphorylated wild type, while the phospho-Ser 2448 was phosphorylated at ϳ20% of wild type level. AMP kinase-mediated phosphorylation of the T2446E and T2446A peptide was greatly reduced compared with the wild type peptide consistent with Thr 2446 being a major AMPK target site, while phosphorylation at the 2448 position virtually abolished the ability of AMPK to phosphorylate the peptides (Fig. 5). DISCUSSION The results presented here provide evidence that Thr 2446 is a regulated phosphorylation site in mTOR, and given the previous evidence that this is an important regulatory region (27,31,43), this phosphorylation is likely to play a regulatory role in mTOR action. The Thr 2446 phosphorylation is likely to be a negative regulation given that Thr 2446 phosphorylation is associated with conditions that reduce signaling through mTOR and our finding that phosphorylation of Thr 2446 correlates with an attenuation of insulin-stimulated phosphorylation of p70 S6 kinase.
We have previously provided evidence that Ser 2448 is phosphorylated downstream of PKB (35). Despite the fact Thr 2446 fits a weak PKB consensus sequence, our evidence indicates   (36,37), and our results suggest that AMPK is likely to be at least one of the kinases regulating phosphorylation at Thr 2446 . There are two lines of evidence for this; first, nutrient deprivation, DNP, and AICAR all activate AMPK and stimulate phosphorylation of Thr 2446 , and second, AMPK can directly phosphorylate the aa 2440 -2551 mTOR peptide. The optimum amino acid sequence for AMPK has not been fully defined, but our peptide phosphorylation studies provide evidence that Thr 2446 is likely to be phosphorylated by the enzyme; while our peptides contain three serines and two threonines the mutation of Thr 2446 residue to glutamic acid or alanine greatly reduced phosphorylation of the peptide by AMPK.
Our data also show that conditions that increase phosphorylation at Thr 2446 tend to decrease phosphoprylation at Ser 2448 and vice versa. This could be in part due to coordinate regulation of the kinases involved. For example insulin will activate PKB while reducing AMPK activity. However, our data suggest a further level of regulation as it suggests that phosphorylation of the two sites is mutually exclusive. This is supported by the finding that PKB does not phosphorylate the aa 2440 -2551 peptide when Thr 2446 is changed to glutamic acid (mimicking phosphorylation), but phosphorylation is normal when Thr 2446 is changed to alanine. Furthermore, while Ser 2448 is clearly not a major AMPK-stimulated phosphorylation site, phosphorylation of the peptide by AMPK is greatly reduced by phosphorylation at the 2448 position, consistent with the notion that phosphorylation at Thr 2446 and Ser 2448 are mutually exclusive events.
The mutual exclusivity probably arises due to the phosphorylation effecting the substrate recognition sequence for the second kinase, and interestingly, there is a precedent for mutually exclusive regulatory phosphorylation sites separated by a single amino acid being involved in integrating signals related to nutrient status in other enzymes. In acetyl-CoA carboxylase this involves Ser 77 , which is an AMPK site, Ser 79 , which is a proetin kinase A target (44), and in the case of hormone-sensitive lipase, it involves Ser 563 as a protein kinase A site and Ser 565 as an AMPK site (45,46).
In summary, we identify Thr 2446 as a novel phosphorylation site in mTOR that is increased by conditions of low nutrient and correlates with negative regulation of mTOR activity. As phosphorylation at this site limits phosphorylation at Ser 2448 and vice versa, we suggest that these two phosphorylation sites might act as a switch to control the positive and negative signals regulating protein translation.