Two motifs in the translational repressor PHAS-I required for efficient phosphorylation by mammalian target of rapamycin and for recognition by raptor.

Mammalian target of rapamycin (mTOR) is the central element of a signaling pathway involved in the control of mRNA translation and cell growth. The actions of mTOR are mediated in part through the phosphorylation of the eukaryotic initiation factor 4E-binding protein, PHAS-I. In vitro mTOR phosphorylates PHAS-I in sites that control PHAS-I binding to eukaryotic initiation factor 4E; however, whether mTOR directly phosphorylates PHAS-I in cells has been a point of debate. The Arg-Ala-Ile-Pro (RAIP motif) and Phe-Glu-Met-Asp-Ile (tor signaling motif) sequences found in the NH2- and COOH-terminal regions of PHAS-I, respectively, are required for the efficient phosphorylation of PHAS-I in cells. Here we show that mutations in either motif markedly decreased the phosphorylation of recombinant PHAS-I by mTOR in vitro. Wild-type PHAS-I, but none of the mutant proteins, was coimmunoprecipitated with hemagglutinin-tagged raptor, an mTOR-associated protein, after extracts of cells overexpressing raptor had been supplemented with recombinant PHAS-I proteins. Moreover, raptor overexpression enhanced the phosphorylation of wild-type PHAS-I by mTOR but not the phosphorylation of the mutant proteins. The results not only provide direct evidence that both the RAIP and tor signaling motifs are important for the phosphorylation by mTOR, possibly by allowing PHAS-I binding to raptor, but also support the view that mTOR phosphorylates PHAS-I in cells.

Many hormones, growth factors, and nutrients stimulate protein synthesis by promoting the phosphorylation of PHAS-I (also known as 4E-BP1), the prototypic member of a family of translational repressor proteins (1,2). Nonphosphorylated PHAS-I binds tightly to the mRNA cap-binding protein, eIF4E, 1 and blocks binding of eIF4E to eIF4G, a scaffolding protein that organizes several other important initiation factors, including eIF3, which links the complex to the 40 S ribosomal subunit (3). When phosphorylated in the appropriate sites, PHAS-I dissociates from eIF4E, allowing formation of the complex needed for the proper positioning of the 40 S ribosome and for efficient scanning of the 5Ј-untranslated region.
PHAS-I is phosphorylated in five sites (4,5), all of which conform to a (Ser/Thr)-Pro motif (see Fig. 1). Insulin promotes the phosphorylation of Thr 36 , Thr 45 , Ser 64 , and Thr 69 (4,6,7). Phosphorylation occurs in an ordered fashion, with phosphorylation of Thr 36 and Thr 45 preceding that of Thr 69 and Ser 64 (5,8). The phosphorylation of Ser 82 does not appear to affect eIF4E binding (9), and although phosphorylation of Ser 64 markedly decreases binding in vitro, mutating Ser 64 to Ala does not significantly change the amount of PHAS-I bound to eIF4E in cells (5). In contrast, Thr to Ala mutation of any of the three Thr-Pro sites increased eIF4E binding (5). Thus, it is clear that the Thr-Pro sites are important in controlling the function of PHAS-I.
Findings with rapamycin provided the first evidence implicating the mammalian target of rapamycin (mTOR) in the control of PHAS-I (6,10). Rapamycin treatment of cells attenuates the phosphorylation of all four insulin-sensitive sites (4,6,7). When presented as a complex with its intracellular receptor, FKBP12, rapamycin binds with high affinity to mTOR (11). mTOR (also known as FRAP or RAFT1) (12)(13)(14) is a founding member of a family of Ser/Thr protein kinases that have catalytic domains homologous to that in phosphatidylinositol 3-OH-kinase (15,16). In reactions conducted in vitro, mTOR phosphorylated purified PHAS-I in Thr 36 , Thr 45 , Ser 64 , and Thr 69 , suggesting that (Ser/Thr)-Pro was a recognition motif for phosphorylation by mTOR (7,17).
Although there is general agreement that PHAS-I phosphorylation in vivo is controlled by the mTOR signaling pathway, it has been argued that mTOR does not directly phosphorylate PHAS-I in cells. Uncertainty as to the role of mTOR as a PHAS-I kinase arose from the demonstration that mTOR phosphorylates S6K-1 in Thr 389 , which is flanked by nonprolyl hydrophobic residues (16). This finding clearly eliminates (Ser/ Thr)-Pro as the sole determinant for phosphorylation and indicates that the features of the substrate that allow recognition by mTOR are more complicated than the primary sequence of amino acids surrounding the phosphorylation sites. mTOR was shown recently (18 -20) to exist in a complex with raptor, a large (M r ϭ 150,000) protein possessing a unique NH 2 -terminal region followed by three HEAT motifs and seven WD-40 domains. Raptor has been shown to bind directly to both PHAS-I and S6K-1, and it has been suggested that raptor presents these proteins to mTOR for phosphorylation (18).
Although the structural motifs in the substrate that allow direct phosphorylation by mTOR have not been fully defined, two sequences in PHAS-I that are required for the efficient phosphorylation of the protein in cells have been described. The first, referred to as the RAIP motif because of the sequence of amino acids involved, is found in the NH 2 -terminal region of PHAS-I (21). The second, referred to as the TOS motif, is formed by the last five amino acids (Phe-Glu-Met-Asp-Ile) in the protein (22). Disrupting the TOS motif by a Phe to Ala point mutation, or removing the RAIP motif either by ⌬16 NH 2terminal truncation or by mutating Ile-Pro to Ala-Ala, markedly decreased phosphorylation of overexpressed PHAS-I in human embryonic kidney 293 cells (21,22). In the present study, we have investigated the influence of the RAIP and TOS motifs on the phosphorylation of PHAS-I by mTOR.
To generate raptor antibodies, a peptide (CEKIEGSKSLAQSWR-MKD) having a sequence identical to positions 36 -53 in human raptor was coupled to keyhole limpet hemocyanin (19), and the peptide-hemocyanin conjugate was used to immunize rabbits as described previously (24). Antibodies were purified using columns containing affinity resins prepared by coupling the peptide to Sulfolink beads (Pierce).
Monoclonal antibody 9E10, which recognizes the Myc epitope tag, and 12CA5, which recognizes the HA epitope tag, were purified from hybridoma culture medium. Monoclonal antibody to the AU1 epitope tag was from Berkeley Antibody Company.
Mutations and Generation of PHAS-I Expression Vectors-Rat PHAS-I cDNA in pBluescript (SK Ϫ ) was used as template for generating mutations (23). cDNA encoding PHAS-I with a Phe 113 to Ala point mutation (F113A PHAS-I) was generated by using the PCR and primers to introduce BamHI sites at the ends. The reverse primer encoded the point mutation. cDNA encoding PHAS-I lacking the NH 2 -terminal sixteen amino acids (⌬16 PHAS-I) was generated by PCR, again introducing BamHI sites at both ends. To introduce mutations changing coding of Ile 15 -Pro 16 into Ala 15 -Ala 16 (RAAA PHAS-I), site-directed mutagenesis was conducted using a Transformer site-directed mutagenesis kit (Clontech) with a mutagenic oligonucleotide. BamHI sites were then added to the ends of the mutant cDNA by PCR. The PCR products encoding the mutant PHAS-I proteins were digested with BamHI and inserted into the BamHI site of pBluescript. Nucleotide sequencing of the coding regions of each construct confirmed that the desired mutations were present and that no secondary mutations were introduced by the mutagenic procedures.
Fragments encoding PHAS-I proteins were excised from pBluescript with BamHI and inserted into the BamHI site of either pET14b (Novagen) for bacterial expression or pCMV-Tag 3a (Stratagene) for expression in mammalian cells. Proper orientation of the inserts was determined by restriction mapping and confirmed by nucleotide sequencing.
Bacterial Expression and Purification of Recombinant PHAS-I Proteins-The pET14b constructs encode NH 2 -terminal His-tagged proteins, which were expressed in bacteria and purified as described previously (25). Protein content was determined using the method of Smith et al. (26). To assess purity and to confirm concentration, samples (2 g) of the recombinant proteins were subjected to SDS-PAGE (27). The proteins were found as single bands of equal Coomassie Blue staining intensity. 2 Raptor Expression Construct-cDNA (4008 bp) encoding HA-tagged human raptor was assembled from three fragments, which were generated by PCR using primers based on the published raptor sequence (18 -20). Fragment 1 (bp 1-1043) was amplified from I.M.A.G.E. clone 3635589, fragment 2 (bp 952-1968) was amplified from reverse-transcribed human skeletal muscle cDNA (7175-1; Clontech), and fragment 3 (bp 1963-4008) was generated from I.M.A.G.E. clone 6057826. BamHI sites at bp 1963-1968 were introduced into both fragment 2 and fragment 3. Creating these sites did not change amino acid coding, and it allowed joining of the pieces. All three fragments were inserted into pCR2.1 (Invitrogen). Next, cDNA encoding the triple HA tag from pKH3 (28) was amplified by PCR and ligated into the SalI and EcoRI sites of pBluescript. Fragment 1 was excised and inserted into the EcoRI and XmaI sites of this construct. Separately, fragment 3 was excised with BamHI and NotI and inserted between these sites in pBluescript (SKϪ). Fragment 2 was then removed with XmaI and BamHI and inserted between these sites in the fragment 3/pBluescript construct. The resulting construct was digested with KpnI and XmaI, which cut pBluescript and the 5Ј end of fragment 2, respectively. Between these sites was inserted the HA-fragment 1 cDNA that had been excised from pBluescript with KpnI and XmaI, thus generating full-length HAtagged raptor. The raptor cDNA was excised from pBluescript with KpnI and NotI and inserted between these sites in pcDNA3. The coding region was sequenced and found to be free of errors.
Overexpression of mTOR, Raptor, and PHAS-I Proteins-293T cells were seeded into plastic tissue culture dishes (2 ϫ 10 4 cells/cm 2 ; Falcon) and cultured in humidified 5.0% CO 2 in air for 24 h in growth medium composed of 10% (v/v) fetal bovine serum in Dulbecco's modified Eagle's medium. For expression of AU1-mTOR, 293T cells were transfected with pcDNA3 alone or pcDNA3 containing an AU1-mTOR insert by using TransIT-LT2 (Mirus Corp., Madison, WI) and 5 g of DNA/100mm-diameter dish as described previously (15). Transfections for expressing PHAS-I proteins and HA-raptor were performed in exactly the same manner, except using the expression vectors described in the preceding paragraphs. Cells were used in experiments 18 -20 h after transfection.
Immune Complex Assay of mTOR Activity-Prior to the kinase assay, washed AU1-mTOR complexes were incubated at 21°C for 90 min without additions or with 5 g of mTAb1 in 20 l of Buffer A (50 mM NaCl, 0.1 mM EGTA, 1 mM dithiothreitol, 0.5 mM microcystin LR, 10 mM Na-HEPES, and 50 mM ␤-glycerophosphate, pH 7.4). The beads were rinsed twice and suspended in 20 l of Buffer A. The kinase reactions were initiated by adding 20 l of Buffer A supplemented with 0.2 mM [␥-32 P]ATP (2000 mCi/mmol), 20 mM MnCl 2 , and 40 g/ml of wild-type or mutant PHAS-I. Reactions were terminated after 30 min by adding SDS sample buffer. Measurements under these conditions reflect the initial rate of phosphorylation, as less than 5% of the available substrates were phosphorylated, and the reactions have been shown to proceed linearly for 60 min (29).
Electrophoretic Analyses-Samples were subjected to SDS-PAGE before proteins were electrophoretically transferred to Immobilon (Millipore) membranes and immunoblotted as described previously (7). The amounts of 32 P incorporated into PHAS-I proteins were determined by scintillation counting of gel slices. Relative levels of 32 P incorporation were determined by phosphorimaging. Signal intensities of bands in immunoblots were determined by scanning laser densitometry.
Other Materials-Rapamycin and LY294002 were from Calbiochem-Novabiochem International. Purified activated Erk2 isoform was prepared as described by Khokhlatchev et al. (30). Glutathione S-transferase (GST)-FKBP12 was expressed in bacteria and purified as described previously (14). [␥-32 P]ATP was from PerkinElmer Life Sciences.

Inhibitory Effects of Mutations of the TOS and RAIP Motifs on Phosphorylation of PHAS-I by mTOR-
The effects of disrupting the TOS and RAIP motifs on phosphorylation of PHAS-I by mTOR were investigated by performing immune complex kinase assays. mTOR was expressed as an AU1 epitope-tagged protein in 293T cells and immunoprecipitated prior to the assay. After incubating the immunopurified mTOR with the antibody, mTAb1, mTOR activity was assessed using purified recombinant PHAS-I proteins as substrates. mTAb1 activates mTOR by binding to an inhibitory regulatory domain located in the COOH-terminal region of mTOR (17,31). Phosphorylation was assessed by 32 P incorporation from [␥-32 P]ATP. As the three Thr-Pro sites are the preferred sites in PHAS-I for phosphorylation by mTOR, phosphorylation was also assessed by immunoblotting with P-Thr36/45 and P-Thr69 antibodies.
The TOS motif, which is formed by the last five COOHterminal amino acids in PHAS-I ( Fig. 1), was disrupted by mutating Phe 113 to Ala (22). The mutant protein, designated F113A PHAS-I, was phosphorylated by mTOR at only 25% of the rate at which mTOR phosphorylated wild-type PHAS-I (Fig. 2). Disrupting the TOS motif decreased the phosphorylation of Thr 36 /Thr 45 and Thr 69 by 72 and 86%, respectively. In contrast, the Phe 113 to Ala mutation attenuated the phosphorylation of neither Thr 36 /Thr 45 nor Thr 69 by mitogen-activated protein kinase (Fig. 3), indicating that phosphorylation sites were not compromised in the recombinant protein.
Two mutant proteins were generated to investigate the influence of the RAIP motif (21). In one the NH 2 -terminal region containing the motif (Fig. 1) was removed by truncation, yielding ⌬16 PHAS-I. The other involved a more conservative approach in which Ile 15 and Pro 16 were each mutated to Ala, resulting in RAAA PHAS-I (21). Although neither the truncation nor the double mutation attenuated the ability of mitogenactivated protein kinase to phosphorylate PHAS-I (Fig. 3), both reduced phosphorylation by mTOR (Fig. 2). ⌬16 PHAS-I was phosphorylated at only one-third of the rate of the wild-type protein. Interestingly, truncation had a more pronounced effect on the phosphorylation of Thr 36 /Thr 45 than on phosphorylation of Thr 69 . The effects of the double mutation were somewhat smaller than those of truncation but comparable in the sense that the decrease in Thr 36 /Thr 45 phosphorylation was greater than the decrease in Thr 69 phosphorylation.
Rapamycin Sensitivity of the Phosphorylation of Wild-type and Mutant PHAS-I Proteins by mTOR-None of the PHAS-I proteins were appreciably phosphorylated when immune complex kinase reactions were conducted with samples from cells transfected with vector alone (Fig. 4), indicating that the phosphorylation reactions required mTOR. Confirmation of this point was provided by results of experiments in which the effects of mTOR activators and inhibitors were investigated. mTAb1 increased the overall phosphorylation, as well as the phosphorylation of Thr 36 /Thr 45 and Thr 69 , in wild-type and all three mutant proteins (Fig. 4). Phosphorylation of all four proteins was abolished by LY294002 (Fig. 4), the phosphatidylinositol 3-OH-kinase inhibitor that also inhibits mTOR (29,32). The phosphorylation of all four proteins was also decreased by rapamycin.
Rapamycin-FKBP12 does not fully inhibit the phosphorylation of PHAS-I by mTOR (29). The phosphorylation of Thr 36 / Thr 45 in wild-type PHAS-I by mTOR is more resistant to inhibition by rapamycin than that of Thr 69 . Interestingly, disrupting either the TOS or the RAIP motifs increased the extent of inhibition of Thr 36 /Thr 45 phosphorylation produced by rapamycin (Fig. 4). To investigate further the influence of these motifs on rapamycin sensitivity, reactions were conducted with wild-type PHAS-I, F113A PHAS-I, and RAAA PHAS-I in the presence of 10 mM GST-FKBP12 and increasing concentrations of rapamycin (Fig. 5). Although the phosphorylation of all three PHAS-I proteins was inhibited by increasing rapamycin, a striking difference with respect to sensitivity to the drug was evident between wild-type PHAS-I and the mutant proteins when the amounts of 32 P incorporation were expressed as a percentages of the respective controls (Fig. 5A). At the highest concentration of rapamycin, which decreased the overall phosphorylation of wild-type PHAS-I by about half, 32 P incorporation into either F113A PHAS-I or RAAA PHAS-I was decreased by ϳ90%.
The distinction between wild-type and mutant proteins was Samples were subjected to SDS-PAGE, and immunoblots were prepared with P-Thr36/45 and P-Thr69 antibodies. Phosphorylation by samples from cells transfected with vector alone was negligible (Fig. 4). 32 P contents of the proteins were determined by scintillation counting of gel slices. 32 P incorporated (mol 32 P/mol PHAS-I protein Ϯ S.E., n ϭ 5) by mTOR were as follows: wild-type (0.024 Ϯ 0.005), F113A (0.006 Ϯ 0.001), RAAA (0.012 Ϯ 0.002), and ⌬16 (0.007 Ϯ 0.002). In addition to total phosphorylation, the intensities of the bands in immunoblots prepared with P-Thr36/45 and P-Thr69 were determined by optical density scanning. The results presented are expressed relative to the 32 P content or the immunoblot signals from wild-type PHAS-I and are means Ϯ S.E. from five experiments.

FIG. 3. Phosphorylation of wild-type and mutant PHAS-I proteins by mitogen-activated protein (MAP) kinase.
Phosphorylation reactions were conducted exactly as described in the legend to Fig.  2, except that 20 l of 5 g/ml activated recombinant Erk2 in Buffer A was substituted for the mTOR beads. Samples were subjected to SDS-PAGE, and immunoblots were prepared with PHAS-I antibodies and with phosphospecific antibodies. even more pronounced when Thr 36 /Thr 45 phosphorylation was assessed by immunoblotting (Fig. 5B). Maximal inhibitory concentrations of rapamycin decreased Thr 36 /Thr 45 phosphorylation in the wild-type protein by only 30% but decreased phosphorylation of these sites in RAAA PHAS-I and F113A PHAS-I by 90%. The dose response curves for inhibition of 32 P incorporation into the mutant proteins (Fig. 5A), as well as phosphorylation of Thr 36 /Thr 45 (Fig. 5B) and Thr69 (Fig. 5C), were almost identical, with half-maximal inhibition occurring at ϳ20 nM rapamycin. In contrast to the effects of disrupting the RAIP and TOS motifs on Thr 36 /Thr 45 phosphorylation, mutation of these motifs had little, if any, effect on the rapamycin sensitivity of Thr 69 phosphorylation (Fig. 5C).
Rapamycin-FKBP12 clearly inhibited phosphorylation of the mutant proteins (Fig. 5), indicating that the RAIP and TOS motifs are not absolutely required for recognition by mTOR. These results predict that if mTOR phosphorylates the proteins in vivo, then phosphorylation of the mutant proteins should still be inhibited by rapamycin. To investigate this prediction, wild-type and mutant PHAS-I proteins were expressed in 293T cells. These rapidly proliferating cells are persistently activated, so that even without stimulation, PHAS-I is highly phosphorylated. As shown in Fig. 6, wild-type PHAS-I overexpressed in 293T cells was found in the highly phosphorylated ␥ form. Relatively little ␥ forms of F113A and RAAA proteins were detected, suggesting that these proteins were less highly phosphorylated than wild-type. Moreover, rapamycin promoted the dephosphorylation of the mutant proteins, as evidenced by the increased mobility of F113A, RAAA, and ⌬16 PHAS-I.
Interactions between PHAS-I Proteins and Raptor-To investigate the possibility that the RAIP and TOS motifs are involved in binding of PHAS-I to raptor, experiments were conducted in which extracts from HA-raptor overexpressing 293T cells were supplemented with purified PHAS-I proteins before immunoprecipitations were performed with anti-HA antibodies. Wild-type PHAS-I was readily detected in complexes containing raptor (Fig. 7, A and B). The fact that PHAS-I binding persisted during washes of the immune complexes is indicative of a relatively high affinity interaction between raptor and PHAS-I. None of the mutant proteins coimmunoprecipitated with HA-raptor (Fig. 7, A and B). Because none of the mutations disrupted binding of the PHAS-I antibodies (Fig. 3), the results are consistent with the interpretation that the RAIP and TOS motifs are required for the efficient binding of PHAS-I to raptor.
As raptor has been proposed to present PHAS-I to mTOR for phosphorylation (18), we investigated the effect of raptor overexpression on the phosphorylation of PHAS-I proteins by mTOR (Fig. 8). Coexpression of raptor with mTOR resulted in a marked increase in the phosphorylation of wild-type PHAS-I by mTOR (Fig. 8A). In contrast, raptor had relatively little effect on the phosphorylation of F113A PHAS-I, RAAA PHAS-I, or ⌬16 PHAS-I proteins.
Observing increased phosphorylation by mTOR after raptor overexpression implies that the mTOR-raptor complex phosphorylates PHAS-I more efficiently than mTOR lacking raptor. However, it was important to confirm that the overexpressed raptor formed a stable complex with mTOR particularly, because certain nonionic detergents have been shown to disrupt this complex (18). When homogenization buffer and buffers used to wash immune complexes were supplemented with Tween 20, as in our standard immunoprecipitation procedure, HA-raptor coimmunoprecipitated with AU1-mTOR (Fig. 9A). When Triton X-100 was substituted for Tween 20 in the homogenization buffer, a comparable amount of mTOR was recovered, but almost no raptor coimmunoprecipitated (Fig. 9A). These findings confirm the previous observations of Hara et al. (18).
To estimate how much overexpressing raptor increased the mTOR-raptor complex AU1 mTOR was immunoprecipitated, and immunoblots were prepared using polyclonal raptor antibodies. Raptor was detected in AU1 immunoprecipitates from cells transfected with AU1-mTOR alone, indicative of a complex between AU1-mTOR and endogenous raptor (Fig. 9B). Assuming the raptor antibodies react equally well with endogenous and HA-tagged raptor, the results indicate that overexpressing HA-raptor increased the total amount of raptor coimmunoprecipitating with mTOR by ϳ10-fold.
Raptor overexpression increased phosphorylation of Thr 36 / Thr 45 in wild-type PHAS-I by ϳ5-fold (Fig. 8B). Phosphorylation of Thr 36 /Thr 45 was lower in the three mutant proteins than in wild-type PHAS-I, both without and with raptor overexpression (Fig. 8B). The effects on Thr 69 phosphorylation were more complicated (Fig. 8C). Without raptor overexpression, mTORmediated phosphorylation of Thr 69 in ⌬16 PHAS-I was lower than phosphorylation of this site in wild-type PHAS-I. Under these conditions phosphorylation of this site in neither of the other mutant proteins was less than that in wild-type. Overexpressing raptor increased phosphorylation of Thr 69 in wild-type PHAS-I by ϳ5-fold (Fig. 8C). Interestingly, raptor enhanced the phosphorylation of Thr 69 in both proteins in which the RAIP motif was disrupted. Overexpressing raptor did not increase phosphorylation of Thr 69 in PHAS-I lacking the TOS motif.

DISCUSSION
The major point of this report is that disrupting either the RAIP or TOS motifs markedly inhibits the ability of mTOR to phosphorylate PHAS-I. The findings provide important information concerning the features involved in substrate recognition by mTOR. Moreover, when considered with previous evidence that the RAIP and TOS motifs are needed for the efficient phosphorylation of PHAS-I in cells (21,22), the present results support the view that mTOR phosphorylates PHAS-I in vivo.
Characterization of a hypophosphorylated PHAS-I fragment that bound tightly to eIF4E in cells undergoing apoptosis led to discovery of the importance of the RAIP motif (21). Amino acid sequencing revealed that the apoptotic fragment was generated by caspase cleavage of the Asp 24 -Gly 25 bond. Tee and Proud (21) subsequently localized the critical region needed for efficient phosphorylation of PHAS-I in cells to the RAIP motif. 293T cells were transfected with AU1-mTOR and/or HA-raptor as described under "Experimental Procedures." AU1-mTOR was immunoprecipitated, and immune complex kinases assays were conducted using purified wildtype and mutant PHAS-I proteins as substrate. 32 P incorporation was estimated by phosphorimaging, and immunoblots were prepared with the P-Thr36/45 and P-Thr69 antibodies. A phosphorimage of 32 P and immunoblots with phosphospecific antibodies are shown (A). The intensities of the bands in immunoblots prepared with P-Thr36/45 (B) and P-Thr69 (C) were expressed as percentages of the respective maximum signals. Mean values Ϯ S.E. of three experiments are presented.
In expression studies in human embryonic kidney 293 cells, phosphorylation of Thr 36 /Thr 45 was essentially abolished by the ⌬16 truncation, whereas Thr 69 phosphorylation persisted (21). In the present experiments, less of the RAAA PHAS-I than wild-type PHAS-I was found in the more slowly migrating ␥ form, confirming that disruption of the RAIP motif decreases phosphorylation in cells (Fig. 6). Interestingly, after raptor overexpression the ⌬16 truncation had a much more pronounced effect on decreasing the phosphorylation of Thr 36 / Thr 45 than on decreasing phosphorylation of Thr 69 by mTOR in vitro (Fig. 8, B and C).
S6K1 is another well characterized target of the mTOR signaling pathway that can be directly phosphorylated by mTOR in vitro (16,29). It has been known for some time that deletion of 30 amino acids from the NH 2 terminus of S6K1 inhibits phosphorylation of the kinase in cells (33,34). The similarity between the sequence Phe-Asp-Ile-Asp-Leu near the NH 2 terminus of S6K1 and the COOH-terminal Phe-Glu-Met-Asp-Ile in PHAS-I was noted by Schalm and Blenis (22). These investigators demonstrated that disrupting these sequences decreased phosphorylation of the respective proteins. Not only did mutation of the TOS motif in S6K1 significantly inhibit the activation of the kinase in cells, but it also prevented the inhibitory effect of S6K1 overexpression on PHAS-I phosphorylation, supporting the view that PHAS-I and S6K1 interact with a common upstream effector (22).
The COOH-terminal region of PHAS-I is very highly conserved among isoforms and across species (1), indicating that it is important in the function of the proteins. Our results confirm that the TOS motif is important for the efficient phosphorylation of PHAS-I but indicate that it is not absolutely required for mTOR signaling. Mutating Phe 113 to Ala decreased phosphorylation in 293T cells, but phosphorylation of the mutant protein was still clearly inhibited by rapamycin treatment of the cells (Fig. 6). Moreover, mTOR phosphorylated F113A PHAS-I, albeit at a reduced rate, in a reaction that was inhibited by either rapamycin-FKBP12 or LY294002 (Fig. 4). Indeed, after disruption of the TOS motif, phosphorylation of Thr 36 /Thr 45 by mTOR became more sensitive to inhibition by rapamycin (Fig.  5B). These results bear a remarkable similarity to very recent findings by Wang et al. (35), who demonstrated that the phosphorylation of Thr 37 /Thr 46 , the human equivalents of Thr 36 / Thr 45 , became more sensitive to inhibition by rapamycin in intact cells when the TOS motif was removed by truncation (35).
It is difficult to formally prove that any kinase phosphorylates a particular substrate in a cell, and the fact that rapamycin inhibits PHAS-I phosphorylation does not necessarily mean that mTOR phosphorylates the protein in vivo. For example, in Saccharomyces cerevisiae the effects of rapamycin are mediated in part by effects on protein phosphatases (36), and the possibility that phosphatase activation contributes to the effects of rapamycin on decreasing the phosphorylation of PHAS-I cannot be eliminated. However, the findings that disrupting either the RAIP or TOS motif decreased phosphorylation of PHAS-I, both in cells and by mTOR in vitro, contribute to the increasing evidence that mTOR is a major PHAS-I kinase. Included in this evidence are findings that mTOR phosphorylates the same sites in PHAS-I that are phosphorylated in a rapamycin-sensitive manner in cells (7,17). Rapamycin treatment of cells has a more pronounced effect on inhibiting phosphorylation of Thr 69 than the phosphorylation of Thr 36 /Thr 45 (5,37). Likewise, the phosphorylation of Thr 69 by mTOR in vitro is more sensitive to rapamycin-FKBP12 than the phosphorylation of Thr 36 /Thr 45 (7,29) (Fig. 4). Very recently Hara et al. (18) demonstrated that PHAS-I associates directly with raptor, a binding partner of mTOR, providing a close physical link between PHAS-I and mTOR.
The previous finding that raptor enhanced phosphorylation of PHAS-I by mTOR (18) led us to investigate the hypothesis that the TOS and/or RAIP motifs facilitate PHAS-I binding to raptor. The finding that purified wild-type PHAS-I, but not PHAS-I proteins with disrupted RAIP and TOS motifs, coimmunoprecipitated with HA-tagged raptor (Fig. 7) supports this hypothesis. Also, overexpression of raptor with mTOR increased the phosphorylation of wild-type PHAS-I but not the mutant proteins by mTOR in vitro (Fig. 8). Although we favor the interpretation that the motifs enhance PHAS-I phosphorylation by allowing the interaction of PHAS-I with raptor, we cannot eliminate the possibility that increasing raptor increases the binding of PHAS-I to mTOR or perhaps other proteins associated with the raptor-mTOR complex. Such proteins might become limiting for activity with overexpression of both raptor and mTOR. This could explain why the percentage increase in PHAS-I kinase activity produced by raptor overexpression appeared to be less than the increase in the raptor-mTOR complex (5-fold versus 10-fold) (see Fig. 8B and Fig. 9B).
In S. cerevisiae TOR proteins are found in two complexes (20). TORC1 contains TOR1 or TOR2, LST8, a protein with seven WD-40 domains, and KOG1, the yeast counterpart to raptor. TORC1 binds rapamycin with high affinity. Another complex, TORC2, contains TOR2, LST8, and the three proteins, AVO1, AVO2, and AVO3 (20). TORC2 does not bind rapamycin, possibly because of the lack of raptor, or because one or more of the AVO proteins interferes with the binding site. The evidence for a mammalian TORC1 is strong, as both raptor and mLST8 have been shown to coimmunoprecipitate with mTOR (20). A human AVO1 homolog, hSIN1, exists and its pattern of expression very closely resembles that of mTOR, although there is no direct evidence of association of mTOR and hSIN1 (20). Thus, whether complexes resembling TORC1 and TORC2 are formed with the single mammalian TOR protein is not known. FIG. 9. Coimmunoprecipitation of raptor and mTOR. A, 293T cells were transfected with vector alone, AU1-mTOR, or AU1-mTOR and HA-raptor. Cells were homogenized in buffer containing Tween 20, as in all other experiments in this report, or in buffer in which Triton X-100 was substituted for Tween 20. AU1-mTOR was immunoprecipitated with AU1 antibodies. Samples of the immunoprecipitates were incubated with mTAb1 before immune complex kinase assays were conducted using purified wild-type PHAS-I as substrate. A phosphorimage is presented ( 32 P). Other samples of the immunoprecipitates were immunoblotted with mTAb2, a polyclonal antibody to mTOR, and with HA antibodies. B, 293T cells were transfected with vector alone or with HA-raptor and/or AU1-mTOR. Immunoprecipitations were conducted with AU1-antibodies, and samples were immunoblotted with polyclonal raptor antibodies.
It is interesting that the resistance of Thr 36 /Thr 45 to rapamycin-FKBP12 was largely lost by disrupting either the RAIP or TOS motifs (Fig. 5B), as these results are suggestive of phosphorylation of PHAS-I by two complexes, one sensitive to rapamycin and the other insensitive. However, the two complexes do not necessarily correspond to mTORC1 and mTORC2. An explanation consistent with the present results is that tight binding to raptor, an interaction requiring both the TOS and RAIP motifs, provides a degree of protection from the inhibitory effects of rapamycin on phosphorylation of Thr 36 / Thr 45 . The present findings also suggest that optimal phosphorylation of Thr 36 /Thr 45 by mTOR depends on the high affinity interaction between PHAS-I and raptor, as phosphorylation was decreased by disrupting either the TOS or RAIP motifs.
The results involving Thr 69 phosphorylation are difficult to reconcile with a simple model. After raptor overexpression, disrupting the TOS and RAIP motifs actually had opposite effects on phosphorylation of this site by mTOR (Fig. 8, B and C). One interpretation is that phosphorylation of Thr 69 was mediated by different protein kinases. However, we do not believe that this is the case, as there is very good evidence that the Thr 36 /Thr 45 and Thr 69 sites are phosphorylated by mTOR. For example, dose response curves for the inhibition of phosphorylation of Thr 36 /Thr 45 by wortmannin, LY294002, and caffeine were identical to those for inhibition of Thr 69 (29). Moreover, mTOR harboring an Asp 2338 to Ala mutation in the catalytic domain is capable of phosphorylating neither Thr 36 / Thr 45 nor Thr 69 (7,29).
Phosphorylation of PHAS-I in cells occurs in an hierarchal manner (5,8), with the phosphorylation of Thr 36 and Thr 45 preceding that of Thr 69 and Ser 64 . Inspection of the stoichiometry of phosphorylation and the electrophoretic mobility of PHAS-I phosphorylated in Thr 36 /Thr 45 by mTOR in vitro reveals that the phosphorylation of the protein by mTOR also involves an ordered mechanism. For example, in Fig. 4 note that after incubating mTOR with mTAb1 approximately onethird of the P-Thr36/45 antibody reactivity in wild-type PHAS-I was up-shifted. Because a previous study (1) has established that phosphorylation Thr 36 and Thr 45 does not significantly retard the electrophoretic mobility of PHAS-I, the up-shift in P-Thr 36 /Thr 45 immunoreactivity must have been caused by phosphorylation of another site. In this experiment, 32 P incorporation indicated that mTOR introduced a maximum of 0.034 mol of phosphate per mol of PHAS-I. Therefore, if phosphorylation of the site responsible for the gel shift had been equally likely to occur in nonphosphorylated PHAS-I (at least 96.6% of the total) and Thr 36 /Thr 45 -phosphorylated protein, then the up-shifted fraction of P-Thr36/45 reactivity should have been negligible (less than 3.4% of the total). The site responsible for the shift is most likely Thr 69 , whose phosphorylation is known to promote the ␣ to ␤ shift (1). In a previous study, we found that mutating Thr 36 and Thr 45 to Ala markedly diminished the rate of Thr 69 phosphorylation by mTOR in vitro (29), providing further support for the conclusion that phosphorylation of Thr 36 and Thr 45 is needed for optimal phosphorylation of Thr 69 .
Regardless of the explanation, the results are consistent with the view that the RAIP motif suppresses Thr 69 phosphorylation. As a working hypothesis, we propose that phosphorylation of Thr 36 /Thr 45 , or the ⌬16 and the Ile 15 -Pro 16 to Ala 15 -Ala 16 mutations, reverse the inhibitory influence of the RAIP motif. We also propose that interaction of PHAS-I with raptor, mediated by the TOS motif, is still required for the phosphorylation of Thr 69 , though the strength of this interaction is insufficient to allow coimmunoprecipitation of PHAS-I with raptor. In this model, disrupting either the TOS or RAIP motifs would be expected to decrease phosphorylation of Thr 36 and Thr 45 , the initial phosphorylation events that are enhanced by tight binding to raptor. Disrupting the RAIP motif would obviate the need of Thr 36 /Thr 45 phosphorylation for Thr 69 phosphorylation, thereby allowing increased phosphorylation of Thr 69 when the raptor-mTOR complex was increased by raptor overexpression. On the other hand, disrupting the TOS motif would be expected to decrease Thr 69 phosphorylation by reducing both the interaction of PHAS-I with raptor and the phosphorylation of Thr 36 /Thr 45 .