Immunopurified Mammalian Target of Rapamycin Phosphorylates and Activates p70 S6 Kinase α in Vitro

, p70 S6 kinase a (p70 a ) is activated in vivo through a multisite phosphorylation in response to mitogens if a sufficient supply of amino acids is available or to high concentrations of amino acids per se . The immunosuppressant drug rapamycin inhibits p70 a activation in a manner that can be overcome by coexpression of p70 a with a rapamycin-resistant mutant of the mammalian target of rapamycin (mTOR) but only if the mTOR kinase domain is intact. We report here that a mammalian recombinant p70 a polypeptide, extracted in an inactive form from rapamycin-treated cells, can be directly phosphorylated by the mTOR kinase in vitro predominantly at the rapamycin-sensitive site Thr-412. mTOR-cata-lyzed p70 a phosphorylation in vitro is accompanied by a substantial restoration in p70 a kinase activity toward its physiologic substrate, the 40 S ribosomal protein S6. Moreover, sequential phosphorylation of p70 a by mTOR and 3-phosphoinositide-dependent protein kinase 1 in vitro resulted in a synergistic stimulation of p70 a activity to levels similar to that attained by serum stimulation in vivo . These results indicate that mTOR is likely to function as a direct activator of p70 in vivo , although the relative contribution of mTOR-catalyzed p70 phosphorylation in each of the many circumstances that engender p70

p70 S6 kinase ␣ (p70␣), 1 whose major substrate is the 40 S ribosomal protein S6, plays a critical role in the translation of a subclass of mRNAs that contain a short oligopyrimidine sequence immediately following the transcriptional start site (1). p70␣ is activated in response to insulin/mitogens in vivo through a multisite phosphorylation of serine and threonine residues (2). Several sets of independently regulated p70␣ phosphorylation sites have been identified (3)(4)(5)(6); one set consists of Ser/Thr-Pro motifs, five of which are clustered in a psuedosubstrate autoinhibitory domain in the noncatalytic carboxyl-terminal tail (Ser-434, Ser-441, Ser-447, Ser-452, and Thr-444 in p70␣), and two others, Thr-390 and Ser-394, are located in a 65-amino acid segment immediately carboxyl-terminal to the kinase catalytic domain. A second set of regulated phosphorylation sites, Thr-412 and Ser-427, exhibit the sequence motif Phe-Ser/Thr-Phe/Tyr. Thr-252, located on the activation loop in catalytic subdomain VIII, is the site at which 3-phosphoinositide-dependent protein kinase 1 (PDK1) phosphorylates p70␣ (7,8). Among these, the phosphorylation of Thr-252, Ser-394, and Thr-412 is necessary for the activation of p70␣ kinase catalytic function; the attainment of physiologic levels of p70␣ activity results from a strongly synergistic, positive site-site interaction between the phosphorylated Thr-252 and Thr-412 residues (7).
In addition to its regulation by insulin and mitogens through PI-3 kinase-dependent pathways, p70␣ can also be activated by increasing concentrations of extracellular amino acids in the absence of serum or mitogens to the level attained by maximal mitogen stimulation (9)(10)(11). Moreover, a threshold level of cellular amino acids is necessary for p70␣ to be susceptible to activation by mitogens. Withdrawal of amino acids from the nutrient medium results in a rapid, selective deactivation of p70␣, which becomes unresponsive to mitogens; readdition of amino acids restores the mitogen responsiveness of p70␣ (9).
The immunosuppressant drug rapamycin inhibits p70␣ in vivo (12,13). This is achieved indirectly by the ability of a rapamycin-FKBP12 complex to bind to the mTOR polypeptide and inhibit mTOR kinase activity; mTOR mutants unable to bind the rapamycin-FKBP12 complex can rescue p70␣ from rapamycin-induced dephosphorylation and inhibition but only if the mTOR catalytic domain is intact (14,15). As for the biochemical steps by which the mTOR kinase controls p70␣ phosphorylation and activity, evidence is available in support of two independent, but nonexclusive mechanisms. The possibility that mTOR inhibits an inactivating p70␣-phosphatase is supported by both indirect and direct experiments. Thus, a doubly deleted p70␣ mutant (p70␣-⌬2-46/⌬CT104) can be activated by mitogens and inhibited by low concentrations of wortmannin but is insensitive to inhibition by rapamycin (3) or amino acid withdrawal (9); these features are most readily explained if mitogens and PI-3 kinase control p70␣-kinases, whereas amino acid sufficiency and mTOR negatively regulate a p70␣-phosphatase. A recent report provides direct evidence implicating protein phosphatase 2A in this role (16). Conversely, Burnett et al. (17) reported that mTOR can directly phosphorylate prokaryotic recombinant fragments of p70␣ in vitro at sites important to activation, including Thr-412. The latter finding was surprising, inasmuch as all sites of mTOR-catalyzed phosphorylation on the eukaryotic initiation factor-4E binding protein 1 (eIF-4E BP1) reside in Ser/Thr-Pro motifs (18,19). We therefore inquired whether mTOR can phosphorylate and/or activate a precisely folded full-length p70␣ polypeptide expressed in mammalian cells and dephosphorylated and inactivated in vivo by the pretreatment with rapamycin.
Cell Lysis and Immunoprecipitation-Recombinant p70␣ was prepared from HEK293 cells that were transfected with various p70␣ cDNAs and extracted after serum starvation for 16 h and the following treatment with 0.2 M rapamycin for 30 min prior to harvest. Endogenous mTOR was extracted from HEK293 cells, and recombinant mTOR was extracted from HEK293 cells transfected with mTOR cDNAs. All cells were lysed in ice-cold buffer A (20 mM Tris (pH 7.4), 20 mM NaCl, 1 mM EDTA, 20 mM ␤-glycerophosphate, 5 mM EGTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 2 g/ml aprotinin, 1 M leupeptin), and the supernatants were obtained after centrifugation at 10,000 ϫ g for 20 min at 4°C. The anti-mTOR immunocomplex and the anti-recombinant p70␣ immunocomplex were each prepared using separate supernatants incubated 2 h at 4°C with the specific antibodies. The two supernatants were then mixed together, and protein G-Sepharose beads were added immediately and incubated for 45 min at 4°C. Normal mouse immunogloblin was added to cell supernatant to provide control for the anti-mTOR immunocomplex, whereas the anti-HA or anti-FLAG antibodies were added to cell supernatant prepared from mock-transfected cells to provide the controls corresponding to recombinant HA-tagged mTOR or FLAG-tagged p70␣ immunocomplexes. The protein G beads containing immunocomplexes were washed twice with buffer A containing 0.5 M NaCl (the high salt wash buffer) followed by two washes with either the same high salt wash buffer or the high salt wash buffer containing 1% Nonidet P-40 or the RIPA buffer. The RIPA buffer consisted of 20 mM Tris (pH ϭ 7.4), 1% Triton X-100, 0.1% SDS, 0.1% deoxycholate, and 150 mM NaCl. The immunocomplexes were further washed twice with buffer B (10 mM Hepes (pH ϭ 7.4), 50 mM ␤-glycerophosphate, 50 mM NaCl) and subjected to the kinase assay.
Kinase Assays and Immunoblot-The kinase reaction was started by the addition of buffer C (10 mM Hepes (pH ϭ 7.4), 50 mM NaCl, 50 mM ␤-glycerophosphate, 10 mM MnCl 2 , 100 M ATP (10 Ci of [␥-32 P] ATP). The reaction was incubated for 30 min at 30°C and terminated by the addition of the SDS sample buffer. The ability of mTOR to stimulate the kinase activity of p70␣ toward S6 in vitro was measured using a two step kinase reaction. In the first step, after p70␣ was immunoprecipitated with mTOR on protein G-Sepharose beads, the immunoprecipitate was washed twice with the high salt wash buffer containing 1% Nonidet P-40 and twice with buffer B, and the immunoprecipitate was incubated in buffer C with nonradioactive ATP for the indicated times. The first kinase reaction was terminated by washing the beads twice with the ice-cold high salt wash buffer and twice with buffer D (20 mM MOPS (pH 7.4), 10 mM ␤-glycerophosphate, 1 mM dithiothreitol) and subjected to the second kinase assay. The samples were incubated in the S6 kinase assay mixture (50 mM MOPS (pH 7.2), 12 mM MgCl 2 , 2 mM EGTA, 10 mM ␤-glycerophosphate, 0.5 M protein kinase inhibitor, 1 mM dithiothreitol, 0.5 A 260 units of 40 S ribosomal subunit, and 60 M ATP (5 Ci of [␥-32 P] ATP) for 15 min at 30°C, and the reaction was terminated by the addition of the SDS sample buffer. To measure the effects of GST-PDK1 on mTOR-catalyzed activation of p70␣, a three step kinase assay was employed. The first kinase reaction was performed as described above, and the reaction was terminated by washing the beads with the ice-cold buffer E (50 mM Tris (pH 7.4), 0.1 mM EGTA). The second kinase reaction was initiated by adding buffer F (50 mM Tris (pH 7.4), 0.1 mM EGTA, 1 mM dithiothreitol, 2 M protein kinase inhibitor, 10 mM MgCl 2 , 1 mg/ml bovine serum albumin, 100 M ATP) containing either purified GST-PDK1 (7) or control buffer and continued for 30 min at 30°C. The second kinase reaction was terminated by washing the beads with the ice-cold high salt wash buffer twice and buffer D twice and subjected to the S6 kinase assay as described above.
After the kinase reaction was terminated, the reaction mixtures were separated on SDS-polyacrylamide gel electrophoresis and transferred onto a polyvinylidene difluoride membrane. The membrane was analyzed by autoradiography using x-ray film and a BAS-2000 Bioimaging analyzer (Fuji). Then, the membrane was immunoblotted with the indicated antibody as the first antibody and visualized by using ECL method.

RESULTS AND DISCUSSION
To obtain a precisely folded full-length p70␣ polypeptide that was expressed in mammalian cells and dephosphorylated and inactivated in vivo, HEK293 cells were transiently transfected with a full-length HA-tagged p70␣, and cells were deprived of serum and treated with rapamycin (0.2 M) for 30 min prior to harvest. HA-tagged p70␣ was immunopurified on protein G-Sepharose beads, together with either a control immunoglobulin or an anti-mTOR immunocomplex; each were prepared separately from extracts of HEK293 cells. To define optimal conditions for detection of mTOR-catalyzed p70␣ phosphorylation, the protein G-Sepharose beads were washed in several ways prior to the kinase assay ( Fig. 1A). Although mTORcatalyzed p70␣ phosphorylation was detectable with all washing conditions, the mTOR autophosphorylation and kinase activity toward p70␣ is substantially enhanced when the immunoprecipitate is washed with the high salt wash buffer containing 1% Nonidet P-40 or the RIPA buffer, compared with that prepared after washing with the high salt wash buffer without detergent. The result was unexpected, as we have previously shown that washing of mTOR immunoprecipitates with 1% Nonidet P-40 reduces greatly the ability of mTOR to catalyze eIF-4E BP1 phosphorylation (20). These differences in mTOR-catalyzed p70␣ phosphorylation are not because of differences in the recovery of mTOR polypeptide, as demonstrated by the immunoblot with the anti-mTOR antibody (Fig. 1A). mTOR-catalyzed 32 P incorporation into p70␣ is detectable within 5 min after initiation of the kinase reaction and increases over 30 min (Fig. 1B); no 32 P incorporation into the wild-type p70␣ substrate (i.e. p70␣ autophosphorylation) is detectable in the absence of mTOR (Fig. 1B, lanes 6 and 7). Moreover, mTOR-catalyzed 32 P incorporation into the kinaseinactive, ATP binding site mutant of p70␣ (p70␣-Lys-123 3 Met) is similar in extent to that seen with wild-type p70␣ substrate (data not shown). These results show that the 32 P incorporation into p70␣ occurring in the presence of mTOR is catalyzed by a mTOR-associated kinase and is not because of a stimulation of p70␣ autophosphorylation.
To confirm that the phosphorylation of p70␣ by mTOR is dependent on the intrinsic kinase activity of mTOR, we compared the ability of a recombinant wild-type or kinase-negative mutant of mTOR to phosphorylate p70␣ in vitro (Fig. 1C). In contrast to the robust 32 P incorporation into p70␣ catalyzed by recombinant wild-type mTOR, no phosphorylation of p70␣ is detectable on incubation with the kinase-negative mutant of mTOR. We next examined the effects of mTOR kinase inhibitors on mTOR-catalyzed p70␣ phosphorylation in vitro (Fig.  1D). Incubation of mTOR immunoprecipitates with a rapamycin-FKBP complex severely inhibits mTOR autophosphorylation, as well as the phosphorylation of p70␣, whereas neither rapamycin or FKBP singly have any effect. Wortmannin, at a concentration previously shown to inhibit mTOR kinase toward eIF-4E BP in vitro also inhibits mTOR-catalyzed p70␣ phosphorylation. These results indicate that the phosphorylation of p70␣ in vitro requires the intrinsic kinase activity of mTOR. To examine the effects of stimulation by mitogens and depletion of amino acids on the mTOR kinase activity, serum-deprived cells were treated with or without 10% serum in the presence of amino acids or incubated for up to 2 h in the amino acid-free buffer; mTOR was immunoprecipitated and assayed for kinase activity toward p70␣. No significant alterations in the mTOR kinase toward p70␣ resulted from these treatments (data not shown); thus the effects of these perturbations on mTOR kinase activity toward p70␣, if any, do not survive immunoprecipitation and washing.
We employed a panel of anti-p70␣ phosphopeptide antibod- ies (21) to examine whether mTOR catalyzed the phosphorylation in vitro of sites on p70␣ known to be phosphorylated in vivo. The phospho-specific immunoreactivity at all sites examined appeared to be increased by mTOR; however, this was most unmistakable with Thr-412 (Fig. 1, A and B), which exhibits no phospho-specific immunoreactivity prior to incubation with mTOR. Overall phospho-specific immunoreactivity at Thr-444/Ser-447 is also substantially increased over the initial level, whereas the response at Ser-434 is equivocal in that the modest apparent increase in overall phospho-specific immunoreactivity at Ser-434 may be entirely attributable to the upshift of a fraction of p70␣ polypeptides resulting from mTOR-catalyzed phosphorylation at other sites with a consequent spreading of Ser-434-P immunoreactivity over a greater area.
In view of the limitation of immunoblot for quantitative analysis, we compared the extent of mTOR-catalyzed 32 P incorporation into wild-type p70␣ with that observed using equal amounts of several p70␣ mutants as substrates. The p70␣ mutants examined were (i) p70␣-5A in which five Ser/Thr-Pro sites (Ser-434, Ser-441, Thr-444, Ser-447, and Ser-452) in the carboxyl-terminal autoinhibitory domain are substituted by Ala; (ii) p70␣-Thr-412 3 Ala; (iii) p70␣-⌬CT104 in which the carboxyl-terminal 104 amino acids are deleted and the protein terminates after Ser-421; (iv) p70␣-⌬CT104/Thr-412 3 Ala. The quantitative importance of Thr-412 as a site of mTORcatalyzed p70␣ phosphorylation is clearly evident in Fig. 2; mutation of p70␣ Thr-412 to Ala reduces mTOR-catalyzed 32 P incorporation into full-length p70␣ by about 80% and into p70␣-⌬CT104 by a similar extent; the mTOR-catalyzed 32 P incorporation into the p70␣-⌬CT104/Thr-412 3 Ala mutant is less than 10% of that seen with full-length p70␣ wild-type (Fig.  2B). Thus, Thr-412 is a dominant site of mTOR-catalyzed p70␣ phosphorylation in vitro. Overall mTOR-catalyzed 32 P incorporation into the p70␣-5A mutant is diminished by about 25% compared with p70␣ wild-type, whereas 32 P incorporation into the ⌬CT104 mutant is diminished by 50 -60%. These results indicate that a portion of mTOR-catalyzed p70␣ phosphorylation is directed to the carboxyl-terminal tail, at least half of which is into the Ser/Thr-Pro sites mutated in the 5A variant. This is consistent with the results of the anti-Thr-444/Ser-447-P immunoblots (Fig. 1, A and B, and Fig. 2A). The lesser total 32 P incorporation into p70␣-⌬CT104 as compared with p70␣-5A suggests that there might be phosphorylation site(s) other than the five Ser/Thr-Pro sites within the carboxyl-terminal 104 amino acids. One such site may be Ser-427, located in a Phe-Ser-Phe motif similar to that surrounding Thr-412. Another possible explanation is that the absence of the carboxyl-terminal 104 amino acids may impair the ability of mTOR to phosphorylate Thr-412.
The phosphorylation of p70␣ Thr-412 is known to be critical for its S6 kinase activity and substitution of this residue with an acidic amino acid results in a substantial increase in "basal" S6 kinase activity (5,21). We therefore inquired whether the p70␣ kinase activity is increased by mTOR-catalyzed phosphorylation in vitro. As shown in Fig. 3, the p70␣ kinase activity shows a time-dependent increase on incubation with mTOR (Fig. 3B), which parallels the extent of mTOR-catalyzed phosphorylation at Thr-412 and Thr-444/Ser-447 detected by immunoblot (Fig. 3A). In contrast, no S6 kinase activity is detectable in the absence of mTOR. To establish whether the activation of p70␣ in vitro requires the intrinsic kinase activity of mTOR, the recombinant wild-type and kinase-negative mutant of mTOR were employed for the assays (Fig. 3C). As in Fig. 3A, incubation of p70␣ with wild-type mTOR significantly increased the S6 kinase activity, whereas no activation is detected on incubation of p70␣ with kinase-negative mTOR.
These results clearly indicate that the kinase activity of p70␣, which had been fully inactivated in vivo by the treatment of cells with rapamycin, was restored, at least in part, by the phosphorylation in vitro catalyzed by the kinase activity intrinsic to the mTOR catalytic domain.
PDK1-catalyzed phosphorylation of Thr-252 on the p70␣ activation loop is a critical and probably final step in the physiologic activation of p70␣ in vivo (7,8,21). The ability of PDK1 to phosphorylate Thr-252 is regulated primarily by the accessibility of the p70␣ activation loop to PDK1, which in turn is  12,74. B, the phosphorylation of p70␣ mutants by mTOR was carried in the separate experiments, and 32 P incorporated into FLAG-tagged p70␣ mutants was quantified. 32 P incorporated into each mutant was expressed as a percentage of that into wild-type p70␣. Data are the mean Ϯ S.D. of three experiments.
controlled by a series of prior p70␣ phosphorylations. Phosphorylation of the multiple Ser/Thr-Pro motifs clustered in the psuedosubstrate autoinhibitory segment of the p70␣ carboxylterminal tail serves to disocclude the catalytic domain, greatly enhancing access to PDK1. A similar effect can be achieved by deletion of the p70␣ carboxyl-terminal tail (to give p70␣-⌬CT104). At any level of PDK1 activity, the extent of Thr-252 phosphorylation of p70␣-⌬CT104 is substantially greater than with a similar amount of full-length p70␣ polypeptide. In addition, the S6 kinase activity generated by any extent of PDK1catalyzed Thr-252 phosphorylation is significantly higher for p70␣-⌬CT104 as compared with full-length p70␣. Displacement of the p70␣ carboxyl-terminal tail is also necessary for the phosphorylation of Thr-412 in vivo, and modification of Thr-412 itself significantly enhances the ability of PDK1 to phosphorylate Thr-252. In addition, the simultaneous phosphorylation of Thr-412 and Thr-252 appears to generate a synergistic activation of p70␣. Thus, the substitution of Thr-412 by Glu in p70␣-⌬CT104 alone gives a 6-fold increase in S6 kinase activity, and the PDK1 catalyzed phosphorylation of p70␣-⌬CT104 Thr-252 alone gives a 15-fold increase, but the two modifications together give at least a 240-fold increase in S6 kinase activity over the unmodified p70␣-⌬CT104 polypeptide (7). The importance of the strong positively cooperative effect of Thr-252 and Thr-412 phosphorylation for the physiologic activation of p70␣ is illustrated by the response of p70␣ to the inhibitors rapamycin and wortmannin; these agents each cause a rapid dephosphorylation of Thr-412 but a slower and lesser dephosphorylation of Thr-252. Despite the preservation of Thr-252 phosphorylation, S6 kinase activity in the presence of rapamycin or wortmannin declines in parallel to Thr-412 dephosphorylation (21). In view of the ability of mTOR to catalyze the in vitro phosphorylation of p70␣ Thr-412 as well as sites within the autoinhibitory segment in the p70␣ carboxyl-terminal tail and the potential effects of such phosphorylations on the response of p70␣ to PDK1, we compared the p70␣ activation achieved in vitro by mTOR or PDK1 alone to that achieved by sequential phosphorylation by mTOR and PDK1 and to that achieved in vivo by stimulation of the cells with 10% serum. As shown in Fig. 4, mTOR alone increased the S6 kinase activity of p70␣ in vitro by more than 10-fold, whereas PDK1 alone  16, 82.8. B, the activation of p70␣ by mTOR was carried out in the separate experiments, and 32 P incorporated into S6 was quantified. 32 P incorporated into S6 at the indicated times was expressed as a percentage of that at 45 min. Data are the mean Ϯ S.D. of three experiments. C, FLAG-tagged wild-type p70␣ was immunoprecipitated without (Ϫ) or with either HA-tagged wild-type (W) or kinase negative mTOR. The immunoprecipitates were subjected to the two step kinase assay as described in Fig. 4A 4. Synergistic activation of p70␣ by mTOR and PDK1. HA-tagged wild-type p70␣ that had been treated by 0.2 M rapamycin (lanes 3-10) was immunoprecipitated with (ϩ) or without (Ϫ) endogenous mTOR, and the immunoprecipitates were subjected to the three step kinase assay as described under "Experimental Procedures." In lanes 11-14, HA-tagged wild-type p70␣ was stimulated with 10% serum in vivo, and a different amount of HA-p70␣ (100, 80, 66.6, and 50% (lanes 11-14, respectively) of the amount of p70␣ used in lanes [3][4][5][6][7][8][9][10] was immunoprecipitated and subjected to the S6 kinase assay. The samples were analyzed by autoradiography (top) and immunoblot with the indicated antibodies (bottom hardly activated p70␣, presumably reflecting the relatively poor access of Thr-252 to PDK1 in full-length, inactive p70␣ as seen previously (7). In contrast, phosphorylation of p70␣ by PDK1 after a prior phosphorylation by mTOR increased the p70␣ activity by 10-fold over that engendered by mTOR alone, to a level roughly 70-fold greater than that generated by PDK1 acting alone. Moreover, the S6 kinase activity generated in vitro by the sequential action of mTOR and PDK1 is indistinguishable from that achieved in vivo by stimulation of cells with 10% serum (Fig. 4).
The present results demonstrate that mTOR can catalyze directly the phosphorylation and activation of p70␣ in vitro. In addition, mTOR can activate p70␣ in a synergistic manner with PDK1 in vitro, and it is likely that this occurs in vivo. Nevertheless, the nature of the physiologic inputs that control mTOR-catalyzed p70␣ phosphorylation and the relative contribution of mTOR-catalyzed p70␣ phosphorylation to overall p70␣ regulation in vivo are not clear. Pretreatment of 3T3-L1 cells with insulin has been reported to cause a modest (1.3-2fold) increase in the ability of mTOR to catalyze eIF-4E BP1 phosphorylation in vitro. Moreover, coexpression with active protein kinase B may enhance mTOR kinase, although evidence for direct activation of mTOR by protein kinase B is lacking. On this basis, mTOR has been proposed to be an intermediate in the insulin/PI-3 kinase-dependent activation of p70␣ (22). To the contrary, the ability of the rapamycin-resistant p70␣-⌬2-46/⌬CT104 mutant to undergo insulin-stimulated, wortmannin-inhibitable Thr-412 phosphorylation and activation in the presence of concentrations of rapamycin far in excess of those required for complete inhibition of endogenous wild-type p70␣ and mTOR indicates that insulin-responsive kinases exist that are capable of p70␣ (Thr-412) phosphorylation and activation, other than mTOR (9).
Thus, the contribution of mTOR to insulin-stimulated p70␣ Thr-412 phosphorylation in vivo is unsettled but may be minor. Another possible role for the mTOR kinase is as the mediator of the amino acid-stimulated phosphorylation and activation of p70␣; as yet however, direct evidence supporting amino acid regulation of the mTOR kinase activity is lacking (9). A plausible synthesis for the operation of the mTOR kinase in vivo is that insulin-and amino acid-induced signals each converge independently in the regulation of mTOR, although insulin also controls an alternative, mTOR-independent set of p70␣-kinases. In turn, mTOR controls p70␣ through direct phosphorylation, as well as through the negative regulation of a p70␣ phosphatase.
Gingras et al. (19) reported that mTOR phosphorylates eIF-4E BP1 in vitro primarily on Thr-37 and Thr-46; these phosphorylations do not themselves result in the release of eIF-4E but are required for the further phosphorylation on several carboxyl-terminal serum-sensitive sites, and these latter phosphorylations result in the release of eIF-4E. In the case of p70␣, mTOR alone gives some activation, but by phosphorylating the Ser/Thr-Pro sites within the carboxyl-terminal 104 amino acids to improve access of PDK1 and by phosphorylating Thr-412, mTOR strongly promotes the ability of PDK1 to phosphorylate Thr-252. In both instances, mTOR phosphorylation acts primarily in a "priming" role rather than a sole activator.
An intriguing and unresolved aspect of mTOR function is the apparent ability of its single catalytic domain to catalyze phos-phorylation of Ser/Thr-Pro sites, such as those on eIF-4E BP1 and p70␣ Thr-444/Ser-447, as well as Phe-Ser/Thr-Phe/Tyr sites, such as p70␣ Thr-412. Although these two kinds of mTOR kinase activity appear differentially sensitive to detergent, they are both inhibited by rapamycin/FKBP-12 in vitro and by mutations in the mTOR kinase domain. Assuming both activities are physiologically meaningful, such a breadth in substrate specificity is relatively unprecedented among the protein kinases. Finally, it should be noted that Phe-Ser/Thr-Phe/Tyr motifs homologous to p70␣ Thr-412 are found in many kinases of the AGC (protein kinase A, G, and C) subclass, and it is of interest to note the recent report that PKC␦ activation and phosphorylation at Ser-662 (in the context Phe-Ser-Phe) is inhibitable by rapamycin (23). In conclusion, the present study demonstrates the in vitro activation of p70␣ by mTOR-catalyzed phosphorylation involving p70␣ Thr-412, a critical site conserved in the other AGC kinase subfamily members.