Regulation of an activated S6 kinase 1 variant reveals a novel mammalian target of rapamycin phosphorylation site.

A critical step in S6 kinase 1 (S6K1) activation is Thr(229) phosphorylation in the activation loop by the phosphoinositide-dependent protein kinase (PDK1). Thr(229) phosphorylation requires prior phosphorylation of the Ser/Thr-Pro sites in the autoinhibitory domain and Thr(389) in the linker domain, consistent with PDK1 more effectively catalyzing Thr(229) phosphorylation in a variant harboring acidic residues in these positions (S6K1-E389D(3)E). S6K1-E389D(3)E has high basal activity and exhibits partial resistance to rapamycin and wortmannin, and its activity can be further augmented by mitogens, effects presumably mediated by Thr(229) phosphorylation. However, PDK1-induced Thr(229) phosphorylation is reported to be constitutive rather than phosphatidylinositide 3,4,5-trisphosphate-dependent, suggesting that S6K1-E389D(3)E activity is mediated through a distinct site. Here we use phosphospecific antibodies to show that Thr(229) is fully phosphorylated in S6K1-E389D(3)E in the absence of mitogens and that regulation of S6K1-E389D(3)E activity by mitogens, rapamycin, or wortmannin parallels Ser(371) phosphorylation. Consistent with this observation, a dominant interfering allele of the mammalian target of rapamycin, mTOR, inhibits mitogen-induced Ser(371) phosphorylation and activation of S6K1-E389D(3)E, whereas wild type mTOR stimulates both responses. Moreover, in vitro mTOR directly phosphorylates Ser(371), and this event modulates Thr(389) phosphorylation by mTOR, compatible with earlier in vivo findings.

Cell growth is tightly coordinated by nutrient and energy availability as well as local growth factor concentrations (1,2). Studies in Drosophila and the mouse have shown that the 40 ribosomal protein S6 kinases play a critical role in cell growth as a function of nutrient and energy availability (3)(4)(5). The few dS6K-deficient flies that survive to adults do so after a severe delay in development (6). Furthermore, such flies are half the size of wild type flies, due to a reduction in cell size rather than cell number. In contrast, S6K1 1 -deficient mice are viable and fertile, although S6K2 is compensatorily up-regulated in these animals (3). Despite the presence of S6K2, S6K1-deficient mice exhibit a conspicuous reduction in body size during embryogenesis and adulthood. Such mice are hypoinsulinemic and glucose-intolerant, due to a reduction in both insulin secretion and insulin content in ␤ cells (7). These deficits are due to a sharp reduction in pancreatic endocrine mass, which is accounted for by a selective decrease in ␤ cell size rather than cell number. This phenotype closely parallels that of protein malnutritioninduced type 2 diabetes mellitus (7).
The role of S6K1 in cell growth has been illuminated by an increasing knowledge of the molecular mechanisms and the signaling components that control its activation (8). S6K1 activation involves a complex interplay between ordered phosphorylation events, which occur within distinct intramolecular regulatory domains. The first identified S6K1 phosphorylation sites, Ser 411 , Ser 418 , Thr 421 , and Ser 424 , are clustered within an autoinhibitory domain in the carboxyl terminus (9). These sites contain a proline in the ϩ1-position, and results involving substitution with uncharged or acidic amino acids suggest that their function is to modulate kinase activity (10). In contrast, Thr 229 , in the activation loop, and Ser 371 and Thr 389 , in the conserved linker domain, appear critical for activation (11)(12)(13), since replacement of any of these three sites with an alanine residue abolishes kinase activity (11,13). In the full-length enzyme, phosphorylation of Thr 229 and Thr 389 is tightly regulated by growth factors and inhibited by rapamycin and wortmannin treatment (10,12). A significant degree of resistance to both inhibitors is achieved by substituting acidic residues for Thr 389 and for the carboxyl-terminal Ser/Thr-Pro sites in the autoinhibitory domain (11,14). Interestingly, the element of sensitivity displayed by S6K1-E389D 3 E in the presence of either inhibitor is roughly equivalent to its ability to be activated by mitogens, an effect thought to be mediated by Thr 229 phosphorylation (11,14).
Recently, it has been demonstrated that Thr 229 phosphorylation is mediated by PDK1 (15,16), whereas Thr 389 as well as Ser 411 , Thr 421 , and Ser 424 phosphorylation have been shown to be regulated by mTOR (17). Efficient Thr 229 phosphorylation appears to require Thr 389 phosphorylation and the Ser/Thr-Pro FIG. 1. Effect of insulin and dominant negative PDK1 on Thr 229 phosphorylation. A, schematic representation of wild-type S6K1 and S6K1-E389D 3 E. The kinase domain and autoinhibitory domain are shown in solid black and hatched bars, respectively. The Ser/Thr-Pro sites in linker and autoinhibitory domains, which were mutated to acidic residues, are indicated with single-letter amino acid codes. B, exponentially growing HEK 293 cells were transiently transfected with either Myc-S6K1 or Myc-S6K1-E389D 3 E, deprived of serum for 16 h, and then either extracted directly or following a 30-min treatment with 100 nM insulin. Myc-S6K1 and Myc-S6K1-E389D 3 E were immunoprecipitated and measured for S6 kinase activity using 40 S ribosomes as a substrate. The level of Thr 229 phosphorylation was followed by Western blot analysis using the anti-phospho-Thr 229 antibody, and the blot was stripped and reprobed with the anti-Myc 9E10 antibody (see "Experimental Procedures"). C, exponentially growing HEK 293 cells were transiently transfected with Myc-S6K1-E389D 3 E-GST alone or in the presence of Myc-PDK1-KI. All measurements were carried out as in B after purification of Myc-S6K1-E389D 3 E-GST on glutathione beads (see "Experimental Procedures"). Myc-PDK1-KI expression levels were evaluated by Western blot analysis with the anti-Myc 9E10 antibody (see "Experimental Procedures"). The results are typical of at least two independent experiments. sites in the autoinhibitory domain (18), since PDK1 more effectively catalyzes Thr 229 phosphorylation of S6K1-E389D 3 E than it does wild type S6K1 (16,18). PDK1 was initially identified as the kinase responsible for phosphorylation of the activation loop site, Thr 308 , in PKB (19,20). Since this reaction has an a priori requirement for phosphatidylinositide 3,4,5trisphosphate (19,20), the assumption has been that the corresponding reaction in S6K1 is also mediated as a downstream effect of PI3K (12,21). In support of this hypothesis, membrane-targeted alleles of the PI3K catalytic subunit, p110, constitutively activate PKB and S6K1 (22)(23)(24)(25)(26)(27)(28). However, no significant changes were observed in the ability of PDK1, derived from cells treated with either mitogens or the PI3K inhibitor, wortmannin, to phosphorylate S6K1-E389D 3 E in vitro (16). Furthermore, in contrast to S6K1, S6K1-E389D 3 E has elevated basal levels of Thr 229 phosphorylation (18), leading to the hypothesis that PDK1 is constitutively active and only requires disruption of the carboxyl and amino termini in S6K1 to access Thr 229 (16,18).
The findings above raised the possibility that mitogen-induced increases in S6K1-E389D 3 E activity, as well as its sensitivity to wortmannin and rapamycin, were mediated through a phosphorylation site distinct from Thr 229 . To test this possibility, phosphospecific antibodies to Thr 229 were generated to analyze its role in regulating S6K1-E389D 3 E activity by mitogens, rapamycin, and wortmannin. These results led to the finding that S6K1-E389D 3 E activation is regulated through a distinct phosphorylation site, which is rapamycin-and wortmannin-sensitive, and which serves as a direct substrate for mTOR.

EXPERIMENTAL PROCEDURES
Plasmid Construction-Unless otherwise stated, all S6K1 constructs were expressed as fusion proteins from a cytomegalovirus-driven expression vector, which was amino-terminally Myc-or HA-tagged as described elsewhere (18). Where referred to, S6K1 is the cytoplasmic isoform. HA-tagged mTOR and mTOR-KI (D2537E) have been described (2). Myc-tagged PDK1-KI (K111Q, D205N, D223N) was a kind gift from M. Frodin.
Cell Culture and Transfection-Human embryonic kidney 293 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. For transfection, 10 6 HEK 293 cells were seeded per 10-cm culture dish, and then, 24 h later, the cells were transiently transfected with 1-20 g of the appropriate construct, using a modified calcium phosphate precipitation procedure (18). Where required, the final DNA concentration was brought to 20 g with empty vector. After 12 h, the cells were washed twice with serum-free Dulbecco's modified Eagle's medium and maintained in same media for an additional 24 h prior to treatment and extraction. Where indicated, cells were treated with either 100 nM insulin or 10% serum for 30 min and were pretreated with 20 nM rapamycin or 250 nM wortmannin for 15 min prior to insulin stimulation and extraction. Cells were extracted for S6K1 as described previously (18) or for mTOR in 20 mM Tris, pH 7.5, 20 mM NaCl, 20 mM ␤-glycerophosphate, 1 mM EDTA, 0.5 mM dithiothreitol, 5 mM EGTA, and 10 g/ml leupeptin, and samples were centrifuged at 12,000 ϫ g for 20 min at 4°C prior to flash freezing in liquid N 2 and storage at Ϫ70°C.
Immunoblotting and Kinase Assays-Determination of protein concentrations, S6 kinase assays, immunoblotting, and chemiluminescence were carried out as previously described (2). Where indicated, the acrylamide concentrations were decreased from 10 to 6%, which allowed resolution of the S6K1 phosphoderivatives. Phosphospecific antibodies against Thr 389 and Thr 421 /Ser 424 of S6K1 were obtained from New England Biolabs and used as directed. mTOR kinase assays were performed as follows. HA-mTOR and Myc-or HA-S6K1 were separately extracted from transiently transfected 293 cells. The mTOR and S6K1 immunocomplexes were each prepared separately, incubated with the specific antibodies, mixed together, incubated with protein G beads, and washed extensively. Where indicated, the immunocomplex was incubated with GST-FKBP12 and rapamycin in a buffer containing 10 mM Hepes, 50 mM NaCl, 50 mM ␤-glycerophosphate, 10 mM MnCl 2 , and 1 mM ATP at 30°C for 60 min. The reaction mixture was washed to remove excess ATP and then further incubated with 40 S ribosomes in a kinase buffer containing 20 mM MOPS, pH 7.4, 10 M ␤-glycerophosphate, 0.5 mM dithiothreitol, 10 mM MgCl 2 , and [␥-32 P]ATP. Following SDS-PAGE, the upper or middle part of the gel was cut, transferred to membrane, and immunoblotted with the specific antibodies against mTOR or S6K1, respectively. The lower part of SDS-PAGE was analyzed by autoradiography or storage phosphorimaging (Molecular Dynamics, Inc., Sunnyvale, CA) and ImageQuant software (Molecular Dynamics) to measure S6 phosphorylation.
Generation of Anti-phosphopeptide Antisera to S6K1-Phosphospecific antibodies directed against Thr 229 and Ser 371 were produced by immunizing New Zealand White rabbits with the following synthetic phosphopeptides coupled to keyhole limpet hemocyanin: Thr 229 (GTVTHT*FCGTI) and Ser 371 (FTRQTPVDS*PDDSTLS). Nonphosphospecific antibodies were removed by adsorption to a peptide column generated from the nonphosphorylated peptide sequence. Phosphospecific antibodies that passed through this column were passed over a column composed from the phosphopeptide sequence and after washing antibodies were eluted with 100 mM glycine, pH 2.5, and then dialyzed against Tris-buffered saline. The specificity of the Thr 229 phosphospecific antibodies was confirmed by probing a Western blot containing wild type S6K1 and S6K1-A229.

RESULTS
Effect of Insulin on S6K1-E389D 3 E-The S6K1 variant, S6K1-E389D 3 E, which harbors acidic residues at both Thr 389 and in the autoinhibitory domain Ser/Thr-Pro sites (Fig. 1A), has increased basal kinase activity, which can be further augmented by serum or insulin stimulation (14,30). The assumption has been that the increase in basal activity is through increased phosphorylation of Thr 229 in the activation loop, facilitated by the acidic amino acid substitutions in the linker and autoinhibitory domain (11). The additional increase in insulin-induced S6K1-E389D 3 E activity was thought to be catalyzed by the activation of PI3K, leading to increased production of phosphatidylinositide 3,4,5-trisphosphate and activation of PDK1, the Thr 229 kinase (18). However, the activity of PDK1 appears to be constitutive and refractile to mitogen stimulation (19) and, with regard to S6K1 activation, phosphatidylinositide 3,4,5-trisphosphate-independent (15,16). To assess the role of Thr 229 phosphorylation in mediating S6K1-E389D 3 E activation, a Myc-tagged variant was ectopically expressed in 293 cells, and following insulin stimulation, the state of Thr 229 phosphorylation was monitored with a phosphospecific antibody (see "Experimental Procedures"). As a control, Thr 229 phosphorylation was followed in parallel in a Myc-tagged wild type S6K1 (Myc-S6K-WT), transiently expressed in 293 cells. Consistent with earlier studies by twodimensional phosphopeptide analyses (18), Thr 229 phosphorylation and S6 kinase activity in Myc-S6K-WT were increased following insulin treatment (Fig. 1B). However, Thr 229 phosphorylation and S6 kinase activity in Myc-S6K1-E389D 3 E were already elevated in serum-deprived cells (Fig 1B). It should be noted that the extent to which basal activity of Myc-S6K1-E389D 3 E is elevated varies as a function of expression level (14,30). Following insulin treatment, Thr 229 phosphorylation did not increase further, despite the fact that kinase activity toward S6 was increased ( Fig 1B). Consistent with PDK1 mediating Thr 229 phosphorylation, a Myc-tagged kinase inactive variant of PDK1 (Myc-PDK1-KI) (31), when co-expressed with Myc-S6K1-E389D 3 E, lowered basal and insulin-induced Thr 229 phosphorylation as well as S6 kinase activity (Fig. 1C). Taken together, these findings indicate that increased Thr 229 phosphorylation in S6K1-E389D 3 E is not responsible for an insulininduced increase in S6 kinase activity and that other phosphorylation sites or activating mechanisms must mediate this response.
Role of Ser 371 Phosphorylation-In previous studies, four additional residues, Thr 367 , Ser 371 , Ser 404 , and Thr 447 , were identified as insulin-induced phosphorylation sites (11,13). However, when either the serine or threonine residues at 367, 404, and 447 were individually replaced by either alanine or by an acidic amino acid, these variants showed no difference in kinase activity with mitogen treatment, as compared with wild type S6K1 (11,13). However, conversion of Ser 371 to either alanine or aspartic acid abolished kinase activity (13). Analysis of Ser 371 phosphorylation by two-dimensional phosphopeptide mapping following tryptic digestion is difficult, since this phosphorylation site migrates with several incomplete tryptic digestion products in a poorly resolved area of the chromatogram (13). To resolve this, an anti-phospho-Ser 371 -specific antibody was developed (see "Experimental Procedures"), which recognizes phosphorylated Ser 371 . Insulin stimulation of 293 cells transiently expressing Myc-S6K-WT led to an approximately 2-fold increase in Ser 371 phosphorylation in the kinase ( Fig.  2A), compatible with previous observations by two-dimensional phosphopeptide chromatography (13). This increase is paralleled by an increase in Thr 421 /Ser 424 phosphorylation ( Fig. 2A), as monitored with a previously described anti-phospho-Thr 421 / Ser 424 -specific antibody (32). These sites, like Ser 371 , also have a proline in the ϩ1-position and a hydrophobic residue in the Ϫ2-position. To ensure the specificity of the anti-phospho-Ser 371 antibody, its ability to recognize phosphorylated Ser 371 was examined in the Myc-S6K1-S371A variant, having an alanine substituted at this site. The results show that the antiphospho-Ser 371 -specific antibody did not recognize the Ser 371 variant, despite increased phosphorylation of Thr 421 /Ser 424 ( Fig. 2A). It is also evident that the alanine substitution has little effect on overall S6K1 phosphorylation as evidenced by the decreased mobility of the protein in low acrylamide gels (see "Experimental Procedures"), consistent with earlier find- ings (13). In addition, phosphorylation of Ser 371 in the kinaseinactive allele of S6K1, S6K1-KI, was clearly observed ( Fig.  2A), indicating that Ser 371 phosphorylation is not an autophosphorylation event. To determine whether Ser 371 phosphorylation is still regulated in S6K1-E389D 3 E, its level of phosphorylation was monitored in Myc-S6K1-E389D 3 E following transient expression in either quiescent or insulin-stimulated 293 cells. The results show that although basal levels of Ser 371 are elevated in quiescent cells, they are further increased following insulin stimulation, but as shown in Fig. 1A, there is no corresponding effect on Thr 229 phosphorylation (Fig. 2B). These data are consistent with Ser 371 being the site responsible for mediating S6K1-E389D 3 E activation.
Effect of Rapamycin and Wortmannin on Thr 229 and Ser 371 Phosphorylation-In previous studies, we have shown that S6K1-E389D 3 E retains an element of sensitivity to rapamycin and wortmannin, which is roughly equivalent to its ability to be activated by mitogens (14). This effect would not appear to be mediated through Thr 229 phosphorylation, since ectopically expressed PDK1 is resistant to both inhibitors (16). To assess the effect of the two inhibitors on insulin-induced Thr 229 and Ser 371 phosphorylation, 293 cells were transfected with Myc-tagged S6K1 or S6K1-E389D 3 E, and the level of phosphorylation at each site was followed as a function of the indicated treatment (Fig. 3). As observed previously, insulin-induced Thr 229 phosphorylation in S6K1 was blocked by the two inhibitors, and in each case, the level of Thr 229 phosphorylation closely paralleled S6 kinase activity (Fig. 3). In contrast to S6K1, basal Thr 229 phosphorylation was elevated in S6K1-E389D 3 E and was refractile to either insulin, wortmannin, or rapamycin treatment ( Figs. 1 and 3), consistent with the fact that none of these agents were found to influence PDK1 activity (16). However, S6K1-E389D 3 E kinase activity increased in response to insulin and was inhibited by pretreatment with rapamycin and wortmannin, demonstrating that the insulin-induced, rapamycinand wortmannin-sensitive components, which regulate S6K1-E389D 3 E activity, are not controlled by Thr 229 phosphorylation. Like Thr 229 phosphorylation, Ser 371 phosphorylation closely paralleled kinase activity in S6K1. More importantly, Ser 371 phosphorylation is inhibited by both rapamycin and wortmannin. However, in contrast to Thr 229 phosphorylation, Ser 371 phosphorylation of S6K1-E389D 3 E increased roughly 2-fold in the presence of insulin, and this increase was suppressed by either rapamycin or wortmannin (Fig. 3). The change in Ser 371 activity closely paralleled kinase activity, consistent with Ser 371 phosphorylation representing the regu- lated component in the activation of S6K1-E389D 3 E. Thus, Ser 371 phosphorylation is rapamycin-and wortmannin-sensitive and appears to be the common element whereby the two inhibitors modulate S6K1-E389D 3 E activity.
The Role of mTOR in Regulating Ser 371 Phosphorylation in Vivo-The findings that Ser 371 phosphorylation is blocked by rapamycin, which directly inhibits mTOR function (33), and that mTOR has been shown to mediate Ser/Thr-Pro phosphorylation in S6K1 as well as 4EBP-1 (17,34), raised the possibility that mTOR directly controls Ser 371 phosphorylation. In order to examine this possibility, a kinase-inactive variant, HA-mTOR-KI, was co-transfected with a Myc-tagged S6K1-E389D 3 E reporter into 293 cells. As shown in Fig. 4A, high basal kinase activity was detected in S6K1-E389D 3 E, and serum stimulation led to a small but significant increase in kinase activity and in Ser 371 phosphorylation. Both responses were significantly suppressed by the dominant interfering allele of mTOR (HA-mTOR-KI). Under these conditions, there was no effect on Thr 229 phosphorylation of ectopically expressed HA-mTOR-KI (data not shown). Given that HA-mTOR-KI blocked Ser 371 phosphorylation, we tested whether an HA-tagged wild type allele of mTOR (HA-mTOR-WT) could raise S6K activity. Co-expression of HA-mTOR-WT in 293 cells with S6K1-E389D 3 E slightly increased high basal state of S6K1-E389D 3 E kinase and Ser 371 phosphorylation (Fig. 4B). Furthermore, the increase in Ser 371 phosphorylation was inhibited by rapamycin treatment. Taken together, these findings indicate that Ser 371 is regulated by mTOR in vivo.
In Vitro Phosphorylation of Ser 371 by mTOR-Recent studies have shown that Thr 389 , Ser 411 , and Thr 421 /Ser 424 are direct in vitro substrates of mTOR (17). To test whether mTOR would also phosphorylate Ser 371 in vitro, Myc-S6K-WT derived from 293 cells pretreated with rapamycin was used as a direct substrate for HA-tagged mTOR immunopurified from 293 cells.
The results show that Myc-mTOR-WT induces increased Ser 371 phosphorylation in vitro as assessed by Western blot analysis with the Ser 371 anti-phosphospecific antibody following electrophoresis on low acrylamide gels (Fig. 5A), consistent with in vivo findings ( Fig. 2A). In contrast, HA-mTOR-KI had no effect. More importantly, in the presence of rapamycin and FKBP12, but not in the presence of either component alone, phosphorylation of Ser 371 by mTOR is abolished (Fig. 5A), consistent with the in vivo finding (Fig. 4). In parallel, incubation of S6K1-E389D 3 E with HA-mTOR-WT also led to increased S6K1-E389D 3 E activation and Ser 371 phosphorylation, in an FKBP-12/rapamycin-sensitive manner (Fig. 5B and data not shown). Furthermore, the extent of both responses was similar to those observed in vivo (compare Figs. 4B and 5B). Therefore, Ser 371 phosphorylation appears to be directly regulated by mTOR in vitro and in vivo.
Effect of Ser 371 on S6K1 Activation-As shown previously (13), substitution of an alanine or an aspartate for Ser 371 blocks serum-or insulin-induced Thr 389 phosphorylation and S6K1 activation (Fig. 6A). However, substitution of an acidic residue at Thr 389 in the S6K1-E389D 3 E background fails to rescue kinase activity (13), suggesting that Ser 371 phosphorylation contributes directly to S6K1 activation independent of its role in regulating Thr 389 phosphorylation. To test this possibility in vitro, either S6K1 or S6K1-S371A, from 293 cells pretreated with rapamycin, were incubated with either HA-mTOR-WT or HA-mTOR-KI. Both S6K1 variants displayed basal levels of phosphorylated Thr 229 , which were not altered by incubation with either mTOR variant (Fig. 6B). However, incubation of either S6 kinase variant with wild type, but not kinase-inactive, mTOR led to increased Thr 389 phosphorylation, with the extent of Thr 389 phosphorylation much higher in S6K1-S371A than in wild type S6K1 (Fig. 6B). However, to achieve the same level of activity as S6K1-WT, S6K1-S371A apparently requires much higher levels of Thr 389 phosphorylation (Fig. 6B), consistent with detailed titration studies (data not shown). Although unexpected, these findings are compatible with Ser 371 phosphorylation regulating Thr 389 phosphorylation and with its ability to directly affect S6K1 activity. DISCUSSION A number of signaling components, including PI3K (27,35,36), PKB (24,37,38), PDK1 (15,16), protein kinase C (39 -43), Rac-1 (44,45), mTOR/FRAP (2,33), and, more recently, NEK6/7 (46), have been implicated in the regulation of S6K1 activity and consequently cell growth. Of these, mTOR and PDK1 have been identified, biochemically, molecularly and genetically as essential regulators of S6K1 function (4,5,15,16,47,48). The role of PDK1 as the S6K1 Thr 229 kinase was initially based on in vitro kinase assays and in vivo ectopic co-expression studies and is consistent with recent findings showing that S6K1 activity is abolished in ES cells deficient in PDK1 (47). The intimate interplay between PI3K, PKB, and PDK1 has led to two widely held postulates concerning S6K1 activation: that PKB regulates Thr 389 phosphorylation through controlling mTOR activity (49) and that PDK1 phosphorylation of Thr 229 is PI3K-dependent (12,15). However, neither model appears to be correct. In the case of PKB, recent studies from this laboratory have demonstrated that activated alleles of PKB only induce S6K1 activation when these alleles are constitutively targeted to the membrane (22). Furthermore, an interfering mutant of PKB, which blocks insulin-induced reporter PKB activation, has no effect on S6K1 activation (22). Finally, the ability of a number of agonists to alter mTOR kinase activity toward Thr 389 remains controversial, despite the ability of these agents to induce potent Thr 389 phosphorylation (see Refs. 2, 48, and 49, and see below). In the case of Thr 229 phosphorylation, the argument against a direct PI3K Ͼ PDK1 Ͼ S6K1 connection is much clearer. First, PDK1 Thr 229 in vitro kinase activity is unaffected when PDK1 is derived from cells treated with insulin, wortmannin, or rapamycin (16). Indeed, the phosphorylation of Thr 229 in S6K1-E389D 3 E is constitutive and unaffected by inhibition of PI3K (Fig. 1B). Consistent with these findings, the ability of PDK1 to phosphorylate Thr 229 in vitro is unaffected by lipids (15). The assumption that PI3K directly controls PDK1-mediated Thr 229 phosphorylation is derived from the fact that this step is wortmannin-sensitive (35). However, the effect of wortmannin does not appear to be directly through inhibition of Thr 229 phosphorylation but rather due to inhibition of Thr 389 phosphorylation, which is a prerequisite for the constitutively active PDK1 to access and phosphorylate the Thr 229 site (16,18). Consistent with these findings, the Drosophila PDK1 input to S6K activation is also PI3K-independent (5). However, S6K activation in Drosophila KI67 cells is wortmannin-resistant, and in the fly it is PI3K-independent, although it is still dependent on TOR. It may be that in more highly developed eukaryotes, signaling pathways have evolved to a higher level of complexity. Indeed, in several mammalian cell types, S6K activation has been reported to be PI3K-independent (42,50,51). To resolve this point, it will be important to establish the molecular mechanism by which PI3K signals to S6K.
Consistent with the findings above, a series of recent studies have led to a potential molecular mechanism by which Thr 389 phosphorylation may regulate Thr 229 phosphorylation. Initially, in a yeast two-hybrid screen it was found that the kinase domain of PDK1 interacts with a hydrophobic motif in the protein kinase C-related kinase-2 (PRK2) (52). Although PRK2 is not a downstream effector of PI3K, the sequence, which interacts with PDK1, is homologous to a conserved motif found in the AGC family of serine kinases, including the sequence surrounding Thr 389 in S6K1 (11,53). This sequence was termed the PDK1-interacting fragment, and the pocket in the kinase domain of PDK1 to which it binds was termed the PDK1interacting fragment-binding pocket (54). Subsequently, it was demonstrated that PDK1 is able to bind directly with the analogous phosphorylated hydrophobic motif in p90RSK, enabling PDK1 to phosphorylate the T-loop of p90RSK (31). Likewise, recent studies suggest a similar mechanism by which PDK1 binds to S6K1 to mediate Thr 229 phosphorylation. In this case, a carboxyl-terminal truncated version of S6K1 was em- ployed as a substrate for PDK1, since full-length S6K1 is a poor substrate (54). The absence of this domain or conversion of the four phosphorylation sites within the autoinhibitory domain (Fig. 1A) to acidic residues raises basal levels of Thr 389 phosphorylation such that these S6K1 variants become substrates for PDK1 (15,16,18,54). Whereas the carboxyl terminal truncated S6K1 variant was activated and interacted with PDK1, wild type S6K1 failed to interact or to be activated by PDK1. Interaction and activation of the carboxyl-terminal truncated S6K1 variant was greatly facilitated by conversion of Thr 389 to an acidic residue (14,18). These findings are consistent with the order of phosphorylation, initially determined by phophopeptide mapping of different variants of S6K1 (14).
From previous studies, the phosphorylation of Thr 229 appeared to be the only mitogen input required for insulin-induced activation of S6K1-E389D 3 E (16,18). However, such changes in activity were not reflected by changes in Thr 229 phosphorylation (Fig. 1B), suggesting the existence of an additional wortmannin-sensitive input. This input appears to be Ser 371 phosphorylation mediated by mTOR, since this event parallels S6K1 activation by mitogens and inactivation by rapamycin or wortmannin. This site was previously thought to be rapamycin-resistant (13), a discrepancy that probably arose from the high basal levels of Ser 371 phosphorylation combined with the difficulties in quantitating phosphopeptide maps. Ser 371 is flanked by a proline in the ϩ1-position (13), reminiscent of the mTOR-directed phosphorylation sites in the autoinhibitory domain of S6K1 and in 4EBP-1 (10,28,34,55). Ser 371 resides in the linker domain of S6K1, the highly conserved element connecting the catalytic and autoinhibitory domains. Ser 371 homologous phosphorylation sites have been identified in PKB and the protein kinase Cs. In PKB, the Ser 371 homologous site, Thr 450 , appears to be constitutively phosphorylated (56) yet dispensable for activity (57). In contrast, in the protein kinase Cs it appears to be an autophosphorylation site, necessary for the maturation of kinase activity (13). In S6K1, this site appears to be critical for kinase activity and, unlike the protein kinase Cs, not regulated by autophosphorylation, since insulin-induced Ser 371 phosphorylation and Thr 389 phosphorylation are unaffected in a kinase-inactive S6K1 variant ( Figs. 2A and 6A). These studies imply that although this site is conserved in these kinases, its use has been differentially exploited during evolution.
Previous studies had shown that substitution of an alanine for Ser 371 inhibits Thr 389 phosphorylation and S6K1 activation (13). Furthermore, this effect was not rescued by substituting an acidic residue for Thr 389 , although such a substitution is known to raise kinase activity (11). These findings implied that Ser 371 phosphorylation had at least two distinct functions, to regulate Thr 389 phosphorylation and to directly participate in kinase activation (13). The fact that the ability of mTOR to phosphorylate Thr 389 in vitro in the S6K1-S371A variant is not impaired by the alanine substitution suggests that the role of Ser 371 phosphorylation in regulating Thr 389 phosphorylation is distinct from that of participating in catalyzing the phosphorylation reaction at this site. Potentially, Ser 371 phosphorylation may be required to localize the kinase such that Thr 389 can be phosphorylated by mTOR or possibly to stabilize Thr 389 phosphorylation, as has been proposed for Thr 229 phosphorylation (58). Although the in vivo mechanism by which Ser 371 phosphorylation regulates Thr 389 phosphorylation needs to be resolved, it is clear that increased levels of Ser 371 phosphorylation parallel increased kinase activity when an acidic residue is substituted at Thr 389 (Figs. 4 -6). In addition, higher levels of Thr 389 phosphorylation are required in vitro to achieve similar levels of S6K1 activity in the S6K1-S371A variant as compared with the wild-type kinase, further supporting a direct role for Ser 371 phosphorylation in augmenting kinase activity.