Critical Role of T-Loop and H-Motif Phosphorylation in the Regulation of S6 Kinase 1 by the Tuberous Sclerosis Complex*

The tuberous sclerosis gene products Tsc1 and Tsc2 behave as tumor suppressors by restricting cell growth, a function conserved among metazoans. Recent evi-dence has indicated that hyperactivation of S6 kinase 1 (S6K1) may represent an important biochemical change in the development of tuberous sclerosis-associated lesions. We show here that deletion of either Tsc1 or Tsc2 or expression of the Rheb (Ras homolog enriched in brain) GTPase leads to hyperphosphorylation of S6K1 at a subset of regulatory sites, particularly those of two essential residues functionally conserved among AGC superfamily serine/threonine kinases, i.e. the activation loop (T-loop; Thr-229) and the hydrophobic motif (H-motif; Thr-389). These sites are reciprocally and dose-de-pendently regulated when S6K1 is coexpressed with the Tsc1-Tsc2 complex. Mutations that render S6K1 mTOR (mammalian target of rapamycin)-resistant also protect S6K1 activity and phosphorylation from down-regula-tion by Tsc1/2. We demonstrate that two disease-associ-ated mutations in Tsc2 fail to negatively regulate S6K1 activity concomitant with a failure to modify T-loop and H-motif phosphorylation. Finally, we identify one path-ological Tsc2 mutation that retains its ability to negatively regulate S6K1, suggesting that, in some cases, tuberous sclerosis may develop independently of S6K1 hyperactivation. These results also highlight the

The tuberous sclerosis gene products Tsc1 and Tsc2 behave as tumor suppressors by restricting cell growth, a function conserved among metazoans. Recent evidence has indicated that hyperactivation of S6 kinase 1 (S6K1) may represent an important biochemical change in the development of tuberous sclerosis-associated lesions. We show here that deletion of either Tsc1 or Tsc2 or expression of the Rheb (Ras homolog enriched in brain) GTPase leads to hyperphosphorylation of S6K1 at a subset of regulatory sites, particularly those of two essential residues functionally conserved among AGC superfamily serine/threonine kinases, i.e. the activation loop (T-loop; Thr-229) and the hydrophobic motif (Hmotif; Thr-389). These sites are reciprocally and dose-dependently regulated when S6K1 is coexpressed with the Tsc1-Tsc2 complex. Mutations that render S6K1 mTOR (mammalian target of rapamycin)-resistant also protect S6K1 activity and phosphorylation from down-regulation by Tsc1/2. We demonstrate that two disease-associated mutations in Tsc2 fail to negatively regulate S6K1 activity concomitant with a failure to modify T-loop and H-motif phosphorylation. Finally, we identify one pathological Tsc2 mutation that retains its ability to negatively regulate S6K1, suggesting that, in some cases, tuberous sclerosis may develop independently of S6K1 hyperactivation. These results also highlight the importance of dual control of T-loop and H-motif phosphorylation of S6K1 by the Tsc1-Tsc2 complex.
Tuberous sclerosis is a hyperproliferative disorder resulting in the appearance of benign tumors in multiple organ systems including brain, skin, lungs, heart, eyes, kidneys, pancreas, and the skeleton (1,2). Initially, linkage analysis of affected families suggested that two distinct loci participated in the manifestation of the disease (3). Meanwhile, TSC1 and TSC2 were identified by positional cloning as the genes whose mutations cause tuberous sclerosis (4,5). In metazoans, Tsc1 and Tsc2 form a signaling complex that performs a cell growth suppressive function. Studies in Drosophila have demonstrated that loss-of-function mutations in either dTsc1 or dTsc2 increase cell size and do so cell autonomously (6,7). Conversely, combined overexpression of dTsc1 and dTsc2 reduces cell size, whereas expression of either product individually is without effect (6), which is consistent with a model in which the Tsc1-Tsc2 heterodimer is the functional configuration. Homozygous disruption of either Tsc1 or Tsc2 in mice is embryonically lethal, whereas mice heterozygous for either allele display increased organ size and a propensity for tumor development (8,9).
Recently, it has been demonstrated that a deficiency in either of the tuberous sclerosis gene products leads to hyperactivation of S6K1. 1 In the absence of an S6K1 crystal structure, the model of S6K1 activation is based on extensive mutational and structure-function studies and by inferences from the structure and analysis of other AGC kinases, including Akt (10,11). It is postulated that the unphosphorylated autoinhibitory domain (AID) folds upon and occludes the amino-terminally positioned kinase domain, perhaps performing the following two functions: 1) limiting access of the substrate to the catalytic site; and 2) burying additional activating phosphorylation sites within the protein's interior (see Fig. 1A). Activation is achieved through a coordinated and sequential series of phosphorylations beginning with phosphorylation of the carboxylterminal AID. Up to six serine/threonine and proline sites, designated (S/T)P, are localized to the AID and undergo phosphorylation in response to activating stimuli. Substitution of four of these sites with alanine compromises the serum-induced activation of S6K1 (12), whereas exchange of these sites with phosphomimic acidic residues only slightly increases the basal activity (12,13). It is therefore plausible that phosphorylation of the AID is necessary but not sufficient for S6K1 activation. Subsequently, two sites conserved among kinases of the AGC superfamily of serine/threonine kinases are phosphorylated, namely the activation loop (T-loop) site at position Thr-229 and the hydrophobic motif (H-motif) site at position Thr-389. Mutation of Thr-229 to either alanine or glutamate abolishes kinase activity (13)(14)(15). However, whereas substitution of Thr-389 with alanine abolishes kinase activity (13,16,17), glutamate substitution of the H-motif increases basal kinase activity (13,16,17). Additional phosphorylation sites have been mapped to Thr-367, Ser-371, and Ser-404 (13,18). However, it is less clear just how the phosphorylation of these sites participates in the collective regulation of S6K1.
For S6K1, T-loop and H-motif phosphorylation is the net result of the integration of two major input pathways, nutrient sufficiency and growth factor adequacy. The nutrient sufficiency pathway senses the availability of glucose and amino acids as well as mitochondrial function and requires a complex comprised of mTOR, the regulatory associated protein of mTOR (Raptor), and G␤L (19,20). Raptor appears to physically recognize the TOS (target of rapamycin signaling) motif in S6K1 and another mTOR substrate, the eIF4E-binding protein (4EBP), and may function to present such substrates to the mTOR kinase (21). mTOR has been shown to phosphorylate S6K1 in vitro both at the H-motif site, Thr-389, and at the AID sites, Ser-411 and Thr-421/Ser-424, although Thr-389 appears to be the major site of phosphorylation (22). Small interfering RNA depletion of mTOR, Raptor, or G␤L compromises nutrient stimulation of S6K1 (19), demonstrating a requirement for each component of the complex in productive signaling. It is unclear precisely how nutrient availability is sensed by the mTOR-Raptor-G␤L complex. Regulation of S6K1 by the growth factor adequacy pathway involves activation of phosphatidylinositol 3-kinase (PI3K) and, thus, the generation of phosphatidylinositol-3,4,5-trisphosphate (PIP 3 ) in cellular membranes. S6K1 is activated by ectopic expression of PI3K (14) and is inhibited by the PI3K inhibitor wortmannin (12,14), indicating that PI3K is both necessary and sufficient for S6K1 activity.
In response to growth factors, PI3K phosphorylates phosphoinositides at the D3 position of the inositol ring, producing PIP 3 . The PH (pleckstrin homology) domains of phosphoinositide-dependent kinase 1 (PDK1) and Akt bind PIP 3 , resulting in the membrane localization and activation of these proteins (reviewed in Ref. 23). In mitogen-stimulated cells, S6K1 is phosphorylated at the T-loop site, Thr-229, by PDK1 (24,25). A phosphopeptide modeled after the phosphorylated H-motif of S6K1 binds much more efficiently to PDK1 than the unphosphorylated peptide (26), potentially explaining the strongly synergistic nature of H-motif and T-loop phosphorylation of S6K1. Moreover, when coexpressed in cells, PDK1 binds much more efficiently to an S6K1 variant bearing a phosphomimic H-motif substitution (27), suggesting that the phosphorylated H-motif provides a surface with which PDK1 makes physical contact. Akt is also phosphorylated and activated by PDK1. In response to growth factors, Akt phosphorylates Tsc2 on as many as five sites (28,29), thereby inhibiting the intrinsic GTP exchange activity of Tsc2 for the small GTP-binding protein Rheb (30 -32). Thus, in response to growth factors, Rheb accumulates in the GTP-liganded and, thus, active configuration. Although Rheb activation is sufficient to activate S6K1 even in the absence of amino acids (33), whether or not Rheb signals to S6K1 indirectly through mTOR remains a debated issue. Given the requirement of T-loop and H-motif phosphorylation for S6K1 activation and given the finding that Tsc1 or Tsc2 lossof-function results in hyperactivation of S6K1, we set out to determine what role individual phosphorylations play in the collective control of S6K1 by the Tsc1-Tsc2 complex.
Plasmid Constructs-Amino-terminally HA-tagged, wild type rat S6K1, E389D3E-S6K1, D4E-S6K1, ⌬CT-S6K1, and ⌬NT⌬CT-S6K1 cloned into pRK7 were generously provided by John Blenis (Harvard University) and Jim Jefferson (Pennsylvania State University College of Medicine). Human Tsc1 and Tsc2 cloned into pEFP2 were provided by Jeff DeClue (NCI, National Institutes of Health). Amino-terminally FLAG-tagged Tsc1 was constructed by site directed mutagenesis (QuikChange site-directed mutagenesis kit, Stratagene) using pEFP2-Tsc1 as a template. Sense and antisense primers were designed to incorporate the FLAG epitope as a loop structure, flanked on either side by sequence that annealed to the template. FLAG-tagged Tsc2 was constructed similarly using pEFP2-Tsc2 as a template. Amino-terminally Myc-tagged human Rheb was supplied by Paul Worley (Johns Hopkins University).
Cell Culture and Transient Transfection-Mouse embryo fibroblasts (MEFs) from Tsc1-and Tsc2-null embryos and wild type MEFs were generously provided by David Kwiatkowski (Harvard University). MEFs were cultured in Dulbecco's modified Eagle's medium supplemented with fetal calf serum (10%, v/v) and penicillin, streptomycin, and ciprofloxacin. For experiments requiring insulin stimulation, MEFs were cultured in Dulbecco's modified Eagle's medium containing low serum (0.5% fetal calf serum, v/v) overnight and stimulated with insulin as indicated the following morning.
For immunoprecipitation, lysates normalized for cell protein were incubated with 4 -8 g of anti-S6K1 antibody or with 2 l of 12CA5 ascites and mixed end-over-end for 1-2 h at 4°C. Protein A-agarose beads were added for an additional hour, and immune complexes were isolated by centrifugation. Immunoprecipitates were washed twice with lysis buffer and heated for 5 min to 100°C in 1ϫ sample buffer.
Assay of S6K1 Kinase Activity-The assay of S6K1 activity has been described elsewhere (34). In brief, immunopurified endogenous or exogenous S6K1 was incubated with a synthetic peptide (AKRRRLSS-LRA), and 32 P incorporation into the peptide substrate was measured by liquid scintillation counting.

RESULTS
In Drosophila, loss of Tsc1 or Tsc2 leads to increased activity of dS6K and, as a result, a significant increase in the size of affected cells. To establish whether or not Tsc1 and Tsc2 regulation of S6K1 is conserved in mammalian cells, we assayed the kinase activity of S6K1 in vitro in MEFs isolated from Tsc1or Tsc2-null mice or their wild type counterparts. Whereas treatment of wild type MEFs with insulin induced a marked increase in S6K1 activity, the basal activity of S6K1 isolated from Tsc1-and Tsc2-null MEFs was high and was not increased further by insulin (Fig. 1B). In fact, basal S6K1 activity in the knock-out MEFs was equivalent to the insulin-stimulated activity measured in wild type MEFs, suggesting that deletion of either Tsc1 or Tsc2 renders S6K1 constitutively activated. Nevertheless, the high basal activity of S6K1 in Tsc1-and Tsc2-null MEFs remained sensitive to rapamycin (40 nM) and to LY294002 (20 M), which inhibits mTOR and PI3K equally efficiently at this concentration (35). In whole cell extracts from Tsc1-and Tsc2-null MEFs, basal phosphorylation of ribosomal protein S6 (rpS6), an endogenous S6K1 substrate, was equivalent to the insulin-stimulated levels observed in wild type MEFs (Fig. 1C, cf. lane 4 versus lanes 7 and 13). These data are consistent with genetic epistasis analyses conducted in Drosophila, indicating that mTOR lies downstream of the Tsc1-Tsc2 complex in the regulation of S6K1 (6, 7).
Whereas overexpression of Tsc1-Tsc2 reduces H-motif phosphorylation of coexpressed S6K1 (29), Tsc2-null cells display augmented H-motif phosphorylation of S6K1 relative to wild type cells (36). It remains to be resolved, however, whether T-loop phosphorylation (Thr-229) or phosphorylation of other regulatory sites in the AID is influenced by the loss of Tsc1 or Tsc2. To address this question, S6K1 was immunoprecipitated from Tsc1-or Tsc2-null MEFs as well as from wild type MEFs under different treatment conditions. The phosphorylation status of S6K1 was assessed using antiphosphopeptide antibodies specific for the phosphorylation sites Thr-229, Ser-371, Thr-389, Ser-411, and Thr-421/Ser-424 (see Fig. 1A). In all cell types, the pattern of basal and insulin-stimulated S6K1 activity (Fig. 1B) tightly correlated with H-motif phosphorylation at Thr-389 (Fig. 2). Importantly, whereas phosphorylation of Thr-389 in Tsc1-and Tsc2-null MEFs was inhibited by rapamycin, phosphorylation of this site was strongly resistant to inhibition by wortmannin. In contrast, S6K1 prepared from wild type MEFs displayed a reduction in H-motif phosphorylation when treated with either rapamycin or wortmannin. The wortmannin resistance of H-motif phosphorylation in MEFs devoid of Tsc1 or Tsc2 suggests that PI3K regulates S6K1 upstream of Tsc1-Tsc2 and that the effect of the loss of Tsc1-Tsc2 is dominant in the stimulation of S6K1 by PI3K. T-loop phosphorylation of S6K1 at Thr-229 under these conditions mirrors phosphorylation of the H-motif in that this phosphorylation was largely rapamycin-sensitive and wortmannin-resistant in Tsc1-and Tsc2-null cells. A similar correlation was observed for phosphorylation of the AID sites Ser-411 and Thr-421/Ser-424. Unlike the regulation of the sites described above, the phosphorylation of Ser-371 was neither robustly stimulated by insulin nor inhibited by rapamycin or wortmannin. Thus, phosphorylation of Ser-371 is unregulated under these conditions and appears to some extent constitutive. Analysis of the phosphorylation status of rpS6 (Fig. 2B) indicates that the cumulative effect of this multisite regulation was constitutive activation of S6K1 when Tsc1 or Tsc2 is functionally lost.
Clearly, deletion of the genes encoding Tsc1 or Tsc2 induces maximal phosphorylation of S6K1 at several regulatory sites, including the T-loop and H-motif sites. Consequently, in Tsc1and Tsc2-null cells S6K1 is constitutively activated and refrac-tory to conditions that would otherwise be inactivating, e.g. serum-deprivation. Overexpression of the Tsc1-Tsc2 complex in cells has the converse effect, i.e. S6K1 inhibition (28,29,36) (see below). We therefore sought to address whether or not this inhibition is associated with dephosphorylation of the same subset of Tsc1-Tsc2-regulated sites defined by the aforementioned studies. HEK293T cells exhibit high basal S6K1 activity (data not shown) and were therefore chosen for the subsequent experiments. Cells were transfected with increasing amounts of vectors expressing FLAG-tagged Tsc1 and FLAG-tagged Tsc2 together with HA-tagged S6K1, and S6K1 kinase activity was assayed in vitro at two different ectopic S6K1 gene dosages. Expression of Tsc1-Tsc2 led to a dosage-dependent reduction in cotransfected S6K1 activity (Fig. 3A). This effect was irrespective of the level of ectopically expressed S6K1 used in these experiments. Overexpression of Tsc1-Tsc2 reduced Hmotif phosphorylation (Fig. 3B), paralleling the decrease in S6K1 kinase activity. In contrast, little change in the phosphorylation of the AID sites, Thr-421/Ser-424, was observed upon Tsc1-Tsc2 expression. This was somewhat unexpected, given the robust stimulation of Thr-421/Ser-424 phosphorylation observed in Tsc1-and Tsc2-null MEFs ( Fig. 2A). Additionally, the phosphorylation of S6K1 at Ser-371 also was not significantly altered upon Tsc1-Tsc2 overexpression, which is consistent with this site remaining relatively unregulated by the Tsc1-2 complex. These data suggest that overexpression of Tsc1-Tsc2 inhibits S6K1 primarily through regulation of the H-motif and, potentially, T-loop phosphorylation, with a minimal effect on AID phosphorylation. We reasoned that the S6K1 mutations that artificially preserve or mimic H-motif phosphorylation should render the activity of the corresponding variant resistant to inhibition in cells overexpressing of Tsc1-Tsc2. Unfortunately, any modification of the T-loop site in S6K1, whether it be alanine or glutamate substitution, abolishes kinase activity (13,14,16). Therefore, the necessity and sufficiency of regulation of the T-loop site could not be directly tested in such an assay. A series of HA-tagged truncation and phosphomimic S6K1 mutants were expressed in the presence or absence of FLAG-tagged Tsc1 and FLAG-tagged Tsc2, and the kinase activity of each mutant was assayed in vitro. Because the specific activities of each S6K1 construct differ as a result of each unique mutation, the activity measurements for each construct assayed in the absence of Tsc1-Tsc2 was assigned a value of 100% to allow comparison between constructs (Fig.  4A). The unadjusted kinase activity measurements are presented for reference in Fig. 4B, as is the relative expression of these constructs (Fig. 4C). Mutation of the AID, either by truncation (⌬CT) or by the acidic, phosphomimic substitution of five serum-stimulated (S/T)P sites (D4E), failed to protect S6K1 from inhibition upon Tsc1-Tsc2 overexpression (Fig. 4A). Deletion of the amino terminus (⌬NT) of S6K1 removes the TOS motif, a small recognition/docking site found in mTOR substrates (37) that renders ⌬NT⌬CT rapamycin-resistant (18,38,39). Importantly, the activity of ⌬NT⌬CT was resistant to inhibition in cells in which the Tsc1-Tsc2 complex was expressed ectopically. Finally, acidic, phosphomimic substitution of the H-motif and the AID sites (E389D3E) was also sufficient to render S6K1 resistant to inhibition by Tsc1-Tsc2. This construct has also been shown to retain 50% of its kinase activity in rapamycin-treated cells (13,18) indicating that it, like ⌬NT⌬CT, is regulated, in part, independently of mTOR.
As shown in Figs. 2 and 3, the regulation of wild type S6K1 by Tsc1-Tsc2 is achieved through control of H-motif and T-loop phosphorylation. Furthermore, phosphomimic substitution of the H-motif is sufficient to protect S6K1 from inhibition upon the overexpression of Tsc1-Tsc2. Therefore, it is reasonable to FIG. 2. A subset of S6K1 phosphorylation sites are rapamycinsensitive, but wortmannin-resistant in the absence of Tsc1 or Tsc2. A, cells were cultured in low serum-containing medium overnight and then in the presence or absence of 100 nM wortmannin or 40 nM rapamycin for 1 h. Cells were then stimulated or not with 200 nM insulin for 20 min. Extracts were prepared, and S6K1 was immunoprecipitated. The immune complexes were separated by gel electrophoresis and immunoblotted with anti-S6K1 antibodies or phospho-specific antiserum recognizing the following S6K1 phosphorylation sites: Thr(P)-220 (pT229), Ser(P)-371 (pS371), Thr(P)-389 (pT389), Ser(P)-411 (pS411), and Thr(P)-421/Ser(P)-424 (T421/S424). B, cells were treated and extracts prepared as described for panel A. Extracts were normalized for protein and separated by gel electrophoresis. Proteins were immunoblotted with antibodies raised against S6K1 and rpS6 phosphorylated on Ser-240 and Ser-244. The data are representative of three independent experiments. predict that E389D3E and ⌬NT⌬CT, which are resistant to inhibition by Tsc1-Tsc2, also exhibit H-motif and T-loop phosphorylation that is protected from inhibition by Tsc1-Tsc2 expression. To test this hypothesis, we compared H-motif and T-loop phosphorylation among a panel of S6K1 constructs expressed with or without Tsc1-Tsc2 in HEK293T cells. S6K1 was then immunoprecipitated with anti-HA antibody and immunoblotted using phospho-specific antisera. Whereas the phosphorylation of Thr-229 and Thr-389 in wild type and ⌬CT S6K1 was reduced in cells coexpressing Tsc1-Tsc2, neither T-loop nor H-motif phosphorylation was affected by overexpressed Tsc1-Tsc2 in the E389D3E or ⌬NT⌬CT (Fig. 5A) (note that the anti-phospho-Thr-389 antibody used in these studies recognizes the S6K1 mutant with a glutamate substitution at position 389). These constructs were expressed at similar levels as shown by analysis of whole cell lysates (Fig. 5B). Collectively, these data argue that the inhibition of S6K1 observed in cells FIG. 3. Ectopic expression of the Tsc1-Tsc2 complex dose-dependently inhibits S6K1 kinase activity and H-motif phosphorylation. A, HEK293T cells were transiently transfected with either 10 or 20 ng of HA-S6K1 with or without the indicated amounts of FLAG-Tsc1 and FLAG-Tsc2. Where applicable, equal quantities of Tsc1 and Tsc2 were present in each transfection. Cells were cultured in medium containing 10% serum and harvested 60 h post-transfection. HA-S6K1 was immunoprecipitated with anti-HA antibodies, and S6K1 kinase activity was determined as described under "Experimental Procedures." The data shown represent the average of duplicate samples for each condition. These results are representative of three independent experiments. B, extracts were prepared for each condition as described for panel A. Extracts were normalized for protein and separated by gel electrophoresis. Proteins were immunoblotted with anti-FLAG and anti-HA antibodies as well as with phospho-specific antisera specific for S6K1 phosphorylated on Ser-371 (pS371), Thr-389 (pT389), and Thr-421/Ser-424 (pT421/S424). The data are representative of three independent experiments.
FIG. 4. mTOR-resistant S6K1 constructs are also resistant to inhibition by Tsc1-Tsc2. A, HEK293T cells were transfected with 10 -100 ng of HA-S6K1 so as to obtain equivalent expression among constructs. Cells were cotransfected with 290 ng of FLAG-Tsc1/FLAG-Tsc2. Cells were cultured and extracts prepared as described in the Fig.  3 legend. S6K1 kinase activity was determined in anti-HA immunoprecipitates. The data shown represent the average of duplicate samples for each condition. These results are representative of three independent experiments. For purposes of comparison, the activity values obtained from cells that were not transfected with FLAG-Tsc1/FLAG-Tsc2 were assigned values of 100% for each construct. B, the unadjusted S6K1 kinase activity measurements are presented. C, cell extracts were separated by gel electrophoresis and immunoblotted with anti-HA and anti-FLAG antibodies.
Presumably, disease-associated mutations in Tsc1 and Tsc2 result in the loss of some essential function of the complex. If S6K1 hyperactivation is important for the progression of tuberous sclerosis, those mutations causative for the development of tuberous sclerosis should also show defective regulation of S6K1. To test this possibility, HA-tagged S6K1 was coexpressed with FLAG-tagged Tsc1 alone or in combination with a panel of naturally occurring, disease-associated FLAGtagged Tsc2 mutants in HEK293T cells. S6K1 was immunopre-cipitated, and its kinase activity and the extent of T-loop and H-motif phosphorylation were determined. In this assay, little change in either the kinase activity (Fig. 6A) or in the phosphorylation of the T-loop or the H-motif (Fig. 6C) was observed when Tsc1 or Tsc2 was expressed alone. However, coexpression of Tsc1 and Tsc2 was sufficient to reduce both S6K1 activity as well as T-loop and H-motif phosphorylation, supporting the idea that Tsc1 and Tsc2 are each critical components of the signaling complex. Of the disease-associated Tsc2 substitutions tested in this assay, F615S and Y1571H failed to regulate S6K1 in regard to both kinase activity and T-loop and H-motif phosphorylation. However, N525S Tsc2 was competent to negatively regulate S6K1, as evidenced both by the reduction in coexpressed S6K1 kinase activity and phosphorylation. These differences were not due to variations in the expression level of the Tsc2 constructs (Fig. 6C).
Coexpression of the Rheb GTPase is sufficient to induce S6K1 activation concomitant with H-motif phosphorylation (32). We therefore coexpressed HA-tagged S6K1 with or without Myc-tagged Rheb in HEK293 cells, which, unlike HEK293T cells, have low basal S6K1 activity when serum starved, and we immunoprecipitated S6K1 with anti-HA antibodies. As shown in Fig. 6D, the expression of Rheb induced phosphorylation of both the T-loop and the H-motif. The expression levels of Rheb and S6K1 in whole cell lysates are also shown (Fig. 6E).

DISCUSSION
In this study we have demonstrated that the T-loop and H-motif are primary sites of phosphorylation-dependent regulation of S6K1 by Tsc1-Tsc2. Although Ser-371 is necessary for kinase activation (40), in MEFs the phosphorylation of this site is unaffected by the presence of insulin or by the inhibitors wortmannin or rapamycin ( Fig. 2A). Furthermore, Ser-371 phosphorylation was neither significantly regulated by the loss of Tsc1 or Tsc2 ( Fig. 2A) nor by overexpression of the Tsc1-Tsc2 complex (Fig. 3B). Loss of either Tsc1 or Tsc2 resulted in increased basal AID phosphorylation at Ser-411, whereas the phosphorylation of the adjacent AID sites Thr-421/Ser-424 was similar to wild type levels ( Fig. 2A). Collectively, these sites were sensitive to wortmannin and rapamycin in wild type MEFs, whereas the phosphorylation of these sites were wortmannin-insensitive in Tsc1-or Tsc2-null MEFs. Similar to the phosphorylation of Ser-371, Thr-421/Ser-424 was essentially unregulated in HEK293T cells overexpressing Tsc1-Tsc2 (Fig.  3B). Therefore, the reduction in S6K1 activity associated with overexpression of Tsc1-Tsc2 is less dependent on the regulation of Ser-371 and Thr-421/Ser-424 phosphorylation and, instead, tightly correlates with the regulation of the T-loop and H-motif sites.
How then is the regulation of S6K1 by Tsc1-Tsc2 achieved? In Tsc1-and Tsc2-null MEFs, the T-loop site (Thr-229) and the H-motif site (Thr-389) are constitutively phosphorylated to the levels observed in wild type MEFs stimulated with insulin ( Fig.  2A). Similarly, overexpression of Rheb induces both T-loop and H-motif phosphorylation (Fig. 6D). Conversely, the phosphorylation of these sites is reduced in cells overexpressing the Tsc1-Tsc2 complex (Figs. 3B, 5A, and 6C). Biondi and colleagues (27) have demonstrated that, in vitro, a carboxyl-terminally truncated S6K1 is a much better PDK1 substrate when the S6K1 H-motif is changed to a phosphomimic glutamate. Furthermore, PDK1 interacts with this Thr3 Glu mutant more efficiently. Finally, the PDK1-S6K1 interaction is competed by an H-motif surrogate phosphopeptide, suggesting that PDK1 interacts with the phosphorylated H-motif of S6K1 and that this event increases the efficiency with which PDK1 phosphorylates the T-loop of S6K1. Given that the kinase activity of PDK1 is not increased by growth factors or stimuli that induce PI3K activation (24,41), and given that PDK1 and S6K1 are constitutively associated when expressed in cells (42), it is plausible that loss of Tsc1 or Tsc2, through activation of Rheb, induces H-motif phosphorylation. The phosphorylated H-motif is then recognized by PDK1, which facilitates T-loop phosphorylation of S6K1. Given the importance of T-loop and H-motif phosphorylation in modulating the catalytic activity of AGC kinases, the coordinated regulation of these motifs in S6K1 explains the reciprocal regulation of S6K1 catalytic activity in cells that lack or overexpress Tsc1/2. Furthermore, the sensitivity of both S6K1 activity and T-loop and H-motif phosphorylation to rapamycin indicates that mTOR signaling activity is required for hyperactivation of S6K1 in Tsc1-and Tsc2-deficient MEFs. Conversely, the wortmannin-resistance of S6K1 activity and phosphorylation in these cells demonstrates that the Tsc1/Tsc2 deficiency induces S6K1 activation independently of PI3K.
Independent genetic screens of tuberous sclerosis patients have identified dozens of disease-associated lesions spanning virtually the entire length of the Tsc2 protein (43)(44)(45)(46)(47)(48). Several of these mutations, e.g. F615S and Y1571H, abrogate not only the interaction with Tsc1 but also the tyrosine phosphorylation of Tsc2 (49,50), potentially explaining the loss of function of the respective Tsc2 variant. Although Tsc2 is tyrosine-phosphorylated in response mitogenic stimuli (50), the function of this phosphorylation is unclear. However, the Tsc2 mutation N525S affects neither Tsc1 interaction nor tyrosine phosphorylation of Tsc2 (49), suggesting that the manifestation of tuberous sclerosis may involve biochemical alterations other than Tsc1-Tsc2 interaction and Tsc2 tyrosine phosphorylation. As shown in Fig. 6, A-C, there is a strong correlation between S6K1 regulation and the reported biochemical properties described above, i.e. interaction with Tsc1 and Tsc2 tyrosine phosphorylation. It would therefore appear that regulation of S6K1 involves productive assembly of the Tsc1-Tsc2 complex and involves Tsc2 tyrosine phosphorylation. The observation that N525S negatively regulates S6K1 as well as wild type Tsc2 raises the intriguing possibility that the development of tuberous sclerosis may, in some cases, occur independently of S6K1 hyperactivation. It would therefore be interesting to determine whether or not tuberous sclerosis-associated lesions harboring FIG. 6. Two disease-associated mutations fail to regulate S6K1 kinase activity or T-loop, and H-motif phosphorylation, whereas Rheb stimulates T-loop and H-motif phosphorylation. A, HEK293T cells were transfected with 20 ng of HA-S6K1 with or without 370 ng of FLAG-Tsc1/FLAG-Tsc2 (wild type or mutant). HA-S6K1 activity was determined as described in the Fig. 4 legend. The data shown represent the average of duplicate samples for each condition. These results are representative of three independent experiments. B, extracts were normalized for protein and separated by gel electrophoresis. Proteins were detected by immunoblotting with anti-HA and anti-FLAG antibodies. C, S6K1 was immunoprecipitated with anti-HA antiserum and immunoblotted with anti-HA, anti-Thr(P)229 (pT229), and anti-Thr(P)-389 (pT389). D, HEK293 cells were transfected with 20 ng of HA-S6K1 with or without 370 ng of Myc-Rheb. Extracts were prepared 36 h post-transfection from cells cultured in serum-free medium overnight, and S6K1 was immunoprecipitated with anti-HA antiserum and immunoblotted with anti-HA, anti-Thr(P)-229 (pT229), and anti-Thr(P)-389 (pT389). Results are representative of three independent experiments. E, extracts were normalized for protein and separated by gel electrophoresis. Proteins were detected by immunoblotting with anti-HA and anti-Myc antibodies. the N525S mutation in Tsc2 exhibit normal S6K1 activation.
Inhibition of the mTOR signaling pathway is rapidly emerging as a promising strategy for the treatment of solid tumors, particularly those resulting from constitutive activation of the PI3K/Akt pathway, such as PTEN-inactivating mutations (51)(52)(53). Albeit largely benign, tuberous sclerosis-associated tumors can influence almost every major organ system and are thus responsible for the wide spectrum of physiological complications in affected individuals. The observations that mTOR signaling is up-regulated in these lesions suggest that rapamycin or its second generation derivatives may prove effective in the management of tuberous sclerosis. Although highly specific inhibitors of S6K1 are not currently available, pharmacological inactivation of S6K1 may also have beneficial effects in the treatment of tuberous sclerosis. Our results, however, indicate that not all lesions resulting from Tsc2 mutation may be equally responsive to such a therapeutic approach.