Inactivation of the Tuberous Sclerosis Complex-1 and -2 Gene Products Occurs by Phosphoinositide 3-Kinase/Akt-dependent and -independent Phosphorylation of Tuberin*

The tuberous sclerosis complex (TSC) is a genetic disorder that is caused through mutations in either one of the two tumor suppressor genes, TSC1 and TSC2, that encode hamartin and tuberin, respectively. Interaction of hamartin with tuberin forms a heterodimer that inhibits signaling by the mammalian target of rapamycin to its downstream targets: eukaryotic initiation factor 4E-binding protein 1 (4E-BP1) and ribosomal protein S6 kinase 1 (S6K1). During mitogenic sufficiency, the phosphoinositide 3-kinase (PI3K)/Akt pathway phosphorylates tuberin on Ser-939 and Thr-1462 that inhibits the tumor suppressor function of the TSC complex. Here we show that tuberin-hamartin heterodimers block protein kinase C (PKC)/MAPK- and phosphatidic acid-mediated signaling toward mammalian target of rapamycin-dependent targets. We also show that two TSC2 mutants derived from TSC patients are defective in repressing phorbol 12-myristate 13-acetate-induced 4E-BP1 phosphorylation. PKC/MAPK signaling leads to phosphorylation of tuberin at sites that overlap with and are distinct from Akt phosphorylation sites. Phosphorylation of tuberin by phorbol 12-myristate 13-acetate was reduced by treatment of cells with either bisindolylmaleimide I or UO126, inhibitors of PKC and MAPK/MEK (MAPK/ERK kinase), respectively, but not by wortmannin (an inhibitor of PI3K). This work reveals that both PI3K-independent and -dependent mechanisms modulate tuberin phosphorylation in vivo.

The tuberous sclerosis complex (TSC) 1 is a human genetic disorder that leads to the development of benign hamartomatous tumors in the brain, kidneys, heart, eyes, lungs, and skin. The disorder affects about 1 million individuals worldwide, with an estimated prevalence of one in 6,000 newborns. Common symptoms include seizures, mental retardation, behavior problems, and skin abnormalities (for review see Ref. 1). Genetic studies have mapped this disorder to the TSC1 and TSC2 genes that encode the protein products hamartin (ϳ130 kDa) and tuberin (ϳ200 kDa), respectively.
Hamartin and tuberin negatively regulate cell growth and proliferation as a tuberin-hamartin heterodimer (2,3). Drosophila and mouse genetics show that mutations in TSC1 or TSC2/gigas result in increased cell size (an increase in cell mass/cell growth) and cellular proliferation (an increase in cell number) (4 -6). Genetic epistasis analysis in Drosophila places dTSC1/dTSC2 downstream of dPI3K and dAkt (also referred to as protein kinase B) but upstream of Drosophila ribosomal protein S6 kinase (dS6K) (6 -7). The loss of function or overexpression of several Drosophila genes recently linked the dPI3K and dTOR signaling pathways to the regulation of cell proliferation and cell growth (8), important cellular processes that are inappropriately regulated during tumor formation. These include dPI3K, dAkt, dS6K, and d4E-BP (effectors of dPI3K), and dTOR. Collectively, this work suggested that dTOR and dTSC1/dTSC2 were connected via the PI3K signaling pathway.
Recently, a surge of research from many labs revealed that TSC1/TSC2 functioned to inhibit nutrient-and PI3K-dependent signaling through mTOR (also called FRAP or RAFT), resulting in the inhibition of its downstream targets S6K1 and 4E-BP1 (9 -12). Although the majority of work supports this notion, the TSC1/TSC2 may also function in some circumstances to inhibit PI3K signaling to S6K1 independently of mTOR (see Ref. 13 and for review see Ref. 14). It is clear, however, through the use of different experimental approaches, that PI3K-dependent signaling serves to inhibit the tuberin-hamartin heterodimer through direct Akt-dependent phosphorylation of tuberin at Ser-939 and Thr-1462 (the major Akt phosphorylation sites) (9,15). Together, these observations suggest that the tuberin-hamartin heterodimer inhibits mTOR and that Akt can phosphorylate tuberin, thereby inactivating it. mTOR is regulated by mitogens, nutrients, and energy availability and is a critical upstream regulator of S6K1 and 4E-BP1 (for reviews see Ref. 16). The mTOR inhibitor, rapamycin, induces rapid S6K1 and 4E-BP1 dephosphorylation. Regulation of S6K1 and 4E-BP1 phosphorylation by stimulation of two distinct signaling pathways, PI3K and MAPK, require an mTOR input, as rapamycin can potently inhibit the phosphorylation of these downstream effectors by both PI3K-and MAPK-regulated pathways (16 -18). Hypophosphorylated 4E-BP1 binds to the cap-binding protein eIF4E to inhibit cap-dependent translation (for review see Ref. 19), which is reversed when 4E-BP1 is phosphorylated. The dissociation of hyperphosphorylated 4E-BP1 from eIF4E leads to the binding of eIF4G to eIF4E and the initiation of protein synthesis. It has been shown that both S6K1 and eIF4E/4E-BP1 effect mTORdependent control of mammalian cell size (20). The research summarized above, combined with the Drosophila genetic studies (8), strongly imply that aberrantly high mTOR-dependent signaling to both S6K1 and eIF4E/4E-BP1 contributes to the expansion and size of TSC tumors.
It has been known for some time that phosphorylation of S6K1 and 4E-BP1 are regulated by PI3K and/or by PKC-dependent (phorbol ester-activated) pathways (21)(22)(23)(24). We use phorbol 12-myristate 13-acetate (PMA), a phorbol ester and a tumor promoter, to induce S6K1 and 4E-BP1 phosphorylation via activation of the conventional and novel PKCs. In this report we show that the TSC1/TSC2 inhibits PKC/MAPK signaling toward 4E-BP1 and S6K1. Because PMA does not activate PI3K-dependent signaling in the HEK293E cells used in this study, these data support the model that TSC1/TSC2 inhibits mTOR rather than PI3K-mediated signaling. We report that tuberin is phosphorylated on the Akt-dependent sites (Ser-939 and Thr-1462) and on distinct phosphorylation sites upon PMA treatment, which occurs in the absence of PI3K signaling and is inhibited by the protein kinase inhibitors toward MEK (U0126) and PKC (bisindolylmaleimide I). This study reveals that the PKC/MAPK signaling input to tuberin is distinct from that mediated by PI3K/Akt. Furthermore, our work addresses an important question as to how S6K1 is activated upon PI3K-independent signaling. Our findings provide evidence that cell signaling through converging PI3K and PKC/ MAPK pathways inactivate the tuberin-hamartin heterodimer as a result of tuberin phosphorylation at overlapping sites.

MATERIALS AND METHODS
Chemicals and Materials-Wortmannin and U0126 were purchased from Biomol (Plymouth Meeting, PA), and bisindolylmaleimide I was from Calbiochem. PMA, insulin, and anisomycin were bought from Sigma, and epidermal growth factor (EGF) was from Invitrogen. PA was obtained from Avanti Polar Lipids, Inc. (Alabaster, AL). Radiolabeled reagents were purchased from PerkinElmer Life Sciences, and m 7 GTP-Sepharose was from Amersham Biosciences. All other reagents (unless stated) were obtained from VWR Scientific (West Chester, PA).
Mammalian Tissue Culture, Transfection, and Sample Preparation-Human embryonic kidney 293E (HEK293E) cells were cultured (at 37°C within 5% CO 2 ) and maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Transient transfections were performed by calcium phosphate as described previously (26). Empty pRK7 was used so that all transfection mixes contained 6 or 10 g of total DNA per 6-or 10-cm 2 plate, respectively. Prior to use, PA was prepared as described previously (27). After treatment, cells were washed twice with STE (pH 7.2) buffer and then harvested with lysis buffer (10 mM KPO 4 , 1 mM EDTA, 10 mM MgCl 2 , 50 mM ␤-glycerophosphate, 5 mM EGTA, 0.5% Nonidet P-40, 0.1% Brij 35, 1 mM sodium orthovanadate, 40 mg/ml phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 5 g of pepstatin, pH 7.2). The cell extracts generated were spun at 14,000 rpm for 10 min and then stored at Ϫ80°C. Due to subtle variations of cell passage number, acidity of cell medium, and confluency of cells, prepared cell lysates were routinely analyzed for Akt or ERK1/2 activation to confirm that the cells were properly responding to cell stimulation. The results presented in this paper are representative of at least three experiments.
Analysis of Protein Phosphorylation-For Western blot analysis, lysates were subjected to SDS-PAGE, transferred to nitrocellulose membranes, and blotted with the appropriate antibody followed by horseradish peroxidase-conjugated secondary antibodies. To observe a mobility shift of 4E-BP1 and tuberin, the proteins were resolved on 12.5 and 7.5% polyacrylamide gels, respectively. All immunoblots were detected by enhanced chemiluminescence. Anti-FLAG antibodies (M2) were purchased from Eastman Kodak Co. (New Haven, CT). Anti-HA antibodies was kindly provided by M. Chou (University of Pennsylvania, Philadelphia), and anti-4E-BP1, -tuberin (Thr-1462 phospho-specific), -Akt, and -eIF4E antibodies were bought from Cell Signaling Technology (Beverly, MA). The C-20 anti-tuberin antibody was purchased from Santa Cruz Biotechnology Inc. (Santa Cruz, CA). Anti-p38MAPK antibodies were purchased from R & D Systems Inc. (Minneapolis, MN). Anti-MAPK antibodies were generated as described previously (28). eIF4E was purified by using affinity chromatography on m 7 GTP-Sepharose as described (20). For analysis of hamartin and tuberin within the insoluble pellet, the cell lysates were prepared as described previously (12).
Immunoprecipitation and Immune Complex Kinase Assays-For immunoprecipitation studies of HA-tagged S6K1, cell extracts were immunoprecipitated with anti-HA antibodies bound to protein A-Sepharose (Pharmacia Corp., Peapack, NJ) for 3 h. Immunoprecipitates were washed as described previously (25). S6K1 kinase activity was determined in vitro by using recombinant GST-S6 (32 C-terminal amino acids of ribosomal protein S6) as a substrate, as described previously (25). Quantification of incorporation of the 32 P label was quantified on a Bio-Rad PhosphorImager with ImageQuant software. In vivo 32 P radiolabeling of tuberin was carried out in serum-starved HEK293E cells overexpressing hamartin and tuberin. The cells were incubated in phosphate-free medium for 4 h, and the cells were treated and pulsed with 5 mCi of [ 32 P]orthophosphate for 2 h. FLAG-tagged hamartin and tuberin were immunoprecipitated with an anti-FLAG antibody bound to protein G-Sepharose (Pharmacia Corp.) for 1 h. Immunoprecipitates were washed as for the S6K1 kinase assays (see above) and then subjected to SDS-PAGE and autoradiography.

Co-expression of Hamartin and Tuberin Impairs PMA-induced 4E-BP1 Phosphorylation and eIF4E
Function Independently of PI3K-Previously, we reported that co-expression of both hamartin and tuberin blocked 4E-BP1 phosphorylation upon insulin stimulation of HEK293E cells (12). In this cell line, insulin strongly stimulates signaling through the PI3K pathway but not the MAPK pathway (12). Given that activation of the MAPK pathway by phorbol esters results in the phosphorylation of 4E-BP1 that is dependent on mTOR but independent of PI3K (23, 24), we examined whether hamartin and/or tuberin could affect PMA-induced 4E-BP1 phosphorylation.
Hamartin and tuberin were co-expressed with 4E-BP1 in HEK293E cells that were serum-starved and then stimulated with PMA. By using insulin as a control to stimulate PI3K-dependent signaling, we show that PMA stimulation of this cell line activates the PKC/MAPK signaling pathway but not the PI3K/Akt pathway, as shown by the ability of PMA to stimulate phosphorylation of extracellular signal-regulated kinase (ERK) 1/2 (also known as MAPK) without inducing phosphorylation of Akt on Thr-308 (Fig. 1A). The level of hamartin and tuberin within the cell lysates were compared (Fig. 1B) and revealed that the amount of hamartin and tuberin protein was enhanced when each was co-expressed with the other, as reported previously (12,29). The ability of hamartin and tuberin to impair PMA-induced 4E-BP1 phosphorylation was analyzed by a gel shift assay in conjunction with phospho-specific antibodies that recognize 4E-BP1 when phosphorylated on Thr-37/Thr-46, Ser-65, and Thr-70. The ␣-, ␤-, and ␥-species of 4E-BP1 resolve as three separate bands on high acrylamide SDS gels. The lowest ␣-isoform band and the highest ␥-isoform band contain the least and most phosphorylated species of 4E-BP1, respectively. Overexpression of tuberin without its binding partner, hamartin, partially impaired the phosphorylation of 4E-BP1, which was further enhanced when hamartin was co-expressed with tuberin. Hamartin and tuberin co-expression inhibited the "priming" phosphorylation of 4E-BP1 at Thr-37/46 upon PMA stimulation, similar to the effect of rapamycin. Phosphorylation of Thr-37/46 is necessary for the optimal phosphorylation of the latter sites Thr-70 and Ser-65 (for review see Ref. 19). Ser-65 and Thr-70 phosphorylations are required for the dissociation of 4E-BP1 from eIF4E (reviewed in Ref. 19), where the occurrence of Ser-65 phosphorylation coincides with its release from eIF4E (30). Given the low degree of Ser-65 and Thr-70 phosphorylation on 4E-BP1 upon hamartin and tuberin co-expression, 4E-BP1 should remain bound to eIF4E after PMA stimulation. To verify this, endogenous eIF4E was purified on m 7 GTP-Sepharose, and the level of exogenous 4E-BP1 bound to eIF4E was examined (Fig. 1C). As expected, eIF4E⅐4E-BP1 complexes where not disrupted upon PMA treatment of cells co-expressing hamartin and tuberin. These data reveal that PMA-induced 4E-BP1 phosphorylation and release from eIF4E are potently blocked by tuberin-hamartin heterodimers, in a manner analogous to that of rapamycin (compare lanes 1 and 2 to 9 and 10 in Fig. 1C). Therefore, overexpression of tuberin and hamartin impairs cap-dependent translation by inhibiting mTOR signaling.
Tuberin Mutants (K599M and V796E) Derived from TSC Patients Are Unable to Repress PMA-induced 4E-BP1 Phosphorylation-Unlike wild-type tuberin, patient-derived mutants of tuberin are unable to impair PI3K signaling toward down-stream components of mTOR (10,12). We therefore examined whether two patient-derived tuberin mutants (K599M and V769E) were incapable of blocking PMA-induced 4E-BP1 phosphorylation (Fig. 2). Previously it was shown that the K599M tuberin mutant retained normal chaperone function, i.e. the K599M tuberin mutant still binds to and translocates hamartin from the membrane-bound insoluble fraction to the cytosol. However, the V796E tuberin mutant lacks this chaperone function (29). Consistent with previous work, the K599M tuberin mutant increased the amount of hamartin within the cytosol, whereas the V769E mutant of tuberin did not (Fig. 2). The majority of the V769E tuberin mutant was observed, as reported previously (29), to be present in the insoluble Nonidet P-40 membrane fraction after lysis (Fig. 2, 2nd panel from top). The phosphorylation status of 4E-BP1 upon PMA stimulation was unchanged by co-expressing the V798E tuberin mutant with hamartin when compared with the vector only control, as observed by the similar mobility shift of 4E-BP1 on SDS-PAGE and equivalent levels of Thr-37/46, Ser-65, and Thr-70 phosphorylation (Fig. 2). Overexpression of the K599M tuberin mutant with hamartin did not affect Thr-37/46 or Thr-70 phosphorylation and weakly inhibited Ser-65 phosphorylation of 4E-BP1 when compared with the wild-type tuberin upon PMA stimulation. Therefore, the K599M tuberin mutant still retains some properties of the wild-type tuberin to inhibit 4E-BP1 phosphorylation.
Tuberin and Hamartin Overexpression Impair S6K1 Activation upon PMA Stimulation-To explore further the ability of hamartin and tuberin to repress downstream mTOR signaling targets, the activity of exogenous S6K1 was analyzed from cells stimulated with PMA in the presence of protein kinase inhibitors of PI3K (wortmannin), mTOR (rapamycin), and PKC (bisindolylmaleimide I) (Fig. 3). PMA potently activated S6K1, which was fully blocked by 5 M bisindolylmaleimide I, a PKC inhibitor that abolished the PMA-induced PKC activation of ERK1/2 (as observed by bisindolylmaleimide I to prevent ERK1/2 phosphorylation). Under conditions where wortman- A, cell extracts were subjected to Western blot analysis with phospho-specific antibodies to compare levels of Akt phosphorylation at Thr-308 (␣-Akt(P)-T308) and ERK1/2 phosphorylation (ERK1(P) and ERK2(P)). Equal protein levels of Akt and ERK1/2 between samples are shown. B, ␣-FLAG antibody was employed to determine the protein levels of both exogenous hamartin and tuberin within the prepared cell lysates, as indicated. The extent of exogenous 4E-BP1 phosphorylation was determined with use of the ␣-HA antibody (BP1(␣-HA)) and antibodies that recognize 4E-BP1 when phosphorylated at Thr-37/46, Thr-70, and Ser-65, where indicated. The ␣-, ␤-, and ␥-species of 4E-BP1 are indicated. C, endogenous eIF4E was purified using m 7 GTP-Sepharose as shown with the ␣-eIF4E antibody, and exogenous HA-tagged 4E-BP1 associated with eIF4E was determined with the ␣-HA antibody.
Tuberin and Hamartin Overexpression Inhibit S6K1 Activity Independently of PI3K-To confirm whether hamartin and tuberin co-expression inhibits S6K1 via a PI3K-independent mechanism, it was necessary to activate S6K1 in the complete absence of PI3K signaling. To do this, we stimulated HEK293E cells with either EGF or PMA in the presence of wortmannin (to completely impair basal signaling through PI3K) (Fig. 4). Given that wortmannin blocked insulin-induced Thr-308 phosphorylation of Akt, the EGF-or PMA-induced activation of S6K1 in the presence of wortmannin occurs independently of PI3K. EGF-induced activation of S6K1 was blocked by both rapamycin and U0126 (Fig. 4A, lanes 5 and 6, respectively) and therefore shows a requirement for both mTOR and MEK to activate S6K1 in the absence of PI3K-mediated signaling. Considering that bisindolylmaleimide I did not prevent ERK1/2 phosphorylation or EGF-induced S6K1 activation, whereas the U0126 compound did (Fig. 4A, lanes 7 and 6, respectively), suggests that the PI3K-independent increased activity of S6K1 by EGF must be largely dependent on activation of MAPK signaling rather than PKC signaling. Furthermore, this experiment confirms that the concentration of bisindolylmaleimide I used does not inhibit MAPK signaling to ERK1/2 or S6K1 non-specifically. Hamartin and tuberin co-expression blocked the PI3K-independent activation of S6K1 upon EGF stimulation by ϳ70% (Fig. 4A, lane 8), which was repressed further upon pre-treatment with the U0126 compound (Fig. 4A, lane 9). PMA treatment that activates the PKC/MAPK pathway potently stimulated S6K1 activity 19-fold (Fig. 4A, lane 11), which was fully blocked by rapamycin (Fig. 4A, lane 12) and markedly impaired by either U0126 or bisindolylmaleimide I (Fig. 4A, lanes 13 and 14, respectively), treatments that were sufficient to block ERK1/2 phosphorylation upon PMA treatment. Hamartin and tuberin co-expression repressed the PI3Kindependent PMA activation of S6K1 by ϳ80% (Fig. 4A, lane  15), which was reduced to levels lower than that of the basal with either U0126 or bisindolylmaleimide I (Fig. 4A, lanes 16  and 17, respectively).
We show that hamartin and tuberin co-expression in conjunction with U0126 treatment significantly inhibited EGFand PMA-induced activation of S6K1 more than either hamartin/tuberin co-expression or U0126 treatment alone (see Fig.  4B), suggesting that tuberin-hamartin inhibits signaling through mTOR. Importantly, wortmannin was used in conjunction with either EGF or PMA treatments revealing that hamartin and tuberin represses S6K1 activity independently of PI3K.
Phosphatidic Acid (PA)-mediated mTOR Signaling Is Impaired by Hamartin and Tuberin Co-expression-To further confirm whether tuberin-hamartin inhibits mTOR-dependent signaling, we pre-treated HEK293E cells with wortmannin to inhibit basal PI3K-mediated signaling while promoting mTOR signaling with PA. PA has been shown to activate mTOR signaling through the direct binding of PA to the FKBP12/rapamycin-binding domain of mTOR (31). Co-expression of hamartin and tuberin impaired the activation of S6K1 upon PA treatment (Fig. 5). As reported previously (31), PA did not stimulate either Akt or ERK1/2 phosphorylation, indicating that S6K1 activation upon PA treatment was a result of increased mTOR signaling toward S6K1 (Fig. 5). Therefore, these data further support the idea that the tuberin-hamartin directly inhibits mTOR-dependent signaling.
PKC/MAPK-mediated in Vivo Tuberin Phosphorylation Occurs Independently of PI3K/Akt-The ability of PMA to activate S6K1 independently of PI3K suggests that a PI3K-independent input must inhibit the tuberin-hamartin heterodimer so that S6K1 can become activated by PKC/MAPK signaling. Because the function of tuberin-hamartin is modulated through Akt-dependent phosphorylation of tuberin (9,10,15,32), we wanted to investigate whether endogenous tuberin was also phosphorylated upon activation of the PKC/MAPK signaling pathway under conditions where Akt was inactive (Fig. 6A). Phosphorylation of tuberin retards its mobility on SDS-PAGE, as observed when HEK293E cells were treated with PMA (Fig.  6A, upper panel). Phosphorylation of tuberin by PMA was either impaired or fully prevented by pre-treating the cells with U0126 and bisindolylmaleimide I, respectively. The PMA-induced phosphorylation of tuberin was unaffected by wortmannin and reveals that both PKC and MAPK signaling leads to the phosphorylation of tuberin in the absence of PI3K signaling. Interestingly, insulin treatment, which resulted in increase reactivity of tuberin with both anti-phospho-Thr-1462 and anti-phospho-Akt substrate antibodies (Fig. 6A, 2nd and  3rd panel from top, respectively), did not alter the mobility of tuberin on SDS-PAGE. The anti-phospho-Akt substrate antibody recognizes proteins that are phosphorylated on the Ser or Thr within the RXRXX(S/T) motif (where X is any amino acid). This anti-phospho-Akt substrate antibody was first used to show that Akt phosphorylated tuberin on Ser-939 and Thr-1462 and that both conform to the RXRXX(S/T) consensus sequence (9). Therefore, the observed mobility shift of tuberin  Fig. 1. S6K1 activity assays were carried out as described under "Materials and Methods." The total levels of S6K1 are shown with use of the ␣-HA antibody. Incorporation of 32 P label into GST-S6 substrate was quantified using a PhosphorImager, and the fold activation of S6K1 from untreated conditions is graphed. An autoradiograph of the gel is presented.
upon PMA treatment likely occurs through the phosphorylation of other as yet unidentified sites that are distinct from those phosphorylated by insulin-dependent Akt activation. Surprisingly, PMA treatment caused an increase in tuberin phosphorylation at the Akt consensus phosphorylation RXRXX(S/T) sites (as observed by the increased detection of tuberin with the Akt substrate antibody (Fig. 6A, 2nd panel  from top)) that can be partially attributed to phosphorylation of tuberin at Thr-1462 (Fig. 6A, 3rd panel from top). Detection of PMA-induced phosphorylation with the anti-Akt substrate an-  Fig. 1. S6K1 activity assays were carried out as described under "Materials and Methods." The levels of S6K1 were determined with the ␣-HA antibody (S6K1(␣-HA)). Incorporation of 32 P label into the GST-S6 substrate was quantified using a PhosphorImager, and an autoradiograph of the gel is shown (bottom panel). tibody was insensitive to wortmannin but inhibited by both U0126 and bisindolylmaleimide I, revealing that Akt is not the kinase involved in the phosphorylation of tuberin at Thr-1462 or the RXRXX(S/T) consensus sites upon PMA treatment. This finding reveals that tuberin is phosphorylated in vivo by a protein kinase regulated by the PKC/MAPK signaling pathway that recognizes a similar consensus phosphorylation motif to that of Akt.
To examine the extent of tuberin phosphorylation upon PMA treatment in more detail, we radiolabeled exogenous tuberin in vivo. To confirm whether phosphorylation of tuberin by PMA occurred in the absence of basal PI3K/Akt signaling, we compared the incorporation of radiolabeled phosphate into FLAGtagged tuberin upon PMA treatment in the presence or absence of U0126 or bisindolylmaleimide I, in conjunction with wortmannin (Fig. 6B). To facilitate good incorporation of radiolabeled phosphate into tuberin, hamartin was co-expressed with tuberin to increase the levels of the tuberin-hamartin heterodimer within the cells. PMA markedly enhanced the phosphorylation of tuberin, which was blocked by pre-treating the cells with either U0126 or bisindolylmaleimide I. Interestingly, there was no significant increase of 32 P-radiolabeled incorporation into hamartin upon PMA treatment, revealing that the PKC/MAPK signaling pathway modulates the function of the tuberin-hamartin heterodimer by phosphorylating tuberin rather than hamartin.
The data presented in Fig. 6A suggest that the phosphorylation of tuberin upon PMA treatment occurs at sites that overlap with and are unique to those that are phosphorylated by Akt. To address this possibility further, we measured 32 Pradiolabeled incorporation into a double point mutant of tuberin, which has the two main Akt phosphorylation sites (Ser-939 and Thr-1462) mutated to alanines (referred to as "tuberin(SATA)"), upon PMA and insulin stimulation (Fig. 6C). As expected, insulin stimulation, which activated Akt but not ERK1/2, enhanced the incorporation of 32 P radiolabel into wildtype tuberin but not the tuberin(S939A/T1462A) mutant that lacked the Akt phosphorylation sites. Confirming the notion that tuberin can be phosphorylated at Akt-independent sites, PMA treatment enhanced the incorporation of 32 P radiolabel into the wild-type and the mutant (S939A/T1462A) tuberin protein. PMA-induced phosphorylation of the tuberin(S939A/ T1462A) mutant was blocked by pre-treatment with either U0126 or bisindolylmaleimide I, cell treatments that were sufficient to prevent phosphorylation of ERK1/2. These data indicate that two independent signaling pathways, PI3K/Akt and PKC/MAPK, lead to the phosphorylation of overlapping and distinct sites on tuberin. The arrows labeled p, pp, and ppp indicates increasing degrees of tuberin phosphorylation. Antibodies that recognize tuberin when phosphorylated at the RXRXX(S/T) sites and Thr-1462 were employed (␣-Akt-sub and ␣-Tub(P)-T1462, respectively). Protein levels of tuberin, ERK1/2, and Akt and levels of phosphorylation of ERK1/2 and Akt were determined as in Fig. 1. B, HEK293E cells co-expressing FLAG-tagged hamartin and tuberin (Ham/Tub), where indicated, were serum-starved and then subjected to in vivo radiolabeling as described under "Materials and Methods." During radiolabeling, the cells were treated with either 25 M U0126 or 5 M bisindolylmaleimide I, where indicated, in conjunction with 100 nM wortmannin for 30 min, and then stimulated with 100 ng/ml PMA for 40 min before being lysed. Both exogenous hamartin and tuberin were immunoprecipitated with an ␣-FLAG antibody conjugated to protein G-Sepharose beads, and the levels of 32 P label incorporation were determined by resolving the protein on SDS-PAGE followed by autoradiography. C, HEK293E cells co-expressing FLAG-tagged hamartin with either wild-type tuberin or the tuberin(S939A/T1462A) mutant, where indicated, were subjected to in vivo radiolabeling as described in Fig. 5B. Cells were pre-treated with either 25 M U0126 or 5 M bisindolylmaleimide I for 30 min and stimulated with either 100 ng/ml PMA or 100 nM insulin for 40 or 30 min, respectively, where indicated. An autoradiograph ( 32 P-FLAG-Tub) and a Western blot using the anti-FLAG antibody (Tub(␣-FLAG)) of the immunoprecipitated (IP) FLAG-tagged tuberin protein is shown. Protein levels of ERK1/2 and Akt and levels of phosphorylation of ERK1/2 and Akt were determined as in Fig. 1.
Tuberin(S939A/T1462A) Mutant Impairs PMA-induced S6K1 Activity-To investigate whether tuberin phosphorylation at Ser-939 and Thr-1462 inactivates the function of the protein upon PMA stimulation, we compared the effects of wild-type tuberin and tuberin(S939A/T1462A) to block PMAinduced S6K1 activation. We observed that the S939A/T1462A tuberin mutant was more effective at inhibiting S6K1 activation after stimulation with PMA than the wild-type tuberin (Fig. 7). The S939A/T1462A tuberin mutant inhibited PMAinduced S6K1 activation by 70%, whereas the wild-type only inhibited S6K1 by 35%. This finding indicates that tuberin phosphorylation at Ser-939 and Thr-1462 through the PKC/ MAPK signaling pathway functions to inhibit tuberin.
PMA-induced Tuberin Phosphorylation Is Independent of p38MAPK-The p38MAPK-activated kinase MK2 (also referred to as MAPKAPK2) has been shown to phosphorylate tuberin on Ser-1210 (33). It should be noted that the MK2 phosphorylation site within tuberin is "LYKSLS" (where the last amino acid is Ser-1210), which does not conform to the consensus RXRXX(S/T) sequence. Therefore, it is unlikely that the anti-phospho-Akt substrate antibody would recognize tuberin when phosphorylated on Ser-1210. To determine whether activation of the p38MAPK/MK2 signaling pathway promotes tuberin phosphorylation on the Akt consensus RXRXX(S/T) sites, we stimulated cells with anisomycin (referred to as "Ans"), a stress agonist (Fig. 8). Cell treatments with anisomycin induced p38MAPK phosphorylation but did not result in tuberin phosphorylation on the RXRXX(S/T) motifs. In contrast, insulin and PMA treatments induced tuberin phosphorylation on the RXRXX(S/T) sites that were blocked by treatments with wortmannin (to block PI3K) or bisindolylmaleimide I (to block PKC), respectively. Therefore, activation of the p38MAPK/MK2 signaling pathway does not induce tuberin phosphorylation on the RXRXX(S/T) sites that are phosphorylated upon activation of either the PI3K/Akt or PKC/MAPK signaling pathways. DISCUSSION Previous work (9 -13) clearly depicts that the TSC1/2 inhibits PI3K/Akt-mediated signaling toward S6K1 and 4E-BP1. There has been some controversy, however, as to whether tuberin-hamartin inhibits PI3K signaling through an mTOR-dependent or -independent mechanism (for review see Ref. 14). The data presented here strongly argue that mTOR-dependent signaling is indeed the target of TSC1/2. We report that tuberin-hamartin inhibits 4E-BP1 phosphorylation (Figs. 1B and 2) and S6K1 activity (Figs. 3, 4, A and B, and 5) upon cell treatments that activate the mTOR signaling pathway independently of PI3K. Tuberin-hamartin has also been shown previously to inhibit S6K1 activation upon the re-addition of amino acids to nutrient-starved cells (10 -12) and to be associ-  (30 min). Antibodies that recognizes tuberin when phosphorylated at the RXRXX(S/T) sites and p38 when dually phosphorylated at Thr-180/Tyr182 were used (Tub(P)Akt-sub and ␣-p38MAPK(P)). Protein levels of tuberin and p38 were also determined.
FIG. 9. Model that depicts both PKC/MAPK-and PI3K/Akt-dependent phosphorylation of tuberin and the control of mTOR signaling. Activation of either pathway results in the phosphorylation of tuberin on overlapping RXRXX(S/T) consensus sites (which includes Ser-939 and Thr-1462) that inactivates the tuberin-hamartin heterodimer. Inactivation of the tuberin-hamartin heterodimer promotes mTOR signaling to downstream components, eIF4E/4E-BP1 and S6K1, which is modulated by the small G-protein Rheb. Both nutrients and phosphatidic acid (PA) stimulates mTOR signaling, which is blocked by both rapamycin and tuberin-hamartin. At present it is unclear whether the nutrient signaling pathway functions upstream of Rheb or that Rheb-and nutrient-mediated signaling functions in parallel pathways upstream of mTOR. Kinase inhibitors are shown. BIM, bisindolylmaleimide I. FKBP12/rapamycin, rapamycin bound to its intracellular receptor, FKBP12, binds to and inhibits mTOR. ated with mTOR (10). Collectively, the evidence supports the model that mTOR signaling is the target of TSC.
This work extends our current understanding of how TSC functions as a tumor suppressor within mammalian cells. We report that tuberin-hamartin blocks PKC/MAPK-mediated signaling to the downstream components of mTOR, 4E-BP1, and S6K1. It is known that both S6K1 and eIF4E function to drive cell growth and proliferation (20,34). Given that tuberin is phosphorylated and inactivated by a PKC/MAPK signaling input, which is necessary for S6K1 activity and eIF4E function to drive cap-dependent translation, it is possible that the inappropriate inactivation of tuberin through constitutive PKC/ MAPK signaling may contribute to oncogenesis.
PI3K/Akt signaling inhibits the tumor suppressor function of tuberin-hamartin through the Akt-dependent phosphorylation of tuberin that occurs predominantly at Ser-939 and Thr-1462 (9,10,15,32). Interestingly, phosphorylation of these sites also occurs upon activation of the PKC/MAPK signaling pathway to inactivate the tumor suppressor function of the tuberin-hamartin heterodimer. We propose a model whereby both PI3K/Akt and PKC/MAPK signaling pathways converge on the tuberinhamartin independently (Fig. 9), and phosphorylation of these overlapping RXRXX(S/T) consensus sites within tuberin inactivates the tumor suppressor heterodimer to promote mTOR signaling. In support of our model, we show that the tuberin(S939A/T1462A) mutant, which cannot be phosphorylated by Akt (9), inhibits PKC/MAPK-mediated activation of S6K1 more potently than wild-type tuberin (Fig. 7A). Therefore, phosphorylation of these overlapping RXRXX(S/T) consensus sites within tuberin by PI3K/Akt-and PKC/MAPK-mediated signaling are important for the regulation of the TSC1/2 proteins, tuberin and hamartin. It is probable that additional PKC/MAPK-mediated phosphorylation of tuberin contributes to the inhibition of the tuberin-hamartin heterodimer. One kinase candidate that may phosphorylate tuberin at these RXRXX(S/T) consensus motifs was p90 ribosomal S6 kinase 1 (RSK1). However, overexpression of active RSK1 within HEK293E cells did not enhance the phosphorylation of tuberin (data not shown) implying that it is unlikely that RSK1 is the physiological kinase. Further studies will have to be carried out to identify the kinase(s) responsible for tuberin phosphorylation upon PKC/MAPK-mediated signaling. We are presently trying to determine which basophilic kinase phosphorylates the RXRXX(S/T) consensus sites within tuberin upon activation of PKC/MAPK signaling. Some potential kinase candidates include conventional and novel PKCs and RSK2/3/4.
During the preparation of this manuscript, a number of studies characterized Rheb (Ras homologue enriched in brain) as a molecular target of TSC1/2 within flies and mammals (35)(36)(37)(38)(39). Drosophila genetics using epistasis analysis positioned Rheb downstream of dTSC1/2 and upstream of dTOR (35,36). Research by Zhang et al. (37) identified dTSC2 as a Rheb GTPase-activating protein (GAP). Work from our laboratory demonstrated that Rheb specifically activated mTOR-mediated signaling toward S6K1 and 4E-BP1 that was blocked by tuberin and hamartin overexpression (39). Furthermore, mammalian studies show that tuberin is a RhebGAP (38,39). Interestingly, the ability of tuberin to function as a RhebGAP is significantly enhanced when associated with its binding partner, hamartin (39). These studies reveal that Rheb is an upstream modulator of mTOR and a downstream target of TSC1/2 (see model, Fig. 9) that is conserved between flies and mammals.
How does the phosphorylation of tuberin regulate the tuberin-hamartin heterodimer? It is currently unclear whether Akt-dependent phosphorylation of tuberin disrupts the inter-action of tuberin and hamartin due to conflicting data. Favoring Akt-mediated tuberin phosphorylation leading to dissociation of the tuberin-hamartin heterodimer, insulin stimulation, or Akt overexpression reduced the amount of hamartin that co-immunoprecipitated with tuberin in Drosophila-derived cell lines (6). Furthermore, a phosphomimetic tuberin protein (where the Akt-mediated phosphorylation sites were substituted to acidic residues) associated less favorably with hamartin (10). In contrast, two research groups (9,32) showed that the Akt-dependent phosphorylation of tuberin within mammalian cells was not sufficient to disrupt the TSC tumor suppressor complex. Similarly, we did not observe any disruption of the tuberin-hamartin heterodimer upon PMA-induced tuberin phosphorylation (data not shown). It has also been reported that Akt-mediated phosphorylation of tuberin increases the turnover rate of both tuberin and hamartin (10,32). We carried out [ 35 S]methionine pulse-chase experiments in HEK293E cells overexpressing tuberin to address whether PI3K/Akt-or PKC/MAPK-mediated signaling could enhance the turnover rate of exogenous tuberin, and we saw that neither insulin nor PMA stimulation of HEK293E cells resulted in the significant loss of tuberin over an 8-h period (data not shown). This discrepancy may be a consequence of overexpression that may enhance the stability of tuberin. Many research groups (40 -43) have also shown that 14-3-3 binds to the phosphorylated species of tuberin, and recently, p38MAPK-activated kinase MK2 was observed to phosphorylate tuberin on S1210 and direct 14-3-3 binding to tuberin (33). These studies provide an attractive mechanism where 14-3-3 association with tuberin can rapidly inhibit the tumor suppressor complex and presumably target tuberin for degradation. It is unlikely that MK2 is the candidate kinase that phosphorylates tuberin upon PKC/ MAPK signaling inputs as the p38MAPK/MK2 signaling pathway is not stimulated upon cell treatments with PMA (Fig. 8). Furthermore, stimulation of the p38MAPK signaling pathway does not result in the phosphorylation of tuberin at the consensus RXRXX(S/T) phosphorylation sites (Fig. 8). Given that the basophilic kinases, MK2 and mitogen and stress-activated kinase (MSK)-1/2, are downstream of p38, these p38-regulated kinases are unlikely to mediate tuberin phosphorylation at these RXRXX(S/T) sites.
Our data support the model that tuberin and hamartin function together as a heterodimer to inhibit mTOR signaling and reveal that tuberin is inhibited through both PI3K-dependent and -independent mechanisms. It is interesting to speculate that the tuberin-hamartin heterodimer functions as a pivotal sensor of PI3K/Akt and PKC/MAPK signaling pathways that, depending on the signal strength of either pathway, modulates the extent of mTOR signaling. Much is unknown regarding the mechanism by which tuberin-hamartin functions to inhibit mTOR, although recent research has shown that Rheb is a molecular target of TSC1/2 and an upstream regulator of mTOR. It will be interesting to see how this complex story unravels with further studies.