Cell Cycle-regulated Phosphorylation of Hamartin, the Product of the Tuberous Sclerosis Complex 1 Gene, by Cyclin-dependent Kinase 1/Cyclin B*

Tuberous sclerosis complex is a tumor suppressor gene syndrome whose manifestations can include seizures, mental retardation, and benign tumors of the brain, skin, heart, and kidneys. Hamartin and tuberin, the products of the TSC1 and TSC2 genes, respectively, form a complex and inhibit signaling by the mammalian target of rapamycin. Here, we demonstrate that endogenous hamartin is threonine-phosphorylated during nocodazole-induced G2/M arrest and during the G2/M phase of a normal cell cycle. In vitro assays showed that cyclin-dependent kinase 1 phosphorylates hamartin at three sites, one of which (Thr417) is in the hamartin-tuberin interaction domain. Tuberin interacts with phosphohamartin, and tuberin expression attenuates the phosphorylation of exogenous hamartin. Hamartin with alanine mutations in the three cyclin-dependent kinase 1 phosphorylation sites increased the inhibition of p70S6 kinase by the hamartin-tuberin complex. These findings support a model in which phosphorylation of hamartin regulates the function of the hamartin-tuberin complex during the G2/M phase of the cell cycle.

Tuberous sclerosis complex (TSC) 1 is a tumor suppressor gene syndrome whose manifestations can include seizures, mental retardation, autism, and tumors in the brain, retina, kidney, heart, and skin. Mutations in two genes, TSC1 on chromosome 9q34 (1) and TSC2 on chromosome 16p13 (2), cause TSC. Tuberin, the TSC2 gene product, and hamartin, the TSC1 gene product, are known to interact (3,4) and appear to function as a complex.
Tuberin is phosphorylated by the kinase Akt (protein kinase B) (24,(27)(28)(29)(30). Phosphorylation of tuberin by Akt negatively regulates inhibition of p70S6K by tuberin. Tuberin is also a substrate of the p38 and MK2 kinase cascade (31), mediating its interaction with 14-3-3 (32)(33)(34). Here, we report that endogenous hamartin is phosphorylated during G 2 /M, demonstrating for the first time that hamartin, like tuberin, is regulated by phosphorylation. Results from both in vitro and in vivo experiments indicate that hamartin is a substrate of the cyclin-dependent kinase CDK1 (cdc2), which is active in late cell cycle phases and promotes entry into mitosis when bound to cyclin B1 (reviewed in Refs. 35 and 36). Hamartin is phosphorylated by CDK1 at three residues, the most highly conserved of which lies within the hamartin-tuberin interaction domain. A mutant form of hamartin that cannot be phosphorylated by CDK1 increased the inhibition of p70S6K by the hamartin-tuberin complex. These results suggest that regulation of the hamartin-tuberin complex during G 2 /M may play a role in the integration of cell division with cell size, protein synthesis, and/or growth factor signaling.

EXPERIMENTAL PROCEDURES
Cells and Cell Treatments-HEK293 human embryonic kidney cells (ATCC CRL-1573) were cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum. G 2 /M arrest was induced by treating the cells for 18.5 h with 70 ng/ml nocodazole or Taxol dissolved in Me 2 SO (all from Sigma). For cell cycle synchronization, the cells were treated with 500 M hydroxyurea (Sigma) for 16 h, released in Dulbecco's modified Eagle's medium with 10% fetal bovine serum, and harvested at different time points after release. For CDK1 inhibition, the cells were treated with 20 M alsterpaullone (A.G. Scientific Inc., San Diego, CA).
Site-directed Mutagenesis-Site-directed mutagenesis was performed using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) and confirmed by DNA sequencing. The constructs were cloned into pcDNA 3.1ϩ or pcDNA3.1ϩ/His (Invitrogen).
Immunoblotting and Antibodies-The cells were lysed in RIPA buffer (1ϫ phosphate-buffered saline, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, supplemented by protease and phosphatase inhibitor cocktails I and II (Sigma)) unless otherwise specified. Ten g of total cellular protein were resolved by 5%, 7.5%, or 4 -20% SDS/PAGE and * This work was supported by National Institutes of Health Grant DK51052 and grants from the Tuberous Sclerosis Alliance and The LAM Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  transferred to Immobilon-P membranes (Millipore, Bedford, MA). The following antibodies were used to detect proteins by enhanced chemiluminescence (Amersham Biosciences): anti-tuberin C20 (Santa Cruz, Santa Cruz, CA), anti-hamartin (3), anti-␤-actin (Sigma), anti-phosphotyrosine clone 4G10 (Upstate, Waltham, MA), anti-phosphotyrosine clone PY99 (Santa Cruz), anti-phosphothreonine-proline, anti-p70S6K, and anti-phosphoT389-p70S6K (Cell Signaling, Beverly, MA).
Flow Cytometry Analysis-The cell pellets were fixed in 70% ethanol and stained with 20 g/ml propidium iodide (Sigma) containing 9.5 mg/ml RNase (Sigma). Flow cytometry was performed on a Becton-Dickinson FACScan machine. The percentages of cells in the G 1 , S, and G 2 /M phases of the cell cycle were determined using CellQUEST DNA Acquisition software (Becton-Dickinson).
Immunoprecipitations-The cells were lysed on ice in RIPA or PTY buffer (50 mM HEPES, 50 mM NaCl, 5 mM EDTA, 1% Triton X-100, and 50 mM NaF). 500 -1000 g of total protein were incubated at 4°C with anti-hamartin, anti-tuberin, or anti-TetraHis (Qiagen) antibodies for 1 h with constant rotation. Fifty l of protein A-Sepharose bead slurry (Invitrogen) were added to the immunocomplexes and rotated at 4°C for 16 h. The beads were washed in lysis buffer and boiled in Laemmli buffer (Bio-Rad).
Phosphatase Assays-The hamartin immunoprecipitates were washed with either CIAP buffer (50 mM Tris-HCl, pH 9.0, 1 mM MgCl 2 ) or PP1 buffer (50 mM Tris-HCl, pH 7.5, 1 mM MnCl 2 , 1 mM dithiothreitol). The beads were pelleted, resuspended in 100 l of the respective phosphatase buffer, and incubated at 37°C for 10 min. The respective phosphatase (30 units of calf intestinal alkaline phosphatase (CIAP; Amersham Biosciences) or 0.5 unit of the PP1 (Sigma)) was added to the beads and incubated at 37°C for 20 min. The beads were pelleted and boiled in Laemmli buffer.
CDK1/Cyclin B in Vitro Kinase Assay-Activated CDK1/cyclin B complexes were prepared by mixing His-tagged cyclin B bound to nickel beads with baculovirus-expressed recombinant CDK1, in the presence of 50 M ATP and 10 mM MgCl 2 (37). The CDK1/cyclin B complex was eluted from the beads using 62.5 l of imidazole buffer (10 mM HEPES, pH 7.4, 150 mM NaCl, 100 mM imidazole). The hamartin immunoprecipitates were washed in 500 l of kinase buffer (40 mM HEPES, 10 mM NaCl, 1 mM MgCl 2 , 1 mM MnCl 2 , pH 7) and incubated for 30 min at 30°C in 25 l of kinase buffer containing 100 M ATP and 1 l of activated CDK1/cyclin B complex. The reaction was terminated by boiling the beads in Laemmli buffer.
p70S6K in Vitro Kinase Assay-Myc-S6K1 (a gift of Dr. Richard Lamb), TSC1, and TSC2 were transfected in HEK293 cells, and the cell lysates were prepared in PTY buffer after serum deprivation for 24 h. Ectopic p70S6K was immunoprecipitated with Myc-agarose beads (BD Clontech, Palo Alto, CA). The beads were washed in ADBI buffer (20 mM MOPS, pH 7.2, 25 mM ␤-glycerol phosphate, 5 mM EGTA, 1 mM sodium orthovanadate, 1 mM dithiothreitol), and the p70S6K activity was measured using the S6 kinase assay kit (Upstate). Briefly, the immunocomplexes were incubated at 30°C for 30 min with a substrate peptide containing the p70S6K consensus phosphorylation site in the presence of PKC/PKA inhibitors, 15 mM MgCl 2 , 100 M ATP, and 10 Ci of [␥-32 P]ATP. The reactions were spotted on P81 paper, air-dried, and washed in 0.75% phosphoric acid and acetone, and the radioactivity was measured in a scintillation counter. The immunoprecipitates were incubated for 20 min with CIAP (a serine/threonine/tyrosine phosphatase), PP1 (a threonine/serine phosphatase), or without phosphatase and separated by 7.5% SDS/PAGE. The upshifted hamartin band (arrowhead) was present in the immunoprecipitate with no phosphatase treatment. The upshifted hamartin band was removed after incubation with CIAP or PP1, indicating that during nocodazole treatment hamartin is phosphorylated on threonine and/or serine residues. The results are representative of three independent experiments. B, HEK293 cells were arrested in G 2 /M after treatment with nocodazole for 18.5 h. The cell lysates and immunoprecipitates were separated by 7.5% SDS/PAGE and immunoblotted. An upshift in hamartin mobility was present after nocodazole in both the lysate and immunoprecipitate (arrowhead). Increased anti-phosphothreonineproline immunoreactivity was seen in the immunoprecipitated hamartin after nocodazole treatment. The results are representative of three independent experiments. C, in Cos7 cells, an increase in threonine-proline phosphorylation of hamartin was also present after nocodazole treatment. Similar results were seen in a second independent experiment.

Nocodazole-induced Arrest Results in a Mobility Shift of Endogenous
Hamartin-Under normal growth conditions, hamartin migrates predominantly as a single band in a Western immunoblot. During earlier work in which HEK293 cells were arrested with nocodazole (38), a second more slowly migrating band was observed (Fig. 1A). This band, which represented an apparent increase in the molecular mass of hamartin of ϳ10 kDa, became evident after 16 h of nocodazole treatment when the majority of the cells were in G 2 /M (Fig. 1B). This upshift in hamartin was also seen in HEK293 cells after Taxol treatment (data not shown).
Endogenous Hamartin Is Threonine-phosphorylated during Nocodazole Arrest-To determine whether the slower mobility band of hamartin was the result of phosphorylation, endogenous hamartin was immunoprecipitated and immunoblotted. As expected, an upshifted band was present in the immunoprecipitate after nocodazole-induced G 2 /M arrest ( Fig. 2A, compare first and second lanes). Treatment of the immunoprecipitate with CIAP for 20 min at 37°C abolished the upper band, demonstrating that hamartin is phosphorylated after nocodazole treatment. Treatment of the immunoprecipitate with PP1 removed the slower migrating hamartin band ( Fig. 2A), suggesting that hamartin is phosphorylated at a serine and/or threonine residue after nocodazole treatment.
Proline-directed kinases such as CDK1 are known to be active in mitotic cells, and the hamartin-tuberin complex has FIG. 3. Tuberin prevents hamartin phosphorylation and interacts with phosphorylated hamartin. A, HEK293 cells were lysed in PTY buffer. Hamartin was immunoprecipitated from cells with endogenous hamartin expression, ectopic expression of hamartin alone, or co-expression of both hamartin and tuberin. The immunoprecipitates were separated by 5% SDS/PAGE and immunoblotted with anti-hamartin or antiphosphothreonine-proline antibodies. An increase in phosphothreonine-proline immunoreactivity was seen with nocodazole treatment. Phosphothreonine-proline immunoreactivity was reduced when both hamartin and tuberin were expressed, compared with cells with ectopic expression of hamartin only, suggesting that tuberin co-expression attenuates hamartin threonine-proline phosphorylation. Similar results were seen in three independent experiments. B, HEK293 cells were transfected with His-hamartin and either wild type or a mutant form of tuberin (TSC2R611Q), which does not interact with hamartin. The cells were lysed in PTY buffer after incubation with Me 2 SO or nocodazole and ectopically expressed hamartin was immunoprecipitated (IP) with anti-His antibody. In the lysates, an upshift in hamartin mobility was observed in the nocodazoletreated TSC1/TSC2 cells (lane 2) and in both untreated and nocodazole-treated TSC1/TSC2R611Q cells (lanes 3 and 4, respectively). Similarly, in the hamartin immunoprecipitates, increased reactivity with the phosphothreonine-proline antibody was detected for the nocodazole-treated TSC1/TSC2 cells and in both Me 2 SO and nocodazole-treated TSC1/TSC2R611Q cells. C, HEK293 cells with endogenous expression of hamartin and tuberin were treated with Me 2 SO or nocodazole and lysed in PTY buffer, and either hamartin or tuberin was immunoprecipitated. As expected, a shift in hamartin mobility and increased reactivity to phosphothreonine-proline was observed in the hamartin immunoprecipitates from nocodazole-treated cells. A shift and reactivity to phosphothreonine-proline was also observed in tuberin immunoprecipitates from nocodazoletreated cells, indicating that tuberin can interact with phosphohamartin. D, His-tagged hamartin was co-expressed with tuberin. The cells were treated with nocodazole, and tuberin was co-immunoprecipitated with His-tagged hamartin. The same amount of tuberin co-immunoprecipitated with hamartin from nocodazole-treated or control cells. been previously shown to interact with CDK1 and cyclin B1 (39). To determine whether a threonine-proline residue of hamartin was phosphorylated, the hamartin immunoprecipitate was immunoblotted with a phosphothreonine-proline antibody. An increase in threonine-proline-phosphorylated hamartin was present in nocodazole-treated cells (Fig. 2B), indicating that hamartin is phosphorylated at a threonineproline residue and consistent with the result of the PP1 phosphatase experiment ( Fig. 2A). In contrast, no immunoreactivity was observed with either of two phosphotyrosine-specific antibodies after nocodazole treatment (data not shown). The increase in threonine-proline phosphorylation of hamartin after nocodazole treatment was also seen in Cos7 cells (Fig. 2C). Nocodazole did not induce threonine-proline phosphorylation of tuberin (data not shown).

Ectopically Expressed Hamartin Is Phosphorylated during Nocodazole Arrest, and Co-expression of Tuberin Attenuates
Hamartin Phosphorylation-These experiments demonstrated that endogenous hamartin is threonine-proline-phosphorylated during nocodazole-induced G 2 /M arrest. To determine whether exogenous hamartin was also threonine-phosphorylated after nocodazole treatment, hamartin was overexpressed in HEK293 cells, immunoprecipitated, and immunoblotted with the phosphothreonine-proline antibody. An increase in the phosphorylation of exogenous hamartin was seen in cells treated with nocodazole (Fig. 3A). The phosphorylation of exogenous hamartin was reduced when both tuberin and hamartin were overexpressed, compared with overexpression of hamartin alone (Fig.  3A). Additionally, hamartin co-expressed with a mutant form of tuberin (TSC2R611Q), which does not interact with hamartin (40), is phosphorylated even in the absence of nocodazole (Fig.  3B). This suggests that hamartin may be protected from phosphorylation when complexed with tuberin.
Phosphorylation of Hamartin Does Not Alter the Interaction with Tuberin-An interaction between endogenous tuberin and endogenous phosphorylated hamartin was detected in tuberin immunoprecipitates of nocodazole-treated cells (Fig. 3C). No difference in the amount of co-immunoprecipitated tuberin was observed in His immunoprecipitates from cells expressing TSC2 and His-TSC1 (Fig. 3D). Taken together, these data indicate that phosphorylation of hamartin does not have a major effect on the interaction between hamartin and tuberin.
Endogenous Hamartin Is Phosphorylated during the G 2 /M Phase of the Cell Cycle-Phosphorylation during nocodazoleinduced arrest could indicate regulation of hamartin during G 2 /M phases of the normal cell cycle or alternatively reflect the cell response to microtubular damage. To determine whether endogenous hamartin is phosphorylated during the normal cell cycle, HEK293 cells were synchronized in the G 1 phase with hydroxyurea followed by release into fresh medium. The cells were collected at 0, 8, 9, 10, 11, and 12 h and analyzed by fluorescence-activated cell sorting. Hamartin was immunoprecipitated at each time point and immunoblotted with the hamartin and phosphothreonine-proline antibody (Fig. 4A). An increase in phosphothreonine-proline immunoreactivity was seen, with the highest immunoreactivity 12 h after release when 55% of the cells were in G 2 /M. An upshift in hamartin mobility was not detected in this experiment. This may indicate that fewer sites are phosphorylated during normal G 2 /M compared with prolonged exposure to active CDK1 during nocodazole-induced arrest. To determine whether the phosphorylation was reversible, the experiment was repeated with later time points (Fig. 4B). The increased phosphorylation was detected at 8, 10, and 12 h after release from hydroxyurea, when 48, 53, and 41% of the cells were in the G 2 /M phase, respectively, but FIG. 5. CDK1/cyclin B phosphorylates hamartin in vitro. A, endogenous hamartin or hamartin from HEK293 cells overexpressing either hamartin only or both hamartin and tuberin was immunoprecipitated (IP), incubated for 30 min with activated CDK1/cyclin B complex, separated by 5% SDS/PAGE, and examined by Western immunoblot. The amounts of hamartin were normalized prior to immunoprecipitation. An upshift in hamartin mobility (arrowhead) was present when each immunoprecipitate was incubated with CDK1/cyclin B. B, the hamartin upshift (arrowhead) was seen with wild-type (wt) CDK1 but not with either of two kinase-inactive forms of CDK1 (T161A and N133A). C, the hamartin upshift (arrowhead) was removed when the product of the kinase reaction was incubated for 30 min with CIAP or the serine/threonine phosphatase PP1. Similar results were seen in three independent experiments for each panel.
FIG. 6. CDK1 phosphorylates hamartin in vivo. A, hamartin was immunoprecipitated (IP) from cells with endogenous expression of hamartin or overexpression of hamartin and tuberin and separated by 7.5% SDS/PAGE. Alsterpaullone, a CDK1 inhibitor, blocked the nocodazole-induced phosphorylation of hamartin. The same results were seen in a second independent experiment. B, to determine whether CDK1 was the kinase responsible for hamartin phosphorylation during the normal G 2 /M phase, the cells were synchronized by hydroxyurea, released, and harvested at 8, 10, 12, 16, and 24 h after release. One hour prior to harvest, the cells were treated with alsterpaullone to inhibit CDK1. Hamartin was immunoprecipitated and resolved by 7.5% SDS/ PAGE. The decrease in threonine-proline phosphorylation of hamartin was evident at 12 h in cells treated with alsterpaullone (right panel). The right and left panels of the figure are the same exposure. not at 16 h, when only 27% of the cells were in the G 2 /M phase. A small amount of base-line phosphothreonine proline immunoreactivity was present at the 0-and 24-h time points, which could indicate persistence of the phosphorylation by CDK1 or the activity of another kinase.
Hamartin Is Phosphorylated in Vitro by CDK1/Cyclin B-To determine whether CDK1 phosphorylates hamartin in vitro, immunoprecipitated hamartin was incubated with activated CDK1/cyclin B complexes. The amount of total protein used was normalized to the extent of overexpression prior to hamartin immunoprecipitation. An upshift in the mobility of endogenous hamartin was present after incubation with CDK1/cyclin B (Fig. 5A). A similar upshift was seen after kinase treatment of hamartin immunoprecipitates from cells with ectopic expression of hamartin only or co-expression of hamartin and tuberin (Fig. 5A). To establish the specificity of the CDK1/cyclin B complex on hamartin, the hamartin immunoprecipitates were incubated with two kinase-dead forms of CDK1 (T161A and N133A). The mobility shift was not observed with either kinase-dead form of CDK1 (Fig. 5B). To verify that the change in the mobility of hamartin was the result of phosphorylation, the product of the kinase assay was treated with CIAP or the serine/threonine phosphatase PP1, both of which restored the mobility of hamartin back to its base line (Fig. 5C).
Hamartin Is Phosphorylated in Vivo by CDK1-To determine whether hamartin is phosphorylated in vivo by CDK1, the cells were treated with alsterpaullone, a CDK1 inhibitor, in the presence of nocodazole for 18.5 h. Hamartin was immunoprecipitated at endogenous levels and from cells overexpressing both hamartin and tuberin. The level of threonine-proline phosphorylation of hamartin was significantly reduced in the immunoprecipitates from cells treated with alsterpaullone (Fig. 6A). To determine whether alsterpaullone inhibits the phosphorylation of endogenous hamartin during the normal cell cycle, HEK293 cells were synchronized with hydroxyurea. Alsterpaullone was added 1 h prior to harvest. The cells were harvested 8, 10, 12, 16, and 24 h after hydroxyurea release. The cells treated with alsterpaullone had decreased hamartin phosphorylation at 12 h, when ϳ49% of the cells were in G 2 M (Fig.  6B, right panel), compared with untreated control cells (Fig.  6B, left panel).

Mutation of Thr 417 , Ser 584 , and Thr 1047 Blocks Phosphorylation of Hamartin in Vitro and in Vivo-The preferred consensus phosphorylation sequence for CDK1 is (S/T)PX(K/R) (41).
Hamartin contains three potential CDK1 sites: Thr 417 (TPPR), Ser 584 (SPCK), and Thr 1047 (TPEK) (Fig. 7A). Thr 417 , the most highly evolutionarily conserved of the three potential sites, is within the hamartin-tuberin interaction domain (residues 302-430) (40,42). Thr 1047 is within the ezrin-radixin-moesin interaction domain of hamartin (residues 881-1084) (12). These sites were mutated to alanine either singly (TSC1-417A; TSC1-1047A), doubly (TSC1-417A/1047A), or all three (TSC1AAA). These forms of hamartin were overexpressed with wild-type TSC2, immunoprecipitated, and tested for in vitro phosphorylation by CDK1. As shown in Fig. 7B, both of the single mutants showed an upshift in hamartin after incubation with CDK1/cyclin B, although the degree of the shift was less than that of wild-type hamartin. This result indicated that neither Thr 417 nor Thr 1047 was the sole site of in vitro CDK1 phosphorylation. The double mutant also showed a shift. The

FIG. 7. Hamartin contains three consensus CDK1 phosphorylation sites ((S/T)PX(K/R)).
A, the evolutionary conservation of the three consensus CDK1 phosphorylation sites of hamartin is shown for human, rat, and Drosophila melanogaster. Of the three sites (Thr 417 , Ser 584 , and Thr 1047 ), only Thr 417 is conserved in all three species. Thr 417 is within the tuberin interaction domain of hamartin, and Thr 1047 is within the ezrin-radixin-moesin (ERM) interaction domain, as illustrated. B, to determine which potential CDK1 sites were phosphorylated, the in vitro kinase assay was performed with immunoprecipitated (IP) hamartin from cells expressing wild-type hamartin (wt-TSC1) or one of the following hamartin phosphomutants: T417A (TSC1-417A), T1047A (TSC1-1047A), T417A/T1047A double phosphomutant (TSC1-417A/1047A), and T417A/S584A/T1047A triple phosphomutant (TSC1AAA). The hamartin upshift (arrowhead) was evident after 5% SDS/PAGE for all forms of hamartin except the triple phosphomutant. The relative degree of upshift was less for each single and for the double phosphomutant compared with wild-type hamartin. triple mutant (TSC1AAA), however, did not show a shift, suggesting that the mobility shift detected in vitro reflects phosphorylation at these three sites.
In cells overexpressing the His-tagged triple phosphomutant hamartin (TSC1AAA), no increase in threonine-proline phosphorylation of immunoprecipitated hamartin was detected after nocodazole treatment (Fig. 8A) consistent with the hypothesis that Thr 417 and Thr 1047 are the predominant sites of threonine-proline phosphorylation. However, a more slowly migrating band appeared to be present after nocodazole treatment of cells expressing TSC1AAA, suggesting that sites not recognized by the phosphothreonine proline antibody may also be modified during nocodazole treatment.
To determine whether the phosphomutant form of hamartin (TSC1AAA) can be phosphorylated in vivo, cells transfected with wild-type TSC1 or TSC1AAA were arrested in the G 1 /S phase with hydroxyurea and released. Similarly to Fig. 4, the cells transfected with wild-type TSC1 showed increased hamartin phosphorylation at 8, 10, and 12 h after release, whereas the TSC1AAA phosphomutant did not show increased phosphorylation (Fig. 8B).

Expression of the Triple Phosphomutant Hamartin Does Not Block Nocodazole-induced Arrest and Does Not Alter the Cell
Cycle Profile of HEK293 Cells-To determine whether phosphorylation of hamartin inhibits the ability of cells to arrest in nocodazole or alters cell cycle kinetics, tuberin was overexpressed with wild-type hamartin or triple phosphomutant hamartin (TSC1AAA). No differences were observed in the cell cycle profile of asynchronous or nocodazole-arrested cells (data not shown).
Phosphomutant Hamartin Increases the Inhibition of p70S6K-To study the effect of hamartin phosphorylation on the activation of the mTOR/p70S6K pathway, HEK293 cells were co-transfected with Myc-S6K1 together with wild-type TSC2 and either wild-type TSC1 or TSC1AAA. Following immunoprecipitation of Myc-p70S6K, the activity of p70S6K was measured by an in vitro kinase assay. As previously reported by several groups, the p70S6K activity was decreased by the expression of the hamartin-tuberin complex. Expression of the nonphosphorylatable TSC1AAA further decreased the activity of p70S6K, compared with TSC1/TSC2 transfected cells (Fig.  9A). Western immunoblotting confirmed that the phosphorylation of p70S6K at residue Thr 389 was decreased in cells transfected with hamartin phosphomutant compared with wild type both in the total cell lysates and in p70S6K immunoprecipitates (Fig. 9B). We then tested the tuberin AKT phosphomutant S939A/T1482A (TSC2AA) (27). The magnitude of suppression of p70S6K phosphorylation was similar when TSC1AAA was expressed with wild-type TSC2, compared with wild-type TSC1 with TSC2AA. Co-expression of TSC1AAA with TSC2AA further decreased the phosphorylation of p70S6K (Fig. 9C). DISCUSSION We report here that endogenous hamartin is threonine-phosphorylated during nocodazole-induced G 2 /M phase and also during the G 2 /M phase of the normal cell cycle. Both in vitro and in vivo studies indicate that hamartin is a direct target for phosphorylation by CDK1. In vitro, hamartin is phosphorylated at three sites (Thr 417 , Ser 584 , and Thr 1047 ) by CDK1. In vivo, hamartin phosphorylation is blocked by the CDK1 inhibitor alsterpaullone, and hamartin with the three residues mutated to alanine (TSC1AAA) is not threonine-proline-phosphorylated during nocodazole-induced G 2 /M arrest or during G 2 /M progression of the normal cell cycle. Together, these findings indicate that CDK1 is the kinase responsible for the phosphorylation. Another group has previously reported that the hamartin-tuberin complex co-immunoprecipitates with CDK1, cyclin B1, and cyclin A (39), supporting the hypothesis that hamartin is directly phosphorylated by CDK1. To our knowledge, this is the first indication that endogenous hamartin is phosphorylated and that the hamartin-tuberin complex is specifically regulated during the cell cycle.
The fact that endogenous hamartin is phosphorylated during the normal cell cycle suggests that hamartin phosphorylation has physiologic significance. The most highly conserved consensus CDK1 phosphorylation site of hamartin lies within the hamartin-tuberin interaction domain. However, we did not detect a difference in the interaction between hamartin and tuberin during nocodazole arrest. We also did not detect a difference in the cell cycle profile of asynchronous HEK293 cells or their ability to arrest in nocodazole when tuberin was co-expressed with the three forms of hamartin (data not shown).
Phosphorylated hamartin from cells with endogenous expression of the two proteins co-immunoprecipitates with tuberin, and the amount of tuberin co-immunoprecipitating with hamartin is similar in nocodazole-treated and untreated cells. Co-expression of both tuberin and hamartin reduced the phosphorylation of hamartin both in vitro and in vivo, compared with expression of hamartin alone. Hamartin is constitutively phosphorylated when co-expressed with a patient-derived TSC2 mutant, which has been previously shown not to interact with hamartin (40). These data suggest that tuberin attenuates hamartin phosphorylation and/or that the primary substrate for phosphorylation is hamartin not complexed with tuberin.
Studies from both Drosophila and mammalian cells have demonstrated that the hamartin-tuberin complex negatively regulates mTOR and p70S6K (9,19,(23)(24)(25). During mitosis, the activity of p70S6K is suppressed, in part by direct phosphorylation of the p70S6K autoinhibitory domain by CDK1 (44, FIG. 8. The hamartin triple phosphomutant is not threonineproline-phosphorylated during nocodazole-induced G 2 /M arrest or during normal G 2 /M progression. A, His-tagged wild-type (His-TSC1) or triple phosphomutant hamartin (His-TSC1AAA) was co-expressed with tuberin in HEK293 cells, immunoprecipitated (IP) using anti-His antibody, separated by 7.5% SDS/PAGE, and immunoblotted with anti-hamartin or anti-phosphothreonine-proline antibodies. The hamartin triple phosphomutant was not threonine-proline-phosphorylated during nocodazole treatment, consistent with the in vitro results in Fig. 7. Similar results were seen in three independent experiments. B, HEK293 cells were transfected with His-TSC1/TSC2 or His-TSC1AAA/ TSC2, arrested in G 1 /S with hydroxyurea, and harvested at 8, 10, and 12 h after release. The cells were lysed in RIPA buffer and ectopically expressed hamartin was immunoprecipitated with anti-His antibody. The immunoprecipitated wild-type hamartin showed increased reactivity for the phosphothreonine-proline antibody at 10 and 12 h (similar to Fig. 4). This increase in immunoreactivity was not detected with the phosphomutant form of hamartin (TSC1AAA). 45). It is known that cell cycle progression and cell growth are coordinated, yet separable, in mammalian cells (43) as well as in yeast and Drosophila, but the signaling pathways responsible for this coordination are not yet completely understood.
We hypothesize that phosphorylation of hamartin by CDK1 in the G 2 /M phase regulates the activity, rather than the interaction, of the hamartin-tuberin complex during mitosis. If this is true, the phosphorylation of hamartin by CDK1 could play a central role in the coordination of cell cycle progression, cell growth, and cell size. Supporting this hypothesis, we found that a nonphosphorylatable form of hamartin (TSC1AAA) increased the inhibition of p70S6K by the hamartin-tuberin complex. The activity of p70S6K correlated with phosphorylation at residue Thr 389 , a key phosphorylation site and downstream target of mTOR. Phosphomutants for both TSC1 and TSC2 on CDK1 and AKT sites, respectively, acted together to further inhibit p70S6K phosphorylation compared with either mutant alone.
Mutation of the three CDK1 sites inhibited threonine-proline phosphorylation of hamartin during nocodazole-induced arrest. However, a slower migrating hamartin band was seen after nocodazole treatment of cells expressing the mutant hamartin (Fig. 8A), suggesting that sites not recognized by the phosphothreonine proline antibody may also be modified during nocodazole treatment. These could be nonconsensus serine phosphorylations by CDK1, phosphorylation by other kinases, or other types of post-translational modifications. Additional experiments will be needed to determine the cause and biological relevance of these modifications.
In conclusion, these data demonstrate that hamartin is phosphorylated by CDK1 during the G 2 /M phase of the cell cycle and indicate that hamartin phosphorylation does not alter the hamartin/tuberin interaction but instead regulates p70S6K activity during the G 2 /M phase. The regulation of p70S6K activity is complex, involving an ordered series of phosphorylation events that result in conformational changes. Multiple kinases directly phosphorylate p70S6K, including CDK1 (44). It seems surprising that phosphorylation of p70S6K by CDK1 results in decreased p70S6K activity (45), whereas phosphorylation of hamartin by CDK1 may result in increased p70S6K activity. It is possible that these apparently opposing effects serve to balance p70S6K activity. Alternatively, p70S6K could be temporally regulated by CDK1. Direct phosphorylation by CDK1 inactivates p70S6K during early mitosis. During late mitosis/ early G 1 phase, the phosphorylation of hamartin by CDK1 could release the inhibition on p70S6K, allowing for increased protein synthesis and cellular growth. Further studies will be needed to address these issues. FIG. 9. Hamartin phosphorylation decreases the inhibition of p70S6K. A, wild-type (TSC1) or the triple phosphomutant (TSC1AAA) hamartin was co-transfected with TSC2 and Myc-S6K1 in HEK293 cells. Myc-p70S6K was immunoprecipitated (IP) after 24 h of serum starvation, and the p70S6K activity was measured by in vitro 32 P incorporation. Expression of wild-type hamartin decreased the activity of p70S6K, compared with the vector control (pcDNA). Expression of the triple phosphomutant hamartin further decreased the p70S6K activity by 28% compared with wild-type hamartin. *, p Ͻ 0.05, compared with TSC1. The error bars represent standard deviation of three measurements. B, the lysates from A were immunoblotted with anti-hamartin and anti-tuberin. The lysates and Myc immunoprecipitates were immunoblotted with anti-p70S6K and anti-phosphoT389-p70S6K. In both lysates and immunoprecipitates from cells overexpressing TSC1AAA, there was decreased phosphorylation of p70S6K at Thr 389 compared with wild-type TSC1. Phosphorylation of p70S6K at Thr 389 correlates with the p70S6K activity shown in A. The results are representative of three independent experiments. C, HEK293 cells were co-transfected with Myc-S6K1, TSC1, TSC1AAA, TSC2, and the TSC2 S939A/T1462A AKT phosphomutant (TSC2AA). The cells were serum-starved for 24 h, lysed, and immunoblotted for tuberin, hamartin, P-S6K (T389), total S6K, and ␤-actin. The ratio of P-S6K/S6K was calculated based on densitometric analysis. Cells transfected with TSC1/TSC2 showed a decrease in the amount of S6K phosphorylation, which is further decreased in TSC1/TSC2AA transfectants. A similar degree of reduction (30%) in S6K phosphorylation is observed in TSC1AAA/TSC2 transfectants. The combination of the two mutants (TSC1AAA/TSC2AA) further decreases S6K phosphorylation. The results are representative of three independent experiments.