Cell, Volume 126

Supplemental Data

TSC2 Integrates Wnt and Energy Signals

via a Coordinated Phosphorylation by

AMPK and GSK3 to Regulate Cell Growth Ken Inoki, Hongjiao Ouyang, Tianqing Zhu, Charlotta Lindvall, Yian Wang, Xiaojie Zhang, Qian Yang, Christina Bennett, Yuko Harada, Kryn Stankunas, Cun-yu Wang, Xi He, Ormond A. MacDougald, Ming You, Bart O. Williams, and Kun-Liang Guan

Figure S1. Wnt Activates mTOR Pathway

S1A. Wnt-1 does not activate ERK. Wnt-1 expressing RIE or Rat1 cells were treated

with various inhibitors as indicated. ERK protein and phospho-ERK were detected by

Western blots with specific antibodies.

S1B. Stimulation of S6K phosphorylation by Wnt-1 overexpression is blocked by

rapamycin. Wnt-1 or empty vector expressing HEK293T cells were treated with or

without rapamycin for 30 minutes. Phosphorylation of S6K and Akt was determined by

immunoblot analyses.

S1C. Activation of mTOR signaling by Wnt-1 conditioned media. 293T cells were

treated with Wnt-1 conditioned media or control media for one hour. Phosphorylation of

endogenous proteins was detected by Western blotting with indicated specific phospho-

antibodies. Wnt-1 conditioned media were derived from Wnt-1 stable expressing 293T

cells. Vector indicates the media from 293T cells stably transfected with empty vector.

S1D. Dose dependent activation of S6K by Wnt-3a. Serum starved MEF cells were

stimulated with various concentrations of Wnt-3A for 30 minutes as indicated.

S1E. Rapamycin blocks Wnt-3a-induced phosphorylation of 4EBP1 in MEF cells.

S1F. Time course of Wnt-3a stimulation. Serum starved MEF were treated with Wnt-

3a (300 ng/ml) for indicated times. Active β-catenin was detected by specific antibody

recognizing the unphosphorylated protein. Note that activation of S6K preceeded

accumulation of active β-catenin.

S1G. Wnt-3a stimulates phosphorylation of S6K in C2C12 cells. Serum starved C2C12

cells were treated with Wnt-3a (300 ng/ml) for the indicated times.

S1H. Wnt-3a stimulates S6K phosphorylation in TSC2 expressing LEF cells. Serum

starved LEF cells were treated with Wnt-3a (300 ng/ml) for 30 minutes.

Figure S2. Wnt Stimulates Cell Growth in Rat1 and RIE Cells

S2A. Wnt-1 increases size of Rat1 cells. Vector and Wnt-1 expressing Rat1 cells were

treated with or without rapamycin (10 nM for 48 hours). Cell size of the G1 population

was determined by FACS analysis. Rapamycin decreases cell size at a basal level and

diminishes the effect of Wnt-1 on cell size.

S2B. Wnt-1 increases size of RIE cells. Experiments are similar to panel S2A.

S2C. Wnt10b expression increases cell size. Wnt10b expressing ST2 cells were treated

with rapamycin (5 nM) for 48 hours. Cell size was determined by FACS analysis.

Figure S3. The Effect of Rapamycin on Wnt-1-Expressing Tumor Development in

Nude Mice

The arrows indicate injection of rapamycin. Closed squares and circles denote tumor

growth of mice injected with rapamycin and control, respectively.

Figure S4. Involvement of Wnt Pathway Components in S6K Phosphorylation

S4A. DKK1 does not inhibit insulin response. MC3T3E1 cells were treated with insulin

(200 nM for 10 minutes) in the presence or absence of soluble recombinant human DKK1

(hDKK1, 200 ng/ml, pretreatment for 2 hours) as indicated. Phosphorylation of S6K and

S6 were determined.

S4B. DKK1 inhibits Wnt-3a-induced mTOR signaling. MC3T3E1 cells were

pretreated with various concentrations of hDKK1 as indicated for 2 hours followed by

Wnt-3a treatment for 30 minutes. Phosphorylation of S6K was determined by

immunoblot.

S4C. Axin is important for Wnt-induced mTOR signaling. Hela cells expressing

LKB1were transfected with indicated RNAi. Wnt-3a (300 ng/ml, for 30 minutes)

stimulation was indicated. Phosphorylation of endogenous S6K and S6K protein levels

were shown.

S4D. APC is required for Wnt-induced mTOR signaling. The APC mutant HT29 cells

with inducible expression of β-galactosidase or APC were treated with ZnCl (100 µM) to

induce the expression of β-galactosidase or APC. Serum starved cells were treated with

Wnt-3a (400 ng/ml, 40 minutes) as indicated. Phosphorylation of S6 was determined.

Figure S5. Function of GSK3 in Wnt-Induced mTOR Activation

S5A. LiCl inhibits β-catenin phosphorylation and increases β-catenin protein levels.

HEK293 cells were treated with indicated concentrations of NaCl or LiCl for 2 hours.

Phosphorylation and protein levels of S6K, Akt, and β-caternin were determined by

Western blotting.

S5B. Phosphorylation of S6K by LiCl treatment in MEF and LEF cells.

S5C. GSK3-Inhibitor (GSKI) treatment increases S6K phosphorylation in LEF and

bone marrow stromal (ST2) cells. Cells were treated with 20 µM GSK-I for 6 hours.

S5D. GSK3 is not essential for insulin to stimulate S6K phosphorylation. RIE cells

(top two panels) and MEF cells (bottom two panels) were treated with 20 mM LiCl for 1

hour and then stimulated with 400 nM insulin for 30 minutes. Phosphorylation of S6K

and its protein level were determined by immunoblotting.

S5E. Increase of TSC2 mobility by LiCl, GSK-I, or Wnt-1. Serum starved

cementoblasts were treated with LiCl (20 mM) or GSK-I (20 µM) and mobility of

endogenous TSC2 was determined (top panel). Wnt-1 expressing HEK293 cells also

show increased TSC2 mobility (bottom panel).

Figure S6. Phosphorylation of TSC2 by AMPK and GSK3

S6A. S1341 and S1337 in TSC2 are required for phosphorylation by GSK3β but not

AMPK in vitro. Recombinant GST-TSC2 fragment containing intact or mutated

phosphorylation sites were incubated with immunoprecipitated AMPK in the presence of

32P-ATP (left panels). Phosphorylation of TSC2 F1 was detected by phosphoimager. The

experiments in the right panels were performed similarly to those in Fig. 6B. . Arrow

indicates the GSK3-phosphorylated TSC2 F1. Protein levels for HA-AMPK and HA-

GSK3b-S9A were detected by immunoblot while GST-TSC2 F1 protein levels detected

by coomassie staining. Mutation of S1345A, but not S1341/1337, completely eliminated

AMPK dependent phosphorylation. Mutation of either S1345A or S1341/1337A

abolished TSC2 phosphorylation by GSK3.

S6B. GSK3β phosphorylates TSC2 in a sequential manner. Phosphorylation of various

TSC2 mutants by GSK3β was performed similarly to those in the right panels of S6A.

Mutation of S1341A eliminated TSC2 phosphorylation by GSK3 while mutation of

S1337A decreased TSC2 phosphorylation by GSK3. Mutation of S1333A or T1329A

had minor effects on TSC2 phosphorylation. The lack of effect of two N-terminal

mutations (S1333A or T1329A) is likely due to the fact that the in vitro phosphorylation

is incomplete at the more N terminal residues.

S6C. 2D phosphopeptide mapping of in vitro phosphorylation site. Samples from Fig.

S6B were analyzed by 2D phosphopeptide mapping.

S6D. S1337 and S1341 are phosphorylated in vivo. TSC2 wild type and mutant (2AC)

were transfected into HEK293 cells and labeled with 32P-phosphate. TSC2 protein was

immunoprecipitated and 2D phosphopeptide mapping was performed. Comparison of in

panel a and panel b shows that the intensities of the circled spots (1 to 4) are missing or

reduced in the 2AC mutant . The in vitro phosphorylated TSC2 F1 by GSK3β (panel c)

was mixed with the in vivo labeled TSC2-2AC (panelb) to produce panel d. The arrow

head in panel b indicates a new phosphopeptide created by the mutation of

S1337A/S1341A on TSC2.

S6E. Inhibition of GSK3 by GSK-I decreases the retarded mobility of TSC2 caused by

2DG.

S6F. Schematic presentation of TSC2 mutants and recognition of phospho-TSC2

antibody.

S6G. Characterization of TSC2 phosphorylation by GSK3 in vivo. HEK293 cells were

co-transfected with indicated plasmids. HA-TSC2 was immunoprecipited and

phosphorylation of TSC2 was monitored by phospho-specific antibody (upper panels).

Co-expression of GSK3 increased TSC2 phosphorylation. Mutation of the

phosphorylation sites (2AC) completely eliminated the recognition, suggesting the

specificity of the antibody. The effect of 2DG on TSC2 phosphoryaltion was also

determined. Transfected cells were treated with 2DG (10 mM) for 30 minutes. TSC2

phosphorylation was determined by immunoblot analysis (lower panels).

Figure S7. Phosphorylation of TSC2 by SGK3 Is Important for Cellular Energy

Response

S7A. Inactivation of the mTOR pathway by energy depletion is compromised in

GSK3β-/- cells. Cells were treated with the indicated concentrations of 2DG for 30

minutes. Protein levels and phosphorylation were determined by immunoblots with

specific antibodies.

S7B. Elimination of the GSK3 phosphorylation sites in TSC2 decreases ability of TSC2

to inhibit mTOR signaling. Wild type (Wt) and TSC2-4A (4A) (see Fig.6A for details of

the mutations) were expressed in TSC2-/- LEF cells and stable clones with similar

expression were characterized.

S7C. Glucose deprivation stimulates apoptosis of LEF cells expressing TSC2-4A but

not wild type TSC2. Cells were cultured in medium containing 25 mM (glucose+) or 0

mM (glucose-) glucose medium for 48 hours. Phase contrast photomicrographs are

shown.

S7D. Glucose deprivation induces caspase 3 cleavage in LEF cells expressing vector or

TSC2-4A, but not TSC2. Cells were cultured in medium containing 0 or 25 mM glucose

for 20 hours.

Figure S8. TSC2 Integrates Signals from Energy via AMPK and Wnt via GSK3 to

Regulate Cell Growth

In this model, Wnt signals to stimulate the mTOR pathway via GSK3 and TSC2.

Phosphorylation of TSC2 by GSK3 requires priming phosphorylation by AMPK. TSC2

integrates inputs from growth factors via Akt, RSK, and ERK, cellular energy levels via

AMPK, and Wnt via GSK3 to regulate mTOR, a central cell growth regulator.

Supplemental Information

Supplemental Experimental Procedures

Antibodies, Plasmids, and Materials

Anti-S6K, anti-phospho S6K (T389, S421/424), anti-4EBP1, anti-phospho 4EBP1

(S65, T37), anti-S6, anti-phospho S6 (S240/244), anti-Akt, anti-phospho Akt (S473),

anti-AMPK, anti-phospho AMPK (T172), anti-mTOR, anti-phospho-mTOR (S2448),

anti-phospho GSK3-β (S9), anti-phospho-β-catenin (S33/37/T41), anti-cyclin D, and

anti-cleaved Caspase-3 were obtained from Cell Signaling (Beverly, MA). Anti-Actin,

anti-VEGF and anti-TSC2 antibodies were from Santa Cruz Biotechnololgy (Santa Cruz,

CA). Anti-active-β catenin and anti-GSK3 antibodies were obtained from Upstate

(Charlottesville, VA). Anti-β-catenin, anti-phospho ERK(T202/204), and anti-ERK

antibodies were from Transduction laboratories (Lexington KY). Anti-Axin was obtained

from Zymed (South San Francisco, CA). Anti-HA and anti-FLAG antibodies were from

Covance (Philadelphia, PA) and Sigma (St. Louis, MO), respectively. Anti-phospho-

TSC2 antibody (S1337/1341) was generated using phosphorylated peptide

(CFQPpSQPLpSKSS) as antigen (Covance). Horseradish peroxidase-conjugated IgG

secondary antibodies were obtained from Amersham (Buckinghamshire, UK).

The plasmids expressing HA-tagged S6K1(αII), HA-tagged TSC2, Myc-tagged

TSC1, HA-tagged AMPK (αI), HA-tagged AMPK (αI) D159A and GST-TSC2 (1300-

1367), Flag-Dvl-2 were described previously (Inoki et al., 2003; Tamai et al., 2000).

Flag-β-catenin, Axin, Flag-dn TCF, and TCF reporters were kindly provided by E. R.

Fearon (University of Michigan). Human GSK3β (HA-GSK3β) and GSK3β-S9A (HA-

GSK3β-S9A) constructs were kindly provided by J. R. Woodgett (Ontario Cancer

Institute). HA-GSK3β-K85M/K86I mutant and the other mutant constructs used in this

study were created by site-directed mutagenesis via the Stratagene QuickChange Kit (La

Jolla, CA) and were verified by DNA sequencing.

Recombinant mouse Wnt-3a was purchased from R&D Systems (Minneapolis,

MN). The AMPK inhibitor (compound C) and 5-aminoimidazole-4-carboxamide-1-D-

ribofuranoside (AICAR) were obtained from Merck (Whitehouse Station, NJ) and Tronto

Research Chemicals (North York, ON, Canada), respectively. GSK3 inhibitor (GSK

inhibitor-1/TDZD-8) was purchased from Calbiochem (La Jolla, CA).

RNA Interference

Smart Pool short interfering RNA oligonucleotides toward TSC2, GSK3α,

GSK3β, Axin1, and Axin2 were purchased from Dharmacon (Denver, CO). EGFP

siRNA was (5’-AAGACAAUCGGCUGCUCUGAU-3’) was synthesized by Dharmacon.

Oligonucleotides were transfected into HEK293 cells, and lysates were made 48 hours

post transfection.

Reporter Assay

For luciferase reporter assays, cells were plated in six-well plates. pTOPFLASH,

pFOPFLASH, CMV-β-gal and indicated plasmids were co-transfected as described

previously. TCF transcriptional activity was measured as the ratio of luciferase activity

from the pTOPFLASH vector to the pFOPFLASH vector. All luciferase activities were

normalized by β-galactosidase activity.

Metabolic Labeling and Two-Dimensional Phosphopeptide Mapping

HEK293 cells were co-transfected with HA-TSC2, Myc-TSC1 and indicated

plasmids. The transfected cells were incubated in phosphate and serum free medium for

1 hour before incubation with 0.5 mCi/ml 32P-orthophosphate for 6 hours. Before

harvesting, the labeling cells were treated with or without 40 mM 2-deoxy-glucose for 30

minutes. HA-TSC2 was immunoprecipitated, resolved by SDS-PAGE and transferred to

a PVDF membrane. Phosphorylated TSC2 was visualized by autoradiography.

Phosphopeptide mapping was then performed. In brief, the phosphorylated TSC2 bands

were excised, fixed in methanol and incubated in 500 µl of 0.5% polyvinylpyrrolidone-40

dissolved in 100 mM acetic acid for 30 minutes at 37oC. The samples were then digested

with 20 µg of TPCK-treated trypsin (Sigma) at 37oC in 75 mM ammonium bicarbonate

buffer (pH 8.0 containing 5% acetonitrile). After digestion, samples were dried under

vacuum and suspended in 10 µl of water containing 4% acetonitrile. Samples were

spotted onto a cellulose plate and first dimensional electrophoresis was performed using

1% ammonium bicarbonate buffer pH 8.9. The plates were chromatographed in the

second dimension in chromatography buffer (ie: n-butanol/ pyridine/ acetic acid/ water,

75: 50: 24: 50). The plates were dried, and phosphopeptides were visualized by

autoradiography.

Transgenic Mice

Animal studies were cared for by the Unit for Laboratory Animal Medicine at the

University of Michigan under the approval by the University Committee on Use and Care

of Animals. DNA purification and microinjection of bone specific promoter (osteocalcin)

driving Wnt10b transgene (OCN-Wnt10b) into fertilized C57Bl/6 mouse eggs were

performed by the transgenic animal model core facility at the University of Michigan.

Mice were screened for integration of the Wnt10b transgene by PCR. Wild type C57Bl/6

females were used to establish lines. The detail characterizations of Wnt10b transgenic

mice will be described elsewhere (O. MacDougald). Wnt1 transgenic and LRP6 knockout

mice have been reported previously (Li et al., 2000; Pinson et al., 2000).

Immunohistochemical Staining

The mandibular bone was immersed immediately in fixative-10% neutralized

formalin buffer (Fisher Scientific). Fixation was carried out for 24 hours followed by

demineralization in 0.5M EDTA (PH. 7.33). Serial 7 µM paraffin embedded longitudinal

sections of the mandibular bone from both Wnt10b and wild type animals were processed

in the Histology Core at the School of Dentistry, University of Michigan. The factor

VIII-related antigen Ab-1 (NeoMarkers) and normal goat IgG (R & D) were used as

positive and negative controls, respectively.

Tumor Growth in Nude Mice

A primary Wnt-1 expressing tumor cell line, G105, was established from a

mammary tumor of an MMTV-Wnt-1 transgenic female. The tumor tissue was finely

chopped and further digested in Trypsin-EDTA 0.05% for 40 minutes. The tumor cells

were then cultured in Dulbecco’s modified Eagle’s Medium (DMEM) (Invitrogen;

Carlsbad, CA) supplemented with L-glutamine (2 mM), penicillin (100

units/ml)/streptomycin (1µg/ml) and 15% (v/v) fetal bovine serum at 37°C and in 5%

CO2 atmosphere. The tumorigenicity of G105 cells was assayed by subcutaneous

injection of 1 x 106 cells suspended in 100 µL of serum-free DMEM into right flank of 4

to 6 weeks old female athymic mice (Hsd:Athymic Nude-nu (nu/nu), which were

obtained from the Harlan (Indianapolis, Indiana). Rapamycin was reconstituted in

absolute ethanol at 10 mg/ml and diluted in 5% Tween 80 and 5% Peg-400 before

injection. Tumor cells were allowed to grow for 3 days without treatment. On day 4 after

tumor cell injection, mice were randomized into 2 groups (5 animals per group), and

treatment of rapamycin was initiated. One group was treated with rapamycin,

administered at doses of 1.5 mg/kg/d intraperitoneally for 5 consecutive days. The second

group received daily injections of the carrier solution as controls. Tumor size was

measured with calipers in 3 dimensions at times as indicated. Tumor volume was

calculated using the formula for volume of an ellipsoid: 4/3 x L/2 x W/2 x H/2, where L

= length, W = width, and H = height. All mice were sacrificed by asphyxiation with CO2

on day 28, and tumors were removed and tumor weight (gram) measured.

Quantification and Statistical Analysis

Quantification of signals in Western blotting was performed by pixel analysis via

the NIH Image software suite. For statistical analysis in multiple groups, differences

between groups were assessed by one-way ANOVA followed by Scheffe's test among the

groups.