Glycogen synthase kinase-3β suppresses the expression of protein phosphatase methylesterase-1 through β-catenin

Protein phosphatase 2A (PP2A) is the major tau phosphatase. Its activity toward tau is regulated by the methylation of PP2A catalytic subunit (PP2Ac) at Leu309. Protein phosphatase methylesterase-1 (PME-1) demethylates PP2Ac and suppresses its activity. We previously found that glycogen synthase kinase-3β (GSK-3β) suppresses PME-1 expression. However, the underlying molecular mechanism is unknown. In the present study, we analyzed the promoter of PME-1 gene and found that human PME-1 promoter contains two lymphoid enhancer binding factor-1/T-cell factor (LEF1/TCF) cis-elements in which β-catenin serves as a co-activator. β-catenin acted on these two cis-elements and promoted PME-1 expression. GSK-3β phosphorylated β-catenin and suppressed its function in promoting PME-1 expression. Inhibition and activation of GSK-3β by PI3K-AKT pathway promoted and suppressed, respectively, PME-1 expression in primary cultured neurons, SH-SY5Y cells and in the mouse brain. These findings suggest that GSK-3β phosphorylates β-catenin and suppresses its function on PME-1 expression, resulting in an increase of PP2Ac methylation.


INTRODUCTION
Neuronal microtubule associated protein (MAP) tau is abnormally hyperphosphorylated and aggregated into neurofibrillary tangles (NFTs) in brains of individuals with Alzheimer's disease (AD) and related neurodegenerative diseases [1,2]. The number of NFTs in the brain positively correlates with the degree of dementia [3][4][5]. Hyperphosphorylation of tau not only destroys its biological activity but also converts it into a cytotoxic protein that sequesters normal tau and aggregates into NFTs [6][7][8]. Thus, hyperphosphorylation plays a pivotal role in tau pathogenesis in AD and related conditions.
Protein phosphatase 2A (PP2A) is the major tau phosphatase in human brain [9]. It dephosphorylates tau at multiple sites with various degrees of efficiency [9]. PP2A activity is reduced in AD brain [9,10]. Inhibition of PP2A causes abnormal hyperphosphorylation of tau and AD-like tau pathology in cultured cells and in vivo in rodent brain [11][12][13]. PP2A consists of a core enzyme of scaffold A and a catalytic C subunit and a variable regulatory B subunit; the nature of the B subunit determines the localization and substrate specificity of the holoenzyme [14,15]. Methylation of PP2A catalytic subunit (PP2Ac) at Leu309 is required for the association of core enzyme with PR55 B subunit to dephosphorylate tau [16][17][18][19][20][21].
Then, we overexpressed GSK-3β in SH-SY5Y cells, and then measured the expression of PME-1 by Western blots and by quantitative real-time PCR (qPCR). We found that GSK-3β overexpression suppressed the expression of PME-1, as measured both at the mRNA level ( Figure 1C) and the protein level ( Figure 1B, 1D), suggesting that GSK-3β suppresses the expression of PME-1 in human neuroblastoma cells.
To further determine the role of GSK-3β on PME-1 expression, we knocked down GSK-3β by using siRNA and measured the PME-1 expression in SH-SY5Y cells.
We found that knockdown of GSK-3β increased both the mRNA ( Figure 1F) and protein ( Figure 1E, 1G) levels of PME-1. However, knockdown of GSK-3α by its siRNA did not obviously affect the expression of 3β, demethylated PP2Ac and PME-1 were analyzed by Western blots. (B-D) SH-SY5Y cells were transfected with pCI/HA-GSK-3β. The mRNA level of PME-1 was analyzed by qPCR (C). The protein levels of PME-1 and GSK-3β were analyzed by Western blots (B) and normalized with GAPDH (D). (E-G) SH-SY5Y cells were transfected with siGSK-3β to knockdown the expression of GSK-3β. The protein levels of PME-1 and GSK-3β were analyzed by Western blots (E) and normalized with GAPDH (G). The mRNA level of PME-1 was measured by qPCR (F). (H) SH-SY5Y cells were transfected with siGSK-3α to knockdown the expression of GSK-3α. The protein levels of PME-1 and GSK-3α/β were analyzed by Western blots. Data are presented as mean ± SD (n=3), *P < 0.05, ***P < 0.001. PME-1 ( Figure 1H). These results further confirm that GSK-3β inhibits PME-1 expression.

Human PME-1 promoter contains two putative LEF1/TCF elements
To understand how GSK-3β may suppress PME-1 expression, we first analyzed the promoter of the human PME-1 gene by Genomatix's MatInspector software [25,26]. The bioinformatic analysis revealed an array of putative nuclear factor-binding sites, including two potential LEF1/TCF like elements located on -349 bp --333 bp and +19 bp -+35 bp (Figure 2A), on which βcatenin acts as co-activator of TCF transcription factors to regulate the transcription of target genes.
To determine whether β-catenin binds to the LEF1/TCF elements of PME-1 promoter, we performed the ChIP assay in SH-SY5Y cells. We overexpressed β-catenin tagged with HA at the N-terminus in SH-SY5Y cells and immunoprecipitated β-catenin by anti-HA and then the bound DNA fragments were amplified by PCR to analyze co-immunoprecipitated LEF1/TCF elements of human PME-1 promoter. We found that anti-HA, but not control IgG, was able to co-immunoprecipitate two LEF1/TCF cis-elements 1 and 2 ( Figure 2B), suggesting that βcatenin may regulate PME-1 expression through the LEF1/TCF elements.
To study the regulation of transcription of PME-1, we inserted the promoter region of human PME-1, -1000 to +389, into the pGL4.10 vector to generate the reporter plasmid, pGL4/PME-1-1000 ( Figure 2C), transfected it together with pRL-TK into cells and measured luciferase activity by the dual luciferase assay. We found that the promoter of human PME-1 drove luciferase expression, leading to an increase of luciferase activity by ~50-and ~75 fold in the HEK-293T cells and SH-SY5Y cells, respectively ( Figure 2D).

LEF1/TCF cis-elements in PME-1 promoter enhance the PME-1 expression
To investigate the role of two LEF1/TCF elements in PME-1 expression, we deleted LEF1/TCF elements of human PME-1 promoter in the pGL4/PME-1 ( Figure  3A), and transfected them into HEK-293T cells, and analyzed luciferase activity to reflect the expression level of PME-1. We found that the luciferase activity was lower in pGL4/PME-1-330 than in pGL4/PME-1-350 and similarly it was lower in pGL4/PME-1+36 than that in pGL4/PME-1+1 ( Figure 3A). Thus, deletion of both cis-elements 1 and 2 of LEF1/TCF dramatically suppressed the luciferase activity, suggesting ciselements of LEF1/TCF work as enhancer and promote PME-1 expression. In addition, we found that deletion of -1000 bp --350 bp or 330 bp -+1 bp enhanced luciferase activity, suggesting that these regions of PME-1 promoter may contain the suppressors, the nature of which remain to be determined.
To further verify the role of two LEF1/TCF elements in promoting PME-1 expression, we mutated two LEF1/TCFs ( Figure 3B) and measured the luciferase activity. We found that M1, M2, M3 and M4 caused significant reduction of luciferase activity ( Figure 3B), confirming that LEF1/TCF elements 1 and 2 both promote the expression of PME-1.

β-catenin promotes the expression of PME-1 via LEF1/TCF cis-elements
β-catenin acts as a co-activator of transcription factors TCF/LEF in regulation of target gene expression [27].
To evaluate the role of β-catenin in PME-1 expression, we co-transfected β-catenin or its siRNA (siβ-catenin) with pGL4/PME-1-350 in HEK-293T cells and measured the PME-1 expression using the luciferase reporting system. We found that β-catenin overexpression enhanced, and its knockdown suppressed the luciferase activity significantly ( Figure 4A), indicating the regulation of PME-1 expression by β-catenin.
To determine the role of β-catenin in endogenous PME-1 expression, we overexpressed β-catenin or siβ-catenin in HEK-293T cells and analyzed PME-1 expression by qPCR and Western blots. We found that overexpression of β-catenin elevated the levels of endogenous PME-1 mRNA ( Figure 4D) and protein ( Figure 4C, 4E). Knockdown of β-catenin showed the opposite results in the protein ( Figure 4C, 4E) and mRNA level ( Figure  4D). Taking together, these results indicate that βcatenin also enhances the endogenous PME-1 expression.
To determine the effect of GSK-3β on PME-1 promoter activity, we co-transfected pGL4/PME-1-1000 with GSK-3β or its siRNA and analyzed the luciferase activity. We found that overexpression of GSK-3β significantly decreased the luciferase activity ( Figure  5E), whereas knockdown of GSK-3β with siRNA increased the luciferase activity ( Figure 5E). These results confirm that GSK-3β suppresses PME-1 transcription.
We co-expressed GSK-3β with β-catenin and determined the role of GSK-3β in β-catenin enhanced luciferase activity of pGL4/PME-1-1000. We found that luciferase activity was lower in the cells with cooverexpression of GSK-3β and β-catenin than that with Figure 2. The promoter of human PME-1 contains two putative LEF1/TCF elements. (A) Human PME-1 promoter region has two potential LEF1/TCF like elements (red). The promoter (-1000 to +389) of human PME-1 was analyzed by MatInspector software. Cis-elements are labeled with different color. There are two potential LEF1/TCF like elements. Other cis-elements are NFAT (nuclear factor of activated T cells), E2F (transcription factor family including E2F-and DP-like subunits), c-myb (Cellular homologue of avian myeloblastosis virus oncogene). (B) The SH-SY5Y cells were overexpressed with β-catenin tagged with HA. Monoclonal anti-HA were used to immunoprecipitate βcatenin and co-immunoprecipitated LEF1/TCF cis-elements were amplified by PCR using primers specific to LEF1/TCF cis-elements 1 and 2. (C) Schematic diagram of PME-1 promoter and its luciferase reporter plasmid. Human PME-1 promoter was inserted into pGL4.10 containing luciferase reporter gene to generate pGL4/PME-1-1000. (D) pGL4/PME-1-1000 and pGL4.10 showed in panel C were transfected into HEK-293T or SH-SY5Y cells. The Photinus pyralis luciferase activity and Renilla reniformis luciferase activity were measured subsequently and the Photinus pyralis luciferase activity was normalized with Renilla reniformis luciferase activity. Data are presented as mean ± SD (n=3); *P < 0.05, **P < 0.01. β-catenin overexpression alone ( Figure 5F). These results further suggest that GSK-3β suppresses βcatenin function in promoting PME-1 expression.
GSK-3β is known to phosphorylate β-catenin at Ser33 and promote its degradation by proteasome [28,30,31]. Missense mutation of Ser33 to Tyr in colorectal tumor cell line in SW48 is hyperactive as it avoids degradation and accumulates sufficiently to enter the nucleus [32,33]. We mutated Ser33 to Tyr, β-cateninS33Y and co-expressed it with pGL4/PME-1-1000. We found that compared with wild type β-catenin, β-cateninS33Y increased the luciferase activity significantly ( Figure  5G). Taking together, these results indicate that GSK-3β phosphorylates β-catenin and promotes its degradation, resulting in suppression of β-catenin function in the transcription of PME-1.

Activation of PI3K/AKT signaling suppresses GSK-3β activity and promotes PME-1 expression
Activation of PI3K-AKT signaling leads to inhibition of GSK-3β activity by phosphorylation at Ser9. To inhibit GSK-3β, we treated HEK-293T cells with 2% fetal bovine serum (FBS) for 24 hr to activate PI3K-AKT and analyzed the phosphorylation of GSK-3β and βcatenin by Western blots. We found that FBS induced the increase in phosphorylation of GSK-3β at Ser9 ( Figure 6A, 6B) and increase of Non-pS-β-catenin ( Figure 6A, 6B), compared with the cells cultured without serum. The levels of PME-1 mRNA and protein were increased in FBS treated cells ( Figure 6C). Thus, inhibition of GSK-3β by activation of PI3K/AKT signaling suppressed β-catenin phosphorylation, leading to an increase in the expression of PME-1.
Insulin activates PI3K signaling and suppresses GSK-3β activity [34,35]. To learn the regulation of PME-1 expression by PI3K-GSK-3β in neurons or neuronal like cells, following 2 hr culture without serum, SH-SY5Y cells were treated with or without 100 nM insulin for 0.5 hr or 18 hr and subjected to Western blots. We found that the phosphorylated Ser473-AKT and Ser9-GSK-3β were clearly increased in the cells treated with 100 nM insulin for 0.5 hr and /or 18 hr ( Figure 6D), suggesting activation of PI3K signaling by insulin. The  10. Two LEF1/TCF elements are named as LEF1/TCF-1, located at -349 to -333 and labeled in blue color, and LEF1/TCF-2, located at +19 to +35 and labeled in purple color. The deletions shown as panel A left were co-transfected with pRL-TK into HEK-293T cells. The luciferase activity was measured. (B) LEF1/TCF cis-elements 1 and 2 were mutated and the mutated nucleotides are labelled in red color. pGL4/PME-1-1000 with different mutations were co-transfected with pRL-TK into SH-SY5Y cells. The luciferase activity was analyzed. Data are presented as mean ± SD (n=3), *P < 0.05, ***P < 0.001, ****P < 0.0001. AGING level of PME-1 did not obviously alter in the cells treated with 100 nM insulin for 0.5 hr, but was markedly increased in that for 18 hr ( Figure 6D), supporting that inhibition of GSK-3β by insulin enhanced PME-1 expression in SH-SY5Y cells.
Then, we treated primary cultured neurons with 100 nM insulin for 18 hrs and analyzed PME-1 expression by Western blots ( Figure 6E). We found that insulin treatment significantly increased the levels of non-pS-β-catenin and PME-1 ( Figure 6F), confirming the enhancement of PME-1 expression by PI3K signaling in neuron.
Furthermore, we activated GSK-3β by inhibiting PI3K/AKT signaling with LY294002 in cultured neurons and analyzed PME-1 mRNA by qPCR. We found that LY294002 suppressed the expression of PME-1 in dose-dependent manner ( Figure 6G), confirming that GSK-3β suppresses PME-1 expression in neurons.

Specific inhibition of GSK-3β enhances PME-1 expression in cells and in vivo
ARA014418 specifically inhibits GSK-3β activity by competing with ATP. We investigated whether inhibition of GSK-3β affects the binding of β-catenin with LEF1/TCF cis-elements. We overexpressed βcatenin in SH-SY5Y cells and treated the cells with the ARA014418 for 4.5 hr. We performed ChIP assay and found that ARA014418 treatment increased the levels of co-immunoprecipitated cis-elements of LEF1/TCF-1 and LEF1/TCF-2 by β-catenin ( Figure 7A), suggesting inhibition of GSK-3β enhances the binding of β-catenin to LEF1/TCF elements. into HEK-293T cells, and then the PME-1 expression was determined through luciferase activity. (B) HEK-293T cells were co-transfected with pGL4/PME-1s, β-catenin and pRL-TK. The luciferase activity was measured. (C-E) HEK-293T cells were transfected with β-catenin or siβcatenin and analyzed for PME-1 and β-catenin by Western blots (C, D) or qPCR (E). Quantification (E) of the Western blots are presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. AGING To determine the effect of ARA014418 on PME-1 expression in primary cultured cortical neurons, we treated the neurons with various concentrations of ARA014418 for 4.5 hr and analyzed mRNA level of PME-1 by qPCR. We found that ARA014418 dose-dependently increased PME-1 mRNA level ( Figure 7B).

DISCUSSION
Methylation of PP2Ac plays an important role in the regulation of tau phosphorylation. We previously reported that GSK-3β enhances PP2Ac methylation through suppression of the expression of PME-1 [24]. In the present study, we further investigated the molecular mechanism involved in the regulation of PME-1 expression by GSK-3β. We found that the promoter of human PME-1 contains two LEF1/TCF cis-elements, which act as enhancers to promote the expression of PME-1 and that binding of β-catenin to LEF1/TCF ciselements enhances the expression of PME-1. Furthermore, GSK-3β phosphorylates β-catenin and suppresses its nucleus translocation, leading to suppression of PME-1 transcription. Inhibition of GSK-3β by activated PI3K signaling through its phosphorylation at Ser9 enhanced PME-1 expression (Figure 8). with β-catenin and cultured without Fetal bovine serum (FBS, as control, con) or with 2% FBS for 48 hr. The cell lysates were analyzed for total GSK-3β, β-catenin, PME-1 and the phosphorylation of β-catenin and GSK-3β by Western blots (A). Levels of phosphorylated β-catenin and GSK-3β were normalized with corresponding proteins (B). The mRNA level of PME-1 was measured by qPCR and normalized with GAPDH (C). The protein level of PME-1 was quantified from panel A and normalized with GAPDH. (D) SH-SY5Y cells were treated with 100 nM insulin for 0.5 hr or 18 hr and analyzed by Western blots developed with indicated antibodies. (E, F) Primary cortical neurons were treated with 100 nM insulin for 18 hr. The levels of β-catenin and PME-1 were analyzed by Western blots (E) and normalized with GAPDH (F). (G) Primary cortical neurons were cultured and treated with the indicated concentration LY294002 for 4.5 hr. The PME-1 mRNA level was measured by qPCR and normalized with GAPDH. Data are presented as mean ± SD (n=3); *P < 0.05; **P < 0.01. AGING Transcription factors lymphoid enhancer-binding factors/T-cell factors (LEFs/TCFs) are known to bind to the consensus sequence CCTTTGWW, cis-element of LEF/TCF, and regulate the transcription of target genes [36]. In the present study, bioinformatic analysis with MatInspector revealed that the human PME-1 promoter contains two LEF1/TCF like elements: CTTTGAT at -344bp to -338bp and CCTTTGT at +23bp to +29bp. Deletion or mutation of each of LEF1/TCF motifs decreases significantly luciferase expression driven by human PME-1 promoter, suggesting that these two LEF1/TCF cis-elements promote PME-1 expression. In addition to LEF1/TCF cis-elements, PME-1 promotor contains several cis-elements, including NFAT, E2F, and c-Myb. Deletion of -1000 --350 or -330 -+1 of PME-1 promoter significantly increased luciferase activity, suggesting silencers in these regions. The role of these cis-elements in the regulation of PME-1 transcription remains to be studied. The TCF/LEF is a group of transcription factors which bind to DNA through a high mobility group domain. The mammalian TCF/LEF family comprises four nuclear factors designated TCF7, LEF1, TCF7L1, and TCF7L2 (also known as TCF1, LEF1, TCF3, and TCF4, respectively) [37,38]. These nuclear factors are involved in the Wnt signaling pathway, where they recruit the coactivator βcatenin to enhancer elements of the target genes [27]. Without Wnt stimulation, β-catenin forms a complex hr, for ChIP assay using antibody to β-catenin. The two LEF1/TCFs were amplified by PCR with their specific primers. (B) Primary cortical neurons from embryonic day 18 SD rat were cultured and treated with the indicated concentration ARA014418 for 4.5 hr. The PME-1 mRNA level was measured by qPCR and normalized with GAPDH. (C-E) ARA014418 (5 mM 2 μl/mouse) was intracerebroventricularly injected into hTau transgenic mice for 48 hr. The cortex was homogenized and analyzed by Western blots developed with the indicated antibodies (C) or qPCR for PME mRNA (E). GAPDH was included as a loading control. Levels of phosphorylated β-catenin and GSK-3β were normalized with corresponding proteins (D). Data are presented as mean ± SD (n=4 or 5), *P < 0.05, **P < 0.01. with axin (axis inhibitor), adenomatous polyposis coli (APC), casein kinase 1α (CK1α), and GSK-3β and undergoes phosphorylation-dependent ubiquitination and degradation [28,30,31,39,40]. Furthermore, conserved serine and threonine residues (Ser 33, Ser 37, Thr 41, and Ser 45) at the NH2-terminal domain of βcatenin are phosphorylated by GSK-3β and CK1α. Activation of Wnt signaling leads to the recruitment of Axin to the membrane and functional inactivation of the β-catenin destruction complex. This results in cytoplasmic and nuclear accumulation of β-catenin. In the nucleus, β-catenin associates with TCFs to activate transcription of Wnt signaling target genes. In the present study, ChIP assay showed that β-catenin bound to the two LEF/TCF cis-elements of PME-1 promoter and enhanced the PME-1 expression determined by using luciferase reporter.
PP2A is also known to regulate Wnt signaling by directly dephosphorylating β-catenin at Ser 37 and Thr41 to generate active β-catenin with enhances transcriptional activity [29]. PR55α, a PP2A regulatory subunit which requires methylated PP2Ac to form the holo-enzyme [62], regulates the dephosphorylation of β-catenin. Heat shock protein 105 (HSP105), a component of the β-catenin degradation complex, recruits PP2A to dephosphorylate β-catenin to maintaining a phosphorylation balance [63]. Thus, β-catenin may promote PME-1 expression, leading to demethylation of PP2Ac and hence decreased PP2A activity, which phosphorylation of GSK-3β and inhibition its activity in phosphorylating β-catenin. β-catenin translocates to nucleus and act as co-activator with LEF1/TCF to promote PME-1 expression, which catalyzes the demethylation of PP2Ac.
consequently may result in phosphorylation and reduction of β-catenin by a negative feedback.
In summary, we found here that β-catenin promotes PME-1 transcription. GSK-3β suppresses PME-1 expression by phosphorylation of β-catenin, leading to increase of methylation of PP2Ac. Thus, we speculate that activation of GSK-3β resulting from insulin resistant in AD brain one hand causes hyperphosphorylation of tau, other hand may increase PP2A methylation through PME-1, resulting in the attenuation of hyperphosphorylation of tau.

Animals
The hemizygous human tau transgenic (B6.Cg-Mapttm1(EGFP)Klt Tg(MAPT)8cPdav/J, Tg/hTau) mice with murine tau knockout background [64] obtained from the Jackson Laboratory (Bar Harbor, ME, USA) were used in this study. The mice were housed in cages at 24 ± 1 °C, 50~60% humidity under a 12 h light/dark cycle, with access to food and water ad libitum. The housing, breeding, and animal experiments were in accordance with the approved protocols from Institutional Animal Care and Use Committees of Nantong University and according to the United States PHS Policy on Humane Care and Use of Laboratory Animals.

Primary cortical neuron culture
Rat cerebral cortical neurons were isolated and cultured according to a previously described method [65]. Briefly, fetal cerebral cortices from Sprague Dawley (SD) rats at embryonic day 18 were cut into pieces in cold PBS buffer, digested with 0.25% trypsin at 37 °C for 15 min, and centrifuged at 1000 rpm for 5 min. The pellet was suspended in DMEM with 10% FBS. Following 400 mer filtering, cell density was counted and adjusted to 1 × 10 6 cells/ml. The cells were plated onto poly-D-lysine (100 μg/ml) coated 12-well plates and were maintained in a humidified atmosphere at 37 °C, 5% CO2. The medium was changed to neurobasal supplemented with 2% B27 (Invitrogen) containing 100 U/ml penicillin and 100 μg/ml streptomycin 4 h later. On the second day of culture, the cells were treated with 10 μM Ara-C for 24 h to inhibit non-neuronal cells. On the seventh day, the cells were treated with various concentrations of ARA014418 or LY294002 at for 4.5 h for measurement of PME-1 mRNA or treated with 100 nM insulin for 18 hr for measurement of PME-1 protein.

Quantitative real-time PCR (qPCR)
Total RNA was isolated from primary cultured cortical neurons or from cultured cells using the HP Total RNA Kit (Omega, Norcross, GA, USA). One microgram of total RNA was used for first-strand cDNA synthesis with Oligo-(dT)18 by using the HiScript II Q RT SuperMix for qPCR (Vazyme, Nanjing, JS, China). The target mRNA was amplified by using LightCycler® Multiplex DNA Masters (Roche, Indianapolis, IN) in a LightCycler® 96 system (Roche, Indianapolis, IN) under the condition: 95 °C 10 min, at 95 °C for 30 sec and at 60 °C for 1 min for 40 cycles. Relative levels of target mRNAs were calculated by the comparative CT (threshold cycle) method (2 -ΔΔCT ). The primers used for PME-1 are as follows: human PME-1 forward: 5′GTCGTCCTAAAACCTTCAAGTC Table 1. List of primers used for generating mutated promoter of PME-1 constructs, mutated β-catenin constructs and TCF/LEF constructs.

Dual luciferase assay
HEK-293T or SH-SY5Y cells were co-transfected with pCI/β-catenin, pCI/GSK-3β, pGL4/PME-1s with pRL-TK, and then the cells were lysed in 0.1 ml of passive lysis buffer (Promega, Madison, WI). The luciferase activity was measured by the dual luciferase assay (Promega, Madison, WI) according to manufacturer's manuals. The Photinus pyralis luciferase activity and Renilla reniformis luciferase activity were measured subsequently and the Photinus pyralis luciferase activity was normalized with Renilla reniformis luciferase activity.

Chromatin immunoprecipitation (ChIP)
SH-SY5Y cells were transfected with pCI/β-catenin tagged with HA at the N terminus and treated with or without 20 μM ARAO14418 for 4.5 hr. Chromatin immunoprecipitation (ChIP) assay was performed using the Simple ChIP® Enzymatic Chromatin IP Kit (Cell Signaling, Danvers, MA, USA), following the instructions of the manufacturer. The cells were fixed with 1% formaldehyde in PBS for 10 min at room temperature, and then glycine was added to a final concentration of 125 mM and incubated for 5 min at room temperature. The cells were collected in cold PBS with Protease Inhibitor Cocktail, then lysed with corresponding buffer. The nuclei were then collected and the chromatin was enzymatically digested into fragments ranging from 150 to 900 bp. ChIP was performed using protein G magnetic beads and either 2 mg of β-catenin antibody (Cell Signaling), HA antibody (Sigma), or the nonspecific IgG antibody (#2729; Cell Signaling). Protein-DNA cross-links were reversed, and the DNA was purified. The specific primers (listed in Table 1) were used to amplify regions containing LEF1/TCF-1 and LEF1/TCF-2 cis-elements. LEF1/ TCF-1 region primer forward: 5′GTTTATCCAGCGG GGGTGCGGGTCT3′ and reverse: 5′GAAACAACG AAAAGGCCAAGGGTCTG3′; LEF1/TCF-2 region primer forward: 5′ACGCTTTGCTACAATACCTCTC CG3′ and reverse: 5′CTGGGAACCCGACGACTTTC ACCGG3′.

Immunofluorescence staining
HeLa cells were plated on glass coverslips in 24-well plates and transfected with pCI/HA-GSK-3β; 48 h later, the cells were washed with PBS and fixed with 4% paraformaldehyde in PBS for 30 min at room temperature. After washing with PBS, the cells were blocked with 10% goat serum in PBS containing 0.2% Triton X-100 for 1 hr at 37 °C, and incubated with mouse anti-HA (1:400) and rabbit anti-β-catenin antibody (1:400) overnight at 4 °C. After incubation with secondary antibodies (Alexa 488-conjugated goat anti-rabbit IgG and Alexa 555-conjugated goat antimouse IgG, 1:1000, Invitrogen, Carlsbad, CA, USA) plus Hoechst 33342 at room temperature, the cells were washed with PBS, mounted with SlowFade® Gold Antifade Mountant (Invitrogen, Carlsbad, CA, USA), and visualized with a TCS-SP5 dual photon laser-scanning confocal microscope (Leica, Bensheim, Germany). We used the Image J software to quantify the fluorescence intensity of the nucleus and whole cell. The ratio of fluorescence level in the nucleus and the cytoplasm (total-nucleus) was calculated and presented as average of the cells from 4-5 fields at random.

Stereotactic injection
Mice were deeply anesthetized with 1.25% Avertin (Sigma, St. Louis, MO, USA) and placed in a stereotaxic apparatus (Stoelting, Wood Dale, IL, USA). After craniotomy, ARA014418 (2 μl of 5 mM dissolved in DMSO) was unilaterally injected into lateral ventricle of 6-month-old Tg/hTau mice using a 10 μl Hamilton syringe custom made with a 30-gauge/0.5 inch/ hypodermic needle (Hamilton Syringe Co., Reno, NV, USA). The coordinates were as follows: 0.3 mm posterior, 1.0 mm lateral to bregma, and 2.5 mm ventral to the dura surface. ARA014418 was injected at a rate of 0.66 μl/min, and the needle was kept in position for 3 min additionally before slow withdrawal to prevent leakage of the liquid infused. DMSO was injected into Tg/hTau mice of the same age by using the same approach as a vehicle control. The mice were allowed to completely recover on a soft heating pad before they were returned to their home cages. 48 hr after injection, mice were sacrificed by cervical dislocation. The cortex of mouse was homogenized as described above for Western blots and mRNA assay.

Statistical analysis
The GraphPad Prism 5.0 software package was used for Statistical Analysis. The data are presented as mean ± SD. For multiple-group analysis, data points were compared by one-way ANOVA with Bonferroni post-hoc test. For twogroup comparison, data points were compared by the unpaired two tailed Student's t test.

AUTHOR CONTRIBUTIONS
N.J., R.S., Y.J. and D.C. performed experiments and analyzed results. C.X.G. and K.I. discussed results and edited manuscript. N.J. and F.L. designed experiments, analyzed and interpreted results, and wrote the manuscript.

CONFLICTS OF INTEREST
The authors declare that they have no conflicts of interest with the contents of this article.

FUNDING
This work was supported in part by funds from Nantong University, New York State Office for People with Developmental Disabilities, the Neural Regeneration Co-innovation Center of Jiangsu Province, Jiangsu Government Scholarship for Oversea Studies and by grants from the National Natural Science Foundation of China (31500829) and U.S. Alzheimer's Association (DSAD-15-363172).