The Low Density Lipoprotein Receptor-related Protein 6 Interacts with Glycogen Synthase Kinase 3 and Attenuates Activity*

Glycogen synthase kinase 3 (GSK3) is a widely expressed Ser/Thr protein kinase that phosphorylates numerous substrates. This large number of substrates requires precise and specific regulation of GSK3 activity, which is achieved by a combination of phosphorylation, localization, and interactions with GSK3-binding proteins. Members of the Wnt canonical pathway have been shown to influence GSK3 activity. Through a yeast two-hybrid screen, we identified the Wnt canonical pathway co-receptor protein low density lipoprotein receptor-related protein 6 (LRP6) as a GSK3-binding protein. The interaction between the C terminus of LRP6 and GSK3 was also confirmed by in vitro GST pull-down assays and in situ coimmunoprecipitation assays. In vitro assays using immunoprecipitated proteins demonstrated that the C terminus of LRP6 significantly attenuated the activity of GSK3β. In situ, LRP6 significantly decreased GSK3β-mediated phosphorylation of tau at both primed and unprimed sites. Finally, it was also demonstrated that GSK3β phosphorylates the PPP(S/T)P motifs in the C terminus of LRP6. This is the first identification of a direct interaction between LRP6 and GSK3, which results in an attenuation of GSK3 activity.

Glycogen synthase kinase 3 (GSK3) 2 is a widely expressed protein kinase with high expression in the brain, and specifically within neurons (for a review, see Ref. 1). GSK3 is a unique Ser/Thr protein kinase that phosphorylates both primed (target Ser/Thr is 4 amino acids N-terminal to a prephosphorylated Ser/Thr) and unprimed (target Ser/Thr is flanked by a Pro) substrates (for a review, see Ref. 2). A screen of a rat brain cDNA library revealed that GSK3 is encoded by two independent genes, GSK3␣ and GSK3␤, with molecular masses of 51 and 47 kDa, respectively (3). The two genes display 85% overall sequence identity, which is even higher in the catalytic domain (93%). In the brain, although GSK3␣ mRNA level is higher than GSK3␤, GSK3␤ protein level is higher (4). The poor relationship between transcription and translation in some tissues indicates that these two isoforms are subject to differential regulation, but little is known about the isoform-specific functions.
More than 40 proteins have been reported to be phosphorylated by GSK3, including over a dozen transcription factors (reviewed in Ref. 2). This large number of substrates illustrates the great potential of GSK3 to affect many cellular functions and suggests that the activity of GSK3 must be carefully regulated by individual mechanisms for each substrate. Although the mechanisms regulating GSK3 are not fully understood, precise control appears to be achieved by a combination of phosphorylation, localization, and interactions with GSK3-binding proteins (reviewed in Ref. 2). Protein complexes that contain GSK3 are of major importance in regulating its actions. The best characterized of these complexes is involved in the Wnt canonical pathway, where GSK3binding proteins control access to the GSK3 substrate, ␤-catenin, and generate a high degree of specificity in regulating the actions of GSK3. In the absence of Wnt signal, adenomatous polyposis coli, ␤-catenin, GSK3, casein kinase I (5), and other proteins all bind to the scaffold protein axin to form a complex (6). In this complex, ␤-catenin is phosphorylated by casein kinase I and subsequently by GSK3, which allows the ␤-transducing repeat-containing protein, an F-box protein in the E3 ubiquitin ligase complex, to bind and tag ␤-catenin for proteasomemediated degradation (7,8). When specific Wnts, such as Wnt 3a and Wnt 8, bind to the appropriate Frizzled (Fz) receptors, Dishevelled is activated; this, in concert with the GSK3-binding protein, frequently rearranged in advanced T-cell lymphoma (FRAT), facilitates disruption of the axin-based complex. This decreases the phosphorylation of ␤-catenin and results in ␤-catenin accumulation and activation (for a review, see Ref. 9). This is a classical example of how GSK3-binding proteins can regulate the action of GSK3 toward individual substrates. In addition to the Fz receptors, Wnt canonical signaling requires single span transmembrane proteins that belong to a subfamily of low density lipoprotein (LDL) receptor-related proteins (LRPs): vertebrate LRP5 and -6 and their Drosophila ortholog Arrow (10 -12). Unlike Fz, which is required for multiple Wnt pathways (13,14), Arrow and LRP5/6 appear to be specifically required for Wnt/␤-catenin signaling (12,15). Although the mechanism for LRP5/6 involvement in Wnt signaling has not been fully elucidated, recent studies suggest that the C terminus of LRP5/6 binds to axin and localizes axin and other bound proteins of the destruction complex to the plasma membrane. This is followed by axin degradation and dispersal of proteins of the destruction complex (16,17). Furthermore, a study demonstrated that a single PPP(S/T)P motif, which is reiterated five times in the LRP5/6/Arrow intracellular domain, can fully activate the Wnt pathway (16). Wnt signaling stimulates and requires phosphorylation of the PPP(S/T)P motif (16), although it has not been determined which kinases and/or phosphatases are involved in this modification. Although this is the prevailing model for LRP5/6/Arrow function, it does not explain the observation that the intracellular domain alone of LRP5/6 can activate and potentiate the Wnt 3a-induced LEF-1 signal, although it is not membrane-anchored (18).
Members of the Wnt canonical pathway have also been shown to influence GSK3 activity toward substrates outside of the pathway, with the microtubule-associated protein tau being a notable example (reviewed in Ref. 2). Tau is phosphorylated by GSK3 at both unprimed and primed motifs. It has been demonstrated that primed phosphorylation of tau at Thr 231 by GSK3 plays an essential role in decreasing the ability of tau to both bind and stabilize microtubules (19). In contrast, GSK3-mediated phosphorylation of unprimed sites on tau (e.g. Ser 396/ 404) may be more pathological and contribute to the formation of filamentous inclusions in Alzheimer disease and other neurodegenerative disorders (20). Previously, our group demonstrated that axin potently inhibits tau phosphorylation by GSK3␤, probably by sequestering GSK3␤ away from tau (21). Dishevelled has also been shown to decrease tau phosphorylation by GSK3␤ (22), despite the fact that Dishevelled and GSK3␤ do not interact (8). Interestingly, studies showed that FRAT-1 inhibits GSK3-mediated tau phosphorylation (23). Further, antisense-induced knock-down of Dickkopf-1, a negative modulator of the Wnt pathway that functions through binding to the co-receptor LRP6 (24,25), attenuates the increased tau phosphorylation in ␤-amyloid peptide-treated neurons (26). However, whether LRP6 directly influences GSK3 activity has not been elucidated.
Here we identify LRP6 as a GSK3-binding protein through a yeast two-hybrid assay. The interaction between LRP6 and GSK3 was confirmed by an in vitro GST pull-down assay and coimmunoprecipitation assays. The interaction of LRP6 with GSK3 significantly attenuated the activity of GSK3, and LRP6 was found to be a GSK3 substrate. Overall, this study suggests that LRP6 plays a unique and important role in and out of Wnt canonical signaling through regulating GSK3 activity.
Constructs and Recombinant Protein Preparation-The tau construct that does not contain exons 2 and 3 (T4) was described previously (27). Human GSK3␤-HA was a gift from Dr. J. R. Woodgett. The plasmid encoding glutathione S-transferase (GST) fusion ␤-catenin in a bacterial expression vector was a gift from Dr. W. I. Weis (27). To make the GSK3␣-HA construct, human brain mRNA (Clontech) was used as the template for RT-PCR of GSK3␣ (Superscript TM one-step reverse transcription-PCR (Invitrogen)). The following primers were used in the reaction: forward, 5Ј-GCG AAT TCC GTT ATG AGC GGC GGC GGG CCT TCG-3Ј; reverse, 5Ј-GCG AAT TCC AGC ACA CTG GCG GCC GTA-3Ј. The PCR product was digested with EcoRI and subcloned into the same vector as GSK3␤-HA. The cytosolic fragment of the C terminus of human LRP6 (LRP6-C3) was described previously (18). Using LRP6-C3 in pCMV5A (18) as a template, PCR was performed to subclone the LRP6-C3 fragment into corresponding vectors. The following primers were used for each PCR: forward (5Ј-CGC GGA TCC AGC ATG GGA CCA GCT TCT GTG CCT CTT GGT TAT GTG-3Ј) and reverse (5Ј-GCG CGT CGA CTT CTA GGA GGA GTC TGT ACA GGG AGA GGG TGG CGG TGG GT-3Ј) for subcloning LRP6-C3 into pGEX-6P-2 (Amersham Biosciences) (GST-LRP6-C3); forward (5Ј-CCG GAA TTC GGG ACC AGC TTC TGT GCC TCT TGG TTA TGT-3Ј) and reverse (5Ј-GGG ATC CTC AGG AGG AGT CTG TAC AGG GAG AG-3Ј) for subcloning LRP6-C3 into pEGFP-C1 (BD Biosciences) (GFP-LRP6-C3). The PCR products were digested and ligated into the corresponding cloning sites in vectors. For subcloning LRP6-C3 into pGBD-C1 vector (Clontech), LRP6-C3 was cut out from LRP6-C3 in pCMV5A by BamHI and SalI and ligated into the corresponding cloning sites in the vector. We also mutated all five PPP(S/T)P motifs in GST-LRP6-C3 to PPPAP to make GST-mut-LRP6-C3. The mutatgenesis reactions were carried out using either the GeneEditor TM in vitro site-directed mutagenesis system from Promega or the QuikChange TM site-directed mutagenesis kit from Stratagene. Plasmid integrity was verified using DNA sequencing analysis. GST-LRP6-C3, GST-mut-LRP6-C3, and GST-␤-catenin were purified according to the manufacturer's protocol, with the exception that cleared bacterial lysates were incubated with preblocked Fast Flow Glutathione-Sepharose beads (Amersham Biosciences) for 1 h with rotation at 4°C. For controls, GST was prepared in the same manner. For functional studies, GST-LRP6-C3 was treated on column with 60 units of Prescission protease (Amersham Biosciences) in cleavage buffer (50 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, pH 7.0) overnight at 4°C to remove the GST, and the LRP6-C3 was eluted. Where indicated, the fusion protein was left intact and eluted with glutathione buffer (20 mM Tris, 10 mM reduced glutathione, pH 8). After elution, GST-␤-catenin, GST-LRP6-C3, GST-mut-LRP6-C3, and LRP6-C3 were dialyzed into phosphate-buffered saline. To determine the concentration of the purified protein, the samples were diluted in 2ϫ SDS stop buffer (0.25 M Tris-HCl, pH 7.5, 2% SDS, 25 mM dithiothreitol, 5 mM EDTA, 5 mM EGTA, 10% glycerol, and 0.01% bromphenol blue) and electrophoresed on an 8% SDS-polyacrylamide gel, in addition to aliquots containing known amounts of bovine serum albumin (BSA) (Fisher). The gel was Coomassie-stained, destained, and dried. The gel was scanned and quantitated using UNSCANIT software (Silk Scientific, Inc.), and the amount of recombinant protein in the sample was determined as a function of the BSA standards.
Yeast Two-hybrid Screening-The MATCHMAKER yeast two-hybrid system and an adult human brain cDNA library were purchased from Clontech. The bait for library screening was the intracellular domain of human LRP6 (residues 1416 -1613), which was fused to the GAL4 DNA binding domain by subcloning into the pGBD-C1 vector. The adult human brain cDNA library (Clontech) in the pGAD-C2 vector, which has a transcription activation domain, was used in the screen. Prior to the screening, the LRP6-C3 construct in the pGBD-C1 vector was first transformed into the Y190 yeast hosts (Clontech), and subsequently the yeast Y190 cells carrying LRP6-C3 were transformed with the human brain cDNA library using the lithium acetate/singlestranded carrier DNA/polyethylene glycol protocol (28). A small fraction of the transformation reaction was plated on synthetic dropout medium ϪLeu, ϪTrp, to estimate the total number of transformants. The remainder was plated on synthetic dropout medium ϪLeu, ϪTrp, ϪAde to identify ADE2 reporter gene-positive colonies, which were subsequently transferred to synthetic dropout medium ϪLeu, ϪTrp, ϪHis plates (Teknova) for further selection.
Transient Transfections and Collection-CHO cells were transiently transfected using FuGENE 6 transfection reagent (Roche Applied Science) according to the manufacturer's protocol. Forty-eight hours after transfection, cells were harvested in lysis buffer (0.5% Nonidet P-40, 150 mM NaCl, 10 mM Tris-Cl (pH 7.4), 1 mM EGTA, and 1 mM EDTA, 0.1 M okadaic acid, 0.1 mM phenylmethylsulfonyl fluoride, 10 g/ml each of aprotinin, leupeptin, and pepstatin). After sonication on ice for 10 s, cellular debris was removed by centrifugation. Protein concentrations of supernatants were then determined by the bicinchoninic acid assay (Pierce).
Immunoblotting, Pull-down Assays, and Immunoprecipitation-Cells were collected as described above. After determination of protein LRP6 Binds to GSK3 and Inhibits Activity concentrations, samples were diluted in 2ϫ SDS buffer and electrophoresed on 8% SDS-polyacrylamide gels and transferred to nitrocellulose membranes. Nitrocellulose membranes were blocked for 1 h with 5% nonfat dry milk in TBST (20 mM Tris-HCl, pH 7.6, 137 mM NaCl, 0.05% Tween 20) and then probed with antibody overnight at 4°C. Blots were then rinsed with TBST, incubated with the appropriate horseradish peroxidase-conjugated secondary antibody (Jackson Immunoresearch) for 1 h at room temperature, rinsed with TBST, and developed with enhanced chemiluminescence (Amersham Biosciences). To evaluate the phosphorylation state of tau, the following antibodies were used: AT180 (Pierce/Endogen), which recognizes tau phosphorylated on Thr 231 (29,30) (numbering based on the longest human tau isoform (31)) and PHF-1 (a gift from Dr. P. Davies), which recognizes tau phosphorylated on Ser 396/404 (32,33). Total tau levels were determined using a combination of the phospho-independent monoclonal tau antibodies Tau 5 (a gift from Dr. L. Binder) (34) and 5A6 (35). Levels of transfected GSK3 and GFP-LRP6-C3 were detected using monoclonal HA (Transduction Laboratories) and monoclonal GFP (Roche Applied Science) antibodies, respectively.
The in vitro pull-down assay was carried out as described previously (36). Equal amounts of GST-or GST-LRP6-C3-conjugated glutathione beads were incubated for 2 h with lysates from CHO cells transiently transfected with GSK3 and washed eight times with phosphate-buffered saline containing 350 mM NaCl (high salt) and 0.2% Triton X-100 prior to immunodetection by Western blotting. Alternatively, GST-or GST-LRP6-C3-conjugated beads were incubated with 200 ng of recombinant GSK3␤ (Upstate Biotechnology, Inc., Lake Placid, NY) and washed in the same manner.
For immunoprecipitation, 100 g of the total lysate was diluted to a final concentration of 0.5 mg/ml and subjected to immunoprecipitation at 4°C with preconjugated monoclonal anti-GFP antibody (Roche Applied Science) M-280 sheep anti-mouse magnetic IgG beads (Dynal Biotech) for 2 h on a rotational shaker. Precipitates were washed at least five times and boiled in 2ϫ SDS stop buffer. The supernatants were collected, resolved on 12% SDS-polyacrylamide gels, and blotted as described above. Five micrograms of total protein of each condition were probed to check expression levels.
GSK3 Activity Assays-Cell lysates were incubated for 3 h at 4°C with magnetic IgG beads, which were precoupled with 1.5 g of the monoclonal anti-GSK3␤ antibody (BD Biosciences) (to immunoprecipitate endogenous GSK3␤) or with 1 g of the monoclonal anti-GFP antibody (to immunoprecipitate GFP-LRP6-C3 and the associated GSK3␤). After washing IgG beads containing the immunomobilized GSK3␤ four times with phosphate-buffered saline containing 350 mM NaCl (high salt) and 0.5% Nonidet P-40, the beads were washed two additional times with GSK3 kinase buffer (20 mM Tris-Cl, pH 7.5, 5 mM MgCl 2 , and 1 mM dithiothreitol). GSK3␤ activity assays were performed as described using phosphoglycogen synthase peptide 2 or recombinant tau as substrates (37) with the exception that GSK3␤ activity was normalized to GSK␤ levels in the sample, and the data were expressed as percentages of control (GFP alone). The same immunoprecipitation was performed to detect phospho-Ser 9 GSK3␤ (Cell Signaling) in each condition.
Recombinant in Vitro GSK3␤ Activity Assays-GST-␤-catenin, purified as described above, was bound to washed glutathione beads for 4 h and washed twice with GSK3 kinase buffer. Sixty units of recombinant GSK3␤ (New England Biolabs) was preincubated with or without LRP6-C3 (cleaved from GST as described above) in the presence of 1 mg/ml BSA in GSK3 kinase buffer for 30 min at 4°C. Preliminary experiments were conducted to determine optimal conditions for functional assay readouts, and a 40-fold molar excess of LRP6-C3 over GSK3␤ was used for all experiments. Following preincubation, reaction components were added to final concentrations of 125 M ATP, 0.182 Ci/l [␥-32 P]ATP, and 1 mg/ml BSA in GSK3 kinase buffer. Washed GST-␤catenin bound to beads was then incubated in the reaction mix for 30 min at 30°C. After removing the reaction supernatant and washing the beads twice with GSK3 kinase buffer, the beads were boiled with 30 l of 2ϫ SDS buffer, and the supernatants were resolved on an 8% polyacrylamide gel. The gels were dried and analyzed by phosphorimaging (Storm 840 (Amersham Biosciences) and ImageQuant software). GST-LRP6-C3 and GST-mut-LRP6-C3 in vitro phosphorylation by recombinant GSK3␤ was performed in a similar manner except that the purified GST-LRP6-C3 or GST-mut-LRP6-C3 was incubated with recombinant GSK3␤ alone without binding to the glutathione beads and without the addition of the BSA. For comparison, recombinant tau was incubated with recombinant GSK3␤ under identical conditions. Data Analysis-Data were analyzed using analysis of variance between individual groups. Values were considered significantly different when p was Ͻ0.05. Results were expressed as mean Ϯ S.E.

RESULTS
The C Terminus of LRP6 Directly Interacts with GSK3-Previous studies have shown that the Wnt canonical pathway co-receptors LRP5 and -6 can bind to axin and recruit it to the membrane, resulting in the degradation of axin and propagation of the Wnt signal (16,17). Although this is one mechanism by which LRP5/6 facilitates canonical Wnt signaling, our previous study showed that membrane targeting was not essential for the C terminus of LRP5/6 to constitutively activate the Wnt canonical pathway and synergistically increase LEF-1 activation in response to Wnt 3a (18). These findings suggested that LRP5/6 can facilitate the propagation of the Wnt signal through mechanisms other than recruitment of axin to the membrane. To determine which other proteins may be involved in the LRP5/6 signaling cascades, a yeast twohybrid screen was carried out by using the intracellular domain of human LRP6 (LRP6-C3) as the bait, which was described previously (18). An adult human brain cDNA library was used in the screen. We screened ϳ2.5 million clones and identified four clones as GSK3 by nucleotide sequencing analyses. GSK3 interacted strongly with LRP6-C3 in yeast (Fig. 1). We also identified nine other unique clones, some of which are in the Wnt canonical pathway. However, none of them were axin. To confirm the interaction between LRP6-C3 and GSK3 in vitro, we made a GST-fused LRP6-C3 construct. Purified GST  FEBRUARY 24, 2006 • VOLUME 281 • NUMBER 8 and a GST-LRP6-C3 fusion protein were conjugated to glutathione beads. CHO cells were transfected with GSK3␣ or GSK3␤, and lysates from the transfected cells were incubated with GST-or GST-LRP6-C3conjugated beads. Both GSK3␣ and GSK3␤ were specifically pulled down by LRP6-C3 in vitro (Fig. 2). To rule out an indirect interaction between LRP6-C3 and GSK3, a pull-down assay was performed using recombinant GSK3␤ and recombinant GST-LRP6-C3 (Fig. 2C), which confirmed a direct protein-protein interaction. Further co-immunoprecipitation assays also revealed that LRP6-C3 interacts directly with both GSK3␣ and GSK3␤ (Fig. 3).

LRP6 Binds to GSK3 and Inhibits Activity
LRP6-C3 Decreases GSK3␤ Phosphorylation of a Primed Peptide and Recombinant Tau in Vitro-Since LRP6-C3 interacts with GSK3␤, the effect of this interaction on the activity of GSK3␤ was subsequently evaluated. In this assay, endogenous GSK3␤ was used, because CHO cells express high levels of GSK3␤ and this approach reduced the number of variables in the assay (Fig. 4A). To immunoprecipitate similar amounts of GSK3␤, 30 g of the lysate was used when GSK3␤ was immunoprecipitated directly, and 150 g of the lysate was used when GSK3␤ was co-immunoprecipitated by using the anti-GFP antibody to bring down LRP6-C3 (Fig. 4B). GSK3␤ was immunoprecipitated from GFP-or GFP-LRP6-C3-transfected cell lysates with a monoclonal GSK3␤ antibody and used to 32 P-phosphorylate substrates in vitro. Fractionation and cytochemical analyses revealed that GFP-LRP6-C3 was present in both the cytosol and nucleus (data not shown). When GSK3␤ was immunoprecipitated from CHO cell lysates and used in the assay, there was no significant difference in GSK3␤ activity between the lysates from GFP-and GFP-LRP6-C3-transfected cells (Fig. 4C). This is probably due to the fact that only a small fraction of the GSK3␤ immunoprecipitated was bound by LRP6-C3. Indeed, when the GSK3␤ immunoprecipitates were probed for LRP6-C3, none could be detected at a normal exposure (Fig. 4B), and it was only visible when the blots were significantly overexposed (Fig. 4B). Therefore, to determine the effects of LRP6-C3 on GSK3␤ activity, GFP-LRP6-C3 was immunoprecipitated with the GFP antibody, and GSK3␤ activity was measured in the precipitates. When the values were normalized to the levels of GSK3␤ in the immunoprecipitates, LRP6-C3 significantly inhibited GSK3␤ activity toward the primed peptide substrate (Fig. 4C). Assays were also carried out using recombinant tau as a substrate. A representative phosphorylated recombinant tau autoradiograph is shown in Fig.  4D, and the corresponding quantitated data from three separate experiments are shown in Fig. 4E, demonstrating that LRP6-C3 also attenuates GSK3␤ phosphorylation of an unprimed substrate. These data clearly demonstrate that the C terminus of LRP6 inhibits GSK3␤ activity. Interestingly, only a small amount of the GSK3␤ that co-precipitated with LRP6-C3 was phosphorylated on Ser 9 , compared with the phospho-Ser 9 immunoreactivity that was present in the fractions in which GSK3␤ was immunoprecipitated directly (Fig. 4B), indicating that LRP6 preferentially binds to the dephosphorylated, more active form of GSK3␤. It should be noted that the levels of phospho-Ser 9 GSK3␤ (Ser 9 ) were equivalent in the lysates from GFP-and GFP-LRP6-C3-transfected cells (Fig. 4A).
Recombinant LRP6-C3 Is an in Vitro Substrate of GSK3␤-In a previous study (16), it was revealed that the C terminus of LRP6 was phosphorylated by a hitherto unidentified kinase at the PPP(S/T)P motif, which is reiterated five times. To determine whether GSK3␤ can phosphorylate the C terminus of LRP6, GST-LRP6-C3 was incubated in the presence of recombinant GSK3␤ and [ 32 P]ATP (Fig. 5). These data demonstrate that LRP6-C3 is efficiently phosphorylated by GSK3␤ in vitro, comparable with tau. Under the same conditions, casein kinase I␦ did not phosphorylate LRP6-C3 (data not shown). Previous studies have shown that casein kinase I␣ phosphorylates ␤-catenin (38). Further, to determine whether GSK3␤ specifically phosphorylates the PPP(S/T)P motifs, we substituted all of the Ser/Thr in the PPP(S/T)P motifs to Ala (GST-mut-LRP6-C3), and the GST-mut-LRP6-C3 was incubated in the presence of recombinant GSK3␤ and [ 32 P]ATP under identical conditions (Fig. 5). The results show that the mut-LRP6-C3 is not phosphorylated by GSK3␤, which strongly indicates that GSK3␤ is the kinase that specifically phosphorylates the Ser/Thr in the motif.

LRP6-C3 Decreases GSK3␤ Phosphorylation of Recombinant ␤-Catenin in Vitro-
In order to test the effects of the GSK3␤/LRP6-C3 interaction on GSK3 activity toward the Wnt pathway substrate ␤-catenin, in vitro phosphorylation assays were performed using recombinant ␤-catenin in the absence and presence of LRP6-C3 (Fig. 6). The addition of recombinant LRP6-C3 to the assay significantly attenuated GSK3␤mediated phosphorylation of ␤-catenin. These data are in agreement with our previous in situ findings that overexpression of LRP6-C3 in cells results in a decrease in phospho-␤-catenin levels (18).
The C Terminus of LRP6 Inhibits GSK3␤-mediated Phosphorylation of Tau at both Primed and Unprimed Sites in Situ-Previous studies have provided evidence that GSK3-binding proteins, such as FRAT-1 and FRAT-2, which are the proteins involved in the Wnt canonical pathway, affect GSK3␤-mediated tau phosphorylation (23,37,39). Since LRP6 is a GSK3-binding protein that attenuates its activity in in vitro assays, we next examined the effects of LRP6 on GSK3-mediated  A, CHO cells were transiently transfected with GFP or GFP-LRP6-C3, and cell lysates (5 g) were immunoblotted with the GSK3␤ antibody to detect endogenous GSK3␤, a Ser(P) 9 -GSK3␤ antibody to determine the levels of Ser 9 -phosphorylated GSK3␤ or the GFP antibody to determine the expression of the transfected constructs. B, cell lysates were immunoprecipitated (IP) with either the GSK3␤ antibody to directly precipitate the endogenous GSK3␤ or with the GFP antibody to immunoprecipitate LRP6-C3. The immunoprecipitates were then used in kinase assays. In order to get similar amounts of GSK3␤ in the immunoprecipitates, 30 g of lysate was used when GSK3␤ was immunoprecipitated directly, and 150 g of lysate was used when GFP-LRP6-C3 was immunoprecipitated and GSK3␤ was a co-immunoprecipitating protein. Representative immunoblots show that the levels of GSK3␤ co-immunoprecipitated with LRP6-C3 and the immunoprecipitated GSK3␤ are similar. LRP6-C3 prefers binding to the unphosphorylated GSK3␤, since the levels of phospho-Ser 9 immunoreactivity are low in the immunoprecipitates. The overexposed blot of GFP-LRP6-C3 shows that when GSK3␤ is immunoprecipitated, GFP-LRP-C3 does co-precipitate, although proportionally only a small fraction of the GSK3␤ is interacting with the GFP-LRP-C3. When GFP-LRP6-C3 was immunoprecipitated and therefore all of the GSK3␤ in the assay was bound to LRP6-C3, GSK3␤ phosphorylated both the primed peptide (C) and unprimed recombinant tau (E) to a significantly lesser extent compared with when GSK3␤ was immunoprecipitated directly. Data were normalized to immunoprecipitated GSK3␤ levels and expressed as a percentage of the values obtained with immunoprecipitated GSK3␤ from GFP lysates. Data are present as mean Ϯ S.E. of three separate experiments. **, p Ͻ 0.001 compared with GSK3␤ from GFP lysates. D, representative autoradiograph of phosphorylated recombinant tau protein. FEBRUARY 24, 2006 • VOLUME 281 • NUMBER 8 tau phosphorylation in situ. CHO cells were transiently transfected with tau in the absence or presence of GSK3␤ and in the presence or absence of GFP or GFP-LRP6-C3, and tau phosphorylation was examined by immunoblotting with phospho-specific antibodies (Fig. 7A). As expected, exogenous expression of GSK3␤ resulted in an increased phosphorylation at both the PHF-1 (phospho-Ser 396/404 ) and AT180 (phospho-Thr 231 ) epitopes. However, when GFP-LRP6-C3 was expressed with GSK3␤, GSK3␤ phosphorylation of tau at the unprimed PHF-1 epitope significantly decreased compared with what was observed with GSK3␤ alone, and GSK3␤ phosphorylation of tau at the primed AT180 epitope was completely abolished. GFP alone has no effect on GSK3␤-mediated tau phosphorylation (Fig. 7A). When GFP-LRP6-C3 was expressed alone with tau, there was a significant decrease in the phosphorylation state of tau at the unprimed PHF-1 epitope (Fig.  7B), which indicates that GFP-LRP6-C3 also inhibits the phosphorylation of tau by endogenous GSK3␤. GFP-LRP6-C3 also attenuated GSK3␣-mediated tau phosphorylation in a manner similar to what was observed for GSK3␤ (data not shown).

DISCUSSION
In this study, we show for the first time that the C terminus of LRP6 directly binds to GSK3 and that this interaction significantly attenuates the activity of GSK3. GSK3 is a key protein in the Wnt canonical signaling pathway and plays an essential role in regulating tau phosphorylation. Here we clearly demonstrate in vitro that the C terminus of LRP6 inhibits GSK3␤-mediated phosphorylation of ␤-catenin. Furthermore, in situ studies show that LRP6-C3 inhibits GSK3␤-mediated phosphorylation of both primed and unprimed sites on tau. Finally, we demonstrate that in vitro, GSK3␤ phosphorylates the PPP(S/T)P motifs in LRP6. These are novel and important findings, since they provide insight into the mechanisms by which LRP6 may regulate the phosphorylation of specific GSK3␤ substrates in and out of the Wnt canonical pathway.
Wnt signaling through ␤-catenin has been studied extensively; however, much still remains unknown about this pathway, especially at the cell membrane level. The membrane proteins LRP5/6 are required coreceptors in the Wnt canonical pathway (10 -12). LRP5/6 are the only two proteins from the low density lipoprotein receptor family that have been shown to facilitate signaling through the Wnt canonical pathway, although a minireceptor of LRP1, another member of the LDL receptor family that interacts with the human Fz-1, has been shown to repress Wnt 3a signaling by disrupting the Fz receptor-LRP5 and -6 coreceptor complex (40). It is clear that the main function of canonical Wnts is to nucleate the formation of a physical complex between receptor and coreceptor (11); this was confirmed in a recent study by using a Dickkopf-1-Fz5 fusion protein (41). However, how this membrane complex transduces the signal into the cell is still not completely clear. Studies showed that the intracellular domain of LRP5 and -6 interacts with axin and that the addition of Wnt appears to stimulate the recruitment of axin to LRP5 and -6 at the membrane. Once at the membrane, axin is degraded, which subsequently results in increased ␤-catenin levels and signaling (16,17). Further, it was demonstrated that the PPP(S/T)P motif in the intracellular domain of LRP5/6 is required for signaling (42) and that phosphorylation of this motif creates a docking site for axin (16). Recently, another group showed that FRAT-1 binds to the cytoplasmic domain of LRP5 and that the recruitment of axin and FRAT-1 to the membrane by LRP5 leads to both axin degradation and FRAT-1mediated inhibition of GSK3 (43). Here we clearly demonstrate that the C terminus of LRP6 also interacts directly with GSK3. This direct interaction inhibits GSK3-mediated phosphorylation of ␤-catenin. Although it is apparent that LRP6 can bind all three proteins (axin, FRAT-1, and GSK3), it is unlikely that they all exist in the same complex due to the fact that axin and FRAT-1 competitively bind GSK3 (44). Recently, it was shown that although FRAT is strictly required for maternal Wnt signaling in Xenopus, FRAT is not an essential component of the canonical Wnt pathway in higher organisms (45). These data suggest that FRAT-1 binding to LRP5 is not necessary for LRP5 function in the pathway. All of these studies have suggested that LRP5 and -6 can function in the Wnt canonical pathway through multiple mechanisms.
In the C terminus of LRP6, Ser and Thr represent nearly 30% of the amino acids. Here we show that GSK3␤, but not casein kinase I␦, phosphorylates the C terminus of LRP6 in an in vitro assay. However, the GST-mut-LRP6-C3 cannot be phosphorylated by GSK3␤, which FIGURE 6. Phosphorylation of ␤-catenin by recombinant GSK3␤ is attenuated by recombinant LRP6-C3. GSK3␤ preincubated in the absence or presence of 8 pmol of LRP6-C3 was used to phosphorylate GST-␤-catenin. Phosphor images were quantitated, and the incorporation of [ 32 P]ATP was normalized to total ␤-catenin protein levels as determined by Coomassie staining followed by scanning densitometry. Values are expressed as a ratio of these two values. Data are present as mean Ϯ S.E. of three experiments. *, p ϭ 0.02. FIGURE 7. LRP6-C3 inhibits phosphorylation of tau by GSK3␤ at both primed and unprimed sites in situ. A, representative immunoblots showing total tau levels (Tau5/ 5A6), and tau phosphorylated at the primed AT180 site (pThr 231 ) and the unprimed PHF-1 (pSer 396/404 ) site. CHO cells were transiently transfected with tau and the other constructs as indicated below the panels. Tau protein levels were equivalent in all conditions. Expression of exogenous GSK3␤ resulted in a decrease in the electrophoretic mobility of tau, indicative of increased phosphorylation, and phosphorylation at both the PHF-1 and AT180 epitopes increased as expected. The presence of GFP-LRP6-C3 significantly decreased the GSK3␤-mediated phosphorylation of tau at the PHF-1 epitope and completely abolished the GSK3␤-mediated phosphorylation of tau at the AT180 epitope. GFP alone had no effect on the GSK3␤-mediated phosphorylation of tau. The expression levels of transfected GFP, GFP-LRP6-C3, and GSK3␤ (by blotting for HA) are also shown. ␣-Tubulin was used as the loading control. B, a longer exposure of the PHF-1 blot shown in A to clarify the GFP-LRP6-C3-mediated inhibition of the phosphorylation of the 396/404 epitope of tau by endogenous GSK3␤. strongly suggests that GSK3␤ contributes to the phosphorylation of the PPP(S/T)P motifs of LRP6 to supply a binding site for axin in situ. Therefore, it can be speculated that the interaction of LRP6 with GSK3␤ and the subsequent attenuation of GSK3␤ activity in combination with phosphorylation of LRP6 is responsible for at least part of the function of LRP6 in the canonical pathway. It has recently been shown that the C termini of LRP5 and LRP6, which are not membrane-anchored, can efficiently activate the Wnt canonical pathway by decreasing the phospho-␤-catenin level and increasing LEF-1 activity in situ (18). It thus can be hypothesized that the binding of the cytosolic termini of LRP5 and -6 to GSK3 in the axin complex attenuates GSK3 phosphorylation of ␤-catenin, thus activating the Wnt canonical pathway without extracellular signaling. However, whether the C terminus of LRP6 alone is a physiologically relevant entity is currently unknown. It may occur in some pathological situations to function as an oncogene for constitutively activating the Wnt canonical pathway (46). It is also interesting to note that other low density lipoprotein receptor family members, such as LRP (47) and megalin (48), undergo ectodomain shedding followed by release of the C terminus. Moreover, whether and how other proteins from the axin complex can affect the interaction between LRP6 and GSK3, thus changing the activity of the Wnt signal, needs to be further studied.
Besides the proteins from the Wnt canonical pathway and its namesake substrate glycogen synthase, GSK3 can phosphorylate more than 40 proteins, including over a dozen transcription factors (2). GSK3 activity is precisely regulated by a combination of phosphorylation, localization, and interactions with other proteins. It should be noted that previous studies have shown that GSK3␤ activity can be regulated by its interaction with other proteins. For example, binding of the transcription factor p53 (49) or FRAT-2 (37) increases GSK3␤ activity, whereas binding by FRAT-1 attenuates activity (23,39). In this study, we show that LRP6 is another GSK3-binding protein that inhibits GSK3 phosphorylation of both primed and unprimed substrates, both in and out of the Wnt canonical pathway. It suggests that LRP6 generally attenuates GSK3 activity. In our study, the in situ inhibitory function of LRP6 on GSK3 activity is much stronger than in vitro, which highly suggests that other proteins from the complex may play a role in facilitating the interaction and/or function. It is also worth noting that LRP6-C3 preferentially binds to the active form of GSK3␤, which efficiently phosphorylates LRP6-C3 in the complex. Thus, the inhibitory function of LRP6-C3 on GSK3␤-mediated phosphorylation of ␤-catenin, tau, and glycogen synthase is through protein-protein interactions.
Tau is a well characterized substrate of GSK3 both in situ and in vitro. Up to 25 sites in tau have been shown to be modified by GSK3, including both unprimed and primed Ser/Thr-Pro motifs (2). Modulation of tau phosphorylation is an important regulating mechanism both during development and in the adult brain, and GSK3␤ may play a key role (50 -52). The importance of primed site (e.g. Thr 231 ) phosphorylation is in regulating tau-microtubule interactions (19), and the unprimed phosphorylation of tau at specific sites (e.g. Ser 396 /Ser 404 ) may be involved in the pathological formation of filaments (20), which occur in the tau-neurodegenerative diseases, such as Alzheimer disease. From our study, LRP6 inhibits GSK3-mediated phosphorylation of tau at both primed and unprimed sites, which indicates that LRP6 could modulate both physiological and pathological tau phosphorylation by GSK3␤.
In summary, we demonstrate for the first time that GSK3␤ phosphorylates the PPP(S/T)P motifs in LRP6 and that C terminus of LRP6 directly interacts with GSK3. Further, we show that this interaction inhibits GSK3␤ activity toward ␤-catenin, tau, and glycogen synthase in vitro. In situ GSK3-mediated phosphorylation of both primed and unprimed sites on tau is also decreased by the C terminus of LRP6. The LRP6-modulated decrease in the GSK3-mediated phosphorylation of ␤-catenin may be an essential step in the Wnt canonical signaling pathway. In the case of tau, the interaction of LRP6 with GSK3␤ may inhibit physiological phosphorylation events that regulate tau-microtubule interactions and attenuate phosphorylation at unprimed sites, which have the potential to be of more pathological significance. Overall, LRP6 plays a unique and important role in the regulation of GSK3␤ activity in and out of the Wnt canonical pathway.