Glycogen Synthase Kinase 3 (cid:1) Phosphorylates Tau at Both Primed and Unprimed Sites

Glycogen synthase kinase 3 (cid:1) (GSK3 (cid:1) ) phosphorylates substrates, including the microtubule-associated protein tau, at both primed and unprimed epitopes. GSK3 (cid:1) phosphorylation of tau negatively regulates tau-micro-tubule interactions; however the differential effects of phosphorylation at primed and unprimed epitopes on tau is unknown. To examine the phosphorylation of tau at primed and unprimed epitopes and how this impacts tau function, the R96A mutant of GSK3 (cid:1) was used, a mutation that prevents phosphorylation of substrates at primed sites. Both GSK3 (cid:1) and GSK3 (cid:1) -R96A phosphorylated tau efficiently in situ . However, expression of GSK3 (cid:1) -R96A resulted in significantly less phosphorylation of tau at primed sites compared with GSK3 (cid:1) . Con-versely, GSK3 (cid:1) -R96A phosphorylated unprimed tau sites to a significantly greater extent than GSK3 (cid:1) . Prephosphorylating tau with cdk5/p25 impaired the ability of GSK3 (cid:1) -R96A to phosphorylate tau, whereas GSK3 (cid:1) R96A phosphorylated recombinant tau to a significantly greater extent than GSK3 (cid:1) . Moreover, the amount of tau associated with microtubules was reduced by overexpression of GSK3 (cid:1) but only when tau was phosphorylated at primed sites, as phosphorylation of tau by GSK3 (cid:1) -R96A did not negatively regulate the association of tau with microtubules. These results demonstrate that GSK3 (cid:1) -mediated phosphorylation of tau at primed sites plays a more significant role in regulating the interaction of tau with microtubules than phosphorylation at unprimed epitopes. cells overexpressing HA-GSK3 (cid:1) or HA-GSK3-S9A AT180 immunoreactivity (primed site, phosphorylated Thr-231) was increased significantly; in contrast, expression of HA-GSK3 (cid:1) -R96A resulted in significantly less AT180 im- munoreactivity. Expression of HA-GSK3 (cid:1) or HA-GSK3-S9A also increased PHF-1 immunoreactivity (unprimed sites, phosphorylated Ser- 396/404), but expression of HA-GSK3 (cid:1) -R96A resulted in a much greater increase in tau phosphorylation at this epitope. Actin blots are included to show equal protein loading for all the samples. PHF-1 immunoreactivity was present only in the microtubule-bound fractions. total and GSK3 (cid:1) expression levels, as well as the state of tau, were evaluated in the high speed supernatants prior to use in the microtubule-binding assay. These data show that tau and the GSK3 (cid:1) constructs were all expressed at similar levels.

Glycogen synthase kinase 3␤ (GSK3␤) phosphorylates substrates, including the microtubule-associated protein tau, at both primed and unprimed epitopes. GSK3␤ phosphorylation of tau negatively regulates tau-microtubule interactions; however the differential effects of phosphorylation at primed and unprimed epitopes on tau is unknown. To examine the phosphorylation of tau at primed and unprimed epitopes and how this impacts tau function, the R96A mutant of GSK3␤ was used, a mutation that prevents phosphorylation of substrates at primed sites. Both GSK3␤ and GSK3␤-R96A phosphorylated tau efficiently in situ. However, expression of GSK3␤-R96A resulted in significantly less phosphorylation of tau at primed sites compared with GSK3␤. Conversely, GSK3␤-R96A phosphorylated unprimed tau sites to a significantly greater extent than GSK3␤. Prephosphorylating tau with cdk5/p25 impaired the ability of GSK3␤-R96A to phosphorylate tau, whereas GSK3␤-R96A phosphorylated recombinant tau to a significantly greater extent than GSK3␤. Moreover, the amount of tau associated with microtubules was reduced by overexpression of GSK3␤ but only when tau was phosphorylated at primed sites, as phosphorylation of tau by GSK3␤-R96A did not negatively regulate the association of tau with microtubules. These results demonstrate that GSK3␤-mediated phosphorylation of tau at primed sites plays a more significant role in regulating the interaction of tau with microtubules than phosphorylation at unprimed epitopes.
Tau is a family of microtubule-associated proteins that are expressed predominantly in neurons and are produced by alternative splicing of a single gene (1,2). Although tau likely plays a role in numerous neuronal processes, such as vesicle transport (3,4), microtubule-plasma membrane interactions (5,6), and the intracellular localization of protein such as fyn (7,8), 14-3-3 (9, 10), and protein phosphatase 2A (11,12), the best characterized function of tau is the binding and stabilization of microtubules (13)(14)(15). Tau is a phosphoprotein, and in vitro, numerous studies have demonstrated that site-specific phos-phorylation of tau regulates its ability to bind and stabilize microtubules. Increased phosphorylation of tau in situ tends to negatively regulate tau-microtubule interactions (16 -20), although the effects of site-specific tau phosphorylation on microtubule binding in the cell have not been as thoroughly studied, as they have been in vitro.
Numerous protein kinases phosphorylate tau in vitro; however in situ the list is much more abbreviated. Protein kinase A (21), microtubule affinity regulating kinases (22), cyclin-dependent protein kinase 5 (cdk5) 1 /p35[p25] (23,24), and glycogen synthase kinase 3␤ (GSK3␤) (25)(26)(27)(28)(29) have been shown to phosphorylate tau in situ. Although all these kinases can phosphorylate tau in situ, there is emerging evidence that GSK3␤ may play a major role in regulating tau phosphorylation both in physiological and pathological conditions. GSK3␤ is expressed ubiquitously and found at high levels in brain where it localizes predominantly to neurons (30). GSK3␤ is a critically important protein kinase in brain as it phosphorylates a number of important substrates including transcription factors such as ␤-catenin, cAMP-response element-binding protein, and cytoskeletal proteins such as MAP1B and tau, which regulates their function (31). Although numerous GSK3␤ substrates have been identified, the consensus sites recognized by GSK3␤ are not straightforward. GSK3␤ can phosphorylate unprimed sites at Ser/Thr-Pro motifs. Examples of unprimed substrates are axin and adenomatous polyposis coli gene product (APC) (32)(33)(34). In other cases the substrate must first be phosphorylated at a site that is four amino acids C-terminal to the target site, which primes the substrate for phosphorylation by GSK3␤. This primed motif ((S/T)XXX(S/ T)[P]) often occurs in proline-rich regions of the substrate (32,35). Glycogen synthase (36), eukaryotic protein synthesis initiation factor 2B (37), and ␤-catenin (38) are all primed substrates. Indeed, it appears as if the majority of GSK3␤ substrates are actually primed (35). Interestingly, although GSK3␤ phosphorylates tau at unprimed sites (39,40), prior phosphorylation by cdk5/p25[p25] enhances the phosphorylation of tau at specific sites (e.g. Thr-231) (41,42), indicating that tau may be both a primed and unprimed substrate of GSK3␤.
GSK3␤ activity is regulated by its phosphorylation state. Phosphorylation of GSK3␤ on Tyr-216 increases its activity (43), and phosphorylation at Ser-9 inhibits kinase activity (44,45), and both these events are regulated by growth factor signaling (31). In response to insulin or insulin-like growth factor, Akt is activated, and this kinase phosphorylates GSK3␤ on Ser-9 decreasing its activity (46). Recently, it was demon-* This work was supported by National Institutes of Health Grant NS35060 and by a grant from the Alzheimer's Disease Association. 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  strated that phosphorylation of Ser-9 results in the N-terminal of GSK3␤ acting as a pseudosubstrate, occupying the same binding site used by primed substrates (32,44,47). It was also found that Arg-96 in GSK3␤ is critically important in forming the binding pocket for the phosphate group of the primed substrates (or pseudosubstrate) (32,44,47). Mutation of Arg-96 to Ala decreased the ability of GSK3␤ to phosphorylate primed substrates significantly but had no effect on its ability to phosphorylate unprimed sites (32).
Although there are data to suggest that tau is both a primed and unprimed substrate of GSK3␤ in situ, this has not been demonstrated clearly. Further, how phosphorylation of tau by GSK3␤ on different sites modulates its association with microtubules has not been explored thoroughly. By using wild type GSK3␤ and GSK3␤ with the R96A or S9A mutation we demonstrate that GSK3␤ phosphorylates tau at both primed sites (pThr-231) and unprimed sites (pSer-396/404). Disruption of the GSK3␤ phosphate-binding pocket (R96A mutation) results in a significant decrease in tau phosphorylation at the primed site of Thr-231 and an enhancement of phosphorylation at the unprimed site of Ser-396/404. Further, GSK3␤-R96A phosphorylates recombinant tau to a significantly greater extent than wild type or GSK3␤-S9A; however if tau is first phosphorylated by cdk5/p25 then GSK3␤-R96A is significantly less effective than wild type or GSK3␤-S9A in phosphorylating tau. Finally, GSK3␤-mediated phosphorylation of tau at only the primed sites decreases tau-microtubule interactions.
Plasmid Constructs-The wild type tau (4R-Tau; ϩexon 10, Ϫexons 2 and 3) construct was a generous gift from Dr. M. Hutton (Mayo Clinic, Jacksonville, FL). 4R-Tau was subcloned into mammalian expression vector, pCDNA 3.1(Ϫ) (Invitrogen). By using the wild type tau construct as a template, PCR was performed (forward primer, 5Ј-GCC GAA TTC TAT GGC TGA GCC CCG CCA G-3Ј; reverse primer, 5Ј-CGG GGT ACC TCA CAA ACC CTG CTT GGC-3Ј), and the PCR reaction product was digested with EcoRI and KpnI and ligated into the EcoRI and KpnI sites of the pCDNA 3.1(Ϫ). The integrity of the 4R-tau construct was confirmed by sequence analysis.
Transient Transfections-4R-Tau, wild type HA-GSK3␤, HA-GSK3␤-S9A, or HA-GSK␤-R96A were transiently transfected into CHO cells using FuGENE 6 (Roche Molecular Biochemicals) transfection reagent according to the manufacturer's protocol. 36 h after transfection, the cells were washed with ice-cold phosphate-buffered saline and then collected and processed as described below for the different assays.
Immunoblotting-Cells were collected in lysis buffer (150 mM NaCl, 10 mM Tris-HCl, 1 mM EGTA, 1 mM EDTA, 0.2 mM sodium vanadate, 0.5% Nonidet P-40), containing 1 mM phenylmethylsulfonyl fluoride, 0.1 M okadaic acid, and a 10 g/ml concentration each of aprotinin, leupeptin, and pepstatin. Lysates were sonicated on ice and centrifuged, and protein concentrations in the supernatants were determined using the bicinchoninic acid assay (Pierce). Equal amounts of protein from each sample were electrophoresed on 10% SDS-polyacrylamide gels, transferred to nitrocellulose, and probed with the indicated antibodies. The tau antibodies used in this study were Tau5/5A6 (Tau5 from Dr. L. Binder), which are phospho-independent tau antibodies (49 -51), AT180 (Innogenetics), which recognizes tau when it is phosphorylated at Thr-231 (52,53), and PHF-1 (from Dr. P. Davies), which recognizes tau phosphorylated at Ser-396/404 (54). The monoclonal GSK3␤ was purchased from Transduction Laboratories, the monoclonal cdk5 antibody was from Santa Cruz Biotechnology, the monoclonal ␤-tubulin antibody was from Sigma, the monoclonal ␤-actin antibody was from Chemicon, and the monoclonal HA antibody was from Roche Molecular Biochemicals. After incubation with the appropriate horseradish peroxidaseconjugated secondary antibody (Jackson ImmunoResearch Laboratories), the blots were developed using enhanced chemiluminescence (Amersham Biosciences).
Preparation of Primed Recombinant Tau-To prephosphorylate (prime) tau to use as a GSK3␤ substrate, cdk5 and p25 were coexpressed in CHO cells, and cdk5 was immunoprecipitated. 120 g of the protein extract (1 g/l) was incubated with 2 l of rabbit polyclonal anti-cdk5 antibody (Santa Cruz Biotechnology) overnight at 4°C with gentle agitation. Extracts were then incubated with 40 l of washed protein A-Sepharose (Sigma) for 2 h at 4°C. The phosphorylation reaction was carried out by mixing immunoprecipitated cdk5/p25 with 40 l of kinase buffer containing 20 mM Tris, pH 7.5, 5 mM MgCl 2 , 1 mM dithiothreitol, 250 M ATP, and 0.2 g/l recombinant tau protein (Takara/Panvera, Madison, WI). For the negative control, the reaction was performed without ATP in the kinase buffer. The samples were incubated at 30°C for 30 min with shaking and then spun to pellet the cdk5/p25 immune complex. The supernatants containing the recombinant tau were collected and dialyzed to remove free ATP using Slide-A-Lyzer mini dialysis units (Pierce). Following dialysis, protein concentrations were determined by using the bicinchoninic acid assay. The supernatants were immunoblotted for cdk5 to confirm that there was no cdk5 present.
Measurement of GSK3␤ Activity-To immunoprecipitate the HA-GSK3␤ constructs, 70 g of protein from cellular extracts (1 g/l) prepared in lysis buffer was incubated with 2 l of mouse monoclonal HA antibody overnight at 4°C with gentle agitation. Extracts were then incubated with 30 l of washed protein G-Sepharose (Sigma) for 2 h at 4°C. The immobilized immune complexes were washed twice with lysis buffer and twice with kinase buffer (20 mM Tris, pH 7.5, 5 mM MgCl 2 , and 1 mM dithiothreitol). Kinase activity was measured by mixing immunoprecipitated GSK3␤ with 30 l of kinase buffer containing 20 mM Tris, pH 7.5, 5 mM MgCl 2 , 1 mM dithiothreitol, 250 M ATP, 1.4 Ci of [␥-32 P]ATP (Amersham Biosciences), and 0.1 g/l recombinant tau protein (in some experiments prephosphorylated tau was used). The samples were incubated at 30°C for 30 min, and 30 l of 2ϫ SDS stop buffer was added to each sample to stop the reaction. Samples were placed in a boiling water bath for 5 min, and proteins were separated on 10% SDS-polyacrylamide gels. The gels were vacuum-dried, exposed to a phosphoscreen overnight, and quantitated using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA). The efficiency of HA immunoprecipitation was determined by immunoblotting for GSK3␤.
In indicated experiments, the activity of GSK3␤ was measured using a primed peptide substrate, phosphoglycogen synthase peptide-2 (Upstate Biotechnology) as described previously (55), except that the peptide concentration used was 25 M, and the reaction time was 20 min. A volume of 10 l of the 30-l reaction mixture was then spotted in triplicate on P81 filter paper (Whatman). The filters were washed three times with 0.5% phosphoric acid for 15 min each and one time with 95% ethanol for 15 min and dried, and 32 P incorporation into the peptide was measured by scintillation counting.
Fractionation-Cell were rinsed once with warm phosphate-buffered saline and once with warm extraction buffer (80 mM PIPES, pH 6.8, 1.0 mM MgCl 2 , 2.0 mM EGTA, 30% glycerol, 10 mM benzamidine, 50 g/ml leupeptin, 1.0 mM phenylmethylsulfonyl fluoride, and 0.5 M okadaic acid). Cells were then incubated in 300 l (for a 60-mm plate) of extraction buffer containing 0.1% Triton X-100 for 10 min at 37°C. Cells were transferred into microcentrifuge tubes and centrifuged for 2 min at 16,000 ϫ g. The supernatant was removed and incubated in a boiling water bath for 10 min, whereas the pellet was resuspended in 100 l of 2ϫ SDS stop buffer, without dithiothreitol or dye, and sonicated. Protein determinations were made, and samples were diluted with 2ϫ SDS stop buffer, boiled, electrophoresed on SDS-polyacrylamide gels, and immunoblotted as described above.
Microtubule-binding Assay Using Cell Lysates-The microtubulebinding assay was carried out as described previously with a few modifications (6). Cells were collected and resuspended in 80 mM PIPES/ KOH (pH 6.8) and 1 mM CaCl 2 and protease and phosphatase inhibitors. Cell suspensions were sonicated briefly on ice and incubated on ice for 15 min. Samples were brought to 1.5 mM EGTA and centri-fuged at 100,000 ϫ g for 1 h at 4°C. Equal amounts of protein lysate were used in each binding assay. Supernatants were adjusted to 1 mM GTP and 10 M taxol and incubated with taxol-stabilized microtubules (30 M) prepared from rat brain (56) in a final volume of 30 l for 10 min at 37°C. The mixtures were centrifuged through 100 l of 30% (w/v) sucrose cushions in 80 mM PIPES/KOH containing 1 mM EGTA, 1 mM GTP, and 10 M taxol, at 100,000 ϫ g for 30 min in an airfuge at room temperature. The supernatant was collected and diluted with 2ϫ SDS stop buffer, and the pellet was resuspended in 50 l of 2ϫ SDS stop buffer prior to incubation in a boiling water bath for 5 min. The unbound (supernatant) and bound (pellet) fractions were separated by electrophoresis on a 10% SDS-polyacrylamide gel, transferred to nitrocellulose, and immunoblotted as described above.
Microtubule-binding Assay Using Recombinant Tau-CHO cells were transfected with HA-GSK3␤, HA-GSK3␤-S9A, HA-GSK␤-R96A, or vector only as the control. To immunoprecipitate the HA-tagged constructs, 400 g of protein from cellular extracts was incubated with 4 l of mouse monoclonal HA antibody overnight at 4°C with gentle agitation followed by incubation with 40 l of washed protein G-Sepharose for 2 h at 4°C. The immobilized immune complexes were washed twice with lysis buffer and twice with kinase buffer. The phosphorylation reaction was carried out by mixing the immunoprecipitates with 20 l of kinase buffer and 0.08 g/l recombinant tau protein. The samples were incubated at 30°C for 1 h with shaking and then spun to pellet the GSK3␤ immune complex. Phosphorylated or control tau (100 ng) was diluted in 80 mM PIPES/KOH (pH 6.8), 1.5 mM EGTA. The samples were adjusted to 1 mM GTP and 10 M taxol and incubated with taxol-stabilized microtubules (20 M) in a final volume of 30 l for 10 min at 37°C. The mixtures were centrifuged through 100 l of 30% (w/v) sucrose cushions in 80 mM PIPES/KOH containing 1 mM EGTA, 1 mM GTP, and 10 M taxol, at 100,000 ϫ g for 30 min in an airfuge at room temperature. The supernatant and pellet were collected and processed as described above.

RESULTS
Wild Type and GSK3␤-R96A Phosphorylate Tau Differentially-To determine whether GSK3␤ differentially phosphorylates tau when the phosphate-binding pocket of GSK3␤ is disrupted, cells were transfected with tau alone or with tau and HA-GSK3␤, HA-GSK3␤-R96A, or HA-GSK3␤-S9A. Lysates were collected and immunoblotted with the phosphate-independent tau antibodies Tau5/5A6, AT180, which recognizes tau when Thr-231 is phosphorylated (a primed site), PHF-1, which recognizes tau when Ser-396/404 are phosphorylated (unprimed sites), a GSK3␤ antibody, or a ␤-actin antibody. In the absence of exogenous GSK3␤ tau migrated as a doublet that was not recognized by either AT180 or PHF-1 (Fig. 1). Co-expression of either the wild type or mutant GSK3␤ resulted in a significant decrease in the mobility of tau; however in the presence of HA-GSK3␤ or HA-GSK3␤-S9A tau migrated as a distinct doublet. In contrast, in the presence of HA-GSK3␤-R96A, tau always presented as a blurred band(s); two distinct tau bands were never observed (also see Fig. 5B). This differential migration of tau in the presence of HA-GSK3␤-R96A compared with HA-GSK3␤ or HA-GSK3␤-S9A is likely because of the differential phosphorylation of tau by HA-GSK3␤-R96A. Expression of either HA-GSK3␤ or HA-GSK3␤-S9A with tau resulted in a robust increase in AT180 immunoreactivity and to a lesser extent increased PHF-1 staining. In contrast to these findings, expression of HA-GSK3␤-R96A resulted in a large increase in PHF-1 immunoreactivity and only a minimal increase in AT180 immunoreactivity (Fig. 1). In all cases wild type and mutant GSK3␤ were expressed at similar levels (Fig. 1). These results suggest that tau is both a primed and unprimed substrate of GSK3␤.
GSK3␤-R96A Phosphorylates Recombinant Tau Efficiently-The in vitro activity of the different GSK3␤ constructs toward primed and unprimed substrates was examined. Cells were transfected with HA-GSK3␤, HA-GSK3␤-R96A, or HA-GSK3␤-S9A immunoprecipitated with the HA antibody and used in the in vitro kinase assays. All the GSK3␤ constructs efficiently phosphorylated recombinant tau; however HA-GSK3␤-R96A phosphorylated recombinant tau ϳ4-fold more than HA-GSK3␤ or HA-GSK3␤-S9A ( Fig. 2A). In contrast, and as expected (32), HA-GSK3␤-R96A was extremely inefficient in phosphorylating a primed peptide, showing activity that was less than 20% of the activity of HA-GSK3␤ or HA-GSK3␤-S9A (Fig. 2B). To determine whether GSK3␤ phosphorylates primed tau in a manner similar to the primed peptide, tau was prephosphorylated with cdk5/p25 as described under "Experimental Procedures," prior to use in the in vitro kinase assay. After preparation of the primed or control tau (no ATP in the reaction), no cdk5 was present in the tau fractions as confirmed by immunoblotting for cdk5 (data not shown). Priming of recombinant tau prior to phosphorylation by the GSK3␤ constructs demonstrated that HA-GSK3␤-R96A was not as efficient as HA-GSK3␤ or HA-GSK3␤-S9A in phosphorylating primed tau (Fig. 2C, left panel). In contrast, HA-GSK3␤-R96A phosphorylated the control, mock phosphorylated tau much more efficiently than HA-GSK3␤ or HA-GSK3␤-S9A (Fig. 2C,  right panel). These results demonstrate clearly that both in situ and in vitro tau is both a primed and unprimed substrate of GSK3␤.
Tau Phosphorylated at a Primed Site Is Only Present in the Soluble Fraction-To determine whether there is a differential distribution of tau phosphorylated at primed (AT180) and unprimed (PHF-1) sites, cells were transfected with tau alone or tau with one of the GSK3␤ constructs, and the detergentinsoluble cytoskeleton was separated from the detergent-soluble fraction. These results (Fig. 3) demonstrate clearly that tau phosphorylated at primed sites (AT180) is only present in the soluble fractions, whereas tau phosphorylated at unprimed sites (PHF-1) is present in both the soluble and insoluble fractions. . HA-GSK3␤-S9A and HA-GSK3␤-R96A were expressed at similar levels compared with wild type HA-GSK3␤. The phospho-independent antibodies Tau5/5A6 show total tau levels in the transfected cells were approximately the same for all conditions. Tau5/ 5A6 immunoblotting revealed that the expression of the GSK3␤ constructs reduced the electrophoretic mobility of tau significantly, indicating an increase in phosphorylation. In cells overexpressing HA-GSK3␤ or HA-GSK3-S9A AT180 immunoreactivity (primed site, phosphorylated Thr-231) was increased significantly; in contrast, expression of HA-GSK3␤-R96A resulted in significantly less AT180 immunoreactivity. Expression of HA-GSK3␤ or HA-GSK3-S9A also increased PHF-1 immunoreactivity (unprimed sites, phosphorylated Ser-396/404), but expression of HA-GSK3␤-R96A resulted in a much greater increase in tau phosphorylation at this epitope. Actin blots are included to show equal protein loading for all the samples.

Primed Phosphorylation Sites of Tau Regulate the Intracellular Localization of Tau-
The next experiment was to determine how GSK3␤-mediated phosphorylation of tau at primed and unprimed sites regulates the association of tau with the cytoskeleton. Cells were transfected with tau alone or with tau and one of the GSK3␤ constructs, and the cells were fractionated, and the distribution of tau between the soluble and insoluble fractions was determined (Fig. 4). Co-expression of HA-GSK3␤ or HA-GSK3␤-S9A with tau increased the amount of tau in the soluble fraction compared with what was observed in cells transfected with tau alone, indicating that GSK3␤ changed the binding affinity of tau for the cytoskeleton/microtubules (Fig. 4). In contrast co-expression of HA-GSK3␤-R96A with tau did not result in an obvious increase in soluble tau levels (Fig. 4). These data suggest that phosphorylation of tau at primed sites may regulate the association of tau with microtubules, whereas GSK3␤-mediated phosphorylation at unprimed sites does not significantly impact tau-microtubule interactions.

Phosphorylation of Tau at Unprimed Sites by GSK3␤-R96A
Does Not Decrease Microtubule Binding-To further determine the functional significance of tau phosphorylation by GSK3␤ at primed and unprimed sites, a microtubule sedimentation assay was performed. High speed supernatants from cells transfected with tau alone or tau and HA-GSK3␤, HA-GSK3␤-R96A, or HA-GSK3␤-S9A were incubated with taxol-stabilized microtubules, and the amount of tau bound to the microtubules in the pellet and the amount of tau that remained unbound was measured. In the absence of GSK3␤, all the tau was in the microtubule-bound pellet. In contrast, when tau was co-expressed with HA-GSK3␤ or HA-GSK3␤-S9A a significant amount of tau was observed in the unbound fraction and was recognized by AT180 (Fig. 5A). Interesting, no PHF-1 immunoreactivity was observed in the unbound fractions. Further, co-expression of tau and HA-GSK3␤-R96A did not result in an increase in unbound tau even though there was a significant increase in PHF-1 immunoreactivity in the bound fraction. Fig.  5B shows that the expression of tau and GSK3␤ was equivalent To prime tau, tau was prephosphorylated with cdk5/p25 that was immunoprecipitated from cells in which cdk5 and p25 were co-expressed (left panel). As a control (unprimed tau), tau was incubated with the cdk5/p25 immunoprecipitate but in the absence of ATP (right panel). The panels below each graph show the autoradiographs for the experiment. The data were expressed as a percent of HA-GSK3␤ activity and are representative of results obtained from two to three separate experiments. These findings demonstrate clearly that priming of tau with cdk5/p25 diminishes the ability of HA-GSK3␤-R96A to phosphorylate tau. in all conditions. These results indicate that phosphorylation of tau by GSK3␤ at primed sites is necessary to decrease the affinity of tau for microtubules, whereas phosphorylation at unprimed sites, such as PHF-1, does not significantly effect tau-microtubule interactions.
To further confirm that the phosphorylation of tau at primed sites plays an important role in regulating the interaction of tau with microtubules, recombinant tau was phosphorylated with immunoprecipitated HA-GSK3␤, HA-GSK3␤-S9A, or HA-GSK3␤-R96A or with immunoprecipitates from vector-transfected cells. The phosphorylated recombinant tau was then used in a microtubule-binding assay, and the amount of bound and free tau was determined. These results (Fig. 6) again demonstrate that phosphorylation of tau by HA-GSK3␤ or HA-GSK3␤-S9A increased the amount of unbound tau compared with controls, whereas tau phosphorylation by HA-GSK3␤-R96A did not increase the level of unbound tau. DISCUSSION The aim of this study was to gain insight into the site-specific tau phosphorylation by GSK3␤ and to determine how phosphorylation of specific epitopes on tau affects its function. Our results clearly demonstrate that GSK3␤ phosphorylates tau on both primed and unprimed sites; however it is the phosphorylation of tau at the primed epitopes that decreases tau-microtubule interactions.
It has been demonstrated unequivocally that GSK3␤ phosphorylates tau in situ; however most studies have focused the phosphorylation of unprimed sites such as PHF-1, Tau-1, and AT8 (25,28,(57)(58)(59), although in a transgenic mouse model that expressed both human tau and GSK3␤-S9A an increase in AT180 immunoreactivity was noted (59). In vitro it has been demonstrated that AT180 recognition of tau requires Thr-231 to be phosphorylated, and GSK3␤ can phosphorylate tau on this site only when Ser-235 is already phosphorylated (53). Therefore, AT180 recognizes a primed GSK3␤ epitope. In situ we have shown that disruption of the priming phosphate-binding pocket in GSK3␤ by making the R96A mutation (32,44) impairs the ability of GSK3␤ to phosphorylate tau at the AT180 epitope significantly, confirming the in vitro findings. Further, priming of recombinant tau by prephosphorylating it with cdk5/p25 (41,42) results in a decrease in tau phosphorylation by the GSK3␤-R96A mutant compared with wild type GSK3␤ and GSK3␤ with the S9A mutation. Interestingly, GSK3␤ and GSK3␤-S9A phosphorylated tau to a similar extent; however GSK3␤-R96A phosphorylated recombinant tau much more efficiently than both GSK3␤ and GSK3␤-S9A. These data indicate that the enhancement of tau phosphorylation by the GSK3␤-R96A mutant is not because of the inability of GSK3␤-R96A to be inhibited by the pseudosubstrate of the phosphorylated N-terminal (32,44). In addition, tau phosphorylation at the non-primed PHF-1 epitope was much greater when cells were co-transfected with HA-GSK3␤-R96A compared with HA-GSK3␤ or HA-GSK3␤-S9A. The reason for the enhancement in non-primed tau phosphorylation by the GSK-R96A mutant is unknown, but may be because of an increase in the accessibility of tau to the active site. However, the R96A mutation of GSK3␤ did not affect phosphorylation of axin or ␤-catenin (32).
GSK3␤ is a constitutively active kinase that is inactivated by Ser-9 phosphorylation. Engagement of signaling pathways that activate Akt results in GSK3␤ Ser-9 phosphorylation and down-regulation of GSK3␤ activity. In addition, other protein kinases such as protein kinase C and mitogen-activated protein kinase-activated protein kinase 1 can phosphorylate Ser-9 on GSK3␤ and down-regulate its activity (31). In almost all experiments in this study, GSK3␤-S9A showed slightly, but not significantly, more activity than GSK3␤. The likely reason for this observation is that all measures were done under basal conditions and not in the presence of signals that would activate Akt, and therefore the effects of the S9A mutation on GSK3␤ activity were not prominent.
There is good evidence that in situ GSK3␤-mediated tau phosphorylation decreases the affinity of tau for microtubules. For example, microtubule depolymerization in response to colchicine treatment was greater in cells transfected with tau and GSK3␤ compared with cells transfected with just tau (57,60), and cells co-transfected with tau and GSK3␤ exhibited a reduction in microtubule bundling compared with cells transfected with tau alone (26). Further, in mice transgenic for GSK3␤-S9A and tau increased tau phosphorylation occurred, as well a reduction in the axonopathy that was observed in the mice that just overexpressed human tau (59). The reduction in the axonopathy in response to increased expression of GSK3␤-S9A is likely because of the fact that the increase in tau phosphorylation resulted in a decrease in tau-microtubule interactions, which thus alleviated the inhibition in axonal transport because of tau overexpression (3,59). Although it is clear that tau is phosphorylated by GSK3␤ on numerous sites, the contribution of these different sites to the regulation of tau function in situ has not been explored thoroughly. In this study we show that although GSK3␤-R96A phosphorylates tau in situ at nonprimed sites efficiently, and this results in a decrease in the electrophoretic mobility of tau, this did not result in an increase in the amount of free, non-microtubule-bound tau. This is in contrast to wild type GSK3␤ or GSK3␤-S9A, which did cause an increase in the levels of unbound tau. Further, AT180 immunoreactivity was only found in the soluble or unbound fraction, indicating that phosphorylation within this epitope and/or at other primed sites may play an important role in regulating FIG. 5. Phosphorylation of tau by GSK3␤ at primed sites results in decreased microtubule binding. CHO cells were transiently transfected with tau alone or tau and HA-GSK3␤ (GSK3␤), HA-GSK3␤-S9A (S9A), or HA-GSK3␤-R96A (R96A), and high speed supernatants from the cells were used in a microtubule-binding assay. A, tau is presented in soluble unbound fraction when it was phosphorylated by HA-GSK3␤ or HA-GSK3␤-S9A. Taxol-stabilized microtubules were added to the high speed supernatants, the microtubules were pelleted, and the distribution of tau between the microtubule pellet (Bound) and supernatant (Unbound) was determined by immunoblotting with several tau antibodies. These data demonstrate that tau phosphorylated by GSK3␤-R96A remains bound to the microtubules. In contrast, tau phosphorylated by HA-GSK3␤, and to a greater extent GSK3␤-S9A, shifted into the unbound supernatant. Further, AT180 immunoreactivity was present only in the unbound fractions. PHF-1 immunoreactivity was present only in the microtubule-bound fractions. B, total tau and GSK3␤ expression levels, as well as the phosphorylation state of tau, were evaluated in the high speed supernatants prior to use in the microtubule-binding assay. These data show that tau and the GSK3␤ constructs were all expressed at similar levels.
FIG. 6. In vitro assay demonstrating that phosphorylation of tau at primed sites by GSK3␤ decreases microtubule binding by tau. CHO cells were transiently transfected HA-GSK3␤ (GSK3␤), HA-GSK3␤-S9A (S9A), HA-GSK3␤-R96A (R96A), or vector only (Ϫ). Immunoprecipitations were carried out using an HA antibody, and the precipitates were used to phosphorylate recombinant tau. A microtubule-binding assay was carried out using the phosphorylated recombinant tau and taxol-stabilized microtubules, and the bound and unbound fractions were immunoblotted for the presence of tau (Tau5/5A6) or tubulin. These results demonstrate that phosphorylation of tau in vitro by HA-GSK3␤ or HA-GSK3␤-S9A decreases the association of tau with microtubule, whereas phosphorylation with HA-GSK3␤-R96A does not alter tau-microtubule interactions significantly. the interaction of tau with microtubules. In contrast, PHF-1 immunoreactivity was found exclusively in the bound fractions (Fig. 5A), indicating that phosphorylation of tau at this epitope and/or at other unprimed GSK3␤ sites plays less of a role in regulating tau-microtubule interaction. The finding that GSK3␤ phosphorylation of primed sites on tau may play a more significant role in regulating the function of tau than unprimed sites is quite intriguing given the fact that the majority of GSK3␤ substrates are primed (35). Additionally, recent data have shown that ␤-catenin, which was believed previously to be an unprimed substrate of GSK3␤, is actually primed by casein kinase I␣ (61). Further, different substrates of GSK3␤ require different priming protein kinases, such as protein kinase A for cAMP-response element-binding protein and casein kinase II for glycogen synthase (31) and of course casein kinase I␣ for ␤-catenin (61). Therefore one can speculate that the most functionally relevant GSK3␤-mediated tau phosphorylation events are at primed epitopes and that perhaps protein kinase(s) like cdk5/p25 may play an essential role in regulating GSK3␤mediated tau phosphorylation (41,42). Indeed, GSK3␤ and cdk5/p35[p25] co-purify together with tau and microtubules and were referred to originally as tau protein kinase I and II, respectively (62). A dual kinase system for the regulation of tau phosphorylation by GSK3␤ would give a high degree of control for GSK3␤ substrate selectivity and provide for the needed regulation of tau function in the cell. Further studies are required to fully elucidate the role of GSK3␤-mediated tau phosphorylation at both primed and unprimed sites and how these site-specific phosphorylation events affect tau function.