14-3-3 Connects Glycogen Synthase Kinase-3 (cid:1) to Tau within a Brain Microtubule-associated Tau Phosphorylation Complex*

In a recent study, we reported that in bovine brain extract, glycogen synthase kinase-3 (cid:1) and tau are parts of an (cid:1) 400–500 kDa microtubule-associated tau phosphorylation complex (Sun, W., Qureshi, H. Y., Agarwal-Mawal, D., and H. K. (2002) J. Biol. Chem . 277, 11933–11940). In this study, we find that when purified brain microtubules are subjected to Superose 12 gel filtration column chromatography, the dimeric scaffold protein 14-3-3 (cid:2) co-elutes with the tau phosphorylation complex components tau and GSK3 (cid:1) . From gel filtration fractions containing the tau phosphorylation complex, 14-3-3 (cid:2) , GSK3 (cid:1) , and tau co-immunoprecipitate with each other. From extracts of bovine brain, COS-7 cells, and HEK-293 cells transfected with

Microtubules, the major cytoskeletal structures of eukaryotic cells, are dynamic structures, and their assembly and disassembly is regulated by microtubule-associated proteins (1). In neurons, tau is one of the major microtubule-associated proteins and is mainly found in the axonal compartment (for reviews, see Refs. [1][2][3]. Tau binds to microtubules and stabilizes microtubule structure. Studies suggest that tau regulates microtubule dynamics, axonal transport, and neuronal morphology by binding and stabilizing the microtubule structure (1)(2)(3). There are six tau isoforms, which migrate with sizes 45-65 kDa on an SDS-polyacrylamide gel. These isoforms are phosphorylated on multiple sites in the brain and display a characteristic retarded mobility on an SDS gel upon phospho-rylation (2,3). Tau phosphorylation reduces the affinity of tau for microtubules and is one of the mechanisms that control microtubule structure and dynamics in vivo (1)(2)(3).
In Alzheimer's disease (AD) 1 brain, abnormally hyperphosphorylated tau accumulates and forms paired helical filaments (4,5). Since abnormally phosphorylated tau does not bind to microtubules, abnormal tau phosphorylation in AD brain is thought to cause a loss of tau function, microtubule dysfunction, and neurodegeneration (2,3). It is not understood how abnormally phosphorylated tau accumulates in AD brain, but a defect in the regulatory mechanism that controls tau phosphorylation/dephosphorylation is very likely to be involved. The elucidation of the regulatory mechanism that controls tau phosphorylation in normal brain and the determination of how this regulation fails in AD brain are essential steps in understanding disease ontogeny and developing therapeutic interventions.
Glycogen synthase kinase-3 (GSK3) is an important regulatory enzyme that phosphorylates numerous substrates and regulates diverse physiological processes such as glycogen metabolism, gene expression, apoptosis, signal transduction, and cell fate specification (6 -8). There are two isoforms of GSK3 that are highly expressed in the brain: ϳ51-kDa GSK3␣ and ϳ47-kDa GSK3␤ (9). In transfected cells and transgenic mice, enhanced expression of GSK3␤ leads to tau phosphorylation and microtubule instability (10 -15). In AD brain, GSK3␤ is activated in pretangle neurons and accumulates in paired helical filaments (16,17). These observations suggest that GSK3␤ phosphorylates tau in both normal and AD brain. Previous studies have shown that a large amount of GSK3␤ in brain is associated with microtubules (18 -20), and microtubule-associated GSK3␤ is part of an ϳ400 -500-kDa multiprotein complex containing tau and GSK3␤ (20). These data indicate that GSK3␤ phosphorylates tau within a microtubule-associated multiprotein complex (hereon designated as tau phosphorylation complex). The enormity of the tau phosphorylation complex suggests that within the complex, there may be proteins other than tau and GSK3␤ (20). The identification of all the complex components and the determination of their functions within the complex are essential to understanding the mechanism by which GSK3␤ phosphorylates tau in the brain.
In the brain, ϳ1% of soluble protein is 14-3-3 and has been suggested to be critical for brain function (21). From bovine brain extract, 14-3-3 co-immunoprecipitates with tau (36). In vitro, 14-3-3 binds and changes the tau conformation, thus making tau susceptible for kinase phosphorylation (36). More importantly, a substantial amount of 14-3-3 co-purifies with microtubules from the brain extract (36). These observations suggest that 14-3-3 is an integral part of brain microtubules and is involved in the regulation of tau phosphorylation and microtubule dynamics. However, very little information is available about microtubule-associated 14-3-3. In this study, we have further analyzed microtubule-associated 14-3-3. Herein we report that brain microtubule-associated 14-3-3 is part of the tau phosphorylation complex containing GSK3␤ and tau. Our data indicate that 14-3-3 mediates GSK3␤-tau interaction and facilitates tau phosphorylation by GSK3␤ within the complex.
Cell Culture and Transfection-COS-7 and HEK-293 cells were maintained in Dulbecco's modified Eagle's medium (high glucose) medium (Invitrogen) supplemented with 10% fetal bovine serum. Cells were plated in 100-mm culture dishes, grown to ϳ80% confluency, and transfected by standard calcium phosphate method with various amounts of the appropriate plasmids. For each 100-mm dish, 5-10 g of DNA was mixed with 50 l of CaCl 2 (2.5 M) to give a final volume of 500 l with distilled water. The mixture of DNA and CaCl 2 was added to 500 l of 2ϫ HEPES-buffered saline (1.63% NaCl, 1.188% Hepes, 0.02% Na 2 HPO 4 (pH 7.2)), and the mixture was allowed to settle at 20°C for 30 min. DNA mixture was added to the cells dropwise, and cells were allowed to grow for 12-18 h. The medium was then changed, and cells were incubated for 48 -72 h.
Microtubule Assembly/Disassembly and Partial Purification of 14-3-3 from Microtubule Fractions-Purification of microtubules from a fresh bovine brain extract by the temperature-induced microtubule assembly/disassembly has been described previously (38). Microtubule pellet obtained by centrifugation after first, second, third, and fourth cycles of assembly/disassembly were designated as P1, P2, P3, and P4, and the supernatants were designated as S1, S2, S3, and S4, respectively.
For immunoprecipitation, the supernatant (ϳ200 l) was precleared with ϳ50 l of protein G-agarose beads (Sigma) equilibrated in lysis buffer. The precleared sample was mixed with 10 g of indicated antibody, and the mixture was shaken end-over-end for 6 h at 4°C. After shaking, 30 l of protein G-agarose beads was added to the mixture, and the shaking was continued for another 5 h. The beads were then collected by centrifugation and washed three times (30 min each). The washed beads were dissolved in 50 l of SDS-PAGE sample buffer, boiled, and centrifuged, and 20 l of supernatant was analyzed by immunoblot analysis using the indicated antibody. The immunoprecipitation procedure for generating Fig. 4 is essentially as described (20).
To perform GST pull-down assay, ϳ50 l of glutathione-agarose beads (Sigma) coated with the indicated protein was incubated with 200 l of the cell or brain extract with end-over-end shaking for 14 h at 4°C. After shaking, beads were washed three times with 50 mM Tris-HCl (pH 7.5), 0.5% Nonidet P-40, 1 mM EDTA, and 1 mM DTT. The washed beads were dissolved in 50 l of SDS-PAGE sample buffer, boiled, and centrifuged, and 20 l of the supernatant was analyzed by immunoblot analysis using the indicated antibody. To generate Fig. 7, the GST pull-down assay was carried out as described above, except the brain or cell extract was replaced by the tau sample (50 mM Tris-HCl (pH 7.5), 100 mM NaCl, 1 mM EDTA, 1 mM DTT, 0.05% Tween 20, 0.3% bovine serum albumin, and 50 g/ml tau).
Microtubule-associated 14-3-3 Is Part of a Large Molecular Complex-To further characterize microtubule-associated 14-3-3, we depolymerized P3 microtubules by cold incubation and then subjected them to a phosphocellulose chromatography. 14-3-3 was not recovered within the flow-through fractions and eluted from the column with an NaCl gradient along with the other microtubule-associated proteins (data not shown, but see "Materials and Methods"). We then combined the column fractions containing 14-3-3 and chromatographed through an FPLC Superose 12 gel filtration column. Most of 14-3-3 eluted within fractions 40 -46 with a size of ϳ500-kDa (Fig. 2B). Since the size of dimeric 14-3-3 is ϳ60-kDa (21,22), these data indicated that 14-3-3 is bound to another biological molecule within the brain microtubules.
Identification of Molecules Bound to 14-3-3 within Brain Microtubules-A silver-stained SDS gel of various column fractions from Fig. 2A showed numerous protein bands of various sizes within fractions 40 -46 (data not shown) and did not give us any indication as to the identification of the 14-3-3-bound protein (s). In a previous study, we found that 14-3-3 is associated with tau in bovine brain extract and binds to tau in vitro (36). In a recent study, we showed that within brain microtubules, GSK3␤ and tau are parts of a multiprotein complex that elutes from an FPLC gel filtration column used in this study to generate Fig. 2A with an ϳ400 -500-kDa size (20). We noted a very similar gel filtration behavior between the high molecular size 14-3-3 present within fractions 40 -46 ( Fig. 2A) and the tau phosphorylation complex described by us in a previous study (20). We therefore analyzed various Fig. 2A column fractions for the presence of tau and GSK3␤. As shown in Fig. 2, C and D, tau and GSK3␤ were indeed present within fractions FIG. 1. Co-purification of 14-3-3 with microtubules. Microtubules were purified from a fresh bovine brain extract (H) by repeated cycles of microtubule assembly and disassembly. Samples (10 ml each) were analyzed by SDS-PAGE or immunoblot (IB) analysis using the indicated antibodies. A, SDS-PAGE showing tubulin and other microtubule-associated proteins in indicated fractions. B and C, immunoblots; D, 14-3-3/tubulin ratio. Blots B and C were scanned, and the band intensity values of 14-3-3 and tubulin in various fractions were obtained. The ratio for the indicated fraction was then determined by dividing the 14-3-3 band intensity value by the band intensity value of tubulin in that fraction. Values are the average of three independent determinations. P1, P2, P3, and P4 indicate pellets, whereas S1, S2, S3, and S4 indicate supernatants obtained after first, second, third, and fourth microtubule assembly/disassembly cycles, respectively.

FIG. 2. FPLC gel filtration of microtubule-associated 14-3-3.
Microtubules purified by three cycles of assembly and disassembly were chromatographed through a phosphocellulose column. The effluent fractions containing 14-3-3 were then analyzed by an FPLC Superose 12 gel filtration column calibrated previously with the indicated molecular weight marker proteins. Fractions (1 ml each) were collected, and 20 l from each indicated fraction was immunoblotted using the indicated antibody. A, gel filtration profile. BSA indicates bovine serum albumin. B, C, and D, immunoblots. IB indicates immunoblot. 40 -46, indicating that 14-3-3 has co-eluted with the tau phosphorylation complex from the gel filtration column. We pooled fractions 40 -46 containing 14-3-3 and a portion of the pooled fraction chromatographed through an FPLC Mono S column. SDS-PAGE and immunoblot analyses of various effluent fractions indicated that 14-3-3, tau, and GSK3␤ had co-eluted from the column (data not shown). We then pooled column fractions containing 14-3-3 and chromatographed through a Sepharose 4B gel filtration column. Tau, GSK3␤, and 14-3-3 again co-eluted (data not shown).
To gain more evidence in support of the above idea and to study the interactions of 14-3-3, tau, and GSK3␤ within the tau phosphorylation complex, we first asked whether or not 14-3-3 could bind to GSK3␤ directly. When glutathione-agarose beads coated with GST-14-3-3 were incubated with a brain extract, GSK3␤ specifically precipitated with the GST-14-3-3 beads (Fig. 5A). Although this observation indicated that 14-3-3 associates with GSK3␤ in the brain extract, we could not rule out the possibility that tau, which can bind to both GSK3␤ (20) and 14-3-3 in vitro (36), may have influenced observed 14-3-3 and GSK3␤ association (Fig. 5A). Therefore, we performed a similar GST pull-down assay as described above by using COS-7 cells that express GSK3␤ but not tau. As shown in Fig. 5B, GSK3␤ again came down with GST-14-3-3 from the cell extract. To confirm that it was GSK3␤ that came down with GST-14-3-3 and not any other protein of similar size that may be immunoreactive to our anti-GSK3␤ antibody used to generate Fig. 5, A and B, we transfected HEK-293 cells with HA-GSK3␤. Transfected cells were lysed, and glutathione-agarose beads coated with GST-14-3-3 were incubated with the cell lysates. Incubated beads were washed and immunoblotted by using an anti-HA antibody to test 14-3-3-GSK3␤ binding. As  Fig. 2A were combined, and a portion of the combined fraction was chromatographed through an FPLC Mono S column followed by a Sepharose 4B gel filtration column. An aliquot (20 l) from the peak effluent gel filtration column fraction containing 14-3-3, tau, and GSK3␤ was electrophoresed on a 10% SDS gel. The gel was stained with Coomassie Brilliant Blue for proteins and used to determine the molar ratios. A, protein-stained gel. B, molar ratio. The gel in panel A was scanned, and the band intensity values of various bands were obtained. The molar ratio for the indicated protein was then determined by dividing the band intensity value by the molecular weight of that protein. expected, HA-GSK3␤ bound to GST-14-3-3 (Fig. 5C). Based on these data, we concluded that 14-3-3 directly binds to GSK3␤.
To further confirm the above finding, we performed an in vitro GST pull-down assay. Glutathione-agarose beads coated with GST-GSK3␤ were mixed with bacterially expressed recombinant tau in the presence of a series of 14-3-3 concentrations. Beads were washed, and bead-bound tau was detected by immunoblot analysis using an anti-tau antibody. Comparatively very little tau bound to beads when GST-GSK3␤ was incubated with tau alone (Fig. 7, lane 3). However, when an increasing amount of 14-3-3 was included in the assay mixture, the amount of tau binding to GST-GSK3␤ increased progressively (Fig. 7, lanes 4 -8). When the amount of 14-3-3 was 100 g/ml in the assay mixture, ϳ10-fold more tau bound to GST-GSK3␤ than in the absence of 14-3-3 (compare lanes 3  and 7). Based on these data, we concluded that 14-3-3 pro-motes in vitro GSK3␤-tau binding and is required for a stable association of tau and GSK3␤ in vivo.
14-3-3 Stimulates GSK3␤-catalyzed Tau Phosphorylation-GSK3␤ is one of the kinases implicated to phosphorylate tau in vivo (10 -20). Since we find that 14-3-3 is required for a stable association between GSK3␤ and tau, we examined the influence of 14-3-3 on GSK3␤Ϫcatalyzed tau phosphorylation in vivo. We transfected HEK-293 cells in various combinations with FLAG-tau, Xpress-14-3-3, and HA-GSK3␤ constructs. Transfected cells were lysed, and the cell lysates were analyzed for tau phosphorylation using various tau phosphorylationsensitive antibodies: AT8, PHF1, and 12E8, which recognize tau phosphorylated on Ser 199 /Ser 202 , Ser 396 /Ser 404 , and Ser 262 , respectively (20,36). As shown in Fig. 8, A-C, tau was slightly phosphorylated in cells transfected with FLAG-tau alone (lane 3). This phosphorylation increased in cells co-transfected with FLAG-tau and HA-GSK3␤ as expected (lane 5). In cells that were co-transfected with fixed amounts of FLAG-tau and HA-GSK3␤ but different amounts of Xpress-14-3-3, FLAG-tau phosphorylation increased progressively with the increase in the amount of Xpress-14-3-3 (lanes 7-9). This increase was evident not only by an increased immunoreactivity against all tau phosphorylation-sensitive antibodies tested but also by a retarded mobility of FLAG-tau on the SDS gel, a characteristic feature of hyperphosphorylated tau (2, 3). Thus, 14-3-3 profoundly stimulated GSK3␤-catalyzed tau phosphorylation in vivo. Glutathione-agarose beads coated with GST-GSK3␤ or GST were incubated with tau solution in the presence of indicated amounts of 14-3-3. After incubation, beads were washed and immunoblotted with anti-tau antibody. This experiment was repeated three times with similar results. IB indicates immunoblot.

DISCUSSION
Recently, we reported the existence of a tau phosphorylation complex containing GSK3␤ and tau within brain microtubules (20). Since the observed size of the complex is ϳ400 -500 kDa and the sum of the molecular sizes of tau and GSK3␤ is ϳ97 kDa, we suggested that other proteins may also be present within the complex (20). In this study, we find that microtubule-associated 14-3-3 co-elutes with the tau phosphorylation complex from a gel filtration column, indicating that the size of high molecular size microtubule-associated 14-3-3 is the same as that of tau phosphorylation complex (Fig. 2). 14-3-3 and the tau phosphorylation complex in the microtubule fraction cannot be separated from each other by phosphocellulose, gel filtration, and Mono S chromatographies. Tau, 14-3-3, and GSK3␤ co-immunoprecipitate with each other from column fractions containing the phosphorylation complex (Fig. 4). In vitro, 14-3-3 binds to tau (36) and GSK3␤ (Fig. 5). These and other data presented in this study indicate that 14-3-3 is also a part of the microtubule-associated tau phosphorylation complex.
Tau and GSK3␤ co-immunoprecipitate with each other from brain extracts (Fig. 4) (20). In contrast, tau does not co-immunoprecipitate with GSK3␤ from HEK-293 cell extracts cotransfected with GSK3␤ and tau (Fig. 6A, lane 6). These data indicate that in HEK-293 cells, the interaction of tau with GSK3␤ is weak, whereas in the brain, GSK3␤ stably associates with tau. This in turn suggests that brain contains a factor required for a stable association of GSK3␤ with tau, and this factor may be missing in HEK-293 cells.
Our gel filtration data (Fig. 2) and co-immunoprecipitation analysis (Fig. 4) indicate that within the tau phosphorylation complex, tau, GSK3␤, and 14-3-3 are inseparable. Moreover, in HEK-293 cells, tau associates with GSK3␤ only in the presence of 14-3-3 (Fig. 6). In vitro, ϳ10-fold more tau binds to GSK3␤ in the presence than in the absence of 14-3-3 (Fig. 7). Since 14-3-3 can bind to tau (36) and GSK3␤ (Fig. 5) independently, these data indicate that 14-3-3 is the factor that connects and mediates the association of GSK3␤ with tau within the brain. However, as discussed above, tau does not associate with GSK3␤ in HEK-293 cells transfected with only tau and GSK3␤, although 14-3-3 is known to be widely expressed in various cell lines including HEK-293 cells (21). It is possible that within HEK-293 cells, the endogenous 14-3-3 either is not sufficient or is not available to mediate the interaction of transfected GSK3␤ and tau.
In a previous study, we reported that in vitro GSK3␤ binds to the N-terminal region of tau (20). Consistent with that report, we find that tau comes down with GST-GSK3␤ in a GST pulldown assay (Fig. 7, lane 3). However, tau does not co-immunoprecipitate with GSK3␤ from lysates of HEK-293 cells co-transfected with GSK3␤ and tau (Fig. 6A, lane 6). These observations suggest that in the absence of 14-3-3, GSK3␤ binds to tau with a low affinity. It thus appears that in the brain, GSK3␤ interacts with tau in two different ways: one with low affinity that does not require 14-3-3 and the other with high affinity that requires 14-3-3.
The substrate recognition by GSK3 is regulated by two mechanisms. The first mechanism requires a priming phosphorylation of the substrate (41,42). For example, casein kinase 2 phosphorylates glycogen synthase first and generates a recognition motif for GSK3. GSK3 then phosphorylates casein kinase 2-phosphoprylated glycogen synthase (42). The second mechanism does not require priming phosphorylation. Instead, a scaffold protein bridges GSK3␤ to its substrate within a multiprotein complex (42). In the Wnt signaling pathway, GSK3␤ phosphorylates ␤-catenin within a ␤-catenin destruction complex. ␤-catenin alone is not a good substrate of GSK3␤. The scaffold protein axin connects GSK3␤ to ␤-catenin and facilitates ␤-catenin phosphorylation by GSK3␤ within the complex (6,(42)(43)(44).
Biochemical analyses and studies involving transgenic mice and cultured mammalian cells have established that GSK3␤ phosphorylates tau in the brain (10 -20). The mechanism by which GSK3␤ phosphorylates tau is not clear. Our recent study (20) and the results presented in this study indicate that GSK3␤, tau, and 14-3-3 are parts of a microtubule-associated tau phosphorylation complex. Within the complex, 14-3-3 binds to tau and GSK3␤ simultaneously and assembles the complex. Thus, the role of 14-3-3 within the tau phosphorylation complex appears to be similar to that of axin within the ␤-catenin destruction complex. Furthermore, 14-3-3 binds to tau and changes the tau conformation, making tau susceptible for hyperphosphorylation in vitro (36) and perhaps in vivo (Fig.  8). Since 14-3-3 stimulates tau phosphorylation on Ser 199 , Ser 198 Ser 202 , Ser 262 , Ser 396 , and Ser 404 (Fig. 8), it appears that 14-3-3-induced conformational change occurs within a large part of the C-terminal tau region, which is the main area of in vivo phosphorylation (45). These observations suggest that 14-3-3 not only enhances association of tau and GSK3␤ within the complex but also prepares tau for GSK3␤ action.
We have found a unique multiprotein complex containing tau, GSK3␤, and 14-3-3 within brain microtubules. Thus, a pool of GSK3␤ in the brain is targeted to microtubules through a stable association with tau and 14-3-3. Because the function of this complex is to regulate tau phosphorylation and microtubule dynamics, we named this complex the tau phosphorylation complex. It should be noted that the size of the phosphorylation complex is 400 -500-kDa, whereas the sum of the sizes of tau, GSK3␤, and 14-3-3 dimer is ϳ167-kDa. Therefore, it is possible that there may be proteins other than, tau, GSK3␤, and 14-3-3 within the tau phosphorylation complex. These proteins may play important roles in regulating tau phosphorylation and interactions between various phosphorylation FIG. 8. Effect of 14-3-3 on GSK3␤-catalyzed tau phosphorylation in vivo. HEK-293 cells transfected with the indicated constructs were lysed, and 20 g protein from each lysate was immunoblotted using the indicated antibody. A-C, immunoblot analysis using tau phosphorylation-sensitive monoclonal antibodies, AT8, PHF1, and 12E8, which only cross-react with phosphorylated tau. D-F, immunoblots to show expression levels of FLAG-tau, HA-GSK3␤, and Xpress-14-3-3. Lane 1 represents mock-transfected cells. Similar results were obtained in three different experiments. IB indicates immunoblot. complex components. Studies are ongoing in our laboratory to identify all of the phosphorylation complex components.