Promotion of Hyperphosphorylation by Frontotemporal Dementia Tau Mutations*

Mutations in the tau gene are known to cosegregate with the disease in frontotemporal dementia with parkinsonism linked to chromosome 17 (FTDP-17). How-ever, the molecular mechanism by which these mutations might lead to the disease is not understood. Here, we show that four of the FTDP-17 tau mutations, R406W, V337M, G272V, and P301L, result in tau proteins that are more favorable substrates for phosphorylation by brain protein kinases than the wild-type, largest four-repeat protein (cid:1) 4L and (cid:1) 4L more than (cid:1) 3L. In general, at all the sites studied, mutant tau proteins were phosphorylated faster and to a higher extent than (cid:1) 4L and (cid:1) 4L > (cid:1) 3L. The most dramatic difference found was in the rate and level of phosphorylation of (cid:1) 4L R406W at positions Ser-396, Ser- 400, Thr-403, and Ser-404. Phosphorylation of this mutant tau was 12 times faster and 400% greater at Ser-396 and less than 30% at Ser-400, Thr-403, and Ser-404 than phosphorylation of (cid:1) 4L. The mutated tau proteins polymerized into filaments tau mg/ml brain phosphorylation 8 h ofincubation m The mol of 32 P/mol of tau calculated using (cid:3) - 32 P]ATP of a known specific activity and, as control, the brain extract without exogenous tau. determine the amount of radioactivity bound to tau, 2 (cid:2) l of the incubation on Whatman chro- matographic paper prespotted with 5 (cid:2) l of 20% trichloroacetic acid 2 m M ATP 200 m M NaCl. The were subjected to ascending chromatography to separate the free 32 P from protein-incor-porated 32 P. The was counted, and the mol of phosphate per mol of tau were calculated from the specific radioactivity after subtracting

Pathological tau aggregation in the form of filaments is increasingly recognized as a common denominator underlying a variety of dementias called tauopathies, which include Alzheimer disease (AD), 1 frontotemporal dementia with Parkinsonism linked to chromosome 17 (FTDP-17), amytrophic lateral sclerosis, cortical basal degeneration, dementia pugilistica, Pick disease, progressive supranuclear palsy, and tangle-only dementia. Despite their diverse topographic and phenotypic manifestations, these tauopathies are linked to the progressive accumulation of filamentous, hyperphosphorylated tau inclusions. The discovery of the co-segregation of certain mutations in the tau gene with the disease in inherited cases of FTDP-17 has confirmed that tau abnormalities that are seen in tauopathies as neurofibrillary degeneration are disease-causative (1)(2)(3).
Tau promotes the assembly of and stabilizes microtubules. It is a family of six proteins derived from a single gene by alternative mRNA splicing (4,5). The human brain tau isoforms range from 352 to 441 amino acids. They differ in whether they contain 3 (3L, 3S, or 3) or 4 (4L, 4S, or 4) tubulin binding domains/repeats (R) of 31 or 32 amino acids each near the C terminus and 2 (3L, 4L), 1 (3S, 4S), or no (3,4) inserts of 29 amino acids each in the N-terminal portion of the molecule. In AD, all six isoforms have been reported to be present in a hyperphosphorylated state in paired helical filaments (PHFs), which form neurofibrillary tangles (6 -8). Microtubules are required for axoplasmic transport, and in the tangle-bearing neurons of patients with AD, the microtubule system is destroyed and replaced by PHFs.
In inherited FTDP-17 certain mutations in tau gene cosegregate with the disease. These mutations are missense, intronic, and deletions of single amino acids. The missense mutations result in the substitution of one amino acid, and it has been reported that these mutations compromise tau ability to promote microtubule assembly (9,10) and, in the presence of polyanions, promote its ability to polymerize (11)(12)(13). The intronic 5Ј to exon 10 mutations result in overexpression of 4 R tau (1,3). The effect of the deletion mutations in tau gene in patients is as yet unknown. In vitro ⌬K280 and ⌬N296 result in a change of 3R/4R ratio and a decrease in binding of tau to microtubules (14 -16). The exact molecular mechanism of neurodegeneration in the affected patients is not yet understood. Like patients with AD, FTDP-17 patients show accumulations of hyperphosphorylated tau in the form of neurofibrillary tangles. All the mutations discovered in tau are dominant, suggesting that the effect of these mutations results in a gain of toxic function (17). In AD, hyperphosphorylation of tau appears to precede the appearance of tangles (18,19), and abnormally hyperphosphorylated tau (AD P-tau) sequesters normal tau, microtubule-associated protein 1 and 2 and depolymerizes microtubules in vitro (20 -22).
In the present report, we show that the gain of toxic function by FTDP-17 mutations rests on the increased susceptibility of mutant tau to be hyperphosphorylated as follows. (i) The FTDP-17 mutations R406W, V337M, G272V, and P301L make tau more favorable for hyperphosphorylation, (ii) 4L is a more favorable substrate for kinases than 3L, (iii) the mutated tau proteins polymerize into filaments at lower levels of phosphate incorporation (4 -6 mol of phosphate/mol of tau) than wild-type tau (ϳ10 mol of phosphate per mol of protein), and (iv) upon incorporation of ϳ4 mol of phosphate per mol of protein, the hyperphosphorylated tau is able to sequester normal tau. These findings taken together suggest that mutations in tau might cause neurodegeneration through rendering a molecule that is a favorable substrate for kinases which polymerizes at lower levels of phosphorylation and that becomes "toxic" to the microtubule system by sequestering normal tau.
Plasmid Construction and DNA Mutagenesis-All wild-type and mutant tau genes were cloned in pET17b vector (Novagen) through NdeI-EcoRI restriction sites. The pET17b vector in combination with suitable Escherichia coli strains (DH5alfa for cloning and BL21DE3 for protein expression) represents an inducible T7-promotor expression system (30).
The pET17b plasmids carrying cDNA of the human genes for 3L and 4L isoforms were produced by PCR amplification of tau cDNAs from pRK172/3L or pRK172/4L vectors kindly provided by Dr. Michel Goedert. Specific primers, including translation initiation start (ATG) and stop (TGA) codons and NdeI and EcoRI-cloning restriction sites, were used.
Four pET17b plasmids bearing genes for 4L with FTDP17 point mutation G272V, P301L, V337M, or R406W (Fig. 1A) were produced. The numbering of amino acids refers to the longest human brain tau isoform 441 . Mutagenesis was achieved by inverse PCR amplification of the pET/tau vector with inverted primers, one of which bore the point mutation. Vent thermostable DNA polymerase was used because of its proofreading activity and production of blunt ends. Amplified linearized plasmids were gel-purified, and after treating with T4 polynucleotide kinase in T4 ligase ligation buffer for 30 min at 37°C and religation (T4 ligase ligation buffer plus 5% polyethylene glycol and 10 mM dithiothreitol for 1 h at room temperature) were transformed in DH5alfa cells for screening and plasmid isolation (Qiaprep Miniprep kit, Qiagen). Each selected construct was verified by DNA sequencing with one primer in the middle of the tau gene and two primers complementary to vector sequences flanking the insert (BigDye Terminator Ready Reaction kit, ABI Prism377 Sequencer, PerkinElmer Life Sciences).
Isolation of Recombinant Human Brain Tau Proteins-The tau constructs were subcloned, and recombinant human tau proteins were purified as described (31) except that the perchloric acid extraction was avoided. The phosphocellulose-purified tau was further purified on a Sephacyl 300 column at 4°C.
Hyperphosphorylation of Tau-Hyperphosphorylation was performed using 100,000 ϫ g brain extract from a 20-day-old rat as the source of protein kinases as described previously (32). The reaction was carried out at 35°C in 60 mM Hepes, pH 7.4, 8 mM MgCl 2 , 5 mM EGTA, 2 mM ATP, 2 mM dithiothreitol, 20 nM calyculin A, 1 mM 4-[2-aminoethylamino]-benzenesulfonyl fluoride (AEBSF, a serine protease inhibitor) and from 0.1-0.5 mg/ml tau protein and 0.25-0.5 mg/ml of brain extract. The brain extract was prepared fresh before the phosphorylation reaction. After 2 and 8 h of incubation NaF (17 mM) and ATP (2 mM), respectively, were added. The mol of 32 P/mol of tau was calculated using [␥-32 P]ATP of a known specific activity and, as control, the brain extract without exogenous tau. To determine the amount of radioactivity bound to tau, 2 l of the incubation mixture was dotted on Whatman chromatographic paper prespotted with 5 l of 20% trichloroacetic acid containing 2 mM ATP and 200 mM NaCl. The samples were subjected to ascending chromatography to separate the free 32 P from protein-incorporated 32 P. The latter was counted, and the mol of phosphate per mol of tau were calculated from the specific radioactivity after subtracting the corresponding value of the enzyme alone control.
Mapping of Phosphorylation Sites with Phospho-dependent Antibodies to Tau-The sites phosphorylated in tau were mapped by using the phospho-dependent tau antibodies listed above by Western blots. A 5-l sample of the phosphorylation reaction mixture was taken at 1.5, 3, 4.5, 6, 10, 20, and 22 h of incubation and subjected to SDS-PAGE (50 ng of tau per lane). The rate of phosphorylation was compared at Thr-181, Ser-199, Ser-202, Thr-205, Thr-212, Ser-214, Thr-217, Thr-231, Ser-235, Ser-262, Ser-396, Ser-404, and Ser-422 using specific phospho-dependent antibodies. Once we identified the specific sites that were differentially phosphorylated, we quantitated the specific differences. The phosphorylated proteins were dotted in different amounts (0.5, 1, 5, 10 ng per dot) on nitrocellulose membranes, blocked with bovine serum albumin, and developed with the phospho-dependent antibodies selected above. We employed a mixture of rabbit polyclonal antibodies, 92e, 134d, and 111e, against total tau to assay the levels of total tau. 125 I-Labeled secondary antibodies were used to quantitate the immunoreactivity. The ratio between tau immunoreactivity with a phosphodependent antibody to that of a phospho-independent antibody (92e, 134d, and 111e) was used to normalize the phosphorylation of tau at a particular site.
Self-assembly of Tau-Wild-type and mutant tau self-assembly was induced by hyperphosphorylation, with the brain extract as a source of kinases as described above. The polymerization of mutated tau proteins phosphorylated by different kinases was compared by negative-stain electron microscopy. From the phosphorylation reaction mixture described above, 10 l at different incubation times (0, 0.5, 1, 2, 4, 6, and 22 h) were applied on a 300-mesh carbon-coated grid and negatively stained with 2% phosphotungstic acid as described previously (20,33). The percentage of aggregated protein was determined by centrifuging an aliquot (10 l) of the incubation mixture at 30,000 ϫ g for 20 min in a Beckman TL-100 ultracentrifuge at different times to separate the aggregated from the soluble protein pool. The amount of tau in each fraction was quantified as above by radioimmuno dot blots developed with a mixture of three phospho-independent tau antibodies 92e, 134d, and 111e.
Association of Normal Tau to the Hyperphosphorylated Tau-The binding of the in vitro hyperphosphorylated tau to normal tau was studied by a dot overlay assay (20,34). For the dot overlay assay, in vitro hyperphosphorylated tau proteins (10 -100 ng) were spotted on nitrocellulose and blocked with 5% fat-free dry milk in 100 mM Mes buffer, pH 6.7, for 1 h. After blocking, the membrane was overlaid with normal tau (3-6 g/ml) for 3 h. The unbound tau was removed by washing the membrane, and the bound tau was fixed with 0.5% formaldehyde. After fixation, the bound tau was detected with Tau-1 antibody, which recognizes tau unphosphorylated at Ser-195/198/199/202. Samples in which overlay with normal tau was substituted with bovine serum albumin were used to deduct any background binding of Tau-1 to the hyperphosphorylated tau.
Immunoblots, Protein, and Tau Assays-Protein concentration was estimated by the method of Bensadoun and Weinstein (35). Sample preparation and immunoblots were carried out as described previously (6). The levels of recombinant tau isoforms and AD P-tau were determined by the radioimmuno-slot blot method of Khatoon et al. (36) using a mixture of three phospho-independent tau antibodies, 92e, 134d, and 111e.

RESULTS
Expression and Isolation of 4L, 3L, and Mutant Tau 4L R406W , 4L V337M , 4L G272V , and 4L P301L -Four pET17b plasmids bearing genes for 3L, 4L, and 4L with FTDP17 point mutation G272V, P301L, V337M, or R406W were produced (Fig. 1A). Each selected construct was verified by DNA sequencing with one primer in the middle of the tau gene and two primers complementary to vector sequences flanking the insert (data not shown). The tau proteins (Fig. 1A) were expressed and purified from E. coli. We previously showed that although tau is heat-and acid-stable, these treatments abolished its ability to self-polymerize into short PHF-like filaments (32). For this reason, we purified recombinant proteins without any steps using heat or acid treatment. The isolation procedure yielded ϳ5 mg of Ͼ95% pure protein per 500 ml of bacterial culture (Fig. 1B).
The FTDP17 point mutations in recombinant tau40 protein did not cause any drastic property change compared with 4L. The bacterial expression yield and behavior in the fast protein liquid chromatography purification were identical to those of normal tau. The point mutations did not increase toxicity for E. coli or different apparent thermostabilities of purified protein. All the FTDP-17 mutant tau proteins migrated to the same apparent molecular weight in SDS-PAGE, suggesting no apparent conformational change in the mutant versus the wildtype tau proteins, at least in the denaturing conditions (Fig. 1B).
Comparison of Mutated 4L R406W , 4L V337M , 4L G272V , and 4L P301L with Wild-type 4L and 3L as Substrates for Hyperphosphorylation-We postulated that FTDP-17-mutated tau proteins are more favorable substrates than the wild-type tau for abnormal hyperphosphorylation. To test this hypothesis, we in vitro hyperphosphorylated recombinant wild-type and mutated tau proteins with normal rat brain extract as a source of kinases. Phosphorylation of 4L with the brain extract resulted in ϳ12 mol of phosphate per mol of the protein during ϳ20 h, whereas the extent of phosphorylation was much higher for all the mutated tau proteins examined, i.e. ϳ13-18 mol of phosphate per mol of 4L R406W , 4L V337M, 4L P301L , and 4L G272V ( Fig. 2A). Not only the stoichiometry but also the kinetics of phosphorylation were more favorable for phosphorylation of the mutated tau proteins. The kinetics of phosphorylation was more favorable for 4L than for 3L, although the net levels of phosphorylation were only ϳ10% less for 3L. Phosphorylation is known to retard the mobility of tau in SDS-PAGE (7). We analyzed the electrophoretic patterns of wild-type and mutant tau proteins as a function of the phosphorylation time by Western blots using antibodies against total tau (Fig. 2B). As the phosphorylation reaction progressed, there was an upward mobility shift of tau that corresponded to the higher levels of phosphorylation. For example, at 3 h of phosphorylation two bands were seen, and the proportion of the upper band was higher for mutant tau proteins (percentages calculated from scanning of the blots: 44% for 4L R406W , 46% for 4L V337M, 42% for 4L P301L , and 51% for 4L G272V as compared with 35% for 4L and 27% for 3L). These mobility shift data suggested that mutant tau proteins were probably phosphorylated more readily than 4L and 3L.
The higher phosphorylation stoichiometry of mutant tau proteins compared with wild-type tau indicated an increase in phosphorylation of more than a single site. To identify the phosphorylation sites responsible for hyperphosphorylation, we studied tau phosphorylation at various sites. For each phospho-epitope, the immunoreactivity was determined at each time point of the phosphorylation reaction, and that amount was normalized to the total amount of tau present in the sample (specific immunoreactivity). The specific immunoreac-  2. A, time course of 32 P incorporation into wild-type 4L and 3L and mutated 4L R406W , 4L V337M , 4L G272V , and 4L P301L . Wild-type 4L and 3L and mutated 4L R406W , 4L V337M , 4L G272V , and 4L P301L , 0.5 mg/ml, were incubated with rat brain extract as a source of brain kinases in the presence of ATP to induce hyperphosphorylation. Mutated 4L R406W , 4L V337M , 4L G272V , and 4L P301L incorporated more moles of phosphate per mole of tau than wild-type 4L. B, Western blots of incorporation of phosphate into wild-type 4L, 3L and mutated 4L R406W , 4L V337M , 4L G272V , and 4L P301L . Aliquots of the above phosphorylation reaction mixtures were taken at different incubation times and analyzed by Western blots developed with a mixture of phosphorylation-independent polyclonal antibodies against tau. tivity was obtained for each antibody and each time point and plotted as a function of time. From these plots, three different parameters could be determined; they are the rate of phosphorylation at each particular site from the slope of the curve at the starting time (initial velocity), the highest amount of phosphorylation at that site (from the plateau of the same curve), and the time needed to reach 50% of the total phosphorylation of the particular site in 4L. Because there are no standards for each antibody, only the relative level of phosphorylation, but not the absolute values, can be calculated. The values obtained for 4L at each site were taken as 100%, and all the others were expressed relative to those of 4L. The calculated rates of phosphorylation, average results of three to six independent experiments, are shown in Fig. 3A. In general 4L was a more favorable substrate for kinases than 3L, since the rate of phosphorylation at almost all the sites investigated was lower (less than 100%) for 3L (Fig. 3A). The total level of phosphorylation was slightly lower for 3L than for 4L (from 70 -90%, Fig. 3B) at most of the sites studied. All the mutant tau proteins investigated were found to be more favorable substrates for phosphorylation than 4L in both the rate (Fig 3A) and the total level of phosphorylation (Figs. 2 and 3B). In general, 4L was a more favorable substrate than 3L, and the mutant tau proteins were more favorable than 4L for phosphorylation.
To compare the rates of phosphorylation of tau at each site, we determined the time it took for each mutant and for 3L (Fig. 3C) to achieve 50% of the phosphorylation level of 4L at a particular site. These results suggested that mutations in tau induced conformational changes, making it a more favorable substrate for overall phosphorylation. The most remarkable change was the one induced by the R406W mutation. This mutation reduced the time for half-maximal phosphorylation at Ser-396 from ϳ 8 h in 4L to ϳ2 h in 4L R406W (see panel 4L R406W ). The dramatic increase in the phosphorylation of 4L R406W at Ser-396 (Fig. 3C) resulted in a marked decrease in phosphorylation at positions 400, 403 (data not shown), and 404 (Fig. 3C).
Polymerization of Mutated Taus upon Hyperphosphorylation-We previously showed that polymerization of tau can be induced by hyperphosphorylation when at least 10 mol of phosphate per mol of this protein have been incorporated (32). Because we found (see above in "Results") that the mutated tau proteins were more favorable substrates for phosphorylation, we proceeded to test the hypothesis that the mutated proteins on phosphorylation self-assemble into filaments more readily than does wild-type tau. For this purpose, 3L, 4L, 4L R406W , 4L V337M , 4L G272V , and 4L P301L were phosphorylated as described above, and 10-l aliquots were taken at different incubation times (0, 1/2, 1, 2, 4, 6, and 22 h) were employed for negative stain electron microscopy. No filaments were detected at the 0-min incubation time (data not shown). As early as 1/2 h after phosphorylation, polymerization of 4L V337M into filaments that had portions very well defined of about 4-nm-width was detected (data not shown). At 1 h of phosphorylation, 4L V337M (Fig. 4Ai) appeared to polymerize more than 4L R406W (Fig. 4Aii) and 4L G272V , but filaments were observed in all these tau proteins. At 4 h of phosphorylation, filaments were detectable in 4L P301L , and lateral association of filaments was detectable for 4L R406W (Fig. 4Aiii). At 6 h of phosphorylation, association of filaments into bundles was easily visible (Fig. 4, Aiv). No filaments were detected in 3L or 4L in up to 6 h of incubation or in any of the mutant tau proteins incubated in the absence of ATP (figure not shown). As the phosphorylation progressed, the filaments detected were more abundant, and 10 -15-nm filaments were also found (Fig.  4Aiii). Generally, the 10 -15-nm filaments were associated into bundles. PHF-like structures could be detected in some of the bundles (Fig. 4, Av, inset). At 22 h of phosphorylation, filaments could be seen in both mutated and wild-type tau proteins (data not shown). These findings revealed that mutant tau proteins required fewer moles of phosphate per mole of protein to selfassemble into filaments, probably because the conformation of mutated tau proteins made them more prone to aggregate.
To quantitate aggregation of tau, aliquots of the phosphorylation mixture were taken at different times and centrifuged for 10 min at 20,000 ϫ g to separate the polymerized from the non-polymerized protein. The amount of tau recovered in the supernatant was determined by dot blot assays as described using a mixture of phospho-independent polyclonal antibodies. The amount of 3L and 4L in the supernatant did not decrease more than 10% during 5 h of phosphorylation, whereas by that time, 40 -80% of the mutant tau proteins formed sedimentable aggregates (Fig. 4B).
Association of Normal Tau with the Hyperphosphorylated Tau-Previously we showed that abnormally hyperphosphorylated tau from AD brain sequestered normal tau and disrupted microtubules and that in vitro hyperphosphorylation of tau conferred the ability to bind normal tau (37). 4L, 3L, 4L R406W , 4L V337M , 4L G272V , and 4L P301L were phosphorylated as described above, and 10-l aliquots were taken at different incubation times (0, 1/2, 1, 2, 4, 6, and 22 h). Taus were dotted on a nitrocellulose membranes and overlaid with 5 g/ml 4L. Tau bound to P-tau was detected as described and expressed as an arbitrary number reflecting the increment of Tau-1 immunoreactivity of the membrane overlaid with normal tau minus the corresponding one overlaid with bovine serum albumin as a control. Mutated tau proteins and wild-type tau proteins were able to sequester normal tau upon incorporation of about 4 mol of phosphate per mol of protein, which was achieved by as early as 30 min of phosphorylation in the case of mutant tau proteins. With all the tau proteins employed, the ability to sequester normal tau increased with phosphorylation until it reached a maximum level of binding at about 3-5 mol of phosphate per mol of tau, and then this capacity decreased with further phosphate incorporation (Fig. 5). The maximum level of normal tau binding was different for the different P-tau proteins: The levels of binding ranged from 2.1 (4L V337M ) to 1.4 (4L P301L ) times more than that by 4L, and the latter bound ϳ1.1 times more tau than 3L. As phosphorylation progressed, mutant tau proteins showed a marked decrease in their ability to bind normal tau, probably due to their self-assembly into filaments. DISCUSSION Dementia in AD and in other tauopathies such as FTDP-17 appears to require neurofibrillary degeneration. Like AD, FIG. 4. Tau polymerization and ultrastructural analysis upon phosphorylation. Wild-type 4L and 3L and mutated 4L R406W , 4L V337M , 4L G272V , and 4L P301L , 0.5 mg/ml, were incubated with rat brain extract as a source of brain kinases in the presence of ATP to induce hyperphosphorylation. Two aliquots of the samples were taken at different incubation times, and one was negatively stained for electron microscopy (A: i, 4L V337M ; ii, 4L R406W , 1-h incubation; iii, 4L R406W , 4-h incubation; iv, 4L P301L ; v, 4L G272V , 6-h incubation), and the other was centrifuged to separate soluble from aggregated tau. The soluble tau was quantitated by immuno-dot blot (panel B).
FTDP-17 is characterized histopathologically by hyperphosphorylation and aggregation of tau into tangles of filaments. The mechanism by which mutations in tau lead to neurofibrillary degeneration and, consequently, the disease is not well understood. To date, three silent mutations, 18 missense mutations, and two deletion and eight intronic mutations that co-segregate with FTDP-17 have been described (www.lille. inserm.fr/u422/taumutable.html). Three effects of mutations on the biology of tau have been postulated; they are 1) reduced ability to bind to microtubules (9, 10), 2) altered tau splicing, with the majority of the mutations resulting in increased inclusion of exon 10 in tau mRNA, i.e. an increase in the ratio of 4R/3R tau proteins (1,3,38), and 3) an increase in tau filament formation in the presence of polyanions in vitro (39 -41). The mutations of tau have been shown to be autosomal dominant, and thus, it is likely that the FTDP-17 mutations result in a gain of toxic function of tau. Reduced ability to bind to microtubules would be a loss of function. However, all mutations that affect alternative splicing and result in 4R tau should work through some other mechanism, because 4R tau proteins bind to microtubules with higher affinity than 3R tau proteins. The third postulated mechanism of toxicity is increased formation of tau filaments. However, to date, there is no evidence linking filament formation to neurotoxicity; instead, reports have shown the association of neurotoxicity to hyperphosphorylation of tau in the absence of its polymerization (42,43).
Previously, we showed that unlike normal tau, AD P-tau does not promote but instead inhibits in vitro microtubule assembly by sequestering normal microtubule-associated proteins and that this toxic behavior is due solely to abnormal hyperphosphorylation because in vitro dephosphorylation converts AD P-tau into a normal-like protein (20 -22, 44). Furthermore, hyperphosphorylation promotes self-as-sembly of tau into tangles of PHF/straight filaments (32). These in vitro findings were recently corroborated in cell culture in which treatment with okadaic acid induced thioflavin S-positive hyperphosphorylated tau aggregates (45). Thus, it is possible that in AD, probably due to a decrease in phosphatase activity (46), and in FTDP-17, probably due to mutation, tau becomes a more favorable substrate for kinases, is hyperphosphorylated, sequesters normal tau, disrupts the microtubule system, and self-assembles into tangles of filaments. In the present study, we have found both the rate and the level of phosphorylation of FTDP-17 mutant tau proteins Ͼ 4L Ͼ 3L and that this increase in phosphorylation of mutated tau proteins was reflected at several specific sites. At around 4 mol of phosphate per mol of tau, this protein was able to sequester normal tau, and the ability to sequester normal tau was mutated tau proteins Ͼ 4L Ͼ 3L. Mutated tau proteins acquired the conformation needed to polymerize into filaments at lower levels of incorporated phosphate. Therefore, these results taken together suggest that FTDP-17 mutations of tau render tau more ready to acquire toxic-like behavior induced by abnormal hyperphosphorylation.
We observed that FTDP-17 mutant tau proteins phosphorylated faster and to a higher extent than did 4L in vitro using brain extract as a source of kinases. These results suggested that the FTDP-17 mutations induce a conformational change in the tau proteins that make them more favorable for phosphorylation and that the presence of the extra microtubule binding domain, i.e. the second repeat, similarly leads to hyperphosphorylation. Previously, Jicha et al. (47) showed by circular dichroism that there is a change in the conformation of tau due to the mutations in 4L G272V , 4L P301L , 4L V337M , and 4L R406W . The alteration in tau structure was so significant that it also altered its elution profile in a C8 reverse-phase column chromatography, probably by leading to increased exposure of hydrophobic residues in extended areas of the molecule. The present study suggests that the mutation-induced conformational changes probably make tau more prone to hyperphosphorylation and to assemble at lower levels of phosphate incorporation (4 -6 mol of phosphate/mol of tau) than wild-type tau (ϳ10 mol of phosphate per mol of protein). Phosphorylation-induced polymerization of filamentous protein in the pathogenesis of neurodegenerative diseases is not exclusive to tau and to AD. Recently, it has been shown that ␣-synuclein is phosphorylated in synucleinopathy lesions and that its phosphorylation promotes fibril formation in vitro (48).
Previous studies of the effect of phosphorylation on mutant tau proteins are quite variable. In cellular systems, it has been reported that mutant tau proteins are phosphorylated equally or less than wild-type tau (49 -51). The discrepancy between the findings of these previous studies and our results might be due to the absence of certain kinases in the cell lines used, to the different isoforms of tau transfected, and/or to the fact that tau can be microtubule-bound in the cell system. In the present study, one of the most remarkable differences we found with 4L was the 12-times faster phosphorylation at Ser-396 and more than 4 times greater total phosphorylation at this site in 4L R406W . In cell lines, 4L R406W seems not to be phosphorylated at Ser-396, although another study found it to be phosphorylated to a similar extent as the wild-type tau (49). Consistent with the present study, when 4L R406W was expressed in a neuronal cortical cell model (52), it was found to be phosphorylated at several sites, including Ser-396, and unable to bind to microtubules due to the phosphorylation state. In 4L R406W transgenic mice, the protein was found to be phosphorylated at several sites, including Ser-396, and to be aggregated into filaments (53). In Drosophila (43) Wild-type 4L and 3L and mutated 4L R406W , 4L V337M , 4L G272V , and 4L P301L , 0.5 mg/ ml, were incubated with rat brain extract as a source of brain kinases in the presence of ATP to induce hyperphosphorylation. Aliquots were taken at different times, spotted on nitrocellulose membrane, and overlaid with 5 g/ml 4L to determine the binding of normal to phosphorylated tau proteins as described. Note the differences in the scale. found to be highly phosphorylated at Ser-396 but was not aggregated into filaments. 4L G272V , -V337M , and -P301L were also expressed in mice, and increased tau phosphorylation and tau inclusions were found in these animals (54 -58).
For tau to assemble into filaments or to acquire the ability to bind normal tau, it is possible that phosphorylation of a particular site in a homologous or heterogeneous manner within a portion of the molecule must occur. If it were a particular site, it should have been consistently phosphorylated early in all three mutant tau proteins studied. If the distribution of charges is important, some sites nearby in the molecule could balance the rate of phosphorylation of a particular site with others. Although it is very difficult to distinguish between both alternatives, it appears that the need of a region of tau to be phosphorylated is more compatible with the results presented in the present study. For example, in the case of 4L R406W , the phosphorylation of Ser-404 was impaired, but that of Ser-396 was markedly increased. These findings are compatible with meeting the need for a balance of charges in the C-terminal region of tau more than the need of Ser-404 to be phosphorylated. Except for this effect on 4L R406W , the effect on the other mutated tau proteins seemed to be general for all sites. It appears that these mutations affect conformation-facilitating phosphorylation in general. In the case of 4L, at least 10 mol of phosphate are required for its polymerization (32), and according to phosphorylation kinetics, 10 mol of phosphate per mol of protein are added during about 10 h of phosphorylation (see Fig. 2A). 4L acquires the maximal ability of binding normal tau at about 3-4 h of phosphorylation (see Fig. 5B), i.e. after the incorporation of about 4 mol of phosphate per mol of protein. These results suggest that at least two different conformational states of tau are induced by phosphorylation, one in which the hyperphosphorylated tau is able to bind normal tau and one in which it is able to self-assemble into filaments. These results combined with the data of phosphorylation kinetics (see Fig. 3C) suggest that the conformation of tau needed to sequester normal tau might involve phosphorylation of 4L at positions 199, 202, 205, 212, 235, 262, and 404, and for self-assembly, further phosphorylation at positions 231, 396, and 422. Phosphorylation at Thr-181 or Thr-217, sites with slower kinetics, might control the assembly, or these sites might be phosphorylated on the polymer.
In the case of the mutant tau proteins in general, polymerization and the ability to bind normal tau are acquired earlier in the phosphorylation reaction; that is, lower levels of phosphate for self-assembly (mutated tau, 0.5-4 h, and a stoichiometry of 4 -6 mol of phosphate per mol of protein versus wild-type tau,10 h, and ϳ10 mol of phosphate per mol of protein, Fig. 2A) and for the binding, the same level of phosphate as wild-type tau but achieved faster due to increased rate of phosphorylation (mutated tau, 0.5 h versus wild-type tau, 2 h). Involvement of the different phosphorylation sites in these two different FIG. 6. A hypothetical scheme of the phosphorylation-induced self assembly of wild-type and FDTDP-17 mutated tau proteins. Tau self-assembles mainly through the microtubule binding domain/repeat R3 in 3R tau proteins and through R3 and R2 in 4R tau proteins (R2 and R3 have ␤-structure). Regions of tau molecule both N-terminal and C-terminal to the repeats are inhibitory. Hyperphosphorylation of tau neutralizes these basic inhibitory domains, enabling tau-tau interaction (phosphorylation sites indicated by red Ps). In the case of the C-terminal region beyond Pro-397 (398 -441), a highly acidic segment masks the repeats. Phosphorylation (red Ps) of tau at Ser-396 and/or 404 opens this segment, allowing tau-tau interaction through the repeats. FTDP-17 mutations make tau a more favorable substrate for phosphorylation than the wild-type 4L, which in turn is more readily phosphorylated than 3L. The mutated tau proteins achieve the conformation required to self-assemble at a lower level of incorporated phosphate. Although the FTDP-17 mutant tau proteins have conformations that are more prone to polymerize, in the absence of hyperphosphorylation, the highly basic segments and the C terminus interfere with polymerization. The conformation of the mutant tau proteins (mutation positions indicated in black) is more favorable for the action of kinases than 4L. Upon hyperphosphorylation (phosphorylation positions indicated in red Ps), both wild-type and mutated tau proteins adopt the conformation needed to polymerize into filaments. conformational states of tau in the case of the mutants appears to follow a similar pattern to that of 4L, except that in the case of 4L R406W , Ser-404 is barely phosphorylated, and the rate of Ser-396 phosphorylation is greatly increased. For the mutant tau proteins, it appears that phosphorylation at sites such as Thr-181 and Thr-217 is achieved on the polymer. This further phosphorylation might stabilize the polymers and induce lateral interactions.
The ability of P-tau to bind normal tau in all the tau proteins studied had a maximum at ϳ3.5-4 mol of phosphate per mol of protein and decreased upon higher mol of phosphate incorporation (Fig. 5). It is possible that once tau starts forming the polymer, the protein loses its ability to bind normal tau. These results are in agreement with our observations that AD P-tau binds normal tau more as a monomer than when it is aggregated into filaments. 2 We propose the scenario depicted in Fig. 6. Tau has been previously shown to undergo intermolecular association through the microtubule binding domains, and the flanking regions appear to be inhibitory of tau self-assembly (32,51,59). If we consider tau to have little secondary structure and that segments of tau are separated by prolines that induce a bend in the amino acid chain, the theoretical isoelectric points of these segments are strongly basic at the N-terminal side of microtubule binding domains. We propose that these very basic segments (pI Ͼ 9) in the proline-rich region of tau mask the intermolecular attraction of the microtubule binding domains. Considering that the mutant tau proteins start polymerizing after 1 or 2 h of incubation (4 -6 mol of phosphate per mol of protein), the sites that are phosphorylated at a level of about 50% of the total phosphorylation of 4L in less than 2 h are Thr-212, Ser-235, and Ser-262 in the N-terminal side. The pI value for the segment containing Thr-212 is 12, Ser-235 is 9.18, and Ser-262 is 9.56. Phosphorylation of these sites decreases the theoretical pI and, consequently, the negative effect of the N-terminal region on tau self-assembly. The C-terminal region is basic up to Pro-397, and the rest of the segments are acidic. It is possible that the acidic segment "masks" the interacting part of tau and that the phosphorylation at Ser-396 and/or Ser-404 opens up this segment, allowing the intermolecular interaction through the microtubule binding domains. In FTDP-17 mutant tau, the conformation is more prone to polymerize than the wild-type protein. In short, the FTDP-17 tau mutations induce a conformational change in tau molecule that makes it easier to polymerize and a more favorable substrate for kinases that polymerize at lower levels of phosphorylation. This implies that mutated tau proteins are more sensitive to changes in phosphorylation than wild-type tau and 4L more than 3L. Any imbalance that increases phosphorylation levels might trigger the polymerization of tau.
Clinically, FTDP-17 phenotypes appear to be quite variable, not only between mutations, but also within a single mutation and even within individual families either with intronic (60) or missense mutations (61). Despite the variability, some similarities are observed, as for example parkinsonism is present more frequently in families with intronic mutations (for review, see Ref. 62). This evidence suggests that there is a clinical overlap among the various tauopathies and that modifications of tau by other genetic and/or epigenetic factors, such as different expression levels of tau proteins, stress-activated kinases, and different levels of phosphatase activity, might also be involved in the disease.
In conclusion, the present study suggests that mutations alter tau conformation, making it more susceptible to hyperphosphorylation, and that the conformation is further altered by phosphorylation, making this phosphorylated tau able to sequester normal tau and then assemble into filaments at lower phosphorylation levels than wild-type tau. The autosomal dominant inheritance could be explained by the generation of the toxic form of tau that sequesters normal tau. Re-establishment of tight regulation of phosphorylation levels could arrest or prevent tau-induced neurofibrillary degeneration.