Site-specific Phosphorylation of Tau Accompanied by Activation of Mitogen-activated Protein Kinase (MAPK) in Brains of Niemann-Pick Type C Mice*

Niemann-Pick type C (NPC) disease is characterized by an accumulation of cholesterol in most tissues and progressive neurodegeneration with the formation of neurofibrillary tangles. Neurofibrillary tangles are composed of paired helical filaments (PHF), a major component of which is the hyperphosphorylated tau. In this study we used NPC heterozygous and NPC homozygous mouse brains to investigate the molecular mechanism responsible for tauopathy in NPC. Immunoblot analysis using anti-tau antibodies (Tau-1, PHF-1, AT-180, and AT-100) revealed site-specific phosphorylation of tau at Ser-396 and Ser-404 in the brains of NPC homozygous mice. Mitogen-activated proteinkinase,apotentialserinekinaseknowntophospho-rylatetau,wasactivated,whereasotherserinekinasessuchasglycogensynthasekinase-3 b and cyclin-dependent kinase 5 were inactive. Morphological examination demonstrated that a number

feature of NPC is the accumulation of low density lipoproteinderived cholesterol due to a defect in the sorting/trafficking of cholesterol from lysosomes and late endosomes (1,2). Neuropathologic examination has revealed neuronal distension, swollen axons, and polymorphorous cytoplasmic bodies that react with the cholesterol binding reagent, filipin. Since low density lipoprotein-derived cholesterol is inaccessible to the brain and nervous system, the accumulation of cholesterol in neurons must be derived from an additional source. Recent studies using NPC fibroblasts have concluded that endogenously synthesized cholesterol can contribute to cholesterol accumulation as a result of the circulation of cholesterol between the plasma membrane and endosomal/lysosomal compartments (3).
The gene responsible for NPC, referred to as the Niemann-Pick C1 gene (NPC1), was mapped to a region of chromosome 18 in both human and mice and subsequently cloned (4). Although the function of NPC1 remains undefined, studies demonstrate a crucial role for this protein in cholesterol metabolism (5)(6)(7). NPC mice share many of the pathophysiological abnormalities observed in patients with NPC (8), including the accumulation of cholesterol in tissues and neurodegeneration marked by decreased Purkinje cell numbers. Use of the murine model for NPC has provided important insights into the role of NPC1 in cholesterol metabolism (4). In a previous study it was reported that NPC mice are asymptomatic at birth, with the earliest definitive symptoms of the disease apparent by 4 -6 weeks of age and death ensuing by 10 -15 weeks of age (8).
It has been shown that the brains of NPC patients with neurodegeneration have neurofibrillary tangles (NFTs) without amyloid deposits (9)(10)(11). Interestingly, the presence of NFTs, which are composed of paired helical filaments (PHF), is one of the diagnostic hallmarks of Alzheimer's disease (AD) (12). A major component of PHF is tau, which is a microtubule-associated protein (13,14). It has previously been shown that the phosphorylation of tau prevents it from binding to microtubules (15)(16)(17)(18)(19). Although the phophorylation of tau in AD is the subject of intense investigation, the molecular mechanism responsible for this altered regulation remains to be defined. In this context, it is interesting to note that a perturbation in cholesterol metabolism and NFT formation without amyloid deposits coexist in the brains of NPC patients. This may indicate that a disturbance in cholesterol metabolism is responsible for tauopathy. We have recently demonstrated using cultured neurons that cholesterol deficiency results in axonal degeneration associated with microtubule depolymerization and hyperphosphorylation of tau (20). Therefore, it is important to inves-tigate the molecular mechanisms underlying tau phosphorylation associated with perturbed cholesterol metabolism in NPC brains.
Studies on the phosphorylation state of tau in the brains of NPC mice have not been performed. Evidence indicates that the defect in NPC1 function together with the perturbation of cholesterol metabolism may be an important tool for elucidating the pathways involved in the modulation of tau phosphorylation, NFT formation, and neurodegeneration. In this work, we have determined the phosphorylation state of tau in brains of BALB/c mice carrying the genetic mutation in NPC1. Our results demonstrate that tau is hyperphosphorylated at Ser-396 and Ser-404. The elevation of tau phosphorylation in NPC (Ϫ/Ϫ) mice was accompanied by activation of MAP kinase (MAPK). The activation of MAPK may result from decreased levels of cholesterol in cellular compartments due to defects in NPC1-mediated cholesterol trafficking.

EXPERIMENTAL PROCEDURES
NPC Mice-BALB/c mice carrying the genetic mutation for NPC1 were obtained from The Jackson Laboratory (Bar Harbor, MA). These heterozygous mice were bred to acquire NPC (ϩ/ϩ), NPC (ϩ/-), and NPC (Ϫ/Ϫ) mice used for the study. The genotypes of the mice were determined from genomic DNA isolated from tail-snip DNA using a polymerase chain reaction-based method and oligonucleotide primers described previously (4). Mice used in this study ranged from 6 to 12 weeks of age.
Antibodies-The monoclonal antibody Tau-1 was obtained from Chemicon International (Temecula, CA). The monoclonal antibody PHF-1 was kindly provided by Dr. P. Davies (Albert Einstein College of Medicine). The monoclonal antibodies AT-100 and AT-180 were purchased from Innogenetics (Ghent, Belgium). The rabbit polyclonal anti-NPC1 antibody was generated by immunizing rabbits with a MAPpeptide conjugate and purified from serum using peptide-specific affinity chromatography as previously described (21). Rabbit polyclonal anti-phospho-MAPK, rabbit polyclonal anti-phospho-independent-MAPK, and monoclonal anti-phospho-GSK-3␤ antibody were purchased from New England Biolabs (Beverly, MA). Rabbit anti-p35 antibody, which reacts with the p35 and p25 regulatory subunits of cyclin-dependent kinase 5, was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Fluorescein isothiocyanate-conjugated anti-mouse IgG and biotinylated goat anti-mouse IgG were purchased from Vector (Burlingame, CA).
Protein Preparation-Mouse tissues were homogenized in 10 volumes of Tris-saline (50 mM Tris-HCl, pH 7.4, 150 mM NaCl), containing protease inhibitors (Complete) and phosphatase inhibitors (10 M NaF and 1 mM orthovanadate) using a motor-driven Teflon homogenizer. The homogenates were centrifuged at 3,000 ϫ g for 10 min at 4°C, and supernatants were saved for biochemical analyses. Protein concentrations were determined using the bicinchoninic acid protein assay kit (Pierce). Aliquots of supernatant containing equal amounts of protein were subjected to sodium dodecylsulfate-polyacrylamide gel electrophoresis for immunoblot analysis.
Heat and Alkaline Phosphatase Treatment-Supernatants containing equal amounts of protein were heated to 95°C for 10 min and clarified by centrifugation at 20,630 ϫ g for 15 min. To each of these clarified heat-stable supernatants was added the same volume of saturated ammonium sulfate, and the mixtures were kept at 0°C for 1 h. The 20,630 ϫ g pellets were suspended in 50 mM Tris-HCl buffer, and alkaline phosphatase treatment was performed essentially as described elsewhere (22).
Immunoblot Analysis-Proteins separated using SDS-polyacrylamide gel electrophoresis were electrophoretically transferred onto a polyvinylidene difluoride membrane (Millipore, Bedford, MA). Nonspecific binding was blocked with 5% fat-free milk in phosphate-buffered saline containing 0.1% Tween 20. The blots were then incubated with primary antibodies overnight at 4°C. For the detection of both monoclonal and polyclonal antibodies, appropriate peroxidase-conjugated secondary antibodies were used in conjunction with SuperSignal chemiluminescence (Pierce) to obtain images saved on film.
Lipid Analysis-The concentration of cholesterol and phospholipid in samples were determined using enzymatic methods. Cholesterol was determined using a cholesterol determination kit, LTCII (Kyowa Medex, Tokyo), whereas phospholipid was determined using a phospholipid determination kit, PLB (Wako, Osaka, Japan).
Histological Analyses-The brains isolated from NPC (ϩ/ϩ) and NPC (Ϫ/Ϫ) mice were fixed in 10% formalin and embedded in paraffin for histological analysis. The tissue sections were washed four times with phosphate-buffered saline (PBS) and stained with a filipin solution for 1 h at room temperature (RT) before washing and mounting for microscopy. The fixed cells were examined using a UV filter. For immunohistochemistry, the sections were washed in PBS, incubated with 5% normal goat serum in PBS for 1 h at RT, followed by incubation with primary antibody against PHF-1 (1:20 dilution) for 1 h at RT. The sections were then washed in PBS three times and incubated with fluorescein isothiocyanate-conjugated anti-mouse IgG (1:5,000 dilution) or biotinylated goat anti-mouse IgG (1:200 dilution) for 1 h at RT. Fluorescein isothiocyanate-labeled sections were washed in PBS and observed using a fluorescence microscope. Biotin labeled sections were washed in PBS five times and visualized with 3Ј,3Ј-diaminobenzidine chloride.
Electron Microscopy-The cerebellum of NPC (ϩ/ϩ) and NPC (Ϫ/Ϫ) mice were washed with PBS and fixed with 2.0% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.4, containing 0.8% saccharose for 3 h. The cerebellum were washed in 0.1 M cacodylate buffer containing 0.8% saccharose overnight at RT and then incubated with 1% osmium tetroxide in the same buffer for 3 h. After dehydration, the cells were embedded in epoxy resin. Ultrathin sections were doubly stained with 2% uranyl acetate for 10 min and with a lead-staining solution for 5 min and observed using a transmission electron microscope (JEOL JEM-1200EX).

RESULTS
The expression of NPC1 in the cerebrums, cerebellums, and livers of NPC (ϩ/ϩ), NPC (ϩ/Ϫ), and NPC (Ϫ/Ϫ) mice were determined using immunoblot analysis. Consistent with previous results (21), NPC1 was determined to be a 180-kDa protein in the livers of NPC (ϩ/ϩ) and NPC (ϩ/Ϫ) mice ( Fig. 1). NPC (ϩ/Ϫ) livers expressed approximately half the amount of NPC1 that NPC (ϩ/ϩ) livers expressed, whereas NPC (Ϫ/Ϫ) livers expressed no detectable NPC1 due to a truncation mutation. The lower band seen in the livers was a nonspecific one, because it was also detected using preimmune serum (data not shown). In the cerebrums and cerebellums of NPC (ϩ/ϩ) and NPC (ϩ/Ϫ) mice, NPC1 was determined to be a 165-kDa protein, whereas NPC1 was not present in samples from the NPC (Ϫ/Ϫ) mice (Fig. 1). As with the liver, NPC (ϩ/Ϫ) cerebrums and cerebellums expressed approximately intermediate levels of NPC1. Immunoblot analysis of each sample using preimmune serum or anti-NPC1 in the presence of immunizing peptide resulted in the detection of no bands corresponding to the above molecular weights, indicating that these lower molecular weight bands represent NPC1. The amino acid sequence based on the cDNA of NPC1 predicts a molecular mass of 142 kDa. The difference in the apparent molecular mass, i.e. 165 kDa in the brain and 180 kDa in the liver, is likely due to altered glycosylation of NPC1 (23).
The concentrations of cholesterol in the cerebrum, cerebellum, and liver were not significantly different in 6-week-old NPC(ϩ/ϩ), NPC(ϩ/Ϫ), and NPC(Ϫ/Ϫ) mice (Fig. 2a). However, by 10 -12 weeks of age, NPC(Ϫ/Ϫ) mice had a significantly increased concentration of cholesterol in the cerebrum (2.5fold) and cerebellum (2.8-fold) compared with NPC (ϩ/ϩ) mice. As reported previously, the liver of NPC (Ϫ/Ϫ) mice had an FIG. 1. Immunoblot analysis of NPC1 in NPC (؉/؉), NPC (؉/؊), and NPC (؊/؊) mouse tissues. An equivalent amount of postnuclear supernatant protein was prepared from cerebrum, cerebellum, and liver for immunoblot analysis according to the procedures described under "Experimental Procedures." Proteins were separated using 7.5% SDSpolyacrylamide gel electrophoresis, transferred onto an Immobilon membrane, and probed with anti-NPC1. Film was used to obtain images of the protein bands using enhanced chemiluminescence. approximate 10-fold increase in the amount of cholesterol compared with NPC (ϩ/ϩ) mice (24 -26) (Fig. 2b). Moreover, the concentration of phospholipids in the liver of 10 -12 week old NPC (Ϫ/Ϫ) mice was 1.8-fold higher than those in NPC (ϩ/Ϫ) and NPC (ϩ/ϩ) mice of the same age. The concentrations of phospholipids in the cerebrum and cerebellum were similar between the three genotypes.
In direct support of the increased amount of cholesterol measured in 10 -12-week-old NPC (Ϫ/Ϫ) mice, filipin staining demonstrated an accumulation of cholesterol in neurons, particularly in the Purkinje cells of a NPC (Ϫ/Ϫ) mouse (Fig. 3b,  arrows), that was not observed in the neurons from a NPC (ϩ/ϩ) mouse (Fig. 3a, arrowheads). Moreover, an electron micrograph of cerebellum sections from a 12-week-old NPC (Ϫ/Ϫ) mouse shows polymorphorous cytoplasmic bodies, loosely packed multi-concentric lamellar-like structures in a Purkinje cell, and a smaller type of neuron (Fig. 3, c and d). These features are characteristic of the intracellular accumulations of cholesterol in the brains of patients with NPC (27).
Several well characterized antibodies to tau were used to examine its presence in various tissues of mice from 10 to 12 weeks of age. The immunoblot analysis of tau using Tau-1 antibody, which recognizes dephosphorylated sites of tau at four closely located serine residues, Ser-195, Ser-198, Ser-199, and Ser-202 (28), shows that tau in the cerebrum, cerebellum, and liver was found to have apparent molecular masses between 50 and 70 kDa (Fig. 4a). For samples derived from NPC (Ϫ/Ϫ) and NPC (ϩ/Ϫ) mice, the main bands immunoreactive to Tau-1 appear to exhibit slower electrophoretic mobility than the bands for samples derived from NPC (ϩ/ϩ) mice. (Fig. 4, a  and b). These bands of tau exhibiting slower mobility are known to be characteristic of phosphorylated tau. The increased level of tau phosphorylation in NPC (Ϫ/Ϫ) mice was marked in the cerebellum, which is the most commonly affected region in NPC brains. Thus, we focused on the cerebellum to examine the phosphorylation state of tau. The antibodies used were the site-specific phospho-dependent antibodies PHF1, AT-180, and AT-100, which recognize the phosphorylated tau epitopes Ser-396/Ser-404, Thr-231, and Ser-214/Thr-217, respectively. Only the upper migrating band representing tau was strongly reactive to PHF-1 when samples from the cerebellum of NPC (Ϫ/Ϫ) mice were analyzed. This was not ob- Aliquots of postnuclear supernatants from the cerebrum, cerebellum, and liver were prepared according to the procedures described under "Experimental Procedures." The protein concentration of each sample was determined. The amount of cholesterol and phopholipid from each sample was determined using kits. The amount of cholesterol and phospholipid from each sample was normalized to mg of protein from 6-week-old mice (a) and 10 -12-week-old mice (b). Data represent the mean Ϯ S.E. of three samples for the NPC (ϩ/ϩ) and NPC (Ϫ/Ϫ) mice except for the NPC (ϩ/Ϫ) mice at 12-weeks of age (n ϭ 1). *, p Ͻ 0.05 versus NPC (ϩ/ϩ); **, p Ͻ 0.001 versus NPC (ϩ/ϩ), and ***, p Ͻ 0.01 versus NPC (ϩ/ϩ). served when samples from the cerebella of NPC (ϩ/Ϫ) and NPC (ϩ/ϩ) mice were analyzed (Fig. 4b), indicating that the Ser-396/Ser-404 sites were highly phosphorylated in NPC (Ϫ/Ϫ) mice. In contrast, the intensity of the band reactive to AT-100 and AT-180 was stronger for NPC (ϩ/ϩ) mice than for NPC (ϩ/Ϫ) and NPC (Ϫ/Ϫ) mice. However, the mobility of the AT-100-and AT-180-reactive band for NPC (Ϫ/Ϫ) mice was reduced to that of the band for NPC (ϩ/ϩ) and NPC (ϩ/Ϫ) mice (Fig. 4b).
Alkaline phosphatase treatment, which induces protein dephosphorylation, revealed a detailed pattern of the tau isoforms with significant shifts in electrophoretic mobility (Fig.  4c). After dephosphorylation, tau from the cerebellum was resolved into three major isoforms with apparent molecular masses of 62, 58, and 52 kDa (Fig. 4c). These three isoforms were detected in samples derived from each genotype. Importantly, the heat-stable supernatant of samples from the cerebellum of NPC (Ϫ/Ϫ) and NPC (ϩ/Ϫ) mice showed that the bands immunoreactive with Tau-1 exhibited a slower electrophoretic mobility than those from NPC (ϩ/ϩ) mice (Fig. 4c,  open triangles).
Immunohistochemical analysis using PHF-1 was performed to determine whether tau phosphorylation was also evident in brain slices. Some of the cerebellum granular neurons (Fig. 5a) and cerebrum cortical neurons (Fig. 5c) in the brain slices of NPC (Ϫ/Ϫ) mice at 10 weeks of age were shown to be PHF-1 immunopositive, whereas those in the brain slice of NPC (ϩ/ϩ) mice were immunonegative (Fig. 5b), lending support to the results shown in Fig. 4b. We also performed electron microscopy to determine whether filaments with PHF exist in the neurons of the cerebrum and cerebellum. However, we could not detect PHF formation in these cells (data not shown).
To determine the molecular basis for increased tau phosphorylation, the expression and the phosphorylation state of several well known tau-directed protein kinases, including MAPK, GSK-3␤, and p25, were determined. Immunoblot analysis using the anti-phospho-MAPK antibody, which recognizes only an activated form of MAPK, revealed that brains from 12-weekold NPC (Ϫ/Ϫ) mice have an 11-fold increase in MAPK activity compared with the brains of 12-week-old NPC (ϩ/Ϫ) and NPC (ϩ/ϩ) mice (Fig. 6a). The overall level of MAPK was similar for each of the three genotypes (Fig. 6a). Immunoblot analysis using the anti-phospho-GSK-3␤ antibody showed no alteration in the amount of the active GSK-3␤ in the brains from the three genotypes (Fig. 6a). Immunoblot analysis using the anti-p35 and p25 antibodies also showed no alteration in the conversion of p35 to p25 from the brains of the three genotypes (Fig. 6a). Together, these results indicate that neither GSK-3␤ nor p25 are responsible for increased tau phosphorylation. In contrast, markedly increased levels of MAPK activity, in conjunction protein from the cerebrum, cerebellum, and liver were separated using 10% SDS-polyacrylamide gel electrophoresis. Separated proteins were transferred onto an Immobilon membrane and probed with antibodies against Tau-1 (a). Equivalent amounts of postnuclear protein were also subjected to immunoblot analysis using the site-specific phospho-tau antibodies, PHF-1, AT-100, and AT-180 in addition to Tau-1 (b). Equivalent amounts of postnuclear protein were also not treated (Ϫ) or treated (ϩ) with bacterial alkaline phosphatase (BAP) and subjected to immunoblot analysis using Tau-1 (c). Three independent experiments showed similar results.
with increased phosphorylation of tau, suggest that the MAPK pathway may be responsible for tauopathy in NPC. Additionally, NPC(Ϫ/Ϫ) mice have increased MAPK activity in the cerebellum (3-fold) compared with NPC(ϩ/ϩ) and NPC(ϩ/Ϫ) mice at 6 weeks of age (Fig. 6b); however, cholesterol accumulation was not found (Fig. 2a). DISCUSSION The mechanism by which NPC1 deficiency, associated with altered cholesterol metabolism, contributes to tauopathy remains undefined. In the present study, we have shown for the first time that (i) the accumulation of intracellular cholesterol in NPC mouse brains was a function of age, (ii) the levels of tau phosphorylation at Ser-396/Ser-404 are markedly increased, (iii) the increase in tau phosphorylation is marked in the cerebellum, which is the most commonly affected region of the brain in NPC, and (iv) MAPK is activated in the brains of NPC (Ϫ/Ϫ) mice.
Consistent with previous studies (1,5,24), our results have demonstrated that the concentration of cholesterol measured in the brains of NPC (Ϫ/Ϫ) mice at 6 weeks of age were not elevated compared with the brains of NPC (ϩ/ϩ) mice. However, using both biochemical and morphological analysis of NPC(Ϫ/Ϫ) mice, we demonstrate that the concentration of cholesterol measured in the brains of NPC (Ϫ/Ϫ) mice 10 -12 weeks of age were elevated, similar to other organs (24 -26). Cholesterol accumulation has previously been shown to occur as a result of lipoprotein-cholesterol uptake via the coated-pit pathway (25). But a recent study suggests that NPC cells can accumulate cholesterol in the absence of low density lipoprotein, indicating that endogenously synthesized cholesterol and plasma membrane cholesterol can contribute to the accumulation of lysosomal cholesterol (3). In any event, our results indicate that the accumulation of cholesterol is age-dependent. Cholesterol accumulation is detected in 10-to 12-week-old mice but not detected in 6-week-old (this study) or 7-week-old mice, as shown in other studies (24).
There are several possible mechanisms for the increased phosphorylation of tau at the PHF-1 epitope: (i) a decrease in the level of cholesterol at specific cellular compartments due to a defect in cholesterol trafficking, (ii) the accumulation of cholesterol and other lipids in the lysosomal/late endosomal compartment, or (iii) the direct result of a defect in NPC1 function. Based on the results of our study and those presented previously by other investigators, the evidence indicates that a decrease in the level of available cholesterol due to a defect in cholesterol trafficking is responsible for tau phosphorylation in the brains of NPC (Ϫ/Ϫ) mice. Previous studies show that, despite an accumulation of cellular cholesterol, the rate of cholesterol synthesis and the expression of low density lipoprotein receptors is not down-regulated (24). Moreover, the esterification of excess cholesterol at the endoplasmic reticulum is delayed as a result of cholesterol not being able to gain access to pools responsible for maintaining intracellular cholesterol homeostasis (26,29). Consistent with this hypothsis, our results suggest that additional cellular compartments may be cholesterol-deficient and, therefore, directly responsible for inducing tau phosphorylation and microtubule depolymerization in the axons of cultured neurons (20).
Cholesterol depletion of plasma membrane caveolae have been shown to cause the activation of MAPK (30). Consistent with these results, we have found that MAPK is activated in cholesterol-deficient neurons and that NPC fibroblasts have significantly decreased levels of caveolae cholesterol (data not shown). In the present study, we demonstrate that MAPK activity is elevated in the brains of NPC (Ϫ/Ϫ) mice not only at 10 -12 weeks of age but also at 6 weeks of age, indicating that the activation of MAPK precedes the accumulation of cholesterol. Since tau is known to be a substrate for MAPK and that Ser-396/Ser-404, but not Ser-214, Thr-217, or Thr-231 sites, are phosphorylated by MAPK (31)(32)(33), the assumption that MAPK is responsible for tau phosphorylation in NPC (Ϫ/Ϫ) brains would well explain our present results that tau was phospho- FIG. 6. MAPK is phosphorylated and activated in the cerebellums of NPC (؊/؊) mice. The cerebellums of mice 6 weeks old (b) and 10 -12 weeks old (a) were analyzed for MAPK phosphorylation and activation. MAPK activation was determined by immunoblot analysis of postnuclear supernatants using the anti-phospho-MAPK antibody that primarily recognizes activated MAPK. Total MAPK was similarly determined using the anti-MAPK antibody. The level of activation of GSK-3␤ was determined using anti-phospho-GSK-3␤ antibody. Conversion of p35 to p25 was determined using anti-p35 antibody, which reacts with the p35 and p25 regulatory subunits of cyclin-dependent kinase 5. rylated at Ser-396 and Ser-404 (demonstrated as PHF-1-immunopositive) and was not phosphorylated at Ser-214 and Thr-217 (demonstrated as AT-100-immunopositive) nor at Thr231 (demonstrated as AT-180-immunopositive). These results indicate that a decrease in the amount of cholesterol available at particular cellular compartments may be directly responsible for the stimulation of MAPK activity and the subsequent promotion of site-specific phosphorylation of tau in the brains of NPC (Ϫ/Ϫ) mice and NPC patients. The mechanisms underlying time discrepancy in the activation of MAPK and the elevation of tau phosphorylation, however, remain unclear. But there are several studies demonstrating the relationship between the kinetics of kinases and tau phosphorylation and/or NFT formation. It has been shown that activation of the MAPK cascade is most pronounced during early stages of AD (34), suggesting that activation of MAPK may precede tauopathy found in AD. It has also been suggested that responsible kinases including MAPK work at different stages or varying durations in Alzheimer's disease (32). In addition, tau antigenicity and its behavior has been shown to be influenced by the sequential and convergent influences of multiple kinases (35). These lines of evidence suggest that tau phosphorylation is modulated by a complex mechanism and only allows us to argue that activation of a given kinase may precede tau phosphorylation. However, direct evidence is indeed required to elucidate the precise kinetics by which responsible kinases, including MAPK and phosphatases, induce tau phosphorylation and contribute to NFT formation in vivo.
The remaining question to be answered is why tau is sitespecifically phosphorylated at Ser-396/Ser-404 sites in the brains of NPC (Ϫ/Ϫ) mice, whereas other sites, including Ser-214, Thr-217, and Thr-231 are relatively dephosphorylated compared with the brains of NPC (ϩ/ϩ) and NPC (ϩ/Ϫ) mice. The molecular mechanism underlying this regulation remains unclear, and additional studies are necessary. Our finding suggests that MAPK, but neither GSK-3␤ nor cyclin-dependent kinase 5, modulates the phosphorylation of tau in a site-specific manner. Other proteins, including tau-directed kinases and phosphatases, may also be involved in the regulation of tau phosphorylation and dephosphorylation in the brains of NPC (Ϫ/Ϫ) mice. Consistent with alterations in the activity of specific kinases, our studies have shown that the expression of specific protein kinases and phosphorylated state of caveolin-1 and annexin II are increased in the livers of NPC (Ϫ/Ϫ) mice (36).
The association between cholesterol deficiency and the promotion of tau phosphorylation may provide clues to the pathogenesis of tauopathy in AD, because possession of the allele ⑀4 of apolipoprotein E, which is a key molecule regulating cholesterol metabolism in the central nervous system, has been found to be a strong risk factor for the development of AD (37)(38)(39). Recently, we have demonstrated that oligomeric amyloid ␤-protein reduces the intracellular cholesterol content in neurons, 2 which may result in tau phosphorylation (20). Taken together, these findings may suggest that cholesterol plays a critical role in the tauopathy of AD and NPC.