Assembly of tau in transgenic animals expressing P301L tau: alteration of phosphorylation and solubility

Authors


Address correspondence and reprint requests to Shu-Hui Yen, Department of Neuroscience, Mayo Clinic Jacksonville, 4500 San Pablo Road, Jacksonville, FL 32224, USA. E-mail: Yen.Shu-Hui@Mayo.Edu

Abstract

Transgenic mice (JNPL3), which develop neurofibrillary degeneration and express four-repeat human tau with P301L missense mutation, were characterized biochemically to determine whether the development of aggregated tau from soluble tau involves an intermediate stage. Homogenates from mice of different ages were separated into buffer-soluble (S1), sarkosyl- and salt-extractable (S2) and sarkosyl-insoluble pellet (P3) fractions, and analyzed for human tau distribution, phosphorylation and filament formation. S1 and S2 fractions contained 50–60-kDa tau whereas the S2 fraction also had 64-kDa tau. The level of tau in the P3 fraction increased in an age-dependent manner and correlated positively with the soluble tau concentration. The P3 fraction from 2.5–6.5-month-old mice contained 64- and 50–60-kDa tau, whereas that from 8.5-month and older transgenic animals contained mostly 64-kDa and higher molecular weight tau. The S2 and P3 fractions contained comparable amounts of 64-kDa tau. The 64-kDa tau was predominantly human, and phosphorylated at multiple sites: Thr181, Ser202/Thr205, Thr212, Thr231, Ser262, Ser396/Ser404, Ser409 and Ser422. Most of these sites were phosphorylated to a lesser extent in S2 than in P3 fractions. Tau polymers were detected in P3 fractions from 3-month and older female JNPL3 mice, but not in non-transgenic controls. The results suggest that tau in S2 represents an intermediate from which insoluble tau is derived, and that phosphorylation may play a role in filament formation and/or stabilization.

Abbreviations used
AD

Alzheimer's disease

FTDP-17

frontotemporal dementia with parkinsonism linked to chromosome 17

MT

microtubule

PMSF

phenylmethylsulfonyl fluoride

TBS

Tris-buffered saline

Tau is a microtubule (MT)-associated protein encoded by a single gene on human chromosome 17. Exons 9–12 of the tau gene encode a tandem repeat region important for MT binding and self-interaction. Alternative splicing of exon 10 (E10) gives rise to tau isoforms with three (E10 –) or four (E10 +) repeats, whereas splicing of exons 2 and 3 generates tau with and without one or both N-terminal inserts (0 N, 1 N, 2 N). The adult human brain expresses all six isoforms, whereas fetal human brain expresses only the shortest three-repeat isoform. In adult mice, the four-repeat tau isoforms are predominantly expressed (Kosik et al. 1989). In normal brains most of the tau proteins are soluble, but in Alzheimer's disease (AD) and related disorders, including frontotemporal dementia and parkinsonism linked to chromosome 17 (FTDP-17), they become insoluble and accumulate as filamentous structures. The mechanisms that lead to the transformation of soluble tau into filamentous structures remain unknown.

Genetic analyses have linked mutations in the tau gene to FTDP-17 (Hutton et al. 1998; Poorkaj et al. 1998). To date, over 20 distinct pathogenic mutations in the tau gene have been identified (see review by Lee et al. 2001). In vitro studies have demonstrated that many of the missense mutations disrupt the binding of tau to MT (Hasegawa et al. 1998; Hong et al. 1998; DeTure et al. 2000) and alter the ability of tau to form filaments (Nacharaju et al. 1999; Arrasate et al. 1999; Goedert et al. 1999; Barghorn et al. 2000; Gamblin et al. 2000).

Transgenic mice expressing mutant or wild-type tau have been generated in several laboratories (Gotz et al. 1995, 2001a,b; Brion et al. 1999; Ishihara et al. 1999; Spittaels et al. 1999; Lewis et al. 2000; Probst et al. 2000; Tanemura et al. 2001). The JNPL3 line of transgenic mouse expresses tau with the P301L mutation and is a model of neurofibrillary degeneration, in which there is also neuronal loss, axonal degeneration and motor impairment (Lewis et al. 2000). Filamentous tau inclusions are detected in neurons, and sarkosyl-insoluble tau proteins of 64 kDa and higher molecular weights are found in lysates from the brain and spinal cord of JNPL3 mice. Sarkosyl-insoluble tau is detected in 2–3-month-old hemizygous JNPL3 mice, but neurofibrillary tangles are detected consistently only in older animals. The level of sarkosyl-insoluble tau appears to differ between male and female mice, with higher levels in females (Lewis et al. 2001).

Dephosphorylation studies of samples derived from JNPL3 mice and humans with AD or FTDP-17 have shown that the 64-kDa and higher molecular weight bands contain insoluble hyperphosphorylated tau (Lewis et al. 2000). Phosphorylation has been proposed to play a role in changing the conformation of tau to a state favorable for self-interaction (Jicha et al. 1997; Daly et al. 2000). In vitro studies of AD brain tau have shown that hyperphosphorylated tau has decreased ability to promote MT polymerization, and is even able to strip normal tau from MTs (Alonso et al. 1996; 2001). Multiple phosphorylated sites have been identified in abnormal tau from AD and related disorders (Buee and Delacourte 1999). However, the conditions that govern the in vivo formation of tau filaments have been poorly investigated so far. Furthermore, interpretation of results from studies of autopsy samples can be complicated by the variability in disease severity, agonal state, age, environmental factors and post-mortem delays. These complexities can be reduced or avoided by studies of animal models that develop tauopathy.

In this paper, we report a study of JNPL3 mice, focusing on issues that were not addressed before. We demonstrate that the mice contain a pool of tau that differs from normal and sarkosyl-insoluble tau in solubility and the extent of phosphorylation. In addition, we show that the morphology of insoluble tau derived from three age groups of mice is similar, and that the extent of insoluble tau formation correlates with the level of human tau expression.

Materials and methods

Transgenic mice

Transgenic mice and non-transgenic littermates were bred by mating hemizygous JNPL3 mice with Swiss Webster mice (Taconic, Germantown, NY, USA). Animals were maintained on largely Swiss Webster background with minimal contribution of C57BL6 and DBA2. Mice were genotyped for the tau transgene by PCR between exons 1 and 5 of human tau cDNA against an internal PS-2 control PCR. They were killed 2.5–10.5 months after birth. Ten 2.5–3.5-month-old mice (five female, five male), ten 6–6.5-month-old mice (five female, five male) and seven 8.5–10.5-month-old mice (three female, four male) were studied. Procedures involving animals and their care were approved by the Mayo Clinic Animal Care and Use Committee in compliance with National Institutes of Health guidelines and policies.

Brain tissue extraction

Hemibrains were separated into cortico-limbic (cortex, amygdala and hippocampus) and subcortical (basal ganglia, diencephalon, brain stem and cerebellum) regions, quickly frozen on dry ice and stored at −80°C. Each brain piece was weighed and homogenized in three volumes of Tris-buffered saline (TBS) containing protease and phosphatase inhibitors (25 mm Tris-HCl, pH 7.4, 150 mm NaCl, 1 mm EDTA, 1 mm EGTA, 5 mm sodium pyrophosphate, 30 mmβ-glycerophosphate, 30 mm sodium fluoride, 1 mm phenylmethylsulfonyl fluoride (PMSF)). The homogenates were centrifuged in a Beckman TLA100.2 rotor (Beckman, Palo Alto, CA, USA) at 150 000 g for 15 min at 4°C. Supernatants were collected as S1 fractions, and the pellets (P1) were re-homogenized in three volumes of salt/sucrose buffer (0.8 m NaCl, 10% sucrose, 10 mm Tris/HCl, pH 7.4, 1 mm EGTA, 1 mm PMSF) and centrifuged as above. The resulting pellets were discarded and the supernatants were incubated with sarkosyl (Sigma, St Louis, MO, USA; 1% final concentration) for 1 h at 37°C. The sarkosyl mixtures were then centrifuged in a Beckman TLA100.2 rotor at 150 000 g for 30 min at 4°C. The supernatants (S2 fraction) were collected, and the pellets (P3) were resuspended in 50 µL TBS and used immediately for immunogold electron microscopy or stored at −80°C for western blotting. Some P1 fractions were (twice) resuspended repeatedly with TBS, re-homogenized and centrifuged. The pellets were extracted with salt/sucrose buffer and processed as described above to obtain salt/sarkosyl-soluble fractions and sarkosyl-insoluble fractions, referred to as S2′ and P3′ respectively.

Western blot analysis

Samples were prepared by adding 6 × Laemmli sample buffer to the supernatant and pellet fractions, and boiled for 5 min. The amount of sample loaded per lane was based on the initial wet-weight of frozen brain tissue. The samples were separated by gel electrophoresis on 10% sodium dodecyl sulfate–polyacrylamide gels and transferred to nitrocellulose membranes (Bio-Rad, Hercules, CA, USA). All blots were incubated with a blocking solution containing 5% milk proteins, 0.1% goat serum and 0.1% Tween-20 in TBS for 30 min, before an overnight incubation with various tau antibodies at 4°C. The blots were then washed five times with TBS containing 0.1% Tween-20, and incubated for 1 h at room temperature (25°C) with peroxidase-conjugated, goat anti-rabbit (1 : 4000; Chemicon, Temecula, CA, USA), anti-mouse IgG (1 : 2000; Bio-Rad), anti-mouse IgG + IgM (1 : 1000; Roche, Indianapolis, IN, USA) or anti-mouse IgG3 (g3 chain specific) antibody (1 : 500; Southern Biotechnology Associates, Birmingham, AL, USA). After washing, the blots were developed using the enhanced chemiluminescence system (ECL plus; Amersham Biosciences, Piscataway, NJ, USA). Immunoreactivity of tau proteins was analyzed from scanned films using MCID software (Imaging Research Inc., Ontario, Canada). Recombinant tau in three different concentrations was loaded to ensure that the immunoreactivity detected in brain fractions was within the linear range. To compare the relative amount of tau protein, the same volume of recombinant tau was loaded as a control in all western blot experiments. After the densities of the immunoreactive bands had been measured with MCID, the relative amounts of tau protein in different blots were normalized with recombinant tau standards.

Antibodies

Polyclonal tau antibodies E1, WKS44, WKS45 and WKS46 were raised against tau polypeptides corresponding to amino acid residues 19–33, 162–178, 258–266 and 350–370 respectively (Kenessey et al. 1997). Mouse monoclonal antibodies CP13, TG3, PHF1 and PG5 that recognize sites phosphorylated, respectively, at Ser202/Thr205, Thr231/Ser235, Ser396/Ser404 and Ser409 were provided by Dr P. Davies (Albert Einstein College of Medicine, Bronx, NY, USA; Jicha et al. 1999; Weaver et al. 2000). Tau46, which recognizes tau peptides corresponding to amino acids 404–441, and T49, which is specific for mouse tau, were provided by Dr V. M. Lee (University of Pennsylvania, PA, USA). AT270 and AT180, reactive with sites phosphorylated at Thr181 and Thr231 respectively, were purchased from Pierce Endogen (Rockford, IL, USA). Polyclonal antibodies pT212, pT231, pS262 and pS422, recognizing sites phosphorylated at Thr212, Thr231, Ser262 and Ser422 respectively, were purchased from BioSource International (Camarillo, CA, USA). For western blotting, tau antibodies were used at the following dilutions in blocking solution: affinity-purified rabbit polyclonal E1, 1 : 1000; WKS44, 1 : 1000; WKS45, 1 : 2000; WKS46, 1 : 1000; pT212, 1 : 500; pT231, 1 : 1000; pS262, 1 : 500; pS422, 1 : 1000; Tau46, 1 : 50000; T49, 1 : 500; PHF1, 1 : 1000; CP13, 1 : 200; TG3, 1 : 3; AT270, 1 : 1000; and PG5, 1 : 50. Monoclonal anti-β-tubulin antibody TUB 2.1 (Sigma) was used at 1 : 10000 dilution.

Immunogold electron microscopy

Sarkosyl-insoluble materials were adsorbed on carbon/formvar-coated 400 mesh copper grids (EM Sciences, Fort Washington, PA, USA) for 60 s. The grids were washed three times with filtered TBS, blocked for 15 min in a filtered TBS solution containing 0.04% bovine serum albumin and 2% horse serum, and then incubated for 2 h with primary antibody. Excess antibodies were removed by washing three times in blocking solution, and the grids were incubated for 1 h in Aurogold 5-nm gold-conjugated anti-rabbit or 10-nm gold-conjugated anti-mouse secondary antibody (1 : 20 dilution; Amersham Biosciences). After washing five times with TBS, the grids were stained with 2% uranyl acetate for 45 s and examined with a Philips 208S electron microscope (Philips, Hillsboro, OR, USA). All antibodies were diluted with the blocking reagent. The tau primary antibodies used for electron microscopy were mouse monoclonal Tau46 (1 : 50) and rabbit polyclonal E1 (1 : 20).

Statistical analysis

The correlation between age and human tau distribution, as well as the correlation of age, gender and regions to the relative amount of tau protein in the soluble fraction, were tested by anova and multiple comparison procedures of the Student Newman–Keuls test. The correlation between the levels of soluble and sarkosyl-insoluble tau was tested using Pearson product moment correlation. Data were analyzed with Sigma Stat for Microsoft Windows, version 2.03 (SPSS Science, Chicago, IL, USA), and significance levels were set at p < 0.05.

Results

Gel electrophoretic profiles and relative level of human tau

Brain tissues of JNPL3 mice contained tau proteins with different solubilities that could be separated into buffer-extractable (S1), high-salt-extractable (S2) and sarkosyl-insoluble (P3) fractions. Western blotting with a human tau-specific antibody, E1, demonstrated that the gel electrophoretic profiles of tau in these fractions were different. S1 and S2 fractions from both cortico-limbic and subcortical regions of all animals contained human tau of 50–60 kDa in molecular weight (Fig. 1a). The S2 fraction from subcortical regions of older animals also contained tau proteins of 64 kDa. Tau of this molecular weight was detected weakly in the S1 fraction from few old animals. Tau proteins of 64 kDa and higher molecular weight were detected in the P3 fraction, and were also more abundant in the subcortical than cortico-limbic samples, consistent with previous findings that subcortical regions display more severe neurofibrillary pathology than cortico-limbic regions (Lewis et al. 2001). The generation of both 64-kDa and higher molecular weight tau species appeared to be accompanied by a decrease in 50–60-kDa tau species (compare S2 and P3 fractions), suggesting that formation of 64-kDa tau precedes that of higher molecular tau and is associated with a decreases in tau solubility. As reported before (Lewis et al. 2000), we did not detect the 64-kDa tau in the P3 fraction prepared from transgenic mice expressing wild-type human tau. Neither did we detect 64-kDa tau in an S2 fraction derived from wild-type human tau transgenic animals (data not shown).

Figure 1.

Distribution of human tau in different brain fractions from JNPL3 mice. (a) Immunoblotting of buffer-soluble (S1), high-salt and sarkosyl-soluble (S2) and sarkosyl-insoluble (P3) fractions with a human tau-specific antibody, E1. The fractions were derived from (1) cortico-limbic and (2) subcortical brain regions of ∼3-, ∼6- and ∼9-month (M)-old mice. The location of 64-kDa tau (arrowhead) and recombinant four-repeat tau isoform without the N-terminal inserts (4R0N) is indicated. S1, S2 and P3 fractions were loaded at a ratio of 1 : 5 : 75 (based on wet-weight). (b) Immunoblotting of brain homogenates and fractions from a 9-month-old mouse. Brain homogenate (H), S1, S2 obtained from extraction of TBS-washed P1 (S2′) and P3 derived from TBS-washed samples (P3′) were probed with E1 antibody. The location of 64-kDa tau (arrowhead) and recombinant tau (rT) containing four- and three-repeat tau isoforms without the N-terminal inserts (4R0N and 3R0N respectively) is indicated. (c) The proportion of tau in different fractions from subcortical regions of mice of three different age groups.

To be certain that the 50–60-kDa tau found in the S2 fraction is not entirely due to trapping of S1, we analyzed the S2′ fraction, which was prepared from samples washed with TBS. Similar to S2, the S2′ fraction contained 50–60- and 64-kDa tau (compare Fig. 1(b) with Fig. 1(a), 9 months); as observed in P3, more 64-kDa tau than lower molecular weight tau was detected in the P3′ fraction.

The amount of human tau in subcortical regions of JNPL3 mice with different solubilities was determined by analysis of E1 immunoreactivity in all tau bands in different fractions, followed by correction of background and correction of differences in the total volume of the fractions and the volume of samples loaded on gels. The results (Fig. 1c) showed that human tau in transgenic mouse brain was recovered mostly in the S1 fractions, followed by S2 and P3 fractions. Statistical analysis showed no significant differences between different age groups in the relative amount of tau in different fractions.

Soluble tau protein

The level of tau expression in female and male JNPL3 mice of different ages was compared (Fig. 2a) by probing the S1 fraction with antibody to tau (E1) and tubulin (TUB 2.1), and by quantitation of immunoreactivity on western blots with densitometric scanning. The ratio of E1 to tubulin is shown in Fig. 2(d). Although S1 fractions from the cortico-limbic or subcortical regions from 3- and 6-month-old female mice (regardless of brain region) had an average tau/tubulin ratio higher than that of corresponding brain regions from 9 month-old females, the differences were not statistically significant. This may be due to variations between individual animals within a given age group, small sample sizes or the impact of increasing insoluble tau species. Comparison of cortico-limbic and subcortical regions from the same brain in terms of the level of tau in S1 showed that subcortical regions had significantly a higher tau/tubulin ratio than cortico-limbic regions (p = 0.009, n = 25; Mann–Whitney rank sum test). Without normalization with tubulin, the subcortical regions also showed a higher tau level than the cortico-limbic regions (p = 0.06, n = 25).

Figure 2.

Age, gender and brain region differences in the level of soluble and sarkosyl-insoluble tau in JNPL3 mice. The animals were coded ar. Human tau protein in two brain regions (1, cortico-limbic; 2, subcortical) from three females (F) and three males (Ma) per age group was detected by western blotting with E1 antibody. (a) S1 fraction; each lane contained 0.1 µL of S1 fraction diluted to 10 µL. (b) P3 fraction; samples from mice af and r, il, and other mice were derived from 5, 2.5 and 1.25 mg wet-weight of brain respectively. The adjustment was made to allow detection of insoluble tau in animals with lower levels of tau expression. (c) Immunoblotting of brain fractions with E1 and Tau46. Recombinant (rT) four-repeat and three-repeat tau without the N-terminal inserts encoded by exons 2 and 3 (4R0N and 3R0N respectively) were used as a reference to estimate the contribution of mouse tau in the P3 fraction. (d) and (e) Mean ± SEM values obtained from several animals with correction of differences in sample loading.

Sarkosyl-insoluble tau protein

The relative amount of sarkosyl-insoluble tau in female and male mice of different ages was compared by immunoblotting, using recombinant tau as a reference. Subcortical regions contained more insoluble tau than cortico-limbic regions, and the subcortical regions in male mice had less insoluble tau than age-matched female mice (Figs 2b and e). These findings are consistent with those reported previously (Lewis et al. 2001) and with the observation that transgene expression levels are higher in female mice (McGowan et al.). On average, the level of insoluble tau in subcortical regions of 6-month-old male mice was about 20% of that in age-matched female mice. As observed in studies of soluble tau, the level of insoluble tau varied among mice that were matched for age and sex. The data obtained from western blotting also agrees with previous immunocytochemical studies of JNPL3 mice, in which age- and sex-matched mice were shown to have variable degrees of tau pathology (Lewis et al. 2001).

To investigate whether mouse tau co-aggregated with human tau in JNPL3 mice, we probed the P3 fraction in duplicate with antibodies reactive with both human and mouse tau (Tau46) and antibodies reactive only with human tau (E1) (Fig. 2c). Recombinant human tau was used as a reference. The ratio of E1 immunoreactivity in P3 versus that in recombinant tau, and the ratio of Tau46 immunoreactivity in P3 versus that in recombinant tau were calculated and found to be comparable. The results indicate that tau in the P3 fraction is largely of human origin, as incorporation of substantial amounts of mouse tau in P3 would be reflected by a relatively higher Tau46 (recognizes mouse and human tau) to E1 (human specific) immunoreactivity. In addition, we probed the P3 fraction with a mouse-specific tau antibody and detected very little immunoreactivity (Fig. 2c). Similar analyses of S2 fractions indicated that most of the 64-kDa tau also represented transgenic human tau. In comparison, the S1 fraction contained similar amounts of human and mouse tau, which is consistent with our previous findings (Lewis et al. 2000).

Levels of sarkosyl-insoluble tau with respect to amount of soluble human tau

To determine whether the formation of insoluble tau is influenced by tau expression, the amount of human tau in S1 fractions was plotted against that in P3 fractions without considering the gender or age of the mice (Fig. 3a). There was a positive correlation between the levels of soluble and insoluble tau (R = 0.672, p <  0.001, n = 44; Pearson product moment correlation), indicating that animals with higher levels of soluble tau were more likely to have more insoluble tau. This tendency was also apparent when different age groups were analyzed separately (Figs 3b–d). (2.5–3.5 months old: R = 0.749, p = 0.002, n = 14; 6–6.5 months old: R = 0.810, p < 0.0001, n = 18; 8.5–10.5 months old: R = 0.704, p = 0.011, n = 12).

Figure 3.

Correlation between sarkosyl-insoluble tau and soluble tau. The y axis represents the sarkosyl-insoluble tau levels as a natural logarithm. Data were analyzed based on mice (a) of all ages (b) 2.5–3.5 months old, (c) 6–6.5 months old and (d) 8.5 months and older. The levels of significance were p < 0.001, p = 0.002, p < 0.0001 and p = 0.011 for (a) to (d) respectively.

Tau assembly

Previous studies suggested that neurofibrillary pathology in JNPL3 mice developes later than the earliest detection of sarkosyl-insoluble tau (Lewis et al. 2000). This raised the possibility that sarkosyl-insoluble tau detected in brains of young animals may not be filamentous. This issue was examined by studying P3 fractions with electron microscopy. Structures consisted of tau were identified by immunogold labeling with anti-tau polyclonal (E1) and monoclonal antibodies (Tau46) that recognize the amino- and carboxy-terminals of tau respectively. Tau aggregates with fibrous appearance were detected in P3 fractions from brains of 3-month-old female mice (Fig. 4a), but not in samples from non-transgenic mice (Fig. 4d). These structures had a diameter ranging between 15 and 25 nm (Figs 4b and c), resembling straight filaments observed in AD and FTDP-17. Most of the polymerized tau sturctures were short and no more than 400 nm in length. Only occasionally were longer tau polymers detected. Polymerized tau was also observed in samples from 9-month-old JNPL3 mice (Figs 4e and f); the polymers were often longer and more uniform in diameter than those detected in younger JNPL3 mice. The results demonstrate that the formation of sarkosyl-insoluble tau is accompanied by the assembly of fibrous elements as well as some non-fibrous structures (Fig. 4a, marked by arrowhead) with E1 and Tau46 immunoreactivities. Whether the non-fibrous structures represent precursors (nucleation sites) for tau filaments requires further investigation.

Figure 4.

Immunogold labeling of sarkosyl-insoluble fractions. Samples were prepared from subcortical regions of (a–c) 3-month-old female JNPL3 mouse, (d) 9-month-old female non-transgenic mouse and (e and f) 9-month-old female JNPL3 mouse. (b) High-magnification image of the aggregated tau in the rectangular box shown in (a). (c and f) High-magnification images of labeled structures. All samples were double-labeled with two antibodies, E1 (polyclonal) and Tau46 (monoclonal), followed by incubation with 5- and 10-nm colloidal gold-conjugated secondary antibodies respectively. Besides elongated structures, tau immunoreactivity was also detected on elements that may represent tau at early stages of aggregation or fragmented tau filaments (arrowhead). Scale bar, 100 nm for (a), (c) and (f); 200 nm for (e).

Tau phosphorylation

To investigate the role of phosphorylation in tau aggregation, samples were prepared from subcortical regions of 3-, 6- and 9-month-old female JNPL3 mice, and probed with a panel of antibodies specific for different phosphorylated tau epitopes. Owing to the low levels of tau aggregation in 3-month-old mice, the loading of samples from these mice was increased to twice that of 6- or 9-month-old mice. The dilution of each antibody was adjusted to produce similar levels of immunoreactivity in the P3 fractions. TG3 and PG5 did not label the P3 fraction intensely even at low dilutions. Phosphorylated tau immunoreactivities of different intensities were detected in proteins of 55–60 kDa in both S1 and S2 fractions with different antibodies. Some of the immunoreactivities detected in the 55-kDa region by monoclonal antibodies are likely to be non-specific, as this region was labeled (Fig. 5 marked with arrow) in control blots incubated only with secondary antibody (data not shown). The staining detected with polyclonal antibodies to phosphorylated tau (pT212, pT231, pS262, pS422), however, was specific. Among these four polyclonal antibodies, pS422 had the weakest labeling of 55–60-kDa tau in S1 and S2 fractions, suggesting that the phosphorylation of Ser422 occurs later or slower than that of other sites. Regardless of whether monoclonal or polyclonal anti-phospho-tau antibodies were used, very few phosphorylated tau species were found in the 50–55-kDa range in 6- and 9-month-old mice. In comparison, tau of these molecular weights was more intensely labeled in 3-month-old mice with some anti-phospho-tau antibodies. These tau species were E1 positive and considered to represent intact tau as they had the same or higher molecular weight than the four-repeat recombinant tau standard (Fig. 5).

Figure 5.

Tau phosphorylation in 3-, 6- and 9-month (M)-old female JNPL3 mice. Samples from subcortical brain regions were immunoblotted with 10 different antibodies, including E1, AT270, CP13, pT212, pT231, TG3, pS262, PHF1, PG5 and pS422. The amino acid residues containing different phospho-epitopes are shown, and are numbered according to the tau isoform with 441 amino acids. S1, S2 and P3 fractions were loaded at a ratio of 1 : 5 : 11.25 (based on wet-weight). Such adjustments allow more accurate assessment of the relative amount of 64-kDa protein in S2 versus P3 fractions. The loading of samples (absolute) from 3-month-old mice was twice that of 6- or 9-month-old mice. Western blots of CP13, TG3 and PG5 showed mouse IgG band (approximately 55 kDa; arrow) in S1 and S2 fractions. The location of 64-kDa tau (arrowhead) and recombinant tau (rT), containing four- and three-repeat tau isoforms without the N-terminal inserts, (4R0N and 3R0N) is indicated.

The 64-kDa protein from both S2 and P3 fractions displayed phosphorylated tau immunoreactivity (Fig. 5), but the extent of phosphorylation did not appear to be the same. This was verified by quantitation of the relative amount of phospho-tau in S2 and P3 fractions from 6- and 9-month-old mice, with densitometric scanning of western blots and correction for differences in sample volumes (S2 : P3 = 1 : 2.25). Based on the data derived from E1 immunoblotting of samples from the subcortical brain regions of three 6-month-old female mice (Fig. 6), the S2 and P3 fractions contained comparable amounts of 64-kDa tau (S2-64 kDa/P3-64 kDa ≈1). The ratio of PHF1- and AT270-immunoreactive 64-kDa tau in S2 versus P3 fractions was similar to that determined from E1 immunoblots; however, the amount of 64-kDa tau reactive with other phospho-tau antibodies was less in S2 than P3. The ratio of 64-kDa tau in two fractions ranged from 0.7 to 0.2 with a rank order as follows: pT231 (Thr231) > pS262 (Ser262) > pT212 (Thr212) > CP13 (Ser202/Thr205) > pS422 (Ser422). Similar results were obtained from analysis of 9-month-old mice (data not shown). Quantitative analysis of 64-kDa tau in S2 versus P3 was not performed with 3-month-old mice, because of poor resolution and weak labeling of the 64-kDa band in the S2 fraction. TG3 (Thr231/Ser235) and PG5 (Ser409) were not rank ordered because these antibodies had weaker immunoreactivity than other antibodies, gave little or no labeling of 64-kDa protein in the S2 fraction, and were not suitable for accurate quantitative analysis.

Figure 6.

Ratio of 64-kDa tau in S2 versus P3 fractions from 6-month-old female mice. The ratio of phosphorylated 64-kDa tau in S2 versus P3 was normalized with the ratio derived from E1 immunoblotting (E1-positive S2-64 kDa/P3-64 kDa as 1). The data represent the mean ± SEM of three animals.

Discussion

Numerous animal models have been generated for investigation of the pathogenesis of neurodegenerative disorders such as AD and FTDP-17, and to aid in the development of agents capable of preventing or treating the disease. Our previous studies established that expression of P301L mutant tau in transgenic mice leads to development of neurofibrillary tangles, neuronal loss and accumulation of phoshorylated tau that is resistant to solubilization by sarkosyl (Lewis et al. 2000). In the present study we have further demonstrated the presence of a pool of tau with intermediate solubility in these tau mice. This pool, although not extractable in TBS, is extractable by sarkosyl and is at least 10-fold larger than the pool of sarkosyl-insoluble tau. It is likely that this intermediate pool represents tau in the process of being converted to insoluble polymers, as it contains two groups of tau: one resembles normal tau (buffer extractable) and has a molecular weight of 50–60 kDa; and the other resembles sarkosyl-insoluble tau in which the major species is 64 kDa in size. Consistent with this view, our data showed that animals with lower levels of insoluble tau tend to have less 64-kDa in the S2 fraction (Fig. 1a).

The conversion of tau from a low to high molecular weight form appeared to depend on several factors, such as the level of soluble tau as well as the age and sex of the animal. On average, 3-month-old animals developed less insoluble tau than older animals, and mice containing higher levels of buffer-soluble tau tended to have more sarkosyl-insoluble tau. In this regard, it has been indicated in previous in vitro studies that the assembly of tau filaments requires the presence of critical concentrations of tau, and involves the formation of dimers, changing the conformation of tau from random coil to beta sheet (nucleation) with subsequent elongation (Von Bergen et al. 2001). It has also been demonstrated that the assembly of tau filaments in vitro requires the presence of molecules such as glycosaminoglycans, RNA or polyunsaturated fatty acids (Goedert et al. 1996; Kampers et al. 1996; Perez et al. 1996; Wilson and Binder 1997). Whether such molecules or others with similar abilities are needed to facilitate tau filament assembly in vivo is still unknown. Unlike neurofibrillary tangles in human neurological disorders with tauopathies, preliminary data from immunocytochemical studies of P301L mice did not indicate the presence of heparan sulfate proteoglycan in either tangles or pre-tangles (Yen et al.). Further studies are necessary to determine if other factors are critical to formation of neurofibrillary tangles besides age and tau expression levels.

The shift of tau from 50–60 kDa to 64 kDa in the S2 fraction results from increases in tau phosphorylation at multiple sites. This shift is also observed in human tauopathy in which the 64-kDa protein has also been shown to contain the tau isoform expressed in JNPL3 mice. These data demonstrate that tau hyperphosphorylation occurred in both human disease and in the mice. Generation of conformational and phosphorylated epitopes recognized by TG3 and PG5 respectively has been considered important for distinguishing abnormal tau in AD from normal tau (Vincent et al. 1998; Jicha et al. 1999). These epitopes have been shown previously to be phosphorylated and located at Thr231/Ser235 (TG3) and Ser409 (PG5) respectively. In present study, TG3 and PG5 immunoreactivities were detected in the 64-kDa tau species in the P3 fraction from the JNPL3 mice; however, signals on western blots for these antibodies were weak. One technical problem is that TG3 and PG5 are much less immunoreactive than other antibodies to phosphorylated epitopes in immunoblots. We have attempted to use commercial polyclonal antibodies known to recognize phosphorylated Ser409 and found they are no better than PG5 (monoclonal). Besides the apparent relatively low titer of TG3, it is possible that the conformational epitope recognized by this antibody is not fully preserved during sample extraction or preparation for gel electrophoresis.

The conversion of tau from a sarkosyl-soluble to -insoluble form is associated with increases in the extent of phosphorylation. Based on the ratio of 64-kDa tau phosphorylation in S2 and P3 fractions from 6-month-old mice, Ser396/Ser404 are already highly phosphorylated when the solubility of tau shifts from buffer-extractable to a sarkosyl-soluble pool. At this stage, most Thr181 and Thr231 residues have also already been phosphorylated. Other sites, however, do not reach similar levels of phosphorylation until tau becomes sarkosyl insoluble. A function of phosphorylation of these sites may be to stabilize tau filaments. This is supported by the observation that filaments formed by assembly of recombinant tau are highly soluble, and phosphorylation has been shown to increase the stability of neurofilaments, which are neuronal cytoskeleton insoluble in Triton-X100 (Pant 1995). Phosphorylation may also enable tau polymerization to progress from dimerization/nucleation to assembly phases. The issue of whether tau in the S2 fraction is derived from filaments and/or seeds (dimers, nuclei) could not be resolved by electron microscipic examination of the fraction, as it contains proteins solubilized in sarkosyl. We are exploring the use of different strategies to address this issue. With respect to the type of tau aggregate in JNPL3 mice, our data demonstrate that the majority of tau in the sarkosyl-insoluble fraction is transgenic human tau, even though the level of transgenic tau is roughly the same as that of endogenous tau (Lewis et al. 2000). The results are consistent with those obtained from studies of human brains with the P301L mutation in which mutant tau preferentially accumulates in lesions (Rizzu et al. 2000; Miyasaka et al. 2001).

In previous studies (Lewis et al. 2000), we noticed that there were discrepancies between the age at which insoluble tau and neurofibrillary tangles become detectable. The present immunoelectron microscopic studies of insoluble tau fractions clearly demonstrate that 2–3-month-old JNPL3 mice produce tau polymers with morphology comparable to that of older animals. Thus, the inability to see neurofibrillary tangles probably reflects the small number of filaments in neurons, which were below the sensitivity of detection.

To determine if the observed variability in soluble tau levels reflects differences in the outbred genetic background we are currently investigating soluble and insoluble tau in JNPL3 mice on an in-bred strain (C57B6). The soluble tau level differences observed between males and females are likely to reflect differences in transgene expression, as previous studies of JNPL3 mice with in situ hybridization and northern blot analysis have shown that female mice express higher levels of tau mRNA than their male counterparts (Lewis et al. 2001). It remains to be determined whether there are hormonal influences on tau transgenic expression.

In summary, our studies of mutant tau transgenic mice show that (i) age- and sex-matched mice have variable levels of sarkosyl-insoluble tau; (ii) insoluble tau levels increase with the age of the mice; (iii) mice with higher levels of tau develop more insoluble tau; (iv) tau assembly occurs in mice as young as 2–3 months of age; (v) conversion of soluble cytosolic tau to detergent-insoluble tau is associated with the formation of phosphorylated tau with intermediate solubility; and (vi) tau phospho-epitopes display different relative degrees of phosphorylation in partially soluble (S2) and insoluble (P3) fractions.

Acknowledgements

We thank Dr Peter Davies for antibodies PHF1, CP13, TG3 and PG5; Dr Virginia Lee for Tau46 and T49. This work was supported by the National Institute on Aging (NIA) grants (to S.-H. Yen, M. H. Hutton and D. W. Dickson), the Alzheimer's Disease and Related Disorder Association (ADRDA) grant to S.-H. Yen, Mayo Clinic Alzheimer's Disease Research Center (ADRC) grant to E. McGowan, the Mayo Foundation, The Smith Scholar Program to N. Sahara and J. Lewis, and The John Douglas French Alzheimer's Foundation to J. Lewis.

Ancillary