Address correspondence and reprint requests to Peter Davies Department of Pathology Albert Einstein College of Medicine, 1300 Morris Park Avenue F526, Bronx NY 10461 USA. E-mail: firstname.lastname@example.org
Neurofibrillary tangles are composed of insoluble aggregates of the microtubule-associated protein tau. In Alzheimer's disease the accumulation of neurofibrillary tangles occurs in the absence of tau mutations. Here we present mice that develop pathology from non-mutant human tau, in the absence of other exogenous factors, including β-amyloid. The pathology in these mice is Alzheimer-like, with hyperphosphorylated tau accumulating as aggregated paired helical filaments. This pathologic tau accumulates in the cell bodies and dendrites of neurons in a spatiotemporally relevant distribution.
sodium dodecyl sulfate polyacrylamide gel electrophoresis
Alzheimer's disease (AD) is a progressive neurodegenerative disorder that is characterized by two major pathologies: the extracellular accumulation of β-amyloid in plaques, and the intracellular accumulation of the microtubule-associated protein tau in neurofibrillary tangles (NFTs). The relative importance of, and the relationship between, these pathologies remain unclear. The prevailing hypothesis has been that β-amyloid deposition is the initiator of AD pathogenesis, with cell death and tau pathology occupying secondary roles (Hardy and Higgins 1992; Taylor et al. 2002). Mutations in the tau gene have been discovered (Hutton et al. 1998; Spillantini et al. 1998; Poorkaj et al. 2001) that directly link abnormalities in tau to neurodegenerative disease, yet there are no known tau mutations in AD. Several lines of transgenic mice develop amyloid deposition (reviewed in Wong et al. 2001); however, these mice do not develop tau pathology, and do not exhibit significant evidence of neuronal death. NFT formation has been observed in mice expressing mutant human tau transgenes (reviewed in Gotz 2001), making them good models for inherited tauopathies, but less valuable to the study of pathogenesis in AD. There appear to be functional differences between the normal and mutant tau proteins (Hasegawa et al. 1998; Hong et al. 1998; Dayanandan et al. 1999), and it is possible that the mechanism of tau pathology and cell death differs between AD and the tau mutation cases.
The adult profile of tau isoforms differs between humans and mice. There are six isoforms of human tau that are generated by alternative splicing of a single tau gene (Goedert et al. 1989a, 1989b; Kosik 1989). Tau isoforms are described as either 3 repeat (3R) or 4 repeat (4R), depending on the inclusion or exclusion of an imperfect repeat region coded for by exon 10 (Hutton et al. 1998; Spillantini et al. 1998; Poorkaj et al. 2001). Adult mouse brain contains exclusively the 4R tau isoforms, whereas levels of 3R and 4R are approximately equal in normal adult human brain. Although mouse and human tau sequences are similar, there are 14 amino acid differences in the N-terminal region. To create a model system with a human profile of tau isoforms, we have generated mice that express exclusively the human tau isoforms, herein referred to as htau mice. We show that htau mice undergo an age-related accumulation of AD-relevant phosphorylations on tau in the cell bodies and dendrites of neurons and accumulate aggregated paired helical filaments.
Materials and methods
Generation and analysis of htau mice
Htau mice were generated by crossing mice that express a tau transgene derived from a human PAC, H1 haplotype, known as 8c mice (Duff et al. 2000), with tau knockout (KO) mice that have a targeted disruption of exon one of tau (Tucker et al. 2001). The F1 generation of mice that contained the human tau gene was backcrossed to the KO mice to obtain a population of mice that were homozygous for the mouse tau disruption, but that also carried the human tau transgene. Mice were genotyped by PCR using primers for the human tau transgene (forward 5′-ACTTTGAACCAGGATGGCTGAGCCC-3′, reverse 5′-CTGTGCATGG CTGTCCCTACCTT-3′), the mouse tau gene (forward 5′-CTCAGCATC CCACCTGTAAC-3′, reverse 5′-CCAGTTGTGTATGTCCACCC-3′), and the disrupted tau gene (forward 5′-CAGGCTTTGAACCAGTATGG-3′, reverse 5′- TGAACTTGTGGC CGTTTACG-3′). Mice were maintained on a Swiss Webster/129/SvJae/C57BL/6 background.
Immunohistochemistry was performed on htau mouse brain sagittal vibratome sections collected from multiple mice, of each sex, aged 6 weeks, 3, 6, 7, 9, 10, 11, and 13 months. Immunohistochemistry was also performed on serial coronal sections (every sixth section) of htau mice aged 1, 10, and 11 months to determine the distribution of the cell body staining. Mouse brains were immunostained using standard streptavidin biotin peroxidase methods as described (Duff et al. 2000) and developed with 3,3′-diaminobenzidine (DAB). The following monoclonal antibodies were used to stain free-floating brain sections: CP13 (1 : 25; Ps202; Duff et al. 2000; Lewis et al. 2000), MC1 (1 : 10; Jicha et al. 1997a, 1997b, 1999a; detecting a conformational abnormality of tau, Weaver et al. 2000), and PHF1 (1 : 25; Ps396/404; Greenberg et al. 1992; Otvos et al. 1994). Sections were also immunostained with antibodies to neurofilament (NP18; 1 : 500; Duff et al. 2000). Accumulation of phosphorylated neurofilament was not detected in cell bodies in any region of the brain (data not shown).
Immunoblotting was performed as described (Andorfer and Davies 2000). Briefly, brains were homogenized in Tris-buffered saline (TBS), pH 7.4 containing 1 mm phenylmethylsulphonyl fluoride (PMSF), 1 mm sodium vanadate, 2 mm EGTA, and 10 mm sodium fluoride. Half the homogenate was used for preparation of heat-stable protein fractions by addition of NaCl to 2% and β-mercaptoethanol to 5% and heating at 100°C for 10 min. Lysate from individual mouse brains was separated using 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS–PAGE) and transferred to nitrocellulose membranes. Membranes were probed with CP27 (1 : 25) specific for human tau isoforms (Duff et al. 2000), CP13 (1 : 25) specific for Ps202, CP9 (1 : 25) specific for Pt231 (Jicha et al. 1999b), MC6 (1 : 25) specific for Ps235 (Weaver et al. 2000), and PHF1 (1 : 25) specific for dual Ps396/404.
Sarkosyl insolubility assay
This assay for insoluble tau was performed as described (Greenberg and Davies 1990). Ultracentrifugation and sarkosyl extraction (30 min incubation in 1% sarkosyl) was used to obtain soluble and insoluble (aggregated) fractions of tau from wild-type and 8c mice aged 9 months and htau mice aged 3 and 9 months. Insoluble fractions (pellets) were washed one time with 1% sarkosyl. Soluble and insoluble fractions were immunoblotted as above with Tau-46 (Chemicon, Temecula, CA, USA) and CP13 (Ps202), ET2 (4R tau, 1 : 500; Togo et al. 2002) and RD3 (3R tau, 1 : 1000; Togo et al. 2002; de Silva et al. 2003).
Electron microscopy (EM) on tissue sections was performed as described (Weaver et al. 2000) on mice, aged 9, 10, and 11 months, five mice per age group. Immunolabeling was performed with CP13 (1 : 20) (Ps202) as above and developed with DAB. Some sections were examined without immunostaining.
For EM of tau filaments the sarkosyl-insoluble pellets were suspended in 100 µL of 1 × TBS. A 6-µL aliquot of the suspension of the insoluble material was placed on each carbon coated formvar grid (Electron Microscopy Sciences, Fort Washington, PA, USA) for 10 min, then washed in distilled water (3 × 5 min each), stained for 15 min in uranyl acetate (5% in 50% ethanol), washed again in distilled water, and allowed to dry. All solutions were filtered before using. The uranyl acetate step was performed in the dark.
RT–PCR analysis of tau splice isoforms
RNA was isolated from brains of 6-month-old htau, 8c, and wild-type mice using Trireagent (Molecular Research Center, Cincinnati, OH, USA). RT–PCR was performed using specific primers designed to span either the N-terminal region (across exons 1 and 5), or to span the microtubule binding domain region (across exons 9 and 11), as performed previously on the 8c mice (Duff et al. 2000), with a cycle number of 25. Primers specific to human tau that spanned alternatively spliced exon 10, would amplify either 3R or 4R tau splice isoforms (forward primer: 5′-CTCCAAAATCAGGGGATCGC-3′, reverse primer 5′-CCTTGCTCAGGTCAACTG GT-3′.) Isoforms were identified based on their expected sizes. Products containing exon 10 (4R) were 390 bp, while those without exon 10 (3R) were 253 bp. Mouse-specific primers (forward primer 5′-TCCGCTGTCCTCTTCTGTC-3′, reverse primer 5′-TTCTCGTCATTTCCTGTCC-3′) and human specific primers (forward 5′-TGAACCAGGATGGCTGAGC-3′, reverse 5′-TTGTCATCGCTTCCAGTCC-3′) were used to examine splicing of exons 2 and 3.
Htau mice express exclusively the human isoforms of tau
Htau mice were generated by selective mating of two previously generated mouse lines, 8c and tau KO mice. The 8c mice express a tau transgene, derived from a human PAC driven by the tau promoter, and produce all human tau isoforms. 8c mice do not develop any evidence of tau pathology in the CNS (Duff et al. 2000). The tau KO mice were generated by a targeted disruption, in which cDNA for the enhanced green fluorescent protein (EGFP) was inserted into exon one of tau. The KO mice also do not develop any evidence of pathology. The resulting htau mice express six isoforms of human tau, but do not express mouse tau.
Tau hyperphosphorylation and accumulation in neuronal cell bodies and dendrites
Tau, in AD and several other neurodegenerative diseases, is hyperphosphorylated and redistributed from its normal position in the axon into the neuronal cell body and associated dendrites. The monoclonal antibody CP13, specific to phosphorylated serine 202 (Ps202) on tau, is commonly used to detect tau pathology in both early (pre-tangle) and more advanced stages of NFT accumulation. To characterize the distribution of phosphorylated tau in the htau mice, immunohistochemistry was performed with CP13 on brains collected from multiple male and female htau mice (between 6 weeks and 13 months of age). There was an age-dependent change in the localization of CP13 positive tau in the htau mice. In 6-week-old htau mice CP13 reactivity was high in the axons, but negligible in the cell bodies of the hippocampus (Figs 1a and d) and neocortex (Fig. 1g). The pattern of CP13 reactivity reversed as the htau mice aged. Cell body accumulation of CP-13 reactive tau occurred by 3 months of age (Figs 1b, e and h), and increased with further aging (Figs 1c, f and i). By 9 months of age, reactivity in the htau hippocampus (Fig. 2a) resembled that occurring in early stage NFTs of human brain (Fig. 2b).
A somatodendritic accumulation of tau in aging htau mice was also identified with antibodies used to detect pathologic tau in human brain, including MC1, recognizing an abnormal tau conformation and PHF1, a marker for later stage tangles (Uboga and Price 2000; Weaver et al. 2000) that is specific for phosphorylation at serines 396 and 404 (Ps396/404). These antibodies, in addition to CP13, revealed morphologically abnormal neurons in multiple regions of the older htau brains (9 months and older). MC1-positive cells, were detected in both the neocortex (Fig. 2c) and hippocampus (Fig. 2e) of htau mice aged 9 months. There was extensive reactivity of the neuropil with PHF1 by 3 months of age; however, this late-stage antibody stained few neuronal cell bodies in mice younger than 12 months old. PHF1-positive cells, in both the cortex (Fig. 2d) and hippocampus of 13-month-old htau mice appeared irregularly shaped, often with distorted processes. This pattern of staining was similar in each sex and was consistent between the multiple mice examined at each time point. No cell body staining was observed in age-matched wild-type, 8c or KO mice (data not shown).
The age-related accumulation of tau pathology in the htau mice was also observed by immunoblotting (Fig. 3). The phosphorylation-specific antibodies were used to examine heat stable preparations (enriched for tau) of htau brains from mice aged 2, 6, 9, 13, and 15 months. The blots indicated an age-related increase in phosphorylation at serine 202 (Fig. 3b), threonine 231 (Fig. 3c), and serine 235 (Fig. 3d). In agreement with PHF1 immunohistochemical staining, the immunoblots indicate that total levels of phosphorylation at serines 396/404 do not significantly increase with age in the htau mice (Fig. 3e).
The regional distribution of tau pathology in the htau mice shares similarity with that occurring in AD
Accumulation of phosphorylated tau in the cell bodies and dendrites of neurons occurred in several cell populations that are severely affected in AD, including the entorhinal cortex (Fig. 4b) and ventromedial hypothalamus (Fig. 4e). Cholinergic cell populations in the medial septum (Fig. 4c) and the nucleus of the horizontal limb of the diagonal band (Fig. 4f; homologous to the nucleus basalis of Meynert in humans) also accumulated CP13-positive tau in their cell bodies and dendrites. Although the cell body accumulation of phosphorylated tau was extensive in the hippocampus and in the neocortex of htau mice, there was no significant accumulation in the striatum (Fig. 4a) or the cerebellum (Fig. 4d), areas not usually associated with tau pathology in AD.
Filamentous insoluble tau accumulates in htau mice
NFTs are composed of insoluble twisted filaments, known as paired helical filaments (PHFs) that are composed of aggregated tau. Immunoelectron microscopy performed on htau mouse brain sections revealed an accumulation of CP13-positive aggregates in the somatodendritic compartments of cortical and pyramidal hippocampal cells (Fig. 5a). Aggregates in early AD cases (Fig. 5b) appeared similar when stained with CP13. In htau neurons, the majority of CP13-positive tau was located in the proximal dendrites, suggesting that this was the initial site of tau aggregation. The aggregates had an average width of 15 nm, and were not densely packed (Figs 5d and e). Similar aggregates were observed by electron microscopy performed on unstained sections (Fig. 5c).
PHF-tau can be isolated from human AD brain via the sarkosyl-insolubility assay (Greenberg and Davies 1990), in which tau can be fractionated into two populations, one that contains normal tau (soluble in sarkosyl) and one that contains aggregated tau (insoluble in sarkosyl). We used this method on brains from wild-type, 8c, and htau mice, and found that aggregated tau accumulated in htau mice, but not in age-matched wild-type or 8c mice (Fig. 6). Insoluble tau was present in both 2- and 9-month-old htau mice. This insoluble material was phosphorylated at serine 202 (Fig. 6f), and was reactive with an antibody that is specific for the 3R isoforms of tau (Togo et al. 2002; de Silva et al. 2003; Fig. 6g). Although a 4R-specific antibody (Togo et al. 2002) reacts with soluble tau (Fig. 6c), it does not react with the insoluble tau (Fig. 6h), suggesting that the insoluble tau is composed solely of 3R isoforms.
Electron microscopy (EM) was performed on the sarkosyl-insoluble material extracted from htau mice at several time points (9, 12, and 14 months). PHFs were isolated (Figs 7a–e) from mice at all of these ages. The htau filaments have an average period length of 44.7 nm (n = 13), that is comparable to that of PHF isolated from human AD brain (Table 1).
Table 1. Comparison of measurements (n = 13) of paired helical filaments isolated from htau mice and human AD brain
All measurements are a mean ±SD in nanometers (nm).
43.4 ± 3.4
44.7 ± 3.5
12.8 ± 0.8
12.4 ± 0.7
6.6 ± 1.1
7.0 ± 1.1
Tau isoforms levels
A number of neurodegenerative diseases that develop tau pathology appear to result from an altered profile of tau isoforms (Buee and Delacourte 1999; Buee et al. 2000; Ishizawa et al. 2000; Miyasaka et al. 2001). Altered splicing of the human tau transgene toward the production of more 3R tau has been documented in the 8c mice (Duff et al. 2000). To examine whether the tau pathology in the htau mice might be caused by an altered ratio of tau isoforms, we examined the levels of 3R and 4R expression on western blots loaded with equal levels of total protein (Fig. 8). Immunoblot analysis with antibodies specific to human tau (Fig. 8a), 4R tau (Fig. 8b), and 3R tau (Fig. 8c) was performed on brain homogenates collected from wild-type mice, tau KO mice, 8c mice, htau mice, and normal human brain. Both 8c and htau mice expressed high levels of human tau (Fig. 8a), although expression was highest in the 8c mice. Wild-type mice produced high levels of 4R tau protein (Fig. 8b), and as expected did not produce any 3R isoforms (Fig. 8c), or human tau (Fig. 8a). KO mice do not react with any of the antibodies used (Figs 8a–c). Although 3R isoforms were not expressed in normal adult mice, they were highly expressed in the transgenic mice, both 8c and htau (Fig. 8c). In the 8c mice levels of 3R and 4R tau isoform were higher than in the htau mice and the majority of the 4R isoforms appeared to be contributed by the mouse gene.
We also performed RT–PCR analysis on RNA extracted from the htau mice, 8c mice, and wild-type mice. Primers specific for mouse tau sequence amplified products in the wild-type and 8c mice, but not in the htau mice (Fig. 9a), confirming that the htau mice do not produce mouse tau. Primers specific for the human tau sequences were used to confirm that multiple isoforms of human tau are produced (Figs 9b and c) in the htau mice. In both the 8c and the htau mice, the level of 3R isoform produced was significantly higher than that of the 4R isoform (Fig. 9c), supporting the findings at the protein level described above. Amplified products containing exon 10 (4R) gave a 390-bp band, while those without exon 10 (3R) gave a 253-bp band. We also looked for the presence of isoforms including the N-terminal exons 2 and 3, using primer sets specific for the human isoforms (Fig. 9b). Isoforms containing exons 2 and 3 resulted in a 428-bp band, those containing only exon 2 gave a 341-bp band, and those lacking both exons gave a 253-bp band.
Htau mice develop tau pathology in a time course and distribution that is comparable to that occurring in the early stages of human AD. Hyperphosphorylated, conformationally altered tau accumulates in the cell bodies and dendrites of neurons in the htau mice as they age. The first evidence of tau redistribution from the axons into the cell bodies occurs by 3 months of age. Accumulation of hyperphosphorylated tau begins by 6 months, but increases further by 13 and 15 months of age. Aggregated tau and PHFs were detectable in htau mice aged 9 months via the sarkosyl-insolubility assay and immunoelectron microscopy. High levels of tau are expressed in the htau mice; however, the pathology is not likely to occur solely because of overexpression of tau. The 8c mice are in fact expressing higher levels of human tau and are also expressing high levels of mouse tau, for a combined load of tau that is in excess of that occurring in the htau mice. Yet, the 8c mice do not develop pathology, whereas the htau mice do. An altered ratio of tau isoforms, combined with a loss of mouse tau, may underlie the htau pathology. The main difference between these transgenic mice is the presence of 4R mouse tau in the 8c mice. The RT–PCR results indicate the presence of more 3R isoform mRNA from the human transgene in both 8c and htau mice, that is confirmed by the immunoblots. However, the total 4R tau isoform level in the 8c is high, and thus it is only the htau mice that have an actual imbalance of 3R isoform over 4R isoform. The accumulation of 3R, but not 4R tau in the sarkosyl-insoluble fraction strengthens this hypothesis. Significant functional differences between 3R and 4R tau have been reported, particularly with. respect to microtubule dynamics (Goode et al. 2000; Utton et al. 2001), and a shift in the ratio of normal tau isoforms has been documented in several tauopathies. Although it is not clear why simply shifting the ratio of normal tau isoforms should be as detrimental to neuronal cells as the presence of mutant tau, the htau mice provide an intriguing model system in which to explore this issue further. It is conceivable that 4R mouse tau protects the 8c mice from developing the neurofibrillary pathology that results in the htau mice from an excess of human 3R tau.
Mice expressing mutant tau (Lewis et al. 2000) have provided the most complete mouse models of neurofibrillary tau pathology available thus far. However, the pathology in these mice is predominantly in the hindbrain and spinal cord, areas of less relevance to AD. In contrast, the majority of tau pathology in htau mice is located in the neocortex and hippocampus, is minimal in the brain stem and spinal cord, and there is no evidence of gross motor or behavioral disturbances. Thus the htau mice more closely recapitulate an AD-type spatial distribution of early tau pathology. The distribution of tau pathology in the htau mice also includes selected subcortical brain regions, such as the medial septum and the nucleus of the diagonal band, that supply cholinergic innervation to the hippocampus and neocortex, and are susceptible to extensive atrophy and cell death in AD brains (Whitehouse et al. 1982; Tagliavani and Pilleri 1983; Rogers et al. 1985; Saper et al. 1985; Rasool et al. 1986).
Two independent groups (Gotz et al. 2001; Lewis et al. 2001) have demonstrated that elevated β-amyloid can influence the severity of pathology in mutant tau mice, suggesting a causal relationship between the two pathologies. Htau mice, however, develop pathology from wild-type human tau and no other exogenous factors, including β-amyloid. Some evidence of pathology from wild-type tau has been shown by other groups, including in a Drosophila model combining overexpression of wild-type human tau with manipulation of GSK3β activation (Jackson et al. 2002), and in a mouse model expressing cDNA for the shortest 3R isoform of tau (Ishihara et al. 1999). The htau mice join these models as continuing evidence that β-amyloid may not be an absolute requirement for tau pathology to form from wild-type tau. The htau mice will allow us to address some of the important questions related to the development of tau pathology in AD, including its relationship to β-amyloid deposition and provide us with a means to dissect the sequence of events involved in the build-up of filamentous tau and its relevance to learning and memory deficits.
Supported by grants from The Tepperman Family, Molecular Geriatrics Corp, NIMH 38623, and NS 07098. RdS is supported by the Reta Lila Weston Trust.