Rapid induction of intraneuronal neurofibrillary tangles in apolipoprotein E-deficient mice

Abstract

Cultured hippocampal slices prepared from apolipoprotein E-deficient mice were exposed to an inhibitor of cathepsins B and L and then processed for immunocytochemistry using antibodies against human paired helical filaments. Dense, AT8-immunopositive deposits were found in the subiculum, stratum oriens of hippocampal field CA1, and the hilus of the dentate gyrus. This distribution agrees with that described for tangles in Alzheimer's disease. The appearance of the labeled structures fell into categories that correspond to previously proposed stages in the progression of intraneuronal neurofibrillary tangles in human hippocampus. Electron microscopic analyses confirmed that microtubule disruption and twisted bundles of filaments were present in neurons in the affected areas. These results support the hypothesis that partial lysosomal dysfunction is a contributor to Alzheimer's disease and suggest a simple model for studying an important component of the disease.

Neurofibrillary tangles (NFTs) and amyloid plaques are pathological hallmarks of Alzheimer's disease (AD). Although much has been learned about the formation of plaques, the mechanisms responsible for tangles are still poorly understood. The discovery that NFTs are largely composed of hyperphosphorylated tau protein and fragments of tau (1–7) pointed to age-related changes in kinases and phosphatases as key events in their etiology (8–16). A very different type of mechanism was implicated when it was recognized that neurofibrillary tangles comparable to those found in AD occur in young adults with Niemann-Pick's type C disease (17–19), a single gene mutation resulting in partial lysosomal dysfunction. Other postmortem analyses showed that lysosomal disturbances develop in AD vulnerable neurons in advance of AD pathologies (20–23). Finally, studies using cultured slices from rat brain confirmed that experimentally induced lysosomal dysfunction generates hyperphosphorylated tau fragments (24, 25) and other concomitants of AD (26).

Missing from hypotheses connecting lysosomes to tangles is direct evidence that disturbances of the former are followed by robust formation of the latter. Moreover, there are no results suggesting that predisposing conditions for AD interact with lysosomes, as would be expected from these hypotheses. The only predisposing condition unequivocally associated with late-onset AD is the variation of the apolipoprotein E (apoE) gene. In humans, three alleles (ɛ2, ɛ3, and ɛ4) of the apoE gene encode variants that differ at two residues (27, 28). ApoE4 increases the risk for late-onset familial and sporadic AD dose-dependently, whereas apoE2 and apoE3 seem to function as protective factors (29–34). In vitro experiments showed that apoE3 but not apoE4 variants were able to form stable complexes with the microtubule-associated proteins tau and MAP2c (35). It has been hypothesized that apoE3, by binding to tau, protects tau from being hyperphosphorylated and thus prevents the generation of intracellular neurofibrillary tangles. Indeed, hyperphosphorylation of tau has been reported in apoE-deficient mice (36). Here, we report that experimentally induced lysosomal dysfunction causes rapid and extensive formation of intraneuronal tangles in brain slices prepared from apoE-deficient mice. In addition to confirming an important prediction of the lysosomal hypothesis, the findings provide a simple model for studying an important feature of AD.

Materials and Methods

Preparation of Mouse Hippocampal Slice Cultures.

Ten- to 13-day-old C57BL/6J (wild-type) and C57BL/6J-apoE^tmlUnc (apoE^−/−) mice obtained from the Jackson Laboratory were used to prepare hippocampal slice cultures by using the methods described by Stoppini et al. (37). Each culture cluster plate contained hippocampal slices from either two wild-type or two apoE^−/− mice, and individual wells were used for matched control and experimental treatment groups. After maintaining the slices with normal culture medium (38) in vitro for 12–14 days, slices were incubated with culture medium containing either 20 μM N-CBZ-l-phenylalanyl-l-alanine-diazo-methyl-ketone (ZPAD, Bachem), a selective inhibitor of cathepsins B and L (39), or vehicle (DMSO, 0.04%) for 6 days. The treatment medium was exchanged every other day.

Immunohistochemical Procedures.

For immunocytochemical staining, control and ZPAD-treated slices from both wild-type and apoE^−/− mice were fixed in 4% paraformaldehyde, cryoprotected in 20% sucrose, and sectioned on a freezing microtome at 25 μm parallel to the broad upper surface of the explant. Free-floating sections were blocked with 10% normal horse serum, incubated with monoclonal anti-paired helical filament, AT8 (1:1,000; Innogenetics, Zwijndrecht, Belgium), which recognizes the full-length human tau protein (and tau fragments) phosphorylated at residues Ser-202 and Thr-205, in 5% normal horse serum at 4°C overnight. The sections then were incubated in biotinylated anti-mouse IgG (1:200, Vector Laboratories), followed by the avidin/biotin system (1:100, Vector Laboratories), and the staining was finally revealed by using diaminobenzidine as chromogen. As a control, the primary antiserum was omitted from the initial incubation. The results described here involved detailed analysis of multiple sections from 30 apoE^−/− and 30 wild-type slices treated with ZPAD. A smaller number of vehicle-alone slices in each group also were examined. Double-labeling to confirm the identification of cells as neurons was carried out in four apoE^−/− slices by using an antibody against neuron-specific nuclear protein (NeuN, Chemicon).

Electron Microscopic Analysis.

For electron microscopy, control and ZPAD-treated slices were fixed with 1.5% paraformaldehyde and 1.5% glutaraldehyde at 4°C for 2 h. Slices were postfixed in 2% osmium tetroxide for 1 h, dehydrated in a series of graded alcohols, and embedded in Polybed-812. Semithin (1 μm) sections were cut from these blocks and stained with 0.1% toluidine blue. Sections containing the region of the subiculum and hippocampal field CA1 were used for ultrastructural analysis. After the peripheral regions of the specimen were trimmed away, thin sections were cut with a diamond knife, mounted on Formvar-coated slot grids, stained with uranyl acetate and lead citrate, and examined with a CM10 transmission electron microscope (Philips Electronic Instruments, Mahwah, NJ).

Immunoblotting.

Control and ZPAD-treated slices were collected in ice-cold 10 mM Tris⋅HCl harvest buffer containing 0.32 M sucrose/2 mM EDTA/2 mM EGTA/0.1 mM leupeptin, pH 7.4, and centrifuged and sonicated after being resuspended in lysis buffer (8 mM Hepes/1 mM EDTA/0.3 mM EGTA, pH 8.0). Proteins (100–120 μg) from each sample were denatured by boiling for 5 min with 2.5% (wt/vol) SDS and 3% 2-mercaptoethanol and then subjected to SDS/PAGE on 10% linear gradient gels (40). Resolved proteins were then transferred to nitrocellulose membranes as described by Towbin et al. (41), incubated in 3% gelatin in Tris-buffered saline for 1 h at room temperature followed by incubation with 1% gelatin in Tris-buffered saline with 0.5% Tween 20 containing an antibody that recognizes tau-1 (1:100; Roche Molecular Biochemicals) at room temperature overnight. Antibodies were visualized by using the 5-bromo-4-chloro-3-indolyl-phosphate and nitro blue tetrazolium substrate system. Relative optical densities and areas of immunopositive bands were quantified by using the National Institutes of Health image analysis system.

Results

Slices from apoE^−/− mice (Fig. 1 A and B) were comparable in size and appearance to their wild-type (Fig. 1 C and D) counterparts. Some, but not all, untreated apoE^−/− slices had AT8-immunolabeled cells in the outgrowth zone that developed in the first week after explantation. However, immunopositive neurons were only rarely found within the dendritic and cell body layers of hippocampus itself (Fig. 1A).

Figure 1.

Induction of tangle-like structures in cultured hippocampal slices prepared from apoE^−/− mice. Slices from apoE^−/− mice were incubated with vehicle (A) or ZPAD, an inhibitor of cathepsins B and L (B), for 6 days and then processed for immunocytochemistry using a mAb (AT8) that recognizes hyperphosphorylated tau proteins and neurofibrillary tangles in human tissue. Immunopositive elements are found in the outgrowth regions of the control slice from an apoE^−/− mouse but not within the hippocampus itself. In contrast, the ZPAD-treated slice has numerous, densely labeled cells in stratum oriens of hippocampal field CA1 and in the subiculum. Note that the densely packed neurons in stratum pyramidale of field CA3 and in stratum granulosum of the dentate gyrus are not stained. (Lower) AT8 immunoreactivity in slices from wild-type mice treated with vehicle (C) or ZPAD (D) for 6 days. Note the lack of AT8-immunopositive cells in vehicle-treated slices and the small number of AT8-immunopositive cells in ZPAD-treated slices. (Bar = 200 μm.)

Numerous, densely stained neurons were present within the hippocampus and retrohippocampal cortex in nearly all of the 30 apoE^−/− slices treated with ZPAD for 6 days. These were numerous in the subiculum and hippocampal field CA1 and relatively uncommon in the dentate gyrus (except for the hilus where AT8-immunoreactive neurons were often encountered) and field CA3. Within CA1, the profiles were much more frequent in stratum oriens than in strata pyramidale or radiatum (Fig. 1B). This can be seen clearly in the higher-power micrograph presented in Fig. 2. Fig. 2 also illustrates the extent to which most neurons and their processes were unstained by the paired helical filament antibody, even in apoE^−/− slices treated with ZPAD. Note, for example, that the densely packed cell bodies in stratum pyramidale, as well as the profusion of apical dendritic branches in stratum radiatum, are barely detectable in the micrograph. The pattern seen in Figs. 1 and 2 held for most slices; a common variation involved the presence of significant numbers of labeled cells in strata pyramidale and radiatum of field CA1 and in the hilus of the dentate gyrus (data not shown).

Figure 2.

Types and distribution of phosphorylated tau-immunoreactive neurons in CA1 region after 6 days of ZPAD treatment. Shown is a vertical section that extends across most of the basal (stratum oriens), and the inner third of the apical (stratum radiatum), dendritic fields in field CA1 of a cultured slice that had been exposed to ZPAD for 6 days. The majority of the AT8-immunopositive cells are found in the basal dendritic field. The cell bodies (stratum pyramidale) and apical dendrites of the pyramidal cells, by far the most numerous population of neurons in the section, are with few exceptions unlabeled. The stained elements are not homogeneous. The cells marked as 1 appear to be intact neurons with immunopositive processes and dense deposits accumulating within the cell body. The labeled neuron marked as 2 has swollen and distorted dendrites. The elements marked as 3 appear to be remnants of neurons. (Bar = 50 μm.)

Effects of the type described above were not seen after much longer treatments in prior studies using cultured slices from rat hippocampus (25). Wild-type mice were intermediate between rats and apoE^−/− mice. Slices with clear anatomical landmarks in the CA1/subiculum boundary region were selected for estimating the magnitude of the difference between the two mouse groups. Counting was done in a 0.3-mm² box centered over strata oriens and pyramidale at the CA1/subicular border. The number of AT8-immunopositive profiles in apoE^−/− mice was 194 ± 15 (mean ± SEM, n = 5) and 113 ± 13 (n = 5) in the wild types, a difference that was highly significant (P = 0.005, two-tail t test). These results confirm that the apoE mutation contributes to the formation of intraneuronal NFTs.

Although ZPAD treatment had robust and reliable effects across apoE^−/− slices, the immunopositive cells within a given slice were not homogeneous in appearance. Most of the labeled neurons were shrunken and had “polar caps”; i.e., dense deposits located eccentrically within the somata. Examples of these are marked as 1 in Fig. 2. In many cases, the cells had immunopositive processes extending away from the somata for considerable distances. Neurons with labeled, pathological dendrites (2 in Fig. 2) as well as caps of labeled material unconnected to cell bodies (3 in Fig. 2) were also commonplace throughout stratum oriens and subiculum. The isolated caps may be remnants of neurons. Double-labeling experiments (not shown) confirmed that the shrunken profiles were neurons.

The variety of densely stained elements found in the ZPAD-treated apoE^−/− slices is not unlike the diversity of intraneuronal NFTs found in hippocampus during early-stage AD. This point is illustrated in Fig. 3. The upper panels, from an apoE slice, are higher-power micrographs organized according to a progression along the lines proposed for NFT development in AD (42–44). The steps shown in Fig. 3 are as follows: essentially intact neurons with immunopositive cell bodies and dendrites (A); dense, localized deposits within the cell body accompanied by evident dendritic abnormalities such as clubbing (B and C, arrow); expansion of initial dendritic segments (D and E); loss of dendritic organization, sometimes accompanied by the growth of large filament-filled structures resembling growth cones (F, arrows); and disappearance of the neuron leaving a cap of labeled material (G and H).

Figure 3.

Morphology of neurons stained by an antibody that recognizes neurofibrillary tangles. (Upper) Immunopositive neurons in cultured slices prepared from apoE^−/− mice. The micrographs are ordered according to a proposed sequence of pathological steps. (A) Two neurons in the subiculum with immunopositive cell bodies and primary dendritic branches (white arrows). Note that other neurons in the field are unlabeled (black arrows). (B) Neuron with a dense deposit (cap) in one pole of its cell body. (C) Neuron with pathological swelling (arrow) of a distal dendrite. (D and E) Cells with pathological dendritic expansions proximal to the cell body. (F) Exploded process attached to a dendrite containing fibrous material. Note that the dense cap of immunopositive material covers most of the cell body. (G and H) Dense caps that do not appear to be associated with somata, i.e., remnants of neurons. (Bar = 12.5 μm in A, 10 μm in B, 8 μm in C, 15 μm in D and H, 11 μm in E and G, and 17 μm in F.) (Lower) Immunopositive neurons in the hippocampus from a brain of a patient classified as being in the early stages of AD. The micrographs are again arranged according to a proposed sequence of pathologies. (A) Apparently intact pyramidal neuron with a dense cap and a labeled apical dendrite. (B and C) Neurons with dendritic swellings. (D and E) Dendritic expansions proximal to the cell body. (F and G) Immunopositive caps that do not appear to be attached to intact neurons. (Bar = 50 μm in A, 45 μm in B and D, 30 μm in C, 18 μm in E, 20 μm in F, and 12.5 μm in G.)

The bottom panels of Fig. 3 are from hippocampal field CA1 of an early-stage AD brain. Immunostaining was carried out with the same procedures and antibody used for the sections from the apoE^−/− slices. A shows a typical neuron with a labeled dendrite and cell body cap (arrow). B–D illustrate the dendritic abnormalities that are commonplace at this stage of the disease. Note the apparent clubbing and fragmentation (arrows) at sites removed from the cell body. E shows a swelling proximal to the soma; examples of neuronal remnants are found in F and G.

Electron microscopic analyses of zones with labeled neurons confirmed that the pathological changes detected with paired helical filament antibodies were accompanied by the development of aberrant filaments. Neuronal cell bodies and proximal apical dendrites were often nearly filled with a dense plexus of filamentous material, as shown in the micrograph in Fig. 4A. Closer examination of the filaments from this (Fig. 4B) and other (e.g., Fig. 4C) neurons showed them to be long twisting bundles that frequently crossed each other at different orientations (arrows). Structures of this type were unique to ZPAD-treated slices.

Figure 4.

Electron micrographs of CA1 neurons from apoE^−/− slices incubated with ZPAD for 6 days. (A) Survey micrograph showing a primary dendrite emerging from a neuronal cell body. Filamentous material (arrows) occupies more than half of the cross-section of the dendrite. (B) Enlargement of bundled filaments in the proximal dendrite from A. (C) Another example of a dendrite with filaments that form bundles that crisscross each other (arrows). (Bar = 2 μm in A, 0.75 μm in B, and 0.4 μm in C.)

Several lines of evidence indicate that NFTs assemble from mixtures of tau and tau fragments (4, 45), and it is likely that tau proteolysis is an essential step in tangle formation. Accordingly, immunoblots were used to test whether accelerated breakdown of tau might account for the enhanced build-up of intraneuronal NFTs in the apoE^−/− slices. The antibody tau-1 detected a set of tau isoforms (50–55 kDa) in untreated hippocampal slice cultures from both apoE^−/− and wild-type mice (Fig. 5). Six days of ZPAD treatment caused a 22 ± 3% (mean ± SEM) reduction in levels of native tau in slices from wild-type animals and a 35 ± 1% reduction in apoE^−/− slices (n = 6 for each, P < 0.001, t test).

Figure 5.

Experimentally induced lysosomal dysfunction causes greater tau breakdown in hippocampal slices from apoE^−/− mice. Hippocampal slices cultured from wild-type (apoE^+/+) and apoE^−/− mice were incubated with vehicle (Con) or ZPAD for 6 days. Slices then were homogenized and processed for electrophoresis. Western blots were stained by using anti-nonphosphorylated tau-1 mAb. Note that ZPAD treatment induced a marked decrease in native tau (p50–55) (*, P < 0.001, two-tailed t test, in both apoE^+/+ and apoE^−/− slices) and an increase in the breakdown fragments at p29 kDa. Densitometric analysis (Lower) shows that ZPAD-induced breakdown of tau-1 is significantly greater (#, P < 0.001, t test) in apoE^−/− vs. apoE^+/+ hippocampal tissues.

Discussion

The present studies show that suppression of cathepsins B and L for only 6 days is sufficient to induce tangle-like structures and that apoE has a potent influence on tangle formation. Four types of neuronal staining were recognized by antibodies against human paired helical filaments: cell body and dendritic tree, a loose cap of staining occupying about half the cell body, a very dense cap of staining, and a dense cap without an intact cell body. These groups correspond to stages in tangle development described by Braak and Braak (42, 43) and Thal et al. (figure 6 in ref. 44) on the basis of postmortem studies with the same antibody used in the present experiments. Electron microscopic analyses showed that neurons in immunopositive areas had disrupted microtubules and were filled with twisted filamentous structures. Finally, the formation of the intraneuronal structures in mouse slices was most reliable and pronounced in the same areas in which it occurs in AD brains, i.e., the subiculum, stratum oriens of field CA1, and the hilus of the dentate gyrus. That this complex pattern is similar to that described in mapping studies of AD pathologies strongly suggests that the same mechanisms are responsible for the formation of intraneuronal NFTs in slices and in AD.

How lysosomal dysfunction triggers tangle formation and why this process is facilitated by apoE deficiency are critical subjects for future work. The limited dysfunction induced by inhibitors of cathepsins B and L causes the increase and apparent leakage of proteases into the cytoplasm and fragmentation of tau (24). Cathepsin D has been shown to cleave tau at neutral (cytoplasmic) pH (24, 46), and the fragmentation of tau induced by ZPAD treatment was significantly reduced by inhibitors of cathepsin D (47). Inhibition of cathepsins B and L caused increases in the levels of cathepsin D that were clearly greater in apoE^−/− mice than in wild types (A.P.Y., X.B., and G.L., unpublished observation). These effects could account for the present results because paired helical filaments self-assemble much more rapidly when appropriate tau fragments are present (45). With regard to apoE, the isoforms apoE2 and apoE3 form stable complexes with the microtubule-associated proteins tau and MAP2c, whereas apoE4 does not (35). The absence of the stabilizing influence of apoE2/3 could lead to greater dissociation of tau from microtubule complexes and hence greater exposure to kinases and proteases. Possibly related to this, levels of hyperphosphorylated tau are reported to be elevated in apoE^−/− mice (36) (but see ref. 48), and the present experiments show that breakdown of tau proteins is significantly greater than that found in slices from wild-type mice. Other mechanisms, such as changes in membrane lipid composition or enhanced lipid oxidation that could compromise lysosomal membranes, also might lead to severe lysosomal dysfunction that in turn could cause the accelerated formation of tangle-like structures in apoE^−/− mice. In short, our results demonstrate that the absence of apoE significantly enhances the susceptibility to lysosomal dysfunction and that the combination of these two factors, i.e., apoE deficiency and lysosomal dysfunction, is both necessary and sufficient for the induction of neurofibrillary tangles.

Abbreviations

NFT

neurofibrillary tangle

AD

Alzheimer's disease

apoE

apolipoprotein E

ZPAD

N-CBZ-l-phenylalanyl-l-alanine-diazo-methyl-ketone