Neurofibrillary Tangles and the Deposition of a Beta Amyloid Peptide with a Novel N-Terminal Epitope in the Brains of Wild Tsushima Leopard Cats

Beta amyloid (Aβ) deposits are seen in aged individuals in many of the mammalian species that possess the same Aβ amino acid sequence as humans. Conversely, neurofibrillary tangles (NFT), the other hallmark lesion of Alzheimer’s disease (AD), are extremely rare in these animals. We detected Aβ deposits in the brains of Tsushima leopard cats (Prionailurus bengalensis euptilurus) that live exclusively on Tsushima Island, Japan. Aβ42 was deposited in a granular pattern in the neuropil of the pyramidal cell layer, but did not form argyrophilic senile plaques. These Aβ deposits were not immunolabeled with antibodies to the N-terminal of human Aβ. Sequence analysis of the amyloid precursor protein revealed an amino acid substitution at the 7th residue of the Aβ peptide. In a comparison with other mammalian animals that do develop argyrophilic senile plaques, we concluded that the alternative Aβ amino acid sequence displayed by leopard cats is likely to be related to its distinctive deposition pattern. Interestingly, most of the animals with these Aβ deposits also developed NFTs. The distributions of hyperphosphorylated tau-positive cells and the two major isoforms of aggregated tau proteins were quite similar to those seen in Alzheimer’s disease. In addition, the unphosphorylated form of GSK-3β colocalized with hyperphosphorylated tau within the affected neurons. In conclusion, this animal species develops AD-type NFTs without argyrophilic senile plaques.


Introduction
Neurofibrillary tangles (NFT), one of the diagnostic lesions of Alzheimer's disease (AD), are rarely found in non-human animal brains. Although the etiology of AD is yet to be elucidated, the ''amyloid hypothesis'' is widely accepted to explain its pathogenesis [1]. According to this hypothesis, the age-dependent accumulation of beta amyloid (Ab) peptides in the brain induces a subsequent cascade that culminates in NFT formation. Argyrophilic aggregates of Ab peptide are called senile plaques, which are another diagnostic lesion of AD.
The AD-related alterations that occur in the brains of animals such as monkeys and dogs have been well studied [2], [3], [4], [5], [6]. However, although these animals frequently form senile plaques with aging, they rarely develop NFT [7], [8], [9]. Even in the few reported animal cases of NFT, no pathological examinations were performed to exclude other diseases that could have caused the NFT to develop [10], [11]. Therefore, it has been a major interest whether AD is a humanspecific disease [12], [13].
One of the authors (JKC) has previously reported the occurrence of NFT in the brains of captive cheetahs (Acinonyx jubatus) [11]. Subsequently, we have detected NFT and Ab deposits in the brains of wild Tsushima leopard cats (Prionailurus bengalensis euptilurus). According to the phylogenetic tree of living cat species (Felidae), these two species belong to two closely related lineages that diverged approximately 6.7 million years ago [14]. The NFT of the leopard cats were consistent with the pathological characteristics of human AD and were also accompanied by diffuse granular Ab42 deposits. Interestingly, unlike other animals such as monkeys and dogs [15], aged cheetahs and leopard cats do not develop argyrophilic senile plaques despite the fact that they develop diffuse Ab deposits in their brains. In the present study, analysis of the leopard cat APP gene detected a base substitution, which altered the N-terminal amino acid sequence of the Ab protein. Interestingly, many higher mammals that develop argyrophilic plaques, including dogs and monkeys, possess the same Ab amino acid sequence as humans [16], [17], [18]. The present study provides biological insights into the pathogenesis of AD.

Animal Brains
Most of the animals used in this study were wild animals that had lived exclusively on Tsushima Island, Nagasaki Prefecture, Japan. The Tsushima leopard cat is a subspecies of the leopard cat (Prionailurus bengalensis) [19]. The leopard cat was designated as a national endangered species in 1994 and ever since has been the focus of a conservation program funded by the Japanese government [http://kyushu.env.go.jp/twcc/multilang/english/ pamph.htm].
A retrospective study was performed using paraffin-embedded tissues from 14 individual brains (Table 1). The brains were obtained from routine necropsies performed at the Laboratory of Veterinary Histopathology, Kagoshima University; Laboratory of Veterinary Pathology, Yamaguchi University; the Tsushima Wildlife Center of the Ministry of Environment of Japan; or the Department of Veterinary Pathology, the University of Tokyo. Most of these animals were killed in road accidents. No animal was killed for the purposes of this study. Unfortunately, the precise ages of the animals were not determined except for two individuals (Case No. 1:3-days-old and Case No. 2:3-years-old) that died at a reproduction facility (Table 1). Case No. 13 and 14 had been kept in captivity at a conservation facility for 10 and 15 years, respectively (Table 1).

Histopathology
All brains were fixed in 10% phosphate-buffered formalin, coronally sliced, and then conventionally embedded in paraffin. The paraffin-embedded tissues were cut into 4-mm-thick serial sections. The deparaffinized sections were then stained with HE, periodic acid-methenamine silver (PAM), and the Gallyas-Braak method.

Immunohistochemistry
Consecutive sections were stained using the immunoenzyme technique. In order to deactivate endogenous peroxidase, the deparaffinized sections were immersed in 1% hydrogen peroxide in methanol for 20 minutes and then washed with Tris-buffered saline (TBS). The primary antibodies that were used in this study are listed in Table 2. After incubation with the primary antibody at 4uC overnight, immunolabeled antigens were visualized using the Dako Envision+ System (Dako, Carpinteria, CA, USA). In brief, the sections were incubated with HRP-labeled polymer at 37uC for 40 minutes, reacted with 0.05% 393-diaminobenzidine plus 0.03% hydrogen peroxide in Tris-hydrochloric acid buffer, and then counterstained with hematoxylin. Negative controls were obtained by omitting the primary antibodies.

Indirect Double Immunofluorescence Staining
The double immunofluorescence staining technique was also performed to determine which cell types contained hyperphosphorylated tau. Anti-MAP2, GFAP, and Olig2 antibodies were used as markers of neuronal cells, astrocytes, and oligodendrocytes, respectively. In addition, the coexistence of hyperphosphorylated tau with glycogen synthase kinase 3-b (GSK 3-b), phosphorylated GSK 3-b (Ser9), and ubiquitin was also analyzed.
After incubation with the primary antibodies at 4uC overnight, the sections were washed with TBS. As secondary antibodies, ALEXA594-conjugated goat anti-mouse IgG (Invitrogen, OR, USA) and ALEXA488-conjugated goat anti-rabbit IgG (Molecular Probes, OR, USA) were mixed with TBS (dilution: 1:100 for both antibodies). The sections were incubated with the secondary antibody mixture at 37uC for one hour, mounted with Vectashield (H-1500, Vector Laboratories, Burlingame, CA, USA), and examined under a Leica DMI 3000B fluorescence microscope (Leica Microsystems, Tokyo, Japan).

Electron Microscopic Analysis
Formalin-fixed paraffin-embedded brain tissue from Case No. 13 was deparaffinized, cut into 1-mm cubes, fixed in 2.5% glutaraldehyde 0.1M phosphate buffer (pH 7.4), and then postfixed in 1% osmium tetroxide 0.1 M cacodylate buffer (pH 7.2). The tissues were dehydrated in a graded series of ethanols, treated with QY-1 (Nisshin EM, Tokyo, Japan), and embedded in an epoxy resin (Quetol 651, Nisshin EM, Tokyo, Japan). Ultrathin sections from selected areas were stained with uranyl acetate and lead citrate and examined with a Hitachi H-7500 transmission electron microscope (Hitachi High-Technologies, Tokyo, Japan).

Scoring
The distributions of Ab42 and hyperphosphorylated tau were assessed using the following scoring methods. Ab42 deposition, 2: none, +: diffuse Ab42 deposition in the cerebral cortex, ++: diffuse   Ab42 deposition in the cerebral cortex and hippocampus, and +++: additional distinct plaque-like deposition; hyperphosphorylated tau, 2: none, +: a few AT8-positive cells were found in the parahippocampal gyrus, ++: AT8-positive cells were found in the parahippocampal gyrus and hippocampal CA1 region, and +++: in addition to the parahippocampal gyrus and hippocampal CA1 region, AT8-positive cells were also found in the ectosylvian gyrus and hippocampal CA3 region.

APP Transcript Sequence Analysis
Total RNA was extracted from the formalin-fixed paraffinembedded brain tissue of three leopard cats (Case No. 7, 8, and 14 were selected as they displayed the least postmortem changes) using the RNeasy FFPE kit (Qiagen, Tokyo, Japan). Subsequently, 10 ng of RNA were reverse-transcribed and amplified using the OneStep RT-PCR kit (Qiagen, Tokyo, Japan). For cDNA amplification, we designed a pair of primers covering exons 11-12 (forward primer 59-AGATCCGGTCCCAGGTTATG-39) and exons 16-17 (reverse primer 59-GTCGACCTCCACGA-CACC-39) of the domestic cat (Felis catus) APP gene (EN-SFCAG00000001556). The PCR products were electrophoresed on 2% agarose gel and then purified using the QIAquick Gel Extraction kit (Qiagen, Tokyo, Japan). Direct DNA sequencing was accomplished using the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, CA, USA) on the 37306l DNA Analyzer (Applied Biosystems, CA, USA).

Results
In immunohistochemical examinations, the brains of 6 individuals were found to be positive for Ab42, including 5 brains that were positive for hyperphosphorylated tau ( Table 1). Note that all of the individuals that possessed hyperphosphorylated tau also displayed Ab42 deposits, but the reverse was not true. The hyperphosphorylated tau-positive cells first appeared in the parahippocampal gyrus, and they subsequently spread through regions CA1 to CA3 of the hippocampus and into the ectosylvian gyrus (temporal lobe) in the more severely affected cases ( Figure 1A). On the other hand, Ab42 was initially diffusely deposited in the parietal and temporal cortices and subsequently spread to the hippocampal region. In some cases, Ab42 was deposited in a speckled pattern ( Figure 1B); however, these deposits were very diffuse, and none of these cases were stained with PAM staining. Also, Ab42 was not deposited in the vascular walls.
In general, Ab42 was granularly deposited in the neuropil of the pyramidal cell layer (Figure 2A). Interestingly, these deposits were not immunolabeled with anti-AbN1 antibody or anti-AbpN3 antibody on sequential sections ( Figure 2B, 2C). In addition to Ab peptides, aggregates of hyperphosphorylated tau were observed in neurites and perikarya ( Figure 2D). These aggregates displayed intense staining for both the 3 repeat (3R) and 4 repeat (4R) tau isoforms ( Figure 2E, 2F). Colocalization of pan tau antibody with hyperphosphorylated tau, 3R-tau and 4R-tau, also pan Ab antibody with Ab42 were confirmed ( Figure S1). With Gallyas-Braak staining, argyrophilic NFT and neuropil threads were abundantly observed in the areas containing hyperphosphorylated tau-positive cells (Figure 3). Ultrastructurally, some neuronal somata and neurites had been filled with bundles of filaments ( Figure 4A). These filaments formed paired structures with diameters of 10-20 nm. Most of the filaments were straight ( Figure 4B), and some constrictions were also observed suggesting helical structure ( Figure 4C).
Double immunofluorescence staining examinations revealed that aggregates of hyperphosphorylated tau had developed not only in neuronal cells but also in some oligodendrocytes ( Figure 5A, 5C). Note that there were no astrocytic plaques ( Figure 5B), which was also confirmed by the Gallyas-Braak method. In addition, the unphosphorylated form of GSK-3b, which is the major kinase involved in tau phosphorylation, colocalized with hyperphosphorylated tau in neuronal somata, whereas staining for phosphorylated-GSK-3b was negative in these cells ( Figure 6A, 6B). The hyperphosphorylated tau-positive cells were markedly positive for ubiquitin, which was distributed in a granular pattern ( Figure 6C). The aggregation of hyperphosphorylated tau was also confirmed by AT100 antibody (Figure 6D), which detects different phosphorylation site from AT8 antibody (AT8: Ser202/Thr205 and AT100: Ser212/Thr214).
Both 3R-tau and 4R-tau were expressed in the brains of the adult individuals regardless of the presence or absence of NFT formation ( Figure 7B, 7C). However, only the 3R-tau protein was expressed in the brain of the neonatal (3-days-old) leopard cat ( Figure 7A).

Discussion
Many animal species develop Ab deposits, especially higher mammals (e.g., monkeys, dogs, bears, camels, and horses) [4], [20], [21], [22], [23]. Most of these animals develop argyrophilic plaques, and some even develop mature plaques with amyloid cores. In contrast, felids seldom develop argyrophilic senile Figure 7. Sequential sections subjected to immunohistochemical staining of hyperphosphorylated tau (AT8), 3R-tau, and 4R-tau in the CA1 region of a neonate brain (A), an adult brain without AT8-positive aggregates (B), and an adult brain with AT8-positive aggregates (C). (A) Only the 3R-tau isoform was expressed in the brain of the neonatal leopard cat. (B, C) In the adult brains, both the 3R-tau and 4R-tau isoforms were expressed regardless of the presence or absence of AT8-positive aggregates. Bar = 100 mm. doi:10.1371/journal.pone.0046452.g007 plaques, but granular aggregates of Ab peptide are often observed in aged domestic cats and cheetahs [8], [11], [24], [25]. In the felid phylogenetic tree, the leopard cat lineage is located in between the cheetah lineage and the domestic cat lineage [14]. These three species are the most recent to have diverged among the 8 lineages of living felids. The findings obtained in this study further confirm the distinctive pattern of Ab deposition that occurs in the brains of felids (Figure 2A). In addition, we found that the N-terminal epitope of the leopard cat Ab peptide differs from that found in humans and other animals that develop argyrophilic plaques ( Figure 2B, 2C, 8) [2], [16], [17], [26]. It has recently been established that the N-terminal subtype of Ab peptides affects their aggregability, and hence, plaque formation [27], [28]. The alternative N-terminal epitope of the leopard cat might be responsible for the low aggregability of its Ab peptides, which do not seem to produce argyrophilic plaques or vascular deposits.
Most interestingly, nearly all of the individuals that displayed Ab deposition also possessed NFT (Table 1, Figure 3). Considering that non-human animals rarely develop NFT, a high incidence of NFT is likely to be a trait of this species. In order to determine whether these NFT correspond to the NFT found in AD, we investigated their histopathological characteristics. As we found that the distribution of hyperphosphorylated tau-positive cells in leopard cats was quite similar to that observed in human AD patients, we developed a scoring system based on the Braak staging method (Table 1) [29]. Although only a limited number of cases were available for study, subjective assessments suggested that the spread of tau hyperphosphorylation in leopard cats corresponds to the progression of AD [30].
Human diseases that involve the development of intracellular aggregates of tau protein, such as Pick's disease (PiD), corticobasal degeneration (CBD), progressive supranuclear palsy (PSP), argyrophilic grain disease (AGD), and AD, are termed tauopathies. In CBD and PSP, tau often aggregates in astrocytic processes, forming lesions called astrocytic plaques and tufted astrocytes [31], [32]. PiD and AGD produce distinctive tau inclusions in neuronal cell bodies, which are known as Pick bodies and argyrophilic grains, respectively. In the brains of leopard cats, hyperphosphorylated tau aggregates in neuronal cells and some oligodendrocytes, but not in astrocytes ( Figure 5). Oligodendrocytic tau inclusions are most prominently found in PSP and CBD, and to a lesser degree, in AD brains [33]. In sections of the leopard cat brains that had been stained with the Gallyas-Braak method, NFT and neuropil threads were observed (Figure 3), whereas no Pick bodies or argyrophilic grains were found. Ultrastructurally, the NFT were composed of paired filaments with diameters of 10-20 nm ( Figure 4B, 4C), which is consistent with the properties of NFT in AD brains [34]. However, most of the filaments exhibited a straight laminar structure, rather than the helical structure that is often seen in AD brains [35].
Tau protein is associated with microtubules and promotes their polymerization and stabilization. Exon 10 of the tau gene encodes the second of four microtubule-binding repeat domains; therefore, the alternative splicing of exon 10 results in tau isoforms with either three (3R-tau) or four (4R-tau) microtubule-binding sites. Although the pathomechanism is yet to be elucidated, the dominant tau isoform in inclusions varies among diseases [36], [37]. PiD develops coiled filaments composed of 3R-tau alone [38], [39], whereas in CBD, PSP, and AGD the filamentous aggregates are composed of 4R-tau alone [40], [41]. In the brains of the leopard cats, both 3R-tau and 4R-tau aggregated in neuronal cells ( Figure 2E, 2F, S1B, S1C) [36], [42], [43]. In addition, we found that only the 3R-tau isoform was expressed in the brain of the neonatal leopard cat (Fig. 7A). In human brains, tau expression shifts from the 3R-tau isoform alone to both the 3R-tau and 4R-tau isoforms between post-natal day 15 and 35 [44], [45], [46]. Therefore, the age-related expression pattern of tau isoforms and the components of the tau inclusions found in the leopard cat brains correspond to those of human aging.
As mentioned above, animals that develop argyrophilic senile plaques, such as monkeys and dogs, are known to display the same Ab peptide amino acid sequence as humans. On the other hand, Ab deposition has never been demonstrated in non-transgenic wild-type rodents, such as rats and mice. Rodent Ab displays three amino acid differences in its N-terminal region compared with human Ab, which are presumed to account for the absence of amyloid deposits in wild-type rodents [30]. As it was also the case in the leopard cat, the different amino acid residues are intensively located in the N-terminal region of the Ab protein [18]. The findings of the present study indicate that the leopard cat Ab peptide has an intermediate aggregative nature between those of human Ab and rodent Ab (Table 3). There is increasing evidence to suggest that weakly aggregative forms of Ab are more important for neurodegeneration than classical argyrophilic plaques consisting of a mass of aggregated Ab protein [51], [52], [53], [54], [55], [56]. However, wild-type rodents with non-aggregative Ab do not develop NFT. In a study using PS16APP transgenic mice, the agedependent accumulation of small Ab aggregates was found to be related to decreased GSK-3b phosphorylation, which resulted in tau phosphorylation [57]. Since AD-type NFT have never been observed in the brains of monkeys or dogs with senile plaques, it is generally considered that such non-human animals die before NFT develop [13]. The leopard cats that were examined in this study are only found on Tsushima Island, Japan. This subspecies Table 3. Senile plaque and neurofibrillary tangle formation in humans, leopard cats, and rodents. The lack of genetic diversity in this subspecies should be taken into consideration as a potential factor in the peculiar AD pathology seen in these animals. (TIF)