[1] 2w REGULAR PAPER

[1] 191w We investigated isoform composition of aggregated tau protein in brains with Pick's disease (PiD), corticobasal degeneration (CBD) and progressive supranuclear palsy (PSP) by immunoblot analysis of sarkosyl-insoluble fractions of brain homogenates. We also examined the adjacent brain tissues immunohistochemically with a rabbit antibody, Ex10, which specifically recognizes exon 10 of tau. The Ex10 recognizes tau isoforms with four microtubule-binding repeats (4Rtau) but not those with three microtubule-binding repeats (3Rtau). Sarkosyl-insoluble tau from the brains of patients with CBD and PSP consisted of 4Rtau. Insoluble tau from the PiD brains contained both 3Rtau and 4Rtau, where 3Rtau predominated over 4Rtau. In brain tissues of CBD and PSP, Ex10 immunostained all neuronal and glial tau-positive structures. They included pre-tangles, astrocytic plaques, tuft-shaped astrocytes, and oligodendroglial coiled bodies. In PiD brains, astrocytic inclusions were also positive for 4Rtau. However, the majority of, if not all, Pick bodies and oligo-dendroglial tau inclusions were negative for 4Rtau. Such results suggest that, in neurons and oligodendroglia, tau isoforms involved in the pathological processes differ between CBD/PSP and PiD, and are thus disease specific. This contrasts with the astrocytic tau isoforms that accumulate similarly in all three disorders.

[1] 119w The microtubule-associated protein tau aggregates as cytoplasmic inclusions in neurons and glial cells in a variety of neurodegenerative diseases. These include Alzheimer's disease (AD), Pick's disease (PiD), corticobasal degeneration (CBD) and progressive supranuclear palsy (PSP), which are now collectively referred to as tauopathies. In brain, six tau isoforms are produced by alternative splicing of mRNA. Three isoforms contain three microtubulebinding repeats in the C-terminal region (three-repeat tau: 3Rtau). The other three contain four microtubule-binding repeats by addition of the exon 10-derived 31-amino acid sequence (four-repeat tau: 4Rtau). Both 3Rtau and 4Rtau include isoforms with or without the insertion of 29 or 58 amino acids, corresponding to the exon 2 or exons 2 and 3 sequences, near the N terminus.

[2] 108w Recently, a number of exonic and intronic mutations in the tau gene have been identified in the families of hereditary frontotemporal dementia and parkinsonism linked to chromosome 17 (FTDP-17) [12,21]. Biochemical analyses of sarkosyl-insoluble tau from brains of FTDP-17 patients indicated that different pathogenic mutations caused aggregation of distinct tau isoforms [4,11]. Such results suggest that isoform-specific aggregation of tau may also play a role in the pathogenesis of sporadic tauopathies. Although a variety of tau-positive inclusions occur in neurons and glial cells in the brains of PiD, CBD and PSP [2,14,15,16,22,23], tau isoforms aggregated in these structures have not yet been fully identified. The aim of this

[3] 30w Tetsuaki Arai • Kenji Ikeda • Haruhiko Akiyama • Yasuo Shikamoto • Kuniaki Tsuchiya • Saburo Yagishita • Thomas Beach • Joseph Rogers • Claudia Schwab • Patrick L. McGeer

[4] 84w Distinct isoforms of tau aggregated in neurons and glial cells in brains of patients with Pick's disease, corticobasal degeneration and progressive supranuclear palsy study was to investigate isoform composition of aggregated tau in PiD, CBD and PSP brains. We employed immunoblot analysis of sarkosyl-insoluble tau preparations and immunohistochemistry with a rabbit anti-tau antibody specific for exon 10. Our results indicate that tau isoforms aggregated in neurons and oligodendroglia differ between PiD and CBD/PSP, while tau isoforms accumulated in astrocytes are common to these diseases.

[1] 264w Our results for CBD and PSP agree with a previous report [20], further supporting a notion that the abnormal accumulation of tau in both neuronal and glial cells in these diseases consists essentially of 4Rtau. Immunohistochemistry with a 4Rtau-specific antibody, Ex10, has confirmed that all tau-positive structures such as pre-tangles, tuftshaped astrocytes, astrocytic plaques and oligodendroglial coiled bodies, contain 4Rtau. In the immunoblot of CBD case 1, a minor band was detected at a position corresponding to 3Rtau without an insert (Fig. 1A, lane 2). Whether this 3Rtau band is derived from abnormally phosphorylated doublet remains unknown. In this case, weak unclear bands of nonphosphorylated tau were also seen before dephosphorylation (Fig. 1A, lane 1). Contamination of normal tau is less likely since addition of recombinant 3Rtau to brain homogenates from CBD cases did not result in the appearance or enhancement of nonphosphorylated tau bands in the sarkosyl-insoluble fraction (data not shown). The secondary deposition of a small amount of normal tau to aggregated tau inclusions might take place in the cells [1] in some CBD cases. In PiD, sarkosyl-insoluble tau comprises predominantly 3Rtau. This may be consistent with a previous report [5], in which the authors argued for 3Rtau-specific abnormality in PiD. However, we also detected a considerable amount of 4Rtau. Nondephosphorylated tau from our PiD samples was negative for Tau-1, indicating that both 3Rtau and 4Rtau were derived from abnormally phosphorylated aggregated tau but not from contaminated normal tau. Such a result is supported by Ex10 immunohistochemistry of adjacent brain tissues, in which we found 4Rtau accumulation in some abnormal tau-positive structures.

[2] 144w Among a large number of 4Rtau-negative PiB, occasional NFT and neuropil threads were stained positively for 4Rtau in the cortex of PiD patients. This is not surprising since Hof et al. have already reported frequent coexistence of NFT and PiB in the cortex of PiD [10] and NFT are known to contain all six tau isoforms [8]. Under such circumstances, significance of a few PiB-like round inclusions positive for 4Rtau is uncertain. At the light mi-croscopic level, it is hard to distinguish PiB from tangentially cut NFT. Nevertheless, it remains possible that a small number of PiB contain 4Rtau. At present, the origin of 4Rtau-positive neuropil threads in PiD, either glial or neuronal, is also unclear, although neuropil thread formation in axons or dendrites has been suggested in the neuropil of PiD brains [18,19]. These issues require an immunoelectron microscopic study for further elucidation.

[3] 75w The question arises as to whether 4Rtau detected in our PiD samples was derived from concomitant NFT. This is unlikely, however, since we did not detect any abnormal band in the immunoblot analysis of sarkosyl-insoluble fraction from the temporal neocortex of an AD case at a very early stage of the disease (data not shown). In this tissue, we found only a few NFT, which was comparable with non-demented elderly people and our PiD cases.

[4] 306w In our PiD cases we found a number of astrocytic inclusions that were positive for 4Rtau. They were clearly distinguished on a morphological basis from astrocytic plaques and tuft-shaped astrocytes, which are considered to be histopathological hallmarks of CBD and PSP, respectively [16]. The astrocytic inclusions in PiD are similar to those referred to as thorn-shaped astrocytes found in postencephalitic parkinsonism of Economo type, dementia pugilistica, AD, PSP and some aged controls [14]. Although some investigators have suggested that the occurrence of the astrocytic inclusions is of pathogenetic significance in PiD [6,15], presence of 4Rtau in these inclu- Fig. 4A-D Double immunostaining for Ex10 (purple) and AT8 (brown) in the temporal cortex of patients with PiD. A The majority of PiB are stained brown, indicating that they lack 4Rtau. Occasional Ex10-positive round inclusions are seen (arrow) among Ex10-negative PiB. B A few neurofibrillary tangles are labeled for Ex10 (arrow). C Some neuropil threads are stained for Ex10 (arrow). D AT8-positive thorn-shaped structures with a few short and thick processes are labeled for Ex10 (arrows). E,F High-power photomicrographs of thorn-like shaped structures doubly immunostained for AT8 (purple) and anti-GFAP (brown). They are present in the cytoplasm of GFAP-positive astrocytes. G Double immunostaining for Ex10 (purple) and AT8 (brown) in the temporal white matter of a PiD patient. A number of inclusions of curved or circular shape (arrows) are stained brown, indicating that they lack 4Rtau. H High-power photomicrograph of the inclusions in the white matter immunostained with AT8 (purple). Tissues are counterstained with 0.1% Kernechtrot to identify the nucleus of glial cells. I Double immunostaining for anti-human tau (pool2) (purple) and anti-human C4d (brown). A curve-shaped inclusion is seen in a C4d-positive oligodendroglia (PiB Pick bodies). A-D and G are at the same magnification. Bars A (also for B-D,G) 50 µm; E (also for F,H,I) 20 µm

[5] 37w sions sharply contrasts with 3Rtau-dominant abnormality in PiD. Thorn-shaped astrocytes in AD are also positive for 4Rtau (data not shown). These results rather indicate lack of disease specificity of this inclusion and relevance to reactive astrocytic responses.

[6] 97w On the other hand, many oligodendroglial inclusions observed in the white matter of the PiD brains are negative for Ex10, indicating that they consist exclusively of 3Rtau. Oligodendroglial tau inclusions have been reported in a variety of tauopathies such as CBD, PSP, dementia with argyrophilic grains and subacute sclerosing panencephalitis [2,3,13,22,23]. However, despite their morphological resemblance, oligodendroglial tau inclusions in CBD and PSP contain 4Rtau. We speculate that, in contrast to astrocytic tau inclusions, isoform specific pathology takes place in both neurons and oligodendroglia, where 3Rtau-dominant abnormality characterizes PiD, while 4Rtau-dominant processes occur in CBD and PSP.

[1] 135w Diagnoses of all cases were initially made on the basis of clinical signs and symptoms and were confirmed in every case by routine neuropathological examination. Tissues from the parietal region of three patients with CBD, the frontal region of two patients with PSP, and the temporal region of three patients with PiD, as well as three controls were used. A fresh-frozen brain block was taken from each case and cut into two pieces. One piece was homogenized and used for immunoblot analysis. The other piece was fixed in 70% ethanol containing 150 mM NaCl for 2 days and used for immunohistochemistry. Ethanol-fixed temporal cortex of one more PiD case was also available. This particular PiD case was complicated by the presence of AD pathology in the form of numerous neurofibrillary tangles (NFT) and senile plaques.

[2] 159w Sarkosyl-insoluble tau was extracted from brain homogenate as described previously [8]. The sarkosyl-insoluble pellet was washed twice with the same sarkosyl-containing buffer to remove possible contamination with normal tau. Samples were then dephosphorylated by incubation with E. coli alkaline phosphatase (type III; Sigma Fine Chemicals) [8]. All six isoforms of recombinant human tau protein were expressed in E. coli BL21 (DE3) [7] and were used as control samples. The dephosphorylated and nondephosphorylated samples as well as a mixture of six isoforms of recombinant tau were loaded on 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and blotted onto a polyvinylidene difluoride membrane (Millipore Co.). Membranes were incubated overnight with either anti-human tau (pool2), a phosphorylation-independent polyclonal antibody to tau [17], or anti-Tau-1 (Cedarlane, Canada), a phosphorylation-dependent antibody that recognizes only normal tau proteins [9]. Following incubation with a biotinylated secondary antibody, labeling was detected using the avidin-biotin-horseradish peroxidase complex (ABC) system (Vector Laboratory, USA) coupled to 4-chloro-1naphthol as the chromogen.

[3] 141w We developed two polyclonal anti-tau antibodies, Ex10 and c-tau, by immunizing rabbits with synthetic peptides conjugated to keyhole limpet hemocyanin. Ex10 was raised against the first 16 amino acids of exon 10 (KVQIINKKLDLSNVQS) so that it recognizes 4Rtau but not 3Rtau. C-tau was raised against the first 19 amino acids of exon 11 (KVTSKCGSLGNIHHKPGGG) to detect all six tau isoforms. Specificity of these antibodies was established by immunoblot analysis of the six isoforms of recombinant human tau. Labeling with these antibodies was abolished by absorption with the unconjugated antigen peptides. AT8, a monoclonal antibody specific to phosphorylated Ser 202, was purchased from Innogenetics (Belgium) and was used to identify tau pathology in tissue sections. A rabbit anti-glial fibrillary acidic protein (GFAP) (Dakopatts, Denmark) antibody and an anti-human C4d (Quidel, USA) were used to identify astrocytes and complement activated oligodendroglia [24], respectively.

[4] 96w Following cryoprotection in 15% sucrose in 0.01 M phosphatebuffered physiological saline (PBS) pH 7.4, brain blocks were cut on a freezing microtome into 30 µm thickness and treated as freefloating sections. For single staining, primary antibody labeling was detected using the ABC system coupled to a diaminobenzidine (DAB) reaction intensified with nickel ammonium sulfate to yield a purple precipitate. For double staining, primary antibody labeling in the second cycle was detected in the same way as single staining except that nickel ammonium sulfate was eliminated from the DAB solution so that a brown precipitate was obtained.

[5] 112w Immunolabeling with anti-human tau (pool2), a phosphorylation-independent polyclonal antibody to tau, showed that sarkosyl-insoluble tau extracted from all patients of CBD and PSP appeared as two major bands of 68 and 62 kDa (Fig. 1A, lanes 1, 3 and 6; B, lanes 1 and 4). After the alkaline phosphatase treatment of sarkosyl-insoluble materials, tau appeared as two major bands that align with recombinant 4Rtau with or without the 29 amino-acid Nterminal insert (exon 2) in all cases (Fig. 1A, lanes 2, 4 and 7; B, lanes 2 and 5). In a CBD case, a very faint band appeared at a position corresponding to 3Rtau without an N-terminal insert (Fig. 1A, lane 2).

[6] 160w Immunoblots of PiD samples with anti-human tau (pool2) showed diffuse smear-like background (Fig. 1C, lanes 1 to 4, 6 and 7). With a monoclonal phosphorylation-dependent antibody, Tau-1, the background labeling appeared only after dephosphorylation (Fig. 1C, lanes 9 and 10), indicating that the smear-like staining was derived from abnormally phosphorylated tau rather than nonspecific binding of the polyclonal antibody. In PiD, nondephosphorylated sarkosyl-insoluble tau migrated as two major bands of 64 and 58 kDa (Fig. 1C, lanes 1, 3 and 6). These bands were negative for Tau-1 (Fig. 1C, lane 9). After dephosphorylation, tau appeared as two major bands that align with 3Rtau with or without the 29 aminoacid insert (exon 2). In addition, minor bands were present at positions corresponding to 3Rtau with the 58 aminoacid insert (exons 2 and 3) and 4Rtau with or without the 29 amino-acid insert (Fig. 1C, lanes 2, 4, 7 and 10). No band was detected in the sarkosyl-insoluble fraction from control brains.

[7] 379w Ex10 labeled neuronal and glial tau-positive structures in the brains of CBD and PSP. In the parietal region of CBD A Nondephosphorylated tau from three CBD brains migrates as two major bands of 68 and 62 kDa (lanes 1, 3 and 6). Dephosphorylation reveals two major bands that align with recombinant tau isoforms (lane 5 and 8) of four microtubule-binding repeats (referred to as 4R) with or without the 29 amino acid N-terminal insert (exon 2) (lanes 2, 4 and 7). In addition, a faint band is detected in case 1 at a position corresponding to a tau isoform of three microtubule-binding repeats (referred to as 3R) without an N-terminal insert (lane 2). B Nondephosphorylated tau from the PSP brains migrates as two major bands of 68 and 62 kDa (lane 1 and 4). Dephosphorylation reveals two major bands that align with 4Rtau with or without the 29 amino acid insert (exon 2). C Immunoblots of sarkosyl-insoluble tau stained with anti-human tau (pool2) are shown from lane 1 to 8, and those with Tau-1 from lane 9 to 11. Nondephosphorylated tau from the PiD brains is immunolabeled with anti-human tau (pool2) as two major bands of 64 and 58 kDa (lanes 1, 3 and 6), but not with Tau-1 at all (lane 9). Dephosphorylated tau is stained as two major and one minor bands that align with all recombinant 3Rtau, as well as two or three minor bands that correspond to 4Rtau with both anti-human tau (pool2) (lanes 2, 4 and 7) and Tau-1 (lane 10). Smear-like diffuse background labeling is seen in all three PiD samples. Negative staining of lane 9 indicates the smear-like labeling is derived from phosphorylated tau aggregates in the PiD brains (CBD corticobasal degeneration, PSP progressive supranuclear palsy, PiD Pick's disease, 3/4Rtau tau isoforms with three/four microtubule-binding repeats) patients, Ex10 positively stained pre-tangles, argyrophilic threads (Fig. 3A) and astrocytic plaques (Fig. 3B) in the cortex, and oligodendroglial coiled bodies in the white matter. In the frontal region of PSP, Ex10 labeled pre-tangles, argyrophilic threads and tuft-shaped astrocytes in the cortex (Fig. 3C), and oligodendroglial coiled bodies in the white matter (Fig. 3D). Ex10-positive structures in CBD and PSP were comparable in both number and morphology with those positive for AT8 in all cases examined.

[8] 186w In PiD, Ex10 revealed fewer positive structures than those seen in AT8-stained nearby sections. Therefore, further analysis was performed with double immunostaining for Ex10 and AT8, where Ex10 was stained purple and AT8 brown. As shown in Fig. 4A, the vast majority of Pick bodies (PiB) were negative for Ex10 and, therefore, stained brown for AT8. Occasional Ex10-positive, round inclusions were seen among Ex10-negative PiB (Fig. 4A, arrow). In a PiD case complicated with AD-type changes, numerous NFT and neuropil threads were stained positively for Ex10 but, again, many PiB were Ex10 negative. In other three cases of PiD, NFT (Fig. 4B) and neuropil threads (Fig. 4C) occurred only occasionally and were even fewer than Ex10-positive round inclusions. The numbers of NFT and neuropil threads were far smaller than those of AD in these cases. Although the presence of AD-type pathology in most PiD brain seemed to be incidental, age-related findings, occurrence of Ex10-positive structures in the same tissue sections provided positive control for Ex10 staining and confirmed that negative staining of many PiB for Ex10 was not due to pre-or postmortem problems of these cases.

[9] 127w In the cortex of PiD, Ex10 also stained AT8-positive structures that took thorn-like shape with a few short and thick processes (Fig. 4D). Double immunostaining for AT8 and GFAP revealed that they were present in the cytoplasm of GFAP-positive astrocytes (Fig. 4E, F). Such astrocytic inclusions were found in all PiD cases examined but the frequency varied from case to case. In the white matter, a large number of inclusions with circular or arced shape were seen in all PiD cases. These structures were positive for AT8 but negative for Ex10 (Fig. 4G) and were often associated with nuclei of oligodendroglia (Fig. 4H). Double staining with anti-human tau (pool2) and anti-C4d showed that such tau-positive inclusions frequently occurred in the cell bodies of C4d-positive oligodendroglia (Fig. 4I).

[1] 45w Each of the six isoforms of recombinant tau was loaded on SDS-PAGE and immunoblotted with either c-tau or Ex10. Figure 2A illustrates that c-tau recognizes all six isoforms. Ex10, on the other hand, recognizes all three isoforms of 4Rtau but none of 3Rtau (Fig. 2B).