Introduction

Multiple system atrophy (MSA) is an adult onset neurodegenerative disease characterized by varying degrees of parkinsonism features, cerebellar ataxia, corticospinal dysfunction and autonomic dysfunction [73, 74], but the defining feature of MSA is the presence of glial cytoplasmic inclusions (GCIs) [52]. MSA brains show varying extent of demyelination and atrophy, especially in the cerebellum, pons, and medulla, as well as the loss of pigmented cells in the substantia nigra pars compacta [73, 74]. Previously, patients with MSA were classified under three clinical categories: olivopontocerebellar atrophy, striatonigral degeneration and Shy-Drager syndrome [28]. However, it is now widely accepted that MSA is a single disease entity and this notion has led to the consensus recommendation that the classification of patients with MSA should be simplified to MSA-C (MSA with prominent cerebellar ataxia) or MSA-P (MSA with prominent parkinsonism) depending on the relative predominance of clinical and pathological abnormalities [26].

GCIs are found in oligodendrocytes and they can have a flame-, triangular-, or sickle-shaped appearance [52]. GCIs can be observed throughout the white matter, but they are more abundant in the basal ganglia, substantia nigra, pontine nucleus, medulla and cerebellum [2, 12, 39]. Ultrastructural analysis of GCIs demonstrated that they are composed of a meshwork of randomly arranged, loosely packed filaments with cross-sectional diameters of 15–30 nm [1, 34, 35, 44, 52]. Using immunological, ultrastructural and biochemical approaches, it has been demonstrated that these fibrils are predominantly comprised of polymerized α-synuclein (α-syn) [3, 11, 12, 68, 69, 70].

α-syn fibrils are also the major component of other types of pathological lesions, such as Lewy bodies (LBs) and Lewy neurites, which are defining features of Parkinson's disease (PD) and dementia with LBs (DLB) [62, 63]. Pathological inclusions comprised of α-syn are present in a spectrum of neurodegenerative disorders that are collectively referred to as α-synucleinopathies [14].

α-syn is predominantly expressed in central nervous system neurons, where it is localized mainly at presynaptic terminals [21, 31, 32, 75]. It is unclear if α-syn is normally expressed at low levels in oligodendrocytes in adult brain, but it can be detected in cultured rodent oligodendrocytes [57]. It is possible that an abnormal increase in expression of α-syn in these cells leads to the formation of GCIs, although α-syn mRNA does not seem to be increased in the white matter of MSA brains [49]. It is likely that other changes in the cytoplasmic milieu, perhaps the accumulation of α-syn-interacting proteins, must occur to promote the fibrillogenesis of α-syn. Some reports have demonstrated that the microtubule-binding protein tau, which interacts with α-syn in vitro [33], can be detected in GCIs. However, these findings are controversial since immunostaining analyses of GCIs with anti-tau antibodies have been reported to range from entirely negative [34, 44] to a substantial proportion of GCIs being labeled [1, 8, 38, 43, 51, 54, 56, 66].

Recently, it was reported that the 14-3-3 proteins are present in the majority of classical and cortical LBs in PD and DLB [36]. The 14-3-3 family of proteins is ubiquitously expressed in mammalian cells, but they are especially abundant in brain, where they account for ~1% of total soluble protein [7]. There are at least seven mammalian isoforms of 14-3-3 [7, 61], which exist as dimers, and they are thought to be adaptors that serve to modulate signal transduction pathways by preferentially binding to phosphorylated motifs [18, 61]. Notably, 14-3-3 proteins have been shown to bind to α-syn in living cells [47].

The presence of 14-3-3 and tau in GCIs was analyzed in 21 cases of MSA to determine if they were interacting proteins that might play a role in GCI formation. These proteins were found in a majority of GCIs in some brains. However, in most cases, only a minority or none of the GCIs were positive for 14-3-3 and tau, suggesting that their accumulation in GCIs is secondary to the fibrillogenesis of α-syn.

Materials and methods

Antibodies

Anti-14-3-3β (clone H-8) murine monoclonal antibody (mAb) was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Antibody 17026 is a rabbit antiserum raised against full-length recombinant tau and it detects all isoforms of tau [30]. PHF-1 [48] and PHF-6 [29] are phosphorylation-dependent mAbs that specifically react with tau phosphorylated at Ser-396/404 and Ser-231, respectively. Tau-1 is a dephosphorylation-specific mAb that recognizes an epitope within amino acid residues 189–209 in the longest isoform of tau [6, 65]. Murine anti-α-syn mAbs LB 509 and Syn 202 have been described previously [5, 23].

Cell culture

HEK293 cells were obtained from the American Type Culture Collection (Rockville, MD) and cultured in Dulbecco's modified medium-high glucose 4.5 g/l (DMEM) supplemented with 10% FBS, 100 U/ml penicillin, 100 U/ml streptomycin, and 2 mM l-glutamine. HEK 293 cells stably expressing human α-syn have been described previously [53].

Gel electrophoresis and Western blotting

Cells were harvested in phosphate-buffered saline (PBS; 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4), lysed in 2% SDS, 62.5 mM TRIS, pH 6.8, and protein concentrations were determined using the bicinchoninic acid (BCA) assay (Pierce, Rockford, IL). The cell extracts were diluted in SDS-sample buffer (10 mM TRIS, pH 6.8, 1 mM EDTA, 40 mM DTT, 1% SDS, 10% sucrose), and the proteins were resolved on slab gels by SDS-polyacrylamide gel electrophoresis. Proteins were electrophoretically transferred to nitrocellulose membranes (Schleicher and Schuell, Keene, NH) in buffer containing 48 mM TRIS, 39 mM glycine and 10% methanol. Membranes were blocked with 5% skimmed milk powder dissolved in TRIS-buffered saline-Tween (20 mM TRIS, pH 7.7, 137 mM NaCl and 0.1% Tween-20), incubated with primary antibodies followed by anti-mouse horseradish peroxidase (HRP)-conjugated antibodies and visualized by enhanced chemiluminescence as recommended by the manufacturer.

Tissue collection and processing

The harvesting, fixation and further processing of the tissue specimens studied here were conducted as previously described [12, 58]. Briefly, tissue blocks were removed at autopsy and fixed by immersion in 70% ethanol with 150 mM NaCl or 10% buffered formalin for 24–36 h. The samples were dehydrated through a series of graded ethanols to xylene at room temperature and infiltrated with paraffin at 60°C according to a previously described schedule [67]. The blocks were then cut into multiple, near serial 6-μm sections for immunohistochemical staining. The diagnostic assessment of all MSA cases was performed in accordance with published guidelines [20, 26, 42].

Immunohistochemistry

Immunohistochemistry was carried out using the avidin-biotin complex (ABC) detection system (Vector Laboratories, Burlingame, CA) and 3,3'-diaminobenzidine (DAB) as described [12]. Briefly, sections were deparaffinized and re-hydrated, endogenous peroxidases were quenched with 5% hydrogen peroxide in methanol for 30 min and sections were blocked in 0.1 M TRIS with 2% donor horse serum (TRIS/DHS) for 5 min. To try to improve antigen detection, some selected sections were pre-treated with 88% formic acid (FA) for 1 or 5 min or heated in a microwave oven for 5 min prior to treatment with hydrogen peroxide. All antibodies were diluted in TRIS/DHS. Primary antibodies were incubated overnight at 4°C. After washing, sections were sequentially incubated with species specific, biotinylated secondary antibodies for 1 h and ABC complex for 1 h. Bound antibody complexes were visualized by incubating sections in a solution containing 100 mM TRIS, pH 7.6, 0.1% Triton X-100, 1.4 mM diaminobenzidine (DAB), 10 mM imidazole and 8.8 mM hydrogen peroxide.

Evaluation of immunohistochemical data

The cerebellum and pons of MSA brains, which are rich in GCIs, were evaluated for the purpose of this study. The abundance of immunoreactive GCIs was assessed using a semi-quantitative assessment strategy, as described [12]. The regional density of immunoreactive inclusions was graded in each region as follows: 3, frequent; 2, moderate; 1, few; and 0, none. Sections were graded independently by two observers (B.I.G and M.E.M.).

Results

Specificity of 14-3-3β antibody

To study the presence of 14-3-3 proteins in GCIs, a pan-14-3-3 mAb (clone H-8) that recognizes a conserved epitope within the N terminus of 14-3-3 proteins was used. The specificity of this antibody was determined by Western blot analysis of HEK 293 cells extracts (Fig. 1). The anti-14-3-3 mAb specifically recognized a major immunoband of ~30 kDa and a less intense band with a slightly higher molecular mass, consistent with the predicted migration of 14-3-3 proteins on SDS-PAGE [18, 61]. Since 14-3-3 proteins have some sequence homology to α-syn [47], the possible cross-reactivity of this antibody with α-syn was assessed by Western blot analysis using extracts of HEK 293 cells stably overexpressing human α-syn, and this showed that α-syn was readily detected with the anti-α-syn mAb LB509, but there was no cross-reactivity of the anti-14-3-3 mAb with α-syn (Fig. 1).

Fig. 1.
figure 1

Specificity of anti-14-3-3β antibody. Cell lysates (5 μg) from HEK 293 cells or HEK 293 cells stably expressing human α-syn (293/α-syn) were resolved on 13% SDS-polyacrylamide gels. Proteins were electrophoretically transferred onto nitrocellulose membranes that were probed with anti-14-3-3β mAb H-8 or anti-α-syn mAb LB509. The mobility of molecular mass markers is indicated on the left of each panel (α-syn α-synuclein, mAb monoclonal antibody)

Full size image

Immunohistochemical detection of 14-3-3 and tau in GCIs

The abundance of GCIs was assessed in each MSA brain using the Syn 202 mAb (Fig. 2A), because it detected these lesions most robustly [12]. Tissue sections from the pons and cerebellum were used to determine the presence of tau and 14-3-3 in GCIs, since these regions are rich in these lesions (Table 1). In one case (case 1) there was a high proportion of GCIs that stained positive for 14-3-3 (Fig. 2B, Table 1), but in the majority of cases only a moderate or low number of GCIs were 14-3-3 positive (Table 1). There were also a few cases were no GCIs were detected with the anti-14-3-3 mAb.

Fig. 2.
figure 2

Immunostaining of GCIs with antibodies to α-syn, 14-3-3 and tau. Immunolabeling of cerebellar sections of multiple system atrophy brains with antibodies Syn 202 (A), 14-3-3β (H-8) (B), 17026 (C), and Tau-1 (D). Higher magnifications of GCIs labeled with each antibody, respectively, are shown in the lower insets. Staining with PHF-6 is shown in the upper inset of D (GCI glial cytoplasmic inclusion). Bar A–D 160 μm; insets 80 μm

Full size image
Table 1. Summary of MSA subjects studied. Semiquantitative estimate of the density of immunoreactive glial cytoplasmic inclusions (using the antibodies indicated) in cerebellum and pons were graded as follows: 3, frequent; 2, moderate; 1, few; 0, none (MSA multiple system atrophy, PMI post-mortem interval, n/a not available, Dx diagnosis, MSA-P MSA with prominent parkinsonism, MSA-C MSA with prominent cerebellar ataxia, E ethanol, F formalin, Age age at death)
Full size table

In a subset of MSA cases, some GCIs were immuno-detected with the anti-tau antibody 17026. This polyclonal antibody, raised against full-length recombinant human tau, was selected because it recognizes all the alternatively spliced isoforms of tau and it is phosphorylation independent [30]. There was no case where the density of tau-positive inclusions was very abundant (Table 1), although in some fields a significant number of GCIs were stained by 17026 (Fig. 2C), but in most cases infrequent or no GCIs were stained with 17026. Tau-positive inclusions were also strongly immunoreactive with the Tau-1 mAb specific for non-phosphorylated tau (Fig. 2D). Conversely, GCIs were unstained with the phosphorylation-specific antibody PHF-1 and only weakly reactive with the phosphorylation-specific antibody PHF-6 (Fig. 1D, upper inset). Antigen retrieval with FA treatment or heating in microwave oven did not enhance the staining with the anti-14-3-3 or anti-tau antibodies. In cases (e.g., case 1) where GCIs were frequently stained for 14-3-3 or tau, double-labeling analysis demonstrated that these proteins could accumulate together in GCIs. However, in most cases the major of GCIs were rarely double-labeled for 14-3-3 and tau (data not shown).

There was no direct association between the accumulation of tau in GCIs and the formation of tau inclusions in other region of the brain, such as the hippocampus. For example, cases with the most abundant tau-positive GCIs (cases 1, 3, 12 and 19) had little or sparse tau pathology in the entorhinal cortex (data not shown). Conversely, cases with abundant tau pathology in the entorhinal cortex (cases 5, 7, 17, and 20) (data not shown) had no or few tau-positive GCIs (see Table 1).

Discussion

Increasing evidence suggests that the polymerization of α-syn and the formation of pathological α-syn inclusions leads to the impairment of cellular functions and the degeneration of affected cells. For example, the pathological mutation A53T, which has been identified in at least 12 families [4, 41, 50, 55, 64], increases the propensity of α-syn to polymerize in vitro [10, 22, 45]. Although patients carrying the A53T mutation were initially diagnosed with familial PD, re-examination revealed that the distribution of α-syn pathology is significantly more widespread throughout the neuraxis than in classical idiopathic PD [13, 64]. Furthermore, the clinical presentation of affected individuals with the A53T mutation includes several features (e.g., dementia, prominent myoclonus and urinary incontinence) that are not characteristic of or common to classical idiopathic PD [27, 64]. Expression of human A53T α-syn in transgenic mice results in severe motor impairment that coincides with abundant α-syn pathological inclusions, but mice expressing equivalent levels of wild-type human α-syn do not display these impairments or pathological changes [24].

The events that result in the formation of pathological inclusions comprised of the wild-type α-syn protein remain enigmatic. It is possible that the interaction of α-syn with other proteins can induce the formation of inclusions. Here, the presence of two α-syn-interacting proteins, tau and 14-3-3, were investigated in the GCIs of 21 patients with MSA. Both proteins were detected in a subset of GCIs, but not in all the MSA cases. In some studies it has been reported that tau can be detected in at least a significant number of GCIs [1, 8, 38, 43, 51, 54, 56, 66], but in other studies the presence of tau in GCIs was not observed [34, 44]. In the cohort studied here, some GCIs in a subset of MSA cases were tau immunoreactive, but in some cases the α-syn-positive GCIs demonstrated a paucity of tau immunoreactivity (Table 1). Furthermore, the majority of these MSA cases showed no tau-positive GCIs in the cerebellum. These results indicate that the controversy of whether tau is present in GCIs can be explained, at least in part, by the specific subset of patients and the brain regions analyzed in prior independent studies, and this might reflect intrinsic pathological heterogeneity in MSA, not unlike the clinical variability in this disorder. Furthermore, previous analyses of GCIs with anti-phosphorylation specific antibodies suggested that tau was hypophosphorylated in these lesions [8]. Consistent with the results by Cairns et al. [8], GCIs containing tau demonstrated a paucity of staining with PHF-1, weak staining with PHF-6 (Fig. 2D, upper inset), and robust labeling with Tau-1 (Fig. 2D).

The accumulation of tau in pathological lesions in oligodendrocytes is also observed in other neurodegenerative disorders, such as coiled bodies in supranuclear palsy and corticobasal degeneration [9, 15]. Coiled bodies are typically morphologically distinct from GCIs. They commonly encircle the nucleus and extend into proximal cellular processes. These coiled bodies are comprised of bundles of tau filaments and they are not stained with antibodies to α-syn [69]. Alternatively spliced isoforms of tau containing the exon 10 insert (i.e., four microtubule binding repeats, 4R-tau) are the predominant components of coiled bodies [17, 60]. In one study, where the isoforms of tau in GCIs were analyzed biochemically, it was determined that 4R-tau isoforms were the major tau species in GCIs [54]. Unfortunately, frozen tissue from cases with sufficient tau-positive GCIs was not available to us to perform a similar biochemical analysis.

In some MSA cases, 14-3-3 proteins accumulated in a significant proportion of GCIs. However, in other cases, 14-3-3 could not be detected in any GCI. Recently, Kawamoto et al. [37] reported the presence of 14-3-3 proteins in GCIs; however, in their cohort of patients a high percentage of, but not all, GCIs were immunopositive for 14-3-3 proteins. These findings suggest that interaction between 14-3-3 and α-syn is not necessary for the formation of GCIs. Moreover, due to the temporal limitations of analyzing post-mortem specimens, it is difficult to ascertain at which stage of GCIs formation 14-3-3 proteins are incorporated. Once available, transgenic mouse models of MSA with an age-dependent accumulation of GCIs will be better suited to address this issue. 14-3-3 polypeptides are believed to act as adaptor proteins that bind to phosphorylated motifs [18, 61], and 14-3-3 has been shown to interact with α-syn in cells, while α-syn has also been shown to be a phosphoprotein [46] that is hyperphosphorylated in pathological lesions of PD, DLB and MSA [19]. It has been suggested that phosphorylation of α-syn may play a role in promoting fibril formation [19], but the mechanism to account for how this occurs remains to be elucidated. It is possible that 14-3-3 accumulates in GCIs by binding to phosphorylated α-syn. Similar to 14-3-3 and tau, synphilin, another α-syn-binding protein, also is found only in a subset of α-syn inclusions [71].

The proteins α-syn and tau share many similar properties. Both can adopt an unfolded confirmation in solutions [59, 72] and they can polymerize into fibrils leading to the formation of pathological lesions [14, 16]. They are also predominantly neuronal proteins, although they are likely expressed at low levels in oligodendrocytes [40, 57] and they can accumulate in these cells to form inclusions. The presence of tau in some GCIs is consistent with our resent findings that α-syn fibrils formation can be an initiator of tau fibrillogenesis [25]. The changes in cellular physiology that result in the formation of α-syn and tau inclusions are still unclear, but it is possible that the accumulation of interacting proteins in these inclusions may contribute to cell death by leading to the sequestration of key molecular mediators of homeostatic processes, thereby further impairing the regulation of normal cellular functions.