Address correspondence and reprint requests to Dr Catriona A. McLean, Department of Pathology, The University of Melbourne, Victoria 3010, Australia. E-mail: email@example.com
Intracellular inclusions containing α-synuclein (αSN) are pathognomonic features of several neurodegenerative disorders. Inclusions occur in oligodendrocytes in multiple system atrophy (MSA) and in neurons in dementia with Lewy bodies (DLB) and Parkinson's disease (PD). In order to identify disease-associated changes of αSN, this study compared the levels, solubility and molecular weight species of αSN in brain homogenates from MSA, DLB, PD and normal aged controls. In DLB and PD, substantial amounts of detergent-soluble and detergent-insoluble αSN were detected compared with controls in grey matter homogenate. Compared with controls, MSA cases had significantly higher levels of αSN in the detergent-soluble fraction of brain samples from pons and white matter but detergent-insoluble αSN was not detected. There was an inverse correlation between buffered saline-soluble and detergent-soluble levels of αSN in individual MSA cases suggesting a transition towards insolubility in disease. The differences in solubility of αSN between grey and white matter in disease may result from different processing of αSN in neurons compared with oligodendrocytes. Highly insoluble αSN is not involved in the pathogenesis of MSA. It is therefore possible that buffered saline-soluble or detergent-soluble forms of αSN are involved in the pathogenesis of other αSN-related diseases.
αSN has also been identified in the predominantly oligodendroglial inclusions of multiple system atrophy (MSA) (Arima et al. 1998; Spillantini et al. 1998; Tu et al. 1998; Wakabayashi et al. 1998). MSA is also characterized by macroscopic atrophy of regions such as the putamen and basis pontis with microscopic neuronal loss and astrocytosis. Whether αSN is expressed in normal oligodendrocytes and has a physiological role in these cells is unknown. The origin of the αSN in the pathological inclusions is therefore uncertain.
αSN in brain homogenate can be studied by a combination of differential centrifugation and Western blotting which separates multiple pools of protein on the basis of their solubility. The technique provides a different perspective to the more widely used immunohistochemical methods that demonstrate the cellular location of protein, in particular the aggregated protein in inclusions. It is not known what form of αSN is damaging to cells and so all pools of αSN, not just those in inclusions, need to be investigated. Recent work in AD has demonstrated a correlation between buffered saline-soluble Aβ and AD severity that does not exist with highly insoluble Aβ in amyloid plaques (Lue et al. 1999; McLean et al. 1999), and similarities may exist in αSN-related diseases. Western blotting can also differentiate between different sized forms of the protein of interest.
We have previously used differential centrifugation and Western blotting analysis to study the solubility and species of αSN in cortical grey matter of DLB, PD, AD and controls (Culvenor et al. 1999). Full-length (16 kDa) αSN, a 12-kDa species and a 6-kDa putative NAC (non-Aβ component of AD amyloid) fragment were identified in the buffered saline-soluble fraction. Sodium dodecyl sulfate (SDS)-insoluble material was only found in DLB cases. Further investigation in a larger series of DLB cases showed statistically significant increases in levels of SDS-soluble and SDS-insoluble αSN compared with controls. Significant changes in Tris-buffered saline (TBS)-soluble levels of αSN were not detected (Campbell et al. 2000). MSA is another αSN-related disease affecting oligodendrocytes as well as neurons. The aim of this study was to compare the amount, solubility and molecular weight species of αSN in MSA, DLB, PD and normal aged controls in order to identify disease-related changes. Comparison of two distinct but related diseases has the potential to improve understanding of the types of changes in αSN that are consistently associated with neurodegeneration and therefore more likely to be involved in pathogenesis. This study shows that SDS-insoluble αSN found in DLB and PD is not present in MSA. Highly insoluble αSN is therefore not involved in the pathogenesis of MSA and may not be necessary for the pathogenesis of other αSN-related diseases. The critical pathogenic species of αSN may lie in the buffered saline-soluble or SDS-soluble fractions.
Case selection and description
Fresh-frozen right hemibrains and formalin-fixed left hemibrains were obtained from the Australian Brain Bank that is supported by the National Health and Medical Research Council Australian Neuroscience and Mental Illness Research Network. A diagnosis of MSA was made on standard pathological criteria including the presence of αSN immunoreactive glial cytoplasmic inclusions (GCI) in white matter in conjunction with appropriate clinical features (Gilman et al. 1999). The topographic distribution of inclusions in the four selected MSA cases were in keeping with three being of an olivopontocerebellar atrophy type (OPCA) and one of a striatonigral degenerative subtype (SNA). A diagnosis of DLB was based on the consensus guidelines including the presence of Lewy bodies (LB) in the neocortex in conjunction with appropriate clinical features (McKeith et al. 1996). PD was diagnosed on clinical features and the presence of classical neuropathological findings of LBs, neuronal degeneration and depigmentation in the substantia nigra. The regions selected for biochemical analysis were chosen on the basis of areas of known protein aggregation as determined by histological analysis and the criteria given in Table 1. Control samples were taken from the same sites for comparison. Control cases had no clinical history of any neurological disorder and had no abnormal neuropathology on examination. For each disease process, sections from the same site as those chosen for biochemical analysis were taken from the contralateral formalin-fixed hemibrain and examined for the density of inclusions.
Table 1. Cases, sites selected and mean post-mortem interval
Pons, frontal white matter, cerebellar white matter
65 ± 11
Pons, frontal white matter, cerebellar white matter
67 ± 7.7
Dementia with Lewy bodies
Medial temporal cortex
73 ± 12
Medial temporal cortex
61 ± 16
77 ± 2
78 ± 7
Antibodies and recombinant protein
Polyclonal antibodies were raised in rabbits to the human αSN C-terminal domain (amino acids 116–131; Ab 97/8), the N-terminal domain (amino acids 1–18; Ab 97/5) and to the non-Aβ component of Alzheimer's disease amyloid (NAC) region (amino acids 75–91; Ab 42580) and characterized (Culvenor et al. 1999). Recombinant full-length human αSN was expressed in E. coli using the pRSETB vector (Invitrogen, Groningen, the Netherlands) and partially purified according to the method of Jakes et al. (1994).
Formalin-fixed tissue blocks were embedded in paraffin. Sections were treated with 80% formic acid for 5 min, 3% hydrogen peroxide for 5 min and incubated with blocking buffer (50 mm Tris-HCl, 175 mm NaCl pH 7.4 with 20% serum corresponding to species for secondary antibody) before incubation with the primary antibody. Ab 97/8 was used at 1 : 2000 dilution. Secondary reagents linked to horseradish peroxidase (Dako, Glostrup, Denmark) were used and visualized with diaminobenzidine. Sections were counterstained with haematoxylin.
Histological quantitation of inclusions
Intracellular inclusions seen on immunostained sections were quantitated as inclusions per high-power field (hpf). Granular immunoreactive fragments (< 3 µm in diameter) in tissue section were not included in the quantitation, being at the limit of accurate cell localization by light microscopy. Quantitation was performed on cortical Lewy bodies from the medial temporal cortex in DLB, substantia nigra Lewy bodies in PD and glial inclusions in MSA.
Homogenization and differential centrifugation of brain tissue
Fresh-frozen hemibrains stored at −70°C were brought to −20°C and samples from the anatomical sites specified were obtained (Table 1). Samples were homogenized (0.1 g/mL) by sonication at 4°C in TBS/sucrose buffer containing protease inhibitors [50 mm Tris-HCl pH 7.4, 175 mm NaCl, 1 m sucrose, 2 mm phenylmethylsulfonylfluoride (PMSF), 5 mm EDTA, 2 µg/mL aprotinin, 2 µg/mL pepstatin A, 5 µg/mL leupeptin, 2 µg/mL antipain] and centrifuged at 90 000 g for 2 h at 4°C. The overlying layer of crude myelin (negligible in cortical sites) was removed for later analysis and the supernatant fraction taken as the TBS-soluble fraction. The pellet was rinsed twice with TBS, resuspended in TBS and centrifuged at 100 000 g for 1 h at 4°C and the supernatant discarded. The remaining pellet was resuspended (at 24°C) in 5% SDS/TBS buffer and centrifuged at 100 000 g for 30 min at 25°C. The supernatant fraction was termed the SDS-soluble fraction. The pellet was resuspended in 8 m urea/8% SDS and termed the SDS-insoluble fraction.
Myelin was purified according to the modified protocol of Petratos et al. (1998). Briefly, white matter from an MSA case and a control was homogenized in a Teflon-glass homogenizer in 0.29 m sucrose (5% wt/vol), layered on 0.85 m sucrose in a 2 : 3 ratio and centrifuged for 45 min at 90 000 g. Crude myelin at the interface was collected, diluted 1 : 1 with 0.24 m sucrose and centrifuged at 17 000 g for 30 min. The pellet was resuspended in distilled water (osmotic shock) and re-pelleted by centrifugation at 17 000 g for 30 min. This was followed by a second discontinuous sucrose gradient, osmotic shock, two further rinses in distilled water, a third gradient, osmotic shock and two rinses to obtain myelin of high purity. Protein concentration was determined by bicinchoninic acid (BCA) protein assay (Pierce, Rockford, IL, USA) using 2% SDS to reduce lipid interference (Morton and Evans 1992). Western blotting was performed using 20 µg protein aliquots loaded on 10% Tris-Tricine SDS-polyacrylamide gel electrophoresis (PAGE) (Culvenor et al. 1999; Li et al. 1999) and Ab 97/8.
Samples of purified myelin were prepared for transmission electron microscopy by fixation in 2.5% glutaraldehyde/PBS for 45 min at room temperature. After washing four times for 5 min with PBS, samples were fixed in 2% osmium tetroxide, washed in distilled water, dehydrated in graduated acetone and embedded in Epoxy-Araldite (50 : 50) resin. Thin sections were stained with uranyl acetate and lead citrate and examined in a Siemens 102 electron microscope.
Protein concentration was determined by BCA assay. Western blot analysis was performed using various amounts of protein to obtain appropriate signal levels. For quantitation blots, 20 µg of TBS-soluble fraction and 15 µg of SDS-soluble and SDS-insoluble fractions were used for white matter sites and pons. These aliquots were mixed with sample buffer (63 mm Tris-HCl pH 6.8, 2% SDS, 5% β-mercaptoethanol, 5% glycerol, 1 mg/mL bromophenol blue), boiled for 5 min and analysed using 10% Tris-Tricine SDS-PAGE. TBS-soluble fraction samples were also analysed using native 10% Tris-glycine PAGE (electrophoresis buffer: 25 mm Tris, 192 mm glycine pH 8.3, in the absence of boiling, SDS, β-mercaptoethanol and urea). Proteins were electrophoretically transferred to nitrocellulose membranes (Biorad, Hercules, CA, USA) at 380 mA for 45 min at 4°C. The membranes were boiled for 5 min in PBS and blocked in 0.5% hydrolysed casein/PBS. The membranes of all three fractions were then incubated with primary antibody Ab 97/8 (1 : 10 000).
For qualitative analysis, bound primary antibody was detected using horseradish peroxidase-conjugated antirabbit immunoglobulin (Dako, Glostrup, Denmark) and enhanced chemiluminescence (ECL, Amersham Pharmacia Biotech, Buckinghamshire, UK). For quantitation [125I]-protein A (ICN Biochemicals, Irvine, California, USA) was used to detect bound primary antibody and the blot exposed to a phosphorimaging plate. Quantitation of the full-length αSN band was performed using macbas v2.5 software (Fuji Film Co., Tokyo, Japan). Adjustment for loading error was made by densitometry of Coomassie blue stained gel. Quantitation data were transformed logarithmically (base 10) to reduce positive skew. Two-tailed Student's t-tests were then used to assess the significance of the difference between means, and Pearson product moment correlation coefficients were determined. Statistical significance was confirmed with non-parametric tests. Each blot included a standard set of sample dilutions to ensure the linearity of antibody binding and detection.
α-Synuclein in normal controls and in disease
Brain samples from MSA, DLB, PD cases and normal aged controls (Table 1) were homogenized and fractionated by solubility into TBS-soluble, SDS-soluble and SDS-insoluble fractions. Western blotting using 10% Tris-Tricine SDS-PAGE and antibodies to the C-terminal domain (Ab 97/8), to the central NAC region (Ab 42580) and to the N terminus (Ab 97/5) was performed.
Comparable αSN levels in white matter (frontal) and grey matter (medial temporal cortex), pons and substantia nigra were achieved by loading six times more total protein from white matter samples. The amount of αSN per unit of protein is much lower in white matter. In the TBS-soluble fraction full-length αSN resolved as a 16-kDa band in all disease and control samples. A band at approximately 19 kDa was also detected at greater relative levels in white matter than in grey matter. A 12-kDa band and 6-kDa putative NAC fragment were also detected at low levels in all TBS-soluble samples examined using Ab 42580 (Fig. 1a). The 6-kDa band had equivalent migration when it appeared in white and grey matter, pons and substantia nigra.
In the SDS-soluble fraction, Ab 42580 detected full-length αSN and smaller amounts of 12-kDa species in all disease categories and control grey matter but did not detect any α-synuclein species in control white matter or pons (Fig. 1b). A smear above 24 kDa was also seen in disease.
SDS-insoluble αSN was not detected in white matter from controls or in MSA cases (Fig. 1c). A small amount of SDS-insoluble αSN was detected in the pons of MSA cases as a 16-kDa full-length protein and a 36-kDa band with Ab 97/8. Antibody Ab 42580 detected full-length αSN and a smear from the stacking gel down to 24 kDa but did not detect the 6-kDa species in MSA. In contrast, DLB and PD (but not controls) contained large quantities of SDS-insoluble full-length, 24- and 6-kDa αSN species using Ab 42580 and 36-kDa species using Ab 97/8.
A 12-kDa band was strongly detected in MSA and DLB by Ab 42580 as well as by antibody (Ab 97/5) to the N-terminus, and had the same rate of migration in frontal white, grey (medial temporal cortex) and pontine samples (Fig. 1d). Antibody to the C-terminal domain (Ab 97/8) did not detect 12- or 6-kDa species (Fig. 1e). Antibody 97/8 did, however, detect a band at approximately 36 kDa in the SDS-soluble and SDS-insoluble fractions of DLB and MSA cases. The 6-kDa putative NAC was not detected in any SDS-soluble fraction.
α-Synuclein inclusion quantitation
The mean density of intraneuronal Lewy body inclusions in the medial temporal cortex of DLB cases was 10 per hpf. In the substantia nigra of PD, inclusion density was 7 per hpf and the mean density of glial inclusions in MSA frontal white matter was 19 per hpf.
α-Synuclein in purified myelin
The crude myelin fraction formed on top of the 1 m sucrose following TBS extraction contained substantial amounts of αSN. This αSN contributed to the SDS-soluble fraction when 1 m sucrose was omitted from the buffer used in the initial 100 000 g centrifugation step. αSN became detectable in the SDS-soluble fraction of control white matter, however SDS-insoluble αSN remained undetectable with this modified protocol (not shown). To test whether the presence of αSN in crude myelin represented a specific association or contamination, myelin was purified. αSN levels dropped progressively with purification suggesting that αSN is not strongly bound to myelin (Fig. 2a). The MSA case had considerably more residual full-length αSN than the control by Western blot and the 36-kDa species was detected in MSA but not in control myelin. Standard ultrastructural analysis of the purified myelin showed myelin membrane figures with only occasional synaptic, cytoplasmic and lysosomal material present (Fig. 2b). Immunoelectron microscopy using post-embedding labelling of lightly fixed LR white resin-embedded brain tissue and purified myelin did not detect αSN in association with myelin using the same conditions that detected strong labelling of LB in DLB tissue (data not shown).
Comparison of α-synuclein solubility in MSA and in normal controls
The pattern of αSN solubility varies considerably between individual cases of MSA and in the distribution within each brain. Quantitation of total αSN in crude homogenate by Western blot with Ab 97/8 revealed a significant increase in MSA frontal white matter (p = 0.01), pons (p = 0.02) and cerebellar white matter (p = 0.02) compared with normal aged controls (Fig. 3a). In the TBS-soluble fraction Ab 97/8 detected a significantly increased αSN level in MSA frontal white matter compared with controls (p = 0.04). Significant differences in mean αSN level did not exist between MSA and controls in pons (p = 0.84) or in cerebellar white matter (P = 0.98) (Fig. 3b). Three MSA cases had decreased TBS-soluble αSN in the pons but the fourth case was increased above normal. On average, MSA cases had significantly increased levels of SDS-soluble αSN compared with controls in the frontal white matter (p = 0.003), pons (p = 0.001) and cerebellar white matter (p = 0.004) (Fig. 3c). These results were confirmed with Ab 42580. The fourth MSA case, which had high TBS-soluble αSN levels in the pons, also had low levels of SDS-soluble αSN and relatively few pontine GCI were detected by immunohistochemistry in this case. In individual MSA cases there was a significant inverse correlation between TBS-soluble and SDS-soluble levels of αSN in frontal white matter (r = − 0.89, p = 0.02). Similar trends were observed in cerebellar white matter (r = − 0.80, p = 0.08) and pons (r = − 0.76, p = 0.13), but did not reach significance with this number of cases. In preliminary analysis of single case–control data from other brain regions a small increase in SDS-soluble αSN was found in corpus callosum. Differences in αSN levels were not found in MSA occipital, cingulate or temporal cortex compared with controls. αSN in DLB white matter was indistinguishable from that in normal controls (data not shown).
Species of α-synuclein under native gel conditions
To further investigate potential alterations in the TBS-soluble pool of αSN, aliquots of TBS-soluble fractions from normal and MSA white matter as well as normal and DLB grey matter were analysed on 10% Tris-glycine native PAGE. In all cases a consistent pattern of seven bands was apparent with the slowest migrating doublet being the most abundant (Fig. 4a). The lowest band in brain homogenate migrated close to the level of partially purified recombinant αSN (representing the monomeric protein). Differences in the level of TBS-soluble αSN manifested proportionally in all bands. All bands were converted to monomer by addition of SDS (Fig. 4b).
This study shows that SDS-insoluble αSN is not found in MSA, contrasting with significant levels of SDS-insoluble αSN found in DLB and PD. This suggests that as SDS-insoluble αSN is not required for the pathogenesis of MSA, it may not be necessary for the pathogenesis of other αSN-related diseases. αSN is likely to have a central pathogenic role in diseases where it aggregates to form inclusions, most strongly indicated by the occurrence of causative mutations in the αSN gene in familial PD (Polymeropoulos et al. 1997; Krüger et al. 1998). The form of the αSN protein that is deleterious is unknown. The occurrence of end-stage MSA in the absence of SDS-insoluble αSN directs the search for the pathogenic species of αSN towards either the less aggregated/membrane associated αSN in the SDS-soluble fraction or even the TBS-soluble pool of αSN.
TBS-soluble αSN is abundant in the cortex and the level appears to be independent of disease (Campbell et al. 2000). However, potential changes in a small subpopulation of diseased neurons may be masked in this biochemical analysis. The species of TBS-soluble αSN present are also independent of the disease process in our study of pons, medial temporal cortex and substantia nigra. Quantitation of TBS-soluble αSN in white matter of MSA shows an increase compared with controls. This overall increase in TBS-soluble αSN, observed in MSA white matter, appears to be proportional in all bands separated by native gel electrophoresis. These multiple bands may reflect oligomerization of αSN with itself or with other proteins, but there was no evidence of abnormal forms of αSN in the TBS-soluble fraction of MSA or DLB. Increased levels of TBS-soluble αSN were not observed in MSA pons, possibly due to the competing effect of decreased αSN production following pontine neuronal loss.
Increased expression of αSN could be directly toxic by overstimulating the normal function of αSN. Alternatively it could enhance the aggregation of αSN into forms that are the abnormal toxic species by providing the critical concentration of αSN required for nucleation dependent self-aggregation, similar to that observed in vitro (Wood et al. 1999). Research directly testing toxicity of αSN is limited. External administration of aggregated αSN and NAC peptide was toxic in vitro (El-Agnaf et al. 1998), but differs significantly from the intracellular expression and aggregation of αSN in vivo. Overexpression of αSN in cell culture caused toxicity although the mechanism is unknown (Ostrerova et al. 1999). Either overstimulation of normal αSN function or aggregation (which was not examined) may have been responsible.
The mean density of glial inclusions visible by light microscopy in MSA white matter is greater but comparable to that seen in the intraneuronal LB of DLB and PD. These intracellular inclusions of MSA, DLB and PD would be expected to contain the most insoluble αSN in the brain homogenates. Our results therefore imply that GCI contain only SDS-soluble αSN but LB contain SDS-insoluble αSN (and possibly SDS-soluble αSN as well). The difference in the maximum insolubility of αSN in GCI and LB may result from differences in the aggregation of αSN in the oligodendroglial or neuronal environment in disease. Differences in the time course of disease development in MSA and DLB may also be involved although survival following clinical onset is comparable (Ben-Shlomo et al. 1997; Olichney et al. 1998). The absence of the 6-kDa putative NAC fragment in the SDS-soluble and SDS-insoluble fractions of MSA brain homogenate suggests that, in contrast to LB, GCI do not contain the 6-kDa species. It will be important to establish whether the differences in end-stage aggregates are merely a byproduct of events downstream of the pathogenic insult or if they signify fundamental differences in the preceding pathogenic events in MSA, DLB and PD.
Whether αSN incorporated into inclusions can be actively involved in pathogenesis is unknown. Some evidence has suggested that sequestration of αSN in inclusions may even be protective. Investigation of the substantia nigra of PD demonstrated that cells with LB were less likely to die than cells without LB (Tompkins and Hill 1997). If inclusions themselves are not pathogenic this would suggest that any sequestered αSN in the SDS-soluble fraction is also not pathogenic.
The major αSN abnormality in MSA compared with controls is the significant accumulation of SDS-sensitive aggregates and the presence of a 36-kDa SDS-stable species that may be an αSN dimer. Consistent with this, the presence of SDS-soluble αSN in putamen of MSA but in not controls was reported by Dickson et al. (1999), although there were minor differences in the fractionation buffers. These differences may account for the presence of SDS-soluble αSN in control cortex, not observed by Dickson et al. We have found that 1% triton, used by Dickson et al. (1999), is capable of extracting this αSN component (data not shown).
One MSA case in our study had relatively low levels of SDS-soluble αSN in the pons. This is consistent with variability in the topographical distribution of neuropathology observed in MSA. Few inclusions were visible by immunohistochemistry in the pons of this case (SND subtype). A small amount of SDS-insoluble material (representing SDS-resistant aggregated protein) was detected in the pons of the other MSA cases (OPCA subtypes). Although this may indicate that a small component of MSA inclusions is SDS-insoluble, some of the αSN aggregates in the pons are in neurons. The SDS-insoluble material detected may therefore be purely neuronal in origin.
Crude myelin fractions from MSA white matter were found to contain more αSN than those from normal controls, and the 36-kDa putative dimer was only present in MSA. This αSN did not copurify although some was still present after several cycles of purification. αSN is therefore not strongly associated with myelin in homogenized tissue. It is not known whether αSN can come into contact with myelin in vivo. Myelin pathology is a feature of MSA (Matsuo et al. 1998). Whether this is a direct result of αSN-mediated damage or secondary to the primary pathogenic insult, which may be neuronal or glial, is unknown.
αSN appears to be centrally important in the pathogenesis of MSA, DLB and PD. Mutations in the αSN gene in familial PD and the protein's presence in GCI and LB strongly suggest that physicochemical alterations in αSN can lead to disease. These alterations lead to a transition from normal soluble αSN to insoluble aggregates. Protein aggregation is associated with cell damage but whether aggregation itself is the key pathogenic event remains to be established. The absence of SDS-insoluble αSN in MSA demonstrated in this study suggests that highly aggregated species are not necessary for pathogenesis. Further research into the less aggregated and soluble forms of αSN is necessary in order to elucidate the molecular pathogenesis of αSN-related diseases and open potential avenues for therapy.
We thank Tina Cardamone and Jo Merriner for assistance with histochemistry, Dr Roberto Cappai and Denise Galatis for recombinant αSN, and Dr David Small for discussion. This work was supported in part by grants from the National Health and Medical Research Council of Australia and the University of Melbourne.