Similar neuronal imprint and no cross-seeded fibrils in α-synuclein aggregates from MSA and Parkinson’s disease

Aggregated alpha-synuclein (α-syn) is a principal constituent of Lewy bodies (LBs) and glial cytoplasmic inclusions (GCIs) observed respectively inside neurons in Parkinson’s disease (PD) and oligodendrocytes in multiple system atrophy (MSA). Yet, the cellular origin, the pathophysiological role, and the mechanism of formation of these inclusions bodies (IBs) remain to be elucidated. It has recently been proposed that α-syn IBs eventually cause the demise of the host cell by virtue of the cumulative sequestration of partner proteins and organelles. In particular, the hypothesis of a local cross-seeding of other fibrillization-prone proteins like tau or TDP-43 has also been put forward. We submitted sarkosyl-insoluble extracts of post-mortem brain tissue from PD, MSA and control subjects to a comparative proteomic analysis to address these points. Our studies indicate that: (i) α-syn is by far the most enriched protein in PD and MSA extracts compared to controls; (ii) PD and MSA extracts share a striking overlap of their sarkosyl-insoluble proteomes, consisting of a vast majority of mitochondrial and neuronal synaptic proteins, and (iii) other fibrillization-prone protein candidates possibly cross-seeded by α-syn are neither found in PD nor MSA extracts. Thus, our results (i) support the idea that pre-assembled building blocks originating in neurons serve to the formation of GCIs in MSA, (ii) show no sign of amyloid cross-seeding in either synucleinopathy, and (iii) point to the sequestration of mitochondria and of neuronal synaptic components in both LBs and GCIs.

INTRODUCTION Synucleinopathies, such as Parkinson's disease (PD), dementia with Lewy bodies (DLB), or multiple system atrophy (MSA), are a group of neurodegenerative disorders of unknown etiology with very diverse clinical and pathological presentations. However, these diseases share certain standard features, the most significant being the aggregation of alpha-synuclein (α-syn), present in cytoplasmic inclusions bodies (IBs) in the cells of affected brain regions. These IBs are divergent in nature and cellular localization across these synucleinopathies. Lewy bodies (LBs) are found in neurons for PD and DLB, while glial cytoplasmic inclusions (GCIs) are detected in oligodendrocytes for MSA 1,2 . LBs are eosinophilic inclusions comprising filamentous structures in a dense core surrounded by a peripheral halo 3 containing numerous membranous components and dysmorphic organelles 4 . GCIs are packed triangle or sickle-shaped, filamentous structures 5 . α-syn is the principal filament constituent of both IBs 2,5,6 .
Besides their histopathological characterization, the origins, the mechanisms of formation, and the pathophysiological role of IBs remain unelucidated. In agreement with early studies showing that IB-harboring neurons show no apparent signs of apoptosis 7,8 and that there is no correlation between the extent of LB pathology and the extent of neuronal depletion in PD [9][10][11] , it appears likely that the α-syn fibrils accumulated in the IBs are not toxic by themselves, and that their compaction represents the result of a neutralization process rather. Instead, the concomitant and cumulative incorporation of mitochondria and proteins essential for cell function into the growing IBs could sporadically be detrimental to the host cell 12 . At the same time, amyloidogenic partner proteins like tau or TDP-43 could also, in some instances, be locally cross-seeded and cause the cell's demise [13][14][15] . In both cases, the catabolism of IBs released by the dying host cells in the extracellular space could then cause a local leak of small IB fragments and α-syn fibrils from the IB mass, fragments that in turn would be capable of being taken up by the neighboring cells in which they would trigger a de novo aggregation of endogenous α-syn, the formation of novel IBs, and so on. Indeed, in mice, the intracerebral injection of IBs extracted from human DLB brains leads to the development and to the spread of a synucleinopathy with neuropathological and cytological features that are strikingly identical to the one observed after injections of pure recombinant α-syn fibrils 16,17 . Thus, if this IB-dependent intercellular spread mechanism holds, it is tempting to speculate that the composition of IBs present in a host cell could reflect the history of the spread, i.e., show proteins of the donor cell transferred together with the IB fragment seeds and eventually associated with the α-syn fibrils mass in the neo-formed IB. Such a theoretical possibility of retracing the spread history of IBs could probably help to solve the enigma concerning GCIs in MSA. These inclusions feature a high load of fibrillar α-syn, which contrasts with the very low physiological abundance of the protein in mature oligodendrocytes 18,19 . To date, the origin of these oligodendroglial IBs remains to be unraveled.
To this end, we used a different approach in the present study. We purified and isolated aggregated α-syn and its associated insoluble proteomes from PD and MSA brains using their sarkosylinsolubility. We used Sarkospin, a previously developed procedure for purifying pathological protein aggregates by sedimentation 17,28,29 . By adapting Sarkospin to α-syn, we sought to specifically isolate the aggregated forms of the protein from their physiological monomeric and oligomeric counterparts. The latter separation allowed us to scrutinize the accompanying insoluble proteomes associated with each disease and shed light on the cellular origin and the components of IBs.

RESULTS
Extraction of pathological alpha-synuclein from synucleinopathy brains and separation from its regular counterparts by an adapted Sarkospin procedure Intending to extract and identify the insoluble proteomes associated with aggregated α-syn for each synucleinopathy, we adapted the Sarkospin procedure to the purification of α-syn pathological assemblies found in human post-mortem tissue samples of PD and MSA subjects. We previously developed this method to extract aggregated TDP-43 29 . It was adapted by using ultracentrifugation of the sample mixed within a sucrose cushion after its solubilization in sarkosyl at 37°C with simultaneous nuclease treatment (Fig. 1a). The resulting supernatant contains sarkosyl-soluble material. The insoluble entities are collected in a dry pellet that is compatible with a further identification or quantification of its protein content by immunoblotting or mass spectrometry. Above all, the extraction of pathological assemblies through insolubility allows a direct comparison to control samples by applying the same procedure to healthy brain samples. We thus proceeded with parallel Sarkospin extractions, sedimentations, and filter-trap assays of sarkosyl-insoluble aggregates (Fig.  1b, c) from the brains of human control, PD, and MSA subjects (n = 3 independent subject brains per control or disease group).
The results indicate that the adapted Sarkospin procedure efficiently separated pathological aggregated α-syn from its soluble counterparts. As observed in Fig. 1b, high molecular weight α-syn species that are filter-trapped on nitrocellulose membrane are specifically present in the pellet fractions of PD and MSA samples (MJFR1, nc). In contrast, small mono-and oligomeric forms of the protein that get to pvdf membranes are observable mainly in all groups' supernatants (MJFR1, pvdf).
Another marker of pathological forms of the protein, namely its S129-phosphorylation (EP1536Y positive signal), indicates that large insoluble pS129-α-syn-containing aggregates are isolated in the pellet fractions for synucleinopathy samples (EP1536Y, nc) as opposed to the physiological EP1536Y-positive soluble entities found in all supernatants (EP1536Y, pvdf). Also, α-syn aggregates purified in synucleinopathy sample pellets have proved to be peculiarly resistant to denaturation (SDS, heat) and proteolysis (Proteinase K) while supernatants comprised only sensitive monomeric bands ( Supplementary Fig. 2a, b). Notably, actin, yet physiologically forming large assemblies, is found only in supernatant Sarkospin fractions and witnesses the procedure's efficient solubilization (Fig. 1a, b).
By quantification on filter-trap immunoblots for n = 9 brains, of Sarkospin supernatant and pellet fractions content in the proteins mentioned above, it appeared that over 80% of the nc-trapped αsyn in brain extracts from PD and MSA patients was pelleted, and less than 5% in brain extracts from control subjects (Fig. 1c, MJFR1 nc). In PD and MSA samples, S129-phosphorylated α-syn coincided with the insoluble α-syn pool since 90% of the pS129-αsyn signal (Fig. 1c, EP1536Y nc) was found in the pellets. In clear contrast, this figure was less than 5% for the control subjects. Noteworthy, a small subset of insoluble aggregates passes to pvdf membranes, but the majority (60−70% for PD and MSA, 99% for CTL) of α-syn present on this support is from supernatants.
Collectively, these results indicate that the novel Sarkospin procedure efficiently extracts pathological α-syn aggregates from human brain samples and physically separates these entities from physiological mono-and oligomeric α-syn forms present in healthy brain samples. Thus, we attained a purification method compatible with a subsequent proteomic analysis to extract and identify insoluble proteins from synucleinopathy subject brains and their direct comparison with parallel Sarkospin extractions from control patient brains.
Prime enrichment in alpha-synuclein and unidentified insoluble proteins in PD and MSA Sarkospin pellet fractions Proteomic analysis of Sarkospin pellets extracted from control, PD, and MSA samples (n = 3 independent subject brains per group) identified a total of 1022 proteins that resisted sarkosyl solubilization ( Fig. 2a We calculated bibliographic LB and GCI scores for the identification in this dataset of the proteins previously detected by proteomic analysis in the literature as components of pathological cytoplasmic inclusions. These indices relate to the number of mass spectrometry studies identifying the indicated protein in LBs and GCIs out of six 3,6,[23][24][25]27 and four 5,6,23,25 publications, respectively. Interestingly many of the most significantly enriched candidates were not identified in previous studies and had a null bibliographic LB and GCI score (smallest-sized red and blue bubbles close to α-syn on Fig. 2a, b, respectively). Besides, many proteins previously identified (i.e., with high bibliographic LB or GCI indices) are found in the extracts. Still, they are not significantly enriched in synucleinopathy pellets compared to control patient samples with no synucleinopathy (large dark bubbles on Fig. 2a, b). This could be explained by the absence in previous studies of a comparator group with equivalent fractions from healthy brains allowing to rule out an unselective enrichment. Indeed, for proteins physiologically highly expressed in the brain, their high abundance could simply lead to an increased detection probability whatever the IB purification method, irrespective of the IB pathology.
We challenged this possibility for all the 1022 proteins present in the insoluble proteome. Figure 3 represents the absolute abundance of each protein in control and disease samples, with color-coding the score of identification in the literature (bibliographic LB and GCI scores, Fig. 3a, c) or in our proteomic analysis (Z score, Fig. 3b, d), and allows to highlight the association between these parameters. These results unambiguously show that most of the proteins in LB and GCI proteomes previously reported in the literature are abundant in both control and pathological insoluble fractions (colored "tip of the spear", Fig. 3a, c). Instead, selecting the proteins showing a significant enrichment in IB vs. control brain fractions allows the identification of pathology-related proteins independently of their abundance level, as a result, even rare proteins can be detected as significantly enriched in PD or MSA pellets as compared to controls (colored "side of the spear", Fig. 3b, d).
Therefore, the present Sarkospin procedure, coupled with a subsequent comparative proteomic analysis, is an adequate method for unraveling insoluble proteomes and unbiasedly refining the list of candidates by considering their fold enrichment with regard to control samples. This approach defines the insoluble proteomes specific for each synucleinopathy and identifies LB and GCI components.
Major overlap between PD and MSA insoluble proteomes A total of 206 unique proteins were gated previously as significantly enriched in at least one of the synucleinopathy types compared to controls (Fig. 2a, b, red and blue, DataSet 4). Among this group of proteins, 84 are significantly enriched in PD as well as in MSA compared to controls with FC > 1.5 and p < 0.05 (DataSet 5). Inversely, among the 206 disease-associated proteins, only seven proteins were found to be selectively enriched in one synucleinopathy compared to the other, with three PD-specific and four MSA-specific (Fig. 4, red and blue, DataSet 6, 7). This highlights the extreme narrowness of the PD-and MSA-specific insoluble proteomes and indicates that the most substantial part of synucleinopathy-associated insoluble proteomes is shared between these two pathologies.
In other words, these data show the tremendous overlap of PD and MSA-associated insoluble proteomes. More specifically, in the most enriched proteins for both diseases or even for MSA specifically, a consistent proportion of candidates have been identified to play a role in PD pathogenesis (DataSet 3, 5, 7). Namely, for example, NipSnap-1 (average FC = 7, p = 8.0 × 10 −3 ) was shown to be implicated in Parkin-related mitophagy 30 ; Septin-5 (average FC = 4.6, p = 2.5 × 10 −3 ) was associated with dopamine-dependent neurotoxicity in early-onset PD also linked to Parkin 31 ; even Protein phosphatase methylesterase 1 which we find to be MSA-specific (Fig. 4a) has previously been also implicated in PD pathogenesis as enriched in PD subject substantia nigra 32 and playing a role in α-syn phosphorylation through PP2A activity regulation 33 .   Fig. 1 Extraction of pathological alpha-synuclein and separation from its normal counterparts from synucleinopathy brains by an adapted Sarkospin procedure. a Schematic representation of the adapted Sarkospin protocol. Human control, PD, or MSA brain samples were homogenized and solubilized with sarkosyl at 37°C in the presence of Benzonase before a single ultracentrifugation step within a sucrose cushion to separate supernatant and pellet fractions. b, c Biochemical analysis of Sarkospin fractions extracted from human brain samples. Supernatants and pellets were subjected to filter trap on layered nitrocellulose (nc, top) and pvdf (bottom) membranes immunoblotted against human α-synuclein (MJFR1), S129-phosphorylated α-synuclein (EP1536Y) or ß-actin (actin). b Representative pictures of the immunoblots for all samples, membranes, and antibodies. c Quantifications of each protein's percentage residing in Sarkospin pellet for all human samples used in this experiment. Bars represent mean with standard deviations for n = 3 biologically independent human brain samples per group.  Notably, from the four MSA-specific candidates we identify, none shows a glial-specific expression or a glial tropism ( Fig. 4a−d,  DataSet 7). Besides protein phosphatase methylesterase 1 described above (Fig. 4a), Phospholipid transporting ATPase IA acts as aminophospholipid translocase at the neuronal plasma membrane (Fig.  4d); Calnexin plays a role in receptor-mediated endocytosis at the synapse (Fig. 4c); and Vesicle-fusing ATPase is directly implicated in vesicle-mediated transport and fusion of vesicles to membranes (Fig. 4b).
Altogether, while the absence of oligodendroglial makers in the MSA-specific insoluble proteome is unexpected, the massive overlap of PD and MSA insoluble proteomes could suggest a common neuronal origin of these insoluble entities in both PD and MSA.
To put this possibility under scrutiny, we compared the lists of 130 and 160 proteins enriched respectively in PD and MSA pellets, to the published insoluble proteome found in an experimental neuronal synucleinopathy modelized in mouse neuronal cultures using preformed α-syn fibrils (PFFs) as seeds 12 ( Supplementary   Fig. 3). Strikingly, 38 and 58 insoluble proteins are shared between these PFF-treated neurons and PD and MSA respectively, with several -such as NipSnap-1 for PD, and Arginase-1 for MSA, showing a substantial enrichment in both proteomics studies ( Supplementary Fig. 3a, b). This striking overlap with yet completely independent studies and samples types validates several candidates, and brings a strong support to the notion that GCIs in MSA are made of α-syn amyloid building blocks preassembled in neurons To confirm the enrichment of these candidates in PD, MSA, or PFF-treated neurons insoluble proteomes, we ran an independent Sarkospin assay and subjected the fractions to the dot blot analysis used for the validation of Sarkospin (Fig. 1), with antibodies directed to ten proteins of interest, after validating their specificity on western blots (Supplementary Figs. 4 and 5, "Methods" section). The latter analysis allowed to eventually validate the significant enrichment of proteins such as NipSnap-1 and Septin-5 in PD pellets or Arginase-1, Fumarase, and ATP8A1 in MSA pellets. Spear charts representing mass spectrometry analysis of the protein content of all n = 9 Sarkospin pellets. Each of the 1022 master proteins identified in the present proteomic study is plotted as a circle positioned upon the total abundance of the protein (log 10 Abundance, A.U.) in control samples and its total abundance in PD (a, b) and MSA (c, d) samples. a, c Spear charts are respectively color-coded upon the protein LB and GCI scores described in Fig. 2. b, d Spear charts are color-coded respectively upon the protein Z score for PD and MSA, with Z = FC × (−log 10 p) for each disease group.
Gene ontology analysis is supportive of the prime neuronal origin of GCIs and indicative of the absence of cross-seeded amyloid partners like tau or TDP-43 To better decipher the cellular localization or the implication in specific pathways and processes of the proteins identified in pathological insoluble proteomes, possibly indicating their origin and reason of enrichment, we ran a gene ontology (GO) analysis. The results of the GO overrepresentation test were performed on the list of 199 overlap proteins (Fig. 4, black) compared to the list of all 1022 proteins identified. This statistical test allows identifying clusters of biological processes, molecular functions, and cellular components specifically overrepresented in synucleinopathy insoluble proteomes by quantifying their respective enrichment and significance compared to controls (Fig. 5, fold change, p-value, Z score, and #proteins, DataSet 8).
First of all, a clear-cut result is that no cross-seeded amyloidogenic partner proteins are associated with fibrillar α-syn in PD and MSA.
The bubble chart representation of GO analysis in Fig. 5 compellingly shows that the most significantly enriched and importantly represented GO clusters for synucleinopathies indicate a neuronal imprint, with a remarkable synaptic and mitochondrial membrane trend (Fig. 5). Indeed, all biological processes, molecular functions, and cellular components reaching maximal fold change and statistical significance are related to synaptic and mitochondrial localization or pathways. Surprisingly, while even MSA-enriched insoluble proteomes are represented in these GO clusters, no protein identification nor GO cluster reflects a glial localization or tropism. Instead, all components of the latter clusters are enriched in neuronal cells. More specifically, mitochondrial actors (Mitochondrial proton/calcium exchanger protein, Fumarate hydratase, Pyruvate carboxylase, Sidoreflexin, Very longchain specific acyl-CoA dehydrogenase, Cytochrome c oxidase) and even to a greater extend synaptic proteins (NipSnap-1, Septin-5, Synapsin-1, Synapsin-2, Synaptoporin, Glutamate receptor 2) are among the most enriched proteins in both PD and MSA pellets (Fig. 4).
Out of the 199 proteins used in the present GO analysis, 52 and 50 have been identified previously as components of LBs and GCIs, respectively, in at least one of the ten available proteomic studies 3,5,6,20,21,[23][24][25]27 . Therefore, the insoluble proteomes of PD and MSA identified by Sarkospin most likely represent the core components of IBs. Their association with aggregated α-syn is tight enough to resist the harsh sarkosyl solubilization of the extraction procedure.
Collectively, the primary neuronal composition of synucleinopathy insoluble proteomes-and of MSA-specific one to a greater extend-the identification of PD-related proteins in MSA-enriched candidates and the synaptic origin of these components indicate the prime neuronal source of preassembled building blocks serving to the formation of oligodendroglial GCIs in MSA. Our conclusion is ideally in line with the experimental observation that in mice inoculated with α-syn PFFs, there is an initial florid neuronal synucleinopathy that develops to progressively decrease PD vs. MSA  with time and leave the place to the progressive appearance of oligodendrocytes harboring GCI-like inclusions 34 . This is also in line with the notion that overexpression of α-syn in oligodendrocytes fails to cause the buildup of GCIs 28 . Finally, our observations do not support the idea that cross-seeding processes involving amyloidogenic proteins other than α-syn occur in PD and MSA.

DISCUSSION
The development of a novel procedure by adaptation of our formerly published Sarkospin method 28,29 (Fig. 1) allowed the extraction and purification of pathological aggregates from synucleinopathy subject brain tissues. The data presented here confirm that α-syn is the only amyloid and by far the most enriched protein present in PD and MSA sarkosyl-insoluble fractions of brain samples 6 (Fig. 2). A growing body of evidence tends to suggest that the conformation of amyloid α-syn itself would be the molecular basis of the distinct phenotypic "strain" features associated to PD and MSA, respectively [35][36][37][38] . The presence of this neuronal protein in oligodendrocyte inclusions is the core point of the enigma we sought to address in the present study 39 : what are the origin and the reason for the formation of LBs and GCIs? Characterizing the proteomes of LBs and GCIs has been repeatedly attempted in previous studies in order to understand which proteins are involved in their formation and their packing. Besides α-syn, mass spectrometry analysis of these IB-enriched fractions identified hundreds of proteins that potentially play a key role in the pathogenesis or the origin and mechanisms of formation of LBs and GCIs. These proteins include structural and cytoskeletal elements (neurofilaments, tubulins, tubulin polymerization promoting protein, TPPP/p25), α-synuclein-binding proteins (14-3-3, synphillin-1), components of the ubiquitinproteasome system, α-ß-crystallin, heat shock proteins, or DJ-1. However, these candidates' relevance to the molecular pathology of IBs can be limited by the ability of the experimental procedures to isolate IBs with sufficient purity and yield from the surrounding structures and protein contaminants. Recently, a purification method including a partial proteolysis step allowed the refinement of this list of proteins and demonstrated a high burden of synaptic vesicle-related proteins of the IBs 25 . These candidates are implicated in clathrin-mediated endocytosis (clathrin, AP-2 complex, dynamin), retrograde transport (dynein, dynactin, spectrin), or synaptic vesicle fusion (synaptosomal-associated protein 25, vesicle-associated membrane protein 2, syntaxin-1).
Using extracts from control, PD, and MSA brains, the comparative proteomic analysis we performed identified proteins enriched in PD and MSA brain samples (Fig. 2), with novel undescribed candidates, and refinement by excluding a list of previously identified proteins as probable false positives and/or protein not tightly associated with the sarkosyl-insoluble α-syn amyloids of IBs (Fig. 2) 3,5,6,20,21,[23][24][25]27 .
The protein enrichment in the respective PD or MSA insoluble proteome showed a tremendous overlap between them, with only few proteins being significantly different (Fig. 4). Lastly, the latter results associated with a gene ontology analysis revealed compellingly that both insoluble proteomes are composed of a vast majority of neuronal proteins, with a significant synaptic overrepresentation (Fig. 5), as well as numerous mitochondrial proteins. These data are in total agreement with previous proteomic studies 25 and microscopic analysis 4 which revealed a burden of synaptic-vesicle-related proteins, and membranous structures, such as vesicles or dysmorphic mitochondria. Noteworthy, our study independently validated many synaptic proteins found in the previously mentioned study 25 . These results strengthen the hypothesis of the neuronal origin of GCIs. It is tempting to speculate that in MSA α-syn could primarily get aggregated in neurons, then released as preformed inclusions bodies, eaten-up by surrounding oligodendrocytes, and eventually matured/stored as inert GCIs 34 . The fact that oligodendrocytespecific proteins are not enriched in MSA samples by our Sarkospin procedure is in line with the idea that the oligodendrocyte contribution is late, with glial proteins only secondarily and loosely associated with the GCIs and not properly with the amyloid core of the inclusion, previously pre-assembled in neurons. Two hypotheses could then explain the sparse neuronal inclusions observed in MSA brains: (i) these inclusions would contain the highly pathogenic α-syn aggregates that would not have-yet-killed the host neurons or (ii) these IBs would consist of less pathogenic amyloid assemblies, possibly similar to PDassociated α-syn aggregates, with a different amyloid conformation.
Noteworthy, the aim of our procedure was to refine the lists of candidate proteins to the ones very tightly associated with the inclusion bodies (the "core" of LBs/GCIs). Doing so, a substantial number of proteins loosely associated are solubilized by sarkosyl treatment and therefore not identified in our study. In other words, the absence from our lists of a protein previously identified as a component of the inclusions does not disprove its association with the pathologic IBs, but suggests instead that this association is relatively loose.
Also, our study is in line with the description of the delayed formation of GCI-like inclusions in oligodendrocytes upon PFF treatment 34 . In this model, oligodendrocytes possibly tend to form inclusions in order to sequester the exogenous pre-assembled entities. This mechanism-reminiscent of our hypothesis of an uptake of aggregates by oligodendrocytes from neighboring neurons-would drastically differ from a proper amyloid replication and pathogenic templating, the latter happening solely in neuronal cells where monomeric α-syn is available for templating. As to the overexpression-based models causing an upregulation of α-syn in oligodendrocytes, we have recently shown, using phosphorylated pS129-α-syn as surrogate marker of aggregation in this type of MSA model, that previous investigations were misled by the fact that second to overexpression the α-syn monomers get phosphorylated in the absence of any aggregation. In addition, in these overexpression-based models, the phosphorylated monomers associate with myelin giving a false impression of homo-multimerization in western blots or separation by centrifugation 28 .
Of particular importance, the list of proteins identified in the present study is restricted to the production of peptides identifiable by mass spectrometry, deriving from the presence of tryptic cleavage sites. Suppose one considers the high representation of neurodegenerative disease-associated proteins in partially trypsin-resistant proteins. In that case, it is possible that their enrichment here was underestimated, or even that some escaped identification. However, the conditions used here are known to reveal TDP-43 and tau 29 , and the absence of detection of crossseeded amyloid partners does not thus derive from an insufficient proteolytic cleavage.
Extending the present study to other brain regions and at different pathology stages would be of particular interest. In particular, using amygdala or putamen extracts for focusing on LBs or GCIs protein compositions, respectively, should help refining the present lists and identifying less represented candidates. Also, Sarkospin extraction and purification of synucleinopathy subject brain tissue yet devoid of any inclusion 40 , and the insoluble proteomes' comparison to the ones we defined here are of high importance for understanding the formation and composition of IBs. The Sarkospin procedure we developed here allows testing the seeding ability 36 or the pathogenicity of aggregated proteins 37 , an asset to understand the role of the inclusions in the disease process. Also, facing these insoluble proteomes' components to lists of identified interactants of mono-oligomeric physiological α-syn, or α-syn amyloid binders would help understand why the presence of these proteins in the inclusions and to decrypt their mechanisms of pathological association. The substantial lipidic membranes content of the LBs was recently investigated 4 . Thus, in a future study, we plan to take the advantage of the Sarkospin fractionation in a lipid floating layer, and a pellet containing lipids still associated to the IBs despite the harsh sarkosyl treatment. Performing a lipidomic analysis on both fractions similar to the present proteomic analysis shall identify the lipids specific to one or the other synucleinopathy, as well as their possible tight association with the IBs. The latter would keep, as for proteins, imprints from the past pathogenic mechanisms, and represent possible leads in understanding the disease processes.

Ethics statement
Samples were obtained from brains collected in a Brain Donation Program of the Brain Bank "GIE NeuroCEB" (Neuro-CEB BB-0033-00011). The consents were signed by the patients themselves or their next of kin in their name, in accordance with the French Bioethical Laws. The Brain Bank GIE NeuroCEB has been declared at the Ministry of Higher Education and Research and has received approval to distribute samples (agreement AC-2013-1887).

Human brain samples
Human cortices (cingulate gyrus) were dissected from freshly frozen postmortem brain samples from n = 3 control, sporadic PD or MSA subjects respectively. The choice of cingulate gyrus samples was based on a prior biochemical analysis of different brain regions from the CTL, PD, and MSA subjects in this series by immunoblotting against pS129 phosphorylated ( Supplementary Fig. 1a) and aggregated ( Supplementary Fig. 1b) α-syn. On the basis of these data, we found that the cingulate gyrus was the most comparable region in terms of total amyloid α-syn load for these PD and MSA subjects. In addition, the pathological records regarding the contralateral hemispheres indicated the presence of substantial amounts of LBs and GCIs in this area for the PD and MSA subjects, respectively. We thus proceeded with a comparative proteomic analysis of the insoluble proteomes using this brain region.

Human subject anonymized information
Human subject anonymized information is provided in Table 1.

Sarkospin procedure
For the extraction and purification of aggregates from brains samples, the Sarkospin procedure was adapted from previously published protocols 28,29 . Brain tissue samples were homogenized at 10% (w/v) in solubilization buffer (SB): 10 mM Tris pH 7.5, 150 mM NaCl, 0.1 mM EDTA, 1 mM DTT, Complete EDTA-free protease inhibitors (Roche) and PhosSTOP phosphatase inhibitors (Roche) using a gentleMACS Octo Dissociator (Miltenyi Biotec) with M Tubes, and the Protein extraction program. Samples were mixed 1:1 with SB 4% (w/v) N-lauroyl-sarcosine (sarkosyl, Sigma), 2 U.μl −1 Benzonase (Novagen), and 4 mM MgCl 2 , reaching a final volume of 500 μl. Sarkospin solubilization was then performed by incubating the samples at 37°C under constant shaking at 600 rpm (Thermomixer, Eppendorf) for 45 min. Solubilized samples were then mixed 1:1 with SB 40% (w/v) sucrose, without sarkosyl, MgCl 2 , or Benzonase, in 1 ml polycarbonate ultracentrifuge tubes (Beckman Coulter) and centrifuged at 250,000 × g for 1 h at room temperature with a TLA 120.2 rotor using an Optima XP benchtop ultracentrifuge (Beckman Coulter). The supernatant was collected by pipetting. For filter trap and SDS-PAGE immunoblot analysis, pellets were resuspended directly in the tube with 100 µL of the buffer corresponding to the supernatant (SB 1% sarkosyl 20% sucrose), and mixed with the same buffer in a fresh tube for reaching 1 ml (equal volumes to supernatant). For proteomics analysis, pellets were resuspended in 100 µL SB, and equalized for their total protein concentration quantified by bicinchoninic acid (BCA) assay (Thermo), prior to denaturation in Laemmli buffer.

Analysis of the protein contents of Sarkospin fractions by filter trap
For filter trap assays, native Sarkospin fractions were spotted onto layered nitrocellulose and PVDF 0.2 µm membranes (Protran, GE) using a dot blot vacuum device (Whatman). Membranes were fixed for 30 min in PBS with PFA 0.4% (v/v) (Sigma) final concentration. After three washes with PBS, membranes were blocked with 5% (w/v) skimmed powder milk in PBS-Tween 0.5% (v/v) and probed with primary and secondary antibodies in PBS-Tween with 4% (w/v) BSA (Antibody Table). Immunoreactivity was whether visualized by chemiluminescence or infrared using Clarity ECL and Chemidoc (Biorad) or Odissey systems (Li-Cor) respectively.

Proteinase K treatments and western blot
For PK resistance assays, equal volumes of solubilized homogenates, Sarkospin supernatants or pellets fractions were treated or not with 1 µg. ml −1 Proteinase K (Sigma) from 0 to 60 min at 37°C. At the end of the indicated time, samples were added Laemmli 1x prior to denaturation at 95°C for 5 min, and loaded on Mini-Protean TGX 12% gels (Biorad) followed by SDS-PAGE electrophoresis. Gels were whether stained for total protein amount (silver stain, Biorad) or transferred on nitrocellulose 0.2 µm membranes with Trans-Blot Turbo transfer system (Biorad) using the Mixed molecular weight program. Membranes were fixed with PFA, and proteins were immunolabelled as described for filter trap.

Mass spectrometry analysis of Sarkospin pellets
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD024998.

Sample preparation and protein digestion
Protein samples were solubilized in Laëmmli buffer and 10 µg per sample were deposited onto SDS-PAGE gel for concentration and cleaning purpose. Separation was stopped once proteins have entered resolving gel. After colloidal blue staining, bands were cut out from the SDS-PAGE gel and subsequently cut in 1 mm × 1 mm gel pieces. Gel pieces were destained in 25 mM ammonium bicarbonate 50% acetonitrile (ACN), rinsed twice in ultrapure water, and shrunk in ACN for 10 min. After ACN removal,

Database search and results processing
Data were searched by SEQUEST through Proteome Discoverer 2.3 (Thermo Fisher Scientific Inc.) against the Homo sapiens Reference Proteome Set (from Uniprot 2019-03; 73,645 entries). Spectra from peptides higher than 5000 Da or lower than 350 Da were rejected. The search parameters were as follows: mass accuracy of the monoisotopic peptide precursor and peptide fragments was set to 10 ppm and 0.6 Da respectively. Only b-and y-ions were considered for mass calculation. Oxidation of methionines (+16 Da) and protein N-terminal Acetylation (+42 Da) were considered as variable modifications and carbamidomethylation of cysteines (+57 Da) as fixed modification. Two missed trypsin cleavages were allowed. Peptide validation was performed using Percolator algorithm 41 and only "high confidence" peptides were retained corresponding to a 1% False Positive Rate at peptide level. Peaks were detected and integrated using the Minora algorithm embedded in Proteome Discoverer. Proteins were quantified based on unique peptides intensities. Normalization was performed based on total protein amount. Protein ratio were based on protein abundances calculated as the sum of the three most intense peptides. Two-ways ANOVA tests were calculated considering the three values of protein abundance in each comparison. Quantitative data were considered for proteins quantified by a minimum of two peptides, fold changes above 1.5 and a statistical p-value lower than 0.05.

Gene ontology analysis
Gene ontology analysis was performed online using Panther Geneontology website: http://pantherdb.org. Lists of Uniprot accession IDs of the different gated proteins (PD and MSA groups) were used as sample Homo sapiens list, and were searched for statistical overrepresentation test against the list of all proteins identified in the present MS study, with biological process, molecular function, and cellular components annotations. Results of multiple comparison corrected t-tests gave the indicated fold enrichment, number of proteins, and p value for the overrepresentation of each gene ontology cluster identified. Z scores were calculated as Z = fold change × (−log 10 (p value)).

Antibodies used in this study
Antibodies used in this study and their respective information are detailed in Table 2.