Brain region-specific susceptibility of Lewy body pathology in synucleinopathies is governed by α-synuclein conformations

The experimental use of human brain samples was approved under the protocol number RA032611/1 by the University College London and the REC reference 18/LO/0721 (‘Queen Square Brain Bank for Neurological Disorders’).

Single-mutation αS expression plasmids were generated from the pcDNA4 or pcDNA3/aS plasmid9 using the QuikChange II site-directed mutagenesis kit and appropriate primers.

Cells were cultured at 37 °C in 5% CO₂. Human BE(2)-M17 neuroblastoma cells (called M17D, ATCC number CRL-2267) were cultured in DMEM/F12 GlutaMAX (Thermo Fisher Scientific), 10% fetal bovine serum (Sigma) and 1% penicillin/streptomycin. Prior to the transfection, cells were seeded at a density of 1.5 × 106 cells/6 cm dish. HEK cells were cultured in DMEM with 10% fetal bovine serum (Thermo Fisher Scientific) and 1% penicillin/streptomycin and l-glutamine (2 mM, Gibco). Transfections with αS wild type and fPD mutants (A30P, E46K, G51D, A53T, H50Q) were carried out using Lipofectamine 2000 according to the manufacturer’s instructions. Cells were harvested 48 h after transfection, pelleted, snap frozen and further processed by cross-linking. HEK293 αS C-term Strep II-tagged lysates were purchased from GeneScript.

Frozen brain or cell pellet samples were manually homogenized in 1 × PBS/protease inhibitor/phosphatase inhibitor (Sigma, Thermo Fisher Scientific). Tissue suspensions were lysed by sonication (Sonic Dismembrator model 300, setting 40 for 15 s at 4 °C or Fisher Scientific Model 705 Sonic Dismembrator, sonication of amplitude 5% for 15 s at 4 °C). Samples were ultracentrifuged at 100,000g for 1 h at 4 °C. The resultant supernatant contained soluble proteins. To separate membrane proteins from insoluble proteins, the pellet from the homogenized brain tissue was resuspended in 1 ml OG-RIPA buffer (0.5% Nonidet P-40 substitute (NP-40, Pan Reac), 0.5% sodium deoxycholate (Sigma), 0.1% sodium dodecyl sulfate (SDS, Sigma), and 10 mM calcium chloride dihydrate (Sigma) with 2% n-octyl-ß-d-glucoside (abcam) and centrifuged at 175,000g for 30 min at 4 °C. The supernatant containing membrane-associated proteins was collected. The remaining pellet containing insoluble proteins was resuspended in 200 µl 8 M urea (Sigma)/5% SDS (Sigma)/PBS (Sigma), sonicated for 30 s (5 s on, 5 s off setting, amplitude 20%, room temperature (RT), Fisher Scientific Model 705 Sonic Dismembrator), and boiled for 10 min at 100 °C. Protein concentrations were measured with the Pierce BCA protein assay kit (Thermo Fisher Scientific) according to the manufacturer’s instructions.

Freshly collected and washed erythrocytes were resuspended in threefold volume of ACK lysing buffer (Lonza, Walkersville MD, USA). (NH₄)₂SO₄ to a final concentration of 25% was added and incubated at 4 °C for 30 min. The lysate was centrifuged (20,000g, 20 min), and the supernatant brought up to 50% (NH₄)₂SO₄. The pellet was washed several times in 55% (NH₄)₂SO₄ to remove excess hemoglobin. The sample was centrifuged at 20,000g for 20 min and the pellet resolubilized in 50-fold volume of 50 mM phosphate buffer, pH 7.0, 1 M (NH₄)₂SO₄. Five millilitre of the resultant solution was injected onto a 5 ml HiTrap phenyl hydrophobic interaction column (GE Healthcare) equilibrated with 50 mM phosphate buffer, pH 7.0, 1 M (NH₄)₂SO₄. αS was eluted with a 1–0 M (NH₄)₂SO₄ gradient in 50 mM phosphate buffer, pH 7.0 (αS eluted at ~ 0.75 M (NH₄)₂SO₄).

Twenty microgram of total protein in a total volume of 25 µl was incubated with 10 µl 5 mM DSG (Thermo Fisher Scientific, final concentration 1.43 mM, DSG was first dissolved in 50 µl of anhydrous DMSO and further in PBS pH 7.4 to a final volume of 1 ml) for 30 min at 37 °C shaking or rotating. The reaction was quenched with 3.5 µl 1 M Tris–HCl, pH 7.6 (Sigma), for 5 min shaking at RT. Samples were analyzed in biological duplicates and technical triplicates. The cross-linker disuccinimidyl suberate (DSS, Thermo Scientific Pierce) was used for validation.

Antibodies used were 2F12 to αS[12] (MABN1817, Merck, WB 1:5000), SOY1 to αS (MABN1818, Merck), and anti-DJ-1 [5] (or GeneTex, WB 1:2000). Anti-Tau (EP2456Y, abcam, WB 1:5000), anti-β tubulin III (T2200, Sigma-Aldrich, WB 1:5000), anti-heat shock protein 70 (HSP70, ab181606, abcam, WB 1:1000), anti-14-3-3 beta (ab15260, abcam, WB 1:500), anti-β actin (ab8227, Abcam, WB 1:5000), and anti-HSP90 (HEK: ab clone 16F1, ADI-SPA-83, Enzo, WB 1:1000, huBrain: ab PA3-013, Invitrogen, WB 1:500).

Multi-array 96 HB plates (MSD) were coated with the capture antibody anti-αS 2F12 (200 ng 2F12[12] in PBS, 30 µl/well) and incubated over night at 4 °C. The next day, the remaining liquid was removed, and the plates were blocked with 5% Blocker A (MSD) in PBS-Tween 0.1% (150 µl/well, Sigma) shaking for 2 h at RT. Plates were washed 5× with PBS-Tween 0.1% (150 µl/well). Standards (recombinant αS, highest concentration 1 ng, ratio for serial dilution 1:4, 30 µl/well) and protein (30 µl/well, soluble protein fraction 1:500, membrane (OG-RIPA) and insoluble (UREA/SDS fraction 1:100) were diluted in 1% Blocker A in PBS-Tween 0.1% and the plates were incubated shaking for 2 h at RT. The remaining liquid was removed and the plates were washed 5× with PBS-Tween 0.1% (150 µl/well). The sulfo-tagged detection antibody SOY1 (50 ng SOY1 in 1% Blocker A (MSD)/PBS-Tween 0.1%, 30 µl/well) was added and the plate incubated shaking for 1 h at RT in the dark. The remaining liquid was removed, and the plates were washed 5× with PBS-Tween 0.1% (150 µl/well). 2× read buffer (MSD)/MilliQ water was added (150 µl/well) and the plates were immediately measured using an MSD Sector 2400 imager or MSD Model No. 1300, QuickPlex SQ 120 according to the manufacturer’s instructions. All samples and standards were analyzed in technical duplicates.

Immunoprecipitated αS fractions were injected on a Superdex 200 (10/300 GL) column (GE Healthcare) at room temperature and eluted with 50 mM NH4Ac (pH 7.4) while measuring (in-line) the conductivity and the 280-nm absorption of the eluate at 0.7 ml/min. For size estimation, a gel filtration standard (Bio-Rad) was run on the column, and the calibration curve was obtained by semilogarithmic plotting of molecular weight versus the elution volume divided by the void volume. Immunoblotting was performed to identify the fractions containing αS multimers and monomers.

After SEC separation, samples containing αS multimers or monomers were exchanged into 10 mM ammonium acetate using Zeba spin desalting columns (Thermo Fisher), lyophilized, and resuspended in 10 mM ammonium acetate at a concentration of approximately 10 μM. αS samples were added to a 1 mM path length quartz cuvette for far-UV CD and analyzed using a J-1500 CD spectrometer (Jasco) at 25 °C. Temperature control with an accuracy of 0.1 °C was achieved with a heating/cooling accessory equipped with a Peltier element (PFD-425S) connected to a water thermostatic bath. Buffer spectra were recorded and subtracted.

Immunoprecipitated and SEC-separated αS multimers and monomers were loaded at a flow rate of 0.15 ml/min onto a Superdex 200 3.2/300 GL Increase column (GE Healthcare) previously equilibrated in 50 mM ammonium acetate, pH 7.4. The column was connected in line to a Dawn 8 + MALS detector (Wyatt Technology) using a laser emitting at 690 nM and by refractive index measurement using an Optilab T-rex (Wyatt Technology Corp.).

Five milligram per millilitre recombinant αS in PBS was aggregated for 3 days at 37 °C with nutation to form thioflavin T-positive fibrils. To generate soluble PFFs, αS fibrils were diluted to 1 mg/ml and sonicated at power level 50 for 3 × 10 s using a Sonic Dismembrator model 300 (Fisher). Aliquots of the resultant material were flash-frozen in liquid nitrogen and stored at − 80 °C.

All reagents were purchased from Thermo Fisher Scientific unless otherwise noted. Transfected M17D were incubated with 0.5 µg/ml αS PFF. After 48 h, the medium was aspirated, and cells were harvested by scraping and washed 2× in cold PBS (500g spin for 5 min). Cell pellets were resuspended in 1× PBS/protease inhibitor and sonicated with a Sonic Dismembrator model 300 (microtip setting 40; 2 × 15 s). Cells were spun at 20,000g for 10 min. The supernatant was kept (“cytosol”) and the pellet was incubated in 1% Triton X-100 at 4 °C for 30 min under nutation. After a 20,000g spin for 10 min, the supernatant was kept (“membrane fraction”) and the pellet dissolved in 5% SDS at 100 °C for 10 min to give the “insoluble αS” fraction. Relative αS content in each fraction was analyzed by SDS-PAGE/Western blot and ELISA. For the SCDi experiments, M17D/αS-WT::YFP[30] cells were monitored on an IncuCyte Zoom machine (Essen Bioscience). Cells were seeded at a density of 0.1*10°6 cells/ml in a 12-well plate. Twenty-four hours after seeding, a fraction of the cells was treated with 10 µM SCDi or DMSO. Twenty-four hours after the treatment, some cells were seeded with 0.5 µg/ml αS^F. Another 48 h and 96 h after seeding, SCDi or DMSO was added again. The experiments were carried out in biological duplicates. The IncuCyte settings and analysis were performed analogous to Imberdis et al. [30]. For the seeding experiment, M17D human neuroblastoma cells constitutively expressing wt αS were analyzed in biological and technical duplicates, and cross-linking was performed analogous to brain samples. Cells were seeded at a density of 0.1*10°6 cells/ml in a 12-well plate. Twenty-four hours after seeding, some cells were seeded with 0.5 µg/ml αS^F.

To detect amyloid fibril growth, a discontinuous assay was used. Aliquots (10 μl) were removed from each purified αS sample (lyophilized from 50 mM ammonium acetate, pH 7.4, and agitated at 37 °C at a concentration of 75 μM in 20 mM Bis–Tris propane, 100 mM LiCl, pH 7.4) and added to 2 ml of a 10 μM thioflavin T (ThT) solution in 10 mM glycine buffer, pH 9. Fluorescence was directly quantified on a Varian Eclipse fluorescence spectrophotometer at 20 °C by exciting at 444 nM and scanning the emission wavelengths from 460 to 550 nM with slit widths set at 5 nM (PMT at 750 V).

HEK293 αS C-term Strep II-tagged lysates were cross-linked and immunoprecipitated as described above and the αS^H and αS^U were separated via SEC and run on an SDS-gel. Gel pieces containing the bands of interest were cut out and lyophilized prior to mass spectrometry analysis. Gel slices were subjected to gel digestion. In brief, slices were washed consecutively with water, 50% acetonitrile (ACN), and 100% ACN. Proteins were reduced with 20 mM DTT in 50 mM ammonium bicarbonate and alkylated with 40 mM acrylamide (in 50 mM bicarbonate). The slices were washed again and dehydrated with ACN. Proteolysis was performed with 330 ng chymotrypsin or trypsin (sequencing grade Promega, Mannheim, Germany) at 37 °C overnight. The peptide extracts were dried in a vacuum concentrator and stored at − 20 °C. Dried peptides were dissolved in 10 µl 0.1% formic acid (solvent A). Two microlitres was injected onto a C18 analytical column (self-packed 300 mM length, 100 µM inner diameter, ReproSil-Pur 120 C18-AQ, 3 µM, Dr. Maisch GmbH, Ammerbuch-Entringen, Germany). Peptides were separated during a linear gradient from 2 to 35% solvent B (90% acetonitrile, 0.1% formic acid) within 90 min at a flow rate of 300 nl/min. The nanoHPLC was coupled online to an Orbitrap Lumos mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). Peptide ions between 330 and 1600 m/z were scanned in the Orbitrap detector with a resolution of 60,000 (maximum injection time 50 ms, AGC target 400,000). Precursor ions (threshold 25,000) were subjected to higher energy collision-induced dissociation within a 2.5 s cycle and fragments were analyzed in the Orbitrap detector (maximum injection time 22 ms, resolution = 15,000). Fragmented peptide ions were excluded from repeat analysis for 15 s.

Lewy body pathology density was assessed semiquantitatively on formalin-fixed paraffin-embedded tissues at 20× magnification (Leica DM3000 microscope), using revised McKeith staging scheme of αS pathology [42], with the stages ranging from 0.5 to 4. Stage 0.5 was assigned if a single Lewy body was seen in a region of interest after screening several nearby fields. Stage 1 was assigned if a single Lewy body was consistently seen per each field after screening several nearby fields, and Stage 1.5 if a single Lewy body was seen per field with rare additional two Lewy bodies per single nearby field. Stage 2 was assigned if two to three Lewy bodies were seen per field, and Stage 2.5, if twio to three Lewy bodies were seen per field and rare additional four Lewy bodies per nearby field. Stage 3 was assigned if 4–10 Lewy bodies were consistently seen per field and Stage 3.5, if 11–20 Lewy bodies were seen per field. Stage 4 was assigned if more than 20 Lewy bodies were seen per field. The following anatomical regions were assessed: amygdala, transentorhinal cortex, peri- and parastriate cortex, and cortical regions of the insula, anterior cingulate gyrus, middle temporal gyrus, Heschl’s gyrus, anterior middle frontal gyrus, and inferior parietal lobule.

Western Blot: the ratio of protein multimers to monomers was analyzed using the Image Studio software western analysis according to the manufacturer’s instructions (background subtraction: median, border with 3, top/bottom). Data analysis (except for correlation and regression analysis) was performed using GraphPad Prism 7 (GraphPad Software, La Jolla, CA, USA). Statistical significance was determined by Mann–Whitney test (p < 0.05, two-tailed). Samples are displayed as mean ± SD. Linear regression analysis was performed using R. Additional R software packages included “devtools” (https://​github.​com/​r-lib/​devtools), “easyGgplot2” (https://​github.​com/​kassambara/​easyGgplot2) and “Hmisc” (https://​github.​com/​harrelfe/​Hmisc). Samples were coded as follows. Gender: male = 1, female = 2. McKeith: neocortical = 3, limbic = 2, brainstem = 1, controls (no McKeith staging) were set to 0. CERAD: none = 0, sparse = 1, mild = 2, moderate = 3, frequent = 4. Braak LB staging: controls were set to 0, PD and DLB patients were coded according to evaluated Braak LB staging. ELISA: raw data from the MSD Discovery Workbench was imported into Excel 2010 (Microsoft) for background calculation and subtraction. Values from Excel were imported into GraphPad Prism 7 (GraphPad Software, La Jolla, CA, USA) for interpolation together with the known standard concentrations (standard curve: sigmoidal, 4PL, X is log (concentration, no special handling for outliers). The final concentration of each analyzed sample was calculated in Excel according to the dilution factor. Mass spectrometry: raw data processing and analysis of database searches were performed with Proteome Discoverer software 2.4.0.305 (Thermo Fisher Scientific). The protein-specific peptide identification was done with an in-house Mascot server version 2.6.1 (Matrix Science Ltd, London, UK) from Proteome Discoverer. MS2 data were searched against a database of common contaminants (cRAP of the Global Proteome Machine), and human SwissProt sequences. Precursor ion m/z tolerance was 10 ppm, and fragment ion tolerance 0.02 Da. Tryptic peptides with up to two missed cleavages were searched, propionamide was set as a static modification of cysteines, and oxidation as a dynamic modification of methionine. Mascot results from searches against SwissProt were sent to the Percolator algorithm_ENREF_3 version 3.02 as implemented in Proteome Discoverer 2.4. If the data did not permit use of Percolator, PSMs were evaluated by a target-decoy procedure. Spectra without high confidence (FDR 1%) matches were sent to a second round Mascot search with semi-specific enzyme cleavage. The false discovery rate of proteins in these samples was 0.6%.

To address the question of whether the equilibrium of αS^U and αS^H also plays an important role in sporadic synucleinopathies, we adapted our established [4, 12, 29] multimer assay (using cross-linking and Western blot analysis of lysate) for frozen human post-mortem tissue (Online Resource Figs. 1, 2, 3, 4, 5). This protocol revealed a prominent cytosolic ~ 80 kDa and 14 kDa αS species (Fig. 1a; Online Resource Figs. 2, 3, 4, 5) as described previously [12, 29, 38]. The two cytosolic αS species corresponded to a 86 kDa helically folded homo-multimer and a 14 kDa monomer, respectively, as confirmed by circular dichroism (CD) spectroscopy, Multi-angle light scattering and Western blotting were conducted on purified 14 kDa and 86 kDa species from human brain tissue and HEK cells (Fig. 1a, b, f; Online Resource Figs. 6, 7). Mass spectrometry and Western Blot analysis of the 86 kDa species showed only the presence of αS compared to controls, indicating its homo-multimeric nature (Online Resource Table 1, Online Resource Figs. 8, 9, 10, 11). We confirmed that the cross-linking procedure did not lead to an artificial production of αS^H using recombinant E. coli-derived protein in contrast to brain tissue (which led only to cross-linked αS^U), HEK cells or red blood cells (yielding cross-linked αS^H) and non-denaturing purification from both human tissue and bacterial cells (leading to the same species obtained under cross-linking conditions, Fig. 1c–e). Figure 1 shows the biophysical analysis of physiological αS found in the cytosol. Insoluble or amyloid-type aggregated forms of pathological αS are not found in the cytosol (they represent typically only ~ 0.5% of total protein in this subcellular compartment) as we have shown before in human brain [54]) and are therefore not displayed.

Next, we assessed the ratio of cytosolic helical and unfolded αS (αS^H/αS^U) in four different brain regions (entorhinal cortex, cingulate cortex, frontal cortex and striatum) of 28 age-matched individuals, classified as neurological controls, 15 DLB patients and 15 sporadic PD patients (Table 1). Representative pictures of Western blot analyses from controls and DLB and PD patients are displayed in Fig. 2a and Online Resource Fig. 12.

We found that DLB patients exhibited a significant reduction of αS^H/αS^U in the frontal cortex (p < 0.001, Fig. 2b) and cingulate cortex (p = 0.004, Fig. 2c). PD patients exhibited a significant decrease of αS^H/αS^U in the frontal cortex (p = 0.001, Fig. 2b). We did not observe significant changes between patients and controls in the entorhinal cortex, a region typically affected by Lewy body (LB) pathology earlier in the disease course than neocortical regions (Online Resource Fig. 13a), the striatum which served as a control region (due to low LB burden, Online Resource Fig. 13b) or the internal control protein DJ1 (Online Resource Fig. 13c–f). Analysis of variance with regard to clinical and neuropathological features (summary Table 1) revealed a significant difference between the reduction of the αS^H/αS^U and increased αS Braak LB staging in the frontal cortex (p < 0.001) and cingulate cortex (p = 0.02, Fig. 2d, e). A similar difference was observed between increased McKeith staging and reduced αS^H/αS^U ratios (frontal cortex p < 0.001; cingulate cortex p = 0.02, Online Resource Fig. 14). A significant difference was also detected between decreased αS^H/αS^U ratios and increased Braak neurofibrillary tangle (NFT) staging in the entorhinal cortex (p = 0.04, Online Resource Fig. 14). All other clinical features such as gender, age or post-mortem interval (Table 1) were not associated with the reduction of αS^H/αS^U (data not shown). Hence, these data indicate a region-specific destabilization of physiological αS in brain regions affected by LB pathology.

Given the apparent disease and region specificity of the αS^H destabilization, and the discussed “prion-like” aggregation transmission of pathological αS in a highly region-specific manner, we analyzed whether the αS^H form of αS, being resistant to spontaneous aggregation, can also protect against the self-templating “prion-like” aggregation mediated by αS^F. Therefore, we performed a thioflavin T-based aggregation assay on purified αS and separated αS^H multimers and αS^U monomers. As the cross-linker interfered with ThT assays (Online Resource Fig. 15a), we prepared native, non-cross-linked αS^H from human blood and αS^U from E. coli, given the difficulty of native αS purification from brain [38], and assessed its susceptibility toward “prion-like” aggregation. Our ThT-aggregation assay showed that αS^H—in contrast to monomeric αS^U—did not form ThT-bound fibrils (Fig. 3a). The addition of recombinant “prion-like” αS^F did stimulate the aggregation in the αS^U monomer, but not in the αS^H multimer, hence indicating that the cytosolic helical structure found in human brain is not susceptible to putative “prion-like” aggregation transmission of amyloid protein species (Fig. 3a). Furthermore, addition of αS^F did not lead to destabilization of purified αS^H. In addition, we analyzed M17D cells with high αS^H/αS^U ratios and found that the addition of αS^F to wt cells did not result in a significant decrease of αS^H/αS^U ratios (Online Resource Fig. 15b), indicating that while stabilization of αS^H protects from detrimental effects, αS^F (amyloid itself) is not causal for destabilization of αS^H. Taken together, these data suggest that a decrease in αS^H/αS^U ratios potentially leads to a raised susceptibility in neurons for “prion-like” transmission of αS amyloid. To test this, we stimulated “prion-like” aggregation in human neuronal cells (M17D) expressing αS constructs with varying αS^H/αS^U ratios (wt or previously described αS^H destabilizing fPD mutations) by addition of preformed αS^F to the cell media. We then assessed the amount of (Triton-X) insoluble αS found in the cell pellets by αS ELISA after 2 days of seeding. While no differences in forming insoluble αS were found in cells expressing fPD mutations without treatment (Fig. 3b), the addition of αS^F to the growth medium led to a significant increase of aggregated αS in the neuronal cell line across all fPD mutations, indicating an increased “prion-like” aggregation susceptibility compared to wild type (wt, Fig. 3c). This directly indicates that all fPD mutations lead to greater susceptibility toward “prion-like” aggregation in neuronal cells, while their effect on spontaneous aggregation is comparatively small. In accordance with previous published data [14], we detected a decrease of αS^H/αS^U ratio in the lysate of M17D cells expressing different familial PD SNCA mutations (A30P, G51D, A53T, E46K and H50Q, respectively) compared to wt (Fig. 3d). The decrease of αS^H/αS^U ratios caused by these mutations correlated significantly with the susceptibility toward “prion-like” aggregation as measured by insoluble αS in αS^F-seeded M17D cells (p = 0.002, Fig. 3d). In contrast, M17D cells with higher αS^H/αS^U ratios exhibited decreased seeding and aggregation capabilities, indicating that increased αS^H stabilization in neuronal cells is protective against “prion-like” propagation (Fig. 3d). To establish a causal link between αS^H stability and “prion-like” aggregation resistance, we next performed rescue experiments with a stearoyl-coenzyme A desaturase inhibitors (SCDi), a known pharmacological stabilizer of αS multimers [19, 30, 45] leading to an increase in the relative amount of protective αS^H in our M17D cells (Fig. 3e). We then seeded M17D cells expressing YFP-tagged wt αS and followed the αS inclusion formation under αS^F addition via automated confocal microscopy (IncuCuyte). Here, the multimer-stabilizer SCDi prevents the formation of pathological αS inclusions even in the presence of αS^F (Fig. 3f), indicating a rescue of “prion-like” susceptibility through the stabilization of physiologically αS^H. This protective stabilization even extended to the spontaneous aggregation seen in wt αS expressing neuronal cells without αS^F addition (Fig. 3f).

To validate the association between the region-specific distribution and the proposed “prion-like” aggregation transmission theory of αS, we carried out an analysis of nine brain regions from 10 controls, 6 DLB and 13 sporadic PD patients (Table 2). All brain regions for each individual were analyzed in biological and technical duplicates (n = 1044 single samples). Compared to the analysis of the four brain regions, which used PD patients mostly at their extremes (Braak 6), we were able to include selected PD patients with Braak stages 4 and 5 and additional clinical information such as the cognitive status or Thal phases (Table 2).

The brain regions selected for the analysis reflect the typical temporal development of LB pathology across the limbic and neocortical regions, with early affected areas [amygdala, cortex of parahippocampal gyrus (PHG) and anterior cingulate cortex (ACC)] in Braak stage 4, later affected areas [cortex of insula, middle temporal (MTG) and anterior middle frontal gyri (AMFG)] corresponding to Braak stage 5 and first-order sensory association area [inferior parietal lobule (IPL)] in Braak stage 6. Also, the cortex of the transverse temporal gyrus (Heschl’s gyrus) and occipital cortex (OC, not collected from primary visual cortex), which are regions typically spared from LB pathology or affected late in disease course, were included in the analysis (Fig. 4). We investigated the changes across all nine brain regions and used the slope or trendline calculated for the regions for further comparisons. Interestingly, in this extended brain region analysis, PD patients had significantly higher slopes across the nine brain regions compared to controls (Fig. 4a, p = 0.006). The slopes dropped in PD patients with dementia and further declined in DLB patients which were all reported to be demented compared to PD patients without dementia (Fig. 4, p = 0.003). It became obvious that in controls, αS^H/αS^U ratios in the later affected brain areas, according to the classical Braak staging, were not higher compared to PD samples per se (Fig. 4b, d–f). The important changes worked out in the analysis of the nine brain regions is as follows: five of six DLB patients exhibited decreased slopes, meaning lower αS^H/αS^U ratios, in the later affected brain regions according to the classical Braak staging (Fig. 4c). PD patients Braak 4, Braak 5 and PD patients Braak 6 without dementia had increasing slopes, thus higher αS^H/αS^U ratios, especially in the IPL, Heschl’s gyrus and OC (Fig. 4d–f). Interestingly, some PD Braak 6 patients, those with dementia, also showed decreased slopes across the nine brain regions similar to DLB patients resulting in low αS^H/αS^U ratios in the IPL, Heschl’s gyrus and OC (Fig. 4f). We performed a linear regression analysis with all available clinical variables (Table 2) and could support those findings by identifying dementia as the only significant predictor of the outcome variable, the αS^H/αS^U ratios (p = 0.03, R² = 0.3). Overall, in PD patients Braak 4 and 5 with and without dementia, the slopes across the nine brain regions remained positive with higher αS^H/αS^U ratios in the later affected brain regions (Fig. 4d–e). In PD and DLB patients Braak 6, the slopes clearly separated those patients with (decreased slopes) and without dementia (increased slopes, Fig. 4c, f). Thus, the disequilibrium of αS^H/αS^U ratios in the cortical regions across all nine brain regions seems to be critical for PD and DLB patients. Thus, we assume that the individual equilibrium in PD and DLB patients and especially the cortical αS^H/αS^U ratios are disturbed during the disease. These findings were further supported by applying a semiquantitative LB score across the nine brain regions showing that with increasing Braak stages in PD patients and Braak 6 DLB patients, the LB score increases in the later affected brain regions, especially insula, IPL, Heschl’s gyrus and OC (Fig. 4g). The increased means of the semiquantitative LB score comparing controls, PD and DLB patients was significantly different from the decreasing mean αS^H/αS^U slopes compared across all groups (p = 0.02, Fig. 4h). Thus, when the disease progresses and LB scores raise, αS^H/αS^U ratios decline, especially in demented individuals. These findings further support the region-specific distribution and the proposed “prion-like” aggregation transmission theory of αS.

In summary, when the equilibrium of αS^H/αS^U shifts toward the aggregation-prone αS^U in PD and DLB patients, our data demonstrate that the likelihood of fibril formation, and subsequently LB inclusions, increases. Our current findings provide a novel mechanism, in which the equilibrium of physiological aggregation-resistant αS multimers and physiological, aggregation-prone αS monomers is disturbed in sporadic PD and DLB patients, governing the regions affected in the brain and clinical outcome (Fig. 5). Local and individual changes might modify the differential caudo-rostral progression in these synucleinopathies, if the pathology originates in the brain stem as currently described (Fig. 5). The data imply that brain stem pathology would either arrive in neuronal tissue with stepwise increase via midbrain toward the cortices, leading to a slow progression and late involvement of the cortex (PD pattern), or a fast neuropathological progression from brain stem areas to cortical brain regions in DLB patients.