Background: Aggregation of α-synuclein (α-syn) is a prominent feature of Parkinson's disease (PD) and other synucleinopathies. Currently, α-syn seed amplification assays (SAAs) using cerebrospinal fluid (CSF) represent the most promising diagnostic tools for synucleinopathies. However, CSF itself contains several compounds that can modulate the aggregation of α-syn in a patient-dependent manner, potentially undermining unoptimized α-syn SAAs and preventing seed quantification. Methods: In this study, we characterized the inhibitory effect of CSF milieu on detection of α-syn aggregates by means of CSF fractionation, mass spectrometry, immunoassays, transmission electron microscopy, solution nuclear magnetic resonance spectroscopy, a highly accurate and standardized diagnostic SAA, and different in vitro aggregation conditions to evaluate spontaneous aggregation of α-syn. Results: We found the high-molecular weight fraction of CSF (> 100,000 Da) to be highly inhibitory on α-syn aggregation and identified lipoproteins to be the main drivers of this effect. Direct interaction between lipoproteins and monomeric α-syn was not detected by solution nuclear magnetic resonance spectroscopy, on the other hand we observed lipoprotein-α-syn complexes by transmission electron microscopy. These observations are compatible with hypothesizing an interaction between lipoproteins and oligomeric/proto-fibrillary α-syn intermediates. We observed significantly slower amplification of α-syn seeds in PD CSF when lipoproteins were added to the reaction mix of diagnostic SAA. Additionally, we observed a decreased inhibition capacity of CSF on α-syn aggregation after immunodepleting ApoA1 and ApoE. Finally, we observed that CSF ApoA1 and ApoE levels significantly correlated with SAA kinetic parameters in n = 31 SAA-negative control CSF samples spiked with preformed α-syn aggregates. Conclusions: Our results describe a novel interaction between lipoproteins and α-syn aggregates that inhibits the formation of α-syn fibrils and could have relevant implications. Indeed, the donor-specific inhibition of CSF on α-syn aggregation explains the lack of quantitative results from analysis of SAA-derived kinetic parameters to date. Furthermore, our data show that lipoproteins are the main inhibitory components of CSF, suggesting that lipoprotein concentration measurements could be incorporated into data analysis models to eliminate the confounding effects of CSF milieu on α-syn quantification efforts.
Keywords: α-synuclein; Lipoproteins; Cerebrospinal fluid; Seed amplification assays; RT-QuIC
Supplementary Information The online version contains supplementary material available at https://doi.org/10.1186/s13024-023-00613-8.
Parkinson's disease (PD), dementia with Lewy bodies (DLB) and multiple system atrophy (MSA) are neurodegenerative diseases pathologically characterized by the presence of intracellular α-syn inclusions in vulnerable brain regions and are commonly referred to as synucleinopathies. Seed amplification assays (SAAs), known as protein misfolding cyclic amplification (PMCA) [[
In this work, we comprehensively characterized the inhibitory effect of CSF on α-syn aggregation: we first identified a high-molecular weight (HMW) fraction of CSF exerting most of the inhibitory effect. Subsequently, we selected putative inhibitors based on their relative abundance in the CSF HMW fraction and specifically analysed their interactions with α-syn, their effect on α-syn aggregation, and possible impact on α-syn SAAs.
In the current work we thoroughly characterized the inhibitory effect of human CSF on α-syn aggregation by means of several complementary techniques. The study design of the present work is summarised in Fig. 1. At first, we collected more evidence supporting the fact that CSF is naturally capable of inhibiting α-syn aggregation in both SAA reaction mix and phosphate buffered saline (PBS), both in seeded and unseeded conditions, and in a patient-dependent manner (Fig. 1A.1). Starting from the observation that two CSF samples collected from normal-pressure hydrocephalus (NPH) patients with marked difference in proteins content produced different α-syn ThT fluorescence profiles in PBS (Fig. 1A.2), we performed fractionation of a pool of CSF collected from neurological controls by mean of centrifugal filters (Fig. 1B.1). We then analysed different fractions by means of mass spectrometry and Protein aggregation assays (Fig. 1B.2–3). We determined that the fraction corresponding to molecular weight above 100 kDa, rich in apolipoproteins, albumin and transthyretin (TTR), retained most of the inhibitory effect of neat CSF. By means of Protein aggregation assays, Western blot (WB), dot blot (DB), transmission electron microscopy (TEM), and solution nuclear magnetic resonance (NMR) spectroscopy we determined that high-density and low-density lipoproteins (HDL and LDL) are highly inhibitory against α-syn aggregation and likely interact with α-syn oligomeric species (Fig. 1.C.1–3). Subsequently, we tested the impact of varying (within physiological ranges) concentrations of albumin, TTR, HDL and LDL on ultrasensitive diagnostic SAA (Fig. 1C.4). We then used immunoprecipitation (IP), to deplete the two most abundant CSF apolipoproteins (ApoA1 and ApoE) from a newly made CSF pool and tested the effect with Protein aggregation assays (Fig. 1D.1). Finally, we measured total protein content, ApoA1 and ApoE concentrations in SAA-negative CSF samples of a small cohort of neurological controls and repeated SAA on the same samples spiked with preformed aggregates to evaluate possible correlation between SAA kinetic parameters and CSF total protein content, ApoA1, and ApoE concentrations (Fig. 1D.2).
Escherichia coli BL21 (DE3) Gold were transformed with a pT7-7 vector cloned with the gene encoding α-syn. The overnight preculture of transformed cells was diluted 100-fold in LB medium and induced at an OD
The suspension was then ultra-centrifuged at 20 000 rpm (Type 70 Ti rotor, Beckman Coulter) for 25 min, and the pellet was collected and resuspended with 90 mL precooled ultrapure water containing 38 μL of 1 M MgCl
For
For TTR, Escherichia coli BL21(DE3) RIPL PLysS cells were transformed with pET-28a(+) plasmid encoding TTR gene. The cells were cultured in LB Medium containing 0.1 mg/mL of Kanamycin, grown at 37 °C, until OD
The neurological control (NC) CSF samples used in this work had been previously collected and stored at -80 °C according to international guidelines [[
CSF from 19 different NC subjects (10 females and 9 males, average age = 70 y, standard deviation = 8 y) were pooled (CSF pool 1) reaching a total volume of 8 mL that was split in 2 aliquots of 3 mL and 10 aliquots of 0.2 mL.
A second CSF pool of 5 mL was prepared from different CSF samples collected form n = 10 NC subjects (5 females and 5 males, average age = 69 y, standard deviation = 5 y). This second pool was then split in 10 aliquots of 0.5 mL each, which were then used for ApoA1 and ApoE immunodepletion experiments and protein aggregation assays.
Two aliquots of 0.5 mL relative to 31 CSF samples collected from NC subjects (7 females and 24 males, average age = 69 y, standard deviation = 8 y), were selected for ApoA1 and ApoE ELISAs, total protein measurement, and SAA experiments.
CSF collected from healthy control (HC) and PD subjects was used to perform initial α-syn seed spiking experiments and to test the impact of adding HDL, LDL, HSA, and TTR at Amprion Inc. (San Diego, CA, U.S.). The HC1-6 samples used in α-syn seed spiking experiments shown in Fig. 2A have been purchased from Biochemed Services (Winchester, VA, U.S.) and were collected from 3 male and 3 female subjects (age = 26–39 y). With reference to the experiments summarised in Fig. 8, HC22 (female, age = 35 y) and HC24, HC66, HC67 (all males, age = 30–35 y) were also purchased from Biochemed Services. All neat HC CSF samples tested negative in α-syn SAA. With reference to the same experiments, PD10, PD34, PD62 CSF samples (all males, age = 72–79) belonged to PD patients with an Hoehn and Yahr (H&Y) stage of 2 and were purchased from PrecisionMed (Carlsbad, CA, U.S.). PD47 (PD patient, female, age = 62 y, H&Y = 1.5) was instead purchased from BioIVT (Westbury, NY, U.S.). All neat PD CSF samples tested positive in α-syn SAA.
An aliquot of 3 mL of CSF pool 1 was resuspended in 1.5 mL of PBS 3 × in order to have 4.5 mL of human pooled CSF in PBS 1x. This volume was then subjected to a series of filtrations using Amicon® Ultra-4 molecular weight cut-off (MWCO) filters. The procedure used to fractionate human CSF is schematized in Fig. 4A. The aliquots collected in this way contained the different constituents of the starting 4.5 mL of CSF in PBS with different concentration factors, the volume and the relative concentration factors (with respect to fraction 1) of the aliquots depicted in Fig. 4A are summarised in Table S1. Aliquots 2, 3, 4 and 5 were washed 3 times with PBS before storage. The ability to interact with α-syn monomers was then tested for all the CSF fractions. The different concentration factors were adjusted by diluting the samples with PBS (see Supplementary Material Table S1).
Preformed α-syn aggregates were generated by incubating 1 mg/mL of α-syn in PBS for one week at 37 °C under vigorous double orbital shaking (500 rpm) in a sealed 1.5 mL polypropylene vial. The final products were subjected to cycles of sonication (20 s tip sonication, 20 s rest) with an amplitude of 12 μm. The polypropylene vial had been immersed in ice for the whole duration of the sonication procedure. The aggregates were then diluted at 0.25, 2.5, 25, 250 and 2500 pg/μL, considering the initial monomer concentration as reference. The generated α-syn aggregates were then aliquoted and stored at -80 °C.
Samples were analysed as previously reported [[
The protein aggregation experiments used to characterize interaction between α-syn and CSF constituents were performed with a programmable BMG LABTECH ClarioStar® fluorometer in Greiner clear-bottom 96-well plates (cat# 655,906). The ThT fluorescence was read from the bottom using excitation and emission wavelengths of 450 and 480 nm, respectively. An incubation temperature of 37 °C was used for all the experiments. Slightly different gain values were used to avoid overflow of the analog-to-digital converter. In each experiment, lyophilized recombinant α-syn was thawed in 3 mM NaOH at the concentration of 3.5 mg/mL. The solution was brought to physiological pH by diluting it with concentrated PBS (4x) and distilled water. In all the experiments described, the final reaction volume was 200 μL, α-syn final concentration was 0.7 mg/mL and ThT final concentration was 10 μM. To avoid bacterial contamination, 0.1% NaN
158 μL of the solution containing monomeric α-syn was poured in wells, each of them containing 6 glass beads of 1 mm diameter. Depending on the experiment, 40 μL of human pooled/NPH CSF, 40 μL of PBS or 40 μL of CSF fractions were added. In seeded experiments, 2 μL of PBS containing 0, 0.25, 2.5, 25, 250, and 2500 pg/μL of α-syn aggregates were added. Plates were sealed and subjected to cycles of shaking (1 min double-orbital shaking at 500 rpm, 14 min rest) inside the fluorometer.
To compare the results with those of a previously published paper, the experiment was performed in the exact same way as it is described in Bellomo et al. [[
In these experiments 40 μL of solutions containing different dilutions (all the products were diluted with distilled H
During the ultrasensitive SAA experiments performed at Amprion Inc., fluorescence readings were collected every 30 min to estimate kinetic parameters with high accuracy. The following 4 parameter sigmoid was used to fit the raw fluorescence readings:
Graph
where F
Graph
where P
The average background fluorescence produced by three wells containing the analyte without α-syn was subtracted prior to the analysis from the ThT intensity profiles relative to the same analyte in the presence of α-syn. While analysing data relative to the sole α-syn the background fluorescence from well containing only the reaction buffer was subtracted. Each ThT kinetic trace was then fitted with a double sigmoid function using Origin Pro v9.0. In the fitting model, A2 fits the fluorescence value of the second plateau, A1 fits the fluorescence value of the first plateau and A0 fits the baseline fluorescence. The time parameters t1 and t2 fit the first and the second inflection points, respectively, while d1 and d2 represent the slopes of the sigmoids. In the non-linear fitting procedure used, the following bounds were applied: 0 < A0 < 1000, 500 < A1 < 5000, A2 > 2000, 0 < t1 < 100 h and t2 > 0. For some kinetic traces, a decrease in fluorescence was observed after reaching the second plateau. This known phenomenon is caused by the sequestration of ThT molecules by mature fibrils and by the sedimentation of HMW insoluble aggregates [[
Equal amounts of assay products (volumes containing 2 µg of α-syn) were added with Laemmli's sample buffer without sodium dodecyl sulphate (SDS), without boiling them to prevent solubilization of SDS-sensitive aggregates. Samples were separated through SDS-PAGE on 4–20% polyacrylamide gels (Bio-Rad) and transferred into PVDF membranes (0.45 μm, Bio-Rad) by wet transfer at 100 V constant for 90 min using 25 mM Tris–HCl, 192 mM glycine, 20% methanol, and 0.015% SDS. Membranes were fixed with 4% PFA for 30 min prior to blocking with 5% non-fat milk in TBS-T (TBS with 0.1% Tween 20) for 1 h at room temperature. After blocking, filters were incubated with primary antibody against α-syn (
For dot blotting aliquots of products obtained in SAAs corresponding to 300 ng of monomeric α-syn in the initial reaction mixtures were spotted (2 μL/spot) on nitrocellulose membrane pre-equilibrated with TBS-T. Samples were dried at RT and fixed with PFA (0.4% in PBS) for 30 min, and then filters were blocked with 2% non-fat milk (in TBS-T). The membranes were incubated overnight at 4 °C with OC (1:1000) or A11 (1:1000) conformational antibodies [[
Immunoprecipitation (IP) was performed to deplete ApoA1 and ApoE from CSF pool sample. 50 µL of settled Immobilized Protein A/G (100 µL resin slurry, Pierce™ Protein A/G Plus Agarose, ThermoFisher Scientific™, USA) and 60 µg of anti-ApoA1 antibody (MIA1404, ThermoFisher Scientific™, USA) were combined in a 2-mL tube. The beads-antibody slurry was incubated for 4 h at 4 °C (constant rotation). The tube was centrifuged at 1,000 × g for 2 min at 4 °C and the supernatant was discarded. The bead pellet was washed with Phosphate Buffered Saline (PBS) twice. Similarly, 50 µL of settled Immobilized Protein A/G (100 µL resin slurry, Pierce™ Protein A/G Plus Agarose, ThermoFisher Scientific™, USA) and 20 µg of anti-ApoE antibody (PA5-27,088, ThermoFisher Scientific™, USA) were combined a 2-mL tube. The beads-antibody slurry was incubated for 4 h at 4 °C (constant rotation). The tube was centrifuged at 1,000 × g for 2 min at 4 °C and the supernatant was discarded. The bead pellet was washed with PBS twice. Subsequently, ApoA1-antibody-bound beads and ApoE-antibody-bound beads were combined in a single 2-mL tube. The solution was gently mixed and centrifuged at 1000 × g for 2 min at 4 °C. The supernatant was discarded and 400 µL of CSF pool was added to the tube with beads. The sample underwent overnight incubation at 4 °C (constant rotation). At the end of incubation, the CSF-beads slurry was centrifuged (
CSF levels of ApoA1 and ApoE were assessed with use of the commercially available ELISA kits—Human Apolipoprotein AI ELISA Kit, ab108803 and Human Apolipoprotein E ELISA Kit, ab108813 (Abcam, UK) in n = 31 SAA-negative NC CSF samples. Both assays were performed according to the manufacturer protocol. All the standard curve points and CSF samples were run in duplicate. Optical density (OD) was read at 450 nm (wavelength correction 570 nm) by the Clariostar (BMG Labtech, Germany) plate reader. The 4-parameter logistic model was applied to generate a standard curve and interpolate concentrations of analysed CSF samples.
Total protein content was evaluated in the same samples with use of the Pierce™ 660 nm Protein Assay Reagent cat. 22,660 (Thermo Scientific™, USA). The assay was performed according to the manufacturer protocol. A set of BSA dilutions of known concentrations served as the standard to calculate total protein content of CSF samples. All the standard curve points and CSF samples were run in duplicate.
One-way analysis of variance (ANOVA) and Tukey post-hoc test for mean comparisons were applied while assessing differences among fitted kinetic parameters for different samples. Correlation between added seed mass and T2 parameters were computed according to Spearman. Two-tailed Student's t-test was applied when comparing adjusted integrated densities of dot blot images. Standard error of mean (SEM) is reported in each image showing bar plots and/or average fluorescence profiles. In the analysis of mass spectrometry data, an false discovery rate (FDR) of 1% was imposed and the criterion used to accept protein identification included probabilistic score sorted by the software. Correlations among SAA parameters and Log2-transformed CSF levels of ApoA1, ApoE and total protein were computed by means of Pearson's correlation coefficients. Ward linkage criterion was applied for hierarchical clustering of correlations.
A detailed description of the study design (experiments type and workflow) is present in Sect. 2.1 of Material and Methods and graphically summarised in Fig. 1.
Graph: Fig. 1 Scheme summarizing the pipeline and techniques used in the present work. Icon legend. TEM: transmission electron microscopy; ThT: thioflavin-T protein aggregation assay; SAA: α-synuclein seed amplification assay; 1H NMR: solution proton magnetic resonance spectroscopy; MS: mass spectrometry; CENTR. FILTERS: centrifugal filters; 1H-15N NMR: 2D proton-nitrogen solution nuclear magnetic resonance spectroscopy; DB: dot blot assay; WB: Western blot assay; IP: immunoprecipitation; ELISA: enzyme-linked immunosorbent assay
CSF inhibition of α-syn aggregation in SAAs remains poorly characterized, although it has been already reported in the literature [[
Graph: Fig. 2 CSF inhibits α-syn aggregation in unseeded and seeded conditions in a patient-dependent manner. A Six different human HC CSF samples (40 µL) spiked with 20 fg synthetic preformed α-syn fibrils (seeds) and analysed by diagnostic SAAs conditions: 0.3 mg/mL (19.6 µM) of recombinant α-syn in 100 mM PIPES pH 6.5 and 500 mM NaCl, 200 µL final volume. B Protein aggregation assay performed using 0.7 mg/mL of recombinant α-syn in PBS with (black) and without (red) 40 μL of NC CSF pool (final volume of 200 μL). C Seed amplification assay in PBS of different amounts of synthetic seeds (0.5, 5, 50, 500 and 5,000 pg) with and without NC pooled CSF (only 3 seed masses shown). D Graphical description of the fitting function used. A2 fits the fluorescence value of the second plateau, A1 fits the fluorescence value of the first plateau and A0 fits the baseline fluorescence. The time parameters t1 and t2 fit the first and the second inflection points, respectively, while d1 and d2 represent the slopes of the sigmoids. E Protein aggregation assay performed using 0.7 mg/mL of recombinant α-syn in PBS (final volume of 200 μl). Six glass beads with a diameter of 1 mm were added in each well. The shaking/incubation protocol consisted in 1 min shaking at 500 rpm and 14 min rest at 37 °C. The experiment was performed in quintuplicate; three replicates were used to produce the above reported average aggregation profile, the other two replicates were collected from the plate at t = 35 h and t = 165 h, and analysed by TEM to produce the representative images shown in the bottom of panel E). All ThT fluorescence traces are represented as average intensity over 3 replicates with error bars representing SEM
CSF from cognitively unimpaired normal-pressure hydrocephalous (NPH) is usually available in large volumes and is often used for assay development or as negative controls in diagnostic SAAs. Although considered negative controls due to lack of detectable α-syn seeds, dilution of CSF components has been reported to be common in NPH patients [[
Graph: Fig. 3 Analysis of NPH CSF samples. A Protein aggregation assay performed using 0.7 mg/mL of recombinant α-syn in PBS with 40 μl of CSF from 2 NPH subjects (40 µL). The two ThT fluorescence traces are represented as average intensity over 3 replicates with error bars representing the SEM. B Portion of 1D 1H NMR spectra relative to the two NPH CSF samples. C Relative concentration (emPAI score multiplied by protein molecular weight) ratio of total protein and the three most abundant protein constituents measured by nLC-nESI HRMS/MS in neat NPH1 and NPH2 CSF samples. Approximately 200 and 400 different proteins were detected in NPH1 and NPH2, respectively. Albumin was the most abundant protein followed by apolipoproteins and complement proteins. Apolipoproteins scores were summed together, with ApoA1 and ApoE being the most abundant (~ 85% of the total). Complement C3 and C4 were found as the most abundant complement proteins (~ 65% of the total)
After demonstrating that CSF composition is donor-dependent and correlates with inhibition of α-syn aggregation in vitro, we fractionated the previously used NC-CSF pool to investigate which CSF components were behind this inhibitory effect. Fractionation was performed using conical centrifugal molecular weight cut-off (MWCO) filters (for details about fractionation procedure and final concentration factors see Fig. 4A and Table S1). We obtained 6 samples after fractionation: whole CSF, CSF constituents of molecular weight (MW) above 100 kDa (> 100 kDa), CSF constituents of MW between 100 and 50 kDa (100-50 kDa), CSF constituents of MW between 50 and 10 kDa (50-10 kDa), CSF constituents of MW between 10 and 3 kDa (10-3 kDa), and CSF constituents of MW below 3 kDa (< 3 kDa). We then analysed the inhibitory effect of each of these 6 fractions on the spontaneous aggregation of α-syn in the previously mentioned PBS conditions (i.e., those of Fig. 2B, C, E). There were clear differences in α-syn aggregation depending on the MW of the CSF fraction. Whole CSF and all the fractions with MW > 10 kDa drastically inhibited α-syn aggregation, while 10-3 kDa and < 3 kDa fractions showed comparable aggregation to the reaction without CSF components (PBS control) (Fig. 4B). We estimated the second fluorescence plateau (A2) using the double sigmoidal model and compared the results to the maximum fluorescence readings (F
Graph: Fig. 4 Different CSF fractions differently affect α-syn aggregation. A Scheme of the CSF fractionation procedure. From a starting aliquot of 4.5 mL of CSF in PBS 1x, we collected 6 aliquots containing compounds of different molecular weight and froze them in liquid nitrogen. After every filtration with centrifugal filters, the flow-through of the filtered fraction was passed to a filter with smaller cut-off. B The addition of CSF fractions, whole NC-CSF pool, and PBS (40 μL) was analysed by ThT protein aggregation assay to evaluate effects on α-syn spontaneous aggregation. Background signal was corrected by subtracting the average fluorescence of three replicates containing PBS, whole CSF and CSF fractions without α-syn. All ThT fluorescence traces are represented as average intensity over 3 replicates with error bars representing the SEM. C Mean fitted A2 parameters (fitting was not possible for samples with whole CSF and the > 100 kDa fraction) and maximum fluorescence values (Fmax) estimated from individual ThT traces. Two scales of fluorescence intensity were used to better compare the results. Represented values correspond to the average of three replicates with error bars reflecting the SEM. D Relative concentration (emPAI score multiplied by protein molecular weight) of the most abundant protein constituents measured by nLC-nESI HRMS/MS. Apolipoproteins scores were summed together, with ApoA1 and ApoE being the most abundant (~ 85% of the total). Scores for fractions 10–3 and < 3 kDa are not shown since the protein content of these fractions was negligible with respect to the others
After identifying lipoproteins as main candidates to explain the inhibitory effect of CSF on α-syn aggregation, we evaluated if the effect is reproduced when using purified lipoprotein in the absence of the other CSF components. We used serum purified high-density lipoprotein (HDL) (LP3, Millipore-Sigma) and serum purified human serum albumin (HSA, A1653, Millipore-Sigma) as control. Albumin levels were quite similar between CSF fractions, and we have already shown that HSA only partially reduces α-syn aggregation at plasma concentrations (43 mg/mL) [[
Graph: Fig. 5 HDL reduces α-syn aggregation more efficiently than HSA. A ThT protein aggregation assay performed using 0.7 mg/mL of recombinant α-syn in PBS pH 7.4 in the presence of different concentrations of HSA (0, 0.3, 6.7 and 43 mg/mL) and HDL (0, 0.12 and 0.57 mg/mL). To remove the background fluorescence, the average fluorescence of three replicates containing the same amount of HSA and HDL without α-syn was subtracted prior to the analysis. The data represent the average fluorescence of three replicates with error bars representing the SEM. B Fmax and fitted A2 parameters fitted from individual traces and averaged on the three replicates are shown. C Inverted and window/level-adjusted image of the dot-blot assay performed on the final reaction products of HSA and HDL-containing samples. Dot-blots were probed with OC (detection of fibrils) and A11 (detection of amorphous oligomers) conformational antibodies. D Adjusted (background subtracted) integrated density measured in a circular region of interest (0.35 cm2) surrounding dots relative to samples 1–5 with error bars representing the standard deviation of the background noise. The average grey level is significantly lower (p < 0.001) for samples with HDL with respect to all the HSA concentrations tested by applying two-tailed Student's t-test both for OC and A11 antibodies
After confirming that purified HDL at plasma concentration replicates the inhibition of α-syn aggregation shown by whole CSF and by the > 100 kDa CSF fraction, we evaluated if HDL could retain the same level of inhibition when tested within a range of concentrations from 1 to 0.003 mg/mL (including the physiological ones of human CSF). Interestingly, we observed a dose-dependent partial inhibition of α-syn aggregation when adding 0.003 and 0.03 mg/mL, while 0.3 and 1 mg/mL completely blocked the formation of ThT-reactive aggregated species (Fig. 6A). The partial inhibition was most noticeable as a delay in the second inflection point (t2), although it was also observed as a reduction in fluorescence of the first plateau (A1) (Fig. 6A inset). ApoE and ApoA1 represent 50–60% of total CSF apolipoprotein, and their respective reported concentration in CSF is approximately 0.01 mg/mL and 0.004 mg/mL [[
Graph: Fig. 6 HDL reduces α-syn aggregation even at CSF physiological (ca. 0.03 mg/mL) and sub-physiological levels by preventing the formation of transient oligomeric/protofibrillary species. A Protein aggregation assay performed using 0.7 mg/mL of recombinant α-syn in PBS with 0, 0.003, 0.03, 0.3 and 1 mg/mL of added human serum HDL. To remove the background fluorescence, the average fluorescence of three replicates containing the same amounts of HDL without α-syn was subtracted prior to the analysis. All ThT fluorescence traces are represented as average intensity over 3 replicates with error bars representing SEM. B The presence of monomeric α-syn (14–18 kDa) in samples collected at different timepoints of the spontaneous aggregation process were monitored by WB using Syn211 antibody. Monomeric α-syn decreases as t increases due to the formation of fibrils. C In a similar way, a WB with Syn211 was performed on the reaction products obtained after 180 h, at different HDL concentrations with and without α-syn
CSF HDL has been reported to have an intermediate size between serum HDL and serum low density lipoprotein (LDL) [[
Graph: Fig. 7 HDL and LDL impede α-syn aggregation more efficiently than TTR. Lipoproteins exert their anti-aggregation properties by interlacing to early protofibrillary/oligomeric species. A-B Fitted kinetic parameters of a Protein aggregation assay performed using 0.7 mg/mL of recombinant α-syn in PBS with different concentrations of LDL, HDL and TTR. The average fluorescence of three replicates containing the same amounts of LDL, HDL and TTR without α-syn was subtracted prior to the analysis. C-H TEM images relative to the final products obtained using (C-D) α-syn alone (0.7 mg/mL), (E) LDL 0.3 mg/mL alone, (F) HDL 0.3 mg/mL alone, and a combination of both α-syn 0.7 mg/mL + LDL 0.3 mg/mL (G) or HDL 0.3 mg/mL (H)
We have shown lipoprotein inhibition of α-syn aggregation by ThT, detection of oligomeric and amorphous species by conformational antibodies, and the measurement of monomeric α-syn by WB, but these techniques do not allow the observation of α-syn fibrillary species in the presence of lipoprotein. Thus, we used transmission electron-microscopy (TEM) to evaluate final products of aggregation reactions containing α-syn alone, α-syn with LDL or HDL, and HDL and LDL without α-syn (Fig. 7C-H). In agreement with ThT fluorescence traces and the kinetic parameters collected for these reactions, there was a marked decrease in the amount of α-syn fibrils in reactions with LDL and HDL compared to the α-syn alone reaction. Interestingly, the few small α-syn fibrils observed were intertwined with lipoproteins (Fig. 7G, H), which can be easily seen in the α-syn reaction with HDL (Fig. 7H). Similar structures, although with less clear intertwining, have been observed by co-incubating α-syn with NC-CSF (Supplementary Material Fig. S9).
To determine if the inhibitory effect of lipoproteins on α-syn spontaneous aggregation also affects the seeded aggregation of α-syn intrinsic to SAAs, we used a highly sensitive and specific diagnostic α-syn SAA that can amplify α-syn seeds directly from PD/DLB/MSA CSF while showing no spontaneous aggregation with healthy control (HC) CSF [[
Graph: Fig. 8 HDL and LDL significantly modulate α-syn aggregation in SAAs. A-B Representative SAAs traces performed on a neat PD (PD47) CSF sample and on a neat HC (HC22) CSF sample (1 × physiological concentration) and the same samples spiked with 0.006 mg/mL (2 × physiological concentration) and 0.024 mg/mL (5 × physiological concentration) HDL or LDL. The average kinetic traces with error bars representing the SEM calculated on three replicates of wells containing CSF additioned with HDL and LDL are shown in panels A and B, respectively. C-D Average time-to-threshold (TTT) values measured in all the PD samples. The average TTT and SEM were calculated by assuming a TTT of 125 h (maximum TTT observed) for replicates in which aggregation was not considered significant (Fmax < 5000 a.u.). One-way ANOVA coupled with Tukey post-hoc test was applied to assess the statistical significance of the observed relative differences of all the individual measured traces among neat, 2 × HDL/LDL and 5 × HDL/LDL. Significant differences were marked with * with *** indicating a p-value < < 0.001. D Summary of the final SAA outcome for the analysed PD and HC samples. The outcome was categorized as: positive (+) when 3/3 replicates were determined positive by the probabilistic algorithm, inconclusive (?) when 2/3 replicates were determined positive by the probabilistic algorithm, and negative (-) when just 1/3 or 0/3 replicates were determined positive by the probabilistic algorithm
To further confirm the role of lipoproteins in the patient-dependent CSF inhibition on α-syn aggregation, we depleted ApoA1 and ApoE from a newly made pooled CSF (see Materials and Methods Sect. 2.3, 2.11, and Supplementary Fig. S11) by immunoprecipitation (IP). We then analysed α-syn spontaneous aggregation in the presence of immunodepleted CSF (IP CSF), CSF subjected to the IP procedure without antibodies (IP CSF no Ab), and neat CSF (neat CSF) from the same pooled CSF sample. The results of this experiment are summarised in Fig. 9. As expected, depletion of the two most abundant CSF apolipoproteins (IP CSF) resulted in a significantly faster α-syn spontaneous aggregation as compared with IP CSF no Ab and neat CSF.
Graph: Fig. 9 Experiments on ApoA1- and ApoE-immunodepleted CSF. A Protein aggregation assay performed using 0.7 mg/mL of recombinant α-syn in PBS with 40 μL of: CSF subjected to ApoA1 and ApoE IP (IP CSF), CSF subjected to IP without antibodies (IP CSF no Ab) and neat CSF belonging to different aliquots of the same NC pool. To remove the background fluorescence, the average fluorescence of three replicates containing the same reaction mix without α-syn was subtracted prior to the analysis. All ThT fluorescence traces are represented as average intensity over 3 replicates with error bars representing the SEM. B Average t2 fitted parameters with error bars representing the SEM. P-values were calculated by applying one-way ANOVA followed by Tukey post-hoc test
Lastly, we measured levels of the two most abundant CSF apolipoproteins (ApoA1 and ApoE) in n = 31 SAA-negative NC CSF samples. Although albumin represents most of total CSF proteins, total protein measurements were also considered, as both CSF ApoA1 and albumin derive from peripheral blood [[
Graph: Fig. 10 CSF ApoA1, ApoE and total protein content significantly correlate with SAA time variables. A representative SAA ThT fluorescence traces relative to samples producing the shortest (NC1, TTT = 12.8 h), median (NC2, TTT = 15.8 h) and longest (NC3, TTT = 20.3 h) TTT averaged on three replicates. All ThT fluorescence traces are represented as average intensity over 3 replicates with error bars representing the SEM. B Heatmap summarizing correlations (Pearson's) between SAA kinetic parameters and Log2-transformed CSF ApoA1, ApoE, ApoA1 + ApoE, and total protein. Hierarchical clustering was performed by using Ward linkage criterion. C-F scatter plots with detail of Pearson's correlation coefficients and relative p-value for TTT vs Log2-transformed CSF concentrations (originally in μg/mL) of ApoA1 (C), ApoE (D), total protein (E), and ApoA1 + ApoE (F). Linear regression lines with their 95% confidence intervals (dotted lines) are also displayed
Inhibition of spontaneous and seeded fibrilization of α-syn by human CSF has been briefly reported in recent literature, but not yet characterized [[
We used TEM to observe the ultrastructure of the α-syn fibrils in the presence of HDL and LDL and found both lipoproteins to colocalize with the fibrils, suggesting a direct interaction with α-syn aggregates. Although solution NMR has previously proven to be a suitable tool to identify proteins interacting with monomeric α-syn [[
We then evaluated the effect of lipoproteins on the amplification of α-syn seeds from human PD CSF samples in an ultrasensitive SAA. In these experiments, the reactions were supplemented with HDL and LDL to mimic CSF samples with 2X or 5X the physiological levels of HDL and LDL. We found that both HDL and LDL could significantly change the kinetic parameters of the amplification reaction starting from 2X concentration. At 2X LDL concentration, the SAA outcome of one of four PD samples changed from positive to inconclusive, while the same happened for two of four PD samples at both 5X HDL and LDL. Nevertheless, the likelihood of encountering CSF samples containing such high levels of lipoproteins is low, which is in line with the impressive sensitivity and specificity consistently reported by diagnostic SAAs [[
Our results describe a novel interaction between lipoproteins and α-syn that inhibits the formation of α-syn fibrils and could have relevant biological implications in vivo. Our findings also have direct and important implications for the future development and improvement of α-syn SAAs, which are currently the most promising diagnostic tool for synucleinopathies. The donor-specific inhibition of CSF on α-syn aggregation reported here offers an explanation for the lack of correlation seen between kinetic parameters and clinical progression. Moreover, our data reveals apolipoprotein as the main inhibitory components of CSF, suggesting that measurements of apolipoproteins and/or total protein content could be incorporated into data analysis models to eliminate confounding effects of CSF milieu on α-syn seed quantification efforts using kinetic parameters.
We are grateful to Dr. Maya Petricciuolo for the support provided during the preparation of TEM grids. We also thank the anonymous Reviewers for their constructive criticism and suggestions.
G.B., C.L. and M.Fr. designed the research; G.B., S.P., L.G., D.R. performed the protein aggregation assays to characterize the interaction between α-syn and CSF; L.C.M., Y.M., and C.M.F. planned and performed the ultrasensitive seed amplification experiments at Amprion. D.R. and S.B. expressed and purified recombinant α-synuclein; D.R. and S.G. expressed and purified recombinant transthyretin; G.B., L.C., D.R., and M.L. performed solution NMR experiments; S.G., D.R., and S.P. performed CSF pooling and fractionation; L.G. and S.P. performed dot blot and Western blot experiments; G.P. performed the mass spectrometry analyses; F.M. and C.M.G.D.L. performed the transmission electron microscopy measurements; A.L.W., D.C., and G.B. performed CSF immunoprecipitation, Western blot experiments, total protein content and ELISA ApoA1 and ApoE measurements; M.Fr., C.L., and L.P. provided the needed reagents and analytic tools; G.B., E.R., and M.Fi. performed the literature search; C.L., L.P., M.Fr. and M.L. supervised the project; G.B. analysed the data wrote the first draft; E.R, Y.M., and C.M.F., revised English language and grammar; All the authors reviewed the final version of the manuscript.
GB is supported by the Postdoctoral Fellowship for Basic Scientists grant of the Parkinson's Foundation (Award ID: PF-PRF-934916). SP is funded by the Associazione Italiana Ricerca Alzheimer Onlus (Airalzh) grant AGYR2020. This work was partially supported by the Italian Ministry of Health (RRC) to FM. This work was also supported by Regione Toscana (CERM-TT and BioEnable), the Italian Ministero dell'Istruzione, dell'Università e della Ricerca through the "Progetto Dipartimenti di Eccellenza 2023–2027 (DICUS 2.0)" to the Department of Chemistry "Ugo Schiff" of the University of Florence, and the Recombinant Proteins JOYNLAB laboratory. The authors acknowledge the support and the use of resources of Instruct-ERIC, a landmark ESFRI project, and specifically the CERM/CIRMMP Italy centre, as well as the project "Potentiating the Italian Capacity for Structural Biology Services in Instruct-ERIC, Acronym "ITACA.SB" (Project no. IR0000009) within the call MUR 3264/2021 PNRR M4/C2/L3.1.1, funded by the European Union – NextGenerationEU. ALW is supported by the Marie Skłodowska-Curie grant agreement No. 86019 – MIRIADE project (European Union's Horizon 2020 research and innovation program).
All the relevant data generated or analysed during this study are included in this published article and its supplementary information files. Raw fluorescence and mass spectrometry data are available to the corresponding authors upon reasonable request.
All the procedures involving human subjects were performed following the Helsinki Declaration. All patients and/or their legal representatives gave informed written consent for the lumbar puncture, CSF collection, assessment, analysis, and the inclusion in the study, that was approved by the local Ethics Committee (CEAS n 1287/08), University of Perugia. CSF samples were obtained with the informed consent of all participants.
Not applicable.
The authors declare the following competing financial interest(s): Prof. Parnetti served as Member of Advisory Boards for Fujirebio, IBL, Roche and Merck. Dr. Concha, Ms. Farris, and Mr. Ma are inventors on several patents related to SAA technology (PMCA) and are associated to Amprion Inc., a biotech company focused on the commercial utilization of SAA for diagnosis. All the other authors declare no financial and non-financial competing interests.
Graph: Additional file 1:Fig. S1. Silver staining on α-syn used in Protein aggregation experiments. Two replicate silver staining experiments performed on a 4-20% SDS-PAGE gel with 1 μg of purified α-syn after one (lane 3) and two (lane 2) size-exclusion chromatography steps. Fig. S2. T 50 measured in Six different human CSF samples spiked with 20 fg of seeds. The measured T 50 parameters were globally different, as assessed by one-way analysis of variance (ANOVA) and Fisher's LSD post-hoc test for mean comparisons. *0.01< p <0.05; **0.001< p <0.01: *** p <0.001. Fig. S3. Quantitative regression analysis of the seed masses in the absence of human CSF. Measured t2 parameters for the seeded experiment vs the quantity of seeds added; the horizontal axis is displayed in log10 scale. The t2 values displayed result from the average of three replicates, error bars reflect the standard deviation of the mean value. The data were fitted with a natural log function, the correlation between t2 and the added seed masses was assessed by means of Pearson's correlation coefficient (r). Fig. S4. NMR titrations of α-syn with CSF fractions. Intensity decreases of the signals of two-dimensional (2D) 15 N– 1 H HSQC experiments acquired at 950 MHz at T = 283 K on 15 N labelled α-syn (100 μM) in PBS after the addition of: (A) whole pooled CSF in PBS, (B) < 3 kDa CSF fraction in PBS and (C) > 100 kDa CSF fraction in PBS. The residues experiencing the largest decreases in signal intensity (smaller by one or more standard deviations with respect to the average value) are highlighted in light blue. The intensity ratios corresponding to overlapping peaks are highlighted in red (their values were not considered in the calculation of the average decreases and standard deviations). Fig. S5. CSF pH drift. The pH change due to the exposure of CSF to air was monitored over time in 500 μL of undiluted pooled CSF (A) and in the presence of PBS (400 μL CSF + 200 μL PBS 3x) in polypropylene vials with a Thermo Scientific Orion pH-meter equipped with a glass 6 mm diameter pHenomenal MIC 220 Micro electrode. Right before each measurement, the sample was vortexed for 20 sec and left open to air for another 20 sec. Fig. S6. Different CSF fractions differently affect α-syn aggregation. Mean fitted t2 parameters of samples with 40 μl of PBS/CSF fractions. The values displayed result from the average of three replicates with error bars reflecting the SEM. For whole CSF and the >100 kDa fraction the total duration of the experiment is shown due to the absence of appreciable aggregation. Fig. S7. Raw images of the dot-blot assays. A-B) Native images of the dot-blot assay performed on α-syn alone replicates at different timepoints with OC (A) and A11 (B) conformational antibodies. C-D) Native image of the dot-blot assay performed on the HSA and HDL containing samples with OC (C) and A11 (D) conformational antibodies. Fig. S8. WB experiments to track α-syn aggregation in the presence of human HDL. A) α-Syn aggregation patterns in samples collected at different timepoints of the spontaneous aggregation process, was monitored by WB using Syn211 antibody (4-20% SDS-PAGE, 2 μg protein loaded). Monomeric α-syn decreases as t increases due to the formation of fibrils. B) In a similar way, a WB with Syn211 was performed on the reaction products obtained after 180 h, at different HDL concentrations with and without α-syn (exposure time 210 s). C) The experiment was then repeated by doubling the amount of sample loaded into the gel to better highlight the presence of oligomeric species (exposure time 30 s). Under these conditions, chosen to better visualize the signal at 150-200 kDa, the α-syn monomer bands at 1 and 0.3 mg/mL HDL may not quantitively reflect monomer concentration (overloaded lanes). Fig. S9. Representative TEM images of α-syn incubated with CSF. Representative TEM images obtained by analyzing samples obtained by the co-incubation of α-syn 0.7 mg/mL at 37 °C with pooled human CSF (1:5 ratio with respect to total reaction volume). Samples were subjected to cycles of incubation (13 min) and shaking (double-orbital, 2 min) at 500 rpm. Fig. S10. NMR titrations of α-syn with HDL, LDL and TTR. A) Intensity decreases of the signals of two-dimensional (2D) 15 N– 1 H HSQC experiments acquired at 950 MHz at T = 283 K on 15 N labelled α-syn (100 μM) in PBS after the addition of 0.57 mg/mL serum-derived HDL. The intensity ratios corresponding to overlapping peaks are highlighted in red. B) Intensity decreases of the signals of two-dimensional (2D) 15 N– 1 H HSQC experiments acquired at 950 MHz at T = 283 K on 15 N labelled α-syn (100 μM) in PBS after the addition of 1 mg/mL serum-derived LDL. C) Intensity decreases of the signals of two-dimensional (2D) 15 N– 1 H HSQC experiments acquired at 950 MHz at T = 283 K on 15 N labelled α-syn (100 μM) in PBS after the addition of 3 mg/mL TTR. Fig. S11. WB experiments performed on immunodepleted CSF. WB experiments were performed with anti-ApoA1 (MIA1404) and anti-ApoE (PA5-27088) antibodies on neat CSF and supernatants (400 μL CSF, 100 μL slurry) resulting from immunoprecipitation procedure performed using different quantities of the same antibodies (IP CSF samples), immunoprecipitation performed without antibodies (CSF IP no Ab). The conditions relative to IP CSF ApoA1 60 µg and IP CSF ApoE 20 µg were then selected for Protein aggregation assays. Table S1. Concentration factors and final volumes of the CSF fractions.
• CV
- Coefficient of variation
• DLB
- Dementia with Lewy bodies
• EDTA
- Ethylenediaminetetraacetic acid
• ELISA
- Enzyme-linked immunosorbent assay
• emPAI
- Exponentially modified PAI
• FDR
- False discovery rate
• Fmax
- Maximum fluorescence
• Fmin
- Minimum fluorescence
• HC
- Healthy control
• HDL
- High-density lipoprotein
• HMW
- High molecular weight
• HSA
- Human serum albumin
• HSQC
- Heteronuclear single quantum correlation
• IMAC
- Immobilized metal ion affinity chromatography
• IP
- Immunoprecipitation
• IPTG
- Isopropil-β-D-1-tiogalattopiranoside
• LB
- Luria–Bertani
• LDL
- Low-density lipoprotein
• MSA
- Multiple system atrophy
• MW
- Molecular weight
• MWCO
- Molecular weight cut-off
• NC
- Neurological controls
- nLC-nESI HRMS/MS
- NanoLiquidChromatography coupled to High Resolution Mass Spectrometry equipped with a nanoelectrospray interface
• NMR
- Nuclear magnetic resonance
• NPH
- Normal pressure hydrocephalus
• OD600
- Optical density at 600 nm wavelength
• PBS
- Phosphate buffered saline
• PD
- Parkinson's disease
• PGDS
- Prostaglandin-D synthase
• PIPES
- Piperazine-N,N′-bis(2-ethanesulfonic acid)
• PMCA
- Protein misfolding cyclic amplification
• PMSF
- Phenylmethylsulfonyl fluoride
• PVDF
- Polyvinylidene fluoride
• RFU
- Relative fluorescence units
• ROI
- Region of interest
• RT-QuIC
- Real-time quaking-induced conversion
• SAA
- Seed amplification assay
- SDS-PAGE
- Sodium Dodecyl Sulphate—PolyAcrylamide Gel Electrophoresis
• SEM
- Standard error of the mean
• T50
- Time to reach the half maximum
• TEM
- Transmission electron microscopy
• ThT
- Thioflavin-T
• TTR
- Transthyretin
• WB
- Western blot
- α-syn
- α-Synuclein
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
By Giovanni Bellomo; Silvia Paciotti; Luis Concha-Marambio; Domenico Rizzo; Anna Lidia Wojdaƚa; Davide Chiasserini; Leonardo Gatticchi; Linda Cerofolini; Stefano Giuntini; Chiara Maria Giulia De Luca; Yihua Ma; Carly M. Farris; Giuseppe Pieraccini; Sara Bologna; Marta Filidei; Enrico Ravera; Moreno Lelli; Fabio Moda; Marco Fragai; Lucilla Parnetti and Claudio Luchinat
Reported by Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author