Substitution of Met-38 to Ile in γ-synuclein found in two patients with amyotrophic lateral sclerosis induces aggregation into amyloid

Significance Understanding how synuclein proteins form amyloid in vitro and in cells is crucial to understand disease. Previous studies showed that the P1 region (residues 36–42) of α-synuclein controls amyloid formation. We here report a single nucleotide polymorphism in the P1 region of γ-synuclein (γSyn) (Met38 to Ile) found in two individuals with amyotrophic lateral sclerosis. Both individuals have a second polymorphism in the same gene (Glu110 to Val) that is commonly found in the general population. We show that Ile38-containing γSyn forms amyloid rapidly in vitro, while Met38 does not aggregate into amyloid and Val110 is protective, slowing aggregation. The results highlight the critical role of the P1 sequence in tipping the balance between a protein’s propensity for amyloid formation.


In silico analysis
The Syn and Syn sequences were aligned using the UniProt align tool available at www.uniprot.org/align(1).The aggregation propensity and solubility of the sequences were analysed by using the online tools Zyggregator (2) and CamSol (3) at pH 7.0 or 4.0.

NAC peptide synthesis
Peptides (NH2-EQVTNVGGAVVTGVTAVAQKTVEGAGSIAAATGFV-COOH and NH2-EQANAVSEAVVSSVNTVATKTVEEAENIAVTSGVV-COOH for Syn-NAC and Syn-NAC, respectively) were synthesised on a CEM Liberty Blue peptide synthesiser with microwave assistance using default coupling cycles for the first 15 residues and double couplings for the rest of the peptide.The synthesis was performed on 0.1 mmol scale using Valine preloaded HMPB-NovaPeg resin (0.56 mmol/g), DMF as a solvent, 20% (v/v) piperidine in DMF for the deprotection and 5 eq. of Fmoc-protected amino acid, 5 eq.DIC and 5 eq.Oxyma for couplings.Cleavage from resin was accomplished using TFA:H2O:TIS:EDT, 92.5 : 2.5 : 2.5 : 2.5 (5 mL x 3 h) and peptides were precipitated using cold ether.Peptides were purified on preparative HPLC Agilent 1260 infinity system equipped with UV detector, fraction collector and Kinetex EVO 5 µm C18 100Å 21.2 x 250 mm reverse phase column using a gradient of 10 -30% (v/v) acetonitrile with 0.1% (v/v) ammonia, followed by lyophilisation.The identity of the peptides was confirmed by high-resolution mass spectrometry on Bruker Maxis Impact spectrometer using electrospray ionisation.Immediately before use the peptide was resuspended in the appropriate experimental buffer and an aliquot of this was diluted thousand-fold into water.The A205 of the diluted sample was measured using a dual beam UV-1800 spectrophotometer (Shimadzu) with extinction coefficients of 104,320 and 96,120 M -1 cm - 1 for Syn-NAC and Syn-NAC, respectively (4).The remaining undiluted peptide was diluted in reaction buffer to the appropriate concentration.

ThT fluorescence assays
Four ThT assay buffers were prepared by mixing Tris and sodium acetate, to give Tris/sodium acetate buffers at pH 4.5, pH 5.5, pH 6.5 and pH 7.5 each with a total 20 mM ionic strength.
(Syn or Syn) or NAC peptide solution in each assay buffer containing 20 μM ThT (each well containing a total of 100 µL solution) in sealed 96-well flat bottom assay plates (Corning, nonbinding surface 3651) in a FLUOstar Omega plate reader (BMG Labtech) at 37°C with a single teflon coated poly-ball per well (PolyScience Inc, 17649-100) and continuous shaking at 600 rpm for 72 h.ThT fluorescence was exited at 444 nm, emission detected at 480 nm.
Experiments, in general, were performed three times with four replicates.Data were processed in MATLAB (r2022b) where the raw data was normalised to the 95 th percentile of the signal amplitude for samples that reached a plateau.Where appropriate, slopes in the plateau phase were removed by fitting the plateau phase data and subtracting the fitted values.The tlag was calculated as the x-intercept of the tangent to the inflection point and the t50 as the time taken to reach y = 0.5 after normalisation.If a plateau was not reached within the experimental time frame or if an enhanced ThT signal was not observed at all then the signal was normalised to the 95 th percentile of the average pH independent signal of the same variant in that microplate (all proteins showed a ThT positive signal at pH 4.5).Where necessary tip sonication was performed after 72 h using a 3 mm ultrasonic probe (Sonics, 630-0422) at 20 % amplitude with a VCX-130PB ultrasonic processor (Sonics) and the sample was redeposited into a fresh assay plate of the same type and incubated for a further 48 h whilst ThT signal was monitored.

Negative stain TEM
For negative stain TEM images, 5 µL of each sample was added to a glow discharged carbon coated copper grid and incubated for 20 sec.Samples were dried by dabbing with filter paper and the grid was then washed two times by submersion into a water droplet with drying steps in-between each wash.This process was repeated with two droplets of 1 % (w/v) uranylacetate to negatively stain the samples.Imaging was performed on either the FEI Tecnai T12 electron microscope or the FEI Tecnai F20 electron microscope.The NAC peptide samples were centrifuged and resuspended in 20 μl of water prior to deposition.

Mutagenesis, expression, and purification
Syn variants with single amino acid substitutions were generated by Q5 site directed mutagenesis (NEB), starting with a plasmid containing human cDNA encoding M38/E110 variant of Syn in a pET23a vector.Recombinant expression of the proteins was performed in Escherichia coli BL21 (DE3) cells and purification performed as described previously (5) before lyophilisation and storage at -20°C.Immediately before an experiment, the protein was dissolved in the desired buffer.The purity and correct mass of all proteins was confirmed by mass spectrometry (ESI-MS) and SDS-PAGE.

Strep-tagged Syn variants and in-tube aggregation assay
A DNA fragment encoding Strep-tag was added in-frame just upstream the stop codon of each of four cDNAs encoding the human Syn variants using a standard PCR technique and the resulting DNA fragments were cloned into the bacterial expression vector pCS19 (6).
Expression of recombinant proteins in KU98 cells was induced by 0.5 mM IPTG for 5 h.Bacteria cells were then collected, lysed, and the cleared lysates incubated in a boiling water bath for 20 min.Denatured protein precipitates were separated from the Syn-containing supernatant as described previously (7) and Strep-tagged Syn variants were purified using Strep-tactin columns (IBA-GmbH, Goettingen, Germany) according to the manufacturer's instructions.A solution of each protein (900 µL of 0.92 mg/mL protein) in diluted BXT elution buffer (final concentrations: 33 mM TrisHCl, 50 mM NaCl, 0.33 mM EDTA, 1.66 mM biotin, pH 8) was prepared in duplicate in sterile 2 mL test tubes and incubated at 37 o C with constant shaking at 190 rpm for 13 days.

Differential pelleting assay
Replicate samples from the endpoint of each ThT assay were pooled and 100 μL from each was loaded into a TLA100 ultracentrifuge tube (Beckman Coulter 343775) and subjected to ultracentrifugation at 100,000g for 30 min in a Beckman Coulter Optima MAX-XP Ultracentrifuge.The soluble fraction after ultracentrifugation was collected.Both the noncentrifuged (whole) and the soluble fraction were diluted 1 in 6 into the appropriate ThT assay buffer before being diluted further with SDS loading dye.The samples were depolymerised by boiling in 2% (w/v) SDS for 10 min and loaded onto 15% Tris-Tricine SDS-PAGE gels.The gels were stained using InstantBlue Coomassie protein stain (Abcam ISB1L) and imaged using an Alliance Q9 Imager (Uvitec).Band intensities were quantified using ImageJ version 1.52a.The percentage of pelleted protein was quantified from the intensity of the band corresponding to the whole fraction minus the intensity of the band corresponding to the soluble fraction.
Alternately for the NAC peptides, quantification was performed as above except that the soluble fraction was diluted 1 in 10 in 0.1% (v/v) trifluoroacetic acid (TFA) in a conical HPLC vial (Thermo Scientific C4010-13) and 10 μl was loaded onto a Nucleosil C4 300Å 250 x 4.6 mm HPLC column (Chromex) that had been equilibrated in 0.1% (v/v) TFA.Protein was eluted over 15 min on a gradient of 5-80% (v/v) acetonitrile using a Nexera LC-40 system (Shimadzu).
Absorbance over a range of wavelengths from 190 nm to 350 nm was detected using a photodiode array detector (Shimadzu) and the absorbance at 205 nm was used in all calculations.10, 7.5, 5, 2.5 and 1 μl of freshly dissolved 8 μM Syn-NAC peptide (dissolved and initially diluted to 80 μM in water followed by dilution to 8 μM in 0.1% (v/v) TFA) was loaded onto and eluted from the same column, identically as before, to form a set of standards.The area under the curve (AUC) of the peak for each run at 9.5 min was calculated in MATLAB (r2022b) and the set of standards was fit to a straight line using the concentrations 8, 20, 40, 60 and 80 μM as the x axis coordinates.Finally, given that the original sample at the start of the ThT experiment was 80 μM, the concentration of the soluble protein was converted to percentage of sample in the pellet.

CryoEM sample preparation and data collection
Syn-I38/E110 fibrils were formed at pH 6.5 from a 100 µM starting monomer concentration in 1.5 mL Eppendorf tubes with incubation for three weeks at 37˚C with shaking at 600 rpm, followed by two weeks incubating quiescently at room temperature.Prior to preparing cryo-EM grids, the fibrils were concentrated 4x-times by centrifuging 100 µL of the reaction at 13,000g for 10 min and resuspending the pellet in 25 µL of the resulting supernatant.The sample was applied to Tergeo plasma cleaned (Pie Scientific, 60 s) Lacey carbon 300 mesh grids, blotted and frozen in liquid ethane using a Vitrobot Mark IV (FEI) with a 0 sec wait and 5.5 sec blot time, respectively.The Vitrobot chamber was maintained at close to 100% humidity and 6˚C.The cryoEM dataset was collected at the University of Leeds Astbury Centre Biostructure Laboratory using a Titan Krios electron microscope (Thermo Fisher) operated at 300 kV with a Falcon4 detector in counting mode and Selectris energy filter set to a 10 e-V slit width.A nominal magnification of 130,000x was used yielding a pixel size of 0.95 Å.A total of 5,280 movies were collected with a nominal defocus range of -1.3 to -2.5 µm and a total dose of ~41 e-/Å2 over an exposure of 5 s, corresponded to a dose rate of ~7.4 e-/pixel/s.The movies were collected as EER fractions with 1539 raw hardware frames.

CryoEM data processing
The raw EER movies were fractionated into 38 fractions, aligned and summed using motion correction in RELION-4 (8) with a dose per frame of 1.1 e-/Å2.CTF parameters were estimated for each micrograph using CTFFIND4 (9).Images that did not contain fibrils were removed by quickly screening through 10 Å lowpass-filtered micrographs paneled at a 0.2x scale, leaving 1,579 micrographs for further processing (SI Appendix Figure S5A).Fibrils from roughly 100 micrographs were manually picked, and fibril segments were extracted and used to train a picking model in crYOLO (10) to automatically pick from all the images with an interbox spacing of 14.4 Å (three helical repeats).The resulting 584,854 segments were extracted 2x binned (with a 570 Å box size) and 483,631 segments remained after picking artefacts such as carbon edges and ice contamination were removed using multiple rounds of 2D classification using the VDAM algorithm in RELION-4 (SI Appendix Figure S5B).
There was no evidence of twisting features in the fibrils, both in 2D class averages or in the raw micrographs.A selection of 338,457 fibrils showing signs of internal features were reextracted unbinned (with a 285 Å box size) and subsequent 2D classification showed clear layer lines in the class averages with a regular ~4.8 Å spacing, consistent with a cross- structure in the fibrils (SI Appendix Figure S5C).To generate the intensity plots in SI Appendix Figure S5D, the grey values were measured along a straight line drawn parallel to the fibril axis for a single class average image in Fiji/ImageJ (11), with two examples shown for different class averages.3D classification using a featureless cylinder as a starting template and helical twist searches around 359.5-360.0˚did not yield any promising models to pursue further.

Rate-zonal density-gradient ultracentrifugation
The protocol was adapted from (12).For each experiment sucrose was dissolved to 50, 40, 30, 20 and 10 % (w/v) into the appropriate buffer at the pH used for the ThT assays.750 μL of each solution was then layered into an OptiSeal polypropylene centrifuge tube (Beckman Coulter 361621) by pipetting equal amounts of each solution gently down the side of the tube with the 50% (w/v) solution first, to create a discrete sucrose gradient.750 μL of proteincontaining sample was added last.The sample was then subjected to ultracentrifugation at 113,000 x g for 4 h in a TLA110 rotor.150 μL volumes were taken from the top down.Only the middle 150 μL from each sucrose fraction was analysed.Fractions were imaged by SDS PAGE and negative stain TEM as described above, except that the grids were held in each water droplet for 30 sec.The rate at which a particle moves through a solvent during centrifugation, the settling velocity, is defined by Stokes' law (equation 1): Equation 1.
where p is the particle density, p0 is the solvent density, g is the acceleration force, D is the particle size and N0 is the solvent viscosity.
When the particle density is larger than the solvent density, as is the case for the differential pelleting assay, the settling velocity is proportional to the particle size.However, in a ratezonal density-gradient experiment, solvents with a variety of densities are used.When the solvent density becomes equal to the particle density (p = p0), V = 0 and the particle will stop moving through the solvent.Equilibration over a long period of time (4 h here) will result in the separation of particles by density.

Liposome formation
25 mg of dry DMPS powder (Avanti polar lipids) was resuspended in 1 ml of 80:20 v/v chloroform and methanol.A film of lipid was created in a round bottom test tube by evaporating off the solvents using a stream of nitrogen followed by overnight vacuum desiccation.Where possible the following steps were performed above the lipid transition temperature (Tm) of DMPS.The lipid film was resuspended in 890 μl of 20 mM sodium phosphate buffer, pH 6.5, briefly vortexed for 30 s and incubated for 30 minutes.The multilamellar vesicles were then subject to five freeze thaw cycles before being extruded by being passed through a 100 nm membrane 21 times.The resulting liposomes were stored at 4 C and were used within 48 h.

Fibril formation in the presence of liposomes
Amyloid formation in the presence of DPMS LUVs was monitored using ThT fluorescence under quiescent conditions, using 50 μM initial monomer, 30 C, with no beads in 20mM sodium phosphate buffer, pH 6.5.All other parameters were the same as in the fibril formation assays as described in the absence of liposomes.

Liposome binding by circular dichroism
Far-UV CD spectra were acquired in 1-mm path length quartz cuvettes (Hellma) using a ChirascanTM plus CD Spectrometer (Applied Photophysics).CD spectra were acquired using a 1-nm bandwidth and 1-s time step, and data were collected at 1-nm increments at 30 °C.
Three scans (190−260 nm) were acquired and averaged per sample per repeat of the experiment.Binding of Syn variants to liposomes was calculated using a previously described ( 13) fitting protocol and equation 2.
Equation 2 Where  is the concentration of DMPS divided by the concentration of synuclein protein,   is the fraction of bound protein calculated by subtracting the CD signal of free protein from the observed CD signal and dividing that by the difference in CD signal of protein in the free and saturated bound state and L is the number of lipids interacting with one protein monomer.

Cell culture experiments
Plasmid constructs for expression of four Syn variants in eukaryotic cells were generated by PCR from corresponding plasmids in pCS19 vector (see above) using the primers: forward 5′-CCCCTCGAGACAACCAAAATGGATGTCTTCAAGA-3′ and reverse 5′-CCGAATTCTCTAGATCAGTCTCCC-3′.The PCR fragments were digested with XhoI/EcoRI, cloned into pcDNA3.1 mammalian expression vector and inserts were verified by DNA sequencing.
Twenty-four hours prior to transfection, cells were plated in different well-plate formats, at a density of 50-60% confluency prior to transfection.The day after, cells were transfected with equimolar ratios of plasmids with cDNA encoding the different variants of human Syn.
Transfections were performed with calcium phosphate.Briefly, 3 h prior to transfection, fresh cell medium was added to the cells.DNA was diluted in 1 × HBS buffer (25 mM 4-(2hydroxyethyl)-1-piperazineethanesulfonic acid, 140 mM NaCl, 5 mM KCl, 0.75 mM Na2HPO4 2H2O, 6 mM dextrose, pH 7.1) and 2.5 M CaCl2 solution was added dropwise with vigorous mixing.After incubation for 20 min, the mixture was added dropwise to the cells.After 16 h the cells were fed with fresh medium.
Images were acquired using a 63× objective and analysed using LAS AF v.
For survival experiments, transgenes for UAS-SNCG variants or control UAS-mCD8::GFP were expressed under transcriptional regulation of Elav-Gal4.Vials containing approximately 10 flies were passaged to fresh food and survival censused on every second day.

Eye phenotype
UAS-SNCG variants or control UAS-mCD8::GFP transgenes were crossed to eye specific GMR-Gal4.Flies were aged up to 3 weeks at 25°C (n=10), at which point each were collected and frozen at -20°C in 1.5 mL microcentrifuge tubes, for imaging.Eyes were examined and images captured using Zeiss Stemi 508 with Zeiss AxioCam ERc 5s Rev.2.

Rapid iterative negative geotaxis (RING)
We conducted a modified version of RING assay (16).Flies expressing UAS-SNCG variants or control UAS-mCD8::GFP were reared and maintained at 25°C.For RING analysis groups of 10 flies per genotype were passaged into the empty vials and left to acclimatize for 10 min.
Vials were mounted into a custom-built frame composed of a vial carrier and vertical rails, allowing parallel analysis of six vials from a consistent height (20 cm) and impact.Digital if the amount of protein in the whole sample is equal to that in the soluble sample.(C) Ratezonal density-gradient ultracentrifugation.Since Syn monomers equilibrate in the 0 and 10% (w/v) sucrose fractions (Supplementary Figure S7) we assume that all sample that equilibrates between 0 and 10% (w/v) sucrose is monomer.This assay can separate monomer from oligomer, but not dense oligomers from fibrils which both may pellet in the 50 % (w/v) sucrose fraction.Performing density gradient centrifugation in a discontinuous sucrose gradient subsequent to differential pelleting (which removes fibrils ( 20)) can thus be used to separate fibrils from other dense aggregates.Schematic diagrams of SDS PAGE gels are shown (all bands are considered to be of exactly equal intensity in these particular examples), alongside a table indicating how the example data would be interpreted.Created using biorender.com.
2.2.1 (LeicaMicrosystems) software.The percent of cells containing Syn aggregates within the population of successfully transfected cells was then determined by counting.At least 100 transfected cells were assessed per variant in each of 4 independent experiments.The percent of cells containing large (>2 µm) vs, small (<2 µm) Syn aggregates were also assessed.Statistical analyses of the data were performed using the one-way ANOVA with Turkey post-hog test for multiple comparisons or Student's t-test for independent variables.Differences were considered statistically significant at *p < 0.05.Statistical analyses of the data were performed using the one-way ANOVA or Student's t-test for independent variables.The data are presented as mean ± standard deviation and represent the results from at least four independent experiments.Differences were considered statistically significant at *p < 0.05.
images were collected for analysis at 10 sec post-impact (Panasonic Lumix DC-F782).The assay was repeated five times.Images were analysed in ImageJ, with height climbed of individual flies recorded.The average distance travelled by all flies in a single vial, averaged across five consecutive RING trials is considered 1 technical replicate.Immunofluorescent stainingDrosophila heads were removed and fixed for 20 min in 4% (v/v) paraformaldehyde (PFA) in phosphate buffered saline containing 0.1% (v/v) Triton-x100 (PBS-Tx).Whole brains were dissected and fixed for a further 20 min in 4% (v/v) PFA in PBS-Tx.Brains were blocked in 10% (v/v) goat serum in PBS-Tx for 30 min, then incubated with primary rabbit polyclonal antihuman γSyn-specific SK109 antibody(17) overnight at 4°C.Following washing with PBS-Tx, brains were incubated overnight with fluorescent secondary antibody (1:200, goat anti-rabbit IgG Alexa Fluor 488, ThermoFisher A-11008).Maximal orthogonal z-projection fluorescent micrographs were collected using a Zeiss Cell Observer Spinning Disc confocal microscope using Zen Blue software (Version 2.6).flies of other three lines survived marginally, but statistically significantly better than control flies (p<0.001 for both Mantel-Cox log-rank and Gehan-Breslow-Wilcoxon tests).fluorescence signal.(B) Kinetics of fibril formation of I38/E110 at pH 7.5 and I38/V110 at pH 4.5 monitored by ThT fluorescence in which samples taken after 72 h of incubation with beads and shaking (A) were then either subjected to sonication (orange) or were not sonicated (brown) and then incubated for a second 48 h period under quiescent conditions.Three replicates of each are shown (some of which are overlapping).(C) Pelleting assay demonstrating that after sonication and an additional 48 h incubation no further pelletable material is formed, regardless of whether the sample had been sonicated.morphologies (scale bars represent 10 nm).(C) Unbinned 2D class averages, in which the ladder typical for cross- structures can be clearly seen (scale bar represents 20 Å). (D) Plot of grey value along the length of two representative unbinned 2D class averages (shown in (C)) demonstrating that the peaks are aligned with the 4.8 Å repeating units (vertical lines) classical of cross- amyloid.(maroon) LPRs with 100 nm DMPS liposomes over 150 h.(B) Negative stain TEM images of Syn variants after 150 h incubation with 100 nm DMPS liposomes at 0,4,8,16 and 60:1 LPR, as indicated.(C) Far UV CD spectra of 25 μM Syn variants mixed with 100 nm DMPS liposomes at 0, 10, 20, 25, 30, 35, 40, 45, 50, 60, 80 and 100:1 LPR.(D) The MRE at a wavelength of 222 nm for each LPR.The binding curve (in red) was fitted to find the Kd and L values (not possible for the I38 proteins which did not saturate at any of the lipid to protein ratios).average height climbed normalised to the average height climbed by the same set of flies of each line at the age of 20 days as 100%.(*p<0.05;**p<0.01;****p<0.0001,Kruskal-Wallis ANOVA with post-hoc Dunn's test, n=6 groups of flies per genotype).(E) Kaplan-Meier curves analysis revealed that with the exception of γSyn I38/V110, pan-neuronal expression of γSyn variants modestly improve survival of flies compared with GFP expressing controls (p<0.001for both Mantel-Cox log-rank and Gehan-Breslow-Wilcoxon tests; n=128 for γSyn M38/E110V, n=142 for M38/E110, n=97 for I38/V110, n=277 for I38/E110, n=213 for GFP).