α-Synuclein Fibril-induced Inclusion Spread in Rats and Mice Correlates with Dopaminergic Neurodegeneration

Dynamic light scattering profiles short α-syn fibrils

α-Syn fibrils seed the formation of inclusions in neurons in culture and in the mouse and rat brain. To generate fibrils, we purified mouse and human α-syn protein from bacteria to a high purity and incubated monomer protein with shaking until turbid. After ultracentrifugation at 100,000 × g, >90% of the α-syn protein was found in the pellet fraction (P100), with <10% remaining in the supernatant (Figure 1A, S100 fraction). Analysis via non-denaturing blue-native PAGE efficiently separated monomeric protein that migrated at ~56 kDa from fibrils that all migrated above 500 kDa in size (Figure 1B). Increasing the amount of sonication time did not convert any of the higher-molecular weight conformations into monomer based on native PAGE analysis.

Examination of the fibrils by transmission electron microscopy showed typical filaments. Increasing total times of sonication generally reduced fibril length (e.g., 30 seconds to 240 seconds, Figure 1C). Analysis of monomer protein by electron microscopy did not reveal the presence of aggregated protein or filaments, consistent with native PAGE results. We measured the lengths of thousands of particles from different preparations of fibrils and plotted the distributions as histograms (Figure 1 D–G). While an average α-syn fibril length can be calculated from a distribution, breakage of fibrils into smaller fragments becomes much less efficient as the fibrils become shorter. Only a few nanometers average length (e.g., ~four α-syn monomer subunits) was different in doubling sonication times from 60 seconds, to 120 seconds, to 240 seconds (Figure 1H). ~Two α-syn monomer subunits likely represent one nanometer of fibril (Serpell et al., 2000). These results are potentially consistent with past studies demonstrating longer protein fibrils (e.g., composed of insulin protein) are stiff and brittle and subject to fracture, whereas shorter fibrils become increasingly fracture-resistant because they can bend and absorb otherwise shearing forces (Fitzpatrick et al., 2015; Knowles and Buehler, 2011).

Thioflavin binding demonstrated that increasing sonication times and shortening fibrils did not destroy the amyloid conformation of the α-syn fibrils (Figure 2A), and that the shortest α-syn fibrils we could generate could seed the formation of new fibrils in vitro (Figure 2B). Consistent with these results, circular dichroism demonstrated that these short α-syn fibrils remained structurally like the longer fibrils (Figure 2C). To determine whether other biophysical characteristics were changing with extensive sonication, we compared dynamic light scattering profiles. Light scattering changed with increasing sonication, and intensity distributions were used to calculate concentration (Figure 2D,E). Increasing sonication time progressively reduced the molecular weight and increased concentration of the fibrils in a linear manner (Figure 2E). To test whether the fibril preparations were functionally increasing in concentration with respect to seeding capability of new fibrils, we included the same weight of each fibril preparation together with monomer protein (1:20 w/v) into a thioflavin binding assay and observed a proportional increase in seeding capacity, in close agreement with the light scattering predictions (Figure 2F).

Concentration-dependent fibril seeding phenotypes in primary neurons

The addition of fibrils to primary hippocampal neurons seeds the recruitment of endogenous α-syn into inclusions that can be identified with an antibody to pS129-α-syn (Volpicelli-Daley et al. 2011). A recent study demonstrated that reducing fibril length to less than 50 nm and transfecting these fibrils into SH-SY5Y cells caused robust inclusion formation when the cells were transiently transfected with α-syn (Tarutani et al., 2016). However, past studies have not described fibril concentration (i.e., number and length of particles) effects with respect to inclusion formation from endogenously expressed protein in cultured neurons. Past studies have also utilized both human and mouse α-syn to template new inclusions in cells that over-express α-syn (Peelaerts et al., 2015). Further, a significant species barrier has been recently identified for human fibrils templating new mouse fibrils in primary neurons (Luk et al., 2016).

To test different concentrations of fibrils with respect to the concentration of α-syn endogenously expressed in the neurons (Supplemental Figure 1), we titrated 49 nm (average) human and mouse fibrils (30 second sonication condition, Figure 1) to induce inclusions in primary neurons. We verified by silver-stain denaturing PAGE analysis that preparations of human and mouse fibrils were comparable to one another and contained the same amount of α-syn protein (Figure 3A). Fibrils were added to neurons at seven days in vitro (DIV), and cultures were immunostained fourteen days later. Immunofluorescence was performed using antibodies to tau, to visualize axons, NeuN to visualize neuronal nuclei, and to pS129-α-syn to visualize inclusions. Monomeric α-syn does not template new inclusions and pS129-α-syn signal in these cultures was used as a threshold cut-off (Volpicelli-Daley et al., 2011). Signal from pS129-α-syn was also defined in α-syn knockout neurons to ensure only authentic inclusions derived from α-syn were analyzed (Supplemental Figure 2). We found that femptomolar concentrations of the fibrils were sufficient to induce widespread inclusions throughout the cultures (Figure 3B–F). Inclusions could be broadly categorized as larger inclusions in the soma or thin filaments in axons.

As the endogenous substrate (i.e., α-syn) is critical to the formation of new fibrils in neurons, we measured the concentration of endogenous α-syn in the neurons at the time of fibril addition and compared this to the amount of α-syn fibrils added. We calculated ~110 ng of α-syn per well, or ~0.3 pg of α-syn protein per neuron on average, representing nearly 0.2% of the total amount of protein in the culture (Supplemental Figure 1). Notably, this concentration is higher than that reported in purified synaptosomes from rat brain (Wilhelm et al., 2014). To quantify pathology in the neurons at different ratios of fibril seeds (49 nm average) to substrate, we developed an automated imaging platform for unbiased inclusion quantification. The addition of fibrils at a ~1:50 ratio of fibril weight to endogenous α-syn (w/v, 0.1 pM rodent α-syn, ~2 ng mL⁻¹ Figure 4A,B) induced inclusions in ~1% of the possible tau-positive process area in the culture and ~1% of NeuN bodies (Figure 3D). At concentrations of fibrils greater than 1 pM (i.e., 1:5 ratio of w/v of α-syn in fibrils to total α-syn in the neurons), human fibrils seeded comparably to mouse fibrils. There was no difference in the amount of soma inclusions that could be quantified between 10 pM and 100 pM fibril exposures of either human or mouse fibrils (fibril weight exceeding the amount of endogenous α-syn), indicating a ceiling effect. ~12–15% of all NeuN bodies were susceptible to developing an inclusion and ~4% of all tau-positive area could be identified with inclusions (Figure 4A,B and Figure 3F). We did not observe a significant loss of tau-positive process area under any of the fibril exposures or loss of NeuN-positive bodies (Figure 4C,D and 4G,H), suggesting the overt loss of cells was unlikely to introduce a confounding variable to the analysis.

Fibril length has been proposed as a critical determinant for fibril seeding but has not been demonstrated in cells that express endogenous levels of α-syn (Tarutani et al., 2016). We determined whether reducing fibril length at higher concentrations of α-syn would change inclusion profiles and potentially expand the number of neurons that could develop pathology. Unsonicated fibrils (but subjected to one freeze-thaw cycle, ~195 nm on average) were less efficient than all of the sonicated fibrils (Figure 4E,F). Shortening fibrils beyond 100 nm did not induce more inclusions either in the soma or axons. In summary, these results show 1) a narrow dynamic range for inclusion formation in concentration gradients (10 fM to 100 pM tested) of exogenous fibrils, 2) the lack of a species barrier in human fibrils cross-seeding mouse inclusions in neurons, 3) a general lack of effect of fibril length (195 to 29 nm range tested) for seeding new inclusions in primary neurons.

Fibril-induced α-syn pathology and dopaminergic neurodegeneration in rodents

α-Syn fibril injections have been described in several mouse strains including C57BL/6 and C3H (Luk et al., 2012), as well as rats (Paumier et al., 2015). To test fibril length and strains of rodents in inclusion formation and spread in the brain, we injected a series of mice (C57Bl/6J and C3H/HeJ) and rats (Taconic Sprague Dawley) unilaterally in the dorsal striatum with a constant amount of α-syn protein in all conditions. Six months after injection, we analyzed the number of pS129-α-syn inclusions that formed in neurons near the injection site in the striatum as well as in brain nuclei that are known to project to the striatum. In all groups of rodents, an equivalent mass of monomer α-syn protein was injected, as well as sections analyzed from α-syn knockout rodents (Supplementary Figure 3), to distinguish authentic pS129-α-syn inclusions from off-target and background staining. Except for monomer α-syn injections that never generated inclusions, all rodent brains demonstrated inclusions of variable number in the dorsal striatum, amygdala, piriform cortex, and thalamus.

In the striatum of C57BL/6J, the abundance of inclusions increased dramatically with the injection of short 29 nm fibrils compared to the same weight of 195 nm fibrils (Figure 5A–D). These inclusions were morphologically similar to those formed in primary neuron cultures (Figure 3), and soma inclusions were always adjacent or overlapping with NeuN-positive cells in the striatum (Figure 5F–H). Neuritic inclusions in the striatum were mostly negative for tyrosine hydroxylase (TH, Figure 5I), but occasionally overlapped with TH expression in shorter inclusions (Figure 5J). The somatic inclusions localized to striatal projection neurons and we did not detect an instance of a ChAT interneuron with an inclusion (Figure 5K–L).

The amygdala has prominent projections to the site of our fibril injections in the dorsal striatum (McDonald, 1991) and neurons throughout the amygdala developed robust neuritic and somatic inclusions adjacent to NeuN-positive nuclei (Figure 6A). The amygdala inclusion load corresponded well to the inclusion load in the striatum, with higher concentrations of short fibrils inducing more robust pathology. A small fraction of dopaminergic neurons in the SNpc also developed inclusions (Figure 6B1–3). Confocal analysis of cell bodies in the SNpc revealed inclusions that tended to spiral around the nucleus of dopaminergic neurons (Figure 6B4–6). Every cell in the SNpc with an inclusion was TH-positive and TH staining did not appear reduced in these cells.

Several cortical regions demonstrated inclusions after injection of the striatum. The piriform cortex and somatosensory cortex (Figure 6C1–3) were prioritized for analysis because these brain regions avoided the needle tract of fibril injection. Confocal analysis of these cortical areas showed that all the inclusions were associated with NeuN positive cells, with fibril-spiral inclusions in pyramidal neurons in the piriform cortex (Figure 6C4–5). The inclusion burden in shorter-fibril (29 nm) injected C57BL/6J was more robust compared to longer-fibril (195 nm) injected animals. Long-fiber injected C57BL/6J demonstrated very sparse inclusions in every brain region. In contrast to C57BL/6J, C3H/HeJ mice appeared to be insensitive to fibril length, as both longer-fibrils and shorter fibrils produced very similar and robust inclusion profiles in the different brain regions (Figure 7). Thus, the 29 nm short-fibril injections of C57BL/6J and C3H/HeJ produced inclusions of similar abundance. But, (Figure 5–7), the long fibrils (195 nm) only produced robust inclusions in C3H/HeJ mice.

Fibril injections in rats also induces pS129-α-syn inclusions (Paumier et al., 2015). We performed unilateral injections into the dorsolateral striatum in Sprague Dawley rats with the same fibril preparations used in mice and analyzed brain sections six-months post-injection. We observed a remarkably similar distribution of inclusions as compared to C57BL/6J mice (Figure 8). Inclusions appeared in the amygdala, piriform cortex, and SNpc and the abundance of inclusions was most robust with the higher concentration of the short fibrils fibrils (Figure 8A4, B4, C4, D4). In contrast, longer fibrils (i.e., 195 nm) produced very sparse to no inclusions in these areas, similar to C57BL/6J. Monomer α-syn protein injections in the rats failed to produce detectable inclusions in any of these brain regions. Overall, we can confirm that shorter fibrils injected into the brain are more efficient in seeding inclusions in Sprague Dawley rats and C57BL/6J. The C3H/HeJ strain appears to be more sensitive to inclusion formation than the other rodents tested, with longer-fibrils performing equivalently to shorter fibrils in seeding inclusions.

Striatal injections of fibrils have been reported to cause a ~30% loss of dopaminergic neurons in the SNpc in rats and mice six-months post-injection (Luk et al., 2012; Paumier et al., 2015). We determined whether varying the inclusion burden in the rodent brain via manipulation of fibril length and using different mouse strains or rats would correlate with dopaminergic neurodegeneration. We developed a series of rats and mice and injected fibrils of various lengths into the striatum and counted inclusions through the SNpc six-months post-injection (Figure 9A). We ranked the pathology in three other brain regions (piriform cortex, amygdala, and thalamus) from zero (no detectable inclusions), one (one or a few inclusions), to five (the highest burden of inclusions), in mice and rats (Figure 9B). We also counted TH+ (Nissl counterstained) positive cells in ipsilateral SNpc and contralateral SNpc using unbiased stereology (see Methods). In many of these animals, a clear reduction in the number of TH-positive cells in the ipsilateral SNpc could be observed (Figure 9C). In counting the number of TH stained cells through the SNpc and plotting the percent loss (ipsilateral SNpc to contralateral SNpc) against the inclusion burden (Figure 9A–B), we observed a trend where animals that had the highest numbers of inclusions in the SNpc (Figure 9D) and brain regions connected to the striatum (Figure 9E) typically had higher levels of dopaminergic neurodegeneration. In these animals, the same amount (mass) of α-syn was injected, but little to no neurodegeneration could be observed in animals that failed to develop inclusions.

pS129-α-Syn inclusion burden positively correlates with dopaminergic neurodegeneration

Our fibril injections into the striatum resulted in a few hundred neurons in the SNpc with inclusions we could detect versus the thousands of neurons that remain (Figure 9A). Formation of pS129-α-syn inclusions may occur first in axons, followed by inclusion appearance in the soma over time. In vitro inclusion formation occurs more readily in axons compared to the soma, likely because of the high concentration of α-syn at the presynaptic terminal relative to the cell body (Volpicelli-Daley et al., 2011). We determined whether fibril exposures at the dopamine neuron cell body would 1) increase the number of inclusions in the SNpc, 2) cause toxicity unrelated to inclusion formation or fibril length, and 3) whether inclusions would spread in other brain regions different from striatum injections. We injected 49 nm mouse α-syn fibrils into the SNpc or striatum in a series of rats and analyzed brain sections six-months post-injection. In both SNpc or striatum injections, a comparable number of SNpc neurons demonstrated inclusions. In addition, spiny projection neurons in the striatum developed inclusions both with striatal and SNpc injections. When fibrils are injected in the SNpc, inclusion spread to spiny projection neurons may represent a straightforward assay to measure inclusion formation distant from the injection site. The amygdala also showed an indistinguishable inclusion load between the two groups of animals (Figure 10B2 and C2). One difference between the two injection sites was the obviously less abundant inclusion spread to the cortex, with sparse or no inclusions detectable in the piriform or somatosensory cortices in SNpc injected rats compared to striatum injected rats (Figure 10). This may reflect the relative sparsity of cortical projections to the SNpc.

Stereological counts for TH+ SNpc cells after injection in the SNpc of rats revealed a heavy lesion (on average 50% loss, but greater than 95% in a few rats, see Figure 11A1) due to the injection of short (49 nm) mouse α-syn fibrils. We did not observe an obvious loss of cells or loss of TH signal in the adjacent ventral-tegmental area. Plotting the absolute counts of TH+ SNpc cells revealed the possibility of contralateral TH cell loss in some of the rats (Figure 11A2). However, the extent of neurodegeneration in the ipsilateral SNpc did not appear to correspond well with the extent of contralateral neurodegeneration, revealing the complexity of the relationship of ipsilateral to contralateral neurodegeneration (Figure 11A3). Injection of the comparable mass of monomer α-syn failed to produce any detectable loss of cells in any of the rats, suggesting that the surgical procedure of fibril injection or other components included in the injection besides fibrils were unlikely to elicit a neurodegeneration phenotype, at least on their own (Figure 11A,B). Plotting neurodegeneration as a function of striatum inclusion count in the SNpc fibril-injected rats revealed a strong positive correlation (r=~0.8) between spiny projection neurons with inclusions and the extent of neurodegeneration in the SNpc (Figure 11C).

Generation of recombinant α-syn

Mouse and human α-syn, encoded in the inducible bacterial expression plasmid pRK172 and purified by maxi-prep (Qiagen), was transformed into chemically-competent BL21 (DE3) Codon Plus cells (Clontech). A single colony was selected for growth in terrific-broth (Fisher) supplemented with 100 μg mL⁻¹ ampicillin (Sigma). Growth was monitored to log-phase, protein production induced by IPTG for 2 hours, 37°C, and pellets were collected at OD₆₀₀ = 0.80. Bacterial pastes were weighed and ~5 g was combined with 25 mL of lysis buffer consisting of 750 mM NaCl, 10 mM Tris, pH 7.6, 1 mM EDTA, 1 mM PMSF, and 1× bacterial protease inhibitor cocktail (RPI). Homogenates were sonicated at 70% power (Fisher Model 500 Dismembrator) for 1 min 20 sec (5 seconds sonication pulses with 25 seconds of off time, on ice), and tubes placed in boiling water for 15 minutes. After ~15 minutes on ice, samples were centrifuged for 25 minutes at ~10,000 × g for 25 minutes at 4° C, twice. Supernatants were loaded into SnakeSkin dialysis tubing (3.5 kDa MWCO, Fisher) and placed into 4 L of ice-cold buffer (10 mM Tris, pH 7.6 with 50 mM NaCl, 1 mM EDTA, PMSF) with stirring overnight. Next, supernatants were centrifuged at 100,000 × g for 1 hour at 4 C, and supernatants concentrated to ~5 mL using Amicon Ultra15 3.5 MWCO columns. Concentrate was next passed through a HiLoad 16/600 Superdex Column, 1 × 120ml (GE Healthcare) with 10 mM Tris, pH 7.6 with 50 mM NaCl, 1 mM EDTA as running buffer. Fractions containing α-syn protein, as assessed by SDS-PAGE and Coomassie stain, were combined and loaded into SnakeSkin dialysis tubing (5 kDa MWCO, Fisher) and placed into 4 L of ice-cold buffer (10 mM Tris, pH 7.6 with 25 mM NaCl, 1 mM EDTA, PMSF) with stirring overnight. After dialysis the fractions were loaded at one time into a HiPrep Q HP 16/10 Column, 1 × 20 mL (GE Healthcare) with a loading buffer (10 mM Tris pH 7.6, 25mM NaCl, l mM EDTA, 1 mM PMSF) and eluted with a gradient application of high-salt buffer (10 mM Tris, 1M NaCl pH 7.6, 1 mM EDTA, 1 mM PMSF). Samples containing α-syn were identified by SDS-PAGE and Coomassie blot, and dialyzed into the same dialysis buffer (excluding PMSF and EDTA) and concentrated as before. Concentration of protein (monomer, undiluted) was determined by BCA assay (Pierce), purity assessed by SDS-PAGE and Coomassie blot. Endotoxin levels were determined using a Pierce LAL chromogenic endotoxin quantification kit. For primary neuron experiments, a maximum of 1.3 EU was added and for in vivo experiments, a maximum of 6.3 or 12.6 EU was injected (mice and rats, respectively).

Generation of fibrils, sonication, and biosafety procedures

Monomer protein was diluted to a 5 mg mL⁻¹solution of mouse or human α-syn protein (buffered with 50 mM Tris, pH 7.5, and 150 mM KCl) were included in 1.5 mL polypropylene centrifuge tubes shaken at 700 R.P.M. in an Eppendorf thermomixer for seven days at 37°C. The solution becomes turbid (i.e., cloudy) during the incubation. The fibrils were then aliquoted and frozen in liquid nitrogen or at −80 C. Immediately before each experiment, fibrils were thawed and diluted in phosphate-buffered saline (pH 7.5) to the desired amount (if necessary), and a 1/8 inch probe tip was inserted to a few millimeters from the bottom of the tube, and 30% power (Fisher Scientific Sonic Dismembrator FB120110) was applied for the indicated time (from 15 sec to 240 sec) in 15 second intervals with 1 sonication pulse and 1 second wait. In between 15 second intervals, fibrils rested in a metal block at room temperature for 2 min to dissipate heat in the solution. Since sonication aerosolizes fibrils, all sonication steps were performed in a biosafety-level 2 cabinet with the operator wearing disposable wrist guards and double-gloved. For all procedures, monomer α-syn was kept on ice, and fibril α-syn preparations were kept at room temperature, each for up to 6 hours before discarding. Clean-up and inactivation of the fibrils on contaminated surfaces was accomplished using a 1% SDS solution as described (Bousset et al., 2016; Thomzig et al., 2014).

Biophysical measures

For circular dichroism, 0.1 mg mL⁻¹ of monomer or fibril samples were loaded onto a photospectrometer cell with a 0.2 mm path length and scanned with the CD DSM-20 system (Olis), with a wavelength from 190 to 260 nm. For dynamic light scattering, estimated molecular weight of monomers and sonicated fibrils, 0.05 mg mL⁻¹ of α-syn samples were measured with Titan DynaPro (Wyatt Technology) at 25°C. Data were collected and analyzed using the Dyna V6.3.4 software package, with the solvent (phosphate-buffer saline) background signal subtracted from each sample. Molecular weights of particles were estimated from intensity distributions. Thioflavin measurements with recorded in 100 μM preparations of monomer (1.4 mg mL⁻¹), combined with the indicated amount of fibrils, together with 20 ng mL⁻¹ of thioflavin T (Sigma), mixed in 20 μL reactions in 384 well plates. Plates were incubated at 37°C, shaken at 800rpm, and emission at 483 nm (440 nm excitation) was recorded at the indicated interval. For transmission electron microscopy, 3 μL of 0.1 mg mL⁻¹ samples were applied to glow-discharged 400 mesh, carbon-only, copper grids (Electron Microscopy Sciences) and negatively stained with 1% uranyl acetate (Polysciences). The grids were imaged in a FEI Tecnai F20 electron microscope (Eindhoven) operated at 200 kV with nominal magnification at 65,000× and a defocus range of −1.0 μm to −1.27 μm. Images were collected on a Gatan Ultrascan 4000 CCD camera.

Animals

All animal protocols were approved by the University of Alabama at Birmingham Institutional Animal Care and Use Committee. C57BL/6J mice, C3H/HeJ, and timed-pregnant mice, and α-syn knockout mice (Strain 016123), were obtained from Jackson laboratories. Sprague Dawley outbred rats were obtained from Taconic Farms.

Primary neuron culture

At days 16–17 gestation in isoflurane-anesthetized mice, animals were sacrificed by cervical dislocation and embryos removed to ice. After a five-minute incubation on ice, brains were removed and hippocampi dissected into Hibernate E medium (BrainBits). Tissue was next digested with papain (Worthington Biochemical) in HBSS buffer supplemented with HEPES (10 mM, pH 7.4) and 100 mM sodium pyruvate with 1% penicillin/streptomycin (pen/strep, Invitrogen). Primary neurons were dissociated from tissue pieces into Neurobasal media (Gibco) supplemented with 1× B27 reagent (Gibco), 5 mM GlutaMAX (Gibco) and 10% fetal bovine serum (Invitrogen) and cells plated to poly-d-lysine (Sigma) coated surface at a density of 5 × 10⁴ cells per cm². 2 hours later, media was replaced with Neurobasal media, containing only GlutaMAX and B27 supplement.

Intracranial injection and processing

Intracranial injections of α-syn fibrils or monomer control (equivalent protein weights, 10 μg or 20 μg per injection each for mouse and rat, respectively) were conducted in ~8–10 week old mice or rats anesthetized with vaporized isoflurane and dexdomitor on a gas mask fitted to a digital stereotaxic frame (David Kopf). Rodents were wrapped in an infrared core-warming stage with monitoring of pulse-oxygen and respiration (Kent Scientific PhysioSuite). Solutions for injection were drawn to a 32-gauge custom needle (Hamilton) with a 110° bevel fitted to a gas-tight syringe and controlled by a digital pump (Harvard Apparatus). The injection of 2 μL (mouse) or 4 μL (rat) occurred over the course of 20 minutes to prevent solution running back up the needle track. Solutions were injected into the right dorsal striatum in mice (and rats) at the following empirically derived coordinates: 1.0 (0.7 rat) mm anterior and 1.85 (3.0 rat) mm lateral to the Bregma, and 3.0 (5.5 rat) mm ventral relative to the skull, with the needle bevel facing laterally. Solutions were injected into the right SNpc at the following empirically derived coordinates: 3.1 (4.65 rat) mm posterior and 1.5 (2.25 rat) mm lateral to Bregma, and 4.6 (7.45 rat) mm ventral relative to skull, with the needle bevel facing laterally. Scalp incisions were closed by suture. Six months later, rodents were deeply anesthetized with isofluorane and transcardially perfused with PBS (pH 7.4) followed by freshly prepared 4% PFA buffered in PBS. Brains were removed, post-fixed for 24 hours in the 4% PFA and PBS solution, floated into 30% sucrose PBS solution for up to three days, and frozen in isopentane solution (−50°C) and stored at −80°C.

Immunofluorescence and Immunohistochemistry

Primary neurons were affixed in 4% paraformaldehyde (Electron Microscopy) with 4% sucrose for 1 hour at room temperature, and rinsed four-times with phosphate-buffered saline (PBS). A blocking buffer of 3% bovine-serum albumin (Sigma) with 0.1% Triton X-100 in PBS was applied for 1 hour at room temperature, and primary antibodies applied in the same buffer with pS129-α-syn (20 ng mL⁻¹, clone 81A, IgG2a, Biolegend), NeuN (200 ng mL⁻¹, Clone A60, IgG1, EMD Millipore) and Tau (1 μg mL⁻¹, rabbit polyclonal, Dako), overnight at 4 °C. Neurons were rinsed and then a solution of secondary antibodies (Goat anti-IgG2a AlexaFluor 555 (4 μg mL⁻¹), Goat anti-rabbit AlexaFluor 488 (4 μg mL⁻¹), and Goat anti-IgG1 AlexaFluor 647 (4 μg mL⁻¹) with Hoechst 33342 stain (10 μg mL⁻¹) for 3 hours at room temperature. Coverslips were adhered with Prolong Gold (Invitrogen) mounting media.

Brain sections were cut to 40 μm on a freezing microtome and incubated in an antigen recovery solution of 10 mM sodium citrate, pH 6.0, supplemented with 0.05% tween, at 37 °C for 1 hour with gentle rocking. Sections were rinsed with tris-buffered saline (TBS, pH 7.4) and combined with blocking buffer (5% normal goat serum with 0.3% Triton X-100 in TBS). Sections were rinsed and combined with primary antibodies pS129-α-syn (20 ng mL⁻¹, clone 81A, IgG2a, Biolegend), NeuN (200 ng mL⁻¹, Clone A60, IgG1, EMD Millipore), tyrosine hydroxylase (400 ng mL⁻¹, rabbit polyclonal, EMD Millipore), and choline acetyltransferase (1 μg mL⁻¹, rabbit polyclonal, EMD Millipore) for 24 hours at 4° C with gentle rocking. Sections were rinsed and combined with secondary antibodies described above for 24 hours at 4 °C with gentle rocking. Sections were mounted to superfrost slides and mounted to coverslips with Prolong Gold (Invitrogen). Confocal microscopy was carried out with a Leica SP5 and imaging processing with LAS X software (Leica).

For immunohistochemistry with 3,3′-diaminobenzidine (DAB) and Nissl staining, brain sections were treated as above except for a quench for 30 min at room temperature with 0.3% H₂O₂ in methanol following antigen retrieval. Secondary antibodies (Jackson Immunologic), conjugated to biotin were used and sections treated with the ABC kit (Vector) followed by development with Impact DAB reagent (Vector). Sections were dehydrated progressively in ethanol and then Histo-Clear (national Diagnostics) and mounted to coverslips with Permount reagent (Fisher). Some sections were rehydrated with immersion to water and then treated with cresyl violet solution (0.1% cresyl violet, 0.08% acetic acid) for one-minute in a microwave, and then rinsed in water and dehydrated with ethanol, and immersed back into Histo-clear and mounted onto slides. Slides were analyzed on an Olympus BX61 wide field microscope.

Unbiased approaches, stereology, and statistics

Primary neurons were analyzed by triple-fluorescence microscopy on a Zeiss Cell Observer with an automated stage controlled by Zen 2.1 (Carl Zeiss) software. Grids for image acquisition were determined automatically and processed for signal related to tau and pS129-α-syn (238 μm frames), and NeuN (750 μm frames). Automatic measurements were recorded for the area of each condition occupied by a tau-positive process, the area within that process occupied by phosphorylated α-syn (α-syn/tau area ratio), and the total number of NeuN bodies detected within each image, and finally the number of phosphorylated α-syn bodies adjacent to NeuN bodies (i.e., soma inclusions). Signal for pS129-α-syn generated in parallel experiments from α-syn KO neurons was used to establish thresholds of background. Thousands of processes and features were collected for each condition reported in at least three independent experiments.

Unbiased stereological estimations of the total number of TH+ (Nissl counterstain) in the SNpc was performed using an optical fractionator probe (Stereologer software, Stereology Resource Center) by an investigator blinded to the experiment identity. A 4x objective was used to identify borders of the SNpc as a reference space, identified in part through Nissl contrast stain, at all midbrain levels. Sections used for counting covered the entire SNpc and were equally spaced 120 μm apart, with the frame placed randomly on the first counting area and systematically moved through all counting areas until the entire region was sampled. Guard zone height used was 2 μm to avoid artifacts on the cut surface of sections. Sampling grid size (e.g., 200 μm) was adjusted as needed, due to the heterogeneity of lesion size, to allow 1 to 5 objects on average counted at each of 100 to 200 x–y locations (minimum 100 objects counted for each observation, based on mean section thickness). Cells that were large (~30–50 μm) and Nissl+ but not TH positive within the borders of the SNpc were also counted (less than 1% of the total counts in each experimental group reported herein, consistent with past observations that α-syn fibril exposures do not suppress TH expression in cells that harbor inclusions (Volpicelli-Daley et al., 2016)). For counting soma inclusions through the SNpc or striatum, the rare-object counting probe was utilized. Only somatic perinuclear (e.g., adjacent to Nissl bodies) phosphorylated α-syn inclusions were counted because neuritic serpentine inclusions likely extended through multiple sections, with some facing perpendicular and nearly invisible to the cut surface, thereby invalidating stereological estimations.

A ranking based approach was also taken to estimate the amount of soma pathology in other brain areas that included the cortex, amygdala, and dorsal striatum. A score of 0 to 5 was assigned by a rater blinded to the experimental condition, with 0 representing no observable inclusions in any sections (e.g., Fig 5A), a score of 1 representing only a few neurons on average harboring inclusions per section (e.g., Fig 5B, Fig 6B2, Fig 8C2), a score of 2 to 3 with a mid-range load of inclusions (e.g., Fig 8A3 and Fig 8B3, respectively), and 4 to 5 representing sections with the highest amount of inclusions observed in any group (e.g., Fig 5D and Fig 6A3). Average scores for each animal were determined from 6–10 sections from each brain region.

Statistical analysis and graphs were performed and created with Graphpad Prism 5.0 software. Experimental groups were confirmed for normality using the Kolmogorov-Smirnov test and homogeneity of variances using Levene’s test. Two-tailed one-way ANOVA tests were used to calculate p values, where less than 0.05 was considered significant, and for those groups with significant differences, a post-hoc test (Tukey’s) was applied to each possible combination. Linear regression lines are shown for some plots, and Spearman correlation values were selected for analyzing ranked data.