Impairment in stimulus-response learning as a cognitive biomarker in a model of synucleinopathy

Abstract

Translating findings from preclinical assessments of cognitive function relevant to synucleinopathies into successful clinical trials poses significant challenges. In this study, we examined the effectiveness of human-relevant touchscreen cognitive assessments in detecting the impact of α-synuclein pathology in a mouse model of synucleinopathy on the emergence of stimulus-response learning impairments as a potential cognitive biomarker. We generated two types of α-synuclein pre-formed fibrils, each displaying distinct biophysical characteristics in vitro and biochemical properties when stereotaxically injected in vivo. These two types of fibrils were capable of triggering impairments in stimulus-response learning in pre-motor, prodromal disease stages, akin to what is observed in humans with synucleinopathy. This work supports the use of touchscreen cognitive assessments in conjunction with robust preclinical rodent models for identifying targets and testing therapeutic strategies for synucleinopathic diseases.

Introduction

Neurological diseases affecting cognition are leading sources of ill health and disability worldwide, making them key public health priorities. Animal models are critical for understanding and developing treatments for these diseases, but present significant challenges as robust, validated animal models must be combined with appropriate, translational, and reliable behavioural assessments to understand how pathology affects cognition [1–4]. One class of neurological disease for which such modelling is currently underdeveloped are the synucleinopathies, which include Parkinson’s Disease (PD), Lewy Body Dementia, and Multiple Systems Atrophy. In these diseases, cognitive deficits are highly prevalent, often appear in early prodromal disease stages [5], and have significant impacts on quality of life, disability, and caregiver burden [6, 7].

Synucleinopathies are characterized by misfolding, aggregation and propagation of α-synuclein (aSyn) [8]. In its native form, aSyn is a small soluble protein that is predominantly expressed in the presynaptic nerve terminals of mammalian brain tissue [9], and which carries out key regulatory functions that include lipid metabolism and transfer [10–12] as well as synaptic transmission and plasticity [13–16]. In disease states, however, aSyn is known to misfold and aggregate into toxic proteinaceous and membranous inclusions known as Lewy bodies and Lewy neurites, where a large portion of aSyn is phosphorylated at the serine residue 129 (pS129) [17–19]. Several lines of evidence support a prion-like cell-to-cell transmission of misfolded aSyn [20–25]. According to this model, disease-related conformers of aSyn can act as a template or seed to trigger host aSyn to misfold, become pathological, and spread across the central nervous system [26–28]. Notably, brain atrophy and aSyn deposition occur predominately in cortico-basal ganglia- thalamic circuits [29, 30], which are key for maintaining cognitive functions [31] and are sensitive to changes in dopaminergic transmission [32, 33].

To study synucleinopathies in preclinical models, a widely used and reliable approach is the injection of aSyn pre-formed fibrils (PFF) in the M83 hemizygous transgenic mouse model, which overexpresses human mutant α-synuclein (A53T) and develops progressive neurological symptoms [24, 25, 34, 35]. In this model, aSyn pathology can be induced by injecting fragmented aSyn PFFs composed of human recombinant misfolded aSyn species into the brain, spinal cord, or gut, which initiates the seeding of pathological species near the site of injection, and which can spread within interconnected tissues [24, 25, 36]. Recently, it has been shown that this model is also capable of recapitulating the progressive brain atrophy and aSyn deposition in cortico-basal ganglia-thalamic circuits commonly seen in patients [35], along with the development of early-onset deficits in cognitive flexibility [34, 35]. Notably, misfolded proteins can assume different conformations that likely contribute to different patterns of diffusion, aggregation, and pathology [37, 38]. Indeed, similar to the well-documented prion strains [39, 40], aSyn can form multiple conformational strains that can be propagated and selected in vivo using the PFF model [37, 41, 42], enabling the comprehensive study of aSyn pathology under varying misfolding conditions.

One highly translational cognitive construct that is poorly studied in models of synucleinopathy is the acquisition of stimulus-response (S-R) learning. S-R learning is a cognitive process essential for adaptive decision-making and day-to-day functioning, which is commonly impaired in patients with synucleinopathies [43–47], and could serve as a cognitive biomarker. To study S-R learning preclinically in a way that maintains high translational potential, we developed the touchscreen-based Visuomotor Conditional Learning (VMCL) task [48–50]. Notably, VMCL in rodents has been shown to be dependent on relevant nodes within the cortico-basal ganglia-thalamic circuits [51–54], and is sensitive to changes in dopaminergic transmission [55], making it highly relevant for the study of synucleinopathies. Here, we tested whether a combined approach of the robust aSyn PFF model and rodent touchscreen VMCL task could be used to model S-R learning impairments relevant to synucleinopathies.

Materials & methods

Expression, purification, and fibrilization of α-synuclein

Human aSyn PFF-1 and PFF-2 are available from a commercial source (StressMarq Biosciences Cat# SPR-322 and SPR-317, respectively), which are expressed, purified, and fibrilized using methods previously described [36, 56]. In brief, wild-type human aSyn (native, untagged) in a PET24a vector was expressed in E.coli BL-21 grown in Terrific Broth with kanamycin. Cultures were grown shaking at 37 ^oC to an OD₆₀₀ between 0.6-0.9 before induction with 0.2 mM IPTG for overnight expression at room temperature (Type 1, aSyn PFF-1) or 1 mM IPTG for 4 h expression at 37 ^oC (Type 2, aSyn PFF-2) before cells were pelleted at 4000 x g, washed with 1X PBS pH 7.4 and frozen −80 ^oC. Pellets were boiled at 90 ^oC for 15 min in 10 mM Tris pH 7.6, 1 mM EDTA and 750 mM NaCl with protease inhibitor cocktail (Sigma Cat #S8820-2TAB) and insoluble material was removed by centrifuging 20,000 x g at 4 ^oC. Supernatant was incubated with PEI on ice for 20 min and re-centrifuged to precipitate excess nucleic acid. Protein was dialyzed into 10 mM Tris pH 7.6, 1 mM EDTA and 40 mM NaCl overnight at 4 ^oC. Monomers were purified using anion exchange with a gradient from 40-500 mM NaCl. Fractions containing aSyn visible on SDS-PAGE were buffer-exchanged into 1X PBS pH 7.4. If necessary, monomers were passed through an endotoxin removal column (Pierce) to ensure an endotoxin level <1 EU/mg protein as determined by a ToxinSensor Chromogenic LAL Endotoxin Assay Kit (GenScript). Purified, low endotoxin monomers were concentrated to 5-8 mg/mL using a 5 kDa MWCO concentrator (Amicon), filter sterilized (0.2 µm), aliquoted and frozen −80 ^oC. Concentration and purity >95% were confirmed by nanodrop and SDS-PAGE. For fibrilization reactions, monomers were thawed, mixed, and centrifuged 20,000 x g at 4 ^oC to remove any pre-existing aggregate. Supernatant containing monomers was pooled, re-filter sterilized (0.2 µm), and diluted to 5 mg/mL with 1X PBS pH 7.4 if necessary. aSyn was shaken at 37 ^oC, 1000 rpm for 7 days using a Thermomixer C with a heated lid. Final wild-type human aSyn PFFs were aliquoted and frozen at −80 ^oC.

Characterization of α-synuclein PFFs in vitro

Sedimentation assay

PFFs were centrifuged at 20,000 x g at room temperature for 30 min. The supernatant was removed and the pelleted PFFs were washed in the same volume of 1X PBS pH 7.4, followed by a re-centrifugation. The wash was removed and final pellet of PFFs were re-suspended in 1X PBS pH 7.4. Supernatant, wash, and pellet samples were re-suspended in loading buffer containing SDS and β-mercaptoethanol and run at 200 V for 50 min on a 12% Tris-Glycine SDS-PAGE gel and stained with Coomassie Brilliant Blue G.

Transmission electron microscopy

Samples were prepared for examination in the transmission electron microscope using the ‘direct application method’ [57]. Negative stain transmission electron microscopy images of aSyn PFF-1 and PFF-2 were acquired at 80 Kv on carbon coated 400 mesh copper grids using phosphotungstic acid and uranyl acetate stain.

Thioflavin-T monomer seeding assay

Monomers (100 μM), PFFs (10 μM), or both were shaken at 600 rpm at 37 ^oC in a sealed Lumox 96-well multiwell plate (Sarstedt Cat# 94.6000.024) for 48 h in the presence of 25 μM ThT. Fluorescence was measured at regular intervals on a Molecular Devices Gemini XPS Microplate reader using Softmax Pro Software version 6.5.1 at an excitation wavelength of 450 nm and an emission wavelength of 485 nm.

Far UV circular dichroism

Samples of both aSyn PFF-1 and PFF-2 were initially sonicated 10 cycles at high frequency using a Bioruptor Pico to ensure solubility yet still maintain fibrillar structure. All samples were diluted with 1X PBS buffer to 1.6 mg/mL from 2.0 mg/mL for far-UV analysis. The auto-sampler temperature was set to 20 °C for sample storage prior to injection into the smart cell for analysis at 20 °C. Samples were analyzed on the Chirascan Q100 with monochromator wavelength set at 260.0 nm, bandwidth of 1.0 nm and a step of 0.5 nm, sampling 2 time points per second. Sample spectra were collected in triplicate, averaged, normalized, smooth and converted into CD units of ∆ε. The CDNN algorithm was used to fit the far-UV CD data for prediction of secondary structure.

Characterization of α-synuclein PFFs in vivo

Animals

M83 hemizygous mice (M83 + /− mice; B6; C3H-Tg[SNCA]83Vle/J) were obtained from Jackson Laboratory (004479) and maintained in-house on a C57BL/C3H background. These transgenic mice express one copy of the human aSyn transgene array bearing the familial A53T mutation in the SNCA gene under the control of the mouse prion protein promoter, along with endogenous mouse aSyn [58].

Male and female M83 hemizygous (+/−) and littermate controls (−/−), aged 3–6 months, were pseudorandomly assigned to receive intrastriatal injections of wild-type human aSyn PFF-1, PFF-2, or PBS via stereotaxic surgery, with littermates from the same cage distributed across treatment groups to minimize selection and litter-related bias. The distribution of ages was balanced across groups to ensure comparability. For details, refer to the section on stereotaxic surgery for α-synuclein inoculation in the supplementary information.

Experimental design

Subjects were housed in 28 × 18 cm plastic shoebox cages at 22-23 °C, 50 ± 10% humidity with rise tops, and a 12:12 h reverse light-dark cycle (lights turned off at 09:00). Food and water were provided ad libitum until 1 week prior to behavioural testing, at which point mice were food restricted to 85–90% of their free-feeding body weight. Daily food rations were provided with pellets commercially available from Bio-Serv in Flemington, New Jersey (0.5 g Cat#F0171; 1 g Cat#F0173). Experiments were performed during the dark cycle (between 09:00-18:00) 5-7 days/ week. Mice were single housed following stereotaxic surgery to prevent infection of the surrounding incision area.

Automated touchscreen cognitive assessment

Apparatus

Cognitive testing was conducted within automated Bussey-Saksida Mouse Touchscreen Systems Model 80614-20 (Lafayette Instruments, Lafayette, IN). Experiments were carried out inside sound-attenuating cabinets which consist of a standard operant chamber and a touch-sensitive 12.1-inch monitor (screen resolution 600 ×800), as previously described [59]. Throughout testing, the touchscreen was permanently covered by a black Plexiglas 3-holes mask, which limited subjects’ interaction with the screen to relevant task locations (three 7x7cm square windows). A liquid reward dispensing magazine was situated at the back of the chamber and linked to a liquid reward dispenser pump (reinforcer used: Strawberry milkshake from Neilson Dairy). A light emitting diode (LED) illuminated the food magazine during reward delivery and a tone generator triggered auditory tones used throughout touchscreen training. Animal activity was recorded via infrared photobeams located at the front and back of the chamber and at the entry to the reward magazine. The schedule design, control of the apparatus via Whisker control system and data collection was controlled by ABET II Video Touch software V21.02.26.

Touchscreen habituation and pre-training

Between 2-8 weeks post-injection (WPI), mice were habituated to the touchscreen apparatus. The stages of acclimation and appetitive conditioning is freely available within the standard operating procedure at touchscreencognition.org. In brief, mice were assigned to a specific touchscreen operant chamber and gradually exposed to the apparatus, the reinforcer, and the tone. Subsequently, they learnt to interact with specific areas of the screen to receive a reward and initiate new trials by nose poking the reward receptacle, with each pretraining stage repeated until mice passed a pre-set criterion.

Visuomotor conditional learning (VMCL) task

The touchscreen VMCL task is freely available at 10.58064/ceae-s252 and was run as previously described [49, 50]. In brief, sessions began with the delivery of a reinforcer (800 ms pulse feed time, 20 µL) in the reward receptacle. Upon exiting of the reward tray, one of two discriminatory stimuli (white icicle (image A) or grey equal sign (image B)) was presented in the centre window. Through trial and error, subjects learnt that presentation image A required an immediate nose poke to the right flanking window for a correct response and delivery of reward, while the presentation of image B required a nose poke to the left flanking window. Both discriminatory stimuli were presented equally throughout the session (15 trials of each, for a total of 30 new trials maximum), and were presented pseudo-randomly so that no image was presented more than 3 times in a row. The time to respond to the discriminatory stimuli was uncapped, but the time to respond to the left or right flanking window was limited to 5-seconds to engender fast and automatic S-R learning behaviour.

A correct response was associated with a light in the reward receptacle, the delivery of 20uL of reinforcer, and a 1000 ms tone, followed by a 20-second intertrial interval (ITI). An incorrect response, or a failure to respond within 5-second limited hold (termed a ‘missed’ trial) led to a time-out period, indicated by a 5-second illumination of the house light, followed by an ITI. To counteract the development of side biases and ensure that subjects received a consistent number of rewards per session, any trial immediately succeeding a time out period was designated a ‘correction trial’ in which the same image was re-presented. Correction trials were consecutively presented until a correct choice was made. The VMCL task involved a maximum of 30 new (non-correction) trials in each maximum 60-minute session for 20 sessions (5-7 sessions per week, maximum of 1 session per day).

The primary outcomes of VMCL were (a) accuracy (percentage of correct responses): $\frac{\mathrm{\#}\mathit{correct\; trials}}{(\mathit{\#total\; trials}-\mathrm{\#}\mathit{missed\; trials})}$, (b) Percentage of missed trials: $\frac{\mathrm{\#}\mathit{missed\; trials}}{\mathit{\#total\; trials}}$, (c) number of correction trials, (d) perseveration index: $\frac{\mathrm{\#}\mathit{correction\; trials}}{\mathit{\#incorrect\; trials}}$, which uses the number of correction trials as an indication of perseverative behaviour, corrected for the number of correction trials that would be expected based on performance (as poorly performing animals require more correction trials). Importantly, incorrect trials are defined as errors made on the first choice of a new trial; errors occurring during correction trials are not counted as incorrect trials. (e) correct touch latency: time from nose-poking the discriminatory stimulus in the central window to nose poking the correct flanking window), (f) incorrect touch latency: time from nose-poking the discriminatory stimulus in the central window to nose poking the incorrect flanking window, and (g) reward collection latency: time to nose poke into the reward magazine after giving a correct response on the touchscreen. All data are presented as the mean + $\mathrm{SEM}$, and plotted in 5 blocks of 4 sessions.

Motor function battery

Various motor assessments, including forelimb grip force, accelerating rota-rod, and CatWalk XT gait analysis, were conducted at 8 WPI (prior to cognitive testing) and at 16 WPI (following cognitive testing) to evaluate the order of symptom development. The experimenter was blind to the experimental condition during motor assessments, but not during data analysis. More details about the assessment of different motor functions can be found in the supplementary information section.

Tissue collection and processing

At 16 + WPI, tissue was collected either for whole brain histology, including immunofluorescence staining, or dissected and stored −80 °C for biochemistry including western blotting, proteinase K digestion assay, and dot blot assay. For histology, mice were anesthetized with ketamine (100 mg.kg⁻¹)-xylazine (20 mg.kg⁻¹) and transcardially perfused with ice-cold phosphate-buffered saline (PBS) followed by 4% paraformaldehyde (PFA). Brains were extracted and kept overnight in 4% PFA at 4 °C, and then transferred into 30% sucrose solution where they stayed until they were flash-frozen on dry-ice in Tissue-Tek O.C.T. Compound (Sakura, Cat. #4583) for long-term storage at −80 °C. Tissue was sliced on a cryostat (Leica CM1950s) in 25 µm coronal sections. A subset of sections was reserved for immunofluorescence analyses used in this study. For biochemistry, brains were collected from mice cervically dislocated and rapidly dissected on ice, flash frozen, and store at −80 °C. More details about the above procedures are described in the methods section of supplementary information.

Data analysis

Analyses were conducted on GraphPad Prism 9 for macOS Version 9.5.1. Shapiro-Wilk test of normality and the Rout method for identifying outliers (Q = 0.1%) were conducted prior to each analysis. If normality failed, data were transformed prior to statistical testing [60]. Differences between two means were examined using paired or unpaired t-tests. When examining 3 or more comparisons, tests of analysis of variance (ANOVA) were employed (one- or two- way with repeated measures), followed by Šídák’s multiple comparisons if a significant interaction was revealed. Alternatively, if a dataset had missing values due to the identification of outliers, we compared groups using a repeated measures mixed-effects ANOVA model. If the assumption of homogeneity of variances was violated, Welch’s ANOVA was used instead, followed by unpaired t-tests with Welch’s correction for post hoc comparisons. Sample sizes were estimated using a partial eta squared (ηp 2) power analysis for repeated measures two-way ANOVA (power = 0.8, ɑ = 0.05) [61]. We acknowledge the importance of investigating potential sex differences, and although male and female mice were tested in a balanced design [62], they were combined in statistical analyses due to insufficient power to detect sex differences [63]. Analyses where p values (or adjusted p values) < 0.05 were considered significant.

Results

Generation and characterization of recombinant α-synuclein pre-formed fibrils in vitro

To examine the effect of aSyn pathology on S-R learning, we first characterized two human recombinant aSyn PFFs in vitro, generated by changing the time and temperature for protein expression in E. coli. The first fibril type, referred to as PFF-1, was generated from monomers expressed in E.coli at room temperature overnight. The second fibril type, referred to as PFF-2, was generated from monomers expressed in E.coli at 37 ^oC for four hours (Fig. 1A). Although fibrils were shaken under the same conditions (1000 rpm at 37 ^oC), they displayed distinct biophysical properties when analysed in vitro. Both fibril types were mainly insoluble and appeared as elongated fibrils of mixed lengths under transmission electron microscopy (Fig. 1B, C). Far UV circular dichroism measurements indicated a high β-sheet content in both fibrils compared to disordered monomers, but with some small differences: lower β-sheet/turn content (40.4% vs 43.0%) and random coil content (31.4% vs 33.4%), but higher α-helix content (23.7% vs 19.6%) was revealed in PFF-1 relative to PFF-2 (Fig. 1D). Differences were observed in Thioflavin-T (ThT) binding between the preparations, with aSyn PFF-1 triggering a stronger ThT signal both after fibrilization and during a seeding assay with additional monomers (Fig. 1E, F). Taken together, despite originating from the same monomeric source expressed in E.coli under different conditions, differences in seeding capability and secondary structures support the notion that PFF-1 and PFF-2 may represent distinct conformational states of aSyn PFFs in vitro.

Fig. 1. Generation and characterization of recombinant human α-synuclein pre-formed fibrils in vitro.

A Schematic of α-synuclein pre-formed fibril production. α-synuclein type 1 pre-formed fibrils (PFF-1) were generated from monomers expressed in E.coli at room temperature overnight, while α-synuclein type 2 pre-formed fibrils (PFF-2) were generated from monomers expressed in E.coli grown at 37 °C for four hours. Following this, both fibril precursors underwent monomer purification and identical fibrilization protocols and were characterized in vitro. B SDS-PAGE of PFF-1 and PFF-2 after sedimentation revealed similar populations of insoluble (pellet) vs soluble (supernatant) materials, both consisting of predominantly insoluble species. C Transmission Electron Microscopy (TEM) of unsonicated PFF-1 and PFF-2 displayed similar architecture and range of fibril sizes, both appearing as elongated fibrils of mixed length. D Far UV circular dichroism (UV-CD) measurements identified a significant β-sheet presence in fibrils compared monomers with subtle differences between PFF-1 and PFF-2s. Thioflavin T (ThT) binding assay revealed a stronger binding capacity of PFF-1 (E) relative to PFF-2 (F) after fibrilization and during a seeding assay with additional monomer.

Inoculation of α-synuclein pre-formed fibrils in M83 mice in vivo

To evaluate the impact of these distinct aSyn PFFs in vivo, we inoculated 3–6-month-old hemizygous (+/-) M83 male and female mice (Fig. 2A). These mice possess one copy of the A53T-mutant human aSyn transgene array and were chosen as they (a) refrain from exhibiting phenotypes until 20 months of age [42, 64], a timepoint far exceeding our experimental timeline, and (b) can propagate human recombinant aSyn [65–67]. Equal amounts (12.5 µg) of PFF-1 or PFF-2 were inoculated unilaterally into the right dorsal striatum. As negative controls, M83 + /- and M83-/- mice were injected with an equivalent volume of phosphate-buffered saline (PBS) solution. None of the PBS-inoculated mice developed discernable phenotypes at any point during the experiment, and direct comparisons revealed no significant differences in behaviour (Supplementary Fig. 1). For this reason, control groups were combined for behavioural testing. Furthermore, we confirmed that the age distribution was balanced across groups to ensure comparability. We also statistically verified that starting age had no effect on subsequent behavioral performance, as assessed by regression analysis of percent correct accuracy (Supplementary Fig. 2).

Fig. 2. Experimental design for inoculating α-synuclein pre-formed fibrils in M83 mice in vivo.

A M83 mice underwent stereotaxic surgery for inoculation of aSyn PFF-1, aSyn PFF-2 or PBS control into the right dorsal striatum. At 8 weeks post injection, they subsequently underwent a motor test battery, including assessments of motor strength, motor coordination and gait. Following completion, they were assessed for the cognitive ability to acquire stimulus-response associations, with the ‘Visuomotor Conditional Learning’ at 9-12 weeks post injection. At 16 weeks post-injection, they repeated the motor test battery, and then were processed for pathology and biochemistry. B Brain homogenates from aSyn PFF-1 and PFF-2 inoculated mice both displayed resistance to proteinase K but exhibited variable banding patterns upon digestion, as assessed by Western Blot using an antibody for total aSyn. C, D Furthermore, the same brain homogenates were evaluated by dotblots using an antibody with high affinity for aggregated aSyn (chBIIB054), revealing that while both fibril types exhibited significant binding, aSyn PFF-2 brain homogenates presented with a stronger signal (Welch’s ANOVA test, W_{2, 5.34} = 7.03, p < 0.05, in which Welch-corrected unpaired t-tests revealed to be driven by a significant difference between PBS control and PFF-2 (p < 0.05). Graphics made with Biorender.com.

To evaluate if PFF-1 and PFF-2 would show different biochemical properties upon inoculation in vivo, we conducted a comparative analysis of their susceptibility to proteinase K digestion. Sensitivity to proteinase K cleavage typically indicates an unfolded and/or denatured structure of aSyn, whereas resistance suggests a more compact and aggregated conformation [68]. Critically, the pattern of peptide fragments generated by proteinase K digestion can also serve to discriminate conformational strains [37, 69]. In support of aSyn PFF-1 and PFF-2 potentially existing as distinct conformational strains in vivo, we found that brain homogenates from aSyn-inoculated mice of both fibril types were resistant to proteinase K but exhibited somewhat distinct banding patterns upon digestion (Fig. 2B). While PFF-1 digestion generally produced 1 digestion band, PFF-2 digestion often generated 2 stronger bands. Furthermore, while both fibril types exhibited significant binding to chBIIB054, an antibody with high affinity for aggregated aSyn [70], aSyn PFF-2-seeded brain homogenates presented with a stronger signal following dotblot analyses (Fig. 2C, D). Taken together, these results suggest that aSyn PFF-1 and PFF-2 may represent distinct aSyn strains upon inoculation in vivo.

Propagation of aggregated α-synuclein following inoculation in vivo

Relative to controls, both aSyn PFF-1- and PFF-2-inoculated mice also exhibited significant pS129 pathology at 16 weeks post-injection (WPI) (Fig. 3). Similar to previous studies [35, 71, 72], immunohistochemical analyses revealed significant aSyn pS129 inclusions in regions that encompassed, but were not limited to, the injection site (dorsal striatum), the orbitofrontal cortex, piriform cortex, medulla and substantia nigra (Fig. 3C). Qualitatively, inclusions and staining were similar between PFF-1 and PFF-2. To quantify immunoreactivity for pS129 further, detergent-soluble and insoluble aSyn fractions of protein extracts were taken from the striatum (Fig. 3D–I) and cortex (Supplementary Fig. 3) and quantified with immunoblots. pS129 levels from mice inoculated with PFFs were expressed as the percentage change from the levels found in mice inoculated with PBS (controls). However, pS129 levels in cortex and striatum did not differ significantly between PFF-1 and PFF-2 (Fig. 3E, F), as did human aSyn levels (Fig. 3H, I). Together, this work demonstrates that while both types of fibrils induced aSyn pathology, no differences were observed between them with respect to pS129 levels.

Fig. 3. Propagation of aggregated α-synuclein following inoculation of pre-formed fibrils in vivo.

A aSyn PFF-1, PFF-2 or PBS control was unilaterally injected into the right dorsal striatum. Brains for immunohistochemistry (B, C) and immunoblots (D-I) were collected at 16 weeks post injection. B Immunostained brain slices imaged at 63X on a Leica Stellaris 5 Confocal Microscope for aSyn phosphorylated at S129 (green= pS129; blue= Hoescht), scale bar 25 µm. C Although inoculation was localized within the dorsal striatum, pathology spread to many areas including orbitofrontal cortex, piriform cortex, medulla, and substantia nigra. 40X, Leica Stellaris 5 Confocal Microscope, scale bar 50 µm. D–I To quantify this pathology further, protein extract from the striatum of PFF- and PBS-inoculated mice were extracted and quantified with immunoblots. D–F The levels of pS129 and total human aSyn found in the detergent-soluble (RIPA) fractions from PFF-1 and PFF-2 samples were expressed as the percentage fold change relative to PBS samples. No significant differences were found between PFF-1 and PFF-2 when comparing pS129 (unpaired t-test, p > 0.05) or human aSyn (unpaired t-test, p > 0.05). G–I The levels of pS129 and total human aSyn found in the detergent-insoluble fractions (UREA solubilization of RIPA-insoluble pellets) were expressed in the same way as the detergent-soluble fractions (RIPA). No significant differences were found between PFF-1 and PFF-2 when comparing pS129 (unpaired t-test, p > 0.05) or human aSyn (unpaired t-test, p > 0.05). Graphics made with Biorender.com.

Inoculation of α-synuclein PFF-2, but not PFF-1, induce motor deficits within 16 weeks

To track the progression of symptom development following aSyn inoculation, we then compared motor function prior to (8 WPI) and after cognitive testing (16 WPI). These experiments revealed that aSyn PFF-2-inoculated mice, but not PFF-1-inoculated mice, presented with severe motor deficits in the time frame of the experiment (Fig. 4). For example, the forelimb grip force of aSyn PFF-1 (N = 25, n = 11♂, n = 14♀), PFF-2 (N = 19, n = 7♂, n = 11♀), and PBS-control mice (N = 30, n = 17♂, n = 13♀) was compared. While no differences were revealed between time points for PBS controls and aSyn PFF-1-inoculated mice, aSyn PFF-2 mice exhibited a significant decline in forelimb grip strength at 16 WPI relative to 8 WPI (Fig. 4A, B). Similarly, when examining the latency to fall on an accelerating rota-rod between groups, PFF-2-inoculated mice presented with a significantly shorter latency to fall compared to PFF-1 and PBS control mice at 16 weeks, but not 8 weeks (Fig. 4C–E).

Fig. 4. Inoculation of aSyn PFF-2, but not PFF-1, induce motor deficits within 16 weeks.

A Schematic of the forelimb grip force test, used to measure muscle strength. Subjects grasped a smooth, triangular pull bar with both forelimbs and the force exerted in Newtons (N) was measured. B aSyn PFF-2 but not PFF-1 inoculated mice displayed a significant reduction in forelimb muscle strength at 16 weeks post injection (WPI) relative to 8 (2-way RM ANOVA, Group x Session: main effect of group (F_2,72 = 4.409, p < 0.05), main effect of session (F_1,72 = 13.51, p < 0.001), and significant interaction (F_2,72 = 6.677, p < 0.01), in which Šídák’s multiple comparisons revealed to be driven by a significant difference within between 8 and 16 weeks within the PFF-2 group (adj. p < 0.0001). C Schematic of the accelerating Rota-rod, used to measure motor coordination and motor learning. Subjects were placed on a stationary rod which began to accelerate linearly from 5-36 revolutions per minute over 5 min, and latency to fall (s) was calculated automatically. D No significant difference between groups was found at 8 WPI (2-way RM ANOVA, Group x Session: main effect of session (F_9,675 = 15.41, p < 0.0001), but no main effect of group (p > 0.05). A significant interaction was first observed, (F_{18, 675} = 1.767, p < 0.05), but Šídák’s multiple comparisons revealed no significant differences between groups (adj. p > 0.05)). However, (E) aSyn PFF-2 mice exhibited a shorter latency to fall at 16 WPI relative to aSyn PFF-1 and PBS controls (2-way RM ANOVA, Group x Session: Main effect of group (F_2,71 = 7.436, p < 0.01), main effect of session (F_9,639 = 3.205, p < 0.001), but no interaction (p > 0.05). To examine this main effect further, Šídák’s multiple comparisons revealed a significant difference between aSyn PFF-2 and PBS controls (adj. p < 0.01) and aSyn PFF-2 and PFF-1 (adj. p < 0.01)). F Schematic of the catwalk XT Gait Analysis, used to measure gait and locomotion. Footprints were captured while subjects voluntarily traversed a glass-plated CatWalk runway, towards a dark goal box at the end. G A significant difference between 8 and 16 weeks was revealed in the stride length of aSyn PFF-2 inoculated mice (2-way RM ANOVA, Group x Session: no main effect of group (p > 0.05), but a main effect of session (F_1,24 = 4.493, p < 0.05), and significant interaction (F_{2,24= 3.646}, p < 0.05), in which Šídák’s multiple comparisons revealed a significant difference between timepoints within the aSyn PFF-2 group (adj. p < 0.01). H A significant difference between 8 and 16 weeks was revealed in the swing speed of aSyn PFF-2 inoculated mice (2-way RM ANOVA, Group x Session: no main effect of group (p > 0.05), but a main effect of session (F_{1,24= 6.537}, p < 0.05), and significant interaction (F_{2,24= 4.300}, p < 0.05) in which Šídák’s multiple comparisons revealed a significant difference between timepoints within the aSyn PFF-2 group (adj. p < 0.01). I A significant difference between 8 and 16 weeks was revealed in the step cycle of aSyn PFF-2 inoculated mice (2-way RM ANOVA, Group x Session: no main effect of group (p > 0.05), but a main effect of session (F_{1,24= 9.232}, p < 0.01) and significant interaction (F_{2,24= 4.035}, p < 0.05) in which Šídák’s multiple comparisons revealed a significant difference between timepoints within the aSyn PFF-2 group (adj. p < 0.01). Data presented as Mean + SEM, group x session Two-way RM ANOVA, ** p < 0.01, **** p < 0.0001. Graphics made with Biorender.com.

Furthermore, we also tested a subset of subjects on the Catwalk XT for gait analysis (Fig. 4F–I). aSyn PFF-1 (N = 9, n = 6♂, n = 3♀), PFF-2 (N = 8, n = 6♂, n = 2♀), and PBS-control M83 (+/−) mice (N = 10, n = 6♂, n = 4♀) were assessed as they freely traversed the catwalk runway, and automated gait calculations for stride length (Fig. 4G), swing speed (Fig. 4H) and step cycle (Fig. 4I) were compared. Consistent with the patterns of motor dysfunction described above, aSyn PFF-2 inoculated mice, but not PFF-1 or PBS control mice, displayed abnormal gait at 16 WPI relative to 8 WPI (Fig. 4G–I). These data therefore demonstrate severe motor dysfunction at 16 WPI for aSyn PFF-2-inoculated mice. In stark contrast, however, aSyn PFF-1-inoculated mice displayed motor strength, coordination, and gait equivalent to PBS controls throughout the entirety of experimental timeline. Taken together, this work suggests that aSyn PFF-1 and PFF-2 strains progress at different rates to the final stages of disease.

Inoculation of both α-synuclein PFF-1 and PFF-2 induce stimulus-response learning impairments prior to motor dysfunction

Finally, we directed our attention towards assessing cognitive function during the interval between motor assessments (9-12 WPI), preceding the presentation of major motor impairments. The VMCL task is a well-established instrumental paradigm which assesses the acquisition and expression of S-R contingencies through the repeated pairing of visual stimuli (presented on a touch-sensitive screen) with arbitrary motor responses (nose pokes to the right- or left-flanking positions). Designed to be solvable by establishing distinct S-R associations, rather than action-outcome strategies or by simple pavlovian conditioning processes alone [54], it has previously been shown to be dependent on the striatum [53, 54], cingulate cortex [51], and nigrostriatal dopamine [55], all of which are highly disrupted in patients with synucleinopathies.

At 9-12 WPI, mice inoculated with aSyn PFF-1 (N = 25, n = 11♂, n = 14♀), PFF-2 (N = 23, n = 11♂, n = 12♀), and PBS-control mice (N = 30, n = 14♂, n = 16♀) were tested on the VMCL task (Fig. 5). Relative to PBS controls, both aSyn PFF-1- and PFF-2-inoculated mice were significantly impaired at acquiring the VMCL task, reflected in a 10–20% reduction in their final % correct accuracy in block 5 (Fig. 5B). With chance level performance at 50% (Fig. 5C) and the highest controls levels at ~80% (Fig. 5D), this reduction represents about 30-60% of the behavioural range. These mice also performed a greater number of correction trials, achieved a higher perseveration index, and took longer to make incorrect choices, a pattern of behaviour that may suggest a decreased sensitivity to negative feedback (Fig. 5F–I). PFF-inoculated mice were not globally impaired, however no significant differences between groups were found in the percent missed trials or the latencies to make correct choices and collect reward (Fig. 5E, H, and J), indicating that motivation and general activity were not affected. Taken together, these results suggest that despite behaving as different strains of aSyn, both PFFs induced a robust cognitive deficit, indicating a compromised ability to learn associations between stimuli and responses, with a pattern of behaviour which may suggest abnormal processing of feedback. Furthermore, given the variable level of impact on motor function observed within our experimental timeline, this work also highlights the sensitivity of VMCL task to detect cognitive deficits in prodromal disease stages under variable misfolding conditions.

Fig. 5. Inoculation of both α-synuclein pre-formed fibrils induce severe stimulus-response learning cognitive impairments in M83 mice.

A Schematic of Visuomotor Conditional Learning (VMCL) task. Cognitive assessments were conducted within touchscreen systems equipped with a touch-sensitive screen, a reward magazine attached to a reward pump for delivery of strawberry milkshake liquid reward and ABET cognition software (above). VMCL was employed to evaluate the acquisition of stimulus-response (S-R) contingencies, with subjects learning the conditional rule: “if visual stimulus A is presented, make motor response to the right-flanking window; if visual stimulus B is presented, make motor response to the flight-flanking window” (below). All subjects underwent VMCL testing for 20 sessions, 5-7 sessions per week. B M83 mice inoculated with aSyn PFF-1 and PFF-2 were significantly impaired at acquiring VMCL, as demonstrated by lower percent correct responses relative to PBS-inoculated mice (2-way RM ANOVA, Group x Session: main effect of group (F_2,60=10.53, p < 0.001), main effect of session (F_{2.686, 161.2=10.53}, p < 0.0001) and significant interaction (F_8,240=3.131, p < 0.01), in which Šídák’s multiple comparisons revealed to be driven by a significant difference between PBS control and PFF-1 in sessions 2 (adj. p < 0.05), 4 (adj. p < 0.05) and 5 (adj. p < 0.05), and a significant difference between PBS control and PFF-2 in sessions 2 (adj. p < 0.01), 3 (adj. p < 0.001), 4 (adj. p < 0.01), and 5 (adj. p < 0.01)). C While all groups began at chance in Block 1 (1-way ANOVA, p > 0.05), (D) both aSyn-inoculated groups were significantly impaired relative to controls in Block 5 (1-way ANOVA, F_8.764, p < 0.001, in which Šídák’s multiple comparisons revealed to be driven by a significant difference between PBS controls and PFF-1 (adj. p < 0.05), and between PBS controls PFF-2 (adj. p < 0.001)). E No significant difference was found in the percentage of missed trials across groups (RM Mixed-Effects Model, Group x Session: main effect of session (F_{2.050,118.4= 20.55}, p < 0.0001), but no main effect of group or interaction (p > 0.05)), (F) but aSyn PFF-1 and PFF-2 mice exhibited a greater number of correction trials (2-way RM ANOVA, Group x Session: main effect of group (F_{2,60= 6.008}, p < 0.01), main effect of session (F_{3.240,194.4= 66.22}, p < 0.0001) but no interaction (p > 0.05)), and (G) an elevated perseveration index compared to controls (RM Mixed-Effects Model, Group x Session: main effect of group (F_2,60=4.648, p < 0.05), main effect of session (F_{3.191,171.5=24.48}, p < 0.0001), but no interaction (p > 0.05)). Furthermore, comparing task latencies, no significant difference was found in the latency to make correct choices (H) (2-way RM ANOVA: Group x Session: p > 0.05), but aSyn PFF-1 and PFF-2 mice took significantly longer to make incorrect choices (I) across VMCL acquisition compared to PBS controls (2-way RM ANOVA: Group x Session: main effect of group (F_2,60=5.384, p < 0.01), main effect of session (F_{3.202,192.1= 62.37}, p < 0.0001), and significant interaction (F_8,240=1.985, p < 0.05), in which Šídák’s multiple comparisons revealed to be driven by a significant difference between PBS control and PFF-1 in sessions 2 (adj. p < 0.01), 3 (adj. p < 0.01) and 5 (adj. p < 0.05), and a significant difference between PBS control and PFF-2 in sessions 3 (adj. p < 0.05) and 5 (adj. p < 0.05)). No significant difference was found for the latency to collect rewards (J) (2-way RM ANOVA, Group x Session: main effect of session (F_{2.275,135.9=17.67}, p < 0.0001), but no main effect of group or interaction (p > 0.05)). Data presented as Mean + SEM, * p < 0.05, ** p < 0.01, *** p < 0.001. Graphics made with Biorender.com.

Discussion

Our primary goal was to test the validity of a cognitive test for detecting impairments in stimulus-response learning, a domain highly relevant to synucleinopathies. Translating findings from tests of cognitive function relevant to synucleinopathies from preclinical to clinical settings has proven difficult, partly due to the dissimilar and poorly replicable cognitive testing methods commonly used in preclinical studies. Moreover, most cognitive tests used in models of neurodegenerative diseases fail to test cognitive domains that are relevant for human disease. Here, we began to address these issues within the context of S-R learning, a highly relevant cognitive function commonly disrupted in patients with synucleinopathies [44–47, 73], which relies on vulnerable cortico-basal ganglia- thalamic brain circuits [33, 53, 74].

Considerable clinical and pathological heterogeneity exists within and among synucleinopathies, which has been linked to the existence of distinct conformational strains of pathological aSyn that cause different disease manifestations [37, 41]. To address this, we generated two distinct strains of human recombinant aSyn PFFs by varying preparation conditions. Both fibril types shared key misfolded protein characteristics, including insolubility and high β-sheet content. However, differences in their secondary structures and ThT binding capacity in vitro, suggest that the preparation methods yield fibrils with distinct biophysical properties [75].

To evaluate the impact of aSyn PFFs in vivo, we subsequently injected them unilaterally into the dorsal striatum of M83 mice hemizygous for the human aSyn A53T mutation [58]. Although these mice do not display any noticeable phenotypes until over 20 months of age [65, 67, 76], inoculating them at 3-5 months resulted in hyperphosphorylation of intraneuronal aSyn accumulations resembling inclusions, along with proteinase K resistance and elevated binding with the chBIIB054 antibody [70]. These results align well with those previously observed using the aSyn PFF in the M83 hemizygous mouse model [34, 35, 77], and support the ability of aSyn PFFs to induce robust aSyn pathology relevant to synucleinopathies. Notably, however, we also observed somewhat distinct banding patterns following proteinase K digestion, suggesting that these aSyn PFFs may have distinct characteristics upon inoculation in vivo, potentially leading to varying levels of pathology in rodents.

Our motor and cognitive assessments, which followed previous experimental designs [34, 35], provided insights into how distinct aSyn PFF strains may differentially impact neurological function. This approach enabled us to track the progression of symptom onset, considering that cognitive deficits may precede motor dysfunction in patients with synucleinopathies [78]. The observed phenotypic variability, particularly the motor dysfunction seen with aSyn PFF-2 but not PFF-1, highlights the potential for strain-specific pathological effects. It is important to note that the M83 hemizygous model should not be interpreted as a staging model for Parkinson’s disease, but rather as a model of synucleinopathy. In line with this, prior work using microPET imaging has shown no significant changes in dopamine transmission in these mice [79], and motor deficits are likely driven by high levels of α-synuclein overexpression in the spinal cord [80]. Therefore, while dopaminergic mechanisms may modulate some behavioural outcomes, spinal cord pathology likely plays a primary role in the motor phenotype observed here. We acknowledge that our interpretation of cognitive impairments is not directly linked to a specific pathological stage. Future work combining behavioural, and pathological measures will be important for defining disease phases more objectively in this model. These findings support the notion that variations in aSyn PFF structure can drive distinct outcomes, potentially mirroring the heterogeneity seen in synucleinopathies. Moreover, the association between higher aSyn levels and motor impairments in PFF-2-inoculated mice underscores the importance of understanding strain-specific differences when modeling disease progression [37, 38, 41]. Our findings are consistent with prior studies showing that α-synuclein pathology can disrupt striatal synaptic plasticity and impair both motor and cognitive functions, including visuospatial and associative learning, in rodent models [81–83]. Notably, differences in the site and form of α-synuclein expression have been linked to distinct patterns of symptom onset and synaptic dysfunction, including early memory deficits following hippocampal α-syn overexpression [84].

Lastly, cognitive testing was assessed using a touchscreen apparatus, a method widely recognized for its efficacy and replicability [3, 4]. This method enables the implementation of behavioural paradigms in rodents that closely mimic those used in clinical settings, facilitating high cross-species translation [3, 48–50, 85]. To study S-R learning preclinically, we developed the VMCL task [48–50], which has recently been shown to rely on brain circuits and neuromodulators vulnerable in synucleinopathies including the striatum [53, 74], cingulate cortex [51], and nigrostriatal dopamine [55]. Importantly, S-R learning is highly conserved between rodents and humans [86–89] and is commonly impaired in patients diagnosed with synucleinopathies [44–47, 73], although this relationship can be influenced by medication state [90] and the severity of the disease [73, 91].

Notably, we observed significant impairments in acquiring the VMCL task in both aSyn PFF-1 and PFF-2 inoculated mice, including impaired reduction in incorrect choice latencies across learning (Fig. 5H), and patterns of behaviour that may suggest abnormal processing of negative feedback, without changes in motivation or motor function as measured by correct choice and reward latencies. This latter finding suggests that the observed impairments were due to learning the task itself, although we cannot definitively rule out an explanation in terms of performance of the task (an impairment in decision-making at choice, despite intact but unexpressed learning of the rule). The data clearly indicate that the VMCL task is sufficiently sensitive to detect cognitive impairments in this mouse model, prior to onset of detectable motor dysfunction, in both protein misfolding conditions tested. Furthermore, the idea of altered feedback in these models chimes with recent studies that have demonstrated altered feedback learning in PD patients [92–94], where the processing of positive and negative feedback is influenced by dopaminergic transmission [93]. Further investigation is needed to determine whether abnormal dopaminergic transmission mediates the cognitive deficits observed following aSyn inoculation.

Collectively, while our findings highlight the potential influence of distinct α-synuclein PFF strains on cognitive and motor outcomes, the primary contribution of this study is in validating the translational potential of the VMCL task for assessing cognitive deficits associated with synucleinopathy.

Together, through the comparison of two distinct aSyn PFFs, we highlighted the efficacy of the aSyn PFF in the M83 hemizgous mouse model combined with the use of translational touchscreen cognitive assessments in the study of S-R learning relevant to synucleinopathies. Despite evidence that our two fibril types displayed distinct characteristics in vitro and in vivo, both induced a severely compromised ability to make associations between stimuli and responses, occurring prior to the onset of motor dysfunction, and thus replicating clinical conditions [5]. Therefore, the combination of robust animal models and appropriate battery of cognitive touchscreen testing can be successfully deployed for the development of much-needed target identification and therapeutic strategy testing to improve the cognitive function of patients diagnosed with devastating synucleinopathies.

Author contributions

Oren Princz-Lebel, Lisa M. Saksida, Marco A.M. Prado, Vania F. Prado and Timothy J. Bussey designed the experiments. Oren Princz-Lebel, Jacob A. McPhail, Ariel Louwrier, Anoosha Attaran, Rodrigo Sandoval Contreras, Vladislav Novikov, Samina Panjwani, Anthony Chu, Cherisse Tan, Claire A. Lemieux, Miguel Skirzewski and Meira M.F. Machado performed experiments. Oren Princz-Lebel, Marco A.M. Prado and Timothy J. Bussey wrote the manuscript, with the support and review of all authors.

Data availability

Data are available in the following repositories: motor behaviours and biochemical and quantifications- https://figshare.com/account/mycontent/projects/246104. Visuomotor Conditional Learning (VMCL) touchscreen task are available in the MouseBytes repository, at: https://mousebytes.ca/data-link?linkguid=6bfd7a16-52d4-475d-8f2a-e490e601d541 (10.58064/qh7h-8w97).

Competing interests

Jacob A. McPhail generated and characterized the recombinant alpha-synuclein pre-formed fibrils while an employee of StressMarq Biosciences, and Ariel Louwrier is the founder of StressMarq Biosciences. The authors declare that they have no other competing interests. Timothy J. Bussey and Lisa M. Saksida have established a series of targeted cognitive tests for animals, administered via touchscreen with a custom environment known as the “Bussey-Saksida touchscreen chamber”. Cambridge Enterprise, the technology transfer office of the University of Cambridge, supported commercialisation of the Bussey-Saksida chamber, culminating in a license to Campden Instruments. Any financial compensation received from commercialisation of the technology is fully invested in further touchscreen development and/or maintenance. All other authors declare no competing interests.

Ethics approval and consent to participate

All procedures were conducted in accordance with the relevant guidelines and regulations, including the Canadian Council on Animal Care (CCAC) guidelines, and were approved by the Animal Care and Veterinary Services (ACVS) at Western University (protocol #2020-162 and #2020-163). All methods were performed in compliance with the ARRIVE guidelines.

Supplementary information

The online version contains supplementary material available at 10.1038/s41398-025-03795-5.