Allosteric Modulation of Pathological Ataxin-3 Aggregation: A Path to Spinocerebellar Ataxia Type-3 Therapies

Abstract

Spinocerebellar ataxia type 3 (SCA3) is a rare neurodegenerative disorder caused by the expansion of a polyglutamine (polyQ) repeat in ataxin-3 (Atx3) for which no disease-modifying therapies are available. The presence of protein inclusions enriched in polyQ-expanded Atx3 in neurons suggests that inhibiting its self-assembly may provide targeted therapies. Here, it is demonstrated that the supramolecular tweezer CLR01 binds to a lysine residue on a positively charged patch of the Atx3 catalytic Josephin domain, decreasing conformational fluctuations of the distal helical hairpin, without altering its ubiquitin hydrolase activity. This reduces exposure of the aggregation-prone region that initiates Atx3 self-assembly, ultimately delaying Atx3 amyloid fibril formation and reducing the secondary nucleation rate, a process linked to fibril proliferation and toxicity. CLR01's effects translate into the reversal of synapse loss in SCA3 cultured cortical neuron model, improve locomotor function in a Caenorhabditis elegans SCA3 model, and delay disease onset with reduced severity of motor symptoms in a SCA3 mouse model. These insights reveal a novel allosteric site for developing CLR01-inspired therapies targeting pathological aggregation while preserving essential functional sites. They also highlight that targeting allosteric sites in amyloid-forming proteins may provide new opportunities for safe therapeutic strategies for various protein misfolding disorders.

Keywords: amyloid; molecular therapies; molecular tweezer; polyglutamine; preclinical models; protein dynamics.

Figure 1

CLR01 interacts with Atx3 and induces conformational changes in the JD. a) Chemical structures of the molecular tweezer CLR01 and the control molecule CLR03. b) ESI‐MS spectra of wild‐type Atx3 (Atx3 13Q) and CLR01 at a 1:1 molar ratio. The insets shows the deconvoluted spectra and highlight the two CLR01 binding events. c) Cartoon representation of three states of JD: experimental NMR models of the JD in open (PDB ID: 1YZB,^{[ 46 ]} yellow) and closed (PDB ID: 2AGA,^{[ 93 ]} orange) conformations. A representative structure (light pink) of the “half‐open” conformation of the JD is shown (most populated cluster from the GaMD simulations in complex with CLR01). d) Amino acid sequence of JD with residues undergoing chemical shifts in the NMR HSQC spectra upon CLR01 binding in red. Secondary structure elements in the JD are shown below the amino acid sequence with helix α1 in green, the aggregation‐prone segment (α4–β1) in bold and orange, and helix α6 in teal. e) Cartoon representation of the JD open conformation with the residues affected by chemical‐shift‐perturbation (CSP) upon addition of CLR01 shown as red spheres. K128, a key residue in the interaction between CLR01 and JD, is highlighted as pink spheres. Other regions are colored as in panel (d). f) Cartoon representation of the third most populated cluster from the GaMD simulations of JD (open conformation) with a single CLR01 molecule (magenta sticks) bound at K128 (pink spheres). Residues displaying a chemical shift in the HSQC spectrum are highlighted as red spheres; R182 and other positively charged residues, corresponding to predicted CLR01 binding sites according to GaMD simulations and Gibbs binding energy estimations, are shown as sticks. This structure highlights a state where the hairpin (yellow surface) moves toward the aggregation‐prone region (orange).

Figure 2

CLR01 binding modulates the Atx3 aggregation pathway, delaying amyloid fibril assembly and decreasing the secondary nucleation rate. a,d) Representative ThT progress curves of 5 µm Atx3 13Q (a) or Atx3 77Q (d) amyloid self‐assembly kinetics in the absence or presence of CLR01 or CLR03. Error bars correspond to the standard deviation of three replicates. b,c) Bar plots show mean ± standard error of the mean (SEM) for (b) Atx3 13Q amyloid formation halftime (t ₅₀) and (c) aggregation rate (v ₅₀) derived from fitting of ThT fluorescence curves.^{[ 99 ]} Control samples were compared with CLR01 or CLR03‐treated samples at the indicated molar ratios. Individual points represent independent replicates (N = 5; each representing the mean values from 3–5 technical replicates). e,f) Same as (b, c) for Atx3 77Q, with 4 independent replicates. Statistical analysis was performed using a linear mixed‐effects model (LME) for Atx3 13Q (b, c), and repeated‐measures one‐way ANOVA for Atx3 77Q (e,f), followed by Dunnett's multiple comparisons test (Table S2, Supporting information). g) Time‐dependent DLS intensity distributions during the aggregation of 5 µm Atx3 13Q alone or in the presence of 25 µm CLR01 or CLR03. Different colors show different time points (size distributions measured in triplicate). Vertical lines: scattering intensities predicted by a nucleation‐and‐growth model for monomers (thinner lines) and amyloid fibrils (thicker lines) at the end of 48 h incubation (refer to Figure S7 in the Supporting Information, for other time points); h) same as (g) for Atx3 77Q. i) Fitted nucleation‐and‐growth rate constants for Ax3 13Q and Atx3 77Q; error bars correspond to standard errors (Table 1).

Figure 3

CLR01 reduces the assembly of mature fibrils of mutant Atx3 and dissociates preformed protofibrils and mature fibrils. a) Morphological analysis of 5 µm Atx3 13Q or Atx3 77Q aggregates in the absence or presence of 50 µm CLR01 or CLR03 at the endpoint of the aggregation assay (66 h), monitored by negative staining TEM. Scale bars correspond to 100 nm; orange rectangles denote Atx3 curvilinear protofibrils; red rectangles highlight small protofibrils formed in the presence of CLR01, red triangles point to Atx3 spherical oligomers; white arrowheads mark clusters of spherical aggregates; and white open arrows point to mature Atx3 Q77 fibrils. b) Particle size histograms and c) scatter plots of all particle measurements with median ± 95% CI, from the TEM images shown in (a) (3–4 images per condition were analyzed). Descriptive statistics are shown in Table S3 (Supporting information). d) Morphological analysis of 5 µm Atx3 13Q or Atx3 77Q protofibrils formed at the endpoint of the aggregation assay (t = 66 h), before (T0) and after three days of incubation (T3d) at 37 °C in the absence (Control) or presence of 400 µm CLR01 or CLR03 by negative‐staining TEM. Scale bars correspond to 100 nm. e) Morphological analysis of 5 µm 77Q mature SDS‐resistant fibrils from a 160 h aggregation assay before (T0) and after ten days of incubation (T10d) at 37 °C in the absence (Control) and presence of 400 µm CLR01 by negative‐staining TEM. Scale bars correspond to 200 nm.

Figure 4

CLR01 reverts the loss of glutamatergic synapses in cortical neurons caused by pathological Atx3. Rat cortical neurons were transfected with plasmids encoding eGFP, wild‐type eGFP‐Atx3 28Q, or mutant eGFP‐Atx3 84Q. a) Neurons were immunolabeled for MAP2, PSD95, and vGLUT1. Excitatory synapses were detected as PSD95 puncta that colocalized with vGLUT1; scale bar = 10 µm. b) Integrated fluorescence intensity of PSD95 puncta colocalized with vGLUT1 (n = 33–36 neurons per condition in 3 independent experiments). Statistical analyses were performed using Kruskal–Wallis and Dunn's post‐hoc test (Table S4, Supporting Information). c) Neurons incubated with 10 µm CLR03 or CLR01 were immunolabeled for MAP2, PSD95, and VGluT1 clusters; scale bar = 10 µm. d) Integrated fluorescence intensity of PSD95 puncta colocalized with VGluT1 (n = 35–41 neurons per condition in each of 3 independent experiments. Statistical analyses were performed using Kruskal–Wallis and Dunn's post‐hoc tests (Table S5, Supporting Information). In panels (c) and (d), boxes show the 25th and 75th percentiles, whiskers range from the minimum to the maximum values, and the horizontal line shows the median value. Mean is represented by “+.” e) GFP‐Atx3 accumulates in the cell body of a fraction of transfected cortical cultures treated with CLR03 or CLR01; scale bars = 300, 15 µm. f) Percentage of cells with GFP‐Atx3 aggregates (n = 3 independent experiments, 31–35 neurons per condition). Statistical analyses were performed using two‐way ANOVA and Sidak's multiple comparisons post‐hoc tests (Table S6, Supporting Information). p‐values are indicated in the graphs shown in panels (b), (d), and (f). Data are represented as mean ± SEM.

Figure 5

CLR01 treatment alleviates motor deficits in a C. elegans model of SCA3 even when treatment is initiated at a postsymptomatic disease phase. a) CLR01 treatment improved motor dysfunction of mutant Atx3 animals. AT3q130 animals were treated with different concentrations of CLR01 for four days, and the percentage of locomotion‐defective animals was determined. CLR01 significantly impacted motor phenotype between 0.0001 and 1 µm; the maximum efficiency (%) was observed at 0.1 µm at which 67% phenotypic rescue was achieved compared to the control strain (Table S8, Supporting Information). b) Treatment with 0.1 µm CLR01 for four days did not impact motor phenotype in the wild‐type strain (N2). c) Postsymptomatic treatment with CLR01 improves the animals' motor defects after 2 or 4 days of treatment (days 6 and 8 after hatching). (a–c) Data are represented as mean ± SD of 4 independent experiments, with ≈50 animals per condition/per assay (total number of animals of ≈250). A one‐way ANOVA test was applied, followed by a bilateral Dunnet test for post‐hoc comparisons (a and c). In the comparison between the two groups (b and c before treatment), independent‐samples t‐test was used (Tables S9, S10, Supporting Information). d–g) Impact of early life chronic CLR01 treatment on mutant human Atx3 aggregation pattern, labeled with an Atx3‐specific antibody in the C. elegans model of SCA3 throughout animals’ development and adulthood. Pictures of the animal's heads were obtained by confocal microscopy (Olympus FV1000) and analyzed using the MeVisLab software.^{[ 106 ]} The graphs show a pool of 3 independent assays, with at least four animals per condition in each trial. p‐values were calculated using independent‐samples t‐test (Table S10, Supporting Information). Scale bars = 20 µm.

Figure 6

CLR01 chronic administration improves motor function and neuropathology of SCA3 mice. a) The latency to cross a swimming tank, representing both the strength and the movement coordination ability of the animals, is higher in vehicle‐treated SCA3 mice. The severity of this symptom improves over time with CLR01 treatment, which has a delayed onset and is effective up to 16 weeks of age. b) CLR01 improves SCA3 mice balance, decreasing their latency to cross a 12 mm squared beam. c) The gait quality of SCA3 animals is strikingly improved by CLR01 treatment, delaying the onset of ataxic gait by eight weeks. d) Representative images of cresyl violet staining in the cervical spinal cord of healthy and pyknotic cells. e) Vehicle‐treated SCA3 mice showed a substantial increase in the number of pyknotic cells, which was rescued by daily CLR01 administration for 18 weeks. f) Representative pictures of the motor‐neuron cell bodies (black arrows) in the ventral horn of the cervical spinal cord. g) CLR01 treatment increases the number of motor neurons in the cervical spinal cord. Sample size and statistical details are provided in Table S11 (Supporting Information). The scale bar for lower magnification pictures represents 200 µm, and for higher magnification, 100 µm. p‐values are shown in the graph. (a, b) Data are represented as mean ± SEM (n = 13–19 per group). A repeated measures ANOVA test was applied, followed by Tukey test for multiple comparisons. (c) Data are represented as the % of animals within each category (n = 13–19 per group). A Kruskal–Wallis H test was applied. (e and g) Data are represented as the median (min; max) with all datapoints shown (N = 5–6 animals per group; n = 4–7 slices per mouse). A one‐way ANOVA test was applied, followed by a Tukey test for multiple comparisons. Black lettering represents the comparisons between WT‐vehicle and SCA3‐vehicle. Red lettering represents the comparisons between SCA3‐vehicle and SCA3‐CLR01.

Figure 7

Schematic representation of the effects of CLR01 on Atx3 self‐association. Binding of CLR01 to K128 (colored dark blue) within a positively charged surface patch opposite to the aggregation‐prone region and ubiquitin binding site (orange) triggers a conformational change in the helical hairpin (yellow) of the Josephin domain. The CLR01‐induced conformational change reshapes the Atx3 aggregation pathway and traps spherical oligomers while reducing secondary nucleation rates. This modulation of the self‐assembly pathway correlates with the reversion of synapse dysfunction in SCA3 cell models and a delay in the onset of motor impairment in SCA3 animal models. Image created in BioRender (https://BioRender.com/lgwm68m).