Sacs^–/– mice carrying a deletion of the Sacs gene were used as previously described (Girard et al., 2012; Lariviere et al., 2015; Ady et al., 2018). Heterozygous Sacs^+/– mice were bred to obtain litter-matched Sacs^–/– and wildtype (WT) mice. Roughly equal numbers of male and female mice were used in all experiments. Sex differences were tested for but not observed. Experiments were performed at postnatal day 40 (P40; N = 4 for WT and N = 4 for Sacs^–/– mice), P150 (N = 7 for WT and N = 8 for Sacs^–/– mice), P270 (N = 4 for WT and N = 4 for Sacs^–/– mice), and P365 (N = 4 for WT and N = 4 for Sacs^–/– mice). Breeding and animal procedures were approved by the McGill University Animal Care Committee and were in accordance with the rules and regulations established by the Canadian Council on Animal Care.

Brain tissue from WT and Sacs^–/– mice was prepared as previously described (Ady et al., 2018). In brief, mice were deeply anesthetized with 2,2,2-tribromoethanol (Avertin) via intraperitoneal injection, and tissue fixation was carried out via intracardiac perfusion. An initial flush was performed with phosphate-buffered saline (PBS, 0.1M, pH 7.4) and 5.6 μg/ml heparin salt. This was followed by perfusion with 40 ml of 4% paraformaldehyde (PFA) in phosphate buffer (PB, pH 7.4), which was also used for post-fixation storage of the extracted brains for a further 3 days at 4°C on a shaker at 70 RPM. If further storage was required, brains were stored in PBS with 0.5% sodium azide until slicing.

The cerebellum was sliced coronally using a Vibratome 3000 sectioning system (Concord, ON, Canada) to produce 100 μm thick slices. Immunohistochemistry was carried out on free-floating slices and two sections per location were used for each animal. Briefly, slices were incubated for half an hour in blocking solution (containing 1 × PBS, pH 7.4, 0.4% Triton X, 5% bovine serum albumin (BSA), and 0.05% Sodium Azide) followed by a 3-day incubation with the primary antibodies in blocking solution and 90 minutes for secondary antibodies in blocking solution. Following the final incubation, slices were washed in PBS and mounted using ProLong Gold Antifade mounting medium (ThermoFisher Scientific, Waltham, MA, United States), protected from light, and kept at 4°C. All chemicals were purchased from Millipore Sigma (Oakville, Canada) unless otherwise indicated.

Lobule immunohistochemistry: To label zebrin-positive cells, we used goat Aldolase C antibody (1:300; Catalogue number SCl-12065; Santa Cruz Biotechnology, Dallas, TX, United States) in combination with Alexa-488 conjugated anti-goat antibody (A11055; Life Technologies, Burlington, ON, Canada). In addition, we used rabbit anti-calbindin antibody (1:500, CB-38a; Swant, Marly, Switzerland) as a Purkinje cell marker in combination with Alexa-594 conjugated anti-rabbit secondary antibody (R37119; Life Technologies). All secondary antibodies were used at 1:1,000 dilution. In a separate set of experiments to validate the use of calbindin staining to determine Purkinje cell numbers in Supplementary Figure 1, we performed immunohistochemistry as previously described for calbindin, followed by a counterstaining step using NeuroTrace 435/455 blue fluorescent Nissl stain (1:80; 1-hour incubation; N21479; Life Technologies).

Cerebellar nuclei immunohistochemistry: To label Purkinje cell terminal puncta in the CN, sagittal cerebellar slices were prepared and stained with rabbit anti-calbindin antibody (1:500, CB-38a, Swant) in combination with an Alexa-594 conjugated anti-rabbit secondary antibody (711-585-152, Jackson Immunoresearch, West Grove, PA, United States). To label zebrin-positive Purkinje cell puncta, mouse aldolase-C antibody (1:500, ab190368, Abcam, Toronto, Canada) was used in combination with Alexa-488 anti-mouse secondary antibody (715-545-150, Jackson Immunoresearch). CN neurons were labeled with guinea pig anti-NeuN antibody (1:500, abN90, Millipore, Burlington, MA, United States) in combination with DyLight 405 anti-guinea pig secondary antibody (1:500, 106475003, Jackson Immunoresearch). In separate experiments on additional slices from the same experimental animals, guinea pig vesicular GABA transporter (VGAT) antibody (1:200, 131 004, Synaptic Systems, Gottingen, Germany) with Alexa-488 anti-guinea pig secondary antibody (1:1000, 706-545-148, Jackson Immunoresearch) was used in conjunction with rabbit anti-calbindin antibody (Swant, as before), to label GABAergic terminals in the CN, shown in Supplementary Figure 2. For each animal, eight sagittal slices were collected at spaced intervals to ensure sampling across the fastigial and interposed CN. We used four WT and three Sacs^–/– P90 mice. Number of cells: WT anterior = 28 cells, WT posterior = 26 cells, Sacs^–/– mice anterior = 25 cells, Sacs^–/– mice posterior = 46 cells.

Zebrin stripe imaging: Images were acquired using an Axio Observer Z1 fluorescent microscope equipped with Zen blue software (Zeiss, Oberkochen, Germany) at a 20× magnification and tiled together for a whole lobule visualization. Gain and contrast were kept constant for all conditions during the imaging process. To compare across conditions, we used anatomical markers in the cerebellum to ensure that we were comparing data at the same brain location. We selected the cerebellar slices using images from the coronal view of the interactive Allen Mouse Brain atlas viewer¹ (Allen Institute of Brain Science, mouse coronal Atlas, 2011). For the analysis of the anterior lobules, we selected sections that matched images between 115 and 117; and for the posterior lobules, we selected sections that matched images between 127 and 130. We used two sections per lobule per animal. Mean fluorescence intensity and band thickness were measured using ImageJ 1.51F software. Briefly, the mean intensity was measured by tracing a box that contained both soma and dendrites of each zebrin-positive band. A similar box (size and position) was used to measure the zebrin-negative bands. To measure the thickness of the bands, a line containing only the zebrin-positive label was traced parallel to the length of the lobule, midway of the length of the dendritic tree. Complementary measurements were done in the zebrin-negative bands (refer to Supplementary Figure 3 for illustration of analysis).

Purkinje cell counts: Images were acquired at a 20× magnification. Images were taken in one focal plane where the label of both primary antibodies was seen. In a separate set of experiments, we confirmed using a Nissl stain that calbindin gave us a good measurement of total Purkinje cell number (Supplementary Figures 1, 3), at an age when Purkinje cell death was observed. Thus, calbindin labeling was used to count total Purkinje cell numbers, while zebrin labeling was used to discern zebrin-positive (that were co-labeled for calbindin and zebrin) cells from zebrin-negative cells (that were labeled only for calbindin), even when Purkinje cell death was evident. We measured Purkinje cell density by measuring the number of Purkinje cells as a function of the length of the Purkinje cell layer, reporting numbers per 100 μm. The same measurement was used for zebrin-positive and zebrin-negative cells. Measurements were performed on sections that were identified using coordinates from the Allen Mouse Brain Atlas as described above. Briefly, anterior lobule III and possibly the boundary of lobule IV were identified between positions 115 and 117, and posterior lobules (lobules VIII and IX) were identified between positions 127 and 130. Total Purkinje cell length per condition was similar and is reported in Table 1. Representative images of the immunostaining are shown in pseudo-color.

Cerebellar nuclei imaging: Images were acquired using an LSM800 confocal microscope (Zeiss) at 1024 × 1024 resolution with a 63× objective and consistent imaging conditions and settings throughout. Images stacks of CN cells were obtained, and the cross-section of each cell at the widest point was chosen for analysis. The number of Purkinje cell synapses onto each CN cell was quantified by counting the number of calbindin-positive puncta surrounding large CN cells (>15 μm diameter). Puncta were only included in counts if they were immediately adjacent to the NeuN-positive CN cells. Puncta that were not touching the CN cell (located more than 0.5 μm away from the NeuN-positive area) were determined to not be forming a synapse with that CN cell and were not included in counts (refer to Supplementary Figure 2 for examples). Zebrin-positive Purkinje cell synapses were identified by counting the number of the previously identified calbindin-positive puncta that were also zebrin-positive (Supplementary Figure 2). The number of zebrin-negative Purkinje cell synapses was then determined by subtracting the number of zebrin-positive puncta from the total number of calbindin-positive puncta. The data were then collated to show the number of zebrin-positive or zebrin-negative Purkinje cell puncta per large CN cell. All imaging was acquired and analyzed blind to condition, and images are presented in pseudo-color.

Comparisons were made using paired or unpaired Student’s t-tests for normally distributed data or Mann Whitney U tests for when data were not normally distributed using JMP 12 (SAS, Cary, NC, United States) software. Data are represented by box and whisker plots, showing the median (horizontal line within boxes), second and third quartiles (rectangles) ± 1 SD (whiskers), or by an average ± SEM.

The anterior-posterior differences in Purkinje cell firing properties have been associated with the expression profile of zebrin (Larouche and Hawkes, 2006): zebrin-positive cells are enriched in posterior lobules and fire at a lower frequency than zebrin-negative cells, which are enriched in anterior lobules (Xiao et al., 2014; Zhou et al., 2014). Since abnormal zebrin expression has been observed in rodent models of other forms of ataxia (Sawada et al., 2009; Sarna and Hawkes, 2011; Bailey et al., 2014; White et al., 2021), the changes in anterior-lobule firing that we previously reported (Ady et al., 2018) may likewise arise from abnormal expression of zebrin in the cerebellum in Sacs^–/– mice. Conversely, intact neurotransmission from Purkinje cells to their downstream targets is required for proper zone formation in the cerebellum (White et al., 2014), and given that we have previously shown that Purkinje cells in Sacs^–/– mice both fire at reduced frequencies and have deficits in their innervation of the CN (Ady et al., 2018), we wondered whether changes in Purkinje cell properties could themselves lead to disrupted zebrin patterning.

Anterior lobules (III and IV) are comprised of mostly zebrin-negative Purkinje cells, with only three thin, sagittally oriented zebrin-positive stripes in the vermis. We rationalized that if changes we observed in firing rates were due to changes in zebrin expression, they might arise either from the expansion of the width of zebrin stripes or from aberrant ectopic zebrin expression in regions that are normally zebrin-negative, which would be associated with lower Purkinje cell firing rates. We used immunohistochemistry in coronal slices from WT and Sacs^–/– mice at disease onset (P40-50) to examine zebrin expression throughout anterior lobules III/IV (Figure 1A and Supplementary Figure 3). We found no significant differences in the width of zebrin-positive stripes when comparing WT and Sacs^–/– cerebellum (N = 4 mice for WT and Sacs^–/–, 2 sections/brain analyzed; Figure 1B, see Table 2 for individual P-values for each band). If developmental changes in zebrin patterning occurred in Sacs^–/– mice, the intensity of zebrin staining in both zebrin-negative and -positive bands might be altered, since changes in molecular compartmentalization have been reported in other mice (Miyazaki et al., 2012; Bailey et al., 2014). We determined the intensity of zebrin staining in both zebrin-positive and zebrin-negative stripes in anterior lobules and found no significant differences in Sacs^–/– mice compared to WT (Student’s t-test for all comparisons other than posterior bands 2_L, 1_R, and 2_R when Mann Whitney U tests were performed, not significantly different; zebrin-positive bands: band 2_L: P = 0.217, band 1: P = 0.075, band 2_R: P = 0.109; zebrin-negative bands: band 2_L: P = 0.10, band 1_L: P = 0.525, band 1_R: P = 0.675, band 2_R: P = 0.706; Figures 1C,D). This suggests that the reduction in Purkinje cell firing rates that we observed in anterior lobules at P40 (Ady et al., 2018) does not arise from abnormal expression of zebrin in these cells or from changes in the developmental patterning or expression of zebrin. Previously, it has been reported that Purkinje cell death is not observed at P30 but is evident at P90 (Lariviere et al., 2015). To determine whether cell death is observed at P40, we determined the density of Purkinje cell bodies and found that no differences were observed in Sacs^–/– mice compared to WT (Student’s t-test, P = 0.99; Figure 1E). Similarly, there were no detectable differences in either zebrin-positive (Student’s t-test, P = 0.37; Figure 1F) or zebrin-negative (Student’s t-test, P = 0.89; Figure 1G) cell density.

Zebrin-positive Purkinje cells are more prominent in posterior lobules. We predicted that zebrin expression would be unchanged in posterior Sacs^–/– cerebellar vermis, since we observed no Purkinje cell firing deficits (Ady et al., 2018) and no later Purkinje cell loss (Lariviere et al., 2015, 2019) in this region. To address this, we measured zebrin expression in posterior lobules VIII and IX (Figure 2A and Supplementary Figure 3) and observed no differences in zebrin band width (Figure 2B and Table 3), or zebrin intensity in either zebrin-positive (Student’s t-test for all comparisons, not significantly different; zebrin-positive bands: band 3_L: P = 0.878, band 2_L: P = 0.89, band 1: P = 0.543, band 2_R: P = 0.368, band 3_R: P = 0.47; Figure 2C) or zebrin-negative (Student’s t-test for all comparisons, not significantly different; zebrin-negative bands: band 2_L: P = 0.515, band 1_L: P = 0.467, band 1_R: P = 0.43, band 2_R: P = 0.401; Figure 2D) bands from posterior lobules in WT and Sacs^–/– mice. Similarly, we observed no differences in the total cell density in posterior lobules (Student’s t-test, P = 0.52; Figure 2E) or in the densities of zebrin-positive (Student’s t-test, P = 0.54; Figure 2F) or zebrin-negative (Student’s t-test, P = 0.70; Figure 2G) cells.

Although Purkinje cell death is known to occur in the predominantly zebrin-negative anterior lobules of the Sacs^–/– mouse cerebellum (Lariviere et al., 2015, 2019), the molecular profile of surviving neurons has not been determined in Sacs^–/– mice. To ascertain this, we examined an age when Purkinje cell death is known to occur that is still relatively early during disease progression (P150; Figure 3). We used immunohistochemistry to label Purkinje cells with calbindin for total Purkinje cells, and zebrin to discern zebrin-positive (which are dual labeled) and zebrin-negative (which only express calbindin) cells. We measured the width of bands in the anterior zebrin-negative lobules and found no difference in Sacs^–/– mice compared to WT (Figure 3B and Table 2). Interestingly, the width of two of the three zebrin-positive bands in the anterior lobules at P150 was slightly expanded in Sacs^–/– mice compared to WT (Figure 3C and Table 2). We next measured the width of zebrin-negative (Figure 3D and Table 3) and zebrin-positive (Figure 3E and Table 3) bands in posterior lobules at P150 and found a small increase in one zebrin-negative band (band 1_L; Figure 3D and Table 3), with no other statistically significant differences in Sacs^–/– mice posterior bands.

To determine if changes in band width reflect changes in Purkinje cell density at P150, we next measured this in WT and Sacs^–/– mice (Figure 3F). We found that the total density of calbindin-labeled Purkinje cell somata was reduced in anterior lobules in P150 Sacs^–/– mice, consistent with previous reports (Student’s t-test, P < 0.001; Figure 3F) (Lariviere et al., 2015). To determine whether this reduction was observed equally in zebrin-positive and -negative Purkinje cells, we then examined cells of each molecular profile separately. We observed a significant reduction in anterior lobule zebrin-negative Purkinje cells (Student’s t-test, P < 0.0001; Figure 3G, left). Although the number of zebrin-positive cells in anterior lobules is low (Figure 3A, left), we did not observe any significant differences in their density in Sacs^–/– mice (Student’s t-test, P = 0.68; Figure 3G). Since cell death has not been reported in posterior lobules in Sacs^–/– mice (Lariviere et al., 2015, 2019), we expected no differences in the proportion of zebrin-positive and -negative Purkinje cells there. Consistent with this, we observed that the densities of both zebrin-positive and -negative cells in posterior lobules were not significantly different in P150 Sacs^–/– mice (Student’s t-test for all measurements: P = 0.82 for total cell density; P = 0.44 for zebrin-negative cell density; and P = 0.70 for zebrin-positive cell density; Figure 3H). These results suggest that Purkinje cell death occurs primarily in zebrin-negative neurons in anterior lobules at an early stage of disease progression (P150) and that although zebrin-positive bands expand, they appear to do so to occupy the space left by the zebrin-negative cell death.

ARSACS is a progressive neurodegenerative disease, and in mouse models, increased levels of Purkinje cell death have been observed with aging (Lariviere et al., 2015, 2019). Therefore, we wondered whether advanced stages of disease progression extend cell death to include previously resilient zebrin-positive cells. To determine this, we measured zebrin-positive and -negative Purkinje cells in anterior and posterior lobules at advancing disease stages. First, we examined P270 (9 months; Figure 4A). We found that in anterior lobules, one zebrin-negative band was reduced, while two were unchanged (Figure 4B and Table 2) and two of the three zebrin-positive bands were wider in Sacs^–/– mice (Figure 4C and Table 2), just as we observed at P150. In posterior lobules, however, there was only a small reduction observed in one zebrin-positive band (band 2_L; Figure 4E and Table 3), with no other differences were observed in the remaining zebrin-positive and zebrin-negative bands (Figures 4D,E and Table 3). Similar to our findings at P150, the total cell density was reduced in anterior but not posterior lobules in Sacs^–/– mice (Student’s t-test, P = 0.02; Figure 4F), which was reflected in a reduction in zebrin-negative, but not zebrin-positive cell density in anterior lobules (Student’s t-test for both measurements; zebrin-negative cell density: P = 0.01; zebrin-positive density: P = 0.52; Figure 4G). However, no differences in total zebrin-positive or -negative cells were observed in posterior lobules (Student’s t-test for all comparisons; total: P = 0.13; zebrin-negative: P = 0.26; zebrin-positive: P = 0.08; Figure 4H).

Finally, we examined an even more advanced stage of disease progression at 1 year (P365; Figure 5A) and found results similar to those at P150 and P270, but more severe. We now observed that two out of four anterior zebrin-negative bands were reduced in Sacs^–/– mice compared to WT (Figure 5B and Table 2), and this was accompanied by significant increases in the band width of all three anterior zebrin-positive bands in Sacs^–/– mice at P365 (Figure 5C and Table 2), suggesting that Purkinje cells become more disorganized in anterior lobules as the disease progresses. However, restructuring was predominantly observed in anterior lobules, as even at this later disease stage, posterior bandwidths appeared largely normal in Sacs^–/– mice for zebrin-negative (Figure 5D and Table 3) and most zebrin-positive (Figure 5E and Table 3) bands, other than a small increase in zebrin-positive band 1 (Figure 5E and Table 3). The density of Purkinje cells in Sacs^–/– mice at P365 was again only affected in anterior, but not posterior lobules (Student’s t-test for both comparisons; anterior: P = 0.003; posterior: P = 0.48; Figure 5F). Similar to what we observed at earlier ages, this change appeared to arise predominantly from changes in anterior zebrin-negative, but not zebrin-positive cell density (Student’s t-test for both comparisons; zebrin-negative cell density: P = 0.004; zebrin-positive cell density: P = 0.75; Figure 5G). In agreement with results from overall posterior cell density at P365 (Figure 5F), no change in posterior zebrin-negative and -positive cell densities was observed in Sacs^–/– mice at P365 (Student’s t-test for both comparisons, zebrin-negative cell density: P = 0.67; zebrin-positive cell density: P = 0.60; Figure 5H). These results agree with our hypothesis that cell death is largely limited to anterior lobules even at relatively advanced disease stages and that this is manifested by changes in cell density of zebrin-negative cells. However, although no changes in zebrin-positive numbers are observed, anterior lobules become more disorganized at P365 in Sacs^–/– mice, likely because zebrin-positive bands expand into space where zebrin-negative cells have died.

Given that we have examined the progression of Purkinje cell loss in a mouse model of ARSACS over time, we wondered whether this represented change arising from aging or from disease progression. We compared the change in anterior lobule Purkinje cells across time, from P40 to P365, which allows us to normalize to WT cells, and thus graphically represent the contribution arising exclusively from disease progression. We observed a gradual decrease of Purkinje cell numbers in anterior lobules (Figure 6A), which was reflected in the proportion of zebrin-negative cells over time (Figure 6B), while the proportion of zebrin-positive cells remained remarkably similar at all ages (Figure 6C).

Purkinje cell death is one of the most prominent pathophysiological changes in both ARSACS brains and animal models of ARSACS, but other changes have also been observed. For instance, although the numbers of cells in the CN of the cerebellum are unaltered in Sacs^–/– cerebellum, the number of puncta, likely reflecting the number of synaptic inputs onto these cells, is reduced at the onset of disease in Sacs^–/– mice (Ady et al., 2018). Changes in the innervation of target neurons in the CN will likely have a profound impact on cerebellar function. This raises the question of whether this reduction arises from changes in numbers of inputs from all Purkinje cells or whether specific subsets of Purkinje cells also show preferential changes in their innervation of the CN?

We first determined whether the decrease in the number of puncta, which we used as a proxy for the number of potential Purkinje cell terminals innervating the CN neurons, that we previously observed at disease onset (Ady et al., 2018), was observed as the disease progressed (at P90). We measured the number of calbindin-positive Purkinje cell puncta made onto large projection neurons in the fastigial and interposed nuclei (Figure 7A) since these receive innervation from the vermis (Voogd and Ruigrok, 2004). We found that the number of puncta was reduced in Sacs^–/– mice (Student’s t-test; P < 0.0001; Figures 7A,B), similar to what is observed at earlier ages (Ady et al., 2018). We have previously demonstrated that in the CN of both WT and Sacs^–/– mice, there is a high degree of colocalization between calbindin-positive puncta and puncta that are stained with VGAT at P40 (Ady et al., 2018). This indicates that these previously identified calbindin-positive puncta are likely functional GABAergic Purkinje cell terminals. We found a high degree of co-labeling in WT and Sacs^–/– mice at P90 (Supplementary Figure 2). Thus, the reduction in the number of calbindin-positive puncta in the CN of ARSACS mice at P90 appears to reflect a loss of functional Purkinje cell presynaptic terminals rather than a shift in the proportion of inputs arising from Purkinje cell terminals compared to other inputs.

It has previously been described that the pattern of zebrin projections respects the anterior – posterior (or rostral – caudal) division of the CN (Hawkes and Leclerc, 1986; Sugihara, 2011), with the anterior regions of the nuclei receiving predominantly zebrin-negative input and posterior nuclei receiving predominantly zebrin-positive input (Sugihara, 2011). We observed this division in both WT and Sacs^–/– mice (Figure 7C). Thus, to determine whether changes were made in the number of zebrin-positive or -negative puncta in the CN, we analyzed neurons with respect to their anterior – posterior location within both CN nuclei.

In anterior CN, which receives largely zebrin-negative Purkinje cell innervation, we observed a reduction in the number of zebrin-negative puncta onto large CN neurons compared to WT (WT: N = 4; Sacs^–/– N = 3; Student’s t-test, P < 0.0001; Figures 8A,B). The number of zebrin-positive Purkinje cell puncta in anterior CN was low in both WT and Sacs^–/– mice and did not significantly differ between genotype (Mann Whitney U test, P = 0.81; Figures 8A,B), which suggests that no major rewiring of puncta occurs from zebrin-positive cells. Interestingly, although the total number of zebrin-negative puncta made onto neurons was lower in posterior CN compared to anterior CN, there was no significant reduction in the number of zebrin-negative puncta in Sacs^–/– cerebellum compared to WT (Mann Whitney U test, P = 0.15; Figures 8C,D). We also observed unchanged numbers of zebrin-positive Purkinje cell puncta in posterior CN in Sacs^–/– mice (Student’s t-test, P = 0.16; Figures 8C,D). These data suggest that reduction in the innervation of targets in the CN by Purkinje cell axons is predominantly arising from alterations in zebrin-negative Purkinje cells in the anterior vermis. Taken together, our results are consistent with a specific population of Purkinje cells, which do not express zebrin and reside in anterior vermis, that are uniquely vulnerable to loss of sacsin. Vulnerable cells display firing deficits (Ady et al., 2018), show reduced innervation of downstream targets (although Purkinje cells in other projection regions may also be affected in this altered innervation pattern, Figures 8A,B) and also are vulnerable to Purkinje cell death as the disease progresses (Figures 3–6).