Mutations within TTBK2 cause the adult-onset, neurodegenerative disease SCA11. However, the etiology of SCA11 is poorly defined. SCA11 is somewhat unusual among SCAs, in part because the reported causal mutations are base pair insertions or deletions within the coding region of TTBK2 (Houlden et al., 2007; Johnson et al., 2008; Lindquist et al., 2017), rather than the expansion of CAG repeats, which is the genetic cause of most SCA subtypes (Hersheson et al., 2012). To test the requirements for TTBK2 in maintaining neural function within the adult brain, we obtained a conditional allele of Ttbk2 (Ttbk2^{tm1c(EUCOMM)Hmgu}) from the European Mutant Mouse Cell Repository, (referred to from here as Ttbk2^fl). We then crossed Ttbk2^fl mice to a mouse line expressing tamoxifen-inducible Cre recombinase driven by a ubiquitously expressed promoter, Ubc-Cre-ERT2 (Ruzankina et al., 2007). Using this model, we induce recombination of Ttbk2 in all tissues of the mouse, including the brain, upon injection with tamoxifen (TMX). Because morphogenesis of the mouse cerebellum is complete by P21 (Marzban et al., 2014), we chose this time to begin our TMX injections. For all of our experiments, Control animals are either siblings with the same genotype (Ttbk2^fl/fl;Ubc-Cre-ERT2⁺) injected with oil vehicle only, or Ttbk2^fl/fl;Ubc-Cre-ERT2^- sibling mice injected with the same dose of TMX. We found no phenotypic differences between Control condition animals or pre-induction Ttbk2^fl/fl;Ubc-Cre-ERT2+ animals at P21 (Figure 1—figure supplement 1A-C). Consistent with other conditional mutants where cilia are globally removed in adulthood (Davenport et al., 2007), 4-month-old Ttbk2^c.mut mice exhibit obesity (Figure 1—figure supplement 1D, D’, E: 32.29 g ± 1.86 for Control vs. 46.33 g ± 2.04 for Ttbk2^c.mut) as well as cystic kidneys (Figure 1—figure supplement 1F). Loss of TTBK2 protein was confirmed with western blot analysis on cerebellum lysates from Ttbk2^c.mut animals and littermate Controls (Figure 1—figure supplement 2G).

Because the cerebellum is critical for motor coordination and SCA11 is associated with motor deficits, we evaluated locomotor behavior in the Ttbk2^fl/fl;Ubc-Cre-ERT2⁺,TMX treated animals (referred to from here as Ttbk2^c.mut) relative to littermate Controls. Within 3 weeks following induction of recombination with TMX, Ttbk2^c.mut mice exhibited apparent locomotor deficiencies when observed in their cage (Figure 1—video 1 ). To further examine motor coordination in Ttbk2^c.mut mice, we employed a rotarod performance test. Ttbk2^c.mut mice exhibited a shorter latency to fall compared to the littermate Controls in each trial, for both the accelerating rotarod analysis as well as the steady speed rotarod analysis (Figure 1A,B). These results indicate that Ttbk2^c.mut mice are impaired in their motor coordination, consistent with motor deficits observed in multiple mouse models of SCA (Lalonde and Strazielle, 2019; Klockgether et al., 2019).

To assess whether the motor behavioral changes we observed in the Ttbk2^c.mut animals are a consequence of changes to neuronal architecture in the adult brain, we examined Ttbk2^c.mut mice at 4 months of age (3 months post TMX). The brains of Ttbk2^c.mut mice have slightly smaller olfactory bulbs, but the overall gross morphology of the cortex and cerebellum was unchanged (Figure 1—figure supplement 1H). SCA11 pathology is associated with degeneration of the cerebellar neurons. We therefore examined the architecture and connectivity of neurons within the cerebellum to assess whether the Ttbk2^c.mut animals exhibited phenotypes similar to those described for mouse models of other subtypes of SCA. Within the cerebellum, PCs are the major source of functional neuronal output, and receive excitatory inputs primarily from parallel fibers and climbing fibers. Parallel fibers extend from the granule neurons, a population of densely packed neurons found directly beneath PCs (Ichikawa et al., 2016). Climbing fibers extend from neurons of the Inferior Olivary Nuclei (ION) in the medulla (Kano et al., 2018). These connections are essential for PC function, and dysfunction or loss of these connections, particularly the VGLUT2+ excitatory synapses from the climbing fibers, has been shown in various mouse models of SCA to be linked to pathology and disease progression (Duvick et al., 2010; Ebner et al., 2013; Furrer et al., 2013; Smeets and Verbeek, 2016; Smeets et al., 2015).

To assess the changes throughout Ttbk2^c.mut cerebella, we looked at the different layers of the cerebellar folia. We observed thinning of the molecular layer of the cerebellum, which comprises elaborate dendrites extended from the PCs (Figure 1C,D. 175 μm ± 3.422 for Control vs. 160.9 μm ± 2.527 for Ttbk2^c.mut). More dramatically, on examination of the synaptic marker VGLUT2, we found a marked reduction in these puncta throughout the Ttbk2^c.mut cerebellum compared to Controls (Figure 1C,E 527.3 puncta ± 12.68 for Control vs. 297.8 puncta ± 15.65 for Ttbk2^c.mut) despite a lack of obvious gross morphological changes in the dendritic arbors of Ttbk2^c.mut PCs (Figure 1—figure supplement 2A,B). Additionally, Ttbk2^c.mut animals injected with tamoxifen beginning at P45 exhibit the same loss of VGLUT2+ synapses (Figure 1—figure supplement 3A-B), showing that loss of TTBK2 after all connections and development of the cerebellum are complete, leads to degeneration of these synapses. We confirmed that this loss of VGLUT2 synapses was not caused by loss of dendritic spines, with use of a Golgi stain to view individual dendrites and spines on Ttbk2^c.mut PCs (Figure 1F,G; 1.236 ± 0.22 spines per μm for Control vs. 1.368 ± 0.32 spines per μm for Ttbk2^c.mut).

To further explore the role of TTBK2 in PC function, we examined calcium receptor abundance in these neurons. PCs require an intracellular calcium modulation network as a means of signaling function. Within this network inositol 1,4,5-trisphosphate receptors (IP3Rs) are key calcium channel regulators needed for calcium release from the surrounding endoplasmic reticulum (ER) throughout the PC (Sarkisov and Wang, 2008). Precise regulation of IP3R activity is critical, and a balance of calcium channel release is imperative to the overall function of the cerebellum. Mutations in the IP3R1 gene have been linked to SCA15 and SCA29, while overexpression of IP3R1 underlies phenotypes within SCA2 and SCA3 (Tada et al., 2016). Because of these links to other SCA-related phenotypes, we therefore examined IP3R expression throughout the Ttbk2^c.mut animals. We found that levels of IP3R in PCs are reduced in Ttbk2^c.mut compared to Controls starting at P45, with expression strongly reduced throughout the PCs 3 months after TMX (Figure 1H). Thus, like mouse models of other SCA subtypes, the PCs of Ttbk2^c.mut animals exhibit defects in calcium modulation consistent with dysfunction of these cells.

Next, we tested whether loss of Ttbk2 affects other cell types linked to the pathology of SCA, in addition to the PCs. Climbing fibers extend from neurons of the inferior olivary nucleus (ION) in the medulla. These fibers traverse the brain stem, enter the cerebellar cortex, and innervate the PC dendrites (Watanabe and Kano, 2011). As we saw a reduction of the VGLUT2+ synaptic terminals between these climbing fibers and PC dendrites, we examined the soma of the ION neurons from which these climbing fibers extend. In several subtypes of SCA, including SCA1, 2, 3, 6, and 7 (Seidel et al., 2012), the pathology of the disorder is characterized in part by the loss of ION soma; a characteristic also observed in mouse models of these diseases. Neurons within the ION can be identified by dual expression of Calbindin and NeuN in the medial ventral region of the medulla (Figure 2—figure supplement 1A,B). When we looked at this population of neurons, we did not notice a loss of these cells. However, we did find that the perikarya of the neurons within the ION were smaller in Ttbk2^c.mut animals compared to Controls. We therefore used NeuN to label the perikarya of neurons within the ION and measured the area of the somata of ION neurons, and found a significant reduction in the area of ION neuron soma in Ttbk2^c.mut mice at 4 months of age compared to Controls (Figure 2A,B; 180.4 μm² ± 1.93 for Control vs. 109 μm² ± 1.01 for Ttbk2^c.mut). Neuronal shrinkage is a phenotype that has been noted in patients with SCA1 as well as Friedrich’s ataxia, and is thought to precede neuronal apoptosis (Nagaoka, 2003; Dell'Orco et al., 2015; Kemp et al., 2016). This implies that, in addition to the PCs themselves, the neurons sending critical inputs to the PCs are perturbed in Ttbk2^c.mut mice.

Throughout the brain, astrocytes and glia also play important roles in maintaining synaptic connectivity and strength. In the cerebellum, the processes of the BG are interspersed with PC dendrites in the molecular layer, with BGs enwrapping the excitatory synapses of the PCs (Leung and Li, 2018). As defects in BG morphology have been linked to the etiology of SCA7 (Furrer et al., 2011), we examined the BGs in Ttbk2^c.mut. To assess the morphology of BGs in the Ttbk2^c.mut animals and evaluate whether defects in these cells may contribute to the phenotype, we used GFAP to visualize BG fibers that extend throughout the cerebellar folia. We found that the numbers of glial fibers were moderately reduced in Ttbk2^cmut cerebellar folia compared to littermate Controls (Figure 2C,D; 11.44 BG fibers ± 0.29 for Control vs. 7.64 BG fibers ± 0.22 for Ttbk2^cmut), suggesting that loss of Ttbk2 has modest effects on the morphology of BGs. Taken together, these data suggest that loss of Ttbk2 affects several cell types in the cerebellum and medulla, underscoring the widespread importance of Ttbk2 within these tissues.

Dysfunction and eventual atrophy of the PCs in the cerebellum is the primary pathology underlying SCA11 in human patients (Houlden et al., 2007). In our conditional Ttbk2 mutant mice, the most prominent phenotype is altered connectivity of the PCs with additional cellular changes seen in the BGs as well as ION neurons. To determine the degree to which these defects are the result of cell autonomous vs. non-cell-autonomous requirements for Ttbk2 in the PCs, we used the PC-specific Cre line Pcp2-Cre, which drives recombination specifically in PCs within the cerebellum beginning at P6 (Zhang et al., 2004). At P30, Ttbk2^fl/fl;Pcp2-Cre⁺ (referred to from here as Ttbk2^Pcp2) animals have normal cerebellar structure throughout, with molecular layer thickness comparable to that of littermate Control animals (Figure 3A,C. 202.7 μm ± 3.51 in P30 Control vs. 191.5 μm ± 3.21 in P30 Ttbk2^Pcp2). The VGLUT2+ synapses between climbing fibers and PCs are not significantly changed between P30 Control and Ttbk2^Pcp2 animals (Figure 3A,D; 548.2 puncta ± 13.36 in P30 Control vs. 538.9 puncta ± 18.14 in P30 Ttbk2^Pcp2). This indicates that despite postnatal loss of Ttbk2, initial connections between PCs and climbing fibers are established normally. By P90, however, Ttbk2^Pcp2 animals exhibited phenotypes largely recapitulating those observed in the Ttbk2^c.mut animals. At P90, numbers of primary cilia are significantly reduced on PCs of Ttbk2^Pcp2 animals (Figure 3—figure supplement 1A,B; 48.93 ± 7.86 percent PCs ciliated in Control vs. 18.67 ± 11.07 percent PCs ciliated in P90 Ttbk2^Pcp2). While molecular layer thickness between the P90 Control and Ttbk2^Pcp2 was not changed (Figure 3B,C; 202.9 μm ± 2.11 in P90 Control vs. 197.4 μm ± 4.18 in P90 Ttbk2^Pcp2), we see a significant decrease in VGLUT2 puncta throughout the cerebellum (Figure 3B,D; 549.9 puncta ± 15.47 in P90 Control vs. 476.9 puncta ± 15.82 in P90 Ttbk2^Pcp2), indicating that PCs have started to lose these important connections from the climbing fiber synapses.

We then assessed motor coordination of P30 and P90 Ttbk2^Pcp2 animals using the rotarod performance test and did not observe significant changes in P30 animals. However, by P90, the Ttbk2^Pcp2 animals consistently exhibited reduced latency to fall on both the accelerating rotarod as well as the steady speed rotarod performance tests (Figure 3E and F). These data show that loss of Ttbk2, specifically from PCs, causes neurodegenerative phenotypes.

Our data from the Ttbk2 global conditional knockouts revealed that the morphology of BGs was modestly perturbed (Figure 2C,D). As defects in BGs have been shown to non-cell autonomously contribute to the degenerative phenotypes observed in SCA7 (Furrer et al., 2011), we tested whether deletion of TTBK2 specifically from these cells could also result in loss of synapses and other degenerative changes to the PCs. We crossed Ttbk2^fl/fl animals to a Slc1a3-CreER mouse (Wang et al., 2012) to produce Ttbk2^fl/fl; Slc1a3-CreER+ mice to induce recombination of the Ttbk2 allele specifically within glial cells. Following the same TMX injection protocol used for the Ttbk2^c.mut experiments, we did not see changes to the VGLUT2+ synapses on PC dendrites (Figure 3G–I). These data indicate that the PC phenotypes observed in the Ttbk2^c.mut mice are primarily cell autonomous.

In the first 3 months following TMX injections, the phenotypes exhibited by the Ttbk2^c.mut mice consisted mainly of altered synaptic connectivity between PC and ION climbing fibers, and accompanying deficits in motor coordination (Figure 1). However, when we assessed the cerebellar phenotypes of animals at 6 months of age (5 months following TMX injection), we found gaps in the molecular layer where PCs appear to be absent (Figure 4A). We quantified this observation by counting PC soma within a defined region of the primary fissure, and confirmed that the number of PCs is reduced in 6-month-old Ttbk2^c.mut mice compared to littermate Controls of the same age, as well as compared to 4-month-old Ttbk2^c.mut mice (Figure 4B; 18.5 ± 0.29 PCs per 500 μm for 4-month Control vs. 18.42 PCs ± 0.34 for 4-month Ttbk2^c.mut; 18.67 PCs ± 0.43 for 6-month Control vs. 11.92 PCs ± 0.74 for 6-month Ttbk2^c.mut). We found more PC gaps in folia of 6-month-old Ttbk2^c.mut animals compared to Controls, and that most gaps are enriched at the inner folia. PC gaps were not found on folia X, which is consistent with data showing folia X being resistant to neurodegeneration (Figure 4C and Figure 4—figure supplement 1A) (Tolbert et al., 1995).

A postmortem examination of a SCA11 affected individual revealed Tau aggregates in regions of the brain outside of the cerebellum (Houlden et al., 2007). We therefore looked for pathological Tau aggregates in the cortex in both the 4-month-old and 6-month-old Ttbk2^c.mut animals. We could not detect the accumulation of phosphorylated Tau in the Control or 4-month-old Ttbk2^c.mut cortex. However, in the cortex of the 6-month-old Ttbk2^c.mut animals, we noticed a small number of neurons with some accumulation of phosphorylated Tau, recognized by an antibody specific to Ser202 and Thr205 phosphorylated tau (Figure 4—figure supplement 2A). This mild accumulation is compared to neurons in a mouse model of Alzheimer’s, JNPL3(P301L), which expresses a mutated form of Tau (Lewis et al., 2000) and is therefore positive for accumulation of phosphorylated Tau (Figure 4—figure supplement 2B). Thus, dysfunction of PCs after loss of Ttbk2, accompanied by Tau accumulation outside of the cerebellum, recapitulates currently described SCA11 phenotypes.

Our prior work demonstrated that mutations associated with SCA11, which result in the production of a truncated protein, interfere with the function of full-length TTBK2. In particular, these mutations dominantly interfere with cilia formation in embryos and cultured cells (Bowie et al., 2018). Throughout the adult cerebellum and other regions of the hindbrain, neurons possess primary cilia (Figure 5A–C). Within 20 days following administration of TMX to induce recombination (P45), the number of ciliated cells in the cerebellum declined dramatically in Ttbk2^c.mut mice: from a mean of 22.46 cilia per 32 mm² field ± 0.7626 in Control animals to 2.36 cilia per 32 mm² field ± 0.3103 in Ttbk2^c.mut animals (Figure 5D,E). This loss of cilia was observed throughout the cerebellum, brain stem, and other areas of the brain such as the hippocampus and the cortex (Figure 5—figure supplement 1A,B). Thus, loss of cilia coincides with the behavioral changes we identified in Ttbk2^c.mut mice, yet precedes the cellular changes ocurring throughout the cerebellum of Ttbk2^c.mut mice as they age.

Ttbk2 is essential both for the initiation of cilium assembly as well as the structure and stability of cilia (Bowie et al., 2018; Goetz et al., 2012). Given this critical link between TTBK2 and primary cilia in all cell types examined in both developing and adult tissues, we tested whether loss or dysfunction of cilia via a different genetic mechanism causes convergent phenotypes to those of the Ttbk2^c.mut mice. For these studies we turned to conditional mutants of another key ciliary protein, Intraflagellar Transport Protein 88 (IFT88). IFT88 is a component of the IFTB particle required for assembly of the ciliary axoneme as well as anterograde trafficking within the cilium (Pazour et al., 2000). Our previous work shows that IFT88 functions downstream of TTBK2 in cilium initiation (Goetz et al., 2012), with TTBK2 being required for IFT recruitment. In the developing and postnatal brain, IFT88 is important for cilia structure in the hippocampus and cortex (Willaredt et al., 2008) and when knocked out in these specific neuron populations results in memory deficits (Berbari et al., 2014). Additionally, Ift88 null mutants exhibit nearly identical embryonic phenotypes to those of Ttbk2 null mutants (Murcia et al., 2000). When we knocked out Ift88 using the same approach described for Ttbk2^c.mut animals, we observed that the numbers of cilia were significantly reduced in Ift88^c.mut cerebella at 3 months post TMX treatment, although more cilia remain in Ift88^c.mut cerebella compared to the Ttbk2^c.mut animals with the same treatment (Figure 6A,C; 17.31 cilia per 32 mm² field ± 0.65 for Control vs. 12.13 cilia per 32 mm² field ± 0.50 for Ift88^c.mut). Western blot analysis of cerebellar tissue from Ift88^c.mut mice reveals that a small amount of IFT88 protein perdures in brain tissue (Figure 6B). This could, in part, help to explain why we do not see a full loss of cilia throughout the cerebellum similar to that observed in the Ttbk2^c.mut mice. Regardless, the cilia that do remain in Ift88^c.mut animals are shorter in length than Controls (Figure 6D; 2.31 μm ± 0.10 for Control vs. 1.70 μm ± 0.08 for Ift88^c.mut).

To further characterize the remaining cilia in the brains of Ift88^c.mut mice, we examined additional markers of the ciliary membrane, including adenylate cyclase 3 (AC3) (Guadiana et al., 2016). We observed that within the WT cerebellum, there exist cilia that are AC3+ as well as AC3+/ARL13B+ (Figure 6E, arrowhead, Figure 6F). In our Ift88^c.mut animals, we observed that the numbers of AC3+ cilia were strongly reduced (Figure 6G: 7.47 AC3+ cilia per 32 mm² field ± 0.41 in Control vs. 2.17 AC3+ cilia per 32 mm² field ± 0.22 in Ift88^c.mut). This analysis suggests that IFT88 is required forlocalization of specific signaling molecules such as AC3 to neuronal primary cilia throughout the cerebellum.

We then examined cerebellar structure and circuitry in Ift88^c.mut animals. Similar to our findings in Ttbk2^c.mut animals, no changes to PC number were evident at 3 months post TMX treatment. Molecular layer thickness was reduced in Ift88^c.mut animals (Figure 7A,B; 182.3 μm ± 2.5 in Control vs. 169.3 μm ± 3.04 in Ift88^c.mut). Similarly, VGLUT2 puncta were reduced in Ift88^c.mut compared to Controls (Figure 7A,C; 515.7 puncta ± 20.58 in Control vs. 395.6 puncta ± 13.7 in Ift88^c.mut). We also tested Ift88^c.mut animals on the rotarod performance test to uncover any motor coordination deficits, given that these mice exhibit similar cellular changes to those of the Ttbk2^c.mut animals. These tests revealed that Ift88^c.mut animals also have a shorter latency to fall time on the steady speed rotarod performance test, but not on the accelerating rotarod performance test (Figure 7D,E), while Ttbk2^c.mut animals have a shorter latency to fall time on both accelerating and steady speed rotarod performance tests (Figure 1A,B). Steady speed rotarod analysis is thought to more accurately detect motor coordination deficits, while the accelerating rotarod test can also be affected by mouse fatigue (Monville et al., 2006). Taken together, these data show that loss of IFT88 from the adult brain results in impaired ciliary structure and similar defects in cerebellar architecture and locomotor behavior to those observed in the animals lacking TTBK2.

We further assessed whether the Ift88^c.mut animals also lose PCs as they age, as was the case for the Ttbk2^c.mut mice (Figure 4). 6-month-old Ift88^c.mut animals show gaps throughout the PC layer, and have reduced numbers of PC soma (Figure 8A,B; 17.5 ± 0.44 per 500 μm in 4-month-old Control vs. 16.92 PC soma ± 0.31 in 4-month-old Ift88^c.mut. 17.17 PC soma ± 0.55 in 6-month-old Control vs. 12.67 PC soma ± 0.43 in 6-month-old Ift88^c.mut). Coupled with these findings, the molecular layer thickness is further reduced in 6-month-old Ift88^c.mut animals (Figure 8C–E; 173.9 μm ± 2.28 in 6-month-old Control vs. 158.0 μm ± 1.63 in 6-month-old Ift88^c.mut), as well as VGLUT2 puncta counts being diminished in 6-month-old Ift88^c.mut (Figure 8C,D,F; 645.9 puncta ± 26.83 in 6-month-old Control vs. 461.4 puncta ± 25.42 in 6-month-old Ift88^c.mut).

The use and care of mice as described in this study was approved by the Institutional Animal Care and Use Committees of Duke University (Approval Number A218-17-09). All animal studies were performed in compliance with internationally accepted standards.

Ttbk2^c.mut mice were produced by crossing Ttbk2^{tm1a(EUCOMM)Hmgu} mice to ACTB:FLPe (Jax stock #003800). The following mice were purchased from Jackson Laboratories: Ift88^flox (stock #022409), Ubc-CreER (stock #007001), Slc1a3-CreER (stock #012586) and Pcp2-Cre (Jax stock #010536). Slides used from JNPL3(P301L) mice were a gift from Dr. Carol Colton at Duke University.

PCR genotyping was performed on all mice before experiments to confirm the presence of floxed alleles and Cre. Ttbk2-floxed allele, primers used: 5’ ATACGGTTGAGATTCTTCTCCA, 3’ AGGCTGTACTGTAACTCACAAT (WT band 978 bp, floxed band 1241 bp). Ift88-floxed allele, primers used: 5’ GCCTCCTGTTTCTTGACAACAGTG, 3’ GGTCCTAACAAGTAAGCCCAGTGTT (WT band 350 bp, floxed band 370 bp). Universal Cre (Ubc-CreER, Pcp2-Cre), primers used: 5’ GATCTCCGGTATTGAAACTCCAGC, 3’ GCTAAACATGCTTCATCGTCGG (transgene band 650 bp).

Tamoxifen powder (Sigma T5648) was dissolved in corn oil (Sigma C8267) to a desired concentration of 20 mg/mL. Mice were given five consecutive 100 μL intraperitoneal injections of 20 mg/mL tamoxifen starting at P21. Control mice were given corn oil vehicle only.

To harvest tissues from adult mice, animals were deeply anesthetized with 12.5 mg/mL avertin and transcardially perfused with 10 mL of phosphate buffered saline (PBS) followed by 20 mL of 4% paraformaldehyde (PFA). Whole brains were dissected out and left to incubate for 24 h in 4% PFA at 4°C. For cryosectioning, tissue was cryoprotected in 30% sucrose overnight and embedded in Tissue Freezing Medium (General Data TFM-5). Cerebella were then cut sagittally down the middle, and embedded in Tissue Freezing Medium (General Data TFM-5). Tissue was sectioned at 20–30 μm thickness on a Leica Cyrostat (model CM3050S).

Western blot methods were done as previously described (Bouskila et al., 2011) (Bowie et al., 2018). Briefly, for tissue which was being used to quantify levels of TTBK2, a buffer containing 50 mM Tris/HCl, pH 7.5, 1 mM EGTA, 1 mM EDTA, 1 mM sodium orthovanadate, 10 mM sodium-2-glycerophosphate, 50 mM sodium fluoride, 5 mM sodium pyrophosphate, 0.27 M sucrose, 1 mM benzamidine and 2 mM PMSF, supplemented with 0.5% NP-40 and 150 mM NaCl was used. For other tissue samples a buffer containing 10 mM Tris/Cl pH 7.5, 150 mM NaCl, 0.5 mM EDTA, 1% Triton, 1 mM protease inhibitors (Sigma #11836170001) and 25 mM β-glycerol phosphate (Sigma 50020) was used. Total protein concentration was determined using a BSA Protein Assay Kit (Thermo Fisher #23227). For western blots, 15 μg of protein lysate was used for detection.

All quantification of cerebellar tissue was done using ImageJ software. Images taken for quantification of cilia abundance were 10 μm z-stacks taken at 63x in four distinct folia regions of the cerebellum, two rostral and two caudal (specifically, the outer edge of folia I/II, the internal zone between folia III and IV, the tip of folia VI, and the outer edge of folia IX on a sagittal section were imaged for cilia quantification). The Purkinje cell layer was placed into the middle of the image with equal distance above and below for quantification. Per animal, four sections were scored each and three animals were included in all quantifications. These cilia were the same population taken for cilia length measurements as well.

For the molecular layer thickness, images were taken at 20x along the entirety of a primary fissure. A line was drawn from the base of the molecular layer to the pial surface, and a measurement was recorded. For this same line, the top of the line measurement was then brought down to the distal extent of the VGLUT2 synapse area, and a measurement recorded. For consistency, only the caudal side of the folia was measured. For the VGLUT2 puncta analysis, the ‘Analyze Particles’ function in ImageJ was used. Each image for the VGLUT2 puncta analysis was taken at 63x on the caudal side of the primary fissure, a 10 μm z-stack was made, and the image quantified. For the quantification, each stack was made into a black and white image, where the VGLUT2 puncta were black against a white background. Thresholding was performed, and the Analyze Particle function used. These measurements were routinely tested against user ROI counting to confirm accuracy. Four cerebellar slices were imaged per animal, and three animals were included in the analysis.

For the area measurements of the ION nuclei, the ION was identified by cells that were positive for both NeuN and Calbindin as well as location within ventral medulla in which these cells reside. Images used for the NeuN area analysis were taken at 20x. A 10 μm z-stack image was made, and using the line tool, outlines were carefully drawn around the NeuN positive neuron and the area recorded. Per animal, over 150 cells were measured and three animals were included in the quantification.

Glial fibers were assessed as previously described (Furrer et al., 2011). Briefly, a 100 μm horizontal line was drawn 50 μm below the pial surface of the primary fissure folia. Glial fibers which crossed this 100 μm were scored. Per animal, 36 measurements were made and three animals were included in the quantification.

The following antibodies and dilutions were used in this study: mouse anti-ARL13B (NeuroMabs N295B/66, 1:500), rabbit anti-ARL13B (gift from Tamara Caspary, 1:500, and Proteintech 17711–1-AP, 1:500), mouse anti-gamma-Tubulin (Sigma T6557, 1:1000), rabbit anti-Calbindin D28K (Cell Signaling Technologies 13176S, 1:250), guinea pig anti-Calbindin D28K (Synaptic Systems 214–004, 1:200), guinea pig anti-VGLUT2 (EMD Millipore AB2251, 1:2500), rabbit anti-NeuN (Abcam ab177487, 1:1000), DAPI (Sigma D9542, 1x), rabbit anti-AC3 (Santa Cruz SC-588, 1:10 - discontinued), rabbit anti-AC3 (Abeomics 34–1003, 1:100), chicken anti-GFAP (EMD Millipore AB5541, 1:500), rabbit anti-FoxP2 (Abcam, ab106046, 1:400), rabbit anti-IP3 (Abcam, ab108517, 1:200), and mouse anti-AT8 (Thermo Scientific MN1020, 1:100).

For immunostaining cerebellar tissue, sections were rinsed in 1xPBS to remove OCT and permeabilized in 0.2% PBS-T (PBS + 0.2% Triton X-100) for 10 min, and then rinsed 3 × 5 min in PBS before the blocking step. Blocking solution contained 5% serum, 1% BSA made up in 0.1% PBS-T, and sections were incubated at room temperature in blocking solution for 1 h. Primary antibodies were used at indicated dilutions and incubated at 4°C overnight. Following primary antibody incubation, slides were rinsed 3 × 5 min in 1xPBS and secondary antibodies were used to detect epitopes. All secondary antibodies were supplied from Life Technologies. Secondary antibodies incubated for 1–3 h at room temperature. Following secondary antibody incubation, slides were rinsed 3 × 5 min in 1xPBS and mounted with either ProLong Gold antifade reagent (Invitrogen P23930).

For Golgi staining of cerebellar tissue, FD Rapid Golgistain kit (FD Neurotechnologies) was used, according to manufacturer’s instructions. 100 μm sections were made after staining. Imaging of Golgi stained PCs was completed using a Zeiss Axio Imager with a 40x objective. Z-stacks containing entire PCs were used for quantification, and proximal dendrites were chosen that did not intersect with other dendrites for clear quantification. Using ImageJ software, spines were counted per length of measurement and reported as number of spines per micron.

A rotarod performance test was completed with help from the Duke University Mouse Behavioral and Neuroendocrine Core Facility. Testers were blind to mouse genotype before beginning any experiments. The accelerating rotarod testing was performed the day before steady state rotarod testing. All accelerating tests were conducted at a speed that increased from 4 RPM to 40 RPM over 5 min. All steady speed tests were conducted at 32 RPM. Four trials were conducted per test. A trial was stopped after 300 s maximum time had elapsed for mice that did not fall off the rotarod during testing. Mice were aborted from the trial run if they held onto the rotarod for three full rotations. Mice were given 30 min between trials to rest, and four trials were completed per test.

Statistical analyses, p-values, and experimental numbers for all experiments are outlined in respective figure legends. Analyses were performed using Graph Pad Prism 8.