Mapping SCA1 regional vulnerabilities reveals neural and skeletal muscle contributions to disease

Spinocerebellar ataxia type 1 (SCA1) is a fatal neurodegenerative disease caused by an expanded polyglutamine tract in the widely expressed ataxin-1 (ATXN1) protein. To elucidate anatomical regions and cell types that underlie mutant ATXN1-induced disease phenotypes, we developed a floxed conditional knockin mouse ( f-ATXN1 146Q/2Q ) with mouse Atxn1 coding exons replaced by human ATXN1 exons encoding 146 glutamines. f-ATXN1 146Q/2Q mice manifested SCA1-like phenotypes including motor and cognitive deficits, wasting, and decreased survival. Central nervous system (CNS) contributions to disease were revealed using f-ATXN1 146Q/2Q ; Nestin-Cre mice, that showed improved rotarod, open field, and Barnes maze performance by 6-12 weeks-of-age. In contrast, striatal contributions to motor deficits using f-ATXN1 146Q/2Q ; Rgs9-Cre mice revealed that mice lacking ATXN1 146Q/2Q in striatal medium-spiny neurons showed a trending improvement in rotarod performance at 30 weeks-of-age. Surprisingly, a prominent role for muscle contributions to disease was revealed in f-ATXN1 146Q/2Q ; ACTA1-Cre mice based on their recovery from kyphosis and absence of muscle pathology. Collectively, data from the targeted conditional deletion of the expanded allele demonstrated CNS and peripheral contributions to disease and highlighted the need to consider muscle in addition to the brain for optimal SCA1 therapeutics.


INTRODUCTION
Neurodegenerative diseases are often characterized by a variety of clinical phenotypes.Thus, two fundamental issues facing neurodegenerative disease research are 1) the identification of the anatomical/cellular features underlying each disease symptom, and 2) assessing the extent to which disease processes in these regions are similar at a molecular level.Progress in these areas provides vital information for developing treatments to ensure maximum therapeutic efficacy.In addition, mapping the anatomical basis of disease-associated phenotypes/symptoms is a means of further understanding CNS function at a systems level.Spinocerebellar ataxia type 1 (SCA1), an autosomal dominant neurodegenerative disease, is one of the neurodegenerative diseases caused by expansion of a CAG trinucleotide tract that encodes a polyglutamine stretch in the resulting ataxin-1 (ATXN1) protein (1,2).An initial symptom of SCA1 is typically gait ataxia.As the disease progresses, common symptoms include dysarthria, muscle wasting, and cognitive deficits such as executive functioning difficulties.Late-stage disease involves bulbar dysfunction, which is thought to cause the swallowing and breathing difficulties that underlie premature death (3)(4)(5)(6)(7).The typical pathological pattern is olivopontocerebellar atrophy with loss of cerebellar Purkinje cells, major neuronal loss in the dentate nuclei, and extensive olivary neuronal loss.
Basal pontine neuronal atrophy is variable and, in some cases, severe.Other brainstem areas affected are the red nuclei, the vestibular nuclei, and motor cranial nerve nuclei (4,7).
In the spinal cord, the anterior horns, posterior columns, and spinocerebellar tracts also have atrophy with variable sparing of the pyramidal tracts.The pars compacta of the substantia nigra is relatively spared, but the striatum including the putamen, pallidum and thalamus can have mild involvement.The basal forebrain cholinergic nuclei, cerebral cortex, and hippocampus may have mild neuronal loss (4,7).Previous studies demonstrate the utility of Atxn1 knockin mice, in which an expanded stretch of CAG repeats is inserted into one Atxn1 allele, as a model for studying the spectrum of SCA1-like phenotypes (8).In this study, we describe the generation and characterization of a conditional SCA1 knockin mouse model in which the two murine Atxn1 coding exons, 7 and 8, in one allele were replaced with the two human ATXN1 exons encoding the entire human ATXN1 protein; i.e. human exon 8, containing an expanded 146 CAG repeat, and human exon 9 which includes the stop codon and 3'UTR.LoxN recombination sites flanked the human ATXN1 exons such that the human coding exons of the expanded ATXN1 146Q/2Q allele are deleted in the presence of Cre recombinase.Heterozygous floxed humanized ATXN1 146Q/2Q (f-ATXN1 146Q/2Q ) mice manifested a very similar collection of SCA1-like phenotypes as seen in Atxn1 154Q knockin mice (8), i.e. a progressive neurological disorder featuring motor incoordination, cognitive deficits, wasting with kyphosis, and decreased survival.The large, expanded allele of both the original SCA1 knockin mouse and the one generated in this study render the mice an excellent and accurate genetic model for juvenile onset SCA1, which typically shows much broader phenotypes, but still allows studies of regional vulnerability to disease.To this end, we crossed f-ATXN1 146Q/2Q mice with Nestin-Cre and Rgs9-Cre mice to reveal contributions of all CNS cells and striatal medium spiny neurons (MSNs) to SCA1-like phenotypes, respectively.Further, f-ATXN1 146Q/2Q were crossed to ACTA1-Cre mice, to assess the contribution of skeletal muscle to SCA1-like phenotypes.By biochemical and image analyses comparing f-ATXN1 146Q/2Q ;ACTA1-Cre mice in which f-ATXN1 146Q was deleted from muscle in f-ATXN1 146Q/2Q mice, we showed that ATXN1[146Q] induced a strong peripheral muscle specific phenotype.The findings from studying each of these sub-populations revealed that muscle involvement is prominent in this disease model while the striatal contribution is less prominent and might be a contributor in the late stage to disease progression.

Generation and characterization of the conditional knockin f-ATXN1
146Q/2Q SCA1 mouse model.A conditional knockin mouse model designated f-ATXN1 146Q/2Q was developed to investigate brain region and peripheral tissue contributions to SCA1-like phenotypes as well as to obtain a platform for testing human ATXN1 gene targeting therapies.The entire coding region of one allele of the mouse Atxn1 gene was replaced with that of the human ATXN1 gene in a two-step process.First, the 3' end of the Atxn1 gene (20kb) including the last two exons through the end of the transcript (Figure 1A) was deleted from the genome of a mouse ES cell line (C57BL/6N) and replaced by Frt recombination sites and selectable markers (Figure 1B).Next, the syntenic sequence from the human genome (31kb) was inserted using site-specific recombination at flanking FRT sites (Figure 1C).The ES cell was injected into C57BL6 embryos yielding f-ATXN1 146Q/-mice, which were then crossed with WT mice to yield f-ATXN1 146Q/2Q mice.With respect to coding sequences, these mice have one mouse WT allele and one disease-causing human allele encoding ATXN1 containing 146 glutamine repeats.
To evaluate the ability of the ATXN1 146Q allele to be deleted by Cre recombinase, f-ATXN1 146Q/2Q mice were crossed with Sox2-Cre mice (9).This Sox2-Cre mouse mediates efficient Cre-mediated recombination at gastrulation in all epiblast-derived cells including the primordial germ cells.Using CAG repeat-specific and recombination-specific PCR primers (Figure S1A), recombination and deletion of the ATXN1 146Q allele was detected in all tissues examined from a f-ATXN1 146Q/2Q ;Sox2-Cre mouse (Figures S1B-D).

Impact of deleting the ATXN1 146Q allele from the CNS on SCA1-like phenotypes.
Nestin-Cre mice express Cre recombinase in neuronal and macroglial precursors and thus throughout the CNS (11).f-ATXN1 146Q/2Q ;Nestin-Cre mice were examined to assess the extent to which deletion of the f-ATXN1 146Q allele from CNS neuronal and glial cells improved other SCA1-like neurological phenotypes manifested by f-ATXN1 146Q mice.f-ATXN1 146Q/2Q ;Nestin-Cre mice had a remarkable and significant improvement in performance in the open-field test (locomotion) compared to f-ATXN1 146Q/2Q mice at 12 weeks-of-age (Figure 2A).f-ATXN1 146Q/2Q ;Nestin-Cre mice also had normalized cognitive performance as evaluated by the Barnes Maze test.Compared to controls WT-Atxn1 2Q/2Q   and WT-Atxn1 2Q/2Q ;Nestin-Cre mice, f-ATXN1 146Q/2Q mice showed a significant delay in their first entry into the escape hole zone while f-ATXN1 146Q/2Q ;Nestin-Cre mice did not (Figure 2B).f-ATXN1 146Q/2Q ;Nestin-Cre mice had significant improvement in survival (Figure 2C) and a small but trending improvement in weight compared to f-ATXN1 146Q/2Q , significant only for mice at 38 weeks.Overall, their weight remained reduced compared to wild type mice (Figure 2D).In addition, at 36 weeks-of-age, f-ATXN1 146Q/2Q displayed a hindlimb clasping phenotype characteristic of mice with pathology in a variety of brain regions (Figure 2E) (12).The hindlimb clasping phenotype was absent in f-ATXN1 146Q/2Q ;Nestin-Cre mice at this age.
Naïve f-ATXN1 146Q/2Q mice have a significant deficit in performance on the accelerating rotarod at 6 weeks, which progresses in severity with age (Figures 3A, 3B and S4).Notably, f-ATXN1 146Q/2Q ;Nestin-Cre mice with the ATXN1 146Q allele deleted throughout the CNS (11) displayed improved rotarod performance, practically to wild-type levels, at all ages tested, from 6 to 30 weeks-of-age (Figures 3A and S4B).Previous studies in which the cerebella of 5 week old SCA1 knockin mice were injected with adeno-associated viruses expressing inhibitory RNAs targeting Atxn1 showed rescued rotarod performance, indicating a critical role of the cerebellum in the motor coordination deficit seen in SCA1 (13,14).Of note, in SCA1 patients, changes in striatal volume are associated with an age-associated decline in motor performance (15,16).Thus, as a first step in assessing regions of the CNS in addition to the cerebellum where dysfunction might contribute to altered rotarod performance with increasing age, f-ATXN1 146Q/2Q mice were crossed to Rgs9-Cre mice to delete the ATXN1 146Q allele from striatal MSNs (Figure S5) (17,18).Interestingly, deletion of ATXN1 146Q from striatal MSNs trended towards partial improvement of rotarod performance that did not manifest until 30 weeks-of-age in f-ATXN1 146Q/2Q ;Rgs9-Cre mice relative to f-ATXN1 146Q/2Q mice (Figures 3B and S4C).While f-ATXN1 146Q/2Q mice significantly declined over time, ATXN1 146Q/2Q ;Rgs9-Cre were not significantly different compared to WT-Atxn1 2Q/2Q or WT-Atxn1 2Q/2Q ;Rgs9-cre mice.Deletion of ATXN1 146 from striatal MSNs significantly improved dopamine-and cAMP-regulated phosphoprotein-32 (DARPP-32) expression in the striatum of f-ATXN1 146Q/2Q ;Rgs9-Cre mice (Figures S6A and B).
Previous studies show extensive somatic expansion of the ATXN1 CAG repeat in SCA1 patient and knockin mice striatum (19,20).Analyses of 35 week f-ATXN1 146Q/2Q mice also revealed high levels of somatic expansion in striatum.CAG expansion was also observed in other tissues/brain regions, most notably the medulla (Figures 3C and D), with f-ATXN1 146Q/2Q mice exhibiting a brain regional expansion profile similar to that of an SCA1 patient (20).
Striatal expansion was significantly reduced in f-ATXN1 146Q/2Q ;Rgs9-Cre mice (Figures 3C and   D), demonstrating that expansions occur in MSNs.Residual striatal expansion in f-ATXN1 146Q/2Q ;Rgs9-Cre mice may be explained by incomplete Cre-mediated excision of the expanded CAG repeat-containing allele in MSNs and/or CAG expansion in other striatal cell types (Figure S5A).We speculate that expansion of expanded ATXN1 CAG in the striatum contributes to mild deficits in motor performance with disease progression in f-ATXN1 146Q/2Q mice.That CAG expansion occur in the striatal MSNs is interesting.However, the data indicate that such expansions may contribute to striatal pathogenesis only late in the course of disease.

Deletion of the f-ATXN1
146Q allele from skeletal muscle reveals a direct muscle-specific pathogenic effect of ATXN1[146Q].It has been noted that muscle wasting is observed in SCA1 patients, and in mice muscle myopathy could be contributing to kyphosis (21).Figure 4A shows that f-ATXN1 146Q/2Q mice manifest a severe kyphosis by 20 weeks-of-age with increasing severity to 50 weeks-of-age.We examined the CNS impact of ATXN1[146Q] on kyphosis by deleting ATXN1[146Q] from all CNS neurons, including spinal motor neurons, by crossing f-ATXN1 146Q/2Q mice with Nestin-Cre mice.Image analysis showed that kyphosis in f-ATXN1 146Q/2Q ;Nestin-Cre mice remained severe (Figure 4B).To examine whether kyphosis mice were crossed with transgenic ACTA1-Cre (HSA-Cre79) mice to delete the ATXN1 146Q allele from muscle (22).In f-ATXN1 146Q/2Q ;ACTA1-Cre mice, expression of the ATXN1 146Q allele was significantly reduced in skeletal muscle, including the tongue and diaphragm but not in the heart, cerebral cortex, cerebellum, medulla, or lung (Figures S3C and S3E).
mice restoring the spinal configuration similar to that seen in wild type mice at 20 and 50 weeks-of-age (Figures 4C and 4D).Using an approach to calculate a kyphosis index (KI), the KI of f-ATXN1 146Q/2Q mice is significantly reduced compared to the KI of WT-Atxn1 2Q/2Q mice (Figures 4E and 4F).Notably, deletion of the ATXN1 146Q allele from muscle restored the KI to that of WT-Atxn1 2Q/2Q mice.In addition, the grip strength test used to measure the neuromuscular function as maximal muscle strength showed significant rescue when ATXN1 146Q was deleted from muscle in the ACTA1-Cre mice but not the Nestin-Cre mice (Figure 4G and H).Interestingly, there was a slight but significant difference between the WT-Atxn1 2Q/2Q vs WT-Atxn1 2Q/2Q ; Nestin-Cre mice in grip strength at 18 wks.Of note, the Nestin-Cre transgenic line trended slighter smaller (Figure 2D) presumably due to this Cre line being affected by mild hypopituitarism which could impair grip strength (23).These data support the concept that kyphosis and the deficit in grip strength in f-ATXN1 146Q/2Q mice are the result of muscle dysfunction induced by expression of ATXN1[146Q] in muscle.
Consistent with an ATXN1[146Q]-induced muscle phenotype, tibialis anterior (TA) and extensor digitorum longus (EDL) muscle masses were lower in f-ATXN1 146Q/2Q mice compared to WT-Atxn1 2Q/2Q ;ACTA1-Cre or f-ATXN1 146Q/2Q ;ACTA1-Cre mice at both 18 and 30 weeks (Figures 5A and 5G).Additional characterization of the effects of ATXN1[146Q] on skeletal muscle function revealed a progressive, muscle specific myopathy.Further evaluation of muscle strength was performed using an in vivo electrode-based assay that measures dorsiflexor muscle torque via direct stimulation of the peroneal nerve.Dorsiflexor (EDL + TA) torque was significantly lower at 18 wks and trending lower at 30 weeks-of-age, in f-ATXN1 146Q/2Q mice compared to WT-Atxn1 2Q/2Q ;ACTA1-Cre mice, despite correcting for body mass or muscle mass (Figures 5B, 5C and 5H, 5I).To further investigate the ATXN1[146Q]-induced muscle weakness phenotype, we used an ex vivo assay that assesses EDL contractile function independent of endogenous motor neuron activation.At 18 weeks-of-age, EDL specific force production was not different between genotypes suggesting that intrinsic muscle contractility, independent of motor neuronal activation, was not affected in f-ATXN1 146Q/2Q mice (Figure 5D).However, at 30 wks, EDL specific force was lower in the f-ATXN1 146Q/2Q mice, a phenotype that was fully corrected with deletion of the ATXN1 146Q allele from muscle (Figure 5J).The in vivo but not ex vivo muscle strength deficits in f-ATXN1 146Q/2Q mice at 18 weeks suggest the presence of a neuronal pathology potentially related to disrupted motor neuron activation or neuromuscular junction (NMJ) dysfunction.Consistent with possible alterations in NMJ function, at eight weeks-of-age increased expression of mRNAs encoding the NMJ proteins MuSK and CHRNA1 (24) were detected (Figure S6C).At 40 weeks-of-age expression of Musk and Chrna1 RNAs were restored to normal levels in f-ATXN1 146Q/2Q ;ACTA1-Cre mice (Figure 6A) but not in f-ATXN1 146Q/2Q ;Nestin-Cre mice (Figure 6B).Furthermore, the reduction in EDL specific force observed at 30 weeks in f-ATXN1 146Q/2Q mice could indicate a progressive myopathy stemming from a combination of chronic muscle disuse and alterations in intrinsic muscle contractile function.Importantly, all muscle deficits were restored to WT-Atxn1 2Q/2Q ;ACTA1-Cre levels in mice with the ACTA1-Cre induced deletion of the ATXN1 146Q allele.Together, these results indicate the presence of a progressive myopathy that eventually affects intrinsic muscle function.
Histological analysis of muscle cross sections at 18 and 30 weeks revealed that total fiber number of the TA was not different between genotypes, but the f-ATXN1 146Q/2Q TA muscles had significantly smaller fiber cross-sectional area compared to WT-Atxn1 2Q/2Q ;ACTA1-Cre TA muscles (Figures 5E,F and 5M and 5K,L and 5N).Muscle fiber size was completely rescued in f-ATXN1 146Q/2Q ;ACTA1-Cre mice (Figures 5F and 5L).These data support the presence of a f-ATXN1 146Q/2Q specific skeletal muscle pathology that is characterized by muscle weakness and smaller muscle size.
Proper nuclear localization of mutant ATXN1 is critical for many disease-like phenotypes including motor dysfunction, cognitive deficits, and premature lethality (25).Like in f-ATXN1 146Q/2Q mice, smaller dorsiflexor muscle mass (Figure 7A), and smaller fiber size (Figure 7F) indicating the presence of a skeletal muscle myopathy are seen in Atxn1 175Q/2Q mice as early as 12 weeks-of-age.As in f-ATXN1 146Q/2Q mice, Atxn1 175Q/2Q mice had the same number of muscle fibers as WT-Atxn1 2Q/2Q mice (Figure 7E).Notably, this phenotype was partially corrected in Atxn1 175Q-K772T/2Q mice in which ATXN1's nuclear localization sequence is disrupted.The Atxn1 175Q-K772T/2Q muscles demonstrate partially recovered muscle mass (Figure 7A), and muscle fiber size (Figures 7F and 7G).Thus, as with the other SCA1-like disease phenotypes, we conclude that nuclear localization of expanded ATXN1 is important for the muscle-specific phenotypes caused by expanded ATXN1.
Deleting the f-ATXN1 146Q allele from muscle in the f-ATXN1 146Q/2Q ;ACTA1-Cre mice showed no improvement in open field, rotarod and hind limb clasping phenotypes (Figures S7A and C).In contrast, survival and wasting were significantly improved in f-ATXN1 146Q/2Q ;ACTA1-Cre mice (Figures S7D and E) supporting the concept that mutant ATXN1 induced muscle pathology contributes to these SCA1-like phenotypes.

Discussion
Neurodegenerative diseases are typically characterized by prominent pathology in specific brain regions/neuronal populations; yet many of these disorders impact other neural structures, circuits, and cell populations that lack obvious pathology.In SCA1, cerebellar Purkinje cell degeneration is a frequent and prominent pathological feature.However, ATXN1 is widely expressed and SCA1 patients present with symptoms linked to multiple different brain regions (7).In particular, individuals with large repeat expansion and juvenile onset have broader phenotypes including cognitive deficits, striatal dysfunction, and muscle wasting (10) highlighting that as the repeat grows more and more cells become vulnerable to disease.A question facing the understanding of SCA1 pathogenesis, as well as many other neurodegenerative diseases, is the extent to which vulnerability in specific cell populations contribute to each disease-associated phenotype.In this study, we addressed this question in SCA1 using a conditional mouse model with broad spatial mutant gene expression that matches that of humans.We found that deletion of the f-ATXN1 146Q allele from CNS neuronal and glial cells rescued SCA1-like neurological phenotypes seen in the f-ATXN1 146Q/2Q mice.Analysis of a key disease phenotype, deficit in motor performance on the rotarod, showed that this phenotype is very nicely rescued in Nestin-Cre mice and that it is predominantly of cerebellar origin because rescue in striatal MSNs was minimal and only very late in disease course.Additionally, we found direct pathological effects of f-ATXN1 146Q on muscle and pinpointed this contributes to wasting and kyphosis.
Like the Atxn1 154Q/2Q knockin mice (8), f-ATXN1 146Q/2Q mice developed neurological phenotypes similar to those seen in SCA1 patients.Motor incoordination on the rotarod was seen as early as 7 weeks-of-age in f-ATXN1 146Q/2Q mice and cognitive deficits became apparent as assessed by the Barnes maze test later at 24 weeks-of-age.Confirming the neural basis of these SCA1-like phenotypes, substantial rescue of rotarod performance (Figures 3A   and S4B), performance in the open field test (Figure 2A), and on the Barnes maze (Figure 2B) was observed in f-ATXN1 146Q/2Q mice crossed to Nestin-Cre mice.The Nestin-Cre line was designed to direct Cre recombinase expression to neuronal and macroglial precursors providing recombination exclusively in the CNS (11).Analysis of the tissue pattern of f-ATXN1 146Q deletion in f-ATXN1 146Q/2Q ;Nestin-Cre mice confirmed the CNS specificity of Nestin-Cre recombination (Figures S3A, S3B and S3D).As an initial effort to further dissect the anatomical basis of a key SCA1-like phenotype manifested by the f-ATXN1 146Q/2Q mice, we selected motor incoordination as measured by rotarod performance.Interestingly, while injection of an Atxn1-targeting siRNA into the cerebellum was able to improve rotarod performance at 10 weeks-of-age in Atxn1 154Q/2Q knockin mice (13), deletion of f-ATXN1 146Q from the striatum by crossing f-ATXN1 146Q/2Q mice to Rgs9-Cre mice decreased the decline in rotarod performance seen late in disease progression at 30 weeks-of-age.Correspondingly at diagnosis, magnetic resonance imaging (MRI) of SCA1 patients shows that loss of cerebellar volume is essentially complete.As SCA1 patients age, MRI analyses show a progressive loss in striatal volume (15,16).Moreover, a recent study found striatal volume to be a predictor of motor decline with increasing patient age after onset of ataxia (16).Our findings indicate that a relative time course of disease in cerebellum and striatum in f-ATXN1 146Q/2Q mice parallel the MRI findings in SCA1 patients.
Previous studies show that in the striatum somatic instability of expanded ATXN1 and HTT are similar to each other in knockin mice and patients (19,20,26).Here, we show that in f-ATXN1 146Q/2Q mice, striatal ATXN1 CAG expansion occurs in MSNs (Figures 3C and 3D).In contrast to HD where repeat expansion in striatal MSNs drives onset of disease, data presented here along with previous data indicate that in SCA1 expanded ATXN1 expression in the cerebellum drives disease onset and that striatal ATXN1 repeat expansion only contributes to age-dependent progression of motor deficits well after disease onset (Figures 3B and S4C).Thus, we reason that somatic instability alone does not drive HD and SCA1, and that the protein context and corresponding function of the protein in which the repeat expansion occurs is critical for driving the course of disease pathogenesis.
As disease progresses in SCA1 patients, muscle wasting is often seen (27)(28)(29)(30).Signs of motor neuron pathology were suggested as being linked to muscle wasting (31).In this study, we provide evidence that expanded ATXN1 has a direct pathological effect on skeletal muscles.
At 18 weeks, we observed that in vivo torque deficits were not corrected even after normalizing muscle strength by muscle size.This suggests a muscle force-generating pathology that could be stemming from a variety of muscle contraction-related elements ranging from motor neuron activation to myosin-actin cross bridge cycling.Interestingly, ex vivo force at 30 weeks was significantly lower in f-ATXN1 146Q/2Q mice than WT-Atxn1 2Q/2Q; ACTA1-Cre mice, which suggests that the muscle pathology could progress to more specific disruptions in intrinsic contractile function.Notably, every muscle deficit was restored to that seen in WT-Atxn1 2Q/2Q mice in f-ATXN1 146Q/2Q mice in which ACTA1-Cre induced deletion of the ATXN1 146Q allele.Overall, the data point to the presence of a neuromuscular pathology perhaps involving dysfunction at the NMJ.That NMJ function is impaired in f-ATXN1 146Q mice is supported by the finding that expression of Musk and Chrna1 RNA is altered in muscle of f-ATXN1 146Q mice.Further investigation is necessary to explore the contractile-specific effects of the muscle-specific ATXN1[146Q] mutation.
Notably, deletion of f-ATXN1 146Q from muscle substantially reduced kyphosis that, in mice, can be associated with paraspinal skeletal muscle impairment (21) as well as restored normal dorsiflexor strength, muscle mass and fiber size.As seen for other SCA1-like phenotypes, motor dysfunction, cognitive deficits, and premature lethality (25), proper nuclear localization of expanded ATXN1 was shown to be critical for muscle pathogenesis.
In conclusion, we demonstrate the utility of the f-ATXN1 146Q/2Q conditional mouse model in linking pathogenesis in specific anatomical regions/cell populations with SCA1phenotypes.The results reveal that neural and peripheral pathological effects of expanded ATXN1 contribute independently to disease presentation.This finding has important implications for design and administration of optimal SCA1 therapeutics, especially in juvenile cases or late in disease for adult onset cases.Notably, we show that the rotarod motor performance deficit is more complex in that while initially driven by pathology of cerebellar Purkinje cells, pathology in striatal MSNs might contribute to this phenotype at late stages of disease.Lastly, data presented provide strong evidence of muscle specific pathology in f-ATXN1 146Q/2Q mice.Thus, an ideal SCA1 therapeutic needs to subdue mutant ATXN1 toxicity in the CNS and peripheral muscle.

Sex as a biological variable
This study examined male and female mice and similar results were obtained for both sexes.

Mice
All mice were housed and managed by Research Animal Resources under specific pathogen-free conditions in an Association for Assessment and Accreditation of Laboratory Animal Care International approved facility.The mice had unrestricted access to food and water except during behavioral testing.In all experiments, equal numbers of male and female mice were used.All mice were age matched within experiments and littermate controls were used when possible.All mice were maintained on a C57BL/6 genetic background.WT-Atxn1 2Q/2Q (C57Bl6/J) mice, ACTA1-Cre/HSACre79 (B6.Cg-Tg(ACTA1cre)79Jme/J RRID:IMSR_JAX:006149) mice, and Nestin-Cre (B6.Cg-Tg(Nes-Cre)1Kln/J RRID:IMSR_JAX:003771), tdTomato (B6N.129S6-Gt(ROSA)26Sortm1(CAG-tdTomato*,-EGFP*)Ees /J RRID:IMSR_JAX:023537) mice were obtained from The Jackson Laboratory.The Rgs9-Cre mice were a generous gift from Dr. X. William Yang, University of California Los Angeles.
On day 8, selected clones were picked onto 96 well gelatin iMEF plates.When cells were 80-90%confluent, ES cells were frozen in duplicate 96 well format (80% compete media, 10% additional ES FBS, 10% DMSO) and a third onto 48 well gelatin-only plate, in which cells were allowed to expand and harvested for DNA analysis of each clone by PCR with the following primers: i6Atxn1 Junction (F1: 5'-ACACGTGGCTGCAATTTGTC ; R1: 5'-GTAACGCGCTTGCTGCTTG) and 3' of Atxn1 Junction (F1: 5'-AGCGTATCCACATAGCGT; R1: 5'-CTTGCCCATTGCATACCAGG).The puromycin cassette in clone 40 from this first transfection was replaced by human genomic DNA syntenic to the deleted mouse genomic sequences by Flp recombinase using the approach as described (33).This 31kb of human sequence (BAC RP11-413J6) extends from sequences 5' of human ATXN1 exon 8 (5'-TAATGTTACACCAGGCTAAA ) to a sequence 3' of the human ATXN1 polyA site (5'-AGGTGGAATCCCCTGCACCC ), and is flanked by FRT sites, a Lox sequence just 3' of the 5' FRT site, contains a 146Q expanded CAG repeated in exon 8, and at the 3' end has a promoter-FRT cassette that drives expression of the neomycin resistance gene upon recombination into the modified mES cell.After transfection, mES cells were transferred to 10cm, gelatin coated iMEF-Neo plates.Cells were allowed to recover for 48 hrs and were then selected with 125ug/mL of G418 (50mg/mL Gibco) for 7 days.Selected clones (13 total) were picked and expanded.The clones were frozen with 80% complete media, 10% additional ES-FBS, and 10% DMSO.

Analyses of CAG instability
Genomic DNA was extracted from dissected striatum, medulla, cortex, hippocampus and cerebellum using the Qiagen DNeasy Blood & Tissue kit according to the manufacturer's protocol.Tail DNA was prepared using Wizard SV Genomice DNA kit (Promega, A2360).The

Total mouse/human RT-QPCR assay:
Total ATXN1 RNA levels were measured using primers and probes that measures both human and mouse RNA levels in the same reaction.The assay was designed with a forward primer that binds to Exon 6 of the mouse sequence which is common for all genotypes.
(mAtxn1 Forward 5'AAGAAAGACACCACCAGAACC) The reverse primer was designed to bind to an identical sequence shared by mouse exon 7 and human exon 8. (m+hATXN1 Reverse 5'GATTTCTGTAGGGGATCCAGGC) This will generate two highly similar but unique amplicons using a single set of primers.Quantitation of the two amplicons will be achieved by utilizing two probes.Probes were designed to bind to unique sequences within the amplicons conferring specificity for human or mouse sequence: mAtxn1 Ex6-Ex7 (56-FAM/CCACT GCCA/ZEN/GCCTAAAGAACCCA/3IABkFQ) or hAtxn1 Ex8 (5HEX/CCAGAGCTG/ZEN/C TGTTGGCGGATTGTA /3IABkFQ).QPCR was optimized for primer concentration, probe concentration and checked for reaction efficiency.Efficiency for mouse reaction 1.926, human 1.925.

CT Scan
Mice were anesthetized with 2.5% isoflurane and CT scans were performed using a Sophie G8 uPET/uCT imaging system (PerkinElmer).The x-ray source for the CT scans were set to 59 kVp and 100uA and the resulting voxel size was 200um isotropic.3-dimensional CT images were analyzed using Imaris 9.8 (Oxford Instruments).Kyphosis index was determined based on where the distance is calculated from a horizontal line drawn from the center of the C7 vertebrae to the center of the pelvis, and a vertical line from the apex spine curvature to the intersection of the horizontal line (21).Measurement points were manually added in 3-dimensional space using the Imaris spots module.and the time spent (sec) in the goal versus other 3 quadrants of the maze (+1, opposite, -1) was assessed for the probe test.

Open Field
An open field arena was used to assess exploratory locomotor activity and consisted of a white rectangular box (20"W x 20"L x 10" H) illuminated by overhead LED lights (~150 lux).
Mice were allowed to explore the open field arena for 30 min and ANY-maze video tracking software was used to measure the total distance traveled (m) and average speed (m/sec) in individual animals.The apparatus was cleaned with 70% ethanol between animals.

Grip Strength
Grip strength testing was conducted using a digital meter (Bioseb).Mice were gently lifted by the tail and the front paws lowered onto a wire grid so that the front paws grabbed on, and then they were gently pulled back horizontal to the grip bar at a constant speed until the grip released.This was repeated for four trials and grip strength (g) was averaged across the trials.
Animal weights were used to calculate strength/weight ratios.

In Vivo Torque
Mice were anesthetized with isoflurane and maximal isometric torque of the anterior crural muscles was measured.Sterilized platinum needle electrodes were placed through the skin near the left common peroneal nerve and connected to a stimulator (Models S48, Grass Technologies, West Warwick, RI) and stimulus isolation unit (SIU5, Grass Technologies, West Warwick, RI).The contractile function of the anterior crural muscles was assessed by measuring isometric torque (variable voltage (3-10V), 150-ms train, and 0.1-ms pulses) every minute until peak torque was achieved.

Ex vivo muscle preparation
EDL force production was assessed according to methods described previously (39).measures data, sphericity was not assumed; thus, a Geisser-Greenhouse correction was used.
When ANOVA findings were significant, (p<0.05), the analysis was followed by multiple comparisons testing using Tukey's (all pairwise comparisons, >2 groups), Šídák's (all pairwise comparisons, 2 groups) or Dunnett's (pairwise comparisons back to one control group, >2 groups) correction for multiple comparisons.Survival analysis, plotted as Kaplan-Meyer curves, was assessed using log-rank Mantel-Cox and Gehan-Breslow-Wilcoxon tests.

Study approval
The University of Minnesota Institutional Animal Care and Use Committee approved all animal use protocols.
This study was supported by NIH grants R01NS022920 (H.T.O.), R35NS127248 (H.T.O.), R01AR049899 (J.M.E.), R01NS049206 (V.C.W.) and R01NS027699 (H.Y.Z.).The authors thank the Yun You and the Mouse Genetics Laboratory at the University of Minnesota and the MGH Mission Driven Service Core for DNA Fragment Analysis and Christine Chau with creating the graphical abstract using BioRender.

Figure 1 .AFigure 2 . 12 WeightFigure 3 .Figure 4 .Figure 5 .Figure 6 .Figure 7 .
Figure 1.Generation and characterization of the f-ATXN1 146Q/2Q conditional mouse model.(A) Organization of the mouse WT-Atxn1 (blue) and human ATXN1 (black) genes, with the only two exons encoding the ATXN1 protein indicated by boxes larger and darker than the non-coding exons.The size (kb) and location of the mouse genomic sequences (blue) replaced by the human genomic sequences (black) in the f-ATXN1 146Q allele are indicated.(B) The portion of the Atxn1 gene encompassing the two coding exons was replaced with an FRT-recombination recipient cassette in mouse ES cells and then that cassette was replaced with the portion of the human ATXN1 genomic sequences syntenic to the deleted mouse sequence using FLP recombinase.(C) The inserted human sequences in the resulting ATXN1 146Q allele are flanked by LOX recombination sites, as shown.Mating mice with this allele to lines expressing CRE recombinase removes the human ATXN1 insertion, as shown.(D) Barnes Maze performance at 24 weeks of age, unpaired t test.(E) Mouse survival plotted as Kaplan-Meyer curves with median lifespan labeled for each genotype.Log-rank (Mantel Cox) ****(p<0.0001)and Gehan-Breslow-Wilcoxon ****(p<0.0001).(F) Body weight measurements between 6 and 36 weeks of age, RM two-way ANOVA with Geisser-Greenhouse correction and Šídák's post hoc test.(G) Representative photographs of 42-week-old WT-Atxn1 2Q/2Q and f-ATXN1 146Q/2Q showing kyphosis.Significance of results is denoted as * (p<0.05),** (p<0.01),*** (p<0.001), and **** (p<0.0001).