Driving mitochondrial fission improves cognitive, but not motor deficits in a mouse model of Ataxia of Charlevoix-Saguenay

Autosomal-recessive spastic ataxia of Charlevoix-Saguenay (ARSACS) is caused by loss-of-function mutation in the SACS gene, which encodes sacsin, a putative HSP70-HSP90 co-chaperone. Previous studies with Sacs knock-out (KO) mice and patient-derived fibroblasts suggested that SACSIN mutations inhibit the function of the mitochondrial fission enzyme dynamin-related protein 1 (Drp1). This in turn resulted in mitochondrial hyperfusion and dysfunction. We experimentally tested this hypothesis by genetically manipulating the mitochondrial fission/fusion equilibrium, creating double KO (DKO) mice that also lack positive (PP2A/Bβ2) and negative (PKA/AKAP1) regulators of Drp1. Neither promoting mitochondrial fusion (Bβ2 KO) nor fission (Akap1 KO) influenced progression of motor symptoms in Sacs KO mice. However, our studies identified profound learning and memory deficits in aged Sacs KO mice. Moreover, this cognitive impairment was rescued in a gene dose-dependent manner by deletion of the Drp1 inhibitor PKA/Akap1. Our results are inconsistent with mitochondrial dysfunction as a primary pathogenic mechanism in ARSACS. Instead, they imply that promoting mitochondrial fission may be beneficial at later stages of the disease when pathology extends to brain regions subserving learning and memory.


Introduction
First identi ed in the Charlevoix-Saguenay region of Quebec, where it was maintained due to founder effects, ARSACS is now recognized as one of the most common recessive spastic ataxia worldwide (1,2).
ARSACS is caused by loss-of-function variants in the SACS gene and strikes homozygous carriers in early life.The Sacs knockout (KO) mouse recapitulates many of the cardinal features of ARSACS, including gait abnormalities and loss of cerebellar Purkinje cells (3).Sacsin, the protein encoded by the SACS gene, is a large (~520 kD), multi-domain protein with a predicted function as a HSP70/90 co-chaperone (1).
An early report demonstrated mitochondrial localization of sacsin, as well as an interaction with the mitochondrial ssion enzyme Drp1 (4).Mitochondria in neurons with absent or reduced expression of sacsin were abnormally elongated, suggesting that the protein is necessary for proper assembly of the mitochondrial ssion machinery.Studies with patient-derived broblasts con rmed this conclusion (4,5).
Intermediate lament phenotypes have also been reported, including abnormal bundling of neuro laments in the soma and dendrites, hypophosphorylation of neuro lament heavy polypeptide (NFH), as well as vimentin cages in patient broblasts.In broblasts, alterations in the autophagylysosomal system were seen as evidence that sacsin has an important role in proteostasis (3,6).
With this report, we asked whether dysregulation of Drp1-mediated mitochondrial ssion underlies ARSACS pathology, or whether other molecular events, such as intermediate lament aggregation may be primary disease drivers.To this end, we took a genetic approach, crossing Sacs KO mice with germline KOs of two well-established regulators of Drp1, PP2A/Bβ2 and PKA/Akap1.Mitochondrial ssion is regulated by reversible phosphorylation of Drp1 at a highly conserved Ser residue that is phosphorylated by protein kinase A (PKA) and dephosphorylated by two phosphatases, PP2A and PP2B (7).Phosphorylation of Ser637 inhibits Drp1-dependent mitochondrial ssion leading to mitochondrial elongation by unopposed fusion, while dephosphorylation activates the ssion enzyme, shortening mitochondria (8, 9).We previously reported on mice that lack a regulatory subunit of PP2A (Bβ2), which targets the phosphatase to mitochondria because it includes an alternatively spliced mitochondrial localization sequence (10,11).Bβ2 is expressed throughout the central and peripheral nervous system, including the cerebellum, but undetectable in non-neuronal cells (Fig. 1).Intriguingly, a non-coding CAG repeat expansion in a promoter region of the gene encoding Bβ2 (PPP2R2B) causes spinocerebellar ataxia type-12 (SCA12) (12,13).In brains of Bβ2 KO mice, Drp1 is hyperphosphorylated at Ser637, and, consistent with Drp1 inactivation, mitochondria are elongated.Bβ2 KO mice are also protected from cerebral ischemic stroke, likely as a consequence of increased bioenergetic reserves (spare respiratory capacity) (11).We further reported that A-kinase anchoring protein 1 (AKAP1) recruits PKA to the outer mitochondrial membrane to phosphorylate and inactivate Drp1 (14).AKAP1 is tethered to the outer-mitochondrial membrane (15) and ubiquitously expressed (Fig. 1).Akap1 KO mice exhibit smaller mitochondria in neurons and glia, along with exacerbated stroke outcomes (16).AKAP1/Bβ2 double KO (DKO) mice show normal Drp1 regulation, mitochondrial morphology, and stroke sensitivity, indicating that PKA and PP2A exert their effects via a shared effector, Drp1 (11).
We generated Sacs/Bβ2 and Sacs/Akap1 DKO mice and tested for ARSACS disease modi cation at the behavioral level.We reasoned that the Akap1 KO could reverse mitochondrial elongation reported in Sacs KO neurons (4).The Bβ2 KO, on the other hand, might attenuate neurodegeneration (as it does in ischemic stroke), if the mitochondrial elongation observed in Sacs KO mice is an adaptive, rather than a disease-driving mechanism.More precipitous cerebellar decline in either Sacs/Bβ2 or Sacs/Akap1 DKO model would also be informative, as it would support the notion of ARSACS as a mitochondrial disease.
To our surprise, neither DKO in uenced deterioration of motor performance in ARSACS-model mice with age.However, we uncovered a striking decline in cognitive function in older Sacs KO mice.Equally striking, this decline was rescued by promoting mitochondrial ssion by deleting the Drp1 inhibitor PKA/Akap1.

Results
Human whole-tissue mRNA sequencing data retrieved from the Broad Institute (gtexportal.org)indicates wide-spread expression of SACS and AKAP1, whereas Bβ2 (PPP2R2B) expression is largely con ned to the brain.Within brain regions, SACS expression is uniform, while Bβ2 expression is relatively low and AKAP1 expression is relatively high in the cerebellum (Fig. 1).The three gene products are therefore in the right place to functionally interact.
We initially examined DKOs of Sacs and the ssion driver Bβ2 (Fig. 2A), both of which cause mitochondrial elongation when deleted alone (4,11).We set up crosses to yield Sacs wild-type and disease-relevant, homozygous null (-/-) mice combined with Bβ2 alleles of all three genotypes (+/+, +/-, -/-).Heterozygous Bβ2 KOs were included because they afford partial protection from ischemic stroke (11).The same cohorts of mice were analyzed at 3 and 6 months of age using the three-day accelerating Rotarod test, which measures motor coordination and motor learning.As reported before (3), Sacs deletion by itself signi cantly impaired performance at both ages.We included both male and female mice in these and subsequent experiments but detected no sex differences.Data from both sexes were therefore pooled for statistical analysis, with data points for male and female mice differentiated by symbol outline colors in each bar graph for transparency.
At three months of age, all six mouse genotypes learned similarly to stay on the accelerating rod (similar slope of day-to-day performance increase), and heterozygous and homozygous deletion of PP2A/Bβ2 did not improve the rotarod performance in Sacs -/-mice (Fig. 2B, 2C).At six months of age, the same cohort of mice did not show a clear learning pattern (Fig 2D).Heterozygous and homozygous deletion of PP2A/Bβ2 did not improve the Rotarod performance in Sacs -/-mice at this age either (Fig 2E).
Mice with Akap1 +/-genotype were included because Akap1 heterozygosity improves neuroanatomical and metabolic symptoms in a mouse model of Bardet-Biedl syndrome (17).3 months old mice were examined for motor-coordination and -learning using the 3-day accelerating Rotarod test.Motorperformance of mice of all genotypes improved over time, but Sacs -/-mice performed consistently worse than Sacs +/+ mice (Fig. 3B).Notably, while homozygous Akap1 deletion did not improve motor function in Sacs -/-mice, mice carrying one copy of the Akap1 gene performed at a level not signi cantly different from Sacs +/+ mice (Fig. 3C).
Encouraged by the nding that Akap1 heterozygosity might alleviate motor de cits, we examined aged (13-16 months old) Sacs/Akap1 DKO mice, when Sacs KO symptoms are more pronounced.At this age, Sacs KO mice displayed severe impairments on the Rotarod.However, latency to fall was unaffected by the Akap1 genotype (Fig. 4A, B).Time crossing the balance beam, an indicator of motor coordination, was increased in Sacs -/-mice, with no signi cant effect of the Akap1 genotype (Fig. 4C).Distance traveled in the open eld test was reduced in Sacs KO mice; again, without apparent in uence of Akap1 gene dose (Fig. 4D).Likewise, the wire hang test indicated severely impaired grip strength in Sacs -/-mice, but no improvement when one or both Akap1 alleles had been deleted (Fig. 4E).
Next, the same cohort of aged mice were subjected to an associative learning and memory paradigm, contextual fear conditioning.In this test, mice are placed in a context with novel visual, odor, and tactile cues and then subjected to a foot shock.24 h later, mice are re-introduced into the same context and the time anticipating the foot shock ("freezing") is recorded.Compared to Sacs +/+ , Sacs -/-mice displayed a highly signi cant de cit in associating the context with the foot shock they received the day prior.Remarkably, deletion of Akap1 improved the learning and memory performance in a gene-dose-dependent manner in Sac -/-mice, with homozygous Akap1 deletion resulting in near normal contextual condition (Fig. 4F).

Discussion
This study con rms a previous report that Sacs KO mice faithfully replicate the natural history of ARSACS (3).By analyzing DKOs with established regulators of Drp1, we also provide evidence against dysregulation of the mitochondrial ssion enzyme as a primary disease driver in ARSACS.As in other neurodegenerative diseases, mitochondrial dysfunction is increasingly implicated in ARSACS pathology (18-20).For instance, MitoQ, a mitochondria-targeted antioxidant, was recently shown to improve motor coordination and delay Purkinje cell death in Sacs KO mice (21).In light of the present results, mitochondrial dysfunction in ARSACS is unlikely due to an imbalance of mitochondrial ssion and fusion, but rather a secondary consequence of improper folding and aggregation of one or more of the critical clients of the sacsin co-chaperone complex.In a speculative scenario, aggregated neuro laments in Sacs KO Purkinje neurons interfere with tra cking of mitochondria along neurites and recycling of dysfunctional mitochondria by mitophagy.
We also report for the rst time that loss of sacsin is associated with profound impairments in learning and memory in older mice.Cerebellar ataxias, including ARSACS, manifest not only with motor symptoms, but also with a spectrum of neuropsychiatric and learning disorders including dyslexia, attention de cit hyperactivity disorder, autism spectrum disorders, panic disorder, schizophrenia, and intellectual disabilities.Coined as "cognitive dysmetria" or "cerebellar cognitive affective syndrome", this was recognized independently in the late 1990's by Nancy Andreasen (22)(23)(24), Jeremy Schmahmann (25), and others (26).Case studies of ARSACS, speci cally, listed a variety of non-motor symptoms, such as low motivation (apathy), dysphoria, but also paranoid ideation, irritability, and marked cognitive dysfunction, including anosognosia (27,28).
Whereas fMRI studies indicate that the cerebellum participates in the retrieval of episodic memory and other cognitive tasks (24), there remain questions how cerebellar disorders impair cognition.On the one hand, cerebellar nuclei project, directly or indirectly, to various brain areas involved in higher-order cognition, including the prefrontal cortex.Also, cerebellar lesions due to accidents or surgical resections can present with non-motor symptoms similar to cerebellar disorders.On the other hand, most cerebellar ataxia disease genes, including SACS, are ubiquitously expressed (Fig. 1), and cerebral atrophy commonly follows cerebellar atrophy in hereditary cerebellar ataxias.
Further studies using conditional Sacs KO mice are needed to pinpoint the cellular and anatomical origins of non-motor symptoms of ARSACS.Also, future studies should address the temporal relationship between cognitive and motor symptoms in ARSACS and the mechanism by which Akap1 deletion improves cognitive decline in Sacs KO mice.

Mice
Mouse work was performed in accordance with the guidelines of the animal ethics committee of the University of Iowa.Mice were group-housed in a colony maintained with a standard 12 h light/dark cycle and given food and water ad libitum.Experiments were performed on age-matched mice of both sexes as indicated in bar graphs.Experiments were conducted according to the Guide for the Care and Use of Laboratory Animals, as adopted by the National Institutes of Health, and with approval of the University of Iowa AAALC-accredited Institutional Animal Care and Use Committee.
The Sacs -/-mice were a kind gift of Bernard Brais, McGill (3).The AKAP1 -/-mouse line was kindly provided by Dr. Stanley McKnight at University of Washington (29), and Bβ2 -/-mice were generated at the U. Iowa Mouse KO Core Facility (11).We generated mice with heterozygous or homozygous deletion of PP2A/Bβ2 or Akap1 that were either wild-type or null at the Sacs locus.Mice were in the C57BL/6J background and were backcrossed to C57BL/6J mice imported from the Jackson Laboratory (Bar Harbor, ME) every 6-10 generations to prevent genetic drift.Mice of all genotypes were born in expected Mendelian ratios and were fertile, except for Akap1 -/-females who are infertile (29).All mice achieved a normal lifespan, with many individuals surviving beyond two years.However, Sacs -/-mice with or without deletion of Bβ2 or Akap1 displayed progressive gait abnormality as documented (3).

Behavioral Testing
General.5-7 days of habituation and handling was done prior to behavioral assessment.Mice were allowed to acclimatize to the testing environment for 30 minutes prior to the starting of the experiment trials on the day of testing.All genotypes were tested on the same day in a randomized order with experimenters blinded to genotype and sex.
Accelerating rotarod.Mice were placed on a rotating rod (IITC Life Woodland Hills, CA) with gradual speed increase from 4 to 40 rpm over 5 minutes, in which the latency to fall was recorded.Testing consisted of four trials on each day for three consecutive days, with 5-10 minutes between each trial.The average of four daily trials was recorded for each mouse.
Balance beam.The balance beam apparatus was acquired from MazeEngineers Inc (Skokie, IL).The balance beam test was performed as previously described (30).A 12 mm-wide beam was used in our tests.Mice were placed at one end of the beam and the latency for crossing the beam to the other end was recorded by MediaRecorder software (Noldus).Mice were trained for three times in each day for two consecutive days and tested for three times on day 3.The best performance (minimum latency to cross beam) was recorded on the testing day.
Wire hang test.The test began with the mice placed on an elevated wire cage top, which was then inverted and suspended above the home cage.The time it took for the animal to fall was recorded.This test was conducted two times on one day, with 5-10 minutes between each trial.The best performance (longer hang time) was recorded.(C) Sacs -/-mice performed consistently worse than Sacs +/+ mice, with no effect by homozygous deletion of Akap1.However, at day 2 and 3, heterozygous deletion of Akap1 in Sacs -/-mice performed at a level not signi cantly different from Sacs +/+ mice.See Fig. 2 for data presentation and analysis.
Akap1 KO rescues cognitive, but not motor function in aged Sacs-/mice.(A, B) According to the 3-day accelerating Rotarod test, aged (13-16 months old) Sacs -/-mice performed worse than Sacs +/+ mice regardless of Akap1 genotype.Motor learning was not evident from linear regression analysis.(C) Sacs -/- mice took signi cantly more time to cross the balance beam and homozygous deletion of Akap1 did not improve their performance.However, Akap1 +/-mice showed partial improvement.(D) Signi cant muscle weakness in Sacs -/-mice compared to Sacs +/+ mice was detected by the wire hang test; however, heteroor homozygous loss of Akap1 had no effect.(E) In the 10-min open eld test, deletion of Akap1 did not improve the distance-traveled de cit of Sacs -/-mice.(F) Sacs -/-displayed weaker recall of contextual fear memory.Deletion of Akap1 improved the learning and memory performance in a gene dose-dependent manner.See Fig. 2 for data presentation and analysis.Akap1 genotype data in (D, E) was pooled and analyzed by Student's T-test because there was no signi cant Sacs/Akap1 genotype interaction.

Figures
Figures