Progressive gait and motor deficits in a rat model of Alexander disease

Abstract

Alexander disease is a leukodystrophy caused by gain-of-function mutations in the gene for Glial Fibrillary Acidic Protein (GFAP) which result in accumulation and aggregation of GFAP protein, astrocyte dysfunction, and ultimately developmental delay, failure to thrive, and intellectual and motor impairment. A Gfap^+/R237H rat model, designed to mimic the common R239H human variant, meets normal milestones during early postnatal development, but declines dramatically as the rats mature. At severe stages of disease, Gfap^+/R237H rats exhibit cognitive and motor deficits and increased mortality. Here we provide a more detailed analysis of the Gfap^+/R237H rat with respect to onset of motor impairments and increasing loss of function. We show that Gfap^+/R237H rats develop abnormal open field activity as they mature but stabilize with age, and that motor deficits are apparent as early as 4 weeks of age, as demonstrated by poor rotarod performance. We use automated gait analysis to further characterize subtle differences at this early age and demonstrate the progression and persistence of impairment at late stages of disease. In addition, we find evidence for changes in cerebellar size, suggesting a potential neuroanatomical correlate to the observed deficits. The rat model provides a novel system in which to investigate aspects of impaired motor function and central nervous system pathology that are directly relevant to the human disease.

1. Introduction

Alexander Disease (AxD) is a rare neurological disorder and one of the few primary disorders of astrocytes [1–4], as it is caused by mutations in GFAP, which encodes the major intermediate filament protein of astrocytes [5]. Onset can occur at any age, and in most cases the disease is progressive and fatal [6]. AxD has been classified by age of onset (neonatal, infantile, juvenile, and adult) [7, 8], or as cerebral and bulbospinal forms along with a third intermediate category [9]. Statistical analysis of symptoms has yielded two main types. Type I has early onset of symptoms, including developmental delay, seizures, megalencephaly, failure to thrive, and gradual loss of intellectual function, with frontal lobe predominance and leukodystrophy. Type II occurs at any age, demonstrates lesions in structures of the posterior fossa with correlating clinical features including bulbar, pseudobulbar, cerebellar, or pyramidal tract signs; autonomic dysfunction; and abnormal gait [6]. Type I may also be subdivided into four sub-categories that are distinguishable, in part, by the trajectories with which individuals acquire or fail to meet motor milestones [10].

The hallmark pathological feature of AxD is the presence of Rosenthal fibers, which are cytoplasmic aggregates of GFAP and other intermediate filaments, small heat shock proteins including αB-crystallin and HSP27, ubiquitin, and other proteins [11, 12]. It is currently unknown whether Rosenthal fibers are toxic and how they contribute to disease, but evidence suggests that, similar to other proteinopathies, misfolded proteins interfere with protein clearance pathways leading to oxidative stress, neuroinflammation, and astrocyte dysfunction [2, 4]. Rosenthal fibers may more directly interfere with cellular functions such as mitosis and cytokinesis, and astrocytes in AxD often have lobulated or multiple nuclei [11]. There is currently no cure or effective treatments for Alexander disease. However, because AxD is a monogenic disorder resulting from gain-of-function mutations, it may be possible to develop therapies, such as antisense oligonucleotide or gene editing technologies, to correct the abnormal gene function [13]. In addition, AxD represents a unique model for astrocytic pathology that can contribute to our understanding of chronic gliosis and astrocyte dysfunction in other neurological diseases.

To improve our understanding of AxD and GFAP pathology, we developed a Gfap^+/R237H rat model using CRISPR-Cas9 mutagenesis to reproduce the severe R239H mutation observed in patients with early-onset AxD at the orthologous GFAP-Arg237 position in the rat (hereafter referred to as R237H). The R237H rat model displays widespread astrocyte pathology, white matter deficits, and motor impairment. In this report we characterize motor deficits, including rotarod performance, open field activity, and gait kinematics at different stages of disease. These studies provide important insights into the effects of astrocyte pathology on motor function and offer a clinically relevant animal model with useful functional outcome measures for testing experimental therapeutic interventions.

2. Methods and procedures

2.1. Generation of the rat AxD model

R237H rats were generated as previously described [14], and maintained as heterozygotes on an outbred Sprague-Dawley genetic background (Charles River CD IGS rat). Rats were bred at the University of Wisconsin-Madison (UW-Madison), and pregnant dams shipped to the University of California Davis (UC Davis) for behavioral phenotyping of offspring at the UC Davis M.I.N.D. Institute Intellectual and Developmental Disabilities Research Center (IDDRC). Rats were housed in an environmentally controlled vivarium adjacent to the behavioral testing rooms under a 12:12 hour light-dark cycle with food and water available ad libitum. All animal studies were approved by the Institutional Animal Care and Use Committees at UC Davis (Protocol #21523, January 8, 2020) or under the College of Letters and Sciences and Vice Chancellor Office for Research at UW-Madison (protocol G005354, approved December 21, 2015) and comply with the ARRIVE guidelines.

2.2. Distributions of animals to individual experiments

Table 1 shows the numbers of males and females, R237H and wild type (WT) rats, and age at testing for each experimental study. Table 1 also lists the cohort identification number (Cohort ID) that rats were grouped under for behavioral testing. Same sex rats were housed 2 per cage, with one R237H and one WT when possible. Rats were evaluated in the following behavioral tests: open field, rotarod, elevated plus maze, and DigiGait gait analyses. All testing occurred during the light phase of the circadian cycle. Not all animals were used in every test, and the age of testing was dependent on animal availability and specific questions about task performance at developmental ages that became of interest after these studies were initiated. In addition, limitations of facilities, staffing, and animal availability during the initial months of the COVID pandemic resulted in disruptions in our ability to test all animals in all behavioral tests, and in some cases dictated ages of rats that were available for testing. Separate groups of WT and R237H rats were used for cerebellar morphology and histopathology as described in Methods for each of the analyses.

Table 1.

Experimental layout showing numbers of male and female rats, litters and testing ages for behavioral experiments

Cohort ID	# of Litters	# of Subjects		Age at testing	Test
		WT	R237H
1	6	12 male, 12 female	13 male, 13 female	16 weeks	Open Field
2	9	10 male, 10 female	10 male, 10 female	3 & 8 weeks	Open Field
		9 male, 11 female	9 male, 9 female	5 weeks	EPM
		15 male, 5 female	7 male, 13 female	4 weeks	Rotarod
3–4	7	8 male, 8 female	8 male, 8 female	4 weeks	DigiGait
5–6	9	8 male, 9 female	7 male, 6 female	16 weeks	DigiGait

2.3. Statistical analysis

Data were analyzed using SPSS statistical package (IBM SPSS Statistics, Version 29.0.2.0). Behavioral data were analyzed using a mixed-effects analysis of variance (ANOVA) with litter as a random variable to control for possible contribution of litter to behavioral outcomes [15], or by repeated measures analysis of variance (RMANOVA). Genotype and sex were included in all analyses. Levene’s test for homogeneity of variance was carried out and variables violating assumptions of homogeneity were analyzed via the Welch’s t-test. All tests were 2-tailed with the minimum probability to establish statistical significance set at p<0.05. All data in graphs represent means ± standard error of the mean $(\stackrel{-}{X}\pm \text{SE})$ unless indicated otherwise.

Body weight did not differ at weaning, but by 6 weeks of age male and female R237H rats gained significantly less weight than WT rats (P<0.001), and this difference persisted throughout the study (i.e., 26 weeks of age, see Fig. 1). Failure to thrive and delayed somatic growth are common features of AxD and are recapitulated in the rat model. However, body weight was not used as a covariate in the statistical analyses of behavioral results because body weight effects are a direct consequence of the R237H gene mutation, and not a random, “uncontrolled”, pre-existing variability that can be controlled for by analysis of covariance (ANCOVA). Specifically, ANCOVA may be considered inappropriate when the distribution of covariate values are clearly related through causation or correlation to the dependent variable [16]. Nevertheless, statistical analyses using body weight as a covariate were carried out and are presented as Supplementary Material. These results did not reveal any new significant differences between groups and while some comparisons were no longer significant, the overall conclusions were not changed for the behavioral experiments. While there are no adequate statistical means to separate the effects of body weight and genotype on behavior in these experiments, the possible consequences of weight differences are considered within the results and discussion for each behavioral experiment.

Fig. 1.

Body weights from postnatal week 2 through week 26 for male (A) and female (B) wild type (WT) and R237H rats. Data are from Cohort 1 drawn from 6 litters. Body weight did not differ at weeks 2 and 3, but R237H rats weighed significantly less than WT rats thereafter (p<0.001). This difference between genotypes was significant for both males and females.

Experimenters were blinded to genotype throughout behavioral testing, although the smaller body size of R237H versus WT rats was apparent from approximately 5–6 weeks of age (see Fig. 1, also Fig. 1 in reference [14]).

2.4. Open field test

Rats in Cohort 1 were tested in the open field at 16 weeks of age. Infrared beam-breaks were used to record total distance traveled (cm), horizontal activity, vertical activity, and total time in the center of the open field (sec). Rats in Cohort 2 were tested using the same procedure at 3 and 8 weeks of age and were previously reported [14]. Briefly, rats were placed into the open field monitoring system for 30 minutes and activity recorded in 5-minute bins (Integra, Omnitech Electronics, Inc., Columbus, OH). The apparatus was made of clear Plexiglas (40.4 cm square, 10 cm margin width). Room illumination was set to 30 lux. After testing, rats were returned to their home cages. Two data points in the 8-week group (1 WT female and 1 R237H female) were excluded from statistical analysis due to data loss during acquisition.

2.5. Accelerating rotarod

Motor coordination, balance, and learning were tested in the accelerating rotarod [17]. Rats in Cohort 2 were tested at 4 weeks of age [postnatal day (PND) 29–32]. Rats were first trained on the rod at a constant speed of 5 rpm for 120 seconds. If the rat fell off, it was placed back onto the rod for the remaining time. Each rat was given three training trials with approximately 45 minutes between each trial. Following the training day, the rats were tested for three consecutive days, three trials per day. During these tests the rod accelerated from 5 to 40 rpm over a 5-minute period. If the rat fell off, that time was recorded, and the trial was complete.

2.6. Elevated plus maze

The elevated plus maze is frequently used in rats and mice to assess anti-anxiety effects of pharmacological agents, as well as to uncover brain regions and circuitry related to anxiety [18, 19]. The maze is composed of two open arms without side walls (10 × 50 cm), two closed arms (10 × 50 cm) with 30.5 cm high side walls, and a center compartment (10 × 10 cm). The arena is elevated 100 cm above the ground. Rats were tested at 5 weeks of age (PND 35–38). The lighting was set to 5–10 lux in the closed arms and 10–20 lux in the open arms. Rats in Cohort 2 were individually placed in the center of the arena and allowed to explore for 5 minutes and then returned to their home cage. Rats were tracked and recorded using Ethovision software. Time spent on the open and closed arms and center compartment were measured. Less time on the open arms provides evidence for increased anxiety [18].

2.7. Gait analysis

The DigiGait^™ Imaging System (Version 14.5, Mouse Specifics, Inc., Framingham, MA) was used to analyze gait in WT and R237H rats. Briefly, rats were placed in the DigiGait^™ apparatus and allowed to acclimate for 3–5 minutes. Following acclimation, the treadmill belt was turned on at a speed of 5 cm/s. The speed of the belt was slowly increased to 15 cm/s over the course of 2–3 minutes. Once a speed of 15 cm/s was reached, video of the rat’s gait was recorded from a camera located below the treadmill belt. Videos in which all 4 paws were in view the entire time and no paws slipped below the back bumper were collected to analyze 6–8 strides (3–6 seconds). If rats could not complete the task on the first day, they were tested up to two more times. The treadmill belt was cleaned with 70% EtOH between animals and daily upon completion of testing. DigiGait^™ Analysis Software generates multiple indices of gait dynamics, symmetry, and variability that can be employed to compare gait between genotypes. Fourteen of these indices were used for statistical analysis in assessing both male and female R237H and WT rats. These indices are described in Table 2 and depicted graphically in Figure 5. Temporal gait parameters analyzed were stride frequency (steps/s), stride duration (s), stance duration (s), swing duration (s), brake duration (s), and propel duration (s). Spatial measures were stride length (cm), stance width (cm), step angle (deg), absolute paw angle (deg), paw area (cm²), and paw placement positioning (cm). Ataxia coefficient and hindlimb paw drag were also measured. Ataxia coefficient is a ratiometric variable reflecting stride length variability, and hindlimb paw drag is a measure of the robustness of paw lift off of the belt during propulsion. Redundant variables, including those that express data in percentages and those focused on variability of gait performance (i.e., coefficient of variation, CV), were not included in the main analyses. Sex was included as a variable in the statistical analyses. Gait analysis was performed at 4 weeks (Cohorts 3–4) and 16 weeks (Cohorts 5–6) of age. Rats were not tested at 8 weeks of age because the severity of the phenotype in R237H rats prevented them from reliably walking on the DigiGait treadmill.

Table 2.

Description of DigiGait indices used to evaluate gait patterns in R237H and WT rats (Modified from E.R. Berryman, 2009, [67])

	Parameter	Description
1.	Stride frequency (steps/sec)	The average number of steps taken per second.
2.	Stride duration (sec)	The amount of time to complete one full stride for one limb.
3.	Stance duration (sec)	Duration in which the paw remains in contact with the belt.
4.	Swing duration (sec)	Duration of the swing phase when there is no paw contact with the belt.
5.	Braking duration (sec)	The time between initial paw contact with the belt to the maximal paw contact at the end of the swing phase.
6.	Propel duration (sec)	Time required for accelerating forward motion from maximal paw contact with the belt to immediately before the swing phase. A longer duration might indicate less strength and less control.
7.	Stride length (cm)	The distance between initial contacts of the same paw in one complete stride.
8.	Stance width (cm)	The distance between the two front paws or the two hind paws as measured from the middle paw area.
9.	Step angle (deg)	The angle made between left and right hind paws as a function of stride length and stance width. Increased step angle argued to reflect ataxia after prenatal alcohol exposure ([68])
10.	Absolute paw angle (deg)	Absolute value of the angle that the paw makes relative to the long axis of the direction of movement. Otherwise known as “splay angle”.
11.	Paw area (cm2)	The maximal paw area in contact with the treadmill during the stance phase of the step cycle.
12.	Paw placement positioning (cm)	This is a measure related to balance. It measures the extent of overlap between ipsilateral fore and hind paws during stance phase.
13.	Paw drag	Propulsion metric calculated to reflect the robustness of paw lift off from the belt. This may be higher when the hind limb is impaired from lifting the paw off the belt during early propulsion.
14.	Ataxia Coefficient	An index of step-to-step variability calculated as [(MAX Stride Length – MIN stride Length)/MEAN Stride Length] for each limb.

2.8. Cerebellum measures

We previously reported water content in different regions of the CNS as a measure of edema, with most areas showing increased water, with the exception of cerebellum [14]. To better demonstrate the inverse changes in cerebellum compared to other areas in the CNS, here we report water content as a percent change of R237H rats compared to WT, as well as wet weights for the different brain regions in these same animals (N=6, males). In addition, area measures of cerebellum were taken from brains fixed in 4% paraformaldehyde (n = 7 R237H males, 7 R237H females, 6 WT male, 6 WT females). Forty-micron midsagittal sections were mounted on slides in aqueous media, imaged with a dissection microscope, and traced in ImageJ to determine area. All animals were 8 weeks of age.

To determine whether differences in cerebellar size were apparent at 16 weeks of age, a separate set of tissues were collected, weighed, and imaged at the midline for area measures from fresh unfixed cerebella. Brains were removed after euthanasia, and whole cerebella were dissected on ice and weighed. They were then split at the sagittal midline and placed in 3 cm petri dishes for immediate imaging of the cut face. Area measures were determined from each cerebellar half using ImageJ and averaged for each rat. Weight and area measures were taken from 5 R237H and 6 WT males, and weight measures for 6 R237H, 6 WT females.

2.9. Histology and immunofluorescence

To assess general histopathology, female R237H and WT rats at 16 weeks of age (n = 4 per genotype) were anesthetized with isoflurane and transcardially perfused with phosphate buffered saline (PBS). Brains were removed and immersion fixed in methacarn before paraffin embedding, sectioning (4 μm), and staining with hematoxylin and eosin (H&E) by the University of Wisconsin Carbone Cancer Center Experimental Animal Pathology Laboratory. Brightfield images were acquired with a SPOT camera (Diagnostic Instruments) on a Nikon Microphot.

For immunolabeling, male and female rats at 8 weeks of age (n = 4 per genotype) were perfused with PBS followed by 4% paraformaldehyde before removing brains for overnight immersion fixation. After cryoprotecting in graded sucrose solutions (to 30%), 40 μm sections were collected from the sagittal midline on a sliding microtome and stored in cryoprotectant (25% polyethylene glycol, 25% glycerol in 0.1 M phosphate buffer pH 7.4). Floating sections were rinsed in PBS before blocking and permeabilizing with 5% normal serum and 0.3% Triton-X-100 (TX100) in PBS. Sections were incubated with primary antibodies diluted in 1% BSA/0.3% TX100 in PBS, including rabbit anti-Calbindin D-28K (1:1000, Chemicon AB1778) and rabbit anti-GFAP (1:500, DAKO Z0334), for 72 hours at 4°C. Sections were washed in PBS with 0.05% TX100 before adding secondary antibodies (Alexafluor-488 or −568 conjugated goat anti-rabbit, Thermo Fisher Scientific, Invitrogen A11034 and A11036) diluted 1:500 in the same diluent and incubated overnight at 4°C. Sections were rinsed, incubated with 1 μg/ml DAPI, washed again in PBS/TX100, and mounted with ProLong Gold mounting media (Thermo Fisher Scientific, Invitrogen) before imaging with a Nikon A1R-HD or Leica Stellaris 8 confocal microscope system within the Waisman Center Cellular and Molecular Neuroscience Core. Images are presented as maximum intensity projections.

3. Results

3.1. Body weight gain

R237H rats fail to thrive after weaning and by 8 weeks of age are gaunt, frail, and demonstrate significant motor impairments [14]. At this age, a small percentage of animals die of unknown causes, but those that survive past 12 weeks of age begin to gain weight, become fertile, and survive well into adulthood (past 1 year, [14]). In the present study, animals showing weight loss were given diet gel and subcutaneous fluids, and although actual life span was not examined, less than 5% of R237H rats died. Representative body weights from postnatal week 2 through 26 weeks of age are shown in Fig. 1 (summarized from Cohort 1: 12 WT male, 12 WT female, 12 R237H male, and 13 R237H female rats). A RMANOVA demonstrated significant main effects of age (F_1,26=2069.7, p<0.0001), genotype (F_1,26=373.6, p<0.0001), sex (F_1,26=223.4, p<0.0001), but not for litter (F_5,26=0.83, p=0.540). Significant interactions were found for age × genotype (F_1,26=251.3, p<0.001), age × sex (F_1,26=273.4, p<0.001) and age × genotype × sex (F_1,26=83.0, P<0.001). There were no significant interactions between litter and age, sex, or genotype. Individual comparisons at each week showed that at 2 and 3 (weaning) weeks of age there were no weight differences between genotypes, between males and females, and no sex × genotype interaction. By 6 weeks of age large differences between WT and R237H rats and between males and females were found that remained significant throughout the 26-week measurement period (p<0.001, each comparison at each week from week 6 through 26).

3.2. Open field activity

Behavioral deficits in the rat model of AxD are minimal at 3 weeks of age, but by 8 weeks, R237H rats are severely motor-impaired, as we have previously demonstrated by open field activity and poor performance in various motor tests [14]. Here, we analyzed open field activity at 16-weeks of age (Fig. 2) for comparisons with a more detailed gait analysis performed in the current study (section 3.5 below), and include the effects of sex differences for all 3 age groups. For direct comparisons of open field activity among age groups, Fig. 2 includes the 16-week-old group as well as previously reported data for R237H and WT rats at 3 and 8 weeks of age [14]. Further analysis with sex as a variable for each age group is included in supplemental Fig. S1.

Fig. 2.

Open field activity for WT and R237H rats. Horizontal and vertical activity measured by beam breaks over 30-min at 3, 8, and 16 weeks. Total distance traveled (cm) and time spent (s) in the center of the arena are also shown. (Males and females tested, n=10 per group at 3 and 8 weeks; n=12 for WT and 13 for R237H rats at 16 weeks, ****p<0.0001).

At 3 weeks of age there were no differences in open field activity between groups due to genotype or sex, and no significant interactions. At 8 weeks R237H rats showed reduced horizontal (F_1,34=29.95, p<0.0001) and vertical activity (F_1,34=42.84, p<0.0001), traveled a shorter total distance (F_1,34=27.55, p<0.0001), and center time was shorter (F_1,34=24.49, p<0.0001), compared to WT. There was also a significant sex × genotype interaction for both vertical activity (F_1,34=5.00, p<0.05) and center time (F_1,34=4.42, p<0.05). Subsequent analysis of vertical activity within the sexes at 8 weeks showed that R237H males (128.8±29.5) differed significantly (p<0.0001) from WT males (1328.9±96.9), and R237H females (445.1±208.5) differed significantly (p<0.01) from WT female rats (1034.1±165.2; supplemental Fig. S1). Analysis of center time showed that R237H males (230.3±25.9 s) differed significantly (p<0.0001) from WT males (697.1 ±97.6 s), while R237H females (269.7±51.7 s) and WT females (458.3±64.8 s) did not differ significantly (p=0.058, supplemental Fig. S1). Male WT rats showed longer center time (p<0.05) than female WT rats, while female R237H rats did not differ significantly from male R237H rats (Fig. S1). There were no significant sex differences and no other significant interactions for horizontal activity and total distance.

In comparison, at 16 weeks, R237H rats showed lower vertical activity than WT rats (F_1,46=18.37, p<0.0001) and shorter center time (F_1,46=21.04, p<0.0001). Horizontal activity and total distance did not differ. There was a sex difference in vertical activity (F_1,46=16.12, p<0.001) and center time (F_1,46=13.9, p<0.001) regardless of genotype, with females showing less vertical activity (1536±94.2) than males (2139±162.8) and shorter center times (461.5±20.9 s) than males (630.7±52.9 s). There were also sex × genotype interactions for vertical activity (F_1,46=8.92, p<0.01) and center time (F_1,46=4.86, p<0.05). Male WT rats differed from male R237H rats in vertical activity (p<0.0001) and center time (p<0.0001) (Fig. S1). Male WT rats showed higher vertical activity (p<0.0001) and longer center time (p<0.001) than female WT rats, while female R237H rats did not differ from male R237H rats (Fig. S1).

3.3. Rotarod

Motor function and coordination were tested on the accelerating rotarod over 3 consecutive days as shown in Fig. 3. We previously reported that at severe stages of disease (i.e., 8 weeks of age) R237H rats cannot perform the rotarod test [14]. Therefore, we tested rats during the fourth postnatal week (PNDs 29–32), an earlier stage in disease progression. Analysis by RMANOVA (genotype × sex × trial) showed a difference between WT and R237H (F_1,36=18.12, p=0.0001), but no significant sex or sex × genotype interaction. There was a significant trial effect (F_8,288=4.54 p<0.0001), and a trial × genotype interaction (F_8,288=2.19, p<0.05). Individual comparisons for each trial showed that WT and R237H rats did not differ on the first two trials but differed significantly on most trials thereafter, with R237H rats falling from the rod earlier (Fig. 3). Individual data points are shown in supplemental Fig. S2.

Fig. 3.

Rotarod performance for WT and R237H rats. Rats were tested 3 times on each of 3 consecutive days on an accelerating rotarod to assess motor function and coordination. (n=15 male and 5 female WT; 7 male and 13 female R237H rats. Individual data points provided in supplemental Fig. S2, *p<0.05, **p<0.01, ***p<0.001).

3.4. Elevated plus maze

As shown in Fig. 4, there were no differences between groups in time spent in the open arm (p=0.46), closed arm (p=0.27), or center of the elevated plus maze (p=0.26). In addition, calculation of the ratio of open to closed arms failed to show differences between WT and R237H rats (p=0.37; data not shown). There were also no significant sex differences or sex × genotype differences between groups. However, velocity (cm/s) and subsequently total distance traveled on the maze were less in R237H rats compared to WT rats (p<0.001). Results separated by sex are shown in supplemental Fig. S3.

Fig. 4.

Elevated plus maze performance for WT and R237H rats. (A) Time (s) in the open and closed arms and in the center of the maze are shown. (B) Mean velocity (cm/s) is shown as a measure of overall activity. (n = 9 male and 11 female WT; n = 9 male and 9 female R237H; ***p<0.001).

3.5. Gait analysis

Gait abnormalities are a common feature in AxD, and on gross observation the R237H rats show severe impairments in which the hind paws frequently splay out laterally and posteriorly [14]. To quantify these observational differences, we used the DigiGait system to measure temporal and spatial indices of gait performance. The gait parameters analyzed are listed in Table 2 and depicted in the gait cartoon in Fig. 5.

Fig. 5.

Graphical depiction of locomotor variables quantified by DigiGait software. (A) Camera depiction from the side and beneath wild type (top) and R237H (bottom) rats. Paws are color coded for clarity. Bottom image depicts hindlimb paw drag found in some R237H rats. (B) Graph showing paw position during one stride for front right paw (top) and hind right paw (bottom). Stride (s) is time to complete one complete stride. Stance time (s) is time paw is in contact with treadmill and is comprised of deceleration after paw contacts treadmill (Brake) and acceleration of the paw leading to lift off from the treadmill (Propulsion). Swing is the duration (s) that the paw is raised off the treadmill before initiating the next stride. (C) Cartoon showing paw measurements used for analyses (i.e., paw angle, paw area, step angle, stride length, stride width).

Thirty-nine rats were available for testing in DigiGait at 4 weeks of age, and 44 at 16 weeks of age. Eight WT and 13 R237H rats did not generate useable treadmill data for DigiGait (3 WT and 4 R237H at 4 weeks; 5 WT and 9 R237H at 16 weeks), even after retesting, and their data were omitted from statistical analysis. The difference in the number of R237H and WT rats omitted from analysis was not statistically significant (X²_(1,82)=1.60, p=0.21).

Temporal indices of gait for forelimb and hindlimb at 4 weeks and 16 weeks of age are shown in Fig. 6. At 4 weeks of age, when the phenotype is only weakly evident, swing duration, which is the time (s) that the paw is off the treadmill belt during midstride, was longer for both forelimbs and hindlimbs for R237H vs WT rats (p<0.001). No other differences were found at 4 weeks of age for temporal gait variables. In contrast, by 16 weeks, stride frequency (steps/sec) was lower for R237H versus WT rats for both forelimbs and hindlimbs (p<0.01). R237H rats also demonstrated increased stride duration in both forelimbs and hindlimbs (p<0.01) compared to WT rats. Swing duration also differed between genotypes for forelimb (p<0.01) and hindlimb (p<0.0001). Stance duration, which is the total time of paw contact with the belt, did not differ between groups (data not shown). However, stance duration is actually derived from two distinct measures, brake duration and propel duration, which can be analyzed separately. Brake duration is time from initial paw contact to maximum paw contact with the belt during deceleration and was decreased in the forelimbs for R237H rats (p<0.01). Propel duration, the time of belt contact from the end of the brake phase until paw lifts off from the belt, was increased in the forelimb for R237H rats (p<0.01). The decreased brake duration and the increased propel duration may have served to offset one another, resulting in no significant difference in stance duration between groups.

Fig. 6.

Temporal gait measures for WT and R237H rats at 4 and 16 weeks of age. (A) stride frequency (steps/s), (B) stride duration (s), (C) swing duration (s), (D) brake duration (s), (E) propel duration (s), and (F) paw drag, for the two age groups. (Males and females tested; n=8 at 4 weeks; n=8 male and 9 female WT and 7 male and 6 female R237H rats at 16 weeks; **p<0.05, **p<0.01, ***p<0.001, ****p<0.0001).

Hindlimb paw drag differed for WT and R237H rats at 16 weeks (p<0.001) but not at 4 weeks of age. Paw drag is a measure derived from computing slopes (cm²/sec) from the gait profile just preceding the beginning of swing and may reflect how quickly rats transition from the stance phase to the swing phase. It may also reflect the robustness of the lift off stage of the paw from the belt. If a hindlimb is impaired from lifting the paw off the belt during early propulsion, then these values may be less steep. The significantly shallower negative slope of the R237H rats compared to WT rats at 16 weeks of age indicates slower transition (i.e., increased paw drag) to the swing phase. This difference was not evident in 4-week-old rats. These findings are consistent with the longer hindlimb propulsion duration in R237H rats at 16 weeks, suggesting more time was required to produce the force needed to propel forward potentially due to hindlimb muscular weakness. In quadrupeds the hindlimbs produce most of the propulsive forces while the forelimbs produce most of the braking force [20].

Fig. 7 summarizes the analysis of spatial gait parameters. At 4 weeks of age there were no differences between genotypes for the forelimb measures, and only step angle differed for the hindlimb, with R237H rats showing a larger hindlimb step angle (i.e., hindlimb splay) compared to WT rats (p<0.01). Analysis of gait at 16 weeks revealed several differences between R237H and WT rats, particularly for hindlimb gait. Stride length, the length (cm) a paw travels through a given stride, was increased in R237H rats at 16 weeks for both forelimb (p<0.01) and hindlimb (p<0.01). Increased stride length coupled with increased swing time (see Fig. 6) in mutant rats would indicate abnormal gait that is not likely the result of lower body weight. Step angle and stride width (data not shown) did not differ between genotypes at 16 weeks. Hindlimb absolute paw angle, the angle the paw makes with the long axis of the direction of motion of the animal, was increased for R237H rats versus WT at 16 weeks (p<0.0001). Paw area and ataxia coefficient did not differ between genotypes at 16 weeks of age (data not shown). Paw placement positioning, which is sensitive to disturbance in balance, did not differ between WT and R237H rats. Considered together, the longer propulsion times and increased paw drag reflects more time required by R237H rats to propel into the next stride compared to WT.

Fig. 7.

Spatial gait measures for WT and R37H rats at 4 and 16 weeks of age. (A) stride length (cm), (B) step angle (deg), (C) absolute paw angle (deg), and (D) paw placement positioning (cm), for the two age groups. (Males and females tested; n=8 at 4 weeks; n=8 male and 9 female WT and 7 male and 6 female R237H rats at 16 weeks; **p<0.01, ****p<0.0001).

3.6. Sex differences in gait

No significant sex differences were found at 4 weeks of age for forelimb or hindlimb gait. There was a significant interaction between sex and genotype for forelimb (F_1,27=4.77, p<0.05) and hindlimb stride frequency (F_1,28=4.58, p<0.05), and hindlimb propel duration (F_1,28=4.66, p<0.05) (supplemental Fig. S4). Tests of simple main effects for these interactions using Holm-Šídák’s multiple comparisons did not show significant differences. The lack of sex differences in gait at 4 weeks is likely related to the fact that size differences, as measured by body weight, between sexes at this developmental age (i.e., 3–4 weeks postnatal) were small (i.e., <5.0%) and not statistically significant at 3 weeks (F_1,42=2.66, p=0.11). At 16 weeks, sex differences were found for the forelimb including stride duration (F_1,26=5.71, p<0.05; male =0.74±0.01s; female =0.66±0.02s), brake duration (F_1,26 = 5.97, p<0.05; male =0.22±0.04s; female =0.17±0.03s), stride length (F_1,26=5.69, p<0.05; male =11.04±0.28cm; female =9.93±0.35cm) and paw place positioning (F_1,26=6.06, p<0.05; male=0.83±0.09cm; female =0.47±0.09cm), with smaller values for females compared to males for each variable (Figs S4, S5). Sex differences were also found for forelimb stride frequency (F_1,26=4.815, p<0.05; male =1.37±0.03 steps/s; females =1.50±0.13 steps/s), which was increased for females, likely compensating for their shorter stride. With the exception of brake duration, post-test comparisons demonstrated significant differences were between WT males and females (Figs. S4, S5). There was a sex by genotype interaction for forelimb step angle (F_{1, 25}=5.03. p<0.05), with male R237H rats showing a reduction (63.7±4.0 degrees) compared to WT (72.9±2.7 degrees) and females demonstrating no change (Fig. S5). One caveat of the study is that the 16-week group did not have a balanced number of animals per sex and genotype that could perform the task sufficiently for analysis. Although there is a main effect of sex for some measures, direct comparisons show most differences are between WT rats, and there are no consistent differences to suggest R237H males or females have a more severe phenotype.

3.7. Cerebellar hypoplasia in R237H rats

We previously demonstrated pathology in the R237H rat spinal cord, including myelin deficits in the corticospinal tract [14]. We also found increased water content suggesting edema in spinal cord and most brain regions analyzed, with the exception of cerebellum where water content was decreased (Fig. 8A, males at 8 weeks of age). Here we show that cerebellar mass is decreased in R237H compared to WT rats (Fig. 8B). Although the brains of R237H rats are smaller in general [14], other brain regions analyzed did not show a decrease in wet weight. R237H rats show an overall reduction in brain mass by 8% at 8 weeks of age, whereas the reduction in cerebellum is 19%. Area measures from sections taken at the sagittal midline also show a decrease in size for R237H rats (F_1,22=25.8, p<0.0001) with no effect of sex (Fig. 8C–D). All 10 lobules are present, and the lack of enlarged sulci suggests that the reduction in mass is not due to atrophy. Measures using fresh tissue at 16 weeks gave similar results with significant effects of genotype (F_1,19=29.7, p<0.0001) and sex (F_1,19-=22.9, p=0.0001), and a significant sex by genotype interaction (F_1,19-=22.9, p=0.0001). Both mass and midline area were reduced in male R237H rats (supplemental Fig. S6A,B). At this age, female R237H rats showed a trend for reduced cerebellar mass, but they were not significantly different from WT females in post-test comparisons (p = 0.058, Fig. S6A). The molecular and granule cell layers are intact, and Purkinje cells are present (Fig. 8E). Astrocytes in the white matter and deep cerebellar nuclei show GFAP accumulation and prominent Rosenthal fibers (Fig. 8F, supplemental Fig. S6C), while Bergmann glia are relatively spared [14].

Fig. 8.

Cerebellar pathology in the R237H rat. (A,B) Percent change in ratio of water content over dry brain region mass (g/g) (A) and wet weights (B) in R237H rats versus wild type at 8 weeks of age (n = 6 males, multiple t-tests, ***p < 0.001, ****p < 0.0001, error bars = standard deviation; OB – olfactory bulb, CX – cortex, HC – hippocampus, MB – midbrain, CB – cerebellum, BS – brainstem, SC – spinal cord). (C,D) Representative contrast images of cerebellum (40 μm) at the sagittal midline in R237H rats and wild type controls (C) and area measures for size comparison (D, **p < 0.01, two-way ANOVA, Holm-Šídák’s post-tests, n = 7 R237H males, 7 R237H females, 6 WT male, 6 WT females). (E) Calbindin immunofluorescence (red) labels Purkinje cells in the molecular layer in both R237H and wild type cerebella. DAPI labeled nuclei (blue) highlight the granule cell layer. (F) GFAP immunofluorescence in the cerebellar nuclei of wild type and R237H rats at 8 weeks of age (n = 2 R237H males, 2 R237H females, 1 WT male, 3 WT females, representative images shown in E-F). Scale bars apply to all images in each panel. Panel (A) represents re-analyzed data from Fig. 4N in [14].

4.0. Discussion

This study characterized motor performance and gait in an R237H rat model of Alexander disease [14]. Earlier results were both replicated and extended, providing a more detailed description and timeline of motor dysfunction. R237H rats show dramatic delays in weight gain after weaning, a developmental period normally characterized by conspicuous somatic growth. Although body condition improves over time, deficits in motor function persist and most notably disturbances of gait.

R237H and WT rats did not differ in body weight at 2 weeks of age, and no differences in open field activity were found between genotypes at 3 weeks of age. However, an enduring failure to match WT weight gains was evident by 6 weeks of age and continued throughout developmental maturation in both male and female R237H rats. Failure to thrive and developmental delay are common features of AxD, particularly with infantile and juvenile onset [3]. At 8 weeks of age, when R237H rats are frail and gaunt, males and females showed lower activity in the open field across most measures. At 16 weeks of age, when body condition has improved and body weight increased, differences in the open field were only found for males with reduced vertical activity and center time in R237H rats compared to WT. No significant differences were found between female R237H and WT rats at this age, indicating improved activity in R237H rats of both sexes. The parallel between improved body condition and mobility suggests that failure to thrive contributes to motor impairment in the R237H model. Neuroinflammation is prominent in both animal models and human AxD [21, 22], and cytokines related to cachexia and sickness behavior are elevated in the R237H rat central nervous system (CNS) [23–25]. Whether CNS stress and immune responses affect appetite and metabolism, or deficits in oral motor function lead to deficient weight gain is not clear. These questions and the effects of nutrition and body condition on neurological deficits and motor function in AxD are the focus of ongoing studies.

The open field procedure assesses general mobility and basic motor function and is particularly useful for severely impaired animals. However, the accelerating rotarod, which assesses balance, grip strength, as well as motor coordination and learning, can detect more subtle differences in motor function [26, 27]. We previously reported that at 8 weeks of age, R237H rats were weak, had great difficulty crossing a horizontal ladder, and could not stay on the rotarod [14]. Here we tested rats on the rotarod over 3 consecutive days at 4 weeks of age and show that deficits in balance and overall motor coordination are apparent at this earlier age. Wild type and R237H rats did not differ significantly in time on the rotarod on the first 2 trials on day 1 of testing, suggesting that baseline performance of R237H rats was similar to WT rats. However, WT rats showed rapid improvement in performance by trial 3 on day 1 while R237H did not improve, and this difference in performance was apparent in subsequent trials. Improved performance over repeated testing in the accelerating rotarod reflects motor skill learning [28]. Therefore, the observation that R237H rats showed little or no evidence of improvement indicates impaired motor learning. Several brain regions associated directly with motor function are activated during rotarod performance based on increased c-Fos levels, including the cerebellum, motor cortex, cingulate cortex and dorsal striatum [29]. However, neural circuitry in brain regions not typically classified as motor regions, including hippocampus (dentate, CA1 and CA3), amygdala, and infralimbic cortex [30] also show c-Fos activation during rotarod performance. These observations are relevant in view of decreased numbers of neurons in the dentate gyrus reported earlier [23] and the decrease in size of the cerebellum reported here (Fig. 8), which may contribute to rotarod impairment.

Given the contrast in results from the open field test showing no deficits at 3 weeks of age, and rotarod demonstrating impairment at 4 weeks, we wanted to use a more nuanced test to assess changes in normal mobility. Therefore, DigiGait was used to quantify gait disturbances at different stages of disease. In this report, we found both spatial and temporal anomalies in the R237H rat for both forelimb and hindlimb gait. At 4 weeks, forelimb and hindlimb swing duration and hindlimb step angle were increased in R237H rats compared to WT. These 4-week gait deficits, although mild, indicate motor disturbance early in the development of AxD pathology, and are consistent with the motor deficits found with rotarod testing in R237H rats at this age. Adult patterns of gait are mostly complete by postnatal day 27 in rats [31], and by way of comparison, brain development in rodents at 4 weeks of age and musculo-skeletal development at 3 weeks have been compared to human development at 2 years of age [32].

At 16 weeks of age gait deficits were found across several measures in both male and female rats indicating progression of motor deficits into adulthood. These included stride length, duration, and frequency in both forelimbs and hindlimbs. This could be the result of reduced muscular strength and/or central effects on neural mechanisms of gait. Paw drag, a measure of the robustness of the lift-off stage of the hind-paw from the treadmill belt, was abnormal in R237H rats compared to WT rats at 16 weeks of age, but not at 4 weeks. Similar deficits are observed in a mouse model of spastic paraplegia [33], and given the phenotypic overlap with AxD [6, 34, 35], these results suggest that gait analysis is a useful measure in detecting clinically relevant phenotypes such as rigidity and spasticity [36, 37]. Open field shows differences in vertical activity at 16 weeks, but not in horizontal activity or distance traveled, suggesting gait analysis is a more sensitive measure and confirming that motor impairment persists, even with improved body condition at this age. Swing duration, in particular, was a consistent metric, showing robust and persistent differences in both forelimbs and hindlimbs. These results also demonstrate the value of gait analysis as a translational metric that should be useful in the development of therapeutics for AxD.

No differences were found between R237H and WT rats in the elevated plus maze, suggesting that anxiety is not a major feature of the rat R237H model of AxD. Therefore, the finding that R237H rats spent less time in the center of the open field, and consequently more time in the margins in the open field (thigmotaxis), may not indicate increased anxiety in R237H rats. It should be noted that anxiety has not previously been reported as a clinical feature of AxD, and when it has been noted in anecdotal reports it is difficult to determine whether it is a simple co-morbidity.

Possible sex differences in the open field were found where male R237H rats showed less vertical activity and reduced center time compared to male WT rats, while this was not seen in females. Effects of sex were also detected in the different measures of gait analysis but mostly between WT animals, and there was no consistent pattern to suggest one sex was more impaired than the other in R237H rats. Initial studies in a mouse model of AxD reported subtle sex-dependent deficits in contextual fear conditioning. Specifically, 10-week-old female AxD model mice showed less freezing to context than female WT, but this was not found for male mice [38, 39]. Ex vivo diffusion imaging in the brains of 8-week-old R237H rats found changes in fractional anisotropy (FA), mean diffusivity (MD), and neurite density index (NDI) in several brain regions [40]. Some were sex-dependent, with female R237H rats showing a greater change in FA than male R237H rats in neocortex, as well as larger changes in MD and NDI in hippocampus. It should be noted that the available literature on human AxD does not show a consistent effect of sex on basic measures of disease severity such as age of onset [6, 41]. Nevertheless, our results suggest that future studies should consider potential sex differences in motor function in AxD.

As R237H rats mature past the critical period between 6 to 12 weeks of age when animals are severely impaired and show increased mortality, they begin to gain weight and their overall appearance improves. This raises questions regarding whether testing those that survive skews comparisons across age groups as well as questions concerning mechanisms of apparent recovery. In the present study, mortality was less than 5% of the total number of rats used, making it very unlikely that a “survivor” effect could account for the apparent improvement in neurological function found at 16 weeks in R237H rats. Why animals start gaining weight at this later stage is not clear. GFAP mutations lead to protein accumulation and aggregation, and protein degradation pathways, including ubiquitin-proteasome and autophagy-lysosomal systems, are likely relevant to disease progression [42–44]. Rosenthal fibers are labeled with ubiquitin and the selective autophagy adaptor p62/SQSTM1 [14], and although shifts in the balance of GFAP turnover could reduce the pathological burden, GFAP levels in the R237H rat brain are the same at 8 and 16 weeks of age [14]. It is interesting that improvement in symptoms (not necessarily motor) is occasionally seen in human AxD, though the number of such case reports is very small [45–47]. All of these examples involved individuals with adult-onset, two with vomiting as the primary problem, and others with onset precipitated by head trauma [48] or alcohol abuse [49] from which some recovery might have been expected after the acute injury subsided. Hence the relevance of the improvement phase in the rat model for the human disease remains to be determined. Future studies to analyze expression of small heat shock proteins and activation of protective stress pathways [23, 50, 51], which could potentially alleviate protein aggregation or oxidative stress without reducing GFAP, may lead to a better understanding of disease severity in older animals and AxD pathology in general.

The R237H rat exhibits extensive spinal cord pathology including myelin deficits and degenerating axons in the corticospinal tract, enlarged gray matter neurites with debris accumulation, and reduced expression of KCNJ10, AQP4, GJA1, and SLC1A2, indicative of general astrocyte dysfunction [14]. Whether axonal changes result from local pathology or interference in upper motor neuron circuits is not clear, but disruption of spinal cord pattern generators alone could be responsible for the observed motor phenotypes [52–54]. In this report, we also note cerebellar hypoplasia. Cerebellar involvement is common in AxD, with signal abnormalities apparent in MRI [6, 9, 55] and pathology often located in the dentate nucleus [56]. Astrocyte populations in the cerebellum are heterogeneous among the different layers, with highly specialized Bergmann glia in the molecular layer, velate astrocytes in the granule cell layer, and fibrous astrocytes in the white matter. Conditional ablation of astrocytes, including Bergmann glia, in adult mice leads to loss of granule cells and motor impairment with a staggering gait and paralysis [57], and during early postnatal development leads to disruption of the normally discrete cerebellar layers and smaller cerebellar size [58]. Interestingly, GFAP-null mice have been reported to be impaired in LTD and eyeblink conditioning but are not impaired in the rotarod [59]. In the rat model of AxD, Bergmann glia are relatively spared with little pathology, while velate astrocytes in the granular layer demonstrate a uniquely dense accumulation of GFAP around the nucleus [14]. Here we show that GFAP pathology is most prominent in fibrous astrocytes of the cerebellar white matter and protoplasmic astrocytes within the cerebellar nuclei (Fig. S6), suggesting potential effects on the circuitry connecting the cerebellum to other brain regions.

Animal models with severe cerebellar atrophy, including Purkinje and granule cell loss, demonstrate impaired motor function, but postural sensorimotor learning on the rotarod is not necessarily abolished [60]. Adaptation, even after complete ablation of the cerebellar cortex, is thought to be coordinated by the deep cerebellar nuclei and suggests recruitment of other brain regions involved in the complex circuitry of cerebellar signaling [61]. Synaptic plasticity is apparent in both the molecular and granular layers as well as the deep cerebellar nuclei. However, climbing fibers from the inferior olivary nucleus transmit error signals to modulate synapses between parallel fibers (originating from granule neurons) and Purkinje cells to adapt to changing sensory input [62]. Ablation of the inferior olive specifically leads to deficits in motor learning on the rotarod [63], and the pathology observed in the dentate nucleus and cerebellar peduncles, as reported here, or alternatively in the inferior olive could lead to the deficits in learning the rotarod task observed in the R237H rat.

The dentate nucleus is important for fine motor movement, and dysfunction may be reflected by reduced cerebellar size and ataxia. It is also part of the triangle of Guillain-Mollaret (GMT), a circuit consisting of the dentate nucleus, the red nucleus, and the inferior olivary nucleus, and GMT damage can cause palatal myoclonus and ocular movement disorders [64], both of which are common clinical features in type II AxD [6]. A recent retrospective analysis of MRI signs in AxD patients found T2 hyperintensity of the central inferior olivary nucleus in 87% of those tested, with no difference between type I and II disease [65]. The dentate nucleus is located at the roof of the fourth ventricle and may be particularly susceptible to periventricular astrocyte pathology which is prominent in AxD. White matter pathology and axonal damage in the GMT circuitry could also affect neurons in both cerebellar and brainstem nuclei. More extensive analysis to measure cerebellar layers and quantify different cell populations in both the vermis and lateral hemispheres at different ages will be necessary to determine pathological processes and confirm whether reduced cerebellar mass in the R237H rat is the result of hypoplasia or atrophy. These studies along with a more detailed analysis of brainstem nuclei will be the focus of future efforts to understand the role of different cerebellar afferents in motor deficits, potentially including nystagmus and palatal myoclonus, in the model.

Body-weight differences likely contributed to some, but not all, differences in motor performance found between WT and R237H rats. Previous measures of reduced grip strength in R237H rats [14] are plausibly related to muscle mass. However, the extent to which body weight per se may have contributed to the additional motor deficits reported here is difficult to establish. In the open field test the lighter R237H rats showed generally lower activity than WT, and it is unclear how body weight differences would predict this outcome. On the accelerating rotarod heavy animals are typically reported to fall sooner [66]. The fact that the lighter R237H fell sooner than the heavier WT rats indicates that differences in performance between groups were not due solely to body weight. The possible effects of body weight and size on gait are also difficult to predict. For example, step angle and swing duration differed between WT and R237H rats at 4 weeks of age when body weight did not show significant differences. Similarly, at 16 weeks, the smaller R237H rats had larger mean stride length and stride duration compared to WT, and how this may relate to body weight is unclear. Finally, hindlimb paw drag appears to be a gait anomaly that is not easily ascribed solely to size differences.

5.0. Conclusions

In this report we demonstrate a progression of motor impairment in the R237H rat from an early symptomatic period, through a severe stage of decline, followed by apparent stabilization of disease. We show that automated gait analysis captures the degree and nature of impairment at different ages and may be a useful correlate for the human condition. Swing duration may be a particularly useful metric, in that changes were evident at an early age and persisted with disease progression. These results further demonstrate the utility of the model for preclinical studies, including relevant motor and cognitive outcome measures for testing therapeutics. In addition, we identify cerebellar hypoplasia as a future area of study in the model to connect functional deficits with regional pathology. We have recently shown that astrocyte pathology in hippocampus leads to a dramatic neuroinflammatory response with loss of synaptic and mitochondrial proteins [23], and given the heterogenous nature of astrocyte pathology in AxD, understanding the impact of reactive gliosis and the response to mutant GFAP in other regions of the CNS will be important in interpreting the variable presentations in the human disease.

Supplementary Material

1

Supplemental Fig. S1. Open field activity for male and female WT and R237H rats. The number of infrared beam breaks are shown as a measure of horizontal and vertical activity in the open field arena over the 30-minute test period for animals at 3, 8, and 16 weeks of age. Total distance traveled (cm) is also shown as well as the time spent (s) in the center of the arena. (n=10 per group at 3 and 8 weeks; n=12 for WT and 13 for R237H rats at 16 weeks, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, Holm-Šídák’s post-tests).

2

Supplemental Fig. S2. Rotarod performance for WT and R237H rats. Rats were tested on 3 consecutive days on an accelerating rotarod to assess motor function and coordination. Graph shows the same data as Figure 3, but includes individual data points (n=15 male and 5 female WT; 7 male and 13 female R237H rats, *p<0.05, **p<0.01, ***p<0.001).

3

Supplemental Fig. S3. Elevated plus maze performance for male and female WT and R237H rats. (A) The amount of time spent (s) in the open and closed arms and in the center of the maze is shown as a measure of anxiety. (B) Mean velocity (cm/s) on the elevated plus maze for WT and R237H rats is shown as an indication of the ability to perform the test. (n=9 WT and 9 R237H males, n=11 WT and 9 R237H females, ***p<0.001).

4

Supplemental Fig. S4. Temporal gait measures at 4 and 16 weeks of age for male and female WT and R237H rats. (A) stride frequency (steps/s), (B) stride duration (s), (C) swing duration (s), (D) brake duration (s), (E) propel duration (s), and (F) paw drag, for the two age groups. (Males and females tested; n=8 at 4 weeks; n=8 male and 9 female WT, and 7 male and 6 female R237H rats at 16 weeks; *p<0.05, **p.0.01, ***p<0.001, ****p<0.0001, Holm-Šídák test; † indicates significant effect of sex, ‡ indicates significant sex × genotype interaction, two-way ANOVA).

5

Supplemental Fig. S5. Spatial gait measures at 4 and 16 weeks of age for male and female WT and R237H rats. (A) stride length (cm), (B) step angle (deg), (C) absolute paw angle (deg), and (D) paw placement positioning (cm), for the two age groups. (Males and females tested; n=8 at 4 weeks; n=8 male and 9 female WT, and 7 male and 6 female R237H rats at 16 weeks; *p<0.05, **p.0.01, ***p<0.001, Holm-Šídák test; † indicates significant effect of sex, ‡ indicates significant sex × genotype interaction, two-way ANOVA).

6

Supplemental Fig. S6. Cerebellar pathology in the R237H rat at 16 weeks of age. (A) Dorsal view of cerebellum wild type and R237H rats (males shown) and corresponding wet weight at 16 weeks of age (n = 6 WT and 5 R237H males; n = 6 WT and 6 R237H females, 2-way ANOVA with sex and genotype as variables, ****p<0.0001, Holm-Šídák post-test, comparison between female rats which did not reach significance, p=0.058; † indicates significant effect of sex, ‡ indicates significant sex × genotype interaction, two-way ANOVA). (B) Images of acutely isolated cerebella and area measures at the sagittal midline (n = 6 WT, 5 R237H male rats at 16 weeks, ***p<0.001, t-test). (C) H&E staining shows Rosenthal fiber accumulation (arrows) in astrocytes within the cerebellar nuclei and white matter adjacent to the 4^th ventricle (females at 16 weeks of age, n = 4). Error bars = standard deviation. Scale bars apply to all images in each panel.

Data Availability:

The datasets for the current study are available from the corresponding author on reasonable request.