Dysregulation of alternative splicing in spinocerebellar ataxia type 1

Abstract

Spinocerebellar ataxia type 1 is caused by an expansion of the polyglutamine tract in ATAXIN-1. Ataxin-1 is broadly expressed throughout the brain and is involved in regulating gene expression. However, it is not yet known if mutant ataxin-1 can impact the regulation of alternative splicing events. We performed RNA sequencing in mouse models of spinocerebellar ataxia type 1 and identified that mutant ataxin-1 expression abnormally leads to diverse splicing events in the mouse cerebellum of spinocerebellar ataxia type 1. We found that the diverse splicing events occurred in a predominantly cell autonomous manner. A majority of the transcripts with misregulated alternative splicing events were previously unknown, thus allowing us to identify overall new biological pathways that are distinctive to those affected by differential gene expression in spinocerebellar ataxia type 1. We also provide evidence that the splicing factor Rbfox1 mediates the effect of mutant ataxin-1 on misregulated alternative splicing and that genetic manipulation of Rbfox1 expression modifies neurodegenerative phenotypes in a Drosophila model of spinocerebellar ataxia type 1 in vivo. Together, this study provides novel molecular mechanistic insight into the pathogenesis of spinocerebellar ataxia type 1 and identifies potential therapeutic strategies for spinocerebellar ataxia type 1.

spinocerebellar ataxia type 1, neurodegeneration, alternative splicing, Purkinje cell

Introduction

Spinocerebellar ataxia type 1 (SCA1) is a dominantly inherited, progressive, neurodegenerative genetic disorder characterized by motor loss and severe degeneration of specific neuronal populations, including cerebellar Purkinje cells (PCs) [1]. This disease is caused by a CAG-trinucleotide repeat expansion in the ATXN1 gene, leading to an expanded polyglutamine (polyQ) tract in the ATAXIN-1 protein [1]. Although ataxin-1 is broadly expressed throughout the brain, PCs are one of the populations predominantly affected by this disease. It is not well known why PCs are one of the most susceptible neurons to degeneration in SCA1. To understand the importance of PC degeneration in SCA1, transgenic mouse models were developed that express mutant ataxin-1 specifically in the PCs of the cerebellum, and recapitulate key aspects of neurodegeneration and motor deficits observed in SCA1 patients [2, 3]. The precise mechanism by which mutant ataxin-1 causes SCA1 still remains elusive, but it is known that nuclear ataxin-1 function is crucial for the development of SCA1 [4, 5]. In the nucleus, ataxin-1 can regulate gene expression [4–8]. Ataxin-1 alone may not bind directly to DNA, but it can regulate transcription by binding to transcription factors [9–11]. Ataxin-1 also interacts with RNAs and various splicing factors [11–15], which play a role in multiple steps in the creation of alternative splicing events [16, 17]. However, it has not yet been shown whether the polyQ-expanded mutant ataxin-1 expression leads to misregulated alternative splicing (mAS) events in SCA1.

The biological consequences of splicing dysregulation in the context of neurodegenerative diseases has been well established by research in various fields, including spinal muscular atrophy (SMA) and transactivating response region DNA binding protein 43 (TDP43) [18–23]. SMA is predominantly caused by a deletion or mutation in the SMN1 gene, and although there is a nearly identical copy, SMN2, a skipping of exon 7 in SMN2 prevents it from compensating for the loss of SMN1 [20, 21]. The FDA approved use of the antisense oligonucleotide SPINRAZA that targets intronic splicing silencer N1 and prevents exon 7 from being skipped [24], thereby creating a cure for a neurodegenerative disease caused by alternative splicing. Another example of splicing dysregulation can be seen in TDP43. TDP43 is a major protein component involved in disease pathogenicity in a variety of neurodegenerative diseases, including amyotrophic lateral sclerosis and frontotemporal lobar degeneration. Mutations in TDP43 have been shown to cause splicing dysregulation that may lead to neurodegeneration [18, 19, 25]. Taken together, these examples of splicing dysregulation in neurodegeneration underscore the importance of understanding whether mAS events occur in SCA1, as this might help identify critical mechanisms and pathways in which mutant ataxin-1 leads to disease pathology in SCA1.

To better understand the role of mAS in SCA1, we first utilized the PC-specific SCA1 transgenic mouse model (B05) that expresses mutant ataxin-1 only in the PCs of the cerebellum [2]. We found that mAS occurs throughout SCA1 disease progression and that the predominant form of mAS events are ``skipped exons'' of target genes. We also show that the majority of mAS events do not overlap with differentially expressed genes (DEGs) (<25% overlap), and that SCA1 B05 mAS events are predominantly located in PCs where mutant ataxin-1 is expressed. We further show that mAS reveals overall different, novel biological pathways compared to DEG analysis alone, and that several genes with misregulated skipped exons are disease causative. In addition, we found that mAS events also occur in the cerebella of SCA1 knock-in mice. Finally, we found evidence that a subset of mAS events in SCA1 occur via interaction with the splicing factor Rbfox1, which can modulate degenerative phenotypes in a Drosophila model of SCA1 in vivo.

Materials and methods

Mouse husbandry

All animal care procedures were approved by the Yale University Institutional Animal care and Use Committee (IACUC; Approval Code: 2021-11342, Approval Date: 21 December 2021) as previously published [26]. Mice were kept in a 12-h light/dark cycle with standard chow and ad libitum access to water. Two mouse models were utilized for described experiments: SCA1 B05 (Pcp2::ATXN1 [82Q]) mice [2] were maintained on a pure FVBN/J background and SCA1 knock-in (Atxn1^154Q/2Q) mice [27] on a pure C57BL/6J background were reanalyzed from previously published datasets [7]. Both male and female mice were utilized unless otherwise stated.

RNA extraction

Whole mouse cerebella were macro-dissected, flash frozen, and stored at −80°C until further processing. RNA was extracted using the Qiagen RNeasy Mini Kit (Qiagen, 74106) per the manufacturer’s protocol. DNA was removed with DNase I as described in the manufacturer’s protocol. All RNA was extracted from male mice.

Real-time quantitative PCR

cDNA was synthesized using oligo-dT primers and the iScript cDNA synthesis kit (BioRad, 1708891). RT-qPCR was performed using iQ SYBR Green Supermix (Bio-Rad, 1708880). All samples were loaded in triplicate and relative fold change was normalized to Gapdh and Actb using Bio-Rad CFX manager software (BioRad, 1845000). See also Supplementary Table 9 for primers used.

RNA sequencing

RNA was sent to the Yale Center for Genomic Analysis for processing. RNA integrity was measured with capillary electrophoresis (Agilent BioAnalyzer 2100, Agilent Technologies, Santa Clara, CA, USA) and RIN values were checked to ensure RNA integrity. Libraries were generated and quantified before pooling and samples were sequenced on a NovaSeq using HiSeq 100 bp paired-end read strategy with 30 M reads per samples. Sequencing data can be found at GEO accession # (GSE232368).

RNA sequencing data analysis

Sequencing reads were mapped to the mouse genome version mm10 using STAR version 2.7.9 [28]. Reads were counted for each gene using HTSeq version 0.13.5 [29]. Differential expression analysis was carried out using DESeq2 package in R [30]. Genes were called differentially expressed based on an FDR adjusted P-value < 0.05. Euler and Venn diagrams were created using the eulerr package in R [31]. Gene Ontology biological process analysis was conducted using enrichGO tool from the clusterProfiler v4.8.1 package [32].

RNA splicing event data analysis

Misregulated alternative splicing event analysis was carried out on STAR aligned reads using rMATS version 3.1.0 [33]. A cutoff of FDR adjusted P-value < 0.05 was used to establish significance in the analysis. rMATS reports alternative splicing events as inclusion in one sample group versus inclusion in another sample group. In this context, inclusion is defined by the presence of the alternatively spliced feature. For example, in the case of a skipped exon with higher inclusion in SCA1 B05, the proportion of reads mapped to a skipped exon compared to surrounding exons is higher in SCA1 B05 than wild-type. Gene Ontology (GO) biological process analysis was conducted using enrichGO tool from the clusterProfiler v4.8.1 package [32].

Single cell analysis

Lists of total misregulated alternatively spliced genes and the top 50 differentially expressed genes (DEGs) were input as a search query in the single cell dataset from Kozareva et al. [34]. The search was done using the Single Cell Portal hosted by The Broad Institute. Violin plots were generated from the adult cerebellum data, which was grouped by clusters representing cell types and subsampled from over 600 000 cells to 100 000 cells. The expression for each gene plotted in each cell type cluster was collapsed by the mean value across cells.

Toluidine blue staining and quantification of Purkinje cells

Mice were anesthetized with isoflurane (Henry Schein, 11695-6776-2) prior to intracardial perfusion with phosphate buffered saline (PBS) and 4% paraformaldehyde. Brains were postfixed overnight in 4% paraformaldehyde before incubation in 20% and 30% sucrose in PBS for 24 h each. Samples were frozen in O.C.T. compound (VWR, 4583) and sliced into 30-μm sections on a cryostat (Leica). Free-floating sections were washed in PBS and PBS with 0.1% Triton-X and subsequently mounted onto slides. Slides were dipped in 0.5% Toluidine Blue O (Sigma-Aldrich, T3260) in PBS for three seconds and washed in PBS until residual stain was removed. Slides were dried prior to cover slipping with Permount (Fisher Scientific, SP15-100). Brightfield images were taken at 20× magnification on an Olympus VS200 microscope. Images were analyzed on Olympus OlyVIA V3.4.1 (Build 26606) by a blinded observer. For Purkinje cell (PC) number, the total number of PCs were quantified across a 250-μm distance in lobes 5/6. For molecular layer thickness analysis, the molecular cell layer was measured from the base of the PC to the midline at 3 points 100-μm apart in lobes 5/6. Three midline brain sections were analyzed per mouse.

Drosophila genetic interaction study

Targeted misexpression experiments were carried out using the Glass Multiple Repeat (GMR-Gal4) enhancer, which directs the expression of transgenes of interest in cells including the differentiating retinal neurons of the fly eye. The transgenic flies were crossed with GMR-Gal4 and the lines showing phenotypes in the fly eye were used for further study and analysis. We used Gal4/UAS targeted misexpression system for the study [35]. GMR-Gal4/+; UAS-lacZ, UAS-ATXN1^82Q was used to sample the SCA1 pathology in the fly eye. All the fly stocks and the progeny of the result of genetic interaction in experimental fly lines were raised on a standard fly food and maintained at a temperature of 25°C.

Drosophila stock lines used in the study

The transgenic fly lines utilized in the study are: GMR-Gal4/CyO; +/+ (BL 1104), UAS-lacZ/UAS-lacZ; +/+ (BL 8529), +/+; UAS-lacZ/UAS-lacZ (BL 8530), +/+; UAS-ATXN1^82Q/+ (BL 33818), UAS-RBFOX1/UAS-RBFOX1; +/+ (BL 82374), and +/+; UAS-Rbfox1^RNAi/UAS-Rbfox1^RNAi (BL 32476).

Adult fly eye imaging

The adult flies were screened at two days after eclosion and prepared for fly eye imaging by freezing them at −80°C. The bright field images of fly eye were captured using a color camera (Leica M170 HD) at 10× magnification and mounted on a Leica M125 microscope using Z-sectioning approach. The images were then saved in .tiff format and analyzed using Fiji (v2.3.0) [36]. The fly eyes were quantified manually for red pigmentation by splitting each fly eye into three compartments (C1, C2, C3). Each compartment was checked for the presence or absence of the red color in the eye and assigned the numeric values of 0 or 1. A value of 0 indicates the absence of red color pigment while a value of 1 indicates the presence of red color pigment. The values for each of the compartments were summed for a final value and converted to percentage. The eyes were also quantified manually for black spots. Individual fly eyes were checked for the presence or absence of black spots in area that corresponds to more than 25% of the fly eye and assigned numeric values of 0 or 1. A value of 0 indicates the absence while a value of 1 indicates the presence of black spots in area ≥ 25% of the fly eye. Statistical analysis and graphs were done using GraphPad prism software.

Results

Progressive cerebellar degeneration in SCA1 B05 mice

To test whether mAS events occur in SCA1, we analyzed splicing events in the cerebellum of the PC-specific SCA1 B05 mouse model at multiple time points (Fig. 1). First, we aimed to verify the degree of neurodegeneration across early and late time points in the SCA1 B05 cerebellum (Fig. 2). We observed progressive thinning of the molecular layer starting at 12-weeks of age (P = 0.028) and continued reduction observed until 52-weeks of age (P < 0.0001) in the SCA1 B05 relative to wild-type (WT) controls (Fig. 2B). Further, significant PC loss was observed at 52-weeks of age (P < 0.0001) in the SCA1 B05 relative to WT controls (Fig. 2C). Together, we concluded that our SCA1 B05 mouse line displays similar degrees of PC pathology as previously reported [2, 3, 37] and that more than 50% PCs persist until 52-weeks of age. This validation of the SCA1 B05 model, as well as the presence of PCs at 52-weeks of age, allowed us to analyze splicing events throughout disease progression.

Figure 1

Schematic of project workflow. In brief, SCA1 B05 mice and littermate WT controls were collected across stages of disease, including 5-, 12-, 20-, and 52-weeks of age. Cerebellar samples were prepared for immunohistological analysis and RNA sequencing in order to investigate differential gene expression and alternative splicing events.

Figure 2

Pathological characterization of SCA1 B05 mouse cerebellum. (A) Representative images of WT (top row) and SCA1 B05 (bottom row) cerebellar lobes 5/6 stained with toluidine blue at 5-, 12-, 20-, and 52-weeks of age. Arrow indicates PC and double-pointed arrow indicates molecular layer thickness. Scale bar = 100 μm. (B and C) Quantification of molecular layer thickness (B) and PC number (C). Points in graphs indicate average measurements per mouse (n = 3 per genotype). Data are represented as mean ± SEM, p-value plotted in graphs, ns = not significant, by two-way ANOVA with Šídák’s multiple comparisons test.

Alternative splicing events are misregulated in SCA1

To investigate how alternative splicing events occur in SCA1, we first performed bulk RNA sequencing (RNA-seq) of the SCA1 B05 cerebellum and WT littermate controls at 5-, 12-, 20-, and 52-weeks of age (Fig. 3A, Supplementary Fig. 1A, Supplementary Table 1). We first aimed to confirm that mutant ataxin-1 expression leads to differential gene expression. We observed an increase in total DEGs over time from 422 DEGs at 5-weeks of age to 2590 DEGs at 52-weeks of age (Fig. 3A, Supplementary Fig. 1A). Down-regulated DEGs at 5-, 12-, 20-, and 52-weeks highly overlap (Fig. 3B, Supplementary Fig. 1B), while the set of up-regulated DEGs is largely unique for each time point (Fig. 3B, Supplementary Fig. 1B). Gene Ontology (GO) biological process analysis of the shared down- and up-regulated DEGs at 12-, 20-, 52-weeks of age showed down-regulation of synaptic structure and ion channel genes but up-regulation of neuronal death, gliogenesis, and extracellular structure GO hits (Fig. 3C). Together, these data validate that our sequencing analysis reveals progressive DEG changes in the SCA1 B05 relative to WT.

Figure 3

Generation and analysis of differentially expressed genes in SCA1 B05 mouse model. (A) Venn diagram showing overlap of total DEGs in SCA1 B05 relative to WT controls across timepoints at 12-, 20-, and 52-weeks of age. (B) Venn diagram showing overlap of down-regulated (left) and up-regulated (right) DEGs in SCA1 B05 relative to WT controls across timepoints at 12-, 20-, and 52-weeks of age. (C) Top 10 GO term hits for DEGs that were down-regulated across timepoints (left) at 12-, 20-, and 52-weeks of age and DEGs that were up-regulated across timepoints (right) at 12-, 20-, and 52-weeks of age in the SCA1 B05 cerebellum. GO terms are plotted with the circle circumference indicating gene count and the color indicating adjusted p-value, with x-axis indicating gene-ratio. Also see Supplementary Fig. 1.

Next, we were interested in identifying whether mAS events occur across time points in the SCA1 B05 cerebellum. We investigated this using the rMATS tool [33] (Supplementary Tables 2–5), which can identify five separate alternative splicing events, including: skipped exons (SE), alternative 5′ splice sites (A5SS), alternative 3′ splice sites (A3SS), mutually exclusive exons (MXE), and retained introns (RI). The number of statistically significant mAS events are summarized in Table 1A. Similar to the trend observed in our DEG analysis, the number of mAS events increases progressively over time between 5- and 20-weeks of age in the SCA1 B05 model (Fig. 4A). Comparing the mAS events to the set of DEGs at each time point reveals that fewer than 25% of the differentially spliced genes are also differentially expressed (Fig. 4B). The most abundant type of differential splicing event observed is the skipped exon (Fig. 4A, Table 1A); therefore, we primarily focus on this gene set for the remaining mAS analysis unless otherwise stated.

Table 1

Misregulated alternative splicing events in SCA1 mouse cerebellum.

A
SCA1 B05
	5 weeks		12 weeks		20 weeks		52 weeks
Splicing event	WT Inclusion	SCA1 B05 Inclusion	WT Inclusion	SCA1 B05 Inclusion	WT Inclusion	SCA1 B05 Inclusion	WT Inclusion	SCA1 B05 Inclusion
Skipped Exon (SE)	124	116	108	101	263	222	257	201
Alternative 5′ Splice site (A5SS)	14	16	14	12	19	25	19	20
Alternative 3′ Splice site (A3SS)	25	25	15	31	27	51	33	28
Mutually Exclusive Exons (MXE)	9	3	12	30	23	30	16	29
Retained Introns (RI)	2	3	3	3	7	9	10	8
B
12-week SCA1 KI
Splicing Events		WT Inclusion			SCA1 KI Inclusion
Skipped Exon (SE)		49			28
Alternative 5′ Splice Site (A5SS)		5			0
Alternative 3′ Splice Site (A3SS)		10			4
Mutually Exclusive Exons (MXE)		7			4
Retained Intron (RI)		5			7

(A) Number of misregulated alternatively splicing events observed at 5-, 12-, 20-, and 52-week time points from SCA1 B05 mouse cerebellum. (B) Number of misregulated alternative splicing event observed in the 12-week-old SCA1 KI mouse cerebellum. Data is split by the type of events reported by rMATS and on the basis of increased WT or SCA1 inclusion.

Figure 4

Alternative splicing events are misregulated specifically in PCs of SCA1 B05 mice. (A) Total number of differential splicing events observed across timepoints in the SCA1 B05 mouse cerebellum relative to WT controls. Color indicates type of splicing event, including skipped exon, A5SS, A3SS, mutually exclusive exons, and retained introns. (B) Percent of misregulated alternatively spliced genes that are also differentially expressed at different time points. (C and D) Normalized violin plots showing expression of cell type-specific expression of genes containing misregulated skipped exons in (C) 12-week SCA1 B05 and (D) 12-week SCA1 KI mouse cerebellum. Also see Supplementary Fig. 2.

Since our analysis utilized a cerebellar bulk RNA-seq dataset using the SCA1 B05 model, it is unclear whether the changes that we observed are PC-specific or if mAS events occur in other cell types as well. To investigate this question, we explored the publicly available mouse cerebellar cortex single cell dataset [34] in the context of our gene expression and splicing changes. We show that the majority of genes with misregulated skipped exons are associated with PCs at all time points in our PC-specific model of SCA1 (Fig. 4C, Supplementary Fig. 2). Analysis of DEG changes at 20-weeks of age showed that down-regulation was most notably seen in the PCs where mutant ataxin-1 is expressed in the SCA1 B05 model (Supplementary Fig. 3A), while up-regulation was observed across diverse cerebellar cell types (Supplementary Fig. 3B).

To further examine whether mAS were due to the presence of mutant ataxin-1 or specific to PCs in a more physiologically relevant condition, we reanalyzed a previously published bulk RNA-seq dataset from our lab that was generated from the SCA1 knock-in (SCA1 KI; Atxn1^154Q/+) mouse model (GEO Accession #: GSE122099) (Supplementary Table 6) [7]. In the SCA1 KI model, mutant ataxin-1 protein is expressed endogenously throughout the entire cerebellum, which is in contrast to the PC-specific SCA1 B05 model. rMATS analysis of the 12-week SCA1 KI dataset showed that mAS events are similar to the B05 model and that skipped exons are again the most common event to occur (Table 1B, Supplementary Table 7). Analysis of the mAS skipped exon events from the 12-week SCA1 KI cerebellum samples show that this mAS event occurs at a higher level in PCs compared to the other cell types (Fig. 4D). However, the degree of mAS events in the SCA1 KI is less specific to PCs compared to the B05 model, as the average expression of each gene is much more similar across cell types. Analysis of DEGs of the SCA1 KI model showed a similar pattern of relatively higher degree of down-regulation in PCs and up-regulation broadly across different cell types (Supplementary Fig. 4).

Misregulated alternative splicing events reveal new biological pathways affected in SCA1

To understand what pathways are affected by the mAS events in SCA1, we performed GO analysis of both the mAS skipped exons and DEGs, respectively (Supplementary Fig. 5A–D). As SCA1 is a progressive neurodegenerative disease, we expected to see pathways change over time. However, whether the mAS event pathways were similar to DEG pathways remained unknown. Interestingly, the majority of GO terms from mAS events do not overlap with GO terms from down- or up-regulated DEGs across all time points (Fig. 5A and B, Supplementary Fig. 5A–D). For example, GO analysis of mAS skipped exons at 20-weeks of age shows no overlap with pathways affected by down-regulated DEGs (Fig. 5A and B) or with up-regulated DEGs (Supplementary Fig. 5C). However, there are few exceptions, such as synapse organization at 12-weeks of age and regulation of membrane potential at 52-weeks of age (Supplementary Fig. 5B and D). We also performed GO analysis on the other mAS events that were not skipped exons at all time points and observed no overlap with mAS skipped exons or either down- or up-regulated DEGs (Supplementary Fig. 5E–H).

Figure 5

Differential biological pathways affected by mAS events and DEGs in SCA1. (A and B) Top 10 GO term hits for mAS skipped exon genes (A) and down-regulated DEGs (B) in 20-week-old SCA1 B05 mouse cerebellum relative to WT controls. GO terms are plotted with the circle circumference indicating gene count and the color indicating adjusted P-value, with x-axis indicating gene-ratio. (C) Venn diagram of total number of mAS skipped exon genes that are commonly and uniquely altered across all timepoints in SCA1 B05 mouse. (D) Venn diagram of total number of mAS skipped exon genes that are commonly and uniquely altered between the SCA1 KI model at 12-weeks of age and mAS skipped exon events that occurred at all timepoints in the B05 model. See also Supplementary Fig. 5.

Next, we analyzed whether mAS events commonly occurred across all time points and identified seven genes: Inpp4a, Itpr1, Tenm4, Anks1b, Trpc3, Snap25, and Cacna1g (Fig. 5C). Of these seven genes, three of them (Trpc3, Snap25, and Cacna1g) were also mAS events in the 12-week SCA1 KI dataset (Fig. 5D). Cacna1g and Trpc3 were of interest to us since these are both ion channel genes and causative for two other SCA-related diseases, SCA42 [38] and SCA41 [39], respectively. We chose to further analyze Trpc3 because it has been shown that the splicing of exon 9 forms a truncated Trpc3 isoform (Supplementary Fig. 6A) that has an increased activity and is one of the predominant isoforms found in the cerebellum [40, 41]. rMATS analysis of the exon inclusion in the set of WT and SCA1 B05 samples for Trpc3 shows that the expression of each exon is higher in WT than B05 (Supplementary Fig. 6B), indicating overall down-regulation of Trpc3 transcripts in SCA1. However, the relative inclusion of exon 9 is more frequent in B05 (average inclusion level 0.61) than WT (average inclusion level 0.26) (Supplementary Table 4, Supplementary Fig. 6B).

Altered expression of splicing factors leads to misregulated alternative splicing events

We next sought to investigate potential molecular mechanisms by which mAS events occur in SCA1. We hypothesize that changes in the expression level or activity of splicing factors are likely the cause of splicing dysregulation in SCA1. Thus, we first filtered our SCA1 B05 cerebellum DEG and mAS datasets at all time points with mouse datasets of splicing factors [42] (Fig. 6A, Supplementary Table 8) and found that 17 splicing factors were transcriptionally affected (Table 2). Next, we filtered our list of transcriptionally misregulated splicing factors against known ataxin-1 physical interactors [11]. We found that two of our transcriptionally misregulated splicing factors directly bind ataxin-1, which are Rbfox1 and Rbfox2 (Fig. 6B, Table 2). Through these analyses, we uncovered only one splicing factor, Rbfox1, that was both a DEG/mAS event and ataxin-1 physical interactor. The expression of the splicing factor Rbfox1 is significantly down-regulated at 12-, 20-, and 52-weeks, is a “mutually exclusive exons” mAS event at 20-weeks, and directly binds to ataxin-1 (Table 2). We validated Rbfox1 down-regulation at 20-weeks of age through real-time quantitative PCR (RT-qPCR) (Fig. 6C). We then utilized a database containing human RBFOX1 targets [43], which we mapped to mouse orthologs using the HUGO Gene Nomenclature Committee Comparison of Orthology Predictions tool [44–47]. The list of predicted mouse Rbfox1 targets was then compared against our total mAS gene list at all time points and showed that a significant portion (20.44%) of Rbfox1 targets (hypergeometric distribution; P < 0.0001) were mAS events in SCA1 (Fig. 6D).

Figure 6

Rbfox1 mediates mutant ataxin-1-associated mAS events and modulates neurodegenerative phenotypes in a Drosophila model of SCA1. (A) Venn diagram showing overlap of known mouse splicing factors with mAS events and DEGs of the SCA1 B05 mouse cerebellum at all timepoints. (B) Venn diagram showing overlap of transcriptionally misregulated splicing factors and ataxin-1 physical interactors. (C) RT-qPCR validation of Rbfox1 mRNA level in the 20-week SCA1 B05 cerebellum relative to WT littermate controls, normalized to housekeeping genes Gapdh, and Actb (n = 4 per genotype). Data are represented as mean ± SEM, P = 0.0087 by student’s t-test. (D) Venn diagram that shows the overlap (20.44%; P < 0.0001) of Rbfox1 targets and total mAS events of the SCA1 B05 mouse cerebellum at all time points. (E-H) Genetic interaction study between Rbfox1 and SCA1. (E and F) Light microscopy images of adult Drosophila eyes at 2 days after eclosion show the depigmentation and necrotic death (marked by arrows) phenotypes of the external eyes. Scale bar = 100 μm. (G and H) Quantification of red eye percentage (left) and black spots (right). See also Supplementary Fig. 7. The plotted data in the quantification of red eye percentage (left) is the total percentage of the entire eye field with red pigment for the respective genotypes. The plotted data in the quantification of black spots (right) is the total percentage of eyes in a genotype containing black spots in 25% or more of the eye area. Data are represented as mean ± SEM, p-value plotted in graph, by two-way ANOVA with Šídák’s multiple comparisons test. Flies were raised at 25°C and genotypes are: (i) Control (GMR-Gal4 > UAS-lacZ; UAS-lacZ), (ii) SCA1 (GMR-Gal4 > UAS-lacZ; UAS-ATXN1^82Q), (iii) RBFOX1 (GMR-Gal4 > UAS-RBFOX1; UAS-lacZ), (iv) SCA1; RBFOX1 (GMR-Gal4 > UAS-RBFOX1; UAS-ATXN1^82Q), (v) control (GMR-Gal4 > UAS-lacZ; UAS-lacZ), (vi) SCA1 (GMR-Gal4 > UAS-lacZ; UAS-ATXN1^82Q), (vii) Rbfox1^RNAi (GMR-Gal4 > UAS-lacZ; UAS-Rbfox1^RNAi), and (viii) SCA1; Rbfox1^RNAi (GMR-Gal4 > UAS-Rbfox1^RNAi, UAS-ATXN1^82Q). Points in graphs indicate measurements per eye. Over expression = OE.

Table 2

Transcriptional dysregulation of splicing factors in SCA1 B05 mouse cerebellum.

Gene/protein	5-week DEG	5-week mAS	12-week DEG	12-week mAS	20-week DEG	20-week mAS	52-week DEG	52-week mAS	Atxn1 physical interactor
Khdrbs3	0.474	-	-	-	-	-	-	-	-
Rbfox1	-	-	−0.394	-	−0.288	MXE	−0.531	-	Yes
Ptbp1	-	-	-	-	0.449	-	0.229	-	-
Srsf7	-	-	-	-	0.270	-	-	-	-
Srrm4	-	-	-	-	0.353	-	−0.206	-	-
Jmjd6	-	-	-	-	0.354	-	-	-	-
Srsf5	-	-	-	-	-	RI	0.355	-	-
Rbfox3	-	-	-	-	-	-	−0.256	A3SS	-
Srsf2	-	-	-	-	-	-	0.266	-	-
Prpf40a	-	-	-	-	-	-	0.202	-	-
Prmt7	-	-	-	-	-	-	−0.267	-	-
Tardbp	-	A5SS	-	-	-	-	-	RI; A3SS	-
Sf3b1	-	-	-	-	-	SE	-	-	-
Mbnl1	-	-	-	-	-	SE	-	-	-
U2af1	-	-	-	-	-	SE; MXE	-	-	-
Ddx5	-	-	-	-	-	RI	-	-	-
Rbfox2	-	-	-	-	-	-	-	SE	Yes

Splicing factors that were transcriptionally affected in SCA1 B05 mouse cerebellum across time points are shown. DEG data represents the log2 fold change rounded to the nearest third decimal place. Data of mAS events is presented as what mAS event occurred. Splicing factors that are known to be physical interactors with ataxin-1 are noted. Differentially Expressed Genes = DEG, misregulated alternative splicing = mAS, Skipped Exons = SE, Alternative 5′ Splice Site = A5SS, Alternative 3′ Splice Site = A3SS, Mutually Exclusive Exons = MXE, Retained Introns = RI.

To test the effect of Rbfox1 function toward SCA1 pathology in vivo, we used a Drosophila model of SCA1 in which targeted expression of mutant ataxin-1 causes retinal degeneration in the eye [48]. Light microscopy images of SCA1 flies show retinal degeneration with a loss of red pigment and a yellow-colored eye as well as necrotic patches on the eye marked by arrows (Fig. 6E and F, Supplementary Fig. 7). Co-overexpression of RBFOX1 with SCA1 caused a more severe retinal degeneration with loss of eye pigment (P = 0.0292) and necrotic patches (P < 0.0001) than SCA1 alone, while RBFOX1 overexpression itself did not cause a loss of red pigment but caused some necrotic patches to occur (Fig. 6E and G). In contrast, the reduction of Drosophila  Rbfox1 expression in SCA1 flies caused a significant rescue of retinal degeneration phenotypes of SCA1 with improved eye pigmentation (P < 0.0001) and reduced necrotic spots (P = 0.0018) (Fig. 6F and H). Together, these genetic interaction studies suggest that Rbfox1 and its splicing targets can modulate the pathogenesis of SCA1 in vivo.

Discussion

Previous studies have shown that mutant ataxin-1 physically interacts with diverse transcription factors and splicing factors [9–15]. Although many studies demonstrate that SCA1 leads to dysregulation of gene expression [7, 26, 49–51], it has not yet been shown whether it affects splicing events. In this study, we show that mAS events begin to occur as early as 5-weeks of age in the SCA1 mouse cerebellum and exaggerate throughout disease progression, which we believe contributes to altered regulation of gene expression and pathology in SCA1 mouse cerebellum.

Cell typing studies using SCA1 B05 RNA-seq data identified up-regulated DEGs in a range of neuronal and glial cerebellar cell types; however, a higher degree of down-regulated genes relative to up-regulated genes was specific to PCs [51]. This cell type-specific increase in down-regulation makes sense as ataxin-1 is proposed to be primarily a transcriptional repressor [9]. The universal up-regulation of DEGs across multiple cell types in the cerebellum could be a secondary response to the PC pathology of the B05 model. It is worth mentioning that a majority of genes with differential splicing events identified in SCA1 B05 mouse cerebellum did not overlap with DEGs, and this was reflected in the GO pathway analysis of mAS events and DEGs. Analysis of the top 10 GO pathways between mAS events and DEGs were vastly different with the exceptions of synapse organization at 12-weeks of age and regulation of membrane potential at 52-weeks of age. Together, this data highlights the importance of the identification and role of genes with altered splicing events in the pathogenesis of SCA1.

In an effort to further understand how these mAS events may be pathologically important in SCA1, we identified three genes (Trpc3, Cacna1g, and Snap25) that are mAS events in all time points in the SCA1 B05 model as well as in the SCA1 KI model. Of those genes, Trpc3 was of particular interest to us because it is a SCA-causative gene for SCA41 [39] and previously literature has examined the molecular differences between its two isoforms [40, 41]. The mAS event in the SCA1 B05 model creates a less active Trpc3 isoform with exon 9, thereby creating a significant impact on its functionality. It has been shown that in Trpc3^−/− mice the mGluR1-mediated slow excitatory postsynaptic currents (EPSCs) at the parallel fiber to PC synapses was absent [52]. Analysis of slow EPSCs in SCA1 B05 mice was significantly reduced by 5-weeks and 12-weeks of age but not at 3-weeks of age [53]. Down-regulation of Trpc3 alone may not account for this marked decrease in EPSC activity, and further understanding the role of mAS events in Trpc3 and other ion channel genes will be important to understand their contribution in disease pathogenesis.

SCAs are a heterogenous group of disorders caused by seemingly unrelated genes, yet display similar clinical phenotypes such as gait imbalance and ataxia. Our analysis revealed that Trpc3 was not the only SCA-causative gene that had mAS events. Cacna1g and Itpr1 are also ion channel genes as well as mAS events at all time points. Cacna1g is reported as a causative gene in SCA42 [38, 54] and Itpr1 is SCA-causative for SCA15 [55, 56] and SCA29 [57]. These findings show that SCA1 can regulate the expression of other SCA-causative genes. Future studies investigating the significance of splicing dysregulation in disease pathogenesis and shared clinical features among SCAs are of high interest.

We report that 17 splicing factors are transcriptionally affected in SCA1. These splicing factors may have their expression level changed as shown through DEG analysis or their activity level may be affected by mAS events. Among these splicing factors, Rbfox1 stood out due to its down-regulation at multiple time points and its known physical interaction with ataxin-1 [11]. Although Rbfox1 is also a mutually exclusive exon mAS event at 20-weeks of age, the inclusion level is less than 10% (Supplementary Table 4). Therefore, we suspect that its down-regulation across multiple time points in SCA1 might be the major contributor for the mAS events in SCA1 B05 cerebellum. Importantly, alteration of Rbfox1 expression levels modulates SCA1 neurodegenerative phenotypes in a Drosophila model by showing that overexpression of RBFOX1 worsened phenotypes, while reduction of Rbfox1 was able to rescue degeneration phenotypes. This suggests that Rbfox1 down-regulation in the cerebellum of SCA1 B05 mice might be a potential protective mechanism. However, previous literature has shown that complete loss of Rbfox1/2 proteins in the mouse cerebellum causes highly irregular firing of PCs and motor loss at 10-weeks of age, suggesting that they are essential for PC health [58]. It is worth mentioning that there are opposite effects of the same protein in SCA1 pathology in two different species of SCA1 models. For example, loss or decreased expression of the transcription repressor Capicua ameliorates the SCA1-associated neuropathological phenotypes in mice [59], while decreased expression of Capicua worsens mutant ataxin-1-induced eye phenotypes in Drosophila [9]. Therefore, the precise role (e.g. protective, pathogenic, or both) of Rbfox1 in the pathogenesis of SCA1 in mammalian systems is required in further studies. Interestingly, Rbfox1 has also been reported to interact with ataxin-2 [11, 60] and furthermore, Rbfox1 splicing of Atxn2 is modulated by ataxin-1 [61]. Consistently, in our dataset, mAS of Atxn2 occurs at 20-weeks of age, potentially indicating ataxin-1/Rbfox1 as a regulator of other SCA-causing genes. Once again, future studies are warranted for the functional impact and therapeutic potential of Rbfox1 and other splicing factors affected in SCA1.

Finally, disease pathogenesis is not restricted to the cerebellum in SCA1 as there are multiple brain regions that are affected in SCA1 human and mouse models, including the brainstem and cortex [7, 26, 27, 62–67]. Correspondingly, there are vast transcriptional changes reported in these brain regions of SCA1 mouse models [7, 26, 49]. However, it has not been investigated yet whether and how alternative RNA splicing might be dysregulated in these SCA1-affected brain regions beyond cerebellum. Therefore, it is worth investigating whether these potential mAS events can causally contribute to the pathogenesis in broad brain regions of SCA1. In addition, future experiments need to investigate whether RNA splicing dysregulation occurs in brain regions affected by other SCAs and/or polyQ diseases, and their contribution to corresponding disease pathogenesis. Overall, our studies highlight the significance of mAS events in SCA1 pathology, underscoring novel pathways and targets of interest for potential therapeutic intervention, and highlights the need for future RNA splicing analysis in other neurodegenerative diseases.

Acknowledgements

We would like to thank all members of the Lim laboratory for providing valuable comments and feedback. Figure 1 was created with BioRender. We also thank the Bloomington Drosophila Stock Center for the Drosophila strains and the Yale Center for Genome Analysis for performing the RNA sequencing experiments.

Conflict of interest statement

The authors report no competing interests.

Funding

This work was supported by National Institute of Health grants NS083706 (J.L.), AG076154 (J.L.), AG074609 (J.L.), AG066447 (J.L.), T32 NS007224 (K.L.), and the Lo Graduate Fellowship for Excellence in Stem Cell Research (K.L.).

Data availability

Raw data was generated at Yale University. Data supporting the findings of the study are available from the corresponding author on reasonable request. RNA sequencing data can be accessed at GEO accession # (GSE232368).

Author notes

Victor Olmos and Evrett N. Thompson contributed equally.