Spinocerebellar ataxia type 3 (SCA3) is a progressive, typically late-onset autosomal dominant neurodegenerative disease [1
]. SCA3 is one of a larger group of diseases, termed, the polyglutamine (polyQ) diseases [2
]. These diseases all share a common pathogenic mechanism; an expanded CAG repeat in the coding sequence of nine genes, and in the case of SCA3, the CAG expansion is located in the penultimate exon (exon 10) of ATXN3
]. Healthy individuals have a stable repeat range of 7–44, while SCA3 patients usually have 54 or more repeats. SCA3 is known to have an unstable pre-mutation range of 45–53 repeats, and while these individuals are typically asymptomatic, they have the ability to pass on an expanded allele in what is known as ‘genetic anticipation’. As with other polyQ diseases, the pathogenic severity and age of onset is typically inversely correlated to the size of the expansion: the larger the expansion, the more severe the pathogenesis and the earlier the age of onset [5
]. The ATXN3
encodes for a 361 amino acid (aa), 45 kDa protein (ENST00000558190.6), termed ataxin-3. The ataxin-3 protein is known to act as an isopeptidase and is well documented in cell deubiquitination, as well as proteasomal protein degradation [2
The expanded CAG repeat located in exon 10 of ATXN3
results in the addition of an extended glutamine tract in ataxin-3, directly leading to conformational changes that give the protein a toxic gain of function(s), as well as subjecting the protein to formation of neuronal nuclear inclusions [7
]. Although SCA3 is clinically heterogeneous in presentation, the main feature is progressive ataxia, which in turn affects speech, balance and gait of the affected individual [3
]. Despite arising from a single variant gene, the pathogenesis of SCA3 has been difficult to characterize, as several toxic pathways and mechanisms have been proposed to play a role in the disease.
Several studies that use antisense oligonucleotides (AOs) to modify the mRNA of ATXN3
by attempting to remove the CAG containing exon have been conducted [8
]. Until now, van Roon-Mom and colleagues have published two reports detailing the removal of the CAG containing exon in the ATXN3
]. These studies show removal of the CAG containing exon, and production of a functional truncated protein using a modified 2′-O
-methoxy-ethyl nucleotide (2′-MOE) on a phosphorothioate (PS) backbone. Growing concerns regarding safety of PS oligonucleotides, such as hepatotoxicity, liver necrosis and altered regulation of protein and metabolic pathways [11
] and thus, the consequences and safety of long-term exposure to AOs on a PS backbone [11
] may limit applicability of these compounds. Phosphorothioate backbone AOs have resulted in severe injection site reactions, thrombocytopenia and renal toxicity in clinical studies [16
]. An in vitro study by Flynn et al. (2018), demonstrated severe, sequence-independent backbone-specific effects of 2′-O
-methyl modified bases on a phosphorothioate backbone (2′-Me PS) AOs, including altered distribution of nuclear proteins, the appearance of abnormal but highly structured nuclear inclusions and aggregates and global disturbance of the transcriptome [18
An AO chemistry currently in clinical use is the phosphorodiamidate morpholino oligomer (PMO) [19
]. This chemistry has a backbone of methylene morpholine rings, with phosphorodiamidate group linkages and an oligomer of this chemistry has been granted accelerated approval by the FDA. Eteplirsen
(Sarepta Therapeutics, Ma) is designed to excise dystrophin exon 51 during pre-mRNA processing to restore functional protein expression in a subset of boys with Duchenne muscular dystrophy (DMD). The drug restored modest dystrophin expression in patient muscle where previously no, or only traces of dystrophin were evident [20
]. The PMO chemistry is reported to have excellent biological stability, and to be safe and well tolerated, with no serious adverse effects reported in the Eteplirsen
treated children and young men to date [21
Here, we describe efficient removal of the CAG containing exon 10 to produce a truncated ataxin-3 protein, lacking the polyglutamine tract, an isoform reported by Toonen et al. (2017) to be functionally active [8
]. Our study shows that by using the PMO chemistry, not only is exon 10 skipping enhanced at the RNA level, but also significant downregulation of the protein with higher number of glutamine repeats and an increase in production of the truncated protein is observed, when compared to the use of the 2′-Me PS AO chemistry. With robust splice switching efficiency and an established long-term safety profile, the PMO oligomers described here are presented as lead pre-clinical candidates to treat SCA3 patients.
The pathogenesis of SCA3 is attributed to the expanded polyQ tract that confers a toxic gain of function to the protein [4
]. Therefore, the removal of the polyQ tract may provide a therapeutic strategy to delay onset or reduce severity of SCA3. As a consequence of exon 10 removal, the induced ataxin-3 isoform is missing the UIM3 domain. As reported by van Roon-Mom and colleagues, the 291aa, 34kDa protein, although missing the UIM3 domain, still binds ubiquitin in a similar manner to the full-length ataxin-3 [8
]. Additionally, two naturally occurring isoforms lacking UIM3 have been shown to actually have higher rates of deubiquitination, relative to the major isoform containing the UIM3 [29
]. With recent FDA approvals of AO therapeutics to restore gene expression to treat spinal muscular atrophy and DMD, pre-mRNA splicing intervention could be applied to downregulate or modify expression of toxic-gain-of-function diseases. This proof of concept study may be relevant to other polyQ diseases where the polyQ tract is found in a removable and dispensable exon. However, this would need to be the subject of functional studies to determine the effect of removing regions of other polyQ proteins. For example, Huntington’s disease is unlikely to be amenable to exon skipping, as the polyQ tract is located in the initial exon [3
]. Here, we show removal of ATXN3
exon 10, and consequently the polyQ repeat from the normal and expanded ataxin-3 protein, using two different AO chemistries. Additionally, these in vitro experiments demonstrate proof of concept and can provide pre-clinical candidate molecules for eventual in vivo administration. Dosage regimens, clearance rates and delivery methods will need to be optimised to assess the effects of PMOs compared to other chemistries.
We confirm apparent off-target effects of AOs on a PS backbone, transfected in vitro. The 2′-Me PS AO targeting ATXN3
induced sequestration of the paraspeckle protein NONO, while no such consequences were observed when cells were treated with the same sequences synthesised as PMOs, in keeping with our previously reported findings and those of the Crooke group [12
]. While these studies were only conducted in vitro, van Roon-Mom and colleagues described activation of the innate immune system as the result of intracerebroventricular administration of 2′-Me AOs into mice [15
]. Marked upregulation of Oasl2 and Bst2 proteins, both of which are involved in interferon signalling, particularly in response to viral infections [30
], was a major finding [15
]. Several other studies have described adverse effects of AOs on a PS backbone, including thrombocytopenia, severe injection site reaction, cytotoxicity and induction of double-stranded DNA breaks [33
The safety of PMOs in the clinic is evident from long term treatment (240 weeks) of DMD patients with the PMO Eteplirsen
. No evidence of serious adverse reactions were reported [33
], while the same cannot be said for the 2′-Me PS AO drug Drisapersen.
Every drug on a the PS backbone tested to date has been reported to elicit off-target effects [33
]. This may be explained by the crucial difference that PMOs, unlike 2′-Me PS AOs, are uncharged and therefore do not readily interact with proteins [38
]. We, and others have shown that PMOs generally outperform 2′-Me PS AOs for splice switching applications [37
], possibly due to the superior stability, specificity and binding affinity of PMOs, compared to 2′-Me PS AOs [38
]. Although PMO-treated cells were transfected at higher concentrations than the 2′-Me PS AOs that are transfected as lipoplexes, these high concentrations are required for gymnotic PMO uptake but were nevertheless well tolerated by cells. Although the delivery of 2′-Me PS AOs is greatly enhanced through the use of cationic liposome preparations, concentrations of 600 nM and above lead to widespread cell death in vitro.
Although PMOs outperformed 2′-Me PS AOs in vitro, the PMOs do have several limitations. When administered in vivo, PMOs are rapidly cleared by the renal system and must therefore be dosed at relatively high levels. This is compounded by poor uptake and delivery of PMOs into target tissues [42
]; due to their uncharged nature PMOs cannot readily move through cellular membranes [43
]. These limitations are currently subject to extensive research in developing technologies, such as conjugation to cell penetrating peptides and various physical and chemical methods to increase uptake and reduce the rapid clearance [42
]. Despite these draw backs, from our in vitro data we believe PMOs to have greater therapeutic potential for SCA3 and other diseases, such as muscular disorders that are amenable to splice switching [45
]. In saying that, it is important to note that in vitro studies and in vivo animal models do not always translate into successful treatments for patients. This is mainly due to sequence differences in the cases of in vivo models, with a prime example being a mouse model of Duchenne muscular dystrophy, whereby subtle changes in sequences can drastically affect AO efficiency [46
We show significant knockdown of the full-length ataxin-3 isoforms encoded by both alleles, as well as modification of the ataxin-3 protein to produce a truncated protein, missing the polyQ tract. While the impact of global ataxin-3 knockdown is under debate [8
], the role of ataxin-3 in the ubiquitin-proteasome machinery is well-established, and there are conflicting views as to whether ataxin-3 is essential in maintaining normal cellular function [3
]. Figiel and colleagues (2011) created a functional Atxn3
knockout mouse that showed no obvious phenotype, with a life span comparable to that of the wildtype mouse [48
]. Another group reported similar findings, with no adverse effects on Atxn3
knockout mouse life span or fertility and no apparent abnormalities, but they did report apparent increased anxiety and increased levels of ubiquitinated proteins in the Atxn3
knockout model [51
]. Separately, AO treated transgenic SCA3 mice (expressing a human full-length, 84Q ATXN3
gene), showed that knockdown of ataxin-3 was well tolerated with no signs of astrogliosis or microgliosis [52
]. However, our data show a high proportion of the truncated protein as a consequence of ATXN3
exon 10 skipping, and thus this may provide sufficient functional protein to support ubiquitination and protein degradation [53
]. With that being said, the Paulson group conducted in vitro experiments to assess the effects of ATXN3 knockout using Atxn3
null mouse embryonic stem cells. They found that the loss of Atxn3 caused dysregulation in signalling pathways that included depression of Wnt and BMP4 pathways, as well as elevated growth factor pathways [54
]. In contrast, the same group showed that knockdown of an expanded ATXN3
in a transgenic mouse model (MJD-Q84.2) using a 2′-MOE AO rescued the phenotype with no apparent adverse effects, thus suggesting in vivo treatment and knockdown of ATXN3 may be feasible [55
]. Moreover, in the current study it is believed that the removal of the CAG repeat alone without removal of key functional domains would result in limited downstream effects. Further studies investigating the long-term impact of ataxin-3 knockdown in vivo will be required to determine if skipping of exon 10 has potential as a treatment for SCA3.
In conclusion, while this study was a preliminary in vitro investigation, PMOs consistently produced a significantly higher proportion of the truncated protein, missing the toxic polyQ repeat, relative to 2′-Me PS AOs. With increasing numbers of AO therapeutics being approved for clinical use, our results suggest that the lead PMOs may be an attractive therapeutic option for the treatment of spinocerebellar ataxia type 3.
4. Materials and Methods
4.1. AO Design and Synthesis
Splice-switching AOs were designed to target and anneal to splicing motifs at the intron/exon boundaries as well as predicted exon splice enhancer sequences identified using the web-based application, Human Splicing Finder 3.0
]. In addition, specificity of the AOs for the ATXN3
target motifs was confirmed via BLAST analysis to identify potential off-target annealing. 2′-Me PS AOs were obtained from TriLink Biotechnologies (Maravai LifeSciences, San Diego, CA, USA), while PMOs were purchased from Gene Tools, LLC (Philomath, OR, USA). Nomenclature of AOs is according to Aung-Htut, McIntosh et al. (2019) and indicates gene, exonic target with annealing coordinates relative to the intron:exon:intron arrangement (Table 1
and Table 2
4.2. Cell Culture
Primary dermal fibroblasts were cultured from a skin biopsy taken from a healthy volunteer, after informed consent, and the project received approval from the Human Research Ethics Committee at Murdoch University (approval number, 2013/156). Healthy human fibroblasts were cultured in Dulbecco’s Modified Essential Medium (Gibco; Life Technologies, Melbourne, Australia), supplemented with 15% fetal bovine serum (FBS) (Scientifix, Cheltenham, Australia). The SCA3 fibroblast cell line (GM6151b) was obtained from Coriell Cell Repositories (Camden, NJ, USA) and cultured in minimal essential medium (MEM), supplemented with 15% FBS (Scientifix), 1% Glutamax (Gibco) and 1x penicillin/streptomycin.
All cell strains were either transfected with 2′-Me PS AO, Lipofectamine 3000 (Life Technologies) lipoplexes in Opti-MEM (Gibco) according to manufacturer’s instructions, or with PMOs, using Endo-Porter (Gene Tools, LLC) according to manufacturer’s instructions in MEM (Gibco), supplemented with 7.5% FBS (Scientifix) [58
]. Cells were harvested 24 h following transfection for transcript analysis or after 48 h for protein studies.
4.4. RNA Extraction and RT-PCR Assays
Total RNA was extracted using the MagMax™ nucleic acid isolation kits (ThermoFisher Scientific, Melbourne Australia) in accordance with the manufacturer’s instructions. Transcripts were amplified using the one-step SuperScript® III reverse transcriptase, with 50 ng of total RNA as the template. To amplify the ATXN3 transcript, exon 7-F (5′GTCCAACAGATGCATCGACCAA3′) and exon 11-R (5′AGCTGCCTGAAGCATGTCTTCTT3′) primers were used (Gene Works, Adelaide, Australia). The cycling reactions included 55 °C for 30 min, 94 °C for 2 min, with 28 cycles of 94 °C 30s, 55 °C 30s and 68 °C 1.5 min. The PCR products were fractionated on 2% agarose gels in Tris-Acetate-EDTA buffer.
4.5. Western Blotting
Cell lysates (~800,000 cells) were prepared in 100 μL of 125 mM Tris-HCl, pH 6.8, 15% SDS, 10% Glycerol, 1.25 µM PMSF (Sigma-Aldrich, Sydney, Australia) and 1 × protease inhibitor cocktail (Sigma-Aldrich) and subsequently sonicated 6 times (1 s pulses) prior to the addition of bromophenol blue (0.004%) and dithiothreitol (2.5 mM). Samples were heated at 94 °C for 5 min, cooled on ice and centrifuged at 14,000× g for 2 min before loading onto the gel.
Total protein (25 µg), determined by a BCA assay (ThermoFisher Scientific), was loaded onto NuPAGE Novex 4–12% Bis/Tris gradient gels (ThermoFisher Scientific). Samples were subsequently fractioned at 200 volts for one hour. Proteins were then transferred onto a Pall Fluoro Trans® polyvinylidene fluoride membrane at 350 mA for one hour. Following blocking in 5% skim milk in TSBT for one hour, the membrane was incubated with either mouse monoclonal anti-ataxin-3 antibody (Millipore, cat. No. MAB5360, Billerica, USA) at 1:500 dilution or rabbit polyclonal anti-β-tubulin antibody (ThermoFisher Scientific, cat. No. PA1-41331) at 1:1000 dilution in 5% skim milk in TSBT, overnight at 4 °C. For detection of pathogenic polyQ stretches, the membrane was probed with mouse monoclonal anti-polyQ antibody (Millipore, cat. no. MAB1574, Billerica, USA), at 1:1000 dilution in 5% skim milk in TSBT, overnight at 4 °C.
For immunodetection, polyclonal goat anti-rabbit or anti-mouse immunoglobulins/HRP (Dako, cat. no P0448 and D0447 respectively, Sydney, Australia) at a dilution of 1:10,000 and Luminata Crescendo Western HRP substrate (Merk Millipore, Sydney, Australia) were used. The blots were exposed for a serial scan of 20 s using the Fusion FX gel documentation system (Vilber Lourmat, Marne-la-Vallée, France).
Approximately 7500 patient fibroblasts were seeded into each well of an 8 well chamber slide (Ibidi, Martinsried, Germany) and incubated for 24 h, prior to transfection. Following transfection, with 2′-Me PS AOs or PMOs, cells were fixed in ice-cold acetone:methanol (1:1) for 5 min and then air dried.
Fixed cells were incubated in PBS containing 1% Triton X-100 for 10 min at room temperature to permeabilise the nuclear membrane, and then in PBS to remove excess Triton X-100. Mouse anti-NONO monoclonal antibody (a gift from Prof. Archa Fox, The University of Western Australia) was diluted in PBS containing 0.05% Tween20 and applied to cells for one hour at room temperature. NONO was detected using AlexaFluor488 anti-mouse IgG (ThermoFisher Scientific, cat no. A-11001) (1:400) after incubation for one hour at room temperature, and subsequently counterstained with Hoechst 33342 (Sigma-Aldrich) for nuclei detection (1 mg/mL diluted, 1:125).
4.7. Densitometric and Statistical Analysis
Densitometric analysis was conducted using ImageJ (version 1.8.0_112) imaging software (NIH, Bethesda, MD, USA) [59
value refers to unpaired two tailed student’s t tests. A p
value < 0.05 was considered statistically significant.