Repeat expansion disease: Progress and puzzles in disease pathogenesis

A shared mechanism for DM1 and DM2

One of the most striking developments in the repeat disease field has been the realization that expanded repeats in transcripts can cause cellular toxicity and neurodegeneration by altering the splicing machinery. In 1992, a CTG repeat expansion in the 3’ UTR of a protein kinase gene was found to be the cause of myotonic dystrophy type 1 (DM1), the most common form of adult muscular dystrophy ^6–8. This discovery posed a puzzling question: How does a dominantly inherited repeat expansion in the non-coding region of a gene produce a disease phenotype that affects many different tissues? Although some evidence emerged for reduced dosage of the DM1 protein kinase as the cause of certain DM1 features, haploinsufficiency could not account for most facets of DM1 pathophysiology ⁹. As the DM1 gene is located in a gene-rich region, some argued that DM1 resulted from altered expression of the DM1 protein kinase in combination with altered expression of adjacent genes, and proposed that the CTG repeat expansion mimicked a contiguous gene syndrome ¹⁰. This “field theory” of DM1 pathogenesis gained support when knock-out mice lacking the gene downstream of DM1, SIX5, were found to develop cataracts, which are also a feature of DM1 ^11,12.

However, the field theory of DM1 pathogenesis could not easily be reconciled with a DM1 phenocopy caused by a different genetic locus ¹³. This disorder, known as myotonic dystrophy type 2 (DM2), has a phenotype similar to DM1. A distinction is occasional congenital presentation and mental retardation in DM1 patients who receive large CTG repeat expansions because of maternal anticipation ^14,15. DM2 is caused by expansion of a CCTG tetranucleotide repeat in the zinc-finger protein 9 (ZNF9) gene ¹⁶, the normal function of which does not match any of the genes at the DM1 locus. Hence, the parsimonious conclusion was that RNA transcripts containing a CUG repeat expansion – whether in a triplet repeat – or a CCUG repeat – initiate a shared pathogenic cascade.

A role for RNA toxicity in DM1 had been first suggested by RNA foci in cells from patients with DM1 ¹⁷, and indistinguishable RNA foci were found in samples from patients with DM2 ¹⁶. The validity of the RNA toxicity model was strongly supported by a mouse model of myotonic dystrophy in which a 250 CTG repeat in the 3’ UTR of an unrelated transgene - skeletal actin -was shown to cause myotonia and myopathy ¹⁸. In another study, preceding the identification of DM2, altered splicing owing to increased function of the CUG-binding protein (CUG-BP) emerged as a plausible mechanism for the RNA gain-of-function toxicity ¹⁹. Subsequent work has implicated reduced function of muscleblind 1 (Mbnl1), owing to its sequestration into CUG repeat rich foci, as a contributing factor in the splicing abnormalities in DM1 and DM2 ²⁰. Mbnl1 is one of a family of three proteins (Mbnl1, Mbnl2, and Mbnl3) that have been shown to bind specifically to double-stranded RNA hairpins formed by CUG repeats ²¹. Although increased CUG-repeat containing RNAs appear to increase CUG-BP levels, the production of CUG and CCUG ribonuclear foci by expanded CUG-repeat containing RNAs results in Mbnl1 sequestration and depleted function. Therefore, the model is of increased CUG-BP function combined with decreased Mbnl1 function. Studies of the Drosophila melanogaster muscleblind mutant show that loss-of-function of Mbnl prevents terminal differentiation of retinal and muscle cells ^22,23. In mammals, Mbnl1 serves an analogous function, as it favors the splicing of target genes into adult isoforms so, in DM1 and DM2, reduced Mbnl1 function allows fetal isoform production to persist in adult tissues, and affects the expression level of many target transcripts ²⁴. A very recent study found that loss of Mbnl1 can explain the majority of splicing alterations in DM1, and identified extracellular matrix genes as a common target of expanded CUG-repeat containing RNAs, linking DM1 pathogenesis with other connective tissue diseases and muscular dystrophies ²⁴. The pleiotropic phenotype of the myotonic dystrophies is thus believed to result from altered splicing of genes whose protein products function in pathways that are linked to disease features. For example, altered splicing of the chloride channel gene and insulin receptor genes are linked to myotonia and glucose intolerance, respectively.

FXTAS: two mechanisms at the FMR1 locus

Advances in our understanding of myotonic dystrophy pathogenesis set the stage for the characterization of other RNA gain-of-function repeat diseases. One of the most intriguing of these has been the fragile X-tremor ataxia syndrome (FXTAS) ^25,26. This disorder occurs primarily in male carriers of the fragile X syndrome premutation allele, and appears to result from an entirely different molecular mechanism than fragile X syndrome. CGG expansions exceeding ~200 repeats produce a mental retardation phenotype by reducing expression of the FMR1 gene, whereas CGG expansions of 55 – 200 repeats result in higher expression of a transcript containing the CGG tracts. The RNA molecules with the expanded CGG tract initiate a cascade of events that culminate in central nervous system (CNS) neurodegeneration, which is characterized by ubiquitin-positive inclusions in the nuclei of neurons and glia ²⁷. Studies in D. melanogaster showed that CGG repeat expansions are sufficient to produce neurodegeneration ²⁸, and features of FXTAS histopathology were also recapitulated in a knock-in mouse model ²⁹. In light of the CUG-BP model of myotonic dystrophy, researchers investigated the function of CGG RNA binding proteins. Through a combination of biochemical and genetic approaches, mainly in the fly, three proteins were found that bind CGG repeats and have reduced function in the disease models: Pur α, hnRNP A2 / B1, and CUGBP1 ^30,31. The extensive work on myotonic dystrophy and FXTAS, which combined molecular approaches with model organism studies and proteomics, has emphasized that mutant transcripts can produce neuronal dysfunction by disturbing the balance and availability of RNA-binding proteins. Altered function of these proteins seems to be the crux of the molecular pathology in these diseases (Figure 1).

Toxic RNAs in CAG/polyglutamine disease?

If uninterrupted CUG repeat expansions in RNA can produce neurotoxicity, is it possible that other types of repeat expansions also exert their toxic effects at the RNA level? In the case of the CAG / polyglutamine repeat diseases (Table 1), extensive work has shown that protein toxicity rather than RNA toxicity is principally responsible for the disease. Evidence against CAG-RNA toxicity include studies of SCA1 and SBMA transgenic mice in which production of mutant protein and mutant RNA did not yield a neurodegenerative phenotype when the mutant protein could not enter the nucleus, (either because of mutation in the nuclear localization signal or lack of ligand binding)^30–33. Furthermore, in D. melanogaster no evidence for general CAG RNA toxicity was found³⁴. However, the recent identification of muscleblind as a modifier of disease toxicity in the SCA3 fly model prompted Li et al. to reexamine this question and, contrary to earlier work, they demonstrated that untranslated RNAs containing tracts of 100 or 250 CAGs can cause retinal degeneration and neuronal dysfunction in D. melanogaster, although interrupted CAGCAA tracts were non-pathogenic ³⁵. These researchers did not find evidence for altered splicing involving muscleblind, so the basis of CAG-RNA toxicity and of the modifier effect remains unclear. Therefore, whether CAG-RNA toxicity contributes to any, or all, of the nine canonical polyglutamine repeat diseases remains highly controversial. At the same time, considerable progress has been made in understanding the mechanistic basis of polyglutamine proteotoxicity, and sophisticated models of disease pathogenesis, involving protein modification and interaction, have now been proposed.

Neuroprotective Autophagy

Autophagy might provide neuroprotection through accelerated turnover of misfolded disease proteins. Indeed, it has been shown that poly(A) binding protein 2 with an expanded alanine tract (the disease protein in oculopharyngeal muscular dystrophy) and at least some polyglutamine-expanded proteins are delivered to autophagic vacuoles^38,47 and degraded by autophagy in vitro^48,49. Further support for this view comes from studies in Drosophila models of SBMA and HD, in which genetic ablation of autophagy leads to greater accumulation of polyglutamine disease proteins and increased neurodegeneration^50,51 Moreover, pharmacological augmentation of autophagy using small molecules such as rapamycin, lithium, or trehalose, or genetic augmentation by over-expression of HDAC6, results in accelerated turnover of expanded polyglutamine protein and reduced neurodegeneration in Drosophila and mouse models of polyglutamine disease^{52,53,54,55,56–58}.

A caveat to these studies is that virtually all have been performed in models based on over-expression of exogenous mutant protein, which is precisely the scenario in which autophagy would be expected to have greatest effect. It remains important to examine the role of autophagy further in animal models that more faithfully recapitulate endogenous protein expression levels and patterns, using knock-in or BAC transgenic approaches, for example. Nevertheless, the view of autophagy as a neuroprotective process is consistent with the emerging appreciation that autophagy is cytoprotective in numerous contexts - such as conditions of oxidative stress, growth factor deficiency or nutrient limitation - through accelerated turnover of damaged organelles and maintenance of metabolic homeostasis by mobilization of intracellular energy stores^59,60.

Is autophagy a disease target?

Despite compelling evidence that autophagy affords neuroprotection by degrading disease proteins, the question remains: why is this protection incomplete? Is it possible that autophagy is not only a modifier of disease but also a target of disease? It has been suggested that the increased numbers of neuronal autophagic vacuoles in some repeat expansion diseases may reflect a defect in autophagy flux rather than autophagy induction⁶¹. It is now appreciated that there is significant basal autophagy in many mammalian tissues. The demand for basal autophagy differs among tissues; it is particularly important in the liver and in post-mitotic cells such as neurons and myocytes^62,63,64,65. Indeed, neurons are especially vulnerable to perturbations in the autophagy-lysosomal system – not only because they are highly metabolically active, but also because of their unique cellular architecture. In neurons, lysosomes are concentrated in the soma adjacent to the nucleus. Therefore, autophagosomes produced in dendrites, axons or synaptic terminal regions must be transported substantial distances to enable fusion with lysosomes, which makes autophagy in neurons particularly vulnerable to defects in vesicular trafficking⁶⁶. The importance of basal autophagy in the CNS was shown by conditional knockout of key autophagy genes, which resulted in neurodegeneration with accumulation of ubiquitin-positive inclusions similar to that seen human neurodegenerative diseases (including many repeat expansion diseases) ^67,68. The increased vulnerability of neurons to impairment of the autophagy-lysosomal system might account for the high frequency of neurological phenotypes produced by mutations that target the endosomal-lysosomal system⁶⁹.

It remains to be shown convincingly whether autophagy is impaired in repeat expansion disease. Given the broad range of cellular abnormalities associated with polyglutamine diseases and other repeat expansion diseases, it would not be surprising to find impairment of autophagy. Impaired autophagy - a sort of ‘cellular indigestion’ – could result from overburdening the autophagy-lysosomal system with misfolded, aggregated protein that is difficult to degrade. Alternatively, autophagy could be directly impaired if a disease protein is important for the process of autophagy. As noted above, huntingtin associates with components of the endosomal-lysosomal system and is trafficked with autophagosomes in axons, which suggests that huntingtin might normally regulate autophagy^{8, 70}. Recently, it was shown in vitro and in a mouse knock-in model of HD that polyglutamine expansion in huntingtin impairs the activity of the GTPase Rab11 and leads to a defect in endosome recycling ^71,72. Thus, is it possible that polyglutamine expansion impairs a normal function of huntingtin and contributes to impaired autophagy? This would be consistent with the emerging theme that altered native protein function underlies polyglutamine disease pathogenesis.

Non-polyglutamine determinants of disease

A striking feature of polyglutamine diseases is the selective vulnerability of the CNS despite widespread expression of many polyglutamine disease proteins in non-neural cell types. Yet, there is striking divergence in clinical phenotype among the polyglutamine diseases; neurologists can easily distinguish the movement disorder of HD from the weakness in SBMA or the ataxia in SCA1. The disease-specific features reflect selective loss of different populations of neurons: despite wide expression within the CNS, polyglutamine expansion in huntingtin selectively affects striatal neurons and cortical neurons, whereas the same genetic mutation in the androgen receptor or ataxin-1 targets motor neurons and Purkinje neurons, respectively. On the basis of these observations it was predicted that features other than polyglutamine, unique to each disease protein, must influence pathogenesis⁷³. This prediction has been borne out in recent advances that have highlighted the importance of host protein context in polyglutamine disease pathogenesis. Principal among these have been insights into the influence of post-translational modification in pathogenesis and, in at least one case, in determining cell type specificity. Here we summarize some important insights into the role of post-translational phosphorylation, acetylation, and sumoylation in polyglutamine disease pathogenesis (Table 2).

Phosphorylation of ataxin-1 maintains a balance

The importance of post-translational modification in polyglutamine disease was first illustrated by the Orr and Zoghbi laboratories in a series of papers examining ataxin-1 phosphorylation. Orr and colleagues detected polyglutamine length-dependent phosphorylation of ataxin-1 that mapped to serine 776⁷⁴. Their interest was heightened after observing that an antibody specific for phospho-S776-ataxin-1 preferentially stained pathological, nuclear-localized ataxin-1 in their SCA1 mouse model. They generated transgenic mice expressing polyglutamine-expanded ataxin-1 with an alanine substituted for serine at position 776 (S776A) to prevent phosphorylation. The mice expressing polyglutamine-expanded ataxin-1-S776A had minimal behavioral and histopathological abnormalities, illustrating the importance of this phosphorylation in pathogenesis.

Phosphorylation alters the conformation of a target protein and can profoundly influence target protein function, often by regulating protein-protein interactions⁷⁵. Ataxin-1 interacts with two discrete heterotypic protein complexes, one defined by the transcription factor Capicua and the other defined by the RNA-binding protein RBM17. Phosphorylation of ataxin-1 regulates the balance of ataxin-1 association with these distinct functional complexes. The evidence suggests that polyglutamine expansion upsets this balance and drives pathogenesis. Specifically, polyglutamine expansion promotes phosphorylation of S776, which favors association with the RBM17 complex and so might contribute to SCA1 neuropathology through a gain-of-function mechanism. Concomitantly, polyglutamine expansion attenuates the formation and function of the Capicua complex, contributing to SCA1 through a partial loss-of-function mechanism^76,77. This discovery has important implications, as this model suggests that polyglutamine disease pathogenesis might involve subtle alteration of native protein function rather than an entirely novel gain of function. This insight highlights the importance of understanding the normal interactions and functions of each disease protein to understand the pathogenesis of each polyglutamine disease.

Phosphorylation and huntingtin trafficking

Huntingtin is subject to phosphorylation at multiple serines and phosphorylation of serine 421 (S421) has emerged as a particularly important determinant of HD pathogenesis. Phosphorylation of S421 has been reported to be carried out by multiple kinases, including the serine/threonine kinase Akt and the serum / glucocorticoid-induced kinase (SGK) ^78–80. Significant levels of phospho-S421-huntingtin are present in normal human and mouse striatum, but are reduced by polyglutamine expansion. The basis of reduced phosphorylation is unclear, but is presumed to be owing to altered huntingtin conformation with polyglutamine expansion. Through the use of phosphorylation site mutants and manipulation of the relevant kinases, phosphorylation of S421 has been shown to substantially reduce toxicity in cultured striatal neurons and in a rodent model of Huntington’s disease^78–81. What is the basis for this? Huntingtin normally associates with vesicle membranes and microtubules and contributes to endocytosis and transport of endocytic vesicles^82–87. Indeed, huntingtin might mediate retrograde transport of signaling endosomes that carry brain-derived neurotrophic factor (BDNF)⁸⁷. This is crucial, since BDNF signaling supplied by cortico-striatal projection neurons to the striatum - the brain region most severely affected in HD - is an important survival factor ^88,89. Although it has been known for some time that polyglutamine expansion impairs the ability of huntingtin to support BDNF signaling, the basis for this was unclear⁸⁷. The recent discovery that the phosphorylation status of S421 modulates the recruitment of motor proteins to endocytic vesicles and is a key determinant of huntingtin regulation of vesicular transport, and is altered by polyglutamine expansion, offers an explanation. Indeed, constitutive phosphorylation of S421 can overcome the BDNF signaling defect caused by polyglutamine expansion, and can suppress toxicity in HD models^90,91. These data highlight a mechanism by altered post-translational modification owing to repeat expansion can affect the ratio of native protein-protein interactions. Therefore, as in the case of SCA1, the pathogenesis of HD appears to be mediated, at least in part, by alteration of native protein function.

Additional phosphorylation sites in huntingtin have been implicated as modifiers of disease, although less is known about how they are influenced by polyglutamine expansion or the mechanisms by they influence pathogenesis. For example, phosphorylation at S343 and S536 is associated with reduced cleavage of huntingtin, and at threonine 3 is associated with reduced aggregation of huntingtin N-terminal fragments^92–94. Notably, the phosphorylation status of S13 and S16 has recently been implicated an important determinant of HD pathogenesis. Gu and colleagues generated transgenic mice that express full-length mutant huntingtin with S13 and S16 mutated to either aspartate (which mimics constitutive phosphorylation) or alanine (which is resistant to phosphorylation) ⁹⁵. Both mutant proteins retain normal huntingtin function, as shown by their ability to rescue huntingtin knock-out phenotypes. Mice expressing the phosphorylation-resistant protein exhibited typical neurodegeneration with associated motor and behavioral phenotypes. By contrast, mice expressing the phosphomimetic protein did not exhibit neurodegeneration, illustrating the dramatic impact of this post-translational modification on toxicity. How might phosphorylation S13 and S16 influence pathogenesis? Thompson and colleagues recently found that these serines are phosphorylated by the I-kappaB kinase (IKK) complex. Through promoting modification of adjacent lysine residues, phosphorylation by IKK targets huntingtin for degradation by a process that requires both the proteasome and lysosome ⁹⁶. Phosphorylation has also been implicated in influencing the toxicity of SCA3, DRPLA and SBMA, for example: phosphorylation of ataxin-3 at S256 by glycogen synthase kinase 3β reduces aggregation of polyglutamine-expanded atxain-3⁹⁷; phosphorylation of polyglutamine-expanded androgen receptor by MAPK at S516 is associated with increased cleavage and toxicity in cell models of SBMA ⁹⁸; and S734 of atrophin-1 is phosphorylated by Jun-N-terminal kinase, although the significance of this in the pathogenesis of DRPLA has not yet been examined.

Acetylation as a determinant of stability

Post-translational protein acetylation takes place on the ε-amino group of lysines and prevents this group becoming positively charged, thus impacting the electrostatic properties of the protein (reviewed in ⁹⁹). Acetylation has diverse consequences, including altered protein-DNA and protein-protein interactions. Acetylation can compete with ubiquitination at lysines residues, such that acetylation increases protein stability⁹⁹. For example, acetylation of ataxin-7 at K257 was recently found to stabilize the protein and promote accumulation of the ataxin-7 N-terminal truncation product that is believed to be the toxic, disease-causing species ¹⁰⁰. Acetylation can also, occasionally, reduce protein stability¹⁰¹.

Krainc and colleagues found that huntingtin stability is regulated by acetylation at K444¹⁰². Interestingly, they found that acetylation facilitates trafficking of mutant huntingtin into autophagosomes, thus serving as a signal for degradation by the autophagy-lysosomal system, rather than influencing proteasomal degradation. Acetylation accelerates clearance of the mutant huntingtin by autophagy and reverses its toxic effects in mouse primary striatal and cortical neurons and in a C. elegans model of HD. When polyglutamine-expanded huntingtin was rendered resistant to acetylation, the protein steadily accumulated and increased degeneration in cultured neurons and in mouse brain. This study provided the first evidence that acetylation can target proteins for autophagic degradation. However, the mechanism of targeting for degradation remains to be determined, as does the extent to which proteins other than huntingtin utilize this mechanism.

Sumoylation as a determinant of cell type vulnerability

Sumoylation is a reversible post-translational modification in which a small ubiquitin-like modifier (SUMO) is covalently conjugated to a lysine residue in a target protein. It provides an efficient way to modulate the subcellular localization, activity and stability of a wide variety of substrates. A role for sumoylation in polyglutamine disease was first suggested because of SUMO immunoreactivity in pathological inclusions characteristic of HD, SCA3 and DRPLA. Thompson and colleagues reported that sumoylation of huntingtin at K6 and K9 stabilizes N-terminal fragments of huntingtin, reduces its ability to form aggregates, and promotes its capacity to repress transcription ¹⁰³. They also noted that sumoylation of a polyglutamine-expanded huntingtin fragment exacerbated neurodegeneration in a Drosophila model of HD.

Recently, huntingtin sumoylation was found to depend upon the small guanine nucleotide -binding protein Rhes by a mechanism independent of Rhes GTPase activity ¹⁰⁴. Rhes preferentially binds polyglutamine-expanded huntingtin and promotes sumoylation, including at K9, K15 and K91. Sumoylation of mutant huntingtin at K15 and K91 correlated with reduced aggregation and increased cytotoxicity. Rhes is expressed only in the striatum, and endogenous Rhes co-purified with huntingtin in extracts from transgenic mice, suggesting that the interaction of these proteins takes place under physiological conditions. Thus, Rhes-huntingtin interactions may account for the localized neuropathology of HD, which is largely restricted to the corpus striatum. Other polyglutamine disease proteins - including ataxin-1, androgen receptor, TBP-1, and atrophin-1 – are sumoylated under physiological conditions, and in some cases this has been correlated with increased toxicity in cell culture disease models. The extent to which sumoylation influences pathogenesis in these diseases is yet to be explored, however.

The two faces of SCA8

Spinocerebellar ataxia type 8 (SCA8) is a slowly progressive, dominantly inherited disease in which affected patients display pronounced signs of ataxia in combination with cortical spinal tract findings ¹⁰⁵. The phenotype is highly variable, and cerebellar atrophy can be dramatic, though poorly correlated with the degree of ataxia. The unearthing of the SCA8 repeat expansion utilized a repeat-directed cloning method to pull out an expanded CTG repeat from a patient with ataxia, and this allowed researchers to demonstrate linkage to the repeat mutation in members of another very large ataxia family ¹⁰⁶. The CTG repeat occurred in a gene that was transcribed and not translated, leading the team to propose that expression of the untranslated, expanded CTG repeat produced neurotoxicity at the RNA level. This hypothesis was not uniformly accepted, and required testing in a mouse model for its validation. Again both Drosophila and mouse models were generated, and the sufficiency of CTG expansion RNA toxicity was demonstrated ^107,108. However, further analysis of the SCA8 BAC transgenic mice revealed an unexpected finding – 1C2-positive intra-nuclear inclusions of the type that would be detected in polyglutamine expansion diseases. As CTG is read as CAG in the anti-sense orientation, careful reexamination of the anti-sense sequence indicated potential translation of a protein consistently almost entirely of polyglutamine. Based upon a series of experiments with materials from SCA8 patients, SCA8 BAC mice, and cell culture models, these investigators demonstrated the existence of a transcript (ATXN8) encoding the predicted polyglutamine-rich ataxin-8 protein, and proposed that both polyglutamine pathology and RNA gain-of-function toxicity combine to yield the SCA8 phenotype. Recent work from this group has reinforced the CUG RNA toxicity of the ataxin-8 anti-sense transcript (ATXN8OS), and has found that the ATXN8OS transcript may sequester Mbnl1 protein ¹⁰⁹. Mbnl1 sequestration was then shown to alter the splicing of gene whose product is a transporter of the inhibitory neurotransmitter GABA. The restricted expression of ATXN8OS is consistent with splicing defects occurring in only the cerebellum, although the extent of altered splicing beyond the GABA transporter gene in cerebellum is yet to be determined.

Whither HDL2?

Another repeat disease that is vying for a place among the difficult-to-classify repeat disorders is Huntington’s disease like-2 (HDL2). HDL2, as its name implies, appears to be a remarkably close phenocopy of Huntington’s disease ¹¹⁰. HDL2 is caused by a CTG repeat expansion occurring within the junctophilin-3 (JPH3) gene, such that affected individuals have 40 or more CTGs ¹¹⁰. Junctophilin-3 is a member of a family of proteins that link plasma membrane voltage sensors with intracellular ion channels ¹¹¹. It remains unclear how the JPH3 CTG repeat expansion produces molecular pathology, as it occurs within the variably spliced exon 2A of the JPH3 gene, and three splice variants have been reported: one placing the CTG repeat in a polyleucine translational frame, one placing the CTG repeat in a polyalanine translational frame, and one localizing the CTG repeat to the 3’ UTR ¹¹². Database queries do not reveal any evidence for transcription of the CAG repeat-containing strand; however, the presence of ubiquitin-positive neuronal intranuclear inclusions (that incidentally do NOT contain the huntingtin protein causal in HD) suggests the existence of a misfolded protein – whether it be polyglutamine or some other poly-amino acid tract. Careful study of HDL2 brain material has revealed evidence for RNA foci, and expression of JPH3 with an expanded CUG tract yields RNA foci containing Mbnl1 ¹¹². Furthermore, altered splicing of the genes encoding tau and amyloid precursor protein occurs in HDL2 frontal cortex. Thus, it appears that CUG RNA toxicity, together with some form of proteotoxicity, may combine in HDL2 to produce neurodegeneration. Hence, SCA8 and HDL2 appear to belong to a category of diseases involving simultaneous RNA gain-of-function and protein gain-of-function molecular pathology (Figure 2). In the case of HDL2, a third possible simultaneous mechanism is partial loss-of-function of junctophilin-3 protein, as JPH3 null mice and heterozygous null mice display impaired coordination and shortened lifespan ¹¹³, and HDL2 patients express reduced levels of JPH3 ¹¹². JPH3 loss-of-function, however, does not yield RNA foci or ubiquitin-positive inclusions in JPH3 knock-out mice, indicating that JPH3 haploinsufficiency alone can not account for HDL2.