GGC repeat expansions within new open reading frames are translated into toxic polyglycine proteins in oculopharyngodistal myopathy
Manon Boivin, Jiaxi Yu, Nobuyuki Eura, Léa Schmitt, David Pietri, Erwan Grandgirard, Patrice Goetz-Reiner, Damien Plassard, Chadia Nahy, Anne Maglott, Bastien Morlet, Chao Gao, Elise Lefebvre, Muriel Philipps, Pascal Eberling, Angélique Pichot, Paola Rossolillo
TL;DR
This study shows that GGC repeat expansions in previously unknown DNA regions cause toxic proteins linked to a rare muscle and brain disease.
Contribution
The discovery that GGC expansions in small ORFs produce toxic polyglycine proteins is novel and expands understanding of disease mechanisms.
Findings
GGC expansions in new ORFs are translated into polyglycine proteins that form p62-positive inclusions.
Polyglycine proteins cause muscle and neurological issues in multiple model systems.
The compound TMPyP4 reduces expression of these toxic proteins, suggesting a potential therapy.
Abstract
A total of 3–6% human genome is composed of microsatellite sequences, which are short DNA elements composed of two to six nucleotide motifs repeated in tandem. Expansion of a subset of these microsatellites is the leading cause of >60 diseases. However, most of these mutations are located in sequences annotated as noncoding, which raises questions about their pathogenicity. Here we found that GGC repeat expansions causing oculopharyngodistal myopathy with or without oculopharyngeal myopathy leukoencephalopathy are located within previously unrecognized open reading frames (ORFs), resulting in their translation into new polyglycine-containing proteins. Antibodies developed against these proteins stain the p62-positive inclusions typical of these diseases. Moreover, expression of these polyglycine proteins causes locomotor and skeletal muscle alterations associated with neurodegeneration…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 10
Figure 11
Figure 12
Figure 13
Figure 14
Figure 15
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9- —https://doi.org/10.13039/501100002915Fondation pour la Recherche Médicale (Foundation for Medical Research in France)
- —https://doi.org/10.13039/501100001665Agence Nationale de la Recherche (French National Research Agency)
- —https://doi.org/10.13039/501100001809National Natural Science Foundation of China (National Science Foundation of China)
- —https://doi.org/10.13039/501100004826Natural Science Foundation of Beijing Municipality (Beijing Natural Science Foundation)
- —https://doi.org/10.13039/501100005090Beijing Nova Program
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsGenetic Neurodegenerative Diseases · Glycogen Storage Diseases and Myoclonus · Amyotrophic Lateral Sclerosis Research
Main
A total of ~98% human genome is comprised of sequences annotated as noncoding, with half of them composed of repetitive DNA elements, including microsatellites, which are two- to six-nucleotide DNA motifs repeated in tandem. These microsatellites, estimated between ~1.5 and 2 million in humans, occupy 3–6% of our genome and are a source of genetic variation as they are highly heterogeneous in size and sequences^1^. Expansion of a subset of these microsatellites over a threshold size is also the leading cause of various human pathologies, including cancer and inherited diseases^2,3^. In that aspect, >60 neurodevelopmental, neuromuscular, and neurodegenerative disorders are known to be caused by expansions of trinucleotide, tetranucleotide, pentanucleotide or hexanucleotide repeats. Remarkably, this number is rapidly increasing as advances in long-read and whole human-genome sequencing have revealed ~20 new pathogenic microsatellite expansions causing human genetic diseases in recent years^4–12^. When embedded within a coding sequence, in-frame repeat expansions are translated into a mutant protein containing a stretch of repeated amino acids. The archetype of this mechanism is the polyglutamine (polyQ) group of diseases, where expansions of CAG repeats, embedded within the open reading frames (ORFs) of diverse genes, are translated into toxic polyQ-containing proteins, ultimately resulting in neuronal cell dysfunction and death. However, a majority of microsatellite expansions, notably most of the recently discovered ones, are located in genomic sequences annotated by default as noncoding (5′-untranslated region (5′UTR) and 3′UTR, introns, antisense RNAs, long noncoding RNAs, etc.), thus questioning their pathogenicity^13^.
Oculopharyngodistal myopathy (OPDM; OMIM 164310) is a rare adult-onset and slowly progressive neuromuscular disease first described in 1977 (ref. ^14^), while oculopharyngeal myopathy with leukoencephalopathy (OPML; OMIM 618637) is an autosomal dominant disorder with oculopharyngeal myopathy, diffuse limb weakness and leukoencephalopathy described more recently^4^. Key clinical features of OPDM and OPML comprise ptosis, external ophthalmoplegia, dysphagia and dysarthria associated with facial and distal limb muscle weakness. Their histopathology is characterized by the presence of large cytoplasmic rimmed vacuoles and rare, but typical, eosinophilic intranuclear inclusions, which are p62-positive and ubiquitin-positive but of unknown origin^4–10,15^. The genetic causes of OPDM and OPML were uncovered recently as similar expansions of ~50 to 200–300 repeats of the trinucleotide GGC sequence located within diverse genomic regions, transcribed but annotated as noncoding and embedded in at least six different genes (LOC642361, LRP12, GIPC1, NOTCH2NLC, RILPL1 and ABCD3)^4–11^. Consequently, these pathologies are now classified into at least six subtypes according to the gene hosting the pathogenic GGC repeat expansion (LOC642361, OPML; LRP12, OPDM1; GIPC1, OPDM2; NOTCH2NLC, OPDM3; RILPL1, OPDM4; ABCD3, OPDM5). Of interest, recent clinical studies indicate that OPDM and OPML have a much wider clinical spectrum than previously thought, with evidence of variable neurological manifestations and reports of movement disorders, tremor, ataxia, visual disturbance, peripheral neuropathy, etc. Finally, OPDM3 shares the same genetic cause with neuronal intranuclear inclusion disease (NIID)^4,5,10^. NIID is a neurological disease characterized by variable muscle weakness associated with heterogeneous dysfunctions of the central and peripheral nervous system. These genetic similarities and clinical overlaps suggest that OPDM, OPML and NIID belong to a new continuum of neurological diseases, which possibly share a common pathophysiological mechanism^16,17^. However, a loss-of-function mechanism is unlikely as expression of the genes hosting these GGC repeat expansions is unaltered in tissue samples from individuals with OPDM, NIID or OPML, with observation of increased or unchanged mRNA levels and, for coding genes, normal protein expression^6–10^. These observations exclude a classical promoter-silencing mechanism but raise questions about how GGC repeat expansions, located within genomic regions annotated as noncoding, can lead to the formation of protein inclusions and cause muscle and neuronal dysfunction.
Here we found that the GGC repeats located in the long ‘noncoding’ LOC642361 RNA, as well as in the ‘noncoding’ sequences of the GIPC1, NOTCH2NLC and RILPL1 genes, are embedded within previously unrecognized ORFs, resulting in expression of new proteins where each GGC repeat encodes for a glycine amino acid. Consequently, these GGC repeat expansions are translated into new polyglycine-containing proteins. Antibodies developed against these proteins confirmed their expression in patients, notably their localization in p62-positive inclusions in muscle sections of individuals with OPDM and OPML. Moreover, expression of these polyglycine proteins in cell and animal models is sufficient to induce formation of the characteristic OPDM/OPML p62-positive inclusions, as well as muscle fiber atrophy associated with neurodegeneration and neuroinflammation, thus recapitulating key clinical features of OPDM, OPML and NIID. Of interest, side-by-side comparison of these diverse polyglycine proteins reveals unexpected variations in their biological properties and toxicity, highlighting a key contribution of the specific amino acid sequences flanking their common polyglycine core. Finally, we tested various pharmacological compounds and identified the cationic porphyrin TMPyP4 as a proof-of-concept therapeutic for these neurological disorders.
Overall, this study highlights the richness and complexity of the human genome, notably the existence of numerous uncharted small ORFs in sequences originally annotated as noncoding, resulting in translation of their embedded microsatellite mutations into new and toxic proteins.
Results
GIPC1, antisense RILPL1 and LOC642361 GGC repeats are translated into polyglycine
As noncanonical translation of repeat expansions is an established model of pathogenicity in microsatellite diseases^18,19^, we investigated the potential translation of the GGC repeats causing OPDM and OPML. Three representative noncoding sequences, namely the 5′UTR of GIPC1, the antisense transcript of RILPL1 and the LOC642361 long noncoding RNA (lncRNA), which are the cause of OPDM2, OPDM4 and OPML, respectively, were cloned with ~50 GGC repeats and fused to green fluorescent protein (GFP) in the three possible frames potentially encoded by these repeats (glycine, alanine and arginine; Fig. 1a,b, Supplementary Note 1 and Supplementary Fig. 1a–f). Notably, transfection into HEK293 cells followed by fluorescence-activated cell sorting (FACS) analysis, direct observation of the GFP fluorescence and western blotting consistently indicate that the GIPC1 GGC repeats are predominantly translated in the glycine frame, while GFP expression in the alanine or arginine frames is negligible (Fig. 1c–e). Similar results were obtained with the antisense RILPL1 and LOC642361 noncoding RNAs (Extended Data Fig. 1a–f). In contrast, analysis of the sense RILPL1 RNA with a CCG expansion shows no detectable translation in any frames (Supplementary Fig. 1g–i). Controls indicate that the lack of GGC repeat translation into polyalanine or polyarginine is not caused by differences in RNA expression, a potential toxicity leading to cell loss or another bias impairing observation of GGC repeats translation in the alanine or arginine frames (Supplementary Note 1 and Supplementary Fig. 1j–l). Finally, the treatment of cell extracts with lysostaphin, a glycyl–glycine endopeptidase, cleaves these proteins into smaller products, thus confirming the presence of a polyglycine stretch within them (Extended Data Fig. 1g–i). Overall, these results indicate that the GIPC1, antisense RILPL1 and LOC642361 GGC repeat expansions, while located in sequences annotated as noncoding, are translated into new polyglycine-containing proteins that have yet to be characterized.Fig. 1GIPC1, RILPL1 and LOC642361 GGC repeats are translated into polyglycine.a, Scheme of the GGC repeat expansions located within the GIPC1, antisense RILPL1 and LOC642361 sequences. b, Scheme of the construct expressing ~50 pure GGC repeats cloned with their upstream host sequences and fused to the GFP in their three potential encoded frames. This plasmid also contains an independent Cherry expression cassette. c, GFP and Cherry FACS analysis (top) with its quantification (bottom) of HEK293 cells transfected for 24 h with a plasmid expressing 50 GGC repeats embedded within the GIPC1 5′UTR sequence fused to the GFP in the glycine, alanine or arginine frames, while the Cherry is expressed independently. Bar heights represent the mean and error bars represent the mean ± s.e.m. Sample size, n = 3 independent biological replicates. d, As in c but with microscopy fluorescence analysis. Scale bars = 10 µm. e, As in c but with immunoblotting analyses (Extended Data Fig. 1 and Supplementary Fig. 1). Full-length blots are provided as Source data. Gly., glycine; Ala., alanine; Arg., arginine.Source data
GIPC1, antisense RILPL1 and LOC642361 GGC repeats are embedded in small ORFs
To uncover how these microsatellites are translated, constructs with 50 GGC repeats embedded in the upstream GIPC1, antisense RILPL1 or LOC642361 sequences and fused to GFP in the glycine frame were transfected into HEK293 cells and their corresponding encoded GFP-tagged polyglycine proteins were immunoprecipitated and analyzed by mass spectrometry to determine their N-terminal sequences (Fig. 2a). In all three sequences, translation starts with a typical acetylated methionine (M^ac^), which corresponds to initiation at standard start codons located upstream of the GGC repeats (Fig. 2a and Supplementary Fig. 2a–d). Translation initiations of the RILPL1 antisense RNA and of the LOC642361 lncRNA occur at classical ATG start codons, while translation of the GIPC1 5′UTR occurs in the absence of any ATG start codon but instead initiates at a CTG near-cognate start codon located ahead of the repeats (Supplementary Fig. 2a and Supplementary Note 2). Near-cognate start codons (CTG, GTG, ACG, TTG) are codons differing from the cognate AUG start codon by one nucleotide, but that can nonetheless initiate translation through mispairing with the initiator methionine-tRNA^20,21^. Western blotting, fluorescence observation and FACS analyses show that the deletion of these ATG or CTG start codons abolishes polyglycine expression, highlighting the requirement of these start codons to drive GGC repeats translation (Fig. 2b and Supplementary Fig. 2e–p). Next, we noted that the GIPC1 5′UTR and the LOC642361 lncRNA sequences with a control size of repeats (~10 GGC) are similarly translated, but into small and unstable peptides, which are hardly detected without inhibition of the cell degradation pathways (Extended Data Fig. 2a–c). In contrast, these microproteins become stable when carrying an expansion of their polyglycine stretch (Extended Data Fig. 2d,e). Overall, these data reveal that the 5′UTR of the GIPC1 gene, the antisense transcript of RILPL1 and the LOC642361 long noncoding RNA contain previously unrecognized ORFs, which are translated independently of the length of their GGC microsatellites. Their translation initiates at either cognate AUG or near-cognate start codons located ahead of the GGC repeats and in the glycine frame, resulting in expression of new proteins where each GGC repeat encodes for a glycine amino acid. In the absence of a GGC expansion and thus with a normal size of glycine stretch, their encoded peptides are of small size (<100 amino acids) and thus unstable and hardly observable. In contrast, when encompassing an expansion over ~50 GGC repeats, these ORFs are translated into new polyglycine-containing proteins, which were named uGIPpolyG, asRILpolyG and LOC6polyG for upstream of GIPC1, antisense of RILPL1 and LOC642361-encoded polyglycine proteins, respectively (Fig. 2c–e, Supplementary Note 2 and Supplementary Fig. 2q–y).Fig. 2GIPC1, RILPL1 and LOC642361 GGC repeats are embedded in small ORFs.a, Scheme of HEK293 cells transfected for 24 h with a plasmid expressing 50 GGC repeats cloned with their upstream GIPC1, antisense RILPL1 and LOC642361 sequences and fused to GFP in the glycine frame, followed by GFP-immunoprecipitation and mass spectrometry analysis to determine the N-terminal sequences of their encoded polyglycine proteins. b, Immunoblot against the GFP, Cherry or GAPDH of proteins extracted from HEK293 cells transfected as in a but with WT or mutant (∆CTG/ATG start codons) constructs expressing 50 GGC repeats cloned with their upstream GIPC1, antisense RILPL1 and LOC642361 sequences and fused to GFP in the glycine frame. c–e, Schemes and amino acid sequences of the new ORFs and their encoded polyglycine proteins identified in the GIPC1 5′UTR (c), RILPL1 antisense transcript (d) and LOC642361 lncRNA (e) (Extended Data Fig. 2 and Supplementary Fig. 2). Full-length blots are provided as Source data. WT, wild type; sORF, short ORF; ter., terminal.Source data
Polyglycine proteins colocalize with p62 inclusions in OPDM/OPML muscle sections
To confirm that these GGC repeat expansions are translated into new polyglycine-containing proteins in individuals with OPDM/OPML, we developed antibodies against their distinct N-terminal or C-terminal sequences (Supplementary Note 3 and Supplementary Fig. 3a–l). Immunofluorescence staining performed on skeletal muscle sections of individuals with OPDM2, OPDM4 and OPML revealed the presence of their respective polyglycine proteins (uGIPpolyG, asRILpolyG and LOC6polyG) within the p62-positive cytoplasmic rimmed vacuoles and intranuclear inclusions typical of these diseases (Fig. 3a–c and Supplementary Fig. 3m–o). Moreover, as OPDM3 and NIID have an identical genetic cause, namely an expansion of GGC repeats in the 5′UTR of the NOTCH2NLC gene, and as this expansion was recently found to belong to a small ORF translated in a polyglycine protein, uN2CpolyG^22–24^, we developed antibodies against this protein and uncovered its presence within the typical p62-positive inclusions in muscle sections of individuals with OPDM3 (Fig. 3d and Supplementary Fig. 3p). No or only faint staining was observed in non-OPDM individuals (Fig. 3a–d and Supplementary Fig. 3m–p), as without a GGC repeat expansion and thus without a polyglycine stretch, these microproteins do not aggregate, and their small sizes impair their stability and their detection. Moreover, as each of these antibodies is directed against a specific ORF sequence encoding a distinct polyglycine protein, these antibodies are specific to each OPDM subtype and indeed do not stain p62-positive inclusions in other OPDM/OPML subtypes (Extended Data Fig. 3a–d). These controls confirm the specificity of our antibodies and support the existence of distinct ORFs, each encoding a unique polyglycine protein, which are consequently specific to each OPDM subtype. Finally, as various microsatellite expansions have been reported to be RAN translated in their three potential frames^18,19^, and as a short expansion of GCN repeats in PABPN1 is translated into a protein with an extended polyalanine stretch that causes oculopharyngeal muscular dystrophy^25^, we also investigated a potential translation of the GIPC1 GGC repeats in the alanine frame. However, two independent antibodies developed against a putative GIPC1 polyalanine protein do not stain intranuclear inclusions or rimmed vacuoles in muscle sections of individuals with OPDM2, arguing against translation of GGC repeats in the alanine frame (Supplementary Fig. 3q–s). Overall, these data confirm that the GGC repeat expansions causing OPDM and OPML are embedded in previously unrecognized ORFs and consequently translated into new polyglycine-containing proteins.Fig. 3. Polyglycine proteins are present in typical OPDM/OPML p62-positive inclusions.a, Top, partial amino acid sequence of the uGIPpolyG protein encoded by the expanded GGC repeats embedded in the GIPC1 5′UTR sequence causing OPDM2. The peptide sequences against which the uGIP antibody is directed are indicated in bold and underlined. Bottom, immunofluorescence staining against p62 and the uGIPpolyG protein on skeletal muscle sections of individuals with OPDM2 or age-matched control individuals. b, As in a but with the asRILpolyG protein stained in individuals with OPDM4. c, As in a but with the LOC6polyG protein stained in individuals with OPML. d, As in a but with the uN2CpolyG protein stained in individuals with OPDM3. Scale bar = 10 µm (a–d; Extended Data Fig. 3 and Supplementary Fig. 3).
Expression of polyglycine proteins forms inclusions and is pathogenic in muscle cells
To study further these polyglycine proteins, we cloned their cDNAs with an expansion of 100 GGN glycine-encoded codons, a strategy preventing repeat instability (Fig. 4a, Supplementary Note 4 and Supplementary Fig. 4a). As a control, expression of 100 pure GGC RNA repeats, deprived of any translation start codon, is not toxic, dismissing a potential RNA toxicity mechanism (Supplementary Note 4 and Supplementary Fig. 4b–e). Of interest, expression of these diverse polyglycine proteins in human LHCN-M2 differentiated muscle cells followed by immunofluorescence revealed that they form cytoplasmic and intranuclear inclusions, which are p62-positive and thus reminiscent of the OPDM, OPML and NIID histopathological features (Fig. 4b,c and Supplementary Fig. 4f). Live imaging suggests that cytoplasmic polyglycine inclusions may be too large to penetrate nuclei but may directly aggregate and grow within the nucleus from soluble polyglycine species and/or microaggregates (Supplementary Videos 1 and 2). Correlative light and electron microscopy (EM) shows that these polyglycine inclusions appear as round-shaped electron-dense deposits composed of filamentous structures without membrane boundaries (Fig. 4d), which is consistent with observations in individuals with OPDM and NIID. Of interest, we noted that these different polyglycine proteins present some differences in their localization, with the OPDM4 asRILpolyG protein being systematically more nuclear than the other (Fig. 4e). Similarly, protein expression assessed by immunoblotting revealed further unexpected variations and different protein half-life, with the uN2CpolyG (OPDM3/NIID) protein consistently less observed (Fig. 4f, Supplementary Note 4 and Supplementary Fig. 4g). As polyglycine proteins accumulate in cellular inclusions that may correspond to insoluble protein aggregates, which classically escape to immunoblot detection performed on the soluble cell fraction, we also performed dot blot analysis of the cell lysate pellet sonicated in 2× Laemmli buffer (Fig. 4f). This assay exposed further disparities between these polyglycine proteins, with the uN2CpolyG, asRILpolyG and LOC6polyG proteins notably more present in the insoluble protein fraction. These results were confirmed by quantification of the localization of these diverse polyglycine proteins in LHCN-M2 muscle cells, notably their presence in inclusions versus a diffuse localization (Extended Data Fig. 4a). Next, immunoprecipitation of these diverse polyglycine proteins followed by mass spectrometry unveiled specific interactants, notably some that are exclusive to peculiar polyGly proteins (Extended Data Fig. 4b and Supplementary Data 1). In that aspect, the uN2CpolyG protein interacts with the KU70/KU80 dimer involved in DNA repair, while the LOC6polyG interacts with ribosomal proteins. These interactions are independent of their glycine stretches, suggesting that the newly identified ORFs may encode functional microproteins, whose physiological importance remains to be thoroughly studied. Finally, expression of these diverse polyglycine proteins is toxic and causes LHCN-M2 muscle cell death, but with some differences with a higher toxicity of the uN2CpolyG, asRILpolyG and LOC6polyG proteins, notably compared to the uGIPpolyG protein or an artificial ATG polyGly protein expressing 100 glycines with no OPDM flanking sequences (Fig. 4g). Live cell tracking indicates that cell death was observed both in cells with polyGly inclusions and in cells showing a diffuse localization of these polyglycine proteins (Extended Data Fig. 4c). Moreover, formation of polyGly inclusions can be abrupt, with a diffuse localization observed during dozens of hours and sudden aggregation in minutes. Notably, some polyGly aggregates were even observed after cell death. Finally, no overt signs of apoptosis were noted, suggesting that these proteins are toxic by another pathway (Supplementary Fig. 4h,i). In conclusion, these diverse polyglycine proteins share the common properties of forming p62-positive cellular inclusions and inducing cell death, recapitulating key features of OPDM and OPML. However, these polyglycine proteins also show different biological properties (localization, half-life, aggregation, toxicity, etc.), suggesting a modulation of their central and common polyglycine core by their specific flanking amino acid sequences.Fig. 4. Expression of polyglycine proteins forms inclusions and is pathogenic in muscle cells.a, Scheme of the constructs encoding GFP-tagged polyGly, uGIPpolyG, uN2CpolyG, asRILpolyG or LOC6polyG protein cloned with an optimized expansion of 100 GGN repeats. b, GFP fluorescence and immunofluorescence against the desmin and lamin A/C proteins of LHCN-M2 cells differentiated into myotubes for 4 days and expressing either GFP-tagged ATG polyG, uGIPpolyG, uN2CpolyG, asRILpolyG or LOC6polyG. Scale bars = 10 µm. c,d, GFP fluorescence with immunofluorescence against p62 and lamin A/C (c) or CLEM (d) of LHCN-M2 cells expressing asRILpolyG-GFP and differentiated into myotubes for 4 days. Scale bars, as indicated. e, Quantification of nuclear versus cytoplasmic localization of the polyglycine proteins studied in b. Bar heights represent the mean. Error bars represent the mean ± s.e.m. Sample size, n = 3 independent biological replicates. f, SDS–PAGE gel and immunoblot against the GFP or the GAPDH of soluble proteins (top), and dot blot against the GFP or Ponceau staining of the insoluble proteins (bottom) extracted from 48-h differentiated LHCN-M2 muscle cells expressing either the GFP or GFP-tagged ATG polyG, uGIPpolyG, uN2CpolyG, asRILpolyG or LOC6polyG. g, Cell viability of LHCN-M2 muscle cells differentiated for 3 days and expressing GFP or GFP-tagged ATG polyG, uGIPpolyG, uN2CpolyG, asRILpolyG or LOC6polyG. Bar heights represent the mean and error bars represent the mean ± s.e.m. Sample size, n = 6 independent biological replicates. Unpaired two-tailed t test compared to the GFP control condition (Extended Data Fig. 4 and Supplementary Fig. 4). Full-length blots are provided as Source data. CLEM, correlative light and EM.Source data
Polyglycine proteins form inclusions and are pathogenic for muscles in mice
To determine the physiological impact of these polyglycine proteins, we expressed them through a recombinant adeno-associated viral (rAAV) strategy in mouse skeletal muscles (Fig. 5a). Histological analyses of the tibialis anterior muscles up to 10 months after rAAV injection show that polyglycine proteins are toxic and promote muscle fiber atrophy with the presence of internalized or centralized nuclei, but with some striking differences among these proteins. Indeed, expression of the OPDM4 asRILpolyG, OPML LOC6polyG and OPDM3 uN2CpolyG proteins causes histological changes in 3–5 months after rAAV injection, while the OPDM2 uGIPpolyG protein shows a lesser toxicity with some muscle changes detected only 9 months after rAAV injection (Fig. 5b,c, Supplementary Note 5 and Supplementary Fig. 5a). Similarly, expression of ATG polyGly, a protein deprived of any OPDM natural bordering sequences, shows a limited and delayed pathogenicity. Analyses of p62 staining revealed numerous p62-positive cytoplasmic and intranuclear inclusions, as observed in OPDM patients (Fig. 5d and Supplementary Fig. 5b). These inclusions are eosinophilic, which is especially apparent in the uN2CpolyG expressing mice (Fig. 5b). Of interest, all OPDM polyglycine proteins form inclusions, but with some notable differences, with observation of frequent OPML LOC6polyG and OPDM3 uN2CpolyG aggregates, while ATG polyG and OPDM2 uGIPpolyG inclusions are less represented (Fig. 5e). Moreover, the localization of these polyglycine proteins varies, with the OPDM4 asRILpolyG protein more observed in nuclei compared to the other polyGly proteins (Fig. 5e). Single nuclei RNA sequencing revealed an increase in macrophages and B-cells, as well as in regenerative muscle fibers, in OPDM versus control mice (Extended Data Fig. 5a,b). These results indicate signs of inflammation and muscle regeneration consistent with myopathic changes in OPDM mice. However, these alterations were mild, with limited transcriptomic changes and only minor changes in muscle fiber types (Supplementary Fig. 5c,d and Supplementary Data 2). Correspondingly, animal performances were only slightly altered in rotarod and open field locomotor tests (Supplementary Fig. 5e,f). These data indicate that expression of polyglycine proteins in mice causes progressive muscle fiber atrophy and histological changes reminiscent of OPDM, but with specific and limited myopathic alterations, at least in the time frame and AAV-driven mouse models analyzed here. Finally, expression of the asRILpolyG and LOC6polyG proteins are remarkably deleterious as these mice die suddenly around 5–6 months or 8–9 months after rAAV injection, respectively (Fig. 5f). These mice present dilated cardiomyopathy with the presence of numerous p62-positive inclusions in cardiomyocytes (Extended Data Fig. 5c). Abundance of these aggregates mirrors their toxicity with rare ATG polyG and uGIPpolyG inclusions, an intermediate situation for uN2CpolyG, while the asRILpolyG and LOC6polyG proteins form numerous large aggregates associated with notable myopathic changes (Extended Data Fig. 5c). These data are reminiscent of the cardiac dysfunctions reported in OPDM and NIID^26,27^ and led to investigate the toxicity of these polyglycine proteins in other tissues, notably the central nervous system (CNS), especially in regards of the neurological manifestations reported in individuals with OPDM2, OPDM3/NIID and OPML^4,5,10,28–30^.Fig. 5. Expression of polyG proteins forms inclusions and is pathogenic in mouse muscles.a, Scheme of the AAV strategy to study OPDM/OPML polyglycine toxicity in mouse skeletal muscles. Panel a was created with BioRender.com. b, H&E staining of TA frozen sections of 5-month-old AAV-injected male mice expressing GFP, ATG polyG-GFP or GFP-tagged OPDM2 uGIPpolyG, OPDM3 uN2CpolyG, OPDM4 asRILpolyG or OPML LOC6polyG. The last image shows a representative p62 immunohistochemistry, which reveals numerous protein inclusions. Scale bars = 50 µm. c, Quantification of mouse TA muscle fiber area 5 months (top) or 9 months (bottom) postinjection of AAV expressing the OPDM/OPML polyglycine proteins and controls shown in a. Bar heights represent the mean. Error bars represent the mean ± s.e.m. Sample size, n = 4 mice per condition and with at least 1,000 muscle fibers counted per animal. One-way ANOVA with Tukey’s post hoc test. d, GFP fluorescence and immunofluorescence against p62 with counterstaining of membranes by fluorescence-conjugated WGA and nuclear DNA by DAPI on frozen TA muscle sections 5-months postinjection of AAV expressing the OPDM/OPML polyglycine proteins and controls described in a. Scale bars = 10 µm. e, Quantification of GFP-positive inclusions in TA frozen sections of controls and OPDM/OPML polyglycine-expressing mice. Bar heights represent the mean and error bars represent the mean ± s.e.m. Sample size, n = 3 mice per condition, with at least 1,000 muscle fibers counted per animal. f, Kaplan–Meier survival curve of controls and OPDM/OPML polyglycine-expressing mice. Sample size, n = 8 mice per condition (Extended Data Fig. 5 and Supplementary Fig. 5). ANOVA, analysis of variance; H&E, hematoxylin and eosin; WGA, wheat germ agglutinin; TA, tibialis anterior.
Polyglycine proteins form inclusions and are pathogenic for the CNS in mice
Expression of polyglycine in the mouse CNS is toxic, resulting in progressive motor performances and coordination changes, associated with a reduced lifespan (Fig. 6a,b and Extended Data Fig. 6a–e). However, we noted some differences among these diverse polyglycine proteins, with mice expressing the ATG polyG or the OPDM2 uGIPpolyG protein showing a milder pathogenicity and longer lifespan compared to mice expressing the OPML LOC6polyG or the OPDM3 uN2CpolyG proteins. Next, p62 staining revealed that these polyglycine proteins form cytoplasmic and intranuclear inclusions, recapitulating key histopathological features of OPDM, OPML and NIID (Fig. 6c, Supplementary Note 6 and Supplementary Fig. 6a). Abundance of these polyglycine inclusions mirrors their toxicity, and their accumulation is age dependent (Fig. 6c and Extended Data Fig. 6f). Finally, polyglycine expression leads to neuroinflammation and neuronal cell death, notably loss of Purkinje cells (Fig. 6d and Supplementary Fig. 6b,c). Overall, these data confirm that expression of polyglycine-containing proteins is toxic and recapitulate key features of OPDM, OPML and NIID, notably myopathic changes and neurodegeneration associated with the presence of typical p62-positive inclusions. Moreover, side-by-side analysis of these diverse polyglycine proteins revealed some notable differences in their expression, localization and toxicity, highlighting the importance of their specific flanking amino acid sequences to modulate the toxic properties of their central polyglycine core.Fig. 6. Expression of polyglycine proteins forms inclusions and is pathogenic in mouse CNS.a, Scheme of the AAV strategy to study OPDM/OPML polyglycine toxicity in mouse CNS. Panel a was created with BioRender.com. b, Kaplan–Meier survival curve of controls and OPDM/OPML polyglycine-expressing mice. Sample size, n = 10 mice per condition. c, Top, immunohistochemistry against p62 with cresyl violet (Nissl) counterstaining of various mouse brain areas 3-month postinjection of AAV expressing GFP, ATG polyG-GFP or GFP-tagged OPDM2 uGIPpolyG, OPDM3 uN2CpolyG or OPML LOC6polyG. Scale bars = 50 µm. Bottom, quantification of p62-positive inclusions. Bar heights represent the mean. Error bars represent the mean ± s.e.m. Sample size, n = 3 mice per condition, with at least 200 nuclei counted per brain region and per animal. d, Top, immunofluorescence against p62 and calbindin on the cerebellum 3-months postinjection of AAV expressing controls or OPDM/OPML GFP-tagged polyG proteins. Scale bars = 50 µm. Bottom, quantification of PC number. Bar heights represent the mean. Error bars represent the mean ± s.e.m. Sample size, n = 3 mice per condition, with 4 mm^2^ cerebellum area counted per animal (Extended Data Fig. 6 and Supplementary Fig. 6). PC, Purkinje cell.
Porphyrin TMPyP4 alleviates aggregation and toxicity of polyglycine proteins
To alleviate the toxicity of these polyglycine proteins, we tested various compounds and identified one, the cationic porphyrin TMPyP4, that efficiently reduces their abundance and toxicity in cell cultures (Fig. 7a,b, Supplementary Note 7 and Supplementary Fig. 7a–e). RNA sequencing and mass spectrometry revealed that TMPyP4 induces only limited changes and no global transcriptomic or proteomic alterations (Supplementary Fig. 7f–i and Supplementary Data 3 and 4). Pathway analysis revealed that TMPyP4 acts principally on translation (Supplementary Data 5), which is consistent with its known inhibitory function on the translation of GC-rich microsatellites^31,32^. Next, to investigate TMPyP4 effects in animals, we developed Drosophila expressing polyglycine proteins. Ubiquitous expression of uGIPpolyG (OPDM2) leads to a progressively reduced mobility and shortened lifespan, while expression of asRILpolyG (OPDM4) was particularly toxic with no, or very few, animals surviving to the adult stage (Supplementary Note 7 and Extended Data Fig. 7a,b). Expression of these polyglycine proteins in Drosophila eyes led to ommatidial degeneration and loss of rhabdomeres, but with a higher toxicity of the OPDM4 asRILpolyG protein compared to the OPDM2 uGIPpolyG protein (Extended Data Fig. 7c and Supplementary Fig. 7j). These results are consistent with observations in cells and mice, highlighting in a third model the importance of the specific amino acid sequences flanking the common polyglycine core of these proteins to modulate their pathogenicity. Notably, TMPyP4 corrects polyglycine toxicity, restoring normal eye structure and rhabdomeres in uGIPpolyG and asRILpolyG-expressing Drosophila (Fig. 7c). Overall, these data highlight that expression of polyglycine proteins reproduces the locomotor and neurodegenerative clinical features observed in the OPDM, OPML and NIID disorders, and that modulating the expression and toxicity of these polyglycine proteins could be of therapeutic interest for these neurological diseases (Fig. 8).Fig. 7. The porphyrin TMPYP4 alleviates aggregation and toxicity of polyglycine proteins.a, Dot blot against GFP or Ponceau staining of the insoluble proteins extracted from 48-h differentiated LHCN-M2 muscle cells expressing GFP-tagged ATG polyG, OPDM2 uGIPpolyG, OPDM3 uN2CpolyG, OPDM4 asRILpolyG or OPML LOC6polyG and treated overnight with the indicated drug concentration. b, Cell viability of LHCN-M2 muscle cells differentiated for 3 days and expressing GFP, ATG polyG-GFP or GFP-tagged OPDM2 uGIPpolyG, OPDM3 uN2CpolyG, OPDM4 asRILpolyG or OPML LOC6polyG and treated overnight with no or 0.3, 1 or 3 µM of TMPyP4. Bar heights represent the mean. Error bars represent the mean ± s.e.m. Sample size, n = 6 independent biological replicates. Unpaired two-tailed t test compared to each nontreated control condition. c, TMPyP4 ameliorates ommatidial degeneration in Drosophila models of polyG-expanded proteins. Left, fly eyes representative EM images of 20-day-old OPDM2 uGIPpolyG or OPDM4 asRILpolyG-expressing Drosophila fed with no, 30, 100 or 200 μM of TMPyP4. Scale bars = 5 μm in columns 1 and 3, and 2 μm in columns 2 and 4. Right, quantification analysis revealed that TMPyP4 significantly preserved ommatidial integrity in OPDM polyglycine-expressing flies. Bar heights represent the mean. Error bars represent the mean ± s.e.m. All quantitative data are given as the number of rhabdomeres per ommatidium. Sample size, n = 40–56 ommatidia from three flies per condition. One-way ANOVA with Bonferroni post hoc test (Extended Data Fig. 7 and Supplementary Fig. 7).Source dataFig. 8Model of polyG toxicity in OPDM/OPML neurological diseases.Expanded GGC repeats within sequences originally annotated as noncoding are embedded in previously undescribed ORFs, resulting in expression of new polyglycine-containing proteins that form protein inclusions and are toxic for neuronal and muscle cells. Of interest, toxicity and biological properties of their central polyglycine core are modulated by their bordering sequences, which are specific to each hosting ORF. Moreover, these data recall the neurodegenerative NIID, FXTAS and SCA4 disorders, suggesting existence of a wider neurological spectrum of diseases caused by polyGly proteins. FXTAS, fragile X-associated tremor/ataxia syndrome; SCA4, spinocerebellar ataxia 4.
Discussion
OPDM, NIID and OPML are neurological diseases caused by similar GGC repeat expansions, but embedded in sequences annotated as noncoding in diverse genes (LOC642361, LRP12, GIPC1, NOTCH2NLC, RILPL1 and ABCD3). Here we found that some of these GGC repeats are embedded in previously uncharted ORFs and are translated into new polyglycine-containing proteins that form p62-positive protein inclusions and are toxic in cell and animal models. In addition, this work clarifies some prerequisites for these GGC microsatellite expansions to be translated into stable and detectable polypeptides, notably the necessity for these repeats to be (1) located in an RNA transcript, which may include ill-described sequences annotated by default as noncoding, but with the requirement that this RNA is exported within the cytoplasm where translation occurs; (2) embedded within an ORF, with the crucial point to be in frame with an upstream cognate ATG or near-cognate (ACG, CTG, GTG or TTG) start codon with its associated Kozak motif and (3) of sufficient size for the encoded polypeptide to be stable and thus reliably detected in tissues (Supplementary Note 8).
Overall, these data are reminiscent of the fragile X-associated tremor/ataxia syndrome and spinocerebellar ataxia 4, where GGC repeat expansions, respectively, located in a small upstream ORF (uORF) of the FMR1 gene or within the main ORF of the ZFHX3 protein, are translated into polyglycine-containing proteins, which are toxic and accumulate in p62-positive inclusions^12,33–36^. Altogether, these observations support the existence of a new group of human disorders, the polyG (or polyGly) diseases, where similar expansions of GGC repeats are embedded in diverse, previously poorly characterized, ORFs and consequently translated into various polyglycine-containing proteins, which form protein inclusions and are toxic for muscle and neuronal cells. Moreover, this work reinforces the proposition that OPDM, OPML, NIID, fragile X-associated tremor/ataxia syndrome and spinocerebellar ataxia 4 belong to a new continuum of neuromuscular and neurodegenerative diseases with similar genetic causes, overlapping clinical and histopathological presentations, and a shared mechanism of pathogenicity (Fig. 8). This recalls the pioneering discovery of the polyQ group of diseases where similar expansions of CAG repeats, embedded within the main ORFs of diverse genes, are translated in various polyQ-containing proteins that form protein aggregates and are pathogenic in various neurodegenerative disorders.
However, this work also raises several questions, notably whether other microsatellite expansions, embedded in yet unrecognized ORFs and thus translated into potentially toxic proteins, remain to be identified (Supplementary Note 8). Patent candidates would be the GGC repeat expansions found in the LRP12 and ABCD3 genes, causing OPDM1 and OPDM5, respectively^4,11^. Whether these microsatellite mutations located in sequences annotated as noncoding are nonetheless translated into new and toxic proteins is an exciting question for future studies. It also remains to be determined how these polyglycine proteins are toxic. They form cellular inclusions, which is consistent with the known self-aggregation properties of glycine homopolypeptides that form amyloid-like fibrils^37,38^. However, whether these polyglycine proteins are toxic under their soluble or aggregated form is unknown. Similarly, it is unclear whether their localization is important for their toxicity, and, in that aspect, how these polyglycines accumulate in the nucleus in the absence of an evident nuclear localization signal remains to be clarified (Supplementary Note 8). Another point of interest is the side-by-side comparisons of these diverse polyglycine proteins, which reveal that their abilities to form inclusions and promote cell death originate from their central and common polyglycine core, while their localization, expression, half-life, aggregation and interactions with other proteins are modulated by their specific N-terminal and C-terminal flanking sequences, originating from their hosting ORFs. It remains to be investigated thoroughly how these flanking sequences modulate the toxicity of these polyglycine proteins.
In conclusion, this work highlights a unified pathogenic mechanism for the skeletal muscle and CNS dysfunctions observed in individuals with OPDM, NIID and OPML, where GGC repeat expansions are embedded in previously unrecognized ORFs and consequently translated into new and toxic polyglycine-containing proteins. Consistent with a shared mechanism of toxicity, this work also provides proof of concept that a common therapeutic approach may be worth pursuing for these neurological disorders.
Methods
Human samples
Human muscle samples were sampled with the informed consent of individuals and families and approved by the Institutional Review Board of the Peking University First Hospital, First Affiliated Hospital of Fujian Medical University and National Center of Neurology and Psychiatry. This study was approved by the ethics committees of Peking University First Hospital, First Affiliated Hospital of Fujian Medical University and National Center of Neurology and Psychiatry, and all procedures were conducted in accordance with relevant guidelines and regulations. Muscle biopsy samples from patients with OPDM, OPML and NIID and age-matched control participants were examined. All clinical materials were obtained for diagnostic purposes after informed consent was provided. Before this study, all samples had been analyzed using routine histology techniques and EM. Fresh-frozen samples were stored at −80 °C until use.
Mice
All animal work were performed with approval from the IGBMC/ICS Animal Care Committee and the French agency for research on animals, Direction générale de la recherche et de l’innovation (DGRI), authorization APAFIS33864-2021111217327782. C57BL/6 wild-type male mice were retro-orbitally AAV-injected at 2 months and then housed for 6–8 months in a temperature-controlled room (19–22 °C) with a 12-h light/12-h dark cycle and free access to food and water. Mice were killed by carbon dioxide (CO_2_) inhalation to dissect the different skeletal muscles, heart and brain, which were subsequently frozen for molecular biology, freezing using prechilled isopentane or paraformaldehyde (PFA)-fixed and embedded in paraffin for histology.
Cell cultures
U2OS and HEK293 cells were grown in DMEM containing 1 g l^−1^ glucose with 10% FCS and gentamicin at 37 °C in 5% CO_2_. LHCN-M2 cells were grown in DMEM containing 4.5 g l^−1^ glucose with 20% FCS, without PyrNa/M199, supplemented with 25 µg ml^−1^ fetuin, 5 mg ml^−1^ hEGF, 0.5 mg ml^−1^ human bFGF, 5 µg ml^−1^ human insulin, 0.2 µg ml^−1^ dexamethasone and gentamicin at 37 °C in 5% CO_2_. Differentiation of the LHCN-M2 cells was induced by serum removal. U2OS T-Rex cells (Thermo Fisher Scientific) stably expressing Nup50-Cherry were lipofectamine-transfected with Pci1-linearized pcDNA3 expressing Nup50 fused to the mCherry and selected for neomycin resistance for 2 weeks.
Constructs
Human GIPC1 exon 1, antisense RILPL1 and LOC642361 lncRNA sequences upstream of their GGC repeats were cloned into pcDNA3.1 fused to GFP lacking its start codon (ΔATG), with each insert cloned in all three reading frames. Mutations of the ATG or CTG start codons, or within ORFs, were achieved by inverse PCR or by oligonucleotide ligations. GIPC1 uORF, antisense RILPL1 and LOC642361 small ORFs with either 12 or 100 optimized GGN repeats were synthesized by GenScript and fused to GFP into a pAAV2-CAG vector. To ensure repeat expansions stability, all GGC repeat-containing plasmids were transformed into STBL3 bacterial strain (Invitrogen), and all constructs were confirmed by sequencing.
Cell transfection and treatments
For transient transfection, cells were plated and transfected the following day in medium with 0.1% FCS or without serum for LHCN-M2 cells for 5 h using Lipofectamine 2000 (Thermo Fisher Scientific). After 5 h to 4 days post-transient transfection, cells were analyzed by live imaging, immunofluorescence, real-time qPCR (RT-qPCR), cell viability, dot blotting or western blotting. For treatments, LHCN-M2 cells were incubated overnight with indicated concentration of SRPIN340, H-89, fluphenazine TMPyP4, 5,10,15,20-tetra(4-pyridyl)-21H,23H-porphine, 5,10,15,20-tetraphenyl-21H,23H-porphine (Sigma-Aldrich), 5,10,15,20-tetrakis(4-aminophenyl)-21H,23H-porphine, 5,10,15,20-tetrakis(4-ethynylphenyl)-21H,23H-porphine, 5,10,15,20-tetra(pyridin-2-yl)porphyrin, (porphyrin-5,10,15,20-tetrayltetrakis(benzene-4,1-diyl))tetraboronic acid, 4,4′,4″,4‴-(21H,23H-porphine-5,10,15,20-tetrayl)tetrakis-phenol (BLDpharm). For cycloheximide treatment, HEK293 cells were treated 1 day post-transfection with 50 µg ml^−1^ of cycloheximide for 1, 3, 8 or 24 h.
Cell viability assay
LHCN-M2 cells were transiently transfected for 72 h with the different polyglycine-expressing constructs and treated overnight with the indicated drug concentration. After the addition of 0.5 µM of TO-PRO-3 (Thermo Fisher Scientific), live cells were imaged using the CX7 Cellular Imaging System (25 fields per well at ×10 magnification), followed by a cell-to-cell analysis using Cellomics HCS Studio software (CellHealth Bioapplication). Transfected cells were detected using GFP staining and dead cells were identified using TO-PRO-3 intensity within cell mask.
FACS analysis
HEK293 cells transfected for 24 h with the different frame constructs were trypsinized, centrifuged for 5 min at 100g and resuspended in 500 µl of PBS. Cells were analyzed by the BD LSRFortessa X-20 and results were construed by FlowJo.
Western blotting
Proteins were denatured for 3 min at 95 °C, separated on 4–12% Bis–Tris gel (NuPAGE), transferred on nitrocellulose membranes (Amersham Protan), blocked with 5% nonfat dry milk in Tris-buffered saline with 0.1% Tween-20 (TBS-T), incubated with anti-GFP (Abcam, ab290; 1:10,000), anti-GFP (Abcam, ab1218; 1:10,000), mCherry (Abcam, ab167453; 1:5,000), GAPDH (Abcam, ab8245; 1:10,000), KU70 (SantaCruz, sc-56129; 1:5,000), KU80 (Abcam, ab119935; 1:10,000), RPL10A (Thermo Fisher Scientific, MA5-27171; 1:3,000), RPL36 (Abcam, ab241584; 1:10,000), HA (Abcam, ab130275; 1:5,000), uGIP pAb or uN2C pAb (rabbit polyclonal homemade; 1:1,000), uN2C 4D12, asRIL 2D8 or 4B9 or LOC6 2E8 (mouse monoclonal homemade; 1:1,000) in TBS-T with 5% nonfat dry milk overnight at 4 °C. The membranes were washed thrice and incubated with antirabbit or mouse peroxidase antibody Cell Signaling Technology, 7074S or 7076S; 1:10,000) 1 h in TBS, followed by washing and ECL Prime chemiluminescence revelation kit (Millipore).
Dot blotting
LHCN-M2 cells transfected with ATG polyG-GFP, uGIPex2polyG-GFP, uGIPex4polyG-GFP, uN2CpolyG-GFP, asRILpolyG-GFP or LOC6polyG-GFP constructs during 48 h were scraped and centrifuged for 10 min at 700g at 4 °C. The cell pellet was frozen, resuspended in 200 µl of RIPA, frozen and centrifuged for 10 min at 20,000g at 4 °C. The pellet was resuspended in 200 µl of 2× Laemmli buffer, sonicated for 5 s at 20% amplitude and warmed for 3 min at 95 °C. Proteins were directly loaded on nitrocellulose membranes (Amersham Protan), washed twice with Towbin buffer, blocked with 5% nonfat dry milk in TBS-T and incubated with anti-GFP (Abcam, ab290; 1:10,000) in TBS-T with 5% nonfat dry milk overnight at 4 °C. The membranes were washed thrice and incubated with antirabbit peroxidase antibody (Cell Signaling Technology, 7074S; 1:10,000) 1 h in TBS, followed by washing and ECL Prime chemiluminescence revelation kit (Millipore).
Lysostaphin treatment
HEK293 cells transfected with uGIPpolyG-GFP, asRILpolyG-GFP or LOC6polyG-GFP constructs were scraped and centrifuged for 10 min at 700g at 4 °C. The cell pellet was resuspended in 300 µl of RIPA and centrifuged for 10 min at 20,000g at 4 °C. Thirty microliters of supernatant extract were incubated with 10 ng µl^−1^ of lysostaphin (Prospec, ENZ-269) for 1–20 min at 37 °C. Laemmli buffer was added to the mix and proteins were analyzed by western blot.
AAV production and retro-orbital injection
Recombinant AAV were generated by triple-transfection of HEK293T/17 cell line with the pAAV expression plasmids (expressing—GFP, ATG polyG-GFP, uGIPex2polyG-GFP, uGIPex4polyG-GFP, uN2CpolyG-GFP, asRILpolyG-GFP or LOC6polyG-GFP), the auxiliary plasmid pHelper (Agilent) encoding the adenovirus helper functions and the capsid plasmid pUCmini-iCAP-PHP.eB (Addgene, 103005) or pMyoAAV-4A. The pMyoAAV-4A was previously generated by the IGBMC facility using available literature^39^. The rAAV were collected from cell lysate and treated with Benzonase (Merck) at 100 U ml^−1^. Recombinant vectors were purified by Iodixanol gradient ultracentrifugation (OptiPrep, Axis Shield), followed by dialysis and concentration (Amicon Ultra-15 Centrifugal Filter Device 100K) against sterile PBS (Dulbecco’s PBS containing 0.5 mM MgCl_2_). Particles were quantified by RT-PCR and vector titers were expressed as viral genomes per ml (vg ml^−1^). The 2-month-old C57BL/6 male mice were injected retro-orbitally with 100 µl of sterile NaCl with 1.5 × 10^13^ vg kg^−1^ of AAV.
Mouse phenotyping
Rotarod test (Bioseb) was performed with three testing trials during which the rotation speed accelerated from 4 to 40 rpm in 5 min. Trials were separated by 10–15-min interval. The average latency was used as index of motor coordination performance.
Notched bar test
Mice were tested under 100-lux lighting on a 2-cm-wide and 50-cm-long natural wooden piece notched bar comprising 12 platforms of 2 cm, spaced by 13 gaps of 2 cm and bearing a 6-cm^2^ terminal platform. Animals had to cross the notched bar twice for training and thrice for the test. Every instance of a back paw passing through the gap was counted as an error, and the global error percentage was calculated.
Open field test
Mice were tested in automated open fields, each of which was virtually divided into central and peripheral regions. The open fields were placed in a room homogeneously illuminated at 120 lux. Each mouse was placed in the periphery of the open field and allowed to explore freely the apparatus for 30 min, with the experimenter out of the animal’s sight. The distance traveled, the number of rears and time spent in the central and peripheral regions were recorded over the test session. The number of entries and the percent time spent in the center area are used as an index of emotionality/anxiety.
Immunofluorescence on PFA-fixed cells
Glass coverslips containing plated cells were fixed for 15 min in PBS with 4% PFA, washed with PBS and incubated in PBS with 0.5% Triton X-100 for 5 min. The coverslips were incubated during 1 h with primary antibody against p62 (Abcam, ab56416; 1:1,000), p62 (Abcam, ab109012; 1:1,000), desmin (Abcam, ab32362; 1:500), lamin A/C (Abcam, ab238303; 1:1,000), uGIP pAb or uN2C pAb (rabbit polyclonal homemade; 1:100) and uN2C 4D12, asRIL 2D8 or 4B9 or LOC6 2E8 (mouse monoclonal homemade; 1:100). After washing with PBS, the coverslips were incubated with donkey antimouse or donkey antirabbit secondary antibodies conjugated with Alexa 488, CY3 or CY5 (Jackson Immunoresearch; 1:500) for 1 h, washed twice with PBS and incubated for 3 min in PBS/DAPI (1:10,000 dilution). Coverslips were rinsed twice before mounting in Pro-Long media (Molecular Probes).
Immunofluorescence or immunochemistry on PFA-fixed tissue sections
For immunochemistry followed by cresyl violet counterstaining, buffers were DEPC-treated and autoclaved. Brain sections were deparaffinized for 10 min in Sub-X (Leica) and dehydrated as follows: ethanol 100% (10 min), ethanol 90% (5 min), ethanol 70% (5 min) and rinsed in water. Antigen retrieval was done in pressure cooker in 10 mM Tris pH 9, 1 mM EDTA or 10 mM sodium citrate pH 6. For immunochemistry, endogenous peroxidase activity was blocked for 15 min with 3% H_2_O_2_. Slides were blocked for 1 h with PBS, 0.5% Triton X-100 and 5% horse serum for immunofluorescence of PBS with 0.1% Tween-20 and 5% BSA for immunochemistry followed by overnight incubation at 4 °C with primary antibody against Calbindin (Cell Signaling Technology, 13176S; 1:800), GFAP (Abcam, ab68428; 1:10,000), p62 (Abcam, ab56416; 1:1,000), p62 (Cell Signaling Technology, 23214S; 1:500) or tyrosine hydroxylase (Abcam, ab112; 1:2,000). For immunofluorescence, slides were washed with PBS containing 0.1% Triton X-100, incubated with donkey antimouse or donkey antirabbit secondary antibodies conjugated with Alexa 488 or CY3 (Jackson Immunoresearch; 1:500) for 1 h, washed twice with PBS containing 0.1% Triton X-100 and incubated for 3 min in PBS/DAPI (1:1,000 dilution). Slides were rinsed twice in PBS before mounting in Pro-Long media (Molecular Probes). For immunochemistry, slides were washed with PBS containing 0.1% Tween-20, incubated with horse antimouse or antirabbit coupled to peroxidase (Vector, MP-7402 or MP-7401) for 30 min, washed with PBS containing 0.1% Tween-20 and then revealed by DAB EqV substrate (Vector, SK-4103) under binocular magnifier. The reaction was stopped by immersing the slide in PBS. Then, the slides were washed for 15 min in water, and stained in 1% cresyl violet solution at 55 °C for 10 min. Slides were washed in water, quickly dehydrated in 100% ethanol, immersed in Sub-X and mounted in CV Ultra mounting medium (Leica).
Immunofluorescence or immunochemistry on isopentane-frozen sections
For immunochemistry, endogenous peroxidase activity was blocked for 15 min with 3% H_2_O_2_. Muscle sections were blocked for 1 h with PBS and 3% BSA and incubated overnight at 4 °C with primary antibody directed against p62 (Abcam, ab109012; 1:1,000), p62 (Cell Signaling Technology, 23214S; 1:500), lamin A/C (Abcam, ab238303; 1:1,000), lamin B1 (Proteintech, 12987-1-AP; 1:500), type I fibers (DSHB, BA-D5; 1:50), type IIa fibers (DSHB, SC-71; 1:50), type IIb fibers (DSHB, BF-F3; 1:50), uGIPpolyGly pAb or uN2C pAb (rabbit polyclonal homemade; 1:100) and uGIPpolyAla 1A7 or 3G4, uN2C 4D12, asRIL 2D8 or 4B9 or LOC6 2E8 (mouse monoclonal homemade; 1:100). After washing with PBS, the slides were incubated with donkey antimouse or donkey antirabbit secondary antibodies conjugated with Alexa 488, CY3 or CY5 (Jackson Immunoresearch; 1:500), goat antimouse IgM DyLight 405, IgG2b CY3 and IgG1 CY5 (Jackson Immunoresearch; 1:100) and wheat germ agglutinin conjugated to Alexa 555 (1:300) for 1 h, washed twice with PBS and incubated for 3 min in PBS/DAPI (1:1,000 dilution) for immunofluorescence. Concerning immunochemistry, slides were washed in PBS, incubated with horse antirabbit coupled to peroxidase (Vector, MP-7401) for 30 min, washed in PBS and revealed by DAB EqV substrate (Vector, SK-4103). The reaction was stopped by immersing the slide in PBS. Then, the slides were washed for 15 min in water, hematoxylin and eosin stain after the constructor’s instructions (Abcam, ab245880). For succinate dehydrogenase activity, tissue sections were incubated in a succinate dehydrogenase reaction mixture (1.5 mM nitroblue tetrazolium, 5 mM EDTA, 48 mM succinic acid, 750 µM sodium azide, 30 mM methyl–phenylmethlyl sulfate, phosphate buffered to pH 7.6). Slides were immersed in Sub-X and mounted in CV Ultra mounting medium (Leica).
Image acquisition
Slides were imaged by spinning disk Yokogawa CSU W1 mounted with Leica Dmi 8 microscope with ×20 or ×63 objectives or by Zeiss Axioscan 7 scanner with ×20 objective for immunohistochemistry or ×40 objective for immunofluorescence. For live-cell imaging, U2OS Nup50-mCherry was plated in glass-bottom plates and transfected for 5 h before the acquisition started. Images were taken every 30 min for 15 h with the Yokogawa CSU W1 mounted with Leica DMi 8 microscope with ×63 objectives for videos or for 65 h with the Zeiss Axio Observer Z1 with ×20 objectives for cell death quantification (n = 3, >100 cells counted).
Correlative light and EM
Cells were grown to near confluency directly on carbon-coated sapphire disks (3 × 0.05 mm; engineering office M. Wohlwend GMBH). The sapphire disks were then transferred to 300-µm deep flat carriers and subjected to high-pressure freezing with the HPM10 BALTEC apparatus. Automated freeze substitution (AFS) was performed in the chamber of an AFS2 device (Leica Microsystems GmbH). The samples were kept at −90 °C in dry acetone containing 0.1% uranyl acetate for 24 h. The temperature was gradually increased to −45 °C at a rate of 5 °C per hour, followed by 5 h at −45 °C. The samples were washed with pure acetone and infiltrated with graded concentrations of Lowicryl HM20. Polymerization was achieved by ultraviolet light exposure at −25 °C for 48 h, followed by an additional 9 h at room temperature (20 °C). Ultrathin sections were cut 90 nm with Leica Ultracut, picked up on 200 mesh copper grids coated with a carbon film. Sections were viewed on a spinning-disk Yokogawa X1 microscope equipped with a Nikon TI2 microscope to locate the fluorescence signal. Then, after poststained for 10 min in 2% aqueous UA and 5 min in lead citrate, sections were viewed on a Tecnai G2-20 transmission EM operated at 120 kV, and images were acquired on a TVIPS TemCam F416 camera.
Muscle fiber segmentation
Hematoxylin and eosin staining or wheat germ agglutinin channel for fluorescence images were used to segment muscle fibers. The muscle fibers were segmented from Qupath with Cellpose. Then, the fiber classification was calculated with a Python script and reimported into Qupath to display the result.
Transgenic constructs and Drosophila strains
The uORFs of human GIPC1 and antisense RILPL1 with either 10 or 100 GGN repeats were subcloned into the attB-pUAST vector and integrated into the attP2 site of phiC31 stocks through standard microinjection procedures. Transgenic Drosophila lines expressing either GFP vector control, uGIP-GFP, uGIPpolyG-GFP, asRIL-GFP or uRILpolyG-GFP were successfully generated. All Gal4 driver lines were acquired from the Bloomington Drosophila Stock Center. The fly strains were maintained under standard conditions at 25 °C on cornmeal agar medium, with a regulated 12-h light/12-h dark cycle.
Fly EM analysis
Samples were collected and fixed in 2.5% glutaraldehyde at 4 °C overnight. Subsequently, the samples were sectioned using a Leica EM UC6/FC6 Ultramicrotome. To verify the proper orientation and quality of the sections, they were stained with toluidine blue. The selected sections were then transferred to copper grids and counterstained with uranyl acetate and lead citrate to enhance contrast. Finally, the prepared samples were imaged using an EM.
Fly climbing assay and lifespan assay
Flies were separated by sex within 24 h of hatching and transferred to fresh vials every 4 days throughout the experimental period. Vials containing 20–30 flies of the same genotype were used. For climbing assays, flies were gently tapped to the vial bottom, and the number crossing the 5 cm mark within 15 s was recorded. Each trial was repeated five times, with mean and s.e. calculated. Lifespan assays used two sex-matched and age-matched groups. Climbing ability and lifespan were assessed and recorded every 5 days.
Fly porphyrin TMPyP4 administration
Porphyrin TMPyP4 was obtained from Selleck Chemicals and stored as a 1 mM stock solution at −20 °C. For Drosophila studies, the compound was dissolved in sterile water to achieve final concentrations of 30 μM, 100 μM and 200 μM. Before each experiment, fresh dilutions were prepared in Drosophila cornmeal agar medium. The drug was administered starting from the egg stage, and adult flies were transferred to fresh vials containing porphyrin TMPyP4 every 3 days.
Quantification and statistical analysis
To eliminate bias, image or animal analyses were either completely automated or blinded. All statistical analyses were performed using Excel (Microsoft) and an online web statistical calculator (Astatsa). Experiments are represented as either mean ± s.e.m. or box-and-whisker plots with box upper and lower limits representing the 25th and 75th quartiles, respectively, and the whiskers depicting the lowest and highest data points and the horizontal line through the box represent the median. The statistical tests used are two-tailed unpaired Student t test or one-way analysis of variance with post hoc Tukey’s honestly significant difference test. Sample sizes were determined by past experiments and to minimize the number of mice used. No statistical method was used to determine whether data meet assumptions of the statistical approach. Detailed statistical information, including the statistical test, measures, number ‘n’ of animals, biological replicates and/or assays, are indicated in Figs. 1, 4, 5, 6 and 7 and their respective legends.
Additional methods
Details of RT-qPCR, mass spectrometry and RNA sequencing are presented in Supplementary Note.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at 10.1038/s41588-026-02507-z.
Supplementary information
Supplementary InformationSupplementary Notes 1–8, Supplementary Figs. 1–7, Supplementary Methods and supporting data for the Supplementary Figs. 1, 3, 4 and 7. Reporting Summary Peer Review File Supplementary DataSupplementary Data 1–5. Supplementary Video 11st example of live imaging of U2OS cells expressing Cherry-Nup50 labeling cell nuclei and OPDM4 asRILpolyG-GFP to follow intranuclear inclusions formation. Supplementary Video 22nd example of live imaging of U2OS cells expressing Cherry-Nup50 labeling cell nuclei and OPDM4 asRILpolyG-GFP to follow intranuclear inclusions formation.
Source data
Source Data Figs. 1,2, 4 and 7 and Extended Data Figs. 1, 2 and 4Unprocessed western blots and gels for Figs. 1e, 2b, 4f and 7a and Extended Data Figs. 1c,f–i, 2a–d and 4d.