GGGGCC repeat expansion in C9ORF72 compromises nucleocytoplasmic transport

Abstract

GGGGCC (G₄C₂) repeat expansion in a noncoding region of C9ORF72 is the most common cause of sporadic and familial forms of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD)^1,2. The basis for pathogenesis is unknown. To capture the consequences of G₄C₂ repeat expansion in a tractable genetic system, we generated transgenic fly lines expressing 8, 28 or 58 G₄C₂ repeat-containing transcripts that do not have a translation start site (AUG) but contain an open-reading frame for green fluorescent protein (GFP) to detect repeat-associated non-AUG (RAN) translation. These transgenic animals show dosage-dependent, repeat length-dependent degeneration in neuronal tissues and RAN translation of dipeptide repeat (DPR) proteins as observed in patients. This model was used in a large-scale, unbiased genetic screen ultimately leading to the identification of 18 genetic modifiers that encode components of the nuclear pore complex (NPC) as well as the machinery that coordinates the export of nuclear RNA and the import of nuclear proteins. Consistent with these results we found morphological abnormalities in the architecture of the nuclear envelope in cells expressing expanded G₄C₂ repeats in vitro and in vivo. Moreover, we identified a substantial defect in RNA export resulting in retention of RNA in the nuclei of Drosophila cells expressing expanded G₄C₂ repeats and also in mammalian cells, including aged iPSC-derived neurons from C9ORF72 patients. These studies show that a primary consequence of G₄C₂ repeat expansion is the compromise of nucleocytoplasmic transport through the nuclear pore, revealing a novel mechanism of neurodegeneration.

To explore pathogenic mechanisms of disease initiated by C9ORF72 repeat expansion in a genetically tractable model organism, we used PhiC31 integrase mediated insertion of 8, 28 or 58 copies of G₄C₂ repeats into specific genomic loci of Drosophila. These repeats are not expressed at baseline but are transcribed when GAL4 is introduced in trans by genetic cross (Fig. 1a). The GFP coding sequence without a start codon was placed in frame with a potential poly(GP) dipeptide repeat (Fig. 1a), although RAN translation of the sense strand may also occur in two other reading frames producing poly(GR) and poly(GA).

Figure 1. G₄C₂ repeats induces length- and dosage-dependent degeneration in Drosophila.

a, Constructs expressing 8, 28, or 58 copies of G₄C₂ repeats. b, G₄C₂-58 causes a rough eye phenotype. c, Two copies of G₄C₂-58 expressed in motor neurons results in a decrease in larval size (left panel) and locomotor activity (right panel). d, Quantification of the distance traveled by larvae expressing repeats in motor neurons. Values are mean ± s.e.m., n = 3 trials, ** p < 0.01, by One Way ANOVA, Tukey’s Post Hoc test. e, Expression of two copies of G₄C₂-58 in motor neurons reduces bouton number. Scale bar: 25 μm. f, Quantification of bouton number and muscle size in larvae expressing G₄C₂ repeats. Values are mean ± s.e.m., n = 6, ** p < 0.01, by One Way ANOVA, Tukey’s Post Hoc test.

We expressed these constructs in the Drosophila eye using GMR-GAL4 and observed a length- and dosage-dependent rough eye phenotype (Fig. 1b) similar to that reported recently^3,4. Toxicity was also observed when these repeats were expressed in other tissues. Expression of G₄C₂-58, but not G₄C₂-8, in motor neurons using OK371-GAL4 led to small larvae with significantly impaired locomotor activity (Fig. 1c, d). The neuromuscular junctions (NMJs) were examined with presynaptic and postsynaptic markers (Fig. 1e). Expression of G₄C₂-58 resulted in a significant decrease in bouton number and total muscle area when compared with GFP or G₄C₂-8 (Fig. 1f). Additionally, active zones within the NMJ were markedly reduced in larvae expressing G₄C₂-58 compared with controls (Extended Data Fig. 1a, b). Expression of G₄C₂-58 by the pan-neuronal driver elav-GAL4 also resulted in dosage-dependent locomotor defects and NMJ abnormalities (Extended Data Fig. 1c–f). Moreover, expression of G₄C₂ repeats by the muscle-specific driver MHC-GAL4 led to age-dependent defects in indirect flight muscle resulting in permanent abnormal wing position as compared to controls (Extended Data Fig. 1g).

In this fly model, RAN translation occurs in a G₄C₂ repeat length-dependent manner. Poly(GP)-GFP DPRs were detected in flies expressing G₄C₂-58 and to a lesser extent G₄C₂-28, but not in flies expressing equivalent levels of G₄C₂-8 (Extended Data Fig. 2a, b). Indeed, nuclear and cytoplasmic inclusions of poly(GP)-GFP were present in tissues expressing G₄C₂-58 but not G₄C₂-8 such as brain, muscle and salivary gland cells (Extended Data Fig. 2c–g). To determine whether DPRs are sufficient to drive a degenerative phenotype, we used codons alternate to G₄C₂ to generate new transgenic flies directly expressing poly(GA), poly(GR), or poly(GP), with an AUG start codon and N-terminal GFP (Supplementary Table 1; Extended Data Fig. 3a). Expression of neither GFP-(GP)₄₇ nor GFP-(GA)₅₀ elicited a degenerative eye phenotype (Extended Data Fig. 3a, b). Thus, while poly(GP)-GFP serves as a useful marker of cells expressing G₄C₂-58 RNA and producing DPR proteins, it is unlikely to contribute significantly to the degenerative phenotype. By contrast, expression of GFP-(GR)₅₀ was highly toxic resulting in greater than 95% lethality with a few escapers having severely degenerated eyes (Extended Data Fig. 3a). These results are consistent with recent reports indicating that poly(GR) is toxic in cultured cells and Drosophila^3,5,6. In fact, immunoblotting with antibodies against DPRs showed that flies expressing G₄C₂-58 RNA produce poly(GR) in addition to poly(GP) DPRs, although neither poly(GA) nor poly(PR) was detected (Extended Data Fig. 3c, d). Thus, G₄C₂-58 toxicity may be mediated through toxic RNA or poly(GR) produced in our fly model, or a combination of these mechanisms.

To gain unbiased insight into the pathogenic mechanisms underlying G₄C₂-58 toxicity, we performed a large-scale genetic screen to identify genetic loci whose partial loss of function may significantly modify the G₄C₂-58-induced rough eye phenotype. To this end, flies expressing G₄C₂-58 in the eye were crossed with 372 chromosomal deficiency lines spanning the entirety of the 2^nd and 3^rd chromosomes, representing ~80% of the fly genome. In total, 5.1% of the deficiencies were found to suppress the rough eye phenotype whereas 8.8% of these enhanced the rough eye phenotype (Extended Data Table 2).

Two deficiency lines on the 2^nd chromosome, Df(2R)1725 and Df(2R)1735, each in heterozygotes had no eye phenotype by themselves but led to a marked enhancement of the G₄C₂-58 rough eye phenotype (Fig. 2a). The genomic regions covered by these deficiencies partially overlap (Fig. 2b), suggesting one or more genes in this overlapping region are responsible for the observed enhancement. After systematic evaluation of this candidate interval with classical loss of function (LOF) alleles and RNAi lines, we identified the gene Nup50, whose partial LOF enhanced the G₄C₂-58 phenotype (Fig. 2c). The effect of the Nup50 was specific to G₄C₂-58 since there was no rough eye phenotype present when Nup50 was knocked down in eyes expressing GFP (Fig. 2c). Nup50 is a component of the nuclear pore and also plays a critical role in promoting protein nuclear import through interaction with Importin β and Ran GTPase⁷, Consistent with the impact of Nup50 LOF on G₄C₂-58-mediated degeneration, a dominant-negative form of Ran (Ran^T24N)⁸ strongly enhanced the G₄C₂-58 rough eye phenotype (Extended Data Fig. 4a). Moreover, the nuclear import factors Nup153 and Transportin, which work in concert with Nup50 and Ran⁷, were also identified as enhancers of the G₄C₂-58 rough eye phenotype (Extended Data Fig. 4a). These genetic analyses suggest that protein import is compromised by G₄C₂-58 expression.

Figure 2. Genetic screen identifies multiple modifiers of G₄C₂-58 toxicity in the nucleocytoplasmic transport pathway.

a, G₄C₂-58 expression driven by GMR-GAL4 causes a rough eye phenotype (top right) that was enhanced by either Df(2R)1725/+ or Df(2R)1735/+(middle). b, Overlapping genomic region of deficiency lines Df(2R)1725 or Df(2R)1735. c, RNAi knockdown (Nup50^20824/GD and Nup50^100564/KK) and a genetic allele (Nup50^KG09557) identify Nup50 as an enhancer. d, The deficiency Df(3R)Antp17 (green) but not others (grey) suppressed the G₄C₂-58 eye phenotype. e and f, Identification of Ref as a suppressor. g, Table summarizing modifier genes and their known functions. h, Suppressors (green) and enhancers (red) of G₄C₂-58 toxicity in the nucleocytoplasmic trafficking pathway.

The strongest suppressor identified in the genetic screen was Ref1, revealed by interrogation of the suppressor deficiency Df(3R)Antp17 (Fig. 2d, e). This deficiency also suppressed the strong rough eye phenotype produced by two copies of G₄C₂-58, as did heterozygosity for the null allele Ref1⁰²²⁶⁷ (Fig. 2f). Ref1, and its human ortholog ALYREF, are RNA-binding proteins that associate with the 5′ end of mRNAs to prevent degradation by the nuclear exosome and, together with the TREX component CHTOP, facilitate delivery of fully processed mRNAs to the nuclear pore receptor NXF1 that mediates their export through the nuclear pore^9–12. Consistent with the impact of Ref1 LOF on G₄C₂-58-mediated degeneration, partial LOF of NXF1 or CHTOP enhanced the rough eye phenotype (Fig. 2g, h, Extended Data Fig. 5). The fact that the G₄C₂-58 phenotype is exacerbated by LOF in NXF1 or CHTOP, but partially rescued by LOF of Ref1 perhaps highlights the dual role played by Ref1 in partitioning target RNAs between the nuclear pore and the nuclear exosome⁹. Indeed, in support of this interpretation we identified LOF in EXOSC3 and EXOSC10, which encode components of the nuclear exosome, as strong enhancers of G₄C₂-58-related toxicity (Fig. 2g, h, Extended Data Fig. 5). Notably, LOF mutations in EXOSC3 in humans cause a congenital form of motor neuron disease¹³.

The genes encoding cap-binding proteins NCBP1, NCBP2, and ARS2, that mediate recruitment of the TREX complex to the 5′ end of RNA to initiate RNA export¹⁴, were also identified as enhancers of G₄C₂-58-mediated degeneration (Fig. 2g, h, Extended Data Fig. 5). Moreover, we found that LOF of either Nup107 or its binding partner Nup160, both of which are nuclear pore components responsible for exporting some RNAs^15–17, suppressed the G₄C₂-58 rough eye phenotype (Extended Data Fig. 5). One particularly notable enhancer was the Drosophila ortholog of human GLE1, a critical mediator of RNA export at the nuclear pore¹⁸ (Fig. 2g, h, Extended Data Fig. 5). Interestingly, complete LOF of human GLE1 causes congenital motor neuron disease¹⁹, whereas partial LOF of human GLE1 is associated with adult-onset ALS²⁰.

The identification of TREX and nuclear pore proteins as genetic modifiers strongly implies that not only is nuclear import compromised by G₄C₂-58 expression, but also nuclear export of RNAs and proteins. Consistent with this latter notion we found that LOF of Crm1, a gene encoding a major receptor for the export of RNAs and proteins, resulted in significant enhancement of the G₄C₂-58 phenotype (Extended Data Fig. 4b). Moreover, expression of G₄C₂-58 in motor neurons caused exquisite sensitivity to leptomycin B, an inhibitor of nuclear export (Extended Data Fig. 4c). In total, the genetic screen and follow up analyses identified 18 modifier genes within the pathway of nucleocytoplasmic transport, and RNA export in particular, highlighting this system as an important target of G₄C₂-58-related toxicity (Fig. 2g, h, Supplementary Table 3). In particular, the identification of 4 strong suppressors of G₄C₂-58 indicates that compromised nucleocytoplasmic transport is an important causal pathway responsible for degeneration.

In follow up to our genetic studies, RFP-Nup107 was expressed in salivary glands using Fhk-GAL4. In cells expressing G₄C₂-8, Nup107 labels a distinct nuclear boundary that is morphologically indistinguishable from cells expressing GFP (Fig. 3a). In contrast, in cells expressing G₄C₂-58, the nuclear envelope exhibited a wrinkled appearance and in many nuclei, Nup107 was found to form inclusions near the nuclear envelope (Fig. 3a). To further explore nuclear architecture, the nuclear envelope was visualized by immunostaining of endogenous Lamin C. 43.7% of cells expressing G₄C₂-58 showed an abnormal, “frayed” nuclear envelope phenotype, whereas this phenotype was observed in only 7.1% of cells expressing G₄C₂-8 (Fig. 3b, c, Extended Data Fig. 6a). Thus, G₄C₂-58 expression causes defects in the architecture of the nuclear envelope and mislocalization of nucleoporins. Moreover, cells expressing one copy of G₄C₂-58 showed a significant increase in the ratio of nuclear to cytoplasmic RNA in comparison to cells expressing GFP or G₄C₂-8 (Fig. 3d, e). Expression of 2 copies of G₄C₂-58 resulted in a more pronounced retention of nuclear RNA (Fig. 3d, e). Importantly, depletion of Ref1 by RNAi partially suppressed the nuclear RNA retention phenotype in cells expressing G₄C₂-58, reducing nuclear RNA density to control levels (Fig. 3d, e). Thus, G₄C₂ length-dependent degeneration in Drosophila is accompanied by nuclear retention of RNA, consistent with the results of the unbiased genetic screen.

Figure 3. Drosophila salivary gland cells expressing G₄C₂-58 exhibit nuclear envelope abnormalities and accumulation of nuclear RNA.

a, The effect of G₄C₂-58 expression on Nup107 localization. Scale bar: 50 μm. b, G₄C₂-58 expression causes abnormal nuclear envelope morphology. Scale bar: 50 μm. c, Quantification of the frayed nuclear envelope phenotype in cells expressing either G₄C₂-8 (n = 251) or G₄C₂-58 (n = 127). Values are mean ± s.e.m., ** p < 0.001 by Student’s t test. d, Accumulation of total nuclear RNA relative to cytoplasmic RNA (red). Knockdown of Ref1 led to a partial rescue of nuclear RNA accumulation. Scale bar: 25 μm. e and f, Quantification of relative intensity of total nuclear RNA vs. cytoplasmic RNA. Values are mean ± s.e.m., n = 30 cells, * p < 0.01, ** p < 0.001 by One Way ANOVA, Tukey’s Post Hoc test.

Similar retention of nuclear RNA was observed in HeLa and HEK293T cells upon transient expression of G₄C₂-58, but not G₄C₂-8 (Fig. 4a–e, Extended Data Fig. 6b). Specifically, this export defect was illustrated using an alternate approach to monitor the fate of newly synthesized RNA by metabolic labeling. In HeLa cells expressing G₄C₂-58, we observed a significant decrease in the export of nascent RNA in comparison to control cells expressing GFP or G₄C₂-8, although the total levels of metabolically labeled transcripts were comparable (Fig. 4a–c, Extended Data Fig. 6c). Moreover, polyA⁺ mRNA as detected by an oligo-dT probe accounts for at least some portion of the retained RNA in mammalian cells expressing G₄C₂-58 (Fig. 4d, e).

Figure 4. Accumulation of nuclear RNA in human cells expressing G₄C₂ repeat expansion.

a, Reduced accumulation of newly synthesized RNA in the cytoplasm of cells transfected with G₄C₂-58 (white arrows) (imaged at 60 min). Yellow arrows: untransfected cells. b, Quantification of cytoplasmic RNA intensity following metabolic labeling of newly synthesized RNA (0 min: n > 24, 30 min: n > 7, 60 min: n > 25). ** p < 0.01, *** p < 0.001 by One Way ANOVA, Tukey’s Post Hoc test. c, Total RNA intensity following labeling of newly synthesized RNA at time point 0 min. d, PolyA⁺ RNA was measured by FISH in transiently transfected HeLa cells. Scale bar: 50 μm. e, Quantification of cells containing polyA⁺ RNA puncta, n = 30 cells per condition. f, Total cellular RNA in 2-month old iPSCs-derived cortical neurons. Scale bar: 25 μm. g, Quantification of the cytoplasmic RNA ratio in patient cortical neurons vs. controls. Values are mean ± s.e.m., n > 15 cells per line, ** p < 0.01 by Student’s t test.

Advances in induced pluripotent stem (iPS) cell technology permit the generation of neurons from patients including those with FTD and ALS²¹. To further examine the impact of expanded G₄C₂ toxicity on nucleocytoplasmic transport in a more relevant human cell type from C9ORF72 patients, we examined the subcellular distribution of total RNA in 2-month-old cortical neurons, which are affected in FTD, a frequent clinical feature of C9ORF72-related disease²². These neurons were differentiated from previously characterized^23–25 as well as newly generated and characterized iPS cell lines derived from 5 C9ORF72 patients and 3 controls (Extended Data Figs. 7 and 8, Supplementary Tables 4 and 5). Consistent with our observations in fly, HEK293T and HeLa cells, this analysis confirmed a significant increase in the nuclear to cytoplasmic ratio of RNA in C9ORF72 patient cells in comparison to controls (Fig. 4f, g). On average, C9ORF72 neurons showed a 35% increase in the nuclear to cytoplasmic ratio of RNA density (Fig. 4g). Interestingly, this difference in the ratio of nuclear to cytoplasmic RNA was not observed in fibroblasts obtained from C9ORF72 patients when compared to controls, consistent with low levels of C9ORF72 expression in fibroblasts²⁵ (Extended Data Fig. 9a–c). Thus, neurons derived from patients with C9ORF72-related disease mirror the nucleocytoplasmic transport defects identified by unbiased approaches in a genetic model of disease.

The NPC is the largest macromolecular complex in eukaryotic cells, consisting of multiple copies of more than 30 different proteins²⁶. Some structural components of the NPC have an exceedingly long half-life measured in years²⁷. Indeed, nuclear pores are a particularly intriguing target for age-related diseases affecting post-mitotic cells such as neurons, putting them at risk of accumulating damage over extended periods of time²⁷. Nucleocytoplasmic transport through the nuclear pore is vital to cell viability, and primary defects in this function cause of a host of human diseases of varying phenotypes ranging from cancer to neurological disease^26,28,29.

Important questions regarding the pathogenesis of C9ORF72-related disease remain; in particular the relative contribution of expanded G₄C₂ RNA vs. DPRs or other mechanisms, which could cause the nuclear pore defect. Also to be determined is whether defects in nucleocytoplasmic transport and other reported defects such as sequestration of specific RNA binding proteins, nucleolar stress and ER stress are parallel processes or directly connected with each other. The rapid pace of discovery in this area of disease biology suggests that answers to these questions will come soon.

Methods

Generation of Drosophila lines

To generate transgenic Drosophila expressing G₄C₂-8, G₄C₂-28 and G₄C₂-58, G₄C₂ repeats of the respective length were cloned downstream of the UAS promoter and upstream of the EGFP sequence with the start codon removed in the plasmid pUAST-ATTB. Transgenic Drosophila lines were generated by BestGene Inc. such that the transgene was inserted using the PhiC31 integrase into either the attP2 site on chromosome 3 (loci 68A4) or the attP40 site on chromosome 2 (loci 25C6). GFP-(GP)₄₇, GFP-(GA)₅₀, and GFP-(GR)₅₀ plasmids³⁰ were subcloned into the pUAST-ATTB plasmid and inserted into the attP2 site by BestGene Inc. Flies were raised at 25°C on a standard diet. A complete listing of Drosophila stocks that modify the G₄C₂-58 phenotype are listed in Supplementary Table 3. For genetic interaction studies, the combined stock (GMR-GAL4/Cyo; UAS-G₄C₂-58/Tm6,Tb) was crossed with deficiency stocks or individual mutants and phenotypic analysis was performed on flies aged 24–48 hours in both males and females. Knock down efficiency of RNAi lines was measured by quantitative RT-PCR (Extended Data Fig. 10, Supplementary Table 6). For nuclear envelope morphology studies, the recombined stocks (Fkh-GAL4, UAS-GFP/TM6,Tb; Fkh-GAL4, UAS-G₄C₂-8/TM6,Tb; and Fkh-GAL4,UAS-G₄C₂-58/TM6,Tb) were crossed with UAS-mRFP-Nup107 transgenic flies or immunostained for endogenous Lamin C.

Drosophila stocks

Deficiency stocks were obtained from the Bloomington Drosophila Stock Center deficiency kit. RNAi lines were obtained from either the Bloomington Drosophila Stock Center or the Vienna Drosophila RNAi Center. UAS-RAN^T24N and Fkh-Gal4 flies were kindly provided by Drs. K. S. McKim and E. Baehrecke respectively.

Drosophila eye, muscle and neuronal phenotype analysis

Phenotypic analysis of G₄C₂ repeat expression in the Drosophila eye, muscle, and salivary gland was assessed by crossing G₄C₂-8, G₄C₂-28 and G₄C₂-58 lines to GMR-GAL4, MHC-GAL4, and FKH-GAL4, respectively. Neuronal expression was achieved by crossing G₄C₂-8, G₄C₂-28 and G₄C₂-58 lines to elav-GAL4 (pan neuronal driver) or OK371-GAL4 (motor neuron driver). Eye phenotypes were imaged by light microscopy and muscle phenotypes were visually assessed by wing posture (n=30 controls, 21 G₄C₂-8, 50 G₄C₂-28 and 24 G₄C₂-58). Neuromuscular synaptic bouton number, active zone density and crawling ability were measured as previously described^31,32 with slight modification. To count type 1b synaptic bouton number, each genotype was double stained presynaptically with anti-HRP-Cy3 (1:200, Jackson Immunoresearch) and postsynaptically with anti-Disc large 1 (1:50, DSHB). Synaptic boutons of muscle 4 in abdominal segment 2, 3 and 4 (A2–A4) were imaged with a Marianas spinning disc microscope and maximum projection images were used to count synaptic boutons. Active zone area and presynaptic area stained by anti-Bruchpilot (NC82, 1:100, DSHB) and anti-HRP-Cy3 were measured with ImageJ and the ratio of active zone area/presynaptic area was calculated. Muscle 4 in abdominal segment 2, 3 and 4 (A2–A4) was analyzed. To examine crawling ability, 4–7 wandering 3^rd instar larvae for each group were collected, washed and placed onto a 3% agarose gel in a 10 cm dish. After 5 min acclimation, larval crawling behavior was recorded by a digital camera for 30 sec (15 fps). Each group was tested three times. Moving distances of each larva were manually measured with ImageJ.

Immunofluorescence analysis of salivary gland cells

Dissected salivary glands from wandering third instar larvae were fixed with 4% paraformaldehyde (PFA) at room temperature for 15 min, washed three times with phosphate-buffered saline (PBT, 0.1% Triton X-100 in PBS), and blocked for 1 hr at room temperature with 10% normal goat serum (Sigma) diluted in PBT. Tissues were then incubated overnight at 4°C with mouse anti-Lamin C (1:30, LC28.26; Developmental Studies Hybridoma Bank (DSHB)). After three washes in PBT (10 min each), Tissues were incubated for 1 hr at room temperature with secondary antibodies (goat anti-mouse Alexa Fluor 568, 1:200; Invitrogen) diluted in the blocking solution. Tissues were washed three times in PBS and mounted with VECTASHIELD (Vector Laboratories). Fluorescence signals were examined with an Olympus IX70 Microscope or Leica TCS SP5 II laser scanning confocal microscope.

Visualization of RAN products in muscle cells

To visualize poly(GP)-GFP in the thorax expressing G₄C₂-58, flies were cleared overnight using a modified ScaleA2 (2 M urea, 10% glycerol, 0.1% Triton-X). RAN peptides were visualized using light sheet microscopy. To visualize RAN products at high magnification, thoraxes were dissected and stained with the same immunofluorescence protocol described above for salivary glands but using anti-Lamin (1:100, ADL 67.10; Developmental Studies Hybridoma Bank (DSHB)) to visualize the nucleus and phalloidin (1:40, Life Technologies) to visualize the muscle structure.

Immunoblots

Adult flies were frozen with dry ice and vortexed to remove the heads or thorax. Samples from each genotype were homogenized in RIPA buffer with proteinase inhibitor cocktail added. Urea buffer (7 M urea, 2 M thiourea, 4% CHAPS, 30 mM Tris pH8.5) was used when indicated to visualize the non-RIPA soluble protein fraction. Sample was mixed with 4X sample buffer (1 M Tris-HCl [pH 6.8], 8% SDS, 40% glycerol, 0.1% bromophenol blue) and boiled for 5 min, separated on a SDS gel and transferred to a membrane. The membrane was blocked, probed with primary antibody, and incubated with secondary antibody. The signal was visualized with either chemiluminescent substrate (SuperSignal West Pico; Pierce) or by using an Odyssey Fc (Li-Cor). Primary western blot antibodies were anti-GFP (AB3080, Millipore or SC-9996 Santa Cruz Biotechnology), anti-poly(GP), poly(GA) or poly(GR) antibodies^33,34, anti-β-actin antibody (4967; Cell Signaling Technology) or anti-actin antibody (sc-1616, Santa Cruz Biotechnology).

RNA in situ hybridization

Fixation of dissected salivary glands, cortical neurons, HeLa, or 293T cells (obtained from ATCC) was performed using 1% paraformaldehyde on ice for 5 min, followed by a second fixation in 1% paraformaldehyde plus 0.05% NP40 for an additional 5 min. Samples were then transferred to 70% ethanol and stored at −20°C until needed. The probes for global RNA in either Drosophila or human cells were prepared from genomic DNA derived from w¹¹¹⁸ flies or 293T cells. Both probes were labeled by nick translation using either alexafluor 594 dUTP. To hybridize the probes, 70 ng of the labeled probes (the human probe was combined with 2 μg of human cot1 DNA) were suspended in 10 μl of hybridization buffer consisting of 50% formamide, 2X SSC, and 10% dextran sulfate. Fixed cells were dehydrated in 70%, 80%, and 100% ethanol for 2 min each prior to hybridization. Probes suspended in hybridization buffer were denatured at 70°C for 5 min and then applied to the dehydrated cells and hybridized at 37°C overnight. Samples were washed following hybridization in 50% formamide, 2X SSC at 37°C for 5 min, then briefly rinsed in room temperature PBS and mounted on slides with DAPI. To detect poly(A) mRNA, HeLa cells were fixed with 4% paraformaldehyde followed by ice cold methanol and 70% ethanol for 10min each. The cells were treated with 1 ng/ l μL of 5′-labeled Cy3-oligo-dT(20) at 37°C for 1 h. Samples were washed in room temperature PBS and mounted on slides with DAPI. Images were captured using either widefield fluorescence microscopy or a Marianas spinning disc microscope. An extended depth of focus function was used to combine all components of individual microscope fields.

Pulse-chase of newly synthesized RNA

HeLa cells were transfected with plasmids expressing GFP, G₄C₂-8 or G₄C₂-58 that had been subcloned into pcDNA 3.1 using Fugene HD (Promega). After 48 h incubation, cells were treated with to 1 mM 5-ethynyl uridine (EU) was for 1 h. RNA was then visualized at indicated time points using the Click-iT detection kit (Invitrogen) as recommended by the manufacturer. To visualize GFP signal, cells were co-stained using monoclonal anti-GFP (SC-9996, Santa Cruz Biotechnology) as a primary antibody.

iPSC lines

iPSC lines from 2 control subjects and 4 G₄C₂ repeat expansion carriers have been previously characterized^23–25,35 (please see Extended Data Table 4). Generation of integration-free induced pluripotent stem cells (iPSCs) from fibroblasts of one G₄C₂ expansion carrier and one control subject, using episomal plasmids with the reprogramming factors Oct4, Sox2, Klf4, Lin28 and L-Myc were performed as described before³⁶, which was approved by the University of Massachusetts Medical School Institutional Biosafety Committee. The use of human fibroblasts was approved by the UCSF Institutional Review Board and informed consent was obtained from all subjects. After reprogramming, characterization of iPSCs shows that cells have normal karyotype, express pluripotent markers and have the capacity to differentiate into cells of the three germ layers as described in Almeida et al., (2012, 2013)^23,35. All the iPSC lines used here were tested regularly with no mycoplasma contamination.

Cortical neuronal cultures

4 iPSC lines from 3 control subjects and 6 iPSC lines from 5 C9ORF72 carriers were differentiated to cortical neurons as described earlier^23,35 (please also see Extended Data Table 5). Neuronal cultures were aged for 8-weeks before fixation for RNA FISH.

Immunostaining for iPSCs and iPSCs-derived neurons

For immunostaining cells were fixed in 4% paraformaldehyde for 15 minutes and then permeabilized with 0.3 % Triton X-100 for 5 minutes. Cells were blocked with 5% bovine serum albumin for 30 minutes; cells were incubated with primary antibodies overnight at 4 °C. The following primary antibodies were used: mouse anti-OCT4 1:100 (Santa Cruz Biotechnology), goat anti-NANOG 1:100 (R&D Systems), mouse anti-SSEA4 1:100 (Abcam), rabbit anti-desmin 1:100 (Thermo Scientific), mouse anti-βIII-tubulin 1:500 (Promega), mouse anti-α-fetoprotein 1:200 (R&D Systems), mouse anti-MAP2 1:500 (Sigma), goat anti ChAT 1:200 (EMD Millipore), rabbit anti-VGLUT1 1:500 (Synaptic Systems). After incubation with primary antibodies cells were washed with PBS three times and incubated with Alexa Fluor secondary antibodies 1:500 (Invitrogen) for 1 h at room temperature followed by counter staining with DAPI.

Fluorescence in situ hybridization

FISH in iPSCs derived from control and C9ORF72 expansion carrier using a Cy3-conjugated (GGCCCC)₄ probe was performed as described in Almeida et al., (2013)²³.

Quantification

For salivary glands, 30 cells from at least 5 individual salivary glands for each genotype were used for quantitative analysis. ImageJ software was used to measure the ratio of nuclear to cytoplasmic RNA by comparing the total RNA measured in the nucleus to that of the cytoplasm. DAPI was used to mark the nucleus. For cortical neurons, the ratio of nuclear to cytoplasmic RNA density was calculated by measuring the density of the RNA signal in both the nucleus and cell body using ImageJ software. DAPI was used to mark the nuclear boundary of neurons. For cortical neurons, at least 15 neurons were measured from each individual patient (Controls 1–3 (n=16, 16, 31) and Patients 1–5 (n=16,16,16,30,15)). Error bars for all quantification are standard error.

Extended Data

Extended Data Figure 1. Expression of G₄C₂ repeats induces length-dependent phenotypes in Drosophila.

a, G₄C₂ Drosophila motor neurons using the OK371-GAL4 driver leads to a significant reduction in active zones as immunostained by the anti-Bruchpilot antibody NC82 and anti-HRP. Scale bar: 50 μm. b, Quantification of active zones. Values are mean ± s.e.m. ** p < 0.01, One way ANOVA, Tukey’s Post Hoc test. c, Pan Neuronal expression of G₄C₂-58 repeats induces dosage-dependent decrease in larval size (left) and locomotor activity measured in 30 seconds (right) when G₄C₂-58 is expressed in all neurons using the elav-GAL4 driver. d, Quantification of the distance traveled by third instar larvae reveals expressing two copies of G₄C₂-58 results in a significant deficit in locomotor activity. Values are mean ± s.e.m. n=3 trials, more than 4 larvae per group in each trial. ** p < 0.01, One way ANOVA, Tukey’s Post Hoc test. e, Pan Neuronal expression of G₄C₂-58 repeats in Drosophila neurons using the elav-GAL4 driver leads to a significant reduction in the bouton number. Bouton number was quantified by examining the presynaptic (anti-HRP) and postsynaptic (anti-DLG1) markers (left). Scale bar: 50 μm. f, Quantification of buoton number (left) and muscle size (right) reveal both are significantly reduced in Drosophila larvae expressing G₄C₂-58 repeats. Values are mean ± s.e.m. n ≥ 5, ** p < 0.01, One way ANOVA, Tukey’s Post Hoc test. g, Expression of G₄C₂-28 and G₄C₂-58 but not G₄C₂-8 in the muscle using the MHC-GAL4 driver leads to loss of wing control in adult flies (n > 20 for all genotypes). This phenotype was assessed by examining the permanent wing posture of live adult flies.

Extended Data Figure 2. RAN translation is observed in the Drosophila expressing G₄C₂ repeats.

a, Western blot revealing translation of RAN poly-dipeptides in flies expressing G₄C₂-58 in the eye. RAN poly-dipeptides were not found in flies expressing G₄C₂-8 or control flies. There was minimal expression of GFP positive product observed in flies expressing G₄C₂-28. GFP expressing flies (lane 1) were used as a positive control for the anti-GFP antibody. b, Western blot showing production of RAN product when G₄C₂-28 and G₄C₂-58 but not G₄C₂-8 repeats are expressed in the muscle. RAN products were visualized with anti-GFP antibody (left) and anti-poly(GP) antibody (right). c, The RAN product poly-GP-GFP from flies expressing G₄C₂-58 in the muscle form large visible inclusions as visualized under light sheet fluorescent microscopy (left) and by confocal microscopy (right). Scale bar: 50 μm. d, Expression of G₄C₂-58 in the salivary gland cells results in the formation of large nuclear inclusions and smaller cytoplasmic inclusions. Scale bar 50 μm. e–f, Expression of G₄C₂-58 in the ventral ganglion by OK371 driver results in the formation of nuclear and cytoplasmic inclusions, whereas GFP shows diffused nuclear and cytoplasmic localization. Lamin staining shows nuclear membrane, and CD8-RFP shows plasma membrane. Scale bars: 25 μm. g, Expression of G₄C₂-58 in pan neuronal cells by elav driver results in the nuclear and cytoplasmic inclusions. Scale bar: 25 μm.

Extended Data Figure 3. Ectopic expression of poly(GR) but not poly(GA) or poly(GP) peptides are toxic in Drosophila.

a, Transgenic Drosophila were generated that express ATG driven poly(GA), poly(GR) and poly(GP) peptides with an N-terminal GFP tag (top). Expression of GFP-(GA)50 and GFP-(GP)47 were nontoxic when expressed in the eye with GMR-GAL4 whereas GFP-(GR)50 expression resulted in >95% lethality with surviving adults having severely degenerated eyes (bottom). b, Western blot showing the expression of G₄C₂-58, GFP-(GA)50, GFP-(GR)50 and GFP-(GP)47 as visualized in muscle by anti-GFP antibody. c, Western blot showing the expression of poly(GP) in muscle of flies expressing G₄C₂-58 and GFP-(GP)47 but not GFP-(GA)50, GFP-(GR)50 and control flies as visualized by anti-GP antibody. d, Dot blot analysis of RAN peptides in muscle revealing expression of poly(GA) only in GFP-(GA)50 flies, expression of poly(GR) in G₄C₂-58 and GFP-(GR)50 flies. As expected, anti-sense DPR poly(PR) was not found in any of the lysates. The background protein signal was used as a loading control.

Extended Data Figure 4. Nuclear import and export is altered by G₄C₂-58 expression.

a, A threonine to asparagine substitution at residue 24 in the Ran protein abolishes the affinity for GTP and reduces its affinity for GDP. Hence, the Ran (T24N) is always in either a nucleotide free state or in its inactive, GDP-bound state, and acts as dominant negative. RANT24N expression driven by GMR-GAL4 causes a mild eye phenotype when expressed in the absence of G₄C₂-58 (upper row, right panel). The G₄C₂-58 rough eye phenotype is strongly enhanced by dominant negative RANT24N expression (middle row, left panel). The G₄C₂-58 eye phenotype is significantly enhanced by knockdown of Nup153 by two independent RNAi lines (middle row, two right panels). The G₄C₂-58 eye phenotype is also mildly enhanced by knockdown of transportin (Trn) (bottom row). b, Knockdown of Crm1 in flies expressing G₄C₂-58 induces a mild enhancement of the G₄C₂-58 eye phenotype (left vs. middle). Crm1 knockdown in the absence of G₄C₂-58 repeats does not produce a rough eye phenotype (left). c, Expression of two copies of G₄C₂-58 in the Drosophila motor neurons leads to reduced viability (50%). Chemical inhibition of Crm1 with Leptomycin B (500 nM) enhances G₄C₂-58 toxicity resulting in reduced viability (23%). Leptomycin B does not impede viability (100%) in Drosophila expressing GFP.

Extended Data Figure 5. Phenotypes of additional suppressors and enhancers of G₄C₂-58.

a, Phenotypes demonstrating suppression of the G₄C₂-58 rough eye phenotype by RNAi knock down of identified genes. b, Phenotypes demonstrating enhancement of the G₄C₂-58 rough eye phenotype by RNAi knock down of identified genes. c, Knock down of identified modifier genes shows little or no phenotype in the absence of G₄C₂ repeat expression.

Extended Data Figure 6. Impairment of Nucleocytoplasmic shuttling in Drosophila and cultured human cell lines.

a, G₄C₂-58 expression driven by FKH-GAL4 causes an abnormal nuclear envelope as shown by Lamin C staining (bottom) in comparison to G₄C₂-8 (top). Scale bar: 10 μm. b, Transfection of 293T cells with G₄C₂-58 (bottom) but not G₄C₂-8 (top) leads to an increase in nuclear RNA puncta as visualized with a total RNA FISH probe. Non-transefected cells (absence of GFP signal) do not show an increase in nuclear RNA in either G₄C₂-8 or G₄C₂-58 transfected cell. Scale bars: 25 μm. c, Enlarged images showing slowed accumulation of newly synthesized RNA in the cytoplasm of HeLa cells expressing G₄C₂-58. Scale bar: 25 μm.

Extended Data Figure 7. Characterization of newly generated integration-free iPSCs lines.

a, iPSC lines from a control subject (Line 11) and a G₄C₂ repeat expansion carrier (Line 3 and Line 8) express pluripotent markers SSEA-4, Nanog and Oct-4. Bars: 50 μm. b, qRT-PCR analysis ofexpression levels of pluripotent stem cell markers SOX2 and Nanog in these iPSCs lines showing no statistical differences between these lines and human embryonic stem cell line H9. c, After differentiation into cortical neurons about 90% of cells in these cultures are MAP2-positive neurons. d, Quantification of average percentage of MAP2-positive neurons and there is no difference between control and C9ORF72 cultures. e, Quantification of average percentage of VGLUT-positive excitatory neurons among all neurons and there is no difference between control and C9ORF72 cultures. Scale bars in all panels: 50 mm.

Extended Data Figure 8. Karyotyping analysis and pluripotency of newly generated iPSC lines.

a, G-band staining showing a normal karyotype for all the lines analyzed. b, After in vitro spontaneous differentiation of control and C9 carrier iPSC lines, cells were staining for α-fetoprotein (AFP, endoderm), desmin (mesoderm), βIII-tubulin (ectoderm), and Hoechst (nuclei). All lines showed differentiation towards derivates of three germ layers. Scale bars: 20 mm.

Extended Data Figure 9. Accumulation of nuclear RNA is not seen in fibroblasts derived from patients with G₄C₂ repeat expansion.

a and b, Total cellular RNA was measured by FISH in fibroblasts derived from either control (a) subjects or patients (b) with G₄C₂ repeat expansion. Scale bars: 25 μm. c, Quantification shows no statistical difference in the observed nuclear to cytoplasmic RNA ratio in patient vs. control fibroblasts. n=16 for all lines.

Extended Data Figure 10. qRT-PCR Analysis.

a–f. qRT-PCR analysis demonstrating knockdown of selected modifiers in Drosophila eyes. mRNA levels of selected modifier (a, d), GAL4 (b, e), and G₄C₂-58 (c, f) assayed by qRT-PCR in progeny resulting from wild type (w1118), classical mutant allele or UAS-RNAi lines of selected modifiers mated with either GMR-GAL4 or GMR-GAL4/Cyo;UAS-G₄C₂-58-GFP/TM6 to induce knockdown of the selected gene. RNA was obtained from whole Drosophila head lysates. Gene expression levels are mean ± S.D., *p < 0.05, **p<0.01, ***p<0.001 by One Way ANOVA, Tukey’s Post Hoc test. g–h. qRT-PCR analysis demonstrating knockdown of Ref1 and nup50 in Salivary Gland. mRNA levels of selected modifier (left), GAL4 (middle), and G₄C₂-58 (right) assayed by qRT-PCR in progeny resulting from either P(PZ)Ref102267 (g), or nup5020824/GD (h) mated with FKH-GAL4,UAS-G₄C₂-58-GFP/TM6. RNA was obtained from salivary gland lysates. Gene expression levels are mean ± S.D., *p < 0.05 by Student’s t-test.