The expanding biology of the C9orf72 nucleotide repeat expansion in neurodegenerative disease

Abstract

A nucleotide repeat expansion (NRE) within the chromosome 9 open reading frame 72 (C9orf72) gene was the first of this type of mutation to be linked to multiple neurological conditions, including amyotrophic lateral sclerosis and frontotemporal dementia. The pathogenic mechanisms through which the C9orf72 NRE contributes to these disorders include loss of C9orf72 function and gain-of-function mechanisms of C9orf72 driven by toxic RNA and protein species encoded by the NRE. These mechanisms have been linked to several cellular defects — including nucleocytoplasmic trafficking deficits and nuclear stress — that have been observed in both patients and animal models.

A GGGGCC (G₄C₂) nucleotide repeat expansion (NRE) in the first intron of the chromosome 9 open reading frame 72 (C9orf72) gene represents the most common genetic abnormality in familial amyotrophic lateral sclerosis (ALS) and in frontotemporal dementia (FTD)^1,2; this NRE has also been detected in 5–20% of patients with sporadic ALS³ (BOX 1). The actual size of the NRE varies dramatically among patients with ALS and in patients with FTD carrying the mutation (here termed C9 ALS/FTD). The NRE can contain as many as hundreds or thousands of repeats; however, the presence of less than 30 repeats is generally not associated with disease. To date, the size of the expansion does not appear to be related to differences in disease parameters, such as severity of disease or age of disease onset. Because this mutation is so commonly detected in both ALS and FTD and in both familial presentations and apparent sporadic presentations of these disorders, understanding the underlying pathophysiology that it causes is essential for both disease management and the development of future biomarkers and drug therapies.

Box 1 ∣. Origins of the C9orf72 nucleotide repeat expansion.

The chromosome 9 open reading frame 72 (C9orf72) nucleotide repeat expansion (NRE) was first discovered through genomic sequencing of families with high penetrance of familial amyotrophic lateral sclerosis (ALS) or familial frontotemporal dementia (FTD)^1,2. Geographically separated and apparently unrelated families harbouring a C9orf72 mutation showed a shared haplotype, suggesting that this mutation arose from a single founder¹. However, the origin of the C9orf72 NRE has been reconsidered following the identification of patients carrying C9orf72 mutations with no apparent family history of ALS, FTD or the shared founder haplotype⁹⁷. Pre-mutation carriers (asymptomatic carriers who have relatively larger numbers (≥30) of repeats than the standard population) can produce symptomatic offspring that carry grossly expanded C9orf72 nucleotide repeats⁹⁷. This large expansion of repeats is proposed to occur either during parent–offspring transmission or in the early stages of embryonic development^98,99. Multiple lines of evidence indicate that there is a range of C9orf72 repeat sizes in both asymptomatic and symptomatic individuals, showing that C9orf72 repeats are intrinsically unstable¹⁰⁰. Therefore, it is possible that large spontaneous expansions may be inherited by offspring when the repeat sizes present in their asymptomatic parent approach an instability threshold.

Even a relatively small number of repeats (~70) can be symptomatic²⁰, and asymptomatic carriers with few repeats (~30) do show C9orf72-related pathology (but not TAR DNA-binding protein 43 (TDP43) mislocalization)⁵¹. Several studies of in vitro C9orf72 NRE models have suggested that phenotypes depend on the repeat size, and it was recently suggested that the age of onset is also dependent on the repeat size¹⁰¹; however, this effect may be a result of the actions of additional genetic risk factors or epigenetic differences⁹⁷. For example, single-nucleotide polymorphisms in the transmembrane protein 106B (TMEM106B) gene were shown to be neuroprotective and associated with later disease onset in patients with FTD (but not ALS) who are carriers of the C9orf72 NRE^102,103. Future genetic and epigenetic analyses of C9orf72 ALS and C9orf72 FTD carriers and their families will be essential.

C9orf72 encodes a protein that can be expressed as two (long and short) isoforms (FIG. 1). Initial protein bioinformatics approaches had indicated that C9orf72 lacked structural homology with any known proteins; however, recent analyses have suggested that C9orf72 contains a DENN (differentially expressed in normal and neoplastic cells) domain^4,5, which may allow it to function as a GDP–GTP nucleotide exchange factor (GEF)^6,7. Experiments in transgenic mice and zebrafish have demonstrated that C9orf72 is expressed at high levels in neurons^8,9, and the human proteome map indicates that C9orf72 is highly expressed in the brain¹⁰. A recent immunocytochemistry study suggests that the short isoform of C9orf72 localizes to the nuclear membrane, whereas the long isoform is predominantly cytoplasmic¹¹. The biological significance of this subcellular distribution is unclear, particularly as the cellular function of C9orf72 remains largely unknown. However, there is evidence to indicate that C9orf72 may participate in cellular trafficking¹².

Figure 1 ∣. The C9orf72 nucleotide repeat expansion.

This schematic illustrates the chromosome 9 open reading frame (C9orf72) gene, showing the site of the nucleotide repeat expansion (NRE; red), and three annotated transcript variants. Transcripts V2 and V3 encode the long form of the C9orf72 protein, whereas transcript V1 encodes the short form. Exons are shown as blue boxes. Transcript variants V1 and V3 contain the NRE in the first intron of the gene and therefore generate pre-mRNA that contains the repeat expansion and can produce RNA foci. The presence of the C9orf72 NRE in these transcripts also results in decreased use of the endogenous promoter of C9orf72 transcription (depicted) and increased use of alternative promoters identified between the endogenous promoter and the repeats. The NRE is located within the predicted promoter region of variant V2 and therefore may contribute to the reduced transcription and highly decreased levels of mRNA.

There are three primary mechanisms through which the C9orf72 NRE has been proposed to contribute to neural injury: through diminished C9orf72 protein levels and consequent loss of C9orf72 function, through the generation of expanded toxic RNA species, and through the accumulation of dipeptide repeat proteins (DPRs) encoded by the NRE.

In this Review, we explore the current literature supporting these three proposed drivers of C9orf72 NRE-linked disease. We present each of the possible disease mechanisms and describe the cellular defects that have been observed in patient cells and tissues and in transgenic C9orf72 NRE models. We describe the overlapping pathways that have been identified as being altered in disease and then briefly discuss therapeutic strategies being explored to offset the cellular dysfunctions that are linked to the C9orf72 NRE mutation.

C9orf72 loss of function

A loss-of-function mechanism in C9 ALS/FTD was first suggested by the observation that there are reduced levels of all three C9orf72 mRNA variants in patient tissue samples^2,13-17. Although the transcript levels were highly variable among patients in these studies, the findings were consistent with early immunostaining studies that showed decreased nuclear levels of C9orf72 in patient fibroblasts¹. Staining with newer and more reliable isoform-specific antibodies confirmed a reduction in the levels of the long isoform of C9orf72, but suggested that there was an increase in levels of the short isoform in patient cortex samples¹¹. Notably, the short isoform of C9orf72 re-localized to the plasma membrane in motor neurons of patients with C9 ALS (although no significant difference was observed in the motor cortex)¹¹. The biological significance of this re-localization remains unclear. However, there is evidence to indicate that C9orf72 may be involved in RAB-GTPase-mediated cellular trafficking¹². Moreover, the interaction of C9orf72 with nuclear pore complex (NPC) components is inversely correlated with levels of TAR DNA-binding protein 43 (TDP43) cytoplasmic inclusions in human tissue¹¹. Together, these findings provide a glimpse into the possible role of C9orf72 loss-of-function mutations in neurodegeneration.

Mechanisms driving C9orf72 loss of function.

Several mechanisms, including epigenetic silencing and transcriptional instability, might lead to the transcriptional silencing of C9orf72 (REFS 16,18,19) (BOX 2). The presence of the NRE leads to increased local epigenetic modifications that cause a promoter region located upstream of and distal to the NRE (responsible for driving the production of the transcript variants V1 and V3) to contain fewer transcriptional start sites. Furthermore, the NRE results in increased numbers of transcriptional start sites immediately upstream of the NRE²⁰ (BOX 2; FIG. 1). The promoter region that is located within or downstream of the NRE (and that drives the production of transcript V2; FIG. 1) also contains fewer transcriptional start sites²⁰. An increase in repressive DNA and histone epigenetic mechanisms initiated by the NRE (BOX 2) underlies these alterations and changes the transcriptional landscape of the mutant allele. This process leads to the production of truncated and/or divergent transcripts^18,20-22.

Box 2 ∣. Genome instability and loss of function.

Genomic regions that contain repetitive DNA sequences, such as nucleotide repeat expansions (NREs), have been linked to approximately 40 neurological or neuromuscular genetic disorders^104-107. In tissues from patients with these disorders, the molecular mass of NRE-containing regions of DNA can be variable, indicating genome instability — a tendency for the genomic region to be prone to DNA breaks, mutations and/or expansion or contractions. Often, this instability occurs in a tissue- or cell-specific manner^108,109.

Several mechanisms are proposed to contribute to genome instability, including increased sensitivity to spontaneous mutations, recombination and/or accumulation of single-stranded (ss) or double-stranded (ds) DNA breaks¹¹⁰. It is currently unclear how such mechanisms contribute to cell type-specific local genome instability^99,110,111. However, there is evidence to indicate that the formation of secondary or tertiary nucleotide structures such as hairpins, parallel and antiparallel G-quadruplexes (G4)^112-118, R loops^119-124 and I-motifs increases the exposure of ssDNA to damaging environmental agents and can initiate repeat expansion or contraction. For example, upon local dsDNA melting, the formation of these alternative structural states on at least one of the two ssDNAs can lead to insertion or removal of repeat tracts as a result of DNA repair processes⁹⁹. Additionally, bidirectional transcription, which may be mechanistically linked to R loop formation^125,126, may contribute to expansion or contraction of repeats¹²⁵. Given that the chromosome 9 open reading frame (C9orf72) NRE sequence adopts many of these structural features (see the figure, top panel), it is unsurprising that genome instability was identified in patients carrying the mutation².

Several other proteins linked to neurodegeneration have also been linked to the maintenance of genome fidelity. For example, unresolved R loops require the actions of an RNA:DNA helicase, such as the amyotrophic lateral sclerosis (ALS)-related protein senataxin^127,128, to resolve their stability¹²⁹. These proteins may also provide mechanistic insight into the age-related genome defects and damage, as well as chromosomal spatial reorganization, that may lead to degeneration.

There is increasing evidence to indicate that the cell can respond to the presence of the C9orf72 NRE by epigenetically silencing the mutant C9orf72 allele. Specifically, increased repressive cytosine hypermethylation at the C9orf72 NRE locus has been reported^{19,98,130-133}, and has been proposed to spread outward to the flanking DNA⁹⁸ (see the figure, bottom panel). DNA methylation can be neuroprotective and act as a disease modifier^132,133, suggesting that DNA hypermethylation may act against the progression of disease. There is also evidence to suggest that there is increased repressive histone hypermethylation at the C9orf72 NRE locus in patients^130,131. These observations are consistent with repressive epigenetic changes observed in other NRE-linked diseases¹³⁴. However, the epigenetic effects of repressive methylation may be more complicated in the case of the C9orf72 NRE. As outlined above, the formation of alternative DNA structures may alter heterochromatin and transcriptional levels at the NRE locus^135,136. Cytosine hypermethylation can stabilize the formation of G-quadruplexes, thus increasing transcriptional silencing in vitro^137,138. Therefore, there may be a battle between the repressive epigenetic mechanisms acting to decrease local genomic and transcriptional instability induced by alternative nucleotide structures and further modulation of these epigenetic effects by these same alternative nucleotide structures (FIG. 2). Thus, epigenetic modifications aimed to alleviate the potential deleterious effects of the NRE may become compromised over time. However, functional studies specifically testing for a role of epigenetics in downstream cellular toxicity have not yet been performed.

The structurally polymorphic nature of the C9orf72 NRE (BOX 2) may also directly impair polymerase processivity, causing RNA polymerases to stall, slip or release during transcriptional initiation or elongation. This impaired polymerase processivity leads to reduced transcript levels and/or aborted or truncated transcripts^18,21,23-27. In vivo, additional transcriptional elongation factors can rescue or dismantle stalled polymerase complexes²⁸. However, the need to maintain transcriptional fidelity across the NRE may sequester these essential proteins, thus increasing transcriptional and genomic instability.

Bidirectional transcription also arises as a result of the C9orf72 NRE and is common to many NRE-linked diseases^29-31. Although the mechanisms underlying the bidirectional production of repeat-containing transcripts are poorly understood, it is known that bidirectional transcription occurs at most CpG islands throughout the genome and can produce varying lengths of RNA transcripts^32,33. Thus, a GC-rich sequence, such as the C9orf72 NRE, may produce short abortive transcripts through bidirectional transcription. These short transcripts could act as siRNAs and drive gene repression^34,35, as described for other NREs^36,37. One could also speculate that such short abortive transcripts could hijack the molecular machinery through which microRNAs function, modulating the processing of RNAs containing sequences that are complimentary to the NRE^38,39.

C9orf72 loss of function and neurodegeneration.

It is unclear whether the NRE-driven reduction in C9orf72 expression levels causes neurodegeneration. In zebrafish, c9orf72 knockdown resulted in severe axonal and behavioural deficits that were rescued by the expression of exogenous human C9orf72 (REF. 8). Similarly, deletion of the Caenorhabditis elegans C9orf72 orthologue alfa-1 resulted in age-dependent motility deficits⁴⁰. However, knockdown of C9orf72 in rat primary cortical and motor neurons did not affect neuronal survival over a 2-week period⁴¹. Similarly, adult mice treated with C9orf72-targeting antisense oligonucleotides (ASOs) exhibited reduced C9orf72 RNA levels in the brain and spinal cord but no neuropathology or behavioural deficits⁴². In mice studied up to 24 months of age, C9orf72 ablation in neurons and glia did not result in any motor neuron dysfunction, pathology or altered survival⁴³. The differences between findings in mice and other species might be explained by the high degree of homology between human C9orf72 and its mouse orthologue (98%) when compared with its homology to zebrafish c9orf72 (76%) or C. elegans alfa-1 (50%). Mouse and human C9orf72 protein may exhibit similar endogenous function owing to their high protein homology compared to other species; thus, loss-of-function studies in mice might be more relevant to the human disease condition.

The idea that loss of C9orf72 function drives disease pathology was also challenged by studies of patients with C9 ALS/FTD who exhibit >50% reduction of C9orf72 RNA levels but exhibit a phenotype similar to patients with C9 ALS/FTD who have higher C9orf72 RNA levels^44,45. However, recent work that used mouse models carrying a bacterial artificial chromosome (BAC) that expresses human C9orf72 containing the NRE provided support for a possible C9orf72 loss-of-function mechanism. In these studies, several gain-of-function mechanisms and additional pathological features associated with disease were reproduced (see below). However, the mice showed no neurodegenerative phenotypes, suggesting that gain-of-function mechanisms alone cannot explain disease pathogenesis^22,46. This result hints that C9orf72 loss of function contributes to neurodegeneration.

C9orf72 gain-of-function mechanisms

A considerable body of in vivo and in vitro evidence suggests a gain-of-function cellular mechanism as a cause of C9orf72 NRE-mediated neurodegeneration; however, it remains unclear which of the proposed gains in function is the most important.

Sense and antisense NRE RNA foci.

Repeat expansion mutations generate expanded RNA sequences that have been linked to disease pathology. Evidence for their role in C9 ALS/FTD derives from the presence of RNA aggregates containing the (G₄C₂) repeat encoded by the NRE in nuclei within the spinal cord and cortex of patients with C9 FTD². These RNA foci were subsequently observed in the cerebellum, motor cortex, frontal cortex, hippocampus, lumbar motor neurons, interneurons and glia of patients with C9 ALS/FTD^{2,13,30,42,47-50}. (G₄C₂) RNA foci are also found in motor and cortical neurons generated from induced pluripotent stem cells (iPSCs) derived from patients with C9 ALS/FTD^13,14,20,42. In both patient tissue and iPSC-derived neurons the foci are predominantly nuclear^13,50.

The bidirectional transcription of the C9orf72 NRE^30,31 also results in the generation of RNA aggregates. RNA foci containing the (C₄G₂) repeat encoded by the antisense transcript of the NRE are observed in the frontal cortex, Purkinje cells, cerebellum, hippocampus, spinal cord motor neurons, astrocytes, microglia and oligodendrocytes of patients with C9 ALS/FTD^30,42,48-50, where they are mainly nuclear⁵⁰. (G₄C₂) and (C₄G₂) RNA foci can be detected in the same nuclei and have been reported to colocalize^49,50. One study showed that (G₄C₂) and (C₄G₂) foci appear in higher numbers in Purkinje neurons and motor neurons than in granule neurons and hippocampal CA4 subfield and dentate gyrus neurons⁴⁹. However, another study found that (G₄C₂) RNA foci predominantly exist in cerebellar granule neurons, whereas (C₄G₂) RNA foci are found more frequently in cerebellar Purkinje neurons and motor neurons⁴⁹. It is important to note that a direct comparison of the levels of sense and antisense RNA foci using RNA fluorescence in situ hybridization is impossible unless alternative quantitative methods are used.

The appearance of C9orf72 NRE RNA foci sometimes correlates with pathological disease features. Age of disease onset correlated with the number of (G₄C₂) and (C₄G₂) RNA foci in a small number of patients with C9 FTD⁵⁰. Furthermore, 30–60% of (G₄C₂) or (C₄G₂) RNA foci-containing cells in the frontal cortex and hippocampus showed TDP43 mislocalization. However, less than 20% of these cells showed p62 mislocalization, calling into question the correlation between RNA foci and pathological inclusions⁵⁰. Another study showed RNA foci and p62 inclusions but no TDP43 mislocalization in an asymptomatic C9orf72 NRE carrier⁵¹. Thus, although it is tempting to speculate that RNA foci are toxic to cells, there is limited evidence to indicate that the RNA foci themselves cause cellular dysfunction. This conclusion is supported by the observations of RNA foci in the absence of neurodegenerative phenotypes in C9orf72 BAC mice models^22,46. This result indicates that RNA gain-of-function mechanisms probably require an additional component to manifest neurodegenerative phenotypes. As outlined below, the sequestration of RNA-binding proteins (RBPs) by the expanded RNA within NRE RNA foci can disrupt various molecular cascades⁵² (BOX 2; FIG. 2; Supplementary information S1 (table)) and is therefore likely to contribute to disease pathogenesis.

Figure 2 ∣. Cellular processes impaired by the C9orf72 nucleotide repeat expansion.

Cellular features and pathways that are altered by the RNA and/or dipeptide repeat proteins (DPRs) generated from the chromosome 9 open reading frame (C9orf72) nucleotide repeat expansion (NRE) mutation are shown. Black and grey arrows indicate pathological features linked to the C9orf72 RNA or DPRs, respectively. The repeat-containing RNA, which forms RNA foci, has been implicated in sequestering key RNA-binding proteins (RBPs). This process may lead directly to altered nucleolar function, changes in RNA processing or maturation and nucleocytoplasmic transport defects. DPRs have also been shown to lead to nucleocytoplasmic transport defects, as well as altering autophagy, nucleolar function and increasing endoplasmic reticulum (ER) stress. As shown in the right part of the image, RNA and DPRs generated from the C9orf72 NRE mutation converge on pathways driving nucleocytoplasmic trafficking defects, including altered nuclear pore complex dynamics, import factors and nucleolar stress. C9orf72 RNA can sequester nucleolar proteins, leading to nucleolar stress, or bind to nuclear pore complex proteins and disrupt nucleocytoplasmic trafficking. Cytoplasmic C9orf72 RNA can be non-canonically translated though repeat-associated non-ATG-dependent translation (RANT), leading to the production of DPRs. Specific DPRs can impair nucleocytoplasmic trafficking and, when imported into the nucleus, can associate with nucleolar proteins to cause nucleolar stress. Additionally, both the C9orf72 RNA and specific DPRs have been proposed to affect RNA splicing and/or editing.

Protein sequestration.

Protein sequestration by RNA is an established mechanism of toxicity in some NRE disorders. For example, the myotonic dystrophy (DM1) NRE results in the transcription of CUG repeats, which sequester the muscleblind-like protein 1 (MBNL1) protein and lead to MBNL1 loss of function⁵³.

To determine the RBPs that may be sequestered by RNA encoded by the C9orf72 NRE, several laboratories have used various methods to enable an unbiased identification of the ribonucleoproteins that interact with the transcribed C9orf72 NRE in different tissues^13,18,54-56. Many unique C9orf72•RBP interactomes have been identified (FIG. 3; Supplementary information S2 (table)). Comparisons among these studies identified heterogeneous nuclear ribonucleoproteins (hnRNPs) as an overrepresented protein family that bind to the C9orf72 NRE RNA. Proteins in this family contain an RNA recognition motif that is shared with several other proteins, including TDP43 and fused in sarcoma (FUS; also known as TLS). In particular, isoforms of hnRNP H can bind to the (G₄C₂) RNA, which may lead to globally increased alternative splicing of intron exon cassettes in the motor cortex and cerebellum of patients with C9 ALS/FTD⁵⁷.

Figure 3 ∣. Identification of a diverse C9orf72•protein interactome.

The diagram compares five different chromosome 9 open reading frame (C9orf72)•RNA-binding protein (RBP) interactomes discovered using RNA pulldown or protein arrays^13,18,54-56. As shown, many unique proteins were identified that shared no overlap among the studies (see Supplementary information S2 (table)). Numbers indicate the number of protein candidates found to overlap between the indicated studies. The protein isoforms of heterogeneous nuclear ribonucleoprotein (hnRNP) H showed the most overlap between studies. Other proteins identified in at least three studies included splicing factor proline/glutamine-rich (SFPQ), interleukin enhancer-binding factor 2 (ILF2) and myelin basic protein (MBP). Of the overlapping RBPs, only hnRNP H has been shown to contribute to the increased alternative splicing events observed in patients carrying a C9orf72 nucleotide repeat expansion (NRE)⁵⁷. Most RBP candidates have not been explored for potential functional consequences in vivo. Regions assigned no number designate comparisons in which there was no protein overlap.

Many additional proteins, including Purα, adenosine deaminase RNA-specific, B2 (ADARB2), nucleolin, hnRNP A1, hnRNP A2/B1, hnRNP A3, hnRNP H, ALYREF and RanGAP1, colocalize with C9orf72 (G₄C₂) RNA foci^{13,18,20,49,54,56,58,59}. However, only a handful of these proteins have been further studied, and their biological relevance has not been validated. It is also important to note that the C9orf72•RBP interactome is largely dependent on the underlying structures formed by the sense C9orf72 RNA, which may include either G-quadruplex or hairpin structures. For example, nucleolin and RanGAP1 preferentially bind to the (G₄C₂) C9orf72 RNA G-quadruplex^18,59, whereas hnRNP H isoforms bind to the RNA G-quadruplex or hairpin structure with similar affinities¹⁸. Thus, the structural variability of the (G₄C₂) RNA may affect divergent RNA processing pathways and have a critical role in a multihit RNA gain-of-function model in which there are broad RNA processing alterations.

DPR pathology.

Repeat-containing RNAs that escape the nucleus can form cytoplasmic RNA foci and can also be translated into DPRs through a non-canonical translational mechanism known as repeat-associated non-ATG-dependent translation (RANT)⁶⁰ (BOX 3). RANT has been identified in vitro and in vivo in many neurological diseases, including spinocerebellar ataxia type 8 (REF. 60), myotonic dystrophy type 1 (REF. 60), fragile X syndrome⁶¹, Huntington disease and C9 ALS/FTD^{13,14,30,31,48,62-65}. RANT has also been proposed to be an important contributor to NRE-linked disease progression⁶⁶.

Box 3 ∣. The production of dipeptide repeats.

The mechanisms that underlie repeat-associated non-ATG-dependent translation (RANT) and the cellular conditions that promote it are poorly understood. Two conditional constraints have been identified for the production of detectable dipeptide repeat (DPR) levels: the nucleotide repeat expansion (NRE) must reach a threshold length before RANT will occur, and the NRE must be able to adopt RNA secondary structures such as hairpins⁶⁰. Notably, all transcribed NRE sequences that have nearly self-complimentary base pairing can adopt stable or dynamic RNA secondary structures, which implies that most NREs identified to date meet this criterion.

Recently, the mechanism underlying RANT has been proposed to parallel other mechanisms of non-traditional translation, such as the translation of RNA sequences that contain internal ribosome entry sites (IRESs)¹³⁹. However, IRES-mediated translation classically uses an AUG start site, and only a few IRES sequences have been examined in the absence of AUG start sites and/or for translation in all three reading frames^140,141.

An unusual translation feature found to occur in other NREs is ribosomal frameshifting, caused by the shifting of the ribosomal complex units by one nucleotide in either direction during translational elongation. Large tracts of (CAG) repeats were reported to lead to increased ribosomal frameshifting in several neurological disorders^142-145. Experiments on (CAG) repeats demonstrated that there was an equal probability of the occurrence of a frameshift at any (CAG) codon. This effect results in a mixture of frameshift hybrids in which the frequency of ribosomal frameshifting is proportional to the (CAG) length¹⁴⁶. However, whether frameshifting occurs during RANT of NREs has not been shown, and the extent to which it contributes to the diversity of DPRs generated in vitro or in vivo is unknown.

In the case of the chromosome 9 open reading frame (C9orf72) NRE, the formation of diverse stable RNA structures could further impair ribosomal elongation and also presumably increase the probability of frameshifting. Importantly, the formation of RNA G-quadruplexes can enhance ribosomal frameshifting in vitro and in vivo, and this process may affect the translation of and relative abundance of specific DPRs generated from the G-quadruplex-forming C9orf72 NRE¹⁴⁷. Furthermore, the presence of an RNA G-quadruplex has been shown to increase IRES-mediated translation; however, the formation of multiple RNA G-quadruplexes on a transcript can have an inhibitory effect on translation¹⁴⁸. Therefore, the structures that enhance the recruitment of the ribosomal complex to the NREs to facilitate RANT can also modulate translation by either blocking its progression or inducing ribosomal frameshifting during translation elongation in a NRE length-dependent manner.

The sense (G₄C₂) RNA NRE of C9orf72 encodes three DPRs, glycine–proline (GP), glycine–alanine (GA) and glycine–arginine (GR), whereas the antisense (C₄G₂) RNA encodes proline–glycine (PG), proline–arginine (PR) and proline–alanine (PA). Of these, (GP (or PG)) is the most abundant in C9 ALS/FTD neurons, perhaps because it can be produced from both the sense- and the antisense-expanded RNA³⁰. However, (PA), (PR) and (PG) DPRs have been identified in the hippocampus, cerebellum, amygdala, medulla, thalamus, frontal cortex, spinal cord and motor cortex of patients with C9 ALS/FTD^30,48, with the hippocampus and cerebellum (granule cells) showing the highest levels of DPRs. C9 ALS/FTD spinal cord and brainstem tissue contains the lowest DPR levels^30,67. Notably, (GP), (GA) and (GR) DPRs colocalize with p62 but do not correlate with TDP43 mislocalization or neurodegeneration in the neocortex, spinal cord or cerebellum⁶⁴. (GP) DPRs were also detected in C9 ALS/FTD iPSC-derived cortical and motor neuron cultures^13,14, and the localization of endogenous C9 ALS/FTD DPRs is predominantly cytoplasmic^{13,30,31,48,63,64,68}. Interestingly, for most cells in C9orf72 NRE patient tissue, RNA foci and RANT products were mutually exclusive⁴⁸.

The localization of these DPRs to regions such as the cerebellum and lateral geniculate nucleus is confusing because these brain regions are not associated with obvious clinical or pathological aspects of ALS or FTD. There are changes in the RNA profile of the cerebellum of patients with C9 ALS/FTD, the significance of which is currently unclear⁵⁷. However, cerebellar (GP) levels were correlated with cognitive dysfunction in a cohort of patients with C9 ALS/FTD⁶⁹. Imaging studies also suggest that there are subtle cerebellar anatomical abnormalities in patients with C9 FTD⁷⁰. As noted above, DPRs have only occasionally been observed in spinal cord motor neurons^30,67, in which p62 and TPD43 cytoplasmic aggregates are common. This observation has led to confusion about the role of these DPRs in endogenous disease. Analysis of three post-mortem C9orf72 NRE carriers who died prematurely of unexpected causes revealed DPR accumulation with minimal TDP43 mislocalization, suggesting that RANT might precede TDP43 pathology⁷¹. However, it remains unclear whether physiologically relevant levels of DPRs can lead to neurodegeneration. Indeed, recent evidence from C9orf72 NRE BAC mouse models show that DPR pathology does not correlate with a neurodegenerative phenotype^22,46. It will be imperative to standardize antibody use and protocols in future studies to better understand DPR pathobiology in C9 ALS/FTD.

What is the primary gain-of-function mechanism?

Whether ‘toxic’ protein-sequestering RNA or RANT DPRs are the primary drivers of C9 ALS/FTD neurotoxicity is highly debated (see Supplementary information S1 (table)). Exogenous overexpression of 38 or 72 copies of (G₄C₂) ((G₄C₂)_{38 or 72}) produced RNA foci in the SH-SY5Y neuronal cell line and primary mouse hippocampal neuronal cultures and the presence of these foci correlated with high levels of activated caspase 3 and a length-dependent increase in apoptotic activity⁵⁸. (G₄C₂)_{38 or 72} overexpression also caused length-dependent apoptotic cell death in transfected zebrafish embryos⁵⁸.

Ubiquitous (G₄C₂)₃₀ transgene expression in Drosophila melanogaster was developmentally lethal, and specific expression of (G₄C₂)₃₀ in the eye or motor neurons led to progressive neurodegeneration⁵⁶. (G₄C₂)₃₀ expression in D. melanogaster motor neurons also induced an age-dependent reduction in locomotor activity⁵⁶. In support of a role for RBP sequestration in these effects, overexpression of Purα, which interacts with (G₄C₂) RNA, suppressed (G₄C₂)₃₀-mediated neurodegeneration in the D. melanogaster eye; however, it was unclear whether Purα expression attenuates motor neuron degeneration⁵⁶.

In rodents, neural-specific expression of (G₄C₂)₆₆ caused widespread CNS neurodegeneration⁷². These mice also exhibited both (G₄C₂) RNA foci and (GP) and (GA) DPRs⁷². (G₄C₂)₆₆-expressing mice exhibited cortical neuron and Purkinje cell degeneration and astrogliosis⁷² and developed significant motor deficits⁷². Although these studies support a gain-of-function mechanism of degeneration, they do not separate RNA and DPR toxicity and may not accurately recapitulate the molecular events and/or toxicity associated with physiological expression levels.

The development of neurons differentiated from iPSCs derived from patients with C9 ALS/FTD has enabled the recapitulation of endogenous allelic expression of mutant C9orf72 RNA. C9orf72 iPSC-derived motor neurons exhibited reduced C9orf72 mRNA expression, reduced numbers of (G₄C₂) RNA foci, reduced (GP) DPR levels and altered gene expression^13,14,20. Similarly, iNeurons (neurons reprogrammed directly from patient fibroblasts) from patients with C9 ALS/FTD exhibited sense and antisense RNA foci and (GP) and (PR) DPRs⁷³. C9orf72 iPSC-derived cortical neurons exhibited enhanced sensitivity to autophagy¹⁴ and iPSC-derived motor neuron cultures were sensitive to endoplasmic reticulum stressors¹⁸. Young C9 ALS/FTD iPSC-derived neuronal cultures exhibited hyperexcitability^74,75 followed by an age-dependent shift to hypoexcitability^20,74. Notably, cortical hyperexcitability has also been identified in symptomatic carriers of the C9orf92 NRE⁷⁶. Consistent with this observation, younger C9orf72 iPSC-derived motor neuron cultures were more sensitive to extracellular glutamate stimulation¹³. Acute treatment with ASOs targeting the (G₄C₂) repeat, the intronic region upstream of the NRE or the C9orf72 coding sequence RNA reduced the number of RNA foci and/or mRNA expression and mitigated the sensitivity of C9 ALS/FTD iPSC-derived neurons to glutamate despite the presence of (GP) DPRs¹³. These studies point to the involvement of sense (G₄C₂) RNA in toxicity, but do not rule out (GR) or (GA) DPR toxicity, toxicity from newly translated or undetectable (GP) DPRs or toxicity from the (C₄G₂) NRE.

An RNA-based gain-of-function mechanism has also been suggested by longitudinal imaging of primary neuronal cultures. A (G₄C₂)₄₂ intronic RNA expressed in mouse cortical and motor neuron cultures reduced neuronal survival and enhanced the cumulative risk of death⁴¹. Interestingly, RNA toxicity was dose-dependent and exhibited neural-subtype specificity because (G₄C₂)₄₂ expression was not toxic to hippocampal neurons⁴¹. Notably, DPRs were not detected in intronic (G₄C₂)₄₂ RNA-expressing cells⁴¹. However, one cannot exclude a contribution of undetectable levels of DPRs.

In other studies, DPR gain-of-function mechanisms have been demonstrated. Overexpression of ATG-(PR)-(C₄G₂) or ATG-(GP)-(G₄C₂) constructs in HEK293 cells enhanced cell death compared with (G₄C₂) or (C₄G₂) expression alone³⁰. Similarly, U2OS cells showed reduced viability when incubated with (GR)₂₀ or (PR)₂₀, and U2OS cells and cultured human astrocytes exhibited morphological changes when incubated with (GR) or (PR)⁷⁷. Purified (PR)₂₀ was toxic to U2OS cells at high concentrations, whereas similar concentrations of purified (GR)₂₀, which has a short half-life⁷⁷, conferred toxicity if reapplied every 2 hours.

Longitudinal imaging of mouse primary neurons and human neurons showed that overexpression of ATG-(PR)₅₀ enhanced the cumulative risk of death and reduced survival of mouse primary motor, cortical and hippocampal neurons in a dose-dependent manner⁴¹. ATG-(GR)₅₀ was also toxic to mouse primary motor neurons; however, its effects were less severe than ATG-(PR)₅₀, and it did not reduce survival of cortical or hippocampal neurons⁴¹. ATG-(PR)₅₀ overexpression was also neurotoxic to human iPSC-derived neuronal cultures and motor neurons generated from human fibroblasts⁴¹. Although mouse primary motor neuron survival was not significantly reduced by ATG-(GA)₅₀ expression in one study, other studies found that ATG-(GA)_50–175 expression reduces dendritic branching⁷⁸, induces endoplasmic reticulum stress⁷⁹, impairs Notch signalling⁸⁰, enhances apoptotic signalling via activated caspase 3 and caspase 7, and is toxic to primary mouse cortical neurons in vitro^78,79.

DPR-based gain of function was also demonstrated in two in vivo D. melanogaster studies. Targeted expression of ATG-(GR)_{36 or 100} and ATG-(PR)_{36 or 100} caused degeneration when expressed in the adult eye and reduced survival when expressed ubiquitously⁴⁷. Expression of ATG-(PA)₃₆ and ATG-(GA)₃₆ did not result in eye degeneration; however, ATG-(GA)₁₀₀ expression caused survival deficits that occurred much later than ATG-(GR)₁₀₀ or ATG-(PR)₁₀₀ (REF. 47).

In further support of DPR gain-of-function mechanisms, D. melanogaster with targeted expression of (G₄C₂)_{36, 108 or 299} modified with stop codons to prevent the translation of long DPR products did not exhibit a degenerative eye phenotype⁴⁷ even though the short G₄C₂ RNAs could still form non-canonical structures and RNA foci. D. melanogaster expressing (G₄C₂)_{36 or 103} repeats in the absence of stop codons did exhibit neurodegeneration and production of DPRs. These DPRs could be detected in the brain, but it is unclear whether the appearance of the DPRs occurred concomitantly with eye degeneration⁴⁷. In agreement with the idea that C9orf72 NRE toxicity is protein-based, treatment with a general inhibitor of translation led to decreased toxicity in D. melanogaster expressing (G₄C₂)_{36 or 103} repeats. Additionally, ELISA (enzyme-linked immunosorbent assay) methods that detect low levels of DPRs⁷³ demonstrate that there is a correlation between neurodegeneration and DPR levels in D. melanogaster C9orf72 expression models, with no or minimal correlation with C9orf72 RNA foci⁸¹. This evidence therefore supports the notion that RANT, and maybe specific DPR species, is the primary driver of C9orf72 NRE neurotoxicity, but still does not reconcile the recent BAC mouse models of C9orf72 that show the presence of DPRs and RNA foci with no neurodegenerative phenotype^22,46.

In patients with C9orf72 NRE ALS, DPRs are rarely observed in the spinal cord, despite the frequent presence of cytoplasmic TDP43 in this tissue⁶⁷. However, DPRs are frequently found in the frontal cortex of patients with C9 ALS/FTD, suggesting that a tissue-specific mechanism underlies RNA or DPR toxicity⁸². For example, (GA) and (GP) were found in higher levels in the cerebellum of patients with C9orf72 FTLD and in patients with FTLD-motor neuron disease than in patients with C9orf72 ALS, and (GP) DPR levels positively correlated with cognitive impairment⁶⁹. Thus, unknown intrinsic cellular mechanisms that promote RANT might dictate RNA or DPR neurotoxicity.

In summary, it is clear that RNA or DPR accumulation can be toxic in a dose-dependent manner. It is possible that there is a convergence of RNA and DPR toxicity. Indirect evidence for this idea was found when co-expression of intronic (G₄C₂)₄₂ and ATG-(PR)₅₀ synergistically increased the cumulative risk of death of primary mouse cortical neurons⁴¹. However, many of these experiments relied on overexpression or application of the RNA or DPRs at concentrations that probably exceed physiological levels. Therefore, it is important to understand the relative contribution of the RNA and/or DPR toxicity in relevant endogenous systems such as iPSCs and the human brain. To this end, it will be necessary to fully characterize and identify methods to specifically inhibit RANT. Furthermore, whether NRE length correlates with toxicity remains unclear. Recent work has shown that an asymptomatic (G₄C₂)₃₀ carrier exhibited sense and antisense RNA foci, DPRs and p62 inclusions but did not exhibit TDP43 mislocalization, indicating that even short repeats can initiate C9orf72-related pathology⁵¹.

Dysfunctional pathways in C9 ALS/FTD

How does the C9orf72 NRE actually lead to cellular dysfunction? In recent years multiple theories regarding the pathophysiological events that underlie C9orf72 NRE cytotoxicity have emerged (Supplementary information S1 (table)). These events provide a basic understanding of disease pathophysiology and may provide new targets for intervention.

Nuclear stress.

RNA containing the C9orf72 NRE accumulates in the nucleus, where it can perturb many pathways and nuclear bodies. These perturbations are largely dependent on the RBP•C9orf72 NRE interactome that can alter the function of essential proteins, such as those involved in RNA processing and maturation. Recent transcriptome profile analyses of brain tissue from patients harbouring the C9orf72 NRE confirmed that RNA processing is impaired and that there is increased alternative splicing and alternative polyadenylation, including mis-splicing of other known ALS- or FTD-linked genes⁵⁷.

The nucleolus — which is necessary for ribosomal RNA biogenesis — is composed of several proteins that can interact with C9orf72 NRE RNA and specific DPRs, and nucleolar function is impaired in patients carrying the C9orf72 NRE and models of C9orf72 NRE^{18,22,41,77,83}. Moreover, (GR)_n and (PR)_n colocalize with the nucleolus and can induce nucleolar stress and cell death at micromolar concentrations⁷⁷. Similar nucleolar stress was observed in in vitro models overexpressing (G₄C₂)-containing RNA or scrambled RNAs designed to produce DPR-only cellular toxicity^41,84. C9orf72 NRE BAC mouse models also recapitulate nucleolar stress pathology²². The primary cause of this nucleolar stress remains unknown; however, evidence suggests that a dual toxicity model may be applicable^13,41 and that the nucleolus could be a convergence point for C9orf72 NRE RNA and DPR gain of function (FIG. 2).

Nucleocytoplasmic trafficking.

A new direction in our understanding of C9orf72 NRE pathophysiology came from four independent studies that implicated dysfunctional nucleocytoplasmic trafficking as a key event in C9 ALS/FTD. In the first study, a large interactome screen showed that (G₄C₂) RNA interacts with RanGAP1, a GTPase-activating protein that stimulates the hydrolysis of RanGTPase (Ran)^13,59. Ran is necessary for the active transport of proteins that contain a classical nuclear localization sequence (NLS). RanGAP1 was shown to be a potent genetic modifier (suppressor) of a neurodegenerative phenotype (rough eye) in (G₄C₂)₃₀-expressing D. melanogaster⁵⁹. C9 ALS iPSC-derived neurons and D. melanogaster C9orf72 NRE models exhibited a disrupted Ran gradient⁵⁹ and showed reduced localization or sluggish fluorescent recovery after photobleaching (FRAP) for NLS-containing reporter molecules, implying that there are deficits in nuclear protein import⁵⁹ and disruption of functional nucleocytoplasmic transport. Indeed, TDP43, which contains a classical NLS sequence, was slightly enriched in the cytoplasm of C9 ALS iPSC neurons⁵⁹. Genetic crosses and therapeutics designed to modify nucleocytoplasmic trafficking and enhance protein import or reduce nuclear export rescued the (G₄C₂)-mediated rough eye phenotype⁵⁹. ASOs targeting the NRE sense strand rescued the nuclear transport deficits and were neuroprotective against the rough eye phenotype in the D. melanogaster model and rescued the Ran gradient and reduced cytoplasmic TDP43 localization in ALS iPSC neurons⁵⁹. Notably, in these studies, the flies expressed only the sense strand RNAs and DPRs were undetectable during the time of neurodegeneration, suggesting that the observed deficits were primarily caused by NRE sense strand (G₄C₂) RNA.

A second study carried out an unbiased loci deletion screen in a novel D. melanogaster model that expresses (G₄C₂)₅₈ and exhibits sense strand RANT products⁸⁵. Eighteen genetic modifiers of the rough eye phenotype in genes that encode for NPCs and components of the nucleocytoplasmic trafficking pathway were identified⁸⁵. One potent suppressor of the rough eye phenotype was AlyRef, a protein involved in mRNA export^49,54,85. Consistent with the idea that AlyRef has a role in C9 ALS pathogenesis, (G₄C₂)₅₈-expressing D. melanogaster cells and C9 ALS iPSC motor neurons showed significant nuclear RNA retention, which was amplified when additional copies of (G₄C₂)₅₈ transgenes were added⁸⁵. These studies therefore support a model of dysfunction at the NPC, wherein protein import of classical NLS-containing proteins and export of RNAs is impaired.

NPC proteins were also found to be abnormally aggregated in C9 ALS iPSC neurons and throughout the motor cortex of many patients with C9 ALS⁵⁹. For example, RanGAP1 showed perinuclear aggregation and intranuclear accumulation⁵⁹. Nup107 (a modifier of neurodegeneration in the (G₄C₂)₅₈ fly model⁸⁵) and Nup205 showed similar aggregation and nuclear retention in the motor cortex of patients with C9 ALS⁵⁹, and nuclear membrane abnormalities were also observed in the (G₄C₂)₅₈ fly model⁸⁵. The mechanism (or mechanisms) underlying this aggregation is unknown. Furthermore, whether the aggregated NPC proteins affect both nuclear import and export as well as whether this pathology is reversible remains to be ascertained.

A third study showed that DPRs can also affect nucleocytoplasmic transport. Overexpression of ATG-(PR)₅₀ in yeast resulted in toxicity that could be modified by 35 genetic enhancers and 27 suppressors⁸⁶. Notably, 11 of these suppressors were involved in nucleocytoplasmic trafficking, including six importin proteins⁸⁶. iNeurons from two patients with C9 ALS also showed a reduction and/or mislocalization of a nuclear RanGEF protein, RCC1 (REF. 86), that could similarly disrupt nucleocytoplasmic trafficking. Interestingly, in contrast to the studies above, this study implicated the antisense (C₄G₂) RNA in toxicity, via (PR) accumulation⁸⁶. Therefore, although both the sense and antisense NRE lead to nucleocytoplasmic trafficking deficits, they could act through independent mechanisms.

In the fourth and most recent study, cytoplasmic protein aggregates were shown to interfere with nucleocytoplasmic transport⁸⁷. This study demonstrated that the aggregation of overexpressed artificial β-sheet proteins or mutant huntingtin or TDP43 fragments in the cytoplasm and not the nucleus lead to nucleocytoplasmic transport defects in HEK293T cells. Altered NPC pathology was also only observed when the β-sheet proteins aggregated in the cytoplasm. This work suggested that cytoplasmic protein aggregation, a pathological feature shared by several neurological disorders, can itself lead to impaired nucleocytoplasmic transport⁸⁷.

The primary cause of NPC dysfunction and nucleocytoplasmic transport defects in patients carrying the C9orf72 NRE and in models of C9orf72 NRE is currently unclear. The interpretation of some results may be hampered by the non-physiological overexpression systems used. It seems likely that there are many nuclear trafficking proteins that are affected by the (G₄C₂) RNA, DPRs and/or cytoplasmic protein aggregation. However, it is also likely that nucleocytoplasmic transport and NPC dysfunction in C9 ALS/FTD is a central event and that it occurs via gain-of-function mechanisms. Furthermore, it is possible that there may be a node at which different mechanisms converge. For example, (G₄C₂) RNA might initiate the cascade by interacting with RNA export factors in the nucleus and nuclear- or NPC-bound components. Simultaneously, (G₄C₂)_exp and (C₄G₂)_exp might escape the nucleus and through RANT produce DPRs that aggregate in the cytoplasm. Arginine-rich DPRs could bind to importins and be transported to the NPC to induce further disruptions. These events could seed NPC aggregation that further compromises nuclear transport while enhancing the nuclear retention of RNAs and the cytoplasmic accumulation of NLS-containing proteins, such as TDP43. This process might lead to enhanced cryptic exon inclusion due to TDP43 mislocalization⁸⁸. Furthermore, RNA retention and nuclear import of arginine-rich DPRs could disrupt nuclear bodies such as the nucleolus^18,41. Finally, the short isoform of endogenous C9orf72, which localizes to the nuclear envelope and biochemically interacts with RanGTPase and importins¹¹, may also contribute to nucleocytoplasmic transport defects through a loss-of-function mechanism. Therefore, it is enticing to speculate that loss of function of C9orf72 might confer additional sensitivity to cells with gain-of-function-mediated deficits in the nucleocytoplasmic transport pathway.

Therapeutic interventions

The prospects for therapeutic interventions for nucleotide expansion disease are exciting. To date, sense ASOs have rescued C9orf72-linked disease phenotypes in human cellular and animal preclinical models (detailed above) and have demonstrated mitigation of C9orf72 pathophysiology^13,20,42. Substantial progress has been made with preclinical studies of ASOs targeting another NRE-linked disease, myotonic dystrophy⁸⁹, and Phase I trials have been successfully completed with ASOs targeting superoxide dismutase 1 (SOD1) in familial ALS^90,91. There are also ongoing studies using ASOs for spinal muscular atrophy⁹¹. Future studies will provide additional information on the efficacy of ASO therapies.

Recent progress has also been made in the design of nucleotide structure-specific small molecules that can alter the progression of disease. For example, small molecules designed to specifically recognize and bind to nucleotides in unique RNA spatial conformations, such as RNA hairpins, have been shown to rescue RNA-induced pathological defects and prevent the downstream accumulation of DPRs^73,92. Other structurally specific drugs, such as the G-quadruplex-binding porphyrin, TMPyP4, can destabilize C9orf72 RNA G-quadruplex•RBP interactions⁹³ and can rescue transport defect pathologies and the rough eye phenotypes observed in the transgenic D. melanogaster (G₄C₂)₃₀ overexpression model⁵⁹. Genetic modification or compounds such as KPT-276 that inhibit nuclear export can rescue neurodegeneration and nucleocytoplasmic trafficking deficits in (G₄C₂)-overexpressing flies, suggesting that modifiers of nucleocytoplasmic trafficking might be an effective alternative therapeutic strategy^59,85. Similar molecules have recently been shown to be efficacious in rodent models of inflammatory demyelination⁹⁴. Inhibiting nuclear export of expanded (G₄C₂) RNA might prevent NPC dysfunction and toxic DPR accumulation.

Pharmacodynamic read outs to assess drug efficacy still remain a challenge when developing therapies for neurodegenerative disorders. SOD1 levels in the cerebrospinal fluid have been used as a pharmacodynamic marker to measure the efficacy of ASO therapeutics in patients with ALS⁹⁵. Similarly, an ELISA-based detection assay for the RANT (GP) DPR is currently in development. To date, studies have successfully quantified extracellular (GP) DPR in the cerebrospinal fluid of 14 patients with C9 ALS/FTD⁷³; perhaps this assay could be used as a longitudinal biomarker for novel C9orf72 NRE-targeting ASOs or small molecules.

Recent advances in genome editing using CRISPR/Cas technology have revolutionized the development of model organisms as a result of its speed and efficacy in engineering DNA. The possibility that CRISPR/Cas technology or similar genome editing might also be used as the basis of therapy to eliminate a fatal human disease is exciting; however, the ethical debate regarding treating genomic diseases in humans will be important⁹⁶. Although improvements have been made to reduce off-target effects, this technology is still in its infancy, and its efficacy in vivo must continue to be rigorously examined before its use as genome therapy tool could be considered.

Concluding thoughts and perspectives

The evidence that C9orf72 NRE generated DPRs, RNA and cytoplasmic protein aggregates that may interfere with NPC and nucleocytoplasmic transport presents opportunities to enhance our understanding, and potential mitigation, of neurodegeneration. For example, does altered nucleocytoplasmic transport precede the cytoplasmic accumulation of TDP43 inclusions in patients⁶⁷? Can we create tools to mitigate DPR-based toxicity? Is the NPC more sensitive in neurons than glia and/or differentially regulated in neuron subtypes, or in neurons versus glia? What is the underlying mechanism of nuclear pore aggregation? Does the short isoform of C9orf72 contribute to nuclear pore pathology? To answer these questions, it is necessary to gain a better understating of the chain of events and the primary mechanistic contributor that leads to the observed patient pathologies.

Substantial research focus has been placed on the toxic gain-of-function mechanisms of the C9orf72 NRE. This focus was partially due to difficulties in understanding the role and function of C9orf72 in normal biology. Although RNA and DPRs from the C9orf72 NRE mutation have revealed many pathological disease features that are initiated by the gain-of-function mechanism, it is unknown which features directly correlate with neurodegeneration. However, uncoupling the C9orf72 NRE phenotypes caused by the RNA versus DPR gain-of-function mechanisms remains a challenge. Quantitative measurements of endogenous RNA and DPR levels in vivo are critical to develop accurate interpretations of in vitro models. Moreover, current RNA and/or DPR overexpression models must be interpreted with strong caution because there can often be a disconnect between phenotypic observations in overexpression systems and actual patient phenotypes.

GDP–GTP nucleotide exchange factor.

(GEF). A protein that functions to stimulate the release of guanosine diphosphate (GDP) from a GTPase to allow subsequent binding of the guanosine triphosphate (GTP) to the active site.

Epigenetic silencing.

Covalent DNA or histone modifications, such as DNA or histone methylation, that act to repress or silence a genomic region by reducing access of the transcriptional machinery to the DNA.

Polymerase processivity.

A relative measure of the functional efficiency and rate of polymerase to process (replicate or transcribe) nucleic acid information.

Bidirectional transcription.

Transcription that occurs simultaneously on both the positive and the negative strands of DNA, where the direction of RNA polymerase progression along each strand is either convergent or divergent.

Haplotype.

A combination of alleles at different loci in the genome that tend to be inherited together because they show high linkage disequilibrium (often because they are physically close).

Induced pluripotent stem cells.

(iPSCs). Cells that are created from differentiated cell types — for example, fibroblasts — that are reprogrammed by a cocktail of transcription factors (or other approaches) back to a pluripotent state. These cells can then be differentiated into cells of distinct lineages — for example, neurons.

R loops.

A nucleic acid structure that occurs when a single strand of RNA invades a double-stranded DNA, forming an RNA:DNA hybrid that is stabilized through Watson–Crick base pairs, and leaves a displaced single-stranded DNA molecule.

dsDNA melting.

The separation of base-paired DNA strands through the disruption of Watson–Crick pairing by physical or thermodynamic means.

Interactomes.

Sets of physical interactions occurring between two or more components.

G-quadruplex.

A structure composed of G-quartets, four planar guanine molecules that are stabilized by Hoogsteen base pairing, that are stacked on top of each other and stabilized by a central cation and π-stacking interactions between the G-quartets.

Hairpin.

A nucleic acid molecule that contains a stem region, which can be composed of complimentary base pairing and be interspersed with single-stranded or loop elements, and contains a tight looped region (hairpin turn) at the end of the base-paired stem region.

Repeat-associated non-ATG-dependent translation.

(RANT). The non-canonical translation of a repetitive RNA sequence initiated by an AUG start codon.

Nucleocytoplasmic trafficking.

The bidirectional protein transport between the cytoplasm and the nuclear matrix through the nuclear pore complex.

Genetic modifier.

A genetic variation in (cis) or outside (trans) a gene or genetic locus that alters the phenotypic expression of the gene.

CRISPR/Cas technology.

The essential components of the bacterial adaptive immunity, Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) and CRISPR-associated (Cas) genes, that are currently being utilized and developed for precise editing, regulating and targeting of genes in model systems and organisms.