Problems at the nuclear pore

Expansion of a repetitive DNA sequence is associated with neurodegeneration. Three studies identify genes involved in nuclear import and export that can mediate the toxicity this expansion causes. See Article p.56 & Letter p.129

Advances in molecular genetics and stem-cell technology are transforming our understanding of disease. Such progress is desperately needed in amyotrophic lateral sclerosis, a paralysing neurodegenerative disease that is almost uniformly fatal. It is therefore welcome news that three studies (two in this issue^1,2 and one in Nature Neuroscience³) have converged on a molecular mechanism that seems to underlie a familial form of the disease.

Amyotrophic lateral sclerosis (ALS) is typically sporadic, but around 10% of cases are familial. Although mutations in more than a dozen genes can be involved, those in C9ORF72 are by far the most common cause of ALS, being responsible for approximately 40% of familial cases⁴. Mutations in C9ORF72 occur in a section of DNA comprising six tandemly repeated bases: four guanines (G) followed by two cytosines (C). This G₄C₂ hexanucleotide sequence is typically repeated two or three times, but can be expanded to tens or even thousands of repeats in people with C9ORF72-associated ALS (C9-ALS)^5,6.

There are two leading models proposing how G₄C₂-hexanucleotide-repeat expansion (HRE) leads to neurodegeneration. One, the toxic RNA model, posits that G₄C₂ RNAs transcribed from the expansion bind to crucial RNA-binding proteins or other cellular factors, which prevents them from functioning normally. The other model proposes that, through an unusual form of translation, these expanded RNA molecules produce toxic dipeptide repeat proteins (DPRs) — such as strings of glycine and arginine (GR) amino acids, or of proline and arginine (PR). Central to both hypotheses is the fact that RNAs harbouring the HRE assemble into G-quadruplex structures that confer abnormal molecular behaviours⁷. There is experimental evidence to support both hypotheses^{8,9,10,11,12,13,14,15,16,17}, although it is unclear whether DPRs are expressed at sufficient levels to contribute to toxicity¹⁸.

In the first of the three studies, Zhang et al (page 56).¹ engineered Drosophila melanogaster fruit flies to express HREs of 30 repeats (termed (G₄C₂)₃₀) in the flies' eyes. G₄C₂ expansion in this system causes neurodegeneration and results in a 'rough-eye' trait that can be scored to identify mutations in other genes that lessen or worsen toxicity. Reasoning that proteins that bind to the G₄C₂ RNA are probably mediators of G₄C₂-associated toxicity, the authors crossed flies expressing (G₄C₂)₃₀ with flies harbouring mutations in genes encoding G₄C₂-binding proteins identified through previous biochemical screens^9,10. Mutations that activated a gene called RanGAP strongly suppressed rough eye and neurodegeneration in (G₄C₂)₃₀ flies. The RanGAP protein, which binds to G₄C₂ RNA, is located on the cytoplasmic face of the nuclear membrane, and is part of one of around 2,000 nuclear-pore complexes that control the flow of proteins and RNA in and out of the nucleus. Zhang and colleagues then identified several other nuclear-import genes involved in (G₄C₂)₃₀-elicited neurotoxicity.

Freibaum et al (page 129).² reached similar conclusions using a different strategy. They crossed flies expressing a (G₄C₂)₅₈ HRE with flies missing defined chromosomal segments spanning from tens of genes to more than a hundred, and, using an iterative approach, homed in on genes whose deletion altered G₄C₂ toxicity. The authors identified numerous nuclear-import factors whose inactivation worsened the rough-eye trait of (G₄C₂)₅₈ flies. The comprehensive screen revealed that changes to nuclear-export proteins also enhanced neurodegeneration and rough eye, suggesting that alterations in both nuclear import and export contribute to G₄C₂-mediated toxicity, at least in their system.

Both groups then probed exactly how G₄C₂ HREs disrupt nuclear trafficking. Zhang et al. showed that RanGAP bound G₄C₂ HREs in vitro, and corroborated this finding in vivo using neurons taken from patients with C9-ALS. HRE expression disrupted the nuclear import of fluorescent test substrates and of normal nuclear proteins — most notably TDP-43, which forms misfolded aggregates in the degenerating motor neurons of most people with ALS. Freibaum and colleagues observed nuclear-membrane irregularities in HRE-expressing cells and demonstrated that G₄C₂ HREs inhibit nuclear RNA export, an effect that was relieved by reducing the expression of genes that suppressed G₄C₂-mediated toxicity. Together, these findings established a strong connection between defective nuclear trafficking and neurodegeneration (Fig. 1).

Figure 1: Models of G₄C₂-mediated neurotoxicity.

The neurodegenerative disease amyotrophic lateral sclerosis can be caused by the expansion of a repetitive six-base DNA sequence (GGGGCC; G₄C₂) in the gene C9ORF72. Transcription of this G₄C₂ hexanucleotide repeat expansion (HRE) produces RNA that assembles into harmful G-quadruplex structures that can remain in the nucleus or be exported to the cytoplasm. Three studies^1,2,3 provide evidence suggesting that the G₄C₂ HRE RNA may cause cell toxicity in several ways: by binding to target proteins in nuclear-pore complexes in the nucleus, thereby disrupting nuclear export of RNA; by binding to the nuclear-import protein RanGAP outside the nucleus, preventing normal nuclear import; and by undergoing an abnormal form of translation to form toxic dipeptide repeat proteins (DPRs) that interfere with nuclear import, possibly through interactions with karyopherin proteins. Other effects of G₄C₂ HRE at both the RNA and protein levels remain unclear.

Do DPRs contribute to G₄C₂ toxicity? Both studies detected G₄C₂-derived DPRs, but neither could show whether these DPRs contributed significantly to toxicity. In the third study discussed here, Jovičić et al.³ addressed this point directly, performing a genetic screen to identify genes that lessened or worsened the toxicity caused by a PR₅₀ DPR in the yeast Saccharomyces cerevisiae. Because PR₅₀ was expressed from synthetic DNA and not from a G₄C₂ HRE, toxicity should derive from the DPR itself, rather than from its parent RNA. Six of the strongest suppressors of PR₅₀-associated toxicity in the researchers' screen encoded members of the karyopherin family of nuclear-import proteins. The screen also suggested that the genesis of ribosomes (the cellular machinery that produces proteins) goes awry in PR₅₀-expressing yeast. In the future, even more leads are likely to be mined from these genetic data.

These three studies take us to a higher plane of understanding of C9ORF72-associated ALS, with a focus placed squarely on the nuclear pore. For the future, the newly identified toxicity-suppressing genes will need to be tested in mammalian models of G₄C₂ expansion and DPR toxicity, probably using recently developed mouse strains¹⁹. The findings also raise the question of whether nuclear-trafficking defects contribute to neurotoxicity in other types of ALS. Neurons have a limited ability to replace damaged nuclear-pore complexes, and age-dependent decreases in nuclear integrity have been postulated as a risk factor for ageing-related disease²⁰. Thus, enhancers of nuclear import should be tested in other ALS models, particularly those in which TDP-43 aggregation is observed.

The genetic studies have not resolved whether one mechanism of toxicity predominates in C9-ALS. At face value, the data suggest that G₄C₂-containing RNAs and G₄C₂-derived DPRs elicit toxicity through an overlapping set of nuclear-pore proteins. However, it remains possible that DPRs contribute to neurotoxicity directly in flies. This question could be answered by investigating whether the nuclear-import enhancers picked up in the G₄C₂ screens can rescue neurodegeneration in flies expressing toxic DPRs. It will also be important to further characterize G₄C₂–RanGAP interactions, and to determine whether DPRs bind nuclear-pore proteins. Finally, because both DPRs and G₄C₂ HREs reportedly disrupt a subnuclear structure called the nucleolus^10,13, the relationship between this mechanism and nuclear-membrane defects should be deciphered.

Can our understanding of toxic G₄C₂ RNA be leveraged for therapy? Zhang et al. reversed the rough-eye trait by feeding (G₄C₂)₃₀ flies either a compound that disrupts G₄C₂–RanGAP binding or a small-molecule inhibitor of nuclear export. The three studies also identified other genes that may be 'druggable', including those encoding proteins that oppose RanGAP. Development and preclinical testing of modulators of nuclear import or export is certainly warranted. No doubt, genetic studies such as the three discussed here will identify other nodes of therapeutic interest.

¹