C9orf72 Hexanucleotide Repeat Expansion

C9orf72 (Chromosome 9 open reading frame 72) is a protein-coding locus on human chromosome 9p21.2 (11 exons spanning ~27.5 kb) that encodes a 481-amino-acid (~54 kDa) differentially expressed in normal and neoplastic cells (DENN)-domain protein required for endolysosomal trafficking, macroautophagy initiation, and innate immune homeostasis. In non-neuronal myeloid cells and neurons alike, the C9orf72 protein forms a stable heterotrimeric complex (CSW complex) with Smith–Magenis syndrome chromosomal region candidate gene 8 (SMCR8) and WD repeat-containing protein 41 (WDR41). Within this C9orf72–SMCR8–WDR41 holoenzyme, the tandem DENN domains of C9orf72 and SMCR8 act as a GTPase-activating protein (GAP) and guanine nucleotide exchange regulator for small Rab (Rab8a, Rab11a, Rab39b) and ARF (ARF1, ARF6) GTPases, while WDR41 anchors the complex to lysosomal membranes via the cationic amino acid transporter PQLC2 and coordinates ULK1-dependent autophagosome maturation and STING type I interferon degradation.

In 2011, Rosa Rademakers, Bryan Traynor, and colleagues discovered that a non-coding hexanucleotide repeat expansion (HRE) of the sequence (GGGGCC)n (abbreviated G4C2) located within Intron 1 of C9orf72 (between non-coding Exons 1a and 1b) accounts for ~30–40% of familial amyotrophic lateral sclerosis (fALS), ~25% of familial frontotemporal dementia (fFTD), and ~5–10% of apparently sporadic ALS and FTD in populations of European ancestry. While unaffected individuals carry 2 to 24 (GGGGCC)n repeats (with 2, 5, and 8 repeats representing the most common alleles), affected C9-ALS/FTD patients harbor hundreds to >4,000 repeats (400 bp to >25 kb) accompanied by pronounced somatic mosaicism between peripheral blood leukocytes and post-mortem cerebral cortex or spinal motor neurons.

Promoter Alternative Splicing and C9orf72 Haploinsufficiency

The human C9orf72 locus undergoes alternative promoter usage and alternative splicing to generate three principal mRNA transcript variants (V1, V2, and V3):

• Variant 1 (V1, NM_145005): Initiates at Exon 1a upstream of the (GGGGCC)n repeat and splices to an alternative donor site in Exon 5, encoding a short 222-amino-acid isoform (C9-S, ~24 kDa).
• Variant 2 (V2, NM_018325): Initiates at Exon 1b immediately downstream of the (GGGGCC)n repeat (accounting for ~85–90% of total C9orf72 transcripts in healthy human brain) and encodes the full-length 481-amino-acid protein (C9-L).
• Variant 3 (V3, NM_001256054): Initiates at Exon 1a upstream of the repeat, transcribes directly through Intron 1 containing the (GGGGCC)n expansion, and encodes the full-length 481-amino-acid protein.

When the Intron 1 (GGGGCC)n tract expands beyond ~65–100 repeats, the 100%-GC-content sequence forms cotranscriptional DNA–RNA hybrid R-loops and stable intramolecular/intermolecular G-quadruplex (G4) structures that stall RNA polymerase II elongation and recruit repressive histone H3 lysine 9 (H3K9me3) and DNA methylation across the upstream CpG island. Consequently, transcription from the major downstream Exon 1b promoter (V2) is suppressed on the expanded allele, reducing total full-length C9orf72 protein levels by ~40–50% (haploinsufficiency) in patient frontal cortex and spinal cord—impairing lysosomal clearance of aggregated proteins and sensitizing microglia to hyperactive STING innate immune signaling.

Bidirectional G-Quadruplex RNA Foci and RBP Sequestration

Although V2 transcription is reduced, bidirectional transcription initiating from Exon 1a (V1/V3 sense strand) and from an antisense promoter downstream of the repeat generates both sense (GGGGCC)n RNA and antisense (CCCCGG)n (C4G2) RNA. Inside the nucleus, sense (GGGGCC)n transcripts fold into parallel four-stranded RNA G-quadruplexes stabilized by Hoogsteen hydrogen-bonded guanine tetrads and monovalent K⁺ ions, while antisense (CCCCGG)n transcripts form i-motif and C-rich hairpin structures.

These structured repeat RNAs undergo multivalent liquid–liquid phase separation to form nuclear RNA foci that sequester essential RNA-binding proteins (RBPs), including hnRNP H (which binds G4 motifs to regulate cassette exons), SRSF1 (which mediates NXF1-dependent nuclear export of repeat RNAs into the cytoplasm), Pur-α, ADARB2, and the nucleolar chaperone nucleolin (NCL), causing widespread transcriptome-wide splicing defects prior to neuronal loss.

Repeat-Associated Non-AUG (RAN) Translation and Nucleocytoplasmic Collapse

Once SRSF1 and the TREX/NXF1 pathway export spliced Intron 1 lariats or unspliced V1/V3 sense and antisense repeat RNAs into the cytoplasm, the stable G4 and hairpin secondary structures stall scanning 43S pre-initiation ribosomal complexes. Even in the complete absence of a canonical AUG start codon, stalled ribosomes initiate Repeat-Associated Non-AUG (RAN) translation (often utilizing near-cognate CUG or AGG codons in the upstream flanking sequence and enhanced by integrated stress response eIF2α phosphorylation) across all six sense and antisense reading frames, synthesizing five dipeptide repeat proteins (DPRs): poly-GA, poly-GP, poly-GR, poly-PR, and poly-PA.

The two arginine-rich species—poly-GR and poly-PR—carry a high positive charge density that mimics arginine-rich nuclear localization signals (R-motifs) and interacts via cation–π and electrostatic forces with low-complexity aromatic domains of biomolecular transcriptional condensates and phase-separated organelles:

1. Nuclear Pore Complex (NPC) Blockade: Poly-PR and poly-GR bind the phenylalanine-glycine (FG) hydrogel meshwork of central channel nucleoporins (Nup54, Nup62, Nup98) and sequester Importin-α/Importin-β (Karyopherin-β1), collapsing the Ran-GTP/Ran-GDP nucleocytoplasmic gradient and mislocalizing nuclear TDP-43 (TARDBP) into the cytoplasm—where TDP-43 forms phosphorylated aggregates and triggers cryptic exon inclusion in STMN2 (stathmin-2) and UNC13A.
2. Nucleolar and Ribosomal Arrest: Poly-PR partitions into the granular component of the nucleolus, displacing nucleophosmin (NPM1), suppressing rRNA biogenesis, and binding directly to the polypeptide exit tunnel of the 80S ribosome to arrest global protein synthesis.

Because C9orf72 pathology combines gain-of-toxic-function (sense and antisense repeat RNAs plus DPRs) with loss-of-normal-function (V2 haploinsufficiency), therapeutic design requires allele- or transcript-selective targeting. In Phase I/II clinical trials (BIIB078 / tominersen-class and WVE-004), gapmer antisense oligonucleotides (ASOs) engineered to bind Intron 1 upstream of the (GGGGCC)n repeat successfully lowered cerebrospinal fluid poly-GP levels by >50–80% via RNase H1-mediated cleavage of sense V1/V3 transcripts; however, these first-generation sense-only ASOs failed to improve clinical ALSFRS-R outcomes, likely because they spared toxic antisense (CCCCGG)n RNA / poly-PR, or partially lowered endogenous C9orf72 protein levels. Current next-generation modalities focus on dual sense/antisense ASOs, small-molecule G4 binders, RAN translation inhibitors (PKR/eIF2B modulators), and AAV delivery of STMN2 or TDP-43-responsive splicing switches.

How can an intronic repeat that is spliced out of the mRNA still be translated into toxic proteins?

Under normal splicing, introns are excised as branched lariat RNAs, debranched by DBR1, and rapidly degraded inside the nucleus. In C9orf72 expansion carriers, however, the (GGGGCC)n G-quadruplex tract stalls spliceosomal processing—causing retention of Intron 1 in mature polyadenylated V1/V3 transcripts—and also stabilizes excised circular intronic lariats against exonucleolytic decay. Both intron-retained mRNAs and stabilized intronic RNAs bind SRSF1 and exit into the cytoplasm for RAN translation.

Why must C9orf72 antisense oligonucleotides target Intron 1 rather than coding exons?

Over 85% of functional full-length C9orf72 protein is translated from Variant 2 (V2), which initiates transcription at Exon 1b downstream of the Intron 1 (GGGGCC)n repeat and therefore lacks the expansion entirely. If an ASO targets coding Exons 2–11, RNase H1 destroys healthy V2 transcripts from both alleles, inducing severe C9orf72 loss-of-function neuroinflammation. Targeting Intron 1 selectively degrades repeat-containing V1/V3 pre-mRNAs while preserving V2.

Why is poly-GP measured in clinical trials if poly-GR and poly-PR are the most neurotoxic DPRs?

Arginine-rich poly-GR and poly-PR are highly basic, insoluble, and rapidly sequestered inside neuronal nucleoli, ribosomes, and stress granules, resulting in sub-femtomolar concentrations in lumbar cerebrospinal fluid (CSF). By contrast, poly-GP (translated from both sense and antisense strands) is uncharged, metabolically stable, and freely soluble, allowing ultrasensitive immunoassay quantification in CSF as a direct pharmacodynamic readout of target engagement.

1. DeJesus-Hernandez, M., Mackenzie, I. R., Boeve, B. F., et al. (2011). "Expanded GGGGCC hexanucleotide repeat in noncoding region of C9ORF72 causes chromosome 9p-linked FTD and ALS." Neuron, 72(2), 245–256. https://doi.org/10.1016/j.neuron.2011.09.011
2. Renton, A. E., Majounie, E., Waite, A., et al. (2011). "A hexanucleotide repeat expansion in C9ORF72 is the cause of chromosome 9p21-linked ALS-FTD." Neuron, 72(2), 257–268. https://doi.org/10.1016/j.neuron.2011.09.010
3. Mori, K., Weng, S.-M., Arzberger, T., et al. (2013). "The C9orf72 GGGGCC repeat is translated into aggregating dipeptide-repeat proteins in FTLD/ALS." Science, 339(6125), 1335–1338. https://doi.org/10.1126/science.1232927
4. Balendra, R., & Isaacs, A. M. (2018). "C9orf72-mediated ALS and FTD: multiple pathways to disease." Nature Reviews Neurology, 14(9), 544–558. https://doi.org/10.1038/s41582-018-0047-2