The ALS/FTD-related C9orf72 hexanucleotide repeat expansion forms RNA condensates through multimolecular G-quadruplexes

Abstract

Amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) are neurodegenerative diseases that exist on a clinico-pathogenetic spectrum, designated ALS/FTD. The most common genetic cause of ALS/FTD is expansion of the intronic hexanucleotide repeat (GGGGCC)_n in C9orf72. Here, we investigate the formation of nucleic acid secondary structures in these expansion repeats, and their role in generating condensates characteristic of ALS/FTD. We observe significant aggregation of the hexanucleotide sequence (GGGGCC)_n, which we associate to the formation of multimolecular G-quadruplexes (mG4s) by using a range of biophysical techniques. Exposing the condensates to G4-unfolding conditions leads to prompt disassembly, highlighting the key role of mG4-formation in the condensation process. We further validate the biological relevance of our findings by detecting an increased prevalence of G4-structures in C9orf72 mutant human motor neurons when compared to healthy motor neurons by staining with a G4-selective fluorescent probe, revealing signal in putative condensates. Our findings strongly suggest that RNA G-rich repetitive sequences can form protein-free condensates sustained by multimolecular G-quadruplexes, highlighting their potential relevance as therapeutic targets for C9orf72 mutation-related ALS/FTD.

Fig. 1. G-quadruplexes structural features and mechanistic hypothesis.

a Structural characteristics of G-quadruplexes—The G4 scaffold is composed of a G-tetrad, a planar tetrameric assembly of guanine (G) bases held together by Hoogsteen H-bonding. G4-tetrads stack to form a typical G4-structure, which is stabilised by monovalent cations  that can coordinate the G-tetrad, which is dependent on the size of the cation (Li⁺» Na⁺> K⁺). G-tetrads can assemble into G4-structures either within the same DNA/RNA strand (unimolecular) or by condensation of multiple independent DNA/RNA strands (multimolecular, i.e., tetramolecular from four strands). b Proposed mechanism of aggregation in ALS/FTD pathological aggregates—The (GGGGCC)_n repeat sequence can aggregate in the absence of proteins due to the formation of multimolecular G4s (mG4s).

Fig. 2. DNA (GGGGCC)_n forms mG4s at different repeat lengths.

a (GGGGCC)_n CD spectra—(GGGGCC)_n (n = 4, 6, 9, 12) were annealed at 500 µM, 250 µM, 100 µM and 50 µM under mG4-forming conditions (see methods) and further diluted to 5 µM prior CD analysis. All the samples present a positive peak at about 263 nm and a negative peak at 240 nm that can be associated to a parallel G4 conformation. b (GGGGCC)_n agarose gel electrophoresis—SYBR safe stain. (GGGGCC)_n (n = 2-11) were annealed at 250 µM under mG4-forming conditions (see methods) and the gel was stained with SYBR safe. This gel includes the same samples used for the following NMM staining displayed in (d) and has been added for reference to aid in the visualisation of the location of the bands. c (GGGGCC)₁₁ agarose gel electrophoresis—(GGGGCC)₁₁ was annealed in the presence of different KCl concentrations. As KCl concentration increases, so does the molecularity of the species involved, indicating that the formed multimolecular structures have a strong KCl dependence. In the schematics, the green ball represents the K⁺ ion. On the right, the same samples are stained with NMM, while on the left the sequences are FAM labelled. For reference a 100 base pairs ladder has also been added, which was stained with SYBR safe. d (GGGGCC)_n agarose gel electrophoresis—NMM stain. (GGGGCC)_n (n = 2-11) were annealed at 250 µM in mG4-forming conditions. The lanes containing any of the (GGGGCC)_n were fluorescent upon staining with NMM, implying the presence of G4s in the higher molecularity species formed under these conditions. The absence of staining for both the ladder and the non-G4 ssDNA control confirm the specificity of the dye for G4-containing species.

Fig. 3. DNA (GGGGCC)_n aggregates in a G4-dependent fashion.

a (GGGGCC)_n aggregates—Brightfield imaging of (GGGGCC)_n (n = 2–12) annealed under mG4-forming conditions at 250 µM (see methods). 100 µm scalebar. Micrographs in (a) are representative from n = 2 independent replicates. b FAM labelled (GGGGCC)₁₀—FAM-labelled (GGGGCC)₁₀ was annealed under mG4-forming conditions (see methods). The image is split to show the brightfield channel (on the right) and the FAM (on the left). 10 µm scalebar. Data for (b) is a representative micrograph from n = 2 independent replicates. c (GGGGCC)₆ NMM staining experiment—(GGGGCC)₆ was annealed under mG4-forming conditions and stained with NMM (see methods). The image is split to show the brightfield channel (on the right) and the NMM fluorescence channel (on the left). Data from (c) is a representative micrograph from n = 2 independent replicates. 10 µm scalebar. d (GGGGCC)_n phase diagram in KCl—(GGGGCC)_n was annealed under mG4-forming conditions. At lower annealing concentrations and lower repeat lengths, the sequence does not present the ability to aggregate. The higher the repeat length and the higher the concentration, the more likely it is to observe aggregation. r(GGGGCC)_6,10 were also annealed under mG4s forming conditions and presented aggregation at lower concentrations than their DNA equivalents. A colour gradient is utilised in the phase diagram to symbolise a higher likelihood of finding aggregates for a given sequence. e (GGGGCC)₆ NMM photooxidation experiment—Upon laser excitation the NMM dye bound to the mG4s photo-oxidises the guanine in the quadruplex, leading to the disassembly of the structure. The region of interest highlighted in the figure (ROI) was irradiated for 1 min and the second image was subsequently acquired, revealing disassembly of the aggregates within the ROI due to photooxidation. The image is split to show the brightfield channel (on the left) and the NMM (on the right). 10 µm scalebar. Data from (e) is a representative micrograph from n = 2 independent replicates.

Fig. 4. Role of base pairing and monomolecular G4s in the aggregation mechanism.

a Circular Dichroism spectra of mutants—Controls of n = 6 were annealed under mG4-forming conditions (see methods). Sequences forming G4-structures present the characteristic peaks at ~263 nm and 240 nm, whilst the (CCCCGG)_n control fails to generate any significant CD signal. b Mutants agarose gel electrophoresis—SYBR safe stain—Controls of n = 6 were annealed under mG4-forming conditions. c (GGGGCC)_n agarose gel electrophoresis—NMM stain—Controls of n = 6 were annealed under mG4-forming conditions. All G4-forming sequences ((GGGGAA)₆, (GGGGTT)₆, (GGGGCC)₆) presented fluorescence bands upon NMM staining, confirming the presence of G4-structures as observed by CD. Ladder, non-G4 ssDNA control, and (CCCCGG)₆ do not yield any NMM staining and are not visible in the gel, confirming the specificity of the dye for G4-containing species. d Aggregation of the mutant sequences—Controls of n = 6 were annealed under mG4-forming conditions. Aggregates were only detected under confocal microscopy for the (GGGGAA)₆ and (GGGGCC)₆ sequences, which are the only ones able to form mG4s. 100 µm scalebar.

Fig. 5. (GGGGCC)₁₀ aggregation in presence of 10 µM PDS.

(GGGGCC)₁₀ was annealed at 150 µM under mG4-forming conditions. The sample containing the G4-ligand presents less aggregates and of a smaller size with respect to the non-PDS containing controls. 100 µm scalebar. The micrographs are representative micrographs of n = 2 independent experiments.

Fig. 6. (GGGGCC)_n ability to aggregate in its RNA form and to interact with TDP-43 RRM1-2.

a RNA (GGGGCC)_n aggregates in a G4-dependent fashion, r(GGGGCC)₆ NMM staining experiment—r(GGGGCC)₆ was annealed under mG4-forming conditions (see methods). It was then subjected to NMM staining prior to acquisition. The image is split to show the brightfield channel (on the right) and the NMM fluorescence channel (on the left). 10 µm scalebar. b r(GGGGCC)₆ NMM photo-oxidation experiment—Upon laser excitation the NMM dye bound to the mG4s photo-oxidises the guanine in the quadruplex, leading to the disassembly of the structure (see methods). The region of interest (ROI) highlighted in the figure was irradiated for 1 min and the second image was subsequently acquired, showing condensate disassembly due to photo-oxidation. 10 µm scalebar. Micrographs from panels a and b are representative of n = 2 respective independent replicates. c Brightfield and NMM-stained imaging of TDP-43 RRM1–2 and (GGGGCC)₁₁—(GGGGCC)₁₁ was annealed under mG4-forming conditions (1–300 µM) and then incubated with 10 µM TDP-43 RRM1–2, revealing a synergist aggregation effect between the (GGGGCC)n repeats and TDP-43. 100 µm scalebar. Micrographs in panel c are representative of 3 technical replicates and n = 2 independent biological replicates.

Fig. 7. C9orf72 mutant iPSC-derived motor neurons exhibit an increased number of G4 foci per cell soma and increased percentage of G4 foci within the nucleus, when compared to healthy controls.

a Confocal representative images of iPSC-derived motor neurons labelled with nuclear stain DRAQ5 (red) and NMM (yellow). Upon staining with NMM, both control and C9orf72 mutant cells exhibit fluorescent G4 foci. 50 μm scale bar. b Quantification of the number of NMM foci per cell soma in control and mutant motor neurons. C9orf72 mutant cells exhibit a significant increase in the number of NMM foci per cell. P-value = 0.0140 from two-tailed Mann-Whitney U test. c Quantification of the percentage of NMM foci in the nucleus in control and mutant motor neurons. Mutant cells exhibit a significant increase in the percentage of NMM foci within the nucleus. P-value = 0.0038 from unpaired two-tailed t test. Data for panels b and c are from n = 2 healthy and n = 2 mutant cell lines examined over four repeats with nine fields of view per repeat and presented as mean ± SD, *p  <  0.05, **p  <  0.01.