G‐quadruplex‐binding small molecules ameliorate C9orf72 FTD/ALS pathology in vitro and in vivo

1Expansions of a GGGGCC repeat within the first intron of the C9orf72 gene are the most common genetic cause of both amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD), two rapidly progressive and incurable neurodegenerative disorders (Dejesus‐Hernandez et al, 2011; Renton et al, 2011). While the general population carries < 30 GGGGCC (G4C2) repeats, with approximately 90% of individuals carrying < 8 repeats, large hexanucleotide repeat expansions (HRE), typically between 800 to > 4,000, are causative of ALS and FTD (Beck et al, 2013; van Blitterswijk et al, 2013). HREs are transcribed and the resulting RNA forms nuclear foci and can also be translated in all reading frames into dipeptide repeat proteins (DPRs) through a non‐canonical process termed repeat‐associated non‐ATG (RAN) translation (Ash et al, 2013; Gendron et al, 2013; Lagier‐Tourenne et al, 2013; Mizielinska et al, 2013; Mori et al, 2013a,b; Zu et al, 2013). Both repeat RNA and DPRs have been proposed to drive pathogenesis: foci can sequester RNA‐binding proteins (RBPs) and therefore impair their function (Haeusler et al, 2016), while DPRs have been proven to be toxic in numerous disease models (Mizielinska et al, 2014; Wen et al, 2014; Zhang et al, 2016).

2G4C2 RNA can fold to form the highly stable non‐canonical G‐quadruplex (G‐Q) conformation (Fratta et al, 2012), a four‐stranded structure formed by the stacking of planar tetrads of four non‐sequential guanosine residues (G‐quartets). RNA G‐Qs are able to form in vivo (Biffi et al, 2014), are enriched in RNA 5′ and 3′ UTRs (Huppert et al, 2008) and are known to be involved in regulating numerous RNA functions, including splicing, RNA transport and translation (Simone et al, 2015). As G‐Qs can directly affect translation (Bugaut & Balasubramanian, 2012), and G4C2 G‐Qs have been shown to specifically sequester disease‐relevant RBPs (Haeusler et al, 2014; Conlon et al, 2016), they may play an important role in both RNA foci and DPR toxicity.

3Small molecules binding to both DNA and RNA G‐Qs have been identified (Di Antonio et al, 2012), and due to the different conformation of RNA and DNA G‐Q molecules, ligands preferentially targeting RNA G‐Qs have also been developed (Biffi et al, 2014). Identification of molecules that specifically target C9orf72 repeat RNA could have therapeutic potential by shielding pathogenic interactions of the C9orf72 expanded RNA with RBPs, and/or by interfering with RAN translation. We report here a screen identifying molecules with selectivity for the G4C2 G‐Q RNA and show they are able to reduce both RNA foci formation and RAN translation in C9orf72 iPSC‐neuron models and C9orf72 repeat‐expressing flies.

4In order to identify small molecules that preferentially stabilise RNA G4C2 G‐Qs, we adapted a FRET‐based G‐Q melting assay (Guyen et al, 2004; Collie et al, 2012), to specifically report G4C2 G‐Q stabilisation. We have previously identified several novel G‐Q‐binding chemotypes in the chemical library from the anti‐parasitic drug discovery programme at Georgia State University based on non‐conjugated aromatic diamidines (Ohnmacht et al, 2014). Here, we screened 138 small molecules, 104 from this library and 34 previously established G‐Q‐binding compounds (Schultes et al, 2004; Moore et al, 2006). We measured their ability to stabilise (G4C2)4 oligonucleotides composed of either RNA or DNA folded into G‐Qs. 44/138 small molecules increased the melting temperature (Tm) of (G4C2)4 RNA by greater than 13°C. Twelve of these showed at least 5°C greater stabilisation of (G4C2)4 RNA than (G4C2)4 DNA (Fig 1A). Strikingly, three of these molecules (DB1246, DB1247, DB1273, green circles in Fig 1A) had very similar chemical structures, differing by only two atoms (Fig 1B), indicating a genuine structure–function relationship. No other compounds in the compound set have similar features of two linked five‐membered rings.

5We therefore took these three small molecules forward for further testing. A stabilisation dose response for both sense (G4C2)4 (Fig 1C) and antisense (Appendix Fig S1) (G2C4)4 RNA and DNA oligonucleotides confirmed the preferential stabilisation of RNA G4C2 G‐Qs by DB1246, DB1247 and DB1273. Circular dichroism (CD) spectroscopy confirmed that the (G4C2)4 RNA formed the expected parallel G‐Q structure, with a minimum at 236 nm and a maximum at 264 nm. Each of the three small molecules caused the appearance of a characteristic additional induced CD signal in a separate region of the spectrum (350–550 nm; Fig 1D), while the small molecules alone gave no CD signal (Appendix Fig S2), thus confirming direct binding of DB1246, DB1247 and DB1273 to (G4C2)4 RNA G‐Qs. We derived the Tm from our CD denaturation curves, which confirmed that (G4C2)4 RNA G‐Qs were stabilised in the presence of each small molecule (Appendix Fig S3). These compounds bind to the RNA G4C2 repeat G‐Q with high affinities, with Kd values in the range ca 200–400 nM (measurements by fluorescence anisotropy, Appendix Fig S4). These data identify, using FRET, CD and fluorescence anisotropy techniques, three structurally related small molecules DB1246, DB1247 and DB1273 that bind and stabilise RNA G4C2 G‐Qs.

6We next determined whether the small molecules could alleviate the key pathologies associated with C9orf72 G4C2 repeat expansion in patient‐derived iPSC‐neurons. We first characterised three patient iPSC lines (described in Appendix Table S1). We confirmed the presence of G4C2 repeat expansions by Southern blotting, which were maintained following differentiation into either motor or cortical neurons (Appendix Fig S5). Cortical neurons were prepared using an established differentiation protocol (Shi et al, 2012; Sposito et al, 2015). Spinal motor neurons were generated using dual‐SMAD and GSK3β inhibition followed by caudal and ventral patterning to the pMN domain and finally promoting cell cycle exit using a Notch antagonist (Hall et al, 2017), yielding 90% pure motor neuron cultures (Fig EV1). The efficiency of differentiation did not differ between C9orf72 and control lines (Fig EV1). We also confirmed that RNA foci were specifically observed in patient iPSC‐motor and iPSC‐cortical neurons (Appendix Fig S6). We next performed a dose response in one patient iPSC‐cortical neuron line to investigate the effect of DB1246, DB1247 and DB1273 on RNA foci formation. For each of the small molecules, a 1 μM treatment for 4 days reduced RNA foci burden (Fig EV2). We therefore treated cortical neurons derived from all three independent C9orf72 repeat expansion iPSC lines with 1 μM of DB1246, DB1247 or DB1273 for 4 days. Each small molecule significantly reduced RNA foci burden by approximately 50% (Fig 2). The same treatment on iPSC‐motor neurons derived from the three independent patient lines showed that DB1246 and DB1273 reduced RNA foci burden, again by approximately 50%, while DB1247 did not significantly reduce RNA foci burden. These data show that small molecules that bind RNA G4C2 G‐Qs can reduce RNA foci in both iPSC‐motor and iPSC‐cortical neurons.

7We next addressed whether DB1246, DB1247 or DB1273 could reduce the other major pathology in C9FTD/ALS—dipeptide repeat proteins. We established an MSD ELISA for poly(GP) and showed that poly(GP) is specifically detected in C9orf72 repeat expansion iPSC‐motor and iPSC‐cortical neurons (Appendix Fig S7). Treatment with 1 or 4 μM of DB1246, DB1247 or DB1273 for 4 days did not reduce poly(GP) levels in iPSC‐motor neurons (Appendix Fig S8), indicating a differential response of RNA foci and poly(GP) to the small molecules. We therefore investigated 7‐day treatments with a range of concentrations (8, 12 and 16 μM). We focused on iPSC‐motor neurons due to their shorter differentiation protocol compared to cortical neurons. We found that the two small molecules that reduced RNA foci in iPSC‐motor neurons, DB1246 and DB1273, also significantly reduced poly(GP) levels (Fig 3A). DB1273 was the most effective, significantly reducing poly(GP) at all concentrations, with greater than 50% reduction at 16 μM. Importantly, expression levels of C9orf72 transcripts were not affected by the same treatment (Fig 3B), indicating a direct effect on G4C2 repeat G‐Q RNA, rather than a more general effect on transcription. We also measured the expression levels of MCM2 as it has a G‐Q within its core promoter region (Huppert & Balasubramanian, 2007), and its expression is reduced by the G‐Q‐binding small molecule TMPyP4 (Liu et al, 2014). The levels of MCM2 were unaffected (Fig 3B), indicating specificity of the small molecules for G4C2 RNA G‐Qs. We assessed the toxicity of DB1246, DB1247 and DB1273 in iPSC‐motor neurons using a concentration range from 0.05 to 40 μM. Importantly, no toxicity was observed in the dose range that reduces poly(GP) levels (Fig 3C). A similar toxicity profile was also observed in human fibroblasts (Appendix Fig S9). These data show that small molecules that bind RNA G4C2 G‐Qs can specifically reduce the levels of dipeptide repeat proteins generated by endogenous RAN translation of G4C2 repeats in cultured patient neurons, without causing toxicity.

8We treated Drosophila with adult‐onset ubiquitous expression of 36 G4C2 repeats (Mizielinska et al, 2014) for 7 days with the most effective small molecule, DB1273, by feeding it to adult flies in liquid food. While food intake was no different across treatment groups (Appendix Fig S10), DB1273 led to a significant and dose‐dependent decrease in poly(GP) levels (Fig EV3A and B). Liquid food did not support the flies long enough to determine whether DB1273 was also able to ameliorate the decreased survival of adult (G4C2)36 flies. We therefore delivered DB1273 to larvae in solid food, since effects on survival can be observed more rapidly in larvae and larvae eat more than adult flies, allowing delivery of an increased amount of compound. Constitutive, ubiquitous expression of (G4C2)36 in Drosophila larvae is toxic, with 24% of larvae dying before the pupal stage (Fig 3D). Treatment of first‐instar larvae with DB1273 for 7 days led to a significant increase in survival to the pupal stage (Fig 3D). We have shown that poly(GR) is the DPR responsible for toxicity in our G4C2 repeat expansion flies (Mizielinska et al, 2014). We therefore developed a poly(GR) MSD immunoassay (Fig EV4). Poly(GR) was specifically detected in (G4C2)36 larvae and treatment with DB1273 significantly decreased poly(GR) levels by 33% (Fig 3E). The level of expression of the G4C2 transgene in treated larvae was not affected (Appendix Fig S11) indicating a specific effect on RAN translation rather than an effect on the transgene itself. In addition, the same treatment in control larvae did not cause toxicity (Appendix Fig S12). We used the inherent fluorescence of DB1273 to examine the biodistribution of DB1273 in our treated larvae. DB1273 was broadly distributed throughout the gut, including the epithelium, while in other tissues, including the central nervous system, infrequent fluorescent puncta were consistently observed in all three replicates (Fig EV5). Overall, these data show that small molecules that bind RNA G4C2 G‐Qs can reduce dipeptide repeats levels in vivo and importantly this leads to functional benefit.

9We identify and characterise three small molecules that share structural similarity and the ability to bind and stabilise G4C2 RNA G‐Qs. We demonstrate that these molecules can reduce the frequency of nuclear RNA foci, and the levels of DPRs in C9orf72 patient iPSC‐derived neurons. Furthermore, we provide evidence of in vivo efficacy by showing that they can reduce DPRs and improve the survival of GGGGCC repeat‐expressing Drosophila. Importantly, we observed a significant reduction in poly(GR), the DPR responsible for toxicity in our fly model (Mizielinska et al, 2014) as well as other models (Moens et al, 2017), and thus likely more clinically relevant than poly(GP). We note that despite limited biodistribution, the small molecule DB1273 was still able to exert a small but significant beneficial in vivo effect on survival. This provides proof of principle for targeting G‐Q RNA in C9FTD/ALS and indicates that optimising CNS penetrance will lead to even greater efficacy.

10DNA and RNA C9orf72 HREs have been shown to form both G‐Qs and hairpin structures (Fratta et al, 2012; Reddy et al, 2013; Haeusler et al, 2014; Su et al, 2014; Sket et al, 2015). It is widely accepted that HREs exert a toxic gain of function, either linked to RNA‐mediated sequestration of RBPs in RNA foci, or through translation into toxic DPRs (Mizielinska & Isaacs, 2014). Previous work has shown that targeting the hairpin conformation of HREs, through small molecules, is a successful approach to reduce both RNA foci and DPRs (Su et al, 2014). Here, we demonstrate significant reductions of both RNA foci and DPRs are also possible by targeting the HRE G‐Q conformation.

11The effect of our small molecules on G4C2 RNA in patient neurons is consistent with the presence of RNA G‐Q structures in vivo, as shown using G‐Q‐specific antibodies (Biffi et al, 2014), including specifically in C9orf72 patient lines (Conlon et al, 2016). Interestingly, it has been reported that most predicted RNA G‐Qs are unfolded in eukaryotic cells (Guo & Bartel, 2016). As the C9orf72 HRE represents a very favourable sequence for G‐Q formation, our small molecules may enhance and stabilise folding of G4C2 RNA into G‐Qs in vivo, as has been reported for other RNA G‐Q‐binding ligands (Biffi et al, 2014). Small molecule‐mediated stabilisation of G4C2 RNA G‐Qs could decrease RNA foci either through facilitating their degradation or masking the HRE RNA from the FISH probe, with either scenario indicating effective target engagement. The decrease we observe in DPRs is consistent with reports that G‐Qs in the 5′ UTRs of several genes inhibit translation, with inhibition enhanced by the addition of G‐Q‐binding small molecules (Bugaut & Balasubramanian, 2012; Biffi et al, 2014). While our compounds were able to reduce both RNA foci and poly(GP) DPRs, the effect on foci was more rapid and obtained at lower small‐molecule concentrations. This may be due to the high stability of poly(GP), which requires 10 days of anti‐C9orf72 antisense oligonucleotide treatment to be significantly reduced in C9orf72 iPSC‐neurons (Gendron et al, 2017).

12RNA and DNA G‐Qs have numerous differences in terms of both conformation, binding partners and regulation, and small molecules have been developed to differentiate between them (Di Antonio et al, 2012). We successfully adopted a DNA/RNA parallel screening approach to identify small molecules preferentially targeting HRE RNA. In support of preferential binding to RNA G‐Qs, we observed no impact on transcription levels of MCM2, a transcript with a DNA G‐Q in its core promoter, which was previously shown to be reduced by TMPyP4 (Liu et al, 2014), a generic G‐Q‐binding molecule that can also bind G4C2 RNA in vitro (Zamiri et al, 2014). In summary, these results provide proof of principle for the further development of drugs that selectively bind G4C2 RNA G‐Qs as a therapeutic strategy for C9FTD/ALS.

13The FRET assay was performed as previously described (Guyen et al, 2004; Collie et al, 2012). Briefly, DNA and RNA dual‐labelled oligonucleotides (5′‐FAM and 3′‐TAMRA) of sequence (G4C2)4 and (G2C4)4 (IDT, Leuven, Belgium) were initially dissolved as a 100 μM stock in water and diluted to 1 μM in 10 mM sodium cacodylate buffer (pH 7.35) and annealed by heating to 98°C for 10 min, followed by cooling to room temperature in the heating block. Small molecules were stored as 10 mM stock solutions in DMSO; final solutions (2× concentrations) were prepared by dilution in sodium cacodylate buffer (pH 7.35). 96‐well plates (Bio‐Rad) were prepared by aliquoting 50 μl of the annealed oligonucleotide into each well, followed by 50 μl of the compound solutions, which were tested at both 1 and 2 μM. Fluorescence measurements were made on a DNA Engine Opticon (MJ Research) with excitation at 450–495 nm and detection at 515–545 nm. Fluorescence readings were taken at intervals of 0.5°C over the range 30–100°C, with a constant temperature being maintained for 30 s prior to each reading to ensure a stable value. Final analysis of the data was carried out using a custom script written in Origin 7.0 (OriginLab Corp., Northampton, MA).

14Circular dichroism experiments were performed at temperatures between 15°C and 95°C, with a 1°C/min temperature gradient, using a Jasco J715 spectropolarimeter (Jasco Hachioji, Tokyo, Japan) equipped with a Jasco peltier temperature control system. The CD spectrum from 180 to 600 nm was measured for 1 μM pre‐folded (G2C4)4 RNA, either alone or in the presence of each small molecule (2 μM) in 10 mM sodium cacodylate (pH 7.35). The T m was calculated by fitting the curve to the Van't Hoff equation using Grafit 5 (Erithacus Software). A CD spectrum of the buffer was recorded and subtracted from all raw signals before plotting using GraphPad Prism 5.

15Primary fibroblast lines were generated from skin biopsies, which were obtained under informed consent. Ethical permission for this study was obtained from the National Hospital for Neurology and Neurosurgery and the Institute of Neurology joint research ethics committee. Fibroblasts were reprogrammed as previously described (Sposito et al, 2015). iPSCs were cultured in feeder‐free conditions on Geltrex‐coated plates in Essential 8 medium (Thermo Scientific). iPSCs were passaged with 0.5 mM EDTA (Thermo Scientific). Cortical neuron differentiation was as previously described (Shi et al, 2012; Sposito et al, 2015). Briefly, iPSCs were plated at 100% confluency and the media switched to neural induction media (1:1 mixture of N‐2 and B‐27‐containing media supplemented with the SMAD inhibitors Dorsomorphin and SB431452 (Tocris)). Medium consists of DMEM/F‐12 GlutaMAX, 1× N‐2, insulin, L‐glutamine, non‐essential amino acids, β‐mercaptoethanol, penicillin, streptomycin, Neurobasal, 1× B‐27 (Thermo Scientific). Media were changed daily during neural induction, and at the end of the induction period (day 10–12), the converted neuroepithelium was replated onto laminin‐coated plates using dispase (Thermo Scientific) and maintained in neural maintenance media (a 1:1 mix of N‐2 and B‐27) which was replaced every 2–3 days. At around days 25–35, neuronal precursors were passaged further with accutase (Thermo Scientific) and plated for the final time at day 35 onto poly‐ornithine and laminin‐coated plates (Sigma), before being used in experiments between days 55 and 65. Motor neuron differentiation was as recently described (Hall et al, 2017). Briefly, iPSCs were plated at 100% confluency and the media switched to motor neuron induction media (1:1 mixture of N‐2 and B‐27‐containing media supplemented with the SMAD inhibitors Dorsomorphin and SB431452 (Tocris) and GSK‐3β inhibitor CHIR99021). After a 6‐day induction period, the converted neuroepithelium was replated onto laminin‐coated plates using dispase (Thermo Scientific) and maintained in a 1:1 mix of the described N‐2 and B‐27 media supplemented with 0.5 μM retinoic acid (Sigma) and 1 μM Purmorphamine (Calbiochem) for a further 7 days, then maintained in a 1:1 mix of the described N‐2 and B‐27 media supplemented with 0.1 μM Purmorphamine (Calbiochem) for 4 days. Motor neuron precursors were replated onto laminin‐coated plates using EDTA (Thermo Scientific) and maintained in a 1:1 mix of the described N‐2 and B‐27 media supplemented with 0.1 μM compound E (Millipore) before being used at day 30.

16iPSC‐derived neurons were fixed in 4% methanol‐free formaldehyde (Pierce) for 10 min at room temperature, dehydrated in a graded series of alcohols, air‐dried and rehydrated in phosphate‐buffered saline (PBS), permeabilised for 10 min in 0.1% Triton X‐100 in PBS, briefly washed in PBS and incubated for 30 min in pre‐hybridisation solution (40% formamide, 2× SSC, 1 mg/ml tRNA, 1 mg/ml salmon sperm DNA, 0.2% BSA, 10% dextran sulphate, 2 mM ribonucleoside vanadyl complex) at 67°C. Hybridisation solution (40% formamide, 2× SSC, 1 mg/ml tRNA, 1 mg/ml salmon sperm DNA, 0.2% BSA, 10 % dextran sulphate, 2 mM ribonucleoside vanadyl complex, 0.2 ng/μl (C4G2)4 LNA probe, 5′ TYE563‐labelled, Exiqon) was incubated with the cells for 2 h at 67°C. Cells were washed for 30 min at 67°C with high‐stringency buffer (2× SSC, 0.1% Triton X‐100) and then for 20 min each, in 0.2× SSC buffer. Nuclei were stained by DAPI. Coverslips were then dehydrated in 70% and 100% EtOH and mounted onto slides in Vectashield for iPSC‐cortical neurons or Dako mounting medium for iPSC‐motor neurons. Images were acquired using an LSM710 confocal microscope (Zeiss) using a plan‐apochromat 40×/1.4 NA oil immersion objective.

17Five to ten z‐stacks were acquired for each field and at least four independent fields were imaged. For iPSC‐cortical neurons, maximum intensity projections of each z‐stack were analysed in Fiji‐ImageJ and nuclear RNA foci automatically counted by using the analyse particles function to identify nuclei and the find maxima function to identify RNA foci in each nucleus. For iPSC‐motor neurons, blinded manual counting of maximum intensity projections was performed. At least 70 neurons were counted for each independent differentiation.

18A poly(GP) Meso Scale Discovery (MSD) immunoassay was established using our previously generated rabbit anti‐poly(GP) antibody (Mizielinska et al, 2014). A poly(GR) MSD immunoassay was established using newly generated affinity purified rabbit anti‐(GR)7 antibodies (Eurogentec). The assays were performed as previously described (Su et al, 2014). Briefly, iPSC‐neurons were lysed in RIPA buffer with protease inhibitors (Roche complete mini EDTA‐free) and then sonicated using a Soniprep 150 (Renaissance Scientific Limited) probe sonicator. Lysates were centrifuged at 16,000 g to remove insoluble material. Drosophila L3 larvae were frozen in liquid nitrogen, and two larvae per replicate were homogenised in ice‐cold RIPA buffer (Sigma) with protease inhibitors (Roche complete mini EDTA‐free). Lysis was allowed to proceed on ice for 10 minutes, before lysates were centrifuged at 21,000 g for 20 min at 4°C and supernatant was collected in fresh tubes. The protein concentration of the lysates was determined using the DC protein assay (Bio‐Rad) following manufacturer's instructions and equal amounts of protein used in the immunoassay. Capture was performed with either unlabelled anti‐poly(GP) or anti‐poly(GR) antibodies. Detection on an MSD sector imager utilised either sulfo‐tagged anti‐poly(GP) antibody or biotinylated anti‐poly(GR) antibody followed by sulfo‐tagged streptavidin. For iPSC‐neurons, prior to analysis, the average reading from a calibrator containing no peptide was subtracted from each reading. For Drosophila larvae, an individual wild‐type (w1118) value was chosen at random and subtracted from all samples to correct for background, and resultant negative values were considered to be 0. A four‐parameter logistic regression curve was fit to the values obtained using the peptide calibrators using GraphPad Prism, and concentrations interpolated. Specificity was confirmed with a peptide cross‐reactivity assay using (GP)7, (PR)7 or (GR)7 synthetic peptides (Biogenes) at a concentration of 100 ng/ml. The lower limit of detection was calculated after fitting a four‐parameter logistic regression curve using the MSD workbench 4.0 software.

19iPSC‐motor neurons in 96‐well plates were treated for 7 days with 0, 0.1, 0.2, 0.3, 0.6, 1.3, 2.5, 5, 10, 20 and 40 μM of DB1246, DB1247, DB1273 or cisplatin as a positive control. Cell viability was measured using the XTT assay kit II (Roche) according to the manufacturer's instructions and absorbance measured with a Tecan Spark 10M plate reader.

20iPSC‐derived motor neurons were treated with each of the three small molecules (DB1246, DB1247, DB1273) at 16 μM final concentration for 7 days and compared to the DMSO vehicle. Total RNA was extracted in duplicate for each condition using TRIzol (Invitrogen) and treated with DNase I (Roche). cDNA was synthesised using 1 μg of total RNA for all samples, with a SuperScript III first‐strand cDNA synthesis kit (Invitrogen) and an equimolar mixture of oligo dT and random hexamer primers. Real‐time qRT–PCR was carried out using Power SYBR Green Master Mix (Applied Biosystems) for MCM2. TATA‐binding protein (TBP) and Glyceraldehyde 3‐phosphate dehydrogenase (GAPDH) were used as housekeeping genes to normalise across different samples. Additionally, expression of three alternative C9orf72 transcript variants (V1, V2, V3) was measured using two custom‐designed fluorescent LNA PrimeTime® probes (IDT) and previously described isoform‐specific primers (Fratta et al, 2013). The amplified transcripts were quantified using the comparative Ct method and presented as normalised fold expression (ΔΔC t). Oligonucleotide sequences are provided in Appendix Table S2.

21All fly stocks were maintained at 25°C on a 12‐h:12‐h light:dark cycle at 60% constant humidity and on standard sugar–yeast–agar (SYA) medium (agar, 15 g/l; sugar, 50 g/l; autolyzed yeast, 100 g/l; nipagin, 100 g/l; and propionic acid, 2 ml/l). A fly line expressing 36 GGGGCC repeats (36R), under the upstream activating sequence (UAS) promoter, was used (Mizielinska et al, 2014). Expression in adults was driven using the inducible daughterless‐GeneSwitch (da‐GS) driver which was kindly provided by Veronique Monnier (Tricoire et al, 2009). Expression in larvae was driven with the constitutive daughterless‐GAL4 (daGAL4) driver (Bloomington Drosophila Stock Center).

22Two days after eclosion, adult daGS>36R flies were fed with liquid food administered via capillaries containing 100 μM RU486 to induce expression of the repeats, or with no RU486 as a control, for 7 days. In addition, flies induced with RU486 were treated with vehicle or DB1273 at 200 μM or 500 μM concentrations. Flies were housed at a density of five flies per vial in plastic vials filled with 2 ml of 1% agar, to ensure humid conditions. Vials were sealed with Parafilm perforated with four holes using a 26‐G needle to ensure adequate air circulation. Two graduated 10‐μl disposable glass capillary tubes were held in place through the Parafilm using truncated 200‐μl pipette tips. The liquid food consisted of 5% (wt/vol) sucrose, 2% yeast extract and blue food dye to aid visualisation of the liquid in the capillaries. A mineral oil overlay was used at the top of the capillary to minimise evaporation. An identical chamber without flies was included to determine evaporative losses. Capillaries were replaced with fresh liquid food as required, and the volume of food ingestion was measured over the course of 7 days. After 7 days, flies were snap‐frozen in liquid nitrogen for analysis of abdominal poly(GP) levels by immunoblotting.

23To investigate DB1273 toxicity in vivo, control daughterless‐GAL4 (daGAL4)/+ first‐instar larvae were placed into vials containing 90% SYA food, supplemented with vehicle (water and DMSO at a final concentration of 0.3%) or DB1273 to a 0.5 or 1 mM concentration, at a density of 50 larvae per vial, and the number reaching the pupal and adult stage of development were counted. To investigate DB1273 efficacy, daGAL4>36R first‐instar larvae were placed into vials containing 90% SYA food, supplemented with either vehicle or DB1273 to a 1 mM concentration, at a density of 50 larvae per vial. The number reaching the pupal stage of development was counted 7 days later by an observer blinded to the experimental conditions. Third‐instar larvae treated with vehicle or DB1273 to a 1 mM concentration were frozen for RNA and protein analysis.

24Fly abdomens were homogenised in 20 μl per abdomen of 2× Laemmli sample buffer (4% SDS, 20% glycerol, 120 mM Tris–HCl pH 6.8, 200 mM DTT with bromophenol blue) and heated to 95 °C for 10 min. Samples were centrifuged at 21,000 g for 2 min, and supernatants were separated on Nu‐PAGE 4‐12% Bis–Tris gels in MES running buffer, and then transferred onto nitrocellulose membranes in a Tris–Glycine transfer buffer with 20% methanol. Membranes were incubated with rabbit anti‐poly(GP) (Mizielinska et al, 2014) 1:1,000 or mouse anti‐α‐tubulin, Sigma T6199, 1:2,000 followed by horseradish peroxidase‐tagged secondary antibody (anti‐rabbit HRP, ab6721 or anti‐mouse HRP, ab6789, Abcam, 1:10,000). Specific binding was detected with Luminata Forte (Millipore) and imaged with an ImageQuant LAS4000 (GE Healthcare Life Sciences). Intensity of bands was quantified using Fiji‐ImageJ Software.

25For iPSC‐neuron small‐molecule effects on RNA foci and poly(GP) levels, we normalised to the control value for each replicate (independent induction) and performed a one‐sample t‐test of treatment replicates versus the control value of 1. For comparison of motor neuron differentiation efficiency between C9orf72 and control iPSCs, a Mann–Whitney U‐test was performed due to unequal variances. For Drosophila small‐molecule treatments, poly(GP) levels were normalised to the loading control, α‐tubulin, and a one‐way ANOVA performed with Dunnett's post hoc test across the treatment groups. For comparing lethality/viability in Drosophila, a chi‐square test was used. For quantitative PCR, a t‐test was used for comparison of two groups and a one‐way ANOVA with Dunnett's post hoc test for comparison of more than two groups. Statistical analyses were performed in GraphPad Prism 5.

26RS, RB, TGM, RP, AH, ELC, NSW, TN, AJN, JMP, WDW, DWB, HZ, LP, SW, GP, SN, RP, PF and AMI designed and/or interpreted aspects of the different experiments. RS, RB, TGM, EP, KMW, NSW, TN, JGJ, SA, MC, M‐TK, AJN, JSM, MAI, CES, TV, AAF, ACa, ACh, HH, SW and PF performed experiments. RS, RB, PF and AMI wrote the manuscript with input from all co‐authors.

27The authors declare that they have no conflict of interest.

28Amyotrophic lateral sclerosis and frontotemporal dementia are devastating degenerative diseases with no treatments currently available. A mutation in the C9orf72 gene is the most common cause of both amyotrophic lateral sclerosis and frontotemporal dementia, accounting for approximately 10% of all cases. The C9orf72 mutation is termed a repeat expansion as it consists of six DNA bases, GGGGCC, that are repeated thousands of times. The GGGGCC repeat DNA is transcribed into repeat RNA, which we have previously shown folds into a distinct secondary structure termed a G‐quadruplex. The repeat RNA is then translated into repetitive proteins, termed dipeptide repeat proteins, which are extremely neurotoxic. Therefore, reducing the levels of these dipeptide repeat proteins is a potential therapeutic strategy.

29We performed a screen to identify compounds that specifically target the C9orf72 RNA G‐quadruplex. We identified three compounds with similar chemical structures that were able to selectively bind the C9orf72 repeat G‐quadruplex RNA. We then treated C9orf72 patient‐induced pluripotent stem cell (iPSC)‐derived neurons, and fruit flies harbouring the C9orf72 repeat expansion, with these compounds. We found that the compounds reduced the levels of the damaging dipeptide repeat proteins in both the iPSC‐neurons and the fruit flies. Furthermore, they improved the survival of the C9orf72 repeat fruit flies.

30These results provide proof of principle that targeting the C9orf72 GGGGCC repeat RNA G‐quadruplex is a potential treatment strategy for amyotrophic lateral sclerosis and frontotemporal dementia caused by C9orf72 repeat expansion.