Length-dependent RNA foci formation and Repeat Associated non-AUG dependent translation in a C. elegans G 4 C 2 model.

GC-rich repeat expansion mutations are implicated in several neurodegenerative diseases and can lead to repeat associated non-AUG-dependent (RAN) translation and concentrations of nuclear RNA foci. To model C9orf72 ALS/FTD, we engineered C. elegans to express pure GGGGCC (G 4 C 2 ) repeats of varying lengths and observed RAN translation and nuclear RNA foci. RNA foci were observed in animals expressing ≥20 G 4 C 2 repeats while RAN translation occured in animals expressing ≥33 G 4 C 2 repeats. These findings show that in C. elegans , RAN translation can occur even in the absence of C9orf72 intronic sequence normally surrounding the repeat. Given that the currently accepted repeat threshold for C9 disease is >30 repeats, our data are consistent with a model in which RAN peptides are key drivers of C9orf72 disease pathology. Copyright:

Figure 1. Characterization of a . A) Description of G 4 C 2 expression clones and their expression products. Characteristics of these expression clones are listed above the G 4 C 2 repeat domain. B) In situ hybridization to detect G 4 C 2 -containing mRNA in muscle cell nuclei. Arrowheads point to examples of RNA foci, which were only observed in muscle cell nuclei (dashed circle). Scale bar = 1 micron. C) Quantification of the percentage of animals with observable RNA foci. The number above each bar represents the number of animals examined. **** - p<0.0001, Fisher’s exact test with Bonferroni correction. D) Canonical translation or RAN translation was detected using GFP expression. Note in the ‘+ATG’ lines how GFP expression transitions from soluble to puncta as the number of repeats increase. This is consistent with the production of a (Gly-Ala)x-GFP fusion protein with increasing number of repeats, which is known to form aggregates in C. elegans (Rudich et al. 2017). Without a start ATG, GFP expression is not observed until repeats ≥33, which fits the definition of RAN translation (Zu et al. 2011). All images were taken using identical exposure times. Scale bar = 100 microns. E) Quantification of RAN translated GFP. Data were normalized to the mean of the 5 repeat measurements. Graph shows the mean ± S.D. with individual data points shown. **-p<0.01, ***-p<0.001, ****-p<0.0001, One-way ANOVA with Welch test.

Description

A six nucleotide repeat (GGGGCC; G 4 C 2 ) in the first intron of the C9orf72 gene is the most common genetic cause of Amyotrophic Sclerosis (ALS) and Frontotemporal Dementia (FTD) (DeJesus-Hernandez et al. 2011; Murray et al. 2011; Renton et al. 2011). The formation of nuclear localized repeat-containing RNA foci is a hallmark of C9orf72 pathogenesis (Mizielinska et al. 2013). Moreover, unconventional repeat associated non-AUG dependent (RAN) translation of sense G 4 C 2 and antisense G 2 C 4 repeat containing mRNA gives rise to six distinct dipeptide proteins (DPRs) (Mori et al. 2013), some of which are highly neurotoxic in multiple model systems, including C. elegans (Rudich et al. 2017; Snoznik et al. 2021). While prior studies have shown G 4 C 2 repeats can give rise to RNA foci and RAN translation in C. elegans (Kramer et al. 2016; Sonobe et al. 2021), the contextual requirements for these processes are not defined. For example, how many G 4 C 2 repeats are required for RNA foci formation and RAN translation? Is the sequence surrounding the G 4 C 2 repeat in intron 1 of the C9orf72 sequence required for RAN translation?

To address these questions, we developed transgenic C. elegans expressing defined numbers of pure G 4 C 2 repeats lacking any of the surrounding C9orf72 intron sequence (Figure 1A). To clone these 100% GC repeats, we commercially synthesized a (G 4 C 2 ) 5 clone flanked by unique restriction sites (HindIII/BamHI) and expanded the repeat length using a recursive directional ligation strategy (McDaniel et al. 2010) into the standard C. elegans pPD95.79 GFP expression vector. To achieve repeat sizes ≥33, we discovered that we had to modify the pPD95.79 vector by reversing the origin of replication (Ori), likely due to alterations in the stability of G-rich leading versus lagging strands during replication (Thys and Wang 2015). With the Ori flipped, we were able to clone up to 120 G 4 C 2 repeats, which we verified by restriction digest and DNA sequencing. We cloned a muscle-specific myo-3 promoter (with or without a start ATG) upstream of the repeat. GFP coding sequence was downstream of the repeat and was located in the ‘Glycine-Alanine’ (GA) DPR reading frame. We used these plasmids to generate transgene arrays using standard microinjection approaches. Repeats in vivo appeared stable, since in the ‘+ATG’ clones, we observed robust GFP expression that showed increased aggregate formation with increasing repeat number, consistent with our previous observations of GA-GFP (Rudich et al. 2017). We did not observe notable changes in viability, fecundity, or behavior for any of the lines. We also failed to observe any notable phenotypes when the 120 repeat clone was injected at higher concentrations (100ng/µl). Note that unlike previous C. elegans G 4 C 2 models (Kramer et al. 2016; Sonobe et al. 2021), our models lack any neighboring C9orf72 sequence context.

We first examined the properties of RNA foci formation in Day 1 adults using in situ hybridization to detect sense strand RNA foci (Figure 1B,C). We observed that expression of high repeat numbers gave rise to robust nuclear foci only in the muscle cells expressing the G 4 C 2 mRNA. We failed to detect RNA foci in muscle using a G 4 C 2 antisense probe, suggesting these C. elegans G 4 C 2 lines do not undergo antisense transcription as is observed with native G 4 C 2 repeats in mammals (Zu et al. 2013). We discovered that the threshold for sense RNA foci formation was less than 20 repeats, as no foci formation was observed in the 5 repeat animals, but foci were observed in 20, 33, 50, 70, and 120 repeat animals. Qualitatively, RNA foci appear to be more discreet and numerous as the number of repeats increase, although we note that this could be a consequence of the number of available binding sites for the nucleic acid probe, which is 4 repeats in length.

Next, we examined whether or not the G 4 C 2 transgenes lacking a start ATG supported RAN translation, despite the absence of surrounding C9orf72 intron sequence (Figure 1D,E). All of the G 4 C 2 clones produced robust GFP expression in the presence of a start ATG, indicating all of the clones are intact and capable of supporting translation. However, in the absence of a start ATG, we did not observe GFP expression in lines with 5 or 20 G 4 C 2 repeats. However, in animals expressing 33, 50, 70, and 120 G 4 C 2 repeats, significant GFP expression in muscle cells was observed. The levels of GFP expression in the -ATG lines undergoing RAN translation were qualitatively weaker that those observed in the +ATG lines. Additionally, we found that GFP expression in the -ATG lines rarely formed puncta as observed in the +ATG lines with ≥33 repeats. This could be due to the lower levels of GFP expression in the -ATG lines, since aggregate formation may be concentration dependent (Fung et al. 2003). Alternatively, this could be due to the initiation of translation within the body of the repeat leading to the production of GA repeats below the threshold for aggregation. Nevertheless, our data show that C. elegans exhibit repeat-associated non-ATG dependent translation from a pure G 4 C 2 repeat transgene lacking any additional C9orf72 -specific sequences.

We find that the pure G 4 C 2 sequence is sufficient to support RNA foci formation and RAN translation even in the absence of any flanking C9orf72 sequences in C. elegans . This is significant because all previous C. elegans G 4 C 2 models utilized a C9orf72 ‘minigene’ containing a single repeat length with flanking human intronic sequence. These intronic sequences may play an important role since RAN translation is greatly enhanced by the presence of non-repeat alternative start codons within the C9orf72 intron (Green et al. 2017; Sonobe et al. 2021). Our findings show that even in the absence of these intronic sequences, the pure repeat sequence is sufficient to support both RNA foci formation and RAN translation.

G 4 C 2 RAN thresholds in C. elegans closely match the hypothesized disease threshold in human ALS patients of >30 repeats (DeJesus-Hernandez et al. 2011). This is consistent with the hypothesis that RAN peptides have a primary pathological role in the development of C9 disease. Indeed, we previously found that arginine-rich dipeptide repeats are toxic in both C. elegans and mammals through similar genetic pathways (Snoznik et al. 2021). Overall, our G 4 C 2 models provides a robust visual phenotype for future genetic screens aimed at defining the molecular mechanisms of G 4 C 2 RAN translation. Such screens may provide new pathological insights and treatments targets for the C9orf72 related neurodegenerative diseases.

Methods

Construction of G . A (GGGGCC) 5 sequence flanked by HindIII and BamHI restriction sites was commercially synthesized in the pMA vector (ThermoFisher Scientific). Repeats were expanded using the recursive directional ligation (RDL) strategy to generate pure uninterrupted repeats, as previously described (Mizielinska et al. 2014). Briefly, the origin vector was linearized with BspQI. An insert of (G 4 C 2 ) n was isolated via BspQI/EcoO1091 digestion and inserted into the origin vector. Further rounds of RDL were carried out to generate repeats of the indicated sizes. Repeats were subcloned into the pPD95.79 vector (AddGene) using HindIII/BamHI digestion. This vector was modified to reverse the origin of replication, which we discovered to be necessary to subclone >33 G 4 C 2 repeats. A 2.5Kb myo-3 promoter fragment +/- the start ATG was subcloned into the HindIII site to generate the final expression vector. Final expression clones were verified by Sanger sequencing (Genewiz). Chromatographs were visually inspected to verify the number of GGGGCC repeats. N2 worms were injected with the G 4 C 2 expression plasmid (20 ng/µl) and either rol-6 (80 ng/µl) or myo-2p::mCherry (2.5 ng/µl) as a transformation marker. At least 2 extrachromosomal array lines were analyzed for each repeat construct to verify phenotypic consistency. All strains were grown on NGM plates with OP50 bacteria.

. RNA foci were detected using in situ hybridization using a Cy5-labeled (G 2 C 4 ) 4 locked nucleic acid (LNA) probe (IDT). Day 1 adult worms grown at 25ºC were fixed in 4% paraformaldehyde followed by washes in 70%, 90%, and 100% EtOH. The LNA probe was denatured at 80 ºC for 75 seconds and then added to hybridization buffer (50% formamide, 2X SSC, 50 mM NaPO4, 10% dextran sulfate) at 10 ng/µl. Probes were hybridized overnight at 37 ºC. Worms were then washed twice for 30 minutes with 2X SSC at 37 ºC. Next, worms were washed in 4X SSC with 0.1% Triton X100 containing 1 ug/ml Hoechst 33258, followed by two washes with 2X SSC at room temperature for 5 minutes. Worms were then mounted on slides containing antifade solution. Images were acquired using a Leica DMI4000B widefield microscope and deconvolved using identical settings. Each image is from a deconvolved Z-stack where DNA and RNA foci were observed in the same plane. For quantification, individual animals were scored as either foci positive or foci negative by an observer blinded to genotype. Animals was scored as ‘Foci positive’ if they contained ≥5 foci of Cy5 signal that co-localized with the nuclear DNA signal.

GFP imaging . Worms expressing the indicated number of G 4 C 2 repeats were cultured at 25 ºC. 24 hours after worms reached the L4 stage, animals were anesthetized with 10 mM levamisole. For images, worms were arranged on the surface of a clean NGM agar plate and images were captured using M205FA fluorescence dissecting scope and a DFC345FX digital camera using identical exposure times within the '+ATG' or '-ATG' groups. For quantification, worms were anesthetized in 10mM levamisole and placed on a slide and GFP and DAPI images were captured with a 10X lens and GFP filter (Leica L5 ET set) and DAPI filter (Leica A4 ET set) on a DMI4000B with a DFC 340FX camera (N=8 per genotype, identical settings for all genotypes). To account for autofluorescence, the DAPI image was subtracted from the GFP image. For the resulting subtracted image, GFP fluorescence was quantified along each of three 100 micron lines per animal centered on the midline at the animal’s head (immediately posterior to pharynx), midbody (vulva region), and tail (immediately posterior to intestine). Intensity measurements were summed along and between lines to get a total GFP measurement per animal). Image subtraction and intensity measurements were carried out in Leica Advanced Fluorescence Software, v2.1.0.

Reagents

C. elegans Strains:

+/+; drEx503 [myo-3p+ATG::(G ]

+/+; drEx505[myo-3p+ATG::(G ]

+/+; drEx275 [myo-3p+ATG::(G

+/+; drEx301 [myo-3p+ATG::(G

oxIs322; drEx287 [myo-3p+ATG::(G

oxIs322; drEx298 [myo-3p+ATG::(G

+/+; drEx480 [myo-3p-NO ATG::(G

+/+; drEx477 [myo-3p-NO ATG::(G

+/+; drEx477 [myo-3p-NO ATG::(G

drIs42 [myo-3p-NO ATG::(G

oxIs322; drEx290 [myo-3p-NO ATG::(G

drIs41 [myo-3p-NO ATG::(G

Strain	Genotype
OG1247	+/+; drEx503 [myo-3p+ATG::(G 4 C 2 ) 5 -gfp (20ng/ul); rol-6 (80ng/ul) ]
OG1249	+/+; drEx505[myo-3p+ATG::(G 4 C 2 ) 20 -gfp (20ng/ul); rol-6 (80ng/ul) ]
OG756	+/+; drEx275 [myo-3p+ATG::(G 4 C 2 ) 33 -gfp (20ng/ul); rol-6 (80ng/ul)]
OG794	+/+; drEx301 [myo-3p+ATG::(G 4 C 2 ) 50 -gfp (20ng/ul); rol-6(80ng/ul)]
OG779	oxIs322; drEx287 [myo-3p+ATG::(G 4 C 2 ) 70 -gfp (20ng/ul); rol-6 (80ng/ul)]
OG791	oxIs322; drEx298 [myo-3p+ATG::(G 4 C 2 ) 120 -gfp (20ng/ul); rol-6 (80ng/ul)
OG1148	+/+; drEx480 [myo-3p-NO ATG::(G 4 C 2 ) 5 -gfp (20ng/ul); rol-6 (80ng/ul)]
OG1145	+/+; drEx477 [myo-3p-NO ATG::(G 4 C 2 ) 20 -gfp (20ng/ul); rol-6 (80ng/ul)]
OG759	+/+; drEx477 [myo-3p-NO ATG::(G 4 C 2 ) 33 -gfp (20ng/ul); rol-6 (80ng/ul)]
OG967	drIs42 [myo-3p-NO ATG::(G 4 C 2 ) 50 -gfp (20ng/ul); rol-6 (80ng/ul)]
OG782	oxIs322; drEx290 [myo-3p-NO ATG::(G 4 C 2 ) 70 -gfp (20ng/ul); rol-6 (80ng/ul)]
OG977	drIs41 [myo-3p-NO ATG::(G 4 C 2 ) 120 -gfp (20ng/ul); myo-2p::mCherry (2.5ng/ul); pBluescript (77.5ng/ul)]