Antisense RNA foci in the motor neurons of C9ORF72-ALS patients are associated with TDP-43 proteinopathy
GGGGCC repeat expansions of C9ORF72 represent the most common genetic variant of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia. We and others have proposed that RNA transcribed from the repeat sequence is toxic via sequestration of RNA-binding factors. Both GGGGCC-repeat (sense) and CCCCGG-repeat (antisense) molecules are detectable by fluorescence in situ hybridisation as RNA foci, but their relative expression pattern within the CNS and contribution to disease has not been determined. Blinded examination of CNS biosamples from ALS patients with a repeat expansion of C9ORF72 showed that antisense foci are present at a significantly higher frequency in cerebellar Purkinje neurons and motor neurons, whereas sense foci are present at a significantly higher frequency in cerebellar granule neurons. Consistent with this, inclusions containing sense or antisense derived dipeptide repeat proteins were present at significantly higher frequency in cerebellar granule neurons or motor neurons, respectively. Immunohistochemistry and UV-crosslinking studies showed that sense and antisense RNA molecules share similar interactions with SRSF2, hnRNP K, hnRNP A1, ALYREF, and hnRNP H/F. Together these data suggest that, although sense and antisense RNA molecules might be expected to be equally toxic via their shared protein binding partners, distinct patterns of expression in various CNS neuronal populations could lead to relative differences in their contribution to the pathogenesis of neuronal injury. Moreover in motor neurons, which are the primary target of pathology in ALS, the presence of antisense foci (χ2, p < 0.00001) but not sense foci (χ2, p = 0.75) correlated with mislocalisation of TDP-43, which is the hallmark of ALS neurodegeneration. This has implications for translational approaches to C9ORF72 disease, and furthermore interacting RNA-processing factors and transcriptional activators responsible for antisense versus sense transcription might represent novel therapeutic targets.
KeywordsC9ORF72 Amyotrophic lateral sclerosis RNA foci Dipeptide repeat protein Immunohistochemistry
GGGGCC hexanucleotide repeat expansions in C9ORF72 represent the most common genetic variant of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) [7, 27]. The mechanism of pathogenesis is unknown, but it has been suggested that a gain-of-function toxicity may be mediated via sequestration of RNA recognition motif (RRM) containing proteins by RNA foci [2, 3]. It has been observed that RNA foci are formed, not only from sense, but also from antisense transcription of the repeat expansion [7, 16, 21]. The relative contribution of GGGGCC-repeat (sense) and CCCCGG-repeat (antisense) RNA molecules to disease pathogenesis is unknown, but is likely to have significant implications for subsequent translational research. Work by Haeusler et al.  recently suggested that, with a small number of exceptions, the protein binding partners of the two species of RNA foci are similar.
Another suggested mechanism of pathogenesis is direct toxicity of one or more of five dipeptide repeat proteins (DPRs) translated in different reading frames from either the sense  or antisense  RNA molecules. Poly(Gly-Ala) (GA) and poly(Gly-Arg) (GR) are translated from sense RNA molecules; poly(Pro-Ala) (PA) and poly(Pro-Arg) (PR) are translated from the antisense RNA molecules and poly(Pro-Gly) (PG) is translated from both molecules. Several recent studies have described how these proteins might disrupt ribosomal RNA biogenesis and pre-mRNA splicing [15, 22] or form toxic aggregates . If DPRs are key to pathogenesis, then aberrant nuclear export of repeat RNA sequences, which is necessary to facilitate access to cytoplasmic translation machinery, may be an attractive therapeutic target. We have previously identified interactions between sense RNA repeat sequences and mRNA export adaptor proteins which might have a role in inappropriate licencing for nuclear export .
We conducted extensive immunohistochemistry (IHC) in tissue from C9ORF72-ALS cases to determine the distribution of each species of RNA foci within various CNS neuronal populations known to degenerate in C9ORF72-disease . Blinded examination of serial sections showed that antisense foci are present at a higher frequency in cerebellar Purkinje neurons and motor neurons, whereas sense foci are present at a higher frequency in cerebellar granule neurons. Similar examination in neuronal populations of the hippocampal dentate gyrus and CA4 subfield did not reveal a consistent distinction, with significant variability between cases. Moreover, neuronal inclusions containing DPRs translated from sense RNA are present at a higher frequency in cerebellar granule neurons, whereas neuronal inclusions containing DPRs translated from antisense RNA are present at a higher frequency in motor neurons. Notably, motor neurons are the primary target of pathology in ALS. Furthermore we examined the distribution of RRM-containing proteins predicted to bind one or both of sense and antisense foci with specific attention to colocalisation with antisense RNA foci. Direct and specific binding to the antisense/sense repeat sequence was examined by UV crosslinking using purified recombinant proteins. Finally, we studied the relative association of each species of RNA foci with the hallmark of ALS neurodegeneration, namely mislocalisation of TDP-43 in motor neurons . We add novel insights to this field—in particular our focus on neuropathology has allowed us to contextualize the sense and antisense RNA foci within framework of the human disease.
Materials and methods
Clinical details of C9ORF72+ cases used in pathological analysis
Age at onset (Years)
Disease duration (Months)
Site of onset
Post-mortem delay (h)
A 5′ TYE-563-labelled LNA (16-mer fluorescent)-incorporated DNA probe was used against the sense (Exiqon, Inc.; batch number 607323) and the antisense RNA hexanucleotide repeat (Exiqon, Inc.; batch number 610331). Slides were prepared and RNA foci were visualised as described previously  using a Leica SP5 confocal microscope system with a ×63/1.4 oil immersion objective lens. Briefly prehybridisation was followed by overnight hybridization at 66 °C in a humid atmosphere. A single wash at room temperature with 2 × SSC/0.1 % Tween-20 preceded three washes at 65 °C with 0.1 × SSC. Slides were then mounted in DAPI Vectashield or processed further for dual staining of RNA and protein.
RNA-binding UV-crosslinking assays
RNA-binding assays were carried out as described previously [12, 13]. Recombinant proteins were expressed and purified from E.coli (Supplementary Table 1). Magoh, SRSF2 9-101, ALYREF, hnRNP A1-like2, hnRNP K, and hnRNP F were expressed in E. coli and purified by Ion Metal Affinity Chromatography in 1 M NaCl containing buffers to remove potentially bound RNA from E. coli. hnRNP K was further purified by ion exchange chromatography using a Mono Q column (GE healthcare).
(GGCCCC)5 and (CCCCGG)5 RNAs were 5′ end labelled with [ɣ32P]-ATP using T4 polynucleotide kinase (Fermentas). Reaction mixes were made up in RNA binding buffer [15 mM HEPES pH 7.5, 500 mM NaCl, 5 mM MgCl2, 10 % (v/v) glycerol, 0.05 % (v/v) Tween-20] with 50 ng radiolabelled RNA and 2 µg purified recombinant protein. Mixes were incubated for 10 min at room temperature before being UV irradiated on ice at full power. Complexes were analysed by SDS/PAGE and stained with Coomassie blue before being vacuum dried and exposed on a phosphoimage screen.
The following antibodies were used for IHC anti-TDP-43 (Proteintech 10782-2-AP), anti-hnRNP H/F (Abcam ab10689), anti-hnRNP A1 (Abcam ab5832, 9H10 clone), anti-SRSF2 (Abcam ab11826), anti-ALYREF (Sigma, clone 11G5), anti-nucleolin (Proteintech 10556-1-AP), and anti-hnRNP K (Abcam ab52600). Poly-GA was detected with anti-GA antibodies (mouse, clone 5F2) as previously described . Poly-GR, poly-PA, poly-PR, and poly-PG were detected with antibodies provided by Stuart Pickering-Brown (Proteintech, Manchester, UK). For anti-hnRNP A1 and anti-SRSF2, antigen retrieval was performed by microwaving for 10–30 min in EDTA at pH 8.0. For all other antibodies, antigen retrieval involved 10–20 min microwave in trisodium citrate at pH 6.5 except for anti-hnRNP H/F where no specific antigen retrieval was performed. After incubation with the primary antibodies overnight at 4 °C in DEPC-treated PBS/5 % BSA slides were washed in DEPC PBS and incubated in fluorescent species-specific secondary antibodies. When dual staining of protein and RNA was performed, RNA FISH was performed first after which slides were immediately transferred to PBS/5 % BSA for protein staining.
Relative distribution of sense and antisense RNA foci
Mean and standard deviation (SD) of number of sense and antisense RNA foci per nucleus in Purkinje neurons, granule neurons, motor neurons, dentate gyrus neurons, and CA4 subfield neurons in four C9ORF72-ALS patients
Dentate gyrus neurons
CA4 subfield neurons
Relative distribution of DPRs derived from sense and antisense RNA sequences
Mean and standard deviation (SD) of number of inclusions per cell containing poly-GA and poly-PA protein, in granule neurons and motor neurons from six patients with C9ORF72-disease
In our previous study  we showed that, at a cellular level, there was no significant correlation between the presence of sense RNA foci and the presence of sense RNA-derived poly-GA inclusions. In this study, we examined the relationship between the presence of antisense RNA foci and poly-PA inclusions in fifteen motor neurons from four C9ORF72-ALS cases (Supplementary Table 4). As for the sense species, there was no significant correlation between the two observations (χ2, p = 0.83).
Cellular distribution of RNA foci and RRM-containing proteins
We used confocal microscopy to validate in vivo some of the RRM-containing proteins that were found to interact with (CCCCGG)4 repeat RNA . We and others have previously demonstrated colocalisation of SRSF2, hnRNP A1, hnRNP H/F, and ALYREF with sense RNA foci [3, 17]. We set out to investigate the cellular distribution of the same proteins with respect to antisense RNA foci, and we also examined nucleolin and hnRNP K which are proposed to be specific binding partners of sense and antisense foci, respectively .
Cellular distribution of RNA foci and TDP-43
The precise mechanisms of neuronal injury in C9ORF72-disease appear complex, and are likely to involve RNA gain-of-function toxicity mediated by sense and antisense transcription of the GGGGCC repeat expansion (reviewed in [2, 4]). The small number of cases examined in this study and the wide variability in the phenotype of C9ORF72-related disease prohibits informative comparison between foci distribution and clinical phenotype, but a useful proxy is the pathological hallmark of ALS neurodegeneration: nuclear loss of TDP-43 . We have demonstrated that antisense but not sense foci are significantly associated with nuclear loss of TDP-43 in motor neurons. This intriguing observation suggests that antisense RNA foci may occupy a key position in the cascade of disease pathogenesis. Moreover, examining the differences and similarities between the two species of RNA foci may shed light on important mechanisms leading to neurodegeneration.
We have demonstrated colocalisation of antisense RNA foci with SRSF2, hnRNP A1, hnRNP H/F, ALYREF, and hnRNP K, but not nucleolin. This is consistent with the work of Haeusler et al. . UV-crosslinking studies confirmed that each of these interactions is direct and specific, with the exception of hnRNP K. Conflicting results between the two methodologies may arise because IHC is unable to distinguish between direct and indirect interaction. There is significant potential for indirect binding: many RRM proteins co-exist and interact within nuclear speckles. Notably of the proteins we have examined, SRSF2 colocalisation with both sense  and antisense foci was observed with the highest frequency, and this protein is the core component of nuclear speckles .
Both the IHC and the UV-crosslinking studies in this report suggest that the binding partners of sense and antisense RNA foci are not significantly different. This is also reported by others . Many of these identified binding partners are localised, with SRSF2, to nuclear speckles, nuclear domains implicated in the storage, and supply of splicing factors to active transcription sites . Neuromuscular diseases, including type 1 myotonic dystrophy (DM1), have been associated with depletion of normal components of nuclear speckles [1, 29]. Sequestration of these proteins by sense or antisense RNA foci and consequent disruption of the normal function of these essential nuclear organelles might be a key event in the pathophysiology of C9ORF72-mediated neurodegeneration. If so, our results would predict that both species of RNA foci should be equally toxic. This is consistent with observed toxicity of sense foci in various model systems [8, 16, 17, 21, 28]. This led us to ask whether the key difference might not be in the interactions of the foci themselves, but in the neuronal populations in which sense and antisense foci are expressed.
In all cases, the relative frequency of sense and antisense foci varied consistently and significantly between neuronal populations. Importantly in motor neurons, the primary target of pathology in ALS, antisense foci are more abundant than sense foci. Therefore, we suggest that the key event determining toxicity leading to TDP-43 mislocalisation, of antisense as opposed to sense RNA foci, might be a propensity to produce antisense foci mediated by cell-specific transcriptional regulation. Alternatively sense RNA foci might be degraded at a higher rate than antisense RNA foci. In-vitro studies have suggested that both sense and antisense RNA sequences form complex secondary structures including G-quadruplexes and hairpin loops . These secondary structures may help stabilize the RNA foci and prevent degradation.
It is interesting that two populations of relatively large neurons, motor neurons and cerebellar Purkinje neurons, exhibited antisense RNA foci at a higher frequency than sense RNA foci, in contrast to the smaller cerebellar granule neurons. This suggests that our observations may be related to some property correlated with neuronal size. However, in the hippocampus, neither the larger CA4 subfield neurons nor the smaller dentate granule neurons exhibited either species of RNA foci at a consistently higher frequency.
We observed antisense foci in the cytoplasm of motor neurons, which is consistent with aberrant nuclear export and may be a key step in the facilitation of proposed repeat associated non-ATG translation to produce DPR species . We made a similar observation with respect to sense foci  and we suggest that interaction between repeat RNA and mRNA export adaptors, such as ALYREF, might override the normal nuclear retention of pre-mRNA species. Recent studies consistent with a key role for DPRs in the pathogenesis of C9ORF72-mediated neurodegeneration [15, 20, 22] suggest that this represents an attractive therapeutic target.
We have demonstrated that the frequency of sense and antisense foci is usually correlated i.e. a patient with more sense foci will also have more antisense foci. This is particularly interesting in case 1 from our analysis (Table 2) who displayed a relatively high frequency of sense and antisense RNA foci in the cerebellum and CA4 subfield neurons of the hippocampus, which are both extra-motor areas. Case 8 also exhibited a relatively high frequency of sense and antisense RNA foci in CA4 subfield neurons; the frequency of RNA foci in the cerebellum of case 8 was not quantified. In contrast to the other cases examined these patients displayed extra-motor disease clinically as well as pathologically: clinical FTD was diagnosed with (case 1) and without (case 8) ALS (Table 1). This is consistent with a correlation between the development of RNA foci in specific neuronal subtypes and clinical presentation, but this hypothesis will require validation in a larger number of FTD and ALS cases.
Finally, varying frequency of the expression of sense and antisense repeat RNA has implications for the formation of specific DPRs. Our observations of all five DPRs are consistent with our conclusions relating to the expression of sense and antisense RNA foci. In cerebellar granule neurons, where sense RNA foci are more abundant, there is a higher frequency of sense-RNA derived DPR inclusions; and in motor neurons where antisense RNA foci are more abundant, there is a higher frequency of antisense RNA-derived DPR inclusions. Therefore, we suggest that, at least in these neuronal populations, translation of the sense and antisense derived proteins occurs in different quantities depending of the relative availability of RNA repeat molecules. The results of the present study potentially explain the observations of others that inclusions containing poly-GA protein are much more abundant in certain neuronal populations including cerebellar granule neurons . However, Davidson et al.  failed to demonstrate antisense RNA derived DPR inclusions in Purkinje neurons of the cerebellum and dentate gyrus neurons of the hippocampus. This contrasts with our demonstration of antisense RNA foci in both of these populations, particularly in the cerebellar Purkinje neurons which we found to show a preference for exhibiting antisense rather than sense RNA foci. This variation between neuronal populations might be explained by variability in control of nuclear export of repeat RNA species; in this context, it is interesting that mutations in hGle1, a mRNA export adaptor, have recently been shown to cause selective death of motor neurons .
In our previous study we showed that there was no significant correlation between the presence or absence of nuclear sense RNA foci in cerebellar granule neurons and whether or not those cells contain a cytoplasmic inclusion positive for poly-GA . Similarly in this study we have shown that there is no significant correlation between the presence or absence of nuclear antisense RNA foci in motor neurons, and whether or not those cells contain an inclusion positive for poly-PA. This suggests that our population level conclusion in this study, that neuronal populations have a propensity to produce either sense or antisense RNA derived foci and DPRs, does not apply at a cellular level. Thus, whilst both RNA foci and DPRs are derived from the same RNA molecules, the processes by which this RNA is stabilised into a focus or exported for translation are probably different or even mutually exclusive—indeed work from Gendron et al.  suggested that this might be the case. So if motor neurons have a preference for antisense transcription of the C9ORF72 expansion then the motor neuron population will express higher levels of antisense RNA derived foci and DPRs, but not necessarily within the same individual cells.
Our work highlights that any therapeutic approach to C9ORF72-ALS should consider the presence of antisense RNA foci in motor neurons. An antisense oligonucleotide approach has been proposed as a therapeutic option in C9ORF72-disease [8, 28]. We suggest that both sense and antisense RNAs should be targeted, as has been proposed by others . Indeed, in relation to the relative selective vulnerability of motor neurons in C9ORF72-mediated pathology in vivo, targeting the antisense foci may be even more important than targeting sense foci. A limitation of our study is the reliance on post-mortem tissue which represents end-stage disease and may exclude the most vulnerable cells which have already been lost; as such we await validation of our findings in model systems.
We acknowledge grants from the European Community’s Seventh Framework Programme (FP7/2007-2013) under the EuroMOTOR project, Grant Agreement No 259867 and the EU Joint Programme—Neurodegenerative Disease Research (JPND) projects, SOPHIA and STRENGTH, supported through the following funding agencies under the aegis of JPND—www.jpmd.eu: United Kingdom, Medical Research Council to PJS and JK. PJS is an NIHR Senior Investigator. JCK and JRH are supported by MND Association/MRC Lady Edith Wolfson Fellowship awards ([MR/K003771/1] and [G0 800380] respectively). Biosample collection was supported by the MND Association and the Wellcome Trust (PJS). We are grateful to all of the patients with ALS and control individuals without neurological disease who donated biosamples for research.
Conflict of interest
The authors declare that they have no conflict of interest.
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed consent was obtained for all individual participants included in the study.
- 6.Davidson YS, Barker H, Robinson AC, Thompson JC, Harris J, Troakes C et al (2014) Brain distribution of dipeptide repeat proteins in frontotemporal lobar degeneration and motor neurone disease associated with expansions in C9ORF72. Acta Neuropathol Commun 2:70PubMedCentralPubMedCrossRefGoogle Scholar
- 9.Gendron TF, Bieniek KF, Zhang YJ, Jansen-West K, Ash PE, Caulfield T et al (2013) Antisense transcripts of the expanded C9ORF72 hexanucleotide repeat form nuclear RNA foci and undergo repeat-associated non-ATG translation in c9FTD/ALS. Acta Neuropathol 126:829–844PubMedCentralPubMedCrossRefGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.