Screening a UK amyotrophic lateral sclerosis cohort provides evidence of multiple origins of the C9orf72 expansion

An expanded hexanucleotide repeat in the C9orf72 gene is the most common genetic cause of amyotrophic lateral sclerosis and frontotemporal dementia (C9ALS/FTD). Although 0–30 hexanucleotide repeats are present in the general population, expansions >500 repeats are associated with C9ALS/FTD. Large C9ALS/FTD expansions share a common haplotype and whether these expansions derive from a single founder or occur more frequently on a predisposing haplotype is yet to be determined and is relevant to disease pathomechanisms. Furthermore, although cases carrying 50–200 repeats have been described, their role and the pathogenic threshold of the expansions remain to be identified and carry importance for diagnostics and genetic counseling. We present clinical and genetic data from a UK ALS cohort and report the detailed molecular study of an atypical somatically unstable expansion of 90 repeats. Our results across different tissues provide evidence for the pathogenicity of this repeat number by showing they can somatically expand in the central nervous system to the well characterized pathogenic range. Our results support the occurrence of multiple expansion events for C9ALS/FTD.


Introduction
Expansions of a GGGGCC hexanucleotide repeat in the first intron and/or promoter region of C9orf72 are the most frequent known monogenic cause of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) in populations of European descent (DeJesus-Hernandez et al., 2011;Gijselinck et al., 2012;Majounie et al., 2012;Renton et al., 2011). Repeats ranging in size from 0 to 30 are found in the general population, whereas pathogenic expansions range between 500 and 4500 repeats, with considerable somatic instability DeJesus-Hernandez et al., 2011;van Blitterswijk et al., 2013b). The finding of a conserved haplotype spanning approximately 200 kb around pathogenic expansions may result from either a common founder, or a "risk" haplotype predisposing to expansion Majounie et al., 2012;Mok et al., 2012;Pliner et al., 2014). How the C9orf72 expansion leads to disease is as yet unknown, proposed mechanisms include the following: (1) loss of C9orf72 function; (2) RNA toxicity associated with the presence of nuclear repeat expansion RNA foci; and (3) protein toxicity caused by dipeptide repeats originating from the translation of the repeat expansion (Ling et al., 2013).
We present genetic and clinical data from C9orf72 screening in a UK ALS cohort. We confirm the clinical features previously associated with C9orf72 expansions and report an atypical expansion with approximately 90 repeats in blood which shows somatic instability within and between tissues. Our analysis of multiple tissues shows the repeat significantly expanded in the central nervous system (CNS) to over 3000 hexanucleotides.
Our data, together with recently reported cases, strongly support the hypothesis that C9orf72 repeat expansions occur in different individuals in the context of the permissive risk haplotype (Dols-Icardo et al., 2014;Van Blitterswijk et al., 2013b). This finding sheds light on the origins of the C9orf72 repeat expansion and has important implications for pathomechanisms of C9orf72 ALS/FTD, diagnostic testing, and genetic counseling for C9orf72 expansions.

Methods
We screened a cohort of 452 ALS patients who had not been screened for other genetic causes of ALS, from University College London Partners Motor Neuron Disease clinics (London, UK). Patients gave written consent, and the project was approved by the local ethical review committee. DNA was extracted from blood and flash-frozen tissues using standard techniques.
Fibroblasts were generated from a 3-mm skin punch biopsy taken under local anesthetic following informed consent. Biopsies were dissected into 1 mm pieces and cultured at 37 C, 5% CO 2 in DMEM, 10% FBS, 1% L-Glutamine, 50U/mL penicillin, and 50 mg/mL streptomycin until fibroblasts were seen to grow out from the explants. Media changes were performed every 3 days. When fibroblasts reached confluency, they were detached from culture dishes using TrypleE (Invitrogen) and transferred to larger culture vessels for further expansion. Cells (1 Â 10 7 ) were used for genomic DNA extraction using the standard techniques.
Repeat-primed polymerase chain reaction expansion screening and single-nucleotide polymorphism (SNP) genotyping was performed on blood-derived DNA as previously described Majounie et al., 2012). Genetic screening data for 350 patients were previously reported . Two Southern hybridization methods for expansion sizing were performed using: (1) an oligonucleotide probe consisting of 5 GGGGCC repeats ; or (2) a single-copy 1 kb probe  that anneals adjacent to the repeat (see Supplementary Table 1 for PCR-primer sequences used to derive the single-copy probe). Three different restriction enzyme digestions were used in the single-copy probe method to produce "normal" bands of different sizes: EcoRI/BamHI double digest: 2.4 kb; BsU36I: 6.2 kb; EcoRI: 8 kb (Fig. 1A); the 6.2 kb and 8 kb bands compress the large somatic smears in expansion-positives for unambiguous detection, whereas the 2.4-kb band allows accurate sizing.

Clinical features and C9orf72 screening of a UK ALS cohort
We screened C9orf72 hexanucleotide repeats in 452 ALS patient blood-derived DNA samples using repeat-primed polymerase chain   reaction and revealed expansions of >30 repeats in 30 patients (6.6%), hereafter referred to as C9þ individuals. The mean age at onset of C9þ cases was 54.6 years compared with 59.3 years in noncarriers. Site of disease onset was bulbar in 26% of C9þ cases compared with 27% in noncarriers (Table 1).

Sizing of C9orf72 repeat reveals one exception to very large expansions
Southern hybridizations were performed on blood DNA from 27 C9þ cases (19 previously reported) see Supplementary Fig. 1 for details on the additional 8) and confirmed the presence of somatically unstable expanded repeats with sizes between 1100 and 4200. One exception was case 61776 which showed a doublet indicating a small expansion of the C9orf72 repeat . To better discriminate the size of the expanded repeat, we performed single-copy probe Southern blotting and established that this individual carried an expansion of approximately 90 repeats in blood. The EcoRI/BamHI double digest, electrophoresed as a smear indicating somatic instability (70e120 repeats) (Fig. 1B).

Case 61776 shows typical C9orf72 molecular and histopathologic features in postmortem brain
The clinical features of case 61776, outlined in Supplementary Material were typical of ALS, with no family history of the disease, bulbar onset at 57 years, and a 6-year disease course before death.
Postmortem neuropathologic analysis revealed mild frontal atrophy and a discoloration of the anterior horn in the spinal cord. Numerous p62-positive and TDP-43 negative star-like neuronal cytoplasmic inclusions were seen in the granule cells of the dentate fascia, hippocampal subregions, and cerebellar cortex ( Fig. 2A and  B). These were shown to contain the dipeptide proteins ( Fig. 2H and I). An occasional TDP-43-positive inclusion was seen in oligodendrocytes of the pencil fibers of the striatum and substantia nigra ( Fig. 2D and F). The neurons of the XII cranial nerve nucleus were depleted with a single TDP-43 neuronal cytoplasmic inclusion seen Fig. 2. Immunohistochemical analysis of case 61776. Small p62-positive "star-like" inclusions were present in the granule cells of the dentate fascia (A) and the cerebellar cortex (B). Larger p62-positive inclusions were seen in the CA4 subregion of the hippocampus and also contained intranuclear inclusions (C). TDP-43-positive oligodendroglial inclusions were seen in the striatum (D) and substania nigra (F). TDP-43-positive neuronal cytoplasmic inclusions were present in the 12th nerve nucleus (E) and spinal cord (G). The "star-like" p62positive inclusions were also positive with anti-glycine proline (H) and anti-glycine arginine dipeptide antibodies (I). Bar represents 50 mm in A and B; 20 mm in E, G, H, and I; and 10 mm in C, D, and F. (Fig. 2E and G). Thoracic spinal cord showed severe depletion of the anterior horn neurons together with degeneration of both the crossed and uncrossed corticospinal tracts. Occasional TDP-43positive cytoplasmic inclusions were observed in surviving motor neurons along with fine neuropil threads and occasional oligodendroglial cells with fine filamentous intracytoplasmic inclusions.
RNA foci containing sense and antisense C9orf72 expansion transcripts are a characteristic feature of C9orf72 ALS/FTD. To investigate the presence of RNA foci in this case, fluorescence in situ hybridization was performed. We detected abundant sense and antisense foci in the frontal cortex (Fig. 3) and also in the hippocampus, cerebellum, and spinal cord (data not shown).

An expansion of the C9orf72 hexanucleotide repeat has occurred in the CNS
We performed Southern blots on DNA derived from postmortem flash-frozen CNS samples from case 61,776. Frontal cortex, cerebellum, and spinal cord-derived DNA showed very large C9orf72 repeat expansions (approximately 950e3000þ repeats), not differing from typical C9orf72 cases, with the exception of the presence in the spinal cord DNA of a band representing the 90 repeat expansion together with the larger repeat (Fig. 4A).
In the absence of parental DNA, to answer whether the hexanucleotide repeat had expanded specifically in the CNS, or whether there had been a retraction in a regenerating tissue such as blood, where a selective pressure can be envisaged, we also analyzed the C9orf72 repeat size in tissues and cell types with diverse embryological derivations, including dura mater (mesoderm or a proposed dual mesoderm-ectoderm origin) (Adeeb et al., 2012;Kaplan et al., 2005), cauda equina (ectoderm), and skinderived fibroblasts (mesoderm) (Fig. 4A and B). All samples showed the presence of the smaller 90 repeat expansion similar to that in blood (mesoderm). These results suggest the repeat has expanded in size in the developmental lineage after the differentiation between central and peripheral nervous system and making the possibility of a retraction to the same size in multiple tissues less likely.

Assessing the threshold for somatic repeat expansion instability
Previous reports have indicated that 20e30 hexanucleotide repeat expansions, commonly referred to as "intermediate" repeats, play a pathogenic role (Gómez-Tortosa et al., 2013). To assess if similar somatic instability occurs in these repeats, we performed Southern blotting on 8 samples with 20e27 repeat expansions and showed no detectable instability in this size range (Fig. 4C).

A permissive haplotype and multiple expansion events
We then assessed if case 61776 carried the risk haplotype found to associate with all known cases of C9orf72 ALS/FTD and performed genetic analysis of 36 of the 42 SNPs that define this "Finnish risk" haplotype (Majounie et al., 2012). Results showed that case 61776 carried the risk allele on a consecutive stretch of 22 SNPs spanning 105 kb upstream to 26 kb downstream of the C9orf72 gene (Supplementary Table 2), therefore not differing from other C9orf72 positive cases.

Atypical C9orf72 repeat expansions occur at low frequency in blood
To estimate the frequency of atypical repeat expansions, we reviewed all C9orf72 Southern blotting data published until the beginning of March 2014. Of 345 C9þ ALS/FTD probands (where expansion size was investigated by Southern blotting), 13 showed atypical expansions of 50e200 repeats (3.8%), including 2 cases where the 50e200 repeat allele occurred in conjunction with a large expansion (Buchman et al., 2013;Waite et al., 2014). A number of these blots were performed on DNA derived from lymphoblastoid cell lines which are known to poorly represent the range of repeats in vivo Hübers et al., 2014). If only Southern blots performed on blood DNA are considered, 6/195 expansions fall within the 50e200 repeat range (3.1%) (Supplementary Table 3).

Discussion
We present clinical features and C9orf72 expansion frequency and sizing data for a cohort of 452 UK ALS patients. In accordance with previous reports, the age at onset in the C9þ group is younger than C9À cases (Majounie et al., 2012). In our cohort, bulbar onset prevalence was not increased in C9þ cases. The frequency of C9orf72 repeat expansions has been found to vary between different geographical regions (Pliner et al., 2014;Woollacott and Mead, 2014); in our cohort, the frequency (w7%) is similar to that reported in previous UK studies (Cooper-Knock et al., 2012).
Although the presence of C9orf72 repeat expansion has been tested in numerous studies, only a minority have used Southern blotting to size the hexanucleotide expansion Buchman et al., 2013 Moss et al., 2014;Hübers et al., 2014, p. 72;Ishiura et al., 2012;Mann et al., 2013;Meisler et al., 2013;Murray et al., 2013;Takada et al., 2012;Van Blitterswijk et al., 2013a, 2013bWaite et al., 2014). Overall, these have shown that in DNA extracted from the most commonly analyzed tissues, such as blood, cerebellum, and frontal cortex, most of the expansions range between several hundred to several thousand repeats. The largest smears were typically observed in blood, and a larger, tighter band commonly observed in frontal cortex, compared with a smaller, more diffuse smear in the cerebellum Van Blitterswijk et al., 2013b). In rare cases, expansions of approximately 50e200 repeats have been reported in blood DNA (Buchman et al., 2013;Dobson-Stone et al., 2013;Dols-Icardo et al., 2014;Van Blitterswijk et al., 2013b). The pathogenicity of such size repeats and the corresponding repeat size in CNS tissues in some of these cases remains to be determined.
Our analysis of the C9orf72 repeat expansion size in blood confirms the presence of several hundred to several thousand repeats in most of the cases, but we also describe a single case carrying approximately 90 repeats in blood and other tissues. The postmortem analysis on this case shows the typical C9orf72 pathology pattern and interestingly, Southern blotting of frontal cortex, cerebellum, and spinal cord regions showed an expansion much larger than in blood and in the typical range of C9orf72 cases.
Our Southern blot analysis performed on cells and tissues from diverse embryological origins strongly suggests that an expansion from 90 repeats to 950e3000þ repeats occurred in the CNS. Indeed, the large expanded repeat is present only in brain and spinal cord, whereas the 90 repeat expansion is present both in ectodermally derived cauda equina and in more "distantlyrelated" tissues such as fibroblasts, blood and dura mater that originate from the mesoderm and mesoderm and/or ectoderm. These findings are consistent with those recently published by Van Blitterswijk et al. (2013b) who presented a case (P20) with a (C) Southern blots following BamHI/EcoRI digestions performed on individuals with repeat sizes of, in order on the blot, 2þ20, 2þ22, 2þ24, 5þ26, 10þ25, 2þ23, 2þ23, and 5þ27 repeats (sized by PCR), and 2 independent blood DNA extractions from case 61776 (mosaic band marked with *) whose nonexpanded allele has 5 repeats. No mosaicism was shown in the cases with 20e27 repeats. The position of the bands did not reflect their size relative to each other in all cases: the electrophoresis of the bands where more DNA was loaded was retarded in the agarose gel. Abbreviations: Neg, negative control; PCR, polymerase chain reaction; Pos, positive control. small expansion in several non-CNS tissues and large expansions in CNS tissue.
In both cases, the "small" expanded alleles are very similar in size among tissues, making the possibility of a retraction of the very large expansion to one of approximately 90 repeats unlikely. Indeed our findings demonstrate this would involve multiple retraction events (e.g., in the mesoderm and ectodermal peripheral nervous system progenitors) and, further, all retraction events would need to lead to the same final size. Dols-Icardo et al. (2014) have described a family where 3 brothers carry repeats ranging from 116 to 148 in blood, with offspring of 2 of the brothers inheriting repeats of 120 and 1401 repeats. This finding brings further support to the "multiple origin" hypothesis of C9orf72 large expansions.
The presence of the typical Finnish haplotype in the case here described supports the possibility that a permissive allele, predisposes to the hexanucleotide repeat expansion.
Another possible explanation, compatible with the single founder hypothesis, is that the original expansion occurred to a size of approximately 90 repeats, which expanded further in multiple cases. Given the instability of such repeats within the individual described here, and between generations (Dols-Icardo et al., 2014), this appears unlikely.
Whether retractions may also occur in C9orf72 cases remains a possibility. Van Blitterswijk et al. (2013b) have reported 2 cases in which multiple tissues have large expansions with the exception of a smaller band in either skeletal muscle (P14) or testes (P25). In these cases, single retraction events in those tissues may be possible.
The pathogenic threshold of the C9orf72 expansion remains to be determined. The findings reported here underline the necessity to study both peripheral and CNS tissues to identify this. Expansions ranging from 20 to 22 repeats have been associated with FTD (Gómez-Tortosa et al., 2013), but further segregation data and screening of large numbers of patients and controls are needed to clarify their significance. Whether these repeats expand somatically also remains to be addressed, but it is important to note that alleles of 20e27 repeats, in contrast to case 61776, did not show somatic instability in our study. The presence of somatic instability in blood does not offer direct evidence of pathogenicity, but is an important feature to consider, because in many neurologic disorders caused by nucleotide repeat expansions, pathogenic expansions are characterized by somatic instability, often with greater instability in tissues directly related to disease (Brouwer et al., 2009;Kennedy et al., 2003).
In diagnostic testing for Huntington disease, spinocerebellar ataxias types 1, 2, 3, 7, Friedreich ataxia, dentatorubralpallidoluysian atrophy, and X-linked spinal bulbar muscular atrophy, our laboratory has screened over 40,000 alleles and observed somatic instability in DNA extracted from blood in all pathogenic repeats but none from alleles with repeats in the nonpathogenic range. Taken together, these data suggest the threshold for instability may lie between >30 and 90 repeats.
Our review of all published C9þ cases underlines the fact that atypical expansions detected in blood represent approximately 3.1% of cases. Genetic counseling for these patients is currently challenging, because CNS tissues have not been examined in all published cases and the pathogenic repeat threshold and the likelihood of expansion in the brain specifically is unknown. Future studies addressing further aspects of the hexanucleotide expansions, such as the methylation state of the repeat itself and surrounding regions, may also contribute to the discrimination between pathogenic and nonpathogenic expansions (Xi et al., 2013).
We provide additional evidence that the C9orf72 expansion has occurred on multiple occasions on a permissive haplotype. That somatic instability may play a fundamental role in certain cases highlights the importance the understanding of this mechanism may have and may point to possible therapeutic strategies.