Novel genes associated with amyotrophic lateral sclerosis: diagnostic and clinical implications

MATR3

In 2014, four mutations (p.S85C, p.F115C, p.P154S, and p.T622A) in MATR3 were identified by exome sequencing in four families of European descent with either ALS alone or with a combination of ALS and dementia.⁸ Since 2014, 11 additional variants have been described, predominantly occurring in patients with sporadic ALS.^9–12 In the ExAC database, three variants (p.E664A, p.N787S, and c.48+1G>T) had a reported allele frequency of 0·03–0·05%, p.F115C was reported once, but none of the other variants were listed. Overall, the contribution of MATR3 to the development of ALS or ALS and frontotemporal dementia is relatively rare, with no significant correlation observed between phenotype and genotype.

In patients with ALS with MATR3 mutations, upper and lower motor neurons are affected and survival duration ranges from 2–12 years.¹⁰ The concomitant clinical presentation of ALS and myopathic features in individuals with the p.S85C mutation is important because these patients are initially diagnosed with vocal cord and pharyngeal dysfunction with asymmetric distal myopathy, but the presentation of pyramidal tract signs and progressive respiratory failure at end-stage disease usually warrants re-diagnosis.⁸ By contrast to TDP43 and FUS, whereby mutations cause relocalisation of the mutant protein from the nucleus to cytoplasm, studies^8,13 have shown that the subcellular localisation of mutant MATR3 is generally unaffected. Furthermore, MATR3-positive inclusions were occasionally observed in histopathological sections from patients with MATR3 mutations, and in one individual with C9orf72 expansion.⁸

MATR3 is a 125 kDa nuclear protein with RNA and DNA binding domains that appears to primarily regulate gene expression.¹⁴ Transgenic mice overexpressing human MATR3 protein develop hindlimb paralysis and muscle atrophy, indicating that neuromuscular function is sensitive to MATR3 levels.¹⁵ The protein forms a complex with two other ALS-associated RNA-binding proteins, TDP43⁸ and FUS,^16,17 in a RNA-dependent manner and the p.S85C mutation enhances this interaction.⁸ Thus, overlap might occur in the upstream regulatory proteins or downstream effector targets among ALS-RNA binding proteins. Elucidation of this potentially shared set of proteins might identify molecules suitable for therapeutic intervention.

CHCHD10

CHCHD10 was first linked to ALS in a study¹⁸ of a large French family who had a complex phenotype of ALS, ataxia, mitochondrial myopathy, parkinsonism, and sensorineural hearing loss. Exome sequencing identified a p.S59L mutation within CHCHD10.¹⁸ Subsequently, 20 additional missense variants, clustered in exon 2—which encodes an internal hydrophobic helical segment important for mitochondrial membrane binding¹⁹—have been reported in a broad range of neurodegenerative disorders, including ALS and frontotemporal dementia,^18,20–28 frontotemporal lobar degeneration,²⁹ parkinsonism,^26,27 Alzheimer’s disease,³⁰ autosomal dominant mitochondrial myopathy,³¹ adult-onset spinal muscular atrophy,³² and Charcot-Marie-Tooth type 2.³³ The pathogenicity of p.S59L,¹⁸ p.R15L,^20–22 and p.G66V²⁰ has been validated in family studies, whereby the mutations were shown to segregate with ALS. Additionally, the mutations were absent in the ExAC database. However, mutations in CHCHD10 appear to be a relatively rare cause of ALS, but might be more frequent among patients diagnosed with frontotemporal dementia.^27,34

CHCHD10 is a 14 kDa nuclear-encoded, mitochondrial protein localised to the mitochondrial intermembrane space. The protein is important for the maintenance of mitochondrial dynamics and cellular bioenergetics.³⁵ Patient fibroblasts expressing mutant CHCHD10 protein (p.S59L) have a fragmented mitochondrial network and disrupted mitochondrial cristae.¹⁸ These effects are similar to abnormalities in mitochondrial dynamics induced by mutations in TDP43.³⁶ CHCHD10 also interacts with TDP43, which promotes retention of TDP43 in the nucleus,³⁷ but this localisation is disrupted in the presence of CHCHD10 mutations, causing an accumulation of TDP43 in the cytoplasm and synaptic damage.³⁷ Further study is necessary to investigate the mechanistic association between these proteins and their involvement in mitochondrial dysfunction and TDP43 proteinopathy. This insight could identify therapeutic targets susceptible to manipulation by small molecules, to rescue the observed cellular defects involved in ALS.

TUBA4A

TUBA4A was implicated as a novel gene for familial ALS on the basis of exome sequencing data obtained from a large cohort of European and American patients with ALS and controls.³⁸ This finding was replicated in an independent Belgian cohort,²⁶ but not in Asian patients with ALS.³⁹ All variants were absent or had very low frequency in the ExAC database and had adequate segregation data, with the exception of p.K430N. The overall frequency of TUBA4A mutations suggests it is a rare cause of ALS. Little information is available about the clinical presentation, prognosis, or neuropathological evaluation of patients with TUBA4A variants, and although patients often present with typical features of ALS, some also present with features of frontotemporal dementia.^26,38

The main cytoskeletal scaffold in cells is comprised of microtubules, composed of polymerised α-tubulin and β-tubulin subunits. In primary motor neurons, expression of missense mutation TUBA4A interferes with tubulin dimerisation, resulting in a weakened microtubule network.³⁸ Mutations have been found to cluster in the protein domain responsible for the interaction with other tubulin subunits and the axonal transport proteins dynein and kinesin.⁴⁰ This finding highlights the crucial role of cytoskeletal and axonal transport defects in the pathogenesis of ALS. Therapeutic approaches enhancing cytoskeletal integrity might be crucial for halting progression or reversing the disease course.

TBK1

A whole exome sequencing study⁴¹ revealed that TBK1 was implicated in ALS. Enrichment of nonsynonymous variants in patients with ALS compared with healthy controls was found across the entire coding region.⁴¹ This finding was validated by another whole exome sequencing study,⁴² which reported segregation of the pathogenic variants within affected families. Mutations in TBK1 are found in about 1% of patients with familial ALS and in approximately 1% of patients with sporadic ALS.^42–53 The clinical phenotypes associated with TBK1 mutations are heterogeneous, with variable age of onset, differing progression, and irregular length of survival time.^39,43–46 Extrapyramidal symptoms, ataxia, and psychiatric symptoms have also been reported in some patients with TBK1 mutations.⁴⁶ Neuropathological examination of CNS tissue from patients with a TBK1 mutation showed SQSTM1/p62 and TDP43-positve inclusions,^42,45,46 which are indicative of abnormal TDP43 protein aggregation and defective protein clearance pathways.⁴⁷ Since these inclusions are also observed in other patients with ALS without TBK1 mutations,⁴⁸ this suggests that a common disease mechanism might exist, and a broad treatment approach to restore defective proteostasis might also benefit patients with TBK1 mutations.⁴⁹

TBK1 is a homodimeric multidomain protein with a kinase domain, a ubiquitin-like domain, and two coiled-coil domains.⁵⁰ The protein acts as an interaction platform for multiple proteins and regulates the activities of downstream protein targets involved in key cellular processes that have been implicated in ALS, including neuroinflammation, ubiquitin-proteasome systems, and autophagy pathways involving other genes also associated with ALS—ie, OPTN, SQSTM1/p62, VCP, and UBQLN2.⁵⁰ Most pathogenic variants identified in TBK1 are concentrated within the kinase and the coiled-coiled domains,⁴² suggesting that these mutations might operate by altering these downstream regulatory pathways. We identified some variants (p.K291E, p. I305T, p.L306I, p.H322Y, p.T3221I, p.R444Q, and p.A535T) in the ExAC database that had a frequency higher than our estimated threshold of 0·0033%, suggesting that they are unlikely to be pathogenic. Pathogenicity of the other variants will require further investigation in families and in cells or animal models.

NEK1 and C21orf2

Heterozygous loss-of-function mutations in NEK1 have been implicated in sporadic ALS.⁴¹ NEK1 interacts with two proteins known to be associated with ALS, ALS2 and VAPB,⁴⁵ which are involved in endosomal and endoplasmic reticulum lipid trafficking; this interaction provides some functional evidence of NEK1 involvement in ALS pathogenesis. Two independent case-control studies^51,52 provided corroborative evidence that NEK1 is associated with ALS, and indicated that it might account for up to 2% of all ALS cases. Clinical descriptions of patients with NEK1 mutations are scarce, but patients appear to present with typical ALS without dementia.^51,52

Concomitant with the identification of NEK1, a large case-control study⁵³ using the genome-wide association approach found that C21orf2 was associated with increased ALS risk. NEK1 and C21orf2 interact with each other and are involved in microtubule assembly, DNA damage response and repair, and mitochondrial function.^54,55 Additional genetic replication studies in independent cohorts and functional and clinical studies for both NEK1 and C21orf2 are required to fully understand the contribution of these variants in the pathogenesis of ALS.

CCNF

CCNF was identified as a causative gene for ALS on the basis of exome sequence analysis⁵⁶ of a large family of European descent who had ALS, frontotemporal dementia, or both diseases, with an autosomal dominant pattern of inheritance. The authors reported additional, potentially pathogenic variants in CCNF in familial cases (all absent or less than the 0·0033% threshold in the ExAC database), with an overall mutation frequency that ranged between 0·6 and 3·3% in white populations.⁵⁷ Clinically, these patients presented with either typical ALS, ALS with frontotemporal dementia, or frontotemporal dementia alone.⁵⁶

CCNF is the substrate-recognising component of the Skp1-cullin-F-box E3 ubiquitin-ligase complex, which is responsible for tagging proteins with ubiquitin and marking them for degradation via the ubiquitin-proteasome system.⁵⁷ Neuronal cells overexpressing mutant CCNF show an increase in ubiquitin-tagged proteins, which include TDP43. This increase suggests that these variants affect the proteosomal degradation pathway by either aberrantly tagging all proteins with ubiquitin or failing to transfer ubiquitin-tagged proteins to the proteasome complex for removal.⁵⁶ This finding indicates that mutations in CCNF might lead to abnormal proteostasis, which might be exacerbated by TDP43 proteinopathy. Therefore, therapies that enhance protein clearance or reduce ubiquitination might be viable approaches to treatment.