Genetics of ALS – genes and modifier

Abstract

Purpose of review

Amyotrophic lateral sclerosis (ALS) is a complex genetic disorder, and the pace of discoveries is very rapid. This review aims at briefly summarizing our current knowledge, and at discussing the progress of the last two years.

Recent findings

Common variation in numerous genes and variants in some nuclear-encoded mitochondrial genes were linked to an increased or modified risk of ALS, respectively. Mitochondrial function, i.e. specific mitochondrial haplotypes and loss-of-function variants in mitochondria-related genes, was identified as potent modifier of ALS survival, but not risk. Pioneering analyses of copy number variations in ALS-related genes revealed an increased load in ALS, but causality is unclear. A rare hyperactive variant of ER stress associated transcription factor CREB3 was linked to both substantially decreased ALS risk and slower disease progression. Furthermore, variants in IGFBP7 were linked to rare "ALS reversals", but existence of such phenotypes is controversial.

Summary

Common variation increasing ALS risk contributes to our understanding of sporadic ALS, and novel structural variants have the potential to at least partly explain the missing heritability in ALS. Identification of mitochondrial function and ER stress signaling as potent disease modifiers provide valuable starting points for therapeutic approaches beyond targeting single causative genes.

INTRODUCTION

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease primarily affecting motoneurons in motor cortex, brainstem and spinal cord. In general, the disease starts focally at around 60 years of age, and rapidly spreads from the initial site of onset to neighboring areas, affecting most voluntary muscles at end-stage. The progressive loss of motoneurons is associated with increasing weakness, muscular atrophy, and paralysis, leading to respiratory failure and death typically 2–5 years after symptom onset. Less than 50% of ALS patients additionally develop cognitive and/or behavioral impairment during the course of the disease, and 5–10% progress to concomitant frontotemporal dementia (FTD). Therefore, and because of partially shared genetic causes, ALS and FTD are the two extremes of the ALS-FTD disease spectrum. The neuropathological hallmark of ALS are nuclear clearance and neuronal cytoplasmic inclusions of the RNA-binding protein TDP-43 in affected tissues [1,2]. The only exception to the almost universal TDP-43 pathology in ALS are rare cases caused by mutations in SOD1 or FUS[3].

Historically, ALS is classified in sporadic (sALS) and familial (fALS) cases that cannot be distinguished clinically. Roughly 5–10% of ALS patients report a family history of the disease, and, depending on the population studied, 50–85% are explained by known mutations in causative ALS genes. However, many monogenic causes of ALS show reduced penetrance, and pathogenic mutations in ALS genes are found in 10–20% of sporadic cases [4], including de novo mutations, especially in FUS[5], SOD1[6] and SPTLC1[7]. The contribution of genetics to ALS is much higher than expected, and studies based on data of twins discordant for ALS, trios or parent-child dyads repeatedly estimated the heritability of ALS around 40–60% [8,9]. First-degree relatives of ALS patients without a known genetic cause of the disease have a roughly two-fold increased lifetime risk of developing ALS compared to the general population [9]. Therefore, ALS is a complex disorder with a strong genetic component, and mostly undefined contribution of environmental factors [1]. The classification in sALS and fALS does not adequately reflect the complex genetics of this disease. A recent classification of ALS associated genes according to effect size provides more detailed information about the contribution of genetic variants in individual patients, and ranges from rare monogenic causes of the disease (large effect size) to very common risk factors (small effect size) that do not cause disease alone. Estimating contribution of individual genes, or even variants in the future, to the development of ALS is critical for calculating polygenic risk scores, and improve genetic counselling [4].

Discovery of genes and variants associated with ALS is closely linked to advances in sequencing technologies and data analyses. Especially whole-exome and whole-genome sequencing in the last 15 years greatly improved our understanding of ALS genetics, and multiplied numbers of causative genes and risk factors [10]. In order to keep up with the rapid developments in ALS genetics, this review aims at providing an overview of our current knowledge of ALS genetics, and summarizes the progress made and trends of the last two years. 

Box 1.

no caption available

STATE-OF-THE-ART IN AMYOTROPHIC LATERAL SCLEROSIS GENETICS

The genes involved in ALS have manifold functions, but can usually be assigned to few cell biological topics, such as RNA metabolism, proteostasis, the cytoskeleton including axonal transport, mitochondrial function and/or DNA repair mechanisms. The mode of inheritance is mostly autosomal dominant, and only very few genes display autosomal recessive or X-linked patterns [4]. The types and location of ALS associated variants may be specific for the respective gene. For example, while pathogenic missense mutations are found throughout SOD1 protein [11], causative missense variants cluster in the C-terminal disordered region of TDP-43 [12]. Fig. 1 summarizes our current knowledge of ALS associated genes and their contribution to disease.

FIGURE 1.

Rare and common variants in various genes contribute to the development of ALS. (a) Genes in which rare variants are the monogenic cause of, or increase the risk for, ALS. Estimated odds ratios are indicated. (b) Established risk genes for ALS in which common variants contribute to the development of ALS. Respective frequencies of variants in sALS patients are shown (data from [4] and references therein; mean odds ratios were used in (a) in case more than one estimate was available).

By far the most frequent genes responsible for ALS worldwide are C9orf72 and SOD1. While a hexanucleotide repeat expansion in C9orf72 is the most common cause in populations of European ancestry (≈30–60% fALS, 5–10% sALS), it is very rare in Asian populations (≈2% fALS, 0.3% sALS). Variants in SOD1 are a frequent cause of ALS in both populations, being the most common cause of ALS in Asians (≈30% fALS, 1.5% sALS), and the second most common in Europeans (≈12–20% fALS, 1–2% sALS). Interestingly, variants in VABP are extremely rare in European and Asian ALS patients, but seem to be the most common cause of fALS (≈30%) in South America. Variants in TARDBP, FUS and TBK1 are much rarer, but each gene may explain up to 5% of fALS, depending on the population studied. Mutations in other ALS genes are extremely rare, and pathogenic variants are usually found in 1%, or even less, of fALS patients [4,13]. Besides these disease-causing germline variants, a first report indicates that somatic mosaicism may be responsible for a substantial proportion of sporadic ALS and FTD (>20%), either by an overall higher load of such variants in patients compared to controls, and/or by known pathogenic variants found with low allele frequency in the respective tissues [14].

In contrast to the numerous genetic causes and risk factors associated with ALS, very few genetic variants have been described that modify the ALS phenotype, such as age-at-onset, progression or survival. Variants in ATXN2 and UNC13A are the only well described genetic ALS disease modifiers known to date [10]. ATXN2 encodes an RNA-binding protein harboring a CAG repeat in exon 1 that, if expanded, is causative for spinocerebellar ataxia type 2 [15]. Intermediate repeat lengths are a relatively common risk factor for ALS, and are found in roughly 1–2% of healthy individuals and ≈5% of ALS patients. Orthologues of ATXN2 have been identified as potent modifier of TDP-43 toxicity in yeast and flies [16], and reduction or depletion of Atxn2 in transgenic ALS model mice expressing human wildtype TDP-43 substantially rescued the phenotype by reducing TDP-43 pathology, improving motor functions and extending lifespan [17]. In human ALS, survival of patients with intermediate ATXN2 CAG repeat lengths is shorter than survival of patients with shorter repeats [18]. Hence, reducing expression of ATXN2 may be beneficial for both ALS patients with and without expanded CAG in ATXN2. However, a recent clinical trial targeting ATXN2 expression by antisense oligonucleotides was discontinued due to the lack of clinical benefits for the patients (ALSpire Study: NCT04494256).

UNC13A encodes a protein involved in synaptic vesicle maturation and priming, and consequently plays an important role in neurotransmission and synaptic plasticity. Genome-wide association studies (GWAS) repeatedly linked very common variation in UNC13A to a slightly increased risk of ALS and/or FTD, and shorter survival of patients carrying risk alleles have been repeatedly reported in independent cohorts. Mechanistically, risk alleles of UNC13A are more prone to miss-splicing caused by loss of nuclear TDP-43. Here, an intronic sequence (so-called cryptic exon) is included into mature UNC13A mRNAs leading to a frameshift, and consequently degradation by nonsense-mediated mRNA decay. Associated reduction of UNC13A protein in turn may contribute to neurodegeneration by generally reducing neuronal health and neurotransmission, and weakening of synapses and neuromuscular junctions [19]. Clinical trials aiming at preventing inclusion of the UNC13A cryptic exon, e.g. by splice-modifying antisense oligonucleotides, are currently in preparation.

Other genetic modifiers of human ALS gained much less attention than ATXN2 and UNC13A, and most of them require replication in larger/independent cohorts of patients. These include, but are not limited to, variants in CHGB[20,21], CTIF[22] and the RNA gene NEAT1[23] for age-at-onset, and in CAMTA1 for survival [24].

NOVEL RISK GENES FOR AMYOTROPHIC LATERAL SCLEROSIS

In the last two years, summary statistics of the latest large ALS GWAS screen [25] were further exploited. A transcriptome-wide association study (TWAS) [26^▪▪] focusing on relatively common variation [minor allele frequency (MAF) of variants >1%] could identify variants in 108 genes associated with ALS (Fig. 2a), including some established ALS genes that may cause the disease alone when harboring a very rare deleterious variant, such as C9orf72[27], TBK1[28,29], and potentially NEK1[30,31]. Effect sizes of variants in these 108 genes are most likely small, but the data will help understanding the genetic architecture of ALS, especially when no high-effect-size variants are involved. Interestingly, functional annotation of these 108 genes using the Enrichr database [32] did not reveal any significant enrichment of these genes in specific Gene Ontology categories or pathways, underscoring the high heterogeneity in ALS. The same study [26^▪▪], as well as another one [33^▪] using highly similar datasets, performed a proteome-wide association study (PWAS) to identify proteins whose expression levels were associated with ALS. Results from both studies were almost identical, and highly similar to a comparable study from 2023 [34]. The results of all three studies are summarized in Fig. 2b.

FIGURE 2.

Novel genes associated with ALS identified by transcriptome- (TWAS) and proteome-wide association studies (PWAS). (a) List of genes increasing the risk for ALS when carrying specific variants identified by a TWAS approach [26^▪▪]. Note that this study was restricted to common variation (minor allele frequency > 1%). (b) Genes in which variation increases the risk for ALS identified by three independent PWAS approaches [26^▪▪,33^▪,34]. Genes that have been previously associated with ALS are marked in green.

Besides TWAS and PWAS approaches, Mendelian Randomization was used to discover novel genetic links to ALS, especially in genes coding for cathepsins and mitochondrial proteins. Two independent studies could not detect a causal link between variants impacting expression level of cathepsins and ALS, but for other neurodegenerative diseases [35,36]. Results for mitochondria related genes are less consistent, likely because of different datasets and methods used, and further validation is required. One study could show that variants in nuclear-encoded genes C1QBP, PDK1 and MRMT3 are associated with an increased risk for ALS, while variants in COX5B and MUL1 slightly decrease risk [37]. Other studies linked variants leading to higher expression of nuclear-encoded ACLY[38], as well as to higher and lower expression of nuclear-encoded genes MRM1 and MYO19, respectively, to an increased risk for ALS [39]. Furthermore, an elegant study convincingly showed that ALS risk is not associated with mitochondrial haplotypes [40^▪▪].

A very important study addressed the impact of oligogenicity, i.e. presence of more than one rare (MAF < 0.01) disease associated variant in the same individual, on ALS risk and clinical outcome. Here, the authors found, at least based on our current knowledge of ALS genetics, that oligogenic events are of relevance in ≈6% of patients, and ALS risk is higher in individuals carrying more than one rare ALS associated variant. Interestingly, oligogenicity had little influence on age-at-onset or survival [41^▪▪], further supporting independent genetics of ALS risk and survival reported previously [25,42,43].

NOVEL STRUCTURAL VARIANTS

So far, relatively little is known about the contribution of structural variants to ALS risk, likely because rearrangements, inversions, insertions or deletions of larger DNA segments are hard to detect in short-read whole-exome and whole-genome sequencing data. However, such variants may at least partly explain the missing heritability in ALS. An interesting study [44^▪▪] determined copy numbers of 131 genes previously associated with ALS in a small cohort of patients and controls. Copy number variations (CNVs), either of whole genes, or of smaller regions within a gene, were present in most genes and most probands. However, CNV load was higher in patients than in controls based on both the number of such events and the cumulative lengths of CNVs [44^▪▪]. This pioneering report provides a solid basis for studying genome-wide CNVs in large cohorts of patients.

The contribution of CNVs in SMN1 and SMN2 to ALS risk is debated for more than 20 years, and inconsistent or even contrary results were reported from independent patient cohorts (summarized in [45]). Recently, results differing from previous studies were reported, i.e. no difference of CNVs in SMN1 between patients and controls, but an enrichment of individuals carrying a single copy of SMN2 among ALS patients [46]. Thus, the contribution of CNVs in SMN1/SMN2 to ALS risk is still not clear.

NOVEL CLINICAL MODIFIER

Genetic modifier of ALS clinical characteristics discovered in the last 24 months include some very interesting candidates. However, independent replication is required for most of them.

CNVs in SMN1 and/or SMN2 are not only controversial considering ALS risk (see above), but also in terms of their impact on survival. In a recent report [46], homozygous deletion of SMN2 was associated with shorter survival while no effect on survival was detected for copy numbers of SMN1. Both findings are supported by some studies, but not by others (summarized in [45]), and the role of SMN1/SMN2 copy numbers in ALS remains enigmatic.

Very promising candidate modifier were identified when analyzing mitochondria and mitochondria-related genes in ALS [40^▪▪]. Here, a specific mitochondrial haplotype (0_0_0_2_2_2) was associated with longer survival of patients in three independent cohorts. Interestingly, mitochondrial haplotypes were not associated with disease risk, indicating that mitochondrial dysfunction in ALS is rather a consequence than the cause of the disease. Mitochondrial haplotypes may determine copy numbers of the mitochondrial DNA [47], and higher copy numbers in ALS were associated with shorter survival. By analyzing rare loss-of-function variants in nuclear encoded mitochondrial genes, ALS patients carrying such variants in ACADM or DNA2 showed 50% shorter survival than other patients. Here too, variants in these genes did not increase the risk for ALS [40^▪▪]. Hence, mitochondrial function may modify ALS disease progression after onset, but is not or rather weakly involved in ALS risk.

Another ALS disease modifier linked to pathogenic events was recently identified by focusing on transcriptomics of vulnerable glutamatergic corticospinal neurons from ALS patients and model mice. Here, the authors found that vulnerable neurons undergo endoplasmatic reticulum (ER) stress, and identified the ER stress responsive transcription factor CREB3 and its regulatory network as a resilience marker. Indeed, a hyperactive rare variant of CREB3 (p.Arg119Gly; rs11538707) was associated with both a substantial decreased risk (≈40%) for developing ALS and less severe disease progression in patients [48^▪▪]. Hence, enhancing the CREB3-dependent pro-survival arm of ER stress response in a therapeutic approach may be beneficial for patients.

Last but not least, an interesting GWAS compared ALS patients to a few “ALS patients” that initially met ALS diagnostic criteria, but then seemed to show substantial and sustained clinical improvement. Here, variants leading to decreased expression of IGFBP7 were associated with the reversal phenotype [49^▪]. Interestingly, IGFBP7 expression is increased in Alzheimer's disease, and has been linked to memory impairment [50]. However, the role of IGFBP7 in reversing ALS phenotypes is controversial. It is not yet clear if extremely rare “ALS reversals” suffer from typical ALS, or rather from another, yet undescribed disease mimicking ALS diagnostic criteria. Nevertheless, IGFBP7 may be of relevance in ALS pathogenesis, but further studies are required to judge the impact of this gene on ALS phenotypes.

CONCLUSION

In the last 2 years, >100 genes have been associated with an increased risk for ALS. Considering allele frequency of most of the variants in these genes, effect sizes are likely small, but such data is valuable for better understanding the etiology of sALS. While no novel monogenic causes have been identified, first studies addressing the contribution of CNVs and somatic mosaicism to the development of ALS are promising, and both have the potential to at least partly explain the missing heritability in ALS. Advanced sequencing methods facilitating longer reads and ultra-deep sequencing for the detection of somatic variants with very low allele frequencies are available, and will further improve our understanding of ALS genetics in the near future.

Progress was also reported regarding genetic modifier of ALS phenotypes, and mitochondrial function as well as ER stress response were linked to survival and/or disease risk, respectively. However, whether improvement of these cellular processes indeed results in benefits for the patients has yet to be shown. Development of such treatments is of high priority to mitigate the disease in the majority of patients not carrying a causative mutation in a common ALS gene that may be targeted directly by other approaches, e.g. by antisense oligonucleotides.

Acknowledgements

We would like to thank Albert C. Ludolph (Department of Neurology, Ulm University, Ulm, Germany) for the invitation writing this review, and for critically reading the manuscript.

Financial support and sponsorship

A.F. is supported by the German Research Foundation (DFG; grant #521487152) and the German Society for Muscular Diseases (“Deutsche Gesellschaft für Muskelkranke e.V.” [DGM]).

Conflicts of interest

The authors declare no conflicts of interest associated with this manuscript.

REFERENCES AND RECOMMENDED READING

Papers of particular interest, published within the annual period of review, have been highlighted as:

• ▪ of special interest

• ▪▪ of outstanding interest