DNA methylation in amyotrophic lateral sclerosis: where do we stand and what is next?

ABSTRACT

Genes involved in immune response, inflammation and metabolism are among those most likely affected by changes in DNA methylation (DNAm) and expression levels in amyotrophic lateral sclerosis (ALS) tissues. Unfortunately, it is still largely unclear whether any of these changes precede the onset of disease symptoms or whether most of them are the result of the muscular and metabolic changes that follow symptoms onset. In this article the author discusses the strengths and limitations of the available studies of DNAm in ALS and provides some suggestions on what, in his opinion, could be done in the near future for a better understanding of the DNAm changes occurring in ALS, their link with environmental exposures and their potential clinical utility.

Plain language summary

Article highlights.

DNA methylation (DNAm) studies in amyotrophic lateral sclerosis

• Global changes in DNAm levels occur in both whole blood and spinal cord DNA samples from patients with amyotrophic lateral sclerosis (ALS).

• Genes involved in immune response, inflammation and metabolism are among those most likely affected by changes in DNAm and expression levels in ALS tissues.

Can DNAm in peripheral tissues provide insights into individuals at risk of developing ALS in the coming years?

• Whether or not certain changes in DNAm precede symptoms onset and may be informative about individuals at risk of developing ALS symptoms in the coming years is a still unsolved question.

Environmentally induced DNAm changes in ALS, what do we really know?

• Some investigators suggest links between environmental exposure end the observed DNAm changes in ALS tissues but can still say little about causality.

Is there any sex difference in DNAm levels?

• Limited data are still available on sex differences in DNAm levels in ALS tissues.

The potential clinical utility of the ALS methylome

• Recent study points to a potential link between DNAm signatures and ALS age of onset, progression rate and survival. Validation of these potential biomarkers is required.

DNAm in the different forms of ALS presentation

• Comparisons of DNAm data across different forms of ALS presentation are lacking and will be warranted in future investigations.

Future perspective

• Efforts should be made to collect large cohorts of presymptomatic carriers of ALS-linked mutations, and longitudinal investigations should be performed in those subjects searching for early ALS methylation biomarkers.

• Similar longitudinal investigations and follow-up studies are required in large cohorts of human samples to identify early ALS methylation signatures and clarify if DNAm biomarkers of ALS progression rate and survival can be detected at symptoms onset or before.

• Those studies can help identify early ALS related methylomic changes, their links with environmental exposures, lifestyles, dietary or medical interventions and their potential clinical utility.

1. Introduction

Amyotrophic lateral sclerosis (ALS) is a devastating neuromuscular disease resulting from the degeneration of upper and lower motor neurons. Clinical presentations and disease progression vary among ALS patients, but most of them dye within 2–3 years from symptoms onset. The majority of ALS is sporadic (sALS) and familial forms account for almost 10% of the cases (fALS). Mutations in SOD1, FUS and TARDBP genes and a hexanucleotide repeat expansion in C9orf72, have been found in almost 50% of fALS and 10% of sALS cases. Rare variants in additional genes, oligogenic and polygenic inheritance patterns and complex gene–environment interactions likely explain the rest of ALS. In addition, numerous nongenetic risk factors have been proposed to increase ALS risk, including exposure to metals and pesticides, occupational settings such as manufacturing, construction and agriculture, professional soccer, air pollution, electromagnetic fields, head traumas, dietary habits, microbiome composition, smoking and low levels of education [1,2].

In 2020, I comprehensively reviewed the literature related to the epigenetic mechanisms involved in neuromuscular disorders, and it was clear that epigenetic mechanisms play a role in ALS progression [3]. At that time, numerous investigators had observed a global increase in DNA methylation (DNAm), assessed as quantification of 5-methylcytosine (5-mC) levels, in both blood and spinal cord DNA samples from familial and sporadic ALS patients [4–6]. Those studies were paralleled by epigenome-wide methylation studies (EWAS) in blood and spinal cord tissues, which revealed major genes and pathways epigenetically dysregulated in individuals suffering from this condition [7–9]. In this article I will briefly summarize the studies on DNAm in ALS patients, focusing on the most recent ones conducted in the last 5 years. Next, I will discuss the strengths and limitations of the available literature and what, in my view, should be done in the coming years to further address this issue.

2. DNAm studies in ALS

DNAm represents one of the most stable and investigated epigenetic marks and consists in the transfer of a methyl group from S-adenosylmethionine (SAM) to cytosine by either de novo (DNMT3A and DNMT3B) or maintenance (DNMT1) DNA methyltrasferases (DNMTs), thus forming 5-mC. DNAm occurs in several genome regions, including repetitive DNA elements, coding sequences and regulatory regions. DNAm has several physiological roles, including cell differentiation, maintenance of cell identity and genome stability and regulation of gene expression levels. Usually, methylation in the gene promoter region is linked to repression of transcription. In neurons, DNAm is required for neurogenesis, neuronal maturation, synaptic plasticity, learning and memory. Specific changes in DNAm occur with aging and are linked to increased risk of various human diseases [10,11].

Studies performed over the first two decades of the 21st century revealed an increased level of 5-mC in both spinal cord and blood DNA samples from ALS patients [4–6]. Those studies were paralleled by observations of increased DNMT levels in motor neurons of ALS animal models, as well as in postmortem brain and spinal cord tissues of ALS patients [12]. Also changes in mitochondrial DNA (mtDNA) methylation were detected in spinal cord and skeletal muscles of transgenic SOD1 mice and in human blood samples from SOD1-mutant ALS patients [13,14]. DNAm age acceleration was observed in blood DNA of ALS patients [9,15,16], and both EWAS and expression studies in ALS blood and spinal cord samples revealed global changes in DNAm and a dysregulated expression of several genes primarily involved in immune response, inflammation and metabolism [7–9]. Concerning the promoters of major ALS genes, namely SOD1, C9orf72, FUS and TARDBP, they resulted almost completely demethylated in samples from ALS patients, except for C9orf72 hexanucleotide repeat expansion carriers that showed hypermethylation of the C9orf72 promoter region in 10–30% of the cases [5,17–20].

The most recent studies on DNAm in ALS performed over the last 5 years are summarized in Tables 1 & 2 which show, respectively, investigations performed on blood samples from ALS patients (Table 1), and studies performed in brain, spinal cord, skeletal muscles and motor neurons from ALS patients or ALS animal models (Table 2). Collectively, these studies confirm previous findings of impaired DNAm in ALS tissues [21–39]. Indeed, increased levels of 5-mC and/or increased DNMT activity was observed in ALS motor neurons as well as in spinal cord and skeletal muscles of transgenic SOD1 mutant mice [36,38], and numerous studies revealed the potential of DNAm signatures from either whole blood DNA or from cell-free DNA extracted from the blood plasma to discriminate between ALS and disease-free status [21,25,34]. The genes and pathways epigenetically dysregulated in ALS tissue are increasingly defined. Indeed, a large-scale meta-analysis of methylomic data in blood samples from 6763 ALS patients and 2943 healthy controls revealed altered methylation levels of genes related to immune function, metabolism and cholesterol biosynthesis in the first group [24]. Another recent large-scale study took advantage of machine learning approaches to elaborate genomic, epigenomic and transcriptomic data obtained from human fALS induced pluripotent stem cell-derived motor neurons (hiPSC MNs), and from large-scale gene expression datasets of blood and postmortem spinal cord tissues from ALS patients. Authors observed altered methylation and expression of genes related to synaptic function in fALS hiPSC MNs, as well as altered expression of genes related to toll-like receptor cascade, immune function, autophagy and protein metabolism in ALS blood, and of genes related to synaptic function, vesicle trafficking, toll-like receptor cascade, immune function and TP53-mediated regulation of metabolic functions in ALS spinal cord tissues [39]. Moreover, the acquired somatic mutations in ALS tissues were linked to an age-related decline of DNA repair mechanisms [39]. Some authors observed DNAm age acceleration in ALS patients [22,25], while others observed an increased epigenetic drift and the accumulation of rare epivariations in blood samples from individuals with sALS [31]. Changes in mtDNA methylation were confirmed in ALS patients and inversely correlated with disease duration from onset [23,32].

Table 1.

Summary of recent DNA methylation studies in blood samples from patients with amyotrophic lateral sclerosis.

Author	Sample	Number	Type of study	Main results	Ref.
Nabais et al., 2020.	Whole blood	Study cohort: 782 ALS cases and 613 controls; replication cohort: 1159 ALS cases and 637 controls	Genome-wide methylation	A DNAm score significantly classified case-control status	[21]
Zhang et al., 2020.	Whole blood	Whole blood from 249 ALS patients, including 200 sporadic and 49 familial cases	Genome-wide methylation	DNAm age acceleration was associated with ALS age at onset and survival	[22]
Stoccoro et al., 2020.	Whole blood	Fourteen SOD1-ALS; 13 C9orf72-ALS; 36 sporadic ALS (sALS); 51 controls	Quantification of mitochondrial DNA (mtDNA) D-loop methylation levels and of mtDNA copy number	Reduced mtDNA D-loop methylation levels in ALS patients respect to controls, in particular in SOD1-ALS and sALS. Inverse correlation between D-loop methylation levels and the mtDNA copy number	[23]
Hop et al., 2022.	Whole blood	A total of 6763 ALS patients; 2943 controls	Genome-wide methylation	The study revealed 42 differentially methylated genes between ALS and control samples, enriched for pathways related to metabolism, cholesterol biosynthesis, and immunity	[24]
Ruf et al., 2022.	Whole blood	Sixty-five sALS; 24 C9orf72-ALS; 10 FUS-ALS; 11 C9orf72-presymptomatic mutation carriers; 6 FUS-presymptomatic mutation carriers; 71 controls	Genome-wide methylation	DNAm age acceleration was observed in symptomatic ALS cases in comparison to healthy controls. Methylation changes in ALS blood are associated with disease status and are not present before the onset of symptoms	[25]
Freydenzon et al., 2022.	Whole blood	A total of 438 ALS cases and 417 controls	Genome-wide methylation	DNAm near the PEX11B gene was associated with self-reported cadmium exposure	[26]
Cai et al., 2022.	Whole blood	Thirty-two sALS patients; 32 healthy controls	Genome-wide methylation	Authors identified 34 significant differentially methylated positions and 12 differentially methylated regions in ALS patients respect to controls. The methylation levels of ELOVL2 and ARID1B were positively associated with disease age of onset and duration, respectively	[27]
Cai et al., 2022.	Whole blood	Forty-five ALS patients; 32 healthy controls	Promoter methylation and expression levels of LGALS1	Increased mRNA levels and reduced promoter methylation of LGALS1 in ALS patients respect to controls	[28]
Tazelaar et al., 2023.	Whole blood	Twenty-two ALS-discordant monozygotic twin pairs	Genome-wide methylation	The methylation analysis identified three candidate genes, namely SND1, LBX1, and GRN, as differentially methylated between affected and unaffected twins, but none reached genome-wide significance	[29]
Yazar et al., 2023.	Whole blood	Seven ALS-discordant monozygotic twin pairs	Genome-wide methylation	Authors identified one CpG site in GRIK1 showing hypermethylation in the affected co-twins	[30]
Brusati et al., 2023.	Whole blood	Sixty one sALS patients; 61 healthy controls	Genome-wide methylation	Increased epigenetic drift was detected in sALS patients compared with controls, and rare epivariations were exclusively enriched in sALS cases	[31]
Stoccoro et al., 2024.	Whole blood	Twelve SOD1-ALS; 13 C9orf72-ALS	Quantification of mitochondrial DNA (mtDNA) D-loop methylation levels and of mtDNA copy number	D-loop methylation levels inversely correlated with disease duration	[32]
Yang et al., 2024.	Whole blood	Forty-one sALS patients; 27 healthy controls	Genome-wide methylation	Authors identified numerous differentially methylated regions and/or positions associated with disease progression rate and survival	[33]
Caggiano et al., 2024.	Cell-free DNA extracted from the blood plasma	Eighty-eight ALS patients; 14 possible ALS or primary lateral sclerosis (PLS) patients; 76 healthy controls; asymptomatic gene carriers; 15 individuals with other neurodegenerative diseases	Targeted methylation sequencing	Cell-free DNAm accurately predicted ALS status and disease severity	[34]

Table 2.

Summary of recent DNA methylation studies in brain, spinal cord, skeletal muscles and motor neurons from amyotrophic lateral sclerosis patients and animal models.

Author	Sample	Number	Type of study	Main results	Ref.
Zhang et al., 2020.	Cortical or spinal cord samples	Cortical or spinal cord samples from 18 patients, including 1 familial and 17 sporadic cases	Genome-wide methylation	Frontal cortex-based and spinal cord-based DNAm age acceleration was associated with ALS age at onset and survival	[22]
Savage et al., 2020.	Cerebellum and motor cortex	Two familial ALS (fALS); 2 sALS; 2 controls	Methylation analysis of the RC-L1 loci	Reduced methylation of RC-L1 loci in the motor cortex of ALS patients compared with control brains	[35]
Appleby-Mallinder et al., 2021.	Spinal cord, motor cortex and prefrontal cortex	Spinal cord (10 controls, 10 sALS, 10 C9orf72-ALS); motor cortex (8 controls, 21 sALS, 12 C9orf72-ALS); frontal cortex (11 controls, 30 sALS, 12 C9orf72-ALS)	Immunohistochemical quantification of 5-methylcytosine (5-mC) and 5-hydroxymethylcytosine (5-hmC) levels	Increased 5-mC and 5-hmC levels in ALS lower motor neurons compared with controls, but no difference in spinal cord glia. Motor neurons with TDP-43 pathology showed lower 5-mC and 5-hmC levels than those without TDP-43 pathology	[36]
Hartung et al., 2021.	Motor neurons from human induced pluripotent stem cells (hiPSC MNs)	Three healthy control cell lines and 3 FUS-ALS cell lines	Promoter methylation and expression levels of FUS. Expression levels of DNMTs	Differentiated mutant FUS motor neurons showed higher DNMT levels and increased FUS methylation and lower expression compared with the differentiated motor neurons from control cell lines	[37]
Martin et al., 2022.	Spinal cord and skeletal muscles from SOD1-mutant ALS mice; postmortem human brain and spinal cord samples	Different lines of transgenic mice expressing mutant or wild-type human SOD1 (hSOD1); matched non transgenic mice; 12 ALS patients; 12 matched controls	Quantification of 5-mC levels and DNMT levels; Genome-wide methylation	Increased DNMT activity and 5-mC levels in spinal cord and skeletal muscles of hSOD1 mutant mice, and hypermethylation of cytoskeletal-related genes in skeletal muscles. Increased 5-mC in human ALS spinal cord and motor cortex	[38]
Catanese et al., 2023.	Human induced pluripotent stem cell -derived motor neurons (hiPSC MNs); public gene expression and mutation databases	hiPSC MNs from 8 fALS donors with mutations in SOD1 (n = 2), C9orf72 (n = 2), FUS (n = 2), TARDBP (n = 2); hiPSC MNs from 3 healthy controls; blood transcriptome dataset (GSE112681) from 397 ALS patients and 645 controls; spinal cord transcriptome dataset (GSE137810) from 154 ALS patients and 49 controls; somatic mutation dataset from 773 ALS patients and 93 healthy controls	Machine-learning based integration of genomic, transcriptomic, and epigenomic data	Altered methylation and expression of genes related to synaptic function in fALS hiPSC MNs. Altered expression of genes related to toll-like receptor cascade, immune function, autophagy, and protein metabolism in ALS blood. Altered expression of genes related to synaptic function, vesicle trafficking, toll-like receptor cascade, immune function, and TP53-mediated regulation of metabolic functions in ALS spinal cord tissues. Acquired somatic mutations in ALS tissues were linked to aging and age-related decline of DNA repair mechanisms	[39]

Some investigators made attempts to link the observed changes in DNAm to the exposure to environmental factors [26,31]. For example, Freydenzon and colleagues collected blood methylomic data from 438 Australian ALS patients and 417 healthy controls and looked for correlation between DNAm and self-reported exposure to various environmental factors, lifestyles and working conditions, including farming or rural living, type of employment, exposure to toxins or chemicals, exercise, smoking, previous injuries or illnesses and medication. The authors replicated several known associations between smoking status and DNAm and found potential associations between metal exposure and changes in DNAm, in particular cadmium or mercury exposure and working in metallurgy-related fields [26]. More recently, Brusati and colleagues used the comparative genomic database, a publicly available database, to search for chemical compounds potentially associated with the genes that showed altered methylation levels in ALS blood DNA samples [31]. Numerous chemicals were potentially associated with the observed ALS blood epigenomic signatures, including benzo(a)pirene, sodium arsenite, silicon dioxide, nickel, air pollution, smoking, pesticides and many others, albeit causality could not be inferred [31].

Another very interesting suggestion from recent studies is the possibility to use DNAm data as predictors of disease age at onset and progression rate. Studies in ALS families have shown that both nuclear and mtDNA levels differ from presymptomatic to symptomatic mutation carriers and change with disease progression [5,14,25,32], and this might be attributable to the fact that ALS symptoms, including muscle wasting, paralysis and hypermetabolism, have a profound effect on DNAm changes [25,32]. More recently, some investigators suggested that DNAm data can be informative on disease age of onset, progression rate, and survival. For example, Zhang and colleagues [22] performed a whole blood EWAS in 200 sALS and 49 fALS patients observing a strong association between DNAm age acceleration and disease age of onset, and an inverse association between DNAm age acceleration and the median survival age of their patients [22]. Another EWAS in whole blood DNA samples from 32 sALS patients and 32 healthy matched controls suggested that the methylation levels of ELOVL2 were associated with disease age of onset, and those of ARID1B with disease duration [27]. A similar study in whole blood DNA samples from 41 sALS patients and 27 matched controls identified 948 differentially methylated positions (DMPs) and 298 differentially methylated regions associated with ALS progression rate, as well as 590 DMPs and 197 differentially methylated regions associated with disease survival [33]. Moreover, in a recent preprint, Caggiano and colleagues suggest that also DNAm signatures from cell-free DNA extracted from the blood plasma of ALS patients might be informative of disease severity [34]. A summary of DNAm data from ALS tissues is provided (Figure 1). Altogether these data, albeit still preliminary and pending validation, are extremely encouraging and suggestive of potential clinical applications of DNAm in ALS. However, there are still several points that, in my opinion, require to be further addressed and that will be discussed in the next sessions.

Figure 1.

A summary of the amyotrophic lateral sclerosis methylome. Studies performed in blood or central nervous system samples from patients with ALS have revealed increased global DNAm levels (5-mC content) and altered methylation and expression of genes related to synaptic function, metabolism and immune response in the central nervous system, as well as of genes related to metabolism, cholesterol biosynthesis, immune response and inflammation in the blood. Furthermore, the ALS blood methylome can discriminate between ALS patients and controls. Most of the DNAm changes observed in the blood of ALS patients occur after disease onset, and potential associations with disease age of onset and disease survival have been suggested. Some but not all authors observed DNAm age acceleration in ALS. The genetic background, environmental exposures and the metabolic changes occurring after disease onset are likely contributors of the observed DNAm changes.

ALS: Amyotrophic lateral sclerosis; DNAm: DNA methylation.

3. Can DNAm in peripheral tissues provide insights into individuals at risk of developing ALS in the coming years?

Whether or not certain changes in DNAm precede symptoms onset and may be informative about individuals at risk of developing ALS symptoms in the coming years is an interesting and still unsolved question. Humans with known pathogenic mutation in ALS causative genes, but still presymptomatic, represent individuals at risk to develop disease symptoms in the coming years. Other models of a presymptomatic ALS stage are transgenic mice before the onset of symptoms. Recently, Ruf and colleagues performed a methylome analysis in blood cell DNA from 65 sALS and 34 fALS patients (24 C9orf72 expansion carriers; 10 carriers of pathogenic FUS mutations), presymptomatic carriers of C9orf72 expansions (n = 11) or pathogenic FUS mutations (n = 6), and 72 healthy matched controls, observing numerous DMPs associated with either disease status, C9orf72 or FUS mutation status [25]. Furthermore, the application of a classifier model that predicts disease status using blood cell methylomic data, classified all but one presymptomatic mutation carrier as healthy. Based on their results, the authors concluded that most blood cell DNAm changes in ALS patients result from disease symptoms, rather than from the presence of ALS-associated genetic mutations and are not present before the onset of symptoms. Therefore, according to their data, authors concluded that symptoms such as muscle wasting, paralysis and disease-related metabolic changes, could have a stronger effect on DNAm than disease-causative mutations [25]. Results by Ruf and colleagues [25] are similar to those that we previously observed in SOD1 ALS families [5]. Indeed, we collected blood cell DNA samples from five ALS families with SOD1 mutations, including six ALS patients, 13 presymptomatic SOD1 mutation carriers, and 16 healthy family members among non-carriers of SOD1 mutations. We then quantified 5-mC levels in blood cell DNA, observing that they were significantly higher in ALS patients than in presymptomatic mutation carriers or in non-carrier family members. Notably, pre-symptomatic mutation carriers and non-carrier family members showed very similar methylation levels in blood cell DNA. By contrast, in ALS patients the global DNAm levels increased with the duration of the disease from symptoms onset [5]. Another study in ALS families included 29 SOD1 ALS patients, 13 presymptomatic SOD1 mutation carriers, 18 C9orf72 ALS patients, six presymptomatic C9orf72 expansion carriers, four FUS ALS patients, five presymptomatic FUS mutation carriers, 3 TARDB ALS patients, 4 presymptomatic TARDBP mutation carriers, and 32 healthy family members without known pathogenic ALS mutations. The study revealed changes in mtDNA methylation levels in SOD1 mutation carriers, both symptomatic and in presymptomatic stage, respect to healthy controls. No such changes were observed in fALS samples with C9orf72, FUS, or TARDBP mutations [14]. Also studies in SOD1 transgenic ALS mice revealed certain changes in mitochondrial DNMT activities and mtDNA methylation that precede the onset of symptoms [13].

Altogether these studies suggest the existence of gene-specific epigenetic signatures in ALS but have failed to identify common epigenetic signatures that precede the onset of disease symptoms. Considering that the most recent large-scale gene expression studies in ALS tissues have confirmed that most changes occur in pathways related to inflammation, immune system and metabolism, it is not unlikely that most of the epigenetic changes observed in ALS tissue result from disease symptoms [25,39]. However, only a few presymptomatic carriers of ALS-linked mutations have so far been investigated, thus limiting the power to detect early methylomic changes. Furthermore, longitudinal DNAm studies are missing in those individuals [5,13,25]. Interestingly, when Ruf and coworkers [25] applied a classifier model that predicts disease status using blood cell methylomic data, one of the presymptomatic C9orf72 expansion carriers was not correctly classified as healthy, suggesting that its methylome was closer to that of symptomatic patients than to those of the other presymptomatic carriers. Was this a subject close to convert to a symptomatic stage? Only longitudinal studies could help addressing this question.

Furthermore, presymptomatic carriers of ALS-causative mutations or transgenic ALS animal models are only partially representative of the sporadic cases, which represent most of ALS. Recent evidence points to a long pre-symptomatic ALS stage and suggest a disease continuum ranging from clinically silent, to prodromal, to clinically manifest [40–43]. Changes in both blood metabolic profiles and neurofilament light chain levels are emerging as potential biomarkers of the presymptomatic disease stages and such biomarkers might be used in the near future to identify individuals in the preclinical or prodromal ALS stages [41–43]. Longitudinal epigenetic studies in those individuals are needed to identify earliest epigenetic signatures that might precede the onset of symptoms. Therefore, further investigation and longitudinal studies in larger cohorts of presymptomatic individuals are required prior to exclude that certain common epigenetic signatures might precede and be informative of symptoms onset in the near future.

4. Environmentally induced DNAm changes in ALS, what do we really know?

Numerous studies on major neurodegenerative diseases, such as Alzheimer's disease (AD) or Parkinson's disease (PD), are highlighting potential gene-environment interactions occurring early in life or throughout the life course and resulting in epigenetic modifications of genes coding for proteins involved in the main pathogenetic mechanisms of those conditions, but limited data are still available for ALS [44,45]. Among the numerous potential AD risk factors capable of inducing DNAm changes, there is substantial evidence from studies in rodents and primates suggesting that early-life exposure to lead (Pb) induces long-lasting epigenetic changes in the developing animal brain, resulting in an increased expression of the APP gene ultimately leading to AD-like pathology in adulthood [46,47]. APP codes for the precursor protein of the amyloid-beta (Aβ) peptide, the neurotoxic peptide that accumulates and forms amyloid plaques in the AD brain and represents a causative gene for familial AD [46,47]. Similarly, studies in neuronal cell cultures, animal models and in human AD blood samples, have shown that reduced levels of folate and other B-group vitamins result in hypomethylation and increased expression of PSEN1 and BACE1 genes, both involved in the amyloidogenic cleavage of APP that generates the Aβ peptide. Indeed, AD-like features were observed in rodent brains following a B-vitamins deprivation resulting in PSEN1 hypomethylation and overexpression [48–50]. Pesticide exposure is a well-known PD risk factor and increasing evidence from neuronal cell cultures and PD animal models suggests that exposure to various pesticides can induce DNAm changes in several PD-related genes [51–53]. Unfortunately, the promoters of major ALS genes, including SOD1, C9orf72, FUS and TARDBP, are almost completely demethylated in samples from ALS patients and recent large-scale studies failed to observe a dysregulated expression of major ALS genes in ALS tissues, thus preventing us from having candidate genes to test in cellular and animal models exposed to different environmental conditions [24,39]. Only certain C9orf72 hexanucleotide repeat expansion carriers show hypermethylation of the C9orf72 promoter region, but this is believed to be a compensatory mechanism to counteract the deleterious effects of large repeat expansions [54].

Anyway, several attempts have been made to link methylomic data observed in ALS patients to environmental exposures. Hop and coworkers used published polymethylation scores as proxies for several exposures and traits, observing that polymethylation scores for high density lipoprotein cholesterol (HDL-C), monocyte cell proportion and granulocyte cell proportion were positively associated with ALS [24]. By contrast, an inverse association with ALS was observed for polymethylation scores for alcohol intake, body mass index and other white blood cells proportions [24]. Freydenzon and colleagues observed seven epigenome-wide significant CpGs associated with self-reported exposure to cadmium, mercury and metallurgy [26], while Brusati and coworkers observed that the methylomic signatures observed in ALS blood DNA samples could result from exposure to more than 30 different chemical compounds [31]. Several investigators have also reported epigenetic age acceleration in ALS and/or linked DNAm age acceleration to age of disease onset [9,15,16,22,25]. Various environmental factors such as smoking, excessive alcohol use, poor nutrition and others have been linked to increased DNAm age [55]. However, recent large-scale studies could not unambiguously confirm DNAm age acceleration in ALS patients, so that this point is still controversial [24,31]. Moreover, DNAm age acceleration has been linked to increased risk of various human diseases, not exclusively of ALS [55]. Overall, despite that it is not unlikely that several environmental exposures within the course of life can leave DNAm changes potentially contributing to ALS onset and progression, the studies performed so far and attempting to link the ALS methylome to environmental exposures can say little about causality [24,26,31]. Once again, we need large-scale longitudinal studies to understand which epigenetic changes precede and might contribute to disease onset and which are instead the result of the profound metabolic changes that occur after disease symptoms are manifest. This point will be discussed further in the next sessions.

5. Is there any sex difference in DNAm levels?

Sex differences in DNAm levels have often been reported in neurodegenerative conditions [56,57], but little is still known about ALS. Most recent studies performed on blood DNA samples from ALS patients, listed in Table 1, have included sex-matched case-control cohorts, and sex adjustment was performed in data analysis, but only a few of them show data on sex comparisons of blood methylation levels [21–34], and the same is true for previous investigations in ALS tissues [4–9]. In the genome-wide investigation conducted by Zhang and coworkers [22], the authors observed that the association between DNAm age acceleration and disease age of onset and survival was significant in both males and females [22]. Others observed that the mtDNA D-loop methylation levels were similar between sexes in ALS patients [23]. Similarly, no differences in the promoter methylation levels of LGALS1 were observed between ALS males and females [28]. Studies performed in brain or spinal cord tissues are often limited to a few samples (Table 2), and do not show sex comparisons of DNAm data. Additional studies are warranted to further investigate potential sex differences in DNAm levels from ALS tissues, especially from the tissues primarily affected by the disease.

6. The potential clinical utility of the ALS methylome

Several investigators observed increase DNAm age in ALS patients [9,15,16,22,25], and among them Zhang and colleagues observed a strong association between DNAm age acceleration and disease age of onset, and an inverse association between DNAm age acceleration and the median survival age of ALS patients [22]. Similarly, Ruf and coworkers observed increased DNAm age in their cohort of ALS patients, and a trend for increased DNAm age in presymptomatic carriers of both C9orf72 and FUS mutations [25]. Unfortunately, the largest methylomic study in ALS samples by Hop and colleagues [24], that included more than 9.700 blood DNA samples from ALS samples and healthy controls, failed to confirm the association between DNAm age acceleration and disease survival. Moreover, due to a substantial heterogeneity in their data, authors could not unambiguously support a role for DNAm age acceleration in ALS [24]. Also, the recent case–control study by Brusati and colleagues failed to observe DNAm age acceleration in ALS cases [31]. However, despite that a role for DNAm age acceleration in ALS age of onset and survival is still questioned, Hop and coworkers identified five DMPs associated with disease survival that survived correction for multiple confounding factors [24]. Authors also used polymethylation scores as proxies for various traits, observing association between blood cell composition and disease survival [24]. Other recent studies comprising few dozens of matched case-control samples have suggested potential associations with ALS age of onset, progression rate and survival [27,33], and very recent and still unpublished data suggest that also the methylation status of cell free DNA samples from the blood plasma could be informative of disease severity [34].

Collectively, these data point to potential clinical applications of DNAm data in ALS, but again one of the missing points is the longitudinal availability of DNAm data. How many of these potential epigenetic biomarkers of disease progression rate and survival can be detected at symptoms onset? Are these positions already differentially methylated before the onset of symptoms? Is their methylation status changing within disease progression? What is the correlation between the DNAm changes and the metabolic changes that occur within disease progression? All these questions need to be addressed in the near future in order to clarify if the analysis DNAm data, especially in the preclinical stages of ALS, could be informative of the success of interventions aimed at slow down the disease progression rate.

7. DNAm in the different forms of ALS presentation

The clinical presentation of ALS is heterogeneous as is the course of the disease, and patients may present with variable degrees of upper or lower motor neuron involvement. In most patients, symptoms begin in the upper or lower limbs, and affected individuals may complain of muscle twitching and weakness in one arm or leg. However, about 20–25% of ALS patients present with bulbar symptoms, including dysarthria, dysphagia and dysphonia. Less frequently (1–3% of the cases), ALS presents with respiratory muscle weakness as the initial manifestation [58]. Such heterogeneity at presentation could be reflected by different methylomic signatures, but studies addressing this question are still limited. Indeed, as discussed in the previous section, longitudinal methylomic data collections from presymptomatic to end-stages disease are lacking, and it is extremely important to clarify this point. Several investigations conducted in the second decade of the 21st century have reported increased 5-mC levels in the blood or spinal cord DNA of ALS patients compared with controls [4–7]. Among these, Tremolizzo and coworkers searched for differences in the 5-mC content in blood DNA from ALS patients with bulbar onset or limb onset, but no difference was observed between the two groups [4]. More recently, the blood methylomic study by Zhang and colleagues revealed that the observed association between DNAm age acceleration and ALS age of onset and survival was not driven by subgroups stratified by site of onset and remained significant in both patients with bulbar onset and those with limb onset [22]. In contrast, the study by Cai and coworkers not only revealed reduced promoter methylation and increased mRNA levels of the LGALS1 gene in ALS patients compared with controls, but also showed differences in LGALS1 promoter methylation levels between ALS patients with bulbar onset and limb onset [28]. Beyond these studies, comparisons of DNAm data across different forms of ALS presentation are lacking and will be warranted in future investigations.

8. Conclusion

In conclusion, the numerous investigations of DNAm in ALS samples have confirmed numerous epigenetic changes in ALS tissues. In particular, genes involved in immune response, inflammation and metabolism are among the most likely affected by changes in DNAm and expression levels in ALS tissues. A still open question is whether these changes precede the onset of disease symptoms or are likely a consequence of the profound muscular and metabolic changes that occur after symptoms onset. Further studies in presymptomatic individuals at higher genetic ALS risk, as well as in individuals with a biomarker-based definition of preclinal disease stage will be helpful for the identification of early epigenetic biomarkers of the disease, for a better understanding of the contribution of environmental factors to the observed DNAm changes, as well as for the design of potential interventions aimed at lowering disease progression.

9. Future perspective

In my view, the studies performed so far in ALS samples have clearly demonstrated that DNAm changes occur in ALS tissues. Numerous metabolic genes and genes involved in inflammatory pathways and immune response are among those most affected by changes in DNAm and gene expression, and changes observed in whole blood DNA samples often mirror those observed in spinal cord samples. Are these changes cause or consequence of the disease? Can they represent traces left on the genome by early-life exposure to environmental risk factors? Can they be reverted by dietary interventions, lifestyles or medications to slow down disease progression? These are still open questions to address in the coming few years. To shed light on early ALS-related DNAm changes, I believe that efforts should be made to collect large cohorts of presymptomatic carriers of ALS-linked mutations, such as those in SOD1, C9orf72, FUS and TARDBP genes, as they represent individuals at high genetic risk of conversion into a symptomatic stage. Longitudinal DNAm studies in those individuals could help to identify both shared and gene-specific epigenetic changes that precede the onset of disease symptoms as well as their correlation with disease progression rate and survival. Those studies should be paralleled by large scale longitudinal epigenome-wide investigations, including an elevated number of samples whose blood-based biomarkers could be indicative of individuals in presymptomatic ALS stages. Follow-up studies in those cohorts and comparison studies between converters and nonconverters into a symptomatic stage might represent the best strategy to identify early epigenetic biomarkers of clinical utility and see if and how these biomarkers can predict disease onset and severity (Figure 2). Once discovered in these cohorts, early ALS methylation signatures could be used as proxies of environmental exposure, and validation studies could be performed in both neuronal cell cultures and animal models. Similarly, they could represent promising biomarkers to investigate the potential of dietary, lifestyle and medication interventions aimed at slowing down disease progression.

Figure 2.

Proposed strategies to further address DNA methylation in amyotrophic lateral sclerosis. Longitudinal investigations with repeated measures of genome-wide DNA methylation in blood samples from individuals at higher ALS risk, including (A) Carriers of familial ALS (FALS) gene mutations in a presymptomatic stage and (B) Large cohorts of individuals with a biomarker-based definition of preclinical ALS will be helpful for the identification of early epigenetic biomarkers of the disease, including those resulting from specific gene mutations and those shared among familial and sporadic ALS cases. Repeated investigations at symptoms onset and within disease progression could also allow discriminating the epigenetic changes occurring before and after disease onset and reveal which of them can predict disease onset and severity. Collectively, these studies can help identify early ALS related methylomic changes, their links with environmental exposures, lifestyles, dietary or medical interventions and their potential clinical utility.

Author contributions

F Coppedè conceived and designed the perspective and wrote the article.

Financial disclosure

The authors have no financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Competing interests disclosure

The authors have no competing interests or relevant affiliations with any organization or entity with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Writing disclosure

No writing assistance was utilized in the production of this manuscript.