Background
Amyotrophic lateral sclerosis (ALS) is a devastating neurodegenerative disorder, characterized by the death of motor neurons as well as paralysis of voluntary muscles [
1]. The majority of ALS patients die within 3~5 years after diagnosis, mostly due to respiratory failure [
2]. The pathological mechanisms underlying ALS are multifarious and complex, with a sophisticated interaction between genetic and environmental factors [
3]. To date, no effective therapies for ALS have been found, and approved drugs only improved survival to a limited extent [
4]. Therefore, exploring the pathogenesis of ALS and developing novel therapeutic strategies are necessary and urgent.
Currently, compelling evidence implicates that dysregulated immunities are associated with ALS [
5]. It has been established that neuroinflammation is overactivated in ALS, accompanied by microglia transformation, astrocyte proliferation, perivascular infiltration of monocytes and T cells, and dysregulated immune-related genes [
6‐
9]. Typical hallmarks of autoimmunity occurring in the pathogenesis of ALS have also been reported, such as the presence of circulating immune complexes and the evidence of higher frequency of specific histocompatibility types [
10]. Moreover, epidemiologic studies presented that several pre-existing autoimmune disorders are associated with an increased risk of ALS [
11]. And intermediate alleles of
C9orf72, the most common genetic cause of ALS, were suggested to be associated with systemic autoimmune diseases, indicating the role of
C9orf72 in immunity regulation [
12]. Furthermore, mice harboring loss-of-function mutations in the ortholog of
C9orf72 cause fatal autoimmune diseases [
13]. These associations raise the possibility of shared genetic or environmental risk factors, or clues to modifiable triggers that might thereby affect ALS incidence. Therefore, a systematic study is necessary to decipher whether shared polygenic risk variants exist between ALS and autoimmune diseases, and whether specific molecular biological pathways are involved.
Recently, a novel statistical method to investigate genetic overlapping between polygenic traits using summary data from genome-wide association studies (GWAS) have been developed and utilized extensively in several human traits and diseases [
14‐
16]. By incorporating GWAS results from multiple disorders and phenotypes, this method could provide insights into the genetic pleiotropy (defined as a single gene or variant being associated with more than one distinct phenotype) and increased statistical power to discover less significant associations [
14‐
16]. Applying this approach, we systematically evaluated the shared genetic risk between ALS and autoimmune diseases and further conducted functional enrichment analysis.
Discussion
In the current study, we investigated the pleiotropy between ALS and autoimmune disorders using summary statistics from large GWAS and the conditional FDR statistical method. We identified significant genetic enrichment for ALS as a function of CeD and MS and moderate enrichment of UC and T1D. Besides, we identified 13 significant shared loci, with 5 validated in the replication ALS GWAS, and 3 were annotated as related to gene expression in tissues from both Braineac and GTEx. These results clarified the shared genetic architecture between ALS and autoimmune diseases, suggested that ALS pathogenesis might be mediated by the dysfunctioning of the immune system, and provided a better understanding for the pleiotropy of ALS.
Neurodegenerative disorders are complex diseases characterized by loss of neurons and axons in the central nervous system. Recently, there is increasing recognition that inflammation and immune response play important roles in the pathogenesis of neurodegeneration [
42]. Activated microglia, which is a key regulator of brain inflammation, is also an important cause for neuronal loss in models of neurodegenerative diseases [
43]. Previous studies have investigated pleiotropy between AD, PD and autoimmune diseases [
15,
16], and identified shared risk genes. However, the pleiotropy between ALS and autoimmune disorders is still not clear. Our results supported the hypothesis of shared genetic risk between ALS and autoimmune diseases and supplemented current knowledge for the correlation between neurodegeneration and autoimmunity.
We observed a substantial genetic enrichment between CeD and ALS. CeD is an autoimmune disorder that occurs in genetically predisposed individuals who develop an immune reaction to gluten [
44]. Meanwhile, gluten sensitivity has also been shown to induce neurologic manifestations [
45]. For example, a recent case-control designed study suggested that ALS might be associated with autoimmunity and gluten sensitivity [
44]. However, such links still need to be confirmed further, as subsequent studies reported inconsistent results [
46]. We also identified a shared risk gene
GPX3, which could have functional relevance to both diseases.
GPX3 is an antioxidant molecule functionally related to
SOD1 [
47], the first causative gene for ALS. In a mass spectrometric screen of sera of
SOD1H46R rats compared to wild-type controls,
GPX3 expression was increased by 1.3 fold in the pre-symptomatic stage, while decreased by 0.74 fold in the end stage of the disease [
48]. Meanwhile,
GPX3 activity was reduced in CeD patients (
P < 0.001) in a recent pilot study, suggesting its potential role in the pathogenesis of CeD [
49]. In contrast, minimal enrichment was observed between ALS and CD, which has partially shared genetic basis and pathogenesis with CeD [
50]. However, the strongest signal rs2076756 (
NOD2) in the conjunctional Manhattan plot was identified between ALS and CD.
NOD2 is a member of the pattern-recognition receptor family and can recognize muramyl dipeptide in cytomembrane to activate the NF-κB pathway and stimulate inflammatory factor response [
51]. Recently,
NOD2 was identified as a potent autophagy inducer as well, suggesting its potential role in neurodegeneration [
52]. In addition, moderate enrichment was observed between ALS and UC, and
G2E3-
SCFD1 was identified as shared risk genes between ALS and IBD. These results together suggested a potential link between ALS and chronic inflammation of the gastrointestinal tract, though such link was not consistently detected based on current results. Further explorations are warranted to elucidate the correlation.
We also observed genetic enrichment between ALS and MS. Both ALS and MS are complex neurological disorders with sophisticated interactions of environmental toxicity and genetic predisposition [
53]. By far, the origin of MS and ALS is still unknown, but the progressive central axonal degeneration is seen in both MS and ALS. And there is evidence suggesting shared cellular mechanisms affecting the disease progression, particularly glial responses in the two disorders [
54]. One explanation would be that the two conditions share some common genes which predispose to both MS and ALS. Here, we identified six shared risk genes including
TRIP11,
ATXN3,
GGNBP2,
SLC9A8,
DENND6B, and
ANO5 (Table
1). The six shared genes were enriched in two significant cellular components, namely organelle subcompartment (GO:0031984,
P = 0.0008) and endomembrane system (GO:0012505,
P = 0.0032) based on results from ConsensusPathDB. Therefore, the membrane system may implicate the overlapping factors related to MS and ALS, and the shared genes might be responsible for the link.
Association between ALS and diabetes has been observed in epidemiological studies [
55,
56]. Recently, a causal protective role of type 2 diabetes on ALS was noticed in the European population [
57]. In contrast, the association between T1D and ALS was still less understood, and T1D might increase the risk for ALS based on a recent retrospective population-based study [
11,
56], though such results were still awaiting further replications [
58]. In our study, we observed enrichment between ALS and T1D, and identified three shared risk genes
C9orf72,
ZNF184, and
ATXN2. Repeat expansions in
C9orf72 is a frequent cause of ALS, and
C9orf72 carriers tend to have autoimmune diseases more frequently, suggesting autoimmune inflammation may be intrinsically linked to ALS pathophysiology [
59]. Meanwhile, a recent study found that mutations disrupting the normal function of
C9orf72 cause mice to develop features of autoimmunity [
12]. Thus,
C9orf72 might serve as an important factor related to inflammation and autoimmunity [
12].
ZNF184 is in the extended major histocompatibility complex (MHC) region, which plays a complex but important role in both neurodegenerative and autoimmune diseases.
ATXN2 has been reported to be associated with several autoimmune diseases like T1D [
60], CD [
61], and CeD [
62] by GWAS. Meanwhile, high-length repeats of CAG trinucleotide in
ATXN2 was identified as a risk factor for ALS as well [
63,
64]. Taken together, genetic correlation exists between ALS and T1D, and genes such as
C9orf72,
ZNF184, and
ATXN2 might function as potential links between the two diseases.
Compared with CeD, MS, T1D, and UC, we did not detect enrichment between ALS and RA, psoriasis, asthma, CD, and SLE. The results were to a large extent in agreement with a previous epidemiologic study which investigated whether ALS incidence was higher in people with prior autoimmune diseases [
11]. This study found that patients with CeD, younger-onset diabetes, MS, asthma, and SLE were at higher risk of ALS, and the rate ratio for UC was borderline significant (
P = 0.05), and no significant difference was found for RA, psoriasis and CD. As for the different results for asthma and SLE between our study and this epidemiological study, further explorations were warranted to elucidate their correlation with ALS. Moreover, we noted that both positive and negative genetic correlations were observed between ALS and autoimmune disorders (Additional file
1: Table S2). We identified a significant and positive genetic correlation between ALS and CeD, MS, RA, and SLE, indicating that the direction of effect of risk-increasing and protective alleles is consistently aligned between ALS and these autoimmune disorders at genome wide. In contrast, we found a significant and negative genetic correlation between ALS and IBD, UC, and CD, implying these diseases may possess divergent biological mechanisms from the other autoimmune disorders. Such results were in line with previous epidemiological research, which detected no significant difference in the incidence rate of ALS among patients with prior UC and CD compared with controls [
11]. Additionally, IBD, including UC and CD, was characterized by chronic inflammation of the gastrointestinal tract, which fulfilled some of the criteria required for classification as autoimmune disorders, while the extent of the involvement remained to be determined [
65]. Taken together, immune and autoimmune mechanisms are complex and may vary from disease to disease, so a deeper investigation into the molecular mechanisms involved in these diseases is necessary to further understand the correlation.
Combining cFDR analyses results and eQTL analyses results from Braineac and GTEx, we identified three risk genes, namely
SLC9A8,
ATXN3, and
GGNBP2.
SLC9A8 is a kind of sodium-hydrogen exchangers (NHEs) that exchange extracellular Na
+ for intracellular H
+. Ionic homeostasis dysregulation has been proposed as the main trigger of the pathological cascade which brings to motor-neuron loss [
66]. A recent study also observed a marked increase in the persistent component of the Na
+ current from pre-symptomatic SOD1
G93A mice [
67,
68], suggesting the potential role of
SLC9A8 in ALS etiology.
ATXN3 provides instructions for making an enzyme called ataxin-3, which is involved in destroying excessive or damaged proteins. A trinucleotide repeat expansion in
ATXN3 could cause spinocerebellar ataxia type 3, a neurologic disorder that is characterized by progressive ataxia. Mutations in
ATXN3 affect neurons and other types of brain cells and alter transcription of multiple signal transduction pathways including depressed Wnt and elevated growth factor pathways [
69]. Moreover, GWAS has identified suggestively significant SNPs in
ATXN3 as associated with ALS [
17]. These findings suggest the potential involvement of
ATXN3 in ALS. Previous studies have suggested the potential role of
GGNBP2 in ALS through gene-based association analysis and summary statistics-based Mendelian randomization (SMR) analysis [
26,
70]. Meanwhile, we noticed that
GGNBP2 is an important tumor suppresser involved in several kinds of cancers [
71]. Cancer and neurodegenerative diseases are like the two sides of a coin. The pathways that cause neuronal apoptosis, like mitogen-activated protein kinase (MAPK) signaling, can cause uncontrolled neuronal growth as well. Epidemiological studies have shown that the overall risk of cancer was significantly reduced in patients with ALS [
72]. Therefore,
GGNBP2 might act as potential genetic links between ALS and cancer.
Using the pleiotropic genes for functional enrichment analysis in ConsensusPathDB, we identified four enriched pathways, all of which were somehow involved in the pathogenesis of ALS. Membrane trafficking has been implicated in virtually every aspect of neuronal function, particularly neuronal maintenance and degeneration [
73]. Intracellular membrane trafficking defects affecting key neuronal functions may be an early determinant of motor neuron loss in ALS [
74]. Meanwhile, the molecular regulation of intracellular and extracellular vesicle trafficking is an important pathway in ALS pathogenesis [
75], and mutation in the vesicle-trafficking protein has been implicated to cause late-onset spinal muscular atrophy and ALS. Fragmentation of the Golgi apparatus has been detected in motor neurons of ALS patients and animal models [
76], and mutant
SOD1 was shown to inhibit secretory protein transport from the ER to Golgi apparatus [
77].
Combining results from previous publications that estimate pleiotropy between AD, PD and autoimmune diseases using the conditional FDR method, we found that all three neurodegenerative disorders were strongly or moderately enriched on immune-mediated enteropathy like CeD, CD, and UC. Gut microbiota has been shown to play an essential role in IBD and CeD [
78], while in recent years, gut microbiome alterations were found to be closely related to neurodegeneration as well [
79]. The microbiota-gut-brain axis composed of endocrinological, immunological, and neural mediators was even proposed due to its involvement in neurodegenerative diseases [
80]. Our pleiotropic analyses together with previous studies further emphasized that gut microbiota plays an important role in the pathogenesis of neurodegenerative disorders. In addition, differential enrichment was also observed between autoimmune disorders and AD, PD, and ALS respectively. The diversity suggested the variation between the pathogenesis of different neurodegenerative disorders.
Strengths and limitations
Using the pleiotropy-based statistical method, we identified novel significant SNPs associated with ALS and autoimmune disorders. These SNPs were mostly suggestively significant in the original studies but overlooked due to the limitation of the current GWAS sample size. Moreover, the current results have clinical implications. Since we combined GWAS summary statistics from different but related diseases, the findings may increase our understanding of the pathogenetic mechanisms influenced by pleiotropic loci and facilitate novel treatment strategies in clinical trials. However, the GWAS used in the current study were mostly performed on participants of European ancestry; thus, the findings of shared genetic architecture might be biased and not applicable to other populations. Meanwhile, there was a potential sample overlap between each GWAS. Such overlap might bring some bias to the statistical analysis, although the bias will be minimal. Meanwhile, there was sample overlap between the discovery ALS GWAS and replication ALS GWAS. To strengthen the validity of the identified pleiotropic SNPs, we utilized a more stringent significance threshold (P = 1E−05) instead of the commonly used threshold in replication (P = 0.05). In addition, the sample size for some diseases like CeD and SLE was relatively smaller, which might influence the results to some extent. Besides, although we identified several pleiotropic SNPs, how these SNPs or genes involved in the pathogenesis of ALS and autoimmune diseases were still little understood. Further functional explorations will provide better understandings.