Introduction

Amyotrophic lateral sclerosis (ALS, OMIM #105400) is a rare and devastating neurodegenerative disorder of unknown etiology characterized by rapidly progressive paralysis leading to death due to respiratory failure, typically within 3–5 years of symptom onset. Population-based epidemiological studies of the disease show that between 1.6% and 5.7% of cases are familial in nature, whereas the remaining 95% occur sporadically throughout the population [1],[2]. Various genes that cause familial ALS have been identified, including copper/zinc superoxide dismutase [3], dynactin 1 [4],[5], alsin [6], senataxin [7], and vesicle-associated protein B [8]. In contrast to familial ALS, the genetics of the sporadic form are poorly understood and it is not known if sporadic ALS is monogenic (i.e., a single-gene disorder), polygenic (i.e., multiple interacting genes), multi-factorial (i.e., interacting genetic and environmental factors), or arises from some unknown non-genetic cause [9].

The neuropathology of ALS is characterized by the abnormal accumulation of insoluble ubiquitin proteins in the cytoplasm of degenerating motor neurons [10]. Recently, TAR DNA-binding protein 43 (TARDBP, OMIM #605078) was recognized as a major constituent of these neuronal cytoplasmic inclusions [11],[12],[13]. TDP-43 protein is evolutionary conserved and its structure consists of a glycine-rich domain and two RNA recognition motifs [14]. The exact function of TDP-43 remains unclear, though it is known to bind DNA and RNA (such as human immunodeficiency virus type 1 TAR DNA sequence motifs) [15], and to be involved in the regulation of messenger RNA splicing and exon skipping [16]. Four studies have independently reported TARDBP mutations in familial forms of ALS displaying autosomal dominant inheritance and in sporadic ALS. An A315T mutation in exon 6 of TARDBP was reported to segregate with disease within a multi-generational ALS family and this sequence variant was not found in 1,505 control subjects [17]. In other study, a missense mutation M337V in exon 6 of TARDBP segregated with disease within an ALS family across two generations. In addition, two missense mutations (Q331K and G294A) were identified in two of 372 sporadic ALS cases screened for mutations (representing 0.5% of the total cohort). None of the three mutations were identified in 1,262 control subjects [18]. In a separate study of 200 individuals with ALS from France and Quebec, eight additional sequence variants were identified in TARDBP including the previously identified A315T missense mutation and A382T in small families and D169G, G287S, G348C, R361S, N390D and N390S in individual sporadic cases [19]. In the latest study, two additional missense mutations (G290A and G298S) were identified in familial ALS cases [20]. None of the sequence variants identified in sporadic ALS patients have been seen in more than one case.

In this study, we undertook mutational screening of TARDBP in a cohort of 279 North American sporadic ALS cases and in a cohort of 806 ethnically-matched controls. The same case-control cohort had been previously genotyped using Illumina HumanHap550 BeadChips which assay over 550,000 tagged SNPs across the human genome [9]. This genotype data was analyzed to identify (a) genomic structural variants of the TARDBP locus relevant to the pathogenesis of sporadic ALS, (b) genetic association of SNPs within TARDBP with disease status, and (c) common haplotypes across the TARDBP locus that alter risk of motor neuron degeneration. Finally, we sequenced TARDDP in the African samples of the Human Gene Diversity Panel (n = 173) to examine genetic diversity of this gene in non-Caucasian populations.

Materials and Methods

Results

Table 1 details the sequence variants found in the 279 ALS cases, the 806 North American controls and in the 173 African samples that were screened as part of this study. Pathogenic mutations were not identified in any of the 279 ALS cases. Five of the ALS samples (ND09546, ND10023, ND09582, ND10379, ND10379) carried the synonymous variation p.A66A (c.198 T>C) in exon 2 of TARDBP, but this sequence variant was likely to be a benign polymorphism as it was found in eight of the Caucasian controls and in a single Bedouin sample of the HGDP. Furthermore, the previously reported pathogenic sequence variants were not present in any of the case or control samples, though a synonymous variant involving codon 315 (p.A315A, c.945G>A) was identified in a single Caucasian control individual (ND05681).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Sequence variations found in the TARDBP gene in the 279 ALS cases, 806 North American controls and 173 African samples.

https://doi.org/10.1371/journal.pone.0002450.t001

Nineteen control samples were excluded due to poor quality genotyping on the Illumina HumanHap SNP chips (i.e. call rates <95%). Thus the final cohort for which genotyping chip data was available consisted of 279 ALS cases and 787 control individuals. Copy number analysis of the TARDBP locus using data generated with Illumina Infinium II HumanHap550 BeadChips did not reveal structural abnormalities affecting the TARDBP gene (±75kb) in any of the 279 cases.

Statistical analysis of association was done for 42 tagging SNPs within the TARDBP gene and the flanking 100kb, irrespective of Hardy-Weinberg disequilibrium or minor allele frequency. One SNP (rs12059717 located 17.kb from the gene) was excluded due to low call rate. None of the remaining 41 SNPs were significantly associated with altered risk of developing ALS under the additive model. Similarly, no haplotype significantly altering disease risk was identified (Supplementary Tables S1 and S2).

Discussion

In this study we evaluated the role of TARDBP in the pathogenesis of sporadic ALS by undertaking sequence mutational analysis, as well as evaluating genomic structural variation, genetic association and haplotype association in a cohort of 279 North American ALS patients and 806 neurologically normal controls. We did not find any genetic or genomic evidence that TARDBP was associated with sporadic disease. Although the size of our cohort is not large enough to exclude this gene as the underlying etiology of occasional rare cases of sporadic ALS, our data indicates that mutations in TARDBP are not a common cause of sporadic motor neuron degeneration in the North American population (i.e. no greater than 0.4% of all cases). Although mutations in the TARDBP gene are rare, the protein may still play an important role in ALS. In fact, several studies have linked ALS pathogenesis to perturbations of TDP-43 nuclear trafficking, solubility and intracellular accumulation [25]–[27].

Our data agree with previous reports of a negative association between TARDP and ALS in 237 sporadic patients [28]. Furthermore, we did not find any case or control samples in our series that carried any of the previously reported pathogenic mutations (A315T, A382T or M337V) that have been reported in familial cases, or any of the 8 variants (D169G, G287S, G294A, Q331K, G348C, R361S, N390D and N390S) that each have been reported in single sporadic cases. Although the published segregation data strongly supports the pathogenicity of the M337V and A315T mutations within these families, the lack of further examples of the eight mutations identified within sporadic ALS cases suggests that the true pathogenic nature of these variants remains to be established. Interestingly, all the variants found and considered by the authors as pathogenic, with the exception of D169G in exon 4, are situated in exon 6; all non-synonymous but benign variants identified by us, lie outside of this exon. These mutations may alter the normal function/transport of TDP-43 or they may cause a toxic gain of function of the protein by altering its C-terminal [19].

Structural variants of the TARDBP gene were not a cause of sporadic ALS in our dataset. Although the density of SNPs on the Illumina HumanHap550 is not sufficient to exclude small deletions or insertions (42 SNPs over 212.9Kb, 1 SNP per 5Kb), our sequence analysis did excluded the presence of small deletions/insertions within the coding sequence and flanking introns of TARDBP. Finally, neither genetic association nor haplotype association with disease was observed across the TARDBP locus. Although this analysis was based on a relatively small number of markers, these 42 SNPs were tagging SNPs that are representative of neighboring sequence variation and extract a larger amount of genetic information than a similar number of randomly chosen SNPs.

We sequenced TARDBP in a cohort of 173 African samples to determine the variability of the gene in a non-Caucasian population and because this population carried the highest likelihood of finding changes. As expected, given the greater genetic variation known to exist within African populations, SNPs were more commonly present in this cohort compared to the Caucasian controls (6,9% compared to 1,1%). This observation is consistent with the older age of these populations with the consequent accumulation of sequence variants over a greater number of generations [29], [30]. Interestingly, we observed only two non-synonymous SNPs within TARDBP in both the African and Caucasian controls (i.e. D65E and A90V). We speculate that the biological importance of this gene has resulted in negative selective pressure against variants that alter TDP-43 amino-acid sequence.

In conclusion, our comprehensive genetic and genomic assessment of TARDBP failed to identify disease associated variants in the North American patients studied here. Taken together with previous reports, we conclude that mutations in TARDBP are not a common cause of sporadic motor neuron degeneration.

Supporting Information