Mutations in the FUS/TLS Gene on Chromosome 16 Cause Familial Amyotrophic Lateral Sclerosis

See allHide authors and affiliations

Science  27 Feb 2009:
Vol. 323, Issue 5918, pp. 1205-1208
DOI: 10.1126/science.1166066

This article has a correction. Please see:


Amyotrophic lateral sclerosis (ALS) is a fatal degenerative motor neuron disorder. Ten percent of cases are inherited; most involve unidentified genes. We report here 13 mutations in the fused in sarcoma/translated in liposarcoma (FUS/TLS) gene on chromosome 16 that were specific for familial ALS. The FUS/TLS protein binds to RNA, functions in diverse processes, and is normally located predominantly in the nucleus. In contrast, the mutant forms of FUS/TLS accumulated in the cytoplasm of neurons, a pathology that is similar to that of the gene TAR DNA-binding protein 43 (TDP43), whose mutations also cause ALS. Neuronal cytoplasmic protein aggregation and defective RNA metabolism thus appear to be common pathogenic mechanisms involved in ALS and possibly in other neurodegenerative disorders.

Amyotrophic lateral sclerosis (ALS) is a progressive, uniformly fatal, age-dependent degenerative disorder of motor neurons. Its incidence (0.6 to 2.6 per 100,000 humans) peaks in the sixth decade of life (1). Death from ALS is typically 2 to 5 years after onset and is usually a consequence of respiratory paralysis. Familial cases account for about 10% of ALS. Of these, about 20% are caused by mutations in the superoxide dismutase 1 (SOD1) gene (2). A small number of both familial and apparently sporadic cases are caused by mutations in various other genes, including TAR DNA-binding protein 43 (TDP43) (35), although in the majority of familial ALS (FALS) cases the causative gene is unknown.

In a family of Cape Verdean origin (F577) (fig. S1A), four members developed distinctive ALS with onset in the proximal upper extremities and spreading to the lower extremities, but not the bulbar region. The maternal grandparents of the proband were first cousins; the family originates from a small island of roughly 6000 inhabitants, raising the possibility that the inheritance pattern is recessive. To pursue this, we conducted loss-of-heterozygosity (LOH) mapping and identified a major LOH cluster within a previously reported chromosome 16 ALS locus. Five smaller regions of homozygosity were observed in other chromosomal regions (table S1).

The major LOH cluster spanned approximately 4 Mb and contained 56 candidate ALS genes. Genomic sequencing of these genes revealed a sequence variant in the index ALS cases in exon 15 of the fused in sarcoma/translated in liposarcoma (FUS/TLS) gene (6). This variant, a base pair C1551G missense mutation (7), substituting glutamine for histidine at amino acid residue 517 (Table 1), was heterozygously present in asymptomatic individuals and homozygously present in four individuals with FALS and three individuals who were asymptomatic; one was just entering the age of risk for ALS, whereas the other two were below that age. The variant was not detected in 1446 control DNA samples from North America; a single heterozygote was observed in 66 DNA samples (132 chromosomes) from Cape Verde.

Table 1.

FUS/TLS mutations in ALS cases, with onset and disease duration data. Base numbering begins with the start codon; amino acid numbering begins with the methionine start codon.

View this table:

We next fully sequenced all of the 15 FUS/TLS exons in two families genetically linked to chromosome 16 (F55, fig. S1B, and NUFMS9900, fig. S1C), (8). In family 55 (F55), we detected a missense mutation (C1561G) that substituted glycine for arginine at residue 521; this was coinherited with ALS (in five of five affected members for whom DNA was available but in no unaffected individuals). Incomplete penetrance is evident in F55: two mutation carriers lived past the average age of onset without developing ALS. In family NUFMS9900, a G1553A mutation substituting lysine for arginine at residue 518 was present in 10 of 10 available affected members. These variants were not detected in 1446 control subjects. Haplotype analysis of six single-nucleotide polymorphisms (SNPs) across the FUS/TLS locus does not exclude a common founder among apparently unrelated R521G pedigrees (9).

We sequenced all 15 exons in 81 other unrelated FALS cases and 293 sporadic ALS (SALS) DNA samples, and an additional 209 ALS families were screened for mutations in exon 15. Overall, in 17 different FALS families, 13 different mutations were detected, including 10 in exon 15, two in exon 5, and one in exon 6 (Table 1 and Fig. 1). The most common mutations were R521C and R521G, each present in 3 of the 17 families. No mutations were found in the SALS DNA set. None of the exon 15 variants was observed in 1446 control individuals sequenced. All of the mutated residues in exon 15 of FUS/TLS were highly conserved (fig. S1D); each of the 5 arginines in exon 15 is mutated (Fig. 1).

Fig. 1.

Positions of the FUS/TLS mutations superimposed on the exon and domain organization of the gene. FUS/TLS is encoded by 15 exons that span a genomic region of 11.6 kb. S,Y,Q,G–rich denotes a region rich in the amino acids serine, tyrosine, glutamine, and glycine; G-rich and RGG-rich regions are enriched in either glycine or the arginine-glycine-glycine motif, respectively (11). Further details for each mutation are in Table 1.

Autopsy tissue from a single patient from F55 (R521G mutation) showed loss of motor neurons in the anterior horn of the spinal cord and the hypoglossal nucleus. Myelin pallor in the anterior corticospinal tracts, macrophages surrounding shrunken Betz cells in the motor cortex (neuronophagia), and increased lipofuscin staining in neurons were also noted. The distribution of FUS/TLS was assessed using immunostaining of frozen brain and spinal cord sections from this affected individual. Both control and patient sections revealed immunostaining of FUS/TLS within the nuclei (of neurons and of nonneuronal cells). However, in F55 patient sections there was additional prominent cytoplasmic staining (Fig. 2). Further immunostaining revealed diffuse ubiquitin positivity in nuclei in the patient's tissue but not control tissue (fig. S2), suggesting that one or more nuclear proteins are misfolded.

Fig. 2.

Cytoplasmic retention of mutant FUS/TLS. (A) FUS/TLS immunostaining of sections of frontal cortex from an F55 familial ALS (FALS) patient (four top panels) as compared to a control (four bottom panels). (Left) Merged image of NeuN immunostaining (green) and lipofuscin autofluorescence (red). (Right) FUS/TLS is identified by the brown 3,3′-diaminobenzidine tetrahydrochloride reaction product. Scale bars, 20 μm. (B) Immunostaining of spinal cord from an F55 FALS patient (top row) compared to a control (bottom). Staining patterns for NeuN (red), FUS/TLS (green), and nuclei (blue) are shown individually and in merged images (right) (9). DAPI, 4′,6′-diamidino-2-phenylindole. Scale bars, 10 μm.

Thus, the R521G missense mutation in FUS/TLS led to aberrant trafficking with subsequent cytoplasmic retention of the mutant protein. To evaluate this, we analyzed the subcellular distribution of wild type (WT), R521G (F55), or H517Q (F577) FUS/TLS fused to green fluorescent protein (GFP) 24 hours after transient transfection in N2A and SKNAS cells. With immunofluorescence, the R521G (F55) mutant FUS/TLS–GFP transfections showed dense cytoplasmic staining (Fig. 3A). The ratio of the fraction of cells with exclusively nuclear staining versus cells with combined cytoplasmic and nuclear staining was approximately 0.80/0.20 for the WT protein. In contrast, the ratio for R521G was roughly 0.50/0.50. These ratios were similar in the N2A and SKNAS cells. Immunofluorescence did not detect cytoplasmic retention in cells transfected with the H517Q-GFP (F577) mutant. Subcellular localization of FUS/TLS was additionally studied by compartmental fractionation of SKNAS cells transfected with WT, R521G, or H517Q FUS/TLS–GFP fusion proteins. Immunoblotting of fractions followed by immunostaining with an antibody to GFP demonstrated a substantially higher ratio of soluble cytosolic to soluble nuclear FUS/TLS for both mutants (Fig. 3B). Additionally, a higher ratio of total insoluble to soluble nuclear FUS/TLS protein was also seen for both mutants, although it is more pronounced for the R521G mutant; this reflected both an increase in total insoluble FUS and a decrease in soluble nuclear FUS (fig S3).

Fig. 3.

Mislocalization of mutant FUS/TLS. (A) SKNAS (top) or N2A (bottom) cells transfected with WT or mutant (F55 = R521G or F577 = H517Q) recombinant FUS/TLS–GFP fusion protein, counterstained with DAPI. Top row, merged images; lower row, DAPI only. The percentage of cells observed with nuclear-only (dark bars) and any cytosolic (light bars) FUS/TLS staining is indicated in the bar graphs at right. The error bars indicate SD (from total number of cells counted on three cover-slips). Asterisks indicate statistically significant differences between FUS constructs [P < 0.0001; analysis of variance (ANOVA) followed by Holms test for multiple comparisons]. Scale bars, 10 μm. (B) Cell fractionation studies. SKNAS cells transfected with WT or mutant (F55 = R521G or F577 = H517Q) recombinant FUS/TLS–GFP fusion protein were harvested and fractionated at 24 hours for analysis by immunoblotting using an antibody to GFP, with binding quantified by chemiluminescence. Lamin and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) loading controls are shown below. Densitometric ratios are at right. The error bars indicate SD from three triplicate immunoblot measurements. Asterisks indicate statistically significant differences between WT and mutant FUS constructs (P < 0.0001, ANOVA followed by Holms test for multiple comparisons).

The major defined RNA-interacting domains of FUS/TLS are located in the mid-region of the protein, from amino acids 280 to 370, encoded by exons 9 to 11 (10, 11); sequences of target RNA domains recognized by FUS/TLS have been reported. To show that the FALS-associated FUS/TLS mutations detected here do not alter the RNA-binding domain of FUS, we performed in vitro RNA-binding experiments with recombinant histidine (His)–tagged mutant and WT FUS/TLS proteins and RNA 24-nucleotide oligomer containing GGUG motifs and known to bind FUS/TLS (10). Binding of the RNA oligomers was similar for mutant and WT FUS/TLS protein (fig. S4).

FUS/TLS is a nucleoprotein that functions in DNA and RNA metabolism (1215). It has also been implicated in tumorigenesis (6, 16, 17) and RNA metabolism. FUS/TLS knockout mice show perinatal mortality (18) or male sterility and radiation sensitivity (19). FUS/TLS–deficient neurons show decreased spine arborization with abnormal morphology. In hippocampal neuronal slice cultures, the protein is found in RNA granules that are transported to dendritic spines for local RNA translation in response to metabotropic glutamate receptor (mGluR5) stimulation (20).

We detected 13 FUS/TLS mutations in patients with FALS but none in patients with SALS. We estimate that FUS/TLS mutations are detected in about 5% of FALS; this is comparable to the frequency of TDP43 gene mutations in ALS but less than that for SOD1 (mutated in ∼20% of FALS cases). The FUS/TLS mutations described here led to cytoplasmic retention and apparent aggregation of FUS/TLS. This is reminiscent of several models of the pathogenesis of FALS that are mediated by the aggregation of mutant superoxide dismutase (21) and the mislocalization in ALS of both mutant and WT TDP43 (4, 22). FUS/TLS has also been reported to be a major nuclear aggregate–interacting protein in a model of Huntington's disease (23). Genes implicated in other motor neuron diseases also involve aspects of DNA and RNA metabolism [table S5 in (24)]; understanding the convergent pathophysiologies of these genetic variants will provide insights into new targets for therapies for the motor neuron diseases.

Supporting Online Material

Materials and Methods

Figs. S1 to S4

Table S1


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article