A yeast functional screen predicts new candidate ALS disease genes

Abstract

Amyotrophic lateral sclerosis (ALS) is a devastating and universally fatal neurodegenerative disease. Mutations in two related RNA-binding proteins, TDP-43 and FUS, that harbor prion-like domains, cause some forms of ALS. There are at least 213 human proteins harboring RNA recognition motifs, including FUS and TDP-43, raising the possibility that additional RNA-binding proteins might contribute to ALS pathogenesis. We performed a systematic survey of these proteins to find additional candidates similar to TDP-43 and FUS, followed by bioinformatics to predict prion-like domains in a subset of them. We sequenced one of these genes, TAF15, in patients with ALS and identified missense variants, which were absent in a large number of healthy controls. These disease-associated variants of TAF15 caused formation of cytoplasmic foci when expressed in primary cultures of spinal cord neurons. Very similar to TDP-43 and FUS, TAF15 aggregated in vitro and conferred neurodegeneration in Drosophila, with the ALS-linked variants having a more severe effect than wild type. Immunohistochemistry of postmortem spinal cord tissue revealed mislocalization of TAF15 in motor neurons of patients with ALS. We propose that aggregation-prone RNA-binding proteins might contribute very broadly to ALS pathogenesis and the genes identified in our yeast functional screen, coupled with prion-like domain prediction analysis, now provide a powerful resource to facilitate ALS disease gene discovery.

Fig. 1.

A yeast functional screen identifies human RRM proteins with properties similar to FUS and TDP-43. (A) When expressed in yeast, TDP-43 and FUS form multiple cytoplasmic aggregates (Upper) and are toxic (Lower). (B) We designed a yeast functional screen to identify additional human RRM proteins that aggregate and are toxic in yeast. A library of 133 different human ORFs encoding the proteins was cloned into yeast expression vectors as galactose-inducible YFP fusions. We individually transformed each of these plasmids into yeast cells and assessed their effect on aggregation and toxicity by fluorescence microscopy and spotting assays, respectively. (C) Examples of various localization patterns in yeast cells of human RRM proteins. Some proteins were localized diffusely throughout the cytoplasm (TUT1 and DND1) and others were localized diffusely in the nucleus (DNAJC17). Some formed multiple foci in the nucleus (RBM39) and several others resembled FUS and TDP-43, which formed multiple cytoplasmic foci (EWSR1, TAF15, HNRNPA0, and DAZ1). (D) Spotting assays to assess the toxicity of human RRM proteins. Transformants were grown on synthetic media containing either glucose (control, RRM gene ‘‘off’’) or galactose (to induce expression of candidate ORFs, RRM gene ‘‘on’’). Some proteins were very toxic when overexpressed (DAZ1, HNRNPA0, FUS, and TDP-43) whereas others were moderately toxic (EWSR1 and TAF15) and others were not toxic (PPIE and DNAJC17). See Table 1 and Dataset S1 for toxicity and aggregation scores.

Fig. 2.

Missense mutations in TAF15 in patients with ALS. (A) Comparison of FUS and TAF15 demonstrates similar domain architecture. Both proteins contain a single RRM, a glycine-rich domain, a predicted prion domain, RGG domains, and a C-terminal PY motif, which can function as an NLS (18). Similar domains are also present in TDP-43. The TAF15 variants identified in exons 13–16 in ALS cases are highlighted in red and the variant, R388H, identified in both cases and controls is highlighted in black. (B–E) DNA sequence analysis of TAF15 in North American Caucasian patients with ALS identified three missense variants. (B) A TAF15 mutation in an ALS case: c.1258 G > A, predicted to lead to p.G391E. (C) A TAF15 mutation in an ALS case: c.1308 C > T, predicted to lead to p.R408C. (D) An additional TAF15 variant c.1504G > A, predicted to lead to p.G473E, identified in the ALS cohort from Mayo Clinic. (E) An additional TAF15 variant c.1189T > C, predicted to lead to p.M368T, identified in an ALS cohort from Sweden. (F) Sequence alignment of amino acids 358–416 and 463–483 of TAF15 from diverse vertebrate species indicates that the mutated residues in TAF15 are highly conserved. Identical amino acids have a black background and mutation sites are red.

Fig. 3.

A functional assay to distinguish potentially damaging TAF15 variants from benign variants. We performed an unbiased assessment of all TAF15 missense variants identified from sequencing of ALS cases (M368T, G391E, R408C, and G473E) and controls (R388H). (A) Primary rat embryonic neuron cultures were transfected with myc-tagged WT or mutant TAF15 and stained with α-myc (red). Transfection of WT TAF15 or the R388H variant found in controls results in localization within the nucleus and cytoplasm of neurons in a diffuse pattern. In contrast, the ALS-linked mutant forms of TAF15 showed a striking accumulation of cytoplasmic foci (arrows) in dendrites and axons. (Scale bar, 20 μm.) (B) Quantitation of transfected WT or mutant TAF15 that accumulates in cytoplasmic puncta. Four of four TAF15 variants found in ALS cases (M368T, G391E, R408C, and G473E) showed cytoplasmic puncta formation whereas the one variant found in both cases and controls (R388H) behaved like WT. *P < 0.01 (cytoplasmic puncta formation of TAF15 variants compared with WT, Student's t test). Error bars show mean ± SEM. (C) Because the transfection efficiency in the primary rat spinal cord neuron cultures was not high enough to detect TAF15 overexpression by immunoblot, we transfected the same constructs of WT and TAF15 mutants into HEK293T cells. Immunoblotting with α-myc antibodies was used to detect myc-tagged TAF15 and α-actin antibody was used as a loading control. In contrast to the primary neuronal cultures, we did not observe a significant difference in aggregation between WT TAF15 and the variants in HEK293T cells.

Fig. 4.

TAF15 is an aggregation-prone protein like TDP-43 and FUS. (A) Following TEV protease cleavage to remove the N-terminal GST tag, FUS, TAF15, and TDP-43 proteins were processed for SDS-PAGE and Coomassie stained to confirm purity and expected molecular weight. (B) GST-TDP-43, GST-FUS, or GST-TAF15 (3 μM) were incubated in the presence or absence of TEV protease at 25 °C for 0–90 min with agitation. Note that very little aggregation occurs in the absence of TEV protease. The extent of aggregation was determined by turbidity. Values represent means ± SEM (n = 3). (C) GST-TDP-43, GST-FUS, or GST-TAF15 (3 μM) were incubated in the presence of TEV protease at 25 °C for 0–60 min. At the indicated times, reactions were processed for sedimentation analysis. Pellet and supernatant fractions were resolved by SDS-PAGE and stained with Coomassie Brilliant Blue. The amount of protein in the pellet fraction was determined by densitometry in comparison with known quantities of the appropriate protein. Values represent means ± SEM (n = 3). A human RRM protein, DND1, which did not aggregate and was not toxic in yeast (Fig. 1 C and D), was also soluble and did not form aggregates in this assay. (D) GST-TDP-43, GST-FUS, or GST-TAF15 (3 μM) were incubated in the presence of TEV protease at 25 °C for 0–60 min. At various times, reactions were processed for EM. Small arrows denote small pore-shaped oligomers and large arrows denote linear polymers. (Scale bar, 500 nm.) (E) Gallery of TDP-43, FUS, and TAF15 oligomers formed during aggregation reactions. (Scale bar, 50 nm.) (F) Following TEV protease cleavage to remove the N-terminal GST tag, TAF15 wild-type (WT), G391E, and R408C proteins were processed for SDS-PAGE and Coomassie stained to confirm purity and expected molecular weight. (G) ALS-linked TAF15 variants G391E and R408C displayed accelerated aggregation kinetics whereas the R388H variant found in both cases and controls aggregated with similar kinetics to WT TAF15.

Fig. 5.

TAF15 confers neurodegeneration and dysfunction in Drosophila. (A) Toxicity of RRM proteins in the eye. TAF15 causes degeneration and disruption of the retinal structure, akin to TDP-43 (also see ref. 24). Control is driver line alone, gmr-GAL4/+. TDP-43 is gmr-GAL4/UAS-TDP-43-YFP. TAF15 is gmr-GAL4/UAS-TAF15 (grown at 29 °C). (B) Progressive loss of climbing behavior upon expression of TAF15 selectively in motor neurons using the motor neuron-specific D42-GAL4 driver. (C) Up-regulation of two other RRM proteins does not cause neurodegeneration in Drosophila. As a specificity control for the neurodegenerative phenotype conferred by up-regulation of TDP-43 and TAF15 in Drosophila, we tested the effects of up-regulating the fly homologs of two other human RRM proteins in the eye. Up-regulation of tsu and CG17187, which are Drosophila homologs of human genes RBM8A and DNAJC17, respectively, did not cause neurodegeneration. (D) Expression of TAF15 in the nervous system reduces lifespan (blue, compared with normal in purple). Up-regulation of TAF15 variants G391E and R408C causes more rapid death (red and green, compared with WT TAF15 in blue). (E) Immunoblot showing TAF15 WT and mutant expression levels in transgenic flies. β-Tubulin levels were used as loading control.

Fig. 6.

TAF15 accumulates in the cytoplasm in some sporadic spinal cord neurons of patients with ALS. (A) Immunohistochemistry for TAF15 in a postmortem control spinal cord neuron indicates nuclear localization, whereas, in addition to nuclear localization, TAF15 accumulated in cytoplasmic puncta of spinal cord neurons of patients with ALS (B–H). (Scale bar, 25 μm.)