Self-assembly of FUS through its low-complexity domain contributes to neurodegeneration

Abstract

Aggregation of fused in sarcoma (FUS) protein, and mutations in FUS gene, are causative to a range of neurodegenerative disorders including amyotrophic lateral sclerosis (ALS) and frontotemporal dementia. To gain insights into the molecular mechanism whereby FUS causes neurodegeneration, we generated transgenic Drosophila melanogaster overexpressing human FUS in the photoreceptor neurons, which exhibited mild retinal degeneration. Expression of familial ALS-mutant FUS aggravated the degeneration, which was associated with an increase in cytoplasmic localization of FUS. A carboxy-terminally truncated R495X mutant FUS also was localized in cytoplasm, whereas the degenerative phenotype was diminished. Double expression of R495X and wild-type FUS dramatically exacerbated degeneration, sequestrating wild-type FUS into cytoplasmic aggregates. Notably, replacement of all tyrosine residues within the low-complexity domain, which abolished self-assembly of FUS, completely eliminated the degenerative phenotypes. Taken together, we propose that self-assembly of FUS through its low-complexity domain contributes to FUS-induced neurodegeneration.

Introduction

Neurodegenerative disorders are pathologically characterized by intracellular accumulation of fibrillar inclusions, which are composed of disease-relevant proteins, e.g. tau and α-synuclein (1). Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disorder that affects upper and lower motor neurons, resulting in muscular atrophy and weakness. Fused in sarcoma/translated in liposarcoma (FUS) gene has been identified in the pedigrees with autosomal dominantly inherited familial ALS (fALS) linked to ALS type 6 (2,3). More than 50 point or truncation mutations in FUS gene have been reported in fALS (4,5). FUS-immunoreactive cytoplasmic inclusions in neurons are the pathological hallmark of FUS-linked ALS (2,3), frontotemporal lobar degeneration (FTLD) (6,7), neuronal intermediate filament inclusion disease (8) and basophilic inclusion body disease (9), collectively referred to as FUS proteinopathies (5). However, it remains unclear how FUS-positive inclusions are formed in the neuronal cytoplasm and cause neurodegeneration in FUS proteinopathies.

FUS is a ubiquitously expressed RNA-binding protein that belongs to the FET (FUS, Ewing's sarcom (EWS), TATA-binding protein associated factor 15 (TAF15) protein family characterized by the presence of an RNA-recognition motif (RRM), a zinc finger domain, nuclear export signal (NES) and a proline-tyrosine nuclear localization signal (PY-NLS) (10). A series of studies have revealed that FUS is involved in multiple steps of RNA processing, e.g. splicing and transcription (11–13). FUS deficiency in mice resulted in perinatal or early death before adulthood in the inbred background (14,15). FUS deficient mice in the outbred background grew into adulthood and exhibited vacuolation in the neuropil of hippocampus, hyperactivity and reduction in anxiety-like behavior, but not FTLD- or ALS-like phenotypes (16). These results indicate that loss of function of FUS may not be a major mechanism underlying the neurodegeneration in FUS proteinopathies.

The majority of mutations in FUS gene linked to fALS are located in the PY-NLS domain and disrupt the nuclear import of FUS mediated by Transportin (17). In vivo overexpression of fALS mutant FUS in rodents causes cytoplasmic accumulation of FUS and progressive degeneration of motor neurons (18–21). Overexpression of fALS mutant FUS in Drosophila melanogaster also exhibited neurodegeneration of photoreceptor neurons, whereas deletion of its NES domain suppressed the degeneration (22). These observations suggest that fALS mutations in FUS gene may cause neurodegeneration through cytoplasmic mislocalization of FUS. However, the molecular mechanism whereby cytoplasmic FUS causes neurodegeneration has remained unknown.

Here, we generated lines of transgenic D.melanogaster (TG fly) overexpressing human FUS in the retinal photoreceptor neurons and found that fALS-linked mutant FUS TG flies exhibited severe retinal degeneration compared with those expressing wild-type (wt) FUS through an increase in the level of cytoplasmic FUS. We also found that the carboxy-terminal truncation mutations exacerbated the FUS-induced degenerative phenotypes by sequestrating wt FUS into cytoplasmic aggregates. Notably, overexpression of FUS with mutations at the low-complexity (LC) domain that was incapable of self-assembly, did not elicit any degenerative phenotypes. These results strongly suggest that the self-assembly of FUS contributes to the neurodegeneration induced by FUS.

Results

FALS mutations in FUS exacerbated FUS-induced retinal degeneration by increasing cytoplasmic localization of FUS

To elucidate the pathomechanism underlying FUS proteinopathies, we established D.melanogaster overexpressing human wt (FUS wt), or P525L fALS mutant FUS located within the carboxy-terminal PY-NLS (FUS P525L) (Fig. 1A), using the GAL4-UAS system with random insertion method, and crossed them with the gmr-GAL4 line to overexpress proteins specifically in the retinal cells including photoreceptor neurons. By the semi-quantification of protein levels of FUS wt in the photoreceptor neurons of FUS wt TG flies, the expression level of FUS wt in one photoreceptor neuron was estimated to be ∼26.3 pg (Supplementary Material, Fig. S1A). If the volume of Drosophila photoreceptor neurons were comparable to those of average-sized neurons, the concentration of FUS wt was estimated to be ∼40–260 μm. Histopathological observation of the retina of FUS wt TG flies revealed mild but discrete degeneration with thinning of retina, disturbance in the ommatidial alignment and vacuolization (Fig. 1B and C). Overexpression of FUS P525L induced more severe retinal degeneration compared with FUS wt (Fig. 1B and C). Immunoblot analyses of TG flies showed that FUS P525L was expressed at similar levels to FUS wt (Fig. 1D). PY-NLS in FUS is required for the nuclear localization through recognition by Transportin, which mediates protein nuclear import (23,24). To determine whether P525L mutation alters the subcellular localization of FUS in the Drosophila retinal cells, we immunostained retina of TG flies and found that FUS P525L was localized not only in the nuclei, but also in the cytoplasm as an aggregate-like structure, whereas FUS wt was predominantly localized in the nuclei (Fig. 1E). These results raised the possibility that P525L mutation exacerbated FUS-induced retinal degeneration through an increase in the cytoplasmic localization of FUS.

Figure 1.

 FALS mutations in FUS exacerbated FUS-induced retinal degeneration through an increase in the level of cytoplasmic localization of FUS. (A) Schematic structure of human FUS protein. FUS harbors QGSY rich region, Gly-rich region (G rich), an RRM, a Zinc finger domain (Zn) or a proline-tyrosine nuclear localization signal (PY-NLS). P525L mutation (red) is located within PY-NLS. (B) H&E stained sections of compound eyes of 5-day-old gmr-driven LacZ, FUS wt or P525L TG flies. Scale bar = 50 μm. (C) Quantification of retinal thickness of each TG fly. FUS P525L TG flies exhibited significantly severe retinal degeneration compared with FUS wt TG flies (44.3 ± 1.2 μm in FUS wt and 13.5 ± 0.72 μm in FUS P525L), n = 10, mean ± S.E. (***); P < 0.001. (D) Comparison of expression levels of FUS (upper panel) or α-tubulin (bottom panel) as a loading control. Immunoblotting analyses of 5-day-old gmr-driven LacZ, FUS wt or FUS P525L TG flies. Relative expression levels of FUS are indicated under the panels (n = 3). (E) Immunofluorescence labeling of retinal cells of 5-day-old gmr-driven LacZ, FUS wt or P525L TG flies using anti-FUS antibody (400–450) (green) or DRAQ-5 (magenta, nuclear marker). Scale bar = 5 μm.

It has been reported that the RNA-binding ability of FUS may regulate FUS-induced neurodegeneration in the Drosophila model (25). To examine whether fALS mutations of FUS change the RNA-binding ability, we carried out an electrophoresis mobility shift assay (EMSA) to detect the binding of FUS to specific RNAs (11). We found that recombinant FUS P525L protein bound to synthesized RNA in a concentration dependent manner at a similar extent to FUS wt protein (Supplementary Material, Fig. S1D–F), suggesting that P525L mutation does not change the RNA-binding ability. To confirm the specificity of interaction of FUS with RNA, we also examined the FUS lacking the RNA-recognition motif (FUS ΔRRM, amino acid residues 286–368 of human FUS deleted) and found that FUS ΔRRM showed decreased binding to specific RNA compared with GST-FUS wt (Supplementary Material, Fig. S1G and H). The reason why GST-FUS ΔRRM retained partial interaction with specific RNA is unclear, but it is possible that another RNA-recognition motif, i.e, residues 449–518 of human FUS (26), might have worked as a supplementary RNA-binding region in the GST-FUS ΔRRM protein. We have also performed EMSA using a negative control RNA, which is expected to show no binding to FUS (CCUC, ref. 11). This RNA never interacted with GST-FUS wt, suggesting that GGUG sequence is necessary for the interaction with FUS (Supplementary Material, Fig. S1H).

Double expression of carboxy-terminally truncated FUS and FUS wt, but not carboxy-terminally truncated FUS alone, elicited severe retinal degeneration

To investigate the effects of PY-NLS in FUS, we generated TG fly lines expressing carboxy (C)-terminally truncated mutant FUS lacking the 13-amino-acid (FUS ΔC) or 32-amino-acid (FUS R495X) residues in PY-NLS (Fig. 2A). R495X truncation mutation has been linked to fALS with early disease onset and rapid disease progression (27–29). Unexpectedly, FUS ΔC or R495X TG flies exhibited very mild retinal degeneration, and the thickness of the retina of FUS ΔC or R495X TG flies was larger than that of FUS wt TG flies (Fig. 2B and D). Immunoblot analyses of the lysates of heads of TG flies using an antibody against amino acid residues 400–450 of FUS [anti-FUS(400–450); Fig. 2A] showed that the expression levels of FUS ΔC and R495X was ∼2.2 and ∼3.1 times higher than that of FUS wt, respectively (Fig. 2F). Another anti-FUS antibody, which recognizes the C terminus of FUS [anti-FUS(500–526); Fig. 2A], did not react with any polypeptides in the lysates of heads of FUS ΔC or R495X TG flies (Fig. 2F). These data suggest that the intactness of the C terminus of FUS is necessary for the degeneration induced by FUS. Surprisingly, cross of FUS ΔC or R495X TG flies with FUS wt TG flies caused severe retinal degeneration and reduction in the thickness of retina compared with FUS wt, ΔC or R495X single TG flies (Fig. 2C and D). To elucidate the mechanism whereby FUS wt/FUS ΔC or R495X double TG flies exhibited severe retinal degeneration, we performed immunofluorescence labeling of the retina of TG flies. We found that FUS wt, which is labeled by anti-FUS(500–526) or anti-FUS(52–400), was predominantly localized at the nuclei in FUS wt TG flies, whereas FUS ΔC or R495X, which was positive only for anti-FUS(52–400), was predominantly localized at the cytoplasmic aggregates (Fig. 2G). In sharp contrast, FUS wt, which is specifically immunolabeled by anti-FUS(500–526), was localized not only in the nuclei, but also in the cytoplasmic aggregates in the retinal cells of FUS wt/FUS ΔC or R495X double TG flies (Fig. 2G). These observations suggested that FUS wt was sequestered from nuclei into cytoplasmic aggregates by co-expression with FUS ΔC or R495X, which caused the severe degeneration of the retina of FUS wt/FUS ΔC or R495X double TG flies. To compare the degenerative phenotypes with the same transgene number, we also generated FUS wt/FUS wt or FUS wt/FUS P525L double TG flies, both of which exhibited extremely severe degenerative phenotypes in the retina compared with FUS wt single TG flies (Fig. 2C, Supplementary Material, Fig. S1B and C). Immunoblot analysis using anti-FUS(400–450) showed that the expression level of FUS in the heads of FUS wt/FUS wt double TG flies was at a similar level to that of FUS wt single TG flies (Fig. 2E). Immunoblot analyses also showed that FUS wt/FUS ΔC or R495X double TG flies were comparable to that of FUS wt single TG flies (Fig. 2F). The reason why the expression levels of FUS wt, ΔC or R495X were decreased in the double TG flies compared with FUS wt, ΔC or R495X single TG flies is unclear; one possibility would be that FUS negatively regulated its own expression in the retina of double TG flies through the interaction with its exon 7 of FUS mRNA in similar manner to that documented in HeLa cells (30). We also found that anti-FUS (52–400) and anti-FUS (400–450) antibodies revealed a band of smaller size corresponding to C-terminally truncated FUS mutant, in addition to a larger sized faint band corresponding to FUS wt (Fig. 2F). We ensured that FUS ΔC retained the RNA-binding ability in vitro by EMSA (Supplementary Material, Fig. S1D and F).

Figure 2.

Double expression of C-terminally truncated FUS and FUS wt exhibited severe degeneration in the retina of Drosophila. (A) Schematic structures of C-terminal truncation mutant FUS. The epitopes of anti-FUS antibodies are indicated below. (B) H&E stained sections of compound eyes of 5-day-old gmr-driven FUS wt, ΔC or R495X TG flies. Scale bar = 50 μm. (C) H&E stained sections of compound eyes of 5-day-old gmr-driven FUS wt/FUS wt, FUS wt/FUS ΔC or FUS wt/R495X double TG flies. Scale bar = 50 μm. (D) Quantification of retinal thickness of each TG fly. FUS ΔC or R495X TG flies exhibited significant thicker retinal thickness than FUS wt TG flies (64.0 ± 1.2 μm in FUS ΔC, 62.5 ± 1.5 μm in FUS R495X) and FUS wt/FUS wt, FUS wt/FUS ΔC or FUS wt/R495X double TG flies exhibited significant thinner retinal thickness than FUS wt TG flies (14.7 ± 1.3 μm in FUS wt/FUS wt, 21.6 ± 1.2 μm in FUS wt/FUS ΔC, 16.0 ± 0.91 μm in FUS wt/FUS R495X), n = 10, mean ± S.E. (*); P < 0.05, (**); P < 0.01, (***); P < 0.001. (E) Comparison of expression levels of FUS and α-tubulin. Immunoblot analyses of 5-day-old gmr-driven LacZ, FUS wt, or FUS wt/FUS wt TG flies using anti-FUS(400-450) (upper panel) or anti-α-tubulin antibody (bottom panel). Relative expression levels of FUS by anti-FUS (400–450) are indicated under the panels (n = 3). (F) Comparison of expression levels of FUS (upper three panels) or α-tubulin (bottom panel). Immunoblotting analyses of 5-day-old gmr-driven LacZ, FUS wt, ΔC, R495X, FUS wt/FUS ΔC, or FUS wt/FUS R495X TG flies using anti-FUS(400–450) (first upper panel), anti-FUS(500–526) (second upper panel), anti-FUS(52–400) (third upper panel), or anti-α tubulin antibody (bottom panel). Relative expression levels of FUS by anti-FUS(400–450) are indicated under the panels (n = 3). (G) Immunofluorescence labeling of retinal cells of 5-day-old gmr-driven FUS wt, ΔC, R495X, FUS wt/FUS ΔC, or FUS wt/FUS R495X TG flies using anti-FUS(500–526) (green), anti-FUS(52–400) (red), or DRAQ-5 (blue) as a nuclear marker. Scale bar = 5 μm.

To examine whether Drosophila Cabeza (Caz), a homologous molecule to human FUS, was also localized to the cytoplasmic aggregates, we have immunohistochemically studied the retina of LacZ, FUS wt or FUS ΔC TG flies with an anti-Caz antibody, and found that Caz was mostly localized in the nuclei in FUS wt TG flies, and the subcellular localization was not relocated to cytoplasmic aggregates in the FUS ΔC TG flies (Supplementary Material, Fig. S2).

C-terminally truncated mutant FUS sequestered wt FUS proteins from nuclei into cytoplasmic aggregates in HEK293 cells

To further investigate the subcellular distribution of FUS proteins, we examined the effects of C-terminal truncation mutations of FUS in mammalian cells. By immunofluorescence labeling, we found that endogenous human FUS, amino (N)-terminally FLAG-tagged FUS wt (FLAG-FUS wt) or N-terminally 3xMyc-tagged FUS wt (3xMyc-FUS wt) was predominantly localized within the nuclei in HEK293 cells (Fig. 3A, C and E), in a similar pattern to FUS wt in the retinal cells of TG flies (Fig. 1E). In contrast, FLAG-FUS P525L was localized at cytoplasmic aggregates of ∼1 μm in diameter in HEK293 cells (Supplementary Material, Fig. S3), which are positive for T cell intracellular antigen-1-related protein (TIAR) characteristic of stress granules (17,31,32) (Supplementary Material, Fig. S3). Next, we found that FLAG-tagged FUS ΔC (FLAG-FUS ΔC) was predominantly localized at cytoplasm (Fig. 3A), and occasionally formed TIAR-positive cytoplasmic aggregates in HEK293 cells (Supplementary Material, Fig. S3). Notably, endogenous FUS, which is specifically immunolabeled by anti-FUS(500–526), was localized not only in the nuclei, but also in the cytoplasmic aggregates in HEK293 cells expressing FLAG-FUS ΔC (Fig. 3A). These observations suggest that endogenous FUS was relocalized from nuclei to be sequestered into cytoplasmic aggregates by co-expression of FLAG-FUS ΔC. To test this idea, we co-transfected FUS proteins with two different epitope-tags into HEK293 cells. 3xMyc tagged FUS ΔC or R495X (3xMyc-FUS ΔC or R495X) was also localized at the cytoplasmic aggregates in HEK293 cells, together with FLAG-FUS wt (Fig. 3C). In contrast, FLAG-FUS wt was localized not only in the nuclei, but also in the cytoplasmic aggregates in HEK293 cells doubly expressing FLAG-FUS wt and 3xMyc-FUS ΔC, or R495X (Fig. 3C). Upon immunoblotting, anti-FUS(400–450) antibody revealed comparable levels of total FUS protein in HEK293 cells expressing FLAG-FUS wt to that in FLAG-FUS wt and 3xMyc-FUS ΔC or 3xMyc-FUS R495X doubly transfected cells (Fig. 3B). Moreover, co-expression of 3xMyc-FUS wt and FLAG-FUS ΔC or FLAG-FUS P525L in HEK293 cells also caused coalescence of 3xMyc-FUS wt at the cytoplasmic aggregates (Fig. 3E). Upon immunoblotting, anti-FUS(400–450) antibody revealed comparable levels of total FUS protein in HEK293 cells expressing 3xMyc-FUS wt to that in 3xMyc-FUS wt and FLAG-FUS ΔC doubly transfected cells (Fig. 3D). These data strongly supported the view that FUS wt in the nuclei was sequestered into cytoplasmic aggregates by the co-expression of C-terminally truncated mutant FUS.

Figure 3.

C-terminally truncated mutant FUS sequestered wt FUS protein from nuclei into cytoplasmic aggregates. (A) Immunofluorescence labeling of non-transfected HEK293 cells or HEK293 cells overexpressing FLAG-FUS ΔC using anti-FUS(500-526) (red), anti-FLAG antibody (green), or DRAQ-5 (blue) as a nuclear marker. Scale bar = 10 μm. Top-right insets are enlarged views of boxed regions. (B) Immunoblotting analyses of non-transfected HEK293 cells, or HEK293 cells overexpressing FLAG-FUS wt, 3xMyc-FUS ΔC, 3xMyc-FUS R495X, FLAG-FUS wt/3xMyc-FUS ΔC, or FLAG-FUS wt/3xMyc-FUS R495X using anti-FLAG antibody (first upper panel), anti-Myc antibody (second upper panel), anti-FUS(400–450) (third upper panel) or α-tubulin (bottom panel). (C) Immunoblotting analyses of non-transfected HEK293 cells, or HEK293 cells overexpressing 3xMyc-FUS wt, FLAG-FUS ΔC, or 3xMyc-FUS wt/FLAG-FUS ΔC using anti-Myc antibody (first upper panel), anti-FLAG antibody (second upper panel), anti-FUS(400–450) (third upper panel) or α-tubulin (bottom panel). (D) Immunofluorescence labeling of HEK293 cells overexpressing FLAG-FUS wt, 3xMyc-FUS ΔC, 3xMyc-FUS R495X, FLAG-FUS wt/3xMyc-FUS ΔC, or FLAG-FUS wt/3xMyc-FUS R495X using anti-FLAG antibody (green), anti-Myc antibody (red), or DRAQ-5 (blue). Scale bar = 10 μm. Top-right insets are enlarged views of boxed regions. (E) Immunofluorescence labeling of HEK293 cells overexpressing 3xMyc-FUS wt, FLAG-FUS ΔC, 3xMyc-FUS wt/FLAG-FUS ΔC, or 3xMyc-FUS wt/FLAG-FUS P525L using anti-FLAG antibody (green), anti-Myc antibody (red), or DRAQ-5 (blue). Scale bar = 5 μm.

The self-assembly of FUS through its LC domain is required for the retinal degeneration induced by FUS

Based on the finding that C-terminally truncated mutant FUS has sequestered FUS wt into cytoplasmic aggregates in HEK293 cells, we hypothesized that the self-assembly of FUS in the cytoplasm may be crucial for the neurodegeneration induced by FUS. To examine whether FUS is self-assembled in the cytoplasm, we utilized the bi-molecular fluorescence/luminescence complementation (BiFC/BiLC) technique. We first developed BiFC assay with N- or C-terminal fragment of Venus (V1 or V2), a yellow fluorescent protein, fused to the N terminus of FUS. When V1-FUS and V2-FUS interacted with each other, V1 and V2 fragments would reconstitute a functional Venus protein and exhibit fluorescence (Fig. 4A). As a negative control, we also generated N- or C-terminal fragment of Venus without FUS (V1-only, V2-only). V1-only was mostly localized in the cytoplasm (Fig. 4G), and V1-only did not interact with V2-FUS, emitting no fluorescence (Fig. 4A). We found that single expression of full-length Venus-tagged FUS wt (Venus-FUS) exhibited intense fluorescence in the nuclei of HEK293 cells (Fig. 4B), whereas single expression of V1-FUS or V2-FUS did not (data not shown). Moreover, yellow fluorescence was found in the nuclei of cells co-expressing V1-FUS and V2-FUS, whereas only faint fluorescence was detected in the nuclei of cells co-expressing V1-only and V2-FUS (Fig. 4B). The relative fluorescence intensity in HEK293 cells co-expressing V1-FUS and V2-FUS was significantly higher (∼4.2-fold) than that in V1-only and V2-FUS co-expressing cells (Fig. 4C). Co-expression of V1-FUS P525L and V2-FUS P525L showed yellow fluorescence in the cytoplasm of HEK293 cells, in an identical manner to full-length Venus-tagged FUS P525L (Venus-FUS P525L) (Fig. 4D). The relative fluorescence intensity in HEK293 cells co-expressing V1-FUS P525L and V2-FUS P525L was significantly higher (∼2.7-fold) than that in cells co-expressing V1-only and V2-FUS P525L (Fig. 4E). Immunoblot analyses showed equal levels of BiFC-tagged FUS proteins expressed in HEK293 cells (Fig. 4F). To quantitatively assess the self-assembly of FUS, we next used a NanoBiT BiLC assay. In the NanoBiT system, the reversible interaction between LgBiT: SmBiT allows for the detection of dose-dependent protein interaction. We fused LgBiT- or SmBiT- to the N terminus of FUS (Fig. 4H), and found that co-expression of LgBiT-FUS and SmBiT FUS in HEK293 cells exhibited the luminescence, whereas single expression of LgBiT-FUS or SmBiT FUS did not elicit the luminescence (Fig. 4I). These data strongly suggested that fragments of FUS were able to self-assemble. Notably, double expression of LgBiT-FUS wt and SmBiT-FUS R495X also activated the luminescence (Fig. 4I). This result supported the notion that FUS R495X sequestered FUS wt from nuclei to cytoplasm and assembled with each other.

Figure 4.

Self-assembly of FUS wt and FUS P525L in the nuclei and cytoplasm, respectively, in HEK293 cells. (A) Schematic diagram of BiFC assay to visualize the self-assembly of FUS. If V1-FUS and V2-FUS are assembled together, V1 and V2 fragments would reconstitute a functional Venus protein and exhibit fluorescence. (B) Fluorescence imaging of HEK293 cells overexpressing Venus-FUS wt, V1-FUS wt/V2-FUS wt, or V1-only/V2-FUS wt. DRAQ-5 (blue) is a nuclear marker. Scale bar = 10 μm. (C) Relative fluorescence intensity of Venus protein in (B) (n = 50–52), (***); P < 0.001. (D) Fluorescence imaging of HEK293 cells overexpressing Venus-FUS P525L, V1-FUS P525L/V2-FUS P525L or V1-only/V2-FUS P525L. DRAQ-5 (blue) is a nuclear marker. Scale bar = 10 μm. (E) Relative fluorescence intensity of Venus protein in (D) (n = 26–40), (***); P < 0.001. (F) Immunoblotting analyses of non-transfected HEK293 cells, or HEK293 cells overexpressing Venus, V1-only, V2-only, V1-only/V2-only, Venus-FUS wt, V1-FUS wt, V2-FUS wt, V1-FUS wt/V2-FUS wt, V1-only/V2-FUS wt, Venus-FUS P525L, V1-FUS P525L, V2-FUS P525L, V1-FUS P525L/V2-FUS P525L or V1-only/V2-FUS P525L using anti-FUS(400–450) (upper panel), anti-GFP antibody (middle panel) or anti-α-tubulin (bottom panel). (G) Immunofluorescence imaging of HEK293 cells overexpressing V1-only using anti-GFP amino-terminus. DRAQ-5 (blue) was used as a nuclear marker. Scale bar = 10 μm. (H) Schematic diagram of BiLC assay to measure the self-assembly of FUS. If LgBiT-FUS and SmBiT-FUS were assembled together, LgBiT and SmBiT fragments would reconstitute a functional NanoBiT protein and exhibit luminescence. (I) Relative luminescence of HEK293 cells transiently expressing LgBiT-FUS, SmBiT-FUS, LgBiT-FUS/SmBiT-FUS, LgBiT-FUS P525L/SmBiT-FUS P525L, LgBiT-allS FUS/SmBiT-alls FUS or LgBiT-FUS/SmBiT-FUS R495X (n = 5), (***); P < 0.001.

To further examine whether the self-assembly of FUS is required for the FUS-induced neurodegeneration, we focused on the N-terminal glutamine-glycine-serine-tyrosine (QGSY) rich and Gly-rich regions. These regions were recently recognized as a low-complexity (LC) sequence domain, which has been implicated in the self-assembly of FUS; allS mutant FUS (allS-FUS), in which 27 tyrosine residues in the LC domain were replaced with serine, was incapable of the self-assembly through the LC domain (33). Upon the BiLC experiment, we found that allS mutation significantly reduced the ability of the self-assembly of FUS (Fig. 4I). We generated TG fly lines expressing allS-FUS wt or allS-FUS P525L (Fig. 5A). AllS-FUS wt or P525L TG flies exhibited no degenerative phenotypes with normal thickness of the retina (Fig. 5B and C). Immunoblot analyses showed that allS mutant FUS migrated slower than FUS wt, and that the expression levels of allS-FUS wt and P525L was ∼0.38 and ∼1.6 times that of FUS wt, respectively (Fig. 5D). Immunofluorescence analysis revealed that allS-FUS wt was predominantly localized at nuclei, whereas allS-FUS P525L was localized in the cytoplasm (Fig. 5E). Moreover, we found that allS-FUS wt retained the RNA-binding ability in vitro by EMSA (Supplementary Material, Fig. S1D and F). Taken together, although allS-FUS P525L was predominantly localized in cytoplasm, the allS mutation abolished the retinal degeneration of TG flies, suggesting that the self-assembly of FUS through the LC domain is indispensable for the neurodegeneration induced by FUS.

Figure 5.

AllS mutation of FUS abolished FUS-induced retinal degeneration. (A) Schematic structures of allS mutant FUS. (B) H&E stained sections of compound eyes of 5-day-old gmr-driven allS-FUS wt or P525L TG flies. Scale bar = 50 µm. (C) Quantification of retinal thickness of each TG fly. allS-FUS wt or P525L TG flies exhibited no degenerative phenotype with normal thickness of the retina (70.5 ± 0.75 μm in allS-FUS wt, 73.1 ± 1.0 μm in allS-FUS P525L) (n = 10), mean ± S.E. (***); P < 0.001. (D) Comparison of expression levels of FUS (upper panel) or α-tubulin (bottom panel). Immunoblotting analyses of 5-day-old gmr-driven LacZ, FUS wt, allS-FUS wt, FUS P525L or allS-FUS P525L TG flies. Relative expression levels of FUS are indicated under the panels (n = 3). (E) Immunofluorescence labeling of retinal cells of 5-day-old gmr-driven allS-FUS wt or allS-FUS P525L TG flies using anti-FUS(400–450) (green) or DRAQ-5 (magenta). Scale bar = 5 μm.

To further investigate the role of LC domain of FUS in neurodegeneration, we also examined the mutant FUS lacking the QGSY rich region, or both the QGSY rich and Gly rich regions (FUS ΔQGSY, or FUS ΔQGSY-G, Supplementary Material, Fig. S4A). Histopathological observation of the retina showed no degenerative phenotypes in FUS ΔQGSY or ΔQGSY-G TG flies (Supplementary Material, Fig. S4B). Immunoblot analyses showed that the expression levels of FUS ΔQGSY and ΔQGSY-G was 0.7 and 1.7 times that of FUS wt, respectively (Supplementary Material, Fig. S4C). These data suggested that the QGSY rich region of FUS is required for the FUS-induced neurodegeneration.

Previous biochemical analyses of FUS proteins extracted from post-mortem human brains have shown that the distribution of FUS was shifted toward the detergent-insoluble fractions in the brains of patients with FTLD-FUS compared with those in FTLD-TDP or control brains (34). To examine whether the self-assembly of FUS through its LC domain causes the insolubility of FUS, we examined the solubility of wt or mutant FUS proteins expressed in HEK293 cells. The ratios of insoluble FLAG-FUS wt and P525L were ∼10.1 and 7.2%, respectively, whereas insoluble FLAG-allS-FUS wt or P525L proteins were hardly detected on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (Supplementary Material, Fig. S4D and E). These data indicate that the self-assembly of FUS through its LC domain is linked to the acquisition of insolubility of FUS protein.

The self-assembly of FUS is necessary for the sequestration of FUS wt from nuclei into cytoplasmic aggregates

Double expression of C-terminally truncated mutant FUS with FUS wt caused severe retinal degeneration through the sequestration of FUS wt from nuclei into cytoplasmic aggregates (Fig. 2). To investigate whether the self-assembly of FUS through its LC domain is required for this sequestration, we generated TG flies doubly expressing allS-FUS and FUS ΔC in the retina (allS-FUS × FUS ΔC). Histopathological analyses revealed very mild retinal degeneration and thinning of the retina in allS-FUS × FUS ΔC double TG flies compared with LacZ TG flies (Fig. 6A and B), which was in sharp contrast with the severe degeneration in FUS wt × FUS ΔC double TG flies (Fig. 2). Immunoblot analyses using three different antibodies showed that anti-FUS(400–450) recognized both allS-FUS and FUS ΔC, whereas anti-FUS(500–526) exclusively labeled allS-FUS, and anti-FUS(52–400) exclusively labeled FUS ΔC in allS-FUS × FUS ΔC double TG flies (Fig. 6C). We found that allS-FUS × FUS ΔC double TG flies exhibited a decrease in the expression levels of allS-FUS and FUS ΔC, compared with allS-FUS or FUS ΔC single TG flies (Fig. 6C). To examine whether allS-FUS was sequestered into cytoplasmic aggregates in allS-FUS × FUS ΔC double TG flies, we carried out immunofluorescence labeling of allS-FUS with anti-FUS(500–526) and FUS ΔC with anti-FUS(52–400), and found that allS-FUS was predominantly observed in the nuclei and was hardly detected in the cytoplasmic aggregates, whereas FUS ΔC was observed in the cytoplasmic aggregates in allS-FUS × FUS ΔC double TG flies (Fig. 6D). These data suggest that the failure of sequestration of allS-FUS by FUS ΔC from nuclei into cytoplasmic aggregates might have caused the lack of exacerbation of retinal degeneration in allS-FUS × FUS ΔC double TG flies.

Figure 6.

 C-terminally truncated mutant FUS did not sequester allS mutant FUS protein from nucleus into cytoplasm. (A) H&E stained sections of compound eyes of 5-day-old gmr-driven allS-FUS wt/FUS ΔC double TG flies. Scale bar = 50 µm. (B) Quantification of retinal thickness of each TG fly. allS-FUS/FUS ΔC double TG fly exhibited mild retinal degeneration as well as FUS ΔC TG flies (63.0 ± 1.8 μm in allS-FUS/FUS ΔC) (n = 10), mean ± S.E. (*); P < 0.05. (**); P < 0.01. n.s. means no significant difference. (C) Immunoblotting analyses of 5-day-old gmr-driven LacZ, allS-FUS wt, FUS ΔC, or allS-FUS wt/FUS ΔC TG flies using anti-FUS (400–450) (first upper panel), anti-FUS (500–526) (second upper panel), anti-FUS(52–400) (third upper panel), or anti-α tubulin antibody (bottom panel). (D) Immunofluorescence labeling of retinal cells of 5-day-old gmr-driven allS-FUS wt or allS-FUS wt/FUS ΔC TG flies using anti-FUS(500–526) (green), anti-FUS (52–400) (red) or DRAQ5 (blue). Dashed lines indicate nuclear outline. Arrows indicate cytoplasmic aggregates. Scale bar = 5 µm.

We found that FLAG-FUS P525L or FUS ΔC was localized at the TIAR-positive stress granules in HEK293 cells (Supplementary Material, Fig. S3). To examine whether the self-assembly of FUS through its LC domain is required for the localization of FUS in stress granules, we expressed FLAG-tagged allS mutant FUS without or with P525L mutation (FLAG-allS-FUS wt and FLAG-allS-FUS P525L, respectively) in HEK293 cells and found that neither FLAG-allS-FUS wt nor FLAG-allS-FUS P525L was localized into the TIAR-positive granules (Supplementary Material, Fig. S5A). Moreover, when the transfected HEK293 cells were treated with 0.5 m arsenite to induce formation of stress granules, the induced stress granules were negative for allS-FUS wt or allS-FUS P525L (Supplementary Material, Fig. S5B). These data suggest that the self-assembly of FUS through its LC domain also is necessary for the recruitment of FUS into stress granules.

Discussion

In this study, we have shown the following: (i) overexpression of human FUS in the retina of Drosophila caused degeneration; (ii) a fALS mutation P525L exacerbated FUS-induced retinal degeneration by increasing the cytoplasmic localization of FUS; (iii) C-terminal truncation mutations of FUS attenuated the FUS-induced retinal degeneration despite the cytoplasmic localization; (iv) double overexpression of C-terminal truncation mutant FUS with FUS wt exacerbated the FUS-induced retinal degeneration by sequestrating FUS wt from nuclei into cytoplasmic aggregates; (v) allS mutant FUS, which was incapable of the self-assembly, completely abolished the FUS-induced retinal degeneration; and (vi) allS mutant FUS did not relocate FUS from nuclei into the cytoplasmic aggregates. Based on these data, we propose that the self-assembly of FUS through its LC domain within the cytoplasm contributes to neurodegeneration caused by FUS (Fig. 7).

Figure 7.

Schematic model depicting the possible mechanism of FUS-induced degeneration in the Drosophila retina. Wild-type FUS is predominantly localized at the nucleus (A). In contrast, fALS mutant FUS, e.g. P525L, or C-terminally truncated FUS, e.g. R495X, is mislocalized in the cytoplasm due to the impediment of the shuttle-back of FUS into the nucleus (B and C). Mislocalized FUS proteins in the cytoplasm are self-assembled via the LC domain and form aggregates, leading to neurodegeneration.

In this study, we generated series of FUS TG flies by the random insertion method. Site-directed insertion method may be superior to the random insertion method as a manipulation technique to generate TG flies, in terms of normalization of the expression levels of transgene and elimination of the positional effects of gene insertion. To avoid the positional effects of gene insertion, we generated more than three lines of each TG fly, and tried to normalize the expression levels of TG proteins by the semi-quantification of the protein levels. We have estimated that the expression level of FUS wt in one photoreceptor neuron was ∼26.3 pg (40–260 μm) (Supplementary Material, Fig. S1A). It has been reported that the concentration of endogenous FUS protein in HeLa cells is ∼2 μm (35), suggesting that the FUS wt TG flies expressed 20–130 times higher levels of FUS protein compared with the Hela cells. We observed mostly no degenerative phenotype in the retina of allS-FUS P525L TG flies (Fig. 5), despite the similar levels of FUS overexpression. Also, we were able to detect the difference in the magnitude of the degenerative phenotype in the retina of FUS P525L compared with FUS wt TG flies despite similar high levels of TG proteins (Fig. 1). From these data, we concluded that the comparison of the degenerative phonotype in FUS TG flies is feasible and reasonable in an overexpression paradigm as conducted in this study.

Recent in vitro studies revealed that the LC domain of FUS is required for the reversible formation of hydrogel (33,36) or phase-separated liquid-droplets (37,38); however, it was unclear how the LC domain of FUS is involved in the mechanism of neurodegeneration. In this study, using BiFC assay in HEK293 cells, we found that FUS wt and P525L were self-assembled in the nuclei and cytoplasm, respectively (Fig. 4). We also found that allS mutation of FUS, which eliminated the self-assembly of FUS (Fig. 4), completely abolished the degenerative phenotype in the retina of Drosophila induced either by FUS wt or P525L (Fig. 5). Based on these observations, we conclude that the self-assembly of FUS is required for the neurodegeneration induced by FUS. Recently, a set of ALS- or FTLD-related RNA-binding proteins, e.g. TDP-43, hnRNPA1, EWS or TAF15, also were reported to form phase-separated liquid droplets through their LC domains as in FUS protein (37,39,40). These results lead us to speculate that the self-assembly of the disease-related RNA binding proteins may be a common mechanism underlying the neurodegeneration in ALS or FTLD. ALS-linked mutations within the LC domain of FUS or TDP-43 have been reported to increase the formation of irreversible gels/aggregates from reversible hydrogels or liquid-liquid phase separated structures (35,41,42). These findings suggest that the formation of the irreversible gels/aggregate or reversible hydrogels may be involved in the mechanism of neurodegeneration induced by FUS through its LC domain.

Although endogenous FUS is predominantly localized in the nuclei, FUS-immunoreactive cytoplasmic inclusions are observed in neurons of patients with FUS proteinopathies. fALS-linked mutations within PY-NLS have been reported to disrupt Transportin-mediated nuclear import of FUS and lead to neurodegeneration (17,22). In this study, we observed that fALS-linked P525L mutant FUS was localized in cytoplasm and caused severe neurodegeneration compared with FUS wt in the retina of Drosophila (Fig. 1). These data support the hypothesis that cytoplasmic mislocalization of FUS is one of the key mechanisms of neurodegeneration caused by FUS. However, we found that FUS R495X TG flies exhibited very mild retinal degeneration despite the cytoplasmic localization of R495X mutant FUS (Fig. 2B and D). Our findings are corroborated by the recent finding that the R495X mutant FUS caused only mild ommatidial phenotype in the retina and no significant reduction in climbing ability in Drosophila (43). These results suggest that C-terminal truncation mutation attenuated the FUS-induced toxicity in the cytoplasm, although it does not dovetail with the fact that R495X mutation is associated with a very aggressive disease course of ALS in humans. We found that the C-terminal truncation mutant FUS (FUS ΔC, R495X) sequestered FUS wt from nuclei into cytoplasmic aggregates in the retina of Drosophila (Fig. 2G) as well as in HEK293 cells (Figs 3 and 4I), and that double expression of FUS wt with either FUS ΔC or R495X in the retina of Drosophila caused severe retinal degeneration compared with FUS wt TG fly (Fig. 2C and D). Based on these observations, we speculated that the sequestration of FUS wt from nuclei into cytoplasmic aggregates caused by the R495X mutant FUS was the key mechanism that accounts for the reappearance of the severe toxicity linked to full-length FUS. We also found that allS mutant FUS was not sequestered by FUS ΔC in the retina of Drosophila (Fig. 6), and that allS-FUS × FUS ΔC double TG flies exhibited very mild retinal degeneration at similar levels to FUS ΔC TG flies (Fig. 6). These data support the notion that C-terminal truncation mutation of FUS may cause neurodegeneration by augmenting the sequestration of FUS wt into cytoplasmic aggregates through its LC domain. We found that Caz, a Drosophila homolog of FUS, was not sequestered into the cytoplasmic aggregates by the putative interaction with FUS ΔC in the retina of FUS ΔC TG flies. Interestingly, Caz lacks the amino-terminal QGSY-rich region, which comprises the predominant portion of the LC domain (44). We reasoned that FUS ΔC was not capable of sequestering Caz into the cytoplasmic aggregates because of the lack of the QGSY-rich region in Caz. This may also suggest that loss-of-function of Caz is not the major cause of FUS-induced neurodegeneration in the FUS wt TG flies.

The mechanism whereby the self-assembled FUS in cytoplasmic aggregates then induces neurodegeneration is unclear. One possibility is that the cytoplasmic aggregates induce neuronal toxicity through perturbation of the function of ribonucleoproteins in RNA processing. Previous report that fALS-linked mutations within the LC domain impaired the new protein synthesis in cultured neurons may support this hypothesis (41). Another possibility is that the self-assembled FUS per se causes neuronal toxicity. Our finding that the C-terminal truncation mutants of FUS attenuated the FUS-induced degeneration (Fig. 2B and D) might suggest the involvement of as yet unknown interactors with the C terminus of FUS in the mechanism of degeneration. Transportin 1 is known to interact with the C terminus of FUS and regulate the nuclear transport of FUS (17). Previous immunohistochemical analyses revealed that Transportin 1 was colocalized with FUS-immunopositive inclusions in the brains of patients with FUS proteinopathies (45,46), suggesting that Transportin 1 itself may be involved in the formation of FUS inclusions and related toxicity through interaction with the C terminus of FUS. It is also possible that the cytoplasmic mislocalization of FUS may compromise the physiological function of FUS in the nuclei, leading to the abnormality in RNA processing. Previous report showing that RNA-binding-incompetent mutations in FUS block the neurodegeneration in the retina of Drosophila (25) may support this view. However, the observations that FUS deficiency in mice did not exhibit the FTLD- or ALS-like phenotypes (14–16) may favor the possibility of a mechanism of neurodegeneration independent of its physiological function. We cannot rule out the possibility that the carboxy-terminal truncation cause a change of tertiary structure of FUS into a less-toxic conformation.

In sum, our present findings suggest that the self-assembly of FUS in the cytoplasm through its LC domain induces neurodegeneration in FUS proteinopathies. Also, our observations in a series of allS mutant FUS TG flies raises hopes that suppression of self-assembly of FUS may be a potential therapeutic strategy. Further investigations using lines of FUS TG flies and BiFC system we have created in this study will unveil the molecular mechanism underlying the self-assembly leading to FUS proteinopathies.

Materials and Methods

Fly stocks and generation of TG flies

TG flies were raised using standard procedures at 25 degrees. GAL4-UAS system was used for tissue specific overexpression. gmr-GAL4 and UAS-LacZ lines were purchased from Bloomington Drosophila Stock Center. To generate the lines of human wt or mutant FUS TG fly, respective constructs were injected into w¹¹¹⁸ embryos using the random insertion method as described previously in (47). We have checked the expression levels of TG FUS proteins by immunoblotting and selected a single line that expressed the transgene products at closer levels to FUS wt in the head of FUS wt TG flies. The number of clones of each single TG flies analyzed in this work is 4 for FUS wt, 3 for FUS P525L, 3 for FUS R495X, 6 for FUS ΔC, 3 for allS-FUS wt, 3 for allS-FUS P525L, 4 for FUS ΔQGSY, or 4 for FUS ΔQGSY-G (Supplementary Material, Fig. S6).

Plasmid construction

Human FUS cDNA was purchased from Life Technologies. AllS mutant FUS cDNA was kindly gifted from Dr Steven L. McKnight and Dr Masato Kato at University of Texas Southwestern Medical Center. XhoI site was added to the forward primer and HindIII site was added to the reverse primer for human FUS and the XhoI/HindIII fragment was subcloned into pColdI vector (Takara). Following primers were used. Forward: 5’-CGGTGCTCGAGGGTGTTGGAACTTC-3’ and reverse: 5’-AAGCTTTTCCAGAACCTGGGGAGCC-3’. Site-directed mutagenesis was carried out for P525L mutant. Respective primer pairs were used as follows. P525L forward: 5’-GACAGGATCGCAGGGAGAGGCTGTATTAA-3’, P525L reverse: 5’-TTAATACAGCCTCTCCCTGCGATCCTGTC-3’. ΔC, R495X, ΔQGSY, ΔQGSY-G and ΔRRM mutants were generated by inverse PCR methods. Respective primer pairs were used as follows. ΔC forward: 5’-TAATTAGCCTGGCTCCCCAGGTTC-3’, ΔC reverse: 5’-GGAATCCATCTTGCCAGGGCCAAAG-3’, R495X forward: 5’-TAATTAGCCTGGCTCCCCAGGTTC-3’, R495X reverse: 5’-GAAGCCTCCACGGTCCCCGC-3’, ΔQGSY forward: 5’-GGTGGTGGAGGTGGAGGTGGAGGTG-3’, ΔQGSY reverse: 5’-CATGTCCGCGCACGCGCGCACAGGC-3’, ΔQGSY-G forward: 5’-CCTCGGGACCAAGGATCACGTCATG-3’, ΔQGSY-G reverse: 5’-CATGTCCGCGCACGCGCGCACAGGC-3’, ΔRRM forward: 5’-GCTACTCGCCGGGCAGACTTTAATC-3’ and ΔRRM reverse: 5’-GTTGTTGTCTGAATTATCCTGTTCG-3’. For generation of TG fly, XhoI/XbaI fragment of FUS/pColdI was subcloned into pUAST vector.

For the mammalian expression of human FUS wt, BamHI site was added to forward primter and XhoI site was added to reverse primer for human FUS and the BamHI/XhoI fragment was subcloned into pcDNA5 FLAG A or pcDNA5 myc A vectors. Following primers were used. Forward: 5’-CGGGGATCCCCATGGCCTCAAAC-3’ and reverse: 5’-GCCGCTCGAGTTAATACGGCCTCTC-3’. For the mammalian expression of P525L, ΔC and R495X mutants, reverse primers were as follows instead. P525L reverse: 5’-GCCGCTCGAGTTAATACAGCCTCTC-3’, ΔC reverse: 5’-GCCGCTCGAGTTAGGAATCCATCTTGC-3’ and R495X reverse: 5’-CCCGCTCGAGTCAGAAGCCTCCACG-3’.

For BiFC assay, HindIII site was added to forward primer and BamHI site was added to reverse primer for Venus, and BamHI site was added to forward primer and XhoI site was added to reverse primer for human FUS. Respective HindIII/BamHI fragment for Venus and BamHI/XhoI fragment for FUS were subcloned into pcDNA3.1 vector. Following primers were used. Venus or V1 forward: 5’-AAAAAAAGCTTGCCACCATGGTGAGCAAGGGCG-3’, V2 forward: 5’-AAAAAAAGCTTGCCACCATGAAGAACGGCATCAAGGCC-3’, Venus or V2 reverse: 5’-GCATCGGATCCCTTGTACAGCTCG-3’, V1 reverse: 5’-AAAAAGGATCCCTGCTTGTCGGCGGTGATATAG-3’, V1 only reverse: 5’-AAAAAGGATCCTTACTGCTTGTCGGCGGTGATATAG-3’, FUS wt or P525L forward: 5’-AAAAAGGATCCATGGCCTCAAACGATTATACC-3’, FUS wt reverse: 5’-CTAGACTCGAGTTAATACGGCCTCTCC-3’ and FUS P525L reverse: 5’-CTAGACTCGAGTTAATACAGCCTCTCC-3’.

For BiLC assay, human FUS was first subcloned into XhoI/XbaI site of pBiT1.1-N[TK/LgBiT] vector (Promega) or pBiT2.1-N[TK/SmBiT] vector (Promega) using following primer pairs. Forward primer used for FUS: 5’-AAAAACTCGAGCGGTATGGCCTC-3’, and reverse primer used for FUS wt: 5’-AAAAATCTAGATTAATACGGCCTCTCCCTG-3’, for FUS P525L: 5’-AAAAATCTAGATTAATACAGCCTCTCCCTG-3’, or for FUS R495X: 5’-AAAAATCTAGATTAGAAGCCTCCACGGTCCC-3’. Respective fragment for LgBiT-FUS or SmBit-FUS was then subcloned into XhoI/XbaI site of pcDNA3.1 vector.

Cell cultures and transfection

HEK293 cells (ATCC CRL-1573) were cultured in Dulbecco’s modified Eagle medium with 10% fetal bovine serum. FUS expressing vectors were transfected using Polyethylenimine ‘Max’ (Polyscience inc) according to manufacturer’s protocol. Cells were analyzed 48 h after transfection.

Antibodies

The following antibodies were used as primary antibody. Anti-FUS antibodies [anti-FUS(52–400) (Proteintech, 60160–1-Ig], anti-FUS(400–450) (Bethyl, A300–293A) and anti-FUS(500–526) (Bethyl; A300–294A)] were used for immunohistochemistry and immunoblotting analyses. Anti-FLAG (Sigma; M2, or Cell Signaling Technology; no. 2368), anti-Myc (Cell Signaling Technology; 9B11), anti-α-tubulin (Sigma; DM1A), anti-GFP (Life Technologies; A11122), or anti-GFP amino-terminus (Sigma; G1544) was used for immunohistochemistry or immunoblotting analyses. Anti-Caz antibody (kindly gifted from Dr Y. Azuma and Dr M. Yamaguchi) was used for immunohistochemistry of Caz (48).

Immunohistochemistry

Immunohistochemistry of TG flies were performed as described previously (49). Briefly, heads of 5-day-old flies were fixed with 4% paraformaldehyde (PFA) in PBS containing 0.1% Triton X-100 at room temperature for 2 hrs. Fixed heads were embedded in paraffin and cut in coronal sections at 4 µm thickness. H&E staining was performed to evaluate retinal structures and retinal thickness was quantified with Image J software. For immunostaining of retinal cells, sections were treated with microwave for 10 min in citrate buffer (pH = 6.0) for antigen retrieval. After blocking with 10% calf serum in PBS, sections were incubated with primary antibodies for overnight at 4 degrees. For immunofluorescence, sections were incubated with a mixture of Alexa fluorophore conjugated secondary antibodies against mouse or rabbit IgG and DRAQ-5 as a nuclear marker. Retinal cells were observed with Leica confocal microscopy SP5.

Immunocytochemistry

For immunocytochemistry of HEK293 cells, cells were fixed with 4% PFA in PBS for 30 min at room temperature. After blocking with 10% calf serum in PBS containing 0.1% Triton X-100, cells were incubated with primary antibodies for overnight at 4 degrees. For immunofluorescence, cells were incubated with a mixture of Alexa fluorophore conjugated secondary antibodies against mouse or rabbit IgG and DRAQ-5 as a nuclear marker. Cells were observed with Leica confocal microscopy SP5.

Immunoblotting

Ten heads of 5-day-old flies or transfected HEK293 cells were lysed in a Laemmli sample buffer. Lysate was separated by 10% SDS-PAGE containing 2% SDS and transferred to polyvinylidene difluoride membranes. After probing with primary antibodies, the immunoblots were developed using a chemiluminescence kit (Wako) and visualized by LAS-4000 mini (GE healthcare). We quantified the band intensities of FUS proteins derived from the head of each TG fly line by immunoblotting using the anti-FUS antibody (400–450) and calculated the average expression levels of each TG fly line in more than three independent experiments. If we did not obtain statistically significant difference in the expression levels between individual samples using the one-way ANOVA test followed by post-hoc comparison using Tukey-Kramer test, we described the results as ‘comparable expression levels’.

BiFC/BiLC assay

Generation of expression vector for amino-terminal (1–158 a.a., V1) or carboxy-terminal (159–239 a.a., V2) Venus fused FUS was described in plasmid construction. 0.3 μg of plasmid cDNA was transfected into HEK293 cells using Polyethylenimine ‘Max’. After 16-h incubation, cells were fixed with 4% PFA in PBS for 15 min at room temperature. DRAQ-5 was used as a nuclear marker. Cells were observed with Leica confocal microscopy SP5 and fluorescent intensity of Venus was measured using Image J software.

Generation of expression vector for LgBiT or SmBiT Venus fused FUS was described in plasmid construction. 8.0 × 10³ cells of HEK293 were plated onto 96-well plate. 1.8 μg of plasmid cDNA with 20 ng of firefly luciferase cDNA (pGL4.54[luc2/TK], Promega) was transfected into HEK293 cells using FUGENE6 according to manufacturer’s protocol. After 20-h incubation, NanoBiT luminescence was measured using the Nano-Glo Live Cell Substrate (Promega) according to manufacturer’s protocol by multi-plate luminometer GloMax Navigator (Promega). After measurement of the NanoBiT luminescence, cells were re-incubated for 1 h and the firefly luminescence was measured using the ONE-Glo EX luciferase Assay Substrate (Promega) according to manufacturer’s protocol by multi-plate luminometer GloMax Navigator (Promega). The measured value of NanoBiT luminescence was normalized by the measured value of firefly luminescence in each well.

Statistics

All quantitative data were shown as mean ± S.E. One-way or two-way ANOVA were used for statistical analysis.

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

We thank Bloomington Drosophila Stock Center for GMR-GAL4 and UAS-LacZ TG lines. We thank Dr Steven L. McKnight and Dr Masato Kato at University of Texas Southwestern Medical Center for allS-mutant FUS cDNA. We thank Dr Yumiko Azuma and Dr Masamitsu Yamaguchi at Kyoto Institute of Technology for anti-Caz antibody. We also thank Dr Masayuki Miura and Dr Takahiro Chihara at the University of Tokyo for valuable suggestions and technical supports.

Conflict of Interest statement. None declared.

Funding

This work was supported by Grants-in-Aid for Scientific Research (C) (15K08297 to T.H.), Grant-in-Aid for Scientific Research on Innovative Areas (Brain Protein Aging and Dementia Control) (17H05687 to T.H.) and for JSPS Fellows (14J06189 to K.M.) from the Ministry of Education, Culture, Sports, Science and Technology, Japan.

Author notes

The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint First Authors.