Mutant FUS proteins that cause amyotrophic lateral sclerosis incorporate into stress granules

Abstract

Mutations in the RNA-binding protein FUS (fused in sarcoma) are linked to amyotrophic lateral sclerosis (ALS), but the mechanism by which these mutants cause motor neuron degeneration is not known. We report a novel ALS truncation mutant (R495X) that leads to a relatively severe ALS clinical phenotype compared with FUS missense mutations. Expression of R495X FUS, which abrogates a putative nuclear localization signal at the C-terminus of FUS, in HEK-293 cells and in the zebrafish spinal cord caused a striking cytoplasmic accumulation of the protein to a greater extent than that observed for recessive (H517Q) and dominant (R521G) missense mutants. Furthermore, in response to oxidative stress or heat shock conditions in cultures and in vivo, the ALS-linked FUS mutants, but not wild-type FUS, assembled into perinuclear stress granules in proportion to their cytoplasmic expression levels. These findings demonstrate a potential link between FUS mutations and cellular pathways involved in stress responses that may be relevant to altered motor neuron homeostasis in ALS.

Figure 1.

Domain structure of FUS and expression of normal and mutant FUS in HEK-293 cell lines. (A) The functional domains of FUS (www.uniprot.org) include a Gln-Gly-Ser-Tyr (QGSY)-rich region (blue), a Gly-rich domain (red), an RNA recognition motif (RRM; green), an Arg-Gly (RG)-rich region (yellow), which contains a RanBP2-type zinc finger domain (ZF; orange) and a putative NLS (light green). Labeled are fALS-linked mutants within exons 14 and 15, with those investigated here (R495X, H517Q and R521G) highlighted in red. Truncated FUS constructs of this study included an ALS-linked mutant (R495X) and an experimental mutant (G515X, blue), which removed the sequence encoded by exon 15. These residues are underlined in the primary sequence. (B) GFP-FUS stable lines and naïve HEK-293 cells were cultured with (+) or without (−) doxycycline for 40 h, and lysates containing 3 µg total protein/lane were analyzed by western blot using anti-GFP, anti-FUS and anti-GAPDH (loading control) antibodies. (C) Confocal microscopy of HEK-293 cell lines following induction with doxycycline for 40 h and staining with the nuclear dye DAPI (blue) and an anti-GFP antibody. A strong cytoplasmic GFP-FUS signal was observed for the R495X and G515X mutants with a truncated NLS, whereas GFP-FUS(WT, R521G and H517Q) exhibited a predominately nuclear localization. Scale bar, 10 µm. (D) Quantitative analysis of GFP fluorescence from live cells (see Materials and Methods) showed that the ratio of cytoplasmic:nuclear GFP fluorescence intensity was significantly higher for GFP-FUS(R521) compared with GFP-FUS(WT), and for the truncation mutants GFP-FUS(R495X and G515X) compared with the other GFP-FUS proteins (WT, R521G and H517Q). Asterisks indicate statistically significant differences between cell lines (n = 11–28 cells analyzed per line) as determined by the Kruskal–Wallis ANOVA followed by a Dunn's multiple comparison test (*P < 0.05, **P < 0.01).

Figure 2.

Incorporation of mutant FUS into stress granules in HEK-293 cells. Each GFP-FUS HEK-293 stable cell line was induced with doxycycline for 40 h, treated with 0.5 mm sodium arsenite for 1 h, and then probed with anti-GFP (green) and anti-TIAR (red) antibodies, and the nuclear dye DAPI (blue). Cytoplasmic aggregates containing GFP-FUS were detected with anti-GFP for the R521G, R495X and G5151X lines, but not for the WT and H517Q lines. Composite images indicate that the accumulated GFP-FUS(R521G, R495X and G515X) co-localized with the TIAR stress granule marker. Scale bar, 10 µm.

Figure 3.

Recruitment of GFP-FUS into perinuclear stress granules was more rapid and extensive for truncation mutants. GFP-FUS(R521G, R495X or G515X) was induced for 4 days in HEK-293 cells, and a time series of GFP intensities was acquired in live cells at 37°C over 30 min following the addition of 0.5 mm sodium arsenite. Differential interference contrast (DIC) images (left) indicate cell positions, and confocal fluorescence images (right) show maximum-intensity z-projections obtained from serial 0.2 µm slices using a 60× objective. GFP fluorescence revealed a rapid (within 4–7 min) formation of FUS(R495X and G5151X) aggregates, whereas FUS(R521G) formed smaller, less intense aggregates with a slower time course. Scale bar, 10 µm. The full time course appears in Supplementary Material, Videos Fig3video2.avi (G515X), Fig3video3.avi (R495X) and Fig3video4.avi (R521G). FUS WT expression under the same conditions did not accumulate in the cytoplasm (Supplementary Material, Fig3video5.avi).

Figure 4.

Formation of cytoplasmic GFP-FUS aggregates was reversible after heat shock. GFP-FUS(WT, H517Q, R521G, R495X or G515X) was induced for 3 days in HEK-293 cells, and a time series of GFP intensities (acquired as in Fig. 3) was detected in live cells over the indicated time course (hh:mm) upon shifting from 37–42.5°C. Cytoplasmic FUS inclusions assembled most rapidly (by 7–12 min) in cells expressing GFP-FUS(R495X and G5151X) and to a lesser extent in cells expressing GFP-FUS(R521G and H517Q). Aggregate formation was reversible within minutes of a temperature shift back to 37°C, as shown for GFP-FUS(R521G, R495X and G5151X). Scale bar, 10 µm. The full time course appears in Supplementary Material, Videos Fig4video6.avi (G515X), Fig4video7.avi (R495X), Fig4video8.avi (R521G), Fig4video9.avi (H517Q) and Fig4video10.avi (WT).

Figure 5.

Cytoplasmic GFP-FUS in stress granules did not co-localize with adjacent P-bodies. HEK-293 cell lines stably expressing GFP-FUS proteins were treated as described in Figure 2, except that 1 mM sodium arsenite was administered for 2 h and cells were probed with anti-Hedls/GE-1, a marker of P-bodies (red). Scale bar, 10 µm. None of the GFP-FUS proteins co-localized with the P-body marker. However, P-bodies localized in close proximity to stress granules containing the GFP-FUS(R521G, R495X and G515X) variants, and in some cases P-bodies appeared docked to these stress granules (see inset for G515X; scale bar, 1 µm).

Figure 6.

Expression of GFP-FUS variants in spinal cord and body wall muscle of zebrafish embryos. Zebrafish eggs were injected at the 1–2 cell stage with mRNAs encoding GFP-FUS variants. Embryos were fixed at 25 hpf, and the intact body wall region was immunostained with anti-GFP (green) and a nuclear marker (Draq5, blue). Confocal image stacks (40× objective, Δz = 1.0 µm) were acquired in a longitudinal orientation (red box) from the lateral body muscle (m) extending medially through the spinal cord (sc) and notochord (n). Shown are representative medial and lateral slices. For embryos expressing GFP-FUS(WT, H517Q and R521G), the GFP signal was predominantly nuclear, whereas for embryos expressing the R495X truncation mutant, the GFP localized mostly to the cytoplasm. Scale bar, 50 µm.

Figure 7.

Expression of GFP-FUS variants in spinal cord of zebrafish embryos. Zebrafish eggs were injected with mRNAs encoding GFP-FUS variants, and embryos were processed for confocal microscopy (100×) as described in Figure 6. Shown are representative 0.9 µm slices (left panels) and 0.4 µm slices acquired using a 3.44× optical zoom (right panel). The higher magnification clearly showed the nuclear expression of GFP-FUS(WT, H517Q and R521G) variants and cytoplasmic accumulation of the R495X or G515X truncation mutants in the spinal cord. Scale bars, 10 µm.

Figure 8.

Heat shock increased the cytoplasmic localization of the H517Q mutant and triggered accumulation of R521G, R495X and G515X into stress granules in vivo. Confocal slices of zebrafish spinal cord from 25 hpf embryos incubated at 42.5°C for 45 min followed by immediate fixation and double immunostaining with anti-GFP (green) and anti-TIAR (red). For a group of spinal cord cells from embryos expressing the H517Q mutant, the GFP redistributed to the cytoplasm in a granular pattern without incorporating prominently into stress granules, whereas this occurred only rarely for cells expressing GFP-WT FUS. For embryos expressing the R521G, R495X and G515X mutants, the GFP assembled into aggregates that co-localized with anti-TIAR staining. Nuclei were stained with Draq5. Left panel obtained at 100×, right panels obtained using 3.44× optical zoom. Scale bars, 10 µm.