Mechanism of karyopherin-β2 binding and nuclear import of ALS variants FUS(P525L) and FUS(R495X)

Abstract

Mutations in the RNA-binding protein FUS cause familial amyotropic lateral sclerosis (ALS). Several mutations that affect the proline-tyrosine nuclear localization signal (PY-NLS) of FUS cause severe juvenile ALS. FUS also undergoes liquid-liquid phase separation (LLPS) to accumulate in stress granules when cells are stressed. In unstressed cells, wild type FUS resides predominantly in the nucleus as it is imported by the importin Karyopherin-β2 (Kapβ2), which binds with high affinity to the C-terminal PY-NLS of FUS. Here, we analyze the interactions between two ALS-related variants FUS(P525L) and FUS(R495X) with importins, especially Kapβ2, since they are still partially localized to the nucleus despite their defective/missing PY-NLSs. The crystal structure of the Kapβ2·FUS(P525L)^PY-NLS complex shows the mutant peptide making fewer contacts at the mutation site, explaining decreased affinity for Kapβ2. Biochemical analysis revealed that the truncated FUS(R495X) protein, although missing the PY-NLS, can still bind Kapβ2 and suppresses LLPS. FUS(R495X) uses its C-terminal tandem arginine-glycine-glycine regions, RGG2 and RGG3, to bind the PY-NLS binding site of Kapβ2 for nuclear localization in cells when arginine methylation is inhibited. These findings suggest the importance of the C-terminal RGG regions in nuclear import and LLPS regulation of ALS variants of FUS that carry defective PY-NLSs.

Figure 1

Structure of Kapβ2·FUS(P525L)^PY-NLS complex. (A) Sequence of FUS PY-NLS and the P525L mutation site. PY-NLS residues that are modeled in the Kapβ2·FUS(P525L)^PY-NLS crystal structure are underlined. (B) The overall structure of FUS(P525L)^PY-NLS (yellow) bound to Kapβ2 (purple). (C) Comparison of the Kapβ2-bound FUS(P525L)^PY-NLS (yellow) with WT FUS^PY-NLS (cyan). (D) Details of PY-NLS epitope-3 (FUS(P525L)^PY-NLS is in yellow; WT FUS^PY-NLS in cyan) interacting with Kapβ2 (purple and pink). Representative contacts ≤ 4.1 Å are shown as dashed lines. (E) Stereo view of the simulated annealing (SA) composite omit map 2Fo—Fc map contoured at 1.0σ overlaid onto the modeled mutant PY-NLS peptide.

Figure 2

Interactions of FUS with Importins. (A) FUS constructs used in (B–E). (B) Pull-down binding assay showing interactions between MBP-FUS, MBP-FUS(1–500), MBP-FUS(R495X), MBP-FUS PY-NLS or Impα with GST-Impβ or GST-Kapβ2 immobilized on beads. Control experiments and unbound proteins in the pull-down experiments are shown in Supplementary Fig. 1A. (C) Pull-down binding assay of MBP-FUS(1–500) with immobilized GST-Kapβ2 in the presence of MBP-FUS PYNLS. (D) Pull-down binding assay to probe interactions between MBP-FUS(1–500) and GST-Kapβ2, -Impα or -Impα/β. (E) Pull-down binding assay of MBP-FUS(1–500) with immobilized GST-Impβ in the presence of MBP-IBB. Proteins in (B–E) are visualized by Coomassie-stained SDS-PAGE.

Figure 3

Turbidity assays of FUS in the presence of importins. (A) Left panel: turbidity assays with 8 μM MBP-FUS(1–500) in the presence of buffer, 8 µM Kapβ2, 8 µM Kapβ2·PYNLS and 8 µM Kapβ2 + 8 µM RanGTP. Right panel: turbidity assays of 8 µM MBP-FUS (full length) in the presence of buffer, 8 µM Kapβ2 or Kapβ2·M9M. (B) Left panel: turbidity assays of 8 μM MBP-FUS(1–500) in the presence of buffer, 8 µM Impβ, 8 µM Impβ + 8 µM IBB, 8 µM Impα/β, and 8 µM Impβ + 8 µM RanGTP. Right panel: turbidity assays of 8 µM MBP-FUS (full length) in the presence of buffer, 8 µM Impβ, 8 µM Impβ + 8 µM IBB and control of IBB alone. For turbidity assays In (A,B), importins and other proteins were added to the MBP-FUS proteins prior to addition of the Tev protease at time = 0 min, and OD_{395 nm} of the solutions measured 60 min after addition of the Tev protease. The experiments were performed at room temperature, OD_{395 nm} normalized to measurements of MBP-FUS proteins + buffer + Tev at time = 60 min, the mean of 3 replicate experiments, ± SD are shown. (C) Turbidity assays of 8 µM of three different MBP-FUS constructs in the presence of buffer or 2–16 µM Kapβ2 at 10 °C. Kapβ2 is added prior to Tev (added at time = 0 min) and OD_{395 nm} is recorded 60 min after Tev addition. OD_{395 nm} is normalized to measurements of respective MBP-FUS construct + buffer + Tev. Mean of 3 or 4 replicate experiments, ± SD is shown.

Figure 4

Interactions of Kapβ2 with various FUS fragments. (A) FUS constructs used in (B,C). (B,C) Pull-down binding assay showing the interactions of immobilized GST-Kapβ2 with various MBP, MBP-M9M and MBP-FUS constructs of various lengths. The MBP-FUS constructs in (B) have systematic deletions of regions from the C-terminus. Fragments containing residues 1–500 is missing the PY-NLS, 1–452 is missing RGG3-PYNLS, 1–430 is missing ZnF-RGG3-PYNLS, 1–370 is missing RGG2-ZnF-RGG3-PYNLS and 1–265 contains only LC-RGG1. All but one MBP-FUS constructs in (C) have no PY-NLS and systematic deletions of regions from the N-terminus. Fragments containing residues 121–500 is missing the LC and PY-NLS, 277–500 is missing LC-RGG1 and PY-NLS, 371–500 is missing LC-RGG1-RRM and PYNLS, 453–500 contains only RGG3 and 278–385 contains only the RRM. (D) Pull-down binding assay showing immobilized GST-Kapβ2 and GST-Impβ binds MBP-FUS(1–500) but not if all arginine residues in RGG2 and RGG3 were substituted by lysine (RtoK). (E) Pull-down binding assay of various immobilized GST-Importins with MBP-FUS 371–500. Proteins in (B–E) are visualized by Coomassie-stained SDS-PAGE.

Figure 5

Localization of full-length FUS, FUS(R495X) and FUS(1–500) in HeLa cells. (A) Confocal microscopic images of live HeLa cells expressing EYFP₂-FL FUS, EYFP₂-FUS(R495X) or EYFP₂-FUS(1–500). (B) Bar diagram of relative percentage of nuclear and cytoplasmic fluorescence intensity in cells, with the mean ± SEM, n = 10–14. Significant differences compared with the corresponding control samples are indicated ***p < 0.001. ns, not significant (Ordinary one-way ANOVA test). (C) Fluorescence microscopic images of fixed HeLa cells expressing Flag-FL FUS or -FUS(R495X), with or without methylation inhibitor AdOx (20 μM) treatment. FUS is visualized by immunofluorescence (Alexa 488 secondary antibody). (D) Bar diagram of relative percentage of nuclear and cytoplasmic fluorescence intensity of cells in (C). More than 30 cells were analyzed for each experiment. Error bars represent standard deviation. ***Indicates adjusted P values < 0.001.

Figure 6

Localization of FUS(371–500) in the presence and absence of importin inhibitors. (A) Confocal microscopic images of live HeLa cells expressing EYFP₂ or EYFP_2-FUS(371–500), with or without methylation inhibitor AdOx (20 μM treatment. Hoechst 33,342 was used as nuclear counter stain. Scale bar = 10 μm. (B) Bar diagram of relative percentage of nuclear and cytoplasmic fluorescence intensity in cells is shown with the mean ± SEM, n = 10–14. (C) Localization of EYFP₂-FUS(317–500) in the presence of importin β1 inhibitor (Importazole), importin α inhibitor (expressed Bimax2 peptide) and Kapβ2 inhibitor (expressed M9M peptide). The HeLa cells were first treated with AdOx. Expression of the peptide inhibitors were either monitored directly (RFP-BiMax2, second row) or by direct immunofluorescence (MBP-M9M via Alexa 565 s antibody, second row). Scale bar = 10 μm. (D) Bar diagram of relative percentage of nuclear (Nuc) and cytoplasmic (Cyto) fluorescence intensity in cells is shown with the mean ± SEM, n = 10–14. Ordinary one-way ANOVA test was performed for statistical analysis using GraphPad software. Significant differences compared with the corresponding control samples are indicated ****p < 0.001. ns, not significant.