Interactions between ALS-linked FUS and nucleoporins are associated with defects in the nucleocytoplasmic transport pathway

Abstract

Nucleocytoplasmic transport (NCT) decline occurs with aging and neurodegeneration. Here, we investigated the NCT pathway in models of amyotrophic lateral sclerosis-fused in sarcoma (ALS-FUS). Expression of ALS-FUS led to a reduction in NCT and nucleoporin (Nup) density within the nuclear membrane of human neurons. FUS and Nups were found to interact independently of RNA in cells and to alter the phase-separation properties of each other in vitro. FUS-Nup interactions were not localized to nuclear pores, but were enriched in the nucleus of control neurons versus the cytoplasm of mutant neurons. Our data indicate that the effect of ALS-linked mutations on the cytoplasmic mislocalization of FUS, rather than on the physiochemical properties of the protein itself, underlie our reported NCT defects. An aberrant interaction between mutant FUS and Nups is underscored by studies in Drosophila, whereby reduced Nup expression rescued multiple toxic FUS-induced phenotypes, including abnormal nuclear membrane morphology in neurons.

Figure 1.. The Ran gradient is perturbed in motor neurons expressing ALS-linked FUS.

Immunofluorescence imaging of Ran revealed enhanced cytoplasmic versus nuclear Ran localization in FUS M511Nfs iPSC-derived motor neurons compared to control and FUS R521G motor neurons. Neurons at DIV 36–42 were used and the indicated number of cells (below) were analyzed across at least three independent differentiations (each differentiation represents an experiment). (a-c) Four control (C1–C4) and two ALS (M511Nfs and R521G) lines were employed (a) Representative immunofluorescence images for neurons derived from two control and two ALS iPSC lines. Staining includes Ran (green), the motor neuron marker Islet1/2 (red), the pan-neuronal marker Tuj1 (anti-TUBB3; grey), and DAPI (blue). Scale bar= 10 μm. The Ran N/C ratio of the chosen representative neurons: C1 (1.25 and 1.53); C4 (2.12, 0.99); M511Nfs (0.64, 0.69); R521G (1.29). (b,c) Quantification of Ran N/C ratio in all six lines (pooled controls, n=151; M511Nfs, n=117; R521G, n=63 cells) reveals a significant decrease for FUS M511Nfs neurons. Each point in the bar graph represents the average across all cells for an independent experiment. Within each experiment, the ALS lines were normalized to the control lines. One-way ANOVA with Dunnett’s multiple comparisons test, ***p=0.0003, n.s. p=0.0967. F (2, 328)= 7.632. Error bars represent s.e.m. (c) Frequency distribution graphs of Ran N/C ratio for FUS M511Nfs (left) and R521G (right) neurons compared to controls. (d-f) Similar analysis as A-D using FUS M511Nfs motor neurons compared to the isogenic, corrected control (M511Nfs*Cor). (d) Immunofluorescence staining as in (a). Scale bar= 10 μm. The Ran N/C ratio of the chosen representative neurons: M511Nfs*Cor (1.19); M511Nfs (0.55, 0.58, 0.91). Abbreviation within the zoom insets: N (nucleus), C (cytoplasm). (e) Quantification of the Ran N/C ratio as in (b). Ran N/C is significantly lower in M511Nfs versus M511Nfs*Cor neurons (M511Nfs*Cor, n=103; M511Nfs, n=90 cells). Unpaired two-tailed t-test, **p=0.0063. t=2.762, df=191. (f) Quantification of the Ran N/C ratio in ISL1/2 negative M511Nfs and M511Nfs*Cor neurons (non-motor neurons). Ran N/C is significantly lower in M511Nfs versus M511Nfs*Cor neurons (M511Nfs*Cor, n=32; M511Nfs, n=29 cells). Unpaired two-tailed t-test, ****p<0.0001. t=4.594, df=59. (g) A Western blot analysis of total Ran protein levels in neurons derived from the indicated line; data from three independent differentiations are shown. Tuj1 was used as loading control. Source data are provided as a Source Data file. (h) Quantification of (g). Unpaired two-tailed t-test, n.s.= not significant. t=0.09654, df=4. For all quantifications, data distribution was assumed to be normal but this was not formally tested.

Figure 2.. Diminished nuclear pore signal and nucleocytoplasmic transport in ALS-FUS human neurons.

(a) Immunofluorescence images of Pom121 (red) and Nup62 (green), representative of n=5 experiments, reveals weaker signals in FUS M511Nfs nuclei compared to nuclei from the isogenic control neurons. Binary images (white) derived from the fluorescence images are shown. DAPI (blue). Scale bar = 5 μm. (b) An illustration for the cluster analysis used to quantify the continuity of Pom121 and Nup62 staining around the perimeter of the nucleus. (c-f) Results of the cluster analysis for Pom121 (scaffold Nup; c, e) and Nup62 (central channel Nup; d, f) for n=152 M511Nfs*Cor and n=144 M511Nfs nuclei that were analyzed across five independent differentiations (each differentiation represents an experiment) from neurons at DIV 36–45. (c, d): significantly higher cluster numbers in FUS M511Nfs versus M511Nfs*Cor nuclei through analysis of Pom121 (Unpaired two-tailed t-test, **p=0.0095, t=3.391, df=8) and Nup62 (Unpaired two-tailed t-test, p=0.0137, t=3.144, df=8; error bars represent s.e.m.). (e, f): results of the cluster analysis presented in a population pyramid format show the cluster number for every neuron analyzed. Each point in a bar graph represents the average across all cells in that experiment, normalized to M511Nfs*Cor. (g-i) Results of the FRAP experiment with the NLS-tdTomato-NES shuttling reporter in FUS M511Nfs*Cor neurons (black; n=29 cells) and M511Nfs neurons (red; n=36 cells) across three independent differentiations. Neurons at DIV 56–60 were used for FRAP experiments. (g) FRAP recovery curves. (h) The area under the recovery curve was significantly lower in FUS M511Nfs neurons compared to M511Nfs*Cor neurons (Unpaired two-tailed t-test, **p=0.0036, t=6.112, df=4; error bars represent s.e.m.). Each point in the bar graph represents an independent differentiation. (i) The initial slopes of the FRAP recovery curves (comprising the first 50 seconds from panel g) were determined using a non-linear regression analysis and found to be significantly slower in FUS M511Nfs neurons compared to M511Nfs*Cor neurons (Unpaired two-tailed t-test, **p=0.0033, t=6.271, df=4; error bars represent s.e.m.). For all quantifications, data distribution was assumed to be normal but this was not formally tested.

Figure 3.. Wild-type and mutant FUS interact with Nup62 in iPSCs.

(a) Left: FUS IP. Nup62 was pulled down with wild-type (P525L*Cor) and mutant (P525L) FUS. Middle: Nup62 IP. Both wild-type and mutant FUS were pulled down with Nup62. In both IPs, interactions of FUS and Nup62 were retained and significantly enhanced upon RNase treatment, suggesting the interactions were not mediated through RNA. Right: IgG and input controls for IP. Source data are provided as a Source Data file. (b) Quantification for FUS IP. Densitometry analysis of FUS and Nup62 signals in the Western blot shown in (a). Two-way ANOVA with Tukey’s multiple comparisons test. P525L*Cor: ****p<0.0001; P525L: ***p=0.0002. F (1, 8)=0.7543. (c) Quantification for Nup62 IP, as described for (b). Two-way ANOVA with Tukey’s multiple comparisons test. P525L*Cor: *p=0.0497; P525L: *p=0.0204. F (1, 8)=0.2072. Only significant comparisons were denoted. (b,c) Each point in the bar graph represents a biological replicate (n=3). Error bars represent s.e.m. Data distribution was assumed to be normal but this was not formally tested.

Figure 4.. Nup62 and FUS alter the phase separation properties of each other in a manner that is modulated by RNA in vitro.

Fluorescence images showing phase separation of 1.5 μM MBP-FUS WT (red) doped with ≤2.5% Alexa555-labelled MBP-FUS WT and 1.5 μM MBP-Nup62 (green) doped with ≤ 2.7% Alexa488-labelled MBP-Nup62. Full-length FUS and Nup62 proteins were used herein. (a) MBP-FUS forms dynamic spherical droplets (Supplemental Video 1). Co-mixing MBP-FUS with MBP-Nup62 promotes the formation of FUS/Nup62-containing amorphous assemblies that appear less dynamic (Supplemental Video 2). (b) Co-mixing as in (a) in the presence of RNA (mass ratio of RNA:FUS:Nup62 ≈ 1.5:1:1) prevents the formation of amorphous assemblies, giving way to spherical droplets MBP-FUS with some MBP-Nup62 co-localization. Images were taken at the indicated timepoints after initiation of phase separation by 3.6% PEG. Results are representative of n=4 independent experiments. Scale bar= 50 μm (inset = 5 μm).

Figure 5.. Mutant FUS interacts with F/G Nup in the cytoplasm of human neurons.

(a) Immunofluorescence images of FUS (green) and F/G Nup marker mAb414 (red) in human neurons are representative of n=3 experiments. Tuj1 (grey). DAPI (blue). Scale bar= 10 μm. (b) Proximity ligation assay (PLA) using the FUS and mAb414 antibodies in (a). (b) Maximum intensity projections of the all z-planes for cell soma (delineated with blue dotted line). (c) Quantification of (b). A significant increase of total PLA signal in FUS M511Nfs neurons compared to control neurons; ****p<0.0001. t=6.630, df=141. (d) Maximum intensity projections of the center ten optical planes, allowing for quantification of nuclear (delineated with yellow dotted line surrounding DAPI) versus cytoplasmic PLA signals detected in between the blue and yellow lines. PLA signal at the nuclear membrane is counted as “nuclear” here. (e) Quantification of (d). PLA C/N ratio revealed significantly higher cytoplasmic PLA signal in FUS M511Nfs neurons compared to control neurons; ****p<0.0001. t=8.384, df=97. The PLA C/N ratio of the chosen representative neurons: M511Nfs*Cor (0.33, 0.30); M511Nfs (1.05, 1.00, 1.42). (f-h) Similar PLA C/N analyses were performed with FUS and Nup62 (f), Nup98 (g), Nup153 (h) antibodies. ****p<0.0001. (f) t=5.240, df=112. (g) t=5.142, df=114. (h) t=6.797, df=102. (i) PLA analysis with mAb414 and Nup62 antibodies show a loss of nuclear Nup PLA signal (including the nuclear membrane) and an increase of cytoplasmic Nup PLA signal in FUS M511Nfs neurons compared to the control. PLA signal (grey); DAPI (Blue). The PLA C/N ratio of the chosen representative neurons: M511Nfs*Cor (0.49, 0.27, 0.18); M511Nfs (0.71, 1.49). (j) Quantification of (i). M511Nfs*Cor, n=69; M511Nfs, n=68 cells; **p=0.0018. t=2.832, df=135. All the images were chosen to be representative of the mean quantified value. Scale bar= 10 μm. (c,e,f,h,j) Box and whiskers plot showing all cells analyzed from three independent differentiations. Plots indicate the 25^th (above) and 75^th (below) percentiles, with a line within showing the median; whiskers above and below the box indicate the maximum and minimum values. Unpaired two-tailed t-test was used for all comparisons. Data distribution was assumed to be normal but this was not formally tested.

Figure 6.. Nuclear abnormalities associated with ALS-linked FUS are rescued by knock-down of Nup62 in vivo

(a) Immunofluorescence imaging of larval ventral nerve chord in Drosophila expressing FUS WT-RFP (red, top) and ALS mutant FUS P525L-RFP (red, bottom), with and without Nup62 knockdown. Lamin C (grey) staining was used to mark the nuclear membrane. Nuclear membrane invaginations, which appear as wrinkled membranes, were observed in P525L-expressing neurons (denoted by yellow arrowheads). Zoom images show normal nuclear membrane morphologies for all conditions except P525L. Scale bar= 10μm. (b) Quantification of percentage of neurons that have nuclear membrane defects in all four groups (Each point in the bar graph represents an animal. FUS WT, n=4; FUS WT + Nup62 RNAi, n=7; FUS P525L, n=6; FUS P525L + Nup62 RNAi n=8) reveals a significant increase in percentage of neurons with nuclear morphology defects in P525L-expressing neurons compared to WT. Upon Nup62 knockdown, there was a significant decrease in percentage of P525L-expressing neurons with nuclear morphology defects. Each point represents data from one larva. One-way ANOVA with Tukey’s multiple comparisons test was used for analysis (F-value=14.94, df=3). Only significant comparisons were denoted. WT-RFP vs P525L-RFP ****p-value<0.0001; P525L-RFP vs P525L-RFP + Nup62 RNAi ***p-value=0.0003. Error bars indicate s.e.m. (c) Quantification of nuclear FUS-P525L intensities and (d) nuclear-cytoplasmic (N/C) ratio of FUS-P525L intensities with and without Nup62 knockdown (P525L-RFP, n=29; P525L + Nup62 RNAi, n=47) indicates no difference in either nuclear levels (p-value=0.128) (c), or nuclear-cytoplasmic FUS distribution (p-value=0.145) (d) in P525L-expressing neurons upon Nup62 knockdown. Each point in (c) and (d) represent individual neurons. Unpaired two-tailed Student’s t-test was used for analysis (For [c]: t=1.539, df = 74; and for [d]: t=1.473, df = 74). Error bars indicate s.e.m. (b-d) Data distribution was assumed to be normal but this was not formally tested. (e) Post-mortem human spinal cord from two ALS-FUS cases (H517Q and R521C) compared to a neurologically normal control demonstrate abnormally misshapen nuclei (i.e., deviating from a round shape and/or containing membrane invaginations), which are denoted by yellow arrow-heads. Images are representative of FUS inclusion-bearing neurons within n=1 staining experiments. Tissues were stained with anti-FUS antibody and a nuclear dye, revealing cytoplasmic mislocalization of ALS-FUS. Top panel: immunohistochemistry. Bottom panel: confocal immunofluorescence images of a neuron from the FUS R521C case with FUS accumulation (red) proximal to the nucleus (blue). Scale bar= 10μm.

Figure 7.. Modulation of Nup expression rescues FUS-induced toxicity in vivo.

(a-c) Kaplan-Meier survival curves of flies expressing (a) FUS WT, (b) R518K mutant and (c) R521C mutants with and without Nup62 RNAi in adult motor neurons (D42-gal4). Non-transgenic (w1118) flies were used as controls. Log-Rank (Mantel-Cox) statistical test was used for analysis of survival curves. ****p-value<0.0001; FUS-WT vs FUS-WT + Nup62 RNAi p-value=0.0397; R518K vs R518K + Nup62 RNAi p-value=0.0128. Data distribution was assumed to be normal but this was not formally tested. (d) Tabular representation of number of days it took for 25%, 50% and 75% death respectively for each genotype. The number of days that Nup62 siRNA confers for survival for each genotype are shown in red. (e) Percentage of flies expressing FUS WT, R518K and R521C in motor neurons through development, with and without Nup62 RNAi, knockdown that eclosed/hatched with normal straight wings (green) or abnormal crumpled wings (red) (w1118 Control n = 99; FUS WT n = 49; FUS WT + Nup62 RNAi n = 56; R518K n=88; R518K + Nup62 RNAi n = 72; R521C n = 76; R521C + Nup62 RNAi n = 68).