Widespread FUS mislocalization is a molecular hallmark of amyotrophic lateral sclerosis

Abstract

Mutations causing amyotrophic lateral sclerosis (ALS) clearly implicate ubiquitously expressed and predominantly nuclear RNA binding proteins, which form pathological cytoplasmic inclusions in this context. However, the possibility that wild-type RNA binding proteins mislocalize without necessarily becoming constituents of cytoplasmic inclusions themselves remains relatively unexplored. We hypothesized that nuclear-to-cytoplasmic mislocalization of the RNA binding protein fused in sarcoma (FUS), in an unaggregated state, may occur more widely in ALS than previously recognized. To address this hypothesis, we analysed motor neurons from a human ALS induced-pluripotent stem cell model caused by the VCP mutation. Additionally, we examined mouse transgenic models and post-mortem tissue from human sporadic ALS cases. We report nuclear-to-cytoplasmic mislocalization of FUS in both VCP-mutation related ALS and, crucially, in sporadic ALS spinal cord tissue from multiple cases. Furthermore, we provide evidence that FUS protein binds to an aberrantly retained intron within the SFPQ transcript, which is exported from the nucleus into the cytoplasm. Collectively, these data support a model for ALS pathogenesis whereby aberrant intron retention in SFPQ transcripts contributes to FUS mislocalization through their direct interaction and nuclear export. In summary, we report widespread mislocalization of the FUS protein in ALS and propose a putative underlying mechanism for this process.

Keywords: RNA binding protein; amyotrophic lateral sclerosis (ALS); fused in sarcoma FUS; intron retention.

Figure 1

Nuclear-to-cytoplasmic mislocalization of FUS in human and mouse VCP-mutant ALS models. (A) Subcellular localization of FUS determined by immunocytochemistry in iPSCs (D0), neural precursors (NPCs; D7) and patterned motor neuron precursors (pMNs; D14). The ratio of the average intensity of the FUS immunolabelling in the nucleus (N) versus cytoplasm (C) was automatically determined in both CTRL and VCP mutant lines cells. Data shown are average N/C ratios (±SD) per field of view from four control and four VCP mutant lines. P-value from unpaired t-test with Welch’s correction. (B) Analysis of the subcellular localization of FUS in motor neurons in the ventral spinal cord of wild-type, VCP^A232E and SOD1^G93A mice. Motor neuron cytoplasm was identified by ChAT staining, nuclei were counterstained with DAPI. Images were acquired as confocal z-stacks using a Zeiss 710 confocal microscope with a z-step of 1 μm, processed to obtain a maximum intensity projection and analysed using Fiji. Motor neurons were identified by ChAT staining, and the nuclear and cytoplasmic area were manually drawn based on DAPI and ChAT staining, respectively. For each region of interest, the average FUS immunoreactivity intensity was measured, background was subtracted, and the ratio between nuclear and cytoplasmic average intensity was calculated. Data shown are nuclear/cytoplasmic (N/C) ratio (mean ± SD) per cell from four wild-type, four SOD1^G93A and three VCP^A232E mice. Scale bar = 20 μm. Linear mixed effects analysis of the relationship between FUS localization and either VCP or SOD1 mutation to account for idiosyncratic variation due to animal differences: SOD1 mutation does not affect FUS localization [2(1) = 0.6387, P = 0.4242], lowering it by about −0.8152 ± 1.1152 (standard errors), while VCP mutation does affect FUS localization [2(1) = 8.3145, P = 0.003933], lowering it by −4.2175 ± 0.9731 (standard errors).

Figure 2

Nuclear-to-cytoplasmic mislocalization of FUS in human sporadic ALS. (A) Analysis of the subcellular localization of FUS in motor neurons in the ventral spinal cord of healthy controls (n = 8) and patients with sporadic ALS (sALS) (n = 12). Motor neuron cytoplasm was identified by ChAT immunolabelling, nuclei were counterstained with DAPI. N/C ratio of FUS immunoreactivity was measured as described in Fig. 1B. Only motor neurons with a visible nucleus were considered for the analysis. We used R and lme4 (Bates et al., 2015) to perform a linear mixed effects analysis of the relationship between FUS localization and sALS disease. Scale bar = 20 μm. Data shown are N/C ratio (mean ± SD) per cell from at least eight cases per group. (B) Nuclear and cytoplasmic areas of motor neuron nuclei and cytoplasm in sALS (n = 12) versus control (n = 8). Data shown are mean ± SD. P-values from non-parametric Mann-Whitney’s test. (C) Single-cell measurement of nuclear (right) and cytoplasmic area (left) shows no correlation with FUS nuclear to cytoplasmic ratio as shown with scatter plots. PCC = Pearson correlation coefficient.

Figure 3

FUS binds to an aberrantly retained intron within the SFPQ transcript, which is exported from the nucleus. (A) Bar plots depicting the level of enrichment in FUS-binding to the retained intron compared with non-retained introns within the same gene. Analysis of the 167 previously reported as aberrantly retained introns in ALS. (B) Genome browser view of FUS iCLIP crosslinking events along the SFPQ and FUS transcript. Grey boxes highlight the location of retained introns. (C) Dot plot showing the levels of cytoplasmic intron retention (IR) SFPQ transcripts measured by qPCR. IR transcript levels were measured using an IR-specific primer pair and normalized over the expression levels of a constitutive SFPQ exon, as described previously (Luisier et al., 2018). n = 5 control lines and n = 4 VCP lines, mean ± standard error of the mean (SEM) from two independent replicates per line, each analysed in technical duplicate. P = 0.0041 from Welch’s test. [D(i)] Single molecule fluorescence in situ hybridization (smFISH) showing SFPQ intron-retaining transcript in the cytoplasm. The top panel shows the nuclear staining only for context; the probe clearly detects intron-retaining transcripts in the cytoplasm. [D(ii)] Quantification and statistical analysis of smFISH performed on three control lines and three VCP mutant lines using Welch’s test. (E) Schematic diagram of proposed model. Cartoon summarizing the functional consequence of aberrant IR in SFPQ transcript. The 9 kb intron 9 in SFPQ is retained and exported to the cytoplasm. The FUS protein binds extensively to this retained intron. The FUS protein itself is abundantly relocalized from the nucleus to the cytoplasm in diverse ALS models (human iPSC and mouse transgenic), as well as post-mortem samples from patients with sporadic ALS.