FUS is an N1- and N6-methyladenosine-binding protein

TL;DR

This study shows that the FUS protein binds to specific RNA modifications, which may contribute to neurological diseases like ALS and FTLD.

Contribution

The discovery that FUS binds to N1- and N6-methyladenosine in RNA provides a novel mechanism for FUS-related neurodegeneration.

Findings

FUS binds to methylated adenosines in CAG repeat RNA, causing cytoplasmic redistribution.

Reducing m1A and m6A levels decreases FUS-RNA co-localization in cells.

Binding to methylated RNA makes FUS less mobile in the cytosol.

Abstract

Nucleotide repeat expansions contribute to a number of neurological disorders. Mutations and augmented expression in fused in sarcoma (FUS) can result in amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration (FTLD). Here we reveal that FUS is an N1- and N6-methyladenosine (m1A- and m6A)-binding protein, where the protein interacts with the methylated adenosines in CAG repeat expansion RNA, thereby leading to the protein’s cytoplasmic redistribution in SH-SY5Y cells. We also found that ectopically expressed FUS co-localizes with CAG repeat RNA in the cytosol. This co-localization is diminished upon genetic depletion of m6A and m1A writer proteins (i.e. METTL3 and TRMT61A), pharmacological inhibition of METTL3, and ectopic overexpression of m1A and m6A eraser proteins (i.e. ALKBH3 and FTO). Moreover, binding to methylated CAG repeat RNA renders the ectopically expressed FUS protein less dynamic in cells. Together, our study underscores a critical role for m1A and m6A in enhancing FUS–RNA interaction, which results in aberrant subcellular distribution and attenuated mobility of the protein in cells. These findings unveil a novel mechanism underlying neurodegenerative disorders emanating from elevated expression of FUS and suggest targeting FUS-methylated adenosine interactions as a potential therapeutic strategy for FUS proteinopathy.

Graphical Abstract

Figures

FUS binds to m1A- and m6A-containing RNA in vitro. (A and B) Representative e;ectrospray ionization (ESI)-MS (A) and MS/MS (B) for the [M + 2H]2+ ions of a tryptic peptide of FUS, AAIDWFDGK. (C) Fluorescence anisotropy for measuring the binding affinities of FUS toward the unmodified and the corresponding m1A- and m6A-containing RNA probes used in the quantitative proteomic experiments. (D) Fluorescence anisotropy for measuring the binding affinities of recombinant FUS toward unmodified as well as m1A- and m6A-containing (CAG)7 repeat RNA. Error bars represent the standard deviation (SD) (n = 3).

FUS binds to CAG repeat RNA in SH-SY5Y cells. (A) Representative images of SH-SY5Y cells ectopically expressing FUS and CAG repeat RNA. Scale bars, 10 μm. (B) Co-localization of FUS with CAG repeat RNA in SH-SY5Y cells. (C) The m1A/rA and m6A/rA ratios in ectopically expressed CAG repeat RNAs isolated from SH-SY5Y cells (n = 3 independent biological replicates). (D) Quantification of foci area and percentage of FUS co-localized with CAG repeat RNA (n = 50). Data represent the mean ± SD. The P-values were calculated using one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test. ns, P > 0.05; *0.01 < P < 0.05; *0.0001 < P < 0.001; P < 0.0001.

1A and m6A writer and eraser proteins modulate FUS–CAG repeat RNA interactions in SH-SY5Y cells. (A) FISH and fluorescence microscopy for assessing the co-localization of FUS with CAG repeat RNAs in SH-SY5Y cells with ectopic expression of ALKBH3 and FTO, or knockdown of TRMT61A and METTL3. Scale bars, 10 μm. (B) The m1A/rA and m6A/rA ratios in SH-SY5Y cells with ectopic expression of ALKBH3 and FTO, or knockdown of TRMT61A and METTL3 (n = 3 independent biological experiments). (C) Quantification of foci area and percentage of EGFP–FUS co-localized with CAG repeat RNA (n = 50). Data in (B) and (C) represent the mean ± SD. The P-values were determined using one-way ANOVA with Tukey’s multiple comparisons test. ns, P > 0.05; *0.01 < P < 0.05; 0.001 < P < 0.01; *0.0001 < P < 0.001; P < 0.0001.

FUS binds to methylated CAG repeat RNA to form stress granules. (A) FISH and immunofluorescence microscopy for monitoring the co-localization of EGFP–FUS, CAG repeat RNAs, and endogenous G3BP1 in SH-SY5Y cells. Scale bars, 10 μm. (B) Quantification of the percentage of EGFP–FUS co-localized with G3BP1 (n = 50). (C) Quantification of the percentage of CAG repeat RNA co-localized with G3BP1 (n = 50). Data represent the mean ± SD. The P-values were determined using one-way ANOVA with Tukey’s multiple comparisons test. ns, P > 0.05; *0.01 < P < 0.05; 0.001 < P < 0.01; *0.0001 < P < 0.001; P < 0.0001.

Both m1A and m6A in CAG repeat RNA modulate the dynamic properties of FUS protein in cells. (A) FRAP images of EGFP–FUS in live cells expressing (CAG)0 or (CAG)38. Scale bars, 10 μm. (B) Recovery rate curves for FUS after bleaching (n = 6). (C) Diffusion coefficients of FUS based on data shown in (B) (n = 6). (D) FRAP images of EGFP–FUS in live cells transfected with synthetic RNA. Scale bars, 10 μm. (E) Recovery rate curves for FUS after bleaching (n = 3). (F) Diffusion coefficients of FUS determined from data shown in (E) (n = 3). The data represent the mean ± SD. The P-values were determined using one-way ANOVA with Tukey’s multiple comparisons test. ns, P > 0.05; *0.01 < P < 0.05; 0.001 < P < 0.01; * 0.0001 < P < 0.001; P < 0.0001.

Equations

$$\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} \begin{eqnarray*} {\mathrm{D\ }} = {\mathrm{\ }}\frac{{{{R}^2}}}{{4{{t}_{1/2}}}} \end{eqnarray*}\end{document}$$

$$\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} \begin{eqnarray*} {\mathrm{MF\ }} = {\mathrm{\ }}\frac{{{{F}_\infty } - {{F}_0}}}{{{{F}_{pre}} - {{F}_0}}} \end{eqnarray*}\end{document}$$

Full text

Introduction

The RNA-binding protein fused in sarcoma (FUS), also known as translocated in liposarcoma (TLS), is closely associated with neurodegenerative disorders, including amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration (FTLD) [1, 2]. Most ALS/FTLD-associated FUS mutations occur in its C-terminal proline tyrosine-nuclear localization signal (PY-NLS) sequence, which is thought to increase the protein’s tendency to mislocalize to the cytoplasm and aggregate, and reduce its ability to bind to nuclear RNAs [3]. Additionally, under various stress conditions, FUS is redistributed to the cytoplasm and forms stress granules with other RNA-binding proteins, including G3BP1 and TDP-43 [4].

In addition to missense mutations in FUS protein, overexpression of wild-type FUS in neurons is associated with ALS. For instance, some ALS patients carry mutations in the 3′-untranslated region (3′UTR) of the FUS gene, resulting in elevated expression of FUS protein [5, 6]. FUS overexpression is toxic to yeast cells, Drosophila [7, 8], and human neuronal cells, where it induces progressive motor neuron degeneration and death by activating the mitochondrial apoptosis pathway [9, 10]. Augmented levels of wild-type FUS alter nuclear functions in cells and recapitulate ALS/FTLD-like behavior in mouse models [11]. These results suggest that elevated expression of wild-type FUS can also contribute to neurodegenerative disorders, though the precise mechanisms remain unclear.

Trinucleotide repeat expansions constitute an important genetic cause of various neurological disorders, including certain forms of intellectual disability, spinocerebellar ataxia (SCAs), and Huntington’s disease (HD) [12, 13]. CAG repeat expansions are relatively common in the genome [14], and intermediate CAG repeat length expansion in the ATXN2 gene is significantly associated with an increased risk of sporadic ALS [15]. In addition, CAG repeat expansions in the first exon of the human huntingtin (HTT) gene lead to the production of huntingtin protein containing an elongated polyglutamine tract that is highly toxic to neuronal cells [16]. Moreover, certain repeat expansion RNAs can promote phase separation and sequester RNA-binding proteins, thereby disrupting their normal functions and contributing to the pathogenesis of neurological diseases [17–19].

TDP-43 (TAR DNA-binding protein 43) is an RNA-binding protein that carries a prion-like domain and plays a critical role in RNA splicing, processing, and regulation [20]. Our previous study demonstrated that TDP-43 can bind to N1-methyladenosine (m^1^A) in CAG repeat RNA, and the binding triggers the protein’s cytoplasmic redistribution, truncation, and co-localization with stress granules, thereby contributing to neurodegeneration [21]. Similar to TDP-43, FUS harbors five RNA-binding domains and has been identified as a component of stress granules in patient cells [22, 23]. Furthermore, our previous quantitative proteomics experiments led to the identification of FUS as a candidate m^1^A-binding protein [24]. In this study, we demonstrated that FUS is capable of binding directly to m^1^A- and m^6^A-containing RNA. In addition, ectopically expressed FUS interacts with m^1^A- or m^6^A-modified CAG repeat RNA in cells, where the interactions promote the cytoplasmic redistribution of ectopically expressed FUS protein and impair its dynamic properties in SH-SY5Y cells. These findings unveil a novel mechanism for neurodegenerative disorders emanating from FUS proteinopathy and suggest a potential therapeutic strategy for these diseases.

Materials and methods

Expression and purification of recombinant FUS

Plasmid for expressing recombinant 6× His-maltose-binding protein (MBP)-tagged full-length FUS protein was obtained from Addgene (#98 651). pRK7-EGFP-FUS plasmid was constructed by inserting the enhanced green fluorescent protein (EGFP) sequence into pRK7-FUS plasmid at the XhoI and NdeI restriction recognition sites. The plasmid was transformed into competent Rosetta (DE3) pLysS Escherichia coli cells, and protein expression was induced by incubating cells with 1 mM isopropyl-β-d-1-thiogalactopyranoside (Sigma) at 16°C for 16 h. The cells were subsequently harvested by centrifugation and lysed by sonication in 20 ml of ice-cold phosphate-buffered saline (PBS) containing 10% (v/v) glycerol and 1 mM phenylmethylsulfonyl fluoride (Sigma) for 15 min. The cell lysate was then centrifuged at 10 000 g for 15 min, and the supernatant was collected and filtered using a 0.45 μm syringe filter. The 6× His-MBP-tagged FUS protein was purified from the supernatant by using a HisTrap column (Cytiva), following the manufacturer’s recommended procedures. Protein purity was verified by sodium dodecylsulfate (SDS)–polyacrylamide gel electrophoresis (PAGE) analysis (Supplementary Fig. S1) and stored in PBS containing 20% (v/v) glycerol at −80°C until use.

In vitro binding assay

The RNA probes (5′-biotin-CCGUUCCGCCCXGGCCGCGCCCAGCUGGAAUGCA-3′ and 5′-biotin-CAGCAGCAGCXGCAGCAGCAG-3′, X = A, m^1^A, or m^6^A; Integrated DNA Technologies) were labeled with T4 RNA ligase 1 (NEB) and pCp-Cy3 (Jena Bioscience), following the manufacturer’s protocol. Briefly, 10 μM RNA in a 30 μl solution containing 3 μl of 10× T4 RNA ligase buffer, 3 μl of 10 mM ATP, 1 μl of 100 mM dithiothreitol (DTT), 3 μl of T4 RNA ligase (1000 U μl^−1^), and 100 μM pCp-Cy3. After a 12 h incubation, the Cy3-labeled probes were purified by using Bio-Spin P-30 columns (Bio-Rad). The Cy3-labeled probes were subsequently annealed by heating the solution at 95°C for 5 min, followed by cooling slowly to room temperature over 3 h.

Fluorescence anisotropy-based binding assay was performed following previously described procedures [25]. The aforementioned RNA probes (5 nM) were incubated, for 30 min on ice, with the indicated concentrations of recombinant FUS protein in 20 μl of binding buffer containing 10 mM Tris–HCl (pH 7.5), 1 mM EDTA, 100 mM KCl, 0.1 mM DTT, and 10 μg ml^−1^ bovine serum albumin (BSA). Fluorescence anisotropy was subsequently recorded on a BioTek Synergy H1 Multimode Reader (Agilent Technologies, La Jolla, CA, USA), with the excitation and emission wavelengths being 530 and 590 nm, respectively. The dissociation constant (Kd) values were calculated with GraphPad Prism 8 using non-linear regression for curve fitting with the one-binding-site model.

Cell culture and shRNA/siRNA knockdown

SH-SY5Y cells (ATCC) were maintained in Dulbecco’s modified Eagle’s medium (DMEM; 11995-065, Gibco) containing 10% fetal bovine serum (FBS), 100 U ml^−1^ penicillin, and 100 μg ml^−1^ streptomycin under standard cell culture conditions (37°C, 5% CO_2_). The cells were confirmed to be free of mycoplasma contamination by using the LookOut Mycoplasma PCR Detection Kit (MP0035, Sigma-Aldrich).

The small interfering RNA (siRNA) sequences for TRMT61A are listed in Supplementary Table S1. The cells cultured in 6-well plates were transfected with 100 pmol siRNAs per well for 48 h using RNAiMAX (Invitrogen) following the manufacturer’s protocol, where non-targeting siRNA (Dharmacon, D-001210–02-20) was used as control.

The sequences for short hairpin RNAs (shRNAs) against METTL3 and TRMT61A are listed in Supplementary Table S1. All shRNAs and control non-targeting shRNAs were cloned into the AgeI/EcoRI site of the pLKO.1 vector (Addgene, plasmid #10 878) and confirmed by Sanger sequencing. Cells were transfected with pLKO.1/puro-shRNAs together with pLTR-G (Addgene plasmid #17 532) envelope plasmid and pCMV-dR8.2 dvpr (Addgene plasmid #8455) package plasmid using PolyFect transfection reagent (QIAGEN). Viral particles were collected 48 h later and filtered through a 0.45 μm sterile filter. After that, the cells were transfected with lentiviral constructs expressing shRNA for 24 h, and selected with puromycin for 7 days.

For rescue experiments, cells stably expressing METTL3 or TRMT61A shRNA were transfected with expression plasmids encoding the corresponding wild-type or catalytically inactive enzyme, i.e. METTL3-D395A [26] and TRMT61A-D181A [27], using Transit-X2 (Mirus) according to the manufacturer’s instructions. The cells were harvested at the indicated time points after transfection for further experiments.

Cross-linking and immunoprecipitation followed by reverse transcription–quantitative PCR

SH-SY5Y cells were seeded in 10 cm dishes. After 24 h, the cells were transfected with 8 μg of Flag-FUS plasmid using TransIT-X2 (Mirus) following the manufacturer’s instructions and incubated at 37°C, 5% CO_2_ for another 24 h. When the confluence reached ~80%, the culture medium was removed, and the cells were washed gently with 5 ml of PBS at room temperature, and then with 6 ml of ice-cold PBS, followed by irradiation once with 150 mJ cm^−2^ in a UV-C cross-linker (Spectronics Spectrolinker, XL-1000) at 254 nm. The cells were harvested, and the cell pellets were snap-frozen in liquid nitrogen and stored at −80°C until use. For each immunoprecipitation sample, 50 μl of protein G Dynabeads (Invitrogen) were conjugated with 2 μg of Flag antibody at 4°C for 30 min. One sample was incubated with immunoglobulin G (IgG) antibody as the control. The cells were lysed in 1 ml of lysis buffer containing 50 mM Tris–HCl (pH 7.4), 100 mM NaCl, 1% IGEPAL CA-630, 0.1% SDS, 0.5% sodium deoxycholate, 1:100 (v/v) protease inhibitors at 4°C. RNase I (1.0 μl) was diluted with 100 μl of lysis buffer, and 10 μl of the resulting solution was added to the lysate together with 2 μl of Turbo DNase. RNA was digested for 3 min with shaking at 37°C and 1100 rpm. The samples were subsequently incubated on ice for 3 min. The samples were then pre-cleared by centrifugation at 21 000 g at 4°C for 10 min. The supernatants were collected for immunoprecipitation with the antibody-conjugated beads at 4°C for 2 h. The beads were washed, and the 3′ termini of the immunoprecipitated RNA were dephosphorylated by incubating with T4 PNK (NEB, with 3′ phosphatase activity) at 37°C for 20 min with rotation at 1100 rpm. An adapter sequence (pre-adenylated L3-App) was conjugated with the 3′ terminus of the RNA by using T4 RNA ligase at 16°C with rotation at 1100 rpm overnight. The RNA molecules were isolated from the beads by treating with 10 μl of proteinase K (Sigma) at 37°C for 20 min. Total RNA was extracted from the immunoprecipitates using Trizol and the relative levels of HTT, TBP, and FOS transcripts were quantified by reverse transcription–quantitative PCR (RT–qPCR) using Luna^®^ Universal qPCR Master Mix (NEB) on the CFX96 RT–qPCR detection system (Bio-rad). Primers used for RT–qPCR are listed in Supplementary Table S1. The levels of target mRNA in the CLIP (cross-linking and immunoprecipitation) samples were normalized to the input samples.

STM2457 treatment

SH-SY5Y cells were seeded in 6-well plates at 25% confluency and maintained in a humidified incubator at 37°C with 5% CO_2_. After 24 h, the cells were treated with 10 μM STM2457 and incubated for an additional 24 h [28]. The cells were subsequently harvested or processed for downstream experiments.

Western blot

METTL3 and TRMT61A knockdown cells were cultured in a 6-well plate, and the cells were harvested when their confluency levels reached 65–80%. The cells were lysed with CelLytic M cell lysis reagent (Sigma-Aldrich) and, after centrifugation, the supernatant was collected for western blot analysis. Antibodies recognizing human METTL3 (Proteintech, 67733-1-lg, 1:5 000) and TRMT61A (Thermo Fisher, A305-858A-T, 1:1 000) were used as primary antibodies for western blot analysis. Goat anti-rabbit IgG (whole molecule)–peroxidase antibody (Sigma, #A0545) and mouse (m-)IgGκ binding protein (BP)–horseradish peroxidase (HRP) (Santa Cruz Biotechnology, sc-516102) were used as secondary antibodies.

Extraction of CAG repeat mRNA and liquid chromatography–tandem mass spectrometry analysis

Total RNA was extracted from SH-SY5Y cells using TRI reagent (Sigma). The RNA sample (∼5 μg) was subsequently incubated with 200 μM 5′-biotinylated 5× CTG oligodeoxyribonucleotide in 2 volumes of hybridization buffer (50 mM Tris–HCl, pH 7.0, 750 mM NaCl, 1 mM EDTA, 1% SDS, and 15% formamide) at 37°C for 2 h, then at room temperature for 2 h. Streptavidin-conjugated agarose beads (Thermo Scientific) were added to the resulting mixture, and the suspension was incubated at 4°C for 2 h. The oligodeoxyribonucleotide-bound beads were then washed four times with 2× SSC buffer (0.30 M NaCl, 15 mM sodium citrate, pH 7.0) and the beads were resuspended in RNase-free water for digestion. The RNA was digested on-beads with 1 U of nuclease P1 in 25 μl of buffer containing 25 mM NaCl and 2.5 mM ZnCl_2_ at 37°C for 2 h. Antarctic phosphatase (0.5 U) and 3 μl of 1.0 M NH_4_HCO_3_ were subsequently added to the mixture. After incubation at 37°C for an additional 2 h, the digestion mixture was dried and reconstituted in 100 μl of H_2_O. [^13^C_5_]Adenosine, [D_3_]m^6^A, and [D_3_]m^1^A were used as internal standards for the quantifications of rA, m^6^A, and m^1^A, respectively. The enzymes in the digestion mixture were removed by extraction using chloroform:isoamyl alcohol (24:1), and salt in the samples was removed by acetonitrile precipitation. The resulting supernatant was dried, and the dried residues were reconstituted in 20 μl of H_2_O and injected for liquid chromatography–tandem mass spectrometry (LC–MS/MS) analysis on a TSQ Altis triple-quadrupole mass spectrometer (Thermo). The trapping and analytical columns were packed with porous graphitic carbon and Zorbax SB-C18 stationary phase materials, respectively. Formic acid (0.1%, v/v) in water and formic acid (0.1%, v/v) in acetonitrile were employed as mobile phases A and B, respectively, and a gradient of 0–15% B in 10 min, 15–95% B in 30 min, and 95% B in 10 min was used. The mass spectrometer was operated in the multiple-reaction monitoring (MRM) mode, where the neutral loss of a ribose from the [M + H]^+^ ions of rA, m^6^A, m^1^A, and their stable isotope-labeled counterparts was monitored. The calibration curves for the quantifications of the three nucleosides are shown in Supplementary Fig. S2.

RNA-fluorescence in situ hybridization and immunofluorescence microscopy

SH-SY5Y cells expressing the indicated RNA were fixed at 24 h following transfection and permeabilized by incubating for 10 min in pre-cooled methanol containing 10% (v/v) acetic acid. RNA was detected using a Cy3-labeled DNA probe (5′-Cy3-CTGCTGCTGCTGCTGCTGCTGCTG-3′). Hybridization and washing buffers were obtained from Biosearch Technologies and used following the manufacturer’s protocol. For immunofluorescence detection of proteins following RNA-fluorescence in situ hybridization (FISH), methanol-fixed cells were stained using an antibody against G3BP1 (Proteintech, 66486-1-lg, 1:100), and Alexa Fluor 488- or 594-labeled secondary antibodies (Invitrogen, 1:250). The samples were subsequently stained with 4′,6-diamidino-2-phenylindole (DAPI) and imaged using confocal microscopy as described above. The data were analyzed using Zen 2 Blue (version 2.3).

To quantify foci size, we randomly selected 50 foci from different cells, and measured their areas using ImageJ. The extent of co-localization of EGFP–FUS with CAG repeat expansion RNA was determined by normalizing the fluorescence intensity of EGFP–FUS that is co-localized with Cys3 signal detected in the FISH channel against the total fluorescence intensity of EGFP–FUS in each cell. To determine the degree of co-localization of EGFP–FUS with G3BP1 protein in cells expressing CAG repeat expansion RNA, we quantified the fluorescence intensity of EGFP–FUS that is co-localized with G3BP1 granules and normalized that to the total fluorescence intensity of the EGFP–FUS in each cell. Likewise, the extent of co-localization of CAG repeat expansion RNA with G3BP1 was determined by normalizing the fluorescence intensity of Cys3 signal (in the FISH channel) in G3BP1 aggregates to the total fluorescence intensity of Cys3 signal in each cell. ImageJ was used for all the quantification, and a total of 50 cells were randomly selected for each co-localization analysis.

Fluorescence recovery after photobleaching assay

Fluorescence recovery after photobleaching (FRAP) experiments were performed using an LSM 880 laser scanning confocal microscope (Zeiss 880 Inverted Airyscan Fast) coupled with a temperature-, humidity-, and CO_2_-controlled top-stage incubator for live-cell imaging. Photobleaching was conducted using a 488 nm line from an Argon laser at 50% power. Three regions of interest (ROIs) were defined for these experiments: ROI-1 was the photobleached circular region in the droplet; ROI-2 was a circular, unphotobleached region of similar size in the same droplet; and ROI-3 was defined as background and drawn outside of the droplet, where its signal was subtracted from those of ROI-1 and ROI-2. Raw data were processed and plotted using Prism.

The diffusion coefficient (D) was determined by using the following equation:

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} \begin{eqnarray*} {\mathrm{D\ }} = {\mathrm{\ }}\frac{{{{R}^2}}}{{4{{t}_{1/2}}}} \end{eqnarray*}\end{document}

R is the radius of the bleached area, t1/2 is the time required for a bleached spot to recover half between initial and steady-state fluorescence intensities [29].

The mobile fraction (MF) was calculated as:

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} \begin{eqnarray*} {\mathrm{MF\ }} = {\mathrm{\ }}\frac{{{{F}_\infty } - {{F}_0}}}{{{{F}_{pre}} - {{F}_0}}} \end{eqnarray*}\end{document}

F pre is the fluorescence intensity before photobleaching, F0 is the intensity immediately after bleaching, and F∞ is the plateau intensity after recovery.

Results

FUS is an m1A- and m6A-binding protein

A previous stable isotope labeling by amino acids in cell culture (SILAC)-based affinity screening led to the identification of FUS as a candidate m^1^A-binding protein (Fig. 1A, B) [24]. We first investigated whether FUS can bind directly to m^1^A- and m^6^A-harboring RNA by conducting fluorescence anisotropy analyses. The results showed that recombinant FUS exhibits stronger binding to m^1^A- and m^6^A-containing RNA probes than the control RNA probe used in the quantitative proteomic experiments. The dissociation constant (K_d_) values were 3.1 ± 0.5, 7.6 ± 1.7, and 28.8 ± 5.2 nM for the m^1^A-, m^6^A-, and the corresponding unmodified A-containing probes, respectively (Fig. 1C). Likewise, FUS exhibits preferential binding to m^1^A- and m^6^A-containing CAG repeat RNAs over their unmodified counterpart, with the Kd values being 6.7 ± 1.0, 9.1 ± 1.5, and 26.6 ± 2.1 nM, respectively (Fig. 1D).

FUS binds to m1A- and m6A-containing RNA in vitro. (A and B) Representative e;ectrospray ionization (ESI)-MS (A) and MS/MS (B) for the [M + 2H]2+ ions of a tryptic peptide of FUS, AAIDWFDGK. (C) Fluorescence anisotropy for measuring the binding affinities of FUS toward the unmodified and the corresponding m1A- and m6A-containing RNA probes used in the quantitative proteomic experiments. (D) Fluorescence anisotropy for measuring the binding affinities of recombinant FUS toward unmodified as well as m1A- and m6A-containing (CAG)7 repeat RNA. Error bars represent the standard deviation (SD) (n = 3).

Methylated CAG repeat RNA interacts with FUS in cells

We next asked whether FUS binds to m^1^A- and m^6^A-containing RNA in SH-SY5Y cells, and how the interactions modulate the biochemical and biophysical properties of FUS. We found that, in the absence of CAG repeat RNA expression, ectopically expressed EGFP–FUS is evenly distributed throughout the cell (Fig. 2A). In SH-SY5Y cells with ectopic co-expression of EGFP–FUS and (CAG)22 or (CAG)38 RNA, we observed a co-localization of EGFP–FUS with the CAG repeat RNA in the cytoplasm, and the extents of co-localization increase with the length of the CAG repeat RNA, as revealed by RNA-FISH and fluorescence microscopy analyses (Fig. 2B). We also examined whether the expression of CAG repeat RNA could alter the subcellular distribution of endogenous FUS protein. Our immunofluorescence experiment showed that, in SH-SY5Y cells, endogenous FUS is primarily distributed in the nuclei, and the subcellular distribution of FUS is not perturbed by ectopic expression of (CAG)38 RNA (Supplementary Fig. S3).

*FUS binds to CAG repeat RNA in SH-SY5Y cells. (A) Representative images of SH-SY5Y cells ectopically expressing FUS and CAG repeat RNA. Scale bars, 10 μm. (B) Co-localization of FUS with CAG repeat RNA in SH-SY5Y cells. (C) The m1A/rA and m6A/rA ratios in ectopically expressed CAG repeat RNAs isolated from SH-SY5Y cells (n = 3 independent biological replicates). (D) Quantification of foci area and percentage of FUS co-localized with CAG repeat RNA (n = 50). Data represent the mean ± SD. The P-values were calculated using one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test. ns, P > 0.05; *0.01 < P < 0.05; ***0.0001 < P < 0.001; ***P < 0.0001.

We observed previously that m^1^A levels in CAG repeat RNA increase with CAG repeat length in HEK293T cells [21]. We asked whether a similar observation can be made for SH-SY5Y cells. Our LC–MS/MS results showed that the m^1^A level in (CAG)38 RNA is higher than that in (CAG)22 RNA, whereas the m^6^A levels are similar in (CAG)22 and (CAG)38 RNA (Fig. 2C).

To further substantiate the impact of m^1^A and m^6^A modifications in CAG repeat RNA on foci formation and CAG repeat RNA–FUS protein interaction, we transfected SH-SY5Y cells with synthetic (CAG)7 RNA carrying no modification, or with the central adenosine being replaced completely with m^1^A or m^6^A. Our results showed that the size of the foci and co-localization of EGFP–FUS with the m^1^A- and m^6^A-containing CAG repeat RNA were significantly higher than those with the corresponding unmodified RNA (Fig. 2D). This result suggests that m^1^A and m^6^A modifications promote the interactions of RNA with FUS in cells, and this interaction stimulates the formation of FUS granules in the cytosol. In this vein, Yoneda et al. [30] showed, by using RNA pull-down and western blot analysis, that m^6^A-modified RNA binds more strongly to FUS than does unmodified RNA. Our results are consistent with the previous observation, and further substantiate that FUS exhibits preferential binding not only to m^6^A-, but also to m^1^A-carrying RNA over their unmodified counterpart.

m1A and m6A writer and eraser proteins modulate FUS–CAG repeat RNA interactions

We next asked how the co-localization of FUS with CAG repeat RNA is modulated by m^1^A and m^6^A writer and eraser proteins. We knocked down the m^1^A and m^6^A “writer” proteins (i.e. TRMT61A and METTL3, respectively), and overexpressed wild-type and catalytically inactive forms of the m^1^A and m^6^A “eraser” proteins (i.e. ALKBH3 and FTO, respectively) in SH-SY5Y cells (Supplementary Fig. S4). Our LC–MS/MS data showed that the levels of m^1^A and m^6^A in CAG repeat RNA were significantly diminished in cells depleted of TRMT61A and METTL3, respectively (Fig. 3B). Overexpression of wild-type forms of the m^1^A and m^6^A “eraser” proteins, but not their catalytically inactive counterparts, led to reduced levels of m^1^A and m^6^A, respectively, in CAG repeat RNA. Similar to what we observed for HEK293T cells [21], the m^6^A level in CAG repeat RNA also decreased in SH-SY5Y cells overexpressing wild-type ALKBH3, suggesting that ALKBH3 is also capable of removing m^6^A from CAG repeat RNA in SH-SY5Y cells (Fig. 3B).

*1A and m6A writer and eraser proteins modulate FUS–CAG repeat RNA interactions in SH-SY5Y cells. (A) FISH and fluorescence microscopy for assessing the co-localization of FUS with CAG repeat RNAs in SH-SY5Y cells with ectopic expression of ALKBH3 and FTO, or knockdown of TRMT61A and METTL3. Scale bars, 10 μm. (B) The m1A/rA and m6A/rA ratios in SH-SY5Y cells with ectopic expression of ALKBH3 and FTO, or knockdown of TRMT61A and METTL3 (n = 3 independent biological experiments). (C) Quantification of foci area and percentage of EGFP–FUS co-localized with CAG repeat RNA (n = 50). Data in (B) and (C) represent the mean ± SD. The P-values were determined using one-way ANOVA with Tukey’s multiple comparisons test. ns, P > 0.05; *0.01 < P < 0.05; **0.001 < P < 0.01; ***0.0001 < P < 0.001; ***P < 0.0001.

We next employed RNA-FISH and fluorescence microscopy assays to assess how genetic manipulations of RNA-modifying enzymes influence the interaction between FUS and CAG repeat RNA. The size and proportion of FUS foci that are co-localized with CAG repeat RNA were significantly reduced in cells upon METTL3 depletion or overexpression of wild-type FTO, compared with cells overexpressing the catalytically inactive FTO mutant (Fig. 3A–C), where His231 and Asp233 in FTO are both mutated to Ala [31]. Moreover, the size of FUS foci and percentage of FUS co-localized with CAG repeat RNA were significantly diminished in cells treated with STM2457, a METTL3 inhibitor [28], compared with those treated with dimethylsulfoxide (DMSO) (Supplementary Fig. S5).

A recent study showed that m^6^A reduction can relieve FUS-associated ALS granules, but did not demonstrate a direct interaction between FUS and m^6^A-modified RNA [32]. Our findings are in keeping with the observations made in that study, and further document that FUS recognizes m^6^A-modified CAG repeat RNA and is recruited by the RNA to form granules in cells, a process that is regulated by m^6^A writer and eraser proteins, i.e. METTL3 and FTO.

The m^1^A level in CAG repeat RNA is also critical for the interaction between FUS and the CAG repeat RNA, as manifested by attenuated co-localizations of FUS with CAG RNA in cells depleted of TRMT61A, or overexpressing wild-type ALKBH3, but not its catalytically inactive mutant [21] (Fig. 3A–C). In addition, our RT–qPCR results revealed that genetic depletion of METTL3 or TRMT61A did not result in a diminished level of (CAG)38 RNA (data not shown), excluding the possibility that the attenuated FUS–CAG repeat RNA interactions detected in the knockdown cells are due to diminished expression of the repeat RNA.

To further determine whether the catalytic activities of METTL3 and TRMT61A are required for the aforementioned effects on CAG repeat RNA–FUS interactions, we conducted rescue experiments by re-expressing wild-type or catalytically inactive METTL3 and TRMT61A in the corresponding knockdown cells. Our fluorescence microscopy and RNA-FISH experiments revealed that reconstitution with the wild-type enzymes results in augmented co-localizations of EGFP–FUS with CAG repeat RNA, whereas the catalytically dead enzymes fail to do so (Supplementary Fig. S6A, B). In this vein, our LC–MS/MS analysis confirmed that only the wild-type TRMT61A and METTL3, but not their mutant counterparts, restored m^1^A and m^6^A levels in CAG repeat RNA, respectively (Supplementary Fig. S6C, D).

FUS interacts with endogenous m1A- and m6A-containing transcripts

We next performed CLIP-qPCR to assess the interactions of FUS with endogenous CAG repeat-containing mRNAs, as well as other transcripts harboring known m^6^A modification sites. In this vein, both HTT and TBP genes contain endogenous CAG repeat tracts in their coding regions and are widely used for examining CAG repeat-associated molecular mechanisms [33, 34]. We observed a significant reduction in the enrichment of TBP mRNA in FUS CLIP samples prepared from TRMT61A-depleted cells relative to control shRNA-treated cells (Supplementary Fig. S7A). Likewise, the enrichment of HTT mRNA was markedly decreased upon genetic depletion of TRMT61A or METTL3 (Supplementary Fig. S7B). Similarly, the enrichment of FOS mRNA, which is known to carry m^6^A modification [35, 36], was significantly diminished in the FUS CLIP samples generated from cells depleted of METTL3 (Supplementary Fig. S7C). Together, these results support that FUS interacts with endogenous m^1^A- and m^6^A-containing transcripts in cells.

m1A and m6A drive FUS to stress granules

Stress granules are phase-separated compartments in the cytosol containing RNAs and RNA-binding proteins [37]. We next examined whether FUS–CAG repeat RNA interactions contribute to the formation of stress granules in cells. In the absence of CAG repeat RNA expression, G3BP1 and FUS were evenly distributed in cells. In SH-SY5Y cells expressing (CAG)22 or (CAG)38 RNA, a fraction of endogenous G3BP1 formed granules that are co-localized with FUS and CAG repeat RNA (Fig. 4A, B). In addition, EGFP–FUS and CAG repeat RNA exhibit higher degrees of co-localization with G3BP1 granules in cells expressing (CAG)38 than those expressing (CAG)22 RNA (Fig. 4). Notably, genetic depletion of TRMT61A or METTL3 in SH-SY5Y cells with ectopic expression of (CAG)38 RNA led to diminished EGFP–FUS and CAG repeat RNA in stress granules, indicating that m^1^A and m^6^A in (CAG)38 RNA promote the recruitment of FUS to stress granules (Fig. 4; Supplementary Figs. S8 and S9).

*FUS binds to methylated CAG repeat RNA to form stress granules. (A) FISH and immunofluorescence microscopy for monitoring the co-localization of EGFP–FUS, CAG repeat RNAs, and endogenous G3BP1 in SH-SY5Y cells. Scale bars, 10 μm. (B) Quantification of the percentage of EGFP–FUS co-localized with G3BP1 (n = 50). (C) Quantification of the percentage of CAG repeat RNA co-localized with G3BP1 (n = 50). Data represent the mean ± SD. The P-values were determined using one-way ANOVA with Tukey’s multiple comparisons test. ns, P > 0.05; *0.01 < P < 0.05; **0.001 < P < 0.01; ***0.0001 < P < 0.001; ***P < 0.0001.

We obtained similar results for SH-SY5Y cells transfected with synthetic CAG repeat RNAs. The above-mentioned synthetic m^1^A- and m^6^A-contianing CAG repeat RNAs elicited greater recruitment of FUS into G3BP1-positive stress granules compared with the unmodified control RNA (Supplementary Fig. S10), further supporting a role for m^1^A and m^6^A in facilitating the partition of FUS into stress granules.

m1A and m6A alter dynamic properties of FUS

We next asked whether m^1^A and m^6^A in CAG repeat expansion RNA modulate the dynamic properties of FUS protein. To this end, we performed FRAP experiments on EGFP–FUS in cells with and without expression of CAG repeat expansion RNA (Fig. 5A). Our results showed that the extent of fluorescence recovery and the diffusion coefficient of EGFP–FUS were significantly lower in cells expressing (CAG)38 compared with those expressing (CAG)0 (Fig. 5B). Depletion of TRMT61A or METTL3, however, restored fluorescence recovery and the diffusion coefficient of EGFP–FUS in cells with expression of (CAG)38 RNA (Fig. 5B, C). Moreover, overexpression of wild-type ALKBH3 and FTO, but not their catalytically inactive mutants, restored fluorescence recovery of FUS to a greater extent than their catalytically inactive counterparts (Supplementary Fig. S11). These results indicate that CAG repeat expansion RNA can render the FUS protein less dynamic, and this process is promoted by m^1^A and m^6^A in the repeat RNA.

*Both m1A and m6A in CAG repeat RNA modulate the dynamic properties of FUS protein in cells. (A) FRAP images of EGFP–FUS in live cells expressing (CAG)0 or (CAG)38. Scale bars, 10 μm. (B) Recovery rate curves for FUS after bleaching (n = 6). (C) Diffusion coefficients of FUS based on data shown in (B) (n = 6). (D) FRAP images of EGFP–FUS in live cells transfected with synthetic RNA. Scale bars, 10 μm. (E) Recovery rate curves for FUS after bleaching (n = 3). (F) Diffusion coefficients of FUS determined from data shown in (E) (n = 3). The data represent the mean ± SD. The P-values were determined using one-way ANOVA with Tukey’s multiple comparisons test. ns, P > 0.05; *0.01 < P < 0.05; **0.001 < P < 0.01; *** 0.0001 < P < 0.001; ***P < 0.0001.

We also transfected cells with synthetic RNAs to assess the effects of m^1^A and m^6^A in CAG repeat RNA on modulating the biophysical properties of FUS. We observed that the fluorescence from EGFP–FUS recovered more slowly after photobleaching when cells were transfected with synthetic CAG repeat RNA harboring an m^1^A or m^6^A modification than those transfected with the corresponding unmodified RNA (Fig. 5E). Furthermore, the diffusion coefficient of FUS in SH-SY5Y cells was markedly attenuated upon transfection with m^1^A- or m^6^A-modified RNA (Fig. 5F). This result demonstrates the direct effects of m^1^A and m^6^A in CAG repeat RNA on modulating the dynamic properties of FUS.

Discussion

Mutations in the PY-NLS domain of FUS result in its cytoplasmic mislocalization in cells and alter its binding affinity toward RNA [38, 39]. Multiple mutations in the NLS domain of FUS have been identified in both familial and sporadic ALS patients [40]. Furthermore, elevated levels of wild-type FUS have been observed in some ALS patients [5, 6], where overexpression of FUS can also result in its mislocalization in cells and give rise to neurodegeneration [11].

Protein aggregation in motor neurons has been linked to neurodegenerative disorders [41]. RNA-binding proteins, e.g. TDP-43 and FUS, can form granules in cells, leading to ALS/FTLD-like behaviors in various organisms [42, 43]. Our previous study revealed a length-dependent elevation in the level of m^1^A in CAG repeat expansion RNA, where the m^1^A can enhance the binding of the RNA with TDP-43, promoting its mislocalization, truncation, and formation of gel-like aggregates [21]. These results support that RNA modification plays a crucial role in the formation of toxic neuronal granules.

Several studies suggest that FUS may function as an m^6^A reader protein [30, 32, 44]; however, it remained unclear whether the protein is capable of binding directly with m^6^A in RNA. Here, we reveal that recombinant FUS protein is a methylated adenosine reader protein, where it binds preferentially to both m^1^A and m^6^A in RNA (Fig. 1). This property of FUS differs from that of TDP-43, which can only recognize m^1^A, but not m^6^A in RNA [21]. We also observe that m^1^A and m^6^A in ectopically expressed CAG repeat RNA and endogenous mRNAs promote their interactions with FUS, and these interactions elicit cytoplasmic redistribution of ectopically expressed FUS (Figs. 2 and 3). Moreover, we demonstrate that elevated levels of m^1^A and m^6^A in CAG repeat RNA rendered FUS protein less dynamic in cells and promoted stress granule formation, as manifested by the observations made for cells with ectopic overexpression of the m^1^A and m^6^A eraser proteins (i.e. ALKBH3 and FTO), with genetic depletion of their writer proteins (i.e. TRMT61A and METTL3), with pharmacological inhibition of METTL3, and with transfection of synthetic CAG repeat RNA with or without m^1^A and m^6^A (Figs. 4 and 5; Supplementary Figs. S5–S10). These findings suggest that the epitranscriptomic m^1^A and m^6^A modifications may exert their neurotoxic effects partly through binding with FUS protein. On the grounds that modulating the writer and eraser enzymes of m^1^A or m^6^A alone could influence the FUS–CAG repeat RNA interaction and affect the subcellular distribution as well as the dynamic properties of ectopically expressed FUS protein, our results support that these two methylated adenosine modifications assume non-redundant functions in regulating the biophysical properties of FUS protein.

It is worth noting that the aforementioned observations were made with ectopically expressed FUS, and cytoplasmic redistribution elicited by CAG repeat expansion RNA was not detectable for endogenous FUS (Supplementary Fig. S3). Nevertheless, overexpression of FUS is known to contribute to ALS [5, 6]. In addition, FUS overexpression is toxic to human neuronal cells, where it triggers motor neuron degeneration and death by activating the mitochondrial apoptosis pathway [9, 10]. Elevated expression of wild-type FUS alters nuclear functions in cells and elicits ALS/FTLD-like behavior in mice [11]. On the basis of our findings, treatment with METTL3 and TRMT61A inhibitors could help reduce the aggregation and cytoplasmic redistribution of FUS protein in neuronal cells, providing potential benefits for ALS/FTLD cases associated with FUS overexpression.

Supplementary Material

gkag194_Supplemental_File