The motor neuron m6A repertoire governs neuronal homeostasis and FTO inhibition mitigates ALS symptom manifestation

Abstract

Amyotrophic lateral sclerosis (ALS) is a swiftly progressive and fatal neurodegenerative ailment marked by the degenerative motor neurons (MNs). Why MNs are specifically susceptible in predominantly sporadic cases remains enigmatic. Here, we demonstrated N⁶-methyladenosine (m⁶A), an RNA modification catalyzed by the METTL3/METTL14 methyltransferase complex, as a pivotal contributor to ALS pathogenesis. By conditional knockout Mettl14 in murine MNs, we recapitulate almost the full spectrum of ALS disease characteristics. Mechanistically, pervasive m⁶A hypomethylation triggers dysregulated expression of high-risk genes associated with ALS and an unforeseen reduction of chromatin accessibility in MNs. Additionally, we observed diminished m⁶A levels in induced pluripotent stem cell derived MNs (iPSC~MNs) from familial and sporadic ALS patients. Restoring m⁶A equilibrium via a small molecule or gene therapy significantly preserves MNs from degeneration and mitigates motor impairments in ALS iPSC~MNs and murine models. Our study presents a substantial stride towards identifying pioneering efficacious ALS therapies via RNA modifications.

Fig. 1. m⁶A RNA modification levels are downregulated in ALS; loss of m⁶A methyltransferases METTL3  and METTL14 leads to neurodegeneration.

a Timeline of MN differentiation in control (Ctrl) and ALS (SOD1^+/L144F, C9ORF72^{exp~800 G4C2}, TDP43^G298S) iPSC lines. CPA (cyclopiazonic acid) was applied to accelerate MN degeneration (annotated as basal time point day 0). NF neurotrophic factors. b–d MN degeneration index and m⁶A RNA methylation levels of Ctrl and ALS iPSC~MNs. m⁶A percentages of Ctrl/ALS types at indicated time points were normalized to those at day 0. ALS iPSC~MNs undergo dramatic degeneration from day 7 to day 21 post-CPA treatment, while a significant reduction in m⁶A level was already exhibited at day 4. The degeneration index measures the neurite fragmentation. e–g Quantification of the m⁶A ratio in mRNAs from day 4 post-CPA treatment of Ctrl and ALS (SOD1^+/L144F, C9ORF72^{exp~800 G4C2}, TDP43^G298S) iPSC lines. The mRNAs were extracted by poly(A) purification. The panels at the right show representative images of an m⁶A dot blot and methylene blue staining (for loading controls) (b–g: n = 3 independent experiments). h Timeline of STM2457 (20 μM)-mediated inhibition of m⁶A modification during MN differentiation of wild-type iPSC lines. STM2457 was applied for 6 days to accelerate MN degeneration. i and j Inhibition of METTL3-mediated m⁶A modification in human wild-type iPSC~MNs results in a sharp decline in m⁶A levels after four days of treatment for STM2457, as assayed by m⁶A ELISA in mRNA (i) and mRNA m⁶A dot blot (j) (i, j: n = 3 independent experiments). m⁶A methylation was normalized to the vehicle (DMSO), which served as a non-stressed control. Note that dramatic neurite degeneration was observed on day 6 in k upon STM2457 treatment. Scale bar, 200 µm. l Quantification of the degeneration index at an indicated time point normalized to the vehicle control (n = 6 independent experiments). Illustrations in a and h created in BioRender. Chen, J. (2025) https://BioRender.com/3nizfu7. All Data are presented as mean ± SD, significant P values from two-tailed t-tests. N.S. non-significant. Source data are provided as a Source data file.

Fig. 2. Phenotypic characterization of ChAT-Cre; Mettl14^floxed mutant mice.

a Body weight of male and female ChAT-Cre; Mettl14^floxed and littermate control mice. Data are presented as mean ± SD. b Kaplan–Meier survival curves reflect that both male and female ChAT-Cre; Mettl14^floxed mice die prematurely compared to littermate controls. Immunostaining (c) and quantification (d) of lumbar ChAT^on MN numbers reveal a gradual reduction starting after P100 and a significant loss of ChAT-Cre; Mettl14^floxed MNs at P160 (Data are presented as mean ± SD, with significant P values from two-tailed t-tests. NS non-significant. n = 3 in ChAT-Cre; Mettl14^floxed and control mice at P30, P70, P160, and P250, respectively; n = 5 in ChAT-Cre; Mettl14^floxed mice and n = 4 in control mice at P100, Scale bar, 100 µm). Preferential loss of the cholinergic C-bouton nerve terminals of MNs in the ChAT-Cre; Mettl14^floxed mice from P70 (e), with respective quantification in (f). Scale bar, 50 µm. Data are presented as mean ± SD, n = 3 mice in ChAT-Cre; Mettl14^floxed and control at P30, P100, P160, and P250, respectively; n = 5 in ChAT-Cre; Mettl14^floxed mice and n = 3 in control mice at P70, with significant P values from two-tailed t-tests. N.S. non-significant. Source data are provided as a Source data file.

Fig. 3. Molecular characterization of ChAT-Cre; Mettl14^floxed mutant mice.

Images illustrate microglial activation, as determined by immunostaining for Iba1 (a), and quantification (b) of lumbar Iba1^on numbers at the ventral region, revealing significant microglial activation in ChAT-Cre; Mettl14^floxed mice compared to littermate controls. Staining was repeated on n = 5 mice. c Tdp43 (green) is localized in the nucleus of the MNs of control mice. In the ChAT-Cre; Mettl14^floxed mutant mice, numerous Tdp43 inclusions exist in the cytoplasm (arrow). High magnifications of the highlighted Tdp43 aggregates in MNs are shown in the rightmost panels. Respective quantification is presented in d, (n = 3 mice). e, g Representative z-stack confocal images of neuromuscular junctions (NMJs) in gastrocnemius (GA) muscles dissected from P150 ChAT-Cre; Mettl14^floxed and littermate control mice. Motor nerves were visualized using a combination of SV2/NF(2H3) (green) and post-synaptic AChRs with α-BTX (magenta). Arrowheads identify denervated synapses, abnormal axonal swellings (e), and smaller endplates (g). f, h Quantification of the denervation ratio of NMJs and endplate area from (e, g). (n = 3 mice, quantification for all NMJs from all views of captured images). Scale bar, 100 µm. Data are presented as mean ± SD with significant P values from two-tailed t-tests. Source data are provided as a Source data file.

Fig. 4. m⁶A-deficient mice display motor deficits that recapitulate ALS.

a Schematic illustration of the behavioral tests conducted to assess motor functions. Created in BioRender. Chen, J. (2025) https://BioRender.com/mrtnwlk. Locomotor coordination on an accelerating rotarod is displayed as the rotation speed at which mice fell off (b) and the latency to fall (c). (Ctrl: n = 3/3, 8/9, 5/10, 11/10, and 10/10 mice; ChAT-Cre; Mettl14^floxed: n = 3/3, 7/11, 7/15, 9/12, and 8/10 mice at P40, P70, P100, P130, and P160 from males/females, respectively). There was a significant decrease in locomotor activity at P70 and thereafter for males and at P100 and thereafter for females. d Forelimb grip strengths for ChAT-Cre; Mettl14^floxed male and female mutant mice. (Ctrl: n = 3/6, 12/12, 12/14, 12/10, and 6/10 mice and ChAT-Cre; Mettl14^floxed: n = 3/6, 10/14, 10/17, 8/12, and 7/9 mice at P40, P70, P100, P130, and P160 from males/females, respectively). e Travel pathways (red) of representative trajectory diagrams filmed for 10 and 60 minutes in the open field test arena (square perimeter) for the early-onset (P70) and disease progression (P160) stages of ChAT-Cre; Mettl14^floxed and littermate control mice. f Total distance traveled in the open field test. (Ctrl: n = 3/3, 7/9, 6/8, 8/9, and 8/9 mice; ChAT-Cre; Mettl14^floxed: n = 3/3, 6/11, 6/11, 7/12, and 7/9 mice at P40, P70, P100, P130, and P160 from males/females, respectively). g Schematic illustration of the behavioral tests from the treadmill conducted to assess motor function. Created in BioRender. Chen, J. (2025) https://BioRender.com/mrtnwlk. h, i Stride width (usually mediated by INs) is not compromised, whereas stride length (mediated by MNs) is drastically reduced in the ChAT-Cre; Mettl14^floxed mice (Speed = 15 cm/s). n = 6 mice (3 from P180 and 3 from > P210 mice). Data are presented as mean ± SD, significant P values from two-tailed t-tests. N.S. non-significant. Source data are provided as a Source data file.

Fig. 5. Nanopore direct RNA-seq identifies ALS risk genes as m⁶A modified.

a Timeline of in vitro differentiation and maturation from mESC~MNs. Created in BioRender. Chen, J. (2025) https://BioRender.com/xap28jf. b The heatmap from qPCR verification shows that postnatal upregulation of functionally relevant genes. Slc5a7 serves as a positive control as it is constantly expressed after the postmitotic stage. c Upper panel: Immunostaining of Syn1 in Day 7 and Day 12 mESC~MNs. Smi32 labels the MNs and neurites. Scale bar, 100 µm. n = 3 independent experiments. All quantification information is provided in the Methods and Source data file. Lower panel: Quantifications reveal a significant increase in neurite thickness (left), neurite complexity (middle), and mature structure revealed by the puncta number (right) in the Day 12 mature mESC~MNs. d Overview of the experimental and analysis workflow for conducting Nanopore direct RNA-seq on mature mESC~MNs. Created in BioRender. Chen, J. (2025) https://BioRender.com/cq89p4h. e The heatmap from Nanopore direct RNA-seqshows that several feature adult genes (ChAT, Spp1, Bag, Slc5a7, Fos, and Syn1) are more enriched in the Day12 mature mESC~MNs compared to embryonic genes (Isl1, Olig2, and Mnx1). f Motif preference of m⁶A peaks identifies the DRACH consensus motif (D = A, T, or G, R = A or G, and H = A, T, or C). Metagene profile of enrichment of m⁶A-modified sites across the mRNA transcriptome. 5’UTR, 5’ untranslated region; CDS, coding sequence; 3′UTR, 3′ untranslated region. Replicates 1, 2, and 3 (R1, R2, and R3) represent the triplicate biological repeats. g KEGG pathway analysis of the m⁶A-modified MN epitranscriptome reveals distinct biological pathways related to neurodegenerative diseases. Terms of interest in this study are highlighted in bold purple. h Schematic for analyzing the m⁶A-modified MN epitranscriptome, showing that 41.98% of m⁶A-modified genes are ALS risk genes. i, j The verification of the predicted m⁶A-modified sites in Tardbp and Atp13a2 by m⁶A pull-down qPCR of selected high m⁶A-modified sites and low m⁶A-modified sites. Points represent individual biological experiments. All data are presented as mean ± SD, n = 3, with significant P values from two-tailed t-tests. N.S. non-significant. Source data are provided as a Source data file.

Fig. 6. Identification of m⁶A-modified genes contributing to MN degeneration.

a Overview of the experimental workflow for single nucleus multiomics. Created in BioRender. Chen, J. (2025) https://BioRender.com/esjho7r. b Immunostaining showing localization of Sun1-sfGFP-myc in MNs that carry R26-CAG-LSL-Sun1-sfGFP-myc together with a Cre driver. Scale bar, 200 µm. The staining experiment was independently repeated. (n = 3 mice). c Uniform manifold approximation and projection (UMAP) representation of all nuclei that passed quality filtering. Dimensionality reduction and clustering were performed based on gene expression (RNA, left), chromatin accessibility (ATAC, middle), and weighted nearest neighbor (WNN) integration of RNA and ATAC data (right). Clusters are color-coded and annotated using label transfer prediction, referencing Blum et al., 2021. d Major cell type proportions are unaffected at P100~P120 in Sun1^sfGFP; ChAT-Cre; Mettl14^floxed mice, a stage before MN degeneration. e Schematic for cross-referencing DEGs, particularly those down-regulated in Sun1^sfGFP; ChAT-Cre; Mettl14^floxed MNs, and the m⁶A-modified MN epitranscriptome. The resulting data reveals distinct biological pathways (Gene Ontology, right) and KEGG (f) that might cause MN degeneration in the Sun1^sfGFP; ChAT-Cre; Mettl14^floxed mice. Terms of interest in this study are highlighted in bold and purple. Schematic for analyzing the dot-plot data (g), with the outcome (h) showing ALS disease risk genes displaying significant changes in expression in Sun1^sfGFP; ChAT-Cre; Mettl14^floxed mice in each cholinergic neuronal subtype.

Fig. 7. Increase of repressive histone modification marks and closed chromatin regions in ChAT-Cre; Mettl14^floxed mice.

a Schematic for cross-referencing of DEGs, particularly those up-regulated in Sun1^sfGFP; ChAT-Cre; Mettl14^floxed MNs, and the m⁶A-modified MN epitranscriptome, with the outcome revealing distinct biological pathways (Gene Ontology, right) that might increase repressive histone modification and the DNA damage response (highlighted in bold purple) in the Sun1^sfGFP; ChAT-Cre; Mettl14^floxed mice. b–e Representative images illustrate a dramatic increase in repressive H3K9me3 mark (b) and DNA damage γH2AX (d) signals. Quantifications of lumbar H3K9me3^on (c) and lumbar γH2AX^on (e) signal intensities in the ventral regions from the spinal cord of P120 ChAT-Cre; Mettl14^floxed mice compared to littermate controls (Ctrl: n = 5 mice, ChAT-Cre; Mettl14^floxed: n = 6 mice, quantification for all MN nuclei from all views of captured images; Data are presented as mean ± SD with significant P values from two-tailed t-tests; Scale bars, 20 µm). f The bar plot shows changes in the number of peaks and distribution of their annotated genomic locations in cholinergic neuronal subtypes derived from Sun1^sfGFP; ChAT-Cre; Mettl14^floxed mice and control (Ctrl) snATAC-seq data. Source data are provided as a Source data file.

Fig. 8. An m⁶A eraser inhibitor efficiently rescues human ALS iPSC~MNs from premature death by restoring dysregulated genes caused by hypo-m⁶A.

a Schematic illustration of the m⁶A biogenesis pathway and the applied FTO inhibitor (FB23-2) with their corresponding targeting pathways. Created in BioRender. Chen, J. (2025) https://BioRender.com/3nizfu7. b Representative images of FB23-2 rescuing the MN degeneration associated with ALS. Scale bar, 200 µm. Quantifications of m⁶A mRNA methylation levels (c) and degeneration index values (d) at an indicated time point and compared to the CPA treatment. Note the significant rescue of the degeneration index upon applying FB23-2 to CPA-stressed C9ORF72^exp, SOD1^+/L144F, TDP43^G298S, and sALS iPSC~MNs. Data are presented as mean ± SD, n = 3, significant P values from two-way ANOVA. N.S. non-significant. e Heatmaps of normalized expression level between stress-treated (CPA) ALS-relevant lines with or without subsequent FB23-2 treatment, revealing restorations of many ALS risk genes with m⁶A modifications (highlighted with rectangles) to control (vehicle) levels for the FB23-2-treated groups. A z-score normalization was performed on the normalized read counts across samples for each gene after stress treatment (CPA) with or without subsequent FB23-2 treatment. Samples were normalized to the vehicle control to reveal the normalized expression level. Notably, following FTO inhibitor treatment, these genes were restored to levels comparable to controls among those ALS iPSC~MNs are highlighted in bold purple. Source data are provided as a Source data file.

Fig. 9. Gene therapy of adult SOD1^G93A mice by overexpressing Fto-shRNA is sufficient to protect neuromuscular function and delay disease onset.

a, b Overview of the experimental strategy. Created in BioRender. Chen, J. (2025) https://BioRender.com/3m2vxum. c Western blot reveals that Fto protein level is reduced in mouse lumbar spinal cords after scAAV9-shFto injection (n = 3 mice). d and e Kaplan-Meier survival curves with log-rank test revealing prolongation of the onset of weight decline in SOD1^G93A mice (from ∼140 to ∼155 days), with lifespans extended by ∼10% (from ∼170 days to ∼187 days), following scAAV9-shFto injection. m⁶A levels is upregulated (f), MN number is rescued (f, with quantification in h), and gliosis is reduced (g, with quantification in i) after scAAV9-shFto injection of SOD1^G93A mice at P140. Scale bars, 100 µm. (mean ± SD, n = 4 mice; two-tailed t-tests). j The CMAP amplitude is reduced in SOD1^G93A mice at P60 and gradually declines further over time, whereas scAAV9-shFto treatment significantly ameliorates neuromuscular function at P160 (mean ± SD, Ctrl: n = 5/1, 4/3, 3/3, and 4/3 mice, Ctrl; scAAV9-shFto: n = 5/1, 5/2, 5/2, and 5/2 mice; SOD1^G93A: n = 2/3, 5/6, 4/5, and 4/3 mice; SOD1^G93A; scAAV9-shFto: n = 3/2, 5/3, 4/3, and 4/3 mice at P60, 120, 140, and 160 from males/females respectively; two-tailed t-tests) (right). k, l Motor coordination and muscle strength are enhanced by scAAV9-shFto injection, as assayed by rotarod test at P60~P160 (k) (mean ± SD, Ctrl: n = 6/2, 4/3, 4/3, and 4/3 mice, Ctrl; scAAV9-shFto: n = 5/1, 5/2, 5/2, and 5/2 mice; SOD1^G93A: n = 3/4, 5/6, 5/6, and 5/5 mice; SOD1^G93A; scAAV9-shFto: n = 5/2, 5/3, 5/3, and 5/3 mice at P60, 120, 140, and 160 from males/females respectively; two-tailed t-tests), and by grip strength test (l) (mean ± SD, Ctrl: n = 6/2, 4/3, 4/3, and 4/1 mice, Ctrl; scAAV9-shFto: n = 5/1, 5/2, 5/2, and 5/2 mice; SOD1^G93A: n = 4/4, 5/6, 5/6, and 4/3 mice; SOD1^G93A; scAAV9-shFto: n = 5/2, 5/3, 5/3, and 5/3 mice at P60, 120, 140, and 160 from males/females respectively; two-tailed t-tests). N.S. non-significant. Illustrations in j–l were created in BioRender. Chen, J. (2025) https://BioRender.com/mrtnwlk. Source data are provided as a Source data file.

Fig. 10. An increase of H3K9me3 is observed in MNs of SOD1^G93A and is reduced after scAAV9-shFto treatment.

a Representative image illustrating H3K9me3 marks (yellow arrowheads) in the ventral horn of different sets of Ctrl and SOD1^G93A mice, with scAAV9-shFto intrathecal injections. b Quantifications of lumbar H3K9me3^on MNs (Ctrl: n = 7 mice, Ctrl; scAAV9-shFto: n = 3 mice, SOD1^G93A: n = 4 mice, SOD1^G93A; scAAV9-shFto: n = 4 mice, quantification for all MN nuclei from all views of captured images; Scale bars, 20 µm). Data are presented as mean ± SD with significant P values from two-way ANOVA. N.S. non-significant. Source data are provided as a Source data file.