Depletion of Mettl3 in cholinergic neurons causes adult-onset neuromuscular degeneration

Abstract

Motor neuron (MN) demise is a hallmark of several neurodegenerative diseases, including amyotrophic lateral sclerosis (ALS). Post-transcriptional gene regulation can control RNA's fate, and defects in RNA processing are critical determinants of MN degeneration. N⁶-methyladenosine (m⁶A) is a post-transcriptional RNA modification that controls diverse aspects of RNA metabolism. To assess the m⁶A requirement in MNs, we depleted the m⁶A methyltransferase-like 3 (METTL3) in cells and mice. METTL3 depletion in embryonic stem cell-derived MNs has profound and selective effects on survival and neurite outgrowth. Mice with cholinergic neuron-specific METTL3 depletion display a progressive decline in motor behavior, accompanied by MN loss and muscle denervation, culminating in paralysis and death. Reader proteins convey m⁶A effects, and their silencing phenocopies METTL3 depletion. Among the m⁶A targets, we identified transactive response DNA-binding protein 43 (TDP-43) and discovered that its expression is under epitranscriptomic control. Thus, impaired m⁶A signaling disrupts MN homeostasis and triggers neurodegeneration conceivably through TDP-43 deregulation.

Keywords: ALS; CP: Neuroscience; FTD; METTL3; N6-methyladenosine; RNA epigenetics; RNA metabolism; TDP-43; age-related neurodegeneration; amyotrophic lateral sclerosis; frontotemporal dementia; m6A; methyltransferase-like 3; motor neurons; transactive response DNA-binding protein-43.

Figure 1.. Depletion of Mettl3 in cholinergic neurons results in an adult-onset neuromuscular phenotype.

(A) Spinal cords from two-week-old control and Mettl3-cKO mice were stained with anti-METTL3 (green) and anti-ChAT (red) antibodies. Scale bar, 25 μM. (B-C) Progressive loss of body weight (g) in male (n=6-8 / genotype) and female (n=6-8 / genotype) Mettl3-cKO mice. (D-E) Kaplan-Meier survival curve for Mettl3-cKO mice. Log-rank (Mantel-Cox) test (P<0.0001), Median all mice: 272, n=27; Median male: 274.5, n=12; Median female: 271, n=15. (F) Inverted grid (n=21-25 / genotype). (G) Rotarod (n=12-13 / genotype). For (F-G), data were analyzed by a non-linear regression curve fit using the least sum-of-squares method (P<0.0001). (H) Open Field: ambulatory distance (cm) (n=9-10 / genotype, 7 months). There is a significant main effect of genotype (F_1,16 =26.776, P<0.001). (I) Open field: center time (sec) (n=9-10 / genotype, 7 months). There is a significant main effect of genotype (F_1,16 =5.250, P=0.036). For (H-I), data were analyzed by Two-Way RM ANOVA. See also Figure S1.

Figure 2.. MNs are differentially affected by Mettl3 depletion.

(A) Representative images of ChAT immunostaining in ventral lumbar segment 4 and 5 (L4-L5) of the spinal cord and oculomotor MNs in the brainstem from control and Mettl3-cKO mice at end-stage (P250). Scale bar, 50 μm. Quantification of (B) MN number per hemi-section, (P=0.0006) and (C) oculomotor MNs, (ns). (D) Immunostaining of synaptophysin (red) and α-Bungarotoxin (green) in the tibialis anterior (TA) from control and Mettl3-cKO mice at P100 and P250. Scale bar, 10μM. Quantification of neuromuscular junctions (NMJs) innervation status at (E) P100, (ns) and (F) P250, (P<0.0001). Data are means ± SEM of n=3-4 independent experiments analyzed by unpaired t-test. See also Figure S2.

Figure 3.. METTL3 depletion results in the selective degeneration of ES-MNs.

Fields of GFP-expressing ES-MNs (A) and Tuj1-expressing ES-INs (D). False colors are assigned to individual neurons by MetaMorph. Scale bar, 200 μm. Quantification of (B) MNs (Day 1 and Day 7; P<0.0001), (C) MN-neurite outgrowth (Day 1, P=0.0001; Day 7; P<0.0001), (E) INs (Day 1, P>0.9999; Day 7, P=0.1482), and (F) IN-neurite outgrowth (Day 1, P=0.1750; Day 7, P=0.7086). (G-N) Quantification of MNs and MN-neurite outgrowth, respectively, for Mettl3 (G, Day 8, P=0.038; H Day 8, P=0.018), Ythdf1 (I, Day 8, P=0.0004; J, Day 8 P=0.0004), Ythdf2 (K, Day 8 P=0.1384; L, Day 8 P=0.3187), and Ythdf3 (M, Day 8, P=0.0029; N, Day 8, P=0.0005). Data are means ± SEM of n ≥ 3 independent experiments analyzed by unpaired t-test. See also Figure S3.

Figure 4.. Nanopore-based direct RNA sequencing identifies TARDBPas an m⁶A target.

(A). Metagene distribution of m⁶A+ (green) and randomly selected (red) 5-mers, including 0.95 confidence intervals (shadow). (B) Frequency of A-containing m⁶A+ 5-mers (black) and any A-containing tested 5-mer (gray). The red dashed line represents the expected frequency for 5-mers composed of 5 nucleotides randomly extracted with probability 0.25. (C) As in (B) for DRACH 5-mers. (D) Scatterplot reporting the number and median significance of identified m⁶A+ 5-mers (GMM logit adjusted p-value < 0.05) for each of the 1,777 genes coding m⁶A+ transcripts. In red are 8 top-ranking genes for the number and significance of m⁶A 5-mers (thresholds set at 95% of the TARDBP values – grey dashed lines). Green triangles identify genes saturated at the axis maximum values. (E) Stoichiometry of the 5 m⁶A sites detected on TARDBP. (F) Heatmap showing the WT modification probability for each of the 5 TARDBP m⁶A sites (rows) in each read (columns). (G) Number of TARDBP reads simultaneously modified in the indicated combinations of 2, 3, 4 and 5 m⁶A sites. The co-occurrence significance reports the number of times the co-occurrence exceeds a null model obtained by shuffling 1,000 times the probabilities of each site. See also Figure S4 and Table S1.

Figure 5.. Loss of m⁶A impairs TDP-43 autoregulation.

(A) TDP-43 binds to the TDP-43 binding region (TDPBR in green) in the 3’UTR (blue) of TARDBP pre-mRNA. Exons are depicted as boxes, highlighting coding regions (black), and introns as lines. (B) m⁶A-RIP and qRT-PCR for Tardpb in ES-MNs. The graph depicts the depletion of Tardpb in Mettl3-KO conditions (relative to IgG). Hprt is used as a negative control. Data are means ± SEM of 3 independent experiments analyzed by unpaired t-test (P=0.0003). (C) Western blot of endogenous TDP-43 levels from doxycycline-induced and non-induced HEK-293T cells transfected with siRNA^NTC or siRNA^METTL3. GAPDH was used as a loading control. (D-E) Quantification of endogenous TARDBP mRNA (D) and TDP-43 protein (E) expression in RNAi-transfected cells with induction (+Dox) of HA-TDP-43 expression relative to non-induced cells (−Dox, dotted line). Data are means ± SEM of n=4 independent experiments analyzed by unpaired t-test (D, P=0.4055; E, P=0.0207). (F) Western blot of mouse TDP-43 levels in the cytoplasmic or nuclear fraction of Mettl3-WT and Mettl3-KO ES-MN lysates. Histone H3 and PKC were used as nuclear or cytoplasmic loading controls, respectively. (G-H) Quantification of TDP-43 protein levels in nuclear (G) or cytoplasmic (H) fractions from F. Data are means ± SEM of n=3 independent experiments analyzed by unpaired t-test (G, P=0.0070; H, P=0.0059). (I) Western blot of mouse TDP-43 levels after immunoprecipitation of TDP-43 in total cell lysates from ES-INs or ES-MNs differentiated from Mettl3-WT or Mettl3-KO. ß-actin was used as a loading control. (J-K) Quantification of TDP-43 protein levels in ES-MNs (J) or ES-INs (K) from I. Data are means ± SEM of n=3 independent experiments analyzed by unpaired t-test (J, P<0.0001; K, P=0.5164). (L) Model by which the loss of m⁶A marks on TARDBP disrupts the interaction between TDP-43 and its transcript and impairs autoregulation. See also Figure S5.