Autism-related traits in myotonic dystrophy type 1 model mice are due to MBNL sequestration and RNA mis-splicing of autism-risk genes

Figures

• 

  Fig. 1. ASD-risk gene mis-splicing in human DM1 prefrontal cortex. a, Age and sex distribution of DM1 and control (CTRL) prefrontal cortex samples used by Otero et al.35 to generate RNA-seq data. Box plot shows the lower, middle and upper quartiles. Whiskers show minimum and maximum. Ordinary one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test: NS, adjusted P (Padj) > 0.6. b, The number and percentage of AS event types significantly mis-spliced in DM1. SE, skipped exon; MXE, mutually exclusive exon; A5SS/A3SS, alternative 5′/3′ splice site; RI, retained intron event. c, Strong positive correlation between DMPK-CTG lengths and mean |ΔPSI| for all mis-spliced events. The 90th percentile of DMPK-CTG length was selected for this analysis, as it shows the strongest correlation between repeat sizes and splicing dysregulation score in DM1 (ref. 35). d, Enrichment analysis for mis-spliced ASD-risk gene sets in DM1. e, Sashimi plots quantitatively visualize splice junctions in DM1 (n = 8) and CTRL (n = 8) RNA-seq samples for selected AS events. f, High-confidence mis-splicing of ASD-risk genes in DM1 (n = 21) versus CTRL (n = 8). Data are presented as mean values PSI ±…

• 

  Fig. 2. Microexon mis-splicing in DM1 and Mbnl cDKO frontal cortices. a, The number and percentage of AS event types significantly mis-spliced in Mbnl cDKO (n = 3 males). b, ASD-risk gene set enrichment analysis for mis-spliced genes in Mbnl cDKO. c, Overlap between mis-spliced ASD-risk gene sets in DM1 and Mbnl cDKO. d, Scn2a MXE mis-splicing in Mbnl cDKO (n = 3) versus WT (n = 3) RNA-seq. e, miE enrichment analysis for mis-spliced ASD-risk gene sets. f, Ank2 miE mis-splicing in in Mbnl cDKO RNA-seq. g, Orthologous mis-spliced miEs in DM1 (n = 21) and Mbnl cDKO (n = 3). h, Schematic of Ank2 miE to the A3SS coordinate splicing and modeled structures of mouse Ank2 polypeptides. The aa sequences changed by AS are shown by a magenta box. The S901 phosphorylation site is bolded. Diamonds represent OR (b,e). Error bars depict 95% confidence intervals. Data are presented as mean values ± s.d. (d,g). #FDR < 0.10, *FDR < 0.05, **FDR < 0.01, ***FDR < 0.001, ****FDR < 0.0001 (b,d,e,g). Source data

• 

  Fig. 3. MBNL proteins govern developmental splicing transitions in ASD-risk genes. a, MBNL1, MBNL2 and MBNL3 gene expression levels in developing brains of five species: human (n = 49), rhesus (n = 17), mouse (n = 51), rabbit (n = 57) and rat (n = 57). Total MBNL expression relative to newborn/postnatal day 0 (P0d) (top). Mean MBNL expression for five species at each developmental stage (bottom). Significant differences between MBNL1 and MBNL2 expression at different developmental stages were determined by a two-tailed t-test; **P < 0.01. b, Strong positive correlation between Mbnl gene expression in developing WT mouse cortex and mean |ΔPSI | for MBNL-sensitive AS events (top) and miEs only (bottom) in SFARI genes. Black diagonal lines are the linear regression lines, and light gray areas are the 95% confidence intervals. P value for a two-tailed test. c, Splicing change in the developing mouse cortex for four MBNL-sensitive AS events in high-confidence ASD-risk genes. Embryonic days 14.5–16.5, postnatal days 0–30, and postnatal months 4–21 (n = 2 for each time point). d, Dmd miE splicing transitions during mouse cortical development. e, ASD-risk gene mis-splicing in Mbnl1 KO (n = 2),…

• 

  Fig. 4. Mis-splicing in the Mbnl2 KO hippocampus. a, The Allen Mouse Brain Atlas shows the normalized color-coded Mbnl2 expression level (from blue, low to red, high) derived from the informatics data processing of in situ hybridization results (mouse.brain-map.org/gene/show/69724)79. b, Scn2a MXE, Nrxn1 miE and Shank3 miE mis-splicing in Mbnl2 KO (n = 5) and littermate WT (n = 5) frontal cortex (FCx) and hippocampus (Hipp). Two-tailed t-test: ***P = 0.0001, ****P < 0.0001. c, The number and percentage of AS event types significantly mis-spliced in Mbnl2 KO hippocampus (n = 3 females). d, Ank2 miE mis-splicing in Mbnl2 KO RNA-seq. e, ASD-risk gene mis-splicing in Mbnl2 KO (n = 3) versus WT (n = 3) hippocampus RNA-seq. f, ASD-risk gene set enrichment analysis for mis-spliced genes in Mbnl2 KO. Diamonds represent OR. Error bars depict 95% confidence intervals. g, Percent of overlapping events annotated within ASD-risk genes among DM1 prefrontal cortex, Mbnl cDKO frontal cortex and Mbnl2 KO hippocampus. h, Examples of overlapping mis-splicing events in DM1 (n = 21), Mbnl cDKO (n = 3) and Mbnl2 KO (n = 3). One-way ANOVA followed by Tukey’s multiple comparisons test. NS, Padj > 0.05, ****Padj…

• 

  Fig. 5. MBNL proteins directly regulate Ank2 miE splicing. a, The number and percentage of AS event types significantly mis-spliced in Mbnl DKD CAD cells (n = 3). b, ASD-risk gene set enrichment analysis for mis-spliced genes in Mbnl DKD cells. c, MBNL-binding motif enrichment in mis-spliced ASD-risk genes. d, miE enrichment analysis for mis-spliced ASD-risk gene sets. e, Ank2 miE mis-splicing in Mbnl DKD RNA-seq. f, MBNL2-CLIP-seq reads (orange boxes, combined n = 3 females) in the Ank2 miE downstream intron containing three conserved (magenta) and one suboptimal (black) motifs. g, Schematic of limited heterologous Atp2a1 splicing minigenes and regulation by MBNL proteins. MBNL-binding sequences (magenta) identified in mouse Ank2 and human ANK2 and their mutants (blue). h, Heterologous Atp2a1 splicing minigene regulation by MBNL proteins in HeLa cells (n = 4). Unpaired two-tailed t-test: ****P < 0.0001. Data are presented as mean ± s.d. (e,h). Diamonds represent OR (b–d). Error bars depict 95% confidence intervals. #FDR ≤ 0.10, *FDR < 0.05, **FDR < 0.01, ***FDR < 0.001, ****FDR < 0.0001. Source data

• 

  Fig. 6. MBNL and SRRM regulate the same miEs. a, Age and sex distribution of ASD and CTRL prefrontal cortex samples in PsychENCODE RNA-seq data. ASD was confirmed by the Autism Diagnostic Interview-Revised (n = 8) or supported by records (n = 2). Box plot shows the lower, middle and upper quartiles. Whiskers show minimum and maximum. Two-tailed t-test; NS, P = 0.83. b, The number and percentage of AS event types significantly mis-spliced in ASD. c, ANK2 miE mis-splicing in ASD. Sashimi plots of ASD (n = 8) and CTRL (n = 8) RNA-seq. #FDR = 0.056. d, The number and percentage of AS event types significantly mis-spliced in Srrm DKD N2a cells (n = 2). e, Ank2 miE mis-splicing in Srrm DKD RNA-seq (n = 2). f, Overlapping mis-spliced miEs in Srrm DKD CAD and Mbnl DKD N2a cells. g, Selected miE splicing analysis in Mbnl DKD, Srrm DKD and Mbnl;Srrm QKD N2a cells (n = 4). h, Capillary electrophoresis analysis of Ank2 miE and A3SS. CI, combination index (g,h). One-way ANOVA followed by Tukey’s multiple comparisons test; NS, Padj > 0.05, *Padj < 0.05, **Padj < 0.01, ***Padj < 0.001, ****Padj < 0.0001. Data are presented as mean ± s.d (c,g–h). Source data

• 

  Fig. 7. Ank2 miE regulation by MBNL and SRRM proteins. a, SRRM4-CLIP-seq reads coverage (purple box) in the Ank2 miE upstream intron containing the UGC motif. b, Schematic of heterologous Atp2a1-Ank2 miE splicing minigenes and regulation by MBNL and SRRM proteins. MBNL (magenta) and SRRM (purple) binding sequences identified in mouse Ank2 and their mutants (blue). c, Proposed secondary structures for normal and mutant Ank2 RNAs. The optimal thermodynamic stability of the structure is expressed in Gibbs free energy (∆G) in kcal mol−1 for the reaction at 37 °C using mfold software. d, Experimental design. e, Atp2a1-Ank2 miE splicing analysis in N2a experiment (n = 3). Data are presented as mean values ± s.d. One-way ANOVA followed by Tukey’s multiple comparisons test; NS, Padj > 0.05, *Padj < 0.05, **Padj < 0.01, ****Padj < 0.0001. f, SRRM and MBNL proteins directly bind to Ank2 and synergistically promote miE inclusion. Source data

• 

  Fig. 8. Social behavior deficits in two DM1 mouse models. a, Scheme of the three-chamber habituation and sociability test. b,e, Ratio of time the test mice spent in the chamber with a novel animal (stranger) and object during the sociability test. Time spent in the middle chamber is not included. b, WT mice (n = 11), Dmpk-(CTG)480/WT (n = 11), and Dmpk-(CTG)480/480 (n = 11). Paired two-tailed t-test; NS, P = 0.38 (t = 0.9134, d.f. 10), *P = 0.013 (t = 2.997, d.f. 10), **P = 0.0025 (t = 4.012, d.f. 10). c, Representative mouse movement during the habituation phase. d, Normalized mean values for 13 parameters measured in the automated open field test. WT mice (n = 11), Mbnl2 KO (n = 12). Two-tailed t-test; NS, P = 0.83 (t = 0.2131, d.f. 21). e, WT mice (n = 12), Mbnl2 KO (n = 12); two-tailed t-test; NS, P = 0.43 (t = 0.8193, d.f. 11), *P = 0.023 (t = 2.641, d.f. 11). f, Scheme of the free dyadic social interaction test. g, Representative mouse interactions. h, Mouse interactions analysis. WT (n = 15), Mbnl2 KO (n = 14). Interaction time and brief interaction; two-tailed t-test; ***P = 0.00098 (t = 3.696, d.f. 27), ****P < 0.0001 (t = 9.842, d.f. 27). Average length; two-tailed Mann–Whitney…

• 

  Extended Data Fig. 1. Mis-splicing in DM1 prefrontal cortex. a, Differential AS analysis in DM1 (N = 21) compared to control (CTRL; N = 8) prefrontal cortex. The percentage of mis-spliced AS events (left) and genes (right) in the DM1. b, Strong positive correlation between DMPK-CTG lengths and mean |ΔPSI| for mis-spliced retained intron (RI) events. Note, the 90th percentile of DMPK-CTG length was selected, since it displays the strongest correlation between repeat sizes and splicing dysregulation score in DM135. The black diagonal line represents the linear regression line, and the light gray area indicates the 95% confidence interval. P-value for a two-tailed test. c, Negative correlation between mean ΔPSI for mis-spliced RI events (n = 158) and their host transcript relative steady-state levels (baseMean ≥ 1) in DM1. FC-fold change. The black diagonal line represents the linear regression line, and the light gray area indicates the 95% confidence interval. P-value for a two-tailed test. d, Circular RNA species in CSTPP1 intron 7 (I7) in DM1 (N = 21) and CTRL (N = 8) prefrontal cortex as well as control and HeLa treated with RNase R RNA-seq samples35,92. Box plot shows I7 relative to exon 7 (E7)…

• 

  Extended Data Fig. 2. miE mis-splicing in Mbnl cDKO frontal cortex. a, Differential AS analysis in frontal cortex of Mbnl cDKO compared to littermate WT control (N = 3 males). The percentage of mis-spliced AS events (left) and genes (right) in the Mbnl cDKO. b, ASD-relevant gene set enrichment analysis for mis-spliced genes in Mbnl cDKO (N = 3). Diamonds represent the odds ratio. Error bars depict the 95% confidence interval. c, Quantification of coordinate Ank2 miE and A3SS mis-splicing in Mbnl cDKO (N = 3) and WT (N = 3) RNA-seq samples. d, Examples of miE mis-splicing in ASD-risk genes. Sashimi plots of Mbnl cDKO (N = 3) and WT (N = 3) RNA-seq samples for Nrxn1 miE, Shank3 miE, and Dmd miE. Modeled structures of mouse protein isoform regions containing a sequence encoded by miE (magenta box). c,d, Data are presented as mean values ± SD. b,c,d, #FDR < 0.10, *FDR < 0.05, **FDR < 0.01, ***FDR < 0.001, ****FDR < 0.0001. Source data

• 

  Extended Data Fig. 3. MBNL-mediated splicing transitions in ASD-risk genes during brain development. a, Mbnl1, Mbnl2 and Mbnl3 gene expression levels during WT mouse cortex development. Dots represent the mean values (N = 2). b, Strong positive correlation between Mbnl gene expression in developing WT mouse cortex and mean |ΔPSI | for MBNL-sensitive AS events (top) and miEs only (bottom) in MSSNG-2017 and MSSNG-2022 genes. The black diagonal lines represent the linear regression lines, and the light gray areas indicate the 95% confidence intervals. P-value for a two-tailed test. c, Proportion of MBNL-sensitive AS events differentially spliced (gray) in the developing cortex (E14.5 d, E16.5 d vs P4m, P21m). d, Scn2a MXE and Ank2 miE splicing transitions during WT cortical development. e, RNA mis-splicing in Mbnl1 KO (N = 2), Mbnl2 KO (N = 2) and Mbnl cDKO (N = 2) mouse E18.5 cortical neuron RNA-seq samples. Box plot shows the lower, middle, and upper quartiles. Whiskers show minimum and maximum. Kruskal–Wallis test followed by Dunn’s multiple comparison test (n = mis-spliced AS events); nsPadj = 0.11, ****Padj < 0.0001. Source data

• 

  Extended Data Fig. 4. Mis-splicing in the mouse Mbnl2 KO brain. a, Nissl bodies from the Allen Mouse Brain Atlas at the same slice position as Mbnl1 and Mbnl2 RNA in situ hybridization (ISH) results. Staining Nissl bodies serve as a reference for the ISH data. For ISH, specific digoxigenin tagged RNA probes were used to label cells expressing Mbnl1 (mouse.brain-map.org/gene/show/36037) and Mbnl2 (mouse.brain-map.org/gene/show/69724) transcripts79. b, Relative MBNL1, MBNL2, and MBNL3 expression in multiple human and mouse brain regions. CNS - central nervous system. c, Schematic of the 3-primer Scn2a RT–PCR assay. d, Scn2a MXE and Nrxn1 miE mis-splicing in Mbnl2 KO (N = 5) and littermate WT (N = 3) cerebellum. Two-tailed t-test; nsP > 0.05. This result was repeated twice. e, Differential AS analysis in frontal hippocampus of Mbnl2 KO compared to littermate WT control (N = 3 females). The percentage of mis-spliced AS events (left) and genes (right) in the Mbnl2 KO. f, Examples of mis-splicing in ASD-risk genes in Mbnl2 KO RNA-seq. g, ASD-relevant gene set enrichment analysis for mis-spliced genes in Mbnl2 KO (N = 3). Diamonds represent the odds ratio. Error bars depict the 95% confidence interval. *FDR…