Characterization of Rbm20 expression in the brain

A. Dot plot of the expression of a hand-curated list of RBPs across different neuronal neocortical populations. RBPs were chosen based on the presence of an RNA recognition motif (RRM) in their sequence and on the ranking of their gini-index value (only the first 20 RBPs displaying the highest gini-index value are displayed (see methods)). RBPs expression was measured by Ribo-TRAP sequencing and expressed as RPKM values normalized over the mean expression across different neuronal populations.

B. Sagittal section of the mouse brain used for immunohistochemistry of RBM20 (grey). The somatosensory cortex and the olfactory bulb regions, where RBM20 is expressed, are highlighted. Scale bar: 1 mm.

C. Immunohistochemistry of RBM20 expression (green) in Parvalbumin positive interneurons (red) in the neocortex.

D. Schematic illustration of the olfactory bulb circuitry and cell types (left). GL: glomerular layer, EPL: external plexiform layer, MCL: mitral cell layer, GCL: granule cell layer, (left). RBM20 expression (green) is specific to the mitral cell layer and glomeruli layer of the olfactory bulb identified in the Vglut2Cre :: Rpl22HA mouse line (HA staining in red) (middle and right). Scale bar 100 µm.

E. Western blot showing RBM20 expression levels in cortex (CX), olfactory bulb (OB) and heart (HR) samples of WT and constitutive KO mice. RBM20 band at 150-170 kDa is indicated with an arrow. GAPDH detection is used as loading control.

F. Immunofluorescence of RBM20 (green) expression in the mitral cell layer (MCL) of P0 and P35 Tbx21Cre::Ai9tdTomato mice. Mitral cells and tufted neurons are labelled with the tdTomato reporter (red). Scale bar 100 µm.

Identification of RBM20 direct targets in the heart and olfactory bulb

A. Sub-nuclear localization of RBM20 (green) in heart cardiomyocytes (left) and mitral cells of the olfactory bulb (right) of WT mice at P35. Scale bar 10 µm.

B. Schematic illustration of HA epitope-tagging of endogenous RBM20 in mice. A cassette was inserted into the Rbm20 locus containing a strong synthetic 3’ splicing acceptor site (3’SA) introduced into sequences derived from Rbm20 exon 14 and in frame fusion of a histidine-biotin-histidine-3xHA tag, followed by a poly-adenylation signal (up). Schematic representation of the resulting RBM20HA protein where the last exon of the protein is fused to a histidine-biotin-histidine-3xHA tag (down).

C. Western blot showing the validation of RBM20 expression in the olfactory bulb and cortex tissues of WT and Rbm20HA tagged mice. GAPDH is used as a loading control.

D, E (D) Quantification of the percentage of peaks identified in the heart (red) and (E) olfactory bulb (blue) tissue in each genomic feature: introns, untranslated regions (3’UTR, 5’UTR), coding sequence (CDS), others (promoters, intergenic regions, non-coding regions). The absolute number of the peaks identified in each genomic feature is reported on top of the corresponding bar.

F. Bar plot showing the percentage of peaks identified in distal (>500 bp) or proximal (< 500 bp) intronic regions in the heart (red) and olfactory bulb (blue). The absolute number of the peaks identified is reported on top of the corresponding bar.

Identification of RBM20 direct mRNA targets in the heart and olfactory bulb

A. Tracks illustrating RBM20 CLIP-seq signal for Ttn, CamkIId, Cacna1c, and Nav2. Read density obtained for heart samples (red traces) and olfactory bulb (green traces) with the corresponding input samples (overlaid traces in grey). CLIP peaks, considered statistically significant by IDR (score >540, equivalent to IDR<0.05) are marked by black arrowheads. RBM20-dependent alternative exons previously reported for cardiomyocytes are labeled in red. Note that in the olfactory bulb, RBM20 binding sites are identified on CamkIId and Cacna1c pre-mRNAs. However, these binding sites are distal (> 500 bp) to the alternative exons. See TableS1 for coordinates.

B. Illustration of the GO categories of RBM20 mRNA targets in the heart (IDR<0.05). Cellular Component analysis with Bonferroni correction (p-value ≤ 0.05). The number of genes found in each category is displayed on top of each bar. Minimal number of genes identified in each category: 5 genes.

C. Illustration of the GO categories of RBM20 mRNA targets in the olfactory bulb (IDR<0.05). Cellular Component analysis with Bonferroni correction (p-value ≤ 0.05). The number of genes found in each category is displayed on top of each bar. Minimal number of genes identified in each category: 5 genes.

Differential gene expression and alternative exon incorporation rates in Rbm20 conditional knock-out cells

A. Volcano plot of differential gene expression in RiboTrap-isolated mRNAs from Rbm20WT vs. Rbm20ΔVglut2 olfactory bulb (P35). Significantly regulated genes shown in red, cut-off fold-change (FC) of 1.5, adjusted p-value <0.01, total number of up- and down-regulated noted on top. Note that Rbm20 itself is strongly reduced, outside the axis limits, and not represented in this plot (see Table S4).

B. Volcano plot of the differential gene expression in RiboTrap-isolated mRNAs from Rbm20WT vs. Rbm20ΔPV mouse neocortex (P35) as in panel A. The Y-chromosomal genes Ddx3y, Uty, Kdm5d, Eif2s3y were highly differentially expressed due to the larger number of Rbm20 mutant males used in the Ribotrap isolations (3 wild-type females and 1 wild-type male vs. 4 knock-out male mice were used for this experiment). These genes and Rbm20 itself were excluded from the plot (see Table S4 for complete data).

C. Volcano plot representing differentially-included exons in Rbm20ΔVglut2 RiboTrap-isolated mRNAs from olfactory bulb neurons. The dotted lines correspond to FC values of 1.5 and −1.5 and −log10 (p-value) of 1.3. Significantly regulated exons (FC 1.5 and p<0.05) are shown in red.

D. Volcano plot representing differentially-included exons in Rbm20ΔPV RiboTrap-isolated mRNAs from cortical interneurons. The dotted lines correspond to FC values of 1.5 and −1.5 and −log10 (p-value) of 1.3. Significantly regulated exons (FC 1.5 and p-value <0.05) are shown in red.

E. Number of exons differentially expressed in Vglut2+ cells isolated from the olfactory bulb of Rbm20ΔVglut2 mice and number of exons with significant RBM20 CLIP peaks in binding sites within indicated distances. Equivalent information is provided for exons divided by annotations for specific features of regulation: alternative poly-adenylation (last exons), alternative transcription start sites (TSS), complex events and cassette exons.

Long pre-mRNAs are depleted in Rbm20ΔVglut2 mitral cells

A. Metagene coverage plots of CLIP peaks across all RBM20-bound introns. Peak density across exons from the same transcripts are shown for comparison.

B. Total number of genes up- or down-regulated (FC > 1.5 and adjusted p-value <0.01) in glutamatergic cells from the olfactory bulb of Rbm20ΔVglut2 mice. The fraction of genes with significant RBM20 CLIP peaks is indicated in blue.

C. Illustration of gene ontologies enriched amongst up- and down-regulated genes (cellular component analysis with Bonferroni correction (p-value 0.05). Minimum number of 5 genes identified per category.

D. Correlation of differential gene expression in Rbm20ΔVglut2 cells, intron length and CLIP-seq data. Genes were ranked by FC in differential gene expression and mean total intron length (left) and mean number of intronic CLIP peaks (right) for blocks of 100 genes were plotted. Ranks of the genes meeting FC cut-off for down- or up-regulation are highlighted in color. Spearman’s coefficients and p-values are indicated.

E. Boxplot showing the total intron length per gene (expressed in log10 scale) for categories of downregulated genes (all, with or without RBM20 binding sites), non-regulated genes, and up-regulated genes in Rbm20ΔVglut2 in RiboTrap isolates from the olfactory bulb. Only annotated genes are plotted (number of genes shown at the bottom). P-values from Wilcoxon test are indicated. Medians: all down-regulated genes 83.2 kb, down-regulated genes with peaks: 152.4 kb; down-regulated genes without peaks: ∼33.6 kb; all non-regulated genes: 21.5 kb; all up-regulated genes: 7.4 kb.

F. Boxplot (log10 scale) illustrating the total length of introns found in genes identified in our RBM20 Ribo-TRAP dataset (grey) compared to introns presenting RBM20 binding sites (green). RBM20 bound introns exhibit a higher intron length. P-values from Wilcoxon test are indicated. Medians: expressed introns: 1.4 kb; introns with peaks: 30.5 kb.

G. Plot representing the cumulative probability distribution of intron length between the two groups of introns as in panel F. P-value (Kolmogorov-Smirnov test) is indicated.

Characterization of Rbm20 expression in the brain

A. Dot plot of the expression of a hand-curated list of RBPs across different neuronal neocortical populations. RBPs were chosen based on the presence of an RNA recognition motif (RRM) in their sequence and their expression in the neocortex. RBPs expression was measured by Ribo-TRAP sequencing and expressed in RPKM values normalized over the mean expression across different neuronal populations.

B. Fluorescent in situ hybridization (FISH) on P25 brain slices from mice with genetic marking of cell populations (Pvalbcre::Ai9 cre-dependent tdTomato expression). Red: tdTtomato mRNA, blue: DAPI. Scale bar 100 μm;

C. Fluorescent in situ hybridization for Rbm20 transcripts in tdTomato-marked cells for different cell classes (cre-dependent Ai9 tdTomato reporter crossed to the indicated cre-recombinase expressing lines: Camk2, Pvalb, Sst, Vip) in P23-26 somatosensory cortex cells in layer 5. Green: Rbm20 mRNA, red: tdTtomato mRNA, blue: DAPI. Scale bar 10 μm.

D. Quantification of Rbm20 mRNA expression as in C, expressed as mean number of fluorescent dots per cell from three animals per genotype. P-value < 0.01, one-way Anova.

E. Schematic illustration of the olfactory bulb circuitry and cell types. GL: glomerular layer, EPL:external plexiform layer, MCL: mitral cell layer, GCL: granule cell layer, (left). Fluorescent in situ hybridization (FISH) on brain slices for Rbm20 (green), Tbr2 (gray), Vglut2 (red) mRNAs (middle). The insets show the example of a cell expressing all three markers (right). Scale bar: 100 μm; scale bar insets: 10 μm.

F. Pie charts indicating the quantification of the percentage of cells of the MCL and GL expressing Vglut2, Rbm20 and Tbr2 transcripts. Amongst the Rbm20+ cells, the percentage of neurons presenting co-localization with glutamatergic markers was calculated.

G. Quantification of the percentage of neurons of the mitral cell layer (MCL) and glomerular layer (GL) expressing high or low Rbm20 mRNA levels, co-localizing with Tbr2 and Vglut2 markers. The absolute number of high and low Rbm20 expressing neurons identified in both the mitral cell layer and the glomeruli layer is reported in brackets.

H. Fluorescent in situ hybridization (FISH) on brain slices of Rbm20 (green), Gad1 (gray), Vglut2 (red) mRNAs. Scale bar 100 µm. The arrow in the inset on the right of the panel shows an example of a cell expressing low Rbm20 levels and co-localizing with Gad1 marker but not with Vglut2 marker. Scale bar inset: 10 μm.

Generation of Rbm20 – HA tagged mouse line

A. Schematic illustration of the “COIN allele” strategy used for generation of Rbm20HA knock-in mice. The COIN allele module was inserted in the Rbm20 locus with the CRISPR-CAS9 system. Upon Cre-mediated recombination of the loxP sites, the COIN module is inverted and the presence of a strong synthetic 3’ splicing acceptor site (3’SA) results in the expression of the tagged Rbm20 isoform. While the strategy was achieved to obtain conditional tagging in specific cell types the cre-mediated inversion frequency was too low for use in vivo.

B. Immunohistochemistry of RBM20HA protein in mitral cells of wild type and Rbm20HA mice. HA (gray), DAPI (blue). Scale bar: 10 μm.

C. Targeted proteomic analysis on heart samples of wild type (WT), Rbm20WT/HA heterozygous, and Rbm20HA/HA homozygous knock-in mice. Three proteotypic peptides were quantified. Plotted results represent means of all three peptides normalized to wild-type samples. Four biological replicates per genotype. The mean of the –log2 ratio (light/heavy) peptides was calculated and displayed. Note that an assessment of RBM20 expression level in the Rbm20HA mice by Western blot was not possible as the C-terminal HA-epitope reduced binding affinity of the anti-RBM20 antibody raised against the C-terminus of the protein.

Identification of RBM20 binding sites on transcript mRNAs

A – B Correlation analysis of normalized counts (reads per million [rpm] +1) of called CLIP peaks between seCLIP replicates of the heart (2 samples) and the olfactory bulb (3 samples). Gray shades represent density of the data points. Pearson coefficients are indicated in above the corresponding plots.

C. Enrichment of the TCTT motif at cross-link-induced truncation sites (CITS) in both heart and olfactory bulb tissues. The enrichment is calculated from the frequency of the TCTT motif starting at each position of the inferred cross-linked site, normalized by frequency of the same motif in adjacent flanking regions.

D. Motif finding analysis performed with DREME on heart and olfactory bulb seCLIP datasets. The first 5 statistically significant motifs are reported, ranked based on the enrichment p-value (E-value; right of each panel). The E-value is defined as the p-value times the number of candidate motifs tested. The enrichment p-value is calculated using Fisher’s Exact Test for enrichment of the motif in the positive sequences, calculated after erasing sites that match previously found motifs. Note that the G-rich motif (rank 3) is commonly found to be non-specifically recovered in seCLIP datasets.

Normal morphological differentiation of mitral cells in the absence of RBM20.

A, B. Immunostaining of RBM20 (green), RPL22-HA (gray) and DAPI (blue) in the cortex and olfactory bulb of P35 PvalbCre::Rpl22HA/HA and (B) Vglut2Cre::Rpl22 HA/HA mice, compared to corresponding littermates carrying the conditional Rbm20fl/fl alleles (Rbm20ΔPV and Rbm20ΔVglut2, lower panels).

C. RBM20-positive (green) mitral cells retrogradely labeled through injection of rAAV2-SYN-Cre virus (red) in piriform cortex of a Ai9tdTOM mice at P35. Scale bar 100 µm.

D. Schematic illustration of the rAAV2-DIO-eGFP virus injection into the Piriform cortex of Vglut2Cre mice for retrograde labelling of olfactory bulb mitral cells. Representative image of the site of viral injection in the piriform cortex. Scale bar 500 µm.

E. GFP+ mitral cell (gray) retrogradely labeled through injection of rAAV2-SYN-DiO-GFP virus in the piriform cortex of Vglut2Cre and Vglut2cre::Rbm20fl/fl(Rbm20ΔVglut2) mice. Tracing of the neuronal arborization is displayed on the right. Scale bar 500 µm.

F. Quantification of the absolute number of branches and the mean length of the neuronal tufts in reconstructed mitral cells of Vglut2Cre and Rbm20ΔVglut2 mice.

Quality control analysis of Ribo-TRAP RNA-sequencing samples

A – B. Fold-enrichment (FC) of markers specific to inhibitory cortical neurons or glutamatergic neurons for WT and cKO samples. For PV samples, the following markers were tested: Pvalb, Gad67, Vgat, Vglut1, Gfap. For glutamatergic: Vglut1, Vglut2, Gad67, Pcdh21, Gfap, Vgat and Tbr2.

C. Coverage plot indicating the percentage of read bases at a given position of the transcript. No sample displayed 3’ or 5’ coverage bias across the transcript length.

D. Bar plot of each biological replicate where for each sample the following parameters are indicated: the proportion of number of reads uniquely mapped, mapped to multiple loci, mapped to too many loci or unmapped reads for all the samples. All samples show highly similar values across biological replicates, as well as across brain region and genotype, suggesting a high consistency and homogeneity of the RNA-seq data.

E. Bar plot representing the relative percentage of reads falling on genomic features for all the biological samples.

F. PCA of genes expressed in each olfactory bulb sample (n=5 biologically independent samples per genotype). Variance explained by the principal components 1 and 2 (PC1 and PC2) is indicated. Gene expression values were normalized by Variance Stabilizing Transformation (VST).