Fig 2: Mimicking HNRNPM-dependent linear-splicing events in cells inhibits exon inclusion and cell growth.(A, top panel) Schematic showing outcomes of splicing reporter assay. Wildtype (WT) or mutant (MUT) HNRNPM-binding sites at the USP33 or APMAP genes and identified by eCLIP were cloned into a splicing reporter. The percentages of green fluorescent protein (GFP) (indicating exon inclusion) and dsRed fluorescent proteins (RFP) (indicting exon skipping) single- (green and red bars) and double-positive cells (yellow bars) out of all fluorescent cells are shown in a barplot in the lower panel. Shown are the mean and standard deviation of three experiments. Student’s t-test was used to calculate p-values between the indicated groups. Ns: not significant; ** p<0.01; **** p<0.0001. (B) Gene ontology analysis of transcripts that are either changed in expression or mis-spliced upon HNRNPM knockdown. Both mis-spliced circular (green dots) and linear (red dots) RNAs are shown. Top enriched gene ontology categories (yellow dots) and the genes present in each category are indicated with colored lines. (C) Cell proliferation and exon inclusion outcomes in splice-switching-antisense-oligonucleotides (SSO)-treated cells as compared to scrambled SSO-treated cells. For proliferation assays, growth is shown relative to cells treated with the scrambled non-targeting SSO. The mean and standard deviation of three biological replicates are shown. p-values for growth were determined using Student’s t-test. *p<0.05. A representative splicing gel together with the PSI of the HNRNPM-regulated isoforms are shown below each barplot. As controls, a HNRNPM targeting SSO is also included (lane 2). For each gene (ZNF548, PRKAB2, and EED), multiple SSOs targeting different sequences were used. ZNF548 SSO-1, ZNF548 SSO-2, and PRKAB SSO-1 are control SSOs that do not induce inclusion events. For the EED event, both 2-O-methyl (2OMe) and 2-methoxy-ethyl (MOE)-based SSO backbone chemistries were used to control for potential toxicities due to SSO transfection. SSO backbone chemistries used are indicated at the top of each panel. (D) Disease-free survival plot of The Cancer Genome Atlas (TCGA) prostate adenocarcinoma (PRAD) patients, stratified by the total number of events per patient where the PSI of a given HNRNPM-regulated exon exceeds that of the median PSI within the patient population. Patients with more exon inclusion events (top 75th percentile) are shown in blue, whereas patients with less exon inclusion events (lower 25th percentile) are shown in red. (E) WD40 domain 5 of EED when the HNRNPM-regulated exon 10 is included. Structure of the HNRNPM-regulated WD40 domain in WT EED (gray), superimposed on the predicted structure of the HNRNPM-dependent EED isoform (blue). The new peptide generated by the splicing event is depicted in orange. (F) Closeup view of hydrogen bonding interactions that stabilize the WD40 beta sheet in either WT or the new EED isoform. (G) Western blot of H3K27me3 (EED target), HNRNPM, and EED upon treatment with the scrambled, HNRNPM-targeting SSO (HNRNPM) or SSOs that promote the inclusion of EED exon 10 (EED SSO-1 and SSO-2). Backbone chemistry of the SSOs used is indicated. Untreated cells are included as controls. HNRNPM. Shown also are histone 3 and actin B loading controls. (H) Disease-free survival curves of prostate cancer patients when expression in transcripts per million (TPM) of EED is low (blue) or high (red).

Fig 3: HNRNPM binds guanidine and uridine (GU)-rich elements within long introns.(A) Normalized read density of HNRNPM eCLIP at the HNRNPM and PRAM1 genes. Input (yellow) and immunoprecipitation (red) tracks are shown. High-confidence (p<0.05, fold enrichment >2) peaks are shown in the lowest track. (B) Distribution of HNRNPM-bound peaks across different genomic features. Relative frequency of total of specific events is indicated in the legend. (C) Distribution of gene lengths in all expressed genes (red), expressed and non-HNRNPM-bound genes (green) and expressed and HNRNPM-bound genes (blue). Log10(Gene length in base pairs) is plotted. (D) Histogram plot of nucleotide frequencies within called, high-confidence HNRNPM-bound peaks. (E) Motif analysis of HNRNPM-bound peaks. Shown are the top three (from top to bottom; p=1e-726, p=1e-625, and p=1e-597) motifs found within intragenic HNRNPM-bound sites.

Fig 4: ALS-causing mutations in MATR3 cause aberrant RNA-binding of MATR3 and hnRNPM to transcriptomic transcripts. a Schematic of RNA-immunoprecipitation to probe for physical interaction between MATR3 WT, F115C and S85C and hnRNPM. b–d Fold change differences in RNA targets immunoprecipitated with FLAG-MATR3 and V5-HNRNPM from cells expressing MATR3 WT and ALS-causing mutations F115C and S85C (n = 3-4 per group; One-way ANOVA). (b, left) DYRK1A, (c, left) SMYD3, and (d, left) ZNF644 mRNA show significantly higher enrichment when immunoprecipitated with F115C and S85C mutants compared to MATR3 WT. (B, right) DYRK1A, (C, right) SMYD3, and (D, right) ZNF644 mRNA also show significantly higher enrichment when immunoprecipitated with hnRNPM in cells expressing F115C and S85C groups compared to MATR3 WT. e Top 20 unique GO:Biological Process terms that are enriched in gene ontology assessment of shared transcriptomic targets from K562 cells and f HepG2 cells. Green bars indicate biological processes unique to the cell type. Orange bars indicate biological processes commonly enriched in both cell types. Error bars indicate S.E.M. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

Fig 5: Regulation of Nrxn3 exon SS4 splicing by Rbfox/LASR in MeCP2 KO and T158M mice.a Alternative splicing of Nrxn3 in the cerebral cortex of MeCP2 WT, KO, and T158M detected by RT-PCR. The average percentages of exon inclusion with standard deviations are shown below the gel images (n = 4 biological replicates; **p < 0.01, two-sided Student’s t test). b Schematic representation of Nrxn3 minigenes and RT-PCR analysis of in vivo splicing of Nrxn3 minigenes (wt and mut) in HEK293T cells transfected with a vector or a FLAG-RBFOX2 expression construct. The average percentages of exon inclusion with standard deviations are shown below (n = 3 biologically independent experiments). c RT-PCR analysis of in vivo splicing of Nrxn3 wt minigene in WT or MeCP2 KO HEK293T cells transfected with a vector or a FLAG-RBFOX2 expression construct. The average percentages of exon inclusion with standard deviations are shown below (n = 3 biologically independent experiments). d, e CLIP-RT-qPCR analyses of Rbfox proteins and hnRNP M binding to Nrxn3 pre-mRNAs in MeCP2 KO (d) or T158M (e) mice compared to WT mice. f ChIP-qPCR analysis of MeCP2 binding to Nrxn3 genomic locus in WT mice. g, h ChIP-qPCR analyses of Rbfox proteins and hnRNP M associating with Nrxn3 genomic locus in MeCP2 KO (g) or T158M (h) mice compared to WT mice. Error bars in d–h represent standard deviations (n = 3 biologically independent experiments; *p < 0.05, **p < 0.01, ***p < 0.001, two-sided Student’s t test). Source data are provided as a Source data file.