Fig 2: HNRNPC and NFIX as direct targets for miR-744. To validate miR-744 mediated regulation of potential target genes, SKOV3 and OVCAR3 cells were transfected with miR-744 mimic or non-targeting siRNA (NT, negative control for cell death) as described in Fig. 3. After 48 h, potential target gene expression was analysed by qRT-PCR. The relative mRNA expression of potential target genes was normalized to PPIA and NT employing the Livak method55 (a). Significantly downregulated potential targets were further analysed by the Luciferase Reporter Assay (b and c) to proof direct binding of the miRNA to its predicted target sequence. Anti-miR-744 was used to confirm the binding sides of miR-744 to the target mRNAs (d). The binding sites in the 3′UTR (Table 3), based on data of microRNA.org34 and TargetScanHuman35, were cloned into a pMirGLO vector. Relative luciferase activity was measured 72 h after co-transfection of the pMirGLO vector with miR-744 mimic, miRNA inhibitor anti-miR-744 or non-targeting siRNA (NT) in HEK293T cells. The relative luciferase expression was normalized to the expression after co-transfection with NT. For Western Blot analysis, SKOV3 and OVCAR3 cells were harvested 60 h after miR-744 transfection (e). The representative blots were stained with antibodies against NFIX and HNRNPC. GAPDH was used as loading control. Statistical differences were tested using paired t-test [n = 3 biological and technical replicates; mean ± SD; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001].

Fig 3: CMTR1 interacts with splicing factor ELAVL1 in gastric cancer cells. a GO pathway analysis via Metascape program (https://metascape.org) of 204 CMTR1-interacting proteins derived from BioGRID database (https://thebiogrid.org). b Venn diagram showing the identification of CMTR1-binding partners via over-lapping analysis of CMTR1-interactng proteins derived from BioGRID database, established splicing factors, and RBPs binding with CD44 pre-mRNA in POSTAR3 (http://postar.ncrnalab.org) or ENCORI (https://rnasysu.com/encori/) database. c Co-IP and western blot assays indicating the interaction of CMTR1 with splicing factors in AGS cells. d Schematic depiction (upper panel) and cross-linking RIP assay (lower panel) showing the interaction of ELAVL1, HNRNPC, HNRNPU, or PTBP1 with CD44 pre-mRNA containing alternative splicing sites around exon 10 in AGS cells. e RT-PCR assay (left panel) and western blot (right panel) assays indicating the differential expression levels of CD44v6 and CD44s in gastric cancer AGS and MKN-45 cells stably transfected with empty vector (mock), ELAVL1, scramble shRNA (sh-Scb), sh-ELAVL1 #1, or sh-ELAVL1 #2. f Representative images (upper panel) and quantification (lower panel) of immunofluorescence assay showing the co-localization (arrowheads) of ELAVL1 with CMTR1 in HGC-27 cells. g BiFC assay indicating interaction between CMTR1 and ELAVL1 (arrowheads) within HGC-27 cells co-transfected with pBiFC-VN173-CMTR1 and pBiFC-VC155- ELAVL1, with nuclei staining by DAPI. h In vitro binding (left panel), co-IP (right panel) and western blot assays showing the interaction between GST-tagged CMTR1 and MBP-tagged ELAVL1 proteins, and that in AGS cells transfected with full-length or truncations of HA-tagged CMTR1 and Flag-tagged ELAVL1 constructs. Fisher's exact test for over-lapping analysis in b. Data are representative of three independent experiments in c-h

Fig 4: Binding of hnRNP C is necessary to inhibit back-splicing events.(A) Representative RT-PCR image (top) and densitometric analysis (bottom) showing endogenous circSMN9-6 expression in D341 cells transfected with siNC or siHNRNPC. Actin was used as a loading control (n = 3; means ± SD; Student’s t test; P ≤ 0.05). (B) Schematic representation of the wild-type circSMN9-6 minigene (WT). Gray boxes indicate exons, red boxes represent Alu elements, black lines denote introns, and red arrows indicate divergent primers used for the detection of minigene-derived circSMN9-6. The nucleotide sequence highlights HNRNPC-binding sites within SMN2 intron 5. (C) Representative RT-PCR image (top) and densitometric analysis (bottom) showing minigene-derived circSMN9-6 expression in HEK293T cells transfected with either empty vector (EV) or the circSMN9-6 WT minigene, with or without RNase R treatment. Actin was used to normalize RNA input (n = 3; mean ± SD; one-way ANOVA; *P < 0.05 and ****P < 0.0001). (D) Schematic representation of mutant circSMN9-6 minigenes. BS1, BS2, and BS1/BS2 indicate mutations in HNRNPC-binding site 1, HNRNPC-binding site 2, and both sites, respectively. (E) Representative RT-PCR image (top) and densitometric analysis (bottom) showing circSMN9-6 expression derived from mutant minigenes in HEK293T cells transfected with BS1, BS2, or BS1/BS2 constructs, following transfection with siNC or siHNRNPC. Actin was used as a loading control (n = 3; means ± SD; Student’s t test; **P < 0.01).

Fig 5: HnRNP C is a general repressor of circRNA biogenesis in group 3 MB cells.(A) Percentage of circRNAs generated from exons, introns, intergenic regions, and antisense transcripts in siNC and siHNRNPC samples. (B) Number of unique circRNAs detected in siNC- and siHNRNPC-treated D341 cells. Median with interquartile range is shown (Student’s t test; **P < 0.01). (C) Number of circRNA-producing genes in siNC and siHNRNPC samples. (D) Circular to linear read-count ratio per circRNA in control and HNRNPC-depleted cells. P value is shown (Student’s t test). (E) Percentage of hnRNP C–regulated genes in terms of back-splicing (BS; n = 352), alternative splicing (AS; n = 648), and gene expression (GE; n = 572). Fisher’s exact test; *P = 0.03, ****P = 8.17 × 10−51, and ####P = 6.5 × 10−60. (F) Number of up-regulated and down-regulated circRNAs upon HNRNPC depletion. (G) Gene set enrichment analysis of the genes generating up-regulated circRNAs was performed using Enrichr (https://maayanlab.cloud/Enrichr/). (H) Validation of six up-regulated exonic circRNAs contained in the list of the top 8 most up-regulated circRNAs in HNRNPC-depleted cells (n = 4; means ± SD; Student’s t test versus siNC; *P < 0.05, **P < 0.01, and ***P < 0.001).