Fig 2: Motif analysis and interacting RBPs identified in aUTRs extracted from APA isoforms that show differential APA profiles when cytoplasmic APA isoforms are compared between: (A) isoforms that lengthen in SW620 compared to 1CT and (B) that shorten in SW480 compared to 1CT and (C) lengthen in SW620 compared to SW480. In all cases the top 2 significant overrepresented motifs were interrogated using the TOMTOM motif comparison tool to identify the corresponding interacting human RNA binding proteins with the default significance threshold set at E-value <10. The top two RBP hits for each motif are shown. No motifs were found for comparisons of SW620 compared to 1CT cells shortening in former and SW620 compared to SW480 cells shortening in former and SW480 compared to 1CT cells lengthening in former. Green boxes highlight hnRNPC hits. (D) hnRNPC is overexpressed in the nuclear (red) and cytoplasmic (purple) fractions of SW620 compared to SW480 and 1CT cells at the RNA level. Top 12 tracks show reads from sequencing the 3' ends of extracted nuclear RNAs (tracks 1–6) or cytoplasmic RNAs (tracks 7–12). Bottom two tracks show reads from full RNA-seq of SW620 and 1CT cytoplasmic polyadenylated mRNA. C1, C2: cytoplasmic fractions repeat 1 and 2. N1, N2: nuclear fractions repeat 1 and 2. Read numbers are scaled, tracks 1–6 max = 208 RPM, tracks 7–12 max = 350 RPM, tracks 13–14 max = 1200 RPM.

Fig 3: Comparison of APA profile shifts in RNAseq samples extracted from matched patient normal and tumour samples (34). (A) Bar chart comparing the mean read numbers aligning to the terminal exon of hnRNPC per million total aligned reads in normal and tumour samples. (B) Bar chart comparing the mean ratios of reads aligning to the terminal exon of the short and long isoforms of MTHFD1L in normal and tumour samples. (C–F) Bar charts comparing the mean ratios of reads aligning to the terminal exon of the short isoform and the aUTR for genes (C) PA2G4, (D) CLCN7, (E) SMAD3 and (F) FBRSL1 for normal and tumour samples. All bar chart error bars show the S.E.M. P-values (Student's t-test) test the significance of the difference in the means.

Fig 4: HNRNPC regulates the effect of the TFAP2A/DDR1 axis on collagen fiber arrangement and immune cell infiltration by recognizing the m6A modification of TFAP2A mRNA by VIRMA. Note: (A) RT-qPCR detection of the expression levels of VIRMA, HNRNPC, TFAP2A, and DDR1 in each group of EO771 cells; (B) Masson staining detection of collagen fiber arrangement in each group of EO771 cells in BC-MPS (×100); (C) Tumor volume of each group of breast cancer mice; (D) Picture of tumor harvested from each group of breast cancer mice; (E) Tumor weight of each group of breast cancer mice; (F) RT-qPCR detection of the expression levels of VIRMA, HNRNPC, TFAP2A, and DDR1 in each group of breast cancer mice tumor tissue; (G) Immunohistochemistry detection of Collagen IV content in ECM of each group of breast cancer mice tumor tissue (×200); (H) Masson staining detection of collagen fiber arrangement in each group of breast cancer mice tumor tissue (×400); (I) Immunohistochemistry detection of CD8 positive cell number in each group of breast cancer mice tumor tissue (×200); (J) CD8 + T cell migration experiment to detect the number of CD8 + T cell migration in each group of breast cancer mice tumor tissue after decellularized ECM treatment. * indicates P < 0.05 when compared with the two groups. All cell experiments were repeated 3 times with 5 mice in each group

Fig 5: VIRMA and HNRNPC participate in the m6A regulation of TFAP2A. Note: (A) Cuilab database predicted the m6A sites on TFAP2A mRNA, and the abscissa represents the location of the m6A sites on mRNA, and the ordinate represents the prediction score of the m6A sites, and the higher the score, the greater the possibility of m6A modification at the site; (B) Correlation analysis of TFAP2A and m6A regulatory factors HNRNPA2B1, HNRNPC, VIRMA, and WTAP in breast cancer-related TCGA dataset, including 1096 breast cancer tissues; (C) RT-qPCR detected the expression levels of HNRNPA2B1, HNRNPC, VIRMA, and WTAP in breast cancer cells (EO771, EMT6, and 4T1.2); (D) Schematic diagram of m6A regulatory mechanism; (E) RT-qPCR and Western blot detected the expression levels of VIRMA and HNRNPC in each group of EO771 cells after silencing; (F) RT-qPCR detected the expression levels of TFAP2A in each group of EO771 cells after silencing of VIRMA and HNRNPC; (G) RT-qPCR detected the expression levels of VIRMA, HNRNPC, and TFAP2A in each group of EO771 cells; (H) Me-RIP detected the m6A methylation level of TFAP2A mRNA in each group of EO771 cells; (I) PAR-CLIP experiment detected the binding of VIRMA to TFAP2A mRNA in each group of EO771 cells; (J) RIP experiment detected the m6A modification recognition of TFAP2A mRNA by HNRNPC in EO771 cells. * indicates P < 0.05, ** indicates P < 0.01 compared with the control group. All cell experiments were repeated 3 times