IGF2BP2 binding to CPSF6 facilitates m6A-mediated alternative polyadenylation of PUM2 and promotes malignant progression in ovarian cancer

Abstract

Background: N6-methyladenosine (m6A) and alternative polyadenylation (APA) are common posttranscriptional regulatory mechanisms in eukaryotes. However, the m6A-dependent mechanism of APA regulation in ovarian cancer (OC) is still unclear.

Methods: The correlation between m6A and APA was analyzed by using RNA methylation sequencing of OC cells and single-cell sequencing of clinical samples from public databases. To explore the core regulatory factors that served as a bridge between m6A and APA, we employed RNA pull-down with biotin-labelled m6A, immunoprecipitation, mass spectrometry, western blot, protein purification and GST pull-down assays. Furthermore, the important target genes were screened by PAS-seq, eCLIP-seq, RIP-seq and meRIP-seq, and verified by RT-qPCR, 3'RACE, RNA stability, and dual luciferase reporter assays. Multiple phenotypic experiments were conducted to evaluate the function of the IGF2BP2-PUM2 axis in vitro and in vivo.

Results: This study found that the m6A was correlated with the APA and affected the 3'end processing in OC. The APA regulator CPSF6 tended to bind the m6A-modified transcripts in OC cells. Mechanistically, we demonstrated that the m6A reader IGF2BP2 KH1-4 domains could directly bind to the CPSF6-RS domain to regulate the 3'end processing of OC. Furthermore, sequencing revealed that the m6A was highly enriched in the 3'UTR near the proximal polyadenylation signal (PAS), which promotes the use of proximal PAS and leads to 3'UTR shortening. PUM2 was carried m6A and recognized by IGF2BP2, and CPSF6 was recruited at the proximal polyadenylation signal (pPAS) to generate the short-3'UTR transcript. The short PUM2 transcript was more stable than the long transcript, which promoted the malignant progression of OC.

Conclusions: We revealed a novel mechanism in which the m6A could regulate the APA processing of pre-mRNAs by crosstalk of IGF2BP2 and CPSF6. This study provides a potential strategy for the effective treatment of OC.

Highlights: The interaction between m6A and APA is mediated by the m6A regulator IGF2BP2 and the APA factor CPSF6. The transcripts harboring m6A modification tend to use the proximal polyadenylation signal (PAS) in ovarian cancer (OC). PUM2 promotes the malignant progression of OC through its m6A methylation and APA processing.

Keywords: CPSF6; IGF2BP2; N6‐methyladenosine; PUM2; alternative polyadenylation; ovarian cancer.

FIGURE 1

The relationship between m6A and APA in OC. (A) Combined analysis of meRIP‐seq and scRNA‐seq of APA data in OC. (B) Peak analysis of m6A modification and APA processing. (C) The peak map of meRIP‐seq. (D) The binding motif of meRIP‐seq. (E, F) PAS‐seq analysis of METTL3 knockdown in OVCAR3, the volcano map of long and short genes (E), and the proximal and distal PAS uses S‐shaped curve (F). (G) The PAS distribution of APA genes after METTL3 knockdown.

FIGURE 2

The APA factor CPSF6 interacts with the m6A regulator IGF2BP1/2/3 complex in OC cells. (A) CPSF6 was related to m6A modification according to the RNA‒protein pulldown assay. (B) The binding proteins of CPSF6 were detected by IP/MS in OVCAR3 cells. (C) List of the top 10 proteins for CPSF6‐IP. (D) IP assay using a CPSF6 antibody with or without RNaseA to detect the IGF2BP1, IGF2BP2 and IGF2BP3 proteins in OC cells. (E) IP assay of IGF2BP1/2/3 with or without RNaseA to detect CPSF6. (F) The distribution of proteins was detected by nucleocytoplasmic separation. (G) The colocalization of the IGF2BP1/2/3 complex and CPSF6 was detected by an immunofluorescence assay.

FIGURE 3

CPSF6 RS domain directly binds to the C‐terminal KH domains of IGF2BP2. (A) Co‐IP assay of Flag‐CPSF6 truncated and Myc‐IGF2BP1/2/3 in 293T cells. (B, C) Schematic diagram of CPSF6 truncated (B) and IGF2BP2 truncated (C) proteins. (D) Western blot analysis of GST‐CPSF6‐RS or His‐IGF2BP1/2/3 proteins of GST‐input and pulldown samples. (E) Co‐IP assay of Myc‐IGF2BP2 truncated and Flag‐CPSF6 in 293T cells. (F) The structure and site of the CPSF6 RS‐IGF2BP2 interaction complex were predicted by AlphaFold3.

FIGURE 4

Joint analysis of multiple sequencing for screening target genes in OVCAR3 cells. (A, B) PAS‐seq analysis of IGF2BP2 knockdown in OVCAR3, the volcano map of long and short genes (A), and the proximal and distal PAS uses a S‐shaped curve (B). (C) The PAS distribution of short and long APA genes after IGF2BP2 knockdown. (D) The volcano plot of IGF2BP2 binding genes by RIP‐seq. (E) Overlap of the CPSF6 binding target genes. (F) The enrichment motif analysis of CPSF6 eCLIP‐seq. (G) Venn diagram showing the overlap of the multiple sequencing results. (H) The peak map of CPSF6 binding sites. (I) Joint analysis of m6A‐seq and CPSF6 eCLIP‐seq.

FIGURE 5

METTL3 and IGF2BP2 regulate the APA process to affect the stability of PUM2 with m6A‐dependent and CPSF6‐binding manner in OC cells. (A, B) The expression ratio of long and short transcripts of target genes after knockdown of METTL3 (A) and IGF2BP2 (B) in OC cells, respectively. (C–F) The interaction between 3′UTR of PUM2 mRNA and CPSF6 (C), METTL3 (D), IGF2BP2 (E), and m6A (F) antibodies was validated by using eCLIP‐seq, RIP‐PCR and meRIP‐seq. (G) Distribution of the CPSF6 binding sites across PUM2 mRNA transcript as identified by eCLIP‐seq. (H) Distribution of m6A peaks across PUM2 mRNA as identified by meRIP‐seq. (I) IGV tracks showing the enrichment of PUM2 mRNA according to PAS‐seq after METTL3 and IGF2BP2 knockdown in OVCAR3 cells. (J) The abundance of PUM2 long and short transcripts was identified by 3′RACE assay in OC cells with METTL3 or IGF2BP2 knockdown. (K) The expression of PUM2 was detected by western blot after knocking down METTL3 and IGF2BP2 in OC cells. (L) Decay curves and half‐life (T1/2) of PUM2 mRNA in OVCAR3 cells with METTL3 and IGF2BP2 knockdown were derived from mRNA stability profiling. (M) The interaction between CPSF6 and PUM2 mRNA was detected by RIP‐qPCR following METTL3 and IGF2BP2 knockdown in OVCAR3 cells. (N) Scheme of m6A sites WT and MUT nucleotide sequence surrounding PUM2 pPAS in 3′UTR, and dual luciferase reporter assay performed by cotransfecting PUM2 3′UTR WT or MUT and IGF2BP2 plasmid.

FIGURE 6

Downregulation of PUM2 expression inhibits the growth and metastasis of OC cells in vitro and in vivo. (A) The knockdown efficiency of PUM2 in OC cells was determined by western blot. (B–D) The growth, proliferation, and metastasis of OC cells were detected by CCK‐8 (B), colony formation (C) and Transwell (D) assays after knocking down PUM2. (E–G) The subcutaneous tumorigenesis model of gross xenograft anatomy (E), growth curve (F) and tumour weight (G) were tested in which PUM2 was knocked down in OVCAR3 cells. (H–J) For the intraperitoneal metastasis model, PUM2‐knockdown OVCAR3 cells were injected into nude mice, and the graphs show the gross anatomy of the abdominal cavity (H), number of metastatic nodules (I) and metastatic tumour weight data (J). (K) HE staining was performed to evaluate tissue morphology, and IHC was performed to visualize Ki‐67‐ and Caspase‐3‐positive staining in xenografted tumours. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 7

PUM2 serves as a functionally essential target of IGF2BP2 in OC cells. (A–C) The ability of growth, proliferation, and metastasis was detected using CCK8 (A), clone formation (B) and transwell (C) when knockdown IGF2BP2 in OC cells and then overexpressing PUM2. (D–F) Overexpression of PUM2 reversed the growth (D), proliferation (E), and metastasis (F) of OC cells in which IGF2BP2 inhibitor CWI1‐2 treatment. (G) Detection of protein expression levels of PUM2, METTL3, and IGF2BP2 in 12 pairs of OC and normal fallopian tube epithelial tissue samples. (H) Pearson's correlation plot of the expression of METTL3 and IGF2BP2 with that of PUM2 in OC patients from the TCGA database. *p < 0.05, **p < 0.01, ***p < 0.001. ns, not significant.