Nuclear accumulation of YTHDF1 regulates mRNA splicing in the DNA damage response

Abstract

YTH domain–containing family protein 1 (YTHDF1), a reader of N6-methyladenosine (m⁶A), has been implicated in regulating RNA metabolism in the cytosol. Here, we report a role of YTHDF1 within the nucleus in response to genotoxic stress. Upon radiation, YTHDF1 is phosphorylated at serine-182 in an ataxia telangiectasia and Rad3-related–dependent manner. This phosphorylation inhibits exportin 1–mediated nuclear export of YTHDF1, resulting in its accumulation within the nucleus. Nuclear YTHDF1 enhances the binding capacity of serine- and arginine-rich splicing factor 2 to a group of m⁶A-modified exons, leading to increased exon inclusion. Specifically, YTHDF1 promotes splicing and expression of DNA repair genes, such as BRCA1 and TP53BP1, thereby mitigating excessive DNA damage. Depletion of YTHDF1 sensitizes cancer cells to radiation treatment. Together, our study reveals a crucial role of YTHDF1 in m⁶A-mediated messenger RNA splicing in the DNA damage response, proposing it as a potential target for radiation therapy.

In response to x-ray irradiation, YTHDF1 accumulates in the nucleus to regulate the splicing and expression of DNA repair genes.

INTRODUCTION

Our genomes are constantly subjected to various sources of endogenous (e.g., DNA replication errors or reactive oxygen species generated as by-products of cellular metabolism) and exogenous (e.g., ionizing radiation) assaults. To protect our genomes, cells have evolved a sophisticated network of biochemical pathways, collectively termed the “DNA damage response” (DDR) (1). DDR orchestrates DNA repair with cell cycle checkpoint activation and other cellular responses, including apoptosis, transcription, and pre-mRNA splicing (2). Defects in DDR lead to cell genomic instability, which is closely associated with cancer development. However, these defects also provide therapeutic vulnerabilities due to the heightened dependence of cancer cells on DDR (3–6).

N6-methyladenosine (m⁶A), the most prevalent mRNA modification, has been defined as another layer of gene expression regulation. m⁶A modification is dynamically regulated by methyltransferases (“writers”) and demethylases (“erasers”). The writers, comprising core methyltransferase components methyltransferase-like 3 (METTL3), METTL14, and their cofactors (WTAP, RBM15, HAKAI, VIRMA, and ZC3H13), promote the deposition of m⁶A on mRNAs (7–9). The removal of m⁶A relies on erasers, including fat mass and obesity-associated (FTO) and AlkB homolog 5 (ALKBH5) (10, 11). The reader proteins, such as YTH domain proteins and insulin-like growth factor 2 mRNA-binding proteins (IGF2BPs), regulate the metabolism of m⁶A-modified mRNAs, including mRNA splicing, degradation, and translation (12–14). YTH proteins have a deep pocket recognizing the methyl group of m⁶A, while such feature is absent in IGF2BPs (12–14). Dynamically regulated by writers, erasers, and readers, m⁶A modification plays crucial roles in various physiological processes (15, 16).

Recently, the role of m⁶A mRNA modification in DDR has emerged (1). m⁶A has been observed at single-strand breaks induced by ultraviolet (UV) irradiation, aiding in the recruitment of DNA polymerase κ for DNA repair (17). METTL3-mediated m⁶A modification also influences the repair of double-strand breaks (DSBs). Upon DSBs, the ataxia telangiectasia mutated (ATM) phosphorylates METTL3, leading to its recruitment to DSB sites and resulting in m⁶A modification of nascent RNA. m⁶A-modified RNAs form DNA-RNA hybrids at DSBs, modulating homologous recombination (HR)–mediated DSB repair (18). Beyond its direct role in DNA repair, m⁶A affects the expression of genes involved in DNA repair. METTL14 has been shown to promote UV-induced global genome repair by regulating m⁶A-mediated translation of damaged DNA binding protein 2 (19). Furthermore, m⁶A mRNA methylation influences the expression of DNA repair genes, affecting the sensitivity of cancer cells to chemotherapy (20, 21). However, the mechanisms through which the m⁶A machinery orchestrates gene expression in the DDR remain to be explored.

Here, we report that YTH domain–containing family protein 1 (YTHDF1), an m⁶A reader, is a nucleus-cytosol shuttling protein. In response to DNA damage, phosphorylation of YTHDF1 induces its nuclear retention. Nuclear YTHDF1 recruits the splicing factor serine- and arginine-rich splicing factor 2 (SRSF2) and promotes exon inclusion of DDR genes, facilitating DNA repair. Furthermore, high expression of YTHDF1 in cancer cells increases resistance to radiation and results in poor cancer treatment outcomes.

RESULTS

X-ray irradiation induces nuclear localization of YTHDF1

To explore the potential function of m⁶A modification in DDR, a dose of 4–gray (Gy) x-ray, which is sufficient to induce DNA damage in various cell lines, was used to irradiate human embryonic kidney (HEK) 293T cells (22). Using immunofluorescence staining, we examined the subcellular localization of the m⁶A machinery before and after x-ray irradiation. Under normal conditions, the m⁶A writer METTL3, the erasers FTO/ALKBH5, and the readers YTHDC1/2 were predominantly localized in the nucleus, while the readers YTHDF1/2/3 were localized in the cytosol. One hour after x-ray irradiation, the m⁶A writers, erasers, and most readers exhibited no notable change in their intracellular localization (fig. S1A). However, the YTHDF1 protein displayed a notable nuclear localization (Fig. 1A). This phenomenon was consistently validated by subcellular protein fractionation experiments (Fig. 1B) and was also confirmed in U2OS cells (fig. S1B). These findings suggested the possible involvement of YTHDF1 in DDR.

Fig. 1. ATR-mediated phosphorylation promotes YTHDF1 nuclear localization in response to x-ray irradiation.

(A) Immunofluorescence detection of YTHDF1 using anti-YTHDF1 in HEK293T cells with or without x-ray treatment (a dose of 4 Gy). Nuclei were counterstained with Hoechst (blue). Scale bars, 10 μm. CTL, control. DAPI, 4′,6-diamidino-2-phenylindole. (B) Immunoblotting analysis of total, cytosolic (Cyto), and nuclear (Nuc) fractions of HEK293T cells with or without x-ray treatment (a dose of 4 Gy). (C) Immunoblotting analysis confirming the efficiency of ATR and ATM inhibitors (ATRi and ATMi). HEK293T cells treated with KU60019 (10 μM) or VE-821 (10 μM) for 24 hours and x-ray were subjected to immunoblotting analysis. (D) Immunoblotting analysis confirming the efficiency of DNA-PKcs inhibitor (DNA-PKcsi). HEK293T cells treated with NU7441 (2 μM) for 24 hours and x-ray were subjected to immunoblotting analysis. (E) Subcellular localization of FLAG-tagged YTHDF1 under the same conditions as (C) and (D). Lentiviral transduction was used for the expression of YTHDF1-FLAG. (F) Validation of ATR knockdown efficiency by immunoblotting in HEK293T cells. (G) Intracellular localization of YTHDF1 in control and ATR knockdown cells with x-ray treatment (a dose of 4 Gy). (H) Coimmunoprecipitation experiment confirming the interaction between ATR and YTHDF1. Lysates from HEK293T cells with or without x-ray treatment (a dose of 4 Gy) were immunoprecipitated with immunoglobulin G (IgG) or anti-ATR and then immunoblotted with anti-YTHDF1 antibody. IP, immunoprecipitation. (I) Intracellular localization of alanine mutants of YTHDF1 in HEK293T cells. A dose of 4-Gy x-ray irradiation was applied. (J) In vitro kinase assay using antibody-affinity–isolated ATR complex with purified YTHDF1 or YTHDF1-S182A. The phosphorylation was detected using a p-serine antibody (top). YTHDF1 and ATR were detected by Western blot (middle) and Coomassie blue staining (bottom). GFP, green fluorescent protein.

Phosphorylation of YTHDF1 at serine-182 by ATR promotes its nuclear localization

ATM, ataxia telangiectasia and Rad3–related (ATR), and DNA-dependent protein kinase catalytic subunit (DNA-PKcs) are the principal DDR kinases situated at the forefront of the DDR in eukaryotic cells, mediating the DDR signaling cascade in response to various types of DNA lesions (2). To determine which of these mediators contribute to the nuclear localization of YTHDF1, we treated HEK293T cells with ATM inhibitor KU60019 (10 μM), ATR inhibitor VE-821 (10 μM), or DNA-PK inhibitor NU7441 (2 μM) for 24 hours. ATR inhibitor, but neither ATM inhibitor nor DNA-PKcs inhibitor, effectively blocked the nuclear localization of YTHDF1 (Fig. 1, C to E, and fig. S1C). Similar results were obtained when we used short hairpin RNAs (shRNAs) to knock down ATR expression (Fig. 1, F and G). Hydroxyurea or camptothecin (CPT), both of which induce replication stress and activate ATR, could also concentrate YTHDF1 in the nucleus (fig. S1D). Collectively, these data provided evidence that ATR activation is indispensable for the nuclear accumulation of YTHDF1 under x-ray irradiation.

We proposed that YTHDF1 is phosphorylated in an ATR-dependent manner. Our coimmunoprecipitation experiments detected an association between YTHDF1 and ATR following x-ray irradiation (Fig. 1H). Prior research had identified several phosphorylation sites on YTHDF1 (PhosphoSitePlus). Six phosphorylation sites, Y116, S182, T273, S291, S350, and S398, were reported by at least two different references. To determine which of these residues mediate YTHDF1’s nuclear localization, we substituted each residue with alanine via site-directed mutagenesis. When these mutants were expressed in HEK293T cells, the S182A mutant failed to translocate into the nucleus upon x-ray irradiation (Fig. 1I).

To investigate whether YTHDF1 serves as a direct substrate for ATR, we conducted an in vitro kinase assay using purified ATR and recombinant YTHDF1. Data showed that ATR phosphorylated wild-type (WT) YTHDF1 but not YTHDF1-S182A (Fig. 1J), affirming that S182 of YTHDF1 is indeed a genuine phosphorylation site for ATR. Collectively, these findings strongly suggested that ATR-mediated phosphorylation of YTHDF1 at S182 facilitates its nuclear localization.

Phosphorylation of YTHDF1 impedes exportin 1–dependent nuclear export

Examination of the amino acid sequence in proximity to S182 unveiled the presence of a consensus nuclear export signal (NES), which exhibited high evolutionary conservation across various species (Fig. 2A). This NES contains three hydrophobic amino acids: methionine-187 (M187), leucine-190 (L190), and isoleucine-192 (I192). Mutagenesis of two hydrophobic residues within this putative NES leads to the nuclear localization of YTHDF1 (Fig. 2B). Nuclear export of NES-bearing proteins is typically mediated by exportin 1 [XPO1, also known as chromosomal maintenance 1 (CRM1)]. The physical interaction between CRM1 and YTHDF1 was validated by coimmunoprecipitation assays (Fig. 2C and fig. S1E). To corroborate the presence of an NES in YTHDF1, cells were subjected to leptomycin B (LMB), a drug known to impede the interaction between CRM1 and NES-bearing proteins. In response to 1-hour LMB treatment, YTHDF1 exhibited nuclear accumulation in HEK293T cells (Fig. 2D), suggesting that CRM1 mediates the nuclear export of YTHDF1. Similar findings were validated in U2OS cells (fig. S1B).

Fig. 2. Phosphorylation blocks CRM1-dependent YTHDF1 nuclear export.

(A) The consensus NES of YTHDF1 in representative species. The hydrophobic residues are shown within red frames, while the phosphorylation site is shown within a blue frame. (B) Subcellular localization of FLAG-tagged YTHDF1 mutants in HEK293T cells. Nuclei were counterstained with Hoechst (blue). Scale bars, 10 μm. (C) Coimmunoprecipitation analysis of CRM1 interaction with YTHDF1. Lysates from HEK293T cells expressing FLAG-CRM1 were subjected to immunoprecipitation using either IgG or anti-FLAG antibody, followed by immunoblotting with an anti-YTHDF1 antibody. (D) Intracellular localization of endogenous YTHDF1 in HEK293T cells with or without 1-hour LMB treatment. Scale bars, 10 μm. (E) Coimmunoprecipitation analysis of FLAG-CRM1 interaction with YTHDF1 with or without x-ray treatment (a dose of 4 Gy). (F) Coimmunoprecipitation analysis of endogenous CRM1 interaction with YTHDF1 with or without x-ray treatment (a dose of 4 Gy). (G) Coimmunoprecipitation analysis of CRM1 interaction with YTHDF1 or S182A mutant. HEK293T cells transfected with the indicated plasmids were treated with or without x-ray irradiation (4 Gy). Cell lysates were subjected to immunoprecipitation using an anti–hemagglutinin (HA) antibody, followed by immunoblotting with both anti-HA and anti-FLAG antibodies.

Considering YTHDF1’s nuclear retention under x-ray irradiation, we postulated that the interaction between CRM1 and YTHDF1 might be compromised. This hypothesis was confirmed by coimmunoprecipitation assays in HEK293T cells. YTHDF1-CRM1 binding was notably reduced when cells were exposed to x-ray irradiation (Fig. 2, E and F). Moreover, YTHDF1-WT exhibited diminished binding affinity to CRM1 than YTHDF1-S182A in response to x-ray irradiation (Fig. 2G). These data strongly suggested that S182 phosphorylation attenuates the interaction between YTHDF1 and CRM1.

Nuclear YTHDF1 promotes DNA repair

Phosphorylated histone H2AX (γH2AX) is a specific and sensitive molecular marker for monitoring DNA damage (23). To study the role of YTHDF1 in DDR, we generated YTHDF1–knockout (KO) HEK293T cell lines and assessed DNA damage by detecting γH2AX foci (Fig. 3A and fig. S2A). Depletion of YTHDF1 led to an increased accumulation of γH2AX foci, both under normal conditions and after x-ray irradiation (Fig. 3, B and C, and fig. S2B). Knockdown of YTHDF1 also increased x-ray–induced γH2AX formation in U2OS cells (fig. S2C). In addition, YTHDF1-KO cells displayed increased tail moment in neutral comet assays (Fig. 3, D and E). These findings suggested that the absence of YTHDF1 impairs DNA damage repair.

Fig. 3. YTHDF1 is required for DNA repair.

(A) Immunoblotting analysis of γH2AX levels in WT and YTHDF1-KO HEK293T cells treated with or without x-ray (4 Gy). (B) Immunofluorescence detection of γH2AX foci under the same conditions as (A). Scale bars, 10 μm. (C) Quantification of the number of γH2AX foci per cell in (B). (D) Neutral comet assays for detecting DNA damage under the same conditions as (A). Scale bars, 10 μm. (E) Quantification of the relative tail moments in (D) (n > 50 cells). (F) Schematic of the reporter system for the analysis of HR-mediated DSB repair. (G) Schematic of the reporter system for the analysis of NHEJ-mediated DSB repair. (H) Quantification of the frequency of HR-mediated double-strand break repair in WT and YTHDF-KO cells. RFP, red fluorescent protein. (I) Quantification of the frequency of NHEJ-mediated double-strand break repair in WT and YTHDF-KO cells. (J) Cell proliferation assay of WT and YTHDF1-KO cells treated with or without x-ray (8 Gy; n = 5) using cell counting kit–8 (CCK-8). (K) Immunoblotting analysis of γH2AX level in the indicated cells treated with or without x-ray (4 Gy). YTHDF1-KO cells were reconstituted with either WT YTHDF1 or S182A mutant. (L) γH2AX foci number in WT cells, YTHDF-KO cells, or KO cells reconstituted with WT YTHDF1 or S182A mutant under x-ray treatment (4 Gy). (M) The relative tail moment in neutral comet assays under the same conditions as (L). (N) Cell proliferation assays of WT cells, YTHDF-KO cells, or KO cells reconstituted with WT YTHDF1 or S182A mutant under x-ray treatment (8 Gy) using CCK-8. All values represent the means ± SEM of three independent experiments. Statistical significance was determined using unpaired two-tailed Student’s t test (*P < 0.05, **P < 0.01, and ***P < 0.001). ns, not significant.

DSBs can be repaired through error-free HR or error-prone nonhomologous end joining (NHEJ) (24). To investigate whether YTHDF1 regulates these repair pathways, we used HR and NHEJ reporter systems. The HR reporter introduced DSBs using the I–Sce I endonuclease at an integrated direct repeat green fluorescent protein (DR-GFP) gene (Fig. 3F), while the NHEJ reporter used the two broken ends of a GFP gene separated by an adenovirus exon digested with Hind III (Fig. 3G). Cells were transfected with each reporter along with the p-Cherry expression vector as an internal control. The results demonstrated that YTHDF1 depletion notably inhibited both HR-mediated and NHEJ-mediated DSB repair (Fig. 3, H and I, and fig. S2D). Accordingly, YTHDF1 deletion markedly reduced cell viability after x-ray irradiation (Fig. 3J), affirming that YTHDF1 plays a crucial role in promoting DNA repair to facilitate cell survival. Reexpression of YTHDF1-WT, but not YTHDF1-S182A, effectively reduced γH2AX foci formation (Fig. 3, K and L, and fig. S2E) and comet tail moment (Fig. 3M and fig. S2F) and improved cell viability (Fig. 3N). These results suggested that nuclear localization of YTHDF1 is essential for an effective DDR.

YTHDF1 interacts with spliceosome components

To uncover the mechanisms through which YTHDF1 facilitates DNA repair, we immunoprecipitated YTHDF1 interactome from cells expressing FLAG-tagged YTHDF1 that were treated with x-ray, followed by mass spectrometry (MS) analysis (Fig. 4A). After excluding nonspecific interactions in the control group, we successfully identified a total of 180 potential YTHDF1-binding proteins (table S1). Our MS analysis captured CRM1 (XPO1) in the YTHDF1 interactome, supporting a role of CRM1 in YTHDF1 nuclear exportation (table S1). Several importins, importin subunit α–1, importin subunit β–1, and importin-5, were identified to interact with YTHDF1, implying that YTHDF1 undergoes nuclear import (table S1).

Fig. 4. YTHDF1 interacts with spliceosome components within nuclear speckles.

(A) Schematic representation illustrating the strategy for the identification of YTHDF1-interacting proteins by LC/MS-MS. m/z, mass/charge ratio. (B) Functional enrichment analysis of 180 YTHDF1-interacting proteins based on the Database for Annotation, Visualization and Integrated Discovery analysis. ER, endoplasmic reticulum. (C) Overview of the assembly of the major spliceosome complex. (D) The spliceosome components that interact with YTHDF1. (E) Coimmunoprecipitation assay demonstrating the interaction between YTHDF1 and SF3B1, SF3B3, and SRSF2. HEK293T cells expressing FLAG-tagged YTHDF1 were treated with or without x-ray (4 Gy). Lysates were immunoprecipitated using IgG or anti-FLAG antibodies and subsequently analyzed by immunoblotting with the indicated antibodies. RNA digestion was performed using RNase A (5 μg/ml). (F) Impairment of the interaction between YTHDF1 and spliceosome components upon mutation of S182A. HEK293T cells expressing FLAG-YTHDF1 or S182A mutant were exposed to x-ray irradiation, followed by immunoprecipitation using an anti-FLAG antibody. (G) Colocalization of YTHDF1 with SRSF2 within nuclear speckles in response to x-ray treatment (4 Gy). Scale bars, 10 μm. a.u., arbitrary units.

Bioinformatic analysis revealed a notable enrichment of YTHDF1-interacting proteins in the ribosome pathway, consistent with YTHDF1’s known role in translational regulation (13). The spliceosome pathway also exhibited enrichment, ranking second only to the ribosome pathway (Fig. 4B, fig. S3A, and table S2). More than a dozen spliceosome factors were captured, including key components of the U2 small nuclear ribonucleoprotein (snRNP) complex, such as SF3B1 and SF3B3. Notably, SRSF2 (also known as SC35), a critical component of nuclear speckles (25, 26), was also identified as a YTHDF1-interacting partner (Fig. 4, C and D). Independent coimmunoprecipitation experiments further confirmed the association of SC35, SF3B1, and SF3B3 with both FLAG-YTHDF1 (Fig. 4E) and endogenous YTHDF1 (fig. S3B). hnRNPH3, a splicing factor that was not identified in our MS experiment, did not interact with YTHDF1, implying that YTHDF1 specifically interacts with certain splicing factors. The interactions were not disrupted by ribonuclease A (RNase A) digestion (fig. S3C), excluding RNA-dependent association. As anticipated, these interactions were only observed with WT YTHDF1 and not with the S182A mutant (Fig. 4F). Immunofluorescence experiments demonstrated colocalization of YTHDF1 with SRSF2 in nuclear speckles in response to x-ray irradiation (Fig. 4G). These findings strongly suggested the role of nuclear YTHDF1 in pre-mRNA splicing.

YTHDF1 enhances the splicing of DNA repair genes in an m⁶A-dependent manner

Given the aforementioned evidence, we used RNA sequencing (RNA-seq) to quantify the impact of YTHDF1 on alternative splicing (AS) events. Depletion of YTHDF1 affected various types of AS events, including skipped exons (SEs), alternative 5′ splice sites, alternative 3′ splice sites, retained introns, and mutually exclusive exons (fig. S3D). Among these AS events, we focused on SEs due to their prevalence and abundance (fig. S3D). By using the percent spliced-in (PSI) metric, we quantified the inclusion level of individual cassette exons. Most cassette exons exhibited decreased PSI in response to YTHDF1 depletion, indicating YTHDF1’s propensity to promote cassette exon inclusion.

To identify YTHDF1-binding sites, we performed m⁶A sequencing (m⁶A-seq), along with cross-linking and immunoprecipitation sequencing (CLIP-seq), on nuclear YTHDF1. The m⁶A-seq identified 41,920 m⁶A peaks across within 8745 transcripts (Fig. 5A). Metagene analysis showed that the m⁶A peaks are primarily located around the stop codon (fig. S3E). The canonical RGACH motif was identified in m⁶A-seq peaks (fig. S3F) (27). The CLIP-seq identified 8257 YTHDF1-binding peaks within 5803 transcripts. By integrating m⁶A-seq data with CLIP-seq data, we identified 3129 m⁶A–containing transcripts directly bound by YTHDF1 (Fig. 5A and table S3). Most RNAs bound by YTHDF1 belonged to the mRNA and long non-coding RNA (lncRNA) categories (Fig. 5B), constituting 62.7 and 32.2%, respectively. The binding regions of YTHDF1 in mRNAs were highly enriched in coding sequence (CDS) regions, with less enrichment in untranslated regions (UTRs) (Fig. 5C). The mRNA-binding regions of nuclear YTHDF1 are distinct from cytosolic YTHDF1 as previously reported (13), suggesting that they have different functions.

Fig. 5. YTHDF1 enhances the retention of m⁶A-marked exons.

(A) Venn diagram illustrating the overlap between m⁶A-containing transcripts (m⁶A targets) and YTHDF1-binding transcripts (CLIP targets). (B) Percentage representation of diverse RNAs bound by YTHDF1 based on CLIP-seq. Most binding clusters of YTHDF1 are located within mRNAs. (C) Distribution of YTHDF1-binding sites across four mRNA segments (intron, CDS, 5′UTR, and 3′UTR). (D) Boxplot depicting PSI changes following YTHDF1-KO for non-m⁶A exons, m⁶A-targeted exons, and YTHDF1-targeted exons. (E) Violin plot displaying exon expression changes following YTHDF1-KO for non-m⁶A exons, m⁶A-targeted exons, and YTHDF1-targeted exons. The upper and lower quartiles, as well as the median, are indicated for each group. (F) Functional enrichment analysis of transcripts in (C). (G) Integrative Genomics Viewer (IGV) tracks exhibiting the read coverage of BRCA1 and TP53BP1 genes from RNA-seq data of control (purple) and YTHDF1-KO (green) cells, along with CLIP-seq data (gray) and m⁶A data (blue). The RRACH sites located within m⁶A peaks are indicated. The normalized read number in the peak summit was indicated. (H) RT-PCR results and statistical analysis of exon inclusion level of BRCA1 and TP53BP1 in WT cells, YTHDF-KO cells, or KO cells reconstituted with WT YTHDF1 or S182A mutant under x-ray treatment (4 Gy). The ACTB gene was used as an internal control. (I) The mRNA level of BRCA1, TP53BP1, and RAD51 in WT cells, YTHDF-KO cells, or KO cells reconstituted with WT YTHDF1 or S182A mutant under x-ray treatment (4 Gy). (J) Immunoblotting analysis of WT cells, YTHDF-KO cells, or KO cells reconstituted with WT YTHDF1 or S182A mutant under x-ray treatment (4 Gy). Statistical significance was determined using the Mann-Whitney test (*P < 0.05, **P < 0.01, and ***P < 0.001).

The m⁶A-modified exons and YTHDF1-targeted exons showed a statistical reduction in PSI compared to non-m⁶A exons following YTHDF1 depletion (Fig. 5D and table S4), indicating a role of YTHDF1 in exon retention. Consequently, sequencing read coverage of m⁶A-containing exons and YTHDF1-targeted exons statistically diminished compared to non-m⁶A exons following YTHDF1 depletion (Fig. 5E and table S5), indicative of a preference for exon skipping.

Kyoto Encyclopedia of Genes and Genomes pathway analysis of YTHDF1 target genes enriched several key mediators involved in DNA repair, such as breast cancer gene 1 (BRCA1), tumor protein p53 binding protein 1 (TP53BP1), and RAD51 recombinase (RAD51) (Fig. 5F and table S6). The corresponding RNA-seq, CLIP-seq, and m⁶A-seq data for these genes are presented in Fig. 5G and fig. S3G. Reverse transcription polymerase chain reaction (RT-PCR) experiments confirmed that YTHDF1 depletion led to the skipping of alternative exons in these DNA repair genes (BRCA1, TP53BP1, and RAD51) (Fig. 5H and fig. S3H). Knockdown of either METTL3 or METTL14 induces exon skipping of YTHDF1 target genes (fig. S3, I and J), indicating that YTHDF1 relies on m⁶A modification to regulate AS. The defective splicing of target mRNAs in YTHDF1-KO cells could be rescued by WT YTHDF1 but not S182A mutant (Fig. 5H and fig. S3H). The skipping of exons may introduce frameshifts or premature termination, which induce the degradation of these mRNAs. We detected reduced expression of BRCA1, TP53BP1, and RAD51 in YTHDF1-KO cells, which could be rescued by WT YTHDF1 but not S182A mutant (Fig. 5, I and J). Collectively, these findings support the critical role of nuclear YTHDF1 in promoting exon inclusion of DNA repair genes following x-ray irradiation.

YTHDF1 regulates AS through SRSF2

SRSF family members recognize pre-mRNA cis-acting elements known as exonic splicing enhancers, which facilitate the selection of splice sites (28). Given our confirmation of a physical interaction between YTHDF1 and SRSF2 (Fig. 4E), we hypothesized that YTHDF1 regulates AS through SRSF2. First, we evaluated changes in PSI using RNA-seq after knockdown of SRSF2 (Fig. 6A). As anticipated, YTHDF1-targeted exons exhibited a more substantial decline in ΔPSI compared to nontargeted exons following SRSF2 knockdown (Fig. 6B and table S7).

Fig. 6. YTHDF1 facilitates SRSF2 binding to m⁶A-modified transcripts.

(A) Verification of SRSF2 knockdown (shSRSF2 #1 and shSRSF2 #2) efficiency in HEK293T cells. (B) Boxplot illustrating the PSI change in non-m⁶A exons, m⁶A-containing exons, and YTHDF1-targeted exons following SRSF2 knockdown (shSRSF2 #1). (C) Violin plot displaying the binding affinity of SRSF2 in YTHDF1-KO cells relative to the control for non-m⁶A exons, m⁶A-containing exons, and YTHDF1-targeted exons. The upper and lower quartiles, as well as the median, are indicated for each group. (D) Immunoblotting analysis of m⁶A-modified RNAs captured by FLAG-SRSF2 in the WT and YTHDF1-KO cells. WT or KO cells expressing FLAG-SRSF2 were cross-linked with UV, treated with micrococcal nuclease, and immunoprecipitated by FLAG antibody. The immunoprecipitates were subjected to immunoblotting analysis with m⁶A antibody. (E and F) IGV tracks exhibiting the read coverage of BRCA1 (E) and TP53BP1 (F) genes from RNA-seq data and SRSF2 RIP sequencing (RIP-seq) data. The m⁶A peaks are indicated. The normalized read number in the peak summit was indicated. (G and H) RT-PCR results and statistical analysis of exon inclusion level of BRCA1 (G) and TP53BP1 (H) in control and SRSF2-knockdown cells. The ACTB gene was used as an internal control. Statistical significance was determined using the Mann-Whitney test (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001).

To investigate whether YTHDF1 regulates the RNA binding capacity of SRSF2, we conducted RNA immunoprecipitation (RIP) assays for SRSF2 in control cells and YTHDF1-depleted cells under x-ray treatment. The binding affinity of SRSF2 to YTHDF1-targeted exons substantially diminished in YTHDF1-KO cells (Fig. 6C and table S8). This suggests that the altered RNA binding of SRSF2 can be attributed to the loss of SRSF2 recruitment by YTHDF1. Consistent with the sequencing analysis, the binding affinity of SRSF2 to m⁶A-modified RNAs notably decreased upon YTHDF1-KO (Fig. 6D).

The YTHDF1-targeted exons in representative genes (BRCA1, TP53BP1, and RAD51) were notably down-regulated upon SRSF2 knockdown (Fig. 6, E and F, and fig. S3K). Their binding to SRSF2 was also reduced in YTHDF1-KO cells (Fig. 6, E and F, and fig. S3K). Knockdown of SRSF2 induced exon skipping in these genes and decreased their expression (Fig. 6, G and H, and fig. S3, L and M). Collectively, these findings suggested that YTHDF1 regulates AS events by modulating SRSF2’s access to their target mRNAs.

YTHDF1 deficiency sensitizes tumors to radiotherapy

Cancer cells with deficiencies in DDR often exhibit increased sensitivity to DNA damage–based treatments such as cisplatin or irradiation. Clinical data have indicated that YTHDF1 expression is elevated in a variety of human cancers. Moreover, a higher level of YTHDF1 expression is associated with poorer survival rates in patients undergoing radiotherapy (fig. S4A). Therefore, we hypothesized that targeting YTHDF1 might sensitize tumors to radiotherapy or other DNA-damaging drugs. We tested this in a hepatocellular carcinoma model. Knockdown of YTHDF1 notably increased the sensitivity of HepG2 cells to x-ray irradiation (Fig. 7, A to C). We also confirmed that YTHDF1 knockdown sensitizes tumor cells to DNA-damaging drug CPT (fig. S4, B and C). Further, we demonstrated that tumor growth was more effectively suppressed in mice with YTHDF1-depleted xenograft tumors when subjected to x-ray treatment compared to the control group (Fig. 7, D to F). Notably, YTHDF1-deficient xenograft tumors exhibited increased formation of γH2AX (Fig. 7, G and H) and higher rates of cell apoptosis when exposed to radiation (Fig. 7, I and J). Splicing defects and reduced expression of BRCA1 and TP53BP1 were observed in YTHDF1-deficient tumor cells (fig. S4, D and E). We depleted YTHDF1 using a different shRNA and obtained consistent results (fig. S4, F to H). These results indicate that depletion of YTHDF1 notably sensitizes cancer cells to x-ray–induced DNA damage, ultimately leading to cell death and improved therapeutic outcomes in this mouse xenograft model system.

Fig. 7. YTHDF1 depletion enhances sensitivity to radiotherapy.

(A) Verification of YTHDF1 knockdown efficiency in HepG2 cells. (B) Cell proliferation assays conducted under normal conditions for control and YTHDF1-knockdown cells using CCK-8 (n = 5). (C) Cell proliferation assays for control and YTHDF1-knockdown cells subjected to x-ray treatment (8 Gy; n = 5) using CCK-8. (D) Experimental design depicting the radiotherapy procedure in a mouse model. (E) Representative image of the tumors from control (shCTL) and YTHDF1-knockdown (shDF1) groups with or without x-ray irradiation (4 Gy). (F) Quantitative analysis of the tumor size and tumor volume in (E) (n = 4). (G) Representative images of γH2AX staining in xenograft tumors subjected to x-ray treatment (4 Gy) and untreated controls with or without YTHDF1 depletion. Scale bar, 100 μm. (H) Quantitative analysis of γH2AX-positive cells in the samples shown in (G). (I) Representative images of terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL) staining in xenograft tumors subjected to x-ray treatment (4 Gy) and untreated controls with or without YTHDF1 depletion. Scale bar, 100 μm. (J) Quantitative analysis of the percentage of apoptotic cells (TUNEL-positive) in the samples presented in (I). All data are presented as means ± SEM from multiple independent experiments. Statistical significance was determined using an unpaired two-tailed Student’s t test (*P < 0.05, **P < 0.01, and ***P < 0.001).

DISCUSSION

The YTHDF family proteins, including YTHDF1, YTHDF2, and YTHDF3, are well-known m⁶A readers that have been implicated in mediating the effects of m⁶A in the cytosol, primarily affecting mRNA translation and degradation (12, 13, 29). Here, we identified a previously unrecognized function of YTHDF1 in the nucleus. Specifically, in the DDR, YTHDF1 was found to be enriched in spliceosome components and potentially localized to nuclear speckles containing splicing factors. As a result, YTHDF1 associates with m⁶A-modified exons and promotes their inclusion during pre-mRNA splicing. Collectively, our findings suggest a mechanism of m⁶A-dependent pre-mRNA splicing mediated by YTHDF1 during the DDR, which enhances the expression of DNA repair genes (fig. S5A).

Emerging evidence has indicated that m⁶A modification plays a role in regulating pre-mRNA AS. The role of m⁶A in mRNA splicing was also shown in Drosophila Sxl splicing, which is regulated by HAKAI and other key components of the m⁶A complex (30). In mammalian cells, acute depletion of METTL3 disrupts the inclusion of alternative introns/exons in the nascent transcriptome (31), and FTO-dependent m⁶A demethylation is involved in the regulation of RNA AS events in adipocytes (32). It is proposed that m⁶A affects the splicing outcome by influencing the RNA binding ability of pre-mRNA splicing factors, such as SRSF family members (32). Our data demonstrated that YTHDF1 acts as a bridge between SRSF2 and m⁶A in the context of DDR. This was supported by several key observations: a physical interaction between YTHDF1 and SRSF2, depletion of either SRSF2 or YTHDF1 leading to exon skipping in targeted transcripts, and a notable reduction in the affinity of m⁶A-marked transcripts for SRSF2 upon YTHDF1 depletion. These findings collectively suggest that YTHDF1 recruits SRSF2 to m⁶A-containing transcripts to facilitate efficient pre-mRNA splicing. Intriguingly, YTHDC1, a nuclear m⁶A reader protein, was reported to promote exon inclusion in targeted mRNAs through recruiting SRSF3 (33), underscoring the intricate mechanisms through which m⁶A modulates mRNA splicing, a process that could potentially vary depending on cell type or context (16).

Our data reveal that YTHDF1 is quickly transported to the cytosol by CRM1 under normal conditions. Upon x-ray treatment, activated ATR signaling phosphorylates YTHDF1 at S182, which prevents its nuclear export. Notably, S182 is not within ATR consensus motif, and whether other factors are needed for this phosphorylation process warrants further investigation. We have identified an NES within YTHDF1 responsible for regulating its cytosolic distribution. A recent interactome analysis aimed at identifying CRM1 binding partners has identified YTHDF1-3 as robust CRM1-interacting proteins (34), strongly suggesting that all three members of the YTHDF1 family exhibit cytosol-to-nucleus shuttling properties. Furthermore, accumulating evidence suggests that the shuttling of YTHDF family members is dynamically regulated and responds differently to various extracellular signals. For instance, heat shock stress induces nuclear localization of YTHDF2 (35), whereas x-ray irradiation, as demonstrated in our study, leads to nuclear localization of YTHDF1. Our research provides compelling evidence that ATR-dependent phosphorylation of YTHDF1 at S182 retains YTHDF1 within the nucleus by inhibiting the interaction between YTHDF1 and CRM1. These findings suggest that posttranslational modifications play a pivotal role in mediating the dynamic relocalization of YTHDF family members. Although YTHDF family members may function redundantly under certain conditions (29), studies have also highlighted distinct functions of these proteins. Sequence alignment showed that the YTH domains are highly conserved among YTHDF1/2/3, while the other regions are less conserved (fig. S5B). The amino acid difference in the N terminus of YTHDF proteins would enable distinct functions (36).

YTHDF1 has been found to be up-regulated in various human cancers, including colorectal cancer, gastric cancer, ovarian cancer, and hepatocellular carcinoma (37–40). It has been reported that YTHDF1 can promote carcinogenesis by facilitating the translation of oncogenic genes, such as TCF7L2, FZD7, or eIF3C (37, 38, 40). Given that YTHDF1 promotes DNA repair, it may contribute to resistance to chemotherapy and radiation-based cancer treatments. Clinical studies have shown that higher expression of YTHDF1 correlates with lower survival probability in various patients with cancer undergoing cisplatin and/or radiation therapy (fig. S4A). Our studies have demonstrated that depletion of YTHDF1 notably increases the sensitivity of hepatocellular carcinoma cells to irradiation both in vitro and in vivo (Fig. 7). In addition, YTHDF1 has been reported to promote chemoresistance of colorectal cancer cells and ovarian cancer cells (41, 42). Consequently, YTHDF1 may hold potential as a therapeutic target, and its inhibition or depletion could potentially improve the outcomes of cancer treatments.

In summary, our study reveals a role for YTHDF1 in mRNA splicing in the DDR. The combination of YTHDF1-regulated mRNA splicing, degradation, and translation allows for more precise and effective control of protein production during stresses. The relocalization of YTHDF1 from the cytosol to the nucleus and its functional transition shed light on our understanding of the regulation of RNA m⁶A modification.

MATERIALS AND METHODS

Cell culture and treatment

HEK293T and HepG2 cells were obtained from the American Type Culture Collection and cultured in Dulbecco’s modified Eagle’s medium (Gibco) with 10% fetal bovine serum (Biowest) and 1% penicillin/streptomycin (Sangon Biotech) at 37°C and 5% CO₂.

To generate DNA damage, cells were irradiated with 4- or 8-Gy x-ray and harvested 1 hour later. For chemical treatment, cells were treated with KU60019 (10 μM; Selleck), VE-821 (10 μM; Selleck), or NU7441 (2 μM; Selleck) for 24 hours.

Cell counting kit–8 (CCK-8) was used for cell viability assays. Briefly, cells were seeded at a density of 5.5 × 10³ cells per 100 μl per well in a 96-well plate. For the measurement, 10 μl of CCK-8 reagent was added to each well, and the absorbance at 450 nm was measured using a microplate reader. Cell viability was quantified by comparing the measurements to the baseline from day 1.

Cellular fractionation

Cells were collected and lysed with swelling buffer (25 mM Hepes, 0.5 mM MgCl₂, 10 mM KCl, and 0.5% NP-40). The mixture was incubated at 4°C for 15 min and centrifuged at 1000g for 5 min. The supernatant containing the cytoplasmic proteins was collected. The remaining cell pellet was lysed with radioimmunoprecipitation assay (RIPA) lysis buffer on ice for 30 min and centrifuged at 12,000g for 10 min. The supernatant containing the nuclear proteins was collected.

Generation of YTHDF1-KO cell line

To achieve YTHDF1 ablation, single-guide RNA targeting human YTHDF1 was subcloned into the vector lentiCRISPR v2 (52961, Addgene) at the Bsm BI site. HEK293T cells were transfected with the construct using Lipofectamine 2000 transfection reagent, followed by selection with puromycin (1 μg/ml) for 5 days.

Mice experiments

Mice were maintained and bred in specific pathogen–free conditions at the Animal Center of Zhejiang University. All animal studies were performed in compliance with the Guide for the Care and Use of Laboratory Animals by the Medical Experimental Animal Care Commission of Zhejiang University. For xenograft experiments, a total of 1 × 10⁶ control (shCTL) or YTHDF1-depleted (shYTHDF1) cells in 100 μl of phosphate-buffered saline (PBS) were mixed with an equal volume of Matrigel and then subcutaneously injected into the dorsal flank of nonobese diabetic CRISPR Prkdc Il2r Gamma mice. Mice bearing shCTL or shYTHDF1 cells were randomly divided into control and x-ray treatment groups. The mice were treated with a dose of 4-Gy irradiation therapy twice a week from day 14. The mice were euthanized on day 31, and the tumor’s weight and volume were measured.

Immunofluorescence

Cells cultured on coverslips on 24-well plates were fixed with 4% polyformaldehyde solution and then permeabilized with 0.1% Triton X-100 at 4°C for 15 min. Cells were blocked with 5% bovine serum albumin in PBS and then exposed to a primary antibody at room temperature for 2 hours. Following the primary antibody incubation, the cells were treated with fluorescence-labeled secondary antibodies for 1 hour. Last, coverslips were mounted using Hoechst (Invitrogen, USA) and imaged with a Nikon A1R Confocal Microscope.

Antibodies used in the experiments are as follows: YTHDF1 (1:500; 17479-1-AP, Proteintech), YTHDF2 (1:300; 24744-1-AP, Proteintech), YTHDF3 (1:100; 25537-1-AP, Proteintech), METTL3 (1:400; ab195352, Abcam), FTO (1:400; 27226-1-AP, Proteintech), YTHDC1 (1:300; 14392-1-AP, Proteintech), YTHDC2 (1:300; 27779-1-AP, Proteintech), γH2AX (1:100; 16-202A, Sigma-Aldrich), FLAG (1:500; F1804, Sigma-Aldrich), and SRSF2 (1:100; ab11826, Abcam).

Immunoblotting

Cells were lysed using RIPA lysis buffer (Beyotime, China). The proteins were separated by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene difluoride (PVDF) membranes (Millipore). The membranes were subjected to blocking and incubation with primary antibodies, followed by a secondary antibody and detection using a chemiluminescent detection system (Amersham Imager).

Antibodies used in the experiments are as follows: YTHDF1 (1:2000; 17479-1-AP, Proteintech), γH2AX (1:2000; 16-202A, Sigma-Aldrich), ATM pS1981 (1:1000; ab36810, Abcam), ATM (1:5000; ab32420, Abcam), hemagglutinin (HA) (1:5000; AE008, ABclonal), ATR (1:1000; 13934, Cell Signaling Technology), ATR pS428 (1:1000; 2853, Cell Signaling Technology), DNA-PKcs (1:1000; ab70250, Abcam), DNA-PKcs pS2056 (1:1000; ab18192, Abcam), SF3B1 (1:3000; 27684-1-AP, Proteintech), SF3B3 (1:3000; 14577-1-AP, Proteintech), SRSF2 (1:1000; ab204916, Abcam), α-tubulin (1:5000; A6830, ABclonal), β-actin (ACTB) (1:4000; 23660-1-AP, Proteintech), lamin B1 (1:1000; 13435, Cell Signaling Technology), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (1:3000; 60004-1-Ig, Proteintech).

In vitro kinase assay

A previously established protocol was used for ATR kinase assay (43, 44). His-tagged YTHDF1-WT and YTHDF1-S182A were overexpressed and purified from Escherichia coli. Subsequently, they were incubated with affinity-purified ATR complex extracted from irradiated HEK293T cells in kinase buffer [20 mM Hepes (pH 7.5), 50 mM NaCl, 50 mM phosphatase inhibitors, 10 mM MgCl2, 1 mM dithiothreitol (DTT), 50 μM adenosine 5′-triphosphate, and proteinase inhibitor] at 30°C for 30 min. The reaction was halted by adding an equal volume of SDS loading buffer. The mixture was then subjected to immunoblotting with a phosphoserine antibody (P5747, Sigma-Aldrich).

HR or NHEJ reporter assay

The analysis of the frequency of HR and NHEJ was conducted following previously established procedures. To evaluate HR repair of DNA-DSBs, the plasmids DR-GFP and I–Sce I (functional endonuclease) were cotransfected into HEK293T cells at a ratio of 3:1. Transient expression of I–Sce I endonuclease generates DNA-DSBs at the integrated GFP gene sequences and successful repair through HR results in GFP expression. (45). NHEJ reporter uses the two broken ends of a GFP gene separated by an adenovirus exon digested with Hind III. The NHEJ reporter plasmid was digested with Hind III for 6 hours, and linearized DNA was then transfected into HEK293T cells (46). For both experiments, mCherry was cotransfected to serve as a transfection control. Three days posttransfection, cells were harvested, resuspended in 0.5 ml of PBS, and analyzed by flow cytometry (Beckman Coulter, CytoFLEX LX).

Neutral comet assay

The neutral comet assay was conducted using the Single Cell Gel Electrophoresis Kit (Trevigen) following the manufacturer’s instruction. Cell samples were mixed with low-melting agarose at a 1:10 ratio, and the mixture was pipetted onto microscope slides. After agarose solidification, slides were immersed in lysis solution for 30 min and then immersed in the neutral electrophoresis solution (pH 8.0) for another 30 min. Electrophoresis was conducted in the electrophoresis solution at a voltage of 1 V/cm for 15 min. Following electrophoresis, the agarose was stained with propidium iodide and visualized using a fluorescence microscope. ImageJ was used to quantify the comet tail moment.

Coimmunoprecipitation

HEK293T cells transfected with vector, pYTHDF1-WT-FLAG, pYTHDF1-S182A-FLAG, or pCRM1-HA as indicated in the figures were subjected to x-ray irradiation. Cells were lysed in lysis buffer [50 mM tris-HCl (pH 7.4), 0.25% deoxycholate, 150 mM NaCl, 1% NP-40, 1 mM EDTA, 1 mM glycerophosphate, protease inhibitor, and phosphatase inhibitor] for 30 min on ice and centrifuged. The supernatant was then subjected to a 4-hour incubation at 4°C with either anti-FLAG or anti–immunoglobulin G (IgG) beads with rotation. The samples were washed three times with NT2 buffer [200 mM NaCl, 50 mM Hepes (pH 7.6), 2 mM EDTA, 0.05% NP-40, and 0.5 mM DTT], denatured with SDS loading buffer, and subjected to immunoblotting with corresponding antibodies.

Immunoprecipitated complex from x-ray–treated cells was used for MS analysis. The proteins bound to the beads were eluted using a FLAG peptide solution (100 μg/ml; PEP-087, Invitrogen) for 2 hours at 4°C. The immunoprecipitated proteins were subjected to MS using Thermo Fisher Scientific Q Exactive HF-X. Data analysis was performed with MaxQuant Software.

RNA sequencing

Total RNAs were isolated using TRIzol Reagent (Invitrogen). mRNAs were enriched from total RNAs using Dynabeads Oligo(dT)₂₅ (Invitrogen). mRNAs were fragmented into approximately 100- to 200-nucleotide fragments using RNA Fragmentation Reagents (AM8740, Ambion). These fragments were then collected for library construction and high-throughput sequencing.

m⁶A sequencing

Dynabeads Protein G (Invitrogen) coated with 3 μg of anti-m⁶A antibody (202 003, Synaptic Systems) were incubated with mRNA fragments at 4°C for 4 hours. Following incubation, the beads were washed three times with immunoprecipitation buffer [10 mM tris-HCl (pH 7.4), 150 mM NaCl, and 0.1% NP-40]. The RNA was then eluted using 200 μl of elution buffer [5 mM tris-HCl (pH 7.4), 0.05% SDS, 1 mM EDTA, proteinase K, and RNase inhibitor] and precipitated with ethanol. The purified RNA was subsequently used for library construction and high-throughput sequencing.

Cross-linking and immunoprecipitation

HEK293T cells stably expressing FLAG-tagged YTHDF1 were treated with 100 μM 4SU (T4509, Sigma-Aldrich) for 16 hours. Cells were exposed to a 365-nm UV light source (Vilber) to induce cross-linking. The cell nuclei were isolated and lysed using NP-40 lysis buffer [50 mM Hepes (pH 7.5), 150 mM KCl, 2 mM EDTA, 0.5% NP-40, 1 mM NaF, 0.5 mM DTT, protease inhibitors, and RNase inhibitor]. The cleared cell lysates were treated with RNase T1 (1 U/μl) at 22°C for 15 min, followed by immunoprecipitation using anti-FLAG M2 magnetic beads (M8823, Sigma-Aldrich). The samples obtained from the immunoprecipitation step were subjected to a second round of treatment with RNase T1. Cross-linked RNAs were dephosphorylated by calf intestinal phosphatase. The complexes were then separated by SDS-PAGE and transferred onto PVDF membrane. The remaining protein on the membrane was digested by 2 mg of proteinase K. RNA was isolated by acidic phenol/chloroform extraction and ethanol precipitation. The recovered RNA was then subjected to small RNA library construction and deep sequencing.

The CLIP-seq reads for the YTHDF1 were preprocessed to remove the adapter sequences using fastp software (version 0.20.0) (47) and then aligned to the human genome (version hg38). Clusters of reads were identified using PARalyzer (v1.1) (48). Each cluster was annotated on the basis of gene information from Ensembl (release 72).

RNA immunoprecipitation

Control or YTHDF1-KO cells were transfected with FLAG-SRSF2 plasmid. Three 15-cm plates of cells were used for each RIP experiments. Cells were lysed in lysis buffer [150 mM NaCl, 10 mM Hepes (pH 7.6), 2 mM EDTA, 0.5% NP-40, 0.5 mM DTT, a 1:100 protease inhibitor cocktail, and RNase inhibitor (400 U/ml)]. Cell lysates were incubated on ice for 30 min, followed by centrifugation. The supernatant was immunoprecipitated with anti-FLAG beads at 4°C for 4 hours. The beads were washed with NT2 buffer and then treated with micrococcal nuclease (2 U/ml). The beads were digested with proteinase K (4 mg/ml; New England Biolabs) at 50°C for 40 min. RNAs recovered from the supernatant were used for library generation and deep sequencing.

Peak calling

The m⁶A peaks were identified using household scripts as described previously (35). For RIP sequencing (RIP-seq), the target binding regions were identified using MACS2 software (version 2.1.1) with default options.

AS analysis

The PSI estimation method (49) was used to analyze AS events based on RNA-seq data. The sorted bam files were used for AS analysis with rMATS (v4.1.2). The output data were processed to identify cassette alternative exons, which are exons that can be either included or excluded from the transcripts. Mann-Whitney test was performed to evaluate the statistical difference of ΔPSI values between groups.

RNA isolation, cDNA synthesis, and quantitative RT-PCR

Total RNA was isolated from HEK293T cells using TRIzol reagent (Invitrogen). cDNA synthesis was carried out with the Evo M-MLV RT-PCR Kit (Accurate Biology) using random primers. Real-time PCR was conducted using the SYBR Green Pro Taq HS quantitative PCR (qPCR) Mix on a LightCycler 480 Real-Time PCR System (Roche). Relative gene expression was normalized to ACTB or GAPDH expression and determined using the 2^(−ΔΔCT) method. RT-qPCR primers are listed in table S9.

Semiquantitative PCR

PCR reactions were conducted in a 25-μl reaction volume using PCR Master Mix (Yeasen). For all the PCR assays detecting AS changes, we used 35 to 40 cycles. All PCR reactions were performed in three independent biological replicates. The primer pairs used for detecting various AS events are listed in table S9.

Motif identification

The m⁶A-enriched regions within each m⁶A-immunoprecipitation sample were identified using MACS2 software (version 2.2.7.1) (50), with the corresponding input sample serving as control. The enriched m⁶A motifs within these clusters were analyzed using MEME (v5.5.4) (51).

Statistical analysis

All statistical analyses were carried out with GraphPad Prism 9. Comparisons between the two groups were performed using a two-tailed unpaired Student’s t test, and the statistical significance was indicated by asterisks (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001). Each experiment was repeated three times for statistical robustness.

Supplementary Materials

The PDF file includes:

Figs. S1 to S5

Legends for tables S1 to S9

Other Supplementary Material for this manuscript includes the following:

Tables S1 to S9