Nuclear Aurora kinase A switches m6A reader YTHDC1 to enhance an oncogenic RNA splicing of tumor suppressor RBM4

Abstract

Aberrant RNA splicing produces alternative isoforms of genes to facilitate tumor progression, yet how this process is regulated by oncogenic signal remains largely unknown. Here, we unveil that non-canonical activation of nuclear AURKA promotes an oncogenic RNA splicing of tumor suppressor RBM4 directed by m⁶A reader YTHDC1 in lung cancer. Nuclear translocation of AURKA is a prerequisite for RNA aberrant splicing, specifically triggering RBM4 splicing from the full isoform (RBM4-FL) to the short isoform (RBM4-S) in a kinase-independent manner. RBM4-S functions as a tumor promoter by abolishing RBM4-FL-mediated inhibition of the activity of the SRSF1-mTORC1 signaling pathway. Mechanistically, AURKA disrupts the binding of SRSF3 to YTHDC1, resulting in the inhibition of RBM4-FL production induced by the m⁶A-YTHDC1-SRSF3 complex. In turn, AURKA recruits hnRNP K to YTHDC1, leading to an m⁶A-YTHDC1-hnRNP K-dependent exon skipping to produce RBM4-S. Importantly, the small molecules that block AURKA nuclear translocation, reverse the oncogenic splicing of RBM4 and significantly suppress lung tumor progression. Together, our study unveils a previously unappreciated role of nuclear AURKA in m⁶A reader YTHDC1-dependent oncogenic RNA splicing switch, providing a novel therapeutic route to target nuclear oncogenic events.

Fig. 1

A non-canonical function of nuclear AURKA in driving RBM4 aberrant RNA splicing. a Representative immunohistochemistry (IHC) staining showing AURKA expression in non-small cell lung cancer (NSCLC) specimens and adjacent normal tissues. Right images show ×20 and ×40 magnification of region indicated by the black box in the corresponding image. Scale bar, 50 μm. b UniProt Keywords analysis of AURKA-regulated splicing events of SE type. −log₁₀ P-values were plotted for each enriched functional classification. c Comparison of AURKA-regulated splicing events of SE type with AURKA RNA-immunoprecipitation sequencing (RIP-seq) data and one RNA splicing-related gene set. Two candidates, SRRM1 and RBM4. d Schematics of human RBM4 pre-mRNA and protein. The full-length RBM4 (RBM4-FL) contains 1–4 exons, whereas exon 3 of RBM4 can be skipped to generate a new short isoform RBM4-S. RBM4-S only contains an N-terminal RNA-binding domain. e Validation of RBM4 splicing change by semi-quantitative RT-PCR in DOX-induced AURKA knockdown NCI-H460 cells. PSI (Percent Spliced In) = RBM4-FL/RBM4-FL + RBM4-S. DOX, 2 μg/ml. f RT-qPCR assay was conducted to detect RBM4-FL and RBM4-S mRNA expression levels in DOX-induced AURKA knockdown NCI-H460 cells. g RBM4 splicing reporter was overexpressed in control and AURKA knockdown NCI-H460 cells. Western blot was conducted to assay for the splicing change of RBM4. h Empty vector (VEC), AURKA-WT (wild-type), AURKA-NLS (nuclear-localized sequence), and AURKA-NES (nuclear export sequence) were transfected into endogenous AURKA depleted NCI-H460 cells. The localization of AURKA (green) was detected by immunofluorescence (IF) assay with anti-AURKA antibody. The nuclei were stained with DAPI (blue). Scale bar, 10 μm. i RBM4 splicing change in VEC, AURKA-WT, AURKA-NLS, and AURKA-NES reconstitution NCI-H460-shAURKA cells was measured by RT-PCR. j Binding of RBM4 pre-mRNA with AURKA protein was detected by RIP assay in NCI-H460 cells. k Colocalization of RBM4 pre-mRNA (red) with AURKA protein (green) was detected by RNA fluorescence in situ hybridization (FISH) assay in NCI-H460 cells. The nucleus was stained with DAPI (blue). Scale bar, 10 μm. Data are shown as means ± SD. P-values were calculated with two-tailed unpaired Student’s t-test and P < 0.05 is considered statistically significant

Fig. 2

RBM4-S functions as a tumor promoter by abolishing RBM4-FL-mediated inhibition of SRSF1-mTORC1 activity. a Flag-labeled RBM4-FL and RBM4-S were transfected into A549 cells, respectively. The localization of Flag-RBM4-FL and Flag-RBM4-S was detected by IF assay with anti-Flag antibody (green). The nuclei were stained with DAPI (blue). Scale bar, 10 μm. The effects of RBM4-FL and RBM4-S on the proliferation of A549 cells. Cells were stably transfected with Vector, RBM4-FL, RBM4-S plasmids, and the proliferative ability of cells was analyzed by b cell counting kit-8 (CCK8) and c colony formation assays. d, e Immunodeficient mice were subcutaneously inoculated with equal number of Vector, RBM4-FL, and RBM4-S stably expressing A549 cells (2 × 10⁶ cells per mouse, n = 10). Photograph and tumor volume were shown. f Validation of the expression of AURKA and two variants of RBM4 by RT-PCR. g, h Colony formation and CCK8 proliferative assays were conducted to examine cell proliferation ability. i Validation of the expression of AURKA and exogenous Flag-RBM4-S by western blot. j, k Colony formation and CCK8 proliferative assays were performed to examine cell proliferation ability. l SRSF1 was transiently expressed in control and Flag-RBM4-FL overexpressed A549 cells, and SRSF1 was transiently depleted in control and Flag-RBM4-S overexpressed A549 cells. The protein expression levels of Flag-RBM4-FL or Flag-RBM4-S, SRSF1, P-4E-BP1, 4E-BP1, P-S6K, and S6K were assessed by western blot assay and normalized to GAPDH. m Increasing amounts of Flag-RBM4-S were co-expressed with Flag-RBM4-FL in HEK-293T cells. Flag-RBM4-FL/S, SRSF1, P-4E-BP1, 4E-BP1, P-S6K, and S6K protein expression levels were assessed by western blot and normalized to GAPDH. Data are shown as means ± SD. P-values were calculated with two-tailed unpaired Student’s t-test and P < 0.05 is considered statistically significant

Fig. 3

Nuclear AURKA inhibits m⁶A-YTHDC1-SRSF3-mediated RBM4 exon inclusion. a RBM4 gene was truncated into four parts, each of which was applied to an in vitro transcription assay with Biotin-labeled as indicated. Sense 1: −412/+463, Sense 2: −510/+403, Sense 3: −280/+499, Sense 4: −496/+403. b The streptavidin beads-immobilized biotin-labeled RNA truncations were incubated with A549 cell lysates and an RNA pull-down assay was performed. c Four potential m⁶A sites located in the Sense 2 (−510/+403) region of RBM4 pre-mRNA. d RNA methylation modification was detected by the methylated RNA-immunoprecipitation (MeRIP)-qPCR assay in A549 cells. e The PCR products of RBM4 truncated sense and anti-sense regions were applied to an in vitro transcription assay with Biotin-labeled. Binding of these RNA truncations with YTHDC1/GAPDH proteins was examined by an RNA pull-down assay in A549 cells. f RBM4 splicing change in YTHDC1-WT (wild-type) and YTHDC1-DM (m⁶A binding region mutant) reconstitution endogenous YTHDC1 depleted A549 cells was detected by RT-PCR. g Binding of RBM4 pre-mRNA with IgG or exogenous Flag-VEC/Flag-YTHDC1-WT/Flag-YTHDC1-DM protein was detected by RIP-qPCR assay in A549 cells. h The level of YTHDC1 or i METTL3 was depleted in control and SRSF3 overexpressed A549 cells. Relative mRNA abundance of SRSF3, YTHDC1/METTL3, and RBM4 splicing was examined by RT-PCR. j The level of METTL3 or k YTHDC1 or l SRSF3 was depleted in A549-Tet-On-shAURKA cells with or without DOX induction. Relative mRNA abundance of AURKA, METTL3/YTHDC1/SRSF3, and RBM4 splicing was examined by RT-PCR. Data are shown as means ± SD. P-values were calculated with two-tailed unpaired Student’s t-test and P < 0.05 is considered statistically significant

Fig. 4

Nuclear AURKA interacts with m⁶A reader YTHDC1 and disrupts the binding of SRSF3 to YTHDC1. a The interaction between endogenous AURKA and YTHDC1 in both the absence and presence of RNase was analyzed by a co-immunoprecipitation (Co-IP) assay. b The interaction between exogenous HA-AURKA and Flag-YTHDC1 was measured by Co-IP assay in HEK-293T cells. c Nuclear/cytoplasmic protein fractions of A549 cells were subjected to immunoprecipitation (IP) and immunoblotting (IB) using antibodies as indicated. d GST-fused AURKA full-length (1–403 aa) and truncations (333–403 aa, 1–333 aa, 1–283 aa, 1–233 aa, 1–183 aa, 1–131 aa) were co-expressed with Flag-YTHDC1 in HEK-293T cells. Protein interaction was analyzed by the GST pull-down assay. e The protein interaction between AURKA and YTHDC1 was analyzed by a Co-IP assay in A549 cells treated with DMSO and three kinds of AURKA kinase inhibitors (AKI603/MLN8237/VX680). AKI603, 1 μM; MLN8237, 0.2 μM; VX680, 0.5 μM. f GST-fused AURKA full-length (1–403 aa) was co-expressed with Flag-SRSF3 in HEK-293T cells. Protein interaction was analyzed by the GST pull-down assay. g Co-IP analysis of the protein interaction between YTHDC1 and SRSF3 in NCI-H460 cells transfected with increasing concentrations of plasmid encoding AURKA after MG132 treatment. MG132, 20 mM. h Co-IP analysis of the protein interaction between SRSF3 and YTHDC1 in NCI-H460 cells transfected with different siRNAs to knockdown AURKA after MG132 treatment. MG132, 20 mM

Fig. 5

Nuclear AURKA bridges the interaction between hnRNP K and YTHDC1 and promotes m⁶A-YTHDC1-hnRNP K-dependent RBM4 exon skipping. a Combination of AURKA-interacting proteins from SILAC and Biogrid database were compared with core spliceosomal proteins. b The level of hnRNP K, hnRNP A1 and hnRNP U was depleted respectively in A549-Tet-On-shAURKA cells with or without DOX induction. Relative mRNA abundance of AURKA, hnRNPs, and RBM4 splicing was examined by RT-PCR. c Binding of RBM4 pre-mRNA with IgG or hnRNP K protein was detected by RIP-qPCR assay in A549 cells. d Two potential hnRNP K binding sites located in the Sense 2 (−510/+403) region of RBM4 pre-mRNA. RBM4 splicing reporters with the indicated mutations (mut1 and mut2) were generated. e RBM4 splicing reporters containing wild-type (WT) and mutations (mut1 or mut2) were overexpressed in A549 cells with siNC or sihnRNP K. RT-PCR was conducted to detect the splicing change of RBM4. f The PCR products of negative control GAPDH (NC) and hnRNP K binding sites (WT/mut1/mut2) were applied to an in vitro transcription assay with Biotin-labeled. Binding of these RNAs with hnRNP K/GAPDH proteins was detected by an RNA pull-down assay in A549 cells. g Co-IP analysis of the protein interaction between YTHDC1 and hnRNP K in NCI-H460 cells transfected with increasing concentrations of plasmid encoding AURKA after MG132 treatment. MG132, 20 mM. h Co-IP analysis of the protein interaction between hnRNP K and YTHDC1 in NCI-H460 cells transfected with different siRNAs to knockdown AURKA after MG132 treatment. MG132, 20 mM. Data are shown as means ± SD. P-values were calculated with two-tailed unpaired Student’s t-test and P < 0.05 is considered statistically significant

Fig. 6

AURKA nuclear translocation inhibitors reverse RBM4 splicing towards RBM4-S to restrain tumor growth. a Establishment of small-molecule chemical libraries and drug screening model. The localization of enhanced green fluorescent protein (EGFP)-fused AURKA was analyzed via inverted fluorescence microscope. Scale bar, 10 μm. PHA, 10 μM; JNJ, 10 μM. b The localization of endogenous AURKA protein (green) was detected by IF assay in A549 and NCI-H460 cells treated with DMSO, JNJ, and PHA, respectively. The nuclei were stained with DAPI (blue). Scale bar, 10 μm. c The splicing change of RBM4 was validated by RT-PCR in A549 and NCI-H460 cells treated with DMSO, JNJ and PHA. d Empty vector, AURKA-NLS, and AURKA-NES were overexpressed in NCI-H460 cells treated with DMSO, JNJ, and PHA. RT-PCR was conducted to assay for the splicing change of RBM4. e–g Immunodeficient mice were subcutaneously inoculated with equal number of NCI-H460 cells (6 × 10⁶ cells per mouse). When tumor size reached 100 mm³, animals were randomized into four groups (n = 5/group) and treated as follows: PBS, VX680 (20 mg/kg/d), PHA (40 mg/kg/d) combined with VX680 (20 mg/kg/d), JNJ (20 mg/kg/d) combined with VX680 (20 mg/kg/d). The agents were administered by gavage and the treatment continued for 10 days. Photograph of tumors was shown. Tumor weight and volume were shown. Data are shown as means ± SD. P-values were calculated with two-tailed unpaired Student’s t-test and P < 0.05 is considered statistically significant

Fig. 7

Clinical relevance of AURKA-RBM4 aberrant splicing axis. a Relative protein expression levels of AURKA and RBM4-FL/S in eight pairs of NSCLC specimens (T) and adjacent normal tissues (N) were subjected to western blot analysis. b Linear regression analysis of a correlation between AURKA relative expression level and the PSI value of RBM4 in eight pairs of clinical specimens was shown. R² = 0.2960, P = 0.0360. c Representative images of nuclear AURKA, RBM4, and SRSF1 IHC staining in lung tumor specimens. Scale bars, 100 μm or 50 μm. d Schematic diagram of the mechanism: Nuclear AURKA disrupts the binding of SRSF3 to m⁶A reader YTHDC1, thereby inhibiting the production of m⁶A-YTHDC1-SRSF3-mediated tumor suppressive isoform RBM4-FL. In turn, nuclear AURKA bridges the interaction between hnRNP K and YTHDC1, thus facilitating the production of m⁶A-YTHDC1-hnRNP K-mediated tumor-promoting isoform RBM4-S. Blocking AURKA nuclear translocation, can reverse the oncogenic splicing of RBM4 and suppress lung tumor progression. Pearson’s correlation test in c. P < 0.05 is considered statistically significant