NSUN2-mediated m5C modification drives alternative splicing reprogramming and promotes multidrug resistance in anaplastic thyroid cancer through the NSUN2/SRSF6/UAP1 signaling axis

Abstract

Rationale: Anaplastic thyroid carcinoma (ATC) is an extraordinarily aggressive form of thyroid cancer, frequently presenting with locally advanced infiltration or distant metastases at the time of initial diagnosis, thus missing the optimal window for surgical intervention. Consequently, systemic chemotherapy and targeted therapies are vital for improving the prognosis of ATC. However, ATC exhibits significant resistance to conventional treatments, highlighting the need to elucidate the biological mechanisms underlying this drug resistance and identify novel therapeutic targets to overcome it. Methods: We conducted a comprehensive analysis of both bulk and single-cell RNA sequencing (scRNA-seq) data from ATC samples to screen for m⁵C modification-related genes associated with multidrug resistance (MDR). We then performed IC₅₀ assays, flow cytometry, and employed a spontaneous tumorigenic ATC mouse model with Nsun2 knockout to demonstrate that NSUN2 promotes MDR in ATC. To investigate the mechanisms of NSUN2-mediated drug resistance, we generated NSUN2-knockout ATC cell lines and performed transcriptomic, proteomic, and MeRIP-seq analyses. Additionally, RNA sequencing and alternative splicing analyses were conducted to determine global changes upon NSUN2 knockout. We further explored the underlying mechanisms of the NSUN2/SRSF6/UAP1 axis through glycoprotein staining, denaturing IP ubiquitination, nuclear-cytoplasmic fractionation, and PCR. Lastly, we evaluated the synergistic effects of a small-molecule NSUN2 inhibitor with anticancer agents both in vitro and in vivo. Results: Our findings reveal that NSUN2 expression correlates significantly with MDR in ATC. NSUN2 operates as a "writer" and ALYREF as a "reader" of m⁵C on SRSF6 mRNA, inducing alternative splicing reprogramming and redirecting the splice form of the UAP1 gene from AGX1 to AGX2. As a result, AGX2 enhances the N-linked glycosylation of ABC transporters, stabilizing them by preventing ubiquitination-mediated degradation. Furthermore, an NSUN2 inhibitor reduces NSUN2 enzymatic activity and diminishes downstream target expression, presenting a novel, promising therapeutic approach to overcome MDR in ATC. Conclusions: These findings suggest that the NSUN2/SRSF6/UAP1 signaling axis plays a vital role in MDR of ATC and identify NSUN2 as a synergistic target for chemotherapy and targeted therapy in ATC.

Keywords: 5-methylcytosine; NSUN2; alternative splicing reprogramming; anaplastic thyroid cancer; multidrug resistance.

Figure 1

Bioinformatics analysis reveals a correlation between NSUN2 and MDR in ATC. A) Spearman correlation of gene expression and drug sensitivity in ATC patients from the CTRP database. B-C) Summary of enriched pathways among differentially expressed genes (DEGs) between high- and low-NSUN2-expression groups. D) TSNE plot exhibiting the identified clusters of the scRNA-seq data from ATC (n = 10). E) Scatterplot depicting the log2 fold change (log2FC) for NSUN2-positive versus NSUN2-negative cells and its correlation with predicted IC₅₀ Area Under the Curve (AUC) values for each drug. F) Butterfly diagram illustrating correlations among NSUN2 expression, drug sensitivity, and GO pathway analysis. G) TSNE plot displaying the identified clusters in scRNA-seq data from ATC (n = 10). H) Single-cell drug sensitivity analysis using Beyondcell on GEO dataset samples. I) Correlation between NSUN2 expression and the Beyondcell score.

Figure 2

NSUN2-induced MDR in ATC depends on its methyltransferase activity. A) Western blot analysis of NSUN2 expression in Cal-62 cells with NSUN2 knockout. B) Dot blot assay examining the effect of NSUN2 knockout on m⁵C levels in mRNA transcriptomes of Cal-62 cells. C) Colorimetric m⁵C quantification assay confirming alterations in m⁵C levels following NSUN2 knockout in Cal-62 cells. D-F) Cell growth inhibition assays evaluating the impact of NSUN2 knockout on Cal-62 cell sensitivity to doxorubicin, cisplatin, and lenvatinib. G) Relative protein expression of NSUN2 in Cal-62 cells after treatment with the indicated vectors. H) Dot blot assay measuring m⁵C abundance in mRNA transcriptomes of Cal-62 cells following vector treatment. I) Colorimetric m⁵C quantification assay confirming m⁵C changes in Cal-62 cells treated with the indicated vectors. J-L) Cell growth inhibition assay assessing IC⁵⁰ values in Cal-62 cells transfected with the indicated vectors. M) Schematic depicting the generation of spontaneous ATC mouse models, illustrating both the targeting strategy (left) and breeding scheme (right). N) Mouse genotyping results, with DNA gel band sizes of 130/324 bp for TPO-cre, 185/307 bp for Braf^V600E, 290/370 bp for Trp53^flox/flox, and 151/212 bp for Nsun2^flox/flox. O) Diagram showing the administration of doxorubicin, cisplatin, and lenvatinib in genetically engineered mice. mATC (murine anaplastic thyroid carcinoma). P) Scatter plot displaying the final tumor weights for the indicated groups. Q) Scatter plot presenting the final tumor volumes for the indicated groups. R) Survival curves for mice with ATC for the indicated groups. The data are presented as the mean ± SD. All *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 3

NSUN2-mediated m⁵C modification drives alternative splicing reprogramming by targeting SRSF6 in ATC. A) Volcano plot illustrating changes in protein levels in Cal-62 cells following NSUN2 knockout, as determined by TMT-MS. B) GO enrichment analysis of TMT-MS data from Cal-62 cells after NSUN2 knockout. C) Volcano plots showing the percent splicing inclusion (ΔPSI) of differentially spliced genes in Cal-62 cells after NSUN2 knockout. D) Violin plots highlighting significant ΔPSI changes in Cal-62 cells after NSUN2 knockout. E) Heatmap of TMT-MS data indicating differentially expressed genes among SF3A/B, the U2AF core complex, and the hnRNP family in Cal-62 cells following NSUN2 knockout. F) Statistical analysis of SRSF6 protein levels from TMT-MS data in Cal-62 cells following NSUN2 knockout. G) Western blot analysis of the indicated proteins in Cal-62 cells with NSUN2 knockout. H) Integrative Genomics Viewer (IGV) tracks of MeRIP-seq data surrounding the SRSF6 locus in Cal-62 cells after NSUN2 knockout. I) MeRIP-qPCR measuring m⁵C enrichment in SRSF6 mRNA in Cal-62 cells with NSUN2 knockout. J) MeRIP-qPCR assessing m⁵C enrichment in SRSF6 mRNA in Cal-62 cells treated with indicated vectors. K) Nuclear and cytoplasmic distribution of SRSF6 mRNA in Cal-62 cells upon NSUN2 knockout, evaluated by qRT-PCR. L) Nuclear and cytoplasmic distribution of SRSF6 mRNA in Cal-62 cells treated with the indicated vectors. M) Western blot analysis of the indicated proteins in Cal-62 cells treated with the indicated vectors. N) Canonical m⁵C binding sites within SRSF6. Top, motif sequence at m⁵C binding sites; bottom, identified m⁵C sites in SRSF6. O) MeRIP-qPCR quantifying m⁵C enrichment in SRSF6 mRNA in Cal-62 cells treated with the indicated vectors. P) Nuclear and cytoplasmic distribution of SRSF6 mRNA in Cal-62 cells treated with the indicated vectors. The data are presented as the mean ± SD. All *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 4

NSUN2 promotes MDR in ATC by enhancing the N-glycosylation of ABC transporters. A) Gene Ontology (GO) enrichment analysis of transcriptional data from Cal-62 cells following NSUN2 knockout. B-C) GSEA enrichment analysis of transcriptional data from Cal-62 cells after NSUN2 knockout. D) Glycoprotein staining in Cal-62 cells after NSUN2 knockout is shown on the right. Coomassie blue staining for total protein detection is shown on the left. E-F) Western blotting analysis of indicated protein levels in Cal-62 cells treated with or without tunicamycin (TM), swainsonine (SS). Cell lysate samples were treated with or without PNGase F. “G” means N-glycosylation and “DG” means decreased N-glycosylation. G) Western blot analysis of indicated protein levels in Cal-62 cells with NSUN2 knockout. Cell lysates were treated with or without PNGase F. H) Representative IHC images of NSUN2, ABCG2, and ABCC1 expression in ATC tissues (n = 35). I-J) The chi-square test was used to analyze the correlation between NSUN2 and ABCG2 or ABCC1 expression. K-L) Correlation analysis of IHC scores for NSUN2 and ABCG2 or ABCC1. M-N) The potential N-Glycosylated asparagine sites of ABCG2 and ABCC1 predicted with NetNglyc server. O-P) ABCG2 and ABCC1 expression were detected by western blot in Cal-62 cells transfected with the N557Q, N388Q, N596Q, or N71Q, N19Q, N310Q, N(71+310)Q mutant plasmids, respectively. Q-S) Flow cytometry analysis of intracellular accumulation of doxorubicin, cisplatin, and lenvatinib in Cal-62 cells following NSUN2 knockout. Quantification of mean fluorescence intensity is shown on the right. The data are presented as the mean ± SD. All *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 5

NSUN2 enhances ABC transporters stability through N-glycosylation. A) Cal-62 cells were treated with TM or the indicated concentrations of MG132, followed by western blot analysis using the specified antibodies. B) Cal-62 cells with NSUN2 knockout were treated with indicated concentrations of MG132, followed by western blot analysis with indicated antibodies. C) Western blotting analysis of indicated protein levels in Cal-62 cells treated with TM and MG132. Cell lysate samples were treated with PNGase F. D) Quantification of ABCC1 and ABCG2 protein levels performed using ImageJ; data are presented as mean ± SD. E-F) Cal-62 cells treated with TM and MG132 were subjected to ABCC1 or ABCG2 immunoprecipitation (IP), followed by western blot analysis with anti-ubiquitin antibodies. G) Western blotting analysis of indicated protein levels in Cal-62 cells having NSUN2 knockout and treated with MG132. Cell lysate samples were treated with PNGase F. H) Quantification of ABCC1 and ABCG2 protein intensities using ImageJ; data are presented as mean ± SD. I-J) Cal-62 cells having NSUN2 knockout and treated with MG132 were subjected to ABCC1 or ABCG2 immunoprecipitation (IP), followed by western blot analysis with anti-ubiquitin antibodies. K-L) Wild-type (WT) or the indicated ABCG2/ABCC1 mutants expressed in Cal-62 cells were treated with 125 μg/mL cycloheximide (CHX) for specified intervals and analyzed by western blot. M-N) Protein intensities of ABCG2 or ABCC1 were quantified using Image J and normalized to β-Actin levels. The data are presented as the mean ± SD. All *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 6

NSUN2 promotes N-glycosylation by mediating the alternative splicing of UAP1 through SRSF6. A) UAP1 promotes the biosynthesis of UDP-GlcNAc, a key donor in glycosylation reactions, particularly in N-glycosylation and O-GlcNAc modifications. B) Sashimi plots illustrating changes in UAP1 splicing events in Cal-62 cells following NSUN2 knockout. C) RT-PCR analysis of UAP1 splicing in control versus NSUN2 knockout Cal-62 cells. The number of skipped exons is depicted for each transcript. D) RT-PCR showing the splicing events of UAP1 in spontaneous mouse ATC tumors with or without Nsun2 knockout. E) RT-PCR showing the splicing events of UAP1 in Cal-62 cells treated with indicated vectors (upper panel). The number of skipped exons is depicted for each transcript. The reconstruction of SRSF6 was determined by western blot (lower panel). F) Glycoprotein staining in Cal-62 cells treated with indicated vectors is shown on the right. Coomassie blue staining for total protein detection is shown on the left. G) Western blot analysis of indicated protein levels in Cal-62 cells treated with indicated vectors. H-J) Flow cytometry analysis of intracellular accumulation of doxorubicin, cisplatin, and lenvatinib in Cal-62 cells treated with indicated vectors. Quantification of mean fluorescence intensity is shown on the right. K-M) Cell growth inhibition assay evaluating the effects of the indicated vectors on Cal-62 cell sensitivity to doxorubicin, cisplatin, and lenvatinib. N-P) Tumor growth curves evaluating the impact of the indicated vectors treatment on Cal-62 cells response to doxorubicin, cisplatin, and lenvatinib. Q-S) Tumor final weights evaluating the impact of the indicated vectors treatment on Cal-62 cells response to doxorubicin, cisplatin, and lenvatinib. The data are presented as the mean ± SD. All *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 7

ALYREF functions as an m⁵C “reader” that recognizes and shuttles SRSF6 from nucleus to cytoplasm. A-B) GSEA enrichment analysis indicating that ALYREF expression correlates with chemotherapy resistance. C) Heatmap depicting genes associated with doxorubicin and cisplatin resistance in Cal-62 cells with NSUN2 knockout or ALYREF, YBX1, and SRSF2 knockdown. D) RIP-qPCR assay detecting ALYREF binding to SRSF6 in Cal-62 cells. E) Western blotting analysis of the indicated proteins obtained from SRSF6[m⁵C] pulldown assay. F) RIP-qPCR comparing ALYREF-WT and ALYREF-K171A binding to SRSF6 in Cal-62 cells treated with indicated vectors. G) RIP-qPCR assessing ALYREF binding to SRSF6 mRNA in Cal-62 cells following NSUN2 knockout. H) RIP-qPCR evaluating ALYREF binding to SRSF6 mRNA in Cal-62 cells treated with indicated vectors. I) Nuclear and cytoplasmic distribution of SRSF6 mRNA in Cal-62 cells with ALYREF knockdown, evaluated by qRT-PCR assay. J) Nuclear and cytoplasmic distribution of SRSF6 mRNA in Cal-62 cells overexpressing ALYREF or ALYREF-K171A mutant. K) Nuclear and cytoplasmic distribution of SRSF6 mRNA in Cal-62 cells treated with indicated vectors. L-N) Cell growth inhibition assay evaluating the impact of indicated vectors treatment on Cal-62 cell response to doxorubicin, cisplatin, and lenvatinib. O) Western blotting analysis of the indicated protein levels in Cal-62 cells treated with indicated vectors. P) Immunofluorescence staining of ALYREF in NSUN2-knockout Cal-62 cells (left) and quantitative analysis of nuclear/cytoplasmic ALYREF distribution (right). Q) Western blotting analysis (left) and corresponding quantification (right) of nuclear and cytoplasmic ALYREF in control and NSUN2-knockout Cal-62 cells are shown, with PARP1 and TUBULIN as nuclear and cytoplasmic markers, respectively. The data are presented as the mean ± SD. All *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 8

Small-molecule NSUN2 inhibitor enhances multidrug sensitivity in ATC. A) Dot blot analysis assessing m⁵C levels in mRNA from Cal-62 cells with NSUN2 knockout, with or without NSUN2 inhibitor treatment. B) RT-qPCR-based MeRIP demonstrating m⁵C enrichment in SRSF6 mRNA in Cal-62 cells with NSUN2 knockout, with or without NSUN2 inhibitor. C) Nuclear and cytoplasmic distribution of SRSF6 mRNA in Cal-62 cells with NSUN2 knockout, with or without NSUN2 inhibitor, evaluated by qRT-PCR assay. D) Western blot analysis of indicated protein levels in Cal-62 cells with NSUN2 knockout, with or without NSUN2 inhibitor. E) RT-PCR shows the splicing events of UAP1 in Cal-62 cells and ACT-1 cells treated with or without NSUN2 inhibitor. F) Glycoprotein staining in Cal-62 cells and ACT-1 cells treated with or without NSUN2 inhibitor (right). Coomassie blue staining for total protein detection is shown on the left. G) Western blot analysis of indicated protein levels in Cal-62 cells and ACT-1 cells treated with or without NSUN2 inhibitor. Cell lysates were treated with or without PNGase F. H-J) Flow cytometry analysis of intracellular accumulation of doxorubicin, cisplatin, and lenvatinib in Cal-62 cells treated with or without NSUN2 inhibitor. Quantification of mean fluorescence intensity is shown on the right. K-M) Cell growth inhibition assay assessing the effect of combining NSUN2 inhibitor with doxorubicin, cisplatin, or lenvatinib on Cal-62 cells. N) Schematic illustrating the administration of NSUN2 inhibitor, doxorubicin, cisplatin, and lenvatinib in genetically engineered mice. O) Scatter plot depicting final tumor weights in the specified groups. P) Scatter plot showing final tumor volumes in the specified groups. Q) Survival curves for mice with ATC across the indicated groups. The data are presented as the mean ± SD. All *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 9

Schematic model of the function and mechanism of NSUN2 in ATC multidrug resistance.