Genome-wide investigation of differentially expressed alternative splicing and RNA-binding protein genes association with AML drug resistance

Abstract

Background

Acute myeloid leukemia (AML) is a hematologic malignancy with persistently poor outcomes. Despite advances in chemotherapy, targeted therapy, and hematopoietic stem cell transplantation, drug resistance and relapse remain major obstacles, with the global 5-year survival rate for adults with AML below 50%. The molecular mechanisms underlying relapse and drug resistance are not fully understood.

Methods

Bone marrow mononuclear cells were isolated from four patients with AML relapse (reAML), three patients with AML remission, and five healthy donors. RNA sequencing was performed, followed by differentially expressed genes (DEGs) analysis, overlapping analysis, weighted gene co-expression network analysis (WGCNA) and co-expression analysis, alternative splicing (AS) analysis and functional enrichment analysis.

Results

We identified 593 upregulated and 999 downregulated genes specifically in reAML group compared with both AML remission and healthy groups. Notably, 33 RNA-binding protein (RBP) genes were consistently up-regulated and 30 down-regulated in reAML group. WGCNA and AS analyses revealed marked splicing alterations in reAML group, accompanied by multiple differential regulatory alternative splicing genes (RASGs) and events (RASEs).

Conclusions

This study provides a systematic landscape of RBP dysregulation and aberrant alternative splicing in AML relapse and drug resistance. These findings highlight RBPs and alternative splicing events (ASEs) as potential clinical biomarkers and therapeutic targets. Our work opens new avenues for predicting relapse risk and overcoming drug resistance in AML patients.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-026-05455-6.

Background

Acute myeloid leukaemia (AML) is a malignant disorder characterised by abnormal growth and differentiation of haematopoietic stem cells (HSCs), in which immature myeloid precursors (myeloblasts) accumulate in the bone marrow and peripheral blood, interfering with normal haematopoietic function [1]. Over the past decades, much effort has been devoted to improving the understanding of AML pathophysiology, and many promising new therapies targeting diverse cellular mechanisms have either been approved or are progressing through clinical trials [2–4]. However, in the last decades, the incidence of AML has gradually increased globally, and the five-year survival rate of patients with AML remains poor, highlighting the urgent need for novel and synergistic therapies [5, 6].

The pathophysiological process involved in AML is extremely intricate, and includes an anomaly in the cell proliferation, differentiation, and apoptosis pathways. Currently, the pathogenesis of AML has been deeply investigated, and the corresponding standard therapy combined with a targeted strategy has already been widely adopted. For instance, the FLT-3 inhibitor midostaurin is commonly used in combination therapy for AML patients with FLT-3 mutations [7]. However, the leading cause of AML-related deaths remains treatment failure due to refractory or recurrent AML, which is caused by drug resistance [8]. Studies have shown that the emergence of numerous drug resistance mechanisms is highly related to the alteration of drug transport and metabolism, such as impaired drug uptake, enhanced drug degradation or efflux, loss of prodrug metabolism activation, and changes in target enzymes [9]. Therefore, renewing appropriate pharmaceuticals targeting AML drug resistance-associated genes is a current focus in various treatment strategies, including frontline, relapsed/refractory and minimal residual disease [8], which would play a crucial role in improving the survival rate and in prolonging overall survival of AML patients.

The regulation mechanism of alternative splicing (AS) has a profound influence on gene expression in higher eukaryotes. Most mammalian pre-mRNAs are alternately spliced in a manner that alters the open reading frame, which not only enables the regulation of gene expression levels but also ensures a high degree of gene-product diversity [10]. In humans, AS occurs in almost all genes containing multiple introns, wherein RNA polymerase II plays a vital role in pre-mRNA splicing [11]. It is noteworthy that approximately 15% of human genetic diseases and cancers are strongly associated with abnormal alternative splicing [12]. In particular, recurrent mutations in spliceosome machinery and genome-wide aberrant splicing events have been recognised as prominent components of AML. Research revealed that aberrant splicing can lead to expression alterations of tumour suppressors and oncogenes and to dysregulation of the apoptosis signalling network, thus influencing the drug resistance of AML cells [13]. Therefore, a better understanding of AS characteristics in AML drug resistance and progression is necessary to design appropriate therapeutic strategies recognising specific alternatively spliced or mutated oncogenic targets.

RNA-binding proteins (RBPs) are proteins that can regulate gene expression and function by binding to RNA with highly dynamic spatio-temporal regulatory features [14]. In the cellular dynamics of RNA transcription and post-transcriptional AS, RNA modification, transport, translation, and degradation, there are a variety of specific RBPs that bind to RNA to exhibit its substantial regulatory function. Abnormal expressions or depletions of RBPs play a key role in the drug resistance and progression of AML, and are, therefore, potential therapeutic targets [15]. For example, Sun et al. demonstrated that circMYBL2, a non-coding RNA from mybl2, can promote the binding of RBP PTBP1 to FLT3-ITD mRNA, therefore upregulating the expression level of FLT3-ITD proteins and further enhancing the resistance of AML cells to quizartinib both in vitro and in vivo [16]. Interestingly, it was also reported that depletion in AML cells of PIWIL4, a germ stem cell-associated RBP belonging to the RNase H-like superfamily, can lead to uncontrolled R-loop formation, transcriptional stagnation, DNA damage, greater sensitivity to pharmacological inhibition of the anti-thrombin receptor (ATR) pathway, and further pharmacologically actionable dependency in AML cells [17, 18]. However, the overall regulatory mechanism by which RBPs form drug resistance in AML remains unclear.

In this study, we hypothesised that the differential expression of RBPs is involved in post-transcriptional regulation in relapsed AML patients undergoing medication, thereby causing the occurrence of AS in mRNA precursors, and speculated that the regulatory axis of RBP-AS may play a vital molecular regulatory role in the formation of drug resistance in AML. Herein, we selected bone marrow blood from patients with AML relapse (reAML), patients with AML remission (AML), and Healthy donors as research objects. Subsequently, by using modern RNA sequencing (RNA-seq) and bioinformatics analysis methodologies, the differentially expressed RBPs and AS genes were obtained, and further mechanistic analyses were performed by applying gene ontology (GO), the Kyoto Encyclopedia of Genes and Genomes (KEGG), protein-protein interaction (PPI), and weighted gene co-expression network analysis (WGCNA). Ultimately, we aimed to explore the probable underlying mechanisms of aberrant RBPs and AS regulators in guiding the formation of drug resistance in AML and expected to lay a solid foundation for further exploration of novel drugs targeting AML resistance.

Materials and methods

Patient samples

Bone marrow blood samples from 4 patients with AML relapse, 3 patients with AML remission and 5 control donors were obtained at the Affiliated Hospital of Shandong University of Traditional Chinese Medicine, China. Control samples were obtained from healthy donors of hematopoietic stem cells. Informed consent was obtained in accordance with the Declaration of Helsinki. Bone marrow blood samples were acquired with the informed consent of the donors. The use of human tissues was approved by the Medical Institutional Ethics Committee of the Affiliated Hospital of Shandong University of Traditional Chinese Medicine, China. Approval no., (2023) Review No.(143)-KY. Clinical trial number: not applicable. In addition, all methods in this study were performed in accordance with the Declaration of Helsinki. Bone marrow mononuclear cells were isolated by density gradient centrifugation using Lymphocyte separation medium (Solarbio, China).

RNA extraction and sequencing

The RNA extraction and sequencing were performed in collaboration with Wuhan Ruixing Biotechnology Co., Ltd, China. Total RNA was treated with RQ1 DNase (Promega, USA) to eliminate genomic DNA contamination. RNA quality and concentration were assessed by measuring the A260/A280 ratio using a SmartSpec Plus spectrophotometer (BioRad, USA), and RNA integrity was further confirmed via 1.5% agarose gel electrophoresis.

For RNA-seq library preparation, 1 µg of total RNA per sample was used. mRNA was enriched using VAHTS mRNA Capture Beads (Vazyme, N401, USA). The purified mRNA was subsequently treated with RQ1 DNase (Promega, USA) to remove any residual DNA. Directional RNA-seq libraries were then constructed using the VAHTS Universal V8 RNA-seq Library Prep Kit for Illumina (Vazyme, NR605, USA). In brief, polyadenylated mRNAs were purified and fragmented, and the resulting fragments were reverse-transcribed into double-stranded cDNA. After end repair and A-tailing, the cDNA was ligated with the Adaptor (Vazyme, N323, USA). The ligated products were purified and size-selected to 300–500 bp, followed by amplification, purification, and quantification. The final libraries were stored at − 80 °C until sequencing. Strand-specificity was achieved by not amplifying the dUTP-labeled second cDNA strand.

Library preparation was carried out according to the manufacturer’s instructions, and sequencing was performed on the Illumina Novaseq 6000 platform using a 150 nt paired-end protocol.

RNA-seq raw data cleaning and alignment

Raw reads containing more than 2-N bases were first discarded. Then adaptors and low-quality bases were trimmed from raw sequencing reads using a FASTX-Toolkit (Version 0.0.13). The short reads less than 16 nt were also dropped. After that, clean reads were aligned to the GRCh38 genome by HISAT2 with an allowance of 4 mismatches [19]. Uniquely mapped reads were used for gene reads number counting and fragments per kilobase of transcript per million fragments mapped (FPKM ) calculation [20].

Differentially expressed genes (DEGs) analysis

Differentially expressed genes (DEGs) were identified using the R Bioconductor package DESeq2 [21]. Genes with an adjusted P value < 0.05 and a fold change > 2 or < 0.5 were considered significantly differentially expressed.

Immune-related genes

1,793 immune-related genes (https://www.immport.org/shared/genelists) were retrieved from the ImmPort database.

WGCNA and co-expression analysis

To comprehensively investigate the gene expression landscape, weighted gene co-expression network analysis (WGCNA) was performed to cluster genes exhibiting similar expression patterns using default parameters [22]. All expressed genes were included as input. For each identified module, the eigengene was computed to represent the overall expression pattern of the genes within that module. To examine the regulatory relationship between lncRNAs and their host mRNAs, Pearson’s correlation coefficients (PCCs) were calculated. Based on the PCC values, the relationships were classified into three categories: positively correlated, negatively correlated, or non‑correlated.

Alternative splicing analysis

Alternative splicing events (ASEs) and regulatory alternative splicing events (RASEs) between samples were identified and quantified using the ABLas pipeline, following previously described methods [23, 24]. According to the ABLas detection, the types of ASEs were classified based on splice junction reads, including exon skipping (ES), alternative 5’ splice site (A5SS), alternative 3’ splice site (A3SS), mutually exclusive exons (MXE), mutually exclusive 5’ UTRs (5pMXE), mutually exclusive 3’ UTRs (3pMXE), cassette exons, as well as combined events such as A3SS&ES and A5SS&ES.

Functional enrichment analysis

To determine the functional categories of differentially expressed genes (DEGs), Gene Ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were annotated using the KOBAS 2.0 server [25–28]. A hypergeometric test and the Benjamini–Hochberg false discovery rate (FDR) controlling procedure were applied to assess the significance of enrichment for each term.

Statistical analysis

Statistical differences between two groups (i.e., reAML vs. Healthy & reAML vs. AML) were analysed using a student’s t-test with SPSS Statistics software v.28.0. Data were expressed as means ± standard error of mean (SEM). *p < 0.05 was considered as statistically significant.

Results

RNA-seq analysis of differential gene expression in AML drug-resistant patients

To assess the profile of DEGs in AML relapse (reAML) patients, we performed RNA-seq analysis of bone marrow blood samples from reAML patients, AML remission patients (AML), and Healthy individuals. The results showed that the gene expression profile of reAML patients was significantly altered. Specifically, reAML patients had a total of 2,222 up-regulated genes and 2,413 down-regulated genes compared with Healthy controls. A total of 656 genes were up-regulated, and 1,115 genes were down-regulated compared with AML samples (Fig. 1A and B). Subsequent overlapping analysis revealed a total of 593 overlapped up-regulated genes and 999 overlapped down-regulated genes in the reAML group in comparison to the Healthy and AML groups (Fig. 1C).

Fig. 1.

RNA-seq analysis of differential gene expression in AML drug-resistant patients. A Correlation heatmap showing the relativity among all samples. B Volcano plot right showing all DEGs between reAML patients and Healthy samples. Volcano plot left showing all DEGs between reAML patients and AML remission samples. C Venn diagram showing the overlapping result of up-regulated DEGs and down-regulated DEGs in two comparison groups (reAML vs. Healthy & reAML vs. AML). D Bar plot exhibiting the most enriched GO biological process results of the co-upregulated DEGs in two comparison groups (reAML vs. Healthy & reAML vs. AML). E Bar plot exhibiting the most enriched GO biological process results of the co-downregulated DEGs in two comparison groups (reAML vs. Healthy & reAML vs. AML). F Venn diagram showing the overlapping results of immune-associated genes and down-regulated DEGs in two comparison groups (reAML vs. Healthy & reAML vs. AML)

Further analyses showed that the overlapped up-regulated GO biological processes were mainly enriched in signal transduction, angiogenesis, cell differentiation, and negative regulation of osteoblast differentiation, while the overlapped down-regulated processes were mainly enriched in immunoglobulin production, immune response, the cell surface or B cell receptor signalling pathway, inflammatory response, and T cell activation (Fig. 1D and E). In addition, the overlapped up-regulated KEGG pathways were largely cancer-associated pathways, such as the Rap1 signalling pathway, transcriptional misregulation in cancer, and the PI3K-Akt and MAPK signalling pathways, while the overlapped down-regulated pathways were largely immune-related and inflammatory pathways, such as hematopoietic cell lineage, primary immunodeficiency, the T cell receptor signalling pathway, graft-versus-host disease, Th17 cell differentiation, and natural killer (NK) cell-mediated cytotoxicity (Supplementary Fig. 1A and B). The GO and KEGG results both indicated that the overlapped down-regulated genes were mainly focused on the regulation of immunity and inflammation. Subsequently, we used the known immune genes in the “Immport” immune gene bank as a control and further screened 237 total overlapped down-regulated immune-related genes (Fig. 1F).

Of particular interest, four overlapped up-regulated genes (EGFL7, CBX2, SOX4, and CDK6) and four overlapped down-regulated genes (LCN2, CYBB, RNAS, and CD48) were found to be significantly altered in their expression levels in the reAML group (p < 0.001) (Supplementary Fig. 1C). Similarly, the clinical data on acute myeloid leukaemia (LAML) from the TCGA database showed that the expression levels of IGF2BP2, CDK6, and SOX4 were significantly up-regulated in patients with AML relapse (p < 0.05) (Supplementary Fig. 2). This suggested that the expression profile of these DEGs is considerably changed in the disease state of AML relapse, which may be one of the prime reasons for progression and drug resistance in AML relapse patients.

RNA-seq analysis of differential RBP expression in AML drug-resistant patients

In order to clarify the differential expression of RBP genes in reAML patients, we conducted overlapping analysis of overlapped up-regulated and down-regulated genes, as well as RBP genes. The results demonstrated that 33 RBP genes were overlapped up-regulated and 30 RBP genes were overlapped down-regulated (Fig. 2A). In-depth protein-protein interaction (PPI) analysis confirmed a close relationship between the overlapped RBP genes (Fig. 2B). In particular, the expression levels of two overlapped up-regulated (NPM1 & IGF2BP2) and two overlapped down-regulated (MYH9 & LGALS3) RBP genes were significantly varied (Fig. 2C). The results indicated that these differentially expressed RBPs are closely related to the development of AML and are very likely involved in the regulation of AML relapse, which may also be one of the key regulatory mechanisms of the progression of AML relapse and drug resistance.

Fig. 2.

RNA-seq analysis of differential RBP expression in AML drug-resistant patients. A Venn diagram showing the overlapping results among the overlapped up-regulated DEGs and down-regulated DEGs in two comparison groups (reAML vs. Healthy & reAML vs. AML) and RBPs. B Protein-protein interaction (PPI) plot showing the relationships among the overlapping genes from (A). C Box plot showing the expression levels of NPM1, IGF2BP2, MYH9, and LGALS3 in the indicated groups. FPKM, fragments per kilobase per million mapped reads. Error bars represent means ± SEM. ***p < 0.001; **p < 0.01; *p < 0.05

WGCNA of differential AS in AML drug-resistant patients

To explore the variations of AS in reAML patients, we then carried out in-depth analyses of bone marrow blood samples from the reAML, AML, and Healthy groups. The results demonstrated significant changes in AS in the reAML group in comparison to the AML and Healthy groups, and multiple differential regulatory alternative splicing events (RASEs) were also identified (Fig. 3A and B). We further conducted WGCNA of all the detected regulatory alternative splicing genes (RASGs) and found that these genes were divided into 9 modules (Fig. 3C). According to the different gene expressions in the modules, the turquoise and yellow modules were selected for specific gene expression analyses. The results showed that the expression level of RASG in the turquoise module was increased in the AML and Healthy groups, and decreased in the reAML group. However, in the yellow module, the expression level of RASG was down-regulated in the AML and Healthy groups, and up-regulated in the reAML group (Fig. 3C). The above results implied that both RASGs and RASEs are altered in reAML patients, showing that they are likely another important factor in the progression of AML relapse and resistance.

Fig. 3.

WGCNA of differential AS in AML drug-resistant patients. A PCA base on AS ratio of detected AS events in all samples. The ellipse for each group is the confidence ellipse. B Bar plot presents the number of AS events in different groups. C Hierarchical clustering heatmap from WGCNA showing AS ratio levels of all detected AS event. D Hierarchical clustering heat map showing the AS pattern of all samples. E Bar plot showing the WCGNA expression pattern in turquoise (above) and yellow modules (below).

Clinical data analysis of differential RAS gene expression in AML drug-resistant patients

In order to identify the biological processes that closely related to AML in the turquoise and yellow modules, we performed GO analysis on RASGs for both modules. The results clarified that the GO biological processes in the turquoise module were largely enriched in RNA splicing, mRNA processing, splicesomal complex assembly, positive regulation of DNA template transcription, DNA repair, phagocytosis, and protein stabilisation. However, the GO biological processes in the yellow module were mainly enriched in positive regulation of DNA-templated transcription and both the positive and negative regulation of transcription by RNA polymerase II (Fig. 4A). To further identify the specific RASGs that were differentially expressed, we conducted an in-depth analysis of the relative expression levels of RASGs in both the turquoise and yellow modules. The results confirmed that three genes (GRK6, FLNA, and NCOR2) in the turquoise module and three genes (INF2, MGAT1, and NT5C2) in the yellow module in the reAML group were significantly altered compared with the AML and Healthy groups (Fig. 4B). Therefore, it is speculated that the differentially expressed RASGs may play a vital role in the regulation of key RASEs in the progression of AML relapse and drug resistance by altering numerous intricate biological processes, such as RNA splicing, transcriptional regulation, and protein stabilisation.

Fig. 4.

Clinical data analysis of differential RAS gene expression in AML drug-resistant patients. A Bar plot exhibiting the most enriched GO biological process results of the turquoise module (left) and yellow module (right). B Box plots showing the ratio level of 3 genes in turquoise module (GRK6, FLNA, and NCOR2 above) and 3 genes in yellow module (MGAT1, NT5C2, and INF2 below). Error bars represent means ± SEM. ***p < 0.001; **p < 0.01; *p < 0.05

Discussion

After pharmacotherapy, patients with AML may develop genetic and epigenetic changes that further lead to the enhancement of drug resistance, increasing the difficulties of subsequent therapies [29, 30]. For example, cytarabine (Ara-C) acts in vivo through activation by deoxycytidine kinase phosphorylation to cytarabine triphosphate. However, a majority of relapsed or drug-resistant AML patients have mutations in the deoxycytidine kinase gene, resulting in increased resistance to pharmacotherapy and eventual treatment failure [31]. Recently, studies have shown that mRNA methylation under the regulation of methyltransferase 3 (METTL3) may be involved in the regulation of drug resistance in AML [32, 33]. Here, we report the significant alteration in global gene expression profiles in mononuclear cells from bone marrow blood in relapsed or drug-resistant AML patients compared with AML remission and Healthy controls.

Of note, we found that the significantly overlapped up-regulated genes (i.e. EGFL7, CBX2, SOX4, and CDK6) in relapsed or drug-resistant AML patients are closely associated with biological processes related to AML progression and resistance, such as aberrant cell differentiation and transcription dysregulation. For example, early studies have demonstrated that the specific expression of SOX4 induces the initiation and development of leukaemia in zebrafish, mice, and humans [34, 35]. Recently, He et al. have reported that SOX4 regulates the expression of ARHGAP9, a drug-resistance-related gene, by directly binding to its promoter, thus interfering the proliferation, cycle, and apoptosis of AML cells and accelerating the disease progression and drug resistance of patients with AML [36]. Notably, the significantly overlapped down-regulated genes (i.e. LCN2, CYBB, RNAS, and CD48) are thought to dominate the regulation of the immune microenvironment and the response to inflammation during AML relapse or drug resistance. For instance, Zhang et al. demonstrated that CD48 overexpression can lead to the activation of NK cells and the enhancement of their killing activity against AML cells in vitro [37]. In contrast, the decreased expression of CD48 in AML patients helps AML cells avoid NK cells’ recognition [38]. Therefore, this study suggests that genes that are significantly differentially expressed in patients with relapsed or drug-resistant AML may serve as novel biomarkers for monitoring AML prognosis and targeted therapeutics.

RBPs are key controllers in mediating post-transcriptional gene regulation (PTGR), and they regulate the intracellular fate of individual transcripts via interactions with RNA-binding domains (RBDs) [39]. Herein, we report that numerous RBP genes are significantly differentially expressed in patients with AML relapse in comparison to AML remission and Healthy controls, strongly correlating with AML progression and drug resistance. Similarly, it has also been demonstrated that normal and leukaemic cells can be distinguished on the basis of differential RBP expression, indicating their ability to define cellular identity [40, 41]. Thus, by analysing AML progression- and drug resistance-related RBPs, we can evaluate the disease status of AML patients and infer survival probability. This, in turn, may lead to the identification of new therapeutic targets and personalised precision oncology in predicting AML prognosis.

RAS signalling is an important mechanism for regulating the diversities of gene expression and the proteome amongst eukaryotes [42]. However, abnormal expression of RAS in cells may lead to the development of various diseases, such as atherosclerosis as well as lung and pancreatic adenocarcinoma [43, 44]. Research has also shown that aberrant RAS is one of the leading factors involved in the regulation of AML drug resistance, and it is usually closely associated with poor prognosis and drug resistance in AML [45]. For instance, aminoacyl-tRNA synthetase-interacting multifunctional protein 2 (AIMP2) is a non-enzymatic component required for the multi-tRNA synthetase complex, while the exon 2 skipping alternatively spliced variant of AIMP2 (AIMP2-DX2) compromises AIMP2 activity and is strongly correlated with major cancer signalling pathways, poor prognosis, and drug resistance in AML. Indeed, further analysis of the clinical patient cohort reveals that AIMP2-DX2 positive AML patients show decreased sensitivities to anticancer drugs and overall survival [46]. Here, we unravel the significant difference of RASGs and RASEs in patients with AML relapse compared with AML remission and Healthy subjects, especially for both intronR and A5SS alternative splicing types. Further GO analysis proved that the differential expression of RASGs very likely participates in the regulation of key RASEs in AML by affecting large numbers of sophisticated biological processes, such as RNA splicing, transcriptional regulation, and protein stabilisation, thus increasing the drug resistance of AML cells. Our findings strongly suggest that differentially expressed RASGs and RASEs are closely related to the regulation of drug resistance in AML.

In summary, this study demonstrates that dysregulated RBPs and aberrant AS are closely associated with disease progression and drug resistance in AML, supporting a regulatory axis in which RBP dysfunction drives AS alterations that contribute to chemo-resistance. Notably, abnormally expressed RBPs and RASEs may serve as potential prognostic biomarkers or therapeutic targets. Elucidating the precise mechanisms by which RBPs and AS regulate drug resistance will be essential to unravel the complex molecular landscape of resistance in AML and to inform the development of more effective therapies targeting resistance-related genes.

Nevertheless, several limitations should be acknowledged. The present study is based solely on bioinformatic analyses and lacks experimental validation in cellular or animal models. Specifically, the roles of key RBPs (IGF2BP2, NPM1, MYH9, and LGALS3) in regulating alternative splicing remain to be experimentally verified, and the impact of AS on drug-resistant phenotypes, including proliferation, apoptosis, and half-maximal inhibitory concentration (IC₅₀), has yet to be confirmed. Addressing these gaps will be a primary focus of our future research, with the goal of providing robust experimental evidence linking RBP dysregulation and aberrant AS to drug resistance in AML.

Conclusions

In this study, we demonstrate a mechanistic link between dysregulated RNA-binding proteins, aberrant alternative splicing, and drug resistance in relapsed AML. Specifically, our integrated analysis reveals that altered expression of 63 RBP genes correlates with widespread splicing changes in relapsed patients, supporting a regulatory axis hypothesis: differential RBP expression leads to aberrant alternative splicing, which in turn drives drug resistance. This regulatory framework positions RBPs as upstream drivers of splicing-mediated chemo-resistance. An in-depth mechanistic understanding of how specific RBPs modulate drug resistance-associated splicing events may inform the development of splicing-modulating therapies and RBP-targeted strategies for overcoming relapse in AML.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AML

Acute myeloid leukaemia

reAML

Patients with AML relapse

HSCs

Haematopoietic stem cells

DEGs

Differentially expressed genes

RBPs

RNA-binding proteins

mybl2

myb-related protein B gene

PTBP1

Polypyrimidine tract-binding protein 1

FLT3-ITD

Internal tandem duplication mutations within FMS-like tyrosine kinase-3

AS

Alternative splicing

ATR

Anti-thrombin receptor

PCCs

Pearson’s correlation coefficients

ASEs

Alternative splicing events

RASGs

Regulatory alternative splicing genes

RASEs

Regulatory alternative splicing events

RNA-seq

RNA sequencing

GO

Gene ontology

KEGG

Kyoto encyclopedia of genes and genomes

PPI

Protein-protein interaction

WGCNA

Weighted gene co-expression network analysis

ES

Exon skipping

A5SS

Alternative 5' splice site

A3SS

Alternative 3' splice site

MXE

Mutually exclusive exons

5pMXE

Mutually exclusive 5' UTRs

3pMXE

Mutually exclusive 3' UTRs

FPKM

Fragments per kilobase per million mapped reads

PTGR

Post-transcriptional gene regulation

Ara-C

Cytarabine

METTL3

Methyltransferase 3

NK

Natural killer

RBDs

RNA-binding domains

AIMP2

Aminoacyl-tRNA synthetase-interacting multifunctional protein 2

AIMP2-DX2

Exon 2 skipping alternatively spliced variant of AIMP2

Author contributions

YM Zhang and XY Dong contributed equally. JY Wang, YM Zhang and XY Dong conceived and designed the study. JY Wang, YM Zhang and XY Dong analysed the data and wrote the manuscript. JY Wang and YM Zhang supervised the study and revised the manuscript. W Zheng, J Xu and ZZ Wang analysed clinical samples and conducted the experiments. SM Ding and YY Wang collected clinical samples. All authors read and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (No. 82104609), Natural Science Foundation of Shandong Province (No. ZR2023LZY002), Medical and Health Technology Development Plan of Shandong Province (No. 202003041150) and High-level Talents Project of Affiliated Hospital of Shandong University of Traditional Chinese Medicine (No. rc0000002-41).

Data availability

The raw sequence data reported in this paper have been deposited in the Genome Sequence Archive (Genomics, Proteomics & Bioinformatics 2025) in National Genomics Data Center (Nucleic Acids Res 2025), China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (GSA-Human: HRA013447) that are publicly accessible at *https://ngdc.cncb.ac.cn/gsa-human* [47, 48].

Declarations

Ethics approval and consent to participate

The studies involving human bone marrow blood samples were reviewed and approved by the Medical Institutional Ethics Committee of the Affiliated Hospital of Shandong University of Traditional Chinese Medicine, China. Approval No., (2023) Review No. (143)-KY. Clinical trial number: not applicable. The informed consent to participate was obtained from all of the participants in the study. In addition, all methods in this study were performed in accordance with the Declaration of Helsinki.

Consent for publication

All listed authors have actively participated in the study and have read and approved the submitted manuscript.

Competing interests

The authors declare no competing interests.