DKC1: a robust prognostic factor and potential therapeutic target in multiple myeloma

Abstract

Background

Multiple myeloma (MM) is an incurable plasma cell malignancy. Dyskerin pseudouridine synthase 1 (DKC1), a nucleolar protein, is essential for RNA modification and cellular homeostasis, yet its role in MM remains unclear.

Methods

Prognostic significance of DKC1 in MM patients was evaluated using the MMRF CoMMpass and GEO datasets. Functional effects of DKC1 knockdown or overexpression were investigated via in vitro proliferation, apoptosis assays and in vivo xenografts. Transcriptomic profiling and CMC-based pseudouridine (Ψ) mapping were used to define DKC1-mediated regulation of ATF5.

Results

Elevated DKC1 expression was identified as an independent prognostic marker of poor outcomes in MM patients. Decision tree analysis demonstrated that integrating DKC1 expression further refined prognostic stratification beyond the ISS system. Functional assays revealed that DKC1 promoted MM cell proliferation, survival and colony formation, while DKC1 knockdown or pharmacologic inhibition with pyrazofurin significantly reduced MM cell proliferation and colony formation, increased apoptosis in vitro, and suppressed tumor growth in xenograft models. RNA sequencing analysis identified ATF5 as a downstream target of DKC1, and subsequent experimental validation confirmed that DKC1 exerts part of its function through ATF5. We further demonstrated that DKC1 knockdown reduces ATF5 mRNA stability through impaired pseudouridylation. Site-specific Ψ modifications on ATF5 mRNA confirmed a direct post-transcriptional regulatory mechanism.

Conclusions

DKC1 drives MM progression by promoting ATF5 stability through pseudouridylation, thereby enhancing myeloma cell proliferation and survival. These findings highlight that DKC1 may be used as a potential biomarker for risk stratification and a promising therapeutic target in MM.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-026-07916-6.

Introduction

Multiple myeloma (MM) ranks as the second most prevalent hematologic malignancy, constituting about 2% of all cancers and 10% of hematologic cancers. Globally, the incidence of MM has increased by 167% from 1990 to 2021 [1]. It is hypothesized that MM is a progressive disorder and the vast majority of MM patients start from monoclonal gammopathy of undetermined significance (MGUS) and smoldering MM (SMM) [2]. Moreover, among patients with MM, plasma cell leukemia (PCL) is recognized as an aggressive and clinically high-risk subtype [3]. Over time, various risk stratification systems were introduced. Durie-Salmon staging system was used historically since 1975 based on clinical features. In 2005, based on levels of albumin (ALB) and beta2-microglobulin (B2M), the international staging system (ISS) was introduced to classify patients into three stages, acting as a simpler and more reliable prognostic tool. Then, this score was integrated with lactate dehydrogenase level and the presence of high-risk cytogenetic abnormalities [4]. When it comes to treatment, significant advances have been made over the past few decades, including the introduction of proteasome inhibitors, immunomodulatory drugs, monoclonal antibodies, CAR T-cell therapy, and autologous stem cell transplantation. Despite these developments, MM remains incurable, primarily due to the emergence of drug resistance and disease relapse [5, 6], highlighting the urgent need for novel therapeutic strategies.

Dyskerin pseudouridine synthase 1 (DKC1) gene encodes a protein named dyskerin, which is a L-shaped protein of 514 residues with a molecular weight of 58 kDa. Mutations in DKC1 cause the premature aging syndrome X-linked dyskeratosis congenita (X-DC), an incurable disorder that typically leads to bone marrow failure [7]. When dyskerin acts as a pseudouridine synthase, it exerts its function by forming a heterotetrameric complex that binds H/ACA motif-containing RNAs, such as the H/ACA-box small nucleolar RNAs (snoRNAs), and Cajal body RNAs (scaRNAs) to mediate site-specific pseudouridylation [8, 9]. Through this complex, dyskerin targets rRNAs and snRNAs, contributing to rRNA modification and precursor mRNA splicing [9, 10]. Moreover, the RNA component of telomerase (TERC) also contains H/ACA motifs, and dyskerin is essential for regulating its accumulation to ensure proper telomerase activity for telomere extension [11, 12]. While impaired telomerase function contributes to X-DC, mutations affecting the pseudouridine synthase catalytic domain of dyskerin cause more severe disease phenotypes, highlighting the critical role of dyskerin’s enzymatic function beyond telomerase regulation [13]. Despite these advances, investigations into mRNA pseudouridylation remain limited, in part because mRNAs are less abundant compared with rRNAs and snRNAs. Consequently, the functional consequences of dyskerin-mediated pseudouridylation on mRNA processing and its potential impact on human disease are still not fully understood.

DKC1 overexpression has been reported in various malignancies, including neuroblastoma, colorectal cancer, breast cancer, endometrial cancer [14] and hepatocellular carcinoma, where its high expression is consistently associated with poor prognosis [15–19], suggesting that DKC1 may function as an oncogene in these cancers. In contrast, in spontaneous pituitary tumors and cutaneous squamous cell carcinoma, DKC1 acts as a tumor suppressor by impairing p27 IRES-dependent translation and suppressing mevalonate pathway activation, respectively, thereby inhibiting tumorigenesis [20, 21]. Together, these findings indicate that DKC1 plays context-dependent roles in cancer.

To date, in MM, the contribution of DKC1 to disease progression, as well as its role in mRNA regulation, remains largely uncharacterized. In this study, we identified DKC1 as a critical driver of MM and an independent prognostic marker for patient outcomes. Importantly, DKC1 may complement the ISS staging system to improve the accuracy of risk stratification. Functional assays demonstrated that DKC1 is required for MM cell proliferation and survival, while its inhibition suppresses growth and induces apoptosis in vitro and in vivo. Mechanistically, DKC1 exerts its oncogenic effects by stabilizing ATF5 through pseudouridylation, revealing a novel regulatory axis in MM pathogenesis. Collectively, these findings establish DKC1 as a robust prognostic factor and a promising therapeutic target in MM.

Methods

Bioinformatics analysis

The transcriptome and clinical data of Multiple Myeloma Research Foundation (MMRF) CoMMpass data were retrieved from The National Cancer Institute (NCI) Genomic Data Commons (GDC) dataset (accessed in July 2024), and GSE136324, GSE2113, GSE164701, GSE164703 and GSE164706 datasets were obtained from Gene Expression Omnibus (GEO) database (accessed in August 2024). Protein expression data were collected from a published article reporting proteomic profiling of MM and related plasma cell disorders [22]. The “Combat” algorithm in the R package “SVA” was used to remove batch effects when merging different GEO datasets. DKC1 expression difference between groups was analyzed using “ggpubr” package. Patients were divided into two groups upon median DKC1 level, and Kaplan–Meier (KM) survival analysis was conducted to explore the correlation between DKC1 and survival in all patients using “survminer” and “survival” packages. To evaluate the prognostic significance of various factors in MM, we performed both univariate and multivariate Cox regression analyses. The variables included in the analyses were DKC1 level, ISS stage, B2M, ALB, age, and sex for both overall survival (OS) and progression-free survival (PFS). Variables with a P-value of less than 0.05 in the multivariate Cox regression analysis were considered independent prognostic factors. Moreover, to assess whether incorporating DKC1 provides incremental prognostic value beyond clinical covariates alone, we compared the clinical covariates-only model with the full model (clinical covariates and DKC1 expression) for both OS and PFS. Predictive performance was quantified using the concordance index (C-index) and integrated Brier score (IBS) [23]. We then used the “rms” package to develop a nomogram and assessed the predictive nomogram accuracy through calibration curve plotting, with 1000 bootstrap resamples for internal validation. To classify categorical outcomes or predict continuous values, decision tree was used. It facilitates statistical analysis by utilizing the Classification and Regression Tree (CART) algorithm, which functions as a modeling technique structured in the form of a tree-like diagram. We built a decision tree to predict risk stratification based on B2M, ALB, DKC1 levels using “rpart” (Recursive Partitioning and Regression Trees) package [24]. Meanwhile, internal validation was conducted using bootstrap validation with 1000 repetitions, and predictive performance was assessed by time-dependent receiver operating characteristic (time-dependent ROC) curve using the “timeROC” package [25]. The correlation between genes was assessed using the Spearman rank correlation method. The R software version was 4.2.1.

Cell lines and cell culture

The MM cell lines (RPMI8226 and MM.1S) were obtained from Zhong Qiao Xin Zhou Biotechnology Co., Ltd (Shanghai, China). They were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium (Gibco, Grand Island, NY, USA), supplemented with 10% fetal bovine serum (FBS; 10100147, Gibco) and 100 U/mL penicillin–streptomycin (15140122, Gibco). Cells were maintained in a humidified incubator at 37 °C with 5% CO₂ and monitored for mycoplasma infection every six months by using MycoBlue Mycoplasma Detector (D101-02, Vazyme, Nanjing, China). Peripheral blood samples from three healthy donors were collected after informed consent, and peripheral blood mononuclear cells (PBMCs) were isolated using human lymphocyte separation medium (Ficoll, P8900, Solarbio, Beijing, China). Similarly, bone marrow samples were collected from ten MM patients (plasma cell proportion 67.95% ± 15.48%) and five healthy controls after informed consent. Bone marrow mononuclear cells were isolated using the same separation medium. The collection and use of human samples were approved by the Ethics Committee of the Second Qilu Hospital of Shandong University (No. KYLL2024737).

Transfection of lentiviral vectors

The shRNA sequence targeting the DKC1 gene (GCTCAGTGGCTGTATGATA) and ATF5 gene (GCTGGAACAGATGGAAGACTT), along with the control sequence (CCTAAGGTTAAGTCGCCCTCG) were cloned into the GL427 vector (OBiO Technology, Shanghai, China) for gene knockdown. The coding sequences of the DKC1 gene (NM_001363.5) and the ATF5 gene (NM_001193646.2) were cloned into the GL107 vector (OBiO Technology) for overexpression studies, with an empty vector used as the control. After lentiviral transduction, stable DKC1 knockdown and overexpression cell lines were selected using puromycin (P8230, Solarbio), while ATF5 knockdown and overexpression cell lines were generated through blasticidin (B9300, Solarbio) selection.

Real-time quantitative polymerase chain reaction (RT-qPCR)

Total RNA was extracted with SteadyPure Quick RNA Extraction Kit (AG20213, Accurate Biology, Wuhan, China) and subsequently reverse transcribed with Hifair® 1st Strand cDNA Synthesis SuperMix (11141ES10, Yeasen, Shanghai, China), according to the manufacturer’s instructions. Finally, ChamQ Universal SYBR qPCR Master Mix (Q711-02, Vazyme) was used in QuantStudio™ 5 Real-Time PCR System (Thermo Fisher Scientific, Waltham, USA). The relative mRNA expression of the target genes was quantified using the 2^−ΔΔCt method, with GAPDH as the reference gene. The primer sequences used were listed in Table S1.

Western blot analysis

Total protein was extracted using RIPA buffer (R0010, Solarbio) and after determining protein concentrations by Bicin-choninic Acid (BCA) Protein Assay Kit (ZJ101, Epizyme Biotech, Shanghai, China), equal amounts of protein (30 µg per lane) were separated via SDS-PAGE and transferred onto PDVF membranes (ISEQ00010, Millipore, MA, USA). Then, membranes were blocked with 5% non-fat milk, and incubated with primary antibodies and secondary antibodies before imaged with chemiluminescence imaging system (Tanon, Shanghai, China). Band density was quantified using Image J v1.54f software and normalized to β-actin. The following primary antibodies were used: DKC1 (A4407, ABclonal, Wuhan, China), ATF5 (AP0692, ABclonal), Bax (5023S, CST, Danvers, MA, USA), Bcl-2 (A19693, ABclonal) and β-actin (AC026, ABclonal).

EdU assay

Cell proliferation was assessed using an EdU assay kit (C0075S, Beyotime, Shanghai, China) according to the manufacturer’s protocols. Cells were seeded into 6-well plates, and cell proliferation was assessed 2 hours after the addition of EdU solution. To investigate the impact of pyrazofurin (PF, SML1502, Sigma-Aldrich, St. Louis, USA) [26] on cell proliferation, cells were treated with PF at final concentrations of 0.2, 0.4, and 0.8 µM for 72 h, with an equivalent volume of vehicle used as the control (0 µM PF). Finally, fluorescence signals were captured using a CytoFLEX flow cytometer (Beckman Coulter, Brea, CA, USA).

Soft agar colony formation assay

According to the literature [27], 0.55% and 0.3% agar, prepared by mixing agar with 2× RPMI 1640 medium, were used to create two distinct layers in a six-well plate. The 0.55% agar was added to form the bottom layer, while the 0.3% agar, containing 1 × 10³ cells, was added to form the top layer. Then 100 μL of medium was added twice weekly over the upper layer of agar to avoid desiccation. Once the colonies were visible, photographs of the wells were taken using imager and inverted microscope respectively. Colonies were counted using image J software.

Cell cycle and apoptosis analysis

Cells were collected and washed twice using cold PBS. Apoptotic cells were detected using the Annexin V-PE/7-AAD apoptosis detection kit (A213-02, Vazyme). To assess the effect of PF on the apoptosis in MM cells and patient-derived PBMCs, different concentrations of PF were added. Cell cycle distribution was assessed using a Cell Cycle Detection Kit (CCS012, MultiSciences, Hangzhou, China) following the manufacturer’s instructions. Finally, apoptotic cells and cell cycle phases were analyzed and quantified using a CytoFLEX flow cytometer.

Cell invasion assay

1 × 10⁵ cells were resuspended with serum-free medium and seeded into the upper chamber of transwell insert (3422, Corning, NY, USA) coated with Matrigel (356234, BD Biosciences, San Jose, CA, USA) diluted 1:8 in RPMI 1640 medium. Meanwhile, 600 μL of medium containing 20% FBS was added to the lower chamber as a chemoattractant. After incubation at 37 °C for 24 hours, cells across the membrane were fixed with 4% paraformaldehyde and stained with 0.1% crystal violet (C0121; Beyotime). Finally, five random fields of cells were photographed using a microscope (Leica, Wetzlar, Germany) and the purple-stained cells were quantified with Image J software [28].

MM mouse model

NCG mice (6–8 weeks old) were purchased from GemPharmatech Co., Ltd. (Jiangsu, China). MOCK1- and shDKC1–RPMI8226 cells (8 × 10⁶ cells per mouse) were injected subcutaneously into the right axillary region of 6 mice per group. When the tumor volume reached 150 mm³, tumor establishment was confirmed, and this time point was defined as Day 0. Tumor volumes were measured every two days and calculated using the formula: V = L×W²/2 (L = long diameter, W = short diameter). When the tumors reached a diameter of 15 mm or if the mice were unable to eat or drink during the observation period, they were sacrificed by cervical dislocation.

For PF treatment, luciferase-labeled RPMI8226 cells (8 × 10⁶) were injected subcutaneously into the right axillary region of mice. Once tumor volume reached 150 mm³ (Day 0), mice were randomly assigned into two groups (n = 5 per group). Mice in the control group were treated with vehicle, whereas those in the experimental group received PF treatment (5 mg per kg, intraperitoneal injection, every 3rd day) [29]. To monitor tumor development, the mice were intraperitoneally injected with 3 mg of D-luciferin (abs42017259, Absin, Shanghai, China) per mouse on Days 7 (D7) and 11 (D11). Tumor progression was then observed using a live bioluminescence imaging system (IVIS Spectrum, PerkinElmer, Waltham, MA). Similarly, when the tumors reached a diameter of 15 mm or if the mice were unable to eat or drink during the observation period, they were sacrificed by cervical dislocation.

The animal experiments were approved by the Ethics Committee of the Second Qilu Hospital of Shandong University (no. KYLL2024737) and were conducted in accordance with legal requirements and national guidelines.

Hematoxylin and Eosin (H&E) staining and Immunohistochemical (IHC) analysis

HE staining was performed to evaluate the effects of PF on the histopathological features of visceral organs in mice. Paraffin sections were dewaxed and rehydrated, followed by nuclear staining with Hematoxylin for 5 minutes and treatment with a differentiating solution. The cytoplasm was counterstained with Eosin for 10 seconds to 2 minutes. Sections were then dehydrated through graded ethanol, cleared in xylene, and mounted with resin.

IHC was performed to evaluate Ki67, Ly6G and CD68 expression. Paraffin-embedded tissue sections were deparaffinized, rehydrated, and subjected to antigen retrieval. After blocking endogenous peroxidase activity, sections were incubated overnight at 4 °C with primary antibody. After incubation with secondary antibody for 50 minutes, target signals were visualized using diaminobenzidine (DAB, ZLI-9018, ZSGB-BIO, Beijing, China) as the chromogen. Counterstaining was performed with hematoxylin. The following primary antibodies were used: anti-Ki67 (GB121141, Servicebio, Wuhan, China), anti-CD68 antibody (GB113109, Servicebio), anti-Ly6G antibody (GB11229, Servicebio). TUNEL staining was performed to detect apoptotic cells using a commercial kit (G1507, Servicebio) according to the manufacturer’s instructions.

Image acquisition was performed using a digital pathology scanner (NanoZoomer S60, Japan).

mRNA sequencing

mRNA sequencing was performed on MOCK1 and DKC1 knockdown RPMI8226 cells, with three biological replicates per group, by Sinotech Genomics Co., Ltd. (Shanghai, China). After RNA extraction and purification, as well as the preparation of RNA libraries, paired-end sequencing was carried out using a next-generation sequencing platform based on Illumina technology. Gene abundance was expressed as fragments per kilobase of exon per million mapped reads (FPKM). Differential expression analysis of mRNA between MOCK1 and DKC1 knockdown groups was conducted using the R package edgeR. RNAs with |log₂(FC)| > 1, a q-value < 0.05, and a mean FPKM > 1 in one group were considered significantly altered and selected for subsequent analysis. Moreover, we performed a Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis via enrich R package.

mRNA stability assay

MM cells were exposed to actinomycin D (5 µg/mL, HY-17559, MedChemExpress, NJ, USA) for 0, 0.5, 1.5, 2, 4 and 8 hours and mRNA levels at each time point were normalized to the mRNA abundance of 0 h. The mRNA half-life (t_1/2) was calculated using nonlinear regression (one-phase decay) in GraphPad Prism v9.0.0 (GraphPad Software, La Jolla, CA, USA).

RNA immunoprecipitation (RIP) assay

RIP assay was conducted using RNA immunoprecipitation kit (Bes5101; BersinBio, Guangzhou, China). MM cell lysates were incubated with an anti-DKC1 antibody (sc-373956, Santa Cruz Biotechnology, CA, USA) or an IgG control antibody, and RNA-protein complexes were captured using magnetic beads. After multiple washes to remove non-specific bindings, RNA was extracted and reverse transcribed into cDNA. The enrichment of ATF5 mRNA bound to DKC1 was evaluated by RT-qPCR.

CMC-based detection of pseudouridine (Ψ) at specific genomic loci

To identify Ψ sites within the target mRNA, we first employed a N-cyclohexyl-N′-β-(4-methylmorpholinium) ethylcarbodiimide p-tosylate (CMC)-based method as described in prior study [30]. CMC (C106402, Sigma) selectively react with Ψ residues in RNA, causing reverse transcription (RT) termination [31, 32], thereby leading to elevated cycle threshold (Ct) values during RT-qPCR. By comparing Ct values between CMC-treated (CMC+) and untreated (CMC-) groups, regions potentially harboring Ψ modifications can be predicted. To ensure broad coverage, we designed seven primer pairs spanning the entire coding sequence (CDS) of the gene (primer sequences are listed in Table S2).

Subsequently, to achieve single-nucleotide resolution, we employed Mn^{2 +} -dependent RT condition, which enable read-through of Ψ sites rather than termination while introducing characteristic mutations or deletions at or near the modified bases. These mutation signatures, when analyzed by Sanger sequencing, enabled precise mapping of Ψ sites [33]. After fragmentation, EDTA denaturation, and CMC treatment, 10 μL RNA for each group was annealed with 1 µL of 100 µM random hexamer primers (SO142, Thermo Fisher Scientific, Waltham, USA) at 65 °C for 5 min and immediately chilled on ice. The RNA-primer mix was then added to 8 µL of freshly prepared RT buffer (125 mM Tris-HCl, pH8.0, 15 mM MnCl₂, 187.5 mM KCl, 1.25 mM dNTPs, 25 mM DTT, with a final Mn²⁺ concentration 6 mM). Then after incubation at 25 °C for 2 min, 1 µL of SuperScript II reverse transcriptase (18064014, Thermo Fisher Scientific) was added. RT was performed at 25 °C for 10 min and 42 °C for 3 h, followed by termination of the reaction at 70 °C for 15 min. The resulting cDNA was used as a template for RT-qPCR, followed by TA cloning and Sanger sequencing to identify Ψ sites.

Statistical analysis

Experimental data analyses were performed using GraphPad Prism v9.0.0. Data were presented as mean ± SD, and unpaired Student’s t-test, One-way ANOVA, two-way ANOVA, Wilcox, or Kruskal–Wallis test were used to compare differences between groups based on the distribution of data. A P-value less than 0.05 was considered statistically significant.

Results

Higher DKC1 expression is correlated with adverse clinical features

To investigate the role of DKC1 in MM, we analyzed public datasets to explore its associations with clinical characteristics. Analysis of the GSE2113 dataset revealed a stepwise increase in DKC1 expression from MGUS to MM, peaking in PCL (Fig. 1A). Integrated analysis of multiple datasets (GSE164701, GSE164703, GSE164706) validated this finding, showing that PCL patients exhibited significantly higher DKC1 expression than MM patients (Fig. S1A). qPCR analysis further demonstrated that DKC1 mRNA expression was significantly elevated in MM patients compared to controls (Fig. S1B), and clinical characteristics of the patients were summarized in Table S3. Consistently, protein-level analysis demonstrated a similar upward trend from healthy controls to MGUS, MM, and PCL (Fig. 1B). Moreover, in the MMRF CoMMpass dataset, DKC1 expression was significantly elevated in patients with recurrent MM compared to primary MM (Fig. 1C). DKC1 expression was also higher in patients with advanced stages (Fig. 1D).

Fig. 1.

DKC1 expression is correlated with adverse clinical features and serves as a biomarker for prognostic evaluation and disease staging in MM. A DKC1 mRNA expression in MM and PCL patients compared to MGUS patients from GSE2113 dataset. B DKC1 protein expression across healthy controls, MGUS, MM, and PCL patients. C, D DKC1 mRNA expression analysis from the MMRF CoMMpass dataset. C DKC1 mRNA expression in primary and recurrent patients. D DKC1 mRNA expression across different ISS stages. E Kaplan-Meier survival analysis of OS in MMRF CoMMpass dataset for DKC1-high (n = 392) and DKC1-low (n = 393) patients. F Kaplan-Meier survival curves for PFS in GSE136324 dataset for DKC1-high (n = 218) and DKC1-low (n = 218) patients. G, H Univariate (left) and multivariate (right) Cox regression analyses for OS (G) and PFS (H), using data from MMRF CoMMpass dataset and GSE136324 dataset respectively. I Decision tree analysis based on DKC1, B2M and ALB for risk stratification in MM patients. J Time-dependent ROC curve evaluating the predictive performance of the model over time. K Estimation of variable importance based on the decision tree model. Abbreviations: plasma cell leukemia (PCL); multiple myeloma (MM); monoclonal gammopathy of undetermined significance (MGUS); overall survival (OS); progression-free survival (PFS); β2-microglobulin (B2M); albumin (ALB); receiver operating characteristic (ROC)

We further explored the prognostic significance of DKC1 expression in all patients. The results of MMRF CoMMpass dataset showed that patients with higher DKC1 expression had poorer OS (Fig. 1E). Meanwhile, analysis of all patients in GSE136324 dataset revealed that higher DKC1 expression was correlated with poorer PFS (Fig. 1F).

DKC1 serves as a biomarker for prognostic evaluation and disease staging in MM

We further explored the prognostic significance of DKC1 expression. Univariate Cox regression analyses of MMRF CoMMpass dataset indicated that DKC1 expression, ISS stage and B2M were associated with OS, and multivariate analyses revealed that DKC1 expression and ISS stage served as independent prognostic factors (Fig. 1G). Meanwhile, analysis of GSE136324 dataset demonstrated that DKC1 expression, ISS stage, B2M, ALB, age and sex were associated with PFS, whereas DKC1 expression, B2M, ALB and age emerged as independent predictors of PFS (Fig. 1H). Thus, DKC1 served as an independent prognostic factor for both OS and PFS. We then developed a prognostic nomogram integrating DKC1 expression and ISS stage for OS (Fig. S2A), and the calibration curves demonstrated good concordance between predicted and observed outcomes (Fig. S2B). Similarly, we constructed nomogram incorporating DKC1 expression, B2M, ALB and age for PFS (Fig. S2C, and the calibration curves showed strong concordance between the predicted and observed outcomes (Fig. S2D). Besides, to determine whether DKC1 adds prognostic value beyond clinical variables, we compared clinical-only model with full model incorporating DKC1, using C-index and IBS for evaluation. For OS, adding DKC1 to the ISS stage model improved the C-index from 0.666 (SE = 0.027) to 0.706 (SE = 0.027) and lowered the IBS from 0.127 to 0.124. For PFS, the addition of DKC1 to the clinical covariates-only model (ALB, age, B2M) resulted in a slight increase in the C-index from 0.635 (SE = 0.0197) to 0.640 (SE = 0.0186) and a reduction in the IBS from 0.170 to 0.166. These results indicate that incorporating DKC1 improves the prognostic precision of both OS and PFS models. Moreover, it was widely acknowledged that B2M and ALB served as key determinants in establishing the ISS staging system. We applied a decision tree to evaluate the role of DKC1 expression in the staging. The patients were divided into four distinct risk stratification groups according to B2M and DKC1 levels, but ALB was automatically excluded from this classification by the model (Fig. 1I). Bootstrap internal validation demonstrated that the model maintained stable discriminative performance, with area under the curve (AUC) values of 0.78 (95% CI, 0.68–0.88), 0.79 (95% CI, 0.69–0.88), and 0.77 (95% CI, 0.67–0.88) at 1, 2, and 3 years, respectively, as shown in the time-dependent ROC curve (Fig. 1J). Moreover, the “rpart” package used for decision tree analysis allowed for the estimation of variable importance scores. The result showed that DKC1 had the highest score compared with B2M and ALB (Fig. 1K). Together, these results highlight the notable prognostic and staging value of DKC1.

DKC1 is required for MM cell proliferation/cell cycle progression, survival and invasion

Given the findings above, we conducted gain- and loss-of-function studies to directly determine its impact on MM cells. The knockdown (RPMI8226 and MM.1S cells) and overexpression (RPMI8226 cells) efficiencies were confirmed by RT-qPCR and Western blot (Fig. S3A). Cell proliferation assessed by EdU assay (Fig. 2A) and colony formation measured by soft agar assay (Fig. 2B) in DKC1-knockdown (shDKC1) RPMI8226 and MM.1S cells were significantly reduced compared to control cells (MOCK1). In contrast, DKC1 overexpression (oeDKC1) in RPMI8226 cells led to a marked increased proliferation compared to control cells (MOCK2). Meanwhile, treatment of DKC1 inhibitor pyrazofurin (PF) significantly inhibited the growth and colony formation of RPMI8226 cells (Fig. 2C, D). Consistently, cell cycle analysis revealed that DKC1 knockdown significantly reduced the proportion of cells in the S phase, with DKC1 overexpression exhibiting the opposite trend (Fig. S3B). Moreover, DKC1 knockdown induced apoptosis in the RPMI8226 and MM.1S cells (Fig. 3A, B). Consistently, Bcl-2 protein level decreased, whereas Bax protein level increased (Fig. S3C), supporting the pro-apoptotic effect of DKC1 suppression. Conversely, DKC1 overexpression reduced the proportion of apoptotic cells at various time points under nutrient deprivation conditions, with no significant difference of apoptosis rates in 10% FBS (Fig. 3C). Furthermore, the DKC1 inhibitor PF induced apoptosis of MM cells (Fig. 3D), without affecting apoptosis of PBMCs from patients (Fig. S3D). Besides, the transwell assay showed that knockdown of DKC1 inhibited cell invasive ability, while ectopic DKC1 expression stimulated this capability (Fig. S3E).

Fig. 2.

DKC1 promotes proliferation of MM cells. A, B EdU staining (A) and soft agar colony formation (B) of MM cells following DKC1 knockdown (shDKC1) in RPMI8226 and MM.1S cells or DKC1 overexpression (oeDKC1) in RPMI8226 cells, with corresponding control cells (MOCK1 for knockdown, MOCK2 for overexpression) (scale bar = 200 µm). C, D EdU staining (C) and soft agar colony formation (D) of MM cells treated with PF at 0.2, 0.4, and 0.8 µM, with an equivalent volume of vehicle used as the control (0 µM PF) (scale bar = 200 µm). Data are presented as mean ± SD from three independent experiments. Abbreviations: pyrazofurin (PF)

Fig. 3.

DKC1 reduces apoptosis of MM cells. A, B Flow cytometric analysis of apoptosis in shDKC1 and MOCK1 MM cells, with (A) RPMI8226 cells and (B) MM.1S cells, using Annexin V and 7-AAD staining. C Flow cytometric analysis of apoptosis in oeDKC1 and MOCK2 MM cells following culture with or without FBS for 24, 48, and 72 h. D Apoptosis of RPMI8226 cells was assessed after treatment with PF at 0.2, 0.4, and 0.8 µM for 72 h, with an equivalent volume of vehicle used as the control (0 µM PF). Data are presented as mean ± SD from three independent experiments

DKC1 inhibits MM tumor development in vivo

To explore the function of DKC1 in vivo, shDKC1- and MOCK1–RPMI8226 cells were subcutaneously injected into NCG mice (Fig. 4A). The results showed that the tumors derived from the shDKC1 group grew more slowly, thereby forming significantly smaller masses (Fig. 4B, C). while the tumor weight was lower than those in MOCK1 group (0.70 ± 0.37 g vs. 2.52 ± 0.71 g) (Fig. 4D). Besides, Ki67 staining showed a lower proportion of positive cells in the shDKC1 group, indicating reduced proliferation (Fig. 4E), whereas TUNEL staining revealed an increased number of positive cells, suggesting enhanced apoptosis in these tumors (Fig. 4F).

Fig. 4.

Both DKC1 knockdown and PF treatment suppress tumor growth in MM mouse model. A Experimental schematic comparing the MOCK1 and shDKC1 groups in the MM mouse model. B-D Representative tumor photographs (B), tumor growth curves (C) and tumor weights (D) in the MOCK1 and shDKC1 groups, with six mice per group. For panels (C), ** indicates p = 0.0056 at day 2, **** indicates p < 0.0001 at days 4, 6, 8, respectively for MOCK1 (blue) vs shDKC1 (red). E, F Representative immunohistochemical staining of Ki67 (E) and TUNEL (F) in MOCK1 and shDKC1 groups at × 20 magnification (scale bar = 100 µm) and × 40 magnification (scale bar = 50 µm). Data were obtained from 3 mice per group. G Schematic representation of the PF treatment regimen in the MM mouse model. Briefly, MM xenograft models were established and mice were randomized to two groups. Mice in the experimental group received PF (5 mg/kg, intraperitoneal injection, every third day), whereas those in the control group were administered vehicle. H-K Representative tumor photographs (H), tumor growth curves (I), tumor weights (J) and bioluminescence monitoring (K) of tumors in PF and vehicle groups, with five mice per group. For panel (I), * indicates p = 0.0218 at day 8, **** indicates p < 0.0001 at days 10, 12, respectively for vehicle (blue) vs PF (red). L, M Representative immunohistochemical staining of Ki67 (L) and TUNEL (M) in vehicle and PF groups at × 20 magnification (scale bar = 100 µm) and × 40 magnification (scale bar = 50 µm). Data were obtained from 3 mice per group. All quantitative data are presented as mean ± SD. Abbreviations: pyrazofurin (PF)

To evaluate the in vivo effect of PF, we administered drug and vehicle treatments to mice after tumor formation (Fig. 4G). The treated mice had smaller tumor volumes (Fig. 4H). Additionally, tumor growth was much slower, and tumor weight was lower (0.64 ± 0.09 g vs. 1.76 ± 0.32 g) (Fig. 4I, J). To further validate these findings, bioluminescence monitoring was performed, which consistently showed reduced tumor growth in the treated mice (Fig. 4K). Ki67 and TUNEL staining in the treated tumors also demonstrated decreased proliferation and increased apoptosis, respectively (Fig. 4L, M). Besides, no significant differences were observed in Ly6G (neutrophil marker) and CD68 (macrophage marker) expression levels between the treated and control groups (Fig. S4). Furthermore, H&E staining revealed that PF treatment caused no significant damage to the organs of the treated mice (Fig. S5).

ATF5 is a downstream effector of DKC1 in MM cells

To explore the mechanisms underlying DKC1 function, we performed mRNA sequencing on the shDKC1- and MOCK1–RPMI8226 cells. The volcano plot highlights the genes with the most significant expression differences between the two groups (Fig. 5A). KEGG pathway analysis revealed downregulated pathways including amino acid biosynthesis, metabolic pathways, and glycine, serine, and threonine metabolism, as well as upregulated pathways such as the mRNA surveillance pathway, nucleocytoplasmic transport, and spliceosome (Fig. S6).

Fig. 5.

ATF5 is the potential downstream effector of DKC1 in MM cells. A Volcano plot of differentially expressed mRNAs between MOCK1 and shDKC1 RPMI8226 cells. B Venn diagram showing overlapping genes from GO enrichment analysis between MOCK1 and shDKC1 cells involved in apoptosis and cell proliferation pathways. C Heatmap of the top 30 overlapping genes identified from GO enrichment analysis, shown with three replicates per group. D Scatter plot showing the relationship between ATF5 and DKC1 mRNA expression, using data from the MMRF CoMMpass dataset. E, F ATF5 mRNA expression (E) and ATF5 protein levels (F) in DKC1 knockdown and overexpression cells, compared to their respective control groups. G, H mRNA stability assay showing the decay rate of ATF5 mRNA in MOCK1 and shDKC1 cells. I RIP assay showing DKC1 binding to ATF5 mRNA. Data are presented as mean ± SD from three independent experiments, each with three technical replicates (wells) for qPCR. Abbreviations: gene ontology (GO)

Since our experiments revealed that DKC1 affected various cellular functions, such as growth and apoptosis, we used a Venn diagram to display the distribution and overlapping patterns of genes associated with the two key biological processes (apoptotic process and cell population proliferation) in the Gene Ontology (GO) enrichment. A total of 307 genes were included in this analysis. Notably, 70 genes were identified to be simultaneously associated with the two processes (Fig. 5B). Among them, activating transcription factor 5 (ATF5) exhibited the largest expression variation (log₂FC = −3.46, q-value = 3.0 × 10⁻⁶⁶) (Fig. 5C). Besides, bioinformatics analysis of CoMMpass dataset indicated that the expression of ATF5 was positively correlated with the expression of DKC1 (Spearman r = 0.17, p = 2.7 × 10⁻⁶) (Fig. 5D). Consistently, RT-qPCR and Western blot results confirmed the reduced expression of ATF5 following DKC1 knockdown in RPMI8226 and MM.1S cells, while the increased expression of ATF5 in DKC1-overexpressed RPMI8226 cells (Fig. 5E, F). We thus assessed the mRNA stability of ATF5, and our results unraveled that DKC1 knockdown resulted in a reduced ATF5 mRNA half-life (Fig. 5G, H). Moreover, RIP assay revealed that DKC1 binds to ATF5 mRNA (Fig. 5I), and AlphaFold3 modeling predicted potential interaction sites between DKC1 and ATF5 mRNA (Fig. S7).

Since DKC1 regulates the stability of mRNA through pseudouridylation, we aimed to identify the pseudouridine (Ψ) modification sites on ATF5 mRNA. Accordingly, seven primer pairs (Primer1 - Primer7) were designed, and Ct values corresponding to each region were compared between CMC-treated (CMC+) and untreated (CMC-) groups. Based on the Ct differences observed across all regions, we selected the top four regions—amplified by primer 1 (Region1: 382–467), 5 (Region5: 835–965), 6 (Region6: 974–1145), and 7 (Region7: 1114–1264)—as candidates likely to harbor Ψ sites (Table S4).

RT read-through strategy for site-specific detection revealed putative Ψ modification sites at nucleotide positions 404, 860, 1096, and 1190 within ATF5 mRNA. Importantly, knockdown of DKC1 resulted in a drastic reduction of these Ψ modifications, as evidenced by the nearly complete absence of RT-induced mutations or deletions at these sites (Fig S8A-D). Together, these findings demonstrate that DKC1 modifies ATF5 mRNA via pseudouridylation, thereby promoting its stability.

To further verify whether DKC1 regulates oncogenic activity of MM cells via ATF5, rescue experiments were performed. The results showed that overexpression of ATF5 partially reversed the effects of DKC1 knockdown on cell proliferation and apoptosis, indicating that ATF5 mediates, at least in part, the downstream effects of DKC1 depletion (Fig. 6A–C). Conversely, knockdown of ATF5 partially abrogated the proliferative and anti-apoptotic effects induced by DKC1 overexpression (Fig. 6A, B and D), Furthermore, ATF5 overexpression restored Bcl-2 protein level while suppressed Bax expression in DKC1-depleted cells. In contrast, ATF5 knockdown decreased Bcl-2 and increased Bax expression in cells overexpressing DKC1 (Fig. S9A-I). Together, these findings indicated that ATF5 functioned as a critical downstream effector of DKC1 required for malignant phenotype of MM cells.

Fig. 6.

ATF5 rescues the cellular phenotypes induced by DKC1 knockdown in MM cells. A, B EdU assay (A) and soft agar colony formation assay (B) were performed in MOCK1, shDKC1, and shDKC1+oeATF5 groups, alongside MOCK2, oeDKC1, and oeDKC1+shATF5 groups. C Apoptosis analysis in MOCK1, shDKC1, and shDKC1+oeATF5 groups. D Apoptosis analysis following 72 h serum starvation in MOCK2, oeDKC1, and oeDKC1+shATF5 groups. Data are presented as mean ± SD from 3 independent experiments. E Model illustrating the DKC1–ATF5-Bcl-2 axis. Reduced DKC1 expression decreases ATF5 mRNA stability, downregulates Bcl-2, and consequently inhibits proliferation while promoting apoptosis in MM cells

Discussion

Despite improvements in rates of remission and survival from advances in treatments including stem cell transplantation, novel chemotherapeutic agents, and immunotherapy strategies, MM remains an incurable disease. Recently, CAR-T cell therapy has shown promising efficacy in relapsed or refractory MM patients, achieving high initial remission rates. However, its long-term clinical benefits remain limited, with 5-year PFS and OS rates of 21.0% and 49.1%, respectively, and 83.8% of patients experienced disease progression and/or death after initial remission [34]. These limitations highlight the pressing need for innovative therapeutic approaches that exploit alternative mechanisms or novel targets. Our study identified DKC1 as a robust independent prognostic biomarker for patient survival. Moreover, DKC1 knockdown markedly inhibited proliferation, induced apoptosis of MM cells and significantly delayed tumor progression in xenograft models. These effects were recapitulated by pharmacological DKC1 inhibition, further supporting its candidacy as a therapeutic target in MM.

For MM patients, precise staging is critical for tailoring the most appropriate initial therapy, thereby optimizing both the depth and durability of therapeutic response. Although several validated clinical staging systems are currently in use, a considerable proportion of patients stratified as low-risk at diagnosis ultimately experience early relapse [35]. Nowadays, the advent of high-throughput sequencing technologies has facilitated more precise molecular stratification, enabling the identification of patient subgroups characterized by distinct gene expression signatures. To date, more than 20 such signature systems have been reported [36], with prominent examples including MM profiler (EMC92/SYK92) [37], comprising 92 genes, and MyPRS (UAMS GEP70) [38], comprising 70 genes. Both signatures have demonstrated the ability to identify high-risk patients with poor prognosis. Therefore, given the limitations of existing staging systems and the growing emphasis on molecular profiling, we sought to evaluate whether DKC1 could enhance prognostic precision. In our study, DKC1 emerged as a potential complementary marker to the ISS staging system. Consistently, nomograms incorporating independent prognostic variables, including DKC1 expression, demonstrated good predictive accuracy for both OS and PFS. Notably, the inclusion of DKC1 enhanced the C-index and reduced the IBS, indicating improved predictive performance compared with clinical variables alone. Collectively, these results underscore the prognostic significance of DKC1 and support its utility as a reliable biomarker for refining risk stratification.

PF is a nucleoside analog initially recognized for its ability to inhibit the orotidine monophosphate decarboxylase function of uridine monophosphate synthase [39]. It has been previously used for the treatment of colorectal cancer [40], acute myeloid leukemia [41] and metastatic sarcoma [42], but demonstrated limited therapeutic efficacy. Recently, Rocchi et al. identified PF as the most potent drug inhibiting DKC1 pseudouridine synthase activity through computational screening, and further analysis of the PubChem database revealed that cell lines with high DKC1 expression (e.g., RPMI8226, DMS 273 and 786-O) were significantly more sensitive to PF than those with low expression [26]. In our study, PF effectively suppressed MM progression in vivo. Consistently, in a previous clinical trial, PF showed preliminary efficacy in MM patients, with 4 of 14 responding [43]. These results suggest that PF exert antitumor effects in MM, possibly due to the unique characteristics of malignant plasma cells, including high DKC1 expression and a pronounced reliance on protein synthesis to sustain immunoglobulin production, which may contribute to their sensitivity to PF. At the same time, DKC1 is involved in RNA pseudouridylation and telomerase RNA stabilization, and its inhibition may impact rapidly proliferating healthy tissues. Our in vivo study showed that PF treatment did not induce detectable damage to major organs (e.g., heart, liver, and kidney), suggesting a favorable preliminary safety profile. However, the potential toxicity associated with DKC1 inhibition warrants further investigation, particularly in the context of long-term or high-dose treatment. Beyond PF, in 2025, Roman et al. developed a novel DKC1 inhibitor, R1D2-10, and demonstrated that its combination with paclitaxel synergistically enhanced efficacy while reducing paclitaxel-associated toxicity in triple-negative breast cancer cells. This agent may represent a promising candidate for further investigation in upcoming research [44].

DKC1 encodes dyskerin, a core component of the H/ACA ribonucleoprotein (RNP) complex, which is involved in the pseudouridylation of functional RNAs [45]. Ψ, the most abundant natural RNA modification, can occur at up to 9% of uridine (U) sites in total RNA [46]. The U-to-Ψ conversion rearranges the chemical structure by freeing a nitrogen atom in the base and introducing an additional hydrogen bond donor [47]. Karikó K et al. demonstrated that incorporating naturally occurring Ψ into mRNA improved its biological stability, thereby enhancing translation efficiency both in vivo and in vitro [48]. This increased stability can be explained by the fact that Ψ is more thermodynamically stable than U in short RNA duplexes, partly due to its ability to coordinate a water molecule between the nucleobase and sugar-phosphate backbone, which stabilizes local RNA structure and enhances base stacking [49]. Besides, Karikó K et al. also found that Ψ-modified RNA avoids activation of human Toll-like receptor (TLR) 3, TLR7, and TLR8, as well as human primary dendritic cells, thereby reducing immune-mediated clearance [48]. Most recently, it was shown that this effect is likely due to the inability of RNase T2 and PLD exonucleases to efficiently process Ψ-modified RNA into ligands capable of activating TLRs [50]. This immune evasion may also contribute to the prolonged persistence of RNA.

In addition, our previous analysis found that high DKC1 expression in endometrial cancer is associated with an immune-cold tumor microenvironment, which is characterized by T cell exclusion [14]. In the current study, IHC revealed no significant differences in neutrophil or macrophage infiltration between the control and DKC1 inhibition groups. These observations suggest that DKC1 may modulate the tumor immune microenvironment through mechanisms independent of these innate immune cells, although the lack of observed differences may be influenced by the use of immunodeficient mice.

Previously, Ψ was thought to be predominantly present in tRNAs, rRNAs, and snRNAs [46]. Recently, progress in pseudouridine profiling methods has revealed that pseudouridylation occurs in a regulated manner in mRNAs and noncoding RNAs across yeast and humans [51], and notably, these modifications exhibit a strong preference for coding regions (CDS) in mRNAs [52–55]. During the pseudouridylation process, DKC1 serves as the catalytic subunit of the H/ACA RNP complex, while a class of noncoding RNAs directs the complex to specific target sites through base-pairing interactions [56]. However, unlike the RNA-independent pathway, where pseudouridine synthases (PUS) recognize well-defined sequence or structural motifs, RNA-guided pseudouridylation remains difficult to predict. This difficulty arises from the flexible and degenerate nature of RNA-target pairing, complicating mechanistic studies.

To further investigate the functional mechanism of DKC1 in MM, we performed RNA sequencing analyses and identified ATF5 as a promising downstream target. Notably, knockdown of DKC1 led to a significant destabilization of ATF5 mRNA, resulting in a decrease in its expression at both transcript and protein levels. Importantly, we provide the evidence of pseudouridylation modifications within the CDS of ATF5, revealing a novel layer of post-transcriptional regulation. Furthermore, our rescue experiments demonstrated that ATF5 acts as a pivotal downstream effector mediating the oncogenic functions of DKC1 in MM, underscoring its critical role in disease progression.

ATF5 is a transcription factor of the bZIP family and is classified as a member of the ATF4 subfamily, sharing sequence homology with ATF4 [57]. ATF5 promotes tumorigenesis in various cancers, such as glioma [58] and breast cancer [59]. Downregulation of ATF5 leads to apoptosis, possibly due to its direct regulatory role in anti-apoptotic proteins such as Bcl-2 [60] and Mcl-1 [61]. For Bcl-2, ATF5 primarily functions through direct binding to its promoter, thereby regulating its transcription. Besides, the regulation of Bcl-2 appears to be highly cell type-specific. Experimental evidence demonstrates its critical role in glioblastoma and breast cancer cells, whereas in untransformed astrocytes and human mammary epithelial cells, this regulation is absent [60]. Consistent with this, our results revealed that Bcl-2 protein levels markedly declined upon ATF5 depletion in MM cells, underscoring a context-dependent regulatory mechanism.

Moreover, our KEGG analysis revealed that the biosynthesis of amino acids pathway was downregulated in DKC1-knockdown cells, echoing a recent study showing that DKC1 acts through hnRNP A1 to modulate amino acid metabolism [62]. In contrast, mRNA surveillance pathway was upregulated, which is in line with previous reports demonstrating that pseudouridylation inhibits nonsense-mediated mRNA decay (NMD), prevents premature translation termination, and allows the production of full-length functional proteins [63].

DKC1 may have numerous additional targets, warranting further investigation of its downstream genes and signaling pathways regulated by DKC1, which will provide a deeper understanding of the molecular mechanisms underlying MM sensitivity to DKC1 inhibition and may identify additional targets for combination therapies.

Overall, our study provides novel insights into the role of DKC1 in MM. However, a limitation of the present work is the relatively small number of clinical cases analyzed. Future studies involving larger patient cohorts are needed to validate these findings.

Conclusion

Our study identifies elevated DKC1 expression as a potential biomarker of poor prognosis in MM and a valuable complement to the ISS staging system. Mechanistically, DKC1 drives MM progression by stabilizing ATF5 mRNA through its pseudouridine synthase activity, establishing ATF5 as a key downstream effector of DKC1’s oncogenic function (Fig. 6E). Together, these findings define the DKC1–ATF5 axis as a critical pathway in MM pathogenesis and highlight its potential as a promising therapeutic target.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

MM

Multiple myeloma

DKC1

Dyskerin pseudouridine synthase 1

Ψ

Pseudouridine

MGUS

Monoclonal gammopathy of undetermined significance

PCL

Plasma cell leukemia

ALB

Albumin

B2M

Beta2-microglobulin

ISS

International staging system

X-DC

X-linked dyskeratosis congenita

snoRNA

Small nucleolar RNAs

scaRNAs

Cajal body RNAs

MMRF

Multiple myeloma research foundation

GDC

Genomic data commons

KM

Kaplan–Meier

PBMCs

Peripheral blood mononuclear cells

RT-qPCR

Real-time quantitative polymerase chain reaction

PF

Pyrazofurin

H&E

Hematoxylin and eosin

IHC

Immunohistochemical

CMC

N-cyclohexyl-N′-β-(4-methylmorpholinium) ethylcarbodiimide p-tosylate

PBS

Phosphate-buffered saline

OS

Overall survival

ATF5

Activating transcription factor 5

PFS

Progression-free survival

CDS

Coding sequence

PUS

Pseudouridine synthases

Author contributions

C.Z. and D.X. conceived and designed the project; C.S. performed bioinformatics analysis and experiments; C.Z., Y.J. and D.K. supervised the study; L.C., Y.H., A.L., Y.W., W.Z., J.X., Y.G., W.Z., L.S. provided technical assistance; C.S. analyzed data and wrote the manuscript; C.Z.，Y.J. and D.X. wrote and/or revised the manuscript. All authors contributed to the article and approved the submitted version.

Funding

This study was supported by the Key Research and Development Program of Shandong Province (2021CXGC011101); Weihai Zhengsheng Biotechnology Foundation; Natural Science Foundation of Shandong Province (ZR2023MH341); Swedish Cancer Society (22 1989 Pj); Cancer Society in Stockholm (231402) and Karolinska Institutet (2022-01889).

Data availability

All data relevant to this study are provided in this manuscript and the supplementary information. The RNA-seq data was deposited in the GEO dataset under the accession number GSE308784. The publicly available data used in this study can be accessed from the GDC (https://portal.gdc.cancer.gov/) and GEO (https://www.ncbi.nlm.nih.gov/geo/) databases.

Declarations

Ethics approval and consent to participate

The human samples and animal experiments were approved by the Ethics Committee of the Second Qilu Hospital of Shandong University (no. KYLL2024737) and were conducted in accordance with legal requirements and national guidelines.

Consent for publication

All authors have agreed to publish this manuscript.

Competing interests

The authors declare no conflict of interest.