Inhibition of Aurora B induces senescence and potentiates immunotherapy in hepatocellular carcinoma

Abstract

Background

Hepatocellular carcinoma (HCC) is a global health challenge, with limited treatment options for advanced-stage patients. Although recent approvals of targeted therapies and immune checkpoint inhibitors (ICIs) have expanded the therapeutic landscape, their clinical benefits are often constrained by modest response rates and acquired resistance. Despite ongoing efforts to identify new therapeutic targets, only few inhibitors have progressed to clinical trials, benefiting a small subset of patients. These limitations underscore the need for more effective therapeutic strategies.

Methods

We integrated data from DepMap and TCGA to identify HCC-specific vulnerabilities. A kinome-wide CRISPR screen validated essential targets. Pharmacological inhibition (AZD1152) and genetic knockout of Aurora B (AURKB) were employed for functional validation. Senescence was assessed via SA-β-gal staining, transcriptomics, and SASP analysis. Immune interactions were evaluated using co-culture assays with T cells and flow cytometry for MHC I and immune markers. In vivo efficacy was tested in immunocompetent murine HCC models.

Results

Aurora B was identified as a top candidate essential for HCC survival, overexpressed in tumors, and correlated with poor prognosis. A kinome-wide CRISPR screen further confirmed its essential role in HCC cell survival. Pharmacologic inhibition or genetic knockout of Aurora B markedly suppressed proliferation of HCC cells and induced robust cellular senescence, characterized by cell-cycle arrest, DNA damage, and SASP. These senescent cells exhibited heightened susceptibility to T-cell-mediated cytotoxicity, associated with increased immunoproteasome activity and upregulated MHC I expression. Co-culture assays revealed that senescent tumor cells promoted interferon-γ production and enhanced cytotoxicity in CD8⁺ T cells as well as PD-1 expression. In both in vitro and in vivo models, combining Aurora B inhibition with anti-PD-1 therapy markedly suppressed tumor growth.

Conclusion

This study identifies Aurora B as a therapeutically actionable vulnerability in HCC, whose inhibition induces senescence and enhances tumor immunogenicity. The combination of Aurora B inhibitor with anti-PD-1 therapy yields synergistic anti-tumor effects, supporting its potential as a promising strategy to improve immunotherapy outcomes in HCC.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13402-026-01225-2.

Introduction

Hepatocellular carcinoma (HCC) is a leading cause of cancer-related mortality globally [1], with systemic therapy options for advanced-stage patients remaining limited [2]. Advances in early diagnosis, including imaging refinement and biomarker-based surveillance strategies, as well as evolving staging systems, have improved patient stratification and treatment selection [3, 4]. For over a decade, receptor tyrosine kinase (RTK) inhibitors such as sorafenib [5] and lenvatinib [6], were the only first-line treatments. In 2020, the IMbrave150 phase III trial revolutionized HCC therapy by demonstrating that the combination of anti-PD-L1 atezolizumab and anti-VEGF bevacizumab outperforms sorafenib [7, 8]. This led to the rapid approval of this therapy by the United States Food and Drug Administration (FDA) and other health authorities in over 70 countries [9]. Likewise, the anti-PD-1 sintilimab plus bevacizumab biosimilar (IBI305) combination showed significant survival benefits in HBV-related advanced HCC compared to sorafenib [10]. In 2022, the FDA approved another breakthrough regimen: the dual immune checkpoint blockade of anti-PD-L1 durvalumab and anti-CTLA4 tremelimumab, which demonstrated 30.7% 36-month overall survival (OS) and 25.2% 48-month OS in advanced HCC patients in the phase III HIMALAYA trial [11]. Additional emerging systemic therapies and combination strategies, including novel immunotherapy-based regimens and targeted agents, continue to expand the therapeutic landscape [12, 13]. Despite these advances, most therapies still offer only modest survival benefits, highlighting the urgent need to identify novel therapeutic strategies and more effective drug combinations for HCC treatment.

With the advancement of next-generation sequencing, numerous driver mutations involved in hepatocarcinogenesis and progression have been identified, such as mutations in the TERT promoter, TP53, CTNNB1, ARID1A and AXIN1 [14]. However, most of these genetic changes in HCC remain undruggable. Fibroblast growth factor 19 (FGF19) receptor FGFR4 and c-Met are the only molecular targets that have progressed to clinical evaluation in HCC. FGFR4 inhibitor fisogatinib (BLU-554) achieved an objective response rate (ORR) of only 17% in FGF19-positive patients [15], modestly improved to 50% in combination with anti-PD-L1 therapy [16]. c-Met inhibitor tivantinib failed to improve survival in MET-high patients [17].

Recent functional genetic screens have revealed key regulators critical to HCC biology, presenting opportunities for novel therapies or enhancing standard treatments [18]. Genetically engineered immunocompetent in vivo and in vitro models identified cladribine as an effective agent when combined with Lenvatinib [19]. EGFR inhibition also boosted cabozantinib efficacy [20]. Ferroptosis, triggered by MALT1 inhibition and GPX4 destabilization, synergized with sorafenib or regorafenib to suppress HCC progression [21]. Additionally, tumor microenvironment (TME) factors, including immune cell infiltration and stromal interactions, influence tumor progression and treatment responses [9]. Despite advances in HCC treatment, durable responses remain limited, underscoring the need for novel targets.

Emerging evidence suggests that dysregulation of the cell cycle and induction of cellular senescence can actively remodel the immune microenvironment by promoting the secretion of senescence-associated secretory phenotype (SASP) factors, which modulate immune cell recruitment and function [22]. Given that mitotic kinases are central regulators of cell cycle progression and chromosomal stability, and that their dysregulation can influence both tumor cell proliferation and senescence-associated immune signaling, targeting such pathways may provide a dual opportunity to inhibit tumor growth while modulating the immune microenvironment.

In this study, we performed an unbiased, integrative analysis based on gene expression, patient prognosis, and cancer cell dependency to identify novel therapeutic targets in HCC, and subsequently identified Aurora B kinase as a candidate for further investigation. Given its central role in mitotic regulation, Aurora B provides a mechanistic link between cell cycle-associated senescence and modulation of the tumor immune microenvironment. Our findings highlight Aurora B inhibition as a potential strategy to enhance treatment efficacy and nominate promising candidates for further translational development.

Materials and methods

Cell culture

The human cell lines, Hep3B, Huh7, SK-Hep1, Huh6, SNU449, PLC/PRF/5, SNU398, and HEK293 were obtained from American Type Culture Collection (ATCC). MHCC97H and HCCLM3 cell lines were provided by the Liver Cancer Institute of Zhongshan Hospital. The mouse cell line, Hep53.4 was provided by the CLS-Cell Lines Service (Eppelheim, Germany), Myc^OE/Pten^KO was obtained from the Netherlands Cancer Institute (NKI) [23]. These cell lines were cultured in a controlled environment at 37 °C with 5% CO₂ in Dulbecco’s modified Eagle medium supplemented with 10% fetal bovine serum (FBS), 2 mmol/L glutamine, and 1% penicillin/streptomycin (Gibco). All the cell lines were authenticated using short tandem repeat profiling and regularly subjected to polymerase chain reaction-based assays to test for the presence of mycoplasma.

Compounds and antibodies

AZD1152 (S1147) was purchased from Selleck Chemicals. Antibody against HSP90 alpha/beta (sc-13119) was purchased from Santa Cruz. Antibody against Lamin B1 (12987) was purchased from Proteintech. Antibody against γH2AX (ab81299) was purchased from Abcam. Antibodies against Aurora B (3094T), PSMB9/LMP2 (87667T), PSMB8/LMP7 (13635T) were purchased from Cell Signaling Technology.

Gene dependency

Gene dependency analysis was performed using data from the DepMap Portal. Genome-wide CRISPR-Cas9 loss-of-function screening data were downloaded from DepMap, and Chronos gene effect scores (or CERES scores, if applicable) were extracted for liver cancer cell lines. Gene effect scores were used to quantify the impact of gene knockout on cellular fitness, with lower scores indicating stronger dependency. In the DepMap framework, a score of 0 represents the median effect of non-essential genes, whereas a score of approximately − 1 corresponds to the median effect of common essential genes. Liver cancer cell lines were selected based on lineage annotation, and dependency scores for the gene of interest were compiled for visualization. In the plot, each dot represents the dependency score of the target gene in an individual liver cancer cell line, and the lines indicate the dependency trends of reference genes across the same cell lines.

Public cohorts and data processing

The Cancer Genome Atlas Liver Hepatocellular Carcinoma (TCGA-LIHC) cohort included 373 HCC cases with transcriptomic and clinical data. Gene expression data (raw counts) were obtained from the Genomic Data Commons portal (https://portal.gdc.cancer.gov/), and survival information was retrieved from the TCGA Pan-Cancer Clinical Data Resource (TCGA-CDR). Cases with available AURKB expression data and corresponding survival information were included in the subsequent analyses. Raw counts were similarly converted into TPM values.

The Liver Cancer-RIKEN, JP project (LIRI-JP) cohort included 231 hepatocellular carcinoma (HCC) cases from a Japanese population predominantly infected with HBV/HCV. Gene expression data (raw counts) and corresponding clinical information were downloaded from the International Cancer Genome Consortium (ICGC) data portal (https://dcc.icgc.org/projects/LIRI-JP). Raw count data were converted to transcripts per million (TPM) values for downstream analyses.

AURKB expression and survival analyses

AURKB expression levels were transformed into logTPM values for visualization and statistical analysis. Differences in AURKB expression between normal and tumor tissues were compared in the indicated cohorts. For overall survival analysis, patients in each cohort were dichotomized into AURKB-high and AURKB-low groups using the optimal expression cutoff determined separately for each dataset. Kaplan-Meier curves were plotted and compared using the log-rank test. Hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated using Cox proportional hazards regression.

Cell proliferation assays

For long-term cell proliferation assay, cells were cultured and seeded into 6-well plates at a density of 1.5 to 4 × 10⁴ cells per well, adjusted according to the growth rate of each cell line. The cells were then cultured in medium supplemented with the specified drugs for a duration of 10 to 14 days. Medium was changed twice a week. After the incubation period, the cells were fixed with 4% formaldehyde in phosphate-buffered saline (PBS) and subsequently stained with a 0.1% crystal violet solution diluted in water. Images were acquired using an ImageScanner III (GE Healthcare) at a resolution of 300 dots/inch. For cell viability assay, cells were seeded into 96-well plates at a density of 2 × 10³ cells per well. Cell viability was detected using CellTiter-Blue assay (Promega, G8081) according to the manufacturer’s recommendations.

CRISPR guide RNA generation

Oligonucleotides containing guide RNA sequences of AURKB-1 (ATTCTAGAGTATGCCCCCCG) and AURKB-2 (CATCAACCCATACTGCAGGT) flanked by 20 to 30 nucleotides of overlapping backbone sequence were obtained from Tsingke Biotech. Guide RNA sequences were cloned into LentiCRISPRv2.1 using BsmBI sites, using the Gateway cloning strategy.

Protein lysate preparation and western blots

Cells were washed with PBS and lysed with RIPA buffer supplemented with Complete Protease Inhibitor (Roche) and Phosphatase Inhibitor Cocktails II and III (Sigma). Protein quantification was performed with the BCA Protein Assay Kit (ThermoFisher Scientific). All lysates were freshly prepared and processed with BioRad Gel Electrophoresis Systems.

Senescence-associated β-galactosidase staining

Indicated cell lines were seeded into 6-well plates at a density of 5-10 × 10⁴ cells per well, depending on growth rate and the design of the experiment. About 24 h later, AZD1152 were added at the indicated concentrations. β-Galactosidase activity in the cells was assessed using the Histochemical Staining kit (CS0030-1KT) obtained from Sigma-Aldrich. The detection of β-galactosidase was performed following the manufacturer’s instructions.

Time-lapse live imaging

To allow visualization of chromosomes, cells were transduced with a histone H2B-GFP (LV-GFP, Addgene plas mid#25999). Cells were then plated 24 h before starting the microscope acquisition. DMSO or AZD1152 (0.5 µM) were added in the medium 1 h before starting the movie. Cells were filmed over 96 h and pictures were taken every 8 min. For each condition filmed, 5 different fields were selected. In each field, we randomly chose and followed cells entering in mitosis.

Cell cycle analyses

Cells were plated (1 × 10⁵ cells per plate) on 10 cm plates and treated with DMSO or AZD1152 (0.5 µM) for 2, 4, 6, 12 h. Cells were incubated in 0.5 mL PBS containing 50 mg/mL propidium iodine and 100 mg/mL RNase and incubated at 4 °C for 30 min. Samples were then analyzed on the BD LSRFortessa Analyzer, and data were analyzed with the FlowJo software (BD) after excluding doublet cells.

RNA isolation and real-time quantitative reverse-transcription PCR (RT-qPCR)

RNA was extracted from cells using the FastPure Cell/Tissue Total RNA Isolation Kit V2 (Vazyme, RC112-01) according to the manufacturer’s instructions [24]. The concentration of RNA was determined with a Thermo Scientific NanoDrop One (ThermoFisher Scientific). The RNA was subsequently reverse transcribed into cDNA using the Color Reverse Transcription Kit with gDNA remover (EZBioscience, A0010CGQ). RT-qPCR was performed using the 2× Color SYBR Green qPCR Master Mix (EZBioscience, A0012-R2) on a QuantStudio™ 7 Flex Real-Time PCR machine (Applied Biosystems^®). The 2 − ΔΔCt method was used to quantify gene expression levels. The relative expression levels of genes were obtained through sequential normalization of the values to those of the experimental controls.

In vivo tumor models and treatment protocols

Huh7 and PLC/PRF/5 cells (1 × 10⁷ cells per mouse) were subcutaneously injected into the right posterior flanks of 6-week-old BALB/c nude mice (male, 10 mice per group). Hep53.4 and Myc^OE/Pten^KO cells (1 × 10⁷ cells per mouse) were subcutaneously injected into the right posterior flanks of 6-week-old C57BL/6J mice (male, 10 mice per group). After tumor establishment, the mice were randomly assigned to different treatment groups: vehicle, AZD1152 (50 mg/kg, intraperitoneal injection, 3 times per week), PD-1 antibody (50 µg/mouse, intraperitoneal injection, twice per week), or a drug combination in which each compound was administered at the same dose and schedule as the single agent. Tumor volume was calculated using the modified ellipsoidal formula: tumor volume = 1/2 x length x width², based on caliper measurements. The data are presented as the mean ± standard error of the mean.

Ethics declaration

All animal experiments were conducted in accordance with institutional guidelines and were approved by the Animal Welfare and Ethics Committee, Renji Hospital, Shanghai Jiao Tong University School of Medicine under protocol number RJ2025-142B.

Immunohistochemistry

Formalin-fixed, paraffin-embedded samples were obtained from xenograft and syngeneic tumors and stained with hematoxylin and eosin, antibodies against PCNA (Abcam, ab2426), SA-β-gal (Cell Signaling Technology, 9449), PanCK (Proteintech, 26411-1-AP), CD8 (Abcam, ab217344). After incubation with the primary antibodies, positive cells were visualized using diaminobenzidine as a chromogen.

RNA sequencing

RNA was isolated from cell lines with different experimental conditions using TRIzol (Invitrogen), and complementary DNA libraries were subsequently prepared and sequenced on an Illumina HiSeq2500 platform, generating 65-base pair single-end sequence reads. The obtained reads were aligned to the GRCh38 human reference genome. GSEA was used as previously described to perform gene set enrichment analysis. The FRIDMAN_SENESCENCE_UP gene set was used to evaluate the enrichment of senescence associated genes in the treatment of AZD1152. The SAUL_SEN_MAYO gene set was used to evaluate the cytokine expression. The HALLMARK gene set was used to assess the enrichment of interferon gamma response sequentially treated with AZD1152. The Enrichment scores were corrected for gene set size (normalized enrichment score). The P value estimates the statistical significance of the enrichment score for a single gene set as described previously.

Immunofluorescence and image analysis

For immunofluorescence microscopy, cells were seeded on glass coverslips and cultured in the presence of 0.5 µM AZD1152 for 3 days. Cells were fixed in 2% paraformaldehyde for 20 min; permeabilized with 0.2% Triton X-100 for 5 min; blocked with PBS containing 2% bovine serum albumin (Sigma-Aldrich) for 45 min; and subsequently incubated with H3K9Me3 antibody (Thermo Fisher Scientific, 49-1008) and goat anti-rabbit Alexa Fluor 546 (Invitrogen, 1:200) for 1 h, respectively. Nuclei were stained with 4,6-diamidino-2-phenylindole. Samples were mounted on glass slides in Mowiol (Sigma-Aldrich) after 3 washing steps with PBS. Images were acquired with a Leica TCS SP5 confocal microscope with a 63×/1.4 oil objective. Image processing was performed using ImageJ (National Institutes of Health) software.

Flow cytometry

For immune cell profiling analysis, syngeneic tumor tissues were perfused with PBS and then dissociated into single-cell suspension using PercollTM (Cytiva) and the gentleMACS Octo Dissociator, following the manufacturer’s instructions. The cell suspension was passed through a 100-µm cell strainer (Corning) and then centrifuged at 300 g for 10 min at 4 °C and washed 3 times in FACS buffer. Cells were then incubated with the specified antibodies on ice for 30 min, and then washed with FACS buffer. The signal was detected by using LSRFortessa X-20 flow cytometer (BD). Analyses were carried out using FlowJo software. For HCC and mouse cell lines, cells were harvested and washed 3 times in FACS buffer, then incubated with the specified antibodies on ice for 30 min. For IFN-γ staining, OVA-transfected HCC cells were pretreated with AZD1152 or DMSO control for 48 h and then washed extensively to remove residual drug. Activated OT-I CD8⁺ T cells were subsequently added and co-cultured with the pretreated tumor cells for 4 h in the presence of GolgiStop and GolgiPlug (BD) before harvesting. Cells were then collected, stained with viability dye and surface antibodies, fixed and permeabilized, and subjected to intracellular staining with anti-IFN-γ antibody according to the manufacturer’s instructions. The antibodies utilized for this flow cytometry analysis included: anti-mouse CD45 PE-Cy7 (BD Biosciences, 552848), anti-mouse CD3e FITC (BD Biosciences, 553061), anti-mouse CD8a BV421 (BD Biosciences, 563898), anti-mouse CD4 FITC (BD Pharmingen, 557307), anti-mouse NK-1.1 APC (BD Biosciences, 550627), anti-mouse CD279/PD-1 PE (BD Biosciences, 551892), anti-CD11b FITC (BD Biosciences, 557396), anti-Mouse F4/80 BV421 (BD Biosciences, 565411), anti-mouse Ly-6 C APC (BioLegend, 128016), anti-mouse IFN-γ FITC (BioLegend, 505806), anti-mouse H-2Kb FITC (BioLegend, 114606), anti-Human CD3 RB705 (BD Pharmingen, 757064), anti-human IFN-γ PE-Cy (BD Pharmingen, 557643), anti-human CD279 APC (BD Pharmingen, 558694), anti-human HLA-ABC FITC (BioLegend, 311404), LIVE/DEAD Fixable Near-IR Dead Cell Stain Kit (Thermofidher, L10119).

In vitro killing assays

Activated OT-I T cells were cocultured with OVA transfected cells in 96-well plates at E: T ratios of 1:1. The killing effects were detected 48 h after coculture using the CellTiter-Blue cell viability assay (Promega, G8081), following the manufacturer’s instructions.

Data availability

The raw sequencing data generated in this study have been deposited in the Genome Sequence Archive (GSA) database.

Statistical analysis

All statistical analyses were performed using GraphPad Prism software (GraphPad). All in vitro data are reported as the mean ± SD, and all in vivo data are presented as mean ± SEM. Statistical significance was determined using an unpaired two-tailed Student’s t-test. Significance was assumed with *p < 0.05; **p < 0.01; ***p < 0.001.

Results

Aurora B is a novel therapeutic target for hepatocellular carcinoma treatment

HCC is an aggressive malignancy with few effective therapeutic options and dismal clinical outcomes. Although recent advances in systemic therapies, such as tyrosine kinase inhibitors and immune checkpoint inhibitors, have expanded treatment strategies, their overall efficacy remains limited. Identifying novel molecular targets that can suppress HCC progression and enhance patient survival is urgently needed.

To explore novel molecular targets of HCC, we established a four-step pipeline to identify novel therapeutic targets for HCC (Fig. 1A). Using data from the Dependency Map (DepMap) and The Cancer Genome Atlas (TCGA), we systematically analyzed genes that are essential for liver cancer cell survival, overexpressed in tumor tissues, and associated with poor patient prognosis, while also assessing their targetability and clinical relevance. Through this integrative approach, we obtained 1,021 essential genes for liver cancer cells in step 1 (Fig. 1B), 945 genes overexpressed in tumor tissues in step 2 (Fig. 1C), and 1,279 genes associated with poor prognosis in step 3 (Fig. 1D). By overlapping these datasets, we obtained 26 candidate genes for further evaluation of therapeutic potential (Fig. 1E). Among these, eight genes have reported specific inhibitors, six of which have been tested in clinical trials (AURKB, TOP2A, TTK, PLK1, CDK1, and BIRC5) (Fig. 1F). Testing these inhibitors in eight HCC cell lines and three non-cancerous cell lines revealed that pharmacological inhibition of these targets significantly suppressed HCC cell proliferation (Fig. 1G). Given our focus on translational applicability, we prioritized these druggable targets for experimental validation. This prioritization strategy may exclude other biologically relevant candidates lacking clinically available inhibitors, which warrant further investigation in future studies.

Fig. 1.

Discovery of novel therapeutic targets for hepatocellular carcinoma. (A) Workflow for identifying novel therapeutic targets for HCC. (B) Gene dependency analysis across 22 liver cancer cell lines. (C) Comparison of mRNA expression between tumor and normal liver tissues. (D) Analysis of the association between gene expression and prognosis in HCC. (E) Overlap of genes that are essential for liver cancer cell survival, exhibit elevated expression in tumors, and correlate with poor prognosis. 26 candidate genes are accord with this criterial. (F) Among these 26 genes, 8 genes have specific inhibitors, 6 of which have been evaluated in clinical trials. (G) Screening of kinase inhibitors in cancer versus non-cancer cell lines

We further validate the functional relevance of candidate targets in a kinome-wide CRISPR-Cas9 dropout screen in Hep3B and Huh7 cells (Fig. 2A). Besides the extensively studied target PLK1 [25–27] and CDK1 [28–30] in HCC, AURKB exhibited consistent depletion across both cell lines (Fig. 2B). In our four-step pipeline analysis, AURKB showed as an essential dependency across all liver cancer cell lines (Fig. 2C). Its mRNA expression was markedly elevated in tumor tissues compared to adjacent non-tumor tissues (Fig. 2D), and high AURKB levels were significantly associated with poor prognosis in HCC patients (Fig. 2E). Pharmacological inhibition of Aurora B with AZD1152 potently suppressed HCC cell proliferation (Fig. 2F), and demonstrated better selectivity towards cancerous cells (Fig. 1G). Consistently, genetic depletion of AURKB also impaired cell growth (Fig. 2G-H). In vivo, AZD1152 treatment significantly reduced tumor growth in subcutaneous xenograft models of Huh7 and PLC/PRF/5 cells (Fig. 2I). Collectively, these findings identify Aurora B kinase as a key vulnerability in HCC.

Fig. 2.

Aurora B presented as potential therapeutic target for hepatocellular carcinoma treatment. (A) Schematic of kinome-wide CRISPR-Cas9 screen performed in Huh7 and Hep3B cells. (B) Dropout screening identifies AURKB as a critical dependency for proliferation in both cell lines. (C) Dependency score of AURKB across 22 liver cancer cell lines. (D) mRNA level of AURKB in tumor versus normal tissues. (E) Prognostic analysis of AURKB expression in HCC patients. (F) Colony formation assays of liver cancer cell lines treated with the Aurora B inhibitor AZD1152. (G) Hep3B, Huh7 and PLC/PRF/5 cells were transfected with control or 2 independent small interfering RNAs targeting AURKB, the efficiency of AURKB knockout in liver cancer cell lines was evaluated by Western blot. (H) Colony formation assays of AURKB knockout cell lines. (I) Tumor growth curves of Huh7 and PLC/PRF/5 xenografts in BALB/c nude mice treated with vehicle or AZD1152 (50 mg/kg). Tumor volumes measured every 3 days. *P < 0.05, ****P < 0.001

Inhibition of Aurora B induces senescence in liver cancer cells

Aurora B is a key regulator of mitosis, and its inhibition disrupts proper chromosome segregation and cell division [31]. Consistent with this role, Aurora B inhibition induced mitotic arrest, as demonstrated by live-cell imaging of Hep3B and Huh7 cells stably expressing GFP-tagged histone H2B, which showed a marked prolongation of mitotic duration (Fig. 3A). In line with this observation, flow cytometry analysis revealed a pronounced accumulation of cells with 4n DNA content in Hep3B, Huh7, and PLC/PRF/5 cells following treatment (Fig. 3B, S1A-B), consistent with G2/M-phase accumulation and/or the formation of tetraploid cells due to Aurora B inhibition [32]. Transcriptomic analysis revealed significant enrichment of senescence-associated signaling pathways (Fig. 3C), suggesting activation of a senescence program following Aurora B inhibition. Cellular senescence is a stable form of cell cycle arrest characterized by distinct features, including increased β-galactosidase activity, chromatin remodeling, DNA damage accumulation, lamin B1 reduction, and the secretion of senescence-associated secretory phenotype (SASP) factors [33–37]. Consistent with this, pharmacological or genetic inhibition of Aurora B induced robust senescence phenotypes. We observed increased β-galactosidase activity, as confirmed by SA-β-gal staining (Fig. 3D, S1C), along with the formation of senescence-associated heterochromatin foci (SAHF), indicating chromatin reorganization (Fig. 3E, S1D-E). In addition, Aurora B inhibition led to DNA damage and reduced lamin B1 levels (Fig. 3F, S1F), accompanied by activation of the SASP, as evidenced by elevated expression of SASP factors (Fig. 3G). We observed that RB, a key regulator of senescence-associated cell cycle arrest [38, 39], was upregulated following AZD1152 treatment (Fig. S1G). RB knockout attenuated Aurora B inhibition-induced senescence phenotypes, including decreased SA-β-galactosidase activity (Fig. 3H) and DNA damage (Fig. S1G), indicating that Aurora B inhibition-induced senescence is mediated by RB. In vivo, AZD1152-treated tumors exhibited marked proliferation arrest and strong SA-β-gal positivity, confirming robust senescence induction (Fig. 3I-K). Collectively, these results demonstrate that Aurora B inhibition suppresses HCC cell proliferation and induces a stable senescence program both in vitro and in vivo.

Fig. 3.

Inhibition of Aurora B induced senescence in liver cancer cells. (A) Representative live cell images of Hep3B and Huh7 cells expressing GFP-Histone 2B in the presence or absence of AZD1152 monitored by time-lapse microscopy. (B) Cell cycle distribution of AZD1152-treated liver cancer cells analyzed by flow cytometry. Representative FACS plots of Huh7 cells (upper panel) and quantification of cell cycle distribution in Hep3B, Huh7, and PLC/PRF/5 cells (lower panel). (C) GSEA showing upregulation of senescence-associated genes in AZD1152-treated cells. (D) Senescence induction following AZD1152 treatment detected by SA-β-gal staining (scale bars, 50 μM). (E) Representative images of H3K9me3 staining in Hep3B and Huh7 cells exposed to AZD1152 for 3 days (scale bar, 10 μM). (F) Western blot analysis of lamin B1 and γH2AX in AZD1152-treated cells. (G) GSEA showing SASP genes expression in AZD1152-treated cells. (H) Senescence induction following AZD1152 treatment detected by SA-β-gal staining in RB wild type or deficient cells (scale bars, 50 μM). (I-K) Representative H&E, PCNA, and SA-β-gal staining of formalin-fixed sections from subcutaneous tumors (scale bars, 100 μM)

Aurora B inhibition enhances T cell-mediated clearance of HCC cells by promoting MHC I upregulation

AZD1152 exhibited an acceptable safety profile and achieved a clinical response rate (CR+CRi+PR) of 23% in patients with advanced acute myeloid leukemia [40], whereas clinical trials in solid tumors reported predominantly stable disease without objective responses [41]. This discrepancy may reflect the more complex tumor microenvironment and the barriers to effective immune cell infiltration characteristic of solid malignancies. In current study, Aurora B inhibitor AZD1152 effectively suppressed HCC cell proliferation in vitro and inhibited tumor growth in nude mice, exhibiting potent anti-tumor activity. Since Aurora B inhibition induces cellular senescence, its potential to reshape the tumor microenvironment-particularly in the context of immune modulation-deserves further investigation.

Transcriptomic analysis revealed significant upregulation of interferon gamma (IFNγ) response genes (Fig. S2A), including increased expression of human leukocyte antigen (HLA) class I molecules (Fig. S2B) in Aurora B inhibited cells. Consistent with these findings, protein-level analyses confirmed that Aurora B inhibition enhanced MHC I expression in both human (Fig. S2C-D) and mouse HCC cells (Fig. 4A), in which Aurora B inhibition also suppressed proliferation by inducing senescence and DNA damage (Fig. S2E-G), suggesting enhanced immune recognition. Given the immunogenic nature of senescent cells and their role in promoting antitumor immunity [42, 43], we induced senescence in OVA-transfected mouse HCC cells and co-cultured senescent and non-senescent cells with OT-1 T cells. Analysis of T cell responses showed enhanced CD8⁺ T cell activation, as evidenced by increased IFNγ production when co-cultured with senescent HCC cells (Fig. 4B). Although AZD1152 treatment reduced baseline tumor cell viability, senescent cells exhibited greater susceptibility to T cell-mediated killing, as indicated by a larger reduction in viability upon co-culture (Fig. 4C). To determine whether cellular senescence mediates the link between Aurora B inhibition and enhanced immunogenicity, we assessed the effect of RB deficiency and found that it abrogated the upregulation of MHC molecules induced by Aurora B inhibition (Fig. S2H), indicating that activation of the senescence program is required for this effect. Blocking surface MHC I or IFNγ attenuated T cell-mediated killing in AZD-treated cells, indicating that the enhanced clearance of senescent cells remains dependent on MHC I- and IFNγ-mediated mechanisms (Fig. 4D-E).

Fig. 4.

Inhibition of Aurora B facilitated T cell-mediated killing of HCC cells by inducing MHC I up-regulation through boosting immunoproteasome. (A) Flow cytometric analysis of MHC I surface expression. (B) Flow cytometric detection of IFNγ in AZD1152-treated cells. (C) T cell-mediated cytotoxicity assay in mouse liver cancer cells treated with AZD1152. (D) T cell-mediated cytotoxicity assay in AZD1152 treated liver cancer cells with MHC I blockade. (E) T cell-mediated cytotoxicity assay in AZD1152 treated liver cancer cells with IFNγ blockade. (F) Western blot analysis of immunoproteasome proteins after AZD1152 treatment. (G) Flow cytometric analysis of MHC I surface expression in cells treated with AZD1152, ONX0914, or their combination. (H) T cell-mediated cytotoxicity assay in liver cancer cells treated with AZD1152, ONX0914, or their combination. *P < 0.05, **P < 0.01 and ***P < 0.001

IFNγ is known to induce expression of the immunoproteasome subunits LMP2, LMP7, and MECL-1, thereby enhancing MHC I presentation and promoting cytotoxic T lymphocyte recognition [44–47]. Hep3B and Huh7 cells typically exhibit low HLA-I surface expression due to deficient expression of low molecular weight proteins (LMP2, LMP7) and TAP1 [45]. In senescent cells, several components of the immunoproteasome complex were notably upregulated (Fig. 4F, S3A). Pharmacologic inhibition of the immunoproteasome using ONX-0914 abrogated AZD1152-induced MHC I upregulation (Fig. 4G, S3B) and also partially abolished CD8⁺ T cell-mediated clearance of senescent cells (Fig. 4H). Collectively, these findings demonstrate that Aurora B inhibition enhances immunogenicity of HCC cells by promoting immunoproteasome-mediated MHC I upregulation, which facilitates T cell activation and tumor cell clearance. Disruption of antigen presentation machinery reverses this immune visibility, underscoring its critical role in senescent cancer cell elimination.

Combination of AZD1152 and anti-PD-1 suppresses tumor growth by enhancing T cell activation and function in vivo

Senescent cells secrete SASP factors [48], which can either support immune clearance or promote immune evasion and tumor progression, depending on the context [49–53]. In our system, senescent HCC cells not only augmented CD8⁺ T cell IFNγ production but also induced PD-1 upregulation (Fig. 5A). Based on our observation that Aurora B inhibition induces tumor cell senescence and upregulates MHC class I expression in vitro, we hypothesized that combining Aurora B inhibition with PD-1 blockade may synergistically enhance antitumor immunity. This therapeutic potential was evaluated in both inflamed and non-inflamed tumor models, in which the Aurora B inhibitor AZD1152 was well tolerated, with no obvious adverse effects observed (Fig. S4A-B). In the inflamed Hep53.4 model, combining the Aurora B inhibitor AZD1152 with anti-PD-1 therapy resulted in robust tumor growth suppression compared to either monotherapy (Fig. 5B). While AZD1152 induced senescence in tumor cells, only the combination treatment achieved pronounced tumor growth arrest (Fig. 5B-C). Analysis of the tumor microenvironment revealed that AZD1152 monotherapy had minimal impact on CD8⁺ T cell infiltration, whereas the combination therapy significantly increased CD8⁺ T cell infiltration (Fig. 5D). Moreover, AZD1152 treatment enhanced CD8⁺ T cell activation, as indicated by elevated CD107a and IFNγ expression (Fig. 5E-F), without significantly affecting CD8⁺ T cell senescence or infiltration in treated tumors (Fig. 5G). In the non-inflamed Myc^OE/Pten^KO model, only the combination of AZD1152 and anti-PD-1 led to significant tumor suppression, though the effect was less pronounced than in the inflamed model (Fig. 5H). Both AZD1152 monotherapy and the combination induced tumor cell senescence and upregulated MHC I (Fig. 5I-J). Regarding immune cell infiltration, the combination therapy promoted CD3⁺ T cell infiltration but did not significantly alter the proportions of CD8⁺ T cells (Fig. 5K, S4C). Functional assessment showed that CD8⁺ T cell activation, measured by CD107a and IFNγ expression, was enhanced by the combination treatment (Fig. 5L-M). Likewise, treated tumors showed no obvious CD8⁺ T cell senescence or diminished infiltration (Fig. 5N). Collectively, these findings indicate that Aurora B inhibition primes tumor cells for immune recognition through senescence induction and MHC I upregulation, thereby enhancing T-cell-mediated antitumor immunity. The therapeutic benefit of combining Aurora B inhibition with PD-1 blockade is primarily driven by augmented T-cell activation.

Fig. 5.

Combination of AZD1152 and anti-PD-1 therapy suppresses tumor growth by enhancing T cell activation and function in vivo. (A) Flow cytometric analysis of PD-1 surface expression on T cells. (B) Hep53.4 subcutaneous tumor-bearing mice were treated with vehicle, AZD1152 (50 mg/kg), anti-PD-1 (50 μg/mouse), or the combination. Tumor volumes were measured every 3 days. (C) Representative images of H&E, PCNA, and SA-β-gal staining of formalin-fixed Hep53.4 tumor sections (Scale bars, 100 μM). (D) Flow cytometry analysis of tumor-infiltrating CD8⁺ T cells within the CD3⁺ T cell population. (E-F) Functional analysis of CD8⁺ T cells by CD107a and IFNγ staining. (G) Representative images of PanCK, CD8, and SA-β-gal staining of formalin-fixed Hep53.4 tumor sections (Scale bars, 10 μM). (H) Myc^OE/Pten^KO subcutaneous tumor-bearing mice were treated with vehicle, AZD1152 (50 mg/kg), anti-PD-1 (50 μg/mouse), or the combination. Tumor volumes were measured every 3 days. (I) Representative images of H&E, PCNA, and SA-β-gal staining of formalin-fixed Myc^OE/Pten^KO tumor sections (Scale bars, 100 μM). (J) Flow cytometry analysis of MHC I expression. (K) Flow cytometry analysis of CD8⁺ T cell infiltration within CD3⁺ T cells. (L-M) Functional assessment of CD8⁺ T cells via CD107a and IFNγ staining. (N) Representative images of PanCK, CD8, and SA-β-gal staining of formalin-fixed Myc^OE/Pten^KO tumor sections (Scale bars, 10 μM). *P < 0.05, **P < 0.01 and ***P < 0.001

Discussion

This study identifies Aurora B kinase as a promising therapeutic target in HCC, demonstrating that its inhibition not only suppresses tumor growth by inducing cellular senescence but also enhances anti-tumor immunity by upregulating MHC class I antigen presentation and promoting CD8⁺ T cell infiltration and activation. Combining Aurora B inhibition with anti-PD-1 therapy achieves superior tumor control by synergistically enhancing T cell-mediated clearance of senescent tumor cells, providing a rationale for clinical evaluation of this combinatorial strategy in HCC treatment. Recent studies further emphasize that tumor-intrinsic stress responses, including therapy-induced senescence, can reshape immune surveillance by altering antigen presentation, cytokine signaling, and immune cell recruitment, thereby linking tumor biology directly to immunotherapy responsiveness [54, 55].

Senescence is a cellular response to various stress signals, characterized by stable withdrawal from the cell cycle and significant changes in cell morphology and physiology [33]. This study demonstrates that inhibition of Aurora B kinase suppresses HCC growth by halting mitosis and inducing senescence, consistent with previous research highlighting its role in reducing cell proliferation, epithelial-mesenchymal transition (EMT), and metastasis [56–61]. These findings further support the relevance of targeting Aurora B in liver cancer as a therapeutic strategy. Unlike apoptotic cells, which are irreversibly eliminated, senescent cells remain metabolically active, influencing their microenvironment [62]. Emerging evidence indicates that senescent tumor cells actively communicate with surrounding stromal and immune compartments through SASP factors, which can exert both immune-stimulatory and immune-suppressive effects depending on context [63]. Under certain conditions, such as in B cell lymphoma, senescent tumor cells can bypass this growth arrest, resume proliferation, and drive tumor progression through mechanisms like WNT signaling activation [64]. Thus, rather than serving solely as a terminal state, senescence may act as a transitional phase, allowing cells to acquire increased adaptability and resilience, contributing to tumor heterogeneity and resistance to therapy [53]. While therapies such as CDK4/6 or Aurora B inhibitors induce senescence and trigger immune responses, persistent senescent cells may evade immune clearance, causing chronic inflammation, fibrosis or relapse [53]. This dual nature of senescence highlights the importance of the dynamic interplay between tumor cells and immune components, where ineffective clearance of senescent cells can promote immunosuppression and disease progression [65]. Senolytic agents, which target anti-apoptotic proteins like BCL-2, BCL-xL, and MCL-1 [66–72], offer a strategy to eliminate these cells. In HCC, CDK4/6 plus XPO1 inhibition induces CRBN-dependent senescence, which is targetable by CRBN-based PROTACs [73]. SASP-driven recruitment of CCR2⁺ myeloid cells can suppress NK cell activity and enhance tumor growth [74]. Targeting SASP signaling, such as through co-inhibition of Aurora A and JAK2, can reprogram the immune microenvironment to favor CD8⁺ and NK cell infiltration while reducing immunosuppression [75].

Aurora B inhibition induces senescence in HCC cells, enhances MHC I expression, and activates CD8⁺ T cells, shifting the balance toward immune-mediated destruction by increasing tumor immunogenicity. When combined with anti-PD-1 therapy, Aurora B inhibition leads to synergistic tumor control, consistent with previous evidence that Aurora kinase inhibitors can enhance immune checkpoint inhibitor (ICI) efficacy by remodeling the immune microenvironment (e.g., Aurora A inhibition plus anti-CTLA-4 in HPV-related cancers) [57]. High Aurora B levels are consistently linked to aggressive HCC behavior and poor prognosis [58]. While studies on senescence-associated gene signatures in HCC suggest variable immune infiltration, it remains unclear whether these patterns favor anti- or pro-tumor immunity [51]. Elevated Aurora B correlates with an immunosuppressive microenvironment, characterized by increased Tregs, M0 macrophages, and checkpoint expression [57]. This study directly connects Aurora B inhibition to enhanced immunogenicity, converting immunologically “cold” tumors into “hot” targets for immune clearance.

The combination of Aurora B inhibition and anti-PD-1 therapy outperforms monotherapies, demonstrating robust tumor control and validating senescence induction as a strategy to potentiate immunotherapy responses. Given that Aurora B overexpression correlates with poor prognosis in HCC [37, 39, 40], its inhibition targets both proliferative and immune pathways, offering a strong rationale for clinical trials, particularly in patients with high Aurora B and PD-L1 expression. Clinical trials combining Aurora B inhibitors with PD-1/PD-L1 inhibitors in HCC should consider baseline Aurora B/PD‑1 expression and CD8⁺ T cell infiltration as stratification markers.

This study compellingly demonstrates that Aurora B inhibition triggers senescence, remodels the tumor microenvironment by boosting antigen presentation, and synergizes with PD-1 blockade to enhance CD8⁺ T cell-mediated tumor clearance. It advances the therapeutic frontier by combining cell-cycle arrest with immunotherapy, offering a novel and clinically promising paradigm for HCC and beyond.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

HCC

Hepatocellular carcinoma

RTK

Receptor tyrosine kinase

ICI

Immune checkpoint inhibitor

VEGF

Endothelial growth factor

OS

Overall survival

ORR

Objective response rate

FDA

Food and Drug Administration

HBV

Hepatitis B virus

FGFR4

Fibroblast growth factor receptor 4

FGF19

Fibroblast growth factor 19

GPX4

Glutathione peroxidase 4

TME

Tumor microenvironment

SASP

Senescence-associated secretory phenotype

DepMap

Dependency Map

TCGA

The Cancer Genome Atlas

IFNγ

Interferon gamma

LMP

Low-molecular-weight protein

HLA

Human leukocyte antigen

MHC

Major histocompatibility complex

PD-1

Programmed cell death protein 1

PD-L1

Programmed death-ligand 1

EMT

Epithelial-mesenchymal transition

Author contributions

Conceptualization X. H. M.; Methodology, X. H. M., S. S. W., H. W., C. Y., X. F. C., L. L.; Validation, Y. C.; Investigation, X. H. M., S. S. W., X. F. C., Writing – Original Draft, X. H. M., S. S. W; Writing – Review & Editing, X. H. M., W. X. Q.; Visualization, X. H. M.; Funding Acquisition, X. H. M., C. E. Y.,; Supervision, X. H. M., W. X. Q., Y. C.

Funding

This work was funded by grants from National Natural Science Foundation of China (82303000, 82303796), Shanghai “Post-Qi-Ming-Xing Plan” (23YF1443400), the Shanghai Natural Science Foundation (22ZR1480900), Shanghai Cancer Institute (ZZ-GZR-24-05, ZZ-GZR-24-08).

Declarations

Competing interests

The corresponding author is an editor of this journal and was not involved in the editorial handling of this manuscript.