Background
Primary liver cancer is a malignant tumor with a high incidence and poor prognosis worldwide. Hepatocellular carcinoma (HCC) constitutes 85–90% of primary liver cancers, has a poor prognosis, and ranks as the second leading cause of cancer-related deaths in China, with only 12.1% of patients surviving 5 years. Over half of the global HCC cases and deaths occur in China [
1,
2]. Additionally, the development of HCC involves complex interactions among various factors, pathways, and systems. Despite the helpfulness of staging systems such as Barcelona Liver Cancer Staging (BCLC), Chinese Liver Cancer Staging (CNLC), and tumour–node–metastasis (TNM) in assessing prognosis and outcome, the high heterogeneity of HCC leads to significant survival differences among patients at the same clinical stage [
1]. Therefore, different classification and prognostic indicators are needed to facilitate individualised, comprehensive HCC treatment to further enhance patient survival.
Adaptive immune responses play a crucial role in preventing tumour occurrence and development [
3]. CD8
+ T cells are the main immune effector cells that mediate anti-tumour cell infiltration. However, continuous stimulation by tumour antigens causes infiltrating CD8
+ T cells to eventually lose their effector functions and memory characteristics. This phenomenon, known as CD8
+ T Cell Exhaustion (CD8
+ TEX) [
4,
5], results in a shift in functionality, such as diminished effector capabilities, sustained expression of inhibitory receptors, epigenetic and transcriptional profile changes, and metabolic alterations [
6,
7]. CD8
+ TEX poses a formidable obstacle to current anti-cancer immunotherapy, as they lose the ability to produce anti-tumour cytokines, along with compromised proliferation and cytotoxicity [
8]. In patients with HCC, variations in prognosis may largely hinge on the extent of uncontrolled CD8
+ T cell exhaustion [
9,
10]. Previous studies indicate that CD8
+ T cells in the intermediate stages of exhaustion exhibit the most favourable response to treatment with immune checkpoint inhibitors, which can effectively restore T cell cytotoxic functions [
5,
11]. Therefore, a complementary diagnosis based on the CD8
+ TEX stage (or CD8
+ TEX subtype) is expected to positively enhance the therapeutic effectiveness of immune checkpoint inhibitors [
12].
In this study, we comprehensively analysed CD8+ TEX-related genes prognostic for HCC and explored their value in predicting patient outcomes. We constructed a prognostic risk model for HCC and analysed and validated its predictive efficacy. The significance of screening and characterizing CD8+ TEX markers as well as potential HCC therapies, was analysed. Concurrently, the biological function of the CD8+ TEX-related genes in HCC was validated at the cellular level, providing new insights into individualized clinical treatment.
Materials and methods
Data sources
Ribonucleic acid sequencing (RNA-seq) data of 365 patients with HCC and their corresponding clinical information were downloaded from The Cancer Genome Atlas (TCGA; portal.gdc.cancer.gov) and Genotype-Tissue Expression (GTEx; gtexportal.org/home/) databases. Additionally, RNA-seq data and relevant clinical information from 240 Japanese patients with HCC were downloaded from the International Cancer Genome Consortium (ICGC) database to serve as an external validation cohort. Gene expression data were standardized using the ‘Sanger box’ tool (
http://sangerbox.com/). Based on previous studies [
13], specific pathways associated with CD8
+ T cell exhaustion were identified, including ‘REACTOME_TNF_SIGNALING’, ‘REACTOME_INTERLEUKIN_2_SIGNALING’, and ‘REACTOME_INTERFERON_GAMMA_SIGNALING’ from the MsigDB database. The toxicity-related genes from the ‘KEGG_NATURAL_KILLER_CELL_MEDIATED_CYTOTOXICITY’ pathways were used as CD8
+ T cell exhaustion-related genes, resulting in the acquisition of 270 T cell exhaustion-related genes. The network of the top 20 interacting proteins of model genes (Spearman correlation coefficient ≥ 0.9) was mapped using the String online database [
14] in combination with Cytoscape (
https://cytoscape.org) software. Furthermore, cBioPortal (cbioportal.org/) was used to analyse the mutation characteristics of the model genes in HCC based on the TCGA PanCancer Atlas [
15].
Liver cancer cell lines and experimental antibodies
MHCC97L, MHCC97H, HCCLM3, Huh-7, Hep3B, SK-Hep-1and LO2 cells were provided by the Hepatology Institute of Shanghai Fudan University, and SNU368 was obtained from the Korean Cell Line Bank (Seoul, Republic of Korea). Anti-MAD2L2 (BM5428) and anti-STAM (A00864-1) antibodies were purchased from BosterBio (Pleasanton, CA, USA), while Anti-TBL1XR1 (MBS850368), Anti-ANXA5 (MBS474163), Anti-FKBP1A (MBS9404059) and Anti-PPM1G (MBS626576) antibodies were purchased from MyBioSource (San Diego, CA, USA).
Immunohistochemical method
Post-operative paraffin tissue samples were collected from 20 patients with HCC who underwent surgery at our hospital between October 2021 and December 2022. The inclusion criteria were: postoperative pathological confirmation of hepatocellular carcinoma, no prior anti-cancer treatment before surgery, complete clinical data, and collection of both cancer tissue and adjacent normal liver tissue samples (> 2 cm away from the tumour). The exclusion criteria were: complications with other malignant tumours and co-occurrence of cardiovascular and respiratory diseases. The Ethics Committee of the General Hospital of Ningxia Medical University approved this study (No. KYLL-2022–0413). All study participants provided written informed consent.
The HCC tissue sections were de-waxed using xylene and an alcohol gradient. Rehydration was performed after 2 h in 60℃ ovens. After antigen repair, incubation at room temperature with 3% hydrogen peroxide was performed for 30 min. The samples were then immersed in phosphate-buffered saline for 1 h after vibration washing. Subsequently, the primary antibody was added and incubated at a constant temperature (4 ℃) overnight. The following day, the sections were allowed to reach room temperature, and secondary antibodies were applied after a vibration wash. They were placed at room temperature for 2 h. Finally, alcohol gradient dehydration, xylene transparency, and sealing were performed, followed by microscopic observation.
Prognosis models for depleted CD8+ T cell-related genes
Differentially expressed genes (DEGs) were included in a univariate Cox regression analysis to identify prognostic genes with statistically significant differences. Subsequently, based on the R (R Foundation for Statistical Computing, Vienna, Austria) language package ‘glmnet’ [
16], the DEGs related to HCC prognosis were analysed via LASSO regression with tenfold cross-validation. The model’s risk score was calculated for each patient using the coefficients from the least absolute shrinkage and selection operator (LASSO) regression analysis and each gene’s expression level. The risk score is computed as: ∑β
x*Exp
x (β
x represents the coefficient of each gene screened by LASSO regression analysis, Exp
x represents the expression level of these genes). The median risk score was used as a cut-off value to divide the training set into high- and low-risk groups. Survival analysis was conducted by generating Kaplan–Meier curves using the ‘survival’ package, and time-dependent receiver operating characteristic (ROC) curves in the ‘time-ROC’ package were used to evaluate the predictive efficiency of 1-, 3-, and 5-year survival.
Nomogram construction and verification
The RMS package [
17] in R language software and Cox regression analysis were used to construct a column graph for prognosis prediction based on patient sex, age, Child–Pugh grade, alpha fetoprotein (AFP) level, pathological tissue grade, T stage, and risk score. Calibration curve and consistency index (C-index) were used to evaluate the validity of the nomogram.
GO/KEGG enrichment analysis
The DAVID database (
https://david.ncifcrf.gov/) was used for Gene Ontology (GO) function and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis [
18‐
21]. Gene function analysis included biological process (BP), molecular function (MF), and cellular component (CC), with a primary focus on KEGG pathways. The main enrichment results were visualized when the false discovery rate was < 0.05 and
P < 0.05.
Difference analysis of tumour immune microenvironment
The INESTIMATE algorithm was used to infer the matrix components and levels of immune cells, and tumour purity was estimated for each sample. CIBERSORT [
22] is an analytical tool that uses gene expression data to estimate the composition of immune cells in a mixed-cell population. The components of the 22 immune cells in all samples were evaluated using the ‘CIBERSORT’ function package of the R software, and then the correlation between the risk scores of the prognostic model and the components of the 22 immune cells was analysed. Concurrently, the correlation between the risk score of the prognostic model and co-immune checkpoints was analysed.
Drug sensitivity analysis
The ‘onco-Predict’ package was used to predict the treatment response to small molecular compounds in patients belonging to high- and low-risk groups. Drug sensitivity was predicted according to the half-maximal inhibitory concentration (IC50), and the differences in drug sensitivity between the high- and low-risk groups were evaluated. PubChem (
https://pubchem.ncbi.nlm.nih.gov/) was used for visualising the 3D conformations of the drugs.
Cell culture and transfection
All cells were cultured at 37 °C with 5% CO2 in Dulbecco’s Modified Eagle Medium (DMEM) medium containing 10% fetal bovine serum. The small interfering RNAs (siRNAs) targeting STAM, ANXA5, and MAD2L2 were purchased from GenePharma (Shanghai, China). siRNAs were transfected into the MHCC97H, MHCC97L, and HCCLM3 cell lines using lipo2000. A follow-up experiment was conducted 48 h later.
A total of 3000 cells were evenly distributed in 6-well plates and incubated for 2 weeks. The cells were then fixed with 4% paraformaldehyde and stained with 0.5% crystal violet staining solution. The cell clusters were counted.
Cell proliferation assay
Cell viability was analysed using a cell counting kit-8 (CCK8) (Biomake, China) according to the manufacturer’s protocol. Cells were cultured in 96-well plates (3 × 103 cells/well) and tested at 12, 24, 48, 72, and 96 h after plating. A volume of 100 μL CCK-8 working liquid (90 μL DMEM + 10 μL CCK-8 solution) was added to each well of the cells to be tested and cultured for 1 h. The absorbance at 450 nm was measured with an enzyme-labeled instrument.
Wound healing assay
Cells were cultured in 6-well plates. When the cell density reached 90%, the cell surface was marked with a 200μL plastic straw head and then they were cultured with serum-free DMEM medium. The scratches were photographed at different time points (0 h and 48 h) using an inverted microscope (Olympus, Tokyo, Japan), and cell migration was recorded.
Transwell assay
A transwell chamber with an 8μm aperture (Solarbio, Beijing, China) was used to detect migration capacity. For the cell migration experiment, 8 × 104 cells were resuspended in 200 μL serum-free DMEM and placed in the upper chamber, and 600 μL DMEM containing 10% fetal bovine serum was placed in the lower chamber. The cells were cultured at 5% CO, and 37 °C for 48 h, and the upper chamber cells were wiped with a cotton swab. Cells in the lower chamber were fixed with 4% paraformaldehyde for 20 min and stained with 0.5% crystal violet for 30 min. The cells were photographed with a microscope in the upper left, lower left, upper right, lower right, and middle fields of view.
Western blotting
Cells were lysed with radio-immuno-precipitation assay (Beytime, Beijing, China), and then centrifuged at high speed at 4 °C to obtain the protein solution. The protein concentration was determined using the bicinchoninic acid (KeyGEN BioTECH, Jiangsu, China) method. Sample buffer was added and boiled for 10 min and stored at –20 °C for subsequent use. Proteins were isolated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Proteins were then transferred to PVDF (Merck Millipore, Burlington, MA, USA) and incubated in 5% skim milk at room temperature for 1 h. PVDF membranes were incubated with a specific primary antibody at 4 °C overnight. After washing the PVDF membrane three times, a secondary antibody was incubated at 37 °C for 1 h, and the chemiluminescence method was used for detection.
Statistical analysis
All mRNA expression data were normalized and log-transformed using log2 (data + 1). Statistical analyses were performed using R software (V3.8.3). The Mann–Whitney U test was used to analyse the difference in expression between HCC and normal tissues, while the paired sample t-test was used for paired sample analysis (ns, P ≥ 0.05; *, P < 0.05; **, P < 0.01; ***, P < 0.001). The R package (ggplot2; version 3.3.3) was used for data visualization. The chi-square test or Fisher 's exact test was used to analyse the clinicopathological features of patients with low- and high-risk prognoses for liver cancer. The Kaplan–Meier method was used to compare the overall survival rate of patients with low- and high-risk prognoses for liver cancer, and the rank sum test was used to compare the differences between groups. Statistical significance was set at P < 0.05.
Discussion
This study aimed to explore the role of CD8
+ TEX cells in hepatocellular carcinoma (HCC) and to develop a prognostic model based on their unique characteristics and molecular patterns. HCC ranks as the sixth most common malignancy and the fourth leading cause of cancer-related deaths worldwide, signifying its ongoing prominence as a major health concern [
2]. Despite notable advancements in anti-HCC therapy, the long-term prognosis for patients with HCC remains unfavourable because of a limited understanding of the underlying mechanisms of tumour development and the absence of personalized treatment options for advanced HCC [
23]. Existing staging systems, such as TNM staging and the Japanese comprehensive staging of liver cancer, mainly consider the influence of physical indicators, such as tumour burden, on patient prognosis. However, patients at the same stage often experience different prognoses. Risk stratification and individualized treatment for distinct patients warrant further exploration [
24]. Therefore, early and precise prognosis prediction for patients with HCC has attracted considerable attention. CD8
+ TEX cells, characterized by unique molecular patterns and transcriptional characteristics, are closely related to the occurrence and progression of HCC, making them promising diagnostic and prognostic biomarkers as well as potential therapeutic targets.
CD8
+ T cells play a crucial role in the antitumour immune response of the body, and patients with HCC who exhibit an antigen-specific CD8
+ T cell response tend to have better overall survival. Tumour tissue-infiltrating CD8
+ T cells are closely associated with postoperative survival and tumour recurrence [
25‐
27]. However, HCC has a high degree of molecular complexity and genetic heterogeneity, leading to the recognition of different tumour antigens in HCC tissues. Specific CD8
+ T cells recognize these tumour-associated antigens in autologous tumour tissues, and this process is associated with tumour malignancy. Notably, Flecken T et al. [
11] showed that specific CD8
+ T-cell recognition of tumour-associated antigens was associated with prolonged progression-free survival in patients with HCC. Additionally, several studies have highlighted significant heterogeneity among different CD8
+ T-cell immune responses to tumour-associated antigens within different cohorts of patients with HCC [
11,
28,
29]. In the tumour microenvironment, persistent stimulation by antigens stimulation leads to the gradual loss of effector function and memory characteristics. Consequently, tumour-infiltrating CD8
+ T cells gradually become CD8
+ TEX cells, exhibiting reduced functionalities [
6]. The proportion and surface molecular marker distribution of CD8
+ TEX cells may vary in different tissues and diseases. However, no systematic research has explored the differences in molecular marker genes on the surface of CD8
+ TEX cells and their correlation with the prognosis of patients with HCC.
In this study, our results revealed that the prognosis for patients with HCC in the high-risk group was notably worse compared to the low-risk group. Additionally, the risk score was identified as an independent risk factor for the prognosis of HCC patients. Differentially expressed genes in the high-risk group were mainly associated with the regulation of the immune response, the hypoxia-inducible factor-1 (HIF-1) signalling pathway, and the tumour PD-1/PD-L1 immune checkpoint pathway. This underscores the effectiveness of CD8
+ TEX marker genes in HCC as potential targets for antitumour therapy, including immune checkpoint blockade [
30]. We also identified a strong correlation between the high-risk group of HCC patients and several clinicopathological features, such as BMI, AFP level, T stage, histological grade, pathological grade, and micro-vascular invasion. This suggests that CD8
+ TEX cells promote the malignant progression of HCC and participate in the invasive development of the HCC metastasis process. Therefore, exploring individual differences and epigenetic modifications in CD8
+ TEX cells in HCC enables targeted intervention for personalized, comprehensive treatment, enhancing survival prognosis and developing more effective treatment methods.
In the present study,
MAD2L2, STAM, ANXA5, TBL1XR1, FKBP1A, and
PPM1G were identified as the prognostic differential genes associated with CD8
+ TEX in HCC. Some of these genes have been studied in the context of model genes. For instance,
MAD2L2 is an important component of the mitotic checkpoint complex that interacts with various proteins and is critically involved in various cellular functions, including DNA synthesis, mitosis, and phenotypes caused by changes in cell exhaustion [
31].
STAM participates in intracellular signal transduction, mediates DNA synthesis and growth factor signal transduction under the stimulation of IL-2 and other cytokines, and plays a role in T cell development [
32]. As a member of the calcium-dependent phospholipid-binding protein family, annexin A5 plays a regulatory role in physiological and pathological processes, such as cell signal transduction, inflammation, growth, and proliferation, and is involved in tumour progression, invasion, metastasis, and drug resistance, and the treatment processes [
33]. Over-expression of
ANXA5 promotes malignancy and lymph node metastasis in mouse HCC cells and is a potential marker of malignant tumours and lymph node metastasis [
34].
TBL1XR1 is a transcriptional cofactor containing F-box and WD-40 domains, involved in the regulation of various signal transduction pathways, and plays an important role in the epithelial-mesenchymal transition, drug resistance, proliferation and play an important role in metastasis [
35‐
37].
FKBP1A is a member of the immunophilin protein family, which prevents TGF-
β receptors from being activated by ligands, and mediates immune response regulation and protein folding and transport processes [
38].
PPM1G belongs to the PP2C family of protein phosphatases, and the PP2C family is a negative regulator of cell stress response pathways.
PPM1G also plays an important role in regulating cell cycle progression [
39]. In previous studies, the biological significance of
TBL1XR1,
FKBP1A, and
PPM1G in tumours has been well-documented. The present study demonstrated that the expression levels
of ANXA5, MAD2L2, and
STAM in liver cancer cell lines were significantly higher than those in normal hepatocytes. Moreover, the knockdown of these genes significantly inhibited the proliferation and migration of liver cancer cells, highlighting their potential influence on the malignant progression of HCC. Consequently, these CD8
+ TEX characteristic genes emerge as promising candidates for targeted therapies in HCC.
In addition, we screened six potential small-molecule compounds: BX.795, Dasatinib, Elesclomol, BMS-754807, Doxorubicin, and Epothilone-B. BX-795, functioning as a PDK-1/TBK-1 inhibitor, blocks PDK1/Akt signalling within tumour cells, thereby inhibiting their proliferation and inducing apoptosis [
40]. Dasatinib, a multi-target kinase inhibitor, inhibits the proliferation, invasion, and migration of liver cancer cells and is a potential drug for HCC targeted therapy [
41]. Elesclomol, known for its safety in clinical applications, exhibits strong anti-cancer activity. Furthermore, when combined with taxane anti-cancer drugs, it improves tumour-free survival and tumour radical cure rates [
42]. BMS-754807, an IGF1R/IR inhibitor, induces G2/M arrest in tumour cells and inhibits tumour cell proliferation [
43]. Similarly, Doxorubicin and epothilone-B have been extensively studied as anti-tumour therapies. In the present study, patients in the low-risk group were more sensitive to BX.795, Dasatinib and Elesclomol, whereas those in the high-risk group were more sensitive to BMS.754807, doxorubicin, and epothilone. The culmination of these findings underscores the potential of these small-molecule compounds as therapeutic agents for HCC; however, further analysis is needed in the near future.
However, there are still certain limitations in this study, and there is still a lack of real-world data to validate this prediction model. In addition, there is a lack of in-depth exploration and in vivo experimental verification of the role and mechanism of CD8+ TEX-related gene in HCC. Therefore, in future research, we will continue to explore the mechanism of CD8+ TEX-related gene in HCC.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.