Characterization of T cell proliferation-associated hepatocellular carcinoma subtypes, predictive signatures, and candidate targets

Introduction

At the global level, hepatocellular carcinoma (HCC) is the 6^th most common and 3^rd deadliest cancer globally. HCC patients often exhibit underlying chronic liver disease as a consequence of alcohol abuse, non-alcoholic steatohepatitis (NASH), or infections with the hepatitis B virus (HBV) or hepatitis C virus (HCV) (1). Among these etiologies, HBV is a DNA virus that can induce chronic necroinflammatory responses, thereby promoting hepatocyte mutations and leading to the development of HCC. HCV, an RNA virus, contributes to tumorigenesis through repeated cycles of hepatic injury, regeneration, and fibrosis. In the case of NASH, the pathological progression toward HCC is primarily characterized by hepatic steatosis, inflammation, and fibrosis (2). Different stages of HCC necessitate different treatment strategies, with the selection of locoregional treatment strategies being dependent in large part on liver function and on the location and overall burden of tumors (3). When tumors are only present within the liver, surgical resection, transarterial chemoembolization, and/or radiofrequency ablation can be effective, whereas unresectable cases necessitate systemic treatments that target the tumor microenvironment (TME) (4). HCC is a complex multifactorial disease that entails tumor suppressor gene inactivation and oncogenic activation together with dysregulated signaling pathway activity, altered cellular differentiation, and angiogenic activity (5). As the onset of HCC is generally insidious, radical treatment is only accessible to under 30% of HCC patients at the time of initial diagnosis. Systemic antitumor treatment has the potential to provide advanced HCC patients with a better prognosis (6). Even with newer systemic therapies, however, these patients currently face low survival rates. As HCC has a very complex pathogenesis, researchers have explored many therapies targeting specific factors of pathways of interest (7). A detailed understanding of the molecular mechanisms that underlie the onset of HCC is thus vital to enable the exploration of treatment-related biomarkers that may guide the better management of this malignancy.

There are many different types of cells within the TME, including the populations of tumor, stromal, and immune cells (8). These include natural killer (NK) cells, which are chosen with increasing frequency as relevant targets for cancer treatment. T cells are also widely studied in therapeutic contexts. The complex dynamics of the TME are inextricably linked to processes such as oncogenesis and metastasis to distant sites, with the continuous bidirectional interactions between tumor cells and their environment shaping their overall adaptive fitness. This stable communication between tumors and the TME has been found to be associated with prognostic outcomes (9). Current technical limitations have limited most studies conducted to a focus only on a handful of T cell proliferation-related genes (TRGs) and cell types, failing to provide any insight into the antitumor activity that can arise from the complex, coordinated effects of multiple genes. A systematic understanding of the infiltration of the TME by different cell types and the roles that different TRGs play in this setting may thus offer key insights into the mechanistic basis for HCC development.

T cells can readily localize to tumors wherein they can exert direct antitumor cytolytic activity of play supporting roles by recruiting other immune cell types. Adoptive T cell therapies, which entail the treatment of patients with allogeneic or autologous T cells, have achieved promising activity in recent studies. Through their in vivo expansion and differentiation into memory cell subsets, T cells can also facilitate persistent antitumor monitoring (10). Given their unique antigen-specific cytotoxic functionality, T cells are major targets of efforts to treat and prevent cancer (11,12). In one recent report, 33 TRGs were identified as drivers of the proliferation of T cells in a genome-wide screen (13), though the specific roles that these genes play in HCC have yet to be documented. Efforts to elucidate the impacts of TRGs on cancer patient outcomes are essential to guide new therapeutic development, underscoring the importance of exploring the association between TRGs and prognostic outcomes in patients.

In the present study, an effort was made to explore TRGs in HCC with a focus on their genetic dysregulation, prognostic relevance, and expression. Through these efforts, different HCC patient molecular subtypes and predictive risk signatures were established, providing a means of accurately gauging patient outcomes. Candidate therapeutic biomarkers with prospects for clinical approval were also developed, and levels of certain key TRGs was assessed in samples of tumor and adjacent normal tissue from patients with HCC as well as HCC cell lines to validate these findings. Taken together, based on the TRGs, specific molecular subtypes and predictive signatures were identified, together with potential therapeutic targets for the treatment of HCC. We present this article in accordance with the MDAR and TRIPOD reporting checklists (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-50/rc).

Methods

Data collection

Clinical and transcriptomic data corresponding to HCC patients was accessed through public repositories including the Genomic Data Commons Data Portal (https://portal.gdc.cancer.gov), the Cancer Genome Atlas (TCGA)-HCC project, and the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/, ID: GSE52018 and GSE76427). For details on the HCC patients within these datasets, see Table S1. Overall survival (OS) data were obtained from the TCGA-HCC, GSE52018, and GSE76427 datasets, allowing for the development and validation of HCC patient molecular classifications and a prognostic risk signature. Patients were excluded from these analyses if they lacked complete clinical or follow-up-related data.

HCC-related genetic and transcriptional TRG changes

One recent report established a list of 33 TRGs (13), which are fully detailed in Table S2. Mutation-related information for these genes was obtained from the TCGA, while their potential biological functions were explored through Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses in R. TRG expression levels in normal and tumor tissues were compared with Wilcoxon signed-rank tests and the R limma package. Kaplan-Meier (KM) and univariate Cox regression analyses were also utilized to assess the prognostic performance of these TRGs.

TRG consensus clustering

Different molecular classes of HCC patients were established through consensus clustering based upon TRG expression levels. The clustering parameter, k, was optimized to establish a categorization scheme in which intragroup connections were maximized while intergroup connections were minimized. The two resultant molecular classifications of HCC were then distinguished through a principal component analysis (PCA) approach with the R ‘stats’ package. Differences in survival for these TRG-related clusters were compared through KM curves and the log-rank test with the R ‘survival’ and ‘survminer’ packaged. Clinical parameters were compared between these TRG-based clusters, and differentially expressed genes (DEGs) were identified by comparing the two with a set of established criteria [fold change (FC) >1.5, P<0.05]. Immune cell infiltration and immune-related pathways were identified in TRG clusters through a gene set variation analysis (GSVA) and single-sample gene set enrichment analysis performed with the gsva R package. Prognosis-related DEGs (PRDEGs) were assessed via univariate Cox regression analyses.

Classification of patients into DEG-based gene clusters

Those DEGs identified when comparing TRG clusters were utilized for consensus clustering for the classification of HCC patients into three separate groups. Heatmaps, boxplots, and Wilcoxon signed-rank tests were employed to assess the levels of TRG expression and clinical parameters in these different gene clusters. KM curves were also employed to contrast survival outcomes between these clusters with the log-rank test.

TRG-based prognostic signature development and validation

After identifying genes differentially expressed among these three gene clusters, least absolute shrinkage and selection operator (LASSO) regression and multivariate Cox regression approaches were used to develop a predictive risk signature with the following general formula:

Riskscore=∑i=1nβi∗λi[1]

where n, βi, and λi respectively correspond to the number of candidate genes, coefficient values, and expression levels. Risk scores were then leveraged to separate HCC patients into low- and high-risk groups. Associations between risk scores, survival duration, and status were assessed. Independent prognostic factors in these HCC patients were also identified through univariate and multivariate Cox regression analyses used to assess both the risk scores as well as patient clinical characteristics. The ability of the TRG-based prognostic risk signature to predict survival outcomes in individuals with HCC was additionally assessed in the training, testing, and independent validation (GSE52018) cohorts with the KM and receiver operating characteristic (ROC) approaches.

Nomogram scoring system development and validation

A predictive nomogram was designed based on HCC patient risk scores and clinical parameters according to the outcomes of independent prognostic analyses. In the resultant nomogram, individual variables were matched with corresponding scores, and these scores were summed to produce a total score. The performance of this nomogram was then evaluated based on 1-, 3-, and 5-year survival-related time-dependent ROC curves. The concordance between actual and nomogram-predicted HCC patient outcomes was assessed with calibration plots.

Patient samples

Tumors and paracancerous tissues at least 1 cm from the tumor margin were obtained from 20 HCC patients that underwent surgical resection of their tumors at The First Affiliated Hospital of Wannan Medical College from January 2023 through December 2024. All patients provided written informed consent, and samples were stored at −80 ℃. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Ethics Committee of The First Affiliated Hospital of Wannan Medical College (No. WNMC-AWE-2023121).

Quantitative polymerase chain reaction (qPCR)

To validate differences in key TRG expression levels in HCC, qPCR assays were performed. After using TRIzol (Invitrogen, Carlsbad, USA) to extract total RNA, cDNA was prepared with a PrimeScript RT kit (Vazyme, Nanjing, China). TB Green Premix Ex Taq II (GenStar, Shenzhen, China) and a LightCycler480 instrument (Applied Biosystems, Boston, MA, USA) were then used for qPCR analyses. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used for normalization, and relative expression was quantified via the 2^−ΔΔCt method. Differential expression was compared with Student’s t-tests. Utilized primers are shown below: SOX12, forward: 5'-AAGAGGCCGATGAACGCATT-3', reverse: 5'-TAGTCCGGGTAATCCGCCAT-3'; LPCAT1, forward: 5'-CGCCTCACTCGTCCTACTTC-3', reverse: 5'-TTCCCCAGATCGGGATGTCTC-3'; MMP1, forward: 5'-AAAATTACACGCCAGATTTGCC-3', reverse: 5'-GGTGTGACATTACTCCAGAGTTG-3'; CLEC3B, forward: 5'-CCCAGACGAAGACCT TCCAC-3', reverse: 5'-CGCAGGTACTCATACAGGGC-3'; KRT17, forward: 5'-GGTGGGTGGTGAGATCAATGT-3', reverse: 5'-CGCGGTTCAGTTCCTCTGTC-3'; GAPDH, forward: 5'-GGGAAGGTGAAGGTCGGAGT-3', reverse: 5'-GGGGTCATTGATGGCAACA-3'.

Cell culture

The HepG2 and Huh-7 HCC cell lines were obtained from the American Type Culture Collection (ATCC) and cultured in Dulbecco’s Modified Eagle Medium (DMEM) containing 10% fetal bovine serum (FBS; Lonsera, Sydney, Australia) with 1% penicillin/streptomycin at in a 37 ℃ 5% CO₂ incubator.

Western blotting

Protease and phosphatase inhibitor-containing RIPA buffer (Beyotime, Shanghai, China) was used to extract total protein from appropriate tissue and cell samples. Western blotting was then performed as in past reports (14), using primary antibodies specific for keratin 17 (KRT17; 1:1,000, Abcam, Cambridge, UK), GAPDH (1:20,000, Proteintech, Wuhan, China), AKT (1:1,000, CST, Boston, USA), p-AKT (1:2,000, CST), mTOR (1:1,000, CST), p-mTOR (1:1,000, CST), Bcl-2 (1:1,000, CST), Bax (1:1,000, Proteintech), and cleaved caspase-3 (1:1,000, CST).

siRNA transfection

KRT17-specific and control siRNAs were obtained from GenePharma Co., Ltd. (Shanghai, China), with the following sequences: si-NC (5'-UUCUCCGAACGUGUCACGUTT-3') and si-KRT17 (5'-GCCAGUACUACAGGACAAUTT-3'). After culturing cells until 50% confluent, Lipofectamine 3000 (Invitrogen, Shanghai, China) was used as directed to transfect them. KRT17 knockdown efficiency was assessed by plating these cells for 48 h in 6-well plates, after which they were washed and used in subsequent analyses.

Transwell assays

The upper chambers of a Transwell insert were filled with 200 µL of serum-free DMEM containing 5×10⁴ HCC cells in which KRT17 had or had not been knocked down, while the lower chamber contained 600 µL of DMEM with 10% FBS. After incubating these plates for 48 h, methanol was used to fix the cells on the bottom of the membrane, and they were visualized with 0.1% crystal violet. A microscope was then utilized to visualize five random fields of view.

Viability and colony formation assays

The selected HCC cell lines in which KRT17 had or had not been silenced were added to 96-well plates (2,000/well) and cultured for 0, 24, 48, 72, or 96 h, after which the Cell Counting Kit-8 (CCK-8) reagent (Beyotime, Shanghai, China) was introduced and incubated for a further 1.5 h before measuring the absorbance at 450 nm.

Colony formation assays were performed using cells in which KRT17 had or had not been knocked down. These cells were added to 6-well plates (3,000/well) and cultured for 10 days, fixed for 15 min with 4% paraformaldehyde, and stained at room temperature for 30 min using crystal violet.

Wound healing assay

After adding cells to 6-well plates (1×10⁶/well) and allowing cells to attach overnight, parallel scratch wounds were generated in the monolayer with a pipette tip. After three PBS washes, the wells were filled with serum-free medium and incubated in a standard tissue culture incubator. The distances over which the cells had migrated after 24 h were analyzed using ImageJ software and compared using t-tests and GraphPad Prism (n=3).

Statistical analysis

All statistical analyses were performed using R version 4.2.1 and GraphPad Prism 9.0. Statistical significance was defined as P<0.05.

Results

Identification of TRG clusters using consensus clustering

The clinical data for all patients included in this study are presented in Table S1. A total of 33 TRGs were assessed in these analyses (Table S2). The GO and KEGG pathways in which these TRGs were enriched were also assessed (Figure 1A,1B). Most of the analyzed TRGs exhibited differential expression between normal and tumor samples (P<0.05), with CXCL12, FOSB, IL1RN, GPD1, CYP27A1, and AKR1C4 being downregulated while the remaining genes exhibited upregulation (Table S3). A network was constructed to explore interactions among these TRGs and their prognostic relevance in HCC (Figure 1C). These analyses revealed significant differences in TRG expression when comparing the HCC and normal control samples, highlighting a potentially central role for these genes in HCC development.

Figure 1 Consistent clustering of the two TRG clusters. (A) GO functional enrichment analysis of TRGs. (B) KEGG analyses of the functional enrichment of TRGs. (C) TRG interactions in HCC. (D) Consensus matrix heatmap with two clusters (k=2) and associated correlations. (E) A Kaplan-Meier curve revealing the better prognostic outcomes for those patients in TRG cluster B relative to cluster A (P<0.001). (F) PCA plots indicating effective transcriptomic separation between the two subtypes. (G) ssGSEA results revealing levels of immune cell infiltration associated with different TRG clusters. (H) Comparisons of clinical parameters and TRG expression between the two subtypes. *, P<0.05; ***, P<0.001. BP, biological process; CC, cellular component; GO, Gene Ontology; HCC, hepatocellular carcinoma; KEGG, Kyoto Encyclopedia of Genes and Genomes; MDSC, myeloid-derived suppressor cell; MF, molecular function; PCA, principal component analysis; ssGSEA, single sample Gene Set Enrichment Analysis; TCGA, The Cancer Genome Atlas; TRG, T cell proliferation-related gene.

Expression data for these 33 TRGs were leveraged for consensus clustering analyses in which HCC patients were classified into two TRG-based clusters (Figure 1D, Figure S1). In univariate Cox regression analyses, TRG cluster B was associated with more favorable prognostic outcomes than TRG cluster A (Figure 1E). The PCA indicated good separation between these two HCC patient subtypes (Figure 1F). To clarify the prognostic relevance of these 33 TRGs, KM and Cox regression analyses were conducted (Figure S2), while a GSVA approach was employed to assess biological pathways related to these two subtypes of disease (Figure S3). The CIBERSORT algorithm was also employed to assess the correlative relationships between the two patient subsets and 23 different types of immune cells (Figure 1G). This approach revealed greater levels of activated CD4 T cell, NKT cell, and Th2 cell infiltration in subtype A samples as compared to those of subtype B, whereas the opposite pattern was observed for neutrophils, eosinophils, and Th1 cells. Furthermore, distinct clinical characteristics and TRG expression levels were also evident for these two patient subgroups (Figure 1H).

Classification of patients into gene clusters based on the DEGs between the TRG clusters

After establishing a list of DEGs based on the comparison of patients in TRG clusters A and B, the expression profiles for these genes were utilized to separate HCC patients into three distinct gene clusters. Relationships between TRG clusters, these new gene clusters, DEGs, and clinical parameters are detailed in Figure 2A. DEG expression patterns were also analyzed (Figure 2B). The OS of patients in these three clusters was also compared, demonstrating that patients in gene clusters C and A respectively exhibited the best and worst 5-year prognosis (Figure 2C, P<0.001). These clusters were additionally compared to analyze DEGs, and a LASSO approach was used to select PRDEGs with potential utility for predictive signature construction, with 9 signature genes ultimately being selected through screening (Table S4). The LASSO regression process is detailed in Figure 2D,2E. Multivariate Cox regression was further used to screen a list of signature genes (SOX12, LPCAT1, MMP1, CLEC3B, and KRT17), the corresponding coefficients for which are presented in Table S5.

Figure 2 Identification of three gene clusters based on DEGs. (A) Heatmap showing the relationships between the three gene clusters and clinical parameters. (B) DETRG expression levels in these three gene clusters. (C) Kaplan-Meier curves revealing that patients in clusters C and A respectively exhibited the best and worst prognostic outcomes (P<0.001). (D,E) LASSO and stepwise Cox analyses were employed to assess PRDEGs with potential utility for predictive signature construction. *, P<0.05; **, P<0.01; ***, P<0.001. DEGs, differentially expressed genes; DETRG, differentially expressed T cell proliferation-related gene; LASSO, least absolute shrinkage and selection operator; PRDEGs, prognosis-related DEGs; TCGA, The Cancer Genome Atlas; TNM, tumor-node-metastasis.

TRG-related predictive signature development and validation

The expression levels and corresponding coefficients for each gene in the risk signature were used to compute risk scores for each HCC patient. These risk scores were then used to stratify patients into the low- and high-risk groups. A Sankey diagram was generated to assess the relationships between risk scores, TRG clusters, gene clusters, and survival outcomes for these patients (Figure 3A). Boxplots were constructed depicting the risk scores for the two TRG clusters (Figure 3B) and three gene clusters (Figure 3C). TRGs exhibiting differential expression are presented in Figure 3D, while a heatmap was generated to assess differences in the expression of five signature-related genes (Figure 3E). High-risk HCC patients presented with greater mortality risk (Figure 3F). Correlations between risk scores and immune cell infiltration were also examined (Figure 3G), revealing six immune cell types that were correlated with risk scores (resting NK cells, neutrophils, M2 macrophages, resting mast cells, and both resting and activated memory CD4⁺T cells). Figure 3H shows the relationships between risk signature genes and immune cell abundance. The low-risk group also exhibited higher Stromal and ESTIMATE scores (Figure S4).

Figure 3 TRG-related prognostic signature development. (A) A Sankey diagram showing the correlative relationships between risk scores, TRG clusters, gene clusters, and HCC patient survival. (B,C) Boxplots representing the risk scores corresponding to the two TRG clusters as well as the three gene clusters. (D) TRGs exhibiting differential expression between low- and high-risk groups. (E) Differences in the expression of the 5 signature genes. (F) High-risk HCC patients presented with greater mortality risk. (G) Correlations between risk scores and immune cell infiltration. (H) Relationships between signature genes and the abundance of immune cells. **, P<0.01; ***, P<0.001. HCC, hepatocellular carcinoma; NK, natural killer; TRG, T cell proliferation-related gene.

Risk scores and scatter plots demonstrated that higher scores were related to a higher risk of death, and the expression patterns of the five PRDEGs were assessed with heatmaps (Figure 4A-4C). The predictive performance of risk scores when assessing patient survival were compared using area under the curve (AUC) values in the training cohort (TCGA + GSE76427) (Figure 4D, P<0.001, 1-year AUC =0.744, 3-year AUC =0.788, 5-year AUC =0.803), testing cohort (TCGA + GSE76427) (Figure 4E, P=0.02, 1-year AUC =0.716, 3-year AUC =0.632, 5-year AUC =0.602), and GSE2018 (Figure 4F, P<0.001, 1-year AUC =0.781, 3-year AUC =0.713, 5-year AUC =0.670). All three of these datasets thus demonstrated that the survival of low-risk HCC patients tended to be significantly longer than that of high-risk patients, reaffirming the value of risk scores as predictors of prognosis. Based on these results, a nomogram was developed for these HCC patients (Figure 4G). Age, sex, clinical stage, and risk score were found to be significant prognostic factors, and these were used for the subsequent construction of a nomogram for the prediction of HCC prognosis. Using this model, for example, the 1-, 3-, and 5-year survival rates of a 55-year-old stage III/IV HCC patient were 91.8%, 82.1%, and 71.6%, respectively. Calibration plots confirmed the good predictive performance of this nomogram (Figure 4H).

Figure 4 Risk score utility and assessment. (A-C) Ranked dot and scatter plots showing risk scores and prognostic outcomes, together with heatmaps highlighting prognostic risk gene expression patterns in the training, test, and GSE52018 cohorts. (D-F) Kaplan-Meier analyses demonstrating a significantly lower rate of OS for high-risk patients as compared to low-risk patients in all three cohorts. The 1-, 3-, and 5-year survival of these patients was compared with ROC curves. (G) A nomogram used to predict 1-, 3-, and 5-year HCC patient OS. (H) Calibration curves corresponding to nomogram accuracy and specificity when used to predict patient 1-, 3-, and 5-year OS. ***, P<0.001. AUC, area under the curve; HCC, hepatocellular carcinoma; OS, overall survival; ROC, receiver operating characteristic; TNM, tumor-node-metastasis.

Evaluation of the expression and predictive relevance of prognostic signature TRGs

Expression profiles for five prognostic TRGs were assessed in HCC patients and normal control samples from the TCGA dataset (Figure 5A). Differences in prognosis between the groups with low and high levels of expression for these five TRGs in this dataset were assessed with KM curves (Figure 5B-5F), highlighting the differences in prognosis as a function of the expression of all five of these TRGs in HCC. To confirm these results, tumors and paracancerous samples from 20 HCC patients were collected between January 2023 and December 2024, and the mRNA levels of these five PRDEGs were compared (Figure 5G). The results were largely consistent with those from the TCGA dataset. Of these five genes, SOX12, LPCAT1, MMP1, and CLEC3B have been studied at length, whereas relatively little is known about KRT17. As such, KRT17 was chosen as a target for further research. In qPCR and Western blotting experiments, KRT17 upregulation was observed in most of the 20 HCC patient tumor tissue samples relative to their matched normal tissues (Figure 5G,5H).

Figure 5 Evaluation of the expression and prognostic relevance of the five prognostic signature-associated TRGs. (A) Evaluation of the expression of the five prognostic signature-associated TRGs in tumors and normal tissue samples from the TCGA database. (B-F) Kaplan-Meier analyses of the prognostic relevance of these five TRGs based on TCGA data. (G) qPCR analyses demonstrating the expression of these five prognostic genes at the mRNA level in 20 paired HCC patient tumor and paracancerous tissues. (H) Western blotting was employed to detect KRT17 in 20 paired HCC patient tumor and adjacent tissues. *, P<0.05; **, P<0.01; ***, P<0.001. HCC, hepatocellular carcinoma; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; KRT17, keratin 17; N, normal; qPCR, quantitative polymerase chain reaction; T, tumor; TCGA, The Cancer Genome Atlas; TRG, T cell proliferation-related gene.

Immune checkpoint and drug sensitivity

The boxplot showed that the expression of most immune checkpoints was significantly correlated with risk scores (Figure S5A). We found that the high-risk group had a significantly lower half-maximal inhibitory concentration (IC₅₀) for common chemotherapy drugs, suggesting that the risk score may also be potentially related to drug sensitivity (Figure S5B).

KRT17 knockdown impairs HCC cell migration and proliferation in vitro

HCC cells in which KRT17 had been silenced were next established, confirming the efficiency of knockdown (Figure 6A,6B). In a CCK-8 assay, the silencing of KRT17 was found to have significantly impaired the proliferation of both HepG2 and HuH-7 HCC cells relative to controls (Figure 6C). In a colony formation assay, silencing KRT17 similarly impaired colony formation (Figure 6D), while the migration and invasion of these cells were compromised by KRT17 knockdown in Transwell and wound healing assays (Figure 6E,6F). Together, these findings indicated that KRT17 can activate the proliferative and migratory activity of HCC cells in vitro.

Figure 6 KRT17 silencing inhibits HuH-7 and HepG2 cell migration and proliferation. (A,B) The efficiency of knockdown in the indicated cell lines. (C) A CCK-8 assay was used to quantify the impact of KRT17 silencing on proliferative activity in these HCC cell lines. (D) The effect of KRT17 silencing on colony formation was quantified through a colony formation assay (the cells were stained with 0.1% crystal violet and photographed) (scale bar: 1 cm). (E) The impact of knocking down KRT17 on HCC cell migration was interrogated via wound healing assay (observed and photographed using an inverted microscope) (scale bar: 1,000 µm). (F) Transwell invasion assays were employed to determine how knocking down KRT17 affects HCC cell invasivity (cells were stained with 0.1% crystal violet and observed and photographed under an inverted microscope) (scale bar: 50 µm). *, P<0.05; **, P<0.01; ***, P<0.001. CCK-8, Cell Counting Kit-8; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; KRT17, keratin 17; HCC, hepatocellular carcinoma.

KRT17 knockdown induces HCC cell apoptosis and modulates Akt/mTOR signaling activity

In vitro analyses revealed that KRT17 exerts regulatory effects on the apoptotic death of HCC cells. Specifically, KRT17 silencing resulted in an increase in apoptosis relative to control cells (Figure 7A). Consistently, Western blotting revealed increased pro-apoptotic Bax and caspase-3 levels upon KRT17 knockdown, whereas the opposite was true for anti-apoptotic Bcl-2 (Figure 7B-7D).

Figure 7 Knocking down KRT17 induces in vitro HCC cell apoptosis, with KRT17 mediating Akt/mTOR signaling activity in HCC cells. (A-D) Western blotting was used to detect proteins related to apoptosis in HepG2 and HuH-7 cells in which KRT17 was knocked down. (E-G) Western blotting and corresponding quantification of p-Akt and p-mTOR levels in the indicated cell lines in which KRT17 was knocked down. *, P<0.05; **, P<0.01; ***, P<0.001. Akt, protein kinase B; HCC, hepatocellular carcinoma; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; KRT17, keratin 17; mTOR, mammalian target of rapamycin.

To clarify the mechanisms whereby KRT17 can influence HCC cell growth, further Western blotting was performed in which the knockdown of KRT17 was found to limit Akt phosphorylation (Figure 7E,7F). This coincided with a decline in the phosphorylation of mTOR, which functions downstream of Akt (Figure 7G). In light of these results, KRT17 may play control HCC cell growth, invasion, and apoptotic death through the modulation of Akt/mTOR signaling.

Discussion

Most research efforts to date have centered on individual TRGs or cell types, hampering any systematic efforts to fully elucidate the combinatory effects of multiple TRGs and their relationship with the infiltration and composition of the TME. There has also not been sufficient effort focused on the utility of these TRGs as predictors of prognostic outcomes in HCC patients or on screening for new candidate therapeutic biomarkers.

TRG expression patterns were herein leveraged to stratify HCC patients into two separate clusters. Those patients grouped in TRG cluster B survived for longer on average, and immune cell infiltration characteristics differed markedly between these clusters. Transcriptomic differences were also apparent when comparing these subsets of patients. The genes differentially expressed between these two clusters were further used to define three gene clusters. Through these efforts, TRGs were confirmed to offer utility as predictors of HCC patient clinical outcomes. A prognostic TRG score was thereby developed, and its predictive performance was quantified. Patient classification into two groups was then conducted using risk scores, and significant differences in terms of clinical parameters, TME characteristics, and prognosis were noted between the low- and high-risk groups. The integration of the TRG score and clinical features then led to the establishment of a nomogram that afforded even better performance and enabled practical use of these TRG risk scores. The prognostic model devised herein can allow for the stratification of HCC patients based on their prognosis, while also offering insight into the mechanisms that govern this malignancy and potential avenues towards its more effective and targeted treatment.

In HCC, the TME is highly complex and dynamic (15). Multiple studies have emphasized the critical role that the interplay between the TME and tumor cells plays in HCC evasion of the immune system (16). Indirect or direct approaches to targeting the TME thus have the potential to revolutionize the treatment of this cancer (17). Six immune cell types were found to be associated with HCC patient risk scores (resting NK cells, neutrophils, M2 macrophages, resting mast cells, and both resting and activated memory CD4⁺T cells), and specific signature genes were also closely related to particular types of immune cells. Within the TME, immune cells can crucially shape HCC-related immune defenses. Tumor-associated macrophages are broadly classified into the pro-tumorigenic M2 and anti-tumorigenic M1 subtypes. Of these, M1 macrophages can release a range of inflammatory cytokines with anticancer activity (18), whereas the immunosuppressive activity of M2 macrophages and associated remodeling of the matrix can support more robust tumor growth (19). Tumor-induced M2 polarization has also been identified as a relevant therapeutic target, with canonical Wnt/β-catenin signaling driving HCC tumor immunosuppression, progression, migration, and metastasis (20). In line with these past reports, M2 macrophage levels in the high-risk group of HCC patients with a poorer prognosis were elevated.

Through LASSO and multivariate Cox-based screening approaches, SOX12, LPCAT1, MMP1, CLEC3B, and KRT17 were ultimately selected as key TRGs. A qPCR approach was used to assess the levels of these five key TRGs in HCC cell lines and normal human liver samples. This strategy revealed the significant upregulation of SOX12, LPCAT1, MMP1, and KRT17 in HCC cells, while CLEC3B was significantly downregulated in these tumor cells relative to normal tissue controls. SOX12 overexpression was significantly associated with microvascular invasion and the loss of tumor encapsulation. In HCC, SOX12 is also reportedly associated with aberrant CD8⁺ T cell and Treg infiltration (21). Through its functions in the MAPK pathway, CLEC3B is positively correlated with the inhibition of proliferation such that it may serve as a tumor suppressor in the context of interactions between tumors and the immune system (22). LPCAT1 has been identified as the enzyme primarily responsible for phosphatidylcholine (PC) saturation. Overexpression of LPCAT1 has been found to induce PC remodeling and affect membrane fluidity in many cancer types, leading to the sustained activation of oncogenic receptor tyrosine kinases (23,24). LPCAT1 levels have been found to be associated with alpha-fetoprotein (AFP), tumor-node-metastasis (TNM) stage, Eastern Cooperative Oncology Group (ECOG) scores, and tumor grade (24). The zinc-dependent endopeptidase MMP1 is closely related to HCC cell migration and invasive growth. LPCAT1 and CLEC3B have also been found to offer utility as predictive biomarkers related to immune cell infiltration in HCC (25), while a link between MMP1 and poor survival outcomes for those with advanced HCC has been documented (26). Previous studies have demonstrated that the TME in HCC is influenced by factors such as HBV, HCV, and NASH (27,28). These influences involve a complex interplay of pro-inflammatory and anti-inflammatory cytokines, components of the extracellular matrix, and various immune cell subpopulations that collectively drive malignant progression and inhibit apoptosis within the TME (28). HBV and HCV exert carcinogenic effects through two primary mechanisms (29). The direct mechanism involves viral integration into the host genome, leading to the formation of mutational hotspots. The indirect mechanism is mediated through alterations in the expression of specific RNAs and the secretion of oncogenic proteins, which together facilitate viral replication, invasion, and subsequent oncogenesis. NASH-associated HCC is characterized by the presence of immunosuppressive cells with enhanced functionality that is consistent with T cell exhaustion. Notably, the immunosuppressive features of the TME in NASH-related HCC are primarily driven by infiltrating immune cells rather than by the tumor cells themselves. Compared to NASH-related HCC, HBV- and HCV-induced HCC exhibit significantly lower levels of T cell exhaustion, which may explain their greater responsiveness to immune checkpoint blockade therapies (30). In this study, five prognostic genes associated with TRG subtypes were identified, and these genes were found to play important roles in the pathogenesis and progression of HCC. The findings derived from multi-gene analyses using bioinformatics approaches were largely consistent with conclusions from previous single-gene studies, further supporting the relevance of TRG subtype-related prognostic genes in HCC development. These results suggest that, similar to TRG subtypes, HBV, HCV, and NASH influence the TME of HCC through specific pathogenic genes and immune cells.

Prior researches have established KRT17 as a promising prognostic biomarker in some forms of cancer (31,32), but its relevance in HCC has yet to be clarified. Here, 20 paired HCC patient tissue samples were analyzed, revealing a significant difference in KRT17 levels when comparing the tumors and paracancerous tissues. In osteosarcoma cells, KRT17 reportedly favors growth through AKT/mTOR pathway activation (33), while it similarly activates AKT signaling and the epithelial-mesenchymal transformation (EMT) in esophageal cancer cells to fuel their migration and proliferation (34). In line with these published studies, the silencing of KRT17 in HCC cell lines inhibited their proliferation and migration while inducing apoptotic death in vitro through the AKT/mTOR signaling pathway. Based on these results, KRT17 appears to play an oncogenic role in HCC and may be a promising target for the treatment of this cancer type on our findings that KRT17 may function as an oncogene in HCC and may become a new therapeutic target for HCC.

There are multiple limitations to this study. For one, the large-scale analyses were dependent on publicly accessible datasets, and all analyzed samples were obtained in a retrospective manner such that case selection bias may have had some impact on these results. Large-scale prospective analyses and further in vitro and in vivo experimentation will be vital to confirm these findings. Additionally, in most of these datasets, information pertaining to relevant clinical variables including surgery, neoadjuvant chemotherapy, and chemoradiotherapy was not provided, which may have had some bearing on the prediction of HCC patient prognosis.

Conclusions

Based on these results, T cell proliferation-related molecular subtyping and associated predictive models can be effectively used to forecast outcomes for individuals with HCC. The novel candidate biomarkers identified herein were also verified through in vitro experimentations. These findings provide novel insights that warrant further investigation for the development of precision targeted therapy for patients with HCC.

Acknowledgments

We would like to thank the TCGA database for the HCC dataset and the GEO database for GSE52018 and GSE76427 datasets.

Footnote

Reporting Checklist: The authors have completed the MDAR and TRIPOD reporting checklists. Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-50/rc

Data Sharing Statement: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-50/dss

Peer Review File: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-50/prf

Funding: This study was supported by the National Natural Science Foundation of China (No. 82073947).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-50/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Ethics Committee of The First Affiliated Hospital of Wannan Medical College (No. WNMC-AWE-2023121), and informed consent was obtained from all the patients.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Yang X, Yang C, Zhang S, et al. Precision treatment in advanced hepatocellular carcinoma. Cancer Cell 2024;42:180-97. [Crossref] [PubMed]
2. Huang DQ, El-Serag HB, Loomba R. Global epidemiology of NAFLD-related HCC: trends, predictions, risk factors and prevention. Nat Rev Gastroenterol Hepatol 2021;18:223-38. [Crossref] [PubMed]
3. Rinaldi L, Vetrano E, Rinaldi B, et al. HCC and Molecular Targeting Therapies: Back to the Future. Biomedicines 2021;9:1345. [Crossref] [PubMed]
4. Huang J, Wu Q, Geller DA, et al. Macrophage metabolism, phenotype, function, and therapy in hepatocellular carcinoma (HCC). J Transl Med 2023;21:815. [Crossref] [PubMed]
5. Wang Y, Deng B. Hepatocellular carcinoma: molecular mechanism, targeted therapy, and biomarkers. Cancer Metastasis Rev 2023;42:629-52. [Crossref] [PubMed]
6. Yu B, Ma W. Biomarker discovery in hepatocellular carcinoma (HCC) for personalized treatment and enhanced prognosis. Cytokine Growth Factor Rev 2024;79:29-38. [Crossref] [PubMed]
7. Zheng J, Wang S, Xia L, et al. Hepatocellular carcinoma: signaling pathways and therapeutic advances. Signal Transduct Target Ther 2025;10:35. [Crossref] [PubMed]
8. Dey P, Kimmelman AC, DePinho RA. Metabolic Codependencies in the Tumor Microenvironment. Cancer Discov 2021;11:1067-81. [Crossref] [PubMed]
9. Zhao X, Li K, Chen M, et al. Metabolic codependencies in the tumor microenvironment and gastric cancer: Difficulties and opportunities. Biomed Pharmacother 2023;162:114601. [Crossref] [PubMed]
10. Ma L, Heinrich S, Wang L, et al. Multiregional single-cell dissection of tumor and immune cells reveals stable lock-and-key features in liver cancer. Nat Commun 2022;13:7533. [Crossref] [PubMed]
11. Leon E, Ranganathan R, Savoldo B. Adoptive T cell therapy: Boosting the immune system to fight cancer. Semin Immunol 2020;49:101437. [Crossref] [PubMed]
12. Waldman AD, Fritz JM, Lenardo MJ. A guide to cancer immunotherapy: from T cell basic science to clinical practice. Nat Rev Immunol 2020;20:651-68. [Crossref] [PubMed]
13. Legut M, Gajic Z, Guarino M, et al. A genome-scale screen for synthetic drivers of T cell proliferation. Nature 2022;603:728-35. [Crossref] [PubMed]
14. Chan S, Wang X, Wang Z, et al. CTSG Suppresses Colorectal Cancer Progression through Negative Regulation of Akt/mTOR/Bcl2 Signaling Pathway. Int J Biol Sci 2023;19:2220-33. [Crossref] [PubMed]
15. Kang JH, Zappasodi R. Modulating Treg stability to improve cancer immunotherapy. Trends Cancer 2023;9:911-27. [Crossref] [PubMed]
16. Xue R, Zhang Q, Cao Q, et al. Liver tumour immune microenvironment subtypes and neutrophil heterogeneity. Nature 2022;612:141-7. [Crossref] [PubMed]
17. Samstein RM, Arvey A, Josefowicz SZ, et al. Foxp3 exploits a pre-existent enhancer landscape for regulatory T cell lineage specification. Cell 2012;151:153-66. [Crossref] [PubMed]
18. Wu SY, Fu T, Jiang YZ, et al. Natural killer cells in cancer biology and therapy. Mol Cancer 2020;19:120. [Crossref] [PubMed]
19. Chu X, Tian Y, Lv C. Decoding the spatiotemporal heterogeneity of tumor-associated macrophages. Mol Cancer 2024;23:150. [Crossref] [PubMed]
20. Zhou C, Weng J, Liu C, et al. Disruption of SLFN11 Deficiency-Induced CCL2 Signaling and Macrophage M2 Polarization Potentiates Anti-PD-1 Therapy Efficacy in Hepatocellular Carcinoma. Gastroenterology 2023;164:1261-78. [Crossref] [PubMed]
21. Luo X, Huang W, Li S, et al. SOX12 Facilitates Hepatocellular Carcinoma Progression and Metastasis through Promoting Regulatory T-Cells Infiltration and Immunosuppression. Adv Sci (Weinh) 2024;11:e2310304. [Crossref] [PubMed]
22. Sun J, Xie T, Jamal M, et al. CLEC3B as a potential diagnostic and prognostic biomarker in lung cancer and association with the immune microenvironment. Cancer Cell Int 2020;20:106. [Crossref] [PubMed]
23. Lee H, Zhuang L, Gan B. Rewiring cancer cell death: LPCAT1 shapes lipid composition and ferroptosis resistance. Cell Death Differ 2024;31:1101-3. [Crossref] [PubMed]
24. Liu Q, Zhang X, Qi J, et al. Comprehensive profiling of lipid metabolic reprogramming expands precision medicine for HCC. Hepatology 2025;81:1164-80. [Crossref] [PubMed]
25. Xie XW, Jiang SS, Li X. CLEC3B as a Potential Prognostic Biomarker in Hepatocellular Carcinoma. Front Mol Biosci 2020;7:614034. [Crossref] [PubMed]
26. Liu H, Lan T, Li H, et al. Circular RNA circDLC1 inhibits MMP1-mediated liver cancer progression via interaction with HuR. Theranostics 2021;11:1396-411. [Crossref] [PubMed]
27. Koshy A. Evolving Global Etiology of Hepatocellular Carcinoma (HCC): Insights and Trends for 2024. J Clin Exp Hepatol 2025;15:102406. [Crossref] [PubMed]
28. Borgia M, Dal Bo M, Toffoli G. Role of Virus-Related Chronic Inflammation and Mechanisms of Cancer Immune-Suppression in Pathogenesis and Progression of Hepatocellular Carcinoma. Cancers (Basel) 2021;13:4387. [Crossref] [PubMed]
29. Zheng S, Chan SW, Liu F, et al. Hepatocellular Carcinoma: Current Drug Therapeutic Status, Advances and Challenges. Cancers (Basel) 2024;16:1582. [Crossref] [PubMed]
30. Li M, Wang L, Cong L, et al. Spatial proteomics of immune microenvironment in nonalcoholic steatohepatitis-associated hepatocellular carcinoma. Hepatology 2024;79:560-74. [Crossref] [PubMed]
31. Liang W, Liu H, Zeng Z, et al. KRT17 Promotes T-lymphocyte Infiltration Through the YTHDF2-CXCL10 Axis in Colorectal Cancer. Cancer Immunol Res 2023;11:875-94. [Crossref] [PubMed]
32. Tang S, Liu W, Yong L, et al. Reduced Expression of KRT17 Predicts Poor Prognosis in HER2(high) Breast Cancer. Biomolecules 2022;12:1183. [Crossref] [PubMed]
33. Quinn JJ, Jones MG, Okimoto RA, et al. Single-cell lineages reveal the rates, routes, and drivers of metastasis in cancer xenografts. Science 2021;371:eabc1944.
34. Li C, Teng Y, Wu J, et al. A pan-cancer analysis of the oncogenic role of Keratin 17 (KRT17) in human tumors. Transl Cancer Res 2021;10:4489-501. [Crossref] [PubMed]