PSMD4 promotes malignant phenotypes and is associated with angiogenesis-related signaling and immune remodeling in hepatocellular carcinoma Authors: Qiuyi Wang, Hanxuan Yang, Yanan Pang, Chanyan Zhu Published: 2026 ABSTRACT BackgroundProteasome 26 S non-ATPase subunit 4 (PSMD4) is implicated in various cancers, but its role in hepatocellular carcinoma (HCC) is unclear. This study investigates PSMD4 expression, prognostic value, and functions in HCC progression, angiogenesis, and immune regulation.MethodsTCGA-LIHC was used as the training cohort and ICGC-LIRI as an external validation cohort to assess PSMD4 expression and prognostic value. Human Protein Atlas (HPA) immunohistochemistry data were used to evaluate PSMD4 protein expression. Survival, ROC, nomogram, functional enrichment, immune infiltration, single-cell, and cell–cell communication analyses were performed. Single-cell RNA-seq data from GSE146409 were used to assess cell-type-specific PSMD4 expression and intercellular communication patterns. Spatial transcriptomics data from HRA000437 were used to visualize the spatial distribution of PSMD4 expression in HCC tissues. The biological role of PSMD4 was validated in Huh7 and HepG2 cells using qRT-PCR, CCK-8, colony formation, wound-healing, and western blot assays.ResultsPSMD4 was significantly upregulated in HCC in both TCGA-LIHC and ICGC-LIRI cohorts, with positive protein expression confirmed by HPA immunohistochemistry. High PSMD4 expression was associated with poorer overall survival and showed moderate predictive performance for survival outcomes. Enrichment analyses linked PSMD4 to cell-cycle progression, DNA replication, genomic instability, and angiogenesis-related pathways. Single-cell analysis indicated that PSMD4 was mainly enriched in malignant hepatocytes and SPP1⁺ tumor-associated fibroblasts, while spatial transcriptomics analysis showed that PSMD4 expression was concentrated in tumor-dense regions. PSMD4 knockdown inhibited HCC cell proliferation, colony formation, and migration, and reduced VEGFA, RAF1, and MEK1/2 expression. Moreover, high PSMD4 expression was associated with decreased antitumor immune infiltration and suppression of immune activation-related genes.ConclusionPSMD4 is upregulated in HCC and is associated with unfavorable prognosis. Functional assays suggest that PSMD4 contributes to malignant phenotypes and angiogenesis-related signaling in HCC cells, while computational immune analyses indicate an association between high PSMD4 expression and an immunosuppressive immune landscape. These findings suggest that PSMD4 may serve as a potential prognostic biomarker and candidate therapeutic target in HCC.Supplementary InformationThe online version contains supplementary material available at 10.1007/s12672-026-05405-2. Introduction Hepatocellular carcinoma (HCC) is a predominant form of primary liver cancer and a major global cause of cancer-related deaths. Its incidence continues to rise, particularly in regions with high prevalence of hepatitis B and C infections [1, 2]. Despite advances in surgical resection, locoregional therapies, and targeted agents, the overall prognosis for HCC remains dismal, driven by frequent late-stage detection, high postoperative recurrence, and suboptimal responses to available therapies [3, 4]. Alpha-fetoprotein (AFP), the most widely used serum biomarker for HCC, lacks sufficient sensitivity and specificity, particularly in early-stage or AFP-negative cases, thereby limiting its utility for early detection and risk stratification [1, 2]. Emerging evidence indicates that angiogenesis and immune evasion are critical drivers of tumor growth and metastasis, as well as key contributors to the limited efficacy of current immunotherapies and anti-angiogenic treatments [5–8]. Consequently, research efforts have intensified to identify novel molecular mechanisms and biomarkers—encompassing coding and non-coding RNAs, immune landscapes, and multi-omics models—aimed at enhancing clinical decision-making in HCC [4, 9–13]. However, significant challenges persist, and long-term survival rates for HCC patients remain unsatisfactory. Proteasome 26 S non-ATPase subunit 4 (PSMD4), also known as Rpn10 or S5A, is an essential ubiquitin receptor within the 26 S proteasome complex, playing a central role in substrate recognition and targeting protein degradation. Previous studies have demonstrated that PSMD4 regulates tumor proliferation, metastasis, and drug resistance in various solid tumors, such as breast and lung cancers [14–16]. However, its expression profile, clinical relevance, and functional mechanisms in HCC remain largely unexplored and warrant a systematic investigation. To date, no study has comprehensively analyzed PSMD4 expression and its prognostic relevance in HCC using large-scale, multi-dimensional datasets [17]. Moreover, the potential involvement of PSMD4 in multiple hallmark processes—such as angiogenesis, immune regulation, and tumor microenvironment remodeling—remains unexplored in HCC [18–22]. In contrast to the established immunoregulatory roles of other proteasomal subunits, the potential involvement of PSMD4 in shaping the immune landscape of HCC is largely unexplored. This study systematically evaluated the expression landscape and clinical relevance of PSMD4 in HCC through integrative multi-omics analyses. Transcriptomic datasets from TCGA-LIHC and ICGC-LIRI, together with protein expression data from the Human Protein Atlas (HPA), were used to assess PSMD4 expression at both RNA and protein levels. The prognostic relevance of PSMD4 was evaluated using survival, ROC, and nomogram analyses. Functional enrichment, single-cell transcriptomic, cell–cell communication, and immune infiltration analyses were performed to explore biological processes and tumor microenvironmental features associated with PSMD4 expression. In vitro assays, including CCK-8, colony formation, wound-healing, and western blot analyses, were conducted to examine the functional effects of PSMD4 knockdown in HCC cells. Collectively, this study suggests that PSMD4 is an upregulated and prognostically relevant molecule in HCC. Functional assays support its contribution to malignant cellular phenotypes, whereas computational analyses indicate associations between high PSMD4 expression, angiogenesis-related signaling, and an immunosuppressive immune landscape. Materials and methods Data acquisition and preprocessing RNA-seq expression profiles and corresponding clinical information of HCC were obtained from The Cancer Genome Atlas Liver Hepatocellular Carcinoma cohort (TCGA-LIHC, https://portal.gdc.cancer.gov/projects/TCGA-LIHC) through the Genomic Data Commons (GDC) Data Portal and used as the training cohort. To externally validate the expression pattern and prognostic relevance of PSMD4, transcriptomic and clinical data from the International Cancer Genome Consortium Liver Cancer-RIKEN, Japan cohort (ICGC-LIRI, https://dcc.icgc.org/projects/LIRI-JP) were collected and used as an independent validation cohort. For single-cell analysis, preprocessed and cell-type-annotated scRNA-seq data(GSE146409) of HCC were retrieved from the TISCH database (Tumor Immune Single-cell Hub, http://tisch.comp-genomics.org/), which provides standardized expression matrices with completed quality control, batch correction, dimensionality reduction, clustering, and cell type annotation. For spatial expression analysis, spatial transcriptomics data (HRA000437) were downloaded from the Genome Sequence Archive for Human (GSA-Human) at the National Genomics Data Center (NGDC, https://ngdc.cncb.ac.cn/gsa-human/browse/HRA000437). These spatial data were derived from a published study and had undergone preprocessing, spatial clustering, and regional annotation; they were used in this study solely to visualize the spatial distribution pattern of PSMD4. To validate PSMD4 expression at the protein level, immunohistochemistry (IHC) staining images of PSMD4 in liver cancer tissues were obtained from the Human Protein Atlas (HPA, https://www.proteinatlas.org/ENSG00000159352-PSMD4/cancer/liver+cancer#img). Patients with incomplete survival information were excluded from survival analyses. For downstream analyses, gene expression data were normalized and transformed into log2-transformed expression values when appropriate. PSMD4 expression levels were extracted from each cohort, and patients were stratified into high- and low-PSMD4 expression groups according to the median PSMD4 expression value within each cohort. Differential expression, survival, and prognostic analyses were subsequently performed based on these groupings. Differential gene expression analysis and diagnostic value assessment Gene expression differences between malignant and non-malignant liver tissues were assessed using empirical Bayes moderation implemented in the limma framework. Significant transcripts were visualized through volcano plots and ranked by fold-change magnitude. Additional analysis of the PSMD gene family was conducted, with genes ranked by log₂ fold change and p-values. To quantify the diagnostic relevance of PSMD4, ROC curves were generated and AUC values computed via the “pROC” package in R. Multiple comparisons were corrected using the Benjamini-Hochberg (BH) method. Survival and prognostic analysis Kaplan–Meier survival curves were generated to examine the association between PSMD4 expression and overall survival (OS), with statistical significance assessed via the log-rank test. Patients were divided into risk categories based on a composite risk score derived from gene expression profiles and survival outcomes. Risk stratification and survival distribution were illustrated using corresponding plots. To evaluate the independent prognostic value of PSMD4, both univariate and multivariate Cox regression models were applied. A prognostic nomogram incorporating significant covariates was developed to estimate 1-, 3-, and 5-year survival probabilities. Model performance was assessed using calibration curves to compare predicted and observed survival outcomes. Functional enrichment analysis and mutation landscape To investigate the biological processes linked to elevated PSMD4 expression, GSEA was implemented using the “clusterProfiler” package in R. Significant enrichment of GO terms and KEGG pathways was determined based on adjusted p-values (p.adjust < 0.05). Moreover, genomic alterations were assessed by comparing somatic mutation profiles and copy number variations (CNVs) between PSMD4 high- and low-expression groups using TCGA mutation datasets. CNV amplifications and deletions were visualized through bar graphs and heatmaps, revealing genomic alterations associated with PSMD4 expression. Cell–cell communication and single-cell expression analysis To investigate the role of PSMD4 within the tumor microenvironment, intercellular communication analysis was performed using the CellChat framework. The analysis focused on interactions among ligands, receptors, and transcription factors. Differential ligand–receptor networks were constructed and visualized, and Sankey diagrams were used to illustrate pathway flow dynamics. In addition, single-cell RNA sequencing data were used to assess the cell-type-specific expression patterns of PSMD4 across hepatocytes, immune cells, fibroblasts, and cholangiocytes. Spatial transcriptomics data were further integrated to map the tissue-level distribution of PSMD4 expression. In vitro functional assays The Huh7 and HepG2 cell lines used in this study were purchased from Ruibite Biotechnology (Shanghai) Co., Ltd. Huh7 and HepG2 cells were cultured under standard conditions and transduced with lentiviral vectors carrying either shRNA targeting PSMD4 (shPSMD4) or a negative control (shNC). Stable knockdown cell lines were selected using puromycin, and knockdown efficiency was verified by Western blotting. Cell proliferation was assessed at multiple time points using the CCK-8 assay. Colony formation ability was evaluated by seeding cells at low density, staining colonies with crystal violet, and counting the number of colonies formed. Wound healing assays were performed to assess migratory capacity, with scratch closure monitored at 0, 24, and 48 h. Finally, Western blotting was performed to evaluate the protein levels of PSMD4 and key angiogenesis-related markers, including VEGFA, RAF1, and MEK1/2, with GAPDH serving as the loading control. Immune infiltration and immune landscape analysis To characterize the immune contexture linked to PSMD4 expression, single-sample Gene Set Enrichment Analysis (ssGSEA) was utilized to estimate the infiltration levels of 28 distinct immune cell subsets. Comparative analysis was then conducted to identify differences in immune cell enrichment between PSMD4 high- and low-expression groups. To further investigate the relationship between PSMD4 and immune-related factors, heatmaps were generated across multiple LIHC datasets to analyze the correlation between PSMD4 expression and genes involved in antigen presentation (e.g., the HLA family), immune stimulatory molecules (e.g., CD86, TNFRSFs), chemokines (e.g., CXCL9, CXCL10), and their receptors (e.g., CXCR3, CCR5). Statistical analysis Statistical analyses of experimental data were performed using GraphPad Prism (v9.5), with results expressed as mean ± standard deviation. All computational and bioinformatics analyses were carried out in R (v4.5.0). Kaplan–Meier estimators and log-rank tests were applied for survival analysis, while Spearman’s rank correlation was used to assess pairwise associations. Results PSMD4 is upregulated in hepatocellular carcinoma and validated in independent cohorts To characterize the expression profile of PSMD4 in HCC, differential expression analysis was first performed using the TCGA-LIHC training cohort. The volcano plot showed that PSMD4 was located within the significantly upregulated gene region, indicating increased PSMD4 expression in HCC tissues compared with normal liver tissues (Fig. 1A). Consistently, analysis of the ICGC-LIRI validation cohort confirmed that PSMD4 was also positioned among the upregulated genes, supporting the reproducibility of PSMD4 overexpression across independent HCC datasets (Fig. 1B). We next compared the differential expression patterns of PSMD family members in the two cohorts. In the TCGA-LIHC cohort, PSMD4 exhibited the most prominent upregulation among PSMD family genes (Fig. 1C). A similar expression pattern was observed in the ICGC-LIRI cohort, where PSMD4 again showed the highest degree of upregulation (Fig. 1E). Ranked distribution plots of differentially expressed genes further demonstrated that PSMD4 was consistently located within the upregulated region in both cohorts (Fig. 1D, F), suggesting that PSMD4 represents a markedly elevated PSMD family member in HCC. To further assess PSMD4 expression at the protein level, immunohistochemistry images from the Human Protein Atlas (HPA) database were examined. Representative staining images showed positive PSMD4 protein expression in liver cancer tissues, with staining mainly observed in the cytoplasm, membrane, and/or nucleus of tumor cells (Fig. 1G, H). Collectively, these transcriptomic and protein-level findings indicate that PSMD4 is consistently upregulated in HCC, suggesting its potential involvement in HCC tumorigenesis. Fig. 1Upregulation of PSMD4 in HCC. A, B Volcano plots of differentially expressed genes in the TCGA-LIHC training cohort and ICGC-LIRI validation cohort, with PSMD4 highlighted; C, E Differential expression of PSMD family members in the two cohorts; D, F Ranked distribution plots of differentially expressed genes; G, H Representative HPA immunohistochemistry images of PSMD4 protein expression in liver cancer tissues Upregulation of PSMD4 in HCC. A, B Volcano plots of differentially expressed genes in the TCGA-LIHC training cohort and ICGC-LIRI validation cohort, with PSMD4 highlighted; C, E Differential expression of PSMD family members in the two cohorts; D, F Ranked distribution plots of differentially expressed genes; G, H Representative HPA immunohistochemistry images of PSMD4 protein expression in liver cancer tissues High PSMD4 expression is associated with poor prognosis in HCC To evaluate the prognostic significance of PSMD4 in HCC, patients were stratified into high- and low-risk groups in the TCGA-LIHC training cohort and the ICGC-LIRI validation cohort. In the TCGA-LIHC cohort, the proportion of deceased patients was higher in the high-risk group than in the low-risk group, increasing from 30% to 41% (Fig. 2A). A similar trend was observed in the ICGC-LIRI cohort, where the death rate increased from 6% in the low-risk group to 19% in the high-risk group (Fig. 2D). Risk score distribution analysis showed that patients were progressively ordered from low to high risk in both cohorts (Fig. 2B, E). Survival status plots further indicated that deceased cases tended to accumulate in the high-risk region, suggesting that elevated PSMD4 expression may be associated with unfavorable clinical outcomes (Fig. 2C, F). Kaplan–Meier survival analysis demonstrated that patients with high PSMD4 expression had significantly poorer overall survival than those with low PSMD4 expression in the TCGA-LIHC cohort (Fig. 2G). This prognostic association was further validated in the ICGC-LIRI cohort, where high PSMD4 expression was also associated with reduced overall survival (Fig. 2J). Time-dependent ROC analysis was performed to assess the predictive performance of PSMD4 for patient survival. In the TCGA-LIHC cohort, the AUC values for predicting 1-, 2-, and 4-year survival were 0.640, 0.628, and 0.592, respectively (Fig. 2H). In the ICGC-LIRI cohort, the corresponding AUC values were 0.626, 0.640, and 0.651, indicating that PSMD4 had moderate prognostic predictive ability (Fig. 2K). Furthermore, nomogram models integrating PSMD4 expression with clinicopathological variables were constructed to estimate 1-, 3-, and 5-year survival probabilities in both cohorts (Fig. 2I, L). Collectively, these findings indicate that high PSMD4 expression is associated with poor prognosis and may serve as a potential survival prediction marker in HCC. Fig. 2Prognostic value and survival prediction performance of PSMD4 in HCC. A–C Survival status, risk score, and survival time distribution in the TCGA-LIHC cohort; D–F Survival status, risk score, and survival time distribution in the ICGC-LIRI cohort; G, J Kaplan–Meier survival curves. H, K Time-dependent ROC curves for overall survival prediction; I, L Nomograms for predicting 1-, 3-, and 5-year survival probability Prognostic value and survival prediction performance of PSMD4 in HCC. A–C Survival status, risk score, and survival time distribution in the TCGA-LIHC cohort; D–F Survival status, risk score, and survival time distribution in the ICGC-LIRI cohort; G, J Kaplan–Meier survival curves. H, K Time-dependent ROC curves for overall survival prediction; I, L Nomograms for predicting 1-, 3-, and 5-year survival probability PSMD4 expression is associated with prognosis-related pathways and genomic alterations in HCC To uncover the biological mechanisms through which PSMD4 may affect HCC prognosis, we conducted GSEA based on genes differentially expressed between patients with high and low PSMD4 expression. The results indicated that elevated PSMD4 levels were strongly linked to pathways involving DNA replication, cell cycle control, and genomic instability, including the G2/M checkpoint and mismatch repair (Fig. 3A). Additional GO and KEGG enrichment analyses revealed that PSMD4-related genes participated in diverse biological functions such as retinol metabolism, spindle organization, and chromosomal segregation, and were particularly overrepresented in pathways associated with cell cycle progression and glycolysis (Fig. 3B). We further examined the genomic landscape associated with PSMD4 upregulation. Comparative analyses of somatic mutations and copy number alterations demonstrated that the high-expression group exhibited higher mutation frequencies in critical oncogenes such as TP53 and CTNNB1, alongside widespread CNV amplifications and deletions (Fig. 3C). These data suggest that high PSMD4 expression is associated with genomic instability-related features and malignant progression-related pathways in HCC. Fig. 3Association of PSMD4 expression with functional enrichment and genomic alterations in HCC. A GSEA of differentially expressed genes between high- and low-PSMD4 expression groups; B GO and KEGG enrichment analyses of PSMD4-associated genes; C Somatic mutation and copy number variation profiles stratified by PSMD4 expression Association of PSMD4 expression with functional enrichment and genomic alterations in HCC. A GSEA of differentially expressed genes between high- and low-PSMD4 expression groups; B GO and KEGG enrichment analyses of PSMD4-associated genes; C Somatic mutation and copy number variation profiles stratified by PSMD4 expression PSMD4 knockdown suppresses malignant phenotypes of hepatocellular carcinoma cells To investigate the functional role of PSMD4 in hepatocellular carcinoma cells, PSMD4 was knocked down in Huh7 and HepG2 cells. Wound-healing assays showed that PSMD4 knockdown markedly reduced the wound closure rate at 24 and 48 h compared with the control group, indicating impaired migratory ability in both Huh7 and HepG2 cells (Fig. 4A, B). Colony formation assays further demonstrated that PSMD4 knockdown significantly decreased the number of colonies formed by Huh7 and HepG2 cells, suggesting that PSMD4 contributes to the clonogenic capacity of hepatocellular carcinoma cells (Fig. 4C, D). Consistently, CCK-8 assays showed that PSMD4 knockdown suppressed cell growth over time, with lower OD values observed in the shPSMD4 group than in the shNC group, particularly at later time points (Fig. 4E, F). PSMD4 knockdown efficiency was then validated at both mRNA and protein levels. qRT-PCR analysis showed that PSMD4 mRNA expression was significantly reduced in the shPSMD4 group compared with the shNC group in both Huh7 and HepG2 cells (Fig. 4G). Consistently, western blot analysis and densitometric quantification based on three independent biological replicates confirmed decreased PSMD4 protein expression after PSMD4 knockdown (Fig. 4H, I). Moreover, PSMD4 knockdown reduced the protein expression levels of VEGFA, RAF1, and MEK1/2, suggesting a potential association between PSMD4 and angiogenesis-related RAF/MEK signaling in HCC cells (Fig. 4H, I). Collectively, these results indicate that PSMD4 knockdown suppresses the proliferation, migration, and colony-forming ability of HCC cells, accompanied by reduced VEGFA, RAF1, and MEK1/2 protein expression. Fig. 4Effects of PSMD4 knockdown on malignant phenotypes and angiogenesis-related protein expression in HCC cells. A, B Wound-healing assays in Huh7 and HepG2 cells; C, D Colony formation assays; E, F CCK-8 proliferation assays; G qRT-PCR analysis of PSMD4 mRNA expression; H, I Representative western blot images and densitometric quantification of PSMD4, VEGFA, RAF1, and MEK1/2 protein levels. Data are presented as mean ± SD from three independent biological replicates. Relative mRNA expression was calculated using the 2−ΔΔCt method. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 Effects of PSMD4 knockdown on malignant phenotypes and angiogenesis-related protein expression in HCC cells. A, B Wound-healing assays in Huh7 and HepG2 cells; C, D Colony formation assays; E, F CCK-8 proliferation assays; G qRT-PCR analysis of PSMD4 mRNA expression; H, I Representative western blot images and densitometric quantification of PSMD4, VEGFA, RAF1, and MEK1/2 protein levels. Data are presented as mean ± SD from three independent biological replicates. Relative mRNA expression was calculated using the 2−ΔΔCt method. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 PSMD4 expression is associated with pro-angiogenic intercellular communication in the HCC microenvironment To investigate the cellular context and functional relevance of PSMD4 within the tumor microenvironment (TME), we analyzed intercellular communication patterns based on single-cell transcriptomic data. Heatmap analysis revealed that multiple ligand–receptor interactions were enriched among hepatocellular carcinoma cells, fibroblasts, and immune cell subsets, involving key signaling molecules such as VEGFA, EDN1, and EREG (Fig. 5A). Subsequent Sankey diagram analysis identified several ligand–receptor–transcription factor signaling axes—most notably VEGFA–EPHB2–RELA and EDN1–EDNRB–STAT1—which are widely recognized as critical mediators of angiogenesis (Fig. 5B, C). These pathways exhibited heightened activity in PSMD4 high-expressing cell clusters. Cell-type-specific expression analysis further showed that PSMD4 was primarily enriched in malignant hepatocytes and SPP1⁺ tumor-associated fibroblasts, both of which are known to contribute to angiogenesis and tumor progression (Fig. 5D). Spatial transcriptomics data additionally confirmed that PSMD4 expression was highly concentrated in tumor-dense regions, suggesting a potential association with localized intercellular communication within the TME. Fig. 5PSMD4-associated intercellular communication patterns in the HCC microenvironment. A Heatmap of ligand–receptor interaction scores among major cell types; B, C Sankey diagrams showing ligand–receptor–transcription factor signaling axes; D Violin plot showing PSMD4 expression across cell types PSMD4-associated intercellular communication patterns in the HCC microenvironment. A Heatmap of ligand–receptor interaction scores among major cell types; B, C Sankey diagrams showing ligand–receptor–transcription factor signaling axes; D Violin plot showing PSMD4 expression across cell types High PSMD4 expression is associated with reduced antitumor immune infiltration and immune activation-related gene expression in HCC To explore the relationship between PSMD4 expression and the immune landscape of HCC, we performed computational immune infiltration and immune-related gene expression analyses. High PSMD4 expression was associated with lower estimated infiltration levels of several antitumor immune cell populations, including dendritic cells, B cells, NK cells, and CD8⁺ T cells (Fig. 6A). In addition, multiple immune activation-related genes showed reduced expression in the high-PSMD4 group, including antigen-presentation genes such as HLA-A, HLA-B, and TAP1 (Fig. 6B), costimulatory molecules such as CD86, IL2RA, and members of the TNFRSF family (Fig. 6C), chemokines such as CXCL9, CXCL10, and CCL5 (Fig. 6D), and chemokine receptors such as CXCR3 and CCR5 (Fig. 6E). These findings suggest that high PSMD4 expression is associated with an immunologically suppressed tumor microenvironment characterized by reduced antitumor immune infiltration and decreased immune activation-related gene expression. Fig. 6High PSMD4 expression is associated with an immunosuppressive tumor microenvironment in HCC. A Estimated infiltration levels of antitumor immune cell populations stratified by PSMD4 expression; B–E Expression of immune-related genes in the high- and low-PSMD4 expression groups, including antigen-presentation genes, costimulatory molecules, chemokines, and chemokine receptors High PSMD4 expression is associated with an immunosuppressive tumor microenvironment in HCC. A Estimated infiltration levels of antitumor immune cell populations stratified by PSMD4 expression; B–E Expression of immune-related genes in the high- and low-PSMD4 expression groups, including antigen-presentation genes, costimulatory molecules, chemokines, and chemokine receptors Discussion This study systematically characterized the expression pattern, prognostic relevance, biological function, and tumor microenvironmental associations of PSMD4 in HCC. By integrating transcriptomic data from the TCGA-LIHC training cohort and the ICGC-LIRI validation cohort, we found that PSMD4 was consistently upregulated in HCC tissues. This finding was further supported at the protein level by immunohistochemistry data from the HPA. Clinically, high PSMD4 expression was associated with poorer overall survival and showed moderate predictive performance for survival outcomes in both cohorts. Functionally, PSMD4 knockdown suppressed HCC cell proliferation, colony formation, and migration, accompanied by reduced expression of VEGFA, RAF1, and MEK1/2. In addition, immune landscape analyses suggested that high PSMD4 expression was associated with decreased antitumor immune infiltration and reduced expression of immune activation-related genes. Collectively, these findings suggest that PSMD4 may contribute to HCC progression through its association with malignant cellular phenotypes, angiogenesis-related signaling, and immune remodeling. The consistent upregulation of PSMD4 across independent cohorts strengthens the robustness of its association with HCC. As a ubiquitin receptor of the 26 S proteasome complex, PSMD4 participates in substrate recognition during ubiquitin-mediated protein degradation, a process closely linked to cell-cycle control, stress responses, protein homeostasis, and tumor progression. Although several proteasome-related components have been implicated in cancer biology, the role of PSMD4 in HCC has not been fully defined. In the present study, PSMD4 was not only elevated in HCC tissues but also associated with unfavorable survival outcomes, supporting its potential value as a prognostic biomarker. However, because the current clinical evidence is mainly derived from public datasets and retrospective analyses, the clinical utility of PSMD4 requires further validation in larger prospective cohorts and independent clinical specimens. Previous studies have increasingly implicated proteasome regulatory subunits in tumor progression, prognosis, and treatment resistance [23–25]. For example, PSMD14/POH1, a deubiquitinating subunit of the 19 S regulatory particle, has been reported to regulate cancer-related signaling by modulating the stability or activity of oncogenic substrates [26, 27]. Other PSMD family members have also been associated with malignant progression and unfavorable prognosis in different cancer types [24, 28]. These findings support the broader concept that dysregulated proteasome regulatory components may contribute to tumor biology beyond their canonical roles in bulk protein degradation. Compared with these PSMD family members, PSMD4 has been less extensively investigated in HCC. Therefore, our study expands the current understanding of proteasome regulatory subunits by identifying PSMD4 as a prognostically relevant molecule associated with malignant phenotypes, angiogenesis-related signaling, and immune remodeling in HCC. Mechanistically, our enrichment analyses indicated that PSMD4-associated genes were involved in cell-cycle regulation, DNA replication, mismatch repair, chromosomal segregation, and glycolysis-related pathways. These biological processes are closely related to uncontrolled proliferation and genomic instability, two fundamental features of HCC progression. Consistent with these bioinformatic findings, in vitro knockdown of PSMD4 significantly inhibited proliferation, clonogenic growth, and migration in Huh7 and HepG2 cells. Moreover, PSMD4 knockdown reduced the expression of VEGFA, RAF1, and MEK1/2, suggesting a potential link between PSMD4 and angiogenesis-related RAF/MEK signaling. This observation is biologically plausible because the ubiquitin–proteasome system is known to fine-tune angiogenic signaling by controlling the stability and activity of key pro-angiogenic regulators, including VEGF-related signaling components [29–31]. Given that PSMD4 functions as a ubiquitin receptor within the 26 S proteasome, we propose a working model in which elevated PSMD4 may affect the turnover of upstream regulatory proteins involved in oncogenic signaling, thereby indirectly sustaining RAF/MEK pathway activity and VEGFA expression. However, our current data do not establish direct proteasome-dependent regulation of VEGFA, RAF1, or MEK1/2 by PSMD4. Further rescue experiments, pathway inhibitor assays, ubiquitination analyses, and proteasome inhibition experiments are required to define the direct substrates and mechanistic basis of PSMD4-associated signaling in HCC. In addition to its association with tumor-intrinsic malignant phenotypes, PSMD4 was also related to immune remodeling in the HCC microenvironment. High PSMD4 expression was associated with lower estimated infiltration levels of several antitumor immune cell populations, including CD8⁺ T cells, NK cells, B cells, and dendritic cells. Meanwhile, genes involved in antigen presentation, immune costimulation, chemokine signaling, and immune cell recruitment, such as HLA family genes, CD86, CXCL9, CXCL10, CCL5, CXCR3, and CCR5, were downregulated in the high-PSMD4 group. These findings suggest that PSMD4 may be linked to an immunologically “cold” tumor microenvironment characterized by impaired immune recognition and weakened immune cell recruitment. However, these immune-related findings are based on computational deconvolution and correlation analyses rather than direct experimental validation. Therefore, future studies using flow cytometry, tumor–immune cell co-culture systems, and IHC staining for immune markers such as CD8, CD4, FOXP3, CD68, or CD163 in clinical HCC specimens are required to determine whether PSMD4 directly contributes to immune suppression in HCC. These immune-related findings may also have potential implications for immunotherapy response in HCC. Current systemic treatment strategies for advanced HCC increasingly rely on immune checkpoint blockade, often in combination with anti-angiogenic therapy [32–34]. The clinical benefit of atezolizumab plus bevacizumab in unresectable HCC highlights the therapeutic relevance of the interaction between angiogenesis and antitumor immunity [32, 33, 35, 36]. In this context, the association of high PSMD4 expression with VEGFA-related signaling and reduced antitumor immune infiltration suggests that PSMD4 may help identify tumors with angiogenesis-associated and immune-suppressed features. However, this hypothesis remains exploratory. Future studies should evaluate whether PSMD4 expression is associated with response or resistance to immune checkpoint inhibitors, anti-VEGF therapy, or combination regimens in clinically annotated HCC cohorts. Several limitations should be acknowledged. First, although TCGA-LIHC and ICGC-LIRI were used as training and external validation cohorts, respectively, the clinical analyses were still based on retrospective public datasets. Additional validation in larger multicenter cohorts, GEO datasets, and paired clinical HCC specimens is needed. Second, the functional experiments were performed in Huh7 and HepG2 cells, and further validation in additional HCC cell lines, patient-derived models, and in vivo tumor models would strengthen the causal evidence. Third, although PSMD4 knockdown reduced VEGFA, RAF1, and MEK1/2 expression, rescue experiments, pathway inhibitor assays, ubiquitination analyses, and proteasome inhibition experiments were not performed; therefore, the direct mechanistic link between PSMD4 and angiogenesis-related signaling remains unclear. Fourth, the immune-related findings were derived from computational deconvolution and correlation analyses and were not experimentally validated by flow cytometry, tumor–immune cell co-culture assays, or IHC staining of immune markers in clinical specimens. Finally, although HPA data supported PSMD4 protein expression in liver cancer tissues, independent IHC validation of PSMD4, Ki67, CD31, and CD8 in a clinical HCC cohort is still required. Future studies incorporating animal models, clinical tissue validation, and immunotherapy-response cohorts will be important to strengthen the translational relevance of PSMD4 in HCC. In summary, our study identifies PSMD4 as an upregulated and prognostically relevant molecule in HCC. Functional assays suggest that PSMD4 contributes to malignant cellular phenotypes, while bioinformatic analyses indicate that high PSMD4 expression is associated with angiogenesis-related signaling and an immunosuppressive immune landscape. These findings support the potential value of PSMD4 as a prognostic biomarker and candidate therapeutic target, while highlighting the need for further mechanistic and translational validation. Electronic Supplementary Material Below is the link to the electronic supplementary material. Supplementary Material 1. Supplementary Material 1.