Multi-omics analysis of NEDD1 in hepatocellular carcinoma: biological function, prognostic value, and clinical significance

Abstract

The neural precursor cell expressed developmentally down-regulated protein 1 (NEDD1) is implicated in tumorigenesis, but its role in hepatocellular carcinoma (HCC) remains unclear. This study aims to explore the oncogenic role, regulatory mechanisms, and tumor microenvironment interactions of NEDD1 in HCC. Multi-omics analyses were performed using public datasets (TCGA, GEO) and in-house clinical samples. These included expression and survival analysis, epigenetic (DNA methylation) and post-translational (phosphorylation) profiling, functional pathway enrichment, and drug sensitivity prediction. Functional validation was conducted via NEDD1 knockdown in HCC cells and a subcutaneous xenograft model. The co-expression and spatial distribution of NEDD1 and its predicted partner MZT2B were investigated using single-cell (GSE140228) and spatial transcriptomic (HRA000437) datasets. NEDD1 was significantly overexpressed in HCC tissues and correlated with poor prognosis. Its overexpression was potentially linked to promoter hypomethylation and aberrant phosphorylation. NEDD1 knockdown suppressed HCC cell proliferation, migration, and tumor growth in vivo. Notably, NEDD1 expression positively correlated with immune checkpoint molecules (PD-1, CTLA-4), and low NEDD1 expression was associated with better predicted response to immunotherapy. Single-cell and spatial transcriptomics revealed that NEDD1 and MZT2B co-expression was highly enriched in specific macrophage subsets (e.g., APOE+) and exhibited cell context-dependent heterogeneity, suggesting they may constitute a dynamic functional module within the HCC microenvironment. This multi-omics study suggests NEDD1 as a potential prognostic biomarker and therapeutic target in HCC. We propose a novel model wherein the NEDD1-MZT2B module may operate in both tumor cells and immunosuppressive macrophages, potentially influencing disease progression and immunotherapy response.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-42505-z.

Introduction

Hepatocellular carcinoma (HCC) is the third leading cause of cancer-related fatalities, following lung and colorectal cancers, and is among the most prevalent cancers globally. Some HCC patients may incidentally discover liver masses, often detected during cross-sectional imaging for late-stage HCC due to symptoms such as abdominal pain, weight loss, or liver function deterioration. It is estimated that this incidental diagnosis occurs in approximately 50% of cases worldwide, especially in developing countries¹. Therefore, there is a pressing requirement for novel biomarkers to screen for HCC or monitor treatment progress, which could improve the detection and management rates of early-stage HCC patients and potentially extend their survival^2–5. This imperative has fueled extensive research into discovering new therapeutic targets at various levels, such as plant-derived protein fractions (e.g., active proteins from Adenium obesum leaf extract)⁶, miRNA-mRNA axes involved in cell death regulation (e.g., the miR-3130-5p/FDX1 axis)⁷, and regulatory non-coding RNAs (e.g., hsa_circ_0102231)⁸. These studies underscore the importance and diversity of exploring novel molecular mechanisms in HCC.

The neural precursor cell expressed developmentally down-regulated protein 1 (NEDD1) gene (NCBI Gene ID: 121441) is located on chromosome 12q22, and plays a crucial role in the localization of proteins to the centrosome in cells. The centrosome is a significant organelle in animal and lower plant cells, closely associated with cell mitosis and cytokinesis. NEDD1 is essential for centriole replication and spindle assembly, facilitating functional spindle assembly through phosphorylating NEDD1 and promoting microtubule nucleation around chromatin^9,10. Therefore, NEDD1 may be a mark for controlling cell proliferation¹¹. Previous research has demonstrated that a human gastric cancer xenograft model with NEDD1 knockdown significantly extends survival¹², and NEDD1 can promote the proliferation, migration, and tumorigenic development of lung adenocarcinoma cells, facilitating tumor immune evasion and accelerating tumor progression^13,14. However, the role of NEDD1 in the incidence and development of HCC remains understudied, crucially, it is still unknown whether the regulatory mechanisms, functional roles, and clinical implications of NEDD1 in HCC—a cancer with distinct etiologies and tumor microenvironment—differ substantially from its roles in other cancers (e.g., lung or gastric cancer).

To address these gaps, this study aims to systematically elucidate the role of NEDD1 within the specific context of HCC. We integrated bioinformatics, immunohistochemistry, real-time quantitative PCR, and functional cellular assays to delineate the expression pattern of NEDD1 in liver tissues and its association with the prognosis of primary HCC. Furthermore, through functional pathway enrichment, DNA methylation, protein phosphorylation analyses, and leveraging single-cell sequencing alongside spatial transcriptomic datasets, we investigated the dynamic and cell-type-specific expression patterns of NEDD1 and its related gene MZT2B within the HCC immune microenvironment, as well as their potential to constitute a functional module. This study also sought to explore how NEDD1 influences the tumor immune microenvironment and to link its expression signature to immunotherapy response and potential therapeutic strategies. Collectively, our work not only suggests the promotive effects of NEDD1 on the proliferation, migration, and tumorigenesis of HCC cells, but more importantly, provides the first systematic insight into its potentially regulatory networks and dynamic interactions with the tumor microenvironment in HCC, setting it apart from its known roles in other cancers.

Methods

Patient information and clinical sample collection

HCC tissues and adjacent normal liver tissues from 26 surgical patients were obtained from the Hepatobiliary Surgery Department of the First Affiliated Hospital of Guangxi Medical University. Written informed consent was obtained from all participants prior to their enrollment in the study. All patients were diagnosed with HCC through surgical pathology. The study protocol was approved by the Ethics Committee of the First Affiliated Hospital of Guangxi Medical University (Approval No. 2024-E828-01). Nine cases were evaluated by immunohistochemistry (Supplementary Table S1), and 17 cases were evaluated by RT-qPCR (Supplementary Table S2). These methods were used to verify the differential expression of NEDD1 between cancerous and adjacent tissues. All experimental procedures involving human participants were performed in accordance with the tenets of the Declaration of Helsinki.

Data collection and processing

The pan-cancer analysis dataset The Cancer Genome Atlas (TCGA)-GTEx and the HCC dataset TCGA-LIHC were sourced from the TCGA database (https://portal.gdc.cancer.gov/). The GTEx database was also used. From the Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/) database, we downloaded GSE76427, GSE144269, GSE39791, GSE112790, and GSE54236 to analyze the differential expression and survival between HCC tissue and normal liver tissue based on their expression and clinical data. Immunohistochemical sections of HPA038591 (HCC) and HPA038591 (normal liver tissue) were obtained from the Human Protein Atlas (HPA, https://www.proteinatlas.org/).

Survival analysis of NEDD1

Kaplan-Meier survival analysis was performed on the TCGA-LIHC, GSE76427, GSE144269, and GSE54236 datasets using the survival package. The optimal cutoff values for high/low expression groups were determined using the R package “survminer.” Univariate Cox survival analysis was performed on NEDD1 in TCGA-LIHC and traditional clinical variables using the survival package. Samples in the univariate group that met the set p-value threshold were included in the multivariate Cox model to construct a model. A forest plot was generated using the forest plot package.

Differential gene analysis and enrichment analysis

Tumor samples were extracted from TCGA-LIHC and HCC divided into two groups using a cutoff value of 0.5. The group with the lowest 50% NEDD1 expression was used as the reference group for differential analysis using the limma package. Genes with |log2FoldChange|>2 and adjusted P-values < 0.05 were considered significantly different. A volcano plot was created for visualization. GO pathway enrichment analysis and Gene Set Enrichment Analysis (GSEA) analysis were performed on the high-expression group of NEDD1 in TCGA-LIHC. GSEA analysis was based on the gene set “c2. cp. kegg. v7.5.1.symbols.gmt”, and the gene set enrichment score (ES) was calculated. Significance tests and multiple hypothesis tests were conducted on the ES values of the gene set.

Methylation and phosphorylation

In TCGA-LIHC, the expression level and promoter methylation degree of NEDD1 were standardized using z-core. A z-core value greater than 0 was marked as the high expression group or high methylation group (above standard). Box plots were used to graphically represent the methylation levels at these specific genomic sites. The Wilcoxon rank-sum test was used to compare HCC tissue with normal liver tissue. From the Proteomics Public Database (PDC, https://proteomic.datacommons.cancer.gov/pdc/), we downloaded the dataset PDC000199 and compared the differential expression of phosphorylation sites of the NEDD1 protein between HCC tissue and normal tissue using the Wilcoxon rank sum test.

Prediction of anti-cancer drugs

Clinical data were downloaded from the Cancer Immunohistochemistry Atlas (TCIA), and the correlation between NEDD1 and immunotherapy was visualized using R packages “limma” and “ggpubr”. Spearman correlation analysis was used to determine the relationship between NEDD1 gene expression and dose-response curve (area under the curve [AUC]) values in the CTRP database.

Cell culture and transfection

Normal hepatocyte MIHA was obtained from Fenghui Biotech Co., Ltd. (Hunan, China). HCC cell lines 97H and HUH7 were procured from Pricella Biotechnology Co. (Wuhan, China) or Servicebio (Wuhan, China). MIHA cells were cultured in RPMI-1640, and 97H and HUH7 cells were cultured in DMEM, both containing 10% fetal bovine serum (FBS). The culture temperature was 37°C, and the environment was 5% CO2. The si-RNA of NEDD1 was purchased from Sangon Biotech (China). The si-1 sequence are 5 ‘-AAGUAACUUGUGUAACAUA-3’ and 5 ‘-UAUGUUACACAAGUUACUU-3’. The si-2 sequences are 5 ‘- UCGGUAAUGGAAUAGUAA-3’ and 5 ‘- UUACUAUUCCAUU AUCCGA-3’. The si-3 sequences are 5 ‘-GUGAACAAACGAAGUGUUA-3’ and 5 ‘-UAACACUUCGUUUGUUCAC-3’.

Predicting the impact of NEDD1 knockout on cells

Genome-wide screening data using CRISPR-Cas9 were downloaded from the DepMap database (https://depmap.org/portal/download/). Approximately 17,000 candidate genes were analyzed using the CERES algorithm to calculate the dependency score¹⁵. A negative score indicates cell growth inhibition and/or death after NEDD1 gene knockout. Scores of 0 and − 1 represent the median effect of non-essential genes and common core essential genes, respectively.

Reverse transcription quantitative polymerase chain reaction (RT-qPCR)

Total RNA was extracted from cultured cells and clinical samples using Trizol reagent (Invitrogen). RNA integrity and concentration were assessed using a microspectrophotometer. Reverse transcription of RNA into cDNA and quantitative PCR reagents were purchased from Takara Bio. The PCR primers for NEDD1 were obtained from Sangon Biotech, with the sequences: F: 5 ‘- AGTTGGTCACATCTGGTGCT-3’ and R: 5 ‘- AGTGACTGGCAACTCCAGC-3’.

Western blotting

Cells and tissues were lysed using Solarbio’s protease inhibitor and RIPA buffer at 4 °C. Electrophoresis was performed using a 10% SDS-PAGE gel, followed by transfer to a polyvinylidene fluoride (PVDF) membrane. The membrane was blocked with 5% skim milk powder for 50 min, then incubated overnight at 4 °C with primary antibodies: NEDD1 (1:1000, Proteintech, 13993-1-AP) and Beta Actin (1:20000, Proteintech, 66009-1-Ig). After incubation with secondary antibodies (diluted to 1:5000, Proteintech, catalogue numbers SA00001-1 and SA00001-2) for 60 min, proteins were detected using the ECL (Enhanced Chemiluminescence) technique.

CCK cell proliferation assay

Cells were seeded into 96-well plates at a density of 2500 cells per well, with at least three replicates per group.Measurements were taken on days 0, 1, 2, 3, and 4 using a multifunctional enzyme-linked immunosorbent assay (Varioskan™ LUX) reader to measure the optical density (OD) value at 450 nm.

Clone formation

Cells were seeded into 12-well plates at a density of 1800 cells per well, with at least three replicates per experimental group. The culture period lasted 10–14 days, and colonies with at least 50 cells were counted.

Cell migration experiment

Cells were seeded into culture plate inserts (PC membrane, 12 μm for 97 H, 8 μm for HUH7), with 200,000 cells (97 H) or 50,000 cells (HUH7) per well. Cells were fixed after 24 h of culture.

Nude mouse xenograft experiment

All animal experimental protocols were reviewed and approved by the Ethics Committee for Research of the First Affiliated Hospital of Guangxi Medical University (Approval No. 2025-E0214). All procedures were performed in strict accordance with the institution’s guidelines and regulations for the care and use of laboratory animals. The reporting of this study is in compliance with the ARRIVE guidelines. BALB/c male nude mice (4 weeks old, weighing 16–18 g, n = 10) were purchased from the Animal Center of Guangxi Medical University. Human HCC HUH7 cells were inoculated subcutaneously into the armpit of nude mice to establish a human primary HCC implantation tumor model. HUH7 cells in the logarithmic growth phase were suspended at a concentration of 2 × 10⁶ cells per 0.2 ml and inoculated subcutaneously. Tumor long diameter (a) and short diameter (b) were measured every 3 days using vernier calipers. Tumor volume was calculated using the formula:

Mice were euthanized by cervical vertebra dislocation when the tumor volume reached approximately 1000 mm³ (or the longest diameter reached ~ 1 cm) or after 10 weeks of growth.

All animal experimental procedures were performed in accordance with the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals.

Immunohistochemistry

Paraffin sections from surgical specimens or nude mouse implant tumors were dehydrated and blocked. Sections were incubated with primary antibodies (NEDD1, 1:100, CUSABIO, CSB-PA822820DSR1HU; PCNA, 1:1500, Proteintech, 10205-2-AP; E-cadherin, 1:5000, Proteintech, 20874-1-AP; N-cadherin, 1:2000, Proteintech, 22018-1-AP; Vimentin, 1:2500, Proteintech, 10366-1-AP) at 4 °C overnight. After incubation with secondary antibodies at room temperature for 1 h, sections were observed under a microscope and stained for imaging. Staining results were analyzed using ImageJ software to calculate the average optical density (AOD) of NEDD1-positive staining, which was determined by dividing integrated optical density by the stained area.

Single gene single-cell sequencing analysis

The single-cell dataset GSE140228 was downloaded from the GEO database, and after removing samples of intrahepatic cholangiocarcinoma (ICC), the scTIME database (http://scTIME.sklehabc.com) was used to perform subsequent analysis on GSE140228-HCC (see Supplementary Table S3). GSE140228-HCC is used to detect the relationship between the expression levels of NEDD1 and MZT2B genes in different cell subtypes, and to draw relevant scatter plots.

Investigation into the infiltration of immune cells

The CIBERSORTx website (https://cibersortx.stanford.edu/) was used to quantify the infiltration ratios of NK cells, monocytes, and macrophages in TCGA-Liver Hepatocellular Carcinoma (LIHC) samples^16,17 using the CIBERSORT algorithm. Survival analysis was conducted using the survival package, comparing groups with high and low immune cell infiltration levels.

Download scRNA seq data for human primary HCC

The initial treatment HCC spatial transcriptome dataset HRA000437 was downloaded from the China National Center for Bioinformation (CNCB, https://www.cncb.ac.cn/). HCC slice data containing both HCC tumor and adjacent normal liver tissue were obtained. The expression of NEDD1 and its interacting genes were analyzed in both regions.HCC Among the available samples, tissue slices from four patients met our analysis criteria. According to the original publication¹⁸, HCC slices containing both HCC and normal liver tissue adjacent to the cancer were labeled with “L”. Accordingly, the selected slices were designated as HRA000437-L-HCC1, HRA000437-L-HCC2, HRA000437-L-HCC3, and HRA000437-L-HCC4. Clinical data corresponding to these samples are provided in Supplementary Table S4.

Spatial transcriptomics (ST) analysis

The Sparkle online analysis website (https://grswsci.top/) was used to process and analyze the downloaded spatial transcriptome sequence data. Deconvolution analysis based on ST and single-cell transcriptomics data was performed to evaluate cell composition. A feature scoring matrix was constructed by calculating the average expression level of the top 25 specifically expressed genes at each locus. The expression of NEDD1 and MZT2B genes at each locus was visualized using the “SpatialEigenPlot” function in the Seurat software package.

Statistical analysis

Data were processed using a Perl script, and statistical analysis and mapping were performed using R language software and the “Bioconductor” package. Data were compared using paired t-tests, one-way ANOVA, and Bonferroni post-hoc tests. A p-value less than 0.05 was considered statistically significant (*P < 0.05; **P < 0.01; ***P < 0.001).

Results

NEDD1 overexpression in HCC

We conducted a pan-cancer analysis using the TCGA GTEx database and observed significantly higher expression of NEDD1 in 21 types of cancer tissues compared to adjacent normal tissues (Fig. 1A, P < 0.05). The “gganatogram” package was used to visualize the median values of each organ tumor compared to the normal population (Supplementary Figure S1). Elevated levels of NEDD1 protein were also detected in the HPA database (Supplementary Figure S2). In the TCGA-LIHC dataset, NEDD1 expression was significantly higher in HCC tissues (Fig. 1B). Similar results were obtained in the GSE76427, GSE144269, GSE39791, and GSE112790 datasets (Fig. 1C–F).

Fig. 1.

Examination of NEDD1 expression levels in hepatocellular carcinoma tissues. (A) The mRNA expression levels of NEDD1 across various cancers were obtained from the TCGA-GTEx database; (B–F) NEDD1 expression in the TCGA-LIHC, GSE76427, GSE144269, GSE39791, and GSE112790 datasets; P < 0.05; P < 0.01; *P < 0.001.

The correlation between NEDD1 and the prognosis of HCC

Kaplan-Meier survival analysis indicated that high expression of NEDD1 significantly correlated with poor prognosis (overall survival [OS], disease-specific survival [DSS], and progression-free interval [PFI]) in TCGA-LIHC (Fig. 2A–C). To elucidate the correlation between NEDD1 expression status and the prognosis of HCC, patients with primary HCC in TCGA-LIHC were divided into different groups based on pathological stage (I&Ⅱ, Ⅲ&Ⅳ), age (≤ 60, > 60), gender (female, male), race (Asian, white, black or African American), Alpha-Fetoprotein (AFP, ≤ 400, > 400), Body Mass Index (BMI,≤ 25, > 25), fibrosis Ishak score (0, 1/2, 3/4, 5, 6), and vascular invasion (yes, no). The results showed that high NEDD1 expression was significantly associated with poor prognosis in clinical subgroups older than 60 years, male, Asian, BMI > 25, and fibrotic Ishak score of 6 (Fig. 2D–H). These results suggest that NEDD1 may be an independent prognostic factor for HCC, prompting us to further investigate its expression regulation and downstream functional mechanisms.

Fig. 2.

Correlation between NEDD1 expression and prognosis in primary HCC. (A–C) Overall survival, Disease-specific survival, and Progression-free interval associated with NEDD1 expression in TCGA-LIHC; (D–H) Overall survival associated with NEDD1 expression in various subgroups of TCGA-LIHC (age > 60 years, male, Asian, BMI > 25, fibrosis Ishak score 6).

Diagnostic and prognostic correlation analysis of NEDD1 in various datasets

We used ROC analysis to assess the diagnostic potential of NEDD1. The results indicated that the AUC value of NEDD1 was 0.828 in the TCGA-LIHC database, 0.798 in the GSE76427 dataset, and 0.861 in the GSE144269 dataset (Fig. 3A–C), suggesting that NEDD1 has good prognostic efficacy. Similar results were observed in other datasets and TCGA-LIHC clinical subgroup analysis (Supplementary Figure S3). To further support the prognostic value, we analyzed and generated time-dependent ROC curves in the TCGA-LIHC database, as well as in GSE76427 and GSE144269 (Fig. 3D–F). Univariate and multivariate Cox regression analysis showed that NEDD1 expression and pathological grading were independent risk factors for poor prognosis in primary HCC (Fig. 3G). In addition, we constructed a nomogram containing prognostic factors such as NEDD1 expression, pathological stage, age, gender, race, AFP levels, BMI levels, fibrosis Ishak score, and vascular invasion to predict the probability of 1-year, 3-year, and 5-year OS in HCC patients (Fig. 3H). The calibration curve showed that the survival rate predicted by the nomogram was close to the actual survival rate (Fig. 3I). These results indicate that NEDD1 is a potential prognostic factor for primary HCC.

Fig. 3.

Diagnostic and prognostic analysis of NEDD1. (A–C) Diagnostic ROC curves for NEDD1 in TCGA-LIHC, GSE76427, and GSE144269; (D–F) Time-dependent ROC curves for NEDD1 in TCGA-LIHC, GSE76427, and GSE144269; (G) Univariate and multivariate Cox regression analysis of NEDD1 expression levels; (H,I) Prognostic histograms and calibration curves showing the correlation between NEDD1 expression and multiple clinical variables.

Enrichment analysis of functional pathways of NEDD1

Based on TCGA-LIHC, using the NEDD1 low-expression group as a reference, we identified 467 downregulated and 56 upregulated genes in the high-expression group (Fig. 4A). Gene Ontology (GO) enrichment analysis of these differentially expressed genes showed that NEDD1 is enriched in biological processes (BP) and major cellular components (CC), including cell adhesion through cell membrane adhesion molecules, adhesion of homophilic cells through cell membrane adhesion molecules, and the complex of chain proteins (Fig. 4B, C, Supplementary Table S5). GSEA analysis using the “c2. cp. kegg. v7.5.1.symbols.gmt” dataset revealed that individuals with elevated NEDD1 expression levels showed significant associations with several conventional pathways, including cell cycle, focal adhesion, TGF-β signaling, cell adhesion molecules (CAMs), and MAPK signaling; Conversely, there was a negative association with pathways such as glutathione metabolism, cysteine and methionine metabolism, arachidonic acid metabolism, linoleic acid metabolism, and fatty acid metabolism (Fig. 4D, E, Supplementary Table S6)^19–21.

Fig. 4.

Enrichment analysis of NEDD1. (A) Volcano plot of differentially expressed genes in the high NEDD1 expression group; (B,C)GO enrichment analysis of differentially expressed genes in the high NEDD1 expression group; (D) Signaling pathways positively correlated with high NEDD1 expression; (E) Signaling pathways negatively correlated with high NEDD1 expression.

DNA methylation and protein phosphorylation analysis of NEDD1

To explore the potential mechanism of NEDD1 overexpression in HCC, we analyzed its epigenetic regulation and post-translational modification status. Analysis of the TCGA-LIHC dataset revealed that the DNA methylation level of the NEDD1 promoter region in HCC tissue was significantly reduced compared to normal liver tissue (Fig. 5A). After z-score standardization analysis of the expression level and promoter methylation degree of the NEDD1 gene, four subgroups were obtained. Kaplan-Meier survival analysis showed that the survival (OS, DSS, and PFI) of the 5UTR low methylation and NEDD1 high-expression groups was significantly worse than that of the high methylation and low-expression groups (Fig. 5B–D), suggesting a potential association between NEDD1 5UTR hypomethylation and poor prognosis, which may contribute to carcinogenesis in HCC patients. Further annotation of the 450k methylation chip probe sites using the ChAMPdata package revealed that the NEDD1 promoter region in HCC and normal liver tissues of TCGA-LIHC had 9 methylation sites, with most showing low methylation except for 2 sites with high intensity (Fig. 5E, F). A Wilcoxon test was used to analyze the differences between cancer and normal tissue, revealing that the methylation levels of cg21721825, cg11021222, cg10240725, cg06222550, and cg07549630 in the HCC group were significantly different from those in the normal group, suggesting that these loci may be related to the prognosis of HCC (Fig. 5G–K). These correlative analyses imply, but do not prove, a direct regulatory role of promoter hypomethylation in NEDD1 overexpression. Using the PDC000199 dataset to examine the phosphorylation status of NEDD1, it was found that the phosphorylation status of 7 sites in HCC tissue was significantly higher compared to normal liver tissue (Fig. 6A–G, p < 0.05). Univariate Cox proportional hazards analysis identified the s523 site as a risk factor significantly linked to an increased likelihood of HCC, suggesting that s523 may be the most important phosphorylation site of the NEDD1 protein (Fig. 6H).

Fig. 5.

Methylation of the NEDD1 gene. (A) DNA methylation level of the NEDD1 promoter region in TCGA-LIHC; (B–D) OS, DSS, and PFI analyses for four subgroups based on NEDD1 gene expression and promoter methylation levels in TCGA-LIHC; (E,F) Methylation levels at nine sites in the NEDD1 promoter region in HCC tissues versus normal liver tissues in TCGA-LIHC; (G–K) Methylation levels at five specific sites in the NEDD1 promoter region in HCC and normal liver tissues.

Fig. 6.

Phosphorylation of the NEDD1 protein. (A–G) Phosphorylation levels of seven NEDD1 sites (s403, s404, s412, s430, s517, t522, s523) in the PDC000199 dataset; (H) Single-factor Cox survival analysis for each phosphorylation site.

Prediction of anticancer drugs for NEDD1

Given the carcinogenic properties and unique phosphorylation pattern of NEDD1, we further evaluated its potential as a therapeutic target and predicted related drugs. Using the CTRP and GDSC2 drug databases, we analyzed the expression status of NEDD1 and the AUC or half maximal inhibitory concentration (IC50) values of drug response (Fig. 7A–B). The intersection of CTRP and GDSC2 results suggesting that topotecan, axitinib, and pevonidistat were predicted to have lower IC50/AUC values in the NEDD1 high-expression group in silico, suggesting their potential therapeutic efficacy (Fig. 7C). Topotecan and axitinib have been shown to have therapeutic effects on various cancers^22–26, and pevonidistat has been observed to have anti-tumor effects on HCC cells in the laboratory^27,28. Given that NEDD1 expression in the TCGA-LIHC dataset was strongly positively correlated with the key immune-related molecules PD-1 (PDCD-1, R = 0.262, P < 0.001) and CTLA-4 ( R = 0.268, P < 0.001) (Fig. 7D-E), TCIA was further analyzed to observe the efficacy of immunotherapy in groups with high and low NEDD1 expression. The violin plot suggested that the NEDD1 low-expression group had higher immune scores and better treatment outcomes when receiving immunotherapy, whether it was immunotherapy alone (anti-PD-1 or anti-CTLA4 therapy, Fig. 7F, G) or combination immunotherapy (Fig. 7H). This observation implies that high NEDD1 expression might define an immune microenvironment less responsive to checkpoint blockade, possibly through mechanisms involving the recruitment or functional modulation of immunosuppressive cells like macrophages (as suggested in Fig. 10).

Fig. 7.

Drug prediction related to NEDD1. (A,B) CTRP and GDSC2 drug databases predicting the correlation between NEDD1 expression levels and drug response; (C) Number of overlapping drugs predicted by CTRP and GDSC2; (D,E) Scatter plot showing the correlation between NEDD1 expression levels and PD-1 or CTLA4; (F–I) Immunotherapy analysis of PD-1 or CTLA4 expression in relation to NEDD1 expression in TCIA.

Fig. 10.

Expression profiles of NEDD1 and its associated genes in a single-cell dataset. (A) The gene most significantly interacting with NEDD1 in the STRING database was MZT2B; (B,C) Analysis of MZT2B expression in the TCGA-LIHC dataset and its association with prognosis. (D) UMAP distribution of cell subtypes in the GSE140228-HCC dataset; (E) Correlation scatter plot showing the relationship between NEDD1 and MZT2B gene expression levels across various cell subtypes in the GSE140228-HCC dataset, with the top 10 cell subtypes identified based on their correlation coefficient (R). (F–H) Survival analysis of infiltrating NK cells, monocytes, and macrophages in the TCGA-LIHC dataset.

Verification of NEDD1 expression and functional experiments

To verify the biological function of NEDD1 and its potential as a therapeutic target, we conducted a series of loss-of-function experiments in both cellular and animal models. We evaluated a preliminary cohort of 26 paired HCC and adjacent normal tissues using various experimental methods. Immunohistochemistry results from 9 cases and RT-qPCR results from 17 cases suggesting that the expression of NEDD1 in HCC tissues was significantly higher than in adjacent normal liver tissues (Fig. 8A–D). Although validated in a modest-sized clinical cohort, the consistent overexpression of NEDD1 in HCC tissues across our IHC/RT-qPCR data and multiple public transcriptomic datasets strengthens this observation. To investigate the function of NEDD1 in HCC cell lines, we first used the DepMap database to predict the effect of NEDD1 knockdown on cells (Fig. 8E). The results showed that the CERES scores of NEDD1 in many cell lines were negative and all were less than − 1, implying that cell growth was inhibited or cells died after NEDD1 knockdown, suggesting that NEDD1 may be a common core essential gene. We then used 97 H HCC cells, which are consistent with the HBV infection background of Chinese individuals, and HUH7 HCC cells, which are widely used in xenotransplantation animal models, for subsequent cell function experiments. RT-qPCR confirmed that NEDD1 expression was significantly higher in 97 H and HUH7 HCC cell lines than in MIHA normal liver cell lines (Fig. 8F). Subsequently, we established NEDD1 knockdown models for 97 H and HUH7 cells and confirmed the knockdown efficiency by Western blot and RT-qPCR (Fig. 8G, H, Supplementary File S1). Cell Counting Kit-8 (CCK8) and clonogenesis experiments showed that cell proliferation was significantly inhibited in the si-NEDD1 group compared to the control group (si-NC) (Fig. 8I, J). Transwell experiments showed that the number of migrating cells in the si-NEDD1 group was significantly lower than that in the control group (Fig. 8K). Results from the subcutaneous xenograft tumor model (using the HUH7 cell line) in immunodeficient mice implied that NEDD1 knockdown significantly inhibited cancer cell proliferation, with tumor size and weight significantly reduced at the endpoint (Fig. 8L). Finally, we detected the expression of EMT markers and PCNA protein levels in hepatoma cells of nude mice implanted with tumors after NEDD1 downregulation using immunohistochemistry. The results implied that compared to the NC group, the expression of the epithelial marker E-cadherin was significantly increased after NEDD1 downregulation, while the expressions of the mesenchymal markers N-cadherin and Vimentin, as well as the proliferation marker PCNA, were significantly decreased (Fig. 9A–D).

Fig. 8.

Validation of NEDD1 expression and its effect on the biological behavior of HCC cells. (A–D) Immunohistochemical and RT-qPCR evaluation of NEDD1 expression in HCC tissue and adjacent normal liver tissue; (E) CRISPR-Cas9 screens using the DepMap database to identify the top 200 cell lines based on CERES scores; (F) Comparative analysis of NEDD1 gene expression in normal liver cells, 97 H, and HUH7 cells; (G,H) Western blot and RT-qPCR experiments to validate NEDD1 knockdown efficiency; (I,J) Viability and colony-forming ability of 97 H and HUH7 cells post-transfection (CCK8 assay and clonal formation assay); (K) Transwell assay to analyze the effect of NEDD1 on the migration of 97 H and HUH7 cells; (L) Representative tumor images showing differences in tumor mass and size at the end of the experiment between si-NC and si-1 experimental groups in the HUH7 cell line. P < 0.05; *P < 0.01; **P < 0.001.

Fig. 9.

Expression of epithelial-mesenchymal transition markers and proliferation markers in transplanted tumors. (A–D) Expression levels of E-cadherin, N-cadherin, Vimentin, and PCNA in the si-NC and si-1 groups.

Expression profiles of NEDD1 and its associated genes in single-cell and spatial transcriptomic datasets

The aforementioned experiments suggest the driving effect of NEDD1 on the malignant phenotype of HCC cells. To further explore the cellular interaction network of NEDD1 in the complex tumor microenvironment (TME), we turned to single-cell and spatial transcriptomic analysis, aiming to identify its key interacting partners and affected cell subpopulations. Initially, we identified genes predicted to interact with NEDD1 using the STRING database (https://string-db.org/). The results implied that MZT2B had the highest degree of relevance among the genes examined (Fig. 10A). In the TCGA-LIHC dataset, MZT2B exhibited differential expression patterns in HCC tissues, with higher expression levels significantly linked to worse prognosis (Fig. 10B, C). Then, we evaluated the co-expression of NEDD1 and MZT2B across all available immune cell types in the GSE140228-HCC single-cell dataset (comprising 40 cell types, Fig. 10D), and found that their co-expression was most significantly concentrated in the following cell subgroups : NK-IL7R (Pearson, R = 1, P < 0.001), Mono-FCGR3A (Pearson, R = 1, P < 0.001), and macrophage-apoe (Pearson, R = 0.957, P = 0.011) (Fig. 10E). Using the TCGA-LIHC dataset, survival analysis was conducted to investigate immune cell infiltration (based on the CIBERSORT algorithm, Fig. 10F–H). The results suggested a significant correlation between poor prognosis and macrophage cells characterized by high invasiveness (Fig. 10H, P = 0.001). The macrophage-APOE cell subgroup has been shown to be associated with metastasis and immunotherapy in various cancers^29–31. Thus, the co-expression of NEDD1 and MZT2B within the macrophage-APOE cell subtype may hold greater clinical relevance. Subsequently, we employed deconvolution analysis to elucidate malignant and non-malignant regions across four spatial transcriptome sections of the HRA000437-L-HCC dataset (Fig. 11A, F,K, P). The SpatialFeaturePlot tool was used to depict the spatial expression patterns of the NEDD1 and MZT2B genes across distinct microregions (Fig. 11B, C,G, H,L, M,Q, R). The findings implied that NEDD1 expression was markedly elevated in malignant areas relative to non-malignant areas (Fig. 11D, I,N, S), suggesting that NEDD1 expression may be primarily derived from tumor cells. Similarly, the same results were observed for MZT2B (Fig. 11E, J,O, T).

Fig. 11.

Expression of NEDD1 and MZT2B in the spatial transcriptome. (A,F,K,P) Spatial maps showing malignant and non-malignant regions in the HRA000437-L-HCC dataset; (B,C,G,H,L,M,Q,R) Expression of NEDD1 and MZT2B genes in the four slice datasets. (D,I,N,S) Histograms of the expression of NEDD1 in each slice dataset; (E,J,O,T) Histograms of the expression of MZT2B in each slice dataset.

Sample-specific co-localization of NEDD1 and MZT2B in macrophages

To investigate whether NEDD1 and MZT2B co localize within macrophage APOE subtypes at the spatial level, we conducted further analysis. On the basis of annotated malignant and non malignant regions, we performed more detailed cellular component annotations on the micro regions of four spatial slices (HCC1-4) (Fig. 12A, D,G, J). Subsequently, we calculated the spatial correlation of AUC for specific gene sets (NEDD1, MZT2B, and APOE) within each spatial point, with a particular focus on regions annotated as rich in macrophages or malignant cells. Interestingly, we observed significant heterogeneity between samples. In HCC1, HCC2, and HCC3 slices, we observed that specific gene sets were active in regions enriched with macrophages (Fig. 12B, C,E, F,H, I). However, in HCC4 slice (Fig. 12K, L), no such gene set activity was detected in macrophage-enriched regions. Conversely, in malignant cell-enriched regions of HCC2, HCC3, and HCC4, specific gene sets (NEDD1, MZT2B, and APOE) were active in these regions (Fig. 12E, F,H, I,K, L). Overall, the co expression of NEDD1 and MZT2B is not universally present in macrophages of all samples, but is detected in specific macrophage (APOE subtype) regions of some samples and tumor cell regions of most samples. This suggests that the co expression of the two may be a dynamic functional module associated with tumor progression, and their cellular carriers (tumor cells or tumor associated macrophages) may vary due to local microenvironment differences.

Fig. 12.

Evaluation of Spatial Features for the NEDD1, MZT2B, and APOE Gene Sets in HCC Sections. (A,D,G,J) Spatial maps showing detailed cellular component regions in the HRA000437-L-HCC dataset; (B,E,H,K) Specific gene sets (NEDD1, MZT2B, and APOE) score active landscapes in microregions; (C,F,I,L) The correlation between AUC scores of specific gene sets (NEDD1, MZT2B, and APOE) and various cellular components.

Discussion

HCC is one of the most common and highly malignant tumors globally, posing significant challenges for early diagnosis and effective treatment. The identification of efficient biomarkers for the initial diagnosis and precise prognosis of HCC is crucial for improving patient outcomes. For early HCC screening, clinically established biomarkers including AFP and protein induced by vitamin K absence or antagonist-II (PIVKA-II) still face challenges due to insufficient sensitivity and specificity in certain clinical scenarios. Therefore, the diagnosis and treatment of HCC remain significant challenges, and there is a crucial need for novel molecules associated with immune targeting to enhance diagnostic and treatment rates³².

NEDD1 was initially discovered in neural precursor cells of mice, and along with NEDD4 and NEDD9, it has been implicated in tumorigenesis and tumor progression. Current research on NEDD1 primarily focuses on its role in the formation of spindle apparatuses and centrosomes during cell mitosis. The carboxyl-terminal domain of NEDD1 interacts with γ-tubulin following phosphorylation by Cdk1, Pik1, and Nek9^33–37. This interaction recruits γ-tubulin to the centrosome, forming a large complex with other proteins known as the γ-tubulin ring complex (γ-TuRC)^38,39. Therefore, NEDD1 plays a role in microtubule nucleation at the centrosome. Numerous studies have highlighted the importance of NEDD1 in cell mitosis⁴⁰.

Tumorigenesis is associated with abnormal cell division, and the dysregulated expression of NEDD1, which is involved in mitosis, may contribute to tumor development. Our research found that the expression of NEDD1was significantly higher in various cancer tissues compared to normal tissues, and multiple datasets indicated that high expression of NEDD1 is associated with poor survival and prognosis in HCC patients. We evaluated the diagnostic and prognostic value of NEDD1 in HCC using ROC curve analysis, univariate, and multivariate Cox regression models. Our findings suggest that NEDD1 has significant potential as a novel diagnostic and prognostic biomarker for HCC. Furthermore, GSEA (KEGG) enrichment analysis revealed that the expression of NEDD1 is related with classical pathways such as the cell cycle, focal adhesion, TGF-β signaling, cell adhesion molecules CAMs, and MAPK signaling. These results indicate that abnormal NEDD1 expression may influence the occurrence and development of HCC through these pathways.

A growing body of research indicates a strong relevance between DNA methylation and tumorigenesis^41,42. In certain tumor cells, oncogenes are activated in a low methylation state^43,44, and changes in low methylation in tumor cells can lead to immune evasion and immune escape⁴⁵. Similarly, widespread DNA hypomethylation has been observed in HCC⁴⁶. Unlike previous studies on NEDD1, this study, for the first time, suggests the potential link between the NEDD1 gene and DNA methylation. We found that the DNA methylation level of the NEDD1 promoter region in HCC tissues was significantly reduced after mining the dataset. Additionally, our research revealed that individuals with low levels of methylation and high expression experienced significantly poorer survival rates compared to those with high methylation and low expression. This finding suggest that methylation could regulate the NEDD1 gene, with reduced methylation levels enhancing NEDD1 expression. This conclusion is supported by integrating RNA sequencing data across various datasets; the gene’s elevated expression in HCC samples, as opposed to healthy liver tissues, points to NEDD1 functioning as an oncogenic factor. Protein phosphorylation, a post-translational modification⁴⁷, is intricately connected to the initiation, progression, and advancement of tumors^48–51. Using the PDC000199 dataset, we validated s523 as the primary phosphorylation site of NEDD1 protein in HCC tissues.The efficacy of cancer immunotherapy or targeted therapy is closely associated with gene mutations and abnormal gene expression^52–56. Using the TCIA database, we explored potential differences in immunotherapy response between cohorts with high and low NEDD1 expression levels. The results showed that the low-expression group had better therapeutic effects. Additionally, we predicted the IC50 or AUC of drugs related to NEDD1 expression using a database. Topotecan, axitinib, and pevonidistat suggested better therapeutic effects in the high-expression group, providing a theoretical basis for chemotherapy or targeted therapy in these patients.

At the experimental technical level, we first validated the clinical relevance of NEDD1 by confirming its significant upregulation in paired clinical HCC samples. In addition, our loss-of-function experiments demonstrated that inhibiting NEDD1 effectively suppresses the proliferation, migration, and tumorigenic ability of HCC cells, suggesting that NEDD1 is essential for maintaining the malignant phenotype of HCC cells and may play a carcinogenic role. Moreover, various subtypes of immune cells within the tumor immune microenvironment were strongly associated with tumor prognosis and the efficacy of immunotherapy^57–60. For instance, Peng Zhao and colleagues demonstrated that the frequency of APOE-positive macrophages was elevated in patients with triple-negative breast cancer who did not respond to immune checkpoint inhibitors (ICIs). In a tumor-bearing mouse model, the combination of an APOE inhibitor with ICIs showed the most potent therapeutic efficacy. Yanru Qin and team found that, in metastatic esophageal squamous cell carcinoma, the differences in the tumor microenvironment, compared to primary tumors, were primarily driven by interactions between APOC1 ⁺ / APOE-positive macrophages and tumor or stromal cells. Additionally, Meng-Su Zeng and colleagues revealed a strong correlation between APOE-positive macrophage infiltration and immune cell-associated fibroblast (iCAF) infiltration in HCC^30,61,62. Unlike previous studies on NEDD1, this study found through single-cell transcriptome analysis that NEDD1 and its interacting gene MZT2B exhibited a co expression trend in multiple immune cell subgroups, including Macrophage APOE, suggesting its potential role in the tumor immune microenvironment. More importantly, spatial transcriptome data provides us with a crucial and more physiological perspective to understand the complexity of these interactions. We found that the co expression of NEDD1 and MZT2B exhibited significant inter sample and intra regional heterogeneity in space. In some samples, its co localization is associated with the enrichment region of macrophages (including APOE+subtypes); In other samples, it is more significantly concentrated in the area of malignant epithelial tumor cells. This cell background dependent expression pattern is a typical feature of the complexity and plasticity of the tumor microenvironment⁶³. Research has shown that key signaling pathways in the tumor ecosystem, such as TGF - β and inflammatory signaling, can simultaneously affect tumor cells and stromal cells, leading to the synergistic regulation of the same molecular module in different cell types⁶⁴. Tumor cells and macrophages can promote each other through shared metabolic reprogramming pathways⁶⁵. Therefore, our data supports a more integrated view: NEDD1 and MZT2B constitute a functional module activated in the liver cancer microenvironment. Within tumor cells, this module may directly participate in driving cell cycle progression or invasion (consistent with the known mitotic function of NEDD1); In tumor associated macrophages (TAMs), especially in the APOE+subpopulation, the activation of this module may indicate a specific pro tumor expression, which may be related to immune regulation, matrix remodeling, or lipid metabolism. This is consistent with reports of APOE+macrophages promoting metastasis and immune therapy resistance in other cancer types^30,61,62. The coexistence of this module in two cell types may reflect a synergistic adaptation or bidirectional signaling between tumor cells and immune cells, jointly shaping an immunosuppressive microenvironment and promoting disease progression⁶⁵. Future research needs to use techniques such as multi-color immunofluorescence to validate this spatial relationship at the protein level and elucidate the specific downstream effects of this module in different cell types.

Limitations and future perspectives

This study has several limitations. First, the direct regulatory effect of NEDD1 promoter hypomethylation on its overexpression, as well as the physical interaction between NEDD1 and MZT2B, require further experimental validation (e.g., reporter assays, ChIP, co-immunoprecipitation, co-modulation). Second, the therapeutic efficacy of the candidate drugs (e.g., topotecan, axitinib) identified from database screening in HCC models and the mechanistic link to NEDD1 expression await confirmation by in vitro and in vivo pharmacodynamic experiments. Furthermore, future studies are warranted to address the relatively small clinical cohort size via larger-scale, multi-center validation, and to further complement the current findings through both gain-of-function experiments and orthotopic/metastatic models to confirm NEDD1’s oncogenic sufficiency. Despite these limitations, our comprehensive multi-omics analysis systematically delineates the prognostic value, epigenetic features, immune microenvironment localization, and functional impact of NEDD1 in HCC, providing a solid hypothetical foundation and clear directions for subsequent in-depth mechanistic and translational research.

Conclusion

In summary, this study delineates a multi-layered regulatory network of NEDD1 and its role in HCC (Fig. 13). Firstly, hypomethylation of the promoter region may be a key upstream epigenetic event driving the overexpression of NEDD1 in HCC (Result 5), while abnormal phosphorylation at the s523 site may further modify and activate its protein function (Result 5). High expression of NEDD1 is directly associated with poor prognosis (Results 2, 3) and may drive tumor cell proliferation and migration by enriching in pathways such as the cell cycle and TGF-β (Result 4). Furthermore, our analysis suggests a positive correlation between NEDD1 and immune checkpoint molecules (PD-1, CTLA-4), and a link between high expression and resistance to immunotherapy (Result 6), implying its involvement in shaping an inhibitory immune microenvironment. This hypothesis has been further explored at the single-cell and spatial transcriptomic levels: we found a tight interaction pair between NEDD1 and MZT2B (Result 8), and the expression of this NEDD1-MZT2B functional module exhibits cell context-dependent heterogeneity—it can be active in some tumor cells, driving their intrinsic malignant behaviors, and can also be co-expressed in a specific subpopulation of APOE+ tumor-associated macrophages (Results 8, 9). Given the known pro-metastatic and immunosuppressive functions of APOE+ macrophages^30,61,62, we speculate that the activated NEDD1-MZT2B module in macrophages may be a key link connecting the intrinsic carcinogenicity of tumor cells with the remodeling of the immune microenvironment. This phenomenon of tumor cells and immune cells sharing core modules may represent a novel, bidirectional intercellular communication mode in HCC, jointly driving disease progression. This comprehensive analysis lays the foundation for understanding the role of NEDD1 and its related genes in the tumor immune microenvironment. These findings suggest that NEDD1 may be a potential target for the diagnosis and treatment of HCC.

Fig. 13.

A Model of NEDD1 Regulatory Network and Function in HCC. This schematic summarizes an integrated mechanism: NEDD1 overexpression, potentially regulated by promoter hypomethylation, drives tumor malignancy via key pathways and correlates with poor prognosis. The NEDD1-MZT2B module shows context-specific expression in tumor cells and APOE+ macrophages, linking to immune checkpoint upregulation and therapy resistance, thereby contributing to an immunosuppressive microenvironment.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Yu Chen: Data curation, formal analysis, validation, visualization, original draft. Zuyin Wan and Haixiang Xie: Conceptualization, investigation, software, methodology, validation. Tao Peng and Xin Zhou: Funding acquisition, project administration, supervision, review and editing, resources.

Funding

This work was funded by the Joint Project on Regional High-Incidence Diseases Research of Guangxi Natural Science Foundation (No. 2024GXNSFDA010029). Additional funding was provided by the Guangxi Science and Technology Program under Grant (No. AD25069077), the Key Laboratory of early Prevention & Treatment for regional High Frequency Tumor (Guangxi Medical University)-Ministry of Education (No. 2025GXNSFAA069704 and No. GXMULJZ202402), Guangxi Research Basic Ability Improvement Project for Young and Middle-aged Teachers (No. 2024KY0128) and the National Funded Postdoctoral Researcher Program (No. GZC20230583).

Data availability

The single-cell dataset GSE140228 used in this study can be downloaded from the GEO database (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE140228); The spatial transcriptome dataset HRA000437 can be downloaded from the CNCB database with the accession number HRA000437 (https://ngdc.cncb.ac.cn/gsa-human/browse/HRA000437). The other datasets used are either included in the Materials and Methods section or listed in supplementary tables. For further inquiries, please contact the corresponding author.

Declarations

Competing interests

The authors declare no competing interests.

Statement of ethics

The acquisition of human tumor samples for this study was approved by the Ethics Committee of the First Affiliated Hospital of Guangxi Medical University, with the ethical approval number [2024-E828-01]. The animal experiments were carried out following the guidelines and regulations established by the Ethics Committee for Research at the First Affiliated Hospital of Guangxi Medical University [2025-E0214].

Declaration of AI-assisted technologies in the writing process

During the preparation of this work, the author utilized Grammarly to enhance the readability and linguistic quality of the manuscript. All scientific content, data interpretation, and conclusions were independently completed by the author.