All RNA expression patterns of NFκB inhibitors in HCC and pan-cancer were taken from The University of Alabama at Birmingham Cancer Data Analysis Portal (UALCAN) and the Tumour Immune Estimation Resource (TIMER; http://cistrome.shinyapps.io/timer/), whose data were obtained from the TCGA database. TIMER (15,16) predicts tumour-infiltrating immune cells using gene expression patterns. In the present study, it was used it to assess NFKBIE and immunological infiltration. UALCAN (https://ualcan.path.uab.edu/index.html) (17) is a resource that is available for free that can be used to evaluate TCGA gene expression data. In the present study, UALCAN was used to compare the NFκB inhibitor family and its relationship with clinical and pathological characteristics and its prognostic value.

KEGG (18) is the main public database on pathways. Pathway analysis of differential protein coding genes was performed using the KEGG database, and the significance of differential gene enrichment in each pathway entry was calculated using the hypergeometric distribution test. The fold change (FC) was set to 1.5, whilst adjusted P<0.05 was statistically significant.

Kaplan-Meier plots/Liver cancer mRNA-seq/Auto select best cutoff (http://kmplot.com/analysis) were used to estimate HCC survival rates at 7/5/2024 (19).

Data from >4,000 microarrays are combined in MERAV (http://merav.wi.mit.edu/), enabling reliable cross-sectional comparisons of several tissues and cell lines (20).

NFKBIE protein expression information was obtained from the HPA (https://www.proteinatlas.org/search/NFKBIE). The HPA is a database based on proteomics data that contains protein expression in both normal and malignant tissues (21).

LinkedOmics (https://linkedomics.org/) (22) is a web-based analytical tool for comparing multiomics cancer datasets. This database was used for KEGG analysis and analysis of the TOP50 genes positively and negatively associated with NFKBIE.

STRING (https://string-db.org/. version 12.0) database is a protein interaction network database based on public database and literature information. It gathers several public databases, including UniProt, KEGG, NCBI, and Gene Ontology, to integrate these data and generate a comprehensive protein interaction network database (23–25).

The Genomics of Drug Sensitivity in Cancer (GDSC, version 8.5, www.cancerRxgene.org) database currently encompasses data on drug sensitivity from nearly 75,000 experiments, detailing the responses of approximately 700 cancer cell lines to 138 anti-cancer drugs. Graphical representations of the data are provided throughout, along with links to related resources, and all datasets are fully downloadable. GDSC offers a unique resource, comprising an extensive collection of drug sensitivity and genomic datasets, to facilitate the discovery of new therapeutic biomarkers for cancer treatment (26).

CTRP (https://portals.broadinstitute.org/ctrp. version 2.1) integrates the intricate relationship between cancer cell line genetics, lineage, and other cellular characteristics with small molecule sensitivity to expedite the discovery of personalized cancer treatments. By seamlessly integrating comprehensive drug sensitivity data with detailed genomic information, CTRP serves as a robust resource for gaining profound insights into the complex interplay between cancer genetics and drug responses (27–29).

The gene expression data of ~4,000 clinical samples are contained in the HCCDB (version, http://lifeome.net/database/hccdb/about.html), which is a database that provides accurate and reproducible differential expression profiles for HCC (30). The present study used this database to look at NFKBIE gene networks co-expressed with HCC as well as with normal tissues.

GSCALite (http://bioinfo.life.hust.edu.cn/web/GSCALite/.) is web server for dynamic analysis and visualization of cancer gene sets and drug sensitivity correlations that will be of wide practical value to cancer researchers. The present study used this database to analyze the signaling pathways of NFKBIE effects and studies related to drug resistance (31).

An HCC tissue microarray (cat. no. HLivH160CS02) consisting of 80 matched tumours and normal samples was purchased from Shanghai Outdo Biotech Co., Ltd. Following the dewaxing of paraffin-embedded sections, antigens were repaired (20X Tris-EDTA antigen repair solution; Wuhan Servicebio Technology Co., Ltd.; cat. no. G1203), and endogenous peroxidase activity was quenched (3% hydrogen peroxide solution, incubated at room temperature for 25 min away from light). Subsequent steps included 3% BSA (Wuhan Servicebio Technology Co., Ltd.; cat. no. GC305010) at room temperature for 30 min blocking to prevent non-specific binding, incubation with the primary overnight at 4°C and (NFKBIE polyclonal antibodies 1:200; cat no. 11273-1-AP; Proteintech Group, Inc.) secondary antibodies at room temperature (Wuhan Servicebio Technology Co., Ltd.; cat. no. GB23303; 1:500) and development of the signal using DAB chromogen (Wuhan Servicebio Technology Co., Ltd.; cat. no. G1212). Finally, hematoxylin stained nucleus was used for comparison. The prepared slides were then digitized using a Pannoramic MIDI scanner (3DHISTECH), ensuring high-resolution imaging of the histological details.

Total RNA was isolated from cell samples with a total of 3×10⁵ cells using the RNeasy Mini Kit (Qiagen GmbH) according to the manufacturer's instructions. RNA purity and quantity were assessed using a NanoDrop spectrophotometer (Thermo Fisher Scientific, Inc.). Complementary DNA (cDNA) was synthesized from the isolated RNA using the PrimeScript RT Reagent Kit (Takara Biotechnology Co., Ltd.), followed by cDNA amplification with the SYBR Premix Ex Taq II (Takara Biotechnology Co., Ltd.). The amplification reactions were conducted on an ABI StepOne Plus Real-Time PCR System (Applied Biosystems; Thermo Fisher Scientific, Inc.). NFKBIE-Forward primer sequence: TCTGGCATTGAGTCTCTGCG; NFKBIE-Reversed primer sequence: AGGAGCCATAGGTGGAATCAG. GAPDH-Forward primer sequence: GAGAAGTATGACAACAGCCTCAA. GAPDH-Reversed primer sequence: GCCATCACGCCACAGTTT. The 2^−ΔΔCq method was used for quantitative analysis (32). The PCR cycling conditions were set as follows: initial denaturation at 94–95°C for 1–3 min; 25–35 cycles of denaturation at 94°C to 95°C for 15–30 sec, annealing at 4–5°C below the primer's melting temperature (~50–65°C) for 20–40 sec and extension at 72°C for ~1 min per 1 kb of target DNA. This was followed by a final extension at 72°C for 5–10 min, and the samples were then held at 4°C. All protocols for RNA extraction, cDNA synthesis, and qPCR were conducted strictly following the manufacturers' guidelines. These experiments were replicated three times to ensure reliability and reproducibility of the results.

HEP-3B and PLC/PRF/5 cells were purchased from the American Type Culture Collection. Huh-7 cells were purchased from JCRB Cell Bank. HCCLM3 and MHCC97H cells were purchased from the Shanghai Institute of Biochemistry and Cell Biology (Chinese Academy of Sciences; Shanghai, China). All cells were cultured in DMEM (Gibco; Thermo Fisher Scientific, Inc.) or RPMI 1640 (Gibco; Thermo Fisher Scientific, Inc.) with 10% (Bioind) and 1% ampicillin (Biosharp Life Sciences) at 37°C in a 5% CO₂ incubator.

Short-interfering (si)RNA technology was used to knockdown NFKBIE. The sequences of the siRNAs [General Biol (Anhui) Co., Ltd.] used were as follows: NFKBIE scrambled siControl: 5′-UUCUCCGAACGUGUCACGUTT-3′ and 3′-ACGUGACACGUUCGGAGAATT-5′; NFKBIE siRNA-1: 5′-UCAAGGAACCACAGGAGAATT-3′ and 3′-UUCUCCUGUGGUUCCUUGATT-5′; NFKBIE siRNA-2: 5′-GGACCGGCAUGGUGACACATT-3′ and 3′-UGUGUCACCAUGCCGGUCCTT-5′; and NFKBIE siRNA-3: 5′-GGAAACUGCUGCUGUGUACTT-3′ and 3′-GUACACAGCAGCAGUUUCCTT-5′. Briefly, HCCLM3 and Huh-7 were first digested with Trypsin-EDTA Solution (Biosharp Life Sciences; cat. no. BL512A), then ~2.5×10⁵ cells were counted in each 6-well plate and, after the cells were attached to the wall, pre-configured NFKBIE-siControl (100 nM), NFKBIE-si1 (100 nM), and NFKBIE-si2 (100 nM), were added into the corresponding 6-well plates for 8 h at 37°C. The co-transfection reagent used was Lipofectamine^® 2000 Transfection Reagent (Thermo Fisher Scientific, Inc.). The culture medium was changed 8 h after transfection, and the NFKBIE protein level was measured by western blotting after 48 h. HCCLM3 and Huh-7 cells were transfected with the method and the NFKBIE siRNA knockdown effect was verified by western blotting, and then RNA-sequencing (RNA-seq) was performed by Shanghai OE Biotech Co., Ltd. (Shanghai, China). Wound healing and migration experiments were performed on HCCLM3 and Huh-7 cells after NFKBIE knockdown.

The same number of cells (1×10⁶) were treated with cell lysate (Biosharp Life Sciences), the sample buffer (Bio-Rad 2X Laemmli Sample Buffer; Bio-Rad Laboratories, Inc.; cat no. 1610737) was added and the proteins were denatured and stored at −80°C. The 5×10⁵ cells yielded 100 ul protein solution, 10 µl per well. The proteins were separated on 10% SDS-PAGE gel and transferred onto a PVDF membrane (MilliporeSigma; 0.45 µm). The membrane was blocked in skimmed milk at room temperature for 30–60 min and incubated with the primary antibody followed by the second antibody (OriGene Technologies, Inc.; cat. no. ZF-0516; 1:5,000) for 40–60 min at room temperature. ECL developer (Beyotime Institute of Biotechnology; cat. no. P0018FS) was used. In this study, protein concentration was determined using the BL367A assay kit provided by Biosharp Life Sciences. Initially, BCA Reagent A and BCA Reagent B were mixed in specified ratios to prepare the working solution. Protein samples and a series of BSA standards with known concentrations were then prepared and added to a 96-well plate. The BCA working solution was added to each well, and the plate was incubated at room temperature for 30 min to 2 h. Following incubation, absorbance at 562 nm was measured using a spectrophotometer. Protein concentrations were calculated by comparing these measurements against a standard curve. ImageJ software (National Institutes of Health, version 1.48) was used to calculate grayscale values. NFKBIE polyclonal antibodies (1:1,000; cat no. 11273-1-AP; Proteintech Group, Inc.) and GAPDH monoclonal antibodies (1:10,000; cat no. 60004-1-Ig; Proteintech Group, Inc.) were used as primary antibodies.

A total of 3×10⁵ HCCLM3/Huh-7 cells were initially cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and incubated at 37°C in a 5% CO₂ atmosphere until reaching 90% confluence. A sterile 200 µl pipette tip was employed to create a consistent scratch across the cell monolayer at near full confluence, focusing on maintaining uniform scratch width. After scratching, cells were washed with phosphate-buffered saline (PBS) to remove dislodged cells and debris, followed by the replacement of the culture medium with DMEM containing 1% FBS, to minimize proliferation and concentrate on migration. Photographic documentation was conducted immediately post-scratch (0 h) and at specified intervals (48 h), utilizing an inverted microscope with consistent magnification and focus. The extent of wound closure was assessed by measuring scratch width changes using ImageJ software (National Institutes of Health, version 1.48). The assay was replicated a minimum of three times to ensure data reliability. Statistical differences between groups were analysed using Student's t-test or one-way ANOVA with Tukey's multiple comparisons test, to determine the impact of the experimental conditions on the cells' migratory capabilities.

The complete medium was prepared using DMEM (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% FBS (Bioind) and 1% penicillin-streptomycin (Biosharp). For both the experimental and control groups, cells were digested with trypsin and counted to achieve a concentration of 2×10⁴ cells per 200 µl. These cells were then seeded into culture chambers (Falcon) with serum-free medium inside the chambers and complete medium outside. The cells were incubated at 4°C in an atmosphere of 5% CO₂ for 48 h. After incubation, cells were fixed with 4% paraformaldehyde (Biosharp) and stained with crystal violet. Subsequently, the upper layer of cells in the chamber was gently wiped away, and the cells were observed and photographed under an inverted microscope. This experiment was repeated at least three times. Cell counting was performed using Image J, and data analysis was conducted using GraphPad Prism 8.0 software. Statistical significance between the experimental and control groups was assessed using t-tests or analysis of variance (ANOVA).

The libraries were sequenced (TruSeq Stranded mRNA LTSample Prep Kit, Illumina, RS-122-2101; Agencourt AMPure XP, BECKMAN COULTER, A63881; Qubit RNA Assay Kit, Life Technologies, Q32852; Qubit dsDNA Assay Kit, Life Technologies, Q328520; Bioanalyzer 2100 RNA-6000 Nano Kit, Agilent, 5067-1511; Bioanalyzer 2100 DNA-1000 Kit Agilent, 5067-1504; SuperScript II Reverse Transcriptase, Invitrogen, 18064014;) on an Illumina HiSeq X Ten platform (cat. no. DOE20221816; Illumina, Inc.) and 150 bp paired-end reads were generated. The Agilent 2100 bioanalyzer was used to measure the length and quality of the library. The base error rate in Illumina sequencing was determined by the Phred score, which was calculated by the model to predict the probability of error in base discrimination. Raw data (raw reads) in fastq format were firstly processed using Trimmomatic (version 0.36) (33) and the low-quality reads were removed to obtain the clean reads. Subsequently, ~14,000 clean reads for each sample were retained for analysis. The clean reads were mapped to the human genome (GRCh38) using HISAT2 (version 2.2.1.0) (34). The fragments per kilobase of transcript per million mapped reads (35) of each gene was calculated using Cufflinks (version 2.2.1) (36), and the read counts of each gene were obtained by HTSeq count (version 0.9.1) (37). The concentration of RNA samples can be quickly measured using NanoDrop spectrophotometers and the purity of RNA can be assessed by absorption ratios (A260/A280 and A260/A230). An A260/A280 ratio close to 2.0 usually indicates less protein contamination, while an A260/A230 ratio greater than 2.0 indicates less pollution from organic solvents or other impurities.

Differential expression analysis was performed using the DESeq (version 1.18.0) R package (38). P<0.05 and FC >1.5 or <0.5 were set as the threshold for significantly differential expression. Hierarchical cluster analysis of DEGs was performed to demonstrate the expression pattern of genes in different groups and samples. Gene Ontology (http://geneontology.org) (39) enrichment and KEGG (18) pathway enrichment analysis of DEGs were performed respectively using R (version 4.3.3, Comprehensive R Archive Network) based on the hypergeometric distribution.

Kaplan-Meier plots were used to establish survival curves. Correlations between gene expression levels were assessed using Spearman's correlation coefficient. An unpaired Student's t test was used to evaluate the statistical significance of differences between two independent groups. Ordinary one-way ANOVA with Tukey's multiple comparisons test was used to assess the statistical significance between three independent groups. P<0.05 was considered to indicate a statistically significant difference.

A total of five NFκB inhibitors were identified and their transcript levels were compared at the pan-cancer level: NFKBIB, NFKBID and NFKBIE were significantly overexpressed in HCC tissue compared with normal tissue, and NFKBIA and NFKBIZ expression was not significantly different between HCC and normal tissues (Figs. 1 and S1).

The predictive ability of NFκB inhibitors for HCC were analysed using Kaplan-Meier plots. According to the results, there was no significant association between NFKBIA, NFKBIB, NFKBID and NFKBIZ expression and prognosis (Fig. 2A). However, a low expression of NFKBIE was significantly associated with a better prognosis for patients with HCC, regardless of overall survival, disease specific survival and relapse-free survival, compared with a high expression (Fig. 2B). Significant correlations were demonstrated between the RNA expression level of NFKBIE and several clinical characteristics, including the stage of cancer and tumour protein P53 (TP53) mutation status, cancer stages, tumour grade, nodal metastasis status, patient's weight, and patient's age (Fig. 2C). Data from MERAV and the HPA database also demonstrated that HCC had significantly higher transcriptional and protein levels of NFKBIE than normal tissue samples (Fig. 2D and E).

The 50 genes with the strongest associations were mainly enriched in the ‘Ribosome’, ‘NFκB signalling pathway’ and ‘Primary immunodeficiency’ (Fig. 3A). Heatmaps were created to demonstrate all 50 genes that exhibited the strongest associations with NFKBIE using the LinkedOmics database (Fig. 3B and C). Furthermore, the HCCDB indicated notable differences in the co-expression gene network between HCC and normal tissue (Fig. 3D and E).

V-rel avian reticuloendotheliosis viral oncogene homolog A (RELA), v-rel avian reticuloendotheliosis viral oncogene homolog (REL), conserved helix-loop-helix ubiquitous kinase (CHUK), inhibitor of kappa light polypeptide gene enhancer in B-cells, kinase gamma (IKBKG), inhibitor of kappa light polypeptide gene enhancer in B-cells, kinase beta (IKBKB), and v-rel avian reticuloendotheliosis viral oncogene homolog B (RELB) were mainly enriched in ‘Shigellosis’, ‘Kaposi sarcoma-associated herpesvirus infection’, ‘C-type lectin receptor signaling pathway’, ‘NFκB signaling pathway’, and ‘antifolate resistance’ (Fig. 4A). These proteins also had the strongest interactions with NFKBIE protein (Fig. 4B). In addition, data from Gene Set Cancer Analysis (GSCALite) indicated that NFKBIE suppression notably reduced the PI3K/AKT, TSC/mTOR, and other signaling pathways (Fig. 4C). Moreover, a drug sensitivity analysis using the Genomics of Drug Sensitivity in Cancer and Cancer Therapeutics Response Portal databases revealed that low expression of NFKBIE was correlated with resistance to Z-LLNle-Cho (Fig. 4D) and dabrafenib (Fig. 4E). These results provide new therapeutic ideas for patients with HCC.

In HCC, memory B cells had the greatest expression of NFKBIE, whereas numerous T cells had the lowest expression of NFKBIE (Fig. 5A) (14). Using the TIMER database, B cells and macrophage cells with TP53 mutation had higher levels of immune infiltration. (Fig. 5B). NFKBIE was also significantly negatively associated with endothelial cell and haematopoietic stem cell immune infiltration. Conversely, B cell, M1 macrophage and myeloid dendritic cell immune infiltration were found to be significantly positively correlated with NFKBIE (Fig. 5C-G). Low expression of NFKBIE combined with immune cell infiltration has a better prognosis (Fig. S2).

Silencing NFKBIE regulates HCC through ‘antigen processing and presentation’. To assess the role that NFKBIE serves in HCC, siRNA was used to knockdown NFKBIE expression and then RNA-seq analysis was performed, and a volcanic map of DEGs was shown (Fig. 6A). Among the top 20 elevated genes, the ‘P53 signalling pathway’ was significantly enriched (Fig. 6B). It was demonstrated that knockdown of NFKBIE expression significantly affected ‘antigen presentation’ (HSA04612) and ‘hepatocellular carcinoma’ (HSA05225) (Fig. 6C). Additionally, among the top 20 elevated genes, the ‘longevity-regulating pathway’ was significantly enriched (Fig. 6D). Similar results were demonstrated using data from the HPA database, and NFKBIE was notably associated with ‘antigen presentation’ with high confidence (Fig. 6E).

The expression of NFKBIE was notably the lowest in Huh-7 cells and the highest in HCCLM3 cells (Fig. 7A). After siRNA-mediated reduction in NFKBIE expression (Fig. 7B), the proliferation and migration of HCCLM3 and Huh-7 cells were significantly reduced, compared with the control (Fig. 7C and D). Furthermore, IHC of tissue microarray demonstrated NFKBIE overexpression in different stages of HCC (Figs. 7E and S3).