Abstract

   Hepatocellular carcinoma (HCC) progression is closely linked to the
   role of macrophages. This study utilized single-cell RNA sequencing and
   genomic analysis to explore the characteristic genes of macrophages in
   HCC and their impact on patient prognosis. We obtained single-cell
   se-quencing data from seven HCC samples in the GEO database. Through
   principal component analysis and t-SNE dimensionality reduction, we
   identified 2,000 highly variable genes and per-formed clustering and
   annotation of 17 cell clusters, revealing 482 macrophage-related
   feature genes. A LASSO regression model based on these genes was
   developed to predict the prognosis of HCC patients, with validation in
   the TCGA-LIHC cohort demonstrating model accuracy (AUC = 0.78, 0.72,
   0.71 for 1-, 3-, and 5-year survival rates, respectively).
   Additionally, patients in the high-risk group exhibited elevated tumor
   stemness scores, although no significant differences were observed in
   microsatellite instability (MSI) and tumor mutational burden (TMB)
   scores. Immune-related analyses revealed that FCER1G expression was
   downregulated in HCC and was associated with key pathways such as
   apoptosis and ferroptosis. Reduced FCER1G expression significantly
   affected HCC cell proliferation and migration. Our prognostic model
   provides new insights into precision and immunotherapy for HCC and
   holds significant implications for future clinical applications.

Supplementary Information

   The online version contains supplementary material available at
   10.1038/s41598-025-88071-8.

   Keywords: Hepatocellular carcinoma, Single-cell RNA sequencing,
   Macrophages, Risk scoring model, FCER1G

   Subject terms: Cancer, Biomarkers, Oncology, Gastrointestinal diseases

Introduction

   Hepatocellular carcinoma (HCC) is the third leading cause of
   cancer-related mortality globally, accounting for 90% of primary liver
   cancer cases^[34]1–[35]3. With approximately 830,000 deaths annually
   from primary liver cancer, incidence and mortality rates are projected
   to rise, potentially exceeding one million new cases per year in the
   near future^[36]4. Most HCC cases are diagnosed at advanced stages,
   rendering curative treatments like surgery unfeasible. As a result,
   despite the promising potential of immunotherapy for advanced HCC,
   patient prognosis remains poor^[37]5,[38]6. The high heterogeneity and
   metastatic frequency of HCC, particularly among resectable patients,
   contribute to recurrent metastasis in 70% of cases post-surgery^[39]7.
   This underscores the need to elucidate molecular mechanisms underlying
   HCC progression and metastasis and to identify biomarkers for prognosis
   and therapy. Such insights will improve diagnostic approaches,
   therapeutic strategies, and personalized precision medicine.

   The tumor microenvironment (TME) in HCC is a complex ecosystem composed
   of various cell types, including tumor cells, immune cells, and
   fibroblasts, which interact to drive tumor invasion, growth, and
   metastasis^[40]8,[41]9. The heterogeneity of the TME and its cellular
   components not only plays a critical role in tumor detection,
   treatment, and prognosis but also represents a source of potential
   biomarkers^[42]10,[43]11. Notably, dysfunction in the innate and
   adaptive immune systems can lead to the formation of an
   immunosuppressive TME, facilitating immune evasion by tumor
   cells^[44]12,[45]13. Macrophages, key components of the TME, have
   garnered significant attention due to their diversity and phenotypic
   plasticity^[46]14. Tumor-associated macrophages (TAMs) are closely
   associated with tumor metastasis, invasion, angiogenesis, and
   immunosuppression^[47]15,[48]16. TAMs are typically classified into
   pro-inflammatory, anti-tumor M1-type and anti-inflammatory,
   pro-tumorigenic M2-type macrophages^[49]17,[50]18. Identifying
   TAM-specific biomarkers is essential for targeted immunotherapy, though
   research in this area remains relatively limited^[51]19.

   In this study, we performed cell classification and annotation on
   single-cell RNA sequencing (scRNA-seq) data to identify distinct cell
   subtypes within HCC, with a focus on macrophage signature genes.
   Combined with the TCGA cohort of HCC, we developed a risk scoring model
   to predict patient prognosis and immunotherapy outcomes. Using various
   analytical techniques, we identified and characterized the immune and
   clinical features of patients classified as high- or low-risk based on
   the risk score. Additionally, we explored the prognostic role of risk
   score-related genes in HCC and, most importantly, validated in vitro
   that suppression of the key gene FCER1G enhances HCC cell proliferation
   and migration.

Materials and methods

Data acquisition

   Transcriptomic and clinical data for liver cancer were obtained from
   The Cancer Genome Atlas (TCGA) ([52]https://portal.gdc.cancer.gov/,
   accessed August 13, 2024), comprising 374 liver cancer and 50 normal
   samples. Clinical variables included survival time, survival status,
   age, gender, and TNM stage. Additionally, the cRNA-seq dataset
   [53]GSE112271, containing tumor samples from seven regions of two
   patients, was retrieved from the Gene Expression Omnibus (GEO) database
   ([54]https://www.ncbi.nlm.nih.gov/, accessed August 13, 2024).

Single-cell sequencing analysis

   Single-cell RNA sequencing (scRNA-seq) was conducted for
   high-resolution transcriptomic profiling at the single-cell level. Data
   were rigorously filtered to remove low-quality samples, followed by
   further analysis using the Seurat package^[55]20. Dimensionality
   reduction techniques such as PCA and t-SNE were employed for clustering
   on default parameters, and cell types were annotated using the SingleR
   package for precise identification of key genes^[56]21.

Pseudotime and cell-cell interaction network analysis

   Pseudotime analysis of tumor cell subpopulations was performed using
   Monocle3 (v0.2.3)^[57]22, and results were visualized in
   two-dimensional UMAP^[58]23. The CellChat package was employed to
   assess ligand-receptor interactions in scRNA-seq data on default
   parameters^[59]24, identifying potential signaling pathways based on
   ligand-receptor expression patterns across cell subtypes.

Model construction and validation

   Key prognostic genes were identified using univariate Cox regression,
   and a prognostic model was built with Lasso regression^[60]25. Risk
   scores were calculated for each patient, enabling comprehensive
   evaluation of their prognostic relevance in liver cancer. Kaplan-Meier
   survival analysis was conducted for high- and low-risk groups, and
   differential gene expression was illustrated with heatmaps. Predictive
   performance was assessed using ROC curves, and a nomogram was developed
   to estimate 1-, 3-, and 5-year patient prognoses. Calibration curves
   were employed to validate the accuracy of the clinical model.

Functional enrichment analysis and protein-protein interaction network
construction

   Functional enrichment of ubiquitination-related genes (URG) was
   performed using DAVID (v.6.8, [61]https://david.ncifcrf.gov/, accessed
   August 13, 2024) with p < 0.05 and FDR < 25%, where the KEGG database
   comes from Kanehisa laboratory^[62]26. The STRING database (v.11.0,
   [63]https://string-db.org, accessed August 13, 2024) was used to
   construct the PPI network.

Immunoassays

   The Cibersort algorithm quantified the infiltration of 22 immune cell
   types in each liver cancer patient, and the relationship between immune
   cells and risk scores was analyzed^[64]27,[65]28. Specifically, the
   gene expression profile was uploaded to the Cibersort web tool for
   calculation to obtain the immune score of each sample
   ([66]https://cibersortx.stanford.edu/,, accessed August 13, 2024). MSI
   and TMB were used to evaluate immunotherapy efficacy in high- and
   low-risk groups^[67]29. A tumor stemness index was introduced to assess
   tumor cell differentiation, and the TIDE algorithm evaluated the
   relationship between risk scores and immunotherapy response. Spearman
   correlation analysis was used to examine the link between risk scores
   and immune checkpoints^[68]30.

Protein structure prediction and gene mutation analysis

   Protein structure prediction was conducted by uploading gene symbols to
   Genecard ([69]https://www.genecards.org/, accessed August 13, 2024),
   and mutations were analyzed by identifying SNPs, ONPs, insertions, and
   deletions. Mutation analysis between high- and low-risk groups helped
   elucidate liver cancer’s molecular mechanisms.

Cell culture and transfection

   The hepatocellular carcinoma cell line HepG2 was kindly donated by
   Professor Tang from the School of Basic Medical Sciences, Chengdu
   Medical College^[70]31. HepG2 cells were cultured in DMEM supplemented
   with 10% FBS (Gibco) and 1% PS (Invitrogen) at 37 °C with 5% CO₂. Cells
   were transfected with siRNA targeting FCER1G using Lipofectamine 2000
   (BGI, China) with a final concentration of 20 nM.: si-FCER1G-F:
   UUCUCAUAGCUGGUUAUAGCU; si-FCER1G-R: CUAUAACCAGCUAUGAGAAAU.

Western blot analysis

   Western blot analysis was performed 48 h after transfection with siRNA.
   HepG2 cells were lysed in 1% SDS lysis buffer, and protein
   concentration was measured with a BCA Protein Assay Kit (Thermo
   Scientific). Immunoblots were probed with primary antibodies ((Snail 2
   (proteintech, 12129-1-AP), Snail 1 (abcam, ab216347), Twist 1
   (proteintech, 25465-1-AP), E-cad (abcam, ab231303), N- cad(abcam,
   ab76011), β-Actin(abcam, ab8226)), and bands were visualized using ECL
   reagent (Smart-Lifesciences).

Real-time quantitative PCR (RT-qPCR)

   RT-qPCR was performed following the method described by Shao et
   al.^[71]32. RT-qPCR was performed using TRIzol-extracted RNA and cDNA
   synthesized with the iScript cDNA Synthesis Kit (Bio-Rad). PCR
   amplification was done with Fast SYBR Green Master Mix (Bio-Rad), with
   18 S as the reference gene. Gene expression levels were calculated
   using the 2 − ΔΔCt method. The 18 S gene served as the reference gene
   for normalization. The relative expression levels of target genes were
   calculated using the 2 − ΔΔCt method (ΔCt = Ct of 18 S − Ct of the
   target gene). Primer sequences used were as follows: FCER1G-F:
   TTCCAGCAGTGGTCTTGCTC; FCER1G-R: TGCCTTTCGCACTTGGATCT; 18 S F:
   GTAACCCGTTGAACCCCATT; 18 S R: CCATCCAATCGGTAGTAGCG.

Colony formation assay

   Long-term cell survival was assessed using the colony formation assay.
   Briefly, 1000 cells were seeded in 6-well plates and allowed to grow
   for two weeks. Cells were fixed with 4% paraformaldehyde for 15 min and
   stained with 0.5% (w/v) crystal violet (Sigma-Aldrich) for
   visualization. Colonies were scanned using an Odyssey Scanner (LI-COR,
   Lincoln, NE, USA) and quantified using Image J software.

Cell viability assay

   Cell viability was measured 48 h post-transfection using the Cell
   Counting Kit-8 (CCK-8) assay (MedChem Express, Monmouth Junction, NJ,
   USA). Si-NC or si-FCER1G cells were seeded into 96-well plates at a
   density of 2 × 10^4 cells per well and incubated for 12 h.
   Subsequently, 100 µL of DMEM and 10 µL of CCK-8 solution were added to
   each well and incubated for 1 h. Absorbance at 450 nm was measured
   using a microplate reader (Thermo Scientific, Waltham, MA, USA). All
   experiments were performed in triplicate.

Wound healing assays

   Cell migration was assessed using the wound healing assay. Si-NC or
   si-FCER1G cells were seeded into 6-well plates at a density of 4 × 10^5
   cells per well and incubated for 12 h. Upon reaching 70–80% confluence,
   a wound was created using a 200-µL pipette tip. The cells were cultured
   in DMEM medium supplemented with 1% FBS. Wound closure was monitored at
   0, 12, and 24 h using an inverted microscope.

Statistical analysis

   Differences between groups were analyzed using the Wilcoxon test, and
   correlations were assessed using Spearman’s rank correlation test.
   Survival analysis was performed using Kaplan-Meier curves and the
   log-rank test. Cox regression analysis was conducted with the
   “survival” package in R to calculate hazard ratios (HR) and 95%
   confidence intervals (CI). All statistical tests were two-sided, with
   statistical significance set at P < 0.05. Analyses were conducted using
   R software (version 4.2.2).

Results

Single-cell sequencing analysis

   We obtained single-cell sequencing data from the GEO database
   ([72]GSE112271), comprising seven liver cancer samples for further
   analysis. Initially, we examined each sample’s nFeature and nCount and
   calculated the percentage of mitochondrial gene expression relative to
   total gene expression (Fig. [73]1A-D). Based on these metrics, we
   applied stringent filtering criteria (nFeature_RNA > 50 &
   percent_MT < 5%) to select high-quality cells, ensuring reliable
   downstream analyses. Next, we identified the top 2,000 highly variable
   genes exhibiting significant expression differences across cells, which
   were critical for cell type identification (Fig. [74]1E). Principal
   component analysis (PCA) and t-SNE dimensionality reduction were then
   performed. Clustering analysis successfully segregated the samples into
   17 distinct cell clusters (Fig. [75]1F, G), with the top 10
   heterogeneous genes specifically expressed in these clusters
   (Fig. [76]1H). Further annotation revealed six major cell types:
   adipocytes, endothelial cells, epithelial cells, erythrocytes,
   fibroblasts, and macrophages (Fig. [77]1I). A volcano plot visualized
   the key differential genes between the cell clusters (Fig. [78]1J).
   Cluster 3, mainly composed of macrophages, was characterized by
   differential expression of genes such as PILRA, TREM2, IGFBP7, and
   CRIP2.

Fig. 1.

   [79]Fig. 1
   [80]Open in a new tab

   Single-cell sequencing analysis identifies macrophage-related genes.
   (A–D) Violin plots depicting the distribution of nFeature_RNA,
   nCount_RNA, percent_MT, and percent_HC across 10 HCC samples. (E)
   Volcano plot illustrating the top 1,500 highly variable genes. (F)
   Heatmap showing the expression of key differential genes across
   different clusters. (G) PCA and t-SNE dimensionality reduction
   separating cells into 17 distinct clusters. (H) Bubble plot
   highlighting the main differential genes in various cell types. (I)
   Cell annotation using the SingleR package, categorizing cells into 8
   major groups. (J) Volcano plot displaying key differential genes across
   different cell types.

Extraction of macrophage-specific genes and cell-cell communication analysis

   Recognizing the critical role of macrophages in liver cancer
   development, we identified 482 genes closely associated with
   macrophages (Fig. [81]2A). Pathway enrichment analysis revealed
   significant involvement of these genes in tuberculosis, phagosome, and
   hematopoietic cell lineage pathways (Fig. [82]2B), predominantly
   localized to the Golgi membrane and participating in cytokine-mediated
   signaling pathways (Fig. [83]2C). A protein-protein interaction network
   constructed using STRING ranked key genes through Cytohubba’s EPC,
   Degree, and MNC algorithms on default parameters, with CD4 identified
   as the central node, followed by FCER1G (Fig. [84]2D-G). Cell-cell
   communication networks inferred via CellChat highlighted the importance
   of the MHC-II signaling pathway mediated by CD4 in macrophage
   communication (Fig. [85]2H, I, J, K). Additionally, adhesion molecules
   such as CD99 and ITGB2, previously identified in pathway enrichment,
   played essential roles in macrophage interaction with other cells
   (Fig. [86]2L, M).

Fig. 2.

   [87]Fig. 2
   [88]Open in a new tab

   Extraction of macrophage-specific genes and cell-cell communication.
   (A) Volcano plot of 482 differentially expressed genes in macrophage
   clusters. (B) KEGG pathway enrichment analysis of the 482 genes. (C) GO
   term enrichment analysis of the 482 genes. (D) Protein-protein
   interaction network for the 482 genes. (E–G) Central nodes identified
   in the protein interaction network using the EPC, Degree, and MNC
   topological network methods, respectively. (H–I) Cell-cell
   communication networks among 6 cell types, with connections colored by
   communication quantity (H) or strength (I). (J) Specific
   ligand-receptor interactions within macrophage clusters. (K–M) MHC-II
   signaling pathway (K), CD99 signaling pathway (L), and ITGB2 signaling
   pathway (M) in macrophage interactions with other cells.

Construction of a prognostic model based on macrophage-specific genes using
the LASSO-COX algorithm

   To identify macrophage-specific genes with prognostic relevance, we
   conducted univariate survival analysis, narrowing down 482 genes to 76
   (P < 0.01) (Fig. [89]3A). Using LASSO regression, we selected 10 key
   genes—P2RY6, DLGAP3, TNFRSF11A, FCER1G, LAMB1, NCF2, CSF3R, PLA2G7,
   STC2, and HMOX1—to construct a prognostic risk score (λ = 0.7145)
   (Fig. [90]3B, C).

Fig. 3.

   [91]Fig. 3
   [92]Open in a new tab

   Development of a risk score model using the LASSO algorithm and
   integration with clinical parameters. (A) Univariate Cox analysis
   identifying 76 prognostic genes. (B, C) Gene selection using the
   LASSO-Cox algorithm for constructing the optimal model. (D) Expression
   patterns and survival time relationships of 10 risk score genes. (E)
   Kaplan-Meier analysis demonstrating poorer prognosis in the high-risk
   group compared to the low-risk group. (F) ROC curve analysis showing
   AUC values for the risk score model at 1, 3, and 5 years. (G) Nomogram
   integrating risk scores and clinical factors to predict 1-, 3-, and
   5-year survival rates. (H) Kaplan-Meier analysis indicating worse
   prognosis for patients with higher nomogram scores. (I) ROC curve
   showing AUC values for the nomogram scores at 1, 3, and 5 years. (J)
   Calibration curve demonstrating the model’s accuracy in predicting
   patient survival rates.
   graphic file with name M1.gif

   Heatmap analysis revealed generally consistent expression patterns for
   the selected genes, except for RAMP3 (Fig. [93]3D). Comparative
   survival analysis indicated significantly higher mortality in the
   high-risk group compared to the low-risk group (Fig. [94]3E). The
   model’s prognostic performance was assessed using ROC curves, with AUC
   values of 0.78, 0.72, and 0.71 for 1-, 3-, and 5-year survival,
   respectively (Fig. [95]3F). A nomogram combining clinical parameters
   with the risk score was developed to predict 1-, 3-, and 5-year
   outcomes (Fig. [96]3G). KM analysis indicated worse outcomes for
   patients with higher nomogram scores (Fig. [97]3H), and further ROC
   analysis showed improved AUC values of 0.79, 0.78, and 0.78 for 1, 3,
   and 5 years, respectively (Fig. [98]3I). Calibration curves confirmed
   the model’s robust predictive performance, and also suggested that the
   1-year survival rate would be overestimated, while the 3- and 5-year
   survival rates would be underestimated in the survival prediction of
   low-risk populations (Fig. [99]3J).

Immune landscape of the risk score model

   We explored the association between the risk score and immune cell
   types by analyzing 22 immune cell populations using the CIBERSORT
   algorithm. The results revealed correlations between the risk score and
   macrophages, monocytes, and resting mast cells (Fig. [100]4A).
   Significant differences were also observed in T cells, macrophages,
   monocytes, and resting mast cells between high- and low-risk groups
   (Fig. [101]4B). High-risk patients exhibited significantly higher tumor
   stemness scores compared to low-risk patients (Fig. [102]4C), while MSI
   and TMB scores showed no differences (Fig. [103]4D, E). Immunotherapy
   response analyses using ESTIMATE and TIDE algorithms revealed a lower
   stromal score in the high-risk group (Fig. [104]4F) and reduced “Merck
   18” score and expected immunotherapy response (Fig. [105]4G, H).
   Additionally, immune checkpoint markers such as IL1A, IL12A, CD80,
   TNFRSF18, SELP, IL1B, HAVCR2, and CD276 were significantly associated
   with the risk score (Fig. [106]4I).

Fig. 4.

   [107]Fig. 4
   [108]Open in a new tab

   Analysis of risk scores and immune landscape. (A) Cibersort algorithm
   calculated immune cell scores for 22 immune cell types in TCGA cohort
   samples, with Spearman correlation analysis against risk scores. (B)
   Differences in immune cell scores between high-risk and low-risk groups
   as defined by risk scores. (C) Higher tumor stemness scores observed in
   the high-risk group. (D, E) No significant differences in MSI and TMB
   scores between high- and low-risk groups. (F) Estimate algorithm
   calculating three immune scores in TCGA cohort samples, with
   differential analysis between risk groups. (G, H) TIDE algorithm
   assessing “Merck 18” scores (G) and expected immune response to
   treatment (H) in high- and low-risk groups. (I) Correlation between
   risk scores and 60 immune checkpoints. * denotes p < 0.05, ** denotes
   p < 0.01, *** denotes p < 0.001.

Prognostic and single-cell analysis of risk score genes

   To further elucidate the role of the 10 risk score genes in
   hepatocellular carcinoma (HCC), Kaplan-Meier (KM) survival analysis was
   conducted to assess their association with patient outcomes. The
   results demonstrated that patients in the TCGA cohort could be
   effectively stratified based on the expression levels of these genes,
   with significant differences in prognosis (Fig. [109]5A). Additionally,
   single-cell t-SNE expression heatmaps revealed distinct expression
   patterns for all 10 genes, with FCER1G showing the highest expression
   level and concentration (Fig. [110]5B). To gain deeper insights into
   their roles in cellular development, we conducted pseudotime analysis
   across six annotated cell types (Fig. [111]5C). Focusing specifically
   on the macrophage developmental trajectory, we identified four distinct
   developmental stages within the macrophage population (Fig. [112]5D,
   E). The top 10 differentially expressed genes along this trajectory
   were highlighted, including ALB, APOA2, HP, RBP4, TTR, APCS, GSTA2,
   HULC, TMSB4X, and TYROBP (Fig. [113]5F). Among the risk score genes,
   PLA2G7, HMOX1, and FCER1G were found to closely align with macrophage
   developmental pathways, suggesting their critical involvement in
   macrophage differentiation and function (Fig. [114]5G).

Fig. 5.

   [115]Fig. 5
   [116]Open in a new tab

   Prognostic significance and single-cell analysis of risk score genes.
   (A) Kaplan-Meier survival analysis of 10 risk score genes. (B)
   Expression distribution of 10 risk score genes in single-cell t-SNE
   plots. (C) 3D pseudotime trajectories of six cell types. (D, E) 2D cell
   developmental trajectories of macrophages, highlighting four main cell
   clusters. (F) Top 10 characteristic developmental genes in macrophage
   developmental trajectories. (G) Expression changes of 10 risk score
   genes during macrophage development.

Bioinformatics analysis of FCER1G’s potential role in HCC

   FCER1G (Fc Epsilon Receptor Ig) is a tetrameric protein-coding gene
   (Fig. [117]6A) whose protein and RNA levels were significantly
   downregulated in HCC tumor samples (Fig. [118]6B, C). Abnormal
   phosphorylation modifications at sites such as S61, Y65, T66, T75, and
   Y76 were observed in HCC tumor samples (Fig. [119]6D), despite no
   differences in immunohistochemistry results (Fig. [120]6E). STRING
   analysis revealed FCER1G’s interactions with MAPK signaling pathway
   proteins, JAK family proteins, and leukocyte differentiation antigens
   (Fig. [121]6F). Gene mutation landscape analysis showed that the low
   expression group of FCER1G had a higher mutation frequency of genes
   such as TP53, SDK1, and NBEA, while the high expression group had a
   higher mutation frequency of genes such as AXIN1, OR2J3, and MYH11
   (Fig. [122]6G). In addition, the transcription level of FCER1G was also
   associated with various pathways, including apoptosis, ferroptosis,
   TP53 pathway, PI3K/AKT pathway, tumor inflammatory response, etc. (
   Fig. [123]6H ).

Fig. 6.

   [124]Fig. 6
   [125]Open in a new tab

   Bioinformatics analysis of FCER1G. (A) Protein structure of FCER1G from
   the PDB database. (B) Downregulation of FCER1G protein levels in liver
   cancer tissues from the CPTAC database. (C) Downregulation of FCER1G
   transcript levels in liver cancer tissues from the TCGA-LIHC cohort.
   (D) Abnormal phosphorylation of FCER1G protein in liver cancer tissues
   from CPTAC phosphoproteomics data. (E) Immunohistochemical detection of
   FCER1G expression in normal and HCC tissues. (F) Protein interaction
   network of FCER1G from the STRING database. (G) Mutation analysis of
   TCGA-LIHC cohort grouped by FCER1G transcript levels. (H) ssGSEA
   algorithm calculating enrichment scores for each pathway, with Spearman
   correlation between FCER1G transcript levels and pathway scores.

Inhibition of FCER1G expression enhances proliferation and migration of HepG2
cells

   To investigate FCER1G’s role in HCC, we knocked down its expression
   using si-FCER1G, confirmed by real-time PCR (Fig. [126]7A). Cell
   viability assays (CCK-8) and colony formation experiments demonstrated
   that FCER1G suppression enhanced HepG2 cell proliferation
   (Fig. [127]7B, C). Additional, parallel experiments in Huh7and PLC/PRF5
   also confirmed this finding (Fig. [128]7D-G). EMT marker analysis
   revealed that si-FCER1G increased Snail 1, TWIST 1, and N-cadherin
   expression, while reducing E-cadherin levels, promoting cell migration
   (Fig. [129]7H). This was confirmed by wound healing assays, which
   showed accelerated wound closure in si-FCER1G-treated HepG2 cells
   (Fig. [130]7I).

Fig. 7.

   [131]Fig. 7
   [132]Open in a new tab

   Effects of FCER1G silencing on HepG2 cell proliferation and migration.
   (A) Quantitative PCR verification of si-FCER1G knockdown efficiency in
   HepG2 cells. (B, C) CCK-8 (B) and colony formation assays (C) showing
   the impact of si-FCER1G on HepG2 cell proliferation. (D, E) CCK-8 (D)
   and colony formation assays (E) showing the impact of si-FCER1G on
   HepG2 cell proliferation. (F, G) CCK-8 (F) and colony formation assays
   (G) showing the impact of si-FCER1G on HepG2 cell proliferation. (H)
   Western blotting of HepG2 cell lines transfected with si-NC or
   si-FCER1G using EMT-related antibodies. (I) Wound healing assay using
   yellow pipette tips on HepG2 cells in 6-well plates, with images and
   relative quantification of wound width at 0, 24, and 48 h.

Discussion

   In this study, we conducted an extensive analysis of single-cell
   sequencing data from liver cancer samples, focusing on the
   identification and functional characterization of macrophage-associated
   genes. We selected high-quality cells, clustering them into 17 distinct
   populations and annotating six major cell types. Among these, the
   macrophage cluster stood out, revealing 482 differentially expressed
   genes. Pathway enrichment and protein-protein interaction network
   analyses highlighted key genes such as CD4 and FCER1G. Furthermore, we
   developed a prognostic model based on macrophage-specific genes using
   the LASSO-COX algorithm, demonstrating significant correlations between
   risk scores, immune landscapes, and patient outcomes.

   While previous studies, such as Chen et al.‘s work on
   ubiquitination-related genes and T-cell exhaustion using RNA sequencing
   data^[133]33, focused on different immune subsets, our study broadens
   this knowledge by identifying macrophage-related prognostic genes in
   liver cancer. These findings provide new insights into the tumor
   microenvironment’s complexity, with genes like CD4 and FCER1G emerging
   as central hubs in cell communication networks. These genes likely
   mediate complex macrophage-immune cell interactions through MHC-II
   signaling, influencing immune responses within the tumor
   microenvironment^[134]34,[135]35. Our macrophage-specific gene-based
   prognostic model demonstrates moderate predictive accuracy (AUC values
   ​​at 1, 3, and 5 years are 0.79.0.78, 0.78, respectively) for
   hepatocellular carcinoma (HCC) patient outcomes, aligning with findings
   from similar studies^[136]33,[137]36–[138]38. There were deviations in
   the 1-, 3-, and 5-year survival predictions, which not only points out
   the shortcomings of this study, but also suggests that future research
   should focus on optimizing and improving the prediction model. Notably,
   immune landscape analysis revealed correlations between risk scores and
   specific immune cell populations, particularly macrophages and T cells,
   suggesting that macrophage-related gene expression may reflect tumor
   immune status and influence prognosis and therapeutic responses.

   FCER1G (Fc Epsilon Receptor Ig), a protein-coding gene linked to
   autoreactive T-cell immunity^[139]39,[140]40, is implicated in diseases
   such as atopic dermatitis and Alzheimer’s disease^[141]41,[142]42. In
   liver cancer, our study highlights FCER1G as a critical factor, with
   its downregulation correlating with tumor progression. The association
   of FCER1G with key pathways, including apoptosis and the P53 signaling
   pathway, suggests a potential tumor-suppressive role^[143]43,[144]44.
   Experimental validation in HepG2 cells demonstrated that silencing
   FCER1G promotes cell proliferation and migration, reinforcing its role
   in inhibiting tumor growth.

   Despite these important findings, some limitations must be
   acknowledged. The single-cell data analyzed were derived from a
   specific cohort, Validation in larger, diverse, and multi-center
   datasets is necessary to confirm. Additionally, while our prognostic
   model is promising, its clinical utility should be further assessed
   through prospective studies. Moreover, additional mechanistic research
   is needed to elucidate the precise roles of FCER1G and other key genes
   in liver cancer biology.

Conclusion

   Our study provides a comprehensive analysis of macrophage-related genes
   in liver cancer, revealing their prognostic significance and potential
   as therapeutic targets. By integrating single-cell sequencing data with
   functional experiments, we have laid the groundwork for future
   investigations into the complex interactions within the tumor
   microenvironment and their implications for liver cancer treatment.

Electronic supplementary material

   Below is the link to the electronic supplementary material.
   [145]Supplementary Material 1^ (1.6MB, pdf)

Author contributions

   Author Contributions: YZ: Conceptualization, Data curation, Writing –
   original draft. DK: Data curation, Methodology, Writing – original
   draft. LJ: Writing – original draft, Software. NL: Writing – original
   draft, Investigation. CW: Conceptualization, Writing – review &
   editing. MQ: Conceptualization, Project administration, Funding
   acquisition Writing – review & editing.

Funding

   This work was supported by The School-level fund of Chengdu medical
   college (CYZYB23-09), the National Natural Science Foundation of China
   (82472328), the Sichuan Science and Technology Program (24ZDYF0165),
   Sichuan Provincial Administration of Traditional Chinese Medicine
   Innovation Team Project (2023ZD02).

Data availability

   Single cell data comes from the [146]GSE112271 dataset of the GEO
   database, bulk RNA data comes from the TCGA project, and proteomics
   data comes from the CPTAC project. All public data are fully described
   in the methods section.

Declarations

Competing interests

   The authors declare no competing interests.

Footnotes

   Publisher’s note

   Springer Nature remains neutral with regard to jurisdictional claims in
   published maps and institutional affiliations.

   Deyu Kong and Yiping Zhang have contributed equally to this work and
   share first authorship.

References