Introduction

Malignant melanoma (MM) is an aggressive tumor characterized by high invasiveness, metastatic potential, and mortality rates. Epidemiological data indicate that melanoma accounts for approximately 3% of all malignant tumors and 7%–20% of skin cancers. In recent years, the incidence of MM has significantly risen, particularly in Western developed countries, where the annual increase in cases is estimated at 4%–6%, second only to lung cancer1. The poor prognosis of MM is largely attributed to its high invasiveness, metastatic potential, and early onset of metastasis2. While early surgical excision can achieve a 99% 5-year survival rate, this dramatically decreases by at least 20% once distant metastasis occurs3. For metastatic melanoma, treatment often involves a combination of radiotherapy, chemotherapy, immunotherapy, and targeted therapies, which aim to address resistance mechanisms and minimize adverse side effects4,5. Immunotherapy has emerged as a promising strategy for triggering anti-tumor immunity, significantly improving clinical outcomes when used alone or in combination. However, challenges remain due to severe side effects and the development of primary and acquired resistance6,7.

Hypoxia is a hallmark of solid tumors, including melanoma, and induces adaptive responses such as the upregulation of glycolytic enzymes and angiogenesis factors, mediated by hypoxia-inducible factors (HIFs)8. This metabolic reprogramming enables melanoma cells to survive in nutrient-limited conditions by enhancing autophagy and nutrient scavenging9. Immune cells within the tumor microenvironment also modulate melanoma metabolism via paracrine signaling and metabolic competition, contributing to tumor progression and treatment resistance10. HIF proteins are heterodimers composed of oxygen-regulated α subunits (HIF-1α, HIF-2α, and HIF-3α) and a stable β subunit (aryl hydrocarbon receptor nuclear translocator, or ARNT). HIF-2α, the second member of this family, is implicated in a variety of cancers, including renal, pancreatic, hepatocellular, colorectal, lung, neuroblastoma, osteosarcoma, breast, bladder, and oral squamous cell carcinomas11. It plays a critical role in tumor proliferation, metastasis, angiogenesis, chemotherapy and radiotherapy resistance, and poor clinical outcomes12,13. Recent advancements in gene chip technologies have enabled the detection of HIF-2α expression across multiple cancers, providing insights into its role in cancer initiation and progression14. Bioinformatics analysis of microarray data has further established a strong correlation between elevated HIF-2α expression and poor prognosis in various cancers. The integration of gene chip technology and bioinformatics offers a powerful tool for analyzing cancer-related genes. However, research on the role of HIF-2α in melanoma remains limited. This study utilizes public databases and bioinformatics analysis to investigate the aberrant expression of HIF-2α in melanoma, exploring its potential mechanisms, associated signaling pathways, and impact on the tumor microenvironment. Additionally, the study evaluates the potential of HIF-2α as a novel diagnostic and therapeutic target for melanoma, offering new perspectives for clinical treatment.

Materials and methods

Data acquisition

Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/info/datasets.html), maintained by the National Center for Biotechnology Information (NCBI), was used for data acquisition. Single-cell RNA sequencing data from GSE72056 were downloaded from the public GEO database for analysis. Additionally, transcriptome data from GSE98394, GSE112509, and GSE15605 were obtained. A total of 203 samples were included, consisting of 153 primary melanoma cases and 50 benign skin biopsies (melanocytic nevi). The expression differences of HIF2α were analyzed after data integration and normalization. Furthermore, processed gene expression data from The Cancer Genome Atlas (TCGA) Skin Cutaneous Melanoma (SKCM) cohort15 (https://portal.gdc.cancer.gov/) were downloaded. This database, as the largest cancer genomics resource, contains a comprehensive collection of gene expression, miRNA expression, lncRNA expression, copy number variation, DNA methylation, and SNP data. A total of 473 SKCM samples were included for transcriptome analysis.

Differential expression analysis

The Limma package, an R software tool for differential expression analysis, was used to identify genes that were significantly differentially expressed between groups. Molecular differences between melanoma and control samples were analyzed using the “Limma” package in R. Differentially expressed genes were identified by comparing the melanocytic nevi and melanoma groups, with a threshold of P-value < 0.05 for statistical significance. Volcano plots and heatmaps were generated to visualize the differential expression of genes.

Single-cell data quality control

During data preprocessing, the expression profiles were first read using the Seurat package. Cells were filtered based on several criteria: total UMI count, number of expressed genes, and the proportions of mitochondrial and ribosomal gene expression. The mitochondrial and ribosomal gene expression proportions refer to the percentage of total gene expression attributed to mitochondrial and ribosomal genes, respectively. Cells with a high proportion of mitochondrial or ribosomal gene expression, indicative of low RNA content and potential entry into cell death processes, were considered low-quality and excluded from further analysis16. Additionally, the median absolute deviation (MAD) was employed as a quality control indicator. According to established guidelines, values exceeding 3 MAD from the median were considered outliers and removed from the dataset17. Following these steps, the quality control of the single-cell data was completed.

Dimension reduction, clustering, and cell annotation

The LogNormalize method was applied for global standardization of the expression data. Specifically, each cell’s total expression was scaled to 10,000 transcripts, followed by a log transformation. The cell cycle score was calculated using CellCycleScoring. The FindVariableFeatures function was used to identify highly variable genes, and the ScaleData function was employed to correct for gene expression fluctuations caused by differential mitochondrial gene expression, ribosomal gene expression, and cell cycle variations. Principal component analysis (PCA) was performed on the expression matrix using the RunPCA function, and relevant principal components were selected for subsequent analysis. UMAP (Uniform Manifold Approximation and Projection) was applied for nonlinear dimensionality reduction18. Cell types were identified and annotated by querying CellMarker19, PanglaoDB, and relevant literature. Automated annotation was also conducted using SingleR software to assign cell types and corresponding marker genes from the tissues under study.

Ligand–receptor interaction analysis

CellChat20 was used for quantitative inference and analysis of intercellular communication networks derived from single-cell data. This tool employs network analysis and pattern recognition methods to predict the major signaling inputs and outputs of cells and to explore how cells coordinate their functions within the tumor microenvironment. In this analysis, standardized single-cell expression profiles were used as input, with cell subtypes obtained from single-cell analysis serving as the basis for examining cell-to-cell interactions. Interaction strength was quantified using two indicators: interaction weights and interaction counts (frequency). These metrics allowed us to evaluate the degree of activity and mutual influence of each cell type within the disease context.

Gene set enrichment analysis (GSEA)

Gene Set Enrichment Analysis (GSEA) is a method used to determine whether predefined gene sets are significantly enriched in high-throughput gene expression data21. GSEA compares gene expression profiles between different conditions to identify biological pathways or functions that are up-regulated or down-regulated in specific groups. Patients were classified into high and low expression groups based on the expression levels of key genes. Differences in signaling pathways between these groups were further analyzed using GSEA. The background gene set used was the version 7.0 annotated gene set from the MsigDB database22, which includes pathway-specific annotations. Differentially enriched pathways between groups were identified, and gene sets with an adjusted P-value < 0.05 were ranked based on their consistency score. GSEA is commonly used to explore the correlation between disease classification and biological significance.

Gene set variation analysis (GSVA)

Gene Set Variation Analysis (GSVA) is a nonparametric, unsupervised method used to assess gene set enrichment in transcriptomic data, aiming to reveal differences in biological function between samples23. GSVA scores gene sets by converting the expression levels of individual genes into pathway-level changes, thus assessing the activity of specific biological processes or signaling pathways. This method does not require predefined sample groupings and can be applied to continuous data, making it particularly suitable for various high-throughput gene expression analyses. GSVA calculates the activity of each sample within a specified gene set and generates a score matrix representing the biological pathway activity of each sample. These scores reflect the underlying biological characteristics of the samples and can provide insights into disease classification, prognostic assessment, and biomarker discovery. Given its flexibility and unsupervised nature, GSVA has become an essential tool in modern bioinformatics. In this study, gene sets were downloaded from the Molecular Signatures Database20, and GSVA was used to score each gene set to evaluate potential changes in biological function across different samples.

Key gene-related non-coding RNA networks

MicroRNAs (miRNAs) are non-coding RNA molecules approximately 22 nucleotides in length, which regulate gene expression by binding to the 3’ untranslated region (UTR) of target mRNAs, leading to their degradation or inhibition of translation. miRNAs are involved in numerous biological processes, including cell proliferation, differentiation, apoptosis, and metabolism24. To further investigate the regulatory roles of miRNAs in key genes, we identified key gene-associated miRNAs using the miRCode database and visualized the gene-miRNA interaction network using Cytoscape software.

PCR validation of the impact of hypoxia and HIF2α silencing on melanoma A375 cells EMT

Cell Grouping and Culture: The A375 cells were provided by Jiangsu KeyGEN BioTECH Co.,Ltd. Normoxia group: A375 cells were cultured in DMEM supplemented with 10% FBS, 100 U/mL penicillin, and 100 μg/mL streptomycin. Cells were maintained at 37℃ in a 5% CO2 incubator. Hypoxia group: Cells were cultured in a hypoxia incubator under 95% N2 and 5% CO2 conditions, at 37℃. The cells were continuously exposed to this hypoxic environment for 12 h.

ShRNA-NC and ShRNA-HIF2α Transfection:Transfections were carried out following the manufacturer’s protocol for the Lipo3000 transfection reagent. A mixture of 2 μg plasmid DNA and 4μL Lipo3000 was prepared in 250μL serum-free medium and incubated at room temperature for 20 min. Meanwhile, the culture medium in 6-well plates was replaced with 2 mL serum-free medium. After incubation, the plasmid-Lipo3000 mixture was added to the wells, thoroughly mixed, and incubated for 6 h. Subsequently, the medium was replaced with serum-containing medium, and cells were cultured for an additional 24 h. To evaluate transfection efficiency, a GFP-labeled control plasmid was co-transfected with the target plasmid. 24 h post-transfection, cells were visualized under a fluorescence microscope, and multiple random fields were imaged. Transfection efficiency, calculated as the percentage of GFP-positive cells among total cells using ImageJ software, remained stable at 80%-85% across diverse experimental groups.

Real-Time Quantitative PCR Analysis of HIF-2α, E-cadherin, N-cadherin, and Vimentin Expression:Total RNA was extracted from the processed cells using the Trizol method, followed by reverse transcription to cDNA. PCR was performed using 1μL cDNA, 7μL 0.1% DEPC-treated water, 10μL SYBR PCR Master Mix, 1μL forward primer, and 1μL reverse primer, for a final volume of 20μL per reaction. Reactions were performed in triplicate for each gene. PCR conditions were as follows: 95℃ for 30 s (initial denaturation), followed by 40 cycles of 95℃ for 5 s (denaturation), 60℃ for 30 s (annealing), and 72℃ for 40 s (extension). Cycle threshold (Ct) values were determined based on amplification curves. Relative gene expression was calculated using the 2-ΔΔCt method, with GAPDH as the internal control. Primers used are listed in Table 1.

Table 1 Primers used for quantitative PCR.
Full size table

Statistical analysis

All statistical analyses were performed using R software (version 4.3.0), with a significance threshold set at P < 0.05. PCR data were analyzed using SPSS 20.0 software. Data are expressed as the mean ± standard deviation (\({\overline{\text{x}}} \pm {\text{s}}\)). One-way ANOVA was used to assess overall statistical differences, with pairwise comparisons performed using the SNK test. A P-value of < 0.05 was considered statistically significant.

Results

Differential gene expression analysis

Differential gene expression between the control and disease groups was analyzed using the limma package. Genes with a p-value < 0.05 were considered significant. A total of 7,053 differentially expressed genes were identified, including 3,721 upregulated genes and 3,332 downregulated genes. Volcano plots of all differentially expressed genes and heatmaps of the top 20 most significantly different genes are presented (Fig. 1A,B).

Fig. 1
figure 1

(A) Volcano plot of differentially expressed genes. (B) Heatmap of the top 20 differentially expressed genes with significant differences. (C) Differential expression of HIF2α between disease and control samples. (D) UMAP plot of single-cell annotation. (E) Bubble plot of marker genes for each single-cell subtype. (F) Proportion of each single-cell subtype represented in a bar chart.

Full size image

Expression of HIF2α in tumors

To investigate the expression of HIF2α in melanoma, transcriptome data from GSE98394, GSE112509, and GSE15605 were downloaded from the GEO database. The dataset comprised 203 samples, including 153 cases of primary melanoma and 50 cases of benign skin biopsies (melanocytic nevi). After combining and correcting the data, a significant difference in HIF2α expression between melanocytic nevi tissues and melanoma tissues was observed, with significantly higher levels of HIF2α in tumor tissues (Fig. 1C).

Single-cell data processing and HIF2α expression in single cells

To eliminate low-quality cells and reduce data noise, we considered the data quality across multiple samples during single-cell quality control. Cells with fewer than 50 captured genes were filtered out. The filtering criteria were as follows: (nFeature_RNA > 50 & percent.mt ≤ median + 3MAD & nFeature_RNA ≤ median + 3MAD & nCount_RNA ≤ median + 3MAD), where nFeature_RNA represents the number of detected genes, nCount_RNA refers to the total UMI count of the cell, and percent.mt denotes the percentage of mitochondrial reads. A total of 4,531 cells were retained. Following data normalization, homogenization, and PCA processing to minimize technical noise and batch effects, the single-cell data were clustered into 14 distinct clusters using UMAP. These clusters were annotated into six cell types: T cells, B cells, Melanocytes, Macrophages, Fibroblasts, and Endothelial cells (Fig. 1D). A bubble plot of marker genes for each cell type (Fig. 1E) and a bar plot showing the cell proportions (Fig. 1F) are provided. We then used DotPlot, FeaturePlot, and VlnPlot functions from the Seurat package to examine the expression of HIF2α in single cells (Fig. 2A,B) and analyzed the variation in HIF2α expression across different cell subtypes (Fig. 2C).

Fig. 2
figure 2

HIF2α expression in single cells. (A) Expression abundance of HIF2α in single cells. (B) Expression of HIF2α across different single-cell subtypes. (C) Differential abundance of HIF2α-active cells across single-cell subtypes.

Full size image

Ligand–receptor interaction analysis (CellChat)

Single cells were classified into HIF2α-positive and HIF2α-negative groups based on HIF2α expression using the CellChat package. The ligand–receptor interactions between these groups were then analyzed based on the single-cell expression profiles. Complex interactions were observed between HIF2α-positive and HIF2α-negative cell subtypes (Fig. 3A,D). Among these, Macrophages with high HIF2α expression exhibited significant interactions with other cell types (Fig. 3B), with the relevant ligands depicted (Fig. 3C). For Macrophages lacking HIF2α, stronger interactions with other subtypes were observed (Fig. 3E), and the corresponding ligands are shown (Fig. 3F).

Fig. 3
figure 3

Cell–cell communication in HIF2α-active cells. (A) Network diagram of interactions between HIF2α-positive single-cell subtypes. (B) Interaction frequencies among HIF2α-positive single-cell subtypes. (C) Ligand-driven interactions between HIF2α-positive macrophages and other HIF2α-positive single-cell subtypes. (D) Network diagram of interactions between HIF2α-negative single-cell subtypes. (E) Interaction frequencies among HIF2α-negative single-cell subtypes. (F) Ligand-driven interactions between HIF2α-negative macrophages and other HIF2α-negative single-cell subtypes.

Full size image

Immune regulatory network of HIF2α

We analyzed the relationship between HIF2α and various immune factors, including immunosuppressors, immunostimulators, chemokines, and receptors. Using Pearson correlation analysis, we found that HIF2α was significantly positively correlated with numerous immune factors, suggesting that HIF2α may contribute to the formation of a complex immunoregulatory network within the tumor microenvironment. This network could both promote anti-tumor immune responses and maintain an immunosuppressive state. These findings imply that HIF2α not only plays a role in immune cell infiltration but may also influence immune escape mechanisms, offering valuable insights for potential tumor immunotherapy applications (Fig. 4A–E).

Fig. 4
figure 4

Correlation between HIF2α and various immune factors. (A) Correlation between HIF2α and chemokines. (B) Correlation between HIF2α and receptors. (C) Correlation between HIF2α and MHC. (D) Correlation between HIF2α and immunosuppressive factors. (E) Correlation between HIF2α and immune stimulatory factors.

Full size image

Signaling pathways involved in HIF2α

We performed GSEA and GSVA to investigate the specific signaling pathways associated with HIF2α expression and explore its molecular mechanisms in melanoma progression. GSEA results revealed that HIF2α is enriched in pathways related to the ERK1/2 cascade, positive regulation of autophagy, p38 MAPK cascade, and hypoxia-related pathways (Fig. 5A). GSVA analysis further highlighted enrichment of KRAS signaling, IL2_STAT5 signaling, and autophagy pathways in HIF2α-positive samples (Fig. 5B). These results suggest that HIF2α influences melanoma progression through its regulation of tumor microenvironment interactions. Additionally, to investigate the association between HIF2α and metabolic pathways, ssGSEA scores were calculated for all cells, and a heatmap of metabolic pathways was generated based on HIF2α activity (Fig. 5C). The analysis revealed that HIF2α-positive cells exhibited higher activity in specific metabolic pathways, such as those associated with C3 metabolism.

Fig. 5
figure 5

Signaling pathways involving HIF2α. (A) GSEA analysis of HIF2α. (B) GSVA analysis of HIF2α. (C) Heatmap of metabolic pathways involving HIF2α.

Full size image

Comprehensive network analysis of disease gene regulation and clinical correlation

We investigated the correlation between HIF2α and genes associated with hypoxia, autophagy, and epithelial–mesenchymal transition (EMT) pathways. Genes related to hypoxia and autophagy (e.g., ATG5, ATG7, BECN1) were identified from the REACTOME_AUTOPHAGY pathway, while EMT-related genes (e.g., CDH1, VIM, SNAI1) were extracted from the literature. The correlations between HIF2α and these genes were visualized as scatter plots (Figs. 6A–H and 7A–I). Co-expression networks of HIF2α with genes involved in these pathways were constructed for single cells (Figs. 8A–J and 9A–I).

Fig. 6
figure 6

Correlation between HIF2α and genes involved in hypoxia and autophagy regulation pathways from transcriptomic data. (A) Correlation of HIF2α and ATG5. (B) Correlation of HIF2α and ATG7. (C) Correlation of HIF2α and ATG9A. (D) Correlation of HIF2α and BECN1. (E) Correlation of HIF2α and PINK1. (F) Correlation of HIF2α and SQSTM1. (G) Correlation of HIF2α and TNFSF10. (H) Correlation of HIF2α and BNIP3.

Full size image
Fig. 7
figure 7

Correlation between HIF2α and genes involved in epithelial–mesenchymal transition (EMT) regulation pathways from transcriptomic data. (A) Correlation of HIF2α and BMP6. (B) Correlation of HIF2α and HIF1A. (C) Correlation of HIF2α and GAS5. (D) Correlation of HIF2α and IL6. (E) Correlation of HIF2α and SNAI1. (F) Correlation of HIF2α and SNAI2. (G) Correlation of HIF2α and STAT3. (H) Correlation of HIF2α and TGFB3. (I) Correlation of HIF2α and TNF.

Full size image
Fig. 8
figure 8

Co-expression of HIF2α with genes involved in hypoxia and autophagy regulation pathways in single cells. (A) Co-expression of HIF2α and ATG5. (B) Co-expression of HIF2α and ATG7. (C) Co-expression of HIF2α and BECN1. (D) Co-expression of HIF2α and MAP1LC3A. (E) Co-expression of HIF2α and PINK1. (F) Co-expression of HIF2α and SQSTM1. (G) Co-expression of HIF2α and ATG9A. (H) Co-expression of HIF2α and CSNK2B. (I) Co-expression of HIF2α and TOMM5. (J) Co-expression of HIF2α and TNFSF10.

Full size image
Fig. 9
figure 9

Co-expression of HIF2α with genes involved in epithelial–mesenchymal transition (EMT) regulation pathways in single cells. (A) Co-expression of HIF2α and CDH1. (B) Co-expression of HIF2α and CDH2. (C) Co-expression of HIF2α and KRT8. (D) Co-expression of HIF2α and STAT3. (E) Co-expression of HIF2α and SNAI2. (F) Co-expression of HIF2α and TWIST1. (G) Co-expression of HIF2α and VIM. (H) Co-expression of HIF2α and HIF1A. (I) Co-expression of HIF2α and TGFB3.

Full size image

Furthermore, miRNA-target relationships for HIF2α were predicted using the miRCode database, yielding 49 miRNAs and 49 corresponding mRNA-miRNA pairs, which were visualized using Cytoscape (Fig. 10A).

Fig. 10
figure 10

Regulatory network involving HIF2α. (A) Reverse prediction of non-coding RNA networks associated with HIF2α.

Full size image

Finally, clinical data from TCGA were analyzed to assess the relationship between HIF2α expression and clinical parameters. We found that HIF2α expression was significantly associated with M stage and T stage, particularly in male patients. Elevated HIF2α levels were observed in advanced T and M stages associated with metastasis (Fig. 11A–F). Additionally, survival analysis indicated a significantly lower survival rate in patients with high HIF2α expression (Fig. 11G).

Fig. 11
figure 11

Differential expression of HIF2α in relation to clinical features and staging. (A) Differential expression of HIF2α across different age groups. (B) Differential expression of HIF2α in relation to survival status. (C) Differential expression of HIF2α across different genders. (D) Differential expression of HIF2α in male patients across T stages. (E) Differential expression of HIF2α in male patients across M stages. (F) Differential expression of HIF2α in male patients across clinical stages. (G) Survival analysis of patients with high vs. low HIF2α expression.

Full size image

Effects of hypoxia and HIF-2α knockdown on HIF-2α,E-cadherin, N-cadherin and vimentin mRNA expression in A375 cells

RT-PCR analysis revealed that compared to normoxic conditions, hypoxia induced a modest yet statistically significant upregulation of the mRNA levels of HIF-2α, N-cadherin, and Vimentin (P < 0.01), while E-cadherin mRNA levels showed a modest but statistically significant downregulation (P < 0.01). No notable changes in gene expression were observed in the NC group under hypoxic conditions. However, HIF-2α knockdown under hypoxia led to a modest but statistically significant decrease in the mRNA levels of HIF-2α, N-cadherin, and Vimentin (P < 0.01), while causing a modest yet statistically significant increase in E-cadherin mRNA levels (P < 0.01) in A375 cells (Table 1, Fig. 12A–D).

Fig. 12
figure 12

Effects of Hypoxia and HIF-2α Knockdown on HIF-2α, E-cadherin, N-cadherin, and Vimentin mRNA Expression in A375 Cells. (A) Hypoxia significantly increased the mRNA levels of HIF-2α in A375 cells (*P < 0.01); In the NC group, no significant changes in mRNA expression were observed under hypoxic conditions. HIF-2α knockdown under hypoxia led to a significant reduction in the mRNA levels of HIF-2αcompared to the NC group(#P < 0.01). (B) Hypoxia induced a modest yet statistically significant reduction in the mRNA levels of E-cadherin in A375 cells (*P < 0.01);In the NC group, no significant changes in mRNA expression were observed under hypoxic conditions.HIF-2α knockdown under hypoxia led to a modest but statistically significant upregulation of E-cadherin mRNA levels, as compared to the NC group(#P < 0.01). (C,D) Hypoxia induced a modest yet statistically significant upregulation in the mRNA levels of N-cadherin and Vimentin in A375 cells (*P < 0.01); In the NC group, no significant changes in mRNA expression were observed under hypoxic conditions.HIF-2α knockdown under hypoxia led to a modest but statistically significant reduction of N-cadherin and Vimentin mRNA levels, as compared to the NC group (#P < 0.01).

Full size image

Disscussion

Recent advances in melanoma treatment, particularly with targeted therapy and immunotherapy, have significantly improved patient survival and disease progression. Targeted therapies, such as BRAF inhibitors (vemurafenib, dabrafenib) combined with MEK inhibitors (trametinib), have delayed progression and reduced resistance in many cases25. Similarly, immune checkpoint inhibitors like PD-1 inhibitors (pembrolizumab, nivolumab) have become standard treatment for advanced melanoma, leading to improved overall survival and, in some cases, long-term remission26. Despite these successes, challenges such as resistance and immune evasion remain significant hurdles27. Beyond the therapeutic approaches themselves, the tumor microenvironment (TME)—particularly its role in immune evasion—has become a critical factor in understanding the efficacy of treatments and mechanisms of resistance.

The tumor microenvironment (TME) is typically characterized by hypoxia, with adaptive responses to hypoxic conditions driven by various transcription factors. Among these, hypoxia-inducible factor (HIF) activation represents a hallmark genetic event in melanoma. The HIF complex comprises an oxygen-regulated α subunit (HIF-1α, HIF-2α, or HIF-3α) and a constitutively expressed, stable β subunit (HIF-1β)28. Hypoxia drives the stabilization and nuclear translocation of HIF-1α and HIF-2α, which subsequently interact with transcriptional coactivators CREB-binding protein (CBP) and p300. These complexes bind to hypoxia response elements (HREs) located within the promoter regions of over 300 target genes29. HIF-1α and HIF-2α share 48% amino acid sequence identity and exhibit similar protein structures, differing primarily in their N-terminal transactivation domain (N-TAD). In contrast, HIF-3α shares similarity with HIF-1α and HIF-2α in its helix-loop-helix and PAS domains but lacks the C-terminal transactivation domain (C-TAD)30. Both HIF-1α and HIF-2α are overexpressed in melanoma31, where their protein expression drives melanoma invasion and aggressiveness. Additionally, evidence suggests that HIF-1α and HIF-2α can independently activate SRC, further promoting melanoma metastasis32. The role of HIF-2α in melanoma remains underexplored, with existing data often inconclusive. In a murine melanoma model, selective stabilization of HIF-2α was found to reduce tumor growth and angiogenesis. This effect was mediated by tumor-associated macrophages, which secreted the soluble form of the VEGF receptor (sVEGFR-1), thereby inhibiting the bioactivity of VEGF33. Hao et al. reported that HIF-2α overexpression induces stemness in melanoma cells by suppressing p21 expression34. However, another study employing mass spectrometry to analyze nuclear interactions of HIF-2α in melanoma cells identified key proteins involved in melanoma progression, such as MITF and SOX10, as HIF-2α-binding partners. Interestingly, HIF-2α was found to reduce the invasive potential of melanoma cells in cell line models35. The discrepancies in HIF-2α's effects could be attributed to differences in cell types, organs, or the balance between various HIF subtypes. Our analysis using TCGA clinical data further explored the relationship between HIF2α expression and clinical outcomes in melanoma patients. We found that high HIF2α expression was associated with increased tumor growth, metastasis, and reduced survival, particularly in male patients. This highlights the potential of HIF2α as a key regulator in melanoma progression and underscores the need for further investigation into its role in therapeutic resistance and immune evasion.

Melanoma cancer stem cells are capable of evading immune surveillance by either eliminating anti-tumor immune cells or selectively recruiting immune cells with immunosuppressive and tumor-promoting properties36,37. As immunotherapy has shown significant promise in treating skin melanoma, there has been a growing focus on the interactions between melanoma cells and infiltrating immune cells38,39. Melanoma cells communicate with immune cells through molecular signals that maintain the TME and affect tumor growth, metastasis, invasiveness, angiogenesis, and drug resistance40,41. These signals include growth factors, chemokines, cytokines, and extracellular vesicles containing mRNA, miRNA, DNA, proteins, and lipids. By releasing these molecules into the extracellular space, tumor cells influence the biological characteristics of surrounding cells42,43. Notably, complex crosstalk between tumor stem cells and their environment is critical for tumor stem cell self-renewal, metabolic maintenance, and resistance to treatment44,45. A recent study by Amy Y. Huang et al.46 identified HIF-2α as a novel hypoxia-related target and tested its potential in combination with anti-PD-1 therapy in preclinical models. This combination therapy delayed tumor growth in high immune models but showed limited efficacy in low immune models. Our study further explored the regulatory role of HIF-2α in cell–cell communication using bioinformatics analysis of single-cell transcriptomic data. We found that HIF-2α regulates the ligand–receptor interactions between macrophages and other cell subtypes. In HIF-2α-positive cells, macrophages were not only the most abundant subtype but also exhibited significantly higher communication complexity and intensity. This enhanced communication is likely driven by HIF-2α, which upregulates inflammation-related factors like VEGF and IL-6, as well as immune-suppressive factors like PD-L1. These findings suggest that HIF-2α may promote macrophage functional polarization, remodeling the TME to support tumor growth and immune evasion. In contrast, in HIF-2α-negative cells, macrophage communication was notably reduced, highlighting the importance of HIF-2α in regulating macrophage interactions. These findings not only elucidate the mechanistic role of HIF-2α within the tumor microenvironment but also provide a potential theoretical basis for developing anti-tumor therapies targeting HIF-2α.

The hypoxic microenvironment is closely associated with poor prognosis in cancer and resistance to radiotherapy and chemotherapy47. Under hypoxia,HIF-2α is stabilized by blocking proline hydroxylation, allowing it to dimerize with ARNT and bind to hypoxia response elements (HRE) to activate genes associated with tumor progression48. HIF-2α upregulates angiogenic factors like VEGF, promoting angiogenesis to supply oxygen and nutrients for tumor expansion49. In melanoma, hypoxia promotes tumor growth and resistance to treatment50. HIFs plays a key role in melanoma progression through multiple signaling pathways that regulate tumor growth, metastasis, and immune evasion51,52. Our study used Gene Set Enrichment Analysis (GSEA) and Gene Set Variation Analysis (GSVA) to explore the molecular mechanisms by which HIF-2α influences melanoma. The results showed that HIF-2α expression is linked to the enrichment of several key pathways, especially those related to autophagy, hypoxic response, and cell cycle regulation. Specifically, GSEA revealed that HIF-2α is enriched in pathways like the ERK1/ERK2 cascade, positive regulation of autophagy, and the modulation of the p38MAPK cascade, all of which are crucial for tumor initiation, progression, and metastasis. Further GSVA analysis showed that HIF-2α is associated with increased activity in the KRAS signaling pathway, IL2-STAT5 signaling pathway, and autophagy pathways. These findings suggest that HIF-2α not only regulates melanoma cell growth but also promotes tumor progression through interactions within the TME, particularly with immune cells and metabolic pathways. For instance, the KRAS signaling pathway is involved in cell proliferation, metastasis, and resistance development, while IL2-STAT5 signaling may influence immune evasion mechanisms in the TME. Additionally, single-cell expression profiling combined with GSVA revealed that HIF2α-positive cells exhibit heightened activity of C3-specific metabolic markers. Correlation analysis further showed that these cells display elevated activity in redox regulation and glycolysis pathways, suggesting that HIF2α influences melanoma cell growth and resistance by reprogramming metabolic states. These findings underscore the multifaceted role of HIF2α in melanoma, spanning cell-intrinsic and microenvironmental mechanisms.

Hypoxia can influence tumor invasiveness by activating signaling pathways involved in invasion, angiogenesis, and epithelial–mesenchymal transition (EMT), with EMT being a critical mechanism of tumor metastasis53. During the EMT, HIFs exerts its influence through various mechanisms, orchestrating the morphological and functional reprogramming of epithelial cells and facilitating their transition from an epithelial to a mesenchymal phenotype54. The results of this study demonstrate high expression of HIF-2α in melanoma. In single-cell analysis, a co-expression network of genes involved in epithelial–mesenchymal transition (EMT) was constructed with HIF-2α. The analysis revealed a negative correlation between HIF-2α expression and CDH1, and a positive correlation with the expression of KRT8, CDH2, VIM, SNAI1, SNAI2, TWIST1, GAS5, and STAT3. These findings suggest that HIF-2α promotes the epithelial–mesenchymal transition (EMT) and accelerates tumor malignancy. PCR results further support that hypoxia induces EMT in melanoma cells, with HIF-2α playing a pivotal role in this process. This underscores HIF-2α as a key mediator of EMT and suggests that targeting it could reduce the invasiveness and metastatic potential of melanoma cells. Studies have demonstrated that common malignant skin tumors, such as cutaneous squamous cell carcinoma and malignant melanoma, exhibit significant autophagy activity55. As an intracellular“clean-up” mechanism, activation of the autophagy pathway enables tumor cells to eliminate damaged or redundant organelles, evade apoptosis, and sustain survival within the tumor microenvironment, particularly under stress conditions such as nutrient deprivation and hypoxia. Hypoxia-induced autophagy plays a crucial role in regulating cell growth, dissemination, and resistance to treatment. Hypoxia regulates the activity of AMPK, mTOR, Beclin-1, and ATGs, thereby controlling autophagy in human malignancies56. The study by Jie Sun et al. demonstrated that hypoxia-induced BNIP3 reprograms cellular metabolism through the regulation of mitophagy, thereby promoting the progression and metastasis of uveal melanoma57. In this study, hypoxia- and autophagy-related genes were identified from the MsigDB database, and co-expression networks between autophagy-associated genes and HIF-2α were constructed using transcriptomic and single-cell analyses. The results revealed a significant positive correlation between HIF-2α and autophagy-related genes, including ATG5, ATG7, ATG9A, BECN1,LC3, PINK1, SQSTM1, and TNFSF10. These findings underscore the role of HIF-2α in regulating autophagy activity in melanoma cells, thereby influencing their biological functions and adaptive mechanisms.

In conclusion, our study strongly suggests the critical role of HIF-2α in melanoma progression. By regulating key processes like autophagy, immune evasion, and tumor metabolism, HIF-2α influences the growth, metastasis, and malignancy of melanoma. The results provide a strong theoretical foundation for targeting HIF-2α in anti-tumor therapies. However, The experimental results presented in this paper still have certain limitations. Further functional experiments and clinical trials are needed to validate these findings and explore the clinical feasibility of HIF-2α as a therapeutic target for melanoma. Future studies could incorporate in vivo models and clinical trials to evaluate the potential of HIF-2α-targeted therapy, offering new strategies for precision melanoma treatment.