Ginsenosides exhibit potential therapeutic potential in photoaging by regulating keratinocyte ferroptosis

Abstract

Background

Skin photoaging is a chronic inflammatory state induced by ultraviolet (UV) radiation. Ferroptosis has recently emerged as a critical mechanism implicated in the pathogenesis of photoaging. Ginsenosides, active compounds in ginseng, hold potential for treating skin photoaging, but their molecular mechanisms in regulating ferroptosis remain unclear.

Methods

We utilized single-cell RNA sequencing (scRNA-seq) to analyze murine skin after UVB irradiation, focusing on keratinocyte subpopulations and ferroptosis-related genes. Molecular docking assessed ginsenoside binding to ferroptosis regulators. Embryotoxicity of ginsenosides was evaluated using zebrafish embryos. Ferroptosis gene expression in UVB-exposed keratinocytes with or without ginsenoside treatment was analyzed by Western blot and RT-qPCR.

Results

scRNA-seq revealed a significant increase in keratinocyte numbers and ferroptosis signals following UVB irradiation. We observed that UVB exposure reduced SLC7A11 and GPX4 expression while increasing HMGB1 in keratinocytes. Molecular docking experiments further indicated that ginsenosides C-K and C-Mc can target ferroptosis-related proteins, and are non-toxic in zebrafish embryos. Treatment with ginsenosides effectively attenuated skin photoaging by modulating the SLC7A11/GPX4 axis and suppressing HMGB1 expression, thereby reducing UVB-induced reactive oxygen species levels and restoring keratinocyte viability. Additionally, we also identified three differentiation trajectories of keratinocytes and found that bone marrow-derived cell Gas6 signaling may influence ferroptosis via the Axl/Tyro3 pathway.

Conclusion

Our findings suggest that ginsenosides may represent promising therapeutic candidates for alleviating UVB-induced skin photoaging via modulation of the ferroptosis-associated SLC7A11/GPX4 axis and HMGB1 expression, providing novel mechanistic insights and therapeutic strategies against skin photoaging.

Graphical abstract

1. Introduction

Photoaging results from abnormal skin manifestations caused by repeated ultraviolet radiation exposure, characterized by roughness, thickening, dryness, sagging, wrinkle formation and excessive pigmentation [1]. UV radiation damages DNA and promotes extracellular matrix (ECM) degradation through oxidative stress and chronic inflammation, thereby accelerating skin aging [[2], [3], [4]]. Keratinocytes form a protective barrier after proliferation and become resistant to UV radiation [5]. Skin photoaging is closely associated with keratinocytes. However, repeated UV radiation can activate nuclear factor-κB, upregulate proinflammatory cytokines in keratinocytes [6], and release metalloproteinases to degrade ECM [7,8], which in turn causes skin photoaging and potentially induce skin malignant tumors [9]. Therefore, comprehensively elucidating the role of keratinocytes is critical for advancing therapeutic approaches to address skin photoaging.

Keratinocytes undergo various cell death including ferroptosis and autophagy, which contribute to inflammatory skin diseases [10,11]. Ferroptosis is a key cell death induced by UV radiation [12]. Therefore, regulating ferroptosis is expected to be a potential therapeutic strategy for preventing and treating skin photoaging. Currently, scRNA-seq has been employed to investigate the heterogeneity of keratinocytes under certain inflammatory skin conditions [[13], [14], [15]]. The relationship between the regulation of ferroptosis in keratinocytes and skin photoaging remains unclear.

Rare ginsenosides possess notable biological activities [[16], [17], [18], [19], [20], [21]]. Major ginsenosides such as Rk1, Rb1, Rg1, Rg2, and Rg5 promote cell regeneration and protect against oxidative stress in skin cell models [[18], [19], [20], [21]]. However, the minor ginsenosides C-K and C-Mc have relatively better biological activities than the naturally occurring major ginsenosides [16,17]. Previous studies have shown that ginsenosides C-K activates Nrf2-mediated antioxidant pathways to mitigate oxidative damage [16]. Furthermore, ginsenosides C-Mc regulates the MAPK/AP-1/NF-κB pathway to inhibit MMP production while upregulating Nrf2/ARE expression, thereby maintaining endogenous oxidation balance [17]. Nrf2-mediated antioxidant responses play a key role in inhibiting ferroptosis. Ferroptosis induced cell damage and death primarily by inhibiting of the SLC7A11/GPX4 pathway [22]. Knockdown of Nrf2 significantly decreases the expression of SLC7A11 and GPX4, thereby promoting ferroptosis [[23], [24], [25]]. In addition, high mobility group protein b1 (HMGB1), a chromatin-binding nuclear protein, has emerged as a key mediator linking oxidative stress and ferroptosis [26,27]. HMGB1 can exacerbate the ferroptosis process by regulating glutathione metabolism, destabilizing GPX4 and triggering inflammatory responses, and inhibition or knockdown of HMGB1 expression can alleviate ferroptosis and protect cells from injury [28]. However, the precise mechanisms by which C-K and C-Mc regulate ferroptosis in the context of photoaging remain to be fully elucidated.

Our study elucidated the dynamic changes in keratinocytes during photoaging and the regulatory effects of ginsenosides on UVB- irradiated ferroptosis in HaCaT cells. The scRNA-seq and DNA methylation analyses revealed the differentiation trajectories of keratinocyte subsets and elucidated their associations with ferroptosis during photoaging. These analyses also highlighted the complex communication mode between keratinocytes and myeloid cells, in which Gas6 was identified as a key ligand driving the interaction, especially its upregulation in UVB-stimulated samples. Gas6 is linked to SLC7A11, GPX4 and HMGB1 in the regulation of ferroptosis [11,29]. UVB exposure suppressed SLC7A11/GPX4 expression and increased HMGB1 levels, whereas ginsenoside treatment reversed these alterations and ameliorated ferroptosis-related indicators. These findings provide new insights into the mechanisms by which ginsenosides modulate photodamage and support their therapeutic potential in regulating keratinocyte function.

2. Materials and methods

2.1. Data sources

The single-cell transcriptomics dataset GSE173385, the DNA methylation profile GSE51954, and two mRNA gene expression profiles, GSE38308 and GSE41078, were obtained from the Gene Expression Omnibus database. The GSE173385 dataset contains skin samples from one normal mouse, one UV-irradiated mouse, and one mouse protected with Vitamin D. GSE51954 analyzed skin samples from 39 sun-protected controls and 39 sun-exposed subjects. GSE38308 included 21 pairs of sun-exposed and sun-protected skin samples from northern Chinese women, whereas GSE41078 consisted of 10 normal non-irradiated skin samples and 10 UVB-induced skin samples.

2.2. Data processing

The raw single-cell RNA count matrix were processed using the Seurat package to screen low-quality cells (300 < nCount _ RNA <6000, nFeature _ RNA <7000, min.cells >3, and percent.mt < 15), and double cells were removed using the DoubletFinder package (v2.0.3). The SCTransform function was used to normalize the gene expression counting matrix, perform principal component analysis and cluster, and the resolution was set to 0.6. The UMAP model was visualized by the RunUMAP function, and the differentially expressed genes were evaluated using the FindAllMarkers function. The standard was an average logarithmic scale fold change ≥0.25, p < 0.05. The methylation data were annotated using the load function of the ChAMP [30] package to remove probes for SNP regions and sex chromosomes. The criteria for identifying differentially methylated probes were |delta beta| > 0.2 and false discovery rate (FDR) < 0.05. Use GEOquery to download data from the GEO database and normalize inter-sample reading counts. The final gene expression value was the average of multiple probes, and the differentially expressed genes were compared using the limma package. The standard was log₂ (foldchange) > 1 or < −1 and the adjusted p value was <0.05.

2.3. Computational analysis of data

We first utilized the AUCell [31] package to compute the signature score of the ferroptosis gene set using default settings. For trajectory analysis, Monocle [32] and CytoTRACE [33] programs were employed to identify cell interconversion and evolutionary paths, with plot cell trajectory generating smooth expression curves and CytoTRACE complementing Monocle's unsupervised analysis with default settings. Finally, pseudotime-dependent genes underwent GO and KEGG pathway enrichment analysis using the clusterProfiler [34] package with default parameters. Gene Set Enrichment Analysis was performed to evaluate overall gene enrichment differences (adjusted p < 0.05 by Fisher's exact test), and Gene Set Variation Analysis (GSVA) was performed using the GSVA [35] package with standard settings. Additionally, we employed CellPhoneDB [30] to identify significant ligand-receptor pairs (p < 0.01) across cell types. NicheNet [36] was used to predict potential ligands driving keratinocyte phenotypes, ranking top ligands by Pearson correlation and ligand-target weights based on differentially expressed genes in keratinocytes (adjusted p < 0.05, log₂FC > 0.5).

2.4. Analysis of DNA methylation

The EpiDISH Bioconductor package was employed to estimate DNA methylation proportions for epithelial cells, fibroblasts, and immune cells using the reference dataset "centEpiFibIC.m". Additionally, relative proportions of endothelial cells, keratinocytes, myeloid cells, and HFCs were determined using the "CimpleG" function within the CimpleG [37] package (v0.0.5).

2.5. Characteristics of cell-type infiltration

We utilized the online tool CIBERSOFTx [38] to derive a reference signature matrix from our single-cell RNA-seq dataset, enabling the deconvolution of cell-type proportions from bulk datasets with default parameters for our defined 7 major cell types.

2.6. Cell culture and treatment

HaCaT cells were obtained from a skin biopsy healthy male donor (MCTT, Seoul, Korea) and grown in DMEM (1 % penicillin-streptomycin and 10 % heat-inactivated FBS) in 5 % CO₂ incubator at 37 °C. For experiments, cells were exposed to UVB at a dose of 125 mJ/cm² using a Bio-Link BLX-312 UV irradiation system (Vilber Lourmat GmbH, France) and then treated with ginsenoside C-K or C-Mc (ChemFaces, Wuhan, China) for 6 h. Control groups were subjected to the same protocol without UVB exposure.

2.7. Zebrafish embryotoxicity test

Zebrafish eggs were fertilized for 3 h, then placed in 6-well plates and incubated at 28 °C ± 0.5 °C. The eggs were randomly divided into 8 groups, including control group, C-K(1 μM,10 μM,20 μM)group and C-Mc(1 μM,10 μM,20 μM)group, with 10 fish eggs in each group. After drug treatment, embryos were observed at 0, 24, 48, and 72 h using a fluorescence stereomicroscope (Leica M205F) for abnormal developmental phenotypes or morphological defects. These defects were defined as fertilized egg coagulation, lack of somites, unclear tail bud/yolk sac separation, or lack of heartbeat.

2.8. Statistical analysis

Statistical analyses and graph generation were performed using R (v 4.2.1) and GraphPad Prism (v 8.0). All experiments were repeated three times, with data expressed as mean ± SD. One-way ANOVA was used for multiple group comparisons, followed by Tukey's test for pairwise comparisons. A p-value <0.05 was considered statistically significant.

3. Results

3.1. Dynamic changes of cell population during UVB-irradiated photoaging

We performed scRNA-seq on untreated, UV-irradiated, and vitamin D-treated cells, identifying cell types linked to photoaging. We clustered 20,728 single cells into eight groups, visualized by UMAP. Differentially expressed marker genes and unbiased cluster-specific signatures were employed to annotate cell types, including endothelial cells, epithelial cells, fibroblasts, hair follicle cells (HFCs), keratinocytes, melanocytes, myeloid cells, and T cells, using classic markers from previous publications and the CellMarker database (Fig. 1A–B). All cell types were present across samples, albeit at varying relative abundances. UV exposure significantly increased T cell infiltration and myeloid cell enrichment while reducing endothelial cell and fibroblast proportions, notably increasing epithelial cells and keratinocytes compared to untreated and vitamin D-treated samples.

Fig. 1.

The single-cell transcriptional profiles depict the dynamic changes in UVB-irradiated skin cells. (A) UMAP plot illustrating 8 identified cellular clusters. (B) Bubble diagram displaying primary markers across different cell types. (C) UMAP plot illustrating calculated scores for ferroptosis gene sets across cell clusters. (D) Line chart demonstrating tissue prevalence estimated for each cell type using Ro/e score. (E) Dot plot presenting differentially enriched pathways in the overall cell type.

To analyze ferroptosis across cell populations, we examined the average expression level of ferroptosis signatures (Fig. 1C). Keratinocytes and HFCs exhibited higher expression of ferroptosis signatures, suggesting a potential involvement of keratinocytes in the ferroptosis process. Endothelial cells and fibroblasts were predominantly found in the UV-irradiated group, while epithelial cells and HFCs were more abundant in the C5 and VD groups (Fig. 1D). Melanocytes were excluded from analysis due to their small numbers. To explore functional variations in the regulatory network of UV-induced cell populations, we utilized hallmark gene sets to examine pathway differences among endothelial cells, epithelial cells, fibroblasts, HFCs, keratinocytes, myeloid cells, and T cells in normal and UV-irradiated tissues. Interestingly, myeloid cells exhibited enrichment in metabolic and signaling pathway activities (Fig. 1E). Moreover, UV-irradiated keratinocytes and fibroblasts demonstrated higher upregulation of ferroptosis compared to cells from normal skin (Fig. 1E), with keratinocytes exhibiting the highest ferroptosis scores, consistent with their role as primary epidermal targets of UVB-induced damage [5].

3.2. Epigenetic alterations in keratinocytes during photoaging

DNA methylation profiling provides robust insights into epigenomic changes in photoaging. The results of DNA methylation data showed that there was no significant difference in keratinocyte proportions between the normal and photoaging groups, but the photoaging group, consisting of 39 samples, exhibited a higher keratinocyte abundance, which was consistent with scRNA-seq results (p = 0.68, Fig. 2A). Following differential analysis, we identified 52 hypermethylated differentially methylated positions (DMPs) were identified, with 25 located in the intergenic region, and the remaining 27 DMPs corresponding to 27 genes. Additionally, 93 hypomethylated DMPs were identified, with 63 located in the intergenic region, and the remaining 30 DMPs corresponding to 30 genes. The majority of hypermethylated and hypomethylated DMPs are located within shore regions, comprising 44.2 % and 45.2 % of the total, respectively (Fig. 2B). There was overall hypermethylation in the photoaging group, but not in the normal group. These DMPs correspond to hypermethylated genes, indicating that DNA methylation may play a key role in the pathogenesis of photoaging (Fig. 2C). Comparing differentially methylated genes (DMGs) from DNA methylation profiling with differentially expressed genes (DEGs) from keratinocytes in the scRNA-seq dataset revealed 5 overlapping genes (Fig. 2D). GSEA analysis for deeper insight showed genes with high methylation in photoaging samples were enriched in tissue morphogenesis and growth factor binding pathways (Fig. 2E). These samples also exhibited significant enrichment in skin development pathways, consistent with the notion that lifelong exposure to sunlight plays a crucial role in the development of skin aging characteristics [39].

Fig. 2.

DNA methylation profiles reveal differentiation underlying methylation transformations in photoaging. (A) Boxplot illustrating the estimation of cell proportions in the normal and UVB-irradiated groups using a deconvolution algorithm. (B) Bar plot displaying the distribution of DMPs based on their distance from CGI, with CGI referring to CpG islands, the shore representing regions up to 2 kb from the CpG island, the shelf representing regions 2–4 kb from the CpG island, and opensea areas encompassing the rest of the genome. (C) Heatmap of differentially methylated genes (DMGs). (D) Venn diagram illustrating the overlap between DMGs and differentially expressed genes in keratinocytes. (E) GSEA displaying the gene ontology enrichment pathways of DNA methylation profiles in photoaging.

3.3. Heterogeneity of keratinocyte subsets

To understand keratinocyte transcriptome heterogeneity, we subclustered 3694 cells into three UMAP-defined groups (Fig. 3A). Kera-C1-Krts keratinocytes predominantly expressed several keratin family markers, including Krt1, Krt10, Krt16, and Krt25. Kera-C2-Klks subtypes exhibited enrichment of kallikrein family genes (Klk11, Klk5, Klk8, Klk7), which regulate extracellular matrix organization and are involved in early tumor progression [40]. Kera-C3-Lces subtype expressed high levels of terminal keratinocyte-associated markers (Lce1a, Lce1b, Lce3, Lce6) and played an important role in the antibacterial activity of the skin [41]. Furthermore, Kera-C1-Krts keratinocytes exhibited a higher ferroptosis score than Kera-C2-Klks and Kera-C3-Lces subtypes, suggesting that Kera-C1-Krts keratinocytes have a more ferroptosis-like phenotype (Fig. 3B). Kera-C3-Lces exhibited a higher inflammatory score compared to both subtypes, suggesting the involvement of Kera-C3-Lces in inflammatory response-related pathways [41]. Gene ontology analysis further corroborated our findings (Fig. 3C).

Fig. 3.

Lineage and characteristics of keratinocyte subclusters are depicted. (A) UMAP plot displaying 3694 keratinocytes. (B) Boxplot illustrating ferroptosis phenotype scores (left) and inflammation scores (right) across three keratinocyte subpopulations. (C) Gene ontology analysis of Kera-C1-Krts, Kera-C2-KlKs and Kera-C3-Lces cell subsets. (D) Development trajectory of keratinocyte subclusters is shown. Boxplot at the top illustrates stemness across three keratinocyte subclusters, calculated using the CytoTRACE algorithm. (E) Heatmap illustrates differentially expressed genes in keratinocytes based on the pseudotime trajectory. Color key ranges from blue to red, indicating relative expression levels from low to high. (F) Gene Set Enrichment Analysis (GSEA) reveals KEGG enrichment pathways in both Cell Feature 1 and Cell Feature 2.

The cell lineage trajectory of keratinocytes revealed that the Kera-C2-Klks cluster was at the origin of the pseudotime trajectory, subsequently bifurcating into both Kera-C1-Krts (as cell feature 1) and Kera-C3-Lces subgroups (as cell feature 2) (Fig. 3D). We utilized the BEAM function to illustrate branch-dependent genes' temporal expression patterns for keratinocytes using a heatmap (Fig. 3E). KEGG enrichment analysis revealed that cell feature 2 was primarily enriched in inflammatory responses to wounding and antibiotics, along with enhanced negative regulation of cell development and regulation of fatty acid metabolic processes, consistent with our earlier findings regarding the Kera-C3-Lces cluster (Fig. 3F). Cell feature 1 exhibited significant enrichment in skin development, keratinocyte differentiation, response to oxidative stress, and wound healing. We observed enrichment of fatty acid metabolism in both cell feature 1 and cell feature 2, suggesting that regulation of fatty acid metabolism might mediate the phenotypic and functional shifts during keratinocyte photoaging. These findings suggest a potential direction for the differentiation of keratinocytes during the development of photoaging.

3.4. Key regulatory signals in photoaging

Receptor-ligand-mediated intercellular interactions play a critical role in both physiological and pathological processes. We utilized CellPhoneDB for cell communication analysis to explore the intercellular regulatory networks among various cell types identified in our study. Complex communication patterns were observed between keratinocytes and other cell types (Fig. 4A). Notably, communication between fibroblasts, myeloid cells, and keratinocytes was particularly prominent, as quantified by over 100 key ligand-receptor interactions through communication-related heatmaps. Additionally, endothelial cells, epithelial cells, and T cells also established similar communication links with keratinocytes (Fig. 4B). The identification of these cell types further underscores the broad spectrum of molecular interactions involved (Fig. 4C).

Fig. 4.

Cell-Cell Communication of Major Cell Types, with a Focus on Keratinocytes. (A) Heatmap illustrating significant ligand-receptor interactions between various cell groups. (B) Detailed Overview of Ligand-Receptor Interactions for Each Major Cell Type. (C) Heatmap displaying correlation scores for the regulation of genes (columns) by potential ligands (rows). (D) Heatmap depicting selected ligand-receptor pairs between keratinocytes and other cell types. (E) Boxplot depicting Gas6 expression levels in the control (n = 12) and UVB-irradiated (n = 12) groups. ∗p < 0.05. (F) Scatter plot illustrating the Pearson correlation between keratinocyte and myeloid cell compositions in a bulk dataset (n = 20).

Subsequently, we employed NicheNet analysis to investigate the external regulatory factors influencing keratinocytes. Our results indicated that key ligands expressed by keratinocytes were significantly associated with marker genes of other cell types, with Gas6 identified as a potential ligand driving keratinocyte communication (Fig. 4C). According to CellPhoneDB analysis, myeloid cells established direct communication with keratinocytes via highly enriched Gas6-Axl and Gas6-Tyro3 ligand-receptor pairs (Fig. 4D). Notably, Gas6 was significantly upregulated in UVB-irradiated samples (p < 0.05), suggesting its potential key role in the intercellular signaling remodeling induced by photoaging (Fig. 4E). Keratinocytes exhibited higher expression levels of ferroptosis-related genes, including SLC7A11, GPX4 and HMGB1. Gas6-Axl has been reported to regulate Nrf2, thereby influencing antioxidant stress and providing indirect evidence for the role of Gas6 in modulating the expression of SLC7A11 and GPX4 [29]. In the UVB- irradiated cell model, HMGB1 is commonly associated with ferroptosis induction [11], whereas Gas6 can inhibit HMGB1-mediated inflammation. These two factors may interact in the context of photoaging. Additionally, the ratio of keratinocytes to bone marrow cells varied across different samples, although the distribution of these cell populations did not show significant differences relative to the distribution of photoaging phenotypes (Fig. 4F).

3.5. Molecular docking of ginsenosides and in vivo toxicity assessment in zebrafish embryos

To identify candidate ginsenosides with potent ferroptosis inhibitory potential, we conducted molecular docking of representative ginsenosides (Rk1, Rb1, Rg1, Rg2, Rg5, C-K, and C-Mc) with ferroptosis-related targets, including SLC7A11, GPX4, and HMGB1. Ginsenosides C-K and C-Mc exhibited the lowest binding energies to all targets, indicating the highest binding affinities (Table S2, Fig. 5A). The toxicity of C-K and C-Mc was further evaluated in zebrafish embryos exposed to 1, 10, and 20 μM during early development. Zebrafish embryos were selected as a model for in vivo toxicity assessment due to their rapid development and high genetic homology to humans [42,43]. The survival rate and morphological characteristics of embryos were observed at 0 h, 24 h, 48 h and 72 h after fertilization (Fig. 5B). In all experimental groups, normal gastrula formation was observed at 24 h, normal somatic cell development was observed at 48 h, and complete body axis and normal morphological characteristics were observed at 72 h. The survival rate of all embryos was maintained at 100 %, and no developmental defects were found. These results indicate that ginsenosides C-K and C-Mc do not exhibit embryotoxic effects.

Fig. 5.

Molecular Docking and Zebrafish embryo toxicity test. (A) Molecular docking analysis demonstrated strong binding affinities of ginsenosides C-K and C-Mc to SLC7A11, GPX4, and HMGB1. (B) The effects of ginsenosides C-K and C-Mc (1, 10, and 20 μM) on zebrafish embryos at 0 h, 24 h, 48 h, and 72 h post-fertilization.

3.6. Ginsenosides inhibits UVB-irradiated ferroptosis via the SLC7A11/GPX4 signaling axis

To elucidate the specific mechanism by which ginsenosides C-K and C-Mc regulate cellular ferroptosis, we used UVB-induced HaCaT cells to induce ferroptosis. The results showed that UVB irradiation significantly suppressed the mRNA expression of SLC7A11 and GPX4, while C-K and C-Mc treatments at 10 μM and 20 μM significantly restored their transcriptional levels (Fig. 6A–B). UVB irradiation also resulted in a significant increase in HMGB1 mRNA levels, while C-K and C-Mc treatments effectively reduced the transcriptional levels of HMGB1 (Fig. 6C), suggesting that they could affect oxidative damage in cells by regulating the transcriptional activity of key antioxidant genes. Further Western blot analysis results validated these transcriptional changes, showing that UVB-induced SLC7A11 and GPX4 protein expression decreased, while HMGB1 protein expression was significantly upregulated. Interventions of C-K and C-Mc could significantly restore protein levels of SLC7A11 and GPX4, inhibiting abnormal elevation of HMGB1 (Fig. 6D–E). In addition, UVB irradiation significantly increased ROS levels and reduced viability in HaCaT cells. Treatment with ginsenosides C-K and C-Mc markedly attenuated ROS accumulation and rescued cell viability following UVB exposure (Fig. 6F–I). Mechanistically, ginsenosides C-K and C-Mc attenuate ferroptosis in HaCaT cells by activating the SLC7A11/GPX4 axis and suppressing HMGB1 expression (Fig. 6J).

Fig. 6.

Ginsenoside C-K and C-Mc increased in SLC7A11 and GPX4 and inhibited HMGB1 expression in UVB-irradiated HaCaT cells. HaCaT cells were exposed to UVB (125 mJ/cm²) and treated with ginsenoside C-K or C-Mc (10 or 20 μM) for 6 h. (A–C) mRNA levels of SLC7A11, GPX4, and HMGB1 were quantified and normalized to GAPDH. (D–E) The protein levels of SLC7A11, GPX4, and HMGB1 were assessed by Western blot. (F–G) Intracellular ROS was measured by flow cytometry. (H–I) Cell viability was evaluated following UVB and ginsenoside treatment. (J) Ginsenosides C-K and C-Mc inhibit UVB-induced ferroptosis in HaCaT cells by modulating the SLC7A11/GPX4 axis and HMGB1. Data are presented as mean ± SD. # and ∗ indicate significant differences versus control groups as defined: #p < 0.05, ##p < 0.01, and ###p < 0.001 vs. nonirradiated control. ∗ p < 0.05, ∗∗ p < 0.01, and ∗∗∗ p < 0.001 vs. UVB-irradiated control.

4. Discussion

Keratinocytes exhibited a robust and dynamic response during UVB-irradiated photoaging. Our findings revealed that the abnormal expression of SLC7A11, GPX4 and HMGB1 in keratinocytes following UVB irradiation is highly correlated with ferroptosis. Ginsenoside Rb1 enhances the antioxidant capacity of keratinocytes by activating the Nrf2 pathway, thereby inhibiting UVB-induced ROS accumulation [44]. Ginsenoside Rg3 mitigates the release of pro-inflammatory factors such as IL-6 and TNF-α by suppressing the NF-κB signaling pathway, thus alleviating UVB-irradiated inflammatory responses [45]. In terms of epigenetic regulation, ginsenoside Compound K has been shown to reverse UV- irradiated DNA methylation abnormalities and restore the expression of antioxidant genes like SOD2 and CAT [46]. This study further elucidates the multiple mechanisms of ginsenosides in preventing and treating photoaging. Specifically, ginsenosides C-K and C-Mc inhibit UVB-irradiated ferroptosis by modulating the SLC7A11/GPX4 signaling axis and suppressing HMGB1 activity.

Ultraviolet radiation can accelerate photoaging by inducing ferroptosis in keratinocytes through inhibition of the SLC7A11/GPX4 pathway. Consistent with our findings, UVB exposure significantly downregulates SLC7A11 and GPX4 expression [[47], [48], [49]]. SLC7A11 is essential for maintaining intracellular glutathione homeostasis and supporting GPX4 function [50]. Additionally, GPX4 prevents ferroptosis-related membrane damage by catalyzing the reduction of lipid peroxides, and its loss of function leads to lipid peroxide accumulation and triggers cell death [51]. Our study further revealed that UVB irradiation induces abnormally high expression of HMGB1 in keratinocytes, which is highly correlated with ferroptosis. Ferroptosis can trigger HMGB1 release and acetylation [44,45], consistent with previous findings of elevated HMGB1 levels in UVB-exposed keratinocytes [46]. HMGB1 activation can induce NLRP3 inflammasome formation, leading to oligomerization of NLRP3, caspase-1 activation, and subsequent cleavage and secretion of pro-inflammatory cytokines IL-1β and IL-18 [[52], [53], [54], [55]]. The inflammatory signals mediated by HMGB1 may amplify skin damage effects [11]. Our research indicates that ginsenosides exhibit significant antagonistic effects against ferroptosis in UVB-irradiated HaCaT cells, emphasizing the importance of the SLC7A11/GPX4 signaling pathway and HMGB1 in reducing ferroptosis in photoaged keratinocytes.

The Gas6-Axl/Tyro3 signaling axis exerts key regulatory functions in the photoaging microenvironment. The interaction network established between myeloid cells and keratinocytes via the Gas6 ligand-receptor pair may indirectly inhibit ferroptosis by activating the Nrf2-SLC7A11/GPX4 pathway, while also antagonizing HMGB1-mediated inflammatory responses. This finding provides a theoretical basis for targeting Gas6 signaling to reshape the skin microenvironment. Notably, ginsenosides C-K and C-Mc exhibit significant ferroptosis inhibitory effects in UVB- irradiated HaCaT cells, with their mechanism closely related to restoring SLC7A11/GPX4 expression and inhibiting abnormal HMGB1 activation. This effect is associated with the activation of the Nrf2-SLC7A11/GPX4 axis, wherein Nrf2, as a central transcription factor for antioxidant defense, may be indirectly upregulated via the Gas6-Axl signaling pathway [24]. In addition, UVB irradiation increased ROS levels and decreased HaCaT cell viability, while ginsenosides C-K and C-Mc reduced ROS accumulation and improved cell survival. Single-cell RNA sequencing data support the hypothesis that ultraviolet radiation exposure disrupts skin homeostasis through epigenetic interference [56]. Transcriptional heterogeneity analysis of keratinocyte subpopulations offers new insights into photoaging research. The high ferroptosis score and enrichment of oxidative stress pathways in the Kera-C1-Krts subpopulation suggest its sensitivity to UV damage, while the inflammatory phenotype of the Kera-C3-Lces subpopulation further validates the dual role of keratinocytes in regulating photoaging-related inflammation [57]. The upregulation of Gas6 ligands in keratinocytes, particularly in the Kera-C1-Krts subpopulation after UVB irradiation, suggests that it may regulate Nrf2 activity through paracrine action, thereby synergizing with ginsenosides to exert protective effects. DNA methylation analysis reveals epigenetic reprogramming characteristics of keratinocytes during photoaging. Although overall cell proportions do not change significantly, the enrichment of highly DMPs and their overlap with DEGs in photoaged samples suggest that epigenetic regulation may exacerbate ferroptosis sensitivity by silencing antioxidant-related genes such as SLC7A11. Ginsenosides may reverse this trend through epigenetic reprogramming, providing new evidence for multi-target intervention. The unique advantages of ginsenosides lie in their natural sources and safety. These compounds do not show toxicity in zebrafish embryos, supporting their clinical transformation potential. UVB exposure significantly reshapes the dynamic balance of skin cell populations, characterized by increased T cell infiltration, enrichment of myeloid cells, and an elevated proportion of keratinocytes. This suggests that the synergistic effects of the inflammatory microenvironment and oxidative stress may be central drivers of the photoaging pathological process. Notably, the high expression of ferroptosis-related genes in keratinocytes underscores the critical role of ferroptosis in UV-induced skin damage, consistent with the previously proposed oxidative stress-inflammation-cell death cascade hypothesis [11]. Ginsenosides may synergistically inhibit the inflammatory microenvironment and promote skin barrier repair by modulating Gas6-Axl/Tyro3-mediated interactions between keratinocytes and myeloid cells.

Our study has certain limitations. First, although our findings indicate a potential association of ferroptosis markers (SLC7A11, GPX4, and HMGB1) with keratinocyte ferroptosis, whether these markers directly participate in the mechanisms underlying skin photoaging requires further validation through gene knockdown or overexpression experiments. Second, the heterogeneous regulation of fatty acid metabolism in keratinocyte subpopulations suggests that metabolic reprogramming may play a critical role in photoaging, but the specific changes of metabolites and functional effects remain to be further studied. Despite these limitations, our study is the first to combine epigenetic regulation, single-cell resolution cellular heterogeneity, and natural compound intervention, providing novel insights into the mechanisms of photoaging and the development of therapeutic strategies.

5. Conclusion

This study elucidated the multidimensional mechanism of UVB driving photoaging through the ferroptosis-epigenetic-cellular communication network. The high expression of ferroptosis regulatory genes in keratinocytes is closely related to photoaging. We have confirmed that ginsenosides C-K and C-Mc target the upregulation of SLC7A11/GPX4 expression and inhibit abnormal activation of HMGB1, thereby suppressing photoaging. Therefore, our research provides a novel molecular mechanism model for UV-irradiated skin aging and lays the foundation for evaluating ginsenosides as potential therapeutic agents for photoaging.

Declaration of competing interest

The authors declare no competing financial interests or personal relationships.

Appendix A. Supplementary data

The following are the Supplementary data related to this article.

Data availability

Data will be made available on request.