Resveratrol alleviates intervertebral disc degeneration by regulating ferroptosis of nucleus pulposus cells

Abstract

Intervertebral disc degeneration (IVDD) is a major contributor to chronic low back pain, yet its molecular mechanisms remain incompletely understood. Emerging evidence suggests that ferroptosis, an iron-dependent form of regulated cell death, may play a key role in the degeneration of nucleus pulposus cells (NPCs). However, its involvement in IVDD has not been systematically explored. Resveratrol (Res), a natural polyphenol with antioxidant and cytoprotective effects, shows therapeutic potential, but its regulatory role in ferroptosis during IVDD is unclear. In this study, bulk transcriptome analysis identified 22 ferroptosis-related differentially expressed genes, among which five hub genes (GPX4, FTH1, HMOX1, NCOA4, and TFRC) were strongly linked to ferroptosis and oxidative stress pathways. Single-cell transcriptome data further revealed dynamic expression of these genes in NP cells, showing decreased GPX4 and FTH1 and increased HMOX1, NCOA4, and TFRC along the differentiation trajectory. High-dimensional weighted gene co-expression network analysis (hdWGCNA) confirmed significant associations between ferroptosis-related gene modules and IVDD pathology. Validation using human clinical samples, NP cell models, and animal experiments demonstrated that Res alleviates IVDD by modulating hub gene expression and suppressing NP cell ferroptosis. Overall, this study highlights ferroptosis as a key mechanism driving IVDD and identifies resveratrol as a promising therapeutic candidate targeting NP cell ferroptosis.

Introduction

Intervertebral disc degeneration (IVDD) is one of the most common chronic degenerative diseases and a leading cause of chronic low back pain, spinal dysfunction, and decreased quality of life¹. With the aging population, the incidence of IVDD has been steadily increasing². Data show that from 1990 to 2015, the global incidence of IVDD rose by 54%³. Over 600 million people worldwide are affected by this condition, and the number continues to grow⁴, significantly impairing the quality of life of patients while placing a heavy burden on society and the economy⁵. The pathogenesis of IVDD is multifactorial, involving aging, genetic predisposition, smoking, obesity, mechanical stress, and other factors^6–9. Extensive research has identified multiple molecular mechanisms contributing to IVDD, including inflammation, oxidative stress, Cellular Senescence, extracellular matrix degradation, and mitochondrial dysfunction^8,10–12. However, the complex regulatory interactions among these pathways and their precise contributions to nucleus pulposus cell (NPC) degeneration remain incompletely understood, highlighting the need for further mechanistic studies.

Current treatments for IVDD mainly focus on symptom relief, including pain management and surgical intervention. However, these approaches are often ineffective in delaying or reversing the pathological progression of the intervertebral disc¹³. Consequently, there is an urgent need to develop etiological treatment strategies. In recent years, research has highlighted that programmed cell death (PCD) of NPCs is a key pathological feature in the development of IVDD. Various forms of PCD, including apoptosis, autophagy, pyroptosis, and necroptosis, have been identified^14–16. Targeting PCD has emerged as a potential therapeutic approach for alleviating IVDD. For instance, studies by Chen et al. showed that silencing Grem1 could inhibit apoptosis of NPCs, significantly delaying the progression of IVDD in rats¹⁷. Zhang et al. found that glutamine treatment could alleviate IVDD by enhancing autophagy in NPCs¹⁸. Based on these findings, interventions targeting PCD in NPCs are considered a promising new direction for treating IVDD.

Ferroptosis, a recently discovered form of cell death distinct from other PCDs, is characterized by iron accumulation, lipid peroxidation, and plasma membrane damage^19,20. It is driven by iron-dependent lipid peroxidation, accompanied by mitochondrial phenotypic changes and dysfunction, ultimately leading to cell death^21,22. Ferroptosis has been implicated in the pathogenesis of various diseases. For example, Cui et al. demonstrated that ferroptosis is the main cause of hepatocyte death in alcoholic and non-alcoholic fatty liver disease²³, and Liang et al. identified ferroptosis as a key mechanism in epilepsy²⁴. Similarly, ferroptosis is involved in the pathological process of IVDD. Glutathione peroxidase 4 (GPX4) is a crucial regulator of ferroptosis, inhibiting lipid peroxidation by reducing lipid hydroperoxides to their hydroxyl derivatives, thus preventing ferroptosis^25,26. Yang et al. found that the protein expression of GPX4 was significantly decreased in human nucleus pulposus (NP) tissue from IVDD patients²⁷. Additionally, Zhu et al. showed that overexpression of the deubiquitinase USP11 could significantly alleviate IVDD by upregulating Sirt3, inhibiting oxidative stress-induced ferroptosis, while deletion of Sirt3 exacerbated ferroptosis in NPCs and accelerated IVDD progression²⁸. These studies underscore the critical role of ferroptosis in the onset and progression of IVDD, suggesting that targeting ferroptosis could offer an effective strategy for treating IVDD.

Resveratrol(Res) is a natural polyphenolic compound originally isolated from the roots of Polygonum cuspidatum and is widely found in grapes, peanuts, soybeans, and berries²⁹. In recent years, Res has gained significant attention for its antioxidant, anti-aging, anti-inflammatory, anticancer, and metabolic regulatory properties^30–32. Notably, recent studies have highlighted Res’s potential in regulating ferroptosis. Wang et al. found that Res could activate the SIRT3/FoxO3a pathway, upregulating SOD2 and catalase expression, reducing ROS and lipid peroxidation (LPO) levels, and compensating for the GSH/GPX4 pathway to inhibit ferroptosis and alleviate intestinal ischemia-reperfusion injury³³. Zhang et al. reported that Res could protect MIN6 cells from acrylaldehyde-induced ferroptosis by modulating the PERK pathway related to endoplasmic reticulum stress³⁴. These findings confirm the significant role of Res in regulating ferroptosis. However, whether Res can alleviate IVDD by modulating ferroptosis remains unclear. Therefore, exploring the molecular mechanisms by which Res regulates ferroptosis in IVDD is crucial for the development of targeted therapeutic agents.

In this study, we integrated bulk and single-cell transcriptomic analyses to comprehensively elucidate the role of ferroptosis in intervertebral disc degeneration (IVDD) and to uncover the potential therapeutic mechanism of resveratrol (Res). Bulk RNA-seq data were analyzed to identify ferroptosis-related differentially expressed genes (DEGs) and to determine hub genes through protein–protein interaction (PPI) network construction and enrichment analyses. Single-cell RNA-seq data were further used to explore the cell-type-specific expression patterns, dynamic transcriptional changes, and pseudotime trajectories of these hub genes in NPCs. In addition, high-dimensional weighted gene co-expression network analysis (hdWGCNA) was performed to identify gene modules most strongly correlated with IVDD pathology and ferroptosis activity. The potential binding interactions between Res and hub proteins were predicted using molecular docking, and the regulatory effects of Res on ferroptosis-related genes were validated in clinical NP samples, in vitro NPC models, and in vivo rat models of IVDD. This integrative strategy provides a systematic understanding of ferroptosis in IVDD and supports Res as a promising therapeutic agent for preventing disc degeneration (Fig. 1).

Fig. 1. Schematic overview of the study design.

This study integrates bulk and single-cell transcriptome analyses with experimental validation to investigate how Res alleviates IVDD via ferroptosis regulation. Hub ferroptosis-related hub genes were identified, their dynamic expression in NPCs was characterized, and experimental validation confirmed that Res modulates these genes to inhibit NPC ferroptosis, thereby mitigating IVDD. The figure summarizes the study workflow and main findings.

Results

Differential gene expression analysis of bulk transcriptome

The GSE167199 dataset was downloaded from the GEO database and analyzed (Supplementary Table 1), which includes 3 normal samples and 3 IVDD samples. The normal NP tissues were derived from patients with spinal cord injury, while the IVDD NP tissues were obtained from IVDD patients. Differential expression analysis was performed using the limma package, revealing a total of 2542 differentially expressed genes, including 1149 upregulated and 1393 downregulated genes (Fig. 2A, Supplementary Table 2). A heatmap was generated to visualize the top 20 differentially expressed genes from both the upregulated and downregulated groups (Fig. 2B).

Fig. 2. Determinating of hub genes.

A Volcano plot of differentially DEGs; B Heatmap of the top 20 up-regulated and down-regulated DEGs; C Venn diagram showing the intersection of DEGs, FRGs and relative targets of Res; D PPI network constructed using the STRING database of the 22 FRDEGs; E Venn diagram showing the intersection of FRDEGs and FPGs; F PPI network constructed using Cytoscape.

Hub genes

FRGs were selected from the FerrDb database, resulting in 483 genes. To explore the potential targets of Res, drug target information was collected from several databases, including TCSMP (148 targets), SwissTargetPrediction (100 targets), Drugbank (36 targets), TargetNet (606 targets), and relevant literature (17 targets). After deduplication, 637 Res-related drug targets were identified (Supplementary Table 3). Subsequently, the intersection of differentially expressed genes, FRGs, and Res-related targets was analyzed, revealing 22 FRDEGs (Supplementary Table 4, Fig. 2C). To further explore the interactions among these genes, a PPI network was constructed using the STRING database (Fig. 2D). Additionally, the intersection of the 22 ferroptosis-related differential genes(FRDEGs) with genes in the ferroptosis pathway identified 5 hub genes: GPX4, FTH1, HMOX1, NCOA4, and TFRC (Fig. 2E). Figure 2F shows the interaction network of these 5 hub genes with other FRDEGs, while Fig. 3A displays their positions within the ferroptosis pathway. These findings suggest that these 5 hub genes may play a central role in Res’s regulation of the ferroptosis pathway, providing important targets and theoretical foundations for further research.

Fig. 3. Enrichment analysis.

A Ferroptosis signaling pathway obtained from the KEGG PATHWAY Database, with the locations of the five hub genes highlighted in red. Note: Some gene names in the KEGG pathway are shown as their functional equivalents or synonyms—FTH1 is represented as Ferritin (including FTH1 and FTL subunits), HMOX1 as HO-1, and TFRC as TFR1; B GO analysis results, bar chart of the top 5 results of BP, CC, and MF; C Bar chart of the first 20 results of KEGG analysis; D Single gene GSVA analysis results of 5 hub genes.

Enrichment analysis of FRDEGs

The 22 FRDEGs were subjected to GO and KEGG enrichment analysis. The GO enrichment analysis revealed 97 BP, 34 CC, and 23 MF (Supplementary Table 5). The top 5 BP terms were: response to hypoxia, response to decreased oxygen levels, response to oxygen levels, cellular response to hypoxia, and cellular response to decreased oxygen levels, suggesting that hypoxic environments and oxidative stress may be involved in the process of ferroptosis. The top 5 CC terms were: autolysosome, secondary lysosome, tertiary granule lumen, basal part of cell, and secretory granule lumen, indicating that ferroptosis may be regulated through autophagy and lysosomal functions. The top 5 MF terms included RNA polymerase II-specific DNA-binding transcription factor binding, DNA-binding transcription factor binding, transition metal ion transmembrane transporter activity, transcription coregulator binding, and NAD + ADP-ribosyltransferase activity, suggesting that these genes might influence iron metabolism and cell fate via transcriptional regulation and metal ion transport mechanisms (Fig. 3B).

KEGG analysis identified 134 enriched pathways (Supplementary Table 6). Based on the combined P-values and count numbers, the 22 intersecting genes were primarily involved in , HIF−1 signaling pathway, p53 signaling pathway, necroptosis, and Lipid and atherosclerosis (Fig. 3C). These results indicate that ferroptosis is closely associated with cell death, immune responses, and disturbances in iron homeostasis, particularly in pathological states of iron metabolism imbalance, which could exacerbate cellular damage. These enrichment findings provide significant clues for understanding the molecular mechanisms of ferroptosis and its role in related diseases, offering potential targets and intervention strategies for future research.

Single-gene GSEA analysis of hub genes

To explore the potential functional roles of the 5 hub genes in IVDD, single-gene Gene Set Enrichment Analysis (GSEA) was performed for each gene. The results revealed that all 5 hub genes were enriched in the ferroptosis pathway (Fig. 3D). These findings further suggest that these 5 hub genes may contribute to the promotion of IVDD through the regulation of ferroptosis.

Quality control results of single-cell transcriptome

The single-cell RNA sequencing dataset GSE244889 related to IVDD was downloaded from the GEO database, including 4 normal samples and 3 IVDD samples. The normal samples were derived from trauma patients with Pfirrmann grades I-II, while the IVDD samples came from patients with degenerative spine diseases with Pfirrmann grades III-IV. The dataset included a total of 54,373 cells. To ensure the accuracy of subsequent analyses, several quality control steps were implemented. First, low-quality cells with fewer than 300 expressed genes and cells with total counts exceeding 100,000 were excluded. Additionally, cells with mitochondrial gene proportions greater than 10% and red blood cell gene proportions higher than 3% were removed. After these initial quality control steps, 51,131 high-quality cells were retained. Further doublet removal was performed using the DoubletFinder package, leaving 47,959 cells for downstream analysis.

Cell annotation

Dimensionality reduction and clustering analysis identified 13 distinct cell populations (Supplementary Fig. 1). Using marker genes for each population (Supplementary Fig. 2, Supplementary Table 7), we successfully annotated the following 8 cell types: NPC, Neutrophil, T cell, Endothelial cell, B cell, Macrophage, Pro-Myelocyte, and Smooth muscle cell (Fig. 4A). Figure 4B displays the marker genes for each cell type, while Fig. 4C shows the spatial distribution of key marker genes on the UMAP plot.

Fig. 4. Analysis of single-cell transcriptome atlas.

A UMAP plot of cell annotation; B Bubble diagram of marker genes for various cells; C Spatial location of the hub genes in single-cell atlas; D UMAP images of some key marker genes; E Bubble diagram of 5 hub genes for various cells.

Hub gene localization and differential gene expression analysis in single-cell transcriptome atlas

To investigate the distribution of hub genes across different cell populations, we used the FeaturePlot function to visualize the expression of each hub gene in the various cell types. As shown in Fig. 4D, E, the 5 hub genes show relatively higher expression in NPCs, implying that these genes may participate in the pathological process of IVDD through NPCs.

To further analyze gene expression differences between cell populations, we used the FindAllMarkers function in the Seurat package to identify differentially expressed genes across all cell types, calculating their log2 fold changes (logFC) and statistical significance (Supplementary Table 8, Supplementary Fig. 3). We then employed FindMarkers to compare gene expression differences between control and IVDD groups (Supplementary Table 9, Supplementary Fig. 4). Finally, after performing differential analysis for all cell populations, we focused on NPCs and used the FindMarkers function to identify differentially expressed genes between control and IVDD groups (Supplementary Table 10). Figure 5A illustrates the DEGs in NPCs between the control and IVDD groups.

Fig. 5. Pathway enrichment analysis and pathway scoring.

A Volcano plot of differentially expressed genes between the control group and IVDD group in NPCs; B Bubble chart of the top 10 results of KEGG analysis of differentially expressed genes in NPCs; C Bar graph of GSVA scores of each pathway in the control group and IVDD group in NPCs; D Expression of ferroptosis signaling pathway in the control group and IVDD group.

Enrichment analysis and pathway scoring in single-cell transcriptome atlas

GO enrichment analysis revealed that the differentially expressed genes were primarily enriched in BP such as extracellular exosome, extracellular vesicle, extracellular organelle, cytosol, and vesicle (Supplementary Table 11, Supplementary Fig. 5). In CC, these genes were mainly associated with translational initiation, RNA binding, mRNA metabolic processes, macromolecule catabolic processes, and cellular protein metabolic processes. MF enrichment indicated significant association with cellular macromolecule catabolic processes, cell adhesion molecule binding, cadherin binding, structural molecule activity, and structural constituent of ribosome.

KEGG pathway analysis showed that these genes were significantly enriched in the TNF signaling pathway, proteasome, and ferroptosis (Supplementary Table 12, Fig. 5B). Gene Set Variation Analysis (GSVA) further revealed differences in pathway activity between the IVDD and control groups in NPCs (Fig. 5C). Specifically, ferroptosis-related pathway activity was significantly elevated in IVDD NPCs (Fig. 5D), suggesting a potential role of ferroptosis in the pathological process of IVDD.

Pseudotime analysis

We performed pseudotime trajectory analysis on NPC differentiation using the Monocle2 package. The analysis revealed that NPC differentiation in IVDD progressed along a pseudotime axis, divided into 7 distinct states (Fig. 6A). Further analysis showed that the expression changes of 22 FRDEGs followed a significant temporal pattern along the differentiation trajectory (Fig. 6B). This suggests that these genes may play crucial roles in ferroptosis during IVDD development. Notably, two of the hub genes, GPX4 and FTH1, which are associated with antioxidant activity and iron homeostasis, exhibited progressively decreased expression during differentiation, indicating their potential role in regulating NPC ferroptosis and participating in early differentiation or homeostasis maintenance. Conversely, HMOX1, NCOA4, and TFRC showed increasing expression during differentiation, suggesting that these genes may play key roles in iron metabolism and cell death, particularly in NPC ferroptosis and later stages of differentiation (Fig. 6C). These results further highlight the involvement of these 5 hub genes in the regulation of ferroptosis in NPCs, thereby influencing IVDD.

Fig. 6. Pseudotime analysis.

A Pseudotime analysis of NPC states and differentiation over time; B Heat map of the expression of 22 FRDEGs over time; C Differentiation of expression of 5 hub genes over time.

To further evaluate the robustness of the pseudotime patterns across subjects, we additionally performed pseudotime trajectory analyses separately for each individual sample, including 4 control samples and 3 IVDD samples. The results demonstrated that the expression trends of the hub genes along the differentiation trajectory in IVDD samples were generally consistent with those observed in the integrated analysis (Supplementary Fig. 7A–D). In contrast, the control samples exhibited relatively subtle expression changes or trends opposite to those observed in the integrated and IVDD analyses (Supplementary Fig. 7E–G). These findings further support the robustness and biological relevance of the identified hub genes in IVDD-associated ferroptosis progression.

hdWGCNA

Using hdWGCNA, we identified several core gene modules associated with IVDD. A soft power threshold of 9 was determined to be optimal based on power analysis (Fig. 7A). Under this threshold, genes were grouped into 12 modules (Fig. 7B), and a correlation matrix was used to visualize the co-expression relationships between modules (Supplementary Fig. 6). KEGG enrichment analysis revealed that modules 6, 7, and 10 were significantly enriched in the ferroptosis pathway, indicating that these modules might play important roles in the onset and progression of IVDD through regulation of ferroptosis-related mechanisms (Fig. 7C). Correlation analysis with IVDD phenotype further revealed that modules 10 and 12 were significantly associated with the IVDD phenotype, suggesting their potential critical roles in disc degeneration (Fig. 7D). Analysis of the core genes in these modules identified several FRGs, such as RPL8 in module 6, ATF3, NR4A1, ATF4, SMAD7, MEG3, MEG8, CREB5, and ACSL3 in module 7, GPX4 and ATP5MC3 in module 10, and AKR1C2, CDO1, WWTR1, and ISCU in module 12 (Fig. 7E). These findings further underscore the pivotal role of ferroptosis in the pathogenesis and progression of IVDD.

Fig. 7. hdWGCNA.

A Screening of soft threshold and differentiation of gene modules; B 12 gene modules and the top 10 key genes in each module; C Network diagram of the KEGG enrichment analysis results of the song module genes constructed by cytoscape; D Bubble chart of the correlation between each module and IVDD; E The top 25 key genes of modules 6, 7, 10, and 12.

Molecular docking

To investigate the molecular mechanisms underlying Res’s therapeutic potential in IVDD, we performed molecular docking analyses with five key hub genes (GPX4, FTH1, HMOX1, NCOA4, and TFRC). The results showed that Res exhibited high binding affinity for these target proteins, with all binding energies lower than −5 kcal/mol (Fig. 8).

Fig. 8. Molecular docking.

A–E Molecular docking results of GPX4, FTH1, HMOX1, NCOA4, and TFRC with Res.

Clinical sample analysis

To explore the role of ferroptosis in IVDD, we examined the expression of the five hub genes in clinical samples. Western blot analysis revealed that, compared to the control group, the protein levels of GPX4 and FTH1 were significantly reduced in the IVDD group, while HMOX1, NCOA4, and TFRC were significantly upregulated (P < 0.05, Fig. 9A). The qPCR results corroborated the Western blot findings (P < 0.05, Fig. 9B). These results suggest that the five hub genes play a pivotal role in regulating ferroptosis during the progression of IVDD.

Fig. 9. Experimental verification.

A Protein expression and relative expression of the five core genes in the control and IVDD groups in clinical samples; B mRNA expression of the five hub genes in the control and IVDD groups in clinical samples; C Effects of different concentrations of Res on NPC proliferation at various intervention time points; D Effects of different concentrations of Res on NPC proliferation under IL-1β intervention; E Protein expression and relative expression of the five core genes in the control, IL-1β, and IL-1β+Res groups; F mRNA expression of the five hub genes in the control, IL-1β, and IL-1β+Res groups; G. X-ray images and relative disc height of intervertebral discs in the control, model, and model+Res groups; H HE staining results of intervertebral discs in the control, model, and model+Res groups.

Res promotes NPC proliferation

First, we assessed the effect of Res on NPCs proliferation. The results showed that after 24 and 48 h of treatment, Res at a concentration of 0.01 μmol/L did not show significant statistical differences compared to the control group. However, concentrations of 0.1 μmol/L, 1 μmol/L, 10 μmol/L, and 100 μmol/L of Res all promoted NPC proliferation (P < 0.05), with the most pronounced effect observed at 10 μmol/L. Moreover, after 48 h of intervention, the proliferative effects at all concentrations were less pronounced than at 24 h (Fig. 9C). Subsequently, we evaluated the impact of Res on NPC proliferation after 24 h of IL-1β treatment (10 ng/mL). The results showed that the model group exhibited significantly inhibited NPC proliferation compared to the control group (P < 0.05). Compared to the model group, Res at concentrations of 0.01 μmol/L, 0.1 μmol/L, 1 μmol/L, 10 μmol/L, and 100 μmol/L significantly alleviated the inhibitory effect of IL-1β on NPC proliferation (P < 0.05), with the most pronounced effect observed at 10 μmol/L (Fig. 9D). Thus, 10 μmol/L Res was selected for subsequent experiments.

Res inhibits ferroptosis in NPCs and alleviates IVDD

To explore the effect of Res on ferroptosis in NPCs, we first examined its impact in an in vitro IVDD model. After IL-1β treatment, Western blot analysis revealed that, compared to the control group, the protein expression levels of GPX4 and FTH1 were significantly reduced, while HMOX1, NCOA4, and TFRC were significantly upregulated in the model group (P < 0.05), indicating elevated ferroptosis in NPCs. However, Res treatment alleviated the IL-1β-induced downregulation of GPX4 and FTH1, while inhibiting the upregulation of HMOX1, NCOA4, and TFRC (P < 0.05, Fig. 9E). The qPCR results supported these findings (P < 0.05, Fig. 9F).

Next, we investigated the effect of Res in an in vivo IVDD model. X-ray analysis showed that the intervertebral disc height was significantly reduced in the model group, but Res treatment rescued the disc height reduction (P < 0.05, Fig. 9G). HE staining results revealed that, compared to the control group, the model group exhibited disrupted disc structure, with annular fibrosus rupture and nuclear extrusion. In contrast, the Res treatment group displayed a more intact and orderly disc structure, with better alignment of the annular fibrosus and more preserved NP morphology (Fig. 9H). These findings indicate that Res can alleviate IVDD by inhibiting ferroptosis NPCs.

Discussion

With the aging global population, the incidence of IVDD has significantly increased, becoming a pressing public health issue. Current treatment strategies for IVDD mainly focus on symptom relief, with a lack of effective disease-modifying therapies to delay or reverse its progression. Therefore, exploring new targeted therapeutic strategies is imperative.

In recent years, with advancements in understanding the mechanisms of cell death, various forms of programmed cell death (PCD), including apoptosis, pyroptosis, autophagy, and metal ion-dependent cell death (such as ferroptosis and copper death), have been shown to play critical roles in disease pathology^35–38. Ferroptosis, characterized by iron dysregulation and lipid peroxidation accumulation, involves key mechanisms such as oxidative stress and the abnormal accumulation of iron-dependent lipid peroxides³⁹. Studies have demonstrated that ferroptosis plays a key role in the pathogenesis of IVDD. GPX4, a critical enzyme regulating lipid peroxidation metabolism, converts lipid peroxides to lipid alcohols, thereby maintaining cellular membrane integrity⁴⁰. Research by Yang et al. revealed that the expression of GPX4 protein is significantly downregulated in the NP tissue of IVDD patients²⁷; Li et al. further confirmed that inhibiting ferroptosis in NPCs alleviates IVDD, suggesting its potential therapeutic role⁴¹. Ferroptosis has thus emerged as a promising target for IVDD treatment.

Res, a polyphenolic compound derived from natural plants, is known for its anti-inflammatory, antioxidant, anti-aging, and anti-cancer properties^42,43. Recent studies have increasingly highlighted the role of Res in regulating various forms of PCD, including ferroptosis^44–46. However, it remains unclear whether Res mitigates IVDD progression by modulating ferroptosis. Therefore, this study integrated bulk RNA-seq and single-cell transcriptome analysis, along with clinical samples and in vitro/in vivo experiments, to systematically investigate the potential mechanisms by which Res regulates ferroptosis to treat IVDD.

First, differential analysis of the GSE167199 transcriptome data, combined with FRGs and IVDD target genes, identified 22 ferroptosis-associated differentially expressed genes. By intersecting these with ferroptosis signaling pathway genes, five hub genes (GPX4, FTH1, HMOX1, NCOA4, and TFRC) were identified, suggesting their pivotal role in ferroptosis in IVDD. Functional enrichment analysis revealed that these 22 genes were significantly enriched in ferroptosis, HIF-1, p53 signaling pathways, and BP related to reactive oxygen species (ROS) levels and oxidative stress. Gene set variation analysis (GSVA) further confirmed that all five hub genes were significantly enriched in the ferroptosis pathway, indicating their involvement in the onset and progression of IVDD through ferroptosis regulation.

Subsequently, single-cell transcriptome data (GSE244889) was used for further analysis, revealing the presence of eight cell types in the samples. Gene localization analysis indicated that the five core genes were highly expressed in NPCs. Differential gene functional enrichment analysis of NPCs also pointed to the ferroptosis pathway, with pathway scoring showing significantly higher ferroptosis activity in the IVDD group compared to the control group. Pseudotime analysis demonstrated dynamic changes in the expression levels of the 22 ferroptosis-associated genes during NPC differentiation, where GPX4 and FTH1 expression decreased while HMOX1, NCOA4, and TFRC expression increased, further confirming the key roles of these genes in ferroptosis regulation. Additionally, hdWGCNA identified four gene modules significantly enriched in the ferroptosis pathway, two of which were closely associated with IVDD. These findings reinforce the critical role of ferroptosis in IVDD development (Figs. 7–9).

To verify the mechanism by which Res alleviates IVDD by regulating ferroptosis, we first analyzed clinical samples. The expression patterns of the five core genes were consistent with the transcriptomic data: GPX4 and FTH1 were significantly downregulated in IVDD samples, while HMOX1, NCOA4, and TFRC were significantly upregulated. In vitro experiments showed that Res modulated the expression of these core genes to inhibit ferroptosis in NPCs. Animal experiments further validated that Res significantly improved the pathological progression of IVDD.

In addition to directly modulating ferroptosis-related genes, resveratrol is known to influence broader upstream signaling pathways that regulate oxidative stress and inflammation. Previous studies have demonstrated that Res activates the SIRT1/SIRT3 axis, which enhances mitochondrial function and suppresses ROS accumulation, thereby indirectly attenuating ferroptotic injury^33,47,48. Moreover, Res can activate the Nrf2-mediated antioxidant pathway, promoting the transcription of downstream antioxidant enzymes such as HMOX1 and GPX4, both essential regulators of ferroptosis^49,50. These findings suggest that the observed anti-ferroptotic effects of Res may reflect its integrated regulation of the SIRT1/SIRT3–Nrf2–GPX4 signaling network, providing multifaceted cytoprotection for NPCs.

Furthermore, molecular docking results revealed relatively strong binding affinities between Res and 5 hub ferroptosis-related proteins. We acknowledge that such binding energies indicate potential interactions but do not confirm strong or biologically functional binding. Nevertheless, the docking analysis provides a theoretical basis for understanding how Res may influence ferroptosis-related molecular events. Specifically, Res binding to TFRC might modulate iron uptake, interaction with NCOA4 could affect ferritinophagy and iron release, and its potential association with GPX4 might stabilize this enzyme, reducing lipid peroxidation and oxidative stress. These hypotheses offer valuable directions for future experimental validation to elucidate the precise molecular mechanisms underlying Res-mediated protection against IVDD.

We acknowledge several limitations of this study. First, although both bulk and single-cell transcriptomic analyses and clinical validation were based on human NP tissues, the in vitro experiments were conducted using an immortalized NPC cell line rather than primary human NPCs, which may not fully recapitulate the in vivo cellular microenvironment. Second, although our results demonstrated that Res modulates ferroptosis-related genes and attenuates ferroptosis-associated changes in NPCs, the current evidence does not fully establish a direct causal relationship between ferroptosis inhibition and the protective effects of Res. It remains possible that Res may simultaneously regulate other signaling pathways or forms of programmed cell death, such as apoptosis, pyroptosis, or necroptosis, thereby contributing to improved NPC viability. Therefore, further mechanistic studies using pathway-specific inhibitors or activators, rescue experiments, gene knockdown or overexpression approaches, and additional ferroptosis-specific functional assays are needed to more definitively validate the role of ferroptosis in Res-mediated protection against IVDD. Third, the molecular docking analyses provide only theoretical predictions, and additional biochemical and structural studies are required to confirm these interactions. Finally, the clinical validation sample size was relatively small, and future studies with larger cohorts, in vivo functional experiments, and pathway-specific interventions are needed to comprehensively validate our findings.

In conclusion, this study comprehensively clarifies the molecular mechanism by which Res alleviates IVDD through the regulation of ferroptosis, offering novel therapeutic targets and directions for the treatment of IVDD. Additionally, it provides strong support for the clinical application of Res in degenerative diseases.

Methods

Acquisition and preprocessing of bulk RNA-seq data

The original TPM expression matrix of IVDD NP tissue was obtained from the Gene Expression Omnibus (GEO) database (GSE167199). This dataset includes 3 IVDD and 4 control NP tissue samples from individuals aged 29–65 years. To reduce the dynamic range of the data and make it more normally distributed, the TPM values for all samples were log2-transformed using the formula of log ₂(TPM + 1).

Identification of ferroptosis-related genes and drug targets of Res

Ferroptosis-related genes(FRGs) were retrieved from the FerrDb database, which contains genes involved in the regulation of ferroptosis along with their mechanisms of action. To identify potential drug targets of Res, we collected relevant targets from several databases, including the TCSMP database (https://old.tcmsp-e.com/tcmsp.php/), SwissTargetPrediction (http://www.swisstargetprediction.ch/), DrugBank (https://go.drugbank.com/), and TargetNet (http://targetnet.scbdd.com/), as well as from the existing literature. The targets obtained from these sources were intersected and deduplicated to generate the final set of Resveratrol drug targets.

Differential gene expression analysis in bulk RNA-seq data

To analyze DEGs expression between the normal and IVDD groups, we employed the limma package (version 3.56.2) in R⁵¹. The lmFit function was used for multiple linear regression, followed by the eBayes function to compute moderated t-statistics, F-statistics, and log-odds of differential expression via empirical Bayes moderation of standard errors towards a common value. Genes with a log fold change (logFC) > 0.5 and a p-value < 0.05 were classified as upregulated, while genes with logFC < −0.5 and a p-value < 0.05 were classified as downregulated. Differentially expressed genes were visualized using volcano plots and heatmaps generated by the ggplot2 package.

Identification of hub genes

To identify key genes potentially regulated by Res in ferroptosis, an intersection analysis was performed using the intersect() function to identify genes that were both differentially expressed and involved in ferroptosis regulation. In this study, ferroptosis-related genes (FRGs) refer to all genes reported to participate in the regulation of ferroptosis, including those influencing iron metabolism, lipid peroxidation, and oxidative stress. Additionally, we obtained a ferroptosis pathway genes (FPGs) set from the KEGG-defined ferroptosis pathway in the GSEA database, representing the core genes directly involved in the canonical ferroptosis signaling pathway, such as GPX4, FTH1, and TFRC. We then intersected the ferroptosis-related differentially expressed genes (FRGs) with the FPGs to identify core hub genes with significant biological roles in ferroptosis, which were further validated as possible therapeutic targets for Res in treating IVDD.

Construction of the protein-protein interaction (PPI) network

To investigate the interactions among ferroptosis-related differentially expressed genes, we imported these genes into the STRING database (version 12.0, https://cn.string-db.org/). A protein-protein interaction (PPI) network was constructed, and the network was visualized using Cytoscape software (version 3.7.2).

Enrichment analysis of FRDEGs

To gain a deeper understanding of the biological functions and potential mechanisms of ferroptosis-related differentially expressed genes, we conducted Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses. GO analysis was performed using the clusterProfiler package (version 4.2.2) in R⁵², providing functional annotations for the genes in the categories of cellular component (CC), molecular function (MF), and biological process (BP). KEGG pathway analysis was performed using the KEGG.db and clusterProfiler packages to identify the main biological pathways associated with the differentially expressed genes. The significance threshold was set at p-value < 0.05, with multiple comparisons adjusted using the Benjamini–Hochberg method. The results were visualized using dot plots, bar plots, and enrichment maps to reveal the potential biological functions and pathways of FRGs in IVDD.

Single-gene gene set enrichment analysis of hub genes

To further investigate the potential biological roles of ferroptosis-related differentially expressed genes in IVDD, we performed single-gene gene set enrichment analysis (GSEA). This was conducted using the clusterProfiler package in R. In this approach, gene sets were constructed based on the expression levels of each differentially expressed gene and compared with the KEGG gene sets to evaluate the enrichment of these genes in relevant pathways. The significance threshold was set at p-value < 0.05, with multiple hypothesis testing corrected using False Discovery Rate (FDR). The results were evaluated by the enrichment score (ES) and normalized enrichment score (NES), and visualized using enrichment plots and bar plots to further explore the potential roles of FRGs in IVDD.

Acquisition and quality control of single-cell data

Single-cell RNA-seq data were obtained from the GEO database (GSE244889), which includes 3 IVDD and 3 control NP tissue samples (Supplementary Table 13). Quality control and preprocessing were performed using the Seurat package (version 5.1.0)⁵³. First, genes expressed in fewer than three cells were excluded, and cells expressing fewer than 300 genes were removed to ensure high data quality. To further control for quality, we calculated the proportion of mitochondrial and erythrocyte genes using the PercentageFeatureSet function. Cells with mitochondrial gene proportions greater than 10%, erythrocyte gene proportions greater than 3%, gene counts outside the range of 300–700, or total counts greater than 100,000 were excluded. Subsequently, we used the DoubletFinder package (version 2.0.4) to identify and remove doublets⁵⁴. We set nExp to 0.08 × nCells²/10,000, pN to 0.25, and pK to 0.09 for doublet detection and removal.

Cell type annotation

For cell type annotation, we first normalized the counts expression matrix using the NormalizeData function in Seurat with the LogNormalize algorithm. We then identified the top 2000 highly variable genes (HVGs) using the FindVariableFeatures function with the vst method. After scaling and centering the HVGs using the ScaleData function, we performed principal component analysis (PCA) with the RunPCA function and selected the top 18 principal components for dimensionality reduction. To eliminate batch effects across different samples, we integrated the data using the Harmony algorithm. We constructed a neighbor graph using the FindNeighbors function (dims = 1:18) and performed unsupervised clustering with the FindClusters function at a resolution of 0.2 to identify cell clusters. We visualized cell distributions using UMAP with the RunUMAP function. Marker genes for each cluster were identified using the FindAllMarkers function (min.pct = 0.25, logfc.threshold = 0.5), and the clusters were annotated based on known cell type-specific markers from the literature and databases.

Hub gene localization and differential gene expression analysis in single-cell transcriptome atlas

To investigate the distribution of hub genes across different cell populations, we visualized the expression of each hub gene in individual cell clusters using the FeaturePlot function. This analysis helps to understand the localization and potential function of hub genes within specific cell types.

Next, to explore gene expression differences between cell populations, we performed differential gene expression analysis using the FindAllMarkers function in Seurat. This function calculates the log2 fold change (logFC) and statistical significance of each gene across cell populations. After performing differential analysis across all cell clusters, we focused on the control and IVDD groups to examine gene expression differences within each group. We specifically analyzed the cartilage cell population, comparing differential gene expression between the two groups. Genes with log2 fold change > 0.5 and adjusted p-values < 0.05 were considered upregulated, and those with log2 fold change < −0.5 and adjusted p-values < 0.05 were considered downregulated.

Enrichment analysis and pathway scoring in single-cell transcriptome atlas

To further explore the biological roles of differentially expressed genes, we conducted GO and KEGG pathway enrichment analysis using the clusterProfiler package. The enrichGO function was used for GO analysis, and the enrichKEGG function for KEGG pathway analysis. Additionally, to assess the activity of DEGs in specific pathways, we performed Gene Set Variation Analysis (GSVA). We used the gsva function to calculate pathway scores for each sample, evaluating the activity level of predefined gene sets across different pathways. GSVA results help identify key pathways associated with IVDD and assess differences in pathway activity between groups, providing further insight into the potential biological functions of DEGs in IVDD.

Pseudotime analysis

To investigate the dynamic changes and fate transitions of NPCs during the IVDD process, we first extracted all NPCs from the Seurat object using the subset() function. Unsupervised pseudotime trajectory analysis was then performed on this NPC subset using the Monocle2 package (version 2.28.0)⁵⁵. We set the LowerDetectionLimit parameter to 0.5 and used the negbinomial.size() parameter for appropriate processing of the single-cell RNA-seq data. The count data and metadata from the Seurat object were used to construct a CellDataSet object. We estimated size factors and dispersion using the EstimateSizeFactors and estimateDispersions functions, which are essential for subsequent analysis. Genes with significant variation across the trajectory were identified using the DispersionTable function. To define cell progression, we performed dimensionality reduction using the DDRTree algorithm in the ReduceDimension function and extracted key feature dimensions. Cells were ordered along the pseudotime trajectory using the OrderCells function based on their gene expression patterns, revealing their differentiation processes. Differential analysis of genes associated with trajectory changes at different nodes was performed using the BEAM function, identifying genes related to cell state transitions. Finally, the results were visualized with the plot_cell_trajectory function to display the dynamic changes of cells and their relative positions during the IVDD process.

In addition to the integrated analysis, pseudotime trajectory analyses were further performed separately for each individual sample (4 control samples and 3 IVDD samples) to evaluate the robustness and consistency of the observed gene expression patterns across subjects.

hdWGCNA

To identify highly co-expressed gene modules associated with IVDD, we applied hdWGCNA. This method integrates single-cell data to identify core gene modules associated with IVDD and explore their biological functions. We performed the analysis using the hdWGCNA package (version 0.2.22)⁵⁶. First, the SetupForWGCNA function was used to construct the hdWGCNA object, and cells were clustered using the K-nearest neighbor (KNN) algorithm implemented in the MetacellsByGroups function. The gene expression from each cluster was summarized into metacells, which simplified the data and reduced its complexity. The expression matrix was set using the SetDatExpr function, and a gene-gene correlation adjacency matrix was constructed to infer co-expression relationships. To reduce noise and retain strong connections, the correlation matrix was adjusted by raising it to a specific power. The soft power threshold was determined using the TestSoftPowers function, selecting a minimum power that ensured a scale-free topology fit of at least 0.8. Gene co-expression networks were constructed using the ConstructNetwork function, and modular analysis identified multiple gene modules. The ModuleEigengenes function was used to compute the module eigengenes, which were then correlated with clinical phenotypes (e.g., IVDD status). Finally, KEGG pathway enrichment analysis was performed on the gene modules using the enrichKEGG function, and the results were visualized using Cytoscape to display the network of modules and their associated biological pathways.

Mechanism of Res in the treatment of IVDD

Molecular docking

To explore the potential mechanisms of Res in treating IVDD, we conducted molecular docking to investigate the interactions between Res and five core genes (GPX4, FTH1, HMOX1, NCOA4, and TFRC) identified through bulk and single-cell transcriptome analyses. The three-dimensional structures of the target proteins were retrieved from the Protein Data Bank (PDB) (https://www.rcsb.org) and preprocessed using PyMOL (version 2.6.2), including the removal of water molecules and co-crystallized ligands. Subsequently, AutoDockTools (version 1.5.7) was used to add polar hydrogen atoms and save the processed protein structures in PDBQT format for docking analysis. Potential binding sites were predicted using the Proteins.plus website (https://proteins.plus/), which provided detailed information on the possible interaction regions, helping to identify the protein binding pockets. The molecular structure of Res was downloaded from the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP, https://old.tcmsp-e.com/tcmsp.php) and optimized using Chem3D, followed by conversion to PDBQT format. Molecular docking simulations were performed using AutoDock Vina, with the search space defined by the predicted binding pockets to focus the simulations on the relevant interaction sites. The docking results were evaluated based on binding affinity and binding modes, and the best docking conformations were visualized using PyMOL to illustrate the interaction patterns between Res and the target proteins.

Collection of clinical samples

This study was approved by the Ethics Committee of the Affiliated Traditional Chinese Medicine Hospital of Southwest Medical University [KY2021051-FS01] and adhered to the ethical guidelines outlined in the Declaration of Helsinki. Clinical NP tissue samples included 2 IVDD and 2 control samples (Supplementary Table 14). IVDD samples were collected from patients with disc herniation, while non-degenerative tissue samples were obtained from patients with vertebral fractures who had no history of IVDD. All patients signed informed consent forms and underwent disc excision or spinal fusion surgeries. During surgery, intervertebral disc tissues were immediately frozen in liquid nitrogen and stored at −80 °C for later use.

Reagents and materials

Resveratrol (product name: Resveratrol, CAS: 501-36-0; Cat. No. HY-16561, Batch No. 579789) was purchased from MedChemExpress (MCE, Shanghai, China). The product was provided with a Certificate of Analysis (COA), HPLC, HNMR, LC-MS, and elemental analysis reports to ensure identity and purity. According to the COA, the purity of Resveratrol used in this study was 99.94%. Structural verification was performed by HNMR and LC-MS.

Human immortalized NPCs (iCell-0028a) were obtained from iCell Bioscience Inc. Interleukin-1β (IL-1β, HY-P7028) was also purchased from MCE. MEM/F12 medium (Cat. No. A4192001) and fetal bovine serum (FBS, A5670701) were acquired from Gibco. Antibodies against GPX4 (TD6701), FTH1 (PA4412), TFRC (T56618), HMOX1 (T55113), and NCOA4 (TD4255) were obtained from Abmart. Relevant primers were synthesized by Amresco. CCK8 assay kit (96992) and BCA protein quantification kit (71285-M) were purchased from Sigma. Reverse transcription kits (HY-K0510A) and qPCR kits (HY-K0501) were obtained from MedChemExpress.

Preparation of reagents

Res (10 mg) was dissolved in 224.03 μL DMSO to prepare a 100 mmol/L stock solution, which was stored at −80 °C. For experimental use, Res was diluted to the desired working concentration in MEM/F12 medium containing 10% FBS. The IL-1β solution was prepared in PBS to the required working concentration and used immediately.

Culture and treatment of human NPCs

NPCs were cultured in MEM/F12 medium containing 10% FBS at 37 °C in a 5% CO₂ incubator. Cells were regularly monitored under a microscope for growth. When cell confluence reached ~80%, cells were passaged. Only cells from passages 3–4 were used in the experiments. To establish an in vitro IVDD model, NPCs were treated with IL-1β for 24 h at an appropriate concentration. In the Res treatment group, cells were exposed to Res at a specified concentration for the required duration after IL-1β treatment.

CCK8 assay

To evaluate the effect of Res on NPC proliferation, NPCs were seeded at a density of 5000 cells/well in a 96-well plate and treated with various concentrations of Res (0, 0.01, 0.1, 1, 10, 100 μmol/L) for 24 and 48 h. In the IVDD model, NPCs were pre-treated with IL-1β for 24 h, followed by Res treatment at different concentrations. Each group included six replicates. After treatment, the culture medium was removed, and 100 μL of diluted CCK8 solution (10x) was added to each well and incubated at 37 °C for 1 h. Absorbance at 450 nm was measured using a multifunctional microplate reader.

Western blot

For clinical samples, tissue proteins were extracted using RIPA lysis buffer containing a protease inhibitor. After ultrasonic disruption, samples were centrifuged at 12,000 rpm for 10 min at 4 °C, and the supernatant was collected. Protein concentration was measured using a BCA protein assay kit.

NPCs were divided into normal, model (IL-1β), and treatment (IL-1β + Res) groups. NPCs were seeded at 2 × 10⁵ cells/well in a 6-well plate. In the treatment group, cells were exposed to 10 ng/mL IL-1β for 24 h, followed by Res treatment at the optimal concentration for 24 h. The model group was treated with MEM/F12 medium after IL-1β exposure, and the normal group was maintained in MEM/F12 medium throughout. After treatment, cells were washed three times with PBS, and total protein was extracted and quantified.

Proteins were separated by SDS-PAGE, transferred to PVDF membranes, and incubated with the corresponding primary and secondary antibodies. Protein bands were visualized using a chemiluminescence imaging system, and band intensities were analyzed with ImageJ software.

qPCR analysis

Total RNA was extracted from clinical samples and NPCs using TRIzol reagent according to the manufacturer’s instructions, and RNA concentration and purity were determined by spectrophotometry. The RNA (1 µg) was reverse transcribed into cDNA using a reverse transcription kit. Quantitative PCR was performed, and the relative expression levels of target genes were calculated using the 2^(−ΔΔCt) method. Statistical analysis of the relative expression levels of different genes in various sample groups was performed, and the results were visualized using bar charts.The primer sequences used in the experiments are listed in Supplementary Table 15.

X-ray and HE staining

All experimental rats were provided by the Experimental Animal Center of Southwest Medical University. All procedures involving animals were conducted in accordance with the ARRIVE guidelines and relevant national and institutional regulations for the care and use of laboratory animals. The animal study protocol was reviewed and approved by the Ethics Committee for Animal Experiments of Southwest Medical University (Approval No.20240412-001).

Eight-week-old male Sprague-Dawley (SD) rats were randomly divided into three groups: sham surgery, IVDD model, and Res treatment (n = 8). Randomization was performed using a random number table, and data analysis was conducted in a blinded manner. The IVDD model was established by percutaneous puncture of the L3-L5 intervertebral discs using an 18 G needle. In the sham group, the same procedure was performed without disc puncture. One month after modeling, rats in the Res treatment group received resveratrol (40 mg/kg/day) dissolved in distilled water by oral gavage for 4 weeks, while the sham and IVDD groups received equal volumes of vehicle. All animals were anesthetized prior to imaging and euthanasia using intraperitoneal injection of pentobarbital sodium (50 mg/kg). Anesthesia depth was confirmed by absence of pedal reflex. At the end of the experiment, rats were euthanized by overdose of pentobarbital sodium (150 mg/kg, intraperitoneally), ensuring unconsciousness prior to cessation of cardiac and respiratory function, in accordance with AVMA guidelines. After euthanasia, L3-L5 intervertebral discs were harvested, fixed in 4% paraformaldehyde, decalcified in EDTA, dehydrated, embedded in paraffin, sectioned, and stained with hematoxylin and eosin (HE) for histopathological evaluation. X-ray imaging of the tail was performed under anesthesia, and disc height index (DHI) was measured using ImageJ software.

Data processing and statistical analysis

Statistical analyses were performed using GraphPad Prism 9. Normality and homogeneity of variance tests were conducted before analysis to ensure the applicability of the chosen methods. For clinical samples, independent samples t-tests were performed, and for NPC data, one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test was used to assess group differences. All experimental data are expressed as mean ± standard deviation (Mean ± SD). Statistical significance was set at P < 0.05.

Author contributions

Xiaoqiang Wang: Writing – original draft, Formal analysis. Jiale Lv: Writing – original draft. Daqian Zhou and Weiye Cai: Data curation. Chao Song: Resources, Supervision, Visualization. Zhijiang Fu: Writing – review & editing. Zongchao Liu: Conceptualization, Funding acquisition, Writing – review & editing. All authors have reviewed and approved this paper for publication.

Data availability

The bulk RNA sequencing data analyzed in this study were obtained from the Gene Expression Omnibus (GEO) database under the accession number GSE167199. The single-cell RNA sequencing dataset used in this study was also retrieved from the GEO database under the accession number GSE244889. All data are in the manuscript or supporting information files.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41540-026-00761-9.