STAT3/POSTN/GSTP1/JNK Axis Orchestrates Ferroptosis in Nucleus Pulposus Cells: A Potential Therapeutic Target for Intervertebral Disc Degeneration

Abstract

Objective

This study aimed to investigate the role of the STAT3/POSTN/GSTP1/JNK axis in ferroptosis and extracellular matrix (ECM) metabolic imbalance in nucleus pulposus cells (NPCs) during intervertebral disc degeneration (IDD) and to explore therapeutic strategies targeting this axis.

Methods

Using integrated multiomics sequencing, transcriptional regulation assays (chromatin immunoprecipitation, dual‑luciferase reporter), protein interaction analysis (CoIP), and other molecular biology approaches, we systematically elucidated the regulatory role of the STAT3/POSTN/GSTP1/JNK axis in ferroptosis of NPCs during IDD. Functional validation was performed in POSTN‑edited cell and rat models as well as in a needle‑puncture‑ induced rat IDD model. A small‑molecule candidate targeting this axis was identified through virtual screening, molecular docking, and molecular dynamics simulations.

Results

Periostin (POSTN) expression increased during ferroptosis and induced ferroptosis and ECM metabolic imbalance in NPCs in a concentration- and time-dependent manner. STAT3 was identified as a transcriptional regulator of POSTN and functionally coupled with POSTN to form a self-amplifying positive feedback loop, accelerating ferroptosis progression. Furthermore, POSTN impaired the binding of the GSTP1/JNK complex, leading to the depletion of cellular glutathione. Chemical screening identified pristimerin (PN) as a potential GSTP1-targeting compound, targeting the STAT3/POSTN/GSTP1/JNK axis and delaying IDD progression.

Conclusion

This study identified the important role of the STAT3/POSTN/GSTP1/JNK axis in regulating ferroptosis and ECM metabolism in NPCs and highlighted PN as a promising candidate therapeutic agent for IDD. These findings provide new insights into the molecular mechanisms underlying IDD and offer new targeted therapeutic avenues for IDD.

INTRODUCTION

Low back pain is a common, chronic musculoskeletal disorder. Intervertebral disc degeneration (IDD) is considered the primary cause and imposes a substantial burden on both individuals and society [1,2]. Chronic progression can lead to various spinal disorders including disc herniation, spinal stenosis, and spondylolisthesis [3,4]. IDD is characterized by an imbalance between the anabolic and catabolic processes. This is often accompanied by a gradual loss of water content in the nucleus pulposus (NP) and a marked reduction in extracellular matrix (ECM) components [5,6]. However, the specific molecular mechanisms underlying the initiation and progression of IDD remain unclear and require further investigation.

Programmed cell death (PCD) is a genetically regulated process crucial for maintaining homeostasis [7,8]. Multiple forms of PCD, including apoptosis, necroptosis, and ferroptosis, have been observed during the context of IDD initiation and progression [9,10]. Cells undergoing ferroptosis exhibit distinct organelle morphological features, with the core changes concentrated in mitochondria: mitochondrial shrinkage with reduced or absent cristae, increased mitochondrial membrane density, and outer mitochondrial membrane rupture [11,12]. Ferroptosis is an important contributing factor in IDD progression [13-15]. Collectively, these findings indicate that ferroptosis is a potential emerging therapeutic target for IDD.

Periostin (POSTN) is a nonstructural matricellular protein. It is typically expressed at low levels in most normal adult tissues and cells but is highly upregulated at sites of injury and under pathological conditions [16]. POSTN expression is elevated in IDD and positively correlated with the severity of degeneration [17,18]. POSTN can promote IDD progression by activating the Wnt/β-catenin and nuclear factor-kappa B (NF-κB) signaling pathways, thereby inducing inflammation, apoptosis, and ECM metabolic imbalance in NP cells (NPCs). Wu et al. [19] have reported that under mechanical stress, a Piezo1-induced, self-amplifying loop involving NF-κB and POSTN can accelerate NPCs senescence, thereby exacerbating IDD. This collective evidence suggests that POSTN may serve as a potential molecular target for the prevention and treatment of IDD.

Furthermore, Ma et al. [20] and Wang et al. [21] have reported that increased POSTN expression is associated with increased cellular susceptibility to ferroptosis. However, the specific mechanism through which POSTN induces ferroptosis in IDD remains unclear. Therefore, this study aimed to systematically elucidate the molecular mechanism by which POSTN induces ferroptosis in NPCs, including its upstream transcriptional regulation, downstream effector pathways, and the associated ECM metabolic imbalance, and to identify small‑molecule compounds targeting this axis for potential IDD therapy.

MATERIALS AND METHODS

1. Collection of Human NP Samples

Human NP samples were obtained from individuals undergoing spinal surgery. Patient selection was based on conditions including spinal trauma, idiopathic scoliosis, or intervertebral disc (IVD) herniation. Exclusion criteria encompassed spinal infections, immunosuppressive diseases, hypertension, and diabetes. A total of 24 NP samples were collected from 14 male and 10 female participants, aged 16 to 72 years. The extent of disc degeneration was evaluated using the Pfirrmann grading system. Samples were categorized into 2 groups based on degeneration severity: a mild group (Pfirrmann grade II) and a severe group (Pfirrmann grade IV). The study protocol was approved by the Ethics Committee of Lanzhou University Second Hospital (approval No.: 2024A-258), and informed consent was obtained from all participants.

2. Culture and Treatment of NPCs

The immortalized human NPC cell line was purchased from Shanghai YaJi Biological Technology Co., Ltd. (0028a). The cells were cultured in DMEM/F12 medium (11320033, Gibco, USA) supplemented with 15% fetal bovine serum (BL205B, Biosharp) and 1% streptomycin-penicillin (15140122, Gibco). They were maintained in an incubator at 37°C with 5% CO₂. Upon reaching full confluence, the cells were passaged using a 0.25% trypsin-EDTA solution (25300054, Gibco) and subsequently reseeded at an optimal density in T25 flasks.

For recombinant POSTN (rPOSTN) treatment, NPCs were treated with rPOSTN at various concentrations (5, 10 μg/mL) for 24 hours for the concentration‑gradient experiments, or with 5-μg/mL rPOSTN for various durations (0, 12, 24 hours) for the time‑course experiments. For ferroptosis induction, NPCs were treated with Erastin (S7242, Selleck, USA) (2 μM, 24 hours). For pathway intervention experiments, the STAT3 inhibitor Stattic (HY-13818, MCE, USA) was used at 5 μM for 12 hours, and the STAT3 activator Colivelin (HY-P1061, MCE) was used at 0.5 μM for 12 hours. For pristimerin (PN) treatment, NPCs were treated with PN (HY-N1937, MCE) at 100 and 200 nM for 24 hours.

3. Lentiviral Transfection

The lentiviral transfection procedure was performed as previously described. Briefly, when cell density reached approximately 30%, transfection was carried out in culture flasks using lentiviruses (GeneChem, China) at a multiplicity of infection of 100. Transfection efficiency was assessed by Western blotting analysis.

4. Western Blotting

Total protein was extracted from NPCs using radio immunoprecipitation assay buffer (P0013B, Beyotime, China). Protein concentration was determined by the bicinchoninic acid (BCA) method. Subsequently, proteins were separated by 10% sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis and transferred onto polyvinylidene difluoride membranes. After blocking with 3% bovine serum albumin for 30 minutes, the membranes were incubated with the following primary antibodies: POSTN (1:1,000, ab315104, Abcam, USA), acyl-CoA synthetase long chain family member 4 (ACSL4; 1:1,000, T510198, Abmart), glutathione peroxidase 4 (GPX4; 1:1,000, T56959, Abmart), prostaglandin-endoperoxide synthase 2 (PTGS2; 1:1,000, 27308-1-AP, Proteintech, USA), FERRITIN (1:1,000, T55648, Abmart), collagen II (1:500, PTH2739, Abmart), matrix metalloproteinase-3 (MMP3; 1:1,000, T55188, Abmart), STAT3 (1:1,000, T55292, Abmart), phosphorylated STAT3 (p-STAT3; 1:1,000, T56566, Abmart), interleukin 6 (IL6; 1:1,000, 21865-1-AP, Proteintech), glutathione S-transferase Pi 1 (GSTP1; 1:2,000, 15902-1-AP, Proteintech), phospho-c-Jun N-terminal kinase (p-JNK; 1:2,000, 80024-1-RR, Proteintech), and GAPDH (1:5,000, M20006, Abmart). The membranes were then incubated with secondary antibodies: anti-rabbit immunoglobulin G (IgG; 1:5,000, SA00001-2, Proteintech) and anti-mouse IgG (1:7,000, SA00001-1, Proteintech). Finally, proteins on the membranes were detected using an enhanced chemiluminescence detection reagent (Invitrogen, USA).

5. Immunofluorescence Staining

NPCs were seeded in flat-bottom 6-well plates at a density of 5×105 cells per well. Subsequently, the cells were fixed with 4% paraformaldehyde, permeabilized with 0.5% Triton X-100 in phosphate-buffered saline (PBS) for 10 minutes, and blocked with PBS containing 1% bovine serum albumin. Following this, the cells were incubated with appropriate primary antibodies at 4°C overnight. After washing, the cells were treated with Alexa Fluor 488/594-conjugated Goat Anti-Rabbit IgG (1:300, SA00013-2/SA00013-4, Proteintech) for 1 hour at room temperature. Imaging was then performed using a laser scanning microscope (Olympus, Japan).

6. Immunohistochemistry Staining

IVD specimens from both human and rat sources were collected. The specimens were fixed in 4% paraformaldehyde (G1101, Servicebio, China), decalcified (except for human IVD specimens) using a decalcification solution (G1105, Servicebio, China), dehydrated, and subsequently embedded in paraffin. Sections were cut at a thickness of 5 μm. The sections were first deparaffinized and rehydrated, followed by antigen retrieval using a citrate-based antigen retrieval buffer (MVS-0100, MXB, China). Subsequent steps were performed according to the instructions of the ready-to-use Immunohistochemistry Ultrasensitive SP Kit (KIT-9710, MXB, China). The primary antibodies used in the experiments included: POSTN (1:200, 19899-1-AP, Proteintech), ACSL4 (1:200, T510198, Abmart), GPX4 (1:200, T56959, Abmart), STAT3 (1:200, T55292, Abmart), and GSTP1 (1:100, 15902-1-AP, Proteintech). Histological fields were randomly selected from each section, and imaging was performed using an Olympus BX63 microscope (Olympus, Japan).

7. Histological Staining

Tissue samples were fixed in 4% formaldehyde, decalcified (except for human IVD specimens), dehydrated, embedded in paraffin, and sectioned at a thickness of 5 μm. The sections were then subjected to histological staining and evaluation using hematoxylin and eosin (H&E) staining as well as Safranin O/Fast Green (S-O) staining.

8. Ferrous Iron Detection

Fe²⁺ levels were measured according to the manufacturer’s instructions for the FerroOrange probe (F374, Dojindo, Japan).

9. Measurement of Lipid Peroxidation

Lipid peroxidation levels were measured using a C11 BODIPY probe (D3861, Thermo Fisher, USA) according to the manufacturer’s instructions.

10. Measurement of Malondialdehyde

Measurement of malondialdehyde (MDA) levels were measured using an MDA assay kit (S0131, Beyotime, China) according to the manufacturer’s instructions.

11. Measurement of GSH

GSH levels were measured using Monochlorobimane (HY-101899, MCE) and N-(9-Acridinyl)maleimide (HY-W011618, MCE) according to the manufacturer’s instructions.

12. GST Activity Assay

GST activity was measured using a GST activity kit (Solarbio, China) according to the manufacturer’s protocol. It was defined as the amount of enzyme that was required to reflect the ability to reduce GSH and 1-chloro-2, 4-dinitrobenzene (CDNB). The changes in absorbance of the GSH and CDNB were recorded at 340 nm for 10 seconds and 310 seconds respectively. GST activity was expressed as nmol per min per mg of total protein concentration.

13. Transmission Electron Microscopy Analysis

Transmission electron microscopy (TEM) was employed to examine the mitochondrial morphology in NPCs. Cells were washed with PBS, fixed with 4% glutaraldehyde, and dehydrated through an ethanol gradient. Following embedding in epoxy resin and rinsing with propylene oxide, ultrathin sections were stained with uranyl acetate and lead citrate. Images were acquired using a Hitachi TEM system operating at 80 kV.

14. RNA-seq Data Analysis

Total RNA was extracted from NPCs with or without POSTN overexpression (3 independent biological replicates per group) and subjected to high-throughput transcriptome sequencing on the Illumina HiSeq platform by Qinglian Biotechnology (Beijing) Co., Ltd. Following quality control and read alignment, differentially expressed genes (DEGs) were identified using DESeq2 software. DESeq2 employs median‑of‑ratios normalization and the Wald test with Benjamini-Hochberg correction for multiple testing. Genes with |log₂(fold change [FC])|≥1 and adjusted p-value (false discovery rate [FDR])<0.05 were considered significantly differentially expressed. Functional enrichment analysis of DEGs was performed using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases.

15. Proteomic Sequencing and Data Analysis

NP tissues from wild-type (WT) and POSTN^-/- rats (3 biological replicates per group) were lysed in sodium dodecyl sulfate‑ Tris buffer (4% SDS, 100 mM Tris-HCl, pH 7.6) and sonicated on ice. After centrifugation (4°C, 15,000 g, 10 minutes), the supernatant protein concentration was determined using the BCA method. Equal amounts of protein from each sample were reduced with 5-mM dithiothreitol at 56°C for 30 minutes, alkylated with 15-mM iodoacetamide at room temperature in the dark for 15 minutes, and digested with trypsin using the filteraided sample preparation method. The resulting peptides were desalted, quantified, and analyzed on a Vanquish Neo UHPLC system coupled to an Astral high-resolution mass spectrometer (Thermo Fisher Scientific) operating in data-independent acquisition (DIA) mode. Raw DIA data were searched against the UniProt Rattus norvegicus database using DIA-NN software, with the FDR set to <1% at both protein and peptide levels. Label-free quantification was performed using the MaxLFQ algorithm. Differentially expressed proteins (DEPs) were identified based on |log₂(fold change)|≥1 and adjusted p-value (FDR) <0.05. Functional enrichment analysis of DEPs was conducted using Fisher exact test with the total identified proteins as background.

16. Coimmunoprecipitation Assay

The coimmunoprecipitation (CoIP) assay was performed according to the manufacturer’s instructions using a protein A/G Co-Immunoprecipitation Kit (22202-100, Beaver, China).

17. Chromatin Immunoprecipitation (ChIP) Assay and ChIP-Polymerase Chain Reaction

The ChIP assay was performed using a High-Sensitivity ChIP Kit (ab185913, Abcam) according to the manufacturer’s instructions. Following this protocol, the eluted DNA was subjected to ChIP-polymerase chain reaction (ChIP-PCR). The primer sequences used in the ChIP assay were as follows: F1, 5´-TCTCCTACATGGTTCTTCAGCA-3´; R1, 5´-GGTGTTATTGCAGTTCACATCA-3´; F2, 5´-CAATAATCTTGAGCACACAGACA-3´; R2, 5´-GAACAATTGACTCACTGCATGTT-3´; F3, 5´-TGAAAAGGAAAAGTAGCTCAATG-3´; R3, 5´-GGCTTTGGAGAGCTCAGAAT-3´.

18. Dual-Luciferase Assay

Full-length STAT3 cDNA was cloned into the pIRES2 vector (Clontech, USA) to construct the STAT3 expression plasmid pIRES2-STAT3. Wild-type or mutant fragments of the POSTN promoter at the corresponding sites were cloned into the pGL3 plasmid (Promega, USA) to construct the respective POSTN reporter pGL3 plasmids. The recombinant constructs were then transfected into cells using Lipofectamine 3000 (Invitrogen). After 48 hours of culture, cells were collected, and luciferase activity was measured using the Dual-Luciferase Reporter Assay System (Promega).

19. Molecular Docking

Molecular docking was used to study the molecular affinity of the small molecule hypericin, pristimerin and verticinone to the protein target GSTP1. The protein crystal structures were predicted from Alphafold3, and 3-dimensional structures of small molecules were downloaded from the PUBCHEM database. We performed the molecular docking work by employing AutoDock Vina 1.1.2 software. Prior to docking, PyMol 2.5 was used to process all receptor proteins (including removal of water molecules, salt ions and small molecules). ADFRsuite 1.0 was used to convert all processed small molecules and receptor proteins into the PDBQT format required for docking with AutoDock Vina 1.1.2. The docked conformation with the highest molecular docking score was considered to be the binding conformation for subsequent molecular dynamics (MD) simulations. We analyzed and compared the binding site (BS) poses, chemical bond lengths and chemical bond angles of the original crystal ligand to the protein by redocking the original crystal ligand and the protein using the original crystal ligand of the protein target as a positive reference. Finally, the consistency of the binding mode can indicate the correctness of the molecular docking scheme.

20. Molecular Dynamics Simulation

Based on the docking results, the best-posed complex was subjected to MD simulation studies using the Groningen Machine for Chemicals Simulations (GROMACS) 5.0 package with a CHARMM36 force field under periodic boundary conditions for molecules. Ligand topology files were generated using the CHARMM General Force Field. The charge of the system was neutralized by the addition of the ions. The energy was minimized using a steepest-gradient method to remove any close contacts. The particle mesh Ewald method was employed for energy calculation and for electrostatic and Van der Waals interactions. The systems were equilibrated in the constant Number of particles, Volume, and Temperature (NVT) ensemble for 50,000 steps, followed by equilibration in the constant Number of particles, Pressure, and Temperature (NPT) ensemble for an additional 50,000 steps. Finally, 100 ns MD simulations were performed at 300 K with a 2.0 fs time step, and coordinates were saved every picosecond for analysis.

21. Cellular Thermal Shift Assay

Cell lysates treated with dimethyl sulfoxide or PN were divided into 9 equal aliquots and heated across a temperature gradient from 37°C to 69°C. Two freeze-thaw cycles were performed using liquid nitrogen, and the supernatants were collected for subsequent analysis.

22. Establishment of IDD Rat Model and Magnetic Resonance Imaging

An IDD model was established by puncturing the rat tail IVD, based on a previously reported method and our laboratory experience [22]. The intervertebral space between the 7th/8th caudal vertebrae was located. The skin and muscle layers were incised using a sterile disposable scalpel to expose the white annulus fibrosus. A 21G needle was then inserted vertically into the NP to a depth of approximately 5 mm. The needle was rotated clockwise by 360°, held in place for 30 seconds, and subsequently injected with the corresponding lentivirus using a microsyringe. For PN treatment groups, PN was administered intraperitoneally at 5 mg/kg and 10 mg/kg twice per week for 8 weeks after IDD induction. Eight weeks after surgery, the rats were anesthetized and subjected to imaging using a 3.0 T MRI system (Siemens, Germany).

23. Construction of POSTN Knockout Rat Model

Based on our team’s previous studies [23], to establish the POSTN knockout rat model, CRISPR/Cas9 gene-editing technology was employed. The precise genomic locus of the rat POSTN gene was identified, and a specific single guide RNA (sgRNA) targeting this region was designed. The sgRNA, together with Cas9 protein, was coinjected into rat zygotes to induce DNA doublestrand breaks, resulting in targeted gene disruption. The microinjected zygotes were cultured in vitro for further development and subsequently transferred into the uteri of pseudopregnant surrogate females using standard embryo transfer techniques. After birth, genomic DNA from offspring was extracted for PCR screening and sequencing to identify individuals with successful biallelic POSTN deletion. The entire process adhered strictly to the animal use and care guidelines of the affiliated institution, ensuring ethical and scientific integrity.

24. Statistical Analysis

All experimental data were obtained from at least 3 independent replicates. Data are expressed as mean±standard deviation. Normality of data was assessed by the Shapiro-Wilk test, and homogeneity of variances was evaluated by Levene test. For comparisons between the 2 groups, an independent samples t‑test was used for normally distributed data with equal variances; otherwise, the Mann-Whitney U-test was applied. For comparisons among multiple groups, 1‑way analysis of variance (ANOVA) followed by Tukey post hoc test was performed when data met the assumptions of normality and equal variance; when these assumptions were violated, Welch ANOVA with the Games-Howell post hoc test or the Kruskal-Wallis test with Dunn multiple comparison correction was used as appropriate. Statistical analysis was conducted using GraphPad Prism ver. 10.5.0 (San Diego, USA). A p-value of less than 0.05 was considered statistically significant.

RESULTS

1. POSTN is Upregulated in Ferroptosis

High-throughput sequencing analysis of NPCs treated with recombinant POSTN (rPOSTN) versus normal NPCs was performed, with a focus on identifying DEGs using thresholds of |log₂(FC)| ≥1 and p<0.05. POSTN expression showed a positive correlation with ACSL4, PTGS2, and MMP3, and a negative correlation with GPX4 (Fig. 1A and B). ACSL4 is a rate‑limiting enzyme that promotes ferroptosis by promoting the synthesis of polyunsaturated fatty acids and lipid peroxidation; its upregulation, together with that of PTGS2, constitutes a classical hallmark of ferroptosis activation. GPX4 is an important negative regulator of ferroptosis responsible for detoxifying lipid peroxides, and its downregulation indicates impaired ferroptosis defense [10,24]. MMP3 is a major marker of ECM catabolism, and its elevation reflects enhanced ECM degradation. GO enrichment analysis indicated that DEGs were primarily associated with the positive regulation of STAT proteins, extracellular iron ion influx, and lipid metabolic processes. KEGG analysis highlighted significant enrichment of JAK-STAT signaling and ferroptosis (Fig. 1C).

Fig. 1.

POSTN is upregulated in ferroptosis. (A and B) Volcano plot and heat map of differentially expressed genes (DEGs) from transcriptome sequencing. The screening criteria were | log₂(fold change) | ≥1 and p<0.05. (C) GO and KEGG enrichment analysis of the DEGs. (D) Representative T2-weighted MRI images of spine from patients with Pfirrmann grade II and grade IV. (E) H&E and S-O staining of nucleus pulposus (NP) specimens from patients with different Pfirrmann grades. (F and H) Immunohistochemistry staining and quantitative analysis of POSTN, ACSL4, and GPX4. (G) Immunofluorescence (IF) staining of POSTN in NP specimens from patients with different Pfirrmann grades. (I and J) Western blotting and quantitative analysis of POSTN, ACSL4, and GPX4 in NP specimens from patients with different Pfirrmann grades. (K and L) Representative T2- weighted MRI images of rat IVDs and H&E and S-O staining of rat NP specimens from different groups. (M) IF staining of POSTN, ACSL4, and GPX4 in rat IVDs from different groups. MRI, magnetic resonance imaging; POSTN, periostin; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; H&E, hematoxylin and eosin; S-O, Safranin O/Fast Green; IVD, intervertebral disc; BP, biological process; ACSL4, acyl-CoA synthetase long chain family member 4; GPX4, glutathione peroxidase 4; WT, wild-type; IDD, intervertebral disc degeneration; DAPI, 4´,6-diamidino-2-phenylindole. Data are presented as mean±standard deviation from at least 3 independent experiments. ***p<0.001.

Human NP samples were collected and rigorously assessed using MRI, H&E staining, and S-O staining. Results showed that in the mild group, ECM staining was uniform, and the fibrous structures were loosely organized and relatively ordered, with sparsely distributed cells and large intercellular spaces. The severe group exhibited reduced proteoglycan and collagen fiber content, a significant decrease in NPCs number, and abnormal cell proliferation and cluster formation (Fig. 1D and E). Western blotting and immunohistochemistry showed that, compared with the mild group, POSTN and ACSL4 were increased in the severe group, whereas GPX4 showed the opposite trend (Fig. 1F–J).

A POSTN^-/-rat model was established using CRISPR/Cas9 technology, and a rat IDD model was induced via needle puncture. MRI results revealed a significant reduction in the T2 signal in the IVD of the WT+IDD group. In contrast, the POSTN^-/-+ IDD group showed an enhanced T2 signal. H&E and S-O staining confirmed obvious NP fibrosis, reduced number of ECM components and NPCs, and decreased disc height in the WT+ IDD group. However, in the POSTN^-/-+IDD group, NP fibrosis was reduced, the number of NPCs and ECM components was increased, and the height of the intervertebral space was greater (Fig. 1K and L). IF results showed that POSTN knockout decreased ACSL4 expression and increased GPX4 expression (Fig. 1M).

To further investigate POSTN expression levels during ferroptosis, NPCs were treated with Erastin, a classic ferroptosis‑specific inducer, at a concentration of 2 μM for 24 hours. Western blotting results confirmed that Erastin treatment induced ferroptosis in NPCs (Supplementary Fig. 1A and B). Western blot-ting also showed that POSTN levels were significantly increased in the Erastin-treated group (Supplementary Fig. 1C and D). The IF results were consistent with the Western blotting results (Supplementary Fig. 1E and F).

2. POSTN Induces Ferroptosis and ECM Metabolic Imbalance in NPCs

To investigate whether exogenous POSTN influenced ferroptosis in NPCs, rPOSTN was added. TEM revealed that POSTN treatment altered mitochondrial structure, resulting in shrunken inner membranes with increased electron density and reduced cristae (Fig. 2A). The NPCs were treated with increasing concentrations of rPOSTN. Western blotting showed that higher rPOSTN concentrations progressively upregulated ACSL4, PTGS2, and MMP3 expression and downregulated GPX4, FERRITIN, and collagen II (Fig. 2B and C). The IF staining results were consistent with the Western blotting (Fig. 2D–F). Assays for lipid peroxidation, MDA, and Fe²⁺ levels yielded the corresponding results (Fig. 2G–K). Furthermore, similar to the concentration gradient, extending the duration of rPOSTN treatment increased ferroptosis levels in NPCs (Fig. 2L and M; Supplementary Fig. 1G–N). To distinguish the effects of rPOSTN, POSTN was overexpressed via lentiviral transfection to assess ferroptosis in NPCs. The results showed that the lentiviral vector (LV)-POSTN exhibited upregulation of MMP3, ACSL4, and PTGS2 and downregulation of collagen II, GPX4, and FERRITIN (Fig. 2N and O).

Fig. 2.

POSTN induces ferroptosis and extracellular matrix metabolic imbalance in nucleus pulposus cells (NPCs). (A) Representative transmission electron microscopy images of mitochondrial morphology in NPCs from different treatment groups. (B and C) Western blotting and quantitative analysis of collagen II, MMP3, ACSL4, PTGS2, GPX4, and FERRITIN expression in different experimental groups. (D–F) Immunofluorescence staining and quantitative analysis of ACSL4 and GPX4 expression in different experimental groups. (G and H) Lipid peroxidation levels and corresponding statistical analysis under different POSTN concentrations. (I) Measurement of malondialdehyde (MDA) levels under different POSTN concentrations. (J and K) Fe²⁺ levels and corresponding statistical analysis under different POSTN concentrations. (L and M) Western blotting and quantitative analysis of collagen II, MMP3, ACSL4, PTGS2, GPX4, and FERRITIN expression under different POSTN treatment durations. (N and O) Western blotting and quantitative analysis of collagen II, MMP3, ACSL4, PTGS2, GPX4, and FERRITIN in various experimental groups. POSTN, periostin; rPOSTN, recombinant POSTN; MMP3, matrix metalloproteinase-3; ACSL4, acyl-CoA synthetase long chain family member 4; PTGS2, prostaglandin-endoperoxide synthase 2; GPX4, glutathione peroxidase 4; LVNC, lentiviral vector–negative control; LV-POSTN, lentiviral vector–periostin. Data are presented as mean±standard deviation from at least 3 independent experiments. *p<0.05, **p<0.01, ***p<0.001.

3. POSTN Knockout Alleviates Ferroptosis and ECM Metabolic Imbalance in NPCs

A POSTN knockout NPCs model was used in subsequent studies. Western blotting showed that in the Erastin+POSTN^-/-group, MMP3, ACSL4, and PTGS2 levels were decreased, while collagen II, GPX4, and FERRITIN were increased. These alleviating effects were reversed upon re-addition of rPOSTN to the Erastin+POSTN^-/- group (Fig. 3A and B). TEM revealed that the Erastin+POSTN^-/- group mitigated ferroptosis-induced alterations in the mitochondrial structure (Fig. 3C). The IF results were consistent with those of Western blotting (Fig. 3D–F). Fluorescence assays for lipid peroxidation and Fe²⁺ also confirmed these results (Fig. 3G–J).

Fig. 3.

POSTN knockout alleviates ferroptosis and extracellular matrix metabolic imbalance in nucleus pulposus cells (NPCs). (A and B) Western blotting and quantitative analysis of collagen II, MMP3, ACSL4, PTGS2, GPX4, and FERRITIN expression in different experimental groups. (C) Representative transmission electron microscopy images of mitochondria in NPCs from different groups. (D–F) Immunofluorescence staining and quantitative analysis of ACSL4 and GPX4 expression in different experimental groups. (G and H) Lipid peroxidation levels and corresponding statistical analysis in different experimental groups. (I and J) Fe²⁺ levels and corresponding statistical analysis in different experimental groups. POSTN, periostin; rPOSTN, recombinant POSTN; MMP3, matrix metalloproteinase-3; ACSL4, acyl-CoA synthetase long chain family member 4; PTGS2, prostaglandin-endoperoxide synthase 2; GPX4, glutathione peroxidase 4; DAPI, 4´,6-diamidino-2-phenylindole. Data are presented as mean±standard deviation from at least 3 independent experiments. *p<0.05, **p<0.01, ***p<0.001.

4. POSTN Induces Ferroptosis and ECM Metabolic Imbalance in NPCs by Activating the IL6/STAT3 Pathway

The STAT3 inhibitor Stattic was used to examine changes in ferroptosis levels and ECM-related marker proteins following IL6/STAT3 pathway inhibition in NPCs. Western blotting indicated that POSTN induced IL6 and p-STAT3 expression. This induction was reduced upon the addition of Stattic (Fig. 4A and B). The IF results were consistent with this finding (Fig. 4C and D). Fluorescent assays for lipid peroxidation and Fe²⁺ levels further confirmed this observation (Fig. 4E–H). Regarding ferroptosis and ECM metabolic markers, IF showed that the inhibition of IL6/STAT3 signaling decreased ACSL4 and increased GPX4 levels (Fig. 4I–K). Western blotting results corroborated these changes (Fig. 4L and M). In addition to pathway inhibition, we found that the addition of the agonist Colivelin increased the expression of IL6 and p-STAT3 (Fig. 4N and O). The IF for p-STAT3 aligned with the Western blotting results (Supplementary Fig. 2B and C). Western blotting for ferroptosis and ECM markers showed that activation of the IL6/STAT3 pathway increased MMP3, ACSL4, and PTGS2, while decreasing collagen II, GPX4, and FERRITIN (Fig. 4P; Supplementary Fig. 2A). IF of ACSL4 and GPX4, along with measurements of lipid peroxidation and Fe²⁺ levels, supported these results (Supplementary Fig. 2D–J). Furthermore, in vivo experiments using POSTN^-/- rats showed that POSTN knockout decreased p STAT3 expression, as detected using IF (Fig. 4Q).

Fig. 4.

POSTN induces ferroptosis and extracellular matrix metabolic imbalance in nucleus pulposus cells by activating the IL6/STAT3 pathway. (A and B) Western blotting and quantitative analysis of STAT3, phosphorylated STAT3 (p-STAT3), and IL6 expression in different experimental groups. (C and D) Immunofluorescence (IF) staining and quantitative analysis of p-STAT3 expression in different experimental groups. (E and F) Lipid peroxidation levels and corresponding statistical analysis in different experimental groups. (G and H) Fe²⁺ levels and corresponding statistical analysis in different experimental groups. (I–K) IF staining and quantitative analysis of ACSL4 and GPX4 expression in different experimental groups. (L and M) Western blotting and quantitative analysis of collagen II, MMP3, ACSL4, PTGS2, GPX4, and FERRITIN expression in different experimental groups. (N and O) Western blotting and quantitative analysis of STAT3, p-STAT3, and IL6 expression in different experimental groups. (P) Western blotting results of collagen II, MMP3, ACSL4, PTGS2, GPX4, and FERRITIN in different experimental groups. (Q) IF staining of p-STAT3 in rat intervertebral disc tissues from various groups. POSTN, periostin; rPOSTN, recombinant POSTN; IL6, interleukin 6; MMP3, matrix metalloproteinase-3; ACSL4, acyl-CoA synthetase long chain family member 4; PTGS2, prostaglandin-endoperoxide synthase 2; GPX4, glutathione peroxidase 4; DAPI, 4´,6-diamidino-2-phenylindole; IDD, intervertebral disc degeneration; ns, not significant. Data are presented as mean±standard deviation from at least 3 independent experiments. *p<0.05, **p<0.01, ***p<0.001.

5. STAT3 Transcriptionally Regulates POSTN Expression to Induce Ferroptosis in NPCs

Multiple databases were used to identify potential transcription factors of POSTN and genes associated with ferroptosis. Venn analysis revealed 12 potential transcription factors (Fig. 5A) that regulate POSTN expression and are linked to ferroptosis. Subsequent integration of JASPAR and UCSC databases identified STAT3 as a candidate transcription factor for POSTN (Fig. 5B). The structure of STAT3 is shown in Fig. 5D. Correlation analysis of the GEPIA database revealed a significant positive correlation between STAT3 and POSTN expression (R=0.61, p<0.05) (Fig. 5C). Based on this analysis, we proposed STAT3 as a candidate transcription factor of POSTN. Bioinformatic analysis predicted the sequence logo of the STAT3-binding motif within the POSTN promoter sequence (Fig. 5E). The JASPAR database identified 3 potential STAT3 BSs (Fig. 5F). ChIP-PCR showed that STAT3 bound to all 3 sites (Fig. 5G and H), with the highest binding activity at BS3. A dual-luciferase reporter assay targeting BS3 showed that mutation of BS3 significantly reduced STAT3-induced luciferase activity (Fig. 5I). These results indicate that STAT3 binds to the POSTN promoter sequence and activates its transcription.

Fig. 5.

STAT3 transcriptionally regulates POSTN expression to induce ferroptosis in nucleus pulposus cells. (A) Venn diagram analysis of hTFtarget, HumanTFDB, HOCOMOCO, and DisGeNet databases. (B) Prediction of transcription factors for POSTN using JASPAR and UCSC databases. (C) Correlation analysis of STAT3 and POSTN expression via the GEPIA database. (D) Structure of STAT3. (E) Sequence logo of the STAT3 binding motif in the POSTN promoter sequence predicted by the JASPAR database. (F) Putative STAT3 binding sites in the POSTN promoter predicted by the JASPAR database. (G and H) ChIP-PCR assay and statistical analysis. (I) Statistical analysis of the dual-luciferase reporter gene assay. (J) H&E and S-O staining of IVD from rats in different groups. (K) Immunofluorescence (IF) staining of POSTN in rat IVD from different groups. (L) H&E and S-O staining of IVD from rats in different groups. (M and N) IF staining of ACSL4 and GPX4 in rat IVD from different groups. (O and P) Western blotting and quantitative analysis of ACSL4, PTGS2, GPX4, and FERRITIN in different experimental groups. (Q and R) Lipid peroxidation levels and statistical analysis in different experimental groups. (S and T) Fe²⁺ levels and statistical analysis in different experimental groups. ChIP, chromatin immunoprecipitation; PCR, polymerase chain reaction; IgG, immunoglobulin G; H&E, hematoxylin and eosin; S-O, Safranin O/Fast Green; IVD, intervertebral disc; POSTN, periostin; ACSL4, acyl-CoA synthetase long chain family member 4; PTGS2, prostaglandin-endoperoxide synthase 2; GPX4, glutathione peroxidase 4; IDD, intervertebral disc degeneration; LV-NC, lentiviral vector–negative control; LV-shSTAT3, lentiviral vector–short hairpin STAT3; LV-STAT3, lentiviral vector–STAT3. Data are presented as mean±standard deviation from at least 3 independent experiments. *p<0.05, **p<0.01, ***p<0.001.

In vivo IF experiments showed that STAT3 knockdown decreased POSTN fluorescence intensity (Fig. 5J and K). STAT3 overexpression exacerbated IDD and promoted ferroptosis progression; however, knocking out POSTN expression in this context alleviated these effects (Fig. 5L–N). In vitro, Western blotting, IF, and assays for lipid peroxidation and Fe²⁺ indicated that STAT3 regulates POSTN expression to induce ferroptosis in NPCs (Fig. 5O–T; Supplementary Fig. 3A–C). In vitro experiments involving STAT3 knockdown combined with POSTN overexpression yielded results that supported the above conclusions (Supplementary Fig. 3D–L). Furthermore, we found that STAT3 expression was upregulated during ferroptosis, and increasing STAT3 expression elevated ferroptosis levels (Supplementary Fig. 4A–L).

In summary, our study demonstrates that STAT3, a candidate transcription factor, binds to the POSTN promoter and regulates its expression, thereby promoting ferroptosis in NPCs. Based on our previous findings, we propose that STAT3 and POSTN form a self-amplifying positive feedback loop that mutually reinforces each other to drive the initiation and progression of ferroptosis in NPCs, thereby accelerating IDD.

6. POSTN Impairs the GSTP1/JNK Complex to Regulate Ferroptosis and GSH Levels in NPCs

NP specimens from WT and POSTN^-/- rats underwent proteomic sequencing to identify POSTN downstream proteins (Fig. 6A and B). Venn analysis of DEPs from proteomic sequencing and prior transcriptomic sequencing indicated that GSTP1 was most likely the downstream regulatory protein of POSTN (Fig. 6C). GO and KEGG enrichment analyses showed that GSTP1 was primarily enriched in processes such as glutathione metabolism, iron ion homeostasis, and ferrous ion binding (Fig. 6D and E). GSEA results supported these findings (Fig. 6F and G).

Fig. 6.

POSTN impairs the GSTP1/JNK complex to regulate ferroptosis and glutathione (GSH) levels in nucleus pulposus cells. (A and B) Volcano plot and heat map of differentially expressed genes (DEPs) from proteomic sequencing. (C) Venn diagram analysis of transcriptomic sequencing and proteomic sequencing data. (D and E) GO and KEGG enrichment analysis of DEPs. (F and G) Gene set enrichment analysis. (H and I) Western blotting and corresponding quantitative analysis of ACSL4, PTGS2, GPX4, and FERRITIN in different experimental groups. (J) Fe²⁺ levels in different experimental groups. (K and L) GSH levels assessed by MCB and N-9 in different experimental groups. (M) Coimmunoprecipitation with anti-GSTP1 and Western blotting analysis of p-JNK and GSTP1 in different experimental groups. (N) Representative immunofluorescence images and statistical analysis of JNK and GSTP1 protein expression. GSTP1, glutathione S-transferase Pi 1; JNK, c-Jun N-terminal kinase; p-JNK, phospho-JNK; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; MCB, molecular and cellular biology; N-9, nonoxynol-9; POSTN, periostin; ACSL4, acyl-CoA synthetase long chain family member 4; PTGS2, prostaglandin-endoperoxide synthase 2; GPX4, glutathione peroxidase 4; LV-NC, lentiviral vector–negative control; LV-POSTN, lentiviral vector–periostin; LV-GSTP1, lentiviral vector–GSTP1; DAPI, 4´,6-diamidino-2-phenylindole. Data are presented as mean±standard deviation from at least 3 independent experiments. *p<0.05, **p<0.01, ***p<0.001.

GSTP1 catalyzes the conjugation of GSH to various electrophilic and hydrophobic substances, facilitating their excretion as water-soluble compounds and reducing intracellular macromolecular damage. JNK reportedly interacts with GSTP1 to form a complex that is involved in disease progression, and its activation is implicated in reactive oxygen species (ROS) accumulation during ferroptosis. Western blotting revealed that POSTN overexpression significantly decreased GSTP1 and increased p-JNK levels (Supplementary Fig. 5G and H). POSTN overexpression also suppressed GSTP1 enzymatic activity (Supplementary Fig. 5I). Compared with LV-POSTN, LV-POSTN+ LV-GSTP1 exhibited reduced ferroptosis levels (Fig. 6H and I). Fe²⁺ level assays and GSH detection methods confirmed these results (Fig. 6J–L; Supplementary Fig. 5A and B). CoIP was used to assess the effect of POSTN on the GSTP1/JNK complex, which indicated a stable GSTP1/JNK complex in the LV-NC group. However, POSTN expression impaired this interaction (Fig. 6M). IF confirmed this finding (Fig. 6N). Therefore, we conclude that POSTN impairs GSTP1/JNK complex association. Additionally, we evaluated GSTP1 expression in IDD and its protective role against ferroptosis (Supplementary Fig. 5C–F; Supplementary Fig. 5J and K).

7. PN May Enhance GSTP1 Expression and Promote the GSTP1/JNK Complex to Alleviate Ferroptosis and GSH Depletion

To enhance the translational potential of this study, a natural compound library was screened to identify molecules that could modulate GSTP1 with strong binding affinity. Based on molecular mechanics generalized born surface area, 3 natural compounds with strong affinities for GSTP1 were identified: PN, hypericin, and verticinone (Fig. 7A and B). Molecular docking and dynamic simulations were used to assess the stability of this interaction (Fig. 7C–I; Supplementary Fig. 5L–O). PN was selected as the potential natural compound targeting GSTP1 with a stable interaction, exhibiting an molecular mechanics poisson– boltzmann surface area value of -25.71 kcal/mol. Cellular thermal shift assay results support the interaction hypothesis (Fig. 7J and K).

Fig. 7.

Pristimerin (PN) may enhance GSTP1 expression and promote the GSTP1/JNK complex to alleviate ferroptosis and glutathione (GSH) depletion. (A) Workflow for the virtual screening of natural small-molecule compounds. (B) Top 20 compounds ranked by MM-GBSA calculations. (C–F) Molecular dynamics simulations of pristimerin, hypericin, and verticinone: protein backbone RMSD, ligand RMSD relative to the protein backbone, RMSF and Rg. (G–I) Molecular docking of pristimerin, hypericin, and verticinone. (J and K) Cellular thermal shift assay and quantitative analysis of the interaction between pristimerin and GSTP1. (L) Representative transmission electron microscopy images of different experimental groups. (M and N) Western blotting and quantitative analysis of ACSL4, PTGS2, GPX4, and FERRITIN in different experimental groups. (O and P) GSH levels assessed by MCB and N-9 probes and corresponding statistical analysis in different experimental groups. (Q and R) Fe²⁺ levels and corresponding statistical analysis in different experimental groups. (S) Coimmunoprecipitation using an anti-GSTP1 and Western blotting analysis of p-JNK and GSTP1 in different experimental groups. (T) Representative immunofluorescence (IF) images and statistical analysis of JNK and GSTP1 protein expression. (U) Magnetic resonance imaging images of rat intervertebral discs (IVDs) and H&E and S-O staining of nucleus pulposu specimens from different groups. (V–X) IF staining of GSTP1, JNK, ACSL4, and GPX4 in rat IVDs from different groups. GSTP1, glutathione S-transferase Pi 1; JNK, c-Jun N-terminal Kinase; MM-GBSA, molecular mechanics generalized born surface area; RMSD, root mean square deviation; RMSF, root mean square fluctuation; Rg, radius of gyration; p-JNK, phospho-JNK; ACSL4, acyl-CoA synthetase long chain family member 4; PTGS2, prostaglandin-endoperoxide synthase 2; GPX4, glutathione peroxidase 4; MCB, molecular and cellular biology; N-9, nonoxynol-9; H&E, hematoxylin and eosin; S-O, Safranin O/Fast Green; IgG, immunoglobulin G; DMSO, dimethyl sulfoxide; IP, immunoprecipitation; PCC, Pearson correlation coefficient; DAPI, 4´,6-diamidino-2-phenylindole; IDD, intervertebral disc degeneration. Data are presented as mean±standard deviation from at least 3 independent experiments. *p<0.05, **p<0.01, ***p<0.001.

TEM showed that PN restored mitochondrial structure (Fig. 7L). Western blotting analysis indicated that PN reduced ferroptosis in NPCs (Fig. 7M and N). Fe²⁺ level assays and fluorescent GSH detection confirmed these results (Fig. 7O–R). GST activity assay showed PN may enhance GSTP1 function (Supplementary Fig. 5P). CoIP and IF were used to evaluate the effect of PN on the GSTP1/JNK complex, which showed that PN promoted their association (Fig. 7S and T). In vivo experiments were performed to evaluate the therapeutic potential of the PN. MRI showed increased water content in the PN-treated groups (5 and 10 mg/kg). Histological staining revealed reduced structural disruption and fibrosis along with increased IVD height and ECM components after PN treatment (Fig. 7U). IF revealed decreased GSTP1 and increased JNK levels in IDD, which were reversed by PN, accompanied by reduced ferroptosis (Fig. 7V–X). Furthermore, representative HE staining of major organs (heart, liver, spleen, lung, and kidney) from PN‑treated rats revealed no obvious pathological alterations at 10 mg/kg (Supplementary Fig. 5Q).

DISCUSSION

Ferroptosis is an iron- and ROS-dependent regulated cell death characterized by Fe²⁺ accumulation, lipid peroxidation, and GSH depletion [25,26]. It also induces mitochondrial morphological alterations such as shrinkage and reduced cristae [11]. GO and KEGG analysis of single-cell RNA sequencing data from human NPCs by Zhang et al. [27] revealed significant enrichment of the ferroptosis pathway. Wu et al. [28] have found that scavenging ROS and chelating excess Fe²⁺ protected NPCs from ferroptosis and ECM degradation. Our previous research demonstrated an important role for POSTN in IDD, where it promotes NPCs inflammation, apoptosis, and ECM degradation by activating Wnt/β-catenin and NF-κB signaling [17,18,29]. Therefore, this study investigated how POSTN contributes to IDD by regulating ferroptosis and its underlying mechanisms. This study provides novel evidence that STAT3 binds to the POSTN promoter to regulate transcription. POSTN and STAT3 form a functional positive feedback loop in which they mutually reinforce each other’s expression, thereby accelerating the progression of IDD. POSTN downregulates GSTP1 and disrupts the GSTP1/JNK complex, thereby promoting ferroptosis. Our study suggests that targeting the STAT3/POSTN/GSTP1/JNK axis may represent a promising therapeutic strategy for IDD.

This study provides novel evidence that POSTN plays an important role in inducing ferroptosis in NPCs by increasing lipid peroxidation and Fe²⁺ accumulation. In vitro experiments have shown that POSTN induces ferroptosis in NPCs by mediating the IL6/STAT3 signalling pathway. This role of POSTN in mediating the IL6/STAT3 pathway aligns with previous reports by Chen et al. [30] and Wang et al. [31] Furthermore, in vivo experiments revealed that POSTN knockout significantly reduced p-STAT3 activity, suppressed ferroptosis, and delayed IDD progression. These findings provide novel insights into the role of POSTN in IDD and suggest POSTN as a potential molecular target for therapeutic intervention.

STAT3 is involved in diverse biological processes, including cell proliferation, survival, differentiation, and angiogenesis [32,33]. Transient activation transduces transcriptional signals from cytokines and growth factor receptors in the plasma membrane to the nucleus. Our study provides the novel evidence that STAT3 acts as an upstream transcription factor of POSTN. It should be noted that this STAT3–POSTN relationship constitutes a functional rather than a direct physical interaction: STAT3 transcriptionally activates POSTN expression, and POSTN in turn promotes STAT3 phosphorylation (p‑STAT3) through the IL6/STAT3 signaling cascade, thereby forming a self‑amplifying loop. The changes described herein predominantly reflect alterations in p‑STAT3 (the active form). This finding aligns with those of Nakamura et al. [34], who showed that microgravity-mediated STAT3 signaling induces POSTN expression in chondrocytes. Similarly, O’Meara et al. [35] identified a crucial role of STAT3/POSTN signalling in mammalian heart regeneration. However, these studies did not validate STAT3’s transcriptional regulation of POSTN. We also found that STAT3 expression is upregulated in IDD and influences ferroptosis. Ouyang et al. [36] have reported that STAT3 binds to and regulates consensus DNA response elements in the promoters of negative ferroptosis regulators, establishing a STAT3-ferroptosis axis. These reports support a role of STAT3 in IDD and ferroptosis and highlight its potential as a molecular target for the prevention and mitigation of IDD.

Furthermore, using wild-type and POSTN knockout rats, we identified GSTP1 as a downstream protein of POSTN through proteomic sequencing and Venn analysis. It has been reported that GSTP1 binds to JNK via its C-terminal domain to form a dimeric complex in vivo [37,38]. Under normal conditions, GSTP1 is highly expressed and catalyzes the conjugation of GSH to various electrophilic and hydrophobic substances, forming water-soluble compounds that are excreted, thereby reducing damage to intracellular macromolecules. However, under cellular stress, GSTP1 expression is downregulated while JNK expression increases, leading to dissociation of the GSTP1/JNK complex. In this study, POSTN overexpression resulted in decreased GSTP1, increased JNK expression, elevated Fe²⁺ levels, and GSH depletion, indicating that POSTN significantly disrupts the GSTP1/JNK interaction. This finding is consistent with several recent reports [39-41] that have elucidated the regulatory roles of the GSTP1/JNK complex in various diseases. Additionally, GSTP1 expression was found to be downregulated in IDD-associated ferroptosis. Overexpression of GSTP1 upregulated GPX4 expression, which aligns with reports on the regulatory relationship between GSTP1 and GPX4. Tian et al. showed that GSTP1 overexpression reversed DEX-induced GPX4 downregulation [42]. Li et al. [43] similarly demonstrated that adenoviral overexpression of GSTP1 reversed Ang II-induced GPX4 reduction in mice, indicating that GSTP1 can regulate GPX4 activity. Moreover, it has been reported that the GSTP1 inhibitor Eza attenuates the protective effects of Galangin against ferroptosis, particularly regarding GPX4 expression [40]. These studies collectively suggest that GSTP1 acts as an upstream regulator of GPX4, modulating its activity to mediate changes in ferroptosis levels. However, Zhang et al. [44] reported that GSTP1 exerts its antiferroptotic effect independently of GPX4. They found that GSTP1 acts through GSH conjugation of 4-HNE and selenium-independent GPX activity toward lipid peroxides, with 4-HNE serving as a substrate for GSTP1. This finding presents a notable discrepancy compared with previous reports. Therefore, whether GSTP1 functions independently of the canonical GPX4-dependent mechanism to counteract ferroptosis remains controversial, and no further studies have been reported. Hence, GSTP1 may represent an important molecular target for ferroptosis-based therapies, and its precise molecular mechanism in ferroptosis and IDD warrants further investigation.

Targeting key genes in the PCD pathway effectively alleviates IVD cell death and slows down cell degeneration [10]. Thus, by targeting the STAT3/POSTN/GSTP1/JNK axis, we identified the natural compound PN, which binds to GSTP1 and enhances its activity. The results showed that PN may enhance GSTP1 expression, promote its binding to JNK, and effectively alleviate ferroptosis in NPCs. This discovery suggests potential avenues for therapeutic interventions that target molecules in IDD.

Beyond the mechanistic findings, our results may carry translational potential. First, POSTN was markedly upregulated in degenerate NP tissue and its expression correlated positively with degeneration severity, suggesting its possible utility as a serum or imaging biomarker for IDD assessment—although this prospect requires validation in larger, multicenter cohorts with matched serum samples. Second, from a therapeutic perspective, the present study provides preclinical proof of concept that either genetic deletion of POSTN or pharmacological restoration of GSTP1 activity by PN can suppress ferroptosis in NPCs and attenuate ECM degradation in vivo. However, several obstacles must be overcome before clinical translation. Drug delivery to the avascular NP remains a major challenge. Although intraperitoneal administration of PN elicited therapeutic effects in our rat model, the efficiency with which PN penetrates the annulus fibrosus after systemic delivery is unclear. Intradiscal injection may circumvent this barrier, yet its invasiveness and potential iatrogenic disc damage warrant caution. The development of minimally invasive, sustained release delivery systems—such as thermosensitive injectable hydrogels or nanocarriers—may therefore represent an important step toward clinical application of PN. In addition, the dose–toxicity relationship of PN has not been established: the 2 in vivo doses tested here (5 and 10 mg/kg) were chosen empirically, and no systematic toxicity assessments were performed. Future studies should include pharmacokinetic/toxicokinetic profiling and safety pharmacology evaluations to define the therapeutic window.

An inevitable limitation of this study is that rats differ significantly from humans in their IVD functions and biomechanics. Although various animal models have been developed to study IDD, none can fully replicate the biomechanical environment of the human IVD. Second, trauma to the IVD and certain procedural factors during needle puncture-induced IDD modelling may affect the accuracy of the experimental outcomes. With regard to the in vitro experiments, the in vitro mechanistic experiments in this study were primarily conducted using an immortalized NPC line; subsequent translational studies will further incorporate primary human NPCs and NP organoid models to determine the generalizability of this regulatory axis. At the clinical level, the human NP specimens were obtained from a cohort of 24 cases, and the comparison was confined to mild (Pfirrmann grade II) and severe (grade IV) degeneration, with no grade III samples included. This design limits the analysis of dynamic molecular changes across the full spectrum of IDD progression. Although the baseline characteristics were well balanced between groups, the relatively small sample size may still affect statistical power and generalizability. Future multicenter studies with larger cohorts encompassing all Pfirrmann grades are warranted to further validate and extend these findings. From a genetic model perspective, a global POSTN knockout rat model was used for in vivo functional validation, rather than an NPCspecific conditional knockout model, which represents a limitation. Global knockout may indirectly influence the degenerative phenotype through effects on tissues outside the IVD, making it difficult to completely exclude non–cell‑autonomous contributions of NPCs. However, the core conclusions of this study are supported by complementary evidence from NPC‑specific POSTN knockdown/overexpression functional experiments in vitro and in vivo rescue experiments, supporting an intrinsic role of POSTN in NPCs. In future studies, we will further construct NPC‑specific POSTN conditional knockout mouse models to exclude nonautonomous effects and more precisely validate the specificity of this regulatory axis. Finally, regarding the pharmacological aspect, this study has supported a possible interaction between PN and GSTP1 and its therapeutic effects through multilevel validation; however, the precise BSs have not yet been resolved. In subsequent dedicated drug‑optimization studies, techniques such as binding‑residue mutagenesis and isothermal titration calorimetry will be employed to delineate the binding mode at the atomic level, providing a basis for structural refinement and clinical translation of this compound.

CONCLUSION

Our study identified the STAT3/POSTN/GSTP1/JNK axis as a novel ferroptosis‑regulatory pathway in IDD. We confirmed a positive feedback loop between STAT3 and POSTN that contributes to ferroptosis in NPCs and IDD progression. This study also identified the small natural compound, PN, as a potential therapeutic agent targeting this axis. These findings offer novel insights into the pathogenesis of IDD and suggest potential therapeutic strategies.

Supplementary Materials

Supplementary Figs. 1-5 are available at https://doi.org/10.14245/ns.26520334.0167.

Supplementary Fig. 1.

POSTN expression is upregulated during ferroptosis and induces ferroptosis and extracellular matrix metabolic imbalance in nucleus pulposus cells in a time-dependent manner. (A and B) Western blotting and quantitative analysis of collagen II, MMP3, ACSL4, PTGS2, GPX4, and FERRITIN expression in different experimental groups. (C and D) Western blotting and quantitative analysis of POSTN expression in different experimental groups. (E and F) IF staining and quantitative analysis of POSTN expression in different experimental groups. (G–I) Immunofluorescence staining and quantitative analysis of ACSL4 and GPX4 expression in different experimental groups. (J and K) Lipid peroxidation levels and corresponding statistical analysis under different POSTN treatment durations. (L) MDA levels under different POSTN treatment durations. (M and N) Fe²⁺ levels and corresponding statistical analysis under different POSTN treatment durations. POSTN, periostin; rPOSTN, recombinant POSTN; MMP3, matrix metalloproteinase-3; ACSL4, acyl-CoA synthetase long chain family member 4; PTGS2, prostaglandin-endoperoxide synthase 2; GPX4, glutathione peroxidase 4; DAPI, 4´,6-diamidino-2-phenylindole. Data are presented as mean±standard deviation from at least 3 independent experiments. *p<0.05, **p<0.01, ***p<0.001.

Supplementary Fig. 2.

POSTN induces ferroptosis and extracellular matrix metabolic imbalance in nucleus pulposus cells by activating the IL6/STAT3 signaling pathway. (A) Quantitative analysis of Western blotting for collagen II, MMP3, ACSL4, PTGS2, GPX4, and FERRITIN in different groups. (B and C) Immunofluorescence (IF) staining and quantitative analysis of p-STAT3 expression in different groups. (D–F) IF staining and quantitative analysis of ACSL4 and GPX4 expression in different groups. (G and H) Lipid peroxidation levels and corresponding statistical analysis in different groups. (I and J) Fe²⁺ levels and corresponding statistical analysis in different groups. POSTN, periostin; IL6, interleukin 6; MMP3, matrix metalloproteinase-3; ACSL4, acyl-CoA synthetase long chain family member 4; PTGS2, prostaglandin-endoperoxide synthase 2; GPX4, glutathione peroxidase 4; p-STAT3, phosphorylated STAT3; DAPI, 4´,6-diamidino-2-phenylindole. Data are presented as mean±standard deviation from at least 3 independent experiments. *p<0.05, **p<0.01, ***p<0.001.

Supplementary Fig. 3.

STAT3 transcriptionally regulates POSTN expression, thereby inducing ferroptosis in nucleus pulposus cells. (A–C) Immunofluorescence (IF) staining and quantitative analysis of ACSL4 and GPX4 expression in different groups. (D and E) Western blotting and quantitative analysis of ACSL4, PTGS2, GPX4, and FERRITIN in different groups. (F–H) IF staining and quantitative analysis of ACSL4 and GPX4 expression in different groups. (I and J) Lipid peroxidation levels and corresponding statistical analysis in different groups. (K and L) Fe²⁺ levels and corresponding statistical analysis in different groups. POSTN, periostin; rPOSTN, recombinant POSTN; LV-NC, lentiviral vector–negative control; LV-shSTAT3, lentiviral vector–short hairpin STAT3; LV-STAT3, lentiviral vector–STAT3; ACSL4, acyl-CoA synthetase long chain family member 4; PTGS2, prostaglandin-endoperoxide synthase 2; GPX4, glutathione peroxidase 4; DAPI, 4´,6-diamidino-2-phenylindole. Data are presented as mean±standard deviation from at least 3 independent experiments. *p<0.05, **p<0.01, ***p<0.001.

Supplementary Fig. 4.

STAT3 expression in ferroptosis during intervertebral disc degeneration. (A) H&E and S-O staining of nucleus pulposus (NP) specimens from patients with different Pfirrmann grades. (B) Immunofluorescence (IF) staining of STAT3 in NP specimens from patients with different Pfirrmann grades. (C and D) Immunohistochemistry staining and quantitative analysis of STAT3 in NP specimens from patients with different Pfirrmann grades. (E and F) Western blotting and quantitative analysis of STAT3 in NP specimens from patients with different Pfirrmann grades. (G) Representative transmission electron microscopy images of mitochondria in nucleus pulposus cells from different groups. (H and I) Western blotting and quantitative analysis of STAT3 and p-STAT3 in different groups. (J and K) Western blotting and quantitative analysis of ACSL4, PTGS2, GPX4, and FERRITIN in different experimental groups. (L) IF staining of ACSL4 and GPX4 in rat intervertebral discs from different groups. H&E, hematoxylin and eosin; S-O, Safranin O/Fast Green; LV-NC, lentiviral vector–negative control; LVshSTAT3, lentiviral vector–short hairpin STAT3; LV-STAT3, lentiviral vector–STAT3; ACSL4, acyl-CoA synthetase long chain family member 4; PTGS2, prostaglandin-endoperoxide synthase 2; GPX4, glutathione peroxidase 4; p-STAT3, phosphorylated STAT3; DAPI, 4´,6-diamidino-2-phenylindole. Data are presented as mean±standard deviation from at least 3 independent experiments. **p<0.01, ***p<0.001.

Supplementary Fig. 5.

Role of GSTP1 in regulating nucleus pulposus cells ferroptosis and molecular dynamics simulations. (A) Statistical analysis of Fe²⁺ levels in different groups. (B) Statistical analysis of glutathione levels assessed by MCB and N-9 probes in different groups. (C and D) Western blotting and quantitative analysis of GSTP1 in nucleus pulposus (NP) specimens from patients with different Pfirrmann grades. (E and F) Immunohistochemistry staining and quantitative analysis of GSTP1 in NP specimens from patients with different Pfirrmann grades. (G and H) Western blotting and quantitative analysis of p-JNK, JNK, and GSTP1 in different groups. (I and P) Enzyme activity of GSTP1 and quantitative analysis in different groups. (J and K) Western blotting and quantitative analysis of GSTP1, ACSL4, PTGS2, GPX4, and FERRITIN in different groups. (L) Residue energy decomposition of pristimerin. (M and O) Gibbs energy landscape of pristimerin, hypericin, and verticinone. (Q) Representative H&E staining images of heart, liver, spleen, lung, and kidney from rats in the control and pristimerin (PN) (10 mg/kg) groups. GSTP1, glutathione S-transferase Pi 1; JNK, c-Jun N-terminal Kinase; p-JNK, phospho-JNK; ACSL4, acyl-CoA synthetase long chain family member 4; PTGS2, prostaglandin-endoperoxide synthase 2; GPX4, glutathione peroxidase 4; MCB, molecular and cellular biology; N-9, nonoxynol-9; H&E, hematoxylin and eosin; DAPI, 4´,6-diamidino-2-phenylindole; LV-NC, lentiviral vector–negative control; POSTN, periostin; LV-POSTN, lentiviral vector–periostin; LV-GSTP1, lentiviral vector–GSTP1; ns, not significant. Data are presented as mean±standard deviation from at least 3 independent experiments. *p<0.05, **p<0.01, ***p<0.001.