As the most prevalent degenerative arthritis, OA affects multiple joints such as knees, lumbar, hips, facet, and temporomandibular and becomes irreversible when it develops to a certain extent (1). Currently, non-pharmacologic and pharmacologic therapies are employed to treat OA. Among the pharmacological treatments, analgesics and NSAIDs are the current options for treating OA, and their effectiveness has been proven (2). However, these drugs can be used as symptomatic treatment but cannot change the nature of the disease. Therefore, it is necessary to explore the occurrence and progression mechanism of OA and identify targeted therapeutic targets that are beneficial to the early prevention and treatment of OA.

OA is a complex pathological process, and one of its most prominent features is the degradation of articular cartilage (3). As chondrocytes are the only biological constituents of articular cartilage, researchers have revealed that the occurrence of ferroptosis in these cells is closely associated with the progression of OA (4). As a new type of programmed cell death, the conception of “ferroptosis” was initially given by Stockwell et al. in 2012 and has been extensively researched in recent years (5). Excessive production and accumulation of lipid peroxides and reactive oxygen species (ROS) are the typical characteristics of ferroptosis, and its essence is the unbalance between the generation and degradation of intracellular lipid ROS (6). When cellular cysteine (e.g., Erastin) transport proteins are inhibited, intracellular glutathione (GSH) is depleted, eventually resulting in the inactivation of glutathione peroxidase (GPX4) and the accumulation of lipid peroxidation, which can induce cellular ferroptosis after reaching a certain level. Inhibiting GPX4 enzymes (e.g., RSL3) can also directly accelerate this effect, and iron ion chelators can inhibit this process (7). These findings suggest that inhibition of ferroptosis is a new direction to prevent the progression of OA, and exploring the immune infiltration in OA and its relationship with ferroptosis will help analyze the pathogenesis of ferroptosis in OA.

Accumulation of lipid peroxidation is a dangerous factor for the progression of OA, and dysregulation of iron homeostasis mediated by pro-inflammatory cytokines plays a critical part in iron overload-induced OA progression (4, 8). Pro-inflammatory cytokines, including interleukin-1β (IL-1β) and tumor necrosis factor-a (INF-a), are the main components of inflammation and contribute to cartilage damage in OA. Ding et al. revealed that IL-1β or INF-a could lead to iron overload by increasing iron influx and decreasing iron outflow (9). The PI3K/AKT/CONFAB pathway associated with IL-1β and mediated by enzyme-linked receptors can control cellular life processes. It can be activated with the assistance of various growth factors, cytokines, extracellular matrix, etc., and affect the NF-κB pathway closely associated with inflammatory mediators in tissue inflammation, tumor development and invasion, cell proliferation and apoptosis. Li et al. reported that melatonin could prevent steroid-induced osteoporosis by preventing ferroptosis via triggering the PI3K/AKT/mTOR signaling pathway (10). Chen et al. proposed that curcumin triggers ferroptosis by inhibiting the PI3K/Akt/mTOR signaling pathway, thereby inhibiting the proliferation of colorectal cancer cells (11). So far, ferroptosis-related pathways have been related to many degenerative diseases, carcinogenesis, ischemia-reperfusion injury, and stroke (11). However, the mechanism of ferroptosis in OA is still unclear, and as far as we know, there are still no relevant studies exploring FRGs and their regulatory mechanisms in OA progression.

From a comprehensive perspective, exploring ferroptosis’s clinical significance and underlying molecular mechanisms in OA is important. This study collected samples from OA patients and healthy individuals from the GEO database. Subsequently, we used bioinformatics methods to screen OA differentially expressed genes (DEGs) and feature genes linked to ferroptosis and carried out immune infiltration analysis to analyze their correlations, providing new strategies and directions for treating OA. Finally, we found 149 DEGs, with 4 feature genes identified as potential biomarkers of ferroptosis in OA.

Ferroptosis is a new horizon for current research to explore new biomarkers and therapeutic targets, but the research in OA is still incomplete. In this study, we found 4 feature genes related to ferroptosis in OA; they inspired our understanding of the relationship between ferroptosis and the OA mechanism. Also, we looked into new treatment options using immune infiltration analysis, which could help find new OA biomarkers and therapeutic targets.

Although OA is an illness without a clear explanation of its occurrence and progression, several involved research has shown that the pathogenesis of OA is closely related to changes in some specific signaling pathways, abnormal mechanical load, and ageing (17). In this paper, 149 FRGs were significantly expressed in OA, and the GO functional enrichment analysis indicated that many of them were enriched in the biological process, including oxidative stress and fatty acid metabolism pathways. KEGG pathway enrichment analysis revealed that differentially expressed FRGs were mainly enriched in lipid and atherosclerosis, necroptosis, autophagy and PI3K/Akt pathway.

Several recent studies have revealed that the PI3K/Akt signal transduction pathway has a role in several conditions that affect the bones, including osteoporosis, OA, and osteosarcoma. Moreover, it significantly impacts the proliferation, differentiation, and apoptosis of osteoclasts and osteoblasts (18). Sun et al. found that the PI3K and Akt content decreased in OA (18), which may lead to a reduction in the PI3K/Akt pathway activity. Then reduces the activity of its downstream target mTOR, a key autophagy inhibitor, resulting in abnormal activation of cellular autophagy in OA and synergistic activation of massive chondrocyte apoptosis in the joint (19). At the same time, mitochondrial autophagy, a form of selective autophagy, engulfs large amounts of ferritin, leading to excessive Fe²⁺ release, increasing the intracellular pool of unstable iron and thus leading to chondrocyte ferroptosis (4). Several papers have demonstrated that the PI3K/Akt has a critical effect on the occurrence of OA in terms of maintaining cartilage homeostasis, regulating the differentiation and proliferation of osteoblasts in subchondral bone, and regulating synovial inflammation (18). Improving the activity of this pathway in OA may delay the development of the disease and enhance the expression of downstream genes and signal transduction to inhibit the ferroptosis of synovial fibroblasts (SFs) (20), which is also important for the study of ferroptosis. In this way, a novel potential therapy for OA is provided.

In addition, Basu et al. found that human OA cartilage has higher lipid peroxidation levels than normal cartilage (21); this may be due to reduced glutathione (GSH) levels or becoming more oxidized under conditions such as excessive mechanical loading of the joints and ageing, resulting in reduced resistance to oxidative stress. The abnormal reduction in intracellular GSH content eventually inactivates glutathione peroxidase (GPX4) and leads to the accumulation of lipid peroxidation, which induces cellular ferroptosis after reaching a certain level (22), which may shed some light on how differential FRGs may participate in ferroptosis-related OA.

The NOD-like receptors (NLR) family, widely present in the cytoplasm of human cells, can activate the NF-κB signaling pathway after recognizing the corresponding ligands, next activates caspase-1 to produce the pro-inflammatory factors like IL-1β and IL-18, thereby initiating the development of intrinsic and acquired immunity (23). Cytoplasmic DNA produced by cartilage damage in OA can activate Caspase-1, which binds to the inflammasomes generated by the activation of NLRP3, leading to a large generation of reactive oxygen species (ROS), and excessive accumulation can induce ferroptosis. The activation of NLRP3 inflammatory vesicles in the NOD-like signaling pathway and the production of inflammatory factors may positively influence the developing ferroptosis in OA. All of the pathways mentioned include the activation of the NF-κB pathway, suggesting its involvement in the process of ferroptosis in OA, but the exact mechanism needs to be further explored.

Through two machine-learning intersections, we identified 4 feature genes (BRD7, AQP8, IFNA4, and ARHGEF26-AS1), but the extent of their involvement in the development of ferroptosis in OA remains to be further investigated. We plotted ROC curves for each feature gene to validate internally, and all 4 genes’ AUC values were larger than or almost equal to 0.7, indicating that they have certain diagnostic significance. It suggests that these genes may be biological markers of ferroptosis in OA.

GSEA and GSVA analysis of the 4 feature genes revealed that they are enriched in the cell cycle, immune response, protein synthesis, and metabolic pathways. Several studies have shown that an important characteristic of OA is immune cell infiltration. We performed immune cell-related analyses to learn what role these immune cells play in OA. The content of each cell in the experimental and control groups of each data set was extracted and analyzed for differences. It has been shown that synovitis in OA is dominated by macrophage infiltration, reflecting the activation of the innate immune response (24). The predominance of activated NK cells, M0 macrophages, and dendritic cells activated in our results are consistent with this feature. Inflammatory transcription factors induced by autoantibody or protein are enhanced by ferroptosis-catalyzed oxidation, producing more matrix, cytokine/chemokine production and immune cells (25). Immune infiltration analysis outcomes also indicated a positive relationship between OA and CD8(+) T cells, activated NK cells, M0 macrophages, and dendritic cells activated. It has been shown that CD8 T cells have a biphasic role in OA, on the one hand controlling CD8(+) T cells dissolving T cells through the PI3K/Akt/mTORC1/HIF1 pathway; on the other hand, CD8(+) T cells in animal models kill DCs and abnormal T cells in a perforin-dependent manner (26), thus promoting inflammation. In addition, some studies have revealed that ferroptosis can coordinate CD8(+) T cells and thus enhance cellular immunity to achieve anti-tumor effects (27). Therefore, we hypothesize that increased CD8(+) T cells involved in ferroptosis may take part in the pathogenesis of OA. Besides, immune infiltration analysis suggests that CD4(+) T cells memory resting, activated NK cells, M1 macrophages, and dendritic cells resting cells are reduced, which may lead to decreased ferroptosis cell clearance in the osteoarthritic environment, ultimately exacerbating the development of OA.

One of the 4 feature genes, BRD7 is an oncogene widely expressed in various human tissues and has been extensively studied as a nasopharyngeal carcinoma-associated gene (28). It can interact with other transcription factors involved in transcriptional regulation, such as competing with p110 for interaction with p85α to attenuate PI3K activity (29). It links to the previously mentioned PI3K/Akt signaling pathway and explains the downregulation of PI3K activity in OA. Of interest, Zhang et al. found that BRD7 overexpression induced classical ferroptosis, BRD7 knockdown resisted ferroptosis events and that ferroptosis mediated by high expression of BRD7 may be related to its binding to the N-terminal inverse activation domain of p53 thereby promoting translocation expression of the p53 gene in mitochondria (30). Meanwhile, it has been demonstrated that low expression of BRD7 in the disease activates the primary immunodeficiency pathway. Thus, we speculate that overexpression of BRD7 may induce ferroptosis in OA by inhibiting the primary immunodeficiency pathway. However, further investigations need to be accomplished on the precise mechanism.AQP8, located on chromosome 16, region p12 (31), encodes a transmembrane water channel protein and hydrogen peroxide transporter protein implicated in various cell deaths. A study on pancreatitis suggested that the RIPK1/NF-κB/AQP8 axis may inhibit acinar cell necrosis (32). Over-expressed AQP8 in colorectal cancer (CRC) cells can inhibit the PI3K/Akt signaling transduction (31), which may have a similar effect in OA, and in concert with BRD7, leads to downregulation of PI3K activity in OA. We visualized the drug-gene interactions through cytospace to construct a drug regulatory network and eventually identified a possible drug that regulates AQP8, ethinyl estradiol, an estrogen component of almost all oral contraceptives (OCs) (33). Given the high incidence of OA and the significant changes in sex hormones in women after menopause around the age of 50, it has been found that the lack of estrogen decreases the expression of Akt (34), causing increased chondrocytes apoptosis and articular cartilage degeneration, thereby promoting the development of OA. Our search for this drug may then offer a possibility for the treatment of OA through estrogen regulation of AQP8 expression.

We also mapped the ceRNA network with the feature genes at its center. lncRNAs can compete with mRNAs to bind to miRNAs, thereby regulating the expression of 4 feature genes (14). Wang et al. explored targeting lncRNAs for therapy by identifying OA signature lncRNAs (35), suggesting that these miRNAs that bind directly to mRNAs and lncRNAs that regulate mRNA expression indirectly may provide new possible targets for the prevention or treatment in OA.

There are a few non-negligible drawbacks in this study. Above all, the data we chose were only retrieved from the GEO database, which is an insufficient sample size, and IFNA4 also lacks relevant external verification at present. The next step requires validation of each gene and the combined model constructed from the 4 genes in a larger dataset. In addition, although the biological functions of 4 feature genes were validated in cancer or other diseases, it is noteworthy that these disease models have different biological characteristics from OA, and their specific roles in OA need further investigation. What’s more, there is a lack of valid experimental evidence to confirm the therapeutic effect between the drugs found and OA. Finally, more research is needed regarding the specific association of lncRNAs and miRNAs with signature genes.

In summary, this research selected 4 differentially feature genes that can be potential biomarkers of ferroptosis in OA, including AQP8, BRD7, IFNA4, and ARHGEF26-AS1. The possible mechanism of four key genes in osteoarthritis was explored through analysis. The recently identified lncRNAs in the ceRNA network results and the newly discovered AQP8-targeting medication ETHINYL ESTRADIOL may provide fresh concepts for targeted therapy in OA.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by First Affiliated Hospital of Guangxi Medical University Ethical Review Committee. The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by First Affiliated Hospital of Guangxi Medical University Ethical Review Committee.

JY designed the study and revised the manuscript. YQ, LL and MX conducted data analysis and experiments. XH and JM wrote a draft manuscript. All authors contributed to the article and approved the submitted version.