Lipoxin A4 ameliorates knee osteoarthritis progression in rats by antagonizing ferroptosis through activation of the ESR2/LPAR3/Nrf2 axis in synovial fibroblast-like synoviocytes.

Zhehan Hu: Visualization, Validation, Supervision, Software, Resources, Project administration, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization, Methodology, Writing – original draft, Writing – review & editing. Liang Chen: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources. Jihui Zhao: Visualization, Supervision, Resources, Investigation, Data curation. Weiming Zhang: Visualization, Supervision, Project administration, Data curation. Zhuangzhuang Jin: Writing – original draft, Validation, Software, Methodology, Investigation, Data curation. Yuhan Sun: Visualization, Investigation, Conceptualization. Zihan Li: Software, Formal analysis, Conceptualization. Bohan Chang: Validation, Software, Resources, Project administration, Data curation. Peng Shen: Data curation, Formal analysis, Methodology, Resources, Supervision, Visualization, Writing – original draft. Yue Yang: Writing – review & editing, Writing – original draft, Data curation, Investigation, Project administration, Resources, Validation, Visualization.

To assess the potential roles and mechanisms of LPAR3 and ferroptosis in KOA, synovial samples were collected from patients with different KL grades. We initially collected clinical images and performed histological evaluation of the synovium in these patients ( Fig. 1 , Fig. 2 A). ESR2, LPAR3, and GPX4 (critical regulatory proteins for ferroptosis) were detected in the synovial membranes of patients with different KL grades via immunoblotting. We observed that ESR2, LPAR3, and GPX4 levels were downregulated in the synovial tissue with increasing KL grade ( Fig. 1 B and C). Immunohistochemistry (IHC) corroborated the above results and also showed the upregulation of HO-1 and CGRP levels with increasing KL grade ( Fig. 2 ). Fig. 2 Downregulation of ESR2, LPAR3, and GPX4 and upregulation of HO-1 and CGRP levels in synovial tissue with increasing Kellgren–Lawrence (KL) grades. (A) Immunohistochemical staining results for ESR2, LPAR3, GPX4, HO-1, and CGRP. (B–F) Statistical analysis of the immunohistochemical staining results. A comparison of all KL grades with grade 0 is shown (*P < 0.001, *aP = 0.004, *bP = 0.002, *cP = 0.039, *dP = 0.03, *eP = 0.009, ANOVA). Data are expressed as the mean ± 95% confidence interval; n = 3 patients per group. Fig. 2 Downregulation of ESR2, LPAR3, and GPX4 and upregulation of HO-1 and CGRP levels in synovial tissue with increasing Kellgren–Lawrence (KL) grades. (A) Immunohistochemical staining results for ESR2, LPAR3, GPX4, HO-1, and CGRP. (B–F) Statistical analysis of the immunohistochemical staining results. A comparison of all KL grades with grade 0 is shown (*P < 0.001, *aP = 0.004, *bP = 0.002, *cP = 0.039, *dP = 0.03, *eP = 0.009, ANOVA). Data are expressed as the mean ± 95% confidence interval; n = 3 patients per group. We obtained SD rat FLSs ( Fig. S5A ) and chondrocytes ( Fig. S5B ) and constructed a co-culture system of these cells ( Fig. 3 A). We then induced cellular inflammation using IL-1β to mimic the environment of KOA; using Cell Counting Kit-8 (CCK8) to detect cell viability to determine the optimal induced concentration of IL-1β ( Fig. S5C ). Elevated levels of LPAR3, ESR2, and GPX4 were observed in FLSs after treatment with 10 nM LXA 4 ( Fig. 3 B). Simultaneously, the levels of the destructive markers MMP13 and ADAM9 in chondrocytes under co-culture conditions decreased, whereas the levels of the protective marker collagen II increased. PHTPP treatment reversed this process ( Fig. 3 E). Fig. 3 LXA 4 alleviates chondrocyte cell matrix degradation and inflammation by elevating GPX4 levels in fibroblast-like synoviocytes (FLSs). (A) FLSs and chondrocytes were co-cultured using Transwell chambers. (B) Western blot of ESR2, LPAR3, SOD1, and GPX4 in FLSs. Significant differences were found between the normal and IL-1β groups (*P < 0.001; ANOVA), IL-1β and IL-1β+LXA 4 groups ( + P < 0.001; ANOVA), and the IL-1β+LXA 4 and IL-1β+LXA 4 +PHTPP groups ( # P < 0.001, # aP = 0.002; ANOVA). (C) Western blot of NF-κB P65 in the nucleus and cytoplasm of chondrocytes in a co-culture system. Significant differences were found between the normal and IL-1β groups (*P < 0.001; ANOVA), IL-1β and IL-1β+LXA 4 groups ( + P < 0.001; ANOVA), and the IL-1β+LXA 4 and IL-1β+LXA 4 +PHTPP groups ( # P < 0.001, # aP = 0.0023; ANOVA). (D) Effects of PHTPP and LXA 4 treatment on the nuclear translocation of NF-κB p65 in chondrocytes. Cells were immunostained with rabbit anti–NF–κB p65 antibodies (green) and visualized using confocal microscopy. Cytoskeletal structures were stained with phalloidin (red) and nuclei were stained with DAPI (blue). Scale bar = 20 μm. (E) Western blotting for collagen II, MMP13, and ADAM9 in chondrocytes. Significant differences were found between the normal and IL-1β groups (*P < 0.001; ANOVA), IL-1β and IL-1β+LXA 4 groups ( + P < 0.001; ANOVA), and the IL-1β+LXA 4 and IL-1β+LXA 4 +PHTPP groups ( # P < 0.001; ANOVA). Data are expressed as the mean ± 95% confidence interval; n = 3 per group. Fig. 3 LXA 4 alleviates chondrocyte cell matrix degradation and inflammation by elevating GPX4 levels in fibroblast-like synoviocytes (FLSs). (A) FLSs and chondrocytes were co-cultured using Transwell chambers. (B) Western blot of ESR2, LPAR3, SOD1, and GPX4 in FLSs. Significant differences were found between the normal and IL-1β groups (*P < 0.001; ANOVA), IL-1β and IL-1β+LXA 4 groups ( + P < 0.001; ANOVA), and the IL-1β+LXA 4 and IL-1β+LXA 4 +PHTPP groups ( # P < 0.001, # aP = 0.002; ANOVA). (C) Western blot of NF-κB P65 in the nucleus and cytoplasm of chondrocytes in a co-culture system. Significant differences were found between the normal and IL-1β groups (*P < 0.001; ANOVA), IL-1β and IL-1β+LXA 4 groups ( + P < 0.001; ANOVA), and the IL-1β+LXA 4 and IL-1β+LXA 4 +PHTPP groups ( # P < 0.001, # aP = 0.0023; ANOVA). (D) Effects of PHTPP and LXA 4 treatment on the nuclear translocation of NF-κB p65 in chondrocytes. Cells were immunostained with rabbit anti–NF–κB p65 antibodies (green) and visualized using confocal microscopy. Cytoskeletal structures were stained with phalloidin (red) and nuclei were stained with DAPI (blue). Scale bar = 20 μm. (E) Western blotting for collagen II, MMP13, and ADAM9 in chondrocytes. Significant differences were found between the normal and IL-1β groups (*P < 0.001; ANOVA), IL-1β and IL-1β+LXA 4 groups ( + P < 0.001; ANOVA), and the IL-1β+LXA 4 and IL-1β+LXA 4 +PHTPP groups ( # P < 0.001; ANOVA). Data are expressed as the mean ± 95% confidence interval; n = 3 per group. Immunofluorescence and immunoblotting showed that NF-κB p65 was significantly aggregated in the nuclei of chondrocytes after IL-1β intervention in a co-culture system, whereas LXA4 inhibited the nuclear translocation of NF-κB p65 ( Fig. 3 C and D). Next, we explored the role of LPAR3 in KOA and the regulation of ferroptosis in FLSs. Rat FLSs were treated with 10 ng/ml IL-1β and siRNAs targeting the LPAR3 gene were introduced to these cells. The knockdown efficiency of LPAR3 and the corresponding in vitro phenotypic markers were detected using quantitative PCR (q-PCR) ( Fig. 4 E) and immunoblotting assays ( Fig. 4 A). After IL-1β treatment, the levels of synovial inflammation-related destructive markers MMP3 and MMP13, as well as the ferroptosis marker COX2, were significantly increased; the levels of HO-1 and Nrf2 were upregulated, while the levels of SOD1 and GPX4 were downregulated ( Fig. 4 A). The successful introduction of siRNAs resulted in a significant decrease in LPAR3 levels and a further increase in MMP3, ACSL4, MMP13, and COX2 levels, while Nrf2, SLC7A11, HO-1, SOD1, and GPX4 levels were decreased ( Fig. 4 A). To elucidate the mechanism by which LPA3 regulates ferroptosis in FLSs, we used JC-1 staining, FerroOrange staining, 2,7-dichlorofluorescein diacetate (DCFH-DA), the fluorescent probe MitoSOX Red, and the corresponding flow cytometry assay techniques to determine the mitochondrial membrane potential, cellular ROS, intracellular iron, and mitochondrial ROS levels, respectively ( Fig. 4 B). FLSs treated with IL-1β exhibited decreased mitochondrial membrane potential, increased mitochondrial and intracellular ROS levels, and aggregation of intracellular Fe 2+ ( Fig. 4 B and C). These changes were further exacerbated by the successful LPAR3 knockdown. In addition, TEM also showed ferroptosis in the LPAR3-knockdown group, mainly characterized by the presence of smaller mitochondria, increased membrane density, and reduced mitochondrial cristae ( Fig. 4 D). Fig. 4 LPAR3 knockdown promotes ferroptosis and inflammation in fibroblast-like synoviocytes (FLSs). (A) Western blot of COX2, ACSL4, MMP13, Nrf2, MMP3, LPAR3, SLC7A11, HO-1, GPX4, and SOD1 after successful knockdown of the LPAR3 gene in FLSs. Significant differences were found between the normal and IL-1β groups (*P < 0.001; *aP = 0.0392, *bP = 0.0086, *cP = 0.001, *dP = 0.0203, and *eP = 0.0325), IL-1β+KD1 and IL-1β groups ( # P < 0.001, # aP = 0.0025, and # bP = 0.0291), and the IL-1β+KD2 and IL-1β groups ( & P < 0.001, & aP = 0.0404, & bP = 0.0477, & cP = 0.0015, & dP = 0.0015). (B) and (C) Results of incubation with the JC-1 fluorescent probe, FerroOrange probe, DCFH-DA, MitoSOX Red, and corresponding flow cytometric assays ( # aP = 0.0014) are shown. (D) Transmission electron microscopy of the mitochondrial morphology changes in FLSs. (E) qRT-PCR results and corresponding statistical analysis of LPAR3, MMP13, HO-1, Nrf2, and GPX4 expression. A comparison of all groups with the KDNC group is shown (*P < 0.001; *aP = 0.0034, *bP = 0.001, *cP = 0.0064, *dP = 0.0351, *eP = 0.0016). ns, not significant. Data are expressed as the mean ± 95% confidence interval; n = 3 per group. Fig. 4 LPAR3 knockdown promotes ferroptosis and inflammation in fibroblast-like synoviocytes (FLSs). (A) Western blot of COX2, ACSL4, MMP13, Nrf2, MMP3, LPAR3, SLC7A11, HO-1, GPX4, and SOD1 after successful knockdown of the LPAR3 gene in FLSs. Significant differences were found between the normal and IL-1β groups (*P < 0.001; *aP = 0.0392, *bP = 0.0086, *cP = 0.001, *dP = 0.0203, and *eP = 0.0325), IL-1β+KD1 and IL-1β groups ( # P < 0.001, # aP = 0.0025, and # bP = 0.0291), and the IL-1β+KD2 and IL-1β groups ( & P < 0.001, & aP = 0.0404, & bP = 0.0477, & cP = 0.0015, & dP = 0.0015). (B) and (C) Results of incubation with the JC-1 fluorescent probe, FerroOrange probe, DCFH-DA, MitoSOX Red, and corresponding flow cytometric assays ( # aP = 0.0014) are shown. (D) Transmission electron microscopy of the mitochondrial morphology changes in FLSs. (E) qRT-PCR results and corresponding statistical analysis of LPAR3, MMP13, HO-1, Nrf2, and GPX4 expression. A comparison of all groups with the KDNC group is shown (*P < 0.001; *aP = 0.0034, *bP = 0.001, *cP = 0.0064, *dP = 0.0351, *eP = 0.0016). ns, not significant. Data are expressed as the mean ± 95% confidence interval; n = 3 per group. These results suggest that FLSs showed signs of antioxidant imbalance and ferroptosis after applying IL-1β and that LPAR3 knockdown exacerbated the progression of inflammation and ferroptosis in IL-1β-treated FLSs. To further validate the role of LPAR3 in regulating ferroptosis in FLSs in KOA, an overexpression plasmid targeting the LPAR3 gene was introduced into FLSs after induction with 10 ng/ml IL-1β. The efficiency of LPAR3 gene overexpression and the corresponding phenotypic markers in vitro were confirmed via q-PCR ( Fig. 5 E) and western blotting ( Fig. 5 A). The expression of LPAR3 was significantly increased; among the destructive markers, MMP13, ACSL4, and MMP3 expression were significantly decreased, whereas the levels of the antioxidant factors Nrf2, HO-1, SLC7A11 and SOD1, as well as the key regulator of ferroptosis, GPX4, increased ( Fig. 5 A). TEM showed that LPAR3 overexpression increased the number of mitochondrial cristae as well as improved mitochondrial morphology and membrane integrity ( Fig. 5 C). Confocal microscope showed an increase in mitochondrial membrane potential, a decrease in mitochondrial and intracellular ROS levels, and a decrease in intracellular Fe 2+ levels ( Fig. 5 B). Flow cytometry analysis supported this result ( Fig. 5 B). Immunofluorescence analysis showed that LPAR3 overexpression promoted the nuclear translocation of Nrf2 ( Fig. 5 D). Collectively, these results suggest that LPAR3 overexpression reversed the progression of ferroptosis in IL-1β-treated FLSs. Fig. 5 LPAR3 overexpression inhibits ferroptosis and inflammation in fibroblast-like synoviocytes (FLSs) after IL-1β intervention. (A) Western blot of COX2, ACSL4, MMP13, Nrf2, MMP3, LPAR3, SLC7A11, HO-1, GPX4, and SOD1 after successful overexpression of the LPAR3 gene in FLSs. Significant differences were observed between the normal and IL-1β groups (*P < 0.001, *aP = 0.043, *bP = 0.0026, *cP = 0.0018, *dP = 0.01, *eP = 0.075), and the IL-1β+OE-LPAR3 and IL-1β groups ( # P < 0.001, # aP = 0.0302, # bP = 0.0113, # cP = 0.0206). (B) Results of incubation with the JC-1 fluorescent probe, FerroOrange probe, DCFH-DA, MitoSOX Red, and corresponding flow cytometric assays are shown (*P < 0.001, *aP = 0.0036, *bP = 0.0044). (C) Transmission electron microscopy observation of mitochondrial morphology changes in FLSs. (D) Immunofluorescence assay reveals the nuclear translocation of Nrf2 in FLSs. (E) qRT-PCR results and corresponding statistical analyses of LPAR3, HO-1, MMP13, Nrf2, and GPX4 expression (*P < 0.001; *aP = 0.0027, *bP = 0.0011, *cP = 0.0399, *dP = 0.031). ns, not significant. Data are expressed as the mean ± 95% confidence interval; n = 3 per group. Fig. 5 LPAR3 overexpression inhibits ferroptosis and inflammation in fibroblast-like synoviocytes (FLSs) after IL-1β intervention. (A) Western blot of COX2, ACSL4, MMP13, Nrf2, MMP3, LPAR3, SLC7A11, HO-1, GPX4, and SOD1 after successful overexpression of the LPAR3 gene in FLSs. Significant differences were observed between the normal and IL-1β groups (*P < 0.001, *aP = 0.043, *bP = 0.0026, *cP = 0.0018, *dP = 0.01, *eP = 0.075), and the IL-1β+OE-LPAR3 and IL-1β groups ( # P < 0.001, # aP = 0.0302, # bP = 0.0113, # cP = 0.0206). (B) Results of incubation with the JC-1 fluorescent probe, FerroOrange probe, DCFH-DA, MitoSOX Red, and corresponding flow cytometric assays are shown (*P < 0.001, *aP = 0.0036, *bP = 0.0044). (C) Transmission electron microscopy observation of mitochondrial morphology changes in FLSs. (D) Immunofluorescence assay reveals the nuclear translocation of Nrf2 in FLSs. (E) qRT-PCR results and corresponding statistical analyses of LPAR3, HO-1, MMP13, Nrf2, and GPX4 expression (*P < 0.001; *aP = 0.0027, *bP = 0.0011, *cP = 0.0399, *dP = 0.031). ns, not significant. Data are expressed as the mean ± 95% confidence interval; n = 3 per group. We noted that Nrf2 pathway-associated protein levels varied with LPAR3 levels. To further characterize the association between the two, we added a specific inhibitor of Nrf2, ML385 (5 μM), to FLSs overexpressing the LPAR3 gene. Intervention with ML385 did not significantly alter the levels of MMP3 and LPAR3, but up-regulated the levels of the destructive factors MMP13 and COX2, while down-regulating the levels of the protective factors Nrf2, HO-1, SOD1, and GPX4 ( Fig. 6 A). The q-PCR showed similar results to the immunoblot,ML385 intervention down-regulated the mRNA levels of Nrf2, HO-1 and GPX4, while up-regulating the mRNA levels of MMP13 ( Fig. 6 C). Confocal microscopy and flow cytometry analyses showed that ML385 intervention decreased mitochondrial membrane potential, increased mitochondrial and intracellular ROS levels, and increased intracellular Fe 2+ aggregation in the FLSs of the LPAR3 overexpression group ( Fig. 6 B). TEM revealed a reduction in mitochondrial volume, the disappearance of cristae, and membrane disruption after treatment with ML385 ( Fig. 6 D). Taken together, these results suggest that LPAR3 inhibits ferroptosis in FLSs from IL-1β-treated rats by regulating Nrf2 levels and function, reducing synovial inflammation, and alleviating cartilage degeneration. Fig. 6 LPAR3 exerts a therapeutic effect on osteoarthritis depending on the involvement of its downstream antioxidant factor, Nrf2. (A) Western blotting of COX2, MMP13, Nrf2, MMP3, LPAR3, HO-1, GPX4, and SOD1 expression after the successful overexpression of the LPAR3 gene in FLSs treated with ML385. Significant differences were observed between the IL-1β+OENC and IL-1β+OE-LPAR3 groups (*P < 0.001, *aP = 0.0075, *bP = 0.0016), and the IL-1β+OE-LPAR3 and IL-1β+OE-LPAR3+ML385 groups ( # P < 0.001, # aP = 0.0154, # bP = 0.0078, # cP = 0.0015, # dP = 0.0097). (B) Results of incubation with the JC-1 fluorescent probe, FerroOrange probe, DCFH-DA, MitoSOX Red, and corresponding flow cytometric assays are shown (*P < 0.001, *aP = 0.0075, *bP = 0.001). (C) qRT-PCR results and corresponding statistical analyses of LPAR3, HO-1, MMP13, Nrf2, and GPX4 expression are shown (*P < 0.001; *aP = 0.0086, *bP = 0.0148, *cP = 0.0013). (D) Transmission electron microscopy observation of mitochondrial morphology changes in FLSs. ns, not significant. Data are expressed as the mean ± 95% confidence interval; n = 3 per group. Fig. 6 LPAR3 exerts a therapeutic effect on osteoarthritis depending on the involvement of its downstream antioxidant factor, Nrf2. (A) Western blotting of COX2, MMP13, Nrf2, MMP3, LPAR3, HO-1, GPX4, and SOD1 expression after the successful overexpression of the LPAR3 gene in FLSs treated with ML385. Significant differences were observed between the IL-1β+OENC and IL-1β+OE-LPAR3 groups (*P < 0.001, *aP = 0.0075, *bP = 0.0016), and the IL-1β+OE-LPAR3 and IL-1β+OE-LPAR3+ML385 groups ( # P < 0.001, # aP = 0.0154, # bP = 0.0078, # cP = 0.0015, # dP = 0.0097). (B) Results of incubation with the JC-1 fluorescent probe, FerroOrange probe, DCFH-DA, MitoSOX Red, and corresponding flow cytometric assays are shown (*P < 0.001, *aP = 0.0075, *bP = 0.001). (C) qRT-PCR results and corresponding statistical analyses of LPAR3, HO-1, MMP13, Nrf2, and GPX4 expression are shown (*P < 0.001; *aP = 0.0086, *bP = 0.0148, *cP = 0.0013). (D) Transmission electron microscopy observation of mitochondrial morphology changes in FLSs. ns, not significant. Data are expressed as the mean ± 95% confidence interval; n = 3 per group. To investigate why LXA 4 was able to upregulate the level of LPAR3 in FLSs after IL-1β treatment in the co-culture system, we predicted the potential TFs (transcription factors) of LPAR3 ( Homo sapiens ) using the JASPAR database. We found that ESR2 can act as a TF of LPAR3 ( Fig. S3G ), and co-expression analysis showed they were strongly associated ( Fig. S3F ). Dual luciferase reporter gene experiments showed direct targeting of ESR2 to the predicted binding site of the LPAR3 promoter ( Fig. 7 D). Fig. 7 ESR2 is involved in the pathogenesis of osteoarthritis by regulating LPAR3 expression. (A) Western blotting of COX2, MMP13, Nrf2, MMP3, ESR2, LPAR3, HO-1, GPX4, and SOD1 expression. Significant differences were observed between the IL-1β+OENC and IL-1β+OE-ESR2 groups (*P < 0.001, *aP = 0.0473 *bP = 0.16, *cP = 0.0032, *dP = 0.0046, *eP = 0.0035), and the IL-1β+OE-ESR2+KD-NC and IL-1β+OE-ESR2+KD-LPAR3 groups ( # P < 0.001, # aP = 0.0031, # bP = 0.0011, # cP = 0.0252, # dP = 0.0273, # eP = 0.0025). (B) Results of incubation with the JC-1 fluorescent probe, FerroOrange probe, DCFH-DA, MitoSOX Red, and corresponding flow cytometric assays are shown (*P < 0.001, *aP = 0.0242, *bP = 0.0055). (C) qRT-PCR results and corresponding statistical analyses of ESR2, LPAR3, HO-1, MMP13, and GPX4 expression. Significant differences between the OENC and OE-ESR2 groups (*P < 0.001; *aP = 0.0168, *bP = 0.0479) and the OE-ESR2 and OE-ESR2+KD2 groups ( + P < 0.001, + aP = 0.0014, + bP = 0.0011, + cP = 0.0048) are shown. (D) Analysis of the results of the dual-luciferase reporter gene assay ( # P < 0.001, + P < 0.001). (E) Transmission electron microscopy observation of mitochondrial morphology changes in FLSs. ns, not significant. Data are expressed as the mean ± 95% confidence interval; n = 3 per group. Fig. 7 ESR2 is involved in the pathogenesis of osteoarthritis by regulating LPAR3 expression. (A) Western blotting of COX2, MMP13, Nrf2, MMP3, ESR2, LPAR3, HO-1, GPX4, and SOD1 expression. Significant differences were observed between the IL-1β+OENC and IL-1β+OE-ESR2 groups (*P < 0.001, *aP = 0.0473 *bP = 0.16, *cP = 0.0032, *dP = 0.0046, *eP = 0.0035), and the IL-1β+OE-ESR2+KD-NC and IL-1β+OE-ESR2+KD-LPAR3 groups ( # P < 0.001, # aP = 0.0031, # bP = 0.0011, # cP = 0.0252, # dP = 0.0273, # eP = 0.0025). (B) Results of incubation with the JC-1 fluorescent probe, FerroOrange probe, DCFH-DA, MitoSOX Red, and corresponding flow cytometric assays are shown (*P < 0.001, *aP = 0.0242, *bP = 0.0055). (C) qRT-PCR results and corresponding statistical analyses of ESR2, LPAR3, HO-1, MMP13, and GPX4 expression. Significant differences between the OENC and OE-ESR2 groups (*P < 0.001; *aP = 0.0168, *bP = 0.0479) and the OE-ESR2 and OE-ESR2+KD2 groups ( + P < 0.001, + aP = 0.0014, + bP = 0.0011, + cP = 0.0048) are shown. (D) Analysis of the results of the dual-luciferase reporter gene assay ( # P < 0.001, + P < 0.001). (E) Transmission electron microscopy observation of mitochondrial morphology changes in FLSs. ns, not significant. Data are expressed as the mean ± 95% confidence interval; n = 3 per group. ESR2 has been reported to be a novel intracellular receptor for LXA 4 , and molecular docking showed stable and good binding activity between the two (Docking_score = −6.5 kcal/mol) ( Fig. S4 ). Immunoblotting showed low ESR2 expression in the synovium of patients with OA and in FLSs after IL-1β induction, whereas addition of LXA 4 intervention upregulated ESR2 levels in FLSs ( Fig. 3 B). Next, we performed immunoblotting, confocal microscopy, and flow cytometry analyses to determine whether ESR2 regulates ferroptosis through the transcriptional activation of LPAR3. We observed that after ESR2 overexpression, the expression of LPAR3 was also significantly increased. Among the destructive markers, the expressions of COX2, MMP13, and MMP3 were downregulated, while the levels of the antioxidant factors Nrf2, HO-1, and GPX4 were upregulated; the level of SOD1 did not significantly change ( Fig. 7 A). After the successful overexpression of ESR2, we added KD-LAPR3 and observed a significant reversal of the above trends, except that the levels of ESR2 and MPP13 did not significantly change ( Fig. 7 A). The efficiency of ESR2 overexpression and the corresponding phenotypic markers in vitro were confirmed via q-PCR ( Fig. 7 C) and western blotting ( Fig. 7 A). Confocal microscopy imaging and flow cytometry analyses showed that KD-LPAR3 treatment reversed the increase in mitochondrial membrane potential, the decrease in mitochondrial and intracellular ROS levels, and the decrease in intracellular Fe 2+ in FLSs induced by ESR2 overexpression ( Fig. 7 B). TEM showed irregular mitochondrial morphology, loss of cristae, and rupture of the mitochondrial membrane after successful knockdown of LPAR3 ( Fig. 7 E). In this study, intra-articular injection of MIA was used to establish a KOA model in rats. Forty male SD rats were randomly divided into four groups and monitored weekly for changes in body weight ( Fig. 8 A and Fig. S6A ). Gross imaging meatus scores ( Fig. 8 B, left), X-ray imaging ( Fig. 8 B, right), histological assessment (Mankin and OARSI scores) ( Fig. 8 C–E), and synovial histological assessment (microscopy scores) ( Fig. 8 F) revealed more severe cartilage and synovial damage in the OAG group than in the CG group. The OAM treatment group showed a significant therapeutic effect on the tibiofemoral joint cartilage compared to the OAG group; the treatment efficacy observed in this group was inhibited by PHTPP treatment ( Fig. 8 C). Fig. 8 Treadmill exercise suppresses osteoarthritis-related degeneration in MIA-induced rats. (A) Animal experiment design (n = 10 rats/group). (B) Gross images and X-ray radiographs of the rat knee joints. (C) Cartilage stained with hematoxylin and eosin (H&E), toluidine blue, and senna O, and synovial H&E staining. (D) Histologic Mankin scoring results. (E) Histologic OARSI scoring results. (F) Synovial microscopy scoring. Significant differences were found between the CG and OAG groups (*P < 0.001), OAG and OAM groups ( + P < 0.001), and the OAM and OAM + PHTPP groups ( # P < 0.001). Data are expressed as the mean ± 95% confidence interval; n = 10 rats per group. Fig. 8 Treadmill exercise suppresses osteoarthritis-related degeneration in MIA-induced rats. (A) Animal experiment design (n = 10 rats/group). (B) Gross images and X-ray radiographs of the rat knee joints. (C) Cartilage stained with hematoxylin and eosin (H&E), toluidine blue, and senna O, and synovial H&E staining. (D) Histologic Mankin scoring results. (E) Histologic OARSI scoring results. (F) Synovial microscopy scoring. Significant differences were found between the CG and OAG groups (*P < 0.001), OAG and OAM groups ( + P < 0.001), and the OAM and OAM + PHTPP groups ( # P < 0.001). Data are expressed as the mean ± 95% confidence interval; n = 10 rats per group. Gait analysis clearly showed instability and lameness in the OAG group. Observations in rats in this group include walking due to pain and shorter average stride and step lengths. The step width was longer in the OAG group than in the CG group. Mean step length and stride length were longer in the OAM group than in the OAG group and more similar to the CG group; however, this effect was partially reversed with the addition of PHTPP ( Fig. 9 A). Immunohistochemistry and immunoblotting assays revealed the activation and imbalance of the antioxidant system in the synovial tissues of the OAG group compared to the CG group, mainly observed as elevated levels of Nrf2 and HO-1 and reduced levels of GPX4 and SOD1 ( Fig. 9 B and C). ESR2 and LPAR3 levels were also elevated after treadmill exercise ( Fig. 9 B and C). PHTPP treatment partially blocked the therapeutic effects of treadmill exercise ( Fig. 9 B and C). Fig. 9 (A) Gait analysis in rats. Significant differences were observed between the CG and OAG groups (*P < 0.001), OAG and OAM groups ( + P < 0.001, + aP = 0.004, + bP = 0.003, + cP = 0.001), and the OAM and OAM + PHTPP groups ( # P < 0.001, # aP = 0.02, # bP = 0.035). (B) Western blotting of ESR2, LPAR3, HO-1, GPX4, and SOD1 expression in synovial tissue ( + aP = 0.026). (C) Immunohistochemical staining of synovial tissue for GPX4, HO-1, LPAR3, Nrf2, and SOD1 expresson ( + aP = 0.017, + bP = 0.015, + cP = 0.015, # aP = 0.02, # bP = 0.044). Data are expressed as the mean ± 95% confidence interval; n = 3 rats per group. Fig. 9 (A) Gait analysis in rats. Significant differences were observed between the CG and OAG groups (*P < 0.001), OAG and OAM groups ( + P < 0.001, + aP = 0.004, + bP = 0.003, + cP = 0.001), and the OAM and OAM + PHTPP groups ( # P < 0.001, # aP = 0.02, # bP = 0.035). (B) Western blotting of ESR2, LPAR3, HO-1, GPX4, and SOD1 expression in synovial tissue ( + aP = 0.026). (C) Immunohistochemical staining of synovial tissue for GPX4, HO-1, LPAR3, Nrf2, and SOD1 expresson ( + aP = 0.017, + bP = 0.015, + cP = 0.015, # aP = 0.02, # bP = 0.044). Data are expressed as the mean ± 95% confidence interval; n = 3 rats per group.

Transcriptomic and clinical data ( GSE55235 and GSE29746 datasets) were downloaded from the Gene Expression Omnibus database ( http://www.ncbi.nih.gov/geo ). Synovial tissues from 20 healthy joints and 26 synovial tissues from patients with OA were collected from the GSE55235 dataset. FLSs transcriptome data from 11 healthy individuals and 11 OA patients were collected from the GSE29746 dataset. The microarray data were log-transformed, and genes with adjusted P-values  1.5 were filtered as differentially expressed genes (DEGs). The limma package in the R software (version 3.40.2) was used to analyze the differential gene expression levels. The batch effects were removed using the remote BatchEffect function of the limma package. The heatmap package was used to generate the heat maps. Gene Ontology (GO) enrichment analysis ( http://www.geneontology.org/ ), including molecular function (MF), biological pathway (BP), and cellular component (CC), was used to annotate genes with biological functions. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis ( http://www.genome.jp/kegg/pathway.html ) was used to elucidate pathways associated with the DEGs. ClusterProfiler (version 3.18.0) was used to analyze the GO functions of the potential targets and KEGG pathways enriched in the DEGs. GO and KEGG terms with P-vaues <0.05 were considered statistically significant in this study. The NCBI gene database was used to search for the promoter regions of LPAR3 . Transcription factors (TFs) were identified using the JASPAR 2019 TFBS trace from the Genome Browser at the University of California, Santa Cruz. Based on the co-expression analysis between DEGs, we focused on ESR2 and predicted its binding site in the LPAR3 promoter region using the JASPAR database. The TF intersections of the LPAR3 promoter sequences were extracted by analyzing the hTFtarget, HumanTFDB, and GTRD databases and were used to re-validate the predictions in the JASPAR database. This study was approved by the Ethics Committee of Shengjing Hospital of China Medical University (Shenyang, China; Approval No. 2021PS265K). Informed consent was obtained from all patients. We collected intraoperative synovial tissue samples from patients with different KL grades, a procedure that does not involve tissue damage or extraction outside of normal surgery. Moreover, we collected imaging data of patients’ knees on admission and intraoperative arthroscopic gross images. Baseline patient profiles are shown in Table 4 , human and animal ethical approvals are detailed in Fig. S8 . RNA was extracted from FLSs using the RNAiso Plus kit (Vazyme,RC102-01) and then reverse transcribed to obtain cDNA. The qRT-PCR reaction was performed using a SYBR Green PCR kit and an Applied Biosystems 7500 Real-Time PCR system. Primers were designed by Sangon and purchased from China ( Table 1 ). Table 1 PCR primer sequence. Table 1 Gene Direction Sequence Nrf2 FORWARD TGCCTTCCTCTGCTGCCATTAG REVERSE CATTGAACTCCACCGTGCCTTC HO-1 FORWARD AGAGACGCCCCGAGGAAAATC REVERSE TGCCACGGTCGCCAACAG LPAR3 FORWARD CTTGGCGGCTGCGGATTTC REVERSE CCGTCAGGCTTGTGTCTAGGAG MMP13 FORWARD TGATGATGAAACCTGGACAAGCA REVERSE GAACGTCATCTCTGGGAGCA ESR2 FORWARD TCTCACGTCAGGCACATCAGTAAC REVERSE CAGCATCTCCAGCAGCAGGTC β-ACTIN FORWARD GGAGATTACTGCCCTGGCTCCTA REVERSE GACTCATCGTACTC CTGCTTGCTG PCR primer sequence. FLSs were obtained from the synovial membrane of the knee joints of 4-week-old SD rats (100 ± 10 g; specific pathogen-free [SPF]). The tissues were collected in sterile phosphate-buffered saline (PBS). Fat and connective tissue were removed and digested with 1 mg/ml collagenase (Sigma-Aldrich, St. Louis, MO, USA) for 1–2 h at 37 °C. Cells were then separated from the undigested tissue using a 70-μm cell strainer (BD, Durham, NC, USA) and cultured in 25-cm 2 cell culture flasks in DMEM (Dulbecco's modified Eagle medium; Gibco BRL, Grand Island, NY, USA) with 10% fetal bovine serum (Clark, Shanghai, China) and antibiotics (100 U/ml penicillin, 100 μg/ml streptomycin, and 0.1 mg/ml amphotericin B) in a humid atmosphere of 5% CO 2 in air at 37 °C. Upon reaching confluence, the cells were detached using 0.25% trypsin and split in a 1:3 ratio. Cells from the 4th to 6th passages were used in all experiments. Light microscopy analysis confirmed that >95% of the cells were FLSs. Primary articular chondrocytes were obtained from the hip and knee cartilage of 3-week-old male SD rats (80 ± 10 g; SPF). The tissues were collected in sterile PBS. After sequentially digesting the cartilage with 1.5 mg/ml pronase K (Sigma-Aldrich) for 60 min at 37 °C and in 1.2 mg/ml collagenase D (Sigma-Aldrich) for 60 min at 37 °C, primary chondrocytes were inoculated into 25-cm 2 culture flasks. All cells were cultured in DMEM (Gibco BRL) containing 10% fetal bovine serum (Clark) and antibiotics (100 U/ml penicillin, 100 μg/ml streptomycin, and 0.1 mg/ml amphotericin B) and placed in an incubator at 37 °C with 5% CO 2 in the air. Immunocytochemical staining for type II collagen was performed to determine the chondrocyte phenotype. Primary chondrocytes were passaged at a 1:3 ratio. Chondrocytes from passages 1 to 2 were used for subsequent experiments. All cells were washed with PBS, fixed in 4% paraformaldehyde for 20 min at room temperature, permeabilized with 0.5% Triton X-100 for 30 min, and incubated with a blocking solution (5% bovine serum albumin [BSA]) for 30 min at room temperature. The chondrocytes were then incubated with rabbit anti-collagen type II polyclonal antibodies (ab34712, 1:75; Abcam) and rabbit anti–NF–κB p65 polyclonal antibodies (10745-1-AP, 1:75; Proteintech). FLSs were incubated with rabbit anti-vimentin polyclonal antibodies (ab45939, 1:75; Abcam) and rabbit anti-Nrf2 polyclonal antibodies (ab31163, 1:75; Abcam) overnight at 4 °C and then incubated with Alexa Fluor®488-conjugated anti-rabbit antibodies for 60 min at room temperature in the dark. Cell nuclei were stained with 4,6-diamidino-2-phenylindole (DAPI) for 2 min. After washing, cells adhering to the Bioflex membranes were fixed in PBS containing 20% glycerol. Finally, the chondrocytes were observed under a confocal microscope (Olympus, Tokyo, Japan). FLSs were rinsed with ice-cold PBS and lysed with RIPA lysis buffer (Beyotime; P0013C) containing 1% PMSF (Beyotime, China; ST506). The supernatant was collected via centrifugation at 14,000× g for 5 min at 4 °C. Protein concentrations were quantified using a bicinchoninic acid (BCA) assay kit (Beyotime). Then, 30 μg of protein from each sample was separated using 10–15% SDS-PAGE (Sigma-Aldrich) and transferred to a polyvinylidene difluoride (PVDF) membrane. After blocking the membranes with 1% BSA in Tris-buffered saline (TBS) containing 0.1% Tween-20 (TBST) for 2 h at 22 ± 2 °C, they were incubated with the following primary antibodies at 4 °C overnight: rabbit polyclonal anti-ESR2 antibody (14007-1-AP, 1:1000; Proteintech), rabbit polyclonal anti–NF–κB p65 antibodies (10745-1-AP, 1:1000; Proteintech), rabbit polyclonal ACSL4 antibodies (22401-1-AP; 1:2000; Proteintech), rabbit polyclonal anti-MMP13 antibody (ab39012, 1:1000; Abcam), rabbit monoclonal anti-MM3 antibody (ab52915, 1:1000; Abcam), rabbit monoclonal anti-SLC7A11 antibody (ab307601, 1:1000; Abcam), rabbit polyclonal anti-histone H2A.X (AB41012, 1:1000; Absci), rabbit polyclonal anti-LPAR3 antibody (orb394997, 1:1000; Biorbyt), rabbit monoclonal anti-Nrf2 antibody (ab31163, 1:1000; Abcam), rabbit monoclonal anti–HO–1 antibody (ab189491, 1:1000; Abcam), rabbit monoclonal anti-GPX4 antibody (ab125066, 1:1500; Abcam), rabbit monoclonal anti-SOD1 antibody (ab308181, 1:1000; Abcam), rabbit monoclonal anti-COX2 anti-rabbit (ab179800, 1:2000; Abcam). After three rinses with TBST, the membranes were incubated with IgG-horseradish peroxidase (HRP)-conjugated secondary antibodies (1:10,000, Canlife) for 2 h at 22 ± 2 °C. Finally, the blots were visualized via chemiluminescence (MedChemExpress) using an Amersham Imager 800 (GE, Boston, MA, USA) and analyzed using ImageJ software. FLSs were placed in the bottom compartment of a Transwell system, and chondrocytes were placed in the top compartment. All cells were incubated in 0.4-μm inserts. The two cell types were mixed in a 3:1 (FLSs: Chondrocytes) ratio. At the start of the study, cell survival was >95%. FLSs were treated with 10 ng/ml IL-1β (ab200284, Abcam), 10 nM LXA 4 (90410, Cayman), and 5 nM PHTPP (abs814771, Absin), while chondrocytes were treated with 10 ng/ml IL-1β (ab200284, Abcam) only. After 24 h of incubation, cells were collected for further analysis. Nucleo-plasmic separation of lysed chondrocytes was carried out according to the instructions of the Nucleo- and Cytoplasmic Protein Extraction Kit (P0027, Beyotime) and the samples were placed in the −80 for subsequent protein level assays. After incubating FLSs with a JC-1 fluorescent probe (Beyotime) in the dark for 30 min at 37 °C, they were washed twice with a JC-1 staining buffer. To detect JC-1 monomers, images were captured using a confocal microscope (Olympus) at an excitation wavelength of 490 nm and an emission wavelength of 530 nm. JC-1 aggregates were observed at excitation and emission wavelengths of 525 and 590 nm, respectively. Cells were collected and analyzed using a flow cytometer (CytoFLEX LX, Beckman). The ratio of the mean red fluorescence intensity to the mean green fluorescence intensity was used as an indicator of mitochondrial membrane potential. After washing with PBS, FLSs were harvested, centrifuged at 2500 rpm for 2 min, and cold-fixed in 2.5% glutaraldehyde for 30 min. The samples were fixed again in 1% OsO 4 for 1 h, dehydrated, and embedded in epoxy resin. Ultrathin sections (60 nm thick) were stained with uranyl acetate and lead citrate. The sections were visualized using a transmission electron microscope (TEM; Hitachi 800, Tokyo, Japan) to observe cell morphology. FLSs were treated with 5 μM Mito-SOX Red dye (MedChemExpress) and incubated in the dark at 37 °C for 30 min. Subsequently, the FLSs were rinsed twice with PBS, and images were captured using a confocal microscope (Olympus) at excitation and emission wavelengths of 510 and 580 nm, respectively. In addition, the cells were collected, and the average fluorescence intensity of each group was measured using a flow cytometer (CytoFLEX LX, Beckman). Intracellular reactive oxygen species (ROS) were detected using an ROS Detection Kit (S0033, Beyotime) following the manufacturer's instructions. After treating the FLSs, they were collected and incubated with a DCFH-DA working solution for 30 min at 37 °C in the dark. The cells were then washed twice with sterile PBS. Fluorescence microscopy (Eclipse Ci; Nikon, Japan) was used to detect ROS-mediated fluorescence. The cells were also collected and analyzed using a flow cytometer (CytoFLEX LX, Beckman). FerroOrange is a non-fluorescent cell-permeable dye in its unbound state; however, it displays a fluorescent signal when bound to Fe2 + ions. After incubating FLSs with a FerroOrange fluorescent probe (MedChemExpress) for 30 min at 37 °C in the dark, cells were washed twice with a serum-free cell culture solution. Images were captured using a confocal microscope (Olympus). Cells were collected and analyzed using a flow cytometer (CytoFLEX LX, Beckman). Effective LPAR3 siRNA and control siRNA (Hanbio, HH20230622DY-SI01) were obtained from the supplier. When the chondrocytes reached 50% confluence, they were transfected with LPAR3 siRNA using jetPRIME® Transfection Reagent (Polyplus) according to the manufacturer's protocol. The siRNA primers used are listed in Table 2 . Table 2 siRNA sequence. Table 2 Name Sequence siRNA NC sense UUCUCCGAACGUGUCACGU TT antisense ACGUGACACGUUCGGAGAA TT LPAR3 (r)-si-1 sense GCUGGUAAUUGCUGUGGAATT antisense UUCCACAGCAAUUACCAGCTT LPAR3 (r)-si-2 sense GCAUCUACAUGUAUGUCAATT antisense UUGACAUACAUGUAGAUGCTT siRNA sequence. When the cells reached 50% confluence, LPAR3 and ESR2 overexpression plasmids (Hanbio, HH20230723DY-PC01) were mixed with Lipofectamine 3000 and diluted in DMEM according to the manufacturer's protocol. The mixture was incubated at room temperature for 15 min and added to the cells cultured in each dish. After 48 h, the transfected cells were harvested. The primers used for the plasmids are listed in Table 3 . Table 3 Plasmid primer sequences. Table 3 Gene Direction Sequence r-LPAR3 FORWARD GAGACCCAAGCTGGCTAGTTGaattcGCCACCatggactttttctac REVERSE TCACTTAAGCTTGGTACCGAggatccggagctgcttttattgcacac r-ESR2 FORWARD TTAAGCTTGGTACCGAGCTCGGATCCGCCACCAtgtccatctgtacc REVERSE CACTGTGCTGGATATCTGCAGAATTCTTATTTGTCGTCATCATCC TT r-LPAR3-pro FORWARD GATAGGTACCGAGCTCTTACGCGTCTGCTGCTTTGTGACACTGATC REVERSE AGCTTACTTAGATCGCAGATCTCGAGTGGGGGAGAGGAGGCATGA AG Table 4 Patient's basic information situation. Table 4 Basic information KL0 KL1 KL2 KL3 KL4 χ 2 P Biological sex Male 3 2 3 2 2 0.696 0.191 Female 2 3 3 5 6 Age(years) <65 5 5 3 1 1 0.001 <0.001 ≥65 0 0 3 6 7 Obesity gradation Under weight 1 0 1 0 1 0.839 0.486 Normal weight 2 3 4 4 3 Overweight 2 1 1 1 3 Obesity 0 1 0 2 1 Plasmid primer sequences. Patient's basic information situation. The LPAR3 promoter sequence was cloned into the pGL3-basic vector established upstream of F-luciferase, while ESR2 was cloned into the eukaryotic expression vector pcDNA3.1 (Hanbio, HH20230721DY-PC01). Both vectors were then co-transfected into 293T cells, and the luciferase activity in these cells was detected after 48 h using a dual-luciferase reporter kit (Promega, E1910). F-luciferase activity is a response to high or low promoter activity. We used a Cell Counting Kit-8 (CCK-8) assay (GK10001; GLPBIO, USA) to evaluate the optimal concentration of IL-1β for our experiments. Briefly, FLSs (3 × 10 3  cells/well) were seeded into 96-well plates and treated with increasing concentrations of IL-1β (0, 2, 4, 8, 10, 20, and 40 ng/ml) for 24 h at 60% confluence. Then, 10 μL of CCK-8 and 90 μL of complete culture medium were added to each well and incubated at 37 °C for 2 h. Finally, the absorbance was measured at 450 nm using a Gen5 plate reader (BioTek, Winooski, VT, USA). The study protocol was approved by the Animal Care and Use Committee of China Medical University (Shenyang, China; approval number: 2021PS220K). Animal experiments adhered to the ARRIVE guidelines and were conducted in accordance with the National Research Council's Guidelines for the Care and Use of Laboratory Animals. To avoid physiological changes in estrogen and related receptor levels, we used male SD rats in all of our experiments. Sprague–Dawley (SD) rats (230 ± 10 g, SPF level, aged 8 weeks) were purchased from HFK Bioscience Co., Ltd. (Beijing, China). The rats were housed in plastic cages lined with sawdust bedding at 22–25 °C under SPF conditions, with 70% humidity and a 12 h/12 h light-dark cycle. All rats were acclimatized to a ZH-PT treadmill (Zhongshidichuang Science and Technology Development Co., Ltd., Beijing, China) for one week at a speed of 10 m/min for 10 min per day to reduce stress. After adapting to the treadmill exercise, the rats were randomly divided into four groups (n = 10 in each group): control group (CG), which received an intra-articular injection of 50 μl of sterile saline; KOA group (OAG), a KOA model induced via intra-articular injection of MIA (1 mg in 50 μl of sterile saline) through the patellar suprapatellar ligament bilaterally into the knee joint cavities; OAM group, which similar to OAG but was coupled with a moderate-intensity treadmill exercise at 19.3 m/min, 5° inclination, 60 min/day, 5 days/week for 4 weeks; and the OAM + PHTPP group, where the rats were injected intraperitoneally with 100 μg/kg of PHTPP before each treadmill exercise session. Following the experimental cycle, the rats were anesthetized with an intraperitoneal injection of sodium pentobarbital (40 mg/kg). The bilateral ankles of the SD rats were fixed on a pallet, and plain films of the knee joints were obtained using an X-ray machine (MS FX PRO; Bruker, Contiki, Belgium). The radiographs were evaluated based on joint space narrowing and osteophyte formation. Gross images of the rat knee joints were also acquired using a stereomicroscope (XTL-165-CB; Phenix), SkyScan 1276-CT (Bruker), and NRecon version 1.6 software (Bruker). Rat knee samples were stored in 4% paraformaldehyde for seven days. After rinsing in distilled water for 5 h, the samples were immersed in a 10% ethylenediaminetetraacetic acid (EDTA) solution and decalcified at 37 °C for 28 days. The decalcification solution was changed every three days. Decalcified samples were dehydrated in increasing concentrations of ethanol and embedded in paraffin. Sagittal sections (4 μm thick) were cut from the paraffin-embedded tissues and stained with hematoxylin and eosin (H&E), Senna O, and toluidine blue. Femoral and tibial articular cartilage damage was assessed using a modified Mankin score (0–14) and the Osteoarthritis Research Society International (OARSI) score (0–24) [ 23 , 24 ]. In addition, we used synovial microscopy scoring to assess the histological changes in the synovium. Three experienced observers blindly scored the samples. Gait analysis was used to assess the lameness of the limb affected by KOA. Briefly, the animals were placed in a 100 cm × 10 cm open gait arena and allowed to walk freely from one side to the other without external stimuli or food temptation. The tracks of each rat were recorded by dipping the hind paw in red dye and the front paw in blue dye. The outcome measurements were recorded by three independent examiners who were blinded to the experimental conditions. Paraffin-embedded knee sections were deparaffinized in xylene and then deparaffinized stepwise in water using an ethanol gradient. Sections were treated with 10 mM sodium citrate buffer (pH 6.0) for 30 min at 37 °C for antigen recovery. The sections were then incubated with 0.3% H 2 O 2 for 1 h to block endogenous peroxidase activity. Subsequently, the sections were treated with a 3% BSA solution for 30 min at room temperature to block non-specific antigens. Then, the following primary antibodies were added: rabbit polyclonal anti-ESR2 antibody (14007-1-AP, 1:1000; Proteintech), rabbit polyclonal anti-LPAR3 antibody (orb394997, 1:1000; Biorbyt), rabbit monoclonal anti–HO–1 antibody (ab189491, 1:2000; Abcam), rabbit monoclonal anti-GPX4 antibody (ab125066, 1:500; Abcam), rabbit monoclonal anti-SOD1 antibody (ab308181, 1:100; Abcam), rabbit polyclonal anti-CGRP antibody (16630-1-AP, 1:100; Proteintech), overnight at 4 °C in a cold room. Immunohistochemical reactions were visualized using diaminobenzidine followed by counterstaining with hematoxylin. Images of the sections were captured using a light microscope (BX-53; Olympus) and analyzed using Image-Pro Plus 6.0. Data were analyzed using SPSS software (version 21.0; SPSS Inc.). The Shapiro–Wilk test and Levene's test were used to assess the normality and homogeneity of the data, respectively. Statistical significance was determined using one-way analysis of variance (ANOVA). The nonparametric Kruskal–Wallis test was applied to non-normally distributed data. Statistical significance was set at P < 0.05.

In conclusion, we used cell co-culture techniques to demonstrate that LXA 4 ameliorates chondrocyte matrix degradation by inhibiting ferroptosis in FLSs. In vitro , we explored that the ESR2/LPAR3/Nrf2 axis regulates ferroptosis through the overexpression and knockdown of LPAR3. In vivo , we explored the role of ESR2, a novel receptor for LXA 4 , in KOA using PHTPP and briefly proposed a conjecture about pain in KOA. Altering the phenotype of the cartilage microenvironment by targeting LPAR3 may represent a novel therapeutic approach for limiting the progression of KOA.

In a preliminary clinical study using high-throughput bioinformatics analysis combined with initial validation of synovial samples from patients with different KL grades, it was found that KOA synovium undergoes ferroptosis and is chronically exposed to oxidative stress damage, a process that may be associated with the development of neuropathic pain in KOA. One of our previous studies on the negative effects of synovial iron overload on KOA supports these conclusions [ 25 ]. In the present study, ESR2 and LPAR3 were recognized as critical modulators involved in synovial ferroptosis as their protein levels were down-regulated with increasing KL grade. Synoviocytes are an important cell type in the cartilage microenvironment, whose pathological phenotype determines variations in the microenvironmental components, indirectly affecting chondrocyte metabolism [ 26 ]. Cellular compartmentalization performed for painful versus nonpainful areas of the synovium of KOA shows an important effect of changes in the pathological phenotype of FLSs on KOA. It is noteworthy that although bioinformatics analysis predicted no significant difference in LPAR3 in the synovial dataset ( Figs. S2D and S2E ), it was significantly downregulated in the FLSs dataset. This predicts that LPAR3 may act as a characteristic gene in the involvement of FLSs in KOA progression. To investigate whether LXA 4 could affect chondrocyte matrix degradation by regulating ferroptosis in FLSs, we constructed a co-culture system of FLSs and chondrocytes. The results showed that LXA 4 treatment elevated ESR2 and LPAR3 levels to attenuate ferroptosis in FLSs, ultimately alleviating chondrocyte ECM degeneration and inflammatory responses. This conclusion validates our conjecture from the preliminary experiments. Furthermore, it provides a new perspective for treating KOA; to our knowledge, this is the first time that influencing cartilage degeneration by modulating ferroptosis in FLSs has been explored. As the most widely studied receptor for LPA signaling, LPAR3 is regarded as a protective factor against oxidative stress and a mitochondrial "gatekeeper" [ 27 ]. To explore the role played by LPAR3 in FLSs, we used siRNA and plasmids to knock down and overexpress LPAR3 in vitro . In an IL-1β-induced inflammation model, we found that LPAR3 can protect against KOA by regulating ferroptosis and inflammation in FLSs. The LPAR-induced regulation of ferroptosis involves protecting mitochondrial function, improving iron metabolic processes, and elevating the levels of antioxidant molecules. LPAR3 also activates Nrf2-related signaling pathways that regulate oxidative stress and iron metabolism [ 14 ]. The variations in protein levels observed in our experiments correspond with this observation. Furthermore, intervention with the Nrf2 inhibitor ML385 reversed the suppression of ferroptosis and the reduction of inflammation caused by LPAR3 gene overexpression. To elucidate the reasons for LPAR3 upregulation in the co-culture system after LXA 4 intervention, we predicted multiple TFs in the promoter region of LPAR3 ( Homo sapiens ) using the JASPAR database. We selected ESR2 as the predicted upstream TF of the LPAR3 gene because bioinformatics analysis revealed a strong co-expression relationship between ESR2 and LPAR3 and its level was down-regulated in synovial membranes of patients with KOA in preliminary clinical specimen validation. We first performed dual-luciferase experiments to demonstrate the presence of an ESR2 binding site in the promoter sequence of LPAR3 . Notably ESR2 was reported as a novel intracellular receptor for LXA 4 [ 17 ]. In the co-culture system, LXA 4 upregulated both ESR2 and LPAR3 levels in FLSs. In vitro experiments, knockdown of the LPAR3 gene reversed the inhibition of ferroptosis by ESR2 gene overexpression in FLSs. This reinforces our conclusion that ESR2 exerts anti-inflammatory and antioxidant effects in FLSs after IL-1β intervention, and that these effects depend on the normal function of its downstream target, LPAR3. In our in vivo experiments, we used a rat model of intra-articular MIA treated with moderate-intensity treadmill exercise and a intraperitoneal injection of PHTPP. The gross imaging meatus scores, X-ray imaging, histological assessments, and synovial inflammation scores were consistent with our previous findings [ 28 ], confirming that moderate-intensity treadmill exercise effectively alleviated KOA cartilage and synovial lesions. In addition, IHC and immunoblotting of synovial membranes showed that moderate-intensity platform training was effective in correcting the imbalance in antioxidant responses and increasing the levels of ESR2 and LPAR3, consistent with the results obtained in the co-culture system after treatment with LXA4. However, PHTPP, a selective ESR2 antagonist, partially reversed these therapeutic effects. These findings reveal that exercise can alleviate KOA lesions by inhibiting synovial ferroptosis by elevating LXA4 levels and that this therapeutic effect is partially dependent on the function of ESR2, a novel intracellular receptor for LXA4. It is noteworthy that the role of FPR2, the most widely studied membrane receptor of LXA 4 , in "inflammation termination" has been widely demonstrated. In one of our previous studies, we explored the interaction between LXA 4 and the conventional membrane receptors of FLSs in KOA cases using the FPR2-specific antagonist BOC-2, and verified that LXA 4 could reduce the inflammatory response of FLSs by binding to FPR2, thereby treating KOA. In the present study, we further explored the effect of ESR2, the intramembrane receptor for LXA 4 , on the pathological phenotype of FLSs, building on the known finding that LXA 4 ameliorates inflammation in FLSs by binding to its membrane receptor FPR2. This finding refines the potential mechanism by which LXA 4 regulates various diseases through the exertion of estrogen-like effects, providing novel targets for treating KOA, especially the hormone-related type of KOA. Gait analysis was performed to assess the pain behavior of KOA model rats visually. Changes in the step and stride lengths of the affected hind and front paw footprints clearly reflect lameness during walking. The severity of claudication of SD rats was significantly reduced by moderate-intensity treadmill exercise. KOA-related pain has long been considered a nociceptive pain due to the "wear and tear" pathology; however, clinical patients describe pain that is initially activity-related but subsequently stabilizes over time [ 29 ]. CGRP, which is released from the sensory and efferent nerve terminals, is thought to be a major component of neurogenic inflammation and is involved in the hyperalgesia of joints [ 30 ]. CGRP levels in the synovium of specimens from clinical patients increased with increasing KL grade ( Fig. 2 A and F). All these phenomena suggest that neuropathic pain is equally involved in the development of KOA pain [ 8 ], especially in patients with advanced KOA and intractable pain. Since LXA 4 is involved in spinal nociceptive processing, we hypothesized that moderate-intensity exercise could attenuate neurogenic pain induced by synovial involvement during KOA by producing high levels of LXA 4 . However, whether the specific mechanism involved in this process is related to the inhibition of ferroptosis in FLSs by LXA 4 requires further elucidation. Appropriate exercise generates a hypoxic environment in the vasculature on the one hand, promoting platelet aggregation, lipoxygenase activation, and LXA 4 production. Meanwhile, it also generates synovial fluid and lipids in the infrapatellar fat pads, producing LXA 4 in the knee joint [ 7 ]. At high concentrations, LXA 4 in the blood and synovial fluid utilizes its lipophilicity to enter the cytoplasm of FLSs, bind to its receptor ESR2, and promote the transcriptional function of ESR2, which in turn upregulates the expression of LPAR3. Activated LPA signaling promotes the upregulation and nuclear translocation of Nrf2 by binding to LPAR3, ultimately activating antioxidant response elements. The activation of various antioxidant factors facilitates the direct reduction of ROS levels in the cytoplasm and reduces the sources of ROS by improving mitochondrial function. Finally, decreased lipid peroxidation levels inhibit ferroptosis in FLSs. In conclusion, we propose that LXA 4 inhibits ferroptosis in FLSs via the ESR2/LPAR3/Nrf2 axis, ultimately alleviating the pathological progression of KOA ( Fig. 10 ). Fig. 10 Schematic representation of the mechanism by which LXA 4 inhibits ferroptosis in FLSs: by activating the ESR2/LPAR3/Nrf2 axis. Fig. 10 Schematic representation of the mechanism by which LXA 4 inhibits ferroptosis in FLSs: by activating the ESR2/LPAR3/Nrf2 axis. This study has several limitations. First, the MIA-induced rat model of KOA does not fully characterize all aspects of human KOA, especially spontaneous KOA associated with aging. Similarly, although IL-1β serves as the cytokine that is most commonly used to mimic inflammatory conditions in FLSs, intervention modeling using only one inflammatory factor does not accurately match the intricacies of pathogenesis when KOA occurs in vivo. In the future studies, we will continue to explore the modeling of other inflammatory factors, such as TGF-β1, IL-1α, and IL-18, to screen the optimal in vitro modeling of KOA. Second, the entire knee joint is in a complex dynamic environment, and the variation in cartilage pathologic phenotype cannot be explained only by co-culture of FLSs and chondrocytes under in vitro conditions. Finally, we only used gross imaging and plain radiographs to assess knee degeneration in animal experiments. To further characterize the function of LXA 4 in KOA, we recommend using high-resolution micro-computed tomography and magnetic resonance imaging.

Knee osteoarthritis (KOA) is a degenerative disease of the joints primarily characterized by articular cartilage degeneration, synovial inflammation, and subchondral bone remodeling [ 1 ]. Multiple risk factors, including aging, obesity, trauma, and mechanical loading, play important roles in the pathogenesis of KOA [ 2 ]. While KOA was thought to be caused by the wearing and tearing of articular cartilage, it is now recognized as a chronic total joint disease [ 3 ]. Despite being one of the most common causes of chronic pain, the underlying pathogenesis of KOA remains to be further elucidated [ 4 ]. Lipoxin A 4 (LXA 4 ) is a potent anti-inflammatory lipid mediator that promotes the reduction of inflammation and is considered a novel "inflammation turn-off signal" [ 5 ]. Previous studies by our group have demonstrated that LXA 4 is rapidly generated and metabolized at appropriate exercise intensities and can be considered as a potential biomarker for evaluating the efficacy of exercise in treating KOA [ 6 ]. The main mechanism mainly involves attenuating the production of inflammatory mediators in fibroblast-like synoviocytes (FLSs) and promoting synovial macrophage polarization via the upregulation of LXA 4 [ 7 ]. Current KOA treatments mainly focus on relieving pain symptoms. Notably, exercise can effectively relieve KOA pain, with a previous study indicating the involvement of LXA 4 in spinal nociceptive processing, thereby reducing pain due to inflammation [ 8 ]. Given that the synovium is richly innervated by sensory nerves in periarticular tissues, we identified cellular subpopulations based on the synovium originating from both painful and non-painful areas of OA ( Fig. S1 , previously published) [ 9 ]. The proportions of different cell populations in the synovium vary greatly, including synovial FLSs, monocytes, and Schwann cells, implying their significant involvement in KOA pain. Therefore, we hypothesized that LXA 4 influences KOA-induced pain and disease progression by modulating the pathological phenotype of FLSs. Using high-throughput bioinformatics analysis, we found that the lysophosphatidic acid receptor-3 ( LPAR3 ) gene was significantly downregulated in the FLSs originating from KOA patients ( Fig. S2A ). Lysophosphatidic acid (LPA) is a multifunctional bioactive phospholipid involved in neuropathic pain that mediates C-fiber retraction and A-fiber demyelination [ 10 ]. Recent studies have revealed that extracellular matrix (ECM) synthesis and degradation are strictly regulated by LPA signaling [ 11 ]. LPAR3, the most widely studied receptor for LPA signaling, is regarded as the gatekeeper of mitochondrial health and protects cells from oxidative stress [ 12 ]. It has been reported that the reduction of cytokine release by ameliorating oxidative stress in FLSs contributes to alleviating KOA pain [ 13 ]. A recent study on Hutchinson-Gilford progeria syndrome showed that LPAR3 attenuates oxidative stress by improving mitochondrial homeostasis and modulating the expression of the downstream antioxidant factor Nrf2 [ 12 , 14 ]. However, its biological role in KOA remains unclear. We previously identified the estrogen receptor beta (ESR2)-binding sites in the LPAR3 ( Homo sapiens ) promoter using the JASPAR database and found strong co-expression relationship between them ( Fig. S3 ). ESR2 is a classical nuclear receptor located in the nucleus that exerts "genotypic" regulatory effects by controlling the transcription of specific target genes [ 15 ]. Currently, ESR2 has been validated to inhibit oxidative damage induced by high-glycemic conditions and promote intracellular matrix synthesis by activating the p38MAPK pathway [ 16 ]. Interestingly, in endometriosis, LXA 4 was found to exert estrogen-like effects by binding to the estrogen receptor (ER) and was considered a novel ER modulator [ 17 ]. ESR2 is also considered a novel receptor for LXA 4 , and its expression was elevated in LXA 4 -treated embryonic stem cells [ 18 ]. Molecular docking prediction modeling showed strong binding activity between LXA 4 and ESR2 (Docking_score = −6.5 kcal/mol) ( Fig. S4 ). Therefore, we hypothesized that LXA 4 regulates the ESR2-mediated targeting of LPAR3 gene. Ferroptosis is an iron-dependent form of cell death that is regulated by excessive lipid peroxidation [ 19 ]. Several studies have indicated the occurrence of ferroptosis in damaged areas of the synovium in patients with KOA with lipid peroxidation levels higher in these patients than in healthy individuals [ 20 ]. To investigate the relationship between synovial ferroptosis and the pathological progression of KOA, differential analysis of synovial samples from patients with different KL grades in our preliminary experiments revealed that GPX4 was downregulated in synovial tissues from patients with KOA ( Fig. 1 B). GPX4, a critical regulatory protein in ferroptosis, has been shown to intersect with the mechanisms regulating cartilage oxidative stress and ECM degradation [ 21 ]. Although it has been reported that LPAR3 inhibits ferroptosis by attenuating cellular oxidative stress, the potential mechanism by which LPAR3 regulates ferroptosis in the KOA synovium requires further elucidation. Fig. 1 ESR2, LPAR3, and GPX4 levels are downregulated in synovial tissue with increasing Kellgren–Lawrence (KL) grades. (A) Plain radiographs and magnetic resonance imaging (MRI) scans. A–P: anterior–posterior. (B) Western blot of ESR2, LPAR3, and GPX4 expression in human synovial tissues. (C) Statistical analysis of the relative protein expression levels using β-actin as an internal control. A comparison of all KL grades with grade 0 is shown (*P < 0.001, *aP = 0.004, *bP = 0.001, ANOVA). Data are expressed as the mean ± 95% confidence interval; n = 3 patients per group. Fig. 1 ESR2, LPAR3, and GPX4 levels are downregulated in synovial tissue with increasing Kellgren–Lawrence (KL) grades. (A) Plain radiographs and magnetic resonance imaging (MRI) scans. A–P: anterior–posterior. (B) Western blot of ESR2, LPAR3, and GPX4 expression in human synovial tissues. (C) Statistical analysis of the relative protein expression levels using β-actin as an internal control. A comparison of all KL grades with grade 0 is shown (*P < 0.001, *aP = 0.004, *bP = 0.001, ANOVA). Data are expressed as the mean ± 95% confidence interval; n = 3 patients per group. The knee cartilage tissue is a complex interactive environment containing various non-chondrocytes, including synoviocytes, aside from chondrocytes. These cellular components secrete various metabolic and inflammatory factors through paracrine, autocrine, and endocrine pathways. These factors interact with each other to maintain homeostasis of the articular cartilage, constituting the cartilage microenvironment [ 22 ]. However, existing studies were generally limited to exploring the mechanisms involved in chondrocyte ferroptosis, ignoring the effects of changes in the cartilage microenvironment. Hence, in this study, we constructed a co-culture system of FLSs and chondrocytes in a KOA environment to explore the effects of FLSs ferroptosis on cartilage degeneration. Built on previous findings that LXA 4 is rapidly generated and metabolized during moderate-intensity [ 6 ], this study aimed to investigate the potential pathological mechanisms by which exercise attenuates cartilage degeneration and knee pain through LXA 4 . Using bioinformatics and clinical specimen differential analyses, we identified the involvement of LPAR3 in the regulation of the synovial ferroptosis process as a key contributor to KOA advancement. Ferroptosis regulation through the ESR2/LPAR3/Nrf2 axis was investigated by knocking down and overexpressing the LPAR3 gene. A cell co-culture system was also constructed to explore the effects of FLSs ferroptosis on cartilage degradation. We also constructed a rat model of MIA-induced KOA to explore the regulatory role of LXA 4 and its intracellular receptor, ESR2, in KOA pain and progression. Our findings provide a novel therapeutic target for KOA and reveal the potential mechanism underlying the effect of LXA 4 in treating KOA.

lipoxin A 4 estrogen receptor beta lysophosphatidic acid receptor-3 analysis of variance biological pathway cellular component control group differentially expressed genes estrogen receptor fibroblast-like synoviocytes monosodium iodoacetate gene ontology hematoxylin and eosin Kyoto Encyclopedia of Genes and Genomes Kellgren–Lawrence molecular function magnetic resonance imaging phosphate-buffered saline reactive oxygen species Tris-buffered saline transmission electron microscopy transcription factor extracellular matrix matrix metalloproteinase 13 A distegrinin and a metalloprotease 9 Cyclooxygenase 2

The authors declare that there is no conflict of interest.