GATD3A-deficiency-induced mitochondrial dysfunction facilitates senescence of fibroblast-like synoviocytes and osteoarthritis progression

Abstract

Accumulating evidence indicates that cellular senescence is closely associated with osteoarthritis. However, there is limited research on the mechanisms underlying fibroblast-like synoviocyte senescence and its impact on osteoarthritis progression. Here, we elucidate a positive correlation between fibroblast-like synoviocyte senescence and osteoarthritis progression and reveal that GATD3A deficiency induces fibroblast-like synoviocyte senescence. Mechanistically, GATD3A deficiency enhances the binding of Sirt3 to MDH2, leading to deacetylation and decreased activity of MDH2. Reduced MDH2 activity impairs tricarboxylic acid cycle flux, resulting in mitochondrial dysfunction and fibroblast-like synoviocyte senescence. Intra-articular injection of recombinant adeno-associated virus carrying GATD3A significantly alleviates the osteoarthritis phenotype in male mice. This study increases our current understanding of GATD3A function. In particular, we reveal a novel mechanism of fibroblast-like synoviocyte senescence, suggesting that targeting GATD3A is a potential therapeutic approach for osteoarthritis.

Fig. 1. Senescence of FLSs is associated with OA progression.

a Representative images of immunohistochemistry staining of p16^INK4a in normal and OA patients. n = 10 independent biological samples. Scale bar: 100 µm. b Representative images of immunofluorescence staining of p16^INK4a in synovium from normal and OA patients, and quantification of the percentage of p16^INK4a positive FLSs (n  =  25 Nor, n  =  33 OA). Scale bar: 100 µm. c Spearman’s correlation analysis was performed between Mankin score and the percentage of p16^INK4a positive FLSs in OA samples (n  =  33). d Quantitative PCR analysis of CDKN1A and CDKN2A in FLSs. (n  =  25 Nor, n  =  33 OA). e Western blotting analysis of senescent markers. n = 3 independent biological replicates. f Representative images of SA-β-Gal staining for Nor-FLSs and OA-FLSs. n  =  9 independent biological replicates. Scale bar: 100 µm. g Representative images of immunofluorescence staining of p16^INK4a in Nor-FLSs and OA-FLSs, and the percentage of p16^INK4a positive cells. n  =  15 independent biological replicates. Scale bar: 50 µm. h Quantitative PCR analysis (n  =  3 independent biological replicates) and Elisa assay (n  =  6 independent biological replicates) of pro-inflammatory cytokine (IL-1β and TNF-α) in Nor-FLSs and OA-FLSs. i Western blotting assay to detect the protein levels of chondrocytes co-cultured with Nor-FLSs or OA-FLSs. n = 3 independent biological replicates. j Representative images of Safranin O/Fast green staining of knee joints. n  =  10. Scale bar: 200 µm. k Representative images of immunohistochemistry staining of p16^INK4a. n  =  10. Scale bar: 100 µm. l The severity of the OA phenotype was assessed using the OARSI score system. n = 10 independent biological animals. m Representative images of immunofluorescence in knee joints. S, synovium; C, cartilage. Scale bar: 500 µm (Left panel), 50 µm (Right panel). n Quantification of the percentage of p16^INK4a positive FLSs in the synovium of mice. n  =  10 independent biological animals. o Spearman’s correlation analysis was performed between OARSI score and the percentage of p16^INK4a positive FLSs in OA mice. n  =  10. All data were presented as the means ± SD. P values were determined by two-tailed unpaired Student’s t-test. P < 0.05 was considered statistically significant. Source data are provided as a Source Data file.

Fig. 2. Mitochondrial function is impaired in OA-FLSs.

a The heatmap of RNA-Seq analysis for FLSs separated from normal and OA patients. b The biological processes of differentially expressed mRNAs were analysed by GO enrichment analysis. c GSEA of ‘OXPHOS’ gene sets in Nor-FLSs and OA-FLSs. (d) Representative images of Mitotracker staining to observe the mitochondrial network of Nor-FLSs (n  =  12) and OA-FLSs (n  =  11), and the statistical analysis of mean branch length. Scale bar: 40 µm. The white dashed box represents the enlarged image area in the bottom right corner. e Representative images of TEM to observe Nor-FLSs (n  =  5) and OA-FLSs (n  =  12) and the arrows point to the mitochondria. Scale bar: 1 µm. f Representative images of MitoSOX staining to detect mitochondrial superoxide in Nor-FLSs and OA-FLSs. n  =  6 independent biological replicates. Scale bar: 100 µm. g Mitochondrial membrane potential detected by JC-1 assay and the statistical analysis of aggregate-to-monomer ratio (n  =  3 Nor-FLSs, n  =  7 OA-FLSs). h, i OCR of Nor-FLSs and OA-FLSs, and the analysis of mitochondrial respiration. n  =  3 independent biological replicates. j, k PER of Nor-FLSs and OA-FLSs, and the analysis of basal and compensatory glycolysis. n  =  4 independent biological replicates. l Quantification of ATP content in Nor-FLSs and OA-FLSs. n  =  5 independent biological replicates. All data were presented as the means ± SD. P values were determined by two-tailed unpaired Student’s t-test. P < 0.05 was considered statistically significant. Source data are provided as a Source Data file.

Fig. 3. Decreased GATD3A contributes to mitochondrial dysfunction and promotes FLS senescence.

a Volcano plot of RNA-Seq data. b Western blotting analysis of GATD3A in synovium and FLSs (whole cell lysate) from normal and OA patients. c Quantitative PCR analysis of GATD3A in FLSs (n  =  25 Nor, n  =  33 OA). d Pearson’s correlation between GATD3A expression and CDKN2A expression in OA-FLSs (n  =  33). e Representative images of immunofluorescence of GATD3A in synovium. n = 15 independent biological samples. Scale bar: 100 µm. f Western blotting localization of GATD3A. n  =  3 independent biological replicates. g Fluorescent analysis of GATD3A localization, with signal intensity profiles along a line and Pearson’s correlation analysis (n  =  5 TOMM20, n  =  6 ATP5A and GRP75). Scale bar: 5 µm. Mitotracker staining (n = 10 ShNC, n = 12 ShGATD3A, n = 13 Vec, n = 11 GATD3A) (h), TEM (n = 16 ShNC and ShGATD3A, n = 8 Vec, n = 9 GATD3A) (i) were conducted. Scale bar: 40 µm for Mitotracker staining, 1 µm for TEM. OCR (n = 4 ShNC and ShGATD3A, n  =  5 Vec and GATD3A) (j) and PER (n = 5 ShNC and ShGATD3A, n = 6 Vec, n = 4 GATD3A) (k) assays of Nor-FLSs or OA-FLSs after different treatments. l, m Quantification of ATP content in FLSs after different treatments. n  =  6 independent biological replicates. Western blotting analysis (n  =  3 independent biological replicates) (n), SA-β-Gal staining (n = 6 ShNC and ShGATD3A, n  =  8 Vec and GATD3A) (o) and immunofluorescence (n  =  10 ShNC and ShGATD3A, n  =  13 Vec and GATD3A) (p) of FLSs after different treatments. Scale bar: 100 µm for SA-β-Gal staining, 50 µm for immunofluorescence. q, r Elisa assay to detect the pro-inflammatory cytokines. n  =  6 independent biological replicates. s Western blotting analysis of chondrocytes co-cultured with FLSs after different treatments. n  =  3 independent biological replicates. All data were presented as the means ± SD. P values were determined by two-tailed unpaired Student’s t-test or one-way ANOVA with Tukey’s multiple-comparisons (g). P < 0.05 was considered statistically significant. Source data are provided as a Source Data file.

Fig. 4. GATD3A overexpression in FLSs alleviates OA development in mice.

a Experimental design diagram of intra-articular injection of rAAV-Vec or rAAV-GATD3A in mice (Created in BioRender. Shen, K. (2024) https://BioRender.com/w90b978). b Representative images of immunohistochemistry staining of GATD3A in the knee joint from mice injected with rAAV-Vec or rAAV-GATD3A. n = 7 independent biological samples. Scale bar: 50 µm. c Representative images of immunofluorescent staining of vimentin, GATD3A, and p16^INK4a in the knee joint from sham or DMM-induced OA mice after intra-articular injection with rAAV-Vec or rAAV-GATD3A. Enlarged images were shown below. n = 7 independent biological samples. Scale bar: 50 µm. d 3D reconstructed images of micro-CT of knees and osteophytes (arrows pointed), and statistical analyses of the number of osteophytes. n = 7 independent biological animals. e Representative coronal plane images of knee joints obtained from micro-CT data analyses. f Quantification of bone volume per total volume (BV/TV), and trabecular thickness (Tb. Th) obtained from micro-CT data. n = 7 independent biological animals. g Representative images of H&E staining, Safranin O/Fast green staining and Toluidine Blue staining for knee joints from sham or DMM-induced OA mice after intra-articular injection with rAAV-Vec or rAAV-GATD3A. n = 7 independent biological animals. Scale bar: 100 µm. The severities of synovitis and OA phenotype were assessed through synovial inflammation score (h) and OARSI score system (i). n = 7 independent biological animals. All data were presented as the means ± SD. P values were determined by one-way ANOVA with Tukey’s multiple-comparisons. P < 0.05 was considered statistically significant. Source data are provided as a Source Data file.

Fig. 5. GATD3A deficiency reduces TCA cycle flux, altering the metabolic pattern of FLSs.

a Heatmap of label-free proteomics assay for Nor-FLSs infected with ShNC or ShGATD3A lentivirus. b Number of proteins involved in biological processes was analysed through GO enrichment analysis based on proteomic assay. c KEGG pathway enrichment analysis showed the number of proteins interacting with GATD3A within these pathways based on IP/MS. OPLS-DA analysis (d) and KEGG pathway enrichment analysis (e) of the untargeted metabolomics study for Nor-FLSs infected with ShNC or ShGATD3A lentivirus. f Heatmap of significantly different metabolites in the TCA cycle, glutaminolysis and glycolysis based on the untargeted metabolomics study for Nor-FLSs infected with ShNC or ShGATD3A lentivirus. OCR (n  =  5 for glucose supplementation group, and n = 4 for other two groups) (g) and PER (n  =  6) (h) assays of Nor-FLSs treated with glucose or glutamine, and glucose and glutamine deprivation. i Diagram of ¹³C isotopomer patterns with [¹³C₅] glutamine and [U-¹³C₆] glucose as tracer. j Absolute isotopologue distribution of [¹³C₅] glutamine-derived glutaminolysis and TCA cycle metabolites including glutamate, α-KG, succinate, fumarate, malate, oxaloacetate, citrate and aspartate. n  =  4 independent biological replicates. The incorporation of ¹³C atoms was denoted as m + n (number of ¹³C atoms). k Absolute isotopologue distribution of [U-¹³C₆] glucose-derived glycolysis metabolites including pyruvate and lactate. n  =  4 independent biological replicates. The incorporation of ¹³C atoms was denoted as m + n (number of ¹³C atoms). l Cellular malate and oxaloacetate content were determined in Nor-FLSs infected with ShNC or ShGATD3A lentivirus. n  =  5 independent biological replicates. The glucose uptake rate (n  =  5 independent biological replicates) (m) and the cellular lactate content (n  =  4 independent biological replicates) (n) in Nor-FLSs infected with ShNC or ShGATD3A lentivirus. All data were presented as the means ± SD. P values were determined by two-tailed unpaired Student’s t-test or one-way ANOVA with Tukey’s multiple-comparisons (g, h). P < 0.05 was considered statistically significant. Source data are provided as a Source Data file.

Fig. 6. GATD3A interacts with MDH2, a key enzyme of the TCA cycle.

a The heatmap of significantly different enzymes in the TCA cycle based on proteomics assay for Nor-FLSs infected with ShNC or ShGATD3A lentivirus. b IP/MS analysis indicated that MDH2 is an interacting protein that binds to GATD3A. c Co-IP assay to verify the interaction between endogenous GATD3A and MDH2 in FLSs. n  =  3 independent biological replicates. d Representative images of immunofluorescence to verify the co-localisation of endogenous GATD3A and MDH2 in FLSs. n  =  5 independent biological replicates. Scale bar: 10 µm. e Co-IP assay to verify the interaction between exogenous Flag-GATD3A and His-MDH2 in HEK 293 T cells. n  =  3 independent biological replicates. f Representative images of immunofluorescence to verify the co-localisation of exogenous Flag-GATD3A and His-MDH2 in HEK 293 T cells. n  =  5 independent biological replicates. Scale bar: 10 µm. g Design of GATD3A and MDH2 truncations. h Co-IP assay showing the interactions between Flag-GATD3A and His-MDH2 in HEK 293 T cells. n  =  3 independent biological replicates. i Molecular docking predicted the docking sites of GATD3A and MDH2. j Docking sites of GATD3A and MDH2, and the mutations of binding sites between GATD3A and MDH2. k Co-IP assay showing the interactions between Flag-tagged mutated GATD3A and His-tagged mutated MDH2 in HEK 293 T cells. n  =  3 independent biological replicates. Source data are provided as a Source Data file.

Fig. 7. GATD3A deficiency induces mitochondrial dysfunction by increasing Sirt3-mediated MDH2 deacetylation.

a The MDH2 activity in Nor-FLSs infected with ShNC or ShGATD3A. n  =  5 independent biological replicates. b Western blotting analysis of protein acetylation levels in FLSs. n  =  3 independent biological replicates. c Immunoprecipitation assay assessing MDH2 acetylation in FLSs infected with ShNC or ShGATD3A. n  =  3 independent biological replicates. HEK293T cells infected with ShNC or ShGATD3A were transfected with His-MDH2 alone (d), with His-MDH2 and Flag-GATD3A (e), or with His-MDH2 and treated with NAM or TSA (f). Immunoprecipitation with anti-His was conducted to assess His-MDH2 acetylation. n = 3 independent biological replicates. g GATD3A-knockdowned Nor-FLSs were treated with NAM or TSA, and the MDH2 activity was assed. n  =  5 independent biological replicates. h–k OCR (n  =  5 NAM + TSA, n = 6 other four groups) and PER assays (n  =  3 ShGATD3A, n = 4 other groups) of Nor-FLSs after different treatment. l, m Representative images of SA-β-Gal staining (n = 7 ShNC, n = 6 DMSO, n = 4 NAM and TSA, n = 5 NAM + TSA) and immunofluorescence staining (n  =  15) of Nor-FLSs after different treatment. Scale bar: 100 µm for SA-β-Gal staining, 50 µm for immunofluorescence. n Western blotting analysis of senescent markers in differently treated Nor-FLSs. n  =  3 independent biological replicates. o HEK293T cells were transfected with His-MDH2 WT and 4KR acetylation-deficient mutants, and the acetylation of His-MDH2 was detected. n  =  3 independent biological replicates. p Co-IP assay to verify the interaction between MDH2 and Sirt3 in GATD3A-knockdowned Nor-FLSs. n  =  3 independent biological replicates. q Co-IP assay to verify the interaction between His-MDH2 and Sirt3 in GATD3A-knockdowned HEK293T cells after NAM treatment. n  =  3 independent biological replicates. r GATD3A-knockdowned HEK293T cells were co-transfected with His-MDH2, Myc-Sirt3, and Flag-GATD3A to assess the effect of Flag-GATD3A on His-MDH2 binding to Myc-Sirt3 via Co-IP assay. n  =  3 independent biological replicates. All data were presented as the means ± SD. P values were determined by two-tailed unpaired Student’s t-test (a) or one-way ANOVA with Tukey’s multiple-comparisons. P < 0.05 was considered statistically significant. Source data are provided as a Source Data file.

Fig. 8. Oxaloacetate attenuates GATD3A-deficiency-induced mitochondrial dysfunction and OA development.

a Role of oxaloacetate in TCA cycle and malate-aspartate shuttle (Created in BioRender. Shen, K. (2024) https://BioRender.com/w90b978). b The content of oxaloacetate in Nor-FLSs after infection with ShNC or ShGATD3A lentivirus and supplementation with oxaloacetate. n = 5 independent biological replicates. OCR (n  =  4 ShNC, and n = 3 other two groups) (c) and PER (n  =  4 ShGATD3A, and n = 3 other two groups) d assays of Nor-FLSs after infection with ShNC or ShGATD3A lentivirus and supplementation with 1 mM oxaloacetate. e, f Representative images of SA-β-Gal staining (n  =  10) and immunofluorescence staining (n  =  15) of Nor-FLSs after infection with ShNC or ShGATD3A lentivirus and supplementation with 1 mM oxaloacetate. Scale bar: 100 µm for SA-β-Gal staining and 50 µm for immunofluorescence. g Western blotting and immunoprecipitation analysis to detect the senescent markers and the acetylation level of MDH2 in Nor-FLSs after infection with ShNC or ShGATD3A lentivirus and supplementation with 1 mM oxaloacetate. n  =  3 independent biological replicates. h Experimental design diagram of intraperitoneal injection of oxaloacetate in mice (Created in BioRender. Shen, K. (2024) https://BioRender.com/w90b978). i Representative images of immunofluorescent staining in the knee joint from DMM-induced OA mice with or without oxaloacetate treatment. n = 5 independent biological samples. Scale bar: 50 µm. j 3D reconstruction images of micro-CT of the knees and osteophytes (arrows pointed), and the statistical analysis of the number of osteophytes. n  =  5 independent biological animals. k Representative coronal plane images of knee joints. l Quantification of BV/TV, and Tb. Th. n  =  5 independent biological animals. Representative images of H&E staining (m), Safranin O/Fast green staining and Toluidine Blue staining (n) for knee joints. n = 5 independent biological samples. Scale bar: 100 µm. o The severity of synovitis and OA phenotype were assessed through synovial inflammation score and OARSI score system. n  =  5 independent biological animals. MAS, malate-aspartate shuttle; OAA, oxaloacetate. All data were presented as the means ± SD. P values were determined by one-way ANOVA with Tukey’s multiple-comparisons. P < 0.05 was considered statistically significant. Source data are provided as a Source Data file.