EYA1 regulates CIITA phosphorylation to promote EYA1–CIITA–Runx2 complex formation and extracellular matrix integrity in osteoarthritis Authors: Guanghua Nie, Yi Li, Hongmou Zhao, Dong Wang, Yan Zhang, Xinquan Yang, Xiaodong Wen Published: 2026 ABSTRACT AbstractBackgroundOsteoarthritis (OA) involves progressive extracellular matrix (ECM) degradation in articular cartilage. We previously revealed that the m6A demethylase ALKBH5 stabilizes Runx2 mRNA in chondrocytes via YTHDF1, thereby increasing MMP and ADAMTS expression, causing ECM degradation. However, the transcriptional co-regulators of Runx2 remain unidentified.MethodsOA was induced in mice via injection of sodium monoiodoacetate (MIA). The histological architecture and proteoglycan content of the articular cartilage were evaluated using H&E and Safranin O-Fast Green staining, while subchondral bone microarchitecture was assessed by micro-computed tomography. Circulating proinflammatory cytokine levels were determined by ELISA.ResultsImmunoprecipitation and mass spectrometry analyses of articular cartilage from MIA-induced OA mice identified the transcriptional coactivator EYA1 (Eyes Absent Homolog 1) and CIITA (Class II Major Histocompatibility Complex Transactivator) as novel Runx2-interacting proteins. In vitro and in vivo experiments demonstrated that EYA1, CIITA, and Runx2 form a functional EYA1–CIITA–Runx2 complex that binds to the promoters of MMP and ADAMTS genes and activates their expression. Silencing any member of the complex significantly reduced transcription of these matrix-degrading enzymes. Furthermore, we identified EYA1 as a phosphatase that dephosphorylates CIITA at serine 782, a phosphorylation site targeted by Pim3 (provirus integration site for Moloney murine leukemia virus 3) under inflammatory conditions.ConclusionOur findings uncover a previously unrecognized transcriptional mechanism in OA pathogenesis, in which EYA1 promotes ECM degradation by regulating CIITA dephosphorylation and assembling a transcriptional complex with CIITA and Runx2. Targeting this pathway may provide a promising therapeutic strategy for OA.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13062-026-00871-8.Keywords: Osteoarthritis, Extracellular matrix, EYA1, CIITA, Runx2, Pim3 Introduction Osteoarthritis (OA) is the most prevalent form of degenerative joint disease, affecting hundreds of millions of people globally and representing a leading cause of disability, particularly among the elderly [1, 2]. The disease imposes a substantial socioeconomic burden due to its chronic progression and lack of disease-modifying therapies [1, 2]. Pathologically, OA is characterized by a combination of articular cartilage degeneration, synovial inflammation, subchondral bone remodeling, and osteophyte formation [3, 4]. Among these features, the disruption of extracellular matrix (ECM) homeostasis in articular cartilage represents an important pathological event [5, 6]. The ECM, primarily composed of collagen, proteoglycans, and various structural proteins, is essential for maintaining the mechanical properties and integrity of the joint. Its degradation leads to joint instability, pain, and impaired mobility [5–7]. At the cellular level, OA is driven by a complex interplay of mechanical overload, aging, genetic susceptibility, and inflammatory signaling [3]. Chondrocytes—the only cell type residing in articular cartilage—respond to these stressors by undergoing phenotypic changes that impair their reparative capacity and promote catabolic activity [8, 9]. A hallmark of this catabolic shift is the upregulation of matrix-degrading enzymes, particularly matrix metalloproteinases (MMPs) and aggrecanases such as the ADAMTS (a disintegrin and metalloproteinase with thrombospondin motifs) family [8, 9]. While the destructive roles of these enzymes in ECM degradation are well established, the full spectrum of MMP and ADAMTS genes involved, as well as the upstream molecular mechanisms controlling their expression, remain incompletely characterized. Our recent studies using a sodium monoiodoacetate (MIA)-induced OA mouse model identified coordinated upregulation of ALKBH5 (alkylated DNA repair protein AlkB homolog 5), an m6A RNA demethylase, the transcription factor Runx2 (runt-related transcription factor 2), and several ECM-degrading enzymes including MMPs (MMP3/7/9/13) and ADAMTSs (ADAMTS7/10/20) [10]. Mechanistically, ALKBH5 acts in conjunction with YTHDF1 (YTH N6-methyladenosine RNA binding protein 1) to enhance the stability of Runx2 mRNA, thereby sustaining Runx2 protein expression under pathological conditions [10]. Runx2, in turn, promotes transcription of MMP and ADAMTS genes, reinforcing a feed-forward loop of ECM breakdown [10]. Notably, inhibition of ALKBH5 or Runx2 in vivo significantly ameliorated cartilage degradation in MIA-treated mice [10]. These findings uncover a previously unappreciated regulatory pathway linking epitranscriptomic modification to transcriptional control of ECM catabolism and highlight potential therapeutic targets for OA intervention. Transcriptional regulation of gene expression is a highly coordinated process that involves not only DNA-binding transcription factors but also a diverse array of transcriptional cofactors [11, 12]. These regulatory proteins—including transcriptional activators, coactivators, corepressors, and chromatin remodelers—modulate the accessibility of promoter and enhancer regions and facilitate or inhibit the recruitment of RNA polymerase II [11, 12]. Transcription factors typically recognize specific DNA motifs in target gene promoters or enhancers, but their ability to initiate or sustain transcription often depends on interactions with transcriptional coregulators [11, 12]. For instance, coactivators such as histone acetyltransferases p300/CBP (CREB Binding Protein), CIITA (Class II Major Histocompatibility Complex Transactivator), and mediator complex subunits can enhance transcription by promoting histone acetylation and facilitating transcriptional complex assembly, while corepressors can inhibit gene expression through chromatin condensation or recruitment of histone deacetylases [11–13]. The formation of transcriptional complexes is therefore essential for converting signaling cues into precise gene expression patterns during development and disease [11–13]. Although Runx2 has been identified as a pivotal transcription factor that drives the expression of matrix-degrading enzymes, including MMP3/7/9/13 and ADAMTS7/10/20 [10], transcription factors rarely act alone, and the identity of transcriptional coregulators that partner with Runx2 to promote the expression of these catabolic genes in chondrocytes remains largely unknown. Understanding which coactivators or chromatin-associated factors collaborate with Runx2 to activate MMP and ADAMTS gene transcription is critical for elucidating the gene regulatory networks that underlie ECM degradation in OA. Eyes Absent Homolog 1 (EYA1) is a multifunctional protein best known for its dual role as a transcriptional coactivator and phosphatase [14, 15]. Originally identified as a key developmental regulator in Drosophila eye formation, EYA1 is highly conserved across species and plays essential roles in organogenesis, including kidney, ear, and skeletal development [14, 15]. As a coactivator, EYA1 functions by forming complexes with members of the SIX (Sine Oculis Homeobox Homolog) family of transcription factors, facilitating the activation of tissue-specific gene expression [14, 15]. In addition to its transcriptional coactivation role, EYA1 possesses intrinsic phosphatase activity—unusual among transcriptional regulators—which enables it to modulate the phosphorylation status of both nuclear and cytoplasmic substrates [14, 15]. Emerging evidence has implicated EYA1 in various pathological contexts beyond development. Its dysregulation has been associated with several forms of cancer, where it contributes to tumor progression by promoting cell proliferation, survival, and epithelial-to-mesenchymal transition [14, 15]. EYA1’s phosphatase activity, in particular, has been shown to influence key signaling pathways such as DNA damage response and cell cycle control [14, 15]. To elucidate the transcriptional co-regulators that cooperate with Runx2 to control the expression of matrix-degrading enzymes and to better understand the nuclear regulatory networks driving ECM degradation in OA, we investigated the Runx2 protein interactome in articular cartilage. Through immunoprecipitation combined with mass spectrometry in a MIA-induced OA mouse model, we identified CIITA and EYA1 as novel Runx2-interacting partners. Further analysis demonstrated that Runx2, CIITA, and EYA1 assemble into a functional EYA1–CIITA–Runx2 complex, which binds to the promoter regions of MMP and ADAMTS genes to promote their transcription in chondrocytes. We further revealed that the assembly of the EYA1–CIITA–Runx2 complex depends on the phosphatase activity of EYA1. Specifically, EYA1-mediated dephosphorylation of CIITA at Pim3-dependent phosphorylation sites facilitates the formation of the transcriptional complex. These findings uncover a previously unrecognized layer of transcriptional regulation in OA pathogenesis, provide new mechanistic insights into the upstream control of catabolic gene expression, and highlight potential therapeutic targets to suppress or reverse cartilage degradation in osteoarthritis. Materials and methods Animal experiments All animal procedures were conducted in accordance with ethical guidelines and were approved by the Animal Care and Use Committee of Honghui Hospital, Xi’an Jiaotong University (protocol number 2021–069). Eight-week-old C57BL/6 mice (22–25 g) were obtained from an accredited supplier (Beijing Vital River Laboratory Animal Technology Co., Ltd., Beijing, China) and housed in a specific pathogen-free (SPF) facility under controlled environmental conditions (12-hour light/dark cycle, temperature: 18–23 °C, humidity: 40–60%). Mice were given unrestricted access to standard chow and tap water throughout the experimental period. To establish an OA mouse model, animals were randomly assigned to experimental or control groups (n = 8 per group). The sample size was determined based on our preliminary data and previous studies in OA mouse models, where n = 6–10 mice per group are commonly used [16–18]. An a priori power analysis was performed to ensure adequate statistical power. With a sample size of n = 8 per group, the study achieves a statistical power (1 − β) of > 80% to detect an effect size (Cohen’s d) of 1.5 at a significance level (α) of 0.05. Mice were anesthetized by intraperitoneal injection of ketamine (80 mg/kg) and xylazine (10 mg/kg) prior to intra-articular injection. The right knee was sterilized with 70% ethanol, and a single 50 µL injection of 2 mg/mL monosodium iodoacetate (MIA, Sigma-Aldrich, #I2512) was administered into the joint cavity using a 30-gauge needle (Jensen Global Inc, #JG301.0X), targeting the space between the patella and femoral condyles. Sham-operated mice received an equal volume of sterile saline. Animals were monitored postoperatively until full recovery and were observed daily for signs of pain or distress. To evaluate the therapeutic potential of pharmacologic inhibitors, additional groups of mice received intra-articular MIA injections followed by intraperitoneal administration of either DS-1-38 (an EYA1 phosphatase inhibitor, MedChemExpress, #HY-158059) or CADD522 (a Runx2 inhibitor, Selleckchem, #S0790), each at a dose of 10 mg/kg every 5 days. The dosing regimen was determined based on previous studies and our preliminary experiments. Briefly, higher doses of DS-1-38 (20–40 mg/kg), as reported in prior studies [19], induced noticeable adverse effects under our experimental conditions, including reduced food intake, irritability, and body weight loss; therefore, a reduced dose of 10 mg/kg was selected, which showed no apparent toxicity while maintaining inhibitory efficacy on EYA1. For CADD522, the dosage of 10 mg/kg was determined with reference to the study by Kim et al. (2017) [20], which characterized CADD522 as a specific inhibitor of RUNX2-DNA binding and demonstrated its in vivo antitumor activity at similar pharmacological concentrations. Our comparative pilot study confirmed that 10 mg/kg of CADD522 was sufficient to suppress RUNX2 activity in joint tissues while maintaining a superior safety profile. Mice were assigned to one of four treatment groups (n = 8 per group): sham (saline), OA (MIA only), MIA + DS-1-38, and MIA + CADD522. After 21 days of treatment, mice were euthanized using CO₂ inhalation (60% of cage volume per minute), followed by cervical dislocation to ensure complete sacrifice. Joint tissues were subsequently harvested for downstream histological and molecular analyses. Immunoprecipitation and mass spectrometry To identify Runx2-interacting proteins in articular cartilage, cartilage tissues were isolated from the femoral condyles and tibial plateaus of MIA mice. Three independent biological experiments were performed, each using cartilage tissues pooled from three mice (n = 3 per experiment) with equal weights. Tissues were minced and lysed in IP lysis buffer (Thermo Scientific; #87787) supplemented with protease and phosphatase inhibitor cocktail (Cell Signaling; #5872S) on ice for 30 min. Lysates were clarified by centrifugation at 14,000 × g for 15 min at 4 °C. The supernatants were pre-cleared with normal IgG (Table S1) and protein A/G agarose beads (Thermo Scientific; #20421) for 1 h at 4 °C with gentle rotation. Immunoprecipitation was performed by incubating pre-cleared lysates with anti-Runx2 antibody (Table S1) overnight at 4 °C, followed by incubation with protein A/G agarose beads for an additional 2 h. Beads were washed extensively with IP lysis buffer and eluted with SDS sample buffer (Bio-Rad Laboratories; #1610747). Eluted proteins were separated by 12% SDS-PAGE and visualized using a silver staining kit (Thermo Scientific; #24600). Protein bands were excised and subjected to in-gel digestion with sequencing-grade modified trypsin (Promega; #V5111) overnight at 37 °C. The resulting peptides were analyzed by liquid chromatography–tandem mass spectrometry (LC–MS/MS) using an Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Fisher Scientific) coupled to an EASY-nLC 1200 system. Raw mass spectrometry data were searched against the UniProt mouse database using the Mascot search engine. The following stringent filtering criteria were applied to identify high-confidence Runx2-interacting candidates: (1) a False Discovery Rate (FDR) threshold of < 1% at both peptide and protein levels; (2) a minimum requirement of ≥2 unique peptides for each identified protein; and (3) a Mascot ion score cutoff of P < 0.05. Proteins frequently identified as non-specific contaminants were excluded. Only proteins consistently enriched in the anti-Runx2 group compared to the IgG control across all three biological replicates were considered for further analysis. Cell culture and transfection The two distinct immortalized mouse articular chondrocyte (iMAC) cell lines, iMAC1 and iMAC2, were obtained from the same sources as previously reported [10]. Cells were cultured in DMEM/F-12 medium (Gibco, #11330–032) supplemented with 10% fetal bovine serum (FBS; Gibco, #26140–079) and 1% penicillin–streptomycin (Gibco, #15140–122) at 37 °C in a humidified atmosphere containing 5% CO₂. Cell transfection was conducted following the protocol described in the previous study [10], using the shRNAs and plasmids listed in Tables S2 and S3. After transfection, cells were maintained in fresh complete medium for an additional 48 h before being subjected to further analysis. Transfection efficiency and gene modulation were confirmed by RT-qPCR or Western blotting to verify successful knockdown or overexpression of the target gene. Cell treatments Chondrocytes were seeded in DMEM/F-12 medium and allowed to grow until they reached approximately 70–80% confluence. To assess the cellular response to inflammatory stimuli, cells were exposed to recombinant IL-1β (Thermo Scientific, #PMC0811) at final concentrations of 0, 1, 2, 4 or 8 ng/mL or TNF-α (Thermo Scientific, #RMTNFAI) at final concentrations of 0, 10, 20, or 40 ng/mL for 12 h. For co-treatment assays, cells were simultaneously treated with either IL-1β (8 ng/mL) or TNF-α (40 ng/mL) in combination with DS-1-38 at 15, 30 or 60 µM for the same duration. After treatment, cells were rinsed three times with cold PBS to remove residual compounds. Total RNA and protein were then extracted for subsequent analysis of gene and protein expression. Histological staining and OARSI histopathology scoring After euthanasia, right knee joints from sham and MIA-treated mice were collected, fixed in 10% neutral-buffered formalin for 48 h, and decalcified in 10% EDTA for 2–4 weeks with solution changes every 3 days. Tissues were then dehydrated, embedded in paraffin, and sectioned at 5 μm using a rotary microtome (Leica; #RM2235). Sections were mounted on glass slides for subsequent histological analyses. For hematoxylin and eosin (H&E) staining, sections were deparaffinized in xylene and rehydrated through a graded ethanol series (100%, 95%, and 70%). Slides were stained with hematoxylin for 5 min, rinsed in running tap water, and differentiated in 1% acid alcohol (1% HCl in 70% ethanol), followed by bluing in Scott’s tap water substitute. After rinsing, sections were counterstained with eosin (Sigma-Aldrich; #HT110132) for 2 min, dehydrated through ascending ethanol concentrations, cleared in xylene, and mounted with a permanent mounting medium (Thermo Scientific; #9990402). For Safranin O–Fast Green staining, adjacent sections were deparaffinized and rehydrated as described above, followed by staining with Fast Green (from Safranine O-Fast Green FCF Cartilage Stain Kit; Creative Bioarray; #FDIR-D0158) to visualize non-cartilaginous tissues and counterstaining with Safranin O to assess proteoglycan content in articular cartilage. Stained sections were imaged under a light microscope (Leica; #Visoria P) using identical acquisition settings across all groups. Cartilage degeneration was quantitatively evaluated using the Osteoarthritis Research Society International (OARSI) scoring system as previously described [21]. OARSI scoring was performed in a double-blinded manner by two independent observers who were unaware of the experimental grouping and treatment allocations, ensuring objectivity and minimizing bias. Micro-computed tomography (Micro-CT) analysis Subchondral bone in the mouse tibia was evaluated using a high-resolution Micro-CT system (Micro Photonics Inc., #Skyscan-1276) in accordance with established procedures. Mice were anesthetized with an intraperitoneal injection of ketamine (80 mg/kg) and xylazine (10 mg/kg), then placed in a supine position on the scanning platform. To minimize radiation exposure, the knees treated with MIA or saline were imaged individually. Scans were conducted at 50 kV and 200 µA with an isotropic voxel size of 9 μm. Image reconstruction was performed using NRecon software (version 1.6, Micro Photonics Inc.). Enzyme-linked immunosorbent assay (ELISA) Whole blood was obtained by cardiac puncture under anesthesia with ketamine and xylazine. Samples were left at room temperature for 30 min to allow clotting, followed by centrifugation at 2000 g for 10 min to isolate serum. The supernatant was aliquoted and stored at − 80 °C until further use. Serum concentrations of IL-1β, IL-6, and TNF-α were quantified using commercial ELISA kits specific for each cytokine (R&D Systems: IL-1β, #MLB00C; IL-6, #M6000B; TNF-α, #MTA00B), following the protocols provided by the manufacturer. Cytokine levels were determined by comparing sample absorbance values against standard curves constructed from known concentrations. RNA extraction and RT-qPCR analysis Cells were rinsed with PBS and lysed using TRIzol reagent (Invitrogen; #15596026). RNA was pelleted by centrifugation (12,000 g, 10 min, 4 °C), washed with 75% ethanol, recentrifuged (7,500 g, 5 min, 4 °C), air-dried, and resuspended in RNase-free water. RNA quality and concentration were assessed using a NanoDrop spectrophotometer (Thermo Scientific). For cDNA synthesis, 1 µg of total RNA was reverse transcribed using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems; #4368814) following the manufacturer’s protocol. Quantitative PCR was conducted using PowerUp SYBR Green Master Mix (Applied Biosystems; #A25742) on a QuantStudio 3 Real-Time PCR System. Each 20 µL PCR mixture contained 1 µL of cDNA, 10 µL of SYBR mix, and 0.5 µM of forward and reverse primers (Table S4). Gene expression was normalized to β-actin, and relative expression levels were calculated using the 2−ΔΔCt method. Protein extraction and western blot analysis Chondrocytes were rinsed with ice-cold PBS and lysed in culture dishes using RIPA buffer (Thermo Scientific; #89900), supplemented with protease and phosphatase inhibitor cocktails (Roche; #11836170001). Lysates were incubated on ice for 30 min with intermittent vortexing, followed by centrifugation at 12,000 g for 15 min at 4 °C to collect the protein-containing supernatant. Protein concentration was quantified using the BCA assay (Thermo Scientific; #23225). For immunoblotting, 50 µg of protein per sample was mixed with 4 × Laemmli buffer (Bio-Rad; #1610747) containing 10% β-mercaptoethanol (Sigma-Aldrich; #M6250), heated at 95 °C for 5 min, and separated on 10% SDS-PAGE gels. Proteins were transferred to PVDF membranes using a wet transfer system. Membranes were blocked with 5% non-fat milk in TBST (Tris-buffered saline with 0.1% Tween-20) for 1 h at room temperature and then incubated overnight at 4 °C with specific primary antibodies (Table S1) diluted in blocking buffer. After washing, membranes were incubated with HRP-conjugated secondary antibodies (Table S1) for 1 h at room temperature. Detection was performed using enhanced chemiluminescence (ECL) reagents (GE Healthcare; #RPN2232), and signal was captured using a Bio-Rad ChemiDoc MP imaging system. Sequential immunoprecipitation assay To determine whether EYA1, CIITA, and Runx2 are present within the same endogenous protein complex, a sequential immunoprecipitation assay was performed under non-denaturing conditions. Briefly, articular cartilage tissues from OA mice (n = 3) were pooled in equal amounts (0.1 g per sample) and lysed in ice-cold IP lysis buffer supplemented with protease inhibitors. The lysates were clarified by centrifugation to remove debris. Equal amounts of total protein were first incubated overnight at 4 °C with either normal IgG (negative control) or an anti-Runx2 antibody, followed by incubation with protein A/G agarose beads for 4 h at 4 °C. After extensive washing with IP wash buffer, the bound protein complexes were gently eluted under native conditions using low-pH glycine elution buffer (0.1 M glycine-HCl, pH 2.8), and the eluates were immediately neutralized with 1 M Tris-HCl (pH 8.0). The neutralized eluates were then subjected to a second round of immunoprecipitation with an anti-EYA1 antibody overnight at 4 °C, followed by incubation with fresh protein A/G agarose beads. After washing, the second-round immunoprecipitates were eluted in SDS sample buffer and analyzed by immunoblotting using antibodies against EYA1, CIITA, and Runx2. Normal IgG controls were included in parallel throughout the procedure to assess specificity. Chromatin immunoprecipitation (ChIP) and sequential ChIP (Re-ChIP) assays ChIP experiments were performed as previously described [10] with minor modifications. Briefly, cells were crosslinked with 1% formaldehyde (Sigma-Aldrich; #F8775) for 10 min at room temperature and quenched with 125 mM glycine for 5 min. After washing with ice-cold PBS, cells were lysed in ChIP lysis buffer (Santa Cruz Biotechnology; #sc-45000) supplemented with protease inhibitors. Chromatin was sonicated to yield DNA fragments of approximately 200–500 bp. The sheared chromatin was pre-cleared with protein A/G agarose beads and then incubated overnight at 4 °C with specific antibodies (anti-Runx2, anti-EYA1, or anti-CIITA) or normal IgG (Table S1). Immune complexes were captured using protein A/G agarose beads, washed with ChIP wash buffer (Santa Cruz Biotechnology; #sc-45002), and eluted in buffer containing 1% SDS and 0.1 M NaHCO₃. Crosslinks were reversed at 65 °C for 4 h, followed by proteinase K digestion (Thermo Scientific; #EO0491). DNA was purified by phenol–chloroform extraction, precipitated with ethanol, and resuspended in nuclease-free water. Quantitative PCR (qPCR) was performed using target-specific primers (Table S5), and enrichment was normalized to input DNA. For Re-ChIP assays, chromatin complexes obtained from the first immunoprecipitation with anti-EYA1 were eluted under mild conditions using DTT-containing elution buffer (1% SDS, 100 mM NaHCO₃, 10 mM DTT) at 37 °C for 30 min to preserve protein–DNA and protein–protein interactions. The eluates were then diluted and subjected to a second immunoprecipitation with a different antibody (anti-EYA1, anti-CIITA or anti-Runx2). The second-round immunoprecipitates were processed as described above, including washing, reverse crosslinking, DNA purification, and qPCR analysis. IgG controls were included to assess specificity. Re-ChIP signals were normalized to input DNA. Vector construction The coding sequences of murine EYA1, CIITA, and Runx2 were amplified by PCR using Phusion High-Fidelity DNA Polymerase (New England Biolabs, #M0530) and cloned into pCDNA3-Flag empty vector at the BamHI and XhoI restriction sites, or into pCDNA3-MYC empty vector at the KpnI and XhoI sites, to generate Flag- or MYC-tagged expression constructs. The coding sequences of EYA1 and Pim3 were inserted into the pGEX-6P-1 vector using BamHI and EcoRI sites to generate GST-tagged fusion proteins. The full-length CIITA (1–1078 aa) and its truncated forms — CIITAa (1–350 aa), CIITAb (351–700 aa), CIITAc (701–1078 aa), CIITAc1 (701–826 aa), CIITAc2 (827–952 aa), and CIITAc3 (953–1078 aa) — were PCR-amplified and cloned into the pET28a vector at BamHI and XhoI restriction sites to generate His-tagged constructs for protein purification. Site-directed mutagenesis was performed using the QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent Technologies, #210518) to introduce point mutations at ten serine/threonine sites within the CIITAc region, generating the following mutants: T738A, T748A, T764A, S782A, T786A, T802A, S804A, S808A, T814A, and S819A. All constructs were verified by Sanger sequencing (BGI Genomics, China) to ensure sequence fidelity. Protein expression and purification in E. coli All GST- and His-tagged plasmids were transformed into E.coli strain BL21(DE3) (TransGen Biotech, #CD601), and positive colonies were selected by kanamycin (for pET28a) or ampicillin (for pGEX-6P-1) resistance. Bacterial cultures were grown in LB medium at 37 °C until OD₆₀₀ reached 0.6–0.8, then induced with 0.4 mM IPTG (Sigma-Aldrich, #I6758) for 16 h at 18 °C. For GST-tagged proteins, the pellets were resuspended in GST lysis buffer consisting of 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1 mM DTT, 1 mM PMSF, and optionally 0.1% Triton X-100 and 10% glycerol. Cells were lysed by sonication on ice and centrifuged at 15,000 × g for 30 min at 4 °C. The supernatant was incubated with Glutathione Sepharose 4B beads (GE Healthcare, #17-0756-01) for 2 h at 4 °C, followed by extensive washing and elution with 10 mM reduced glutathione in 50 mM Tris-HCl (pH 8.0). For His-tagged proteins, pellets were lysed in His lysis buffer containing 50 mM NaH₂PO₄ or Tris-HCl (pH 8.0), 300 mM NaCl, 10 mM imidazole, 1 mM PMSF, and optionally 0.1% Triton X-100 and 1 mM β-mercaptoethanol. After sonication and centrifugation, clarified lysates were incubated with Ni-NTA agarose beads (Qiagen, #30210) for 2 h at 4 °C. After washing with buffer containing 20 mM imidazole, proteins were eluted with buffer containing 250 mM imidazole. Eluted proteins were dialyzed against PBS or appropriate storage buffer and stored at − 80 °C for further use. Protein purity was assessed by SDS-PAGE and quantified using a BCA protein assay kit (Thermo Scientific, #23227). In vitro kinase assay To assess Pim3-mediated phosphorylation of CIITA and its mutants, in vitro kinase assays were performed using [γ-32P] ATP. Briefly, recombinant GST-tagged Pim3 protein and His-tagged wild-type or mutant CIITA proteins were incubated in kinase reaction buffer containing 20 mM Tris-HCl (pH 7.5), 10 mM MgCl₂, 1 mM DTT, and 10 µCi [γ-32P] ATP in a total volume of 30 µL. The reactions were carried out at 30 °C for 30 min and terminated by adding 6 × SDS loading buffer followed by boiling for 5 min. Samples were separated by SDS-PAGE, and the gels were stained with Coomassie Brilliant Blue to visualize total protein. After staining, the gels were dried and subjected to autoradiography. Statistical analysis Data are presented as mean ± standard deviation (SD). Statistical evaluation was conducted using SPSS software (version 26, IBM, USA). For comparisons involving more than two groups, one-way analysis of variance (ANOVA) was applied, followed by Tukey’s post hoc test to assess pairwise differences. When comparing two groups, an unpaired Student’s t-test was used. A P-value less than 0.05 was considered indicative of statistical significance. Sample sizes were determined based on preliminary results and power calculations to ensure sufficient sensitivity for detecting meaningful differences. Each experiment was independently repeated at least three times to validate reproducibility. Results Runx2 formed a transcriptional complex with CIITA and EYA1 in vivo and in vitro As previously described, our recent study demonstrated that in an inflammatory microenvironment, Alkbh5 and Ythdf1 stabilize Runx2 mRNA through m6A methylation. Runx2, in turn, activates the transcription of matrix-degrading enzymes such as MMPs and ADAMTSs, thereby promoting ECM degradation in chondrocytes and contributing to the progression of OA [10]. However, it remains unclear which transcriptional regulatory proteins interact with Runx2 under inflammatory conditions to cooperatively regulate the expression of these catabolic genes. To address this question, we performed IP assays using cartilage tissue from a previously established MIA-induced OA mouse model (n = 3, pooled in equal amounts). Total protein was extracted and subjected to IP using protein A/G beads conjugated with either control IgG or anti-Runx2 antibody. The IP products were resolved by SDS-PAGE, visualized by silver staining (Fig. 1A), trypsin digestion, and analyzed by mass spectrometry. A total of 69 Runx2-interacting proteins were identified (Table S6), among which two known transcriptional regulators, EYA1 and CIITA, were significantly enriched (Fig. 1B). EYA1 is a transcriptional coactivator with intrinsic phosphatase activity, while CIITA functions as a master regulator of class II major histocompatibility complex gene expression. Runx2 recruited CIITA and EYA1 to form a transcriptional complex in vivo and in vitro. (A) Silver staining of immunoprecipitates obtained using IgG or anti-Runx2-conjugated protein A/G beads from pooled articular cartilage samples (n = 3, equal weight per sample) of MIA-induced OA mice. (B) Mass spectrometry analysis identifying proteins highly enriched in the Runx2 immunoprecipitates. (C–E) Immunoprecipitation assays performed using anti-Runx2 (C), anti-EYA1 (D), or anti-CIITA (E) antibodies in articular cartilage samples from MIA-induced OA mice (n = 3, pooled equally), followed by immunoblotting to detect associated proteins. (F) Sequential immunoprecipitation assay. Protein complexes were first immunoprecipitated using an anti-Runx2 antibody, followed by a second round of IP with an anti-EYA1 antibody. Western blot analysis of the second-round immunoprecipitates confirmed the co-presence of Runx2, EYA1, and CIITA. (G) Co-IP analysis indicating that Runx2 directly interacted with CIITA but not with EYA1. (H) Co-IP showing that CIITA directly interacted with both Runx2 and EYA1 To determine whether Runx2, EYA1, and CIITA form a complex in vivo, we performed IP assays using an independent set of cartilage samples (n = 3, pooled equally). Protein lysates were subjected to immunoprecipitation with IgG, anti-Runx2, anti-EYA1, or anti-CIITA antibodies. Western blot analysis revealed the co-presence of Runx2, EYA1, and CIITA in the immunoprecipitates obtained with anti-Runx2, anti-EYA1, and anti-CIITA antibodies (Fig. 1C and E), suggesting that these proteins are associated in vivo. To further strengthen this observation, we performed sequential immunoprecipitation assays. Specifically, protein complexes were first immunoprecipitated using an anti-Runx2 antibody, followed by a second round of immunoprecipitation with an anti-EYA1 antibody. Western blot analysis of the second-round immunoprecipitates confirmed the co-presence of Runx2, EYA1, and CIITA (Fig. 1F), providing additional evidence that these three proteins exist within the same complex in vivo. To further determine whether this complex also forms in vitro and to elucidate the interaction pattern among the three proteins, we constructed expression vectors containing either Flag or MYC tags for Runx2, EYA1, and CIITA. Co-transfection of Flag-Runx2 with MYC-CIITA or MYC-EYA1 into iMAC1 cells followed by IP using anti-Flag beads showed that Runx2 directly interacts with CIITA but not with EYA1 (Fig. 1G). Similarly, co-transfection of Flag-CIITA with MYC-EYA1 or MYC-Runx2 revealed that CIITA interacts directly with both EYA1 and Runx2 (Fig. 1H). These findings suggest that CIITA may serve as a bridging factor that links EYA1 and Runx2, facilitating the formation of the EYA1–CIITA–Runx2 transcriptional complex. Knockdown of EYA1–CIITA–Runx2 members caused the downregulation of MMPs and ADAMTSs Our previous study demonstrated that Runx2 regulates the transcription of MMP3, MMP10, MMP12, MMP16, ADAMTS8, ADAMTS10, and ADAMTS20 [10]. To further elucidate the roles of EYA1 and CIITA in this regulatory network, we investigated how their knockdown affects the expression of these matrix-degrading genes. Using the immortalized murine chondrocyte cell line iMAC1, we generated four groups: Control knockdown (ControlKD1/2), EYA1 knockdown (EYA1KD1/2), CIITA knockdown (CIITAKD1/2), and Runx2 knockdown (Runx2KD1/2) (Fig. 2A and B). RT-qPCR analysis revealed that silencing EYA1, CIITA, or Runx2 led to a marked reduction in the expression ofMMP3, MMP10, MMP12, MMP16, ADAMTS8, ADAMTS10, and ADAMTS20. In contrast, the expression of MMP8, ADAMTS19, and GBA2—genes whose promoters lack Runx2 binding motifs—remained unaffected. These findings suggest that, similar to Runx2, EYA1 and CIITA are involved in the transcriptional regulation of a subset of matrix-degrading enzymes in chondrocytes. Knockdown of EYA1, CIITA, or Runx2 reduced mRNA expression of MMP3/10/12/16 andADAMTS8/10/20. (A and B) Quantification of mRNA (A) and protein (B) levels of EYA1, CIITA, and Runx2 in ControlKD1/2, EYA1KD1/2, CIITAKD1/2, and Runx2KD1/2 (iMAC1 cell background). (C) Relative mRNA levels of MMP3, MMP10, MMP12, MMP16, and MMP8 in ControlKD1/2, EYA1KD1/2, CIITAKD1/2, and Runx2KD1/2 cells. (D) Relative mRNA levels of ADAMTS8, ADAMTS10, ADAMTS20, ADAMTS19, and GBA2 in the same knockdown cell lines as in (A). All PCR experiments were performed in triplicate, with each experiment conducted using three independent biological samples. ***P < 0.001 The EYA1–CIITA–Runx2 transcriptional complex bound to the promoter of MMP3/10/12/16and ADAMTS8/10/20 to regulate their expression We next investigated whether EYA1 and CIITA are also recruited to the promoters of MMP and ADAMTS genes. ChIP assays were performed in ControlKD1/2, EYA1KD1/2, CIITAKD1/2, and Runx2KD1/2 cells using protein A/G beads conjugated with IgG, anti-EYA1, anti-CIITA, or anti-Runx2 antibodies. ChIP-qPCR results showed that, similar to Runx2, both EYA1 and CIITA bound to the promoter regions of MMP3, MMP10, MMP12, MMP16, ADAMTS8, ADAMTS10, and ADAMTS20 (Figures S1 and S2). However, the EYA1-CIITA-Runx2 complex could not bind to the promoters of MMP8, ADAMTS19, and GBA2 (Figures S2 and S3). Notably, knockdown of any one of the three components—EYA1, CIITA, or Runx2—resulted in a marked reduction in the binding of the remaining two factors to these promoter regions (Figures S1 and S2). To further confirm the endogenous formation of a functional EYA1–CIITA–Runx2 complex and its direct involvement in transcriptional regulation, we performed Re-ChIP assays. A first round of ChIP was conducted using an anti-EYA1 antibody, followed by a second round of immunoprecipitation with IgG, anti-EYA1, anti-CIITA, or anti-Runx2 antibodies. Focusing on representative target genes (MMP3, ADAMTS8, and MMP8), Re-ChIP–qPCR analysis demonstrated that EYA1, CIITA, and Runx2 co-occupied the promoter regions of MMP3 and ADAMTS8, but not MMP8 (Figures S4). Consistent with the single-round ChIP results in Figures S1 and S2, depletion of any one component significantly impaired the recruitment of the remaining factors to promoters of MMP3 and ADAMTS8 (Figures S4). These findings suggest that EYA1, CIITA, and Runx2 cooperatively form a transcriptional complex that binds to the promoters of MMP3, MMP10, MMP12, MMP16, ADAMTS8, ADAMTS10, and ADAMTS20, thereby regulating their expression. Proinflammatory cytokines dose-dependently induced the EYA1–CIITA–Runx2 complex, and EYA1 silencing altered CIITA Our previous study showed that treatment with the proinflammatory cytokines IL-1β and TNF-α upregulates Runx2 expression in iMAC1 and iMAC2 cells in a dose-dependent manner [10]. To determine whether these cytokines also influence the expression of EYA1 and CIITA, we treated iMAC1/2 cells with increasing concentrations of IL-1β (0, 2, 4, and 8 ng/mL) or TNF-α (0, 10, 20, and 40 ng/mL). Western blot analysis revealed that both cytokines dose-dependently increased the protein levels of EYA1, CIITA, and RUNX2 (Fig. 3A and B). Given that EYA1 also possesses phosphatase activity, we next examined whether it modulates the post-translational modification of CIITA or RUNX2. To this end, we treated ControlKD1/2 and EYA1KD1/2 cells with IL-1β (8 ng/mL) for 12 h. Protein analysis showed that EYA1 knockdown significantly impaired the cytokine-induced upregulation of CIITA, while Runx2 protein levels remained unaffected (Fig. 3C and D). Interestingly, in EYA1-depleted cells treated with IL-1β, we observed a shift in the electrophoretic mobility of CIITA. In ControlKD cells, CIITA appeared as a single band, whereas in EYA1KD cells under the same treatment, CIITA displayed two distinct bands, suggesting the presence of a modified form (Fig. 3C and D). Knockdown or pharmacological inhibition of EYA1 altered CIITA modification. (A and B) Dose-dependent effects of IL-1β (0, 2, 4, and 8 ng/mL; A) and TNF-α (0, 10, 20, and 40 ng/mL, B) on the protein levels of EYA1, CIITA, and Runx2 in iMAC1 and iMAC2 cells. (C and D) Effects of EYA1 knockdown combined with 8 ng/mL IL-1β treatment on the protein expression of EYA1, CIITA, and Runx2 in iMAC1 (C) and iMAC2 (D) cells. (E and F) Effects of co-treatment with DS-1-38 (0, 15, 30, and 60 µM) and 8 ng/mL IL-1β on the protein levels of EYA1, CIITA, and Runx2 in iMAC1 (E) and iMAC2 (F) cells. All immunoblot experiments were performed in triplicate To validate this observation, we treated iMAC1/2 cells with various concentrations of the EYA1 phosphatase inhibitor DS-1-38 (0, 15, 30, and 60 µM) in combination with 8 ng/mL IL-1β for 12 h. Western blotting confirmed that DS-1-38 did not affect Runx2 levels but dose-dependently increased the presence of the modified CIITA band (Fig. 3E and F). These findings suggest that either silencing or pharmacological inhibition of EYA1 alters the post-translational modification pattern of CIITA. Given EYA1’s known phosphatase activity, we speculate that CIITA undergoes phosphorylation, and its phosphorylation state is dynamically regulated by EYA1 and one or more kinases in the inflammatory microenvironment. In the absence or inhibition of EYA1, CIITA accumulates in a phosphorylated form, resulting in the appearance of two distinct bands. Recombinant EYA1 reversed CIITA phosphorylation and CIITA interacted with Pim3 in EYA1-depleted cells To further validate our hypothesis that the modified form of CIITA observed in IL-1β-treated EYA1KD cells is due to phosphorylation, we purified recombinant GST-tagged EYA1 protein and performed an in vitro dephosphorylation assay. Protein lysates were prepared from iMAC1/2 cells treated with 60 µM DS-1-38 and 8 ng/mL IL-1β, and subsequently incubated with GST-EYA1 for 1 h. Western blot analysis showed that GST-EYA1 effectively eliminated the modified CIITA band induced by DS-1-38 and IL-1β treatment. In cells cotreated with GST-EYA1, DS-1-38, and IL-1β, the CIITA protein exhibited a single band pattern comparable to that observed in cells treated with IL-1β alone (Fig. 4A and B). Importantly, the expression level of Runx2 remained unaffected by GST-EYA1 treatment (Fig. 4A and B). These results support the hypothesis that CIITA undergoes phosphorylation in IL-1β-treated EYA1-deficient cells. Exogenous GST-EYA1 reversed CIITA modification and CIITA interacted with Pim3 in IL-1β-treated EYA1KDcells. (A and B) GST-EYA1 supplementation restored CIITA modification. Cell lysates from iMAC1 (A) and iMAC2 (B) cells treated with DS-1-38 (60 µM) and IL-1β (8 ng/mL) were incubated with 40 µg GST-EYA1 for 1 h. Protein levels of EYA1, CIITA, and Runx2 were then examined by immunoblotting. (C) Silver staining of immunoprecipitates obtained using IgG or anti-CIITA-conjugated protein A/G beads from IL-1β-treated EYA1KD cells. (D) Mass spectrometry analysis of CIITA immunoprecipitates identified proteins enriched in IL-1β-treated EYA1KD cells. (E and F) Immunoprecipitation showed that CIITA pulls down Pim3 (E) and Pim3 reciprocally pulled down CIITA (F) in IL-1β-treated EYA1KD cells. All experiments shown in this figure were independently performed at least three times with consistent results To identify the kinase responsible for CIITA phosphorylation, we conducted immunoprecipitation using anti-CIITA antibody-conjugated protein A/G beads in IL-1β-treated EYA1KD cells. The immunoprecipitated protein complexes were resolved by SDS-PAGE, visualized by silver staining (Fig. 4C), digested with trypsin, and analyzed by mass spectrometry. A total of 82 CIITA-interacting proteins were identified (Table S7). Among the top enriched candidates were Pim3, YARS1, ARG1, Runx2, and STXBP2 (Fig. 4D). Notably, Pim3 was the only protein kinase identified among these interacting partners (Table S7). To further confirm the interaction between CIITA and Pim3, reciprocal co-immunoprecipitation was performed using anti-CIITA and anti-Pim3 antibodies in IL-1β-treated EYA1KD cells. Immunoblot analysis revealed that CIITA and Pim3 co-precipitated with each other (Fig. 4E and F), indicating a physical interaction. These findings suggest that Pim3 is the likely kinase responsible for CIITA phosphorylation in IL-1β-treated EYA1-deficient chondrocytes. Pim3 phosphorylates CIITA at serine 782 To investigate the phosphorylation of CIITA by Pim3 and identify the specific phosphorylation site(s), we first divided the full-length CIITA protein (1–1078 amino acids) into three fragments: region a (1–350 aa), region b (351–700 aa), and region c (701–1078 aa) (Fig. 5A). We then purified His-tagged versions of full-length CIITA (His-CIITAWT), as well as His-CIITAa, His-CIITAb, His-CIITAc, along with GST-tagged Pim3. In vitro kinase assays showed that Pim3 phosphorylated both His-CIITAWT and His-CIITAc, but not the His-CIITAa or His-CIITAb fragments (Fig. 5B). To further narrow down the phosphorylation site(s), we subdivided region c into three smaller fragments: c1 (701–826 aa), c2 (827–952 aa), and c3 (953–1078 aa) (Fig. 5C), and purified His-CIITAc1, His-CIITAc2, and His-CIITAc3 for additional assays. The results revealed that Pim3 phosphorylated CIITAc1, but not CIITAc2 or CIITAc3 (Fig. 5D). Pim3 phosphorylated CIITA at serine 782 and EYA1 removed Pim3-mediated phosphorylation. (A) Schematic representation of three regions within the CIITA protein: a (amino acids 1–350), b (351–700), and c (701–1078). (B) GST-Pim3 phosphorylated His-CIITAWT and the His-CIITAc. Top: Coomassie Brilliant Blue (CBB) staining; Bottom: Autoradiography to detect phosphorylation. (C) Schematic diagram of three subregions within the CIITAc region: c1 (701–826), c2 (827–952), and c3 (953–1078). (D) GST-Pim3 phosphorylated His-CIITAc and His-CIITAc1. Top: CBB staining; Bottom: Autoradiography. (E) GST-Pim3 failed to phosphorylate the His-CIITAS782A mutant. In vitro kinase assays were performed using GST-Pim3 and a series of CIITA point mutants (T738A, T748A, T764A, S782A, T786A, T802A, S804A, S808A, T814A, S819A). Top: CBB staining; Bottom: Autoradiography. (F) GST-EYA1 dose-dependently removed Pim3-mediated phosphorylation of CIITA. His-CIITAWT and GST-Pim3 were incubated with increasing amounts of GST-EYA1 (0, 15, 30, 45 µg). Left: CBB staining; Right: Autoradiography. All experiments shown in this figure were independently performed at least three times with consistent results To pinpoint the exact phosphorylation site, we individually mutated all serine (S) and threonine (T) residues within the c1 region to alanine (A): T738A, T748A, T764A, S782A, T786A, T802A, S804A, S808A, T814A, and S819A. Among these mutants, only the S782A substitution abolished Pim3-mediated phosphorylation, while mutations at the other sites had no effect (Fig. 5E). These findings indicate that Pim3 specifically phosphorylates CIITA at serine 782 (S782). We next tested whether EYA1 could dephosphorylate CIITA in a dose-dependent manner. In vitro dephosphorylation assays were performed by incubating His-CIITAWT and GST-Pim3 with increasing amounts of GST-EYA1 (0, 15, 30, and 45 µg). The results demonstrated a progressive decrease in CIITA phosphorylation as EYA1 concentration increased, and phosphorylation was completely abolished at 45 µg of GST-EYA1 (Fig. 5F). These results confirm that CIITA is phosphorylated by Pim3 at serine 782 and that this modification is counteracted by EYA1-mediated dephosphorylation. Phosphorylation of CIITA at S782 regulated its interaction with EYA1 and assembly of the EYA1–CIITA–Runx2 transcriptional complex To determine whether phosphorylation of CIITA at S782 regulates its interaction with EYA1 and the assembly of the transcriptional complex, we generated Flag-tagged CIITA constructs, including wild-type (WT), a non-phosphorylatable mutant (S782A), a phosphomimetic mutant (S782D), and control mutants at a non-functional site (T786A and T786D). These constructs were expressed in ControlKD1, EYA1KD1, and Pim3KD1 cells (Figure S5). Immunoprecipitation assays were performed to assess interactions with EYA1 and Runx2, and ChIP assays were used to evaluate co-occupancy of the EYA1–CIITA–Runx2 complex at the promoters of representative target genes (MMP3 and ADAMTS8). Western blot analysis of Flag-tagged CIITA mutants revealed distinct electrophoretic mobility patterns consistent with phosphorylation status. In IL-1β–stimulated ControlKD1 cells, the CIITAS782D mutant exhibited a single upward-shifted band that co-migrated with the phosphorylated form of CIITAWT. In IL-1β–treated EYA1KD1 cells, CIITAWT appeared as two bands, corresponding to phosphorylated and non-phosphorylated forms, whereas the CIITAS782A mutant, as expected, lacked the phosphorylated band (Fig. 6A). Notably, the CIITAS782D mutant again displayed a single upward-shifted band consistent with the phosphorylated form of CIITAWT (Fig. 6A). In contrast, in Pim3KD1 cells, the phosphorylated form of CIITAWT was absent (Fig. 6A). Importantly, mutation at the control site (T786A and T786D) did not alter the electrophoretic pattern of CIITA; both mutants exhibited banding patterns indistinguishable from WT across ControlKD1, EYA1KD1, and Pim3KD1 cells following IL-1β stimulation, indicating that T786 does not affect CIITA phosphorylation (Fig. 6B). Mutations of CIITA S782 altered its interaction with EYA1 and assembly of the EYA1–CIITA–Runx2 transcriptional complex. (A) Effects of CIITAS782 mutations on its interaction with EYA1 and assembly of the EYA1–CIITA–Runx2 transcriptional complex. Flag-CIITAWT, Flag-CIITAS782A, and Flag-CIITAS782D constructs were transfected into ControlKD1, EYA1KD1, and Pim3KD1 cells. Cells were treated with or without IL-1β (8 ng/mL) for 12 h, followed by immunoprecipitation using anti-Flag antibody. Input (top) and immunoprecipitated (IP; bottom) samples were analyzed by immunoblotting with antibodies against EYA1, Flag, and Runx2. (B) Effects of CIITAT786 mutations on its interaction with EYA1 and assembly of the EYA1–CIITA–Runx2 complex. Flag-CIITAWT, Flag-CIITAT786A, and Flag-CIITAT786D constructs were transfected and analyzed as described in (A). (C and D) Effects of CIITAS782 mutations on promoter occupancy of the EYA1–CIITA–Runx2 complex at the MMP3 promoter. Cells described in (A) were subjected to ChIP assays using IgG, anti-Flag (C), anti-EYA1, or anti-Runx2 (D) antibodies. Input and immunoprecipitated DNA were analyzed by qPCR to assess enrichment at the MMP3 promoter. All immunoblot experiments shown in this figure were independently performed at least three times with consistent results. All qPCR experiments were performed with three independent biological replicates, each measured in triplicate. ns: no significant difference; *P < 0.05; **P < 0.01; ***P < 0.001 Immunoprecipitation analysis revealed that CIITAWT efficiently interacted with EYA1 and facilitated the association with Runx2, particularly upon IL-1β stimulation (Fig. 6A). The CIITAS782A mutant exhibited comparable or slightly enhanced binding to EYA1 and Runx2 (Fig. 6A), indicating that loss of phosphorylation at S782 promoted formation of the EYA1–CIITA–Runx2 complex. In contrast, the phosphomimetic CIITAS782D mutant displayed a substantially reduced interaction with EYA1 and Runx2 (Fig. 6A), suggesting that phosphorylation at S782 disrupted complex assembly. Notably, mutations at the control site (T786A and T786D) did not significantly affect the interaction of CIITA with EYA1 or Runx2, as both mutants behaved similarly to WT across all conditions (Fig. 6B), supporting the specificity of S782 as a regulatory phosphorylation site. Consistently, ChIP assays demonstrated that CIITAWT was recruited to the promoters of MMP3 and ADAMTS8 in response to IL-1β, accompanied by increased occupancy of EYA1 and Runx2 (Fig. 6C and D, and S6). The CIITAS782A mutant showed enhanced promoter binding of CIITA and a corresponding increase in Runx2 recruitment, whereas the CIITAS782D mutant exhibited markedly diminished occupancy of all three components (Fig. 6C and D, and S6). Importantly, CIITAT786A and CIITAT786D mutants displayed promoter-binding patterns indistinguishable from CIITAWT (Figures S7 and S8). Furthermore, EYA1 knockdown significantly impaired CIITAWT recruitment to target promoters, an effect that was slightly rescued by the S782A mutant but not by S782D (Fig. 6C and D, and S6). These results demonstrated that phosphorylation of CIITA at S782 negatively regulated the formation of the EYA1–CIITA–Runx2 transcriptional complex and its recruitment to target gene promoters, whereas mutations at the non-functional T786 site have no detectable impact, underscoring the specificity of S782-mediated regulation. Administration of DS-1-38 in MIA-injected mice significantly prevented the incidence of OA Our previous findings demonstrated that pharmacological inhibition of Runx2 with CADD522 significantly prevents the onset and progression of OA [10]. To determine whether DS-1-38 exerts a similar protective effect, we administered DS-1-38 to mice subjected to MIA–induced OA. Animals were randomly assigned to four groups: saline (control), MIA alone, MIA + DS-1-38, and MIA + CADD522 (Fig. 7A). After 21 days of treatment, knee joint tissues were collected and subjected to histological examination. Histological analyses were performed to assess cartilage degeneration and the therapeutic effects of EYA1 and Runx2 inhibition in OA mice. Safranin O–Fast Green staining demonstrated a substantial loss of proteoglycan content in OA cartilage, indicated by diminished Safranin O–positive staining (Fig. 7B). Notably, both DS-1-38 and CADD522 treatments significantly restored Safranin O staining intensity (Fig. 7B), suggesting preservation of ECM components. Furthermore, the treated groups exhibited reduced cartilage degradation and a more intact cartilage structure compared with untreated OA mice (Fig. 7B). Quantitative analysis using the OARSI scoring system confirmed that OA mice displayed significantly elevated scores relative to controls, whereas DS-1-38 and CADD522 treatments markedly reduced OARSI scores (Fig. 7C), indicating attenuation of cartilage degeneration. H&E staining also revealed that OA joints exhibited pronounced structural abnormalities, including surface fibrillation, cartilage erosion, reduced cartilage thickness, and disorganized chondrocyte distribution (Fig. 7D). In contrast, treatment with the EYA1 inhibitor DS-1-38 or the Runx2 inhibitor CADD522 markedly preserved cartilage architecture, as evidenced by a smoother articular surface, increased cartilage thickness, and more organized chondrocyte arrangement (Fig. 7D). In parallel, micro-CT analysis showed significant recovery of subchondral bone integrity in the DS-1-38- and CADD522-treated mice, compared to the structural deterioration observed in the MIA group (Fig. 7E). These results demonstrate that pharmacological inhibition of EYA1 or Runx2 effectively mitigates OA progression and preserves cartilage integrity in vivo. Administration of DS-1-38 in MIA-injected mice prevented OA incidence. (A) Diagram of DS-1-38 administration in MIA-injected mice. (B) Representative Safranin O/Fast Green–stained sections of articular cartilage from sham, MIA (OA), MIA + DS-1-38, and MIA + CADD522 groups. Scale bars = 100 μm. (C) Quantification of cartilage damage severity using OARSI scores for the groups shown in (B). (D) Representative H&E-stained sections of articular cartilage from the indicated groups. Scale bars = 100 μm. (E) Representative micro-CT images from the indicated groups. Scale bars = 300 μm. (F-H) Serum concentrations of IL-1β (F), IL-6 (G), and TNF-α (H) in the indicated groups. (n = 8 for each group). (I–J) mRNA expression levels of MMPs (MMP3, MMP10, MMP12, MMP16, and MMP8) (I) and ADAMTS family members (ADAMTS8, ADAMTS10, ADAMTS20, and ADAMTS19) (J) in cartilage tissues from the indicated groups (n = 8 per group). (K) Protein expression levels of CIITA, Runx2, EYA1, MMP3, MMP10, MMP12, ADAMTS8, ADAMTS10, and ADAMTS20 in cartilage tissues (n = 3 per group; cartilage samples within each group were pooled with equal weights [0.1 g per sample] prior to protein extraction). (L) Immunoprecipitation analysis using anti-CIITA antibody in cartilage tissues from the indicated groups (n = 3 per group; samples were pooled as described in (K)). ns: no significant difference. ***P < 0.001 To assess the systemic inflammatory response, we quantified serum concentrations of proinflammatory cytokines IL-1β, IL-6, and TNF-α across all groups. As expected, cytokine levels were markedly elevated in the MIA, MIA + DS-1-38, and MIA + CADD522 groups compared to the saline group (Fig. 7F and H). However, no significant differences in cytokine expression were observed among the three MIA-challenged groups (Fig. 7F and H), suggesting that neither DS-1-38 nor CADD522 significantly modulates systemic inflammation. We next evaluated the expression of key matrix-degrading enzymes. RT–qPCR analysis demonstrated that the mRNA levels of MMP3, MMP10, MMP12, MMP16, ADAMTS8, ADAMTS10, and ADAMTS20 were markedly reduced in both the DS-1-38 and CADD522 treatment groups, reaching levels comparable to those in the control group (Fig. 7I and J). Consistently, western blot analysis confirmed that the protein levels of these enzymes were also significantly downregulated following DS-1-38 or CADD522 administration (Fig. 7K). Notably, DS-1-38 treatment led to the appearance of a phosphorylated CIITA band, suggesting altered CIITA phosphorylation status (Fig. 7K). To further investigate the underlying mechanism, we performed immunoprecipitation assays using anti-CIITA in cartilage tissues from sham, MIA, MIA + DS-1-38, and MIA + CADD522 groups. Both DS-1-38 and CADD522 treatments markedly reduced the amounts of EYA1 and Runx2 immunoprecipitated with CIITA (Fig. 7L), indicating that these compounds disrupted the assembly of the EYA1–CIITA–Runx2 transcriptional complex. In parallel, ChIP assays were conducted in the same tissues using IgG, anti-EYA1, anti-CIITA, and anti-Runx2 antibodies, followed by qPCR analysis of promoter occupancy at representative target genes (MMP3, ADAMTS8, and MMP8). The results showed that treatment with DS-1-38 or CADD522 significantly decreased the enrichment of EYA1, CIITA, and Runx2 at the promoters of MMP3 and ADAMTS8, with levels approaching those observed in the sham group, while no enrichment was detected at the MMP8 promoter (Figure S9). Collectively, these findings demonstrate that DS-1-38 and CADD522 suppress the expression of cartilage-degrading enzymes by disrupting the EYA1–CIITA–Runx2 transcriptional complex and its recruitment to target gene promoters, thereby conferring protection against OA development in the MIA-induced mouse model. Discussion ECM degradation by chondrocytes is a hallmark of OA pathogenesis and a major contributor to the structural deterioration of articular cartilage [5–9]. The progressive loss of ECM components, such as collagen and aggrecan, disrupts cartilage integrity, compromises joint function, and accelerates disease progression [5–9]. Although ECM degradation has long been recognized as a central feature of OA, the molecular mechanisms governing the transcriptional regulation of ECM-degrading enzymes remain incompletely understood. Our recent study revealed that the m6A RNA demethylase ALKBH5 and the m6A reader YTHDF1 cooperate to enhance Runx2 mRNA stability through m6A modification in an inflammatory microenvironment [10]. In turn, Runx2 directly binds to the promoters of MMP3, MMP10, MMP12, MMP16, ADAMTS8, ADAMTS10, and ADAMTS20, activating their transcription and thereby exacerbating cartilage matrix degradation and OA progression [10]. However, the transcriptional cofactors that collaborate with Runx2 to regulate the expression of these catabolic genes remained unclear. In this study, we uncovered a previously unrecognized transcriptional regulatory mechanism in OA, whereby the transcription factor Runx2 forms a functional complex with the transcriptional regulators CIITA and EYA1 to drive the expression of key matrix-degrading enzymes (Fig. 8). We further demonstrate that CIITA is a downstream substrate of EYA1 phosphatase, and that EYA1 can dephosphorylate CIITA at a Pim3-mediated phosphorylation site under inflammatory conditions (Fig. 8). This finding provides important mechanistic insight into how proinflammatory signaling modulates transcriptional regulators to control ECM degradation in chondrocytes—a key pathological process in OA progression. Schematic model of proinflammatory cytokines mediating the EYA1-CIITA-Runx2 complex and matrix-degrading genes in the pathogenesis of OA. Proinflammatory cytokines upregulate the expression of EYA1, CIITA, and Runx2. Runx2 recruits CIITA and EYA1 to assemble a transcriptional complex. In addition to its role in complex formation, EYA1 functions as a phosphatase that dephosphorylates CIITA at serine 782, a site modified by Pim3. The EYA1–CIITA–Runx2 complex binds to the promoters of several matrix-degrading enzyme genes, including MMP3, MMP10, MMP12, MMP16, ADAMTS8, ADAMTS10, and ADAMTS20, thereby promoting their transcription. Overexpression of these matrix metalloproteinases and aggrecanases leads to excessive ECM degradation in chondrocytes, contributing to the development and progression of OA MMPs and ADAMTSs constitute two large and functionally diverse families of ECM–degrading enzymes [22, 23]. While their collective roles in cartilage destruction during OA have been identified for many years, their individual expression patterns and regulatory mechanisms vary substantially across different tissues, developmental stages, and disease contexts [24–26]. For instance, MMP13, ADAMTS4, and ADAMTS5 are well-known mediators of cartilage breakdown in OA, where their upregulation correlates strongly with disease severity [24–26]. However, other members such as MMP3, MMP10, MMP12, MMP16, ADAMTS8, ADAMTS10, and ADAMTS20 have been less studied and appear to be selectively induced in specific pathological settings, as demonstrated in our recent study using iTRAQ proteomics in a MIA-induced OA mouse model [10]. Previous studies have illustrated that distinct transcriptional complexes can modulate the expression of MMPs and ADAMTSs in a context-dependent manner. For example, during intervertebral disc degeneration (IDD), Li et al. reported that phosphorylated Runx2 (pRunx2), stabilized by USP24 (ubiquitin-specific peptidase 24), forms a complex with p300 and NCOA3 (nuclear receptor coactivator 3) to activate a broad spectrum of ADAMTS genes, including ADAMTS1, 4, 5, 8, 9, 14, and 20, thereby accelerating ECM loss [27]. Similarly, Tseng et al. showed that Runx2 cooperates with PGC-1α (PPARγ coactivator 1α) and CBP (CREB-binding protein) to drive ADAMTS4 and ADAMTS5 expression in degenerative discs [28]. These findings underscore the diversity of Runx2-associated transcriptional machinery and its context-specific role in ECM regulation. Despite growing evidence in other degenerative diseases, the transcriptional mechanisms by which Runx2 controls the expression of a distinct set of MMPs and ADAMTSs in OA chondrocytes remained unclear. Our study fills this critical knowledge gap by demonstrating, for the first time, that Runx2 forms an inducible transcriptional complex with CIITA and EYA1 under inflammatory conditions. This EYA1–CIITA–Runx2 complex is recruited to the promoters of MMP3, MMP10, MMP12, MMP16, ADAMTS8, ADAMTS10, and ADAMTS20, where it coordinately activates their transcription. The identification of this novel transcriptional complex not only deepens our mechanistic understanding of ECM remodeling in OA but also expands the broader framework of MMP/ADAMTS regulation. Our findings may offer mechanistic insights applicable to other biological processes and pathological conditions involving selective ECM turnover. EYA1 is a multifunctional protein that serves both as a transcriptional coactivator and a protein tyrosine phosphatase, and is known to be involved in organ development, cell fate determination, and DNA damage repair [14, 15]. CIITA, on the other hand, is widely recognized as the master regulator of MHC class II gene transcription in immune cells, functioning as a non-DNA-binding transcriptional coactivator [29]. Despite their established roles in developmental biology and immunology, the involvement of EYA1 and CIITA in the pathogenesis of OA has not been previously reported. In this study, we provide the first evidence that EYA1 and CIITA play crucial roles in the transcriptional regulation of matrix-degrading enzymes in chondrocytes. An unexpected yet important discovery of this study is the identification of CIITA phosphorylation at serine 782, a modification mediated by Pim3 kinase and significantly enhanced in the absence or pharmacological inhibition of EYA1. In vitro assays confirmed that recombinant EYA1 is capable of completely dephosphorylating CIITA, suggesting that EYA1 directly controls the phosphorylation status of CIITA. These findings reveal a dual role for EYA1 in OA pathogenesis: as a transcriptional coactivator within the EYA1–CIITA–Runx2 complex and as a phosphatase that post-translationally regulates CIITA, thereby adding a new layer of complexity to the control of inflammation-driven gene expression in chondrocytes. Current clinical management of OA primarily focuses on symptom relief through the use of nonsteroidal anti-inflammatory drugs (NSAIDs), intra-articular corticosteroid injections, and physical therapy [2, 30, 31]. However, these interventions do not halt the progression of cartilage degeneration, and no disease-modifying OA drugs (DMOADs) have been approved to date [2, 30, 31]. Therefore, the identification of molecular targets that can prevent or slow OA progression remains an urgent unmet need. In this study, we demonstrate that pharmacological inhibition of the EYA1–CIITA–Runx2 transcriptional axis using either DS-1-38, a phosphatase inhibitor targeting EYA1, or CADD522, a Runx2 inhibitor, significantly attenuates OA onset and progression in MIA-induced mouse models. Histological and imaging analyses confirmed that both compounds preserved cartilage histological architecture and proteoglycan content—as evidenced by Safranin O staining—and maintained subchondral bone microarchitecture, while markedly reducing the expression of key matrix-degrading enzymes. These results suggest that DS-1-38 and CADD522 act as transcriptionally targeted agents capable of interrupting ECM catabolism in OA joints, thus holding promise as novel DMOAD candidates. Importantly, our data also show that neither DS-1-38 nor CADD522 significantly reduced the levels of proinflammatory cytokines such as IL-1β, IL-6, and TNF-α, indicating that their chondroprotective effects occur downstream of the inflammatory cascade. This highlights a potential limitation of these agents as standalone therapies and underscores the need to explore combination strategies. Given that inflammation plays a pivotal role in initiating and sustaining catabolic gene expression in OA, it is plausible that co-administration of DS-1-38 or CADD522 with anti-inflammatory agents may yield synergistic benefits by simultaneously suppressing upstream inflammatory triggers and downstream matrix degradation pathways. Future studies are warranted to systematically evaluate the efficacy and safety of such combination regimens in preclinical and clinical settings. Despite the strengths of this study, several limitations should be acknowledged. First, although our findings demonstrate a clear regulatory role for the EYA1–CIITA–Runx2 complex in chondrocytes, whether this mechanism operates similarly in human OA cartilage remains to be validated. Second, the in vivo experiments were limited to the MIA model, which primarily reflects chemically induced cartilage damage; future studies employing surgical or aging-related OA models are needed to confirm the generalizability of our results. Third, while we identified Pim3 as a kinase for CIITA, other upstream signaling pathways that regulate Pim3 activity under inflammatory conditions have not yet been explored. Elucidating these pathways may offer further insights into how extracellular signals fine-tune the EYA1–CIITA–Runx2 transcriptional circuit. In conclusion, our study reveals a novel inflammatory signaling mechanism in chondrocytes involving a transcriptional complex formed by EYA1, CIITA, and Runx2. The dynamic modulation of CIITA phosphorylation/dephosphorylation by Pim3 and EYA1 further underscores the regulatory flexibility of this axis. Targeting components of this complex, such as with DS-1-38 or CADD522, may represent a promising therapeutic strategy to halt cartilage degradation in OA. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Acknowledgements Not applicable. Author contributions Guanghua Nie: Conceptualization, Methodology, Investigation, and Data curation. Yi Li: Methodology and Investigation. Hongmou Zhao: Methodology and Investigation. Dong Wang: Methodology and Investigation. Yan Zhang: Methodology. Xinquan Yang: Methodology. Xiaodong Wen: Supervision, Conceptualization, Writing - review & editing. All authors read and approved the final manuscript. Funding This study was supported by General projects of Shaanxi provincial key research and development plan (NO:2022SF-532) and Shaanxi Provincial Health Research Fund (NO:2022A005). Data availability The datasets used in the present study are available from the corresponding author on reasonable request. Declarations Ethical approval All animal experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of Honghui Hospital, Xi’an Jiaotong University (Protocol No. 2021-069), and were performed in accordance with the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals. Consent for publication All authors agree to publish. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Data Availability Statement The datasets used in the present study are available from the corresponding author on reasonable request.