Enolase 2-mediated lactylation-dependent disruption of the GNL3–MDM2–p53 axis in age-related osteoarthritis

Abstract

Background

Age-related osteoarthritis (OA) involves metabolic dysregulation and chondrocyte senescence. This study examined the nonmetabolic role of enolase 2 (ENO2) in OA pathogenesis and its therapeutic potential.

Methods

Human aged and OA cartilage (n = 3 per group) underwent ¹⁸F-FDG positron emission tomography (PET)–computed tomography (CT) imaging, proteomic profiling, and immunohistochemistry. In vitro chondrocyte senescence models were generated by inducing doxorubicin-induced stress and serial passaging. Protein–protein interactions (ENO2–GNL3–MDM2) were validated by co-immunoprecipitation (IP), GST pull-down, and site-directed mutagenesis (E4A-ENO2 and K5R-GNL3 mutants). Lactylation was assessed using lactylomics and immunoprecipitation. The therapeutic effect of the ENO2-specific inhibitor POMHEX was evaluated in C57BL/6 J mice (n = 6 per group) via intra-articular injection for 16 weeks. Outcomes included histology, micro-CT, pain behavior, and gait analysis.

Results

Proteomics revealed ENO2 upregulation in aged human cartilage. In vitro, ENO2 overexpression promoted extracellular matrix catabolism, senescence, and glycolysis, whereas ENO2 knockdown attenuated these processes. Mediated by its Glu-4 residue, nuclear ENO2 bound GNL3 lactylated at Lys-5. This interaction displaced MDM2 from GNL3, resulting in MDM2 destabilization, impaired ubiquitination, p53 accumulation, and persistent senescence. Moreover, p53 transcriptionally activated ENO2, establishing a pathological positive feedback loop. Pharmacological inhibition of ENO2 with POMHEX disrupted ENO2–GNL3 binding, restored p53 degradation, reduced senescence markers in vitro, and mitigated cartilage degradation, subchondral bone sclerosis, and pain in aged mice.

Conclusions

ENO2 promotes OA progression through a lactate-dependent, lactylation-mediated disruption of the GNL3–MDM2–p53 axis, leading to a senescent feedback loop. Targeting ENO2 may represent a novel disease-modifying therapeutic approach for age-related OA.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s11658-026-00960-6.

Introduction

Osteoarthritis (OA) is the most prevalent degenerative joint disease globally, affecting over 30% of individuals aged 65 years and above [1, 2]. Characterized by age-related cartilage degeneration [3, 4] and extracellular matrix degradation [5, 6], OA imposes a significant disease burden, particularly in aging populations [7, 8]. In China, the number of individuals with OA has surpassed 130 million, and the disease burden continues to increase [9, 10]. Current clinical treatments primarily focus on symptom relief using nonsteroidal anti-inflammatory drugs [11], intra-articular injections [12], and joint replacement surgeries [13]. However, these interventions fail to reverse disease progression or repair damaged cartilage [14], highlighting the urgent need for disease-modifying therapies. Recent studies have identified chondrocyte senescence as a central driver of OA progression [15, 16]. Therefore, elucidating the molecular mechanisms underlying senescence and developing targeted interventions are crucial for overcoming the therapeutic challenges in OA.

Metabolic reprogramming in OA has garnered increasing attention [17, 18]. Senescent chondrocytes exhibit significant activation of glycolysis and mitochondrial dysfunction [19–22], leading to lactate accumulation and exacerbated oxidative stress [23, 24]. These metabolic alterations further accelerate cellular senescence and matrix degradation [25, 26]. For instance, dysregulation of the mitochondrial unfolded protein response has been linked to mechanically induced chondrocyte senescence [27, 28]. In addition, the aberrant activation of the Notch signaling pathway can promote cartilage degeneration through clathrin-mediated endocytosis [28]. Metabolic byproducts, such as lactate, not only disrupt cartilage homeostasis through acidification of the microenvironment but also serve as substrates for epigenetic modifications (e.g., lactylation) [20, 29, 30], regulating the expression of inflammation- and senescence-related genes [31]. Despite the widespread recognition of metabolic dysregulation in OA, upstream regulatory factors that integrate metabolic and senescence signaling networks remain to be elucidated.

Enolase 2 (ENO2), a key glycolytic enzyme, catalyzes the conversion of 2-phosphoglycerate to phosphoenolpyruvate (PEP) [32]. It has traditionally been associated with energy metabolism regulation [33]. However, recent studies have revealed noncanonical functions of ENO2 in cancer [34]. Moreover, the nuclear translocation of ENO2 can regulate genes [35]. These findings suggest that ENO2 participates in cell-fate regulation through both metabolism-dependent and independent pathways. However, whether ENO2 coordinates metabolic reprogramming and senescence signaling in OA via similar mechanisms remains unknown.

Therefore, our study aimed to elucidate the multifaceted functions of ENO2 in OA pathogenesis and explore its potential as a novel therapeutic target, thereby seeking new strategies to treat aged osteoarthritic cartilage and mitigate cellular senescence. This study addresses the critical need for senescence-targeting interventions to halt or reverse the progression of OA in aging populations.

Materials and methods

Clinical samples

To investigate the metabolic and pathological changes in articular cartilage and synovial fluid across different age groups, we enrolled patients undergoing joint replacement surgery at the Zhejiang University School of Medicine First Affiliated Hospital between January 2022 and December 2024. All enrolled patients met the following criteria: (1) diagnosis of OA requiring joint replacement; (2) no history of rheumatoid arthritis, gout, or other inflammatory joint diseases; (3) no history of cancer, autoimmune diseases, or severe infectious diseases; and (4) no prior joint surgeries or treatments that could confound the study results.

Articular cartilage and synovial fluid samples were collected during joint replacement surgeries. The cartilage specimens were immediately snap-frozen in liquid nitrogen for subsequent biochemical analysis or fixed in 4% paraformaldehyde for histological evaluation. Synovial fluid samples were centrifuged at 1000g for 10 min to remove cellular debris and stored at −80 °C for lactate level measurements and protein analysis.

In addition, PET–CT imaging data were retrospectively obtained from the Department of Radiology for patients with a clinical diagnosis of tumors, ensuring no overlap with the OA patient cohort. These imaging data were used to evaluate ¹⁸F-FDG accumulation patterns in the articular cartilage.

Histological analysis

Human articular cartilage specimens and mouse knee-joint tissues were fixed in 4% paraformaldehyde and embedded in paraffin for histological analyses. For immunohistochemical staining, 4-μm-thick sections were cut from the paraffin-embedded blocks. Sections were deparaffinized and rehydrated using a graded ethanol series. Antigen retrieval was performed using a citrate buffer in a microwave oven for 20 min. Endogenous peroxidase activity was quenched by incubating with 3% hydrogen peroxide for 10 min. The sections were incubated with 5% goat serum for 1 h at 25 °C to reduce nonspecific binding. The sections were then incubated with primary antibodies against LDHA (1:100, ab52488, Abcam), GLUT1 (1:100, ab115730, Abcam), ENO2 (1:100, ab79757, Abcam), or collagen II (1:100, ab307674, Abcam) overnight at 4 °C. After washing with phosphate-buffered saline (PBS), the sections were incubated with a secondary antibody (goat anti-rabbit IgG 1:200, horseradish peroxidase (HRP)-conjugated, ab6721, Abcam) for 1 h at room temperature. Staining was visualized using 3,3′-diaminobenzidine substrate and counterstained with hematoxylin. The slides were dehydrated, cleared, and mounted using a mounting medium. For safranin O-fast green staining of mouse knee joint sections, paraffin-embedded tissues were sectioned at a thickness of 4 μm. The sections were then deparaffinized and rehydrated. The slides were stained with hematoxylin for 5 min, rinsed with running water, and differentiated using acid alcohol. After washing, the sections were stained with Fast Green dye for 5 min, followed by differentiation in 1% acetic acid. The sections were then stained with safranin O for 10 min, differentiated in 70% alcohol, dehydrated using graded alcohols, cleared in xylene, and mounted. All the histological sections were analyzed and imaged using a light microscope.

Lactate content measurement

Lactate levels in the synovial fluid and cell culture supernatants were measured using an l-Lactate Assay Kit (S0208S, Beyotime). For synovial fluid samples, the collected fluid was centrifuged at 4 °C and 1000–2000g for 10 min to remove cellular debris; the resulting supernatant was then used for analysis. For the cell culture supernatants, the medium was carefully aspirated from adherent cells or collected from suspended cells after centrifugation (300g for 5 min), and the supernatant was used directly in the assay. The assay was performed according to the manufacturer’s instructions. Briefly, 50 μl of each sample or standard was added to a 96-well plate. A total of 50 μl of the working solution was added to each well and mixed thoroughly. The plates were incubated at 37 °C for 30 min. Absorbance at 450 nm was measured using a microplate reader. Lactate concentrations were determined using the standard curve provided with the kit and adjusted for sample dilution factors.

Proteomic analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis

Proteomic profiling was performed on articular cartilage specimens obtained from three young and three elderly patients. Tissues were collected and snap-frozen in liquid nitrogen immediately after surgery. The samples were sent to OEBiotech (Shanghai, China) for proteomic analysis. Briefly, the tissues were homogenized and proteins were extracted using lysis buffer containing protease inhibitors. Protein concentrations were measured using a bicinchoninic acid (BCA) assay. Equal amounts of protein from each sample were digested with trypsin, and the resulting peptides were labeled using a tandem mass tag (TMT) labeling kit according to the manufacturer’s instructions. The labeled peptides were pooled, desalted, and subjected to liquid chromatography–tandem mass spectrometry (LC–MS/MS) analysis. Mass spectrometry data were analyzed using MaxQuant software for peptide identification and quantification. Raw data were searched against the human protein database to identify the proteins. Quantitative data were normalized and log-transformed. Differentially expressed proteins were defined as those with a fold change greater than 1.25 or less than 0.75 between the two groups. Differentially expressed proteins were then subjected to KEGG pathway enrichment analysis using the DAVID Bioinformatics Resources. The results were visualized using heat maps and pathway enrichment plots to identify the key metabolic pathways associated with age-related changes in the cartilage.

Co-immunoprecipitation (Co-IP) and GST pull-down analysis

Co-IP assays were performed to investigate protein–protein interactions. Briefly, the cells were lysed in ice-cold lysis buffer containing 50 mM Tris–HCl, 150 mM NaCl, 1% NP-40, and protease inhibitors. The cell lysates were clarified by centrifugation at 12,000g for 10 min at 4 °C. The supernatants were incubated overnight at 4 °C with primary antibodies or control IgG (ab172730, Abcam). Protein A/G magnetic beads (P2108, Beyotime) were then added and incubated for 2 h. The immune complexes were washed three times with lysis buffer, and the bound proteins were eluted by boiling in loading buffer. The eluted proteins were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) and analyzed by western blotting using appropriate antibodies. To assess direct protein–protein interactions, ENO2–GST (or GST alone) was immobilized on glutathione beads and incubated with cell lysates containing GNL3 or MDM2. After washing, the bound proteins were eluted and analyzed using western blotting.

Western blot analysis

Western blot analysis was performed to detect protein expression in human and mouse tissue specimens as well as in cell samples. Briefly, the tissues and cells were lysed in radioimmunoprecipitation assay buffer supplemented with 1 mM phenylmethanesulfonyl fluoride. The protein concentration was determined using a BCA protein assay kit. Equal amounts of protein were mixed with 5× loading buffer, separated by 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis, and transferred onto polyvinylidene fluoride membranes. Membranes were blocked with 5% bovine serum albumin in Tris-buffered saline containing 0.1% Tween-20 for 1 h at room temperature. The membranes were then incubated with primary antibodies against GLUT1 (1:1000, ab115730, Abcam), LDHA (1:1000, ab52488, Abcam), β-actin (1:1000, cat no. 66009, Proteintech), ENO2 (1:1000, ab79757, Abcam), MMP3 (1:1000, cat no. 17873, Proteintech), MMP13 (1:1000, cat no. 18165, Proteintech), ACAN (1:1000, A8536, Abclonal), collagen II (1:1000, ab188570, Abcam), p53 (1:1000, A25915, Abclonal), p21 (1:1000, cat no. 10355, Proteintech), p16 (1:1000, ab270058, Abcam), GNL3 (1:1000, cat no. 15060, Proteintech), HA tag (1:1000, cat no. 51064, Proteintech), Flag tag (1:1000, cat no. 20543, Proteintech), tubulin (1:1000, A12289, Abclonal), pan-kla (1:1000, PTM-1401RM, PTM Bio), ubiquitin (1:1000, A0162, Abclonal), MDM2 (1:1000, ab259265, Abcam), and histone H3 (1:1000, ab1791, Abcam), at appropriate dilutions overnight at 4 °C. After washing with Tris-buffered saline with Tween 20 (TBST), the membranes were incubated with a goat anti-rabbit IgG secondary antibody conjugated to HRP (1:5000, ab6721, Abcam) for 1 h at room temperature. The protein bands were visualized using enhanced chemiluminescence reagents and analyzed using an imaging system. β-actin or tubulin was used to normalize the expression levels of the target proteins. For cycloheximide chase assays, cells were treated with CHX to inhibit de novo protein synthesis, harvested at sequential time points, and subjected to western blot analysis as described above to determine the degradation kinetics of target proteins, with the relative protein levels quantified and normalized to the internal reference.

Ubiquitination assay

To assess endogenous ubiquitination of MDM2 or ENO2, cells were treated with the proteasome inhibitor MG132 for 6 h prior to harvest to prevent degradation of ubiquitinated proteins. Cells were lysed in radioimmunoprecipitation assay (RIPA) buffer supplemented with protease inhibitors. The lysates were sonicated briefly and centrifuged at 12,000g for 10 min at 4 °C. Supernatants were incubated with anti-MDM2 (ab259265, Abcam) antibody or anti-ENO2 antibody (ab79757, Abcam) overnight at 4 °C, followed by incubation with Protein A/G magnetic beads for 2 h. Beads were washed three times with lysis buffer. Ubiquitinated proteins were eluted by boiling in SDS loading buffer and detected by western blotting using anti-ubiquitin antibody (A0162, Abclonal). For denaturing immunoprecipitation to confirm polyubiquitination, cell lysates were boiled in 1% SDS for 10 min, diluted tenfold with lysis buffer without SDS, and then subjected to immunoprecipitation as described above.

Quantitative PCR (qPCR) and chromatin immunoprecipitation (ChIP)–qPCR

Total RNA was extracted from the cells or tissues using the AFTMag Quick Tissue/Cell RNA Extraction Kit (RK30156, ABclonal). For cell samples, cells were lysed in Mag Buffer RLA. For tissue samples, tissues were homogenized in Mag Buffer RLA with proteinase K (20 mg/mL), depending on the tissue type. RNA was then purified according to the manufacturer’s instructions, which included DNase I treatment to remove genomic DNA contamination. The extracted RNA was quantified using a spectrophotometer, and its integrity was assessed by agarose gel electrophoresis. The thermal cycling conditions were 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min. The primers used for qPCR are presented in Supplementary Table S1. Expression levels of the target genes were normalized to ACTA1. ChIP assays were performed using a ChIP–qPCR Assay Kit (P2078, Beyotime). Cells were fixed with 1% formaldehyde for 10 min at room temperature to cross-link DNA and proteins. Nuclear lysates were prepared and sonicated to shear DNA into fragments of 200–1000 base pairs (bp). The lysates were immunoprecipitated with anti-p53 antibody (1:100, cat no. 10442, Proteintech) or control IgG (ab172730, Abcam) overnight at 4 °C. Protein A/G magnetic beads were used to capture immune complexes. After washing, the cross-links were reversed and DNA was purified according to the kit’s instructions. qPCR was performed on purified DNA using primers specific to the ENO2 promoter region. The ChIP–qPCR results were expressed as fold enrichment over the input.

Cell culture

Primary human chondrocytes were isolated from the articular cartilage specimens obtained during joint replacement surgery. The cartilage was minced and digested with 0.2% collagenase II (9001-12-1, Sigma-Aldrich) in Dulbecco’s modified Eagle medium (DMEM) (11965092, Gibco) containing 10% fetal bovine serum (FBS) (A5670201, Gibco) and 1% penicillin–streptomycin (15140122, Gibco) at 37 °C for 2–3 h. The cells were then filtered through a 70-μm cell strainer, centrifuged at 300g for 5 min, and cultured in DMEM with 10% FBS and 1% penicillin–streptomycin in a humidified incubator at 37 °C with 5% CO₂. Chondrocytes were passaged once they reached 80–90% confluence. The C28/l2 chondrocyte cell line and HEK293T cells were cultured in DMEM supplemented with 10% FBS and 1% penicillin–streptomycin. Cells were maintained at 37 °C with 5% CO₂ and passaged every 2–3 days.

Nuclear and cytoplasmic protein extraction

Nuclear and cytoplasmic protein extraction was performed using a Nuclear and Cytoplasmic Protein Extraction Kit (P0027, Beyotime). Briefly, adherent or suspended cells were washed with phosphate buffered saline (PBS) and collected via centrifugation. The cell pellets were resuspended in ice-cold lysis buffer containing 1 mM phenylmethylsulfonyl fluoride (PMSF) and briefly vortexed. After ice incubation, the cytoplasmic proteins were extracted by centrifugation at 4 °C and 15,000g for 5 min. The supernatant containing cytoplasmic proteins was collected. For nuclear protein extraction, pellets were lysed in nuclear extraction buffer containing phenylmethylsulfonyl fluoride and subjected to intermittent vortexing on ice for 30 min. Following centrifugation under the same conditions for 10 min, the supernatant containing nuclear proteins was collected and stored at −70 °C until use.

Cellular staining assays

For immunofluorescence analysis, the cells were fixed with 4% paraformaldehyde for 15 min at room temperature and permeabilized with 0.1% Triton X-100 for 10 min. After blocking with 5% normal goat serum for 1 h at room temperature, cells were incubated with primary antibody against ENO2 (1:100, ab79757, Abcam) overnight at 4 °C. Following primary antibody incubation, cells were washed three times with PBS and incubated with Alexa Fluor^® 594-conjugated goat anti-rabbit IgG (1:200, ab150080, Abcam) for 1 h at room temperature. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) (C1002, Beyotime) for 5 min. Fluorescence images were acquired using a fluorescence microscope. For senescence detection, cells were stained with the β-galactosidase kit (RG0039, Beyotime) according to the manufacturer’s instructions. Senescent (blue) cells were counted using a light microscope.

Plasmid transfection and luciferase assay

Transient transfections of C28/l2 chondrocytes and HEK293T cells was performed using Lipofectamine 3000 (L3000015; Invitrogen) according to the manufacturer’s protocol. For each transfection, the cells were seeded in six-well plates at 70–80% confluence. The plasmids were provided by Shanghai GeneChem. Briefly, plasmid DNA (1–2 μg) was mixed with Opti-MEM (31985070, Gibco) and Lipofectamine 3000 in a ratio of 1:2 (DNA: lipid). The mixture was incubated at room temperature for 20 min to allow lipid–DNA complex formation. The complexes were then added to the cells in a serum-free medium. After 6 h of incubation at 37 °C, the transfection medium was replaced with complete growth medium. The cells were harvested 48 h post transfection for further analysis. For the luciferase assay, cells were co-transfected with ENO2-promoter-luciferase and p53 plasmids. After 24 h, the luciferase activity was measured and normalized to Renilla luciferase activity. All short hairpin RNA (shRNA), single guide RNA (sgRNA), and overexpression construct sequences are presented in Supplementary Table S1.

Fluorescence-activated cell sorting analysis

Apoptosis was assessed using Annexin V-FITC and PI Apoptosis Detection Kits (C1062S; Beyotime). The cells were harvested, washed with cold PBS, and resuspended in 1× binding buffer. Annexin V-FITC and PI were added according to the manufacturer’s instructions, and the cells were incubated for 15 min in the dark. Samples were analyzed using a flow cytometer, and data were processed using FlowJo software. For C12-FDG staining [36], cells were incubated with C12-FDG (HY-126839, MCE) at a final concentration of 10 μM in serum-free medium for 1 h at 37 °C. After washing with PBS, cells were analyzed on a flow cytometer using fluorescein isothiocyanate (FITC) channels. The data were analyzed to assess cellular senescence. For glucose uptake, cells were incubated with 50 μM 2-NBDG (HY-116215,MCE) in glucose-free medium for 30 min at 37 °C. After washing with PBS, fluorescence was acquired using a flow cytometer (FITC channel) and analyzed using FlowJo.

Enolase activity assay

Enolase enzymatic activity was measured using a commercial enolase activity assay kit (e.g., Solarbio, BC1180) according to the manufacturer’s instructions. Briefly, chondrocytes or cartilage tissue samples were lysed in ice-cold lysis buffer, and the supernatant was collected after centrifugation. Protein concentration was determined by BCA assay, and equal amounts of protein were added to the reaction mix containing the assay buffer and substrate. The reaction was initiated by incubation at 37 °C, and absorbance at 340 nm was continuously monitored using a microplate reader. Enolase activity was calculated on the basis of the rate of NADH production, normalized to total protein content, and expressed as units per milligram of protein.

Seahorse analysis

The extracellular acidification rate (ECAR) was measured using a Seahorse XF96 analyzer. Cells (2 × 10⁴/well) were seeded overnight, switched to Seahorse XF base medium 1 h before assay, and sequentially injected with glucose (10 mM), oligomycin (1 μM), and 2-DG (50 mM). Data were normalized to the cell number.

Protein spectrometry analysis for lactylation sites

To identify the lactylation sites on GNL3, IP was performed to enrich the GNL3 protein from chondrocytes. Briefly, cell lysates were incubated with anti-GNL3 antibody and protein A/G magnetic beads overnight at 4 °C. The beads were then washed and the bound proteins were eluted and digested with trypsin. The resulting peptides were analyzed by LC–MS/MS at the PTM Biotechnology Company. Mass spectrometry data were used to identify the lactylation sites on GNL3, with a focus on lysine residues modified by lactylation.

Molecular docking simulation

Molecular docking simulations were performed to predict the binding interactions between ENO2 and GNL3. Three-dimensional structures of ENO2 and GNL3 were obtained from the Protein Data Bank. Simulations were conducted using AutoDock Vina software. Briefly, the proteins were prepared by removing water molecules and adding hydrogen atoms. The binding sites were defined on the basis of crystal structure analysis. Docking parameters were set in a grid box covering the potential interaction region. The results were visualized using PyMOL to identify the key residues and interaction patterns.

Transcription factor prediction and promoter analysis

To investigate the transcriptional regulation of ENO2, we used computational tools to predict potential transcription factor-binding sites and p53-binding motifs within the ENO2 promoter region. The PROMO database was used to predict transcription factors that may regulate ENO2 expression. The analysis involved retrieving the promoter sequence of ENO2 from the UCSC Genome Browser, inputting this sequence into PROMO, and selecting an appropriate matrix database of human transcription factors to predict potential binding sites. The JASPAR database was used to identify specific p53-binding motifs in the ENO2 promoter. The promoter sequence was scanned using JASPAR, which is configured to detect motifs corresponding to p53. This dual approach provided a comprehensive overview of the transcriptional regulatory elements associated with ENO2, highlighting the potential regulatory interactions that could influence its expression in the context of chondrocyte senescence and OA pathogenesis.

Animals

Male C57BL/6J mice were purchased from the Animal Center of Zhejiang University School of Medicine First Affiliated Hospital. To assess age-related OA, 18-month-old male C57BL/6J mice were used in this study. For intra-articular injections, the mice were anesthetized with ketamine hydrochloride (200 mg/kg) and xylazine (10 mg/kg) via intraperitoneal injection. The drug solution was administered to the knee joint using a 30-gauge needle. The same volume of saline was administered to the control group. Pain-related behaviors were assessed using hot-plate and knee-extension tests. In the hot-plate test, mice were placed on a heated surface and the latency to paw licking or jumping was recorded. In the knee-extension test, the maximum knee extension angle was measured to assess pain-induced joint stiffness. All behavioral tests were conducted by observers who were blinded to the treatment groups. For gait analysis, the hind paws were dipped in red ink and the forepaws in blue ink. The mice then walked along a 20 × 70-cm runway lined with white paper. Stride length (distance between ipsilateral prints) and step length (anterior–posterior distance between contralateral prints) were measured in three consecutive steps. At the end of the experiment, mice were sacrificed via cervical dislocation. Knee-joint tissues were collected for histological analysis and stored in 4% paraformaldehyde or liquid nitrogen until further analysis.

Micro-CT imaging and analysis

For the micro-CT analysis of knee-joint tissues in mice, the following steps were performed. After sacrifice, knee joints were dissected and fixed in 4% paraformaldehyde for 48 h. The fixed knee joints were scanned using a micro-CT scanner at a resolution of 10 µm per pixel. The scanning parameters were set as follows: voltage, 50 kV; current, 100 µA; and integration time, 10 ms. The scanning process captured 360° projections over a rotation of 180°, with each scan taking approximately 30 min per sample. The acquired micro-CT data were reconstructed using the NRecon software to generate three-dimensional images. The reconstruction parameters were optimized to enhance image quality, with a pixel size of 10 µm and a slice thickness of 10 µm. The reconstructed images were analyzed using the CTAn software to quantify osteophyte formation and other structural changes. Three-dimensional visualizations of knee-joint structures were generated using the software’s rendering tools to provide a comprehensive assessment of osteoarthritic changes in mouse knee joints.

Statistical analysis

All data are expressed as the mean ± scanning electron microscopy (SEM). Statistical analyses were performed using GraphPad Prism software. Comparisons between groups were performed using an unpaired Student’s t-test for two groups or a one-way analysis of variance (ANOVA) for multiple groups. Statistical significance was set at p < 0.05.

Results

Senescence-associated glycolytic reprogramming and ENO2 upregulation in human cartilage

OA is defined by age-related cartilage degeneration, and metabolic dysregulation is increasingly linked to the crosstalk between aging and cartilage damage. Clinical observations revealed an interesting phenomenon. The PET–CT scan showed that compared with the young control group, the joints of elderly patients seemed to have a significantly higher uptake of ¹⁸F-FDG. Correspondingly, the medial knee joint showed a higher uptake than the lateral knee joint (Fig. 1A). ¹⁸F-FDG reflects glucose metabolic activity; therefore, this finding points to enhanced glucose use in senescent and osteoarthritic cartilages. Given the impracticality of conducting prospective and retrospective studies on patients undergoing PET–CT scans, we further isolated chondrocytes from clinical cartilage specimens, cultured them in medium supplemented with 2NBDG, and assessed their glucose uptake capacity. Consistent with the PET–CT results, chondrocytes from elderly donors exhibited significantly increased glucose uptake, which also implied their enhanced glycolytic activity (Supplementary Figure S1A). Consistent with this finding, lactate levels in the synovial fluid were also higher in the elderly (Fig. 1B). While pyruvate is the physiological end product of glycolysis, lactate is produced under anaerobic or pathological conditions. Thus, the elevated lactate levels in synovial fluid indicate increased glycolytic activity in senescent cartilage. To elucidate the molecular basis, we compared the proteomes of young and aged cartilage samples. We identified 255 upregulated and 529 downregulated proteins in aged samples (Fig. 1C), and glycolysis/gluconeogenesis was among the most activated pathways (Fig. 1D). Key glycolytic molecules, including GLUT1, a glucose transporter, and LDHA, which produces lactate, were also more abundant in aged cartilage (Supplementary Figure S1B), reinforcing that glycolytic reprogramming is a hallmark of senescent cartilage. We validated these clinical findings in two in vitro chondrocyte senescence models, doxorubicin (DOX)-induced stress senescence and serial passaging (P0→P2)-induced replicative senescence. Both models showed increased lactate secretion and a yellow shift in the culture medium (Fig. 1E, F; Supplementary Figure S1C, 1D). This matched the elevated synovial lactate levels in patients and confirmed that the process from senescence to glycolysis activation and then to lactate accumulation is a universal feature of chondrocyte senescence. Senescent chondrocytes also had higher levels of GLUT1 and LDHA, along with an increased extracellular acidification rate and glucose uptake (Fig. 1G, H, I; Supplementary Figure S1E, S1F, S1G). These data confirmed that glycolytic activity is functionally enhanced in senescent chondrocytes beyond molecular upregulation. We then focused on key glycolytic regulators. A heat map of glycolytic enzymes revealed 16 upregulated proteins in the aged cartilage, with ENO2 standing out (Fig. 1J). Time-course analysis showed that ENO2 mRNA levels gradually increased during DOX-induced senescence (Fig. 1K), linking its expression to the senescence process. ENO2 protein was more abundant in both aged cartilage (Fig. 1L, M, Supplementary Figure S1H) and DOX-induced senescent chondrocytes (Fig. 1N). As an important heteromer of ENO2, ENO1 can form a complex with ENO2 to exert biological functions (Supplementary Figure S1I); however, no altered expression of ENO1 was detected in the aforementioned differential proteomic analysis. Meanwhile, we validated and examined the expression level of ENO1 in clinical cartilage specimens, and the results showed no significant difference in ENO1 expression in aged cartilage (Supplementary Figure S1J). Simultaneously, immunofluorescence showed that ENO2 was upregulated in both the cytoplasm and nucleus of DOX-treated chondrocytes, suggesting a possible nonmetabolic function of ENO2 metabolic enzymes. (Fig. 1O) These results raise questions regarding the nonmetabolic role of ENO2 in chondrocytes and its potential mechanisms in chondrocyte aging.

Fig. 1.

Metabolic reprogramming and ENO2 upregulation in age-related osteoarthritic cartilage. A PET–CT images showing ¹⁸F-FDG uptake in articular cartilage from young and aged individuals. B Lactate levels in synovial fluid from young and aged individuals. C Volcano plot of differentially expressed proteins in cartilage from young versus aged individuals. D KEGG pathway enrichment analysis of differentially expressed proteins. E Lactate production in human primary chondrocytes treated with 1 μM DOX. F Color change of phenol red-containing supernatant after DOX treatment. G Western blot analysis of LDHA and GLUT1 expression in DOX-treated chondrocytes. H Seahorse ECAR measurement in DOX-treated chondrocytes. I Flow cytometry quantification of 2-NBDG uptake in DOX-treated chondrocytes. J Heat map of glycolysis-related differential proteins identified by proteomics. K qPCR validation of glycolytic gene expression in DOX-treated chondrocytes. L Immunohistochemical staining of p16 and ENO2 in cartilage from young and aged individuals. M Western blot analysis of ENO2 expression in cartilage from young and aged individuals. N Western blot analysis of ENO2 expression in DOX-treated chondrocytes. O Immunofluorescence analysis of ENO2 localization in DOX-treated chondrocytes. ^*P ≤ 0.05, ^**P ≤ 0.01, ^***P ≤ 0.001

ENO2 drives catabolism and senescence in human chondrocytes

After identifying ENO2 as the core driver of glycolytic reprogramming in senescent cartilage, we explored its effect on chondrocyte function. First, we increased the expression of ENO2 (Supplementary Figure S2A) and examined the effects. ENO2 upregulation significantly increased the expression of key molecules that promote cartilage matrix degradation (Fig. 2A). At the same time, it reduced the expression levels of key proteins involved in the synthesis and maintenance of the cartilage matrix, indicating that cartilage matrix synthesis was impaired. (Fig. 2B) This meant that ENO2 not only promoted matrix breakdown but also undermined the ability to repair. In addition to matrix damage, higher ENO2 levels increased the expression of senescence-related proteins (Fig. 2C). It further enhanced glycolytic activity, as reflected by the increased expression of glycolysis-related molecules (Fig. 2D). Additional observations showed that in ENO2-overexpressing cells, the main component of cartilage, proteoglycans, decreased, whereas the expression of another type of matrix-degrading molecule increased (Fig. 2E). Functionally, the number of apoptotic cells increased (Fig. 2F) and more cells exhibited senescent features (Fig. 2G, H). Given the harmful effects of elevated ENO2 levels, we investigated whether reducing ENO2 could reverse this damage. We knocked down ENO2 in chondrocytes (Supplementary Figure S2B) and examined its effects on cellular phenotypes. The results showed that the number of molecules that had spiked to promote matrix breakdown in DOX-treated chondrocytes decreased by the downregulation of ENO2 (Fig. 2I), and at the same time, the key factors that help build the cartilage matrix returned to normal levels (Fig. 2J). In addition, proteins linked to cell senescence became less abundant (Fig. 2K) and the expression of molecules involved in glycolysis decreased (Fig. 2L). In addition, in chondrocytes, the expression trends of ACAN and ADAMTS4 proteins, which were altered by DOX stimulation, changed upon ENO2 inhibition (Fig. 2M). Compared with stimulation with DOX alone, DOX and sh-ENO2 dual stimulation resulted in fewer aged (Fig. 2N, P) and apoptotic chondrocytes (Fig. 2O). Moreover, the effects of sg-ENO2 on senescent chondrocytes were consistent with those observed in the above experiments (Supplementary Figure S2C, S2D, S2E). Collectively, these findings indicate that ENO2 promotes cartilage breakdown by boosting matrix catabolism, accelerating cell senescence, and altering glycolytic activity. When ENO2 is inhibited, these degenerative changes are effectively reversed, confirming that ENO2 is a key mediator of OA-related cartilage damage.

Fig. 2.

Functional consequences of ENO2 modulation in human chondrocytes. Western blot analysis of MMP3 and MMP13 in chondrocytes overexpressing. ENO2. B Western blot analysis of collagen II and SOX9 in chondrocytes overexpressing ENO2. C Western blot analysis of p53, p21, and p16 in chondrocytes overexpressing ENO2. D Western blot analysis of GLUT1 and LDHA in chondrocytes overexpressing ENO2. E Immunofluorescence staining of ACAN and ADAMTS4 in chondrocytes overexpressing ENO2. F Flow cytometry quantification of apoptotic cells after ENO2 overexpression. G Flow cytometry detection of C12-FDG after ENO2 overexpression. H β-galactosidase staining of chondrocytes overexpressing ENO2. I Western blot analysis of MMP3 and MMP13 in chondrocytes subjected to ENO2 knockdown or DOX treatment. J Western blot analysis of collagen II and SOX9 in chondrocytes subjected to ENO2 knockdown or DOX treatment. K Western blot analysis of p53, p21, and p16 in chondrocytes subjected to ENO2 knockdown or DOX treatment. L Western blot analysis of GLUT1 and LDHA in chondrocytes subjected to ENO2 knockdown or DOX treatment. M Immunofluorescence staining of ACAN and ADAMTS4 in chondrocytes subjected to ENO2 knockdown or DOX treatment. N β-galactosidase staining of chondrocytes subjected to ENO2 knockdown or DOX treatment. O Flow cytometry quantification of apoptotic cells after ENO2 knockdown or DOX treatment. P Flow cytometry detection of C12-FDG after ENO2 knockdown or DOX treatment. ^*P ≤ 0.05, ^**P ≤ 0.01, ^***P ≤ 0.001

ENO2 drives senescence by binding lactylated GNL3 in the nucleus

ENO2 is typically viewed as a catalytic enzyme involved in glycolysis. Nonetheless, there are indications that its nonmetabolic role also influences cellular processes [32]; however, its nonmetabolic role in chondrocytes remains unclear. In previous experiments, we found that in chondrocytes stimulated with DOX, the expression of ENO2 increased in both the cytoplasm and nucleus (Fig. 1O). Therefore, we speculated that nuclear ENO2 exerts its nonmetabolic function through protein interactions. To substantiate this hypothesis, we conducted additional analysis on the nuclear expression of ENO2, which suggested that DOX induced an increase in ENO2 expression within the nucleus (Fig. 3A). Subsequently, we further searched for proteins that may interact with ENO2 in the nucleus through co-IP and mass spectrometry analysis, thereby GNL3 was discovered owing to its remarkable enrichment level (Supplementary Figure S3A). GNL3, also known as nucleostemin, is a protein located within the nucleolus. It has been reported to assist in regulating cell growth, preserving stem cell characteristics, influencing tumor development, and preventing cell death [37–41]. Fluorescence imaging was performed, and ENO2 and GNL3 were found to sit together in the nucleus (Fig. 3B). We further confirmed the interaction between the two by immunoprecipitation and GST pull-down assays (Fig. 3C, 3D). Computer simulations indicated that Glu-4 (E4), Arg-9 (R9), and Arg-32 (R32) of ENO2 might be the main amino acid sites that interact and bind with GNL3 (Fig. 3E). Mutation of the E4 site of ENO2 to alanine (E4A) abolished its interaction with GNL3 (Fig. 3F). Remarkably, this single mutation rescued all the previously observed pathological phenotypes, restored anabolic gene expression (Fig. 3G), suppressed senescence markers (Fig. 3H), and inhibited catabolic enzymes (Fig. 3I). Accordingly, proteoglycan content was maintained while degradation markers were reduced (Fig. 3J). Consistent with these findings, E4A mutation significantly attenuated cellular senescence (Fig. 3K). These results collectively demonstrate that the physical interaction between ENO2 and GNL3 is essential for ENO2 to drive chondrocyte senescence, and disruption of this interaction abrogates the pro-senescent effect of ENO2 in chondrocytes. Considering that ENO2 is a glycolytic enzyme responsible for promoting the production of lactic acid, we explored the potential impact of lactic acid on the ENO2–GNL3 interaction. Our findings indicated that lactic acid augmented the association between ENO2 and GNL3. In contrast, Galloflavin, which inhibits lactic acid production, interfered with this association (Fig. 3L). We further propose that this interaction is facilitated by the lactylation of GNL3, suggesting a novel mechanism through which metabolism regulates protein binding and cellular senescence. First, we confirmed that GNL3 is indeed subject to lactylation modification (Fig. 3M). To directly investigate whether ENO2 contributes to this post-translational modification of GNL3, we performed a metabolic tracing assay using biotin-conjugated glucose to track the incorporation of glucose-derived metabolites into proteins. We observed robust biotin signals in GNL3 immunoprecipitates from cells grown in the presence of biotin–glucose, confirming that GNL3 is modified by metabolites originating from glucose. Importantly, this modification was substantially diminished in sg-ENO2-knockout cells, suggesting that ENO2 is essential for the integration of glucose-derived metabolites into GNL3, which in turn enables its lactylation (Fig. 3N). Using lactylomics analysis, we identified five potential lactylation sites in GNL3 (Fig. 3O), among which Lys-5 (K5) was the primary lactylation site (Fig. 3P). Building on these findings, mutation of the Lys-5 site to arginine (K5R) significantly weakened the binding between ENO2 and GNL3 (Fig. 3Q). These findings reveal a novel nonmetabolic function of the glycolytic enzyme ENO2 in the nucleus, where it interacts with GNL3 to drive chondrocyte senescence. A critical question that emerged from our findings was how the ENO2–GNL3 complex exerts its detrimental effects.

Fig. 3.

Intranuclear ENO2 directly binds to GNL3 and modulates its lactylation to regulate chondrocyte senescence. A Western blot analysis of ENO2 in nuclear and cytoplasmic fractions of DOX-treated chondrocytes. B Immunofluorescence co-localization of ENO2 and GNL3 in chondrocytes. C Co-IP validation of ENO2-GNL3 interaction. D GST pull-down assay confirming direct binding between ENO2 and GNL3. E Molecular docking simulation of the ENO2–GNL3 interaction interface. F Co-IP analysis of ENO2–GNL3 interaction following site-directed mutagenesis of ENO2. G Western blot analysis of collagen II and SOX9 in chondrocytes overexpressing ENO2-E4A mutant. H Western blot analysis of p53 and p16 in chondrocytes overexpressing ENO2-E4A mutant. I Western blot analysis of MMP13 and MMP3 in chondrocytes overexpressing ENO2-E4A mutant. J Fluorescence imaging of ACAN and ADAMTS4 in chondrocytes overexpressing ENO2-E4A mutant. K β-galactosidase staining of chondrocytes overexpressing ENO2-E4A mutant. L Co-IP analysis of ENO2–GNL3 interaction under lactate or galloflavin treatment. M IP detection of lactylation on GNL3. N Glucose–biotin incorporation assay to assess GNL3 lactylation following ENO2 silencing. O Mass spectrometry identification of potential lactylation sites on GNL3. P IP assessment of GNL3 lactylation after mutation of predicted sites. Q Co-IP analysis of ENO2–GNL3 interaction upon mutation of GNL3 lactylation sites. ^*P ≤ 0.05, ^**P ≤ 0.01, ^***P ≤ 0.001

ENO2 promotes chondrocyte senescence by competitively displacing MDM2 from lactylated GNL3

To determine whether the ENO2–GNL3 interaction regulates GNL3 stability or expression, we overexpressed ENO2 and assessed GNL3 levels. Notably, GNL3 expression was not affected by ENO2 overexpression (Supplementary Figure S3B). We overexpressed GNL3 (Supplementary Figure S3C) and found no affect on extracellular matrix metabolism (Supplementary Figure S3D, S3E, S3F), apoptosis (Supplementary Figure S3G), glycolysis (Supplementary Figure S3H), or glucose uptake (Supplementary Figure S3I, S3J) in chondrocytes. These results suggested that GNL3 alone was insufficient to induce the previously observed pathological changes and required an interaction with ENO2 to exert its effects. Unexpectedly, however, overexpression of GNL3 reduced the expression of senescence markers such as p53, p21, and p16 (Fig. 4A) and decreased the proportion of senescent cells (Fig. 4C). This effect is in direct contrast to the function of ENO2; whereas ENO2 appears to promote senescence, GNL3 may exert a senescence-suppressive effect. A review of existing literature revealed that GNL3 binds to and stabilizes MDM2 [42–45], an E3 ubiquitin ligase known to promote the degradation of p53 [46, 47]. We confirmed that this interaction also occurred in chondrocytes (Fig. 4C). Notably, the overexpression of GNL3 enhanced the stability of MDM2 (Fig. 4D) and subsequently increased its abundance (Fig. 4E). These observations are consistent with a model in which MDM2 stabilization facilitates p53 degradation, thereby attenuating cellular senescence. To elucidate the role of ENO2 in this mechanism, we further demonstrated that the lactylation-deficient form of GNL3 exhibited enhanced binding affinity for MDM2 (Supplementary Figure S4A). This suggests that lactylation of GNL3 enables competitive binding between ENO2 and MDM2 for GNL3 interaction. ENO2 reduced MDM2 expression (Fig. 4F) via a mechanism independent of direct binding (Fig. 4G) and transcriptional regulation (Fig. 4H). Instead, as demonstrated by cycloheximide chase assays, it accelerates the turnover rate of MDM2 (Fig. 4I) by promoting its ubiquitination (Fig. 4J), which ultimately leads to enhanced proteasomal degradation. A critical test of this model involved knockdown of GNL3 (Supplementary Figure S4B). Under these conditions, ENO2 overexpression failed to promote MDM2 degradation (Supplementary Figure S4C), demonstrating that the ability of ENO2 to destabilize MDM2 was strictly dependent on the presence of GNL3, and more specifically, on its lactylated form. We hypothesized that ENO2 binding to lactylated GNL3 competitively displaced MDM2. This model was strongly supported by co-IP experiments. When GNL3 was immunoprecipitated from cells overexpressing ENO2, significantly less MDM2 was recovered from the complex (Fig. 4K). Conversely, disrupting the ENO2–GNL3 interaction (via the E4A mutation) restored MDM2 protein levels (Fig. 4L), slowed its turnover rate (Fig. 4M), and reduced its ubiquitination (Fig. 4N). Consistent with this, immunoprecipitation of GNL3 from cells expressing the interaction-deficient ENO2 (E4A) mutant showed that GNL3–MDM2 binding was restored to normal levels (Fig. 4O). By integrating these findings, we demonstrated that ENO2 utilizes its Glu-4 residue to disrupt the GNL3–MDM2 interaction, effectively displacing MDM2 from its complex with GNL3. This displacement destabilizes MDM2, leading to its accelerated proteasomal degradation. The subsequent reduction in MDM2 protein levels impairs p53 degradation, resulting in p53 accumulation, ultimately promoting chondrocyte senescence.

Fig. 4.

ENO2 disrupted GNL3–MDM2 interaction to stabilize p53. A Western blot analysis of p53, p21, and p16 in chondrocytes overexpressing GNL3. B Flow cytometric C12-FDG quantification in chondrocytes overexpressing GNL3. C Co-IP validation of endogenous GNL3–MDM2 interaction. D Ubiquitination assay of MDM2 upon GNL3 overexpression. E Western blot quantification of MDM2 upon GNL3 overexpression. F Western blot quantification of MDM2 upon ENO2 overexpression. G GST pull-down assay confirming the absence of direct interaction between ENO2 and MDM2. H quantitative PCR with reverse transcription (qRT–PCR) analysis of MDM2 mRNA upon ENO2 overexpression. I Western blot-based MDM2 protein degradation upon ENO2 overexpression. J Ubiquitination assessment of MDM2 upon ENO2 overexpression. K Co-IP evaluation of GNL3–MDM2 interaction upon ENO2 overexpression. L Western blot analysis of MDM2 and p53 following the E4A mutation. M MDM2 protein degradation and N ubiquitination assays following the E4A mutation. O Co-IP analysis of GNL3–MDM2 interaction following the E4A mutation. ^*P ≤ 0.05, ^**P ≤ 0.01, ^***P ≤ 0.001

p53 transcriptionally activates ENO2 to form a positive-feedback loop

On the basis of previous reports that p53 can enhance glycolysis [48, 49], we investigated whether p53 also regulates ENO2 expression through transcriptional mechanisms. Initially, we overexpressed p53 in chondrocytes (Supplementary Figure S5A) and found it significantly increased both ENO2 protein and mRNA levels (Figs. 5A, 5B). To elucidate the underlying mechanism, we first assessed whether p53 stabilizes ENO2 transcripts. Actinomycin D-based transcriptional inhibition assays revealed no significant differences in ENO2 mRNA decay rates (Fig. 5C), indicating that post-transcriptional mRNA stabilization was not involved. Neither the rate of ENO2 protein degradation nor its ubiquitination status was altered by p53 overexpression (Figs. 5D, E), ruling out post-translational regulation. Subsequent experiments using the p53 activator RITA corroborated these findings, showing elevated ENO2 mRNA and protein levels (Figs. 5F and G). Conversely, p53 knockdown using shRNA (Supplementary Figure S5B) or inhibition with PFTα significantly reduced ENO2 expression (Figs. 5H, I, J, K). Consistently, the same results were observed in chondrocytes upon sg-p53-mediated knockdown (Supplementary Figure S5C, S5D, S5E). Bioinformatics analysis using the JASPAR database identified putative p53 binding sites within the ENO2 promoter region (Fig. 5L). Chromatin immunoprecipitation (ChIP) and luciferase reporter assays confirmed specific p53 binding and transactivation of the ENO2 promoter (Figs. 5M and N). To specifically identify the functional binding site of p53 on the ENO2 promoter, we individually mutated the top three potential p53 binding sites predicted by the JASPAR database. We found that mutation of the first site resulted in a significant reduction in p53 binding to the ENO2 promoter (Supplementary Figure S5F). Collectively, these data establish a positive feedback loop wherein p53 transcriptionally upregulates ENO2, which in turn promotes p53 accumulation by disrupting the GNL3–MDM2–p53 axis. This self-reinforcing circuit amplified glycolysis and accelerated chondrocyte senescence, thereby driving cartilage degeneration.

Fig. 5.

p53 transcriptional activation of ENO2 in chondrocytes A Western blot analysis of ENO2 protein in chondrocytes overexpressing p53. B qRT–PCR quantification of ENO2 mRNA in chondrocytes overexpressing p53. C qRT–PCR quantification of ENO2 mRNA decay in the presence of Actinomycin D (Act D). D Western blot analysis of ENO2 protein degradation after cycloheximide treatment. E IP-based ubiquitination assay of ENO2 in chondrocytes overexpressing p53. F Western blot analysis of ENO2 after 1 nM RITA treatment. G qPCR of ENO2 mRNA after RITA treatment. H Western blot analysis of ENO2 following p53 knockdown. I qRT–PCR quantification of ENO2 mRNA following p53 knockdown. J Western blot analysis of ENO2 after 1 μM pifithrin-α hydrobromide (PFTa) treatment. K qRT–PCR quantification of ENO2 mRNA after PFTα treatment. L JASPAR motif prediction for p53 binding to the ENO2 promoter. M ChIP–qPCR detection of p53 occupancy at the ENO2 promoter. N Luciferase assay measuring p53-driven ENO2 promoter activity. ^*P ≤ 0.05, ^**P ≤ 0.01, ^***P ≤ 0.001

Pharmacological ENO2 inhibition breaks the metabolic-senescence loop and attenuates age-related OA

Despite the high prevalence of age-related OA, effective disease-modifying therapies are limited. Having identified the ENO2–GNL3–MDM2–p53 axis as a critical driver of chondrocyte senescence and pathological glycolysis, we evaluated its therapeutic potential. First, we treated DOX-induced senescent chondrocytes with POMHEX, a specific ENO2 inhibitor. POMHEX significantly reduced lactate production (Fig. 6A), disrupted the ENO2–GNL3 complex (Supplementary Figure S6A), and restored MDM2–p53 binding (Supplementary Figure S6B), thereby correcting the underlying molecular imbalance. These changes translated into functional improvements: POMHEX suppressed senescence markers (Fig. 6B), reduced catabolic factors (Fig. 6C), and restored anabolic gene expression (Fig. 6D). In addition, treatment with POMHEX reduced the number of senescent cells (Fig. 6E) and apoptosis (Fig. 6F) induced by DOX. Since ENO1 and ENO2 are homologous, numerous previous studies have demonstrated that POMHEX exhibits very low activity against ENO1 but high specificity for ENO2 [32, 50, 51]. To investigate and rule out the potential contribution of ENO1 to the observed effects of POMHEX, we first knocked out ENO1 in chondrocytes (Supplementary Figure S6C). We found that POMHEX significantly reduced intracellular enolase activity in these ENO1-knockout cells (Supplementary Figure S6D). In contrast, when ENO2 was knocked out, POMHEX had minimal effect on cellular enolase activity (Supplementary Figure S6E). Furthermore, under ENO2-knockdown conditions, POMHEX failed to rescue the DOX-induced functional changes (Supplementary Figure S6F) and increased apoptosis (Supplementary Figure S6G) in chondrocytes.

Fig. 6.

Pharmacological inhibition of ENO2 in vitro and in vivo. A Lactate levels in chondrocytes stimulated with 1 μM DOX and treated with 1 μM POMHEX. B Western blot analysis of p53 and p16 in DOX-treated chondrocytes with or without POMHEX. C Western blot analysis of MMP13 and MMP3 in DOX-treated chondrocytes with or without POMHEX. D Western blot analysis of collagen II and SOX9 in DOX-treated chondrocytes with or without POMHEX. E Flow cytometric C12-FDG assay in DOX-treated chondrocytes with or without POMHEX. F Flow cytometric analysis of apoptosis in DOX-treated chondrocytes with or without POMHEX. G Safranin O staining, collagen II immunohistochemistry, and immunofluorescence staining for MMP3 and MMP13 in articular cartilage from young and aged mice with or without 10 mg/kg POMHEX treatment. H Western blot analysis of collagen II, MMP13, and p53 in articular cartilage from young and aged mice with or without POMHEX treatment. I Representative micro-CT images of knee joints from young and aged mice with or without POMHEX treatment. J Hot-plate, von Frey, and knee-flexion vocalization assays evaluating pain relief in young and aged mice with or without POMHEX treatment. K Gait analysis including step length and stride length in young and aged mice with or without POMHEX treatment. ^*P ≤ 0.05, ^**P ≤ 0.01,^***P ≤ 0.001

Next, we assessed the efficacy of POMHEX in a naturally aging mouse model of OA (18-month-old C57BL/6J mice). To first confirm the in vivo efficacy of POMHEX on its target, we measured enolase activity in chondrocytes isolated from the articular cartilage of mice treated with POMHEX. As expected, intracellular enolase activity was significantly decreased (Supplementary Figure S6H). In addition, co-IP assays performed on cartilage tissue lysates demonstrated that POMHEX treatment reduced the interaction between ENO2 and GNL3 in vivo (Supplementary Figure S6I), which was consistent with our in vitro observations. Weekly intra-articular administration of POMHEX for 16 weeks markedly improved cartilage integrity, as evidenced by smoother surfaces, enhanced proteoglycan content, and elevated expression of cartilage anabolic markers compared with vehicle-treated controls (Fig. 6G, H, Supplementary Fig. S6J, S6K). Micro-CT analysis revealed reduced osteophyte formation and improved subchondral bone architecture (Fig. 6I; Supplementary Figures S6L, S6M), which correlated with significantly lower OA pathology scores. Importantly, POMHEX treatment alleviated pain sensitivity and improved functional outcomes, including gait normalization and increased the mechanical threshold (Fig. 6J, K).

Collectively, these results demonstrate that the pharmacological inhibition of ENO2 disrupts the senescence–glycolysis feedback loop, attenuates structural degeneration, and ameliorates pain and disability in aged OA mice. Thus, our study identified ENO2 as a tractable therapeutic target and proposes a promising strategy for treating age-related OA.

Discussion

This study systematically investigated the role of the glycolytic enzyme ENO2 in age-related OA and revealed its multifaceted contributions to disease progression through metabolic reprogramming, epigenetic regulation, and senescence amplification. Clinical and molecular analyses demonstrated that ENO2 is markedly upregulated in aged OA cartilage, where it undergoes nuclear translocation and facilitates lactylation-dependent interactions. Mechanistically, ENO2 competitively binds to GNL3, disrupting the GNL3–MDM2 complex and stabilizing p53, thereby establishing a self-reinforcing loop. Pharmacological inhibition of ENO2 with POMHEX alleviated cartilage degeneration, osteophyte formation, and pain in aged mice, underscoring its therapeutic potential. These findings bridge metabolic dysregulation and cellular senescence and offer new insights into the pathogenesis of OA.

Our work challenges the conventional view of metabolic enzymes as mere catalysts by uncovering ENO2’s noncanonical roles. While ENO2 has been implicated in cancer progression through β-catenin activation, its function in OA diverges sharply, driving senescence via p53 stabilization rather than proliferation [52]. This tissue-specific duality highlights the context-dependent nature of metabolic enzyme function. Importantly, we identified lactate, a byproduct of ENO2-driven glycolysis, as both a metabolic intermediate and a mediator of post-translational modifications. The lactylation of GNL3 at K5 demonstrates how metabolic flux directly reshapes protein interaction networks, expanding the concept of “metabolic memory” into degenerative diseases. This mechanism aligns with emerging evidence on the regulatory roles of lactate, but extends it to senescence-associated pathways, providing a missing link between glycolysis and epigenetic reprogramming in OA.

The discovery of an ENO2–p53 positive feedback loop represents a paradigm shift in our understanding of age-related OA progression. Unlike transient p53 activation in acute stress responses, chronic ENO2-driven p53 accumulation likely surpasses physiological thresholds, shifting the balance from protective cell cycle arrest to destructive senescence. This mechanism may explain the age-dependent acceleration of OA in which low-grade metabolic disturbances trigger irreversible phenotypic changes. Furthermore, ENO2’s dual localization of orchestrating cytoplasmic glycolysis and nuclear p53 transcription illustrates how the spatial regulation of metabolic enzymes integrates disparate cellular processes. This spatiotemporal coordination offers a framework for studying compartmentalized signaling in other age-related disorders.

Limitations of the study

The limitations of this study include that the drivers of ENO2 nuclear translocation remain elusive. While senescence induces this process, upstream regulators such as oxidative stress or phosphorylation events (e.g., protein kinase B -mediated modifications) need to be explored. Second, although we demonstrated GNL3-K5 lactylation, direct evidence linking ENO2-derived lactate to this modification requires isotopic tracing. Third, POMHEX’s dual inhibition of the enzymatic and scaffolding functions by POMHEX complicates therapeutic optimization, necessitating domain-specific inhibitors to minimize off-target effects. Finally, the naturally aged mouse model, which is physiologically relevant, may not fully recapitulate the human OA heterogeneity, which involves obesity, trauma, and genetic factors. Future studies should address these gaps through single-cell multi-omics of human OA subtypes and spatial metabolomics to map microenvironmental lactate dynamics.

Translational implications have emerged from the conserved yet adaptable nature of the ENO2–GNL3–MDM2 axis. While our focus was on OA, this pathway may operate in other age-related conditions (e.g., intervertebral disc degeneration), suggesting broad relevance. The efficacy of POMHEX in preclinical models highlighted ENO2 as a druggable target, although clinical translation requires refined delivery strategies (e.g., hydrogel-based slow release) to mitigate repeated injection-related cartilage damage.

In this context, we also explored the potential role of ENO1. Proteomic profiling and western blot analysis revealed no significant upregulation of ENO1 in aged osteoarthritic cartilage compared with young control cartilage. This finding stands in contrast to previous reports that have linked ENO1 to the pathogenesis of OA [53]. The discrepancy may be attributed to differences in the model systems used: those studies were primarily performed in the destabilization of the medical mensicus (DMM)-induced OA model [54], whereas our study employed naturally aged mice. Some studies have observed upregulation of ENO1 mainly in synovial tissue rather than in cartilage, and they demonstrated elevated expression and pro-inflammatory function of ENO1 in rheumatoid arthritis but not in osteoarthritis [55]. Taken together, our findings indicate that ENO2, rather than ENO1, is the predominant isoform driving age-related OA progression, although the ENO2–ENO1 interaction we observed may have context-dependent functional significance that warrants further investigation.

In conclusion, this study redefined ENO2 as a pleiotropic regulator that coordinates the metabolic, epigenetic, and senescence pathways in OA. By elucidating the ENO2–p53 feedback loop and lactylation-mediated protein interactions, we provide a mechanistic foundation for targeting the metabolism–senescence axis, a strategy with transformative potential for age-related degenerative diseases. Future efforts to dissect ENO2’s post-translational landscape and tissue-specific roles would further advance precision therapeutics for OA.

Abbreviations

ANOVA

Analysis of variance

ChIP

Chromatin immunoprecipitation

Co-IP

Co-immunoprecipitation

DOX

Doxorubicin

ECAR

Extracellular acidification rate

ENO2

Enolase 2

KEGG

Kyoto Encyclopedia of Genes and Genomes

LC–MS/MS

liquid chromatography–tandem mass spectrometry

OA

Osteoarthritis

PBS

Phosphate-buffered saline

PCR

Polymerase chain reaction

PEP

Phosphoenolpyruvate

SDS–PAGE

Sodium dodecyl sulfate–polyacrylamide gel electrophoresis

Author contribution

Author Feng Hua (co-first author): conceptualization, methodology, investigation, formal analysis, writing—original draft. Author Jiangyu Nan (co-first author): methodology, investigation, data curation, visualization, writing—original draft. Author Rong Wu: investigation, software, validation, writing—review and editing. Author Bin Zhang: investigation, resources, data curation, writing—review and editing. Author Qimeng Liu: methodology, formal analysis, validation, writing—review and editing. Author Tianliang Ma: project administration, resources. Author Zheyu Zhang: supervision, conceptualization, writing—review and editing. Author Yihe Hu and Jie Xie (co-corresponding author): supervision, project administration, writing—review and editing, corresponding author duties. Author Yute Yang (co-corresponding author): supervision, conceptualization, funding acquisition, writing—review and editing, corresponding author duties.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and publication of this article: this study was supported by the National Natural Science Foundation of China (82202733) and the Zhejiang Provincial Natural Science Foundation of China (LY24H060004).

Data availability

The datasets are available from the corresponding author upon reasonable request. All authors have agreed to publish this manuscript.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the First Affiliated Hospital of Zhejiang University School of Medicine (approval no. IIT20230305B; date: 5 March 2023), and informed consent was obtained from all participants prior to enrollment. All animal procedures were conducted in compliance with the International Council for Laboratory Animal Science (ICLAS) guidelines and approved by the Animal Care and Use Committee of the First Affiliated Hospital of Zhejiang University School of Medicine (approval no. 2023 No.555; date: 14 February 2023). All methods are reported in accordance with Animal Research: Reporting of in Vivo Experiments (ARRIVE) guidelines.

Competing interests

The authors declare no competing interests.