Dysregulation of the glycolysis–mitochondrial metabolism axis in osteoarthritis: mechanisms and therapeutic implications Authors: Ren Zhang, Su-Zhen Chen, Shang-Chun Guo, Shi-Cong Tao Published: 2026 ABSTRACT AbstractOsteoarthritis (OA) is a chronic degenerative joint disease affecting more than 595 million people worldwide. Progressive cartilage degeneration is closely linked to metabolic reprogramming in chondrocytes. Unlike other joint-resident cells, chondrocytes rely on a distinctive, largely glycolysis-dependent mode of glucose utilisation that must be tightly regulated to maintain cellular homeostasis. Glucose metabolism spans cytosolic glycolysis and mitochondrial oxidative phosphorylation, and growing research has clarified how each contributes to the pathophysiology of OA. Building on these advances, we synthesise current mechanistic evidence to conceptualise chondrocyte glucose metabolism as an integrated continuum, proposing the glycolysis–mitochondrial metabolism–OA axis as a unifying framework. We outline how axis disruption provides a coherent lens for understanding metabolic dysfunction in OA, and we highlight mechanistic insights with translational potential, including candidate targets that may support future diagnosis and therapy. Together, this literature review aims to sharpen the conceptual link between metabolic homeostasis and cartilage integrity, and to guide the development of mechanism-informed strategies for degenerative joint disease.The translational potential of this articleThis literature review outlines the mechanistic framework of a glycolysis–mitochondrial metabolism–OA axis and highlights actionable biomarkers and integrated therapeutic strategies that may guide the development of dual-target interventions for precision OA treatment.Keywords: Chondrocytes, Glycolysis–mitochondrial metabolism axis, Metabolic reprogramming, Mitochondrial dysfunction, Osteoarthritis, Oxidative stress Introduction Osteoarthritis (OA) is a chronic degenerative joint disease characterised by progressive cartilage degradation, subchondral bone remodelling, and synovial inflammation. The aetiology of OA is primarily associated with sex, obesity, ageing, genetic predisposition, structural abnormalities or malalignment of the joints, and other contributing factors [1]. Its high prevalence and substantial disability burden pose a formidable global public health challenge, affecting over 595 million individuals worldwide, with case numbers projected to rise with the ageing population [2]. Currently, there is no curative treatment for OA. In clinical practice, non-steroidal anti-inflammatory drugs (NSAIDs) only alleviate symptoms, hyaluronic acid exhibits limited reparative effects, and joint replacement surgery is highly invasive. In the face of these therapeutic challenges, elucidating the underlying mechanisms is an essential foundation for breakthroughs in treatment. OA treatments predominantly focus on anti-inflammatory and symptomatic management; however, patients may develop treatment resistance due to irreversible metabolic damage in chondrocytes (e.g. adenosine triphosphate [ATP] insufficiency resulting from aberrant glycolysis and mitochondrial dysfunction), further escalating therapeutic challenges. Chondrocytes exhibit a distinctive metabolic profile, deriving ∼95% of their ATP from glucose metabolism [3]. Even under aerobic conditions, they predominantly rely on glycolysis for energy production, while mitochondria primarily sustain redox balance. Early studies established that this glycolytic preference aligns closely with the hypoxic joint microenvironment [4], while moderate mitochondrial activity helps minimise reactive oxygen species (ROS) generation, thereby protecting the cartilage matrix from oxidative damage [5]. Given the distinctive metabolic programme of glucose utilisation in chondrocytes and the central importance of glucose metabolic homeostasis for their function, current clinical strategies remain markedly limited. Specifically, no established therapeutic approach targets OA driven by dysregulated glucose metabolism. At the same time, existing reviews have largely focused on the effects of either glycolysis or mitochondrial function in isolation [6,7]. Importantly, cellular glucose metabolism is inherently continuous and integrated, with substantial crosstalk and feedback between pathways. Consequently, even at the conceptual level, a coherent framework for a “glycolysis–mitochondrial metabolism axis” has yet to be established, and the full set of interactions and feedback mechanisms across the pathway have not been comprehensively interrogated. The clinical potential of framing glucose metabolism as a unified, continuous process has likewise not been sufficiently explored. To clarify how aberrant glycolysis reshapes downstream mitochondrial metabolism, and how mitochondrial metabolic dysfunction reciprocally aggravates glycolytic abnormalities, we propose a glycolysis–mitochondrial metabolism–OA metabolic axis (Fig. 1). First, we delineate the pivotal enzymes within this axis, describing how they influence mitochondrial glucose metabolism through direct effects or via aberrant metabolic intermediates. Next, we discuss how mitochondrial metabolic derangements promote OA progression. This framework provides a conceptual basis for developing metabolism-targeted therapies aimed at restoring cellular homeostasis, and may offer broader insights for intervention strategies across degenerative diseases. Unlike previous reviews that examined glycolysis or mitochondrial metabolism in isolation, we link glucose metabolism, mitochondrial function, and OA into a unified, comprehensive axis. Defining this axis expands the metabolic perspective on OA, supports the theoretical feasibility of translating mechanistic insights into clinical applications, and enables evaluation of the translational potential of existing mechanistic studies to inform future therapeutic development. Glycolysis–mitochondrial metabolism axis dysregulation in OA. Pathogenic factors, including inflammation, hypoxia, and mechanical stress, can induce glycolytic abnormalities and mitochondrial dysfunction, initiating a vicious cycle. Metabolic products from aberrational glycolysis (e.g., lactate and AGEs) and dysregulated signalling pathways (e.g., PI3K-AKT and HIF-1α) impair mitochondrial dynamics and bioenergetics, while mitochondrial injury destabilises glycolytic homeostasis. This cyclical imbalance promotes ECM degradation, chondrocyte apoptosis, and synovial inflammation, ultimately driving OA progression. Illustration credit: Lina Cao. Glycolytic aberrations in OA Chondrocytes derive ∼95% of their ATP from glucose metabolism [3,8]. The continuous pathways of glucose metabolism collectively maintain chondrocyte homeostasis, which in turn profoundly shapes the pathogenesis and progression of OA. Glycolysis comprises multiple enzyme-catalysed, sequential reactions, and dysregulation at several key nodes can trigger OA-associated metabolic dysfunction, as shown in Fig. 2. Recent studies indicate that glucose metabolic homeostasis is foundational for maintaining chondrocyte function, and its disruption accelerates OA through mechanisms such as chondrocyte hypertrophy and extracellular matrix (ECM) degradation [9]. Although the continuous regulatory mechanisms governing glucose metabolism remain incompletely understood, existing evidence highlights the direct impact of glycolytic abnormalities on mitochondrial function. Key glycolytic enzymes and their roles in OA-related metabolic dysfunction. GLUT-1 mediates excessive glucose uptake, accelerating AGE formation and triggering NF-κB-dependent inflammatory responses. Hyperactivation of HK2 promotes its dissociation from VDAC, leading to mitochondrial Ca2+ overload and mitochondrial fragmentation. PFKFB3 activates PFK1 to enhance glycolytic flux. GAPDH drives metabolic reprogramming towards glycolysis via the Akt/p38 signalling pathway. ENO1 downregulates ECM anabolic markers (aggrecan and type II collagen) and upregulates catabolic enzymes (MMP3 and MMP13). PKM2 translocates to the nucleus and binds the HIF-1α (HIF1A) promoter, amplifying both glycolysis and inflammatory signalling. LDHA promotes lactate accumulation, which induces ROS production. Created with BioRender.com. Below, we expand on this theme by examining the interplay within the glycolysis–mitochondrial metabolism–OA axis, providing both mechanistic insight and translational perspectives for clinical practice. OA-related activity changes in key glycolytic enzymes The distinctive metabolic programme of the OA microenvironment, together with the high energetic demands imposed by inflammation, enhances glycolysis in chondrocytes. Glucose uptake increases, and the activities of a series of glycolytic enzymes are altered, promoting faster and greater flux through glucose metabolism to meet the energy requirements of inflammatory stress. These enzymatic changes reshape downstream mitochondrial function and, in some cases, directly influence the progression of OA. Below, we summarise how alterations in key glycolytic enzymes affect mitochondrial metabolism and contribute to OA pathogenesis. Glucose transporters (GLUTs) GLUTs are membrane proteins that serve as critical mediators for glucose entry into chondrocytes [10]. GLUT1 and GLUT3 ensure efficient glucose uptake, with GLUT1 highly adaptive to environmental conditions [11]. Specifically, GLUT1 is upregulated under hypoxia or glucose deprivation and downregulated in high-glucose conditions, helping to maintain balanced cellular glucose levels [12]. Research indicates that GLUTs are directly involved in OA progression [13]. In human OA articular cartilage, GLUT1 expression correlates closely with disease severity [14]. Sustained overexpression of GLUT1 can drive excessive glucose uptake, accelerating the formation of advanced glycation end products (AGEs) and thereby promoting proteoglycan depletion and chondrocyte dysfunction [9,15]. Paradoxically, complete GLUT1 ablation in chondrocytes induces apoptosis and proteoglycan loss, ultimately resulting in fibrosis, and direct inhibition of GLUT1 exacerbates OA in mice [16,17]. Interestingly, treatment with nitazoxanide can activate AMP-activated protein kinase (AMPK) and inhibit the mechanistic target of rapamycin (mTOR) complex 1 (mTORC1) pathway, thereby downregulating GLUT1. This effectively reverses interleukin (IL)-1β-induced glycolytic dysregulation, exerting a pronounced anti-inflammatory effect in primary chondrocytes from affected OA patients [18]. In macrophages, IL-38 suppresses M1 polarisation and inflammation partly by inhibiting GLUT1 expression [19]. These findings suggest that, in chondrocytes, GLUT1 expression likely reflects an adaptive response to cellular energy demand, and that direct intervention could disrupt this energy balance, further compromising cellular function. Recent findings demonstrate that 5-hydroxymethylfurfural (5-HMF), an anti-inflammatory compound derived from a root used in traditional herbal medicine, ameliorates disease progression in mice with destabilisation of medial meniscus (DMM)-induced OA. This protective effect is associated with significant downregulation of chondrocyte Glut1 expression. Notably, Glut1 deletion abolishes the benefits of 5-HMF, underscoring its pivotal role in chondrocyte metabolism and confirming its potential as a therapeutic target in OA [20]. Hexokinase (HK) HK catalyses the initial step of aerobic glycolysis and serves as a key regulator of glucose metabolism, with profound impacts on glycolysis–mitochondrial metabolism–OA axis dynamics and metabolic imbalance in OA. The HK family encompasses five isozymes in mammals, among which the acetylation status of Lys15 and Lys21 in the N-terminal mitochondrial binding domain of HK1 regulates its mitochondrial association. Upon deacetylation of these residues, HK1 dissociates into the cytosol and forms a complex with S100A8/A9. This complex recruits inducible nitric oxide synthase (iNOS), which mediates S-nitrosylation of glyceraldehyde-3-phosphate dehydrogenase (GAPDH), leading to its inactivation and subsequent promotion of inflammatory responses [21]. HK2 has emerged as the principal regulatory isoform across diverse tissue-derived cell lines, with predominant expression in musculoskeletal and cardiac cells, positioning it as a key contributor to OA pathogenesis [22]. Under physiological conditions, HK2 binds to mitochondrial voltage-dependent anion channel (VDAC), efficiently utilising mitochondrial ATP to initiate glycolysis while maintaining coupling with oxidative phosphorylation (OXPHOS), thereby adapting to the hypoxic joint environment. RNA sequencing has revealed significant upregulation of HK2-related glycolytic genes in OA [23]. In mechanistic studies in HeLa cells, this aberrant activation has been shown to trigger a cascade of events. Excessive glucose phosphorylation promotes lactate accumulation, acidifying the mitochondrial matrix and suppressing the mitochondrial fusion proteins mitofusin (MFN)1 and optic atrophy 1 (OPA1). In parallel, dissociation of HK2 from mitochondria induces Ca2+ overload, activates calpains, and exacerbates mitochondrial depolarisation [24,25]. In studies on bone marrow-derived macrophages, elevated mitochondrial Ca2+ concentrations triggered VDAC oligomerisation, forming macromolecular pores that facilitate the translocation of mitochondrial DNA (mtDNA) and proteins across the outer mitochondrial membrane and into the cytosol. The released mtDNA subsequently recruits NOD-like receptor pyrin domain containing 3 (NLRP3) inflammasome components and promotes their association with VDAC oligomers, ultimately activating the inflammasome and driving the maturation and secretion of pro-inflammatory cytokines (e.g., IL-1β and IL-18) [26]. Recently, mechanical stress was shown to further upregulate HK2 through the piezo-type mechanosensitive ion channel component 1 (PIEZO1)/signal transducer and activator of transcription 1 (STAT1) pathway, forming a vicious cycle of mechanical stimulation and metabolic hyperactivity [27]. Alternatively, during periods of energy stress, studies on HeLa cells have shown that HK1 forms rings around mitochondria that inhibit mitochondrial fission by sequestering dynamin-related protein 1 (Drp1) from mitochondrial fission factor (MFF) and mitochondrial fission 1 protein (Fis1). During energy recovery, mitochondrial fission is associated with the disassembly of these HK1 rings [28]. In prostate cancer cells, inhibiting VDAC1 oligomerisation has been shown to preserve mitochondrial homeostasis; whether this same mechanism operates in chondrocytes has yet to be determined [29]. Considering these findings, mimicking the mitochondrial binding domain of HK2 using short peptides or VDAC1-binding proteins to prevent its dissociation could represent a feasible approach to reducing mitochondrial fragmentation without affecting glycolysis in healthy chondrocytes. Additionally, localised delivery via hydrogel carriers may further address the challenges of short peptide retention in the joint cavity and targeted release. Phosphofructokinase (PFK) PFK1 serves as a pivotal enzyme in an irreversible step of the glycolytic pathway, primarily catalysing the conversion of fructose-6-phosphate (F6P) to fructose-1,6-bisphosphate (F1,6BP). This reaction consumes one molecule of ATP and depends on magnesium (Mg2+) as a cofactor. Intriguingly, Mg2+ uptake modulates mitochondrial respiration, thereby establishing a reciprocal regulatory relationship between PFK and mitochondria [30]. Among the PFK isoforms, muscle-type PFK (PFKM) is predominantly expressed in skeletal muscle and is encoded by one of 14 high-confidence OA risk genes identified so far [31]. Fructose-2,6-bisphosphatase-3 (PFKFB3), the most potent allosteric activator of PFK1, promotes glycolysis by generating fructose-2,6-bisphosphate, and has been identified as an OA biomarker through gene expression and methylation dataset analyses [32]. PFKFB3 drives lactate accumulation, and the most critical lactylation mark in cartilage, histone H4 lysine 12 lactylation (H4K12la), is selectively enriched at the promoter regions of pro-inflammatory genes such as IL6 and MMP9, enhancing their transcription and exacerbating chondrocyte inflammation and matrix degradation. H4K12la also binds the PFKFB3 promoter, further upregulating its expression and establishing a positive feedback loop [33]. In cardiac endothelial cells, PFKFB3 activates PFK1 by increasing F1,6BP accumulation to enhance glycolysis. This metabolic shift activates the polyol pathway, depleting nicotinamide adenine dinucleotide phosphate (NADPH) and reducing mitochondrial antioxidant capacity; however, whether an analogous mechanism operates in chondrocytes remains to be confirmed [34]. Harpagide, a natural glycoside with antioxidant and anti-inflammatory properties, exhibits strong binding affinity to PFKFB3, inhibiting tumour necrosis factor-α (TNF-α)-induced inflammation in rat articular chondrocytes via glycolytic pathways, suggesting therapeutic potential in OA [35]. In summary, aberrant PFKFB3 activity does not follow a simple “activation–damage” trajectory, but involves a dynamic balance between metabolic adaptation and pathological injury. Intervention with PFKFB3-targeting Harpagide has been validated, and exploring whether the polyol pathway it indirectly activates also operates in joint vascular endothelium or subchondral bone could yield new mechanistic insights for OA-targeted therapies. GAPDH In the glycolytic pathway, GAPDH catalyses the oxidative dehydrogenation and phosphorylation of glyceraldehyde-3-phosphate (GAP) to 1,3-bisphosphoglycerate (1,3BPG), oxidising the GAP aldehyde group to a carboxyl while transferring hydrogen to coenzyme nicotinamide adenine dinucleotide (NAD+), generating nicotinamide adenine dinucleotide plus hydrogen (NADH). Research in chondrocytes has shown that GAPDH activation depends on Akt and p38 signalling, leading to enhanced transcription and translation of related enzymes. Concurrent downregulation of the mitochondrial marker citrate synthase (CS) indicates a metabolic shift from OXPHOS to glycolysis, inhibiting the mitochondrial tricarboxylic acid (TCA) cycle [36]. The OA modelling agent mono-iodoacetate (MIA) inhibits GAPDH, blocking chondrocyte glycolysis and leading to apoptosis and functional abnormalities. This triggers cartilage matrix degradation, and histological analyses of MIA-injected models reveal cartilage fibrosis, proteoglycan loss, and synovial inflammation, ultimately causing joint structural damage [37]. Additionally, GAPDH expression correlates positively with some immune cells (e.g., M0 and M2 macrophages) and negatively with others (e.g., CD8+ T cells), suggesting a role in modulating macrophage polarity and inflammatory responses in OA [38]. Consistently, traditional Chinese medicine studies have identified GAPDH as a potential target in OA, with certain small-molecule compounds exhibiting high binding affinity [38,39], providing valuable clues for drug development. Accordingly, GAPDH may serve as a promising diagnostic marker for OA, where glucose metabolism dysregulation is a dominant feature. Its association with macrophage polarisation also points to a potential metabolism–immune crosstalk pathway. By reshaping the immune microenvironment through metabolic reprogramming and clarifying the causal relationship between GAPDH and immune-cell polarisation, future therapies could be designed to simultaneously target both metabolism and immunity in OA. Enolase (ENO) In mammals, ENO has three primary isozymes encoded by distinct genes. Alpha (α)-ENO, also known as ENO1, is widely distributed across tissues and plays a crucial role in glucose metabolism [40]. In glycolysis, ENO catalyses the conversion of 2-phosphoglycerate to phosphoenolpyruvate, a key step that provides high-energy phosphate compounds for subsequent ATP production. The ENO1 isoform is broadly expressed in various cell types, underscoring its importance in maintaining basic cellular energy metabolism. In cancer cells, ENO1 has been shown to localise to the nucleus and bind nucleic acids, influencing gene transcription and mRNA stability. Through these and other mechanisms, ENO1 contributes to the regulation of cellular proliferation, differentiation, and apoptosis [41]. An oxazole analogue (KB2764) reportedly engages ENO1 at Arg253 and Ser254 and, via pyruvate kinase M (PKM)-dependent control of ENO1 phosphorylation, triggers metabolic reprogramming and restores mitochondrial function in HEK293T cells. This shift redirects cellular energy metabolism from glycolysis towards OXPHOS, increases mitochondrial ATP production, reduces ROS, improves mitochondrial quality, and ameliorates ageing-related phenotypes [42]. In HepG2 cells, cinnamaldehyde targets and inhibits ENO1 activity, restraining gluconeogenic backflow and thereby favouring the preferential utilisation of surplus gluconeogenic substrates (e.g., amino acids) within the TCA cycle. This reprogramming is accompanied by increases in mitochondrial membrane potential (Δψm), ATP levels, and the NAD/NADH ratio, consistent with enhanced mitochondrial energetic efficiency and restoration of glucose metabolic homeostasis [43]. Immortalised human chondrocytes (C-28/I2 cells) induced with IL-1β show markedly increased ENO1 expression over time. IL-1β reduces cell viability, increases apoptosis, and elevates inflammatory cytokines (IL-6 and TNF-α); ENO1 silencing partially reverses these effects, suppressing apoptosis, reducing inflammation, and restoring viability [44]. Experiments show that ENO1 inhibition upregulates the ECM synthesis markers aggrecan and collagen II while downregulating matrix metalloproteinase (MMP) enzymes MMP3 and MMP13, indicating protective effects on chondrocyte ECM in OA [45]. Clinically, ENO1 concentrations in synovial fluid are markedly higher in OA patients than in controls without clinical OA signs, and correlate positively with indices of disease severity, including the functional subscore of the Knee injury and Osteoarthritis Outcome Score (KOOS) and visual analogue scale (VAS) pain score. These findings suggest the potential of ENO1 as an adjunct biomarker for the diagnosis of knee OA. Combining ENO1 with the concomitant downregulation of fibrinogen beta chain (FGB) in synovial fluid may further improve diagnostic specificity [46]. Although links between ENO1 and ROS in chondrocytes remain understudied, clear associations exist in breast cancer research [47], offering cross-disciplinary insights into its potential impact in OA. Additionally, KB2764 initiates beneficial metabolic reprogramming by modulating ENO1 phosphorylation [42]. This approach holds promise for delaying chondrocyte senescence and functional decline at the root level and, compared with conventional inhibitors, represents a more promising pathway for achieving “disease modification”. Pyruvate kinase (PK) In the glycolytic pathway, PK converts phosphoenolpyruvate to pyruvate while generating ATP, serving as a critical rate-limiting enzyme [48]. PK has M1 and M2 isoforms, with PKM2 upregulated in OA chondrocytes. Knockdown of PKM2 in these cells inhibits proliferation and promotes apoptosis, positioning PKM2 as a key regulator of the Warburg effect [49]. Further studies revealed that PKM2 dimerisation induces its nuclear localisation, where it binds to the promoters of downstream target genes (e.g., hypoxia inducible factor-1 alpha [HIF1A]), enhancing their expression and activating inflammation-related IL-1β and glycolytic targets. This establishes a positive feedback loop of sustained glycolytic hyperactivity, lactate accumulation, and amplified inflammatory signalling [50]. PKM2 has also been shown to modulate senescence in primary murine chondrocytes by downregulating the senescence marker p16INK4a. Silencing PKM2 increases the transcription of genes encoding CS and isocitrate dehydrogenase, an effect observed in early-passage cells but not in high-passage cells. Collectively, these findings suggest the potential of targeting PKM2 as a therapeutic strategy for age-associated OA [51]. In parallel, PTEN-induced kinase 1 (PINK1) phosphorylates PKM2 at Ser127, inhibiting its nuclear translocation and interaction with β-catenin, thereby promoting mitophagy and anabolic metabolism in cartilage [52]. Collectively, these mechanisms position PKM2 inhibition as a promising strategy to prevent chondrocyte apoptosis and inflammatory damage in OA. Lactate dehydrogenase (LDH) The glycolytic rate-limiting enzyme LDH isoform A (LDHA) catalyses the conversion of pyruvate to lactate under anaerobic or hypoxic conditions, oxidising NADH to NAD+ to sustain glycolysis and enable rapid energy provision. Bioinformatic analyses indicate that LDHA transcript levels are markedly increased in OA chondrocytes. Mechanistically, LDHA binds NADH through a Rossmann-fold motif, accelerating electron transfer from NADH and thereby triggering mitochondrial ROS production. Notably, inhibitors that solely suppress lactate generation fail to reduce ROS, whereas compounds that block NADH binding effectively abrogate ROS induction [53,54]. Elevated LDHA levels further reinforce glycolytic dependence, suppress mitochondrial OXPHOS, and drive lactate accumulation. Acting as a direct substrate for histone lactylation, lactate preferentially promotes histone H3 lysine 18 lactylation (H3K18la), which facilitates a fibroblast-like phenotypic shift in chondrocytes. This shift leads to ECM degradation and cellular dysfunction, thereby accelerating OA progression [55]. In an osteoblast–chondrocyte co-culture system, elevated lactate promotes HIF-1α upregulation and nuclear accumulation, driving upregulation of glycolytic genes and pyruvate dehydrogenase kinase (PDK1). PDK1 inhibits mitochondrial pyruvate dehydrogenase, reducing TCA flux. Lactate also activates extracellular signal-regulated kinase (ERK)1/2 and p38 mitogen-activated protein kinase (MAPK) signalling, further stabilising HIF-1α and sustaining a glycolytic, lactate-producing state [56]. Clinically, LDHA is highly expressed in the synovial tissues and fibroblasts of patients with temporomandibular joint OA [57]. In these patients, LDHA inhibition suppresses glycolytic activity, promotes hyaluronan synthesis, and attenuates inflammation, without marked changes in cell proliferation, cell-cycle progression, or apoptosis [58]. Looking ahead, strategies targeting LDH-mediated metabolism–immune crosstalk and the LDH–mitochondrial interface may support the development of dual-acting agents with both metabolic and anti-inflammatory effects, helping to alleviate the pathological cycle of glycolytic hyperactivation and mitochondrial dysfunction in OA. For instance, inhibitors that merely reduce lactate production do not diminish ROS, whereas compounds that block NADH binding can effectively abrogate ROS generation. Accordingly, therapies that coordinate glycolytic control with mitochondrial regulation may yield greater benefit than approaches focused on either pathway alone, underscoring the need to further delineate the glycolysis–mitochondrial metabolism–OA axis. Mechanistic links between glycolytic dysregulation and mitochondrial function in chondrocytes Mitochondria are dynamic organelles with a double-membrane structure in eukaryotic cells, serving as the energy-producing “powerhouse”. In chondrocytes, mitochondria generate energy through OXPHOS: under oxygen-rich conditions, pyruvate from cytoplasmic glycolysis enters the TCA cycle, producing NADH and flavin adenine dinucleotide to drive ATP synthesis and support anabolic and catabolic processes. The mitochondrial respiratory chain on the inner membrane facilitates ATP production through proton transfer and pumping. Cellular respiration directs pyruvate fate based on oxygen levels: under hypoxia, pyruvate converts to lactate to complete glycolysis; under normoxia, it enters mitochondria for the TCA cycle and OXPHOS [4]. The TCA cycle occurs in the mitochondrial matrix, where mitochondrial CS catalyses the condensation of acetyl-CoA with oxaloacetate to form citrate, initiating the cycle [59]. Glycolysis yields only 2 ATP per glucose molecule, whereas the TCA cycle and OXPHOS produce ∼36 ATP per glucose [60]. Under physiological hypoxia, chondrocytes reduce mitochondrial ATP production to limit ROS generation and thereby maintain cellular homeostasis [61]. Driven by their distinctive metabolic demands, chondrocytes rely predominantly on glycolysis for energy production, with mitochondrial pathways playing a secondary role. This energy-supply profile, which differs from that of most other cell types, may open new avenues for mechanistic investigation in OA. Beyond ATP generation, mitochondria serve as an important hub for metabolic control and signalling integration, largely through mitochondria-derived metabolic intermediates [62], forming a complex metabolic network with glycolysis. Glycolytic aberrations trigger metabolic cascades that disrupt normal mitochondrial function. Thus, maintaining normal mitochondrial function is essential for cellular energy provision, metabolic homeostasis, and regulatory integrity, with profound significance for chondrocyte health. Aberrant glycolytic by-products in OA In OA pathogenesis, heightened energetic demand in chondrocytes can drive glycolytic hyperactivation and disrupt cellular energy homeostasis. Once this balance is perturbed, multiple pathways and molecular mechanisms can influence the coupling between glycolysis and mitochondrial metabolism. Fig. 3 outlines a network in which aberrant accumulation of glycolytic by-products, including lactate, NADH, ROS, and AGEs, promotes mitochondrial injury through diverse mechanisms. By-products of aberrant glycolysis that damage mitochondrial function. Excessive lactate enters mitochondria via transporters, acidifying the matrix to disrupt mitochondrial dynamics, leading to mitochondrial fragmentation. Decreased NAD+/NADH ratios induce electron leakage in the mETC, increasing ROS production. AGEs accumulate through non-enzymatic glycation, reducing Δψm and releasing cytochrome c to trigger caspase-dependent apoptotic cascades. Glycolytic intermediates (G6P, F1,6BP) deplete NADPH via the polyol and hexosamine pathways, impairing mitochondrial antioxidant capacity (reduced glutathione [GSH]). These processes synergistically cause mitochondrial dysfunction, amplifying OA pathological damage. Created with BioRender.com. Lactate In OA, enhanced glycolytic activity in chondrocytes leads to substantial lactate production even under aerobic conditions [49,58]. The intracellular lactate shuttle hypothesis posits that LDH exhibits the highest maximum velocity (Vmax) of any glycolytic enzyme. Additionally, the equilibrium constant (Keq) for its conversion of pyruvate to lactate favours lactate accumulation, readily causing intracellular acidosis [63]. Isolated mitochondria can directly take up and oxidise lactate, indicating the presence of lactate transporters on the mitochondrial membrane [64]. In chick embryo cardiomyocytes, excessive lactate influx acidifies the mitochondrial matrix, impairing mitochondrial dynamics. This inhibits fusion proteins such as MFN1 and OPA1, reducing mitochondrial fusion, while activating fission proteins such as Drp1, promoting excessive fragmentation [[65], [66], [67]]. Fragmented mitochondria exhibit dysfunction, including elevated ROS, decreased Δψm, and increased apoptotic pathway activity. In chondrocytes, these defects impair OXPHOS and ATP production, thereby exacerbating energy metabolism dysfunction [68,69]. Furthermore, the elevation in ROS inhibits the transcription and protein expression of lactate transporter monocarboxylate transporter 4 (MCT4) by reducing its promoter levels of histone H3 trimethylation (H3K4me3), an activating modification. This leads to ineffective cellular extrusion of lactate, resulting in its intracellular accumulation. At high intracellular concentrations (≥25 mM), lactate inhibits chondrocyte viability and induces apoptosis [70]. Lactate also influences mitochondrial respiration by modulating Mg2+ uptake and binding to hydroxycarboxylic acid receptor 1 (HCAR1). This interaction activates the PI3K-Akt pathway, upregulating NADPH oxidase-4 (NOX4) expression and elevating ROS production, thereby inducing mitochondrial oxidative stress and impairing mitochondrial function [53]. As lactate accumulates and enters the mitochondrial matrix, it stimulates the activity of the mitochondrial electron transport chain (mETC); the resulting increase in mitochondrial ATP synthesis inhibits glycolysis while enhancing the utilisation of pyruvate and other alternative respiratory substrates. This reflects both a compensatory response to lactate accumulation and a recognition of lactate functions beyond metabolism [71]. Evidently, its effects on mitochondria are multifaceted. NADH and ROS Although glycolysis generates NADH under normal conditions, aberrantly elevated glycolytic activity in OA reduces the NAD+/NADH ratio in primary chondrocytes from Sprague–Dawley rats, inducing redox imbalance [72,73]. This shift disrupts Δψm, while constrained mitochondrial electron transport further impairs NAD+ regeneration, aggravating the decline in the NAD+/NADH ratio and reinforcing a vicious cycle [74]. NAD+- or NADH-dependent enzymes, including the pyruvate dehydrogenase complex (PDC) and α-ketoglutarate dehydrogenase complex (α-KGDHC) are impacted, potentially leading to electron leakage in the mETC, increased ROS production, and disruption of downstream processes such as the TCA cycle, ultimately contributing to energy metabolism chaos [75]. Additionally, under glycolytic anomalies, the pentose phosphate pathway generates NADPH by transferring hydrogen and electrons to NADP+, while NOX enzymes—particularly its major variant NOX4 in complex with p22phox—emerge as primary intracellular sources of ROS. Concurrently, lactate-HCAR1 binding activates PI3K-Akt signalling to upregulate NOX4, further amplifying ROS [53,76]. In mesenchymal stem cells (MSCs), direct supplementation with the NAD+ precursor nicotinamide mononucleotide (NMN) increases NAD+ levels and elevates the NAD+/NADH ratio. This shift enhances the expression and activity of sirtuin (Sirt)3, a key mitochondrial regulator, thereby improving mitochondrial performance and reversing senescence-associated phenotypes in MSCs [77]. These NMN-mediated improvements in mitochondrial function and redox balance, driven by regulation of the NAD+/Sirt3 pathway, highlight its therapeutic potential for correcting metabolic dysfunction and alleviating cellular damage in OA. Regarding ROS, glycolytic aberrations lead to the accumulation of intermediates such as glucose-6-phosphate (G6P) and F1,6BP, activating the hexosamine or polyol pathways, depleting NADPH, reducing glutathione, and diminishing mitochondrial antioxidant capacity. Mitochondria are the primary cellular source of ROS, with production closely tied to the mETC, where complexes I and III are key sites [78]. The strong oxidising nature of ROS, when dysregulated in chondrocytes, damages biomolecules, leading to DNA mutations, protein dysfunction, and membrane lipid peroxidation [79,80]. Declining mitochondrial antioxidant defences, coupled with elevated ROS generation, prompt further mitochondrial ROS (mtROS) production. Accumulated mtROS activates HIF-1α and increases its protein levels, inducing a Warburg-like phenotype in which metabolism shifts from aerobic oxidation towards glycolysis. This reprogramming increases lactate production, reduces TCA cycle metabolites, impairs mitophagy, and promotes the accumulation of damaged mitochondria, thereby establishing a vicious cycle [81]. Ultimately, persistently high ROS levels induce mitochondrial membrane depolarisation, compromise mitochondrial function, and perpetuate ROS overproduction, triggering oxidative stress and chondrocyte death. This stress initiates a pathogenic OA cycle in which NADH accumulation leads to increased electron leakage, excessive ROS production, mitochondrial damage, and further inhibition of NADH oxidation. AGEs AGEs accumulate in both spontaneous and diabetes-associated OA [82,83]. Under glycolytic aberrations and hyperglycaemia, glucose and its metabolites undergo non-enzymatic reactions with intracellular proteins and lipids to form AGEs [84]. Studies in rabbit chondrocytes show that AGEs reduce Δψm, decrease ATP production, promote cytochrome C release from mitochondria to the cytosol, activate caspase-3, and initiate apoptotic cascades, increasing chondrocyte apoptosis. AGEs also induce heightened ROS production and endoplasmic reticulum stress (ERS), culminating in apoptosis and MMP production [85]. Further evidence from mesangial cells suggests that activation of the AGE receptor (RAGE) promotes ROS generation primarily through NOX activation, and that the natural isoflavone puerarin attenuates high-glucose-induced oxidative stress through this pathway [86,87]. In addition, binding of AGEs to RAGE activates downstream nuclear factor-κB (NF-κB) and Janus kinase (JNK) signalling, thereby amplifying oxidative stress and promoting the release of pro-inflammatory cytokines, including IL-1β, IL-6 and TNF-α, as well as MMP9. Collectively, these events drive chondrocyte apoptosis and ECM degradation, ultimately accelerating OA progression [88]. AGEs also activate the NF-κB pathway by promoting IκB degradation and p65 nuclear translocation, leading to increased target gene transcription of pro-inflammatory factors and matrix-degrading enzymes. This cascade simultaneously inhibits autophagy and exacerbates apoptosis, ultimately contributing to the onset and progression of OA. Plantamajoside, a medicinal plant with antioxidant and anti-inflammatory properties, can reverse this AGE-mediated activation of the NF-κB pathway [89,90]. In chondrocytes, RAGE signalling may promote NLRP3 inflammasome activation through a two-step mechanism, whereby NF-κB activation provides the initial priming signal and ROS induction supplies the activation signal. Together, these cues facilitate inflammasome assembly and activation, further amplifying inflammation through a self-sustaining feedback loop [91,92]. In chondrocytes, AGEs suppress key signalling pathways, including AMPKα, SIRT1, and peroxisome proliferator-activated receptor gamma co-activator-1α (PGC-1α), leading to increased oxidative stress, inflammation, and apoptosis. These changes disrupt mitochondrial integrity, both as a trigger and a consequence of mitochondrial dysfunction [93,94]. AGEs promote OA progression by upregulating mitochondrial fission protein DRP1, inhibiting mitophagy and thereby increasing chondrocyte apoptosis and MMP expression [95]. Additionally, AGEs form collagen cross-links that disrupt ECM homeostasis and activate the NF-κB and MAPK pathways through a GLUT1–AGEs–ERS–inflammatory cytokine axis, driving IL-6 and TNF-α release and promoting chondrocyte hypertrophy [85,96]. These converging pathogenic roles position AGEs as potential therapeutic targets and diagnostic biomarkers in OA. Glycolytic stress and mETC dysfunction: drivers of mitochondrial impairment in OA The mETC, located on the inner mitochondrial membrane, comprises complexes I, II, III, and IV, along with coenzyme Q (CoQ) and cytochrome c. Complex I accepts electrons from NADH and transfers them to CoQ while pumping protons; complex II passes electrons from succinate to CoQ, which shuttles electrons between complexes; complex III transfers electrons from CoQ to cytochrome c and pumps protons; and cytochrome c then delivers electrons to complex IV, which reduces oxygen to water while pumping protons. Electron transfer energy within the mETC establishes a proton electrochemical gradient that drives ATP synthesis via ATP synthase during OXPHOS. Intimately linked to ATP production, the mETC sustains cellular activities [97]. In the context of glycolytic aberrations, it also influences mitochondrial function. Under abnormal glycolysis, excessive activation of ENO1, overaccumulation of lactate and AGEs, and a low NAD+/NADH ratio can all lead to changes in Δψm [44,70,93]. The maintenance of membrane potential depends on proton pumping by the mETC, and its stability is directly linked to mitochondrial material transport, morphological dynamics, and autophagy. The mETC serves as the primary source of mtROS. Under normal conditions, electrons are efficiently transferred through the mETC to oxygen (O2), generating water (H2O), with only a small amount of electron leakage producing superoxide anions (O2-). Under functional impairment (e.g., obstruction of electron transfer in complexes I or III), electron leakage increases, leading to substantial ROS accumulation [98]. Complex I oxidises NADH to generate NAD+; restricted function leads to NADH accumulation, a reduced NAD+/NADH ratio, and inhibition of the TCA cycle [99]. Thus, prolonged exposure of chondrocytes to a hyperglycolytic environment, with increased lactate levels, inhibits anabolic metabolism and reduces the synthesis of ECM components such as collagen and proteoglycans. This shift promotes catabolic metabolism, accelerates cartilage degeneration, and further amplifies glycolysis to compensate for ATP acquisition. In summary, the mETC governs mitochondrial function through signalling pathways involving energy synthesis, redox balance, membrane potential maintenance, and metabolic alterations. Its dysfunction can trigger energy depletion, ROS accumulation, metabolic disorders, and structural damage, ultimately contributing to OA pathogenesis. Targeting mETC complexes or modulating the influence of upstream glycolytic input may represent a key strategy for protecting mitochondrial integrity. Molecular signals driving OA pathogenesis In the pathological progression of OA, glycolytic aberrations and mitochondrial dysfunction interact through multidimensional signalling pathways to form a vicious cycle that jointly drives cartilage degeneration. Fig. 4 provides a concise overview of these interconnected pathways. Below, we systematically dissect how metabolic by-products impair mitochondrial function and examine the roles of the PI3K-AKT, HIF-1α, NF-κB, mTOR, and AMPK pathways in chondrocyte energy metabolism, elucidating their synergistic or antagonistic roles in OA-associated cartilage degeneration. Signalling pathways that regulate glycolysis–mitochondria crosstalk in OA. Hypoxia activate HIF-1α, which upregulates glycolytic genes and inhibits mitochondrial respiration via PDK1-mediated inactivation of the PDC. The PI3K-AKT pathway enhances glycolysis and HK2 while suppressing mitochondrial fusion (MFN1/2). NF-κB promotes GLUT3 and LDHA expression, exacerbating lactate accumulation and mitochondrial damage. mTOR drives glycolytic hyperactivity through c-Myc activation, whereas AMPK protects mitochondrial homeostasis via PGC-1α in early stage and disrupts mitochondrial homeostasis via JNK and Drp1 in late stage. These pathways collectively disrupt metabolic balance, leading to mitochondrial dysfunction in chondrocytes. Created with BioRender.com. At the onset of OA, inflammatory cytokines such as IL-1β and TNF-α activate the PI3K-AKT and NF-κB pathways [100], upregulating key glycolytic enzymes (e.g., HK2 and LDHA) and inhibiting mitochondrial fusion, leading to metabolic reprogramming, impaired OXPHOS, and hypoxic microenvironments [25]. HIF-1α enhances glycolytic gene expression and suppresses PDC activity, forcing cells to rely on glycolysis for energy while hindering mitochondrial biogenesis and repair [101]. The mTOR signalling pathway promotes glycolytic hyperactivity via cellular Myc proto-oncogene (c-Myc) and inhibits mitochondrial biogenesis, whereas AMPK exerts dual roles in energy stress. Through PGC-1α/SIRT1 regulation and mitophagy, transient activation maintains metabolic homeostasis, but sustained imbalance accelerates mitochondrial quality control failure [102,103]. These pathway interactions collectively promote ROS bursts, Δψm collapse, and apoptosis, ultimately culminating in chondrocyte death and ECM degradation. PI3K-AKT pathway In OA, glycolytic abnormalities trigger mitochondrial dysfunction through the PI3K-AKT pathway, forming a positive feedback loop between metabolism, signalling, and mitochondrial impairment. In multiple tumour cell types, AKT directly phosphorylates rate-limiting glycolytic enzymes (e.g., HK1 and HK2), enhancing their activity and promoting glucose uptake, thereby shifting carbohydrate metabolism towards glycolysis [104,105]; whether this mechanism operates in chondrocytes remains to be verified. Next, the inflammatory factor IL-1β stimulates further activation of PI3K-AKT, increasing mtROS, decreasing membrane potential, and upregulating the expression of mitochondrial fission-related proteins MFF and mitochondrial targeting protein 18 (MTP-18). This decline in Δψm triggers the intrinsic apoptotic pathway, exacerbating OA [106,107]. Studies in human umbilical vein endothelial cells have also identified the PI3K-AKT pathway as an inducer of mitochondrial fission through Drp1 [108], and an inhibitor of the Unc-51-like autophagy activating kinase 1 (ULK1) complex, blocking the autophagic tagging and clearance of damaged mitochondria [109]. SIRT3 acts as an endogenous inhibitor of the PI3K-AKT pathway. SIRT3 overexpression significantly reduces IL-1β-induced PI3K-AKT and mTOR phosphorylation, whereas its knockdown enhances pathway activation [100]. The natural flavonoid orientin reportedly delays chondrocyte senescence and OA progression by inhibiting PI3K-AKT, with mechanisms linked to improvements in mitochondrial homeostasis, namely structural preservation, membrane potential restoration, and ROS reduction [110]. Additionally, in cancer cell mechanistic studies, AKT-mediated HK2 phosphorylation was shown to enhance its mitochondrial binding and glycolytic initiation while inducing mitochondrial depolarisation and cell death via Ca2+ overload [22,25,111]. Chronic AKT activation biases cells towards glycolytic energy dependence, reducing reliance on mitochondrial OXPHOS and diminishing mitochondrial respiratory function through accumulation of aberrant metabolites. Additionally, AKT inhibits glycogen synthase kinase 3β (GSK3β), downregulating mitochondrial fusion proteins MFN1 and MFN2, thereby inducing mitochondrial fragmentation and disrupting mitochondrial network architecture and functionality [[112], [113], [114]]. Thus, targeted inhibition of the PI3K-AKT pathway may attenuate OA progression by safeguarding mitochondrial function. HIF-1α pathway In the hypoxic OA environment, HIF-1α serves as a key regulator of glycolysis, promoting transcription of glycolytic enzymes (e.g., GLUT1, HKs, and LDHA) [115], increasing glucose uptake, and accelerating glycolysis to meet energy demands [116]. Regarding mitochondria, HIF-1α activation suppresses mitochondrial aerobic respiration, reducing oxygen consumption and elevating the NADH/NAD+ ratio, which inhibits NADH-sensitive glycolytic enzymes such as GAPDH, particularly under limited pyruvate supply [49,117]. HIF-1α induces expression of PDK1, which phosphorylates and inactivates the PDC, blocking pyruvate entry into mitochondria for TCA cycling and diminishing mitochondrial energy production [101,118]. HIF-1α also suppresses genes involved in mitochondrial biogenesis, impeding mitochondrial renewal and repair, and leading to functional decline [[119], [120], [121]]. Under hypoxia, glycolysis is tightly coupled to the HIF-1α pathway. HIF-1α-driven glycolytic abnormalities not only suppress mitochondrial aerobic respiration and energy output, but also hinder mitochondrial renewal and repair, thereby accelerating mitochondrial dysfunction. Additionally, in gastric mucosal epithelial cells, HIF-1α can increase mtROS production by upregulating Drp1 via an METTL3/IGF2BP3-dependent N6-methyladenosine (m6A) modification mechanism [122]. Therefore, targeting the interaction between HIF-1α and glycolysis, or intervening in downstream mitochondrial damage pathways, holds promise for breaking the vicious cycle involving glycolytic abnormalities, HIF-1α activation, and mitochondrial functional decline, providing new strategies for metabolic-targeted therapy in OA. Notably, the chondroprotective effects of HIF-1α need to be balanced against the pro-degenerative actions of HIF-2α; their antagonism at the transcriptional regulatory level ultimately determines the trajectory of metabolic homeostasis in chondrocytes [123]. NF-κB pathway In the inflammatory milieu of OA, AGEs and cytokines such as IL-1β and TNF-α activate the NF-κB pathway, upregulating glycolytic genes (e.g., GLUT3 and LDHA) to enhance glucose uptake and glycolysis [54,124]. NF-κB shifts cellular metabolism towards glycolysis, with by-products such as ROS further damaging mitochondria. Upon entering the nucleus, NF-κB preferentially upregulates the transcription of genes encoding pro-inflammatory factors such as TNF-α, IL-1β, and IL-6, which directly damage mitochondria through receptor-mediated signalling pathways [125,126]. This pathway also increases iNOS expression, leading to excessive production of nitric oxide, which inhibits mitochondrial respiratory chain complex IV, thereby impairing respiratory function, reducing ATP synthesis, and promoting ROS accumulation. NF-κB reduces the expression and activity of PGC-1α, hindering mtDNA replication [127], while concurrently regulating core autophagy genes (e.g., ATG5, ATG7, and BECN1) that are essential for mitophagy. Dysregulation of mitophagy prevents timely clearance of damaged mitochondria, impairing mitochondrial quality and function [128]. Therefore, NF-κB mediates mitochondrial damage both indirectly—by amplifying glycolytic aberrations—and directly—by regulating pathways of harm. mTOR pathway Functioning as a cellular nutrient and energy sensor, mTOR exists in two distinct complexes: mTORC1 and mTORC2. It is activated by intracellular nutrient levels, energy status, PI3K-AKT signalling, and inflammatory cytokines [129]. In OA, mTOR senses increased energy demands from inflammatory cells and is activated by inflammatory stimuli; pyruvate levels may also influence its activity [18]. In glioblastoma cells, mTORC2 activation promotes glycolysis by increasing the levels of transcription factor c-Myc, which drives expression of glycolytic genes and enhances lactate production [102]. Through mTORC1 activation, enhanced glycolysis significantly inhibits mitophagy, impairing the clearance of damaged mitochondria and exacerbating mitochondrial dysfunction [18,130]. Activated mTORC1, recruited to lysosomes via the Ragulator-Rag complex, regulates mtROS production, promoting the oligomerisation of gasdermin D N-terminal domain to form 20-nm pores that induce pyroptosis and inflammatory cytokine release [131]. A dynamic regulatory relationship exists between mTORC1 activity and the Ca2+ release channels mucolipin (MCOLN)1 and MCOLN3; through distinct feedback mechanisms and subcellular localisations, they jointly modulate the autophagy process [132]. Although Ca2+ signalling is closely related to mitochondrial function, this mechanism has not yet been verified in OA chondrocytes. Collectively, mTOR senses energy and inflammatory signals in the pathological OA microenvironment, driving disease progression through metabolic reprogramming and inflammatory cell death pathways, and represents a novel therapeutic target in OA. AMPK activation and mitochondrial biogenesis In OA, increased cellular energy demand enhances glycolysis; however, an imbalance between energy supply and demand leads to a relative decrease in ATP and a relative increase in AMP, elevating the AMP/ATP ratio. Once this ratio rises above a threshold, AMPK is activated to restore energy homeostasis [133]. Binding of AMP to the AMPK γ-subunit promotes its activation via phosphorylation at Thr172, a process mediated by upstream kinases such as LKB1 and CaMKK2 [134]. Meanwhile, inflammatory cytokines, such as IL-1β and TNF-α, further disrupt cellular energy balance and indirectly facilitate AMPK activation as a stress response. At this early compensatory stage, AMPK is largely protective and helps maintain cartilage homeostasis through multiple pathways. Upon activation, AMPK can inhibit NF-κB and its downstream p65 signalling, reducing inflammation-mediated cartilage matrix degradation [135]. Activators such as metformin also protect chondrocytes via the AMPK/NF-κB axis. Additionally, AMPK antagonises mTOR signalling by phosphorylating tuberous sclerosis complex 2 (TSC2) and RAPTOR to suppress mTORC1, thereby promoting chondrocyte autophagy and maintaining matrix metabolic balance [136,137]. AMPK can also inhibit the PI3K-AKT pathway, alleviating synoviocyte proliferation and MMP13 expression, suppressing chondrocyte hypertrophy and cartilage degradation, and synergising with the FoxO3a pathway to mediate autophagy and mitigate OA-associated damage [138]. Moreover, AMPK may shift chondrocyte metabolism from OXPHOS towards glycolysis to better match energy supply under inflammatory conditions, and it may help mitigate mitochondrial dysfunction, potentially by modulating PGC-1α to promote mitochondrial biogenesis and mitophagy [139]. Consistently, resveratrol alleviates inflammatory injury by upregulating the SIRT1/AMPK pathway, while active vitamin D activates autophagy via the AMPK–mTOR axis; both may help delay OA progression [140,141]. In late-stage OA, as AMPK activation and the coordination among cooperating pathways become disrupted, its detrimental effects gradually emerge. For example, adiponectin reportedly enhances chondrocyte catabolic activity via the AMPK/JNK pathway, accelerating degradation of proteoglycans and type II collagen [142]. Activated AMPK signalling can also induce Drp1-mediated mitochondrial fission [143]. Furthermore, as AMPK cooperates with SIRT1 to promote a more glycolytic metabolic program, this transition may be accompanied by increased production of inflammatory mediators and catabolic factors, reinforcing the inflammatory milieu. In advanced OA, where chondrocyte injury is extensive, such context-dependent AMPK signalling may tilt the metabolic balance towards catabolic outputs, accelerating cartilage matrix breakdown and contributing to OA progression. Oxidative stress and inflammatory responses In OA chondrocytes, excessive glycolytic enzyme activity, altered Δψm, and the HIF-1α signalling-driven decrease in NAD+/NADH ratio collectively inhibit the mitochondrial TCA cycle and promote ROS production. The resulting oxidative stress damages mETC complexes, depletes cellular antioxidants and antioxidant enzymes, induces mtDNA damage, and ultimately leads to mitochondrial destruction. Hypoxia acts as an allosteric regulatory factor that reprogrammes cellular metabolism in chondrocytes, favouring increased glycolysis over OXPHOS to maintain ATP levels [144]. HIF-1 is a pivotal transcription factor and primary regulator of glycolysis, upregulating GLUTs and glycolytic enzymes [145]. In chondrocytes, HIF-1α signalling modulates collagen synthesis and modification through metabolic changes [146]. OA chondrocytes exhibit increased HIF-1α expression, accompanied by apoptosis, senescence, ECM dyshomeostasis, elevated ROS, and autophagy [22]. Additionally, the inflammatory response induced by the NF-κB signalling pathway during aberrant glycolysis affects mitochondrial function in multiple ways. Abnormalities in mitochondrial dynamics Glycolytic aberrations impact mitochondrial dynamics primarily through metabolite dysregulation and signalling pathway modulation. Lactate produced by excessive glycolysis acidifies the mitochondrial matrix, inhibiting fusion proteins MFN1 and OPA1 while activating fission protein Drp1, promoting excessive fragmentation. Excess ROS exacerbates fission via Drp1 phosphorylation, and AGEs disrupt dynamics-related proteins by inhibiting the AMPK/SIRT1/PGC-1α axis. Multiple signalling pathways contribute to this imbalance: PI3K-AKT activation enhances fission; HIF-1α suppresses biogenesis, hindering fusion; NF-κB dysregulates mitophagy, disrupting dynamic equilibrium; AMPK maintains fusion–fission homeostasis via PGC-1α and SIRT1 activation; and mTOR signalling favours fission. Additionally, intracellular Ca2+ dyshomeostasis interferes with normal mitochondrial function, further affecting fusion and fission processes. Human omics evidence for the glycolysis–mitochondrial metabolism–OA axis RNA sequencing analysis of articular cartilage samples revealed that, compared with healthy controls, OA patients exhibit significant upregulation of glycolysis-related genes, including HK2 [23], LDHA [147], PFKFB3, and HK1 [53]. Interestingly, GLUT expression is not markedly different between normal and OA chondrocytes, and is not regulated by insulin [148]. Conversely, mitochondria-associated glucose metabolism genes, including pyruvate dehydrogenase kinase 4 (PDK4), methylenetetrahydrofolate dehydrogenase 2 (MTHFD2), electron transfer flavoprotein dehydrogenase (ETFDH), mitochondrial cysteinyl-tRNA synthetase 2 (CARS2) [149], and MFN2 [150], are significantly downregulated in OA cartilage. Mitochondrial dysfunction is further evidenced by significant differences in mtDNA [151], reduced mtDNA repair capacity, and increased apoptosis in OA chondrocytes relative to those from healthy individuals [152]. Proteomic analyses of synovial fluid and cartilage from OA patients corroborate these findings, showing elevated levels of glycolytic enzymes such as HK2 [23], ENO1 [153], and PKM2 [49]. By contrast, PFKFB3 expression is downregulated in human OA cartilage samples, despite its roles in glycolytic metabolism regulation, ERS alleviation, caspase-3 inhibition, and promotion of aggrecan and type II collagen expression in these tissues [154]. Mitochondrial function assays in human OA chondrocytes reveal reduced activities of mitochondrial respiratory complexes II, III, and V [155,156], and lower levels of PDK4 and MTHFD2 compared with normal controls [149]. Additionally, multiple regulators involved in mitochondrial glucose metabolism, energy production, or antioxidant defence (e.g., SIRT3, isocitrate dehydrogenase 2 [IDH2], superoxide dismutase 2 [SOD2], AMPK, and PGC-1α) are frequently downregulated in OA, contributing to disordered mitochondrial metabolism, insufficient energy supply, and heightened oxidative stress. Together, these alterations represent a key mechanism underlying OA cartilage damage [6]. Metabolomic analyses support these findings. OA patients show abnormal plasma accumulation of glycolytic by-products, including lactate [156] and AGEs, alongside decreased levels of the TCA intermediate α-ketoglutarate [157]. Compared with healthy controls, lactate concentrations in synovial fluid are significantly higher in OA patients with severe cartilage loss [53], and HK2 expression in peripheral blood mononuclear cells is increased in both rheumatoid arthritis and OA cohorts [158]. Serum lactate levels are elevated in OA patients, and serum succinate levels are positively correlated with OA pain [53,159]. In late-stage OA, whole-blood expression profiling shows significant upregulation of PKM2, LDH, and AMPKα [160]. Collectively, these multi-omics data point to aberrant activation of the glycolysis–mitochondrial metabolism axis in human OA. Dysregulated enzymes, metabolites, and mitochondrial proteins in cartilage, synovial fluid, and serum highlight this axis as a potential pathological hallmark of human OA. Dysfunctional glycolysis and mitochondrial metabolism as a driver of OA progression Dysregulation of the glycolysis–mitochondrial metabolism–OA axis is not an isolated metabolic event, but a multidimensional driver of OA pathology, contributing to energy metabolic reprogramming, oxidative stress amplification, and inflammatory signalling crosstalk. The resulting joint damage manifests as a cascade involving chondrocytes, the ECM, and the surrounding microenvironment, with the underlying mechanisms outlined below. Aberrant chondrocyte apoptosis Histological alterations in chondrocyte density and tissue architecture have been associated with OA progression, and the number of apoptotic chondrocytes in OA cartilage positively correlates with the severity of degeneration [161,162]. This apoptosis is not a random event; rather, it is closely linked to persistent perturbations of the glycolysis–mitochondrial metabolism–OA axis, which drives energy imbalance, intensifies oxidative injury, and activates pro-apoptotic signalling. Imbalance within this metabolic–inflammatory signalling network can induce chondrocyte death through a cascade involving impaired energy sensing, oxidative damage, inflammatory responses, and activation of apoptotic pathways. Primarily, metabolic supply chain disruption driven by enhanced glycolysis and mitochondrial hypoactivity destabilises chondrocyte energy homeostasis, triggering energy-dependent apoptosis [163]. As described above, in OA the dissociation of the key glycolytic enzyme HK2 from mitochondrial VDAC promotes mitochondrial Ca2+ overload, activates calpains to cleave pro-apoptotic proteins, and facilitates mitochondrial fragmentation [25]. Excessive glycolysis increases lactate levels and lowers the NAD+/NADH ratio, which can impair Δψm and promote the release of cytochrome c into the cytosol, where it activates caspase-3, inducing chondrocyte apoptosis [53,85]. In parallel, binding of AGEs to RAGE on chondrocytes increases mtROS production and suppresses the AMPK–SIRT1–PGC-1α pathway, compromising mitochondrial quality control [93,94]. Moreover, glycolysis- and hypoxia-driven HIF-1α activation upregulates PDK1, restricting pyruvate entry into mitochondria and further aggravating metabolic imbalance [118]. Clinically, OA cartilage with severe chondrocyte loss is characterised by greater HK2 dissociation, reduced Δψm, and increased caspase activity. Together, enhanced glycolysis and mitochondrial metabolic derangements may accelerate abnormal chondrocyte apoptosis, potentially providing a mechanistic link between chondrocyte loss, cartilage thinning, and OA progression. Imbalanced ECM synthesis and degradation Articular chondrocytes maintain a dense ECM network through tightly controlled synthesis and degradation of key components, including low-molecular-weight ions and macromolecules (e.g., proteoglycans, type II collagen, and non-collagenous proteins), which are vital for ECM spatiotemporal organisation [164]. OA is characterised by classic features of ECM dysfunction, including reduced synthesis of type II collagen and aggrecan, increased proteoglycan degradation, progressive joint space narrowing on radiographs, and fibrosis of the cartilage surface [165,166]. These phenotypic changes reflect an imbalance between anabolic and catabolic processes in chondrocytes, driven in part by dysregulation of the glycolysis–mitochondrial metabolism–OA axis. Weakened TCA cycling reduces α-ketoglutarate levels, inhibiting histone demethylase activity and downregulating ECM synthesis genes such as COL2A1 [167]. Concurrent mitochondrial dysfunction and glycolytic aberrations impair energy-intensive ECM production [163], while impaired OXPHOS leads to ATP insufficiency, directly suppressing post-transcriptional modifications of type II collagen and aggrecan [168]. Glycolytic anomalies elevate the activity of matrix-degrading MMPs [169] and induce inflammatory mediators such as IL-1β and TNF-α [124], further amplifying MMP expression—particularly MMP13 to mediate efficient type II collagen degradation and MMP3 to cleave various ECM proteins (e.g., proteoglycans) [170,171]. Glycolytic dysregulation and mitochondrial dysfunction disrupt this balance in two ways: ATP scarcity impedes the synthesis of type II collagen and proteoglycans, weakening ECM structural support, while metabolic chaos activates inflammatory pathways (e.g., IL-1β/TNF-α), upregulating degradative enzymes such as MMP13/3 to accelerate ECM breakdown. Together, these contribute to progressive joint space narrowing. GLUT1 overactivation or overexpression can promote glucose influx and accumulation of AGEs, which crosslink collagen fibrils and activate the RAGE–PI3K–AKT pathway, thereby driving chondrocytes towards a fibroblast-like phenotype [172]. Transient receptor potential vanilloid 4 (TRPV4), through Ca2+–CaMKII signalling, induces Drp1 mitochondrial translocation and HK2 dissociation from the mitochondrial membrane, leading to mitochondrial dysfunction and ultimately ECM degradation [173]. Characteristic AGE adducts preferentially deposit within cartilage collagen, resulting in articular cartilage calcification and altered tissue stiffness, further promoting OA progression [[174], [175], [176]]. This convergence of reduced synthesis and enhanced degradation culminates in progressive cartilage matrix destruction. Notably, these axis-mediated perturbations are closely associated with clinical OA severity: higher synovial lactate levels and lower α-ketoglutarate concentrations correlate with more pronounced joint space narrowing, while PFKFB3/H4K12la expression correlates with MMP13 levels in osteoarthritic cartilage [33]. Collectively, these findings highlight the potential role of the glycolysis–mitochondrial metabolism–OA axis as a regulator of ECM homeostasis in OA. Current pharmacological and biological approaches targeting glycolysis and mitochondrial metabolism in OA We systematically reviewed interventions targeting key nodes within the proposed glycolysis–mitochondrial metabolism–OA axis, including molecular targets, candidate agents and tools, and stages of evaluation. Table 1 summarises representative strategies aimed at core glycolytic enzymes [20,23,33,35,42,53,90,[177], [178], [179], [180], [181], [182], [183], [184]], while Table 2 compiles interventions directed at mitochondrial metabolism [72,77,110,167,[185], [186], [187], [188], [189], [190]]. Although these studies support the feasibility of modulating pivotal enzymes and pathways, most approaches remain largely unidirectional, addressing glycolysis or mitochondrial metabolism in isolation rather than their reciprocal coupling. Integrating current evidence could enable the development of therapies that coordinate interventions across the axis, upgrading single-pathway modulation into coupled, axis-level strategies. Table 1Studies of glycolysis-targeting agents and interventions.Table 1TargetDrug/toolStudy stageReferenceGlycolysis2-deoxy-D-glucoseBasic research[177]GLUT5-hydroxymethylfurfuralBasic research[20]GLUTBioceramic ScaffoldsBasic research[178]GLUTAnanas comosusBasic research[179]GLUTglucocorticoidBasic research[180]HKLonidamine and Pulsed electromagnetic fieldBasic research[23]HKOxamateBasic research[181]ENO1oxazole analogue (KB2764)Basic research[42]PFKFB3SongorineBasic research[33]PFKFB3HarpagideBasic research[35]LDHGLX351322Basic research[53]LDHOxamateBasic research[181]AGEs–NF-κB pathwayPlantamajosideBasic research[90,182]Glucose metabolismVitamin D3Phase III clinical trial[183]ENO1circNFKB1Basic research[45]LDHBLDHB knockdownBasic research[184] Studies of glycolysis-targeting agents and interventions. Table 2Studies of agents and interventions targeting mitochondrial metabolism.Table 2TargetDrug/toolStudy stageReferencePDK1JX06Basic research[185]NAD+Nicotinamide MononucleotideBasic research[77]citrate synthaseCrocinBasic research[186]α-ketoglutarate dehydrogenaseDM-α-KGBasic research[167]Mitochondrial Respiratory Chain ComplexHydrogel (HTM-N@Gel)Basic research[72]Mitochondrial Respiratory Chain ComplexHonokiol (HKL)Basic research[187]Mitochondrial Respiratory Chain ComplexSIRT3 knockout (Sirt3−/−)Basic research[187]Mitochondrial Respiratory Chain ComplexNicotinamide riboside (NR)Basic research[188]PI3K-AKTOrientinBasic research[110]PDK2PDK2 knockout (PDK2 KO)Basic research[189]α-ketoglutaratePFKFB3 decreasesBasic research[190] Studies of agents and interventions targeting mitochondrial metabolism. A representative mechanistic example illustrates how coordinated modulation of glycolysis and mitochondrial metabolism could be achieved. Inhibitors that merely suppress lactate production may be insufficient to effectively reduce ROS, whereas combining such approaches with NADH-targeting inhibition could be more effective in restraining ROS generation [54]. In macrophages, the glycolytic activator PFKFB3 directly binds GLUT1 within the mitochondrial matrix, inhibiting its activity, reducing α-ketoglutarate production, destabilising mitochondrial metabolic homeostasis, and enhancing M1 macrophage polarisation, thereby amplifying joint inflammation [190]. This provides molecular-level insight into glycolysis–mitochondria crosstalk in inflammatory settings and identifies a potentially actionable point of convergence: silencing PFKFB3 could simultaneously attenuate glycolytic overactivation, restore mitochondrial α-ketoglutarate generation, and promote M2 polarisation. Such coordinated intervention may support metabolic balance, immune homeostasis, and joint protection, offering experimental validation for the glycolysis–mitochondrial metabolism–OA axis concept. We propose that additional cross-regulatory mechanisms are likely embedded within this complex metabolic axis. Identifying such mechanisms may reveal efficient leverage points for precise correction of axis dysfunction, enabling disproportionate therapeutic gains. Future work should therefore more explicitly interrogate the bidirectional regulatory network linking glycolysis and mitochondrial metabolism, expanding the mechanistic repertoire associated with this axis and establishing a robust platform for the rational design of synergistic, whole-axis interventions. Future challenges and therapeutic prospects of targeting the glycolysis–mitochondrial metabolism–OA axis Given the altered joint microenvironment in OA and the specialised adaptation of chondrocytes to the avascular, alymphatic, and low-oxygen nature of cartilage, drug delivery and pharmacokinetic barriers represent a central bottleneck for translating therapies targeting the proposed axis. Systemically delivered agents may exert whole-body metabolic effects and mitigate indirect axis-related injury from systemic metabolic disturbances, but dense cartilage limits penetration, potentially preventing therapeutically effective concentrations. Furthermore, systemic therapy carries a risk of off-target adverse effects, including dysglycaemia and abnormal muscle metabolism. Therefore, simply listing metabolic inhibitors is insufficient to meet the translational needs of this axis. To address this challenge, we propose a synergistic strategy combining systemic glucose–metabolism modulation with local targeted delivery. Systemically, agents such as glucagon-like peptide 1 (GLP-1) receptor agonists and sodium glucose transporter 2 (SGLT2) inhibitors could improve the overall metabolic milieu, reduce glucose burden in synovial fluid, and limit systemic accumulation of toxic metabolic by-products (e.g., AGEs and lactate), thereby creating a foundation for restoring axis function. Locally, suitable delivery platforms could be employed. For example, an HK2-mimetic peptide based on the HK2–VDAC binding motif could be designed for intra-articular injection to penetrate the cartilage matrix. Encapsulation of this peptide in a thermosensitive hydrogel could achieve sustained release with a biodegradability profile that minimises foreign-body reactions. Additionally, dual-target nanocarriers decorated with cartilage-targeting ligands could co-deliver the peptide along with mitochondrial protective agents. Intra-articular injection could enable preferential accumulation of these carriers within chondrocytes. The use of low-toxicity or non-toxic materials may minimise systemic distribution and overt side effects. Together, these components may offer favourable biocompatibility and feasibility within the joint microenvironment. Theoretically, this synergistic approach not only compensates for the limitations of systemic administration (insufficient intra-cartilaginous exposure and prominent adverse effects), but also improves local joint retention time and targeting efficiency, thereby enhancing precision, clinical feasibility, and translational potential of axis modulation. In addition, abnormal metabolites arising from dysregulation of the glycolysis–mitochondrial metabolism–OA axis may serve as diagnostic biomarkers for the proposed metabolic–inflammatory OA subtype. For instance, ENO1 concentration in the synovial fluid of OA patients is approximately 2.5 times higher than that in healthy individuals, and correlates positively with clinical indicators of disease severity, including KOOS functional and VAS pain scores. Thus, ENO1 may represent a potential biomarker for diagnosing knee OA. Combining this measure with FGB downregulation in synovial fluid could further enhance diagnostic specificity [46]. Similarly, pairing lactate with serum IL-6 may be informative, because lactate levels reflect metabolic disturbance, whereas IL-6 indicates inflammatory activity. PFKFB3, a key glycolytic activator, and H4K12la, a downstream histone modification implicated in inflammatory regulation, also represent promising biomarker candidates. Conclusion Aberrant glycolysis, driven by dysregulation of key enzymes and amplified by inflammatory and hypoxic signals, disrupts mitochondrial function through multiple mechanisms. Excessive lactate accumulation acidifies the mitochondrial matrix and impairs fusion–fission dynamics; AGEs induce collapse of Δψm and trigger apoptotic cascades; and disruption of NADH/NAD+ redox homeostasis enhances electron leakage from the mETC, exacerbating oxidative stress. Conversely, mitochondrial dysfunction, characterised by impaired OXPHOS, defective quality control, and disturbed dynamics, reinforces glycolysis dependence through multiple pathways (e.g., HIF-1α, PI3K-AKT, and NF-κB), forming a pathological circuit that promotes ECM degradation and chondrocyte apoptosis. Human multi-omics evidence and pharmacological studies support the clinical potential of this axis. ENO1, lactate, and PFKFB3/H4K12la have emerged as promising diagnostic biomarkers, while targeted interventions (e.g., 5-HMF, Harpagide, and NAD+ precursors) have demonstrated therapeutic effects. The proposed synergistic strategy integrating systemic metabolic regulation with local targeted delivery may provide additional insights for translating mechanistic findings into clinical applications. Overall, identification of the glycolysis–mitochondrial metabolism–OA axis elevates OA research from single-target exploration to network-level regulation, providing a theoretical basis for shifting OA treatment from symptomatic relief towards repair of underlying drivers. Future work may focus on developing dual-target agents that simultaneously modulate metabolism and inflammation, optimising delivery systems to enhance cartilage penetration, and ultimately bridging the gap between basic research and clinical practice to achieve breakthroughs in OA therapy. CRediT authorship contribution statement Ren Zhang: Conceptualisation, Writing – original draft, Writing – review and editing; Su-Zhen Chen: Writing – review and editing; Shang-Chun Guo: Funding acquisition, Resources, Supervision; Shi-Cong Tao: Funding acquisition, Resources, Supervision, Conceptualisation, Writing – review & editing. All of the authors have read and approved the final manuscript. Declaration of generative AI and AI-assisted technologies in the writing process No generative artificial intelligence (AI) or AI-assisted technologies were used in the preparation of this manuscript. Funding The present study was supported by the following: National Natural Science Foundation of China (grant nos. 82072530, 82372566, and 82472595); Shanghai Pujiang Program (grant no. 2019PJD038); Shanghai “Rising Stars of Medical Talent” Youth Development Program (Youth Medical Talents–Specialist Program); Shanghai Jiao Tong University School of Medicine “Two-Hundred Talent” Program (grant no. 2022-017); and Shanghai Sixth People's Hospital Excellent Young Scientist Development Program (grant no. ynyq202101). Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements We acknowledge current and previous members of our laboratories. We thank Lina Cao for designing the schematic diagram, and acknowledge that some of the illustrations were created using BioRender.com. We also thank Michelle Kahmeyer-Gabbe (Edanz, www.liwenbianji.cn) for assistance during editing the English text of this manuscript. Contributor Information Shang-Chun Guo, Email: scguo@shsmu.edu.cn. Shi-Cong Tao, Email: sctao@shsmu.edu.cn. References