Increasing evidence suggests that alterations in chondrocyte metabolism toward glycolysis, associated with mitochondrial dysfunction, are critically linked to OA pathogenesis (8, 9). In this context, PDK-dependent inhibition of PDH activity may be a pivotal mechanism responsible for the glycolytic metabolic shift in catabolic chondrocytes (26). Here, we identified that PDK2 is specifically upregulated in OA chondrocytes and its loss-of-function led to an increase of PDH activity to restore the IL-1β-mediated metabolic shift toward glycolysis in chondrocytes. In addition, PDK2 deficiency showed a protective phenotype in surgically induced OA model, which was accompanied by reduced oxidative stress and cellular senescence. Mechanistically, PDK2 deficiency led to decreased activation of p38 MAPK, along with a sustained activation of AMPK signaling under IL-1β-treated conditions (Fig. 7). Taken together, our data sheds light on the potential of metabolic reprogramming towards OxPhos as a novel therapeutic approach for OA.
Several lines of evidence indicate a critical involvement of mitochondrial dysfunction in the pathogenesis of OA (9, 27). Specifically, chondrocytes from patients with advanced OA exhibit a decrease in respiring mitochondria, as evidenced by decreased rhodamine123 staining (9). Morphologically, these mitochondria are characterized by increased length relative to width, coupled with an overall reduction in count and disrupted morphology (9). This elongation is particularly noteworthy, as it implies heightened mitochondrial fusion, a phenomenon often observed under conditions such as nutrient withdrawal or increased OxPhos (28, 29). Actually, chondrocytes from relatively preserved articular cartilage demonstrate significantly higher mitochondrial respiration capacity compared to those from severely damaged lesions (9). As OA progresses, however, these metabolic adaptations begin to fail. This is evidenced by a diminished capacity of the respiratory chain and by a decrease in the number of mitochondria coupled with an increase in mitochondrial fission, leading to mitochondrial dysfunction (9, 27). Consequently, impaired mitochondrial function can disrupt ATP production and increase oxidative stress in chondrocytes, both of which are key contributors to the pathogenesis of OA (30).
To date, the molecular mechanism behind the metabolic shift towards glycolysis in OA chondrocytes has remained largely unclear. In this study, we demonstrated a significant increase in PDK2 among PDK isoforms under IL-1β-mediated catabolic condition and in OA chondrocytes (Fig. 1). Moreover, PDK2 deficiency led to a decrease in the phosphorylation of PDH under IL-1β-treated catabolic conditions, indicating the inactivation of PDH complex that converts pyruvate to acetyl-CoA (Fig. 4b). These findings suggest that PDK2 may be a key regulator of chondrocyte metabolism under catabolic conditions. Indeed, our data confirmed that PDK2 deficiency, at least partially, enhanced OxPhos in IL-1β-treated chondrocytes, while reducing glycolysis (Fig. 3). PDK, serving as a negative feedback mechanism, is activated by the products of the PDH reaction and TCA cycle, such as NADH, high energy charge, and acetyl-CoA (17). This activation leads to an inactivation of PDH. On the other hand, a decreasing energy charge and increasing pyruvate concentrations inhibit PDK activity, thereby leading to increased PDH activation (31). Although several mechanisms, including lactate dehydrogenase-A (LDH-A), hypoxia-inducible factor 1A (HIF1A), the AKT-mTOR signaling pathway, and pyruvate kinase M2, are known to control the glycolytic shift in chondrocytes (8, 32–34), PDK is known to mediate the Warburg effect—characterized by enhanced aerobic glycolysis (35). It may directly lead to the glycolytic shift observed in OA. Given this, inhibiting PDK2 could be a promising approach for the metabolic reprogramming of chondrocytes.
The expression of PDK isoforms in chondrocytes has not been extensively characterized. Our data revealed an increase in PDK2, whereas other PDK isoforms, such as PDK1, PDK3, and PDK4, showed a decrease in IL-1β-treated catabolic chondrocytes and in vivo OA cartilage (Fig. 1). Consistent with our findings, a recent study reported a significant downregulation of PDK1 mRNA and protein expressions in OA articular cartilage, although it did not specify the expression of other PDK isoforms (36). The lack of any noticeable phenotype in endochondral bone formation in Pdk2-deficient mice also suggests that PDK2 plays a limited role in the physiological maturation of chondrocytes (data not shown). This evidence of PDK2 being specifically expressed in catabolic chondrocytes suggests that targeting PDK2 could minimally affect normal cartilage physiology, while effectively addressing OA conditions, thereby offering a potential advantage in the development of OA drugs targeting PDK2.
Our data revealed that IL-1β-mediated p38 MAPK phosphorylation was significantly reduced in Pdk2-deficient chondrocytes (Fig. 5). Apoptosis signal-regulating kinase 1 (ASK1), which is positioned upstream of p38 MAPK, is a well-known redox-sensitive kinase (37, 38), implying that lower ROS levels in Pdk2-deficient chondrocytes may lead to reduced phosphorylation of p38 MAPK. Furthermore, p38 MAPK signaling itself can induce oxidative stress via MAP kinase-activated protein kinase 2 (MK2), potentially creating a positive feedback loop between p38 MAPK and oxidative stress in catabolic condition (39). Taking one step further, our data showed that p38 inhibitor significantly suppressed ROS generation in WT chondrocytes, whereas this inhibition of ROS was not observed in PDK2 KO chondrocytes (Fig. 6a). This implies that p38 MAPK does not influence ROS generation in conditions prone to OxPhos due to PDK2 deficiency. In other words, p38 MAPK may primarily contribute to an increase in ROS in situations of mitochondrial dysfunction, characterized by reduced OxPhos. Although OA is primarily a degenerative disease, omics data from OA articular cartilage have revealed a sustained increase in inflammatory signatures (40, 41). Our findings indicate that p38 MAPK could be an essential intermediary, linking metabolic alterations to inflammatory gene signature in OA cartilage.
Another significant observation regarding signaling changes associated with PDK2 deficiency is the more gradual reduction in AMPK phosphorylation (Thr172) by IL-1β stimulation. As implied by its name 'AMP-activated protein kinase', AMPK is activated by AMP, which typically increases under metabolic stress conditions (42). Conversely, when metabolic balance is restored and ATP levels rise, this leads to the kinase’s inactivation (43). Beyond metabolic conditions, the regulation of AMPK also involves several upstream kinases; LKB1, CaMKKβ, and TAK1 are key activators, while PKC, AKT, PKA, and PP2A contribute to its inactivation (44). With regards to oxidative stress, although AMPK activation helps its suppression, oxidative stress can, in return, lead to the inactivation of AMPK (45). Thus, the reduced levels of ROS observed in PDK2 deficiency could slow down the inactivation of AMPK, potentially aiding in the maintenance of metabolic homeostasis of chondrocytes under catabolic condition.
In conclusion, the loss-of-function of PDK2, which is upregulated under catabolic conditions of chondrocytes, leads to a metabolic shift towards OxPhos. This shift is associated with a reduction in oxidative stress and cellular senescence, and is protective in the progression of OA. Our findings suggest that metabolic modulation towards OxPhos deserves particular attention as a potential target for OA treatment.