Cross-talk of inflammation and chondrocyte intracellular metabolism in osteoarthritis

Introduction

Osteoarthritis (OA) is the most prevalent joint disease, characterized by articular cartilage degradation, joint stiffness, pain and loss of mobility in large and small joints. Recent estimates of global prevalence of OA point to nearly 654 million individuals over age 40 impacted by OA worldwide(1), directly and indirectly leading to increased morbidity and risk of all-cause mortality(2, 3). Approximately 215,000 patients presented to Emergency Departments in the United States between 2009-2013 due to OA, with an average cost of around $1,300 per patient(4). Even with these conservative estimates, it’s apparent that acute OA is a major financial and health burden that continues to increase as the disease prevalence rises around the world, requiring better understanding of the underlying mechanisms.

The precise pathophysiology of OA remains unclear, but existing research suggests that it is multi-factorial in nature, due to an interplay between biomechanical, molecular and genetic factors(5). Inflammation has been implicated in OA, acting to shift ECM metabolism from a matrix producing to a matrix-degrading state, characterized by the production of MMP’s and ADAMTS enzymes that break down articular cartilage(6-10). Articular cartilage degeneration leads to loss of its protective biomechanics features leading to joint destruction, further inflammation and OA progression(11).

Cell metabolism underlies nearly all cellular processes, providing energy and small molecule substrates to influence the overall cellular phenotype. Altered cellular metabolism as a player in disease states was identified famously by Otto Warburg in the study of tumors, noting increased reliance upon glycolytic metabolism in the setting of abundant oxygen(12). Alterations in cellular metabolism, including in response to inflammation, have been shown to mediate a multitude of cellular phenotypes (13, 14). While targeting inflammation has not yet panned out therapeutically for the treatment of OA, a greater understanding of the effects of inflammation on chondrocyte physiology is necessary to identify new targets. Our work and the work of others has suggested that inflammation can alter intracellular metabolism of chondrocytes, and the impact of this metabolic reprogramming may be critical for OA (9, 10, 15-18).

Chondrocyte metabolism relies upon both glycolysis and oxidative phosphorylation, as well as other metabolic pathways, for energy production. Chondrocytes can utilize a variety of substrates, ranging from simple sugars, to amino acids and fatty acids which are present in the synovial fluid of joints (19). However, chondrocyte metabolism is not utilized only for energy production, but also for the generation of ECM substrates through anabolic processes. Anabolic processes in this review will be considered any process that involves using small molecules to construct larger molecules, such as protein synthesis or gluconeogenesis.

There is extensive evidence that chondrocytes are capable of undergoing metabolic reprogramming in response to different stimuli or environmental conditions, with inflammation being one of them(Table 1)(10, 18, 20-25). While metabolism has traditionally been associated with ATP production, it is known that metabolic enzymes can have non-metabolic functions such as regulation of mRNA stability and transcriptional regulation, especially involved in inflammatory signaling (26, 27). For example, PKM2 is a glycolytic regulator that can also act as kinase to phosphorylate STAT3 and activate STAT3 signaling, highlighting cross-reactivity of metabolic systems with inflammatory signaling systems(28). Metabolites such as α-ketoglutarate can act as a co-factor for chromatin-modifying enzymes to modulate gene expression through epigenetic changes, indicating that metabolites play important roles outside metabolic pathways(29). Finally, enzymes such as GAPDH can bind to the mRNA of TNF-α to regulate its translation and expression(30). This review will focus on the existing work on the relationship between chondrocyte intracellular metabolism and inflammation, and its potential for the development of novel therapeutics.

Glycolytic changes and inflammation

Glycolysis is one of the most important metabolic systems in chondrocytes and the best studied. In glycolysis, glucose undergoes a series of enzymatic reactions to generate two molecules of pyruvate. Pyruvate can then enter the mitochondria to act as substrate for the TCA cycle or be metabolized to lactic acid by the enzyme lactate dehydrogenase (LDH). Stimulation of chondrocytes with inflammatory stimuli, such as IL-1β or damage-associated molecular patterns (DAMPs), leads to an increase in glycolytic activity of primary chondrocytes(9, 10). This has been shown by a variety of modalities including an increase in gene and protein expression of glycolytic enzymes such as LDHA, GLUT1, and PKM2, as well as metabolomics results indicating increased glycolytic flux and lactate levels (31-33). These results are likely reflected in patients as well, as one group found elevated expression of LDHA in TMJOA patients compared to healthy patients(34). Interestingly, this pattern of elevated glycolysis in the setting of inflammation also holds true across other cell types and has been well described in immune cells that exhibit metabolic plasticity(35).

The “purpose” of this increased glycolysis is unclear, but was suspected to be an adaptive response of cells to a high energy-demanding inflammatory environment. Under standard conditions, it was traditionally believed that lactate fermentation occurred in hypoxic environments since hypoxia inhibits mitochondrial electron transport chain activity, an oxygen dependent process(36). However, cells are capable of performing lactate fermentation even in the presence of adequate oxygen, a process known as aerobic glycolysis(37). As stated above, several groups have shown that under conditions of inflammatory stress, chondrocytes exhibit increased levels of aerobic glycolysis, in which glycolytic activity increases in the absence of changes in oxygen concentration. This is reflective of the well-described “Warburg effect”, first observed in tumor cells in the 1920’s(38). NF-κB activation by inflammatory stimuli was able to stimulate the increase in aerobic glycolysis, with NF-κB inhibitors preventing the metabolic shifting (10). TGF-B1 signaling is another pathway shown to promote glycolytic activity in articular chondrocytes(23). Other pathways are also important for this metabolic shift, including AMPK pathway, sirtuins, mTOR and HIF signaling, all of which are involved in inflammatory signaling (39). For example, Sirt5-null cells contained higher levels of TCA metabolites and trended towards decreased glycolytic metabolites(24). Sirt5 also regulated malonyl-CoA levels, an important regulator of lipid metabolism, potentially integrating these systems. HIF1α, a well-known regulator of the hypoxic response, also supports chondrocyte glycolysis, even in the absence of hypoxia(40), while its inhibitor, SETD7, prevents glycolysis(21).

Many hypotheses have been proposed regarding the role of metabolic plasticity in chondrocytes. This metabolic plasticity which appears essential for the chondrocyte inflammatory phenotype, likely occurs through non-energy producing functions of metabolic enzymes. An example of this phenomena is lactate dehydrogenase (LDH), a glycolytic enzyme that has been shown to play a critical role in chondrocyte inflammatory responses(10). LDHA content and activity is upregulated in chondrocytes in response to inflammatory stimuli, as characterized by increased lactate production. Pharmacologic or genetic inhibition of LDHA was potently anti-inflammatory, reducing the inflammatory response generated by IL-1β. Mechanistically, a novel non-metabolic function for LDHA was discovered as a ROS amplifier, leading to significant oxidative stress in the setting of inflammation(10). Reduction in LDHA-mediated ROS amplification via the small molecule FX11 elicited anti-inflammatory activity through the reduction in expression of IκB-ζ, a protein that interacts with NF-κB transcription factors, to drive the expression of cartilage-degrading genes such as MMP’s and interleukins. LDHA deletion in chondrocytes was also protective against post-traumatic OA in mice, highlighting its potential in pre-clinical models. Other groups have shown similar pro-inflammatory properties of LDHA in other disease models (39, 41).

Other examples of glycolytic targeting to alleviate OA have been reported. Inhibition of glycolysis using 2-deoxy-D-glucose (2-DG) was able to alleviate disease in a model of inflammatory arthritis(42). Another group went further and utilized multiple inhibitors of glycolytic activity to demonstrate that they were effective at reducing IL-1β induced cartilage-degrading gene expression in chondrocytes(43). They utilized 2-DG, Dichloroacetate and 4-methylumbelliferone (4-MU) to characterize metabolic changes utilizing seahorse data, highlighting that metabolism is consistently related to inflammatory gene expression. However, another group reported that deletion of Glut1 in chondrocytes increased OA severity in an animal model, suggesting that glucose metabolism is still critical for cartilage health and needs to be specifically targeted(44).

Mitochondrial dysfunction and inflammation

The function of mitochondria in chondrocytes and their critical role in OA pathophysiology has also been well studied. Mitochondria serve complex functions within cells, acting not only as an energy source, but also as regulators of cell survival and oxidative stress. Within the mitochondria exists the machinery required for the TCA cycle, a highly efficient system for generating high energy electron carriers, as well as the electron transport chain (ETC), the site of oxidative phosphorylation. These biochemical processes have been well described (45). It is well recognized that OA chondrocytes contain dysfunctional mitochondria as represented by loss of membrane potential, decreased biogenesis, decreased ATP production and other parameters. For example, mitochondria in OA cartilage have decreased mitochondrial membrane potential, decreased activity of ETC components and higher production of damaging ROS species such as superoxide(46, 47). OA chondrocytes also have decreased capacity for mitochondrial biogenesis, which is associated with increased cartilage degrading activity(48). Mechanical overloading of mouse joints also leads to overproduction of mitochondrial superoxide in cartilage, similar to the effect seen with inflammatory stimulation (49). More examples of mitochondrial dysfunction in OA are well described elsewhere. It is suspected that mitochondrial changes are associated with inflammatory signaling in chondrocytes.

Stimulation with inflammatory cytokines and subsequent NF-κB activation leads to a dramatic decrease in oxidative phosphorylation of human and mouse chondrocytes, which aligns with the previously noted increase in glycolytic activity(10, 50). Furthermore, stimulation with IL-1β or TNF-α inhibits Complex 1 activity in the ETC(51). These factors also significantly increase superoxide production from mitochondria in chondrocytes in vitro, displayed elegantly using a superoxide biosensor(52). This elevated ROS production from mitochondria leads to a potentially damaging state in the cell that overwhelms antioxidant systems such as SOD (53, 54).

Given that dysfunctional mitochondria support pathological processes such as ROS production, inhibition of these systems should protect against degradative changes. Indeed, inhibition of ETC activity using antimycin A or rotenone in the setting of inflammatory stimulation leads to a decrease in inflammatory gene expression via reduction in IkB-ζ protein levels (9). Similarly, amobarbital, an ETC inhibitor, or N-acetyl cysteine, an antioxidant, can protect against OA in a large animal model of intra-articular fracture by reducing pathological ROS (55). Other studies showed that rescue of mitochondrial function with supplementation of culture media with galactose was able to reduce inflammatory responses of chondrocytes, implicating mitochondrial dysfunction, potentially through reduction of oxidative stress (33, 56). Other mitochondrial activity-targeting molecules have also been studied as possible therapeutics in pre-clinical OA models(57). For example, 4-MU was able to improve mitochondrial function of chondrocytes to reduce cartilage degeneration(58). These studies indicate that ETC activity, and mitochondrial dysfunction in general, can be pathological in the setting of inflammation and that restoration of physiologic mitochondrial activity is likely beneficial.

Therapeutics may also be able to re-program mitochondria to improve their function, leading to decreased cartilage degradation. For example, BMP2 signaling has been shown to promote mitochondrial metabolism and oxidative phosphorylation (23) and replacing glucose with galactose in media drives mitochondrial respiratory activity in chondrocytes (56). Similarly, compounds such as DCA and 4-MU which inhibit glycolysis and promote mitochondrial respiration lead to decreased inflammatory gene expression (43). A radical new approach for promoting intracellular ATP production through the insertion of plant photosynthetic components into chondrocytes promoted mitochondrial function while increasing ECM gene and protein production , while decreasing ECM-degrading enzyme production, highlighting the importance of mitochondrial energy production for chondrocyte function(59). Finally, mitochondrial biogenesis is also important for chondrocyte health and AMPK-Sirt1 axis activation, which can be achieved through various well studied small molecules, is likely also protective in OA(48).

Independent of the ETC, the mitochondrial TCA cycle also holds several important enzymes that play many roles in energy production and signaling. One of the best studied of these enzymes is succinate dehydrogenase (SDHA), which has elegantly been shown to be involved in inflammatory regulation of immune cells(60). These enzymes are less well studied in chondrocytes and joint synovial cells but are likely critical for chondrocyte physiology. One group illustrated that 4-aminobutyrate aminotransferase (Abat), an enzyme involved in conversion of gamma-aminobutyric acid to succinate, a TCA cycle metabolite, was involved in promoting OA(22). Accordingly, the Abat inhibitor vigabatrin was effective at preventing OA progression. Further work is required to understand the role of TCA cycle enzymes in the inflammatory response of chondrocytes, especially through the use of metabolomics and proteomics.

Glutamine metabolism and inflammation

In addition to glucose metabolism, chondrocytes have the ability to perform a diverse array of metabolic processes for energy and substrate production, utilizing other sugars, amino acids and fatty acids(25). Glutamine is an amino acid of interest because it is present in the synovial fluid bathing the articular cartilage and may even be elevated on OA joints(61). Glutamine can be utilized by chondrocytes for energy production by entering the TCA cycle, but also for the production of glutathione, nucleotides and other amino acids(62). Glutamine deprivation significantly reduced oxidative phosphorylation and glycolytic activity in chondrocytes, emphasizing its role as an energy source(18). Another group demonstrated that in the absence of glucose, chondrocytes are able to upregulate glutaminolysis to fuel TCA cycle activity, revealing plasticity between these pathways (63). A recent study demonstrated that Sox9, an essential chondrogenic transcription factor, upregulates glutamine uptake and usage, and that glutamine metabolism regulates chondrogenic gene expression through epigenetic modification and regulation of oxidative stress (62), highlighting the relationship of chondrocyte development with glutamine metabolism.

In the context of inflammation, glutamine metabolic enzymes such as glutaminase (GLS) are upregulated with IL-1β stimulation of chondrocytes(18). This result is supported by recent findings demonstrating that murine and porcine chondrocytes post-injury have elevated levels of GLS, the rate limiting step of glutaminolysis(25). Preventing this increase in glutamine metabolism through glutamine deprivation was able to dose dependently decrease chondrocyte inflammatory response(18). Mechanistically, glutamine deprivation reduced oxidative stress in the setting of inflammation, likely from mitochondria, reducing NF-κB activity and IκB-ζ gene expression. Similarly, a small molecule inhibitor of GLS, CB-839, was also anti-inflammatory, identifying the therapeutic potential of this pathway (Figure 1). The role of other amino acid metabolic pathways requires further study in OA chondrocytes.

IκB-ζ at the intersection of metabolic and inflammation

IκB-ζ plays a critical role in the relationship between chondrocyte metabolism and inflammation, as has been displayed multiple times in this review. IκB-ζ has been shown to be an important disease mediator in OA, autoimmune disease, aging and cancer (64-67), but is a relatively understudied component of NF-κB signaling, which promotes OA initiation and progression(10, 68). IκB-ζ gene expression is driven by NF-κB activity, but its protein expression is modulated in a redox sensitive manner, such that it is stabilized in the setting of oxidative stress and evades proteasomal degradation(10). IκB-ζ interacts with NF-κB subunits p50 and p65/RelA to drive the expression of inflammatory and cartilage degrading genes involved in OA (9, 10, 67). Supporting this finding, IκB-ζ deletion nearly abrogated the inflammatory gene expression of chondrocytes, and was shown by another group to inhibit PTOA in vivo(69), highlighting its key role in OA pathology. In this manner, IκB-ζ can act as a redox sensor of the inflammatory response. Furthermore, given the close interplay between metabolism and cellular redox state, it is likely that IκB-ζ plays a role in the mechanism of many metabolic modulators.

Fatty acid metabolism and chondrocyte physiology

While less well studied than glucose metabolism, fatty acid metabolism is also predicted to be extremely important for OA pathophysiology. Obesity and metabolic disease lead to increased cartilage degradation in both human and pre-clinical models(70-72). However, the mechanism behind this effect is unclear, since the presence of metabolic syndrome leads to many systemic derangements and the impact on intracellular metabolism is not well known. The presence of an infrapatellar fat pad that provides a source of cytokines near the joint space highlights the need for greater understanding of the effect of fatty acids on chondrocyte physiology(73, 74). Since the fat pad is extra-articular but can undergo pathologic changes (75-77), it is unlikely that it can directly secrete synovial fatty acids but likely contributes to the pro-inflammatory joint microenvironment. Several studies have shown the pro-inflammatory nature of saturated fatty acids (SFA) such as palmitic acid(78), on chondrocytes, leading to increased ROS and inflammatory gene expression. Furthermore, SFA have been shown to induce mitochondrial and glycolytic dysfunction (79). One study demonstrated that OA chondrocytes have elevated levels of fatty acid oxidation, and another displayed accumulation of lipids in OA chondrocytes(80, 81). Inhibition of Cpt1, the rate limiting step of fatty acid oxidation, with etomoxir improved mitochondrial function, reducing catabolic gene expression and was protective against OA development in a mouse model(81). In another study, L-carnitine supplementation promotes ATP production and ECM synthesis in human articular chondrocytes, though the mechanism by which it does so is still unclear(82). Further characterization of changes in fatty acid oxidation and condensation is required in the context of OA.

Intracellular metabolism and senescence

We have described the existing evidence for the interaction of glycolysis, mitochondrial activity and other metabolic processes in modulating the chondrocyte inflammatory response. Senescent chondrocytes are also implicated in OA pathogenesis, primarily through the production of SASP factors that degrade cartilage and promote joint inflammation(83). However, the mechanism of senescent transformation and the factors that promote it are not well understood. Inflammation can drive senescent changes in chondrocytes through upregulation of key senescent markers such as p21 and p16 that also drive cell cycle arrest(9, 84, 85). In addition, NF-κB activation and downstream IκB-ζ, which we have determined is modulated by intracellular metabolism, are upregulated in senescent cells and are major drivers of SASP(86). While the role of intracellular metabolism and senescence has not been directly studied in chondrocytes, there are many inferences that can be drawn about the inflammation-metabolism-senescence axis. First, the same factors that drive metabolic reprogramming of chondrocytes, such as inflammatory mediators, have also been shown to be drivers of senescence(9, 84, 85). Second, metabolic reprogramming drives oxidative stress, a promoter of senescence(87, 88), and drives NF-κB - IκB-ζ activity, which is necessary for the expression of SASP genes (9). Based on these findings, it is plausible that targeting metabolism may be effective at targeting senescent cells, especially if they have a metabolic signature distinct from healthy chondrocytes. Supporting this finding, it has been demonstrated in other organ systems that senescent cells exhibit signs of metabolic reprogramming, characterized by increased glycolysis and defective mitochondria, as well as high energy utilization(89). One study demonstrated that senescent chondrocytes have increased lipid accumulation and fatty acid oxidation, and inhibition of this pathway reduces senescent transformation(81). Based on these findings, it is predicted that shifting cell metabolism towards a low glycolysis and increased mitochondrial respiration state may be effective at reducing cellular senescence in cartilage as well. Furthermore, metabolic modulators may be effective senolytics by targeting pathways that senescent cells depend upon.

Impact of metabolism on extra-cellular matrix production

Much of this review has focused on the impact of metabolic reprogramming on the reduction of the chondrocyte inflammatory response to reduce cartilage degradation. However, it is also known that metabolic pathways for anabolic functions, specifically for the generation of ECM protein components that compose the structure of articular cartilage, intersect with energy producing pathways such as glycolysis and glutaminolysis(90, 91). For example, it has been shown that glutamine is utilized for collagen hydroxylation in chondrocytes, independent of its role in energy production(92). Furthermore, HIF-1α, a known metabolic modulator, also regulates collagen synthesis (92). It is likely that metabolic reprogramming and modulation by therapeutic small molecules impact ECM production. However, there is little current evidence to suggest that articular cartilage late in life is able to regenerate by existing chondrocytes. Furthermore, ECM proteins are known to have an extremely long half-life, such that prevention of ECM degradation is predicted to have a significant impact on cartilage stability(93, 94). This is highlighted by studies which showed that elimination of viable chondrocytes in adult mice protected against OA or that elimination of senescent chondrocytes is also protective against OA(95, 96). However, metabolic modulators may still hold potential for cartilage regeneration.

For example, one group found that inhibition of LDHA using GSK2837808A was able to promote hyaluronan synthesis in synovial fibroblasts isolated from OA patients(34). Rotenone, an ETC inhibitor shown to be anti-inflammatory in the setting of IL-1β stimulation in one study, led to decreased proteoglycan content in an explant model of human cartilage(50). The hexosamine biosynthetic pathway is also an essential aspect of chondrocyte anabolism, leading to the generation of UDP-GlcNAc, which is a key component of ECM glycosaminoglycans. This biosynthetic pathway branches off of glycolysis and glutamine metabolism and metabolic modulators may allow or prevent flux into anabolic pathways, though they are poorly studied in this context. Use of 4MU, a hyaluronan inhibitor, was protective by modulating glycolysis and mitochondrial respiration, supporting findings in other systems that intracellular hexose pools can influence other metabolic pathways(43). Similarly, the reverse is also likely to be true, though it is less well studied. For example, inhibition of glutamine flux into the hexosamine pathway using an inhibitor of GFAT-1, azaserine, was likely damaging to cartilage(91). Use of FX11, a LDHA inhibitor, also led to decreased gene expression of Col2a and aggrecan through unclear mechanisms(10). However, it is unclear what role intrinsic chondrocyte mediated cartilage regeneration currently plays in the paradigm of OA treatment. In the future, it may be useful to examine metabolic modulators that drive ECM production to potentially support cartilage regeneration systems.

Conclusion

There is an enormous need in the field of OA and joint research to identify novel therapeutics that can be utilized to protect articular cartilage and maintain joint physiology. Given the multitude of factors that drive OA, ranging from biomechanical stress to inflammatory cytokines, it is essential to identify common mechanistic factors between all of these networks. The goal of OA treatment is two-fold: prevention of cartilage degradation and regeneration of damaged cartilage. Studies so far have demonstrated that metabolic manipulators hold great potential as therapeutics for the treatment of OA through reduction of cartilage degradation, though their role in cartilage regeneration is still understudied (Table 2). Inflammatory stressors have been demonstrated to shift all aspects of chondrocyte metabolism, which is critical for inflammatory and cartilage-degrading gene expression. We have provided several examples by which targeting specific metabolic pathways such as glycolysis and ETC, can regulate ECM catabolic activity in chondrocytes and affect OA development in vitro and in vivo. In addition, targeting metabolism may be effective at influencing multiple aspects of joint degeneration such as cartilage degradation and subchondral bone changes. Given that many of these metabolic changes appear to be conserved amongst various cell types(35, 97), understanding of metabolic processes can have implications in the fields of cancer biology, inflammatory diseases and more, all of which are known to have some degree of intracellular metabolic derangement.

While development of therapeutics is the ultimate goal of research into chondrocyte metabolism, the development of novel biomarkers to aid in the diagnosis of osteoarthritis is also of great importance. No such highly sensitive or specific diagnostic testing exists for osteoarthritis, leading to OA exacerbation often being a diagnosis of exclusion. Possible metabolic biomarkers include measurement of glucose, glutamine and downstream metabolic byproducts to identify the metabolic state of joint synovial cells. In addition, these metabolites can be correlated with levels of inflammatory markers within the joint synovial fluid or serum. The identification of metabolic-inflammatory biomarkers for OA can lead to early identification of disease and possible intervention, especially with novel therapies.