Altered metabolic pathways regulate synovial inflammation in rheumatoid arthritis

Summary

Rheumatoid arthritis is characterized by synovial proliferation, neovascularization and leucocyte extravasation leading to joint destruction and functional disability. The blood vessels in the inflamed synovium are highly dysregulated, resulting in poor delivery of oxygen; this, along with the increased metabolic demand of infiltrating immune cells and inflamed resident cells, results in the lack of key nutrients at the site of inflammation. In these adverse conditions synovial cells must adapt to generate sufficient energy to support their proliferation and activation status, and thus switch their cell metabolism from a resting regulatory state to a highly metabolically active state. This alters redox‐sensitive signalling pathways and also results in the accumulation of metabolic intermediates which, in turn, can act as signalling molecules that further exacerbate the inflammatory response. The RA synovium is a multi‐cellular tissue, and while many cell types interact to promote the inflammatory response, their metabolic requirements differ. Thus, understanding the complex interplay between hypoxia‐induced signalling pathways, metabolic pathways and the inflammatory response will provide better insight into the underlying mechanisms of disease pathogenesis.

Hypoxia and metabolism in RA

Rheumatoid arthritis (RA) is a chronic progressive autoimmune disease, and is characterized by proliferation of the synovial membrane leading to degradation of articular cartilage and bone, and thus functional disability 1. An increase in the number of blood vessels with an immature phenotype, displaying incomplete pericyte coverage, has been demonstrated in the RA synovium 2, 3, 4. These blood vessels are dysregulated, and are thought to remain in a ‘plastic state’, thus primed for endothelial cell activation and sprouting, further facilitating immune cell recruitment 2, 3, 5. The increased metabolic demand of the expanding synovial pannus leads to oxidative stress, altered cellular bioenergetics and a hypoxic microenvironment, which further promotes abnormal cell function and synovial invasiveness within the RA joint 6. The original hypothesis that the inflamed joint is hypoxic was based on studies measuring surrogate markers of hypoxia in RA synovial fluids, where an increase in glycolytic metabolites in the joint was observed 7, 8. Subsequent studies have shown that the inflamed joint is profoundly hypoxic 9, 10, 11, levels of which are inversely associated with increased synovitis, dysfunctional vascularity and microscopic inflammation 11, 12. Furthermore, hypoxia‐mediated effects within the RA inflamed joint are not only dependent on HIF‐1α but involve complex interactions between key proinflammatory signalling pathways, hypoxia‐inducible factor 1‐alpha (HIF‐1α), nuclear factor kappa light‐chain‐enhancer of activated B cells (NF‐κB), Notch‐1 intracellular domain (Notch‐1), Janus kinase/signal transducers and activators of transcription (JAK‐STAT) and phosphoinositide 3‐kinase‐protein kinase B (PI3K‐AKT) in synovial tissue and cells 13, 14, 15, 16.

Evidence of a key role for metabolism in the regulation of synovial inflammation has recently emerged, where proliferation and rapid activation of immune and stromal cells requires a switch in cell metabolism from a resting regulatory state to a highly metabolically active state, in order to maintain energy homeostasis 6, 17. Under normoxic conditions, one glucose molecule enters the cell and is oxidized by glycolysis, generating two molecules of pyruvate, and in the presence of oxygen, pyruvate is decarboxylated by pyruvate dehydrogenase (PDH) then enters the tricarboxylic acid (TCA) cycle to produce 36 molecules of adenosine triphosphate (ATP) via a process known as oxidative phosphorylation (OXPHOS) (Fig. 1). In the absence of oxygen, pyruvate is diverted away from the TCA cycle, and instead pyruvate is converted to lactate by lactate dehydrogenase (LDH) in the cytosol, generating two molecules of ATP (Fig. 1). However, these metabolic changes can also occur in the presence of oxygen, a process known as the ‘Warburg effect’. Although glycolysis is less efficient in terms of ATP production compared to the TCA cycle it produces ATP more rapidly, thus better meets the energy demands of activated synovial immune cells. In addition, many glycolytic‐derived intermediate metabolites diverge into parallel pathways to promote nucleotide synthesis through the pentose phosphate pathway (PPP) to support cell proliferation and survival 18 and de‐novo fatty acid synthesis required for the expansion of the endoplasmic and Golgi membranes for the synthesis, trafficking and secretion of proteins 19.

Figure 1.

Schematic illustration of the key metabolic pathways. Glucose enters the cell via glucose transporters and enters the glycolytic pathway. Hexokinase 2 (HK2) converts glucose into glucose 6‐phosphate dehydrogenase (G6PD). Glycolysis generates pyruvate from glucose with the help of pyruvate kinase M2 (PKM2). This process generates energy in the form of adenosine triphosphate (ATP). Pyruvate is then either converted to lactate and secreted out of the cell or decarboxylated by pyruvate dehydrogenase and converted to acetyl CoA, which enters the tricarboxylic acid (TCA) cycle. The TCA cycle generates nicotinamide adenine dinucleotide (NADH), flavin adenine dinucleotide (FADH2) to feed into the electron transport chain (ETC), which produces 36 molecules of ATP. Glycolysis also feeds into the pentose phosphate pathway (PPP) to produce ribose, NADPH and amino acids. Amino acid metabolism can also feed into the TCA cycle to drive ATP production by the ETC. TCA intermediate citrate drives fatty acid synthesis while fatty acid oxidation drives TCA cycle further by generating acetyl CoA.

Evidence for this metabolic switch in RA has been demonstrated by several studies showing an increase in mitochondrial dysfunction in the RA synovium 20, along with increased expression of glucose transporters (GLUTs) 21 and glycolytic enzymes, including hexokinase 2 (HK2), pyruvate kinase M2 (PKM2), glyceraldehyde 3‐phosphate dehydrogenase (GAPDH), glucose‐6‐phosphate isomerase (GPI) and LDH 7, 8, 21, 22, 23, 24, 25. Marked accumulation of metabolic intermediates including lactate, glutamine, succinate 26 and itaconate 27 have been demonstrated in the RA joint which, in turn, can activate synovial cells 21, 28, 29, further perpetuating disease. In animal models of arthritis and in ex‐vivo RA synovial tissue explants, several studies have shown that re‐programming of these pathways with specific inhibitors leads to resolution of inflammation 21, 30, 31, 32, 33, 34.

The RA synovium is a multi‐cellular tissue, and many cell types interact to promote the inflammatory response. While synovial cells including T cells, macrophages, dendritic cells and synovial fibroblasts co‐exist in this microenvironment, they utilize metabolites differently. Indeed, metabolites produced by one cell can have profound regulatory effects on the function of another cell within the RA synovium. This highlights the need to dissect metabolic pathways and utilization of metabolites in different cell types within the RA synovium, as any potential metabolic therapeutic treatment may be cell type‐specific. While there is a significant number of studies examining metabolic pathways in different immune cell subtypes, few studies have investigated these pathways in the context of the individual cells in the RA joint.

The role of T cell metabolism in RA

Growing understanding of the relationship between metabolic processes within T cells and their activation, differentiation and effector functions is emerging. It is now well established that whereas resting T cells primarily utilize oxidative phosphorylation, aerobic glycolysis is promoted within minutes of T cell receptor (TCR) activation independently of transcription or translation via activation of PDHK1 35, followed by a more sustained induction of the glycolytic machinery 34. This rapid metabolic reprogramming towards aerobic glycolysis provides the energy and substrates necessary for the proliferation and effector functions of activated T cells. In addition, the differentiation of T helper (Th) into Th17 versus regulatory T cells (T_reg) cell is profoundly influenced by metabolism. Th17 cell differentiation requires glycolysis and HIF‐1α, whereas HIF‐1α suppresses T_reg cell differentiation 31, 36. Despite these recent advances, little is known about how metabolism influences human T cell responses, in particular those of already differentiated memory T cells that are found at sites of inflammation such as the RA joint. There are a number of site‐specific factors that are likely to impact upon the metabolism of T cells within the inflamed RA joint, including hypoxia and activation of HIF‐1α, altered nutrient availability and the accumulation of metabolites.

The hypoxic environment of the RA joint might be expected to drive glycolysis in T cells via the induction of HIF‐1α. However, in the circulation there is recent evidence to suggest that this is not, in fact, the case. Rather, naive peripheral blood CD4 T cells from RA patients have been shown to have a defect in 6‐phosphofructo‐2‐kinase/fructose‐2,6‐biphosphatase 3 (PFKFB3), which is the rate‐limiting enzyme in the glycolytic pathway 37. This inhibits their ability to utilize glycolysis upon activation and instead diverts glucose‐6‐phosphate towards the PPP, which runs in parallel to glycolysis, PPP supporting mainly nucleotide synthesis 37. The PPP generates the 5‐carbon building blocks for nucleic acids in addition to significant reductive power in the form of nicotinamide adenine dinucleotide phosphate (NADPH) 37. NADPH is required for the process of fatty acid synthesis; indeed, subsequent work by the same group revealed that increased fatty acid synthesis and accumulation of lipid droplets within the cell was a consequence of the reduced glycolytic flux within RA T cells 34. Furthermore, these conditions resulted in the up‐regulation of tyrosine kinase substrate with five SH3 domains (TKS5) expression, which promoted tissue invasiveness of the RA T cells. RA T cell invasiveness could be reversed via inhibition of fatty acid synthesis, suggesting that the fatty acid synthesis pathway may represent a therapeutic target in RA 34. These studies have provided important insights into the metabolic status of CD4 T cells in RA. The role of CD8 T cells in RA has received less attention than that of CD4 T cells; however, future studies will be important to determine whether CD8 T cells in RA are subject to similar metabolic control.

The inflammatory environment of the RA joint leads to the accumulation of various metabolites 26, 28, and it has now become apparent that these metabolites are not simply by‐products of metabolism but also exert important immune modulating effects. Interestingly, lactate has been shown to modulate T cell function directly via specific cell surface lactate transporters 28. Sodium lactate inhibits the migration of CD4 T cells by inhibition of glycolysis via direct inhibition of HK and PFK 28. In addition, sodium lactate also promotes the expression of interleukin (IL)‐17 by CD4 T cells 28. Furthermore, expression of the lactate transporter SLC5a12 in synovial tissue was shown to correlate with the clinical T cell score, and in‐vivo blockade of lactate transporters resulted in the release of T cells from the inflammatory site 28. These data suggest that the elevated lactate concentrations observed in the RA joint inhibit the glycolysis and migration of effector T cells, resulting in their retention at the site of inflammation and also increasing their production of IL‐17. These findings provide a rationale for the therapeutic targeting of specific lactate transporters on T cells in autoimmune diseases such as RA (reviewed in 38). Consistent with this, expression levels of the lactate transporter MCT4 and glycolytic enzymes HK2, GPI, triosephosphate isomerase (TPI), enolase 1 (Eno 1), PKM2 and LDH are significantly reduced in Th17 cells obtained from HIF1α^–/– compared to wild‐type (WT) mice 31, demonstrating the dependence of increased glycolytic enzymatic activity on the oxygen sensing pathway. Furthermore, in animal models of arthritis 3‐bromopyruvate (BrPA), a specific HK2 inhibitor, significantly decreased clinical arthritis scores in SKG mice, paralleled by an increase in the T_reg/Th17 ratio 30. While the exact underlying mechanisms involved are unclear, it is hypothesized that BrPA may alter the T_reg/Th17 ratio through differential regulation of their respective transcription factors forkhead box protein 3 (FoxP3) and retinoic acid receptor‐related orphan nuclear receptor gamma t (ROR‐γt). Finally, the TCA cycle metabolite succinate has been shown to exert proinflammatory effects on murine macrophages via induction of IL‐1β 39 which, in turn, might be expected to drive Th17 responses. Indeed, a recent study showed that deficiency of the succinate receptor GPR91 attenuates the severity of arthritis in Sucnr1^–/– mice, reducing expansion of Th17 cells 40.

The role of monocyte and macrophage metabolism in RA

Activated macrophages promote a number of proinflammatory mechanisms in the RA synovium through abundant secretion of proinflammatory cytokines; in addition, macrophages induce nitric oxide synthase, present antigen to T and B cells and drive bone resorption 41. RA synovial CD68⁺ macrophages correlate strongly with mitochondrial dysfunction and oxidative stress and are inversely related to in‐vivo synovial pO2 levels 11, 12, 20. HIF‐1α is expressed abundantly by macrophages in the RA synovium, compared to osteoarthritis (OA) and healthy control synovial macrophages 42. Differential signalling mechanisms in monocytes and macrophages under hypoxic conditions have been observed where monocytes preferentially utilize NF‐κB1, while macrophages utilize HIF‐1α 43. Furthermore, in collagen‐induced arthritis (CIA) models, decreased infiltration of myeloid cells to the joint, reduced paw swelling and disease development was observed in animals with HIF‐1α‐deficient macrophages 44.

In the context of metabolic changes, studies have demonstrated comprehensively that classically activated M1 macrophages have an ardent appetite for glucose, indicating a reliance on glycolysis, in contrast to M2‐like macrophages which rely on OXPHOS 45, 46. Indeed, recent studies have shown that the TCA cycle is broken at two key steps in M1 macrophages – after citrate and after succinate 39, 46. The majority of research has, however, focused on in‐vitro monocyte‐derived M1 and M2 macrophage models, and not synovial macrophages, due to the difficulty of isolating these cells; however, it is now becoming apparent that the M1 versus M2 paradigm may be an over‐simplification, and that a spectrum of activation states exists between these two poles within the synovial joint. In the context of RA, studies have shown increased lactic acid, citrate and succinate in RA synovial fluids 26, consistent with studies showing increased glycolysis and a broken TCA cycle in M1 macrophages. Lactic acid enhances secretion of IL‐6 and IL‐23 from monocytes and macrophages 47. Accumulation of succinate in macrophages promotes HIF‐1α activation which, in turn, induces IL‐1β production 39 and in animal models of RA, mice lacking the succinate receptor GPR91 show reduced macrophage activation and secretion of IL‐1β 33. Itaconate, a metabolic inhibitor of succinate dehydrogenase which has been shown to regulate succinate levels and secretion of inflammatory cytokines in activated macrophages 48, is also increased in the RA joint, and is associated with disease activity and response to therapy in animal models of arthritis 27. RA macrophages express high amounts of the glycolytic enzyme α‐enolase, which through autoantibody recognition induces secretion of proinflammatory cytokines 49. High concentrations of glucose have also been shown to increase IL‐1β secretion from RA monocytes through an NLRP3‐dependent mechanism 50. Solute carrier family 7 member 5 (SLC7A5), a key amino acid transporter, is increased in RA monocytes and macrophages, silencing of which leads to a significant reduction of IL‐1β 51. Finally, activation of 5′ AMP‐activated protein kinase (AMPK) in macrophages inhibits IL‐6 production, differentiation of M2 macrophages from synovial fluid monocytes and macrophage expression of IL‐6, tumour necrosis factor (TNF)‐α and NF‐κB in animal models of arthritis 52, 53.

More recent data have shown that macrophages from both RA and coronary artery disease (CAD) patients share metabolic abnormalities to promote inflammation. Disease macrophages appear to be in a hypermetabolic state, addicted to glucose consumption and producing more ATP compared to healthy macrophages 54. In addition, it has been demonstrated that macrophages from CAD patients are capable of memorizing both the metabolic and inflammatory signatures of their precursor monocytes, indicating that there is a memory bias towards a hyperinflammatory and hypermetabolic phenotype in disease macrophages 55. There has been speculation that epigenetic regulation may be the cause of this memorized immune response in both monocytes and macrophages 56, 57, 58. Thus, epigenetic reprogramming is emerging as a key mechanism in macrophage immunometabolism.

The role of metabolism in DC activation in RA

Dendritic cells (DCs) are key players in immunity, and link the innate and adaptive immune response through antigen presentation and cytokine production. Resting DCs, which differ from activated DCs as they are less motile, secretory and immunogenic, rely predominantly on OXPHOS for their energetic needs 59, 60, 61. Activation of DCs, either by antigen exposure or stimulation by Toll‐like receptor (TLR) ligands, results in a two‐phased metabolic shift comprised of an initial glycolytic shift with maintained OXPHOS, followed by elevated glycolysis and the cessation of OXPHOS in long‐term activated DCs. While the early increase in glycolysis is associated with elevated GLUT1 expression 61, 62 and lactate production 59, 60, 61, 62, 63, 64, the increased glucose demand by activated DCs is associated with the requirement of glucose‐derived metabolic intermediates into the PPP and fatty acid synthesis to facilitate amino acid and protein synthesis for the secretion of proinflammatory mediators associated with DC activation. Inhibition of glycolysis via HK2 blockade or fatty acid synthase in lipopolysaccharide (LPS)‐stimulated DCs results in a marked reduction in DC activation and immunogenicity 60. Molecularly, TBK‐IKKε/AKT signalling pathways and downstream target mammalian target of rapamycin (mTOR) have been strongly implicated in early activation of the glycolytic shift, which can be prevented by adhesion‐related kinase (ARK) blockade 62. In the context of RA, while also modulating the T_reg/Th17 ratio, BrPA‐mediated blockade of HK2 suppressed DC activation and cytokine expression 30, suggesting that the glycolysis pathway may represent a potential therapeutic target in RA.

During the later stages of DC activation, mTOR‐mediated induction of inducible nitric oxide synthase (iNOS) and stabilization of HIF‐1α is important for the complete commitment of activated DCs to glycolysis 59, 62, with iNOS‐derived nitric oxide (NO) blocking the electron transport chain (ETC) and sequestering OXPHOS 59 while elevating HIF1α‐dependent glycolytic gene expression 62, 63, 65. While these metabolic changes remain to be studied in the complex nutrient and/or oxygen‐deprived microenvironments, such as those found in the inflamed synovium, similar to macrophages, DC can also sense and respond to extracellular metabolites, such as succinate, butyrate and ATP 33, 66, 67, 68, 69 to potentially trigger a co‐ordinated cellular response reflective of the metabolic state of the surrounding microenvironment. A recent study demonstrated that the succinate receptor, GPR91, acts as a chemotactic that facilitates migration of DCs into the lymph nodes, which induces the expansion of Th17 cells and subsequent development of experimental antigen‐induced arthritis 40. This is consistent with previous studies that demonstrated that succinate can promote chemotaxis of DCs through activation of the succinate receptor 66, 67.

Role of metabolism in synovial fibroblast activation in RA

Synovial fibroblasts (FLS) are fundamental to disease progression and are active drivers of joint destruction in RA 70. FLS are characterized by increased proliferation, resistance to apoptosis and are potent producers of proinflammatory cytokines and matrix degradation enzymes, resulting in a highly invasive phenotype 71, 72. Previous studies have demonstrated increased mitochondrial dysfunction, coupled by changes in the ultrastructure of mitochondria in RA‐FLS in response to inflammatory stimuli, in addition to altered expression of mitochondrial genes associated with apoptosis, redox balance and mitochondrial protein transport 21, 23, 73. Metabolic profiling of FLS demonstrated increases in sugar metabolism (glycolysis and PPP) and amino acid metabolism (tyrosine and catecholamine biosynthesis and protein biosynthesis) 74. RA‐FLS also demonstrate a reliance on glutamine metabolism, with in‐vitro studies demonstrating inhibition of RA‐FLS proliferative and invasive functions under glutamine‐deprived conditions 29. Hypoxia, oxidative stress, TLR2 and TNF‐α activation all promote a switch in the metabolic profile of RA‐FLS, where an increase in the glycolysis : OXPHOS ratio is observed, paralleled by increases in surrogate markers of glycolysis; PFKFB3, PKM2 and GLUT1 and a more invasive phenotype 21, 23, 75, 76. Glycolytic inhibitors, including the PFKFB3 inhibitor 3‐(3‐pyridinyl)‐1‐(4‐pyridinyl)‐2‐propen‐1‐one (3PO) 21 and HK2 inhibitor, BrPA 76 significantly inhibit RA‐FLS invasion and migration capacity, secretion of proinflammatory mediators and activation of HIF‐1α, pSTAT‐3, NF‐κB and Notch‐1IC. Furthermore, accumulation of metabolic intermediates, including lactic acid, glutamine and succinate in the RA joint, have all been shown to further induce the RA‐FLS invasive phenotype 21, 76. Other key pathways involved in FLS activation include the PI3K/AKT1/mTOR, blockade of which resulted in repressed RA‐FLS function 77. Studies have shown that mTOR blockade of RA‐FLS invasion is mediated in part through regulation of focal adhesion kinase (FAK) signalling pathways and cytoskeletal rearrangement, a key mechanism involved in RA‐FLS movement 78. Furthermore, a recent study has shown that TNF‐α signalling co‐opts the mTOR pathway, directing specific signalling pathways that regulate the RA‐FLS response to inflammatory stimuli, an effect coupled with nutrient availability of specific amino acids 79. This is consistent with another study showing that IL‐17‐induced RA‐FLS migration is mediated through the amino acid transporter (LAT1) via mTOR 80.

Metabolomic profiling has shown changes in lipid metabolism of RA‐FLS compared to OA cells 76; however, the role of lipids in the regulation of FLS function has not been studied extensively. Interestingly, studies have shown that molecules in the choline pathway, which can interact with lipids, are highly activated in RA‐FLS, with the choline kinase (ChoKα) enzyme 81 and choline transporters 82 increased in RA‐FLS. Furthermore, a ChoKα inhibitor suppressed the migrative/invasive mechanisms of cultured RA‐FLS and in vivo ameliorated inflammation in the KxBN model 81. Thus, RA‐FLS are transformed from a quiescent state to an aggressive, invasive phenotype in this adverse environment through adaptation of metabolic pathways in order to meet their energy demands, which allows them to resist apoptosis and persist within the inflamed joint.

The above findings in FLS, however, are all performed in RA‐FLS which are maintained in culture in vitro, so while they provide some understanding of the metabolic pathways activated in RA‐FLS in response to specific stimuli, we still lack an understanding of the metabolic demands of RA‐FLS subtypes within this adverse inflammatory joint environment. Indeed, it has now become apparent that there are subsets of synovial fibroblasts within the inflamed joint that display pro‐ and anti‐inflammatory phenotypes, with effects also dependent upon positional memory 83, 84, 85. Therefore, further studies are required to examine if different metabolic profiles are observed depending on the RA‐FLS subtype and the anatomical location.

Metabolism of synovial endothelium

Endothelial cells (EC) rely heavily on glycolysis, with 85% of the ATP requirements coming from the conversion of glucose to lactate 86. The three EC subsets (tip, stalk and phalanx cells) differ in their metabolic requirements reflective of their individual functions (migration, proliferation and quiescence, respectively). When ECs are activated, metabolic changes dictate phenotypical differentiation, with tip and stalk cells showing increased glycolytic rates compared to phalanx EC 87. In the context of RA, few studies have extensively studied metabolic regulation of synovial vessels; however, they display similar dysfunctional morphology to that of the tumour vasculature 2. RA synovial blood vessel instability and oxidative damage correlate inversely with synovial pO2 levels 2, and display enhanced expression of the glucose transporter GLUT1 and the glycolytic enzymes GAPDH and PKM2 21. Exposure of ECs to hypoxic conditions and oxidative stress induces tube formation, migration and proinflammatory mediators, including vascular endothelial growth factor (VEGF), angiopoietins, monocyte chemoattractant protein 1 (MCP‐1), IL‐8 and matrix metalloproteinases (MMPs) 2, 75, 88. In RA‐FLS and EC, hypoxia induces activation of intracellular Notch‐1IC and its ligand DLL‐4, interaction of which is critical for EC tip cell selection, and lateral inhibition of the trailing stalk cell 13, 14. Blockade of the glycolytic enzyme, PFKFB3, inhibits angiogenic tube formation, secretion of proinflammatory/angiogenic mediators and key signalling pathways in RA‐FLS and EC 21. In addition, enriched expression of G6PI in synovial endothelial cells has recently been shown, with in‐vitro G6PI loss‐of‐function assays demonstrating the requirement of G6PI in mediating hypoxia‐induced angiogenesis in RA 89. Finally, in animal models of arthritis, succinate has been shown to induce synovial angiogenesis through VEGF‐dependent HIF‐1α pathways 90. Therefore, a deeper understanding of the metabolic perturbations in pathological conditions and their cross‐talk with immune and stromal cells might offer novel therapeutic opportunities, especially in early disease.

Targeting metabolism in rheumatoid arthritis

As highlighted above, studies examining metabolic pathways in RA have identified specific pathways, enzymes or metabolic intermediates that could potentially be targeted; however, many of our current treatment strategies are already known to alter metabolic pathways. For instance, glucocorticoids which are used routinely as first‐line treatment and regulate transcription of many metabolic genes 91 associated with glycolytic, autophagy and mTOR pathways 92, 93. Conventional disease‐modifying anti‐rheumatic drugs (DMARDs) for RA and psoriatic arthritis (PsA), including methotrexate, leflunomide and apremilast, are anti‐metabolic, where they target purine or pyrimidine nucleotide metabolism, the effects of which are known to inhibit both T cells and synovial fibroblast proliferation. Studies have shown that anti‐TNF‐α treatment decreases expression of GLUT1 and key glycolytic enzymes PKM2 and GAPDH in RA synovium in TNFi responders versus non‐responders 21. Tocilizumab, an anti‐IL‐6 receptor antibody, improved endothelial function and inhibited oxidative stress in RA leucocytes 94. Tofacitinib, a JAK1 and JAK3 inhibitor, reduces glycolysis in activated synovial fibroblasts, and in RA synovial explants inhibits key glycolytic enzymes, paralleled by reduced expression of key proinflammatory mediators and synovial fibroblast outgrowths 95. Indeed, interaction of STAT‐3 and PKM2 leads to activation of HIF1α, with subsequent induction of cellular invasive mechanisms creating a vicious PKM2/STAT‐3/HIF1 feedback loop 96. Furthermore, STAT‐3 blockade inhibits Notch signalling, which is involved in endothelial tip cell selection and fibroblast invasive mechanisms 1.

Directly targeting specific metabolic pathways has been demonstrated both in in‐vitro and in‐vivo models of arthritis where blockade of key glycolytic enzymes inhibits proinflammatory mechanisms 21, 30, 55, 97. Targeting metabolic intermediates including lactate acid, succinate, citrate, itaconate and lipids are also promising therapeutic avenues, where their cellular accumulation regulates synovial fibroblast invasiveness 21, T cell differentiation and migration 28, 31, 34, in addition to macrophage polarization 54. Furthermore, the anti‐cancer drug rapamycin, which targets mTOR, plays a critical role in directing T cell differentiation and function, inhibition of which promotes T_reg cell generation 98. mTOR is also a major repressor of autophagy, and in animal models of arthritis, systemic administration of rapamycin was shown to induce autophagy, paralleled by decreased severity of synovitis and reduction in IL‐1β expression 99. Finally, metformin, the anti‐diabetic drug, which acts in part by indirectly activating the energy sensor 5' AMP‐activated protein kinase (AMPK), has been shown to attenuate disease in mouse models of arthritis 100, effects that are mediated by inhibition of mTOR activity, enhanced autophagic flux, suppression of NF‐κB signalling and inflammatory cytokine production 101.

Summary

Lack of nutrients and a poor oxygen supply, paralleled by the increased metabolic demand of the expanding synovial pannus, leads to a bioenergetic crisis. In this environment synovial cells show adaptive survival responses by switching their utilization of specific metabolic pathways (Fig. 2) in order to satisfy their energy demands. This, in turn, activates key transcriptional signalling pathways which further exacerbates inflammation. However, this environment is complex, with many different cell types co‐existing within the inflamed synovium that display different metabolic requirements. Furthermore, metabolites secreted from one cell type have the ability to regulate the pathogenic phenotype of another, thus amplifying the inflammatory response. Understanding the opposing metabolic requirements of the different cell types will provide significant insights into their relevant pathogenic contribution to disease. However, currently the metabolic regulation of specific cell‐types/subsets within the inflamed synovium are poorly understood. Metabolic regulation can also have opposing effects depending on the disease settings. For instance, while Haas et al. elegantly described that high lactate levels at the site of inflammation act to entrap CD4⁺ T cells in RA by inhibiting T cell motility 28, in the tumour microenvironment accumulation of lactate impairs T cell function and hinders their cytotoxicity 102. Thus, lactate serves to boost proinflammatory mechanisms in the inflammatory milieu, yet suppresses immunity in the cancer setting. Enolase has been shown to induce monocyte/macrophage activation in models of RA 49, but promotes T_reg cell development in cancer models 103. Arginine metabolism drives a pro‐glycolytic phenotype in macrophages and DCs 60, 104; however, it promotes a metabolic switch to OXPHOS, exerting anti‐inflammatory effects in T cells 105, 106. Finally, the amino acid transporter SLC7A5, through leucine influx, induces proinflammatory cytokine secretion in RA monocytes and macrophages via mTORC1‐induced glycolytic reprogramming 107, and has also been demonstrated to play a key role in tumour survival and growth 108, 109. Thus, significant additional research is required if therapeutic strategies targeting metabolism are to be successful.

Figure 2.

Schematic illustration of the glycolytic switch in different cell types. The hypoxic conditions of the synovial joint drives hypoxia‐inducible factor 1‐alpha (HIF‐1a)‐induced glycolysis in some of the major cell types of the synovium. HIF‐1a induces expression of some of the key molecular switches to encode proinflammatory and pro‐glycolytic mechanisms. Specifically, 6‐phosphofructo‐2‐kinase/fructose‐2,6‐biphosphatase 3 (PFKFB3) is up‐regulated in endothelial cells and synovial fibroblasts in response to hypoxia. The mammalian target of rapamycin (mTOR) pathway is also involved in synovial fibroblast activation. Pyruvate dehydrogenase (PKM2) plays a central role in the metabolic switch observed in inflammatory macrophages and activated monocytes. DC metabolism can be directed by protein kinase B (AKT) signalling pathways in early activation while mTOR‐mediated induction of inducible nitric oxide synthase (iNOS) is important during later stages of DC activation, while glucose‐6‐phosphate dehydrogenase (G6PD) and PFKFB3 are key players in the metabolic switch observed in effector T cells.

Future research will require in‐depth characterization of immune and stromal cells at the site of inflammation. Isolation of single cell suspensions from the synovium for cell sorting, single cell transcriptomics and CyTOF analysis, paralleled by advanced imaging technology of synovial tissue sections, will improve our understanding with regard to which metabolic pathways are driving pathogenic phenotypes at the site of inflammation. Additional characterization of specific subsets, which have been recently described for synovial fibroblasts and macrophages, will further facilitate the development of new treatments that will specifically target pathogenic cell types as opposed to protective phenotypes. Furthermore, analysis of these pathways in pre‐RA (arthralgia), early disease RA, established RA and in responders versus non‐responders to current treatment strategies will also aid in our understanding of metabolism and its role in disease pathogenesis in RA.

Disclosure

None.