The IL-2 receptor is composed of three subunits: CD25, CD122, and CD132, which are, respectively, referred to as the α, β, and γ_c (also termed the common gamma chain) subunits (24). CD122 and CD132 are the sole mediators of downstream signaling and may form a heterodimer capable of low-affinity binding to IL-2 (16) (Figure 1). The alpha subunit CD25 does not signal, but is needed for high-affinity binding to IL-2. Most Tregs constitutively express all three subunits, while conventional CD4⁺ and CD8⁺ T cells constitutively express the CD122/CD132 dimer and only express CD25 upon activation. Conventional T cells begin producing IL-2 1 h after activation (25) and constitute the primary source of IL-2 in vivo. IL-2 activates three major signaling axes: the STAT5, PI(3)K, and MAPK/ERK pathways. STAT5 is particularly important for Treg development, as it is necessary to initiate Foxp3 expression (26).

The receptors for two other cytokines, IL-15 and IL-7, share subunits with the IL-2 receptor and partially compensate for losses of IL-2 or CD25 (27). The IL-15 receptor is a trimer that is strikingly similar to the IL-2 receptor, sharing the CD122 and CD132 subunits used for downstream signaling. Its alpha subunit CD215, like CD25, does not signal but instead confers high ligand affinity (28). On the other hand, the IL-7 receptor is a dimer composed of CD132 and a unique alpha subunit, CD127, which is capable of activating STAT5 (29).

Germline knockouts of IL-2 or its receptor components yield similar autoimmune phenotypes due to Treg deficiency (14, 30). Mice develop hemolytic anemia and colitis accompanied by thymic involution, lymph node hyperplasia, and splenomegaly, with elevated numbers of activated effector CD4⁺ and CD8⁺ T cells. Analogous findings have been reported in three clinical cases of CD25 loss (15, 31, 32), indicating that human Tregs are similarly dependent on CD25 and IL-2 signaling. These phenotypes are less severe than the scurfy phenotype resulting from Foxp3 deletion (33), most likely because the loss of IL-2 signaling also impacts effector T cells.

The fact that IL-2 and CD25 knockout mice maintain appreciable numbers of Foxp3⁺ cells in both thymus and spleen (34) indicates that IL-2 is not absolutely required for Treg development or subsequent survival, though it may be needed to achieve full suppressor function. The primary compensatory factor appears to be IL-15, as mice lacking both IL-2 and IL-15 are severely deficient in Foxp3⁺ cells (as are mice lacking either of the shared CD122 or CD132 subunits) (35, 36). In the presence of IL-2, however, IL-15 and IL-7 are dispensable for Treg development and function: IL-15^−/−(37), IL-7^−/− (38), and CD127^−/− (27, 35) mice have normal percentages of Foxp3⁺ cells and do not develop autoimmunity.

Although IL-2 signaling is an important component of Treg development (39), its roles following development are less thoroughly explored. It is generally believed that Tregs require constitutive IL-2 signals to survive and maintain Foxp3 expression, much in the same way these signals are needed during thymic development (40, 41). The role of IL-2 in Treg suppressor function has been difficult to address due to its roles in Treg survival during development. To date, the most prominent attempt to evaluate Treg function has been a Bim^−/− IL-2^−/− double knockout (42), in which targeting of the pro-apoptotic protein Bim was intended to decouple the roles of CD25 in Treg survival versus suppressor function. Although this study suggested that IL-2 is needed for full suppressor function, it should be noted that all germline knockout models of IL-2 signaling components are subject to a critical confounding factor: knockout mice develop lethal autoimmunity, which by its very nature is accompanied by immune activation and widespread inflammation. For this reason, it has been difficult to study how constitutive IL-2 signaling influences Treg lineage stability and function post development, much less study its effects on Treg metabolism.

Blocking antibody approaches can be dosed to avoid inducing autoimmunity. Although they are insufficient to address the issue of Treg function, due to off-target effects on effector T cells, these studies do not support IL-2 as a survival factor for Tregs. Anti-CD25 clone 7D4, widely used in commercial Treg magnetic isolation kits (43), induces loss of CD25 for up to 2 weeks following injection, yet Tregs persist and mice fail to develop autoimmunity (44, 45). It is critical to note that this antibody is distinct from anti-CD25 clone PC61, commonly as a tool to deplete Tregs in vivo, which is now recognized to act via opsonization for phagocytosis rather than through IL-2 deprivation (46–48).

Following immune cell activation by antigen or inflammatory signals, aerobic glycolysis and fatty acid synthesis are rapidly induced to support cell proliferation and cytokine secretion (57). This is reflected in the metabolic profiles of relevant immune subsets: effector T cells such as Th1, Th2, and Th17 cells show increased glycolytic rates following activation, as do effector CD8⁺ T cells. Tregs, like memory CD8⁺ T cells, rely on FAO for their basal metabolism but utilize some degree of aerobic glycolysis to properly execute their suppressor functions.

Beyond mere association with immune activation, several causal links have emerged between inflammatory stimuli, glycolysis, and Tregs (Figure 2). In T cells, signals through the TCR, CD28, or IL-2 activate the PI(3)K/Akt/mTOR cascade (58), which induces expression of the glucose transporter Glut1 to facilitate increased glycolysis (59). Akt also inhibits Foxo1 and Foxo3 transcription factors which are important for Foxp3 gene expression (60–62). mTOR engages Hif-1α, which may also be independently activated through toll-like receptor signaling, to promote the expression of key glycolytic genes (63). Hif-1α may also directly bind Foxp3 and target it for proteasomal degradation (64). Reciprocally, forced Foxp3 expression is sufficient to suppress glycolysis and promote FAO in vitro (20). Treg effector molecules such as CTLA4 and PD-1 suppress glycolysis in CD4⁺ T cells by activating PTEN to antagonize PI(3)K signaling and subsequent glycolysis, with PD-1 also actively promoting FAO by increasing expression of CPT1A (65). These data suggest that elevated glycolysis is detrimental to Treg lineage stability and suppressor function.

However, most studies showing detrimental effects of glycolysis on Tregs were performed in vitro, where T cell activation and glycolysis were driven to their maximum extent. Under certain conditions, glycolysis actually supports Foxp3 expression, promotes Treg proliferation, and potentiates suppressor function. Among in vitro induced human Tregs, the glycolytic enzyme Enolase-1 binds the Foxp3 promoter and its CNS2 regions. This represses transcription of a splice isoform containing Exon 2 (Foxp3-E2), which is needed for optimal Treg suppressor function. Engaging glycolysis forces Enolase-1 into the cytoplasm, thereby allowing transcription of Foxp3-E2 (66). Glycolysis also favors expression of the histone methyltransferase EZH2 by repressing inhibitory microRNAs such as miR-101 and miR-26a (67). EZH2 in turn binds Foxp3 to assist suppression of target genes (68), although no experiment has yet confirmed glycolysis-dependent EZH2 expression is essential for Treg lineage stability. The glycolytic metabolite phosphoenol pyruvate (PEP) can also increase Foxp3 expression through an NFAT-dependent mechanism. By inhibiting the calcium ATPase SERCA, PEP increases intracellular Ca²⁺ levels to promote nuclear translocation of NFAT, which facilitates interactions between the Foxp3 promoter and its CNS2 regions (69, 70).

A recent study suggests a possible resolution of these conflicting roles for glycolysis in Tregs. Using a Glut1 transgene to increase glucose uptake and glycolysis, the authors found that although elevated glycolysis boosts tTreg proliferation, it comes at the cost of their ability to execute suppressor functions (20). This suggests that for optimal Treg activity, a balance must be struck between the cell activating effects of glycolysis with its negative effects on the lineage.

Fatty acid oxidation is generally associated with an anti-inflammatory phenotype and maintenance of Treg lineage stability. One mechanism is through simple antagonism of glycolysis: Tregs express high levels of AMPK, which simultaneously promotes FAO while inhibiting mTOR and subsequent glycolysis (71). In the gut, short-chain fatty acids are also known to inhibit mTOR (72). They have the added benefit of stabilizing pTregs by inhibiting histone deacetylases (HDACs) such as HDAC6 and HDAC9 which would otherwise inhibit Foxp3 expression (73, 74). Reactive oxygen species generated as a byproduct of oxidative phosphorylation have been shown to promote Foxp3 stability by increasing activity of the transcription factor NFAT, which binds the CNS2 enhancer of Foxp3 (70, 75). In addition, Foxp3 may experience post-transcriptional modifications such as acetylation, which prevents Foxp3 from being targeted for degradation thereby increasing its half-life (76). Foxp3 acetylation is dependent on nuclear availability of acetyl-CoA, whose supply is increased upon breakdown of fatty acids. As with glycolysis however, under certain conditions FAO may antagonize Treg lineage stability. FAO promotes an increased NAD⁺/NADH ratio, which elevates the activity of the deacetylase SIRT1 (77). By deacetylating Foxp3, SIRT1 promotes Foxp3 poly-ubiquitination and subsequent proteasomal degradation (78).

pTregs and tTregs diverge considerably in their execution of FAO: although pTregs generally rely upon exogenous fatty acids for their metabolic needs (79), it is uncertain whether tTregs can import exogenous fatty acids in vivo (18). While the coming years will likely clarify this issue, available literature suggests one peculiar metabolic feature among tTregs. One of the major roles for mTOR signaling in tTregs is to drive synthesis of endogenous fatty acid stores, primarily along cholesterol biosynthetic pathways (80). Whether these endogenously synthesized fatty acids are then used for energy is not known, although these specific pathways are needed for tTreg proliferation and optimal suppressor function. Memory CD8⁺ T cells constitute the only major T cell subset known to synthesize endogenous fatty acids for subsequent FAO in vivo (18, 81) and rely in part on IL-7 and IL-15 to regulate these processes. IL-7 induces expression of the channel protein aquaporin 9, which facilitates glycerol import for fatty acid synthesis (82). IL-15 increases FAO by stimulating mitochondrial biogenesis and elevating expression of CPT1a, a key regulator of FAO (83). Given that Tregs appear to rely on IL-7 and/or IL-15 in the absence of IL-2, we speculate that tTregs from IL-2 or CD25 knockout mice may experience a shift from glycolysis to FAO, possibly with an associated loss of suppressor function. Whether similar events might occur in pTregs is unknown, although prior literature (11) suggests a loss of suppressor function in pTregs would result in increased fetal resorption among any IL-2 or CD25 knockout mothers which reach breeding age.

One of the most exciting prospects of immunometabolism is developing therapeutic interventions which can selectively target T cell subsets. Since activated effector T cells are more reliant on glycolysis than Tregs, studies have examined whether inhibiting glycolysis might improve outcomes in mouse models of autoimmunity and transplant rejection. Blocking glycolysis with 2-DG (a competitive inhibitor of hexokinase), or with dichloroacetate (an inhibitor of PDHK isoforms) reduced the severity of experimental autoimmune encephalomyelitis with associated decreases in the percentage of Th17, but not Treg, cells (63, 84). Similar outcomes were reported following inhibition of another glycolytic enzyme, acetyl-CoA carboxylase 1 (ACC1), with soraphen A or with T cell specific genetic deletion of ACC1 (79). Furthermore, treatment with metformin (an agonist of AMPK, which increases fatty acid uptake and oxidation) reduced airway inflammation and fibrosis in a murine asthma model (85). In the transplant setting, treatment with 2-DG, metformin, and a glutamine uptake inhibitor DON prolonged allograft survival in heart and skin transplants, in part by suppressing the proliferation of antigen-specific T cells and by increasing the relative frequency of Tregs (86).

Conversely, interventions that promote glycolysis enhance immune function, presumably by increasing the proliferation and function of effector T cells while inhibiting Treg function. Pharmacological blockade or genetic loss of PTEN leads to Akt-dependent inhibition of Foxo3a and subsequent loss of Foxp3 and tumor regression (87). Furthermore, increasing glycolysis through forced expression of the glucose transporter Glut1 in Tregs exacerbated pathology in an adoptive transfer model of colitis (20). Tregs recovered from this system were also found to have lower levels of Foxp3 protein.