Role of Metabolism in the Immunobiology of Regulatory T Cells

CD4⁺CD25⁺FOXP3⁺ regulatory T cells (Tregs) are critically involved in the maintenance of immune tolerance to self and in the control of immune and autoimmune responses (1). Similar to conventional CD4⁺ T cells (Tconvs), Tregs have a high degree of plasticity that associates with different transcriptional programs, which, in turn, are impacted by cellular metabolism.

During the past decade, significant advances have been made in furthering the understanding of the molecular regulation of gene expression in Tregs (1–3). The integration of multiple cell signals can directly affect transcriptional programs and signaling pathways involved in cell proliferation, production of cytokines, and energy metabolism. In this context, it was reported that glycolysis and fatty acid oxidation (FAO) may be used differently by Tregs and Tconvs (4). In vitro, differentiated mouse Tregs display low glycolytic flux and oxidize lipids at higher rates compared with other T cell subsets (via AMP-activated protein kinase) (5). For human Tregs, which are hyporesponsive to TCR stimulation, the high glycolytic rate is supported by high mammalian target of rapamycin (mTOR) activity and is not associated with FAO (4). However, in vitro–proliferating human Tregs engage glycolysis and FAO, whereas Tconvs increase their metabolic activity by switching oxidative phosphorylation (OXPHOS) of the resting condition toward aerobic glycolysis to generate ATP (4). Aerobic glycolysis is far less efficient than OXPHOS and represents an unusual metabolic feature of proliferating T cells and cancer cells, a phenomenon known as the Warburg effect. Despite its low efficiency in energy production, aerobic glycolysis provides materials that are essential to the synthesis of nucleic acids and phospholipids (4, 6).

In vivo, human and mouse Tregs display a high glycolytic rate associated with hyperactivation of the environmental sensor mTOR (7–9). mTOR consists of two multiprotein complexes: mTOR complex (mTORC)1, which contains the regulatory-associated protein of mTOR (RAPTOR), and mTORC2, which contains the rapamycin-insensitive companion of mTOR. mTORC1 is sensitive to the immunosuppressant drug rapamycin and represents an important regulator of cell growth and differentiation (10, 11). Published evidence suggests that mTORC1 can act as a negative regulator of de novo differentiation of Tregs and as a positive determinant for their function (7, 8, 12). Mouse T cells in which mTORC1 has been ablated do not differentiate into Tregs, requiring concomitant inhibition of mTORC2 signaling to generate Tregs (13).

It must be noted that the metabolic differences between Tregs and Tconvs are significant. Although Tregs are highly dependent on mitochondrial metabolism with the flexibility to also oxidize lipid or glucose, Tconvs primarily convert glucose to lactate (4, 5, 14). Tregs appear to have a stronger respiratory capacity and preferentially oxidize glucose-derived pyruvate compared with Tconvs (15). The high expression of carnitine palmitoyltransferase 1a (CPT1a), the rate-limiting enzyme of FAO that allows the entry of acyl groups into the mitochondria, supports the possibility that Tregs can use multiple fuel sources (4, 5). Interestingly, mTOR controls several metabolic processes, including glucose metabolism, as well as fatty acid synthesis, which is important for Tregs to acquire full regulatory function. mTORC1 increases the expression of glucose transporters, including Glut1, on activated T cells and augments the intracellular concentration of glucose, supporting glycolysis (16). TCR and CD28–induced Akt signaling play an important role in Glut1-mediated glucose transport (5). mTOR signaling also induces glycolysis via the oncogene c-MYC, a crucial regulator of metabolic reprogramming in T cells (14). Specific deletion of RAPTOR, an obligatory component of mTORC1, leads to alterations in cholesterol and lipid synthesis in Tregs (8). The role of mTORC1 in lipogenesis is also supported by the findings that rapamycin blocks the expression of genes involved in lipid synthesis and alters nuclear localization of the master regulators of lipid homeostasis: sterol regulatory element-binding proteins (17).

Cell metabolism is central for Treg differentiation and is tightly linked to their function, in addition to supporting responsiveness to cell stimulation. Depending on nutrient availability and microenvironmental cues, Tregs can use alternate substrates and metabolic pathways for energy (Fig. 1). In the last decade, emphasis has been placed on the relationship between immune signaling and metabolic pathways that affect Treg function, particularly the role of the mTOR complex that senses environmental nutrients and growth factors for the modulation of Treg function and differentiation (7, 8, 13, 18). mTORC1 couples TCR and IL-2 signaling to Treg suppressive activity (8) and, metabolically, drives cholesterol and lipid biosynthesis through the induction of genes, including 3-hydroxy-3-methylglutaryl-CoA reductase, squalene epoxidase, and isopentyl-diphosphate δ isomerase 1, which are required for the expression of Treg markers, such as CTLA-4 and ICOS (8, 19).

Freshly isolated Tregs express high levels of mTOR and ATP because they actively proliferate in vivo (9). Their highly proliferative state reflects higher mTOR activity and ATP levels compared with Tconvs that do not proliferate in vivo. The high proliferative state of Tregs makes them refractory to TCR stimulation in vitro and, therefore, anergic. mTOR overactivation in Tregs can also depend on the capacity of Tregs to secrete leptin, an adipocyte-derived cytokine that activates mTOR via its class I cytokine receptor that is expressed on Tregs (Fig. 1B) (20). Inhibition of mTOR activity via leptin neutralization or transient rapamycin treatment reverses Treg hyporesponsiveness, inducing proliferation after TCR stimulation (7, 20). A possible explanation is that Tregs need to transiently reduce their metabolic rate to enter the cell cycle and proliferate, as suggested by recent reports that showed that transient reduction of glycolysis and mTOR activity (via leptin neutralization) in freshly isolated human Tregs before TCR stimulation reversed their anergic state in vitro (7, 20). Together, the data suggest that, although high mTOR activity renders Tregs refractory to TCR stimulation, it is also necessary for Tregs to proliferate over time once the cell cycle is engaged (4, 7, 20).

Another key issue for Tregs that proliferate in vivo is that active glycolytic-lipogenic pathways (18) would allow rapid generation of ATP and the transfer of glucose-derived carbons into metabolic intermediates for synthesis of proteins, nucleic acids, and lipids. Modulation of PI3K signaling can also alter cellular metabolism and FOXP3 expression. Genetic ablation of phosphatase and tensin homolog on chromosome 10, the primary negative regulator of PI3K, results in an increase in glycolysis, loss of Foxp3 expression, and the induction of effector T cells (21, 22). Recently, Wei et al. (23) showed that deletion of Atg7 or Atg5, two essential genes involved in autophagy, led to Treg loss (autophagy deficiency upregulated the metabolic regulators mTORC1 and glycolysis, contributing to a defective Treg function).

Glycolytic enzymes are also regulated by hypoxia-inducible factor-1α (HIF-1α), which is induced by TCR engagement (24, 25). HIF-1α is a transcription factor that, during hypoxia, binds to hypoxia-response elements and determines the transcription of genes that are important for cell survival under low-oxygen conditions (26), including those encoding enzymes required for glycolysis (27). HIF-1α also promotes the expression of Glut1 and enforces ATP synthesis by glycolysis, rather than OXPHOS, by upregulating pyruvate dehydrogenase kinase 1, an enzyme that inhibits the entry of pyruvate into the tricarboxylic acid cycle (28–30). HIF-1α expression is also dependent on external cues that are integrated by mTOR signaling (31), and HIF-1α is required for optimal Treg function because HIF-1α–deficient Tregs fail to control autoimmune colitis (32). However, the mTOR–HIF-1α axis also promotes Th17 differentiation, and lack of HIF-1α can result in diminished Th17 development and enhanced Treg differentiation that can protect mice from autoimmune neuroinflammation (25).

Finally, the key role of OXPHOS in the energy production of Tregs derives from the observation that deletion of regulators of mitochondrial activity, such as peroxisome proliferator-activated receptor γ coactivator 1a or sirtuin 3, inhibit Treg suppressive function in vitro and in vivo (33).

In sum, in vivo Treg metabolism is dynamic and finely regulated to ensure function, being intimately connected to oscillatory cues, such as the strength of TCR signal, cytokine milieu, and nutrient availability.

Earlier studies showed that the expression of genes involved in cell metabolism influences FOXP3 induction and IL-2 signaling (34–36). Using multiple pharmacological inhibitors and activators, it was shown that different metabolic programs could regulate Treg lineage differentiation in vivo and in vitro (25, 37, 38). Specifically, inhibition of glycolysis with the glucose analog 2-deoxyglucose, a prototypical inhibitor of the glycolytic pathway, promoted the induction of mouse Tregs in vitro in the presence of polarizing cytokines, such as TGF-β and IL-2 (5, 25). During glycolysis inhibition, the reduction in the mTOR-dependent HIF-1α transcriptional program resulted in FOXP3 induction (Fig. 1B). Absence of HIF-1α led to increased Treg differentiation and protected mice from autoimmune diseases, as HIF-1α was capable of promoting FOXP3 ubiquitination and subsequent proteasome degradation (Fig. 1B) (25, 39). Chronic treatment with rapamycin (which hampers glycolysis by inhibiting mTOR) induced de novo expression of FOXP3 and Treg expansion from naive CD4⁺ T cells in the presence of a high concentration of IL-2 (12, 40, 41). This strategy to expand Tregs could be seen as an apparent paradox, because rapamycin inhibits mTOR but IL-2 can activate it. It is known that chronic rapamycin treatment alone inhibits Treg proliferation (7, 42, 43), yet rapamycin in the presence of high doses of IL-2 allows Tregs to expand more robustly than with IL-2 alone (40, 41). We reported that this phenomenon might be due to the fact that Tregs need low mTOR phosphorylation (achieved with rapamycin treatment) to enter the cell cycle; after entering the cell cycle, IL-2 can help to reactivate mTOR, which is necessary for Tregs to proliferate over time. This oscillatory phenomenon for Treg expansion would occur in vitro and in vivo (4, 7, 20, 44).

Inhibition of glucose uptake and glucose oxidation by dichloroacetate promoted Treg differentiation (37), and glycolysis appeared necessary for the generation of human inducible Tregs (iTregs) from Tconvs in vitro in the absence of exogenous polarizing cytokines (i.e., TGF-β), drugs (i.e., rapamycin), or strong TCR stimulation (38). We (38) recently reported that inhibition of glycolysis with 2-deoxyglucose led to a differential expression of human FOXP3-splicing variants, including those required for the suppressive function of Tregs, whereas the inhibition of lipid oxidation supported iTreg differentiation by increasing the expression of the FOXP3-splicing forms that support regulatory functions (Fig. 1B). Other groups confirmed the notion that different human FOXP3-splicing variants have different capacities to control the generation and function of Tregs (45–48). In any case, FOXP3 expression is also impacted by pyruvate metabolism, a checkpoint in glucose metabolism (15). Pyruvate dehydrogenase contributes to the transformation of pyruvate into mitochondrial acetyl-CoA for oxidative metabolism. Acetyl-CoA levels can also control FOXP3 stability. The acetylated state of FOXP3 is reciprocally regulated by the histone acetyltransferase p300 and the histone deacetylase sirtuin 1. Acetylation of FOXP3 increases stable protein levels by preventing polyubiquitination and proteasomal degradation (49).

In contrast, a requirement of lipid uptake and oxidation for the expression of FOXP3 is shown by the use of etomoxir, a selective CPT1a inhibitor that significantly affects FOXP3 expression in mouse T cells (5). Lipid metabolism in Treg commitment also was apparent when using pharmacological inhibition of estrogen-related receptor-α, which impairs Th1, Th2, and Th17 responses, as well as Treg differentiation, in vitro. The addition of fatty acids to in vitro cultures rescued differentiation of Tregs, but not Th cells, because estrogen-related receptor-α upregulated Glut1 protein, glucose uptake, and mitochondrial processes (hampering FAO through CPT1a inhibition), thus favoring Tregs and not Th cells (50).

In conclusion, cell metabolism is highly dynamic in vivo and strongly related to in vitro experimental conditions that include TCR signal strength (dose and duration) and the presence of cytokines or drugs (51–53). Also, although the metabolic requirements of in vitro–differentiating Tregs under polarizing conditions have been actively analyzed, less has been done to identify the metabolic determinants of Treg induction in vivo.

Treg metabolism requires cues that include TCR signal strength, cytokine milieu, and nutrient availability. An emerging concept is that Tregs are functionally specialized, and their development, maintenance, and function are influenced by the local environment represented by the local milieu of metabolites, adipocytokines, and gut microbiota (54).

Purine catabolism is an important metabolic process that regulates the balance of proinflammatory ATP and immunosuppressive adenosine. Extracellular nucleotides, such as the ATP released by T cells during TCR stimulation, can contribute to autocrine modulation through the activation of purinergic P2X receptor (P2X7) that inhibits the Treg-suppressive activity through FOXP3 inhibition (55). Stimulation of P2X7 inhibits tissue-specific immunosuppressive potential of Tregs and facilitates conversion into Th17 cells during chronic inflammation. Pharmacological antagonism of P2X receptors or loss of P2X7 in Tregs ameliorates tissue inflammation by preserving Treg function (55). Also, the CD39 ectoenzyme on human Tregs produces AMP from ATP or ADP, which is subsequently converted to extracellular adenosine by the CD73 ectoenzyme expressed on Tconvs. Proper Treg function requires a coordinated expression of the adenosine 2A receptor on activated T cells to enable adenosine-mediated immune suppression (56–58). Adenosine can bind the adenosine 2A receptor and facilitate Treg generation and suppressive function (59).

Other metabolites that can affect Treg function are vitamins A and D, tryptophan, and arginine. Retinoic acid, the bioactive metabolite of vitamin A, promotes, with TGF-β, the conversion of naive T cells into FOXP3⁺ Tregs (60–62). Retinoic acid stabilizes FOXP3 expression and prevents IL-1β/IL-6–driven conversion of Tregs into Th1/Th17 cells (63). The active metabolite of vitamin D, calcitriol, also promotes growth of FOXP3⁺ and IL-10-producing Tregs, increasing the frequency of FOXP3⁺ cells when combined with TGF-β (64, 65). Glutamine is another amino acid that influences FOXP3 expression. Glutaminolysis provides, in active T cells, carbon and nitrogen for other proliferation-associated biosynthetic pathways, such as the hexosamine and nucleotide biosynthetic pathways involved in several T cell functions. Limited availability of extracellular glutamine shifted the balance from Th1 cells to Tregs (66). TCR stimulation of naive CD4⁺ T cells in the presence of low glutamine levels resulted in the conversion into FOXP3⁺ Tregs under Th1-polarizing conditions. Furthermore, TGF-β–induced Tregs exhibited a Treg-specific demethylated region with a methylation status similar to that of Tregs generated in the absence of glutamine (66). The explanation could be that glutamine is catabolized to generate α-ketoglutarate, which, in T cells, decreases generation of FOXP3⁺ cells and supports energy production through the tricarboxylic acid cycle, which is critical for Th1 cell commitment. Conversely, Song et al. (67) reported that glutamine administration in a mouse acute graft-versus-host disease model significantly increased the fraction of Tregs and inhibited graft-versus-host disease–induced inflammation and tissue injury in the intestine, liver, skin, and spleen.

Other metabolites stimulate the aryl hydrocarbon receptor (AHR), which controls the balance between Tregs and Tconvs. For example, kynurenine, a product of tryptophan catabolism generated by IDO, is an AHR agonist important for the generation, expansion, and suppressive function of stable Tregs (68–70). Dietary metabolites that can bind AHR, such as indole-3-carbinol and 3, 3′-diindolylmethane, can increase Treg infiltration into the CNS and ameliorate experimental autoimmune encephalomyelitis (EAE), a model of multiple sclerosis (71). Moreover, in skin grafts, dendritic cells (DCs) that upregulate enzymes that consume different essential amino acids determine the reduction in mTOR activity and the induction of FOXP3 in T cells because naive T cells fail to proliferate in response to Ag and can be converted into Tregs (72).

Finally, Morse et al. (73) showed that an arginine-rich cell-penetrating phosphorodiamidate morpholino oligomer, by targeting FOXP3, inhibited Treg function and induced effector T cell responses, suggesting that a phosphorodiamidate morpholino oligomer antisense-based strategy might be used as a potential tool to improve immunotherapy.

Molecules involved in the regulation of food intake and metabolism at the hypothalamic level, such as leptin and adiponectin, can also affect generation and proliferation of Tregs. The leptin/leptin receptor (LepR) axis controls the metabolic state and functional activity of Tregs. Tregs express LepR and are present as a specific fat-resident population producing leptin (20, 74, 75). Leptin deficiency in mice is associated with an increased frequency of Tregs and protection from multiple autoimmune diseases (20, 76, 77). In addition, a reduction in leptin levels induced by starvation or leptin neutralization increased Treg frequency and reduced inflammation and autoimmune progression (7, 78). The mechanism of action of leptin on Tregs and Tconvs is interesting. LepR engagement induced mTOR activity in both cell compartments; however, in Tconvs, thanks to their glucose metabolism, it supported proliferation and differentiation toward Th1/Th17 inflammatory phenotypes. In contrast, leptin produced by Tregs, in an autocrine loop, activated glycolysis and mTOR, making these cells hyporesponsive to in vitro TCR activation.

Adiponectin is another adipocyte-derived hormone that has effects on immune cells. For example, adiponectin treatment of DCs resulted in decreased expression of CD80, CD86, MHC class II, and IL-12p40; these effects favored an increase in the frequency of FOXP3⁺ Tregs and reduced Tconv proliferation and IL-2 production (79). Adiponectin-deficient mice developed worse EAE and greater CNS inflammation, demyelination, and axon injury, together with a defect in Treg number and function (80). In this context, Ramos-Ramírez et al. (81) showed that adipose tissue–resident Tregs expressed higher levels of adiponectin receptor 1 than Tregs in the spleen, and its expression on adipose tissue Helios⁺ Tregs correlated negatively with epididymal fat. In conclusion, these data suggest that mediators involved in the regulation of food intake affect Treg metabolism.

Microbial-derived molecules, such as short-chain fatty acids and polysaccharide A (PSA) from Bacteroides fragilis, seem to promote expansion and function of intestinal Tregs (82–86). Among all short-chain fatty acids, butyrate is the strongest Treg inducer in vitro through G protein–coupled receptors. Butyrate inhibits histone deacetylase that increases FOXP3 expression (82, 85). Indeed, when Tconvs are cultured in Treg-differentiation conditions, butyrate treatment enhances acetylation at histone H3 lysine 27 at the level of the FOXP3 promoter, conserved noncoding sequence 1 and 3 regions, thus leading to increased FOXP3 expression (82). Moreover, purified PSA increases Treg frequency and the expression of regulatory molecules, such as IL-10 and TGF-β (84), also via a direct interaction with TLR2 (86). In this context, Atarashi et al. (87, 88) showed that colonization of mice with murine fecal-derived Clostridium clusters IV and XIVa, or clusters IV, XIVa, and XVIII isolated from human feces, expanded Tregs and enhanced their activity in the colon. Mechanistically, those bacterial strains produce, in the intestine, a large amount of TGF-β, a major cytokine that promotes Treg differentiation.

As discussed above, mTOR critically integrates multiple environmental stimuli to regulate T cell activation, differentiation, and homeostasis. However, the same upstream stimuli that can activate mTOR in Tconvs can have different effects on the function and differentiation of Tregs (89, 90). In this context, mTOR controls differentiation and function, suggesting that its targeting could modulate Treg responses. Specifically, mTOR inhibition with rapamycin promoted Treg proliferation following acute treatment, whereas chronic treatment with rapamycin required IL-2 for Treg proliferation (7, 40, 41).

This aspect could relate to an increased activation of Akt by chronic rapamycin (91, 92) or to the fact that rapamycin suppressed mTORC2 and partially inhibited mTORC1 (93). Therefore, the positive effects of rapamycin and mTOR deletion on iTreg differentiation could be attributed to concomitant reductions in mTORC1 and mTORC2 activity (13), considering that deletion of rapamycin-insensitive companion of mTOR delays lethality in mice with RAPTOR deficiency in Tregs. It is interestingly to note that genetic Treg-specific deletion of RAPTOR resulted in a reduced proliferation of Tregs (8), further confirming that, in the absence of IL-2, chronic (pharmacological or genetic) mTOR inhibition negatively regulated Treg proliferation. Moreover, deletion of the gene encoding tuberous sclerosis 1, a negative regulator of mTOR, impaired Treg suppressive function and FOXP3 expression (94).

Another consideration is that mTORC1 is a critical positive regulator of metabolic programs for Tregs in vivo, and genetic deficiency of RAPTOR in mice Tregs leads to lymphadenopathy and multiorgan autoimmunity associated with T cell hyperactivity (8). This can be explained by the finding that RAPTOR regulates the expression of CTLA-4 and, in part, ICOS, and links the biogenesis of cholesterol with metabolic pathways that regulate the proliferation of Tregs (8). However, because those data derive from genetic knockout models, it cannot be excluded that they could be ascribed to activation of alternative, compensatory pathways.

However, excessive mTOR signaling can dampen Treg responses. Transient TCR stimulation induces PI3K-Akt-mTOR signaling that antagonizes FOXP3 expression (35). Freshly isolated Tregs from relapsing-remitting multiple sclerosis patients show an mTOR overactivation that correlates with reduced IL-2 signaling, Treg proliferation, and FOXP3 expression (89), suggesting that the proposed mTOR oscillatory activity can be lost in autoimmune conditions and lead to altered Treg homeostasis/proliferation (89).

Finally, in EAE, the inhibition of acetyl-CoA carboxylase 1 (ACC1) promoted the development of Tregs while restraining Th17 cells (18), because Th17 cells, but not Tregs, depend on ACC1-mediated de novo fatty acid synthesis and the glycolytic-lipogenic metabolic pathway to produce phospholipids for cellular membranes, whereas Tregs readily take up exogenous fatty acids (18). These results indicate fundamental differences between Th17 cells and Tregs with regard to their dependency on ACC1-mediated de novo fatty acid synthesis in those autoimmune models (18).

Tregs represent a major obstacle to effective antitumor immunity and immunotherapy. Indeed, the presence of Tregs correlates with poor prognosis in different tumor types (95, 96). The presence of specific metabolites in the microenvironment profoundly affects the suppressive function and lineage stability of Tregs (97). For example, IDO metabolizes tryptophan to kynurenine, an endogenous ligand that is able to activate AHR, which contributes to Treg induction (98, 99). Many types of cancers overexpress IDO, either in tumor cells or in cancer-associated cells, including macrophages, DCs, and endothelial cells. In the tumor microenvironment, IDO activity reduces local tryptophan availability in the proximity of Tregs. A low concentration of tryptophan activates a stress response pathway in Tregs through the protein kinase general control nonderepressing-2, which inhibits mTORC2 and prevents it from phosphorylating Akt. Inhibition of Akt contributes to the maintenance of Treg suppressive function (100, 101). Also, general control nonderepressing-2 activation in T cells may switch CD4⁺ T cell differentiation toward a regulatory-type phenotype (102, 103). Lastly, mouse tumor-draining lymph nodes contain IDO⁺ DCs that activate Tregs that sustain the intratumoral suppressive microenvironment (104). In papillary thyroid carcinomas, IDO expression correlates with Treg density in the tumor site (105). The antitumor activity of Tregs is also linked to their expression of CD39 and CD73, which generate adenosine from extracellular nucleotides. Adenosine is a potent inhibitor of T cell responses, and the A2A receptor is a major anti-inflammatory adenosine receptor involved in protection from tissue damage (106, 107). CD39⁺CD73⁺ iTregs hydrolyze ATP to 5′-AMP and adenosine and mediate suppression of immune cells that express adenosine receptors. These iTregs, expanding in response to tumor Ags and cytokines, such as TGF-β or IL-10, are presumably responsible for the suppression of antitumor immune responses and successful tumor escape (58).

Metabolic intervention could have relevant implications on Tregs in the development of strategies of intervention when immune tolerance is compromised. In this context, mice fed a high-fat diet and treated with the mTOR inhibitor rapamycin displayed significant changes in the inflammatory profiles in adipose tissue and liver, together with increased Treg function (108). Rapamycin protected against insulin resistance, increased energy expenditure, and reduced weight gain in diet-dependent obese mice (108). Moreover, in type 1 diabetes (T1D), rapamycin expanded Tregs and increased their ability to suppress Tconvs (109, 110). Combined treatment with rapamycin and IL-10 also inhibited T1D development and induced Tregs and long-term immune tolerance in NOD mice (111). In EAE, mTOR inhibition at the peak of disease ameliorated the clinical course and reduced CNS demyelination and axonal loss associated with an expansion of Tregs (112), and in vivo transient inhibition of mTOR enhanced Treg proliferation and ameliorated EAE (7). Recent studies in mice and humans reported broad efficacy for metformin, an AMP-activated protein kinase activator classically used for treating hyperglycemia and type 2 diabetes, in the treatment of autoimmune disorders. Treatment with metformin inhibited mTOR phosphorylation and increased FOXP3 expression, positively affecting the Treg/Th17 cell balance in humans with multiple sclerosis and mice with inflammatory bowel disease (113, 114). Additional pathways can influence Treg cells; for example, pioglitazone, a drug used for type 2 diabetes that stimulates peroxisome proliferator-activated receptor-γ, also restored number and function of visceral adipose tissue Tregs (74). The inhibition of ACC1, a key enzyme for the regulation of fatty acid metabolism, also promoted Treg development and impaired Th17 cell formation in EAE (18).

Lastly, the possibility of modulating immune responses through the manipulation of gut microbiota has been considered, particularly by targeting the Bacteroidetes and Firmicutes phyla that appear to stimulate Tregs and restrain Th17 cells in autoimmunity. Reports suggested that colonization of mice with the human commensal B. fragilis induced IL-10 production and FOXP3 expression, preventing colitis (84, 115). Oral administration of PSA from B. fragilis prevented EAE by stimulating DCs to convert naive T cells into Tregs (116, 117). Several Lactobacillus strains also ameliorated experimental colitis in mice via Treg induction (118, 119).

The field that links immunity and metabolism is expanding rapidly. Interestingly, nonimmunological disorders with a strong metabolic component, such as obesity and type 2 diabetes, have been linked to immune dysregulation, suggesting that metabolic alterations can be induced by, or be a consequence of, an altered state of immune tolerance. In addition, immune-mediated disorders, such as multiple sclerosis, display conspicuous metabolic alterations.

Tregs can represent a bridge linking metabolism and immunity as a result of their unique sensitivity to changes in the intracellular and extracellular milieu that reflects in metabolic cell changes. Experimental evidence (120, 121) shows that metabolic imbalance (i.e., overweight and obesity) can increase the risk for developing immune-mediated diseases, such as T1D and multiple sclerosis. Unfortunately, limitations in studying Treg metabolism are represented by their plasticity and differences reflecting source (human versus mouse), in vitro culture conditions (i.e., exogenous cytokines such as IL-2 and TGF-β), and TCR engagement.

Pharmacological approaches that can target Treg metabolism were considered recently, with the hope of using them as a means to restore impaired function (114). However, the field is still in its preliminary stages, particularly because those pharmacological agents might have effects on Tregs, as well as on all other immune cell subsets.

We thank Dr. Marianna Santopaolo and Dr. Alessandra Colamatteo for reviewing the manuscript.

The authors have no financial conflicts of interest.