It has been well appreciated that the bioenergetic demands of T cells increase dramatically over the resting state upon antigen stimulation. Metabolic remodeling from OXPHOS to aerobic glycolysis is a hallmark of T cell activation and is thought to be required to meet the metabolic demands of proliferation (12). The glycolysis not only expedited production of ATP (100 times faster), but also provided necessary intermediates for other metabolic pathways, such as PPP, hexosamine biosynthesis, amino acid biosynthesis, and fatty acid synthesis (FAS) (39). Recently, a role for the metabolic pathways in T cell activation and the differentiation of T cell functional subsets is beginning to be appreciated. Study demonstrated that cellular metabolism controls T cell lineage choices.

As a hub in the cellular metabolic signal network and sensor for environmental cues, mTOR controls multiple T cell lineage fates and functions, including Th17 and Treg cells (40–43) (Figure 2). The mTOR works by forming two complexes (mTORC1 and mTORC2) with a variety of proteins. mTORC1 and mTORC2 share the catalytic subunit mTOR but are distinguished by the scaffold proteins regulatory associated protein of mTOR (RAPTOR) and rapamycin-insensitive companion of mTOR (RICTOR), respectively (41). The mTOR complexes have been shown to be differentially expressed in murine effector T cells. Th17 and Th1 cells mostly express mTORC1, while Th2 cells predominantly express mTORC2 (44). mTORC1 is known to inhibited by AMP activated protein kinase (AMPK), which is regulated by liver kinase B1 (LKB1). In addition, mTORC1 promote Th17 differentiation by PI3K/AKT pathway and to promoting Glut1 cell-surface trafficking and glycolysis (45–47). As outlined above, the activation of mTORC1 signaling, either through PI3K/Akt-mediated signal transduction or suppression of LKB1-AMPK activity, enhance glycolysis and predisposed naïve T cells to differentiate into inflammatory Th17 cell (48, 49). mTOR, on the flip side, is critically required for Treg functional in mice (50, 51). Murine Treg cells have been shown to express less mTORC1 (52). Over-activation of mTORC1 promotes glycolysis, destabilizes Tregs and impairs their suppressive activity, leading to the loss of suppressive function in inflammatory conditions (45). Withal, FOXP3 expression inhibited mTORC1 activity and glycolysis, and enhanced mitochondrial oxidative (53). In a word, mTOR dependent glycolysis promotes Th17 development while hampering Treg cell formation. Study has demonstrated that rapamycin can blocks mTOR-dependent metabolic pathways including glycolytic activity and the production of IL-17 (54).

HIF1α is downstream of mTOR signaling, thereby playing an important role in regulating mTOR-mediated T cell differentiation (42, 55). Study demonstrated that HIF1α was selectively expressed in Th17 cells and regulates the expression of multiple glycolytic enzymes (55). Murine studies have shown a clear role for metabolism in reciprocally regulating the de novo differentiation of Th17 and Treg cells where HIF1α promote Th17 but to inhibit Treg cell differentiation when naive T cells were activated in the presence of TGF-β (52, 56). Deficiency in HIF1α resulted in greatly reduced glycolytic activity, and results in impairment in the Th17 differentiation program, but increased development of Tregs (54). Therefore, the HIF-1α-dependent glycolysis pathway is critical for regulating Th17 and Treg cell differentiation.

Myc is another critical regulator in T cell metabolism and a potential downstream target of mTOR (8, 57). During T cell activation, Myc as a transcription factor for GLUT1 (the main glucose transporter in lymphocytes), hexokinase 2 (HK2), pyruvate kinase (PK), and lactate dehydrogenase A (LDHA) induces glycolysis. Study has indicated that the pyruvate kinase muscle isozyme (PKM2) inhibitor suppresses glycolysis and the differentiation of Th1 and Th17, thereby reducing disease activity and alleviating experimental autoimmune encephalomyelitis (EAE) by inhibiting PK activity (58). Therefore, down-regulation of the transcription factorc-Myc is critical in inhibiting glycolysis (8). Also, Myc can be rapidly translocated to the plasma membrane through the Akt signaling pathway, thereby increasing glycolysis. Study has demonstrated that CD28 up-regulates glycolysis independent of the T cell receptor (TCR) engagement by increasing the Glut1 and c-myc in CD4+ T cells in patients with multiple sclerosis, which was critical for secreting inflammatory cytokines of Th17 cells (59). Additionally, Foxp3 inhibited c-myc in Treg cells, which was relevant to the impaired glycolysis and the enhanced oxidative phosphorylation (60). Collectively, the mTOR dependent glycolytic process is a metabolic checkpoint for T cell lineage choices and alters Th17 and Treg cell differentiation. Furthermore, studies indicated that rapamycin undertook the role in the survival and proliferation of Treg cell (61, 62). Recent research revealed that rapamycin recovery immune tolerance by intervening with metabolic reprogramming of Treg cells (63). The mTOR dependent glycolytic pathway has provided targets for therapeutic intervention of autoimmune and inflammatory diseases elicited by T cells.

Lipid metabolism is another important pathway by which immune cells can potentially be modulated (Figure 2). The development of Th17 cells was dependent on FAS, whereas differentiation of Treg cell required FAO (52, 64). It is well known that free fatty acids (FFA) in cytoplasm are first catalyzed by acyl coenzyme A synthetase (ACS) to produce fatty acyl CoA, then is transported into the mitochondria to produce acetyl-CoA by carnitine palmityl transferase (CPT)-1, CPT-2 and translocase and enter the tricarboxylic acid cycle (TCA). Murine study has demonstrated that Treg cell differentiation was inhibited following treatment with inhibitor of CPT1, while no effect on the differentiation of Th17 cell (52). Additionally, acetyl-CoA carboxylase (ACC) plays an important role in FAS and catabolism, and is a key enzyme catalyzing fatty acid anabolism. In humans and other mammals, it is a tissue-specific enzyme and exists in two gene forms, ACCI and ACC2. ACC1 exists in the cytoplasm of most adipose tissues (liver, adipose) and catalyzes rate-limiting reaction of long chain fatty acid (LCFA) synthesis. ACC2 is distributed in the heart and muscle tissue, and there are about 140 more amino acids than ACCI at the N-terminal. These amino acid sequences promote ACC2 to anchor on the mitochondrial outer membrane, thus catalyzing the oxidation of mitochondrial fatty acids (65). In the presence of ATP and Mg2+, HCO- as carboxyl donor, acetyl-CoA is carboxylated into malonyl monoacyl-coA by ACC, which plays a central role in Th17 and Treg cell differentiation by regulating de novo FAS. Inhibiting ACC using soraphen, a specific inhibitor of ACC isozymes, under Th17 polarizing conditions skews differentiation towards Tregs in mouse (64). This study also demonstrated that inhibition of ACC has the same effect on the differentiation of human T cells (64). Meanwhile, AMPK can indirectly inhibit FAS by hampering sterol regulatory element-binding protein 1 (SREBP1) and inactivating ACC1, thus inhibiting Th17 cell (66, 67).

On the other hand, FAO is generally associated with an anti-inflammatory phenotype of Treg cells and plays a vital role in maintaining the stability of Treg lineage. In murine studies, newly differentiated Treg cells demonstrated the highest the FAO compared with other effector T cell subsets. Blocking FAO significantly reduced the differentiation of Treg cell and downregulated its immunosuppressive function. Treg cells exhibited elevated phosphorylation and activation of AMPK, which is known to promote mitochondrial lipid oxidation and could give rise to increased FAO while inhibiting mTOR (8, 68, 69). In turn, inhibition of mTOR can lead to increased oxidation of exogenous FA, thus promoting Treg differentiation (70, 71). Furthermore, study have demonstrated that promoting FAO makes the Th17/Treg differentiate in favor of Treg cells, and inhibition of FAS can significantly reduce the ratio of Th17/Treg cell in the absence of glycolysis (72). During initial stage of T cell activation, supplement of lipids intensely restrain the production of Th17 cytokines, but not Treg suppressive function (52). Intriguingly, the up-regulation of FAS associates with down-regulation of FAO, which manifests that metabolic pathways interconnected (8). Collectively, the lipid metabolism affected the differentiation of CD4+ T cells, and inhibition of FAS or promotion of FAO modulates the Th17/Treg axis in favor of Treg cells.

Metformin, an AMPK agonist, which exerts anti-inflammatory effects on CD4⁺ T cells and induce Treg expansion and lipids alterations in various models of autoimmune disease, including experimental autoimmune encephalomyelitis and arthritis (73, 74). In the murine study, enhancing FAO via AMPK activation using the AMP analog 5-aminoimidazole-4-carboxamide ribonucleotide (AICAR), promoted the polarization of Treg cells and inhibited Th17 cells both in vitro and in vivo (75). FAO also plays a crucial role in maintaining the functional stability of Treg cells. Foxp3 extend its half-life by preventing from targeted degradation through post-transcriptional modifications, such as acetylation. Foxp3 acetylation is mainly dependent on acetyl-CoA, whose supply is increased upon breakdown of fatty acids (76). In addition, Rapamycin-mediated inhibition of mTOR increases the rate of FAO in skeletal muscle cells both in vivo and in vitro while decreasing flux into anabolic storage pathways by decreasing level of the enzyme, including ACC1, malonyl-CoA and glycerol phosphate acyltransferase (70, 77). These studies suggest that inhibition of mTOR and activation of AMPK play a physiological role in the regulation of fatty acid metabolism and resulted in prooxidative effects following CD4⁺T cell stimulation.

Cell-intrinsic metabolic pathways directly impact cellular fate and function. IL-2 as a pleiotropic cytokine, is a key regulator of T cell metabolic programs and is critical for the development of Tregs in the thymus and the regulation, proliferation, and maintenance of Tregs in peripheral tissues (78, 79). It has been demonstrated that the IL-2 critically drives TGF-β-treated CD4+ cells to differentiate into Foxp3+ Treg cells (79, 80). It is well known that high-affinity IL-2 receptor (IL-2R) is composed of three subunits: CD25 (the IL-2 receptor α chain, IL-2Rα), CD122 (IL-2 receptor β chain, IL-2Rβ) and CD132 (common γ subunit, γc) (81, 82). The co-expression of all three subunits is needed to confer high-affinity IL-2 binding to a cell. The signaling pathways that IL-2/IL-2R control the differentiation and homeostasis of both pro- and anti-inflammatory T cells is fundamental to determining the molecular details of immune regulation (83).

It is accepted that IL-2 activates two major signaling axes: the STAT5 and PI (3)K pathways (84, 85) (Figure 3). IL-2 activation of STAT5-mediated transcriptional programs is vital to the biological actions of IL-2 and is particularly important for Treg development, as it is necessary to initiate Foxp3 expression (86–89). A mice study showed that deficient of STAT5A/B fail to generate Tregs, demonstrating that the crucial role for IL-2-STAT5 in Treg development (90). In humans, loss of function in STAT5B is also associated with a loss of Tregs (91). Furthermore, in Stat5a/b^fl/fl mice, the deletion of the transcription factor STAT5 resulted in enhanced Th17 cell development when the absence of IL-2 or disruption of its signaling (92). STAT5 binding to the IL17a-IL17f locus inhibiting STAT3- mediated transcription of the IL17 gene and thereby suppressing Th17 cell differentiation (93). Another key signaling pathway in IL-2 regulation of Treg cells is mediated by the PI3K-Akt (94). And its biological effects could be mediated by the activation of cytosolic serine/threonine kinases. As an important serine/threonine kinase, mTOR senses various environmental cues, such as IL-2 (95, 96). Whereas IL-2 control of mTORC1 activity is dependent on JAK kinase activity. The IL-2Rβ chain binds JAK1, the IL-2Rγc binds to JAK3 (97). Thus, IL-2 receptor occupancy results in JAK-STAT activation (98). In addition, IL-2 regulates the expression and activation of MYC, mTORC1 and HIF1α/HIF1β, thereby guaranteeing the uptake of glucose (99). Recent studies have confirmed that the Treg dysfunction in the pathogenesis of AID can be restored by IL-2 effects of IL-2 (32, 34, 100–102).

The majority of microorganisms present in our body are resided in the human gut, affecting the balance between pro- and anti-inflammatory immune responses (103). The prominent role of the gut microbiota in the pathogenesis of autoimmune diseases, such as RA, SLE, SS and SSc, has been demonstrated by both human and animal studies (104). As a crucial actor in maintaining immune system functions, the gut microbiota actively participates in T cell metabolic reprogramming (105). Many of these effects are mediated by metabolites produced by the microbes such as short-chain fatty acids (SCFAs, particularly the acetate, propionate and butyrate), bile acids (BAs), and tryptophan (Trp), as well as by the transformation of environmental or host molecules (106, 107) (Figure 4).

The internal balance of fatty acids can also exhibit significant effects on CD4+ T cells, notably regarding the generation of Th17 and Treg cells (108). SCFAs produced by the gut microbiota metabolizing dietary fibers can diffuse into the cytoplasm and serve as a substrate for FAO, further producing acetyl-CoA, which provides fuel for TCA and OXPHOS. However, fatty acids promote T cell differentiation into effector or regulatory T cells depending on immunological milieu. It is well appreciated that SCFAs promote the development of extrathymic Treg cells via induction of Foxp3 in a histone deacetylase (HDAC) dependent manner, while LCFAs contribute to the differentiation of pro-inflammatory Th1 and Th17 cells via the MAPKs p38 and JNK1 (109–112). In this setting, LCFA-rich diets have been confirmed to aggravate T cell-mediated autoimmune pathological process in mice, conversely, mice fed with SCFA are protected (112). Withal, studies indicated that addition of SCFAs such as acetate, propionate, and butyrate can increases intestinal IgA production, cytokine and chemokine production as well as mediate immune responses (113, 114). The mechanism is also related to the regulatory function of mTOR on immune cells (110, 111, 115). SCFAs act on production pathways of energy and induce ATP production and AMP consumption, which are inhibitors and activators of AMPK, respectively. Thus, the inhibitor activity of AMPK on mTOR is repressed, leading to mTOR activation in B cell (116, 117).

BAs also have an essential impact on T cells differentiation. In effector T cells, the derivative of lithocholic acid (LCA), isoalloLCA, stimulates OXPHOS and the production of mitochondrial reactive oxygen species (mitoROS), which leads to the upregulation of FOXP3 through histone (H3K27) acetylation in Foxp3 promoter region, resulting in Treg differentiation. Conversely, another derivative of LCA, 3-oxoLCA, inhibits the differentiation of Th17 cells by directly interacting with the key transcription factor retinoid-related orphan receptor-γt (RORγt) (118). In addition, BAs act through the BA receptor Breg to regulate the function of RORg+ Treg cells, which plays a pivotal role in maintaining the stability of the colonic environment (119). These studies indicate that BA and its metabolites regulate Th17 and Treg cell balance and elucidate the mechanism by which gut microbes govern the host immune response.

In addition, the Trp catabolism plays an important role in the peripheral immune tolerance by preventing autoimmunity. Gut microbiome dysregulation induces autoimmunity in lupus-susceptible mice by altering Trp metabolism (120). The kynurenine (Kyn) pathway is the major catabolic route of peripheral Trp metabolism, which accounts for >90% in mammals (121). Trp be metabolized to Kyn in the presence of indoleamine-pyrrole 2,3-dioxygenase (IDO), an alternative inducible enzyme, which is induced by pro-inflammatory cytokines, such as TNF-α, IL-6 and IFN-γ, and eventually bring about the generation of the nicotinamide adenine dinucleotide (NAD+) (122, 123). The increased Kyn was observed in lupus-prone mice (120). mTOR, a sensor of T cell mitochondrial homeostasis and regulator of T cell differentiation, can be activated by Kyn in vitro. A recent study revealed that PPP was the most significant metabolome changes in peripheral blood lymphocytes of SLE patients by mass spectrometry. In addition, the accumulation of Kyn was related to the increase of PPP activity, and N-acetylcysteine (NAC), a precursor of glutathione, significantly reduced the level of Kyn and also reduces disease activity in SLE patients by inhibiting mTOR (124). The study by Lai et al. showed that sirolimus, the inhibition of mTOR, increased CD4+CD25+FoxP3+ Treg cells and CD8+ memory T cells and restrained IL-4 and IL-17 production by CD4 and CD4-CD8- double-negative T cells (125). Mitochondrial metabolism, including mitochondrial hyperpolarization, ATP production and oxidative stress, play important roles in the abnormal activation of SLE T cells (126). mTOR is widely thought to be involved in the inhibition of autophagy and can be regulated with rapamycin (127). Furthermore, study has indicated that the expression of GATA-3 and CTLA-4 in SLE Treg cells is reduced and the autophagy and inhibitory functions are reduced. Rapamycin can induce autophagy, restore the expression of GATA-3 and CTLA-4, and correct Treg functions (Figure 5) (128). Mitochondrial autophagy is caused by the dynamin-related protein 1 (Drp1), which leads to mitochondrial degradation. The study by Caza et al. demonstrated that Ras-related protein 4-mediated loss of Drp1 leads to mitochondrial accumulation and disease progression in lupus-susceptible mice (126). In conclusion, the abnormal Trp metabolic caused by gut microbial disorder may be one of the mechanisms of autoimmune activation. Also, the butyrate produced by gut microbes exert anti-inflammatory function by promoting differentiation of Treg cell (129). All in all, probiotics may be an option for the prevention and treatment of autoimmune diseases.

As every cell resides in a unique environment, no cell within our body is completely identical metabolically, phenotypically, and functionally, even of the same type, assessing the metabolic state of single immune cells becomes the focus of autoimmune disease (130). Meanwhile, the complexity of immune cells, the variety of different cell types in tissue, and heterogeneous clinical presentation make single-cell technologies a necessary means to uncover role of single cells (131). Recent years, the ability to extract and recognize biologically informative from single cell is improved with developments in the fields of machine learning. The field of immunology is dominated by high dimensional single-cell analysis, consisting of the first s sing-cell DNA sequencing based genomes reported in 2006, to the first realization of single-cell RNA-sequencing (scRNA-seq) based transcriptomics by Tang et al. in 2009, along with the recent Mass Cytometry, Met-Flow and mass spectrometry (MS) imaging that allowed directly measure metabolism in situ (132–138). Among them, scRNA-seq is the most widely used, which can efficiently amplify mRNA in isolated single cells and then conduct high-throughput sequencing, aiming to capture the entire transcriptome without the need to select target genes in advance, which can effectively solve the problem of transcriptome heterogeneity of immune cells and provides a multidimensional assessment of the transcriptional status. scRNA-seq allows individual cells to be defined as distinct metabolic entities and delineates corresponding transcripts from a metabolic perspective.

Caublomme et al. sequenced the transcriptome of 722 Th17 cells and validated four key molecules, Gpr65, Plzp, Toso and Cd5l, that governing the pathogenicity and susceptibility of autoimmune encephalomyelitis by calculating and analyzing a spectrum of cellular states in vivo to in vitro differentiated Th17 cells. Amongst them, Cd5l and Gpr65 associated with known regulatory and pro-inflammatory genes (139). Wagner has recently investigated metabolic differences associated with pathogenic/non-pathogenic Th17 cells by Compass, an algorithm to characterize the metabolic state of cells and relate metabolic heterogeneity to other cellular phenotypes based on single-cell RNA-Seq and flux balance analysis (140). This study also pinpointed the pathogenic Th17 cells is dependent on higher aerobic glycolysis and TCA to produce ATP, while nonpathogenic Th17 cells mainly rely on FAO for energy. The glycolytic enzyme phosphoglycerate mutase (PGAM) inhibits the pathogenicity of Th17 cell, and inhibition of G6PD alleviates neuroinflammation mediated by Th17 in vivo. Miragaia highlighted the metabolic heterogeneity among Tregs in colon, skin, spleen and lymph nodes of colon and skin in mouse by single-cell RNA-seq (141). In addition, a study based on Met-Flow found that metabolic remodeling reshaped the phenotype and function of T-cell subsets. Met-Flow, a high-dimensional flow cytometry approach, allows simultaneous analysis the metabolic state of key proteins and activation molecules on a single-cell in their representative pathways (142). This study also indicated that the expression of CD25, a characteristic surface marker molecule of Treg cells, is dependent on glycolysis and positively correlated with GLUT1 protein level.

Met-Flow overcomes inherent disadvantage of temporal inconsistency between mRNA abundance and protein concentration in traditional metabolic mRNA analysis (143, 144). The establishment of single-cell profiling technologies allows capturing the complex metabolic state of any cell, leading to a greater understanding of the role of metabolic heterogeneity in immune responses, which has taken the field of immunometabolism to a new level. Yet it is clear that additional effort are required to fully grasp the immunometabolism of Th17/Treg cell and for tasks such as metabolic clustering on the single-cell level.