Immunology & Metabolism

When tumor growth exceeds the vasculature’s ability to fully perfuse the tumor microenvironment with oxygen, regions of hypoxia are established and induction of the hypoxia-responsive transcription factor HIF-1α intensifies cancer cell glucose utilization and lactate release (Eales et al., 2016). Using ¹³C-labeled glucose, Hensley et al. (2016)) traced the fate of glucose within healthy lung tissue and tumors of patients with non-small lung cell carcinoma and found that even within a single tumor, heterogeneity in glucose utilization exists. Lesser-perfused regions of the lung tumors were associated with higher glucose metabolism whereas higher-perfused regions could utilize circulating lactate, transported through monocarboxylate transporter 1 (MCT1), as an alternative TCA cycle substrate (Sonveaux et al., 2008). Lactate metabolism in oxygenated cancer cells also increases glutaminolysis (Pérez-Escuredo et al., 2016). How this metabolic heterogeneity in tumor cells relates to intratumoral immune cell function has not been well elucidated, but exposure of NK and T cells to high concentrations of lactate impairs their activation of the transcription factor NFAT and production of the cytokine interferon gamma (IFN-γ) (Brand et al., 2016). Lactic acid also disrupts T cell motility and causes loss of cytolytic function in CD8 T cells (Haas et al., 2015). Moreover, decreasing conversion of pyruvate to lactate by genetic targeting of lactate dehydrogenase A (LDHA) in tumors helps to restore T cell infiltration and function (Brand et al., 2016), linking lactate production to immunosuppression observed in the tumor microenvironment (Figure 1).
Lactate uptake by tumor-associated macrophages (TAMs) also stimulates tumor progression by inducing vascular endothelial growth factor (VEGF) and arginase I (Arg1) expression through HIF-1α (Colegio et al., 2014). Moreover, chronic VEGF signaling in hypoxic areas leads to elevated glycolysis in endothelial cells, resulting in excessive endothelial sprouting and abnormal leaky vasculature (Goveia et al., 2014). Interestingly, inhibition of REDD1, a hypoxia-induced inhibitor of mechanistic target of rapamycin (mTOR), in TAMs increases their rates of glycolysis to a level that competes with neighboring endothelial cells for glucose and suppresses their angiogenic activity. This metabolic tug-of-war over glucose helps restore vascular integrity, improve oxygenation within the tumor, and prevent metastases (Wenes et al., 2016), providing an example of an intimate metabolic relationship that exists between cells in tumors.
Hypoxia also has considerable effects on TIL function, proliferation, and migration (Vuillefroy de Silly et al., 2016). Increases in HIF-1α activity by culturing T cells in physiologic normoxia (∼3%–5% O₂), genetic deletion of von Hippel-Lindau (VHL) factor, or inhibiting activity of the oxygen-sensing prolyl-hydroxylase (PHD) family of proteins, enhances CD8 T cell glycolysis and effector functions and promotes antitumor immunity (Clever et al., 2016, Doedens et al., 2013, Finlay et al., 2012, Wang et al., 2011). HIF-1α is also needed for the production of the metabolite S-2-hydroxyglutarate (S-2HG), which can drive epigenetic remodeling in activated CD8 T cells and enhance interleukin (IL)-2 production and antitumor defenses (Tyrakis et al., 2016).
Thus, one may expect that hypoxia would potentiate HIF-1α activity and TIL effector functions in tumors, however, this is not what is observed. In addition to signals received from IFN-γ, HIF-1α also induces the expression of the suppressive ligand PD-L1 in tumor cells, TAMs, and myeloid-derived suppressor cells (MDSCs) (Noman et al., 2014), and this can lead to TIL suppression via PD-1 (Figure 1). Moreover, recent work in both mouse and human tumors showed that CD8 TILs lose mitochondrial mass, membrane potential, and oxidative capacity, particularly within the most dysfunctional PD-1⁺ CD8 T cells (Scharping et al., 2016). The loss of mitochondrial function in TILs correlated with diminished expression of PPAR-gamma coactivator 1α (PGC1α) over time and a block in their proliferation and IFN-γ production. Perhaps severe hypoxia ultimately diminishes TIL effector functions. Indeed, respiratory supplementation of oxygen or treatment with metformin decreased intratumoral hypoxia and relieved several immunosuppressive features in the tumor microenvironment; the latter also served as an adjunct therapy that enhanced the antitumor effects of PD-1 blockade (Hatfield et al., 2015, Scharping et al., 2017). These findings suggest that remodeling the hypoxic tone in tumors may be an essential component to developing more efficacious forms of immunotherapy for patients.

Apart from hypoxia, the competition for nutrients and metabolites between tumor cells and infiltrating immune cells can be fierce, consequently influencing signal transduction, gene expression, and the metabolic activities of these neighboring cells. For example, tumor cells manipulate surrounding adipocytes to increase lipolysis to whet their appetite for fatty acids (Nieman et al., 2011). Cancer-associated fibroblasts degrade tryptophan that not only starves immune cells in the local environment of an essential amino acid, but also leads to the production of the immunosuppressive metabolite kynurenine (Hsu et al., 2016). Moreover, glucose is a critical substrate for the antitumor functions of effector T cells and M1 macrophages, which both require engagement of aerobic glycolysis for their activation and full effector functions (Buck et al., 2015, O’Neill and Pearce, 2016). Augmented aerobic glycolysis in cancer cells and endothelial cells places immune cells and their neighbors at odds (Figure 1). Glucose deprivation represses Ca²⁺ signaling, IFN-γ production, cytotoxicity, and motility in T cells and pro-inflammatory functions in macrophages (Cham et al., 2008, Chang et al., 2013, Chang et al., 2015, Macintyre et al., 2014). Several recent studies have demonstrated that the glycolytic activities of cancer cells may restrict glucose utilization by TILs, thereby impairing antitumor immunity (Chang et al., 2015, Ho et al., 2015, Zhao et al., 2016). Increasing glycolysis rates of tumor cells through overexpression of the glycolysis enzyme hexokinase 2 (HK2) suppressed glucose-uptake and IFN-γ production in TILs and created more immunoevasive tumors (Chang et al., 2015, Ho et al., 2015). Zhao et al. (2016)) found that glucose restriction imposed by ovarian cancer leaves microRNA repression of the methyltransferase EZH2 intact in CD8 T cells, reducing their survival and functional capacity. Thus, tumor cells can selfishly coerce or outcompete neighboring cells for glucose to supply their own metabolic demands in a manner that simultaneously suppresses immune defenses.
Amino acid deprivation in the tumor microenvironment serves as another metabolic checkpoint regulating antitumor immunity. Glutaminolysis in tumor cells is critical to replenish metabolites through anaplerotic reactions, which could result in competition for glutamine between tumor cells and TILs (Jin et al., 2016, Pérez-Escuredo et al., 2016). Glutamine controls mTOR activation in T cells and macrophages and is also a key substrate for protein O-GlcNAcylation and synthesis of S-2HG that regulate effector T cell function and differentiation (Sinclair et al., 2013, Swamy et al., 2016, Tyrakis et al., 2016). TAMs, MDSCs, and tolerizing dendritic cells (DCs) can suppress TILs through expression of essential amino acid (EAA)-degrading enzymes such as Arg1 and indoleamine-2,3-dioxygenase (IDO) (Figure 1) (Lee et al., 2002, Munn et al., 2002, Rodriguez et al., 2004, Uyttenhove et al., 2003). Indeed, inhibitors of Arg1 and IDO are under investigation as therapeutic targets in clinical trials (Adams et al., 2015). Several recent studies have highlighted the critical roles of other amino acids such as arginine, serine, and glycine in driving T cell expansion and antitumor activity, but how the availability of these fluctuate within the tumor microenvironment is not clear (Geiger et al., 2016, Ma et al., 2017). Currently, a knowledge gap exists on how the availability of various nutrients and metabolites vary across tumor types, genotypes, or even spatially within tumors to affect antitumor immune responses.
Bioactive lipids, modified lipoproteins, and cholesterol metabolism within the tumor are also important mediators of immune cell function. Like macrophages in atherosclerotic plaques, DCs in the tumor can accumulate oxidized lipoproteins through scavenger receptor-mediated internalization and formation of lipid droplets, which can ultimately impair their ability to cross-present tumor antigens and activate T cells (Cubillos-Ruiz et al., 2015, Ramakrishnan et al., 2014). Expression of lectin-type oxidized LDL receptor 1 (LOX-1) selectively marks MDSCs and oxidized lipid uptake and lipoprotein metabolism contributes to their T cell suppressive functions (Condamine et al., 2016). In addition, blocking cholesterol esterification in TILs by targeting ACAT1 pharmacologically or genetically increases intracellular levels of cholesterol and confers superior T cell responses in a model of melanoma (Yang et al., 2016). It is possible that as immune cells adapt to different tumor microenvironments and the limited availability of “immune stimulatory” nutrients, they become more dependent on alternative fuel sources (such as fats or lactate) that are less conducive to supporting antitumor effector functions. In summary, more elaborate knowledge of these forms of metabolic cross talk or competition between cells within tumors is needed before one can begin to think about how to manipulate these relationships in a manner that alters tumor progression.

As TILs adapt to the tumor microenvironment, they progressively lose their ability to respond to T cell receptor (TCR) stimuli, produce effector cytokines, and proliferate—a process termed functional exhaustion or hyporesponsiveness. This is in part due to the upregulation of several inhibitory receptors like PD-1, LAG3, TIGIT, and CTLA-4 that desensitize T cells to tumor antigens (Wherry and Kurachi, 2015) (Figure 1). PD-1, its ligand PD-L1, and CTLA-4 are important checkpoints for T cells in tumors and the targets of a new and powerful class of cancer treatments that elicit effective and durable responses in patients across multiple cancer types (Ribas, 2015). Interestingly, both chronic exposure to antigen or environmental triggers such as glucose deprivation can upregulate PD-1 (Chang et al., 2013, Wherry and Kurachi, 2015). PD-1 not only suppresses TCR, PI3K, and mTOR signaling in T cells, but also dampens glycolysis and promotes fatty acid oxidation (FAO)—features that may enhance the accumulation of suppressive regulatory CD4 T (Treg) cells in tumors (Bengsch et al., 2016, Parry et al., 2005, Patsoukis et al., 2015). Indeed, blockade of PD-1 re-energizes anabolic metabolism and glycolysis in exhausted T cells in an mTORC1-dependent manner (Chang et al., 2015, Staron et al., 2014). This breathes caution into the types of drug combinations one may consider with α-PD-1:PD-L1 blockade or other forms of immunotherapy. Metabolic interventions, such as the use of mTOR inhibitors, must be targeted specifically to avoid unintended compromises of immune cell function. The PD-1:PD-L1 axis may also directly affect the metabolic activity of tumor cells (Figure 1). It was shown that PD-L1 expression correlated with the rates of glycolysis and the expression of glycolytic enzymes in those cells (Chang et al., 2015). Furthermore, checkpoint blockade antibodies including α-PD-L1 led to an increase in extracellular glucose in tumors in vivo that likely contributed to the improved TIL function and subsequent tumor regression observed. On this note, tumor cell-intrinsic PD-1 expression may counterintuitively increase intrinsic mTOR signaling and tumor growth (Kleffel et al., 2015). Collectively, these findings suggest there may be broader role of the PD-1:PD-L1 axis in cellular metabolism that extends beyond T cells.

Improving the proportion of patients that respond to immunotherapy is an intense area of study, ranging from the search for biomarkers of response, target discovery, to testing new combination therapies. Likely, the most effective therapies will coordinately target co-inhibitory receptor-to-ligand interactions and restore a T cells’ ability to utilize metabolic substrates necessary to sustain their effector functions. Although not discussed at length here, other factors impinging on immune cell metabolism that one should consider in designing anticancer immunotherapies are the accessibility of growth factor cytokines that modulate nutrient transporter expression on immune cells and the fact that immune cells migrate into tumors to exert their effector functions, a metabolically demanding process. When cells move, their intracellular architecture, controlled in part by the actin cytoskeleton, is continuously remodeled. In an elegant dissection of PI3K-dependent growth factor signaling in epithelial cells, Hu et al. (2016)) found cytoskeletal dynamics and glycolysis to be uniquely intertwined. The authors showed that addition of growth factors or insulin activated Rac downstream of PI3K, causing a disruption of the actin cytoskeleton. Loss of this structural integrity released bound aldolase from filamentous F-actin, increasing its catalytic activity. Chemical and genetic inhibition of PI3K, Rac, or actin dynamics modulated glycolysis via mobilization of aldolase (Figure 2). It will be interesting in future studies to explore how other glycolytic enzymes or even the mitochondria in immune cells restructure their metabolic activity through changes in cytoskeletal morphology in response to growth factor or pathogen-derived signals and whether this regulatory circuit can be therapeutically targeted, especially because directed cellular migration is such an integral feature of how the disseminated immune system can focus its attention on points of infection or damage.

Additionally, manipulating metabolic enzyme expression to help T cells adapt to metabolic perturbations in the tumor microenvironment may be another viable strategy to improve antitumor immunity (Clever et al., 2016, Doedens et al., 2013, Ho et al., 2015, Scharping et al., 2016), especially for adoptive cell therapy, a personalized form of cancer treatment that allows for the manipulation and expansion of a patient’s antitumor T cells prior to re-infusion. Seemingly paradoxical is the observation that dampening effector T cell differentiation by impairing glycolysis and boosting mitochondrial FAO and oxidative phosphorylation (OXPHOS), metabolic pathways that favor the formation of resting memory T cell populations, potentiates effector T cell survival and functional capacity against tumors used in adoptive cell therapy (Buck et al., 2016, Sukumar et al., 2013). On the one hand, the engagement of aerobic glycolysis by activated T cells generates by-products of intermediary metabolism that supply substrates used to build biomass and fuel proliferation, provides an avenue for cells to support the equilibrium of reducing and oxidizing equivalents used to release energy, such as NAD⁺/NADH, and regulates the efficient production of effector cytokines critical for tumor regression (Buck et al., 2015, Chang et al., 2015, Pearce et al., 2013). Activation initiated by TCR ligation and binding with costimulatory molecules also augments OXPHOS in T cells (Chang et al., 2013, Sena et al., 2013). Mitochondria undergo biogenesis and take on a grossly punctate and dispersed morphology with expanded cristae junctions (Buck et al., 2016, Ron-Harel et al., 2016) (Figure 2). During this process, the mitochondrial proteome remodels itself to increase mitochondrial one-carbon metabolism. Knockdown of SHMT2, the first enzyme in this pathway, impairs CD4 T cell survival and proliferation in vivo (Ron-Harel et al., 2016). The generation of mitochondrial-derived ROS is also critical for the activation and expansion of antigen-specific T cells (Sena et al., 2013). As previously discussed, TILs that cannot sustain mitochondrial function have compromised functionality within the tumor microenvironment, and rescuing mitochondrial biogenesis in effector T cells improves antitumor immunity (Bengsch et al., 2016, Scharping et al., 2016).
However, on the other hand, dampening regulators of glycolytic metabolism, such as mTOR or c-Myc, and increasing mitochondrial FAO-dependent OXPHOS favors the formation of long-lived memory T cells (Araki et al., 2009, Cui et al., 2015, O’Sullivan et al., 2014, Pearce et al., 2009, Pollizzi et al., 2016, van der Windt et al., 2012, Verbist et al., 2016) (Figure 2). More recent evidence postulates that while oxidative metabolism and FAO characterizes the generation of long-lived stable central memory T cells (Cui et al., 2015, O’Sullivan et al., 2014, Pearce et al., 2009, van der Windt et al., 2012), augmenting glycolysis genetically via VHL deletion favors the formation of effector memory T cells instead, which have low levels of TCF-1, a transcription factor that is expressed in stable populations of central memory T cells capable of self-renewal (Phan et al., 2016, Zhou et al., 2010). It was also recently shown that activated lymphocytes unequally eliminate aged mitochondria in sibling cells, and this process can determine differentiation versus self-renewal (Adams et al., 2016). Maintenance of mitochondria in some cells was linked to anabolism, PI3K/mTOR activation, glycolysis, and inhibited autophagy while mitochondrial clearance in others was associated with catabolism, FoxO1 transcription factor activity, and self-renewal. Thus, mitochondrial maintenance can drive differentiation over self-renewal, illustrating how these organelles lie at the center of cell fate decisions.
In addition to mitochondrial stasis versus clearance, memory T cells also have distinct mitochondrial morphology from effector T cells. Effector T cells have more “fissed” mitochondria whereas memory T cells have more “fused” mitochondrial networks with tight cristae suggesting a requirement for mitochondrial fusion in memory T cell metabolism and homeostasis. Consistent with this observation, antigen-specific T cells lacking the inner mitochondrial membrane fusion protein Opa1 fail to generate memory T cells after bacterial infection and have impaired survival in vitro (Buck et al., 2016). The activation, proliferation, and function of Opa1-deficient effector T cells, however, remain intact. Opa1^−/− T cells have augmented rates of glycolysis and possess mitochondria with diminished OXPHOS efficiency and malformed cristae compared to controls. It was shown that in quiescent T cells, such as naive and memory T cells, mitochondrial fusion ensured tight cristae associations that allowed for efficient electron transport chain (ETC) function (Figure 2). Tight cristae result in dense packing of ETC complexes, which have been found to associate in specialized configurations termed respiratory supercomplexes or respirasomes. Supercomplexes facilitate efficient transfer of electrons and minimize proton leak during ATP production (Cogliati et al., 2013). CD8 T cells express high levels of methylation-controlled J protein (MCJ), a member of the DnaJ family of chaperones (Champagne et al., 2016). MCJ localizes to the inner membrane of mitochondria and associates with complex I of the ETC. By doing so, it negatively regulates the assembly of complex I into supercomplexes. MCJ deficiency was found to enhance naive and activated CD8 T cell OXPHOS and a unique attribute was its role in the secretion, but not the translation, of effector cytokines. Increased respiration efficiency improved the survival of MCJ^−/− effector T cells, which also induced superior protective immunity against viral challenge.
Repurposing the knowledge gained from such studies, boosting oxidative capacity and efficiency through enforcement of mitochondrial fusion or dampening glycolysis with 2-DG, extends the survival and antitumor function of CD8 T cells in models of adoptive cell therapy (Buck et al., 2016, Sukumar et al., 2013). Although aerobic glycolysis initiates and sustains effector functions of activated T cells, augmenting metabolic pathways that support long-lived memory T cells improves T cell responses against tumors, demonstrating a need to strike a balance between these processes, seemingly trading off heightened activation and effector functions of TILs with their sustained functionality and increased survival in the tumor microenvironment. Indeed, modification of the signaling domains within chimeric antigen receptor T cells, used in an alternative form of adoptive cell therapy, with 4-1BB augments mitochondrial biogenesis and oxidative metabolism, enhancing their persistence (Kawalekar et al., 2016). As our capability to selectively reprogram T cell metabolism and reinvigorate tumor-specific T cells improves, there is much promise to provide greater therapeutic benefits to more patients, especially to those with previously incurable cancers.

Another unique structural feature of the gut are the Peyer’s patches, aggregates of gut-associated lymphoid tissue (Reboldi and Cyster, 2016). Found in the small intestine, they represent a specialized lymphoid compartment continuously exposed to food- and microbiome-derived antigens. Due to this exposure, Peyer’s patches are rich in germinal centers (GCs), which are comprised of B, T, stromal, and follicular DCs. B cells segregate into zones where they undergo cycles of proliferation and differentiation and compete for signals directing class switch recombination (CSR) and survival from T follicular helper (Tfh) cells, allowing further maturation of the antibody repertoire.
Although the literature on B and Tfh cell metabolism is still developing, it has been shown that B cell activation induced by either α-IgM ligation or lipopolysaccharide (LPS) increases Glut1 expression and glucose uptake downstream of PI3K and mTOR signaling (Caro-Maldonado et al., 2014, Doughty et al., 2006, Jellusova and Rickert, 2016, Lee et al., 2013, Woodland et al., 2008). Glycolysis and OXPHOS are augmented as well as mitochondrial mass (Caro-Maldonado et al., 2014, Doughty et al., 2006, Dufort et al., 2007). Increased glucose acquisition also fuels de novo lipogenesis necessary for B cell proliferation and growth of intracellular membranes. Inhibition of the fatty acid synthesis (FAS) enzyme ATP-citrate lyase in splenic B cells results in reduced expansion and expression of plasma cell differentiation markers (Dufort et al., 2014). Although apoptosis inducing factor (AIF) is required for T cell survival via ETC complex I function and respiration, AIF deficiency in B cells has no impact on their development or survival because of their reliance on glucose metabolism (Milasta et al., 2016). B cells cultured in galactose fail to expand unlike T cells, which can activate and proliferate in the presence of either galactose or glucose (Chang et al., 2013, Milasta et al., 2016). Glycogen synthase 3, which promotes the quiescence and survival of circulating naive B cells, tempers increases in glycolytic metabolism downstream of CD40 costimulatory receptor signaling and sustains the survival of B cells subjected to glucose restriction (Jellusova et al., 2017). On the other hand, the transition to durable humoral immunity by long-lived plasma cells (LLPCs) was shown to be dependent on mitochondrial pyruvate import and metabolism (Figure 3). Glucose supports antibody glycosylation, but LLPCs acquire more glucose than their short-lived counterparts and their long-term survival is dependent on their ability to siphon glucose-derived pyruvate into the mitochondria during times of metabolic stress (Lam et al., 2016).
It is interesting to speculate that with the constant proliferation of GC B cells in the gut and the importance of glucose and glycolysis in activated plasma cells, access to glucose would become limiting for other cells that occupy this microniche. A few studies suggest that Tfh cells have evolved to be uniquely suited to survive under these constraints. It has been shown that Tfh cells have less mTORC1 activation and reduced glycolysis compared to Th1 cells (Ray et al., 2015). In part, this may be due to expression of their lineage defining transcription factor Bcl6, which can suppress glycolysis potentiated by c-Myc and HIF-1α (Johnston et al., 2009, Nurieva et al., 2009, Oestreich et al., 2014). Consistent with this, overexpression of Bcl6 reduces glycolysis in T cells, and inhibition of mTOR using shRNA favors Tfh cell development over Th1 cells in vivo after viral infection (Ray et al., 2015) (Figure 3). However, a more recent study using mice with conditional deletions of mTORC1 and mTORC2 via OX40 and CD4 cre recombinase observed a requirement of mTOR signaling in Tfh cell development and GC formation within Peyer’s patches (Zeng et al., 2016). The former applied retroviral mTOR shRNA, which requires T cells be fully activated prior to knockdown, while this more recent report used mice where mTOR was excised during T cell development or at the moment of T cell activation, which may explain the disparity between the studies.
In addition to possibly limiting quantities of glucose substrate within GCs, these microniches contain areas of hypoxia, resulting in HIF-1α activation (Abbott et al., 2016, Cho et al., 2016). B cells placed under hypoxic conditions had reduced activation-induced deaminase expression and subsequently underwent less CSR to the pro-inflammatory IgG2c isotype when cultured in conditions that promote IgG production (Cho et al., 2016). In contrast, B cells cultured in IgA-promoting conditions during hypoxia were unaffected, yielding comparable levels of IgA to cells kept at normoxia and highlighting how lymphocyte function may be fine-tuned to varying oxygen tension in tissues (Figure 3). B cells isolated from mice with constitutive activation of HIF-1α by deletion of its suppressor VHL had defects in IgG2c production, which was attributed to diminished mTORC1 activation. B cells from Raptor-deficient heterozygotes also yielded fewer IgG antibodies (Cho et al., 2016). A separate study found that the mTOR inhibitor rapamycin dampens CSR, yielding the formation of lower affinity, more cross-reactive B cell antibodies, which offered broad protection against heterosubtypic flu infection (Keating et al., 2013). Both mTORC1 and HIF-1α promote aerobic glycolysis (O’Neill et al., 2016). However, the metabolic activities of the cells cultured under different isotype conditions while under hypoxia were not explored. A separate study examining mitochondrial function, and specifically mitochondrial ROS, found that B cells with augmented mitochondrial mass, respiration, and ROS stratified cells that underwent CSR marked by IgG1 expression apart from CD138⁺ plasma cells (Jang et al., 2015). The differences seen in mitochondrial function between the B cell populations were in part due to differential regulation of heme synthesis by mitochondrial ROS, however, how the mitochondria affects CSR to other isotypes was not assessed. Cytokines initiate CSR to distinct isotypes and signals derived from these growth factors might be responsible for the differences in metabolic signaling and suggest varying requirements to initiate metabolic programs and CSR in B cells. Secretion of IgA predominates the gut and is critical to maintaining barrier protection and bacterial homeostasis (Kurashima and Kiyono, 2017). The apparent stability of CSR to the IgA isotype under hypoxia and impaired pro-inflammatory IgG2c subtype might have evolved to ensure tolerance with the microbiome, while concurrently providing a stringent method of selection of antibodies produced during inflammatory responses to pathogen-derived antigens.

The metabolic relationship between commensals and immune cells in the gut is further illustrated by the finding that SCFAs derived from the fermentation of dietary fiber by gut microbiota promote B cell metabolism and boost antibody responses in both mouse and human B cells (Kim et al., 2016). Supplementation with dietary fiber or the SCFAs acetate, propionate, and butyrate increases intestinal IgA production, as well as systemic IgG during infection (Figure 3). Culturing B cells with SCFAs was shown to raise acetyl-CoA levels and increase mitochondrial mass, lipid content, and FAS leading to increased plasma cell differentiation and metabolic activity (Kim et al., 2016). Part of this phenotype could be attributed to histone deacetylase (HDAC) inhibition, an established effect of SCFA supplementation.
In addition to their effects on B cells, SCFAs can promote the development and function of colonic Treg cells via induction of Foxp3 in a HDAC-dependent manner (Arpaia et al., 2013, Furusawa et al., 2013, Smith et al., 2013) (Figure 3). Treg cells are critical to maintaining commensal tolerance by the immune system through suppression of aberrant T cell responses. Unlike other activated T helper subsets, Treg cells have been described to primarily rely on OXPHOS driven by FAO (Newton et al., 2016). However, signals downstream of TLR ligation can augment glycolysis and proliferation of Treg cells and reduce their ability to suppress T cell responses (Gerriets et al., 2016). Retroviral enforced expression of Foxp3 promotes OXPHOS and dampens glucose uptake and glycolysis, whereas Treg cells transduced with Glut1 decreased Foxp3 expression after adoptive transfer in vivo and fail to suppress T cell-mediated colitis in a model of inflammatory bowel disease. These findings suggest that during inflammation and microbial infection, Treg cells may temporarily lose their regulatory function to give way to robust T cell responses and participate as more conventional effector helper T cells. Increases in NaCl either from supplementation in vitro or diet in vivo inhibit the suppressive function of human Treg cells via serum/glucocorticoid-regulated kinase 1 (SGK1), which integrates signals from PI3K and mTORC2 to regulate sodium-controlled signal transduction (Hernandez et al., 2015). However, a study of human Treg cells found that the glycolytic enzyme enolase-1 was required for their suppressive activity through its control of Foxp3-E2 splice variants (De Rosa et al., 2015). Depending on environmental cues and metabolites, it appears that Treg cell metabolism can be modulated, affecting their function.
As discussed, increases in SCFAs either from diet, infection, or exogenous treatment impinge on metabolic processes including HDAC activation (Rooks and Garrett, 2016). A recent study suggests that activation of the HDAC sirtuin 1 (SIRT1) negatively impacts Th9 cell differentiation (Wang et al., 2016b). The exposure of CD4 T cells to distinct cytokine cocktails differentiates them into separate helper lineages. However, perturbing metabolism also modulates CD4 T cell fate. A yin and yang relationship between Th17 and Treg cell differentiation has been established. Th17 cells are particularly glycolytic and depend on engagement of this pathway downstream of mTOR and HIF-1α activation. Dampening glycolysis through deletion of HIF-1α or with the inhibitor 2-DG in T cells impairs Th17 development and instead promotes Treg cells, even under Th17-inducing culture conditions (Dang et al., 2011, Shi et al., 2011). Suppression of mTOR with rapamycin or genetic ablation also augments production of Treg cells (Delgoffe et al., 2009, Kopf et al., 2007), and pharmacological inhibition of de novo fatty acid synthesis prevents Th17 differentiation and instead enforces a Treg cell phenotype (Berod et al., 2014).
Although the metabolic characteristics of other CD4 T cell subsets have been compared (Michalek et al., 2011), little was known about Th9 cell metabolism. Th9 cells are characterized by their ability to produce IL-9 and can be generated from naive cells in culture using the cytokines transforming growth factor β (TGF-β) and IL-4. They are implicated in autoimmunity, melanoma, and worm infections (Kaplan et al., 2015). Wang et al. (2016b)) found that Th9 cells are highly glycolytic, in part from their active suppression of SIRT1 expression via the kinase TAK1. SIRT1 was previously shown to negatively control HIF-1α as well as mTOR (Lim et al., 2010, Liu et al., 2014). In line with this, Th9 cell development was augmented in SIRT1-deficient T cells whereas retroviral enforced expression of SIRT1 or dampening of aerobic glycolysis by chemical or genetic means impaired Th9 cell differentiation (Wang et al., 2016b). Th9, Th17, and Treg cells all share the cytokine TGF-β for their development but then depend on additional cytokine signals for their eventual fates. Given their divergent metabolic phenotypes, as well as HDAC requirements, it would be interesting to explore further whether variances in intracellular levels of SCFA metabolites, for example, might couple with environment signals to influence their eventual metabolic and developmental pathway.
Apart from its effect on CD4 T cells, the SCFA acetate also has been shown to affect secondary recall responses from CD8 memory T cells (Balmer et al., 2016). Germ-free mice reconstituted with commensal microbes, or oral or systemic infection with bacterial species, elevated serum acetate concentrations. Memory T cells generated in vitro or in vivo cultured with acetate levels observed during these infections secreted more IFN-γ and augmented glycolysis after restimulation. Acetate can be quickly converted into acetyl-CoA, which can condense with oxaloacetate into citrate in the mitochondria to fuel the TCA cycle and OXPHOS, can be used as a substrate for FAS, or participate in post translational modification (PTM) of proteins including histones (Pearce et al., 2013). Balmer et al. (2016)) mechanistically tied their results to lysine acetylation of the glycolytic enzyme glyceraldehyde 3-phosphate dehydrogenase (GAPDH). GAPDH activity has been shown to regulate T cell production of IFN-γ (Chang et al., 2013, Gubser et al., 2013). Although the study demonstrated that the enzymatic activity of GAPDH was altered by acetylation of K217, whether this PTM was critical to acetate-dependent increases in IFN-γ protein was not explored. In a separate report, CD4 T cells deficient in LDHA expression had defects in IFN-γ production, which stemmed from widespread lack of acetylation of the Ifng locus (Peng et al., 2016). LDHA is the critical enzyme that defines aerobic glycolysis, converting pyruvate to lactate. Culturing cells in galactose impairs aerobic glycolysis, as galactose enters glycolysis at a significantly lower rate than glucose via the Leloir pathway (Bustamante and Pedersen, 1977), a result confirmed by tracing galactose metabolism in T cells (Chang et al., 2013). Reducing GADPH engagement from glycolysis in this fashion permits moonlighting function by this abundantly expressed protein. It was shown that GAPDH binds to the 3′UTR of AU-rich containing cytokine mRNAs, preventing their efficient translation (Chang et al., 2013). Peng et al. (2016)) argue against GAPDH posttranscriptional control of T cell function during aerobic glycolysis deficiency via LDHA deletion because modification of the 3′UTR of Ifng did not rescue defects in cytokine production in their system. However, as Peng et al. (2016)) demonstrated, LDHA-deficient cells have defects in Ifng mRNA production, whereas cells forced to respire in galactose remain transcriptionally competent for Ifng as those cultured in glucose. Supplementation with the SCFA acetate rescued their epigenetic defect and cytokine production. These studies show that aerobic glycolysis regulates both transcriptional and translational functions in T cells.
While products generated from the microbiome can modulate the metabolism of immune cells and shift the balance between tolerance and inflammation, there are hints that immune-driven signals central to gut homeostasis may also mediate their effects through metabolic modulation. One such example is the pleotropic anti-inflammatory cytokine IL-10. Most hematopoietic cells produce and sense IL-10 and its importance for maintaining tolerance with the intestinal microbiota is clearly evident from observations that IL-10- or IL-10R-deficient mice develop spontaneous colitis (Kühn et al., 1993, Spencer et al., 1998). IL-10R deficiency in macrophages is also sufficient to recapitulate onset of severe colitis in mice (Shouval et al., 2014, Zigmond et al., 2014). Further, mice with a myeloid cell-specific deficiency in STAT3, which is activated downstream of the IL-10R by JAK1, develop chronic enterocolitis as they age (Takeda et al., 1999). In experiments that shed light on the importance of aerobic glycolysis engagement in DC activation, it was found that treatment of DCs with recombinant IL-10 blocked increases in their glycolytic rate after LPS stimulation (Krawczyk et al., 2010). Cells subjected to IL-10R blockade further upregulated glycolysis after activation compared to controls. It is tempting to speculate that one of the ways IL-10 might be anti-inflammatory is through inhibition of metabolic reprogramming to aerobic glycolysis during innate immune cell activation (Figure 3). Coincidentally, STAT3 was found to localize to mitochondria and interact with ETC complexes, which helped maintain efficient OXPHOS in the heart (Wegrzyn et al., 2009). Whether traditional cell surface cytokine-receptor signaling could modulate levels of mitochondrial STAT3 was not explored. Of interest, CD8 T cells with a conditional deletion of the IL-10R fail to form memory T cells (Laidlaw et al., 2015), which depend on FAO-driven OXPHOS for their generation after infection (Cui et al., 2015, Pearce et al., 2009, van der Windt et al., 2012).
The gut is one example of a tissue that presents distinct metabolic challenges for immune cells, which affect their steady state and protective versus inflammatory responses. Other examples, such as skin, provide the potential for commensal organisms to metabolically affect immune cell function, a topic reviewed elsewhere (Hand et al., 2016). The intestinal tract is constantly subjected to fluctuations in diet and sometimes the intake of invasive pathogens can also deprive metabolic substrates from immune cells. The bacterium Salmonella typhimurium produces a putative type II asparaginase that depletes available asparagine needed for metabolic reprogramming of activated T cells via c-Myc and mTOR (Torres et al., 2016). The use of asparaginase for acute lymphoblastic leukemia treatment highlights the potential for the depletion of extracellular arginine to significantly affect cellular function (DeBerardinis and Chandel, 2016). Lack of dietary vitamin B1 decreases the number of naive B cells in Peyer’s patches due to their dependence on this TCA cycle cofactor, while leaving IgA⁺ plasma cells intact in the lamina propria (Kunisawa et al., 2015). Although ILC metabolism has only recently been explored (Monticelli et al., 2016, Wilhelm et al., 2016), it was found that in settings of vitamin A deficiency, type 2 ILCs sustain their function via increased acquisition and utilization of fatty acids for FAO (Wilhelm et al., 2016) (Figure 3). The internal balance between polyunsaturated fats and saturated fatty acids can also determine the pathogenicity of Th17 cells; cells that help maintain mucosal barrier immunity and contribute to pathogen clearance (Wang et al., 2015). Long chain fatty acids (LCFA) promote Th1 and Th17 cell polarization and mice fed with LCFA have exacerbated T cell-mediated autoimmune responses, whereas mice fed with SCFA are protected (Haghikia et al., 2015). If the internal lipidome of Th17 cells can alter their functional state from protective to inflammatory as does their access to LCFA versus SCFA in the small intestine, it raises the question of how other tissue systems with a rich diversity of fat deposits and cells, such as the adipose tissue, modulate the metabolism and function of resident immune cells. Indeed, a recent study has found that the survival and function of resident memory T cells depends on exogenous lipid uptake and metabolism mediated by fatty acid binding proteins 4 and 5 (Pan et al., 2017).