Fatty acid metabolism in adaptive immunity

Abstract

Fatty acids not only are a key component of cellular membrane structure, but also have diverse functions in biological processes. Recent years have seen great advances in understanding of how fatty acid metabolism contributes to adaptive immune response. Here, we review 3 key processes, fatty acid biosynthesis, fatty acid oxidation and fatty acid uptake, and how they direct T and B cell functions during immune challenges. Then we will focus on the relationship between microbiota derived fatty acids, short-chain fatty acids, and adaptive immunity. Along the way, we will also discuss the outstanding controversies and challenges in the field.

Introduction

Metabolic reprogramming plays an essential role in the development and activation of immune cells in response to pathogen challenges or environmental changes. Fatty acid (FA) acts as a major player in the metabolic adaption of lymphocytes. Here we review the recent advances on the regulation of FA in adaptive immunity, especially its function on cell activation and differentiation. First, FAs are integral part of cellular membranes, which envelop all cells and the associated intracellular organelles. Second, FAs function as a form of energy storage. Energy demand is essential for cellular function in both quiescent and activated states. Although glutamine and glucose are two major sources of energy, FAs can fuel cellular metabolism through fatty acid oxidation (FAO). Third, they are also involved in a wide range of signaling pathways, acting as signaling mediators or substrates for protein modification. For a long time, studies on the link between FAs and immunity were mainly focused on dietary polyunsaturated fatty acid (PUFA) and its derivatives, which have been covered by an excellent recent review [1]. Here, we summarize the latest advances on how FA synthesis, oxidation, and uptake orchestrate T cell activation, differentiation, and memory response, as well as humoral immune response. We will also discuss how different non-PUFA FA species including short- and long-chain FAs, modulate the adaptive immune response. The sheer diversity and versatility of FAs present a significant challenge for us to untangle their immunological functions. We are just starting to appreciate their importance in immune system. Understandably, the physiological effects of certain FA on lymphocytes remain contentious and are under intensive investigation. We will discuss some of the controversies, provide potential explanations, and suggest future investigations.

Fatty acid synthesis – biosynthesis of different fatty acid species could differentially control lymphocyte activation in a lineage specific manner

FA synthesis (FAsyn) takes place in the cytosol, where acetyl-CoA is regarded as a direct precursor of FAsyn [2]. Two molecules of acetyl‐CoA are converted into malonyl‐CoA by acetyl‐CoA carboxylase (ACC), which is the first committed step for de novo FAsyn. ACC, the rate-limiting enzyme for FA synthesis, has two isoforms in mammal cells: ACC1 (also known as ACCα, encoded by Acaca) and ACC2 (also known as ACCβ, encoded by Acacb). They are located in the cytoplasm and mitochondrial outer membrane, respectively [3]. Fatty acid synthase (FASN, encoded by Fasn) then condenses acetyl‐CoA and malonyl‐CoA into palmitic or octadecanoic acid in the presence of NADPH [4]. These de novo‐synthesized saturated FAs can be converted to monounsaturated FAs (MUFAs) by stearoyl-CoA desaturase (SCD, encoded by Scd) [5]. Acetyl-CoA can also be used as a precursor for the synthesis of steroids and isoprenoids, which are not a subject of this review. FAs can be further incorporated into more complex lipids, such as various phospholipids, diacylglycerides (DAGs), and triacylglycerides (TAGs) [6]. Because glucose metabolism can provide precursors (in the form of pyruvate) and reducing agent (in the form of NADPH through pentose phosphate pathway) for FAsyn, and the glycerol backbone for DAG and TAG formation (Figure 1), it is one of the major sources to fuel de novo FAsyn during lymphocyte activation [7]. The whole FAsyn program is under the control of the master transcription factor sterol regulatory element binding protein (SREBP) [8].

Figure 1. Overview of fatty acid synthesis in immune cells.

Upon immune signal stimulation, T and B cells have upregulated fatty acid synthesis (FAsyn) activity in the cytosol. Acetyl-CoA acts as direct precursor for FAsyn, which can be split from cytoplasmic citrate by ATP-citrate lyase (ACLY). Two molecules of acetyl-CoA are converted into malonyl-CoA by acetyl-CoA carboxylase (ACC), which is the first committed step for de novo FAsyn. Fatty acid synthase (FASN) then condenses acetyl-CoA and malonyl-CoA into saturated fatty acids (FAs), like palmitic acid and stearic acid. These de novo-synthesized saturated FAs can be converted to monounsaturated FAs (MUFAs) by stearoyl-CoA desaturase (SCD). Elongation of very long-chain fatty acid (ELOVL) family enzymes are responsible for catalyzing formation of very long-chain FAs from saturated FAs or MUFAs. Long chain FAs are further incorporated into complex lipids including phospholipid (PL), diacylglyceride (DAG), and triacylglyceride (TAG). Glucose metabolism can generate NADPH and glycerol-3-phosphate. NADPH is an essential reducing agent required for FASN mediated reactions. Glycerol-3-phosphate is incorporated into DAG and TAG as scaffold molecule. Abbreviations: tricarboxylic acid cycle (TCA cycle). Furthermore, glycolysis end product pyruvate can be transported into mitochondrial to generate citrate. Citrate can also be generated by glutamine anaplerosis. Thus, glucose metabolism is key for material provision of FAsyn.

Fatty acid synthesis in T lymphocytes

B and T lymphocytes are the two pillars of the adaptive immune system. Both require abundant nutrients to undergo expansion, proliferation, and differentiation upon immune challenge. Activation of both T and B cells is associated with increased FAsyn [9–13]. Over the past few years, several studies have investigated the FAsyn regulation in T cells [14, 15]. Using mass-spectrometry-based metabolomics approach, Wang et al. revealed that T cells accumulate not only metabolites involved in the anabolic pathways of nucleotides and amino acids, but also various metabolites of the FAsyn pathway after antigenic stimulation, which are dependent on the proto-oncogene MYC transcription factor [10]. Since MYC is one of the downstream targets of mechanistic target of rapamycin complex 1 (mTORC1) [16, 17], it implies that mTORC1 may regulate FAsyn in T cells [10]. In CD8⁺ T cells, TCR signal induces an mTOR-dependent FAsyn program. Treatment of rapamycin, a known inhibitor of mTORC1, or genetic deficiency of regulatory‐associated protein of mTOR (RAPTOR), strongly inhibits FAsyn activity in CD8⁺ T cells [18]. mTORC1 promotes T cell FAsyn partly by activating SREBP [19–21]. T cell activation increases SREBP expression in an mTORC1-dependent manner, while loss of SREBP suppresses FA accumulation and reduces CD8⁺ T cell clonal expansion during viral infection [22]. Pharmacological inhibition or genetic deletion of FAsyn-specific enzymes revealed a more complex picture. T cell-specific deletion of Acaca reduces CD4⁺ and CD8⁺ T cells in secondary lymphoid organs, indicating a critical role of FAsyn in T cell homeostasis. ACC1 deficiency suppresses antigen-specific CD8⁺ T cell accumulation, which is associated with impaired cell survival [23].

De novo FAsyn also controls CD4⁺ T cell differentiation fate. Berod et al. showed that CD4⁺ T cells engage a metabolic flux from glucose towards FAsyn under T helper 17 (Th17) differentiation conditions [9]. ACC1 blockade, either by ACC-specific inhibitor or T cell-specific deletion of Acaca, suppresses the differentiation of Th17 cells and attenuates Th17 cell-mediated autoimmune disease, but selectively promotes the differentiation of regulatory T cells (Tregs) [9]. ACC1 deficiency also suppresses CD4⁺ T cell proliferation and T helper 1 (Th1) and T helper 2 (Th2) differentiation [9], suggesting that ACC1-mediated FAsyn may direct the reciprocal differentiation of effector T (Teff) cells and Tregs. It is possible that ACC1 controls the toggle between Teff and Treg partly by suppressing AMP-activated protein kinase (AMPK) [24]. AMPK inhibits mTORC1 signaling, which promotes Teff and suppresses induced Treg generation [25]. However, a recent study demonstrated that Treg-specific deficiency of FASN impairs the functional maturation of Treg specifically in tumor microenvironment, but does not affect Treg-mediated immune homeostasis or autoimmune pathogenesis, indicating a potential differential impact between ACC1-dependent and FASN-dependent FAsyn on Treg differentiation and function. Mechanistically, FASN mediated palmitate synthesis is required for TCR-induced Treg functional maturation and exogenous palmitate can rescue the reduced suppressive activity of FASN-deficient Tregs [26]. Furthermore, pharmacological inhibition of SCD impedes follicular helper T (Tfh) cell differentiation by promoting ER stress and Tfh cell apoptosis in a model of influenza intraperitoneal immunization [27]. In contrast, abnormally increased FA biosynthesis or accumulation appears to be detrimental to Tregs. For instance, increased TAG accumulation, either through increased TAG biosynthesis, or through reduced TAG lipolysis due to lysosomal acid lipase deficiency or inhibition, leads to impaired Treg suppression of Th2 responses through undefined mechanisms [28]. These studies indicate a general requirement of FAsyn for CD4⁺ Teff cell (Teff, including Th1, Th2, Th17, and Tfh) generation, but a potentially context dependent requirement of FAsyn for Tregs. Interestingly, different Teff lineages may have their unique FA metabolic program. An elegant in vivo CRISPR-Cas9 screening assay recently identified key enzymes in the cytidine diphosphate-ethanolamine pathway for de novo synthesis of phosphatidylethanolamine (PE, a type of phospholipid) as a selective post-transcriptional regulator to control Tfh differentiation, but not other Teff cells or Tregs. Mechanistically, PE promotes Tfh lineage by protecting CXCR5, a key chemokine receptor for Tfh trafficking, from internalization and degradation [29]. PE biogenesis is closely linked to the core FAsyn program. Defects in PE biogenesis pathway can lead to activation of SREBP1, and subsequent enhanced FAsyn, which eventually leads to increased accumulation of TAG [30]. Finally, de novo FAsyn drives Teff cell infiltration into joints and promotes rheumatoid arthritis (RA), partly by enhancing the locomotion program mediated by TKS5, a podosome scaffolding protein. Inhibition of FAsyn curbs the hypermobility of T cells from RA patients and protects synovial tissues from inflammation [31]. It remains unclear how a seemingly general increase of FAsyn in RA T cells appears to affect TKS5 specifically. Conversely, myristoylation, a posttranslational modification where myristate, a 14-carbon saturated fatty acid, is attached to N-terminal glycine of protein by the activity of N-myristoyltransferase (NMT), can suppress Teff cell differentiation and prevent T cell mediated arthritis. This is achieved by activating AMP-activated protein kinase (AMPK), whose activation requires myristoylation. AMPK activation suppresses mTORC1 and subsequently inhibits pro-inflammatory Th1 and Th17 differentiation. T cells from RA patients, but not those with systemic lupus erythematosus, psoriatic arthritis, or granulomatosis with polyangiitis, have reduced expression of NMT and subsequently reduced AMPK activity and overactivation of mTORC1, highlighting a disease specific dysregulation of myristoylation [32].

FAsyn also modulates memory T cell generation. Pharmacological inhibition of ACC1 or genetic deletion of Acaca elevates mitochondrial activity and FAO, which enables the long-term persistence of memory CD4⁺ T cells in vivo, indicating that ACC1 negatively controls memory CD4⁺ T cell generation [33]. However, another study showed that the generation of memory CD8⁺ T cells is dependent on FASN mediated de novo FAsyn using a FASN specific inhibitor [34]. It is unclear how ACC1 and FASN may differentially control memory T cell formation. Future investigations, especially ones using genetic models targeting FASN, are warranted. Finally, memory CD8⁺ T cells appear to engage a futile cycle of active synthesis of FA containing lipids such as triglyceride and diglyceride, and T-cell intrinsic lipolysis of these lipids to liberate FAs as energy fuel, which is mediated by the lysosomal acid lipase [34]. It is unclear why memory T cells may use such a seemingly wasteful strategy. Later studies also showed that T cells can take up extrinsic FAs to fuel their metabolism (see below). More studies are needed to explore how T cell intrinsic FAsyn, lipolysis and FA uptake coordinate to modulate T cell function. Taken together, these findings have revealed that FAsyn is critically required for Teff cell activation and differentiation, and it modulates Treg differentiation and function in a context-dependent manner.

Fatty acid synthesis in B lymphocytes

In contrast to T cells, pharmacological inhibition of FASN, or genetic deletion of SCD does not seem to significantly affect major peripheral B2 B cell subsets, except that FASN inhibition depletes B1a B cells, an innate-like B cell lineage mainly located in the peritoneal cavity [13, 35]. These studies suggest that overall B cells may be less dependent on de novo FAsyn for homeostasis than T cells. More researches are required to determine whether FAsyn contributes to B cell development and homeostasis. Like T cells, the metabolic programs for oxidative phosphorylation (OXPHOS), tricarboxylic acid (TCA) cycle, and nucleotide biosynthesis are all strongly augmented in activated B cells [36]. Transcriptome analyses indicate an enrichment of FAsyn-related gene signature in activated primary B cells and B-cell line, including Acaca, Fasn, Scd2, and Elovl6, all enzymes involved in FAsyn [11, 13, 37]. Meanwhile, activated B cells expand their pool of acetyl-CoA, which is likely dependent on the activity of ATP-citrate lyase (ACLY) [12]. Acetyl-CoA for FAsyn is partially generated by ACLY, which catalyzes the cleavage of citrate into oxaloacetate and acetyl-CoA in the cytosol. Cytoplasmic citrate can be transported from mitochondrial matrix, where glucose derived pyruvate feeds into TCA cycle and can be metabolized to the intermediate citrate. Together with acyl-CoA synthetase short-chain family member 2 (ACSS2), which converts acetate to acetyl-CoA for FAsyn, they maintain the cellular acetyl-CoA level [38]. We have recently investigated the functions of SCD in humoral immunity. We found that SCD activity and SCD-generated MUFAs are required for B cell proliferation and class switch reaction in vitro. Pharmacological suppression of SCD or genetic deletion of Scd severely dampens B cell proliferation, class switch reaction and induces excessive cell autophagy and endoplasmic reticulum (ER) stress, all of which can be fully rescued by supplementation of exogenous MUFAs, like oleic acid (OA) or palmitoleic acid [13]. Interestingly, systemic inhibition of SCD activity, but not B cell-specific deletion of Scd, suppresses humoral immune response upon immunization or influenza infection. These data demonstrate that systemic, but not B cell intrinsic, biosynthesis of MUFAs is indispensable for the host humoral immune response. However, how B cells acquire extrinsic MUFA, the source of extrinsic MUFA, and the functions of MUFA in memory B cell response await further investigations.

Fatty acid oxidation

FAO is the primary source of ATP when cells are in a low glucose state. More than twice as many ATP molecules can be generated from FAO as from glucose or amino acid catabolism when they are oxidized in the mitochondria [39]. Alternatively, FA can also be oxidized in peroxisome to generate chain-shortened acyl-CoA and acetyl-CoA without ATP production [40, 41]. To generate energy, FA must be transported from the cytosol to mitochondria, to be metabolized by sequential removal of 2-carbon fragments, and finally, be oxidized to generate acetyl-CoA to feed into the TCA cycle. The transport of long-chain FA (LCFA) to the mitochondrial matrix is regulated by the carnitine shuttle system composed of carnitine O-palmitoyltransferase 1 and 2 (CPT1 and CPT2) and carnitine-acylcarnitine translocase (CACT). Short- and medium-chain fatty acids (SCFA and MCFA) can freely shuttle between mitochondrial membranes. Since CPT1 catalyzes the rate-limiting step in the oxidation of LCFAs, it is considered to be the main target to manipulate FAO [42]. Upon antigenic activation, T cells undergo metabolic reprogramming to augment aerobic glycolysis, a central metabolic pathway that supports biosynthetic and bioenergetic activities, in which glucose-derived pyruvate is converted into lactate in the presence of oxygen. Glycolysis supports Teff cell growth, differentiation, and cytokines production [18, 43–46]. Catabolic metabolism, like FAO, is proposed to support memory T cells generation [47–49]. Pearce et al. showed that activated CD8⁺ T cells lacking TNF receptor associated factor 6 (TRAF6) display defects in several catabolic pathways of metabolism, including mitochondrial FAO, in response to growth factor withdrawal. Mice with T-cell-specific deletion of TRAF6 mount robust effector CD8⁺ T cell response, but have a profound defect in their ability to generate anti-bacterial memory T cells [47]. This defect in memory T cell generation can be partially rectified by treatment with the AMPK activator metformin or mTOR inhibitor rapamycin, which enhances FAO [50, 51]. A moderate inhibition of mTOR pathway alone is also sufficient to boost anti-viral or anti-tumor memory T cell response [47, 52–54]. Mechanistically, memory CD8⁺ T cells have greater mitochondrial mass and mitochondrial spare respiratory capacity (SRC) compared to naïve and effector CD8⁺ T cells. FAO blockade with etomoxir, a CPT1A inhibitor, or shRNA-mediated knockdown of CPT1A, suppresses SRC, and selectively impairs the survival of memory CD8⁺ T cells, but not effector CD8⁺ T cells in vitro [55, 56]. Based on these findings, FAO is considered to be critical for memory CD8⁺ T cell generation and function.

FAO also regulates CD4⁺ T cell lineage differentiation. Teff cell subsets and Treg exhibit distinct metabolic profiles. Th1, Th2, and Th17 cells express higher level of the glucose transporter 1 (Glut1) and are more glycolytic. In contrast, Tregs have lower Glut1 expression and high FAO. Mechanistically, AMPK-mTOR signaling axis reciprocally regulates lipid oxidation and glucose metabolism, which controls Teff and Treg differentiation [57–59]. Consistent with this idea, OA, a type of MUFA enriched in adipose tissue, promotes the suppressive function of Tregs via amplifying FAO-driven OXPHOS metabolism. Reduced OA level is partially responsible for the impaired Treg functions in patients with multiple sclerosis [60]. Therefore, the current model posits that memory T cells and Tregs, unlike Teff cells, primarily rely on FAO for their homeostasis and function. However, there is an ongoing debate on the importance of LCFA-FAO for memory T cells and Tregs because genetic deletion of Cpt1a specifically in T cells does not affect the generation and function of memory T cells and Tregs in vivo [61]. Furthermore, the dose of etomoxir used in early studies on memory T cells can have off-target effects [61]. Finally, Treg suppressive activity is not always correlated with FAO because inhibition of FA binding protein 5 (FABP5, see below for more details) leads to reduced FAO, but enhanced Treg suppressive function [62]. It is conceivable that oxidation of SCFA and MCFA may compensate for the loss of LCFA-FAO. Alternatively, CPT1A deficiency may be functionally compensated by other CPT1 isoforms. Moreover, metabolic plasticity could allow memory T cell differentiation even under constitutive glycolytic metabolism as long as sufficient ATP levels are maintained [63]. Clearly, more researches are needed to resolve these controversies.

How FAO contributes to B cell function remains poorly understood. We demonstrated that MUFA, but not saturated FA, promotes B cell activation in vitro partly through FAO. We showed that etomoxir, at a dose selective to inhibit CPT1A [61, 64], can suppress B cell activation and counteract exogenous MUFA enhanced FAO [13], suggesting that MUFA FAO contributes to B cell activation. A reversible CPT1 inhibitor, ST1326, suppresses the proliferation of human normal B cell and chronic lymphocytic leukemia (CLL) cell upon stimulation [65]. Diffuse large B-cell lymphoma (DLBCL) is the most prevalent B-cell non-Hodgkin lymphoma in adults worldwide [66], which can be generated from germinal center (GC) B cells or activated B cells [67]. Previous studies showed that B-cell receptor (BCR)-dependent DLBCLs are glycolytic [68], while another subtype, OXPHOS-DLBCL, has the ability to oxidize FAs as the main energy source, and is sensitive to pharmacologic perturbation of this pathway [69]. These studies indicate that FAO may support GC B cells and malignant B cells. GC is a vital microstructure for humoral immune response, where antigen-specific B cells proliferate and undergo somatic hypermutation, affinity maturation and antibody class switching. FAO may be a preferred metabolic pathway for some of the GC B cells. Weisel et al. isolated GC B cells from immunized hapten-specific BCR transgenic mice and investigated their metabolic activity compared to naïve and activated non-GC B cells [70]. Using metabolic flux and [¹³C₁₆]-palmitate labeling assays, they demonstrated that GC B cells primarily undergo FAO (both mitochondrial FAO and peroxisome FAO), but not aerobic glycolysis. They further showed that GC B cells do not exhibit any hypoxic signatures, including activation of hypoxia-inducible factor 1-alpha (HIF1α), which typically promotes glycolysis. However, these observations contradict previous studies [71–73] and remain contentious [74]. One potential explanation stems from the heterogeneity of GC B cells and the fact that Weisel and colleagues did not distinguish different GC B cell subsets. GC is compartmented into light zone (LZ) and dark zone (DZ), which is further divided into proliferating DZ (DZp) and differentiation DZ (DZd) [75]. Cells in different areas are believed to dynamically commute and have specific functions and different gene signatures [76, 77]. LZ cells are enriched in genes associated with hypoxia and glycolysis, while DZ cells are enriched in genes promoting FAO [78]. Thus, it is possible that LZ B cells are in a hypoxic environment and preferentially undergo glycolysis, while DZ B cells preferentially utilize FAO (Figure 2). Since peroxisome FAO is much less efficient for energy generation, it is puzzling that the GC B cells appear to utilize more peroxisome FAO than mitochondrial FAO [70]. Perhaps it is because GC B cells utilize some of the FAs that only peroxisome is equipped to oxidize, such as very long-chain FAs [79]. More detailed functional analyses are required to test these hypotheses.

Figure 2. Germinal center and fatty acid metabolism.

Germinal center (GC) is compartmented into light zone (LZ) and dark zone (DZ). B cells acquire antigen from follicular dendritic cells (FDCs), then the ones with high affinity will receive the help from follicular helper T (Tfh) cells in the light zone (LZ), including CD40L stimulation and IL-21. These signals induce upregulation of c-MYC and PI3K-mTORC1 activation, which promote cell growth, glucose uptake, ribosome biosynthesis. Consequently, LZ B cells are enriched with genes associated with glycolytic metabolism associated with a hypoxic environment and enhanced expression of hypoxia-inducible transcription factor HIF1α. Together, they promote B cell entry into a high proliferation state, which happens in the dark zone (DZ). DZ is further divided into proliferating DZ (DZp) and differentiation DZ (DZd). DZd B cells will again enter LZ for further selection. Gene signatures in DZ are more associated with fatty acid oxidation (FAO) with high expression of carnitine O-palmitoyltransferase 1 A (CPT1A). Such distinct gene expression programs between LZ B and DZ B cells are partly controlled by reciprocal signaling between PI3K in LZ B cells and FOXO1 in DZ B cells.

Fatty acid uptake

FA uptake is another important way to meet the needs for nutrient supply and cellular energy demands of immune cells. LCFA has long been thought to diffuse rapidly across phospholipid bilayers [80], but recent evidence shows that FA uptake is facilitated by membrane-associated proteins, including membrane protein CD36 and FA binding proteins (FABPs) [81–85]. CD36 is a well-known multifunctional class B scavenger receptor and lipid transporter of LCFA implicated in metabolic and inflammatory pathologies [86]. FABPs are small intracellular lipid chaperones, which can bind to LCFAs and facilitate the uptake of cellular LCFA and very long-chain FA. They may actively help the transport of lipids to the specific compartment in cells for storage, signaling transduction, membrane synthesis and metabolism. They also exhibit tissue specific expression patterns [81].

Expression of CD36 is partly driven by the nuclear receptor peroxisome proliferator-activated receptor gamma (PPAR-γ) in an mTORC1 dependent manner [21]. Ma et al. showed that tumor infiltrating CD8⁺ T cells in a murine melanoma model have increased CD36 expression, which promotes lipid peroxidation and ferroptosis, and reduces cytotoxic cytokines production in these CD8⁺ T cells. Blocking CD36 in CD8⁺ T cells restore their antitumor activity. Furthermore, they found that deletion of CD36 in CD8⁺ T cells in combination with anti-PD-1 antibodies achieve better anti-tumor effects than either of them alone [87]. Another study demonstrated that dysfunctional CD8⁺ tumor-infiltrating lymphocytes (TILs) have increased CD36 expression and oxidized low-density lipoproteins (OxLDL), which induce lipid peroxidation, exhaustion markers expression and impair cytokine production. Furthermore, overexpression of glutathione peroxidase-4 (GPX4), a unique antioxidant enzyme, can restore the function of these CD8⁺ TILs [88]. This work corroborates the findings of Ma et al., and together, they reveal an overall detrimental effect of CD36 mediated FA uptake on CD8⁺ TILs. However, CD36 mediated FA uptake may also benefit CD8⁺ T cells. Lin et al. identified CD103^hiCD8⁺ resident memory T (T_RM) cells in the tumor microenvironment of gastric adenocarcinoma patients. They have increased CD36 expression, FA content and FA uptake compared to non-T_RM T cells. Cancer cells with high expression of FABP4 and FABP5 can induce T_RM cell apoptosis by depriving them of FAs, indicating that gastric adenocarcinoma cells outcompete tumor T_RM cells for FA access and increased FA uptake in T_RM cells could improve anti-tumor immunity [89]. These opposite roles of FA uptake in anti-tumor T cell immunity could be caused by different tumor types (B16 murine melanoma and MC38 murine colon adenocarcinoma models were used in CD8⁺ TILs studies [87, 88], while humanized patient-derived xenograft (PDX) gastric adenocarcinoma model was utilized in tumor T_RM study [89]). Another possibility is that different local FA compositions in different tumor models may determine how CD36 mediated FA uptake affects CD8⁺ T cell maintenance and function [84–86]. Because only some of the TILs are T_RM cells, it will be interesting to investigate the metabolic signatures of T_RM and non-T_RM TILs in the same tumor models.

The local microenvironment in vivo can dictate the expression of FA transporters [90]. T_RM cells in the skin have elevated expression of FABP4, and FABP5, which promotes their exogenous FA uptake capabilities and sustains their mitochondrial OXPHOS. The expression of FABP4 and FABP5 are critical for skin T_RM long-term survival and immune function against skin vaccina virus infection [91]. Moreover, T_RM cells express different FABP family members in an organ-specific manner. They can re-program their FABP expression pattern in response to specific tissue-derived factors [90]. These findings demonstrated that local FA environment and FA uptake in the microenvironment cooperate to shape the T_RM phenotypes and support their functions.

Like Teff cells, different FA transporters may have distinct impacts on Tregs in a context-dependent manner. Tumor infiltrating Tregs have higher lipid metabolism and CD36 expression than spleen Tregs [92]. CD36 is indispensable for intratumoral Tregs to maintain mitochondrial fitness, and survival by activating PPAR-β pathway. Consequently, Treg-specific deficiency of CD36, or pharmacological inhibition of CD36, enhances the infiltration and function of anti-tumor Teff cells [92]. Hence, tumor infiltrating Tregs may require both endogenously synthesized FAs and exogenous FAs for their suppressive function in tumor microenvironment [26]. In contrast, FABP5 inhibition or acute deletion of FABP5, but not chronic deficiency of FABP5, enhances Treg suppressive activity, despite impaired mitochondrial respiration and FAO. FABP5 inhibition-enhanced Treg function is attributed to mitochondrial DNA release from damaged mitochondrial, which provokes type I IFN signaling and subsequent IL-10 production [62]. It is unclear why inhibition of CD36 and FABP5 has opposite effects on Treg function. They may mediate transportation of different FAs, which may have distinct effects on Tregs. Another compounding factor is that chronic genetic deficiency may have distinct impacts on Tregs compared to acute gene deletion or pharmacological inhibition, which should be considered in all studies on FA metabolism. More importantly, because Tregs and CD8⁺ TILs exert opposite effects in anti-tumor immunity, and FA uptake can potentially promote both Treg function and CD8⁺ T_RM maintenance, any therapeutic manipulation of FA uptake must take this conundrum into account.

Different B cell subsets exhibit different rates of saturated FA uptake. B1 B cells uptake fluorescence-labeled palmitate at a higher rate than marginal zone (MZ) B cells, whose palmitate uptake rate is higher than mature follicular (FO) B cells. Consequently, B1 B cells and MZ B cells, but not FO B cells, require GPX4 to protect against lipid peroxidation [93]. Consistent with these observations, MZ B cells have higher expression of CD36 than FO B cells [94]. Thus, the FA uptake levels on different B cell subsets may indicate their differential metabolic requirements at steady state. Although lack of CD36 has minimal effects on B cell development, CD36 cooperates with TLR2 to support the antibody response upon bacterial lipoprotein antigen challenge [94]. Plasma cells are key antibody secreting cells that differentiate from naïve B cells in a process driven by transcription factor B lymphocyte-induced maturation protein-1 (BLIMP-1, encoded by Prdm1) [95]. Since PPAR-γ activation can induce CD36 expression and most LCFAs act as its ligands [96], lack of PPAR-γ in B cells also dampens GC B cells generation and plasma cell development associated with reduced B-cell lymphoma 6 (BCL6) and BLIMP-1 expression, key transcription factors for GC and plasma cell generation, respectively [97]. FABP3, a carrier protein of PUFAs, is highly expressed in resting B cells and its expression can be further enhanced upon B cell activation. Although FABP3 deficiency does not affect B cell maturation and antibody levels at steady state, it curtails plasma cell differentiation associated with reduced BLIMP‐1 expression [98]. Thus, CD36 and FABP3 mediated FA uptake supports antibody production and plasma cell generation.

Short-chain fatty acid and mucosal immune response

SCFAs, such as acetate (C2), propionate (C3), and butyrate (C4), are primarily bacterial fermentation products of dietary fiber and are enriched in the colon [99]. Several prominent studies showed that some SCFAs promote colonic Tregs, but not colonic Th1 or Th17, differentiation, highlighting the function of SCFAs in maintaining intestinal T cell homeostasis [100–102]. Later studies indicate that SCFAs, especially at higher concentration, can in fact promote Th1 and Th17 differentiation in vitro, but these SCFA treated Th1 and Th17 cells may not be highly inflammatory in vivo, partly because of increased IL-10 production after SCFA treatment [103–105]. Thus, SCFAs can ameliorate colonic inflammation through increased colonic Treg differentiation and/or IL-10 induction. One study proposed that individual SCFA has differential role in promoting peripheral Treg generation: butyrate promotes spleen but not the colonic accumulation of Treg cells, whereas acetate has a diametrically opposite activity and propionate is capable of both [101]. Yet, butyrate was found to have the most potent effect in promoting colonic Tregs among these 3 SCFAs in a separate study [102]. It is unclear what causes such discrepancy. Mechanistically, SCFAs can engage several G-protein coupled receptors (GPRs), such as GPR41, GPR109a, and GPR43 [106–108]. In addition, SCFAs are known to inhibit histone deacetylase (HDAC) [109]. SCFAs induce FOXP3 and IL-10 expression in colonic Tregs by suppressing the activity of HDAC. Their IL-10 induction capacity is also dependent on the activation of BLIMP-1 and mTOR signaling [100–103]. Whether SCFA engagement of GPRs contributes to these phenotypes remains unresolved, with evidence for [100, 105, 110] and against it [101–103]. Beyond Treg and IL-10 induction, SCFAs can also stimulate CD4⁺ T cells to produce IL-22, which has a dual role in host response to inflammatory insults, either protective or exacerbating depending on the context [111]. SCFAs induce IL-22 production from intestinal CD4⁺ T cells through GPR41 and HDAC inhibition. Importantly, SCFA induced IL-22 protects against enteric bacterial infection [112]. In terms of CD8⁺ T cells, Luu et al. found that butyrate increases IFN-γ and granzyme B expression, which is associated with its HDAC inhibition activity, but independent on GPR41 and GPR43 [113]. Moreover, SCFAs enhance memory potential, recall capacity, and OXPHOS of activated CD8⁺ T cells in a manner dependent on GPR41 and GPR43 [114]. Therefore, it appears that SCFAs may affect T cell differentiation and function through context-dependent mechanisms, i.e., different T cell subsets at different differentiation stages utilizing GPR-dependent or independent mechanisms (Figure. 3). Clearly, more research is needed to clarify these context-dependent mechanisms under which SCFAs modulate T cell immunity.

Figure 3. Overview of signaling pathways regulated by short- and long-chain fatty acids in T and B cell.

Short-chain fatty acids (SCFAs), such as acetate (C2), propionate (C3), and butyrate (C4), can modulate colonic T regulatory (Treg) cell generation and induce IL-10 and FOXP3 expression by suppressing histone deacetylase (HDAC) activity, or through activation of their ligands G-protein coupled receptors (GPRs). The induction of IL-10 is also dependent on the activation of BLIMP-1 and mTOR signaling. SCFAs promote CD8⁺ T cell memory response dependent on GPR41 and GPR43. T cell uptake of long-chain fatty acids (LCFAs) is mediated by FA binding proteins (FABPs) and CD36, which facilitate fatty acid oxidation (FAO) in mitochondrial and increase oxidative phosphorylation (OXPHOS), and consequently promote memory T cell generation and Treg differentiation. LCFAs can engage peroxisome proliferator-activated receptor gamma (PPAR-γ), which promotes CD36 expression to form a feed-back loop. They also engage PPAR-β to promote mitochondrial fitness in Tregs. In terms of B cells, low dose SCFAs increase the cellular acetyl-CoA level and induce IgA and IgG production by enhancing mitochondrial oxidative phosphorylation, glucose metabolism, mTOR signaling, and suppressing HDAC activity. However, SCFAs at high dose suppress IgG production through inhibiting HDAC activity, which modulates expression AID and BLIMP-1. FAO promotes B cell activation and GC formation by increasing OXPHOS. During B cell activation, expression of CD36 and FABP3, and activation of PPAR-γ, are required for optimal antibody response and memory B cell generation, possibly through activation of key transcription factors, BCL6 and BLIMP-1. LCFAs are also required to suppress excessive endoplasmic reticulum (ER) stress and autophagy in B cells. Antigenic signals through T cell receptor (TCR) and CD28 (in T cells), or B cell receptor (BCR) and Toll-like receptor (TLR) (in B cells), can activate mTOR-mediated de novo fatty acid biosynthesis (FAsyn), which produces intrinsic LCFAs and more complex FA-containing lipids, such as diacylglyceride (DAG) and phospholipid (PL). These lipids can be further metabolized to release FAs, which can be transported to mitochondrial and undergo FAO.

Like T cells, the impacts of SCFAs on humoral immune response are debatable. Kim and colleagues showed that mice fed with high level of dietary fiber have increased mucosal IgA⁺ plasma cells, intestinal IgA secretion and systemic IgG level. These effects are most likely due to SCFAs produced by intestinal microbial fermentation of the dietary fiber, because antibiotics treatment abolishes the increased antibody level induced by high dietary fiber feed, and oral administration of propionate increases intestinal IgA production when mice are fed with either low fiber diet or regular diet. Mechanistically, SCFAs affect B cells independent of GPR41 and GPR43. Instead, they are converted to acetyl-CoA, which is fed into TCA cycle and promotes OXPHOS and glucose metabolism associated with mTOR activation. They also promote activation-induced cytidine deaminase (AID, encoded by Aicda) and BLIMP-1 expression through their HDAC inhibitory activity [115]. Consistent with these observations, several other studies showed that high fiber diet or microbial SCFAs increase intestinal IgA production [110, 116]. In particular, acetate can not only promote colon IgA production, but also direct selective IgA binding to different microorganisms by orchestrating epithelial-immune cell interaction and altering the localization of intestinal bacteria within the colon [117]. However, the effects of SCFAs on antibody response could be dose dependent. Sanchez et al. showed that low doses SCFAs moderately enhance IgG production, but at higher doses, they suppress antibody production and plasma cell generation. Furthermore, they demonstrated that SCFAs dampen humoral immune response through inhibition of AID and BLIMP-1 expression by increasing miRNA targeting Aicda or Prdm1 [118, 119]. One important issue is to monitor the SCFA concentrations in the colon lumen longitudinally before and after dietary SCFA interventions in order to find the optimal dosing regimen that generates physiologically relevant SCFA concentrations. In addition, whether the impact of SCFA on B cells is dependent on GPRs is controversial. only Gpr43 RNA is detectable on activated mouse B cells, but not Gpr41, Gpr109a or Olfactory receptor 78 (Olfr78). None of them are detectable on activated human B cells [118]. Genetic deletion of GRP41 or GPR43 does not affect intestinal IgA production or IgA⁺ plasma cell frequency with or without SCFA treatment [115, 117]. Blocking GPR43 also does not affect butyrate and propionate-mediated modulation of Aicda and Prdm1, class switch reaction, or plasma cell differentiation [118]. In contrast, one study showed that GPR43 deficient mice have reduced intestinal IgA level upon acetate feeding compared to control mice. Acetate may not act directly on B cells, but through dendritic cells in a GPR43-dependent manner [116]. It is unclear why different groups get different conclusions in terms of the impacts of SCFAs on mucosal IgA production. One needs to note that SCFAs are not the only FAs that modulate intestinal IgA. Dietary palmitic acid can also promote IgA production in colon by stimulating IgA-producing plasma cells [120]. Thus, dietary FA intervention needs more standardization to ensure a controlled and rigorous investigation across different labs. GPR43 may also control systemic humoral response. One study showed that systemic loss of GPR43 has heightened antibody response in the T cell-independent immunization model, but not in a T cell-dependent immunization model [121]. However, it is unclear whether these phenotypes are modulated by SCFAs or other ligands for GPR43. Finally, microbiome-derived acetate can promote the generation of IL-10 producing regulatory B cells with anti-inflammatory effects [122]. Together, these findings reveal a complex effect of SCFAs on B cell responses, which can be affected by the doses of SCFAs, and may have different molecular and cellular mechanisms under different immunological contexts (Figure 3).

Conclusions and future perspectives

As illustrated in this review, we have just started to appreciate the breadth and complexity of FA metabolism in adaptive immune responses. It is clear that FAs are far more than mere energy storage. FA synthesis, oxidation, uptake, and posttranslational modification are all involved in immune cell activation, lineage differentiation, and memory response through a plethora of mediators, signaling pathways or metabolic processes. They are also intimately linked to immunological diseases, from pathogen infection to autoimmune disorders. Hence, one future development is to develop novel therapies targeting FA metabolism to treat immunological diseases, including cancer and autoimmune disorders. In this regard, FAsyn has been a major target for cancer therapy because cancer cells rely on FAs for their rapid proliferation and survival [123]. Because T cells also depend on FAsyn, and potentially FA uptake in certain circumstances, for their effector function, therapies targeting FA metabolism must balance their impacts on both immune cells and cancer cells. For autoimmunity, more investigations are needed to first identify the nature of FA metabolic dysregulation in different autoimmune disorders in order to design appropriate interventions. As we also discussed throughout this review, there are many challenges and unresolved controversies in the field. Different FA species at different doses and different experimental settings may effect distinct outcomes on adaptive immunity. Therefore, we believe that developing rigorous standardized protocols, especially in dietary interventional experiments, will help the field move forward with more rigor and less contention.

Abbreviations

FA

fatty acid

FAO

fatty acid oxidation

SCFA

short-chain fatty acid

LCFA

long-chain fatty acid

MCFA

medium-chain fatty acid

PUFA

polyunsaturated fatty acid

OA

oleic acid

DAG

diacylglyceride

TAG

triacylglyceride

PL

phospholipid

PE

phosphatidylethanolamine

FAsyn

FA synthesis

ACC

acetyl‐CoA carboxylase

FASN

fatty acid synthase

MUFA

monounsaturated FA

SCD

stearoyl-CoA desaturase

mTORC1

mechanistic target of rapamycin complex 1

RAPTOR

regulatory-associated protein of mTOR

SREBP

sterol regulatory element binding proteins

Th

T helper

Tregs

regulatory T cells

Teff

effector T cells

Tfh

follicular helper T

RA

rheumatoid arthritis

AMPK

AMP-activated protein kinase

OXPHOS

oxidative phosphorylation

TCA

tricarboxylic acid

ACLY

ATP-citrate lyase

ACSS2

acyl-CoA synthetase short-chain family member 2

ER

endoplasmic reticulum

CPT1

carnitine O-palmitoyltransferase 1

CACT

carnitine-acylcarnitine translocase

TRAF6

TNF receptor associated factor 6

SRC

spare respiratory capacity

shRNA

small hairpin ribonucleic acid

Glut1

glucose transporter 1

FABP

fatty acid binding protein

CLL

chronic lymphocytic leukemia

DLBCL

diffuse large B-cell lymphoma

GC

germinal center

LZ

light zone

DZ

dark zone

DZp

proliferating DZ

DZd

differentiation DZ

BCR

B-cell receptor

TCR

T cell receptor

HIF1α

hypoxia-inducible factor 1-alpha

PPAR-γ

peroxisome proliferator-activated receptor gamma

TILs

tumor-infiltrating lymphocytes

GPX4

glutathione peroxidase-4

T_RM

resident memory T

PDX

patient-derived xenograft

IFN

interferon

MZ

m=arginal zone

FO

follicular

TLR

Toll-like receptor

BCL6

B-cell lymphoma 6

BLIMP-1

B lymphocyte-induced maturation protein-1

C2

acetate

C3

propionate

C4

butyrate

GPRs

G-protein coupled receptors

HDAC

histone deacetylase

AID

activation-induced cytidine deaminase

Olfr78

olfactory receptor 78

FDC

follicular dendritic cell

NADPH

Dihydronicotinamide-adenine dinucleotide phosphate

OxLDL

oxidized low-density lipoproteins

IL

Interleukin

DNA

deoxyribonucleic acid

CD

cluster of differentiation

PD-1

Programmed cell death protein 1

Ig

Immunoglobulin