c-MAF, a Swiss Army Knife for Tolerance in Lymphocytes

Beyond its well-admitted role in development and organogenesis, it is now clear that the transcription factor c-Maf has owned its place in the realm of immune-related transcription factors. Formerly introduced solely as a Th2 transcription factor, the role attributed to c-Maf has gradually broadened over the years and has extended to most, if not all, known immune cell types. The influence of c-Maf is particularly prominent among T cell subsets, where c-Maf regulates the differentiation as well as the function of multiple subsets of CD4 and CD8 T cells, lending it a crucial position in adaptive immunity and anti-tumoral responsiveness. Recent research has also revealed the role of c-Maf in controlling Th17 responses in the intestine, positioning it as an essential factor in intestinal homeostasis. This review aims to present and discuss the recent advances highlighting the particular role played by c-Maf in T lymphocyte differentiation, function, and homeostasis.


INTRODUCTION
The Maf (musculoaponeurotic fibrosarcoma) gene encodes the transcription factor c-Maf or MAF. Originally identified in natural musculo-aponeurotic fibrosarcoma of chickens infected with the replication-defective retrovirus AS42, the founding member of the Maf family, named v-Maf, was described as an oncogene (1)(2)(3). Using a probe containing the v-Maf sequence, its cellular counterpart, identified as c-Maf, was thereafter cloned from a number of vertebrate genomes (4). In addition to its function as an oncogene, c-Maf was soon found to regulate various cellular differentiation and developmental processes within tissues. In particular, c-Maf expression controls lens fiber cell differentiation, crystalline gene expression, as well as lens development (5)(6)(7). In neural tissue, c-Maf controls the expression of mechanoreceptors involved in touch sensation (8,9). It also regulates the embryonic development of tubular renal cells (10) and the differentiation of chondrocytes during endochondral bone development (11)(12)(13). c-Maf plays a predominant role for the erythropoiesis that accompanies erythroblastic islands formation in fetal liver (14). In porcine and human pancreatic islets (15), c-Maf also regulates glucagon hormone production, thereby establishing pancreatic endocrine function (16). In line with the major contributions of c-Maf in developmental and physiological processes, mice lacking c-Maf are embryonically (14) or perinatally (5,7) lethal depending on the type of C57BL/6 background. Some mice on the BALB/c background live to adulthood (10,13).
In parallel to the discovery of the many roles of c-Maf within tissue development, c-Maf soon emerged as an immune regulator and was initially identified as a Th2 transcription factor. Similar to its function in tissue development, the role attributed to c-Maf within immune regulation broadened over the years and has extended to most, if not all, known immune cell types. While the role of c-Maf has also been studied within innate immune cell types (17)(18)(19) and B lymphocytes (20), we focus on c-Maf within T cell subsets, where c-Maf regulates the differentiation as well as the function of multiple subsets of CD4 T cells, lending it a crucial position in T cell immunity. Recent research has revealed the role of c-Maf in the control of intestinal Th17 responses by regulatory T cells, positioning it as an essential factor in regulatory T cell specification and, more broadly, the maintenance of intestinal homeostasis. This review aims to present and discuss the recent advances highlighting the particular role played by c-Maf in T lymphocyte differentiation, function, and homeostasis.

THE c-MAF TRANSCRIPTION FACTOR
This basic leucine zipper (bZIP) transcription factor belongs to the AP-1 superfamily, which includes Fos, Jun, ATF, and CREB. The Maf transcription factor family is composed of 7 members divided into two subclasses: the large Maf proteins composed of MAFA/L-MAF, MAFB, MAF/c-Maf, and NLR (neural retina leucine zipper), and the small Maf proteins, MAFK, MAFG, and MAFF, which lack the amino-terminal transactivation domain. The Maf family of transcription factors harbors a unique and highly conserved basic region-leucine zipper (bZIP) structure (21). The basic regions of dimeric Maf factors allows them to recognize a palindromic sequence referred to as the Maf Recognition Element (MARE). This sequence is composed of a 7-bp TPA-Responsive Element (TRE) or a 8-bp cyclic AMP-Responsive Element (CRE) core region and a TGC flanking sequence bound by the Extended Homology Region (EHR), exclusively found in Maf proteins (22) (Figure 1). This long recognition sequence thus distinguishes the Maf protein family from other AP-1 family members and contributes to the important functions of the Maf proteins (23).
Thanks to their leucine zipper domain, Maf proteins can form homo-and heterodimers with other compatible bZIP proteins, such as Jun and Fos (24,25). Maf proteins can also interact with other non-bZIP proteins including specific transcription factors, such as Sox family members (11).
Three isoforms exist for human c-Maf: a short form (373 amino acids), a medium form (383 amino acids), and a long form (30 amino acids more than the short form) of 38.5, 39.6, and 42 kDa respectively. In mice, only two isoforms have been reported: the long form (380 amino acids), called MAF-201, and the short form (370 amino acids). So far, distinct c-Maf products have not shown functional differences, but a potential functional specification cannot be excluded.
c-Maf is located on the chromosome 16q23.2 in humans and on chromosome 8 in mice (26,27). c-Maf is translocated in 5-10% and/or overexpressed in 50% of multiple myelomas (MM) (28,29). c-Maf overexpression in MM drives cyclin D2, integrin β7 and ARK5 expression and leads to proliferation, adhesion to bone marrow stroma cells, invasion, and migration of plasma cells (30). c-Maf is also highly expressed in over half of the angioimmunoblastic T-cell lymphomas (AITL) (30,31). Transgenic overexpression of c-Maf in T cells regulates the same gene expression set as in plasma cells and induces T-cell lymphoma development in mice (30), therefore indicating that c-Maf is a bona fide oncogene contributing to the progression of hematological malignancies. c-Maf is also expressed by other cancers, such as renal or head and neck cancer, yet its expression is not systematically correlated with a bad prognosis (32).

INDUCTION OF c-MAF IN T CELLS
c-Maf expression and activity are regulated at transcriptional, post-transcriptional, as well as post-translational levels. Transcription factors and RNA-mediated silencing control the amount of Maf transcripts, while phosphorylation and SUMOylation modify the activity, sub-cellular localization, and half-life of the protein.
In T cells, antigenic stimuli that modulate the stability of the Maf encoding mRNAs and/or the c-Maf protein may also induce the expression of transcriptional activators, or may cooperate with independent transcriptional stimuli, such as cytokinedriven STAT factors to induce c-Maf transcription. The selective use of those pathways by different stimuli and in distinct cell populations provides the potential for tailoring c-Maf expression to different circumstances.
The highest levels of Maf transcripts can be detected in Th17 and Tfh cells. During Th17 cell polarization, both TGF-β and IL-6 are required for maximum induction of c-Maf, which in turn depends on STAT3 phosphorylation, but not STAT1 or STAT6 activation (35,36,38,39). Moreover, after IL-6 stimulation,  c-Maf expression was initially thought to rely on the IL-4/STAT6 signaling pathway during Th2 cell differentiation, as ectopic expression of activated STAT6 in Th1 cells promoted c-Maf expression, along with Th2-specific cytokines and GATA3 expression (34). However, introducing GATA3 into STAT6deficient T cells restored c-Maf expression, therefore suggesting an indirect role of STAT6 in c-Maf induction during Th2 differentiation (40). Of note, the IL-6/STAT3 signaling pathway is central to c-Maf expression during Th2 cell development (35,41). In particular, Th2 cells express activated forms of STAT3 downstream of a STAT6-signaling pathway. Ablation of STAT3 in developing Th2 cells does not preclude GATA3 and IL-4 expression but selectively impairs c-Maf expression (41). This STAT6-to-STAT3 signaling pathway thus reconciles previous contradictory results concerning the role of STAT6 in c-Maf induction and further supports an indirect role of STAT6 in c-Maf expression.
The inducible co-stimulator (ICOS), expressed by activated T cells, promotes expression of c-Maf in murine Th2 cells and in both mouse and human Th17 cells, although the molecular mechanisms beyond this induction are still ill-defined (33,(42)(43)(44).
IL-27, a member of the IL-12/IL-23 heterodimeric family of cytokines produced by APCs, is also a potent inducer of c-Maf during Tr1 cell differentiation (45). Interestingly, IL-27 signals through STAT1/STAT3 has been shown to up-regulate ICOS expression, thus activating two independent pathways that might up-regulate c-Maf.
Prostaglandin E2 (PGE2), a pro-inflammatory lipid mediator abundant at inflammatory sites, has recently been shown to inhibit c-Maf expression in developing Tr1 cells (46). PGE2 did not affect STAT1/3 activation and its inhibitory effect was mediated through the EP4 prostaglandin receptor and cAMP signaling (46).
The expression of a c-Maf specific intergenic long noncoding RNA, called linc-MAF-4, in Th1 cells inhibits Maf transcription through the recruitment and activation of chromatin-modifying complexes, including the PCR2-associated histone methyltransferase, enhancer of zeste homolog 2 (EZH2), and the lysine-specific histone demethylase 1A (LSD1) (47). Up-regulation of linc-MAF-4 in human CD4 T cells is directly involved in the down-regulation of Maf expression and correlates with encephalitogenic Th cell differentiation and annual relapse rate in patients with multiple sclerosis (48).

Post-transcriptional Regulation
c-Maf expression is tightly regulated by small noncoding microRNAs (miRNAs). In particular, Maf contains phylogenetically conserved miR-155 seed matches in its 3 ′ -UTR. Luciferase reporter experiments confirmed that the c-Maf 3 ′ -UTR is a direct target of miR-155 (49). MiR-155 is strongly expressed in activated T cells and genetic invalidation of miR-155 led to increased levels of c-Maf in T cell lines, thus positioning miR-155 as a major regulator of c-Maf expression in vivo (49). The role of miR-155 in suppressing c-Maf expression has been further extended to microglia cells during the response to CNS ischemia (50). c-Maf is also targeted by miR-143 and miR-365 in macrophages (51), and by miR-1290 in laryngeal carcinomas (52). However, expression of those miRNAs has not yet been reported in T lymphocytes.

Post-translational Control of the Biological Activity of c-Maf
Upon TCR activation, the CARMA1-dependent activation of the IKK complex results in the phosphorylation of the N-terminus part of c-Maf, which is required for nuclear translocation and binding to the promoter of target genes (53). In particular, the T cell-specific deficiency of either CARMA1 or its substrate IKKβ strongly reduced the DNA binding activity of c-Maf without affecting c-Maf abundance. This IKK-mediated activation of c-Maf is independent of NF-kB activation (53). Loss of CARMA1/IKK signaling resulted only in a partial decrease of c-Maf phosphorylation, suggesting that c-Maf might be phosphorylated by multiple kinases. c-Maf is phosphorylated by the Ser/Thr glycogen synthase kinase 3β (GSK3) in human multiple myeloma cell lines and in the lens, leading to protein stabilization (54,55). However, whether GSK3 exerts a similar role in T cells is difficult to evaluate as GSK3 inhibition increases expression of c-Maf in this context (56).
Tyrosine phosphorylation of c-Maf is also critical for its recruitment to the IL-4 and IL-21 promoters and for optimal cytokine production. Phosphorylation of c-Maf on tyrosine residues has been shown to be positively and negatively regulated by the TEC tyrosine kinase and the PTPN22 tyrosine phosphatase, respectively (57).
SUMOylation of c-Maf at the lysine 33 residue reduces its ability to bind the Il4 promoter and decreases the transactivating activity of c-Maf in a luciferase reporter assay (58,59). In addition, a recent report indicated that c-Maf SUMOylation is negatively correlated with Il21 expression in CD4 T cells from diabetogenic NOD mice (60). Furthermore, transgenic expression of a SUMO-defective c-Maf selectively inhibited recruitment of Daxx/HDAC2 to the Il21 promoter and enhanced histone acetylation mediated by CREB-binding protein (CBP) and p300. Thus, the SUMOylation status of c-Maf has a stronger regulatory effect on IL-21 than the level of c-Maf expression, through regulation of epigenetic mechanisms (60).

Regulation of IL-10 Secretion in Multiple T Cell Subsets
Multiple roles have been attributed to c-Maf in distinct T cell subsets (Figure 2), revealing context-specific effects of this transcription factor. However, c-Maf positively regulates Il10 expression in virtually all immune cells, including T cells, B cells, macrophages, and dendritic cells (36,37,(61)(62)(63)(64), suggesting a common regulatory function beyond distinct T cell subsetspecific roles. IL-10 is an essential anti-inflammatory cytokine that plays important roles as a negative regulator of immune responses to foreign or self-antigens and prevents excessive inflammation during the course of infection [reviewed in (65)(66)(67)].
Exploring c-Maf binds to consensus MARE motifs in the Il10 promoter (36,37). Although c-Maf can transactivate Il10 by itself to some extent, c-Maf alone is not sufficient to induce optimal Il10 expression in T cells (36,37). Robust IL-10 expression requires interaction with additional transcriptional regulators that vary among T cell subsets. c-Maf cooperates with the aryl hydrocarbon receptor (AhR) to induce IL-10 in regulatory type 1 (Tr1) cells (37). AhR expression is mainly driven by TGF-β (68) and is not expressed in Th1 cells, in which fine-tuning IL-10 expression mostly relies on the interaction of c-Maf with Blimp-1 (64). IL-10 expression in Th2 cells relies on transcription factors STAT6, GATA3, and IRF4 (69, 70) but whether these factors interact directly with c-Maf awaits further investigation.
Thus, the activity of c-Maf on the Il10 enhancer might not only depend on the accessibility of its motif but also on the nature of the other transcription factors that co-bind to that enhancer. In other words, Il10 expression in Th cells relies on several transcriptional programs that, together with c-Maf, are able to integrate various signals from the environment in order to fine-tune this critical immunosuppressive cytokine.
In addition to a direct positive transcriptional regulation of Il10 expression, c-Maf also provides a common mechanism for a negative regulation of IL-2 signaling in vivo in models of Th1, Th2, and Th17 responses (67).
However, although c-Maf uses common mechanisms of gene regulation in distinct cell subsets, the net outcome of c-Mafdeficiency is different in each cell type, thus indicating that c-Maf has context-specific effects on the immune response, over and above its effects on IL-10 and IL-2 signaling.

Th1/Th2 Cells
The cross-regulation between Th1 and Th2 cells is mediated, in part, by the transcription factors that they express. c-Maf was first described as a Th2-specific gene that induces Il4 gene transcription via direct binding to the Il4 but not the Il5 or Il13 locus (71). Transcription factors GATA3, STAT6, and NFAT can synergize with c-Maf to regulate IL-4 expression in Th2 cells (72)(73)(74). Moreover, overexpression of c-Maf skews the immune response toward a Th2 response (75).
Although these pioneer studies concluded to a pro-Th2 role of c-Maf, normal levels of IL-13 and IgE were observed in c-Mafdeficient mice (72). Contrary to a pro-Th2 effect of c-Maf on the immune response, increased Th2 lung pathology, associated with higher numbers of eosinophils in bronchoalveolar lavage fluids, was observed in the HDM allergy model despite a decreased expression of Il4 in T-cell specific c-Maf deficient mice (67). Of note, cells producing both IL-4 and IL-10, but not IL-4 + IL-10 − Th cells, were lost in this allergy model, in keeping with increased pathology. Thus, although c-Maf can activate the Il4 promoter, its net effect over the Th2 inflammatory response is mainly inhibitory.
In contrast with data obtained in naïve Th cells, ectopic expression of c-Maf in mature Th1 cells did not grant them the ability to produce IL-4, but did decrease their production of IFN-γ (75). In a recent study, chronically activated Th1 cells that were cultured with IL-4-producing Th2 cells upregulate Maf expression (76). These cells were shown to down-regulate Ifng expression and express a dampened

Follicular Helper T Cells
Follicular helper T cells (Tfh) are key regulators of T celldependent long-term humoral immunity (78). Tfh cells express BCL6, a transcriptional repressor considered as the critical master regulator of Tfh cell development in vivo (79,80), and also constitute the major source of IL-21, a cytokine necessary for IgG class-switch recombination and antibody affinity maturation (81).
Using retroviral ectopic expression of c-Maf or BCL6 in in vitro-derived human Tfh cells, Kroenke et al. first reported that c-Maf and BCL6 regulate distinct features of Tfh cell functions, with BCL6 required for Tfh cell development and c-Maf for promoting IL-21 secretion (82). However, recent data have shown that c-Maf is expressed early during Tfh cell differentiation and is critical for Tfh cell development in vivo (83). This is in agreement with the finding that Tfh cell differentiation strongly relies on ICOSL/ICOS and IL-6/STAT3 signaling, two pathways known to induce the expression of c-Maf (35,(84)(85)(86).
The relative roles of c-Maf, BCL6, and other transcription factors in initiating and maintaining Tfh cell are still poorly defined. As c-Maf-bound genes in Tfh cells hardly correlate with genes bound by BCL6 or Ascl2 (87,88), it is tempting to speculate that cooperation between c-Maf and BCL6/Ascl2 is required to reach complete Tfh cell fate through orchestration of distinct sets of genes. Although c-Maf does not regulate Bcl6 transcription, a defect in BCL6 expression in CD4 T cells was observed in the absence of c-Maf, thus suggesting that c-Maf could contribute to BCL6 expression in developing Tfh cells (83).
Beside its role in Tfh cell development, c-Maf is also required for adequate IL-4 and IL-21 production through transactivation of the Il4 and Il21 promoters (82,89,90)

. In particular, Sahoo et al. reported that c-Maf promoted IL-4 secretion in Tfh cells through both direct binding to the CNS2 region in the Il4 locus and via induction of IRF4, thus revealing a distinct role of c-Maf in IL-4 secretion between Th2 and Tfh cell subsets (90).
Th17 Cells c-Maf is highly expressed in Th17 cells and impacts several important aspects of their differentiation and function (33,36,39,61,64,67,91) (Figure 3). It physically associates with the transcription factor Sox5 and, together, they bind and activate the promoter of Rorc in conventional CD4 T cells. The Maf-Sox5 interaction thus controls Th17 development via the induction of RORγt as downstream targets of STAT3 (92). c-Maf also positively regulates certain loci in Th17 cells, including several genes known for controlling inflammation (e.g., Il9, Lif, Il10) (33,37,82,89). c-Maf binds the Il21 promoter, inducing the production of IL-21 which subsequently sustains Th17 expansion (83) and stabilization through IL-23R expression (33). Some evidence also suggests that c-Maf-induced IL-21 secretion could trigger a positive feedback loop by activating STAT3, thus further promoting c-Maf and leading to the development of memory Th17 cells (93). Sato et al. suggested that c-Maf can also directly transactivate the Il23r gene which contains a MARElike sequence (93). However, the relation between IL-23R and c-Maf is still controversial as it was recently reported that c-Maf downregulates Il23r expression in a subset of memory Th17 cells (91). c-Maf has been shown to act as a repressor of Il2 in CD4 T cells, which indirectly amounts to increased Th17 differentiation in EAE models. Indeed, T cell specific c-Maf deficiency led to an improvement of the disease through increased IL-2 production and decreased Th17 differentiation (67).
c-Maf also acts as a global negative regulator of genes associated with Th17 function. This was demonstrated by an exhaustive study of the regulatory network for Th17 cell specification, where c-Maf attenuates the expression of proinflammatory loci (e.g., Rora, Runx1, Il1r1, Ccr6, Tnf ) and repressing genes belonging to pathways regulated by other core transcription factors. c-Maf represses IL-22, but not IL-17A production, in a TGF-β-dependent way. It binds to the Il22 promoter and blocks the positive transcriptional effects of other Th17 transcription factors on the Il22 locus, including RORγt and BATF (39).
Importantly, c-Maf maintains tolerance through the regulation of IL-10, as described above. In Th17 cells, c-Maf regulates the balance between the differentiation toward inflammatory or anti-inflammatory Th17 cells (91). c-Maf controls this balance by binding to enhancers or putative enhancers already available in the general landscape of Th17 cells but not through the direct binding to promoters (91). It promotes an immuno-regulatory program (Il10, Ctla4) while repressing pro-inflammatory associated genes (Ifng, Il22, Il12rb2) (91). Interestingly, c-Maf also promotes the expression of genes associated with memory Th17 cell tissue-residency, such as Ccr9 and Cxcr6 (91).
The environmental cues guiding the transition between inflammatory and anti-inflammatory Th7 cells are still not clearly established. The presence of inflammatory cytokines, such as IL-1β, favors an inflammatory state, as described in Aschenbrenner et al. (91), without directly affecting the level of c-Maf in the cell. The levels of STAT3 activating cytokines, such as IL-6, IL-27, or IL-10, may have the opposite effect and push Th17 cells toward a more anti-inflammatory phenotype through c-Maf induction (36,91). Altogether, c-Maf plays a critical role in Th17 differentiation and functions by balancing the inflammatory properties of these cells, and by doing so, it plays a critical role in the regulation of tolerance/inflammation.

Regulatory T Cells (Treg) and Gut Tolerance
Regulatory type 1 (Tr1) cells have emerged as an important subset of T cells with strong immunosuppressive properties but which does not express master transcription factor Forkhead box 3 (Foxp3), contrary to regulatory T cells (Tregs). The protective role of Tr1 cells has been shown in numerous contexts, such as autoimmunity, colitis, graft-versus-host disease, and tissue inflammation (94). The anti-inflammatory effects of Tr1 cells mainly rely on their ability to produce high amounts of IL-10. The cytokine IL-27 is known to promote the expansion and the differentiation of Tr1 cells through the induction of transcription factor c-Maf and co-stimulatory receptor ICOS. In these cells, c-Maf physically interacts with AhR to transactivate the Il10 and Il21 promoters (37,45,95). Of note, the expression of IL-21 further sustains c-Maf expression in Tr1 cells, as a feed-forward loop, thus highlighting the major role of c-Maf, both in the induction and the stabilization of the Tr1 cell fate (37,95).
Distinct Treg populations adopt specialized phenotypes by co-expressing Foxp3 and lineage-defining transcription factors in response to tissue-or inflammatory-driven signals (100). In particular, RORγt + Tregs represent a subset of Tregs that develops in the intestinal tissue of naïve mice in response to signals arising from a complex microbiota. This subset of peripherally induced Tregs has been shown to efficiently protect from intestinal immunopathology in different colitis models (101,102) and to mediate immunological tolerance to the gut pathobiont Helicobacter hepaticus (38). RORγt + Tregs express high levels of c-Maf and genetic ablation of c-Maf leads to a severe defect in the development of this subset (38,(96)(97)(98)(99).
Numerous studies published in the last couple of years highlighted the critical role of c-Maf in the regulation of gut homeostasis. As mentioned above, c-Maf promotes IL-10 production in distinct T cell subsets. Using IL-10 reporter mice, Neumann et al. recently showed that IL-10 production by Tregs was strictly associated with c-Maf expressing Tregs (96). IL-10 is critically involved in regulating intestinal homeostasis as its total invalidation (103) or T cell specific inactivation (104)(105)(106)(107) led to the development of a strong colitis in mice. IL-10 regulation therefore underlies a pivotal aspect of c-Maf control over gut homeostasis. Of interest, c-Maf regulates the development of RORγt + Tregs, which are the major source of IL-10 in this organ (38,97). Thus, c-Maf might control IL-10 production in the intestine both directly, through regulation of Il10 gene transcription, and indirectly, via the differentiation of a high IL-10-producing Treg subset.
Genetic ablation of c-Maf also impairs the differentiation of follicular regulatory T cells (Tfr) in Peyer's patches (98). Tfr cells co-express both Tfh (CXCR5, PD-1, Bcl6) and Treg (Foxp3, CTLA4) markers and represent a regulatory T cell subset that controls the activity of Tfh cells. In the intestine, Tfr cells foster microbiota diversity via the regulation of symbiotic bacteriaspecific IgA affinity (108). However, the functional consequences of Tfr cell deficiency in the gut homeostasis of mice harboring c-Maf-deficient Tregs requires further investigation.
Although the control exerted by c-Maf in gut homeostasis was clearly established by different groups, the extent of the c-Maf deficiency-driven inflammation varied drastically in distinct studies. Xu et al. reported that c-Maf inactivation in Tregs (Maf Treg ) induced a colitis with the apparition of a rectal prolapse in a third of the mice (38). This phenotype was observed after infection of the animals with the pathobiont Helicobacter hepaticus, but was much more reduced in the absence of infection. The same mouse model did not develop spontaneous colitis, and only showed mild signs of immune cell infiltration and colon tissue destruction in other studies that were not using the pathobiont (96,97). Surprisingly, Maf Treg mice were even protected from acute DSS induced colitis by an increased production of IL-17 and IL-22 (96). Lastly, c-Maf inactivation in all T cell compartments (Maf T ) resulted in a spontaneous strong onset of colitis (97). This gradation in colitis development might be related to subtle differences in the microbiota composition of mice bred in distinct animal facilities and to a gradient of inflammation induced by the presence or not of certain pathobionts. But it might also reflect the role of c-Maf in multiple T cell subtypes in the gut (Figure 4).
The absence of RORγt + Tregs in the gut of Maf Treg mice strongly decreases the threshold for the development of colitis. However, unlike Maf Treg mice, RORγt Treg mice do not spontaneously develop colitis, suggesting that c-Maf has a more substantial role than RORγt in the function of intestinal Tregs. Il10 transcripts are still detectable in the colon of Maf Treg mice (97) and Imbratta et al. recently reported that Th17 cells from the colon express high levels of c-Maf and can produce IL-10. In Maf T mice, this additional source of IL-10 is lost and the threshold for the development of colitis decreases further. This phenotype is lost when the animals are treated with antibiotics, further confirming that the microbiota plays a critical role in the development of colitis, probably through the induction of c-Maf and the development of RORγt + Tregs in the colon (97). We can thus hypothesize that the presence of the microbiota induces a level of inflammation that triggers c-Maf expression in both intestinal Tregs and Th17 cells. c-Maf counterbalances this inflammation and promotes resistance to pathobionts through the production of IL-10 and/or the development of RORγt + Tregs (Figure 4).
c-Maf inactivation strongly impaired IL-10 production in both Helios positive and negative Tregs in distinct organs, showing that c-Maf also regulates IL-10 production in Tregs of thymic and peripheral origin (96). However, ablation of c-Maf in the Treg or other T cell compartments did not result in systemic autoimmune disease, nor did it disturb conventional and regulatory T cell homeostasis in lymphoid organs, and c-Maf-deficient Tregs retained their in vitro suppressive capacity (38,96,98). Thus, although the precise role of c-Maf in the suppressive capacity of Tregs in distinct pathophysiological settings awaits further investigation, it seems that this transcription factor is a major player in colon homeostasis and that it drives immunological tolerance to gut pathobionts, thereby restraining inflammatory bowel disease.

ROLE OF c-MAF IN TUMOR INFILTRATING CD8 T CELLS
c-Maf is not expressed in CD8 T cells at steady state. The expression of c-Maf in CD8 T cells was first described in tumor infiltrating lymphocytes (TILs) obtained from a mouse melanoma model and from melanoma patient (109). Overexpression of c-Maf in CD8 T cells leads to a strong repression of IFN-γ and IL-2 production, and to an increased expression of genes associated with T cell exhaustion, i.e., a dysfunctional state of T cells observed during chronic infections or in TILs (110,111). These genes include Bcl6, Pdcd1 (PD-1), Stat3, and Il10, among other genes that are also described to be regulated by c-Maf in other T cell subtypes. When c-Maf was knocked-out in tumor specific CD8 T cells, those cells had a much higher capacity to restrain tumor growth through increased IFNγ production and increased survival (109). Another study using B16 melanoma in mice confirmed that c-Maf was a major inducer of exhaustion associated genes, in cooperation with Prdm1 (Blimp-1). These two transcription factors had a compensative effect for the regulation of the transcription of many inhibitory receptors, such as PD-1, TIGIT, TIM-3, or LAG-3 (112). The double knockouts for c-Maf and Blimp-1 led to a better control of B16 growth, compared to single knockouts or WT mice (112). Frontiers in Immunology | www.frontiersin.org Tc17 cells are a rather rare subtype of CD8 T cells found in gastrointestinal cancers as well as other diseases ranging from fungal infection to bacterial colonization of the skin and response to influenza (113). A study on mouse and human Tc17 cells showed that c-Maf and RORγt were both essential for the development of these cells (113). Similar to what is observed in Th cells, c-Maf expression in CD8 T cells is associated with increased tolerogenic/noninflammatory functions, a mechanism, which is hijacked by the tumor microenvironment to favor immune escape and tumor development.

ROLE OF c-MAF IN INNATE LYMPHOID CELLS AND γδ T CELLS
Innate lymphoid cells (ILCs) are tissue-resident cells which lack antigen specificity and are preprogrammed for effector function, poising them for rapid cytokine production and for tissular immune priming. Similar to T lymphocytes, ILCs can be classified into three specialized functional subclasses, characterized by the expression of lineage-defining transcription factors and effector cytokines (114). ILC1s are defined by T-bet (Tbx21) and IFN-γ production. ILC2s are characterized by their high GATA3 expression and IL-5 and IL-13 secretion. ILC3s are mainly located in the intestinal lamina propria and are defined by RORγt expression. They play a crucial role in maintaining epithelial barrier integrity by expressing large amounts of IL-22 and IL-17A (115). While c-Maf has been shown to be expressed early in ILC specification (116), it is preferentially expressed in intestinal ILC3s, particularly in CCR6 − NKp46 + cells (17,117).
ILCs show functional and phenotypic plasticity in response to the environment (118). This is exemplified by the ability of ILC3s to upregulate T-bet, lose RORγt expression, and convert to IFNγ-producing ILC1s, also called "ex-ILC3s, " which are involved in intestinal pathology (119)(120)(121). Recent analysis of ILC transcriptional regulatory networks showed that c-Maf regulates the ILC3/ILC1 balance in the intestine (18). c-Maf directly promotes ILC3 identity by upregulating canonical type 3 c-Maf signature genes, common to NKp46 + ILC3s, CCR6 + ILC3s, IL-17-producing γδ T cells, and RORγt + Tregs (17). Moreover, c-Maf indirectly sustains ILC3 identity by inducing CD127 expression, which strengthens the IL-7-dependent promotion of RORγt expression in ILCs (17). In turn, c-Maf directly represses ILC1 conversion by binding to the Tbx21 locus and inhibiting T-bet expression, and by restraining the type 1 chromatin accessibility landscape (17). Interestingly, this direct binding of c-Maf at the Tbx21 locus was not observed in Th1 cells and seems to be ILC-specific (67).
IL-17-producing RORγt + γδ T (Tγδ17) cells are innatelike γδ T cells, functionally preprogrammed to produce IL-17, positioning them as the main IL-17-producer in the steady state intestine and in various inflammatory diseases (122,123). A recent study showed that the collaboration between c-Maf and RORγt is required for Tγδ17 induction and maintenance (19). In a similar dynamic to that observed in ILC3s, c-Maf sustains Tγδ17 differentiation both by supporting activating locus modifications and by antagonizing TCF1 accessibility, a negative regulator of Tγδ17 differentiation.
c-Maf therefore emerges as a global type 3 innate cell insulator, which sustains the expression type 3 genes, while constraining the acquisition of a type 1 phenotype by directly repressing type 1 genes, such as Tbx21 and Tcf7, and by restraining chromatin accessibility. This feature distinguishes type 3 innatelike lymphoid cells from the adaptative Th17 cell subset, in which lineage-defining transcription factor expression and chromatin remodeling are regulated indirectly by c-Maf through negative regulation of the IL-2 signaling (67). c-Maf thus seems to occupy a higher position in the regulatory network controlling type 3 innate cells compared to that of Th17 cells.

PERSPECTIVES: MANIPULATING c-MAF EXPRESSION TO TREAT PATIENTS?
In all T cell subtypes, c-Maf plays an important role in regulating tolerance and homeostasis. Targeting c-Maf expression or function could be a good opportunity to enhance or repress a given immune response. As c-Maf is overexpressed in 50% of multiple myeloma, several laboratories developed strategies to find some potent inhibitors, which could be used to enhance the immune response, in an anti-cancer setting for instance. USP5 is important to regulate the degradation of c-Maf (124). Molecules targeting this pathway, such as Mebendazole, showed some potent anti-myeloma activity by inhibiting the USP5/c-Maf axis (125). Another possible way to inhibit c-Maf activity is by preventing its dimerization. This could be achieved by the design of specific peptide able to destabilize homodimers of c-Maf (126). An older study described that glucocorticoids are potential inhibitors of c-Maf in multiple myeloma (127). However, using these compounds to manipulate the immune system is rather inappropriate, as glucocorticoids strongly dampen it. Lowering c-Maf expression at the cellular level before T cell transfer is also one possibility, with the use of miRNA or similar technologies.
Increasing c-Maf expression in Tregs would be of crucial interest to treat autoimmune diseases, especially in the gut where c-Maf is important in several T cell subtypes. The c-Maf expressing RORγt + Foxp3 + Treg population is also present in human colon but do not show any variation in proportion of total CD4 T cells when comparing colon from patients with Crohn's disease and healthy tissues (128). How to manipulate and increase c-Maf expression and RORγt + Tregs remains a challenge that requires further studies. The microbiota appears to play a central role in the regulation of this regulatory population (97,101,128,129) and manipulating c-Maf expression might have dramatic impact on microbiota equilibrium. Thus, a better understanding of the c-Maf-driven Treg/Tfr/microbiota interplay would be required prior to envisage c-Maf-targeted immunotherapy.

AUTHOR CONTRIBUTIONS
CI, HH, FA, and GV wrote the text and designed the figures.

FUNDING
This work was supported by the University of Lausanne, grants from the Swiss Cancer League (GV: 3679-08-2015), the Max Cloëtta Foundation (GV), the Swiss National Foundation (GV 310030_182680), the European Regional Development Fund (ERDF), the Walloon Region (Wallonia-Biomed portfolio, 411132-957270), and the Fonds Jean Brachet, and a research credit from the National Fund for Scientific Research, FNRS, Belgium. FA is a Research Associate at the FNRS, HH was supported by a Belgian FRIA fellowship.