Mechanisms of tolerance

Go to http://www.immunologicalreviews.com to watch interview with Guest Editor Jeffrey A. Bluestone Productive immune responses to foreign antigens are necessary to protect individuals against pathogens and allow survival in the external environment. The immune system has evolved to achieve efficient immunity against foreign antigens without mounting detrimental responses to self-antigens, thus preserving the integrity of the individual's own tissues. This original paradigm of immune tolerance has developed to include not only the spontaneous absence of self-targeted responses in steady state or after an inflammatory response but also the pharmacologically induced state of non-responsiveness to foreign antigens deliberately introduced into individuals, such as organ grafts. Thus, tolerance can be defined as a lack of reactivity to self-antigens or foreign tissue antigens in an organ graft achieved without the need for long-term immunosuppression while retaining immune competence and reactivity to all other foreign antigens. From an evolutionary point-of-view, the critical importance of immune tolerance is illustrated by the multiple non-redundant mechanisms that are in place to establish this immunological state. Nevertheless, the challenges in attaining and maintaining this state are underlined by the prevalence of autoimmune diseases, the as-of-yet absence of routine tolerogenic therapies in organ transplantation, and the necessity to strike a balance between protecting the individual from autoimmunity while preserving responses against cancerous cells and foreign antigens. Central tolerance mechanisms eliminate many potentially autoreactive T cells in the thymus through negative selection and the preferred selection of thymocytes with high affinity to self toward the regulatory T-cell lineage. Yet, it is clear that potentially self-reactive thymocytes can pass this checkpoint and become a part of the peripheral T-cell repertoire. Peripheral tolerance mechanisms limit the activation of mature self-reactive T cells once they have exited the thymus. Autoreactive T cells may be tolerized by clonal deletion, anergy, or immunological ignorance. Their responses can be controlled by regulatory T cells (Tregs), B cells, or other regulatory leukocytes. It is notable that the concept of immunological tolerance was introduced more than 50 years ago by Medawar et al., well ahead of the characterization of basic features of immune responses as we know them now, such as the major histocompatibility complex (MHC)-restricted recognition of antigenic peptides by the T-cell receptor (TCR), the stages of thymic selection, or the existence of T-cell subsets and lineages, such as T-helper 1 (Th1), Th2, Th17 and Tregs. As narrated by Wood et al. in this issue (1), Medawar et al. observed that injection of allogeneic tissue into neonates led to immune tolerance to subsequent allografts (2). This led to the seminal concept that exposing the developing immune system to new antigens, including alloantigens, could induce specific immunological tolerance. The impact of this work has been tremendous and provided the basis to determine the mechanisms of tolerance and develop tolerogenic approaches in the clinic. Interestingly, plasticity of the immature immune system was noted as one of the keys to the success of the neonatal tolerance induction strategy in these early studies. The recent appreciation that there is also a certain degree of plasticity in mature T-cell subsets and dendritic cells (DCs) in the periphery suggests that the multi-level plasticity of the immune system is critical for the establishment of tolerance. Many other parameters have been proven to determine whether immune tolerance can be achieved (Fig. 1), including the route or location of antigen presentation (for example, oral administration of antigen or immunologically privileged sites favor tolerance), the source and form of antigen (dose, with or without adjuvant/inflammation, tissue antigen, etc.), the antigen-presenting cells (APCs) (cell types involved, e.g. B cells, monocytes/macrophages, DCs, non-professional APCs, etc., and their state of maturity and activation), cells responding to the stimulation (naive versus memory, effector versus regulatory T cells, size and avidity of the repertoire, etc.), cytokine milieu, the role of T-cell intrinsic signals versus extrinsic regulation, the genetic background, the environmental milieu, and the plasticity of tolerance (plasticity of T-cell subsets, role of innate immunity which can break tolerance, acknowledgment that tolerance itself is a dynamic state). Reviews in this volume of Immunological Reviews discuss many of these parameters and their influence on the induction or maintenance of immune tolerance. The individual reviews provide in-depth discussion of the data and contributions made in each area of tolerance. In this introduction, I summarize key elements in each review to achieve a coherent overall picture of tolerance. Multifactorial control of immune tolerance. This figure schematically represents the different genetic, molecular and cellular factors regulating immune tolerance. Both central tolerance and peripheral tolerance pathways are influenced by polymorphisms in autoimmunity susceptibility genes, regulatory cell subsets, T-cell intrinsic mechanisms of regulation, pathogenic cell subsets, and a combination of environmental and immunological factors that depend on the site of the immune response and immunological state of the individual. It is the balance of these inflammatory versus regulatory signals that ultimately determines the outcome of autoimmunity versus tolerance. The diversity of T-cell repertoire is insured by random recombination events taking place during thymic T-cell development. The flip side of coin is that these random events generate a large number of thymocytes bearing self-reactive TCRs. Central tolerance is achieved by the deletion of autoreactive thymocytes through negative selection. Thymocytes with a low or intermediate affinity are positively selected, whereas thymocytes with a higher affinity are negatively selected and undergo clonal deletion. Intriguingly, high-affinity interactions can also signal thymocytes to differentiate into the forkhead box protein 3 (Foxp3)-expressing Treg lineage, resulting in a Treg repertoire skewed toward self-recognition, as mentioned by Rudensky in this issue (3). During the last decade, important strides have been made in the understanding of central tolerance with the discovery of the role played by the transcription factor Aire (autoimmune regulator) in negative selection. Anderson et al. (4) reviewed the latest findings on the role of Aire in immune tolerance. Aire is selectively expressed by mTECs in the thymus and drives the expression of tissue-specific antigens (TSAs), thus inducing the negative selection of tissue-specific autoreactive T cells (5). The critical role of Aire in tolerance induction is illustrated by the multi-organ autoimmunity developing in patients and mice deficient for Aire. Aire is expressed on a subset of CD80hi MHC class IIhi mTECs and controls ectopic antigen expression (6). Of note, as reviewed by Manicassamy and Pulendran (7), targeting of MHC class II molecules to different thymic cell subsets showed that medullary DCs are the only cells sufficient to induce negative selection of CD4+ T cells and, as pointed out by Anderson et al. (4), this function may involve a collaborative effort with mTECs to provide TSAs. Thymic DCs and Hassall's corpuscles, which are groups of epithelial cells within the thymic medulla that produce thymic stromal lymphopoietin (TSLP) and condition DCs, are largely responsible for the induction of Foxp3+ Tregs, especially in humans (8). Thus, while Aire does not appear to affect the global development and homeostasis of peripheral natural Tregs, it may affect their repertoire and negative selection. The features of antigen presentation by different thymic APC populations and their role in the negative selection of CD4+ and CD8+ T cells specific for myelin antigens are reviewed by Goverman (9) in this issue of Immunological Reviews. Indeed, TCR-transgenic mice with specificity for myelin proteins have shown that autoreactive T cells with high affinity for their cognate antigens are usually deleted in the thymus during negative selective selection and low avidity cells than can escape thymic selection circulate in the periphery in a state of immunological ignorance, allowing the analysis of mechanisms governing tolerance to central nervous system self-antigens. These findings have relevance to the establishment of tolerance to tissue-specific antigens in the thymus that preclude the development of autoimmune diseases targeting the central nervous system, such as multiple sclerosis and its animal model experimental autoimmune encephalomyelitis (EAE). Additionally, it is likely that these mechanisms are not unique to the central nervous system antigens but are at play to induce central tolerance to other TSAs. As mentioned above, thymocytes that display high affinity for self-MHC complexes are subjected to clonal deletion, whereas thymocytes with low-affinity TCR are positively selected, progress to the single positive stage and are exported to the periphery. While this process allows the negative selection of many potentially autoreactive thymocytes, it is by definition dependent on the adequate presentation of self peptide-MHC complexes, both qualitatively (presentation of the correct peptide) and quantitatively (enough peptide must be presented to allow for strong interactions with the TCR). In the case of proteolipid protein (PLP) and myelin basic protein (MBP), two major components of myelin and targets of the autoimmune response, presentation is complicated by the fact that multiple isoforms are generated by alternative splicing, and that the abundance of different isoforms can vary between the thymus and peripheral tissues (9). For example, isoforms of MBP expressed in the thymus (Golli-MBP) only partially overlap with 'classical MBP' isoforms expressed in the central nervous system. In addition to the presence of the peripheral epitope sequence in the thymus, the neighboring sequences, the abundance of self-antigen/MHC complexes and post-translational modifications of the self-antigen all influence the efficiency of negative selection. In some cases, sequences of Golli-MBP overlapping the sequence of a classical MBP epitope outcompete the latter for binding with MHC molecules in the thymus, thus compromising clonal deletion of thymocytes specific for this classical MBP determinant. Peptides can bind the groove of MHC class II molecules in different ways or 'registers'. Some registers bind MHC molecules with high stability and affinity, resulting in abundant MHC-peptide complexes that can drive negative selection. Conversely, other registers result in poor binding to MHC molecules, allowing only for few and unstable MHC-peptide complexes at the surface of thymic APCs that are inefficient at inducing clonal deletion, as has been described for MBP and insulin in the NOD mouse (10, 11). Acetylation of some epitopes differs between Golli- and classical MBP and may also hinder the negative selection of thymocytes specific for classical MBP. Analysis of the thymic APCs responsible for efficient negative selection of thymocytes specific for a MBP peptide-MHC class II complex revealed that bone marrow-derived APCs were responsible for central tolerance of CD4+ T cells with this specificity after presenting MBP acquired from the host (12). It is tempting to speculate that the source of self-antigen could include transfer from neighboring mTECs, as discussed above. However, the extent of negative selection for this specificity greatly increased during the weeks after birth and mirrored the level of expression of MBP in the central nervous system, suggesting age-dependent tolerance to an exogenously derived and developmentally regulated antigen as well as the source of APCs (13, 14) as described in depth in the review (9). It was recently discovered that expression of Aire is not limited to the thymus and that a radio-resistant population of stromal cells express Aire in the spleen and lymph node (15). Importantly, expression of a model antigen in these extrathymic Aire-expressing cells (eTACs), express a distinct set of Aire-driven antigens. Moreover, expression of self-antigens by these cells led to the proliferation and deletion of adoptively transferred TCR-transgenic CD8+ T cells recognizing the antigen. Finally, the eTACs express high levels of MHC class II molecules, suggesting that they may be able to interact with CD4+ T cells and perhaps participate in intrinsic or extrinsic regulation of CD4+ T-cell responses, raising the possibility that these cells complement each other by inducing central and peripheral tolerance to a non-overlapping set of autoantigens. This could represent a novel and additional function of costimulatory pathways, which are critical for the induction of immune tolerance as reviewed by Dong and Bluestone in this volume (16, 17). The notion that immune responses could be controlled by specialized suppressor T cells was revived by seminal studies from Sakaguchi et al. less than 15 years ago. These studies led to the identification of CD4+CD25+Foxp3+ Tregs and their crucial role in peripheral tolerance, as reviewed by Rudensky in this issue (3). Of note, other Foxp3+ and Foxp3− regulatory T-cell populations have been characterized and also play a role in certain forms of tolerance, as mentioned by Weiner and Roncarolo et al. (18, 19). Tregs efficiently suppress T-cell responses to self and foreign antigens in vitro and in vivo. The discovery of lineage-specific transcription factor Foxp3 greatly contributed to the identification of CD4+CD25+Foxp3+ T cells as a distinct T-cell subset. The development of autoimmune diseases in mouse models ('scurfy' mice) and IPEX (immunodysregulation, polyendocrinopathy, enteropathy, X linked) patients with deficient or mutant Foxp3 was shown to be due to a defective Treg population and established that these cells played a unique role in keeping autoreactive T cells in check. Importantly, the role of Foxp3 in Treg development and function is not limited to the neonatal period, when the immune system is still developing, but is important throughout life, as demonstrated by the rapid development of lethal multi-organ autoimmunity after acute depletion of Foxp3+ Tregs in adult mice (20). A similar outcome was observed when ablation of mature Tregs was performed in germ-free mice, suggesting that Tregs control the response of peripheral T cells to self-MHC complexes and that failure to do leads to fatal autoimmunity even in the absence of commensal microbiota (21). On a molecular level, thymocytes are pre-determined into the Treg lineage prior to Foxp3 expression, but Foxp3 stabilizes the molecular features of Treg precursors and antagonizes other competing signals, such as retinoid-related orphan receptor-γ (RORγ)-mediated interleukin-17 (IL-17) expression, by directly activates or represses the transcription of multiple genes. Foxp3 expression is not only necessary to establish Foxp3-dependent transcriptional program during Treg differentiation but also for its maintenance in the periphery raised the question of the possible instability of Tregs in vivo. Both Rudensky and ourselves, among others, addressed this question and obtained divergent data in two separate lines of reporter mice (3, 16), although the reasons for this discrepancy are not fully understood. Rudensky and colleagues found that Foxp3 expression was stable in Tregs in steady state and during inflammation except under conditions of depleted IL-2 (22). Conversely, we uncovered a significant population of T cells with unstable Foxp3 expression (exFoxp3 cells) (23). TCR repertoire analysis revealed that these cells likely emerged from both unstable Tregs and effector T cells transiently upregulating Foxp3. ExFoxp3 cells were enriched in inflammatory conditions and self-reactive exFoxp3 cells were pathogenic in vivo. The stability of Tregs may be related to the regulatory elements recently discovered within the Foxp3 locus. Indeed, several conserved non-coding sequences (CNS) could be involved in the expression of Foxp3. The influence of CNS1, CNS2 (aka TSDR), and CNS3 elements on Foxp3 expression in the thymus and periphery is beginning to be described, but the transcription factors and immune conditions allowing their potentiation of Foxp3 induction are still largely unknown (3). Finally, it has become increasingly apparent that Tregs play a key role in suppression of immunity in the tumor setting. As Allison and colleagues summarize in their review (24), suppressive activities mediated by Tregs present at the tumor site block concomitant immunity and promote tumor growth. In fact, the combination of regulatory cells and suppressive factors, including anti-inflammatory molecules such as indoleamine 2,3-dioxygenase (IDO) and transforming growth factor-β (TGF-β), produced by the tumors themselves provide an 'immune privileged' site that mitigates immunity and enhances Treg activity. Costimulation pathways greatly influence the outcome of T-cell stimulation and play a central role in immune tolerance. The definition of costimulatory pathways is not limited to the surface receptors that potentiate TCR signals, such as CD28 or inducible costimulator (ICOS), but also including immunoregulatory receptors that belong to the same family as immunostimulatory receptors, such as cytotoxic T-lymphocyte antigen-4 (CTLA-4) and programed death-1 (PD-1), and the soluble factor IL-2, which is intimately linked to and has overlapping functions with some costimulatory receptors. In this issue of Immunological Reviews, contributions by Dong, Malek and Bluestone address the role of costimulation in tolerance (16, 17, 25). Bour-Jordan et al. review the cellular events controlled by CD28 and the signaling pathways downstream of CD28 (16). In addition, the intrinsic end extrinsic pathways of tolerance controlled by CD28 costimulation are reviewed focused on its role in T-cell deletion and/or anergy, its critical role in the development and survival of CD4+Foxp3+ nTregs (26, 27), and overall role in immune homeostasis. Conversely, the related inhibitory receptor CTLA-4 can also affect tolerance in intrinsic and extrinsic manners by tuning the immune system toward immunoregulation. The fulminant lymphoproliferative disease developing in CTLA-4-deficient mice dramatically illustrates the critical role of CTLA-4 in T-cell tolerance and homeostasis. CTLA-4 has an intrinsic role in the control of T-cell responses and seems to be involved in both the control of homeostatic proliferation triggered by self-MHC molecules and the termination of immune responses to self-or foreign antigens. A simplified view of CTLA-4 intrinsic function is that it is more critical for the induction phase than the maintenance phase of tolerance to self-antigens, but the requirement for CTLA-4 in tolerance depends on the biological context (28). The molecular mechanisms of CTLA-4 inhibitory effects are still unclear but may involve competition with CD28 for the CD80/CD86 ligands and blockade of TCR proximal and distal signaling. In addition, the biology of CTLA-4 is complicated by the existence of multiple splice variant isoforms, including a ligand non-binding form as well as a soluble, secreted variant (sCTLA-4), which may be differentially involved in distinct aspects of CTLA-4 function. The genetic association of polymorphisms in the CTLA-4 gene emphasizes the importance of CTLA-4 in tolerance with susceptibility to multiple autoimmune diseases in mice and humans (29). Importantly, as highlighted by Allison and colleagues in their review (24), CTLA-4 expression of T effector cells shut down immunity and can lead to tumor progression. In fact, clinical trials, driven by this group's research efforts, have shown that a CTLA-4 antagonist, anti-CTLA-4 mAb, can enhance anti-tumor activity in the melanoma setting. Finally, CTLA-4 is critical in the extrinsic control of tolerance. CTLA-4 exerts this effect by playing an important role in the suppressive function of Tregs and by affecting the function of DCs through CTLA-4-induced down modulation of B7 expression and production of the immunosuppressive enzyme IDO (30, 31). PD-1 is another member of the CD28 family that is expressed on activated T and B cells and functions as a negative regulator crucial for the maintenance of tolerance in peripheral tissues, as reviewed by Dong and Bluestone (16, 17). PD-1 participates in central tolerance in the thymus and has been shown to be important for both negative and positive selection. In addition, PD-1 is critical for the maintenance of peripheral tolerance and the prevention of autoimmune diseases, including diabetes and EAE (28). Collectively, these data suggest a model in which CTLA-4 and PD-1 play complementary and non-overlapping roles in peripheral tolerance, with CTLA-4 predominantly controlling T-cell responses early in lymphoid organs and PD-1 acting at later stages and in peripheral tissues (32). One important feature of the ligands for PD-1, PD-L1 and PD-L2, is their high expression on somatic tissues, including many tumors (24). The molecular basis of PD-1 is highlighted in the reviews and, like anti-CTLA-4 mAbs, evidence is highlighted demonstrating the therapeutic potential of anti-PD-1 mAbs in tumor immunotherapies (24, 33). Malek and Bluestone review the control of tolerance by the cytokine IL-2 (16, 25). IL-2 is critical for T-cell expansion but also has a non-redundant tolerogenic function as evidenced by the development of lymphoproliferation and multi-organ autoimmunity in mice and humans deficient for IL-2 or the IL-2 receptor (IL-2R). Polymorphisms in the IL-2 or IL-2R genes have also been associated with susceptibility with autoimmunity (34). IL-2 intrinsically promotes T-cell tolerance by favoring Fas-mediated apoptosis. Furthermore, although IL-2 is not strictly required for Treg induction in the thymus, IL-2 is the major cytokine controlling Treg homeostasis in the periphery (35). IL-2 is also important for Treg suppressive function independent of its role on their homeostasis and stability. Thus, IL-2 is required to maintain optimal numbers and competitive fitness in the periphery and control the size of the Treg 'niche'. Finally, Malek et al. (25) review the differences in IL-2 signaling and function between conventional (Tconv) and Treg cells. One important difference is that IL-2 signaling involves signal transducer and activator of transcription 5 (STAT5) in Tregs but STAT5, mitogen-activated protein kinase (MAPK) and phosphoinositol 3-kinase (PI3K) in Tconv. IL-2 directly promotes the transcription of Foxp3 through activation of STAT5, which binds to regulatory sites within the Foxp3 gene. Weak and transient IL-2 signaling appears enough to mediate most of IL-2-dependent activities in Tregs but not Tconv. This, in turn, could offer a therapeutic window to target Tregs but not Tconv by using carefully titrated doses of IL-2 (36). Additional costimulatory pathways can greatly influence immune responses and tolerance, as reviewed by Dong et al. (17). ICOS is a positive regulator involved in selective effector function of different Th cell subsets and important for the generation of CXCR5+ follicular T-helper cells (Tfh), a unique T-cell subset regulating germinal center reactions and humoral immunity (37). B7-H3 is expressed on many cells of lymphoid and non-lymphoid tissues and its binding to an unidentified receptor on activated T cells might provide a negative feedback mechanism for Th1-mediated responses (38). B7S1 (aka B7x or B7-H4) is also widely expressed in lymphoid and non-lymphoid tissues. B7S1 is a novel negative costimulator and regulates the threshold of T-cell activation (39). Dong et al. additionally review intracellular molecular mediators of tolerance (17), including the role of E3 ubiquitin ligases, Cbl-b and GRAIL, crucial regulators of T-cell tolerance involved in the T-cell-intrinsic induction of anergy (40) and Tregs. In addition, growing evidence reveals that tolerant T cells exhibit unique transcriptional features. Many tolerance-associated transcriptional factors have been identified, including Egr-2, Egr-3, Ikaros, CREM, p50, ZEB1, Blimp-1, Tob and Smads, and could be considered as biomarkers of the tolerant state. Several of these factors are transcriptional repressors, directly bind to the IL-2 promoter, and promote anergy by repressing IL-2 transcription. Finally, there is evidence for an epigenetic control of T-cell tolerance, which is not just a consequence of T cell quiescence but relies on mechanisms actively involved in the tolerance that include histone deacetylation, specific patterns of histone methylation, and DNA methylation at CpG sites. Although DCs have long been considered as the prototypical and most efficient APCs for induction of productive immune responses, it is now clear that DCs play an active and crucial role in the induction of tolerance. In fact, ablation of DCs in mice can result in spontaneous autoimmunity (41). As reviewed by Pulendran, Allison and Turka (7, 24, 42) in this issue of Immunological Reviews, many factors influence whether DCs become immunogenic or tolerogenic when they interact with T cells, including the DC subset, the maturation stage of the DC, and the exposure to microbial stimuli and/or other factors in the local environment. In steady state, DCs typically present an immature phenotype characterized by low expression of MHC class II and costimulatory molecules, and they promote tolerance by inducing T-cell anergy, deletion, or generation of Tregs (43). Maturation of DCs usually makes them immunogenic, but some forms of stimulation, including microbial stimulation, can activate DCs yet result in tolerogenic cells. Regarding subsets of DCs, plasmacytoid DCs, IDO+ DCs, and DEC205+ CD8α+ DCs are typically involved in tolerance to self-antigens in steady-state but can become immunogenic after activation. In tissues, such as intestine, lungs, and skin, CD103+ DCs have been associated with tolerance and likely participate in preventing overwhelming immune responses to many foreign antigens encountered at these sites. For example, as discussed by Weiner (19), CD103+ DCs are critical in the establishment of oral tolerance. Tolerogenic DCs are also found in immune privileged sites like the brain or eye, and migration of tissue-resident DCs into regional LNs promote tolerance to tissue antigens. Furthermore, exposure to anti-inflammatory factors can make DCs tolerogenic (44). Conversely, many tolerogenic DCs can exhibit functional plasticity and become immunogenic in response to pro-inflammatory stimuli. Indeed, as discussed by Pulendran, Allison, Turka, and Ferguson (7, 24, 42, 45), DCs express several receptors that can sense common components of foreign pathogens or endogenous injured/dying tissues. Thus, many of these pattern recognition receptors can recognize both pathogen-associated molecular patterns (PAMPs) and damage associated molecular patterns (DAMPs) (46). As reviewed by Turka et al., most Toll-like receptors (TLRs) (except TLR3) use the adapter protein myeloid differentiation factor 88 (MyD88) for signal transduction (42). TLR signaling through MyD88 represents an important impediment to transplantation tolerance through mechanisms that include induction of IL-12 production by APCs and interference with Treg function (47). While these receptors overwhelmingly activate DCs and make them immunogenic, they can also provide tolerogenic signals in given circumstances and/or cell subsets. For example, most TLRs promote IL-12 production by DCs and induction of Th1 pro-inflammatory responses (48). However, binding of some bacterial products to TLR2 favors DC-mediated induction of Th2 or Treg cells. Similarly, activation of TLR9 leads to immunogenic status for most DC subsets but induces IDO production and tolerogenic properties in plasmacytoid DCs. C-type lectin like receptors (CLRs) are other pattern recognition receptors expressed on DCs. Similar to TLRs, activation of CLRs usually promotes productive immune responses. However, antigen targeting to some CLRs, like DEC205, on DCs results in the induction of Tregs. Furthermore, activation of some CLRs, like DC-SIGN, by microbial products can result in either immunogenic or tolerogenic DCs. CLRs can participate in the uptake of necrotic and apoptotic cells, with distinct outcomes as reviewed by Ferguson et al. (45). Great progress has been made in the understanding of signaling pathways that confer tolerogenic properties to DCs, as reviewed by Manicassamy and Pulendran (7). Among these, the Wnt-β-catenin pathway is involved in the production of retinoic acid in intestinal DCs. β-catenin is constitutively active in intestinal DCs, and its deficiency leads to a reduction in enzymes inducing retinoic acid and a concomitant reduction in Foxp3+ Tregs. Finally, interactions of DCs with neighboring cells and the local environment are critical for the generation of tolerogenic DCs. In particular, induction of a tolerogenic state is facilitated by interactions with Tregs and natural killer T (NKT) cells as well as non-hematopoietic cells like intestinal and thymic epithelial cells. Commensal bacteria also favor tolerance by promoting the production of tolerogenic TSLP and TGF-β by intestinal epithelial cells, which in turn influence neighboring DCs. As reviewed by Turka et al. (42), other cells of the innate immune system participate in the outcome of immune responses toward immunity versus tolera