Generation of their clonal TCR is the first step in T cell development

In the thymus the T cell precursors – also called thymocytes – start to express the recombinase-activating gene (RAG) products and begin to rearrange their αβ TCR genes. T cell precursors that enter the thymus express neither CD4 nor CD8 and are, therefore, called double negative (DN) thymocytes. The DN precursors actively proliferate, undergoing approximately 20 cell divisions and assemble the TCR β chain. Only those DN cells that have successfully rearranged a TCR β chain will progress to the next stage and express CD4 and CD8 simultaneously. These immature CD4⁺CD8⁺ double positive (DP) thymocytes start to reassemble TCR α chains and express T cell receptors (TCR). The randomly rearranged TCRs expressed by the DP thymocytes collectively constitute the organism’s unselected TCR repertoire which is also called the germline repertoire. These thymocytes undergo processes of positive and negative selection. Less than 5% of them survive these selection events and are allowed to exit the thymus as naive mature T cells.

In addition to αβ T cells, other lineages including natural killer T (NKT) cells and γδ T cells also develop in the thymus. Although much less is known about thymic selection of γδ T cells there are profound differences in the thymic development of γδ T cells and αβ T cells. Ligand mediated selection events do not seem to be required for the selection of the γδ T cell repertoire. In contrast, instruction for particular effector functions occurs in the thymus for γδ T cells. Given the many current uncertainties about γδ T cell selection and NKT selection in the thymus, this section will deal exclusively with thymic selection of αβ T cells.

Thymocytes are positively selected for their ability to interact with self MHC molecules

The DP T cells interact with thymic cortical epithelial cells that present peptides derived from endogenous proteins bound to major histocompatibility complex (MHC) molecules. Recognition of self-peptide/MHC complexes is vital for the DP T cells for two reasons:

Moreover, recognition of a peptide/MHC complex via the TCR is necessary for the DP T cell to receive a survival signal. This is called positive selection (see Fig 19.2). Experiments have shown that T cells are positively selected by interaction with self-MHC to prepare them for subsequent activation by non-self-peptide/self-MHC complexes, indicating the biological benefit of positive selection.

Accordingly mice that do not express MHC-class II molecules lack CD4⁺ T cells and mice that do not express MHC-class I molecules lack CD8⁺ T cells.

Positive selection occurs predominantly in the thymic cortex

Specialized APC, the cortical thymic epithelial cells (cTECs) are pivotal for positive selection. cTECs differ in their antigen processing machinery (cathepsins, proteasome subunits) from hematopoietic antigen presenting cells. Given that the number of different peptides that can be presented by any particular MHC molecule is much smaller than the number of different TCRs that undergo positive selection, each peptide must be involved in positively selecting many different T cells. Accordingly, experiments have demonstrated that the peptide on which a TCR is positively selected does not need to share sequence similarity with the peptides recognized by that same TCR in the periphery.

Note that positive selection depends on the recognition of self-peptides bound to self-MHC. The TCRs that are positively selected based on low-affinity interactions with self-peptide/MHC form the selected repertoire that ultimately recognizes microbial antigens, to protect the organism from infectious diseases. This is one illustration for the flexibility of antigen recognition by T cells. The peptides that mediate positive selection in the thymus are also presented outside the thymus where they support survival of mature T cells and may also act as co-agonists that enhance T cell activation by agonist peptides.

An immediate question is why T cells which were selected based upon self-recognition usually do not cause damage to the organism. One answer is that the threshold for TCR signaling is lower in immature DP T cells in the thymus than in mature T cells in the periphery. Thus, DP T cells can respond to low-affinity interaction with peptide/MHC complexes that would not trigger mature T cells. One important regulator of this TCR signaling threshold is the microRNA miR-181a that regulates the expression of several phosphatases involved in TCR signaling. DP T cells express much higher levels of miR181a than mature SP T cells.

Lack of survival signals leads to death by neglect

DP T cells whose receptors have a very low affinity for the peptide/MHC complexes they encounter during their approximately 3–4-day lifespan in the thymic cortex do not receive survival signals and undergo apoptosis in the thymus (see Fig. 19.w2). This is called death by neglect. Accordingly, T cells do not progress beyond the DP stage in mice that have been manipulated to lack expression of MHC molecules. Analyses of the germ line repertoire of TCRs have revealed that this unselected repertoire contains more self-MHC reactive TCRs than expected by chance. In other words: co-evolution of TCR and MHC molecules has shaped the germline αβ TCR repertoire to favor the generation of receptors that can interact with self-MHC.

Fig. 19.w2 The correlation between avidity and thymocyte selection

The avidity of a T cell’s interaction with antigenic peptide presented by an APC will depend on the level of expression of the MHC–peptide complex [MHC + peptide] on the APC and both the affinity and surface expression of TCR on the T cell. [MHC + peptide] depends on the affinity of peptide for MHC and the stability of the complex once formed. Current evidence suggests that CD25⁺ regulatory T cells are selected in the thymus and have a relatively high affinity for MHC and peptide.

Thymocytes are negatively selected if they bind strongly to self-peptides on MHC molecules

After positive selection and commitment to the CD4 or CD8 lineage in the thymic epithelium, thymocytes express the chemokine receptor CCR7 and migrate towards the central region of the thymus, the thymic medulla, where the CCR7 ligands CCL19 and CCL20 are produced. In the thymic medulla the thymocytes are probed for another 4–5 days. T cells whose receptors have a high affinity for the peptide/MHC complexes encountered in the medulla are autoreactive and therefore, potentially dangerous. They receive death signals and undergo apoptosis. This is called negative selection (see Fig 19.2). Experimental evidence suggests that the majority of the DP T cells which were positively selected are later eliminated by negative selection.

The relevance of negative thymic selection can be demonstrated by neonatal thymectomy. Mice that are thymectomized within the first three days after birth develop an autoimmune syndrome including sialadenitis, diabetes, autoimmune gastritis and hepatitis.

A library of self antigens is presented to developing T cells in the thymus

The induction of central tolerance requires the presence of autoantigens in the thymus. This poses an obvious problem for thymic selection: Some autoantigens, e.g. insulin are expressed in a tissue-specific manner, are frequently called tissue-restricted antigens (TRAs). The question, then, is (how) do TRAs get into the thymus for presentation to developing T cells? Some might be brought into the thymus by immigrating antigen presenting cells but it is highly unlikely that this would yield a reliable representation of the organism’s TRAs. Moreover, developmentally regulated TRAs, e.g. antigens that are only expressed after puberty, would not gain access to the fetal thymus. The answer is that specialized cells in the thymus, the medullary thymic epithelial cells (mTECs) express proteins that are otherwise strictly tissue restricted. This has been called ectopic or promiscuous gene expression (Fig. 19.3).

Fig 19.3 Medullary thymic epithelial cells express and present tissue restricted antigens

Different antigen presenting cell populations induce thymic tolerance. Dendritic cells and other conventional APCs present phagocytosed or endogenously synthesized proteins to T lymphocytes. T cells that express receptors with high affinity for these ubiquitous proteins will undergo apoptosis. Medullary thymic epithelial cells (mTECs) express the transcriptional autoimmune regulator, AIRE and, therefore, possess the unique capacity of expressing tissue restricted antigens (TRAs) ectopically in the thymus. This allows for intrathymic clonal deletion of T cells expressing TCRs with high affinities for antigens that are expressed in a tissue restricted manner outside the thymus.

(Adapted from Sprent J, Surh CD. Nature Immunol 2003;4:303–304.)

MTECS express several hundreds or even thousands of functionally and structurally highly diverse antigens that represent almost all tissues in the body. Importantly, mTECs express not only tissue restricted antigens but also developmentally regulated antigens. Thus, gene expression in mTECs is uncoupled from spatial and developmental regulation. The exact mechanisms of this promiscuous gene expression are not yet fully understood. It has become clear that not all mTECs express all TRAs. Any particular TRA is expressed by less than 5% of mTECs.

Whereas mTECs are the only cells known to be capable of promiscuous gene expression, they are not the only cells important for negative selection. Thymic dendritic cells can take up TRAs expressed by mTECs and cross-present these TRAs to T cells. Recent intravital imaging studies have yielded the estimate that a thymocyte makes contact with approximately 500 dendritic cells during its sojourn in the thymic medulla. Although we do not have experimentally based quantitative estimates for thymocyte:mTEC contacts, the purging of self-reactive T cells in the thymic medulla is a remarkable achievement.

Subtle quantitative alterations in the thymic expression of TRAs can be consequential. Murine intrathymic expression levels of autoantigens including insulin and myelin antigens correlate inversely with susceptibility to autoimmune diseases, type I diabetes and experimental autoimmune encephalitis (EAE) respectively. Similarly, in humans genetic variants resulting in low levels of intrathymic insulin-expression are strongly associated with susceptibility to type I diabetes.

Qualitative variations in TRAs expressed in mTECs have also been associated with autoimmune disease models. Differential splicing or the expression of embryonic, rather than mature variants of myelin autoantigens have been associated with strain-specific susceptibility to EAE and may well play a role in the pathogenesis of multiple sclerosis in humans.

AIRE controls promiscuous expression of genes in the thymus

What enables mTECs to express a broad array of TRAs independently of spatial or developmental regulation? mTECs express the transcriptional regulator Aire (autoimmune regulator). Aire controls the expression of a large number of TRAs in mTECs (see Fig. 19.3).

Similarly, in humans, point mutations in the gene coding for Aire are the cause of the rare monogenic autosomal recessive autoimmune polyendocrinopathy–candidiasis–ectodermal dystrophy (APECED) syndrome. APECED is characterized by high titres of several different autoantibodies that cause disease, mainly in endocrine organs. Together these findings strongly suggest that the autoimmune manifestations are caused by diminished expression of TRAs in thymic mTECs due to the Aire deficiency. Aire associates with a large number of partners to form distinct complexes that impinge on different steps of transcription. The different functions of Aire’s partner proteins include chromatin structure and DNA-damage response, gene transcription, RNA processing, and nuclear transport. It is still unclear, however, how Aire controls promiscuous gene expression. Moreover, experimental evidence strongly suggests that additional yet unidentified transcriptional regulators must also be involved in promiscuous gene expression by mTECs.

Immunological ignorance occurs if T cells do not encounter their cognate antigen

Immunological ignorance is maintained as long as autoreactive T-lymphocytes do not enter the tissue in which the autoantigen that they recognize is expressed. The autoreactive naive T cells are not tolerized and can be activated upon recognizing their cognate antigen.

The importance of immunological ignorance was demonstrated in mice that express a transgene-encoded TCR which recognizes a peptide derived from the lymphocytic choriomeningitis virus (LCMV). These mice were bred with another transgenic strain that expressed the viral peptide on the surface of their pancreatic islet cells. Surprisingly, diabetes did not develop in the offspring even though in vitro their T cells could kill cells that displayed the viral peptide (Fig. 19.4). The T cells in these double-transgenic mice were therefore, not tolerant in vivo, instead they were ignorant of their target cells.

Fig 19.4 Immunological ignorance

Immunological ignorance was first demonstrated in transgenic mice. (1) One transgenic strain of mice expresses a viral (LCMV) antigen under the control of the rat insulin promotor (RIP) in pancreatic islets cells. These mice remain healthy. (2) Another strain expresses a transgenically encoded T cell receptor specific for the LCMV antigen (LCMV TCR). These mice also remain healthy. (3) F1 mice from a cross between the RIP-LCMV and LCMV TCR mice express the viral antigen in the pancreas and they have T cells that express the transgenic LCMV-specific TCR. The transgenic T cells are not deleted in the thymus of these mice and they respond appropriately to LCMV antigen in vitro but, the mice remain healthy, indicating that the naive LCMV-specific T cells do not get in contact with the LCMV antigen which is expressed in the pancreas. (4) When the F1 mice are infected with LCMV, the LCMV-specific T cells become activated in the secondary lymphatic organs by high concentrations of processed antigen and can then enter the pancreas and destroy the insulin synthesizing cells. Consequently, the mice succumb to immune mediated diabetes.

When the mice were infected with LCMV, the transgenic T cells became activated, invaded the pancreas and destroyed the islet cells. Consequently the mice succumbed to diabetes. Importantly, LCMV-infection did not cause diabetes in those mice that expressed the transgenic TCR but not the LCMV-peptide in the pancreas. Once activated by LCMV infection, the hitherto ignorant autoreactive T cells (autoreactive because the LCMV-derived peptide was transgenically expressed in the pancreas) acquired the capacity to migrate into their target tissue where they recognized and destroyed the LCMV-peptide expressing islet cells (see Fig. 19.4).

Self-reactive T cells and experimental autoimmunity

Mice that express a transgene-encoded TCR specific for a myelin antigen do not develop experimental autoimmune encephalitis (EAE, a mouse model for multiple sclerosis) spontaneously although more than half of their mature naive peripheral CD4⁺ T cells express the transgene-encoded TCR. Only upon immunization with myelin antigen in adjuvant do the mice develop EAE. This system can be probed even further. When the MBP-TCR transgenic mice are bred with RAG deficient mice, all T cells in the MBP-TCR⁺RAG^−/− mice express the transgene-encoded autoreactive TCR. Even then autoimmunity does not develop spontaneously as long as the mice are not immunized with the self antigen and are kept under sterile conditions.

Lymphocyte activation enhances their migration into non-lymphoid tissues

Upon activation in secondary lymphatic organs naive T cells acquire effector functions and express a different set of chemokine receptors and adhesion molecules which then enables them to enter other organs, particularly in the context of inflammation. Naive T cells lacking those surface molecules, however, are excluded from non-lymphoid tissues so that under normal conditions, potentially autoreactive T cells will ignore their antigens, thereby maintaining self-tolerance. Should such cells be activated accidentally they would pose a permanent threat to the organism. DCs that are activated during infection are likely to present not only microbial but also self-antigens. The T cells activated in this process would gain the capacity to enter tissues and thus have lost their ignorance. Consequently, additional mechanisms must ensure the maintenance of immunological self tolerance.

The amount of released self antigen critically affects sensitization

How does the immune system decide which autoreactive T cells may survive ignorantly and which need to be deleted? Antigen dose (Fig. 19.w1) and TCR avidity play a major role. One key experiment used two different strains of mice that expressed ovalbumin (OVA) specifically in the pancreas. One strain expressed OVA at low levels and the other expressed it at high levels. Only in the high-expressing strain was OVA presented to T cells in the draining lymph nodes, resulting in the deletion of adoptively transferred OVA-specific TC cells. In the low-expressing cells the OVA-specific T cells remained ignorant. Further in vitro assays revealed that the low-level OVA-expression was still sufficient to allow recognition and killing of the OVA-expressing pancreatic β cells by the OVA-specific CTL. Therefore, tissue restricted self-antigens need to be expressed at sufficiently high levels to be presented in the draining lymph nodes. These experiments also clearly showed that self-reactive T cells remain dangerous and poised to wreak havoc even if they are temporarily ignorant. In fact, destruction of pancreatic islet cells can induce the release of sufficient amounts of self-antigen to activate OVA-specific CTL in the OVA-low expressing mice.

Fig 19.w1 Abundant self antigens induce peripheral deletion

(1) Immunological ignorance predominantly works if the self antigen is present at relatively low concentrations. The antigen remains in the tissue in which it is expressed and is not transported into the draining lymph node under steady state conditions. (2) If tissue antigens are present at high concentrations they will be brought to draining lymph nodes by DC even under steady state (non-inflammatory) conditions. Under these conditions, the DC are not immunogenic but tolerogenic and induce apoptosis or anergy in T cells that recognize the self Ag.

Dendritic cells can present antigen in a tolerogenic manner

Experiments designed to analyse CD4 and CD8 T-cell responses to antigens that were expressed in a tissue restricted manner (e.g. exclusively in the pancreas or the skin) revealed that self-Ag specific T cells accumulated in the draining lymph nodes as a result of reaction with self-antigen transported by DCs. Depending on the information received from the DC in addition to peptide presentation, the TH cells may become

The critical importance of DCs for the maintenance of tolerance has also been shown in experiments in which conditional DC depletion in mature mice resulted in spontaneous autoimmunity.

Functional maturation of DCs, characterized by strong expression of MHC and co-stimulatory molecules is induced by microbial or self-derived stimuli, which are sometimes called danger signals (Fig. 19.5). In the absence of such stimuli immature DCs express MHC and costimulatory molecules at low levels and antigen presentation induces T cell anergy or deletion depending upon the expression of high or low levels of self-antigen respectively (see Fig. 19.w1). The signals that induce tolerogenic DC maturation as well as the tolerogenic interactions between DC and T cells are only incompletely understood. Still there are some clear candidates. Several molecules have already been identified that are necessary for tolerogenic DC:T cell interactions. These include surface molecules such as E-cadherin, PD-1 L, CD103, CD152 (CTLA-4) and ICOS-L (CD275) and cytokines, including IL-10 and TGF-β.

Fig 19.5 Critical role of APC for T-cell tolerance

Antigen presenting cells (APC) are critical for the decision between T-cell activation or tolerance. Upon antigen uptake the APC become activated if they sense exogenous or endogenous danger signals. Such classically activated APC present antigen in an immunogenic manner to T-lymphocytes and induce T-cell proliferation and effector functions. Such immunogenic DC, which have matured under inflammatory conditions (i.e. they have received signals via their TLRs or other pattern recognition receptors), upregulate co-stimulatory molecules (e.g. CD86, CD40), MHC class II molecules and CCR7. These changes increase their capacity to present antigen to T cells in an immunogenic manner. They also activate NFκB and express proinflammatory cytokines (e.g. IL-12) which instruct T-cell effector functions. If, in contrast, the APC do not receive danger signals they can still undergo homeostatic maturation. Such APC present antigen in a non-immunogenic manner to T cells and induce clonal deletion or anergy. Tolerogenic DC which are either immature or have matured under steady state conditions, express only low amounts of co-stimulatory and MHC-II molecules and do not secrete proinflammatory cytokines.

Tolerogenic DCs mature under steady-state conditions

DC maturation under steady-state conditions can be triggered by disrupting DC-DC adhesion which is mediated by E-cadherin. Similar to DC maturation induced by microbial products, the disruption of DC-DC contacts induces upregulation of MHC class II, co-stimulatory molecules and chemokine receptors and this process requires the activation of β-catenin. In contrast to DCs that have matured upon sensing microbial products, the DCs that have matured under steady-state conditions do not produce pro-inflammatory cytokines. Consequently, when they present antigen to naive TH cells these DCs induce regulatory T cells rather than effector T cells. Activation of the Wnt-β-catenin signaling pathway in DCs is required for the expression of interleukin-10, transforming growth factor-β and retinoic acid-metabolizing enzymes that are important for the induction of Treg differentiation and the suppression of effector T cells. Mice that lack β-catenin expression selectively in DC develop more severe inflammation in a mouse model of inflammatory bowel disease. It is likely that other, currently unknown triggers and signaling pathways are also relevant for steady-state DC maturation.

Moreover, DCs receive instructive signals from tissue cells. Epithelial cells, for example, produce a host of molecules capable of instructing DC development towards immunogenic or tolerogenic effector functions. At steady state conditions, the production of thymic stromal lymphopoietin (TSLP), IL-25, and IL-33 dominates and these cytokines favor tolerogenic DC development. Responding to tissue trauma, the epithelial cells switch to producing IL-1, IL-6, TNF-α, and type I interferons which strongly favor immunogenic DC maturation. Several additional triggers are known that induce DCs to become tolerogenic. These include the uptake of apoptotic cells and certain immunosuppressive cytokines such as IL-10 and TGF-β1 or substances such as prostaglandin E2 or the vitamin D3 1α,25-dihydroxy-metabolite.

DC surface receptors involved in promoting tolerance

DCs can suppress T-cell responses by upregulating indoleamine 2,3-dioxygenase (IDO) which catabolizes tryptophan to kynurenines that are toxic to T cells. IDO-expression in DC can be induced by CD152 (CTLA-4), a co-inhibitory molecule that is constitutively expressed by regulatory T cells and binds CD80 and CD86 which are expressed by the DC.

Certain ligands for toll like receptors (TLR) expressed by dendritic cells can trigger tolerogenic DC development. This is exploited by several pathogens. For example the parasitic helminth Schistosoma mansoni produces lysophosphatidylserines that bind TLR2 to induce tolerogenic DCs which will help to induce regulatory T cells. Still, there is not one single switch for the induction of activating or tolerogenic DCs. Instead, DC fate decision requires the integration of a multitude of sometimes contradictory signals and more than one receptor is usually involved in dealing with microbial or self antigens. For example, targeting antigens in the absence of further DC-maturation stimuli to the endocytic receptor DEC-205 (CD205) on DCs has resulted in tolerance induction in vivo. The same targeting in combination with additional maturation stimuli for the DC has resulted in strong CD4, CD8 and humoral immune responses towards the antigen targeted to CD205. How exactly a DC is instructed to present antigen alongside stimulating or tolerogenic signals is still poorly understood at present. If one considers that during infections or other forms of tissue damage that trigger DC maturation both self-antigens and microbial antigens will be presented by the same dendritic cell it is difficult to understand how the activation of autoreactive T cells is avoided in these situations. Since the vast majority of people never develop autoimmune disease throughout their lifetime despite repeated infections, perhaps it is asking too much of dendritic cells to have the crucial decision whether a T cell response is initiated or curbed depend solely on the DCs’ maturation stage.

The transcription factor FoxP3 controls Treg development

Regulatory T cells express the transcription factor forkhead box P3 (FoxP3), a member of the forkhead/winged-helix family of transcription factors.

The same manifestations occur in FoxP3-null mice. Both scurfy mice and FoxP3-null mice are deficient in CD4⁺CD25⁺ regulatory T cells. In normal mice only CD4⁺CD25⁺ peripheral T cells and CD4⁺CD25⁺ thymocytes express FoxP3. Other T cells, resting or activated, do not express FoxP3. Forced expression of FoxP3 can convert CD4⁺CD25⁻ T cells into CD4⁺CD25⁺ T cells that can suppress the activation of other T cells. Thus, FoxP3 is essential for Treg differentiation and maintenance and is important for Treg effector functions. Consequently, the coexpression of CD4, CD25 and FoxP3 is widely used to identify murine Treg cells.

Defects in FoxP3 result in multi-system autoimmune diseases

Since the discovery of FoxP3 a multitude of studies in animal models of autoimmunity proved that a deficiency in CD4⁺CD25⁺FoxP3⁺ Treg can accelerate the development or increase the severity of autoimmune disease. Conversely, in some models, disease could be prevented or even reversed through adoptive transfer of Treg.

Similarly, Treg isolated from human peripheral blood can suppress T cell proliferation and cytokine production in vitro. The human counterpart of the scurfy mutation is the IPEX (immune dysregulation, polyendocrinopathy, enteropathy, X-linked) syndrome mentioned above, which is caused by defects in FoxP3 expression resulting in a lack of Treg and has similar clinical manifestations to those observed in scurfy mice. IPEX patients suffer from autoimmune diseases, most prominently autoimmune diabetes (type I diabetes, T1D), thyroiditis, hemolytic anemia, and inflammatory bowel disease and allergic manifestations such as eczema. More than 90% of IPEX patients die from autoimmune diabetes at an early age. This demonstrates that (almost) all individuals harbor potentially diabetogenic autoreactive T cells in their peripheral repertoire. Usually, in healthy people, these autoreactive T cells are controlled by Tregs.

In contrast to murine T cells, human T cells express FoxP3 readily and transiently upon TCR signaling. Most of these cells do not possess immunosuppressive capacities. Thus, FoxP3 expression is not a reliable marker for human Treg cells. In fact, the stability of FoxP3-expression in natural Tregs is currently still a matter of debate.

Selection of nTregs is partly related to the affinity for antigen/MHC

Double transgenic mice that express both a transgene-encoded antigen and a transgene-encoded TCR that recognizes that antigen are instructive. In such mice a large percentage of transgenic T cells will develop into nTregs if either:

In contrast, if the TCR’s affinity for the transgenic peptide is low or if the transgenic peptide is present in low concentration in the thymus, few T cells bearing the transgene-encoded TCR will develop into nTregs (Fig. 19.w2). It is currently a matter of debate if these findings reflect positive selection of T cells with strongly self reactive TCRs to the Treg lineage or the resistance of FoxP3⁺ thymocytes to negative selection.

Repertoire analyses have revealed considerable overlap between the TCR repertoires of FoxP3⁺ Treg cells and conventional FoxP3⁻ T cells. Therefore, the TCR’s affinity for self-peptide/MHC is not the only determinant for Treg selection in the thymus. Co-stimulatory signals from the antigen presenting cells are critically important as demonstrated by the fact that mice deficient in costimulatory ligands or receptors such as CD28, CD80, CD86, CD40, or LFA1 (CD11a/CD18) have massively reduced numbers of Tregs.

IL-2 is required for the development of Tregs

The Treg marker CD25 is the IL-2R α chain and is critically important for the development and maintenance of Treg; FoxP3 expression is supported by IL-2. Mice that lack either IL-2, CD25, CD122 (the IL-2R β chain), or CD132 (the γc chain which is shared by the IL-2R, IL-4R, IL-7R, IL-9R, IL-21R, and IL-15R), suffer from lymphocytic infiltration of internal organs and autoimmune disease manifestations such as hemolytic anemia. These mice also display a profound reduction of FoxP3⁺ Tregs. Similarly, CD25 deficiency in humans is clinically indistinguishable from the IPEX syndrome. Moreover, administration of a neutralizing antibody against IL-2 to neonatal mice results in a profound reduction of FoxP3⁺ Tregs and autoimmune disease manifestations. Therefore, IL-2 is essential for the development and maintenance of Treg cells.

In vitro assays of Treg effector functions

The typical in vitro assay for Treg effector function consists of co-culturing Treg cells with effector T cells (responder cells). Usually the cells are stimulated with anti-CD3 antibodies and the proliferation of the responder cells in the presence or absence of Treg cells is measured. In addition other effector functions, e.g. cytokine production, of the responder cells can be assayed. These assays can be performed either with or without APC. The former use soluble anti-CD3, the latter work with plate-bound anti-CD3. In the absence of APC, Treg suppress the proliferation of CD4⁺ effector T cells only weakly, indicating that APC are an important target for Tregs. On the other hand, Treg suppress CD8⁺ effector T cell responses effectively even in the absence of APC, indicating that Treg also exert direct effects on responder T cells. Therefore, the two types of in vitro assays, with or without APC, measure different aspects of Treg function.

In vivo analyses of Treg effector functions

Treg function in vivo can be analyzed by experiments in which Tregs are either depleted or transferred. The complete absence of FoxP3⁺ Treg cells results in catastrophic autoimmunity as evidenced by IPEX syndrome in humans and the scurfy mice. In vivo depletion of Tregs enhances immune responses indiscriminately. This includes not only responses to autoantigens but also enhanced responses towards pathogens and tumors. Correspondingly, the adoptive transfer of CD25⁺CD4⁺ T cells can suppress autoimmunity or allergy and helps to initiate and maintain transplant tolerance. Depletion of Tregs in vivo has been difficult to achieve during ongoing immune responses for want of a specific surface marker. This problem has recently been overcome by the generation of transgenic mice which express the primate receptor for diphtheria toxin under the control of the FoxP3 promoter. In such mice, FoxP3⁺ Treg can be ablated even during ongoing immune responses.

Tregs secrete immunosuppressive cytokines

One possible mechanism of immunosuppression by Tregs would be the secretion of immunosuppressive cytokines (see Fig. 19.8). Indeed, three inhibitory cytokines, IL-10, TGF-β, and IL-35 are produced by Tregs and important for Treg development or effector function. Importantly, IL-10 and TGF-β are not exclusively produced by Tregs but can be produced by many different cell types. Moreover, most in vitro studies found IL-10 or TGF-β produced by Treg non-essential for Treg mediated suppression.

IL-10 is a potent suppressor of macrophage and T cell effector functions and critically important to dampen immune responses. Many cell types including TH2 and TH17 cells normally produce IL-10, and some TH1 cells start to produce it following chronic antigen stimulation. Therefore, the production of IL-10 by Treg is not critically required in many experimental settings. Mice that lack IL-10 expression specifically in Treg do not develop autoimmune disease manifestations spontaneously. Nevertheless, experimentally induced airway hypersensitivity is more severe in these mice than in their wild type littermates and Treg produced IL-10 seems to be most critical for the control of mucosal immune responses to environmental stimuli. The importance of Treg produced IL-10 seems to depend on the triggers and localization of the immune response.

TGF-β has immunosuppressive functions and is critically required for the differentiation of precursors into Treg cells in vivo and in vitro. TGF-β is also necessary to maintain FoxP3 expression of nTreg cells and thus for Treg cell homeostasis. In contrast, the relevance of Treg cell produced TGF-β as a mediator of Treg effector function in vivo remains unproven. In one particularly instructive experimental system Treg from TGF-β1-deficient mice were able to inhibit colonic inflammation that was induced by adoptive transfer of FoxP3-negative T cells. Importantly, administration of an anti-TGF-β mAb abrogated suppression of colitis mediated by TGF-β1-deficient Treg. These results show that TGF-β is absolutely required for suppression of colitis, but does not need to be produced by Tregs.

IL-35 is a heterodimeric member of the IL-12 family. It consists of an Ebi3 (Epstein–Barr virus-induced gene 3) subunit and p35, which is also known as IL-12. In mice IL-35 is selectively expressed by Treg and required for their optimal effector function.

The relevance of IL-35 for human Treg effector functions is less clear. Many different human cell types express p35. Resting human Tregs do not express Ebi3. Together, these findings make it unlikely that IL-35 is relevant for human Treg. However, IL-35 can induce naive human CD4⁺ T cells to produce IL-35 and acquire immunosuppressive capacities without expressing FoxP3.

Tregs also secrete other immunosuppressive soluble mediators, including galectin 10, that may be important for Treg effector functions.

Tregs can deplete IL-2

One important effect of Tregs on responder T cells is to inhibit the induction of mRNA for cytokines, including IL-2. Importantly, the addition of exogenous IL-2 does not rescue IL-2 mRNA production in the responder cells.

The consumption of IL-2 by Tregs can induce IL-2-deprivation mediated apoptosis of effector T cells. However, Tregs can block autoimmune disease even in IL-2 deficient mice and the importance of IL-2 consumption for Treg effector function is still a matter of debate.

Matters are complicated by the fact that IL-2 is important for the apoptosis of antigen-activated T cells. Hence, the lack of IL-2 or IL-2 signaling does not only result in Treg deficiency but also obliterates peripheral clonal deletion of receptor-activated T cells Therefore, the clinical manifestations of the IL-2 deficiency syndrome cannot be attributed completely to the loss of Tregs.

Cytolysis

Treg are cytotoxic to different cell types, including dendritic cells, CD8⁺ T, NK, and B cells. In different experimental systems Treg cytotoxicity depended on granzyme B-, perforin-, or Fas-FasL-dependent pathways. Treg-mediated DC death in lymph nodes has been demonstrated in vivo and limits the onset of CD8⁺ T cell responses.

Modulation of DC maturation and function

Some of the in vitro experiments to assess the suppressive function of Tregs are performed by stimulating the responder T cells with plate-bound antibodies in the absence of antigen presenting cells. Data from such experiments indicate that Tregs can act directly on responder T cells. In vivo, however, the modulation of DC/T interactions is also an important mechanism of Treg-mediated immunosuppression. Depletion of Treg cells in vivo results in increased DC maturation and elevated numbers of DC. Direct interactions between Treg cells and DCs have also been observed by intravital microscopy. During these interactions Treg cells can exert different effects on DC.

Direct cytolysis. In vivo studies have shown that Treg cells can induce the death of antigen-presenting DCs in lymph nodes in a perforin-dependent manner. This depletion of DCs limited the onset of CD8⁺ T cell responses.

Treg cells can also modulate maturation and function of DCs and can probably also modulate the function of mononuclear phagocytes. In vitro, Treg cells can downregulate the expression of co-stimulatory molecules, including CD80 and CD86 by DC. By averting co-stimulatory interactions between DC and effector T cells the Treg cells can efficiently inhibit T-cell priming. CD152 (CTLA4) which is expressed constitutively by Treg cells plays an important role in reducing the DCs costimulatory capacity.

The effects of Treg cells on DC are severely reduced when blocking antibodies against CD152 are added or CD152-deficient Treg cells are used in these assays. Moreover, mice that lack CD152 expression selectively in Treg cells spontaneously develop systemic autoimmune disease. This latter finding clearly illustrates that CD152 is pivotal for Treg effector functions.

Treg cells can induce DCs to express indoleamine 2,3-dioxygenase (IDO), which depletes tryptophan resulting in the suppression of effector T cell responses. This interaction also depends on the interaction between CD152 and CD80/86 (see Fig. 19.8).

This list of Treg cell effector mechanisms is by no means complete. Many more molecules and cellular interactions have been identified that may be important for Treg effector function. Clearly there is not one dominant molecular or cellular interaction by which Tregs exert their effector functions. Instead, Treg cells use different effector functions to prevent or suppress different innate or adaptive immune responses triggered by different stimuli at different anatomical locations. With few exceptions, the dramatic phenotype seen in scurfy mice or IPEX patients who lack FoxP3 does not develop when just one of the pathways mentioned in this section is absent. One such exception is the lack of CD152 which causes widespread lymphocytic infiltration of multiple organs. Moreover, IL-10 deficiency and the IL-2 deficiency syndrome, which cause a similar dramatic phenotype, are not solely attributable to disturbed Treg function.

Can loss of Treg function explain autoimmune disease?

Several pathological mechanisms have been proposed to explain why autoimmunity occurs despite the presence of Tregs. Could Tregs, despite being present in normal or even enhanced numbers have lost (some of) their effector functions? There is some evidence that Tregs can lose their function or that effector T cells become resistant to their action.

Loss of FoxP3-expression has been observed in several experimental systems. Diminished expression of FoxP3 resulted not only in reduced suppressive capacities of such ‘ex-Tregs’ in some cases these cells also acquired effector functions. Given that Tregs are frequently self-reactive such conversion could significantly contribute to the development of autoimmune diseases. Since the majority of these experiments have been performed in vitro, in genetically altered mice or in lymphopenic mice the relevance of Treg conversion into effector cells is currently unclear.

Alternatively, the effector T cells could become resistant against Treg-mediated suppression. Experimental evidence comes from mice deficient for Cbl-B, TRAF6 or NFAT1 and NFAT4. T cells from these mice cannot become anergic and they are also resistant to Treg-mediated suppression. Memory T cells, which have lower requirements for co-stimulation, are more difficult to influence by Tregs than naive TH cells. Certain TNF-family members such as OX40 and 4-1BB and several cytokines, including the gamma-chain cytokines IL-2, -4, -7 and -15 can contribute to resistance of effector T cells against suppression by Tregs.

However, except for the IPEX syndrome (Fig. 19.9), in which FoxP3⁺ Tregs are completely absent, there is currently no convincing evidence that reduced numbers of Tregs would cause autoimmune disease. In contrast, Tregs are frequently found in increased numbers at the site of autoimmune lesions.

Fig 19.9 Loss of Treg effector functions

In the balanced steady state, Tregs prevent overshooting or unwanted immune responses. Mutations in the FoxP3 gene can result in a complete lack of Treg cells and lethal autoimmune disease manifestations. FoxP3 expression is unstable in iTregs and these cells can convert into effector T cells. Impaired Treg function could also contribute to the pathogenesis of autoimmune diseases. Effector T cells can become resistant to Treg mediated suppression.

(Adapted from Buckner JH, Nat Rev Immunol 2010;10:849).

Cytokine withdrawal can induce apoptosis

One mechanism of peripheral deletion results from the lack of growth factors, particularly IL-2, for which all activated T cells compete. Cytokine withdrawal triggers an intrinsic pathway of apoptosis involving activation of the Bim factor. Regulatory T cells can accelerate or induce cytokine-withdrawal induced apoptosis of effector T cells by consuming IL-2.

IL-2 is not only critically important for the proliferation and differentiation of naive T cells. When effector T-cells are re-stimulated with large amounts of antigen, the addition of IL-2 induces apoptosis in the cycling T cells. This form of peripheral deletion has been called activation induced cell death (AICD) or restimulation induced cell death (RICD). Mice that are deficient in IL-2 or IL-2 signaling develop an autoimmune syndrome characterized by an abundance of activated T lymphocytes, multiple autoantibodies and lymphocytic infiltration of several organs. Restimulation induced cell death is one explanation for the observation that the administration of high doses of antigen can induce tolerance. This phenomenon has been called high dose tolerance.

Thus, IL-2 has contradictory effects at different phases of the T-cell response. During priming IL-2 is critically required to support clonal expansion and differentiation and a lack of IL-2 at that stage will induce apoptosis. When already activated, an effector T cell simultaneously encountering high concentrations of antigen and IL-2, will undergo apoptosis. Therefore, IL-2 both initiates and terminates T-cell responses.

T cells can be killed by ligation of Fas

The death of T cells is also mediated by the pathway involving Fas (CD95) and its ligand (FasL or CD95L); engagement of the Fas receptor induces apoptosis. Since T cells express both Fas and its ligand on activation, the interaction between the two molecules can induce apoptosis. The importance of this mechanism for the maintenance of self-tolerance is illustrated by the fact that patients with defective Fas have a severe autoimmune lymphoproliferative disease. A similar phenotype including the accumulation of chronically activated T lymphocytes and the development of autoimmunity is caused by a Fas mutation in lpr mice and by a FasL mutation in gld mice.

Some tissues, such as the anterior chamber of the eye, the CNS and the testes normally express Fas ligand. Consequently, when CD95⁺ effector T cells enter these tissues, they undergo apoptosis and cannot damage the tissue.