How regulatory T cells work

Abstract

Regulatory T (T_reg) cells are essential for maintaining peripheral tolerance, preventing autoimmune diseases and limiting chronic inflammatory diseases. However, they also limit beneficial responses by suppressing sterilizing immunity and limiting anti-tumour immunity. Given that T_reg cells can have both beneficial and deleterious effects, there is considerable interest in determining their mechanisms of action. In this Review, we discuss the basic mechanisms used by T_reg cells to mediate suppression, and discuss whether one or many of these mechanisms are likely to be crucial for T_reg-cell function. In addition, we present the hypothesis that effector T cells may not be ‘innocent’ parties in this suppressive process and might in fact potentiate T_reg-cell function.

Several sophisticated regulatory mechanisms are used to maintain immune homeostasis, prevent autoimmunity and moderate inflammation induced by pathogens and environmental insults. Chief amongst these are regulatory T (T_reg) cells that are now widely regarded as the primary mediators of peripheral tolerance. Although T_reg cells play a pivotal role in preventing autoimmune diseases, such as type 1 diabetes^1,2, and limiting chronic inflammatory diseases, such as asthma and inflammatory bowel disease (IBD)^3,4, they also block beneficial responses by preventing sterilizing immunity to certain pathogens^5,6 and limiting anti-tumour immunity⁷. A seminal advance in the analysis of T_reg cells came with the identification of a key transcription factor, forkhead box P3 (FOXP3), that is required for their development, maintenance and function^8,9. Mice and patients that lack FOXP3 develop a profound autoimmune-like lymphoproliferative disease that graphically emphasizes the importance of T_reg cells in maintaining peripheral tolerance^10-12 (BOX 1). Although FOXP3 has been proposed as the master regulator of T_reg cells that controls the expression of multiple genes that mediate their regulatory activity^13,14, this has been recently challenged raising the possibility that other transcriptional events may operate upstream of and/or concurrently with FOXP3 to mediate T_reg-cell development¹⁵.

Box 1. Scurfy mice: misplaced mechanistic expectations?

Mice that carry a spontaneous loss-of-function mutation (known as Scurfy mice) or a deletion of Foxp3 develop a fatal autoimmune-like disease with hyperresponsive CD4⁺ T cells^9,12. More recently Foxp3:diptheria toxin receptor (DTR) knockin mice have allowed for the selective depletion of T_reg cells following DT treatment¹⁰⁵. These mice have been invaluable for dissecting the role of Foxp3 in T_reg-cell function. Given the profound phenotype in these mice, there is a general expectation that genetic disruption of any key T_reg-cell inhibitory molecule or mechanism would probably result in a Scurfy-like phenotype. Of course, it is also possible that deletion of a key T_reg-cell gene may be more synonymous with DT-mediated T_reg-cell depletion where Foxp3 may still serve to prevent expression of proinflammatory cytokines¹⁰⁵. Nonetheless, this has lead to the notion that if mutant mice don't have a Scurfy-like or a T_reg-cell-depleted phenotype, then the disrupted gene probably isn't important for T_reg-cell function. This may not necessarily be correct. Indeed, it is possible that no mouse lacking a T_reg-cell inhibitory effector molecule will ever be generated that develops a profound, spontaneous autoimmune disease²¹. It should be noted that mutant mice that are Helicobacter spp. and/or Citrobacter rodentium positive may have an exacerbated phenotype, as several studies have shown that opportunistic enteric bacteria can significant exacerbate gut pathology⁴. Ultimately, the occurrence of disease in knockout mice will depend on whether T_reg cells rely on a single or multiple suppressive mechanisms. Given the number of genes induced or modulated by FOXP3, it is probable that a programme of intrinsic and extrinsic regulation is induced that involves multiple proteins^9,13. Therefore, it would not be surprising if deletion of a single molecule does not provoke the profound Scurfy-like phenotype seen in mice that lack Foxp3.

While Foxp3 has proven to be an invaluable marker for murine T_reg cells, its role in human T_reg cells is less straightforward (see BOX 2 for a discussion of T_reg-cell markers). Humans that lack FOXP3 develop immune dysregulation, polyendocrinopathy, enteropathy, X-linked syndrome (IPEX), a severe autoimmune disease that presents early in infancy. Although FOXP3 appears to be required for human T_reg-cell development and function, expression of FOXP3 alone is clearly not sufficient as a significant percentage of human activated T cells express FOXP3 and yet do not possess regulatory activity^16-20. Furthermore, induction of FOXP3 in human T cells by transforming growth factor-β (TGFβ) does not confer a regulatory phenotype, in contrast to their murine counterparts²⁰. Consequently, FOXP3 is not a good marker for human T_reg cells (BOX 2). Whether this distinction is due to intrinsic differences between mouse and human FOXP3 and/or a requirement for an additional cofactor/transcription factor is an important question that needs to be resolved.

Box 2. T_reg-cell markers.

Identifying discriminatory cell surface markers for the characterization and isolation of T_reg cells has always been a critical goal. Although excellent markers exist for murine T_reg cells, this goal has remained elusive for human T_reg cells. Traditionally, murine and human T_reg cells have been characterized as CD4⁺CD25⁺ (also known as interleukin-2 receptor α (IL-2Rα)). Indeed, murine T_reg cells can be effectively isolated based on staining for CD4⁺CD25⁺CD45RB^low expression. However, the purity of isolated human T_reg cells has always been an issue because T cells up-regulate CD25 upon activation¹⁰⁶. Indeed, during the influenza or allergy season a substantial proportion of human CD4⁺ T cells can express CD25. Although the identification of forkhead box P3 (Foxp3) as a key regulator of T_reg-cell development and function has facilitated their identification in the mouse⁸, many activated (non-regulatory) human T cells express FOXP3, precluding it as a useful marker for human T_reg cells^16-20. Consequently, the search for T_reg-cell-specific cell-surface markers, particularly in humans, has continued in earnest with a growing number of candidates proposed (reviewed by Zhao and colleagues¹⁰⁷). For instance, it was shown that the expression of CD127 (also known as IL-7R) is down-regulated on T_reg cells and that this could be used to increase the purity of human T_reg-cell isolation. Indeed, there is a 90% correlation between CD4⁺CD25⁺CD127^low T cells and FOXP3 expression^108,109. In addition, it was recently found that T_reg cells expressed a higher level of folate receptor 4 (FR4) compared with activated effector T cells¹¹⁰. It is also important to recognize that T_reg cells, like their T helper cell counterparts, may be heterogeneous and thus a collection of cell surface markers could facilitate their isolation and functional characterization. Indeed, such heterogeneity has recently been described based on differential expression of HLA-DR or CCR6^102,103. However, the general use of both markers remains to be fully established so it is quite probable that the search for better T_reg-cell markers will continue for some time.

Significant progress has been made over the last few years in delineating the molecules and mechanisms that T_reg cells use to mediate suppression^21,22. In this Review, we outline our current understanding of the mechanisms used by T_reg cells to mediate suppression, and the challenges that lie ahead in defining their mode of action. We also discuss whether T_reg cells are likely to depend on one, a few or many of these mechanisms. In addition, we propose that effector T cells may have a significant role in boosting and/or modulating T_reg-cell function. Unless stated, we focus here primarily on the mechanisms that are used by thymus-derived natural CD4⁺CD25⁺ FOXP3⁺ T_reg cells.

Basic mechanisms of T_reg-cell function

Defining the mechanisms of T_reg-cell function is clearly of crucial importance. Not only would this provide insight into the control processes of peripheral tolerance but it would probably provide a number of potentially important therapeutic targets. Although this quest has been ongoing since interest in T_reg cells was reignited in 1995²³, there has been significant progress in the last few years. From a functional perspective, the various potential suppression mechanisms of T_reg cells can be grouped into four basic ‘modes of action’: suppression by inhibitory cytokines, suppression by cytolysis, suppression by metabolic disruption, and suppression by modulation of dendritic-cell (DC) maturation or function (FIG. 1).

Figure 1. Basic mechanisms used by T_reg cells.

This schematic depicts the various regulatory T (T_reg)-cell mechanisms arranged into four groups centred around four basic modes of action. ‘Inhibitory cytokines’ include interleukin-10 (IL-10), interleukin-35 (IL-35) and transforming growth factor-β (TGF-β). ‘Cytolysis’ includes granzyme-A- and granzyme-B-dependent and perforin-dependent killing mechanisms. ‘Metabolic disruption’ includes high affinity IL-2 receptor α (CD25)-dependent cytokine-deprivation-mediated apoptosis, cyclic AMP (cAMP)-mediated inhibition, and CD39- and/or CD73-generated, adenosine–purinergic adenosine receptor (A2A)-mediated immunosuppression. ‘Targeting dendritic cells’ includes mechanisms that modulate DC maturation and/or function such as lymphocyte activation gene-3 (LAG3; also known as CD223)–MHC-class-II-mediated suppression of DC maturation, and cytotoxic T lymphocyte antigen-4 (CTLA4)–CD80/CD86-mediated induction of indoleamine 2,3-dioxygenase (IDO), which is an immunosuppressive molecule, by DCs.

Suppression by inhibitory cytokines

Inhibitory cytokines, such as interleukin-10 (IL-10) and TGFβ, have been the focus of considerable attention as a mechanism of T_reg-cell-mediated suppression. There has also been significant interest in their ability to generate induced (also known as adaptive) T_reg-cell populations, either naturally in vivo or experimentally as a potential therapeutic modality (BOX 3). Although the general importance of IL-10 and TGFβ as suppressive mediators is undisputed, their contribution to the function of thymus-derived, natural T_reg cells is still a matter of debate²⁴. This is partly due to the general perception that T_reg cells function in a contact-dependent manner^25,26. Indeed, in vitro studies using neutralizing antibodies or T cells that are unable to produce or respond to IL-10 and TGFβ suggested that these cytokines may not be essential for T_reg-cell function^25-28. However, this contrasts with data from in vivo studies^29,30.

Box 3. Induced or adaptive T_reg cells: development and mode of action.

Naturally occurring FOXP3⁺CD4⁺CD25⁺ T_reg cells develop in the thymus and display a diverse T-cell receptor (TCR) repertoire that is specific for self-antigens^111,112. However, T_reg cells can also be ‘induced’, ‘adapted’ or ‘converted’ from effector T cells during inflammatory processes in peripheral tissues, or experimentally generated as a possible therapeutic^29,113,114. For instance, T regulatory 1 cells (Tr1) and T helper 3 cells (Th3) can be generated experimentally by, and mediate their suppressive activity through interleukin-10 (IL-10) and transforming growth factor-β (TGFβ), respectively^114,115. Typically, these regulatory populations do not express FOXP3. In vivo, it has recently been suggested that stimulation of mouse effector T cells by CD103⁺ dendritic cells (DCs) in the presence of TGFβ and retinoic acid induces the generation of Foxp3⁺ T cells in the gut-associated lymphoid tissue (GALT)^116-121. Furthermore, T_reg cells can be preferentially induced in the periphery by exposure to α_Vβ₈-integrin-expressing DCs¹²² or suppressor of cytokine signalling 3 (Socs3)^−/− DCs¹²³. Interestingly, independent of its role in generating induced T_reg cells, TGFβ may also have an important role in helping to maintain Foxp3 expression in natural T_reg cells¹²⁴, a process that can be blocked by IL-4 or interferon-γ (IFNγ)¹²⁵. In contrast to mouse T cells, FOXP3 induction by TCR stimulation in the presence of TGFβ in human T cells does not confer a regulatory phenotype²⁰. The mechanism of action of adaptive T_reg cells may not necessarily be restricted to suppressive cytokines. Indeed, human adaptive T_reg cells (CD4⁺CD45RA⁺ T cells stimulated with CD3- and CD46-specific antibodies) have also been shown to express granzyme B and killing target cells in a perforin-dependent manner¹²⁶. In contrast to natural T_reg cells, induced T_reg cells often have a restricted specificity for particular cell types, tumours or foreign antigens¹²⁷. Therefore, induced T_reg cells may be ideally suited to respond to infectious agents. This may also be of particular importance in the GALT and in the tumour microenvironment where TGFβ drives the conversion of induced T_reg cells^118,128. A significant challenge in deciphering data from in vivo experiments is to assess the contribution of natural T_reg cells versus induced T_reg cells, and to determine whether inhibitory molecules, such as IL-10 or TGFβ, were derived from the former or the latter (or elsewhere).

In allergy and asthma models, evidence suggests that both natural and antigen-specific T_reg cells control disease in a manner that is, in part, dependent on IL-10²⁹ and in some reports dependent on both IL-10 and TGFβ³¹. Adoptive transfer of allergen-specific T_reg cells induced significant IL-10 production by CD4⁺ effector T cells in the lung following allergen challenge and this T_reg-cell-mediated control of disease was reversed by treatment with an IL-10-receptor-specific antibody³². However, suppression of allergic inflammation and airway hyper-reactivity, and increased production of IL-10 still occurred following transfer of IL-10-deficient T_reg cells, suggesting that T_reg cells can suppress the Th2-driven response to allergens in vivo through an IL-10-dependent mechanism, but that the production of IL-10 by T_reg cells themselves is not required for the suppression observed. This contrasts with a recent study suggesting that the T_reg-cell-specific ablation of IL-10 expression resulted in increased lung allergic inflammation and hyperreactivity³³.

This scenario might occur in other disease models. For instance, the effects of IL-10 can only be partially attributed to T_reg-cell-derived IL-10 in the immune response to hepatitis B virus³⁴ and in the allograft tolerance response elicited by splenocytes exposed to non-inherited maternal antigens³⁵. Recently, it was also shown that IL-10 is crucial for the control of various infections in which T_reg cells have been reported to be involved including Mycobacterium tuberculosis³⁶, Toxoplasma gondii³⁷, Leishmania major³⁸, and Trichinella spiralis³⁹. However, T_reg cells were not the source of IL-10 in all of these infection models.

By contrast, several studies have shown that IL-10 production by T_reg cells is essential for the prevention of colitis in mouse models of IBD⁴⁰. Moreover, it appears that the tumour microenvironment promotes the generation of FOXP3⁺ T_reg cells that mediate IL-10-dependent, cell-contact independent, suppression⁴¹. Similarly, in UV-radiation-induced carcinogenesis, IL-10 production by T_reg cells appears to be important for blocking anti-tumour immunity⁴². IL-10 produced by T_reg cells also appears to be crucial for IL-10-mediated tolerance in a model of hepatitis induced by concanavalin A⁴³ and tolerance to bacterial and viral superantigens⁴⁴. In addition, recent papers suggest new roles for T_reg-cell-derived IL-10 in the induction of feto-maternal tolerance⁴⁵ and B-cell-enhanced recovery from experimental autoimmune encephalomyelitis⁴⁶. Collectively, the picture that appears to be emerging is that the relative importance of T_reg-cell-derived IL-10 is very dependent on the target organism or disease and on the experimental system. Furthermore, the T_reg-cell-specific deletion of IL-10 did not result in the development of spontaneous systemic autoimmunity, but did result in enhanced pathology in the colon of older mice and in the lungs of mice with induced airway hypersensitivity, suggesting that the function of T_reg-cell-derived IL-10 may be restricted to controlling inflammatory responses induced by pathogens or environmental insults³³.

While some early in vitro studies using neutralizing antibodies to TGFβ or T_reg cells lacking TGFβ^25,47 indicated that TGFβ was not required for natural T_reg-cell function, other studies, both in vitro and in vivo suggested a critical role for T_reg-cell surface bound TGFβ^48,49. Therefore, the importance of TGFβ for natural T_reg-cell function has also been a controversial topic. Indeed, there has been considerably more focus recently on the importance of TGFβ in the development of induced T_reg cells and perhaps in T_reg-cell maintenance in general (BOX 3). However, there are studies that suggest that TGFβ produced by T_reg cells may directly participate in effector T-cell suppression. For instance, effector T cells that are resistant to TGFβ-mediated suppression cannot be controlled by T_reg cells in an IBD model⁵⁰. In addition, TGFβ produced by T_reg cells has been found to be important in the control of the host immune response to M. tuberculosis³⁶, suppression of allergic responses³¹ and prevention of colitis in an IBD model⁵¹. Interestingly, TGFβ produced by T_reg cells has also been implicated in limiting anti-tumour immunity in head and neck squamous-cell carcinoma⁵² and in follicular lymphoma⁵³ by rendering T cells unresponsive to the tumour. TGFβ also appears to limit the anti-tumour activity of cytokine-induced killer cells⁵⁴.

Membrane-tethered TGFβ can also mediate suppression by T_reg cells in a cell-cell contact-dependent manner⁴⁸. T_reg cells can control islet infiltration of CD8⁺ T cells and delay the progress of diabetes through membrane-tethered TGFβ⁴⁹. However, experiments using mice deficient in TGFβ-receptor (TGFβR) signalling in effector T cells or using TGFβ or TGFβR blocking reagents failed to show that membrane-tethered TGFβ is required for natural T_reg-cell development or function⁴⁷. More recently, however, interest in membrane-tethered TGFβ has re-surfaced with the description of a previously unappreciated role for it in the tumour microenvironment. TGFβ associated with tumour exosome membranes appears to enhance the suppressive function of T_reg cells and skew T cells away from their effector functions and towards regulatory functions⁵⁵. Furthermore, ovalbumin-induced airway inflammation can be attenuated by heme oxygenase-1 through membrane-tethered TGFβ and IL-10 secretion by T_reg cells⁵⁶, a process that activates the Notch1–HES1 (hairy and enhancer of split 1) axis in target cells⁵⁷. Thus, in light of the most current data, it now appears that soluble and/or membrane-tethered TGFβ may have a previously unappreciated role in natural T_reg-cell function.

Recently, a new inhibitory cytokine, IL-35, has been described that is preferentially expressed by T_reg cells and is required for their maximal suppressive activity⁵⁸. IL-35 is a novel member of the IL-12 heterodimeric cytokine family and is formed by the pairing of Epstein–Barr virus-induced gene 3 (Ebi3), which normally pairs with p28 to form IL-27, and p35 (also known as Il12a), which normally pairs with p40 to form IL-12. Both Ebi3 and Il12a are preferentially expressed by murine Foxp3⁺ T_reg cells^58,59, but not resting or active effector T cells, and are significantly upregulated in actively suppressing T_reg cells⁵⁸. As predicted for a heterodimeric cytokine, both Ebi3^−/− and Il12a^−/− T_reg cells had significantly reduced regulatory activity in vitro and failed to control homeostatic proliferation and cure IBD in vivo. This precise phenocopy suggested that IL-35 is required for the maximal suppressive activity of T_reg cells. Importantly IL-35 was not only required but sufficient, as ectopic expression of IL-35 conferred regulatory activity on naive T cells and recombinant IL-35 suppressed T cell proliferation in vitro⁵⁸. Although IL-35 is an exciting addition to the T_reg-cell portfolio, there is clearly much that remains to be defined about this cytokine and its contribution to T_reg-cell function. For instance, it remains to be determined if IL-35 suppresses the development and/or function of other cell types such as DCs and macrophages.

It is now clear that three inhibitory cytokines, IL-10, IL-35 and TGFβ, are key mediators of T_reg-cell function. Although they are all inhibitory, the extent to which they are utilized in distinct pathogenic/homeostatic settings differs suggesting a non-overlapping function, which needs further refinement.

Suppression by cytolysis

Cytolysis mediated through secretion of granzymes had long been considered the forte of natural killer (NK) cells and cytotoxic CD8⁺ T lymphocytes (CTLs) (reviewed by Lieberman in REF. 60). However, many human CD4⁺ T cells exhibit cytotoxic activity. Consistent with this, activated human natural T_reg cells have been shown to express granzyme A. Furthermore, target cell killing was mediated by granzyme A and perforin through adhesion of CD18⁶¹.

By contrast, murine CD4⁺ T cells are not cytolytic and therefore it was surprising that early gene expression arrays showed that granzyme B expression was up-regulated in murine T_reg cells^62,63. Noelle and co-workers were the first to report that granzyme-B-deficient murine T_reg cells had reduced suppressive activity in vitro⁶⁴, but this T_reg-cell-induced apoptosis appeared to be perforin-independent. The notion that T_reg cells might possess cytolytic activity was supported by studies showing that T_reg cells can kill B cells in a granzyme B-dependent and partially perforin-dependent manner that results in the suppression of B-cell function⁶⁵. More recently, T_reg cells were shown to suppress the ability of NK cells and CTLs to clear tumours by killing these cells in a granzyme-B- and perforin-dependent manner⁶⁶. In addition, Noelle and colleagues have unpublished data showing that effector T cells that overexpress a granzyme-B-specific inhibitor, Spi-6, are resistant to T_reg-cell-mediated suppression (Randolph Noelle, personal communication). Using a transplantation model in which T_reg-cell-mediated tolerance is induced by CD154–CD40 co-stimulatory blockade in conjunction with donor lymphocyte-specific transfusion, they have also shown that the T_reg cells that mediate this tolerance are also dependent on granzyme B for their suppressive activity.

Although the majority of research to date regarding T_reg-cell-induced cytolysis has been focused on granzyme-B-mediated mechanisms, a recent study has suggested that activated T_reg cells induce apoptosis of effector T cells through a TRAIL–DR5 (tumour-necrosis factor-related apoptosis inducing ligand–death receptor 5) pathway⁶⁷. Furthermore, it has been suggested that galectins can mediate cytolysis in a granzyme- and perforin-independent manner⁶⁸. These studies emphasize that more work is required to define the cytolytic mechanisms that T_reg cells use to mediate suppression.

Suppression by metabolic disruption

Recently, several intriguing suppressive mechanisms have been described that could collectively be referred to as mechanisms that mediate ‘metabolic disruption’ of the effector T-cell target. A long-standing debate in the T_reg-cell field is whether the high expression of CD25 empowers T_reg cells to ‘consume’ local IL-2 and therefore starve actively dividing effector T cells by depleting the IL-2 they need to survive^26,69. Although previous studies suggested that this was not a bone fide T_reg-cell mechanism^70,71, a recent study has reignited interest in this question by suggesting that T_reg cells induce cytokine (specifically IL-2) deprivation-mediated apoptosis⁷². However, given that a recent report using human T_reg cells suggests that IL-2 depletion alone is not required for T_reg cells to suppress effector T cells⁷³, more work is clearly required to resolve this debate.

Two new T_reg-cell mechanisms have recently been proposed that induce the intracellular or extracellular release of adenosine nucleosides. Concordant expression of the ectoenzymes CD39 and CD73 was shown to generate pericellular adenosine, which suppressed effector T-cell function through activating the adenosine A2A receptor^74-76. Interestingly, binding of adenosine to the A2A receptor appears to not only inhibit effector T-cell functions but also to enhance the generation of adaptive T_reg cells by inhibiting IL-6 expression while promoting TGFβ secretion⁷⁷. In addition, adenosine has also been shown to modulate DC maturation and favour a toleragenic phenotype (Peter Ernst, personal communication). Although TGFβ induces Foxp3 and T_reg-cell differentiation, IL-6 inhibits the generation of T_reg cells and promotes generation of pro-inflammatory Th17 cell development⁷⁸. Thus, inhibiting IL-6 has important implications in T_reg-cell maintenance. T_reg cells were also shown to suppress effector T-cell function directly by transferring the potent inhibitory second messenger cyclic AMP into effector T cells via membrane gap junctions⁷⁹. Although these mechanisms represent interesting additions to the T_reg-cell arsenal, further studies will be required to corroborate these exciting findings and assess their relative use by T_reg cells.

Suppression by targeting dendritic cells

In addition to directly affecting effector T-cell function, T_reg cells might modulate the maturation and/or function of dendritic cells (DCs) required for effector T-cell activation. This has long been considered an attractive idea but there has been only limited data in support⁸⁰. However, intravital microscopy has revealed direct interactions between T_reg cells and DCs in vivo, which was proposed to attenuate effector T-cell activation by DCs^81,82. So in what way might DCs be used as a conduit for T_reg-cell-mediated suppression? Some time ago, cytotoxic T-lymphocyte antigen 4 (CTLA4) was shown to be constitutively expressed by T_reg cells^25,83, and by using either CTLA4-specific blocking antibodies or CTLA4-deficient T_reg cells it was shown that in the absence of functional CTLA4, T_reg-cell-mediated suppression of effector T cells via DCs was reduced^84,85. Importantly, it was also shown that T_reg cells could condition DCs, through a mechanism dependent on interactions between CTLA4 and CD80 and/or CD86, to express indoleamine 2,3-dioxygenase (IDO), which is a potent regulatory molecule that induces the catabolism of tryptophan into pro-apoptotic metabolites that results in the suppression of effector T cells^86,87.

In addition to inducing DCs to produce immunosuppressive molecules, several studies have suggested that T_reg cells may also downmodulate the capacity of DCs to activate effector T cells. Ivars and colleagues first reported that T_reg cells could downregulate the expression of the co-stimulatory molecules CD80 and CD86 on DCs in vitro⁸⁸. Several studies have also reported the immunomodulatory effects of T_reg cells on DC maturation and/or function^85,89-92. Studies with human T_reg cells have also indicated that T_reg cells may also modulate the function of monocytes and macrophages^93,94. Although the precise mechanism by which this is orchestrated remains elusive, this modulation may be mediated through cell-surface molecules such as CTLA4 and/or cytokines such IL-10 and TGFβ.

Recent studies have also suggested that lymphocyte-activation gene 3 (Lag3; also known as CD223) may block DC maturation. Lag3 is a CD4 homologue that binds MHC class II molecules with very high affinity, has a negative regulatory T cell intrinsic function and is required for maximal T_reg-cell suppression^95,96. Binding of Lag3 to MHC class II molecules expressed by immature DCs induces an immunoreceptor tyrosine-based activation motif (ITAM)-mediated inhibitory signalling pathway, which involves FcγRγ– and extracellular-signal-regulated kinase (ERK)-mediated recruitment of SH2-domain-containing protein tyrosine phosphatase 1 (SHP1), which suppresses DC maturation and immunostimulatory capacity⁹⁷. It is noteworthy that human MHC class II⁺ T_reg cells, have been shown to be more suppressive than MHC class II⁻ T_reg cells, raising the possibility that these cells suppress by ligating LAG3 on activated effector T cells⁹⁸. Although more work is required to fully elucidate if, and how, T_reg cells might suppress effector T-cell function through DCs, this mode of action is attractive, as it may be a more efficient way of suppressing immune responses in vivo given the ∼1:8 ratio of T_reg cells to effector T cells, compared with the ∼1:0.8 T_reg cell to DC ratio found in the peripheral lymph nodes (as determined by flow cytometry and cell counting of pooled lymph nodes; CJW and DAAV, unpublished observations). Furthermore, it has recently been shown that neuropilin-1 (Nrp-1) promotes prolonged interactions with T_reg cells and immature DCs⁹⁹. Given that Nrp-1 is differentially expressed on T_reg cells, this may give them an advantage over naïve T cells in modulating DC function.

Lastly, T_reg cells can also influence immune responses by modulating the recruitment and function of other cell types. For instance, T_reg-cell-derived IL-9 has been shown to recruit and activate mast cells which were shown to be essential regulatory intermediaries in the establishment of peripheral allograph tolerance¹⁰⁰.

Complicating issues

Current dogma dictates that a hallmark of T_reg cells is their dependence on direct contact to mediate their inhibitory activity. This has been upheld by in vitro experiments where T_reg cells are unable to suppress effector T cell proliferation when the two populations are separated by a permeable membrane^25,26. These data led to the notion that T_reg-cell-mediated suppression is contact-dependent. However, there are two important issues one should consider when evaluating the T_reg-cell mechanisms outlined above in the context of contact-dependency. First, these assays are really a measure of proximity rather than contact. Indeed, soluble mediators are most effective close to the source of their generation. The close proximity maintains high local cytokine concentrations, which has been shown to be important for the function of interleukin-2 (IL-2)¹⁰¹. Therefore, the dilution effect of diffusion across the well may render a soluble mediator ineffective. One should also consider the importance of proximity for labile mediators that might be very effective when T_reg cells are close to their target cells but not when far away. For instance, adenosine has a half-life of less than ten seconds.

Second, it is not yet clear how much of the regulatory potency of T_reg cells is directed towards DCs/APCs versus effector T cells. Although several studies have shown that T_reg cells can directly suppress effector T cells in vitro in the absence of APCs, there is no direct evidence that contact between T_reg cells and effector T cells is required for suppression in vivo. Indeed, intravital microscopy experiments suggest that T_reg cells are far more frequently found in contact with DCs^81,82. Furthermore, it is still not clear what the primary target is for many of the mechanisms described above. For instance, suppression by cytolysis, adenosine or cAMP could be directed against DCs and/or effector T cells. Inhibitory cytokines could also influence both populations. For example, although IL-35 was shown to directly act on effector T cells, an effect on DCs has not been precluded. The one mechanism that might be considered effector T cell exclusive is IL-2 deprivation-mediated apoptosis. Clearly, more work is needed to determine the primary target of T_reg-cell suppression, particularly in vivo.

How many mechanisms do T_reg cells need?

Although efforts to define the suppressive mechanisms used by T_reg cells continue, an important question looms large. Is it likely that all these molecules and mechanisms will be crucial for T_reg-cell function? There are three broad possibilities.

One, a single, overriding suppressive mechanism is required by all T_reg cells

Until the entire mechanistic panoply of T_reg cells is defined, one cannot completely rule out this possibility. However, this possibility would seem unlikely as none of the molecules and/or mechanisms that have been defined to date, when blocked or deleted, result in the complete absence of regulatory activity — a consequence that one might predict would result in a ‘Scurfy-like’ phenotype (BOX 1). So, although T_reg cells that lack a single molecule, for instance IL-10, IL-35 or granzyme B, exhibit significantly reduced suppressor function, a scurfy phenotype does not ensue. Given that none of the current T_reg-cell mechanisms can exclusively claim this distinction, it seems unlikely that any ‘unknown’ molecules or mechanisms could do so either.

Two, multiple, non-redundant mechanisms are required for maximal T_reg-cell function

In the studies conducted to date, T_reg cells that lack various suppressive molecules have been shown to be functionally defective. This favours a scenario where there are multiple mechanisms that can be used by T_reg cells but they are non-redundant, with each molecule contributing to the mechanistic whole. At present, this possibility would seem plausible. Indeed, this is supported by the recent analysis of mice possessing a T_reg-cell-specific ablation of IL-10 expression, in which enhanced pathology was observed following environmental insult³³. One would predict that at some point we should be able to generate knockout mice that lack a particular set of genes which results in a complete loss of T_reg-cell activity. For this to be truly non-redundant, this list would probably be restricted and small (2–4 genes).

Three, multiple, redundant mechanisms are required for maximal T_reg-cell function

With the plethora of regulatory mechanisms described to date and the possibility of more yet to be identified, it is conceivable that there are multiple mechanisms that function redundantly. Such a redundant system would help to mitigate against effector T-cell escape from regulatory control. Also, given the very small size of the T_reg-cell population, a sizable arsenal may be required at the height of an effector T-cell attack. Of course, it is possible that a semi-redundant scenario exists.

These possibilities have been discussed from the perspective of there being a single homogeneous T_reg-cell population. However, as for helper T cell subsets it remains possible that a few or even many different T_reg-cell subsets exist²⁴. Each of these may rely on one or multiple regulatory mechanisms. Several recent studies have provided support for both phenotypic and functional heterogeneity amongst T_reg cells. For instance, it has recently been shown that a small sub-population of T_reg cells express the chemokine receptor CCR6, which is associated with T cells possessing an effector-memory phenotype¹⁰². CCR6⁺ T_reg cells appeared to accumulate in the central nervous systems of mice with experimental autoimmune encephalomyelitis (EAE) suggesting that they may have a prevalent role in controlling responses in inflamed tissues. Heterogeneous expression of HLA-DR has also been suggested to mark different subpopulations of functionally distinct human T_reg cells¹⁰³. Indeed, HLA-DR positive T_reg cells were found to be more suppressive than their DR negative counterparts. One might speculate that their enhanced inhibitory activity is due to DR-mediated ligation of the inhibitory molecule LAG3 expressed by activated effector T cells^95,96.

So, if multiple suppressor mechanisms exist, how might these be integrated and used productively by T_reg cells in vivo? We would propose the following possible models²¹. First, a ‘hierarchical’ model in which T_reg cells possess many mechanisms that could be used but only one or two that are really crucial and consistently important in a variety of regulatory settings. Second, a ‘contextual’ model where different mechanisms become more or less important depending on the background or context in which the T_reg cells reside and the type of target cell that they have to repress. For example, some cell types may be inhibited primarily by cytokines, whereas others are most effectively suppressed through lysis by T_reg cells. Alternatively, different mechanisms may be more effective in different tissue compartments or in different disease settings. This notion is supported by the recent analysis of mice in which IL-10 expression was specifically ablated in T_reg cells³³. Whereas T_reg-cell-derived IL-10 was not required for the systemic control of autoimmunity, it did seem to be required from the control of inflammatory events at mucosal interfaces such as the lungs and colon. As a clear picture of the available T_reg-cell weaponry emerges, an important challenge will be to determine their relative importance and contribution to T_reg-cell function in different disease models.

A hypothesis: effector T cells potentiate T_reg-cell function?

Most cellular interactions within the immune system are bidirectional, with molecular signals moving in both directions even though the interaction has broader unidirectional intentions (for example, CD4⁺ T-cell help). However, to date the general perception is that T_reg cells suppress and effector T cells capitulate. We hypothesize that this is in fact an incomplete picture and that effector T cells have a very active role in their own functional demise. Three recent observations support this view. First, we have recently examined the molecular signature of activated T_reg cells in the presence and absence of effector T cells and were surprised to find that it was strikingly different, with hundreds of genes differentially modulated as a consequence of the presence of effector T cells (C.J.W. and D.A.A.V., unpublished observations). Second, we have shown that Ebi3 and Il12a mRNA are markedly upregulated in T_reg cells that were co-cultured with effector T cells, supporting the idea that effector T cells may provide signals which boost IL-35 production in trans⁵⁸. Third, we found that T_reg cells were able to mediate suppression of effector T cells across a permeable membrane when placed in direct contact with effector T cells in the upper chamber of a Transwell™ plate (L.W.C. and D.A.A.V., unpublished observations). Interestingly, this suppression was IL-35 dependent, as Ebi3^−/− T_reg cells were unable to mediate this ‘long-distance’ suppression. Collectively, these data suggest that it is the ‘induction’, rather than the ‘function’, of T_reg-cell suppression that is contact-dependent and that effector T cells have an active role in potentiating T_reg-cell-mediated suppression. Therefore, we hypothesize that receptor–ligand interactions between the co-cultured CD4⁺ effector T cells and T_reg cells initiate a signalling pathway that leads to enhanced IL-35 secretion and regulatory activity (FIG. 2). While the molecule that mediates this enhanced T_reg-cell suppression is unknown, it is possible that IL-2 may serve this function¹⁰⁴. Given the contrasting genetic profiles of activated T_reg cells in the presence and absence of effector T cells, it seems possible that this interaction may boost the expression of other regulatory proteins. It may well be that effector T cells unwittingly perform the ultimate act of altruism.

Figure 2. Model for how effector T cells might boost T_reg-cell function.

This occurs in three stages. (a) Initial regulatory T (T_reg)-cell activation induces production of regulatory factors such as interleukin-35 (IL-35). (b) T_reg cells ‘sense’ the presence of recently activated effector T cells through a receptor–ligand interaction (cell surface or soluble). (c) This in turn boosts or potentiates T_reg-cell function resulting in the enhanced production of regulatory mediators, such as IL-35, and perhaps the induction of new mediators.

Concluding remarks

Although significant progress has been made over the last few years in defining the mechanisms that T_reg cells use to mediate their suppressive function, there is clearly much that remains to be elucidated and many questions persist. First, are there more undiscovered mechanisms and/or molecules that mediate T_reg-cell suppression? What is clear is that the transcriptional landscape of T_reg cells is very different from naive or activated effector T cells. There are literally thousands of genes that are upregulated (or downregulated) in T_reg cells compared with effector T cells. Although it seems unlikely that all or many of these will be crucial for T_reg-cell function, it is quite possible that a few undiscovered genes might be important. It should be noted that although we are discussing mechanisms here, it is clear that some of these molecules may perform key T_reg-cell functions, such as T_reg-cell homing and homeostasis, which are likely to indirectly influence their suppressive capacity in vivo but don't directly contribute to their inhibitory activity. It is also possible that some of these unknown molecules may represent more specific markers for the characterization and isolation of T_reg cells, a particularly important issue for the analysis and use of human T_reg cells (BOX 2).

Second, which mechanisms are most important? An important but potentially complex challenge will be to determine if a few mechanisms are important in many T_reg-cell settings or whether different mechanisms are required in different cellular scenarios. At present it is difficult to assess this objectively as these mechanisms have predominantly been elucidated in different labs using distinct experimental systems and thus none have really been compared in side-by-side experiments. Furthermore, only recently have conditional mutant mice been examined that have a regulatory component specifically deleted in T_reg cells³³.

It almost goes without saying that although defining the T_reg-cell mode of action is of great academic importance, it is also essential in order to develop effective approaches for the clinical manipulation of T_reg cells. Given the capacity of T_reg cells to control inflammation and autoimmunity, and their implication in blocking effective anti-tumour immunity and preventing sterilizing immunity, it seems probable that a clear understanding of how T_reg cells work will present definitive opportunities for therapeutic intervention.

Glossary

Peripheral tolerance

The lack of self-responsiveness of mature lymphocytes in the periphery to specific antigens. These mechanisms control potentially self-reactive lymphocytes that have escaped central-tolerance mechanisms. Peripheral tolerance is associated with suppression of production of self-reactive antibodies by B cells and inhibition of self-reactive effector cells, such as cytotoxic T lymphocytes. Regulatory T (T_reg) cells constitute one mechanism of peripheral tolerance.

Type 1 diabetes

A chronic autoimmune disease that is characterized by the T-cell-mediated destruction of β-cells (which secrete insulin) in the pancreas. Individuals with type 1 diabetes develop hyperglycaemia and can develop diabetes-associated complications in multiple organ systems owing to lack of insulin.

Inflammatory bowel disease (IBD)

A T-cell-mediated inflammatory response that affects the gastrointestinal tract. There are two forms of IBD in humans; Crohn's disease, which can affact any part of the gastrointestinal tract but usually desends from the terminal ileum, and ulcerative colitis (UC), which mainly affects the colon. In the mouse model, most of the inflammation is confined to the large intestines. The target antigen for the pathogenic T cells is unknown.

Airway hyper-reactivity

Initiated by exposure to a defined stimulus that is usually tolerated by normal individuals and that causes bronchoconstriction and inflammatory-cell infiltration in allergic individuals.

Experimental autoimmune encephalomyelitis (EAE)

An animal model of the human autoimmune disease multiple sclerosis. EAE is induced in experimental animals by immunization with myelin or peptides derived from myelin. The animals develop a paralytic disease with inflammation and demyelination in the brain and spinal cord.

Exosomes

Small lipid-bilayer vesicles that are released from activated cells. They comprise either plasma membrane or membrane derived from intracellular vesicles.

Adenosine nucleosides

Adenosine (C₁₀H₁₃N₅O₄) is a nucleoside composed of adenine linked to ribose and is a structural component of nucleic acids. It is also the primary molecular component of cAMP (Cyclic adenosine monophosphate – an important intracellular second messenger), AMP, ADP and ATP (a key sourse of chemical energy for many enzymatic reactions).

Ectoenzymes

An enzyme that is outside the cell memebrane and thus can cleave extracellular substratetes. These are typically teathered to the outside of the cell by a transmembrane domain.

Intravital microscopy

This is used for examination of biological processes, such as leukocyte–endothelial-cell interactions, in living tissue. In general, translucent tissues are used, such as the mesentery or cremaster muscle, which can be exposed and mounted for microscopic observation.

Granzymes

A family of serine proteinases that are found primarily in the cytoplasmic granules of cytotoxic T lymphocytes and natural killer cells. They enter target cells through perforin pores, then cleave and activate intracellular caspases and lead to target-cell apoptosis.

Perforin

A component of cytolytic granules that participates in the permeabilization of plasma membranes, allowing granzymes and other cytotoxic components to enter target cells.

Sterilizing immunity

An immune response that leads to the compele removal of the pathogen.

Biographies

DARIO VIGNALI

Dario AA Vignali received his Ph.D. from the London School of Hygiene and Tropical Medicine, University of London, England, and his postdoctoral training from the German Cancer Research Centre in Heidelberg, Germany and Harvard University in Cambridge, Massachusetts, USA. He is currently an Associate Member of the Immunology Department at St. Jude Children's Research Hospital in Memphis, Tennessee, USA. His research focuses on various aspects of T cell biology including TCR:CD3 signalling and cell biology, the development and function of regulatory T cells, and type 1 diabetes.

LAUREN COLLISON

Lauren Collison received her Ph.D. from The University of Texas at Austin where her work focused on phospholipid metabolism in aging T cells. She is currently a postdoctoral fellow in the Immunology Department at St. Jude Children's Research Hospital in Memphis, Tennessee, USA. Her research interests are in immunoregulatory mechanisms and she aims to pursue work in the field of translational immunology.

CREG WORKMAN

Creg J Workman received his Ph.D. at the University of Illinois, Champaign, Illinois. He completed his postdoctoral training at St. Jude Children's Research Hospital in Memphis, Tennessee. He is currently a Scientific Manager in Dario Vignali's laboratory where his research interests include mechanisms of regulatory T cell function and type 1 diabetes.