Regulatory T cells (Tregs) play an indispensable role in the immune system as they are involved in the prevention of autoimmune diseases, allergies, infection-induced organ pathology, transplant rejection as well as graft versus host disease (GvHD) by suppression of effector T cells and other immune cells (Sakaguchi, 2004). However, Tregs can also dampen immune responses against tumors, as described for various types of cancer (Zou, 2006; Curiel, 2008). Conversely, in certain situations, Tregs can also protect against cancer by controlling cancer-associated inflammation (Gounaris et al., 2009). Therefore, Tregs are subject to intense investigations.

The history of Tregs begins in 1970 (Gershon and Kondo, 1970). Unfortunately, due to lack of a molecular marker, research on suppressor T cells was stopped, until many years later Sakaguchi et al. (1995) identified CD25 as a phenotypic marker for suppressive CD4⁺ T cells in mice. These suppressive T cells were named Tregs and were later also found in humans within the CD4⁺CD25^high T cell population (Baecher-Allan et al., 2001; Dieckmann et al., 2001; Jonuleit et al., 2001; Levings et al., 2001; Stephens et al., 2001; Taams et al., 2001). Constitutive expression of CD25 is restricted to Tregs in naïve mice while in humans, only cells that highly express CD25 designate Tregs, and activated T cells acquire intermediate levels of CD25 (Sakaguchi et al., 1995; Baecher-Allan et al., 2001). Importantly, Foxp3 was identified as a lineage-defining transcription factor for Tregs in mice (Fontenot et al., 2003; Hori et al., 2003; Khattri et al., 2003) and humans (Yagi et al., 2004; Roncador et al., 2005). In mice, up to 90% of Tregs express Foxp3, while resting as well as activated conventional T cells do not show detectable levels of Foxp3. Analogous to mice, the majority of human CD4⁺CD25^high Tregs also express Foxp3. However, in contrast to mice, human conventional T cells (Tcons) transiently express intermediate levels of Foxp3 upon activation (Pillai et al., 2007), questioning Foxp3 as a specific marker for human Tregs. While the necessity of Foxp3 for suppressive function of Tregs is undisputed, contradictory reports leave it unclear whether (temporary) Foxp3 expression is sufficient to confer suppressive abilities to human activated T cells. Some studies demonstrated suppressive capacities of Foxp3⁺ activated human T cells (Walker et al., 2003; Pillai et al., 2007) while others did not (Gavin et al., 2006; Wang et al., 2007; Allan et al., 2007). However, activation-induced transient Foxp3 induction in human Tcons was not observed in other studies (Yagi et al., 2004). These discrepancies might be due to differences in culture conditions, such as strength of T cell receptor (TCR) stimulation and presence of IL-2 or TGF-β. For example, Shevach’s group found that TCR-induced Foxp3 expression in human naïve Tcons is indeed dependent on TGF-β, but does not result in a suppressive phenotype (Tran et al., 2007). In addition, his group claimed that a certain antibody clone used for staining of human Foxp3 results in unspecific staining of activated T cells, although this is controversial (Pillai and Karandikar, 2008). Conflicting results may also be ascribed to differences in kinetics and level of Foxp3 expression and, consequently, expression of Foxp3 target genes (Allan et al., 2008; Zheng et al., 2008).

Foxp3⁺ Tregs are not a uniform population but rather comprise several subphenotypes. The majority of human Tregs displays a CCR7⁻CD45RA⁻ effector/memory phenotype (Beyer and Schultze, 2007). However, only the naïve CD45RA⁺ Treg subset maintains Treg phenotype and function upon ex vivo expansion (Hoffmann et al., 2006). A detailed characterization of the human Foxp3⁺ Treg repertoire suggested a further grouping into CD45RA⁺Foxp3^low resting Tregs, CD45RA⁻Foxp3^high activated Tregs, and CD45RA⁻ Foxp3^low cytokine-secreting Tregs, the latter being non-suppressive (Miyara et al., 2009). Further, several subphenotypes of murine Tregs with distinct transcriptional signatures were found in different anatomical locations (Feuerer et al., 2009, 2010). Diverse Treg subpopulations may apply different suppression mechanisms and specifically control certain effector cell types. In this review, we mainly focus on naturally occurring Tregs (nTregs), which are thymus-derived. Yet peripheral tolerance is not only ensured by nTregs but also involves various induced Treg populations (iTregs), which seem to play a role mainly in the intestine (Curotto de Lafaille and Lafaille, 2009).

Tregs can suppress a variety of immune cells including B cells, NK cells, NKT cells, CD4⁺, and CD8⁺ T cells, as well as monocytes and dendritic cells (DCs). In the following, we will focus on suppression of CD4⁺CD25⁻ conventional T cells (Tcons).

Upon cell–cell contact, Tregs inhibit TCR-induced proliferation and IL-2 transcription of Tcons, as shown for murine Tregs already in 1998 (Thornton and Shevach, 1998). Suppression of murine (Ermann et al., 2001) or human (Dieckmann et al., 2001) Tcon proliferation by Tregs can occur directly, i.e., in the absence of antigen presenting cells (APCs). This direct suppression can involve immunosuppressive cytokines or other factors; however, contact-dependent direct suppression has also been described. In addition, Tregs can inhibit Tcons indirectly by influencing the activation status of APCs and therefore activation of Tcons. Based on these properties, a standard assay to assess Treg function is inhibition of responder Tcon proliferation upon stimulation via the TCR in the presence of APCs. A main function of Tregs is suppression of activation and expansion of naïve Tcons, but they can also inhibit activated effector T cells and memory CD4⁺ (Levings et al., 2001) and CD8⁺ (Suvas et al., 2003) T cells. To be suppressive, Tregs themselves have to be TCR-activated in the presence of IL-2 (Takahashi et al., 1998; Thornton and Shevach, 1998; de la Rosa et al., 2004; Thornton et al., 2004a,b) while costimulation via CD28 is dispensable (Takahashi et al., 2000). However, a more recent study questioned the requirement for Treg activation: Szymczak-Workman et al. (2009) showed that TCR-transgenic Tregs were able to suppress Tcons with different antigen specificity in the absence of the Treg-cognate antigen. Differences in the type of APCs used in the assays, the transgenic system, or “pre-activation” by the cell purification procedure might provide an explanation for the controversy. Yet, the study is in line with other reports that show that Tregs, once active, can suppress Tcons independently of antigen, leading to so-called bystander suppression (Thornton and Shevach, 2000; Karim et al., 2005).

Different mechanisms of Treg-mediated suppression have been described, mostly on the basis of in vitro suppression assays. These mechanisms may also operate in vivo depending on the target cell type and activation status as well as the location and cytokine and microorganism milieu of the immune reaction. Thus, the contribution of suppressive mechanisms might be interpreted differently depending on the cell types and their activation state used in in vitro suppression assays. In addition, differences may occur depending on the readout, as suppressing the production of certain effector cytokines or the release of cytotoxic granules in vivo and in vitro can occur without concomitant suppression of proliferation (Mempel et al., 2006; Schmidt et al., 2011; Sojka and Fowell, 2011). For example, using another in vivo model, it was shown that Tcon expansion in response to antigen was suppressed by Tregs relatively late, yet the remaining Tcons still cycled and produced effector cytokines at these late time points (Klein et al., 2003). It was suggested that activation and/or expansion of antigen-specific Tregs may be a prerequisite for suppression of Tcons in vivo, as it might result in sufficient Treg numbers to enable contact with target Tcons at the site of the immune response (Klein et al., 2003).

The described mechanisms of Treg-mediated T cell suppression are summarized in Figure 1 and will be described in more detail in the following sections.

The coinhibitory molecule CTLA-4 is constitutively expressed in murine and human Tregs and exposed on the cell surface upon activation (Read et al., 2000; Takahashi et al., 2000; Dieckmann et al., 2001). A role for CTLA-4 in suppression in vivo has been suggested, since CTLA-4 deficiency or blockade in mice results in spontaneous autoimmunity, which can be ameliorated by Tregs (Bachmann et al., 1999; Takahashi et al., 2000). In addition, CTLA-4 blockade abrogates the protective effects of Tregs in murine colitis models (Read et al., 2000). Also in vitro suppression of murine Tcon proliferation (in the presence of APCs) was shown to be abrogated in some studies by CTLA-4 blockage (Takahashi et al., 2000; Tang et al., 2004). However, CTLA-4 deficient Tregs could still suppress through compensatory mechanisms involving TGF-β and IL-10 in vitro and in vivo (Tang et al., 2004; Read et al., 2006). Similarly, studies with human Tregs do not show a uniform picture: some in vitro studies did not find an involvement of CTLA-4 in Treg-mediated suppression of Tcon proliferation (Baecher-Allan et al., 2001; Levings et al., 2001), while others could partially abrogate suppression by CTLA-4 blockage (Birebent et al., 2004). These discrepancies may be explained by the involvement of CTLA-4 in some but not all aspects of suppression. Our results showed that rapid Treg-mediated suppression of cytokine transcription in human Tcons was unaffected by CTLA-4 blockage, irrespective of the presence or absence of APCs, while suppression of proliferation in the presence of APCs was partially dependent on CTLA-4 (Schmidt et al., 2011).

Regarding CTLA-4 blockade in long-term experiments, it has to be considered that CTLA-4 is also expressed by activated Tcons after 1–2 days of TCR stimulation (Walunas et al., 1994) and thus cell-intrinsic negative signals are involved. These include competition with the costimulatory molecule CD28 for binding to the B7 molecules CD80 (B7.1) and CD86 (B7.2; van der Merwe et al., 1997), transduction of negative signals that induce cell cycle arrest and prevent IL-2 secretion (Krummel and Allison, 1996; Walunas et al., 1996), and limiting T cell contact with APCs (Schneider et al., 2006). Recently, the molecular mechanism of Treg-mediated suppression via CTLA-4 was elucidated in more detail. Downregulation of CD80/CD86 expression on APCs by Tregs (Cederbom et al., 2000) was shown to be partly dependent on CTLA-4 and the adhesion molecule LFA-1, thereby indirectly inhibiting Tcon activation by APCs in vitro (Oderup et al., 2006; Onishi et al., 2008) and in vivo (Wing et al., 2008). Mechanistically, downregulation of CD80/CD86 on target cells by CTLA-4-expressing cells can involve capture of these ligands by CTLA-4, a process called trans-endocytosis (Qureshi et al., 2011). Another study supports the importance of CTLA-4 in in vivo suppression without direct effects on the responder T cells (Friedline et al., 2009). Strikingly, mice with Treg-specific ablation of CTLA-4 spontaneously develop fatal autoimmune disease (Wing et al., 2008), underlining the importance of CTLA-4 expression on Tregs for the maintenance of peripheral tolerance.

In contrast to the studies showing necessity of CTLA-4 for Treg-mediated suppression, Tran et al. (2009) applied a mixed coculture system with murine and human cells and showed that human Tregs can suppress proliferation of murine Tcons by targeting murine DCs through LFA-1:ICAM-1 interaction between Tregs and DCs. Suppression was only evident in the presence of APCs, and it was not abrogated when human CTLA-4 was blocked. Interestingly, murine Tregs could not suppress human Tcons in the presence of human APCs. These results argue for species-specific differences in suppressive molecules, especially regarding CTLA-4.

Certain human and murine DC subsets express the enzyme indoleamine 2,3-dioxygenase (IDO) that catalyzes degradation of tryptophan to kynurenine, leading to starvation of effector T cells and also to direct cell cycle arrest. At the same time, IDO leads to iTreg generation (Fallarino et al., 2003, 2006; Curti et al., 2007). Tregs themselves can, via CTLA-4-induced signaling, further increase IDO expression in DCs. Additionally it was shown that murine Tregs diminish glutathione synthesis in DCs, presumably CTLA-4-mediated, which leads to a redox environment unfavorable for Tcon proliferation (Yan et al., 2009, 2010). In conclusion, the importance of CTLA-4 in Treg-mediated suppression is undisputed, but it does not seem to be the only suppressive mechanism, and also species differences may exist. Importantly, CTLA-4 blockage has already shown promising outcomes in clinical trials for metastatic melanoma, presumably due to its effects on both effector T cells and Tregs (Peggs et al., 2009). Despite the prominent role for CTLA-4 in Treg-mediated inhibition of DC activation, involvement of the costimulatory molecule CD40, the deubiquitinase A20 acting as antigen-presentation attenuator, and the surface receptor neuropilin-1 has also been described (Hanig and Lutz, 2008; Sarris et al., 2008; Song et al., 2008). Further, the adhesion molecule LAG-3 is expressed on Tregs, binds to MHC class II molecules and was proposed to be involved in suppression, since deficiency or blockage of LAG-3 resulted in disturbance of T cell homeostasis in vivo and reduced suppressive activity of Tregs in vitro (Huang et al., 2004; Workman and Vignali, 2005). These effects might be mediated by LAG-3-induced inhibition of DC activation (Liang et al., 2008). However, LAG-3 deficient mice do not show signs of autoimmune disease. Further, MHC class II expression on an activated human Treg subpopulation that exerts rapid contact-dependent suppression seems to be important (Baecher-Allan et al., 2006) as blockage of MHC class II on activated Tregs was shown to abrogate suppression (Peiser et al., 2007).

Tregs do not only reduce antigen presenting activity of APCs but also support an immunosuppressive cytokine milieu by reducing IL-6 while increasing IL-10 production by DCs (Veldhoen et al., 2006). In addition to these mechanisms, Tregs can inhibit Tcon activation by reduction of contact formation between Tcons and DCs (Tadokoro et al., 2006). As already mentioned, there is mutual interplay between Tregs and tolerogenic DCs, the latter acting in induction and expansion of Tregs.

The role of immunosuppressive cytokines in Treg-mediated suppression is still incompletely understood. Despite the importance of TGF-β and IL-10 in several in vivo models, these cytokines seem to be dispensable for other disease models such as autoimmune gastritis (Suri-Payer and Fritzsching, 2006) as well as for most in vitro systems. Treg-mediated suppression via cytokines is described in more detail below.

One efficient way to repress immune responses would be killing of effector cells by Tregs, which was indeed observed in certain settings. It was shown that upon CD3/CD46 activation, human nTregs express the serine protease granzyme (Gzm) A and kill CD4⁺ T cells and other target cells in a perforin-dependent manner, yet independent of the death receptor CD95 (Grossman et al., 2004). In another report, contact-dependent suppression of Tcon proliferation by activated murine Tregs in vitro was described to be partially GzmB-dependent; however, in contrast to the results of Grossman et al. perforin was not involved here (Gondek et al., 2005). As suggested from mouse studies, GzmB-mediated suppression might be important in vivo to maintain transplant tolerance (Gondek et al., 2008), but it might counteract tumor clearance by killing tumor-reactive NK cells and CD8⁺ T cells (Cao et al., 2007). Involvement of the death receptor TRAIL in Treg-mediated suppression of activated mouse Tcons was shown in vitro and in vivo in a transplantation model, but required strong pre-activation of Tcons and Tregs (Ren et al., 2007).

The importance of IL-2 consumption by Tregs due to their high CD25 expression remains controversial. A study by Lenardo’s group proposed that Tregs induce IL-2 deprivation-mediated apoptosis in mouse Tcons, which was dependent on close proximity between the cells (Pandiyan et al., 2007). However, the authors did not observe suppression of IL-2 production during the first 48 h of coculture, which is not in line with many other studies (Thornton and Shevach, 1998; Barthlott et al., 2005; Sojka et al., 2005). Pandiyan et al. (2007) found that T cells deficient in the proapoptotic protein Bim could not be suppressed in vitro and in vivo in a T cell transfer-induced IBD model. In contrast to these results, a recent study by Vignali’s group did not find involvement of cell death induction by Tregs (Szymczak-Workman et al., 2011). Szymczak-Workman et al. used responder Tcons that are resistant to cytokine withdrawal-induced apoptosis due to Bim deficiency or Bim/Puma double knockout or Bcl-2 overexpression and performed in vitro and in vivo assays to conclude that Tregs do not suppress via induction of apoptosis in responder Tcons. In addition to these mouse studies, we (Oberle et al., 2007) and others (Vercoulen et al., 2009) also did not observe apoptosis induction in human Tcons by human Tregs.

IL-2 does not only promote survival but also proliferation of T cells, and IL-2 consumption seems to be involved in reducing the positive effects of IL-2 on proliferation and IL-2 expression, because exogenous IL-2 could abrogate Treg-mediated suppression of proliferation and/or of IL-2 production in vitro (Thornton and Shevach, 1998; Baecher-Allan et al., 2001; de la Rosa et al., 2004). However, during inhibition of murine Tcons by human Tregs, IL-2 consumption was not causative for suppression of proliferation, as blocking human CD25 did not abrogate suppression under optimal stimulation conditions for the human Tregs (Tran et al., 2009). Of note, under suboptimal stimulation conditions, human Tregs required CD25 signals in order to be functionally suppressive, yet the suppression mechanism under these conditions remains elusive and may or may not involve IL-2 consumption (Tran et al., 2009). We found that, although exogenous IL-2 partially abrogated suppression of human Tcon proliferation by human Tregs, rapid suppression of IL-2 transcription by human pre-activated Tregs was not affected (Oberle et al., 2007), which argues for different suppression mechanisms for proliferation and cytokine transcription. Here, IL-2 itself led to increased IL-2 mRNA expression (Oberle et al., 2007), however, for later time points, negative feedback of IL-2 on IL-2 mRNA and protein expression was described (Villarino et al., 2007). These differences in the effect of IL-2 on IL-2 expression may depend on time point, species, amount of IL-2 and cellular activation status and may contribute to the inconsistent results on the role of IL-2 consumption in Treg-mediated suppression. Together, the role of IL-2 consumption remains controversial and may depend on the specific setting and stimulation conditions of the Tregs. IL-2 consumption by Tregs may even enhance certain effector T cell reactions, which was shown for Th17 responses (Pandiyan et al., 2011). In addition, the link of IL-2 consumption to apoptosis is unclear, as exogenous IL-2 could rescue proliferation of suppressed human Tcons without reducing background apoptosis (Vercoulen et al., 2009).

Together, it can be concluded from these studies that IL-2 consumption by Tregs and apoptosis induction in suppressed target cells may only play a role in suppression under certain circumstances.

Tregs may be specialized to suppress a certain CD4⁺ T cell subset in particular by expression of the hallmark transcription factor of this very subset. In this regard, it was shown that a Treg subset upregulated T-bet in response to IFN-γ, which was essential for the control of Th1-mediated inflammation (Koch et al., 2009). Similarly, IRF-4 (which is typically made by Th2 and Th17 cells) expression in Tregs was required for suppression of Th2 responses (Zheng et al., 2009). Moreover, expression of STAT3 (which drives Th17 differentiation) in Tregs was increased and crucial during control of Th17-mediated intestinal pathology (Chaudhry et al., 2009). Adding to that list, loss of the Th2-associated transcription factor GATA-3 in Tregs led to autoimmune disease, in that case accompanied by destabilization of Foxp3 expression and elevated Th1, Th2, and Th17 cytokine levels (Wang et al., 2011). GATA-3 deficient Tregs tended to convert into a Th17 phenotype, and it was further shown that GATA-3 was upregulated upon stimulation in murine as well as human Tregs (Wohlfert et al., 2011). However, the mechanism through which Tregs suppress via expression of effector T cell-specific transcription factors is currently unclear but might involve competition for limiting factors. The transcription factors could directly interact with Foxp3, as described for IRF-4, STAT3, and RORγt. Expression of the effector T cell-specific transcription factors and their regulated genes such as chemokine receptors might also enable Tregs to migrate to and proliferate at the site of immune response. Accordingly, suppressor activity of IRF-4 or STAT3 knockout Tregs in vitro was shown to be unimpaired, and GATA-3 was selectively expressed by Tregs at environmental interfaces.

In contrast, expression or activation of Th-specific transcription factors in Tregs might lead to a loss of Foxp3 expression and suppressive function. For example, STAT3 was described to mediate downregulation of Foxp3, and intriguingly, STAT3 is activated by several cytokines which negatively regulate Tregs and Foxp3, such as IL-6, IL-23, or IL-27 (Yang et al., 2007, 2008; Yao et al., 2007; Huber et al., 2008; Ahern et al., 2010). Thus, loss of STAT3 expression in T cells during T cell transfer-induced colitis and systemic inflammation increased Treg induction and ameliorated disease (Durant et al., 2010).

Plasticity between CD4⁺ T cell subsets might be higher than originally anticipated, although the stability of the Treg lineage in vivo is highly controversial (Zhou et al., 2009; Rubtsov et al., 2010). Yet several lines of evidence indicate that plasticity is particularly pronounced between Th17 cells and iTregs (Lee et al., 2009), and an intermediate state with expression of both RORγt and Foxp3 at the same time was observed (Lochner et al., 2008; Zhou et al., 2008). In certain cytokine milieus, even induction of IL-17 was described in murine and human Tregs, which might retain suppressive function (Xu et al., 2007; Koenen et al., 2008; Osorio et al., 2008; Beriou et al., 2009; Voo et al., 2009).

Hydrolysis of extracellular ATP to ADP or AMP by the ectoenzyme CD39, expressed by all murine Tregs and by about 50% of human Tregs, represents another Treg-mediated anti-inflammatory mechanism (Borsellino et al., 2007). CD39 knockout Tregs showed reduced suppressive capacities in vitro and in vivo (Deaglio et al., 2007). In human Tregs, CD39 expression was suggested to identify a highly suppressive Treg subset (Mandapathil et al., 2009), and suppression of Tcon proliferation by this subset could be partially abrogated by blockage of ectonucleotidase activity (Mandapathil et al., 2010). CD73, also expressed by Tregs, further degrades AMP to adenosine (Kobie et al., 2006). Adenosine signals via the A2A adenosine receptor and may inhibit DCs as well as activated T cells, e.g., by elevation of cyclic AMP (cAMP; Ernst et al., 2010). In human but not mouse T cells, ATP can also directly increase cAMP levels through the P2Y₁₁ receptor (Abbracchio et al., 2006). In vivo, A2A receptor signaling might induce anergy and promote iTreg generation (Zarek et al., 2008). In conclusion, generation of adenosine seems to be important for the suppressive function of certain Treg subsets.

Tregs produce high levels of cAMP and cAMP amounts were shown to rise in suppressed Tcons upon contact with Tregs, presumably via transfer through gap junctions from murine Tregs to Tcons (Bopp et al., 2007). Accumulation of high cAMP amounts in Tregs might require Treg pre-activation (Bazhin et al., 2010). cAMP seems to be an important component of Treg-mediated suppression, since suppression of IL-2 transcription and proliferation could be partly abrogated by a cAMP antagonist or a gap junction inhibitor (Bopp et al., 2007). The same group recently also showed that induction of cAMP in murine DCs upon Treg coculture contributes to suppression of DCs (Fassbender et al., 2010). The authors suggest that in suppressed Tcons, cAMP acts by inducing expression of the inducible cAMP early repressor (ICER), which functions as a repressor, e.g., at the IL-2 and IL-4 gene loci (Bodor et al., 2000). ICER protein and mRNA were previously shown to be increased in suppressed mouse T cells after 17–20 h of Tcon:Treg coculture stimulation (Bodor et al., 2007a) which correlated with an increase in cAMP (Bopp et al., 2007). In these studies, an increase in ICER and cAMP was not yet detectable after 4–5 h. In human Tcons, we also did not observe ICER upregulation upon coculture with Tregs compared to coculture with Tcons within 5 h (Oberle et al., 2007; Oberle and Schmidt, unpublished data). Since NFAT1/4 double knockout Tcons cannot be fully suppressed by Tregs in long-term suppression assays, such as proliferation or IL-2 mRNA and protein expression after 18 h (Bopp et al., 2005), it was proposed that NFAT forms inhibitory complexes on cytokine promoters with transcriptional repressors such as ICER, PPRγ, or p21^SNFT (Bandyopadhyay et al., 2007; Bodor et al., 2007a,b). However, by calculating the percentage of suppression, partial inhibition can be observed although NFAT1/4 double knockout T cells display hyperproduction of IL-2, suggesting that NFAT1/4 double knockout Tcons are less susceptible, but not completely resistant to Treg-mediated suppression. Furthermore, although ICER failed to accumulate in suppression assays with B7 deficient responder T cells (as ICER induction in suppressed T cells was CTLA-4/B7 and GITR-dependent), residual suppression was still detectable (Bodor et al., 2007a). These results argue for additional suppressive mechanisms together with those mediated by CTLA-4 and ICER:NFAT. More recently, it was proposed that NFAT2 together with ICER translocates to the nucleus to confer repression of the IL-2 promoter in murine T cells upon coculture with Tregs (Vaeth et al., 2011). However, depending on the experimental setting, either increased or decreased NFAT2 nuclear translocation was observed in suppressed compared to control mouse T cells, while ICER clearly translocated into the nucleus upon Treg-mediated suppression (Vaeth et al., 2011).

Together with a report showing that Cbl-b deficient T cells are less sensitive to suppression by Tregs (Wohlfert et al., 2004), data on inhibitory NFAT complexes imply that Treg-suppressed Tcons display features of anergy, as both NFAT1 and Cbl-b are crucial in anergy induction (Macian et al., 2002; Soto-Nieves et al., 2009). Anergy induction in suppressed Tcons might occur later than inhibition of cytokine transcription since we did not detect accumulation of ICER mRNA or other anergy-related genes up to 5 h after TCR stimulation in suppressed compared to control human Tcons (Oberle et al., 2007). By comparing gene expression of suppressed and anergized murine Tcons, Sukiennicki and Fowell (2006) found only partial overlap of NFAT-dependent gene expression while most genes did not overlap. However in their study, changes in gene expression were not yet pronounced 12 h after coculture, but only evident after 36 h. This is in contrast to the gene array performed with human T cells in our group, in which, e.g., NFAT and NF-κB target genes were already suppressed in human Tcons within 1.5–3 h of coculture; however, this gene array was performed with pre-activated Tregs (Schmidt et al., 2011). Thus, discrepancies may be due to Treg pre-activation or species.

In contrast to its role in anergy induction, NFAT also plays an important activating role in TCR-induced transcription of T cell effector cytokines. This rapid TCR-induced NFAT activation in Tcons is suppressed by Tregs, which will be described in the next section.

Signaling events and in particular TCR signaling in responder Tcons upon suppression by Tregs has not been studied extensively so far. Yet detailed knowledge might be important for therapeutic interventions in certain T cell-mediated diseases, as resistance of effector T cells to Treg-mediated suppression frequently occurs in autoimmune diseases (Buckner, 2010). However, kinetics of cytokine suppression has been analyzed in detail. Our group previously showed that inhibition of cytokine transcription occurs rapidly within 1–3 h (Oberle et al., 2007). This suppression of cytokine transcription was also observed when stronger TCR/CD28 stimulation was applied and occurred within 30–45 min when pre-activated Tregs were used (Schmidt et al., 2011). Thornton and Shevach (1998) showed potent inhibition of IL-2 mRNA production in murine Tcons by Tregs, though the authors did not analyze earlier than 15 h. Further experiments with murine Tcon:Treg cocultures showed that IL-2 mRNA suppression starts between 5 and 6 h of coculture stimulation (Barthlott et al., 2005) or even later, after 6–12 h (Sojka et al., 2005). In a subsequent paper, the latter group did not observe IL-2 mRNA suppression after 12 h of coculture (Sukiennicki and Fowell, 2006). In contrast to our experiments, these studies were conducted in the presence of APCs and with murine T cells. Thus, discrepancies in IL-2 mRNA suppression kinetics might be due to species differences or experimental setup regarding strength of stimulation and presence of APCs. There may also be differences in responder Tcon susceptibility to suppression, which could be influenced by the cell purification procedure and resting time between purification and initiation of cocultures. In addition, we used pre-activated Tregs, while the others did not. Nevertheless, when Sojka et al. used pre-activated Tregs, they still observed suppression of IL-2 secretors not earlier than 4–6 h after initiation of coculture. However, at earlier points in time, the induction of IL-2 also was still very low (Sojka et al., 2005).

To analyze Treg-mediated effects on Tcons, we favored an in vitro cell culture system without APCs. Thus, we could analyze direct effects of Tregs on Tcons and we obtained pure populations of responder Tcons after coculture. However, it is still a matter of debate whether the main targets of Treg-mediated suppression in vivo are T cells or rather APCs. It remains possible that Tregs utilize distinct mechanisms to directly suppress Tcons or APCs, and both mechanisms might play a role in vivo. Although several studies have shown that murine as well as human Tregs inhibit TCR-induced proliferation of Tcons in the absence of APCs (Dieckmann et al., 2001; Ermann et al., 2001), others found that murine Tregs fail to inhibit Tcons stimulated by anti-CD3 antibodies (Thornton and Shevach, 2000). In general, significant suppression in the absence of APCs is mostly observed at high (1:1) Treg to Tcon ratios, which makes cell density very important to control. Nevertheless, as APC-containing cocultures may represent a more physiological setting than Tcon:Treg cocultures without APCs, we also used another in vitro cell culture system to show that rapid suppression of cytokine transcription can occur in the presence of APCs (Schmidt et al., 2011).

Transcription of several effector T cell cytokines is initiated upon TCR stimulation and costimulation by the cognate antigenic peptide presented on a mature APC or, artificially, by agonistic anti-CD3 and anti-CD28 antibodies. As we observed rapid suppression of cytokine transcription, we questioned whether Tregs directly influence TCR signaling. Until recently, there was no comprehensive study on TCR signaling events in Tcons upon suppression by Tregs. TCR stimulation leads to a cascade of protein phosphorylation and dephosphorylation events, which ultimately activates transcription factors to initiate a distinct transcriptional program (Lin and Weiss, 2001; Smith-Garvin et al., 2009). The most important transcription factors for the induction of cytokine expression are NF-κB, NFAT, and AP-1, which together mediate activation of several cytokine genes, e.g., IL-2 and IFN-γ (Jain et al., 1995; Young, 1996; Sica et al., 1997). Before we describe our results regarding TCR signaling in Treg-suppressed Tcons, we will give a brief introduction into TCR signaling.

Proliferation and activation of T cells does not only require TCR stimulation but also a second, costimulatory signal (Acuto and Michel, 2003). CD28, the most prominent costimulatory receptor, is expressed on naïve and activated T cells and is triggered by its ligands CD80 and CD86, which are expressed on activated APCs and at low levels on long-term activated T cells. Phosphoinositide 3-kinase (PI3K), which generates phosphatidylinositol 3,4,5-trisphosphate (PIP₃), is one of the key effectors downstream of CD28 and facilitates activation of PDK1 and one of its targets, the kinase Akt (protein kinase B, PKB). PTEN is a lipid phosphatase that negatively regulates the PI3K/Akt pathway by dephosphorylation of PIP₃ into PIP₂. TCR-induced activation of the transcription factors AP-1, NFAT, and in particular NF-κB is directly influenced and amplified by CD28 signaling.

Thus far, the influence of Tregs on CD28 signaling in Tcons is unclear, although several hints from the literature support the hypothesis that Tregs might suppress CD28 signaling in addition to TCR signaling: strong activation or strong CD28 costimulation can (partially) overcome suppression of Tcons by Tregs. This was shown for suppression of IL-2 mRNA and protein as well as for proliferation of murine T cells (Thornton et al., 2004a; Sojka et al., 2005), and also for proliferation and IFN-γ secretion of human Tcons (Baecher-Allan et al., 2002), all performed in the presence of APCs. We used CD28 costimulation to achieve detectable phosphorylation of TCR signaling proteins and still observed Treg-mediated suppression. Yet, we have indications that suppression of Ca²⁺ and of proliferation was abrogated upon increasing the strength of stimulation (unpublished observation).

Reversal of Treg-mediated suppression by CD28 costimulation might involve activation of PI3K/Akt, since different studies imply that hyperactivation of the PI3K/Akt pathway renders Tcons resistant to Treg-mediated suppression (Wohlfert and Clark, 2007). For example, T cell-specific TRAF6 deletion led to hyperactivation of the PI3K/Akt pathway, resistance to Treg-mediated suppression and T cell-mediated autoimmune disease (King et al., 2006). However, PI3K/Akt-induced autoimmune disease might also be attributed to impaired CD95-mediated T cell apoptosis (Ohashi and Yeh, 2006). In addition, partial resistance to Treg-mediated suppression was conferred by expression of constitutively active Akt in T cells (Pierau et al., 2009) or PI3K/Akt activation by IL-15 (Ben Ahmed et al., 2009). These authors did not find involvement of apoptosis in contrast to Pandiyan et al. (2007) who claim that Tregs deprive responder Tcons of IL-2 or IL-15 and, thus, reduced Akt activation and survival of murine Tcons was observed. Similarly, a slight downregulation of Akt phosphorylation could be demonstrated in CD8⁺ T cells upon coculture with Tregs, while ZAP-70 was unaffected (Kojima et al., 2005). Furthermore, it was found that effector T cells from juvenile arthritis patients are resistant to suppression due to Akt hyperactivation (Wehrens et al., 2011). In addition, Cbl-b knockout T cells are resistant to suppression (Wohlfert et al., 2004), which might involve functions of Cbl-b in anergy as well as its negative influence on CD28/PI3K signaling. Also NFAT1/4 double knockout T cells, which are partially resistant to suppression (Bopp et al., 2005), do not require CD28 costimulation to be activated (Ranger et al., 1998), suggesting PI3K/Akt hyperactivation in these cells.

Since the PI3K/Akt pathway crosstalks with all main TCR signaling pathways, it might be involved in Treg-mediated rapid suppression. It remains to be analyzed whether hyperactivation of the PI3K/Akt pathway also abrogates Treg-mediated suppression of Ca²⁺, NF-κB, and NFAT. Further, it remains to be shown whether Tregs might act via inhibition of PI3K/Akt signaling in human Tcons. However, it is unclear how this would directly translate into reduced Ca²⁺ store depletion without concomitant reduction of, e.g., PLCγ activity. Analyses of known and unknown pathways downstream of CD28 in Tcons upon suppression by Tregs remain subject to further investigation.

A growing body of evidence indicates that Tregs do not use only one universal mechanism of suppression, but rather an arsenal of different ones. So far, it is unclear how a Treg “decides” which mechanism to apply, and whether it can switch from one to the other and/or apply several modes of suppression simultaneously. Mechanisms of suppression as well as target T cell susceptibility to suppression likely differ depending on tissue site, cell types involved and activation status of target cell and Treg. Furthermore, different Treg subsets exist and further research should reveal whether these are specialized on a particular suppressive mechanism.

In the commonly used in vitro suppression assays, not all aspects of suppression can be detected with a single readout or point in time, as suppression of proliferation does not necessarily reflect suppression of effector cytokine production. In addition, effects of Tregs on APCs have to be addressed if these are included in the assay. Furthermore, it was shown by several groups that Tregs have to be TCR-activated in order to be suppressive, which might be particularly important for assays assessing rapid suppression and, thus, leaving little time for the Tregs to get activated during stimulation of the coculture. Therefore, it is important to control for potential effects of the pre-activating reagent itself. A possible control for effects of the pre-activating reagent is the use of pre-activated Tcons in comparison with pre-activated Tregs, however, human pre-activated Tcons themselves might upregulate Foxp3 and/or acquire suppressive function. Another possibility is the use of pre-activating reagents which can be removed completely after Treg pre-activation, such as covalently platebound anti-CD3 antibodies.

The knowledge of molecules and signaling pathways affected in Tcons upon suppression by Tregs might be crucial for their therapeutic manipulation in the future. In cancer, suppression of effector T cells is deleterious and breaking the suppressive state is highly desirable. In contrast, a suppressed state of autoreactive T cells is warranted during autoimmune disease. Further research is required to elucidate which mechanisms of Treg-mediated suppression or of target T cell resistance to suppression are most important in a particular disease, and possible therapeutic interventions have to be performed extremely carefully.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.