It is now clear that CD4+ CD25+ Tregs can be derived from two sources, namely those developing within the thymus (whose contribution may therefore diminish with age) and those generated in the periphery. Thymically derived Tregs are thought to originate at the transition between the double-positive and single-positive stages following encounter between thymocytes that bear high affinity TCRs for self-peptide with their cognate antigens [165,187], but this assertion has been challenged recently by observations that thymic commitment to a Treg phenotype may occur at an earlier developmental stage . The autoreactive Treg repertoire may be entrained by deletion following interaction with endogenous superantigens and APCs of both thymic and bone marrow origin , but the peripheral Treg repertoire retains a higher frequency of autospecific than alloreactive cells . How Treg precursors commit to a Treg lineage in the thymus is unknown, but recent evidence points (in mice) to an interaction with a gene locus intimately linked with the MHC  (characterization of this locus and the genes involved is awaited) and may involve an important role for IL-2 signalling  and/or CD28 engagement .
Adults rendered temporarily lymphopenic have a greater propensity to develop autoimmune diseases [192,193]. Although this may reflect loss of a significant proportion of thymically derived Tregs (which are hard to regenerate given age-related thymic atrophy [194,195]) leading to loss of self-tolerance, one cannot ignore the fact that not everyone who is made lymphopenic develops autoimmune disease. One possibility is that the important determinant for maintenance of tolerance to self-components may be the relative frequency of effector cells to Tregs, as some chemotherapy agents have an equal effect on both . However, depletion of CD25+ T cells from mice  or the adoptive transfer of naive T cells into lymphopenic recipients  is not sufficient for the development of autoimmune phenomena. The second possibility is that Tregs are generated in the periphery, an attractive notion that is supported by data showing reconstitution of the CD4+ CD25+ Treg population through conversion of CD4+ CD25– T cells [199,200]. Although this is not a robust phenomenon [12,157], there are suggestions that discrepancies may be the result of competition between CD4+ CD25– and CD4+ CD25+ T cells (i.e. that the rate of conversion is related to the relative frequency of the two cell types)  and that the number of Tregs may be linked to the availability of IL-2 (and, by inference, the number of IL-2-producing effector cells) . Studies of human Treg populations have shown these populations to be highly proliferative and senescent in vivo with very short telomeres , which is consistent with their memory/CD45RO+ phenotype . Their susceptibility to apoptosis and short telomeres (with low telomerase activity) means it is unlikely that they are capable of self-renewal; the more likely explanation is that Tregs are generated in the periphery.
Other studies have corroborated the importance of IL-2 for the development of Tregs and are summarized in the review by Malek et al. . Indeed, in an animal model of graft-versus-host disease (GvHD) where autoreactive T cells from donors deficient in Tregs (DO11·10 Rag–/– animals) are infused into athymic antigen-expressing lymphopenic recipients (sOVA transgenic thymectomized, lethally irradiated mice reconstituted with Rag–/– bone marrow), development and recovery from disease is reliant upon generation of antigen-specific effector cells followed by de novo generation of peripheral Tregs in a concomitant, IL-2-dependent manner . The added implication of these findings is that both functional effector cells and Tregs can develop in parallel from the same population of T cells in response to a single antigen in the periphery.
The mechanism(s) by which Tregs are generated in the periphery are unknown. However, there are indications that, in the same manner as with Th1/Th2 cell polarization, the antigenic stimulus (linked to the amount of antigen present as well as the strength of interaction) may determine the commitment to a Treg phenotype (low doses/weaker stimulus result in more Treg generation) [207,208]. The mode of antigen encounter, namely through (immature or suboptimally activated) dendritic cell presentation, also seems important for the conversion of naive T cells to a Treg phenotype [208,209], as may interactions with anti-inflammatory molecules such as thrombospondin-1 . The demonstration that TGF-β-deficient mice have reduced numbers of peripheral [211,212] but not thymic Tregs and that TGF-β assists the conversion of naive CD4+ CD25– T cells into Tregs both in vivo and in vitro[214,215] argue in favour of the importance of this cytokine in the generation and maintenance of the peripheral Treg pool. The potential importance of these observations to human disorders will be discussed in the context of Th17 development below.
In mice, a discrete population of CD4+ helper T cells has been described as the predominant source of IL-17. These cells have been named Th17 cells. The initial basis of this nomenclature is the dichotomous effects of IL-12 and IL-23 (both members of the same family of IL-12 cytokines, sharing a common IL-12p40 subunit but differing second subunits, IL-12p35 and IL-12p19, respectively ) on the cytokine profile of CD4 cells. While IL-12 (signalling through signal transducer and activator of transcription, STAT-4) had been known to allow lineage commitment towards a Th1 phenotype producing IFN-γ as its signature cytokine (via the transcription factor T-bet[8,9]) [5,7,217] and inflammatory diseases had been viewed along the lines of a Th1/Th2 paradigm (e.g. IL-12p40 neutralization in mice ameliorates inflammatory diseases [218,219]), mice deficient in IFN-γ or IFN-γ signalling remained, paradoxically, susceptible to development of EAE and CIA [15,16]. Furthermore, IL-12p40 (lacking both IL-12 and IL-23) and IL-12p19 (lacking IL-23)-deficient mice were protected against EAE and CIA whereas IL-12p35 (IL-12)-deficient strains remained susceptible [220,221] suggesting that IL-23 rather than IL-12 was important in mediating the pathogenesis of these conditions. Shortly after these observations, it was reported that IL-23 stimulates production of IL-17 from a population of memory (but not naive) CD4+ T cells in a manner that does not exhibit elevation of IFN-γ[21,222] and that IL-17 is linked to the inflammation seen in CIA and EAE [20,21,72]. Further evidence that IL-17 was derived in mice from a discrete population of Th cells that were distinct from the Th1 lineage, termed Th17 cells, were provided by publications showing resistance of Th1 and Th2 cells in vitro to proliferation or production of IL-17 following stimulation with IL-23 and that development of IL-17-producing cells was inhibited by the presence of IFN-γ and/or IL-4 in the culture supernatant [223,224]. It has been proposed that production of IL-12 has greater importance for systemic responses and immunity to intracellular pathogens  while IL-23, produced from activated human macrophages and dendritic cells (DC) , has a more important role for mediating mucosal immune pathology through the promotion of Th1 and Th17 cytokine profiles, respectively .
It is now known that IL-6 [17–19] (see below), and not IL-23 [17,18], is critical for the induction of Th17 lineage commitment (which is supported by the fact that the IL-23 receptor is expressed exclusively on activated and memory T cells [222,228]), while IL-23 seems to be important for the selective expansion of these cells and production of IL-17 . Indeed, other cytokines including IL-2, IL-15, IL-18 and IL-21 can also stimulate IL-17 production from (activated) human T cells and peripheral blood mononuclear cells (PBMC) [226,229], while IL-12 potently inhibits it . Development of Th17 cells is dependent upon correct co-stimulation (ICOS and CD28 ) and the absence of IFN-γ and IL-4, both of which are inhibitory . Furthermore, Th17 lineage differentiation can be inhibited by the Th1-specific transcription factor T-bet in the context of IL-4 blockade  and is characterized by the expression of the orphan nuclear receptor RORγT. Because IL-12 specifies lineage commitment to Th1 and has a stimulatory effect on IFN-γ secretion by Th1 cells , IL-12 may play a critically important role as a regulator of the balance between Th1 and Th17 responses. This assertion is supported by in vivo mouse data in which IL-23 and IL-12 had divergent pro- and anti-inflammatory roles in a model of collagen-induced arthritis . It is important to state that, despite these observations, the description of discrete Th17 cells is mouse-specific and to date no committed Th17 cells have been demonstrated in humans.
Figure 2 shows the cytokine network that is thought to be important in the development and expansion of Th17 cells and the dichotomous Th1/Th17 cytokine profile engendered by these cytokines. Three recent papers have shed some light on the mechanisms by which naive precursor T cells commit to a Th17 phenotype in mice [17–19]. The first, a publication by Veldhoen et al., showed that naive CD4+ T cells could be skewed towards a Th17 phenotype in the presence of dendritic cells and Tregs in an inflammatory milieu (lipopolysaccharide stimulation) . Absence of Tregs leads to Th1 differentiation, presumably through interaction of naive T cells with DCs producing IL-12. In the presence of Tregs and DC, the important drivers of Th17 differentiation were Treg-derived TGF-β and DC-derived IL-6, although both TNF-α and IL-1β (both DC-derived) also augmented the commitment to Th17. In this series of experiments, the IL-17-producing cells did not express T-bet or GATA-3 and addition of IL-12 and IL-4 or IL-18 inhibited Th17 development, but the most important determinant of commitment to a Th17 lineage was the presence of TGF-β, without requirement for cell-to-cell contact. These data were corroborated by Bettelli et al., who demonstrated using cells from a Foxp3-GFP knock-in mouse strain that differentiation towards Treg and Th17 phenotypes were mutually exclusive − activation of naive precursor cells using anti-CD3 in the presence of TGF-β lead to production of green fluorescent protein (GFP)+ cells (i.e. Tregs) as per previous observations, but activation in the presence of IL-6 in addition to TGF-β completely abrogated this and led to development of Th17 cells (that were GFP–). The differentiated cells were, respectively, functionally suppressive and inflammatory and development of the Th17 phenotype was independent of IL-23 . The third paper, by Mangan et al. published simultaneously , showed that addition of TGF-β to naive CD4+ T cells resulted in the development of Th17 cells, an effect which was augmented in the presence of neutralizing antibodies to Th1 and Th2 polarizing cytokines (IL-4 and IFN-γ) or the use of CD4 cells from IFN-γ deficient animals . Furthermore, they demonstrated that TGF-β up-regulated expression of the IL-23 receptor (which may explain the responsiveness of the Th17 population to IL-23). As before, supplementation of the culture conditions with exogenous IL-6 resulted in loss of all Foxp3+ cells, while blockade of IL-6 enhanced Treg development. Again, these findings point to mutually exclusive pathways for Th17 and Treg development based on the availability of TGF-β and IL-6. A schematic for naive T cell commitment is represented in Fig. 3. It should be noted, once again, that these data are derived from mice and whether this pathway exists in humans has not been determined.
Figure 2. Model of mouse helper T cell (Th) commitment to Th1, Th17 and T regulatory cell (Treg) phenotypes following encounter with antigen. Production of transforming growth factor (TGF)-β by naturally occurring Tregs leads to lineage commitment of precursor helper T cells (Thp) towards Treg phenotypes. Stimulation of dendritic cells (DC) by microbial antigens causes production of interleukin (IL)-6, IL-23 and/or IL-12. Predominant production of IL-12 promotes commitment of Thp to a Th1 phenotype while IL-6, in combination with Treg-derived TGF-β promotes skewing of Thp towards a Th17 phenotype. IL-23 produced by DCs causes proliferation and cytokine production by Th17 cells.
Download figure to PowerPoint
Figure 3. T helper cell commitment towards specific lineages in mice. T helper cell precursors (Thp) can be skewed towards mutually exclusive Th1, Th2, Th17 and T regulatory cell (Treg) phenotypes on the basis of the cytokine environment. Presence of interleukin (IL)-12 promotes skewing towards Th1 commitment by signalling through signal transduction and activator of transcription (STAT)-4. Th1 cells are characterized by expression of T-bet and produce interferon-γ and tumour necrosis factor-α. Th2 cell commitment is promoted by IL-4 via STAT-6 signalling. Th2 committed cells are characterized by expression of GATA-3. Development of Treg and Th17 phenotypes both require the presence of transforming growth factor-β but the presence of IL-6 preferentially skews the response towards a Th17 phenotype. Tregs are characterized in mice by expression of Foxp3.
Download figure to PowerPoint