Setting the threshold for extra-thymic differentiation of Foxp3⁺ Tregs: TGF-β-dependent and T-cell autonomous

Foxp3⁺ regulatory T cells (Tregs) play a pivotal role in limiting inflammation, autoimmunity, and in maintaining immune homeostasis (reviewed in 1). The relative contribution to these processes of Tregs developing intra-thymically (naturally occurring Tregs) and those developing in the periphery is unknown. Within the gut, specialized DCs support the TGF-β- and retinoic acid-dependent conversion of conventional T cells into Tregs, a process which may play an important role in controlling intestinal homeostasis 2. In lymphoid tissues, analyses of the T-cell receptor (TCR) repertoires expressed by Foxp3⁺ and Foxp3⁻ CD4⁺ T cells indicate a degree of repertoire overlap (40% or more) suggesting that Tregs and conventional T cells can recognize the same antigens 3, 4; however, it is unclear whether this reflects peripheral conversion or intra-thymic events that result in distinct T-cell subsets expressing the same TCR. Recently, the transcription factor Helios has been linked with naturally occurring, but not induced, Tregs 5. This interesting finding needs to be confirmed in further studies, for it suggests that around 30% of peripheral Tregs arise extra-thymically.

Harnessing conversion of conventional T cells into Tregs makes therapeutic sense. The immunotherapeutic potential of diverting potential aggressors into regulators encompasses all unwanted immune responses including autoimmune and transplantation-induced responses. Inducing Foxp3 expression and suppressive ability in naive T cells in vitro is relatively straightforward, and there is a consensus that it can be achieved by optimal TCR stimulation in the presence of exogenous TGF-β or by inhibiting the P13K/AKT/mTOR signalling network. There is less agreement, however, about the requirement or lack of requirement for co-receptor signalling, e.g. through CD28, and whether there is indeed a strict dependence on TGF-β. Furthermore, while conversion of conventional naïve T cells into Tregs, as measured by Foxp3 expression and suppression assays, can occur efficiently in vitro, the level of Tregs induced in vivo is generally much lower, questioning the relevance of these in vitro protocols and how they relate to Treg induction in vivo.

In this issue of the European Journal of Immunology, two separate studies examine the requirements for Treg induction from naive T cells in vitro and in vivo. Oliveira et al. 6 focus on the disparity between in vivo and in vitro findings, and expand on a previous study 7 indicating that conversion of naive T cells into Tregs can be achieved in vivo through sub-immunogenic exposure to antigen. Oliveira et al. 6 utilize three sub-immunogenic regimes, including administration of antigen either orally or with non-depleting CD4-specific antibodies, or by using DCs pulsed with very low amounts of peptide. Their findings confirm that sub-immunogenic regimes can induce Tregs in vivo that are capable of conferring antigen-specific tolerance to skin grafts. When sub-optimal conditions were used in vitro (anti-CD3 coated plates), the conversion of naïve T cells into Tregs was shown to be TGF-β-dependent. Furthermore, Oliveira et al. 6 demonstrate that sub-optimal stimulation induces autologous TGF-β production by the T cells themselves meaning that addition of exogenous TGF-β is not required for the conversion process; TGF-β production can therefore be considered an event that is upstream of Foxp3 expression and on which Foxp3 expression is dependent. Altogether, the findings of Oliveira et al. 6 suggest that in vivo sub-immunogenicity, rather than a requirement for exogenous TGF-β, defines the conditions facilitating TGF-β-dependent Foxp3 induction in a T-cell intrinsic fashion.

In the second, separate study in this issue of the European Journal of Immunology, Gabryšová et al. 8 analyze how signal strength directs the differentiation of Foxp3⁺ T cells in vitro. The authors showed that the differentiation of Foxp3⁺ T cells in vitro is not dependent on the precise source of the signal (i.e. the co-stimulatory molecules involved), but rather the integration of several different signals received by a particular T cell. There has been controversy regarding the role of CD28 signalling in Foxp3⁺ T-cell induction and Gabryšová et al. 8 demonstrate the importance of CD28 signalling in the context of the strength of other signals (low TCR signal, high co-stimulatory signals) concurrently delivered to induce Treg conversion. Furthermore, the authors demonstrate that sub-optimal TCR stimulation can also be achieved by inhibiting the ERK/MAPK and the AKT/mTOR signalling networks, in a manner that is partially dependent on TGF-β-production by T cells, leading to Foxp3 induction. The study by Gabryšová et al. 8 also suggests a further role for TGF-β in maintaining Foxp3 expression in peripherally induced Tregs.

Are the in vitro and in vivo methods described in the studies by Oliveira et al. 6, Gabryšová et al. 8 and elsewhere relevant to physiological conditions and, if so, do these conditions result in the development of stable Foxp3⁺ populations with sustained suppressive functions? When are induced Tregs needed? There is good evidence that IFN-γ -producing pathogen-specific T cells can upregulate IL-10 production, thereby helping to moderate the pathogen-specific response, presumably to limit immunopathology 9. There may be little need, therefore, for induced Tregs. To date, TCR repertoire analyses have revealed no evidence for shared TCRs amongst autoreactive T cells and Tregs 10; however, the number of specificities examined so far is limited, and the situation whereby peripheral conversion contributes to peripheral tolerance remains a strong possibility. The studies by Oliveira et al. 6 and Gabryšová et al. 8 support the scenario whereby T cells that respond weakly to self-antigens, and hence escape negative selection in the thymus, can respond upon antigen encounter in the periphery by converting into Tregs, thereafter playing a role in mediating dominant tolerance to self. This process would not depend on TGF-β expression in any cell other than the responding T cell itself. It is known that naive T cells continually interact with self-peptide/MHC complexes 11. This raises the question as to exactly when a weak or sub-optimal interaction would drive these naïve cells to become Tregs?

Regardless of the physiological role of induced Tregs in peripheral tolerance, the ability to induce Tregs for therapeutic purposes is extremely attractive. The study of Gabrysova et al. 8 demonstrates the potential complexities involved in getting the T-cell signal strength right to achieve this aim. With this in mind, pharmacological interventions designed to inhibit mTOR or other relevant signalling pathways may prove most effective in encouraging Treg induction 7. These “experiments” are already being performed in humans who receive rapamycin (sirolimus) as an immunosuppressant in a variety of conditions, e.g. after solid organ transplantation. Lineage tracing of the relevant T cells by TCR analyses in these patients may prove fruitful in helping to understand the role of induced Tregs and the in vivo context in which they can arise.