A question of stability and commitment

Regulatory T (T_Reg) cells are commonly defined by their expression of the transcription factor forkhead box P3 (FOXP3), but some studies have questioned the stability of T_Reg cells and the lineage specificity of FOXP3. Human activated T cells can upregulate FOXP3 expression without acquiring a regulatory phenotype, and FOXP3⁺ cells can lose FOXP3 expression and acquire an effector cell phenotype during inflammation or lymphopenia. This study reconciles these controversies by proposing the existence of a minor population of FOXP3⁺ T cells that are not committed to the T_Reg cell lineage.

The authors generated mice in which the expression of red fluorescent protein (RFP) identifies FOXP3⁺ cells and their progeny irrespective of continued FOXP3 expression (Foxp3^GFPCreROSA26^RFP mice). In adult mice, 10–20% of RFP⁺CD4⁺ peripheral T cells (but not of RFP⁺CD4⁺ immature thymocytes) were FOXP3⁻, showing that these cells had downregulated FOXP3 expression in the periphery. The RFP⁺FOXP3⁻ (exFOXP3) T cells had an effector or memory phenotype, could produce diverse effector cytokines and expressed low levels of T_Reg cell markers.

These results are consistent with a model in which the instability of FOXP3⁺ cells does not represent 're-programming' of T_Reg cells

The authors also showed that when FOXP3⁻ naive T cells were activated in vitro in the absence of transforming growth factor-β, ∼10% of the cells upregulated FOXP3 expression. However, these activation-induced FOXP3⁺ cells did not upregulate T_Reg cell signature genes, could not suppress T cell proliferation and did not maintain FOXP3 expression after re-stimulation. Together, the results confirm that FOXP3 expression in mouse T cells can be unstable.

Adoptive transfer experiments showed that peripherally induced FOXP3⁺ T cells can be separated into unstable and stable populations. FOXP3⁻CD4⁺ T cells from mice expressing LY5.1 were first transferred into mice expressing LY5.2; after a period of 2–8 weeks, FOXP3⁺CD4⁺ T cells (a population consisting of endogenous LY5.2⁺ cells as well as LY5.1⁺ cells that had upregulated FOXP3 expression) were transferred from these mice into lymphopenic recipients. When analysed 4 weeks later, some of the LY5.1⁺ cells were FOXP3⁻ and had proliferated extensively. The percentage of FOXP3⁻LY5.1⁺ cells and the degree of population expansion inversely correlated with the length of time that the LY5.1⁺ cells were present in LY5.2⁺ mice. In other words, the stability of FOXP3 expression is related to the length of time for which it is expressed. The adoptive transfer of FOXP3⁺ thymocytes showed that thymus-induced peripheral FOXP3⁺ T cells have stable FOXP3 expression.

These findings indicate that FOXP3 stability depends on whether FOXP3 is induced in the thymus or the periphery. In keeping with this, a percentage of peripherally induced FOXP3⁺ T cells, but not of thymus-derived FOXP3⁺ T cells, lost FOXP3 expression and secreted effector cytokines when stimulated under appropriate polarizing conditions. Those peripherally induced FOXP3⁺ cells that lost FOXP3 expression were typically CD25^−/low cells, whereas CD25^hi peripherally induced FOXP3⁺ cells had stable FOXP3 expression. Together, the results show that FOXP3 expression is only unstable in a transient population of CD25^−/low peripherally induced FOXP3⁺ cells. Furthermore, this population was shown to have defective suppressive activity.

Finally, the authors showed that ∼30% of exFOXP3 T cells can reacquire FOXP3 expression and suppressive activity following activation and that this ability correlates with a demethylated Foxp3 locus. By contrast, exFOXP3 cells with a methylated Foxp3 locus could not re-express FOXP3 and had an effector cell phenotype after activation.

These results are consistent with a model in which the instability of FOXP3⁺ cells does not represent 're-programming' of T_Reg cells, but rather a transient minor population of uncommitted FOXP3⁺ cells that do not have suppressive activity and can differentiate into FOXP3⁻ effector cells. Although some T_Reg cells can lose FOXP3 expression, they retain an epigenetic memory of suppressive activity and cannot convert to effector cells; thus the authors refer to them as 'latent' T_Reg cells.