Immunological tolerance is achieved through recessive and dominant mechanisms. In recessive tolerance the fate of self-reactive T cells is controlled in a cell intrinsic manner such that they undergo cell death or become anergic after exposure to self-antigen. In contrast, dominant tolerance is cell extrinsic and is mediated by CD4+ Tregs that express the transcription factor Forkhead P3 (Foxp3) and show increased self-reactivity when compared with conventional T cells (1). Peripheral homeostasis of Tregs is maintained through T cell receptor stimulation and signaling by IL-2 and costimulatory receptors such as CD28 (2, 3). Identifying the full spectrum of signals controlling Treg homeostasis in vivo may lend itself to therapeutic strategies that can be used to manipulate their number or function in order to ameliorate inflammatory diseases or augment an antitumor immune response. In this regard, the recent study by Schreiber and colleagues in the JCI (4) is particularly interesting, since the authors demonstrate that injection of mice with a mAb that binds to TNF receptor superfamily member 25 (TNFRSF25) selectively expands Tregs. The authors also show that administration of the anti-TNFRSF25 mAb inhibits a Th2 inflammatory response in the lungs of antigen-sensitized mice. Another interesting observation reported in that study was the preferential binding of the anti-TNFRSF25 mAb to Tregs (4). Based on this, the authors concluded that the level of TNFRSF25 on Tregs is considerably higher than that found on conventional CD4+ T cells. We have reexamined the expression of TNFRSF25 on conventional CD4+ T cells and Tregs, and our findings are remarkably different to those reported by Schreiber et al. We used two approaches to investigate expression of TNFRSF25 on T cell subsets (Figure (Figure1).1). First, we showed that a different anti-TNFRSF25 antibody stains conventional T cells and Tregs obtained from Foxp3-GFP knockin mice (5) with similar intensity (Figure (Figure1,1, A–C). Second, we demonstrated that soluble recombinant TL1A (sTL1A), the ligand for TNFRSF25, binds equally well to conventional CD4+ T cells and Tregs (Figure (Figure1D).1D). We therefore conclude that Tregs and conventional CD4+ T cells express similar levels of functional TNFRSF25.
We agree with the findings of Schreiber et al. that TNFRSF25 triggering can expand Tregs. In fact, we recently showed that transgenic mice that constitutively express TL1A have increased numbers of Tregs (6). However, we also observed increased activation of conventional CD4+ T cells, elevated levels of IL-13 and IL-17, and small intestinal immune pathology that manifests as goblet cell and paneth cell hyperplasia (6). A similar phenotype was reported by Siegel and colleagues using mice that express higher levels of a TL1A transgene (7). Interestingly, we found that stimulation of T cells with recombinant TL1A attenuates Treg-mediated suppression in vitro and this effect required TNFRSF25 signaling in either conventional CD4+ T cells or Tregs (6). Thus, TNFRSF25 exerts costimulatory effects on conventional CD4+ T cells as well as on Tregs. Furthermore, its effects on promoting Treg turnover are counterbalanced by its ability to attenuate immune suppression and stimulate effector T cells. This conclusion is in line with findings obtained using mice that were rendered deficient in either TNFRSF25 or TL1A. In these mice, the net effect of defective TNFRSF25 signaling is a reduction in the severity of T cell–mediated inflammation (8, 9).
How then can these findings been reconciled? We suspect that the anti-TNFRSF25 mAb used by Schreiber et al. may bind to an epitope that is preferentially expressed on Tregs, since multiple mRNA isoforms of TNFRSF25 have been described (10), and this may have lead to preferential expansion of Tregs in their experiments. Indeed, this notion is supported by a previous study showing differences in TNFRSF25 mRNA isoform expression between Tregs and Th17 cells (8). Thus, when compared with conventional T cells, Tregs expressed more TNFRSF25 mRNA transcripts that encode the shorter variant of TNFRSF25, which lacks the fourth extracellular cysteine-rich repeat. It is noteworthy that the antibody used in our study bound to both full-length TNFRSF25 as well as the short variant lacking the fourth cysteine-rich repeat (Figure (Figure1E).1E). We believe that caution should be applied in interpreting the findings obtained with anti-TNFRSF25 antibodies, as these may differ in their capacity to stimulate Tregs and effector T cells.