Gone baby gone—but how?

In this issue of Blood, Tewalt et al show how the interaction between PD-1 on CD8 T cells, and its ligand PD-L1 on lymphatic endothelial cells (LECs), appears to play a critical role in peripheral tolerance characterized by deletion.¹ 

Although the thymus plays a significant role in T-cell tolerance via deletion of self-reactive CD4 and CD8 cells, the prevalence of T cell–mediated autoimmune diseases proves that this tolerance is not complete, and that other, peripheral tolerance mechanisms must exist. Indeed, several mechanisms of peripheral tolerance have been described, including anergy, in which T cells are rendered nonfunctional by encountering antigen in the absence of appropriate co-stimulation² ; exhaustion, in which T cells persist but are functionally restrained by the presence of inhibitory molecules on the cell surface³ ; and peripheral deletion.⁴ 

A great deal of prior data suggest that deletional tolerance involves antigen cross-presentation by resting dendritic cells that migrate from tissues to draining lymph nodes.⁵  However, newer studies show that the situation might be more complex, with several groups showing that peripheral deletion can be mediated by cells resident in draining lymph nodes.⁶  Interestingly, several cell populations within lymph nodes express the protein Aire (AutoImmune Regulator), which drives promiscuous transcription of self-proteins and which is important in central deletion in the thymus—further suggesting a more robust role for lymph node cells in tolerance than was previously thought.⁷ 

In these studies, Tewalt and colleagues examined CD8 T cells specific for tyrosinase, which are deleted when transferred to black mice (C57BL/6) that express tyrosinase as a self-antigen. Based on the group's previous data showing a role for LECs in this process,⁸  the authors sought to determine the mechanism of deletion, that is, the cell surface molecules involved, as well as the biochemical mechanism. They began by characterizing the expression of cell surface molecules on LECs that are known to be involved in T-cell exhaustion. Even at rest, LECs express the ligands for PD-1 (PD-L1), LAG-3 (class II MHC), BTLA (HVEM), and CD160 (HVEM). Surprisingly, blockade of the PD-1/PD-L1 interaction completely abrogated deletional tolerance, whereas blockade of other immune checkpoints did not affect deletion at all (see figure). This result was confirmed using PD-L1 knockout mice, which also failed to delete adoptively transferred self-reactive CD8 T cells. Bone marrow chimera studies confirmed that the cell type that mediates deletion was radio-resistant, and thus was clearly not a conventional CD11c⁺ dendritic cell.

In addition to dendritic cells, 2 other cell types in the lymph node strongly expressed PD-L1 (and thus could be the deletors in question); LECs and BECs (blood epithelial cells). Of these, LECs were shown to be the most likely culprits, based on their persistent PD-L1 expression as well as their co-expression of tyrosinase. The authors further explored the mechanisms involved. Hypothesizing that deletion involves a lack of co-stimulation, external co-stimulation was introduced into the model via agonist antibodies to 41BB and/or OX40. Convincingly, co-stimulation provided by 41BB ligation reversed deletional tolerance, while anti-OX40 did not. Biochemically, IL-2 seems to be critically involved; CD8 cells undergoing deletional tolerance expressed considerably lower levels of the IL-2 receptor α chain (CD25) than those activated by a specific vaccine. Functionally, blocking CD25 with a monoclonal antibody completely reversed the rescue of self-antigen–specific CD8 T cells observed in PD-L1 knockout mice and also reversed the rescue of self-antigen–specific T cells seen when co-stimulation was artificially induced with the 41BB agonist antibody. In summary, these studies show that deletion of self-reactive CD8 T cells can be mediated by the interaction between PD-L1 on radio-resistant LECs and PD-1 on CD8 T cells, involves IL-2, and that deletion can be reversed by either PD-L1 blockade, or by providing co-stimulation using an anti-41BB agonist antibody.

These studies have timely clinical implications; the importance of the PD-1/PD-L1 pathway in cancer was recently highlighted by 2 publications showing that blocking the PD-1/PD-L1 interaction can result in clinical responses in approximately 20% to 30% of patients with advanced melanoma or kidney cancer, as well as in some patients with lung cancer.^9,10  Here, PD-1 was postulated to reverse T-cell exhaustion,¹¹  as has been observed in animal models of chronic viral infection. But, these new data suggest an alternative (although not mutually exclusive) mechanism; perhaps PD-1 blockade in humans functions by reversing LEC-mediated deletion of tumor-reactive cells?

Like any important study, these results raise additional questions. Because PD-1 and PD-L1 knockout mice develop only limited, late-onset autoimmunity,¹¹  a deficit of LEC-driven PD-1–mediated deletional tolerance that must be compensated for; this is clearly not a universal peripheral tolerance mechanism. Exactly which antigens, which T cells, and under what condtions this pathway operates is an open and fascinating question. The role of additional lymph node stromal populations is also of interest, as other groups have shown that deletion can be mediated by follicular reticular cells (FRCs) as well.⁶  Because clinical data (in addition to copious preclincal data) show that PD-1–mediated tolerance is important in cancer patients; whether tolerance occurs in the tumor itself, or via the LECs in tumor-draining lymph nodes is a key question moving forward. Finally, these results may shed some light on a troubling (but rare) toxicity observed in the PD-1 trials: inflammation of the lung (pneumonitis). One wonders whether LEC-mediated PD-1–dependent deletional tolerance is perhaps particularly important for pulmonary antigens? In summary, peripheral tolerance continues to fascinate, with clinical relevance to the fields of autoimmunity, chronic infection, and cancer. The novel mechanism described here will likely drive fruitful research for years to come.