Immunological tolerance has fascinated many of us for decades. The quasi-random rearrangement of gene segments encoding the Ag-specific receptors of lymphocytes occasionally generates specificities capable of recognizing the body’s own constituents. Because these autoreactive specificities are potentially dangerous to the organism’s health, multiple mechanisms have evolved to eliminate or inhibit the cells displaying them (1). Certain of the tolerance-inducing mechanisms operate within the primary lymphoid organs, the thymus for T cells and the fetal liver or bone marrow for B cells (termed central tolerance), whereas others exert their impact in the periphery, either within peripheral lymphoid organs or in tissues undergoing an immune response (peripheral tolerance).
A long-standing conundrum concerning central tolerance is the extent to which the universe of Ags specifically expressed in peripheral tissues is represented in the thymus and, relatedly, the degree to which it needs to be to achieve organismal tolerance. Many investigators anticipated that central tolerance would deal primarily with Ags restricted to local stromal cells, ubiquitously expressed Ags, and Ags circulating through the thymus, either blood-borne or within mobile hematopoietic cells. However, reports of peripheral tissue Ags (PTAs) being promiscuously expressed by thymic stromal cells began to emerge in the 1990s. For example, some lines of mice expressing an SV40 T Ag transgene transcribed under the dictates of the rat insulin (Ins) gene promoter exhibited transgene transcripts in the thymus, as well as in the pancreas, and these mice were unable to mount a T or B cell response against SV40 T Ag (2). This observation led to the discovery of Ins transcripts and transcripts encoding other pancreas-specific Ags in mouse thymic stromal cells (2). Later, it was found that the INS gene is also transcribed in the human thymus and that variation in its level of expression in that tissue (but not the pancreas) likely explains the ability of certain INS alleles to protect individuals from development of type 1 diabetes (3, 4). Although more and more examples of thymic expression of PTAs accrued over the years (5–7), it was not known how general this phenomenon really was. Nor was the functional relevance of thymic PTA gene expression clear, given reports of leaky, so-called “illegitimate,” transcription of many genes in presumably irrelevant cell types (8, 9).
The Pillars of Immunology article highlighted this month, emanating from Bruno Kyewski’s group (10), reported a series of findings that brought central tolerance to PTAs into the limelight. First, they purified several mouse thymus constituents, including cortical epithelial cells, medullary epithelial cells (mTECs), dendritic cells, macrophages, and thymocytes, and demonstrated that PTA transcripts are expressed primarily within mTECs, although some are also made in cortical epithelial cells. Second, they established the generality of promiscuous thymic expression of PTAs, documenting mTEC transcripts encoding 26 PTAs characteristic of a large variety of peripheral tissues. For one of the acute-phase proteins, serum amyloid P, they performed an elegant set of thymic transplant experiments to show that tolerance to serum amyloid P depends on its expression by thymic stromal cells rather than liver cells in the periphery, their more traditional site of expression.
Beyond these important contributions, this Pillars of Immunology article made some intriguing observations that continue to puzzle us today. One finding was that any particular PTA is expressed in only a minor fraction of mTECs, ranging from 1 in 50–200 to 1 in 20. This result, based on immunostaining of sorted thymic cells, was recently confirmed and extended on the basis of single-cell RNA sequencing analysis of purified mTECs (11, 12). The mechanisms underlying this restrained expression remain unknown, although, teleologically, it would make sense to restrict the number of PTA peptides generated to a number that can be presented effectively by any given mTEC at any given point in its life. Another intriguing finding from the 2001 paper of Derbinski et al. (10) is that spermatogonia express a repertoire of PTA transcripts as rich as that of mTECs. The mechanisms involved in, and physiological relevance of, this promiscuous gene expression have not been extensively studied. It is possible that this type of PTA expression is involved in another, peripheral, mechanism of tolerance induction. Or might it be more involved in genomic or chromatinic processes that unfold during embryonic (or later) development?
Like many groundbreaking studies, this Pillars of Immunology article promptly seeded a new focus of investigation. Two of the major questions that emerged were: what molecular mechanisms subtend PTA expression in mTECs; and what cellular mechanisms underlie the influence of mTECs on T cell tolerance? It turns out that many, although certainly not all, of the PTA transcripts upregulated in mTECs are induced by the transcriptional regulator Aire, explaining why humans and mice with an AIRE/Aire gene mutation develop multiorgan autoimmunity (13). The precise means by which Aire induces transcription of thousands of genes, otherwise specifically transcribed in different peripheral cell types, is still being elucidated. But it is already clear that Aire does not operate like a classical transcription factor that binds to a promoter sequence motif and induces transcriptional initiation; rather, it cooperates with a large set of partners to release stalling of RNA polymerase II just downstream of transcriptional initiation sites (14–16). As concerns the tolerance mechanisms impacted by PTA expression in mTECs, several reports, relying on both TCR-transgenic and nontransgenic systems, documented the ability of PTAs to drive negative selection of self-reactive effector T cells (17–19). Their capacity to also promote positive selection of Foxp3+CD4+ regulatory T cells was demonstrated more recently (20–23). Thus, mTEC expression of PTAs shapes the generation and function of the T cell repertoire by influencing two key arms of tolerance induction. In both cases, the impact can be exerted directly by mTEC presentation of internally generated Ags or indirectly through cross-presentation of mTEC Ags by dendritic cells in the vicinity (22).
By recognizing the potential importance of ectopic thymic expression of PTA transcripts, when many investigators had passed them off as just a curiosity or simply a reflection of molecular noise, Kyewski and collaborators (10) were able to bring to light an important mechanism of T cell tolerance induction. This mechanism has turned out to have fascinating molecular and cellular biology underpinning it and continues to engage a corps of enthusiastic investigators.
Abbreviations used in this article:
medullary epithelial cell
peripheral tissue Ag.
The authors have no financial conflicts of interest.