Direct presentation of inflammation-associated self-antigens by thymic innate-like T cells induces elimination of autoreactive CD8+ thymocytes

You, Yuanyuan; Dunst, Josefine; Ye, Kewei; Sandoz, Patrick A.; Reinhardt, Annika; Sandrock, Inga; Comet, Natalia R.; Sarkar, Rupak Dey; Yang, Emily; Duprez, Estelle; Agudo, Judith; Brown, Brian D.; Utz, Paul J.; Kastenmüller, Wolfgang; Gerlach, Carmen; Prinz, Immo; Önfelt, Björn; Kreslavsky, Taras

doi:10.1038/s41590-024-01899-6

Download PDF

Article
Open access
Published: 11 July 2024

Direct presentation of inflammation-associated self-antigens by thymic innate-like T cells induces elimination of autoreactive CD8⁺ thymocytes

Nature Immunology volume 25, pages 1367–1382 (2024)Cite this article

17k Accesses
14 Citations
52 Altmetric
Metrics details

Subjects

Abstract

Upregulation of diverse self-antigens that constitute components of the inflammatory response overlaps spatially and temporally with the emergence of pathogen-derived foreign antigens. Therefore, discrimination between these inflammation-associated self-antigens and pathogen-derived molecules represents a unique challenge for the adaptive immune system. Here, we demonstrate that CD8⁺ T cell tolerance to T cell-derived inflammation-associated self-antigens is efficiently induced in the thymus and supported by redundancy in cell types expressing these molecules. In addition to thymic epithelial cells, this included thymic eosinophils and innate-like T cells, a population that expressed molecules characteristic for all major activated T cell subsets. We show that direct T cell-to-T cell antigen presentation by minute numbers of innate-like T cells was sufficient to eliminate autoreactive CD8⁺ thymocytes. Tolerance to such effector molecules was of critical importance, as its breach caused by decreased thymic abundance of a single model inflammation-associated self-antigen resulted in autoimmune elimination of an entire class of effector T cells.

Antigen presentation for central tolerance induction

Article 18 September 2024

A guide to thymic selection of T cells

Article 18 July 2023

A multilayered immune system through the lens of unconventional T cells

Article 21 July 2021

Main

The generation of a diverse antigen receptor repertoire through V(D)J recombination inevitably produces self-reactive receptors in some developing lymphocytes. To prevent autoimmunity, these self-reactive cells must be inactivated either at the site of development (central tolerance) or in secondary lymphoid organs (peripheral tolerance). Central tolerance requires the presence of antigens at the site of lymphocyte differentiation. In the thymus, several mechanisms ensure the breadth of thymocyte exposure to peripheral antigens. A wide range of tissue-restricted antigens show ‘ectopic’ expression in medullary thymic epithelial cells (mTECs)^1,2,3, and some mTECs induce large parts of molecular programs of various peripheral cell types^4,5,6,7. The thymus also harbors all major subsets of dendritic cells (DCs) that can present both endogenously expressed and TEC-derived antigens^8,9. Migration of peripheral DCs into the thymus further expands the spectrum of tissue-specific antigens represented in the thymus^8,10. Finally, another professional antigen-presenting cell (APC) type, thymic B cells, also contributes to the induction of T cell tolerance^11,12,13.

Self-reactive thymocytes are inactivated by negative selection¹⁴, eviction of immature cells from the thymus¹⁵ and diversion into lineages with regulatory or innate-like properties¹⁶. Differentiation of innate-like T cells represents the execution of one of the specialized effector programs (for example, type 1 helper T (T_H1) cell-, type 2 helper T (T_H2) cell- or interleukin-17 (IL-17)-producing helper T (T_H17) cell-like^17,18) following antigen encounter in the thymus. These cells can express genes encoding inflammatory mediators and cytotoxic molecules at steady state^18,19,20 and often home to barrier tissues, where they contribute to the first line of defense against pathogens and regulate tissue homeostasis^21,22.

However, many innate-like T cells found in the thymus, including a large fraction of γδT cells and the majority of invariant natural killer T (iNKT) cells, are not newly developed immature cells but mature tissue-resident lymphocytes^23,24. Inflammatory mediators secreted by these cells have a broad impact on other components of the thymic microenvironment. For example, type 2 cytokines derived from innate-like cells support the intrathymic generation of innate-like interferon-γ (IFNγ)-producing CD8⁺ T cells^25,26, regulate the recruitment of thymic eosinophils²⁷, affect TEC function²⁸ and induce the activation of thymic DCs²⁹. Intriguingly, other inflammatory pathways, for example, type 1/type 3 IFN signaling and MyD88-dependent signaling in TECs^30,31, are also constitutively activated in the thymus. Therefore, the thymus has evolved to harbor a complex constitutively active inflammatory network, and thymus-resident innate-like T cells represent one of its central nodes. The reasons for the activation of numerous inflammatory pathways in the organ where pathogen encounter is unlikely remain unclear.

If autoreactive T cells escape central tolerance and encounter antigen in secondary lymphoid organs in the absence of inflammation, they are inactivated by peripheral tolerance mechanisms³². Mature T lymphocytes will only undergo clonal expansion and differentiation into effector cells if they have received co-stimulatory signals that are provided by APCs only in the context of inflammation. The induction of T cell effector programs whose components constitute self-antigens therefore coincides both spatially and temporally with the presence of a pathogen. Breach of tolerance to such antigens would result in a ‘fratricide’ reaction that would curtail any further immune response involving expression of this molecule. It remains unclear how the immune system discriminates such inflammation-associated self-antigens from antigens originating from the pathogen itself.

In this study, we aimed to investigate the mechanisms of tolerance to T cell-derived inflammation-associated self-antigens. Using model inflammation-associated self-antigens, we demonstrate that CD8⁺ T cell tolerance to these molecules can be induced in the thymus. The high efficiency of this tolerance was ensured by multiple redundant cellular sources of these antigens in the thymus, which, in addition to TECs, included thymic innate-like T cells and thymic eosinophils. Expression of a self-antigen by minute numbers of thymic innate-like T cells was sufficient to mediate remarkably efficient elimination of autoreactive CD8⁺ thymocytes that occurred through direct T cell-to-T cell antigen presentation. These results suggest that constitutive expression of inflammation-associated molecules in the thymus may have evolved to ensure the induction of tolerance to this unique class of self-antigens.

Results

Innate-like T cells as libraries of T cell effector antigens

To start unraveling mechanisms of tolerance induction to T cell-derived inflammation-associated self-antigens, we analyzed the expression of a selection of genes encoding T cell effector molecules (Il4, Il13, Ifng, Il17a, Il17f and Gzmb) and their upstream regulators (Eomes and Tbx21) in thymic cell subsets. Analysis of bulk transcriptomics data from thymic cell types available in the Immunological Genome (Immgen) database^33,34 (Fig. 1a) and in the integrated single-cell RNA-sequencing (scRNA-seq) mouse thymus atlas^6,35,36 (Fig. 1b) demonstrated that professional thymic APCs, such as TECs and thymic DCs, express no or low levels of many of these molecules. Instead, the highest levels of expression of most of these molecules in the thymus overlapped with the expression of Zbtb16, the gene encoding the transcription factor PLZF that orchestrates effector programs of many innate-like T cell populations^37,38,39,40 (Fig. 1b). Analysis of the Immgen dataset confirmed predominant expression of effector T cell molecules in iNKT and γδT cell subsets (Fig. 1a). This pattern of expression was also conserved in humans as determined by analysis of data from the human scRNA-seq thymus atlas³⁵ (Extended Data Fig. 1). Thus, in contrast to professional APCs, thymic innate-like T cells express high levels of effector molecules characteristic of all major effector T cell subsets and would represent perfect libraries of T cell effector program antigens.

**Fig. 1: Thymic innate-like T cells express a broad spectrum of inflammation-associated self-antigens.**

CD8⁺ T cell tolerance to a model inflammation self-antigen

To start characterizing tolerance to inflammation-associated self-antigens, we first took advantage of the predominant expression of Il4 by innate-like T cells in the mouse thymus (Fig. 1a,b) and exploited green fluorescent protein (GFP) in Il4-IRES-GFP mice (referred to hereafter as IL-4–GFP mice)⁴¹ as a generic model of antigen expressed by thymic innate-like T cells. Immunization with a major histocompatibility complex (MHC) class I (H2-K^d) GFP_200–208 epitope⁴² resulted in a prominent CD8⁺ T cell response in wild-type (WT) mice, but no response was detected in IL-4–GFP animals (Fig. 2a), indicating a remarkably efficient CD8⁺ T cell tolerance to GFP in this system. By contrast, immunization with an MHC class II (I-A^b) GFP_81–95 epitope⁴³ resulted in comparable CD4⁺ T cell responses in WT and IL-4–GFP mice (Fig. 2b). Thus, expression of IL-4–GFP resulted in strong CD8⁺, but not CD4⁺, T cell tolerance.

**Fig. 2: Redundant role of radioresistant and radiosensitive cells in the induction of CD8⁺ T cell tolerance to a model inflammation-associated self-antigen.**

Next, to assess the contribution of hematopoietic and stromal cells to the induction of CD8⁺ T cell tolerance, we established WT → WT, IL-4–GFP → WT, WT → IL-4–GFP and IL-4–GFP → IL-4–GFP bone marrow (BM) chimeras and immunized them with the GFP_200–208 peptide. A clear expansion of GFP-specific CD8⁺ T cells was observed only in the WT → WT group, whereas the frequency of tetramer⁺ cells in all other groups was similar to that in unimmunized controls (Fig. 2c). Together, these results reveal efficient induction of CD8⁺ T cell tolerance to a model inflammation-associated self-antigen, with a redundant role of hematopoietic and radioresistant cells in this process.

Central tolerance to a model inflammation self-antigen

We next sought to assess if tolerance to IL-4–GFP was already established in the thymus. Analysis of IL-4–GFP expression revealed that only 0.5 ± 0.2% (mean ± s.d.) of hematopoietic cells in the thymus were GFP⁺, of which 76.9 ± 3.3% were innate-like T cells and 17.7 ± 3.4% represented thymic eosinophils (Fig. 3a and Extended Data Fig. 2a,b). GFP expression was also detectable in 0.8 ± 0.2% of mTECs (Extended Data Fig. 2c). Immunofluorescence microscopy revealed that GFP⁺SiglecF⁺ eosinophils and GFP⁺SiglecF^– innate-like lymphocytes were present both in the thymic cortex and medulla, with more GFP⁺ lymphocytes located in the medulla (Fig. 3b and Extended Data Fig. 2d).

**Fig. 3: Efficient central CD8⁺ T cell tolerance to a model inflammation self-antigen.**

We next tested if the expression of IL-4–GFP by hematopoietic and/or stromal cells would result in the inactivation of autoreactive CD8⁺ T cells already in the thymus. We took advantage of Jedi T cell antigen receptor-αβ (Jedi-TCRαβ) mice, in which T cells express prerearranged chains of the H2-K^d GFP_200–208-specific TCR from the endogenous Tcra and Tcrb loci⁴⁴. We generated BM chimeras by transferring a mix of Jedi-TCRαβ and WT or Jedi-TCRαβ and IL-4–GFP BM cells into WT or IL-4–GFP recipients and analyzed T cell development in these four groups of BM chimeras. As expected, in the absence of IL-4–GFP, the Jedi thymocyte compartment in WT + Jedi-TCRαβ → WT chimeras included CD4/CD8-double-negative (DN), CD4/CD8-double-positive (DP) and CD8-single-positive (CD8SP) cells (Fig. 3c). Strikingly, expression of IL-4–GFP solely by thymic hematopoietic cells in the IL-4–GFP + Jedi-TCRαβ → WT group or by radioresistant cells in the WT + Jedi-TCRαβ → IL-4–GFP group in both cases resulted in a virtually complete elimination of GFP-specific CD8SP thymocytes (Fig. 3c). Residual DP thymocytes in chimeras with IL-4–GFP donors and/or recipients upregulated PD-1 expression and contained a large fraction of dying cells (Fig. 3c) and cells positive for active caspase-3 (Fig. 3d), indicating that these autoreactive cells were eliminated by negative selection. Strikingly, in IL-4–GFP + Jedi-TCRαβ → WT chimeras, this elimination of autoreactive thymocytes was mediated by as few as 0.1 ± 0.02% of GFP⁺ hematopoietic cells present in the thymi of these mice (Extended Data Fig. 2e). By contrast, WT + Jedi-TCRαβ → IL-4–GFP chimeras contained, on average, 0.01% of residual GFP⁺ hematopoietic cells, and some had virtually no detectable GFP⁺ thymocytes, indicating that, in this setting, induction of tolerance is likely to be mediated solely by thymic stroma (Extended Data Fig. 2e). The disappearance of Jedi CD8⁺ thymocytes was reflected in the periphery by an overall decrease in Jedi-TCRαβ cells and a near complete loss of GFP-specific conventional CD8⁺ T cells in the spleen (Fig. 3e). As the TCR in Jedi-TCRαβ mice was expressed prematurely at the DN stages of T cell development (Fig. 3c), we performed a similar analysis with Jedi-TCRβ mice that only carry the TCRβ chain of the Jedi TCR and thus rely on recombination of the endogenous Tcra loci at the DP stage of T cell development. A small distinct population of Jedi-TCRβ GFP-specific CD8⁺ T cells was completely eliminated in the presence of antigen (see Supplementary Note 1 and Supplementary Figs. 1 and 2). We conclude that most CD8⁺ T cells that recognize IL-4–GFP are inactivated already in the thymus, and TECs and hematopoietic cells play redundant roles in the induction of central tolerance to this inflammation-associated self-antigen.

iNKT cells and eosinophils can mediate negative selection

As the thymic GFP⁺ population in IL-4–GFP mice, although dominated by innate-like T lymphocytes, represent a complex mix of cells (Fig. 3a), we next aimed to test if antigen expression solely by innate-like T cells was sufficient to induce tolerance. To this end, we established reaggregate thymus organ cultures (RTOCs; Fig. 4a) with small numbers of Jedi DP thymocytes and iNKT cells sorted from WT or IL-4–GFP thymi. The presence of as few as 0.2% of GFP⁺ iNKT cells in these thymus organoids resulted in a near complete deletion of Jedi cells with few remaining cells acquiring the CD4^–CD8α^loCD8β^loPD-1^hi cell-surface phenotype (Fig. 4b). Thus, antigen expression by minute numbers of innate-like T cells is sufficient to eliminate autoreactive thymocytes in the RTOC system.

**Fig. 4: Antigen expression exclusively by small numbers of iNKT cells is sufficient to induce elimination of autoreactive CD8⁺ thymocytes.**

Of note, this highly efficient induction of negative selection was not a unique property of iNKT cells, as small numbers of conventional ACTB–GFP-transgenic CD4SP thymocytes and thymic eosinophils from IL-4–GFP mice likewise induced elimination of Jedi-TCRαβ DP thymocytes in RTOCs (Extended Data Fig. 3a–c). These results suggest that in addition to professional thymic APCs, such as TECs, negative selection of autoreactive CD8⁺ thymocytes can be induced by expression of antigen in a wide range of ‘amateur’ APC cell types, including developing conventional T lymphocytes, thymic innate-like T cells and thymic eosinophils.

Effects of co-stimulation, antigen localization and affinity

As co-stimulation by CD80/CD86 can be important for effective negative selection for some, but not all, TCR specificities^{45,46,47,48,49}, we next assessed the expression of these molecules by thymic eosinophils and innate-like T cells (Extended Data Fig. 4a). Thymic eosinophils expressed high levels of CD80 but not CD86, whereas a large fraction of thymic iNKT cells and some γδT cells displayed both co-stimulatory molecules on their surface. As data from the Immgen database^20,34 suggest that iNKT cells express little if any Cd80 or Cd86 mRNA, and as T cells often acquire CD80 and CD86 from APCs through trogocytosis^50,51, it seems likely that iNKT cells acquire these molecules from professional thymic APCs. Addition of anti-CD80/CD86 to RTOCs containing Jedi-TCRαβ DP thymocytes and WT or IL-4–GFP iNKT cells had no effect on negative selection (Extended Data Fig. 4b), suggesting that co-stimulation might be dispensable for negative selection in the case of Jedi thymocytes. However, as the co-stimulation requirement for negative selection is not absolute and depends on the epitope⁴⁵, and as thymic innate-like T cells are equipped with CD80 and CD86, it is conceivable that co-stimulation could contribute to the induction of tolerance to some self-antigens expressed by these cells.

As many endogenous inflammation-associated self-antigens are secreted molecules, we next tested if innate-like T cells can induce tolerance to a secreted antigen. To this end, we generated retroviral constructs encoding intracellular (i.c.GFP) and secreted (secGFP) versions of GFP and confirmed efficient GFP secretion in the latter setting (Extended Data Fig. 4c). We next transduced thymic iNKT cells with Thy1.1-IRES-i.c.GFP, Thy1.1-IRES-secGFP and Thy1.1-only retroviruses, sorted Thy1.1⁺ iNKT cells and added them to RTOCs together with Jedi-TCRαβ DP thymocytes. A comparable loss of Jedi cells was observed with i.c.GFP and secGFP constructs, and the remaining cells upregulated PD-1 expression and downregulated CD8 expression (Fig. 4c).

Finally, to test how antigen affinity affects negative selection, we took advantage of altered peptide ligands described for the OT-I TCR. It was shown that Q4, T4 and V4 substitutions in the SIINFEKL epitope (N4) reduced OT-I CD8⁺ T cell reactivity by 20-, 70- and 700-fold, respectively⁵². We therefore retrovirally transduced thymic iNKT cells with constructs encoding cytoplasmic ovalbumin (cOVA)-IRES-GFP containing these epitopes and added GFP⁺ iNKT cells to RTOCs together with sorted OT-I DP thymocytes. N4-, Q4- and T4-expressing iNKT cells induced a near complete elimination of OT-I thymocytes (33-, 18- and 9-fold decreases, respectively), and even very low-affinity V4 epitope was sufficient to induce elimination of about half of the OT-I thymocytes (Fig. 4d).

We conclude that expression of either intracellular or secreted antigens by innate-like T cells can be sufficient to induce elimination of autoreactive CD8⁺ thymocytes and that this elimination can take place across a wide range of affinities.

Antigen expression by iNKT cells induces tolerance in vivo

We next sought to test if antigen expression by innate-like T cells is also sufficient to mediate tolerance induction in vivo. As a first approach to this question, we used IL-17A–GFP mice in which, in contrast to the IL-4–GFP system, GFP expression in the thymic hematopoietic compartment was restricted to innate and innate-like lymphocytes (Fig. 5a and Extended Data Fig. 5a,b). Analysis of stromal populations in these mice also revealed GFP expression in a small subset of mTECs (Extended Data Fig. 5c). Therefore, to dissect the roles of GFP-expressing stromal and hematopoietic populations, we established WT → WT, IL-17A–GFP → WT, WT → IL-17A–GFP and IL-17A–GFP → IL-17A–GFP BM chimeras and immunized these mice with the GFP_200–208 peptide. Fully mirroring the results obtained in the IL-4–GFP system (Fig. 2), IL-17A–GFP expression either by radioresistant or radiosensitive cells was sufficient to abrogate CD8⁺ T cell responses to GFP_200–208 (Fig. 5b), whereas no CD4⁺ T cell tolerance to GFP_81–95 MHC class II epitope immunization was observed in IL-17A–GFP mice (Extended Data Fig. 5d). Unexpectedly, GFP-expressing cells were very poorly reconstituted in the thymi of IL-17A–GFP → WT BM chimeras (0.009 ± 0.003% GFP⁺ cells of total thymocytes in IL-17A–GFP → WT BM chimeras versus 0.06 ± 0.04% in steady-state IL-17A–GFP mice and 0.5 ± 0.2% in IL-4–GFP mice; Extended Data Fig. 5e). This near absence of GFP-expressing cells in the thymi of IL-17A–GFP BM chimeras precluded us from assessing negative selection of supraphysiological numbers of autoreactive GFP-specific thymocytes in Jedi:IL-17A–GFP mixed BM chimeras, a setting that in fact resulted in breach of tolerance (see below). As an alternative approach, we therefore turned to RTOC experiments, which demonstrated that iNKT cells from IL-17A–GFP mice can induce efficient elimination of Jedi-TCRαβ DP thymocytes (Fig. 5c).

**Fig. 5: Expression of antigen by innate-like T cells is sufficient to mediate induction of tolerance in vivo.**

The immunization experiments in IL-17A–GFP → WT BM chimeras described above showed that expression of a model inflammation-associated self-antigen solely by innate-like lymphocytes is sufficient to induce tolerance in vivo, and RTOC experiments suggested that it can happen through negative selection of autoreactive thymocytes. To directly test if exclusive antigen expression by thymic innate-like T cells is sufficient to induce inactivation of autoreactive CD8⁺ thymocytes in vivo, we injected sorted thymic IL-4–GFP iNKT cells into the thymi of Jedi-TCRβ mice, which, before injection, contained 0.007 ± 0.003% GFP-specific thymocytes (Fig. 5d). GFP⁺ cells were detectable in the thymi of the recipient mice 7 days after transfer (Fig. 5d), but their frequency was much lower than that in IL-4–GFP mice (Fig. 3a) and similar to what was observed in steady-state IL-17A–GFP mice (Fig. 5a). The presence of these minute numbers of GFP⁺ iNKT cells resulted in a decreased frequency of GFP-specific thymocytes and an acquisition of a CD4^–CD8α^loCD8β^loPD-1^hi cell-surface phenotype by the remaining GFP-specific cells (Fig. 5d). Thus, antigen expression exclusively by small numbers of thymic innate-like T cells is sufficient to induce elimination of autoreactive CD8⁺ thymocytes in vivo.

T cell-to-T cell antigen presentation mediates negative selection

The efficient negative selection of autoreactive thymocytes by small numbers of antigen-expressing innate-like T cells suggested possible cross-presentation by thymic professional APCs. To start exploring this possibility, we sorted DCs, iNKT cells and eosinophils from the thymi of WT, IL-4–GFP and IL-17A–GFP mice and cocultured them with an NFAT reporter cell line⁵³ expressing the Jedi TCR. Reporter expression was upregulated in cocultures with iNKT cells from IL-4–GFP and IL-17A–GFP mice and eosinophils from IL-4–GFP mice but not in cocultures with DCs (Extended Data Fig. 6a, note the higher baseline reporter expression in all DC cocultures), suggesting that IL-4–GFP and IL-17A–GFP may not undergo efficient cross-presentation by thymic DCs.

To directly test if cross-presentation could contribute to negative selection, we studied the fate of Jedi-TCRαβ DP thymocytes in RTOCs established with iNKT cells sorted from the thymi of IL-4–GFP mice either on an H2^b/d background (capable of presentation to Jedi thymocytes) or on an H2^b/b background that makes direct presentation to Jedi thymocytes impossible. Negative selection and phenotypic changes of Jedi cells were completely abrogated when iNKT cells were unable to directly present the GFP peptide to Jedi thymocytes due to this MHC mismatch, even when such iNKT cells were present at highly supraphysiological numbers (Fig. 6a). No evidence for cross-presentation was obtained when MHC-mismatched iNKT cells from IL-17A–GFP mice or MHC-mismatched iNKT cells expressing a secreted version of GFP were used in RTOC experiments (Extended Data Fig. 6b,c).

**Fig. 6: Innate-like T cells can mediate elimination of autoreactive CD8⁺ thymocytes through direct T cell-to-T cell antigen presentation.**

To assess the contribution of direct versus cross-presentation of IL-4–GFP in vivo, we next established three groups of BM chimeras: (1) WT H2^b/d + Jedi-TCRβ H2^b/d → WT H2^b/d (no antigen), (2) IL-4–GFP H2^b/d + Jedi-TCRβ H2^b/d → WT H2^b/d (both direct and cross-presentation is possible) and (3) IL-4–GFP H2^b/b + Jedi-TCRβ H2^b/d → WT H2^b/d (only cross-presentation by APCs of Jedi and recipient origin is possible). Importantly, in the latter setting, more than half of the thymic DCs expressed H2-K^d (Extended Data Fig. 6d) and therefore were theoretically capable of cross-presentation of GFP_200–208 epitope to Jedi cells. Although GFP-specific thymocytes were eliminated when direct presentation was possible, ‘direct presentation-deficient’ IL-4–GFP H2^b/b + Jedi-TCRβ H2^b/d → WT chimeras had numbers and cell-surface phenotypes of GFP-specific thymocytes indistinguishable from those of the no-antigen group (Fig. 6b). Finally, intrathymic injections of H2^b/b (no direct presentation) or H2^b/d (direct presentation is possible) iNKT cells expressing a secreted version of GFP into Jedi-TCRβ H2^b/d recipients demonstrated that PD-1 upregulation and CD8⁺ downregulation was evident only when transferred iNKT cells were capable of direct antigen presentation (Extended Data Fig. 6e). We conclude that in the case of GFP as an innate-like T cell-derived model antigen, it undergoes direct T cell-to-T cell presentation but not cross-presentation by thymic APCs.

Finally, to visualize these T cell-to-T cell antigen presentation events, we analyzed the interactions between fluorescently labeled iNKT cells and Jedi-TCRαβ thymocytes loaded with Calbryte-520 (Cal-520) AM Ca²⁺ sensor dye in RTOCs by confocal microscopy. Jedi DP cells for these experiments were sorted for low levels of PD-1 expression to enrich TCR-expressing cells (Extended Data Fig. 6f). RTOCs were allowed to assemble overnight and were then imaged by confocal microscopy for 13 h (Fig. 6c–g). Both iNKT cells and viable Jedi-TCRαβ DP thymocytes were highly motile in these organoids (Supplementary Videos 1–4). In line with the expression of the Jedi TCR by about 70% of DP thymocytes at the beginning of the culture and their ongoing negative selection, approximately one-third of the contacts between Jedi thymocytes and IL-4–GFP iNKT cells resulted in Ca²⁺ flux in Jedi-TCRαβ DP cells (Fig. 6d–f, Extended Data Fig. 6g and Supplementary Video 1). No such contact-induced Ca²⁺ signaling was observed in RTOCs with WT iNKT cells (Fig. 6d–f and Supplementary Video 2). Contacts between IL-4–GFP iNKT cells and Jedi-TCRαβ DP cells that lead to Ca²⁺ signaling were longer than those that did not result in Ca²⁺ flux (Fig. 6g), and, in some cases, Jedi thymocytes exhibited chasing behavior toward IL-4–GFP iNKT cells (Supplementary Video 3). In few cases, contact-induced Ca²⁺ flux in Jedi-TCRαβ thymocytes was followed by a loss of motility and changes in cellular morphology suggestive of cell death (Supplementary Video 4). Thus, direct antigen presentation by iNKT cells to autoreactive thymocytes was confirmed by visualization of TCR signaling in autoreactive CD8⁺ thymocytes.

Lack of tolerance to IL-17A–GFP results in a fratricide reaction

We next sought to investigate the consequences of failure to disarm CD8⁺ T cells reactive to an antigen expressed by effector T cells. Our initial analysis of PLZF^lu/lu and Tcrd^–/–Cd1d^–/–Mr1^–/– mice, which lack some innate-like T cell populations, did not reveal autoimmune phenotypes, in line with the unexpected abundance of all functional effector subsets in the thymi of these mice (see Supplementary Note 2 and Supplementary Figs. 3 and 4). We therefore next decided to take advantage of the fact that the thymic IL-17A–GFP⁺ compartment was not properly reconstituted in IL-17A–GFP → WT BM chimeras, and GFP⁺ cells were nearly absent in the thymi of these mice, whereas the peripheral IL-17A–GFP effector subsets were less affected in this setting (Extended Data Figs. 5e and 7a). We hypothesized that if we tip the balance between scarce thymic IL-17A–GFP⁺ cells and developing GFP-specific CD8⁺ thymocytes toward the latter cells in this system, some autoreactive CD8⁺ T cells should escape to the periphery and may cause autoimmune elimination of peripheral T_H17 cells. To test this hypothesis, we established the following three groups of BM chimeras: (1) IL-17A–GFP → WT (no overproduction of autoreactive T cells), (2) WT + Jedi-TCRβ → WT (no antigen) and (3) IL-17A–GFP + Jedi-TCRβ → WT (predominantly peripheral antigen and overproduction of autoreactive T cells; Fig. 7a). As expected, small numbers of GFP-specific CD8⁺ T cells with naive cell-surface phenotype were found in peripheral lymphoid organs of WT + Jedi-TCRβ → WT chimeras (Fig. 7b and Extended Data Fig. 7b). By contrast, lymphoid organs of IL-17A–GFP + Jedi-TCRβ → WT chimeras contained strongly expanded GFP-specific CD8⁺ T cells with a CD44⁺PD-1⁺ activated cell-surface phenotype (Fig. 7b and Extended Data Fig. 7b). Thymi of IL-17A–GFP + Jedi-TCRβ → WT BM chimeras also exhibited an accumulation of GFP-specific CD44⁺PD-1⁺CD8⁺ T cells (Extended Data Fig. 7c), possibly reflecting recruitment of activated cells from the periphery. Of note, this was in stark contrast to the previous observations made in IL-4-GFP + Jedi-TCRβ → WT BM chimera thymi that exhibited a complete loss of GFP-specific cells (Fig. 6b and Supplementary Fig. 2). This accumulation of activated autoreactive CD8⁺ effector T cells coincided with a near complete loss of IL-17A–GFP⁺ T_H17 cells (Fig. 7c and Extended Data Fig. 7d,e; gating on V_β4^– cells was applied to exclude Jedi cells from the analysis). These results suggest that breach in tolerance induction results in an autoimmune attack that may eliminate peripheral IL-17A-expressing effector T cells. To formally test this, we performed an in vivo killing assay by transferring T_H17 cells in vitro differentiated from CD4⁺ T cells of IL-17A–GFP mice. Only a fraction of these cells induced the expression of GFP (which correlated well with the production of IL-17A; Extended Data Fig. 7f) and therefore would represent possible targets for activated Jedi cells. These GFP⁺ cells were selectively eliminated in IL-17A–GFP + Jedi-TCRβ → WT, but not in WT + Jedi-TCRβ → WT, chimeras (Fig. 7d). Together, these results demonstrate that failure to induce CD8⁺ T cell tolerance to a single inflammation-associated self-antigen can result in a fratricide reaction that leads to elimination of the entire class of effector T cells expressing this molecule.

**Fig. 7: Decreased thymic abundance of IL-17A–GFP results in autoimmune elimination of peripheral GFP⁺ T_H17 cells.**

Induction of tolerance to an endogenous effector molecule

Finally, we sought to test if expression of an endogenous effector T cell molecule exclusively in the hematopoietic compartment is sufficient to induce CD8⁺ T cell tolerance to this antigen. We focused on IFNγ, the production of which among thymocytes was restricted to T cell/NK cell/innate lymphoid cell (ILC) lineages, as evidenced by Thy1 and NK1.1 expression and lack of MHC class II expression by IFNγ⁺ cells (Extended Data Fig. 8a). We identified an H2-K^b epitope in IFNγ (IFNγ_69–78 QIISFYLRLF), the immunization of which resulted in a strong CD8⁺ T cell response in Ifng^–/–, but not WT, animals (Extended Data Fig. 8b). We next established the following four groups of BM chimeras: (1) WT → WT, (2) Ifng^–/– → WT, (3) WT → Ifng^–/– and (4) Ifng^–/– → Ifng^–/– and immunized them with IFNγ_69–78 and an irrelevant foreign peptide (ovalbumin (OVA_257–264) SIINFEKL). Intracellular staining for tumor necrosis factor (TNF) after a brief peptide restimulation was used as a readout for antigen-specific CD8⁺ T cell responses. All four groups of chimeras showed a response to the foreign epitope immunization (Extended Data Fig. 8c). By contrast, only Ifng^–/– → Ifng^–/– chimeras exhibited a strong response to IFNγ_69–78 immunization, whereas all other groups of chimeras were equally tolerant to this epitope (Fig. 8). Thus, expression of IFNγ solely by hematopoietic cells can induce CD8⁺ T cell tolerance to this molecule. Together with the T cell/NK cell/ILC-restricted expression, these results suggest that antigen expression by these effector subsets is sufficient for induction of tolerance to this endogenous inflammation-associated self-antigen.

**Fig. 8: Expression of IFNγ solely by hematopoietic cells is sufficient to induce CD8⁺ T cell tolerance.**

Discussion

Self/non-self discrimination by the adaptive immune system largely relies on the context of antigen encounter. For T lymphocytes, a variety of tolerance mechanisms ensure inactivation of autoreactive cells in the thymus or in the periphery in the absence of inflammation induced by pathogen invasion. Only if a mature T lymphocyte encounters an antigen in the context of inflammation does it receive all signals required to induce clonal expansion and execution of effector programs. The expression of numerous self-molecules involved in T cell effector programs therefore overlaps in time and space with the emergence of pathogen-derived antigens. In this study, we aimed to investigate how the immune system distinguishes such inflammation-associated self-antigens from pathogen-derived molecules. We demonstrated that CD8⁺ T cell tolerance to inflammation-associated self-antigens can be induced in the thymus through antigen presentation by a variety of thymic cell types. These include two cell types not previously implicated in T cell tolerance induction—thymus-resident innate-like T cells and thymic eosinophils. We showed that minute numbers of thymic innate-like T lymphocytes, cells that express self-antigens characteristic of all major T cell effector programs, were sufficient to efficiently eliminate autoreactive CD8⁺ thymocytes. We demonstrated that CD8⁺ T cell tolerance induction by innate-like T lymphocytes can be achieved without the involvement of professional APCs through direct T cell-to-T cell antigen presentation. We also provided evidence for the contribution of effector lymphocytes to the induction of tolerance to an endogenous inflammation-associated self-antigen. Finally, we showed that decreased abundance of a model inflammation-associated self-antigen in the thymus can result in a spontaneous fratricide reaction, causing complete elimination of peripheral effector T cell subsets expressing this antigen.

This unprovoked and extremely efficient elimination of a whole class of effector T cells after breach of tolerance to a single inflammation-associated self-antigen highlights the ultimate importance of tolerance to this group of molecules. The importance of this tolerance seems to be further reflected by the high redundancy of cell types that express such antigens in the thymus. Such redundant expression was also characteristic for IFNγ that was expressed by thymic innate and innate-like lymphocytes and a small group of TECs. Well in line with our observations with model antigens, both radioresistant and radiosensitive cells were able to mediate CD8⁺ T cell tolerance to this endogenous inflammation-associated self-antigen. Together with the T cell/NK cell/ILC-restricted expression of IFNγ in the hematopoietic compartment, these results provide evidence for the role of effector lymphocytes in the induction of tolerance to this proinflammatory cytokine. Although by themselves these experiments cannot discriminate between central and peripheral tolerance mechanisms, the abundance of IFNγ-expressing cells in the thymus and our results obtained with model antigens suggest that central tolerance is likely to contribute to the neutralization of IFNγ-reactive CD8⁺ T cells.

Although cytotoxic responses to inflammation-associated molecules have not, to our knowledge, been studied in the context of autoimmunity, many human autoimmune diseases are associated with autoantibodies against a broad range of proinflammatory cytokines⁵⁴, suggesting that CD4⁺ T cell tolerance to these self-antigens is broken in such patients. Interestingly, in our study using IL-4–GFP and IL-17A–GFP as model inflammation-associated self-antigens, we observed profound CD8⁺, but not CD4⁺, T cell tolerance. We also did not find any evidence of GFP cross-presentation by thymic DCs in this system, even when innate-like T cells expressed a secreted form of GFP. However, we did not test if secreted GFP may be subjected to MHC class II presentation by professional APCs. Moreover, the situation may be different for some endogenous secreted antigens, for example, if professional thymic APCs would express a receptor for such a molecule. It remains to be investigated if in these scenarios secreted inflammation-associated self-antigens would be taken up and presented by DCs, enabling the induction of CD4⁺ T cell tolerance. Alternatively, it is also possible that induction of CD4⁺ T cell tolerance is less stringent for inflammation-associated self-antigens, making them frequent targets in autoimmunity. Finally, unlike their mouse counterparts, human thymocytes can express MHC class II molecules⁵⁵, and the expression of genes encoding MHC class II chains was relatively high in some of the thymocyte populations that expressed T cell effector genes (Extended Data Fig. 1). It is therefore conceivable that in humans, these cells could also present antigens to developing CD4⁺ thymocytes and therefore contribute to the induction of CD4⁺ T cell tolerance.

Although the role of direct antigen presentation by professional APCs in the elimination of autoreactive thymocytes is well documented^8,10, the contribution of antigen presentation by nonprofessional APCs to this process under physiological circumstances remained unclear. In in vitro systems, peptide-pulsed thymocytes are capable of direct antigen presentation and induction of cell death in autoreactive thymocytes^56,57. In an in vivo setting, mature T cells were shown to mediate the induction of central tolerance to an alloantigen⁵⁸. In addition, it was demonstrated that autoantigens expressed by T cells⁵⁹ and thymic fibroblasts⁶⁰ can contribute to the induction of T cell tolerance. However, it remains unclear if the latter happens through direct antigen presentation of autoantigens or through their cross-presentation by professional APCs.

In this study, we have demonstrated that endogenous antigen expression by minute numbers of thymic-resident innate-like T cells, thymic eosinophils and conventional developing thymocytes was sufficient to eliminate autoreactive CD8⁺ thymocytes. Moreover, we demonstrated that innate-like T cells can induce CD8⁺ T cell tolerance by directly presenting self-antigens to autoreactive thymocytes. These results substantially expand the array of known cell types that can mediate the induction of T cell tolerance and suggest that most cell types that express MHC class I and are present in the thymus are likely able to contribute to the induction of CD8⁺ T cell tolerance. In addition to B cells and eosinophils, mouse and/or human thymi were reported to harbor several NK cell and ILC subsets, plasma cells, mast cells, erythroblasts, megakaryocytes and recirculating memory T cells^35,61,62,63. Thus, in addition to ectopic antigen expression by mTECs, recruitment and/or retention of these ‘ectopic cell types’ may represent a mechanism that ensures representation of hematopoietic system-derived self-antigens in the thymus.

Recent studies have demonstrated that large fractions of iNKT cells and γδT cells in the thymus are tissue-resident cells^23,24. Moreover, thymus-resident innate-like T cells are in the center of a complex ‘ecosystem’ that actively ensures their presence in the thymus and production of inflammatory mediators by these cells. Indeed, a tuft cell-like subset of mTECs is required for iNKT cell accumulation in the thymus⁵, whereas production of IL-4 by these cells is ensured by their CD1d-dependent activation by thymic myeloid cells⁶⁴. In turn, type 2 cytokines secreted by iNKT cells support the intrathymic generation of IFNγ-producing CD8⁺ T cells^25,26, regulate the recruitment of thymic eosinophils²⁷ and induce the activation thymic DCs²⁹. In addition, proinflammatory cytokines of both stromal (type 1 IFNs, IL-15, IL-6 and thymic stromal lymphopoetin) and innate-like T cell (IFNγ) origin were shown to regulate thymic T cell development^65,66, indicating that these mediators are present in the thymic microenvironment in biologically meaningful amounts. Moreover, several inflammatory pathways, including type 1/type 3 IFN signaling and Toll-like receptor and/or IL-1β signaling, are constitutively active in TECs^30,31.

In this study, we demonstrated that at least two cellular components of this inflammatory network (thymic innate-like T cells and eosinophils) were sufficient to mediate the induction of remarkably efficient tolerance to antigens that they express. Taking these observations together, it is tempting to speculate that constitutive activation of a plethora of inflammatory pathways in the thymus may have evolved to ensure induction of tolerance to inflammation-associated self-antigens, a class of molecules that otherwise largely mirror the spatial and temporal distribution of pathogen-derived antigens.

Methods

Mice

IL-4–GFP⁴¹, IL-17A–GFP and Ifng^–/–⁶⁷ mice were purchased from the Jackson Laboratory. Rag2^–/– mice were purchased from Janvier Labs. PLZF^lu/lu⁶⁸, OT-I⁶⁹, ACTB–GFP⁷⁰, Jedi-TCRαβ⁴⁴ and Tcrd^–/–Cd1d^–/–Mr1^–/–⁷¹ mice were described previously. WT C57BL/6J and BALB/c mice were obtained from Janvier Labs or were bred in-house. Jedi-TCRβ-only mice were generated from Jedi-TCRαβ mice by crossing with WT BALB/c mice and were further backcrossed to the BALB/c background for more than ten generations. WT C57BL/6J × BALB/c F1 mice (referred to as CB6F1), IL-4–GFP CB6F1, IL-17A–GFP CB6F1, ACTB–GFP CB6F1 and Jedi-TCRβ CB6F1 mice were generated by crossing mice on C57BL/6J and BALB/c backgrounds. For MHC mismatch experiments, IL-4–GFP CB6F1 mice were crossed with WT C57BL/6J mice to generate IL-4–GFP H2^b/d and IL-4–GFP H2^b/b mice. Tcrd^–/–Cd1d^–/–Mr1^–/– mice were bred and maintained at the Würzburg Institute for Systems Immunology. All other mice were bred and maintained at the Comparative Medicine Biomedicum facility of Karolinska Institutet (Stockholm, Sweden) and were kept under specific pathogen-free conditions. All mouse experiments were performed according to valid ethical permits, which were approved and regularly controlled by the Swedish and/or German Veterinary Authority.

Flow cytometry

Mouse spleen tissue, lymph nodes and thymi were collected, and single-cell suspensions were obtained by mincing through 70-μm cell strainers. Isolation of intestinal intraepithelial lymphocytes (iIELs) was performed as described previously⁷². Isolation of TECs was performed as described previously⁷³ with minor modifications. In short, each thymus was digested in 0.5–1 ml of RPMI 1640 medium with 1 mg ml^–1 collagenase IV and 100 U ml^–1 DNase I for 20 min at 37 °C before staining with APC-labeled anti-EPCAM, followed by enrichment using anti-APC microbeads (Miltenyi Biotec) and magnets from the EasySep Cell Isolation and Separation Technology platform (STEMCELL Technologies). For detection of antigen-specific T cells, single-cell suspensions were incubated with APC- and/or PE-conjugated H2-K^d GFP_200–208 or I-A^b GFP_81–95, PE- or BV421-conjugated CD1d-PBS-57 or APC-conjugated MR1-5-OP-RU tetramers (all provided by the National Institutes of Health tetramer core facility) for 1 h before enrichment and further cell-surface staining (for I-A^b GFP_81–95) or they were directly included into the surface antibody staining mix (for all other tetramers). Tetramer-binding CD4⁺ T cells were enriched by magnetic enrichment as described by Moon et al.⁷⁴. Dead cells were detected using a Live/Dead Fixable Aqua Dead Cell Stain kit for 405-nm excitation (Thermo Fisher Scientific) or the fixable viability dye eFluor780 (eBioscience) according to manufacturers’ instructions and were excluded from further analysis unless stated otherwise.

For the detection of cleaved caspase-3, cells were fixed and permeabilized with BD Cytofix/Cytoperm (BD Bioscience) and stained with rabbit anti-mouse active caspase-3 (5A1E) on ice for 30 min, followed by incubation with PE-conjugated anti-rabbit IgG (H + L) F(ab′)₂ fragment antibody on ice for 30 min. Intracellular cytokine staining was performed using a BD Cytofix/Cytoperm kit (BD Bioscience) following manufacturer’s instructions.

Data were acquired on an LSR Fortessa, FACSCanto II (BD Biosciences) or Cytek Aurora (Cytek Biosciences) flow cytometer and analyzed with FlowJo software v.10 (BD Biosciences). Cell sorting was performed using FACS Aria III and FACS Fusion cell sorters (BD Biosciences).

Flow cytometry staining reagents

Monoclonal antibodies specific for CD4 (GK1.5, RM4-8), CD3e (145-2C11, 17A2), CD8α (53-6.7), CD8β (YT5156.7.7), PD-1 (REA802, 29F.1A12), CD24 (M1/69), CD44 (IM7), CD19 (1D3, 6D5), B220 (RA3-6B2), CD45.1 (A20), CD45.2 (104), F4/80 (BM8), CD11c (N418), NK1.1 (PK136), TCRβ (H57-597), TCRγδ (GL3), H2-K^b (AF 6-885), H2-K^d (SF1-1.1), cKit (2B8), Ter119 (Ter119), SiglecF (E50-2440), MHC class II (M5/114.15), Ly51 (REA988), EPCAM (REA977), TCRV_α2 (B20.1), TCRV_β4 (REA729), TNF (MAb11), Thy1.2 (30-H12), Thy1.1 (OX7), Ly49 (14B11), CD122 (TM-β1), CCR7 (4B12), CD25 (REA568), CD80 (16-10A1), CD86 (GL-1), IFNγ (REA638, XMG1.2), IL-4 (11B11) and IL-17A (TC11-18H10) were purchased from BioLegend, BD Biosciences, Miltenyi Biotec or Thermo Fisher Scientific. Monoclonal antibody targeting caspase-3 (5A1E) and anti-rabbit IgG (H + L) F(ab′)₂ fragments were purchased from Cell Signaling Technology. Streptavidin-APC (405207) was purchased from BioLegend. Biotinylated UEA1 was purchased from Vector Laboratories. All antibodies and staining reagents were used at dilutions specified by the manufacturer or determined experimentally.

Generation of BM chimeras

For all BM chimera experiments, BM cells from donor mice were stained with APC-labeled CD4, CD8, CD3, TCRβ, TCRγδ, NK1.1 and Ter119 antibodies, followed by magnetic depletion of T and NK cells with anti-APC microbeads (Miltenyi Biotec). BM cells (2 × 10⁶–5 × 10⁶) were then transferred i.v. into lethally irradiated (split dose of 5 Gy, twice, using an Xstrahl CIX3 X-ray irradiator) recipients. For mixed BM chimera experiments shown in Fig. 3c, Jedi-TCRαβ BM cells were mixed with WT or IL-4–GFP BM cells at a 3:1 ratio before transfer. For mixed BM chimera experiments shown in Fig. 6b, Jedi-TCRβ H2^b/d BM cells were mixed at a 1:1 ratio with IL-4–GFP H2^b/b or IL-4–GFP H2^b/d BM cells before transfer. For mixed BM chimera experiments shown in Fig. 7, Jedi-TCRβ BM cells were mixed at a 1:20 ratio with IL-17A–GFP or WT BM cells before transfer into WT recipients (all on a CB6F1 genetic background). For mixed BM chimera experiments shown in Supplementary Fig. 2, Jedi-TCRβ or Jedi-TCRαβ BM cells were mixed at a 1:1 ratio with IL-4–GFP or WT BM cells before transfer into WT recipients (all on a BALB/c genetic background).

In vitro T_H2 cell and T_H17 cell differentiation

For T_H17 cell differentiation, CD4⁺ T cells from spleens of IL-17A–GFP mice on a CB6F1 genetic background were isolated by staining with APC-conjugated anti-CD4, followed by magnetic enrichment with anti-APC magnetic beads (Miltenyi Biotec). Enriched CD4⁺ T cells were cultured in six-well plates (precoated with 4 µg ml^–1 anti-CD3e (Biolegend) in PBS at 4 °C overnight) in IMDM supplemented with 10% fetal calf serum (FCS), 50 µM 2-mercaptoethanol, 2 mM l-glutamine, 100 U ml^–1 penicillin, 100 U ml^–1 streptomycin, 0.5 µg ml^–1 anti-CD28 (Biolegend), 50 ng ml^–1 IL-6 (Peprotech), 2 ng ml^–1 TGFβ (Peprotech), 2.5 µg ml^–1 anti-IL-4 (Biolegend) and 2.5 µg ml^–1 anti-IFNγ (Biolegend). The cells were used in in vivo killing assays 4 days after the start of the cultures.

For T_H2 cell differentiation, CD4⁺ T cells from spleens of WT C57BL/6J mice were isolated as described above. Enriched CD4⁺ T cells were cultured in six-well plates (precoated with 4 µg ml^–1 anti-CD3e in PBS at 4 °C overnight) in IMDM supplemented with 10% FCS, 50 µM 2-mercaptoethanol, 2 mM l-glutamine, 100 U ml^–1 penicillin, 100 U ml^–1 streptomycin, 0.5 µg ml^–1 anti-CD28, 10 ng ml^–1 IL-2 (Peprotech), 20 ng ml^–1 IL-4 (Peprotech) and 2.5 µg ml^–1 anti-IFNγ. Induction of theT_H2 molecular program was confirmed by side-by-side analysis of GFP expression in T_H2 cells differentiated from IL-4–GFP CD4⁺ T cells (80–90% of GFP⁺ cells). The cells were used for repetitive transfer into the indicated BM chimeras or in in vivo killing assays 4 days after the start of the cultures.

In vivo killing assays

In vitro-differentiated IL-17A–GFP T_H17 cells were labeled with 1 μM CTV (Thermo Fisher) in PBS with 0.5% bovine serum albumin (BSA) in a water bath at 37 °C for 8 min. Labeled cells were washed and i.v. transferred (0.5 × 10⁶) into the indicated groups of BM chimeras. Disappearance of GFP^hiCTV⁺ cells was assessed by flow cytometry in the spleens of the recipient mice 20 h after transfer.

In vitro-differentiated T_H2 cells were labeled with 1 μM CTV and mixed with 10 μM CTV-labeled freshly isolated naive T cells at a 1:3 ratio. Labeled cells were washed with PBS and i.v. transferred into the indicated groups of BM chimeras (2 × 10⁶ cells per mouse). The ratio between CTV^int (T_H2) and CTV^hi (naive) cells was assessed in the spleens of recipient mice 20 h after transfer by flow cytometry.

Construction of retroviral vectors

DNA fragments encoding full-length cytoplasmic OVA with N4 (SIINFEKL), Q4 (SIIQFEKL), T4 (SIITFEKL) and V4 (SIIVFEKL) versions of the OVA_257–264 epitope were synthesized by Twist Bioscience and cloned into a retroviral pMIG II vector⁷⁵ using BgIII and Xhol sites. To generate the Thy1.1-IRES-i.c.GFP construct, the Thy1.1-coding sequence was cloned into pMIG II using BglII and XhoI sites. To generate the Thy1.1-IRES-secGFP construct, the DNA sequence encoding a secretory signal peptide of the FSH-β subunit was inserted at the 5′ end of the GFP-coding sequence in the Thy1.1-IRES-i.c.GFP construct as described previously⁷⁶ with minor modifications. In brief, 5′- TGAAAAACACGATAATACCATGATGAAGTCGATCCAGCTTTGCATCCTACTCTGGTGCTTGAGAGCAGTCTGCTGCCATATGGTGAGCAAGGGCGAGG -3′ and 5′-CCTCGCCCTTGCTCACCATATGGCAGCAGACTGCTCTCAAGCACCAGAGTAGGATGCAAAGCTGGATCGACTTCATCATGGTATTATCGTGTTTTTCA -3′ oligonucleotides were annealed and mixed with linearized pMIG II vector (digested with NcoI) and incubated at 50 °C for 30 min with NEBuilder HiFi DNA Assembly Master Mix (New England Biolabs). To generate the Thy1.1-only construct, the Thy1.1-IRES-i.c.GFP construct was digested with NotI and NcoI to remove the IRES-i.c.GFP cassette, followed by removal of overhangs using Klenow fragment (New England Biolabs) and self-ligation of the digested vector. To generate the Bcl-2-P2A-Thy1.1-IRES-secGFP construct, the Thy1.1-coding sequence in the Thy1.1-IRES-secGFP construct was replaced with a DNA fragment encoding Bcl-2-P2A-Thy1.1 (synthesized by Twist Bioscience) using BglII and XhoI sites.

Transfection and retrovirus production

HEK293T cells were cultured in DMEM supplemented with 10% FCS, 50 µM 2-mercaptoethanol, 2 mM l-glutamine, 100 U ml^–1 penicillin and 100 U ml^–1 streptomycin at 37 °C and 5% CO₂ until the confluency reached 70%. To test GFP expression and secretion, HEK293T cells were transiently transfected with Thy1.1-IRES-i.c.GFP or Thy1.1-IRES-secGFP constructs by calcium phosphate transfection, as previously described⁷⁷. The culture medium was replaced with DMEM (without phenol red) 6 h after transfection. Supernatants were collected 8, 24 and 48 h after transfection, and fluorescence intensity was measured with a Varioskan LUX Multimode Microplate Reader (Thermo Fisher) at an excitation wavelength of 480 nm and emission wavelength of 520 nm (with a bandwidth of 12 nm).

Retrovirus-containing supernatants were generated by transient cotransfection of HEK293T cells with the retroviral constructs described above and pCL-Eco packaging vector using calcium phosphate transfection. Viral supernatants were collected 48 h after transfection, mixed thoroughly with 5× PEG solution (40% polyethylene glycol 10,000 (Sigma) and 2.4% NaCl dissolved in water) and incubated overnight at 4 °C. The next day, virus particles were precipitated by centrifugation at 2,000g and 4 °C for 1 h, resuspended in fresh culture medium to reach 100-times the concentration and stored at −80 °C.

Immunization experiments

To study antigen-specific CD8⁺ T cell responses, mice were intraperitoneally injected with 200 µg of GFP_200–208 peptide (HYLSTQSAL, H2-K^d epitope; Genscript) dissolved in PBS plus 50 µg of anti-CD40 (clone FGK4.5, BioXcell) and 50 µg of poly(I:C) (Invivogen).

For CD4⁺ T cell peptide immunizations, mice were subcutaneously injected with 100 µl of an emulsion containing 100 µg of GFP_81–95 peptide (HDFFKSAMPEGYVQE, I-A^b epitope; Genscript). Emulsions were prepared by mixing the peptide in PBS in a 1:1 ratio with complete Freund’s adjuvant (Sigma-Aldrich).

For IFNγ and OVA peptide immunization, BM chimeric mice were intraperitoneally injected with 200 µg of IFNγ_69–78 peptide (QIISFYLRLF, H2-K^b epitope; Genscript) and 50 µg of OVA_257–264 peptide (SIINFEKL, H2-K^b epitope; Genscript) with 50 µg of anti-CD40 and 50 µg of poly(I:C).

Cytokine production assays

For characterization of thymic effector subsets in WT and PLZF^lu/lu mice, isolated thymocytes were incubated in RPMI 1640 culture medium supplemented with 10% FCS, 50 µM 2-mercaptoethanol, 2 mM l-glutamine, 100 U ml^–1 penicillin, 100 U ml^–1 streptomycin, 50 ng ml^–1 phorbol 12-myristate 13-acetate (PMA; Sigma), 1 μg ml^–1 ionomycin (Sigma) and 3 μg ml^–1 brefeldin A (Thermo Fisher) at 37 °C and 5% CO₂. After 5 h, thymocytes were collected for surface antibody and viability staining, followed by intracellular staining using a BD Cytofix/Cytoperm kit (BD Biosciences) according to manufacturer’s instructions. For detection of intracellular cytokines in Tcrd^–/–Cd1d^–/–Mr1^–/– mice, isolated thymocytes were cultured in RPMI 1640 culture medium with 25 ng ml^–1 PMA, 1 μg ml^–1 ionomycin and 1 μg ml^–1 brefeldin A for 4 h at 37 °C and 5% CO₂.

For detection of antigen-specific CD8⁺ responses in peptide-immunized Ifng^–/–/WT BM chimeras, spleens were collected 6 days after immunization and processed to a single-cell suspension. After erythrocyte lysis, splenocytes were resuspended in IMDM containing the indicated peptides (final concentration of 10 µg ml^–1) and cultured in a six-well plate at a density of 2 × 10⁶ cells per ml (1 × 10⁷ cells per well) at 37 °C and 5% CO₂. After a 1-h incubation, brefeldin A (3 μg ml^–1) was added to the culture. Cells were collected, stained as described above and analyzed 5 h after the addition of brefeldin A. Antigen-specific CD8⁺ T cells were identified as TCRβ⁺CD8⁺CD44^hiTNF⁺ in Fig. 8 and Extended Data Fig. 8b,c.

Generation of the Jedi TCR-expressing reporter cell line

NFAT-sFT reporter-containing 16.2c11 cells⁵³ (a kind gift of J. Kisielow, Repertoire Immune Medicines) were retrovirally transduced with constructs encoding Jedi TCR, CD8α and CD8β. Retroviral transfection was performed as described previously⁷⁷. Cells were sorted and expanded to generate a CD8α⁺CD8β⁺ Jedi TCR-expressing reporter cell line.

In vitro cocultures with Jedi TCR-expressing reporter cells

Thymic iNKT cells (gated as GFP⁺ CD1d tetramer-binding cells for IL-4–GFP and IL-17A–GFP mice and as CD1d tetramer-binding cells for WT mice), eosinophils (gated as SiglecF⁺SSC^hi) and DCs (gated as CD11c⁺MHC class II⁺) from WT, IL-4–GFP and IL-17A–GFP mice were sorted and mixed with CD8α⁺CD8β⁺ Jedi TCR-expressing 16.2c11 reporter cells in a 10:1 (iNKT cell:Jedi reporter cell and eosinophil:Jedi reporter cell) or 1:1 (DC:Jedi reporter cell) ratio in IMDM supplemented with 10% FCS, 50 µM 2-mercaptoethanol, 2 mM l-glutamine, 100 U ml^–1 penicillin and 100 U ml^–1 streptomycin. Cells were cultured in 96-well V-bottom plates at 37 °C and 5% CO₂ for approximately 20 h and analyzed by flow cytometry.

Retroviral infection of thymic iNKT cells

Single-cell suspensions of thymocytes were stained with APC-labeled antibodies to CD24 and CD8 (anti-CD11c was added for the experiment shown in Extended Data Fig. 6e), followed by magnetic depletion with anti-APC microbeads (Miltenyi Biotec) to enrich for iNKT cells. Enriched thymic iNKT cells were cultured at a density of 1 × 10⁶ cells per ml in IMDM supplemented with 10 ng ml^–1 IL-7 (Peprotech), 100 ng ml^–1 IL-15 (Peprotech), and 0.5 μg ml^–1 CD1d-PBS-57 tetramer. After 2 days, iNKT cells were centrifuged at 400g at room temperature for 5 min and resuspended in fresh IMDM containing 10 ng ml^–1 IL-7, 100 ng ml^–1 IL-15, 4 μg ml^–1 polybrene (Sigma) and 20 μl ml^–1 concentrated virus particles, transferred to plates precoated with 20 μg ml^–1 RetroNectin (Takara) and spin infected at 1,000g and 32 °C for 1 h. One more round of infection was performed 6 h after the first round of infection. Eighteen hours after the second round of infection, the culture medium was replaced with fresh IMDM containing 10 ng ml^–1 IL-7 and 100 ng ml^–1 IL-15. Forty-eight hours after the first round of infection, iNKT cells were used in RTOC and intrathymic transfer experiments.

Immunofluorescence microscopy of thymus sections

Organ sections were prepared as described previously⁷⁸. For GFP detection, slides were incubated with Alexa Fluor 488-conjugated GFP booster (Chromotek), and staining was performed according to the manufacturer’s instructions. For mTEC detection, slides were incubated with biotinylated UEA1 (Vector Laboratories) followed by incubation with APC-conjugated streptavidin (BioLegend) according to the manufacturer’s instructions. Confocal images were acquired on an LSM 800 system (Carl Zeiss) at the Biomedicum Imaging Core at the Karolinska Institute. Images were processed using Zen 2.3 Black Edition (Carl Zeiss) or Imaris (Bitplane) imaging software.

RTOCs

RTOCs were performed as described previously⁷⁹. Thymi from embryonic day 14.5–15.5 CB6F1 embryos were isolated, and single-cell suspensions were prepared by digesting in PBS containing 0.125% trypsin and 1.325 mM EDTA for 15–30 min at 37 °C. In total, 6 × 10⁴–9 × 10⁴ (or 1.6 × 10⁴ for live-cell imaging) Jedi DP thymocytes (gated as CD4⁺CD8⁺ cells) sorted from the thymi of Jedi-TCRαβ mice (on a B10.D2 genetic background) and sorted iNKT cells (gated as CD1d tetramer-binding cells without additional gating on GFP) from the thymi of WT and IL-4–GFP or IL-17A–GFP mice or CD4SP thymocytes from WT and ACTB–GFP mice or thymic eosinophils (sorted as SiglecF⁺SSC^hi cells) from WT and IL-4–GFP mice were mixed with 2 × 10⁵–3 × 10⁵ cells from digested fetal CB6F1 thymi. RTOCs comparing i.c.GFP and secGFP were established as described above but with sorted Thy1.1⁺ thymic iNKT cells (on a CB6F1 background) transduced with the indicated constructs. RTOCs with OT-I cells were established as described above, but digested embryonic day 15.5 WT thymi were reaggregated with OT-I DP thymocytes (sorted as CD4⁺CD8⁺ cells) and sorted GFP⁺ cOVA-IRES-GFP-transduced thymic iNKT cells (all on a C57BL/6J background). Cell pellets were resuspended in less than 1.5 µl of RPMI 1640 (supplemented with 10% FCS, 50 µM 2-mercaptoethanol, 10 mM HEPES, 2 mM l-glutamine, 1× nonessential amino acids, 1 mM sodium pyruvate, 100 U ml^–1 penicillin and 100 U ml^–1 streptomycin) per RTOC. The cell suspension was transferred onto the membrane (Millipore Millicell 0.4-µm cell culture inserts) and placed in a six-well plate (Sarstedt) filled with the same medium as described above. For eosinophil-containing RTOCs, 5 ng ml^–1 IL-33 (Peprotech) was added to the culture medium. For CD80/CD86 blocking experiments, 10 µg ml^–1 anti-CD80 (16-10A1, BioLegend) and anti-CD86 (GL-1, BioLegend) or the corresponding isotype controls (HTK888, BioLegend; RTK2758, BioLegend) were added to the RTOC culture medium. Reaggregation and further culture were performed in humidified chambers placed in the cell culture incubator at 37 °C and 5% CO_2. Flow cytometric analysis was performed on day 5 of the culture. Live-cell imaging was performed 16–20 h after setup.

RTOC live-cell imaging

For live-cell imaging, RTOCs were prepared as described above with minor modifications. Jedi DP thymocytes (gated as CD4⁺CD8⁺PD-1^lo/int cells) isolated from the thymi of Jedi mice and iNKT cells (gated as GFP⁺CD1d tetramer-binding cells) from the thymi of IL-4–GFP mice were sorted. After sorting, Jedi DP thymocytes were loaded with 5 µM Cal-520 AM (AAT Bioquest) in HBSS containing 2% FCS for 1 h at 37 °C. DP Jedi cells were then washed twice with FCS-containing RPMI 1640 culture medium and labeled with 2.5 µM CTV (Thermo Fisher) in PBS/0.5% BSA for 8 min in a water bath at 37 °C, while thymic iNKT cells were labeled with 2.5 µM CTY (Thermo Fisher) under the same labeling conditions. RTOCs were allowed to assemble for 16–20 h before transfer to compartmentalized glass-bottom dishes (ibidi). For imaging, RTOCs were first embedded in PureCol EZ Gel solution (Sigma-Aldrich) for 45 min at 37 °C. RTOC complete medium was then added on top of the collagen matrix. z-stacks of two RTOCs were first imaged for each condition at a rate of 1 frame per min for 1–3 h using a ×20/0.8-NA Plan-Apochromat objective and a fast Airyscan detector on an LSM880 microscope (Carl Zeiss) with the incubation chamber set to 37 °C and 5% CO₂. Afterward, longer acquisitions of 12 h at 1 frame per min were conducted using standard confocal settings.

RTOC live-cell imaging analysis

Maximum intensity projections of confocal datasets were generated and analyzed in Zen Lite software (Carl Zeiss). Cell contacts were defined by colocalization of an iNKT cell and a Jedi thymocyte for at least three frames (one frame = 1 min). Contact identified using maximum intensity projections was then validated using the full three-dimensional datasets. Consecutive detachment and attachment between two cells with short intervals (shorter than five frames) were counted as one contact. In total, 134 contacts (for RTOCs with WT iNKT cells) and 108 contacts (for RTOCs with IL-4–GFP iNKT cells) were analyzed manually for an increase in Cal-520 AM fluorescence to enumerate Ca²⁺ signaling events (Fig. 6e). For quantification of contact length, the number of frames where cell colocalization had been observed was counted. Contacts where cells could not be tracked throughout the entire contact (for example, cells masked by other cells and cells disappearing from the field of view) were excluded from this analysis. Contacts lasting for 2 frames or less were also excluded from quantification of contact length.

For quantification of Cal-520 AM signal intensity over time, ten contacts from each group (WT iNKT cells, IL-4–GFP iNKT cells with Ca²⁺ signaling detected and IL-4–GFP iNKT cells with no Ca²⁺ signaling detected) were randomly selected and analyzed in ImageJ. The Cal-520 AM signal intensity and CTV signal intensity in Jedi cells were quantified by measuring the mean intensity of each signal in the region of interest (ROI; corresponding to a Jedi cell) drawn manually for each frame. For each frame, Ca²⁺ signal intensity of the Jedi cell was calculated as Ca²⁺ signal intensity (per frame) = (mean intensity of Cal-520 AM in ROI)/(mean intensity of CTV in ROI). The baseline Ca²⁺ signal intensity for each Jedi cell was then calculated by averaging the Ca²⁺ signal intensity (per frame) for five frames right before contact initiation. Normalized Ca²⁺ signal intensities plotted in Fig. 6f and Extended Data Fig. 6g were calculated for each frame using the following equation: normalized Ca²⁺ signal intensity = (Ca²⁺ signal intensity (per frame))/(baseline Ca²⁺ signal intensity).

Intrathymic injections

Intrathymic injections were performed as previously described²⁴. Four- to 5-week-old Jedi-TCRβ recipient mice were anesthetized by intraperitoneal injection of a ketamine (80 mg per kg (body weight))/xylazine (8 mg per kg (body weight)) solution. Subcutaneous injection of carprofen (5 mg kg^–¹ (body weight)) was performed 30–60 min before surgery. When fully anesthetized, the fur was shaved, and a small incision in the skin above the sternum was made. For experiments shown in Fig. 5d, up to 1 × 10⁶ iNKT cells (gated as CD1d tetramer-binding cells without additional gating on GFP) sorted from the thymi of IL-4–GFP mice were resuspended in 10 µl of PBS, and 5 µl per thymus lobe was injected. For experiments shown in Extended Data Fig. 6e, total cells from Bcl-2-P2A-Thy1.1-IRES-secGFP transduction cultures containing 6 × 10⁵ transduced (Thy1.1⁺) iNKT cells were resuspended in 10 µl of PBS, and 5 µl per thymus lobe was injected. Bcl-2 was added to the construct to improve survival of intrathymically transferred iNKT cells. The skin incision was closed with sutures, and mice were intraperitoneally injected with atipamezole (4 mg per kg (body weight)). Carprofen (5 mg per kg (body weight)) was subcutaneously injected once a day until 72 h after surgery. Flow cytometric analysis was performed 7 days after injection.

Autoantibody profiling using bead-based protein microarrays

Serum from PLZF^lu/lu mice was tested for the presence of antibodies against autoantigens and proteins from pathogens. A custom microbead-based antigen array was created to profile serum samples for antibodies against cytokines, autoimmune-associated antigens and viral antigens, as previously described⁸⁰. The array was constructed by conjugating antigens to uniquely barcoded carboxylated magnetic beads (MagPlex-C, Luminex). Beads were qualified using prototype plasma or serum samples with known reactivities. Although the available array was constructed using human antigens, protein homology with mouse antigens still allowed for comparisons in mouse autoantibody development^81,82. Mouse serum samples were tested at a dilution of 1:100 in 0.05% PBS-Tween supplemented with 1% (wt/vol) BSA. Bound antibody was detected using R-PE-conjugated Fcγ-specific goat anti-mouse IgG F(ab′)₂ fragment (Jackson ImmunoResearch, 115-116-071) before analysis using a FlexMap3D instrument (Luminex). Binding events were displayed as median fluorescence intensity (MFI). For normalization, the MFI value for the ‘bare bead’ ID (no conjugated antigen) was subtracted from the MFI value for the antigen-conjugated bead ID. Samples were run in duplicate. Serum samples from MR/Fas^lpr/lpr mice (a mouse model for systemic lupus erythematosus) served as positive controls.

Analysis of public expression datasets

AnnData objects containing previously described human and mouse thymus cell atlases³⁵ were downloaded from https://developmental.cellatlas.io/thymus-development and were replotted using Scanpy⁸³ (version 1.9.1). The combined human effector T cell signature used in Extended Data Fig. 1 was generated using the sc.tl.score_genes Scanpy function and the following list of effector program genes: IL13, IL-17A, IL17F, IFNG, GZMB and GZMK. Analysis of the Immgen dataset was described previously³³.

Statistical analysis

Statistical analyses were performed with GraphPad Prism 9 software. Error bars represent standard deviation of biological replicates (or individual RTOCs) unless stated otherwise. Two-tailed unpaired Student’s t-tests or two-tailed Mann–Whitney tests were used to assess the statistical significance of one observed parameter between two experimental groups. An unpaired ANOVA with a Holm–Sidak multiple comparisons test or a Kruskal–Wallis test was used when more than two experimental groups were compared.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Source data are provided with this paper.

References

Kyewski, B. & Derbinski, J. Self-representation in the thymus: an extended view. Nat. Rev. Immunol. 4, 688–698 (2004).
CAS PubMed Google Scholar
Anderson, M. S. et al. Projection of an immunological self shadow within the thymus by the AIRE protein. Science 298, 1395–1401 (2002).
CAS PubMed Google Scholar
Takaba, H. et al. Fezf2 orchestrates a thymic program of self-antigen expression for immune tolerance. Cell 163, 975–987 (2015).
CAS PubMed Google Scholar
Michelson, D. A., Hase, K., Kaisho, T., Benoist, C. & Mathis, D. Thymic epithelial cells co-opt lineage-defining transcription factors to eliminate autoreactive T cells. Cell 185, 2542–2558 (2022).
CAS PubMed PubMed Central Google Scholar
Miller, C. N. et al. Thymic tuft cells promote an IL-4-enriched medulla and shape thymocyte development. Nature 559, 627–631 (2018).
CAS PubMed PubMed Central Google Scholar
Bornstein, C. et al. Single-cell mapping of the thymic stroma identifies IL-25-producing tuft epithelial cells. Nature 559, 622–626 (2018).
CAS PubMed Google Scholar
Givony, T. et al. Thymic mimetic cells function beyond self-tolerance. Nature 622, 164–172 (2023).
CAS PubMed Google Scholar
Březina, J., Vobořil, M. & Filipp, D. Mechanisms of direct and indirect presentation of self-antigens in the thymus. Front. Immunol. 13, 926625 (2022).
PubMed PubMed Central Google Scholar
Guerder, S., Viret, C., Luche, H., Ardouin, L. & Malissen, B. Differential processing of self-antigens by subsets of thymic stromal cells. Curr. Opin. Immunol. 24, 99–104 (2012).
CAS PubMed Google Scholar
Klein, L., Kyewski, B., Allen, P. M. & Hogquist, K. A. Positive and negative selection of the T cell repertoire: what thymocytes see (and don’t see). Nat. Rev. Immunol. 14, 377–391 (2014).
CAS PubMed PubMed Central Google Scholar
Perera, J. et al. Self-antigen-driven thymic B cell class switching promotes T cell central tolerance. Cell Rep. 17, 387–398 (2016).
CAS PubMed PubMed Central Google Scholar
Yamano, T. et al. Thymic B cells are licensed to present self antigens for central T cell tolerance induction. Immunity 42, 1048–1061 (2015).
CAS PubMed Google Scholar
Perera, J., Meng, L., Meng, F. & Huang, H. Autoreactive thymic B cells are efficient antigen-presenting cells of cognate self-antigens for T cell negative selection. Proc. Natl Acad. Sci. USA 110, 17011–17016 (2013).
CAS PubMed PubMed Central Google Scholar
von Boehmer, H. in Advances in Immunology, Vol. 84 (ed. Alt, F. W.) 201–238 (Academic Press, 2004).
Badr, M. E., Zhang, Z., Tai, X. & Singer, A. CD8 T cell tolerance results from eviction of immature autoreactive cells from the thymus. Science 382, 534–541 (2023).
CAS PubMed Google Scholar
Baldwin, T. A., Hogquist, K. A. & Jameson, S. C. The fourth way? Harnessing aggressive tendencies in the thymus. J. Immunol. 173, 6515–6520 (2004).
CAS PubMed Google Scholar
Lee, M. et al. Single-cell RNA sequencing identifies shared differentiation paths of mouse thymic innate T cells. Nat. Commun. 11, 4367 (2020).
CAS PubMed PubMed Central Google Scholar
Lee, Y. J., Holzapfel, K. L., Zhu, J., Jameson, S. C. & Hogquist, K. A. Steady-state production of IL-4 modulates immunity in mouse strains and is determined by lineage diversity of iNKT cells. Nat. Immunol. 14, 1146–1154 (2013).
CAS PubMed Google Scholar
Stetson, D. B. et al. Constitutive cytokine mRNAs mark natural killer (NK) and NK T cells poised for rapid effector function. J. Exp. Med. 198, 1069–1076 (2003).
CAS PubMed PubMed Central Google Scholar
Cohen, N. R. et al. Shared and distinct transcriptional programs underlie the hybrid nature of iNKT cells. Nat. Immunol. 14, 90–99 (2013).
CAS PubMed Google Scholar
Legoux, F., Salou, M. & Lantz, O. Unconventional or preset αβ T cells: evolutionarily conserved tissue-resident T cells recognizing nonpeptidic ligands. Annu. Rev. Cell Dev. Biol. 33, 511–535 (2017).
CAS PubMed Google Scholar
Ribot, J. C., Lopes, N. & Silva-Santos, B. γδ T cells in tissue physiology and surveillance. Nat. Rev. Immunol. 21, 221–232 (2021).
CAS PubMed Google Scholar
Tan, L. et al. Single-cell transcriptomics identifies the adaptation of Scart1⁺ Vγ6⁺ T cells to skin residency as activated effector cells. Cell Rep. 27, 3657–3671 (2019).
CAS PubMed Google Scholar
Wang, H. & Hogquist, K. A. CCR7 defines a precursor for murine iNKT cells in thymus and periphery. eLife 7, e34793 (2018).
PubMed PubMed Central Google Scholar
Verykokakis, M., Boos, M. D., Bendelac, A. & Kee, B. L. SAP protein-dependent natural killer T-like cells regulate the development of CD8⁺ T cells with innate lymphocyte characteristics. Immunity 33, 203–215 (2010).
CAS PubMed PubMed Central Google Scholar
Weinreich, M. A., Odumade, O. A., Jameson, S. C. & Hogquist, K. A. T cells expressing the transcription factor PLZF regulate the development of memory-like CD8⁺ T cells. Nat. Immunol. 11, 709–716 (2010).
CAS PubMed PubMed Central Google Scholar
Cosway, E. J. et al. Eosinophils are an essential element of a type 2 immune axis that controls thymus regeneration. Sci. Immunol. 7, eabn3286 (2022).
CAS PubMed PubMed Central Google Scholar
White, A. J. et al. A type 2 cytokine axis for thymus emigration. J. Exp. Med. 214, 2205–2216 (2017).
CAS PubMed PubMed Central Google Scholar
Breed, E. R. et al. Type 2 cytokines in the thymus activate Sirpα⁺ dendritic cells to promote clonal deletion. Nat. Immunol. 23, 1042–1051 (2022).
CAS PubMed PubMed Central Google Scholar
Otero, D. C., Baker, D. P. & David, M. IRF7-dependent IFN-β production in response to RANKL promotes medullary thymic epithelial cell development. J. Immunol. 190, 3289–3298 (2013).
CAS PubMed Google Scholar
Vobořil, M. et al. Toll-like receptor signaling in thymic epithelium controls monocyte-derived dendritic cell recruitment and T_reg generation. Nat. Commun. 11, 2361 (2020).
PubMed PubMed Central Google Scholar
ElTanbouly, M. A. & Noelle, R. J. Rethinking peripheral T cell tolerance: checkpoints across a T cell’s journey. Nat. Rev. Immunol. 21, 257–267 (2021).
CAS PubMed Google Scholar
Mingueneau, M. et al. The transcriptional landscape of αβT cell differentiation. Nat. Immunol. 14, 619–632 (2013).
CAS PubMed PubMed Central Google Scholar
Heng, T. S. & Painter, M. W. The Immunological Genome Project: networks of gene expression in immune cells. Nat. Immunol. 9, 1091–1094 (2008).
CAS PubMed Google Scholar
Park, J.-E. et al. A cell atlas of human thymic development defines T cell repertoire formation. Science 367, eaay3224 (2020).
CAS PubMed PubMed Central Google Scholar
Kernfeld, E. M. et al. A single-cell transcriptomic atlas of thymus organogenesis resolves cell types and developmental maturation. Immunity 48, 1258–1270 (2018).
CAS PubMed PubMed Central Google Scholar
Savage, A. K. et al. The transcription factor PLZF directs the effector program of the NKT cell lineage. Immunity 29, 391–403 (2008).
CAS PubMed PubMed Central Google Scholar
Kovalovsky, D. et al. The BTB-zinc finger transcriptional regulator PLZF controls the development of invariant natural killer T cell effector functions. Nat. Immunol. 9, 1055–1064 (2008).
CAS PubMed PubMed Central Google Scholar
Alonzo, E. S. et al. Development of promyelocytic zinc finger and ThPOK-expressing innate γδ T cells is controlled by strength of TCR signaling and Id3. J. Immunol. 184, 1268–1279 (2010).
CAS PubMed Google Scholar
Kreslavsky, T. et al. TCR-inducible PLZF transcription factor required for innate phenotype of a subset of γδ T cells with restricted TCR diversity. Proc. Natl Acad. Sci. USA 106, 12453–12458 (2009).
CAS PubMed PubMed Central Google Scholar
Mohrs, M., Shinkai, K., Mohrs, K. & Locksley, R. M. Analysis of type 2 immunity in vivo with a bicistronic IL-4 reporter. Immunity 15, 303–311 (2001).
CAS PubMed Google Scholar
Gambotto, A. et al. Immunogenicity of enhanced green fluorescent protein (EGFP) in BALB/c mice: identification of an H2-K^d-restricted CTL epitope. Gene Ther. 7, 2036–2040 (2000).
CAS PubMed Google Scholar
Malhotra, D. et al. Tolerance is established in polyclonal CD4⁺ T cells by distinct mechanisms, according to self-peptide expression patterns. Nat. Immunol. 17, 187–195 (2016).
CAS PubMed PubMed Central Google Scholar
Agudo, J. et al. GFP-specific CD8 T cells enable targeted cell depletion and visualization of T-cell interactions. Nat. Biotechnol. 33, 1287–1292 (2015).
CAS PubMed PubMed Central Google Scholar
Watanabe, M., Lu, Y., Breen, M. & Hodes, R. J. B7-CD28 co-stimulation modulates central tolerance via thymic clonal deletion and T_reg generation through distinct mechanisms. Nat. Commun. 11, 6264 (2020).
CAS PubMed PubMed Central Google Scholar
Pobezinsky, L. A. et al. Clonal deletion and the fate of autoreactive thymocytes that survive negative selection. Nat. Immunol. 13, 569–578 (2012).
CAS PubMed PubMed Central Google Scholar
Punt, J. A., Osborne, B. A., Takahama, Y., Sharrow, S. O. & Singer, A. Negative selection of CD4⁺CD8⁺ thymocytes by T cell receptor-induced apoptosis requires a costimulatory signal that can be provided by CD28. J. Exp. Med. 179, 709–713 (1994).
CAS PubMed Google Scholar
Buhlmann, J. E., Elkin, S. K. & Sharpe, A. H. A role for the B7-1/B7-2:CD28/CTLA-4 pathway during negative selection. J. Immunol. 170, 5421–5428 (2003).
CAS PubMed Google Scholar
Breed, E. R., Watanabe, M. & Hogquist, K. A. Measuring thymic clonal deletion at the population level. J. Immunol. 202, 3226–3233 (2019).
CAS PubMed Google Scholar
Hwang, I. et al. T cells can use either T cell receptor or CD28 receptors to absorb and internalize cell surface molecules derived from antigen-presenting cells. J. Exp. Med. 191, 1137–1148 (2000).
CAS PubMed PubMed Central Google Scholar
Sabzevari, H. et al. Acquisition of CD80 (B7-1) by T cells. J. Immunol. 166, 2505–2513 (2001).
CAS PubMed Google Scholar
Zehn, D., Lee, S. Y. & Bevan, M. J. Complete but curtailed T-cell response to very low-affinity antigen. Nature 458, 211–214 (2009).
CAS PubMed PubMed Central Google Scholar
Kisielow, J., Obermair, F.-J. & Kopf, M. Deciphering CD4⁺ T cell specificity using novel MHC–TCR chimeric receptors. Nat. Immunol. 20, 652–662 (2019).
CAS PubMed Google Scholar
Cappellano, G. et al. Anti-cytokine autoantibodies in autoimmune diseases. Am. J. Clin. Exp. Immunol. 1, 136–146 (2012).
PubMed PubMed Central Google Scholar
Seong Hoe, P. et al. HLA-DR expression in human fetal thymocytes. Hum. Immunol. 33, 294–298 (1992).
Google Scholar
Pircher, H., Müller, K.-P., Kyewski, B. A. & Hengartner, H. Thymocytes can tolerize thymocytes by clonal deletion in vitro. Int. Immunol. 4, 1065–1069 (1992).
CAS PubMed Google Scholar
Melichar, H. J., Ross, J. O., Taylor, K. T. & Robey, E. A. Stable interactions and sustained TCR signaling characterize thymocyte–thymocyte interactions that support negative selection. J. Immunol. 194, 1057–1061 (2015).
CAS PubMed Google Scholar
Tian, C., Bagley, J., Forman, D. & Iacomini, J. Induction of central tolerance by mature T cells. J. Immunol. 173, 7217–7222 (2004).
CAS PubMed Google Scholar
Webb, S. R. & Sprent, J. Induction of neonatal tolerance to Mlsa antigens by CD8⁺ T cells. Science 248, 1643–1646 (1990).
CAS PubMed Google Scholar
Nitta, T. et al. Fibroblasts as a source of self-antigens for central immune tolerance. Nat. Immunol. 21, 1172–1180 (2020).
PubMed Google Scholar
Throsby, M., Herbelin, A., Pléau, J.-M. & Dardenne, M. CD11c⁺ eosinophils in the murine thymus: developmental regulation and recruitment upon MHC class I-restricted thymocyte deletion. J. Immunol. 165, 1965–1975 (2000).
CAS PubMed Google Scholar
Gabrielli, S. et al. Murine thymic NK cells are distinct from ILC1s and have unique transcription factor requirements. Eur. J. Immunol. 47, 800–805 (2017).
CAS PubMed PubMed Central Google Scholar
Jones, R. et al. Dynamic changes in intrathymic ILC populations during murine neonatal development. Eur. J. Immunol. 48, 1481–1491 (2018).
CAS PubMed PubMed Central Google Scholar
Wang, H. et al. Myeloid cells activate iNKT cells to produce IL-4 in the thymic medulla. Proc. Natl Acad. Sci. USA 116, 22262–22268 (2019).
CAS PubMed PubMed Central Google Scholar
Etzensperger, R. et al. Identification of lineage-specifying cytokines that signal all CD8⁺-cytotoxic-lineage-fate ‘decisions’ in the thymus. Nat. Immunol. 18, 1218–1227 (2017).
CAS PubMed PubMed Central Google Scholar
Xing, Y., Wang, X., Jameson, S. C. & Hogquist, K. A. Late stages of T cell maturation in the thymus involve NF-κB and tonic type I interferon signaling. Nat. Immunol. 17, 565–573 (2016).
CAS PubMed PubMed Central Google Scholar
Dalton, D. K. et al. Multiple defects of immune cell function in mice with disrupted interferon-γ genes. Science 259, 1739–1742 (1993).
CAS PubMed Google Scholar
Buaas, F. W. et al. PLZF is required in adult male germ cells for stem cell self-renewal. Nat. Genet. 36, 647–652 (2004).
CAS PubMed Google Scholar
Hogquist, K. A. et al. T cell receptor antagonist peptides induce positive selection. Cell 76, 17–27 (1994).
CAS PubMed Google Scholar
Okabe, M., Ikawa, M., Kominami, K., Nakanishi, T. & Nishimune, Y. ‘Green mice’ as a source of ubiquitous green cells. FEBS Lett. 407, 313–319 (1997).
CAS PubMed Google Scholar
Ataide, M. A. et al. Lymphatic migration of unconventional T cells promotes site-specific immunity in distinct lymph nodes. Immunity 55, 1813–1828 (2022).
CAS PubMed Google Scholar
Sheridan, B. S. & Lefrançois, L. Isolation of mouse lymphocytes from small intestine tissues. Curr. Protoc. Immunol. 3, 3.19.11 (2012).
Google Scholar
Xing, Y. & Hogquist, K. A. Isolation, identification, and purification of murine thymic epithelial cells. J. Vis. Exp. 8, e51780 (2014).
Google Scholar
Moon, J. J. et al. Tracking epitope-specific T cells. Nat. Protoc. 4, 565–581 (2009).
CAS PubMed PubMed Central Google Scholar
Holst, J. et al. Generation of T-cell receptor retrogenic mice. Nat. Protoc. 1, 406–417 (2006).
CAS PubMed Google Scholar
Roh, J. Y. et al. Modification of enhanced green fluorescent protein for secretion out of cells. Biotechnol. Bioprocess Eng. 18, 1135–1141 (2013).
CAS Google Scholar
Dunst, J. et al. Recognition of synthetic polyanionic ligands underlies ‘spontaneous’ reactivity of Vγ1 γδTCRs. J. Leukoc. Biol. 107, 1033–1044 (2020).
CAS PubMed Google Scholar
Rauschmeier, R. et al. Bhlhe40 function in activated B and T_FH cells restrains the GC reaction and prevents lymphomagenesis. J. Exp. Med. 219, e20211406 (2022).
CAS PubMed Google Scholar
Anderson, G. & Jenkinson, E. J. Fetal thymus organ culture. CSH Protoc. 2007, pdb.prot4808 (2007).
Chang, S. E. et al. New-onset IgG autoantibodies in hospitalized patients with COVID-19. Nat. Commun. 12, 5417 (2021).
CAS PubMed PubMed Central Google Scholar
Degn, S. E. et al. Clonal evolution of autoreactive germinal centers. Cell 170, 913–926 (2017).
CAS PubMed PubMed Central Google Scholar
de Carvalho, R. V. H. et al. Clonal replacement sustains long-lived germinal centers primed by respiratory viruses. Cell 186, 131–146 (2023).
PubMed Google Scholar
Wolf, F. A., Angerer, P. & Theis, F. J. SCANPY: large-scale single-cell gene expression data analysis. Genome Biol. 19, 15 (2018).
PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We thank M. Aichinger and G. Wirnsberger for a discussion that facilitated the initiation of this project. We thank the National Institutes of Health tetramer core facility for the preparation of all tetramers used in this study. This study was supported by the Swedish Research Council (grants 2017-01118 and 2021-01468 to T.K.), Cancerfonden (grants CAN 2018/710 and 21 1602 Pj to T.K.), Barncancerfonden (grants PR2021-017 and PR2023-0091 to T.K.), Radiumhemmets Forskningsfonder (grants 211192 and 231233 to T.K.), a stipend from the German Research Foundation (DU 1964/1-1 to J.D.) and a stipend from the China Scholarship Council (to Y.Y.). 16.2c11 reporter cells were provided by J. Kisielow (Repertoire Immune Medicines). We thank S. Edwards and the Advanced Light Microscopy facility at Science for Life Laboratory for their help setting up the imaging of the RTOCs. We thank J. Coquet, S. Nylén and A. Gigliotti Rothfuchs (Karolinska Insitutet) and members of their laboratories for their help with experiments that were not included in the manuscript. We thank all members of the Kreslavsky laboratory for their support, discussions of the project and critical reading of the manuscript.

Funding

Open access funding provided by Karolinska Institute.

Author information

These authors contributed equally: Yuanyuan You, Josefine Dunst.

Authors and Affiliations

Division of Immunology and Allergy, Department of Medicine Solna, Karolinska Institutet, Karolinska University Hospital, Stockholm, Sweden
Yuanyuan You, Josefine Dunst, Kewei Ye, Annika Reinhardt & Taras Kreslavsky
Center for Molecular Medicine, Karolinska Institutet, Stockholm, Sweden
Yuanyuan You, Josefine Dunst, Kewei Ye, Annika Reinhardt, Natalia R. Comet, Carmen Gerlach & Taras Kreslavsky
Department of Applied Physics, Science for Life Laboratory, KTH Royal Institute of Technology, Stockholm, Sweden
Patrick A. Sandoz & Björn Önfelt
Institute of Immunology, Hannover Medical School, Hannover, Germany
Inga Sandrock & Immo Prinz
Division of Rheumatology, Department of Medicine Solna, Karolinska Institutet, Karolinska University Hospital, Stockholm, Sweden
Natalia R. Comet & Carmen Gerlach
Max Planck Research Group, Würzburg Institute of Systems Immunology, Julius-Maximilians-Universität Würzburg, Würzburg, Germany
Rupak Dey Sarkar & Wolfgang Kastenmüller
Department of Medicine, Division of Immunology and Rheumatology, Stanford University, Stanford, CA, USA
Emily Yang & Paul J. Utz
Epigenetic Factors in Normal and Malignant Hematopoiesis Lab, CRCM, CNRS, INSERM, Institut Paoli Calmettes, Aix Marseille University, Marseille, France
Estelle Duprez
Equipe Labellisée Ligue Nationale Contre le Cancer, Paris, France
Estelle Duprez
Department of Cancer Immunology and Virology, Dana-Farber Cancer Institute, Boston, MA, USA
Judith Agudo
Parker Institute for Cancer Immunotherapy, Dana-Farber Cancer Institute, Boston, MA, USA
Judith Agudo
Department of Immunology, Harvard Medical School, Boston, MA, USA
Judith Agudo
Ludwig Center at Harvard, Boston, MA, USA
Judith Agudo
The Precision Immunology Institute, Icahn School of Medicine at Mount Sinai, New York, NY, USA
Brian D. Brown
Icahn Genomics Institute, Icahn School of Medicine at Mount Sinai, New York, NY, USA
Brian D. Brown
Department of Immunology and Immunotherapy, Icahn School of Medicine at Mount Sinai, New York, NY, USA
Brian D. Brown
Institute for Immunity, Transplantation and Infection, Stanford University School of Medicine, Stanford, CA, USA
Paul J. Utz
Hamburg Center for Translational Immunology (HCTI), University Medical Center Hamburg-Eppendorf, Hamburg, Germany
Immo Prinz
Institute of Systems Immunology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany
Immo Prinz
Center for Infectious Medicine, Department of Medicine Huddinge, Karolinska Institutet, Stockholm, Sweden
Björn Önfelt

Authors

Yuanyuan You
View author publications
Search author on:PubMed Google Scholar
Josefine Dunst
View author publications
Search author on:PubMed Google Scholar
Kewei Ye
View author publications
Search author on:PubMed Google Scholar
Patrick A. Sandoz
View author publications
Search author on:PubMed Google Scholar
Annika Reinhardt
View author publications
Search author on:PubMed Google Scholar
Inga Sandrock
View author publications
Search author on:PubMed Google Scholar
Natalia R. Comet
View author publications
Search author on:PubMed Google Scholar
Rupak Dey Sarkar
View author publications
Search author on:PubMed Google Scholar
Emily Yang
View author publications
Search author on:PubMed Google Scholar
Estelle Duprez
View author publications
Search author on:PubMed Google Scholar
Judith Agudo
View author publications
Search author on:PubMed Google Scholar
Brian D. Brown
View author publications
Search author on:PubMed Google Scholar
Paul J. Utz
View author publications
Search author on:PubMed Google Scholar
Wolfgang Kastenmüller
View author publications
Search author on:PubMed Google Scholar
Carmen Gerlach
View author publications
Search author on:PubMed Google Scholar
Immo Prinz
View author publications
Search author on:PubMed Google Scholar
Björn Önfelt
View author publications
Search author on:PubMed Google Scholar
Taras Kreslavsky
View author publications
Search author on:PubMed Google Scholar

Contributions

Y.Y. performed most of the experiments. J.D. designed and supervised some of the experiments and performed most of the intrathymic transfer experiments and the initial RTOC experiments. K.Y. contributed to the RTOC and peptide immunization experiments. P.A.S. and B.Ö. performed RTOC imaging. N.R.C. and C.G. provided expertise and training for the intrathymic transfer experiments. I.S. and I.P. performed the initial experiments with IL-17A–GFP mice. E.D. provided PLZF^lu/lu mice and collected sera from these animals. J.A. and B.D.B. generated and provided Jedi mice. R.D.S. and W.K. performed analyses of Tcrd^–/–Cd1d^–/–Mr1^–/– mice. E.Y. and P.J.U. performed antigen array experiments. A.R. and K.Y. analyzed the data and prepared the figures. T.K. suggested the project idea, supervised the study and wrote the manuscript. Y.Y., J.D., K.Y., A.R., C.G. and T.K. edited the manuscript.

Corresponding author

Correspondence to Taras Kreslavsky.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Peer review

Peer review information

Nature Immunology thanks Ludger Klein and Jung Hyun Park for their contribution to the peer review of this manuscript. Primary Handling Editor: S. Houston, in collaboration with the Nature Immunology team. Peer reviewer reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Human thymic innate and innate-like lymphocytes express a broad spectrum of inflammation-associated self-antigens.

UMAP highlighting the expression of combined effector T cell gene signature (generated as describe in Methods) (top) and UMAPs highlighting the expression of the indicated genes in the human thymus cell atlas scRNA-seq dataset.

Extended Data Fig. 2 Characterization of GFP-expressing cell types in IL-4-GFP thymi.

a. Gating strategy for characterization of GFP-expressing cell types in IL4-GFP thymi. b. Frequency of GFP⁺ cells in the indicated thymic subsets in IL4-GFP mice (n = 3 mice). c. Frequency of GFP⁺ cells among mTECs (gated as EPCAM⁺CD45^–UEA1⁺Ly51^–) in IL4-GFP mice (n = 3 mice). d. Images of IL4-GFP and WT thymi sections as analyzed by confocal immunofluorescence microscopy as in Fig. 3b. An area of the same image as in Fig. 3b is shown for the IL4-GFP mouse. e. Frequency of GFP⁺ hematopoietic cells in the thymi of mixed BM chimeras shown in Fig. 3c, e. Quantification (left) and representative flow cytometric plots (right) are shown (n = 4 chimeras for [WT+Jedi-TCRαβ]→WT thymi, n = 6 for the other groups). Data are presented as the mean ± s.d.

Source data

Extended Data Fig. 3 Antigen expression exclusively by small numbers of conventional thymocytes or thymic eosinophils is sufficient to induce elimination of autoreactive thymocytes.

a. Schematic representation of RTOC experiments shown in b-c. b-c RTOCs with Jedi-TCRαβ DP thymocytes established and analyzed as in Fig. 4a, b but with WT or Actb-GFP CD4SP thymocytes (b) and WT or IL4-GFP eosinophils (c) used as antigen source. Representative plots showing frequencies of GFP-expressing cells, frequencies of H2-K^d GFP_200-208 tetramer-binding Jedi thymocytes (quantification shown on the right) and expression of the indicated cell surface markers by the latter cells. Representative results of three independent experiments (b) (n = 2 RTOCs for WT CD4SP, n = 5 for Actb-GFP CD4SP) or results of one experiment (c) (n = 2 for WT eosinophils, n = 3 for IL4-GFP eosinophils). Data are presented as the mean ± s.d.

Source data

Extended Data Fig. 4 Assessment of possible effects of co-stimulation on induction of tolerance by innate-like T cells.

a. Expression of CD80 (top) and CD86 (bottom) by the indicated thymic populations (red) in an IL4-GFP mouse. Isotype control for CD80 is shown in grey (same in both rows). Percentage of positive cells and median fluorescent intensity (MFI) in the positive gate are indicated. The following thymic populations were analyzed: B cells (CD19⁺MHC-II⁺), DCs (CD11c⁺MHC-II⁺), eosinophils (IL4-GFP⁺SSC^hi), γδT cells (CD3⁺TCRγδ⁺), iNKT (TCRβ⁺CD1d-PBS-57 tetramer⁺), CD4SP (CD4⁺), CD8SP (CD8⁺), and DP cells (CD4⁺CD8⁺). b. RTOCs were established and analyzed as in Fig. 4b, but blocking antibodies against CD80 and CD86 or isotype control antibodies were added to culture medium. Representative results of two independent experiments (n = 5 RTOCs for WT isotype group, n = 6 for the other groups). Data are presented as the mean ± s.d. with NS: non-significant (P > 0.05), **P < 0.01, ***P < 0.001, and ****P < 0.0001. Data were analyzed by two-way ANOVA with Holm-Sidak’s multiple comparisons test. c. HEK293T cells were transfected with Thy1.1-IRES-i.c.GFP or Thy1.1-IRES-secGFP retroviral constructs. Expression of Thy1.1 and GFP by HEK293T cells 48 hours after transfection (left, contour plot overlay as well as histogram overlays for Thy1.1 and GFP are shown) and the mean value of GFP fluorescence in culture supernatants of the same cells at different timepoints (right) are shown. Error bars represent s.d. of technical replicates (n = 2 for untransfected group, n = 3 for the other groups).

Source data

Extended Data Fig. 5 Characterization of GFP-expressing cell types in IL17A-GFP thymi.

a. Gating strategy for characterization of GFP-expressing cell types in IL17A-GFP thymi. b. Frequency of GFP⁺ cells in the indicated thymic subsets in IL17A-GFP mice (n = 5 mice). c. Frequency of GFP⁺ cells among mTECs (gated as EPCAM⁺CD45^–UEA1⁺Ly51^–) from IL17A-GFP mice (n = 3 mice). d. WT and IL17A-GFP mice (on C57BL/6J background) were immunized s.c. with GFP_81-95 peptide in CFA or left unimmunized (WT only). 14 days after immunization, I-A^b GFP_81-95 tetramer-binding cells were magnetically enriched from pooled lymph nodes and spleens and quantified by flow cytometry. Representative flow cytometry plots gated on CD4 T cells (left) and absolute numbers of CD4⁺CD44⁺Tetramer⁺ T cells (right) are shown. Representative results of two independent experiments (n = 4 mice per group). e. Comparison of the frequencies of GFP⁺ hematopoietic cells in the thymi of IL17A-GFP mice and in IL17A-GFP→WT BM chimeras (analyzed ≥ 7 weeks after reconstitution) (n = 4 for IL17A-GFP mice, n = 6 for chimeras). Data are presented as the mean ± s.d. with NS: non-significant (P > 0.05). Data were analyzed by two-tailed Student’s t test (d).

Source data

Extended Data Fig. 6 Direct presentation by model antigen-expressing populations rather than cross-presentation by professional APCs is responsible for elimination of autoreactive GFP-specific CD8 thymocytes.

a. iNKT cells, eosinophils and DCs were sorted from thymi of WT, IL4-GFP and IL17A-GFP mice and co-cultured overnight with slow fluorescent timer (sFT) NFAT reporter-containing 16.2c11 cells that were engineered to express the Jedi TCR, CD8α and CD8β. Activation of NFAT signaling was measured by sFT-Blue reporter upregulation. b. RTOCs with sorted Jedi DP thymocytes and iNKT cells were established and analyzed as in Fig. 6a, but using iNKT cells sorted from WT mice (on H2^b/d background) or from IL17A-GFP mice on H2^b/d and H2^b/b backgrounds. Results of one experiment (n = 3 RTOCs for IL17A-GFP H-2^b/b group, n = 4 for the other groups). c. Thymic iNKT cells from H2^d/d (BALB/c background) or H2^b/b (C57BL/6J background) mice were transduced with Thy1.1-IRES-i.c.GFP, Thy1.1-IRES-secGFP and Thy1.1-only retroviruses. Sorted Thy1.1⁺ iNKT cells were added to RTOCs together with Jedi-TCRαβ DP thymocytes. Flow cytometric analysis and quantification are shown. Results of one experiment (n = 6 RTOCs for H-2^d/d i.c.GFP and H-2^d/d secGFP groups, n = 7 for the other groups). d. Mixed BM chimeras were established as described in Fig. 7b. Expression of H-2K^b and H2-K^d on thymic DCs (gated as CD11c⁺MHC II⁺) was analyzed by flow cytometry. e. Thymic iNKT cells from H2^b/d (CB6F1 background) or H2^b/b (C57BL/6J background) mice were transduced with Bcl2-P2A-Thy1.1-IRES-secGFP-encoding retroviruses and injected into the thymi of Jedi-TCRβ mice (on CB6F1 background). Expression of CD8α, CD8β and PD1 on CD4^– H2-K^d GFP_200-208 tetramer-binding thymocytes is shown 7 days after transfer. f. Gating strategy showing utilization of low levels of PD1 as a surrogate marker for TCR expression on DP thymocytes from Jedi-TCRαβ mice. g. Normalized Ca²⁺ signaling intensity in Jedi thymocytes interacting with iNKT cells as in Fig. 6f, but values for individual cells are shown. Data are presented as the mean ± s.d. with NS: non-significant (P > 0.05) and **P < 0.01. Data were analyzed by one-way ANOVA with Holm-Sidak’s multiple comparisons test.

Source data

Extended Data Fig. 7 Decreased thymic abundance of IL17A-GFP results in autoimmune elimination of GFP⁺ T_H17 cells.

a. Comparison of the frequencies of GFP⁺ lymphocytes in the thymi, spleens and mLNs of IL17A-GFP mice and in IL17A-GFP→WT BM chimeras (analyzed 8 weeks after reconstitution). One experiment with 2 IL17A-GFP mice and 3 BM chimeras. b–e. WT recipient mice were irradiated and transferred i.v. with T- and NK-cell depleted BM cells from either IL17A-GFP mice, or Jedi-TCRβ mice and WT mice, or Jedi-TCRβ mice and IL17A-GFP mice (all on CB6F1 background). Mesenteric lymph nodes (mLN) (b, d) and thymi (c, e) were analyzed 8 weeks after reconstitution. Same experiment as in Fig. 7a–c. b, c. Frequency and cell surface phenotype of H2-K^d GFP_200-208 tetramer-binding CD8 T cells. d, e. Frequency of IL17A-GFP expressing cells among total lymphocytes and among Vβ4^– (non-Jedi) CD4 T cells. Representative results of two independent experiments (n = 3 chimeras per group). f. T_H17 cells were differentiated in vitro from WT and IL17A-GFP CD4 T cells as in Fig. 7d. Intracellular expression of IL17A cytokine and GFP was analyzed after a short-term PMA/Ionomycin restimulation. Data are presented as the mean ± s.d. with *P < 0.05 and **P < 0.01. Data were analyzed by two-tailed Student’s t test.

Source data

Extended Data Fig. 8 IFNγ as an endogenous T/NK/ILC-derived inflammation-associated self-antigen.

a. Expression of Thy1, NK1.1 and MHC II by IFNγ⁺ and total WT thymocytes after PMA/Ionomycin stimulation. b. WT and Ifng^–/– mice (on C57BL/6J background) were immunized with IFNγ_69-78 peptide using anti-CD40 and Poly(I:C) as adjuvant. 6 days after immunization, splenocytes were stimulated with IFNγ_69-78 peptide for 6 hours as described in Methods and production of TNF by CD8 T cells was analyzed by flow cytometry. c. WT and Ifng^–/– mice (on C57BL/6J background) were lethally irradiated and reconstituted with syngeneic WT or Ifng^–/– BM cells depleted of T- and NK-cells. 11 weeks after reconstitution, the resulting four groups of BM chimeras were immunized with IFNγ_69-78 and OVA_257-264 peptides using anti-CD40 and Poly(I:C) as adjuvant. 6 days after immunization, splenocytes were stimulated with OVA_257-264 peptide for 6 hours as described in Methods and production of TNF by CD8 T cells was analyzed by flow cytometry (n = 3 for WT→WT, n = 4 for the other groups). Representative results of two independent experiments. Data are presented as the mean ± s.d. with NS: non-significant (P > 0.05) and *P < 0.05. Data were analyzed by two-tailed Student’s t test (b) or one-way ANOVA with Holm-Sidak’s multiple comparisons test (c).

Source data

Supplementary information

Supplementary Information

Supplementary Figs. 1–4 and Notes.

Reporting Summary

Peer Review File

Supplementary Video 1

Example of contacts between a CTV- and Cal-520 AM Ca²⁺ sensor dye-labeled Jedi-TCRαβ DP cell (green) and a CTY-labeled IL4–GFP iNKT cell (purple) in an RTOC experiment. The contacts resulted in the induction of Ca²⁺ signaling (yellow) in the Jedi-TCRαβ DP cell. Time (hh:mm:ss) is shown in the bottom right corner.

Supplementary Video 2

Example of a contact between a CTV- and Cal-520 AM Ca²⁺ sensor dye-labeled Jedi-TCRαβ DP cell (green) and a CTY-labeled WT iNKT cell (purple) in an RTOC experiment. Time (hh:mm:ss) is shown in the bottom right corner.

Supplementary Video 3

Example of a long contact between a CTV- and Cal-520 AM Ca²⁺ sensor dye-labeled Jedi-TCRαβ DP cell (green) and a CTY-labeled IL4–GFP iNKT cell (purple) in an RTOC experiment. Induction of Ca²⁺ signaling (yellow) in the Jedi-TCRαβ DP cell and chasing behavior of a Jedi DP cell toward an iNKT cell could be observed. Time (hh:mm:ss) is shown in the bottom right corner.

Supplementary Video 4

Example of contacts between a CTV- and Cal-520 AM Ca²⁺ sensor dye-labeled Jedi-TCRαβ DP cell (green) and a CTY-labeled IL4–GFP iNKT cell (purple) in an RTOC experiment. The contacts resulted in induction of Ca²⁺ signaling (yellow) and possible apoptosis in the Jedi-TCRαβ DP cell. Time (hh:mm:ss) is shown in the top right corner.

Supplementary Data 1

Statistical source data for Supplementary Figs. 1–4.

Source data

Source Data Figs. 1–8

Statistical source data.

Source Data Extended Data Figs. 2–8.

Statistical source data.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

You, Y., Dunst, J., Ye, K. et al. Direct presentation of inflammation-associated self-antigens by thymic innate-like T cells induces elimination of autoreactive CD8⁺ thymocytes. Nat Immunol 25, 1367–1382 (2024). https://doi.org/10.1038/s41590-024-01899-6

Download citation

Received: 19 June 2023
Accepted: 17 June 2024
Published: 11 July 2024
Issue date: August 2024
DOI: https://doi.org/10.1038/s41590-024-01899-6

This article is cited by

Developmental features and unique characteristics of peptide-specific PLZF+ innate-like T cells in mice
- Ahmed Hassan
- Nico Heise
- Hristo Georgiev
Nature Communications (2025)
Emerging Mechanisms and Biomarkers Associated with T-Cells and B-Cells in Autoimmune Disorders
- Azhagu Madhavan Sivalingam
Clinical Reviews in Allergy & Immunology (2025)

Subjects

Abstract

Similar content being viewed by others

Main

Results

Innate-like T cells as libraries of T cell effector antigens

CD8+ T cell tolerance to a model inflammation self-antigen

Central tolerance to a model inflammation self-antigen

iNKT cells and eosinophils can mediate negative selection

Effects of co-stimulation, antigen localization and affinity

Antigen expression by iNKT cells induces tolerance in vivo

T cell-to-T cell antigen presentation mediates negative selection

Lack of tolerance to IL-17A–GFP results in a fratricide reaction

Induction of tolerance to an endogenous effector molecule

Discussion

Methods

Mice

Flow cytometry

Flow cytometry staining reagents

Generation of BM chimeras

In vitro TH2 cell and TH17 cell differentiation

In vivo killing assays

Construction of retroviral vectors

Transfection and retrovirus production

Immunization experiments

Cytokine production assays

Generation of the Jedi TCR-expressing reporter cell line

In vitro cocultures with Jedi TCR-expressing reporter cells

Retroviral infection of thymic iNKT cells

Immunofluorescence microscopy of thymus sections

RTOCs

RTOC live-cell imaging

RTOC live-cell imaging analysis

Intrathymic injections

Autoantibody profiling using bead-based protein microarrays

Analysis of public expression datasets

Statistical analysis

Reporting summary

Data availability

References

Acknowledgements

Funding

Author information

Authors and Affiliations

Contributions

Corresponding author

Ethics declarations

Competing interests

Peer review

Peer review information

Additional information

Extended data

Supplementary information

Source data

Rights and permissions

About this article

Cite this article

Share this article

This article is cited by

Search

Quick links

CD8⁺ T cell tolerance to a model inflammation self-antigen

In vitro T_H2 cell and T_H17 cell differentiation