Classical MHC expression by DP thymocytes impairs the selection of non-classical MHC restricted inna

Conventional T cells are selected by peptide-MHC expressed by cortical epithelial cells in the thymus, and not by cortical thymocytes themselves that do not express MHC I or MHC II. Instead, cortical thymocytes express non-peptide presenting MHC molecules like CD1d and MR1, and promote the selection of PLZF+ iNKT and MAIT cells, respectively. Here, we report an inducible class-I transactivator mouse that enables the expression of peptide presenting MHC I molecules in different cell types. We show that MHC I expression in DP thymocytes leads to expansion of peptide specific PLZF+ innate-like (PIL) T cells. Akin to iNKT cells, PIL T cells differentiate into three functional effector subsets in the thymus, and are dependent on SAP signaling. We demonstrate that PIL and NKT cells compete for a narrow niche, suggesting that the absence of peptide-MHC on DP thymocytes facilitates selection of non-peptide specific lymphocytes.

The study of innate-like T cells, namely invariant natural killer T (iNKT) cells, mucosal-associated invariant T (MAIT) cells, and γδ T cells, is rapidly expanding and gaining interest. Defined by their phenotype, innate-like T cells bear functional T-cell receptor (TCR) rearrangements, yet acquire a memory phenotype during development, and quickly respond to stimulation without undergoing extensive clonal expansion1,2. They are generally thought to recognize conserved antigens3. For these reasons, they are considered as a bridge between adaptive and innate immunity, and are seen as an appealing target for immune therapies. As a prominent member of this group, iNKT cells were initially described as cells with such unconventional phenotype4,5,6,7. Consequently, it was shown that iNKT cells share the same developmental path with conventional αβ T cells until the double-positive (DP) thymocyte stage8. Yet, they follow a different process of thymic selection. Conventional αβ T cells undergo positive and negative selection by thymic epithelial cells (TECs) presenting peptide antigens by classical class I and II major histocompatibility complex (MHC) molecules. In contrast iNKT cells are positively selected on hematopoietic cells (HCs), in particular, another DP thymocyte presenting lipid antigens via the non-classical MHC-like molecule CD1d3,9,10,11. In addition, positive selection of iNKT cells requires a strong TCR signaling, a process known as “agonist selection,” and a secondary co-stimulatory signal provided by homotypic interactions between signaling lymphocyte activation molecule family (SLAMF) receptors12,13. As a result, selected cells upregulate the key transcription factor PLZF, which governs the commitment to the iNKT cell lineage14,15. Subsequently, more innate-like T-cell subsets were described, including PLZF+ γδ T cells and MAIT cells. Similar to iNKT cells, both cell subsets express PLZF and the latter were shown to be positively selected by DP thymocytes16,17. In addition, it was reported that non-classical MHC Ib-restricted cells can be selected by TECs or by HCs. However, they develop an innate-like phenotype only when they are selected on HCs but not on TECs18. Thus, it appears that a common feature of innate-like T cells is being positively selected by thymocytes. Thus, DP–DP interactions are proven to be pivotal for induction of PLZF expression and initiation of an innate-like program in T cells. Yet, the underlining mechanisms governing selection and development of these cells are still elusive.

Strikingly, murine DP thymocytes do not express classical MHC I or MHC II molecules19,20. It is possible that MHC downregulation at this stage could be a mechanism preventing high levels of diversion to a PLZF-expressing innate-like T-cell lineage, which could compromise conventional CD4 and CD8 T-cell repertoire development. Alternatively, it might facilitate greater development and selection of non-classical MHC I-restricted T cells by reducing competition. Two independent groups generated transgenic mice where MHC II expression was forced at the DP thymocyte stage21,22. Both the mouse lines were generated by expressing the human class II MHC transactivator (CIITA) under the control of CD4 or proximal Lck promotor regions. Consequently, in both transgenic mice, MHC II was expressed on DP thymocytes. This lead to an expansion of thymocyte-selected PLZF+ T cells with an innate-like phenotype similar to iNKT cells23,24.

Yet, innate-like T cells, such as iNKT and MAIT cells, are naturally selected on thymocytes expressing non-classical members of the MHC I family. Thus, we wanted to investigate the effects of forced expression of classical MHC I. Recently, the Class I transactivator (CITA, also called Nlrc5) was reported to be a positive transcriptional regulator of MHC I expression25,26,27. In addition, Nlrc5 drives transcription of components necessary for MHC I processing and peptide loading such as β2m, Tap1, and Lmp225,28. Here we show that Nlrc5 is not expressed in DP thymocytes, despite their high expression of CD1d. We describe an inducible transgenic mouse where murine Nlrc5 can be specifically expressed in different cell types. Remarkably, forced Nlrc5 expression in DP thymocytes led to a dramatic upregulation of peptide-presenting MHC I molecules on the cell surface, therefore making them potentially capable of selecting and steering peptide-specific MHC I-restricted T cells into an innate-like T-cell developmental pathway. Indeed, similar to MHC II, MHC I expression on DP thymocytes provoked expansion of PLZF+ innate-like (PIL) T cells. Moreover, both MHC I- and II-restricted PIL T cells displayed a phenotype resembling iNKT cells in that they were found to exist in three functional subsets based on the expression pattern of characteristic transcription factors, cytokines, and surface markers, and their development required the SLAM-associated adaptor protein (SAP). By analyzing bone marrow (BM) chimeras, PIL T-cell selection was attributed to MHC I expression on neighboring DP thymocytes and not to a cell-intrinsic effect. Although conventional CD4 and CD8 T-cell development was not compromised in these mice, the development of iNKT cells was reduced, indicating that PIL T cells and iNKT cells compete for a limited thymic niche. These results suggest that the absence of peptide-presenting MHC class I molecules on DP thymocytes is a consequence of not expressing Nlrc5 and functions to facilitate the development of non-peptide-specific lymphocyte subsets such as iNKT cells.

Generation of an Nlrc5-stopflox transgenic mouse as a tool for conditional expression of MHC class I molecules

It was previously shown that the expression of MHC II on DP thymocytes promotes the development of MHC II-restricted thymocyte-selected CD4 T cells (T-CD4) with an NKT-like phenotype21,22,24,29. Therefore, we hypothesized that MHC I expression on DP thymocytes could have a similar effect by selecting MHC I-restricted NKT-like T cells (Fig. 1a). Normally, murine thymocytes at the DP stage of their development do not express classical MHC I or MHC II molecules19. In search of a suitable approach to drive stable MHC I expression in DP, we selected the transcription factor Nlrc5 (also known as CITA) as a top candidate with such potential function. Nlrc5 is crucial for transcription not only of MHC I genes but also of components necessary for MHC I processing and peptide loading such as β2m, Tap1, and Lmp225,28. In addition, Nlrc5 expression is heavily repressed at the DP thymocyte stage (Fig. 1b data from ImmGen30). To test whether Nlrc5 could drive higher MHC I expression in vitro, we cloned the coding sequence (CDS) of the murine Nlrc5 gene into a lentiviral expression vector. Indeed, transduced HEK 293 cell samples showed that Nlrc5 was sufficient to drive higher MHC I expression in comparison to control samples (Fig. 1c).

To test the role of Nlrc5 in vivo, we generated an inducible transgenic knock-in mouse where Nlrc5 expression can be upregulated in a tissue-specific manner, in cells that express Cre recombinase. In brief, a construct with a loxP-flanked stop cassette in front of the murine Nlrc5 CDS was inserted into the Rosa26 locus (Fig. 1d). This design allows Nlrc5 expression once Cre recombinase is present in the cell. This mouse line was crossed to the CD4-Cre mouse initiating Cre expression at the DP stage. We observed that MHC I (H-2Db and H-2Kb) was highly expressed on DP thymocytes in these mice, compared to controls, which normally do not express it (Fig. 1e and Supplementary Fig. 1a). Single-positive (SP) thymocytes and splenocytes, which normally express high levels of MHC I, displayed only slightly higher levels in these mice. Of note, MHC II expression was not affected by the Nlrc5transgene (Fig. 1e).

The non-classical MHC Ib alleles that present peptides, Qa-1, Qa-2, and H2-M3, showed the same expression pattern on thymocytes as the classical MHC Ia alleles H2-Kb and H2-Db (i.e., low on DP thymocytes) (Supplementary Fig. 1d). Nlrc5 overexpression led to their upregulation as well (Supplementary Fig. 1b). In contrast, gene expression of non-peptide presenting molecules CD1d and MR1 is high in DP thymocytes (Supplementary Fig. 1e) and was not altered by overexpression of Nlrc5 (Supplementary Fig. 1c). This suggests that peptide presentation through MHC Ia/Ib molecules, in particular, is absent in cortical DP thymocytes due to the absence of Nlrc5 expression, whereas lipid and metabolite presentation are preserved.

As an additional control, we obtained plck-CIITA transgenic mice21, which overexpress MHC II on T cells starting from the DP thymocyte stage (Fig. 1e). In the following sections, both strains were analyzed in parallel. For simplicity, we refer to CD4-Cre/Nlrc5-stopflox as “T-MHC I” mouse and pLck-CIITA as “T-MHC II” mouse.

MHC I expression on DP thymocytes resulted in an increase in PIL T cells

The thymus architecture and the general subset distribution in the thymus showed no striking alterations between wild type (WT) and T-MHC I littermates (Fig. 2a–c and Supplementary Fig. 2). There were no significant differences in the frequency of DN, DP, CD4 SP T cells, γδ T cells, or γδ NKT cells (defined as PLZF+ γδ T cells). T-MHC II mice show an increased CD8 SP frequency due to the induction of “memory phenotype” CD8 T cells, as previously described31. In contrast, T-MHC I mice showed, if anything, a slight reduction in CD8 SP T-cell number and frequency, and a small increase in MAIT cell number (Fig. 2a–cand gating in Supplementary Fig. 3a–c). iNKT cells were approximately twofold reduced in number and frequency in both T-MHC I and T-MHC II mice. The reduction in iNKT cell number and frequency was even more prominent in the spleen and the same reduction trend was present in the liver as well (Supplementary Fig. 4a, b). There were no significant differences in the MAIT cell and γδ NKT cell number in the spleen (Supplementary Fig. 4d, e). Surprisingly, Treg cell frequency was not affected in T-MHC II mice but was increased in the thymus and spleen of T-MHC I mice. In the spleen, this difference was also statistically significant in cell number (Supplementary Fig. 4b lower panel). Overall, the T-MHC I transgenic mouse did not show major lymphocyte development alterations, except for a modest increase in Treg and a decrease in iNKT cell frequency and numbers.

Next, we sought to determine whether cells with NKT-like phenotype were expanded in the T-MHC I transgenic mouse similar to that seen with T-MHC II mice (Fig. 1a). PLZF is the major transcription factor endowing innate-like features to the iNKT cell lineage14. Therefore, we established a gating strategy that allowed identification of peptide-specific PIL T cells, by gating on TCRβ+PLZF+ cells and excluding other subsets known to express PLZF, such as iNKT cells, γδ T cells, and MAIT cells (Fig. 2d). With this strategy, we identified a small fraction of PIL T cells present in WT littermates, increasing by fourfold in T-MHC I mice (Fig. 2e, f). Nonetheless, this change was less robust than the 10–15-fold increase of PIL T cells in the T-MHC II.

PIL T-cell expansion is SAP dependent and not mediated by Nlrc5 in a cell-intrinsic way

Even though we did not observe upregulation of surface CD1d in response to Nlrc5 expression in DP thymocytes, it is formally possible that PIL T cells are CD1d dependent. Therefore, we crossed T-MHC I mice to Cd1d−/− mice. Deficiency of CD1d abrogated iNKT cell development, as expected, but if anything, increased PIL T-cell numbers in T-MHC I mice (Fig. 3a, b). Thus, most PIL T cells are likely MHC Ia/Ib restricted, as their numbers are dependent on MHC I expression in DP thymocytes, yet not dependent on CD1d.

Next, we investigated whether MHC I was required on neighboring DP thymocytes for PIL T-cell development, or if it is due to an unknown cell-intrinsic effect of Nlrc5 overexpression. To this end, we established a set of unequal BM chimeric mice, where WT recipients were lethally irradiated and transplanted with a mix of WT and T-MHC I BM at a ratio 1 : 10 from different congenic mice. With this setup, 90% of the neighboring thymocytes would express MHC I and is thus capable of inducing PIL T-cell selection in WT progenitors. A control group received a mix of WT and WT BM from different congenic mice. Indeed, 8 weeks post transplantation, a significant increase in PIL T cells was present among WT BM-derived cells when neighboring thymocytes were expressing MHC I, in comparison to the control group (Fig. 3c, d). A significant increase in PIL T-cell frequency was present also in the spleen. Therefore, this result disfavors the possibility that forced expression of Nlrc5 promotes PILs development in a cell-intrinsic way and confirms the hypothesis that PIL T cells expand because of MHC I upregulation on neighboring DP thymocytes. Notably, the frequency of Treg cells did not increase among the WT BM-derived cells but was only observed among the T-MHC I-derived cells (Supplementary Fig. 5a, b). This indicates that the observed expansion of Treg cells in the T-MHC I mouse is mediated by Nlrc5 expression in a cell-intrinsic way.

A crucial requirement for iNKT cell development is a triggering of SLAM-SAP signaling pathways during agonist selection13. Therefore, we sought to determine whether SAP deficiency has the same impact on PIL T cell as on iNKT cell development. Expectedly, lack of SAP completely abrogated iNKT cell development. It also abrogated the generation of PIL T cells in both the thymus (Fig. 3e, f) and spleen (Supplementary Fig. 5c, d). This indicates that PIL T cells and iNKT cells employ similar signaling pathways during their selection and development.

PIL T cells are found in the same effector subsets as iNKT cells

Thymic iNKT cells exist in three well-defined effector subsets (iNKT1, 2, and 17), analogous to peripherally activated helper T cells32. Using transcription factor staining, we found that PIL T cells also segregate into three similar subsets, which we name as PIL1 (PLZFloTbet+RORγt-), PIL2 (PLZFhiRORγt-), and PIL17 (PLZFintRORγt+) cells (Fig. 4a). The expression of NK1.1 on PIL1, CD4 on PIL2, and CD138 on PIL17 cells was also similar to that of iNKT cell subsets (Supplementary Fig. 6a). Hence, this is suggestive that each PIL fraction might exhibit similar functional properties to the corresponding iNKT cell subset. Several independent studies have shown that the three NKT subsets have quite distinct gene expression programs33,34,35. Indeed, expression of a large panel of functionally relevant molecules, including CD122, CXCR3, CD69, PD1, CCR6, and CD25 was similar among PIL and iNKT subsets (Supplementary Fig. 6b), as was the production of the cytokines interferon-γ (IFNγ), interleukin (IL)-4, and IL-17A after in vitro stimulation (Supplementary Fig. 6c)32,33. Taken together, these data infer that analogous transcriptional and functional programs operate in PIL and iNKT cells.

Notably, both MHC I- and MHC II-restricted PIL T cells were differentiated into three subsets (Fig. 4a, b). However, MHC II-restricted PIL T cells in T-MHC II animals showed a strong skewing toward the PIL2 effector phenotype, at the expense of PIL17. This is consistent with the well-described overproduction of IL-4 by PLZF+ T-CD4s in pLck-CIITA mice31. However, the three subsets were present in similar proportions in WT and T-MHC I mice (Fig. 4a, b).

As mentioned above, PIL T cells might be phenotypically and functionally similar to iNKT cells. Yet, in contrast to iNKT cells, they are selected on classical MHC molecules. Therefore, we inspected their TCR co-receptor expression and TCR Vβ chain usage. As previously reported, PIL T cells developing in T-MHC II mice were found to be largely CD4 SP with a small fraction of DN cells (Fig. 4c). Surprisingly, PIL T cells in T-MHC I mice did not exclusively express CD8, but segregated into CD4 SP, CD8 SP, and DN, similar to those in WT mice, suggesting that PIL T cells in WT mice may be more closely related to MHC I-restricted cells (Fig. 4c). Although we did not observe a severe bias in TCR Vβ chain usage by PIL T cells, there was a noticeable increase in frequency of Vβ3+ and Vβ5.1/Vβ5.2+ cells within the PIL T-cell pool from WT and T-MHC I mice. In addition, a small increase in frequency of Vβ6+ cells was present within the PIL T cells from T-MHC II mice (Fig. 4d).

It was previously shown that thymic iNKT cell subsets display different levels of TCR on their surface, a trait that is more prominent on the BALB/c background33,36. TCR levels are highest on NKT2 cells, lowest on NKT1 cells, and intermediate on NKT17 cells. Moreover, these differences in TCR expression levels, and thus the presumed signaling strength, are reported to be pivotal for iNKT cell subset commitment and differentiation37,38,39. However, surface TCR did not differ significantly between PIL T-cell subsets (Fig. 4e), suggesting that factors other than TCR expression level may dictate subset commitment and differentiation in these cells.

Historical studies in the field of iNKT cell biology used CD44 and NK1.1 as markers to assess iNKT cell maturity and stage of development40. Using these markers, a higher proportion of PIL T cells were found to be CD44low compared to iNKT cells, indicating a less mature stage (Fig. 4f).

Memory phenotype CD8 T-cell development requires CD4 co-receptor engagement in PIL T cells

As previously reported, the frequency of CD8 SP T cells was increased in the thymus of T-MHC II mice (Fig. 2b)21,22,31. This was shown to be caused by IL-4 produced by PIL T cells (called T-CD4 cells in that report), which mediates development of memory phenotype CD8 T (TMP) cells through the induction of eomesodermin (Eomes)41. Thus, a large proportion of CD8 SP T cells in T-MHC II mice express Eomes (Fig. 5a) and have a memory phenotype31. To our surprise, there was no increase in Eomes+ CD8 TMP in T-MHC I mice (Fig. 5b). As interaction of the TCR with MHC II can invoke CD4 co-receptor signaling, we reasoned that the induction of IL-4 production by PIL T cells and, consequently, the increase in CD8 TMPcells might be the result of CD4 co-receptor signaling. To test this hypothesis, we crossed T-MHC I mice with CD8.4 transgenic mice. In these mice, the cytoplasmic tail of the endogenous Cd8a gene was substituted with the cytoplasmic tail of CD442. The CD8.4 transgene did not cause an increase in the total number of PIL T cells in T-MHC I mice (Fig. 5c), but it shifted the proportion of PIL T cells in favor of PLZFhi PIL2 cells (Fig. 5d), which produce IL-4. Consistent with this, the number of CD8 TMP was markedly increased in CD8.4 T-MHC I mice (Fig. 5e). A similar trend was seen in the spleen (Supplementary Fig. 7). Thus, the particular co-receptor involved in sensing MHC ligands on DP thymocytes influences effector subset differentiation of PIL T cells and has secondary effects of CD8 TMPdevelopment.

PIL T cells compete with iNKT cells for a cellular niche within the thymus

Similar to what have been previously reported with T-MHC II mouse43, we observed a two- to threefold reduction in the number of lipid-specific iNKT cells in mice with increased numbers of PIL T cells due to MHC I or MHC II expression on DP thymocytes (Fig. 2b). Of note, the pooled total cell count of PILs + iNKT cells was not significantly different between WT and T-MHC I mice (data not shown). As both peptide and lipid-specific T cells require SAP signaling from the surface SLAM family co-receptors to develop into PIL T cells or iNKT cells, respectively, this suggests that the two cell types may be in competition with each other for a cellular niche within the thymus. To further test this notion, we examined B6xBALB/c F1 mice, which have a slightly larger (twofold) iNKT cell niche32,44. If a similar niche regulates PIL T cells, then we might expect more PIL T cells in F1 mice. Indeed, both the portion and number of PIL T cells in T-MHC I and T-MHC II mice were higher in animals on the F1 background compared to the B6 background (Fig. 6a, b and Supplementary Fig. 8a, b) inferring an existence of a bigger niche for PIL T-cell development on the F1 background. We observed an inverse relationship between the number of iNKT cells and PIL T cells in both B6 and F1 mice (Fig. 6c and Supplementary Fig. 8c). Taken together, with the expansion of PIL T cells in mice that lack iNKT cells (Fig. 3b), these data strongly suggest that PIL T cells and iNKT cells compete for the same cellular niche. As an aside, the reported skewing of iNKT cells towards the PFZFhi NKT2 subset in F1 mice32 was also observed in PIL T cells (Fig. 6d and Supplementary Fig. 8d). This suggests that the same factors control iNKT and PIL T-cell effector subset skewing in different strains, which are more likely to be environmental cytokines, cell-intrinsic factors, or co-stimulatory molecules rather than specific self-antigens.

Here we describe an Nlrc5-transgenic mouse (T-MHC I) allowing cell subset-specific expression of Nlrc5, which is a central regulator of the expression and function of peptide presenting MHC I genes. Consequently, our data show that forced expression of Nlrc5 led to upregulation of MHC I on DP thymocytes. To the best of our knowledge, this is the first transgenic mouse model allowing cell subset-specific MHC I upregulation without altering MHC II regulation. Hence, this mouse line can serve as a powerful tool in future studies where stable MHC I expression is sought. Notably, Nlrc5 expression increased the surface level of some non-classical MHC Ib molecules such as Qa-1 and Qa-2 (peptide-presenting molecules) but did not affect CD1d and MR1 (non-peptide presenting molecules). Therefore, in our mouse model, we assume enhanced self-peptide MHC Ia/Ib presentation by DP thymocytes renders them capable of steering peptide-specific T cells towards an innate-like developmental pathway. Although we do not provide a direct proof for the nature of the antigens that PIL T cells recognize, previous studies showed that DP thymocytes could not select PIL T cells in a foreign antigen-specific TCR transgenic T-MHC II mouse model22. This implies that PIL T-cell selection is self-antigen specific and requires presentation by an opposing DP thymocyte. In addition, a subsequent study showed that PIL T cells from T-MHC II mice have a highly diverse TCR repertoire, implying they may recognize a variety of self-antigens45. Taken together, it is reasonable to consider that PIL T cells could be selected on endogenous peptide self-antigens presented by DP thymocytes, much similar to iNKT cells selected on an endogenous lipid self-antigen presented by DP thymocytes3,9,10,11.

Although DP thymocytes do not normally express classical MHC Ia molecules, a small fraction of PIL T cells were already present in WT mice. Yet, we show that most of these cells are not CD1d or MR1 restricted. Therefore, it remains unclear how these innate-like cells were selected. Interestingly, the small number of PIL T cells present in WT mice more closely resemble PIL T cells from T-MHC I mice than those from T-MHC II mice. This implies that PIL T cells in WT mice might be restricted to MHC Ib molecules, which are found to be still expressed, although at a low level, on DP thymocyte (i.e., H2-M3 and Qa-1). Indeed, it was previously shown that peptide-specific MHC Ib-restricted T cells could be selected on TECs or HCs46. However, those cells acquired an innate-like phenotype only when they are selected on HCs18,46. In addition, this process was shown to be SAP dependent47. Therefore, it is plausible that PIL T cells in WT mice are MHC Ib restricted.

Although MHC I expression on DP thymocytes led to an expansion of PIL T cells, the magnitude of this expansion was lower than that in the T-MHC II mouse. Further, the increased PIL T-cell frequency in the T-MHC I mouse did not result in a notable expansion of CD8 TMP cells, implying functional difference between PILs selected in T-MHC I and T-MHC II mice. Of note, mature murine iNKT cells do not express CD8αβ. In fact, several studies have previously shown that forced expression of CD8αβ is detrimental for iNKT cell development and leads to profound reduction in iNKT cell numbers7,48. Subsequently, another study showed that the silencing of CD8 expression in developing iNKT cells is mediated by the transcription factor Th-POK, whose expression is required for functional maturity of iNKT cells49. Therefore, the co-receptor choice seems to govern the functional phenotype of PIL cells as well. As a result, despite being selected on MHC I molecules, PIL cells developing in T-MHC I mice might be downregulating CD8 expression post selection due to initiation of the innate-like developmental program and expression of Th-POK. In addition, previous reports suggest that TCR signal strength seems to regulate iNKT cell subset commitment and differentiation37,38,50. Stronger TCR signal intensities favor iNKT2 cell development and a lower strength preferentially guides into the iNKT1 cell pathway. Moreover, CD4 co-receptor signaling intensity is higher in comparison to CD851,52,53 and presumably favors PIL2 cell expansion in T-MHC II mice. Therefore, increasing co-receptor signaling intensity by crossing T-MHC I to the CD8.4 mouse42 may have increased the frequency of PIL2 cells and expansion of CD8 TMP cells.

There are several studies suggesting a possible involvement of Nlrc5 expression in pro-inflammatory and type I IFN responses54,55. Although we did not observe any major alterations of conventional T-cell development, forced Nlrc5 expression in our mouse model caused an increase in Treg cell number and frequency. Moreover, here we show that this expansion was mediated in a cell-intrinsic manner by Nlrc5 and it was not because of MHC I expression on neighboring DP thymocytes. Interestingly, Nlrc5 is debatably described as a positive or negative modulator of nuclear factor-κB (NF-κB) signaling56,57,58. NF-κb signaling is also known to be crucial for Treg cell development and function59,60. Taking this into consideration, it is possible that Nlrc5 expression at the DP stage might influence Treg cell selection and development by modulating NF-κB signaling. Further studies will be needed to identify the precise mechanism behind Nlrc5 involvement in Treg cell biology.

Lastly, we provide several lines of evidence supporting the existance of a narrow niche for innate-like T-cell development. First, iNKT cell number and frequency were reduced in both T-MHC I and T-MHC II mice. Second, this inverse correlation between iNKT cell and PIL T-cell numbers was even more profound on a B6xBALB/c F1 background where the niche for iNKT cell development is larger. In addition, CD1d-deficient T-MHC I mice showed an increase in PIL T cells. Innate-like T cells might compete between each other for abundance of cytokines, activation ligands, or could directly inhibit each other’s development and survival via unknown mechanisms. Of note, the local abundance of IL-7, IL-15, and IL-2561,62,63, in the thymus, are crucial factors implicated in iNKT cell development, survival, and terminal subset maturation. Thus, higher PIL T-cell numbers might alter the bioavailability of these cytokines in the thymus. Yet, MAIT cells and γδ NKT cells did not display a reduction trend in the T-MHC I mouse. Therefore, this is suggestive that PIL T cells and iNKT cells probably compete for the same activation ligands, presumably members of the SLAMF receptors.

MHC class I molecules are some of the most widely expressed genes in the body. In contrast to MHC class II molecules, which are expressed primarily by professional antigen presenting cells, MHC class I is generally thought to be expressed by all nucleated cells. Based on this, the absence of MHC I in DP thymocytes is remarkable. The fact that we observed a reduction of lipid-specific iNKT cells when peptide-presenting MHC alleles were forced to be expressed in DP thymocytes suggests an evolutionary rationale for the absence of MHC I in DP thymocytes—to facilitate the development of innate-like T cells that recognize non-peptide ligands. Indeed, iNKT cells do recognize non-peptide ligands such as lipids64. Given that these cells also are semi-invariant means that their rearrangements occur relatively rarely, despite the great numeric abundance of these cells in the body10. Thus, the thymus appears to have a mechanism whereby the development of innate-like T cells from rare TCR gene rearrangements is favored.