Development and developmental potential of Aire-expressing mTECs

The promiscuous expression of tissue-restricted antigens in the thymic medulla, which was discovered from the thymic expression of pancreas-specific insulin and acute-phase liver-specific C-reactive protein (Jolicoeur et al., 1994; Smith et al., 1997; Klein et al., 1998; Klein and Kyewski, 2000), is mediated at least in part by the nuclear protein AIRE expressed in the mTEC^high subpopulation (Derbinski et al., 2001; Anderson et al., 2002). Aire-expressing mTEC^high are mosaics in terms of promiscuous gene expression. The estimated frequency of promiscuously expressed self-antigen genes in Aire-expressing mTEC^high is between 2 and 15% depending on the genes (Derbinski et al., 2008; Sansom et al., 2014). Reaggregate thymus organ culture experiments showed that mTEC^high, including Aire-expressing mTECs, are derived from embryonic and postnatal mTEC^low, indicating that mTEC^low include cells with a developmental potential to give rise to mTEC^high (Gray et al., 2006, 2007; Rossi et al., 2007a; Gäbler et al., 2007). Aire-independent self-antigen expression is also detected primarily in mTEC^high (Derbinski et al., 2005; Sansom et al., 2014) and is in part regulated by the transcription factor Fezf2 (Takaba et al., 2015).

The lack of proliferative potential in Aire-expressing mTECs in the postnatal thymus and the apoptosis of an mTEC cell line by AIRE overexpression suggested Aire-expressing mTECs represent terminally differentiated mTECs (Gray et al., 2007). However, the postnatal increase in mTEC^low suggested that mTEC^low are not solely the progenitors of mTEC^high but include an additional population that accumulates postnatally (Gray et al., 2006). By lineage tracing of Aire-expressing cells, it was revealed that Aire-expressing mTECs are capable of further differentiation into Aire-negative mTEC^low (Nishikawa et al., 2010; Metzger et al., 2013), which includes recently described mimetic mTECs.

Heterogeneity and development of mimetic mTECs

Morphological diversity in mTECs was noted as early as the 19th century by the discovery of Hassall’s corpuscles with concentric whorls of stratified epithelial cells (Hassall, 1849). Findings of ciliated columnar epithelial cells and neurosecretory epithelial cells in the medulla documented further heterogeneity in mTEC morphology (Farr and Rudensky, 1998). Interestingly, these highly diverse and terminally differentiated epithelial cells localized in the medulla are at least in part detected in post-Aire mTECs (White et al., 2010; Metzger et al., 2013). More recent studies reported an mTEC subpopulation resembling intestinal tuft cells in morphology and molecular expression profiles, including expression of type 2 taste receptors and the type 2 cytokine IL-25 (Miller et al., 2018; Bornstein et al., 2018). Thymic tuft cells are partially derived from Aire-expressing mTECs, even though Aire is not necessary for thymic tuft cell development and not all thymic tuft cells derive from Aire-expressing mTECs (Miller et al., 2018). Like intestinal tuft cells, development of thymic tuft cells is dependent on the transcription factor Pou2f3 (Miller et al., 2018; Bornstein et al., 2018). Recent studies further noted the role of Fezf2 in the development of thymic tuft cells (Lammers et al., 2024; Ushio et al., 2024).

Michelson and colleagues reconfirmed the diversity in the post-Aire mTEC^low population by chromatin accessibility assay and RNA-sequencing analysis of individual mTECs (Michelson et al., 2022). Each post-Aire cluster is enriched with the binding motif of transcription factors known to be essential for extrathymic tissues, reflecting the chromatin accessibility of genes specific to each extrathymic tissue driven by that transcription factor. These post-Aire mTECs mimic extrathymic cells and are suggested to be involved in T-cell tolerance to mimetic cell antigens, thereby termed mimetic mTECs (Michelson et al., 2022) (Fig. 1 A).

Mimetic mTECs may extend their functions beyond self-tolerance establishment in conventional T cells. Thymic tuft cells exhibit a role in regulating the development and function of invariant NKT2 cells in the thymus (Miller et al., 2018; Lucas et al., 2020). Another study reported that endocrine mimetic mTECs control the cellularity of the thymus in a ghrelin-dependent manner, and microfold mimetic mTECs regulate the generation of IgA⁺ plasma cells in the thymus (Givony et al., 2023).

Development and developmental potential of CCL21-expressing mTECs

The chemokine CCL21 produced by mTECs is critical for the establishment of T-cell tolerance through the chemoattraction of positively selected thymocytes from the cortex to the medulla (Kurobe et al., 2006; Kozai et al., 2017). CCL21 protein produced in the thymic medulla is also captured by mesenchymal stroma and contributes to neonatal T-cell emigration (James et al., 2021). Positive selection-inducing TCR signals in cortical thymocytes elevate the expression of CCR7, a receptor for CCL21, and positively selected cortical thymocytes are attracted to the medullary region through CCL21-CCR7–mediated chemotactic signals (Ueno et al., 2004). Two molecular species, CCL21Ser and CCL21Leu, with one amino acid difference are encoded in the mouse genome (Nakano and Gunn, 2001; Chen et al., 2002), and CCL21Ser encoded by Ccl21a locus is predominantly expressed in the thymic medulla (Kozai et al., 2017). In mice lacking either Ccr7 or Ccl21a, positively selected mature thymocytes fail to accumulate in the medulla, and T cells fail to establish self-tolerance (Kurobe et al., 2006; Kozai et al., 2017). The additional CCR7-ligand CCL19 has no appreciable role in the cortex-to-medulla migration of developing thymocytes (Link et al., 2007; Kozai et al., 2017). Thus, CCL21-expressing thymocyte-attracting mTECs represent a functional mTEC subset essential for the establishment of self-tolerance in T cells.

CCL21⁺ mTECs are included in the mTEC^low subpopulation and distinct from AIRE⁺ mTECs (Lkhagvasuren et al., 2013; Kozai et al., 2017). Fate-mapping analysis indicated that Ccl21a⁺ mTECs have a developmental potential to generate most mTECs including AIRE⁺ mTECs and thymic tuft cells, and reaggregate thymus organ culture experiments confirmed that embryonic CCL21⁺ mTECs, which are functionally potent to accumulate medullary thymocytes in the embryonic thymus, can give rise to AIRE⁺ mTECs (Ohigashi et al., 2024). Thus, thymocyte-attracting CCL21⁺ mTECs in embryonic thymus include progenitor activity to become self-antigen-displaying mTECs, including AIRE⁺ mTECs. These findings also indicate the functional conversion of thymocyte-attracting mTECs into self-antigen-displaying mTECs contributes to the development of heterogenous mTEC subpopulations.

Fate-mapping analysis further showed approximately two-thirds (66%) of cTECs are derived from Ccl21a⁺ cells, suggesting Ccl21a-transcribing mTECs, which are detectable only in the thymic medulla, include the developmental potential equivalent to bipotent TEC progenitors. Indeed, cTECs derived from Ccl21a⁺ cells are enriched in the perimedullary region of the thymic cortex (Ohigashi et al., 2024). Interestingly, postnatally appearing mTEC-biased progenitors included cells transcribing Ccl21a (Nusser et al., 2022). These results suggest the similarity and potential overlap between the Ccl21a⁺ fraction of cTEC progenitors and postnatal mTEC-biased bipotent progenitors.

In contrast to embryonic CCL21⁺ mTECs, CCL21⁺ mTECs isolated from postnatal thymus failed to show the developmental potential to give rise to AIRE⁺ mTECs, and the gene expression profiles were markedly different between embryonic and postnatal CCL21⁺ mTECs (Ohigashi et al., 2024). Postnatal CCL21⁺ mTECs may include mTEC-biased bipotent progenitors, but the majority of postnatal CCL21⁺ mTECs represent terminally differentiated mTEC^low that lack progenitor potential (Fig. 1 A).

The IL22-IL23 axis

Direct insight into the molecular regulation of thymus repair came from studies of Dudakov and van den Brink (Dudakov et al., 2012), who identified the importance of IL22, a monomeric cytokine belonging to the IL10 family. Following SLI treatment of Il22^−/− mice or transplant of WT bone marrow into lethally irradiated Il22^−/− hosts, they reported diminished cTECs and mTECs as well as reduced CD4⁺CD8⁺ thymocytes, the latter an indicative measure of thymopoiesis (Dudakov et al., 2012). In addition, recombinant IL22 administration enhanced thymus regeneration following damage in WT mice (Dudakov et al., 2012), while increased IL22 levels were observed after damage (Dudakov et al., 2012; Pan et al., 2014). Thus, thymus regeneration is IL22 dependent, and damage provokes increased availability of this key cytokine, the latter also being reported in the context of HSCT in man (Shang et al., 2021). Subsequent studies also demonstrated the importance of IL22 for thymus function in models of Graft versus Host Disease (Dudakov et al., 2017; Pan et al., 2019a). Mechanistically, events upstream of the IL22 requirement mapped to loss of CD4⁺CD8⁺ thymocytes, stimulating IL23 production from DCs, which then triggered IL22 production from Rorγ(t)-dependent ILC3 (Dudakov et al., 2012). Interestingly, in the context of allogeneic HSCT, donor T cells were also reported as an IL22 source (Pan et al., 2019b). Analysis of downstream events indicated IL22 acted directly on thymic stroma by augmenting Foxn1 expression (Pan et al., 2014), regulating the JAK/STAT/Mcl-1 pathway (Pan et al., 2019a) and promoting TEC survival and expansion (Dudakov et al., 2012). Such observations are important as identification of ILC3 as an important controller of thymus recovery from injury highlights the importance of innate immune components in restoring functionality in an organ important in adaptive immunity. Finally, they also fit well with the ability of IL22 to promote regeneration in multiple organs (Lindemans et al., 2015; McGee et al., 2013; Ren et al., 2010), pointing toward common mechanisms of tissue repair in lymphoid and non-lymphoid tissues.

Importance of the TNF receptor superfamily (TNFRSF)

Multiple TNFRSF members regulate steady-state development and function of immune organs. Three receptors in particular, LTβR, RANK, and CD40, regulate TEC development, most notably the mTEC lineage. For example, RANK expression is confined to the mTEC lineage (Hikosaka et al., 2008; McCarthy et al., 2015) and is an essential regulator of AIRE⁺ mTEC development in the embryo (Rossi et al., 2007a), a role it shares with CD40 during postnatal stages (Akiyama et al., 2008; Irla et al., 2008). LTβR, which is expressed by both cTECs and mTECs (Cosway et al., 2017; Hikosaka et al., 2008), influences mTEC cellularity and homeostasis (Boehm et al., 2003; Cosway et al., 2017), including CCL21⁺ mTEC^low (Lkhagvasuren et al., 2013) and FEZF2⁺ mTEC (Takaba et al., 2015). Given their importance in homeostatic control of TEC, several studies have investigated roles of LTβR and RANK in thymus regeneration.

Following thymus injury, RANKL increases in expression (Dudakov et al., 2012; Li et al., 2020b; Lopes et al., 2017), including on lymphoid tissue inducer (LTi)/ILC3 cells after lethal total body irradiation (Dudakov et al., 2012; Lopes et al., 2017) and also on both LTi/ILC3 and CD4⁺ thymocytes after SLI (Lopes et al., 2017). Interestingly, augmentation of RANKL expression is transient, with surface RANKL on LTi/ILC3 and CD4⁺ thymocytes returning to baseline by 20 days after damage (Lopes et al., 2017). This suggests a feedback loop occurs during progression toward thymus recovery that negatively regulates RANKL availability. Such a mechanism may be important, as prolonged RANKL availability disrupts TEC microenvironments by skewing toward the mTEC lineage (Hikosaka et al., 2008; Yin et al., 2017). Importantly, provision of RANKL, which expands mTEC in steady-state thymus (Ohigashi et al., 2011), was shown to augment thymus regeneration, while RANKL blockade impaired regeneration (Lopes et al., 2017). Furthermore, administration of ruxolitinib, a Janus kinase (Jak) inhibitor that blocks the Jak-STAT pathway implicated in TEC survival (Lomada et al., 2016), decreased RANK expression and impaired thymus regeneration following SLI (Li et al., 2020b). Collectively, these findings provide strong evidence for a RANK-RANKL axis in thymus repair that involves provision of RANKL by both innate (LTi/ILC3) and adaptive (CD4⁺) immune cells. Interestingly, additional hemopoietic cells, notably γδT cells (Roberts et al., 2012), CD1d-restricted NKT cells (White et al., 2014), and progenitor T cells (Singh et al., 2020), have also been shown to express RANKL, with the latter shown to be effective in regeneration of the aged thymus. It will be interesting to address whether these additional sources of RANKL are important in the restoration of thymus function following acute injury.

Of particular interest are the effects of RANKL administration on thymus regeneration. Here, increases in both cTECs and mTECs were noted in the context of both SLI and HSCT (Lopes et al., 2017). As RANK is expressed by mTECs but not cTECs (Hikosaka et al., 2008) and RANK⁺ TEC progenitors are committed to the mTEC lineage (Akiyama et al., 2016; Baik et al., 2016), this raises the question of how can RANK stimulation aid in cTEC regeneration? One possibility is that in addition to direct effects on the RANK⁺ mTEC lineage, RANKL may trigger the regeneration of RANK⁻ cTEC by stimulating RANK on non-TEC cells. Several lines of evidence are consistent with this. First, RANK is expressed by thymic ILC3/LTi, where it is upregulated following thymus damage (Lopes et al., 2017). Second, RANK stimulation of LTi/ILC3 upregulates their expression of LTα, which is important for the positive effects of RANKL on thymus regeneration (Lopes et al., 2017). Third, LTβR expression by cTECs increases following thymus damage (Lopes et al., 2017). Whether LTα-LTβR interactions between LTi/ILC3 and cTEC and/or cTEC progenitors explain the ability of RANKL to stimulate cTEC recovery requires further examination. Finally, there is further evidence for the importance of LTβR in the recovery of thymus function following damage, with diminished progenitor recruitment to the thymus of Ltbr^−/− mice after SLI (Shi et al., 2016), and LTβR stimulation using agonistic antibodies enhancing thymopoiesis after HSCT (Lucas et al., 2016). Whether these effects map to stimulation of LTβR on TECs or on other non-TEC stromal cells such as mesenchyme and endothelium that also express LTβR is not fully clear. Collectively, these observations indicate the importance of LTβR and RANK in the recovery of thymus function following damage and suggest their requirement involves mechanisms involving some interplay between the two receptors. That both LTβR and RANK are key TEC regulators in the steady state further suggests common mechanisms operate to control both thymus development and regeneration.

Type 2 immune mediators

In addition to thymus, several organs can regenerate and regain effective functionality following injury, and it is likely that study of these tissues has aided in elucidating mechanisms of thymus regeneration. Of particular relevance here are components of type 2 immunity, a form of immune response that normally occurs during allergic reactions and following helminth infection (Maizels and Gause, 2023; Molofsky and Locksley, 2023). Type 2 immunity is typically characterized by production of the cytokines IL4, IL5, IL9, and IL13, and involvement of multiple innate cells (e.g., ILC2, mast cells, basophils, and eosinophils). Regarding the influence of these immune mediators on tissue microenvironments, expression of the type 2 IL4 receptor plays an important role. Consisting of IL4Rα, IL13Rα1, and IL13Rα2 components, expression is detectable in epithelial, endothelial, and mesenchymal cells, where it operates as a receptor for both IL4 and IL3 (Gieseck et al., 2018; Junttila, 2018).

Several lines of evidence demonstrate the importance of type 2 immunity in tissue regeneration (Gieseck et al., 2018; Gurtner et al., 2023). For example, in eosinophil-deficient ΔdblGATA mice, where eosinophil development is abrogated as a result of deletion of a high-affinity GATA binding site in the GATA1 promotor (Yu et al., 2002), liver regeneration following either partial hepatectomy or carbon tetrachloride-induced injury is defective. Importantly, this requirement for eosinophils maps at least in part to their production of IL4 (Goh et al., 2013). Interestingly, IL4 targets fibro-adipocyte progenitors that express the type 2 IL4R, a process that inhibits their differentiation toward adipocytes and demonstrates the importance this stromal-expressed receptor in tissue homeostasis and regeneration (Heredia et al., 2013; Lee et al., 2015a). Similarly, eosinophils are recruited following nerve injury and help mediate axon regeneration through their production of IL4 (Liebendorfer et al., 2023; Pan et al., 2020, 2022). Relevant to these observations in other tissues, the adult thymus contains multiple hemopoietic components of type 2 immunity at steady state, including IL4/IL13-producing iNKT cells (Lee et al., 2015b; Miller et al., 2018), IL5-producing ILC2 (Jones et al., 2018; Nussbaum et al., 2013), eosinophils (Albinsson et al., 2023; Cosway et al., 2022; Gatti et al., 2023; Guerri et al., 2013), and basophils (Kim et al., 2010). Also present are stromal cells expressing the type 2 IL4R (White et al., 2017) and stromal subsets producing alarmins capable of triggering type 2 immunity, including IL25⁺ thymic tuft cells (Bornstein et al., 2018; Miller et al., 2018) and IL33 (Ferreira et al., 2021; Xu et al., 2022). This availability of multiple cellular and molecular components of type 2 immunity at steady state raises the possibility of their involvement in regulation of thymus following injury. In line with this, thymic macrophages, eosinophils, and neutrophils are increased following low-dose irradiation that promotes apoptosis induction in CD4⁺CD8⁺ thymocytes (Kim et al., 2010). Moreover, experiments in Csf1^op/op, ΔdblGATA, and Cxcr2^−/− mice indicated a role for thymic innate cells in the clearance of apoptotic cells caused by tissue injury (Kim et al., 2010). However, whether such events were then also connected to the recovery of TEC microenvironments following damage was not assessed.

Direct evidence for eosinophil involvement in this process was reported (Cosway et al., 2022), where thymus regeneration was severely impaired in eosinophil-deficient ΔdblGATA mice that received SLI. Here, defective recovery of both cTEC and mTEC lineages occurred, resulting in reduced thymopoiesis. Mechanistically, eosinophil involvement requires their recruitment to the injured thymus (Cosway et al., 2022), which is controlled by CCR3 and its ligand CCL11. This chemokine is produced by several thymic stromal subsets, and at least in the context of intrathymic expression following irradiation, is triggered in thymic stroma by IL4-producing CD1d-restricted NKT cells (Cosway et al., 2022). Such findings point toward the importance of stromal cell expression of the type 2 IL4R, which agrees with defective thymus regeneration also observed in Il4ra^−/− mice (Cosway et al., 2022). Collectively, they suggest a model in which early stages of thymus regeneration involve type 2 cytokine production by NKT cells, which triggers an increase in CCL11 production by type 2 IL4R⁺ thymic stroma, and results in the recruitment of eosinophils to the damaged thymus (Fig. 2 A). ILC2 were identified as important regulators of eosinophil-mediated thymus regeneration through their production of IL5, and events upstream of ILC2 involve the alarmin IL33, whose expression in thymus predominantly maps to mesenchymal cells (Cosway et al., 2023). Involvement of IL33 and ILC2 in thymus regeneration was subsequently confirmed in other studies (Nevo et al., 2024). The requirement for IL33 as a key regulator of ILC2 during recovery from SLI is selective within alarmins, as IL25 administration and IL25 deficiency had no effects on thymus regeneration (Cosway et al., 2023). This is consistent with normal endogenous thymus regeneration following SLI in Pou2f3^−/− mice that lack tuft cells, the sole intrathymic source of IL25 (Cosway et al., 2022; Nevo et al., 2024). Interestingly, thymic tuft cells were reported to aid in thymus regeneration following dexamethasone-induced thymocyte depletion (Nevo et al., 2024), suggesting differing recovery mechanisms following SLI and dexamethasone-induced thymus damage. On the mode of action of eosinophils, IL4 administration improved defective thymus regeneration in ΔdblGATA mice (Cosway et al., 2023), suggesting production of this type 2 cytokine explains at least in part the role of eosinophils in thymus regeneration. These findings extend our understanding of innate immune components in thymus injury and point toward a network in which IL33 production by thymic mesenchyme triggers the expansion of IL5-producing ILC2, which then act upon IL4-producing eosinophils that regulate recovery of thymic stromal microenvironments (Fig. 2 B).

Open in a separate window

Figure 2.

A type 2 cytokine axis controls thymus regeneration. (A and B) In models of thymus damage caused by sublethal irradiation, intrathymic iNKT cells are relatively radioresistant compared to their conventional thymocyte counterparts. This enables iNKT-cell secretion of type 2 cytokines to operate on thymic stroma, where it enhances production of the chemokine CCL11 by mTEC. This in turn causes a rapid surge in the recruitment of CCR3⁺ eosinophils from the periphery into the thymus (A). In subsequent stages of regeneration, the alarmin IL33 controls ILC2 that produces IL5, a cytokine important in eosinophil regulation. ILC2 feeds into the requirement for eosinophils in thymus regeneration, where eosinophil production of IL4 triggers recovery of the intrathymic microenvironments that are important for the restoration of T-cell development (B). Although IL4 production by eosinophils is important for thymus generation, whether this cytokine causes TEC generation directly, or indirectly through the targeting of non-TEC stroma such as mesenchyme (Mes), is not clear.