Proteasomes are responsible for generating peptides presented by class I MHC molecules of the immune system. β5t, a recently identified proteasome component, is specifically expressed in thymic cortical epithelial cells (cTECs) and plays a pivotal role in generating an immunocompetent repertoire of class I MHC-restricted CD8+ T cells. Here, we report that β5t is detectable in the thymus as early as E12.5 mouse embryos. We also found that β5t expression in cTECs was detectable in mice deficient for RelB or Rag2, indicating that β5t in cTECs is expressed in the absence of thymic medulla formation or thymocyte development beyond the CD4−CD8− stage. β5t expression in the embryonic thymus was not detectable in Foxn1-deficient nude mice, although its expression was not reduced in mice deficient for both CCR7 and CCR9, in which fetal thymus colonization by leukocytes is defective. These results indicate that β5t expression in cTECs is dependent on Foxn1 but independent of thymocyte crosstalk or thymic medulla formation.
Proteasomes are multicatalytic protease complexes that are responsible for the regulation of proteolysis in eukaryotic cells and for the generation of antigenic peptides presented by class I MHC molecules 1. The 20S proteasome is responsible for the proteolytic activity of the proteasome and is composed of 28 subunits (two α-rings with α1–α7 subunits and two β-rings with β1–β7 subunits). Among the subunits, β1, β2, and β5 are responsible for the proteolytic activity 2. Interferon-γ induces the production of a new set of catalytic subunits, β1i, β2i, and β5i, to replace their constitutive counterparts, β1, β2, and β5, thereby forming immunoproteasomes, which are proteasome complexes that possess altered proteolytic activity and participate in efficient antigen presentation and immune response 3. We have recently identified a novel subunit of the 20S proteasome, β5t, which is specifically expressed in thymic cortical epithelial cells (cTECs) and plays a pivotal role in the development of CD8+ T cells 4. In cTECs, β5t, instead of β5 or β5i, is incorporated into the 20S proteasome together with β1i and β2i, thereby forming unique proteasome complexes termed thymoproteasomes 4. We have reported evidence supporting the concept that cTECs display a thymoproteasome-specific spectrum of class I MHC-associated self-peptides that are required for the development of an immunocompetent repertoire of CD8+ T cells 5. Thus, β5t expressed by cTECs is indispensable for the development of the self-protective adaptive immune system 6.
The notion that β5t is specifically expressed in cTECs is based on the following findings: (i) the expression of β5t in various adult mouse organs was specifically detected in the thymus by RNA blot analysis and immunoblot analysis 4; (ii) β5t in adult mouse thymus was specifically detected in Ly51+ cells in the immunohistological analysis of thymus sections and the immunoblot analysis of isolated thymic stromal cell subpopulations 4, 5; and (iii) the expression of the fluorescence protein Venus in mice carrying the β5tVenus allele, in which the β5t coding sequence was substituted for cDNA encoding the fluorescence protein Venus to identify β5t-expressing cells, was specifically detected in the thymic cortex by histological analysis and in Ly51+ cTECs by flow cytometry analysis 4. Thus, it is reasonable that β5t expression in adult mice is highly specific for cTECs.
However, it is unknown whether β5t is expressed specifically in cTECs during early ontogeny and how β5t expression is regulated during cTEC development. It was recently reported that β5t in humans is detected not only in cTECs but also in a fraction of thymic DCs 7, raising the possibility that β5t in the mouse may also be expressed by a fraction of thymic DCs. A thorough understanding of the mechanisms regulating β5t expression is important because it is largely unclear how cTECs are generated and how cTEC development branches from the development of thymic medullary epithelial cells (mTECs). This study shows that β5t expression in mouse embryogenesis is initiated as early as E12.5 and β5t expression in cTECs is dependent on Foxn1 but independent of the development of thymocytes or mTECs.
β5t expression in adult mouse is exclusive to cTECs in the thymus
It was reported that β5t is expressed in cTECs 4. Indeed, by analyzing Venus-derived fluorescence expression in β5t+/Venus mice, in which one genomic allele contains the gene encoding the Venus green fluorescence protein instead of the β5t-coding sequence 4, β5t gene expression was detectable only in the thymic cortex and not in the thymic medulla or in other tissues and organs, including skin, lungs, esophagus, intestine, liver, brain, heart, lymph nodes, kidneys, spleen, and pancreas (data not shown). Immunofluorescence analysis of adult mouse tissues using β5t-specific antibody confirmed that β5t expression was specifically detected in the thymus but not in other tissues (data not shown) and was prominently localized in the thymic cortex rather than the thymic medulla, as shown by costaining for keratin and keratin 8 (Fig. 1A). In the thymic cortex, β5t signals were merged with Ly51, ER-TR4, and CD205, which were all expressed by cTECs (Fig. 1A). On the contrary, β5t expression was rarely detected in the thymic medulla and was not merged with G8.8, ER-TR5, MTS-10, or Ulex europaeus agglutinin 1 (UEA-1), which were all specific for mTECs in the immunofluorescence analysis (Fig. 1B). Within the thymic cortex, β5t signals were not merged with fibroblast-specific ER-TR7 or MTS-15, endothelial cell-specific CD31, macrophage-specific CD11b, or DC-specific CD11c (Fig. 1C). These results indicate that in the adult mouse, β5t is exclusively expressed in the thymus by cTECs and not by other thymic stromal cells including DCs as well as mTECs, fibroblasts, endothelial cells, and macrophages.
The failure to detect β5t in DCs even in the thymic cortex contradicts recent results, showing that β5t is detected in a fraction of DCs in the cortex of human thymus sections 7. However, we found that β5t mRNA, which was abundantly detected in cTECs, was undetectable in isolated DCs (<0.15% of the amount in cTECs) (Fig. 1D), indicating that most thymic DCs do not produce β5t. Thus, the β5t expression previously reported in DCs in human thymus sections 7 is not reproduced in mouse and possibly specific for human.
The β5t signals that were sparsely detected in the medullary region (Fig. 1A and B) were detected even in the thymus sections of β5t-deficient mice (Supporting Information Fig. 1), indicating that the sparsely detected β5t signals in the thymic medulla contain nonspecific background signals caused by the polyclonal anti-β5t antibody. However, it is possible that a minor fraction of the cells localized in the thymic medulla also express β5t.
Ontogeny of cTEC-specific β5t expression during embryogenesis
We then visualized the expression of β5t systemically in E14.5 mouse embryos. As shown in Fig. 2A–F, antibody-detected β5t signals were exclusively found in the thymus and merged with most keratin signals. Thus, β5t is exclusively expressed in the thymus by keratin-expressing TECs during embryogenesis. The expression of β5t in mouse embryos was further analyzed to determine the stage when β5t was first expressed. β5t mRNA was undetectable in feeder-cell-free EB3 embryonic stem (ES) cells (<0.01% of the amount in cTECs) (Fig. 2G), suggesting that β5t is not expressed in the inner cell mass of the blastocyst during early embryogenesis. β5t expression was undetectable, or extremely low, in mouse embryos on and before E10.5 (Fig. 2G and H). Both in β5t+/Venus B6-background and in normal B6 mouse embryos, β5t expression in the thymic primordium at the third pharyngeal pouch was detected at E12.5 but not at E11.5 (Fig. 3A), indicating that β5t expression in the thymus emerges at E12.5 in B6-background mice. Interestingly, the β5t signals in the E12.5 thymus tended to be localized at the ventral and outer region rather than the dorsal and inner region of the thymic primordium (Fig. 3A and B) and were merged with cTEC-specific CD205 but not with mTEC-specific claudins 3 and 4 (Fig. 3B), indicating that the thymus-specific expression of β5t is initiated as early as E12.5 exclusively in cTECs that are localized differently from mTECs. The exclusive expression of β5t in cTECs but not mTECs continues up to E14.5 (Fig. 3C) and later gestational stages, as well as newborn and adult mice, including aged mice (data not shown). These results indicate that β5t expression in mouse embryogenesis is initiated as early as E12.5 and exclusive to cTECs.
β5t expression by cTECs is independent of thymic medulla formation or thymocyte development
We next analyzed β5t expression in mutant mice in which thymic medulla formation or thymocyte development is blocked. RelB is a transcription factor that is required for thymic medulla formation 8, 9. In RelB-deficient mice, the development of mTECs and the formation of the thymic medulla are severely impaired 8, 9. However, β5t expression in cTECs was not impaired in RelB-deficient mice (Fig. 4A), indicating that cTEC-exclusive β5t expression is independent of RelB-dependent thymic medulla formation.
We also analyzed β5t expression in the thymus of mice in which thymocyte development is defective. In mice deficient for Rag2 or doubly deficient for TCRβ and TCRδ, thymocyte development is blocked at the CD4−CD8− double negative (DN) 3 stage 10, 11, whereas thymocyte development is blocked by the DN2 stage in human CD3ε-transgenic tgε26 mice 12, 13. Figure 4A shows that β5t expression in the thymus is readily detectable in Rag2-deficient mice, TCRβTCRδ double-deficient mice, and human CD3ε-transgenic tgε26 mice, indicating that developing thymocytes beyond the DN stage are not essential for the cTEC-specific expression of β5t.
The role of developing thymocytes in the cTEC-specific β5t expression was further analyzed in E12.5 embryos of mice that were doubly deficient for CCR7 and CCR9, in which fetal thymus colonization by leukocytes is severely defective due to the lack of chemokine-mediated attraction of T-lymphoid progenitor cells to the fetal thymic primordium 14. Indeed, the number of thymocytes that accumulated in the E12.5 thymic primordium was markedly decreased in mice that were doubly deficient for CCR7 and CCR9 (Fig. 4B). However, we detected unimpaired β5t expression in E12.5 TECs in mice that were doubly deficient for CCR7 and CCR9 (Fig. 4B). These results indicate that thymus colonization by leukocytes or the intrathymic development of thymocytes is not required for the cTEC-specific expression of β5t.
β5t expression by cTECs is Foxn1-dependent
We then examined whether the thymus-specific β5t expression is regulated by Foxn1, a gene that encodes a transcription factor essential for thymus epithelial development 15, 16. To do so, we analyzed β5t expression in BALB/c-nude mice, in which Foxn1 is spontaneously defective. We first examined the ontogeny of β5t expression in BALB/c mouse embryos, because it is known that embryonic development of the thymus is slightly delayed in BALB/c mouse when compared with B6 mice 17. Indeed, we detected β5t expression in BALB/c embryonic thymus as early as E13.5 but not E12.5 (Fig. 5A), 1 day later than the expression in B6 embryonic thymus (Fig. 3A). We then examined β5t expression in the thymic primordium generated in the embryos of E13.5 BALB/c-nude mice. One in every five serial cryosections that contained the entire third pharyngeal pouch was stained for β5t. Unlike the clear detection of β5t as well as CD45 signals in the E13.5 thymic primordium of control BALB/c-nu/+ mice, no β5t expression and very few CD45+ leukocytes were detected in the thymic primordium of E13.5 BALB/c-nu/nu mice (Fig. 5B). Thus, β5t expression detected in mouse embryonic thymus is dependent on Foxn1.
Phenotypic progression of β5t-expressing cTECs during ontogeny
Finally, we examined whether β5t-expressing cTECs undergo developmental changes or β5t expression represent a cTEC population that is unvarying throughout ontogeny. To do so, intracellular β5t+CD45−EpCAM+cTECs were analyzed by flow cytometry for the expression of cell-surface EpCAM, CD205, Ly51, and I-A molecules. Intracellular β5t staining in cTECs was specific, as the β5t signals were severely diminished in cTECs from β5t-deficient mice (Fig. 6A). As shown in Fig. 6B, the amount of EpCAM expressed on the surface of β5t-expressing cTECs decreased during ontogeny from E15.5 to 14 days old. On the other hand, the amount of CD205, Ly51, and I-A expressed on the surface of β5t-expressing cTECs increased during ontogeny (Fig. 6B). These results indicate that β5t-expressing cTECs dynamically change in phenotype during ontogeny.
The present results show that β5t is exclusively expressed in adult mouse thymus by cTECs and not by other thymic stromal cells including DCs as well as mTECs, fibroblasts, endothelial cells, and macrophages. The results also indicate that β5t expression during mouse embryogenesis is initiated as early as E12.5. The cTEC-exclusive expression of β5t is dependent on Foxn1 but independent of RelB-dependent thymic medulla formation or lymphoid cell colonization of the embryonic thymus. The result that β5t expression is highly restricted to a subpopulation of TECs not only in postnatal mice but also during mouse embryogenesis contrasts those of other thymus-specific genes so far described. For example, Aire is a nuclear factor that is strongly expressed in a fraction of mTECs but not in cTECs and is responsible for protection from the type 1 autoimmune polyendocrinopathy syndrome 18–20. However, unlike β5t, Aire is also detectable in various cells other than mTECs, including ES cells (Fig. 2G) and spermatocytes 21, 22, suggesting that Aire is not only an autoimmune regulator but also functional in germ cell development and early embryogenesis. Foxn1, a transcription factor essential for thymus development, is also expressed in skin epithelial cells and involved in hair development 15, 23. On the contrary, as far as we are aware of, β5t expression is exclusive to the TEC lineage. Thus, an understanding of the molecular and cellular mechanisms that regulate β5t expression should be useful to uncover the mechanisms underlying the development of TECs and TEC subpopulations, including cTECs.
We found that the earliest cells that express β5t during mouse embryogenesis emerge at E12.5 in the ventral and outer region of the thymic primordium, the region that is distinct from the dorsal and inner region of the thymic primordium where the development of mTECs is initiated for the formation of the thymic medulla 24. It was previously shown that both cTECs and mTECs are derived from the endodermal epithelium of the third pharyngeal pouch 25, 26 via common progenitor cells 27, 28. It is thus possible that the anatomically distinct inner and outer layers of the third pharyngeal pouch may have been already sorted to generate mTECs and cTECs respectively, by E12.5. As cTECs appear to originate mostly from the outer layer, cTEC development may require a close interaction with mesenchymal cells that encapsulate the thymic primordium.
Our results show that the cTEC expression of β5t is detected even in the absence of RelB-dependent thymic medulla formation. RelB, a member of the NF-κB family of transcription factors, is essential for the development and maturation of mTECs 8, 9, 29. The undisturbed β5t expression in the thymic cortex of RelB-deficient mice indicates that the development of β5t-expressing cTECs can take place independently of RelB-dependent mTEC development and without crosstalk signals from the mature thymic medulla.
Our results also show that cTEC expression of β5t is independent of thymocyte development. In Rag2-deficient mice, TCRβTCRδ double-deficient mice, and human CD3ε-transgenic tgε26 mice, β5t expression in the thymic cortex is readily detectable. It is especially interesting to note that TEC expression of β5t is not impaired in the tgε26 mice in which thymocyte development is arrested by the DN2 stage 12, 13 and the three-dimensional epithelial meshwork of the thymic cortex is severely defective in association with the formation of cysts 30, 31. Thus, the thymic crosstalk signal that is derived from thymocytes beyond the DN2 stage, which is defective in the tgε26 mice and regulates the late process of cTEC development 30–33, is dispensable for the development of cTECs to the stage of β5t expression. That cTEC expression of β5t can occur independently of thymocytes is further supported by our results showing that embryonic β5t expression in the thymus is not diminished in mice doubly deficient for CCR7 and CCR9. The combination of chemokine signals mediated by CCR7 and CCR9 is essential for fetal thymus colonization by leukocytes at the prevascular stage 14. In this regard, our results reinforce the notion that the development of β5t-expressing cTECs does not require crosstalk signals derived from neighboring thymocytes.
Our results further show that the amount of cell surface molecules, such as EpCAM, CD205, Ly51, and I-A, expressed by β5t-expressing cTECs varies during ontogeny (Fig. 6B). Thus, β5t-expressing cTECs dynamically vary in phenotype during ontogeny, suggesting that β5t expression by cTECs does not represent an unvarying status of terminal cellular development for cTECs.
Finally, our results indicate that β5t expression in cTECs is dependent on Foxn1. It is possible that Foxn1 directly contributes to the regulation of β5t transcription. However, Foxn1 is necessary for the development of immature TEC progenitor cells into cTECs and mTECs 16, 34. Since β5t is not detectable at E11.5 thymic primordium (Fig. 3A), β5t is unlikely to be expressed by immature TEC progenitor cells. Thus, it is possible that lack of β5t in nude mouse thymus may reflect the absence of cTECs by the lack of Foxn1.
In conclusion, the present study identified that β5t, a thymoproteasome component exclusively expressed in cTECs, is useful for further studies of the development of cTECs and other TEC lineages. Our results demonstrate that the appearance of β5t-expressing cTECs is independent of thymic medulla formation or thymocyte seeding but is dependent on Foxn1. Thus, it would be important to clarify the transcriptional mechanisms that lead to β5t expression in cTECs, which would pave the way for a better understanding of how the thymic microenvironment is formed during development and regenerated in the elderly or following chemotherapy and/or radiotherapy.
Materials and methods
C57BL/6 (B6), BALB/c, BALB/c-nu/+, and BALB/c-nu/nu mice were obtained from SLC (Shizuoka, Japan). β5tVenus/Venus4, CCR7−/−35, CCR9−/−36, Rag2−/−10, TCRβ−/−TCRδ−/−37, 38, RelB−/−8, and human CD3ε-transgenic tgε26 12 mice were described previously. The day a vaginal plug was first observed was designated as gestation day 0.5 (E0.5). All mice were maintained under specific pathogen-free conditions and all experiments were carried out under the approval of the Institutional Animal Care and Use Committee of the University of Tokushima.
Multicolor confocal microscopy analysis
Fresh tissues were embedded in OCT compound (Sakura Finetek). Frozen sections measuring 5 μm thick were fixed with either acetone or paraformaldehyde and stained with the following antibodies: rabbit anti-mouse β5t 4, mouse anti-keratin (C-11) (BioLegend), mouse anti-keratin 8 (Progen Biotechnik), anti-mouse CD31 (BioLegend), anti-mouse CD45 (eBioscience), anti-mouse EpCAM/CD326 (G8.8) (BD Pharmingen), biotinylated anti-mouse Ly51 (Pharmingen), biotinylated anti-mouse CD205 (eBioscience), biotinylated anti-mouse CD11b (eBioscience), biotinylated anti-mouse CD11c (HL3) (BD Pharmingen), and biotinylated UEA-1 (Vector Laboratories). MTS-10 and MTS-15 were gifts from Dr. Richard Boyd 39, 40, and ER-TR4, ER-TR5, and ER-TR7 were gifts from Dr. Willem van Ewijk 41, 42. The C-terminal fragment of Clostridium perfringens enterotoxin (C-CPE) was biotinylated and used for the detection of claudins 3 and 4 24. Staining was visualized with fluorescein-conjugated anti-rabbit IgG (Molecular Probes), Alexa633-conjugated anti-rat IgG (Molecular Probes), Alexa633-conjugated anti-mouse IgG (Molecular Probes), or Alexa633-conjugated streptavidin (Molecular Probes). Stained sections were mounted with a fluorescence mounting medium (Dako). Images were analyzed with a TSC SP2 confocal laser-scanning microscope and Leica Confocal software version 2.6.
Isolation of thymic stromal cells
For thymic stromal cell preparation, minced thymuses of adult B6 mice were digested with collagenase and DNase I, as described previously 43. CD45−I-A+UEA1− cTECs, CD45−I-A+UEA1+mTECs, and CD45+CD11Chigh DCs were sorted using FACS-Vantage (BD Biosciences), as described previously 44. Multi-color flow cytometry analysis of thymic stromal cells was carried out using FACS-Aria.
Quantitative mRNA analysis
Total cellular RNA was reverse-transcribed with oligo-dT primer and Superscript III reverse transcriptase (Invitrogen). mRNA expression of β5t, Aire, CD11c, and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was quantified by real-time PCR with SYBR Premix Ex Taq (TaKaRa), Light Cycler DX400 (Roche), and the following primer sets: β5t, 5′-CTCTGTGGCTGGGACCACTC-3′ and 5′-TCCGCTCTCCCGAACGTGG-3′; Aire, 5′-ACCCAACAAGTTCGAAGACCC-3′ and 5′-GACAGCCGTCACAACAGATGA-3′; CD11c, 5′-ATGTTGGAGGAAGCAAATGG-3′ and 5′-CCTGGGAATCCTATTGCAGA-3′; and GAPDH, 5′-CCGGTGCTGAGTATGTCGTG-3′ and 5′-CAGTCTTCTGGGTGGCAGTG-3′. Amplified signals were confirmed to be single bands over gel electrophoresis and normalized to GAPDH.
The authors thank Dr. Willem van Ewijk, Dr. Richard Boyd, Dr. Mitsuru Matsumoto, and Dr. Naozumi Ishimaru for reagents and materials and Dr. Shigeo Koyasu for reading the manuscript. A. M. R. is a graduate student supported by a MEXT scholarship. This study was supported by a MEXT Grant-in-Aid for Scientific Research on Priority Area “Immunological Self” (to Y. T.).
Conflict of interest: The authors declare no financial or commercial conflict of interest.